1 accept_memory= [MM] 1 accept_memory= [MM]
2 Format: { eager | lazy 2 Format: { eager | lazy }
3 default: lazy 3 default: lazy
4 By default, unaccepted 4 By default, unaccepted memory is accepted lazily to
5 avoid prolonged boot t 5 avoid prolonged boot times. The lazy option will add
6 some runtime overhead 6 some runtime overhead until all memory is eventually
7 accepted. In most case 7 accepted. In most cases the overhead is negligible.
8 For some workloads or 8 For some workloads or for debugging purposes
9 accept_memory=eager ca 9 accept_memory=eager can be used to accept all memory
10 at once during boot. 10 at once during boot.
11 11
12 acpi= [HW,ACPI,X86,ARM64,RIS 12 acpi= [HW,ACPI,X86,ARM64,RISCV64,EARLY]
13 Advanced Configuration 13 Advanced Configuration and Power Interface
14 Format: { force | on | 14 Format: { force | on | off | strict | noirq | rsdt |
15 copy_dsdt | 15 copy_dsdt | nospcr }
16 force -- enable ACPI i 16 force -- enable ACPI if default was off
17 on -- enable ACPI but 17 on -- enable ACPI but allow fallback to DT [arm64,riscv64]
18 off -- disable ACPI if 18 off -- disable ACPI if default was on
19 noirq -- do not use AC 19 noirq -- do not use ACPI for IRQ routing
20 strict -- Be less tole 20 strict -- Be less tolerant of platforms that are not
21 strictly ACPI 21 strictly ACPI specification compliant.
22 rsdt -- prefer RSDT ov 22 rsdt -- prefer RSDT over (default) XSDT
23 copy_dsdt -- copy DSDT 23 copy_dsdt -- copy DSDT to memory
24 nospcr -- disable cons 24 nospcr -- disable console in ACPI SPCR table as
25 default _seria 25 default _serial_ console on ARM64
26 For ARM64, ONLY "acpi= 26 For ARM64, ONLY "acpi=off", "acpi=on", "acpi=force" or
27 "acpi=nospcr" are avai 27 "acpi=nospcr" are available
28 For RISCV64, ONLY "acp 28 For RISCV64, ONLY "acpi=off", "acpi=on" or "acpi=force"
29 are available 29 are available
30 30
31 See also Documentation 31 See also Documentation/power/runtime_pm.rst, pci=noacpi
32 32
33 acpi_apic_instance= [ACPI,IOAPIC,E 33 acpi_apic_instance= [ACPI,IOAPIC,EARLY]
34 Format: <int> 34 Format: <int>
35 2: use 2nd APIC table, 35 2: use 2nd APIC table, if available
36 1,0: use 1st APIC tabl 36 1,0: use 1st APIC table
37 default: 0 37 default: 0
38 38
39 acpi_backlight= [HW,ACPI] 39 acpi_backlight= [HW,ACPI]
40 { vendor | video | nat 40 { vendor | video | native | none }
41 If set to vendor, pref 41 If set to vendor, prefer vendor-specific driver
42 (e.g. thinkpad_acpi, s 42 (e.g. thinkpad_acpi, sony_acpi, etc.) instead
43 of the ACPI video.ko d 43 of the ACPI video.ko driver.
44 If set to video, use t 44 If set to video, use the ACPI video.ko driver.
45 If set to native, use 45 If set to native, use the device's native backlight mode.
46 If set to none, disabl 46 If set to none, disable the ACPI backlight interface.
47 47
48 acpi_force_32bit_fadt_addr [ACPI,EARLY 48 acpi_force_32bit_fadt_addr [ACPI,EARLY]
49 force FADT to use 32 b 49 force FADT to use 32 bit addresses rather than the
50 64 bit X_* addresses. 50 64 bit X_* addresses. Some firmware have broken 64
51 bit addresses for forc 51 bit addresses for force ACPI ignore these and use
52 the older legacy 32 bi 52 the older legacy 32 bit addresses.
53 53
54 acpica_no_return_repair [HW, ACPI] 54 acpica_no_return_repair [HW, ACPI]
55 Disable AML predefined 55 Disable AML predefined validation mechanism
56 This mechanism can rep 56 This mechanism can repair the evaluation result to make
57 the return objects mor 57 the return objects more ACPI specification compliant.
58 This option is useful 58 This option is useful for developers to identify the
59 root cause of an AML i 59 root cause of an AML interpreter issue when the issue
60 has something to do wi 60 has something to do with the repair mechanism.
61 61
62 acpi.debug_layer= [HW,ACPI,ACPI_ 62 acpi.debug_layer= [HW,ACPI,ACPI_DEBUG]
63 acpi.debug_level= [HW,ACPI,ACPI_ 63 acpi.debug_level= [HW,ACPI,ACPI_DEBUG]
64 Format: <int> 64 Format: <int>
65 CONFIG_ACPI_DEBUG must 65 CONFIG_ACPI_DEBUG must be enabled to produce any ACPI
66 debug output. Bits in 66 debug output. Bits in debug_layer correspond to a
67 _COMPONENT in an ACPI 67 _COMPONENT in an ACPI source file, e.g.,
68 #define _COMPONENT 68 #define _COMPONENT ACPI_EVENTS
69 Bits in debug_level co 69 Bits in debug_level correspond to a level in
70 ACPI_DEBUG_PRINT state 70 ACPI_DEBUG_PRINT statements, e.g.,
71 ACPI_DEBUG_PRINT(( 71 ACPI_DEBUG_PRINT((ACPI_DB_INFO, ...
72 The debug_level mask d 72 The debug_level mask defaults to "info". See
73 Documentation/firmware 73 Documentation/firmware-guide/acpi/debug.rst for more information about
74 debug layers and level 74 debug layers and levels.
75 75
76 Enable processor drive 76 Enable processor driver info messages:
77 acpi.debug_layer=0 77 acpi.debug_layer=0x20000000
78 Enable AML "Debug" out 78 Enable AML "Debug" output, i.e., stores to the Debug
79 object while interpret 79 object while interpreting AML:
80 acpi.debug_layer=0 80 acpi.debug_layer=0xffffffff acpi.debug_level=0x2
81 Enable all messages re 81 Enable all messages related to ACPI hardware:
82 acpi.debug_layer=0 82 acpi.debug_layer=0x2 acpi.debug_level=0xffffffff
83 83
84 Some values produce so 84 Some values produce so much output that the system is
85 unusable. The "log_bu 85 unusable. The "log_buf_len" parameter may be useful
86 if you need to capture 86 if you need to capture more output.
87 87
88 acpi_enforce_resources= [ACPI] 88 acpi_enforce_resources= [ACPI]
89 { strict | lax | no } 89 { strict | lax | no }
90 Check for resource con 90 Check for resource conflicts between native drivers
91 and ACPI OperationRegi 91 and ACPI OperationRegions (SystemIO and SystemMemory
92 only). IO ports and me 92 only). IO ports and memory declared in ACPI might be
93 used by the ACPI subsy 93 used by the ACPI subsystem in arbitrary AML code and
94 can interfere with leg 94 can interfere with legacy drivers.
95 strict (default): acce 95 strict (default): access to resources claimed by ACPI
96 is denied; legacy driv 96 is denied; legacy drivers trying to access reserved
97 resources will fail to 97 resources will fail to bind to device using them.
98 lax: access to resourc 98 lax: access to resources claimed by ACPI is allowed;
99 legacy drivers trying 99 legacy drivers trying to access reserved resources
100 will bind successfully 100 will bind successfully but a warning message is logged.
101 no: ACPI OperationRegi 101 no: ACPI OperationRegions are not marked as reserved,
102 no further checks are 102 no further checks are performed.
103 103
104 acpi_force_table_verification [HW,AC 104 acpi_force_table_verification [HW,ACPI,EARLY]
105 Enable table checksum 105 Enable table checksum verification during early stage.
106 By default, this is di 106 By default, this is disabled due to x86 early mapping
107 size limitation. 107 size limitation.
108 108
109 acpi_irq_balance [HW,ACPI] 109 acpi_irq_balance [HW,ACPI]
110 ACPI will balance acti 110 ACPI will balance active IRQs
111 default in APIC mode 111 default in APIC mode
112 112
113 acpi_irq_nobalance [HW,ACPI] 113 acpi_irq_nobalance [HW,ACPI]
114 ACPI will not move act 114 ACPI will not move active IRQs (default)
115 default in PIC mode 115 default in PIC mode
116 116
117 acpi_irq_isa= [HW,ACPI] If irq_balan 117 acpi_irq_isa= [HW,ACPI] If irq_balance, mark listed IRQs used by ISA
118 Format: <irq>,<irq>... 118 Format: <irq>,<irq>...
119 119
120 acpi_irq_pci= [HW,ACPI] If irq_balan 120 acpi_irq_pci= [HW,ACPI] If irq_balance, clear listed IRQs for
121 use by PCI 121 use by PCI
122 Format: <irq>,<irq>... 122 Format: <irq>,<irq>...
123 123
124 acpi_mask_gpe= [HW,ACPI] 124 acpi_mask_gpe= [HW,ACPI]
125 Due to the existence o 125 Due to the existence of _Lxx/_Exx, some GPEs triggered
126 by unsupported hardwar 126 by unsupported hardware/firmware features can result in
127 GPE floodings that can 127 GPE floodings that cannot be automatically disabled by
128 the GPE dispatcher. 128 the GPE dispatcher.
129 This facility can be u 129 This facility can be used to prevent such uncontrolled
130 GPE floodings. 130 GPE floodings.
131 Format: <byte> or <bit 131 Format: <byte> or <bitmap-list>
132 132
133 acpi_no_auto_serialize [HW,ACPI] 133 acpi_no_auto_serialize [HW,ACPI]
134 Disable auto-serializa 134 Disable auto-serialization of AML methods
135 AML control methods th 135 AML control methods that contain the opcodes to create
136 named objects will be 136 named objects will be marked as "Serialized" by the
137 auto-serialization fea 137 auto-serialization feature.
138 This feature is enable 138 This feature is enabled by default.
139 This option allows to 139 This option allows to turn off the feature.
140 140
141 acpi_no_memhotplug [ACPI] Disable memo 141 acpi_no_memhotplug [ACPI] Disable memory hotplug. Useful for kdump
142 kernels. 142 kernels.
143 143
144 acpi_no_static_ssdt [HW,ACPI,EARLY 144 acpi_no_static_ssdt [HW,ACPI,EARLY]
145 Disable installation o 145 Disable installation of static SSDTs at early boot time
146 By default, SSDTs cont 146 By default, SSDTs contained in the RSDT/XSDT will be
147 installed automaticall 147 installed automatically and they will appear under
148 /sys/firmware/acpi/tab 148 /sys/firmware/acpi/tables.
149 This option turns off 149 This option turns off this feature.
150 Note that specifying t 150 Note that specifying this option does not affect
151 dynamic table installa 151 dynamic table installation which will install SSDT
152 tables to /sys/firmwar 152 tables to /sys/firmware/acpi/tables/dynamic.
153 153
154 acpi_no_watchdog [HW,ACPI,WDT] 154 acpi_no_watchdog [HW,ACPI,WDT]
155 Ignore the ACPI-based 155 Ignore the ACPI-based watchdog interface (WDAT) and let
156 a native driver contro 156 a native driver control the watchdog device instead.
157 157
158 acpi_rsdp= [ACPI,EFI,KEXEC,EARLY] 158 acpi_rsdp= [ACPI,EFI,KEXEC,EARLY]
159 Pass the RSDP address 159 Pass the RSDP address to the kernel, mostly used
160 on machines running EF 160 on machines running EFI runtime service to boot the
161 second kernel for kdum 161 second kernel for kdump.
162 162
163 acpi_os_name= [HW,ACPI] Tell ACPI BI 163 acpi_os_name= [HW,ACPI] Tell ACPI BIOS the name of the OS
164 Format: To spoof as Wi 164 Format: To spoof as Windows 98: ="Microsoft Windows"
165 165
166 acpi_rev_override [ACPI] Override the 166 acpi_rev_override [ACPI] Override the _REV object to return 5 (instead
167 of 2 which is mandated 167 of 2 which is mandated by ACPI 6) as the supported ACPI
168 specification revision 168 specification revision (when using this switch, it may
169 be necessary to carry 169 be necessary to carry out a cold reboot _twice_ in a
170 row to make it take ef 170 row to make it take effect on the platform firmware).
171 171
172 acpi_osi= [HW,ACPI] Modify list 172 acpi_osi= [HW,ACPI] Modify list of supported OS interface strings
173 acpi_osi="string1" 173 acpi_osi="string1" # add string1
174 acpi_osi="!string2" 174 acpi_osi="!string2" # remove string2
175 acpi_osi=!* 175 acpi_osi=!* # remove all strings
176 acpi_osi=! 176 acpi_osi=! # disable all built-in OS vendor
177 177 strings
178 acpi_osi=!! 178 acpi_osi=!! # enable all built-in OS vendor
179 179 strings
180 acpi_osi= 180 acpi_osi= # disable all strings
181 181
182 'acpi_osi=!' can be us 182 'acpi_osi=!' can be used in combination with single or
183 multiple 'acpi_osi="st 183 multiple 'acpi_osi="string1"' to support specific OS
184 vendor string(s). Not 184 vendor string(s). Note that such command can only
185 affect the default sta 185 affect the default state of the OS vendor strings, thus
186 it cannot affect the d 186 it cannot affect the default state of the feature group
187 strings and the curren 187 strings and the current state of the OS vendor strings,
188 specifying it multiple 188 specifying it multiple times through kernel command line
189 is meaningless. This 189 is meaningless. This command is useful when one do not
190 care about the state o 190 care about the state of the feature group strings which
191 should be controlled b 191 should be controlled by the OSPM.
192 Examples: 192 Examples:
193 1. 'acpi_osi=! acpi_ 193 1. 'acpi_osi=! acpi_osi="Windows 2000"' is equivalent
194 to 'acpi_osi="Win 194 to 'acpi_osi="Windows 2000" acpi_osi=!', they all
195 can make '_OSI("W 195 can make '_OSI("Windows 2000")' TRUE.
196 196
197 'acpi_osi=' cannot be 197 'acpi_osi=' cannot be used in combination with other
198 'acpi_osi=' command li 198 'acpi_osi=' command lines, the _OSI method will not
199 exist in the ACPI name 199 exist in the ACPI namespace. NOTE that such command can
200 only affect the _OSI s 200 only affect the _OSI support state, thus specifying it
201 multiple times through 201 multiple times through kernel command line is also
202 meaningless. 202 meaningless.
203 Examples: 203 Examples:
204 1. 'acpi_osi=' can m 204 1. 'acpi_osi=' can make 'CondRefOf(_OSI, Local1)'
205 FALSE. 205 FALSE.
206 206
207 'acpi_osi=!*' can be u 207 'acpi_osi=!*' can be used in combination with single or
208 multiple 'acpi_osi="st 208 multiple 'acpi_osi="string1"' to support specific
209 string(s). Note that 209 string(s). Note that such command can affect the
210 current state of both 210 current state of both the OS vendor strings and the
211 feature group strings, 211 feature group strings, thus specifying it multiple times
212 through kernel command 212 through kernel command line is meaningful. But it may
213 still not able to affe 213 still not able to affect the final state of a string if
214 there are quirks relat 214 there are quirks related to this string. This command
215 is useful when one wan 215 is useful when one want to control the state of the
216 feature group strings 216 feature group strings to debug BIOS issues related to
217 the OSPM features. 217 the OSPM features.
218 Examples: 218 Examples:
219 1. 'acpi_osi="Module 219 1. 'acpi_osi="Module Device" acpi_osi=!*' can make
220 '_OSI("Module Dev 220 '_OSI("Module Device")' FALSE.
221 2. 'acpi_osi=!* acpi 221 2. 'acpi_osi=!* acpi_osi="Module Device"' can make
222 '_OSI("Module Dev 222 '_OSI("Module Device")' TRUE.
223 3. 'acpi_osi=! acpi_ 223 3. 'acpi_osi=! acpi_osi=!* acpi_osi="Windows 2000"' is
224 equivalent to 224 equivalent to
225 'acpi_osi=!* acpi 225 'acpi_osi=!* acpi_osi=! acpi_osi="Windows 2000"'
226 and 226 and
227 'acpi_osi=!* acpi 227 'acpi_osi=!* acpi_osi="Windows 2000" acpi_osi=!',
228 they all will mak 228 they all will make '_OSI("Windows 2000")' TRUE.
229 229
230 acpi_pm_good [X86] 230 acpi_pm_good [X86]
231 Override the pmtimer b 231 Override the pmtimer bug detection: force the kernel
232 to assume that this ma 232 to assume that this machine's pmtimer latches its value
233 and always returns goo 233 and always returns good values.
234 234
235 acpi_sci= [HW,ACPI,EARLY] ACPI S 235 acpi_sci= [HW,ACPI,EARLY] ACPI System Control Interrupt trigger mode
236 Format: { level | edge 236 Format: { level | edge | high | low }
237 237
238 acpi_skip_timer_override [HW,ACPI,EARL 238 acpi_skip_timer_override [HW,ACPI,EARLY]
239 Recognize and ignore I 239 Recognize and ignore IRQ0/pin2 Interrupt Override.
240 For broken nForce2 BIO 240 For broken nForce2 BIOS resulting in XT-PIC timer.
241 241
242 acpi_sleep= [HW,ACPI] Sleep option 242 acpi_sleep= [HW,ACPI] Sleep options
243 Format: { s3_bios, s3_ 243 Format: { s3_bios, s3_mode, s3_beep, s4_hwsig,
244 s4_nohwsig, 244 s4_nohwsig, old_ordering, nonvs,
245 sci_force_en 245 sci_force_enable, nobl }
246 See Documentation/powe 246 See Documentation/power/video.rst for information on
247 s3_bios and s3_mode. 247 s3_bios and s3_mode.
248 s3_beep is for debuggi 248 s3_beep is for debugging; it makes the PC's speaker beep
249 as soon as the kernel' 249 as soon as the kernel's real-mode entry point is called.
250 s4_hwsig causes the ke 250 s4_hwsig causes the kernel to check the ACPI hardware
251 signature during resum 251 signature during resume from hibernation, and gracefully
252 refuse to resume if it 252 refuse to resume if it has changed. This complies with
253 the ACPI specification 253 the ACPI specification but not with reality, since
254 Windows does not do th 254 Windows does not do this and many laptops do change it
255 on docking. So the def 255 on docking. So the default behaviour is to allow resume
256 and simply warn when t 256 and simply warn when the signature changes, unless the
257 s4_hwsig option is ena 257 s4_hwsig option is enabled.
258 s4_nohwsig prevents AC 258 s4_nohwsig prevents ACPI hardware signature from being
259 used (or even warned a 259 used (or even warned about) during resume.
260 old_ordering causes th 260 old_ordering causes the ACPI 1.0 ordering of the _PTS
261 control method, with r 261 control method, with respect to putting devices into
262 low power states, to b 262 low power states, to be enforced (the ACPI 2.0 ordering
263 of _PTS is used by def 263 of _PTS is used by default).
264 nonvs prevents the ker 264 nonvs prevents the kernel from saving/restoring the
265 ACPI NVS memory during 265 ACPI NVS memory during suspend/hibernation and resume.
266 sci_force_enable cause 266 sci_force_enable causes the kernel to set SCI_EN directly
267 on resume from S1/S3 ( 267 on resume from S1/S3 (which is against the ACPI spec,
268 but some broken system 268 but some broken systems don't work without it).
269 nobl causes the intern 269 nobl causes the internal blacklist of systems known to
270 behave incorrectly in 270 behave incorrectly in some ways with respect to system
271 suspend and resume to 271 suspend and resume to be ignored (use wisely).
272 272
273 acpi_use_timer_override [HW,ACPI,EARLY 273 acpi_use_timer_override [HW,ACPI,EARLY]
274 Use timer override. Fo 274 Use timer override. For some broken Nvidia NF5 boards
275 that require a timer o 275 that require a timer override, but don't have HPET
276 276
277 add_efi_memmap [EFI,X86,EARLY] Includ 277 add_efi_memmap [EFI,X86,EARLY] Include EFI memory map in
278 kernel's map of availa 278 kernel's map of available physical RAM.
279 279
280 agp= [AGP] 280 agp= [AGP]
281 { off | try_unsupporte 281 { off | try_unsupported }
282 off: disable AGP suppo 282 off: disable AGP support
283 try_unsupported: try t 283 try_unsupported: try to drive unsupported chipsets
284 (may crash com 284 (may crash computer or cause data corruption)
285 285
286 ALSA [HW,ALSA] 286 ALSA [HW,ALSA]
287 See Documentation/soun 287 See Documentation/sound/alsa-configuration.rst
288 288
289 alignment= [KNL,ARM] 289 alignment= [KNL,ARM]
290 Allow the default user 290 Allow the default userspace alignment fault handler
291 behaviour to be specif 291 behaviour to be specified. Bit 0 enables warnings,
292 bit 1 enables fixups, 292 bit 1 enables fixups, and bit 2 sends a segfault.
293 293
294 align_va_addr= [X86-64] 294 align_va_addr= [X86-64]
295 Align virtual addresse 295 Align virtual addresses by clearing slice [14:12] when
296 allocating a VMA at pr 296 allocating a VMA at process creation time. This option
297 gives you up to 3% per 297 gives you up to 3% performance improvement on AMD F15h
298 machines (where it is 298 machines (where it is enabled by default) for a
299 CPU-intensive style be 299 CPU-intensive style benchmark, and it can vary highly in
300 a microbenchmark depen 300 a microbenchmark depending on workload and compiler.
301 301
302 32: only for 32-bit pr 302 32: only for 32-bit processes
303 64: only for 64-bit pr 303 64: only for 64-bit processes
304 on: enable for both 32 304 on: enable for both 32- and 64-bit processes
305 off: disable for both 305 off: disable for both 32- and 64-bit processes
306 306
307 alloc_snapshot [FTRACE] 307 alloc_snapshot [FTRACE]
308 Allocate the ftrace sn 308 Allocate the ftrace snapshot buffer on boot up when the
309 main buffer is allocat 309 main buffer is allocated. This is handy if debugging
310 and you need to use tr 310 and you need to use tracing_snapshot() on boot up, and
311 do not want to use tra 311 do not want to use tracing_snapshot_alloc() as it needs
312 to be done where GFP_K 312 to be done where GFP_KERNEL allocations are allowed.
313 313
314 allow_mismatched_32bit_el0 [ARM64,EARL 314 allow_mismatched_32bit_el0 [ARM64,EARLY]
315 Allow execve() of 32-b 315 Allow execve() of 32-bit applications and setting of the
316 PER_LINUX32 personalit 316 PER_LINUX32 personality on systems where only a strict
317 subset of the CPUs sup 317 subset of the CPUs support 32-bit EL0. When this
318 parameter is present, 318 parameter is present, the set of CPUs supporting 32-bit
319 EL0 is indicated by /s 319 EL0 is indicated by /sys/devices/system/cpu/aarch32_el0
320 and hot-unplug operati 320 and hot-unplug operations may be restricted.
321 321
322 See Documentation/arch 322 See Documentation/arch/arm64/asymmetric-32bit.rst for more
323 information. 323 information.
324 324
325 amd_iommu= [HW,X86-64] 325 amd_iommu= [HW,X86-64]
326 Pass parameters to the 326 Pass parameters to the AMD IOMMU driver in the system.
327 Possible values are: 327 Possible values are:
328 fullflush - Deprecated 328 fullflush - Deprecated, equivalent to iommu.strict=1
329 off - do not ini 329 off - do not initialize any AMD IOMMU found in
330 the system 330 the system
331 force_isolation - Forc 331 force_isolation - Force device isolation for all
332 devi 332 devices. The IOMMU driver is not
333 allo 333 allowed anymore to lift isolation
334 requ 334 requirements as needed. This option
335 does 335 does not override iommu=pt
336 force_enable - Force e 336 force_enable - Force enable the IOMMU on platforms known
337 to be b 337 to be buggy with IOMMU enabled. Use this
338 option 338 option with care.
339 pgtbl_v1 - Use v1 339 pgtbl_v1 - Use v1 page table for DMA-API (Default).
340 pgtbl_v2 - Use v2 340 pgtbl_v2 - Use v2 page table for DMA-API.
341 irtcachedis - Disable 341 irtcachedis - Disable Interrupt Remapping Table (IRT) caching.
342 342
343 amd_iommu_dump= [HW,X86-64] 343 amd_iommu_dump= [HW,X86-64]
344 Enable AMD IOMMU drive 344 Enable AMD IOMMU driver option to dump the ACPI table
345 for AMD IOMMU. With th 345 for AMD IOMMU. With this option enabled, AMD IOMMU
346 driver will print ACPI 346 driver will print ACPI tables for AMD IOMMU during
347 IOMMU initialization. 347 IOMMU initialization.
348 348
349 amd_iommu_intr= [HW,X86-64] 349 amd_iommu_intr= [HW,X86-64]
350 Specifies one of the f 350 Specifies one of the following AMD IOMMU interrupt
351 remapping modes: 351 remapping modes:
352 legacy - Use legac 352 legacy - Use legacy interrupt remapping mode.
353 vapic - Use virtu 353 vapic - Use virtual APIC mode, which allows IOMMU
354 to inject 354 to inject interrupts directly into guest.
355 This mode 355 This mode requires kvm-amd.avic=1.
356 (Default 356 (Default when IOMMU HW support is present.)
357 357
358 amd_pstate= [X86,EARLY] 358 amd_pstate= [X86,EARLY]
359 disable 359 disable
360 Do not enable amd_ps 360 Do not enable amd_pstate as the default
361 scaling driver for t 361 scaling driver for the supported processors
362 passive 362 passive
363 Use amd_pstate with 363 Use amd_pstate with passive mode as a scaling driver.
364 In this mode autonom 364 In this mode autonomous selection is disabled.
365 Driver requests a de 365 Driver requests a desired performance level and platform
366 tries to match the s 366 tries to match the same performance level if it is
367 satisfied by guarant 367 satisfied by guaranteed performance level.
368 active 368 active
369 Use amd_pstate_epp d 369 Use amd_pstate_epp driver instance as the scaling driver,
370 driver provides a hi 370 driver provides a hint to the hardware if software wants
371 to bias toward perfo 371 to bias toward performance (0x0) or energy efficiency (0xff)
372 to the CPPC firmware 372 to the CPPC firmware. then CPPC power algorithm will
373 calculate the runtim 373 calculate the runtime workload and adjust the realtime cores
374 frequency. 374 frequency.
375 guided 375 guided
376 Activate guided auto 376 Activate guided autonomous mode. Driver requests minimum and
377 maximum performance 377 maximum performance level and the platform autonomously
378 selects a performanc 378 selects a performance level in this range and appropriate
379 to the current workl 379 to the current workload.
380 380
381 amd_prefcore= 381 amd_prefcore=
382 [X86] 382 [X86]
383 disable 383 disable
384 Disable amd-pstate p 384 Disable amd-pstate preferred core.
385 385
386 amijoy.map= [HW,JOY] Amiga joystic 386 amijoy.map= [HW,JOY] Amiga joystick support
387 Map of devices attache 387 Map of devices attached to JOY0DAT and JOY1DAT
388 Format: <a>,<b> 388 Format: <a>,<b>
389 See also Documentation 389 See also Documentation/input/joydev/joystick.rst
390 390
391 analog.map= [HW,JOY] Analog joysti 391 analog.map= [HW,JOY] Analog joystick and gamepad support
392 Specifies type or capa 392 Specifies type or capabilities of an analog joystick
393 connected to one of 16 393 connected to one of 16 gameports
394 Format: <type1>,<type2 394 Format: <type1>,<type2>,..<type16>
395 395
396 apc= [HW,SPARC] 396 apc= [HW,SPARC]
397 Power management funct 397 Power management functions (SPARCstation-4/5 + deriv.)
398 Format: noidle 398 Format: noidle
399 Disable APC CPU standb 399 Disable APC CPU standby support. SPARCstation-Fox does
400 not play well with APC 400 not play well with APC CPU idle - disable it if you have
401 APC and your system cr 401 APC and your system crashes randomly.
402 402
403 apic= [APIC,X86,EARLY] Advan 403 apic= [APIC,X86,EARLY] Advanced Programmable Interrupt Controller
404 Change the output verb 404 Change the output verbosity while booting
405 Format: { quiet (defau 405 Format: { quiet (default) | verbose | debug }
406 Change the amount of d 406 Change the amount of debugging information output
407 when initialising the 407 when initialising the APIC and IO-APIC components.
408 For X86-32, this can a 408 For X86-32, this can also be used to specify an APIC
409 driver name. 409 driver name.
410 Format: apic=driver_na 410 Format: apic=driver_name
411 Examples: apic=bigsmp 411 Examples: apic=bigsmp
412 412
413 apic_extnmi= [APIC,X86,EARLY] Exter 413 apic_extnmi= [APIC,X86,EARLY] External NMI delivery setting
414 Format: { bsp (default 414 Format: { bsp (default) | all | none }
415 bsp: External NMI is 415 bsp: External NMI is delivered only to CPU 0
416 all: External NMIs ar 416 all: External NMIs are broadcast to all CPUs as a
417 backup of CPU 0 417 backup of CPU 0
418 none: External NMI is 418 none: External NMI is masked for all CPUs. This is
419 useful so that a 419 useful so that a dump capture kernel won't be
420 shot down by NMI 420 shot down by NMI
421 421
422 autoconf= [IPV6] 422 autoconf= [IPV6]
423 See Documentation/netw 423 See Documentation/networking/ipv6.rst.
424 424
425 apm= [APM] Advanced Power M 425 apm= [APM] Advanced Power Management
426 See header of arch/x86 426 See header of arch/x86/kernel/apm_32.c.
427 427
428 apparmor= [APPARMOR] Disable or 428 apparmor= [APPARMOR] Disable or enable AppArmor at boot time
429 Format: { "0" | "1" } 429 Format: { "0" | "1" }
430 See security/apparmor/ 430 See security/apparmor/Kconfig help text
431 0 -- disable. 431 0 -- disable.
432 1 -- enable. 432 1 -- enable.
433 Default value is set v 433 Default value is set via kernel config option.
434 434
435 arcrimi= [HW,NET] ARCnet - "RIM 435 arcrimi= [HW,NET] ARCnet - "RIM I" (entirely mem-mapped) cards
436 Format: <io>,<irq>,<no 436 Format: <io>,<irq>,<nodeID>
437 437
438 arm64.no32bit_el0 [ARM64] Unconditiona 438 arm64.no32bit_el0 [ARM64] Unconditionally disable the execution of
439 32 bit applications. 439 32 bit applications.
440 440
441 arm64.nobti [ARM64] Unconditionall 441 arm64.nobti [ARM64] Unconditionally disable Branch Target
442 Identification support 442 Identification support
443 443
444 arm64.nomops [ARM64] Unconditionall 444 arm64.nomops [ARM64] Unconditionally disable Memory Copy and Memory
445 Set instructions suppo 445 Set instructions support
446 446
447 arm64.nomte [ARM64] Unconditionall 447 arm64.nomte [ARM64] Unconditionally disable Memory Tagging Extension
448 support 448 support
449 449
450 arm64.nopauth [ARM64] Unconditionall 450 arm64.nopauth [ARM64] Unconditionally disable Pointer Authentication
451 support 451 support
452 452
453 arm64.nosme [ARM64] Unconditionall 453 arm64.nosme [ARM64] Unconditionally disable Scalable Matrix
454 Extension support 454 Extension support
455 455
456 arm64.nosve [ARM64] Unconditionall 456 arm64.nosve [ARM64] Unconditionally disable Scalable Vector
457 Extension support 457 Extension support
458 458
459 ataflop= [HW,M68k] 459 ataflop= [HW,M68k]
460 460
461 atarimouse= [HW,MOUSE] Atari Mouse 461 atarimouse= [HW,MOUSE] Atari Mouse
462 462
463 atkbd.extra= [HW] Enable extra LEDs 463 atkbd.extra= [HW] Enable extra LEDs and keys on IBM RapidAccess,
464 EzKey and similar keyb 464 EzKey and similar keyboards
465 465
466 atkbd.reset= [HW] Reset keyboard du 466 atkbd.reset= [HW] Reset keyboard during initialization
467 467
468 atkbd.set= [HW] Select keyboard c 468 atkbd.set= [HW] Select keyboard code set
469 Format: <int> (2 = AT 469 Format: <int> (2 = AT (default), 3 = PS/2)
470 470
471 atkbd.scroll= [HW] Enable scroll whe 471 atkbd.scroll= [HW] Enable scroll wheel on MS Office and similar
472 keyboards 472 keyboards
473 473
474 atkbd.softraw= [HW] Choose between sy 474 atkbd.softraw= [HW] Choose between synthetic and real raw mode
475 Format: <bool> (0 = re 475 Format: <bool> (0 = real, 1 = synthetic (default))
476 476
477 atkbd.softrepeat= [HW] 477 atkbd.softrepeat= [HW]
478 Use software keyboard 478 Use software keyboard repeat
479 479
480 audit= [KNL] Enable the audit 480 audit= [KNL] Enable the audit sub-system
481 Format: { "0" | "1" | 481 Format: { "0" | "1" | "off" | "on" }
482 0 | off - kernel audit 482 0 | off - kernel audit is disabled and can not be
483 enabled until the 483 enabled until the next reboot
484 unset - kernel audit i 484 unset - kernel audit is initialized but disabled and
485 will be fully enab 485 will be fully enabled by the userspace auditd.
486 1 | on - kernel audit 486 1 | on - kernel audit is initialized and partially
487 enabled, storing a 487 enabled, storing at most audit_backlog_limit
488 messages in RAM un 488 messages in RAM until it is fully enabled by the
489 userspace auditd. 489 userspace auditd.
490 Default: unset 490 Default: unset
491 491
492 audit_backlog_limit= [KNL] Set the aud 492 audit_backlog_limit= [KNL] Set the audit queue size limit.
493 Format: <int> (must be 493 Format: <int> (must be >=0)
494 Default: 64 494 Default: 64
495 495
496 bau= [X86_UV] Enable the BA 496 bau= [X86_UV] Enable the BAU on SGI UV. The default
497 behavior is to disable 497 behavior is to disable the BAU (i.e. bau=0).
498 Format: { "0" | "1" } 498 Format: { "0" | "1" }
499 0 - Disable the BAU. 499 0 - Disable the BAU.
500 1 - Enable the BAU. 500 1 - Enable the BAU.
501 unset - Disable the BA 501 unset - Disable the BAU.
502 502
503 baycom_epp= [HW,AX25] 503 baycom_epp= [HW,AX25]
504 Format: <io>,<mode> 504 Format: <io>,<mode>
505 505
506 baycom_par= [HW,AX25] BayCom Paral 506 baycom_par= [HW,AX25] BayCom Parallel Port AX.25 Modem
507 Format: <io>,<mode> 507 Format: <io>,<mode>
508 See header of drivers/ 508 See header of drivers/net/hamradio/baycom_par.c.
509 509
510 baycom_ser_fdx= [HW,AX25] 510 baycom_ser_fdx= [HW,AX25]
511 BayCom Serial Port AX. 511 BayCom Serial Port AX.25 Modem (Full Duplex Mode)
512 Format: <io>,<irq>,<mo 512 Format: <io>,<irq>,<mode>[,<baud>]
513 See header of drivers/ 513 See header of drivers/net/hamradio/baycom_ser_fdx.c.
514 514
515 baycom_ser_hdx= [HW,AX25] 515 baycom_ser_hdx= [HW,AX25]
516 BayCom Serial Port AX. 516 BayCom Serial Port AX.25 Modem (Half Duplex Mode)
517 Format: <io>,<irq>,<mo 517 Format: <io>,<irq>,<mode>
518 See header of drivers/ 518 See header of drivers/net/hamradio/baycom_ser_hdx.c.
519 519
520 bert_disable [ACPI] 520 bert_disable [ACPI]
521 Disable BERT OS suppor 521 Disable BERT OS support on buggy BIOSes.
522 522
523 bgrt_disable [ACPI,X86,EARLY] 523 bgrt_disable [ACPI,X86,EARLY]
524 Disable BGRT to avoid 524 Disable BGRT to avoid flickering OEM logo.
525 525
526 blkdevparts= Manual partition parsi 526 blkdevparts= Manual partition parsing of block device(s) for
527 embedded devices based 527 embedded devices based on command line input.
528 See Documentation/bloc 528 See Documentation/block/cmdline-partition.rst
529 529
530 boot_delay= [KNL,EARLY] 530 boot_delay= [KNL,EARLY]
531 Milliseconds to delay 531 Milliseconds to delay each printk during boot.
532 Only works if CONFIG_B 532 Only works if CONFIG_BOOT_PRINTK_DELAY is enabled,
533 and you may also have 533 and you may also have to specify "lpj=". Boot_delay
534 values larger than 10 534 values larger than 10 seconds (10000) are assumed
535 erroneous and ignored. 535 erroneous and ignored.
536 Format: integer 536 Format: integer
537 537
538 bootconfig [KNL,EARLY] 538 bootconfig [KNL,EARLY]
539 Extended command line 539 Extended command line options can be added to an initrd
540 and this will cause th 540 and this will cause the kernel to look for it.
541 541
542 See Documentation/admi 542 See Documentation/admin-guide/bootconfig.rst
543 543
544 bttv.card= [HW,V4L] bttv (bt848 + 544 bttv.card= [HW,V4L] bttv (bt848 + bt878 based grabber cards)
545 bttv.radio= Most important insmod 545 bttv.radio= Most important insmod options are available as
546 kernel args too. 546 kernel args too.
547 bttv.pll= See Documentation/admi 547 bttv.pll= See Documentation/admin-guide/media/bttv.rst
548 bttv.tuner= 548 bttv.tuner=
549 549
550 bulk_remove=off [PPC] This parameter 550 bulk_remove=off [PPC] This parameter disables the use of the pSeries
551 firmware feature for f 551 firmware feature for flushing multiple hpte entries
552 at a time. 552 at a time.
553 553
554 c101= [NET] Moxa C101 synchr 554 c101= [NET] Moxa C101 synchronous serial card
555 555
556 cachesize= [BUGS=X86-32] Override 556 cachesize= [BUGS=X86-32] Override level 2 CPU cache size detection.
557 Sometimes CPU hardware 557 Sometimes CPU hardware bugs make them report the cache
558 size incorrectly. The 558 size incorrectly. The kernel will attempt work arounds
559 to fix known problems, 559 to fix known problems, but for some CPUs it is not
560 possible to determine 560 possible to determine what the correct size should be.
561 This option provides a 561 This option provides an override for these situations.
562 562
563 carrier_timeout= 563 carrier_timeout=
564 [NET] Specifies amount 564 [NET] Specifies amount of time (in seconds) that
565 the kernel should wait 565 the kernel should wait for a network carrier. By default
566 it waits 120 seconds. 566 it waits 120 seconds.
567 567
568 ca_keys= [KEYS] This parameter 568 ca_keys= [KEYS] This parameter identifies a specific key(s) on
569 the system trusted key 569 the system trusted keyring to be used for certificate
570 trust validation. 570 trust validation.
571 format: { id:<keyid> | 571 format: { id:<keyid> | builtin }
572 572
573 cca= [MIPS,EARLY] Override 573 cca= [MIPS,EARLY] Override the kernel pages' cache coherency
574 algorithm. Accepted v 574 algorithm. Accepted values range from 0 to 7
575 inclusive. See arch/mi 575 inclusive. See arch/mips/include/asm/pgtable-bits.h
576 for platform specific 576 for platform specific values (SB1, Loongson3 and
577 others). 577 others).
578 578
579 ccw_timeout_log [S390] 579 ccw_timeout_log [S390]
580 See Documentation/arch 580 See Documentation/arch/s390/common_io.rst for details.
581 581
582 cgroup_disable= [KNL] Disable a partic 582 cgroup_disable= [KNL] Disable a particular controller or optional feature
583 Format: {name of the c 583 Format: {name of the controller(s) or feature(s) to disable}
584 The effects of cgroup_ 584 The effects of cgroup_disable=foo are:
585 - foo isn't auto-mount 585 - foo isn't auto-mounted if you mount all cgroups in
586 a single hierarchy 586 a single hierarchy
587 - foo isn't visible as 587 - foo isn't visible as an individually mountable
588 subsystem 588 subsystem
589 - if foo is an optiona 589 - if foo is an optional feature then the feature is
590 disabled and corresp 590 disabled and corresponding cgroup files are not
591 created 591 created
592 {Currently only "memor 592 {Currently only "memory" controller deal with this and
593 cut the overhead, othe 593 cut the overhead, others just disable the usage. So
594 only cgroup_disable=me 594 only cgroup_disable=memory is actually worthy}
595 Specifying "pressure" 595 Specifying "pressure" disables per-cgroup pressure
596 stall information acco 596 stall information accounting feature
597 597
598 cgroup_no_v1= [KNL] Disable cgroup c 598 cgroup_no_v1= [KNL] Disable cgroup controllers and named hierarchies in v1
599 Format: { { controller 599 Format: { { controller | "all" | "named" }
600 [,{ controll 600 [,{ controller | "all" | "named" }...] }
601 Like cgroup_disable, b 601 Like cgroup_disable, but only applies to cgroup v1;
602 the blacklisted contro 602 the blacklisted controllers remain available in cgroup2.
603 "all" blacklists all c 603 "all" blacklists all controllers and "named" disables
604 named mounts. Specifyi 604 named mounts. Specifying both "all" and "named" disables
605 all v1 hierarchies. 605 all v1 hierarchies.
606 606
607 cgroup_favordynmods= [KNL] Enable or D 607 cgroup_favordynmods= [KNL] Enable or Disable favordynmods.
608 Format: { "true" | "fa 608 Format: { "true" | "false" }
609 Defaults to the value 609 Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS.
610 610
611 cgroup.memory= [KNL] Pass options to 611 cgroup.memory= [KNL] Pass options to the cgroup memory controller.
612 Format: <string> 612 Format: <string>
613 nosocket -- Disable so 613 nosocket -- Disable socket memory accounting.
614 nokmem -- Disable kern 614 nokmem -- Disable kernel memory accounting.
615 nobpf -- Disable BPF m 615 nobpf -- Disable BPF memory accounting.
616 616
617 checkreqprot= [SELINUX] Set initial 617 checkreqprot= [SELINUX] Set initial checkreqprot flag value.
618 Format: { "0" | "1" } 618 Format: { "0" | "1" }
619 See security/selinux/K 619 See security/selinux/Kconfig help text.
620 0 -- check protection 620 0 -- check protection applied by kernel (includes
621 any implied ex 621 any implied execute protection).
622 1 -- check protection 622 1 -- check protection requested by application.
623 Default value is set v 623 Default value is set via a kernel config option.
624 Value can be changed a 624 Value can be changed at runtime via
625 /sys/fs/selinu 625 /sys/fs/selinux/checkreqprot.
626 Setting checkreqprot t 626 Setting checkreqprot to 1 is deprecated.
627 627
628 cio_ignore= [S390] 628 cio_ignore= [S390]
629 See Documentation/arch 629 See Documentation/arch/s390/common_io.rst for details.
630 630
631 clearcpuid=X[,X...] [X86] 631 clearcpuid=X[,X...] [X86]
632 Disable CPUID feature 632 Disable CPUID feature X for the kernel. See
633 arch/x86/include/asm/c 633 arch/x86/include/asm/cpufeatures.h for the valid bit
634 numbers X. Note the Li 634 numbers X. Note the Linux-specific bits are not necessarily
635 stable over kernel opt 635 stable over kernel options, but the vendor-specific
636 ones should be. 636 ones should be.
637 X can also be a string 637 X can also be a string as appearing in the flags: line
638 in /proc/cpuinfo which 638 in /proc/cpuinfo which does not have the above
639 instability issue. How 639 instability issue. However, not all features have names
640 in /proc/cpuinfo. 640 in /proc/cpuinfo.
641 Note that using this o 641 Note that using this option will taint your kernel.
642 Also note that user pr 642 Also note that user programs calling CPUID directly
643 or using the feature w 643 or using the feature without checking anything
644 will still see it. Thi 644 will still see it. This just prevents it from
645 being used by the kern 645 being used by the kernel or shown in /proc/cpuinfo.
646 Also note the kernel m 646 Also note the kernel might malfunction if you disable
647 some critical bits. 647 some critical bits.
648 648
649 clk_ignore_unused 649 clk_ignore_unused
650 [CLK] 650 [CLK]
651 Prevents the clock fra 651 Prevents the clock framework from automatically gating
652 clocks that have not b 652 clocks that have not been explicitly enabled by a Linux
653 device driver but are 653 device driver but are enabled in hardware at reset or
654 by the bootloader/firm 654 by the bootloader/firmware. Note that this does not
655 force such clocks to b 655 force such clocks to be always-on nor does it reserve
656 those clocks in any wa 656 those clocks in any way. This parameter is useful for
657 debug and development, 657 debug and development, but should not be needed on a
658 platform with proper d 658 platform with proper driver support. For more
659 information, see Docum 659 information, see Documentation/driver-api/clk.rst.
660 660
661 clock= [BUGS=X86-32, HW] gett 661 clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
662 [Deprecated] 662 [Deprecated]
663 Forces specified clock 663 Forces specified clocksource (if available) to be used
664 when calculating getti 664 when calculating gettimeofday(). If specified
665 clocksource is not ava 665 clocksource is not available, it defaults to PIT.
666 Format: { pit | tsc | 666 Format: { pit | tsc | cyclone | pmtmr }
667 667
668 clocksource= Override the default c 668 clocksource= Override the default clocksource
669 Format: <string> 669 Format: <string>
670 Override the default c 670 Override the default clocksource and use the clocksource
671 with the name specifie 671 with the name specified.
672 Some clocksource names 672 Some clocksource names to choose from, depending on
673 the platform: 673 the platform:
674 [all] jiffies (this is 674 [all] jiffies (this is the base, fallback clocksource)
675 [ACPI] acpi_pm 675 [ACPI] acpi_pm
676 [ARM] imx_timer1,OSTS, 676 [ARM] imx_timer1,OSTS,netx_timer,mpu_timer2,
677 pxa_timer,time 677 pxa_timer,timer3,32k_counter,timer0_1
678 [X86-32] pit,hpet,tsc; 678 [X86-32] pit,hpet,tsc;
679 scx200_hrt on 679 scx200_hrt on Geode; cyclone on IBM x440
680 [MIPS] MIPS 680 [MIPS] MIPS
681 [PARISC] cr16 681 [PARISC] cr16
682 [S390] tod 682 [S390] tod
683 [SH] SuperH 683 [SH] SuperH
684 [SPARC64] tick 684 [SPARC64] tick
685 [X86-64] hpet,tsc 685 [X86-64] hpet,tsc
686 686
687 clocksource.arm_arch_timer.evtstrm= 687 clocksource.arm_arch_timer.evtstrm=
688 [ARM,ARM64,EARLY] 688 [ARM,ARM64,EARLY]
689 Format: <bool> 689 Format: <bool>
690 Enable/disable the eve 690 Enable/disable the eventstream feature of the ARM
691 architected timer so t 691 architected timer so that code using WFE-based polling
692 loops can be debugged 692 loops can be debugged more effectively on production
693 systems. 693 systems.
694 694
695 clocksource.verify_n_cpus= [KNL] 695 clocksource.verify_n_cpus= [KNL]
696 Limit the number of CP 696 Limit the number of CPUs checked for clocksources
697 marked with CLOCK_SOUR 697 marked with CLOCK_SOURCE_VERIFY_PERCPU that
698 are marked unstable du 698 are marked unstable due to excessive skew.
699 A negative value says 699 A negative value says to check all CPUs, while
700 zero says not to check 700 zero says not to check any. Values larger than
701 nr_cpu_ids are silentl 701 nr_cpu_ids are silently truncated to nr_cpu_ids.
702 The actual CPUs are ch 702 The actual CPUs are chosen randomly, with
703 no replacement if the 703 no replacement if the same CPU is chosen twice.
704 704
705 clocksource-wdtest.holdoff= [KNL] 705 clocksource-wdtest.holdoff= [KNL]
706 Set the time in second 706 Set the time in seconds that the clocksource
707 watchdog test waits be 707 watchdog test waits before commencing its tests.
708 Defaults to zero when 708 Defaults to zero when built as a module and to
709 10 seconds when built 709 10 seconds when built into the kernel.
710 710
711 cma=nn[MG]@[start[MG][-end[MG]]] 711 cma=nn[MG]@[start[MG][-end[MG]]]
712 [KNL,CMA,EARLY] 712 [KNL,CMA,EARLY]
713 Sets the size of kerne 713 Sets the size of kernel global memory area for
714 contiguous memory allo 714 contiguous memory allocations and optionally the
715 placement constraint b 715 placement constraint by the physical address range of
716 memory allocations. A 716 memory allocations. A value of 0 disables CMA
717 altogether. For more i 717 altogether. For more information, see
718 kernel/dma/contiguous. 718 kernel/dma/contiguous.c
719 719
720 cma_pernuma=nn[MG] 720 cma_pernuma=nn[MG]
721 [KNL,CMA,EARLY] 721 [KNL,CMA,EARLY]
722 Sets the size of kerne 722 Sets the size of kernel per-numa memory area for
723 contiguous memory allo 723 contiguous memory allocations. A value of 0 disables
724 per-numa CMA altogethe 724 per-numa CMA altogether. And If this option is not
725 specified, the default 725 specified, the default value is 0.
726 With per-numa CMA enab 726 With per-numa CMA enabled, DMA users on node nid will
727 first try to allocate 727 first try to allocate buffer from the pernuma area
728 which is located in no 728 which is located in node nid, if the allocation fails,
729 they will fallback to 729 they will fallback to the global default memory area.
730 730
731 numa_cma=<node>:nn[MG][,<node>:nn[MG]] 731 numa_cma=<node>:nn[MG][,<node>:nn[MG]]
732 [KNL,CMA,EARLY] 732 [KNL,CMA,EARLY]
733 Sets the size of kerne 733 Sets the size of kernel numa memory area for
734 contiguous memory allo 734 contiguous memory allocations. It will reserve CMA
735 area for the specified 735 area for the specified node.
736 736
737 With numa CMA enabled, 737 With numa CMA enabled, DMA users on node nid will
738 first try to allocate 738 first try to allocate buffer from the numa area
739 which is located in no 739 which is located in node nid, if the allocation fails,
740 they will fallback to 740 they will fallback to the global default memory area.
741 741
742 cmo_free_hint= [PPC] Format: { yes | 742 cmo_free_hint= [PPC] Format: { yes | no }
743 Specify whether pages 743 Specify whether pages are marked as being inactive
744 when they are freed. 744 when they are freed. This is used in CMO environments
745 to determine OS memory 745 to determine OS memory pressure for page stealing by
746 a hypervisor. 746 a hypervisor.
747 Default: yes 747 Default: yes
748 748
749 coherent_pool=nn[KMG] [ARM,KNL,EARLY 749 coherent_pool=nn[KMG] [ARM,KNL,EARLY]
750 Sets the size of memor 750 Sets the size of memory pool for coherent, atomic dma
751 allocations, by defaul 751 allocations, by default set to 256K.
752 752
753 com20020= [HW,NET] ARCnet - COM2 753 com20020= [HW,NET] ARCnet - COM20020 chipset
754 Format: 754 Format:
755 <io>[,<irq>[,<nodeID>[ 755 <io>[,<irq>[,<nodeID>[,<backplane>[,<ckp>[,<timeout>]]]]]
756 756
757 com90io= [HW,NET] ARCnet - COM9 757 com90io= [HW,NET] ARCnet - COM90xx chipset (IO-mapped buffers)
758 Format: <io>[,<irq>] 758 Format: <io>[,<irq>]
759 759
760 com90xx= [HW,NET] 760 com90xx= [HW,NET]
761 ARCnet - COM90xx chips 761 ARCnet - COM90xx chipset (memory-mapped buffers)
762 Format: <io>[,<irq>[,< 762 Format: <io>[,<irq>[,<memstart>]]
763 763
764 condev= [HW,S390] console devi 764 condev= [HW,S390] console device
765 conmode= 765 conmode=
766 766
767 con3215_drop= [S390,EARLY] 3215 cons 767 con3215_drop= [S390,EARLY] 3215 console drop mode.
768 Format: y|n|Y|N|1|0 768 Format: y|n|Y|N|1|0
769 When set to true, drop 769 When set to true, drop data on the 3215 console when
770 the console buffer is 770 the console buffer is full. In this case the
771 operator using a 3270 771 operator using a 3270 terminal emulator (for example
772 x3270) does not have t 772 x3270) does not have to enter the clear key for the
773 console output to adva 773 console output to advance and the kernel to continue.
774 This leads to a much f 774 This leads to a much faster boot time when a 3270
775 terminal emulator is a 775 terminal emulator is active. If no 3270 terminal
776 emulator is used, this 776 emulator is used, this parameter has no effect.
777 777
778 console= [KNL] Output console d 778 console= [KNL] Output console device and options.
779 779
780 tty<n> Use the virtual consol 780 tty<n> Use the virtual console device <n>.
781 781
782 ttyS<n>[,options] 782 ttyS<n>[,options]
783 ttyUSB0[,options] 783 ttyUSB0[,options]
784 Use the specified seri 784 Use the specified serial port. The options are of
785 the form "bbbbpnf", wh 785 the form "bbbbpnf", where "bbbb" is the baud rate,
786 "p" is parity ("n", "o 786 "p" is parity ("n", "o", or "e"), "n" is number of
787 bits, and "f" is flow 787 bits, and "f" is flow control ("r" for RTS or
788 omit it). Default is 788 omit it). Default is "9600n8".
789 789
790 See Documentation/admi 790 See Documentation/admin-guide/serial-console.rst for more
791 information. See 791 information. See
792 Documentation/networki 792 Documentation/networking/netconsole.rst for an
793 alternative. 793 alternative.
794 794
795 <DEVNAME>:<n>.<n>[,options] 795 <DEVNAME>:<n>.<n>[,options]
796 Use the specified seri 796 Use the specified serial port on the serial core bus.
797 The addressing uses DE 797 The addressing uses DEVNAME of the physical serial port
798 device, followed by th 798 device, followed by the serial core controller instance,
799 and the serial port in 799 and the serial port instance. The options are the same
800 as documented for the 800 as documented for the ttyS addressing above.
801 801
802 The mapping of the ser 802 The mapping of the serial ports to the tty instances
803 can be viewed with: 803 can be viewed with:
804 804
805 $ ls -d /sys/bus/seria 805 $ ls -d /sys/bus/serial-base/devices/*:*.*/tty/*
806 /sys/bus/serial-base/d 806 /sys/bus/serial-base/devices/00:04:0.0/tty/ttyS0
807 807
808 In the above example, 808 In the above example, the console can be addressed with
809 console=00:04:0.0. Not 809 console=00:04:0.0. Note that a console addressed this
810 way will only get adde 810 way will only get added when the related device driver
811 is ready. The use of a 811 is ready. The use of an earlycon parameter in addition to
812 the console may be des 812 the console may be desired for console output early on.
813 813
814 uart[8250],io,<addr>[,options] 814 uart[8250],io,<addr>[,options]
815 uart[8250],mmio,<addr>[,option 815 uart[8250],mmio,<addr>[,options]
816 uart[8250],mmio16,<addr>[,opti 816 uart[8250],mmio16,<addr>[,options]
817 uart[8250],mmio32,<addr>[,opti 817 uart[8250],mmio32,<addr>[,options]
818 uart[8250],0x<addr>[,options] 818 uart[8250],0x<addr>[,options]
819 Start an early, polled 819 Start an early, polled-mode console on the 8250/16550
820 UART at the specified 820 UART at the specified I/O port or MMIO address,
821 switching to the match 821 switching to the matching ttyS device later.
822 MMIO inter-register ad 822 MMIO inter-register address stride is either 8-bit
823 (mmio), 16-bit (mmio16 823 (mmio), 16-bit (mmio16), or 32-bit (mmio32).
824 If none of [io|mmio|mm 824 If none of [io|mmio|mmio16|mmio32], <addr> is assumed
825 to be equivalent to 'm 825 to be equivalent to 'mmio'. 'options' are specified in
826 the same format descri 826 the same format described for ttyS above; if unspecified,
827 the h/w is not re-init 827 the h/w is not re-initialized.
828 828
829 hvc<n> Use the hypervisor con 829 hvc<n> Use the hypervisor console device <n>. This is for
830 both Xen and PowerPC h 830 both Xen and PowerPC hypervisors.
831 831
832 { null | "" } 832 { null | "" }
833 Use to disable console 833 Use to disable console output, i.e., to have kernel
834 console messages disca 834 console messages discarded.
835 This must be the only 835 This must be the only console= parameter used on the
836 kernel command line. 836 kernel command line.
837 837
838 If the device connected to the 838 If the device connected to the port is not a TTY but a braille
839 device, prepend "brl," before 839 device, prepend "brl," before the device type, for instance
840 console=brl,ttyS0 840 console=brl,ttyS0
841 For now, only VisioBraille is 841 For now, only VisioBraille is supported.
842 842
843 console_msg_format= 843 console_msg_format=
844 [KNL] Change console m 844 [KNL] Change console messages format
845 default 845 default
846 By default we print me 846 By default we print messages on consoles in
847 "[time stamp] text\n" 847 "[time stamp] text\n" format (time stamp may not be
848 printed, depending on 848 printed, depending on CONFIG_PRINTK_TIME or
849 `printk_time' param). 849 `printk_time' param).
850 syslog 850 syslog
851 Switch to syslog forma 851 Switch to syslog format: "<%u>[time stamp] text\n"
852 IOW, each message will 852 IOW, each message will have a facility and loglevel
853 prefix. The format is 853 prefix. The format is similar to one used by syslog()
854 syscall, or to executi 854 syscall, or to executing "dmesg -S --raw" or to reading
855 from /proc/kmsg. 855 from /proc/kmsg.
856 856
857 consoleblank= [KNL] The console blan 857 consoleblank= [KNL] The console blank (screen saver) timeout in
858 seconds. A value of 0 858 seconds. A value of 0 disables the blank timer.
859 Defaults to 0. 859 Defaults to 0.
860 860
861 coredump_filter= 861 coredump_filter=
862 [KNL] Change the defau 862 [KNL] Change the default value for
863 /proc/<pid>/coredump_f 863 /proc/<pid>/coredump_filter.
864 See also Documentation 864 See also Documentation/filesystems/proc.rst.
865 865
866 coresight_cpu_debug.enable 866 coresight_cpu_debug.enable
867 [ARM,ARM64] 867 [ARM,ARM64]
868 Format: <bool> 868 Format: <bool>
869 Enable/disable the CPU 869 Enable/disable the CPU sampling based debugging.
870 0: default value, disa 870 0: default value, disable debugging
871 1: enable debugging at 871 1: enable debugging at boot time
872 872
873 cpcihp_generic= [HW,PCI] Generic port 873 cpcihp_generic= [HW,PCI] Generic port I/O CompactPCI driver
874 Format: 874 Format:
875 <first_slot>,<last_slo 875 <first_slot>,<last_slot>,<port>,<enum_bit>[,<debug>]
876 876
877 cpuidle.off=1 [CPU_IDLE] 877 cpuidle.off=1 [CPU_IDLE]
878 disable the cpuidle su 878 disable the cpuidle sub-system
879 879
880 cpuidle.governor= 880 cpuidle.governor=
881 [CPU_IDLE] Name of the 881 [CPU_IDLE] Name of the cpuidle governor to use.
