1 .. SPDX-License-Identifier: GPL-2.0 1 .. SPDX-License-Identifier: GPL-2.0
2 2
3 ============================================== 3 ===================================================================
4 The Definitive KVM (Kernel-based Virtual Machi 4 The Definitive KVM (Kernel-based Virtual Machine) API Documentation
5 ============================================== 5 ===================================================================
6 6
7 1. General description 7 1. General description
8 ====================== 8 ======================
9 9
10 The kvm API is a set of ioctls that are issued 10 The kvm API is a set of ioctls that are issued to control various aspects
11 of a virtual machine. The ioctls belong to th 11 of a virtual machine. The ioctls belong to the following classes:
12 12
13 - System ioctls: These query and set global a 13 - System ioctls: These query and set global attributes which affect the
14 whole kvm subsystem. In addition a system 14 whole kvm subsystem. In addition a system ioctl is used to create
15 virtual machines. 15 virtual machines.
16 16
17 - VM ioctls: These query and set attributes t 17 - VM ioctls: These query and set attributes that affect an entire virtual
18 machine, for example memory layout. In add 18 machine, for example memory layout. In addition a VM ioctl is used to
19 create virtual cpus (vcpus) and devices. 19 create virtual cpus (vcpus) and devices.
20 20
21 VM ioctls must be issued from the same proc 21 VM ioctls must be issued from the same process (address space) that was
22 used to create the VM. 22 used to create the VM.
23 23
24 - vcpu ioctls: These query and set attributes 24 - vcpu ioctls: These query and set attributes that control the operation
25 of a single virtual cpu. 25 of a single virtual cpu.
26 26
27 vcpu ioctls should be issued from the same 27 vcpu ioctls should be issued from the same thread that was used to create
28 the vcpu, except for asynchronous vcpu ioct 28 the vcpu, except for asynchronous vcpu ioctl that are marked as such in
29 the documentation. Otherwise, the first io 29 the documentation. Otherwise, the first ioctl after switching threads
30 could see a performance impact. 30 could see a performance impact.
31 31
32 - device ioctls: These query and set attribut 32 - device ioctls: These query and set attributes that control the operation
33 of a single device. 33 of a single device.
34 34
35 device ioctls must be issued from the same 35 device ioctls must be issued from the same process (address space) that
36 was used to create the VM. 36 was used to create the VM.
37 37
38 2. File descriptors 38 2. File descriptors
39 =================== 39 ===================
40 40
41 The kvm API is centered around file descriptor 41 The kvm API is centered around file descriptors. An initial
42 open("/dev/kvm") obtains a handle to the kvm s 42 open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
43 can be used to issue system ioctls. A KVM_CRE 43 can be used to issue system ioctls. A KVM_CREATE_VM ioctl on this
44 handle will create a VM file descriptor which 44 handle will create a VM file descriptor which can be used to issue VM
45 ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVIC 45 ioctls. A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will
46 create a virtual cpu or device and return a fi 46 create a virtual cpu or device and return a file descriptor pointing to
47 the new resource. Finally, ioctls on a vcpu o 47 the new resource. Finally, ioctls on a vcpu or device fd can be used
48 to control the vcpu or device. For vcpus, thi 48 to control the vcpu or device. For vcpus, this includes the important
49 task of actually running guest code. 49 task of actually running guest code.
50 50
51 In general file descriptors can be migrated am 51 In general file descriptors can be migrated among processes by means
52 of fork() and the SCM_RIGHTS facility of unix 52 of fork() and the SCM_RIGHTS facility of unix domain socket. These
53 kinds of tricks are explicitly not supported b 53 kinds of tricks are explicitly not supported by kvm. While they will
54 not cause harm to the host, their actual behav 54 not cause harm to the host, their actual behavior is not guaranteed by
55 the API. See "General description" for detail 55 the API. See "General description" for details on the ioctl usage
56 model that is supported by KVM. 56 model that is supported by KVM.
57 57
58 It is important to note that although VM ioctl !! 58 It is important to note that althought VM ioctls may only be issued from
59 the process that created the VM, a VM's lifecy 59 the process that created the VM, a VM's lifecycle is associated with its
60 file descriptor, not its creator (process). I 60 file descriptor, not its creator (process). In other words, the VM and
61 its resources, *including the associated addre 61 its resources, *including the associated address space*, are not freed
62 until the last reference to the VM's file desc 62 until the last reference to the VM's file descriptor has been released.
63 For example, if fork() is issued after ioctl(K 63 For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will
64 not be freed until both the parent (original) 64 not be freed until both the parent (original) process and its child have
65 put their references to the VM's file descript 65 put their references to the VM's file descriptor.
66 66
67 Because a VM's resources are not freed until t 67 Because a VM's resources are not freed until the last reference to its
68 file descriptor is released, creating addition 68 file descriptor is released, creating additional references to a VM
69 via fork(), dup(), etc... without careful cons 69 via fork(), dup(), etc... without careful consideration is strongly
70 discouraged and may have unwanted side effects 70 discouraged and may have unwanted side effects, e.g. memory allocated
71 by and on behalf of the VM's process may not b 71 by and on behalf of the VM's process may not be freed/unaccounted when
72 the VM is shut down. 72 the VM is shut down.
73 73
74 74
75 3. Extensions 75 3. Extensions
76 ============= 76 =============
77 77
78 As of Linux 2.6.22, the KVM ABI has been stabi 78 As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
79 incompatible change are allowed. However, the 79 incompatible change are allowed. However, there is an extension
80 facility that allows backward-compatible exten 80 facility that allows backward-compatible extensions to the API to be
81 queried and used. 81 queried and used.
82 82
83 The extension mechanism is not based on the Li 83 The extension mechanism is not based on the Linux version number.
84 Instead, kvm defines extension identifiers and 84 Instead, kvm defines extension identifiers and a facility to query
85 whether a particular extension identifier is a 85 whether a particular extension identifier is available. If it is, a
86 set of ioctls is available for application use 86 set of ioctls is available for application use.
87 87
88 88
89 4. API description 89 4. API description
90 ================== 90 ==================
91 91
92 This section describes ioctls that can be used 92 This section describes ioctls that can be used to control kvm guests.
93 For each ioctl, the following information is p 93 For each ioctl, the following information is provided along with a
94 description: 94 description:
95 95
96 Capability: 96 Capability:
97 which KVM extension provides this ioctl. 97 which KVM extension provides this ioctl. Can be 'basic',
98 which means that is will be provided by 98 which means that is will be provided by any kernel that supports
99 API version 12 (see section 4.1), a KVM_ 99 API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
100 means availability needs to be checked w 100 means availability needs to be checked with KVM_CHECK_EXTENSION
101 (see section 4.4), or 'none' which means 101 (see section 4.4), or 'none' which means that while not all kernels
102 support this ioctl, there's no capabilit 102 support this ioctl, there's no capability bit to check its
103 availability: for kernels that don't sup 103 availability: for kernels that don't support the ioctl,
104 the ioctl returns -ENOTTY. 104 the ioctl returns -ENOTTY.
105 105
106 Architectures: 106 Architectures:
107 which instruction set architectures prov 107 which instruction set architectures provide this ioctl.
108 x86 includes both i386 and x86_64. 108 x86 includes both i386 and x86_64.
109 109
110 Type: 110 Type:
111 system, vm, or vcpu. 111 system, vm, or vcpu.
112 112
113 Parameters: 113 Parameters:
114 what parameters are accepted by the ioct 114 what parameters are accepted by the ioctl.
115 115
116 Returns: 116 Returns:
117 the return value. General error numbers 117 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
118 are not detailed, but errors with specif 118 are not detailed, but errors with specific meanings are.
119 119
120 120
121 4.1 KVM_GET_API_VERSION 121 4.1 KVM_GET_API_VERSION
122 ----------------------- 122 -----------------------
123 123
124 :Capability: basic 124 :Capability: basic
125 :Architectures: all 125 :Architectures: all
126 :Type: system ioctl 126 :Type: system ioctl
127 :Parameters: none 127 :Parameters: none
128 :Returns: the constant KVM_API_VERSION (=12) 128 :Returns: the constant KVM_API_VERSION (=12)
129 129
130 This identifies the API version as the stable 130 This identifies the API version as the stable kvm API. It is not
131 expected that this number will change. Howeve 131 expected that this number will change. However, Linux 2.6.20 and
132 2.6.21 report earlier versions; these are not 132 2.6.21 report earlier versions; these are not documented and not
133 supported. Applications should refuse to run 133 supported. Applications should refuse to run if KVM_GET_API_VERSION
134 returns a value other than 12. If this check 134 returns a value other than 12. If this check passes, all ioctls
135 described as 'basic' will be available. 135 described as 'basic' will be available.
136 136
137 137
138 4.2 KVM_CREATE_VM 138 4.2 KVM_CREATE_VM
139 ----------------- 139 -----------------
140 140
141 :Capability: basic 141 :Capability: basic
142 :Architectures: all 142 :Architectures: all
143 :Type: system ioctl 143 :Type: system ioctl
144 :Parameters: machine type identifier (KVM_VM_* 144 :Parameters: machine type identifier (KVM_VM_*)
145 :Returns: a VM fd that can be used to control 145 :Returns: a VM fd that can be used to control the new virtual machine.
146 146
147 The new VM has no virtual cpus and no memory. 147 The new VM has no virtual cpus and no memory.
148 You probably want to use 0 as machine type. 148 You probably want to use 0 as machine type.
149 149
150 X86: <<
151 ^^^^ <<
152 <<
153 Supported X86 VM types can be queried via KVM_ <<
154 <<
155 S390: <<
156 ^^^^^ <<
157 <<
158 In order to create user controlled virtual mac 150 In order to create user controlled virtual machines on S390, check
159 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_ 151 KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
160 privileged user (CAP_SYS_ADMIN). 152 privileged user (CAP_SYS_ADMIN).
161 153
162 MIPS: <<
163 ^^^^^ <<
164 <<
165 To use hardware assisted virtualization on MIP 154 To use hardware assisted virtualization on MIPS (VZ ASE) rather than
166 the default trap & emulate implementation (whi 155 the default trap & emulate implementation (which changes the virtual
167 memory layout to fit in user mode), check KVM_ 156 memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the
168 flag KVM_VM_MIPS_VZ. 157 flag KVM_VM_MIPS_VZ.
169 158
170 ARM64: <<
171 ^^^^^^ <<
172 159
173 On arm64, the physical address size for a VM ( 160 On arm64, the physical address size for a VM (IPA Size limit) is limited
174 to 40bits by default. The limit can be configu 161 to 40bits by default. The limit can be configured if the host supports the
175 extension KVM_CAP_ARM_VM_IPA_SIZE. When suppor 162 extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
176 KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the 163 KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
177 identifier, where IPA_Bits is the maximum widt 164 identifier, where IPA_Bits is the maximum width of any physical
178 address used by the VM. The IPA_Bits is encode 165 address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
179 machine type identifier. 166 machine type identifier.
180 167
181 e.g, to configure a guest to use 48bit physica 168 e.g, to configure a guest to use 48bit physical address size::
182 169
183 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_V 170 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
184 171
185 The requested size (IPA_Bits) must be: 172 The requested size (IPA_Bits) must be:
186 173
187 == ======================================== 174 == =========================================================
188 0 Implies default size, 40bits (for backwa 175 0 Implies default size, 40bits (for backward compatibility)
189 N Implies N bits, where N is a positive in 176 N Implies N bits, where N is a positive integer such that,
190 32 <= N <= Host_IPA_Limit 177 32 <= N <= Host_IPA_Limit
191 == ======================================== 178 == =========================================================
192 179
193 Host_IPA_Limit is the maximum possible value f 180 Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and
194 is dependent on the CPU capability and the ker 181 is dependent on the CPU capability and the kernel configuration. The limit can
195 be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of 182 be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION
196 ioctl() at run-time. 183 ioctl() at run-time.
197 184
198 Creation of the VM will fail if the requested 185 Creation of the VM will fail if the requested IPA size (whether it is
199 implicit or explicit) is unsupported on the ho 186 implicit or explicit) is unsupported on the host.
200 187
201 Please note that configuring the IPA size does 188 Please note that configuring the IPA size does not affect the capability
202 exposed by the guest CPUs in ID_AA64MMFR0_EL1[ 189 exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
203 size of the address translated by the stage2 l 190 size of the address translated by the stage2 level (guest physical to
204 host physical address translations). 191 host physical address translations).
205 192
206 193
207 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATUR 194 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
208 ---------------------------------------------- 195 ----------------------------------------------------------
209 196
210 :Capability: basic, KVM_CAP_GET_MSR_FEATURES f 197 :Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
211 :Architectures: x86 198 :Architectures: x86
212 :Type: system ioctl 199 :Type: system ioctl
213 :Parameters: struct kvm_msr_list (in/out) 200 :Parameters: struct kvm_msr_list (in/out)
214 :Returns: 0 on success; -1 on error 201 :Returns: 0 on success; -1 on error
215 202
216 Errors: 203 Errors:
217 204
218 ====== ================================= 205 ====== ============================================================
219 EFAULT the msr index list cannot be read 206 EFAULT the msr index list cannot be read from or written to
220 E2BIG the msr index list is too big to !! 207 E2BIG the msr index list is to be to fit in the array specified by
221 the user. 208 the user.
222 ====== ================================= 209 ====== ============================================================
223 210
224 :: 211 ::
225 212
226 struct kvm_msr_list { 213 struct kvm_msr_list {
227 __u32 nmsrs; /* number of msrs in entr 214 __u32 nmsrs; /* number of msrs in entries */
228 __u32 indices[0]; 215 __u32 indices[0];
229 }; 216 };
230 217
231 The user fills in the size of the indices arra 218 The user fills in the size of the indices array in nmsrs, and in return
232 kvm adjusts nmsrs to reflect the actual number 219 kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
233 indices array with their numbers. 220 indices array with their numbers.
234 221
235 KVM_GET_MSR_INDEX_LIST returns the guest msrs 222 KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list
236 varies by kvm version and host processor, but 223 varies by kvm version and host processor, but does not change otherwise.
237 224
238 Note: if kvm indicates supports MCE (KVM_CAP_M 225 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
239 not returned in the MSR list, as different vcp 226 not returned in the MSR list, as different vcpus can have a different number
240 of banks, as set via the KVM_X86_SETUP_MCE ioc 227 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
241 228
242 KVM_GET_MSR_FEATURE_INDEX_LIST returns the lis 229 KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
243 to the KVM_GET_MSRS system ioctl. This lets u 230 to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities
244 and processor features that are exposed via MS 231 and processor features that are exposed via MSRs (e.g., VMX capabilities).
245 This list also varies by kvm version and host 232 This list also varies by kvm version and host processor, but does not change
246 otherwise. 233 otherwise.
247 234
248 235
249 4.4 KVM_CHECK_EXTENSION 236 4.4 KVM_CHECK_EXTENSION
250 ----------------------- 237 -----------------------
251 238
252 :Capability: basic, KVM_CAP_CHECK_EXTENSION_VM 239 :Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
253 :Architectures: all 240 :Architectures: all
254 :Type: system ioctl, vm ioctl 241 :Type: system ioctl, vm ioctl
255 :Parameters: extension identifier (KVM_CAP_*) 242 :Parameters: extension identifier (KVM_CAP_*)
256 :Returns: 0 if unsupported; 1 (or some other p 243 :Returns: 0 if unsupported; 1 (or some other positive integer) if supported
257 244
258 The API allows the application to query about 245 The API allows the application to query about extensions to the core
259 kvm API. Userspace passes an extension identi 246 kvm API. Userspace passes an extension identifier (an integer) and
260 receives an integer that describes the extensi 247 receives an integer that describes the extension availability.
261 Generally 0 means no and 1 means yes, but some 248 Generally 0 means no and 1 means yes, but some extensions may report
262 additional information in the integer return v 249 additional information in the integer return value.
263 250
264 Based on their initialization different VMs ma 251 Based on their initialization different VMs may have different capabilities.
265 It is thus encouraged to use the vm ioctl to q 252 It is thus encouraged to use the vm ioctl to query for capabilities (available
266 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd) 253 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
267 254
268 4.5 KVM_GET_VCPU_MMAP_SIZE 255 4.5 KVM_GET_VCPU_MMAP_SIZE
269 -------------------------- 256 --------------------------
270 257
271 :Capability: basic 258 :Capability: basic
272 :Architectures: all 259 :Architectures: all
273 :Type: system ioctl 260 :Type: system ioctl
274 :Parameters: none 261 :Parameters: none
275 :Returns: size of vcpu mmap area, in bytes 262 :Returns: size of vcpu mmap area, in bytes
276 263
277 The KVM_RUN ioctl (cf.) communicates with user 264 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
278 memory region. This ioctl returns the size of 265 memory region. This ioctl returns the size of that region. See the
279 KVM_RUN documentation for details. 266 KVM_RUN documentation for details.
280 267
281 Besides the size of the KVM_RUN communication 268 Besides the size of the KVM_RUN communication region, other areas of
282 the VCPU file descriptor can be mmap-ed, inclu 269 the VCPU file descriptor can be mmap-ed, including:
283 270
284 - if KVM_CAP_COALESCED_MMIO is available, a pa 271 - if KVM_CAP_COALESCED_MMIO is available, a page at
285 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; 272 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons,
286 this page is included in the result of KVM_G 273 this page is included in the result of KVM_GET_VCPU_MMAP_SIZE.
287 KVM_CAP_COALESCED_MMIO is not documented yet 274 KVM_CAP_COALESCED_MMIO is not documented yet.
288 275
289 - if KVM_CAP_DIRTY_LOG_RING is available, a nu 276 - if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at
290 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For 277 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For more information on
291 KVM_CAP_DIRTY_LOG_RING, see section 8.3. 278 KVM_CAP_DIRTY_LOG_RING, see section 8.3.
292 279
293 280
>> 281 4.6 KVM_SET_MEMORY_REGION
>> 282 -------------------------
>> 283
>> 284 :Capability: basic
>> 285 :Architectures: all
>> 286 :Type: vm ioctl
>> 287 :Parameters: struct kvm_memory_region (in)
>> 288 :Returns: 0 on success, -1 on error
>> 289
>> 290 This ioctl is obsolete and has been removed.
>> 291
>> 292
294 4.7 KVM_CREATE_VCPU 293 4.7 KVM_CREATE_VCPU
295 ------------------- 294 -------------------
296 295
297 :Capability: basic 296 :Capability: basic
298 :Architectures: all 297 :Architectures: all
299 :Type: vm ioctl 298 :Type: vm ioctl
300 :Parameters: vcpu id (apic id on x86) 299 :Parameters: vcpu id (apic id on x86)
301 :Returns: vcpu fd on success, -1 on error 300 :Returns: vcpu fd on success, -1 on error
302 301
303 This API adds a vcpu to a virtual machine. No 302 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
304 The vcpu id is an integer in the range [0, max 303 The vcpu id is an integer in the range [0, max_vcpu_id).
305 304
306 The recommended max_vcpus value can be retriev 305 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
307 the KVM_CHECK_EXTENSION ioctl() at run-time. 306 the KVM_CHECK_EXTENSION ioctl() at run-time.
308 The maximum possible value for max_vcpus can b 307 The maximum possible value for max_vcpus can be retrieved using the
309 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION i 308 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
310 309
311 If the KVM_CAP_NR_VCPUS does not exist, you sh 310 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
312 cpus max. 311 cpus max.
313 If the KVM_CAP_MAX_VCPUS does not exist, you s 312 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
314 same as the value returned from KVM_CAP_NR_VCP 313 same as the value returned from KVM_CAP_NR_VCPUS.
315 314
316 The maximum possible value for max_vcpu_id can 315 The maximum possible value for max_vcpu_id can be retrieved using the
317 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION 316 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
318 317
319 If the KVM_CAP_MAX_VCPU_ID does not exist, you 318 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
320 is the same as the value returned from KVM_CAP 319 is the same as the value returned from KVM_CAP_MAX_VCPUS.
321 320
322 On powerpc using book3s_hv mode, the vcpus are 321 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
323 threads in one or more virtual CPU cores. (Th 322 threads in one or more virtual CPU cores. (This is because the
324 hardware requires all the hardware threads in 323 hardware requires all the hardware threads in a CPU core to be in the
325 same partition.) The KVM_CAP_PPC_SMT capabili 324 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
326 of vcpus per virtual core (vcore). The vcore 325 of vcpus per virtual core (vcore). The vcore id is obtained by
327 dividing the vcpu id by the number of vcpus pe 326 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
328 given vcore will always be in the same physica 327 given vcore will always be in the same physical core as each other
329 (though that might be a different physical cor 328 (though that might be a different physical core from time to time).
330 Userspace can control the threading (SMT) mode 329 Userspace can control the threading (SMT) mode of the guest by its
331 allocation of vcpu ids. For example, if users 330 allocation of vcpu ids. For example, if userspace wants
332 single-threaded guest vcpus, it should make al 331 single-threaded guest vcpus, it should make all vcpu ids be a multiple
333 of the number of vcpus per vcore. 332 of the number of vcpus per vcore.
334 333
335 For virtual cpus that have been created with S 334 For virtual cpus that have been created with S390 user controlled virtual
336 machines, the resulting vcpu fd can be memory 335 machines, the resulting vcpu fd can be memory mapped at page offset
337 KVM_S390_SIE_PAGE_OFFSET in order to obtain a 336 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
338 cpu's hardware control block. 337 cpu's hardware control block.
339 338
340 339
341 4.8 KVM_GET_DIRTY_LOG (vm ioctl) 340 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
342 -------------------------------- 341 --------------------------------
343 342
344 :Capability: basic 343 :Capability: basic
345 :Architectures: all 344 :Architectures: all
346 :Type: vm ioctl 345 :Type: vm ioctl
347 :Parameters: struct kvm_dirty_log (in/out) 346 :Parameters: struct kvm_dirty_log (in/out)
348 :Returns: 0 on success, -1 on error 347 :Returns: 0 on success, -1 on error
349 348
350 :: 349 ::
351 350
352 /* for KVM_GET_DIRTY_LOG */ 351 /* for KVM_GET_DIRTY_LOG */
353 struct kvm_dirty_log { 352 struct kvm_dirty_log {
354 __u32 slot; 353 __u32 slot;
355 __u32 padding; 354 __u32 padding;
356 union { 355 union {
357 void __user *dirty_bitmap; /* 356 void __user *dirty_bitmap; /* one bit per page */
358 __u64 padding; 357 __u64 padding;
359 }; 358 };
360 }; 359 };
361 360
362 Given a memory slot, return a bitmap containin 361 Given a memory slot, return a bitmap containing any pages dirtied
363 since the last call to this ioctl. Bit 0 is t 362 since the last call to this ioctl. Bit 0 is the first page in the
364 memory slot. Ensure the entire structure is c 363 memory slot. Ensure the entire structure is cleared to avoid padding
365 issues. 364 issues.
366 365
367 If KVM_CAP_MULTI_ADDRESS_SPACE is available, b 366 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
368 the address space for which you want to return 367 the address space for which you want to return the dirty bitmap. See
369 KVM_SET_USER_MEMORY_REGION for details on the 368 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
370 369
371 The bits in the dirty bitmap are cleared befor 370 The bits in the dirty bitmap are cleared before the ioctl returns, unless
372 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. 371 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information,
373 see the description of the capability. 372 see the description of the capability.
374 373
375 Note that the Xen shared_info page, if configu !! 374 4.9 KVM_SET_MEMORY_ALIAS
376 to be dirty. KVM will not explicitly mark it s !! 375 ------------------------
>> 376
>> 377 :Capability: basic
>> 378 :Architectures: x86
>> 379 :Type: vm ioctl
>> 380 :Parameters: struct kvm_memory_alias (in)
>> 381 :Returns: 0 (success), -1 (error)
>> 382
>> 383 This ioctl is obsolete and has been removed.
377 384
378 385
379 4.10 KVM_RUN 386 4.10 KVM_RUN
380 ------------ 387 ------------
381 388
382 :Capability: basic 389 :Capability: basic
383 :Architectures: all 390 :Architectures: all
384 :Type: vcpu ioctl 391 :Type: vcpu ioctl
385 :Parameters: none 392 :Parameters: none
386 :Returns: 0 on success, -1 on error 393 :Returns: 0 on success, -1 on error
387 394
388 Errors: 395 Errors:
389 396
390 ======= ================================= 397 ======= ==============================================================
391 EINTR an unmasked signal is pending 398 EINTR an unmasked signal is pending
392 ENOEXEC the vcpu hasn't been initialized 399 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute
393 instructions from device memory ( 400 instructions from device memory (arm64)
394 ENOSYS data abort outside memslots with 401 ENOSYS data abort outside memslots with no syndrome info and
395 KVM_CAP_ARM_NISV_TO_USER not enab 402 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
396 EPERM SVE feature set but not finalized 403 EPERM SVE feature set but not finalized (arm64)
397 ======= ================================= 404 ======= ==============================================================
398 405
399 This ioctl is used to run a guest virtual cpu. 406 This ioctl is used to run a guest virtual cpu. While there are no
400 explicit parameters, there is an implicit para 407 explicit parameters, there is an implicit parameter block that can be
401 obtained by mmap()ing the vcpu fd at offset 0, 408 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
402 KVM_GET_VCPU_MMAP_SIZE. The parameter block i 409 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
403 kvm_run' (see below). 410 kvm_run' (see below).
404 411
405 412
406 4.11 KVM_GET_REGS 413 4.11 KVM_GET_REGS
407 ----------------- 414 -----------------
408 415
409 :Capability: basic 416 :Capability: basic
410 :Architectures: all except arm64 !! 417 :Architectures: all except ARM, arm64
411 :Type: vcpu ioctl 418 :Type: vcpu ioctl
412 :Parameters: struct kvm_regs (out) 419 :Parameters: struct kvm_regs (out)
413 :Returns: 0 on success, -1 on error 420 :Returns: 0 on success, -1 on error
414 421
415 Reads the general purpose registers from the v 422 Reads the general purpose registers from the vcpu.
416 423
417 :: 424 ::
418 425
419 /* x86 */ 426 /* x86 */
420 struct kvm_regs { 427 struct kvm_regs {
421 /* out (KVM_GET_REGS) / in (KVM_SET_RE 428 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
422 __u64 rax, rbx, rcx, rdx; 429 __u64 rax, rbx, rcx, rdx;
423 __u64 rsi, rdi, rsp, rbp; 430 __u64 rsi, rdi, rsp, rbp;
424 __u64 r8, r9, r10, r11; 431 __u64 r8, r9, r10, r11;
425 __u64 r12, r13, r14, r15; 432 __u64 r12, r13, r14, r15;
426 __u64 rip, rflags; 433 __u64 rip, rflags;
427 }; 434 };
428 435
429 /* mips */ 436 /* mips */
430 struct kvm_regs { 437 struct kvm_regs {
431 /* out (KVM_GET_REGS) / in (KVM_SET_RE 438 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
432 __u64 gpr[32]; 439 __u64 gpr[32];
433 __u64 hi; 440 __u64 hi;
434 __u64 lo; 441 __u64 lo;
435 __u64 pc; 442 __u64 pc;
436 }; 443 };
437 444
438 /* LoongArch */ <<
439 struct kvm_regs { <<
440 /* out (KVM_GET_REGS) / in (KVM_SET_RE <<
441 unsigned long gpr[32]; <<
442 unsigned long pc; <<
443 }; <<
444 <<
445 445
446 4.12 KVM_SET_REGS 446 4.12 KVM_SET_REGS
447 ----------------- 447 -----------------
448 448
449 :Capability: basic 449 :Capability: basic
450 :Architectures: all except arm64 !! 450 :Architectures: all except ARM, arm64
451 :Type: vcpu ioctl 451 :Type: vcpu ioctl
452 :Parameters: struct kvm_regs (in) 452 :Parameters: struct kvm_regs (in)
453 :Returns: 0 on success, -1 on error 453 :Returns: 0 on success, -1 on error
454 454
455 Writes the general purpose registers into the 455 Writes the general purpose registers into the vcpu.
456 456
457 See KVM_GET_REGS for the data structure. 457 See KVM_GET_REGS for the data structure.
458 458
459 459
460 4.13 KVM_GET_SREGS 460 4.13 KVM_GET_SREGS
461 ------------------ 461 ------------------
462 462
463 :Capability: basic 463 :Capability: basic
464 :Architectures: x86, ppc 464 :Architectures: x86, ppc
465 :Type: vcpu ioctl 465 :Type: vcpu ioctl
466 :Parameters: struct kvm_sregs (out) 466 :Parameters: struct kvm_sregs (out)
467 :Returns: 0 on success, -1 on error 467 :Returns: 0 on success, -1 on error
468 468
469 Reads special registers from the vcpu. 469 Reads special registers from the vcpu.
470 470
471 :: 471 ::
472 472
473 /* x86 */ 473 /* x86 */
474 struct kvm_sregs { 474 struct kvm_sregs {
475 struct kvm_segment cs, ds, es, fs, gs, 475 struct kvm_segment cs, ds, es, fs, gs, ss;
476 struct kvm_segment tr, ldt; 476 struct kvm_segment tr, ldt;
477 struct kvm_dtable gdt, idt; 477 struct kvm_dtable gdt, idt;
478 __u64 cr0, cr2, cr3, cr4, cr8; 478 __u64 cr0, cr2, cr3, cr4, cr8;
479 __u64 efer; 479 __u64 efer;
480 __u64 apic_base; 480 __u64 apic_base;
481 __u64 interrupt_bitmap[(KVM_NR_INTERRU 481 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
482 }; 482 };
483 483
484 /* ppc -- see arch/powerpc/include/uapi/asm/ 484 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
485 485
486 interrupt_bitmap is a bitmap of pending extern 486 interrupt_bitmap is a bitmap of pending external interrupts. At most
487 one bit may be set. This interrupt has been a 487 one bit may be set. This interrupt has been acknowledged by the APIC
488 but not yet injected into the cpu core. 488 but not yet injected into the cpu core.
489 489
490 490
491 4.14 KVM_SET_SREGS 491 4.14 KVM_SET_SREGS
492 ------------------ 492 ------------------
493 493
494 :Capability: basic 494 :Capability: basic
495 :Architectures: x86, ppc 495 :Architectures: x86, ppc
496 :Type: vcpu ioctl 496 :Type: vcpu ioctl
497 :Parameters: struct kvm_sregs (in) 497 :Parameters: struct kvm_sregs (in)
498 :Returns: 0 on success, -1 on error 498 :Returns: 0 on success, -1 on error
499 499
500 Writes special registers into the vcpu. See K 500 Writes special registers into the vcpu. See KVM_GET_SREGS for the
501 data structures. 501 data structures.
502 502
503 503
504 4.15 KVM_TRANSLATE 504 4.15 KVM_TRANSLATE
505 ------------------ 505 ------------------
506 506
507 :Capability: basic 507 :Capability: basic
508 :Architectures: x86 508 :Architectures: x86
509 :Type: vcpu ioctl 509 :Type: vcpu ioctl
510 :Parameters: struct kvm_translation (in/out) 510 :Parameters: struct kvm_translation (in/out)
511 :Returns: 0 on success, -1 on error 511 :Returns: 0 on success, -1 on error
512 512
513 Translates a virtual address according to the 513 Translates a virtual address according to the vcpu's current address
514 translation mode. 514 translation mode.
515 515
516 :: 516 ::
517 517
518 struct kvm_translation { 518 struct kvm_translation {
519 /* in */ 519 /* in */
520 __u64 linear_address; 520 __u64 linear_address;
521 521
522 /* out */ 522 /* out */
523 __u64 physical_address; 523 __u64 physical_address;
524 __u8 valid; 524 __u8 valid;
525 __u8 writeable; 525 __u8 writeable;
526 __u8 usermode; 526 __u8 usermode;
527 __u8 pad[5]; 527 __u8 pad[5];
528 }; 528 };
529 529
530 530
531 4.16 KVM_INTERRUPT 531 4.16 KVM_INTERRUPT
532 ------------------ 532 ------------------
533 533
534 :Capability: basic 534 :Capability: basic
535 :Architectures: x86, ppc, mips, riscv, loongar !! 535 :Architectures: x86, ppc, mips
536 :Type: vcpu ioctl 536 :Type: vcpu ioctl
537 :Parameters: struct kvm_interrupt (in) 537 :Parameters: struct kvm_interrupt (in)
538 :Returns: 0 on success, negative on failure. 538 :Returns: 0 on success, negative on failure.
539 539
540 Queues a hardware interrupt vector to be injec 540 Queues a hardware interrupt vector to be injected.
541 541
542 :: 542 ::
543 543
544 /* for KVM_INTERRUPT */ 544 /* for KVM_INTERRUPT */
545 struct kvm_interrupt { 545 struct kvm_interrupt {
546 /* in */ 546 /* in */
547 __u32 irq; 547 __u32 irq;
548 }; 548 };
549 549
550 X86: 550 X86:
551 ^^^^ 551 ^^^^
552 552
553 :Returns: 553 :Returns:
554 554
555 ========= ============================ 555 ========= ===================================
556 0 on success, 556 0 on success,
557 -EEXIST if an interrupt is already e 557 -EEXIST if an interrupt is already enqueued
558 -EINVAL the irq number is invalid 558 -EINVAL the irq number is invalid
559 -ENXIO if the PIC is in the kernel 559 -ENXIO if the PIC is in the kernel
560 -EFAULT if the pointer is invalid 560 -EFAULT if the pointer is invalid
561 ========= ============================ 561 ========= ===================================
562 562
563 Note 'irq' is an interrupt vector, not an inte 563 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
564 ioctl is useful if the in-kernel PIC is not us 564 ioctl is useful if the in-kernel PIC is not used.
565 565
566 PPC: 566 PPC:
567 ^^^^ 567 ^^^^
568 568
569 Queues an external interrupt to be injected. T !! 569 Queues an external interrupt to be injected. This ioctl is overleaded
570 with 3 different irq values: 570 with 3 different irq values:
571 571
572 a) KVM_INTERRUPT_SET 572 a) KVM_INTERRUPT_SET
573 573
574 This injects an edge type external interrup 574 This injects an edge type external interrupt into the guest once it's ready
575 to receive interrupts. When injected, the i 575 to receive interrupts. When injected, the interrupt is done.
576 576
577 b) KVM_INTERRUPT_UNSET 577 b) KVM_INTERRUPT_UNSET
578 578
579 This unsets any pending interrupt. 579 This unsets any pending interrupt.
580 580
581 Only available with KVM_CAP_PPC_UNSET_IRQ. 581 Only available with KVM_CAP_PPC_UNSET_IRQ.
582 582
583 c) KVM_INTERRUPT_SET_LEVEL 583 c) KVM_INTERRUPT_SET_LEVEL
584 584
585 This injects a level type external interrup 585 This injects a level type external interrupt into the guest context. The
586 interrupt stays pending until a specific io 586 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
587 is triggered. 587 is triggered.
588 588
589 Only available with KVM_CAP_PPC_IRQ_LEVEL. 589 Only available with KVM_CAP_PPC_IRQ_LEVEL.
590 590
591 Note that any value for 'irq' other than the o 591 Note that any value for 'irq' other than the ones stated above is invalid
592 and incurs unexpected behavior. 592 and incurs unexpected behavior.
593 593
594 This is an asynchronous vcpu ioctl and can be 594 This is an asynchronous vcpu ioctl and can be invoked from any thread.
595 595
596 MIPS: 596 MIPS:
597 ^^^^^ 597 ^^^^^
598 598
599 Queues an external interrupt to be injected in 599 Queues an external interrupt to be injected into the virtual CPU. A negative
600 interrupt number dequeues the interrupt. 600 interrupt number dequeues the interrupt.
601 601
602 This is an asynchronous vcpu ioctl and can be 602 This is an asynchronous vcpu ioctl and can be invoked from any thread.
603 603
604 RISC-V: <<
605 ^^^^^^^ <<
606 604
607 Queues an external interrupt to be injected in !! 605 4.17 KVM_DEBUG_GUEST
608 is overloaded with 2 different irq values: !! 606 --------------------
609 <<
610 a) KVM_INTERRUPT_SET <<
611 <<
612 This sets external interrupt for a virtual <<
613 once it is ready. <<
614 <<
615 b) KVM_INTERRUPT_UNSET <<
616 <<
617 This clears pending external interrupt for <<
618 <<
619 This is an asynchronous vcpu ioctl and can be <<
620 <<
621 LOONGARCH: <<
622 ^^^^^^^^^^ <<
623 607
624 Queues an external interrupt to be injected in !! 608 :Capability: basic
625 interrupt number dequeues the interrupt. !! 609 :Architectures: none
>> 610 :Type: vcpu ioctl
>> 611 :Parameters: none)
>> 612 :Returns: -1 on error
626 613
627 This is an asynchronous vcpu ioctl and can be !! 614 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
628 615
629 616
630 4.18 KVM_GET_MSRS 617 4.18 KVM_GET_MSRS
631 ----------------- 618 -----------------
632 619
633 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEA 620 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
634 :Architectures: x86 621 :Architectures: x86
635 :Type: system ioctl, vcpu ioctl 622 :Type: system ioctl, vcpu ioctl
636 :Parameters: struct kvm_msrs (in/out) 623 :Parameters: struct kvm_msrs (in/out)
637 :Returns: number of msrs successfully returned 624 :Returns: number of msrs successfully returned;
638 -1 on error 625 -1 on error
639 626
640 When used as a system ioctl: 627 When used as a system ioctl:
641 Reads the values of MSR-based features that ar 628 Reads the values of MSR-based features that are available for the VM. This
642 is similar to KVM_GET_SUPPORTED_CPUID, but it 629 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
643 The list of msr-based features can be obtained 630 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
644 in a system ioctl. 631 in a system ioctl.
645 632
646 When used as a vcpu ioctl: 633 When used as a vcpu ioctl:
647 Reads model-specific registers from the vcpu. 634 Reads model-specific registers from the vcpu. Supported msr indices can
648 be obtained using KVM_GET_MSR_INDEX_LIST in a 635 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
649 636
650 :: 637 ::
651 638
652 struct kvm_msrs { 639 struct kvm_msrs {
653 __u32 nmsrs; /* number of msrs in entr 640 __u32 nmsrs; /* number of msrs in entries */
654 __u32 pad; 641 __u32 pad;
655 642
656 struct kvm_msr_entry entries[0]; 643 struct kvm_msr_entry entries[0];
657 }; 644 };
658 645
659 struct kvm_msr_entry { 646 struct kvm_msr_entry {
660 __u32 index; 647 __u32 index;
661 __u32 reserved; 648 __u32 reserved;
662 __u64 data; 649 __u64 data;
663 }; 650 };
664 651
665 Application code should set the 'nmsrs' member 652 Application code should set the 'nmsrs' member (which indicates the
666 size of the entries array) and the 'index' mem 653 size of the entries array) and the 'index' member of each array entry.
667 kvm will fill in the 'data' member. 654 kvm will fill in the 'data' member.
668 655
669 656
670 4.19 KVM_SET_MSRS 657 4.19 KVM_SET_MSRS
671 ----------------- 658 -----------------
672 659
673 :Capability: basic 660 :Capability: basic
674 :Architectures: x86 661 :Architectures: x86
675 :Type: vcpu ioctl 662 :Type: vcpu ioctl
676 :Parameters: struct kvm_msrs (in) 663 :Parameters: struct kvm_msrs (in)
677 :Returns: number of msrs successfully set (see 664 :Returns: number of msrs successfully set (see below), -1 on error
678 665
679 Writes model-specific registers to the vcpu. 666 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
680 data structures. 667 data structures.
681 668
682 Application code should set the 'nmsrs' member 669 Application code should set the 'nmsrs' member (which indicates the
683 size of the entries array), and the 'index' an 670 size of the entries array), and the 'index' and 'data' members of each
684 array entry. 671 array entry.
685 672
686 It tries to set the MSRs in array entries[] on 673 It tries to set the MSRs in array entries[] one by one. If setting an MSR
687 fails, e.g., due to setting reserved bits, the 674 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
688 by KVM, etc..., it stops processing the MSR li 675 by KVM, etc..., it stops processing the MSR list and returns the number of
689 MSRs that have been set successfully. 676 MSRs that have been set successfully.
690 677
691 678
692 4.20 KVM_SET_CPUID 679 4.20 KVM_SET_CPUID
693 ------------------ 680 ------------------
694 681
695 :Capability: basic 682 :Capability: basic
696 :Architectures: x86 683 :Architectures: x86
697 :Type: vcpu ioctl 684 :Type: vcpu ioctl
698 :Parameters: struct kvm_cpuid (in) 685 :Parameters: struct kvm_cpuid (in)
699 :Returns: 0 on success, -1 on error 686 :Returns: 0 on success, -1 on error
700 687
701 Defines the vcpu responses to the cpuid instru 688 Defines the vcpu responses to the cpuid instruction. Applications
702 should use the KVM_SET_CPUID2 ioctl if availab 689 should use the KVM_SET_CPUID2 ioctl if available.
703 690
704 Caveat emptor: !! 691 Note, when this IOCTL fails, KVM gives no guarantees that previous valid CPUID
705 - If this IOCTL fails, KVM gives no guarante !! 692 configuration (if there is) is not corrupted. Userspace can get a copy of the
706 configuration (if there is) is not corrupt !! 693 resulting CPUID configuration through KVM_GET_CPUID2 in case.
707 of the resulting CPUID configuration throu <<
708 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e <<
709 after running the guest, may cause guest i <<
710 - Using heterogeneous CPUID configurations, <<
711 may cause guest instability. <<
712 694
713 :: 695 ::
714 696
715 struct kvm_cpuid_entry { 697 struct kvm_cpuid_entry {
716 __u32 function; 698 __u32 function;
717 __u32 eax; 699 __u32 eax;
718 __u32 ebx; 700 __u32 ebx;
719 __u32 ecx; 701 __u32 ecx;
720 __u32 edx; 702 __u32 edx;
721 __u32 padding; 703 __u32 padding;
722 }; 704 };
723 705
724 /* for KVM_SET_CPUID */ 706 /* for KVM_SET_CPUID */
725 struct kvm_cpuid { 707 struct kvm_cpuid {
726 __u32 nent; 708 __u32 nent;
727 __u32 padding; 709 __u32 padding;
728 struct kvm_cpuid_entry entries[0]; 710 struct kvm_cpuid_entry entries[0];
729 }; 711 };
730 712
731 713
732 4.21 KVM_SET_SIGNAL_MASK 714 4.21 KVM_SET_SIGNAL_MASK
733 ------------------------ 715 ------------------------
734 716
735 :Capability: basic 717 :Capability: basic
736 :Architectures: all 718 :Architectures: all
737 :Type: vcpu ioctl 719 :Type: vcpu ioctl
738 :Parameters: struct kvm_signal_mask (in) 720 :Parameters: struct kvm_signal_mask (in)
739 :Returns: 0 on success, -1 on error 721 :Returns: 0 on success, -1 on error
740 722
741 Defines which signals are blocked during execu 723 Defines which signals are blocked during execution of KVM_RUN. This
742 signal mask temporarily overrides the threads 724 signal mask temporarily overrides the threads signal mask. Any
743 unblocked signal received (except SIGKILL and 725 unblocked signal received (except SIGKILL and SIGSTOP, which retain
744 their traditional behaviour) will cause KVM_RU 726 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
745 727
746 Note the signal will only be delivered if not 728 Note the signal will only be delivered if not blocked by the original
747 signal mask. 729 signal mask.
748 730
749 :: 731 ::
750 732
751 /* for KVM_SET_SIGNAL_MASK */ 733 /* for KVM_SET_SIGNAL_MASK */
752 struct kvm_signal_mask { 734 struct kvm_signal_mask {
753 __u32 len; 735 __u32 len;
754 __u8 sigset[0]; 736 __u8 sigset[0];
755 }; 737 };
756 738
757 739
758 4.22 KVM_GET_FPU 740 4.22 KVM_GET_FPU
759 ---------------- 741 ----------------
760 742
761 :Capability: basic 743 :Capability: basic
762 :Architectures: x86, loongarch !! 744 :Architectures: x86
763 :Type: vcpu ioctl 745 :Type: vcpu ioctl
764 :Parameters: struct kvm_fpu (out) 746 :Parameters: struct kvm_fpu (out)
765 :Returns: 0 on success, -1 on error 747 :Returns: 0 on success, -1 on error
766 748
767 Reads the floating point state from the vcpu. 749 Reads the floating point state from the vcpu.
768 750
769 :: 751 ::
770 752
771 /* x86: for KVM_GET_FPU and KVM_SET_FPU */ !! 753 /* for KVM_GET_FPU and KVM_SET_FPU */
772 struct kvm_fpu { 754 struct kvm_fpu {
773 __u8 fpr[8][16]; 755 __u8 fpr[8][16];
774 __u16 fcw; 756 __u16 fcw;
775 __u16 fsw; 757 __u16 fsw;
776 __u8 ftwx; /* in fxsave format */ 758 __u8 ftwx; /* in fxsave format */
777 __u8 pad1; 759 __u8 pad1;
778 __u16 last_opcode; 760 __u16 last_opcode;
779 __u64 last_ip; 761 __u64 last_ip;
780 __u64 last_dp; 762 __u64 last_dp;
781 __u8 xmm[16][16]; 763 __u8 xmm[16][16];
782 __u32 mxcsr; 764 __u32 mxcsr;
783 __u32 pad2; 765 __u32 pad2;
784 }; 766 };
785 767
786 /* LoongArch: for KVM_GET_FPU and KVM_SET_FP <<
787 struct kvm_fpu { <<
788 __u32 fcsr; <<
789 __u64 fcc; <<
790 struct kvm_fpureg { <<
791 __u64 val64[4]; <<
792 }fpr[32]; <<
793 }; <<
794 <<
795 768
796 4.23 KVM_SET_FPU 769 4.23 KVM_SET_FPU
797 ---------------- 770 ----------------
798 771
799 :Capability: basic 772 :Capability: basic
800 :Architectures: x86, loongarch !! 773 :Architectures: x86
801 :Type: vcpu ioctl 774 :Type: vcpu ioctl
802 :Parameters: struct kvm_fpu (in) 775 :Parameters: struct kvm_fpu (in)
803 :Returns: 0 on success, -1 on error 776 :Returns: 0 on success, -1 on error
804 777
805 Writes the floating point state to the vcpu. 778 Writes the floating point state to the vcpu.
806 779
807 :: 780 ::
808 781
809 /* x86: for KVM_GET_FPU and KVM_SET_FPU */ !! 782 /* for KVM_GET_FPU and KVM_SET_FPU */
810 struct kvm_fpu { 783 struct kvm_fpu {
811 __u8 fpr[8][16]; 784 __u8 fpr[8][16];
812 __u16 fcw; 785 __u16 fcw;
813 __u16 fsw; 786 __u16 fsw;
814 __u8 ftwx; /* in fxsave format */ 787 __u8 ftwx; /* in fxsave format */
815 __u8 pad1; 788 __u8 pad1;
816 __u16 last_opcode; 789 __u16 last_opcode;
817 __u64 last_ip; 790 __u64 last_ip;
818 __u64 last_dp; 791 __u64 last_dp;
819 __u8 xmm[16][16]; 792 __u8 xmm[16][16];
820 __u32 mxcsr; 793 __u32 mxcsr;
821 __u32 pad2; 794 __u32 pad2;
822 }; 795 };
823 796
824 /* LoongArch: for KVM_GET_FPU and KVM_SET_FP <<
825 struct kvm_fpu { <<
826 __u32 fcsr; <<
827 __u64 fcc; <<
828 struct kvm_fpureg { <<
829 __u64 val64[4]; <<
830 }fpr[32]; <<
831 }; <<
832 <<
833 797
834 4.24 KVM_CREATE_IRQCHIP 798 4.24 KVM_CREATE_IRQCHIP
835 ----------------------- 799 -----------------------
836 800
837 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQ 801 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
838 :Architectures: x86, arm64, s390 !! 802 :Architectures: x86, ARM, arm64, s390
839 :Type: vm ioctl 803 :Type: vm ioctl
840 :Parameters: none 804 :Parameters: none
841 :Returns: 0 on success, -1 on error 805 :Returns: 0 on success, -1 on error
842 806
843 Creates an interrupt controller model in the k 807 Creates an interrupt controller model in the kernel.
844 On x86, creates a virtual ioapic, a virtual PI 808 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
845 future vcpus to have a local APIC. IRQ routin 809 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
846 PIC and IOAPIC; GSI 16-23 only go to the IOAPI 810 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
847 On arm64, a GICv2 is created. Any other GIC ve !! 811 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
848 KVM_CREATE_DEVICE, which also supports creatin 812 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
849 KVM_CREATE_DEVICE is preferred over KVM_CREATE 813 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
850 On s390, a dummy irq routing table is created. 814 On s390, a dummy irq routing table is created.
851 815
852 Note that on s390 the KVM_CAP_S390_IRQCHIP vm 816 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
853 before KVM_CREATE_IRQCHIP can be used. 817 before KVM_CREATE_IRQCHIP can be used.
854 818
855 819
856 4.25 KVM_IRQ_LINE 820 4.25 KVM_IRQ_LINE
857 ----------------- 821 -----------------
858 822
859 :Capability: KVM_CAP_IRQCHIP 823 :Capability: KVM_CAP_IRQCHIP
860 :Architectures: x86, arm64 !! 824 :Architectures: x86, arm, arm64
861 :Type: vm ioctl 825 :Type: vm ioctl
862 :Parameters: struct kvm_irq_level 826 :Parameters: struct kvm_irq_level
863 :Returns: 0 on success, -1 on error 827 :Returns: 0 on success, -1 on error
864 828
865 Sets the level of a GSI input to the interrupt 829 Sets the level of a GSI input to the interrupt controller model in the kernel.
866 On some architectures it is required that an i 830 On some architectures it is required that an interrupt controller model has
867 been previously created with KVM_CREATE_IRQCHI 831 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
868 interrupts require the level to be set to 1 an 832 interrupts require the level to be set to 1 and then back to 0.
869 833
870 On real hardware, interrupt pins can be active 834 On real hardware, interrupt pins can be active-low or active-high. This
871 does not matter for the level field of struct 835 does not matter for the level field of struct kvm_irq_level: 1 always
872 means active (asserted), 0 means inactive (dea 836 means active (asserted), 0 means inactive (deasserted).
873 837
874 x86 allows the operating system to program the 838 x86 allows the operating system to program the interrupt polarity
875 (active-low/active-high) for level-triggered i 839 (active-low/active-high) for level-triggered interrupts, and KVM used
876 to consider the polarity. However, due to bit 840 to consider the polarity. However, due to bitrot in the handling of
877 active-low interrupts, the above convention is 841 active-low interrupts, the above convention is now valid on x86 too.
878 This is signaled by KVM_CAP_X86_IOAPIC_POLARIT 842 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
879 should not present interrupts to the guest as 843 should not present interrupts to the guest as active-low unless this
880 capability is present (or unless it is not usi 844 capability is present (or unless it is not using the in-kernel irqchip,
881 of course). 845 of course).
882 846
883 847
884 arm64 can signal an interrupt either at the CP !! 848 ARM/arm64 can signal an interrupt either at the CPU level, or at the
885 in-kernel irqchip (GIC), and for in-kernel irq 849 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
886 use PPIs designated for specific cpus. The ir 850 use PPIs designated for specific cpus. The irq field is interpreted
887 like this:: 851 like this::
888 852
889 bits: | 31 ... 28 | 27 ... 24 | 23 ... 1 !! 853 Â bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
890 field: | vcpu2_index | irq_type | vcpu_inde 854 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
891 855
892 The irq_type field has the following values: 856 The irq_type field has the following values:
893 857
894 - KVM_ARM_IRQ_TYPE_CPU: !! 858 - irq_type[0]:
895 out-of-kernel GIC: irq_id 0 is 859 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
896 - KVM_ARM_IRQ_TYPE_SPI: !! 860 - irq_type[1]:
897 in-kernel GIC: SPI, irq_id betw 861 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
898 (the vcpu_index field is ignore 862 (the vcpu_index field is ignored)
899 - KVM_ARM_IRQ_TYPE_PPI: !! 863 - irq_type[2]:
900 in-kernel GIC: PPI, irq_id betw 864 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
901 865
902 (The irq_id field thus corresponds nicely to t 866 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
903 867
904 In both cases, level is used to assert/deasser 868 In both cases, level is used to assert/deassert the line.
905 869
906 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supporte 870 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
907 identified as (256 * vcpu2_index + vcpu_index) 871 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
908 must be zero. 872 must be zero.
909 873
910 Note that on arm64, the KVM_CAP_IRQCHIP capabi !! 874 Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
911 injection of interrupts for the in-kernel irqc 875 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
912 be used for a userspace interrupt controller. 876 be used for a userspace interrupt controller.
913 877
914 :: 878 ::
915 879
916 struct kvm_irq_level { 880 struct kvm_irq_level {
917 union { 881 union {
918 __u32 irq; /* GSI */ 882 __u32 irq; /* GSI */
919 __s32 status; /* not used for 883 __s32 status; /* not used for KVM_IRQ_LEVEL */
920 }; 884 };
921 __u32 level; /* 0 or 1 */ 885 __u32 level; /* 0 or 1 */
922 }; 886 };
923 887
924 888
925 4.26 KVM_GET_IRQCHIP 889 4.26 KVM_GET_IRQCHIP
926 -------------------- 890 --------------------
927 891
928 :Capability: KVM_CAP_IRQCHIP 892 :Capability: KVM_CAP_IRQCHIP
929 :Architectures: x86 893 :Architectures: x86
930 :Type: vm ioctl 894 :Type: vm ioctl
931 :Parameters: struct kvm_irqchip (in/out) 895 :Parameters: struct kvm_irqchip (in/out)
932 :Returns: 0 on success, -1 on error 896 :Returns: 0 on success, -1 on error
933 897
934 Reads the state of a kernel interrupt controll 898 Reads the state of a kernel interrupt controller created with
935 KVM_CREATE_IRQCHIP into a buffer provided by t 899 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
936 900
937 :: 901 ::
938 902
939 struct kvm_irqchip { 903 struct kvm_irqchip {
940 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 904 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
941 __u32 pad; 905 __u32 pad;
942 union { 906 union {
943 char dummy[512]; /* reserving 907 char dummy[512]; /* reserving space */
944 struct kvm_pic_state pic; 908 struct kvm_pic_state pic;
945 struct kvm_ioapic_state ioapic 909 struct kvm_ioapic_state ioapic;
946 } chip; 910 } chip;
947 }; 911 };
948 912
949 913
950 4.27 KVM_SET_IRQCHIP 914 4.27 KVM_SET_IRQCHIP
951 -------------------- 915 --------------------
952 916
953 :Capability: KVM_CAP_IRQCHIP 917 :Capability: KVM_CAP_IRQCHIP
954 :Architectures: x86 918 :Architectures: x86
955 :Type: vm ioctl 919 :Type: vm ioctl
956 :Parameters: struct kvm_irqchip (in) 920 :Parameters: struct kvm_irqchip (in)
957 :Returns: 0 on success, -1 on error 921 :Returns: 0 on success, -1 on error
958 922
959 Sets the state of a kernel interrupt controlle 923 Sets the state of a kernel interrupt controller created with
960 KVM_CREATE_IRQCHIP from a buffer provided by t 924 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
961 925
962 :: 926 ::
963 927
964 struct kvm_irqchip { 928 struct kvm_irqchip {
965 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 929 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
966 __u32 pad; 930 __u32 pad;
967 union { 931 union {
968 char dummy[512]; /* reserving 932 char dummy[512]; /* reserving space */
969 struct kvm_pic_state pic; 933 struct kvm_pic_state pic;
970 struct kvm_ioapic_state ioapic 934 struct kvm_ioapic_state ioapic;
971 } chip; 935 } chip;
972 }; 936 };
973 937
974 938
975 4.28 KVM_XEN_HVM_CONFIG 939 4.28 KVM_XEN_HVM_CONFIG
976 ----------------------- 940 -----------------------
977 941
978 :Capability: KVM_CAP_XEN_HVM 942 :Capability: KVM_CAP_XEN_HVM
979 :Architectures: x86 943 :Architectures: x86
980 :Type: vm ioctl 944 :Type: vm ioctl
981 :Parameters: struct kvm_xen_hvm_config (in) 945 :Parameters: struct kvm_xen_hvm_config (in)
982 :Returns: 0 on success, -1 on error 946 :Returns: 0 on success, -1 on error
983 947
984 Sets the MSR that the Xen HVM guest uses to in 948 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
985 page, and provides the starting address and si 949 page, and provides the starting address and size of the hypercall
986 blobs in userspace. When the guest writes the 950 blobs in userspace. When the guest writes the MSR, kvm copies one
987 page of a blob (32- or 64-bit, depending on th 951 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
988 memory. 952 memory.
989 953
990 :: 954 ::
991 955
992 struct kvm_xen_hvm_config { 956 struct kvm_xen_hvm_config {
993 __u32 flags; 957 __u32 flags;
994 __u32 msr; 958 __u32 msr;
995 __u64 blob_addr_32; 959 __u64 blob_addr_32;
996 __u64 blob_addr_64; 960 __u64 blob_addr_64;
997 __u8 blob_size_32; 961 __u8 blob_size_32;
998 __u8 blob_size_64; 962 __u8 blob_size_64;
999 __u8 pad2[30]; 963 __u8 pad2[30];
1000 }; 964 };
1001 965
1002 If certain flags are returned from the KVM_CA <<
1003 be set in the flags field of this ioctl: <<
1004 <<
1005 The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag r <<
1006 the contents of the hypercall page automatica <<
1007 intercepted and passed to userspace through K <<
1008 case, all of the blob size and address fields <<
1009 <<
1010 The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indic <<
1011 will always use the KVM_XEN_HVM_EVTCHN_SEND i <<
1012 channel interrupts rather than manipulating t <<
1013 structures directly. This, in turn, may allow <<
1014 such as intercepting the SCHEDOP_poll hyperca <<
1015 spinlock operation for the guest. Userspace m <<
1016 to deliver events if it was advertised, even <<
1017 send this indication that it will always do s <<
1018 <<
1019 No other flags are currently valid in the str <<
1020 966
1021 4.29 KVM_GET_CLOCK 967 4.29 KVM_GET_CLOCK
1022 ------------------ 968 ------------------
1023 969
1024 :Capability: KVM_CAP_ADJUST_CLOCK 970 :Capability: KVM_CAP_ADJUST_CLOCK
1025 :Architectures: x86 971 :Architectures: x86
1026 :Type: vm ioctl 972 :Type: vm ioctl
1027 :Parameters: struct kvm_clock_data (out) 973 :Parameters: struct kvm_clock_data (out)
1028 :Returns: 0 on success, -1 on error 974 :Returns: 0 on success, -1 on error
1029 975
1030 Gets the current timestamp of kvmclock as see 976 Gets the current timestamp of kvmclock as seen by the current guest. In
1031 conjunction with KVM_SET_CLOCK, it is used to 977 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
1032 such as migration. 978 such as migration.
1033 979
1034 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CH 980 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
1035 set of bits that KVM can return in struct kvm 981 set of bits that KVM can return in struct kvm_clock_data's flag member.
1036 982
1037 The following flags are defined: !! 983 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
1038 !! 984 value is the exact kvmclock value seen by all VCPUs at the instant
1039 KVM_CLOCK_TSC_STABLE !! 985 when KVM_GET_CLOCK was called. If clear, the returned value is simply
1040 If set, the returned value is the exact kvm !! 986 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
1041 value seen by all VCPUs at the instant when !! 987 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
1042 If clear, the returned value is simply CLOC !! 988 but the exact value read by each VCPU could differ, because the host
1043 offset; the offset can be modified with KVM !! 989 TSC is not stable.
1044 to make all VCPUs follow this clock, but th <<
1045 VCPU could differ, because the host TSC is <<
1046 <<
1047 KVM_CLOCK_REALTIME <<
1048 If set, the `realtime` field in the kvm_clo <<
1049 structure is populated with the value of th <<
1050 clocksource at the instant when KVM_GET_CLO <<
1051 the `realtime` field does not contain a val <<
1052 <<
1053 KVM_CLOCK_HOST_TSC <<
1054 If set, the `host_tsc` field in the kvm_clo <<
1055 structure is populated with the value of th <<
1056 at the instant when KVM_GET_CLOCK was calle <<
1057 does not contain a value. <<
1058 990
1059 :: 991 ::
1060 992
1061 struct kvm_clock_data { 993 struct kvm_clock_data {
1062 __u64 clock; /* kvmclock current val 994 __u64 clock; /* kvmclock current value */
1063 __u32 flags; 995 __u32 flags;
1064 __u32 pad0; !! 996 __u32 pad[9];
1065 __u64 realtime; <<
1066 __u64 host_tsc; <<
1067 __u32 pad[4]; <<
1068 }; 997 };
1069 998
1070 999
1071 4.30 KVM_SET_CLOCK 1000 4.30 KVM_SET_CLOCK
1072 ------------------ 1001 ------------------
1073 1002
1074 :Capability: KVM_CAP_ADJUST_CLOCK 1003 :Capability: KVM_CAP_ADJUST_CLOCK
1075 :Architectures: x86 1004 :Architectures: x86
1076 :Type: vm ioctl 1005 :Type: vm ioctl
1077 :Parameters: struct kvm_clock_data (in) 1006 :Parameters: struct kvm_clock_data (in)
1078 :Returns: 0 on success, -1 on error 1007 :Returns: 0 on success, -1 on error
1079 1008
1080 Sets the current timestamp of kvmclock to the 1009 Sets the current timestamp of kvmclock to the value specified in its parameter.
1081 In conjunction with KVM_GET_CLOCK, it is used 1010 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1082 such as migration. 1011 such as migration.
1083 1012
1084 The following flags can be passed: <<
1085 <<
1086 KVM_CLOCK_REALTIME <<
1087 If set, KVM will compare the value of the ` <<
1088 with the value of the host's real time cloc <<
1089 KVM_SET_CLOCK was called. The difference in <<
1090 kvmclock value that will be provided to gue <<
1091 <<
1092 Other flags returned by ``KVM_GET_CLOCK`` are <<
1093 <<
1094 :: 1013 ::
1095 1014
1096 struct kvm_clock_data { 1015 struct kvm_clock_data {
1097 __u64 clock; /* kvmclock current val 1016 __u64 clock; /* kvmclock current value */
1098 __u32 flags; 1017 __u32 flags;
1099 __u32 pad0; !! 1018 __u32 pad[9];
1100 __u64 realtime; <<
1101 __u64 host_tsc; <<
1102 __u32 pad[4]; <<
1103 }; 1019 };
1104 1020
1105 1021
1106 4.31 KVM_GET_VCPU_EVENTS 1022 4.31 KVM_GET_VCPU_EVENTS
1107 ------------------------ 1023 ------------------------
1108 1024
1109 :Capability: KVM_CAP_VCPU_EVENTS 1025 :Capability: KVM_CAP_VCPU_EVENTS
1110 :Extended by: KVM_CAP_INTR_SHADOW 1026 :Extended by: KVM_CAP_INTR_SHADOW
1111 :Architectures: x86, arm64 !! 1027 :Architectures: x86, arm, arm64
1112 :Type: vcpu ioctl 1028 :Type: vcpu ioctl
1113 :Parameters: struct kvm_vcpu_events (out) !! 1029 :Parameters: struct kvm_vcpu_event (out)
1114 :Returns: 0 on success, -1 on error 1030 :Returns: 0 on success, -1 on error
1115 1031
1116 X86: 1032 X86:
1117 ^^^^ 1033 ^^^^
1118 1034
1119 Gets currently pending exceptions, interrupts 1035 Gets currently pending exceptions, interrupts, and NMIs as well as related
1120 states of the vcpu. 1036 states of the vcpu.
1121 1037
1122 :: 1038 ::
1123 1039
1124 struct kvm_vcpu_events { 1040 struct kvm_vcpu_events {
1125 struct { 1041 struct {
1126 __u8 injected; 1042 __u8 injected;
1127 __u8 nr; 1043 __u8 nr;
1128 __u8 has_error_code; 1044 __u8 has_error_code;
1129 __u8 pending; 1045 __u8 pending;
1130 __u32 error_code; 1046 __u32 error_code;
1131 } exception; 1047 } exception;
1132 struct { 1048 struct {
1133 __u8 injected; 1049 __u8 injected;
1134 __u8 nr; 1050 __u8 nr;
1135 __u8 soft; 1051 __u8 soft;
1136 __u8 shadow; 1052 __u8 shadow;
1137 } interrupt; 1053 } interrupt;
1138 struct { 1054 struct {
1139 __u8 injected; 1055 __u8 injected;
1140 __u8 pending; 1056 __u8 pending;
1141 __u8 masked; 1057 __u8 masked;
1142 __u8 pad; 1058 __u8 pad;
1143 } nmi; 1059 } nmi;
1144 __u32 sipi_vector; 1060 __u32 sipi_vector;
1145 __u32 flags; 1061 __u32 flags;
1146 struct { 1062 struct {
1147 __u8 smm; 1063 __u8 smm;
1148 __u8 pending; 1064 __u8 pending;
1149 __u8 smm_inside_nmi; 1065 __u8 smm_inside_nmi;
1150 __u8 latched_init; 1066 __u8 latched_init;
1151 } smi; 1067 } smi;
1152 __u8 reserved[27]; 1068 __u8 reserved[27];
1153 __u8 exception_has_payload; 1069 __u8 exception_has_payload;
1154 __u64 exception_payload; 1070 __u64 exception_payload;
1155 }; 1071 };
1156 1072
1157 The following bits are defined in the flags f 1073 The following bits are defined in the flags field:
1158 1074
1159 - KVM_VCPUEVENT_VALID_SHADOW may be set to si 1075 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1160 interrupt.shadow contains a valid state. 1076 interrupt.shadow contains a valid state.
1161 1077
1162 - KVM_VCPUEVENT_VALID_SMM may be set to signa 1078 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1163 valid state. 1079 valid state.
1164 1080
1165 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to s 1081 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1166 exception_has_payload, exception_payload, a 1082 exception_has_payload, exception_payload, and exception.pending
1167 fields contain a valid state. This bit will 1083 fields contain a valid state. This bit will be set whenever
1168 KVM_CAP_EXCEPTION_PAYLOAD is enabled. 1084 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1169 1085
1170 - KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set !! 1086 ARM/ARM64:
1171 triple_fault_pending field contains a valid !! 1087 ^^^^^^^^^^
1172 be set whenever KVM_CAP_X86_TRIPLE_FAULT_EV <<
1173 <<
1174 ARM64: <<
1175 ^^^^^^ <<
1176 1088
1177 If the guest accesses a device that is being 1089 If the guest accesses a device that is being emulated by the host kernel in
1178 such a way that a real device would generate 1090 such a way that a real device would generate a physical SError, KVM may make
1179 a virtual SError pending for that VCPU. This 1091 a virtual SError pending for that VCPU. This system error interrupt remains
1180 pending until the guest takes the exception b 1092 pending until the guest takes the exception by unmasking PSTATE.A.