882 882
883 cpufreq.off=1 [CPU_FREQ] 883 cpufreq.off=1 [CPU_FREQ]
884 disable the cpufreq su 884 disable the cpufreq sub-system
885 885
886 cpufreq.default_governor= 886 cpufreq.default_governor=
887 [CPU_FREQ] Name of the 887 [CPU_FREQ] Name of the default cpufreq governor or
888 policy to use. This go 888 policy to use. This governor must be registered in the
889 kernel before the cpuf 889 kernel before the cpufreq driver probes.
890 890
891 cpu_init_udelay=N 891 cpu_init_udelay=N
892 [X86,EARLY] Delay for 892 [X86,EARLY] Delay for N microsec between assert and de-assert
893 of APIC INIT to start 893 of APIC INIT to start processors. This delay occurs
894 on every CPU online, s 894 on every CPU online, such as boot, and resume from suspend.
895 Default: 10000 895 Default: 10000
896 896
897 cpuhp.parallel= 897 cpuhp.parallel=
898 [SMP] Enable/disable p 898 [SMP] Enable/disable parallel bringup of secondary CPUs
899 Format: <bool> 899 Format: <bool>
900 Default is enabled if 900 Default is enabled if CONFIG_HOTPLUG_PARALLEL=y. Otherwise
901 the parameter has no e 901 the parameter has no effect.
902 902
903 crash_kexec_post_notifiers 903 crash_kexec_post_notifiers
904 Run kdump after runnin 904 Run kdump after running panic-notifiers and dumping
905 kmsg. This only for th 905 kmsg. This only for the users who doubt kdump always
906 succeeds in any situat 906 succeeds in any situation.
907 Note that this also in 907 Note that this also increases risks of kdump failure,
908 because some panic not 908 because some panic notifiers can make the crashed
909 kernel more unstable. 909 kernel more unstable.
910 910
911 crashkernel=size[KMG][@offset[KMG]] 911 crashkernel=size[KMG][@offset[KMG]]
912 [KNL,EARLY] Using kexe 912 [KNL,EARLY] Using kexec, Linux can switch to a 'crash kernel'
913 upon panic. This param 913 upon panic. This parameter reserves the physical
914 memory region [offset, 914 memory region [offset, offset + size] for that kernel
915 image. If '@offset' is 915 image. If '@offset' is omitted, then a suitable offset
916 is selected automatica 916 is selected automatically.
917 [KNL, X86-64, ARM64, R 917 [KNL, X86-64, ARM64, RISCV, LoongArch] Select a region
918 under 4G first, and fa 918 under 4G first, and fall back to reserve region above
919 4G when '@offset' hasn 919 4G when '@offset' hasn't been specified.
920 See Documentation/admi 920 See Documentation/admin-guide/kdump/kdump.rst for further details.
921 921
922 crashkernel=range1:size1[,range2:size2 922 crashkernel=range1:size1[,range2:size2,...][@offset]
923 [KNL] Same as above, b 923 [KNL] Same as above, but depends on the memory
924 in the running system. 924 in the running system. The syntax of range is
925 start-[end] where star 925 start-[end] where start and end are both
926 a memory unit (amount[ 926 a memory unit (amount[KMG]). See also
927 Documentation/admin-gu 927 Documentation/admin-guide/kdump/kdump.rst for an example.
928 928
929 crashkernel=size[KMG],high 929 crashkernel=size[KMG],high
930 [KNL, X86-64, ARM64, R 930 [KNL, X86-64, ARM64, RISCV, LoongArch] range could be
931 above 4G. 931 above 4G.
932 Allow kernel to alloca 932 Allow kernel to allocate physical memory region from top,
933 so could be above 4G i 933 so could be above 4G if system have more than 4G ram
934 installed. Otherwise m 934 installed. Otherwise memory region will be allocated
935 below 4G, if available 935 below 4G, if available.
936 It will be ignored if 936 It will be ignored if crashkernel=X is specified.
937 crashkernel=size[KMG],low 937 crashkernel=size[KMG],low
938 [KNL, X86-64, ARM64, R 938 [KNL, X86-64, ARM64, RISCV, LoongArch] range under 4G.
939 When crashkernel=X,hig 939 When crashkernel=X,high is passed, kernel could allocate
940 physical memory region 940 physical memory region above 4G, that cause second kernel
941 crash on system that r 941 crash on system that require some amount of low memory,
942 e.g. swiotlb requires 942 e.g. swiotlb requires at least 64M+32K low memory, also
943 enough extra low memor 943 enough extra low memory is needed to make sure DMA buffers
944 for 32-bit devices won 944 for 32-bit devices won't run out. Kernel would try to allocate
945 default size of memory 945 default size of memory below 4G automatically. The default
946 size is platform depen 946 size is platform dependent.
947 --> x86: max(swiotlb 947 --> x86: max(swiotlb_size_or_default() + 8MiB, 256MiB)
948 --> arm64: 128MiB 948 --> arm64: 128MiB
949 --> riscv: 128MiB 949 --> riscv: 128MiB
950 --> loongarch: 128Mi 950 --> loongarch: 128MiB
951 This one lets the user 951 This one lets the user specify own low range under 4G
952 for second kernel inst 952 for second kernel instead.
953 0: to disable low allo 953 0: to disable low allocation.
954 It will be ignored whe 954 It will be ignored when crashkernel=X,high is not used
955 or memory reserved is 955 or memory reserved is below 4G.
956 956
957 cryptomgr.notests 957 cryptomgr.notests
958 [KNL] Disable crypto s 958 [KNL] Disable crypto self-tests
959 959
960 cs89x0_dma= [HW,NET] 960 cs89x0_dma= [HW,NET]
961 Format: <dma> 961 Format: <dma>
962 962
963 cs89x0_media= [HW,NET] 963 cs89x0_media= [HW,NET]
964 Format: { rj45 | aui | 964 Format: { rj45 | aui | bnc }
965 965
966 csdlock_debug= [KNL] Enable or disabl 966 csdlock_debug= [KNL] Enable or disable debug add-ons of cross-CPU
967 function call handling 967 function call handling. When switched on,
968 additional debug data 968 additional debug data is printed to the console
969 in case a hanging CPU 969 in case a hanging CPU is detected, and that
970 CPU is pinged again in 970 CPU is pinged again in order to try to resolve
971 the hang situation. T 971 the hang situation. The default value of this
972 option depends on the 972 option depends on the CSD_LOCK_WAIT_DEBUG_DEFAULT
973 Kconfig option. 973 Kconfig option.
974 974
975 dasd= [HW,NET] 975 dasd= [HW,NET]
976 See header of drivers/ 976 See header of drivers/s390/block/dasd_devmap.c.
977 977
978 db9.dev[2|3]= [HW,JOY] Multisystem j 978 db9.dev[2|3]= [HW,JOY] Multisystem joystick support via parallel port
979 (one device per port) 979 (one device per port)
980 Format: <port#>,<type> 980 Format: <port#>,<type>
981 See also Documentation 981 See also Documentation/input/devices/joystick-parport.rst
982 982
983 debug [KNL,EARLY] Enable ker 983 debug [KNL,EARLY] Enable kernel debugging (events log level).
984 984
985 debug_boot_weak_hash 985 debug_boot_weak_hash
986 [KNL,EARLY] Enable pri 986 [KNL,EARLY] Enable printing [hashed] pointers early in the
987 boot sequence. If ena 987 boot sequence. If enabled, we use a weak hash instead
988 of siphash to hash poi 988 of siphash to hash pointers. Use this option if you are
989 seeing instances of '( 989 seeing instances of '(___ptrval___)') and need to see a
990 value (hashed pointer) 990 value (hashed pointer) instead. Cryptographically
991 insecure, please do no 991 insecure, please do not use on production kernels.
992 992
993 debug_locks_verbose= 993 debug_locks_verbose=
994 [KNL] verbose locking 994 [KNL] verbose locking self-tests
995 Format: <int> 995 Format: <int>
996 Print debugging info w 996 Print debugging info while doing the locking API
997 self-tests. 997 self-tests.
998 Bitmask for the variou 998 Bitmask for the various LOCKTYPE_ tests. Defaults to 0
999 (no extra messages), s 999 (no extra messages), setting it to -1 (all bits set)
1000 will print _a_lot_ mo 1000 will print _a_lot_ more information - normally only
1001 useful to lockdep dev 1001 useful to lockdep developers.
1002 1002
1003 debug_objects [KNL,EARLY] Enable ob 1003 debug_objects [KNL,EARLY] Enable object debugging
1004 1004
1005 debug_guardpage_minorder= 1005 debug_guardpage_minorder=
1006 [KNL,EARLY] When CONF 1006 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this
1007 parameter allows cont 1007 parameter allows control of the order of pages that will
1008 be intentionally kept 1008 be intentionally kept free (and hence protected) by the
1009 buddy allocator. Bigg 1009 buddy allocator. Bigger value increase the probability
1010 of catching random me 1010 of catching random memory corruption, but reduce the
1011 amount of memory for 1011 amount of memory for normal system use. The maximum
1012 possible value is MAX 1012 possible value is MAX_PAGE_ORDER/2. Setting this
1013 parameter to 1 or 2 s 1013 parameter to 1 or 2 should be enough to identify most
1014 random memory corrupt 1014 random memory corruption problems caused by bugs in
1015 kernel or driver code 1015 kernel or driver code when a CPU writes to (or reads
1016 from) a random memory 1016 from) a random memory location. Note that there exists
1017 a class of memory cor 1017 a class of memory corruptions problems caused by buggy
1018 H/W or F/W or by driv 1018 H/W or F/W or by drivers badly programming DMA
1019 (basically when memor 1019 (basically when memory is written at bus level and the
1020 CPU MMU is bypassed) 1020 CPU MMU is bypassed) which are not detectable by
1021 CONFIG_DEBUG_PAGEALLO 1021 CONFIG_DEBUG_PAGEALLOC, hence this option will not
1022 help tracking down th 1022 help tracking down these problems.
1023 1023
1024 debug_pagealloc= 1024 debug_pagealloc=
1025 [KNL,EARLY] When CONF 1025 [KNL,EARLY] When CONFIG_DEBUG_PAGEALLOC is set, this parameter
1026 enables the feature a 1026 enables the feature at boot time. By default, it is
1027 disabled and the syst 1027 disabled and the system will work mostly the same as a
1028 kernel built without 1028 kernel built without CONFIG_DEBUG_PAGEALLOC.
1029 Note: to get most of 1029 Note: to get most of debug_pagealloc error reports, it's
1030 useful to also enable 1030 useful to also enable the page_owner functionality.
1031 on: enable the featur 1031 on: enable the feature
1032 1032
1033 debugfs= [KNL,EARLY] This para 1033 debugfs= [KNL,EARLY] This parameter enables what is exposed to
1034 userspace and debugfs 1034 userspace and debugfs internal clients.
1035 Format: { on, no-moun 1035 Format: { on, no-mount, off }
1036 on: All functions 1036 on: All functions are enabled.
1037 no-mount: 1037 no-mount:
1038 Filesystem is 1038 Filesystem is not registered but kernel clients can
1039 access APIs a 1039 access APIs and a crashkernel can be used to read
1040 its content. 1040 its content. There is nothing to mount.
1041 off: Filesystem is 1041 off: Filesystem is not registered and clients
1042 get a -EPERM 1042 get a -EPERM as result when trying to register files
1043 or directorie 1043 or directories within debugfs.
1044 This is equiv 1044 This is equivalent of the runtime functionality if
1045 debugfs was n 1045 debugfs was not enabled in the kernel at all.
1046 Default value is set 1046 Default value is set in build-time with a kernel configuration.
1047 1047
1048 debugpat [X86] Enable PAT debu 1048 debugpat [X86] Enable PAT debugging
1049 1049
1050 default_hugepagesz= 1050 default_hugepagesz=
1051 [HW] The size of the 1051 [HW] The size of the default HugeTLB page. This is
1052 the size represented 1052 the size represented by the legacy /proc/ hugepages
1053 APIs. In addition, t 1053 APIs. In addition, this is the default hugetlb size
1054 used for shmget(), mm 1054 used for shmget(), mmap() and mounting hugetlbfs
1055 filesystems. If not 1055 filesystems. If not specified, defaults to the
1056 architecture's defaul 1056 architecture's default huge page size. Huge page
1057 sizes are architectur 1057 sizes are architecture dependent. See also
1058 Documentation/admin-g 1058 Documentation/admin-guide/mm/hugetlbpage.rst.
1059 Format: size[KMG] 1059 Format: size[KMG]
1060 1060
1061 deferred_probe_timeout= 1061 deferred_probe_timeout=
1062 [KNL] Debugging optio 1062 [KNL] Debugging option to set a timeout in seconds for
1063 deferred probe to giv 1063 deferred probe to give up waiting on dependencies to
1064 probe. Only specific 1064 probe. Only specific dependencies (subsystems or
1065 drivers) that have op 1065 drivers) that have opted in will be ignored. A timeout
1066 of 0 will timeout at 1066 of 0 will timeout at the end of initcalls. If the time
1067 out hasn't expired, i 1067 out hasn't expired, it'll be restarted by each
1068 successful driver reg 1068 successful driver registration. This option will also
1069 dump out devices stil 1069 dump out devices still on the deferred probe list after
1070 retrying. 1070 retrying.
1071 1071
1072 delayacct [KNL] Enable per-task 1072 delayacct [KNL] Enable per-task delay accounting
1073 1073
1074 dell_smm_hwmon.ignore_dmi= 1074 dell_smm_hwmon.ignore_dmi=
1075 [HW] Continue probing 1075 [HW] Continue probing hardware even if DMI data
1076 indicates that the dr 1076 indicates that the driver is running on unsupported
1077 hardware. 1077 hardware.
1078 1078
1079 dell_smm_hwmon.force= 1079 dell_smm_hwmon.force=
1080 [HW] Activate driver 1080 [HW] Activate driver even if SMM BIOS signature does
1081 not match list of sup 1081 not match list of supported models and enable otherwise
1082 blacklisted features. 1082 blacklisted features.
1083 1083
1084 dell_smm_hwmon.power_status= 1084 dell_smm_hwmon.power_status=
1085 [HW] Report power sta 1085 [HW] Report power status in /proc/i8k
1086 (disabled by default) 1086 (disabled by default).
1087 1087
1088 dell_smm_hwmon.restricted= 1088 dell_smm_hwmon.restricted=
1089 [HW] Allow controllin 1089 [HW] Allow controlling fans only if SYS_ADMIN
1090 capability is set. 1090 capability is set.
1091 1091
1092 dell_smm_hwmon.fan_mult= 1092 dell_smm_hwmon.fan_mult=
1093 [HW] Factor to multip 1093 [HW] Factor to multiply fan speed with.
1094 1094
1095 dell_smm_hwmon.fan_max= 1095 dell_smm_hwmon.fan_max=
1096 [HW] Maximum configur 1096 [HW] Maximum configurable fan speed.
1097 1097
1098 dfltcc= [HW,S390] 1098 dfltcc= [HW,S390]
1099 Format: { on | off | 1099 Format: { on | off | def_only | inf_only | always }
1100 on: s390 zlib h 1100 on: s390 zlib hardware support for compression on
1101 level 1 and 1101 level 1 and decompression (default)
1102 off: No s390 zli 1102 off: No s390 zlib hardware support
1103 def_only: s390 zlib h 1103 def_only: s390 zlib hardware support for deflate
1104 only (compr 1104 only (compression on level 1)
1105 inf_only: s390 zlib h 1105 inf_only: s390 zlib hardware support for inflate
1106 only (decom 1106 only (decompression)
1107 always: Same as 'on 1107 always: Same as 'on' but ignores the selected compression
1108 level alway 1108 level always using hardware support (used for debugging)
1109 1109
1110 dhash_entries= [KNL] 1110 dhash_entries= [KNL]
1111 Set number of hash bu 1111 Set number of hash buckets for dentry cache.
1112 1112
1113 disable_1tb_segments [PPC,EARLY] 1113 disable_1tb_segments [PPC,EARLY]
1114 Disables the use of 1 1114 Disables the use of 1TB hash page table segments. This
1115 causes the kernel to 1115 causes the kernel to fall back to 256MB segments which
1116 can be useful when de 1116 can be useful when debugging issues that require an SLB
1117 miss to occur. 1117 miss to occur.
1118 1118
1119 disable= [IPV6] 1119 disable= [IPV6]
1120 See Documentation/net 1120 See Documentation/networking/ipv6.rst.
1121 1121
1122 disable_radix [PPC,EARLY] 1122 disable_radix [PPC,EARLY]
1123 Disable RADIX MMU mod 1123 Disable RADIX MMU mode on POWER9
1124 1124
1125 disable_tlbie [PPC] 1125 disable_tlbie [PPC]
1126 Disable TLBIE instruc 1126 Disable TLBIE instruction. Currently does not work
1127 with KVM, with HASH M 1127 with KVM, with HASH MMU, or with coherent accelerators.
1128 1128
1129 disable_ddw [PPC/PSERIES,EARLY] 1129 disable_ddw [PPC/PSERIES,EARLY]
1130 Disable Dynamic DMA W 1130 Disable Dynamic DMA Window support. Use this
1131 to workaround buggy f 1131 to workaround buggy firmware.
1132 1132
1133 disable_ipv6= [IPV6] 1133 disable_ipv6= [IPV6]
1134 See Documentation/net 1134 See Documentation/networking/ipv6.rst.
1135 1135
1136 disable_mtrr_cleanup [X86,EARLY] 1136 disable_mtrr_cleanup [X86,EARLY]
1137 The kernel tries to a 1137 The kernel tries to adjust MTRR layout from continuous
1138 to discrete, to make 1138 to discrete, to make X server driver able to add WB
1139 entry later. This par 1139 entry later. This parameter disables that.
1140 1140
1141 disable_mtrr_trim [X86, Intel and AMD 1141 disable_mtrr_trim [X86, Intel and AMD only,EARLY]
1142 By default the kernel 1142 By default the kernel will trim any uncacheable
1143 memory out of your av 1143 memory out of your available memory pool based on
1144 MTRR settings. This 1144 MTRR settings. This parameter disables that behavior,
1145 possibly causing your 1145 possibly causing your machine to run very slowly.
1146 1146
1147 disable_timer_pin_1 [X86,EARLY] 1147 disable_timer_pin_1 [X86,EARLY]
1148 Disable PIN 1 of APIC 1148 Disable PIN 1 of APIC timer
1149 Can be useful to work 1149 Can be useful to work around chipset bugs.
1150 1150
1151 dis_ucode_ldr [X86] Disable the mic 1151 dis_ucode_ldr [X86] Disable the microcode loader.
1152 1152
1153 dma_debug=off If the kernel is comp 1153 dma_debug=off If the kernel is compiled with DMA_API_DEBUG support,
1154 this option disables 1154 this option disables the debugging code at boot.
1155 1155
1156 dma_debug_entries=<number> 1156 dma_debug_entries=<number>
1157 This option allows to 1157 This option allows to tune the number of preallocated
1158 entries for DMA-API d 1158 entries for DMA-API debugging code. One entry is
1159 required per DMA-API 1159 required per DMA-API allocation. Use this if the
1160 DMA-API debugging cod 1160 DMA-API debugging code disables itself because the
1161 architectural default 1161 architectural default is too low.
1162 1162
1163 dma_debug_driver=<driver_name> 1163 dma_debug_driver=<driver_name>
1164 With this option the 1164 With this option the DMA-API debugging driver
1165 filter feature can be 1165 filter feature can be enabled at boot time. Just
1166 pass the driver to fi 1166 pass the driver to filter for as the parameter.
1167 The filter can be dis 1167 The filter can be disabled or changed to another
1168 driver later using sy 1168 driver later using sysfs.
1169 1169
1170 reg_file_data_sampling= 1170 reg_file_data_sampling=
1171 [X86] Controls mitiga 1171 [X86] Controls mitigation for Register File Data
1172 Sampling (RFDS) vulne 1172 Sampling (RFDS) vulnerability. RFDS is a CPU
1173 vulnerability which m 1173 vulnerability which may allow userspace to infer
1174 kernel data values pr 1174 kernel data values previously stored in floating point
1175 registers, vector reg 1175 registers, vector registers, or integer registers.
1176 RFDS only affects Int 1176 RFDS only affects Intel Atom processors.
1177 1177
1178 on: Turns ON the 1178 on: Turns ON the mitigation.
1179 off: Turns OFF the 1179 off: Turns OFF the mitigation.
1180 1180
1181 This parameter overri 1181 This parameter overrides the compile time default set
1182 by CONFIG_MITIGATION_ 1182 by CONFIG_MITIGATION_RFDS. Mitigation cannot be
1183 disabled when other V 1183 disabled when other VERW based mitigations (like MDS)
1184 are enabled. In order 1184 are enabled. In order to disable RFDS mitigation all
1185 VERW based mitigation 1185 VERW based mitigations need to be disabled.
1186 1186
1187 For details see: 1187 For details see:
1188 Documentation/admin-g 1188 Documentation/admin-guide/hw-vuln/reg-file-data-sampling.rst
1189 1189
1190 driver_async_probe= [KNL] 1190 driver_async_probe= [KNL]
1191 List of driver names 1191 List of driver names to be probed asynchronously. *
1192 matches with all driv 1192 matches with all driver names. If * is specified, the
1193 rest of the listed dr 1193 rest of the listed driver names are those that will NOT
1194 match the *. 1194 match the *.
1195 Format: <driver_name1 1195 Format: <driver_name1>,<driver_name2>...
1196 1196
1197 drm.edid_firmware=[<connector>:]<file 1197 drm.edid_firmware=[<connector>:]<file>[,[<connector>:]<file>]
1198 Broken monitors, grap 1198 Broken monitors, graphic adapters, KVMs and EDIDless
1199 panels may send no or 1199 panels may send no or incorrect EDID data sets.
1200 This parameter allows 1200 This parameter allows to specify an EDID data sets
1201 in the /lib/firmware 1201 in the /lib/firmware directory that are used instead.
1202 An EDID data set will 1202 An EDID data set will only be used for a particular
1203 connector, if its nam 1203 connector, if its name and a colon are prepended to
1204 the EDID name. Each c 1204 the EDID name. Each connector may use a unique EDID
1205 data set by separatin 1205 data set by separating the files with a comma. An EDID
1206 data set with no conn 1206 data set with no connector name will be used for
1207 any connectors not ex 1207 any connectors not explicitly specified.
1208 1208
1209 dscc4.setup= [NET] 1209 dscc4.setup= [NET]
1210 1210
1211 dt_cpu_ftrs= [PPC,EARLY] 1211 dt_cpu_ftrs= [PPC,EARLY]
1212 Format: {"off" | "kno 1212 Format: {"off" | "known"}
1213 Control how the dt_cp 1213 Control how the dt_cpu_ftrs device-tree binding is
1214 used for CPU feature 1214 used for CPU feature discovery and setup (if it
1215 exists). 1215 exists).
1216 off: Do not use it, f 1216 off: Do not use it, fall back to legacy cpu table.
1217 known: Do not pass th 1217 known: Do not pass through unknown features to guests
1218 or userspace, only th 1218 or userspace, only those that the kernel is aware of.
1219 1219
1220 dump_apple_properties [X86] 1220 dump_apple_properties [X86]
1221 Dump name and content 1221 Dump name and content of EFI device properties on
1222 x86 Macs. Useful for 1222 x86 Macs. Useful for driver authors to determine
1223 what data is availabl 1223 what data is available or for reverse-engineering.
1224 1224
1225 dyndbg[="val"] [KNL,DYNAMIC_ 1225 dyndbg[="val"] [KNL,DYNAMIC_DEBUG]
1226 <module>.dyndbg[="val"] 1226 <module>.dyndbg[="val"]
1227 Enable debug messages 1227 Enable debug messages at boot time. See
1228 Documentation/admin-g 1228 Documentation/admin-guide/dynamic-debug-howto.rst
1229 for details. 1229 for details.
1230 1230
1231 early_ioremap_debug [KNL,EARLY] 1231 early_ioremap_debug [KNL,EARLY]
1232 Enable debug messages 1232 Enable debug messages in early_ioremap support. This
1233 is useful for trackin 1233 is useful for tracking down temporary early mappings
1234 which are not unmappe 1234 which are not unmapped.
1235 1235
1236 earlycon= [KNL,EARLY] Output ea 1236 earlycon= [KNL,EARLY] Output early console device and options.
1237 1237
1238 When used with no opt 1238 When used with no options, the early console is
1239 determined by stdout- 1239 determined by stdout-path property in device tree's
1240 chosen node or the AC 1240 chosen node or the ACPI SPCR table if supported by
1241 the platform. 1241 the platform.
1242 1242
1243 cdns,<addr>[,options] 1243 cdns,<addr>[,options]
1244 Start an early, polle 1244 Start an early, polled-mode console on a Cadence
1245 (xuartps) serial port 1245 (xuartps) serial port at the specified address. Only
1246 supported option is b 1246 supported option is baud rate. If baud rate is not
1247 specified, the serial 1247 specified, the serial port must already be setup and
1248 configured. 1248 configured.
1249 1249
1250 uart[8250],io,<addr>[,options 1250 uart[8250],io,<addr>[,options[,uartclk]]
1251 uart[8250],mmio,<addr>[,optio 1251 uart[8250],mmio,<addr>[,options[,uartclk]]
1252 uart[8250],mmio32,<addr>[,opt 1252 uart[8250],mmio32,<addr>[,options[,uartclk]]
1253 uart[8250],mmio32be,<addr>[,o 1253 uart[8250],mmio32be,<addr>[,options[,uartclk]]
1254 uart[8250],0x<addr>[,options] 1254 uart[8250],0x<addr>[,options]
1255 Start an early, polle 1255 Start an early, polled-mode console on the 8250/16550
1256 UART at the specified 1256 UART at the specified I/O port or MMIO address.
1257 MMIO inter-register a 1257 MMIO inter-register address stride is either 8-bit
1258 (mmio) or 32-bit (mmi 1258 (mmio) or 32-bit (mmio32 or mmio32be).
1259 If none of [io|mmio|m 1259 If none of [io|mmio|mmio32|mmio32be], <addr> is assumed
1260 to be equivalent to ' 1260 to be equivalent to 'mmio'. 'options' are specified
1261 in the same format de 1261 in the same format described for "console=ttyS<n>"; if
1262 unspecified, the h/w 1262 unspecified, the h/w is not initialized. 'uartclk' is
1263 the uart clock freque 1263 the uart clock frequency; if unspecified, it is set
1264 to 'BASE_BAUD' * 16. 1264 to 'BASE_BAUD' * 16.
1265 1265
1266 pl011,<addr> 1266 pl011,<addr>
1267 pl011,mmio32,<addr> 1267 pl011,mmio32,<addr>
1268 Start an early, polle 1268 Start an early, polled-mode console on a pl011 serial
1269 port at the specified 1269 port at the specified address. The pl011 serial port
1270 must already be setup 1270 must already be setup and configured. Options are not
1271 yet supported. If 'm 1271 yet supported. If 'mmio32' is specified, then only
1272 the driver will use o 1272 the driver will use only 32-bit accessors to read/write
1273 the device registers. 1273 the device registers.
1274 1274
1275 liteuart,<addr> 1275 liteuart,<addr>
1276 Start an early consol 1276 Start an early console on a litex serial port at the
1277 specified address. Th 1277 specified address. The serial port must already be
1278 setup and configured. 1278 setup and configured. Options are not yet supported.
1279 1279
1280 meson,<addr> 1280 meson,<addr>
1281 Start an early, polle 1281 Start an early, polled-mode console on a meson serial
1282 port at the specified 1282 port at the specified address. The serial port must
1283 already be setup and 1283 already be setup and configured. Options are not yet
1284 supported. 1284 supported.
1285 1285
1286 msm_serial,<addr> 1286 msm_serial,<addr>
1287 Start an early, polle 1287 Start an early, polled-mode console on an msm serial
1288 port at the specified 1288 port at the specified address. The serial port
1289 must already be setup 1289 must already be setup and configured. Options are not
1290 yet supported. 1290 yet supported.
1291 1291
1292 msm_serial_dm,<addr> 1292 msm_serial_dm,<addr>
1293 Start an early, polle 1293 Start an early, polled-mode console on an msm serial
1294 dm port at the specif 1294 dm port at the specified address. The serial port
1295 must already be setup 1295 must already be setup and configured. Options are not
1296 yet supported. 1296 yet supported.
1297 1297
1298 owl,<addr> 1298 owl,<addr>
1299 Start an early, polle 1299 Start an early, polled-mode console on a serial port
1300 of an Actions Semi So 1300 of an Actions Semi SoC, such as S500 or S900, at the
1301 specified address. Th 1301 specified address. The serial port must already be
1302 setup and configured. 1302 setup and configured. Options are not yet supported.
1303 1303
1304 rda,<addr> 1304 rda,<addr>
1305 Start an early, polle 1305 Start an early, polled-mode console on a serial port
1306 of an RDA Micro SoC, 1306 of an RDA Micro SoC, such as RDA8810PL, at the
1307 specified address. Th 1307 specified address. The serial port must already be
1308 setup and configured. 1308 setup and configured. Options are not yet supported.
1309 1309
1310 sbi 1310 sbi
1311 Use RISC-V SBI (Super 1311 Use RISC-V SBI (Supervisor Binary Interface) for early
1312 console. 1312 console.
1313 1313
1314 smh Use ARM semihosting c 1314 smh Use ARM semihosting calls for early console.
1315 1315
1316 s3c2410,<addr> 1316 s3c2410,<addr>
1317 s3c2412,<addr> 1317 s3c2412,<addr>
1318 s3c2440,<addr> 1318 s3c2440,<addr>
1319 s3c6400,<addr> 1319 s3c6400,<addr>
1320 s5pv210,<addr> 1320 s5pv210,<addr>
1321 exynos4210,<addr> 1321 exynos4210,<addr>
1322 Use early console pro 1322 Use early console provided by serial driver available
1323 on Samsung SoCs, requ 1323 on Samsung SoCs, requires selecting proper type and
1324 a correct base addres 1324 a correct base address of the selected UART port. The
1325 serial port must alre 1325 serial port must already be setup and configured.
1326 Options are not yet s 1326 Options are not yet supported.
1327 1327
1328 lantiq,<addr> 1328 lantiq,<addr>
1329 Start an early, polle 1329 Start an early, polled-mode console on a lantiq serial
1330 (lqasc) port at the s 1330 (lqasc) port at the specified address. The serial port
1331 must already be setup 1331 must already be setup and configured. Options are not
1332 yet supported. 1332 yet supported.
1333 1333
1334 lpuart,<addr> 1334 lpuart,<addr>
1335 lpuart32,<addr> 1335 lpuart32,<addr>
1336 Use early console pro 1336 Use early console provided by Freescale LP UART driver
1337 found on Freescale Vy 1337 found on Freescale Vybrid and QorIQ LS1021A processors.
1338 A valid base address 1338 A valid base address must be provided, and the serial
1339 port must already be 1339 port must already be setup and configured.
1340 1340
1341 ec_imx21,<addr> 1341 ec_imx21,<addr>
1342 ec_imx6q,<addr> 1342 ec_imx6q,<addr>
1343 Start an early, polle 1343 Start an early, polled-mode, output-only console on the
1344 Freescale i.MX UART a 1344 Freescale i.MX UART at the specified address. The UART
1345 must already be setup 1345 must already be setup and configured.
1346 1346
1347 ar3700_uart,<addr> 1347 ar3700_uart,<addr>
1348 Start an early, polle 1348 Start an early, polled-mode console on the
1349 Armada 3700 serial po 1349 Armada 3700 serial port at the specified
1350 address. The serial p 1350 address. The serial port must already be setup
1351 and configured. Optio 1351 and configured. Options are not yet supported.
1352 1352
1353 qcom_geni,<addr> 1353 qcom_geni,<addr>
1354 Start an early, polle 1354 Start an early, polled-mode console on a Qualcomm
1355 Generic Interface (GE 1355 Generic Interface (GENI) based serial port at the
1356 specified address. Th 1356 specified address. The serial port must already be
1357 setup and configured. 1357 setup and configured. Options are not yet supported.
1358 1358
1359 efifb,[options] 1359 efifb,[options]
1360 Start an early, unacc 1360 Start an early, unaccelerated console on the EFI
1361 memory mapped framebu 1361 memory mapped framebuffer (if available). On cache
1362 coherent non-x86 syst 1362 coherent non-x86 systems that use system memory for
1363 the framebuffer, pass 1363 the framebuffer, pass the 'ram' option so that it is
1364 mapped with the corre 1364 mapped with the correct attributes.
1365 1365
1366 linflex,<addr> 1366 linflex,<addr>
1367 Use early console pro 1367 Use early console provided by Freescale LINFlexD UART
1368 serial driver for NXP 1368 serial driver for NXP S32V234 SoCs. A valid base
1369 address must be provi 1369 address must be provided, and the serial port must
1370 already be setup and 1370 already be setup and configured.
1371 1371
1372 earlyprintk= [X86,SH,ARM,M68k,S390 1372 earlyprintk= [X86,SH,ARM,M68k,S390,UM,EARLY]
1373 earlyprintk=vga 1373 earlyprintk=vga
1374 earlyprintk=sclp 1374 earlyprintk=sclp
1375 earlyprintk=xen 1375 earlyprintk=xen
1376 earlyprintk=serial[,t 1376 earlyprintk=serial[,ttySn[,baudrate]]
1377 earlyprintk=serial[,0 1377 earlyprintk=serial[,0x...[,baudrate]]
1378 earlyprintk=ttySn[,ba 1378 earlyprintk=ttySn[,baudrate]
1379 earlyprintk=dbgp[debu 1379 earlyprintk=dbgp[debugController#]
1380 earlyprintk=pciserial 1380 earlyprintk=pciserial[,force],bus:device.function[,baudrate]
1381 earlyprintk=xdbc[xhci 1381 earlyprintk=xdbc[xhciController#]
1382 earlyprintk=bios 1382 earlyprintk=bios
1383 1383
1384 earlyprintk is useful 1384 earlyprintk is useful when the kernel crashes before
1385 the normal console is 1385 the normal console is initialized. It is not enabled by
1386 default because it ha 1386 default because it has some cosmetic problems.
1387 1387
1388 Append ",keep" to not 1388 Append ",keep" to not disable it when the real console
1389 takes over. 1389 takes over.
1390 1390
1391 Only one of vga, seri 1391 Only one of vga, serial, or usb debug port can
1392 be used at a time. 1392 be used at a time.
1393 1393
1394 Currently only ttyS0 1394 Currently only ttyS0 and ttyS1 may be specified by
1395 name. Other I/O port 1395 name. Other I/O ports may be explicitly specified
1396 on some architectures 1396 on some architectures (x86 and arm at least) by
1397 replacing ttySn with 1397 replacing ttySn with an I/O port address, like this:
1398 earlyprintk=s 1398 earlyprintk=serial,0x1008,115200
1399 You can find the port 1399 You can find the port for a given device in
1400 /proc/tty/driver/seri 1400 /proc/tty/driver/serial:
1401 2: uart:ST166 1401 2: uart:ST16650V2 port:00001008 irq:18 ...
1402 1402
1403 Interaction with the 1403 Interaction with the standard serial driver is not
1404 very good. 1404 very good.
1405 1405
1406 The VGA output is eve 1406 The VGA output is eventually overwritten by
1407 the real console. 1407 the real console.
1408 1408
1409 The xen option can on 1409 The xen option can only be used in Xen domains.
1410 1410
1411 The sclp output can o 1411 The sclp output can only be used on s390.
1412 1412
1413 The bios output can o 1413 The bios output can only be used on SuperH.
1414 1414
1415 The optional "force" 1415 The optional "force" to "pciserial" enables use of a
1416 PCI device even when 1416 PCI device even when its classcode is not of the
1417 UART class. 1417 UART class.
1418 1418
1419 edac_report= [HW,EDAC] Control how 1419 edac_report= [HW,EDAC] Control how to report EDAC event
1420 Format: {"on" | "off" 1420 Format: {"on" | "off" | "force"}
1421 on: enable EDAC to re 1421 on: enable EDAC to report H/W event. May be overridden
1422 by other higher prior 1422 by other higher priority error reporting module.
1423 off: disable H/W even 1423 off: disable H/W event reporting through EDAC.
1424 force: enforce the us 1424 force: enforce the use of EDAC to report H/W event.
1425 default: on. 1425 default: on.
1426 1426
1427 edd= [EDD] 1427 edd= [EDD]
1428 Format: {"off" | "on" 1428 Format: {"off" | "on" | "skip[mbr]"}
1429 1429
1430 efi= [EFI,EARLY] 1430 efi= [EFI,EARLY]
1431 Format: { "debug", "d 1431 Format: { "debug", "disable_early_pci_dma",
1432 "nochunk", 1432 "nochunk", "noruntime", "nosoftreserve",
1433 "novamap", 1433 "novamap", "no_disable_early_pci_dma" }
1434 debug: enable misc de 1434 debug: enable misc debug output.
1435 disable_early_pci_dma 1435 disable_early_pci_dma: disable the busmaster bit on all
1436 PCI bridges while in 1436 PCI bridges while in the EFI boot stub.
1437 nochunk: disable read 1437 nochunk: disable reading files in "chunks" in the EFI
1438 boot stub, as chunkin 1438 boot stub, as chunking can cause problems with some
1439 firmware implementati 1439 firmware implementations.
1440 noruntime : disable E 1440 noruntime : disable EFI runtime services support
1441 nosoftreserve: The EF 1441 nosoftreserve: The EFI_MEMORY_SP (Specific Purpose)
1442 attribute may cause t 1442 attribute may cause the kernel to reserve the
1443 memory range for a me 1443 memory range for a memory mapping driver to
1444 claim. Specify efi=no 1444 claim. Specify efi=nosoftreserve to disable this
1445 reservation and treat 1445 reservation and treat the memory by its base type
1446 (i.e. EFI_CONVENTIONA 1446 (i.e. EFI_CONVENTIONAL_MEMORY / "System RAM").
1447 novamap: do not call 1447 novamap: do not call SetVirtualAddressMap().
1448 no_disable_early_pci_ 1448 no_disable_early_pci_dma: Leave the busmaster bit set
1449 on all PCI bridges wh 1449 on all PCI bridges while in the EFI boot stub
1450 1450
1451 efi_no_storage_paranoia [EFI,X86,EARL 1451 efi_no_storage_paranoia [EFI,X86,EARLY]
1452 Using this parameter 1452 Using this parameter you can use more than 50% of
1453 your efi variable sto 1453 your efi variable storage. Use this parameter only if
1454 you are really sure t 1454 you are really sure that your UEFI does sane gc and
1455 fulfills the spec oth 1455 fulfills the spec otherwise your board may brick.
1456 1456
1457 efivar_ssdt= [EFI; X86] Name of an 1457 efivar_ssdt= [EFI; X86] Name of an EFI variable that contains an SSDT
1458 that is to be dynamic 1458 that is to be dynamically loaded by Linux. If there are
1459 multiple variables wi 1459 multiple variables with the same name but with different
1460 vendor GUIDs, all of 1460 vendor GUIDs, all of them will be loaded. See
1461 Documentation/admin-g 1461 Documentation/admin-guide/acpi/ssdt-overlays.rst for details.
1462 1462
1463 1463
1464 eisa_irq_edge= [PARISC,HW] 1464 eisa_irq_edge= [PARISC,HW]
1465 See header of drivers 1465 See header of drivers/parisc/eisa.c.
1466 1466
1467 ekgdboc= [X86,KGDB,EARLY] Allo 1467 ekgdboc= [X86,KGDB,EARLY] Allow early kernel console debugging
1468 Format: ekgdboc=kbd 1468 Format: ekgdboc=kbd
1469 1469
1470 This is designed to b 1470 This is designed to be used in conjunction with
1471 the boot argument: ea 1471 the boot argument: earlyprintk=vga
1472 1472
1473 This parameter works 1473 This parameter works in place of the kgdboc parameter
1474 but can only be used 1474 but can only be used if the backing tty is available
1475 very early in the boo 1475 very early in the boot process. For early debugging
1476 via a serial port see 1476 via a serial port see kgdboc_earlycon instead.
1477 1477
1478 elanfreq= [X86-32] 1478 elanfreq= [X86-32]
1479 See comment before fu 1479 See comment before function elanfreq_setup() in
1480 arch/x86/kernel/cpu/c 1480 arch/x86/kernel/cpu/cpufreq/elanfreq.c.
1481 1481
1482 elfcorehdr=[size[KMG]@]offset[KMG] [P 1482 elfcorehdr=[size[KMG]@]offset[KMG] [PPC,SH,X86,S390,EARLY]
1483 Specifies physical ad 1483 Specifies physical address of start of kernel core
1484 image elf header and 1484 image elf header and optionally the size. Generally
1485 kexec loader will pas 1485 kexec loader will pass this option to capture kernel.
1486 See Documentation/adm 1486 See Documentation/admin-guide/kdump/kdump.rst for details.
1487 1487
1488 enable_mtrr_cleanup [X86,EARLY] 1488 enable_mtrr_cleanup [X86,EARLY]
1489 The kernel tries to a 1489 The kernel tries to adjust MTRR layout from continuous
1490 to discrete, to make 1490 to discrete, to make X server driver able to add WB
1491 entry later. This par 1491 entry later. This parameter enables that.
1492 1492
1493 enable_timer_pin_1 [X86] 1493 enable_timer_pin_1 [X86]
1494 Enable PIN 1 of APIC 1494 Enable PIN 1 of APIC timer
1495 Can be useful to work 1495 Can be useful to work around chipset bugs
1496 (in particular on som 1496 (in particular on some ATI chipsets).
1497 The kernel tries to s 1497 The kernel tries to set a reasonable default.
1498 1498
1499 enforcing= [SELINUX] Set initial 1499 enforcing= [SELINUX] Set initial enforcing status.
1500 Format: {"0" | "1"} 1500 Format: {"0" | "1"}
1501 See security/selinux/ 1501 See security/selinux/Kconfig help text.
1502 0 -- permissive (log 1502 0 -- permissive (log only, no denials).
1503 1 -- enforcing (deny 1503 1 -- enforcing (deny and log).
1504 Default value is 0. 1504 Default value is 0.
1505 Value can be changed 1505 Value can be changed at runtime via
1506 /sys/fs/selinux/enfor 1506 /sys/fs/selinux/enforce.
1507 1507
1508 erst_disable [ACPI] 1508 erst_disable [ACPI]
1509 Disable Error Record 1509 Disable Error Record Serialization Table (ERST)
1510 support. 1510 support.
1511 1511
1512 ether= [HW,NET] Ethernet car 1512 ether= [HW,NET] Ethernet cards parameters
1513 This option is obsole 1513 This option is obsoleted by the "netdev=" option, which
1514 has equivalent usage. 1514 has equivalent usage. See its documentation for details.
1515 1515
1516 evm= [EVM] 1516 evm= [EVM]
1517 Format: { "fix" } 1517 Format: { "fix" }
1518 Permit 'security.evm' 1518 Permit 'security.evm' to be updated regardless of
1519 current integrity sta 1519 current integrity status.
1520 1520
1521 early_page_ext [KNL,EARLY] Enforces p 1521 early_page_ext [KNL,EARLY] Enforces page_ext initialization to earlier
1522 stages so cover more 1522 stages so cover more early boot allocations.
1523 Please note that as s 1523 Please note that as side effect some optimizations
1524 might be disabled to 1524 might be disabled to achieve that (e.g. parallelized
1525 memory initialization 1525 memory initialization is disabled) so the boot process
1526 might take longer, es 1526 might take longer, especially on systems with a lot of
1527 memory. Available wit 1527 memory. Available with CONFIG_PAGE_EXTENSION=y.
1528 1528
1529 failslab= 1529 failslab=
1530 fail_usercopy= 1530 fail_usercopy=
1531 fail_page_alloc= 1531 fail_page_alloc=
1532 fail_make_request=[KNL] 1532 fail_make_request=[KNL]
1533 General fault injecti 1533 General fault injection mechanism.
1534 Format: <interval>,<p 1534 Format: <interval>,<probability>,<space>,<times>
1535 See also Documentatio 1535 See also Documentation/fault-injection/.
1536 1536
1537 fb_tunnels= [NET] 1537 fb_tunnels= [NET]
1538 Format: { initns | no 1538 Format: { initns | none }
1539 See Documentation/adm 1539 See Documentation/admin-guide/sysctl/net.rst for
1540 fb_tunnels_only_for_i 1540 fb_tunnels_only_for_init_ns
1541 1541
1542 floppy= [HW] 1542 floppy= [HW]
1543 See Documentation/adm 1543 See Documentation/admin-guide/blockdev/floppy.rst.
1544 1544
1545 forcepae [X86-32] 1545 forcepae [X86-32]
1546 Forcefully enable Phy 1546 Forcefully enable Physical Address Extension (PAE).
1547 Many Pentium M system 1547 Many Pentium M systems disable PAE but may have a
1548 functionally usable P 1548 functionally usable PAE implementation.
1549 Warning: use of this 1549 Warning: use of this parameter will taint the kernel
1550 and may cause unknown 1550 and may cause unknown problems.
1551 1551
1552 fred= [X86-64] 1552 fred= [X86-64]
1553 Enable/disable Flexib 1553 Enable/disable Flexible Return and Event Delivery.
1554 Format: { on | off } 1554 Format: { on | off }
1555 on: enable FRED when 1555 on: enable FRED when it's present.
1556 off: disable FRED, th 1556 off: disable FRED, the default setting.
1557 1557
1558 ftrace=[tracer] 1558 ftrace=[tracer]
1559 [FTRACE] will set and 1559 [FTRACE] will set and start the specified tracer
1560 as early as possible 1560 as early as possible in order to facilitate early
1561 boot debugging. 1561 boot debugging.
1562 1562
1563 ftrace_boot_snapshot 1563 ftrace_boot_snapshot
1564 [FTRACE] On boot up, 1564 [FTRACE] On boot up, a snapshot will be taken of the
1565 ftrace ring buffer th 1565 ftrace ring buffer that can be read at:
1566 /sys/kernel/tracing/s 1566 /sys/kernel/tracing/snapshot.
1567 This is useful if you 1567 This is useful if you need tracing information from kernel
1568 boot up that is likel 1568 boot up that is likely to be overridden by user space
1569 start up functionalit 1569 start up functionality.
1570 1570
1571 Optionally, the snaps 1571 Optionally, the snapshot can also be defined for a tracing
1572 instance that was cre 1572 instance that was created by the trace_instance= command
1573 line parameter. 1573 line parameter.
1574 1574
1575 trace_instance=foo,sc 1575 trace_instance=foo,sched_switch ftrace_boot_snapshot=foo
1576 1576
1577 The above will cause 1577 The above will cause the "foo" tracing instance to trigger
1578 a snapshot at the end 1578 a snapshot at the end of boot up.
1579 1579
1580 ftrace_dump_on_oops[=2(orig_cpu) | =< 1580 ftrace_dump_on_oops[=2(orig_cpu) | =<instance>][,<instance> |
1581 ,<instance>=2(orig_ 1581 ,<instance>=2(orig_cpu)]
1582 [FTRACE] will dump th 1582 [FTRACE] will dump the trace buffers on oops.
1583 If no parameter is pa 1583 If no parameter is passed, ftrace will dump global
1584 buffers of all CPUs, 1584 buffers of all CPUs, if you pass 2 or orig_cpu, it
1585 will dump only the bu 1585 will dump only the buffer of the CPU that triggered
1586 the oops, or the spec 1586 the oops, or the specific instance will be dumped if
1587 its name is passed. M 1587 its name is passed. Multiple instance dump is also
1588 supported, and instan 1588 supported, and instances are separated by commas. Each
1589 instance supports onl 1589 instance supports only dump on CPU that triggered the
1590 oops by passing 2 or 1590 oops by passing 2 or orig_cpu to it.
1591 1591
1592 ftrace_dump_on_oops=f 1592 ftrace_dump_on_oops=foo=orig_cpu
1593 1593
1594 The above will dump o 1594 The above will dump only the buffer of "foo" instance
1595 on CPU that triggered 1595 on CPU that triggered the oops.
1596 1596
1597 ftrace_dump_on_oops,f 1597 ftrace_dump_on_oops,foo,bar=orig_cpu
1598 1598
1599 The above will dump g 1599 The above will dump global buffer on all CPUs, the
1600 buffer of "foo" insta 1600 buffer of "foo" instance on all CPUs and the buffer
1601 of "bar" instance on 1601 of "bar" instance on CPU that triggered the oops.
1602 1602
1603 ftrace_filter=[function-list] 1603 ftrace_filter=[function-list]
1604 [FTRACE] Limit the fu 1604 [FTRACE] Limit the functions traced by the function
1605 tracer at boot up. fu 1605 tracer at boot up. function-list is a comma-separated
1606 list of functions. Th 1606 list of functions. This list can be changed at run
1607 time by the set_ftrac 1607 time by the set_ftrace_filter file in the debugfs
1608 tracing directory. 1608 tracing directory.
1609 1609
1610 ftrace_notrace=[function-list] 1610 ftrace_notrace=[function-list]
1611 [FTRACE] Do not trace 1611 [FTRACE] Do not trace the functions specified in
1612 function-list. This l 1612 function-list. This list can be changed at run time
1613 by the set_ftrace_not 1613 by the set_ftrace_notrace file in the debugfs
1614 tracing directory. 1614 tracing directory.
1615 1615
1616 ftrace_graph_filter=[function-list] 1616 ftrace_graph_filter=[function-list]
1617 [FTRACE] Limit the to 1617 [FTRACE] Limit the top level callers functions traced
1618 by the function graph 1618 by the function graph tracer at boot up.
1619 function-list is a co 1619 function-list is a comma-separated list of functions
1620 that can be changed a 1620 that can be changed at run time by the
1621 set_graph_function fi 1621 set_graph_function file in the debugfs tracing directory.
1622 1622
1623 ftrace_graph_notrace=[function-list] 1623 ftrace_graph_notrace=[function-list]
1624 [FTRACE] Do not trace 1624 [FTRACE] Do not trace from the functions specified in
1625 function-list. This 1625 function-list. This list is a comma-separated list of
1626 functions that can be 1626 functions that can be changed at run time by the
1627 set_graph_notrace fil 1627 set_graph_notrace file in the debugfs tracing directory.
1628 1628
1629 ftrace_graph_max_depth=<uint> 1629 ftrace_graph_max_depth=<uint>
1630 [FTRACE] Used with th 1630 [FTRACE] Used with the function graph tracer. This is
1631 the max depth it will 1631 the max depth it will trace into a function. This value
1632 can be changed at run 1632 can be changed at run time by the max_graph_depth file
1633 in the tracefs tracin 1633 in the tracefs tracing directory. default: 0 (no limit)
1634 1634
1635 fw_devlink= [KNL,EARLY] Create de 1635 fw_devlink= [KNL,EARLY] Create device links between consumer and supplier
1636 devices by scanning t 1636 devices by scanning the firmware to infer the
1637 consumer/supplier rel 1637 consumer/supplier relationships. This feature is
1638 especially useful whe 1638 especially useful when drivers are loaded as modules as
1639 it ensures proper ord 1639 it ensures proper ordering of tasks like device probing
1640 (suppliers first, the 1640 (suppliers first, then consumers), supplier boot state
1641 clean up (only after 1641 clean up (only after all consumers have probed),
1642 suspend/resume & runt 1642 suspend/resume & runtime PM (consumers first, then
1643 suppliers). 1643 suppliers).
1644 Format: { off | permi 1644 Format: { off | permissive | on | rpm }
1645 off -- Don't create 1645 off -- Don't create device links from firmware info.
1646 permissive -- Create 1646 permissive -- Create device links from firmware info
1647 but use it on 1647 but use it only for ordering boot state clean
1648 up (sync_stat 1648 up (sync_state() calls).
1649 on -- Create device 1649 on -- Create device links from firmware info and use it
1650 to enforce pr 1650 to enforce probe and suspend/resume ordering.
1651 rpm -- Like "on", bu 1651 rpm -- Like "on", but also use to order runtime PM.
1652 1652
1653 fw_devlink.strict=<bool> 1653 fw_devlink.strict=<bool>
1654 [KNL,EARLY] Treat all 1654 [KNL,EARLY] Treat all inferred dependencies as mandatory
1655 dependencies. This on 1655 dependencies. This only applies for fw_devlink=on|rpm.
1656 Format: <bool> 1656 Format: <bool>
1657 1657
1658 fw_devlink.sync_state = 1658 fw_devlink.sync_state =
1659 [KNL,EARLY] When all 1659 [KNL,EARLY] When all devices that could probe have finished
1660 probing, this paramet 1660 probing, this parameter controls what to do with
1661 devices that haven't 1661 devices that haven't yet received their sync_state()
1662 calls. 1662 calls.
1663 Format: { strict | ti 1663 Format: { strict | timeout }
1664 strict -- Default. Co 1664 strict -- Default. Continue waiting on consumers to
1665 probe success 1665 probe successfully.
1666 timeout -- Give up wa 1666 timeout -- Give up waiting on consumers and call
1667 sync_state() 1667 sync_state() on any devices that haven't yet
1668 received thei 1668 received their sync_state() calls after
1669 deferred_prob 1669 deferred_probe_timeout has expired or by
1670 late_initcall 1670 late_initcall() if !CONFIG_MODULES.
1671 1671
1672 gamecon.map[2|3]= 1672 gamecon.map[2|3]=
1673 [HW,JOY] Multisystem 1673 [HW,JOY] Multisystem joystick and NES/SNES/PSX pad
1674 support via parallel 1674 support via parallel port (up to 5 devices per port)
1675 Format: <port#>,<pad1 1675 Format: <port#>,<pad1>,<pad2>,<pad3>,<pad4>,<pad5>
1676 See also Documentatio 1676 See also Documentation/input/devices/joystick-parport.rst
1677 1677
1678 gamma= [HW,DRM] 1678 gamma= [HW,DRM]
1679 1679
1680 gart_fix_e820= [X86-64,EARLY] disabl 1680 gart_fix_e820= [X86-64,EARLY] disable the fix e820 for K8 GART
1681 Format: off | on 1681 Format: off | on
1682 default: on 1682 default: on
1683 1683
1684 gather_data_sampling= 1684 gather_data_sampling=
1685 [X86,INTEL,EARLY] Con 1685 [X86,INTEL,EARLY] Control the Gather Data Sampling (GDS)
1686 mitigation. 1686 mitigation.
1687 1687
1688 Gather Data Sampling 1688 Gather Data Sampling is a hardware vulnerability which
1689 allows unprivileged s 1689 allows unprivileged speculative access to data which was
1690 previously stored in 1690 previously stored in vector registers.
1691 1691
1692 This issue is mitigat 1692 This issue is mitigated by default in updated microcode.
1693 The mitigation may ha 1693 The mitigation may have a performance impact but can be
1694 disabled. On systems 1694 disabled. On systems without the microcode mitigation
1695 disabling AVX serves 1695 disabling AVX serves as a mitigation.
1696 1696
1697 force: Disable AVX t 1697 force: Disable AVX to mitigate systems without
1698 microcode mit 1698 microcode mitigation. No effect if the microcode
1699 mitigation is 1699 mitigation is present. Known to cause crashes in
1700 userspace wit 1700 userspace with buggy AVX enumeration.
1701 1701
1702 off: Disable GDS m 1702 off: Disable GDS mitigation.
1703 1703
1704 gcov_persist= [GCOV] When non-zero 1704 gcov_persist= [GCOV] When non-zero (default), profiling data for
1705 kernel modules is sav 1705 kernel modules is saved and remains accessible via
1706 debugfs, even when th 1706 debugfs, even when the module is unloaded/reloaded.
1707 When zero, profiling 1707 When zero, profiling data is discarded and associated
1708 debugfs files are rem 1708 debugfs files are removed at module unload time.
1709 1709
1710 goldfish [X86] Enable the gold 1710 goldfish [X86] Enable the goldfish android emulator platform.
1711 Don't use this when y 1711 Don't use this when you are not running on the
1712 android emulator 1712 android emulator
1713 1713
1714 gpio-mockup.gpio_mockup_ranges 1714 gpio-mockup.gpio_mockup_ranges
1715 [HW] Sets the ranges 1715 [HW] Sets the ranges of gpiochip of for this device.
1716 Format: <start1>,<end 1716 Format: <start1>,<end1>,<start2>,<end2>...
1717 gpio-mockup.gpio_mockup_named_lines 1717 gpio-mockup.gpio_mockup_named_lines
1718 [HW] Let the driver k 1718 [HW] Let the driver know GPIO lines should be named.
1719 1719
1720 gpt [EFI] Forces disk wit 1720 gpt [EFI] Forces disk with valid GPT signature but
1721 invalid Protective MB 1721 invalid Protective MBR to be treated as GPT. If the
1722 primary GPT is corrup 1722 primary GPT is corrupted, it enables the backup/alternate
1723 GPT to be used instea 1723 GPT to be used instead.
1724 1724
1725 grcan.enable0= [HW] Configuration of 1725 grcan.enable0= [HW] Configuration of physical interface 0. Determines
1726 the "Enable 0" bit of 1726 the "Enable 0" bit of the configuration register.
1727 Format: 0 | 1 1727 Format: 0 | 1
1728 Default: 0 1728 Default: 0
1729 grcan.enable1= [HW] Configuration of 1729 grcan.enable1= [HW] Configuration of physical interface 1. Determines
1730 the "Enable 0" bit of 1730 the "Enable 0" bit of the configuration register.
1731 Format: 0 | 1 1731 Format: 0 | 1
1732 Default: 0 1732 Default: 0
1733 grcan.select= [HW] Select which phy 1733 grcan.select= [HW] Select which physical interface to use.
1734 Format: 0 | 1 1734 Format: 0 | 1
1735 Default: 0 1735 Default: 0
1736 grcan.txsize= [HW] Sets the size of 1736 grcan.txsize= [HW] Sets the size of the tx buffer.
1737 Format: <unsigned int 1737 Format: <unsigned int> such that (txsize & ~0x1fffc0) == 0.
1738 Default: 1024 1738 Default: 1024
1739 grcan.rxsize= [HW] Sets the size of 1739 grcan.rxsize= [HW] Sets the size of the rx buffer.
1740 Format: <unsigned int 1740 Format: <unsigned int> such that (rxsize & ~0x1fffc0) == 0.
1741 Default: 1024 1741 Default: 1024
1742 1742
1743 hardened_usercopy= 1743 hardened_usercopy=
1744 [KNL] Under CONFIG_HA 1744 [KNL] Under CONFIG_HARDENED_USERCOPY, whether
1745 hardening is enabled 1745 hardening is enabled for this boot. Hardened
1746 usercopy checking is 1746 usercopy checking is used to protect the kernel
1747 from reading or writi 1747 from reading or writing beyond known memory
1748 allocation boundaries 1748 allocation boundaries as a proactive defense
1749 against bounds-checki 1749 against bounds-checking flaws in the kernel's
1750 copy_to_user()/copy_f 1750 copy_to_user()/copy_from_user() interface.