1181 1093
1182 Running the VCPU may cause it to take a pendi 1094 Running the VCPU may cause it to take a pending SError, or make an access that
1183 causes an SError to become pending. The event 1095 causes an SError to become pending. The event's description is only valid while
1184 the VPCU is not running. 1096 the VPCU is not running.
1185 1097
1186 This API provides a way to read and write the 1098 This API provides a way to read and write the pending 'event' state that is not
1187 visible to the guest. To save, restore or mig 1099 visible to the guest. To save, restore or migrate a VCPU the struct representing
1188 the state can be read then written using this 1100 the state can be read then written using this GET/SET API, along with the other
1189 guest-visible registers. It is not possible t 1101 guest-visible registers. It is not possible to 'cancel' an SError that has been
1190 made pending. 1102 made pending.
1191 1103
1192 A device being emulated in user-space may als 1104 A device being emulated in user-space may also wish to generate an SError. To do
1193 this the events structure can be populated by 1105 this the events structure can be populated by user-space. The current state
1194 should be read first, to ensure no existing S 1106 should be read first, to ensure no existing SError is pending. If an existing
1195 SError is pending, the architecture's 'Multip 1107 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1196 be followed. (2.5.3 of DDI0587.a "ARM Reliabi 1108 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1197 Serviceability (RAS) Specification"). 1109 Serviceability (RAS) Specification").
1198 1110
1199 SError exceptions always have an ESR value. S 1111 SError exceptions always have an ESR value. Some CPUs have the ability to
1200 specify what the virtual SError's ESR value s 1112 specify what the virtual SError's ESR value should be. These systems will
1201 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In t 1113 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1202 always have a non-zero value when read, and t 1114 always have a non-zero value when read, and the agent making an SError pending
1203 should specify the ISS field in the lower 24 1115 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1204 the system supports KVM_CAP_ARM_INJECT_SERROR 1116 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1205 with exception.has_esr as zero, KVM will choo 1117 with exception.has_esr as zero, KVM will choose an ESR.
1206 1118
1207 Specifying exception.has_esr on a system that 1119 Specifying exception.has_esr on a system that does not support it will return
1208 -EINVAL. Setting anything other than the lowe 1120 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1209 will return -EINVAL. 1121 will return -EINVAL.
1210 1122
1211 It is not possible to read back a pending ext 1123 It is not possible to read back a pending external abort (injected via
1212 KVM_SET_VCPU_EVENTS or otherwise) because suc 1124 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1213 directly to the virtual CPU). 1125 directly to the virtual CPU).
1214 1126
1215 :: 1127 ::
1216 1128
1217 struct kvm_vcpu_events { 1129 struct kvm_vcpu_events {
1218 struct { 1130 struct {
1219 __u8 serror_pending; 1131 __u8 serror_pending;
1220 __u8 serror_has_esr; 1132 __u8 serror_has_esr;
1221 __u8 ext_dabt_pending; 1133 __u8 ext_dabt_pending;
1222 /* Align it to 8 bytes */ 1134 /* Align it to 8 bytes */
1223 __u8 pad[5]; 1135 __u8 pad[5];
1224 __u64 serror_esr; 1136 __u64 serror_esr;
1225 } exception; 1137 } exception;
1226 __u32 reserved[12]; 1138 __u32 reserved[12];
1227 }; 1139 };
1228 1140
1229 4.32 KVM_SET_VCPU_EVENTS 1141 4.32 KVM_SET_VCPU_EVENTS
1230 ------------------------ 1142 ------------------------
1231 1143
1232 :Capability: KVM_CAP_VCPU_EVENTS 1144 :Capability: KVM_CAP_VCPU_EVENTS
1233 :Extended by: KVM_CAP_INTR_SHADOW 1145 :Extended by: KVM_CAP_INTR_SHADOW
1234 :Architectures: x86, arm64 !! 1146 :Architectures: x86, arm, arm64
1235 :Type: vcpu ioctl 1147 :Type: vcpu ioctl
1236 :Parameters: struct kvm_vcpu_events (in) !! 1148 :Parameters: struct kvm_vcpu_event (in)
1237 :Returns: 0 on success, -1 on error 1149 :Returns: 0 on success, -1 on error
1238 1150
1239 X86: 1151 X86:
1240 ^^^^ 1152 ^^^^
1241 1153
1242 Set pending exceptions, interrupts, and NMIs 1154 Set pending exceptions, interrupts, and NMIs as well as related states of the
1243 vcpu. 1155 vcpu.
1244 1156
1245 See KVM_GET_VCPU_EVENTS for the data structur 1157 See KVM_GET_VCPU_EVENTS for the data structure.
1246 1158
1247 Fields that may be modified asynchronously by 1159 Fields that may be modified asynchronously by running VCPUs can be excluded
1248 from the update. These fields are nmi.pending 1160 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1249 smi.pending. Keep the corresponding bits in t 1161 smi.pending. Keep the corresponding bits in the flags field cleared to
1250 suppress overwriting the current in-kernel st 1162 suppress overwriting the current in-kernel state. The bits are:
1251 1163
1252 =============================== ============ 1164 =============================== ==================================
1253 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi 1165 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1254 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sip 1166 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1255 KVM_VCPUEVENT_VALID_SMM transfer the 1167 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1256 =============================== ============ 1168 =============================== ==================================
1257 1169
1258 If KVM_CAP_INTR_SHADOW is available, KVM_VCPU 1170 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1259 the flags field to signal that interrupt.shad 1171 the flags field to signal that interrupt.shadow contains a valid state and
1260 shall be written into the VCPU. 1172 shall be written into the VCPU.
1261 1173
1262 KVM_VCPUEVENT_VALID_SMM can only be set if KV 1174 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1263 1175
1264 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_ 1176 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1265 can be set in the flags field to signal that 1177 can be set in the flags field to signal that the
1266 exception_has_payload, exception_payload, and 1178 exception_has_payload, exception_payload, and exception.pending fields
1267 contain a valid state and shall be written in 1179 contain a valid state and shall be written into the VCPU.
1268 1180
1269 If KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled, !! 1181 ARM/ARM64:
1270 can be set in flags field to signal that the !! 1182 ^^^^^^^^^^
1271 a valid state and shall be written into the V <<
1272 <<
1273 ARM64: <<
1274 ^^^^^^ <<
1275 1183
1276 User space may need to inject several types o 1184 User space may need to inject several types of events to the guest.
1277 1185
1278 Set the pending SError exception state for th 1186 Set the pending SError exception state for this VCPU. It is not possible to
1279 'cancel' an Serror that has been made pending 1187 'cancel' an Serror that has been made pending.
1280 1188
1281 If the guest performed an access to I/O memor 1189 If the guest performed an access to I/O memory which could not be handled by
1282 userspace, for example because of missing ins 1190 userspace, for example because of missing instruction syndrome decode
1283 information or because there is no device map 1191 information or because there is no device mapped at the accessed IPA, then
1284 userspace can ask the kernel to inject an ext 1192 userspace can ask the kernel to inject an external abort using the address
1285 from the exiting fault on the VCPU. It is a p 1193 from the exiting fault on the VCPU. It is a programming error to set
1286 ext_dabt_pending after an exit which was not 1194 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1287 KVM_EXIT_ARM_NISV. This feature is only avail 1195 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1288 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper 1196 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1289 how userspace reports accesses for the above 1197 how userspace reports accesses for the above cases to guests, across different
1290 userspace implementations. Nevertheless, user 1198 userspace implementations. Nevertheless, userspace can still emulate all Arm
1291 exceptions by manipulating individual registe 1199 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1292 1200
1293 See KVM_GET_VCPU_EVENTS for the data structur 1201 See KVM_GET_VCPU_EVENTS for the data structure.
1294 1202
1295 1203
1296 4.33 KVM_GET_DEBUGREGS 1204 4.33 KVM_GET_DEBUGREGS
1297 ---------------------- 1205 ----------------------
1298 1206
1299 :Capability: KVM_CAP_DEBUGREGS 1207 :Capability: KVM_CAP_DEBUGREGS
1300 :Architectures: x86 1208 :Architectures: x86
1301 :Type: vm ioctl 1209 :Type: vm ioctl
1302 :Parameters: struct kvm_debugregs (out) 1210 :Parameters: struct kvm_debugregs (out)
1303 :Returns: 0 on success, -1 on error 1211 :Returns: 0 on success, -1 on error
1304 1212
1305 Reads debug registers from the vcpu. 1213 Reads debug registers from the vcpu.
1306 1214
1307 :: 1215 ::
1308 1216
1309 struct kvm_debugregs { 1217 struct kvm_debugregs {
1310 __u64 db[4]; 1218 __u64 db[4];
1311 __u64 dr6; 1219 __u64 dr6;
1312 __u64 dr7; 1220 __u64 dr7;
1313 __u64 flags; 1221 __u64 flags;
1314 __u64 reserved[9]; 1222 __u64 reserved[9];
1315 }; 1223 };
1316 1224
1317 1225
1318 4.34 KVM_SET_DEBUGREGS 1226 4.34 KVM_SET_DEBUGREGS
1319 ---------------------- 1227 ----------------------
1320 1228
1321 :Capability: KVM_CAP_DEBUGREGS 1229 :Capability: KVM_CAP_DEBUGREGS
1322 :Architectures: x86 1230 :Architectures: x86
1323 :Type: vm ioctl 1231 :Type: vm ioctl
1324 :Parameters: struct kvm_debugregs (in) 1232 :Parameters: struct kvm_debugregs (in)
1325 :Returns: 0 on success, -1 on error 1233 :Returns: 0 on success, -1 on error
1326 1234
1327 Writes debug registers into the vcpu. 1235 Writes debug registers into the vcpu.
1328 1236
1329 See KVM_GET_DEBUGREGS for the data structure. 1237 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1330 yet and must be cleared on entry. 1238 yet and must be cleared on entry.
1331 1239
1332 1240
1333 4.35 KVM_SET_USER_MEMORY_REGION 1241 4.35 KVM_SET_USER_MEMORY_REGION
1334 ------------------------------- 1242 -------------------------------
1335 1243
1336 :Capability: KVM_CAP_USER_MEMORY 1244 :Capability: KVM_CAP_USER_MEMORY
1337 :Architectures: all 1245 :Architectures: all
1338 :Type: vm ioctl 1246 :Type: vm ioctl
1339 :Parameters: struct kvm_userspace_memory_regi 1247 :Parameters: struct kvm_userspace_memory_region (in)
1340 :Returns: 0 on success, -1 on error 1248 :Returns: 0 on success, -1 on error
1341 1249
1342 :: 1250 ::
1343 1251
1344 struct kvm_userspace_memory_region { 1252 struct kvm_userspace_memory_region {
1345 __u32 slot; 1253 __u32 slot;
1346 __u32 flags; 1254 __u32 flags;
1347 __u64 guest_phys_addr; 1255 __u64 guest_phys_addr;
1348 __u64 memory_size; /* bytes */ 1256 __u64 memory_size; /* bytes */
1349 __u64 userspace_addr; /* start of the 1257 __u64 userspace_addr; /* start of the userspace allocated memory */
1350 }; 1258 };
1351 1259
1352 /* for kvm_userspace_memory_region::flags * !! 1260 /* for kvm_memory_region::flags */
1353 #define KVM_MEM_LOG_DIRTY_PAGES (1UL 1261 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1354 #define KVM_MEM_READONLY (1UL << 1) 1262 #define KVM_MEM_READONLY (1UL << 1)
1355 1263
1356 This ioctl allows the user to create, modify 1264 This ioctl allows the user to create, modify or delete a guest physical
1357 memory slot. Bits 0-15 of "slot" specify the 1265 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1358 should be less than the maximum number of use 1266 should be less than the maximum number of user memory slots supported per
1359 VM. The maximum allowed slots can be queried 1267 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1360 Slots may not overlap in guest physical addre 1268 Slots may not overlap in guest physical address space.
1361 1269
1362 If KVM_CAP_MULTI_ADDRESS_SPACE is available, 1270 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1363 specifies the address space which is being mo 1271 specifies the address space which is being modified. They must be
1364 less than the value that KVM_CHECK_EXTENSION 1272 less than the value that KVM_CHECK_EXTENSION returns for the
1365 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slot 1273 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1366 are unrelated; the restriction on overlapping 1274 are unrelated; the restriction on overlapping slots only applies within
1367 each address space. 1275 each address space.
1368 1276
1369 Deleting a slot is done by passing zero for m 1277 Deleting a slot is done by passing zero for memory_size. When changing
1370 an existing slot, it may be moved in the gues 1278 an existing slot, it may be moved in the guest physical memory space,
1371 or its flags may be modified, but it may not 1279 or its flags may be modified, but it may not be resized.
1372 1280
1373 Memory for the region is taken starting at th 1281 Memory for the region is taken starting at the address denoted by the
1374 field userspace_addr, which must point at use 1282 field userspace_addr, which must point at user addressable memory for
1375 the entire memory slot size. Any object may 1283 the entire memory slot size. Any object may back this memory, including
1376 anonymous memory, ordinary files, and hugetlb 1284 anonymous memory, ordinary files, and hugetlbfs.
1377 1285
1378 On architectures that support a form of addre 1286 On architectures that support a form of address tagging, userspace_addr must
1379 be an untagged address. 1287 be an untagged address.
1380 1288
1381 It is recommended that the lower 21 bits of g 1289 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1382 be identical. This allows large pages in the 1290 be identical. This allows large pages in the guest to be backed by large
1383 pages in the host. 1291 pages in the host.
1384 1292
1385 The flags field supports two flags: KVM_MEM_L 1293 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1386 KVM_MEM_READONLY. The former can be set to i 1294 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1387 writes to memory within the slot. See KVM_GE 1295 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1388 use it. The latter can be set, if KVM_CAP_RE 1296 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1389 to make a new slot read-only. In this case, 1297 to make a new slot read-only. In this case, writes to this memory will be
1390 posted to userspace as KVM_EXIT_MMIO exits. 1298 posted to userspace as KVM_EXIT_MMIO exits.
1391 1299
1392 When the KVM_CAP_SYNC_MMU capability is avail 1300 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1393 the memory region are automatically reflected 1301 the memory region are automatically reflected into the guest. For example, an
1394 mmap() that affects the region will be made v 1302 mmap() that affects the region will be made visible immediately. Another
1395 example is madvise(MADV_DROP). 1303 example is madvise(MADV_DROP).
1396 1304
1397 Note: On arm64, a write generated by the page !! 1305 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1398 the Access and Dirty flags, for example) neve !! 1306 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1399 KVM_EXIT_MMIO exit when the slot has the KVM_ !! 1307 allocation and is deprecated.
1400 is because KVM cannot provide the data that w <<
1401 page-table walker, making it impossible to em <<
1402 Instead, an abort (data abort if the cause of <<
1403 was a load or a store, instruction abort if i <<
1404 fetch) is injected in the guest. <<
1405 <<
1406 S390: <<
1407 ^^^^^ <<
1408 1308
1409 Returns -EINVAL if the VM has the KVM_VM_S390 <<
1410 Returns -EINVAL if called on a protected VM. <<
1411 1309
1412 4.36 KVM_SET_TSS_ADDR 1310 4.36 KVM_SET_TSS_ADDR
1413 --------------------- 1311 ---------------------
1414 1312
1415 :Capability: KVM_CAP_SET_TSS_ADDR 1313 :Capability: KVM_CAP_SET_TSS_ADDR
1416 :Architectures: x86 1314 :Architectures: x86
1417 :Type: vm ioctl 1315 :Type: vm ioctl
1418 :Parameters: unsigned long tss_address (in) 1316 :Parameters: unsigned long tss_address (in)
1419 :Returns: 0 on success, -1 on error 1317 :Returns: 0 on success, -1 on error
1420 1318
1421 This ioctl defines the physical address of a 1319 This ioctl defines the physical address of a three-page region in the guest
1422 physical address space. The region must be w 1320 physical address space. The region must be within the first 4GB of the
1423 guest physical address space and must not con 1321 guest physical address space and must not conflict with any memory slot
1424 or any mmio address. The guest may malfuncti 1322 or any mmio address. The guest may malfunction if it accesses this memory
1425 region. 1323 region.
1426 1324
1427 This ioctl is required on Intel-based hosts. 1325 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1428 because of a quirk in the virtualization impl 1326 because of a quirk in the virtualization implementation (see the internals
1429 documentation when it pops into existence). 1327 documentation when it pops into existence).
1430 1328
1431 1329
1432 4.37 KVM_ENABLE_CAP 1330 4.37 KVM_ENABLE_CAP
1433 ------------------- 1331 -------------------
1434 1332
1435 :Capability: KVM_CAP_ENABLE_CAP 1333 :Capability: KVM_CAP_ENABLE_CAP
1436 :Architectures: mips, ppc, s390, x86, loongar !! 1334 :Architectures: mips, ppc, s390
1437 :Type: vcpu ioctl 1335 :Type: vcpu ioctl
1438 :Parameters: struct kvm_enable_cap (in) 1336 :Parameters: struct kvm_enable_cap (in)
1439 :Returns: 0 on success; -1 on error 1337 :Returns: 0 on success; -1 on error
1440 1338
1441 :Capability: KVM_CAP_ENABLE_CAP_VM 1339 :Capability: KVM_CAP_ENABLE_CAP_VM
1442 :Architectures: all 1340 :Architectures: all
1443 :Type: vm ioctl 1341 :Type: vm ioctl
1444 :Parameters: struct kvm_enable_cap (in) 1342 :Parameters: struct kvm_enable_cap (in)
1445 :Returns: 0 on success; -1 on error 1343 :Returns: 0 on success; -1 on error
1446 1344
1447 .. note:: 1345 .. note::
1448 1346
1449 Not all extensions are enabled by default. 1347 Not all extensions are enabled by default. Using this ioctl the application
1450 can enable an extension, making it availab 1348 can enable an extension, making it available to the guest.
1451 1349
1452 On systems that do not support this ioctl, it 1350 On systems that do not support this ioctl, it always fails. On systems that
1453 do support it, it only works for extensions t 1351 do support it, it only works for extensions that are supported for enablement.
1454 1352
1455 To check if a capability can be enabled, the 1353 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1456 be used. 1354 be used.
1457 1355
1458 :: 1356 ::
1459 1357
1460 struct kvm_enable_cap { 1358 struct kvm_enable_cap {
1461 /* in */ 1359 /* in */
1462 __u32 cap; 1360 __u32 cap;
1463 1361
1464 The capability that is supposed to get enable 1362 The capability that is supposed to get enabled.
1465 1363
1466 :: 1364 ::
1467 1365
1468 __u32 flags; 1366 __u32 flags;
1469 1367
1470 A bitfield indicating future enhancements. Ha 1368 A bitfield indicating future enhancements. Has to be 0 for now.
1471 1369
1472 :: 1370 ::
1473 1371
1474 __u64 args[4]; 1372 __u64 args[4];
1475 1373
1476 Arguments for enabling a feature. If a featur 1374 Arguments for enabling a feature. If a feature needs initial values to
1477 function properly, this is the place to put t 1375 function properly, this is the place to put them.
1478 1376
1479 :: 1377 ::
1480 1378
1481 __u8 pad[64]; 1379 __u8 pad[64];
1482 }; 1380 };
1483 1381
1484 The vcpu ioctl should be used for vcpu-specif 1382 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1485 for vm-wide capabilities. 1383 for vm-wide capabilities.
1486 1384
1487 4.38 KVM_GET_MP_STATE 1385 4.38 KVM_GET_MP_STATE
1488 --------------------- 1386 ---------------------
1489 1387
1490 :Capability: KVM_CAP_MP_STATE 1388 :Capability: KVM_CAP_MP_STATE
1491 :Architectures: x86, s390, arm64, riscv, loon !! 1389 :Architectures: x86, s390, arm, arm64
1492 :Type: vcpu ioctl 1390 :Type: vcpu ioctl
1493 :Parameters: struct kvm_mp_state (out) 1391 :Parameters: struct kvm_mp_state (out)
1494 :Returns: 0 on success; -1 on error 1392 :Returns: 0 on success; -1 on error
1495 1393
1496 :: 1394 ::
1497 1395
1498 struct kvm_mp_state { 1396 struct kvm_mp_state {
1499 __u32 mp_state; 1397 __u32 mp_state;
1500 }; 1398 };
1501 1399
1502 Returns the vcpu's current "multiprocessing s 1400 Returns the vcpu's current "multiprocessing state" (though also valid on
1503 uniprocessor guests). 1401 uniprocessor guests).
1504 1402
1505 Possible values are: 1403 Possible values are:
1506 1404
1507 ========================== ============ 1405 ========================== ===============================================
1508 KVM_MP_STATE_RUNNABLE the vcpu is !! 1406 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64]
1509 [x86,arm64,r <<
1510 KVM_MP_STATE_UNINITIALIZED the vcpu is 1407 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1511 which has no 1408 which has not yet received an INIT signal [x86]
1512 KVM_MP_STATE_INIT_RECEIVED the vcpu has 1409 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1513 now ready fo 1410 now ready for a SIPI [x86]
1514 KVM_MP_STATE_HALTED the vcpu has 1411 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1515 is waiting f 1412 is waiting for an interrupt [x86]
1516 KVM_MP_STATE_SIPI_RECEIVED the vcpu has 1413 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1517 accessible v 1414 accessible via KVM_GET_VCPU_EVENTS) [x86]
1518 KVM_MP_STATE_STOPPED the vcpu is !! 1415 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64]
1519 KVM_MP_STATE_CHECK_STOP the vcpu is 1416 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1520 KVM_MP_STATE_OPERATING the vcpu is 1417 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1521 [s390] 1418 [s390]
1522 KVM_MP_STATE_LOAD the vcpu is 1419 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1523 [s390] 1420 [s390]
1524 KVM_MP_STATE_SUSPENDED the vcpu is <<
1525 for a wakeup <<
1526 ========================== ============ 1421 ========================== ===============================================
1527 1422
1528 On x86, this ioctl is only useful after KVM_C 1423 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1529 in-kernel irqchip, the multiprocessing state 1424 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1530 these architectures. 1425 these architectures.
1531 1426
1532 For arm64: !! 1427 For arm/arm64:
1533 ^^^^^^^^^^ !! 1428 ^^^^^^^^^^^^^^
1534 <<
1535 If a vCPU is in the KVM_MP_STATE_SUSPENDED st <<
1536 architectural execution of a WFI instruction. <<
1537 <<
1538 If a wakeup event is recognized, KVM will exi <<
1539 KVM_SYSTEM_EVENT exit, where the event type i <<
1540 userspace wants to honor the wakeup, it must <<
1541 KVM_MP_STATE_RUNNABLE. If it does not, KVM wi <<
1542 event in subsequent calls to KVM_RUN. <<
1543 <<
1544 .. warning:: <<
1545 <<
1546 If userspace intends to keep the vCPU in <<
1547 strongly recommended that userspace take <<
1548 wakeup event (such as masking an interru <<
1549 calls to KVM_RUN will immediately exit w <<
1550 event and inadvertently waste CPU cycles <<
1551 <<
1552 Additionally, if userspace takes action <<
1553 it is strongly recommended that it also <<
1554 original state when the vCPU is made RUN <<
1555 if userspace masked a pending interrupt <<
1556 the interrupt should be unmasked before <<
1557 guest. <<
1558 <<
1559 For riscv: <<
1560 ^^^^^^^^^^ <<
1561 1429
1562 The only states that are valid are KVM_MP_STA 1430 The only states that are valid are KVM_MP_STATE_STOPPED and
1563 KVM_MP_STATE_RUNNABLE which reflect if the vc 1431 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1564 1432
1565 On LoongArch, only the KVM_MP_STATE_RUNNABLE <<
1566 whether the vcpu is runnable. <<
1567 <<
1568 4.39 KVM_SET_MP_STATE 1433 4.39 KVM_SET_MP_STATE
1569 --------------------- 1434 ---------------------
1570 1435
1571 :Capability: KVM_CAP_MP_STATE 1436 :Capability: KVM_CAP_MP_STATE
1572 :Architectures: x86, s390, arm64, riscv, loon !! 1437 :Architectures: x86, s390, arm, arm64
1573 :Type: vcpu ioctl 1438 :Type: vcpu ioctl
1574 :Parameters: struct kvm_mp_state (in) 1439 :Parameters: struct kvm_mp_state (in)
1575 :Returns: 0 on success; -1 on error 1440 :Returns: 0 on success; -1 on error
1576 1441
1577 Sets the vcpu's current "multiprocessing stat 1442 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1578 arguments. 1443 arguments.
1579 1444
1580 On x86, this ioctl is only useful after KVM_C 1445 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1581 in-kernel irqchip, the multiprocessing state 1446 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1582 these architectures. 1447 these architectures.
1583 1448
1584 For arm64/riscv: !! 1449 For arm/arm64:
1585 ^^^^^^^^^^^^^^^^ !! 1450 ^^^^^^^^^^^^^^
1586 1451
1587 The only states that are valid are KVM_MP_STA 1452 The only states that are valid are KVM_MP_STATE_STOPPED and
1588 KVM_MP_STATE_RUNNABLE which reflect if the vc 1453 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1589 1454
1590 On LoongArch, only the KVM_MP_STATE_RUNNABLE <<
1591 whether the vcpu is runnable. <<
1592 <<
1593 4.40 KVM_SET_IDENTITY_MAP_ADDR 1455 4.40 KVM_SET_IDENTITY_MAP_ADDR
1594 ------------------------------ 1456 ------------------------------
1595 1457
1596 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR 1458 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1597 :Architectures: x86 1459 :Architectures: x86
1598 :Type: vm ioctl 1460 :Type: vm ioctl
1599 :Parameters: unsigned long identity (in) 1461 :Parameters: unsigned long identity (in)
1600 :Returns: 0 on success, -1 on error 1462 :Returns: 0 on success, -1 on error
1601 1463
1602 This ioctl defines the physical address of a 1464 This ioctl defines the physical address of a one-page region in the guest
1603 physical address space. The region must be w 1465 physical address space. The region must be within the first 4GB of the
1604 guest physical address space and must not con 1466 guest physical address space and must not conflict with any memory slot
1605 or any mmio address. The guest may malfuncti 1467 or any mmio address. The guest may malfunction if it accesses this memory
1606 region. 1468 region.
1607 1469
1608 Setting the address to 0 will result in reset 1470 Setting the address to 0 will result in resetting the address to its default
1609 (0xfffbc000). 1471 (0xfffbc000).
1610 1472
1611 This ioctl is required on Intel-based hosts. 1473 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1612 because of a quirk in the virtualization impl 1474 because of a quirk in the virtualization implementation (see the internals
1613 documentation when it pops into existence). 1475 documentation when it pops into existence).
1614 1476
1615 Fails if any VCPU has already been created. 1477 Fails if any VCPU has already been created.
1616 1478
1617 4.41 KVM_SET_BOOT_CPU_ID 1479 4.41 KVM_SET_BOOT_CPU_ID
1618 ------------------------ 1480 ------------------------
1619 1481
1620 :Capability: KVM_CAP_SET_BOOT_CPU_ID 1482 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1621 :Architectures: x86 1483 :Architectures: x86
1622 :Type: vm ioctl 1484 :Type: vm ioctl
1623 :Parameters: unsigned long vcpu_id 1485 :Parameters: unsigned long vcpu_id
1624 :Returns: 0 on success, -1 on error 1486 :Returns: 0 on success, -1 on error
1625 1487
1626 Define which vcpu is the Bootstrap Processor 1488 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1627 as the vcpu id in KVM_CREATE_VCPU. If this i 1489 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1628 is vcpu 0. This ioctl has to be called before !! 1490 is vcpu 0.
1629 otherwise it will return EBUSY error. <<
1630 1491
1631 1492
1632 4.42 KVM_GET_XSAVE 1493 4.42 KVM_GET_XSAVE
1633 ------------------ 1494 ------------------
1634 1495
1635 :Capability: KVM_CAP_XSAVE 1496 :Capability: KVM_CAP_XSAVE
1636 :Architectures: x86 1497 :Architectures: x86
1637 :Type: vcpu ioctl 1498 :Type: vcpu ioctl
1638 :Parameters: struct kvm_xsave (out) 1499 :Parameters: struct kvm_xsave (out)
1639 :Returns: 0 on success, -1 on error 1500 :Returns: 0 on success, -1 on error
1640 1501
1641 1502
1642 :: 1503 ::
1643 1504
1644 struct kvm_xsave { 1505 struct kvm_xsave {
1645 __u32 region[1024]; 1506 __u32 region[1024];
1646 __u32 extra[0]; <<
1647 }; 1507 };
1648 1508
1649 This ioctl would copy current vcpu's xsave st 1509 This ioctl would copy current vcpu's xsave struct to the userspace.
1650 1510
1651 1511
1652 4.43 KVM_SET_XSAVE 1512 4.43 KVM_SET_XSAVE
1653 ------------------ 1513 ------------------
1654 1514
1655 :Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2 !! 1515 :Capability: KVM_CAP_XSAVE
1656 :Architectures: x86 1516 :Architectures: x86
1657 :Type: vcpu ioctl 1517 :Type: vcpu ioctl
1658 :Parameters: struct kvm_xsave (in) 1518 :Parameters: struct kvm_xsave (in)
1659 :Returns: 0 on success, -1 on error 1519 :Returns: 0 on success, -1 on error
1660 1520
1661 :: 1521 ::
1662 1522
1663 1523
1664 struct kvm_xsave { 1524 struct kvm_xsave {
1665 __u32 region[1024]; 1525 __u32 region[1024];
1666 __u32 extra[0]; <<
1667 }; 1526 };
1668 1527
1669 This ioctl would copy userspace's xsave struc !! 1528 This ioctl would copy userspace's xsave struct to the kernel.
1670 as many bytes as are returned by KVM_CHECK_EX <<
1671 when invoked on the vm file descriptor. The s <<
1672 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will alwa <<
1673 Currently, it is only greater than 4096 if a <<
1674 enabled with ``arch_prctl()``, but this may c <<
1675 <<
1676 The offsets of the state save areas in struct <<
1677 contents of CPUID leaf 0xD on the host. <<
1678 1529
1679 1530
1680 4.44 KVM_GET_XCRS 1531 4.44 KVM_GET_XCRS
1681 ----------------- 1532 -----------------
1682 1533
1683 :Capability: KVM_CAP_XCRS 1534 :Capability: KVM_CAP_XCRS
1684 :Architectures: x86 1535 :Architectures: x86
1685 :Type: vcpu ioctl 1536 :Type: vcpu ioctl
1686 :Parameters: struct kvm_xcrs (out) 1537 :Parameters: struct kvm_xcrs (out)
1687 :Returns: 0 on success, -1 on error 1538 :Returns: 0 on success, -1 on error
1688 1539
1689 :: 1540 ::
1690 1541
1691 struct kvm_xcr { 1542 struct kvm_xcr {
1692 __u32 xcr; 1543 __u32 xcr;
1693 __u32 reserved; 1544 __u32 reserved;
1694 __u64 value; 1545 __u64 value;
1695 }; 1546 };
1696 1547
1697 struct kvm_xcrs { 1548 struct kvm_xcrs {
1698 __u32 nr_xcrs; 1549 __u32 nr_xcrs;
1699 __u32 flags; 1550 __u32 flags;
1700 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1551 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1701 __u64 padding[16]; 1552 __u64 padding[16];
1702 }; 1553 };
1703 1554
1704 This ioctl would copy current vcpu's xcrs to 1555 This ioctl would copy current vcpu's xcrs to the userspace.
1705 1556
1706 1557
1707 4.45 KVM_SET_XCRS 1558 4.45 KVM_SET_XCRS
1708 ----------------- 1559 -----------------
1709 1560
1710 :Capability: KVM_CAP_XCRS 1561 :Capability: KVM_CAP_XCRS
1711 :Architectures: x86 1562 :Architectures: x86
1712 :Type: vcpu ioctl 1563 :Type: vcpu ioctl
1713 :Parameters: struct kvm_xcrs (in) 1564 :Parameters: struct kvm_xcrs (in)
1714 :Returns: 0 on success, -1 on error 1565 :Returns: 0 on success, -1 on error
1715 1566
1716 :: 1567 ::
1717 1568
1718 struct kvm_xcr { 1569 struct kvm_xcr {
1719 __u32 xcr; 1570 __u32 xcr;
1720 __u32 reserved; 1571 __u32 reserved;
1721 __u64 value; 1572 __u64 value;
1722 }; 1573 };
1723 1574
1724 struct kvm_xcrs { 1575 struct kvm_xcrs {
1725 __u32 nr_xcrs; 1576 __u32 nr_xcrs;
1726 __u32 flags; 1577 __u32 flags;
1727 struct kvm_xcr xcrs[KVM_MAX_XCRS]; 1578 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1728 __u64 padding[16]; 1579 __u64 padding[16];
1729 }; 1580 };
1730 1581
1731 This ioctl would set vcpu's xcr to the value 1582 This ioctl would set vcpu's xcr to the value userspace specified.
1732 1583
1733 1584
1734 4.46 KVM_GET_SUPPORTED_CPUID 1585 4.46 KVM_GET_SUPPORTED_CPUID
1735 ---------------------------- 1586 ----------------------------
1736 1587
1737 :Capability: KVM_CAP_EXT_CPUID 1588 :Capability: KVM_CAP_EXT_CPUID
1738 :Architectures: x86 1589 :Architectures: x86
1739 :Type: system ioctl 1590 :Type: system ioctl
1740 :Parameters: struct kvm_cpuid2 (in/out) 1591 :Parameters: struct kvm_cpuid2 (in/out)
1741 :Returns: 0 on success, -1 on error 1592 :Returns: 0 on success, -1 on error
1742 1593
1743 :: 1594 ::
1744 1595
1745 struct kvm_cpuid2 { 1596 struct kvm_cpuid2 {
1746 __u32 nent; 1597 __u32 nent;
1747 __u32 padding; 1598 __u32 padding;
1748 struct kvm_cpuid_entry2 entries[0]; 1599 struct kvm_cpuid_entry2 entries[0];
1749 }; 1600 };
1750 1601
1751 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX 1602 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1752 #define KVM_CPUID_FLAG_STATEFUL_FUNC 1603 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1753 #define KVM_CPUID_FLAG_STATE_READ_NEXT 1604 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1754 1605
1755 struct kvm_cpuid_entry2 { 1606 struct kvm_cpuid_entry2 {
1756 __u32 function; 1607 __u32 function;
1757 __u32 index; 1608 __u32 index;
1758 __u32 flags; 1609 __u32 flags;
1759 __u32 eax; 1610 __u32 eax;
1760 __u32 ebx; 1611 __u32 ebx;
1761 __u32 ecx; 1612 __u32 ecx;
1762 __u32 edx; 1613 __u32 edx;
1763 __u32 padding[3]; 1614 __u32 padding[3];
1764 }; 1615 };
1765 1616
1766 This ioctl returns x86 cpuid features which a 1617 This ioctl returns x86 cpuid features which are supported by both the
1767 hardware and kvm in its default configuration 1618 hardware and kvm in its default configuration. Userspace can use the
1768 information returned by this ioctl to constru 1619 information returned by this ioctl to construct cpuid information (for
1769 KVM_SET_CPUID2) that is consistent with hardw 1620 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1770 userspace capabilities, and with user require 1621 userspace capabilities, and with user requirements (for example, the
1771 user may wish to constrain cpuid to emulate o 1622 user may wish to constrain cpuid to emulate older hardware, or for
1772 feature consistency across a cluster). 1623 feature consistency across a cluster).
1773 1624
1774 Dynamically-enabled feature bits need to be r <<
1775 ``arch_prctl()`` before calling this ioctl. F <<
1776 been requested are excluded from the result. <<
1777 <<
1778 Note that certain capabilities, such as KVM_C 1625 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1779 expose cpuid features (e.g. MONITOR) which ar 1626 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1780 its default configuration. If userspace enabl 1627 its default configuration. If userspace enables such capabilities, it
1781 is responsible for modifying the results of t 1628 is responsible for modifying the results of this ioctl appropriately.
1782 1629
1783 Userspace invokes KVM_GET_SUPPORTED_CPUID by 1630 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1784 with the 'nent' field indicating the number o 1631 with the 'nent' field indicating the number of entries in the variable-size
1785 array 'entries'. If the number of entries is 1632 array 'entries'. If the number of entries is too low to describe the cpu
1786 capabilities, an error (E2BIG) is returned. 1633 capabilities, an error (E2BIG) is returned. If the number is too high,
1787 the 'nent' field is adjusted and an error (EN 1634 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1788 number is just right, the 'nent' field is adj 1635 number is just right, the 'nent' field is adjusted to the number of valid
1789 entries in the 'entries' array, which is then 1636 entries in the 'entries' array, which is then filled.
1790 1637
1791 The entries returned are the host cpuid as re 1638 The entries returned are the host cpuid as returned by the cpuid instruction,
1792 with unknown or unsupported features masked o 1639 with unknown or unsupported features masked out. Some features (for example,
1793 x2apic), may not be present in the host cpu, 1640 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1794 emulate them efficiently. The fields in each 1641 emulate them efficiently. The fields in each entry are defined as follows:
1795 1642
1796 function: 1643 function:
1797 the eax value used to obtain the ent 1644 the eax value used to obtain the entry
1798 1645
1799 index: 1646 index:
1800 the ecx value used to obtain the ent 1647 the ecx value used to obtain the entry (for entries that are
1801 affected by ecx) 1648 affected by ecx)
1802 1649
1803 flags: 1650 flags:
1804 an OR of zero or more of the following: 1651 an OR of zero or more of the following:
1805 1652
1806 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 1653 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1807 if the index field is valid 1654 if the index field is valid
1808 1655
1809 eax, ebx, ecx, edx: 1656 eax, ebx, ecx, edx:
1810 the values returned by the cpuid ins 1657 the values returned by the cpuid instruction for
1811 this function/index combination 1658 this function/index combination
1812 1659
1813 The TSC deadline timer feature (CPUID leaf 1, 1660 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1814 as false, since the feature depends on KVM_CR 1661 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1815 support. Instead it is reported via:: 1662 support. Instead it is reported via::
1816 1663
1817 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEAD 1664 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1818 1665
1819 if that returns true and you use KVM_CREATE_I 1666 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1820 feature in userspace, then you can enable the 1667 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1821 1668
1822 1669
1823 4.47 KVM_PPC_GET_PVINFO 1670 4.47 KVM_PPC_GET_PVINFO
1824 ----------------------- 1671 -----------------------
1825 1672
1826 :Capability: KVM_CAP_PPC_GET_PVINFO 1673 :Capability: KVM_CAP_PPC_GET_PVINFO
1827 :Architectures: ppc 1674 :Architectures: ppc
1828 :Type: vm ioctl 1675 :Type: vm ioctl
1829 :Parameters: struct kvm_ppc_pvinfo (out) 1676 :Parameters: struct kvm_ppc_pvinfo (out)
1830 :Returns: 0 on success, !0 on error 1677 :Returns: 0 on success, !0 on error
1831 1678
1832 :: 1679 ::
1833 1680
1834 struct kvm_ppc_pvinfo { 1681 struct kvm_ppc_pvinfo {
1835 __u32 flags; 1682 __u32 flags;
1836 __u32 hcall[4]; 1683 __u32 hcall[4];
1837 __u8 pad[108]; 1684 __u8 pad[108];
1838 }; 1685 };
1839 1686
1840 This ioctl fetches PV specific information th 1687 This ioctl fetches PV specific information that need to be passed to the guest
1841 using the device tree or other means from vm 1688 using the device tree or other means from vm context.
1842 1689
1843 The hcall array defines 4 instructions that m 1690 The hcall array defines 4 instructions that make up a hypercall.
1844 1691
1845 If any additional field gets added to this st 1692 If any additional field gets added to this structure later on, a bit for that
1846 additional piece of information will be set i 1693 additional piece of information will be set in the flags bitmap.
1847 1694
1848 The flags bitmap is defined as:: 1695 The flags bitmap is defined as::
1849 1696
1850 /* the host supports the ePAPR idle hcall 1697 /* the host supports the ePAPR idle hcall
1851 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1< 1698 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1852 1699
1853 4.52 KVM_SET_GSI_ROUTING 1700 4.52 KVM_SET_GSI_ROUTING
1854 ------------------------ 1701 ------------------------
1855 1702
1856 :Capability: KVM_CAP_IRQ_ROUTING 1703 :Capability: KVM_CAP_IRQ_ROUTING
1857 :Architectures: x86 s390 arm64 !! 1704 :Architectures: x86 s390 arm arm64
1858 :Type: vm ioctl 1705 :Type: vm ioctl
1859 :Parameters: struct kvm_irq_routing (in) 1706 :Parameters: struct kvm_irq_routing (in)
1860 :Returns: 0 on success, -1 on error 1707 :Returns: 0 on success, -1 on error
1861 1708
1862 Sets the GSI routing table entries, overwriti 1709 Sets the GSI routing table entries, overwriting any previously set entries.
1863 1710
1864 On arm64, GSI routing has the following limit !! 1711 On arm/arm64, GSI routing has the following limitation:
1865 1712
1866 - GSI routing does not apply to KVM_IRQ_LINE 1713 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1867 1714
1868 :: 1715 ::
1869 1716
1870 struct kvm_irq_routing { 1717 struct kvm_irq_routing {
1871 __u32 nr; 1718 __u32 nr;
1872 __u32 flags; 1719 __u32 flags;
1873 struct kvm_irq_routing_entry entries[ 1720 struct kvm_irq_routing_entry entries[0];
1874 }; 1721 };
1875 1722
1876 No flags are specified so far, the correspond 1723 No flags are specified so far, the corresponding field must be set to zero.
1877 1724
1878 :: 1725 ::
1879 1726
1880 struct kvm_irq_routing_entry { 1727 struct kvm_irq_routing_entry {
1881 __u32 gsi; 1728 __u32 gsi;
1882 __u32 type; 1729 __u32 type;
1883 __u32 flags; 1730 __u32 flags;
1884 __u32 pad; 1731 __u32 pad;
1885 union { 1732 union {
1886 struct kvm_irq_routing_irqchi 1733 struct kvm_irq_routing_irqchip irqchip;
1887 struct kvm_irq_routing_msi ms 1734 struct kvm_irq_routing_msi msi;
1888 struct kvm_irq_routing_s390_a 1735 struct kvm_irq_routing_s390_adapter adapter;
1889 struct kvm_irq_routing_hv_sin 1736 struct kvm_irq_routing_hv_sint hv_sint;
1890 struct kvm_irq_routing_xen_ev <<
1891 __u32 pad[8]; 1737 __u32 pad[8];
1892 } u; 1738 } u;
1893 }; 1739 };
1894 1740
1895 /* gsi routing entry types */ 1741 /* gsi routing entry types */
1896 #define KVM_IRQ_ROUTING_IRQCHIP 1 1742 #define KVM_IRQ_ROUTING_IRQCHIP 1
1897 #define KVM_IRQ_ROUTING_MSI 2 1743 #define KVM_IRQ_ROUTING_MSI 2
1898 #define KVM_IRQ_ROUTING_S390_ADAPTER 3 1744 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1899 #define KVM_IRQ_ROUTING_HV_SINT 4 1745 #define KVM_IRQ_ROUTING_HV_SINT 4
1900 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5 <<
1901 1746
1902 flags: 1747 flags:
1903 1748
1904 - KVM_MSI_VALID_DEVID: used along with KVM_IR 1749 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1905 type, specifies that the devid field contai 1750 type, specifies that the devid field contains a valid value. The per-VM
1906 KVM_CAP_MSI_DEVID capability advertises the 1751 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1907 the device ID. If this capability is not a 1752 the device ID. If this capability is not available, userspace should
1908 never set the KVM_MSI_VALID_DEVID flag as t 1753 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1909 - zero otherwise 1754 - zero otherwise
1910 1755
1911 :: 1756 ::
1912 1757
1913 struct kvm_irq_routing_irqchip { 1758 struct kvm_irq_routing_irqchip {
1914 __u32 irqchip; 1759 __u32 irqchip;
1915 __u32 pin; 1760 __u32 pin;
1916 }; 1761 };
1917 1762
1918 struct kvm_irq_routing_msi { 1763 struct kvm_irq_routing_msi {
1919 __u32 address_lo; 1764 __u32 address_lo;
1920 __u32 address_hi; 1765 __u32 address_hi;
1921 __u32 data; 1766 __u32 data;
1922 union { 1767 union {
1923 __u32 pad; 1768 __u32 pad;
1924 __u32 devid; 1769 __u32 devid;
1925 }; 1770 };
1926 }; 1771 };
1927 1772
1928 If KVM_MSI_VALID_DEVID is set, devid contains 1773 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1929 for the device that wrote the MSI message. F 1774 for the device that wrote the MSI message. For PCI, this is usually a
1930 BDF identifier in the lower 16 bits. !! 1775 BFD identifier in the lower 16 bits.
1931 1776
1932 On x86, address_hi is ignored unless the KVM_ 1777 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1933 feature of KVM_CAP_X2APIC_API capability is e 1778 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1934 address_hi bits 31-8 provide bits 31-8 of the 1779 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1935 address_hi must be zero. 1780 address_hi must be zero.
1936 1781
1937 :: 1782 ::
1938 1783
1939 struct kvm_irq_routing_s390_adapter { 1784 struct kvm_irq_routing_s390_adapter {
1940 __u64 ind_addr; 1785 __u64 ind_addr;
1941 __u64 summary_addr; 1786 __u64 summary_addr;
1942 __u64 ind_offset; 1787 __u64 ind_offset;
1943 __u32 summary_offset; 1788 __u32 summary_offset;
1944 __u32 adapter_id; 1789 __u32 adapter_id;
1945 }; 1790 };
1946 1791
1947 struct kvm_irq_routing_hv_sint { 1792 struct kvm_irq_routing_hv_sint {
1948 __u32 vcpu; 1793 __u32 vcpu;
1949 __u32 sint; 1794 __u32 sint;
1950 }; 1795 };
1951 1796
1952 struct kvm_irq_routing_xen_evtchn { <<
1953 __u32 port; <<
1954 __u32 vcpu; <<
1955 __u32 priority; <<
1956 }; <<
1957 <<
1958 <<
1959 When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM <<
1960 in its indication of supported features, rout <<
1961 is supported. Although the priority field is <<
1962 KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported <<
1963 2 level event channels. FIFO event channel su <<
1964 the future. <<
1965 <<
1966 1797
1967 4.55 KVM_SET_TSC_KHZ 1798 4.55 KVM_SET_TSC_KHZ
1968 -------------------- 1799 --------------------
1969 1800
1970 :Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM !! 1801 :Capability: KVM_CAP_TSC_CONTROL
1971 :Architectures: x86 1802 :Architectures: x86
1972 :Type: vcpu ioctl / vm ioctl !! 1803 :Type: vcpu ioctl
1973 :Parameters: virtual tsc_khz 1804 :Parameters: virtual tsc_khz
1974 :Returns: 0 on success, -1 on error 1805 :Returns: 0 on success, -1 on error
1975 1806
1976 Specifies the tsc frequency for the virtual m 1807 Specifies the tsc frequency for the virtual machine. The unit of the
1977 frequency is KHz. 1808 frequency is KHz.
1978 1809
1979 If the KVM_CAP_VM_TSC_CONTROL capability is a <<
1980 be used as a vm ioctl to set the initial tsc <<
1981 created vCPUs. <<
1982 1810
1983 4.56 KVM_GET_TSC_KHZ 1811 4.56 KVM_GET_TSC_KHZ
1984 -------------------- 1812 --------------------
1985 1813
1986 :Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM !! 1814 :Capability: KVM_CAP_GET_TSC_KHZ
1987 :Architectures: x86 1815 :Architectures: x86
1988 :Type: vcpu ioctl / vm ioctl !! 1816 :Type: vcpu ioctl
1989 :Parameters: none 1817 :Parameters: none
1990 :Returns: virtual tsc-khz on success, negativ 1818 :Returns: virtual tsc-khz on success, negative value on error
1991 1819
1992 Returns the tsc frequency of the guest. The u 1820 Returns the tsc frequency of the guest. The unit of the return value is
1993 KHz. If the host has unstable tsc this ioctl 1821 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1994 error. 1822 error.
1995 1823
1996 1824
1997 4.57 KVM_GET_LAPIC 1825 4.57 KVM_GET_LAPIC
1998 ------------------ 1826 ------------------
1999 1827
2000 :Capability: KVM_CAP_IRQCHIP 1828 :Capability: KVM_CAP_IRQCHIP
2001 :Architectures: x86 1829 :Architectures: x86
2002 :Type: vcpu ioctl 1830 :Type: vcpu ioctl
2003 :Parameters: struct kvm_lapic_state (out) 1831 :Parameters: struct kvm_lapic_state (out)
2004 :Returns: 0 on success, -1 on error 1832 :Returns: 0 on success, -1 on error
2005 1833
2006 :: 1834 ::
2007 1835
2008 #define KVM_APIC_REG_SIZE 0x400 1836 #define KVM_APIC_REG_SIZE 0x400
2009 struct kvm_lapic_state { 1837 struct kvm_lapic_state {
2010 char regs[KVM_APIC_REG_SIZE]; 1838 char regs[KVM_APIC_REG_SIZE];
2011 }; 1839 };
2012 1840
2013 Reads the Local APIC registers and copies the 1841 Reads the Local APIC registers and copies them into the input argument. The
2014 data format and layout are the same as docume 1842 data format and layout are the same as documented in the architecture manual.
2015 1843
2016 If KVM_X2APIC_API_USE_32BIT_IDS feature of KV 1844 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
2017 enabled, then the format of APIC_ID register 1845 enabled, then the format of APIC_ID register depends on the APIC mode
2018 (reported by MSR_IA32_APICBASE) of its VCPU. 1846 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
2019 the APIC_ID register (bytes 32-35). xAPIC on 1847 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
2020 which is stored in bits 31-24 of the APIC reg 1848 which is stored in bits 31-24 of the APIC register, or equivalently in
2021 byte 35 of struct kvm_lapic_state's regs fiel 1849 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
2022 be called after MSR_IA32_APICBASE has been se 1850 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
2023 1851
2024 If KVM_X2APIC_API_USE_32BIT_IDS feature is di 1852 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
2025 always uses xAPIC format. 1853 always uses xAPIC format.
2026 1854
2027 1855
2028 4.58 KVM_SET_LAPIC 1856 4.58 KVM_SET_LAPIC
2029 ------------------ 1857 ------------------
2030 1858
2031 :Capability: KVM_CAP_IRQCHIP 1859 :Capability: KVM_CAP_IRQCHIP
2032 :Architectures: x86 1860 :Architectures: x86
2033 :Type: vcpu ioctl 1861 :Type: vcpu ioctl
2034 :Parameters: struct kvm_lapic_state (in) 1862 :Parameters: struct kvm_lapic_state (in)
2035 :Returns: 0 on success, -1 on error 1863 :Returns: 0 on success, -1 on error
2036 1864
2037 :: 1865 ::
2038 1866
2039 #define KVM_APIC_REG_SIZE 0x400 1867 #define KVM_APIC_REG_SIZE 0x400
2040 struct kvm_lapic_state { 1868 struct kvm_lapic_state {
2041 char regs[KVM_APIC_REG_SIZE]; 1869 char regs[KVM_APIC_REG_SIZE];
2042 }; 1870 };
2043 1871
2044 Copies the input argument into the Local APIC 1872 Copies the input argument into the Local APIC registers. The data format
2045 and layout are the same as documented in the 1873 and layout are the same as documented in the architecture manual.
2046 1874
2047 The format of the APIC ID register (bytes 32- 1875 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
2048 regs field) depends on the state of the KVM_C 1876 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
2049 See the note in KVM_GET_LAPIC. 1877 See the note in KVM_GET_LAPIC.
2050 1878
2051 1879
2052 4.59 KVM_IOEVENTFD 1880 4.59 KVM_IOEVENTFD
2053 ------------------ 1881 ------------------
2054 1882
2055 :Capability: KVM_CAP_IOEVENTFD 1883 :Capability: KVM_CAP_IOEVENTFD
2056 :Architectures: all 1884 :Architectures: all
2057 :Type: vm ioctl 1885 :Type: vm ioctl
2058 :Parameters: struct kvm_ioeventfd (in) 1886 :Parameters: struct kvm_ioeventfd (in)
2059 :Returns: 0 on success, !0 on error 1887 :Returns: 0 on success, !0 on error
2060 1888
2061 This ioctl attaches or detaches an ioeventfd 1889 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
2062 within the guest. A guest write in the regis 1890 within the guest. A guest write in the registered address will signal the
2063 provided event instead of triggering an exit. 1891 provided event instead of triggering an exit.
2064 1892
2065 :: 1893 ::
2066 1894
2067 struct kvm_ioeventfd { 1895 struct kvm_ioeventfd {
2068 __u64 datamatch; 1896 __u64 datamatch;
2069 __u64 addr; /* legal pio/mmio 1897 __u64 addr; /* legal pio/mmio address */
2070 __u32 len; /* 0, 1, 2, 4, or 1898 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
2071 __s32 fd; 1899 __s32 fd;
2072 __u32 flags; 1900 __u32 flags;
2073 __u8 pad[36]; 1901 __u8 pad[36];
2074 }; 1902 };
2075 1903
2076 For the special case of virtio-ccw devices on 1904 For the special case of virtio-ccw devices on s390, the ioevent is matched
2077 to a subchannel/virtqueue tuple instead. 1905 to a subchannel/virtqueue tuple instead.
2078 1906
2079 The following flags are defined:: 1907 The following flags are defined::
2080 1908
2081 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << 1909 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
2082 #define KVM_IOEVENTFD_FLAG_PIO (1 << 1910 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
2083 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << 1911 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
2084 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIF 1912 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
2085 (1 << kvm_ioeventfd_flag_nr_virtio_cc 1913 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
2086 1914
2087 If datamatch flag is set, the event will be s 1915 If datamatch flag is set, the event will be signaled only if the written value
2088 to the registered address is equal to datamat 1916 to the registered address is equal to datamatch in struct kvm_ioeventfd.
2089 1917
2090 For virtio-ccw devices, addr contains the sub 1918 For virtio-ccw devices, addr contains the subchannel id and datamatch the
2091 virtqueue index. 1919 virtqueue index.
2092 1920
2093 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero len 1921 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
2094 the kernel will ignore the length of guest wr 1922 the kernel will ignore the length of guest write and may get a faster vmexit.
2095 The speedup may only apply to specific archit 1923 The speedup may only apply to specific architectures, but the ioeventfd will
2096 work anyway. 1924 work anyway.
2097 1925
2098 4.60 KVM_DIRTY_TLB 1926 4.60 KVM_DIRTY_TLB
2099 ------------------ 1927 ------------------
2100 1928
2101 :Capability: KVM_CAP_SW_TLB 1929 :Capability: KVM_CAP_SW_TLB
2102 :Architectures: ppc 1930 :Architectures: ppc
2103 :Type: vcpu ioctl 1931 :Type: vcpu ioctl
2104 :Parameters: struct kvm_dirty_tlb (in) 1932 :Parameters: struct kvm_dirty_tlb (in)
2105 :Returns: 0 on success, -1 on error 1933 :Returns: 0 on success, -1 on error
2106 1934
2107 :: 1935 ::
2108 1936
2109 struct kvm_dirty_tlb { 1937 struct kvm_dirty_tlb {
2110 __u64 bitmap; 1938 __u64 bitmap;
2111 __u32 num_dirty; 1939 __u32 num_dirty;
2112 }; 1940 };
2113 1941
2114 This must be called whenever userspace has ch 1942 This must be called whenever userspace has changed an entry in the shared
2115 TLB, prior to calling KVM_RUN on the associat 1943 TLB, prior to calling KVM_RUN on the associated vcpu.
2116 1944
2117 The "bitmap" field is the userspace address o 1945 The "bitmap" field is the userspace address of an array. This array
2118 consists of a number of bits, equal to the to 1946 consists of a number of bits, equal to the total number of TLB entries as
2119 determined by the last successful call to KVM 1947 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
2120 nearest multiple of 64. 1948 nearest multiple of 64.
2121 1949
2122 Each bit corresponds to one TLB entry, ordere 1950 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
2123 array. 1951 array.
2124 1952
2125 The array is little-endian: the bit 0 is the 1953 The array is little-endian: the bit 0 is the least significant bit of the
2126 first byte, bit 8 is the least significant bi 1954 first byte, bit 8 is the least significant bit of the second byte, etc.
2127 This avoids any complications with differing 1955 This avoids any complications with differing word sizes.
2128 1956
2129 The "num_dirty" field is a performance hint f 1957 The "num_dirty" field is a performance hint for KVM to determine whether it
2130 should skip processing the bitmap and just in 1958 should skip processing the bitmap and just invalidate everything. It must
2131 be set to the number of set bits in the bitma 1959 be set to the number of set bits in the bitmap.
2132 1960
2133 1961
2134 4.62 KVM_CREATE_SPAPR_TCE 1962 4.62 KVM_CREATE_SPAPR_TCE
2135 ------------------------- 1963 -------------------------
2136 1964
2137 :Capability: KVM_CAP_SPAPR_TCE 1965 :Capability: KVM_CAP_SPAPR_TCE
2138 :Architectures: powerpc 1966 :Architectures: powerpc
2139 :Type: vm ioctl 1967 :Type: vm ioctl
2140 :Parameters: struct kvm_create_spapr_tce (in) 1968 :Parameters: struct kvm_create_spapr_tce (in)
2141 :Returns: file descriptor for manipulating th 1969 :Returns: file descriptor for manipulating the created TCE table
2142 1970
2143 This creates a virtual TCE (translation contr 1971 This creates a virtual TCE (translation control entry) table, which
2144 is an IOMMU for PAPR-style virtual I/O. It i 1972 is an IOMMU for PAPR-style virtual I/O. It is used to translate
2145 logical addresses used in virtual I/O into gu 1973 logical addresses used in virtual I/O into guest physical addresses,
2146 and provides a scatter/gather capability for 1974 and provides a scatter/gather capability for PAPR virtual I/O.
2147 1975
2148 :: 1976 ::
2149 1977
2150 /* for KVM_CAP_SPAPR_TCE */ 1978 /* for KVM_CAP_SPAPR_TCE */
2151 struct kvm_create_spapr_tce { 1979 struct kvm_create_spapr_tce {
2152 __u64 liobn; 1980 __u64 liobn;
2153 __u32 window_size; 1981 __u32 window_size;
2154 }; 1982 };
2155 1983
2156 The liobn field gives the logical IO bus numb 1984 The liobn field gives the logical IO bus number for which to create a
2157 TCE table. The window_size field specifies t 1985 TCE table. The window_size field specifies the size of the DMA window
2158 which this TCE table will translate - the tab 1986 which this TCE table will translate - the table will contain one 64
2159 bit TCE entry for every 4kiB of the DMA windo 1987 bit TCE entry for every 4kiB of the DMA window.
2160 1988
2161 When the guest issues an H_PUT_TCE hcall on a 1989 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
2162 table has been created using this ioctl(), th 1990 table has been created using this ioctl(), the kernel will handle it
2163 in real mode, updating the TCE table. H_PUT_ 1991 in real mode, updating the TCE table. H_PUT_TCE calls for other
2164 liobns will cause a vm exit and must be handl 1992 liobns will cause a vm exit and must be handled by userspace.
2165 1993
2166 The return value is a file descriptor which c 1994 The return value is a file descriptor which can be passed to mmap(2)
2167 to map the created TCE table into userspace. 1995 to map the created TCE table into userspace. This lets userspace read
2168 the entries written by kernel-handled H_PUT_T 1996 the entries written by kernel-handled H_PUT_TCE calls, and also lets
2169 userspace update the TCE table directly which 1997 userspace update the TCE table directly which is useful in some
2170 circumstances. 1998 circumstances.
2171 1999
2172 2000
2173 4.63 KVM_ALLOCATE_RMA 2001 4.63 KVM_ALLOCATE_RMA
2174 --------------------- 2002 ---------------------
2175 2003
2176 :Capability: KVM_CAP_PPC_RMA 2004 :Capability: KVM_CAP_PPC_RMA
2177 :Architectures: powerpc 2005 :Architectures: powerpc
2178 :Type: vm ioctl 2006 :Type: vm ioctl
2179 :Parameters: struct kvm_allocate_rma (out) 2007 :Parameters: struct kvm_allocate_rma (out)
2180 :Returns: file descriptor for mapping the all 2008 :Returns: file descriptor for mapping the allocated RMA
2181 2009
2182 This allocates a Real Mode Area (RMA) from th 2010 This allocates a Real Mode Area (RMA) from the pool allocated at boot
2183 time by the kernel. An RMA is a physically-c 2011 time by the kernel. An RMA is a physically-contiguous, aligned region
2184 of memory used on older POWER processors to p 2012 of memory used on older POWER processors to provide the memory which
2185 will be accessed by real-mode (MMU off) acces 2013 will be accessed by real-mode (MMU off) accesses in a KVM guest.
2186 POWER processors support a set of sizes for t 2014 POWER processors support a set of sizes for the RMA that usually
2187 includes 64MB, 128MB, 256MB and some larger p 2015 includes 64MB, 128MB, 256MB and some larger powers of two.
2188 2016
2189 :: 2017 ::
2190 2018
2191 /* for KVM_ALLOCATE_RMA */ 2019 /* for KVM_ALLOCATE_RMA */
2192 struct kvm_allocate_rma { 2020 struct kvm_allocate_rma {
2193 __u64 rma_size; 2021 __u64 rma_size;
2194 }; 2022 };
2195 2023
2196 The return value is a file descriptor which c 2024 The return value is a file descriptor which can be passed to mmap(2)
2197 to map the allocated RMA into userspace. The 2025 to map the allocated RMA into userspace. The mapped area can then be
2198 passed to the KVM_SET_USER_MEMORY_REGION ioct 2026 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2199 RMA for a virtual machine. The size of the R 2027 RMA for a virtual machine. The size of the RMA in bytes (which is
2200 fixed at host kernel boot time) is returned i 2028 fixed at host kernel boot time) is returned in the rma_size field of
2201 the argument structure. 2029 the argument structure.
2202 2030
2203 The KVM_CAP_PPC_RMA capability is 1 or 2 if t 2031 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2204 is supported; 2 if the processor requires all 2032 is supported; 2 if the processor requires all virtual machines to have
2205 an RMA, or 1 if the processor can use an RMA 2033 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2206 because it supports the Virtual RMA (VRMA) fa 2034 because it supports the Virtual RMA (VRMA) facility.
2207 2035
2208 2036
2209 4.64 KVM_NMI 2037 4.64 KVM_NMI
2210 ------------ 2038 ------------
2211 2039
2212 :Capability: KVM_CAP_USER_NMI 2040 :Capability: KVM_CAP_USER_NMI
2213 :Architectures: x86 2041 :Architectures: x86
2214 :Type: vcpu ioctl 2042 :Type: vcpu ioctl
2215 :Parameters: none 2043 :Parameters: none
2216 :Returns: 0 on success, -1 on error 2044 :Returns: 0 on success, -1 on error
2217 2045
2218 Queues an NMI on the thread's vcpu. Note thi 2046 Queues an NMI on the thread's vcpu. Note this is well defined only
2219 when KVM_CREATE_IRQCHIP has not been called, 2047 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2220 between the virtual cpu core and virtual loca 2048 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2221 has been called, this interface is completely 2049 has been called, this interface is completely emulated within the kernel.
2222 2050
2223 To use this to emulate the LINT1 input with K 2051 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2224 following algorithm: 2052 following algorithm:
2225 2053
2226 - pause the vcpu 2054 - pause the vcpu
2227 - read the local APIC's state (KVM_GET_LAPI 2055 - read the local APIC's state (KVM_GET_LAPIC)
2228 - check whether changing LINT1 will queue a 2056 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2229 - if so, issue KVM_NMI 2057 - if so, issue KVM_NMI
2230 - resume the vcpu 2058 - resume the vcpu
2231 2059
2232 Some guests configure the LINT1 NMI input to 2060 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2233 debugging. 2061 debugging.
2234 2062
2235 2063
2236 4.65 KVM_S390_UCAS_MAP 2064 4.65 KVM_S390_UCAS_MAP
2237 ---------------------- 2065 ----------------------
2238 2066
2239 :Capability: KVM_CAP_S390_UCONTROL 2067 :Capability: KVM_CAP_S390_UCONTROL
2240 :Architectures: s390 2068 :Architectures: s390
2241 :Type: vcpu ioctl 2069 :Type: vcpu ioctl
2242 :Parameters: struct kvm_s390_ucas_mapping (in 2070 :Parameters: struct kvm_s390_ucas_mapping (in)
2243 :Returns: 0 in case of success 2071 :Returns: 0 in case of success
2244 2072
2245 The parameter is defined like this:: 2073 The parameter is defined like this::
2246 2074
2247 struct kvm_s390_ucas_mapping { 2075 struct kvm_s390_ucas_mapping {
2248 __u64 user_addr; 2076 __u64 user_addr;
2249 __u64 vcpu_addr; 2077 __u64 vcpu_addr;
2250 __u64 length; 2078 __u64 length;
2251 }; 2079 };
2252 2080
2253 This ioctl maps the memory at "user_addr" wit 2081 This ioctl maps the memory at "user_addr" with the length "length" to
2254 the vcpu's address space starting at "vcpu_ad 2082 the vcpu's address space starting at "vcpu_addr". All parameters need to
2255 be aligned by 1 megabyte. 2083 be aligned by 1 megabyte.
2256 2084
2257 2085
2258 4.66 KVM_S390_UCAS_UNMAP 2086 4.66 KVM_S390_UCAS_UNMAP
2259 ------------------------ 2087 ------------------------
2260 2088
2261 :Capability: KVM_CAP_S390_UCONTROL 2089 :Capability: KVM_CAP_S390_UCONTROL
2262 :Architectures: s390 2090 :Architectures: s390
2263 :Type: vcpu ioctl 2091 :Type: vcpu ioctl
2264 :Parameters: struct kvm_s390_ucas_mapping (in 2092 :Parameters: struct kvm_s390_ucas_mapping (in)
2265 :Returns: 0 in case of success 2093 :Returns: 0 in case of success
2266 2094
2267 The parameter is defined like this:: 2095 The parameter is defined like this::
2268 2096
2269 struct kvm_s390_ucas_mapping { 2097 struct kvm_s390_ucas_mapping {
2270 __u64 user_addr; 2098 __u64 user_addr;
2271 __u64 vcpu_addr; 2099 __u64 vcpu_addr;
2272 __u64 length; 2100 __u64 length;
2273 }; 2101 };
2274 2102
2275 This ioctl unmaps the memory in the vcpu's ad 2103 This ioctl unmaps the memory in the vcpu's address space starting at
2276 "vcpu_addr" with the length "length". The fie 2104 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2277 All parameters need to be aligned by 1 megaby 2105 All parameters need to be aligned by 1 megabyte.