1751 on Perform hardened user 1751 on Perform hardened usercopy checks (default).
1752 off Disable hardened user 1752 off Disable hardened usercopy checks.
1753 1753
1754 hardlockup_all_cpu_backtrace= 1754 hardlockup_all_cpu_backtrace=
1755 [KNL] Should the hard 1755 [KNL] Should the hard-lockup detector generate
1756 backtraces on all cpu 1756 backtraces on all cpus.
1757 Format: 0 | 1 1757 Format: 0 | 1
1758 1758
1759 hashdist= [KNL,NUMA] Large hash 1759 hashdist= [KNL,NUMA] Large hashes allocated during boot
1760 are distributed acros 1760 are distributed across NUMA nodes. Defaults on
1761 for 64-bit NUMA, off 1761 for 64-bit NUMA, off otherwise.
1762 Format: 0 | 1 (for of 1762 Format: 0 | 1 (for off | on)
1763 1763
1764 hd= [EIDE] (E)IDE hard dr 1764 hd= [EIDE] (E)IDE hard drive subsystem geometry
1765 Format: <cyl>,<head>, 1765 Format: <cyl>,<head>,<sect>
1766 1766
1767 hest_disable [ACPI] 1767 hest_disable [ACPI]
1768 Disable Hardware Erro 1768 Disable Hardware Error Source Table (HEST) support;
1769 corresponding firmwar 1769 corresponding firmware-first mode error processing
1770 logic will be disable 1770 logic will be disabled.
1771 1771
1772 hibernate= [HIBERNATION] 1772 hibernate= [HIBERNATION]
1773 noresume Don't check i 1773 noresume Don't check if there's a hibernation image
1774 present durin 1774 present during boot.
1775 nocompress Don't compres 1775 nocompress Don't compress/decompress hibernation images.
1776 no Disable hiber 1776 no Disable hibernation and resume.
1777 protect_image Turn on image 1777 protect_image Turn on image protection during restoration
1778 (that will se 1778 (that will set all pages holding image data
1779 during restor 1779 during restoration read-only).
1780 1780
1781 hibernate.compressor= [HIBERNATION] 1781 hibernate.compressor= [HIBERNATION] Compression algorithm to be
1782 used with hib 1782 used with hibernation.
1783 Format: { lzo 1783 Format: { lzo | lz4 }
1784 Default: lzo 1784 Default: lzo
1785 1785
1786 lzo: Select L 1786 lzo: Select LZO compression algorithm to
1787 compress/deco 1787 compress/decompress hibernation image.
1788 1788
1789 lz4: Select L 1789 lz4: Select LZ4 compression algorithm to
1790 compress/deco 1790 compress/decompress hibernation image.
1791 1791
1792 highmem=nn[KMG] [KNL,BOOT,EARLY] forc 1792 highmem=nn[KMG] [KNL,BOOT,EARLY] forces the highmem zone to have an exact
1793 size of <nn>. This wo 1793 size of <nn>. This works even on boxes that have no
1794 highmem otherwise. Th 1794 highmem otherwise. This also works to reduce highmem
1795 size on bigger boxes. 1795 size on bigger boxes.
1796 1796
1797 highres= [KNL] Enable/disable 1797 highres= [KNL] Enable/disable high resolution timer mode.
1798 Valid parameters: "on 1798 Valid parameters: "on", "off"
1799 Default: "on" 1799 Default: "on"
1800 1800
1801 hlt [BUGS=ARM,SH] 1801 hlt [BUGS=ARM,SH]
1802 1802
1803 hostname= [KNL,EARLY] Set the h 1803 hostname= [KNL,EARLY] Set the hostname (aka UTS nodename).
1804 Format: <string> 1804 Format: <string>
1805 This allows setting t 1805 This allows setting the system's hostname during early
1806 startup. This sets th 1806 startup. This sets the name returned by gethostname.
1807 Using this parameter 1807 Using this parameter to set the hostname makes it
1808 possible to ensure th 1808 possible to ensure the hostname is correctly set before
1809 any userspace process 1809 any userspace processes run, avoiding the possibility
1810 that a process may ca 1810 that a process may call gethostname before the hostname
1811 has been explicitly s 1811 has been explicitly set, resulting in the calling
1812 process getting an in 1812 process getting an incorrect result. The string must
1813 not exceed the maximu 1813 not exceed the maximum allowed hostname length (usually
1814 64 characters) and wi 1814 64 characters) and will be truncated otherwise.
1815 1815
1816 hpet= [X86-32,HPET] option 1816 hpet= [X86-32,HPET] option to control HPET usage
1817 Format: { enable (def 1817 Format: { enable (default) | disable | force |
1818 verbose } 1818 verbose }
1819 disable: disable HPET 1819 disable: disable HPET and use PIT instead
1820 force: allow force en 1820 force: allow force enabled of undocumented chips (ICH4,
1821 VIA, nVidia) 1821 VIA, nVidia)
1822 verbose: show content 1822 verbose: show contents of HPET registers during setup
1823 1823
1824 hpet_mmap= [X86, HPET_MMAP] Allo 1824 hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET
1825 registers. Default s 1825 registers. Default set by CONFIG_HPET_MMAP_DEFAULT.
1826 1826
1827 hugepages= [HW] Number of HugeTL 1827 hugepages= [HW] Number of HugeTLB pages to allocate at boot.
1828 If this follows hugep 1828 If this follows hugepagesz (below), it specifies
1829 the number of pages o 1829 the number of pages of hugepagesz to be allocated.
1830 If this is the first 1830 If this is the first HugeTLB parameter on the command
1831 line, it specifies th 1831 line, it specifies the number of pages to allocate for
1832 the default huge page 1832 the default huge page size. If using node format, the
1833 number of pages to al 1833 number of pages to allocate per-node can be specified.
1834 See also Documentatio 1834 See also Documentation/admin-guide/mm/hugetlbpage.rst.
1835 Format: <integer> or 1835 Format: <integer> or (node format)
1836 <node>:<integ 1836 <node>:<integer>[,<node>:<integer>]
1837 1837
1838 hugepagesz= 1838 hugepagesz=
1839 [HW] The size of the 1839 [HW] The size of the HugeTLB pages. This is used in
1840 conjunction with huge 1840 conjunction with hugepages (above) to allocate huge
1841 pages of a specific s 1841 pages of a specific size at boot. The pair
1842 hugepagesz=X hugepage 1842 hugepagesz=X hugepages=Y can be specified once for
1843 each supported huge p 1843 each supported huge page size. Huge page sizes are
1844 architecture dependen 1844 architecture dependent. See also
1845 Documentation/admin-g 1845 Documentation/admin-guide/mm/hugetlbpage.rst.
1846 Format: size[KMG] 1846 Format: size[KMG]
1847 1847
1848 hugetlb_cma= [HW,CMA,EARLY] The si 1848 hugetlb_cma= [HW,CMA,EARLY] The size of a CMA area used for allocation
1849 of gigantic hugepages 1849 of gigantic hugepages. Or using node format, the size
1850 of a CMA area per nod 1850 of a CMA area per node can be specified.
1851 Format: nn[KMGTPE] or 1851 Format: nn[KMGTPE] or (node format)
1852 <node>:nn[KMG 1852 <node>:nn[KMGTPE][,<node>:nn[KMGTPE]]
1853 1853
1854 Reserve a CMA area of 1854 Reserve a CMA area of given size and allocate gigantic
1855 hugepages using the C 1855 hugepages using the CMA allocator. If enabled, the
1856 boot-time allocation 1856 boot-time allocation of gigantic hugepages is skipped.
1857 1857
1858 hugetlb_free_vmemmap= 1858 hugetlb_free_vmemmap=
1859 [KNL] Requires CONFIG 1859 [KNL] Requires CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
1860 enabled. 1860 enabled.
1861 Control if HugeTLB Vm 1861 Control if HugeTLB Vmemmap Optimization (HVO) is enabled.
1862 Allows heavy hugetlb 1862 Allows heavy hugetlb users to free up some more
1863 memory (7 * PAGE_SIZE 1863 memory (7 * PAGE_SIZE for each 2MB hugetlb page).
1864 Format: { on | off (d 1864 Format: { on | off (default) }
1865 1865
1866 on: enable HVO 1866 on: enable HVO
1867 off: disable HVO 1867 off: disable HVO
1868 1868
1869 Built with CONFIG_HUG 1869 Built with CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON=y,
1870 the default is on. 1870 the default is on.
1871 1871
1872 Note that the vmemmap 1872 Note that the vmemmap pages may be allocated from the added
1873 memory block itself w 1873 memory block itself when memory_hotplug.memmap_on_memory is
1874 enabled, those vmemma 1874 enabled, those vmemmap pages cannot be optimized even if this
1875 feature is enabled. 1875 feature is enabled. Other vmemmap pages not allocated from
1876 the added memory bloc 1876 the added memory block itself do not be affected.
1877 1877
1878 hung_task_panic= 1878 hung_task_panic=
1879 [KNL] Should the hung 1879 [KNL] Should the hung task detector generate panics.
1880 Format: 0 | 1 1880 Format: 0 | 1
1881 1881
1882 A value of 1 instruct 1882 A value of 1 instructs the kernel to panic when a
1883 hung task is detected 1883 hung task is detected. The default value is controlled
1884 by the CONFIG_BOOTPAR 1884 by the CONFIG_BOOTPARAM_HUNG_TASK_PANIC build-time
1885 option. The value sel 1885 option. The value selected by this boot parameter can
1886 be changed later by t 1886 be changed later by the kernel.hung_task_panic sysctl.
1887 1887
1888 hvc_iucv= [S390] Number of z/V 1888 hvc_iucv= [S390] Number of z/VM IUCV hypervisor console (HVC)
1889 terminal devi 1889 terminal devices. Valid values: 0..8
1890 hvc_iucv_allow= [S390] Comma-separat 1890 hvc_iucv_allow= [S390] Comma-separated list of z/VM user IDs.
1891 If specified, 1891 If specified, z/VM IUCV HVC accepts connections
1892 from listed z 1892 from listed z/VM user IDs only.
1893 1893
1894 hv_nopvspin [X86,HYPER_V,EARLY] 1894 hv_nopvspin [X86,HYPER_V,EARLY]
1895 Disables the paravirt 1895 Disables the paravirt spinlock optimizations
1896 which allow the hyper 1896 which allow the hypervisor to 'idle' the guest
1897 on lock contention. 1897 on lock contention.
1898 1898
1899 i2c_bus= [HW] Override the 1899 i2c_bus= [HW] Override the default board specific I2C bus speed
1900 or register a 1900 or register an additional I2C bus that is not
1901 registered fr 1901 registered from board initialization code.
1902 Format: 1902 Format:
1903 <bus_id>,<clk 1903 <bus_id>,<clkrate>
1904 1904
1905 i2c_touchscreen_props= [HW,ACPI,X86] 1905 i2c_touchscreen_props= [HW,ACPI,X86]
1906 Set device-properties 1906 Set device-properties for ACPI-enumerated I2C-attached
1907 touchscreen, to e.g. 1907 touchscreen, to e.g. fix coordinates of upside-down
1908 mounted touchscreens. 1908 mounted touchscreens. If you need this option please
1909 submit a drivers/plat 1909 submit a drivers/platform/x86/touchscreen_dmi.c patch
1910 adding a DMI quirk fo 1910 adding a DMI quirk for this.
1911 1911
1912 Format: 1912 Format:
1913 <ACPI_HW_ID>:<prop_na 1913 <ACPI_HW_ID>:<prop_name>=<val>[:prop_name=val][:...]
1914 Where <val> is one of 1914 Where <val> is one of:
1915 Omit "=<val>" entirel 1915 Omit "=<val>" entirely Set a boolean device-property
1916 Unsigned number 1916 Unsigned number Set a u32 device-property
1917 Anything else 1917 Anything else Set a string device-property
1918 1918
1919 Examples (split over 1919 Examples (split over multiple lines):
1920 i2c_touchscreen_props 1920 i2c_touchscreen_props=GDIX1001:touchscreen-inverted-x:
1921 touchscreen-inverted- 1921 touchscreen-inverted-y
1922 1922
1923 i2c_touchscreen_props 1923 i2c_touchscreen_props=MSSL1680:touchscreen-size-x=1920:
1924 touchscreen-size-y=10 1924 touchscreen-size-y=1080:touchscreen-inverted-y:
1925 firmware-name=gsl1680 1925 firmware-name=gsl1680-vendor-model.fw:silead,home-button
1926 1926
1927 i8042.debug [HW] Toggle i8042 deb 1927 i8042.debug [HW] Toggle i8042 debug mode
1928 i8042.unmask_kbd_data 1928 i8042.unmask_kbd_data
1929 [HW] Enable printing 1929 [HW] Enable printing of interrupt data from the KBD port
1930 (disabled by def 1930 (disabled by default, and as a pre-condition
1931 requires that i8 1931 requires that i8042.debug=1 be enabled)
1932 i8042.direct [HW] Put keyboard por 1932 i8042.direct [HW] Put keyboard port into non-translated mode
1933 i8042.dumbkbd [HW] Pretend that con 1933 i8042.dumbkbd [HW] Pretend that controller can only read data from
1934 keyboard and can 1934 keyboard and cannot control its state
1935 (Don't attempt t 1935 (Don't attempt to blink the leds)
1936 i8042.noaux [HW] Don't check for 1936 i8042.noaux [HW] Don't check for auxiliary (== mouse) port
1937 i8042.nokbd [HW] Don't check/crea 1937 i8042.nokbd [HW] Don't check/create keyboard port
1938 i8042.noloop [HW] Disable the AUX 1938 i8042.noloop [HW] Disable the AUX Loopback command while probing
1939 for the AUX port 1939 for the AUX port
1940 i8042.nomux [HW] Don't check pres 1940 i8042.nomux [HW] Don't check presence of an active multiplexing
1941 controller 1941 controller
1942 i8042.nopnp [HW] Don't use ACPIPn 1942 i8042.nopnp [HW] Don't use ACPIPnP / PnPBIOS to discover KBD/AUX
1943 controllers 1943 controllers
1944 i8042.notimeout [HW] Ignore timeout c 1944 i8042.notimeout [HW] Ignore timeout condition signalled by controller
1945 i8042.reset [HW] Reset the contro 1945 i8042.reset [HW] Reset the controller during init, cleanup and
1946 suspend-to-ram t 1946 suspend-to-ram transitions, only during s2r
1947 transitions, or 1947 transitions, or never reset
1948 Format: { 1 | Y | y | 1948 Format: { 1 | Y | y | 0 | N | n }
1949 1, Y, y: always reset 1949 1, Y, y: always reset controller
1950 0, N, n: don't ever r 1950 0, N, n: don't ever reset controller
1951 Default: only on s2r 1951 Default: only on s2r transitions on x86; most other
1952 architectures force r 1952 architectures force reset to be always executed
1953 i8042.unlock [HW] Unlock (ignore) 1953 i8042.unlock [HW] Unlock (ignore) the keylock
1954 i8042.kbdreset [HW] Reset device con 1954 i8042.kbdreset [HW] Reset device connected to KBD port
1955 i8042.probe_defer 1955 i8042.probe_defer
1956 [HW] Allow deferred p 1956 [HW] Allow deferred probing upon i8042 probe errors
1957 1957
1958 i810= [HW,DRM] 1958 i810= [HW,DRM]
1959 1959
1960 i915.invert_brightness= 1960 i915.invert_brightness=
1961 [DRM] Invert the sens 1961 [DRM] Invert the sense of the variable that is used to
1962 set the brightness of 1962 set the brightness of the panel backlight. Normally a
1963 brightness value of 0 1963 brightness value of 0 indicates backlight switched off,
1964 and the maximum of th 1964 and the maximum of the brightness value sets the backlight
1965 to maximum brightness 1965 to maximum brightness. If this parameter is set to 0
1966 (default) and the mac 1966 (default) and the machine requires it, or this parameter
1967 is set to 1, a bright 1967 is set to 1, a brightness value of 0 sets the backlight
1968 to maximum brightness 1968 to maximum brightness, and the maximum of the brightness
1969 value switches the ba 1969 value switches the backlight off.
1970 -1 -- never invert br 1970 -1 -- never invert brightness
1971 0 -- machine default 1971 0 -- machine default
1972 1 -- force brightnes 1972 1 -- force brightness inversion
1973 1973
1974 ia32_emulation= [X86-64] 1974 ia32_emulation= [X86-64]
1975 Format: <bool> 1975 Format: <bool>
1976 When true, allows loa 1976 When true, allows loading 32-bit programs and executing 32-bit
1977 syscalls, essentially 1977 syscalls, essentially overriding IA32_EMULATION_DEFAULT_DISABLED at
1978 boot time. When false 1978 boot time. When false, unconditionally disables IA32 emulation.
1979 1979
1980 icn= [HW,ISDN] 1980 icn= [HW,ISDN]
1981 Format: <io>[,<membas 1981 Format: <io>[,<membase>[,<icn_id>[,<icn_id2>]]]
1982 1982
1983 1983
1984 idle= [X86,EARLY] 1984 idle= [X86,EARLY]
1985 Format: idle=poll, id 1985 Format: idle=poll, idle=halt, idle=nomwait
1986 Poll forces a polling 1986 Poll forces a polling idle loop that can slightly
1987 improve the performan 1987 improve the performance of waking up a idle CPU, but
1988 will use a lot of pow 1988 will use a lot of power and make the system run hot.
1989 Not recommended. 1989 Not recommended.
1990 idle=halt: Halt is fo 1990 idle=halt: Halt is forced to be used for CPU idle.
1991 In such case C2/C3 wo 1991 In such case C2/C3 won't be used again.
1992 idle=nomwait: Disable 1992 idle=nomwait: Disable mwait for CPU C-states
1993 1993
1994 idxd.sva= [HW] 1994 idxd.sva= [HW]
1995 Format: <bool> 1995 Format: <bool>
1996 Allow force disabling 1996 Allow force disabling of Shared Virtual Memory (SVA)
1997 support for the idxd 1997 support for the idxd driver. By default it is set to
1998 true (1). 1998 true (1).
1999 1999
2000 idxd.tc_override= [HW] 2000 idxd.tc_override= [HW]
2001 Format: <bool> 2001 Format: <bool>
2002 Allow override of def 2002 Allow override of default traffic class configuration
2003 for the device. By de 2003 for the device. By default it is set to false (0).
2004 2004
2005 ieee754= [MIPS] Select IEEE St 2005 ieee754= [MIPS] Select IEEE Std 754 conformance mode
2006 Format: { strict | le 2006 Format: { strict | legacy | 2008 | relaxed | emulated }
2007 Default: strict 2007 Default: strict
2008 2008
2009 Choose which programs 2009 Choose which programs will be accepted for execution
2010 based on the IEEE 754 2010 based on the IEEE 754 NaN encoding(s) supported by
2011 the FPU and the NaN e 2011 the FPU and the NaN encoding requested with the value
2012 of an ELF file header 2012 of an ELF file header flag individually set by each
2013 binary. Hardware imp 2013 binary. Hardware implementations are permitted to
2014 support either or bot 2014 support either or both of the legacy and the 2008 NaN
2015 encoding mode. 2015 encoding mode.
2016 2016
2017 Available settings ar 2017 Available settings are as follows:
2018 strict accept binari 2018 strict accept binaries that request a NaN encoding
2019 supported by 2019 supported by the FPU
2020 legacy only accept l 2020 legacy only accept legacy-NaN binaries, if supported
2021 by the FPU 2021 by the FPU
2022 2008 only accept 2 2022 2008 only accept 2008-NaN binaries, if supported
2023 by the FPU 2023 by the FPU
2024 relaxed accept any bi 2024 relaxed accept any binaries regardless of whether
2025 supported by 2025 supported by the FPU
2026 emulated accept any b 2026 emulated accept any binaries but enable FPU emulator
2027 if binary mod 2027 if binary mode is unsupported by the FPU.
2028 2028
2029 The FPU emulator is a 2029 The FPU emulator is always able to support both NaN
2030 encodings, so if no F 2030 encodings, so if no FPU hardware is present or it has
2031 been disabled with 'n 2031 been disabled with 'nofpu', then the settings of
2032 'legacy' and '2008' s 2032 'legacy' and '2008' strap the emulator accordingly,
2033 'relaxed' straps the 2033 'relaxed' straps the emulator for both legacy-NaN and
2034 2008-NaN, whereas 'st 2034 2008-NaN, whereas 'strict' enables legacy-NaN only on
2035 legacy processors and 2035 legacy processors and both NaN encodings on MIPS32 or
2036 MIPS64 CPUs. 2036 MIPS64 CPUs.
2037 2037
2038 The setting for ABS.f 2038 The setting for ABS.fmt/NEG.fmt instruction execution
2039 mode generally follow 2039 mode generally follows that for the NaN encoding,
2040 except where unsuppor 2040 except where unsupported by hardware.
2041 2041
2042 ignore_loglevel [KNL,EARLY] 2042 ignore_loglevel [KNL,EARLY]
2043 Ignore loglevel setti 2043 Ignore loglevel setting - this will print /all/
2044 kernel messages to th 2044 kernel messages to the console. Useful for debugging.
2045 We also add it as pri 2045 We also add it as printk module parameter, so users
2046 could change it dynam 2046 could change it dynamically, usually by
2047 /sys/module/printk/pa 2047 /sys/module/printk/parameters/ignore_loglevel.
2048 2048
2049 ignore_rlimit_data 2049 ignore_rlimit_data
2050 Ignore RLIMIT_DATA se 2050 Ignore RLIMIT_DATA setting for data mappings,
2051 print warning at firs 2051 print warning at first misuse. Can be changed via
2052 /sys/module/kernel/pa 2052 /sys/module/kernel/parameters/ignore_rlimit_data.
2053 2053
2054 ihash_entries= [KNL] 2054 ihash_entries= [KNL]
2055 Set number of hash bu 2055 Set number of hash buckets for inode cache.
2056 2056
2057 ima_appraise= [IMA] appraise integr 2057 ima_appraise= [IMA] appraise integrity measurements
2058 Format: { "off" | "en 2058 Format: { "off" | "enforce" | "fix" | "log" }
2059 default: "enforce" 2059 default: "enforce"
2060 2060
2061 ima_appraise_tcb [IMA] Deprecated. U 2061 ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead.
2062 The builtin appraise 2062 The builtin appraise policy appraises all files
2063 owned by uid=0. 2063 owned by uid=0.
2064 2064
2065 ima_canonical_fmt [IMA] 2065 ima_canonical_fmt [IMA]
2066 Use the canonical for 2066 Use the canonical format for the binary runtime
2067 measurements, instead 2067 measurements, instead of host native format.
2068 2068
2069 ima_hash= [IMA] 2069 ima_hash= [IMA]
2070 Format: { md5 | sha1 2070 Format: { md5 | sha1 | rmd160 | sha256 | sha384
2071 | sha512 | 2071 | sha512 | ... }
2072 default: "sha1" 2072 default: "sha1"
2073 2073
2074 The list of supported 2074 The list of supported hash algorithms is defined
2075 in crypto/hash_info.h 2075 in crypto/hash_info.h.
2076 2076
2077 ima_policy= [IMA] 2077 ima_policy= [IMA]
2078 The builtin policies 2078 The builtin policies to load during IMA setup.
2079 Format: "tcb | apprai 2079 Format: "tcb | appraise_tcb | secure_boot |
2080 fail_securel 2080 fail_securely | critical_data"
2081 2081
2082 The "tcb" policy meas 2082 The "tcb" policy measures all programs exec'd, files
2083 mmap'd for exec, and 2083 mmap'd for exec, and all files opened with the read
2084 mode bit set by eithe 2084 mode bit set by either the effective uid (euid=0) or
2085 uid=0. 2085 uid=0.
2086 2086
2087 The "appraise_tcb" po 2087 The "appraise_tcb" policy appraises the integrity of
2088 all files owned by ro 2088 all files owned by root.
2089 2089
2090 The "secure_boot" pol 2090 The "secure_boot" policy appraises the integrity
2091 of files (eg. kexec k 2091 of files (eg. kexec kernel image, kernel modules,
2092 firmware, policy, etc 2092 firmware, policy, etc) based on file signatures.
2093 2093
2094 The "fail_securely" p 2094 The "fail_securely" policy forces file signature
2095 verification failure 2095 verification failure also on privileged mounted
2096 filesystems with the 2096 filesystems with the SB_I_UNVERIFIABLE_SIGNATURE
2097 flag. 2097 flag.
2098 2098
2099 The "critical_data" p 2099 The "critical_data" policy measures kernel integrity
2100 critical data. 2100 critical data.
2101 2101
2102 ima_tcb [IMA] Deprecated. Us 2102 ima_tcb [IMA] Deprecated. Use ima_policy= instead.
2103 Load a policy which m 2103 Load a policy which meets the needs of the Trusted
2104 Computing Base. This 2104 Computing Base. This means IMA will measure all
2105 programs exec'd, file 2105 programs exec'd, files mmap'd for exec, and all files
2106 opened for read by ui 2106 opened for read by uid=0.
2107 2107
2108 ima_template= [IMA] 2108 ima_template= [IMA]
2109 Select one of defined 2109 Select one of defined IMA measurements template formats.
2110 Formats: { "ima" | "i 2110 Formats: { "ima" | "ima-ng" | "ima-ngv2" | "ima-sig" |
2111 "ima-sigv2 2111 "ima-sigv2" }
2112 Default: "ima-ng" 2112 Default: "ima-ng"
2113 2113
2114 ima_template_fmt= 2114 ima_template_fmt=
2115 [IMA] Define a custom 2115 [IMA] Define a custom template format.
2116 Format: { "field1|... 2116 Format: { "field1|...|fieldN" }
2117 2117
2118 ima.ahash_minsize= [IMA] Minimum file 2118 ima.ahash_minsize= [IMA] Minimum file size for asynchronous hash usage
2119 Format: <min_file_siz 2119 Format: <min_file_size>
2120 Set the minimal file 2120 Set the minimal file size for using asynchronous hash.
2121 If left unspecified, 2121 If left unspecified, ahash usage is disabled.
2122 2122
2123 ahash performance var 2123 ahash performance varies for different data sizes on
2124 different crypto acce 2124 different crypto accelerators. This option can be used
2125 to achieve the best p 2125 to achieve the best performance for a particular HW.
2126 2126
2127 ima.ahash_bufsize= [IMA] Asynchronous 2127 ima.ahash_bufsize= [IMA] Asynchronous hash buffer size
2128 Format: <bufsize> 2128 Format: <bufsize>
2129 Set hashing buffer si 2129 Set hashing buffer size. Default: 4k.
2130 2130
2131 ahash performance var 2131 ahash performance varies for different chunk sizes on
2132 different crypto acce 2132 different crypto accelerators. This option can be used
2133 to achieve best perfo 2133 to achieve best performance for particular HW.
2134 2134
2135 init= [KNL] 2135 init= [KNL]
2136 Format: <full_path> 2136 Format: <full_path>
2137 Run specified binary 2137 Run specified binary instead of /sbin/init as init
2138 process. 2138 process.
2139 2139
2140 initcall_debug [KNL] Trace initcalls 2140 initcall_debug [KNL] Trace initcalls as they are executed. Useful
2141 for working out where 2141 for working out where the kernel is dying during
2142 startup. 2142 startup.
2143 2143
2144 initcall_blacklist= [KNL] Do not exe 2144 initcall_blacklist= [KNL] Do not execute a comma-separated list of
2145 initcall functions. 2145 initcall functions. Useful for debugging built-in
2146 modules and initcalls 2146 modules and initcalls.
2147 2147
2148 initramfs_async= [KNL] 2148 initramfs_async= [KNL]
2149 Format: <bool> 2149 Format: <bool>
2150 Default: 1 2150 Default: 1
2151 This parameter contro 2151 This parameter controls whether the initramfs
2152 image is unpacked asy 2152 image is unpacked asynchronously, concurrently
2153 with devices being pr 2153 with devices being probed and
2154 initialized. This sho 2154 initialized. This should normally just work,
2155 but as a debugging ai 2155 but as a debugging aid, one can get the
2156 historical behaviour 2156 historical behaviour of the initramfs
2157 unpacking being compl 2157 unpacking being completed before device_ and
2158 late_ initcalls. 2158 late_ initcalls.
2159 2159
2160 initrd= [BOOT,EARLY] Specify 2160 initrd= [BOOT,EARLY] Specify the location of the initial ramdisk
2161 2161
2162 initrdmem= [KNL,EARLY] Specify a 2162 initrdmem= [KNL,EARLY] Specify a physical address and size from which to
2163 load the initrd. If a 2163 load the initrd. If an initrd is compiled in or
2164 specified in the boot 2164 specified in the bootparams, it takes priority over this
2165 setting. 2165 setting.
2166 Format: ss[KMG],nn[KM 2166 Format: ss[KMG],nn[KMG]
2167 Default is 0, 0 2167 Default is 0, 0
2168 2168
2169 init_on_alloc= [MM,EARLY] Fill newly 2169 init_on_alloc= [MM,EARLY] Fill newly allocated pages and heap objects with
2170 zeroes. 2170 zeroes.
2171 Format: 0 | 1 2171 Format: 0 | 1
2172 Default set by CONFIG 2172 Default set by CONFIG_INIT_ON_ALLOC_DEFAULT_ON.
2173 2173
2174 init_on_free= [MM,EARLY] Fill freed 2174 init_on_free= [MM,EARLY] Fill freed pages and heap objects with zeroes.
2175 Format: 0 | 1 2175 Format: 0 | 1
2176 Default set by CONFIG 2176 Default set by CONFIG_INIT_ON_FREE_DEFAULT_ON.
2177 2177
2178 init_pkru= [X86] Specify the def 2178 init_pkru= [X86] Specify the default memory protection keys rights
2179 register contents for 2179 register contents for all processes. 0x55555554 by
2180 default (disallow acc 2180 default (disallow access to all but pkey 0). Can
2181 override in debugfs a 2181 override in debugfs after boot.
2182 2182
2183 inport.irq= [HW] Inport (ATI XL a 2183 inport.irq= [HW] Inport (ATI XL and Microsoft) busmouse driver
2184 Format: <irq> 2184 Format: <irq>
2185 2185
2186 int_pln_enable [X86] Enable power li 2186 int_pln_enable [X86] Enable power limit notification interrupt
2187 2187
2188 integrity_audit=[IMA] 2188 integrity_audit=[IMA]
2189 Format: { "0" | "1" } 2189 Format: { "0" | "1" }
2190 0 -- basic integrity 2190 0 -- basic integrity auditing messages. (Default)
2191 1 -- additional integ 2191 1 -- additional integrity auditing messages.
2192 2192
2193 intel_iommu= [DMAR] Intel IOMMU dr 2193 intel_iommu= [DMAR] Intel IOMMU driver (DMAR) option
2194 on 2194 on
2195 Enable intel iommu dr 2195 Enable intel iommu driver.
2196 off 2196 off
2197 Disable intel iommu d 2197 Disable intel iommu driver.
2198 igfx_off [Default Off] 2198 igfx_off [Default Off]
2199 By default, gfx is ma 2199 By default, gfx is mapped as normal device. If a gfx
2200 device has a dedicate 2200 device has a dedicated DMAR unit, the DMAR unit is
2201 bypassed by not enabl 2201 bypassed by not enabling DMAR with this option. In
2202 this case, gfx device 2202 this case, gfx device will use physical address for
2203 DMA. 2203 DMA.
2204 strict [Default Off] 2204 strict [Default Off]
2205 Deprecated, equivalen 2205 Deprecated, equivalent to iommu.strict=1.
2206 sp_off [Default Off] 2206 sp_off [Default Off]
2207 By default, super pag 2207 By default, super page will be supported if Intel IOMMU
2208 has the capability. W 2208 has the capability. With this option, super page will
2209 not be supported. 2209 not be supported.
2210 sm_on 2210 sm_on
2211 Enable the Intel IOMM 2211 Enable the Intel IOMMU scalable mode if the hardware
2212 advertises that it ha 2212 advertises that it has support for the scalable mode
2213 translation. 2213 translation.
2214 sm_off 2214 sm_off
2215 Disallow use of the I 2215 Disallow use of the Intel IOMMU scalable mode.
2216 tboot_noforce [Default Off] 2216 tboot_noforce [Default Off]
2217 Do not force the Inte 2217 Do not force the Intel IOMMU enabled under tboot.
2218 By default, tboot wil 2218 By default, tboot will force Intel IOMMU on, which
2219 could harm performanc 2219 could harm performance of some high-throughput
2220 devices like 40GBit n 2220 devices like 40GBit network cards, even if identity
2221 mapping is enabled. 2221 mapping is enabled.
2222 Note that using this 2222 Note that using this option lowers the security
2223 provided by tboot bec 2223 provided by tboot because it makes the system
2224 vulnerable to DMA att 2224 vulnerable to DMA attacks.
2225 2225
2226 intel_idle.max_cstate= [KNL,HW,ACPI, 2226 intel_idle.max_cstate= [KNL,HW,ACPI,X86]
2227 0 disables inte 2227 0 disables intel_idle and fall back on acpi_idle.
2228 1 to 9 specify maxim 2228 1 to 9 specify maximum depth of C-state.
2229 2229
2230 intel_pstate= [X86,EARLY] 2230 intel_pstate= [X86,EARLY]
2231 disable 2231 disable
2232 Do not enable intel 2232 Do not enable intel_pstate as the default
2233 scaling driver for 2233 scaling driver for the supported processors
2234 active 2234 active
2235 Use intel_pstate dr 2235 Use intel_pstate driver to bypass the scaling
2236 governors layer of 2236 governors layer of cpufreq and provides it own
2237 algorithms for p-st 2237 algorithms for p-state selection. There are two
2238 P-state selection a 2238 P-state selection algorithms provided by
2239 intel_pstate in the 2239 intel_pstate in the active mode: powersave and
2240 performance. The w 2240 performance. The way they both operate depends
2241 on whether or not t 2241 on whether or not the hardware managed P-states
2242 (HWP) feature has b 2242 (HWP) feature has been enabled in the processor
2243 and possibly on the 2243 and possibly on the processor model.
2244 passive 2244 passive
2245 Use intel_pstate as 2245 Use intel_pstate as a scaling driver, but configure it
2246 to work with generi 2246 to work with generic cpufreq governors (instead of
2247 enabling its intern 2247 enabling its internal governor). This mode cannot be
2248 used along with the 2248 used along with the hardware-managed P-states (HWP)
2249 feature. 2249 feature.
2250 force 2250 force
2251 Enable intel_pstate 2251 Enable intel_pstate on systems that prohibit it by default
2252 in favor of acpi-cp 2252 in favor of acpi-cpufreq. Forcing the intel_pstate driver
2253 instead of acpi-cpu 2253 instead of acpi-cpufreq may disable platform features, such
2254 as thermal controls 2254 as thermal controls and power capping, that rely on ACPI
2255 P-States informatio 2255 P-States information being indicated to OSPM and therefore
2256 should be used with 2256 should be used with caution. This option does not work with
2257 processors that are 2257 processors that aren't supported by the intel_pstate driver
2258 or on platforms tha 2258 or on platforms that use pcc-cpufreq instead of acpi-cpufreq.
2259 no_hwp 2259 no_hwp
2260 Do not enable hardw 2260 Do not enable hardware P state control (HWP)
2261 if available. 2261 if available.
2262 hwp_only 2262 hwp_only
2263 Only load intel_pst 2263 Only load intel_pstate on systems which support
2264 hardware P state co 2264 hardware P state control (HWP) if available.
2265 support_acpi_ppc 2265 support_acpi_ppc
2266 Enforce ACPI _PPC p 2266 Enforce ACPI _PPC performance limits. If the Fixed ACPI
2267 Description Table, 2267 Description Table, specifies preferred power management
2268 profile as "Enterpr 2268 profile as "Enterprise Server" or "Performance Server",
2269 then this feature i 2269 then this feature is turned on by default.
2270 per_cpu_perf_limits 2270 per_cpu_perf_limits
2271 Allow per-logical-C 2271 Allow per-logical-CPU P-State performance control limits using
2272 cpufreq sysfs inter 2272 cpufreq sysfs interface
2273 2273
2274 intremap= [X86-64,Intel-IOMMU,E 2274 intremap= [X86-64,Intel-IOMMU,EARLY]
2275 on enable Interr 2275 on enable Interrupt Remapping (default)
2276 off disable Inter 2276 off disable Interrupt Remapping
2277 nosid disable Sourc 2277 nosid disable Source ID checking
2278 no_x2apic_optout 2278 no_x2apic_optout
2279 BIOS x2APIC o 2279 BIOS x2APIC opt-out request will be ignored
2280 nopost disable Inter 2280 nopost disable Interrupt Posting
2281 posted_msi 2281 posted_msi
2282 enable MSIs d 2282 enable MSIs delivered as posted interrupts
2283 2283
2284 iomem= Disable strict checki 2284 iomem= Disable strict checking of access to MMIO memory
2285 strict regions from userspac 2285 strict regions from userspace.
2286 relaxed 2286 relaxed
2287 2287
2288 iommu= [X86,EARLY] 2288 iommu= [X86,EARLY]
2289 off 2289 off
2290 force 2290 force
2291 noforce 2291 noforce
2292 biomerge 2292 biomerge
2293 panic 2293 panic
2294 nopanic 2294 nopanic
2295 merge 2295 merge
2296 nomerge 2296 nomerge
2297 soft 2297 soft
2298 pt [X86] 2298 pt [X86]
2299 nopt [X86] 2299 nopt [X86]
2300 nobypass [PPC/POWERNV] 2300 nobypass [PPC/POWERNV]
2301 Disable IOMMU bypass, 2301 Disable IOMMU bypass, using IOMMU for PCI devices.
2302 2302
2303 iommu.forcedac= [ARM64,X86,EARLY] Con 2303 iommu.forcedac= [ARM64,X86,EARLY] Control IOVA allocation for PCI devices.
2304 Format: { "0" | "1" } 2304 Format: { "0" | "1" }
2305 0 - Try to allocate a 2305 0 - Try to allocate a 32-bit DMA address first, before
2306 falling back to the 2306 falling back to the full range if needed.
2307 1 - Allocate directly 2307 1 - Allocate directly from the full usable range,
2308 forcing Dual Addres 2308 forcing Dual Address Cycle for PCI cards supporting
2309 greater than 32-bit 2309 greater than 32-bit addressing.
2310 2310
2311 iommu.strict= [ARM64,X86,S390,EARLY 2311 iommu.strict= [ARM64,X86,S390,EARLY] Configure TLB invalidation behaviour
2312 Format: { "0" | "1" } 2312 Format: { "0" | "1" }
2313 0 - Lazy mode. 2313 0 - Lazy mode.
2314 Request that DMA un 2314 Request that DMA unmap operations use deferred
2315 invalidation of har 2315 invalidation of hardware TLBs, for increased
2316 throughput at the c 2316 throughput at the cost of reduced device isolation.
2317 Will fall back to s 2317 Will fall back to strict mode if not supported by
2318 the relevant IOMMU 2318 the relevant IOMMU driver.
2319 1 - Strict mode. 2319 1 - Strict mode.
2320 DMA unmap operation 2320 DMA unmap operations invalidate IOMMU hardware TLBs
2321 synchronously. 2321 synchronously.
2322 unset - Use value of 2322 unset - Use value of CONFIG_IOMMU_DEFAULT_DMA_{LAZY,STRICT}.
2323 Note: on x86, strict 2323 Note: on x86, strict mode specified via one of the
2324 legacy driver-specifi 2324 legacy driver-specific options takes precedence.
2325 2325
2326 iommu.passthrough= 2326 iommu.passthrough=
2327 [ARM64,X86,EARLY] Con 2327 [ARM64,X86,EARLY] Configure DMA to bypass the IOMMU by default.
2328 Format: { "0" | "1" } 2328 Format: { "0" | "1" }
2329 0 - Use IOMMU transla 2329 0 - Use IOMMU translation for DMA.
2330 1 - Bypass the IOMMU 2330 1 - Bypass the IOMMU for DMA.
2331 unset - Use value of 2331 unset - Use value of CONFIG_IOMMU_DEFAULT_PASSTHROUGH.
2332 2332
2333 io7= [HW] IO7 for Marvel-b 2333 io7= [HW] IO7 for Marvel-based Alpha systems
2334 See comment before ma 2334 See comment before marvel_specify_io7 in
2335 arch/alpha/kernel/cor 2335 arch/alpha/kernel/core_marvel.c.
2336 2336
2337 io_delay= [X86,EARLY] I/O delay 2337 io_delay= [X86,EARLY] I/O delay method
2338 0x80 2338 0x80
2339 Standard port 0x80 ba 2339 Standard port 0x80 based delay
2340 0xed 2340 0xed
2341 Alternate port 0xed b 2341 Alternate port 0xed based delay (needed on some systems)
2342 udelay 2342 udelay
2343 Simple two microsecon 2343 Simple two microseconds delay
2344 none 2344 none
2345 No delay 2345 No delay
2346 2346
2347 ip= [IP_PNP] 2347 ip= [IP_PNP]
2348 See Documentation/adm 2348 See Documentation/admin-guide/nfs/nfsroot.rst.
2349 2349
2350 ipcmni_extend [KNL,EARLY] Extend th 2350 ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V
2351 IPC identifiers from 2351 IPC identifiers from 32,768 to 16,777,216.
2352 2352
2353 irqaffinity= [SMP] Set the default 2353 irqaffinity= [SMP] Set the default irq affinity mask
2354 The argument is a cpu 2354 The argument is a cpu list, as described above.
2355 2355
2356 irqchip.gicv2_force_probe= 2356 irqchip.gicv2_force_probe=
2357 [ARM,ARM64,EARLY] 2357 [ARM,ARM64,EARLY]
2358 Format: <bool> 2358 Format: <bool>
2359 Force the kernel to l 2359 Force the kernel to look for the second 4kB page
2360 of a GICv2 controller 2360 of a GICv2 controller even if the memory range
2361 exposed by the device 2361 exposed by the device tree is too small.
2362 2362
2363 irqchip.gicv3_nolpi= 2363 irqchip.gicv3_nolpi=
2364 [ARM,ARM64,EARLY] 2364 [ARM,ARM64,EARLY]
2365 Force the kernel to i 2365 Force the kernel to ignore the availability of
2366 LPIs (and by conseque 2366 LPIs (and by consequence ITSs). Intended for system
2367 that use the kernel a 2367 that use the kernel as a bootloader, and thus want
2368 to let secondary kern 2368 to let secondary kernels in charge of setting up
2369 LPIs. 2369 LPIs.
2370 2370
2371 irqchip.gicv3_pseudo_nmi= [ARM64,EARL 2371 irqchip.gicv3_pseudo_nmi= [ARM64,EARLY]
2372 Enables support for p 2372 Enables support for pseudo-NMIs in the kernel. This
2373 requires the kernel t 2373 requires the kernel to be built with
2374 CONFIG_ARM64_PSEUDO_N 2374 CONFIG_ARM64_PSEUDO_NMI.
2375 2375
2376 irqfixup [HW] 2376 irqfixup [HW]
2377 When an interrupt is 2377 When an interrupt is not handled search all handlers
2378 for it. Intended to g 2378 for it. Intended to get systems with badly broken
2379 firmware running. 2379 firmware running.
2380 2380
2381 irqpoll [HW] 2381 irqpoll [HW]
2382 When an interrupt is 2382 When an interrupt is not handled search all handlers
2383 for it. Also check al 2383 for it. Also check all handlers each timer
2384 interrupt. Intended t 2384 interrupt. Intended to get systems with badly broken
2385 firmware running. 2385 firmware running.
2386 2386
2387 isapnp= [ISAPNP] 2387 isapnp= [ISAPNP]
2388 Format: <RDP>,<reset> 2388 Format: <RDP>,<reset>,<pci_scan>,<verbosity>
2389 2389
2390 isolcpus= [KNL,SMP,ISOL] Isolat 2390 isolcpus= [KNL,SMP,ISOL] Isolate a given set of CPUs from disturbance.
2391 [Deprecated - use cpu 2391 [Deprecated - use cpusets instead]
2392 Format: [flag-list,]< 2392 Format: [flag-list,]<cpu-list>
2393 2393
2394 Specify one or more C 2394 Specify one or more CPUs to isolate from disturbances
2395 specified in the flag 2395 specified in the flag list (default: domain):
2396 2396
2397 nohz 2397 nohz
2398 Disable the tick wh 2398 Disable the tick when a single task runs.
2399 2399
2400 A residual 1Hz tick 2400 A residual 1Hz tick is offloaded to workqueues, which you
2401 need to affine to h 2401 need to affine to housekeeping through the global
2402 workqueue's affinit 2402 workqueue's affinity configured via the
2403 /sys/devices/virtua 2403 /sys/devices/virtual/workqueue/cpumask sysfs file, or
2404 by using the 'domai 2404 by using the 'domain' flag described below.
2405 2405
2406 NOTE: by default th 2406 NOTE: by default the global workqueue runs on all CPUs,
2407 so to protect indiv 2407 so to protect individual CPUs the 'cpumask' file has to
2408 be configured manua 2408 be configured manually after bootup.
2409 2409
2410 domain 2410 domain
2411 Isolate from the ge 2411 Isolate from the general SMP balancing and scheduling
2412 algorithms. Note th 2412 algorithms. Note that performing domain isolation this way
2413 is irreversible: it 2413 is irreversible: it's not possible to bring back a CPU to
2414 the domains once is 2414 the domains once isolated through isolcpus. It's strongly
2415 advised to use cpus 2415 advised to use cpusets instead to disable scheduler load
2416 balancing through t 2416 balancing through the "cpuset.sched_load_balance" file.
2417 It offers a much mo 2417 It offers a much more flexible interface where CPUs can
2418 move in and out of 2418 move in and out of an isolated set anytime.
2419 2419
2420 You can move a proc 2420 You can move a process onto or off an "isolated" CPU via
2421 the CPU affinity sy 2421 the CPU affinity syscalls or cpuset.
2422 <cpu number> begins 2422 <cpu number> begins at 0 and the maximum value is
2423 "number of CPUs in 2423 "number of CPUs in system - 1".
2424 2424
2425 managed_irq 2425 managed_irq
2426 2426
2427 Isolate from being 2427 Isolate from being targeted by managed interrupts
2428 which have an inter 2428 which have an interrupt mask containing isolated
2429 CPUs. The affinity 2429 CPUs. The affinity of managed interrupts is
2430 handled by the kern 2430 handled by the kernel and cannot be changed via
2431 the /proc/irq/* int 2431 the /proc/irq/* interfaces.
2432 2432
2433 This isolation is b 2433 This isolation is best effort and only effective
2434 if the automaticall 2434 if the automatically assigned interrupt mask of a
2435 device queue contai 2435 device queue contains isolated and housekeeping
2436 CPUs. If housekeepi 2436 CPUs. If housekeeping CPUs are online then such
2437 interrupts are dire 2437 interrupts are directed to the housekeeping CPU
2438 so that IO submitte 2438 so that IO submitted on the housekeeping CPU
2439 cannot disturb the 2439 cannot disturb the isolated CPU.
2440 2440
2441 If a queue's affini 2441 If a queue's affinity mask contains only isolated
2442 CPUs then this para 2442 CPUs then this parameter has no effect on the
2443 interrupt routing d 2443 interrupt routing decision, though interrupts are
2444 only delivered when 2444 only delivered when tasks running on those
2445 isolated CPUs submi 2445 isolated CPUs submit IO. IO submitted on
2446 housekeeping CPUs h 2446 housekeeping CPUs has no influence on those
2447 queues. 2447 queues.
2448 2448
2449 The format of <cpu-li 2449 The format of <cpu-list> is described above.
2450 2450
2451 iucv= [HW,NET] 2451 iucv= [HW,NET]
2452 2452
2453 ivrs_ioapic [HW,X86-64] 2453 ivrs_ioapic [HW,X86-64]
2454 Provide an override t 2454 Provide an override to the IOAPIC-ID<->DEVICE-ID
2455 mapping provided in t 2455 mapping provided in the IVRS ACPI table.
2456 By default, PCI segme 2456 By default, PCI segment is 0, and can be omitted.
2457 2457
2458 For example, to map I 2458 For example, to map IOAPIC-ID decimal 10 to
2459 PCI segment 0x1 and P 2459 PCI segment 0x1 and PCI device 00:14.0,
2460 write the parameter a 2460 write the parameter as:
2461 ivrs_ioapic=1 2461 ivrs_ioapic=10@0001:00:14.0
2462 2462
2463 Deprecated formats: 2463 Deprecated formats:
2464 * To map IOAPIC-ID de 2464 * To map IOAPIC-ID decimal 10 to PCI device 00:14.0
2465 write the parameter 2465 write the parameter as:
2466 ivrs_ioapic[1 2466 ivrs_ioapic[10]=00:14.0
2467 * To map IOAPIC-ID de 2467 * To map IOAPIC-ID decimal 10 to PCI segment 0x1 and
2468 PCI device 00:14.0 2468 PCI device 00:14.0 write the parameter as:
2469 ivrs_ioapic[1 2469 ivrs_ioapic[10]=0001:00:14.0
2470 2470
2471 ivrs_hpet [HW,X86-64] 2471 ivrs_hpet [HW,X86-64]
2472 Provide an override t 2472 Provide an override to the HPET-ID<->DEVICE-ID
2473 mapping provided in t 2473 mapping provided in the IVRS ACPI table.
2474 By default, PCI segme 2474 By default, PCI segment is 0, and can be omitted.
2475 2475
2476 For example, to map H 2476 For example, to map HPET-ID decimal 10 to
2477 PCI segment 0x1 and P 2477 PCI segment 0x1 and PCI device 00:14.0,
2478 write the parameter a 2478 write the parameter as:
2479 ivrs_hpet=10@ 2479 ivrs_hpet=10@0001:00:14.0
2480 2480
2481 Deprecated formats: 2481 Deprecated formats:
2482 * To map HPET-ID deci 2482 * To map HPET-ID decimal 0 to PCI device 00:14.0
2483 write the parameter 2483 write the parameter as:
2484 ivrs_hpet[0]= 2484 ivrs_hpet[0]=00:14.0
2485 * To map HPET-ID deci 2485 * To map HPET-ID decimal 10 to PCI segment 0x1 and
2486 PCI device 00:14.0 2486 PCI device 00:14.0 write the parameter as:
2487 ivrs_ioapic[1 2487 ivrs_ioapic[10]=0001:00:14.0
2488 2488
2489 ivrs_acpihid [HW,X86-64] 2489 ivrs_acpihid [HW,X86-64]
2490 Provide an override t 2490 Provide an override to the ACPI-HID:UID<->DEVICE-ID
2491 mapping provided in t 2491 mapping provided in the IVRS ACPI table.
2492 By default, PCI segme 2492 By default, PCI segment is 0, and can be omitted.
2493 2493
2494 For example, to map U 2494 For example, to map UART-HID:UID AMD0020:0 to
2495 PCI segment 0x1 and P 2495 PCI segment 0x1 and PCI device ID 00:14.5,
2496 write the parameter a 2496 write the parameter as:
2497 ivrs_acpihid= 2497 ivrs_acpihid=AMD0020:0@0001:00:14.5
2498 2498
2499 Deprecated formats: 2499 Deprecated formats:
2500 * To map UART-HID:UID 2500 * To map UART-HID:UID AMD0020:0 to PCI segment is 0,
2501 PCI device ID 00:14 2501 PCI device ID 00:14.5, write the parameter as:
2502 ivrs_acpihid[ 2502 ivrs_acpihid[00:14.5]=AMD0020:0
2503 * To map UART-HID:UID 2503 * To map UART-HID:UID AMD0020:0 to PCI segment 0x1 and
2504 PCI device ID 00:14 2504 PCI device ID 00:14.5, write the parameter as:
2505 ivrs_acpihid[ 2505 ivrs_acpihid[0001:00:14.5]=AMD0020:0
2506 2506
2507 js= [HW,JOY] Analog joyst 2507 js= [HW,JOY] Analog joystick
2508 See Documentation/inp 2508 See Documentation/input/joydev/joystick.rst.
2509 2509
2510 kasan_multi_shot 2510 kasan_multi_shot
2511 [KNL] Enforce KASAN ( 2511 [KNL] Enforce KASAN (Kernel Address Sanitizer) to print
2512 report on every inval 2512 report on every invalid memory access. Without this
2513 parameter KASAN will 2513 parameter KASAN will print report only for the first
2514 invalid access. 2514 invalid access.
2515 2515
2516 keep_bootcon [KNL,EARLY] 2516 keep_bootcon [KNL,EARLY]
2517 Do not unregister boo 2517 Do not unregister boot console at start. This is only
2518 useful for debugging 2518 useful for debugging when something happens in the window
2519 between unregistering 2519 between unregistering the boot console and initializing
2520 the real console. 2520 the real console.
2521 2521
2522 keepinitrd [HW,ARM] See retain_i 2522 keepinitrd [HW,ARM] See retain_initrd.
2523 2523
2524 kernelcore= [KNL,X86,PPC,EARLY] 2524 kernelcore= [KNL,X86,PPC,EARLY]
2525 Format: nn[KMGTPE] | 2525 Format: nn[KMGTPE] | nn% | "mirror"
2526 This parameter specif 2526 This parameter specifies the amount of memory usable by
2527 the kernel for non-mo 2527 the kernel for non-movable allocations. The requested
2528 amount is spread even 2528 amount is spread evenly throughout all nodes in the
2529 system as ZONE_NORMAL 2529 system as ZONE_NORMAL. The remaining memory is used for
2530 movable memory in its 2530 movable memory in its own zone, ZONE_MOVABLE. In the
2531 event, a node is too 2531 event, a node is too small to have both ZONE_NORMAL and
2532 ZONE_MOVABLE, kernelc 2532 ZONE_MOVABLE, kernelcore memory will take priority and
2533 other nodes will have 2533 other nodes will have a larger ZONE_MOVABLE.
2534 2534
2535 ZONE_MOVABLE is used 2535 ZONE_MOVABLE is used for the allocation of pages that
2536 may be reclaimed or m 2536 may be reclaimed or moved by the page migration
2537 subsystem. Note that 2537 subsystem. Note that allocations like PTEs-from-HighMem
2538 still use the HighMem 2538 still use the HighMem zone if it exists, and the Normal
2539 zone if it does not. 2539 zone if it does not.
2540 2540
2541 It is possible to spe 2541 It is possible to specify the exact amount of memory in
2542 the form of "nn[KMGTP 2542 the form of "nn[KMGTPE]", a percentage of total system
2543 memory in the form of 2543 memory in the form of "nn%", or "mirror". If "mirror"
2544 option is specified, 2544 option is specified, mirrored (reliable) memory is used
2545 for non-movable alloc 2545 for non-movable allocations and remaining memory is used
2546 for Movable pages. " 2546 for Movable pages. "nn[KMGTPE]", "nn%", and "mirror"
2547 are exclusive, so you 2547 are exclusive, so you cannot specify multiple forms.
2548 2548
2549 kgdbdbgp= [KGDB,HW,EARLY] kgdb 2549 kgdbdbgp= [KGDB,HW,EARLY] kgdb over EHCI usb debug port.
2550 Format: <Controller#> 2550 Format: <Controller#>[,poll interval]
2551 The controller # is t 2551 The controller # is the number of the ehci usb debug
2552 port as it is probed 2552 port as it is probed via PCI. The poll interval is
2553 optional and is the n 2553 optional and is the number seconds in between
2554 each poll cycle to th 2554 each poll cycle to the debug port in case you need
2555 the functionality for 2555 the functionality for interrupting the kernel with
2556 gdb or control-c on t 2556 gdb or control-c on the dbgp connection. When
2557 not using this parame 2557 not using this parameter you use sysrq-g to break into
2558 the kernel debugger. 2558 the kernel debugger.
2559 2559
2560 kgdboc= [KGDB,HW] kgdb over c 2560 kgdboc= [KGDB,HW] kgdb over consoles.
2561 Requires a tty driver 2561 Requires a tty driver that supports console polling,
2562 or a supported pollin 2562 or a supported polling keyboard driver (non-usb).
2563 Serial only format: 2563 Serial only format: <serial_device>[,baud]
2564 keyboard only format 2564 keyboard only format: kbd
2565 keyboard and serial 2565 keyboard and serial format: kbd,<serial_device>[,baud]
2566 Optional Kernel mode 2566 Optional Kernel mode setting:
2567 kms, kbd format: kms 2567 kms, kbd format: kms,kbd
2568 kms, kbd and serial 2568 kms, kbd and serial format: kms,kbd,<ser_dev>[,baud]
2569 2569
2570 kgdboc_earlycon= [KGDB,HW,EARL 2570 kgdboc_earlycon= [KGDB,HW,EARLY]
2571 If the boot console p 2571 If the boot console provides the ability to read
2572 characters and can wo 2572 characters and can work in polling mode, you can use
2573 this parameter to tel 2573 this parameter to tell kgdb to use it as a backend
2574 until the normal cons 2574 until the normal console is registered. Intended to
2575 be used together with 2575 be used together with the kgdboc parameter which
2576 specifies the normal 2576 specifies the normal console to transition to.
2577 2577
2578 The name of the early 2578 The name of the early console should be specified
2579 as the value of this 2579 as the value of this parameter. Note that the name of
2580 the early console mig 2580 the early console might be different than the tty
2581 name passed to kgdboc 2581 name passed to kgdboc. It's OK to leave the value
2582 blank and the first b 2582 blank and the first boot console that implements
2583 read() will be picked 2583 read() will be picked.
2584 2584
2585 kgdbwait [KGDB,EARLY] Stop ker 2585 kgdbwait [KGDB,EARLY] Stop kernel execution and enter the
2586 kernel debugger at th 2586 kernel debugger at the earliest opportunity.
2587 2587
2588 kmac= [MIPS] Korina etherne 2588 kmac= [MIPS] Korina ethernet MAC address.
2589 Configure the RouterB 2589 Configure the RouterBoard 532 series on-chip
2590 Ethernet adapter MAC 2590 Ethernet adapter MAC address.
2591 2591
2592 kmemleak= [KNL,EARLY] Boot-time 2592 kmemleak= [KNL,EARLY] Boot-time kmemleak enable/disable
2593 Valid arguments: on, 2593 Valid arguments: on, off
2594 Default: on 2594 Default: on
2595 Built with CONFIG_DEB 2595 Built with CONFIG_DEBUG_KMEMLEAK_DEFAULT_OFF=y,
2596 the default is off. 2596 the default is off.
2597 2597
2598 kprobe_event=[probe-list] 2598 kprobe_event=[probe-list]
2599 [FTRACE] Add kprobe e 2599 [FTRACE] Add kprobe events and enable at boot time.
2600 The probe-list is a s 2600 The probe-list is a semicolon delimited list of probe
2601 definitions. Each def 2601 definitions. Each definition is same as kprobe_events
2602 interface, but the pa 2602 interface, but the parameters are comma delimited.