2278 2106
2279 2107
2280 4.67 KVM_S390_VCPU_FAULT 2108 4.67 KVM_S390_VCPU_FAULT
2281 ------------------------ 2109 ------------------------
2282 2110
2283 :Capability: KVM_CAP_S390_UCONTROL 2111 :Capability: KVM_CAP_S390_UCONTROL
2284 :Architectures: s390 2112 :Architectures: s390
2285 :Type: vcpu ioctl 2113 :Type: vcpu ioctl
2286 :Parameters: vcpu absolute address (in) 2114 :Parameters: vcpu absolute address (in)
2287 :Returns: 0 in case of success 2115 :Returns: 0 in case of success
2288 2116
2289 This call creates a page table entry on the v 2117 This call creates a page table entry on the virtual cpu's address space
2290 (for user controlled virtual machines) or the 2118 (for user controlled virtual machines) or the virtual machine's address
2291 space (for regular virtual machines). This on 2119 space (for regular virtual machines). This only works for minor faults,
2292 thus it's recommended to access subject memor 2120 thus it's recommended to access subject memory page via the user page
2293 table upfront. This is useful to handle valid 2121 table upfront. This is useful to handle validity intercepts for user
2294 controlled virtual machines to fault in the v 2122 controlled virtual machines to fault in the virtual cpu's lowcore pages
2295 prior to calling the KVM_RUN ioctl. 2123 prior to calling the KVM_RUN ioctl.
2296 2124
2297 2125
2298 4.68 KVM_SET_ONE_REG 2126 4.68 KVM_SET_ONE_REG
2299 -------------------- 2127 --------------------
2300 2128
2301 :Capability: KVM_CAP_ONE_REG 2129 :Capability: KVM_CAP_ONE_REG
2302 :Architectures: all 2130 :Architectures: all
2303 :Type: vcpu ioctl 2131 :Type: vcpu ioctl
2304 :Parameters: struct kvm_one_reg (in) 2132 :Parameters: struct kvm_one_reg (in)
2305 :Returns: 0 on success, negative value on fai 2133 :Returns: 0 on success, negative value on failure
2306 2134
2307 Errors: 2135 Errors:
2308 2136
2309 ====== ================================== 2137 ====== ============================================================
2310 ENOENT no such register !! 2138 Â ENOENT Â Â no such register
2311 EINVAL invalid register ID, or no such re !! 2139 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2312 protected virtualization mode on s 2140 protected virtualization mode on s390
2313 EPERM (arm64) register access not allowe !! 2141 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2314 EBUSY (riscv) changing register value no <<
2315 has run at least once <<
2316 ====== ================================== 2142 ====== ============================================================
2317 2143
2318 (These error codes are indicative only: do no 2144 (These error codes are indicative only: do not rely on a specific error
2319 code being returned in a specific situation.) 2145 code being returned in a specific situation.)
2320 2146
2321 :: 2147 ::
2322 2148
2323 struct kvm_one_reg { 2149 struct kvm_one_reg {
2324 __u64 id; 2150 __u64 id;
2325 __u64 addr; 2151 __u64 addr;
2326 }; 2152 };
2327 2153
2328 Using this ioctl, a single vcpu register can 2154 Using this ioctl, a single vcpu register can be set to a specific value
2329 defined by user space with the passed in stru 2155 defined by user space with the passed in struct kvm_one_reg, where id
2330 refers to the register identifier as describe 2156 refers to the register identifier as described below and addr is a pointer
2331 to a variable with the respective size. There 2157 to a variable with the respective size. There can be architecture agnostic
2332 and architecture specific registers. Each hav 2158 and architecture specific registers. Each have their own range of operation
2333 and their own constants and width. To keep tr 2159 and their own constants and width. To keep track of the implemented
2334 registers, find a list below: 2160 registers, find a list below:
2335 2161
2336 ======= =============================== === 2162 ======= =============================== ============
2337 Arch Register Wid 2163 Arch Register Width (bits)
2338 ======= =============================== === 2164 ======= =============================== ============
2339 PPC KVM_REG_PPC_HIOR 64 2165 PPC KVM_REG_PPC_HIOR 64
2340 PPC KVM_REG_PPC_IAC1 64 2166 PPC KVM_REG_PPC_IAC1 64
2341 PPC KVM_REG_PPC_IAC2 64 2167 PPC KVM_REG_PPC_IAC2 64
2342 PPC KVM_REG_PPC_IAC3 64 2168 PPC KVM_REG_PPC_IAC3 64
2343 PPC KVM_REG_PPC_IAC4 64 2169 PPC KVM_REG_PPC_IAC4 64
2344 PPC KVM_REG_PPC_DAC1 64 2170 PPC KVM_REG_PPC_DAC1 64
2345 PPC KVM_REG_PPC_DAC2 64 2171 PPC KVM_REG_PPC_DAC2 64
2346 PPC KVM_REG_PPC_DABR 64 2172 PPC KVM_REG_PPC_DABR 64
2347 PPC KVM_REG_PPC_DSCR 64 2173 PPC KVM_REG_PPC_DSCR 64
2348 PPC KVM_REG_PPC_PURR 64 2174 PPC KVM_REG_PPC_PURR 64
2349 PPC KVM_REG_PPC_SPURR 64 2175 PPC KVM_REG_PPC_SPURR 64
2350 PPC KVM_REG_PPC_DAR 64 2176 PPC KVM_REG_PPC_DAR 64
2351 PPC KVM_REG_PPC_DSISR 32 2177 PPC KVM_REG_PPC_DSISR 32
2352 PPC KVM_REG_PPC_AMR 64 2178 PPC KVM_REG_PPC_AMR 64
2353 PPC KVM_REG_PPC_UAMOR 64 2179 PPC KVM_REG_PPC_UAMOR 64
2354 PPC KVM_REG_PPC_MMCR0 64 2180 PPC KVM_REG_PPC_MMCR0 64
2355 PPC KVM_REG_PPC_MMCR1 64 2181 PPC KVM_REG_PPC_MMCR1 64
2356 PPC KVM_REG_PPC_MMCRA 64 2182 PPC KVM_REG_PPC_MMCRA 64
2357 PPC KVM_REG_PPC_MMCR2 64 2183 PPC KVM_REG_PPC_MMCR2 64
2358 PPC KVM_REG_PPC_MMCRS 64 2184 PPC KVM_REG_PPC_MMCRS 64
2359 PPC KVM_REG_PPC_MMCR3 64 2185 PPC KVM_REG_PPC_MMCR3 64
2360 PPC KVM_REG_PPC_SIAR 64 2186 PPC KVM_REG_PPC_SIAR 64
2361 PPC KVM_REG_PPC_SDAR 64 2187 PPC KVM_REG_PPC_SDAR 64
2362 PPC KVM_REG_PPC_SIER 64 2188 PPC KVM_REG_PPC_SIER 64
2363 PPC KVM_REG_PPC_SIER2 64 2189 PPC KVM_REG_PPC_SIER2 64
2364 PPC KVM_REG_PPC_SIER3 64 2190 PPC KVM_REG_PPC_SIER3 64
2365 PPC KVM_REG_PPC_PMC1 32 2191 PPC KVM_REG_PPC_PMC1 32
2366 PPC KVM_REG_PPC_PMC2 32 2192 PPC KVM_REG_PPC_PMC2 32
2367 PPC KVM_REG_PPC_PMC3 32 2193 PPC KVM_REG_PPC_PMC3 32
2368 PPC KVM_REG_PPC_PMC4 32 2194 PPC KVM_REG_PPC_PMC4 32
2369 PPC KVM_REG_PPC_PMC5 32 2195 PPC KVM_REG_PPC_PMC5 32
2370 PPC KVM_REG_PPC_PMC6 32 2196 PPC KVM_REG_PPC_PMC6 32
2371 PPC KVM_REG_PPC_PMC7 32 2197 PPC KVM_REG_PPC_PMC7 32
2372 PPC KVM_REG_PPC_PMC8 32 2198 PPC KVM_REG_PPC_PMC8 32
2373 PPC KVM_REG_PPC_FPR0 64 2199 PPC KVM_REG_PPC_FPR0 64
2374 ... 2200 ...
2375 PPC KVM_REG_PPC_FPR31 64 2201 PPC KVM_REG_PPC_FPR31 64
2376 PPC KVM_REG_PPC_VR0 128 2202 PPC KVM_REG_PPC_VR0 128
2377 ... 2203 ...
2378 PPC KVM_REG_PPC_VR31 128 2204 PPC KVM_REG_PPC_VR31 128
2379 PPC KVM_REG_PPC_VSR0 128 2205 PPC KVM_REG_PPC_VSR0 128
2380 ... 2206 ...
2381 PPC KVM_REG_PPC_VSR31 128 2207 PPC KVM_REG_PPC_VSR31 128
2382 PPC KVM_REG_PPC_FPSCR 64 2208 PPC KVM_REG_PPC_FPSCR 64
2383 PPC KVM_REG_PPC_VSCR 32 2209 PPC KVM_REG_PPC_VSCR 32
2384 PPC KVM_REG_PPC_VPA_ADDR 64 2210 PPC KVM_REG_PPC_VPA_ADDR 64
2385 PPC KVM_REG_PPC_VPA_SLB 128 2211 PPC KVM_REG_PPC_VPA_SLB 128
2386 PPC KVM_REG_PPC_VPA_DTL 128 2212 PPC KVM_REG_PPC_VPA_DTL 128
2387 PPC KVM_REG_PPC_EPCR 32 2213 PPC KVM_REG_PPC_EPCR 32
2388 PPC KVM_REG_PPC_EPR 32 2214 PPC KVM_REG_PPC_EPR 32
2389 PPC KVM_REG_PPC_TCR 32 2215 PPC KVM_REG_PPC_TCR 32
2390 PPC KVM_REG_PPC_TSR 32 2216 PPC KVM_REG_PPC_TSR 32
2391 PPC KVM_REG_PPC_OR_TSR 32 2217 PPC KVM_REG_PPC_OR_TSR 32
2392 PPC KVM_REG_PPC_CLEAR_TSR 32 2218 PPC KVM_REG_PPC_CLEAR_TSR 32
2393 PPC KVM_REG_PPC_MAS0 32 2219 PPC KVM_REG_PPC_MAS0 32
2394 PPC KVM_REG_PPC_MAS1 32 2220 PPC KVM_REG_PPC_MAS1 32
2395 PPC KVM_REG_PPC_MAS2 64 2221 PPC KVM_REG_PPC_MAS2 64
2396 PPC KVM_REG_PPC_MAS7_3 64 2222 PPC KVM_REG_PPC_MAS7_3 64
2397 PPC KVM_REG_PPC_MAS4 32 2223 PPC KVM_REG_PPC_MAS4 32
2398 PPC KVM_REG_PPC_MAS6 32 2224 PPC KVM_REG_PPC_MAS6 32
2399 PPC KVM_REG_PPC_MMUCFG 32 2225 PPC KVM_REG_PPC_MMUCFG 32
2400 PPC KVM_REG_PPC_TLB0CFG 32 2226 PPC KVM_REG_PPC_TLB0CFG 32
2401 PPC KVM_REG_PPC_TLB1CFG 32 2227 PPC KVM_REG_PPC_TLB1CFG 32
2402 PPC KVM_REG_PPC_TLB2CFG 32 2228 PPC KVM_REG_PPC_TLB2CFG 32
2403 PPC KVM_REG_PPC_TLB3CFG 32 2229 PPC KVM_REG_PPC_TLB3CFG 32
2404 PPC KVM_REG_PPC_TLB0PS 32 2230 PPC KVM_REG_PPC_TLB0PS 32
2405 PPC KVM_REG_PPC_TLB1PS 32 2231 PPC KVM_REG_PPC_TLB1PS 32
2406 PPC KVM_REG_PPC_TLB2PS 32 2232 PPC KVM_REG_PPC_TLB2PS 32
2407 PPC KVM_REG_PPC_TLB3PS 32 2233 PPC KVM_REG_PPC_TLB3PS 32
2408 PPC KVM_REG_PPC_EPTCFG 32 2234 PPC KVM_REG_PPC_EPTCFG 32
2409 PPC KVM_REG_PPC_ICP_STATE 64 2235 PPC KVM_REG_PPC_ICP_STATE 64
2410 PPC KVM_REG_PPC_VP_STATE 128 2236 PPC KVM_REG_PPC_VP_STATE 128
2411 PPC KVM_REG_PPC_TB_OFFSET 64 2237 PPC KVM_REG_PPC_TB_OFFSET 64
2412 PPC KVM_REG_PPC_SPMC1 32 2238 PPC KVM_REG_PPC_SPMC1 32
2413 PPC KVM_REG_PPC_SPMC2 32 2239 PPC KVM_REG_PPC_SPMC2 32
2414 PPC KVM_REG_PPC_IAMR 64 2240 PPC KVM_REG_PPC_IAMR 64
2415 PPC KVM_REG_PPC_TFHAR 64 2241 PPC KVM_REG_PPC_TFHAR 64
2416 PPC KVM_REG_PPC_TFIAR 64 2242 PPC KVM_REG_PPC_TFIAR 64
2417 PPC KVM_REG_PPC_TEXASR 64 2243 PPC KVM_REG_PPC_TEXASR 64
2418 PPC KVM_REG_PPC_FSCR 64 2244 PPC KVM_REG_PPC_FSCR 64
2419 PPC KVM_REG_PPC_PSPB 32 2245 PPC KVM_REG_PPC_PSPB 32
2420 PPC KVM_REG_PPC_EBBHR 64 2246 PPC KVM_REG_PPC_EBBHR 64
2421 PPC KVM_REG_PPC_EBBRR 64 2247 PPC KVM_REG_PPC_EBBRR 64
2422 PPC KVM_REG_PPC_BESCR 64 2248 PPC KVM_REG_PPC_BESCR 64
2423 PPC KVM_REG_PPC_TAR 64 2249 PPC KVM_REG_PPC_TAR 64
2424 PPC KVM_REG_PPC_DPDES 64 2250 PPC KVM_REG_PPC_DPDES 64
2425 PPC KVM_REG_PPC_DAWR 64 2251 PPC KVM_REG_PPC_DAWR 64
2426 PPC KVM_REG_PPC_DAWRX 64 2252 PPC KVM_REG_PPC_DAWRX 64
2427 PPC KVM_REG_PPC_CIABR 64 2253 PPC KVM_REG_PPC_CIABR 64
2428 PPC KVM_REG_PPC_IC 64 2254 PPC KVM_REG_PPC_IC 64
2429 PPC KVM_REG_PPC_VTB 64 2255 PPC KVM_REG_PPC_VTB 64
2430 PPC KVM_REG_PPC_CSIGR 64 2256 PPC KVM_REG_PPC_CSIGR 64
2431 PPC KVM_REG_PPC_TACR 64 2257 PPC KVM_REG_PPC_TACR 64
2432 PPC KVM_REG_PPC_TCSCR 64 2258 PPC KVM_REG_PPC_TCSCR 64
2433 PPC KVM_REG_PPC_PID 64 2259 PPC KVM_REG_PPC_PID 64
2434 PPC KVM_REG_PPC_ACOP 64 2260 PPC KVM_REG_PPC_ACOP 64
2435 PPC KVM_REG_PPC_VRSAVE 32 2261 PPC KVM_REG_PPC_VRSAVE 32
2436 PPC KVM_REG_PPC_LPCR 32 2262 PPC KVM_REG_PPC_LPCR 32
2437 PPC KVM_REG_PPC_LPCR_64 64 2263 PPC KVM_REG_PPC_LPCR_64 64
2438 PPC KVM_REG_PPC_PPR 64 2264 PPC KVM_REG_PPC_PPR 64
2439 PPC KVM_REG_PPC_ARCH_COMPAT 32 2265 PPC KVM_REG_PPC_ARCH_COMPAT 32
2440 PPC KVM_REG_PPC_DABRX 32 2266 PPC KVM_REG_PPC_DABRX 32
2441 PPC KVM_REG_PPC_WORT 64 2267 PPC KVM_REG_PPC_WORT 64
2442 PPC KVM_REG_PPC_SPRG9 64 2268 PPC KVM_REG_PPC_SPRG9 64
2443 PPC KVM_REG_PPC_DBSR 32 2269 PPC KVM_REG_PPC_DBSR 32
2444 PPC KVM_REG_PPC_TIDR 64 2270 PPC KVM_REG_PPC_TIDR 64
2445 PPC KVM_REG_PPC_PSSCR 64 2271 PPC KVM_REG_PPC_PSSCR 64
2446 PPC KVM_REG_PPC_DEC_EXPIRY 64 2272 PPC KVM_REG_PPC_DEC_EXPIRY 64
2447 PPC KVM_REG_PPC_PTCR 64 2273 PPC KVM_REG_PPC_PTCR 64
2448 PPC KVM_REG_PPC_HASHKEYR 64 <<
2449 PPC KVM_REG_PPC_HASHPKEYR 64 <<
2450 PPC KVM_REG_PPC_DAWR1 64 <<
2451 PPC KVM_REG_PPC_DAWRX1 64 <<
2452 PPC KVM_REG_PPC_DEXCR 64 <<
2453 PPC KVM_REG_PPC_TM_GPR0 64 2274 PPC KVM_REG_PPC_TM_GPR0 64
2454 ... 2275 ...
2455 PPC KVM_REG_PPC_TM_GPR31 64 2276 PPC KVM_REG_PPC_TM_GPR31 64
2456 PPC KVM_REG_PPC_TM_VSR0 128 2277 PPC KVM_REG_PPC_TM_VSR0 128
2457 ... 2278 ...
2458 PPC KVM_REG_PPC_TM_VSR63 128 2279 PPC KVM_REG_PPC_TM_VSR63 128
2459 PPC KVM_REG_PPC_TM_CR 64 2280 PPC KVM_REG_PPC_TM_CR 64
2460 PPC KVM_REG_PPC_TM_LR 64 2281 PPC KVM_REG_PPC_TM_LR 64
2461 PPC KVM_REG_PPC_TM_CTR 64 2282 PPC KVM_REG_PPC_TM_CTR 64
2462 PPC KVM_REG_PPC_TM_FPSCR 64 2283 PPC KVM_REG_PPC_TM_FPSCR 64
2463 PPC KVM_REG_PPC_TM_AMR 64 2284 PPC KVM_REG_PPC_TM_AMR 64
2464 PPC KVM_REG_PPC_TM_PPR 64 2285 PPC KVM_REG_PPC_TM_PPR 64
2465 PPC KVM_REG_PPC_TM_VRSAVE 64 2286 PPC KVM_REG_PPC_TM_VRSAVE 64
2466 PPC KVM_REG_PPC_TM_VSCR 32 2287 PPC KVM_REG_PPC_TM_VSCR 32
2467 PPC KVM_REG_PPC_TM_DSCR 64 2288 PPC KVM_REG_PPC_TM_DSCR 64
2468 PPC KVM_REG_PPC_TM_TAR 64 2289 PPC KVM_REG_PPC_TM_TAR 64
2469 PPC KVM_REG_PPC_TM_XER 64 2290 PPC KVM_REG_PPC_TM_XER 64
2470 2291
2471 MIPS KVM_REG_MIPS_R0 64 2292 MIPS KVM_REG_MIPS_R0 64
2472 ... 2293 ...
2473 MIPS KVM_REG_MIPS_R31 64 2294 MIPS KVM_REG_MIPS_R31 64
2474 MIPS KVM_REG_MIPS_HI 64 2295 MIPS KVM_REG_MIPS_HI 64
2475 MIPS KVM_REG_MIPS_LO 64 2296 MIPS KVM_REG_MIPS_LO 64
2476 MIPS KVM_REG_MIPS_PC 64 2297 MIPS KVM_REG_MIPS_PC 64
2477 MIPS KVM_REG_MIPS_CP0_INDEX 32 2298 MIPS KVM_REG_MIPS_CP0_INDEX 32
2478 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64 2299 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2479 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64 2300 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2480 MIPS KVM_REG_MIPS_CP0_CONTEXT 64 2301 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2481 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32 2302 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2482 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64 2303 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2483 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64 2304 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2484 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32 2305 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2485 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32 2306 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2486 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64 2307 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2487 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64 2308 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2488 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64 2309 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2489 MIPS KVM_REG_MIPS_CP0_PWBASE 64 2310 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2490 MIPS KVM_REG_MIPS_CP0_PWFIELD 64 2311 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2491 MIPS KVM_REG_MIPS_CP0_PWSIZE 64 2312 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2492 MIPS KVM_REG_MIPS_CP0_WIRED 32 2313 MIPS KVM_REG_MIPS_CP0_WIRED 32
2493 MIPS KVM_REG_MIPS_CP0_PWCTL 32 2314 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2494 MIPS KVM_REG_MIPS_CP0_HWRENA 32 2315 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2495 MIPS KVM_REG_MIPS_CP0_BADVADDR 64 2316 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2496 MIPS KVM_REG_MIPS_CP0_BADINSTR 32 2317 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2497 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32 2318 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2498 MIPS KVM_REG_MIPS_CP0_COUNT 32 2319 MIPS KVM_REG_MIPS_CP0_COUNT 32
2499 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64 2320 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2500 MIPS KVM_REG_MIPS_CP0_COMPARE 32 2321 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2501 MIPS KVM_REG_MIPS_CP0_STATUS 32 2322 MIPS KVM_REG_MIPS_CP0_STATUS 32
2502 MIPS KVM_REG_MIPS_CP0_INTCTL 32 2323 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2503 MIPS KVM_REG_MIPS_CP0_CAUSE 32 2324 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2504 MIPS KVM_REG_MIPS_CP0_EPC 64 2325 MIPS KVM_REG_MIPS_CP0_EPC 64
2505 MIPS KVM_REG_MIPS_CP0_PRID 32 2326 MIPS KVM_REG_MIPS_CP0_PRID 32
2506 MIPS KVM_REG_MIPS_CP0_EBASE 64 2327 MIPS KVM_REG_MIPS_CP0_EBASE 64
2507 MIPS KVM_REG_MIPS_CP0_CONFIG 32 2328 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2508 MIPS KVM_REG_MIPS_CP0_CONFIG1 32 2329 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2509 MIPS KVM_REG_MIPS_CP0_CONFIG2 32 2330 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2510 MIPS KVM_REG_MIPS_CP0_CONFIG3 32 2331 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2511 MIPS KVM_REG_MIPS_CP0_CONFIG4 32 2332 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2512 MIPS KVM_REG_MIPS_CP0_CONFIG5 32 2333 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2513 MIPS KVM_REG_MIPS_CP0_CONFIG7 32 2334 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2514 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64 2335 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2515 MIPS KVM_REG_MIPS_CP0_ERROREPC 64 2336 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2516 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64 2337 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2517 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64 2338 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2518 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64 2339 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2519 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64 2340 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2520 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64 2341 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2521 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64 2342 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2522 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64 2343 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2523 MIPS KVM_REG_MIPS_COUNT_CTL 64 2344 MIPS KVM_REG_MIPS_COUNT_CTL 64
2524 MIPS KVM_REG_MIPS_COUNT_RESUME 64 2345 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2525 MIPS KVM_REG_MIPS_COUNT_HZ 64 2346 MIPS KVM_REG_MIPS_COUNT_HZ 64
2526 MIPS KVM_REG_MIPS_FPR_32(0..31) 32 2347 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2527 MIPS KVM_REG_MIPS_FPR_64(0..31) 64 2348 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2528 MIPS KVM_REG_MIPS_VEC_128(0..31) 128 2349 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2529 MIPS KVM_REG_MIPS_FCR_IR 32 2350 MIPS KVM_REG_MIPS_FCR_IR 32
2530 MIPS KVM_REG_MIPS_FCR_CSR 32 2351 MIPS KVM_REG_MIPS_FCR_CSR 32
2531 MIPS KVM_REG_MIPS_MSA_IR 32 2352 MIPS KVM_REG_MIPS_MSA_IR 32
2532 MIPS KVM_REG_MIPS_MSA_CSR 32 2353 MIPS KVM_REG_MIPS_MSA_CSR 32
2533 ======= =============================== === 2354 ======= =============================== ============
2534 2355
2535 ARM registers are mapped using the lower 32 b 2356 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2536 is the register group type, or coprocessor nu 2357 is the register group type, or coprocessor number:
2537 2358
2538 ARM core registers have the following id bit 2359 ARM core registers have the following id bit patterns::
2539 2360
2540 0x4020 0000 0010 <index into the kvm_regs s 2361 0x4020 0000 0010 <index into the kvm_regs struct:16>
2541 2362
2542 ARM 32-bit CP15 registers have the following 2363 ARM 32-bit CP15 registers have the following id bit patterns::
2543 2364
2544 0x4020 0000 000F <zero:1> <crn:4> <crm:4> < 2365 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2545 2366
2546 ARM 64-bit CP15 registers have the following 2367 ARM 64-bit CP15 registers have the following id bit patterns::
2547 2368
2548 0x4030 0000 000F <zero:1> <zero:4> <crm:4> 2369 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2549 2370
2550 ARM CCSIDR registers are demultiplexed by CSS 2371 ARM CCSIDR registers are demultiplexed by CSSELR value::
2551 2372
2552 0x4020 0000 0011 00 <csselr:8> 2373 0x4020 0000 0011 00 <csselr:8>
2553 2374
2554 ARM 32-bit VFP control registers have the fol 2375 ARM 32-bit VFP control registers have the following id bit patterns::
2555 2376
2556 0x4020 0000 0012 1 <regno:12> 2377 0x4020 0000 0012 1 <regno:12>
2557 2378
2558 ARM 64-bit FP registers have the following id 2379 ARM 64-bit FP registers have the following id bit patterns::
2559 2380
2560 0x4030 0000 0012 0 <regno:12> 2381 0x4030 0000 0012 0 <regno:12>
2561 2382
2562 ARM firmware pseudo-registers have the follow 2383 ARM firmware pseudo-registers have the following bit pattern::
2563 2384
2564 0x4030 0000 0014 <regno:16> 2385 0x4030 0000 0014 <regno:16>
2565 2386
2566 2387
2567 arm64 registers are mapped using the lower 32 2388 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2568 that is the register group type, or coprocess 2389 that is the register group type, or coprocessor number:
2569 2390
2570 arm64 core/FP-SIMD registers have the followi 2391 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2571 that the size of the access is variable, as t 2392 that the size of the access is variable, as the kvm_regs structure
2572 contains elements ranging from 32 to 128 bits 2393 contains elements ranging from 32 to 128 bits. The index is a 32bit
2573 value in the kvm_regs structure seen as a 32b 2394 value in the kvm_regs structure seen as a 32bit array::
2574 2395
2575 0x60x0 0000 0010 <index into the kvm_regs s 2396 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2576 2397
2577 Specifically: 2398 Specifically:
2578 2399
2579 ======================= ========= ===== ===== 2400 ======================= ========= ===== =======================================
2580 Encoding Register Bits kvm_r 2401 Encoding Register Bits kvm_regs member
2581 ======================= ========= ===== ===== 2402 ======================= ========= ===== =======================================
2582 0x6030 0000 0010 0000 X0 64 regs. 2403 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2583 0x6030 0000 0010 0002 X1 64 regs. 2404 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2584 ... 2405 ...
2585 0x6030 0000 0010 003c X30 64 regs. 2406 0x6030 0000 0010 003c X30 64 regs.regs[30]
2586 0x6030 0000 0010 003e SP 64 regs. 2407 0x6030 0000 0010 003e SP 64 regs.sp
2587 0x6030 0000 0010 0040 PC 64 regs. 2408 0x6030 0000 0010 0040 PC 64 regs.pc
2588 0x6030 0000 0010 0042 PSTATE 64 regs. 2409 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2589 0x6030 0000 0010 0044 SP_EL1 64 sp_el 2410 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2590 0x6030 0000 0010 0046 ELR_EL1 64 elr_e 2411 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2591 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[ 2412 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2592 0x6030 0000 0010 004a SPSR_ABT 64 spsr[ 2413 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2593 0x6030 0000 0010 004c SPSR_UND 64 spsr[ 2414 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2594 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[ 2415 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2595 0x6030 0000 0010 0050 SPSR_FIQ 64 spsr[ !! 2416 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2596 0x6040 0000 0010 0054 V0 128 fp_re 2417 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2597 0x6040 0000 0010 0058 V1 128 fp_re 2418 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2598 ... 2419 ...
2599 0x6040 0000 0010 00d0 V31 128 fp_re 2420 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2600 0x6020 0000 0010 00d4 FPSR 32 fp_re 2421 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2601 0x6020 0000 0010 00d5 FPCR 32 fp_re 2422 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2602 ======================= ========= ===== ===== 2423 ======================= ========= ===== =======================================
2603 2424
2604 .. [1] These encodings are not accepted for S 2425 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2605 KVM_ARM_VCPU_INIT. 2426 KVM_ARM_VCPU_INIT.
2606 2427
2607 The equivalent register content can be 2428 The equivalent register content can be accessed via bits [127:0] of
2608 the corresponding SVE Zn registers ins 2429 the corresponding SVE Zn registers instead for vcpus that have SVE
2609 enabled (see below). 2430 enabled (see below).
2610 2431
2611 arm64 CCSIDR registers are demultiplexed by C 2432 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2612 2433
2613 0x6020 0000 0011 00 <csselr:8> 2434 0x6020 0000 0011 00 <csselr:8>
2614 2435
2615 arm64 system registers have the following id 2436 arm64 system registers have the following id bit patterns::
2616 2437
2617 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <c 2438 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2618 2439
2619 .. warning:: 2440 .. warning::
2620 2441
2621 Two system register IDs do not follow th 2442 Two system register IDs do not follow the specified pattern. These
2622 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_A 2443 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2623 system registers CNTV_CVAL_EL0 and CNTVC 2444 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2624 two had their values accidentally swappe 2445 two had their values accidentally swapped, which means TIMER_CVAL is
2625 derived from the register encoding for C 2446 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2626 derived from the register encoding for C 2447 derived from the register encoding for CNTV_CVAL_EL0. As this is
2627 API, it must remain this way. 2448 API, it must remain this way.
2628 2449
2629 arm64 firmware pseudo-registers have the foll 2450 arm64 firmware pseudo-registers have the following bit pattern::
2630 2451
2631 0x6030 0000 0014 <regno:16> 2452 0x6030 0000 0014 <regno:16>
2632 2453
2633 arm64 SVE registers have the following bit pa 2454 arm64 SVE registers have the following bit patterns::
2634 2455
2635 0x6080 0000 0015 00 <n:5> <slice:5> Zn bi 2456 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2636 0x6050 0000 0015 04 <n:4> <slice:5> Pn bi 2457 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2637 0x6050 0000 0015 060 <slice:5> FFR b 2458 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2638 0x6060 0000 0015 ffff KVM_R 2459 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2639 2460
2640 Access to register IDs where 2048 * slice >= 2461 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2641 ENOENT. max_vq is the vcpu's maximum support 2462 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2642 quadwords: see [2]_ below. 2463 quadwords: see [2]_ below.
2643 2464
2644 These registers are only accessible on vcpus 2465 These registers are only accessible on vcpus for which SVE is enabled.
2645 See KVM_ARM_VCPU_INIT for details. 2466 See KVM_ARM_VCPU_INIT for details.
2646 2467
2647 In addition, except for KVM_REG_ARM64_SVE_VLS 2468 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2648 accessible until the vcpu's SVE configuration 2469 accessible until the vcpu's SVE configuration has been finalized
2649 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE) 2470 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2650 and KVM_ARM_VCPU_FINALIZE for more informatio 2471 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2651 2472
2652 KVM_REG_ARM64_SVE_VLS is a pseudo-register th 2473 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2653 lengths supported by the vcpu to be discovere 2474 lengths supported by the vcpu to be discovered and configured by
2654 userspace. When transferred to or from user 2475 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2655 or KVM_SET_ONE_REG, the value of this registe 2476 or KVM_SET_ONE_REG, the value of this register is of type
2656 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes t 2477 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2657 follows:: 2478 follows::
2658 2479
2659 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORD 2480 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2660 2481
2661 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX && 2482 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2662 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ 2483 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2663 ((vq - KVM_ARM64_SVE_VQ_MIN) 2484 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2664 /* Vector length vq * 16 bytes suppor 2485 /* Vector length vq * 16 bytes supported */
2665 else 2486 else
2666 /* Vector length vq * 16 bytes not su 2487 /* Vector length vq * 16 bytes not supported */
2667 2488
2668 .. [2] The maximum value vq for which the abo 2489 .. [2] The maximum value vq for which the above condition is true is
2669 max_vq. This is the maximum vector le 2490 max_vq. This is the maximum vector length available to the guest on
2670 this vcpu, and determines which regist 2491 this vcpu, and determines which register slices are visible through
2671 this ioctl interface. 2492 this ioctl interface.
2672 2493
2673 (See Documentation/arch/arm64/sve.rst for an !! 2494 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2674 nomenclature.) 2495 nomenclature.)
2675 2496
2676 KVM_REG_ARM64_SVE_VLS is only accessible afte 2497 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2677 KVM_ARM_VCPU_INIT initialises it to the best 2498 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2678 the host supports. 2499 the host supports.
2679 2500
2680 Userspace may subsequently modify it if desir 2501 Userspace may subsequently modify it if desired until the vcpu's SVE
2681 configuration is finalized using KVM_ARM_VCPU 2502 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2682 2503
2683 Apart from simply removing all vector lengths 2504 Apart from simply removing all vector lengths from the host set that
2684 exceed some value, support for arbitrarily ch 2505 exceed some value, support for arbitrarily chosen sets of vector lengths
2685 is hardware-dependent and may not be availabl 2506 is hardware-dependent and may not be available. Attempting to configure
2686 an invalid set of vector lengths via KVM_SET_ 2507 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2687 EINVAL. 2508 EINVAL.
2688 2509
2689 After the vcpu's SVE configuration is finaliz 2510 After the vcpu's SVE configuration is finalized, further attempts to
2690 write this register will fail with EPERM. 2511 write this register will fail with EPERM.
2691 2512
2692 arm64 bitmap feature firmware pseudo-register <<
2693 <<
2694 0x6030 0000 0016 <regno:16> <<
2695 <<
2696 The bitmap feature firmware registers exposes <<
2697 are available for userspace to configure. The <<
2698 services that are available for the guests to <<
2699 sets all the supported bits during VM initial <<
2700 discover the available services via KVM_GET_O <<
2701 bitmap corresponding to the features that it <<
2702 KVM_SET_ONE_REG. <<
2703 <<
2704 Note: These registers are immutable once any <<
2705 run at least once. A KVM_SET_ONE_REG in such <<
2706 a -EBUSY to userspace. <<
2707 <<
2708 (See Documentation/virt/kvm/arm/hypercalls.rs <<
2709 <<
2710 2513
2711 MIPS registers are mapped using the lower 32 2514 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2712 the register group type: 2515 the register group type:
2713 2516
2714 MIPS core registers (see above) have the foll 2517 MIPS core registers (see above) have the following id bit patterns::
2715 2518
2716 0x7030 0000 0000 <reg:16> 2519 0x7030 0000 0000 <reg:16>
2717 2520
2718 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* ab 2521 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2719 patterns depending on whether they're 32-bit 2522 patterns depending on whether they're 32-bit or 64-bit registers::
2720 2523
2721 0x7020 0000 0001 00 <reg:5> <sel:3> (32-b 2524 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2722 0x7030 0000 0001 00 <reg:5> <sel:3> (64-b 2525 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2723 2526
2724 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_M 2527 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2725 versions of the EntryLo registers regardless 2528 versions of the EntryLo registers regardless of the word size of the host
2726 hardware, host kernel, guest, and whether XPA 2529 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2727 with the RI and XI bits (if they exist) in bi 2530 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2728 the PFNX field starting at bit 30. 2531 the PFNX field starting at bit 30.
2729 2532
2730 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) abov 2533 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2731 patterns:: 2534 patterns::
2732 2535
2733 0x7030 0000 0001 01 <reg:8> 2536 0x7030 0000 0001 01 <reg:8>
2734 2537
2735 MIPS KVM control registers (see above) have t 2538 MIPS KVM control registers (see above) have the following id bit patterns::
2736 2539
2737 0x7030 0000 0002 <reg:16> 2540 0x7030 0000 0002 <reg:16>
2738 2541
2739 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32, 2542 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2740 id bit patterns depending on the size of the 2543 id bit patterns depending on the size of the register being accessed. They are
2741 always accessed according to the current gues 2544 always accessed according to the current guest FPU mode (Status.FR and
2742 Config5.FRE), i.e. as the guest would see the 2545 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2743 if the guest FPU mode is changed. MIPS SIMD A 2546 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2744 registers (see KVM_REG_MIPS_VEC_128() above) 2547 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2745 overlap the FPU registers:: 2548 overlap the FPU registers::
2746 2549
2747 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit F 2550 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2748 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit F 2551 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2749 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit 2552 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2750 2553
2751 MIPS FPU control registers (see KVM_REG_MIPS_ 2554 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2752 following id bit patterns:: 2555 following id bit patterns::
2753 2556
2754 0x7020 0000 0003 01 <0:3> <reg:5> 2557 0x7020 0000 0003 01 <0:3> <reg:5>
2755 2558
2756 MIPS MSA control registers (see KVM_REG_MIPS_ 2559 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2757 following id bit patterns:: 2560 following id bit patterns::
2758 2561
2759 0x7020 0000 0003 02 <0:3> <reg:5> 2562 0x7020 0000 0003 02 <0:3> <reg:5>
2760 2563
2761 RISC-V registers are mapped using the lower 3 <<
2762 that is the register group type. <<
2763 <<
2764 RISC-V config registers are meant for configu <<
2765 the following id bit patterns:: <<
2766 <<
2767 0x8020 0000 01 <index into the kvm_riscv_co <<
2768 0x8030 0000 01 <index into the kvm_riscv_co <<
2769 <<
2770 Following are the RISC-V config registers: <<
2771 <<
2772 ======================= ========= =========== <<
2773 Encoding Register Description <<
2774 ======================= ========= =========== <<
2775 0x80x0 0000 0100 0000 isa ISA feature <<
2776 ======================= ========= =========== <<
2777 <<
2778 The isa config register can be read anytime b <<
2779 a Guest VCPU runs. It will have ISA feature b <<
2780 set by default. <<
2781 <<
2782 RISC-V core registers represent the general e <<
2783 and it has the following id bit patterns:: <<
2784 <<
2785 0x8020 0000 02 <index into the kvm_riscv_co <<
2786 0x8030 0000 02 <index into the kvm_riscv_co <<
2787 <<
2788 Following are the RISC-V core registers: <<
2789 <<
2790 ======================= ========= =========== <<
2791 Encoding Register Description <<
2792 ======================= ========= =========== <<
2793 0x80x0 0000 0200 0000 regs.pc Program cou <<
2794 0x80x0 0000 0200 0001 regs.ra Return addr <<
2795 0x80x0 0000 0200 0002 regs.sp Stack point <<
2796 0x80x0 0000 0200 0003 regs.gp Global poin <<
2797 0x80x0 0000 0200 0004 regs.tp Task pointe <<
2798 0x80x0 0000 0200 0005 regs.t0 Caller save <<
2799 0x80x0 0000 0200 0006 regs.t1 Caller save <<
2800 0x80x0 0000 0200 0007 regs.t2 Caller save <<
2801 0x80x0 0000 0200 0008 regs.s0 Callee save <<
2802 0x80x0 0000 0200 0009 regs.s1 Callee save <<
2803 0x80x0 0000 0200 000a regs.a0 Function ar <<
2804 0x80x0 0000 0200 000b regs.a1 Function ar <<
2805 0x80x0 0000 0200 000c regs.a2 Function ar <<
2806 0x80x0 0000 0200 000d regs.a3 Function ar <<
2807 0x80x0 0000 0200 000e regs.a4 Function ar <<
2808 0x80x0 0000 0200 000f regs.a5 Function ar <<
2809 0x80x0 0000 0200 0010 regs.a6 Function ar <<
2810 0x80x0 0000 0200 0011 regs.a7 Function ar <<
2811 0x80x0 0000 0200 0012 regs.s2 Callee save <<
2812 0x80x0 0000 0200 0013 regs.s3 Callee save <<
2813 0x80x0 0000 0200 0014 regs.s4 Callee save <<
2814 0x80x0 0000 0200 0015 regs.s5 Callee save <<
2815 0x80x0 0000 0200 0016 regs.s6 Callee save <<
2816 0x80x0 0000 0200 0017 regs.s7 Callee save <<
2817 0x80x0 0000 0200 0018 regs.s8 Callee save <<
2818 0x80x0 0000 0200 0019 regs.s9 Callee save <<
2819 0x80x0 0000 0200 001a regs.s10 Callee save <<
2820 0x80x0 0000 0200 001b regs.s11 Callee save <<
2821 0x80x0 0000 0200 001c regs.t3 Caller save <<
2822 0x80x0 0000 0200 001d regs.t4 Caller save <<
2823 0x80x0 0000 0200 001e regs.t5 Caller save <<
2824 0x80x0 0000 0200 001f regs.t6 Caller save <<
2825 0x80x0 0000 0200 0020 mode Privilege m <<
2826 ======================= ========= =========== <<
2827 <<
2828 RISC-V csr registers represent the supervisor <<
2829 of a Guest VCPU and it has the following id b <<
2830 <<
2831 0x8020 0000 03 <index into the kvm_riscv_cs <<
2832 0x8030 0000 03 <index into the kvm_riscv_cs <<
2833 <<
2834 Following are the RISC-V csr registers: <<
2835 <<
2836 ======================= ========= =========== <<
2837 Encoding Register Description <<
2838 ======================= ========= =========== <<
2839 0x80x0 0000 0300 0000 sstatus Supervisor <<
2840 0x80x0 0000 0300 0001 sie Supervisor <<
2841 0x80x0 0000 0300 0002 stvec Supervisor <<
2842 0x80x0 0000 0300 0003 sscratch Supervisor <<
2843 0x80x0 0000 0300 0004 sepc Supervisor <<
2844 0x80x0 0000 0300 0005 scause Supervisor <<
2845 0x80x0 0000 0300 0006 stval Supervisor <<
2846 0x80x0 0000 0300 0007 sip Supervisor <<
2847 0x80x0 0000 0300 0008 satp Supervisor <<
2848 ======================= ========= =========== <<
2849 <<
2850 RISC-V timer registers represent the timer st <<
2851 the following id bit patterns:: <<
2852 <<
2853 0x8030 0000 04 <index into the kvm_riscv_ti <<
2854 <<
2855 Following are the RISC-V timer registers: <<
2856 <<
2857 ======================= ========= =========== <<
2858 Encoding Register Description <<
2859 ======================= ========= =========== <<
2860 0x8030 0000 0400 0000 frequency Time base f <<
2861 0x8030 0000 0400 0001 time Time value <<
2862 0x8030 0000 0400 0002 compare Time compar <<
2863 0x8030 0000 0400 0003 state Time compar <<
2864 ======================= ========= =========== <<
2865 <<
2866 RISC-V F-extension registers represent the si <<
2867 state of a Guest VCPU and it has the followin <<
2868 <<
2869 0x8020 0000 05 <index into the __riscv_f_ex <<
2870 <<
2871 Following are the RISC-V F-extension register <<
2872 <<
2873 ======================= ========= =========== <<
2874 Encoding Register Description <<
2875 ======================= ========= =========== <<
2876 0x8020 0000 0500 0000 f[0] Floating po <<
2877 ... <<
2878 0x8020 0000 0500 001f f[31] Floating po <<
2879 0x8020 0000 0500 0020 fcsr Floating po <<
2880 ======================= ========= =========== <<
2881 <<
2882 RISC-V D-extension registers represent the do <<
2883 state of a Guest VCPU and it has the followin <<
2884 <<
2885 0x8020 0000 06 <index into the __riscv_d_ex <<
2886 0x8030 0000 06 <index into the __riscv_d_ex <<
2887 <<
2888 Following are the RISC-V D-extension register <<
2889 <<
2890 ======================= ========= =========== <<
2891 Encoding Register Description <<
2892 ======================= ========= =========== <<
2893 0x8030 0000 0600 0000 f[0] Floating po <<
2894 ... <<
2895 0x8030 0000 0600 001f f[31] Floating po <<
2896 0x8020 0000 0600 0020 fcsr Floating po <<
2897 ======================= ========= =========== <<
2898 <<
2899 LoongArch registers are mapped using the lowe <<
2900 that is the register group type. <<
2901 <<
2902 LoongArch csr registers are used to control g <<
2903 cpu, and they have the following id bit patte <<
2904 <<
2905 0x9030 0000 0001 00 <reg:5> <sel:3> (64-b <<
2906 <<
2907 LoongArch KVM control registers are used to i <<
2908 such as set vcpu counter or reset vcpu, and t <<
2909 <<
2910 0x9030 0000 0002 <reg:16> <<
2911 <<
2912 2564
2913 4.69 KVM_GET_ONE_REG 2565 4.69 KVM_GET_ONE_REG
2914 -------------------- 2566 --------------------
2915 2567
2916 :Capability: KVM_CAP_ONE_REG 2568 :Capability: KVM_CAP_ONE_REG
2917 :Architectures: all 2569 :Architectures: all
2918 :Type: vcpu ioctl 2570 :Type: vcpu ioctl
2919 :Parameters: struct kvm_one_reg (in and out) 2571 :Parameters: struct kvm_one_reg (in and out)
2920 :Returns: 0 on success, negative value on fai 2572 :Returns: 0 on success, negative value on failure
2921 2573
2922 Errors include: 2574 Errors include:
2923 2575
2924 ======== ================================== 2576 ======== ============================================================
2925 ENOENT no such register !! 2577 Â ENOENT Â Â no such register
2926 EINVAL invalid register ID, or no such re !! 2578 Â EINVAL Â Â invalid register ID, or no such register or used with VMs in
2927 protected virtualization mode on s 2579 protected virtualization mode on s390
2928 EPERM (arm64) register access not allowe !! 2580 Â EPERM Â Â Â (arm64) register access not allowed before vcpu finalization
2929 ======== ================================== 2581 ======== ============================================================
2930 2582
2931 (These error codes are indicative only: do no 2583 (These error codes are indicative only: do not rely on a specific error
2932 code being returned in a specific situation.) 2584 code being returned in a specific situation.)
2933 2585
2934 This ioctl allows to receive the value of a s 2586 This ioctl allows to receive the value of a single register implemented
2935 in a vcpu. The register to read is indicated 2587 in a vcpu. The register to read is indicated by the "id" field of the
2936 kvm_one_reg struct passed in. On success, the 2588 kvm_one_reg struct passed in. On success, the register value can be found
2937 at the memory location pointed to by "addr". 2589 at the memory location pointed to by "addr".
2938 2590
2939 The list of registers accessible using this i 2591 The list of registers accessible using this interface is identical to the
2940 list in 4.68. 2592 list in 4.68.
2941 2593
2942 2594
2943 4.70 KVM_KVMCLOCK_CTRL 2595 4.70 KVM_KVMCLOCK_CTRL
2944 ---------------------- 2596 ----------------------
2945 2597
2946 :Capability: KVM_CAP_KVMCLOCK_CTRL 2598 :Capability: KVM_CAP_KVMCLOCK_CTRL
2947 :Architectures: Any that implement pvclocks ( 2599 :Architectures: Any that implement pvclocks (currently x86 only)
2948 :Type: vcpu ioctl 2600 :Type: vcpu ioctl
2949 :Parameters: None 2601 :Parameters: None
2950 :Returns: 0 on success, -1 on error 2602 :Returns: 0 on success, -1 on error
2951 2603
2952 This ioctl sets a flag accessible to the gues 2604 This ioctl sets a flag accessible to the guest indicating that the specified
2953 vCPU has been paused by the host userspace. 2605 vCPU has been paused by the host userspace.
2954 2606
2955 The host will set a flag in the pvclock struc 2607 The host will set a flag in the pvclock structure that is checked from the
2956 soft lockup watchdog. The flag is part of th 2608 soft lockup watchdog. The flag is part of the pvclock structure that is
2957 shared between guest and host, specifically t 2609 shared between guest and host, specifically the second bit of the flags
2958 field of the pvclock_vcpu_time_info structure 2610 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2959 the host and read/cleared exclusively by the 2611 the host and read/cleared exclusively by the guest. The guest operation of
2960 checking and clearing the flag must be an ato 2612 checking and clearing the flag must be an atomic operation so
2961 load-link/store-conditional, or equivalent mu 2613 load-link/store-conditional, or equivalent must be used. There are two cases
2962 where the guest will clear the flag: when the 2614 where the guest will clear the flag: when the soft lockup watchdog timer resets
2963 itself or when a soft lockup is detected. Th 2615 itself or when a soft lockup is detected. This ioctl can be called any time
2964 after pausing the vcpu, but before it is resu 2616 after pausing the vcpu, but before it is resumed.
2965 2617
2966 2618
2967 4.71 KVM_SIGNAL_MSI 2619 4.71 KVM_SIGNAL_MSI
2968 ------------------- 2620 -------------------
2969 2621
2970 :Capability: KVM_CAP_SIGNAL_MSI 2622 :Capability: KVM_CAP_SIGNAL_MSI
2971 :Architectures: x86 arm64 !! 2623 :Architectures: x86 arm arm64
2972 :Type: vm ioctl 2624 :Type: vm ioctl
2973 :Parameters: struct kvm_msi (in) 2625 :Parameters: struct kvm_msi (in)
2974 :Returns: >0 on delivery, 0 if guest blocked 2626 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2975 2627
2976 Directly inject a MSI message. Only valid wit 2628 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2977 MSI messages. 2629 MSI messages.
2978 2630
2979 :: 2631 ::
2980 2632
2981 struct kvm_msi { 2633 struct kvm_msi {
2982 __u32 address_lo; 2634 __u32 address_lo;
2983 __u32 address_hi; 2635 __u32 address_hi;
2984 __u32 data; 2636 __u32 data;
2985 __u32 flags; 2637 __u32 flags;
2986 __u32 devid; 2638 __u32 devid;
2987 __u8 pad[12]; 2639 __u8 pad[12];
2988 }; 2640 };
2989 2641
2990 flags: 2642 flags:
2991 KVM_MSI_VALID_DEVID: devid contains a valid 2643 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2992 KVM_CAP_MSI_DEVID capability advertises the 2644 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2993 the device ID. If this capability is not a 2645 the device ID. If this capability is not available, userspace
2994 should never set the KVM_MSI_VALID_DEVID fl 2646 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2995 2647
2996 If KVM_MSI_VALID_DEVID is set, devid contains 2648 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2997 for the device that wrote the MSI message. F 2649 for the device that wrote the MSI message. For PCI, this is usually a
2998 BDF identifier in the lower 16 bits. !! 2650 BFD identifier in the lower 16 bits.
2999 2651
3000 On x86, address_hi is ignored unless the KVM_ 2652 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
3001 feature of KVM_CAP_X2APIC_API capability is e 2653 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
3002 address_hi bits 31-8 provide bits 31-8 of the 2654 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
3003 address_hi must be zero. 2655 address_hi must be zero.
3004 2656
3005 2657
3006 4.71 KVM_CREATE_PIT2 2658 4.71 KVM_CREATE_PIT2
3007 -------------------- 2659 --------------------
3008 2660
3009 :Capability: KVM_CAP_PIT2 2661 :Capability: KVM_CAP_PIT2
3010 :Architectures: x86 2662 :Architectures: x86
3011 :Type: vm ioctl 2663 :Type: vm ioctl
3012 :Parameters: struct kvm_pit_config (in) 2664 :Parameters: struct kvm_pit_config (in)
3013 :Returns: 0 on success, -1 on error 2665 :Returns: 0 on success, -1 on error
3014 2666
3015 Creates an in-kernel device model for the i82 2667 Creates an in-kernel device model for the i8254 PIT. This call is only valid
3016 after enabling in-kernel irqchip support via 2668 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
3017 parameters have to be passed:: 2669 parameters have to be passed::
3018 2670
3019 struct kvm_pit_config { 2671 struct kvm_pit_config {
3020 __u32 flags; 2672 __u32 flags;
3021 __u32 pad[15]; 2673 __u32 pad[15];
3022 }; 2674 };
3023 2675
3024 Valid flags are:: 2676 Valid flags are::
3025 2677
3026 #define KVM_PIT_SPEAKER_DUMMY 1 /* emul 2678 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
3027 2679
3028 PIT timer interrupts may use a per-VM kernel 2680 PIT timer interrupts may use a per-VM kernel thread for injection. If it
3029 exists, this thread will have a name of the f 2681 exists, this thread will have a name of the following pattern::
3030 2682
3031 kvm-pit/<owner-process-pid> 2683 kvm-pit/<owner-process-pid>
3032 2684
3033 When running a guest with elevated priorities 2685 When running a guest with elevated priorities, the scheduling parameters of
3034 this thread may have to be adjusted according 2686 this thread may have to be adjusted accordingly.
3035 2687
3036 This IOCTL replaces the obsolete KVM_CREATE_P 2688 This IOCTL replaces the obsolete KVM_CREATE_PIT.
3037 2689
3038 2690
3039 4.72 KVM_GET_PIT2 2691 4.72 KVM_GET_PIT2
3040 ----------------- 2692 -----------------
3041 2693
3042 :Capability: KVM_CAP_PIT_STATE2 2694 :Capability: KVM_CAP_PIT_STATE2
3043 :Architectures: x86 2695 :Architectures: x86
3044 :Type: vm ioctl 2696 :Type: vm ioctl
3045 :Parameters: struct kvm_pit_state2 (out) 2697 :Parameters: struct kvm_pit_state2 (out)
3046 :Returns: 0 on success, -1 on error 2698 :Returns: 0 on success, -1 on error
3047 2699
3048 Retrieves the state of the in-kernel PIT mode 2700 Retrieves the state of the in-kernel PIT model. Only valid after
3049 KVM_CREATE_PIT2. The state is returned in the 2701 KVM_CREATE_PIT2. The state is returned in the following structure::
3050 2702
3051 struct kvm_pit_state2 { 2703 struct kvm_pit_state2 {
3052 struct kvm_pit_channel_state channels 2704 struct kvm_pit_channel_state channels[3];
3053 __u32 flags; 2705 __u32 flags;
3054 __u32 reserved[9]; 2706 __u32 reserved[9];
3055 }; 2707 };
3056 2708
3057 Valid flags are:: 2709 Valid flags are::
3058 2710
3059 /* disable PIT in HPET legacy mode */ 2711 /* disable PIT in HPET legacy mode */
3060 #define KVM_PIT_FLAGS_HPET_LEGACY 0x000 !! 2712 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
3061 /* speaker port data bit enabled */ <<
3062 #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x000 <<
3063 2713
3064 This IOCTL replaces the obsolete KVM_GET_PIT. 2714 This IOCTL replaces the obsolete KVM_GET_PIT.
3065 2715
3066 2716
3067 4.73 KVM_SET_PIT2 2717 4.73 KVM_SET_PIT2
3068 ----------------- 2718 -----------------
3069 2719
3070 :Capability: KVM_CAP_PIT_STATE2 2720 :Capability: KVM_CAP_PIT_STATE2
3071 :Architectures: x86 2721 :Architectures: x86
3072 :Type: vm ioctl 2722 :Type: vm ioctl
3073 :Parameters: struct kvm_pit_state2 (in) 2723 :Parameters: struct kvm_pit_state2 (in)
3074 :Returns: 0 on success, -1 on error 2724 :Returns: 0 on success, -1 on error
3075 2725
3076 Sets the state of the in-kernel PIT model. On 2726 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
3077 See KVM_GET_PIT2 for details on struct kvm_pi 2727 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
3078 2728
3079 This IOCTL replaces the obsolete KVM_SET_PIT. 2729 This IOCTL replaces the obsolete KVM_SET_PIT.
3080 2730
3081 2731
3082 4.74 KVM_PPC_GET_SMMU_INFO 2732 4.74 KVM_PPC_GET_SMMU_INFO
3083 -------------------------- 2733 --------------------------
3084 2734
3085 :Capability: KVM_CAP_PPC_GET_SMMU_INFO 2735 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
3086 :Architectures: powerpc 2736 :Architectures: powerpc
3087 :Type: vm ioctl 2737 :Type: vm ioctl
3088 :Parameters: None 2738 :Parameters: None
3089 :Returns: 0 on success, -1 on error 2739 :Returns: 0 on success, -1 on error
3090 2740
3091 This populates and returns a structure descri 2741 This populates and returns a structure describing the features of
3092 the "Server" class MMU emulation supported by 2742 the "Server" class MMU emulation supported by KVM.
3093 This can in turn be used by userspace to gene 2743 This can in turn be used by userspace to generate the appropriate
3094 device-tree properties for the guest operatin 2744 device-tree properties for the guest operating system.
3095 2745
3096 The structure contains some global informatio 2746 The structure contains some global information, followed by an
3097 array of supported segment page sizes:: 2747 array of supported segment page sizes::
3098 2748
3099 struct kvm_ppc_smmu_info { 2749 struct kvm_ppc_smmu_info {
3100 __u64 flags; 2750 __u64 flags;
3101 __u32 slb_size; 2751 __u32 slb_size;
3102 __u32 pad; 2752 __u32 pad;
3103 struct kvm_ppc_one_seg_page_size 2753 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
3104 }; 2754 };
3105 2755
3106 The supported flags are: 2756 The supported flags are:
3107 2757
3108 - KVM_PPC_PAGE_SIZES_REAL: 2758 - KVM_PPC_PAGE_SIZES_REAL:
3109 When that flag is set, guest page siz 2759 When that flag is set, guest page sizes must "fit" the backing
3110 store page sizes. When not set, any p 2760 store page sizes. When not set, any page size in the list can
3111 be used regardless of how they are ba 2761 be used regardless of how they are backed by userspace.
3112 2762
3113 - KVM_PPC_1T_SEGMENTS 2763 - KVM_PPC_1T_SEGMENTS
3114 The emulated MMU supports 1T segments 2764 The emulated MMU supports 1T segments in addition to the
3115 standard 256M ones. 2765 standard 256M ones.
3116 2766
3117 - KVM_PPC_NO_HASH 2767 - KVM_PPC_NO_HASH
3118 This flag indicates that HPT guests a 2768 This flag indicates that HPT guests are not supported by KVM,
3119 thus all guests must use radix MMU mo 2769 thus all guests must use radix MMU mode.
3120 2770
3121 The "slb_size" field indicates how many SLB e 2771 The "slb_size" field indicates how many SLB entries are supported
3122 2772
3123 The "sps" array contains 8 entries indicating 2773 The "sps" array contains 8 entries indicating the supported base
3124 page sizes for a segment in increasing order. 2774 page sizes for a segment in increasing order. Each entry is defined
3125 as follow:: 2775 as follow::
3126 2776
3127 struct kvm_ppc_one_seg_page_size { 2777 struct kvm_ppc_one_seg_page_size {
3128 __u32 page_shift; /* Base page 2778 __u32 page_shift; /* Base page shift of segment (or 0) */
3129 __u32 slb_enc; /* SLB encodi 2779 __u32 slb_enc; /* SLB encoding for BookS */
3130 struct kvm_ppc_one_page_size enc[KVM_ 2780 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
3131 }; 2781 };
3132 2782
3133 An entry with a "page_shift" of 0 is unused. 2783 An entry with a "page_shift" of 0 is unused. Because the array is
3134 organized in increasing order, a lookup can s !! 2784 organized in increasing order, a lookup can stop when encoutering
3135 such an entry. 2785 such an entry.
3136 2786
3137 The "slb_enc" field provides the encoding to 2787 The "slb_enc" field provides the encoding to use in the SLB for the
3138 page size. The bits are in positions such as 2788 page size. The bits are in positions such as the value can directly
3139 be OR'ed into the "vsid" argument of the slbm 2789 be OR'ed into the "vsid" argument of the slbmte instruction.
3140 2790
3141 The "enc" array is a list which for each of t 2791 The "enc" array is a list which for each of those segment base page
3142 size provides the list of supported actual pa 2792 size provides the list of supported actual page sizes (which can be
3143 only larger or equal to the base page size), 2793 only larger or equal to the base page size), along with the
3144 corresponding encoding in the hash PTE. Simil 2794 corresponding encoding in the hash PTE. Similarly, the array is
3145 8 entries sorted by increasing sizes and an e 2795 8 entries sorted by increasing sizes and an entry with a "0" shift
3146 is an empty entry and a terminator:: 2796 is an empty entry and a terminator::
3147 2797
3148 struct kvm_ppc_one_page_size { 2798 struct kvm_ppc_one_page_size {
3149 __u32 page_shift; /* Page shift 2799 __u32 page_shift; /* Page shift (or 0) */
3150 __u32 pte_enc; /* Encoding i 2800 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
3151 }; 2801 };
3152 2802
3153 The "pte_enc" field provides a value that can 2803 The "pte_enc" field provides a value that can OR'ed into the hash
3154 PTE's RPN field (ie, it needs to be shifted l 2804 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
3155 into the hash PTE second double word). 2805 into the hash PTE second double word).
3156 2806
3157 4.75 KVM_IRQFD 2807 4.75 KVM_IRQFD
3158 -------------- 2808 --------------
3159 2809
3160 :Capability: KVM_CAP_IRQFD 2810 :Capability: KVM_CAP_IRQFD
3161 :Architectures: x86 s390 arm64 !! 2811 :Architectures: x86 s390 arm arm64
3162 :Type: vm ioctl 2812 :Type: vm ioctl
3163 :Parameters: struct kvm_irqfd (in) 2813 :Parameters: struct kvm_irqfd (in)
3164 :Returns: 0 on success, -1 on error 2814 :Returns: 0 on success, -1 on error
3165 2815
3166 Allows setting an eventfd to directly trigger 2816 Allows setting an eventfd to directly trigger a guest interrupt.
3167 kvm_irqfd.fd specifies the file descriptor to 2817 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
3168 kvm_irqfd.gsi specifies the irqchip pin toggl 2818 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
3169 an event is triggered on the eventfd, an inte 2819 an event is triggered on the eventfd, an interrupt is injected into
3170 the guest using the specified gsi pin. The i 2820 the guest using the specified gsi pin. The irqfd is removed using
3171 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying 2821 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
3172 and kvm_irqfd.gsi. 2822 and kvm_irqfd.gsi.
3173 2823
3174 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD suppor 2824 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
3175 mechanism allowing emulation of level-trigger 2825 mechanism allowing emulation of level-triggered, irqfd-based
3176 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is 2826 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
3177 additional eventfd in the kvm_irqfd.resamplef 2827 additional eventfd in the kvm_irqfd.resamplefd field. When operating
3178 in resample mode, posting of an interrupt thr 2828 in resample mode, posting of an interrupt through kvm_irq.fd asserts
3179 the specified gsi in the irqchip. When the i 2829 the specified gsi in the irqchip. When the irqchip is resampled, such
3180 as from an EOI, the gsi is de-asserted and th 2830 as from an EOI, the gsi is de-asserted and the user is notified via
3181 kvm_irqfd.resamplefd. It is the user's respo 2831 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
3182 the interrupt if the device making use of it 2832 the interrupt if the device making use of it still requires service.
3183 Note that closing the resamplefd is not suffi 2833 Note that closing the resamplefd is not sufficient to disable the
3184 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only n 2834 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
3185 and need not be specified with KVM_IRQFD_FLAG 2835 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
3186 2836
3187 On arm64, gsi routing being supported, the fo !! 2837 On arm/arm64, gsi routing being supported, the following can happen:
3188 2838
3189 - in case no routing entry is associated to t 2839 - in case no routing entry is associated to this gsi, injection fails
3190 - in case the gsi is associated to an irqchip 2840 - in case the gsi is associated to an irqchip routing entry,
3191 irqchip.pin + 32 corresponds to the injecte 2841 irqchip.pin + 32 corresponds to the injected SPI ID.
3192 - in case the gsi is associated to an MSI rou 2842 - in case the gsi is associated to an MSI routing entry, the MSI
3193 message and device ID are translated into a 2843 message and device ID are translated into an LPI (support restricted
3194 to GICv3 ITS in-kernel emulation). 2844 to GICv3 ITS in-kernel emulation).
3195 2845
3196 4.76 KVM_PPC_ALLOCATE_HTAB 2846 4.76 KVM_PPC_ALLOCATE_HTAB
3197 -------------------------- 2847 --------------------------
3198 2848
3199 :Capability: KVM_CAP_PPC_ALLOC_HTAB 2849 :Capability: KVM_CAP_PPC_ALLOC_HTAB
3200 :Architectures: powerpc 2850 :Architectures: powerpc
3201 :Type: vm ioctl 2851 :Type: vm ioctl
3202 :Parameters: Pointer to u32 containing hash t 2852 :Parameters: Pointer to u32 containing hash table order (in/out)
3203 :Returns: 0 on success, -1 on error 2853 :Returns: 0 on success, -1 on error
3204 2854
3205 This requests the host kernel to allocate an 2855 This requests the host kernel to allocate an MMU hash table for a
3206 guest using the PAPR paravirtualization inter 2856 guest using the PAPR paravirtualization interface. This only does
3207 anything if the kernel is configured to use t 2857 anything if the kernel is configured to use the Book 3S HV style of
3208 virtualization. Otherwise the capability doe 2858 virtualization. Otherwise the capability doesn't exist and the ioctl
3209 returns an ENOTTY error. The rest of this de 2859 returns an ENOTTY error. The rest of this description assumes Book 3S
3210 HV. 2860 HV.
3211 2861
3212 There must be no vcpus running when this ioct 2862 There must be no vcpus running when this ioctl is called; if there
3213 are, it will do nothing and return an EBUSY e 2863 are, it will do nothing and return an EBUSY error.
3214 2864
3215 The parameter is a pointer to a 32-bit unsign 2865 The parameter is a pointer to a 32-bit unsigned integer variable
3216 containing the order (log base 2) of the desi 2866 containing the order (log base 2) of the desired size of the hash
3217 table, which must be between 18 and 46. On s 2867 table, which must be between 18 and 46. On successful return from the
3218 ioctl, the value will not be changed by the k 2868 ioctl, the value will not be changed by the kernel.
3219 2869
3220 If no hash table has been allocated when any 2870 If no hash table has been allocated when any vcpu is asked to run
3221 (with the KVM_RUN ioctl), the host kernel wil 2871 (with the KVM_RUN ioctl), the host kernel will allocate a
3222 default-sized hash table (16 MB). 2872 default-sized hash table (16 MB).
3223 2873
3224 If this ioctl is called when a hash table has 2874 If this ioctl is called when a hash table has already been allocated,
3225 with a different order from the existing hash 2875 with a different order from the existing hash table, the existing hash
3226 table will be freed and a new one allocated. 2876 table will be freed and a new one allocated. If this is ioctl is
3227 called when a hash table has already been all 2877 called when a hash table has already been allocated of the same order
3228 as specified, the kernel will clear out the e 2878 as specified, the kernel will clear out the existing hash table (zero
3229 all HPTEs). In either case, if the guest is 2879 all HPTEs). In either case, if the guest is using the virtualized
3230 real-mode area (VRMA) facility, the kernel wi 2880 real-mode area (VRMA) facility, the kernel will re-create the VMRA
3231 HPTEs on the next KVM_RUN of any vcpu. 2881 HPTEs on the next KVM_RUN of any vcpu.
3232 2882
3233 4.77 KVM_S390_INTERRUPT 2883 4.77 KVM_S390_INTERRUPT
3234 ----------------------- 2884 -----------------------
3235 2885
3236 :Capability: basic 2886 :Capability: basic
3237 :Architectures: s390 2887 :Architectures: s390
3238 :Type: vm ioctl, vcpu ioctl 2888 :Type: vm ioctl, vcpu ioctl
3239 :Parameters: struct kvm_s390_interrupt (in) 2889 :Parameters: struct kvm_s390_interrupt (in)
3240 :Returns: 0 on success, -1 on error 2890 :Returns: 0 on success, -1 on error
3241 2891
3242 Allows to inject an interrupt to the guest. I 2892 Allows to inject an interrupt to the guest. Interrupts can be floating
3243 (vm ioctl) or per cpu (vcpu ioctl), depending 2893 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
3244 2894
3245 Interrupt parameters are passed via kvm_s390_ 2895 Interrupt parameters are passed via kvm_s390_interrupt::
3246 2896
3247 struct kvm_s390_interrupt { 2897 struct kvm_s390_interrupt {
3248 __u32 type; 2898 __u32 type;
3249 __u32 parm; 2899 __u32 parm;
3250 __u64 parm64; 2900 __u64 parm64;
3251 }; 2901 };
3252 2902
3253 type can be one of the following: 2903 type can be one of the following:
3254 2904
3255 KVM_S390_SIGP_STOP (vcpu) 2905 KVM_S390_SIGP_STOP (vcpu)
3256 - sigp stop; optional flags in parm 2906 - sigp stop; optional flags in parm
3257 KVM_S390_PROGRAM_INT (vcpu) 2907 KVM_S390_PROGRAM_INT (vcpu)
3258 - program check; code in parm 2908 - program check; code in parm
3259 KVM_S390_SIGP_SET_PREFIX (vcpu) 2909 KVM_S390_SIGP_SET_PREFIX (vcpu)
3260 - sigp set prefix; prefix address in parm 2910 - sigp set prefix; prefix address in parm
3261 KVM_S390_RESTART (vcpu) 2911 KVM_S390_RESTART (vcpu)
3262 - restart 2912 - restart
3263 KVM_S390_INT_CLOCK_COMP (vcpu) 2913 KVM_S390_INT_CLOCK_COMP (vcpu)
3264 - clock comparator interrupt 2914 - clock comparator interrupt
3265 KVM_S390_INT_CPU_TIMER (vcpu) 2915 KVM_S390_INT_CPU_TIMER (vcpu)
3266 - CPU timer interrupt 2916 - CPU timer interrupt
3267 KVM_S390_INT_VIRTIO (vm) 2917 KVM_S390_INT_VIRTIO (vm)
3268 - virtio external interrupt; external int 2918 - virtio external interrupt; external interrupt
3269 parameters in parm and parm64 2919 parameters in parm and parm64
3270 KVM_S390_INT_SERVICE (vm) 2920 KVM_S390_INT_SERVICE (vm)
3271 - sclp external interrupt; sclp parameter 2921 - sclp external interrupt; sclp parameter in parm
3272 KVM_S390_INT_EMERGENCY (vcpu) 2922 KVM_S390_INT_EMERGENCY (vcpu)
3273 - sigp emergency; source cpu in parm 2923 - sigp emergency; source cpu in parm
3274 KVM_S390_INT_EXTERNAL_CALL (vcpu) 2924 KVM_S390_INT_EXTERNAL_CALL (vcpu)
3275 - sigp external call; source cpu in parm 2925 - sigp external call; source cpu in parm
3276 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) 2926 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
3277 - compound value to indicate an 2927 - compound value to indicate an
3278 I/O interrupt (ai - adapter interrupt; 2928 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
3279 I/O interruption parameters in parm (su 2929 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
3280 interruption subclass) 2930 interruption subclass)
3281 KVM_S390_MCHK (vm, vcpu) 2931 KVM_S390_MCHK (vm, vcpu)
3282 - machine check interrupt; cr 14 bits in 2932 - machine check interrupt; cr 14 bits in parm, machine check interrupt
3283 code in parm64 (note that machine check 2933 code in parm64 (note that machine checks needing further payload are not
3284 supported by this ioctl) 2934 supported by this ioctl)
3285 2935
3286 This is an asynchronous vcpu ioctl and can be 2936 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3287 2937
3288 4.78 KVM_PPC_GET_HTAB_FD 2938 4.78 KVM_PPC_GET_HTAB_FD
3289 ------------------------ 2939 ------------------------
3290 2940
3291 :Capability: KVM_CAP_PPC_HTAB_FD 2941 :Capability: KVM_CAP_PPC_HTAB_FD
3292 :Architectures: powerpc 2942 :Architectures: powerpc
3293 :Type: vm ioctl 2943 :Type: vm ioctl
3294 :Parameters: Pointer to struct kvm_get_htab_f 2944 :Parameters: Pointer to struct kvm_get_htab_fd (in)
3295 :Returns: file descriptor number (>= 0) on su 2945 :Returns: file descriptor number (>= 0) on success, -1 on error
3296 2946
3297 This returns a file descriptor that can be us 2947 This returns a file descriptor that can be used either to read out the
3298 entries in the guest's hashed page table (HPT 2948 entries in the guest's hashed page table (HPT), or to write entries to
3299 initialize the HPT. The returned fd can only 2949 initialize the HPT. The returned fd can only be written to if the
3300 KVM_GET_HTAB_WRITE bit is set in the flags fi 2950 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
3301 can only be read if that bit is clear. The a 2951 can only be read if that bit is clear. The argument struct looks like
3302 this:: 2952 this::
3303 2953
3304 /* For KVM_PPC_GET_HTAB_FD */ 2954 /* For KVM_PPC_GET_HTAB_FD */
3305 struct kvm_get_htab_fd { 2955 struct kvm_get_htab_fd {
3306 __u64 flags; 2956 __u64 flags;
3307 __u64 start_index; 2957 __u64 start_index;
3308 __u64 reserved[2]; 2958 __u64 reserved[2];
3309 }; 2959 };
3310 2960
3311 /* Values for kvm_get_htab_fd.flags */ 2961 /* Values for kvm_get_htab_fd.flags */
3312 #define KVM_GET_HTAB_BOLTED_ONLY ((__u 2962 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
3313 #define KVM_GET_HTAB_WRITE ((__u 2963 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
3314 2964
3315 The 'start_index' field gives the index in th 2965 The 'start_index' field gives the index in the HPT of the entry at
3316 which to start reading. It is ignored when w 2966 which to start reading. It is ignored when writing.
3317 2967
3318 Reads on the fd will initially supply informa 2968 Reads on the fd will initially supply information about all
3319 "interesting" HPT entries. Interesting entri 2969 "interesting" HPT entries. Interesting entries are those with the
3320 bolted bit set, if the KVM_GET_HTAB_BOLTED_ON 2970 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
3321 all entries. When the end of the HPT is reac 2971 all entries. When the end of the HPT is reached, the read() will
3322 return. If read() is called again on the fd, 2972 return. If read() is called again on the fd, it will start again from
3323 the beginning of the HPT, but will only retur 2973 the beginning of the HPT, but will only return HPT entries that have
3324 changed since they were last read. 2974 changed since they were last read.
3325 2975
3326 Data read or written is structured as a heade 2976 Data read or written is structured as a header (8 bytes) followed by a
3327 series of valid HPT entries (16 bytes) each. 2977 series of valid HPT entries (16 bytes) each. The header indicates how
3328 many valid HPT entries there are and how many 2978 many valid HPT entries there are and how many invalid entries follow
3329 the valid entries. The invalid entries are n 2979 the valid entries. The invalid entries are not represented explicitly
3330 in the stream. The header format is:: 2980 in the stream. The header format is::
3331 2981
3332 struct kvm_get_htab_header { 2982 struct kvm_get_htab_header {
3333 __u32 index; 2983 __u32 index;
3334 __u16 n_valid; 2984 __u16 n_valid;
3335 __u16 n_invalid; 2985 __u16 n_invalid;
3336 }; 2986 };
3337 2987
3338 Writes to the fd create HPT entries starting 2988 Writes to the fd create HPT entries starting at the index given in the
3339 header; first 'n_valid' valid entries with co 2989 header; first 'n_valid' valid entries with contents from the data
3340 written, then 'n_invalid' invalid entries, in 2990 written, then 'n_invalid' invalid entries, invalidating any previously
3341 valid entries found. 2991 valid entries found.
3342 2992
3343 4.79 KVM_CREATE_DEVICE 2993 4.79 KVM_CREATE_DEVICE
3344 ---------------------- 2994 ----------------------
3345 2995
3346 :Capability: KVM_CAP_DEVICE_CTRL 2996 :Capability: KVM_CAP_DEVICE_CTRL
3347 :Architectures: all <<
3348 :Type: vm ioctl 2997 :Type: vm ioctl
3349 :Parameters: struct kvm_create_device (in/out 2998 :Parameters: struct kvm_create_device (in/out)
3350 :Returns: 0 on success, -1 on error 2999 :Returns: 0 on success, -1 on error
3351 3000
3352 Errors: 3001 Errors:
3353 3002
3354 ====== =================================== 3003 ====== =======================================================
3355 ENODEV The device type is unknown or unsup 3004 ENODEV The device type is unknown or unsupported
3356 EEXIST Device already created, and this ty 3005 EEXIST Device already created, and this type of device may not
3357 be instantiated multiple times 3006 be instantiated multiple times
3358 ====== =================================== 3007 ====== =======================================================
3359 3008
3360 Other error conditions may be defined by in 3009 Other error conditions may be defined by individual device types or
3361 have their standard meanings. 3010 have their standard meanings.
3362 3011
3363 Creates an emulated device in the kernel. Th 3012 Creates an emulated device in the kernel. The file descriptor returned
3364 in fd can be used with KVM_SET/GET/HAS_DEVICE 3013 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3365 3014
3366 If the KVM_CREATE_DEVICE_TEST flag is set, on 3015 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3367 device type is supported (not necessarily whe 3016 device type is supported (not necessarily whether it can be created
3368 in the current vm). 3017 in the current vm).
3369 3018
3370 Individual devices should not define flags. 3019 Individual devices should not define flags. Attributes should be used
3371 for specifying any behavior that is not impli 3020 for specifying any behavior that is not implied by the device type
3372 number. 3021 number.
3373 3022
3374 :: 3023 ::
3375 3024
3376 struct kvm_create_device { 3025 struct kvm_create_device {
3377 __u32 type; /* in: KVM_DEV_TYPE_x 3026 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3378 __u32 fd; /* out: device handle 3027 __u32 fd; /* out: device handle */
3379 __u32 flags; /* in: KVM_CREATE_DEV 3028 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3380 }; 3029 };
3381 3030
3382 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR 3031 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3383 -------------------------------------------- 3032 --------------------------------------------
3384 3033
3385 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ 3034 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3386 KVM_CAP_VCPU_ATTRIBUTES for vcpu 3035 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3387 KVM_CAP_SYS_ATTRIBUTES for syste <<
3388 :Architectures: x86, arm64, s390 <<
3389 :Type: device ioctl, vm ioctl, vcpu ioctl 3036 :Type: device ioctl, vm ioctl, vcpu ioctl
3390 :Parameters: struct kvm_device_attr 3037 :Parameters: struct kvm_device_attr
3391 :Returns: 0 on success, -1 on error 3038 :Returns: 0 on success, -1 on error
3392 3039
3393 Errors: 3040 Errors:
3394 3041
3395 ===== =================================== 3042 ===== =============================================================
3396 ENXIO The group or attribute is unknown/u 3043 ENXIO The group or attribute is unknown/unsupported for this device
3397 or hardware support is missing. 3044 or hardware support is missing.
3398 EPERM The attribute cannot (currently) be 3045 EPERM The attribute cannot (currently) be accessed this way
3399 (e.g. read-only attribute, or attri 3046 (e.g. read-only attribute, or attribute that only makes
3400 sense when the device is in a diffe 3047 sense when the device is in a different state)
3401 ===== =================================== 3048 ===== =============================================================
3402 3049
3403 Other error conditions may be defined by in 3050 Other error conditions may be defined by individual device types.
3404 3051
3405 Gets/sets a specified piece of device configu 3052 Gets/sets a specified piece of device configuration and/or state. The
3406 semantics are device-specific. See individua 3053 semantics are device-specific. See individual device documentation in
3407 the "devices" directory. As with ONE_REG, th 3054 the "devices" directory. As with ONE_REG, the size of the data
3408 transferred is defined by the particular attr 3055 transferred is defined by the particular attribute.
3409 3056
3410 :: 3057 ::
3411 3058
3412 struct kvm_device_attr { 3059 struct kvm_device_attr {
3413 __u32 flags; /* no flags c 3060 __u32 flags; /* no flags currently defined */
3414 __u32 group; /* device-def 3061 __u32 group; /* device-defined */
3415 __u64 attr; /* group-defi 3062 __u64 attr; /* group-defined */
3416 __u64 addr; /* userspace 3063 __u64 addr; /* userspace address of attr data */
3417 }; 3064 };
3418 3065
3419 4.81 KVM_HAS_DEVICE_ATTR 3066 4.81 KVM_HAS_DEVICE_ATTR
3420 ------------------------ 3067 ------------------------
3421 3068
3422 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ 3069 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3423 KVM_CAP_VCPU_ATTRIBUTES for vcpu !! 3070 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3424 KVM_CAP_SYS_ATTRIBUTES for syste <<
3425 :Type: device ioctl, vm ioctl, vcpu ioctl 3071 :Type: device ioctl, vm ioctl, vcpu ioctl
3426 :Parameters: struct kvm_device_attr 3072 :Parameters: struct kvm_device_attr
3427 :Returns: 0 on success, -1 on error 3073 :Returns: 0 on success, -1 on error
3428 3074
3429 Errors: 3075 Errors:
3430 3076
3431 ===== =================================== 3077 ===== =============================================================
3432 ENXIO The group or attribute is unknown/u 3078 ENXIO The group or attribute is unknown/unsupported for this device
3433 or hardware support is missing. 3079 or hardware support is missing.
3434 ===== =================================== 3080 ===== =============================================================
3435 3081
3436 Tests whether a device supports a particular 3082 Tests whether a device supports a particular attribute. A successful
3437 return indicates the attribute is implemented 3083 return indicates the attribute is implemented. It does not necessarily
3438 indicate that the attribute can be read or wr 3084 indicate that the attribute can be read or written in the device's
3439 current state. "addr" is ignored. 3085 current state. "addr" is ignored.
3440 3086
3441 .. _KVM_ARM_VCPU_INIT: <<
3442 <<
3443 4.82 KVM_ARM_VCPU_INIT 3087 4.82 KVM_ARM_VCPU_INIT
3444 ---------------------- 3088 ----------------------
3445 3089
3446 :Capability: basic 3090 :Capability: basic
3447 :Architectures: arm64 !! 3091 :Architectures: arm, arm64
3448 :Type: vcpu ioctl 3092 :Type: vcpu ioctl
3449 :Parameters: struct kvm_vcpu_init (in) 3093 :Parameters: struct kvm_vcpu_init (in)
3450 :Returns: 0 on success; -1 on error 3094 :Returns: 0 on success; -1 on error
3451 3095
3452 Errors: 3096 Errors:
3453 3097
3454 ====== ================================ 3098 ====== =================================================================
3455 EINVAL the target is unknown, or the co !! 3099 Â EINVAL Â Â Â the target is unknown, or the combination of features is invalid.
3456 ENOENT a features bit specified is unkn !! 3100 Â ENOENT Â Â Â a features bit specified is unknown.
3457 ====== ================================ 3101 ====== =================================================================
3458 3102
3459 This tells KVM what type of CPU to present to 3103 This tells KVM what type of CPU to present to the guest, and what
3460 optional features it should have. This will !! 3104 optional features it should have. Â This will cause a reset of the cpu
3461 registers to their initial values. If this i !! 3105 registers to their initial values. Â If this is not called, KVM_RUN will
3462 return ENOEXEC for that vcpu. 3106 return ENOEXEC for that vcpu.
3463 3107
3464 The initial values are defined as: <<
3465 - Processor state: <<
3466 * AArch64: EL1h, D, A, I and <<
3467 are cleared. <<
3468 * AArch32: SVC, A, I and F bi <<
3469 cleared. <<
3470 - General Purpose registers, includin <<
3471 - FPSIMD/NEON registers: set to 0 <<
3472 - SVE registers: set to 0 <<
3473 - System registers: Reset to their ar <<
3474 values as for a warm reset to EL1 ( <<
3475 <<
3476 Note that because some registers reflect mach 3108 Note that because some registers reflect machine topology, all vcpus
3477 should be created before this ioctl is invoke 3109 should be created before this ioctl is invoked.
3478 3110
3479 Userspace can call this function multiple tim 3111 Userspace can call this function multiple times for a given vcpu, including
3480 after the vcpu has been run. This will reset 3112 after the vcpu has been run. This will reset the vcpu to its initial
3481 state. All calls to this function after the i 3113 state. All calls to this function after the initial call must use the same
3482 target and same set of feature flags, otherwi 3114 target and same set of feature flags, otherwise EINVAL will be returned.
3483 3115
3484 Possible features: 3116 Possible features:
3485 3117
3486 - KVM_ARM_VCPU_POWER_OFF: Starts the 3118 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3487 Depends on KVM_CAP_ARM_PSCI. If no 3119 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3488 and execute guest code when KVM_RUN 3120 and execute guest code when KVM_RUN is called.
3489 - KVM_ARM_VCPU_EL1_32BIT: Starts the 3121 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3490 Depends on KVM_CAP_ARM_EL1_32BIT (a 3122 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3491 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI 3123 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3492 backward compatible with v0.2) for 3124 backward compatible with v0.2) for the CPU.
3493 Depends on KVM_CAP_ARM_PSCI_0_2. 3125 Depends on KVM_CAP_ARM_PSCI_0_2.
3494 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 3126 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3495 Depends on KVM_CAP_ARM_PMU_V3. 3127 Depends on KVM_CAP_ARM_PMU_V3.
3496 3128
3497 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enabl 3129 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3498 for arm64 only. 3130 for arm64 only.
3499 Depends on KVM_CAP_ARM_PTRAUTH_ADDR 3131 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3500 If KVM_CAP_ARM_PTRAUTH_ADDRESS and 3132 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3501 both present, then both KVM_ARM_VCP 3133 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3502 KVM_ARM_VCPU_PTRAUTH_GENERIC must b 3134 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3503 requested. 3135 requested.
3504 3136
3505 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enabl 3137 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3506 for arm64 only. 3138 for arm64 only.
3507 Depends on KVM_CAP_ARM_PTRAUTH_GENE 3139 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3508 If KVM_CAP_ARM_PTRAUTH_ADDRESS and 3140 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3509 both present, then both KVM_ARM_VCP 3141 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3510 KVM_ARM_VCPU_PTRAUTH_GENERIC must b 3142 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3511 requested. 3143 requested.
3512 3144
3513 - KVM_ARM_VCPU_SVE: Enables SVE for t 3145 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3514 Depends on KVM_CAP_ARM_SVE. 3146 Depends on KVM_CAP_ARM_SVE.
3515 Requires KVM_ARM_VCPU_FINALIZE(KVM_ 3147 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3516 3148
3517 * After KVM_ARM_VCPU_INIT: 3149 * After KVM_ARM_VCPU_INIT:
3518 3150
3519 - KVM_REG_ARM64_SVE_VLS may be 3151 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3520 initial value of this pseudo- 3152 initial value of this pseudo-register indicates the best set of
3521 vector lengths possible for a 3153 vector lengths possible for a vcpu on this host.
3522 3154
3523 * Before KVM_ARM_VCPU_FINALIZE(KVM 3155 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3524 3156
3525 - KVM_RUN and KVM_GET_REG_LIST 3157 - KVM_RUN and KVM_GET_REG_LIST are not available;
3526 3158
3527 - KVM_GET_ONE_REG and KVM_SET_O 3159 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3528 the scalable architectural SV !! 3160 the scalable archietctural SVE registers
3529 KVM_REG_ARM64_SVE_ZREG(), KVM 3161 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3530 KVM_REG_ARM64_SVE_FFR; 3162 KVM_REG_ARM64_SVE_FFR;
3531 3163
3532 - KVM_REG_ARM64_SVE_VLS may opt 3164 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3533 KVM_SET_ONE_REG, to modify th 3165 KVM_SET_ONE_REG, to modify the set of vector lengths available
3534 for the vcpu. 3166 for the vcpu.
3535 3167
3536 * After KVM_ARM_VCPU_FINALIZE(KVM_ 3168 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3537 3169
3538 - the KVM_REG_ARM64_SVE_VLS pse 3170 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3539 no longer be written using KV 3171 no longer be written using KVM_SET_ONE_REG.
3540 3172
3541 4.83 KVM_ARM_PREFERRED_TARGET 3173 4.83 KVM_ARM_PREFERRED_TARGET
3542 ----------------------------- 3174 -----------------------------
3543 3175
3544 :Capability: basic 3176 :Capability: basic
3545 :Architectures: arm64 !! 3177 :Architectures: arm, arm64
3546 :Type: vm ioctl 3178 :Type: vm ioctl
3547 :Parameters: struct kvm_vcpu_init (out) 3179 :Parameters: struct kvm_vcpu_init (out)
3548 :Returns: 0 on success; -1 on error 3180 :Returns: 0 on success; -1 on error
3549 3181
3550 Errors: 3182 Errors:
3551 3183
3552 ====== ================================ 3184 ====== ==========================================
3553 ENODEV no preferred target available fo 3185 ENODEV no preferred target available for the host
3554 ====== ================================ 3186 ====== ==========================================
3555 3187
3556 This queries KVM for preferred CPU target typ 3188 This queries KVM for preferred CPU target type which can be emulated
3557 by KVM on underlying host. 3189 by KVM on underlying host.
3558 3190
3559 The ioctl returns struct kvm_vcpu_init instan 3191 The ioctl returns struct kvm_vcpu_init instance containing information
3560 about preferred CPU target type and recommend 3192 about preferred CPU target type and recommended features for it. The
3561 kvm_vcpu_init->features bitmap returned will 3193 kvm_vcpu_init->features bitmap returned will have feature bits set if
3562 the preferred target recommends setting these 3194 the preferred target recommends setting these features, but this is
3563 not mandatory. 3195 not mandatory.
3564 3196
3565 The information returned by this ioctl can be 3197 The information returned by this ioctl can be used to prepare an instance
3566 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT 3198 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3567 VCPU matching underlying host. 3199 VCPU matching underlying host.
3568 3200
3569 3201
3570 4.84 KVM_GET_REG_LIST 3202 4.84 KVM_GET_REG_LIST
3571 --------------------- 3203 ---------------------
3572 3204
3573 :Capability: basic 3205 :Capability: basic
3574 :Architectures: arm64, mips, riscv !! 3206 :Architectures: arm, arm64, mips
3575 :Type: vcpu ioctl 3207 :Type: vcpu ioctl
3576 :Parameters: struct kvm_reg_list (in/out) 3208 :Parameters: struct kvm_reg_list (in/out)
3577 :Returns: 0 on success; -1 on error 3209 :Returns: 0 on success; -1 on error
3578 3210
3579 Errors: 3211 Errors:
3580 3212
3581 ===== ================================ 3213 ===== ==============================================================
3582 E2BIG the reg index list is too big to !! 3214 Â E2BIG Â Â Â Â the reg index list is too big to fit in the array specified by
3583 the user (the number required wi !! 3215 Â Â Â Â Â Â Â Â Â Â Â Â the user (the number required will be written into n).
3584 ===== ================================ 3216 ===== ==============================================================
3585 3217
3586 :: 3218 ::
3587 3219
3588 struct kvm_reg_list { 3220 struct kvm_reg_list {
3589 __u64 n; /* number of registers in re 3221 __u64 n; /* number of registers in reg[] */
3590 __u64 reg[0]; 3222 __u64 reg[0];
3591 }; 3223 };
3592 3224
3593 This ioctl returns the guest registers that a 3225 This ioctl returns the guest registers that are supported for the
3594 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls. 3226 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3595 3227
3596 3228
3597 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated) 3229 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3598 ----------------------------------------- 3230 -----------------------------------------
3599 3231
3600 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR 3232 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3601 :Architectures: arm64 !! 3233 :Architectures: arm, arm64
3602 :Type: vm ioctl 3234 :Type: vm ioctl
3603 :Parameters: struct kvm_arm_device_address (i 3235 :Parameters: struct kvm_arm_device_address (in)
3604 :Returns: 0 on success, -1 on error 3236 :Returns: 0 on success, -1 on error
3605 3237
3606 Errors: 3238 Errors:
3607 3239
3608 ====== =================================== 3240 ====== ============================================
3609 ENODEV The device id is unknown 3241 ENODEV The device id is unknown
3610 ENXIO Device not supported on current sys 3242 ENXIO Device not supported on current system
3611 EEXIST Address already set 3243 EEXIST Address already set
3612 E2BIG Address outside guest physical addr 3244 E2BIG Address outside guest physical address space
3613 EBUSY Address overlaps with other device 3245 EBUSY Address overlaps with other device range
3614 ====== =================================== 3246 ====== ============================================
3615 3247
3616 :: 3248 ::
3617 3249
3618 struct kvm_arm_device_addr { 3250 struct kvm_arm_device_addr {
3619 __u64 id; 3251 __u64 id;
3620 __u64 addr; 3252 __u64 addr;
3621 }; 3253 };
3622 3254
3623 Specify a device address in the guest's physi 3255 Specify a device address in the guest's physical address space where guests
3624 can access emulated or directly exposed devic 3256 can access emulated or directly exposed devices, which the host kernel needs
3625 to know about. The id field is an architectur 3257 to know about. The id field is an architecture specific identifier for a
3626 specific device. 3258 specific device.
3627 3259
3628 arm64 divides the id field into two parts, a !! 3260 ARM/arm64 divides the id field into two parts, a device id and an
3629 address type id specific to the individual de 3261 address type id specific to the individual device::
3630 3262
3631 bits: | 63 ... 32 | 31 ... !! 3263 Â bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3632 field: | 0x00000000 | devic 3264 field: | 0x00000000 | device id | addr type id |
3633 3265
3634 arm64 currently only require this when using !! 3266 ARM/arm64 currently only require this when using the in-kernel GIC
3635 support for the hardware VGIC features, using 3267 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3636 as the device id. When setting the base addr 3268 as the device id. When setting the base address for the guest's
3637 mapping of the VGIC virtual CPU and distribut 3269 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3638 must be called after calling KVM_CREATE_IRQCH 3270 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3639 KVM_RUN on any of the VCPUs. Calling this io 3271 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3640 base addresses will return -EEXIST. 3272 base addresses will return -EEXIST.
3641 3273
3642 Note, this IOCTL is deprecated and the more f 3274 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3643 should be used instead. 3275 should be used instead.
3644 3276
3645 3277
3646 4.86 KVM_PPC_RTAS_DEFINE_TOKEN 3278 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3647 ------------------------------ 3279 ------------------------------
3648 3280
3649 :Capability: KVM_CAP_PPC_RTAS 3281 :Capability: KVM_CAP_PPC_RTAS
3650 :Architectures: ppc 3282 :Architectures: ppc
3651 :Type: vm ioctl 3283 :Type: vm ioctl
3652 :Parameters: struct kvm_rtas_token_args 3284 :Parameters: struct kvm_rtas_token_args
3653 :Returns: 0 on success, -1 on error 3285 :Returns: 0 on success, -1 on error
3654 3286
3655 Defines a token value for a RTAS (Run Time Ab 3287 Defines a token value for a RTAS (Run Time Abstraction Services)
3656 service in order to allow it to be handled in 3288 service in order to allow it to be handled in the kernel. The
3657 argument struct gives the name of the service 3289 argument struct gives the name of the service, which must be the name
3658 of a service that has a kernel-side implement 3290 of a service that has a kernel-side implementation. If the token
3659 value is non-zero, it will be associated with 3291 value is non-zero, it will be associated with that service, and
3660 subsequent RTAS calls by the guest specifying 3292 subsequent RTAS calls by the guest specifying that token will be
3661 handled by the kernel. If the token value is 3293 handled by the kernel. If the token value is 0, then any token
3662 associated with the service will be forgotten 3294 associated with the service will be forgotten, and subsequent RTAS
3663 calls by the guest for that service will be p 3295 calls by the guest for that service will be passed to userspace to be
3664 handled. 3296 handled.
3665 3297
3666 4.87 KVM_SET_GUEST_DEBUG 3298 4.87 KVM_SET_GUEST_DEBUG
3667 ------------------------ 3299 ------------------------
3668 3300
3669 :Capability: KVM_CAP_SET_GUEST_DEBUG 3301 :Capability: KVM_CAP_SET_GUEST_DEBUG
3670 :Architectures: x86, s390, ppc, arm64 3302 :Architectures: x86, s390, ppc, arm64
3671 :Type: vcpu ioctl 3303 :Type: vcpu ioctl
3672 :Parameters: struct kvm_guest_debug (in) 3304 :Parameters: struct kvm_guest_debug (in)
3673 :Returns: 0 on success; -1 on error 3305 :Returns: 0 on success; -1 on error
3674 3306
3675 :: 3307 ::
3676 3308
3677 struct kvm_guest_debug { 3309 struct kvm_guest_debug {
3678 __u32 control; 3310 __u32 control;
3679 __u32 pad; 3311 __u32 pad;
3680 struct kvm_guest_debug_arch arch; 3312 struct kvm_guest_debug_arch arch;
3681 }; 3313 };
3682 3314
3683 Set up the processor specific debug registers 3315 Set up the processor specific debug registers and configure vcpu for
3684 handling guest debug events. There are two pa 3316 handling guest debug events. There are two parts to the structure, the
3685 first a control bitfield indicates the type o 3317 first a control bitfield indicates the type of debug events to handle
3686 when running. Common control bits are: 3318 when running. Common control bits are:
3687 3319
3688 - KVM_GUESTDBG_ENABLE: guest debuggi 3320 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3689 - KVM_GUESTDBG_SINGLESTEP: the next run 3321 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3690 3322
3691 The top 16 bits of the control field are arch 3323 The top 16 bits of the control field are architecture specific control
3692 flags which can include the following: 3324 flags which can include the following:
3693 3325
3694 - KVM_GUESTDBG_USE_SW_BP: using softwar 3326 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3695 - KVM_GUESTDBG_USE_HW_BP: using hardwar !! 3327 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390, arm64]
3696 - KVM_GUESTDBG_USE_HW: using hardwar <<
3697 - KVM_GUESTDBG_INJECT_DB: inject DB typ 3328 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3698 - KVM_GUESTDBG_INJECT_BP: inject BP typ 3329 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3699 - KVM_GUESTDBG_EXIT_PENDING: trigger an im 3330 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3700 - KVM_GUESTDBG_BLOCKIRQ: avoid injecti <<
3701 3331
3702 For example KVM_GUESTDBG_USE_SW_BP indicates 3332 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3703 are enabled in memory so we need to ensure br 3333 are enabled in memory so we need to ensure breakpoint exceptions are
3704 correctly trapped and the KVM run loop exits 3334 correctly trapped and the KVM run loop exits at the breakpoint and not
3705 running off into the normal guest vector. For 3335 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3706 we need to ensure the guest vCPUs architectur 3336 we need to ensure the guest vCPUs architecture specific registers are
3707 updated to the correct (supplied) values. 3337 updated to the correct (supplied) values.
3708 3338
3709 The second part of the structure is architect 3339 The second part of the structure is architecture specific and
3710 typically contains a set of debug registers. 3340 typically contains a set of debug registers.
3711 3341
3712 For arm64 the number of debug registers is im 3342 For arm64 the number of debug registers is implementation defined and
3713 can be determined by querying the KVM_CAP_GUE 3343 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3714 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which 3344 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3715 indicating the number of supported registers. 3345 indicating the number of supported registers.
3716 3346
3717 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP ca 3347 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3718 the single-step debug event (KVM_GUESTDBG_SIN 3348 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3719 3349
3720 Also when supported, KVM_CAP_SET_GUEST_DEBUG2 <<
3721 supported KVM_GUESTDBG_* bits in the control <<
3722 <<
3723 When debug events exit the main run loop with 3350 When debug events exit the main run loop with the reason
3724 KVM_EXIT_DEBUG with the kvm_debug_exit_arch p 3351 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3725 structure containing architecture specific de 3352 structure containing architecture specific debug information.
3726 3353
3727 4.88 KVM_GET_EMULATED_CPUID 3354 4.88 KVM_GET_EMULATED_CPUID
3728 --------------------------- 3355 ---------------------------
3729 3356
3730 :Capability: KVM_CAP_EXT_EMUL_CPUID 3357 :Capability: KVM_CAP_EXT_EMUL_CPUID
3731 :Architectures: x86 3358 :Architectures: x86
3732 :Type: system ioctl 3359 :Type: system ioctl
3733 :Parameters: struct kvm_cpuid2 (in/out) 3360 :Parameters: struct kvm_cpuid2 (in/out)
3734 :Returns: 0 on success, -1 on error 3361 :Returns: 0 on success, -1 on error
3735 3362
3736 :: 3363 ::
3737 3364
3738 struct kvm_cpuid2 { 3365 struct kvm_cpuid2 {
3739 __u32 nent; 3366 __u32 nent;
3740 __u32 flags; 3367 __u32 flags;
3741 struct kvm_cpuid_entry2 entries[0]; 3368 struct kvm_cpuid_entry2 entries[0];
3742 }; 3369 };
3743 3370
3744 The member 'flags' is used for passing flags 3371 The member 'flags' is used for passing flags from userspace.
3745 3372
3746 :: 3373 ::
3747 3374
3748 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX 3375 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3749 #define KVM_CPUID_FLAG_STATEFUL_FUNC 3376 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3750 #define KVM_CPUID_FLAG_STATE_READ_NEXT 3377 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3751 3378
3752 struct kvm_cpuid_entry2 { 3379 struct kvm_cpuid_entry2 {
3753 __u32 function; 3380 __u32 function;
3754 __u32 index; 3381 __u32 index;
3755 __u32 flags; 3382 __u32 flags;
3756 __u32 eax; 3383 __u32 eax;
3757 __u32 ebx; 3384 __u32 ebx;
3758 __u32 ecx; 3385 __u32 ecx;
3759 __u32 edx; 3386 __u32 edx;
3760 __u32 padding[3]; 3387 __u32 padding[3];
3761 }; 3388 };
3762 3389
3763 This ioctl returns x86 cpuid features which a 3390 This ioctl returns x86 cpuid features which are emulated by
3764 kvm.Userspace can use the information returne 3391 kvm.Userspace can use the information returned by this ioctl to query
3765 which features are emulated by kvm instead of 3392 which features are emulated by kvm instead of being present natively.
3766 3393
3767 Userspace invokes KVM_GET_EMULATED_CPUID by p 3394 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3768 structure with the 'nent' field indicating th 3395 structure with the 'nent' field indicating the number of entries in
3769 the variable-size array 'entries'. If the num 3396 the variable-size array 'entries'. If the number of entries is too low
3770 to describe the cpu capabilities, an error (E 3397 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3771 number is too high, the 'nent' field is adjus 3398 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3772 is returned. If the number is just right, the 3399 is returned. If the number is just right, the 'nent' field is adjusted
3773 to the number of valid entries in the 'entrie 3400 to the number of valid entries in the 'entries' array, which is then
3774 filled. 3401 filled.
3775 3402
3776 The entries returned are the set CPUID bits o 3403 The entries returned are the set CPUID bits of the respective features
3777 which kvm emulates, as returned by the CPUID 3404 which kvm emulates, as returned by the CPUID instruction, with unknown
3778 or unsupported feature bits cleared. 3405 or unsupported feature bits cleared.
3779 3406
3780 Features like x2apic, for example, may not be 3407 Features like x2apic, for example, may not be present in the host cpu
3781 but are exposed by kvm in KVM_GET_SUPPORTED_C 3408 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3782 emulated efficiently and thus not included he 3409 emulated efficiently and thus not included here.
3783 3410
3784 The fields in each entry are defined as follo 3411 The fields in each entry are defined as follows:
3785 3412
3786 function: 3413 function:
3787 the eax value used to obtain the ent 3414 the eax value used to obtain the entry
3788 index: 3415 index:
3789 the ecx value used to obtain the ent 3416 the ecx value used to obtain the entry (for entries that are
3790 affected by ecx) 3417 affected by ecx)
3791 flags: 3418 flags:
3792 an OR of zero or more of the following: 3419 an OR of zero or more of the following:
3793 3420
3794 KVM_CPUID_FLAG_SIGNIFCANT_INDEX: 3421 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3795 if the index field is valid 3422 if the index field is valid
3796 3423
3797 eax, ebx, ecx, edx: 3424 eax, ebx, ecx, edx:
3798 3425
3799 the values returned by the cpuid ins 3426 the values returned by the cpuid instruction for
3800 this function/index combination 3427 this function/index combination
3801 3428
3802 4.89 KVM_S390_MEM_OP 3429 4.89 KVM_S390_MEM_OP
3803 -------------------- 3430 --------------------
3804 3431
3805 :Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S39 !! 3432 :Capability: KVM_CAP_S390_MEM_OP
3806 :Architectures: s390 3433 :Architectures: s390
3807 :Type: vm ioctl, vcpu ioctl !! 3434 :Type: vcpu ioctl
3808 :Parameters: struct kvm_s390_mem_op (in) 3435 :Parameters: struct kvm_s390_mem_op (in)
3809 :Returns: = 0 on success, 3436 :Returns: = 0 on success,
3810 < 0 on generic error (e.g. -EFAULT 3437 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3811 16 bit program exception code if th !! 3438 > 0 if an exception occurred while walking the page tables
3812 3439
3813 Read or write data from/to the VM's memory. !! 3440 Read or write data from/to the logical (virtual) memory of a VCPU.
3814 The KVM_CAP_S390_MEM_OP_EXTENSION capability <<
3815 supported. <<
3816 3441
3817 Parameters are specified via the following st 3442 Parameters are specified via the following structure::
3818 3443
3819 struct kvm_s390_mem_op { 3444 struct kvm_s390_mem_op {
3820 __u64 gaddr; /* the guest 3445 __u64 gaddr; /* the guest address */
3821 __u64 flags; /* flags */ 3446 __u64 flags; /* flags */
3822 __u32 size; /* amount of 3447 __u32 size; /* amount of bytes */
3823 __u32 op; /* type of op 3448 __u32 op; /* type of operation */
3824 __u64 buf; /* buffer in 3449 __u64 buf; /* buffer in userspace */
3825 union { !! 3450 __u8 ar; /* the access register number */
3826 struct { !! 3451 __u8 reserved[31]; /* should be set to 0 */
3827 __u8 ar; /* th <<
3828 __u8 key; /* ac <<
3829 __u8 pad1[6]; /* ig <<
3830 __u64 old_addr; /* ig <<
3831 }; <<
3832 __u32 sida_offset; /* offset <<
3833 __u8 reserved[32]; /* ignored <<
3834 }; <<
3835 }; 3452 };
3836 3453
>> 3454 The type of operation is specified in the "op" field. It is either
>> 3455 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
>> 3456 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
>> 3457 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
>> 3458 whether the corresponding memory access would create an access exception
>> 3459 (without touching the data in the memory at the destination). In case an
>> 3460 access exception occurred while walking the MMU tables of the guest, the
>> 3461 ioctl returns a positive error number to indicate the type of exception.
>> 3462 This exception is also raised directly at the corresponding VCPU if the
>> 3463 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
>> 3464
3837 The start address of the memory region has to 3465 The start address of the memory region has to be specified in the "gaddr"
3838 field, and the length of the region in the "s 3466 field, and the length of the region in the "size" field (which must not
3839 be 0). The maximum value for "size" can be ob 3467 be 0). The maximum value for "size" can be obtained by checking the
3840 KVM_CAP_S390_MEM_OP capability. "buf" is the 3468 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3841 userspace application where the read data sho 3469 userspace application where the read data should be written to for
3842 a read access, or where the data that should !! 3470 KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is
3843 a write access. The "reserved" field is mean !! 3471 stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY
3844 Reserved and unused values are ignored. Futur !! 3472 is specified, "buf" is unused and can be NULL. "ar" designates the access
3845 introduce new flags. !! 3473 register number to be used; the valid range is 0..15.
3846 <<
3847 The type of operation is specified in the "op <<
3848 their behavior can be set in the "flags" fiel <<
3849 be set to 0. <<
3850 <<
3851 Possible operations are: <<
3852 * ``KVM_S390_MEMOP_LOGICAL_READ`` <<
3853 * ``KVM_S390_MEMOP_LOGICAL_WRITE`` <<
3854 * ``KVM_S390_MEMOP_ABSOLUTE_READ`` <<
3855 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE`` <<
3856 * ``KVM_S390_MEMOP_SIDA_READ`` <<
3857 * ``KVM_S390_MEMOP_SIDA_WRITE`` <<
3858 * ``KVM_S390_MEMOP_ABSOLUTE_CMPXCHG`` <<
3859 <<
3860 Logical read/write: <<
3861 ^^^^^^^^^^^^^^^^^^^ <<
3862 <<
3863 Access logical memory, i.e. translate the giv <<
3864 address given the state of the VCPU and use t <<
3865 the access. "ar" designates the access regist <<
3866 range is 0..15. <<
3867 Logical accesses are permitted for the VCPU i <<
3868 Logical accesses are permitted for non-protec <<
3869 <<
3870 Supported flags: <<
3871 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` <<
3872 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION`` <<
3873 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` <<
3874 <<
3875 The KVM_S390_MEMOP_F_CHECK_ONLY flag can be s <<
3876 corresponding memory access would cause an ac <<
3877 no actual access to the data in memory at the <<
3878 In this case, "buf" is unused and can be NULL <<
3879 <<
3880 In case an access exception occurred during t <<
3881 in case of KVM_S390_MEMOP_F_CHECK_ONLY), the <<
3882 error number indicating the type of exception <<
3883 raised directly at the corresponding VCPU if <<
3884 KVM_S390_MEMOP_F_INJECT_EXCEPTION is set. <<
3885 On protection exceptions, unless specified ot <<
3886 translation-exception identifier (TEID) indic <<
3887 <<
3888 If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag <<
3889 protection is also in effect and may cause ex <<
3890 prohibited given the access key designated by <<
3891 KVM_S390_MEMOP_F_SKEY_PROTECTION is available <<
3892 is > 0. <<
3893 Since the accessed memory may span multiple p <<
3894 different storage keys, it is possible that a <<
3895 after memory has been modified. In this case, <<
3896 the TEID does not indicate suppression. <<
3897 <<
3898 Absolute read/write: <<
3899 ^^^^^^^^^^^^^^^^^^^^ <<
3900 <<
3901 Access absolute memory. This operation is int <<
3902 KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to all <<
3903 the checks required for storage key protectio <<
3904 user space getting the storage keys, performi <<
3905 memory thereafter, which could lead to a dela <<
3906 Absolute accesses are permitted for the VM io <<
3907 has the KVM_S390_MEMOP_EXTENSION_CAP_BASE bit <<
3908 Currently absolute accesses are not permitted <<
3909 Absolute accesses are permitted for non-prote <<
3910 <<
3911 Supported flags: <<
3912 * ``KVM_S390_MEMOP_F_CHECK_ONLY`` <<
3913 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` <<
3914 <<
3915 The semantics of the flags common with logica <<
3916 accesses. <<
3917 <<
3918 Absolute cmpxchg: <<
3919 ^^^^^^^^^^^^^^^^^ <<
3920 <<
3921 Perform cmpxchg on absolute guest memory. Int <<
3922 KVM_S390_MEMOP_F_SKEY_PROTECTION flag. <<
3923 Instead of doing an unconditional write, the <<
3924 location contains the value pointed to by "ol <<
3925 This is performed as an atomic cmpxchg with t <<
3926 parameter. "size" must be a power of two up t <<
3927 If the exchange did not take place because th <<
3928 old value, the value "old_addr" points to is <<
3929 User space can tell if an exchange took place <<
3930 occurred. The cmpxchg op is permitted for the <<
3931 KVM_CAP_S390_MEM_OP_EXTENSION has flag KVM_S3 <<
3932 <<
3933 Supported flags: <<
3934 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION`` <<
3935 <<
3936 SIDA read/write: <<
3937 ^^^^^^^^^^^^^^^^ <<
3938 <<
3939 Access the secure instruction data area which <<
3940 for instruction emulation for protected guest <<
3941 SIDA accesses are available if the KVM_CAP_S3 <<
3942 SIDA accesses are permitted for the VCPU ioct <<
3943 SIDA accesses are permitted for protected gue <<
3944 3474
3945 No flags are supported. !! 3475 The "reserved" field is meant for future extensions. It is not used by
>> 3476 KVM with the currently defined set of flags.
3946 3477
3947 4.90 KVM_S390_GET_SKEYS 3478 4.90 KVM_S390_GET_SKEYS
3948 ----------------------- 3479 -----------------------
3949 3480
3950 :Capability: KVM_CAP_S390_SKEYS 3481 :Capability: KVM_CAP_S390_SKEYS
3951 :Architectures: s390 3482 :Architectures: s390
3952 :Type: vm ioctl 3483 :Type: vm ioctl
3953 :Parameters: struct kvm_s390_skeys 3484 :Parameters: struct kvm_s390_skeys
3954 :Returns: 0 on success, KVM_S390_GET_SKEYS_NO !! 3485 :Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
3955 keys, negative value on error 3486 keys, negative value on error
3956 3487
3957 This ioctl is used to get guest storage key v 3488 This ioctl is used to get guest storage key values on the s390
3958 architecture. The ioctl takes parameters via 3489 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3959 3490
3960 struct kvm_s390_skeys { 3491 struct kvm_s390_skeys {
3961 __u64 start_gfn; 3492 __u64 start_gfn;
3962 __u64 count; 3493 __u64 count;
3963 __u64 skeydata_addr; 3494 __u64 skeydata_addr;
3964 __u32 flags; 3495 __u32 flags;
3965 __u32 reserved[9]; 3496 __u32 reserved[9];
3966 }; 3497 };
3967 3498
3968 The start_gfn field is the number of the firs 3499 The start_gfn field is the number of the first guest frame whose storage keys
3969 you want to get. 3500 you want to get.
3970 3501
3971 The count field is the number of consecutive 3502 The count field is the number of consecutive frames (starting from start_gfn)
3972 whose storage keys to get. The count field mu 3503 whose storage keys to get. The count field must be at least 1 and the maximum
3973 allowed value is defined as KVM_S390_SKEYS_MA !! 3504 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3974 will cause the ioctl to return -EINVAL. 3505 will cause the ioctl to return -EINVAL.
3975 3506
3976 The skeydata_addr field is the address to a b 3507 The skeydata_addr field is the address to a buffer large enough to hold count
3977 bytes. This buffer will be filled with storag 3508 bytes. This buffer will be filled with storage key data by the ioctl.
3978 3509
3979 4.91 KVM_S390_SET_SKEYS 3510 4.91 KVM_S390_SET_SKEYS
3980 ----------------------- 3511 -----------------------
3981 3512
3982 :Capability: KVM_CAP_S390_SKEYS 3513 :Capability: KVM_CAP_S390_SKEYS
3983 :Architectures: s390 3514 :Architectures: s390
3984 :Type: vm ioctl 3515 :Type: vm ioctl
3985 :Parameters: struct kvm_s390_skeys 3516 :Parameters: struct kvm_s390_skeys
3986 :Returns: 0 on success, negative value on err 3517 :Returns: 0 on success, negative value on error
3987 3518
3988 This ioctl is used to set guest storage key v 3519 This ioctl is used to set guest storage key values on the s390
3989 architecture. The ioctl takes parameters via 3520 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3990 See section on KVM_S390_GET_SKEYS for struct 3521 See section on KVM_S390_GET_SKEYS for struct definition.
3991 3522
3992 The start_gfn field is the number of the firs 3523 The start_gfn field is the number of the first guest frame whose storage keys
3993 you want to set. 3524 you want to set.
3994 3525
3995 The count field is the number of consecutive 3526 The count field is the number of consecutive frames (starting from start_gfn)
3996 whose storage keys to get. The count field mu 3527 whose storage keys to get. The count field must be at least 1 and the maximum
3997 allowed value is defined as KVM_S390_SKEYS_MA !! 3528 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3998 will cause the ioctl to return -EINVAL. 3529 will cause the ioctl to return -EINVAL.
3999 3530
4000 The skeydata_addr field is the address to a b 3531 The skeydata_addr field is the address to a buffer containing count bytes of
4001 storage keys. Each byte in the buffer will be 3532 storage keys. Each byte in the buffer will be set as the storage key for a
4002 single frame starting at start_gfn for count 3533 single frame starting at start_gfn for count frames.
4003 3534
4004 Note: If any architecturally invalid key valu 3535 Note: If any architecturally invalid key value is found in the given data then
4005 the ioctl will return -EINVAL. 3536 the ioctl will return -EINVAL.
4006 3537
4007 4.92 KVM_S390_IRQ 3538 4.92 KVM_S390_IRQ
4008 ----------------- 3539 -----------------
4009 3540
4010 :Capability: KVM_CAP_S390_INJECT_IRQ 3541 :Capability: KVM_CAP_S390_INJECT_IRQ
4011 :Architectures: s390 3542 :Architectures: s390
4012 :Type: vcpu ioctl 3543 :Type: vcpu ioctl
4013 :Parameters: struct kvm_s390_irq (in) 3544 :Parameters: struct kvm_s390_irq (in)
4014 :Returns: 0 on success, -1 on error 3545 :Returns: 0 on success, -1 on error
4015 3546
4016 Errors: 3547 Errors:
4017 3548
4018 3549
4019 ====== =================================== 3550 ====== =================================================================
4020 EINVAL interrupt type is invalid 3551 EINVAL interrupt type is invalid
4021 type is KVM_S390_SIGP_STOP and flag 3552 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
4022 type is KVM_S390_INT_EXTERNAL_CALL 3553 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
4023 than the maximum of VCPUs 3554 than the maximum of VCPUs
4024 EBUSY type is KVM_S390_SIGP_SET_PREFIX an 3555 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
4025 type is KVM_S390_SIGP_STOP and a st 3556 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
4026 type is KVM_S390_INT_EXTERNAL_CALL 3557 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
4027 is already pending 3558 is already pending
4028 ====== =================================== 3559 ====== =================================================================
4029 3560
4030 Allows to inject an interrupt to the guest. 3561 Allows to inject an interrupt to the guest.
4031 3562
4032 Using struct kvm_s390_irq as a parameter allo 3563 Using struct kvm_s390_irq as a parameter allows
4033 to inject additional payload which is not 3564 to inject additional payload which is not
4034 possible via KVM_S390_INTERRUPT. 3565 possible via KVM_S390_INTERRUPT.
4035 3566
4036 Interrupt parameters are passed via kvm_s390_ 3567 Interrupt parameters are passed via kvm_s390_irq::
4037 3568
4038 struct kvm_s390_irq { 3569 struct kvm_s390_irq {
4039 __u64 type; 3570 __u64 type;
4040 union { 3571 union {
4041 struct kvm_s390_io_info io; 3572 struct kvm_s390_io_info io;
4042 struct kvm_s390_ext_info ext; 3573 struct kvm_s390_ext_info ext;
4043 struct kvm_s390_pgm_info pgm; 3574 struct kvm_s390_pgm_info pgm;
4044 struct kvm_s390_emerg_info em 3575 struct kvm_s390_emerg_info emerg;
4045 struct kvm_s390_extcall_info 3576 struct kvm_s390_extcall_info extcall;
4046 struct kvm_s390_prefix_info p 3577 struct kvm_s390_prefix_info prefix;
4047 struct kvm_s390_stop_info sto 3578 struct kvm_s390_stop_info stop;
4048 struct kvm_s390_mchk_info mch 3579 struct kvm_s390_mchk_info mchk;
4049 char reserved[64]; 3580 char reserved[64];
4050 } u; 3581 } u;
4051 }; 3582 };
4052 3583
4053 type can be one of the following: 3584 type can be one of the following:
4054 3585
4055 - KVM_S390_SIGP_STOP - sigp stop; parameter i 3586 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
4056 - KVM_S390_PROGRAM_INT - program check; param 3587 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
4057 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; 3588 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
4058 - KVM_S390_RESTART - restart; no parameters 3589 - KVM_S390_RESTART - restart; no parameters
4059 - KVM_S390_INT_CLOCK_COMP - clock comparator 3590 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
4060 - KVM_S390_INT_CPU_TIMER - CPU timer interrup 3591 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
4061 - KVM_S390_INT_EMERGENCY - sigp emergency; pa 3592 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
4062 - KVM_S390_INT_EXTERNAL_CALL - sigp external 3593 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
4063 - KVM_S390_MCHK - machine check interrupt; pa 3594 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
4064 3595
4065 This is an asynchronous vcpu ioctl and can be 3596 This is an asynchronous vcpu ioctl and can be invoked from any thread.
4066 3597
4067 4.94 KVM_S390_GET_IRQ_STATE 3598 4.94 KVM_S390_GET_IRQ_STATE
4068 --------------------------- 3599 ---------------------------
4069 3600
4070 :Capability: KVM_CAP_S390_IRQ_STATE 3601 :Capability: KVM_CAP_S390_IRQ_STATE
4071 :Architectures: s390 3602 :Architectures: s390
4072 :Type: vcpu ioctl 3603 :Type: vcpu ioctl
4073 :Parameters: struct kvm_s390_irq_state (out) 3604 :Parameters: struct kvm_s390_irq_state (out)
4074 :Returns: >= number of bytes copied into buff 3605 :Returns: >= number of bytes copied into buffer,
4075 -EINVAL if buffer size is 0, 3606 -EINVAL if buffer size is 0,
4076 -ENOBUFS if buffer size is too smal 3607 -ENOBUFS if buffer size is too small to fit all pending interrupts,
4077 -EFAULT if the buffer address was i 3608 -EFAULT if the buffer address was invalid
4078 3609
4079 This ioctl allows userspace to retrieve the c 3610 This ioctl allows userspace to retrieve the complete state of all currently
4080 pending interrupts in a single buffer. Use ca 3611 pending interrupts in a single buffer. Use cases include migration
4081 and introspection. The parameter structure co 3612 and introspection. The parameter structure contains the address of a
4082 userspace buffer and its length:: 3613 userspace buffer and its length::
4083 3614
4084 struct kvm_s390_irq_state { 3615 struct kvm_s390_irq_state {
4085 __u64 buf; 3616 __u64 buf;
4086 __u32 flags; /* will stay unus 3617 __u32 flags; /* will stay unused for compatibility reasons */
4087 __u32 len; 3618 __u32 len;
4088 __u32 reserved[4]; /* will stay unus 3619 __u32 reserved[4]; /* will stay unused for compatibility reasons */
4089 }; 3620 };
4090 3621
4091 Userspace passes in the above struct and for 3622 Userspace passes in the above struct and for each pending interrupt a
4092 struct kvm_s390_irq is copied to the provided 3623 struct kvm_s390_irq is copied to the provided buffer.
4093 3624
4094 The structure contains a flags and a reserved 3625 The structure contains a flags and a reserved field for future extensions. As
4095 the kernel never checked for flags == 0 and Q 3626 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
4096 reserved, these fields can not be used in the 3627 reserved, these fields can not be used in the future without breaking
4097 compatibility. 3628 compatibility.
4098 3629
4099 If -ENOBUFS is returned the buffer provided w 3630 If -ENOBUFS is returned the buffer provided was too small and userspace
4100 may retry with a bigger buffer. 3631 may retry with a bigger buffer.
4101 3632
4102 4.95 KVM_S390_SET_IRQ_STATE 3633 4.95 KVM_S390_SET_IRQ_STATE
4103 --------------------------- 3634 ---------------------------
4104 3635
4105 :Capability: KVM_CAP_S390_IRQ_STATE 3636 :Capability: KVM_CAP_S390_IRQ_STATE
4106 :Architectures: s390 3637 :Architectures: s390
4107 :Type: vcpu ioctl 3638 :Type: vcpu ioctl
4108 :Parameters: struct kvm_s390_irq_state (in) 3639 :Parameters: struct kvm_s390_irq_state (in)
4109 :Returns: 0 on success, 3640 :Returns: 0 on success,
4110 -EFAULT if the buffer address was i 3641 -EFAULT if the buffer address was invalid,
4111 -EINVAL for an invalid buffer lengt 3642 -EINVAL for an invalid buffer length (see below),
4112 -EBUSY if there were already interr 3643 -EBUSY if there were already interrupts pending,
4113 errors occurring when actually inje 3644 errors occurring when actually injecting the
4114 interrupt. See KVM_S390_IRQ. 3645 interrupt. See KVM_S390_IRQ.
4115 3646
4116 This ioctl allows userspace to set the comple 3647 This ioctl allows userspace to set the complete state of all cpu-local
4117 interrupts currently pending for the vcpu. It 3648 interrupts currently pending for the vcpu. It is intended for restoring
4118 interrupt state after a migration. The input 3649 interrupt state after a migration. The input parameter is a userspace buffer
4119 containing a struct kvm_s390_irq_state:: 3650 containing a struct kvm_s390_irq_state::
4120 3651
4121 struct kvm_s390_irq_state { 3652 struct kvm_s390_irq_state {
4122 __u64 buf; 3653 __u64 buf;
4123 __u32 flags; /* will stay unus 3654 __u32 flags; /* will stay unused for compatibility reasons */
4124 __u32 len; 3655 __u32 len;
4125 __u32 reserved[4]; /* will stay unus 3656 __u32 reserved[4]; /* will stay unused for compatibility reasons */
4126 }; 3657 };
4127 3658
4128 The restrictions for flags and reserved apply 3659 The restrictions for flags and reserved apply as well.
4129 (see KVM_S390_GET_IRQ_STATE) 3660 (see KVM_S390_GET_IRQ_STATE)
4130 3661
4131 The userspace memory referenced by buf contai 3662 The userspace memory referenced by buf contains a struct kvm_s390_irq
4132 for each interrupt to be injected into the gu 3663 for each interrupt to be injected into the guest.
4133 If one of the interrupts could not be injecte 3664 If one of the interrupts could not be injected for some reason the
4134 ioctl aborts. 3665 ioctl aborts.
4135 3666
4136 len must be a multiple of sizeof(struct kvm_s 3667 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
4137 and it must not exceed (max_vcpus + 32) * siz 3668 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
4138 which is the maximum number of possibly pendi 3669 which is the maximum number of possibly pending cpu-local interrupts.
4139 3670
4140 4.96 KVM_SMI 3671 4.96 KVM_SMI
4141 ------------ 3672 ------------
4142 3673
4143 :Capability: KVM_CAP_X86_SMM 3674 :Capability: KVM_CAP_X86_SMM
4144 :Architectures: x86 3675 :Architectures: x86
4145 :Type: vcpu ioctl 3676 :Type: vcpu ioctl
4146 :Parameters: none 3677 :Parameters: none
4147 :Returns: 0 on success, -1 on error 3678 :Returns: 0 on success, -1 on error
4148 3679
4149 Queues an SMI on the thread's vcpu. 3680 Queues an SMI on the thread's vcpu.
4150 3681
4151 4.97 KVM_X86_SET_MSR_FILTER !! 3682 4.97 KVM_CAP_PPC_MULTITCE
4152 ---------------------------- !! 3683 -------------------------
4153 <<
4154 :Capability: KVM_CAP_X86_MSR_FILTER <<
4155 :Architectures: x86 <<
4156 :Type: vm ioctl <<
4157 :Parameters: struct kvm_msr_filter <<
4158 :Returns: 0 on success, < 0 on error <<
4159 <<
4160 :: <<
4161 <<
4162 struct kvm_msr_filter_range { <<
4163 #define KVM_MSR_FILTER_READ (1 << 0) <<
4164 #define KVM_MSR_FILTER_WRITE (1 << 1) <<
4165 __u32 flags; <<
4166 __u32 nmsrs; /* number of msrs in bit <<
4167 __u32 base; /* MSR index the bitmap <<
4168 __u8 *bitmap; /* a 1 bit allows the o <<
4169 }; <<
4170 <<
4171 #define KVM_MSR_FILTER_MAX_RANGES 16 <<
4172 struct kvm_msr_filter { <<
4173 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << <<
4174 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << <<
4175 __u32 flags; <<
4176 struct kvm_msr_filter_range ranges[KV <<
4177 }; <<
4178 <<
4179 flags values for ``struct kvm_msr_filter_rang <<
4180 <<
4181 ``KVM_MSR_FILTER_READ`` <<
4182 <<
4183 Filter read accesses to MSRs using the give <<
4184 indicates that read accesses should be deni <<
4185 a read for a particular MSR should be allow <<
4186 filter action. <<
4187 <<
4188 ``KVM_MSR_FILTER_WRITE`` <<
4189 <<
4190 Filter write accesses to MSRs using the giv <<
4191 indicates that write accesses should be den <<
4192 a write for a particular MSR should be allo <<
4193 filter action. <<
4194 <<
4195 flags values for ``struct kvm_msr_filter``: <<
4196 <<
4197 ``KVM_MSR_FILTER_DEFAULT_ALLOW`` <<
4198 <<
4199 If no filter range matches an MSR index tha <<
4200 allow accesses to all MSRs by default. <<
4201 <<
4202 ``KVM_MSR_FILTER_DEFAULT_DENY`` <<
4203 <<
4204 If no filter range matches an MSR index tha <<
4205 deny accesses to all MSRs by default. <<
4206 <<
4207 This ioctl allows userspace to define up to 1 <<
4208 guest MSR accesses that would normally be all <<
4209 covered by a specific range, the "default" fi <<
4210 bitmap range covers MSRs from [base .. base+n <<
4211 <<
4212 If an MSR access is denied by userspace, the <<
4213 whether or not KVM_CAP_X86_USER_SPACE_MSR's K <<
4214 enabled. If KVM_MSR_EXIT_REASON_FILTER is en <<
4215 on denied accesses, i.e. userspace effectivel <<
4216 KVM_MSR_EXIT_REASON_FILTER is not enabled, KV <<
4217 on denied accesses. Note, if an MSR access i <<
4218 load/stores during VMX transitions, KVM ignor <<
4219 See the below warning for full details. <<
4220 <<
4221 If an MSR access is allowed by userspace, KVM <<
4222 the access in accordance with the vCPU model. <<
4223 inject a #GP if an access is allowed by users <<
4224 the MSR, or to follow architectural behavior <<
4225 <<
4226 By default, KVM operates in KVM_MSR_FILTER_DE <<
4227 filters. <<
4228 3684
4229 Calling this ioctl with an empty set of range !! 3685 :Capability: KVM_CAP_PPC_MULTITCE
4230 filtering. In that mode, ``KVM_MSR_FILTER_DEF !! 3686 :Architectures: ppc
4231 an error. !! 3687 :Type: vm
4232 3688
4233 .. warning:: !! 3689 This capability means the kernel is capable of handling hypercalls
4234 MSR accesses that are side effects of inst !! 3690 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
4235 native) are not filtered as hardware does !! 3691 space. This significantly accelerates DMA operations for PPC KVM guests.
4236 RDMSR and WRMSR, and KVM mimics that behav !! 3692 User space should expect that its handlers for these hypercalls
4237 to avoid pointless divergence from hardwar !! 3693 are not going to be called if user space previously registered LIOBN
4238 SYSENTER reads the SYSENTER MSRs, etc. !! 3694 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
4239 <<
4240 MSRs that are loaded/stored via dedicated <<
4241 part of VM-Enter/VM-Exit emulation. <<
4242 <<
4243 MSRs that are loaded/store via VMX's load/ <<
4244 of VM-Enter/VM-Exit emulation. If an MSR <<
4245 synthesizes a consistency check VM-Exit(EX <<
4246 MSR access is denied on VM-Exit, KVM synth <<
4247 extends Intel's architectural list of MSRs <<
4248 the VM-Enter/VM-Exit MSR list. It is plat <<
4249 to communicate any such restrictions to th <<
4250 3695
4251 x2APIC MSR accesses cannot be filtered (KV !! 3696 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
4252 cover any x2APIC MSRs). !! 3697 user space might have to advertise it for the guest. For example,
>> 3698 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
>> 3699 present in the "ibm,hypertas-functions" device-tree property.
4253 3700
4254 Note, invoking this ioctl while a vCPU is run !! 3701 The hypercalls mentioned above may or may not be processed successfully
4255 KVM does guarantee that vCPUs will see either !! 3702 in the kernel based fast path. If they can not be handled by the kernel,
4256 filter, e.g. MSRs with identical settings in !! 3703 they will get passed on to user space. So user space still has to have
4257 have deterministic behavior. !! 3704 an implementation for these despite the in kernel acceleration.
4258 3705
4259 Similarly, if userspace wishes to intercept o !! 3706 This capability is always enabled.
4260 KVM_MSR_EXIT_REASON_FILTER must be enabled be <<
4261 left enabled until after all filters are deac <<
4262 result in KVM injecting a #GP instead of exit <<
4263 3707
4264 4.98 KVM_CREATE_SPAPR_TCE_64 3708 4.98 KVM_CREATE_SPAPR_TCE_64
4265 ---------------------------- 3709 ----------------------------
4266 3710
4267 :Capability: KVM_CAP_SPAPR_TCE_64 3711 :Capability: KVM_CAP_SPAPR_TCE_64
4268 :Architectures: powerpc 3712 :Architectures: powerpc
4269 :Type: vm ioctl 3713 :Type: vm ioctl
4270 :Parameters: struct kvm_create_spapr_tce_64 ( 3714 :Parameters: struct kvm_create_spapr_tce_64 (in)
4271 :Returns: file descriptor for manipulating th 3715 :Returns: file descriptor for manipulating the created TCE table
4272 3716
4273 This is an extension for KVM_CAP_SPAPR_TCE wh 3717 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
4274 windows, described in 4.62 KVM_CREATE_SPAPR_T 3718 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
4275 3719
4276 This capability uses extended struct in ioctl 3720 This capability uses extended struct in ioctl interface::
4277 3721
4278 /* for KVM_CAP_SPAPR_TCE_64 */ 3722 /* for KVM_CAP_SPAPR_TCE_64 */
4279 struct kvm_create_spapr_tce_64 { 3723 struct kvm_create_spapr_tce_64 {
4280 __u64 liobn; 3724 __u64 liobn;
4281 __u32 page_shift; 3725 __u32 page_shift;
4282 __u32 flags; 3726 __u32 flags;
4283 __u64 offset; /* in pages */ 3727 __u64 offset; /* in pages */
4284 __u64 size; /* in pages */ 3728 __u64 size; /* in pages */
4285 }; 3729 };
4286 3730
4287 The aim of extension is to support an additio 3731 The aim of extension is to support an additional bigger DMA window with
4288 a variable page size. 3732 a variable page size.