2603 For example, to add a 2603 For example, to add a kprobe event on vfs_read with
2604 arg1 and arg2, add to 2604 arg1 and arg2, add to the command line;
2605 2605
2606 kprobe_event=p, 2606 kprobe_event=p,vfs_read,$arg1,$arg2
2607 2607
2608 See also Documentatio 2608 See also Documentation/trace/kprobetrace.rst "Kernel
2609 Boot Parameter" secti 2609 Boot Parameter" section.
2610 2610
2611 kpti= [ARM64,EARLY] Control 2611 kpti= [ARM64,EARLY] Control page table isolation of
2612 user and kernel addre 2612 user and kernel address spaces.
2613 Default: enabled on c 2613 Default: enabled on cores which need mitigation.
2614 0: force disabled 2614 0: force disabled
2615 1: force enabled 2615 1: force enabled
2616 2616
2617 kunit.enable= [KUNIT] Enable execut 2617 kunit.enable= [KUNIT] Enable executing KUnit tests. Requires
2618 CONFIG_KUNIT to be se 2618 CONFIG_KUNIT to be set to be fully enabled. The
2619 default value can be 2619 default value can be overridden via
2620 KUNIT_DEFAULT_ENABLED 2620 KUNIT_DEFAULT_ENABLED.
2621 Default is 1 (enabled 2621 Default is 1 (enabled)
2622 2622
2623 kvm.ignore_msrs=[KVM] Ignore guest ac 2623 kvm.ignore_msrs=[KVM] Ignore guest accesses to unhandled MSRs.
2624 Default is 0 (don't i 2624 Default is 0 (don't ignore, but inject #GP)
2625 2625
2626 kvm.eager_page_split= 2626 kvm.eager_page_split=
2627 [KVM,X86] Controls wh 2627 [KVM,X86] Controls whether or not KVM will try to
2628 proactively split all 2628 proactively split all huge pages during dirty logging.
2629 Eager page splitting 2629 Eager page splitting reduces interruptions to vCPU
2630 execution by eliminat 2630 execution by eliminating the write-protection faults
2631 and MMU lock contenti 2631 and MMU lock contention that would otherwise be
2632 required to split hug 2632 required to split huge pages lazily.
2633 2633
2634 VM workloads that rar 2634 VM workloads that rarely perform writes or that write
2635 only to a small regio 2635 only to a small region of VM memory may benefit from
2636 disabling eager page 2636 disabling eager page splitting to allow huge pages to
2637 still be used for rea 2637 still be used for reads.
2638 2638
2639 The behavior of eager 2639 The behavior of eager page splitting depends on whether
2640 KVM_DIRTY_LOG_INITIAL 2640 KVM_DIRTY_LOG_INITIALLY_SET is enabled or disabled. If
2641 disabled, all huge pa 2641 disabled, all huge pages in a memslot will be eagerly
2642 split when dirty logg 2642 split when dirty logging is enabled on that memslot. If
2643 enabled, eager page s 2643 enabled, eager page splitting will be performed during
2644 the KVM_CLEAR_DIRTY i 2644 the KVM_CLEAR_DIRTY ioctl, and only for the pages being
2645 cleared. 2645 cleared.
2646 2646
2647 Eager page splitting 2647 Eager page splitting is only supported when kvm.tdp_mmu=Y.
2648 2648
2649 Default is Y (on). 2649 Default is Y (on).
2650 2650
2651 kvm.enable_vmware_backdoor=[KVM] Supp 2651 kvm.enable_vmware_backdoor=[KVM] Support VMware backdoor PV interface.
2652 Default is 2652 Default is false (don't support).
2653 2653
2654 kvm.nx_huge_pages= 2654 kvm.nx_huge_pages=
2655 [KVM] Controls the so 2655 [KVM] Controls the software workaround for the
2656 X86_BUG_ITLB_MULTIHIT 2656 X86_BUG_ITLB_MULTIHIT bug.
2657 force : Always depl 2657 force : Always deploy workaround.
2658 off : Never deplo 2658 off : Never deploy workaround.
2659 auto : Deploy work 2659 auto : Deploy workaround based on the presence of
2660 X86_BUG_ITL 2660 X86_BUG_ITLB_MULTIHIT.
2661 2661
2662 Default is 'auto'. 2662 Default is 'auto'.
2663 2663
2664 If the software worka 2664 If the software workaround is enabled for the host,
2665 guests do need not to 2665 guests do need not to enable it for nested guests.
2666 2666
2667 kvm.nx_huge_pages_recovery_ratio= 2667 kvm.nx_huge_pages_recovery_ratio=
2668 [KVM] Controls how ma 2668 [KVM] Controls how many 4KiB pages are periodically zapped
2669 back to huge pages. 2669 back to huge pages. 0 disables the recovery, otherwise if
2670 the value is N KVM wi 2670 the value is N KVM will zap 1/Nth of the 4KiB pages every
2671 period (see below). 2671 period (see below). The default is 60.
2672 2672
2673 kvm.nx_huge_pages_recovery_period_ms= 2673 kvm.nx_huge_pages_recovery_period_ms=
2674 [KVM] Controls the ti 2674 [KVM] Controls the time period at which KVM zaps 4KiB pages
2675 back to huge pages. I 2675 back to huge pages. If the value is a non-zero N, KVM will
2676 zap a portion (see ra 2676 zap a portion (see ratio above) of the pages every N msecs.
2677 If the value is 0 (th 2677 If the value is 0 (the default), KVM will pick a period based
2678 on the ratio, such th 2678 on the ratio, such that a page is zapped after 1 hour on average.
2679 2679
2680 kvm-amd.nested= [KVM,AMD] Control nes 2680 kvm-amd.nested= [KVM,AMD] Control nested virtualization feature in
2681 KVM/SVM. Default is 1 2681 KVM/SVM. Default is 1 (enabled).
2682 2682
2683 kvm-amd.npt= [KVM,AMD] Control KVM 2683 kvm-amd.npt= [KVM,AMD] Control KVM's use of Nested Page Tables,
2684 a.k.a. Two-Dimensiona 2684 a.k.a. Two-Dimensional Page Tables. Default is 1
2685 (enabled). Disable by 2685 (enabled). Disable by KVM if hardware lacks support
2686 for NPT. 2686 for NPT.
2687 2687
2688 kvm-arm.mode= 2688 kvm-arm.mode=
2689 [KVM,ARM,EARLY] Selec 2689 [KVM,ARM,EARLY] Select one of KVM/arm64's modes of
2690 operation. 2690 operation.
2691 2691
2692 none: Forcefully disa 2692 none: Forcefully disable KVM.
2693 2693
2694 nvhe: Standard nVHE-b 2694 nvhe: Standard nVHE-based mode, without support for
2695 protected guest 2695 protected guests.
2696 2696
2697 protected: nVHE-based 2697 protected: nVHE-based mode with support for guests whose
2698 state is k 2698 state is kept private from the host.
2699 2699
2700 nested: VHE-based mod 2700 nested: VHE-based mode with support for nested
2701 virtualizatio 2701 virtualization. Requires at least ARMv8.3
2702 hardware. 2702 hardware.
2703 2703
2704 Defaults to VHE/nVHE 2704 Defaults to VHE/nVHE based on hardware support. Setting
2705 mode to "protected" w 2705 mode to "protected" will disable kexec and hibernation
2706 for the host. "nested 2706 for the host. "nested" is experimental and should be
2707 used with extreme cau 2707 used with extreme caution.
2708 2708
2709 kvm-arm.vgic_v3_group0_trap= 2709 kvm-arm.vgic_v3_group0_trap=
2710 [KVM,ARM,EARLY] Trap 2710 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-0
2711 system registers 2711 system registers
2712 2712
2713 kvm-arm.vgic_v3_group1_trap= 2713 kvm-arm.vgic_v3_group1_trap=
2714 [KVM,ARM,EARLY] Trap 2714 [KVM,ARM,EARLY] Trap guest accesses to GICv3 group-1
2715 system registers 2715 system registers
2716 2716
2717 kvm-arm.vgic_v3_common_trap= 2717 kvm-arm.vgic_v3_common_trap=
2718 [KVM,ARM,EARLY] Trap 2718 [KVM,ARM,EARLY] Trap guest accesses to GICv3 common
2719 system registers 2719 system registers
2720 2720
2721 kvm-arm.vgic_v4_enable= 2721 kvm-arm.vgic_v4_enable=
2722 [KVM,ARM,EARLY] Allow 2722 [KVM,ARM,EARLY] Allow use of GICv4 for direct
2723 injection of LPIs. 2723 injection of LPIs.
2724 2724
2725 kvm-arm.wfe_trap_policy= 2725 kvm-arm.wfe_trap_policy=
2726 [KVM,ARM] Control whe 2726 [KVM,ARM] Control when to set WFE instruction trap for
2727 KVM VMs. Traps are al 2727 KVM VMs. Traps are allowed but not guaranteed by the
2728 CPU architecture. 2728 CPU architecture.
2729 2729
2730 trap: set WFE instruc 2730 trap: set WFE instruction trap
2731 2731
2732 notrap: clear WFE ins 2732 notrap: clear WFE instruction trap
2733 2733
2734 kvm-arm.wfi_trap_policy= 2734 kvm-arm.wfi_trap_policy=
2735 [KVM,ARM] Control whe 2735 [KVM,ARM] Control when to set WFI instruction trap for
2736 KVM VMs. Traps are al 2736 KVM VMs. Traps are allowed but not guaranteed by the
2737 CPU architecture. 2737 CPU architecture.
2738 2738
2739 trap: set WFI instruc 2739 trap: set WFI instruction trap
2740 2740
2741 notrap: clear WFI ins 2741 notrap: clear WFI instruction trap
2742 2742
2743 kvm_cma_resv_ratio=n [PPC,EARLY] 2743 kvm_cma_resv_ratio=n [PPC,EARLY]
2744 Reserves given percen 2744 Reserves given percentage from system memory area for
2745 contiguous memory all 2745 contiguous memory allocation for KVM hash pagetable
2746 allocation. 2746 allocation.
2747 By default it reserve 2747 By default it reserves 5% of total system memory.
2748 Format: <integer> 2748 Format: <integer>
2749 Default: 5 2749 Default: 5
2750 2750
2751 kvm-intel.ept= [KVM,Intel] Control K 2751 kvm-intel.ept= [KVM,Intel] Control KVM's use of Extended Page Tables,
2752 a.k.a. Two-Dimensiona 2752 a.k.a. Two-Dimensional Page Tables. Default is 1
2753 (enabled). Disable by 2753 (enabled). Disable by KVM if hardware lacks support
2754 for EPT. 2754 for EPT.
2755 2755
2756 kvm-intel.emulate_invalid_guest_state 2756 kvm-intel.emulate_invalid_guest_state=
2757 [KVM,Intel] Control w 2757 [KVM,Intel] Control whether to emulate invalid guest
2758 state. Ignored if kvm 2758 state. Ignored if kvm-intel.enable_unrestricted_guest=1,
2759 as guest state is nev 2759 as guest state is never invalid for unrestricted
2760 guests. This param do 2760 guests. This param doesn't apply to nested guests (L2),
2761 as KVM never emulates 2761 as KVM never emulates invalid L2 guest state.
2762 Default is 1 (enabled 2762 Default is 1 (enabled).
2763 2763
2764 kvm-intel.flexpriority= 2764 kvm-intel.flexpriority=
2765 [KVM,Intel] Control K 2765 [KVM,Intel] Control KVM's use of FlexPriority feature
2766 (TPR shadow). Default 2766 (TPR shadow). Default is 1 (enabled). Disable by KVM if
2767 hardware lacks suppor 2767 hardware lacks support for it.
2768 2768
2769 kvm-intel.nested= 2769 kvm-intel.nested=
2770 [KVM,Intel] Control n 2770 [KVM,Intel] Control nested virtualization feature in
2771 KVM/VMX. Default is 1 2771 KVM/VMX. Default is 1 (enabled).
2772 2772
2773 kvm-intel.unrestricted_guest= 2773 kvm-intel.unrestricted_guest=
2774 [KVM,Intel] Control K 2774 [KVM,Intel] Control KVM's use of unrestricted guest
2775 feature (virtualized 2775 feature (virtualized real and unpaged mode). Default
2776 is 1 (enabled). Disab 2776 is 1 (enabled). Disable by KVM if EPT is disabled or
2777 hardware lacks suppor 2777 hardware lacks support for it.
2778 2778
2779 kvm-intel.vmentry_l1d_flush=[KVM,Inte 2779 kvm-intel.vmentry_l1d_flush=[KVM,Intel] Mitigation for L1 Terminal Fault
2780 CVE-2018-3620. 2780 CVE-2018-3620.
2781 2781
2782 Valid arguments: neve 2782 Valid arguments: never, cond, always
2783 2783
2784 always: L1D cache flu 2784 always: L1D cache flush on every VMENTER.
2785 cond: Flush L1D on 2785 cond: Flush L1D on VMENTER only when the code between
2786 VMEXIT and VM 2786 VMEXIT and VMENTER can leak host memory.
2787 never: Disables the 2787 never: Disables the mitigation
2788 2788
2789 Default is cond (do L 2789 Default is cond (do L1 cache flush in specific instances)
2790 2790
2791 kvm-intel.vpid= [KVM,Intel] Control K 2791 kvm-intel.vpid= [KVM,Intel] Control KVM's use of Virtual Processor
2792 Identification featur 2792 Identification feature (tagged TLBs). Default is 1
2793 (enabled). Disable by 2793 (enabled). Disable by KVM if hardware lacks support
2794 for it. 2794 for it.
2795 2795
2796 l1d_flush= [X86,INTEL,EARLY] 2796 l1d_flush= [X86,INTEL,EARLY]
2797 Control mitigation fo 2797 Control mitigation for L1D based snooping vulnerability.
2798 2798
2799 Certain CPUs are vuln 2799 Certain CPUs are vulnerable to an exploit against CPU
2800 internal buffers whic 2800 internal buffers which can forward information to a
2801 disclosure gadget und 2801 disclosure gadget under certain conditions.
2802 2802
2803 In vulnerable process 2803 In vulnerable processors, the speculatively
2804 forwarded data can be 2804 forwarded data can be used in a cache side channel
2805 attack, to access dat 2805 attack, to access data to which the attacker does
2806 not have direct acces 2806 not have direct access.
2807 2807
2808 This parameter contro 2808 This parameter controls the mitigation. The
2809 options are: 2809 options are:
2810 2810
2811 on - enable t 2811 on - enable the interface for the mitigation
2812 2812
2813 l1tf= [X86,EARLY] Control m 2813 l1tf= [X86,EARLY] Control mitigation of the L1TF vulnerability on
2814 affected CPUs 2814 affected CPUs
2815 2815
2816 The kernel PTE invers 2816 The kernel PTE inversion protection is unconditionally
2817 enabled and cannot be 2817 enabled and cannot be disabled.
2818 2818
2819 full 2819 full
2820 Provides all 2820 Provides all available mitigations for the
2821 L1TF vulnerab 2821 L1TF vulnerability. Disables SMT and
2822 enables all m 2822 enables all mitigations in the
2823 hypervisors, 2823 hypervisors, i.e. unconditional L1D flush.
2824 2824
2825 SMT control a 2825 SMT control and L1D flush control via the
2826 sysfs interfa 2826 sysfs interface is still possible after
2827 boot. Hyperv 2827 boot. Hypervisors will issue a warning
2828 when the firs 2828 when the first VM is started in a
2829 potentially i 2829 potentially insecure configuration,
2830 i.e. SMT enab 2830 i.e. SMT enabled or L1D flush disabled.
2831 2831
2832 full,force 2832 full,force
2833 Same as 'full 2833 Same as 'full', but disables SMT and L1D
2834 flush runtime 2834 flush runtime control. Implies the
2835 'nosmt=force' 2835 'nosmt=force' command line option.
2836 (i.e. sysfs c 2836 (i.e. sysfs control of SMT is disabled.)
2837 2837
2838 flush 2838 flush
2839 Leaves SMT en 2839 Leaves SMT enabled and enables the default
2840 hypervisor mi 2840 hypervisor mitigation, i.e. conditional
2841 L1D flush. 2841 L1D flush.
2842 2842
2843 SMT control a 2843 SMT control and L1D flush control via the
2844 sysfs interfa 2844 sysfs interface is still possible after
2845 boot. Hyperv 2845 boot. Hypervisors will issue a warning
2846 when the firs 2846 when the first VM is started in a
2847 potentially i 2847 potentially insecure configuration,
2848 i.e. SMT enab 2848 i.e. SMT enabled or L1D flush disabled.
2849 2849
2850 flush,nosmt 2850 flush,nosmt
2851 2851
2852 Disables SMT 2852 Disables SMT and enables the default
2853 hypervisor mi 2853 hypervisor mitigation.
2854 2854
2855 SMT control a 2855 SMT control and L1D flush control via the
2856 sysfs interfa 2856 sysfs interface is still possible after
2857 boot. Hyperv 2857 boot. Hypervisors will issue a warning
2858 when the firs 2858 when the first VM is started in a
2859 potentially i 2859 potentially insecure configuration,
2860 i.e. SMT enab 2860 i.e. SMT enabled or L1D flush disabled.
2861 2861
2862 flush,nowarn 2862 flush,nowarn
2863 Same as 'flus 2863 Same as 'flush', but hypervisors will not
2864 warn when a V 2864 warn when a VM is started in a potentially
2865 insecure conf 2865 insecure configuration.
2866 2866
2867 off 2867 off
2868 Disables hype 2868 Disables hypervisor mitigations and doesn't
2869 emit any warn 2869 emit any warnings.
2870 It also drops 2870 It also drops the swap size and available
2871 RAM limit res 2871 RAM limit restriction on both hypervisor and
2872 bare metal. 2872 bare metal.
2873 2873
2874 Default is 'flush'. 2874 Default is 'flush'.
2875 2875
2876 For details see: Docu 2876 For details see: Documentation/admin-guide/hw-vuln/l1tf.rst
2877 2877
2878 l2cr= [PPC] 2878 l2cr= [PPC]
2879 2879
2880 l3cr= [PPC] 2880 l3cr= [PPC]
2881 2881
2882 lapic [X86-32,APIC,EARLY] E 2882 lapic [X86-32,APIC,EARLY] Enable the local APIC even if BIOS
2883 disabled it. 2883 disabled it.
2884 2884
2885 lapic= [X86,APIC] Do not use 2885 lapic= [X86,APIC] Do not use TSC deadline
2886 value for LAPIC timer 2886 value for LAPIC timer one-shot implementation. Default
2887 back to the programma 2887 back to the programmable timer unit in the LAPIC.
2888 Format: notscdeadline 2888 Format: notscdeadline
2889 2889
2890 lapic_timer_c2_ok [X86,APIC,EAR 2890 lapic_timer_c2_ok [X86,APIC,EARLY] trust the local apic timer
2891 in C2 power state. 2891 in C2 power state.
2892 2892
2893 libata.dma= [LIBATA] DMA control 2893 libata.dma= [LIBATA] DMA control
2894 libata.dma=0 Dis 2894 libata.dma=0 Disable all PATA and SATA DMA
2895 libata.dma=1 PAT 2895 libata.dma=1 PATA and SATA Disk DMA only
2896 libata.dma=2 ATA 2896 libata.dma=2 ATAPI (CDROM) DMA only
2897 libata.dma=4 Com 2897 libata.dma=4 Compact Flash DMA only
2898 Combinations also wor 2898 Combinations also work, so libata.dma=3 enables DMA
2899 for disks and CDROMs, 2899 for disks and CDROMs, but not CFs.
2900 2900
2901 libata.ignore_hpa= [LIBATA] Igno 2901 libata.ignore_hpa= [LIBATA] Ignore HPA limit
2902 libata.ignore_hpa=0 2902 libata.ignore_hpa=0 keep BIOS limits (default)
2903 libata.ignore_hpa=1 2903 libata.ignore_hpa=1 ignore limits, using full disk
2904 2904
2905 libata.noacpi [LIBATA] Disables use 2905 libata.noacpi [LIBATA] Disables use of ACPI in libata suspend/resume
2906 when set. 2906 when set.
2907 Format: <int> 2907 Format: <int>
2908 2908
2909 libata.force= [LIBATA] Force config 2909 libata.force= [LIBATA] Force configurations. The format is a comma-
2910 separated list of "[I 2910 separated list of "[ID:]VAL" where ID is PORT[.DEVICE].
2911 PORT and DEVICE are d 2911 PORT and DEVICE are decimal numbers matching port, link
2912 or device. Basically 2912 or device. Basically, it matches the ATA ID string
2913 printed on console by 2913 printed on console by libata. If the whole ID part is
2914 omitted, the last POR 2914 omitted, the last PORT and DEVICE values are used. If
2915 ID hasn't been specif 2915 ID hasn't been specified yet, the configuration applies
2916 to all ports, links a 2916 to all ports, links and devices.
2917 2917
2918 If only DEVICE is omi 2918 If only DEVICE is omitted, the parameter applies to
2919 the port and all link 2919 the port and all links and devices behind it. DEVICE
2920 number of 0 either se 2920 number of 0 either selects the first device or the
2921 first fan-out link be 2921 first fan-out link behind PMP device. It does not
2922 select the host link. 2922 select the host link. DEVICE number of 15 selects the
2923 host link and device 2923 host link and device attached to it.
2924 2924
2925 The VAL specifies the 2925 The VAL specifies the configuration to force. As long
2926 as there is no ambigu 2926 as there is no ambiguity, shortcut notation is allowed.
2927 For example, both 1.5 2927 For example, both 1.5 and 1.5G would work for 1.5Gbps.
2928 The following configu 2928 The following configurations can be forced.
2929 2929
2930 * Cable type: 40c, 80 2930 * Cable type: 40c, 80c, short40c, unk, ign or sata.
2931 Any ID with matchin 2931 Any ID with matching PORT is used.
2932 2932
2933 * SATA link speed lim 2933 * SATA link speed limit: 1.5Gbps or 3.0Gbps.
2934 2934
2935 * Transfer mode: pio[ 2935 * Transfer mode: pio[0-7], mwdma[0-4] and udma[0-7].
2936 udma[/][16,25,33,44 2936 udma[/][16,25,33,44,66,100,133] notation is also
2937 allowed. 2937 allowed.
2938 2938
2939 * nohrst, nosrst, nor 2939 * nohrst, nosrst, norst: suppress hard, soft and both
2940 resets. 2940 resets.
2941 2941
2942 * rstonce: only attem 2942 * rstonce: only attempt one reset during hot-unplug
2943 link recovery. 2943 link recovery.
2944 2944
2945 * [no]dbdelay: Enable 2945 * [no]dbdelay: Enable or disable the extra 200ms delay
2946 before debouncing a 2946 before debouncing a link PHY and device presence
2947 detection. 2947 detection.
2948 2948
2949 * [no]ncq: Turn on or 2949 * [no]ncq: Turn on or off NCQ.
2950 2950
2951 * [no]ncqtrim: Enable 2951 * [no]ncqtrim: Enable or disable queued DSM TRIM.
2952 2952
2953 * [no]ncqati: Enable 2953 * [no]ncqati: Enable or disable NCQ trim on ATI chipset.
2954 2954
2955 * [no]trim: Enable or 2955 * [no]trim: Enable or disable (unqueued) TRIM.
2956 2956
2957 * trim_zero: Indicate 2957 * trim_zero: Indicate that TRIM command zeroes data.
2958 2958
2959 * max_trim_128m: Set 2959 * max_trim_128m: Set 128M maximum trim size limit.
2960 2960
2961 * [no]dma: Turn on or 2961 * [no]dma: Turn on or off DMA transfers.
2962 2962
2963 * atapi_dmadir: Enabl 2963 * atapi_dmadir: Enable ATAPI DMADIR bridge support.
2964 2964
2965 * atapi_mod16_dma: En 2965 * atapi_mod16_dma: Enable the use of ATAPI DMA for
2966 commands that are n 2966 commands that are not a multiple of 16 bytes.
2967 2967
2968 * [no]dmalog: Enable 2968 * [no]dmalog: Enable or disable the use of the
2969 READ LOG DMA EXT co 2969 READ LOG DMA EXT command to access logs.
2970 2970
2971 * [no]iddevlog: Enabl 2971 * [no]iddevlog: Enable or disable access to the
2972 identify device dat 2972 identify device data log.
2973 2973
2974 * [no]logdir: Enable 2974 * [no]logdir: Enable or disable access to the general
2975 purpose log directo 2975 purpose log directory.
2976 2976
2977 * max_sec_128: Set tr 2977 * max_sec_128: Set transfer size limit to 128 sectors.
2978 2978
2979 * max_sec_1024: Set o 2979 * max_sec_1024: Set or clear transfer size limit to
2980 1024 sectors. 2980 1024 sectors.
2981 2981
2982 * max_sec_lba48: Set 2982 * max_sec_lba48: Set or clear transfer size limit to
2983 65535 sectors. 2983 65535 sectors.
2984 2984
2985 * [no]lpm: Enable or 2985 * [no]lpm: Enable or disable link power management.
2986 2986
2987 * [no]setxfer: Indica 2987 * [no]setxfer: Indicate if transfer speed mode setting
2988 should be skipped. 2988 should be skipped.
2989 2989
2990 * [no]fua: Disable or 2990 * [no]fua: Disable or enable FUA (Force Unit Access)
2991 support for devices 2991 support for devices supporting this feature.
2992 2992
2993 * dump_id: Dump IDENT 2993 * dump_id: Dump IDENTIFY data.
2994 2994
2995 * disable: Disable th 2995 * disable: Disable this device.
2996 2996
2997 If there are multiple 2997 If there are multiple matching configurations changing
2998 the same attribute, t 2998 the same attribute, the last one is used.
2999 2999
3000 load_ramdisk= [RAM] [Deprecated] 3000 load_ramdisk= [RAM] [Deprecated]
3001 3001
3002 lockd.nlm_grace_period=P [NFS] Assig 3002 lockd.nlm_grace_period=P [NFS] Assign grace period.
3003 Format: <integer> 3003 Format: <integer>
3004 3004
3005 lockd.nlm_tcpport=N [NFS] Assign 3005 lockd.nlm_tcpport=N [NFS] Assign TCP port.
3006 Format: <integer> 3006 Format: <integer>
3007 3007
3008 lockd.nlm_timeout=T [NFS] Assign 3008 lockd.nlm_timeout=T [NFS] Assign timeout value.
3009 Format: <integer> 3009 Format: <integer>
3010 3010
3011 lockd.nlm_udpport=M [NFS] Assign 3011 lockd.nlm_udpport=M [NFS] Assign UDP port.
3012 Format: <integer> 3012 Format: <integer>
3013 3013
3014 lockdown= [SECURITY,EARLY] 3014 lockdown= [SECURITY,EARLY]
3015 { integrity | confide 3015 { integrity | confidentiality }
3016 Enable the kernel loc 3016 Enable the kernel lockdown feature. If set to
3017 integrity, kernel fea 3017 integrity, kernel features that allow userland to
3018 modify the running ke 3018 modify the running kernel are disabled. If set to
3019 confidentiality, kern 3019 confidentiality, kernel features that allow userland
3020 to extract confidenti 3020 to extract confidential information from the kernel
3021 are also disabled. 3021 are also disabled.
3022 3022
3023 locktorture.acq_writer_lim= [KNL] 3023 locktorture.acq_writer_lim= [KNL]
3024 Set the time limit in 3024 Set the time limit in jiffies for a lock
3025 acquisition. Acquisi 3025 acquisition. Acquisitions exceeding this limit
3026 will result in a spla 3026 will result in a splat once they do complete.
3027 3027
3028 locktorture.bind_readers= [KNL] 3028 locktorture.bind_readers= [KNL]
3029 Specify the list of C 3029 Specify the list of CPUs to which the readers are
3030 to be bound. 3030 to be bound.
3031 3031
3032 locktorture.bind_writers= [KNL] 3032 locktorture.bind_writers= [KNL]
3033 Specify the list of C 3033 Specify the list of CPUs to which the writers are
3034 to be bound. 3034 to be bound.
3035 3035
3036 locktorture.call_rcu_chains= [KNL] 3036 locktorture.call_rcu_chains= [KNL]
3037 Specify the number of 3037 Specify the number of self-propagating call_rcu()
3038 chains to set up. Th 3038 chains to set up. These are used to ensure that
3039 there is a high proba 3039 there is a high probability of an RCU grace period
3040 in progress at any gi 3040 in progress at any given time. Defaults to 0,
3041 which disables these 3041 which disables these call_rcu() chains.
3042 3042
3043 locktorture.long_hold= [KNL] 3043 locktorture.long_hold= [KNL]
3044 Specify the duration 3044 Specify the duration in milliseconds for the
3045 occasional long-durat 3045 occasional long-duration lock hold time. Defaults
3046 to 100 milliseconds. 3046 to 100 milliseconds. Select 0 to disable.
3047 3047
3048 locktorture.nested_locks= [KNL] 3048 locktorture.nested_locks= [KNL]
3049 Specify the maximum l 3049 Specify the maximum lock nesting depth that
3050 locktorture is to exe 3050 locktorture is to exercise, up to a limit of 8
3051 (MAX_NESTED_LOCKS). 3051 (MAX_NESTED_LOCKS). Specify zero to disable.
3052 Note that this parame 3052 Note that this parameter is ineffective on types
3053 of locks that do not 3053 of locks that do not support nested acquisition.
3054 3054
3055 locktorture.nreaders_stress= [KNL] 3055 locktorture.nreaders_stress= [KNL]
3056 Set the number of loc 3056 Set the number of locking read-acquisition kthreads.
3057 Defaults to being aut 3057 Defaults to being automatically set based on the
3058 number of online CPUs 3058 number of online CPUs.
3059 3059
3060 locktorture.nwriters_stress= [KNL] 3060 locktorture.nwriters_stress= [KNL]
3061 Set the number of loc 3061 Set the number of locking write-acquisition kthreads.
3062 3062
3063 locktorture.onoff_holdoff= [KNL] 3063 locktorture.onoff_holdoff= [KNL]
3064 Set time (s) after bo 3064 Set time (s) after boot for CPU-hotplug testing.
3065 3065
3066 locktorture.onoff_interval= [KNL] 3066 locktorture.onoff_interval= [KNL]
3067 Set time (s) between 3067 Set time (s) between CPU-hotplug operations, or
3068 zero to disable CPU-h 3068 zero to disable CPU-hotplug testing.
3069 3069
3070 locktorture.rt_boost= [KNL] 3070 locktorture.rt_boost= [KNL]
3071 Do periodic testing o 3071 Do periodic testing of real-time lock priority
3072 boosting. Select 0 t 3072 boosting. Select 0 to disable, 1 to boost
3073 only rt_mutex, and 2 3073 only rt_mutex, and 2 to boost unconditionally.
3074 Defaults to 2, which 3074 Defaults to 2, which might seem to be an
3075 odd choice, but which 3075 odd choice, but which should be harmless for
3076 non-real-time spinloc 3076 non-real-time spinlocks, due to their disabling
3077 of preemption. Note 3077 of preemption. Note that non-realtime mutexes
3078 disable boosting. 3078 disable boosting.
3079 3079
3080 locktorture.rt_boost_factor= [KNL] 3080 locktorture.rt_boost_factor= [KNL]
3081 Number that determine 3081 Number that determines how often and for how
3082 long priority boostin 3082 long priority boosting is exercised. This is
3083 scaled down by the nu 3083 scaled down by the number of writers, so that the
3084 number of boosts per 3084 number of boosts per unit time remains roughly
3085 constant as the numbe 3085 constant as the number of writers increases.
3086 On the other hand, th 3086 On the other hand, the duration of each boost
3087 increases with the nu 3087 increases with the number of writers.
3088 3088
3089 locktorture.shuffle_interval= [KNL] 3089 locktorture.shuffle_interval= [KNL]
3090 Set task-shuffle inte 3090 Set task-shuffle interval (jiffies). Shuffling
3091 tasks allows some CPU 3091 tasks allows some CPUs to go into dyntick-idle
3092 mode during the lockt 3092 mode during the locktorture test.
3093 3093
3094 locktorture.shutdown_secs= [KNL] 3094 locktorture.shutdown_secs= [KNL]
3095 Set time (s) after bo 3095 Set time (s) after boot system shutdown. This
3096 is useful for hands-o 3096 is useful for hands-off automated testing.
3097 3097
3098 locktorture.stat_interval= [KNL] 3098 locktorture.stat_interval= [KNL]
3099 Time (s) between stat 3099 Time (s) between statistics printk()s.
3100 3100
3101 locktorture.stutter= [KNL] 3101 locktorture.stutter= [KNL]
3102 Time (s) to stutter t 3102 Time (s) to stutter testing, for example,
3103 specifying five secon 3103 specifying five seconds causes the test to run for
3104 five seconds, wait fo 3104 five seconds, wait for five seconds, and so on.
3105 This tests the lockin 3105 This tests the locking primitive's ability to
3106 transition abruptly t 3106 transition abruptly to and from idle.
3107 3107
3108 locktorture.torture_type= [KNL] 3108 locktorture.torture_type= [KNL]
3109 Specify the locking i 3109 Specify the locking implementation to test.
3110 3110
3111 locktorture.verbose= [KNL] 3111 locktorture.verbose= [KNL]
3112 Enable additional pri 3112 Enable additional printk() statements.
3113 3113
3114 locktorture.writer_fifo= [KNL] 3114 locktorture.writer_fifo= [KNL]
3115 Run the write-side lo 3115 Run the write-side locktorture kthreads at
3116 sched_set_fifo() real 3116 sched_set_fifo() real-time priority.
3117 3117
3118 logibm.irq= [HW,MOUSE] Logitech B 3118 logibm.irq= [HW,MOUSE] Logitech Bus Mouse Driver
3119 Format: <irq> 3119 Format: <irq>
3120 3120
3121 loglevel= [KNL,EARLY] 3121 loglevel= [KNL,EARLY]
3122 All Kernel Messages w 3122 All Kernel Messages with a loglevel smaller than the
3123 console loglevel will 3123 console loglevel will be printed to the console. It can
3124 also be changed with 3124 also be changed with klogd or other programs. The
3125 loglevels are defined 3125 loglevels are defined as follows:
3126 3126
3127 0 (KERN_EMERG) 3127 0 (KERN_EMERG) system is unusable
3128 1 (KERN_ALERT) 3128 1 (KERN_ALERT) action must be taken immediately
3129 2 (KERN_CRIT) 3129 2 (KERN_CRIT) critical conditions
3130 3 (KERN_ERR) 3130 3 (KERN_ERR) error conditions
3131 4 (KERN_WARNING) 3131 4 (KERN_WARNING) warning conditions
3132 5 (KERN_NOTICE) 3132 5 (KERN_NOTICE) normal but significant condition
3133 6 (KERN_INFO) 3133 6 (KERN_INFO) informational
3134 7 (KERN_DEBUG) 3134 7 (KERN_DEBUG) debug-level messages
3135 3135
3136 log_buf_len=n[KMG] [KNL,EARLY] 3136 log_buf_len=n[KMG] [KNL,EARLY]
3137 Sets the size of the 3137 Sets the size of the printk ring buffer, in bytes.
3138 n must be a power of 3138 n must be a power of two and greater than the
3139 minimal size. The min 3139 minimal size. The minimal size is defined by
3140 LOG_BUF_SHIFT kernel 3140 LOG_BUF_SHIFT kernel config parameter. There
3141 is also CONFIG_LOG_CP 3141 is also CONFIG_LOG_CPU_MAX_BUF_SHIFT config
3142 parameter that allows 3142 parameter that allows to increase the default size
3143 depending on the numb 3143 depending on the number of CPUs. See init/Kconfig
3144 for more details. 3144 for more details.
3145 3145
3146 logo.nologo [FB] Disables display 3146 logo.nologo [FB] Disables display of the built-in Linux logo.
3147 This may be used to p 3147 This may be used to provide more screen space for
3148 kernel log messages a 3148 kernel log messages and is useful when debugging
3149 kernel boot problems. 3149 kernel boot problems.
3150 3150
3151 lp=0 [LP] Specify paral 3151 lp=0 [LP] Specify parallel ports to use, e.g,
3152 lp=port[,port...] lp=none,parpo 3152 lp=port[,port...] lp=none,parport0 (lp0 not configured, lp1 uses
3153 lp=reset first paralle 3153 lp=reset first parallel port). 'lp=0' disables the
3154 lp=auto printer drive 3154 lp=auto printer driver. 'lp=reset' (which can be
3155 specified in 3155 specified in addition to the ports) causes
3156 attached prin 3156 attached printers to be reset. Using
3157 lp=port1,port 3157 lp=port1,port2,... specifies the parallel ports
3158 to associate 3158 to associate lp devices with, starting with
3159 lp0. A port s 3159 lp0. A port specification may be 'none' to skip
3160 that lp devic 3160 that lp device, or a parport name such as
3161 'parport0'. S 3161 'parport0'. Specifying 'lp=auto' instead of a
3162 port specific 3162 port specification list means that device IDs
3163 from each por 3163 from each port should be examined, to see if
3164 an IEEE 1284- 3164 an IEEE 1284-compliant printer is attached; if
3165 so, the drive 3165 so, the driver will manage that printer.
3166 See also head 3166 See also header of drivers/char/lp.c.
3167 3167
3168 lpj=n [KNL] 3168 lpj=n [KNL]
3169 Sets loops_per_jiffy 3169 Sets loops_per_jiffy to given constant, thus avoiding
3170 time-consuming boot-t 3170 time-consuming boot-time autodetection (up to 250 ms per
3171 CPU). 0 enables autod 3171 CPU). 0 enables autodetection (default). To determine
3172 the correct value for 3172 the correct value for your kernel, boot with normal
3173 autodetection and see 3173 autodetection and see what value is printed. Note that
3174 on SMP systems the pr 3174 on SMP systems the preset will be applied to all CPUs,
3175 which is likely to ca 3175 which is likely to cause problems if your CPUs need
3176 significantly diverge 3176 significantly divergent settings. An incorrect value
3177 will cause delays in 3177 will cause delays in the kernel to be wrong, leading to
3178 unpredictable I/O err 3178 unpredictable I/O errors and other breakage. Although
3179 unlikely, in the extr 3179 unlikely, in the extreme case this might damage your
3180 hardware. 3180 hardware.
3181 3181
3182 lsm.debug [SECURITY] Enable LSM 3182 lsm.debug [SECURITY] Enable LSM initialization debugging output.
3183 3183
3184 lsm=lsm1,...,lsmN 3184 lsm=lsm1,...,lsmN
3185 [SECURITY] Choose ord 3185 [SECURITY] Choose order of LSM initialization. This
3186 overrides CONFIG_LSM, 3186 overrides CONFIG_LSM, and the "security=" parameter.
3187 3187
3188 machtype= [Loongson] Share the 3188 machtype= [Loongson] Share the same kernel image file between
3189 different yeeloong la 3189 different yeeloong laptops.
3190 Example: machtype=lem 3190 Example: machtype=lemote-yeeloong-2f-7inch
3191 3191
3192 maxcpus= [SMP,EARLY] Maximum n 3192 maxcpus= [SMP,EARLY] Maximum number of processors that an SMP kernel
3193 will bring up during 3193 will bring up during bootup. maxcpus=n : n >= 0 limits
3194 the kernel to bring u 3194 the kernel to bring up 'n' processors. Surely after
3195 bootup you can bring 3195 bootup you can bring up the other plugged cpu by executing
3196 "echo 1 > /sys/device 3196 "echo 1 > /sys/devices/system/cpu/cpuX/online". So maxcpus
3197 only takes effect dur 3197 only takes effect during system bootup.
3198 While n=0 is a specia 3198 While n=0 is a special case, it is equivalent to "nosmp",
3199 which also disables t 3199 which also disables the IO APIC.
3200 3200
3201 max_loop= [LOOP] The number of 3201 max_loop= [LOOP] The number of loop block devices that get
3202 (loop.max_loop) unconditionally pre-c 3202 (loop.max_loop) unconditionally pre-created at init time. The default
3203 number is configured 3203 number is configured by BLK_DEV_LOOP_MIN_COUNT. Instead
3204 of statically allocat 3204 of statically allocating a predefined number, loop
3205 devices can be reques 3205 devices can be requested on-demand with the
3206 /dev/loop-control int 3206 /dev/loop-control interface.
3207 3207
3208 mce [X86-32] Machine Chec 3208 mce [X86-32] Machine Check Exception
3209 3209
3210 mce=option [X86-64] See Document 3210 mce=option [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
3211 3211
3212 md= [HW] RAID subsystems 3212 md= [HW] RAID subsystems devices and level
3213 See Documentation/adm 3213 See Documentation/admin-guide/md.rst.
3214 3214
3215 mdacon= [MDA] 3215 mdacon= [MDA]
3216 Format: <first>,<last 3216 Format: <first>,<last>
3217 Specifies range of co 3217 Specifies range of consoles to be captured by the MDA.
3218 3218
3219 mds= [X86,INTEL,EARLY] 3219 mds= [X86,INTEL,EARLY]
3220 Control mitigation fo 3220 Control mitigation for the Micro-architectural Data
3221 Sampling (MDS) vulner 3221 Sampling (MDS) vulnerability.
3222 3222
3223 Certain CPUs are vuln 3223 Certain CPUs are vulnerable to an exploit against CPU
3224 internal buffers whic 3224 internal buffers which can forward information to a
3225 disclosure gadget und 3225 disclosure gadget under certain conditions.
3226 3226
3227 In vulnerable process 3227 In vulnerable processors, the speculatively
3228 forwarded data can be 3228 forwarded data can be used in a cache side channel
3229 attack, to access dat 3229 attack, to access data to which the attacker does
3230 not have direct acces 3230 not have direct access.
3231 3231
3232 This parameter contro 3232 This parameter controls the MDS mitigation. The
3233 options are: 3233 options are:
3234 3234
3235 full - Enable M 3235 full - Enable MDS mitigation on vulnerable CPUs
3236 full,nosmt - Enable M 3236 full,nosmt - Enable MDS mitigation and disable
3237 SMT on v 3237 SMT on vulnerable CPUs
3238 off - Uncondit 3238 off - Unconditionally disable MDS mitigation
3239 3239
3240 On TAA-affected machi 3240 On TAA-affected machines, mds=off can be prevented by
3241 an active TAA mitigat 3241 an active TAA mitigation as both vulnerabilities are
3242 mitigated with the sa 3242 mitigated with the same mechanism so in order to disable
3243 this mitigation, you 3243 this mitigation, you need to specify tsx_async_abort=off
3244 too. 3244 too.
3245 3245
3246 Not specifying this o 3246 Not specifying this option is equivalent to
3247 mds=full. 3247 mds=full.
3248 3248
3249 For details see: Docu 3249 For details see: Documentation/admin-guide/hw-vuln/mds.rst
3250 3250
3251 mem=nn[KMG] [HEXAGON,EARLY] Set t 3251 mem=nn[KMG] [HEXAGON,EARLY] Set the memory size.
3252 Must be specified, ot 3252 Must be specified, otherwise memory size will be 0.
3253 3253
3254 mem=nn[KMG] [KNL,BOOT,EARLY] Forc 3254 mem=nn[KMG] [KNL,BOOT,EARLY] Force usage of a specific amount
3255 of memory Amount of m 3255 of memory Amount of memory to be used in cases
3256 as follows: 3256 as follows:
3257 3257
3258 1 for test; 3258 1 for test;
3259 2 when the kernel is 3259 2 when the kernel is not able to see the whole system memory;
3260 3 memory that lies af 3260 3 memory that lies after 'mem=' boundary is excluded from
3261 the hypervisor, then 3261 the hypervisor, then assigned to KVM guests.
3262 4 to limit the memory 3262 4 to limit the memory available for kdump kernel.
3263 3263
3264 [ARC,MICROBLAZE] - th 3264 [ARC,MICROBLAZE] - the limit applies only to low memory,
3265 high memory is not af 3265 high memory is not affected.
3266 3266
3267 [ARM64] - only limits 3267 [ARM64] - only limits memory covered by the linear
3268 mapping. The NOMAP re 3268 mapping. The NOMAP regions are not affected.
3269 3269
3270 [X86] Work as limitin 3270 [X86] Work as limiting max address. Use together
3271 with memmap= to avoid 3271 with memmap= to avoid physical address space collisions.
3272 Without memmap= PCI d 3272 Without memmap= PCI devices could be placed at addresses
3273 belonging to unused R 3273 belonging to unused RAM.
3274 3274
3275 Note that this only t 3275 Note that this only takes effects during boot time since
3276 in above case 3, memo 3276 in above case 3, memory may need be hot added after boot
3277 if system memory of h 3277 if system memory of hypervisor is not sufficient.
3278 3278
3279 mem=nn[KMG]@ss[KMG] 3279 mem=nn[KMG]@ss[KMG]
3280 [ARM,MIPS,EARLY] - ov 3280 [ARM,MIPS,EARLY] - override the memory layout
3281 reported by firmware. 3281 reported by firmware.
3282 Define a memory regio 3282 Define a memory region of size nn[KMG] starting at
3283 ss[KMG]. 3283 ss[KMG].
3284 Multiple different re 3284 Multiple different regions can be specified with
3285 multiple mem= paramet 3285 multiple mem= parameters on the command line.
3286 3286
3287 mem=nopentium [BUGS=X86-32] Disable 3287 mem=nopentium [BUGS=X86-32] Disable usage of 4MB pages for kernel
3288 memory. 3288 memory.
3289 3289
3290 memblock=debug [KNL,EARLY] Enable me 3290 memblock=debug [KNL,EARLY] Enable memblock debug messages.
3291 3291
3292 memchunk=nn[KMG] 3292 memchunk=nn[KMG]
3293 [KNL,SH] Allow user t 3293 [KNL,SH] Allow user to override the default size for
3294 per-device physically 3294 per-device physically contiguous DMA buffers.
3295 3295
3296 memhp_default_state=online/offline/on 3296 memhp_default_state=online/offline/online_kernel/online_movable
3297 [KNL] Set the initial 3297 [KNL] Set the initial state for the memory hotplug
3298 onlining policy. If n 3298 onlining policy. If not specified, the default value is
3299 set according to the 3299 set according to the
3300 CONFIG_MEMORY_HOTPLUG 3300 CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
3301 option. 3301 option.
3302 See Documentation/adm 3302 See Documentation/admin-guide/mm/memory-hotplug.rst.
3303 3303
3304 memmap=exactmap [KNL,X86,EARLY] Enabl 3304 memmap=exactmap [KNL,X86,EARLY] Enable setting of an exact
3305 E820 memory map, as s 3305 E820 memory map, as specified by the user.
3306 Such memmap=exactmap 3306 Such memmap=exactmap lines can be constructed based on
3307 BIOS output or other 3307 BIOS output or other requirements. See the memmap=nn@ss
3308 option description. 3308 option description.
3309 3309
3310 memmap=nn[KMG]@ss[KMG] 3310 memmap=nn[KMG]@ss[KMG]
3311 [KNL, X86,MIPS,XTENSA 3311 [KNL, X86,MIPS,XTENSA,EARLY] Force usage of a specific region of memory.
3312 Region of memory to b 3312 Region of memory to be used is from ss to ss+nn.
3313 If @ss[KMG] is omitte 3313 If @ss[KMG] is omitted, it is equivalent to mem=nn[KMG],
3314 which limits max addr 3314 which limits max address to nn[KMG].
3315 Multiple different re 3315 Multiple different regions can be specified,
3316 comma delimited. 3316 comma delimited.
3317 Example: 3317 Example:
3318 memmap=100M@2 3318 memmap=100M@2G,100M#3G,1G!1024G
3319 3319
3320 memmap=nn[KMG]#ss[KMG] 3320 memmap=nn[KMG]#ss[KMG]
3321 [KNL,ACPI,EARLY] Mark 3321 [KNL,ACPI,EARLY] Mark specific memory as ACPI data.
3322 Region of memory to b 3322 Region of memory to be marked is from ss to ss+nn.
3323 3323
3324 memmap=nn[KMG]$ss[KMG] 3324 memmap=nn[KMG]$ss[KMG]
3325 [KNL,ACPI,EARLY] Mark 3325 [KNL,ACPI,EARLY] Mark specific memory as reserved.
3326 Region of memory to b 3326 Region of memory to be reserved is from ss to ss+nn.
3327 Example: Exclude memo 3327 Example: Exclude memory from 0x18690000-0x1869ffff
3328 memmap=64K$0 3328 memmap=64K$0x18690000
3329 or 3329 or
3330 memmap=0x100 3330 memmap=0x10000$0x18690000
3331 Some bootloaders may 3331 Some bootloaders may need an escape character before '$',
3332 like Grub2, otherwise 3332 like Grub2, otherwise '$' and the following number
3333 will be eaten. 3333 will be eaten.
3334 3334
3335 memmap=nn[KMG]!ss[KMG,EARLY] 3335 memmap=nn[KMG]!ss[KMG,EARLY]
3336 [KNL,X86] Mark specif 3336 [KNL,X86] Mark specific memory as protected.
3337 Region of memory to b 3337 Region of memory to be used, from ss to ss+nn.
3338 The memory region may 3338 The memory region may be marked as e820 type 12 (0xc)
3339 and is NVDIMM or ADR 3339 and is NVDIMM or ADR memory.
3340 3340
3341 memmap=<size>%<offset>-<oldtype>+<new 3341 memmap=<size>%<offset>-<oldtype>+<newtype>
3342 [KNL,ACPI,EARLY] Conv 3342 [KNL,ACPI,EARLY] Convert memory within the specified region
3343 from <oldtype> to <ne 3343 from <oldtype> to <newtype>. If "-<oldtype>" is left
3344 out, the whole region 3344 out, the whole region will be marked as <newtype>,
3345 even if previously un 3345 even if previously unavailable. If "+<newtype>" is left
3346 out, matching memory 3346 out, matching memory will be removed. Types are
3347 specified as e820 typ 3347 specified as e820 types, e.g., 1 = RAM, 2 = reserved,
3348 3 = ACPI, 12 = PRAM. 3348 3 = ACPI, 12 = PRAM.
3349 3349
3350 memory_corruption_check=0/1 [X86,EARL 3350 memory_corruption_check=0/1 [X86,EARLY]
3351 Some BIOSes seem to c 3351 Some BIOSes seem to corrupt the first 64k of
3352 memory when doing thi 3352 memory when doing things like suspend/resume.
3353 Setting this option w 3353 Setting this option will scan the memory
3354 looking for corruptio 3354 looking for corruption. Enabling this will
3355 both detect corruptio 3355 both detect corruption and prevent the kernel
3356 from using the memory 3356 from using the memory being corrupted.
3357 However, its intended 3357 However, its intended as a diagnostic tool; if
3358 repeatable BIOS-origi 3358 repeatable BIOS-originated corruption always
3359 affects the same memo 3359 affects the same memory, you can use memmap=
3360 to prevent the kernel 3360 to prevent the kernel from using that memory.
3361 3361
3362 memory_corruption_check_size=size [X8 3362 memory_corruption_check_size=size [X86,EARLY]
3363 By default it checks 3363 By default it checks for corruption in the low
3364 64k, making this memo 3364 64k, making this memory unavailable for normal
3365 use. Use this parame 3365 use. Use this parameter to scan for
3366 corruption in more or 3366 corruption in more or less memory.
3367 3367
3368 memory_corruption_check_period=second 3368 memory_corruption_check_period=seconds [X86,EARLY]
3369 By default it checks 3369 By default it checks for corruption every 60
3370 seconds. Use this pa 3370 seconds. Use this parameter to check at some
3371 other rate. 0 disabl 3371 other rate. 0 disables periodic checking.
3372 3372
3373 memory_hotplug.memmap_on_memory 3373 memory_hotplug.memmap_on_memory
3374 [KNL,X86,ARM] Boolean 3374 [KNL,X86,ARM] Boolean flag to enable this feature.
3375 Format: {on | off (de 3375 Format: {on | off (default)}
3376 When enabled, runtime 3376 When enabled, runtime hotplugged memory will
3377 allocate its internal 3377 allocate its internal metadata (struct pages,
3378 those vmemmap pages c 3378 those vmemmap pages cannot be optimized even
3379 if hugetlb_free_vmemm 3379 if hugetlb_free_vmemmap is enabled) from the
3380 hotadded memory which 3380 hotadded memory which will allow to hotadd a
3381 lot of memory without 3381 lot of memory without requiring additional
3382 memory to do so. 3382 memory to do so.
3383 This feature is disab 3383 This feature is disabled by default because it
3384 has some implication 3384 has some implication on large (e.g. GB)
3385 allocations in some c 3385 allocations in some configurations (e.g. small
3386 memory blocks). 3386 memory blocks).
3387 The state of the flag 3387 The state of the flag can be read in
3388 /sys/module/memory_ho 3388 /sys/module/memory_hotplug/parameters/memmap_on_memory.
3389 Note that even when e 3389 Note that even when enabled, there are a few cases where
3390 the feature is not ef 3390 the feature is not effective.
3391 3391
3392 memtest= [KNL,X86,ARM,M68K,PPC 3392 memtest= [KNL,X86,ARM,M68K,PPC,RISCV,EARLY] Enable memtest
3393 Format: <integer> 3393 Format: <integer>
3394 default : 0 <disable> 3394 default : 0 <disable>
3395 Specifies the number 3395 Specifies the number of memtest passes to be
3396 performed. Each pass 3396 performed. Each pass selects another test
3397 pattern from a given 3397 pattern from a given set of patterns. Memtest
3398 fills the memory with 3398 fills the memory with this pattern, validates
3399 memory contents and r 3399 memory contents and reserves bad memory
3400 regions that are dete 3400 regions that are detected.
3401 3401
3402 mem_encrypt= [X86-64] AMD Secure M 3402 mem_encrypt= [X86-64] AMD Secure Memory Encryption (SME) control
3403 Valid arguments: on, 3403 Valid arguments: on, off
3404 Default: off 3404 Default: off
3405 mem_encrypt=on: 3405 mem_encrypt=on: Activate SME
3406 mem_encrypt=off: 3406 mem_encrypt=off: Do not activate SME
3407 3407
3408 Refer to Documentatio 3408 Refer to Documentation/virt/kvm/x86/amd-memory-encryption.rst
3409 for details on when m 3409 for details on when memory encryption can be activated.
3410 3410
3411 mem_sleep_default= [SUSPEND] Def 3411 mem_sleep_default= [SUSPEND] Default system suspend mode:
3412 s2idle - Suspend-To- 3412 s2idle - Suspend-To-Idle
3413 shallow - Power-On Su 3413 shallow - Power-On Suspend or equivalent (if supported)
3414 deep - Suspend-To- 3414 deep - Suspend-To-RAM or equivalent (if supported)
3415 See Documentation/adm 3415 See Documentation/admin-guide/pm/sleep-states.rst.
3416 3416
3417 mfgptfix [X86-32] Fix MFGPT ti 3417 mfgptfix [X86-32] Fix MFGPT timers on AMD Geode platforms when
3418 the BIOS has incorrec 3418 the BIOS has incorrectly applied a workaround. TinyBIOS
3419 version 0.98 is known 3419 version 0.98 is known to be affected, 0.99 fixes the
3420 problem by letting th 3420 problem by letting the user disable the workaround.
3421 3421
3422 mga= [HW,DRM] 3422 mga= [HW,DRM]
3423 3423
3424 microcode.force_minrev= [X86] 3424 microcode.force_minrev= [X86]
3425 Format: <bool> 3425 Format: <bool>
3426 Enable or disable the 3426 Enable or disable the microcode minimal revision
3427 enforcement for the r 3427 enforcement for the runtime microcode loader.
3428 3428
3429 mini2440= [ARM,HW,KNL] 3429 mini2440= [ARM,HW,KNL]
3430 Format:[0..2][b][c][t 3430 Format:[0..2][b][c][t]
3431 Default: "0tb" 3431 Default: "0tb"
3432 MINI2440 configuratio 3432 MINI2440 configuration specification:
3433 0 - The attached scre 3433 0 - The attached screen is the 3.5" TFT
3434 1 - The attached scre 3434 1 - The attached screen is the 7" TFT
3435 2 - The VGA Shield is 3435 2 - The VGA Shield is attached (1024x768)
3436 Leaving out the scree 3436 Leaving out the screen size parameter will not load
3437 the TFT driver, and t 3437 the TFT driver, and the framebuffer will be left
3438 unconfigured. 3438 unconfigured.
3439 b - Enable backlight. 3439 b - Enable backlight. The TFT backlight pin will be
3440 linked to the kernel 3440 linked to the kernel VESA blanking code and a GPIO
3441 LED. This parameter i 3441 LED. This parameter is not necessary when using the
3442 VGA shield. 3442 VGA shield.
3443 c - Enable the s3c ca 3443 c - Enable the s3c camera interface.
3444 t - Reserved for enab 3444 t - Reserved for enabling touchscreen support. The
3445 touchscreen support i 3445 touchscreen support is not enabled in the mainstream
3446 kernel as of 2.6.30, 3446 kernel as of 2.6.30, a preliminary port can be found
3447 in the "bleeding edge 3447 in the "bleeding edge" mini2440 support kernel at
3448 https://repo.or.cz/w/ 3448 https://repo.or.cz/w/linux-2.6/mini2440.git
3449 3449
3450 mitigations= 3450 mitigations=
3451 [X86,PPC,S390,ARM64,E 3451 [X86,PPC,S390,ARM64,EARLY] Control optional mitigations for
3452 CPU vulnerabilities. 3452 CPU vulnerabilities. This is a set of curated,
3453 arch-independent opti 3453 arch-independent options, each of which is an
3454 aggregation of existi 3454 aggregation of existing arch-specific options.
3455 3455
3456 Note, "mitigations" i 3456 Note, "mitigations" is supported if and only if the
3457 kernel was built with 3457 kernel was built with CPU_MITIGATIONS=y.
3458 3458
3459 off 3459 off
3460 Disable all o 3460 Disable all optional CPU mitigations. This
3461 improves syst 3461 improves system performance, but it may also
3462 expose users 3462 expose users to several CPU vulnerabilities.
3463 Equivalent to 3463 Equivalent to: if nokaslr then kpti=0 [ARM64]
3464 3464 gather_data_sampling=off [X86]
3465 3465 kvm.nx_huge_pages=off [X86]
3466 3466 l1tf=off [X86]
3467 3467 mds=off [X86]
3468 3468 mmio_stale_data=off [X86]
3469 3469 no_entry_flush [PPC]
3470 3470 no_uaccess_flush [PPC]
3471 3471 nobp=0 [S390]
3472 3472 nopti [X86,PPC]
3473 3473 nospectre_bhb [ARM64]
3474 3474 nospectre_v1 [X86,PPC]
3475 3475 nospectre_v2 [X86,PPC,S390,ARM64]
3476 3476 reg_file_data_sampling=off [X86]
3477 3477 retbleed=off [X86]
3478 3478 spec_rstack_overflow=off [X86]
3479 3479 spec_store_bypass_disable=off [X86,PPC]
3480 3480 spectre_bhi=off [X86]
3481 3481 spectre_v2_user=off [X86]
3482 3482 srbds=off [X86,INTEL]
3483 3483 ssbd=force-off [ARM64]
3484 3484 tsx_async_abort=off [X86]
3485 3485
3486 Exceptions: 3486 Exceptions:
3487 3487 This does not have any effect on
3488 3488 kvm.nx_huge_pages when
3489 3489 kvm.nx_huge_pages=force.
3490 3490
3491 auto (default) 3491 auto (default)
3492 Mitigate all 3492 Mitigate all CPU vulnerabilities, but leave SMT
3493 enabled, even 3493 enabled, even if it's vulnerable. This is for
3494 users who don 3494 users who don't want to be surprised by SMT
3495 getting disab 3495 getting disabled across kernel upgrades, or who
3496 have other wa 3496 have other ways of avoiding SMT-based attacks.
3497 Equivalent to 3497 Equivalent to: (default behavior)
3498 3498
3499 auto,nosmt 3499 auto,nosmt
3500 Mitigate all 3500 Mitigate all CPU vulnerabilities, disabling SMT
3501 if needed. T 3501 if needed. This is for users who always want to
3502 be fully miti 3502 be fully mitigated, even if it means losing SMT.
3503 Equivalent to 3503 Equivalent to: l1tf=flush,nosmt [X86]
3504 3504 mds=full,nosmt [X86]
3505 3505 tsx_async_abort=full,nosmt [X86]
3506 3506 mmio_stale_data=full,nosmt [X86]
3507 3507 retbleed=auto,nosmt [X86]
3508 3508
3509 mminit_loglevel= 3509 mminit_loglevel=
3510 [KNL,EARLY] When CONF 3510 [KNL,EARLY] When CONFIG_DEBUG_MEMORY_INIT is set, this
3511 parameter allows cont 3511 parameter allows control of the logging verbosity for
3512 the additional memory 3512 the additional memory initialisation checks. A value
3513 of 0 disables mminit 3513 of 0 disables mminit logging and a level of 4 will
3514 log everything. Infor 3514 log everything. Information is printed at KERN_DEBUG
3515 so loglevel=8 may als 3515 so loglevel=8 may also need to be specified.
3516 3516
3517 mmio_stale_data= 3517 mmio_stale_data=
3518 [X86,INTEL,EARLY] Con 3518 [X86,INTEL,EARLY] Control mitigation for the Processor
3519 MMIO Stale Data vulne 3519 MMIO Stale Data vulnerabilities.
3520 3520
3521 Processor MMIO Stale 3521 Processor MMIO Stale Data is a class of
3522 vulnerabilities that 3522 vulnerabilities that may expose data after an MMIO
3523 operation. Exposed da 3523 operation. Exposed data could originate or end in
3524 the same CPU buffers 3524 the same CPU buffers as affected by MDS and TAA.
3525 Therefore, similar to 3525 Therefore, similar to MDS and TAA, the mitigation
3526 is to clear the affec 3526 is to clear the affected CPU buffers.
3527 3527
3528 This parameter contro 3528 This parameter controls the mitigation. The
3529 options are: 3529 options are:
3530 3530
3531 full - Enable m 3531 full - Enable mitigation on vulnerable CPUs
3532 3532
3533 full,nosmt - Enable m 3533 full,nosmt - Enable mitigation and disable SMT on
3534 vulnerab 3534 vulnerable CPUs.
3535 3535
3536 off - Uncondit 3536 off - Unconditionally disable mitigation
3537 3537
3538 On MDS or TAA affecte 3538 On MDS or TAA affected machines,
3539 mmio_stale_data=off c 3539 mmio_stale_data=off can be prevented by an active
3540 MDS or TAA mitigation 3540 MDS or TAA mitigation as these vulnerabilities are
3541 mitigated with the sa 3541 mitigated with the same mechanism so in order to
3542 disable this mitigati 3542 disable this mitigation, you need to specify
3543 mds=off and tsx_async 3543 mds=off and tsx_async_abort=off too.
3544 3544
3545 Not specifying this o 3545 Not specifying this option is equivalent to
3546 mmio_stale_data=full. 3546 mmio_stale_data=full.
3547 3547
3548 For details see: 3548 For details see:
3549 Documentation/admin-g 3549 Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst
3550 3550
3551 <module>.async_probe[=<bool>] [KNL] 3551 <module>.async_probe[=<bool>] [KNL]
3552 If no <bool> value is 3552 If no <bool> value is specified or if the value
3553 specified is not a va 3553 specified is not a valid <bool>, enable asynchronous
3554 probe on this module. 3554 probe on this module. Otherwise, enable/disable
3555 asynchronous probe on 3555 asynchronous probe on this module as indicated by the
3556 <bool> value. See als 3556 <bool> value. See also: module.async_probe
3557 3557
3558 module.async_probe=<bool> 3558 module.async_probe=<bool>
3559 [KNL] When set to tru 3559 [KNL] When set to true, modules will use async probing
3560 by default. To enable 3560 by default. To enable/disable async probing for a
3561 specific module, use 3561 specific module, use the module specific control that
3562 is documented under < 3562 is documented under <module>.async_probe. When both
3563 module.async_probe an 3563 module.async_probe and <module>.async_probe are
3564 specified, <module>.a 3564 specified, <module>.async_probe takes precedence for
3565 the specific module. 3565 the specific module.
3566 3566
3567 module.enable_dups_trace 3567 module.enable_dups_trace
3568 [KNL] When CONFIG_MOD 3568 [KNL] When CONFIG_MODULE_DEBUG_AUTOLOAD_DUPS is set,
3569 this means that dupli 3569 this means that duplicate request_module() calls will
3570 trigger a WARN_ON() i 3570 trigger a WARN_ON() instead of a pr_warn(). Note that
3571 if MODULE_DEBUG_AUTOL 3571 if MODULE_DEBUG_AUTOLOAD_DUPS_TRACE is set, WARN_ON()s
3572 will always be issued 3572 will always be issued and this option does nothing.
3573 module.sig_enforce 3573 module.sig_enforce
3574 [KNL] When CONFIG_MOD 3574 [KNL] When CONFIG_MODULE_SIG is set, this means that
3575 modules without (vali 3575 modules without (valid) signatures will fail to load.
3576 Note that if CONFIG_M 3576 Note that if CONFIG_MODULE_SIG_FORCE is set, that
3577 is always true, so th 3577 is always true, so this option does nothing.
3578 3578
3579 module_blacklist= [KNL] Do not load 3579 module_blacklist= [KNL] Do not load a comma-separated list of
3580 modules. Useful for 3580 modules. Useful for debugging problem modules.
3581 3581
3582 mousedev.tap_time= 3582 mousedev.tap_time=
3583 [MOUSE] Maximum time 3583 [MOUSE] Maximum time between finger touching and
3584 leaving touchpad surf 3584 leaving touchpad surface for touch to be considered
3585 a tap and be reported 3585 a tap and be reported as a left button click (for
3586 touchpads working in 3586 touchpads working in absolute mode only).
3587 Format: <msecs> 3587 Format: <msecs>
3588 mousedev.xres= [MOUSE] Horizontal sc 3588 mousedev.xres= [MOUSE] Horizontal screen resolution, used for devices
3589 reporting absolute co 3589 reporting absolute coordinates, such as tablets
3590 mousedev.yres= [MOUSE] Vertical scre 3590 mousedev.yres= [MOUSE] Vertical screen resolution, used for devices
3591 reporting absolute co 3591 reporting absolute coordinates, such as tablets
3592 3592
3593 movablecore= [KNL,X86,PPC,EARLY] 3593 movablecore= [KNL,X86,PPC,EARLY]
3594 Format: nn[KMGTPE] | 3594 Format: nn[KMGTPE] | nn%
3595 This parameter is the 3595 This parameter is the complement to kernelcore=, it
3596 specifies the amount 3596 specifies the amount of memory used for migratable
3597 allocations. If both 3597 allocations. If both kernelcore and movablecore is
3598 specified, then kerne 3598 specified, then kernelcore will be at *least* the
3599 specified value but m 3599 specified value but may be more. If movablecore on its
3600 own is specified, the 3600 own is specified, the administrator must be careful
3601 that the amount of me 3601 that the amount of memory usable for all allocations
3602 is not too small. 3602 is not too small.
3603 3603
3604 movable_node [KNL,EARLY] Boot-time 3604 movable_node [KNL,EARLY] Boot-time switch to make hotplugable memory
3605 NUMA nodes to be mova 3605 NUMA nodes to be movable. This means that the memory
3606 of such nodes will be 3606 of such nodes will be usable only for movable
3607 allocations which rul 3607 allocations which rules out almost all kernel
3608 allocations. Use with 3608 allocations. Use with caution!
3609 3609
3610 MTD_Partition= [MTD] 3610 MTD_Partition= [MTD]
3611 Format: <name>,<regio 3611 Format: <name>,<region-number>,<size>,<offset>
3612 3612
3613 MTD_Region= [MTD] Format: 3613 MTD_Region= [MTD] Format:
3614 <name>,<region-number 3614 <name>,<region-number>[,<base>,<size>,<buswidth>,<altbuswidth>]
3615 3615
3616 mtdparts= [MTD] 3616 mtdparts= [MTD]
3617 See drivers/mtd/parse 3617 See drivers/mtd/parsers/cmdlinepart.c
3618 3618
3619 mtouchusb.raw_coordinates= 3619 mtouchusb.raw_coordinates=
3620 [HW] Make the MicroTo 3620 [HW] Make the MicroTouch USB driver use raw coordinates
3621 ('y', default) or coo 3621 ('y', default) or cooked coordinates ('n')
3622 3622
3623 mtrr=debug [X86,EARLY] 3623 mtrr=debug [X86,EARLY]
3624 Enable printing debug 3624 Enable printing debug information related to MTRR
3625 registers at boot tim 3625 registers at boot time.
3626 3626
3627 mtrr_chunk_size=nn[KMG,X86,EARLY] 3627 mtrr_chunk_size=nn[KMG,X86,EARLY]
3628 used for mtrr cleanup 3628 used for mtrr cleanup. It is largest continuous chunk
3629 that could hold holes 3629 that could hold holes aka. UC entries.
3630 3630
3631 mtrr_gran_size=nn[KMG,X86,EARLY] 3631 mtrr_gran_size=nn[KMG,X86,EARLY]
3632 Used for mtrr cleanup 3632 Used for mtrr cleanup. It is granularity of mtrr block.
3633 Default is 1. 3633 Default is 1.
3634 Large value could pre 3634 Large value could prevent small alignment from
3635 using up MTRRs. 3635 using up MTRRs.
3636 3636
3637 mtrr_spare_reg_nr=n [X86,EARLY] 3637 mtrr_spare_reg_nr=n [X86,EARLY]
3638 Format: <integer> 3638 Format: <integer>
3639 Range: 0,7 : spare re 3639 Range: 0,7 : spare reg number
3640 Default : 1 3640 Default : 1
3641 Used for mtrr cleanup 3641 Used for mtrr cleanup. It is spare mtrr entries number.
3642 Set to 2 or more if y 3642 Set to 2 or more if your graphical card needs more.
3643 3643
3644 multitce=off [PPC] This parameter 3644 multitce=off [PPC] This parameter disables the use of the pSeries
3645 firmware feature for 3645 firmware feature for updating multiple TCE entries
3646 at a time. 3646 at a time.
3647 3647
3648 n2= [NET] SDL Inc. RISCom 3648 n2= [NET] SDL Inc. RISCom/N2 synchronous serial card
3649 3649
3650 netdev= [NET] Network devices 3650 netdev= [NET] Network devices parameters
3651 Format: <irq>,<io>,<m 3651 Format: <irq>,<io>,<mem_start>,<mem_end>,<name>
3652 Note that mem_start i 3652 Note that mem_start is often overloaded to mean
3653 something different a 3653 something different and driver-specific.
3654 This usage is only do 3654 This usage is only documented in each driver source
3655 file if at all. 3655 file if at all.
3656 3656
3657 netpoll.carrier_timeout= 3657 netpoll.carrier_timeout=
3658 [NET] Specifies amoun 3658 [NET] Specifies amount of time (in seconds) that
3659 netpoll should wait f 3659 netpoll should wait for a carrier. By default netpoll
3660 waits 4 seconds. 3660 waits 4 seconds.
3661 3661
3662 nf_conntrack.acct= 3662 nf_conntrack.acct=
3663 [NETFILTER] Enable co 3663 [NETFILTER] Enable connection tracking flow accounting
3664 0 to disable accounti 3664 0 to disable accounting
3665 1 to enable accountin 3665 1 to enable accounting
3666 Default value is 0. 3666 Default value is 0.
3667 3667
3668 nfs.cache_getent= 3668 nfs.cache_getent=
3669 [NFS] sets the pathna 3669 [NFS] sets the pathname to the program which is used
3670 to update the NFS cli 3670 to update the NFS client cache entries.
3671 3671
3672 nfs.cache_getent_timeout= 3672 nfs.cache_getent_timeout=
3673 [NFS] sets the timeou 3673 [NFS] sets the timeout after which an attempt to
3674 update a cache entry 3674 update a cache entry is deemed to have failed.
3675 3675
3676 nfs.callback_nr_threads= 3676 nfs.callback_nr_threads=
3677 [NFSv4] set the total 3677 [NFSv4] set the total number of threads that the
3678 NFS client will assig 3678 NFS client will assign to service NFSv4 callback
3679 requests. 3679 requests.
3680 3680
3681 nfs.callback_tcpport= 3681 nfs.callback_tcpport=
3682 [NFS] set the TCP por 3682 [NFS] set the TCP port on which the NFSv4 callback
3683 channel should listen 3683 channel should listen.
3684 3684
3685 nfs.delay_retrans= 3685 nfs.delay_retrans=
3686 [NFS] specifies the n 3686 [NFS] specifies the number of times the NFSv4 client
3687 retries the request b 3687 retries the request before returning an EAGAIN error,
3688 after a reply of NFS4 3688 after a reply of NFS4ERR_DELAY from the server.
3689 Only applies if the s 3689 Only applies if the softerr mount option is enabled,
3690 and the specified val 3690 and the specified value is >= 0.
3691 3691
3692 nfs.enable_ino64= 3692 nfs.enable_ino64=
3693 [NFS] enable 64-bit i 3693 [NFS] enable 64-bit inode numbers.
3694 If zero, the NFS clie 3694 If zero, the NFS client will fake up a 32-bit inode
3695 number for the readdi 3695 number for the readdir() and stat() syscalls instead
3696 of returning the full 3696 of returning the full 64-bit number.
3697 The default is to ret 3697 The default is to return 64-bit inode numbers.
3698 3698
3699 nfs.idmap_cache_timeout= 3699 nfs.idmap_cache_timeout=
3700 [NFS] set the maximum 3700 [NFS] set the maximum lifetime for idmapper cache
3701 entries. 3701 entries.
3702 3702
3703 nfs.max_session_cb_slots= 3703 nfs.max_session_cb_slots=
3704 [NFSv4.1] Sets the ma 3704 [NFSv4.1] Sets the maximum number of session
3705 slots the client will 3705 slots the client will assign to the callback
3706 channel. This determi 3706 channel. This determines the maximum number of
3707 callbacks the client 3707 callbacks the client will process in parallel for
3708 a particular server. 3708 a particular server.
3709 3709
3710 nfs.max_session_slots= 3710 nfs.max_session_slots=
3711 [NFSv4.1] Sets the ma 3711 [NFSv4.1] Sets the maximum number of session slots
3712 the client will attem 3712 the client will attempt to negotiate with the server.
3713 This limits the numbe 3713 This limits the number of simultaneous RPC requests
3714 that the client can s 3714 that the client can send to the NFSv4.1 server.
3715 Note that there is li 3715 Note that there is little point in setting this
3716 value higher than the 3716 value higher than the max_tcp_slot_table_limit.
3717 3717
3718 nfs.nfs4_disable_idmapping= 3718 nfs.nfs4_disable_idmapping=
3719 [NFSv4] When set to t 3719 [NFSv4] When set to the default of '1', this option
3720 ensures that both the 3720 ensures that both the RPC level authentication
3721 scheme and the NFS le 3721 scheme and the NFS level operations agree to use
3722 numeric uids/gids if 3722 numeric uids/gids if the mount is using the
3723 'sec=sys' security fl 3723 'sec=sys' security flavour. In effect it is
3724 disabling idmapping, 3724 disabling idmapping, which can make migration from
3725 legacy NFSv2/v3 syste 3725 legacy NFSv2/v3 systems to NFSv4 easier.
3726 Servers that do not s 3726 Servers that do not support this mode of operation
3727 will be autodetected 3727 will be autodetected by the client, and it will fall
3728 back to using the idm 3728 back to using the idmapper.
3729 To turn off this beha 3729 To turn off this behaviour, set the value to '0'.
3730 3730
3731 nfs.nfs4_unique_id= 3731 nfs.nfs4_unique_id=
3732 [NFS4] Specify an add 3732 [NFS4] Specify an additional fixed unique ident-
3733 ification string that 3733 ification string that NFSv4 clients can insert into
3734 their nfs_client_id4 3734 their nfs_client_id4 string. This is typically a
3735 UUID that is generate 3735 UUID that is generated at system install time.
3736 3736
3737 nfs.recover_lost_locks= 3737 nfs.recover_lost_locks=
3738 [NFSv4] Attempt to re 3738 [NFSv4] Attempt to recover locks that were lost due
3739 to a lease timeout on 3739 to a lease timeout on the server. Please note that
3740 doing this risks data 3740 doing this risks data corruption, since there are
3741 no guarantees that th 3741 no guarantees that the file will remain unchanged
3742 after the locks are l 3742 after the locks are lost.
3743 If you want to enable 3743 If you want to enable the kernel legacy behaviour of
3744 attempting to recover 3744 attempting to recover these locks, then set this
3745 parameter to '1'. 3745 parameter to '1'.
3746 The default parameter 3746 The default parameter value of '0' causes the kernel
3747 not to attempt recove 3747 not to attempt recovery of lost locks.
3748 3748
3749 nfs.send_implementation_id= 3749 nfs.send_implementation_id=
3750 [NFSv4.1] Send client 3750 [NFSv4.1] Send client implementation identification
3751 information in exchan 3751 information in exchange_id requests.
3752 If zero, no implement 3752 If zero, no implementation identification information
3753 will be sent. 3753 will be sent.
3754 The default is to sen 3754 The default is to send the implementation identification
3755 information. 3755 information.
3756 3756
3757 nfs4.layoutstats_timer= 3757 nfs4.layoutstats_timer=
3758 [NFSv4.2] Change the 3758 [NFSv4.2] Change the rate at which the kernel sends
3759 layoutstats to the pN 3759 layoutstats to the pNFS metadata server.
3760 3760
3761 Setting this to value 3761 Setting this to value to 0 causes the kernel to use
3762 whatever value is the 3762 whatever value is the default set by the layout
3763 driver. A non-zero va 3763 driver. A non-zero value sets the minimum interval
3764 in seconds between la 3764 in seconds between layoutstats transmissions.
3765 3765
3766 nfsd.inter_copy_offload_enable= 3766 nfsd.inter_copy_offload_enable=
3767 [NFSv4.2] When set to 3767 [NFSv4.2] When set to 1, the server will support
3768 server-to-server copi 3768 server-to-server copies for which this server is
3769 the destination of th 3769 the destination of the copy.
3770 3770
3771 nfsd.nfs4_disable_idmapping= 3771 nfsd.nfs4_disable_idmapping=
3772 [NFSv4] When set to t 3772 [NFSv4] When set to the default of '1', the NFSv4
3773 server will return on 3773 server will return only numeric uids and gids to
3774 clients using auth_sy 3774 clients using auth_sys, and will accept numeric uids
3775 and gids from such cl 3775 and gids from such clients. This is intended to ease
3776 migration from NFSv2/ 3776 migration from NFSv2/v3.
3777 3777
3778 nfsd.nfsd4_ssc_umount_timeout= 3778 nfsd.nfsd4_ssc_umount_timeout=
3779 [NFSv4.2] When used a 3779 [NFSv4.2] When used as the destination of a
3780 server-to-server copy 3780 server-to-server copy, knfsd temporarily mounts
3781 the source server. I 3781 the source server. It caches the mount in case
3782 it will be needed aga 3782 it will be needed again, and discards it if not
3783 used for the number o 3783 used for the number of milliseconds specified by
3784 this parameter. 3784 this parameter.
3785 3785
3786 nfsaddrs= [NFS] Deprecated. Us 3786 nfsaddrs= [NFS] Deprecated. Use ip= instead.
3787 See Documentation/adm 3787 See Documentation/admin-guide/nfs/nfsroot.rst.
3788 3788
3789 nfsroot= [NFS] nfs root filesy 3789 nfsroot= [NFS] nfs root filesystem for disk-less boxes.
3790 See Documentation/adm 3790 See Documentation/admin-guide/nfs/nfsroot.rst.
3791 3791
3792 nfsrootdebug [NFS] enable nfsroot 3792 nfsrootdebug [NFS] enable nfsroot debugging messages.
3793 See Documentation/adm 3793 See Documentation/admin-guide/nfs/nfsroot.rst.
3794 3794
3795 nmi_backtrace.backtrace_idle [KNL] 3795 nmi_backtrace.backtrace_idle [KNL]
3796 Dump stacks even of i 3796 Dump stacks even of idle CPUs in response to an
3797 NMI stack-backtrace r 3797 NMI stack-backtrace request.
3798 3798
3799 nmi_debug= [KNL,SH] Specify one 3799 nmi_debug= [KNL,SH] Specify one or more actions to take
3800 when a NMI is trigger 3800 when a NMI is triggered.
3801 Format: [state][,regs 3801 Format: [state][,regs][,debounce][,die]
3802 3802
3803 nmi_watchdog= [KNL,BUGS=X86] Debugg 3803 nmi_watchdog= [KNL,BUGS=X86] Debugging features for SMP kernels
3804 Format: [panic,][nopa 3804 Format: [panic,][nopanic,][rNNN,][num]
3805 Valid num: 0 or 1 3805 Valid num: 0 or 1
3806 0 - turn hardlockup d 3806 0 - turn hardlockup detector in nmi_watchdog off
3807 1 - turn hardlockup d 3807 1 - turn hardlockup detector in nmi_watchdog on
3808 rNNN - configure the 3808 rNNN - configure the watchdog with raw perf event 0xNNN
3809 3809
3810 When panic is specifi 3810 When panic is specified, panic when an NMI watchdog
3811 timeout occurs (or 'n 3811 timeout occurs (or 'nopanic' to not panic on an NMI
3812 watchdog, if CONFIG_B 3812 watchdog, if CONFIG_BOOTPARAM_HARDLOCKUP_PANIC is set)
3813 To disable both hard 3813 To disable both hard and soft lockup detectors,
3814 please see 'nowatchdo 3814 please see 'nowatchdog'.
3815 This is useful when y 3815 This is useful when you use a panic=... timeout and
3816 need the box quickly 3816 need the box quickly up again.
3817 3817
3818 These settings can be 3818 These settings can be accessed at runtime via
3819 the nmi_watchdog and 3819 the nmi_watchdog and hardlockup_panic sysctls.
3820 3820
3821 no387 [BUGS=X86-32] Tells t 3821 no387 [BUGS=X86-32] Tells the kernel to use the 387 maths
3822 emulation library eve 3822 emulation library even if a 387 maths coprocessor
3823 is present. 3823 is present.
3824 3824
3825 no4lvl [RISCV,EARLY] Disable 3825 no4lvl [RISCV,EARLY] Disable 4-level and 5-level paging modes.
3826 Forces kernel to use 3826 Forces kernel to use 3-level paging instead.
3827 3827
3828 no5lvl [X86-64,RISCV,EARLY] 3828 no5lvl [X86-64,RISCV,EARLY] Disable 5-level paging mode. Forces
3829 kernel to use 4-level 3829 kernel to use 4-level paging instead.
3830 3830
3831 noalign [KNL,ARM] 3831 noalign [KNL,ARM]
3832 3832
3833 noapic [SMP,APIC,EARLY] Tell 3833 noapic [SMP,APIC,EARLY] Tells the kernel to not make use of any
3834 IOAPICs that may be p 3834 IOAPICs that may be present in the system.
3835 3835
3836 noautogroup Disable scheduler aut 3836 noautogroup Disable scheduler automatic task group creation.
3837 3837
3838 nocache [ARM,EARLY] 3838 nocache [ARM,EARLY]
3839 3839
3840 no_console_suspend 3840 no_console_suspend
3841 [HW] Never suspend th 3841 [HW] Never suspend the console
3842 Disable suspending of 3842 Disable suspending of consoles during suspend and
3843 hibernate operations. 3843 hibernate operations. Once disabled, debugging
3844 messages can reach va 3844 messages can reach various consoles while the rest
3845 of the system is bein 3845 of the system is being put to sleep (ie, while
3846 debugging driver susp 3846 debugging driver suspend/resume hooks). This may
3847 not work reliably wit 3847 not work reliably with all consoles, but is known
3848 to work with serial a 3848 to work with serial and VGA consoles.
3849 To facilitate more fl 3849 To facilitate more flexible debugging, we also add
3850 console_suspend, a pr 3850 console_suspend, a printk module parameter to control
3851 it. Users could use c 3851 it. Users could use console_suspend (usually
3852 /sys/module/printk/pa 3852 /sys/module/printk/parameters/console_suspend) to
3853 turn on/off it dynami 3853 turn on/off it dynamically.
3854 3854
3855 no_debug_objects 3855 no_debug_objects
3856 [KNL,EARLY] Disable o 3856 [KNL,EARLY] Disable object debugging
3857 3857
3858 nodsp [SH] Disable hardware 3858 nodsp [SH] Disable hardware DSP at boot time.
3859 3859
3860 noefi [EFI,EARLY] Disable E 3860 noefi [EFI,EARLY] Disable EFI runtime services support.
3861 3861
3862 no_entry_flush [PPC,EARLY] Don't flu 3862 no_entry_flush [PPC,EARLY] Don't flush the L1-D cache when entering the kernel.
3863 3863
3864 noexec32 [X86-64] 3864 noexec32 [X86-64]
3865 This affects only 32- 3865 This affects only 32-bit executables.
3866 noexec32=on: enable n 3866 noexec32=on: enable non-executable mappings (default)
3867 read doesn't 3867 read doesn't imply executable mappings
3868 noexec32=off: disable 3868 noexec32=off: disable non-executable mappings
3869 read implies 3869 read implies executable mappings
3870 3870
3871 no_file_caps Tells the kernel not 3871 no_file_caps Tells the kernel not to honor file capabilities. The
3872 only way then for a f 3872 only way then for a file to be executed with privilege
3873 is to be setuid root 3873 is to be setuid root or executed by root.
3874 3874
3875 nofpu [MIPS,SH] Disable har 3875 nofpu [MIPS,SH] Disable hardware FPU at boot time.
3876 3876
3877 nofsgsbase [X86] Disables FSGSBA 3877 nofsgsbase [X86] Disables FSGSBASE instructions.
3878 3878
3879 nofxsr [BUGS=X86-32] Disable 3879 nofxsr [BUGS=X86-32] Disables x86 floating point extended
3880 register save and res 3880 register save and restore. The kernel will only save
3881 legacy floating-point 3881 legacy floating-point registers on task switch.
3882 3882
3883 no_hash_pointers 3883 no_hash_pointers
3884 [KNL,EARLY] 3884 [KNL,EARLY]
3885 Force pointers printe 3885 Force pointers printed to the console or buffers to be
3886 unhashed. By default 3886 unhashed. By default, when a pointer is printed via %p
3887 format string, that p 3887 format string, that pointer is "hashed", i.e. obscured
3888 by hashing the pointe 3888 by hashing the pointer value. This is a security feature
3889 that hides actual ker 3889 that hides actual kernel addresses from unprivileged
3890 users, but it also ma 3890 users, but it also makes debugging the kernel more
3891 difficult since unequ 3891 difficult since unequal pointers can no longer be
3892 compared. However, i 3892 compared. However, if this command-line option is
3893 specified, then all n 3893 specified, then all normal pointers will have their true
3894 value printed. This o 3894 value printed. This option should only be specified when
3895 debugging the kernel. 3895 debugging the kernel. Please do not use on production
3896 kernels. 3896 kernels.
3897 3897
3898 nohibernate [HIBERNATION] Disable 3898 nohibernate [HIBERNATION] Disable hibernation and resume.
3899 3899
3900 nohlt [ARM,ARM64,MICROBLAZE 3900 nohlt [ARM,ARM64,MICROBLAZE,MIPS,PPC,RISCV,SH] Forces the kernel to
3901 busy wait in do_idle( 3901 busy wait in do_idle() and not use the arch_cpu_idle()
3902 implementation; requi 3902 implementation; requires CONFIG_GENERIC_IDLE_POLL_SETUP
3903 to be effective. This 3903 to be effective. This is useful on platforms where the
3904 sleep(SH) or wfi(ARM, 3904 sleep(SH) or wfi(ARM,ARM64) instructions do not work
3905 correctly or when doi 3905 correctly or when doing power measurements to evaluate
3906 the impact of the sle 3906 the impact of the sleep instructions. This is also
3907 useful when using JTA 3907 useful when using JTAG debugger.
3908 3908
3909 nohugeiomap [KNL,X86,PPC,ARM64,EA 3909 nohugeiomap [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge I/O mappings.
3910 3910
3911 nohugevmalloc [KNL,X86,PPC,ARM64,EA 3911 nohugevmalloc [KNL,X86,PPC,ARM64,EARLY] Disable kernel huge vmalloc mappings.
3912 3912
3913 nohz= [KNL] Boottime enable 3913 nohz= [KNL] Boottime enable/disable dynamic ticks
3914 Valid arguments: on, 3914 Valid arguments: on, off
3915 Default: on 3915 Default: on
3916 3916
3917 nohz_full= [KNL,BOOT,SMP,ISOL] 3917 nohz_full= [KNL,BOOT,SMP,ISOL]
3918 The argument is a cpu 3918 The argument is a cpu list, as described above.
3919 In kernels built with 3919 In kernels built with CONFIG_NO_HZ_FULL=y, set
3920 the specified list of 3920 the specified list of CPUs whose tick will be stopped
3921 whenever possible. Th 3921 whenever possible. The boot CPU will be forced outside
3922 the range to maintain 3922 the range to maintain the timekeeping. Any CPUs
3923 in this list will hav 3923 in this list will have their RCU callbacks offloaded,
3924 just as if they had a 3924 just as if they had also been called out in the
3925 rcu_nocbs= boot param 3925 rcu_nocbs= boot parameter.
3926 3926
3927 Note that this argume 3927 Note that this argument takes precedence over
3928 the CONFIG_RCU_NOCB_C 3928 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
3929 3929
3930 noinitrd [RAM] Tells the kerne 3930 noinitrd [RAM] Tells the kernel not to load any configured
3931 initial RAM disk. 3931 initial RAM disk.
3932 3932
3933 nointremap [X86-64,Intel-IOMMU,E 3933 nointremap [X86-64,Intel-IOMMU,EARLY] Do not enable interrupt
3934 remapping. 3934 remapping.
3935 [Deprecated - use int 3935 [Deprecated - use intremap=off]
3936 3936
3937 noinvpcid [X86,EARLY] Disable t 3937 noinvpcid [X86,EARLY] Disable the INVPCID cpu feature.
3938 3938
3939 noiotrap [SH] Disables trapped 3939 noiotrap [SH] Disables trapped I/O port accesses.
3940 3940
3941 noirqdebug [X86-32] Disables the 3941 noirqdebug [X86-32] Disables the code which attempts to detect and
3942 disable unhandled int 3942 disable unhandled interrupt sources.
3943 3943
3944 noisapnp [ISAPNP] Disables ISA 3944 noisapnp [ISAPNP] Disables ISA PnP code.
3945 3945
3946 nokaslr [KNL,EARLY] 3946 nokaslr [KNL,EARLY]
3947 When CONFIG_RANDOMIZE 3947 When CONFIG_RANDOMIZE_BASE is set, this disables
3948 kernel and module bas 3948 kernel and module base offset ASLR (Address Space
3949 Layout Randomization) 3949 Layout Randomization).
3950 3950
3951 no-kvmapf [X86,KVM,EARLY] Disab 3951 no-kvmapf [X86,KVM,EARLY] Disable paravirtualized asynchronous page
3952 fault handling. 3952 fault handling.
3953 3953
3954 no-kvmclock [X86,KVM,EARLY] Disab 3954 no-kvmclock [X86,KVM,EARLY] Disable paravirtualized KVM clock driver
3955 3955
3956 nolapic [X86-32,APIC,EARLY] D 3956 nolapic [X86-32,APIC,EARLY] Do not enable or use the local APIC.
3957 3957
3958 nolapic_timer [X86-32,APIC,EARLY] D 3958 nolapic_timer [X86-32,APIC,EARLY] Do not use the local APIC timer.
3959 3959
3960 nomce [X86-32] Disable Mach 3960 nomce [X86-32] Disable Machine Check Exception
3961 3961
3962 nomfgpt [X86-32] Disable Mult 3962 nomfgpt [X86-32] Disable Multi-Function General Purpose
3963 Timer usage (for AMD 3963 Timer usage (for AMD Geode machines).
3964 3964
3965 nomodeset Disable kernel modese 3965 nomodeset Disable kernel modesetting. Most systems' firmware
3966 sets up a display mod 3966 sets up a display mode and provides framebuffer memory
3967 for output. With nomo 3967 for output. With nomodeset, DRM and fbdev drivers will
3968 not load if they coul 3968 not load if they could possibly displace the pre-
3969 initialized output. O 3969 initialized output. Only the system framebuffer will
3970 be available for use. 3970 be available for use. The respective drivers will not
3971 perform display-mode 3971 perform display-mode changes or accelerated rendering.
3972 3972
3973 Useful as error fallb 3973 Useful as error fallback, or for testing and debugging.
3974 3974
3975 nomodule Disable module load 3975 nomodule Disable module load
3976 3976
3977 nonmi_ipi [X86] Disable using N 3977 nonmi_ipi [X86] Disable using NMI IPIs during panic/reboot to
3978 shutdown the other cp 3978 shutdown the other cpus. Instead use the REBOOT_VECTOR
3979 irq. 3979 irq.
3980 3980
3981 nopat [X86,EARLY] Disable P 3981 nopat [X86,EARLY] Disable PAT (page attribute table extension of
3982 pagetables) support. 3982 pagetables) support.
3983 3983
3984 nopcid [X86-64,EARLY] Disabl 3984 nopcid [X86-64,EARLY] Disable the PCID cpu feature.
3985 3985
3986 nopku [X86] Disable Memory 3986 nopku [X86] Disable Memory Protection Keys CPU feature found
3987 in some Intel CPUs. 3987 in some Intel CPUs.
3988 3988
3989 nopti [X86-64,EARLY] 3989 nopti [X86-64,EARLY]
3990 Equivalent to pti=off 3990 Equivalent to pti=off
3991 3991
3992 nopv= [X86,XEN,KVM,HYPER_V, 3992 nopv= [X86,XEN,KVM,HYPER_V,VMWARE,EARLY]
3993 Disables the PV optim 3993 Disables the PV optimizations forcing the guest to run
3994 as generic guest with 3994 as generic guest with no PV drivers. Currently support
3995 XEN HVM, KVM, HYPER_V 3995 XEN HVM, KVM, HYPER_V and VMWARE guest.
3996 3996
3997 nopvspin [X86,XEN,KVM,EARLY] 3997 nopvspin [X86,XEN,KVM,EARLY]
3998 Disables the qspinloc 3998 Disables the qspinlock slow path using PV optimizations
3999 which allow the hyper 3999 which allow the hypervisor to 'idle' the guest on lock
4000 contention. 4000 contention.
4001 4001
4002 norandmaps Don't use address spa 4002 norandmaps Don't use address space randomization. Equivalent to
4003 echo 0 > /proc/sys/ke 4003 echo 0 > /proc/sys/kernel/randomize_va_space
4004 4004
4005 noreplace-smp [X86-32,SMP] Don't re 4005 noreplace-smp [X86-32,SMP] Don't replace SMP instructions
4006 with UP alternatives 4006 with UP alternatives
4007 4007
4008 noresume [SWSUSP] Disables res 4008 noresume [SWSUSP] Disables resume and restores original swap
4009 space. 4009 space.
4010 4010
4011 no-scroll [VGA] Disables scroll 4011 no-scroll [VGA] Disables scrollback.
4012 This is required for 4012 This is required for the Braillex ib80-piezo Braille
4013 reader made by F.H. P 4013 reader made by F.H. Papenmeier (Germany).
4014 4014
4015 nosgx [X86-64,SGX,EARLY] Di 4015 nosgx [X86-64,SGX,EARLY] Disables Intel SGX kernel support.
4016 4016
4017 nosmap [PPC,EARLY] 4017 nosmap [PPC,EARLY]
4018 Disable SMAP (Supervi 4018 Disable SMAP (Supervisor Mode Access Prevention)
4019 even if it is support 4019 even if it is supported by processor.
4020 4020
4021 nosmep [PPC64s,EARLY] 4021 nosmep [PPC64s,EARLY]
4022 Disable SMEP (Supervi 4022 Disable SMEP (Supervisor Mode Execution Prevention)
4023 even if it is support 4023 even if it is supported by processor.
4024 4024
4025 nosmp [SMP,EARLY] Tells an 4025 nosmp [SMP,EARLY] Tells an SMP kernel to act as a UP kernel,
4026 and disable the IO AP 4026 and disable the IO APIC. legacy for "maxcpus=0".
4027 4027
4028 nosmt [KNL,MIPS,PPC,S390,EA 4028 nosmt [KNL,MIPS,PPC,S390,EARLY] Disable symmetric multithreading (SMT).
4029 Equivalent to smt=1. 4029 Equivalent to smt=1.
4030 4030
4031 [KNL,X86,PPC] Disable 4031 [KNL,X86,PPC] Disable symmetric multithreading (SMT).
4032 nosmt=force: Force di 4032 nosmt=force: Force disable SMT, cannot be undone
4033 via the 4033 via the sysfs control file.
4034 4034
4035 nosoftlockup [KNL] Disable the sof 4035 nosoftlockup [KNL] Disable the soft-lockup detector.
4036 4036
4037 nospec_store_bypass_disable 4037 nospec_store_bypass_disable
4038 [HW,EARLY] Disable al 4038 [HW,EARLY] Disable all mitigations for the Speculative
4039 Store Bypass vulnerab 4039 Store Bypass vulnerability
4040 4040
4041 nospectre_bhb [ARM64,EARLY] Disable 4041 nospectre_bhb [ARM64,EARLY] Disable all mitigations for Spectre-BHB (branch
4042 history injection) vu 4042 history injection) vulnerability. System may allow data leaks
4043 with this option. 4043 with this option.
4044 4044
4045 nospectre_v1 [X86,PPC,EARLY] Disab 4045 nospectre_v1 [X86,PPC,EARLY] Disable mitigations for Spectre Variant 1
4046 (bounds check bypass) 4046 (bounds check bypass). With this option data leaks are
4047 possible in the syste 4047 possible in the system.
4048 4048
4049 nospectre_v2 [X86,PPC_E500,ARM64,E 4049 nospectre_v2 [X86,PPC_E500,ARM64,EARLY] Disable all mitigations
4050 for the Spectre varia 4050 for the Spectre variant 2 (indirect branch
4051 prediction) vulnerabi 4051 prediction) vulnerability. System may allow data
4052 leaks with this optio 4052 leaks with this option.
4053 4053
4054 no-steal-acc [X86,PV_OPS,ARM64,PPC 4054 no-steal-acc [X86,PV_OPS,ARM64,PPC/PSERIES,RISCV,LOONGARCH,EARLY]
4055 Disable paravirtualiz 4055 Disable paravirtualized steal time accounting. steal time
4056 is computed, but won' 4056 is computed, but won't influence scheduler behaviour
4057 4057
4058 nosync [HW,M68K] Disables sy 4058 nosync [HW,M68K] Disables sync negotiation for all devices.
4059 4059
4060 no_timer_check [X86,APIC] Disables t 4060 no_timer_check [X86,APIC] Disables the code which tests for
4061 broken timer IRQ sour 4061 broken timer IRQ sources.
4062 4062
4063 no_uaccess_flush 4063 no_uaccess_flush
4064 [PPC,EARLY] Don't flu 4064 [PPC,EARLY] Don't flush the L1-D cache after accessing user data.
4065 4065
4066 novmcoredd [KNL,KDUMP] 4066 novmcoredd [KNL,KDUMP]
4067 Disable device dump. 4067 Disable device dump. Device dump allows drivers to
4068 append dump data to v 4068 append dump data to vmcore so you can collect driver
4069 specified debug info. 4069 specified debug info. Drivers can append the data
4070 without any limit and 4070 without any limit and this data is stored in memory,
4071 so this may cause sig 4071 so this may cause significant memory stress. Disabling
4072 device dump can help 4072 device dump can help save memory but the driver debug
4073 data will be no longe 4073 data will be no longer available. This parameter
4074 is only available whe 4074 is only available when CONFIG_PROC_VMCORE_DEVICE_DUMP
4075 is set. 4075 is set.
4076 4076
4077 no-vmw-sched-clock 4077 no-vmw-sched-clock
4078 [X86,PV_OPS,EARLY] Di 4078 [X86,PV_OPS,EARLY] Disable paravirtualized VMware
4079 scheduler clock and u 4079 scheduler clock and use the default one.
4080 4080
4081 nowatchdog [KNL] Disable both lo 4081 nowatchdog [KNL] Disable both lockup detectors, i.e.
4082 soft-lockup and NMI w 4082 soft-lockup and NMI watchdog (hard-lockup).
4083 4083
4084 nowb [ARM,EARLY] 4084 nowb [ARM,EARLY]
4085 4085
4086 nox2apic [X86-64,APIC,EARLY] D 4086 nox2apic [X86-64,APIC,EARLY] Do not enable x2APIC mode.
4087 4087
4088 NOTE: this parameter 4088 NOTE: this parameter will be ignored on systems with the
4089 LEGACY_XAPIC_DISABLED 4089 LEGACY_XAPIC_DISABLED bit set in the
4090 IA32_XAPIC_DISABLE_ST 4090 IA32_XAPIC_DISABLE_STATUS MSR.
4091 4091
4092 noxsave [BUGS=X86] Disables x 4092 noxsave [BUGS=X86] Disables x86 extended register state save
4093 and restore using xsa 4093 and restore using xsave. The kernel will fallback to
4094 enabling legacy float 4094 enabling legacy floating-point and sse state.
4095 4095
4096 noxsaveopt [X86] Disables xsaveo 4096 noxsaveopt [X86] Disables xsaveopt used in saving x86 extended
4097 register states. The 4097 register states. The kernel will fall back to use
4098 xsave to save the sta 4098 xsave to save the states. By using this parameter,
4099 performance of saving 4099 performance of saving the states is degraded because
4100 xsave doesn't support 4100 xsave doesn't support modified optimization while
4101 xsaveopt supports it 4101 xsaveopt supports it on xsaveopt enabled systems.
4102 4102
4103 noxsaves [X86] Disables xsaves 4103 noxsaves [X86] Disables xsaves and xrstors used in saving and
4104 restoring x86 extende 4104 restoring x86 extended register state in compacted
4105 form of xsave area. T 4105 form of xsave area. The kernel will fall back to use
4106 xsaveopt and xrstor t 4106 xsaveopt and xrstor to save and restore the states
4107 in standard form of x 4107 in standard form of xsave area. By using this
4108 parameter, xsave area 4108 parameter, xsave area per process might occupy more
4109 memory on xsaves enab 4109 memory on xsaves enabled systems.
4110 4110
4111 nr_cpus= [SMP,EARLY] Maximum n 4111 nr_cpus= [SMP,EARLY] Maximum number of processors that an SMP kernel
4112 could support. nr_cp 4112 could support. nr_cpus=n : n >= 1 limits the kernel to
4113 support 'n' processor 4113 support 'n' processors. It could be larger than the
4114 number of already plu 4114 number of already plugged CPU during bootup, later in
4115 runtime you can physi 4115 runtime you can physically add extra cpu until it reaches
4116 n. So during boot up 4116 n. So during boot up some boot time memory for per-cpu
4117 variables need be pre 4117 variables need be pre-allocated for later physical cpu
4118 hot plugging. 4118 hot plugging.
4119 4119
4120 nr_uarts= [SERIAL] maximum numb 4120 nr_uarts= [SERIAL] maximum number of UARTs to be registered.
4121 4121
4122 numa=off [KNL, ARM64, PPC, RIS 4122 numa=off [KNL, ARM64, PPC, RISCV, SPARC, X86, EARLY]
4123 Disable NUMA, Only se 4123 Disable NUMA, Only set up a single NUMA node
4124 spanning all memory. 4124 spanning all memory.
4125 4125
4126 numa_balancing= [KNL,ARM64,PPC,RISCV, 4126 numa_balancing= [KNL,ARM64,PPC,RISCV,S390,X86] Enable or disable automatic
4127 NUMA balancing. 4127 NUMA balancing.
4128 Allowed values are en 4128 Allowed values are enable and disable
4129 4129
4130 numa_zonelist_order= [KNL, BOOT] Sele 4130 numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
4131 'node', 'default' can 4131 'node', 'default' can be specified
4132 This can be set from 4132 This can be set from sysctl after boot.
4133 See Documentation/adm 4133 See Documentation/admin-guide/sysctl/vm.rst for details.
4134 4134
4135 ohci1394_dma=early [HW,EARLY] en 4135 ohci1394_dma=early [HW,EARLY] enable debugging via the ohci1394 driver.
4136 See Documentation/cor 4136 See Documentation/core-api/debugging-via-ohci1394.rst for more
4137 info. 4137 info.
4138 4138
4139 olpc_ec_timeout= [OLPC] ms delay when 4139 olpc_ec_timeout= [OLPC] ms delay when issuing EC commands
4140 Rather than timing ou 4140 Rather than timing out after 20 ms if an EC
4141 command is not proper 4141 command is not properly ACKed, override the length
4142 of the timeout. We h 4142 of the timeout. We have interrupts disabled while
4143 waiting for the ACK, 4143 waiting for the ACK, so if this is set too high
4144 interrupts *may* be l 4144 interrupts *may* be lost!
4145 4145
4146 omap_mux= [OMAP] Override bootl 4146 omap_mux= [OMAP] Override bootloader pin multiplexing.
4147 Format: <mux_mode0.mo 4147 Format: <mux_mode0.mode_name=value>...
4148 For example, to overr 4148 For example, to override I2C bus2:
4149 omap_mux=i2c2_scl.i2c 4149 omap_mux=i2c2_scl.i2c2_scl=0x100,i2c2_sda.i2c2_sda=0x100
4150 4150
4151 onenand.bdry= [HW,MTD] Flex-OneNAND 4151 onenand.bdry= [HW,MTD] Flex-OneNAND Boundary Configuration
4152 4152
4153 Format: [die0_boundar 4153 Format: [die0_boundary][,die0_lock][,die1_boundary][,die1_lock]
4154 4154
4155 boundary - index of l 4155 boundary - index of last SLC block on Flex-OneNAND.
4156 The remain 4156 The remaining blocks are configured as MLC blocks.
4157 lock - Configure 4157 lock - Configure if Flex-OneNAND boundary should be locked.
4158 Once locke 4158 Once locked, the boundary cannot be changed.
4159 1 indicate 4159 1 indicates lock status, 0 indicates unlock status.
4160 4160
4161 oops=panic [KNL,EARLY] 4161 oops=panic [KNL,EARLY]
4162 Always panic on oopse 4162 Always panic on oopses. Default is to just kill the
4163 process, but there is 4163 process, but there is a small probability of
4164 deadlocking the machi 4164 deadlocking the machine.
4165 This will also cause 4165 This will also cause panics on machine check exceptions.
4166 Useful together with 4166 Useful together with panic=30 to trigger a reboot.
4167 4167
4168 page_alloc.shuffle= 4168 page_alloc.shuffle=
4169 [KNL] Boolean flag to 4169 [KNL] Boolean flag to control whether the page allocator
4170 should randomize its 4170 should randomize its free lists. This parameter can be
4171 used to enable/disabl 4171 used to enable/disable page randomization. The state of
4172 the flag can be read 4172 the flag can be read from sysfs at:
4173 /sys/module/page_allo 4173 /sys/module/page_alloc/parameters/shuffle.
4174 This parameter is onl 4174 This parameter is only available if CONFIG_SHUFFLE_PAGE_ALLOCATOR=y.
4175 4175
4176 page_owner= [KNL,EARLY] Boot-time 4176 page_owner= [KNL,EARLY] Boot-time page_owner enabling option.
4177 Storage of the inform 4177 Storage of the information about who allocated
4178 each page is disabled 4178 each page is disabled in default. With this switch,
4179 we can turn it on. 4179 we can turn it on.
4180 on: enable the featur 4180 on: enable the feature
4181 4181
4182 page_poison= [KNL,EARLY] Boot-time 4182 page_poison= [KNL,EARLY] Boot-time parameter changing the state of
4183 poisoning on the budd 4183 poisoning on the buddy allocator, available with
4184 CONFIG_PAGE_POISONING 4184 CONFIG_PAGE_POISONING=y.
4185 off: turn off poisoni 4185 off: turn off poisoning (default)
4186 on: turn on poisoning 4186 on: turn on poisoning
4187 4187
4188 page_reporting.page_reporting_order= 4188 page_reporting.page_reporting_order=
4189 [KNL] Minimal page re 4189 [KNL] Minimal page reporting order
4190 Format: <integer> 4190 Format: <integer>
4191 Adjust the minimal pa 4191 Adjust the minimal page reporting order. The page
4192 reporting is disabled 4192 reporting is disabled when it exceeds MAX_PAGE_ORDER.
4193 4193
4194 panic= [KNL] Kernel behaviou 4194 panic= [KNL] Kernel behaviour on panic: delay <timeout>
4195 timeout > 0: seconds 4195 timeout > 0: seconds before rebooting
4196 timeout = 0: wait for 4196 timeout = 0: wait forever
4197 timeout < 0: reboot i 4197 timeout < 0: reboot immediately
4198 Format: <timeout> 4198 Format: <timeout>
4199 4199
4200 panic_on_taint= [KNL,EARLY] 4200 panic_on_taint= [KNL,EARLY]
4201 Bitmask for condition 4201 Bitmask for conditionally calling panic() in add_taint()
4202 Format: <hex>[,nouser 4202 Format: <hex>[,nousertaint]
4203 Hexadecimal bitmask r 4203 Hexadecimal bitmask representing the set of TAINT flags
4204 that will cause the k 4204 that will cause the kernel to panic when add_taint() is
4205 called with any of th 4205 called with any of the flags in this set.
4206 The optional switch " 4206 The optional switch "nousertaint" can be utilized to
4207 prevent userspace for 4207 prevent userspace forced crashes by writing to sysctl
4208 /proc/sys/kernel/tain 4208 /proc/sys/kernel/tainted any flagset matching with the
4209 bitmask set on panic_ 4209 bitmask set on panic_on_taint.
4210 See Documentation/adm 4210 See Documentation/admin-guide/tainted-kernels.rst for
4211 extra details on the 4211 extra details on the taint flags that users can pick
4212 to compose the bitmas 4212 to compose the bitmask to assign to panic_on_taint.
4213 4213
4214 panic_on_warn=1 panic() instead of WA 4214 panic_on_warn=1 panic() instead of WARN(). Useful to cause kdump
4215 on a WARN(). 4215 on a WARN().
4216 4216
4217 panic_print= Bitmask for printing 4217 panic_print= Bitmask for printing system info when panic happens.
4218 User can chose combin 4218 User can chose combination of the following bits:
4219 bit 0: print all task 4219 bit 0: print all tasks info
4220 bit 1: print system m 4220 bit 1: print system memory info
4221 bit 2: print timer in 4221 bit 2: print timer info
4222 bit 3: print locks in 4222 bit 3: print locks info if CONFIG_LOCKDEP is on
4223 bit 4: print ftrace b 4223 bit 4: print ftrace buffer
4224 bit 5: print all prin 4224 bit 5: print all printk messages in buffer
4225 bit 6: print all CPUs 4225 bit 6: print all CPUs backtrace (if available in the arch)
4226 bit 7: print only tas 4226 bit 7: print only tasks in uninterruptible (blocked) state
4227 *Be aware* that this 4227 *Be aware* that this option may print a _lot_ of lines,
4228 so there are risks of 4228 so there are risks of losing older messages in the log.
4229 Use this option caref 4229 Use this option carefully, maybe worth to setup a
4230 bigger log buffer wit 4230 bigger log buffer with "log_buf_len" along with this.
4231 4231
4232 parkbd.port= [HW] Parallel port nu 4232 parkbd.port= [HW] Parallel port number the keyboard adapter is
4233 connected to, default 4233 connected to, default is 0.
4234 Format: <parport#> 4234 Format: <parport#>
4235 parkbd.mode= [HW] Parallel port ke 4235 parkbd.mode= [HW] Parallel port keyboard adapter mode of operation,
4236 0 for XT, 1 for AT (d 4236 0 for XT, 1 for AT (default is AT).
4237 Format: <mode> 4237 Format: <mode>
4238 4238
4239 parport= [HW,PPT] Specify para 4239 parport= [HW,PPT] Specify parallel ports. 0 disables.
4240 Format: { 0 | auto | 4240 Format: { 0 | auto | 0xBBB[,IRQ[,DMA]] }
4241 Use 'auto' to force t 4241 Use 'auto' to force the driver to use any
4242 IRQ/DMA settings dete 4242 IRQ/DMA settings detected (the default is to
4243 ignore detected IRQ/D 4243 ignore detected IRQ/DMA settings because of
4244 possible conflicts). 4244 possible conflicts). You can specify the base
4245 address, IRQ, and DMA 4245 address, IRQ, and DMA settings; IRQ and DMA
4246 should be numbers, or 4246 should be numbers, or 'auto' (for using detected
4247 settings on that part 4247 settings on that particular port), or 'nofifo'
4248 (to avoid using a FIF 4248 (to avoid using a FIFO even if it is detected).
4249 Parallel ports are as 4249 Parallel ports are assigned in the order they
4250 are specified on the 4250 are specified on the command line, starting
4251 with parport0. 4251 with parport0.
4252 4252
4253 parport_init_mode= [HW,PPT] 4253 parport_init_mode= [HW,PPT]
4254 Configure VIA paralle 4254 Configure VIA parallel port to operate in
4255 a specific mode. This 4255 a specific mode. This is necessary on Pegasos
4256 computer where firmwa 4256 computer where firmware has no options for setting
4257 up parallel port mode 4257 up parallel port mode and sets it to spp.
4258 Currently this functi 4258 Currently this function knows 686a and 8231 chips.
4259 Format: [spp|ps2|epp| 4259 Format: [spp|ps2|epp|ecp|ecpepp]
4260 4260
4261 pata_legacy.all= [HW,LIBATA] 4261 pata_legacy.all= [HW,LIBATA]
4262 Format: <int> 4262 Format: <int>
4263 Set to non-zero to pr 4263 Set to non-zero to probe primary and secondary ISA
4264 port ranges on PCI sy 4264 port ranges on PCI systems where no PCI PATA device
4265 has been found at eit 4265 has been found at either range. Disabled by default.
4266 4266
4267 pata_legacy.autospeed= [HW,LIBATA] 4267 pata_legacy.autospeed= [HW,LIBATA]
4268 Format: <int> 4268 Format: <int>
4269 Set to non-zero if a 4269 Set to non-zero if a chip is present that snoops speed
4270 changes. Disabled by 4270 changes. Disabled by default.
4271 4271
4272 pata_legacy.ht6560a= [HW,LIBATA] 4272 pata_legacy.ht6560a= [HW,LIBATA]
4273 Format: <int> 4273 Format: <int>
4274 Set to 1, 2, or 3 for 4274 Set to 1, 2, or 3 for HT 6560A on the primary channel,
4275 the secondary channel 4275 the secondary channel, or both channels respectively.
4276 Disabled by default. 4276 Disabled by default.
4277 4277
4278 pata_legacy.ht6560b= [HW,LIBATA] 4278 pata_legacy.ht6560b= [HW,LIBATA]
4279 Format: <int> 4279 Format: <int>
4280 Set to 1, 2, or 3 for 4280 Set to 1, 2, or 3 for HT 6560B on the primary channel,
4281 the secondary channel 4281 the secondary channel, or both channels respectively.
4282 Disabled by default. 4282 Disabled by default.
4283 4283
4284 pata_legacy.iordy_mask= [HW,LIBATA] 4284 pata_legacy.iordy_mask= [HW,LIBATA]
4285 Format: <int> 4285 Format: <int>
4286 IORDY enable mask. S 4286 IORDY enable mask. Set individual bits to allow IORDY
4287 for the respective ch 4287 for the respective channel. Bit 0 is for the first
4288 legacy channel handle 4288 legacy channel handled by this driver, bit 1 is for
4289 the second channel, a 4289 the second channel, and so on. The sequence will often
4290 correspond to the pri 4290 correspond to the primary legacy channel, the secondary
4291 legacy channel, and s 4291 legacy channel, and so on, but the handling of a PCI
4292 bus and the use of ot 4292 bus and the use of other driver options may interfere
4293 with the sequence. B 4293 with the sequence. By default IORDY is allowed across
4294 all channels. 4294 all channels.
4295 4295
4296 pata_legacy.opti82c46x= [HW,LIBATA] 4296 pata_legacy.opti82c46x= [HW,LIBATA]
4297 Format: <int> 4297 Format: <int>
4298 Set to 1, 2, or 3 for 4298 Set to 1, 2, or 3 for Opti 82c611A on the primary
4299 channel, the secondar 4299 channel, the secondary channel, or both channels
4300 respectively. Disabl 4300 respectively. Disabled by default.
4301 4301
4302 pata_legacy.opti82c611a= [HW,L 4302 pata_legacy.opti82c611a= [HW,LIBATA]
4303 Format: <int> 4303 Format: <int>
4304 Set to 1, 2, or 3 for 4304 Set to 1, 2, or 3 for Opti 82c465MV on the primary
4305 channel, the secondar 4305 channel, the secondary channel, or both channels
4306 respectively. Disabl 4306 respectively. Disabled by default.
4307 4307
4308 pata_legacy.pio_mask= [HW,LIBATA] 4308 pata_legacy.pio_mask= [HW,LIBATA]
4309 Format: <int> 4309 Format: <int>
4310 PIO mode mask for aut 4310 PIO mode mask for autospeed devices. Set individual
4311 bits to allow the use 4311 bits to allow the use of the respective PIO modes.
4312 Bit 0 is for mode 0, 4312 Bit 0 is for mode 0, bit 1 is for mode 1, and so on.
4313 All modes allowed by 4313 All modes allowed by default.
4314 4314
4315 pata_legacy.probe_all= [HW,LIBATA] 4315 pata_legacy.probe_all= [HW,LIBATA]
4316 Format: <int> 4316 Format: <int>
4317 Set to non-zero to pr 4317 Set to non-zero to probe tertiary and further ISA
4318 port ranges on PCI sy 4318 port ranges on PCI systems. Disabled by default.