4289 KVM_CREATE_SPAPR_TCE_64 receives a 64bit wind 3733 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
4290 a bus offset of the corresponding DMA window, 3734 a bus offset of the corresponding DMA window, @size and @offset are numbers
4291 of IOMMU pages. 3735 of IOMMU pages.
4292 3736
4293 @flags are not used at the moment. 3737 @flags are not used at the moment.
4294 3738
4295 The rest of functionality is identical to KVM 3739 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
4296 3740
4297 4.99 KVM_REINJECT_CONTROL 3741 4.99 KVM_REINJECT_CONTROL
4298 ------------------------- 3742 -------------------------
4299 3743
4300 :Capability: KVM_CAP_REINJECT_CONTROL 3744 :Capability: KVM_CAP_REINJECT_CONTROL
4301 :Architectures: x86 3745 :Architectures: x86
4302 :Type: vm ioctl 3746 :Type: vm ioctl
4303 :Parameters: struct kvm_reinject_control (in) 3747 :Parameters: struct kvm_reinject_control (in)
4304 :Returns: 0 on success, 3748 :Returns: 0 on success,
4305 -EFAULT if struct kvm_reinject_contr 3749 -EFAULT if struct kvm_reinject_control cannot be read,
4306 -ENXIO if KVM_CREATE_PIT or KVM_CREA 3750 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
4307 3751
4308 i8254 (PIT) has two modes, reinject and !rein 3752 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
4309 where KVM queues elapsed i8254 ticks and moni 3753 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
4310 vector(s) that i8254 injects. Reinject mode 3754 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
4311 interrupt whenever there isn't a pending inte 3755 interrupt whenever there isn't a pending interrupt from i8254.
4312 !reinject mode injects an interrupt as soon a 3756 !reinject mode injects an interrupt as soon as a tick arrives.
4313 3757
4314 :: 3758 ::
4315 3759
4316 struct kvm_reinject_control { 3760 struct kvm_reinject_control {
4317 __u8 pit_reinject; 3761 __u8 pit_reinject;
4318 __u8 reserved[31]; 3762 __u8 reserved[31];
4319 }; 3763 };
4320 3764
4321 pit_reinject = 0 (!reinject mode) is recommen 3765 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
4322 operating system that uses the PIT for timing 3766 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
4323 3767
4324 4.100 KVM_PPC_CONFIGURE_V3_MMU 3768 4.100 KVM_PPC_CONFIGURE_V3_MMU
4325 ------------------------------ 3769 ------------------------------
4326 3770
4327 :Capability: KVM_CAP_PPC_MMU_RADIX or KVM_CAP !! 3771 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
4328 :Architectures: ppc 3772 :Architectures: ppc
4329 :Type: vm ioctl 3773 :Type: vm ioctl
4330 :Parameters: struct kvm_ppc_mmuv3_cfg (in) 3774 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
4331 :Returns: 0 on success, 3775 :Returns: 0 on success,
4332 -EFAULT if struct kvm_ppc_mmuv3_cfg 3776 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
4333 -EINVAL if the configuration is inva 3777 -EINVAL if the configuration is invalid
4334 3778
4335 This ioctl controls whether the guest will us 3779 This ioctl controls whether the guest will use radix or HPT (hashed
4336 page table) translation, and sets the pointer 3780 page table) translation, and sets the pointer to the process table for
4337 the guest. 3781 the guest.
4338 3782
4339 :: 3783 ::
4340 3784
4341 struct kvm_ppc_mmuv3_cfg { 3785 struct kvm_ppc_mmuv3_cfg {
4342 __u64 flags; 3786 __u64 flags;
4343 __u64 process_table; 3787 __u64 process_table;
4344 }; 3788 };
4345 3789
4346 There are two bits that can be set in flags; 3790 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
4347 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if 3791 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
4348 to use radix tree translation, and if clear, 3792 to use radix tree translation, and if clear, to use HPT translation.
4349 KVM_PPC_MMUV3_GTSE, if set and if KVM permits 3793 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
4350 to be able to use the global TLB and SLB inva 3794 to be able to use the global TLB and SLB invalidation instructions;
4351 if clear, the guest may not use these instruc 3795 if clear, the guest may not use these instructions.
4352 3796
4353 The process_table field specifies the address 3797 The process_table field specifies the address and size of the guest
4354 process table, which is in the guest's space. 3798 process table, which is in the guest's space. This field is formatted
4355 as the second doubleword of the partition tab 3799 as the second doubleword of the partition table entry, as defined in
4356 the Power ISA V3.00, Book III section 5.7.6.1 3800 the Power ISA V3.00, Book III section 5.7.6.1.
4357 3801
4358 4.101 KVM_PPC_GET_RMMU_INFO 3802 4.101 KVM_PPC_GET_RMMU_INFO
4359 --------------------------- 3803 ---------------------------
4360 3804
4361 :Capability: KVM_CAP_PPC_MMU_RADIX !! 3805 :Capability: KVM_CAP_PPC_RADIX_MMU
4362 :Architectures: ppc 3806 :Architectures: ppc
4363 :Type: vm ioctl 3807 :Type: vm ioctl
4364 :Parameters: struct kvm_ppc_rmmu_info (out) 3808 :Parameters: struct kvm_ppc_rmmu_info (out)
4365 :Returns: 0 on success, 3809 :Returns: 0 on success,
4366 -EFAULT if struct kvm_ppc_rmmu_info 3810 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
4367 -EINVAL if no useful information can 3811 -EINVAL if no useful information can be returned
4368 3812
4369 This ioctl returns a structure containing two 3813 This ioctl returns a structure containing two things: (a) a list
4370 containing supported radix tree geometries, a 3814 containing supported radix tree geometries, and (b) a list that maps
4371 page sizes to put in the "AP" (actual page si 3815 page sizes to put in the "AP" (actual page size) field for the tlbie
4372 (TLB invalidate entry) instruction. 3816 (TLB invalidate entry) instruction.
4373 3817
4374 :: 3818 ::
4375 3819
4376 struct kvm_ppc_rmmu_info { 3820 struct kvm_ppc_rmmu_info {
4377 struct kvm_ppc_radix_geom { 3821 struct kvm_ppc_radix_geom {
4378 __u8 page_shift; 3822 __u8 page_shift;
4379 __u8 level_bits[4]; 3823 __u8 level_bits[4];
4380 __u8 pad[3]; 3824 __u8 pad[3];
4381 } geometries[8]; 3825 } geometries[8];
4382 __u32 ap_encodings[8]; 3826 __u32 ap_encodings[8];
4383 }; 3827 };
4384 3828
4385 The geometries[] field gives up to 8 supporte 3829 The geometries[] field gives up to 8 supported geometries for the
4386 radix page table, in terms of the log base 2 3830 radix page table, in terms of the log base 2 of the smallest page
4387 size, and the number of bits indexed at each 3831 size, and the number of bits indexed at each level of the tree, from
4388 the PTE level up to the PGD level in that ord 3832 the PTE level up to the PGD level in that order. Any unused entries
4389 will have 0 in the page_shift field. 3833 will have 0 in the page_shift field.
4390 3834
4391 The ap_encodings gives the supported page siz 3835 The ap_encodings gives the supported page sizes and their AP field
4392 encodings, encoded with the AP value in the t 3836 encodings, encoded with the AP value in the top 3 bits and the log
4393 base 2 of the page size in the bottom 6 bits. 3837 base 2 of the page size in the bottom 6 bits.
4394 3838
4395 4.102 KVM_PPC_RESIZE_HPT_PREPARE 3839 4.102 KVM_PPC_RESIZE_HPT_PREPARE
4396 -------------------------------- 3840 --------------------------------
4397 3841
4398 :Capability: KVM_CAP_SPAPR_RESIZE_HPT 3842 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4399 :Architectures: powerpc 3843 :Architectures: powerpc
4400 :Type: vm ioctl 3844 :Type: vm ioctl
4401 :Parameters: struct kvm_ppc_resize_hpt (in) 3845 :Parameters: struct kvm_ppc_resize_hpt (in)
4402 :Returns: 0 on successful completion, 3846 :Returns: 0 on successful completion,
4403 >0 if a new HPT is being prepared, t 3847 >0 if a new HPT is being prepared, the value is an estimated
4404 number of milliseconds until prepara 3848 number of milliseconds until preparation is complete,
4405 -EFAULT if struct kvm_reinject_contr 3849 -EFAULT if struct kvm_reinject_control cannot be read,
4406 -EINVAL if the supplied shift or fla 3850 -EINVAL if the supplied shift or flags are invalid,
4407 -ENOMEM if unable to allocate the ne 3851 -ENOMEM if unable to allocate the new HPT,
>> 3852 -ENOSPC if there was a hash collision
>> 3853
>> 3854 ::
>> 3855
>> 3856 struct kvm_ppc_rmmu_info {
>> 3857 struct kvm_ppc_radix_geom {
>> 3858 __u8 page_shift;
>> 3859 __u8 level_bits[4];
>> 3860 __u8 pad[3];
>> 3861 } geometries[8];
>> 3862 __u32 ap_encodings[8];
>> 3863 };
>> 3864
>> 3865 The geometries[] field gives up to 8 supported geometries for the
>> 3866 radix page table, in terms of the log base 2 of the smallest page
>> 3867 size, and the number of bits indexed at each level of the tree, from
>> 3868 the PTE level up to the PGD level in that order. Any unused entries
>> 3869 will have 0 in the page_shift field.
>> 3870
>> 3871 The ap_encodings gives the supported page sizes and their AP field
>> 3872 encodings, encoded with the AP value in the top 3 bits and the log
>> 3873 base 2 of the page size in the bottom 6 bits.
>> 3874
>> 3875 4.102 KVM_PPC_RESIZE_HPT_PREPARE
>> 3876 --------------------------------
>> 3877
>> 3878 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
>> 3879 :Architectures: powerpc
>> 3880 :Type: vm ioctl
>> 3881 :Parameters: struct kvm_ppc_resize_hpt (in)
>> 3882 :Returns: 0 on successful completion,
>> 3883 >0 if a new HPT is being prepared, the value is an estimated
>> 3884 number of milliseconds until preparation is complete,
>> 3885 -EFAULT if struct kvm_reinject_control cannot be read,
>> 3886 -EINVAL if the supplied shift or flags are invalid,when moving existing
>> 3887 HPT entries to the new HPT,
>> 3888 -EIO on other error conditions
4408 3889
4409 Used to implement the PAPR extension for runt 3890 Used to implement the PAPR extension for runtime resizing of a guest's
4410 Hashed Page Table (HPT). Specifically this s 3891 Hashed Page Table (HPT). Specifically this starts, stops or monitors
4411 the preparation of a new potential HPT for th 3892 the preparation of a new potential HPT for the guest, essentially
4412 implementing the H_RESIZE_HPT_PREPARE hyperca 3893 implementing the H_RESIZE_HPT_PREPARE hypercall.
4413 3894
4414 :: <<
4415 <<
4416 struct kvm_ppc_resize_hpt { <<
4417 __u64 flags; <<
4418 __u32 shift; <<
4419 __u32 pad; <<
4420 }; <<
4421 <<
4422 If called with shift > 0 when there is no pen 3895 If called with shift > 0 when there is no pending HPT for the guest,
4423 this begins preparation of a new pending HPT 3896 this begins preparation of a new pending HPT of size 2^(shift) bytes.
4424 It then returns a positive integer with the e 3897 It then returns a positive integer with the estimated number of
4425 milliseconds until preparation is complete. 3898 milliseconds until preparation is complete.
4426 3899
4427 If called when there is a pending HPT whose s 3900 If called when there is a pending HPT whose size does not match that
4428 requested in the parameters, discards the exi 3901 requested in the parameters, discards the existing pending HPT and
4429 creates a new one as above. 3902 creates a new one as above.
4430 3903
4431 If called when there is a pending HPT of the 3904 If called when there is a pending HPT of the size requested, will:
4432 3905
4433 * If preparation of the pending HPT is alre 3906 * If preparation of the pending HPT is already complete, return 0
4434 * If preparation of the pending HPT has fai 3907 * If preparation of the pending HPT has failed, return an error
4435 code, then discard the pending HPT. 3908 code, then discard the pending HPT.
4436 * If preparation of the pending HPT is stil 3909 * If preparation of the pending HPT is still in progress, return an
4437 estimated number of milliseconds until pr 3910 estimated number of milliseconds until preparation is complete.
4438 3911
4439 If called with shift == 0, discards any curre 3912 If called with shift == 0, discards any currently pending HPT and
4440 returns 0 (i.e. cancels any in-progress prepa 3913 returns 0 (i.e. cancels any in-progress preparation).
4441 3914
4442 flags is reserved for future expansion, curre 3915 flags is reserved for future expansion, currently setting any bits in
4443 flags will result in an -EINVAL. 3916 flags will result in an -EINVAL.
4444 3917
4445 Normally this will be called repeatedly with 3918 Normally this will be called repeatedly with the same parameters until
4446 it returns <= 0. The first call will initiat 3919 it returns <= 0. The first call will initiate preparation, subsequent
4447 ones will monitor preparation until it comple 3920 ones will monitor preparation until it completes or fails.
4448 3921
>> 3922 ::
>> 3923
>> 3924 struct kvm_ppc_resize_hpt {
>> 3925 __u64 flags;
>> 3926 __u32 shift;
>> 3927 __u32 pad;
>> 3928 };
>> 3929
4449 4.103 KVM_PPC_RESIZE_HPT_COMMIT 3930 4.103 KVM_PPC_RESIZE_HPT_COMMIT
4450 ------------------------------- 3931 -------------------------------
4451 3932
4452 :Capability: KVM_CAP_SPAPR_RESIZE_HPT 3933 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4453 :Architectures: powerpc 3934 :Architectures: powerpc
4454 :Type: vm ioctl 3935 :Type: vm ioctl
4455 :Parameters: struct kvm_ppc_resize_hpt (in) 3936 :Parameters: struct kvm_ppc_resize_hpt (in)
4456 :Returns: 0 on successful completion, 3937 :Returns: 0 on successful completion,
4457 -EFAULT if struct kvm_reinject_contr 3938 -EFAULT if struct kvm_reinject_control cannot be read,
4458 -EINVAL if the supplied shift or fla 3939 -EINVAL if the supplied shift or flags are invalid,
4459 -ENXIO is there is no pending HPT, o 3940 -ENXIO is there is no pending HPT, or the pending HPT doesn't
4460 have the requested size, 3941 have the requested size,
4461 -EBUSY if the pending HPT is not ful 3942 -EBUSY if the pending HPT is not fully prepared,
4462 -ENOSPC if there was a hash collisio 3943 -ENOSPC if there was a hash collision when moving existing
4463 HPT entries to the new HPT, 3944 HPT entries to the new HPT,
4464 -EIO on other error conditions 3945 -EIO on other error conditions
4465 3946
4466 Used to implement the PAPR extension for runt 3947 Used to implement the PAPR extension for runtime resizing of a guest's
4467 Hashed Page Table (HPT). Specifically this r 3948 Hashed Page Table (HPT). Specifically this requests that the guest be
4468 transferred to working with the new HPT, esse 3949 transferred to working with the new HPT, essentially implementing the
4469 H_RESIZE_HPT_COMMIT hypercall. 3950 H_RESIZE_HPT_COMMIT hypercall.
4470 3951
4471 :: <<
4472 <<
4473 struct kvm_ppc_resize_hpt { <<
4474 __u64 flags; <<
4475 __u32 shift; <<
4476 __u32 pad; <<
4477 }; <<
4478 <<
4479 This should only be called after KVM_PPC_RESI 3952 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
4480 returned 0 with the same parameters. In othe 3953 returned 0 with the same parameters. In other cases
4481 KVM_PPC_RESIZE_HPT_COMMIT will return an erro 3954 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
4482 -EBUSY, though others may be possible if the 3955 -EBUSY, though others may be possible if the preparation was started,
4483 but failed). 3956 but failed).
4484 3957
4485 This will have undefined effects on the guest 3958 This will have undefined effects on the guest if it has not already
4486 placed itself in a quiescent state where no v 3959 placed itself in a quiescent state where no vcpu will make MMU enabled
4487 memory accesses. 3960 memory accesses.
4488 3961
4489 On successful completion, the pending HPT wil !! 3962 On succsful completion, the pending HPT will become the guest's active
4490 HPT and the previous HPT will be discarded. 3963 HPT and the previous HPT will be discarded.
4491 3964
4492 On failure, the guest will still be operating 3965 On failure, the guest will still be operating on its previous HPT.
4493 3966
>> 3967 ::
>> 3968
>> 3969 struct kvm_ppc_resize_hpt {
>> 3970 __u64 flags;
>> 3971 __u32 shift;
>> 3972 __u32 pad;
>> 3973 };
>> 3974
4494 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED 3975 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
4495 ----------------------------------- 3976 -----------------------------------
4496 3977
4497 :Capability: KVM_CAP_MCE 3978 :Capability: KVM_CAP_MCE
4498 :Architectures: x86 3979 :Architectures: x86
4499 :Type: system ioctl 3980 :Type: system ioctl
4500 :Parameters: u64 mce_cap (out) 3981 :Parameters: u64 mce_cap (out)
4501 :Returns: 0 on success, -1 on error 3982 :Returns: 0 on success, -1 on error
4502 3983
4503 Returns supported MCE capabilities. The u64 m 3984 Returns supported MCE capabilities. The u64 mce_cap parameter
4504 has the same format as the MSR_IA32_MCG_CAP r 3985 has the same format as the MSR_IA32_MCG_CAP register. Supported
4505 capabilities will have the corresponding bits 3986 capabilities will have the corresponding bits set.
4506 3987
4507 4.105 KVM_X86_SETUP_MCE 3988 4.105 KVM_X86_SETUP_MCE
4508 ----------------------- 3989 -----------------------
4509 3990
4510 :Capability: KVM_CAP_MCE 3991 :Capability: KVM_CAP_MCE
4511 :Architectures: x86 3992 :Architectures: x86
4512 :Type: vcpu ioctl 3993 :Type: vcpu ioctl
4513 :Parameters: u64 mcg_cap (in) 3994 :Parameters: u64 mcg_cap (in)
4514 :Returns: 0 on success, 3995 :Returns: 0 on success,
4515 -EFAULT if u64 mcg_cap cannot be rea 3996 -EFAULT if u64 mcg_cap cannot be read,
4516 -EINVAL if the requested number of b 3997 -EINVAL if the requested number of banks is invalid,
4517 -EINVAL if requested MCE capability 3998 -EINVAL if requested MCE capability is not supported.
4518 3999
4519 Initializes MCE support for use. The u64 mcg_ 4000 Initializes MCE support for use. The u64 mcg_cap parameter
4520 has the same format as the MSR_IA32_MCG_CAP r 4001 has the same format as the MSR_IA32_MCG_CAP register and
4521 specifies which capabilities should be enable 4002 specifies which capabilities should be enabled. The maximum
4522 supported number of error-reporting banks can 4003 supported number of error-reporting banks can be retrieved when
4523 checking for KVM_CAP_MCE. The supported capab 4004 checking for KVM_CAP_MCE. The supported capabilities can be
4524 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED. 4005 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4525 4006
4526 4.106 KVM_X86_SET_MCE 4007 4.106 KVM_X86_SET_MCE
4527 --------------------- 4008 ---------------------
4528 4009
4529 :Capability: KVM_CAP_MCE 4010 :Capability: KVM_CAP_MCE
4530 :Architectures: x86 4011 :Architectures: x86
4531 :Type: vcpu ioctl 4012 :Type: vcpu ioctl
4532 :Parameters: struct kvm_x86_mce (in) 4013 :Parameters: struct kvm_x86_mce (in)
4533 :Returns: 0 on success, 4014 :Returns: 0 on success,
4534 -EFAULT if struct kvm_x86_mce cannot 4015 -EFAULT if struct kvm_x86_mce cannot be read,
4535 -EINVAL if the bank number is invali 4016 -EINVAL if the bank number is invalid,
4536 -EINVAL if VAL bit is not set in sta 4017 -EINVAL if VAL bit is not set in status field.
4537 4018
4538 Inject a machine check error (MCE) into the g 4019 Inject a machine check error (MCE) into the guest. The input
4539 parameter is:: 4020 parameter is::
4540 4021
4541 struct kvm_x86_mce { 4022 struct kvm_x86_mce {
4542 __u64 status; 4023 __u64 status;
4543 __u64 addr; 4024 __u64 addr;
4544 __u64 misc; 4025 __u64 misc;
4545 __u64 mcg_status; 4026 __u64 mcg_status;
4546 __u8 bank; 4027 __u8 bank;
4547 __u8 pad1[7]; 4028 __u8 pad1[7];
4548 __u64 pad2[3]; 4029 __u64 pad2[3];
4549 }; 4030 };
4550 4031
4551 If the MCE being reported is an uncorrected e 4032 If the MCE being reported is an uncorrected error, KVM will
4552 inject it as an MCE exception into the guest. 4033 inject it as an MCE exception into the guest. If the guest
4553 MCG_STATUS register reports that an MCE is in 4034 MCG_STATUS register reports that an MCE is in progress, KVM
4554 causes an KVM_EXIT_SHUTDOWN vmexit. 4035 causes an KVM_EXIT_SHUTDOWN vmexit.
4555 4036
4556 Otherwise, if the MCE is a corrected error, K 4037 Otherwise, if the MCE is a corrected error, KVM will just
4557 store it in the corresponding bank (provided 4038 store it in the corresponding bank (provided this bank is
4558 not holding a previously reported uncorrected 4039 not holding a previously reported uncorrected error).
4559 4040
4560 4.107 KVM_S390_GET_CMMA_BITS 4041 4.107 KVM_S390_GET_CMMA_BITS
4561 ---------------------------- 4042 ----------------------------
4562 4043
4563 :Capability: KVM_CAP_S390_CMMA_MIGRATION 4044 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4564 :Architectures: s390 4045 :Architectures: s390
4565 :Type: vm ioctl 4046 :Type: vm ioctl
4566 :Parameters: struct kvm_s390_cmma_log (in, ou 4047 :Parameters: struct kvm_s390_cmma_log (in, out)
4567 :Returns: 0 on success, a negative value on e 4048 :Returns: 0 on success, a negative value on error
4568 4049
4569 Errors: <<
4570 <<
4571 ====== ================================ <<
4572 ENOMEM not enough memory can be allocat <<
4573 ENXIO if CMMA is not enabled <<
4574 EINVAL if KVM_S390_CMMA_PEEK is not set <<
4575 EINVAL if KVM_S390_CMMA_PEEK is not set <<
4576 disabled (and thus migration mod <<
4577 EFAULT if the userspace address is inva <<
4578 present for the addresses (e.g. <<
4579 ====== ================================ <<
4580 <<
4581 This ioctl is used to get the values of the C 4050 This ioctl is used to get the values of the CMMA bits on the s390
4582 architecture. It is meant to be used in two s 4051 architecture. It is meant to be used in two scenarios:
4583 4052
4584 - During live migration to save the CMMA valu 4053 - During live migration to save the CMMA values. Live migration needs
4585 to be enabled via the KVM_REQ_START_MIGRATI 4054 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4586 - To non-destructively peek at the CMMA value 4055 - To non-destructively peek at the CMMA values, with the flag
4587 KVM_S390_CMMA_PEEK set. 4056 KVM_S390_CMMA_PEEK set.
4588 4057
4589 The ioctl takes parameters via the kvm_s390_c 4058 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4590 values are written to a buffer whose location 4059 values are written to a buffer whose location is indicated via the "values"
4591 member in the kvm_s390_cmma_log struct. The 4060 member in the kvm_s390_cmma_log struct. The values in the input struct are
4592 also updated as needed. 4061 also updated as needed.
4593 4062
4594 Each CMMA value takes up one byte. 4063 Each CMMA value takes up one byte.
4595 4064
4596 :: 4065 ::
4597 4066
4598 struct kvm_s390_cmma_log { 4067 struct kvm_s390_cmma_log {
4599 __u64 start_gfn; 4068 __u64 start_gfn;
4600 __u32 count; 4069 __u32 count;
4601 __u32 flags; 4070 __u32 flags;
4602 union { 4071 union {
4603 __u64 remaining; 4072 __u64 remaining;
4604 __u64 mask; 4073 __u64 mask;
4605 }; 4074 };
4606 __u64 values; 4075 __u64 values;
4607 }; 4076 };
4608 4077
4609 start_gfn is the number of the first guest fr 4078 start_gfn is the number of the first guest frame whose CMMA values are
4610 to be retrieved, 4079 to be retrieved,
4611 4080
4612 count is the length of the buffer in bytes, 4081 count is the length of the buffer in bytes,
4613 4082
4614 values points to the buffer where the result 4083 values points to the buffer where the result will be written to.
4615 4084
4616 If count is greater than KVM_S390_SKEYS_MAX, 4085 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4617 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re- 4086 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4618 other ioctls. 4087 other ioctls.
4619 4088
4620 The result is written in the buffer pointed t 4089 The result is written in the buffer pointed to by the field values, and
4621 the values of the input parameter are updated 4090 the values of the input parameter are updated as follows.
4622 4091
4623 Depending on the flags, different actions are 4092 Depending on the flags, different actions are performed. The only
4624 supported flag so far is KVM_S390_CMMA_PEEK. 4093 supported flag so far is KVM_S390_CMMA_PEEK.
4625 4094
4626 The default behaviour if KVM_S390_CMMA_PEEK i 4095 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4627 start_gfn will indicate the first page frame 4096 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4628 It is not necessarily the same as the one pas 4097 It is not necessarily the same as the one passed as input, as clean pages
4629 are skipped. 4098 are skipped.
4630 4099
4631 count will indicate the number of bytes actua 4100 count will indicate the number of bytes actually written in the buffer.
4632 It can (and very often will) be smaller than 4101 It can (and very often will) be smaller than the input value, since the
4633 buffer is only filled until 16 bytes of clean 4102 buffer is only filled until 16 bytes of clean values are found (which
4634 are then not copied in the buffer). Since a C 4103 are then not copied in the buffer). Since a CMMA migration block needs
4635 the base address and the length, for a total 4104 the base address and the length, for a total of 16 bytes, we will send
4636 back some clean data if there is some dirty d 4105 back some clean data if there is some dirty data afterwards, as long as
4637 the size of the clean data does not exceed th 4106 the size of the clean data does not exceed the size of the header. This
4638 allows to minimize the amount of data to be s 4107 allows to minimize the amount of data to be saved or transferred over
4639 the network at the expense of more roundtrips 4108 the network at the expense of more roundtrips to userspace. The next
4640 invocation of the ioctl will skip over all th 4109 invocation of the ioctl will skip over all the clean values, saving
4641 potentially more than just the 16 bytes we fo 4110 potentially more than just the 16 bytes we found.
4642 4111
4643 If KVM_S390_CMMA_PEEK is set: 4112 If KVM_S390_CMMA_PEEK is set:
4644 the existing storage attributes are read even 4113 the existing storage attributes are read even when not in migration
4645 mode, and no other action is performed; 4114 mode, and no other action is performed;
4646 4115
4647 the output start_gfn will be equal to the inp 4116 the output start_gfn will be equal to the input start_gfn,
4648 4117
4649 the output count will be equal to the input c 4118 the output count will be equal to the input count, except if the end of
4650 memory has been reached. 4119 memory has been reached.
4651 4120
4652 In both cases: 4121 In both cases:
4653 the field "remaining" will indicate the total 4122 the field "remaining" will indicate the total number of dirty CMMA values
4654 still remaining, or 0 if KVM_S390_CMMA_PEEK i 4123 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4655 not enabled. 4124 not enabled.
4656 4125
4657 mask is unused. 4126 mask is unused.
4658 4127
4659 values points to the userspace buffer where t 4128 values points to the userspace buffer where the result will be stored.
4660 4129
>> 4130 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
>> 4131 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
>> 4132 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
>> 4133 -EFAULT if the userspace address is invalid or if no page table is
>> 4134 present for the addresses (e.g. when using hugepages).
>> 4135
4661 4.108 KVM_S390_SET_CMMA_BITS 4136 4.108 KVM_S390_SET_CMMA_BITS
4662 ---------------------------- 4137 ----------------------------
4663 4138
4664 :Capability: KVM_CAP_S390_CMMA_MIGRATION 4139 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4665 :Architectures: s390 4140 :Architectures: s390
4666 :Type: vm ioctl 4141 :Type: vm ioctl
4667 :Parameters: struct kvm_s390_cmma_log (in) 4142 :Parameters: struct kvm_s390_cmma_log (in)
4668 :Returns: 0 on success, a negative value on e 4143 :Returns: 0 on success, a negative value on error
4669 4144
4670 This ioctl is used to set the values of the C 4145 This ioctl is used to set the values of the CMMA bits on the s390
4671 architecture. It is meant to be used during l 4146 architecture. It is meant to be used during live migration to restore
4672 the CMMA values, but there are no restriction 4147 the CMMA values, but there are no restrictions on its use.
4673 The ioctl takes parameters via the kvm_s390_c 4148 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4674 Each CMMA value takes up one byte. 4149 Each CMMA value takes up one byte.
4675 4150
4676 :: 4151 ::
4677 4152
4678 struct kvm_s390_cmma_log { 4153 struct kvm_s390_cmma_log {
4679 __u64 start_gfn; 4154 __u64 start_gfn;
4680 __u32 count; 4155 __u32 count;
4681 __u32 flags; 4156 __u32 flags;
4682 union { 4157 union {
4683 __u64 remaining; 4158 __u64 remaining;
4684 __u64 mask; 4159 __u64 mask;
4685 }; 4160 };
4686 __u64 values; 4161 __u64 values;
4687 }; 4162 };
4688 4163
4689 start_gfn indicates the starting guest frame 4164 start_gfn indicates the starting guest frame number,
4690 4165
4691 count indicates how many values are to be con 4166 count indicates how many values are to be considered in the buffer,
4692 4167
4693 flags is not used and must be 0. 4168 flags is not used and must be 0.
4694 4169
4695 mask indicates which PGSTE bits are to be con 4170 mask indicates which PGSTE bits are to be considered.
4696 4171
4697 remaining is not used. 4172 remaining is not used.
4698 4173
4699 values points to the buffer in userspace wher 4174 values points to the buffer in userspace where to store the values.
4700 4175
4701 This ioctl can fail with -ENOMEM if not enoug 4176 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4702 complete the task, with -ENXIO if CMMA is not 4177 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4703 the count field is too large (e.g. more than 4178 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4704 if the flags field was not 0, with -EFAULT if 4179 if the flags field was not 0, with -EFAULT if the userspace address is
4705 invalid, if invalid pages are written to (e.g 4180 invalid, if invalid pages are written to (e.g. after the end of memory)
4706 or if no page table is present for the addres 4181 or if no page table is present for the addresses (e.g. when using
4707 hugepages). 4182 hugepages).
4708 4183
4709 4.109 KVM_PPC_GET_CPU_CHAR 4184 4.109 KVM_PPC_GET_CPU_CHAR
4710 -------------------------- 4185 --------------------------
4711 4186
4712 :Capability: KVM_CAP_PPC_GET_CPU_CHAR 4187 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4713 :Architectures: powerpc 4188 :Architectures: powerpc
4714 :Type: vm ioctl 4189 :Type: vm ioctl
4715 :Parameters: struct kvm_ppc_cpu_char (out) 4190 :Parameters: struct kvm_ppc_cpu_char (out)
4716 :Returns: 0 on successful completion, 4191 :Returns: 0 on successful completion,
4717 -EFAULT if struct kvm_ppc_cpu_char c 4192 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4718 4193
4719 This ioctl gives userspace information about 4194 This ioctl gives userspace information about certain characteristics
4720 of the CPU relating to speculative execution 4195 of the CPU relating to speculative execution of instructions and
4721 possible information leakage resulting from s 4196 possible information leakage resulting from speculative execution (see
4722 CVE-2017-5715, CVE-2017-5753 and CVE-2017-575 4197 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4723 returned in struct kvm_ppc_cpu_char, which lo 4198 returned in struct kvm_ppc_cpu_char, which looks like this::
4724 4199
4725 struct kvm_ppc_cpu_char { 4200 struct kvm_ppc_cpu_char {
4726 __u64 character; /* ch 4201 __u64 character; /* characteristics of the CPU */
4727 __u64 behaviour; /* re 4202 __u64 behaviour; /* recommended software behaviour */
4728 __u64 character_mask; /* va 4203 __u64 character_mask; /* valid bits in character */
4729 __u64 behaviour_mask; /* va 4204 __u64 behaviour_mask; /* valid bits in behaviour */
4730 }; 4205 };
4731 4206
4732 For extensibility, the character_mask and beh 4207 For extensibility, the character_mask and behaviour_mask fields
4733 indicate which bits of character and behaviou 4208 indicate which bits of character and behaviour have been filled in by
4734 the kernel. If the set of defined bits is ex 4209 the kernel. If the set of defined bits is extended in future then
4735 userspace will be able to tell whether it is 4210 userspace will be able to tell whether it is running on a kernel that
4736 knows about the new bits. 4211 knows about the new bits.
4737 4212
4738 The character field describes attributes of t 4213 The character field describes attributes of the CPU which can help
4739 with preventing inadvertent information discl 4214 with preventing inadvertent information disclosure - specifically,
4740 whether there is an instruction to flash-inva 4215 whether there is an instruction to flash-invalidate the L1 data cache
4741 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether 4216 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4742 to a mode where entries can only be used by t 4217 to a mode where entries can only be used by the thread that created
4743 them, whether the bcctr[l] instruction preven 4218 them, whether the bcctr[l] instruction prevents speculation, and
4744 whether a speculation barrier instruction (or 4219 whether a speculation barrier instruction (ori 31,31,0) is provided.
4745 4220
4746 The behaviour field describes actions that so 4221 The behaviour field describes actions that software should take to
4747 prevent inadvertent information disclosure, a 4222 prevent inadvertent information disclosure, and thus describes which
4748 vulnerabilities the hardware is subject to; s 4223 vulnerabilities the hardware is subject to; specifically whether the
4749 L1 data cache should be flushed when returnin 4224 L1 data cache should be flushed when returning to user mode from the
4750 kernel, and whether a speculation barrier sho 4225 kernel, and whether a speculation barrier should be placed between an
4751 array bounds check and the array access. 4226 array bounds check and the array access.
4752 4227
4753 These fields use the same bit definitions as 4228 These fields use the same bit definitions as the new
4754 H_GET_CPU_CHARACTERISTICS hypercall. 4229 H_GET_CPU_CHARACTERISTICS hypercall.
4755 4230
4756 4.110 KVM_MEMORY_ENCRYPT_OP 4231 4.110 KVM_MEMORY_ENCRYPT_OP
4757 --------------------------- 4232 ---------------------------
4758 4233
4759 :Capability: basic 4234 :Capability: basic
4760 :Architectures: x86 4235 :Architectures: x86
4761 :Type: vm 4236 :Type: vm
4762 :Parameters: an opaque platform specific stru 4237 :Parameters: an opaque platform specific structure (in/out)
4763 :Returns: 0 on success; -1 on error 4238 :Returns: 0 on success; -1 on error
4764 4239
4765 If the platform supports creating encrypted V 4240 If the platform supports creating encrypted VMs then this ioctl can be used
4766 for issuing platform-specific memory encrypti 4241 for issuing platform-specific memory encryption commands to manage those
4767 encrypted VMs. 4242 encrypted VMs.
4768 4243
4769 Currently, this ioctl is used for issuing Sec 4244 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4770 (SEV) commands on AMD Processors. The SEV com 4245 (SEV) commands on AMD Processors. The SEV commands are defined in
4771 Documentation/virt/kvm/x86/amd-memory-encrypt !! 4246 Documentation/virt/kvm/amd-memory-encryption.rst.
4772 4247
4773 4.111 KVM_MEMORY_ENCRYPT_REG_REGION 4248 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4774 ----------------------------------- 4249 -----------------------------------
4775 4250
4776 :Capability: basic 4251 :Capability: basic
4777 :Architectures: x86 4252 :Architectures: x86
4778 :Type: system 4253 :Type: system
4779 :Parameters: struct kvm_enc_region (in) 4254 :Parameters: struct kvm_enc_region (in)
4780 :Returns: 0 on success; -1 on error 4255 :Returns: 0 on success; -1 on error
4781 4256
4782 This ioctl can be used to register a guest me 4257 This ioctl can be used to register a guest memory region which may
4783 contain encrypted data (e.g. guest RAM, SMRAM 4258 contain encrypted data (e.g. guest RAM, SMRAM etc).
4784 4259
4785 It is used in the SEV-enabled guest. When enc 4260 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4786 memory region may contain encrypted data. The 4261 memory region may contain encrypted data. The SEV memory encryption
4787 engine uses a tweak such that two identical p 4262 engine uses a tweak such that two identical plaintext pages, each at
4788 different locations will have differing ciphe 4263 different locations will have differing ciphertexts. So swapping or
4789 moving ciphertext of those pages will not res 4264 moving ciphertext of those pages will not result in plaintext being
4790 swapped. So relocating (or migrating) physica 4265 swapped. So relocating (or migrating) physical backing pages for the SEV
4791 guest will require some additional steps. 4266 guest will require some additional steps.
4792 4267
4793 Note: The current SEV key management spec doe 4268 Note: The current SEV key management spec does not provide commands to
4794 swap or migrate (move) ciphertext pages. Henc 4269 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4795 memory region registered with the ioctl. 4270 memory region registered with the ioctl.
4796 4271
4797 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION 4272 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4798 ------------------------------------- 4273 -------------------------------------
4799 4274
4800 :Capability: basic 4275 :Capability: basic
4801 :Architectures: x86 4276 :Architectures: x86
4802 :Type: system 4277 :Type: system
4803 :Parameters: struct kvm_enc_region (in) 4278 :Parameters: struct kvm_enc_region (in)
4804 :Returns: 0 on success; -1 on error 4279 :Returns: 0 on success; -1 on error
4805 4280
4806 This ioctl can be used to unregister the gues 4281 This ioctl can be used to unregister the guest memory region registered
4807 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl abov 4282 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4808 4283
4809 4.113 KVM_HYPERV_EVENTFD 4284 4.113 KVM_HYPERV_EVENTFD
4810 ------------------------ 4285 ------------------------
4811 4286
4812 :Capability: KVM_CAP_HYPERV_EVENTFD 4287 :Capability: KVM_CAP_HYPERV_EVENTFD
4813 :Architectures: x86 4288 :Architectures: x86
4814 :Type: vm ioctl 4289 :Type: vm ioctl
4815 :Parameters: struct kvm_hyperv_eventfd (in) 4290 :Parameters: struct kvm_hyperv_eventfd (in)
4816 4291
4817 This ioctl (un)registers an eventfd to receiv 4292 This ioctl (un)registers an eventfd to receive notifications from the guest on
4818 the specified Hyper-V connection id through t 4293 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4819 causing a user exit. SIGNAL_EVENT hypercall 4294 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4820 (bits 24-31) still triggers a KVM_EXIT_HYPERV 4295 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4821 4296
4822 :: 4297 ::
4823 4298
4824 struct kvm_hyperv_eventfd { 4299 struct kvm_hyperv_eventfd {
4825 __u32 conn_id; 4300 __u32 conn_id;
4826 __s32 fd; 4301 __s32 fd;
4827 __u32 flags; 4302 __u32 flags;
4828 __u32 padding[3]; 4303 __u32 padding[3];
4829 }; 4304 };
4830 4305
4831 The conn_id field should fit within 24 bits:: 4306 The conn_id field should fit within 24 bits::
4832 4307
4833 #define KVM_HYPERV_CONN_ID_MASK 4308 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4834 4309
4835 The acceptable values for the flags field are 4310 The acceptable values for the flags field are::
4836 4311
4837 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 4312 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4838 4313
4839 :Returns: 0 on success, 4314 :Returns: 0 on success,
4840 -EINVAL if conn_id or flags is outs 4315 -EINVAL if conn_id or flags is outside the allowed range,
4841 -ENOENT on deassign if the conn_id 4316 -ENOENT on deassign if the conn_id isn't registered,
4842 -EEXIST on assign if the conn_id is 4317 -EEXIST on assign if the conn_id is already registered
4843 4318
4844 4.114 KVM_GET_NESTED_STATE 4319 4.114 KVM_GET_NESTED_STATE
4845 -------------------------- 4320 --------------------------
4846 4321
4847 :Capability: KVM_CAP_NESTED_STATE 4322 :Capability: KVM_CAP_NESTED_STATE
4848 :Architectures: x86 4323 :Architectures: x86
4849 :Type: vcpu ioctl 4324 :Type: vcpu ioctl
4850 :Parameters: struct kvm_nested_state (in/out) 4325 :Parameters: struct kvm_nested_state (in/out)
4851 :Returns: 0 on success, -1 on error 4326 :Returns: 0 on success, -1 on error
4852 4327
4853 Errors: 4328 Errors:
4854 4329
4855 ===== ================================ 4330 ===== =============================================================
4856 E2BIG the total state size exceeds the 4331 E2BIG the total state size exceeds the value of 'size' specified by
4857 the user; the size required will 4332 the user; the size required will be written into size.
4858 ===== ================================ 4333 ===== =============================================================
4859 4334
4860 :: 4335 ::
4861 4336
4862 struct kvm_nested_state { 4337 struct kvm_nested_state {
4863 __u16 flags; 4338 __u16 flags;
4864 __u16 format; 4339 __u16 format;
4865 __u32 size; 4340 __u32 size;
4866 4341
4867 union { 4342 union {
4868 struct kvm_vmx_nested_state_h 4343 struct kvm_vmx_nested_state_hdr vmx;
4869 struct kvm_svm_nested_state_h 4344 struct kvm_svm_nested_state_hdr svm;
4870 4345
4871 /* Pad the header to 128 byte 4346 /* Pad the header to 128 bytes. */
4872 __u8 pad[120]; 4347 __u8 pad[120];
4873 } hdr; 4348 } hdr;
4874 4349
4875 union { 4350 union {
4876 struct kvm_vmx_nested_state_d 4351 struct kvm_vmx_nested_state_data vmx[0];
4877 struct kvm_svm_nested_state_d 4352 struct kvm_svm_nested_state_data svm[0];
4878 } data; 4353 } data;
4879 }; 4354 };
4880 4355
4881 #define KVM_STATE_NESTED_GUEST_MODE 4356 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4882 #define KVM_STATE_NESTED_RUN_PENDING 4357 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4883 #define KVM_STATE_NESTED_EVMCS 4358 #define KVM_STATE_NESTED_EVMCS 0x00000004
4884 4359
4885 #define KVM_STATE_NESTED_FORMAT_VMX 4360 #define KVM_STATE_NESTED_FORMAT_VMX 0
4886 #define KVM_STATE_NESTED_FORMAT_SVM 4361 #define KVM_STATE_NESTED_FORMAT_SVM 1
4887 4362
4888 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 4363 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4889 4364
4890 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 4365 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4891 #define KVM_STATE_NESTED_VMX_SMM_VMXON 4366 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4892 4367
4893 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEAD 4368 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4894 4369
4895 struct kvm_vmx_nested_state_hdr { 4370 struct kvm_vmx_nested_state_hdr {
4896 __u64 vmxon_pa; 4371 __u64 vmxon_pa;
4897 __u64 vmcs12_pa; 4372 __u64 vmcs12_pa;
4898 4373
4899 struct { 4374 struct {
4900 __u16 flags; 4375 __u16 flags;
4901 } smm; 4376 } smm;
4902 4377
4903 __u32 flags; 4378 __u32 flags;
4904 __u64 preemption_timer_deadline; 4379 __u64 preemption_timer_deadline;
4905 }; 4380 };
4906 4381
4907 struct kvm_vmx_nested_state_data { 4382 struct kvm_vmx_nested_state_data {
4908 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS 4383 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4909 __u8 shadow_vmcs12[KVM_STATE_NESTED_V 4384 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4910 }; 4385 };
4911 4386
4912 This ioctl copies the vcpu's nested virtualiz 4387 This ioctl copies the vcpu's nested virtualization state from the kernel to
4913 userspace. 4388 userspace.
4914 4389
4915 The maximum size of the state can be retrieve 4390 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4916 to the KVM_CHECK_EXTENSION ioctl(). 4391 to the KVM_CHECK_EXTENSION ioctl().
4917 4392
4918 4.115 KVM_SET_NESTED_STATE 4393 4.115 KVM_SET_NESTED_STATE
4919 -------------------------- 4394 --------------------------
4920 4395
4921 :Capability: KVM_CAP_NESTED_STATE 4396 :Capability: KVM_CAP_NESTED_STATE
4922 :Architectures: x86 4397 :Architectures: x86
4923 :Type: vcpu ioctl 4398 :Type: vcpu ioctl
4924 :Parameters: struct kvm_nested_state (in) 4399 :Parameters: struct kvm_nested_state (in)
4925 :Returns: 0 on success, -1 on error 4400 :Returns: 0 on success, -1 on error
4926 4401
4927 This copies the vcpu's kvm_nested_state struc 4402 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4928 For the definition of struct kvm_nested_state 4403 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4929 4404
4930 4.116 KVM_(UN)REGISTER_COALESCED_MMIO 4405 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4931 ------------------------------------- 4406 -------------------------------------
4932 4407
4933 :Capability: KVM_CAP_COALESCED_MMIO (for coal 4408 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4934 KVM_CAP_COALESCED_PIO (for coale 4409 KVM_CAP_COALESCED_PIO (for coalesced pio)
4935 :Architectures: all 4410 :Architectures: all
4936 :Type: vm ioctl 4411 :Type: vm ioctl
4937 :Parameters: struct kvm_coalesced_mmio_zone 4412 :Parameters: struct kvm_coalesced_mmio_zone
4938 :Returns: 0 on success, < 0 on error 4413 :Returns: 0 on success, < 0 on error
4939 4414
4940 Coalesced I/O is a performance optimization t 4415 Coalesced I/O is a performance optimization that defers hardware
4941 register write emulation so that userspace ex 4416 register write emulation so that userspace exits are avoided. It is
4942 typically used to reduce the overhead of emul 4417 typically used to reduce the overhead of emulating frequently accessed
4943 hardware registers. 4418 hardware registers.
4944 4419
4945 When a hardware register is configured for co 4420 When a hardware register is configured for coalesced I/O, write accesses
4946 do not exit to userspace and their value is r 4421 do not exit to userspace and their value is recorded in a ring buffer
4947 that is shared between kernel and userspace. 4422 that is shared between kernel and userspace.
4948 4423
4949 Coalesced I/O is used if one or more write ac 4424 Coalesced I/O is used if one or more write accesses to a hardware
4950 register can be deferred until a read or a wr 4425 register can be deferred until a read or a write to another hardware
4951 register on the same device. This last acces 4426 register on the same device. This last access will cause a vmexit and
4952 userspace will process accesses from the ring 4427 userspace will process accesses from the ring buffer before emulating
4953 it. That will avoid exiting to userspace on r 4428 it. That will avoid exiting to userspace on repeated writes.
4954 4429
4955 Coalesced pio is based on coalesced mmio. The 4430 Coalesced pio is based on coalesced mmio. There is little difference
4956 between coalesced mmio and pio except that co 4431 between coalesced mmio and pio except that coalesced pio records accesses
4957 to I/O ports. 4432 to I/O ports.
4958 4433
4959 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl) 4434 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4960 ------------------------------------ 4435 ------------------------------------
4961 4436
4962 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT 4437 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4963 :Architectures: x86, arm64, mips !! 4438 :Architectures: x86, arm, arm64, mips
4964 :Type: vm ioctl 4439 :Type: vm ioctl
4965 :Parameters: struct kvm_clear_dirty_log (in) 4440 :Parameters: struct kvm_clear_dirty_log (in)
4966 :Returns: 0 on success, -1 on error 4441 :Returns: 0 on success, -1 on error
4967 4442
4968 :: 4443 ::
4969 4444
4970 /* for KVM_CLEAR_DIRTY_LOG */ 4445 /* for KVM_CLEAR_DIRTY_LOG */
4971 struct kvm_clear_dirty_log { 4446 struct kvm_clear_dirty_log {
4972 __u32 slot; 4447 __u32 slot;
4973 __u32 num_pages; 4448 __u32 num_pages;
4974 __u64 first_page; 4449 __u64 first_page;
4975 union { 4450 union {
4976 void __user *dirty_bitmap; /* 4451 void __user *dirty_bitmap; /* one bit per page */
4977 __u64 padding; 4452 __u64 padding;
4978 }; 4453 };
4979 }; 4454 };
4980 4455
4981 The ioctl clears the dirty status of pages in 4456 The ioctl clears the dirty status of pages in a memory slot, according to
4982 the bitmap that is passed in struct kvm_clear 4457 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4983 field. Bit 0 of the bitmap corresponds to pa 4458 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4984 memory slot, and num_pages is the size in bit 4459 memory slot, and num_pages is the size in bits of the input bitmap.
4985 first_page must be a multiple of 64; num_page 4460 first_page must be a multiple of 64; num_pages must also be a multiple of
4986 64 unless first_page + num_pages is the size 4461 64 unless first_page + num_pages is the size of the memory slot. For each
4987 bit that is set in the input bitmap, the corr 4462 bit that is set in the input bitmap, the corresponding page is marked "clean"
4988 in KVM's dirty bitmap, and dirty tracking is 4463 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4989 (for example via write-protection, or by clea 4464 (for example via write-protection, or by clearing the dirty bit in
4990 a page table entry). 4465 a page table entry).
4991 4466
4992 If KVM_CAP_MULTI_ADDRESS_SPACE is available, 4467 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4993 the address space for which you want to clear 4468 the address space for which you want to clear the dirty status. See
4994 KVM_SET_USER_MEMORY_REGION for details on the 4469 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4995 4470
4996 This ioctl is mostly useful when KVM_CAP_MANU 4471 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4997 is enabled; for more information, see the des 4472 is enabled; for more information, see the description of the capability.
4998 However, it can always be used as long as KVM 4473 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4999 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is pre 4474 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
5000 4475
5001 4.118 KVM_GET_SUPPORTED_HV_CPUID 4476 4.118 KVM_GET_SUPPORTED_HV_CPUID
5002 -------------------------------- 4477 --------------------------------
5003 4478
5004 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM 4479 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
5005 :Architectures: x86 4480 :Architectures: x86
5006 :Type: system ioctl, vcpu ioctl 4481 :Type: system ioctl, vcpu ioctl
5007 :Parameters: struct kvm_cpuid2 (in/out) 4482 :Parameters: struct kvm_cpuid2 (in/out)
5008 :Returns: 0 on success, -1 on error 4483 :Returns: 0 on success, -1 on error
5009 4484
5010 :: 4485 ::
5011 4486
5012 struct kvm_cpuid2 { 4487 struct kvm_cpuid2 {
5013 __u32 nent; 4488 __u32 nent;
5014 __u32 padding; 4489 __u32 padding;
5015 struct kvm_cpuid_entry2 entries[0]; 4490 struct kvm_cpuid_entry2 entries[0];
5016 }; 4491 };
5017 4492
5018 struct kvm_cpuid_entry2 { 4493 struct kvm_cpuid_entry2 {
5019 __u32 function; 4494 __u32 function;
5020 __u32 index; 4495 __u32 index;
5021 __u32 flags; 4496 __u32 flags;
5022 __u32 eax; 4497 __u32 eax;
5023 __u32 ebx; 4498 __u32 ebx;
5024 __u32 ecx; 4499 __u32 ecx;
5025 __u32 edx; 4500 __u32 edx;
5026 __u32 padding[3]; 4501 __u32 padding[3];
5027 }; 4502 };
5028 4503
5029 This ioctl returns x86 cpuid features leaves 4504 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
5030 KVM. Userspace can use the information retur 4505 KVM. Userspace can use the information returned by this ioctl to construct
5031 cpuid information presented to guests consumi 4506 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
5032 Windows or Hyper-V guests). 4507 Windows or Hyper-V guests).
5033 4508
5034 CPUID feature leaves returned by this ioctl a 4509 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
5035 Functional Specification (TLFS). These leaves 4510 Functional Specification (TLFS). These leaves can't be obtained with
5036 KVM_GET_SUPPORTED_CPUID ioctl because some of 4511 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
5037 leaves (0x40000000, 0x40000001). 4512 leaves (0x40000000, 0x40000001).
5038 4513
5039 Currently, the following list of CPUID leaves 4514 Currently, the following list of CPUID leaves are returned:
5040 <<
5041 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS 4515 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
5042 - HYPERV_CPUID_INTERFACE 4516 - HYPERV_CPUID_INTERFACE
5043 - HYPERV_CPUID_VERSION 4517 - HYPERV_CPUID_VERSION
5044 - HYPERV_CPUID_FEATURES 4518 - HYPERV_CPUID_FEATURES
5045 - HYPERV_CPUID_ENLIGHTMENT_INFO 4519 - HYPERV_CPUID_ENLIGHTMENT_INFO
5046 - HYPERV_CPUID_IMPLEMENT_LIMITS 4520 - HYPERV_CPUID_IMPLEMENT_LIMITS
5047 - HYPERV_CPUID_NESTED_FEATURES 4521 - HYPERV_CPUID_NESTED_FEATURES
5048 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIO 4522 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
5049 - HYPERV_CPUID_SYNDBG_INTERFACE 4523 - HYPERV_CPUID_SYNDBG_INTERFACE
5050 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES 4524 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
5051 4525
5052 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID 4526 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
5053 with the 'nent' field indicating the number o 4527 with the 'nent' field indicating the number of entries in the variable-size
5054 array 'entries'. If the number of entries is 4528 array 'entries'. If the number of entries is too low to describe all Hyper-V
5055 feature leaves, an error (E2BIG) is returned. 4529 feature leaves, an error (E2BIG) is returned. If the number is more or equal
5056 to the number of Hyper-V feature leaves, the 4530 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
5057 number of valid entries in the 'entries' arra 4531 number of valid entries in the 'entries' array, which is then filled.
5058 4532
5059 'index' and 'flags' fields in 'struct kvm_cpu 4533 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
5060 userspace should not expect to get any partic 4534 userspace should not expect to get any particular value there.
5061 4535
5062 Note, vcpu version of KVM_GET_SUPPORTED_HV_CP 4536 Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
5063 system ioctl which exposes all supported feat 4537 system ioctl which exposes all supported feature bits unconditionally, vcpu
5064 version has the following quirks: 4538 version has the following quirks:
5065 <<
5066 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X6 4539 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
5067 feature bit are only exposed when Enlighten 4540 feature bit are only exposed when Enlightened VMCS was previously enabled
5068 on the corresponding vCPU (KVM_CAP_HYPERV_E 4541 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
5069 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only 4542 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
5070 (presumes KVM_CREATE_IRQCHIP has already be 4543 (presumes KVM_CREATE_IRQCHIP has already been called).
5071 4544
5072 4.119 KVM_ARM_VCPU_FINALIZE 4545 4.119 KVM_ARM_VCPU_FINALIZE
5073 --------------------------- 4546 ---------------------------
5074 4547
5075 :Architectures: arm64 !! 4548 :Architectures: arm, arm64
5076 :Type: vcpu ioctl 4549 :Type: vcpu ioctl
5077 :Parameters: int feature (in) 4550 :Parameters: int feature (in)
5078 :Returns: 0 on success, -1 on error 4551 :Returns: 0 on success, -1 on error
5079 4552
5080 Errors: 4553 Errors:
5081 4554
5082 ====== ================================ 4555 ====== ==============================================================
5083 EPERM feature not enabled, needs confi 4556 EPERM feature not enabled, needs configuration, or already finalized
5084 EINVAL feature unknown or not present 4557 EINVAL feature unknown or not present
5085 ====== ================================ 4558 ====== ==============================================================
5086 4559
5087 Recognised values for feature: 4560 Recognised values for feature:
5088 4561
5089 ===== ================================ 4562 ===== ===========================================
5090 arm64 KVM_ARM_VCPU_SVE (requires KVM_C 4563 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
5091 ===== ================================ 4564 ===== ===========================================
5092 4565
5093 Finalizes the configuration of the specified 4566 Finalizes the configuration of the specified vcpu feature.
5094 4567
5095 The vcpu must already have been initialised, 4568 The vcpu must already have been initialised, enabling the affected feature, by
5096 means of a successful KVM_ARM_VCPU_INIT call 4569 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
5097 features[]. 4570 features[].
5098 4571
5099 For affected vcpu features, this is a mandato 4572 For affected vcpu features, this is a mandatory step that must be performed
5100 before the vcpu is fully usable. 4573 before the vcpu is fully usable.
5101 4574
5102 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FI 4575 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
5103 configured by use of ioctls such as KVM_SET_O 4576 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
5104 that should be performed and how to do it are !! 4577 that should be performaned and how to do it are feature-dependent.
5105 4578
5106 Other calls that depend on a particular featu 4579 Other calls that depend on a particular feature being finalized, such as
5107 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG an 4580 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
5108 -EPERM unless the feature has already been fi 4581 -EPERM unless the feature has already been finalized by means of a
5109 KVM_ARM_VCPU_FINALIZE call. 4582 KVM_ARM_VCPU_FINALIZE call.
5110 4583
5111 See KVM_ARM_VCPU_INIT for details of vcpu fea 4584 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
5112 using this ioctl. 4585 using this ioctl.
5113 4586
5114 4.120 KVM_SET_PMU_EVENT_FILTER 4587 4.120 KVM_SET_PMU_EVENT_FILTER
5115 ------------------------------ 4588 ------------------------------
5116 4589
5117 :Capability: KVM_CAP_PMU_EVENT_FILTER 4590 :Capability: KVM_CAP_PMU_EVENT_FILTER
5118 :Architectures: x86 4591 :Architectures: x86
5119 :Type: vm ioctl 4592 :Type: vm ioctl
5120 :Parameters: struct kvm_pmu_event_filter (in) 4593 :Parameters: struct kvm_pmu_event_filter (in)
5121 :Returns: 0 on success, -1 on error 4594 :Returns: 0 on success, -1 on error
5122 4595
5123 Errors: <<
5124 <<
5125 ====== ================================ <<
5126 EFAULT args[0] cannot be accessed <<
5127 EINVAL args[0] contains invalid data in <<
5128 E2BIG nevents is too large <<
5129 EBUSY not enough memory to allocate th <<
5130 ====== ================================ <<
5131 <<
5132 :: 4596 ::
5133 4597
5134 struct kvm_pmu_event_filter { 4598 struct kvm_pmu_event_filter {
5135 __u32 action; 4599 __u32 action;
5136 __u32 nevents; 4600 __u32 nevents;
5137 __u32 fixed_counter_bitmap; 4601 __u32 fixed_counter_bitmap;
5138 __u32 flags; 4602 __u32 flags;
5139 __u32 pad[4]; 4603 __u32 pad[4];
5140 __u64 events[0]; 4604 __u64 events[0];
5141 }; 4605 };
5142 4606
5143 This ioctl restricts the set of PMU events th !! 4607 This ioctl restricts the set of PMU events that the guest can program.
5144 which event select and unit mask combinations !! 4608 The argument holds a list of events which will be allowed or denied.
>> 4609 The eventsel+umask of each event the guest attempts to program is compared
>> 4610 against the events field to determine whether the guest should have access.
>> 4611 The events field only controls general purpose counters; fixed purpose
>> 4612 counters are controlled by the fixed_counter_bitmap.
5145 4613
5146 The argument holds a list of filter events wh !! 4614 No flags are defined yet, the field must be zero.
5147 <<
5148 Filter events only control general purpose co <<
5149 are controlled by the fixed_counter_bitmap. <<
5150 <<
5151 Valid values for 'flags':: <<
5152 <<
5153 ``0`` <<
5154 <<
5155 To use this mode, clear the 'flags' field. <<
5156 <<
5157 In this mode each event will contain an event <<
5158 <<
5159 When the guest attempts to program the PMU th <<
5160 unit mask is compared against the filter even <<
5161 guest should have access. <<
5162 <<
5163 ``KVM_PMU_EVENT_FLAG_MASKED_EVENTS`` <<
5164 :Capability: KVM_CAP_PMU_EVENT_MASKED_EVENTS <<
5165 <<
5166 In this mode each filter event will contain a <<
5167 exclude value. To encode a masked event use: <<
5168 <<
5169 KVM_PMU_ENCODE_MASKED_ENTRY() <<
5170 <<
5171 An encoded event will follow this layout:: <<
5172 <<
5173 Bits Description <<
5174 ---- ----------- <<
5175 7:0 event select (low bits) <<
5176 15:8 umask match <<
5177 31:16 unused <<
5178 35:32 event select (high bits) <<
5179 36:54 unused <<
5180 55 exclude bit <<
5181 63:56 umask mask <<
5182 <<
5183 When the guest attempts to program the PMU, t <<
5184 determining if the guest should have access: <<
5185 <<
5186 1. Match the event select from the guest aga <<
5187 2. If a match is found, match the guest's un <<
5188 values of the included filter events. <<
5189 I.e. (unit mask & mask) == match && !excl <<
5190 3. If a match is found, match the guest's un <<
5191 values of the excluded filter events. <<
5192 I.e. (unit mask & mask) == match && exclu <<
5193 4. <<
5194 a. If an included match is found and an ex <<
5195 the event. <<
5196 b. For everything else, do not filter the <<
5197 5. <<
5198 a. If the event is filtered and it's an al <<
5199 program the event. <<
5200 b. If the event is filtered and it's a den <<
5201 program the event. <<
5202 <<
5203 When setting a new pmu event filter, -EINVAL <<
5204 unused fields are set or if any of the high b <<
5205 select are set when called on Intel. <<
5206 4615
5207 Valid values for 'action':: 4616 Valid values for 'action'::
5208 4617
5209 #define KVM_PMU_EVENT_ALLOW 0 4618 #define KVM_PMU_EVENT_ALLOW 0
5210 #define KVM_PMU_EVENT_DENY 1 4619 #define KVM_PMU_EVENT_DENY 1
5211 4620
5212 Via this API, KVM userspace can also control <<
5213 counters (if any) by configuring the "action" <<
5214 <<
5215 Specifically, KVM follows the following pseud <<
5216 allow the guest FixCtr[i] to count its pre-de <<
5217 <<
5218 FixCtr[i]_is_allowed = (action == ALLOW) && <<
5219 (action == DENY) && !(bitmap & BIT(i)); <<
5220 FixCtr[i]_is_denied = !FixCtr[i]_is_allowed <<
5221 <<
5222 KVM always consumes fixed_counter_bitmap, it' <<
5223 ensure fixed_counter_bitmap is set correctly, <<
5224 a filter that only affects general purpose co <<
5225 <<
5226 Note, the "events" field also applies to fixe <<
5227 and unit_mask values. "fixed_counter_bitmap" <<
5228 if there is a contradiction between the two. <<
5229 <<
5230 4.121 KVM_PPC_SVM_OFF 4621 4.121 KVM_PPC_SVM_OFF
5231 --------------------- 4622 ---------------------
5232 4623
5233 :Capability: basic 4624 :Capability: basic
5234 :Architectures: powerpc 4625 :Architectures: powerpc
5235 :Type: vm ioctl 4626 :Type: vm ioctl
5236 :Parameters: none 4627 :Parameters: none
5237 :Returns: 0 on successful completion, 4628 :Returns: 0 on successful completion,
5238 4629
5239 Errors: 4630 Errors:
5240 4631
5241 ====== ================================ 4632 ====== ================================================================
5242 EINVAL if ultravisor failed to terminat 4633 EINVAL if ultravisor failed to terminate the secure guest
5243 ENOMEM if hypervisor failed to allocate 4634 ENOMEM if hypervisor failed to allocate new radix page tables for guest
5244 ====== ================================ 4635 ====== ================================================================
5245 4636
5246 This ioctl is used to turn off the secure mod 4637 This ioctl is used to turn off the secure mode of the guest or transition
5247 the guest from secure mode to normal mode. Th 4638 the guest from secure mode to normal mode. This is invoked when the guest
5248 is reset. This has no effect if called for a 4639 is reset. This has no effect if called for a normal guest.
5249 4640
5250 This ioctl issues an ultravisor call to termi 4641 This ioctl issues an ultravisor call to terminate the secure guest,
5251 unpins the VPA pages and releases all the dev 4642 unpins the VPA pages and releases all the device pages that are used to
5252 track the secure pages by hypervisor. 4643 track the secure pages by hypervisor.
5253 4644
5254 4.122 KVM_S390_NORMAL_RESET 4645 4.122 KVM_S390_NORMAL_RESET
5255 --------------------------- 4646 ---------------------------
5256 4647
5257 :Capability: KVM_CAP_S390_VCPU_RESETS 4648 :Capability: KVM_CAP_S390_VCPU_RESETS
5258 :Architectures: s390 4649 :Architectures: s390
5259 :Type: vcpu ioctl 4650 :Type: vcpu ioctl
5260 :Parameters: none 4651 :Parameters: none
5261 :Returns: 0 4652 :Returns: 0
5262 4653
5263 This ioctl resets VCPU registers and control 4654 This ioctl resets VCPU registers and control structures according to
5264 the cpu reset definition in the POP (Principl 4655 the cpu reset definition in the POP (Principles Of Operation).
5265 4656
5266 4.123 KVM_S390_INITIAL_RESET 4657 4.123 KVM_S390_INITIAL_RESET
5267 ---------------------------- 4658 ----------------------------
5268 4659
5269 :Capability: none 4660 :Capability: none
5270 :Architectures: s390 4661 :Architectures: s390
5271 :Type: vcpu ioctl 4662 :Type: vcpu ioctl
5272 :Parameters: none 4663 :Parameters: none
5273 :Returns: 0 4664 :Returns: 0
5274 4665
5275 This ioctl resets VCPU registers and control 4666 This ioctl resets VCPU registers and control structures according to
5276 the initial cpu reset definition in the POP. 4667 the initial cpu reset definition in the POP. However, the cpu is not
5277 put into ESA mode. This reset is a superset o 4668 put into ESA mode. This reset is a superset of the normal reset.