4319 4319
4320 pata_legacy.probe_mask= [HW,LIBATA] 4320 pata_legacy.probe_mask= [HW,LIBATA]
4321 Format: <int> 4321 Format: <int>
4322 Probe mask for legacy 4322 Probe mask for legacy ISA PATA ports. Depending on
4323 platform configuratio 4323 platform configuration and the use of other driver
4324 options up to 6 legac 4324 options up to 6 legacy ports are supported: 0x1f0,
4325 0x170, 0x1e8, 0x168, 4325 0x170, 0x1e8, 0x168, 0x1e0, 0x160, however probing
4326 of individual ports c 4326 of individual ports can be disabled by setting the
4327 corresponding bits in 4327 corresponding bits in the mask to 1. Bit 0 is for
4328 the first port in the 4328 the first port in the list above (0x1f0), and so on.
4329 By default all suppor 4329 By default all supported ports are probed.
4330 4330
4331 pata_legacy.qdi= [HW,LIBATA] 4331 pata_legacy.qdi= [HW,LIBATA]
4332 Format: <int> 4332 Format: <int>
4333 Set to non-zero to pr 4333 Set to non-zero to probe QDI controllers. By default
4334 set to 1 if CONFIG_PA 4334 set to 1 if CONFIG_PATA_QDI_MODULE, 0 otherwise.
4335 4335
4336 pata_legacy.winbond= [HW,LIBATA] 4336 pata_legacy.winbond= [HW,LIBATA]
4337 Format: <int> 4337 Format: <int>
4338 Set to non-zero to pr 4338 Set to non-zero to probe Winbond controllers. Use
4339 the standard I/O port 4339 the standard I/O port (0x130) if 1, otherwise the
4340 value given is the I/ 4340 value given is the I/O port to use (typically 0x1b0).
4341 By default set to 1 i 4341 By default set to 1 if CONFIG_PATA_WINBOND_VLB_MODULE,
4342 0 otherwise. 4342 0 otherwise.
4343 4343
4344 pata_platform.pio_mask= [HW,LIBATA] 4344 pata_platform.pio_mask= [HW,LIBATA]
4345 Format: <int> 4345 Format: <int>
4346 Supported PIO mode ma 4346 Supported PIO mode mask. Set individual bits to allow
4347 the use of the respec 4347 the use of the respective PIO modes. Bit 0 is for
4348 mode 0, bit 1 is for 4348 mode 0, bit 1 is for mode 1, and so on. Mode 0 only
4349 allowed by default. 4349 allowed by default.
4350 4350
4351 pause_on_oops=<int> 4351 pause_on_oops=<int>
4352 Halt all CPUs after t 4352 Halt all CPUs after the first oops has been printed for
4353 the specified number 4353 the specified number of seconds. This is to be used if
4354 your oopses keep scro 4354 your oopses keep scrolling off the screen.
4355 4355
4356 pcbit= [HW,ISDN] 4356 pcbit= [HW,ISDN]
4357 4357
4358 pci=option[,option...] [PCI,EARLY] v 4358 pci=option[,option...] [PCI,EARLY] various PCI subsystem options.
4359 4359
4360 Some options 4360 Some options herein operate on a specific device
4361 or a set of d 4361 or a set of devices (<pci_dev>). These are
4362 specified in 4362 specified in one of the following formats:
4363 4363
4364 [<domain>:]<b 4364 [<domain>:]<bus>:<dev>.<func>[/<dev>.<func>]*
4365 pci:<vendor>: 4365 pci:<vendor>:<device>[:<subvendor>:<subdevice>]
4366 4366
4367 Note: the fir 4367 Note: the first format specifies a PCI
4368 bus/device/fu 4368 bus/device/function address which may change
4369 if new hardwa 4369 if new hardware is inserted, if motherboard
4370 firmware chan 4370 firmware changes, or due to changes caused
4371 by other kern 4371 by other kernel parameters. If the
4372 domain is lef 4372 domain is left unspecified, it is
4373 taken to be z 4373 taken to be zero. Optionally, a path
4374 to a device t 4374 to a device through multiple device/function
4375 addresses can 4375 addresses can be specified after the base
4376 address (this 4376 address (this is more robust against
4377 renumbering i 4377 renumbering issues). The second format
4378 selects devic 4378 selects devices using IDs from the
4379 configuration 4379 configuration space which may match multiple
4380 devices in th 4380 devices in the system.
4381 4381
4382 earlydump dump PCI conf 4382 earlydump dump PCI config space before the kernel
4383 changes anyth 4383 changes anything
4384 off [X86] don't p 4384 off [X86] don't probe for the PCI bus
4385 bios [X86-32] forc 4385 bios [X86-32] force use of PCI BIOS, don't access
4386 the hardware 4386 the hardware directly. Use this if your machine
4387 has a non-sta 4387 has a non-standard PCI host bridge.
4388 nobios [X86-32] disa 4388 nobios [X86-32] disallow use of PCI BIOS, only direct
4389 hardware acce 4389 hardware access methods are allowed. Use this
4390 if you experi 4390 if you experience crashes upon bootup and you
4391 suspect they 4391 suspect they are caused by the BIOS.
4392 conf1 [X86] Force u 4392 conf1 [X86] Force use of PCI Configuration Access
4393 Mechanism 1 ( 4393 Mechanism 1 (config address in IO port 0xCF8,
4394 data in IO po 4394 data in IO port 0xCFC, both 32-bit).
4395 conf2 [X86] Force u 4395 conf2 [X86] Force use of PCI Configuration Access
4396 Mechanism 2 ( 4396 Mechanism 2 (IO port 0xCF8 is an 8-bit port for
4397 the function, 4397 the function, IO port 0xCFA, also 8-bit, sets
4398 bus number. T 4398 bus number. The config space is then accessed
4399 through ports 4399 through ports 0xC000-0xCFFF).
4400 See http://wi 4400 See http://wiki.osdev.org/PCI for more info
4401 on the config 4401 on the configuration access mechanisms.
4402 noaer [PCIE] If the 4402 noaer [PCIE] If the PCIEAER kernel config parameter is
4403 enabled, this 4403 enabled, this kernel boot option can be used to
4404 disable the u 4404 disable the use of PCIE advanced error reporting.
4405 nodomains [PCI] Disable 4405 nodomains [PCI] Disable support for multiple PCI
4406 root domains 4406 root domains (aka PCI segments, in ACPI-speak).
4407 nommconf [X86] Disable 4407 nommconf [X86] Disable use of MMCONFIG for PCI
4408 Configuration 4408 Configuration
4409 check_enable_amd_mmconf [X86] 4409 check_enable_amd_mmconf [X86] check for and enable
4410 properly conf 4410 properly configured MMIO access to PCI
4411 config space 4411 config space on AMD family 10h CPU
4412 nomsi [MSI] If the 4412 nomsi [MSI] If the PCI_MSI kernel config parameter is
4413 enabled, this 4413 enabled, this kernel boot option can be used to
4414 disable the u 4414 disable the use of MSI interrupts system-wide.
4415 noioapicquirk [APIC] Disabl 4415 noioapicquirk [APIC] Disable all boot interrupt quirks.
4416 Safety option 4416 Safety option to keep boot IRQs enabled. This
4417 should never 4417 should never be necessary.
4418 ioapicreroute [APIC] Enable 4418 ioapicreroute [APIC] Enable rerouting of boot IRQs to the
4419 primary IO-AP 4419 primary IO-APIC for bridges that cannot disable
4420 boot IRQs. Th 4420 boot IRQs. This fixes a source of spurious IRQs
4421 when the syst 4421 when the system masks IRQs.
4422 noioapicreroute [APIC] Disabl 4422 noioapicreroute [APIC] Disable workaround that uses the
4423 boot IRQ equi 4423 boot IRQ equivalent of an IRQ that connects to
4424 a chipset whe 4424 a chipset where boot IRQs cannot be disabled.
4425 The opposite 4425 The opposite of ioapicreroute.
4426 biosirq [X86-32] Use 4426 biosirq [X86-32] Use PCI BIOS calls to get the interrupt
4427 routing table 4427 routing table. These calls are known to be buggy
4428 on several ma 4428 on several machines and they hang the machine
4429 when used, bu 4429 when used, but on other computers it's the only
4430 way to get th 4430 way to get the interrupt routing table. Try
4431 this option i 4431 this option if the kernel is unable to allocate
4432 IRQs or disco 4432 IRQs or discover secondary PCI buses on your
4433 motherboard. 4433 motherboard.
4434 rom [X86] Assign 4434 rom [X86] Assign address space to expansion ROMs.
4435 Use with caut 4435 Use with caution as certain devices share
4436 address decod 4436 address decoders between ROMs and other
4437 resources. 4437 resources.
4438 norom [X86] Do not 4438 norom [X86] Do not assign address space to
4439 expansion ROM 4439 expansion ROMs that do not already have
4440 BIOS assigned 4440 BIOS assigned address ranges.
4441 nobar [X86] Do not 4441 nobar [X86] Do not assign address space to the
4442 BARs that wer 4442 BARs that weren't assigned by the BIOS.
4443 irqmask=0xMMMM [X86] Set a b 4443 irqmask=0xMMMM [X86] Set a bit mask of IRQs allowed to be
4444 assigned auto 4444 assigned automatically to PCI devices. You can
4445 make the kern 4445 make the kernel exclude IRQs of your ISA cards
4446 this way. 4446 this way.
4447 pirqaddr=0xAAAAA [X86] 4447 pirqaddr=0xAAAAA [X86] Specify the physical address
4448 of the PIRQ t 4448 of the PIRQ table (normally generated
4449 by the BIOS) 4449 by the BIOS) if it is outside the
4450 F0000h-100000 4450 F0000h-100000h range.
4451 lastbus=N [X86] Scan al 4451 lastbus=N [X86] Scan all buses thru bus #N. Can be
4452 useful if the 4452 useful if the kernel is unable to find your
4453 secondary bus 4453 secondary buses and you want to tell it
4454 explicitly wh 4454 explicitly which ones they are.
4455 assign-busses [X86] Always 4455 assign-busses [X86] Always assign all PCI bus
4456 numbers ourse 4456 numbers ourselves, overriding
4457 whatever the 4457 whatever the firmware may have done.
4458 usepirqmask [X86] Honor t 4458 usepirqmask [X86] Honor the possible IRQ mask stored
4459 in the BIOS $ 4459 in the BIOS $PIR table. This is needed on
4460 some systems 4460 some systems with broken BIOSes, notably
4461 some HP Pavil 4461 some HP Pavilion N5400 and Omnibook XE3
4462 notebooks. Th 4462 notebooks. This will have no effect if ACPI
4463 IRQ routing i 4463 IRQ routing is enabled.
4464 noacpi [X86] Do not 4464 noacpi [X86] Do not use ACPI for IRQ routing
4465 or for PCI sc 4465 or for PCI scanning.
4466 use_crs [X86] Use PCI 4466 use_crs [X86] Use PCI host bridge window information
4467 from ACPI. O 4467 from ACPI. On BIOSes from 2008 or later, this
4468 is enabled by 4468 is enabled by default. If you need to use this,
4469 please report 4469 please report a bug.
4470 nocrs [X86] Ignore 4470 nocrs [X86] Ignore PCI host bridge windows from ACPI.
4471 If you need t 4471 If you need to use this, please report a bug.
4472 use_e820 [X86] Use E82 4472 use_e820 [X86] Use E820 reservations to exclude parts of
4473 PCI host brid 4473 PCI host bridge windows. This is a workaround
4474 for BIOS defe 4474 for BIOS defects in host bridge _CRS methods.
4475 If you need t 4475 If you need to use this, please report a bug to
4476 <linux-pci@vge 4476 <linux-pci@vger.kernel.org>.
4477 no_e820 [X86] Ignore 4477 no_e820 [X86] Ignore E820 reservations for PCI host
4478 bridge window 4478 bridge windows. This is the default on modern
4479 hardware. If 4479 hardware. If you need to use this, please report
4480 a bug to <linu 4480 a bug to <linux-pci@vger.kernel.org>.
4481 routeirq Do IRQ routin 4481 routeirq Do IRQ routing for all PCI devices.
4482 This is norma 4482 This is normally done in pci_enable_device(),
4483 so this optio 4483 so this option is a temporary workaround
4484 for broken dr 4484 for broken drivers that don't call it.
4485 skip_isa_align [X86] do not 4485 skip_isa_align [X86] do not align io start addr, so can
4486 handle more p 4486 handle more pci cards
4487 noearly [X86] Don't d 4487 noearly [X86] Don't do any early type 1 scanning.
4488 This might he 4488 This might help on some broken boards which
4489 machine check 4489 machine check when some devices' config space
4490 is read. But 4490 is read. But various workarounds are disabled
4491 and some IOMM 4491 and some IOMMU drivers will not work.
4492 bfsort Sort PCI devi 4492 bfsort Sort PCI devices into breadth-first order.
4493 This sorting 4493 This sorting is done to get a device
4494 order compati 4494 order compatible with older (<= 2.4) kernels.
4495 nobfsort Don't sort PC 4495 nobfsort Don't sort PCI devices into breadth-first order.
4496 pcie_bus_tune_off Disab 4496 pcie_bus_tune_off Disable PCIe MPS (Max Payload Size)
4497 tuning and us 4497 tuning and use the BIOS-configured MPS defaults.
4498 pcie_bus_safe Set every dev 4498 pcie_bus_safe Set every device's MPS to the largest value
4499 supported by 4499 supported by all devices below the root complex.
4500 pcie_bus_perf Set device MP 4500 pcie_bus_perf Set device MPS to the largest allowable MPS
4501 based on its 4501 based on its parent bus. Also set MRRS (Max
4502 Read Request 4502 Read Request Size) to the largest supported
4503 value (no lar 4503 value (no larger than the MPS that the device
4504 or bus can su 4504 or bus can support) for best performance.
4505 pcie_bus_peer2peer Set e 4505 pcie_bus_peer2peer Set every device's MPS to 128B, which
4506 every device 4506 every device is guaranteed to support. This
4507 configuration 4507 configuration allows peer-to-peer DMA between
4508 any pair of d 4508 any pair of devices, possibly at the cost of
4509 reduced perfo 4509 reduced performance. This also guarantees
4510 that hot-adde 4510 that hot-added devices will work.
4511 cbiosize=nn[KMG] The f 4511 cbiosize=nn[KMG] The fixed amount of bus space which is
4512 reserved for 4512 reserved for the CardBus bridge's IO window.
4513 The default v 4513 The default value is 256 bytes.
4514 cbmemsize=nn[KMG] The f 4514 cbmemsize=nn[KMG] The fixed amount of bus space which is
4515 reserved for 4515 reserved for the CardBus bridge's memory
4516 window. The d 4516 window. The default value is 64 megabytes.
4517 resource_alignment= 4517 resource_alignment=
4518 Format: 4518 Format:
4519 [<order of ali 4519 [<order of align>@]<pci_dev>[; ...]
4520 Specifies ali 4520 Specifies alignment and device to reassign
4521 aligned memor 4521 aligned memory resources. How to
4522 specify the d 4522 specify the device is described above.
4523 If <order of 4523 If <order of align> is not specified,
4524 PAGE_SIZE is 4524 PAGE_SIZE is used as alignment.
4525 A PCI-PCI bri 4525 A PCI-PCI bridge can be specified if resource
4526 windows need 4526 windows need to be expanded.
4527 To specify th 4527 To specify the alignment for several
4528 instances of 4528 instances of a device, the PCI vendor,
4529 device, subve 4529 device, subvendor, and subdevice may be
4530 specified, e. 4530 specified, e.g., 12@pci:8086:9c22:103c:198f
4531 for 4096-byte 4531 for 4096-byte alignment.
4532 ecrc= Enable/disabl 4532 ecrc= Enable/disable PCIe ECRC (transaction layer
4533 end-to-end CR 4533 end-to-end CRC checking). Only effective if
4534 OS has native 4534 OS has native AER control (either granted by
4535 ACPI _OSC or 4535 ACPI _OSC or forced via "pcie_ports=native")
4536 bios: Use BIO 4536 bios: Use BIOS/firmware settings. This is the
4537 the default. 4537 the default.
4538 off: Turn ECR 4538 off: Turn ECRC off
4539 on: Turn ECRC 4539 on: Turn ECRC on.
4540 hpiosize=nn[KMG] The f 4540 hpiosize=nn[KMG] The fixed amount of bus space which is
4541 reserved for 4541 reserved for hotplug bridge's IO window.
4542 Default size 4542 Default size is 256 bytes.
4543 hpmmiosize=nn[KMG] The f 4543 hpmmiosize=nn[KMG] The fixed amount of bus space which is
4544 reserved for 4544 reserved for hotplug bridge's MMIO window.
4545 Default size 4545 Default size is 2 megabytes.
4546 hpmmioprefsize=nn[KMG] The f 4546 hpmmioprefsize=nn[KMG] The fixed amount of bus space which is
4547 reserved for 4547 reserved for hotplug bridge's MMIO_PREF window.
4548 Default size 4548 Default size is 2 megabytes.
4549 hpmemsize=nn[KMG] The f 4549 hpmemsize=nn[KMG] The fixed amount of bus space which is
4550 reserved for 4550 reserved for hotplug bridge's MMIO and
4551 MMIO_PREF win 4551 MMIO_PREF window.
4552 Default size 4552 Default size is 2 megabytes.
4553 hpbussize=nn The minimum a 4553 hpbussize=nn The minimum amount of additional bus numbers
4554 reserved for 4554 reserved for buses below a hotplug bridge.
4555 Default is 1. 4555 Default is 1.
4556 realloc= Enable/disabl 4556 realloc= Enable/disable reallocating PCI bridge resources
4557 if allocation 4557 if allocations done by BIOS are too small to
4558 accommodate r 4558 accommodate resources required by all child
4559 devices. 4559 devices.
4560 off: Turn rea 4560 off: Turn realloc off
4561 on: Turn real 4561 on: Turn realloc on
4562 realloc same as reall 4562 realloc same as realloc=on
4563 noari do not use PC 4563 noari do not use PCIe ARI.
4564 noats [PCIE, Intel- 4564 noats [PCIE, Intel-IOMMU, AMD-IOMMU]
4565 do not use PC 4565 do not use PCIe ATS (and IOMMU device IOTLB).
4566 pcie_scan_all Scan all poss 4566 pcie_scan_all Scan all possible PCIe devices. Otherwise we
4567 only look for 4567 only look for one device below a PCIe downstream
4568 port. 4568 port.
4569 big_root_window Try to add a 4569 big_root_window Try to add a big 64bit memory window to the PCIe
4570 root complex 4570 root complex on AMD CPUs. Some GFX hardware
4571 can resize a 4571 can resize a BAR to allow access to all VRAM.
4572 Adding the wi 4572 Adding the window is slightly risky (it may
4573 conflict with 4573 conflict with unreported devices), so this
4574 taints the ke 4574 taints the kernel.
4575 disable_acs_redir=<pci_dev>[; 4575 disable_acs_redir=<pci_dev>[; ...]
4576 Specify one o 4576 Specify one or more PCI devices (in the format
4577 specified abo 4577 specified above) separated by semicolons.
4578 Each device s 4578 Each device specified will have the PCI ACS
4579 redirect capa 4579 redirect capabilities forced off which will
4580 allow P2P tra 4580 allow P2P traffic between devices through
4581 bridges witho 4581 bridges without forcing it upstream. Note:
4582 this removes 4582 this removes isolation between devices and
4583 may put more 4583 may put more devices in an IOMMU group.
4584 config_acs= 4584 config_acs=
4585 Format: 4585 Format:
4586 <ACS flags>@<p 4586 <ACS flags>@<pci_dev>[; ...]
4587 Specify one o 4587 Specify one or more PCI devices (in the format
4588 specified abo 4588 specified above) optionally prepended with flags
4589 and separated 4589 and separated by semicolons. The respective
4590 capabilities 4590 capabilities will be enabled, disabled or
4591 unchanged bas 4591 unchanged based on what is specified in
4592 flags. 4592 flags.
4593 4593
4594 ACS Flags is 4594 ACS Flags is defined as follows:
4595 bit-0 : ACS 4595 bit-0 : ACS Source Validation
4596 bit-1 : ACS 4596 bit-1 : ACS Translation Blocking
4597 bit-2 : ACS 4597 bit-2 : ACS P2P Request Redirect
4598 bit-3 : ACS 4598 bit-3 : ACS P2P Completion Redirect
4599 bit-4 : ACS 4599 bit-4 : ACS Upstream Forwarding
4600 bit-5 : ACS 4600 bit-5 : ACS P2P Egress Control
4601 bit-6 : ACS 4601 bit-6 : ACS Direct Translated P2P
4602 Each bit can 4602 Each bit can be marked as:
4603 '0' – for 4603 '0' – force disabled
4604 '1' – for 4604 '1' – force enabled
4605 'x' – unc 4605 'x' – unchanged
4606 For example, 4606 For example,
4607 pci=config_ 4607 pci=config_acs=10x
4608 would configu 4608 would configure all devices that support
4609 ACS to enable 4609 ACS to enable P2P Request Redirect, disable
4610 Translation B 4610 Translation Blocking, and leave Source
4611 Validation un 4611 Validation unchanged from whatever power-up
4612 or firmware s 4612 or firmware set it to.
4613 4613
4614 Note: this ma 4614 Note: this may remove isolation between devices
4615 and may put m 4615 and may put more devices in an IOMMU group.
4616 force_floating [S390] Force 4616 force_floating [S390] Force usage of floating interrupts.
4617 nomio [S390] Do not 4617 nomio [S390] Do not use MIO instructions.
4618 norid [S390] ignore 4618 norid [S390] ignore the RID field and force use of
4619 one PCI domai 4619 one PCI domain per PCI function
4620 4620
4621 pcie_aspm= [PCIE] Forcibly enabl 4621 pcie_aspm= [PCIE] Forcibly enable or ignore PCIe Active State Power
4622 Management. 4622 Management.
4623 off Don't touch ASPM conf 4623 off Don't touch ASPM configuration at all. Leave any
4624 configuration done by 4624 configuration done by firmware unchanged.
4625 force Enable ASPM even on d 4625 force Enable ASPM even on devices that claim not to support it.
4626 WARNING: Forcing ASPM 4626 WARNING: Forcing ASPM on may cause system lockups.
4627 4627
4628 pcie_ports= [PCIE] PCIe port serv 4628 pcie_ports= [PCIE] PCIe port services handling:
4629 native Use native PCIe servi 4629 native Use native PCIe services (PME, AER, DPC, PCIe hotplug)
4630 even if the platform 4630 even if the platform doesn't give the OS permission to
4631 use them. This may c 4631 use them. This may cause conflicts if the platform
4632 also tries to use the 4632 also tries to use these services.
4633 dpc-native Use native PC 4633 dpc-native Use native PCIe service for DPC only. May
4634 cause conflic 4634 cause conflicts if firmware uses AER or DPC.
4635 compat Disable native PCIe s 4635 compat Disable native PCIe services (PME, AER, DPC, PCIe
4636 hotplug). 4636 hotplug).
4637 4637
4638 pcie_port_pm= [PCIE] PCIe port powe 4638 pcie_port_pm= [PCIE] PCIe port power management handling:
4639 off Disable power managem 4639 off Disable power management of all PCIe ports
4640 force Forcibly enable power 4640 force Forcibly enable power management of all PCIe ports
4641 4641
4642 pcie_pme= [PCIE,PM] Native PCIe 4642 pcie_pme= [PCIE,PM] Native PCIe PME signaling options:
4643 nomsi Do not use MSI for na 4643 nomsi Do not use MSI for native PCIe PME signaling (this makes
4644 all PCIe root ports u 4644 all PCIe root ports use INTx for all services).
4645 4645
4646 pcmv= [HW,PCMCIA] BadgePAD 4646 pcmv= [HW,PCMCIA] BadgePAD 4
4647 4647
4648 pd_ignore_unused 4648 pd_ignore_unused
4649 [PM] 4649 [PM]
4650 Keep all power-domain 4650 Keep all power-domains already enabled by bootloader on,
4651 even if no driver has 4651 even if no driver has claimed them. This is useful
4652 for debug and develop 4652 for debug and development, but should not be
4653 needed on a platform 4653 needed on a platform with proper driver support.
4654 4654
4655 pdcchassis= [PARISC,HW] Disable/E 4655 pdcchassis= [PARISC,HW] Disable/Enable PDC Chassis Status codes at
4656 boot time. 4656 boot time.
4657 Format: { 0 | 1 } 4657 Format: { 0 | 1 }
4658 See arch/parisc/kerne 4658 See arch/parisc/kernel/pdc_chassis.c
4659 4659
4660 percpu_alloc= [MM,EARLY] 4660 percpu_alloc= [MM,EARLY]
4661 Select which percpu f 4661 Select which percpu first chunk allocator to use.
4662 Currently supported v 4662 Currently supported values are "embed" and "page".
4663 Archs may support sub 4663 Archs may support subset or none of the selections.
4664 See comments in mm/pe 4664 See comments in mm/percpu.c for details on each
4665 allocator. This para 4665 allocator. This parameter is primarily for debugging
4666 and performance compa 4666 and performance comparison.
4667 4667
4668 pirq= [SMP,APIC] Manual mp- 4668 pirq= [SMP,APIC] Manual mp-table setup
4669 See Documentation/arc 4669 See Documentation/arch/x86/i386/IO-APIC.rst.
4670 4670
4671 plip= [PPT,NET] Parallel po 4671 plip= [PPT,NET] Parallel port network link
4672 Format: { parport<nr> 4672 Format: { parport<nr> | timid | 0 }
4673 See also Documentatio 4673 See also Documentation/admin-guide/parport.rst.
4674 4674
4675 pmtmr= [X86] Manual setup of 4675 pmtmr= [X86] Manual setup of pmtmr I/O Port.
4676 Override pmtimer IOPo 4676 Override pmtimer IOPort with a hex value.
4677 e.g. pmtmr=0x508 4677 e.g. pmtmr=0x508
4678 4678
4679 pmu_override= [PPC] Override the PM 4679 pmu_override= [PPC] Override the PMU.
4680 This option takes ove 4680 This option takes over the PMU facility, so it is no
4681 longer usable by perf 4681 longer usable by perf. Setting this option starts the
4682 PMU counters by setti 4682 PMU counters by setting MMCR0 to 0 (the FC bit is
4683 cleared). If a number 4683 cleared). If a number is given, then MMCR1 is set to
4684 that number, otherwis 4684 that number, otherwise (e.g., 'pmu_override=on'), MMCR1
4685 remains 0. 4685 remains 0.
4686 4686
4687 pm_debug_messages [SUSPEND,KNL] 4687 pm_debug_messages [SUSPEND,KNL]
4688 Enable suspend/resume 4688 Enable suspend/resume debug messages during boot up.
4689 4689
4690 pnp.debug=1 [PNP] 4690 pnp.debug=1 [PNP]
4691 Enable PNP debug mess 4691 Enable PNP debug messages (depends on the
4692 CONFIG_PNP_DEBUG_MESS 4692 CONFIG_PNP_DEBUG_MESSAGES option). Change at run-time
4693 via /sys/module/pnp/p 4693 via /sys/module/pnp/parameters/debug. We always show
4694 current resource usag 4694 current resource usage; turning this on also shows
4695 possible settings and 4695 possible settings and some assignment information.
4696 4696
4697 pnpacpi= [ACPI] 4697 pnpacpi= [ACPI]
4698 { off } 4698 { off }
4699 4699
4700 pnpbios= [ISAPNP] 4700 pnpbios= [ISAPNP]
4701 { on | off | curr | r 4701 { on | off | curr | res | no-curr | no-res }
4702 4702
4703 pnp_reserve_irq= 4703 pnp_reserve_irq=
4704 [ISAPNP] Exclude IRQs 4704 [ISAPNP] Exclude IRQs for the autoconfiguration
4705 4705
4706 pnp_reserve_dma= 4706 pnp_reserve_dma=
4707 [ISAPNP] Exclude DMAs 4707 [ISAPNP] Exclude DMAs for the autoconfiguration
4708 4708
4709 pnp_reserve_io= [ISAPNP] Exclude I/O 4709 pnp_reserve_io= [ISAPNP] Exclude I/O ports for the autoconfiguration
4710 Ranges are in pairs ( 4710 Ranges are in pairs (I/O port base and size).
4711 4711
4712 pnp_reserve_mem= 4712 pnp_reserve_mem=
4713 [ISAPNP] Exclude memo 4713 [ISAPNP] Exclude memory regions for the
4714 autoconfiguration. 4714 autoconfiguration.
4715 Ranges are in pairs ( 4715 Ranges are in pairs (memory base and size).
4716 4716
4717 ports= [IP_VS_FTP] IPVS ftp 4717 ports= [IP_VS_FTP] IPVS ftp helper module
4718 Default is 21. 4718 Default is 21.
4719 Up to 8 (IP_VS_APP_MA 4719 Up to 8 (IP_VS_APP_MAX_PORTS) ports
4720 may be specified. 4720 may be specified.
4721 Format: <port>,<port> 4721 Format: <port>,<port>....
4722 4722
4723 possible_cpus= [SMP,S390,X86] 4723 possible_cpus= [SMP,S390,X86]
4724 Format: <unsigned int 4724 Format: <unsigned int>
4725 Set the number of pos 4725 Set the number of possible CPUs, overriding the
4726 regular discovery mec 4726 regular discovery mechanisms (such as ACPI/FW, etc).
4727 4727
4728 powersave=off [PPC] This option dis 4728 powersave=off [PPC] This option disables power saving features.
4729 It specifically disab 4729 It specifically disables cpuidle and sets the
4730 platform machine desc 4730 platform machine description specific power_save
4731 function to NULL. On 4731 function to NULL. On Idle the CPU just reduces
4732 execution priority. 4732 execution priority.
4733 4733
4734 ppc_strict_facility_enable 4734 ppc_strict_facility_enable
4735 [PPC,ENABLE] This opt 4735 [PPC,ENABLE] This option catches any kernel floating point,
4736 Altivec, VSX and SPE 4736 Altivec, VSX and SPE outside of regions specifically
4737 allowed (eg kernel_en 4737 allowed (eg kernel_enable_fpu()/kernel_disable_fpu()).
4738 There is some perform 4738 There is some performance impact when enabling this.
4739 4739
4740 ppc_tm= [PPC,EARLY] 4740 ppc_tm= [PPC,EARLY]
4741 Format: {"off"} 4741 Format: {"off"}
4742 Disable Hardware Tran 4742 Disable Hardware Transactional Memory
4743 4743
4744 preempt= [KNL] 4744 preempt= [KNL]
4745 Select preemption mod 4745 Select preemption mode if you have CONFIG_PREEMPT_DYNAMIC
4746 none - Limited to con 4746 none - Limited to cond_resched() calls
4747 voluntary - Limited t 4747 voluntary - Limited to cond_resched() and might_sleep() calls
4748 full - Any section th 4748 full - Any section that isn't explicitly preempt disabled
4749 can be preempt 4749 can be preempted anytime. Tasks will also yield
4750 contended spin 4750 contended spinlocks (if the critical section isn't
4751 explicitly pre 4751 explicitly preempt disabled beyond the lock itself).
4752 4752
4753 print-fatal-signals= 4753 print-fatal-signals=
4754 [KNL] debug: print fa 4754 [KNL] debug: print fatal signals
4755 4755
4756 If enabled, warn abou 4756 If enabled, warn about various signal handling
4757 related application a 4757 related application anomalies: too many signals,
4758 too many POSIX.1 time 4758 too many POSIX.1 timers, fatal signals causing a
4759 coredump - etc. 4759 coredump - etc.
4760 4760
4761 If you hit the warnin 4761 If you hit the warning due to signal overflow,
4762 you might want to try 4762 you might want to try "ulimit -i unlimited".
4763 4763
4764 default: off. 4764 default: off.
4765 4765
4766 printk.always_kmsg_dump= 4766 printk.always_kmsg_dump=
4767 Trigger kmsg_dump for 4767 Trigger kmsg_dump for cases other than kernel oops or
4768 panics 4768 panics
4769 Format: <bool> (1/Y/ 4769 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4770 default: disabled 4770 default: disabled
4771 4771
4772 printk.console_no_auto_verbose= 4772 printk.console_no_auto_verbose=
4773 Disable console logle 4773 Disable console loglevel raise on oops, panic
4774 or lockdep-detected i 4774 or lockdep-detected issues (only if lock debug is on).
4775 With an exception to 4775 With an exception to setups with low baudrate on
4776 serial console, keepi 4776 serial console, keeping this 0 is a good choice
4777 in order to provide m 4777 in order to provide more debug information.
4778 Format: <bool> 4778 Format: <bool>
4779 default: 0 (auto_verb 4779 default: 0 (auto_verbose is enabled)
4780 4780
4781 printk.devkmsg={on,off,ratelimit} 4781 printk.devkmsg={on,off,ratelimit}
4782 Control writing to /d 4782 Control writing to /dev/kmsg.
4783 on - unlimited loggin 4783 on - unlimited logging to /dev/kmsg from userspace
4784 off - logging to /dev 4784 off - logging to /dev/kmsg disabled
4785 ratelimit - ratelimit 4785 ratelimit - ratelimit the logging
4786 Default: ratelimit 4786 Default: ratelimit
4787 4787
4788 printk.time= Show timing data pref 4788 printk.time= Show timing data prefixed to each printk message line
4789 Format: <bool> (1/Y/ 4789 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4790 4790
4791 proc_mem.force_override= [KNL] 4791 proc_mem.force_override= [KNL]
4792 Format: {always | ptr 4792 Format: {always | ptrace | never}
4793 Traditionally /proc/p 4793 Traditionally /proc/pid/mem allows memory permissions to be
4794 overridden without re 4794 overridden without restrictions. This option may be set to
4795 restrict that. Can be 4795 restrict that. Can be one of:
4796 - 'always': tradition 4796 - 'always': traditional behavior always allows mem overrides.
4797 - 'ptrace': only allo 4797 - 'ptrace': only allow mem overrides for active ptracers.
4798 - 'never': never all 4798 - 'never': never allow mem overrides.
4799 If not specified, def 4799 If not specified, default is the CONFIG_PROC_MEM_* choice.
4800 4800
4801 processor.max_cstate= [HW,ACPI] 4801 processor.max_cstate= [HW,ACPI]
4802 Limit processor to ma 4802 Limit processor to maximum C-state
4803 max_cstate=9 override 4803 max_cstate=9 overrides any DMI blacklist limit.
4804 4804
4805 processor.nocst [HW,ACPI] 4805 processor.nocst [HW,ACPI]
4806 Ignore the _CST metho 4806 Ignore the _CST method to determine C-states,
4807 instead using the leg 4807 instead using the legacy FADT method
4808 4808
4809 profile= [KNL] Enable kernel p 4809 profile= [KNL] Enable kernel profiling via /proc/profile
4810 Format: [<profiletype 4810 Format: [<profiletype>,]<number>
4811 Param: <profiletype>: 4811 Param: <profiletype>: "schedule" or "kvm"
4812 [defaults to 4812 [defaults to kernel profiling]
4813 Param: "schedule" - p 4813 Param: "schedule" - profile schedule points.
4814 Param: "kvm" - profil 4814 Param: "kvm" - profile VM exits.
4815 Param: <number> - ste 4815 Param: <number> - step/bucket size as a power of 2 for
4816 statistical t 4816 statistical time based profiling.
4817 4817
4818 prompt_ramdisk= [RAM] [Deprecated] 4818 prompt_ramdisk= [RAM] [Deprecated]
4819 4819
4820 prot_virt= [S390] enable hosting 4820 prot_virt= [S390] enable hosting protected virtual machines
4821 isolated from the hyp 4821 isolated from the hypervisor (if hardware supports
4822 that). If enabled, th 4822 that). If enabled, the default kernel base address
4823 might be overridden e 4823 might be overridden even when Kernel Address Space
4824 Layout Randomization 4824 Layout Randomization is disabled.
4825 Format: <bool> 4825 Format: <bool>
4826 4826
4827 psi= [KNL] Enable or disab 4827 psi= [KNL] Enable or disable pressure stall information
4828 tracking. 4828 tracking.
4829 Format: <bool> 4829 Format: <bool>
4830 4830
4831 psmouse.proto= [HW,MOUSE] Highest PS 4831 psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
4832 probe for; one of (ba 4832 probe for; one of (bare|imps|exps|lifebook|any).
4833 psmouse.rate= [HW,MOUSE] Set desire 4833 psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
4834 per second. 4834 per second.
4835 psmouse.resetafter= [HW,MOUSE] 4835 psmouse.resetafter= [HW,MOUSE]
4836 Try to reset the devi 4836 Try to reset the device after so many bad packets
4837 (0 = never). 4837 (0 = never).
4838 psmouse.resolution= 4838 psmouse.resolution=
4839 [HW,MOUSE] Set desire 4839 [HW,MOUSE] Set desired mouse resolution, in dpi.
4840 psmouse.smartscroll= 4840 psmouse.smartscroll=
4841 [HW,MOUSE] Controls L 4841 [HW,MOUSE] Controls Logitech smartscroll autorepeat.
4842 0 = disabled, 1 = ena 4842 0 = disabled, 1 = enabled (default).
4843 4843
4844 pstore.backend= Specify the name of t 4844 pstore.backend= Specify the name of the pstore backend to use
4845 4845
4846 pti= [X86-64] Control Page 4846 pti= [X86-64] Control Page Table Isolation of user and
4847 kernel address spaces 4847 kernel address spaces. Disabling this feature
4848 removes hardening, bu 4848 removes hardening, but improves performance of
4849 system calls and inte 4849 system calls and interrupts.
4850 4850
4851 on - unconditionall 4851 on - unconditionally enable
4852 off - unconditionall 4852 off - unconditionally disable
4853 auto - kernel detects 4853 auto - kernel detects whether your CPU model is
4854 vulnerable to 4854 vulnerable to issues that PTI mitigates
4855 4855
4856 Not specifying this o 4856 Not specifying this option is equivalent to pti=auto.
4857 4857
4858 pty.legacy_count= 4858 pty.legacy_count=
4859 [KNL] Number of legac 4859 [KNL] Number of legacy pty's. Overwrites compiled-in
4860 default number. 4860 default number.
4861 4861
4862 quiet [KNL,EARLY] Disable m 4862 quiet [KNL,EARLY] Disable most log messages
4863 4863
4864 r128= [HW,DRM] 4864 r128= [HW,DRM]
4865 4865
4866 radix_hcall_invalidate=on [PPC/PSERI 4866 radix_hcall_invalidate=on [PPC/PSERIES]
4867 Disable RADIX GTSE fe 4867 Disable RADIX GTSE feature and use hcall for TLB
4868 invalidate. 4868 invalidate.
4869 4869
4870 raid= [HW,RAID] 4870 raid= [HW,RAID]
4871 See Documentation/adm 4871 See Documentation/admin-guide/md.rst.
4872 4872
4873 ramdisk_size= [RAM] Sizes of RAM di 4873 ramdisk_size= [RAM] Sizes of RAM disks in kilobytes
4874 See Documentation/adm 4874 See Documentation/admin-guide/blockdev/ramdisk.rst.
4875 4875
4876 ramdisk_start= [RAM] RAM disk image 4876 ramdisk_start= [RAM] RAM disk image start address
4877 4877
4878 random.trust_cpu=off 4878 random.trust_cpu=off
4879 [KNL,EARLY] Disable t 4879 [KNL,EARLY] Disable trusting the use of the CPU's
4880 random number generat 4880 random number generator (if available) to
4881 initialize the kernel 4881 initialize the kernel's RNG.
4882 4882
4883 random.trust_bootloader=off 4883 random.trust_bootloader=off
4884 [KNL,EARLY] Disable t 4884 [KNL,EARLY] Disable trusting the use of the a seed
4885 passed by the bootloa 4885 passed by the bootloader (if available) to
4886 initialize the kernel 4886 initialize the kernel's RNG.
4887 4887
4888 randomize_kstack_offset= 4888 randomize_kstack_offset=
4889 [KNL,EARLY] Enable or 4889 [KNL,EARLY] Enable or disable kernel stack offset
4890 randomization, which 4890 randomization, which provides roughly 5 bits of
4891 entropy, frustrating 4891 entropy, frustrating memory corruption attacks
4892 that depend on stack 4892 that depend on stack address determinism or
4893 cross-syscall address 4893 cross-syscall address exposures. This is only
4894 available on architec 4894 available on architectures that have defined
4895 CONFIG_HAVE_ARCH_RAND 4895 CONFIG_HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET.
4896 Format: <bool> (1/Y/ 4896 Format: <bool> (1/Y/y=enable, 0/N/n=disable)
4897 Default is CONFIG_RAN 4897 Default is CONFIG_RANDOMIZE_KSTACK_OFFSET_DEFAULT.
4898 4898
4899 ras=option[,option,...] [KNL] RAS-spe 4899 ras=option[,option,...] [KNL] RAS-specific options
4900 4900
4901 cec_disable [X86] 4901 cec_disable [X86]
4902 Disable the C 4902 Disable the Correctable Errors Collector,
4903 see CONFIG_RA 4903 see CONFIG_RAS_CEC help text.
4904 4904
4905 rcu_nocbs[=cpu-list] 4905 rcu_nocbs[=cpu-list]
4906 [KNL] The optional ar 4906 [KNL] The optional argument is a cpu list,
4907 as described above. 4907 as described above.
4908 4908
4909 In kernels built with 4909 In kernels built with CONFIG_RCU_NOCB_CPU=y,
4910 enable the no-callbac 4910 enable the no-callback CPU mode, which prevents
4911 such CPUs' callbacks 4911 such CPUs' callbacks from being invoked in
4912 softirq context. Inv 4912 softirq context. Invocation of such CPUs' RCU
4913 callbacks will instea 4913 callbacks will instead be offloaded to "rcuox/N"
4914 kthreads created for 4914 kthreads created for that purpose, where "x" is
4915 "p" for RCU-preempt, 4915 "p" for RCU-preempt, "s" for RCU-sched, and "g"
4916 for the kthreads that 4916 for the kthreads that mediate grace periods; and
4917 "N" is the CPU number 4917 "N" is the CPU number. This reduces OS jitter on
4918 the offloaded CPUs, w 4918 the offloaded CPUs, which can be useful for HPC
4919 and real-time workloa 4919 and real-time workloads. It can also improve
4920 energy efficiency for 4920 energy efficiency for asymmetric multiprocessors.
4921 4921
4922 If a cpulist is passe 4922 If a cpulist is passed as an argument, the specified
4923 list of CPUs is set t 4923 list of CPUs is set to no-callback mode from boot.
4924 4924
4925 Otherwise, if the '=' 4925 Otherwise, if the '=' sign and the cpulist
4926 arguments are omitted 4926 arguments are omitted, no CPU will be set to
4927 no-callback mode from 4927 no-callback mode from boot but the mode may be
4928 toggled at runtime vi 4928 toggled at runtime via cpusets.
4929 4929
4930 Note that this argume 4930 Note that this argument takes precedence over
4931 the CONFIG_RCU_NOCB_C 4931 the CONFIG_RCU_NOCB_CPU_DEFAULT_ALL option.
4932 4932
4933 rcu_nocb_poll [KNL] 4933 rcu_nocb_poll [KNL]
4934 Rather than requiring 4934 Rather than requiring that offloaded CPUs
4935 (specified by rcu_noc 4935 (specified by rcu_nocbs= above) explicitly
4936 awaken the correspond 4936 awaken the corresponding "rcuoN" kthreads,
4937 make these kthreads p 4937 make these kthreads poll for callbacks.
4938 This improves the rea 4938 This improves the real-time response for the
4939 offloaded CPUs by rel 4939 offloaded CPUs by relieving them of the need to
4940 wake up the correspon 4940 wake up the corresponding kthread, but degrades
4941 energy efficiency by 4941 energy efficiency by requiring that the kthreads
4942 periodically wake up 4942 periodically wake up to do the polling.
4943 4943
4944 rcutree.blimit= [KNL] 4944 rcutree.blimit= [KNL]
4945 Set maximum number of 4945 Set maximum number of finished RCU callbacks to
4946 process in one batch. 4946 process in one batch.
4947 4947
4948 rcutree.do_rcu_barrier= [KNL] 4948 rcutree.do_rcu_barrier= [KNL]
4949 Request a call to rcu 4949 Request a call to rcu_barrier(). This is
4950 throttled so that use 4950 throttled so that userspace tests can safely
4951 hammer on the sysfs v 4951 hammer on the sysfs variable if they so choose.
4952 If triggered before t 4952 If triggered before the RCU grace-period machinery
4953 is fully active, this 4953 is fully active, this will error out with EAGAIN.
4954 4954
4955 rcutree.dump_tree= [KNL] 4955 rcutree.dump_tree= [KNL]
4956 Dump the structure of 4956 Dump the structure of the rcu_node combining tree
4957 out at early boot. T 4957 out at early boot. This is used for diagnostic
4958 purposes, to verify c 4958 purposes, to verify correct tree setup.
4959 4959
4960 rcutree.gp_cleanup_delay= [KNL] 4960 rcutree.gp_cleanup_delay= [KNL]
4961 Set the number of jif 4961 Set the number of jiffies to delay each step of
4962 RCU grace-period clea 4962 RCU grace-period cleanup.
4963 4963
4964 rcutree.gp_init_delay= [KNL] 4964 rcutree.gp_init_delay= [KNL]
4965 Set the number of jif 4965 Set the number of jiffies to delay each step of
4966 RCU grace-period init 4966 RCU grace-period initialization.
4967 4967
4968 rcutree.gp_preinit_delay= [KNL] 4968 rcutree.gp_preinit_delay= [KNL]
4969 Set the number of jif 4969 Set the number of jiffies to delay each step of
4970 RCU grace-period pre- 4970 RCU grace-period pre-initialization, that is,
4971 the propagation of re 4971 the propagation of recent CPU-hotplug changes up
4972 the rcu_node combinin 4972 the rcu_node combining tree.
4973 4973
4974 rcutree.jiffies_till_first_fqs= [KNL] 4974 rcutree.jiffies_till_first_fqs= [KNL]
4975 Set delay from grace- 4975 Set delay from grace-period initialization to
4976 first attempt to forc 4976 first attempt to force quiescent states.
4977 Units are jiffies, mi 4977 Units are jiffies, minimum value is zero,
4978 and maximum value is 4978 and maximum value is HZ.
4979 4979
4980 rcutree.jiffies_till_next_fqs= [KNL] 4980 rcutree.jiffies_till_next_fqs= [KNL]
4981 Set delay between sub 4981 Set delay between subsequent attempts to force
4982 quiescent states. Un 4982 quiescent states. Units are jiffies, minimum
4983 value is one, and max 4983 value is one, and maximum value is HZ.
4984 4984
4985 rcutree.jiffies_till_sched_qs= [KNL] 4985 rcutree.jiffies_till_sched_qs= [KNL]
4986 Set required age in j 4986 Set required age in jiffies for a
4987 given grace period be 4987 given grace period before RCU starts
4988 soliciting quiescent- 4988 soliciting quiescent-state help from
4989 rcu_note_context_swit 4989 rcu_note_context_switch() and cond_resched().
4990 If not specified, the 4990 If not specified, the kernel will calculate
4991 a value based on the 4991 a value based on the most recent settings
4992 of rcutree.jiffies_ti 4992 of rcutree.jiffies_till_first_fqs
4993 and rcutree.jiffies_t 4993 and rcutree.jiffies_till_next_fqs.
4994 This calculated value 4994 This calculated value may be viewed in
4995 rcutree.jiffies_to_sc 4995 rcutree.jiffies_to_sched_qs. Any attempt to set
4996 rcutree.jiffies_to_sc 4996 rcutree.jiffies_to_sched_qs will be cheerfully
4997 overwritten. 4997 overwritten.
4998 4998
4999 rcutree.kthread_prio= [KNL,BOOT] 4999 rcutree.kthread_prio= [KNL,BOOT]
5000 Set the SCHED_FIFO pr 5000 Set the SCHED_FIFO priority of the RCU per-CPU
5001 kthreads (rcuc/N). Th 5001 kthreads (rcuc/N). This value is also used for
5002 the priority of the R 5002 the priority of the RCU boost threads (rcub/N)
5003 and for the RCU grace 5003 and for the RCU grace-period kthreads (rcu_bh,
5004 rcu_preempt, and rcu_ 5004 rcu_preempt, and rcu_sched). If RCU_BOOST is
5005 set, valid values are 5005 set, valid values are 1-99 and the default is 1
5006 (the least-favored pr 5006 (the least-favored priority). Otherwise, when
5007 RCU_BOOST is not set, 5007 RCU_BOOST is not set, valid values are 0-99 and
5008 the default is zero ( 5008 the default is zero (non-realtime operation).
5009 When RCU_NOCB_CPU is 5009 When RCU_NOCB_CPU is set, also adjust the
5010 priority of NOCB call 5010 priority of NOCB callback kthreads.
5011 5011
5012 rcutree.nocb_nobypass_lim_per_jiffy= 5012 rcutree.nocb_nobypass_lim_per_jiffy= [KNL]
5013 On callback-offloaded 5013 On callback-offloaded (rcu_nocbs) CPUs,
5014 RCU reduces the lock 5014 RCU reduces the lock contention that would
5015 otherwise be caused b 5015 otherwise be caused by callback floods through
5016 use of the ->nocb_byp 5016 use of the ->nocb_bypass list. However, in the
5017 common non-flooded ca 5017 common non-flooded case, RCU queues directly to
5018 the main ->cblist in 5018 the main ->cblist in order to avoid the extra
5019 overhead of the ->noc 5019 overhead of the ->nocb_bypass list and its lock.
5020 But if there are too 5020 But if there are too many callbacks queued during
5021 a single jiffy, RCU p 5021 a single jiffy, RCU pre-queues the callbacks into
5022 the ->nocb_bypass que 5022 the ->nocb_bypass queue. The definition of "too
5023 many" is supplied by 5023 many" is supplied by this kernel boot parameter.
5024 5024
5025 rcutree.nohz_full_patience_delay= [KN 5025 rcutree.nohz_full_patience_delay= [KNL]
5026 On callback-offloaded 5026 On callback-offloaded (rcu_nocbs) CPUs, avoid
5027 disturbing RCU unless 5027 disturbing RCU unless the grace period has
5028 reached the specified 5028 reached the specified age in milliseconds.
5029 Defaults to zero. La 5029 Defaults to zero. Large values will be capped
5030 at five seconds. All 5030 at five seconds. All values will be rounded down
5031 to the nearest value 5031 to the nearest value representable by jiffies.
5032 5032
5033 rcutree.qhimark= [KNL] 5033 rcutree.qhimark= [KNL]
5034 Set threshold of queu 5034 Set threshold of queued RCU callbacks beyond which
5035 batch limiting is dis 5035 batch limiting is disabled.
5036 5036
5037 rcutree.qlowmark= [KNL] 5037 rcutree.qlowmark= [KNL]
5038 Set threshold of queu 5038 Set threshold of queued RCU callbacks below which
5039 batch limiting is re- 5039 batch limiting is re-enabled.
5040 5040
5041 rcutree.qovld= [KNL] 5041 rcutree.qovld= [KNL]
5042 Set threshold of queu 5042 Set threshold of queued RCU callbacks beyond which
5043 RCU's force-quiescent 5043 RCU's force-quiescent-state scan will aggressively
5044 enlist help from cond 5044 enlist help from cond_resched() and sched IPIs to
5045 help CPUs more quickl 5045 help CPUs more quickly reach quiescent states.
5046 Set to less than zero 5046 Set to less than zero to make this be set based
5047 on rcutree.qhimark at 5047 on rcutree.qhimark at boot time and to zero to
5048 disable more aggressi 5048 disable more aggressive help enlistment.
5049 5049
5050 rcutree.rcu_delay_page_cache_fill_mse 5050 rcutree.rcu_delay_page_cache_fill_msec= [KNL]
5051 Set the page-cache re 5051 Set the page-cache refill delay (in milliseconds)
5052 in response to low-me 5052 in response to low-memory conditions. The range
5053 of permitted values i 5053 of permitted values is in the range 0:100000.
5054 5054
5055 rcutree.rcu_divisor= [KNL] 5055 rcutree.rcu_divisor= [KNL]
5056 Set the shift-right c 5056 Set the shift-right count to use to compute
5057 the callback-invocati 5057 the callback-invocation batch limit bl from
5058 the number of callbac 5058 the number of callbacks queued on this CPU.
5059 The result will be bo 5059 The result will be bounded below by the value of
5060 the rcutree.blimit ke 5060 the rcutree.blimit kernel parameter. Every bl
5061 callbacks, the softir 5061 callbacks, the softirq handler will exit in
5062 order to allow the CP 5062 order to allow the CPU to do other work.
5063 5063
5064 Please note that this 5064 Please note that this callback-invocation batch
5065 limit applies only to 5065 limit applies only to non-offloaded callback
5066 invocation. Offloade 5066 invocation. Offloaded callbacks are instead
5067 invoked in the contex 5067 invoked in the context of an rcuoc kthread, which
5068 scheduler will preemp 5068 scheduler will preempt as it does any other task.
5069 5069
5070 rcutree.rcu_fanout_exact= [KNL] 5070 rcutree.rcu_fanout_exact= [KNL]
5071 Disable autobalancing 5071 Disable autobalancing of the rcu_node combining
5072 tree. This is used b 5072 tree. This is used by rcutorture, and might
5073 possibly be useful fo 5073 possibly be useful for architectures having high
5074 cache-to-cache transf 5074 cache-to-cache transfer latencies.
5075 5075
5076 rcutree.rcu_fanout_leaf= [KNL] 5076 rcutree.rcu_fanout_leaf= [KNL]
5077 Change the number of 5077 Change the number of CPUs assigned to each
5078 leaf rcu_node structu 5078 leaf rcu_node structure. Useful for very
5079 large systems, which 5079 large systems, which will choose the value 64,
5080 and for NUMA systems 5080 and for NUMA systems with large remote-access
5081 latencies, which will 5081 latencies, which will choose a value aligned
5082 with the appropriate 5082 with the appropriate hardware boundaries.
5083 5083
5084 rcutree.rcu_min_cached_objs= [KNL] 5084 rcutree.rcu_min_cached_objs= [KNL]
5085 Minimum number of obj 5085 Minimum number of objects which are cached and
5086 maintained per one CP 5086 maintained per one CPU. Object size is equal
5087 to PAGE_SIZE. The cac 5087 to PAGE_SIZE. The cache allows to reduce the
5088 pressure to page allo 5088 pressure to page allocator, also it makes the
5089 whole algorithm to be 5089 whole algorithm to behave better in low memory
5090 condition. 5090 condition.
5091 5091
5092 rcutree.rcu_nocb_gp_stride= [KNL] 5092 rcutree.rcu_nocb_gp_stride= [KNL]
5093 Set the number of NOC 5093 Set the number of NOCB callback kthreads in
5094 each group, which def 5094 each group, which defaults to the square root
5095 of the number of CPUs 5095 of the number of CPUs. Larger numbers reduce
5096 the wakeup overhead o 5096 the wakeup overhead on the global grace-period
5097 kthread, but increase 5097 kthread, but increases that same overhead on
5098 each group's NOCB gra 5098 each group's NOCB grace-period kthread.
5099 5099
5100 rcutree.rcu_kick_kthreads= [KNL] 5100 rcutree.rcu_kick_kthreads= [KNL]
5101 Cause the grace-perio 5101 Cause the grace-period kthread to get an extra
5102 wake_up() if it sleep 5102 wake_up() if it sleeps three times longer than
5103 it should at force-qu 5103 it should at force-quiescent-state time.
5104 This wake_up() will b 5104 This wake_up() will be accompanied by a
5105 WARN_ONCE() splat and 5105 WARN_ONCE() splat and an ftrace_dump().
5106 5106
5107 rcutree.rcu_resched_ns= [KNL] 5107 rcutree.rcu_resched_ns= [KNL]
5108 Limit the time spend 5108 Limit the time spend invoking a batch of RCU
5109 callbacks to the spec 5109 callbacks to the specified number of nanoseconds.
5110 By default, this limi 5110 By default, this limit is checked only once
5111 every 32 callbacks in 5111 every 32 callbacks in order to limit the pain
5112 inflicted by local_cl 5112 inflicted by local_clock() overhead.
5113 5113
5114 rcutree.rcu_unlock_delay= [KNL] 5114 rcutree.rcu_unlock_delay= [KNL]
5115 In CONFIG_RCU_STRICT_ 5115 In CONFIG_RCU_STRICT_GRACE_PERIOD=y kernels,
5116 this specifies an rcu 5116 this specifies an rcu_read_unlock()-time delay
5117 in microseconds. Thi 5117 in microseconds. This defaults to zero.
5118 Larger delays increas 5118 Larger delays increase the probability of
5119 catching RCU pointer 5119 catching RCU pointer leaks, that is, buggy use
5120 of RCU-protected poin 5120 of RCU-protected pointers after the relevant
5121 rcu_read_unlock() has 5121 rcu_read_unlock() has completed.
5122 5122
5123 rcutree.sysrq_rcu= [KNL] 5123 rcutree.sysrq_rcu= [KNL]
5124 Commandeer a sysrq ke 5124 Commandeer a sysrq key to dump out Tree RCU's
5125 rcu_node tree with an 5125 rcu_node tree with an eye towards determining
5126 why a new grace perio 5126 why a new grace period has not yet started.
5127 5127
5128 rcutree.use_softirq= [KNL] 5128 rcutree.use_softirq= [KNL]
5129 If set to zero, move 5129 If set to zero, move all RCU_SOFTIRQ processing to
5130 per-CPU rcuc kthreads 5130 per-CPU rcuc kthreads. Defaults to a non-zero
5131 value, meaning that R 5131 value, meaning that RCU_SOFTIRQ is used by default.
5132 Specify rcutree.use_s 5132 Specify rcutree.use_softirq=0 to use rcuc kthreads.
5133 5133
5134 But note that CONFIG_ 5134 But note that CONFIG_PREEMPT_RT=y kernels disable
5135 this kernel boot para 5135 this kernel boot parameter, forcibly setting it
5136 to zero. 5136 to zero.
5137 5137
5138 rcutree.enable_rcu_lazy= [KNL] 5138 rcutree.enable_rcu_lazy= [KNL]
5139 To save power, batch 5139 To save power, batch RCU callbacks and flush after
5140 delay, memory pressur 5140 delay, memory pressure or callback list growing too
5141 big. 5141 big.
5142 5142
5143 rcutree.rcu_normal_wake_from_gp= [KNL 5143 rcutree.rcu_normal_wake_from_gp= [KNL]
5144 Reduces a latency of 5144 Reduces a latency of synchronize_rcu() call. This approach
5145 maintains its own tra 5145 maintains its own track of synchronize_rcu() callers, so it
5146 does not interact wit 5146 does not interact with regular callbacks because it does not
5147 use a call_rcu[_hurry 5147 use a call_rcu[_hurry]() path. Please note, this is for a
5148 normal grace period. 5148 normal grace period.
5149 5149
5150 How to enable it: 5150 How to enable it:
5151 5151
5152 echo 1 > /sys/module/ 5152 echo 1 > /sys/module/rcutree/parameters/rcu_normal_wake_from_gp
5153 or pass a boot parame 5153 or pass a boot parameter "rcutree.rcu_normal_wake_from_gp=1"
5154 5154
5155 Default is 0. 5155 Default is 0.
5156 5156
5157 rcuscale.gp_async= [KNL] 5157 rcuscale.gp_async= [KNL]
5158 Measure performance o 5158 Measure performance of asynchronous
5159 grace-period primitiv 5159 grace-period primitives such as call_rcu().