5278 4669
5279 4.124 KVM_S390_CLEAR_RESET 4670 4.124 KVM_S390_CLEAR_RESET
5280 -------------------------- 4671 --------------------------
5281 4672
5282 :Capability: KVM_CAP_S390_VCPU_RESETS 4673 :Capability: KVM_CAP_S390_VCPU_RESETS
5283 :Architectures: s390 4674 :Architectures: s390
5284 :Type: vcpu ioctl 4675 :Type: vcpu ioctl
5285 :Parameters: none 4676 :Parameters: none
5286 :Returns: 0 4677 :Returns: 0
5287 4678
5288 This ioctl resets VCPU registers and control 4679 This ioctl resets VCPU registers and control structures according to
5289 the clear cpu reset definition in the POP. Ho 4680 the clear cpu reset definition in the POP. However, the cpu is not put
5290 into ESA mode. This reset is a superset of th 4681 into ESA mode. This reset is a superset of the initial reset.
5291 4682
5292 4683
5293 4.125 KVM_S390_PV_COMMAND 4684 4.125 KVM_S390_PV_COMMAND
5294 ------------------------- 4685 -------------------------
5295 4686
5296 :Capability: KVM_CAP_S390_PROTECTED 4687 :Capability: KVM_CAP_S390_PROTECTED
5297 :Architectures: s390 4688 :Architectures: s390
5298 :Type: vm ioctl 4689 :Type: vm ioctl
5299 :Parameters: struct kvm_pv_cmd 4690 :Parameters: struct kvm_pv_cmd
5300 :Returns: 0 on success, < 0 on error 4691 :Returns: 0 on success, < 0 on error
5301 4692
5302 :: 4693 ::
5303 4694
5304 struct kvm_pv_cmd { 4695 struct kvm_pv_cmd {
5305 __u32 cmd; /* Command to be exec 4696 __u32 cmd; /* Command to be executed */
5306 __u16 rc; /* Ultravisor return 4697 __u16 rc; /* Ultravisor return code */
5307 __u16 rrc; /* Ultravisor return 4698 __u16 rrc; /* Ultravisor return reason code */
5308 __u64 data; /* Data or address */ 4699 __u64 data; /* Data or address */
5309 __u32 flags; /* flags for future e 4700 __u32 flags; /* flags for future extensions. Must be 0 for now */
5310 __u32 reserved[3]; 4701 __u32 reserved[3];
5311 }; 4702 };
5312 4703
5313 **Ultravisor return codes** !! 4704 cmd values:
5314 The Ultravisor return (reason) codes are prov <<
5315 Ultravisor call has been executed to achieve <<
5316 the command. Therefore they are independent o <<
5317 code. If KVM changes `rc`, its value will alw <<
5318 hence setting it to 0 before issuing a PV com <<
5319 able to detect a change of `rc`. <<
5320 <<
5321 **cmd values:** <<
5322 4705
5323 KVM_PV_ENABLE 4706 KVM_PV_ENABLE
5324 Allocate memory and register the VM with th 4707 Allocate memory and register the VM with the Ultravisor, thereby
5325 donating memory to the Ultravisor that will 4708 donating memory to the Ultravisor that will become inaccessible to
5326 KVM. All existing CPUs are converted to pro 4709 KVM. All existing CPUs are converted to protected ones. After this
5327 command has succeeded, any CPU added via ho 4710 command has succeeded, any CPU added via hotplug will become
5328 protected during its creation as well. 4711 protected during its creation as well.
5329 4712
5330 Errors: 4713 Errors:
5331 4714
5332 ===== ============================= 4715 ===== =============================
5333 EINTR an unmasked signal is pending 4716 EINTR an unmasked signal is pending
5334 ===== ============================= 4717 ===== =============================
5335 4718
5336 KVM_PV_DISABLE 4719 KVM_PV_DISABLE
5337 Deregister the VM from the Ultravisor and r !! 4720
5338 been donated to the Ultravisor, making it u !! 4721 Deregister the VM from the Ultravisor and reclaim the memory that
5339 All registered VCPUs are converted back to !! 4722 had been donated to the Ultravisor, making it usable by the kernel
5340 previous protected VM had been prepared for !! 4723 again. All registered VCPUs are converted back to non-protected
5341 KVM_PV_ASYNC_CLEANUP_PREPARE and not subseq !! 4724 ones.
5342 KVM_PV_ASYNC_CLEANUP_PERFORM, it will be to <<
5343 together with the current protected VM. <<
5344 4725
5345 KVM_PV_VM_SET_SEC_PARMS 4726 KVM_PV_VM_SET_SEC_PARMS
5346 Pass the image header from VM memory to the 4727 Pass the image header from VM memory to the Ultravisor in
5347 preparation of image unpacking and verifica 4728 preparation of image unpacking and verification.
5348 4729
5349 KVM_PV_VM_UNPACK 4730 KVM_PV_VM_UNPACK
5350 Unpack (protect and decrypt) a page of the 4731 Unpack (protect and decrypt) a page of the encrypted boot image.
5351 4732
5352 KVM_PV_VM_VERIFY 4733 KVM_PV_VM_VERIFY
5353 Verify the integrity of the unpacked image. 4734 Verify the integrity of the unpacked image. Only if this succeeds,
5354 KVM is allowed to start protected VCPUs. 4735 KVM is allowed to start protected VCPUs.
5355 4736
5356 KVM_PV_INFO !! 4737 4.126 KVM_X86_SET_MSR_FILTER
5357 :Capability: KVM_CAP_S390_PROTECTED_DUMP !! 4738 ----------------------------
5358 <<
5359 Presents an API that provides Ultravisor re <<
5360 via subcommands. len_max is the size of the <<
5361 len_written is KVM's indication of how much <<
5362 were actually written to. len_written can b <<
5363 valid fields if more response fields are ad <<
5364 <<
5365 :: <<
5366 <<
5367 enum pv_cmd_info_id { <<
5368 KVM_PV_INFO_VM, <<
5369 KVM_PV_INFO_DUMP, <<
5370 }; <<
5371 <<
5372 struct kvm_s390_pv_info_header { <<
5373 __u32 id; <<
5374 __u32 len_max; <<
5375 __u32 len_written; <<
5376 __u32 reserved; <<
5377 }; <<
5378 <<
5379 struct kvm_s390_pv_info { <<
5380 struct kvm_s390_pv_info_header header <<
5381 struct kvm_s390_pv_info_dump dump; <<
5382 struct kvm_s390_pv_info_vm vm; <<
5383 }; <<
5384 <<
5385 **subcommands:** <<
5386 <<
5387 KVM_PV_INFO_VM <<
5388 This subcommand provides basic Ultravisor <<
5389 hosts. These values are likely also expor <<
5390 firmware UV query interface but they are <<
5391 programs in this API. <<
5392 <<
5393 The installed calls and feature_indicatio <<
5394 installed UV calls and the UV's other fea <<
5395 <<
5396 The max_* members provide information abo <<
5397 vcpus, PV guests and PV guest memory size <<
5398 <<
5399 :: <<
5400 <<
5401 struct kvm_s390_pv_info_vm { <<
5402 __u64 inst_calls_list[4]; <<
5403 __u64 max_cpus; <<
5404 __u64 max_guests; <<
5405 __u64 max_guest_addr; <<
5406 __u64 feature_indication; <<
5407 }; <<
5408 <<
5409 <<
5410 KVM_PV_INFO_DUMP <<
5411 This subcommand provides information rela <<
5412 <<
5413 :: <<
5414 <<
5415 struct kvm_s390_pv_info_dump { <<
5416 __u64 dump_cpu_buffer_len; <<
5417 __u64 dump_config_mem_buffer_per_1m; <<
5418 __u64 dump_config_finalize_len; <<
5419 }; <<
5420 <<
5421 KVM_PV_DUMP <<
5422 :Capability: KVM_CAP_S390_PROTECTED_DUMP <<
5423 <<
5424 Presents an API that provides calls which f <<
5425 protected VM. <<
5426 <<
5427 :: <<
5428 <<
5429 struct kvm_s390_pv_dmp { <<
5430 __u64 subcmd; <<
5431 __u64 buff_addr; <<
5432 __u64 buff_len; <<
5433 __u64 gaddr; /* For dump s <<
5434 }; <<
5435 <<
5436 **subcommands:** <<
5437 <<
5438 KVM_PV_DUMP_INIT <<
5439 Initializes the dump process of a protect <<
5440 not succeed all other subcommands will fa <<
5441 subcommand will return -EINVAL if a dump <<
5442 completed. <<
5443 <<
5444 Not all PV vms can be dumped, the owner n <<
5445 allowed` PCF bit 34 in the SE header to a <<
5446 <<
5447 KVM_PV_DUMP_CONFIG_STOR_STATE <<
5448 Stores `buff_len` bytes of tweak compone <<
5449 the 1MB block specified by the absolute <<
5450 (`gaddr`). `buff_len` needs to be `conf_ <<
5451 aligned and at least >= the `conf_dump_s <<
5452 provided by the dump uv_info data. buff_ <<
5453 even if an error rc is returned. For ins <<
5454 fault after writing the first page of da <<
5455 <<
5456 KVM_PV_DUMP_COMPLETE <<
5457 If the subcommand succeeds it completes t <<
5458 KVM_PV_DUMP_INIT be called again. <<
5459 <<
5460 On success `conf_dump_finalize_len` bytes <<
5461 stored to the `buff_addr`. The completion <<
5462 derivation seed, IV, tweak nonce and encr <<
5463 authentication tag all of which are neede <<
5464 later time. <<
5465 <<
5466 KVM_PV_ASYNC_CLEANUP_PREPARE <<
5467 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_D <<
5468 <<
5469 Prepare the current protected VM for asynch <<
5470 resources used by the current protected VM <<
5471 subsequent asynchronous teardown. The curre <<
5472 resume execution immediately as non-protect <<
5473 one protected VM prepared for asynchronous <<
5474 a protected VM had already been prepared fo <<
5475 subsequently calling KVM_PV_ASYNC_CLEANUP_P <<
5476 fail. In that case, the userspace process s <<
5477 KVM_PV_DISABLE. The resources set aside wit <<
5478 be cleaned up with a subsequent call to KVM <<
5479 or KVM_PV_DISABLE, otherwise they will be c <<
5480 terminates. KVM_PV_ASYNC_CLEANUP_PREPARE ca <<
5481 as cleanup starts, i.e. before KVM_PV_ASYNC <<
5482 <<
5483 KVM_PV_ASYNC_CLEANUP_PERFORM <<
5484 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_D <<
5485 <<
5486 Tear down the protected VM previously prepa <<
5487 KVM_PV_ASYNC_CLEANUP_PREPARE. The resources <<
5488 will be freed during the execution of this <<
5489 should ideally be issued by userspace from <<
5490 fatal signal is received (or the process te <<
5491 command will terminate immediately without <<
5492 KVM shutdown procedure will take care of cl <<
5493 protected VMs, including the ones whose tea <<
5494 process termination. <<
5495 <<
5496 4.126 KVM_XEN_HVM_SET_ATTR <<
5497 -------------------------- <<
5498 <<
5499 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CO <<
5500 :Architectures: x86 <<
5501 :Type: vm ioctl <<
5502 :Parameters: struct kvm_xen_hvm_attr <<
5503 :Returns: 0 on success, < 0 on error <<
5504 <<
5505 :: <<
5506 <<
5507 struct kvm_xen_hvm_attr { <<
5508 __u16 type; <<
5509 __u16 pad[3]; <<
5510 union { <<
5511 __u8 long_mode; <<
5512 __u8 vector; <<
5513 __u8 runstate_update_flag; <<
5514 union { <<
5515 __u64 gfn; <<
5516 __u64 hva; <<
5517 } shared_info; <<
5518 struct { <<
5519 __u32 send_port; <<
5520 __u32 type; /* EVTCHN <<
5521 __u32 flags; <<
5522 union { <<
5523 struct { <<
5524 __u32 <<
5525 __u32 <<
5526 __u32 <<
5527 } port; <<
5528 struct { <<
5529 __u32 <<
5530 __s32 <<
5531 } eventfd; <<
5532 __u32 padding <<
5533 } deliver; <<
5534 } evtchn; <<
5535 __u32 xen_version; <<
5536 __u64 pad[8]; <<
5537 } u; <<
5538 }; <<
5539 <<
5540 type values: <<
5541 <<
5542 KVM_XEN_ATTR_TYPE_LONG_MODE <<
5543 Sets the ABI mode of the VM to 32-bit or 64 <<
5544 determines the layout of the shared_info pa <<
5545 <<
5546 KVM_XEN_ATTR_TYPE_SHARED_INFO <<
5547 Sets the guest physical frame number at whi <<
5548 page resides. Note that although Xen places <<
5549 32 vCPUs in the shared_info page, KVM does <<
5550 and instead requires that KVM_XEN_VCPU_ATTR <<
5551 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA be use <<
5552 the vcpu_info for a given vCPU resides at t <<
5553 in the shared_info page. This is because KV <<
5554 the Xen CPU id which is used as the index i <<
5555 array, so may know the correct default loca <<
5556 <<
5557 Note that the shared_info page may be const <<
5558 it contains the event channel bitmap used t <<
5559 a Xen guest, amongst other things. It is ex <<
5560 mechanisms — KVM will not explicitly mark <<
5561 time an event channel interrupt is delivere <<
5562 userspace should always assume that the des <<
5563 any vCPU has been running or any event chan <<
5564 routed to the guest. <<
5565 <<
5566 Setting the gfn to KVM_XEN_INVALID_GFN will <<
5567 page. <<
5568 <<
5569 KVM_XEN_ATTR_TYPE_SHARED_INFO_HVA <<
5570 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA f <<
5571 Xen capabilities, then this attribute may b <<
5572 userspace address at which the shared_info <<
5573 will always be fixed in the VMM regardless <<
5574 in guest physical address space. This attri <<
5575 preference to KVM_XEN_ATTR_TYPE_SHARED_INFO <<
5576 unnecessary invalidation of an internal cac <<
5577 re-mapped in guest physcial address space. <<
5578 <<
5579 Setting the hva to zero will disable the sh <<
5580 <<
5581 KVM_XEN_ATTR_TYPE_UPCALL_VECTOR <<
5582 Sets the exception vector used to deliver X <<
5583 This is the HVM-wide vector injected direct <<
5584 (not through the local APIC), typically con <<
5585 HVM_PARAM_CALLBACK_IRQ. This can be disable <<
5586 SHUTDOWN_soft_reset) by setting it to zero. <<
5587 <<
5588 KVM_XEN_ATTR_TYPE_EVTCHN <<
5589 This attribute is available when the KVM_CA <<
5590 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND <<
5591 an outbound port number for interception of <<
5592 from the guest. A given sending port number <<
5593 a specified vCPU (by APIC ID) / port / prio <<
5594 trigger events on an eventfd. The vCPU and <<
5595 by setting KVM_XEN_EVTCHN_UPDATE in a subse <<
5596 fields cannot change for a given sending po <<
5597 removed by using KVM_XEN_EVTCHN_DEASSIGN in <<
5598 KVM_XEN_EVTCHN_RESET in the flags field rem <<
5599 outbound event channels. The values of the <<
5600 exclusive and cannot be combined as a bitma <<
5601 <<
5602 KVM_XEN_ATTR_TYPE_XEN_VERSION <<
5603 This attribute is available when the KVM_CA <<
5604 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND <<
5605 the 32-bit version code returned to the gue <<
5606 XENVER_version call; typically (XEN_MAJOR < <<
5607 Xen guests will often use this to as a dumm <<
5608 event channel delivery, so responding withi <<
5609 exiting to userspace is beneficial. <<
5610 <<
5611 KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG <<
5612 This attribute is available when the KVM_CA <<
5613 support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPD <<
5614 XEN_RUNSTATE_UPDATE flag which allows guest <<
5615 other vCPUs' vcpu_runstate_info. Xen guests <<
5616 the VMASST_TYPE_runstate_update_flag of the <<
5617 hypercall. <<
5618 <<
5619 4.127 KVM_XEN_HVM_GET_ATTR <<
5620 -------------------------- <<
5621 <<
5622 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CO <<
5623 :Architectures: x86 <<
5624 :Type: vm ioctl <<
5625 :Parameters: struct kvm_xen_hvm_attr <<
5626 :Returns: 0 on success, < 0 on error <<
5627 <<
5628 Allows Xen VM attributes to be read. For the <<
5629 see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_A <<
5630 attribute cannot be read. <<
5631 <<
5632 4.128 KVM_XEN_VCPU_SET_ATTR <<
5633 --------------------------- <<
5634 <<
5635 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CO <<
5636 :Architectures: x86 <<
5637 :Type: vcpu ioctl <<
5638 :Parameters: struct kvm_xen_vcpu_attr <<
5639 :Returns: 0 on success, < 0 on error <<
5640 <<
5641 :: <<
5642 <<
5643 struct kvm_xen_vcpu_attr { <<
5644 __u16 type; <<
5645 __u16 pad[3]; <<
5646 union { <<
5647 __u64 gpa; <<
5648 __u64 pad[4]; <<
5649 struct { <<
5650 __u64 state; <<
5651 __u64 state_entry_tim <<
5652 __u64 time_running; <<
5653 __u64 time_runnable; <<
5654 __u64 time_blocked; <<
5655 __u64 time_offline; <<
5656 } runstate; <<
5657 __u32 vcpu_id; <<
5658 struct { <<
5659 __u32 port; <<
5660 __u32 priority; <<
5661 __u64 expires_ns; <<
5662 } timer; <<
5663 __u8 vector; <<
5664 } u; <<
5665 }; <<
5666 <<
5667 type values: <<
5668 <<
5669 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO <<
5670 Sets the guest physical address of the vcpu <<
5671 As with the shared_info page for the VM, th <<
5672 dirtied at any time if event channel interr <<
5673 userspace should always assume that the pag <<
5674 on dirty logging. Setting the gpa to KVM_XE <<
5675 the vcpu_info. <<
5676 <<
5677 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA <<
5678 If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA f <<
5679 Xen capabilities, then this attribute may b <<
5680 userspace address of the vcpu_info for a gi <<
5681 only be used when the vcpu_info resides at <<
5682 in the shared_info page. In this case it is <<
5683 userspace address will not change, because <<
5684 an overlay on guest memory and remains at a <<
5685 regardless of where it is mapped in guest p <<
5686 and hence unnecessary invalidation of an in <<
5687 avoided if the guest memory layout is modif <<
5688 If the vcpu_info does not reside at the "de <<
5689 it is not guaranteed to remain at the same <<
5690 hence the aforementioned cache invalidation <<
5691 <<
5692 KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO <<
5693 Sets the guest physical address of an addit <<
5694 for a given vCPU. This is typically used fo <<
5695 Setting the gpa to KVM_XEN_INVALID_GPA will <<
5696 <<
5697 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR <<
5698 Sets the guest physical address of the vcpu <<
5699 vCPU. This is how a Xen guest tracks CPU st <<
5700 Setting the gpa to KVM_XEN_INVALID_GPA will <<
5701 <<
5702 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT <<
5703 Sets the runstate (RUNSTATE_running/_runnab <<
5704 the given vCPU from the .u.runstate.state m <<
5705 KVM automatically accounts running and runn <<
5706 and offline states are only entered explici <<
5707 <<
5708 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA <<
5709 Sets all fields of the vCPU runstate data f <<
5710 of the structure, including the current run <<
5711 must equal the sum of the other four times. <<
5712 <<
5713 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST <<
5714 This *adds* the contents of the .u.runstate <<
5715 to the corresponding members of the given v <<
5716 permitting atomic adjustments to the runsta <<
5717 to the state_entry_time must equal the sum <<
5718 other four times. The state field must be s <<
5719 runstate value (RUNSTATE_running, RUNSTATE_ <<
5720 or RUNSTATE_offline) to set the current acc <<
5721 adjusted state_entry_time. <<
5722 <<
5723 KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID <<
5724 This attribute is available when the KVM_CA <<
5725 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND <<
5726 vCPU ID of the given vCPU, to allow timer-r <<
5727 be intercepted by KVM. <<
5728 <<
5729 KVM_XEN_VCPU_ATTR_TYPE_TIMER <<
5730 This attribute is available when the KVM_CA <<
5731 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND <<
5732 event channel port/priority for the VIRQ_TI <<
5733 as allowing a pending timer to be saved/res <<
5734 port to zero disables kernel handling of th <<
5735 <<
5736 KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR <<
5737 This attribute is available when the KVM_CA <<
5738 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND <<
5739 per-vCPU local APIC upcall vector, configur <<
5740 the HVMOP_set_evtchn_upcall_vector hypercal <<
5741 used by Windows guests, and is distinct fro <<
5742 vector configured with HVM_PARAM_CALLBACK_I <<
5743 setting the vector to zero. <<
5744 <<
5745 <<
5746 4.129 KVM_XEN_VCPU_GET_ATTR <<
5747 --------------------------- <<
5748 <<
5749 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CO <<
5750 :Architectures: x86 <<
5751 :Type: vcpu ioctl <<
5752 :Parameters: struct kvm_xen_vcpu_attr <<
5753 :Returns: 0 on success, < 0 on error <<
5754 <<
5755 Allows Xen vCPU attributes to be read. For th <<
5756 see KVM_XEN_VCPU_SET_ATTR above. <<
5757 <<
5758 The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST ty <<
5759 with the KVM_XEN_VCPU_GET_ATTR ioctl. <<
5760 <<
5761 4.130 KVM_ARM_MTE_COPY_TAGS <<
5762 --------------------------- <<
5763 <<
5764 :Capability: KVM_CAP_ARM_MTE <<
5765 :Architectures: arm64 <<
5766 :Type: vm ioctl <<
5767 :Parameters: struct kvm_arm_copy_mte_tags <<
5768 :Returns: number of bytes copied, < 0 on erro <<
5769 arguments, -EFAULT if memory cannot <<
5770 <<
5771 :: <<
5772 <<
5773 struct kvm_arm_copy_mte_tags { <<
5774 __u64 guest_ipa; <<
5775 __u64 length; <<
5776 void __user *addr; <<
5777 __u64 flags; <<
5778 __u64 reserved[2]; <<
5779 }; <<
5780 <<
5781 Copies Memory Tagging Extension (MTE) tags to <<
5782 ``guest_ipa`` and ``length`` fields must be ` <<
5783 ``length`` must not be bigger than 2^31 - PAG <<
5784 field must point to a buffer which the tags w <<
5785 <<
5786 ``flags`` specifies the direction of copy, ei <<
5787 ``KVM_ARM_TAGS_FROM_GUEST``. <<
5788 <<
5789 The size of the buffer to store the tags is ` <<
5790 (granules in MTE are 16 bytes long). Each byt <<
5791 value. This matches the format of ``PTRACE_PE <<
5792 ``PTRACE_POKEMTETAGS``. <<
5793 <<
5794 If an error occurs before any data is copied <<
5795 returned. If some tags have been copied befor <<
5796 of bytes successfully copied is returned. If <<
5797 then ``length`` is returned. <<
5798 <<
5799 4.131 KVM_GET_SREGS2 <<
5800 -------------------- <<
5801 <<
5802 :Capability: KVM_CAP_SREGS2 <<
5803 :Architectures: x86 <<
5804 :Type: vcpu ioctl <<
5805 :Parameters: struct kvm_sregs2 (out) <<
5806 :Returns: 0 on success, -1 on error <<
5807 <<
5808 Reads special registers from the vcpu. <<
5809 This ioctl (when supported) replaces the KVM_ <<
5810 <<
5811 :: <<
5812 <<
5813 struct kvm_sregs2 { <<
5814 /* out (KVM_GET_SREGS2) / in <<
5815 struct kvm_segment cs, ds, es <<
5816 struct kvm_segment tr, ldt; <<
5817 struct kvm_dtable gdt, idt; <<
5818 __u64 cr0, cr2, cr3, cr4, cr8 <<
5819 __u64 efer; <<
5820 __u64 apic_base; <<
5821 __u64 flags; <<
5822 __u64 pdptrs[4]; <<
5823 }; <<
5824 <<
5825 flags values for ``kvm_sregs2``: <<
5826 <<
5827 ``KVM_SREGS2_FLAGS_PDPTRS_VALID`` <<
5828 <<
5829 Indicates that the struct contains valid PD <<
5830 <<
5831 <<
5832 4.132 KVM_SET_SREGS2 <<
5833 -------------------- <<
5834 <<
5835 :Capability: KVM_CAP_SREGS2 <<
5836 :Architectures: x86 <<
5837 :Type: vcpu ioctl <<
5838 :Parameters: struct kvm_sregs2 (in) <<
5839 :Returns: 0 on success, -1 on error <<
5840 <<
5841 Writes special registers into the vcpu. <<
5842 See KVM_GET_SREGS2 for the data structures. <<
5843 This ioctl (when supported) replaces the KVM_ <<
5844 <<
5845 4.133 KVM_GET_STATS_FD <<
5846 ---------------------- <<
5847 <<
5848 :Capability: KVM_CAP_STATS_BINARY_FD <<
5849 :Architectures: all <<
5850 :Type: vm ioctl, vcpu ioctl <<
5851 :Parameters: none <<
5852 :Returns: statistics file descriptor on succe <<
5853 <<
5854 Errors: <<
5855 <<
5856 ====== ================================ <<
5857 ENOMEM if the fd could not be created d <<
5858 EMFILE if the number of opened files ex <<
5859 ====== ================================ <<
5860 <<
5861 The returned file descriptor can be used to r <<
5862 binary format. The data in the file descripto <<
5863 organized as follows: <<
5864 <<
5865 +-------------+ <<
5866 | Header | <<
5867 +-------------+ <<
5868 | id string | <<
5869 +-------------+ <<
5870 | Descriptors | <<
5871 +-------------+ <<
5872 | Stats Data | <<
5873 +-------------+ <<
5874 <<
5875 Apart from the header starting at offset 0, p <<
5876 not guaranteed that the four blocks are adjac <<
5877 the offsets of the id, descriptors and data b <<
5878 header. However, all four blocks are aligned <<
5879 file and they do not overlap. <<
5880 <<
5881 All blocks except the data block are immutabl <<
5882 only one time after retrieving the file descr <<
5883 ``lseek`` to read the statistics repeatedly. <<
5884 <<
5885 All data is in system endianness. <<
5886 <<
5887 The format of the header is as follows:: <<
5888 <<
5889 struct kvm_stats_header { <<
5890 __u32 flags; <<
5891 __u32 name_size; <<
5892 __u32 num_desc; <<
5893 __u32 id_offset; <<
5894 __u32 desc_offset; <<
5895 __u32 data_offset; <<
5896 }; <<
5897 <<
5898 The ``flags`` field is not used at the moment <<
5899 <<
5900 The ``name_size`` field is the size (in byte) <<
5901 (including trailing '\0') which is contained <<
5902 appended at the end of every descriptor. <<
5903 <<
5904 The ``num_desc`` field is the number of descr <<
5905 descriptor block. (The actual number of valu <<
5906 larger, since each descriptor may comprise mo <<
5907 <<
5908 The ``id_offset`` field is the offset of the <<
5909 file indicated by the file descriptor. It is <<
5910 <<
5911 The ``desc_offset`` field is the offset of th <<
5912 of the file indicated by the file descriptor. <<
5913 <<
5914 The ``data_offset`` field is the offset of th <<
5915 of the file indicated by the file descriptor. <<
5916 <<
5917 The id string block contains a string which i <<
5918 which KVM_GET_STATS_FD was invoked. The size <<
5919 trailing ``'\0'``, is indicated by the ``name <<
5920 <<
5921 The descriptors block is only needed to be re <<
5922 file descriptor contains a sequence of ``stru <<
5923 by a string of size ``name_size``. <<
5924 :: <<
5925 <<
5926 #define KVM_STATS_TYPE_SHIFT <<
5927 #define KVM_STATS_TYPE_MASK <<
5928 #define KVM_STATS_TYPE_CUMULATIVE <<
5929 #define KVM_STATS_TYPE_INSTANT <<
5930 #define KVM_STATS_TYPE_PEAK <<
5931 #define KVM_STATS_TYPE_LINEAR_HIST <<
5932 #define KVM_STATS_TYPE_LOG_HIST <<
5933 #define KVM_STATS_TYPE_MAX <<
5934 <<
5935 #define KVM_STATS_UNIT_SHIFT <<
5936 #define KVM_STATS_UNIT_MASK <<
5937 #define KVM_STATS_UNIT_NONE <<
5938 #define KVM_STATS_UNIT_BYTES <<
5939 #define KVM_STATS_UNIT_SECONDS <<
5940 #define KVM_STATS_UNIT_CYCLES <<
5941 #define KVM_STATS_UNIT_BOOLEAN <<
5942 #define KVM_STATS_UNIT_MAX <<
5943 <<
5944 #define KVM_STATS_BASE_SHIFT <<
5945 #define KVM_STATS_BASE_MASK <<
5946 #define KVM_STATS_BASE_POW10 <<
5947 #define KVM_STATS_BASE_POW2 <<
5948 #define KVM_STATS_BASE_MAX <<
5949 <<
5950 struct kvm_stats_desc { <<
5951 __u32 flags; <<
5952 __s16 exponent; <<
5953 __u16 size; <<
5954 __u32 offset; <<
5955 __u32 bucket_size; <<
5956 char name[]; <<
5957 }; <<
5958 <<
5959 The ``flags`` field contains the type and uni <<
5960 by this descriptor. Its endianness is CPU nat <<
5961 The following flags are supported: <<
5962 <<
5963 Bits 0-3 of ``flags`` encode the type: <<
5964 <<
5965 * ``KVM_STATS_TYPE_CUMULATIVE`` <<
5966 The statistics reports a cumulative count <<
5967 Most of the counters used in KVM are of t <<
5968 The corresponding ``size`` field for this <<
5969 All cumulative statistics data are read/w <<
5970 * ``KVM_STATS_TYPE_INSTANT`` <<
5971 The statistics reports an instantaneous v <<
5972 decreased. This type is usually used as a <<
5973 like the number of dirty pages, the numbe <<
5974 All instant statistics are read only. <<
5975 The corresponding ``size`` field for this <<
5976 * ``KVM_STATS_TYPE_PEAK`` <<
5977 The statistics data reports a peak value, <<
5978 of items in a hash table bucket, the long <<
5979 The value of data can only be increased. <<
5980 The corresponding ``size`` field for this <<
5981 * ``KVM_STATS_TYPE_LINEAR_HIST`` <<
5982 The statistic is reported as a linear his <<
5983 buckets is specified by the ``size`` fiel <<
5984 by the ``hist_param`` field. The range of <<
5985 is [``hist_param``*(N-1), ``hist_param``* <<
5986 bucket is [``hist_param``*(``size``-1), + <<
5987 value.) <<
5988 * ``KVM_STATS_TYPE_LOG_HIST`` <<
5989 The statistic is reported as a logarithmi <<
5990 buckets is specified by the ``size`` fiel <<
5991 [0, 1), while the range of the last bucke <<
5992 Otherwise, The Nth bucket (1 < N < ``size <<
5993 [pow(2, N-2), pow(2, N-1)). <<
5994 <<
5995 Bits 4-7 of ``flags`` encode the unit: <<
5996 <<
5997 * ``KVM_STATS_UNIT_NONE`` <<
5998 There is no unit for the value of statist <<
5999 the value is a simple counter of an event <<
6000 * ``KVM_STATS_UNIT_BYTES`` <<
6001 It indicates that the statistics data is <<
6002 unit of Byte, KiByte, MiByte, GiByte, etc <<
6003 determined by the ``exponent`` field in t <<
6004 * ``KVM_STATS_UNIT_SECONDS`` <<
6005 It indicates that the statistics data is <<
6006 * ``KVM_STATS_UNIT_CYCLES`` <<
6007 It indicates that the statistics data is <<
6008 * ``KVM_STATS_UNIT_BOOLEAN`` <<
6009 It indicates that the statistic will alwa <<
6010 statistics of "peak" type will never go b <<
6011 statistics can be linear histograms (with <<
6012 histograms. <<
6013 <<
6014 Note that, in the case of histograms, the uni <<
6015 ranges, while the bucket value indicates how <<
6016 bucket's range. <<
6017 <<
6018 Bits 8-11 of ``flags``, together with ``expon <<
6019 unit: <<
6020 <<
6021 * ``KVM_STATS_BASE_POW10`` <<
6022 The scale is based on power of 10. It is <<
6023 CPU clock cycles. For example, an expone <<
6024 ``KVM_STATS_UNIT_SECONDS`` to express tha <<
6025 * ``KVM_STATS_BASE_POW2`` <<
6026 The scale is based on power of 2. It is u <<
6027 For example, an exponent of 20 can be use <<
6028 express that the unit is MiB. <<
6029 <<
6030 The ``size`` field is the number of values of <<
6031 value is usually 1 for most of simple statist <<
6032 unsigned 64bit data. <<
6033 <<
6034 The ``offset`` field is the offset from the s <<
6035 the corresponding statistics data. <<
6036 <<
6037 The ``bucket_size`` field is used as a parame <<
6038 It is only used by linear histogram statistic <<
6039 bucket in the unit expressed by bits 4-11 of <<
6040 <<
6041 The ``name`` field is the name string of the <<
6042 starts at the end of ``struct kvm_stats_desc` <<
6043 the trailing ``'\0'``, is indicated by ``name <<
6044 <<
6045 The Stats Data block contains an array of 64- <<
6046 as the descriptors in Descriptors block. <<
6047 <<
6048 4.134 KVM_GET_XSAVE2 <<
6049 -------------------- <<
6050 <<
6051 :Capability: KVM_CAP_XSAVE2 <<
6052 :Architectures: x86 <<
6053 :Type: vcpu ioctl <<
6054 :Parameters: struct kvm_xsave (out) <<
6055 :Returns: 0 on success, -1 on error <<
6056 <<
6057 <<
6058 :: <<
6059 <<
6060 struct kvm_xsave { <<
6061 __u32 region[1024]; <<
6062 __u32 extra[0]; <<
6063 }; <<
6064 <<
6065 This ioctl would copy current vcpu's xsave st <<
6066 copies as many bytes as are returned by KVM_C <<
6067 when invoked on the vm file descriptor. The s <<
6068 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will alwa <<
6069 Currently, it is only greater than 4096 if a <<
6070 enabled with ``arch_prctl()``, but this may c <<
6071 <<
6072 The offsets of the state save areas in struct <<
6073 of CPUID leaf 0xD on the host. <<
6074 <<
6075 4.135 KVM_XEN_HVM_EVTCHN_SEND <<
6076 ----------------------------- <<
6077 4739
6078 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CO !! 4740 :Capability: KVM_X86_SET_MSR_FILTER
6079 :Architectures: x86 4741 :Architectures: x86
6080 :Type: vm ioctl 4742 :Type: vm ioctl
6081 :Parameters: struct kvm_irq_routing_xen_evtch !! 4743 :Parameters: struct kvm_msr_filter
6082 :Returns: 0 on success, < 0 on error 4744 :Returns: 0 on success, < 0 on error
6083 4745
6084 <<
6085 :: 4746 ::
6086 4747
6087 struct kvm_irq_routing_xen_evtchn { !! 4748 struct kvm_msr_filter_range {
6088 __u32 port; !! 4749 #define KVM_MSR_FILTER_READ (1 << 0)
6089 __u32 vcpu; !! 4750 #define KVM_MSR_FILTER_WRITE (1 << 1)
6090 __u32 priority; !! 4751 __u32 flags;
6091 }; !! 4752 __u32 nmsrs; /* number of msrs in bitmap */
6092 !! 4753 __u32 base; /* MSR index the bitmap starts at */
6093 This ioctl injects an event channel interrupt !! 4754 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
6094 <<
6095 4.136 KVM_S390_PV_CPU_COMMAND <<
6096 ----------------------------- <<
6097 <<
6098 :Capability: KVM_CAP_S390_PROTECTED_DUMP <<
6099 :Architectures: s390 <<
6100 :Type: vcpu ioctl <<
6101 :Parameters: none <<
6102 :Returns: 0 on success, < 0 on error <<
6103 <<
6104 This ioctl closely mirrors `KVM_S390_PV_COMMA <<
6105 for vcpus. It re-uses the kvm_s390_pv_dmp str <<
6106 the command ids. <<
6107 <<
6108 **command:** <<
6109 <<
6110 KVM_PV_DUMP <<
6111 Presents an API that provides calls which f <<
6112 of a protected VM. <<
6113 <<
6114 **subcommand:** <<
6115 <<
6116 KVM_PV_DUMP_CPU <<
6117 Provides encrypted dump data like register <<
6118 The length of the returned data is provided <<
6119 <<
6120 4.137 KVM_S390_ZPCI_OP <<
6121 ---------------------- <<
6122 <<
6123 :Capability: KVM_CAP_S390_ZPCI_OP <<
6124 :Architectures: s390 <<
6125 :Type: vm ioctl <<
6126 :Parameters: struct kvm_s390_zpci_op (in) <<
6127 :Returns: 0 on success, <0 on error <<
6128 <<
6129 Used to manage hardware-assisted virtualizati <<
6130 <<
6131 Parameters are specified via the following st <<
6132 <<
6133 struct kvm_s390_zpci_op { <<
6134 /* in */ <<
6135 __u32 fh; /* target dev <<
6136 __u8 op; /* operation <<
6137 __u8 pad[3]; <<
6138 union { <<
6139 /* for KVM_S390_ZPCIOP_REG_AE <<
6140 struct { <<
6141 __u64 ibv; /* Gu <<
6142 __u64 sb; /* Gu <<
6143 __u32 flags; <<
6144 __u32 noi; /* Nu <<
6145 __u8 isc; /* Gu <<
6146 __u8 sbo; /* Of <<
6147 __u16 pad; <<
6148 } reg_aen; <<
6149 __u64 reserved[8]; <<
6150 } u; <<
6151 }; 4755 };
6152 4756
6153 The type of operation is specified in the "op !! 4757 #define KVM_MSR_FILTER_MAX_RANGES 16
6154 KVM_S390_ZPCIOP_REG_AEN is used to register t !! 4758 struct kvm_msr_filter {
6155 notification interpretation, which will allow !! 4759 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
6156 events directly to the vm, with KVM providing !! 4760 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
6157 KVM_S390_ZPCIOP_DEREG_AEN is used to subseque <<
6158 adapter event notifications. <<
6159 <<
6160 The target zPCI function must also be specifi <<
6161 KVM_S390_ZPCIOP_REG_AEN operation, additional <<
6162 delivery must be provided via the "reg_aen" s <<
6163 <<
6164 The "pad" and "reserved" fields may be used f <<
6165 set to 0s by userspace. <<
6166 <<
6167 4.138 KVM_ARM_SET_COUNTER_OFFSET <<
6168 -------------------------------- <<
6169 <<
6170 :Capability: KVM_CAP_COUNTER_OFFSET <<
6171 :Architectures: arm64 <<
6172 :Type: vm ioctl <<
6173 :Parameters: struct kvm_arm_counter_offset (i <<
6174 :Returns: 0 on success, < 0 on error <<
6175 <<
6176 This capability indicates that userspace is a <<
6177 offset to both the virtual and physical count <<
6178 using the KVM_ARM_SET_CNT_OFFSET ioctl and th <<
6179 <<
6180 :: <<
6181 <<
6182 struct kvm_arm_counter_offset { <<
6183 __u64 counter_offset; <<
6184 __u64 reserved; <<
6185 }; <<
6186 <<
6187 The offset describes a number of counter cycl <<
6188 both virtual and physical counter views (simi <<
6189 CNTVOFF_EL2 and CNTPOFF_EL2 system registers, <<
6190 always applies to all vcpus (already created <<
6191 for this VM. <<
6192 <<
6193 It is userspace's responsibility to compute t <<
6194 on previous values of the guest counters. <<
6195 <<
6196 Any value other than 0 for the "reserved" fie <<
6197 (-EINVAL) being returned. This ioctl can also <<
6198 ioctl is issued concurrently. <<
6199 <<
6200 Note that using this ioctl results in KVM ign <<
6201 writes to the CNTVCT_EL0 and CNTPCT_EL0 regis <<
6202 interface. No error will be returned, but the <<
6203 applied. <<
6204 <<
6205 .. _KVM_ARM_GET_REG_WRITABLE_MASKS: <<
6206 <<
6207 4.139 KVM_ARM_GET_REG_WRITABLE_MASKS <<
6208 ------------------------------------------- <<
6209 <<
6210 :Capability: KVM_CAP_ARM_SUPPORTED_REG_MASK_R <<
6211 :Architectures: arm64 <<
6212 :Type: vm ioctl <<
6213 :Parameters: struct reg_mask_range (in/out) <<
6214 :Returns: 0 on success, < 0 on error <<
6215 <<
6216 <<
6217 :: <<
6218 <<
6219 #define KVM_ARM_FEATURE_ID_RANGE <<
6220 #define KVM_ARM_FEATURE_ID_RANGE_SIZE <<
6221 <<
6222 struct reg_mask_range { <<
6223 __u64 addr; /* Po <<
6224 __u32 range; /* Re <<
6225 __u32 reserved[13]; <<
6226 }; <<
6227 <<
6228 This ioctl copies the writable masks for a se <<
6229 userspace. <<
6230 <<
6231 The ``addr`` field is a pointer to the destin <<
6232 the writable masks. <<
6233 <<
6234 The ``range`` field indicates the requested r <<
6235 ``KVM_CHECK_EXTENSION`` for the ``KVM_CAP_ARM <<
6236 capability returns the supported ranges, expr <<
6237 flag's bit index represents a possible value <<
6238 All other values are reserved for future use <<
6239 <<
6240 The ``reserved[13]`` array is reserved for fu <<
6241 KVM may return an error. <<
6242 <<
6243 KVM_ARM_FEATURE_ID_RANGE (0) <<
6244 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ <<
6245 <<
6246 The Feature ID range is defined as the AArch6 <<
6247 op0==3, op1=={0, 1, 3}, CRn==0, CRm=={0-7}, o <<
6248 <<
6249 The mask returned array pointed to by ``addr` <<
6250 ``ARM64_FEATURE_ID_RANGE_IDX(op0, op1, crn, c <<
6251 to know what fields can be changed for the sy <<
6252 ``op0, op1, crn, crm, op2``. KVM rejects ID r <<
6253 superset of the features supported by the sys <<
6254 <<
6255 4.140 KVM_SET_USER_MEMORY_REGION2 <<
6256 --------------------------------- <<
6257 <<
6258 :Capability: KVM_CAP_USER_MEMORY2 <<
6259 :Architectures: all <<
6260 :Type: vm ioctl <<
6261 :Parameters: struct kvm_userspace_memory_regi <<
6262 :Returns: 0 on success, -1 on error <<
6263 <<
6264 KVM_SET_USER_MEMORY_REGION2 is an extension t <<
6265 allows mapping guest_memfd memory into a gues <<
6266 KVM_SET_USER_MEMORY_REGION identically. User <<
6267 in flags to have KVM bind the memory region t <<
6268 [guest_memfd_offset, guest_memfd_offset + mem <<
6269 must point at a file created via KVM_CREATE_G <<
6270 the target range must not be bound to any oth <<
6271 bounds checks apply (use common sense). <<
6272 <<
6273 :: <<
6274 <<
6275 struct kvm_userspace_memory_region2 { <<
6276 __u32 slot; <<
6277 __u32 flags; 4761 __u32 flags;
6278 __u64 guest_phys_addr; !! 4762 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
6279 __u64 memory_size; /* bytes */ <<
6280 __u64 userspace_addr; /* start of the <<
6281 __u64 guest_memfd_offset; <<
6282 __u32 guest_memfd; <<
6283 __u32 pad1; <<
6284 __u64 pad2[14]; <<
6285 }; <<
6286 <<
6287 A KVM_MEM_GUEST_MEMFD region _must_ have a va <<
6288 userspace_addr (shared memory). However, "va <<
6289 means that the address itself must be a legal <<
6290 mapping for userspace_addr is not required to <<
6291 KVM_SET_USER_MEMORY_REGION2, e.g. shared memo <<
6292 on-demand. <<
6293 <<
6294 When mapping a gfn into the guest, KVM select <<
6295 userspace_addr vs. guest_memfd, based on the <<
6296 state. At VM creation time, all memory is sh <<
6297 is '0' for all gfns. Userspace can control w <<
6298 toggling KVM_MEMORY_ATTRIBUTE_PRIVATE via KVM <<
6299 <<
6300 S390: <<
6301 ^^^^^ <<
6302 <<
6303 Returns -EINVAL if the VM has the KVM_VM_S390 <<
6304 Returns -EINVAL if called on a protected VM. <<
6305 <<
6306 4.141 KVM_SET_MEMORY_ATTRIBUTES <<
6307 ------------------------------- <<
6308 <<
6309 :Capability: KVM_CAP_MEMORY_ATTRIBUTES <<
6310 :Architectures: x86 <<
6311 :Type: vm ioctl <<
6312 :Parameters: struct kvm_memory_attributes (in <<
6313 :Returns: 0 on success, <0 on error <<
6314 <<
6315 KVM_SET_MEMORY_ATTRIBUTES allows userspace to <<
6316 of guest physical memory. <<
6317 <<
6318 :: <<
6319 <<
6320 struct kvm_memory_attributes { <<
6321 __u64 address; <<
6322 __u64 size; <<
6323 __u64 attributes; <<
6324 __u64 flags; <<
6325 }; 4763 };
6326 4764
6327 #define KVM_MEMORY_ATTRIBUTE_PRIVATE !! 4765 flags values for ``struct kvm_msr_filter_range``:
6328 <<
6329 The address and size must be page aligned. T <<
6330 retrieved via ioctl(KVM_CHECK_EXTENSION) on K <<
6331 executed on a VM, KVM_CAP_MEMORY_ATTRIBUTES p <<
6332 supported by that VM. If executed at system <<
6333 returns all attributes supported by KVM. The <<
6334 time is KVM_MEMORY_ATTRIBUTE_PRIVATE, which m <<
6335 guest private memory. <<
6336 <<
6337 Note, there is no "get" API. Userspace is re <<
6338 the state of a gfn/page as needed. <<
6339 4766
6340 The "flags" field is reserved for future exte !! 4767 ``KVM_MSR_FILTER_READ``
6341 4768
6342 4.142 KVM_CREATE_GUEST_MEMFD !! 4769 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
6343 ---------------------------- !! 4770 indicates that a read should immediately fail, while a 1 indicates that
>> 4771 a read for a particular MSR should be handled regardless of the default
>> 4772 filter action.
6344 4773
6345 :Capability: KVM_CAP_GUEST_MEMFD !! 4774 ``KVM_MSR_FILTER_WRITE``
6346 :Architectures: none <<
6347 :Type: vm ioctl <<
6348 :Parameters: struct kvm_create_guest_memfd(in <<
6349 :Returns: A file descriptor on success, <0 on <<
6350 4775
6351 KVM_CREATE_GUEST_MEMFD creates an anonymous f !! 4776 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
6352 that refers to it. guest_memfd files are rou !! 4777 indicates that a write should immediately fail, while a 1 indicates that
6353 via memfd_create(), e.g. guest_memfd files li !! 4778 a write for a particular MSR should be handled regardless of the default
6354 and are automatically released when the last !! 4779 filter action.
6355 "regular" memfd_create() files, guest_memfd f <<
6356 virtual machine (see below), cannot be mapped <<
6357 and cannot be resized (guest_memfd files do <<
6358 4780
6359 :: !! 4781 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
6360 4782
6361 struct kvm_create_guest_memfd { !! 4783 Filter both read and write accesses to MSRs using the given bitmap. A 0
6362 __u64 size; !! 4784 in the bitmap indicates that both reads and writes should immediately fail,
6363 __u64 flags; !! 4785 while a 1 indicates that reads and writes for a particular MSR are not
6364 __u64 reserved[6]; !! 4786 filtered by this range.
6365 }; <<
6366 4787
6367 Conceptually, the inode backing a guest_memfd !! 4788 flags values for ``struct kvm_msr_filter``:
6368 i.e. is coupled to the virtual machine as a t <<
6369 file itself, which is bound to a "struct kvm" <<
6370 underlying memory, e.g. effectively provides <<
6371 to host memory. This allows for use cases wh <<
6372 used to manage a single virtual machine, e.g. <<
6373 migration of a virtual machine. <<
6374 <<
6375 KVM currently only supports mapping guest_mem <<
6376 and more specifically via the guest_memfd and <<
6377 "struct kvm_userspace_memory_region2", where <<
6378 into the guest_memfd instance. For a given g <<
6379 most one mapping per page, i.e. binding multi <<
6380 guest_memfd range is not allowed (any number <<
6381 a single guest_memfd file, but the bound rang <<
6382 4789
6383 See KVM_SET_USER_MEMORY_REGION2 for additiona !! 4790 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
6384 4791
6385 4.143 KVM_PRE_FAULT_MEMORY !! 4792 If no filter range matches an MSR index that is getting accessed, KVM will
6386 --------------------------- !! 4793 fall back to allowing access to the MSR.
6387 4794
6388 :Capability: KVM_CAP_PRE_FAULT_MEMORY !! 4795 ``KVM_MSR_FILTER_DEFAULT_DENY``
6389 :Architectures: none <<
6390 :Type: vcpu ioctl <<
6391 :Parameters: struct kvm_pre_fault_memory (in/ <<
6392 :Returns: 0 if at least one page is processed <<
6393 4796
6394 Errors: !! 4797 If no filter range matches an MSR index that is getting accessed, KVM will
>> 4798 fall back to rejecting access to the MSR. In this mode, all MSRs that should
>> 4799 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
6395 4800
6396 ========== ================================ !! 4801 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
6397 EINVAL The specified `gpa` and `size` w !! 4802 specify whether a certain MSR access should be explicitly filtered for or not.
6398 page aligned, causes an overflow <<
6399 ENOENT The specified `gpa` is outside d <<
6400 EINTR An unmasked signal is pending an <<
6401 EFAULT The parameter address was invali <<
6402 EOPNOTSUPP Mapping memory for a GPA is unsu <<
6403 hypervisor, and/or for the curre <<
6404 EIO unexpected error conditions (als <<
6405 ========== ================================ <<
6406 4803
6407 :: !! 4804 If this ioctl has never been invoked, MSR accesses are not guarded and the
>> 4805 default KVM in-kernel emulation behavior is fully preserved.
6408 4806
6409 struct kvm_pre_fault_memory { !! 4807 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
6410 /* in/out */ !! 4808 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
6411 __u64 gpa; !! 4809 an error.
6412 __u64 size; <<
6413 /* in */ <<
6414 __u64 flags; <<
6415 __u64 padding[5]; <<
6416 }; <<
6417 4810
6418 KVM_PRE_FAULT_MEMORY populates KVM's stage-2 !! 4811 As soon as the filtering is in place, every MSR access is processed through
6419 for the current vCPU state. KVM maps memory !! 4812 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
6420 stage-2 read page fault, e.g. faults in memor !! 4813 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
6421 CoW. However, KVM does not mark any newly cr !! 4814 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
6422 !! 4815 register.
6423 In the case of confidential VM types where th !! 4816
6424 private guest memory before the guest is 'fin !! 4817 If a bit is within one of the defined ranges, read and write accesses are
6425 should only be issued after completing all th !! 4818 guarded by the bitmap's value for the MSR index if the kind of access
6426 guest into a 'finalized' state so that the ab !! 4819 is included in the ``struct kvm_msr_filter_range`` flags. If no range
6427 ensured. !! 4820 cover this particular access, the behavior is determined by the flags
6428 !! 4821 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
6429 In some cases, multiple vCPUs might share the !! 4822 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
6430 case, the ioctl can be called in parallel. !! 4823
6431 !! 4824 Each bitmap range specifies a range of MSRs to potentially allow access on.
6432 When the ioctl returns, the input values are !! 4825 The range goes from MSR index [base .. base+nmsrs]. The flags field
6433 remaining range. If `size` > 0 on return, th !! 4826 indicates whether reads, writes or both reads and writes are filtered
6434 the ioctl again with the same `struct kvm_map !! 4827 by setting a 1 bit in the bitmap for the corresponding MSR index.
6435 !! 4828
6436 Shadow page tables cannot support this ioctl !! 4829 If an MSR access is not permitted through the filtering, it generates a
6437 are indexed by virtual address or nested gues !! 4830 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
6438 Calling this ioctl when the guest is using sh !! 4831 allows user space to deflect and potentially handle various MSR accesses
6439 example because it is running a nested guest !! 4832 into user space.
6440 will fail with `EOPNOTSUPP` even if `KVM_CHEC <<
6441 the capability to be present. <<
6442 4833
6443 `flags` must currently be zero. !! 4834 Note, invoking this ioctl with a vCPU is running is inherently racy. However,
>> 4835 KVM does guarantee that vCPUs will see either the previous filter or the new
>> 4836 filter, e.g. MSRs with identical settings in both the old and new filter will
>> 4837 have deterministic behavior.
6444 4838
6445 4839
6446 5. The kvm_run structure 4840 5. The kvm_run structure
6447 ======================== 4841 ========================
6448 4842
6449 Application code obtains a pointer to the kvm 4843 Application code obtains a pointer to the kvm_run structure by
6450 mmap()ing a vcpu fd. From that point, applic 4844 mmap()ing a vcpu fd. From that point, application code can control
6451 execution by changing fields in kvm_run prior 4845 execution by changing fields in kvm_run prior to calling the KVM_RUN
6452 ioctl, and obtain information about the reaso 4846 ioctl, and obtain information about the reason KVM_RUN returned by
6453 looking up structure members. 4847 looking up structure members.
6454 4848
6455 :: 4849 ::
6456 4850
6457 struct kvm_run { 4851 struct kvm_run {
6458 /* in */ 4852 /* in */
6459 __u8 request_interrupt_window; 4853 __u8 request_interrupt_window;
6460 4854
6461 Request that KVM_RUN return when it becomes p 4855 Request that KVM_RUN return when it becomes possible to inject external
6462 interrupts into the guest. Useful in conjunc 4856 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
6463 4857
6464 :: 4858 ::
6465 4859
6466 __u8 immediate_exit; 4860 __u8 immediate_exit;
6467 4861
6468 This field is polled once when KVM_RUN starts 4862 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
6469 exits immediately, returning -EINTR. In the 4863 exits immediately, returning -EINTR. In the common scenario where a
6470 signal is used to "kick" a VCPU out of KVM_RU 4864 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
6471 to avoid usage of KVM_SET_SIGNAL_MASK, which 4865 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
6472 Rather than blocking the signal outside KVM_R 4866 Rather than blocking the signal outside KVM_RUN, userspace can set up
6473 a signal handler that sets run->immediate_exi 4867 a signal handler that sets run->immediate_exit to a non-zero value.
6474 4868
6475 This field is ignored if KVM_CAP_IMMEDIATE_EX 4869 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
6476 4870
6477 :: 4871 ::
6478 4872
6479 __u8 padding1[6]; 4873 __u8 padding1[6];
6480 4874
6481 /* out */ 4875 /* out */
6482 __u32 exit_reason; 4876 __u32 exit_reason;
6483 4877
6484 When KVM_RUN has returned successfully (retur 4878 When KVM_RUN has returned successfully (return value 0), this informs
6485 application code why KVM_RUN has returned. A 4879 application code why KVM_RUN has returned. Allowable values for this
6486 field are detailed below. 4880 field are detailed below.
6487 4881
6488 :: 4882 ::
6489 4883
6490 __u8 ready_for_interrupt_injection; 4884 __u8 ready_for_interrupt_injection;
6491 4885
6492 If request_interrupt_window has been specifie 4886 If request_interrupt_window has been specified, this field indicates
6493 an interrupt can be injected now with KVM_INT 4887 an interrupt can be injected now with KVM_INTERRUPT.
6494 4888
6495 :: 4889 ::
6496 4890
6497 __u8 if_flag; 4891 __u8 if_flag;
6498 4892
6499 The value of the current interrupt flag. Onl 4893 The value of the current interrupt flag. Only valid if in-kernel
6500 local APIC is not used. 4894 local APIC is not used.
6501 4895
6502 :: 4896 ::
6503 4897
6504 __u16 flags; 4898 __u16 flags;
6505 4899
6506 More architecture-specific flags detailing st 4900 More architecture-specific flags detailing state of the VCPU that may
6507 affect the device's behavior. Current defined !! 4901 affect the device's behavior. The only currently defined flag is
6508 !! 4902 KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
6509 /* x86, set if the VCPU is in system manage !! 4903 VCPU is in system management mode.
6510 #define KVM_RUN_X86_SMM (1 << 0) <<
6511 /* x86, set if bus lock detected in VM */ <<
6512 #define KVM_RUN_X86_BUS_LOCK (1 << 1) <<
6513 /* x86, set if the VCPU is executing a nest <<
6514 #define KVM_RUN_X86_GUEST_MODE (1 << 2) <<
6515 <<
6516 /* arm64, set for KVM_EXIT_DEBUG */ <<
6517 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 < <<
6518 4904
6519 :: 4905 ::
6520 4906
6521 /* in (pre_kvm_run), out (post_kvm_ru 4907 /* in (pre_kvm_run), out (post_kvm_run) */
6522 __u64 cr8; 4908 __u64 cr8;
6523 4909
6524 The value of the cr8 register. Only valid if 4910 The value of the cr8 register. Only valid if in-kernel local APIC is
6525 not used. Both input and output. 4911 not used. Both input and output.
6526 4912
6527 :: 4913 ::
6528 4914
6529 __u64 apic_base; 4915 __u64 apic_base;
6530 4916
6531 The value of the APIC BASE msr. Only valid i 4917 The value of the APIC BASE msr. Only valid if in-kernel local
6532 APIC is not used. Both input and output. 4918 APIC is not used. Both input and output.
6533 4919
6534 :: 4920 ::
6535 4921
6536 union { 4922 union {
6537 /* KVM_EXIT_UNKNOWN */ 4923 /* KVM_EXIT_UNKNOWN */
6538 struct { 4924 struct {
6539 __u64 hardware_exit_r 4925 __u64 hardware_exit_reason;
6540 } hw; 4926 } hw;
6541 4927
6542 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu 4928 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
6543 reasons. Further architecture-specific infor 4929 reasons. Further architecture-specific information is available in
6544 hardware_exit_reason. 4930 hardware_exit_reason.
6545 4931
6546 :: 4932 ::
6547 4933
6548 /* KVM_EXIT_FAIL_ENTRY */ 4934 /* KVM_EXIT_FAIL_ENTRY */
6549 struct { 4935 struct {
6550 __u64 hardware_entry_ 4936 __u64 hardware_entry_failure_reason;
6551 __u32 cpu; /* if KVM_ 4937 __u32 cpu; /* if KVM_LAST_CPU */
6552 } fail_entry; 4938 } fail_entry;
6553 4939
6554 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vc 4940 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
6555 to unknown reasons. Further architecture-spe 4941 to unknown reasons. Further architecture-specific information is
6556 available in hardware_entry_failure_reason. 4942 available in hardware_entry_failure_reason.
6557 4943
6558 :: 4944 ::
6559 4945
6560 /* KVM_EXIT_EXCEPTION */ 4946 /* KVM_EXIT_EXCEPTION */
6561 struct { 4947 struct {
6562 __u32 exception; 4948 __u32 exception;
6563 __u32 error_code; 4949 __u32 error_code;
6564 } ex; 4950 } ex;
6565 4951
6566 Unused. 4952 Unused.
6567 4953
6568 :: 4954 ::
6569 4955
6570 /* KVM_EXIT_IO */ 4956 /* KVM_EXIT_IO */
6571 struct { 4957 struct {
6572 #define KVM_EXIT_IO_IN 0 4958 #define KVM_EXIT_IO_IN 0
6573 #define KVM_EXIT_IO_OUT 1 4959 #define KVM_EXIT_IO_OUT 1
6574 __u8 direction; 4960 __u8 direction;
6575 __u8 size; /* bytes * 4961 __u8 size; /* bytes */
6576 __u16 port; 4962 __u16 port;
6577 __u32 count; 4963 __u32 count;
6578 __u64 data_offset; /* 4964 __u64 data_offset; /* relative to kvm_run start */
6579 } io; 4965 } io;
6580 4966
6581 If exit_reason is KVM_EXIT_IO, then the vcpu 4967 If exit_reason is KVM_EXIT_IO, then the vcpu has
6582 executed a port I/O instruction which could n 4968 executed a port I/O instruction which could not be satisfied by kvm.
6583 data_offset describes where the data is locat 4969 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
6584 where kvm expects application code to place t 4970 where kvm expects application code to place the data for the next
6585 KVM_RUN invocation (KVM_EXIT_IO_IN). Data fo 4971 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
6586 4972
6587 :: 4973 ::
6588 4974
6589 /* KVM_EXIT_DEBUG */ 4975 /* KVM_EXIT_DEBUG */
6590 struct { 4976 struct {
6591 struct kvm_debug_exit 4977 struct kvm_debug_exit_arch arch;
6592 } debug; 4978 } debug;
6593 4979
6594 If the exit_reason is KVM_EXIT_DEBUG, then a 4980 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
6595 for which architecture specific information i 4981 for which architecture specific information is returned.
6596 4982
6597 :: 4983 ::
6598 4984
6599 /* KVM_EXIT_MMIO */ 4985 /* KVM_EXIT_MMIO */
6600 struct { 4986 struct {
6601 __u64 phys_addr; 4987 __u64 phys_addr;
6602 __u8 data[8]; 4988 __u8 data[8];
6603 __u32 len; 4989 __u32 len;
6604 __u8 is_write; 4990 __u8 is_write;
6605 } mmio; 4991 } mmio;
6606 4992
6607 If exit_reason is KVM_EXIT_MMIO, then the vcp 4993 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
6608 executed a memory-mapped I/O instruction whic 4994 executed a memory-mapped I/O instruction which could not be satisfied
6609 by kvm. The 'data' member contains the writt 4995 by kvm. The 'data' member contains the written data if 'is_write' is
6610 true, and should be filled by application cod 4996 true, and should be filled by application code otherwise.
6611 4997
6612 The 'data' member contains, in its first 'len 4998 The 'data' member contains, in its first 'len' bytes, the value as it would
6613 appear if the VCPU performed a load or store 4999 appear if the VCPU performed a load or store of the appropriate width directly
6614 to the byte array. 5000 to the byte array.
6615 5001
6616 .. note:: 5002 .. note::
6617 5003
6618 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXI !! 5004 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR,
6619 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KV 5005 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
6620 operations are complete (and guest stat 5006 operations are complete (and guest state is consistent) only after userspace
6621 has re-entered the kernel with KVM_RUN. 5007 has re-entered the kernel with KVM_RUN. The kernel side will first finish
6622 incomplete operations and then check fo !! 5008 incomplete operations and then check for pending signals. Userspace
6623 !! 5009 can re-enter the guest with an unmasked signal pending to complete
6624 The pending state of the operation is n !! 5010 pending operations.
6625 visible to userspace, thus userspace sh <<
6626 completed before performing a live migr <<
6627 guest with an unmasked signal pending o <<
6628 to complete pending operations without <<
6629 to be executed. <<
6630 5011
6631 :: 5012 ::
6632 5013
6633 /* KVM_EXIT_HYPERCALL */ 5014 /* KVM_EXIT_HYPERCALL */
6634 struct { 5015 struct {
6635 __u64 nr; 5016 __u64 nr;
6636 __u64 args[6]; 5017 __u64 args[6];
6637 __u64 ret; 5018 __u64 ret;
6638 __u64 flags; !! 5019 __u32 longmode;
>> 5020 __u32 pad;
6639 } hypercall; 5021 } hypercall;
6640 5022
6641 !! 5023 Unused. This was once used for 'hypercall to userspace'. To implement
6642 It is strongly recommended that userspace use !! 5024 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
6643 ``KVM_EXIT_MMIO`` (all except s390) to implem <<
6644 requires a guest to interact with host usersp <<
6645 5025
6646 .. note:: KVM_EXIT_IO is significantly faster 5026 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
6647 5027
6648 For arm64: <<
6649 ---------- <<
6650 <<
6651 SMCCC exits can be enabled depending on the c <<
6652 filter. See the Documentation/virt/kvm/device <<
6653 ``KVM_ARM_SMCCC_FILTER`` for more details. <<
6654 <<
6655 ``nr`` contains the function ID of the guest' <<
6656 expected to use the ``KVM_GET_ONE_REG`` ioctl <<
6657 parameters from the vCPU's GPRs. <<
6658 <<
6659 Definition of ``flags``: <<
6660 - ``KVM_HYPERCALL_EXIT_SMC``: Indicates that <<
6661 conduit to initiate the SMCCC call. If thi <<
6662 used the HVC conduit for the SMCCC call. <<
6663 <<
6664 - ``KVM_HYPERCALL_EXIT_16BIT``: Indicates th <<
6665 instruction to initiate the SMCCC call. If <<
6666 guest used a 32bit instruction. An AArch64 <<
6667 bit set to 0. <<
6668 <<
6669 At the point of exit, PC points to the instru <<
6670 the trapping instruction. <<
6671 <<
6672 :: 5028 ::
6673 5029
6674 /* KVM_EXIT_TPR_ACCESS */ 5030 /* KVM_EXIT_TPR_ACCESS */
6675 struct { 5031 struct {
6676 __u64 rip; 5032 __u64 rip;
6677 __u32 is_write; 5033 __u32 is_write;
6678 __u32 pad; 5034 __u32 pad;
6679 } tpr_access; 5035 } tpr_access;
6680 5036
6681 To be documented (KVM_TPR_ACCESS_REPORTING). 5037 To be documented (KVM_TPR_ACCESS_REPORTING).
6682 5038
6683 :: 5039 ::
6684 5040
6685 /* KVM_EXIT_S390_SIEIC */ 5041 /* KVM_EXIT_S390_SIEIC */
6686 struct { 5042 struct {
6687 __u8 icptcode; 5043 __u8 icptcode;
6688 __u64 mask; /* psw up 5044 __u64 mask; /* psw upper half */
6689 __u64 addr; /* psw lo 5045 __u64 addr; /* psw lower half */
6690 __u16 ipa; 5046 __u16 ipa;
6691 __u32 ipb; 5047 __u32 ipb;
6692 } s390_sieic; 5048 } s390_sieic;
6693 5049
6694 s390 specific. 5050 s390 specific.