5160 5160
5161 rcuscale.gp_async_max= [KNL] 5161 rcuscale.gp_async_max= [KNL]
5162 Specify the maximum n 5162 Specify the maximum number of outstanding
5163 callbacks per writer 5163 callbacks per writer thread. When a writer
5164 thread exceeds this l 5164 thread exceeds this limit, it invokes the
5165 corresponding flavor 5165 corresponding flavor of rcu_barrier() to allow
5166 previously posted cal 5166 previously posted callbacks to drain.
5167 5167
5168 rcuscale.gp_exp= [KNL] 5168 rcuscale.gp_exp= [KNL]
5169 Measure performance o 5169 Measure performance of expedited synchronous
5170 grace-period primitiv 5170 grace-period primitives.
5171 5171
5172 rcuscale.holdoff= [KNL] 5172 rcuscale.holdoff= [KNL]
5173 Set test-start holdof 5173 Set test-start holdoff period. The purpose of
5174 this parameter is to 5174 this parameter is to delay the start of the
5175 test until boot compl 5175 test until boot completes in order to avoid
5176 interference. 5176 interference.
5177 5177
5178 rcuscale.kfree_by_call_rcu= [KNL] 5178 rcuscale.kfree_by_call_rcu= [KNL]
5179 In kernels built with 5179 In kernels built with CONFIG_RCU_LAZY=y, test
5180 call_rcu() instead of 5180 call_rcu() instead of kfree_rcu().
5181 5181
5182 rcuscale.kfree_mult= [KNL] 5182 rcuscale.kfree_mult= [KNL]
5183 Instead of allocating 5183 Instead of allocating an object of size kfree_obj,
5184 allocate one of kfree 5184 allocate one of kfree_mult * sizeof(kfree_obj).
5185 Defaults to 1. 5185 Defaults to 1.
5186 5186
5187 rcuscale.kfree_rcu_test= [KNL] 5187 rcuscale.kfree_rcu_test= [KNL]
5188 Set to measure perfor 5188 Set to measure performance of kfree_rcu() flooding.
5189 5189
5190 rcuscale.kfree_rcu_test_double= [KNL] 5190 rcuscale.kfree_rcu_test_double= [KNL]
5191 Test the double-argum 5191 Test the double-argument variant of kfree_rcu().
5192 If this parameter has 5192 If this parameter has the same value as
5193 rcuscale.kfree_rcu_te 5193 rcuscale.kfree_rcu_test_single, both the single-
5194 and double-argument v 5194 and double-argument variants are tested.
5195 5195
5196 rcuscale.kfree_rcu_test_single= [KNL] 5196 rcuscale.kfree_rcu_test_single= [KNL]
5197 Test the single-argum 5197 Test the single-argument variant of kfree_rcu().
5198 If this parameter has 5198 If this parameter has the same value as
5199 rcuscale.kfree_rcu_te 5199 rcuscale.kfree_rcu_test_double, both the single-
5200 and double-argument v 5200 and double-argument variants are tested.
5201 5201
5202 rcuscale.kfree_nthreads= [KNL] 5202 rcuscale.kfree_nthreads= [KNL]
5203 The number of threads 5203 The number of threads running loops of kfree_rcu().
5204 5204
5205 rcuscale.kfree_alloc_num= [KNL] 5205 rcuscale.kfree_alloc_num= [KNL]
5206 Number of allocations 5206 Number of allocations and frees done in an iteration.
5207 5207
5208 rcuscale.kfree_loops= [KNL] 5208 rcuscale.kfree_loops= [KNL]
5209 Number of loops doing 5209 Number of loops doing rcuscale.kfree_alloc_num number
5210 of allocations and fr 5210 of allocations and frees.
5211 5211
5212 rcuscale.minruntime= [KNL] 5212 rcuscale.minruntime= [KNL]
5213 Set the minimum test 5213 Set the minimum test run time in seconds. This
5214 does not affect the d 5214 does not affect the data-collection interval,
5215 but instead allows be 5215 but instead allows better measurement of things
5216 like CPU consumption. 5216 like CPU consumption.
5217 5217
5218 rcuscale.nreaders= [KNL] 5218 rcuscale.nreaders= [KNL]
5219 Set number of RCU rea 5219 Set number of RCU readers. The value -1 selects
5220 N, where N is the num 5220 N, where N is the number of CPUs. A value
5221 "n" less than -1 sele 5221 "n" less than -1 selects N-n+1, where N is again
5222 the number of CPUs. 5222 the number of CPUs. For example, -2 selects N
5223 (the number of CPUs), 5223 (the number of CPUs), -3 selects N+1, and so on.
5224 A value of "n" less t 5224 A value of "n" less than or equal to -N selects
5225 a single reader. 5225 a single reader.
5226 5226
5227 rcuscale.nwriters= [KNL] 5227 rcuscale.nwriters= [KNL]
5228 Set number of RCU wri 5228 Set number of RCU writers. The values operate
5229 the same as for rcusc 5229 the same as for rcuscale.nreaders.
5230 N, where N is the num 5230 N, where N is the number of CPUs
5231 5231
5232 rcuscale.scale_type= [KNL] 5232 rcuscale.scale_type= [KNL]
5233 Specify the RCU imple 5233 Specify the RCU implementation to test.
5234 5234
5235 rcuscale.shutdown= [KNL] 5235 rcuscale.shutdown= [KNL]
5236 Shut the system down 5236 Shut the system down after performance tests
5237 complete. This is us 5237 complete. This is useful for hands-off automated
5238 testing. 5238 testing.
5239 5239
5240 rcuscale.verbose= [KNL] 5240 rcuscale.verbose= [KNL]
5241 Enable additional pri 5241 Enable additional printk() statements.
5242 5242
5243 rcuscale.writer_holdoff= [KNL] 5243 rcuscale.writer_holdoff= [KNL]
5244 Write-side holdoff be 5244 Write-side holdoff between grace periods,
5245 in microseconds. The 5245 in microseconds. The default of zero says
5246 no holdoff. 5246 no holdoff.
5247 5247
5248 rcuscale.writer_holdoff_jiffies= [KNL 5248 rcuscale.writer_holdoff_jiffies= [KNL]
5249 Additional write-side 5249 Additional write-side holdoff between grace
5250 periods, but in jiffi 5250 periods, but in jiffies. The default of zero
5251 says no holdoff. 5251 says no holdoff.
5252 5252
5253 rcutorture.fqs_duration= [KNL] 5253 rcutorture.fqs_duration= [KNL]
5254 Set duration of force 5254 Set duration of force_quiescent_state bursts
5255 in microseconds. 5255 in microseconds.
5256 5256
5257 rcutorture.fqs_holdoff= [KNL] 5257 rcutorture.fqs_holdoff= [KNL]
5258 Set holdoff time with 5258 Set holdoff time within force_quiescent_state bursts
5259 in microseconds. 5259 in microseconds.
5260 5260
5261 rcutorture.fqs_stutter= [KNL] 5261 rcutorture.fqs_stutter= [KNL]
5262 Set wait time between 5262 Set wait time between force_quiescent_state bursts
5263 in seconds. 5263 in seconds.
5264 5264
5265 rcutorture.fwd_progress= [KNL] 5265 rcutorture.fwd_progress= [KNL]
5266 Specifies the number 5266 Specifies the number of kthreads to be used
5267 for RCU grace-period 5267 for RCU grace-period forward-progress testing
5268 for the types of RCU 5268 for the types of RCU supporting this notion.
5269 Defaults to 1 kthread 5269 Defaults to 1 kthread, values less than zero or
5270 greater than the numb 5270 greater than the number of CPUs cause the number
5271 of CPUs to be used. 5271 of CPUs to be used.
5272 5272
5273 rcutorture.fwd_progress_div= [KNL] 5273 rcutorture.fwd_progress_div= [KNL]
5274 Specify the fraction 5274 Specify the fraction of a CPU-stall-warning
5275 period to do tight-lo 5275 period to do tight-loop forward-progress testing.
5276 5276
5277 rcutorture.fwd_progress_holdoff= [KNL 5277 rcutorture.fwd_progress_holdoff= [KNL]
5278 Number of seconds to 5278 Number of seconds to wait between successive
5279 forward-progress test 5279 forward-progress tests.
5280 5280
5281 rcutorture.fwd_progress_need_resched= 5281 rcutorture.fwd_progress_need_resched= [KNL]
5282 Enclose cond_resched( 5282 Enclose cond_resched() calls within checks for
5283 need_resched() during 5283 need_resched() during tight-loop forward-progress
5284 testing. 5284 testing.
5285 5285
5286 rcutorture.gp_cond= [KNL] 5286 rcutorture.gp_cond= [KNL]
5287 Use conditional/async 5287 Use conditional/asynchronous update-side
5288 primitives, if availa 5288 primitives, if available.
5289 5289
5290 rcutorture.gp_exp= [KNL] 5290 rcutorture.gp_exp= [KNL]
5291 Use expedited update- 5291 Use expedited update-side primitives, if available.
5292 5292
5293 rcutorture.gp_normal= [KNL] 5293 rcutorture.gp_normal= [KNL]
5294 Use normal (non-exped 5294 Use normal (non-expedited) asynchronous
5295 update-side primitive 5295 update-side primitives, if available.
5296 5296
5297 rcutorture.gp_sync= [KNL] 5297 rcutorture.gp_sync= [KNL]
5298 Use normal (non-exped 5298 Use normal (non-expedited) synchronous
5299 update-side primitive 5299 update-side primitives, if available. If all
5300 of rcutorture.gp_cond 5300 of rcutorture.gp_cond=, rcutorture.gp_exp=,
5301 rcutorture.gp_normal= 5301 rcutorture.gp_normal=, and rcutorture.gp_sync=
5302 are zero, rcutorture 5302 are zero, rcutorture acts as if is interpreted
5303 they are all non-zero 5303 they are all non-zero.
5304 5304
5305 rcutorture.irqreader= [KNL] 5305 rcutorture.irqreader= [KNL]
5306 Run RCU readers from 5306 Run RCU readers from irq handlers, or, more
5307 accurately, from a ti 5307 accurately, from a timer handler. Not all RCU
5308 flavors take kindly t 5308 flavors take kindly to this sort of thing.
5309 5309
5310 rcutorture.leakpointer= [KNL] 5310 rcutorture.leakpointer= [KNL]
5311 Leak an RCU-protected 5311 Leak an RCU-protected pointer out of the reader.
5312 This can of course re 5312 This can of course result in splats, and is
5313 intended to test the 5313 intended to test the ability of things like
5314 CONFIG_RCU_STRICT_GRA 5314 CONFIG_RCU_STRICT_GRACE_PERIOD=y to detect
5315 such leaks. 5315 such leaks.
5316 5316
5317 rcutorture.n_barrier_cbs= [KNL] 5317 rcutorture.n_barrier_cbs= [KNL]
5318 Set callbacks/threads 5318 Set callbacks/threads for rcu_barrier() testing.
5319 5319
5320 rcutorture.nfakewriters= [KNL] 5320 rcutorture.nfakewriters= [KNL]
5321 Set number of concurr 5321 Set number of concurrent RCU writers. These just
5322 stress RCU, they don' 5322 stress RCU, they don't participate in the actual
5323 test, hence the "fake 5323 test, hence the "fake".
5324 5324
5325 rcutorture.nocbs_nthreads= [KNL] 5325 rcutorture.nocbs_nthreads= [KNL]
5326 Set number of RCU cal 5326 Set number of RCU callback-offload togglers.
5327 Zero (the default) di 5327 Zero (the default) disables toggling.
5328 5328
5329 rcutorture.nocbs_toggle= [KNL] 5329 rcutorture.nocbs_toggle= [KNL]
5330 Set the delay in mill 5330 Set the delay in milliseconds between successive
5331 callback-offload togg 5331 callback-offload toggling attempts.
5332 5332
5333 rcutorture.nreaders= [KNL] 5333 rcutorture.nreaders= [KNL]
5334 Set number of RCU rea 5334 Set number of RCU readers. The value -1 selects
5335 N-1, where N is the n 5335 N-1, where N is the number of CPUs. A value
5336 "n" less than -1 sele 5336 "n" less than -1 selects N-n-2, where N is again
5337 the number of CPUs. 5337 the number of CPUs. For example, -2 selects N
5338 (the number of CPUs), 5338 (the number of CPUs), -3 selects N+1, and so on.
5339 5339
5340 rcutorture.object_debug= [KNL] 5340 rcutorture.object_debug= [KNL]
5341 Enable debug-object d 5341 Enable debug-object double-call_rcu() testing.
5342 5342
5343 rcutorture.onoff_holdoff= [KNL] 5343 rcutorture.onoff_holdoff= [KNL]
5344 Set time (s) after bo 5344 Set time (s) after boot for CPU-hotplug testing.
5345 5345
5346 rcutorture.onoff_interval= [KNL] 5346 rcutorture.onoff_interval= [KNL]
5347 Set time (jiffies) be 5347 Set time (jiffies) between CPU-hotplug operations,
5348 or zero to disable CP 5348 or zero to disable CPU-hotplug testing.
5349 5349
5350 rcutorture.read_exit= [KNL] 5350 rcutorture.read_exit= [KNL]
5351 Set the number of rea 5351 Set the number of read-then-exit kthreads used
5352 to test the interacti 5352 to test the interaction of RCU updaters and
5353 task-exit processing. 5353 task-exit processing.
5354 5354
5355 rcutorture.read_exit_burst= [KNL] 5355 rcutorture.read_exit_burst= [KNL]
5356 The number of times i 5356 The number of times in a given read-then-exit
5357 episode that a set of 5357 episode that a set of read-then-exit kthreads
5358 is spawned. 5358 is spawned.
5359 5359
5360 rcutorture.read_exit_delay= [KNL] 5360 rcutorture.read_exit_delay= [KNL]
5361 The delay, in seconds 5361 The delay, in seconds, between successive
5362 read-then-exit testin 5362 read-then-exit testing episodes.
5363 5363
5364 rcutorture.shuffle_interval= [KNL] 5364 rcutorture.shuffle_interval= [KNL]
5365 Set task-shuffle inte 5365 Set task-shuffle interval (s). Shuffling tasks
5366 allows some CPUs to g 5366 allows some CPUs to go into dyntick-idle mode
5367 during the rcutorture 5367 during the rcutorture test.
5368 5368
5369 rcutorture.shutdown_secs= [KNL] 5369 rcutorture.shutdown_secs= [KNL]
5370 Set time (s) after bo 5370 Set time (s) after boot system shutdown. This
5371 is useful for hands-o 5371 is useful for hands-off automated testing.
5372 5372
5373 rcutorture.stall_cpu= [KNL] 5373 rcutorture.stall_cpu= [KNL]
5374 Duration of CPU stall 5374 Duration of CPU stall (s) to test RCU CPU stall
5375 warnings, zero to dis 5375 warnings, zero to disable.
5376 5376
5377 rcutorture.stall_cpu_block= [KNL] 5377 rcutorture.stall_cpu_block= [KNL]
5378 Sleep while stalling 5378 Sleep while stalling if set. This will result
5379 in warnings from pree 5379 in warnings from preemptible RCU in addition to
5380 any other stall-relat 5380 any other stall-related activity. Note that
5381 in kernels built with 5381 in kernels built with CONFIG_PREEMPTION=n and
5382 CONFIG_PREEMPT_COUNT= 5382 CONFIG_PREEMPT_COUNT=y, this parameter will
5383 cause the CPU to pass 5383 cause the CPU to pass through a quiescent state.
5384 Given CONFIG_PREEMPTI 5384 Given CONFIG_PREEMPTION=n, this will suppress
5385 RCU CPU stall warning 5385 RCU CPU stall warnings, but will instead result
5386 in scheduling-while-a 5386 in scheduling-while-atomic splats.
5387 5387
5388 Use of this module pa 5388 Use of this module parameter results in splats.
5389 5389
5390 5390
5391 rcutorture.stall_cpu_holdoff= [KNL] 5391 rcutorture.stall_cpu_holdoff= [KNL]
5392 Time to wait (s) afte 5392 Time to wait (s) after boot before inducing stall.
5393 5393
5394 rcutorture.stall_cpu_irqsoff= [KNL] 5394 rcutorture.stall_cpu_irqsoff= [KNL]
5395 Disable interrupts wh 5395 Disable interrupts while stalling if set.
5396 5396
5397 rcutorture.stall_gp_kthread= [KNL] 5397 rcutorture.stall_gp_kthread= [KNL]
5398 Duration (s) of force 5398 Duration (s) of forced sleep within RCU
5399 grace-period kthread 5399 grace-period kthread to test RCU CPU stall
5400 warnings, zero to dis 5400 warnings, zero to disable. If both stall_cpu
5401 and stall_gp_kthread 5401 and stall_gp_kthread are specified, the
5402 kthread is starved fi 5402 kthread is starved first, then the CPU.
5403 5403
5404 rcutorture.stat_interval= [KNL] 5404 rcutorture.stat_interval= [KNL]
5405 Time (s) between stat 5405 Time (s) between statistics printk()s.
5406 5406
5407 rcutorture.stutter= [KNL] 5407 rcutorture.stutter= [KNL]
5408 Time (s) to stutter t 5408 Time (s) to stutter testing, for example, specifying
5409 five seconds causes t 5409 five seconds causes the test to run for five seconds,
5410 wait for five seconds 5410 wait for five seconds, and so on. This tests RCU's
5411 ability to transition 5411 ability to transition abruptly to and from idle.
5412 5412
5413 rcutorture.test_boost= [KNL] 5413 rcutorture.test_boost= [KNL]
5414 Test RCU priority boo 5414 Test RCU priority boosting? 0=no, 1=maybe, 2=yes.
5415 "Maybe" means test if 5415 "Maybe" means test if the RCU implementation
5416 under test support RC 5416 under test support RCU priority boosting.
5417 5417
5418 rcutorture.test_boost_duration= [KNL] 5418 rcutorture.test_boost_duration= [KNL]
5419 Duration (s) of each 5419 Duration (s) of each individual boost test.
5420 5420
5421 rcutorture.test_boost_interval= [KNL] 5421 rcutorture.test_boost_interval= [KNL]
5422 Interval (s) between 5422 Interval (s) between each boost test.
5423 5423
5424 rcutorture.test_no_idle_hz= [KNL] 5424 rcutorture.test_no_idle_hz= [KNL]
5425 Test RCU's dyntick-id 5425 Test RCU's dyntick-idle handling. See also the
5426 rcutorture.shuffle_in 5426 rcutorture.shuffle_interval parameter.
5427 5427
5428 rcutorture.torture_type= [KNL] 5428 rcutorture.torture_type= [KNL]
5429 Specify the RCU imple 5429 Specify the RCU implementation to test.
5430 5430
5431 rcutorture.verbose= [KNL] 5431 rcutorture.verbose= [KNL]
5432 Enable additional pri 5432 Enable additional printk() statements.
5433 5433
5434 rcupdate.rcu_cpu_stall_ftrace_dump= [ 5434 rcupdate.rcu_cpu_stall_ftrace_dump= [KNL]
5435 Dump ftrace buffer af 5435 Dump ftrace buffer after reporting RCU CPU
5436 stall warning. 5436 stall warning.
5437 5437
5438 rcupdate.rcu_cpu_stall_notifiers= [KN 5438 rcupdate.rcu_cpu_stall_notifiers= [KNL]
5439 Provide RCU CPU stall 5439 Provide RCU CPU stall notifiers, but see the
5440 warnings in the RCU_C 5440 warnings in the RCU_CPU_STALL_NOTIFIER Kconfig
5441 option's help text. 5441 option's help text. TL;DR: You almost certainly
5442 do not want rcupdate. 5442 do not want rcupdate.rcu_cpu_stall_notifiers.
5443 5443
5444 rcupdate.rcu_cpu_stall_suppress= [KNL 5444 rcupdate.rcu_cpu_stall_suppress= [KNL]
5445 Suppress RCU CPU stal 5445 Suppress RCU CPU stall warning messages.
5446 5446
5447 rcupdate.rcu_cpu_stall_suppress_at_bo 5447 rcupdate.rcu_cpu_stall_suppress_at_boot= [KNL]
5448 Suppress RCU CPU stal 5448 Suppress RCU CPU stall warning messages and
5449 rcutorture writer sta 5449 rcutorture writer stall warnings that occur
5450 during early boot, th 5450 during early boot, that is, during the time
5451 before the init task 5451 before the init task is spawned.
5452 5452
5453 rcupdate.rcu_cpu_stall_timeout= [KNL] 5453 rcupdate.rcu_cpu_stall_timeout= [KNL]
5454 Set timeout for RCU C 5454 Set timeout for RCU CPU stall warning messages.
5455 The value is in secon 5455 The value is in seconds and the maximum allowed
5456 value is 300 seconds. 5456 value is 300 seconds.
5457 5457
5458 rcupdate.rcu_exp_cpu_stall_timeout= [ 5458 rcupdate.rcu_exp_cpu_stall_timeout= [KNL]
5459 Set timeout for exped 5459 Set timeout for expedited RCU CPU stall warning
5460 messages. The value 5460 messages. The value is in milliseconds
5461 and the maximum allow 5461 and the maximum allowed value is 21000
5462 milliseconds. Please 5462 milliseconds. Please note that this value is
5463 adjusted to an arch t 5463 adjusted to an arch timer tick resolution.
5464 Setting this to zero 5464 Setting this to zero causes the value from
5465 rcupdate.rcu_cpu_stal 5465 rcupdate.rcu_cpu_stall_timeout to be used (after
5466 conversion from secon 5466 conversion from seconds to milliseconds).
5467 5467
5468 rcupdate.rcu_cpu_stall_cputime= [KNL] 5468 rcupdate.rcu_cpu_stall_cputime= [KNL]
5469 Provide statistics on 5469 Provide statistics on the cputime and count of
5470 interrupts and tasks 5470 interrupts and tasks during the sampling period. For
5471 multiple continuous R 5471 multiple continuous RCU stalls, all sampling periods
5472 begin at half of the 5472 begin at half of the first RCU stall timeout.
5473 5473
5474 rcupdate.rcu_exp_stall_task_details= 5474 rcupdate.rcu_exp_stall_task_details= [KNL]
5475 Print stack dumps of 5475 Print stack dumps of any tasks blocking the
5476 current expedited RCU 5476 current expedited RCU grace period during an
5477 expedited RCU CPU sta 5477 expedited RCU CPU stall warning.
5478 5478
5479 rcupdate.rcu_expedited= [KNL] 5479 rcupdate.rcu_expedited= [KNL]
5480 Use expedited grace-p 5480 Use expedited grace-period primitives, for
5481 example, synchronize_ 5481 example, synchronize_rcu_expedited() instead
5482 of synchronize_rcu(). 5482 of synchronize_rcu(). This reduces latency,
5483 but can increase CPU 5483 but can increase CPU utilization, degrade
5484 real-time latency, an 5484 real-time latency, and degrade energy efficiency.
5485 No effect on CONFIG_T 5485 No effect on CONFIG_TINY_RCU kernels.
5486 5486
5487 rcupdate.rcu_normal= [KNL] 5487 rcupdate.rcu_normal= [KNL]
5488 Use only normal grace 5488 Use only normal grace-period primitives,
5489 for example, synchron 5489 for example, synchronize_rcu() instead of
5490 synchronize_rcu_exped 5490 synchronize_rcu_expedited(). This improves
5491 real-time latency, CP 5491 real-time latency, CPU utilization, and
5492 energy efficiency, bu 5492 energy efficiency, but can expose users to
5493 increased grace-perio 5493 increased grace-period latency. This parameter
5494 overrides rcupdate.rc 5494 overrides rcupdate.rcu_expedited. No effect on
5495 CONFIG_TINY_RCU kerne 5495 CONFIG_TINY_RCU kernels.
5496 5496
5497 rcupdate.rcu_normal_after_boot= [KNL] 5497 rcupdate.rcu_normal_after_boot= [KNL]
5498 Once boot has complet 5498 Once boot has completed (that is, after
5499 rcu_end_inkernel_boot 5499 rcu_end_inkernel_boot() has been invoked), use
5500 only normal grace-per 5500 only normal grace-period primitives. No effect
5501 on CONFIG_TINY_RCU ke 5501 on CONFIG_TINY_RCU kernels.
5502 5502
5503 But note that CONFIG_ 5503 But note that CONFIG_PREEMPT_RT=y kernels enables
5504 this kernel boot para 5504 this kernel boot parameter, forcibly setting
5505 it to the value one, 5505 it to the value one, that is, converting any
5506 post-boot attempt at 5506 post-boot attempt at an expedited RCU grace
5507 period to instead use 5507 period to instead use normal non-expedited
5508 grace-period processi 5508 grace-period processing.
5509 5509
5510 rcupdate.rcu_task_collapse_lim= [KNL] 5510 rcupdate.rcu_task_collapse_lim= [KNL]
5511 Set the maximum numbe 5511 Set the maximum number of callbacks present
5512 at the beginning of a 5512 at the beginning of a grace period that allows
5513 the RCU Tasks flavors 5513 the RCU Tasks flavors to collapse back to using
5514 a single callback que 5514 a single callback queue. This switching only
5515 occurs when rcupdate. 5515 occurs when rcupdate.rcu_task_enqueue_lim is
5516 set to the default va 5516 set to the default value of -1.
5517 5517
5518 rcupdate.rcu_task_contend_lim= [KNL] 5518 rcupdate.rcu_task_contend_lim= [KNL]
5519 Set the minimum numbe 5519 Set the minimum number of callback-queuing-time
5520 lock-contention event 5520 lock-contention events per jiffy required to
5521 cause the RCU Tasks f 5521 cause the RCU Tasks flavors to switch to per-CPU
5522 callback queuing. Th 5522 callback queuing. This switching only occurs
5523 when rcupdate.rcu_tas 5523 when rcupdate.rcu_task_enqueue_lim is set to
5524 the default value of 5524 the default value of -1.
5525 5525
5526 rcupdate.rcu_task_enqueue_lim= [KNL] 5526 rcupdate.rcu_task_enqueue_lim= [KNL]
5527 Set the number of cal 5527 Set the number of callback queues to use for the
5528 RCU Tasks family of R 5528 RCU Tasks family of RCU flavors. The default
5529 of -1 allows this to 5529 of -1 allows this to be automatically (and
5530 dynamically) adjusted 5530 dynamically) adjusted. This parameter is intended
5531 for use in testing. 5531 for use in testing.
5532 5532
5533 rcupdate.rcu_task_ipi_delay= [KNL] 5533 rcupdate.rcu_task_ipi_delay= [KNL]
5534 Set time in jiffies d 5534 Set time in jiffies during which RCU tasks will
5535 avoid sending IPIs, s 5535 avoid sending IPIs, starting with the beginning
5536 of a given grace peri 5536 of a given grace period. Setting a large
5537 number avoids disturb 5537 number avoids disturbing real-time workloads,
5538 but lengthens grace p 5538 but lengthens grace periods.
5539 5539
5540 rcupdate.rcu_task_lazy_lim= [KNL] 5540 rcupdate.rcu_task_lazy_lim= [KNL]
5541 Number of callbacks o 5541 Number of callbacks on a given CPU that will
5542 cancel laziness on th 5542 cancel laziness on that CPU. Use -1 to disable
5543 cancellation of lazin 5543 cancellation of laziness, but be advised that
5544 doing so increases th 5544 doing so increases the danger of OOM due to
5545 callback flooding. 5545 callback flooding.
5546 5546
5547 rcupdate.rcu_task_stall_info= [KNL] 5547 rcupdate.rcu_task_stall_info= [KNL]
5548 Set initial timeout i 5548 Set initial timeout in jiffies for RCU task stall
5549 informational message 5549 informational messages, which give some indication
5550 of the problem for th 5550 of the problem for those not patient enough to
5551 wait for ten minutes. 5551 wait for ten minutes. Informational messages are
5552 only printed prior to 5552 only printed prior to the stall-warning message
5553 for a given grace per 5553 for a given grace period. Disable with a value
5554 less than or equal to 5554 less than or equal to zero. Defaults to ten
5555 seconds. A change in 5555 seconds. A change in value does not take effect
5556 until the beginning o 5556 until the beginning of the next grace period.
5557 5557
5558 rcupdate.rcu_task_stall_info_mult= [K 5558 rcupdate.rcu_task_stall_info_mult= [KNL]
5559 Multiplier for time i 5559 Multiplier for time interval between successive
5560 RCU task stall inform 5560 RCU task stall informational messages for a given
5561 RCU tasks grace perio 5561 RCU tasks grace period. This value is clamped
5562 to one through ten, i 5562 to one through ten, inclusive. It defaults to
5563 the value three, so t 5563 the value three, so that the first informational
5564 message is printed 10 5564 message is printed 10 seconds into the grace
5565 period, the second at 5565 period, the second at 40 seconds, the third at
5566 160 seconds, and then 5566 160 seconds, and then the stall warning at 600
5567 seconds would prevent 5567 seconds would prevent a fourth at 640 seconds.
5568 5568
5569 rcupdate.rcu_task_stall_timeout= [KNL 5569 rcupdate.rcu_task_stall_timeout= [KNL]
5570 Set timeout in jiffie 5570 Set timeout in jiffies for RCU task stall
5571 warning messages. Di 5571 warning messages. Disable with a value less
5572 than or equal to zero 5572 than or equal to zero. Defaults to ten minutes.
5573 A change in value doe 5573 A change in value does not take effect until
5574 the beginning of the 5574 the beginning of the next grace period.
5575 5575
5576 rcupdate.rcu_tasks_lazy_ms= [KNL] 5576 rcupdate.rcu_tasks_lazy_ms= [KNL]
5577 Set timeout in millis 5577 Set timeout in milliseconds RCU Tasks asynchronous
5578 callback batching for 5578 callback batching for call_rcu_tasks().
5579 A negative value will 5579 A negative value will take the default. A value
5580 of zero will disable 5580 of zero will disable batching. Batching is
5581 always disabled for s 5581 always disabled for synchronize_rcu_tasks().
5582 5582
5583 rcupdate.rcu_tasks_rude_lazy_ms= [KNL 5583 rcupdate.rcu_tasks_rude_lazy_ms= [KNL]
5584 Set timeout in millis 5584 Set timeout in milliseconds RCU Tasks
5585 Rude asynchronous cal 5585 Rude asynchronous callback batching for
5586 call_rcu_tasks_rude() 5586 call_rcu_tasks_rude(). A negative value
5587 will take the default 5587 will take the default. A value of zero will
5588 disable batching. Ba 5588 disable batching. Batching is always disabled
5589 for synchronize_rcu_t 5589 for synchronize_rcu_tasks_rude().
5590 5590
5591 rcupdate.rcu_tasks_trace_lazy_ms= [KN 5591 rcupdate.rcu_tasks_trace_lazy_ms= [KNL]
5592 Set timeout in millis 5592 Set timeout in milliseconds RCU Tasks
5593 Trace asynchronous ca 5593 Trace asynchronous callback batching for
5594 call_rcu_tasks_trace( 5594 call_rcu_tasks_trace(). A negative value
5595 will take the default 5595 will take the default. A value of zero will
5596 disable batching. Ba 5596 disable batching. Batching is always disabled
5597 for synchronize_rcu_t 5597 for synchronize_rcu_tasks_trace().
5598 5598
5599 rcupdate.rcu_self_test= [KNL] 5599 rcupdate.rcu_self_test= [KNL]
5600 Run the RCU early boo 5600 Run the RCU early boot self tests
5601 5601
5602 rdinit= [KNL] 5602 rdinit= [KNL]
5603 Format: <full_path> 5603 Format: <full_path>
5604 Run specified binary 5604 Run specified binary instead of /init from the ramdisk,
5605 used for early usersp 5605 used for early userspace startup. See initrd.
5606 5606
5607 rdrand= [X86,EARLY] 5607 rdrand= [X86,EARLY]
5608 force - Override the 5608 force - Override the decision by the kernel to hide the
5609 advertisement 5609 advertisement of RDRAND support (this affects
5610 certain AMD p 5610 certain AMD processors because of buggy BIOS
5611 support, spec 5611 support, specifically around the suspend/resume
5612 path). 5612 path).
5613 5613
5614 rdt= [HW,X86,RDT] 5614 rdt= [HW,X86,RDT]
5615 Turn on/off individua 5615 Turn on/off individual RDT features. List is:
5616 cmt, mbmtotal, mbmloc 5616 cmt, mbmtotal, mbmlocal, l3cat, l3cdp, l2cat, l2cdp,
5617 mba, smba, bmec. 5617 mba, smba, bmec.
5618 E.g. to turn on cmt a 5618 E.g. to turn on cmt and turn off mba use:
5619 rdt=cmt,!mba 5619 rdt=cmt,!mba
5620 5620
5621 reboot= [KNL] 5621 reboot= [KNL]
5622 Format (x86 or x86_64 5622 Format (x86 or x86_64):
5623 [w[arm] | c[o 5623 [w[arm] | c[old] | h[ard] | s[oft] | g[pio]] | d[efault] \
5624 [[,]s[mp]#### 5624 [[,]s[mp]#### \
5625 [[,]b[ios] | 5625 [[,]b[ios] | a[cpi] | k[bd] | t[riple] | e[fi] | p[ci]] \
5626 [[,]f[orce] 5626 [[,]f[orce]
5627 Where reboot_mode is 5627 Where reboot_mode is one of warm (soft) or cold (hard) or gpio
5628 (pref 5628 (prefix with 'panic_' to set mode for panic
5629 reboo 5629 reboot only),
5630 reboot_type is 5630 reboot_type is one of bios, acpi, kbd, triple, efi, or pci,
5631 reboot_force is 5631 reboot_force is either force or not specified,
5632 reboot_cpu is s 5632 reboot_cpu is s[mp]#### with #### being the processor
5633 to be 5633 to be used for rebooting.
5634 5634
5635 refscale.holdoff= [KNL] 5635 refscale.holdoff= [KNL]
5636 Set test-start holdof 5636 Set test-start holdoff period. The purpose of
5637 this parameter is to 5637 this parameter is to delay the start of the
5638 test until boot compl 5638 test until boot completes in order to avoid
5639 interference. 5639 interference.
5640 5640
5641 refscale.lookup_instances= [KNL] 5641 refscale.lookup_instances= [KNL]
5642 Number of data elemen 5642 Number of data elements to use for the forms of
5643 SLAB_TYPESAFE_BY_RCU 5643 SLAB_TYPESAFE_BY_RCU testing. A negative number
5644 is negated and multip 5644 is negated and multiplied by nr_cpu_ids, while
5645 zero specifies nr_cpu 5645 zero specifies nr_cpu_ids.
5646 5646
5647 refscale.loops= [KNL] 5647 refscale.loops= [KNL]
5648 Set the number of loo 5648 Set the number of loops over the synchronization
5649 primitive under test. 5649 primitive under test. Increasing this number
5650 reduces noise due to 5650 reduces noise due to loop start/end overhead,
5651 but the default has a 5651 but the default has already reduced the per-pass
5652 noise to a handful of 5652 noise to a handful of picoseconds on ca. 2020
5653 x86 laptops. 5653 x86 laptops.
5654 5654
5655 refscale.nreaders= [KNL] 5655 refscale.nreaders= [KNL]
5656 Set number of readers 5656 Set number of readers. The default value of -1
5657 selects N, where N is 5657 selects N, where N is roughly 75% of the number
5658 of CPUs. A value of 5658 of CPUs. A value of zero is an interesting choice.
5659 5659
5660 refscale.nruns= [KNL] 5660 refscale.nruns= [KNL]
5661 Set number of runs, e 5661 Set number of runs, each of which is dumped onto
5662 the console log. 5662 the console log.
5663 5663
5664 refscale.readdelay= [KNL] 5664 refscale.readdelay= [KNL]
5665 Set the read-side cri 5665 Set the read-side critical-section duration,
5666 measured in microseco 5666 measured in microseconds.
5667 5667
5668 refscale.scale_type= [KNL] 5668 refscale.scale_type= [KNL]
5669 Specify the read-prot 5669 Specify the read-protection implementation to test.
5670 5670
5671 refscale.shutdown= [KNL] 5671 refscale.shutdown= [KNL]
5672 Shut down the system 5672 Shut down the system at the end of the performance
5673 test. This defaults 5673 test. This defaults to 1 (shut it down) when
5674 refscale is built int 5674 refscale is built into the kernel and to 0 (leave
5675 it running) when refs 5675 it running) when refscale is built as a module.
5676 5676
5677 refscale.verbose= [KNL] 5677 refscale.verbose= [KNL]
5678 Enable additional pri 5678 Enable additional printk() statements.
5679 5679
5680 refscale.verbose_batched= [KNL] 5680 refscale.verbose_batched= [KNL]
5681 Batch the additional 5681 Batch the additional printk() statements. If zero
5682 (the default) or nega 5682 (the default) or negative, print everything. Otherwise,
5683 print every Nth verbo 5683 print every Nth verbose statement, where N is the value
5684 specified. 5684 specified.
5685 5685
5686 regulator_ignore_unused 5686 regulator_ignore_unused
5687 [REGULATOR] 5687 [REGULATOR]
5688 Prevents regulator fr 5688 Prevents regulator framework from disabling regulators
5689 that are unused, due 5689 that are unused, due no driver claiming them. This may
5690 be useful for debug a 5690 be useful for debug and development, but should not be
5691 needed on a platform 5691 needed on a platform with proper driver support.
5692 5692
5693 relax_domain_level= 5693 relax_domain_level=
5694 [KNL, SMP] Set schedu 5694 [KNL, SMP] Set scheduler's default relax_domain_level.
5695 See Documentation/adm 5695 See Documentation/admin-guide/cgroup-v1/cpusets.rst.
5696 5696
5697 reserve= [KNL,BUGS] Force kern 5697 reserve= [KNL,BUGS] Force kernel to ignore I/O ports or memory
5698 Format: <base1>,<size 5698 Format: <base1>,<size1>[,<base2>,<size2>,...]
5699 Reserve I/O ports or 5699 Reserve I/O ports or memory so the kernel won't use
5700 them. If <base> is l 5700 them. If <base> is less than 0x10000, the region
5701 is assumed to be I/O 5701 is assumed to be I/O ports; otherwise it is memory.
5702 5702
5703 reserve_mem= [RAM] 5703 reserve_mem= [RAM]
5704 Format: nn[KNG]:<alig 5704 Format: nn[KNG]:<align>:<label>
5705 Reserve physical memo 5705 Reserve physical memory and label it with a name that
5706 other subsystems can 5706 other subsystems can use to access it. This is typically
5707 used for systems that 5707 used for systems that do not wipe the RAM, and this command
5708 line will try to rese 5708 line will try to reserve the same physical memory on
5709 soft reboots. Note, i 5709 soft reboots. Note, it is not guaranteed to be the same
5710 location. For example 5710 location. For example, if anything about the system changes
5711 or if booting a diffe 5711 or if booting a different kernel. It can also fail if KASLR
5712 places the kernel at 5712 places the kernel at the location of where the RAM reservation
5713 was from a previous b 5713 was from a previous boot, the new reservation will be at a
5714 different location. 5714 different location.
5715 Any subsystem using t 5715 Any subsystem using this feature must add a way to verify
5716 that the contents of 5716 that the contents of the physical memory is from a previous
5717 boot, as there may be 5717 boot, as there may be cases where the memory will not be
5718 located at the same l 5718 located at the same location.
5719 5719
5720 The format is size:al 5720 The format is size:align:label for example, to request
5721 12 megabytes of 4096 5721 12 megabytes of 4096 alignment for ramoops:
5722 5722
5723 reserve_mem=12M:4096: 5723 reserve_mem=12M:4096:oops ramoops.mem_name=oops
5724 5724
5725 reservetop= [X86-32,EARLY] 5725 reservetop= [X86-32,EARLY]
5726 Format: nn[KMG] 5726 Format: nn[KMG]
5727 Reserves a hole at th 5727 Reserves a hole at the top of the kernel virtual
5728 address space. 5728 address space.
5729 5729
5730 reset_devices [KNL] Force drivers t 5730 reset_devices [KNL] Force drivers to reset the underlying device
5731 during initialization 5731 during initialization.
5732 5732
5733 resume= [SWSUSP] 5733 resume= [SWSUSP]
5734 Specify the partition 5734 Specify the partition device for software suspend
5735 Format: 5735 Format:
5736 {/dev/<dev> | PARTUUI 5736 {/dev/<dev> | PARTUUID=<uuid> | <int>:<int> | <hex>}
5737 5737
5738 resume_offset= [SWSUSP] 5738 resume_offset= [SWSUSP]
5739 Specify the offset fr 5739 Specify the offset from the beginning of the partition
5740 given by "resume=" at 5740 given by "resume=" at which the swap header is located,
5741 in <PAGE_SIZE> units 5741 in <PAGE_SIZE> units (needed only for swap files).
5742 See Documentation/po 5742 See Documentation/power/swsusp-and-swap-files.rst
5743 5743
5744 resumedelay= [HIBERNATION] Delay ( 5744 resumedelay= [HIBERNATION] Delay (in seconds) to pause before attempting to
5745 read the resume files 5745 read the resume files
5746 5746
5747 resumewait [HIBERNATION] Wait (i 5747 resumewait [HIBERNATION] Wait (indefinitely) for resume device to show up.
5748 Useful for devices th 5748 Useful for devices that are detected asynchronously
5749 (e.g. USB and MMC dev 5749 (e.g. USB and MMC devices).
5750 5750
5751 retain_initrd [RAM] Keep initrd mem 5751 retain_initrd [RAM] Keep initrd memory after extraction. After boot, it will
5752 be accessible via /sy 5752 be accessible via /sys/firmware/initrd.
5753 5753
5754 retbleed= [X86] Control mitigat 5754 retbleed= [X86] Control mitigation of RETBleed (Arbitrary
5755 Speculative Code Exec 5755 Speculative Code Execution with Return Instructions)
5756 vulnerability. 5756 vulnerability.
5757 5757
5758 AMD-based UNRET and I 5758 AMD-based UNRET and IBPB mitigations alone do not stop
5759 sibling threads from 5759 sibling threads from influencing the predictions of other
5760 sibling threads. For 5760 sibling threads. For that reason, STIBP is used on pro-
5761 cessors that support 5761 cessors that support it, and mitigate SMT on processors
5762 that don't. 5762 that don't.
5763 5763
5764 off - no mit 5764 off - no mitigation
5765 auto - automa 5765 auto - automatically select a migitation
5766 auto,nosmt - automa 5766 auto,nosmt - automatically select a mitigation,
5767 disabl 5767 disabling SMT if necessary for
5768 the fu 5768 the full mitigation (only on Zen1
5769 and ol 5769 and older without STIBP).
5770 ibpb - On AMD 5770 ibpb - On AMD, mitigate short speculation
5771 window 5771 windows on basic block boundaries too.
5772 Safe, 5772 Safe, highest perf impact. It also
5773 enable 5773 enables STIBP if present. Not suitable
5774 on Int 5774 on Intel.
5775 ibpb,nosmt - Like " 5775 ibpb,nosmt - Like "ibpb" above but will disable SMT
5776 when S 5776 when STIBP is not available. This is
5777 the al 5777 the alternative for systems which do not
5778 have S 5778 have STIBP.
5779 unret - Force 5779 unret - Force enable untrained return thunks,
5780 only e 5780 only effective on AMD f15h-f17h based
5781 system 5781 systems.
5782 unret,nosmt - Like u 5782 unret,nosmt - Like unret, but will disable SMT when STIBP
5783 is not 5783 is not available. This is the alternative for
5784 system 5784 systems which do not have STIBP.
5785 5785
5786 Selecting 'auto' will 5786 Selecting 'auto' will choose a mitigation method at run
5787 time according to the 5787 time according to the CPU.
5788 5788
5789 Not specifying this o 5789 Not specifying this option is equivalent to retbleed=auto.
5790 5790
5791 rfkill.default_state= 5791 rfkill.default_state=
5792 0 "airplane mode". All 5792 0 "airplane mode". All wifi, bluetooth, wimax, gps, fm,
5793 etc. communication is 5793 etc. communication is blocked by default.
5794 1 Unblocked. 5794 1 Unblocked.
5795 5795
5796 rfkill.master_switch_mode= 5796 rfkill.master_switch_mode=
5797 0 The "airplane mode" b 5797 0 The "airplane mode" button does nothing.
5798 1 The "airplane mode" b 5798 1 The "airplane mode" button toggles between everything
5799 blocked and the previ 5799 blocked and the previous configuration.
5800 2 The "airplane mode" b 5800 2 The "airplane mode" button toggles between everything
5801 blocked and everythin 5801 blocked and everything unblocked.
5802 5802
5803 ring3mwait=disable 5803 ring3mwait=disable
5804 [KNL] Disable ring 3 5804 [KNL] Disable ring 3 MONITOR/MWAIT feature on supported
5805 CPUs. 5805 CPUs.
5806 5806
5807 riscv_isa_fallback [RISCV,EARLY] 5807 riscv_isa_fallback [RISCV,EARLY]
5808 When CONFIG_RISCV_ISA 5808 When CONFIG_RISCV_ISA_FALLBACK is not enabled, permit
5809 falling back to detec 5809 falling back to detecting extension support by parsing
5810 "riscv,isa" property 5810 "riscv,isa" property on devicetree systems when the
5811 replacement propertie 5811 replacement properties are not found. See the Kconfig
5812 entry for RISCV_ISA_F 5812 entry for RISCV_ISA_FALLBACK.
5813 5813
5814 ro [KNL] Mount root devi 5814 ro [KNL] Mount root device read-only on boot
5815 5815
5816 rodata= [KNL,EARLY] 5816 rodata= [KNL,EARLY]
5817 on Mark read-only kernel 5817 on Mark read-only kernel memory as read-only (default).
5818 off Leave read-only kerne 5818 off Leave read-only kernel memory writable for debugging.
5819 full Mark read-only kernel 5819 full Mark read-only kernel memory and aliases as read-only
5820 [arm64] 5820 [arm64]
5821 5821
5822 rockchip.usb_uart 5822 rockchip.usb_uart
5823 [EARLY] 5823 [EARLY]
5824 Enable the uart passt 5824 Enable the uart passthrough on the designated usb port
5825 on Rockchip SoCs. Whe 5825 on Rockchip SoCs. When active, the signals of the
5826 debug-uart get routed 5826 debug-uart get routed to the D+ and D- pins of the usb
5827 port and the regular 5827 port and the regular usb controller gets disabled.
5828 5828
5829 root= [KNL] Root filesystem 5829 root= [KNL] Root filesystem
5830 Usually this a a bloc 5830 Usually this a a block device specifier of some kind,
5831 see the early_lookup_ 5831 see the early_lookup_bdev comment in
5832 block/early-lookup.c 5832 block/early-lookup.c for details.
5833 Alternatively this ca 5833 Alternatively this can be "ram" for the legacy initial
5834 ramdisk, "nfs" and "c 5834 ramdisk, "nfs" and "cifs" for root on a network file
5835 system, or "mtd" and 5835 system, or "mtd" and "ubi" for mounting from raw flash.
5836 5836
5837 rootdelay= [KNL] Delay (in secon 5837 rootdelay= [KNL] Delay (in seconds) to pause before attempting to
5838 mount the root filesy 5838 mount the root filesystem
5839 5839
5840 rootflags= [KNL] Set root filesy 5840 rootflags= [KNL] Set root filesystem mount option string
5841 5841
5842 rootfstype= [KNL] Set root filesy 5842 rootfstype= [KNL] Set root filesystem type
5843 5843
5844 rootwait [KNL] Wait (indefinit 5844 rootwait [KNL] Wait (indefinitely) for root device to show up.
5845 Useful for devices th 5845 Useful for devices that are detected asynchronously
5846 (e.g. USB and MMC dev 5846 (e.g. USB and MMC devices).
5847 5847
5848 rootwait= [KNL] Maximum time (i 5848 rootwait= [KNL] Maximum time (in seconds) to wait for root device
5849 to show up before att 5849 to show up before attempting to mount the root
5850 filesystem. 5850 filesystem.
5851 5851
5852 rproc_mem=nn[KMG][@address] 5852 rproc_mem=nn[KMG][@address]
5853 [KNL,ARM,CMA] Remotep 5853 [KNL,ARM,CMA] Remoteproc physical memory block.
5854 Memory area to be use 5854 Memory area to be used by remote processor image,
5855 managed by CMA. 5855 managed by CMA.
5856 5856
5857 rw [KNL] Mount root devi 5857 rw [KNL] Mount root device read-write on boot
5858 5858
5859 S [KNL] Run init in sin 5859 S [KNL] Run init in single mode
5860 5860
5861 s390_iommu= [HW,S390] 5861 s390_iommu= [HW,S390]
5862 Set s390 IOTLB flushi 5862 Set s390 IOTLB flushing mode
5863 strict 5863 strict
5864 With strict flushing 5864 With strict flushing every unmap operation will result
5865 in an IOTLB flush. De 5865 in an IOTLB flush. Default is lazy flushing before
5866 reuse, which is faste 5866 reuse, which is faster. Deprecated, equivalent to
5867 iommu.strict=1. 5867 iommu.strict=1.
5868 5868
5869 s390_iommu_aperture= [KNL,S390] 5869 s390_iommu_aperture= [KNL,S390]
5870 Specifies the size of 5870 Specifies the size of the per device DMA address space
5871 accessible through th 5871 accessible through the DMA and IOMMU APIs as a decimal
5872 factor of the size of 5872 factor of the size of main memory.
5873 The default is 1 mean 5873 The default is 1 meaning that one can concurrently use
5874 as many DMA addresses 5874 as many DMA addresses as physical memory is installed,
5875 if supported by hardw 5875 if supported by hardware, and thus map all of memory
5876 once. With a value of 5876 once. With a value of 2 one can map all of memory twice
5877 and so on. As a speci 5877 and so on. As a special case a factor of 0 imposes no
5878 restrictions other th 5878 restrictions other than those given by hardware at the
5879 cost of significant a 5879 cost of significant additional memory use for tables.
5880 5880
5881 sa1100ir [NET] 5881 sa1100ir [NET]
5882 See drivers/net/irda/ 5882 See drivers/net/irda/sa1100_ir.c.
5883 5883
5884 sched_verbose [KNL,EARLY] Enables v 5884 sched_verbose [KNL,EARLY] Enables verbose scheduler debug messages.
5885 5885
5886 schedstats= [KNL,X86] Enable or d 5886 schedstats= [KNL,X86] Enable or disable scheduled statistics.
5887 Allowed values are en 5887 Allowed values are enable and disable. This feature
5888 incurs a small amount 5888 incurs a small amount of overhead in the scheduler
5889 but is useful for deb 5889 but is useful for debugging and performance tuning.
5890 5890
5891 sched_thermal_decay_shift= 5891 sched_thermal_decay_shift=
5892 [Deprecated] 5892 [Deprecated]
5893 [KNL, SMP] Set a deca 5893 [KNL, SMP] Set a decay shift for scheduler thermal
5894 pressure signal. Ther 5894 pressure signal. Thermal pressure signal follows the
5895 default decay period 5895 default decay period of other scheduler pelt
5896 signals(usually 32 ms 5896 signals(usually 32 ms but configurable). Setting
5897 sched_thermal_decay_s 5897 sched_thermal_decay_shift will left shift the decay
5898 period for the therma 5898 period for the thermal pressure signal by the shift
5899 value. 5899 value.
5900 i.e. with the default 5900 i.e. with the default pelt decay period of 32 ms
5901 sched_thermal_decay_s 5901 sched_thermal_decay_shift thermal pressure decay pr
5902 1 5902 1 64 ms
5903 2 5903 2 128 ms
5904 and so on. 5904 and so on.
5905 Format: integer betwe 5905 Format: integer between 0 and 10
5906 Default is 0. 5906 Default is 0.
5907 5907
5908 scftorture.holdoff= [KNL] 5908 scftorture.holdoff= [KNL]
5909 Number of seconds to 5909 Number of seconds to hold off before starting
5910 test. Defaults to ze 5910 test. Defaults to zero for module insertion and
5911 to 10 seconds for bui 5911 to 10 seconds for built-in smp_call_function()
5912 tests. 5912 tests.
5913 5913
5914 scftorture.longwait= [KNL] 5914 scftorture.longwait= [KNL]
5915 Request ridiculously 5915 Request ridiculously long waits randomly selected
5916 up to the chosen limi 5916 up to the chosen limit in seconds. Zero (the
5917 default) disables thi 5917 default) disables this feature. Please note
5918 that requesting even 5918 that requesting even small non-zero numbers of
5919 seconds can result in 5919 seconds can result in RCU CPU stall warnings,
5920 softlockup complaints 5920 softlockup complaints, and so on.
5921 5921
5922 scftorture.nthreads= [KNL] 5922 scftorture.nthreads= [KNL]
5923 Number of kthreads to 5923 Number of kthreads to spawn to invoke the
5924 smp_call_function() f 5924 smp_call_function() family of functions.
5925 The default of -1 spe 5925 The default of -1 specifies a number of kthreads
5926 equal to the number o 5926 equal to the number of CPUs.
5927 5927
5928 scftorture.onoff_holdoff= [KNL] 5928 scftorture.onoff_holdoff= [KNL]
5929 Number seconds to wai 5929 Number seconds to wait after the start of the
5930 test before initiatin 5930 test before initiating CPU-hotplug operations.
5931 5931
5932 scftorture.onoff_interval= [KNL] 5932 scftorture.onoff_interval= [KNL]
5933 Number seconds to wai 5933 Number seconds to wait between successive
5934 CPU-hotplug operation 5934 CPU-hotplug operations. Specifying zero (which
5935 is the default) disab 5935 is the default) disables CPU-hotplug operations.
5936 5936
5937 scftorture.shutdown_secs= [KNL] 5937 scftorture.shutdown_secs= [KNL]
5938 The number of seconds 5938 The number of seconds following the start of the
5939 test after which to s 5939 test after which to shut down the system. The
5940 default of zero avoid 5940 default of zero avoids shutting down the system.
5941 Non-zero values are u 5941 Non-zero values are useful for automated tests.
5942 5942
5943 scftorture.stat_interval= [KNL] 5943 scftorture.stat_interval= [KNL]
5944 The number of seconds 5944 The number of seconds between outputting the
5945 current test statisti 5945 current test statistics to the console. A value
5946 of zero disables stat 5946 of zero disables statistics output.
5947 5947
5948 scftorture.stutter_cpus= [KNL] 5948 scftorture.stutter_cpus= [KNL]
5949 The number of jiffies 5949 The number of jiffies to wait between each change
5950 to the set of CPUs un 5950 to the set of CPUs under test.
5951 5951
5952 scftorture.use_cpus_read_lock= [KNL] 5952 scftorture.use_cpus_read_lock= [KNL]
5953 Use use_cpus_read_loc 5953 Use use_cpus_read_lock() instead of the default
5954 preempt_disable() to 5954 preempt_disable() to disable CPU hotplug
5955 while invoking one of 5955 while invoking one of the smp_call_function*()
5956 functions. 5956 functions.
5957 5957
5958 scftorture.verbose= [KNL] 5958 scftorture.verbose= [KNL]
5959 Enable additional pri 5959 Enable additional printk() statements.
5960 5960
5961 scftorture.weight_single= [KNL] 5961 scftorture.weight_single= [KNL]
5962 The probability weigh 5962 The probability weighting to use for the
5963 smp_call_function_sin 5963 smp_call_function_single() function with a zero
5964 "wait" parameter. A 5964 "wait" parameter. A value of -1 selects the
5965 default if all other 5965 default if all other weights are -1. However,
5966 if at least one weigh 5966 if at least one weight has some other value, a
5967 value of -1 will inst 5967 value of -1 will instead select a weight of zero.
5968 5968
5969 scftorture.weight_single_wait= [KNL] 5969 scftorture.weight_single_wait= [KNL]
5970 The probability weigh 5970 The probability weighting to use for the
5971 smp_call_function_sin 5971 smp_call_function_single() function with a
5972 non-zero "wait" param 5972 non-zero "wait" parameter. See weight_single.
5973 5973
5974 scftorture.weight_many= [KNL] 5974 scftorture.weight_many= [KNL]
5975 The probability weigh 5975 The probability weighting to use for the
5976 smp_call_function_man 5976 smp_call_function_many() function with a zero
5977 "wait" parameter. Se 5977 "wait" parameter. See weight_single.
5978 Note well that settin 5978 Note well that setting a high probability for
5979 this weighting can pl 5979 this weighting can place serious IPI load
5980 on the system. 5980 on the system.
5981 5981
5982 scftorture.weight_many_wait= [KNL] 5982 scftorture.weight_many_wait= [KNL]
5983 The probability weigh 5983 The probability weighting to use for the
5984 smp_call_function_man 5984 smp_call_function_many() function with a
5985 non-zero "wait" param 5985 non-zero "wait" parameter. See weight_single
5986 and weight_many. 5986 and weight_many.
5987 5987
5988 scftorture.weight_all= [KNL] 5988 scftorture.weight_all= [KNL]
5989 The probability weigh 5989 The probability weighting to use for the
5990 smp_call_function_all 5990 smp_call_function_all() function with a zero
5991 "wait" parameter. Se 5991 "wait" parameter. See weight_single and
5992 weight_many. 5992 weight_many.
5993 5993
5994 scftorture.weight_all_wait= [KNL] 5994 scftorture.weight_all_wait= [KNL]
5995 The probability weigh 5995 The probability weighting to use for the
5996 smp_call_function_all 5996 smp_call_function_all() function with a
5997 non-zero "wait" param 5997 non-zero "wait" parameter. See weight_single
5998 and weight_many. 5998 and weight_many.
5999 5999
6000 skew_tick= [KNL,EARLY] Offset th 6000 skew_tick= [KNL,EARLY] Offset the periodic timer tick per cpu to mitigate
6001 xtime_lock contention 6001 xtime_lock contention on larger systems, and/or RCU lock
6002 contention on all sys 6002 contention on all systems with CONFIG_MAXSMP set.
6003 Format: { "0" | "1" } 6003 Format: { "0" | "1" }
6004 0 -- disable. (may be 6004 0 -- disable. (may be 1 via CONFIG_CMDLINE="skew_tick=1"
6005 1 -- enable. 6005 1 -- enable.
6006 Note: increases power 6006 Note: increases power consumption, thus should only be
6007 enabled if running ji 6007 enabled if running jitter sensitive (HPC/RT) workloads.
6008 6008
6009 security= [SECURITY] Choose a l 6009 security= [SECURITY] Choose a legacy "major" security module to
6010 enable at boot. This 6010 enable at boot. This has been deprecated by the
6011 "lsm=" parameter. 6011 "lsm=" parameter.
6012 6012
6013 selinux= [SELINUX] Disable or 6013 selinux= [SELINUX] Disable or enable SELinux at boot time.
6014 Format: { "0" | "1" } 6014 Format: { "0" | "1" }
6015 See security/selinux/ 6015 See security/selinux/Kconfig help text.
6016 0 -- disable. 6016 0 -- disable.
6017 1 -- enable. 6017 1 -- enable.