6695 5051
6696 :: 5052 ::
6697 5053
6698 /* KVM_EXIT_S390_RESET */ 5054 /* KVM_EXIT_S390_RESET */
6699 #define KVM_S390_RESET_POR 1 5055 #define KVM_S390_RESET_POR 1
6700 #define KVM_S390_RESET_CLEAR 2 5056 #define KVM_S390_RESET_CLEAR 2
6701 #define KVM_S390_RESET_SUBSYSTEM 4 5057 #define KVM_S390_RESET_SUBSYSTEM 4
6702 #define KVM_S390_RESET_CPU_INIT 8 5058 #define KVM_S390_RESET_CPU_INIT 8
6703 #define KVM_S390_RESET_IPL 16 5059 #define KVM_S390_RESET_IPL 16
6704 __u64 s390_reset_flags; 5060 __u64 s390_reset_flags;
6705 5061
6706 s390 specific. 5062 s390 specific.
6707 5063
6708 :: 5064 ::
6709 5065
6710 /* KVM_EXIT_S390_UCONTROL */ 5066 /* KVM_EXIT_S390_UCONTROL */
6711 struct { 5067 struct {
6712 __u64 trans_exc_code; 5068 __u64 trans_exc_code;
6713 __u32 pgm_code; 5069 __u32 pgm_code;
6714 } s390_ucontrol; 5070 } s390_ucontrol;
6715 5071
6716 s390 specific. A page fault has occurred for 5072 s390 specific. A page fault has occurred for a user controlled virtual
6717 machine (KVM_VM_S390_UNCONTROL) on its host p !! 5073 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
6718 resolved by the kernel. 5074 resolved by the kernel.
6719 The program code and the translation exceptio 5075 The program code and the translation exception code that were placed
6720 in the cpu's lowcore are presented here as de 5076 in the cpu's lowcore are presented here as defined by the z Architecture
6721 Principles of Operation Book in the Chapter f 5077 Principles of Operation Book in the Chapter for Dynamic Address Translation
6722 (DAT) 5078 (DAT)
6723 5079
6724 :: 5080 ::
6725 5081
6726 /* KVM_EXIT_DCR */ 5082 /* KVM_EXIT_DCR */
6727 struct { 5083 struct {
6728 __u32 dcrn; 5084 __u32 dcrn;
6729 __u32 data; 5085 __u32 data;
6730 __u8 is_write; 5086 __u8 is_write;
6731 } dcr; 5087 } dcr;
6732 5088
6733 Deprecated - was used for 440 KVM. 5089 Deprecated - was used for 440 KVM.
6734 5090
6735 :: 5091 ::
6736 5092
6737 /* KVM_EXIT_OSI */ 5093 /* KVM_EXIT_OSI */
6738 struct { 5094 struct {
6739 __u64 gprs[32]; 5095 __u64 gprs[32];
6740 } osi; 5096 } osi;
6741 5097
6742 MOL uses a special hypercall interface it cal 5098 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
6743 hypercalls and exit with this exit struct tha 5099 hypercalls and exit with this exit struct that contains all the guest gprs.
6744 5100
6745 If exit_reason is KVM_EXIT_OSI, then the vcpu 5101 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
6746 Userspace can now handle the hypercall and wh 5102 Userspace can now handle the hypercall and when it's done modify the gprs as
6747 necessary. Upon guest entry all guest GPRs wi 5103 necessary. Upon guest entry all guest GPRs will then be replaced by the values
6748 in this struct. 5104 in this struct.
6749 5105
6750 :: 5106 ::
6751 5107
6752 /* KVM_EXIT_PAPR_HCALL */ 5108 /* KVM_EXIT_PAPR_HCALL */
6753 struct { 5109 struct {
6754 __u64 nr; 5110 __u64 nr;
6755 __u64 ret; 5111 __u64 ret;
6756 __u64 args[9]; 5112 __u64 args[9];
6757 } papr_hcall; 5113 } papr_hcall;
6758 5114
6759 This is used on 64-bit PowerPC when emulating 5115 This is used on 64-bit PowerPC when emulating a pSeries partition,
6760 e.g. with the 'pseries' machine type in qemu. 5116 e.g. with the 'pseries' machine type in qemu. It occurs when the
6761 guest does a hypercall using the 'sc 1' instr 5117 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
6762 contains the hypercall number (from the guest 5118 contains the hypercall number (from the guest R3), and 'args' contains
6763 the arguments (from the guest R4 - R12). Use 5119 the arguments (from the guest R4 - R12). Userspace should put the
6764 return code in 'ret' and any extra returned v 5120 return code in 'ret' and any extra returned values in args[].
6765 The possible hypercalls are defined in the Po 5121 The possible hypercalls are defined in the Power Architecture Platform
6766 Requirements (PAPR) document available from w 5122 Requirements (PAPR) document available from www.power.org (free
6767 developer registration required to access it) 5123 developer registration required to access it).
6768 5124
6769 :: 5125 ::
6770 5126
6771 /* KVM_EXIT_S390_TSCH */ 5127 /* KVM_EXIT_S390_TSCH */
6772 struct { 5128 struct {
6773 __u16 subchannel_id; 5129 __u16 subchannel_id;
6774 __u16 subchannel_nr; 5130 __u16 subchannel_nr;
6775 __u32 io_int_parm; 5131 __u32 io_int_parm;
6776 __u32 io_int_word; 5132 __u32 io_int_word;
6777 __u32 ipb; 5133 __u32 ipb;
6778 __u8 dequeued; 5134 __u8 dequeued;
6779 } s390_tsch; 5135 } s390_tsch;
6780 5136
6781 s390 specific. This exit occurs when KVM_CAP_ 5137 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
6782 and TEST SUBCHANNEL was intercepted. If deque 5138 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
6783 interrupt for the target subchannel has been 5139 interrupt for the target subchannel has been dequeued and subchannel_id,
6784 subchannel_nr, io_int_parm and io_int_word co 5140 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
6785 interrupt. ipb is needed for instruction para 5141 interrupt. ipb is needed for instruction parameter decoding.
6786 5142
6787 :: 5143 ::
6788 5144
6789 /* KVM_EXIT_EPR */ 5145 /* KVM_EXIT_EPR */
6790 struct { 5146 struct {
6791 __u32 epr; 5147 __u32 epr;
6792 } epr; 5148 } epr;
6793 5149
6794 On FSL BookE PowerPC chips, the interrupt con 5150 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
6795 interrupt acknowledge path to the core. When 5151 interrupt acknowledge path to the core. When the core successfully
6796 delivers an interrupt, it automatically popul 5152 delivers an interrupt, it automatically populates the EPR register with
6797 the interrupt vector number and acknowledges 5153 the interrupt vector number and acknowledges the interrupt inside
6798 the interrupt controller. 5154 the interrupt controller.
6799 5155
6800 In case the interrupt controller lives in use 5156 In case the interrupt controller lives in user space, we need to do
6801 the interrupt acknowledge cycle through it to 5157 the interrupt acknowledge cycle through it to fetch the next to be
6802 delivered interrupt vector using this exit. 5158 delivered interrupt vector using this exit.
6803 5159
6804 It gets triggered whenever both KVM_CAP_PPC_E 5160 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
6805 external interrupt has just been delivered in 5161 external interrupt has just been delivered into the guest. User space
6806 should put the acknowledged interrupt vector 5162 should put the acknowledged interrupt vector into the 'epr' field.
6807 5163
6808 :: 5164 ::
6809 5165
6810 /* KVM_EXIT_SYSTEM_EVENT */ 5166 /* KVM_EXIT_SYSTEM_EVENT */
6811 struct { 5167 struct {
6812 #define KVM_SYSTEM_EVENT_SHUTDOWN 1 5168 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
6813 #define KVM_SYSTEM_EVENT_RESET 2 5169 #define KVM_SYSTEM_EVENT_RESET 2
6814 #define KVM_SYSTEM_EVENT_CRASH 3 5170 #define KVM_SYSTEM_EVENT_CRASH 3
6815 #define KVM_SYSTEM_EVENT_WAKEUP 4 <<
6816 #define KVM_SYSTEM_EVENT_SUSPEND 5 <<
6817 #define KVM_SYSTEM_EVENT_SEV_TERM 6 <<
6818 __u32 type; 5171 __u32 type;
6819 __u32 ndata; !! 5172 __u64 flags;
6820 __u64 data[16]; <<
6821 } system_event; 5173 } system_event;
6822 5174
6823 If exit_reason is KVM_EXIT_SYSTEM_EVENT then 5175 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
6824 a system-level event using some architecture 5176 a system-level event using some architecture specific mechanism (hypercall
6825 or some special instruction). In case of ARM6 !! 5177 or some special instruction). In case of ARM/ARM64, this is triggered using
6826 HVC instruction based PSCI call from the vcpu !! 5178 HVC instruction based PSCI call from the vcpu. The 'type' field describes
>> 5179 the system-level event type. The 'flags' field describes architecture
>> 5180 specific flags for the system-level event.
6827 5181
6828 The 'type' field describes the system-level e <<
6829 Valid values for 'type' are: 5182 Valid values for 'type' are:
6830 5183
6831 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has 5184 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
6832 VM. Userspace is not obliged to honour thi 5185 VM. Userspace is not obliged to honour this, and if it does honour
6833 this does not need to destroy the VM synch 5186 this does not need to destroy the VM synchronously (ie it may call
6834 KVM_RUN again before shutdown finally occu 5187 KVM_RUN again before shutdown finally occurs).
6835 - KVM_SYSTEM_EVENT_RESET -- the guest has re 5188 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
6836 As with SHUTDOWN, userspace can choose to 5189 As with SHUTDOWN, userspace can choose to ignore the request, or
6837 to schedule the reset to occur in the futu 5190 to schedule the reset to occur in the future and may call KVM_RUN again.
6838 - KVM_SYSTEM_EVENT_CRASH -- the guest crash 5191 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
6839 has requested a crash condition maintenanc 5192 has requested a crash condition maintenance. Userspace can choose
6840 to ignore the request, or to gather VM mem 5193 to ignore the request, or to gather VM memory core dump and/or
6841 reset/shutdown of the VM. 5194 reset/shutdown of the VM.
6842 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV gu <<
6843 The guest physical address of the guest's <<
6844 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCP <<
6845 KVM has recognized a wakeup event. Userspa <<
6846 marking the exiting vCPU as runnable, or d <<
6847 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has <<
6848 the VM. <<
6849 <<
6850 If KVM_CAP_SYSTEM_EVENT_DATA is present, the <<
6851 architecture specific information for the sys <<
6852 the first `ndata` items (possibly zero) of th <<
6853 <<
6854 - for arm64, data[0] is set to KVM_SYSTEM_EV <<
6855 the guest issued a SYSTEM_RESET2 call acco <<
6856 specification. <<
6857 <<
6858 - for RISC-V, data[0] is set to the value of <<
6859 ``sbi_system_reset`` call. <<
6860 <<
6861 Previous versions of Linux defined a `flags` <<
6862 field is now aliased to `data[0]`. Userspace <<
6863 written if ndata is greater than 0. <<
6864 <<
6865 For arm/arm64: <<
6866 -------------- <<
6867 <<
6868 KVM_SYSTEM_EVENT_SUSPEND exits are enabled wi <<
6869 KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If <<
6870 SYSTEM_SUSPEND function, KVM will exit to use <<
6871 type. <<
6872 <<
6873 It is the sole responsibility of userspace to <<
6874 SYSTEM_SUSPEND call according to ARM DEN0022D <<
6875 KVM does not change the vCPU's state before e <<
6876 the call parameters are left in-place in the <<
6877 <<
6878 Userspace is _required_ to take action for su <<
6879 either: <<
6880 <<
6881 - Honor the guest request to suspend the VM. <<
6882 in-kernel emulation of suspension by setti <<
6883 state to KVM_MP_STATE_SUSPENDED. Userspace <<
6884 state according to the parameters passed t <<
6885 the calling vCPU is resumed. See ARM DEN00 <<
6886 for details on the function parameters. <<
6887 <<
6888 - Deny the guest request to suspend the VM. <<
6889 "Caller responsibilities" for possible ret <<
6890 5195
6891 :: 5196 ::
6892 5197
6893 /* KVM_EXIT_IOAPIC_EOI */ 5198 /* KVM_EXIT_IOAPIC_EOI */
6894 struct { 5199 struct {
6895 __u8 vector; 5200 __u8 vector;
6896 } eoi; 5201 } eoi;
6897 5202
6898 Indicates that the VCPU's in-kernel local API 5203 Indicates that the VCPU's in-kernel local APIC received an EOI for a
6899 level-triggered IOAPIC interrupt. This exit 5204 level-triggered IOAPIC interrupt. This exit only triggers when the
6900 IOAPIC is implemented in userspace (i.e. KVM_ 5205 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
6901 the userspace IOAPIC should process the EOI a 5206 the userspace IOAPIC should process the EOI and retrigger the interrupt if
6902 it is still asserted. Vector is the LAPIC in 5207 it is still asserted. Vector is the LAPIC interrupt vector for which the
6903 EOI was received. 5208 EOI was received.
6904 5209
6905 :: 5210 ::
6906 5211
6907 struct kvm_hyperv_exit { 5212 struct kvm_hyperv_exit {
6908 #define KVM_EXIT_HYPERV_SYNIC 1 5213 #define KVM_EXIT_HYPERV_SYNIC 1
6909 #define KVM_EXIT_HYPERV_HCALL 2 5214 #define KVM_EXIT_HYPERV_HCALL 2
6910 #define KVM_EXIT_HYPERV_SYNDBG 3 5215 #define KVM_EXIT_HYPERV_SYNDBG 3
6911 __u32 type; 5216 __u32 type;
6912 __u32 pad1; 5217 __u32 pad1;
6913 union { 5218 union {
6914 struct { 5219 struct {
6915 __u32 5220 __u32 msr;
6916 __u32 5221 __u32 pad2;
6917 __u64 5222 __u64 control;
6918 __u64 5223 __u64 evt_page;
6919 __u64 5224 __u64 msg_page;
6920 } synic; 5225 } synic;
6921 struct { 5226 struct {
6922 __u64 5227 __u64 input;
6923 __u64 5228 __u64 result;
6924 __u64 5229 __u64 params[2];
6925 } hcall; 5230 } hcall;
6926 struct { 5231 struct {
6927 __u32 5232 __u32 msr;
6928 __u32 5233 __u32 pad2;
6929 __u64 5234 __u64 control;
6930 __u64 5235 __u64 status;
6931 __u64 5236 __u64 send_page;
6932 __u64 5237 __u64 recv_page;
6933 __u64 5238 __u64 pending_page;
6934 } syndbg; 5239 } syndbg;
6935 } u; 5240 } u;
6936 }; 5241 };
6937 /* KVM_EXIT_HYPERV */ 5242 /* KVM_EXIT_HYPERV */
6938 struct kvm_hyperv_exit hyperv 5243 struct kvm_hyperv_exit hyperv;
6939 5244
6940 Indicates that the VCPU exits into userspace 5245 Indicates that the VCPU exits into userspace to process some tasks
6941 related to Hyper-V emulation. 5246 related to Hyper-V emulation.
6942 5247
6943 Valid values for 'type' are: 5248 Valid values for 'type' are:
6944 5249
6945 - KVM_EXIT_HYPERV_SYNIC -- synchronou 5250 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
6946 5251
6947 Hyper-V SynIC state change. Notification is u 5252 Hyper-V SynIC state change. Notification is used to remap SynIC
6948 event/message pages and to enable/disable Syn 5253 event/message pages and to enable/disable SynIC messages/events processing
6949 in userspace. 5254 in userspace.
6950 5255
6951 - KVM_EXIT_HYPERV_SYNDBG -- synchrono 5256 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
6952 5257
6953 Hyper-V Synthetic debugger state change. Noti 5258 Hyper-V Synthetic debugger state change. Notification is used to either update
6954 the pending_page location or to send a contro 5259 the pending_page location or to send a control command (send the buffer located
6955 in send_page or recv a buffer to recv_page). 5260 in send_page or recv a buffer to recv_page).
6956 5261
6957 :: 5262 ::
6958 5263
6959 /* KVM_EXIT_ARM_NISV */ 5264 /* KVM_EXIT_ARM_NISV */
6960 struct { 5265 struct {
6961 __u64 esr_iss; 5266 __u64 esr_iss;
6962 __u64 fault_ipa; 5267 __u64 fault_ipa;
6963 } arm_nisv; 5268 } arm_nisv;
6964 5269
6965 Used on arm64 systems. If a guest accesses me !! 5270 Used on arm and arm64 systems. If a guest accesses memory not in a memslot,
6966 KVM will typically return to userspace and as 5271 KVM will typically return to userspace and ask it to do MMIO emulation on its
6967 behalf. However, for certain classes of instr 5272 behalf. However, for certain classes of instructions, no instruction decode
6968 (direction, length of memory access) is provi 5273 (direction, length of memory access) is provided, and fetching and decoding
6969 the instruction from the VM is overly complic 5274 the instruction from the VM is overly complicated to live in the kernel.
6970 5275
6971 Historically, when this situation occurred, K 5276 Historically, when this situation occurred, KVM would print a warning and kill
6972 the VM. KVM assumed that if the guest accesse 5277 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
6973 trying to do I/O, which just couldn't be emul 5278 trying to do I/O, which just couldn't be emulated, and the warning message was
6974 phrased accordingly. However, what happened m 5279 phrased accordingly. However, what happened more often was that a guest bug
6975 caused access outside the guest memory areas 5280 caused access outside the guest memory areas which should lead to a more
6976 meaningful warning message and an external ab 5281 meaningful warning message and an external abort in the guest, if the access
6977 did not fall within an I/O window. 5282 did not fall within an I/O window.
6978 5283
6979 Userspace implementations can query for KVM_C 5284 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
6980 this capability at VM creation. Once this is 5285 this capability at VM creation. Once this is done, these types of errors will
6981 instead return to userspace with KVM_EXIT_ARM 5286 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
6982 the ESR_EL2 in the esr_iss field, and the fau !! 5287 the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA
6983 Userspace can either fix up the access if it' !! 5288 in the fault_ipa field. Userspace can either fix up the access if it's
6984 decoding the instruction from guest memory (i !! 5289 actually an I/O access by decoding the instruction from guest memory (if it's
6985 executing the guest, or it can decide to susp !! 5290 very brave) and continue executing the guest, or it can decide to suspend,
>> 5291 dump, or restart the guest.
6986 5292
6987 Note that KVM does not skip the faulting inst 5293 Note that KVM does not skip the faulting instruction as it does for
6988 KVM_EXIT_MMIO, but userspace has to emulate a 5294 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
6989 if it decides to decode and emulate the instr 5295 if it decides to decode and emulate the instruction.
6990 5296
6991 This feature isn't available to protected VMs <<
6992 have access to the state that is required to <<
6993 Instead, a data abort exception is directly i <<
6994 Note that although KVM_CAP_ARM_NISV_TO_USER w <<
6995 queried outside of a protected VM context, th <<
6996 exposed if queried on a protected VM file des <<
6997 <<
6998 :: 5297 ::
6999 5298
7000 /* KVM_EXIT_X86_RDMSR / KVM_E 5299 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
7001 struct { 5300 struct {
7002 __u8 error; /* user - 5301 __u8 error; /* user -> kernel */
7003 __u8 pad[7]; 5302 __u8 pad[7];
7004 __u32 reason; /* kern 5303 __u32 reason; /* kernel -> user */
7005 __u32 index; /* kerne 5304 __u32 index; /* kernel -> user */
7006 __u64 data; /* kernel 5305 __u64 data; /* kernel <-> user */
7007 } msr; 5306 } msr;
7008 5307
7009 Used on x86 systems. When the VM capability K 5308 Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
7010 enabled, MSR accesses to registers that would 5309 enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
7011 may instead trigger a KVM_EXIT_X86_RDMSR exit !! 5310 will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
7012 exit for writes. 5311 exit for writes.
7013 5312
7014 The "reason" field specifies why the MSR inte !! 5313 The "reason" field specifies why the MSR trap occurred. User space will only
7015 only receive MSR exits when a particular reas !! 5314 receive MSR exit traps when a particular reason was requested during through
7016 ENABLE_CAP. Currently valid exit reasons are: 5315 ENABLE_CAP. Currently valid exit reasons are:
7017 5316
7018 ============================ ================ !! 5317 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM
7019 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR th !! 5318 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits
7020 KVM_MSR_EXIT_REASON_INVAL access to invali !! 5319 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER
7021 KVM_MSR_EXIT_REASON_FILTER access blocked b <<
7022 ============================ ================ <<
7023 5320
7024 For KVM_EXIT_X86_RDMSR, the "index" field tel !! 5321 For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest
7025 wants to read. To respond to this request wit !! 5322 wants to read. To respond to this request with a successful read, user space
7026 writes the respective data into the "data" fi 5323 writes the respective data into the "data" field and must continue guest
7027 execution to ensure the read data is transfer 5324 execution to ensure the read data is transferred into guest register state.
7028 5325
7029 If the RDMSR request was unsuccessful, usersp !! 5326 If the RDMSR request was unsuccessful, user space indicates that with a "1" in
7030 the "error" field. This will inject a #GP int 5327 the "error" field. This will inject a #GP into the guest when the VCPU is
7031 executed again. 5328 executed again.
7032 5329
7033 For KVM_EXIT_X86_WRMSR, the "index" field tel !! 5330 For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest
7034 wants to write. Once finished processing the !! 5331 wants to write. Once finished processing the event, user space must continue
7035 vCPU execution. If the MSR write was unsucces !! 5332 vCPU execution. If the MSR write was unsuccessful, user space also sets the
7036 "error" field to "1". 5333 "error" field to "1".
7037 5334
7038 See KVM_X86_SET_MSR_FILTER for details on the <<
7039 <<
7040 :: <<
7041 <<
7042 <<
7043 struct kvm_xen_exit { <<
7044 #define KVM_EXIT_XEN_HCALL 1 <<
7045 __u32 type; <<
7046 union { <<
7047 struct { <<
7048 __u32 <<
7049 __u32 <<
7050 __u64 <<
7051 __u64 <<
7052 __u64 <<
7053 } hcall; <<
7054 } u; <<
7055 }; <<
7056 /* KVM_EXIT_XEN */ <<
7057 struct kvm_hyperv_exit xen; <<
7058 <<
7059 Indicates that the VCPU exits into userspace <<
7060 related to Xen emulation. <<
7061 <<
7062 Valid values for 'type' are: <<
7063 <<
7064 - KVM_EXIT_XEN_HCALL -- synchronously notif <<
7065 Userspace is expected to place the hyperc <<
7066 field before invoking KVM_RUN again. <<
7067 <<
7068 :: <<
7069 <<
7070 /* KVM_EXIT_RISCV_SBI */ <<
7071 struct { <<
7072 unsigned long extensi <<
7073 unsigned long functio <<
7074 unsigned long args[6] <<
7075 unsigned long ret[2]; <<
7076 } riscv_sbi; <<
7077 <<
7078 If exit reason is KVM_EXIT_RISCV_SBI then it <<
7079 done a SBI call which is not handled by KVM R <<
7080 of the SBI call are available in 'riscv_sbi' <<
7081 'extension_id' field of 'riscv_sbi' represent <<
7082 'function_id' field represents function ID of <<
7083 array field of 'riscv_sbi' represents paramet <<
7084 array field represents return values. The use <<
7085 values of SBI call before resuming the VCPU. <<
7086 spec refer, https://github.com/riscv/riscv-sb <<
7087 <<
7088 :: <<
7089 <<
7090 /* KVM_EXIT_MEMORY_FAULT */ <<
7091 struct { <<
7092 #define KVM_MEMORY_EXIT_FLAG_PRIVATE (1ULL <<
7093 __u64 flags; <<
7094 __u64 gpa; <<
7095 __u64 size; <<
7096 } memory_fault; <<
7097 <<
7098 KVM_EXIT_MEMORY_FAULT indicates the vCPU has <<
7099 could not be resolved by KVM. The 'gpa' and <<
7100 guest physical address range [gpa, gpa + size <<
7101 describes properties of the faulting access t <<
7102 <<
7103 - KVM_MEMORY_EXIT_FLAG_PRIVATE - When set, i <<
7104 on a private memory access. When clear, i <<
7105 shared access. <<
7106 <<
7107 Note! KVM_EXIT_MEMORY_FAULT is unique among <<
7108 accompanies a return code of '-1', not '0'! <<
7109 or EHWPOISON when KVM exits with KVM_EXIT_MEM <<
7110 kvm_run.exit_reason is stale/undefined for al <<
7111 <<
7112 :: <<
7113 <<
7114 /* KVM_EXIT_NOTIFY */ <<
7115 struct { <<
7116 #define KVM_NOTIFY_CONTEXT_INVALID (1 << <<
7117 __u32 flags; <<
7118 } notify; <<
7119 <<
7120 Used on x86 systems. When the VM capability K <<
7121 enabled, a VM exit generated if no event wind <<
7122 for a specified amount of time. Once KVM_X86_ <<
7123 enabling the cap, it would exit to userspace <<
7124 KVM_EXIT_NOTIFY for further handling. The "fl <<
7125 detailed info. <<
7126 <<
7127 The valid value for 'flags' is: <<
7128 <<
7129 - KVM_NOTIFY_CONTEXT_INVALID -- the VM cont <<
7130 in VMCS. It would run into unknown result <<
7131 <<
7132 :: 5335 ::
7133 5336
7134 /* Fix the size of the union. 5337 /* Fix the size of the union. */
7135 char padding[256]; 5338 char padding[256];
7136 }; 5339 };
7137 5340
7138 /* 5341 /*
7139 * shared registers between kvm and u 5342 * shared registers between kvm and userspace.
7140 * kvm_valid_regs specifies the regis 5343 * kvm_valid_regs specifies the register classes set by the host
7141 * kvm_dirty_regs specified the regis 5344 * kvm_dirty_regs specified the register classes dirtied by userspace
7142 * struct kvm_sync_regs is architectu 5345 * struct kvm_sync_regs is architecture specific, as well as the
7143 * bits for kvm_valid_regs and kvm_di 5346 * bits for kvm_valid_regs and kvm_dirty_regs
7144 */ 5347 */
7145 __u64 kvm_valid_regs; 5348 __u64 kvm_valid_regs;
7146 __u64 kvm_dirty_regs; 5349 __u64 kvm_dirty_regs;
7147 union { 5350 union {
7148 struct kvm_sync_regs regs; 5351 struct kvm_sync_regs regs;
7149 char padding[SYNC_REGS_SIZE_B 5352 char padding[SYNC_REGS_SIZE_BYTES];
7150 } s; 5353 } s;
7151 5354
7152 If KVM_CAP_SYNC_REGS is defined, these fields 5355 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
7153 certain guest registers without having to cal 5356 certain guest registers without having to call SET/GET_*REGS. Thus we can
7154 avoid some system call overhead if userspace 5357 avoid some system call overhead if userspace has to handle the exit.
7155 Userspace can query the validity of the struc 5358 Userspace can query the validity of the structure by checking
7156 kvm_valid_regs for specific bits. These bits 5359 kvm_valid_regs for specific bits. These bits are architecture specific
7157 and usually define the validity of a groups o 5360 and usually define the validity of a groups of registers. (e.g. one bit
7158 for general purpose registers) 5361 for general purpose registers)
7159 5362
7160 Please note that the kernel is allowed to use 5363 Please note that the kernel is allowed to use the kvm_run structure as the
7161 primary storage for certain register types. T 5364 primary storage for certain register types. Therefore, the kernel may use the
7162 values in kvm_run even if the corresponding b 5365 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
7163 5366
>> 5367 ::
>> 5368
>> 5369 };
>> 5370
>> 5371
7164 5372
7165 6. Capabilities that can be enabled on vCPUs 5373 6. Capabilities that can be enabled on vCPUs
7166 ============================================ 5374 ============================================
7167 5375
7168 There are certain capabilities that change th 5376 There are certain capabilities that change the behavior of the virtual CPU or
7169 the virtual machine when enabled. To enable t 5377 the virtual machine when enabled. To enable them, please see section 4.37.
7170 Below you can find a list of capabilities and 5378 Below you can find a list of capabilities and what their effect on the vCPU or
7171 the virtual machine is when enabling them. 5379 the virtual machine is when enabling them.
7172 5380
7173 The following information is provided along w 5381 The following information is provided along with the description:
7174 5382
7175 Architectures: 5383 Architectures:
7176 which instruction set architectures pro 5384 which instruction set architectures provide this ioctl.
7177 x86 includes both i386 and x86_64. 5385 x86 includes both i386 and x86_64.
7178 5386
7179 Target: 5387 Target:
7180 whether this is a per-vcpu or per-vm ca 5388 whether this is a per-vcpu or per-vm capability.
7181 5389
7182 Parameters: 5390 Parameters:
7183 what parameters are accepted by the cap 5391 what parameters are accepted by the capability.
7184 5392
7185 Returns: 5393 Returns:
7186 the return value. General error number 5394 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
7187 are not detailed, but errors with speci 5395 are not detailed, but errors with specific meanings are.
7188 5396
7189 5397
7190 6.1 KVM_CAP_PPC_OSI 5398 6.1 KVM_CAP_PPC_OSI
7191 ------------------- 5399 -------------------
7192 5400
7193 :Architectures: ppc 5401 :Architectures: ppc
7194 :Target: vcpu 5402 :Target: vcpu
7195 :Parameters: none 5403 :Parameters: none
7196 :Returns: 0 on success; -1 on error 5404 :Returns: 0 on success; -1 on error
7197 5405
7198 This capability enables interception of OSI h 5406 This capability enables interception of OSI hypercalls that otherwise would
7199 be treated as normal system calls to be injec 5407 be treated as normal system calls to be injected into the guest. OSI hypercalls
7200 were invented by Mac-on-Linux to have a stand 5408 were invented by Mac-on-Linux to have a standardized communication mechanism
7201 between the guest and the host. 5409 between the guest and the host.
7202 5410
7203 When this capability is enabled, KVM_EXIT_OSI 5411 When this capability is enabled, KVM_EXIT_OSI can occur.
7204 5412
7205 5413
7206 6.2 KVM_CAP_PPC_PAPR 5414 6.2 KVM_CAP_PPC_PAPR
7207 -------------------- 5415 --------------------
7208 5416
7209 :Architectures: ppc 5417 :Architectures: ppc
7210 :Target: vcpu 5418 :Target: vcpu
7211 :Parameters: none 5419 :Parameters: none
7212 :Returns: 0 on success; -1 on error 5420 :Returns: 0 on success; -1 on error
7213 5421
7214 This capability enables interception of PAPR 5422 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
7215 done using the hypercall instruction "sc 1". 5423 done using the hypercall instruction "sc 1".
7216 5424
7217 It also sets the guest privilege level to "su 5425 It also sets the guest privilege level to "supervisor" mode. Usually the guest
7218 runs in "hypervisor" privilege mode with a fe 5426 runs in "hypervisor" privilege mode with a few missing features.
7219 5427
7220 In addition to the above, it changes the sema 5428 In addition to the above, it changes the semantics of SDR1. In this mode, the
7221 HTAB address part of SDR1 contains an HVA ins 5429 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
7222 HTAB invisible to the guest. 5430 HTAB invisible to the guest.
7223 5431
7224 When this capability is enabled, KVM_EXIT_PAP 5432 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
7225 5433
7226 5434
7227 6.3 KVM_CAP_SW_TLB 5435 6.3 KVM_CAP_SW_TLB
7228 ------------------ 5436 ------------------
7229 5437
7230 :Architectures: ppc 5438 :Architectures: ppc
7231 :Target: vcpu 5439 :Target: vcpu
7232 :Parameters: args[0] is the address of a stru 5440 :Parameters: args[0] is the address of a struct kvm_config_tlb
7233 :Returns: 0 on success; -1 on error 5441 :Returns: 0 on success; -1 on error
7234 5442
7235 :: 5443 ::
7236 5444
7237 struct kvm_config_tlb { 5445 struct kvm_config_tlb {
7238 __u64 params; 5446 __u64 params;
7239 __u64 array; 5447 __u64 array;
7240 __u32 mmu_type; 5448 __u32 mmu_type;
7241 __u32 array_len; 5449 __u32 array_len;
7242 }; 5450 };
7243 5451
7244 Configures the virtual CPU's TLB array, estab 5452 Configures the virtual CPU's TLB array, establishing a shared memory area
7245 between userspace and KVM. The "params" and 5453 between userspace and KVM. The "params" and "array" fields are userspace
7246 addresses of mmu-type-specific data structure 5454 addresses of mmu-type-specific data structures. The "array_len" field is an
7247 safety mechanism, and should be set to the si 5455 safety mechanism, and should be set to the size in bytes of the memory that
7248 userspace has reserved for the array. It mus 5456 userspace has reserved for the array. It must be at least the size dictated
7249 by "mmu_type" and "params". 5457 by "mmu_type" and "params".
7250 5458
7251 While KVM_RUN is active, the shared region is 5459 While KVM_RUN is active, the shared region is under control of KVM. Its
7252 contents are undefined, and any modification 5460 contents are undefined, and any modification by userspace results in
7253 boundedly undefined behavior. 5461 boundedly undefined behavior.
7254 5462
7255 On return from KVM_RUN, the shared region wil 5463 On return from KVM_RUN, the shared region will reflect the current state of
7256 the guest's TLB. If userspace makes any chan 5464 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
7257 to tell KVM which entries have been changed, 5465 to tell KVM which entries have been changed, prior to calling KVM_RUN again
7258 on this vcpu. 5466 on this vcpu.
7259 5467
7260 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_ 5468 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
7261 5469
7262 - The "params" field is of type "struct kvm_ 5470 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
7263 - The "array" field points to an array of ty 5471 - The "array" field points to an array of type "struct
7264 kvm_book3e_206_tlb_entry". 5472 kvm_book3e_206_tlb_entry".
7265 - The array consists of all entries in the f 5473 - The array consists of all entries in the first TLB, followed by all
7266 entries in the second TLB. 5474 entries in the second TLB.
7267 - Within a TLB, entries are ordered first by 5475 - Within a TLB, entries are ordered first by increasing set number. Within a
7268 set, entries are ordered by way (increasin 5476 set, entries are ordered by way (increasing ESEL).
7269 - The hash for determining set number in TLB 5477 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
7270 where "num_sets" is the tlb_sizes[] value 5478 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
7271 - The tsize field of mas1 shall be set to 4K 5479 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
7272 hardware ignores this value for TLB0. 5480 hardware ignores this value for TLB0.
7273 5481
7274 6.4 KVM_CAP_S390_CSS_SUPPORT 5482 6.4 KVM_CAP_S390_CSS_SUPPORT
7275 ---------------------------- 5483 ----------------------------
7276 5484
7277 :Architectures: s390 5485 :Architectures: s390
7278 :Target: vcpu 5486 :Target: vcpu
7279 :Parameters: none 5487 :Parameters: none
7280 :Returns: 0 on success; -1 on error 5488 :Returns: 0 on success; -1 on error
7281 5489
7282 This capability enables support for handling 5490 This capability enables support for handling of channel I/O instructions.
7283 5491
7284 TEST PENDING INTERRUPTION and the interrupt p 5492 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
7285 handled in-kernel, while the other I/O instru 5493 handled in-kernel, while the other I/O instructions are passed to userspace.
7286 5494
7287 When this capability is enabled, KVM_EXIT_S39 5495 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
7288 SUBCHANNEL intercepts. 5496 SUBCHANNEL intercepts.
7289 5497
7290 Note that even though this capability is enab 5498 Note that even though this capability is enabled per-vcpu, the complete
7291 virtual machine is affected. 5499 virtual machine is affected.
7292 5500
7293 6.5 KVM_CAP_PPC_EPR 5501 6.5 KVM_CAP_PPC_EPR
7294 ------------------- 5502 -------------------
7295 5503
7296 :Architectures: ppc 5504 :Architectures: ppc
7297 :Target: vcpu 5505 :Target: vcpu
7298 :Parameters: args[0] defines whether the prox 5506 :Parameters: args[0] defines whether the proxy facility is active
7299 :Returns: 0 on success; -1 on error 5507 :Returns: 0 on success; -1 on error
7300 5508
7301 This capability enables or disables the deliv 5509 This capability enables or disables the delivery of interrupts through the
7302 external proxy facility. 5510 external proxy facility.
7303 5511
7304 When enabled (args[0] != 0), every time the g 5512 When enabled (args[0] != 0), every time the guest gets an external interrupt
7305 delivered, it automatically exits into user s 5513 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
7306 to receive the topmost interrupt vector. 5514 to receive the topmost interrupt vector.
7307 5515
7308 When disabled (args[0] == 0), behavior is as 5516 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
7309 5517
7310 When this capability is enabled, KVM_EXIT_EPR 5518 When this capability is enabled, KVM_EXIT_EPR can occur.
7311 5519
7312 6.6 KVM_CAP_IRQ_MPIC 5520 6.6 KVM_CAP_IRQ_MPIC
7313 -------------------- 5521 --------------------
7314 5522
7315 :Architectures: ppc 5523 :Architectures: ppc
7316 :Parameters: args[0] is the MPIC device fd; 5524 :Parameters: args[0] is the MPIC device fd;
7317 args[1] is the MPIC CPU number f 5525 args[1] is the MPIC CPU number for this vcpu
7318 5526
7319 This capability connects the vcpu to an in-ke 5527 This capability connects the vcpu to an in-kernel MPIC device.
7320 5528
7321 6.7 KVM_CAP_IRQ_XICS 5529 6.7 KVM_CAP_IRQ_XICS
7322 -------------------- 5530 --------------------
7323 5531
7324 :Architectures: ppc 5532 :Architectures: ppc
7325 :Target: vcpu 5533 :Target: vcpu
7326 :Parameters: args[0] is the XICS device fd; 5534 :Parameters: args[0] is the XICS device fd;
7327 args[1] is the XICS CPU number ( 5535 args[1] is the XICS CPU number (server ID) for this vcpu
7328 5536
7329 This capability connects the vcpu to an in-ke 5537 This capability connects the vcpu to an in-kernel XICS device.
7330 5538
7331 6.8 KVM_CAP_S390_IRQCHIP 5539 6.8 KVM_CAP_S390_IRQCHIP
7332 ------------------------ 5540 ------------------------
7333 5541
7334 :Architectures: s390 5542 :Architectures: s390
7335 :Target: vm 5543 :Target: vm
7336 :Parameters: none 5544 :Parameters: none
7337 5545
7338 This capability enables the in-kernel irqchip 5546 This capability enables the in-kernel irqchip for s390. Please refer to
7339 "4.24 KVM_CREATE_IRQCHIP" for details. 5547 "4.24 KVM_CREATE_IRQCHIP" for details.
7340 5548
7341 6.9 KVM_CAP_MIPS_FPU 5549 6.9 KVM_CAP_MIPS_FPU
7342 -------------------- 5550 --------------------
7343 5551
7344 :Architectures: mips 5552 :Architectures: mips
7345 :Target: vcpu 5553 :Target: vcpu
7346 :Parameters: args[0] is reserved for future u 5554 :Parameters: args[0] is reserved for future use (should be 0).
7347 5555
7348 This capability allows the use of the host Fl 5556 This capability allows the use of the host Floating Point Unit by the guest. It
7349 allows the Config1.FP bit to be set to enable 5557 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
7350 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG 5558 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
7351 accessed (depending on the current guest FPU 5559 accessed (depending on the current guest FPU register mode), and the Status.FR,
7352 Config5.FRE bits are accessible via the KVM A 5560 Config5.FRE bits are accessible via the KVM API and also from the guest,
7353 depending on them being supported by the FPU. 5561 depending on them being supported by the FPU.
7354 5562
7355 6.10 KVM_CAP_MIPS_MSA 5563 6.10 KVM_CAP_MIPS_MSA
7356 --------------------- 5564 ---------------------
7357 5565
7358 :Architectures: mips 5566 :Architectures: mips
7359 :Target: vcpu 5567 :Target: vcpu
7360 :Parameters: args[0] is reserved for future u 5568 :Parameters: args[0] is reserved for future use (should be 0).
7361 5569
7362 This capability allows the use of the MIPS SI 5570 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
7363 It allows the Config3.MSAP bit to be set to e 5571 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
7364 Once this is done the ``KVM_REG_MIPS_VEC_*`` 5572 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
7365 registers can be accessed, and the Config5.MS 5573 registers can be accessed, and the Config5.MSAEn bit is accessible via the
7366 KVM API and also from the guest. 5574 KVM API and also from the guest.
7367 5575
7368 6.74 KVM_CAP_SYNC_REGS 5576 6.74 KVM_CAP_SYNC_REGS
7369 ---------------------- 5577 ----------------------
7370 5578
7371 :Architectures: s390, x86 5579 :Architectures: s390, x86
7372 :Target: s390: always enabled, x86: vcpu 5580 :Target: s390: always enabled, x86: vcpu
7373 :Parameters: none 5581 :Parameters: none
7374 :Returns: x86: KVM_CHECK_EXTENSION returns a 5582 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
7375 sets are supported 5583 sets are supported
7376 (bitfields defined in arch/x86/incl 5584 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
7377 5585
7378 As described above in the kvm_sync_regs struc 5586 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
7379 KVM_CAP_SYNC_REGS "allow[s] userspace to acce 5587 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
7380 without having to call SET/GET_*REGS". This r 5588 without having to call SET/GET_*REGS". This reduces overhead by eliminating
7381 repeated ioctl calls for setting and/or getti 5589 repeated ioctl calls for setting and/or getting register values. This is
7382 particularly important when userspace is maki 5590 particularly important when userspace is making synchronous guest state
7383 modifications, e.g. when emulating and/or int 5591 modifications, e.g. when emulating and/or intercepting instructions in
7384 userspace. 5592 userspace.
7385 5593
7386 For s390 specifics, please refer to the sourc 5594 For s390 specifics, please refer to the source code.
7387 5595
7388 For x86: 5596 For x86:
7389 5597
7390 - the register sets to be copied out to kvm_r 5598 - the register sets to be copied out to kvm_run are selectable
7391 by userspace (rather that all sets being co 5599 by userspace (rather that all sets being copied out for every exit).
7392 - vcpu_events are available in addition to re 5600 - vcpu_events are available in addition to regs and sregs.
7393 5601
7394 For x86, the 'kvm_valid_regs' field of struct 5602 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
7395 function as an input bit-array field set by u 5603 function as an input bit-array field set by userspace to indicate the
7396 specific register sets to be copied out on th 5604 specific register sets to be copied out on the next exit.
7397 5605
7398 To indicate when userspace has modified value 5606 To indicate when userspace has modified values that should be copied into
7399 the vCPU, the all architecture bitarray field 5607 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
7400 This is done using the same bitflags as for t 5608 This is done using the same bitflags as for the 'kvm_valid_regs' field.
7401 If the dirty bit is not set, then the registe 5609 If the dirty bit is not set, then the register set values will not be copied
7402 into the vCPU even if they've been modified. 5610 into the vCPU even if they've been modified.
7403 5611
7404 Unused bitfields in the bitarrays must be set 5612 Unused bitfields in the bitarrays must be set to zero.
7405 5613
7406 :: 5614 ::
7407 5615
7408 struct kvm_sync_regs { 5616 struct kvm_sync_regs {
7409 struct kvm_regs regs; 5617 struct kvm_regs regs;
7410 struct kvm_sregs sregs; 5618 struct kvm_sregs sregs;
7411 struct kvm_vcpu_events events; 5619 struct kvm_vcpu_events events;
7412 }; 5620 };
7413 5621
7414 6.75 KVM_CAP_PPC_IRQ_XIVE 5622 6.75 KVM_CAP_PPC_IRQ_XIVE
7415 ------------------------- 5623 -------------------------
7416 5624
7417 :Architectures: ppc 5625 :Architectures: ppc
7418 :Target: vcpu 5626 :Target: vcpu
7419 :Parameters: args[0] is the XIVE device fd; 5627 :Parameters: args[0] is the XIVE device fd;
7420 args[1] is the XIVE CPU number ( 5628 args[1] is the XIVE CPU number (server ID) for this vcpu
7421 5629
7422 This capability connects the vcpu to an in-ke 5630 This capability connects the vcpu to an in-kernel XIVE device.
7423 5631
7424 7. Capabilities that can be enabled on VMs 5632 7. Capabilities that can be enabled on VMs
7425 ========================================== 5633 ==========================================
7426 5634
7427 There are certain capabilities that change th 5635 There are certain capabilities that change the behavior of the virtual
7428 machine when enabled. To enable them, please 5636 machine when enabled. To enable them, please see section 4.37. Below
7429 you can find a list of capabilities and what 5637 you can find a list of capabilities and what their effect on the VM
7430 is when enabling them. 5638 is when enabling them.
7431 5639
7432 The following information is provided along w 5640 The following information is provided along with the description:
7433 5641
7434 Architectures: 5642 Architectures:
7435 which instruction set architectures pro 5643 which instruction set architectures provide this ioctl.
7436 x86 includes both i386 and x86_64. 5644 x86 includes both i386 and x86_64.
7437 5645
7438 Parameters: 5646 Parameters:
7439 what parameters are accepted by the cap 5647 what parameters are accepted by the capability.
7440 5648
7441 Returns: 5649 Returns:
7442 the return value. General error number 5650 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
7443 are not detailed, but errors with speci 5651 are not detailed, but errors with specific meanings are.
7444 5652
7445 5653
7446 7.1 KVM_CAP_PPC_ENABLE_HCALL 5654 7.1 KVM_CAP_PPC_ENABLE_HCALL
7447 ---------------------------- 5655 ----------------------------
7448 5656
7449 :Architectures: ppc 5657 :Architectures: ppc
7450 :Parameters: args[0] is the sPAPR hcall numbe 5658 :Parameters: args[0] is the sPAPR hcall number;
7451 args[1] is 0 to disable, 1 to en 5659 args[1] is 0 to disable, 1 to enable in-kernel handling
7452 5660
7453 This capability controls whether individual s 5661 This capability controls whether individual sPAPR hypercalls (hcalls)
7454 get handled by the kernel or not. Enabling o 5662 get handled by the kernel or not. Enabling or disabling in-kernel
7455 handling of an hcall is effective across the 5663 handling of an hcall is effective across the VM. On creation, an
7456 initial set of hcalls are enabled for in-kern 5664 initial set of hcalls are enabled for in-kernel handling, which
7457 consists of those hcalls for which in-kernel 5665 consists of those hcalls for which in-kernel handlers were implemented
7458 before this capability was implemented. If d 5666 before this capability was implemented. If disabled, the kernel will
7459 not to attempt to handle the hcall, but will 5667 not to attempt to handle the hcall, but will always exit to userspace
7460 to handle it. Note that it may not make sens 5668 to handle it. Note that it may not make sense to enable some and
7461 disable others of a group of related hcalls, 5669 disable others of a group of related hcalls, but KVM does not prevent
7462 userspace from doing that. 5670 userspace from doing that.
7463 5671
7464 If the hcall number specified is not one that 5672 If the hcall number specified is not one that has an in-kernel
7465 implementation, the KVM_ENABLE_CAP ioctl will 5673 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
7466 error. 5674 error.
7467 5675
7468 7.2 KVM_CAP_S390_USER_SIGP 5676 7.2 KVM_CAP_S390_USER_SIGP
7469 -------------------------- 5677 --------------------------
7470 5678
7471 :Architectures: s390 5679 :Architectures: s390
7472 :Parameters: none 5680 :Parameters: none
7473 5681
7474 This capability controls which SIGP orders wi 5682 This capability controls which SIGP orders will be handled completely in user
7475 space. With this capability enabled, all fast 5683 space. With this capability enabled, all fast orders will be handled completely
7476 in the kernel: 5684 in the kernel:
7477 5685
7478 - SENSE 5686 - SENSE
7479 - SENSE RUNNING 5687 - SENSE RUNNING
7480 - EXTERNAL CALL 5688 - EXTERNAL CALL
7481 - EMERGENCY SIGNAL 5689 - EMERGENCY SIGNAL
7482 - CONDITIONAL EMERGENCY SIGNAL 5690 - CONDITIONAL EMERGENCY SIGNAL
7483 5691
7484 All other orders will be handled completely i 5692 All other orders will be handled completely in user space.
7485 5693
7486 Only privileged operation exceptions will be 5694 Only privileged operation exceptions will be checked for in the kernel (or even
7487 in the hardware prior to interception). If th 5695 in the hardware prior to interception). If this capability is not enabled, the
7488 old way of handling SIGP orders is used (part 5696 old way of handling SIGP orders is used (partially in kernel and user space).
7489 5697
7490 7.3 KVM_CAP_S390_VECTOR_REGISTERS 5698 7.3 KVM_CAP_S390_VECTOR_REGISTERS
7491 --------------------------------- 5699 ---------------------------------
7492 5700
7493 :Architectures: s390 5701 :Architectures: s390
7494 :Parameters: none 5702 :Parameters: none
7495 :Returns: 0 on success, negative value on err 5703 :Returns: 0 on success, negative value on error
7496 5704
7497 Allows use of the vector registers introduced 5705 Allows use of the vector registers introduced with z13 processor, and
7498 provides for the synchronization between host 5706 provides for the synchronization between host and user space. Will
7499 return -EINVAL if the machine does not suppor 5707 return -EINVAL if the machine does not support vectors.
7500 5708
7501 7.4 KVM_CAP_S390_USER_STSI 5709 7.4 KVM_CAP_S390_USER_STSI
7502 -------------------------- 5710 --------------------------
7503 5711
7504 :Architectures: s390 5712 :Architectures: s390
7505 :Parameters: none 5713 :Parameters: none
7506 5714
7507 This capability allows post-handlers for the 5715 This capability allows post-handlers for the STSI instruction. After
7508 initial handling in the kernel, KVM exits to 5716 initial handling in the kernel, KVM exits to user space with
7509 KVM_EXIT_S390_STSI to allow user space to ins 5717 KVM_EXIT_S390_STSI to allow user space to insert further data.
7510 5718
7511 Before exiting to userspace, kvm handlers sho 5719 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
7512 vcpu->run:: 5720 vcpu->run::
7513 5721
7514 struct { 5722 struct {
7515 __u64 addr; 5723 __u64 addr;
7516 __u8 ar; 5724 __u8 ar;
7517 __u8 reserved; 5725 __u8 reserved;
7518 __u8 fc; 5726 __u8 fc;
7519 __u8 sel1; 5727 __u8 sel1;
7520 __u16 sel2; 5728 __u16 sel2;
7521 } s390_stsi; 5729 } s390_stsi;
7522 5730
7523 @addr - guest address of STSI SYSIB 5731 @addr - guest address of STSI SYSIB
7524 @fc - function code 5732 @fc - function code
7525 @sel1 - selector 1 5733 @sel1 - selector 1
7526 @sel2 - selector 2 5734 @sel2 - selector 2
7527 @ar - access register number 5735 @ar - access register number
7528 5736
7529 KVM handlers should exit to userspace with rc 5737 KVM handlers should exit to userspace with rc = -EREMOTE.
7530 5738
7531 7.5 KVM_CAP_SPLIT_IRQCHIP 5739 7.5 KVM_CAP_SPLIT_IRQCHIP
7532 ------------------------- 5740 -------------------------
7533 5741
7534 :Architectures: x86 5742 :Architectures: x86
7535 :Parameters: args[0] - number of routes reser 5743 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
7536 :Returns: 0 on success, -1 on error 5744 :Returns: 0 on success, -1 on error
7537 5745
7538 Create a local apic for each processor in the 5746 Create a local apic for each processor in the kernel. This can be used
7539 instead of KVM_CREATE_IRQCHIP if the userspac 5747 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
7540 IOAPIC and PIC (and also the PIT, even though 5748 IOAPIC and PIC (and also the PIT, even though this has to be enabled
7541 separately). 5749 separately).
7542 5750
7543 This capability also enables in kernel routin 5751 This capability also enables in kernel routing of interrupt requests;
7544 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM 5752 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
7545 used in the IRQ routing table. The first arg 5753 used in the IRQ routing table. The first args[0] MSI routes are reserved
7546 for the IOAPIC pins. Whenever the LAPIC rece 5754 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
7547 a KVM_EXIT_IOAPIC_EOI vmexit will be reported 5755 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
7548 5756
7549 Fails if VCPU has already been created, or if 5757 Fails if VCPU has already been created, or if the irqchip is already in the
7550 kernel (i.e. KVM_CREATE_IRQCHIP has already b 5758 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
7551 5759
7552 7.6 KVM_CAP_S390_RI 5760 7.6 KVM_CAP_S390_RI
7553 ------------------- 5761 -------------------
7554 5762
7555 :Architectures: s390 5763 :Architectures: s390
7556 :Parameters: none 5764 :Parameters: none
7557 5765
7558 Allows use of runtime-instrumentation introdu 5766 Allows use of runtime-instrumentation introduced with zEC12 processor.
7559 Will return -EINVAL if the machine does not s 5767 Will return -EINVAL if the machine does not support runtime-instrumentation.
7560 Will return -EBUSY if a VCPU has already been 5768 Will return -EBUSY if a VCPU has already been created.
7561 5769
7562 7.7 KVM_CAP_X2APIC_API 5770 7.7 KVM_CAP_X2APIC_API
7563 ---------------------- 5771 ----------------------
7564 5772
7565 :Architectures: x86 5773 :Architectures: x86
7566 :Parameters: args[0] - features that should b 5774 :Parameters: args[0] - features that should be enabled
7567 :Returns: 0 on success, -EINVAL when args[0] 5775 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
7568 5776
7569 Valid feature flags in args[0] are:: 5777 Valid feature flags in args[0] are::
7570 5778
7571 #define KVM_X2APIC_API_USE_32BIT_IDS 5779 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
7572 #define KVM_X2APIC_API_DISABLE_BROADCAST_QU 5780 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
7573 5781
7574 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes 5782 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
7575 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_ 5783 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
7576 allowing the use of 32-bit APIC IDs. See KVM 5784 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
7577 respective sections. 5785 respective sections.
7578 5786
7579 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must b 5787 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
7580 in logical mode or with more than 255 VCPUs. 5788 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
7581 as a broadcast even in x2APIC mode in order t 5789 as a broadcast even in x2APIC mode in order to support physical x2APIC
7582 without interrupt remapping. This is undesir 5790 without interrupt remapping. This is undesirable in logical mode,
7583 where 0xff represents CPUs 0-7 in cluster 0. 5791 where 0xff represents CPUs 0-7 in cluster 0.
7584 5792
7585 7.8 KVM_CAP_S390_USER_INSTR0 5793 7.8 KVM_CAP_S390_USER_INSTR0
7586 ---------------------------- 5794 ----------------------------
7587 5795
7588 :Architectures: s390 5796 :Architectures: s390
7589 :Parameters: none 5797 :Parameters: none
7590 5798
7591 With this capability enabled, all illegal ins 5799 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
7592 be intercepted and forwarded to user space. U 5800 be intercepted and forwarded to user space. User space can use this
7593 mechanism e.g. to realize 2-byte software bre 5801 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
7594 not inject an operating exception for these i 5802 not inject an operating exception for these instructions, user space has
7595 to take care of that. 5803 to take care of that.
7596 5804
7597 This capability can be enabled dynamically ev 5805 This capability can be enabled dynamically even if VCPUs were already
7598 created and are running. 5806 created and are running.
7599 5807
7600 7.9 KVM_CAP_S390_GS 5808 7.9 KVM_CAP_S390_GS
7601 ------------------- 5809 -------------------
7602 5810
7603 :Architectures: s390 5811 :Architectures: s390
7604 :Parameters: none 5812 :Parameters: none
7605 :Returns: 0 on success; -EINVAL if the machin 5813 :Returns: 0 on success; -EINVAL if the machine does not support
7606 guarded storage; -EBUSY if a VCPU h 5814 guarded storage; -EBUSY if a VCPU has already been created.
7607 5815
7608 Allows use of guarded storage for the KVM gue 5816 Allows use of guarded storage for the KVM guest.
7609 5817
7610 7.10 KVM_CAP_S390_AIS 5818 7.10 KVM_CAP_S390_AIS
7611 --------------------- 5819 ---------------------
7612 5820
7613 :Architectures: s390 5821 :Architectures: s390
7614 :Parameters: none 5822 :Parameters: none
7615 5823
7616 Allow use of adapter-interruption suppression 5824 Allow use of adapter-interruption suppression.
7617 :Returns: 0 on success; -EBUSY if a VCPU has 5825 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
7618 5826
7619 7.11 KVM_CAP_PPC_SMT 5827 7.11 KVM_CAP_PPC_SMT
7620 -------------------- 5828 --------------------
7621 5829
7622 :Architectures: ppc 5830 :Architectures: ppc
7623 :Parameters: vsmt_mode, flags 5831 :Parameters: vsmt_mode, flags
7624 5832
7625 Enabling this capability on a VM provides use 5833 Enabling this capability on a VM provides userspace with a way to set
7626 the desired virtual SMT mode (i.e. the number 5834 the desired virtual SMT mode (i.e. the number of virtual CPUs per
7627 virtual core). The virtual SMT mode, vsmt_mo 5835 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
7628 between 1 and 8. On POWER8, vsmt_mode must a 5836 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
7629 the number of threads per subcore for the hos 5837 the number of threads per subcore for the host. Currently flags must
7630 be 0. A successful call to enable this capab 5838 be 0. A successful call to enable this capability will result in
7631 vsmt_mode being returned when the KVM_CAP_PPC 5839 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
7632 subsequently queried for the VM. This capabi 5840 subsequently queried for the VM. This capability is only supported by
7633 HV KVM, and can only be set before any VCPUs 5841 HV KVM, and can only be set before any VCPUs have been created.
7634 The KVM_CAP_PPC_SMT_POSSIBLE capability indic 5842 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
7635 modes are available. 5843 modes are available.
7636 5844
7637 7.12 KVM_CAP_PPC_FWNMI 5845 7.12 KVM_CAP_PPC_FWNMI
7638 ---------------------- 5846 ----------------------
7639 5847
7640 :Architectures: ppc 5848 :Architectures: ppc
7641 :Parameters: none 5849 :Parameters: none
7642 5850
7643 With this capability a machine check exceptio 5851 With this capability a machine check exception in the guest address
7644 space will cause KVM to exit the guest with N 5852 space will cause KVM to exit the guest with NMI exit reason. This
7645 enables QEMU to build error log and branch to 5853 enables QEMU to build error log and branch to guest kernel registered
7646 machine check handling routine. Without this 5854 machine check handling routine. Without this capability KVM will
7647 branch to guests' 0x200 interrupt vector. 5855 branch to guests' 0x200 interrupt vector.
7648 5856
7649 7.13 KVM_CAP_X86_DISABLE_EXITS 5857 7.13 KVM_CAP_X86_DISABLE_EXITS
7650 ------------------------------ 5858 ------------------------------
7651 5859
7652 :Architectures: x86 5860 :Architectures: x86
7653 :Parameters: args[0] defines which exits are 5861 :Parameters: args[0] defines which exits are disabled
7654 :Returns: 0 on success, -EINVAL when args[0] 5862 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
7655 5863
7656 Valid bits in args[0] are:: 5864 Valid bits in args[0] are::
7657 5865
7658 #define KVM_X86_DISABLE_EXITS_MWAIT 5866 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
7659 #define KVM_X86_DISABLE_EXITS_HLT 5867 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
7660 #define KVM_X86_DISABLE_EXITS_PAUSE 5868 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
7661 #define KVM_X86_DISABLE_EXITS_CSTATE 5869 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
7662 5870
7663 Enabling this capability on a VM provides use 5871 Enabling this capability on a VM provides userspace with a way to no
7664 longer intercept some instructions for improv 5872 longer intercept some instructions for improved latency in some
7665 workloads, and is suggested when vCPUs are as 5873 workloads, and is suggested when vCPUs are associated to dedicated
7666 physical CPUs. More bits can be added in the 5874 physical CPUs. More bits can be added in the future; userspace can
7667 just pass the KVM_CHECK_EXTENSION result to K 5875 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
7668 all such vmexits. 5876 all such vmexits.
7669 5877
7670 Do not enable KVM_FEATURE_PV_UNHALT if you di 5878 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
7671 5879
7672 7.14 KVM_CAP_S390_HPAGE_1M 5880 7.14 KVM_CAP_S390_HPAGE_1M
7673 -------------------------- 5881 --------------------------
7674 5882
7675 :Architectures: s390 5883 :Architectures: s390
7676 :Parameters: none 5884 :Parameters: none
7677 :Returns: 0 on success, -EINVAL if hpage modu 5885 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
7678 or cmma is enabled, or the VM has t 5886 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
7679 flag set 5887 flag set
7680 5888
7681 With this capability the KVM support for memo 5889 With this capability the KVM support for memory backing with 1m pages
7682 through hugetlbfs can be enabled for a VM. Af 5890 through hugetlbfs can be enabled for a VM. After the capability is
7683 enabled, cmma can't be enabled anymore and pf 5891 enabled, cmma can't be enabled anymore and pfmfi and the storage key
7684 interpretation are disabled. If cmma has alre 5892 interpretation are disabled. If cmma has already been enabled or the
7685 hpage module parameter is not set to 1, -EINV 5893 hpage module parameter is not set to 1, -EINVAL is returned.
7686 5894
7687 While it is generally possible to create a hu 5895 While it is generally possible to create a huge page backed VM without
7688 this capability, the VM will not be able to r 5896 this capability, the VM will not be able to run.
7689 5897
7690 7.15 KVM_CAP_MSR_PLATFORM_INFO 5898 7.15 KVM_CAP_MSR_PLATFORM_INFO
7691 ------------------------------ 5899 ------------------------------
7692 5900
7693 :Architectures: x86 5901 :Architectures: x86
7694 :Parameters: args[0] whether feature should b 5902 :Parameters: args[0] whether feature should be enabled or not
7695 5903
7696 With this capability, a guest may read the MS 5904 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
7697 a #GP would be raised when the guest tries to 5905 a #GP would be raised when the guest tries to access. Currently, this
7698 capability does not enable write permissions 5906 capability does not enable write permissions of this MSR for the guest.
7699 5907
7700 7.16 KVM_CAP_PPC_NESTED_HV 5908 7.16 KVM_CAP_PPC_NESTED_HV
7701 -------------------------- 5909 --------------------------
7702 5910
7703 :Architectures: ppc 5911 :Architectures: ppc
7704 :Parameters: none 5912 :Parameters: none
7705 :Returns: 0 on success, -EINVAL when the impl 5913 :Returns: 0 on success, -EINVAL when the implementation doesn't support
7706 nested-HV virtualization. 5914 nested-HV virtualization.
7707 5915
7708 HV-KVM on POWER9 and later systems allows for 5916 HV-KVM on POWER9 and later systems allows for "nested-HV"
7709 virtualization, which provides a way for a gu 5917 virtualization, which provides a way for a guest VM to run guests that
7710 can run using the CPU's supervisor mode (priv 5918 can run using the CPU's supervisor mode (privileged non-hypervisor
7711 state). Enabling this capability on a VM dep 5919 state). Enabling this capability on a VM depends on the CPU having
7712 the necessary functionality and on the facili 5920 the necessary functionality and on the facility being enabled with a
7713 kvm-hv module parameter. 5921 kvm-hv module parameter.
7714 5922
7715 7.17 KVM_CAP_EXCEPTION_PAYLOAD 5923 7.17 KVM_CAP_EXCEPTION_PAYLOAD
7716 ------------------------------ 5924 ------------------------------
7717 5925
7718 :Architectures: x86 5926 :Architectures: x86
7719 :Parameters: args[0] whether feature should b 5927 :Parameters: args[0] whether feature should be enabled or not
7720 5928
7721 With this capability enabled, CR2 will not be 5929 With this capability enabled, CR2 will not be modified prior to the
7722 emulated VM-exit when L1 intercepts a #PF exc 5930 emulated VM-exit when L1 intercepts a #PF exception that occurs in
7723 L2. Similarly, for kvm-intel only, DR6 will n 5931 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
7724 the emulated VM-exit when L1 intercepts a #DB 5932 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
7725 L2. As a result, when KVM_GET_VCPU_EVENTS rep 5933 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
7726 #DB) exception for L2, exception.has_payload 5934 #DB) exception for L2, exception.has_payload will be set and the
7727 faulting address (or the new DR6 bits*) will 5935 faulting address (or the new DR6 bits*) will be reported in the
7728 exception_payload field. Similarly, when user 5936 exception_payload field. Similarly, when userspace injects a #PF (or
7729 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is 5937 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
7730 exception.has_payload and to put the faulting 5938 exception.has_payload and to put the faulting address - or the new DR6
7731 bits\ [#]_ - in the exception_payload field. 5939 bits\ [#]_ - in the exception_payload field.
7732 5940
7733 This capability also enables exception.pendin 5941 This capability also enables exception.pending in struct
7734 kvm_vcpu_events, which allows userspace to di 5942 kvm_vcpu_events, which allows userspace to distinguish between pending
7735 and injected exceptions. 5943 and injected exceptions.
7736 5944
7737 5945
7738 .. [#] For the new DR6 bits, note that bit 16 5946 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
7739 will clear DR6.RTM. 5947 will clear DR6.RTM.
7740 5948
7741 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 5949 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
7742 -------------------------------------- <<
7743 5950
7744 :Architectures: x86, arm64, mips !! 5951 :Architectures: x86, arm, arm64, mips
7745 :Parameters: args[0] whether feature should b 5952 :Parameters: args[0] whether feature should be enabled or not
7746 5953
7747 Valid flags are:: 5954 Valid flags are::
7748 5955
7749 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE 5956 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
7750 #define KVM_DIRTY_LOG_INITIALLY_SET 5957 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
7751 5958
7752 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is s 5959 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
7753 automatically clear and write-protect all pag 5960 automatically clear and write-protect all pages that are returned as dirty.
7754 Rather, userspace will have to do this operat 5961 Rather, userspace will have to do this operation separately using
7755 KVM_CLEAR_DIRTY_LOG. 5962 KVM_CLEAR_DIRTY_LOG.
7756 5963
7757 At the cost of a slightly more complicated op 5964 At the cost of a slightly more complicated operation, this provides better
7758 scalability and responsiveness for two reason 5965 scalability and responsiveness for two reasons. First,
7759 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64 5966 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
7760 than requiring to sync a full memslot; this e 5967 than requiring to sync a full memslot; this ensures that KVM does not
7761 take spinlocks for an extended period of time 5968 take spinlocks for an extended period of time. Second, in some cases a
7762 large amount of time can pass between a call 5969 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
7763 userspace actually using the data in the page 5970 userspace actually using the data in the page. Pages can be modified
7764 during this time, which is inefficient for bo 5971 during this time, which is inefficient for both the guest and userspace:
7765 the guest will incur a higher penalty due to 5972 the guest will incur a higher penalty due to write protection faults,
7766 while userspace can see false reports of dirt 5973 while userspace can see false reports of dirty pages. Manual reprotection
7767 helps reducing this time, improving guest per 5974 helps reducing this time, improving guest performance and reducing the
7768 number of dirty log false positives. 5975 number of dirty log false positives.
7769 5976
7770 With KVM_DIRTY_LOG_INITIALLY_SET set, all the 5977 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
7771 will be initialized to 1 when created. This 5978 will be initialized to 1 when created. This also improves performance because
7772 dirty logging can be enabled gradually in sma 5979 dirty logging can be enabled gradually in small chunks on the first call
7773 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIA 5980 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
7774 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is al 5981 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
7775 x86 and arm64 for now). 5982 x86 and arm64 for now).
7776 5983
7777 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previou 5984 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
7778 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the imp 5985 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
7779 it hard or impossible to use it correctly. T 5986 it hard or impossible to use it correctly. The availability of
7780 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals tha 5987 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
7781 Userspace should not try to use KVM_CAP_MANUA 5988 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
7782 5989
7783 7.19 KVM_CAP_PPC_SECURE_GUEST 5990 7.19 KVM_CAP_PPC_SECURE_GUEST
7784 ------------------------------ 5991 ------------------------------
7785 5992
7786 :Architectures: ppc 5993 :Architectures: ppc
7787 5994
7788 This capability indicates that KVM is running 5995 This capability indicates that KVM is running on a host that has
7789 ultravisor firmware and thus can support a se 5996 ultravisor firmware and thus can support a secure guest. On such a
7790 system, a guest can ask the ultravisor to mak 5997 system, a guest can ask the ultravisor to make it a secure guest,
7791 one whose memory is inaccessible to the host 5998 one whose memory is inaccessible to the host except for pages which
7792 are explicitly requested to be shared with th 5999 are explicitly requested to be shared with the host. The ultravisor
7793 notifies KVM when a guest requests to become 6000 notifies KVM when a guest requests to become a secure guest, and KVM
7794 has the opportunity to veto the transition. 6001 has the opportunity to veto the transition.
7795 6002
7796 If present, this capability can be enabled fo 6003 If present, this capability can be enabled for a VM, meaning that KVM
7797 will allow the transition to secure guest mod 6004 will allow the transition to secure guest mode. Otherwise KVM will
7798 veto the transition. 6005 veto the transition.
7799 6006
7800 7.20 KVM_CAP_HALT_POLL 6007 7.20 KVM_CAP_HALT_POLL
7801 ---------------------- 6008 ----------------------
7802 6009
7803 :Architectures: all 6010 :Architectures: all
7804 :Target: VM 6011 :Target: VM
7805 :Parameters: args[0] is the maximum poll time 6012 :Parameters: args[0] is the maximum poll time in nanoseconds
7806 :Returns: 0 on success; -1 on error 6013 :Returns: 0 on success; -1 on error
7807 6014
7808 KVM_CAP_HALT_POLL overrides the kvm.halt_poll !! 6015 This capability overrides the kvm module parameter halt_poll_ns for the
7809 maximum halt-polling time for all vCPUs in th !! 6016 target VM.
7810 be invoked at any time and any number of time <<
7811 maximum halt-polling time. <<
7812 6017
7813 See Documentation/virt/kvm/halt-polling.rst f !! 6018 VCPU polling allows a VCPU to poll for wakeup events instead of immediately
7814 polling. !! 6019 scheduling during guest halts. The maximum time a VCPU can spend polling is
>> 6020 controlled by the kvm module parameter halt_poll_ns. This capability allows
>> 6021 the maximum halt time to specified on a per-VM basis, effectively overriding
>> 6022 the module parameter for the target VM.
7815 6023
7816 7.21 KVM_CAP_X86_USER_SPACE_MSR 6024 7.21 KVM_CAP_X86_USER_SPACE_MSR
7817 ------------------------------- 6025 -------------------------------
7818 6026
7819 :Architectures: x86 6027 :Architectures: x86
7820 :Target: VM 6028 :Target: VM
7821 :Parameters: args[0] contains the mask of KVM 6029 :Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
7822 :Returns: 0 on success; -1 on error 6030 :Returns: 0 on success; -1 on error
7823 6031
7824 This capability allows userspace to intercept !! 6032 This capability enables trapping of #GP invoking RDMSR and WRMSR instructions
7825 access to an MSR is denied. By default, KVM !! 6033 into user space.
7826 6034
7827 When a guest requests to read or write an MSR 6035 When a guest requests to read or write an MSR, KVM may not implement all MSRs
7828 that are relevant to a respective system. It 6036 that are relevant to a respective system. It also does not differentiate by
7829 CPU type. 6037 CPU type.
7830 6038
7831 To allow more fine grained control over MSR h !! 6039 To allow more fine grained control over MSR handling, user space may enable
7832 this capability. With it enabled, MSR accesse 6040 this capability. With it enabled, MSR accesses that match the mask specified in
7833 args[0] and would trigger a #GP inside the gu !! 6041 args[0] and trigger a #GP event inside the guest by KVM will instead trigger
7834 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exi !! 6042 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
7835 can then implement model specific MSR handlin !! 6043 can then handle to implement model specific MSR handling and/or user notifications
7836 to inform a user that an MSR was not emulated !! 6044 to inform a user that an MSR was not handled.
7837 <<
7838 The valid mask flags are: <<
7839 <<
7840 ============================ ================ <<
7841 KVM_MSR_EXIT_REASON_UNKNOWN intercept access <<
7842 KVM_MSR_EXIT_REASON_INVAL intercept access <<
7843 invalid accordin <<
7844 KVM_MSR_EXIT_REASON_FILTER intercept access <<
7845 via KVM_X86_SET_ <<
7846 ============================ ================ <<
7847 <<
7848 7.22 KVM_CAP_X86_BUS_LOCK_EXIT <<
7849 ------------------------------- <<
7850 <<
7851 :Architectures: x86 <<
7852 :Target: VM <<
7853 :Parameters: args[0] defines the policy used <<
7854 :Returns: 0 on success, -EINVAL when args[0] <<
7855 <<
7856 Valid bits in args[0] are:: <<
7857 <<
7858 #define KVM_BUS_LOCK_DETECTION_OFF (1 <<
7859 #define KVM_BUS_LOCK_DETECTION_EXIT (1 <<
7860 <<
7861 Enabling this capability on a VM provides use <<
7862 policy to handle the bus locks detected in gu <<
7863 supported modes from the result of KVM_CHECK_ <<
7864 the KVM_ENABLE_CAP. The supported modes are m <<
7865 <<
7866 This capability allows userspace to force VM <<
7867 guest, irrespective whether or not the host h <<
7868 (which triggers an #AC exception that KVM int <<
7869 intended to mitigate attacks where a maliciou <<
7870 locks to degrade the performance of the whole <<
7871 <<
7872 If KVM_BUS_LOCK_DETECTION_OFF is set, KVM doe <<
7873 exit, although the host kernel's split-lock # <<
7874 enabled. <<
7875 <<
7876 If KVM_BUS_LOCK_DETECTION_EXIT is set, KVM en <<
7877 bus locks in the guest trigger a VM exit, and <<
7878 such VM exits, e.g. to allow userspace to thr <<
7879 apply some other policy-based mitigation. Whe <<
7880 KVM_RUN_X86_BUS_LOCK in vcpu-run->flags, and <<
7881 to KVM_EXIT_X86_BUS_LOCK. <<
7882 <<
7883 Note! Detected bus locks may be coincident wi <<
7884 KVM_RUN_X86_BUS_LOCK should be checked regard <<
7885 userspace wants to take action on all detecte <<
7886 <<
7887 7.23 KVM_CAP_PPC_DAWR1 <<
7888 ---------------------- <<
7889 <<
7890 :Architectures: ppc <<
7891 :Parameters: none <<
7892 :Returns: 0 on success, -EINVAL when CPU does <<
7893 <<
7894 This capability can be used to check / enable <<
7895 by POWER10 processor. <<
7896 <<
7897 <<
7898 7.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM <<
7899 ------------------------------------- <<
7900 <<
7901 Architectures: x86 SEV enabled <<
7902 Type: vm <<
7903 Parameters: args[0] is the fd of the source v <<
7904 Returns: 0 on success; ENOTTY on error <<
7905 <<
7906 This capability enables userspace to copy enc <<
7907 indicated by the fd to the vm this is called <<
7908 <<
7909 This is intended to support in-guest workload <<
7910 allows the in-guest workload to maintain its <<
7911 from accidentally clobbering each other with <<
7912 APIC/MSRs/etc). <<
7913 <<
7914 7.25 KVM_CAP_SGX_ATTRIBUTE <<
7915 -------------------------- <<
7916 <<
7917 :Architectures: x86 <<
7918 :Target: VM <<
7919 :Parameters: args[0] is a file handle of a SG <<
7920 :Returns: 0 on success, -EINVAL if the file h <<
7921 attribute is not supported by KVM. <<
7922 <<
7923 KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM <<
7924 more privileged enclave attributes. args[0] <<
7925 SGX attribute file corresponding to an attrib <<
7926 by KVM (currently only PROVISIONKEY). <<
7927 <<
7928 The SGX subsystem restricts access to a subse <<
7929 additional security for an uncompromised kern <<
7930 is restricted to deter malware from using the <<
7931 system fingerprint. To prevent userspace fro <<
7932 by running an enclave in a VM, KVM prevents a <<
7933 default. <<
7934 <<
7935 See Documentation/arch/x86/sgx.rst for more d <<
7936 <<
7937 7.26 KVM_CAP_PPC_RPT_INVALIDATE <<
7938 ------------------------------- <<
7939 <<
7940 :Capability: KVM_CAP_PPC_RPT_INVALIDATE <<
7941 :Architectures: ppc <<
7942 :Type: vm <<
7943 <<
7944 This capability indicates that the kernel is <<
7945 H_RPT_INVALIDATE hcall. <<
7946 <<
7947 In order to enable the use of H_RPT_INVALIDAT <<
7948 user space might have to advertise it for the <<
7949 IBM pSeries (sPAPR) guest starts using it if <<
7950 present in the "ibm,hypertas-functions" devic <<
7951 <<
7952 This capability is enabled for hypervisors on <<
7953 that support radix MMU. <<
7954 <<
7955 7.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE <<
7956 -------------------------------------- <<
7957 <<
7958 :Architectures: x86 <<
7959 :Parameters: args[0] whether the feature shou <<
7960 <<
7961 When this capability is enabled, an emulation <<
7962 to userspace with KVM_INTERNAL_ERROR (except <<
7963 to handle a VMware backdoor instruction). Fur <<
7964 to 15 instruction bytes for any exit to users <<
7965 failure. When these exits to userspace occur <<
7966 instead of the internal struct. They both ha <<
7967 emulation_failure struct matches the content <<
7968 defines the 'flags' field which is used to de <<
7969 that are valid (ie: if KVM_INTERNAL_ERROR_EMU <<
7970 set in the 'flags' field then both 'insn_size <<
7971 in them.) <<
7972 <<
7973 7.28 KVM_CAP_ARM_MTE <<
7974 -------------------- <<
7975 <<
7976 :Architectures: arm64 <<
7977 :Parameters: none <<
7978 <<
7979 This capability indicates that KVM (and the h <<
7980 Memory Tagging Extensions (MTE) to the guest. <<
7981 VMM before creating any VCPUs to allow the gu <<
7982 available to a guest running in AArch64 mode <<
7983 cause attempts to create AArch32 VCPUs to fai <<
7984 <<
7985 When enabled the guest is able to access tags <<
7986 to the guest. KVM will ensure that the tags a <<
7987 hibernation of the host; however the VMM need <<
7988 tags as appropriate if the VM is migrated. <<
7989 <<
7990 When this capability is enabled all memory in <<
7991 ``MAP_ANONYMOUS`` or with a RAM-based file ma <<
7992 attempts to create a memslot with an invalid <<
7993 -EINVAL return. <<
7994 <<
7995 When enabled the VMM may make use of the ``KV <<
7996 perform a bulk copy of tags to/from the guest <<
7997 <<
7998 7.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM <<
7999 ------------------------------------- <<
8000 <<
8001 :Architectures: x86 SEV enabled <<
8002 :Type: vm <<
8003 :Parameters: args[0] is the fd of the source <<
8004 :Returns: 0 on success <<
8005 <<
8006 This capability enables userspace to migrate <<
8007 indicated by the fd to the VM this is called <<
8008 <<
8009 This is intended to support intra-host migrat <<
8010 upgrading the VMM process without interruptin <<
8011 <<
8012 7.30 KVM_CAP_PPC_AIL_MODE_3 <<
8013 ------------------------------- <<
8014 <<
8015 :Capability: KVM_CAP_PPC_AIL_MODE_3 <<
8016 :Architectures: ppc <<
8017 :Type: vm <<
8018 <<
8019 This capability indicates that the kernel sup <<
8020 "Address Translation Mode on Interrupt" aka " <<
8021 resource that is controlled with the H_SET_MO <<
8022 <<
8023 This capability allows a guest kernel to use <<
8024 handling interrupts and system calls. <<
8025 <<
8026 7.31 KVM_CAP_DISABLE_QUIRKS2 <<
8027 ---------------------------- <<
8028 <<
8029 :Capability: KVM_CAP_DISABLE_QUIRKS2 <<
8030 :Parameters: args[0] - set of KVM quirks to d <<
8031 :Architectures: x86 <<
8032 :Type: vm <<
8033 <<
8034 This capability, if enabled, will cause KVM t <<
8035 quirks. <<
8036 <<
8037 Calling KVM_CHECK_EXTENSION for this capabili <<
8038 quirks that can be disabled in KVM. <<
8039 <<
8040 The argument to KVM_ENABLE_CAP for this capab <<
8041 quirks to disable, and must be a subset of th <<
8042 KVM_CHECK_EXTENSION. <<
8043 <<
8044 The valid bits in cap.args[0] are: <<
8045 <<
8046 =================================== ========= <<
8047 KVM_X86_QUIRK_LINT0_REENABLED By defaul <<
8048 LINT0 reg <<
8049 When this <<
8050 is 0x1000 <<
8051 <<
8052 KVM_X86_QUIRK_CD_NW_CLEARED By defaul <<
8053 AMD CPUs <<
8054 that runs <<
8055 with cach <<
8056 <<
8057 When this <<
8058 change th <<
8059 <<
8060 KVM_X86_QUIRK_LAPIC_MMIO_HOLE By defaul <<
8061 available <<
8062 mode. Whe <<
8063 disables <<
8064 LAPIC is <<
8065 <<
8066 KVM_X86_QUIRK_OUT_7E_INC_RIP By defaul <<
8067 exiting t <<
8068 to port 0 <<
8069 KVM does <<
8070 exiting t <<
8071 <<
8072 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this <<
8073 CPUID.01H <<
8074 IA32_MISC <<
8075 Additiona <<
8076 KVM clear <<
8077 IA32_MISC <<
8078 <<
8079 KVM_X86_QUIRK_FIX_HYPERCALL_INSN By defaul <<
8080 VMMCALL/V <<
8081 vendor's <<
8082 system. W <<
8083 will no l <<
8084 hypercall <<
8085 incorrect <<
8086 generate <<
8087 <<
8088 KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By defaul <<
8089 they are <<
8090 whether o <<
8091 according <<
8092 is disabl <<
8093 is not se <<
8094 KVM will <<
8095 they're u <<
8096 KVM will <<
8097 guest CPU <<
8098 KVM_X86_Q <<
8099 disabled. <<
8100 <<
8101 KVM_X86_QUIRK_SLOT_ZAP_ALL By defaul <<
8102 invalidat <<
8103 address s <<
8104 moved. W <<
8105 VM type i <<
8106 ensures t <<
8107 or moved <<
8108 _may_ inv <<
8109 memslot. <<
8110 =================================== ========= <<
8111 <<
8112 7.32 KVM_CAP_MAX_VCPU_ID <<
8113 ------------------------ <<
8114 <<
8115 :Architectures: x86 <<
8116 :Target: VM <<
8117 :Parameters: args[0] - maximum APIC ID value <<
8118 :Returns: 0 on success, -EINVAL if args[0] is <<
8119 supported in KVM or if it has been <<
8120 <<
8121 This capability allows userspace to specify m <<
8122 assigned for current VM session prior to the <<
8123 memory for data structures indexed by the API <<
8124 to calculate the limit to APIC ID values from <<
8125 CPU topology. <<
8126 <<
8127 The value can be changed only until KVM_ENABL <<
8128 value or until a vCPU is created. Upon creat <<
8129 if the value was set to zero or KVM_ENABLE_CA <<
8130 uses the return value of KVM_CHECK_EXTENSION( <<
8131 the maximum APIC ID. <<
8132 <<
8133 7.33 KVM_CAP_X86_NOTIFY_VMEXIT <<
8134 ------------------------------ <<
8135 <<
8136 :Architectures: x86 <<
8137 :Target: VM <<
8138 :Parameters: args[0] is the value of notify w <<
8139 :Returns: 0 on success, -EINVAL if args[0] co <<
8140 VM exit is unsupported. <<
8141 <<
8142 Bits 63:32 of args[0] are used for notify win <<
8143 Bits 31:0 of args[0] are for some flags. Vali <<
8144 <<
8145 #define KVM_X86_NOTIFY_VMEXIT_ENABLED (1 <<
8146 #define KVM_X86_NOTIFY_VMEXIT_USER (1 <<
8147 <<
8148 This capability allows userspace to configure <<
8149 in per-VM scope during VM creation. Notify VM <<
8150 When userspace sets KVM_X86_NOTIFY_VMEXIT_ENA <<
8151 enable this feature with the notify window pr <<
8152 a VM exit if no event window occurs in VM non <<
8153 time (notify window). <<
8154 <<
8155 If KVM_X86_NOTIFY_VMEXIT_USER is set in args[ <<
8156 KVM would exit to userspace for handling. <<
8157 <<
8158 This capability is aimed to mitigate the thre <<
8159 cause CPU stuck (due to event windows don't o <<
8160 unavailable to host or other VMs. <<
8161 <<
8162 7.34 KVM_CAP_MEMORY_FAULT_INFO <<
8163 ------------------------------ <<
8164 <<
8165 :Architectures: x86 <<
8166 :Returns: Informational only, -EINVAL on dire <<
8167 <<
8168 The presence of this capability indicates tha <<
8169 kvm_run.memory_fault if KVM cannot resolve a <<
8170 there is a valid memslot but no backing VMA f <<
8171 address. <<
8172 <<
8173 The information in kvm_run.memory_fault is va <<
8174 an error with errno=EFAULT or errno=EHWPOISON <<
8175 to KVM_EXIT_MEMORY_FAULT. <<
8176 <<
8177 Note: Userspaces which attempt to resolve mem <<
8178 KVM_RUN are encouraged to guard against repea <<
8179 error/annotated fault. <<
8180 <<
8181 See KVM_EXIT_MEMORY_FAULT for more informatio <<
8182 <<
8183 7.35 KVM_CAP_X86_APIC_BUS_CYCLES_NS <<
8184 ----------------------------------- <<
8185 <<
8186 :Architectures: x86 <<
8187 :Target: VM <<
8188 :Parameters: args[0] is the desired APIC bus <<
8189 :Returns: 0 on success, -EINVAL if args[0] co <<
8190 frequency or if any vCPUs have been <<
8191 local APIC has not been created usi <<
8192 <<
8193 This capability sets the VM's APIC bus clock <<
8194 virtual APIC when emulating APIC timers. KVM <<
8195 by KVM_CHECK_EXTENSION. <<
8196 <<
8197 Note: Userspace is responsible for correctly <<
8198 core crystal clock frequency, if a non-zero C <<
8199 <<
8200 7.36 KVM_CAP_X86_GUEST_MODE <<
8201 ------------------------------ <<
8202 <<
8203 :Architectures: x86 <<
8204 :Returns: Informational only, -EINVAL on dire <<
8205 <<
8206 The presence of this capability indicates tha <<
8207 KVM_RUN_X86_GUEST_MODE bit in kvm_run.flags t <<
8208 vCPU was executing nested guest code when it <<
8209 <<
8210 KVM exits with the register state of either t <<
8211 depending on which executed at the time of an <<
8212 take care to differentiate between these case <<
8213 6045
8214 8. Other capabilities. 6046 8. Other capabilities.
8215 ====================== 6047 ======================
8216 6048
8217 This section lists capabilities that give inf 6049 This section lists capabilities that give information about other
8218 features of the KVM implementation. 6050 features of the KVM implementation.
8219 6051
8220 8.1 KVM_CAP_PPC_HWRNG 6052 8.1 KVM_CAP_PPC_HWRNG
8221 --------------------- 6053 ---------------------
8222 6054
8223 :Architectures: ppc 6055 :Architectures: ppc
8224 6056
8225 This capability, if KVM_CHECK_EXTENSION indic 6057 This capability, if KVM_CHECK_EXTENSION indicates that it is
8226 available, means that the kernel has an imple 6058 available, means that the kernel has an implementation of the
8227 H_RANDOM hypercall backed by a hardware rando 6059 H_RANDOM hypercall backed by a hardware random-number generator.
8228 If present, the kernel H_RANDOM handler can b 6060 If present, the kernel H_RANDOM handler can be enabled for guest use
8229 with the KVM_CAP_PPC_ENABLE_HCALL capability. 6061 with the KVM_CAP_PPC_ENABLE_HCALL capability.
8230 6062
8231 8.2 KVM_CAP_HYPERV_SYNIC 6063 8.2 KVM_CAP_HYPERV_SYNIC
8232 ------------------------ 6064 ------------------------
8233 6065
8234 :Architectures: x86 6066 :Architectures: x86
8235 6067
8236 This capability, if KVM_CHECK_EXTENSION indic 6068 This capability, if KVM_CHECK_EXTENSION indicates that it is
8237 available, means that the kernel has an imple 6069 available, means that the kernel has an implementation of the
8238 Hyper-V Synthetic interrupt controller(SynIC) 6070 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
8239 used to support Windows Hyper-V based guest p 6071 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
8240 6072
8241 In order to use SynIC, it has to be activated 6073 In order to use SynIC, it has to be activated by setting this
8242 capability via KVM_ENABLE_CAP ioctl on the vc 6074 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
8243 will disable the use of APIC hardware virtual 6075 will disable the use of APIC hardware virtualization even if supported
8244 by the CPU, as it's incompatible with SynIC a 6076 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
8245 6077
8246 8.3 KVM_CAP_PPC_MMU_RADIX !! 6078 8.3 KVM_CAP_PPC_RADIX_MMU
8247 ------------------------- 6079 -------------------------
8248 6080
8249 :Architectures: ppc 6081 :Architectures: ppc
8250 6082
8251 This capability, if KVM_CHECK_EXTENSION indic 6083 This capability, if KVM_CHECK_EXTENSION indicates that it is
8252 available, means that the kernel can support 6084 available, means that the kernel can support guests using the
8253 radix MMU defined in Power ISA V3.00 (as impl 6085 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
8254 processor). 6086 processor).
8255 6087
8256 8.4 KVM_CAP_PPC_MMU_HASH_V3 !! 6088 8.4 KVM_CAP_PPC_HASH_MMU_V3
8257 --------------------------- 6089 ---------------------------
8258 6090
8259 :Architectures: ppc 6091 :Architectures: ppc
8260 6092
8261 This capability, if KVM_CHECK_EXTENSION indic 6093 This capability, if KVM_CHECK_EXTENSION indicates that it is
8262 available, means that the kernel can support 6094 available, means that the kernel can support guests using the
8263 hashed page table MMU defined in Power ISA V3 6095 hashed page table MMU defined in Power ISA V3.00 (as implemented in
8264 the POWER9 processor), including in-memory se 6096 the POWER9 processor), including in-memory segment tables.
8265 6097
8266 8.5 KVM_CAP_MIPS_VZ 6098 8.5 KVM_CAP_MIPS_VZ
8267 ------------------- 6099 -------------------
8268 6100
8269 :Architectures: mips 6101 :Architectures: mips
8270 6102
8271 This capability, if KVM_CHECK_EXTENSION on th 6103 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
8272 it is available, means that full hardware ass 6104 it is available, means that full hardware assisted virtualization capabilities
8273 of the hardware are available for use through 6105 of the hardware are available for use through KVM. An appropriate
8274 KVM_VM_MIPS_* type must be passed to KVM_CREA 6106 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
8275 utilises it. 6107 utilises it.
8276 6108
8277 If KVM_CHECK_EXTENSION on a kvm VM handle ind 6109 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
8278 available, it means that the VM is using full 6110 available, it means that the VM is using full hardware assisted virtualization
8279 capabilities of the hardware. This is useful 6111 capabilities of the hardware. This is useful to check after creating a VM with
8280 KVM_VM_MIPS_DEFAULT. 6112 KVM_VM_MIPS_DEFAULT.
8281 6113
8282 The value returned by KVM_CHECK_EXTENSION sho 6114 The value returned by KVM_CHECK_EXTENSION should be compared against known
8283 values (see below). All other values are rese 6115 values (see below). All other values are reserved. This is to allow for the
8284 possibility of other hardware assisted virtua 6116 possibility of other hardware assisted virtualization implementations which
8285 may be incompatible with the MIPS VZ ASE. 6117 may be incompatible with the MIPS VZ ASE.
8286 6118
8287 == ========================================= 6119 == ==========================================================================
8288 0 The trap & emulate implementation is in u 6120 0 The trap & emulate implementation is in use to run guest code in user
8289 mode. Guest virtual memory segments are r 6121 mode. Guest virtual memory segments are rearranged to fit the guest in the
8290 user mode address space. 6122 user mode address space.
8291 6123
8292 1 The MIPS VZ ASE is in use, providing full 6124 1 The MIPS VZ ASE is in use, providing full hardware assisted
8293 virtualization, including standard guest 6125 virtualization, including standard guest virtual memory segments.
8294 == ========================================= 6126 == ==========================================================================
8295 6127
8296 8.6 KVM_CAP_MIPS_TE 6128 8.6 KVM_CAP_MIPS_TE
8297 ------------------- 6129 -------------------
8298 6130
8299 :Architectures: mips 6131 :Architectures: mips
8300 6132
8301 This capability, if KVM_CHECK_EXTENSION on th 6133 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
8302 it is available, means that the trap & emulat 6134 it is available, means that the trap & emulate implementation is available to
8303 run guest code in user mode, even if KVM_CAP_ 6135 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
8304 assisted virtualisation is also available. KV 6136 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
8305 to KVM_CREATE_VM to create a VM which utilise 6137 to KVM_CREATE_VM to create a VM which utilises it.
8306 6138
8307 If KVM_CHECK_EXTENSION on a kvm VM handle ind 6139 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
8308 available, it means that the VM is using trap 6140 available, it means that the VM is using trap & emulate.
8309 6141
8310 8.7 KVM_CAP_MIPS_64BIT 6142 8.7 KVM_CAP_MIPS_64BIT
8311 ---------------------- 6143 ----------------------
8312 6144
8313 :Architectures: mips 6145 :Architectures: mips
8314 6146
8315 This capability indicates the supported archi 6147 This capability indicates the supported architecture type of the guest, i.e. the
8316 supported register and address width. 6148 supported register and address width.
8317 6149
8318 The values returned when this capability is c 6150 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
8319 kvm VM handle correspond roughly to the CP0_C 6151 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
8320 be checked specifically against known values 6152 be checked specifically against known values (see below). All other values are
8321 reserved. 6153 reserved.
8322 6154
8323 == ========================================= 6155 == ========================================================================
8324 0 MIPS32 or microMIPS32. 6156 0 MIPS32 or microMIPS32.
8325 Both registers and addresses are 32-bits 6157 Both registers and addresses are 32-bits wide.
8326 It will only be possible to run 32-bit gu 6158 It will only be possible to run 32-bit guest code.
8327 6159
8328 1 MIPS64 or microMIPS64 with access only to 6160 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
8329 Registers are 64-bits wide, but addresses 6161 Registers are 64-bits wide, but addresses are 32-bits wide.
8330 64-bit guest code may run but cannot acce 6162 64-bit guest code may run but cannot access MIPS64 memory segments.
8331 It will also be possible to run 32-bit gu 6163 It will also be possible to run 32-bit guest code.
8332 6164
8333 2 MIPS64 or microMIPS64 with access to all 6165 2 MIPS64 or microMIPS64 with access to all address segments.
8334 Both registers and addresses are 64-bits 6166 Both registers and addresses are 64-bits wide.
8335 It will be possible to run 64-bit or 32-b 6167 It will be possible to run 64-bit or 32-bit guest code.
8336 == ========================================= 6168 == ========================================================================
8337 6169
8338 8.9 KVM_CAP_ARM_USER_IRQ 6170 8.9 KVM_CAP_ARM_USER_IRQ
8339 ------------------------ 6171 ------------------------
8340 6172
8341 :Architectures: arm64 !! 6173 :Architectures: arm, arm64
8342 6174
8343 This capability, if KVM_CHECK_EXTENSION indic 6175 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
8344 that if userspace creates a VM without an in- 6176 that if userspace creates a VM without an in-kernel interrupt controller, it
8345 will be notified of changes to the output lev 6177 will be notified of changes to the output level of in-kernel emulated devices,
8346 which can generate virtual interrupts, presen 6178 which can generate virtual interrupts, presented to the VM.
8347 For such VMs, on every return to userspace, t 6179 For such VMs, on every return to userspace, the kernel
8348 updates the vcpu's run->s.regs.device_irq_lev 6180 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
8349 output level of the device. 6181 output level of the device.
8350 6182
8351 Whenever kvm detects a change in the device o 6183 Whenever kvm detects a change in the device output level, kvm guarantees at
8352 least one return to userspace before running 6184 least one return to userspace before running the VM. This exit could either
8353 be a KVM_EXIT_INTR or any other exit event, l 6185 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
8354 userspace can always sample the device output 6186 userspace can always sample the device output level and re-compute the state of
8355 the userspace interrupt controller. Userspac 6187 the userspace interrupt controller. Userspace should always check the state
8356 of run->s.regs.device_irq_level on every kvm 6188 of run->s.regs.device_irq_level on every kvm exit.
8357 The value in run->s.regs.device_irq_level can 6189 The value in run->s.regs.device_irq_level can represent both level and edge
8358 triggered interrupt signals, depending on the 6190 triggered interrupt signals, depending on the device. Edge triggered interrupt
8359 signals will exit to userspace with the bit i 6191 signals will exit to userspace with the bit in run->s.regs.device_irq_level
8360 set exactly once per edge signal. 6192 set exactly once per edge signal.
8361 6193
8362 The field run->s.regs.device_irq_level is ava 6194 The field run->s.regs.device_irq_level is available independent of
8363 run->kvm_valid_regs or run->kvm_dirty_regs bi 6195 run->kvm_valid_regs or run->kvm_dirty_regs bits.
8364 6196
8365 If KVM_CAP_ARM_USER_IRQ is supported, the KVM 6197 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
8366 number larger than 0 indicating the version o 6198 number larger than 0 indicating the version of this capability is implemented
8367 and thereby which bits in run->s.regs.device_ 6199 and thereby which bits in run->s.regs.device_irq_level can signal values.
8368 6200
8369 Currently the following bits are defined for 6201 Currently the following bits are defined for the device_irq_level bitmap::
8370 6202
8371 KVM_CAP_ARM_USER_IRQ >= 1: 6203 KVM_CAP_ARM_USER_IRQ >= 1:
8372 6204
8373 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual tim 6205 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
8374 KVM_ARM_DEV_EL1_PTIMER - EL1 physical ti 6206 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
8375 KVM_ARM_DEV_PMU - ARM PMU overflo 6207 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
8376 6208
8377 Future versions of kvm may implement addition 6209 Future versions of kvm may implement additional events. These will get
8378 indicated by returning a higher number from K 6210 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
8379 listed above. 6211 listed above.
8380 6212
8381 8.10 KVM_CAP_PPC_SMT_POSSIBLE 6213 8.10 KVM_CAP_PPC_SMT_POSSIBLE
8382 ----------------------------- 6214 -----------------------------
8383 6215
8384 :Architectures: ppc 6216 :Architectures: ppc
8385 6217
8386 Querying this capability returns a bitmap ind 6218 Querying this capability returns a bitmap indicating the possible
8387 virtual SMT modes that can be set using KVM_C 6219 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
8388 (counting from the right) is set, then a virt 6220 (counting from the right) is set, then a virtual SMT mode of 2^N is
8389 available. 6221 available.
8390 6222
8391 8.11 KVM_CAP_HYPERV_SYNIC2 6223 8.11 KVM_CAP_HYPERV_SYNIC2
8392 -------------------------- 6224 --------------------------
8393 6225
8394 :Architectures: x86 6226 :Architectures: x86
8395 6227
8396 This capability enables a newer version of Hy 6228 This capability enables a newer version of Hyper-V Synthetic interrupt
8397 controller (SynIC). The only difference with 6229 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
8398 doesn't clear SynIC message and event flags p 6230 doesn't clear SynIC message and event flags pages when they are enabled by
8399 writing to the respective MSRs. 6231 writing to the respective MSRs.
8400 6232
8401 8.12 KVM_CAP_HYPERV_VP_INDEX 6233 8.12 KVM_CAP_HYPERV_VP_INDEX
8402 ---------------------------- 6234 ----------------------------
8403 6235
8404 :Architectures: x86 6236 :Architectures: x86
8405 6237
8406 This capability indicates that userspace can 6238 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
8407 value is used to denote the target vcpu for a 6239 value is used to denote the target vcpu for a SynIC interrupt. For
8408 compatibility, KVM initializes this msr to KV !! 6240 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
8409 capability is absent, userspace can still que 6241 capability is absent, userspace can still query this msr's value.
8410 6242
8411 8.13 KVM_CAP_S390_AIS_MIGRATION 6243 8.13 KVM_CAP_S390_AIS_MIGRATION
8412 ------------------------------- 6244 -------------------------------
8413 6245
8414 :Architectures: s390 6246 :Architectures: s390
8415 :Parameters: none 6247 :Parameters: none
8416 6248
8417 This capability indicates if the flic device 6249 This capability indicates if the flic device will be able to get/set the
8418 AIS states for migration via the KVM_DEV_FLIC 6250 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
8419 to discover this without having to create a f 6251 to discover this without having to create a flic device.
8420 6252
8421 8.14 KVM_CAP_S390_PSW 6253 8.14 KVM_CAP_S390_PSW
8422 --------------------- 6254 ---------------------
8423 6255
8424 :Architectures: s390 6256 :Architectures: s390
8425 6257
8426 This capability indicates that the PSW is exp 6258 This capability indicates that the PSW is exposed via the kvm_run structure.
8427 6259
8428 8.15 KVM_CAP_S390_GMAP 6260 8.15 KVM_CAP_S390_GMAP
8429 ---------------------- 6261 ----------------------
8430 6262
8431 :Architectures: s390 6263 :Architectures: s390
8432 6264
8433 This capability indicates that the user space 6265 This capability indicates that the user space memory used as guest mapping can
8434 be anywhere in the user memory address space, 6266 be anywhere in the user memory address space, as long as the memory slots are
8435 aligned and sized to a segment (1MB) boundary 6267 aligned and sized to a segment (1MB) boundary.
8436 6268
8437 8.16 KVM_CAP_S390_COW 6269 8.16 KVM_CAP_S390_COW
8438 --------------------- 6270 ---------------------
8439 6271
8440 :Architectures: s390 6272 :Architectures: s390
8441 6273
8442 This capability indicates that the user space 6274 This capability indicates that the user space memory used as guest mapping can
8443 use copy-on-write semantics as well as dirty 6275 use copy-on-write semantics as well as dirty pages tracking via read-only page
8444 tables. 6276 tables.
8445 6277
8446 8.17 KVM_CAP_S390_BPB 6278 8.17 KVM_CAP_S390_BPB
8447 --------------------- 6279 ---------------------
8448 6280
8449 :Architectures: s390 6281 :Architectures: s390
8450 6282
8451 This capability indicates that kvm will imple 6283 This capability indicates that kvm will implement the interfaces to handle
8452 reset, migration and nested KVM for branch pr 6284 reset, migration and nested KVM for branch prediction blocking. The stfle
8453 facility 82 should not be provided to the gue 6285 facility 82 should not be provided to the guest without this capability.
8454 6286
8455 8.18 KVM_CAP_HYPERV_TLBFLUSH 6287 8.18 KVM_CAP_HYPERV_TLBFLUSH
8456 ---------------------------- 6288 ----------------------------
8457 6289
8458 :Architectures: x86 6290 :Architectures: x86
8459 6291
8460 This capability indicates that KVM supports p 6292 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
8461 hypercalls: 6293 hypercalls:
8462 HvFlushVirtualAddressSpace, HvFlushVirtualAdd 6294 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
8463 HvFlushVirtualAddressList, HvFlushVirtualAddr 6295 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
8464 6296
8465 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR 6297 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
8466 ---------------------------------- 6298 ----------------------------------
8467 6299
8468 :Architectures: arm64 !! 6300 :Architectures: arm, arm64
8469 6301
8470 This capability indicates that userspace can 6302 This capability indicates that userspace can specify (via the
8471 KVM_SET_VCPU_EVENTS ioctl) the syndrome value 6303 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
8472 takes a virtual SError interrupt exception. 6304 takes a virtual SError interrupt exception.
8473 If KVM advertises this capability, userspace 6305 If KVM advertises this capability, userspace can only specify the ISS field for
8474 the ESR syndrome. Other parts of the ESR, suc 6306 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
8475 CPU when the exception is taken. If this virt 6307 CPU when the exception is taken. If this virtual SError is taken to EL1 using
8476 AArch64, this value will be reported in the I 6308 AArch64, this value will be reported in the ISS field of ESR_ELx.
8477 6309
8478 See KVM_CAP_VCPU_EVENTS for more details. 6310 See KVM_CAP_VCPU_EVENTS for more details.
8479 6311
8480 8.20 KVM_CAP_HYPERV_SEND_IPI 6312 8.20 KVM_CAP_HYPERV_SEND_IPI
8481 ---------------------------- 6313 ----------------------------
8482 6314
8483 :Architectures: x86 6315 :Architectures: x86
8484 6316
8485 This capability indicates that KVM supports p 6317 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
8486 hypercalls: 6318 hypercalls:
8487 HvCallSendSyntheticClusterIpi, HvCallSendSynt 6319 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
8488 6320
8489 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH 6321 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
8490 ----------------------------------- 6322 -----------------------------------
8491 6323
8492 :Architectures: x86 6324 :Architectures: x86
8493 6325
8494 This capability indicates that KVM running on 6326 This capability indicates that KVM running on top of Hyper-V hypervisor
8495 enables Direct TLB flush for its guests meani 6327 enables Direct TLB flush for its guests meaning that TLB flush
8496 hypercalls are handled by Level 0 hypervisor 6328 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
8497 Due to the different ABI for hypercall parame 6329 Due to the different ABI for hypercall parameters between Hyper-V and
8498 KVM, enabling this capability effectively dis 6330 KVM, enabling this capability effectively disables all hypercall
8499 handling by KVM (as some KVM hypercall may be 6331 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
8500 flush hypercalls by Hyper-V) so userspace sho 6332 flush hypercalls by Hyper-V) so userspace should disable KVM identification
8501 in CPUID and only exposes Hyper-V identificat 6333 in CPUID and only exposes Hyper-V identification. In this case, guest
8502 thinks it's running on Hyper-V and only use H 6334 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
8503 6335
8504 8.22 KVM_CAP_S390_VCPU_RESETS 6336 8.22 KVM_CAP_S390_VCPU_RESETS
8505 ----------------------------- 6337 -----------------------------
8506 6338
8507 :Architectures: s390 6339 :Architectures: s390
8508 6340
8509 This capability indicates that the KVM_S390_N 6341 This capability indicates that the KVM_S390_NORMAL_RESET and
8510 KVM_S390_CLEAR_RESET ioctls are available. 6342 KVM_S390_CLEAR_RESET ioctls are available.
8511 6343
8512 8.23 KVM_CAP_S390_PROTECTED 6344 8.23 KVM_CAP_S390_PROTECTED
8513 --------------------------- 6345 ---------------------------
8514 6346
8515 :Architectures: s390 6347 :Architectures: s390
8516 6348
8517 This capability indicates that the Ultravisor 6349 This capability indicates that the Ultravisor has been initialized and
8518 KVM can therefore start protected VMs. 6350 KVM can therefore start protected VMs.
8519 This capability governs the KVM_S390_PV_COMMA 6351 This capability governs the KVM_S390_PV_COMMAND ioctl and the
8520 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE 6352 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
8521 guests when the state change is invalid. 6353 guests when the state change is invalid.
8522 6354
8523 8.24 KVM_CAP_STEAL_TIME 6355 8.24 KVM_CAP_STEAL_TIME
8524 ----------------------- 6356 -----------------------
8525 6357
8526 :Architectures: arm64, x86 6358 :Architectures: arm64, x86
8527 6359
8528 This capability indicates that KVM supports s 6360 This capability indicates that KVM supports steal time accounting.
8529 When steal time accounting is supported it ma 6361 When steal time accounting is supported it may be enabled with
8530 architecture-specific interfaces. This capab 6362 architecture-specific interfaces. This capability and the architecture-
8531 specific interfaces must be consistent, i.e. 6363 specific interfaces must be consistent, i.e. if one says the feature
8532 is supported, than the other should as well a 6364 is supported, than the other should as well and vice versa. For arm64
8533 see Documentation/virt/kvm/devices/vcpu.rst " 6365 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
8534 For x86 see Documentation/virt/kvm/x86/msr.rs !! 6366 For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
8535 6367
8536 8.25 KVM_CAP_S390_DIAG318 6368 8.25 KVM_CAP_S390_DIAG318
8537 ------------------------- 6369 -------------------------
8538 6370
8539 :Architectures: s390 6371 :Architectures: s390
8540 6372
8541 This capability enables a guest to set inform 6373 This capability enables a guest to set information about its control program
8542 (i.e. guest kernel type and version). The inf 6374 (i.e. guest kernel type and version). The information is helpful during
8543 system/firmware service events, providing add 6375 system/firmware service events, providing additional data about the guest
8544 environments running on the machine. 6376 environments running on the machine.
8545 6377
8546 The information is associated with the DIAGNO 6378 The information is associated with the DIAGNOSE 0x318 instruction, which sets
8547 an 8-byte value consisting of a one-byte Cont 6379 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
8548 a 7-byte Control Program Version Code (CPVC). 6380 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
8549 environment the control program is running in 6381 environment the control program is running in (e.g. Linux, z/VM...), and the
8550 CPVC is used for information specific to OS ( 6382 CPVC is used for information specific to OS (e.g. Linux version, Linux
8551 distribution...) 6383 distribution...)
8552 6384
8553 If this capability is available, then the CPN 6385 If this capability is available, then the CPNC and CPVC can be synchronized
8554 between KVM and userspace via the sync regs m 6386 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
8555 6387
8556 8.26 KVM_CAP_X86_USER_SPACE_MSR 6388 8.26 KVM_CAP_X86_USER_SPACE_MSR
8557 ------------------------------- 6389 -------------------------------
8558 6390
8559 :Architectures: x86 6391 :Architectures: x86
8560 6392
8561 This capability indicates that KVM supports d 6393 This capability indicates that KVM supports deflection of MSR reads and
8562 writes to user space. It can be enabled on a 6394 writes to user space. It can be enabled on a VM level. If enabled, MSR
8563 accesses that would usually trigger a #GP by 6395 accesses that would usually trigger a #GP by KVM into the guest will
8564 instead get bounced to user space through the 6396 instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
8565 KVM_EXIT_X86_WRMSR exit notifications. 6397 KVM_EXIT_X86_WRMSR exit notifications.
8566 6398
8567 8.27 KVM_CAP_X86_MSR_FILTER !! 6399 8.27 KVM_X86_SET_MSR_FILTER
8568 --------------------------- 6400 ---------------------------
8569 6401
8570 :Architectures: x86 6402 :Architectures: x86
8571 6403
8572 This capability indicates that KVM supports t 6404 This capability indicates that KVM supports that accesses to user defined MSRs
8573 may be rejected. With this capability exposed 6405 may be rejected. With this capability exposed, KVM exports new VM ioctl
8574 KVM_X86_SET_MSR_FILTER which user space can c 6406 KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
8575 ranges that KVM should deny access to. !! 6407 ranges that KVM should reject access to.
8576 6408
8577 In combination with KVM_CAP_X86_USER_SPACE_MS 6409 In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
8578 trap and emulate MSRs that are outside of the 6410 trap and emulate MSRs that are outside of the scope of KVM as well as
8579 limit the attack surface on KVM's MSR emulati 6411 limit the attack surface on KVM's MSR emulation code.
8580 6412
8581 8.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID !! 6413 8.28 KVM_CAP_ENFORCE_PV_CPUID
8582 ------------------------------------- !! 6414 -----------------------------
8583 6415
8584 Architectures: x86 6416 Architectures: x86
8585 6417
8586 When enabled, KVM will disable paravirtual fe 6418 When enabled, KVM will disable paravirtual features provided to the
8587 guest according to the bits in the KVM_CPUID_ 6419 guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
8588 (0x40000001). Otherwise, a guest may use the 6420 (0x40000001). Otherwise, a guest may use the paravirtual features
8589 regardless of what has actually been exposed 6421 regardless of what has actually been exposed through the CPUID leaf.
8590 6422
8591 8.29 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG <<
8592 --------------------------------------------- <<
8593 6423
8594 :Architectures: x86, arm64 !! 6424 8.29 KVM_CAP_DIRTY_LOG_RING
>> 6425 ---------------------------
>> 6426
>> 6427 :Architectures: x86
8595 :Parameters: args[0] - size of the dirty log 6428 :Parameters: args[0] - size of the dirty log ring
8596 6429
8597 KVM is capable of tracking dirty memory using 6430 KVM is capable of tracking dirty memory using ring buffers that are
8598 mmapped into userspace; there is one dirty ri !! 6431 mmaped into userspace; there is one dirty ring per vcpu.
8599 6432
8600 The dirty ring is available to userspace as a 6433 The dirty ring is available to userspace as an array of
8601 ``struct kvm_dirty_gfn``. Each dirty entry i !! 6434 ``struct kvm_dirty_gfn``. Each dirty entry it's defined as::
8602 6435
8603 struct kvm_dirty_gfn { 6436 struct kvm_dirty_gfn {
8604 __u32 flags; 6437 __u32 flags;
8605 __u32 slot; /* as_id | slot_id */ 6438 __u32 slot; /* as_id | slot_id */
8606 __u64 offset; 6439 __u64 offset;
8607 }; 6440 };
8608 6441
8609 The following values are defined for the flag 6442 The following values are defined for the flags field to define the
8610 current state of the entry:: 6443 current state of the entry::
8611 6444
8612 #define KVM_DIRTY_GFN_F_DIRTY BIT 6445 #define KVM_DIRTY_GFN_F_DIRTY BIT(0)
8613 #define KVM_DIRTY_GFN_F_RESET BIT 6446 #define KVM_DIRTY_GFN_F_RESET BIT(1)
8614 #define KVM_DIRTY_GFN_F_MASK 0x3 6447 #define KVM_DIRTY_GFN_F_MASK 0x3
8615 6448
8616 Userspace should call KVM_ENABLE_CAP ioctl ri 6449 Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
8617 ioctl to enable this capability for the new g 6450 ioctl to enable this capability for the new guest and set the size of
8618 the rings. Enabling the capability is only a 6451 the rings. Enabling the capability is only allowed before creating any
8619 vCPU, and the size of the ring must be a powe 6452 vCPU, and the size of the ring must be a power of two. The larger the
8620 ring buffer, the less likely the ring is full 6453 ring buffer, the less likely the ring is full and the VM is forced to
8621 exit to userspace. The optimal size depends o 6454 exit to userspace. The optimal size depends on the workload, but it is
8622 recommended that it be at least 64 KiB (4096 6455 recommended that it be at least 64 KiB (4096 entries).
8623 6456
8624 Just like for dirty page bitmaps, the buffer 6457 Just like for dirty page bitmaps, the buffer tracks writes to
8625 all user memory regions for which the KVM_MEM 6458 all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
8626 set in KVM_SET_USER_MEMORY_REGION. Once a me 6459 set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered
8627 with the flag set, userspace can start harves 6460 with the flag set, userspace can start harvesting dirty pages from the
8628 ring buffer. 6461 ring buffer.
8629 6462
8630 An entry in the ring buffer can be unused (fl 6463 An entry in the ring buffer can be unused (flag bits ``00``),
8631 dirty (flag bits ``01``) or harvested (flag b 6464 dirty (flag bits ``01``) or harvested (flag bits ``1X``). The
8632 state machine for the entry is as follows:: 6465 state machine for the entry is as follows::
8633 6466
8634 dirtied harvested re 6467 dirtied harvested reset
8635 00 -----------> 01 -------------> 1X --- 6468 00 -----------> 01 -------------> 1X -------+
8636 ^ 6469 ^ |
8637 | 6470 | |
8638 +-------------------------------------- 6471 +------------------------------------------+
8639 6472
8640 To harvest the dirty pages, userspace accesse !! 6473 To harvest the dirty pages, userspace accesses the mmaped ring buffer
8641 to read the dirty GFNs. If the flags has the 6474 to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage
8642 the RESET bit must be cleared), then it means 6475 the RESET bit must be cleared), then it means this GFN is a dirty GFN.
8643 The userspace should harvest this GFN and mar 6476 The userspace should harvest this GFN and mark the flags from state
8644 ``01b`` to ``1Xb`` (bit 0 will be ignored by 6477 ``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
8645 to show that this GFN is harvested and waitin 6478 to show that this GFN is harvested and waiting for a reset), and move
8646 on to the next GFN. The userspace should con 6479 on to the next GFN. The userspace should continue to do this until the
8647 flags of a GFN have the DIRTY bit cleared, me 6480 flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
8648 all the dirty GFNs that were available. 6481 all the dirty GFNs that were available.
8649 6482
8650 Note that on weakly ordered architectures, us <<
8651 ring buffer (and more specifically the 'flags <<
8652 using load-acquire/store-release accessors wh <<
8653 other memory barrier that will ensure this or <<
8654 <<
8655 It's not necessary for userspace to harvest t 6483 It's not necessary for userspace to harvest the all dirty GFNs at once.
8656 However it must collect the dirty GFNs in seq 6484 However it must collect the dirty GFNs in sequence, i.e., the userspace
8657 program cannot skip one dirty GFN to collect 6485 program cannot skip one dirty GFN to collect the one next to it.
8658 6486
8659 After processing one or more entries in the r 6487 After processing one or more entries in the ring buffer, userspace
8660 calls the VM ioctl KVM_RESET_DIRTY_RINGS to n 6488 calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
8661 it, so that the kernel will reprotect those c 6489 it, so that the kernel will reprotect those collected GFNs.
8662 Therefore, the ioctl must be called *before* 6490 Therefore, the ioctl must be called *before* reading the content of
8663 the dirty pages. 6491 the dirty pages.
8664 6492
8665 The dirty ring can get full. When it happens 6493 The dirty ring can get full. When it happens, the KVM_RUN of the
8666 vcpu will return with exit reason KVM_EXIT_DI 6494 vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
8667 6495
8668 The dirty ring interface has a major differen 6496 The dirty ring interface has a major difference comparing to the
8669 KVM_GET_DIRTY_LOG interface in that, when rea 6497 KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
8670 userspace, it's still possible that the kerne 6498 userspace, it's still possible that the kernel has not yet flushed the
8671 processor's dirty page buffers into the kerne 6499 processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
8672 flushing is done by the KVM_GET_DIRTY_LOG ioc 6500 flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one
8673 needs to kick the vcpu out of KVM_RUN using a 6501 needs to kick the vcpu out of KVM_RUN using a signal. The resulting
8674 vmexit ensures that all dirty GFNs are flushe 6502 vmexit ensures that all dirty GFNs are flushed to the dirty rings.
8675 6503
8676 NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the o !! 6504 NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding
8677 should be exposed by weakly ordered architect !! 6505 ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls
8678 the additional memory ordering requirements i !! 6506 KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling
8679 reading the state of an entry and mutating it !! 6507 KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual
8680 Architecture with TSO-like ordering (such as !! 6508 machine will switch to ring-buffer dirty page tracking and further
8681 expose both KVM_CAP_DIRTY_LOG_RING and KVM_CA !! 6509 KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail.
8682 to userspace. <<
8683 <<
8684 After enabling the dirty rings, the userspace <<
8685 capability of KVM_CAP_DIRTY_LOG_RING_WITH_BIT <<
8686 ring structures can be backed by per-slot bit <<
8687 advertised, it means the architecture can dir <<
8688 vcpu/ring context, so that some of the dirty <<
8689 maintained in the bitmap structure. KVM_CAP_D <<
8690 can't be enabled if the capability of KVM_CAP <<
8691 hasn't been enabled, or any memslot has been <<
8692 <<
8693 Note that the bitmap here is only a backup of <<
8694 use of the ring and bitmap combination is onl <<
8695 only a very small amount of memory that is di <<
8696 context. Otherwise, the stand-alone per-slot <<
8697 be considered. <<
8698 <<
8699 To collect dirty bits in the backup bitmap, u <<
8700 KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG <<
8701 the generation of the dirty bits is done in a <<
8702 the dirty bitmap should be the very last thin <<
8703 considering the state as complete. VMM needs <<
8704 state is final and avoid missing dirty pages <<
8705 after the bitmap collection. <<
8706 <<
8707 NOTE: Multiple examples of using the backup b <<
8708 tables through command KVM_DEV_ARM_{VGIC_GRP_ <<
8709 KVM device "kvm-arm-vgic-its". (2) restore vg <<
8710 command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_RESTO <<
8711 "kvm-arm-vgic-its". VGICv3 LPI pending status <<
8712 vgic3 pending table through KVM_DEV_ARM_VGIC_ <<
8713 command on KVM device "kvm-arm-vgic-v3". <<
8714 <<
8715 8.30 KVM_CAP_XEN_HVM <<
8716 -------------------- <<
8717 <<
8718 :Architectures: x86 <<
8719 <<
8720 This capability indicates the features that X <<
8721 PVHVM guests. Valid flags are:: <<
8722 <<
8723 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR <<
8724 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL <<
8725 #define KVM_XEN_HVM_CONFIG_SHARED_INFO <<
8726 #define KVM_XEN_HVM_CONFIG_RUNSTATE <<
8727 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL <<
8728 #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND <<
8729 #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_ <<
8730 #define KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNST <<
8731 <<
8732 The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag ind <<
8733 ioctl is available, for the guest to set its <<
8734 <<
8735 If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also <<
8736 provided in the flags to KVM_XEN_HVM_CONFIG, <<
8737 contents, to request that KVM generate hyperc <<
8738 and also enable interception of guest hyperca <<
8739 <<
8740 The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indic <<
8741 KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, K <<
8742 KVM_XEN_VCPU_GET_ATTR ioctls, as well as the <<
8743 for event channel upcalls when the evtchn_upc <<
8744 vcpu_info is set. <<
8745 <<
8746 The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicate <<
8747 features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR <<
8748 supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XE <<
8749 <<
8750 The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag ind <<
8751 of the type KVM_IRQ_ROUTING_XEN_EVTCHN are su <<
8752 field set to indicate 2 level event channel d <<
8753 <<
8754 The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indic <<
8755 injecting event channel events directly into <<
8756 KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indica <<
8757 KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attr <<
8758 KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_V <<
8759 related to event channel delivery, timers, an <<
8760 interception. <<
8761 <<
8762 The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG f <<
8763 the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG at <<
8764 and KVM_XEN_GET_ATTR ioctls. This controls wh <<
8765 XEN_RUNSTATE_UPDATE flag in guest memory mapp <<
8766 updates of the runstate information. Note tha <<
8767 the RUNSTATE feature above, but not the RUNST <<
8768 always set the XEN_RUNSTATE_UPDATE flag when <<
8769 which is perhaps counterintuitive. When this <<
8770 behave more correctly, not using the XEN_RUNS <<
8771 specifically enabled (by the guest making the <<
8772 to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDA <<
8773 <<
8774 The KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE f <<
8775 clearing the PVCLOCK_TSC_STABLE_BIT flag in X <<
8776 done when the KVM_CAP_XEN_HVM ioctl sets the <<
8777 KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag. <<
8778 <<
8779 8.31 KVM_CAP_PPC_MULTITCE <<
8780 ------------------------- <<
8781 <<
8782 :Capability: KVM_CAP_PPC_MULTITCE <<
8783 :Architectures: ppc <<
8784 :Type: vm <<
8785 <<
8786 This capability means the kernel is capable o <<
8787 H_PUT_TCE_INDIRECT and H_STUFF_TCE without pa <<
8788 space. This significantly accelerates DMA ope <<
8789 User space should expect that its handlers fo <<
8790 are not going to be called if user space prev <<
8791 in KVM (via KVM_CREATE_SPAPR_TCE or similar c <<
8792 <<
8793 In order to enable H_PUT_TCE_INDIRECT and H_S <<
8794 user space might have to advertise it for the <<
8795 IBM pSeries (sPAPR) guest starts using them i <<
8796 present in the "ibm,hypertas-functions" devic <<
8797 <<
8798 The hypercalls mentioned above may or may not <<
8799 in the kernel based fast path. If they can no <<
8800 they will get passed on to user space. So use <<
8801 an implementation for these despite the in ke <<
8802 <<
8803 This capability is always enabled. <<
8804 <<
8805 8.32 KVM_CAP_PTP_KVM <<
8806 -------------------- <<
8807 <<
8808 :Architectures: arm64 <<
8809 <<
8810 This capability indicates that the KVM virtua <<
8811 supported in the host. A VMM can check whethe <<
8812 available to the guest on migration. <<
8813 <<
8814 8.33 KVM_CAP_HYPERV_ENFORCE_CPUID <<
8815 --------------------------------- <<
8816 <<
8817 Architectures: x86 <<
8818 <<
8819 When enabled, KVM will disable emulated Hyper <<
8820 guest according to the bits Hyper-V CPUID fea <<
8821 currently implemented Hyper-V features are pr <<
8822 Hyper-V identification is set in the HYPERV_C <<
8823 leaf. <<
8824 <<
8825 8.34 KVM_CAP_EXIT_HYPERCALL <<
8826 --------------------------- <<
8827 <<
8828 :Capability: KVM_CAP_EXIT_HYPERCALL <<
8829 :Architectures: x86 <<
8830 :Type: vm <<
8831 <<
8832 This capability, if enabled, will cause KVM t <<
8833 with KVM_EXIT_HYPERCALL exit reason to proces <<
8834 <<
8835 Calling KVM_CHECK_EXTENSION for this capabili <<
8836 of hypercalls that can be configured to exit <<
8837 Right now, the only such hypercall is KVM_HC_ <<
8838 <<
8839 The argument to KVM_ENABLE_CAP is also a bitm <<
8840 of the result of KVM_CHECK_EXTENSION. KVM wi <<
8841 the hypercalls whose corresponding bit is in <<
8842 ENOSYS for the others. <<
8843 <<
8844 8.35 KVM_CAP_PMU_CAPABILITY <<
8845 --------------------------- <<
8846 <<
8847 :Capability: KVM_CAP_PMU_CAPABILITY <<
8848 :Architectures: x86 <<
8849 :Type: vm <<
8850 :Parameters: arg[0] is bitmask of PMU virtual <<
8851 :Returns: 0 on success, -EINVAL when arg[0] c <<
8852 <<
8853 This capability alters PMU virtualization in <<
8854 <<
8855 Calling KVM_CHECK_EXTENSION for this capabili <<
8856 PMU virtualization capabilities that can be a <<
8857 <<
8858 The argument to KVM_ENABLE_CAP is also a bitm <<
8859 PMU virtualization capabilities to be applied <<
8860 only be invoked on a VM prior to the creation <<
8861 <<
8862 At this time, KVM_PMU_CAP_DISABLE is the only <<
8863 this capability will disable PMU virtualizati <<
8864 should adjust CPUID leaf 0xA to reflect that <<
8865 <<
8866 8.36 KVM_CAP_ARM_SYSTEM_SUSPEND <<
8867 ------------------------------- <<
8868 <<
8869 :Capability: KVM_CAP_ARM_SYSTEM_SUSPEND <<
8870 :Architectures: arm64 <<
8871 :Type: vm <<
8872 <<
8873 When enabled, KVM will exit to userspace with <<
8874 type KVM_SYSTEM_EVENT_SUSPEND to process the <<
8875 <<
8876 8.37 KVM_CAP_S390_PROTECTED_DUMP <<
8877 -------------------------------- <<
8878 <<
8879 :Capability: KVM_CAP_S390_PROTECTED_DUMP <<
8880 :Architectures: s390 <<
8881 :Type: vm <<
8882 <<
8883 This capability indicates that KVM and the Ul <<
8884 PV guests. The `KVM_PV_DUMP` command is avail <<
8885 `KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_I <<
8886 dump related UV data. Also the vcpu ioctl `KV <<
8887 available and supports the `KVM_PV_DUMP_CPU` <<
8888 <<
8889 8.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES <<
8890 ------------------------------------- <<
8891 <<
8892 :Capability: KVM_CAP_VM_DISABLE_NX_HUGE_PAGES <<
8893 :Architectures: x86 <<
8894 :Type: vm <<
8895 :Parameters: arg[0] must be 0. <<
8896 :Returns: 0 on success, -EPERM if the userspa <<
8897 have CAP_SYS_BOOT, -EINVAL if args[ <<
8898 created. <<
8899 <<
8900 This capability disables the NX huge pages mi <<
8901 <<
8902 The capability has no effect if the nx_huge_p <<
8903 <<
8904 This capability may only be set before any vC <<
8905 <<
8906 8.39 KVM_CAP_S390_CPU_TOPOLOGY <<
8907 ------------------------------ <<
8908 <<
8909 :Capability: KVM_CAP_S390_CPU_TOPOLOGY <<
8910 :Architectures: s390 <<
8911 :Type: vm <<
8912 <<
8913 This capability indicates that KVM will provi <<
8914 facility which consist of the interpretation <<
8915 the function code 2 along with interception a <<
8916 PTF instruction with function codes 0 or 1 an <<
8917 instruction to the userland hypervisor. <<
8918 <<
8919 The stfle facility 11, CPU Topology facility, <<
8920 to the guest without this capability. <<
8921 <<
8922 When this capability is present, KVM provides <<
8923 on vm fd, KVM_S390_VM_CPU_TOPOLOGY. <<
8924 This new attribute allows to get, set or clea <<
8925 Topology Report (MTCR) bit of the SCA through <<
8926 structure. <<
8927 <<
8928 When getting the Modified Change Topology Rep <<
8929 must point to a byte where the value will be <<
8930 <<
8931 8.40 KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE <<
8932 --------------------------------------- <<
8933 <<
8934 :Capability: KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SI <<
8935 :Architectures: arm64 <<
8936 :Type: vm <<
8937 :Parameters: arg[0] is the new split chunk si <<
8938 :Returns: 0 on success, -EINVAL if any memslo <<
8939 <<
8940 This capability sets the chunk size used in E <<
8941 <<
8942 Eager Page Splitting improves the performance <<
8943 in live migrations) when guest memory is back <<
8944 avoids splitting huge-pages (into PAGE_SIZE p <<
8945 it eagerly when enabling dirty logging (with <<
8946 KVM_MEM_LOG_DIRTY_PAGES flag for a memory reg <<
8947 KVM_CLEAR_DIRTY_LOG. <<
8948 <<
8949 The chunk size specifies how many pages to br <<
8950 single allocation for each chunk. Bigger the <<
8951 need to be allocated ahead of time. <<
8952 <<
8953 The chunk size needs to be a valid block size <<
8954 block sizes is exposed in KVM_CAP_ARM_SUPPORT <<
8955 64-bit bitmap (each bit describing a block si <<
8956 0, to disable the eager page splitting. <<
8957 <<
8958 8.41 KVM_CAP_VM_TYPES <<
8959 --------------------- <<
8960 <<
8961 :Capability: KVM_CAP_MEMORY_ATTRIBUTES <<
8962 :Architectures: x86 <<
8963 :Type: system ioctl <<
8964 <<
8965 This capability returns a bitmap of support V <<
8966 means the VM type with value @n is supported. <<
8967 <<
8968 #define KVM_X86_DEFAULT_VM 0 <<
8969 #define KVM_X86_SW_PROTECTED_VM 1 <<
8970 #define KVM_X86_SEV_VM 2 <<
8971 #define KVM_X86_SEV_ES_VM 3 <<
8972 <<
8973 Note, KVM_X86_SW_PROTECTED_VM is currently on <<
8974 Do not use KVM_X86_SW_PROTECTED_VM for "real" <<
8975 production. The behavior and effective ABI f <<
8976 unstable. <<
8977 <<
8978 9. Known KVM API problems <<
8979 ========================= <<
8980 <<
8981 In some cases, KVM's API has some inconsisten <<
8982 that userspace need to be aware of. This sec <<
8983 these issues. <<
8984 <<
8985 Most of them are architecture specific, so th <<
8986 architecture. <<
8987 <<
8988 9.1. x86 <<
8989 -------- <<
8990 <<
8991 ``KVM_GET_SUPPORTED_CPUID`` issues <<
8992 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ <<
8993 <<
8994 In general, ``KVM_GET_SUPPORTED_CPUID`` is de <<
8995 to take its result and pass it directly to `` <<
8996 documents some cases in which that requires s <<
8997 <<
8998 Local APIC features <<
8999 ~~~~~~~~~~~~~~~~~~~ <<
9000 <<
9001 CPU[EAX=1]:ECX[21] (X2APIC) is reported by `` <<
9002 but it can only be enabled if ``KVM_CREATE_IR <<
9003 ``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are <<
9004 the local APIC. <<
9005 <<
9006 The same is true for the ``KVM_FEATURE_PV_UNH <<
9007 <<
9008 CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not repo <<
9009 It can be enabled if ``KVM_CAP_TSC_DEADLINE_T <<
9010 has enabled in-kernel emulation of the local <<
9011 <<
9012 CPU topology <<
9013 ~~~~~~~~~~~~ <<
9014 <<
9015 Several CPUID values include topology informa <<
9016 0x0b and 0x1f for Intel systems, 0x8000001e f <<
9017 versions of KVM return different values for t <<
9018 should not rely on it. Currently they return <<
9019 <<
9020 If userspace wishes to set up a guest topolog <<
9021 the values of these three leaves differ for e <<
9022 the APIC ID is found in EDX for all subleaves <<
9023 for 0x8000001e; the latter also encodes the c <<
9024 7:0 of EBX and ECX respectively. <<
9025 <<
9026 Obsolete ioctls and capabilities <<
9027 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ <<
9028 <<
9029 KVM_CAP_DISABLE_QUIRKS does not let userspace <<
9030 available. Use ``KVM_CHECK_EXTENSION(KVM_CAP <<
9031 available. <<
9032 <<
9033 Ordering of KVM_GET_*/KVM_SET_* ioctls <<
9034 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ <<
9035 <<
9036 TBD <<
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.