6018 Default value is 1. 6018 Default value is 1.
6019 6019
6020 serialnumber [BUGS=X86-32] 6020 serialnumber [BUGS=X86-32]
6021 6021
6022 sev=option[,option...] [X86-64] See D 6022 sev=option[,option...] [X86-64] See Documentation/arch/x86/x86_64/boot-options.rst
6023 6023
6024 shapers= [NET] 6024 shapers= [NET]
6025 Maximal number of sha 6025 Maximal number of shapers.
6026 6026
6027 show_lapic= [APIC,X86] Advanced P 6027 show_lapic= [APIC,X86] Advanced Programmable Interrupt Controller
6028 Limit apic dumping. T 6028 Limit apic dumping. The parameter defines the maximal
6029 number of local apics 6029 number of local apics being dumped. Also it is possible
6030 to set it to "all" by 6030 to set it to "all" by meaning -- no limit here.
6031 Format: { 1 (default) 6031 Format: { 1 (default) | 2 | ... | all }.
6032 The parameter valid i 6032 The parameter valid if only apic=debug or
6033 apic=verbose is speci 6033 apic=verbose is specified.
6034 Example: apic=debug s 6034 Example: apic=debug show_lapic=all
6035 6035
6036 slab_debug[=options[,slabs][;[options 6036 slab_debug[=options[,slabs][;[options[,slabs]]...] [MM]
6037 Enabling slab_debug a 6037 Enabling slab_debug allows one to determine the
6038 culprit if slab objec 6038 culprit if slab objects become corrupted. Enabling
6039 slab_debug can create 6039 slab_debug can create guard zones around objects and
6040 may poison objects wh 6040 may poison objects when not in use. Also tracks the
6041 last alloc / free. Fo 6041 last alloc / free. For more information see
6042 Documentation/mm/slub 6042 Documentation/mm/slub.rst.
6043 (slub_debug legacy na 6043 (slub_debug legacy name also accepted for now)
6044 6044
6045 slab_max_order= [MM] 6045 slab_max_order= [MM]
6046 Determines the maximu 6046 Determines the maximum allowed order for slabs.
6047 A high setting may ca 6047 A high setting may cause OOMs due to memory
6048 fragmentation. For mo 6048 fragmentation. For more information see
6049 Documentation/mm/slub 6049 Documentation/mm/slub.rst.
6050 (slub_max_order legac 6050 (slub_max_order legacy name also accepted for now)
6051 6051
6052 slab_merge [MM] 6052 slab_merge [MM]
6053 Enable merging of sla 6053 Enable merging of slabs with similar size when the
6054 kernel is built witho 6054 kernel is built without CONFIG_SLAB_MERGE_DEFAULT.
6055 (slub_merge legacy na 6055 (slub_merge legacy name also accepted for now)
6056 6056
6057 slab_min_objects= [MM] 6057 slab_min_objects= [MM]
6058 The minimum number of 6058 The minimum number of objects per slab. SLUB will
6059 increase the slab ord 6059 increase the slab order up to slab_max_order to
6060 generate a sufficient 6060 generate a sufficiently large slab able to contain
6061 the number of objects 6061 the number of objects indicated. The higher the number
6062 of objects the smalle 6062 of objects the smaller the overhead of tracking slabs
6063 and the less frequent 6063 and the less frequently locks need to be acquired.
6064 For more information 6064 For more information see Documentation/mm/slub.rst.
6065 (slub_min_objects leg 6065 (slub_min_objects legacy name also accepted for now)
6066 6066
6067 slab_min_order= [MM] 6067 slab_min_order= [MM]
6068 Determines the minimu 6068 Determines the minimum page order for slabs. Must be
6069 lower or equal to sla 6069 lower or equal to slab_max_order. For more information see
6070 Documentation/mm/slub 6070 Documentation/mm/slub.rst.
6071 (slub_min_order legac 6071 (slub_min_order legacy name also accepted for now)
6072 6072
6073 slab_nomerge [MM] 6073 slab_nomerge [MM]
6074 Disable merging of sl 6074 Disable merging of slabs with similar size. May be
6075 necessary if there is 6075 necessary if there is some reason to distinguish
6076 allocs to different s 6076 allocs to different slabs, especially in hardened
6077 environments where th 6077 environments where the risk of heap overflows and
6078 layout control by att 6078 layout control by attackers can usually be
6079 frustrated by disabli 6079 frustrated by disabling merging. This will reduce
6080 most of the exposure 6080 most of the exposure of a heap attack to a single
6081 cache (risks via meta 6081 cache (risks via metadata attacks are mostly
6082 unchanged). Debug opt 6082 unchanged). Debug options disable merging on their
6083 own. 6083 own.
6084 For more information 6084 For more information see Documentation/mm/slub.rst.
6085 (slub_nomerge legacy 6085 (slub_nomerge legacy name also accepted for now)
6086 6086
6087 slram= [HW,MTD] 6087 slram= [HW,MTD]
6088 6088
6089 smart2= [HW] 6089 smart2= [HW]
6090 Format: <io1>[,<io2>[ 6090 Format: <io1>[,<io2>[,...,<io8>]]
6091 6091
6092 smp.csd_lock_timeout= [KNL] 6092 smp.csd_lock_timeout= [KNL]
6093 Specify the period of 6093 Specify the period of time in milliseconds
6094 that smp_call_functio 6094 that smp_call_function() and friends will wait
6095 for a CPU to release 6095 for a CPU to release the CSD lock. This is
6096 useful when diagnosin 6096 useful when diagnosing bugs involving CPUs
6097 disabling interrupts 6097 disabling interrupts for extended periods
6098 of time. Defaults to 6098 of time. Defaults to 5,000 milliseconds, and
6099 setting a value of ze 6099 setting a value of zero disables this feature.
6100 This feature may be m 6100 This feature may be more efficiently disabled
6101 using the csdlock_deb 6101 using the csdlock_debug- kernel parameter.
6102 6102
6103 smp.panic_on_ipistall= [KNL] 6103 smp.panic_on_ipistall= [KNL]
6104 If a csd_lock_timeout 6104 If a csd_lock_timeout extends for more than
6105 the specified number 6105 the specified number of milliseconds, panic the
6106 system. By default, 6106 system. By default, let CSD-lock acquisition
6107 take as long as they 6107 take as long as they take. Specifying 300,000
6108 for this value provid 6108 for this value provides a 5-minute timeout.
6109 6109
6110 smsc-ircc2.nopnp [HW] Don't us 6110 smsc-ircc2.nopnp [HW] Don't use PNP to discover SMC devices
6111 smsc-ircc2.ircc_cfg= [HW] Device c 6111 smsc-ircc2.ircc_cfg= [HW] Device configuration I/O port
6112 smsc-ircc2.ircc_sir= [HW] SIR base 6112 smsc-ircc2.ircc_sir= [HW] SIR base I/O port
6113 smsc-ircc2.ircc_fir= [HW] FIR base 6113 smsc-ircc2.ircc_fir= [HW] FIR base I/O port
6114 smsc-ircc2.ircc_irq= [HW] IRQ line 6114 smsc-ircc2.ircc_irq= [HW] IRQ line
6115 smsc-ircc2.ircc_dma= [HW] DMA chan 6115 smsc-ircc2.ircc_dma= [HW] DMA channel
6116 smsc-ircc2.ircc_transceiver= [HW] Tra 6116 smsc-ircc2.ircc_transceiver= [HW] Transceiver type:
6117 0: Toshiba Sa 6117 0: Toshiba Satellite 1800 (GP data pin select)
6118 1: Fast pin s 6118 1: Fast pin select (default)
6119 2: ATC IRMode 6119 2: ATC IRMode
6120 6120
6121 smt= [KNL,MIPS,S390,EARLY] 6121 smt= [KNL,MIPS,S390,EARLY] Set the maximum number of threads
6122 (logical CPUs) to use 6122 (logical CPUs) to use per physical CPU on systems
6123 capable of symmetric 6123 capable of symmetric multithreading (SMT). Will
6124 be capped to the actu 6124 be capped to the actual hardware limit.
6125 Format: <integer> 6125 Format: <integer>
6126 Default: -1 (no limit 6126 Default: -1 (no limit)
6127 6127
6128 softlockup_panic= 6128 softlockup_panic=
6129 [KNL] Should the soft 6129 [KNL] Should the soft-lockup detector generate panics.
6130 Format: 0 | 1 6130 Format: 0 | 1
6131 6131
6132 A value of 1 instruct 6132 A value of 1 instructs the soft-lockup detector
6133 to panic the machine 6133 to panic the machine when a soft-lockup occurs. It is
6134 also controlled by th 6134 also controlled by the kernel.softlockup_panic sysctl
6135 and CONFIG_BOOTPARAM_ 6135 and CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC, which is the
6136 respective build-time 6136 respective build-time switch to that functionality.
6137 6137
6138 softlockup_all_cpu_backtrace= 6138 softlockup_all_cpu_backtrace=
6139 [KNL] Should the soft 6139 [KNL] Should the soft-lockup detector generate
6140 backtraces on all cpu 6140 backtraces on all cpus.
6141 Format: 0 | 1 6141 Format: 0 | 1
6142 6142
6143 sonypi.*= [HW] Sony Programmabl 6143 sonypi.*= [HW] Sony Programmable I/O Control Device driver
6144 See Documentation/adm 6144 See Documentation/admin-guide/laptops/sonypi.rst
6145 6145
6146 spectre_bhi= [X86] Control mitigat 6146 spectre_bhi= [X86] Control mitigation of Branch History Injection
6147 (BHI) vulnerability. 6147 (BHI) vulnerability. This setting affects the
6148 deployment of the HW 6148 deployment of the HW BHI control and the SW BHB
6149 clearing sequence. 6149 clearing sequence.
6150 6150
6151 on - (default) En 6151 on - (default) Enable the HW or SW mitigation as
6152 needed. Thi 6152 needed. This protects the kernel from
6153 both syscall 6153 both syscalls and VMs.
6154 vmexit - On systems w 6154 vmexit - On systems which don't have the HW mitigation
6155 available, e 6155 available, enable the SW mitigation on vmexit
6156 ONLY. On su 6156 ONLY. On such systems, the host kernel is
6157 protected fr 6157 protected from VM-originated BHI attacks, but
6158 may still be 6158 may still be vulnerable to syscall attacks.
6159 off - Disable the 6159 off - Disable the mitigation.
6160 6160
6161 spectre_v2= [X86,EARLY] Control m 6161 spectre_v2= [X86,EARLY] Control mitigation of Spectre variant 2
6162 (indirect branch spec 6162 (indirect branch speculation) vulnerability.
6163 The default operation 6163 The default operation protects the kernel from
6164 user space attacks. 6164 user space attacks.
6165 6165
6166 on - unconditionall 6166 on - unconditionally enable, implies
6167 spectre_v2_use 6167 spectre_v2_user=on
6168 off - unconditionall 6168 off - unconditionally disable, implies
6169 spectre_v2_use 6169 spectre_v2_user=off
6170 auto - kernel detects 6170 auto - kernel detects whether your CPU model is
6171 vulnerable 6171 vulnerable
6172 6172
6173 Selecting 'on' will, 6173 Selecting 'on' will, and 'auto' may, choose a
6174 mitigation method at 6174 mitigation method at run time according to the
6175 CPU, the available mi 6175 CPU, the available microcode, the setting of the
6176 CONFIG_MITIGATION_RET 6176 CONFIG_MITIGATION_RETPOLINE configuration option,
6177 and the compiler with 6177 and the compiler with which the kernel was built.
6178 6178
6179 Selecting 'on' will a 6179 Selecting 'on' will also enable the mitigation
6180 against user space to 6180 against user space to user space task attacks.
6181 6181
6182 Selecting 'off' will 6182 Selecting 'off' will disable both the kernel and
6183 the user space protec 6183 the user space protections.
6184 6184
6185 Specific mitigations 6185 Specific mitigations can also be selected manually:
6186 6186
6187 retpoline - r 6187 retpoline - replace indirect branches
6188 retpoline,generic - R 6188 retpoline,generic - Retpolines
6189 retpoline,lfence - L 6189 retpoline,lfence - LFENCE; indirect branch
6190 retpoline,amd - a 6190 retpoline,amd - alias for retpoline,lfence
6191 eibrs - E 6191 eibrs - Enhanced/Auto IBRS
6192 eibrs,retpoline - E 6192 eibrs,retpoline - Enhanced/Auto IBRS + Retpolines
6193 eibrs,lfence - E 6193 eibrs,lfence - Enhanced/Auto IBRS + LFENCE
6194 ibrs - u 6194 ibrs - use IBRS to protect kernel
6195 6195
6196 Not specifying this o 6196 Not specifying this option is equivalent to
6197 spectre_v2=auto. 6197 spectre_v2=auto.
6198 6198
6199 spectre_v2_user= 6199 spectre_v2_user=
6200 [X86] Control mitigat 6200 [X86] Control mitigation of Spectre variant 2
6201 (indirect branch spec 6201 (indirect branch speculation) vulnerability between
6202 user space tasks 6202 user space tasks
6203 6203
6204 on - Uncondition 6204 on - Unconditionally enable mitigations. Is
6205 enforced by 6205 enforced by spectre_v2=on
6206 6206
6207 off - Uncondition 6207 off - Unconditionally disable mitigations. Is
6208 enforced by 6208 enforced by spectre_v2=off
6209 6209
6210 prctl - Indirect br 6210 prctl - Indirect branch speculation is enabled,
6211 but mitigat 6211 but mitigation can be enabled via prctl
6212 per thread. 6212 per thread. The mitigation control state
6213 is inherite 6213 is inherited on fork.
6214 6214
6215 prctl,ibpb 6215 prctl,ibpb
6216 - Like "prctl 6216 - Like "prctl" above, but only STIBP is
6217 controlled 6217 controlled per thread. IBPB is issued
6218 always when 6218 always when switching between different user
6219 space proce 6219 space processes.
6220 6220
6221 seccomp 6221 seccomp
6222 - Same as "pr 6222 - Same as "prctl" above, but all seccomp
6223 threads wil 6223 threads will enable the mitigation unless
6224 they explic 6224 they explicitly opt out.
6225 6225
6226 seccomp,ibpb 6226 seccomp,ibpb
6227 - Like "secco 6227 - Like "seccomp" above, but only STIBP is
6228 controlled 6228 controlled per thread. IBPB is issued
6229 always when 6229 always when switching between different
6230 user space 6230 user space processes.
6231 6231
6232 auto - Kernel sele 6232 auto - Kernel selects the mitigation depending on
6233 the availab 6233 the available CPU features and vulnerability.
6234 6234
6235 Default mitigation: " 6235 Default mitigation: "prctl"
6236 6236
6237 Not specifying this o 6237 Not specifying this option is equivalent to
6238 spectre_v2_user=auto. 6238 spectre_v2_user=auto.
6239 6239
6240 spec_rstack_overflow= 6240 spec_rstack_overflow=
6241 [X86,EARLY] Control R 6241 [X86,EARLY] Control RAS overflow mitigation on AMD Zen CPUs
6242 6242
6243 off - Dis 6243 off - Disable mitigation
6244 microcode - Ena 6244 microcode - Enable microcode mitigation only
6245 safe-ret - Ena 6245 safe-ret - Enable sw-only safe RET mitigation (default)
6246 ibpb - Ena 6246 ibpb - Enable mitigation by issuing IBPB on
6247 ker 6247 kernel entry
6248 ibpb-vmexit - Iss 6248 ibpb-vmexit - Issue IBPB only on VMEXIT
6249 (cl 6249 (cloud-specific mitigation)
6250 6250
6251 spec_store_bypass_disable= 6251 spec_store_bypass_disable=
6252 [HW,EARLY] Control Sp 6252 [HW,EARLY] Control Speculative Store Bypass (SSB) Disable mitigation
6253 (Speculative Store By 6253 (Speculative Store Bypass vulnerability)
6254 6254
6255 Certain CPUs are vuln 6255 Certain CPUs are vulnerable to an exploit against a
6256 a common industry wid 6256 a common industry wide performance optimization known
6257 as "Speculative Store 6257 as "Speculative Store Bypass" in which recent stores
6258 to the same memory lo 6258 to the same memory location may not be observed by
6259 later loads during sp 6259 later loads during speculative execution. The idea
6260 is that such stores a 6260 is that such stores are unlikely and that they can
6261 be detected prior to 6261 be detected prior to instruction retirement at the
6262 end of a particular s 6262 end of a particular speculation execution window.
6263 6263
6264 In vulnerable process 6264 In vulnerable processors, the speculatively forwarded
6265 store can be used in 6265 store can be used in a cache side channel attack, for
6266 example to read memor 6266 example to read memory to which the attacker does not
6267 directly have access 6267 directly have access (e.g. inside sandboxed code).
6268 6268
6269 This parameter contro 6269 This parameter controls whether the Speculative Store
6270 Bypass optimization i 6270 Bypass optimization is used.
6271 6271
6272 On x86 the options ar 6272 On x86 the options are:
6273 6273
6274 on - Uncondition 6274 on - Unconditionally disable Speculative Store Bypass
6275 off - Uncondition 6275 off - Unconditionally enable Speculative Store Bypass
6276 auto - Kernel dete 6276 auto - Kernel detects whether the CPU model contains an
6277 implementat 6277 implementation of Speculative Store Bypass and
6278 picks the m 6278 picks the most appropriate mitigation. If the
6279 CPU is not 6279 CPU is not vulnerable, "off" is selected. If the
6280 CPU is vuln 6280 CPU is vulnerable the default mitigation is
6281 architectur 6281 architecture and Kconfig dependent. See below.
6282 prctl - Control Spe 6282 prctl - Control Speculative Store Bypass per thread
6283 via prctl. 6283 via prctl. Speculative Store Bypass is enabled
6284 for a proce 6284 for a process by default. The state of the control
6285 is inherite 6285 is inherited on fork.
6286 seccomp - Same as "pr 6286 seccomp - Same as "prctl" above, but all seccomp threads
6287 will disabl 6287 will disable SSB unless they explicitly opt out.
6288 6288
6289 Default mitigations: 6289 Default mitigations:
6290 X86: "prctl" 6290 X86: "prctl"
6291 6291
6292 On powerpc the option 6292 On powerpc the options are:
6293 6293
6294 on,auto - On Power8 a 6294 on,auto - On Power8 and Power9 insert a store-forwarding
6295 barrier on 6295 barrier on kernel entry and exit. On Power7
6296 perform a s 6296 perform a software flush on kernel entry and
6297 exit. 6297 exit.
6298 off - No action. 6298 off - No action.
6299 6299
6300 Not specifying this o 6300 Not specifying this option is equivalent to
6301 spec_store_bypass_dis 6301 spec_store_bypass_disable=auto.
6302 6302
6303 split_lock_detect= 6303 split_lock_detect=
6304 [X86] Enable split lo 6304 [X86] Enable split lock detection or bus lock detection
6305 6305
6306 When enabled (and if 6306 When enabled (and if hardware support is present), atomic
6307 instructions that acc 6307 instructions that access data across cache line
6308 boundaries will resul 6308 boundaries will result in an alignment check exception
6309 for split lock detect 6309 for split lock detection or a debug exception for
6310 bus lock detection. 6310 bus lock detection.
6311 6311
6312 off - not enabled 6312 off - not enabled
6313 6313
6314 warn - the kernel 6314 warn - the kernel will emit rate-limited warnings
6315 about appli 6315 about applications triggering the #AC
6316 exception o 6316 exception or the #DB exception. This mode is
6317 the default 6317 the default on CPUs that support split lock
6318 detection o 6318 detection or bus lock detection. Default
6319 behavior is 6319 behavior is by #AC if both features are
6320 enabled in 6320 enabled in hardware.
6321 6321
6322 fatal - the kernel 6322 fatal - the kernel will send SIGBUS to applications
6323 that trigge 6323 that trigger the #AC exception or the #DB
6324 exception. 6324 exception. Default behavior is by #AC if
6325 both featur 6325 both features are enabled in hardware.
6326 6326
6327 ratelimit:N - 6327 ratelimit:N -
6328 Set system 6328 Set system wide rate limit to N bus locks
6329 per second 6329 per second for bus lock detection.
6330 0 < N <= 10 6330 0 < N <= 1000.
6331 6331
6332 N/A for spl 6332 N/A for split lock detection.
6333 6333
6334 6334
6335 If an #AC exception i 6335 If an #AC exception is hit in the kernel or in
6336 firmware (i.e. not wh 6336 firmware (i.e. not while executing in user mode)
6337 the kernel will oops 6337 the kernel will oops in either "warn" or "fatal"
6338 mode. 6338 mode.
6339 6339
6340 #DB exception for bus 6340 #DB exception for bus lock is triggered only when
6341 CPL > 0. 6341 CPL > 0.
6342 6342
6343 srbds= [X86,INTEL,EARLY] 6343 srbds= [X86,INTEL,EARLY]
6344 Control the Special R 6344 Control the Special Register Buffer Data Sampling
6345 (SRBDS) mitigation. 6345 (SRBDS) mitigation.
6346 6346
6347 Certain CPUs are vuln 6347 Certain CPUs are vulnerable to an MDS-like
6348 exploit which can lea 6348 exploit which can leak bits from the random
6349 number generator. 6349 number generator.
6350 6350
6351 By default, this issu 6351 By default, this issue is mitigated by
6352 microcode. However, 6352 microcode. However, the microcode fix can cause
6353 the RDRAND and RDSEED 6353 the RDRAND and RDSEED instructions to become
6354 much slower. Among o 6354 much slower. Among other effects, this will
6355 result in reduced thr 6355 result in reduced throughput from /dev/urandom.
6356 6356
6357 The microcode mitigat 6357 The microcode mitigation can be disabled with
6358 the following option: 6358 the following option:
6359 6359
6360 off: Disable mitig 6360 off: Disable mitigation and remove
6361 performance i 6361 performance impact to RDRAND and RDSEED
6362 6362
6363 srcutree.big_cpu_lim [KNL] 6363 srcutree.big_cpu_lim [KNL]
6364 Specifies the number 6364 Specifies the number of CPUs constituting a
6365 large system, such th 6365 large system, such that srcu_struct structures
6366 should immediately al 6366 should immediately allocate an srcu_node array.
6367 This kernel-boot para 6367 This kernel-boot parameter defaults to 128,
6368 but takes effect only 6368 but takes effect only when the low-order four
6369 bits of srcutree.conv 6369 bits of srcutree.convert_to_big is equal to 3
6370 (decide at boot). 6370 (decide at boot).
6371 6371
6372 srcutree.convert_to_big [KNL] 6372 srcutree.convert_to_big [KNL]
6373 Specifies under what 6373 Specifies under what conditions an SRCU tree
6374 srcu_struct structure 6374 srcu_struct structure will be converted to big
6375 form, that is, with a 6375 form, that is, with an rcu_node tree:
6376 6376
6377 0: Never. 6377 0: Never.
6378 1: At ini 6378 1: At init_srcu_struct() time.
6379 2: When r 6379 2: When rcutorture decides to.
6380 3: Decide 6380 3: Decide at boot time (default).
6381 0x1X: Above 6381 0x1X: Above plus if high contention.
6382 6382
6383 Either way, the srcu_ 6383 Either way, the srcu_node tree will be sized based
6384 on the actual runtime 6384 on the actual runtime number of CPUs (nr_cpu_ids)
6385 instead of the compil 6385 instead of the compile-time CONFIG_NR_CPUS.
6386 6386
6387 srcutree.counter_wrap_check [KNL] 6387 srcutree.counter_wrap_check [KNL]
6388 Specifies how frequen 6388 Specifies how frequently to check for
6389 grace-period sequence 6389 grace-period sequence counter wrap for the
6390 srcu_data structure's 6390 srcu_data structure's ->srcu_gp_seq_needed field.
6391 The greater the numbe 6391 The greater the number of bits set in this kernel
6392 parameter, the less f 6392 parameter, the less frequently counter wrap will
6393 be checked for. Note 6393 be checked for. Note that the bottom two bits
6394 are ignored. 6394 are ignored.
6395 6395
6396 srcutree.exp_holdoff [KNL] 6396 srcutree.exp_holdoff [KNL]
6397 Specifies how many na 6397 Specifies how many nanoseconds must elapse
6398 since the end of the 6398 since the end of the last SRCU grace period for
6399 a given srcu_struct u 6399 a given srcu_struct until the next normal SRCU
6400 grace period will be 6400 grace period will be considered for automatic
6401 expediting. Set to z 6401 expediting. Set to zero to disable automatic
6402 expediting. 6402 expediting.
6403 6403
6404 srcutree.srcu_max_nodelay [KNL] 6404 srcutree.srcu_max_nodelay [KNL]
6405 Specifies the number 6405 Specifies the number of no-delay instances
6406 per jiffy for which t 6406 per jiffy for which the SRCU grace period
6407 worker thread will be 6407 worker thread will be rescheduled with zero
6408 delay. Beyond this li 6408 delay. Beyond this limit, worker thread will
6409 be rescheduled with a 6409 be rescheduled with a sleep delay of one jiffy.
6410 6410
6411 srcutree.srcu_max_nodelay_phase [KNL] 6411 srcutree.srcu_max_nodelay_phase [KNL]
6412 Specifies the per-gra 6412 Specifies the per-grace-period phase, number of
6413 non-sleeping polls of 6413 non-sleeping polls of readers. Beyond this limit,
6414 grace period worker t 6414 grace period worker thread will be rescheduled
6415 with a sleep delay of 6415 with a sleep delay of one jiffy, between each
6416 rescan of the readers 6416 rescan of the readers, for a grace period phase.
6417 6417
6418 srcutree.srcu_retry_check_delay [KNL] 6418 srcutree.srcu_retry_check_delay [KNL]
6419 Specifies number of m 6419 Specifies number of microseconds of non-sleeping
6420 delay between each no 6420 delay between each non-sleeping poll of readers.
6421 6421
6422 srcutree.small_contention_lim [KNL] 6422 srcutree.small_contention_lim [KNL]
6423 Specifies the number 6423 Specifies the number of update-side contention
6424 events per jiffy will 6424 events per jiffy will be tolerated before
6425 initiating a conversi 6425 initiating a conversion of an srcu_struct
6426 structure to big form 6426 structure to big form. Note that the value of
6427 srcutree.convert_to_b 6427 srcutree.convert_to_big must have the 0x10 bit
6428 set for contention-ba 6428 set for contention-based conversions to occur.
6429 6429
6430 ssbd= [ARM64,HW,EARLY] 6430 ssbd= [ARM64,HW,EARLY]
6431 Speculative Store Byp 6431 Speculative Store Bypass Disable control
6432 6432
6433 On CPUs that are vuln 6433 On CPUs that are vulnerable to the Speculative
6434 Store Bypass vulnerab 6434 Store Bypass vulnerability and offer a
6435 firmware based mitiga 6435 firmware based mitigation, this parameter
6436 indicates how the mit 6436 indicates how the mitigation should be used:
6437 6437
6438 force-on: Unconditio 6438 force-on: Unconditionally enable mitigation for
6439 for both k 6439 for both kernel and userspace
6440 force-off: Unconditio 6440 force-off: Unconditionally disable mitigation for
6441 for both k 6441 for both kernel and userspace
6442 kernel: Always ena 6442 kernel: Always enable mitigation in the
6443 kernel, an 6443 kernel, and offer a prctl interface
6444 to allow u 6444 to allow userspace to register its
6445 interest i 6445 interest in being mitigated too.
6446 6446
6447 stack_guard_gap= [MM] 6447 stack_guard_gap= [MM]
6448 override the default 6448 override the default stack gap protection. The value
6449 is in page units and 6449 is in page units and it defines how many pages prior
6450 to (for stacks growin 6450 to (for stacks growing down) resp. after (for stacks
6451 growing up) the main 6451 growing up) the main stack are reserved for no other
6452 mapping. Default valu 6452 mapping. Default value is 256 pages.
6453 6453
6454 stack_depot_disable= [KNL,EARLY] 6454 stack_depot_disable= [KNL,EARLY]
6455 Setting this to true 6455 Setting this to true through kernel command line will
6456 disable the stack dep 6456 disable the stack depot thereby saving the static memory
6457 consumed by the stack 6457 consumed by the stack hash table. By default this is set
6458 to false. 6458 to false.
6459 6459
6460 stacktrace [FTRACE] 6460 stacktrace [FTRACE]
6461 Enabled the stack tra 6461 Enabled the stack tracer on boot up.
6462 6462
6463 stacktrace_filter=[function-list] 6463 stacktrace_filter=[function-list]
6464 [FTRACE] Limit the fu 6464 [FTRACE] Limit the functions that the stack tracer
6465 will trace at boot up 6465 will trace at boot up. function-list is a comma-separated
6466 list of functions. Th 6466 list of functions. This list can be changed at run
6467 time by the stack_tra 6467 time by the stack_trace_filter file in the debugfs
6468 tracing directory. No 6468 tracing directory. Note, this enables stack tracing
6469 and the stacktrace ab 6469 and the stacktrace above is not needed.
6470 6470
6471 sti= [PARISC,HW] 6471 sti= [PARISC,HW]
6472 Format: <num> 6472 Format: <num>
6473 Set the STI (builtin 6473 Set the STI (builtin display/keyboard on the HP-PARISC
6474 machines) console (gr 6474 machines) console (graphic card) which should be used
6475 as the initial boot-c 6475 as the initial boot-console.
6476 See also comment in d 6476 See also comment in drivers/video/console/sticore.c.
6477 6477
6478 sti_font= [HW] 6478 sti_font= [HW]
6479 See comment in driver 6479 See comment in drivers/video/console/sticore.c.
6480 6480
6481 stifb= [HW] 6481 stifb= [HW]
6482 Format: bpp:<bpp1>[:< 6482 Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
6483 6483
6484 strict_sas_size= 6484 strict_sas_size=
6485 [X86] 6485 [X86]
6486 Format: <bool> 6486 Format: <bool>
6487 Enable or disable str 6487 Enable or disable strict sigaltstack size checks
6488 against the required 6488 against the required signal frame size which
6489 depends on the suppor 6489 depends on the supported FPU features. This can
6490 be used to filter out 6490 be used to filter out binaries which have
6491 not yet been made awa 6491 not yet been made aware of AT_MINSIGSTKSZ.
6492 6492
6493 stress_hpt [PPC,EARLY] 6493 stress_hpt [PPC,EARLY]
6494 Limits the number of 6494 Limits the number of kernel HPT entries in the hash
6495 page table to increas 6495 page table to increase the rate of hash page table
6496 faults on kernel addr 6496 faults on kernel addresses.
6497 6497
6498 stress_slb [PPC,EARLY] 6498 stress_slb [PPC,EARLY]
6499 Limits the number of 6499 Limits the number of kernel SLB entries, and flushes
6500 them frequently to in 6500 them frequently to increase the rate of SLB faults
6501 on kernel addresses. 6501 on kernel addresses.
6502 6502
6503 sunrpc.min_resvport= 6503 sunrpc.min_resvport=
6504 sunrpc.max_resvport= 6504 sunrpc.max_resvport=
6505 [NFS,SUNRPC] 6505 [NFS,SUNRPC]
6506 SunRPC servers often 6506 SunRPC servers often require that client requests
6507 originate from a priv 6507 originate from a privileged port (i.e. a port in the
6508 range 0 < portnr < 10 6508 range 0 < portnr < 1024).
6509 An administrator who 6509 An administrator who wishes to reserve some of these
6510 ports for other uses 6510 ports for other uses may adjust the range that the
6511 kernel's sunrpc clien 6511 kernel's sunrpc client considers to be privileged
6512 using these two param 6512 using these two parameters to set the minimum and
6513 maximum port values. 6513 maximum port values.
6514 6514
6515 sunrpc.svc_rpc_per_connection_limit= 6515 sunrpc.svc_rpc_per_connection_limit=
6516 [NFS,SUNRPC] 6516 [NFS,SUNRPC]
6517 Limit the number of r 6517 Limit the number of requests that the server will
6518 process in parallel f 6518 process in parallel from a single connection.
6519 The default value is 6519 The default value is 0 (no limit).
6520 6520
6521 sunrpc.pool_mode= 6521 sunrpc.pool_mode=
6522 [NFS] 6522 [NFS]
6523 Control how the NFS s 6523 Control how the NFS server code allocates CPUs to
6524 service thread pools. 6524 service thread pools. Depending on how many NICs
6525 you have and where th 6525 you have and where their interrupts are bound, this
6526 option will affect wh 6526 option will affect which CPUs will do NFS serving.
6527 Note: this parameter 6527 Note: this parameter cannot be changed while the
6528 NFS server is running 6528 NFS server is running.
6529 6529
6530 auto the serve 6530 auto the server chooses an appropriate mode
6531 automatic 6531 automatically using heuristics
6532 global a single 6532 global a single global pool contains all CPUs
6533 percpu one pool 6533 percpu one pool for each CPU
6534 pernode one pool 6534 pernode one pool for each NUMA node (equivalent
6535 to global 6535 to global on non-NUMA machines)
6536 6536
6537 sunrpc.tcp_slot_table_entries= 6537 sunrpc.tcp_slot_table_entries=
6538 sunrpc.udp_slot_table_entries= 6538 sunrpc.udp_slot_table_entries=
6539 [NFS,SUNRPC] 6539 [NFS,SUNRPC]
6540 Sets the upper limit 6540 Sets the upper limit on the number of simultaneous
6541 RPC calls that can be 6541 RPC calls that can be sent from the client to a
6542 server. Increasing th 6542 server. Increasing these values may allow you to
6543 improve throughput, b 6543 improve throughput, but will also increase the
6544 amount of memory rese 6544 amount of memory reserved for use by the client.
6545 6545
6546 suspend.pm_test_delay= 6546 suspend.pm_test_delay=
6547 [SUSPEND] 6547 [SUSPEND]
6548 Sets the number of se 6548 Sets the number of seconds to remain in a suspend test
6549 mode before resuming 6549 mode before resuming the system (see
6550 /sys/power/pm_test). 6550 /sys/power/pm_test). Only available when CONFIG_PM_DEBUG
6551 is set. Default value 6551 is set. Default value is 5.
6552 6552
6553 svm= [PPC] 6553 svm= [PPC]
6554 Format: { on | off | 6554 Format: { on | off | y | n | 1 | 0 }
6555 This parameter contro 6555 This parameter controls use of the Protected
6556 Execution Facility on 6556 Execution Facility on pSeries.
6557 6557
6558 swiotlb= [ARM,PPC,MIPS,X86,S39 6558 swiotlb= [ARM,PPC,MIPS,X86,S390,EARLY]
6559 Format: { <int> [,<in 6559 Format: { <int> [,<int>] | force | noforce }
6560 <int> -- Number of I/ 6560 <int> -- Number of I/O TLB slabs
6561 <int> -- Second integ 6561 <int> -- Second integer after comma. Number of swiotlb
6562 areas with t 6562 areas with their own lock. Will be rounded up
6563 to a power o 6563 to a power of 2.
6564 force -- force using 6564 force -- force using of bounce buffers even if they
6565 wouldn't be 6565 wouldn't be automatically used by the kernel
6566 noforce -- Never use 6566 noforce -- Never use bounce buffers (for debugging)
6567 6567
6568 switches= [HW,M68k,EARLY] 6568 switches= [HW,M68k,EARLY]
6569 6569
6570 sysctl.*= [KNL] 6570 sysctl.*= [KNL]
6571 Set a sysctl paramete 6571 Set a sysctl parameter, right before loading the init
6572 process, as if the va 6572 process, as if the value was written to the respective
6573 /proc/sys/... file. B 6573 /proc/sys/... file. Both '.' and '/' are recognized as
6574 separators. Unrecogni 6574 separators. Unrecognized parameters and invalid values
6575 are reported in the k 6575 are reported in the kernel log. Sysctls registered
6576 later by a loaded mod 6576 later by a loaded module cannot be set this way.
6577 Example: sysctl.vm.sw 6577 Example: sysctl.vm.swappiness=40
6578 6578
6579 sysrq_always_enabled 6579 sysrq_always_enabled
6580 [KNL] 6580 [KNL]
6581 Ignore sysrq setting 6581 Ignore sysrq setting - this boot parameter will
6582 neutralize any effect 6582 neutralize any effect of /proc/sys/kernel/sysrq.
6583 Useful for debugging. 6583 Useful for debugging.
6584 6584
6585 tcpmhash_entries= [KNL,NET] 6585 tcpmhash_entries= [KNL,NET]
6586 Set the number of tcp 6586 Set the number of tcp_metrics_hash slots.
6587 Default value is 8192 6587 Default value is 8192 or 16384 depending on total
6588 ram pages. This is us 6588 ram pages. This is used to specify the TCP metrics
6589 cache size. See Docum 6589 cache size. See Documentation/networking/ip-sysctl.rst
6590 "tcp_no_metrics_save" 6590 "tcp_no_metrics_save" section for more details.
6591 6591
6592 tdfx= [HW,DRM] 6592 tdfx= [HW,DRM]
6593 6593
6594 test_suspend= [SUSPEND] 6594 test_suspend= [SUSPEND]
6595 Format: { "mem" | "st 6595 Format: { "mem" | "standby" | "freeze" }[,N]
6596 Specify "mem" (for Su 6596 Specify "mem" (for Suspend-to-RAM) or "standby" (for
6597 standby suspend) or " 6597 standby suspend) or "freeze" (for suspend type freeze)
6598 as the system sleep s 6598 as the system sleep state during system startup with
6599 the optional capabili 6599 the optional capability to repeat N number of times.
6600 The system is woken f 6600 The system is woken from this state using a
6601 wakeup-capable RTC al 6601 wakeup-capable RTC alarm.
6602 6602
6603 thash_entries= [KNL,NET] 6603 thash_entries= [KNL,NET]
6604 Set number of hash bu 6604 Set number of hash buckets for TCP connection
6605 6605
6606 thermal.act= [HW,ACPI] 6606 thermal.act= [HW,ACPI]
6607 -1: disable all activ 6607 -1: disable all active trip points in all thermal zones
6608 <degrees C>: override 6608 <degrees C>: override all lowest active trip points
6609 6609
6610 thermal.crt= [HW,ACPI] 6610 thermal.crt= [HW,ACPI]
6611 -1: disable all criti 6611 -1: disable all critical trip points in all thermal zones
6612 <degrees C>: override 6612 <degrees C>: override all critical trip points
6613 6613
6614 thermal.off= [HW,ACPI] 6614 thermal.off= [HW,ACPI]
6615 1: disable ACPI therm 6615 1: disable ACPI thermal control
6616 6616
6617 thermal.psv= [HW,ACPI] 6617 thermal.psv= [HW,ACPI]
6618 -1: disable all passi 6618 -1: disable all passive trip points
6619 <degrees C>: override 6619 <degrees C>: override all passive trip points to this
6620 value 6620 value
6621 6621
6622 thermal.tzp= [HW,ACPI] 6622 thermal.tzp= [HW,ACPI]
6623 Specify global defaul 6623 Specify global default ACPI thermal zone polling rate
6624 <deci-seconds>: poll 6624 <deci-seconds>: poll all this frequency
6625 0: no polling (defaul 6625 0: no polling (default)
6626 6626
6627 threadirqs [KNL,EARLY] 6627 threadirqs [KNL,EARLY]
6628 Force threading of al 6628 Force threading of all interrupt handlers except those
6629 marked explicitly IRQ 6629 marked explicitly IRQF_NO_THREAD.
6630 6630
6631 topology= [S390,EARLY] 6631 topology= [S390,EARLY]
6632 Format: {off | on} 6632 Format: {off | on}
6633 Specify if the kernel 6633 Specify if the kernel should make use of the cpu
6634 topology information 6634 topology information if the hardware supports this.
6635 The scheduler will ma 6635 The scheduler will make use of this information and
6636 e.g. base its process 6636 e.g. base its process migration decisions on it.
6637 Default is on. 6637 Default is on.
6638 6638
6639 torture.disable_onoff_at_boot= [KNL] 6639 torture.disable_onoff_at_boot= [KNL]
6640 Prevent the CPU-hotpl 6640 Prevent the CPU-hotplug component of torturing
6641 until after init has 6641 until after init has spawned.
6642 6642
6643 torture.ftrace_dump_at_shutdown= [KNL 6643 torture.ftrace_dump_at_shutdown= [KNL]
6644 Dump the ftrace buffe 6644 Dump the ftrace buffer at torture-test shutdown,
6645 even if there were no 6645 even if there were no errors. This can be a
6646 very costly operation 6646 very costly operation when many torture tests
6647 are running concurren 6647 are running concurrently, especially on systems
6648 with rotating-rust st 6648 with rotating-rust storage.
6649 6649
6650 torture.verbose_sleep_frequency= [KNL 6650 torture.verbose_sleep_frequency= [KNL]
6651 Specifies how many ve 6651 Specifies how many verbose printk()s should be
6652 emitted between each 6652 emitted between each sleep. The default of zero
6653 disables verbose-prin 6653 disables verbose-printk() sleeping.
6654 6654
6655 torture.verbose_sleep_duration= [KNL] 6655 torture.verbose_sleep_duration= [KNL]
6656 Duration of each verb 6656 Duration of each verbose-printk() sleep in jiffies.
6657 6657
6658 tpm_suspend_pcr=[HW,TPM] 6658 tpm_suspend_pcr=[HW,TPM]
6659 Format: integer pcr i 6659 Format: integer pcr id
6660 Specify that at suspe 6660 Specify that at suspend time, the tpm driver
6661 should extend the spe 6661 should extend the specified pcr with zeros,
6662 as a workaround for s 6662 as a workaround for some chips which fail to
6663 flush the last writte 6663 flush the last written pcr on TPM_SaveState.
6664 This will guarantee t 6664 This will guarantee that all the other pcrs
6665 are saved. 6665 are saved.
6666 6666
6667 tpm_tis.interrupts= [HW,TPM] 6667 tpm_tis.interrupts= [HW,TPM]
6668 Enable interrupts for 6668 Enable interrupts for the MMIO based physical layer
6669 for the FIFO interfac 6669 for the FIFO interface. By default it is set to false
6670 (0). For more informa 6670 (0). For more information about TPM hardware interfaces
6671 defined by Trusted Co 6671 defined by Trusted Computing Group (TCG) see
6672 https://trustedcomput 6672 https://trustedcomputinggroup.org/resource/pc-client-platform-tpm-profile-ptp-specification/
6673 6673
6674 tp_printk [FTRACE] 6674 tp_printk [FTRACE]
6675 Have the tracepoints 6675 Have the tracepoints sent to printk as well as the
6676 tracing ring buffer. 6676 tracing ring buffer. This is useful for early boot up
6677 where the system hang 6677 where the system hangs or reboots and does not give the
6678 option for reading th 6678 option for reading the tracing buffer or performing a
6679 ftrace_dump_on_oops. 6679 ftrace_dump_on_oops.
6680 6680
6681 To turn off having tr 6681 To turn off having tracepoints sent to printk,
6682 echo 0 > /proc/sys/k 6682 echo 0 > /proc/sys/kernel/tracepoint_printk
6683 Note, echoing 1 into 6683 Note, echoing 1 into this file without the
6684 tp_printk kernel cmdl 6684 tp_printk kernel cmdline option has no effect.
6685 6685
6686 The tp_printk_stop_on 6686 The tp_printk_stop_on_boot (see below) can also be used
6687 to stop the printing 6687 to stop the printing of events to console at
6688 late_initcall_sync. 6688 late_initcall_sync.
6689 6689
6690 ** CAUTION ** 6690 ** CAUTION **
6691 6691
6692 Having tracepoints se 6692 Having tracepoints sent to printk() and activating high
6693 frequency tracepoints 6693 frequency tracepoints such as irq or sched, can cause
6694 the system to live lo 6694 the system to live lock.
6695 6695
6696 tp_printk_stop_on_boot [FTRACE] 6696 tp_printk_stop_on_boot [FTRACE]
6697 When tp_printk (above 6697 When tp_printk (above) is set, it can cause a lot of noise
6698 on the console. It ma 6698 on the console. It may be useful to only include the
6699 printing of events du 6699 printing of events during boot up, as user space may
6700 make the system inope 6700 make the system inoperable.
6701 6701
6702 This command line opt 6702 This command line option will stop the printing of events
6703 to console at the lat 6703 to console at the late_initcall_sync() time frame.
6704 6704
6705 trace_buf_size=nn[KMG] 6705 trace_buf_size=nn[KMG]
6706 [FTRACE] will set tra 6706 [FTRACE] will set tracing buffer size on each cpu.
6707 6707
6708 trace_clock= [FTRACE] Set the cloc 6708 trace_clock= [FTRACE] Set the clock used for tracing events
6709 at boot up. 6709 at boot up.
6710 local - Use the per C 6710 local - Use the per CPU time stamp counter
6711 (converted in 6711 (converted into nanoseconds). Fast, but
6712 depending on 6712 depending on the architecture, may not be
6713 in sync betwe 6713 in sync between CPUs.
6714 global - Event time s 6714 global - Event time stamps are synchronize across
6715 CPUs. May be 6715 CPUs. May be slower than the local clock,
6716 but better fo 6716 but better for some race conditions.
6717 counter - Simple coun 6717 counter - Simple counting of events (1, 2, ..)
6718 note, some co 6718 note, some counts may be skipped due to the
6719 infrastructur 6719 infrastructure grabbing the clock more than
6720 once per even 6720 once per event.
6721 uptime - Use jiffies 6721 uptime - Use jiffies as the time stamp.
6722 perf - Use the same c 6722 perf - Use the same clock that perf uses.
6723 mono - Use ktime_get_ 6723 mono - Use ktime_get_mono_fast_ns() for time stamps.
6724 mono_raw - Use ktime_ 6724 mono_raw - Use ktime_get_raw_fast_ns() for time
6725 stamps. 6725 stamps.
6726 boot - Use ktime_get_ 6726 boot - Use ktime_get_boot_fast_ns() for time stamps.
6727 Architectures may add 6727 Architectures may add more clocks. See
6728 Documentation/trace/f 6728 Documentation/trace/ftrace.rst for more details.
6729 6729
6730 trace_event=[event-list] 6730 trace_event=[event-list]
6731 [FTRACE] Set and star 6731 [FTRACE] Set and start specified trace events in order
6732 to facilitate early b 6732 to facilitate early boot debugging. The event-list is a
6733 comma-separated list 6733 comma-separated list of trace events to enable. See
6734 also Documentation/tr 6734 also Documentation/trace/events.rst
6735 6735
6736 trace_instance=[instance-info] 6736 trace_instance=[instance-info]
6737 [FTRACE] Create a rin 6737 [FTRACE] Create a ring buffer instance early in boot up.
6738 This will be listed i 6738 This will be listed in:
6739 6739
6740 /sys/kernel/t 6740 /sys/kernel/tracing/instances
6741 6741
6742 Events can be enabled 6742 Events can be enabled at the time the instance is created
6743 via: 6743 via:
6744 6744
6745 trace_instanc 6745 trace_instance=<name>,<system1>:<event1>,<system2>:<event2>
6746 6746
6747 Note, the "<system*>: 6747 Note, the "<system*>:" portion is optional if the event is
6748 unique. 6748 unique.
6749 6749
6750 trace_instanc 6750 trace_instance=foo,sched:sched_switch,irq_handler_entry,initcall
6751 6751
6752 will enable the "sche 6752 will enable the "sched_switch" event (note, the "sched:" is optional, and
6753 the same thing would 6753 the same thing would happen if it was left off). The irq_handler_entry
6754 event, and all events 6754 event, and all events under the "initcall" system.
6755 6755
6756 trace_options=[option-list] 6756 trace_options=[option-list]
6757 [FTRACE] Enable or di 6757 [FTRACE] Enable or disable tracer options at boot.
6758 The option-list is a 6758 The option-list is a comma delimited list of options
6759 that can be enabled o 6759 that can be enabled or disabled just as if you were
6760 to echo the option na 6760 to echo the option name into
6761 6761
6762 /sys/kernel/traci 6762 /sys/kernel/tracing/trace_options
6763 6763
6764 For example, to enabl 6764 For example, to enable stacktrace option (to dump the
6765 stack trace of each e 6765 stack trace of each event), add to the command line:
6766 6766
6767 trace_options=s 6767 trace_options=stacktrace
6768 6768
6769 See also Documentatio 6769 See also Documentation/trace/ftrace.rst "trace options"
6770 section. 6770 section.
6771 6771
6772 trace_trigger=[trigger-list] 6772 trace_trigger=[trigger-list]
6773 [FTRACE] Add a event 6773 [FTRACE] Add a event trigger on specific events.
6774 Set a trigger on top 6774 Set a trigger on top of a specific event, with an optional
6775 filter. 6775 filter.
6776 6776
6777 The format is is "tra 6777 The format is is "trace_trigger=<event>.<trigger>[ if <filter>],..."
6778 Where more than one t 6778 Where more than one trigger may be specified that are comma deliminated.
6779 6779
6780 For example: 6780 For example:
6781 6781
6782 trace_trigger="sche 6782 trace_trigger="sched_switch.stacktrace if prev_state == 2"
6783 6783
6784 The above will enable 6784 The above will enable the "stacktrace" trigger on the "sched_switch"
6785 event but only trigge 6785 event but only trigger it if the "prev_state" of the "sched_switch"
6786 event is "2" (TASK_UN 6786 event is "2" (TASK_UNINTERUPTIBLE).
6787 6787
6788 See also "Event trigg 6788 See also "Event triggers" in Documentation/trace/events.rst
6789 6789
6790 6790
6791 traceoff_on_warning 6791 traceoff_on_warning
6792 [FTRACE] enable this 6792 [FTRACE] enable this option to disable tracing when a
6793 warning is hit. This 6793 warning is hit. This turns off "tracing_on". Tracing can
6794 be enabled again by e 6794 be enabled again by echoing '1' into the "tracing_on"
6795 file located in /sys/ 6795 file located in /sys/kernel/tracing/
6796 6796
6797 This option is useful 6797 This option is useful, as it disables the trace before
6798 the WARNING dump is c 6798 the WARNING dump is called, which prevents the trace to
6799 be filled with conten 6799 be filled with content caused by the warning output.
6800 6800
6801 This option can also 6801 This option can also be set at run time via the sysctl
6802 option: kernel/trace 6802 option: kernel/traceoff_on_warning
6803 6803
6804 transparent_hugepage= 6804 transparent_hugepage=
6805 [KNL] 6805 [KNL]
6806 Format: [always|madvi 6806 Format: [always|madvise|never]
6807 Can be used to contro 6807 Can be used to control the default behavior of the system
6808 with respect to trans 6808 with respect to transparent hugepages.
6809 See Documentation/adm 6809 See Documentation/admin-guide/mm/transhuge.rst
6810 for more details. 6810 for more details.
6811 6811
6812 trusted.source= [KEYS] 6812 trusted.source= [KEYS]
6813 Format: <string> 6813 Format: <string>
6814 This parameter identi 6814 This parameter identifies the trust source as a backend
6815 for trusted keys impl 6815 for trusted keys implementation. Supported trust
6816 sources: 6816 sources:
6817 - "tpm" 6817 - "tpm"
6818 - "tee" 6818 - "tee"
6819 - "caam" 6819 - "caam"
6820 - "dcp" 6820 - "dcp"
6821 If not specified then 6821 If not specified then it defaults to iterating through
6822 the trust source list 6822 the trust source list starting with TPM and assigns the
6823 first trust source as 6823 first trust source as a backend which is initialized
6824 successfully during i 6824 successfully during iteration.
6825 6825
6826 trusted.rng= [KEYS] 6826 trusted.rng= [KEYS]
6827 Format: <string> 6827 Format: <string>
6828 The RNG used to gener 6828 The RNG used to generate key material for trusted keys.
6829 Can be one of: 6829 Can be one of:
6830 - "kernel" 6830 - "kernel"
6831 - the same value as t 6831 - the same value as trusted.source: "tpm" or "tee"
6832 - "default" 6832 - "default"
6833 If not specified, "de 6833 If not specified, "default" is used. In this case,
6834 the RNG's choice is l 6834 the RNG's choice is left to each individual trust source.
6835 6835
6836 trusted.dcp_use_otp_key 6836 trusted.dcp_use_otp_key
6837 This is intended to b 6837 This is intended to be used in combination with
6838 trusted.source=dcp an 6838 trusted.source=dcp and will select the DCP OTP key
6839 instead of the DCP UN 6839 instead of the DCP UNIQUE key blob encryption.
6840 6840
6841 trusted.dcp_skip_zk_test 6841 trusted.dcp_skip_zk_test
6842 This is intended to b 6842 This is intended to be used in combination with
6843 trusted.source=dcp an 6843 trusted.source=dcp and will disable the check if the
6844 blob key is all zeros 6844 blob key is all zeros. This is helpful for situations where
6845 having this key zero' 6845 having this key zero'ed is acceptable. E.g. in testing
6846 scenarios. 6846 scenarios.
6847 6847
6848 tsc= Disable clocksource s 6848 tsc= Disable clocksource stability checks for TSC.
6849 Format: <string> 6849 Format: <string>
6850 [x86] reliable: mark 6850 [x86] reliable: mark tsc clocksource as reliable, this
6851 disables clocksource 6851 disables clocksource verification at runtime, as well
6852 as the stability chec 6852 as the stability checks done at bootup. Used to enable
6853 high-resolution timer 6853 high-resolution timer mode on older hardware, and in
6854 virtualized environme 6854 virtualized environment.
6855 [x86] noirqtime: Do n 6855 [x86] noirqtime: Do not use TSC to do irq accounting.
6856 Used to run time disa 6856 Used to run time disable IRQ_TIME_ACCOUNTING on any
6857 platforms where RDTSC 6857 platforms where RDTSC is slow and this accounting
6858 can add overhead. 6858 can add overhead.
6859 [x86] unstable: mark 6859 [x86] unstable: mark the TSC clocksource as unstable, this
6860 marks the TSC uncondi 6860 marks the TSC unconditionally unstable at bootup and
6861 avoids any further wo 6861 avoids any further wobbles once the TSC watchdog notices.
6862 [x86] nowatchdog: dis 6862 [x86] nowatchdog: disable clocksource watchdog. Used
6863 in situations with st 6863 in situations with strict latency requirements (where
6864 interruptions from cl 6864 interruptions from clocksource watchdog are not
6865 acceptable). 6865 acceptable).
6866 [x86] recalibrate: fo 6866 [x86] recalibrate: force recalibration against a HW timer
6867 (HPET or PM timer) on 6867 (HPET or PM timer) on systems whose TSC frequency was
6868 obtained from HW or F 6868 obtained from HW or FW using either an MSR or CPUID(0x15).
6869 Warn if the differenc 6869 Warn if the difference is more than 500 ppm.
6870 [x86] watchdog: Use T 6870 [x86] watchdog: Use TSC as the watchdog clocksource with
6871 which to check other 6871 which to check other HW timers (HPET or PM timer), but
6872 only on systems where 6872 only on systems where TSC has been deemed trustworthy.
6873 This will be suppress 6873 This will be suppressed by an earlier tsc=nowatchdog and
6874 can be overridden by 6874 can be overridden by a later tsc=nowatchdog. A console
6875 message will flag any 6875 message will flag any such suppression or overriding.
6876 6876
6877 tsc_early_khz= [X86,EARLY] Skip earl 6877 tsc_early_khz= [X86,EARLY] Skip early TSC calibration and use the given
6878 value instead. Useful 6878 value instead. Useful when the early TSC frequency discovery
6879 procedure is not reli 6879 procedure is not reliable, such as on overclocked systems
6880 with CPUID.16h suppor 6880 with CPUID.16h support and partial CPUID.15h support.
6881 Format: <unsigned int 6881 Format: <unsigned int>
6882 6882
6883 tsx= [X86] Control Transac 6883 tsx= [X86] Control Transactional Synchronization
6884 Extensions (TSX) feat 6884 Extensions (TSX) feature in Intel processors that
6885 support TSX control. 6885 support TSX control.
6886 6886
6887 This parameter contro 6887 This parameter controls the TSX feature. The options are:
6888 6888
6889 on - Enable TSX 6889 on - Enable TSX on the system. Although there are
6890 mitigations f 6890 mitigations for all known security vulnerabilities,
6891 TSX has been 6891 TSX has been known to be an accelerator for
6892 several previ 6892 several previous speculation-related CVEs, and
6893 so there may 6893 so there may be unknown security risks associated
6894 with leaving 6894 with leaving it enabled.
6895 6895
6896 off - Disable TSX 6896 off - Disable TSX on the system. (Note that this
6897 option takes 6897 option takes effect only on newer CPUs which are
6898 not vulnerabl 6898 not vulnerable to MDS, i.e., have
6899 MSR_IA32_ARCH 6899 MSR_IA32_ARCH_CAPABILITIES.MDS_NO=1 and which get
6900 the new IA32_ 6900 the new IA32_TSX_CTRL MSR through a microcode
6901 update. This 6901 update. This new MSR allows for the reliable
6902 deactivation 6902 deactivation of the TSX functionality.)
6903 6903
6904 auto - Disable TSX 6904 auto - Disable TSX if X86_BUG_TAA is present,
6905 otherwise e 6905 otherwise enable TSX on the system.
6906 6906
6907 Not specifying this o 6907 Not specifying this option is equivalent to tsx=off.
6908 6908
6909 See Documentation/adm 6909 See Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6910 for more details. 6910 for more details.
6911 6911
6912 tsx_async_abort= [X86,INTEL,EARLY] Co 6912 tsx_async_abort= [X86,INTEL,EARLY] Control mitigation for the TSX Async
6913 Abort (TAA) vulnerabi 6913 Abort (TAA) vulnerability.
6914 6914
6915 Similar to Micro-arch 6915 Similar to Micro-architectural Data Sampling (MDS)
6916 certain CPUs that sup 6916 certain CPUs that support Transactional
6917 Synchronization Exten 6917 Synchronization Extensions (TSX) are vulnerable to an
6918 exploit against CPU i 6918 exploit against CPU internal buffers which can forward
6919 information to a disc 6919 information to a disclosure gadget under certain
6920 conditions. 6920 conditions.
6921 6921
6922 In vulnerable process 6922 In vulnerable processors, the speculatively forwarded
6923 data can be used in a 6923 data can be used in a cache side channel attack, to
6924 access data to which 6924 access data to which the attacker does not have direct
6925 access. 6925 access.
6926 6926
6927 This parameter contro 6927 This parameter controls the TAA mitigation. The
6928 options are: 6928 options are:
6929 6929
6930 full - Enable T 6930 full - Enable TAA mitigation on vulnerable CPUs
6931 if TSX i 6931 if TSX is enabled.
6932 6932
6933 full,nosmt - Enable T 6933 full,nosmt - Enable TAA mitigation and disable SMT on
6934 vulnerab 6934 vulnerable CPUs. If TSX is disabled, SMT
6935 is not d 6935 is not disabled because CPU is not
6936 vulnerab 6936 vulnerable to cross-thread TAA attacks.
6937 off - Uncondit 6937 off - Unconditionally disable TAA mitigation
6938 6938
6939 On MDS-affected machi 6939 On MDS-affected machines, tsx_async_abort=off can be
6940 prevented by an activ 6940 prevented by an active MDS mitigation as both vulnerabilities
6941 are mitigated with th 6941 are mitigated with the same mechanism so in order to disable
6942 this mitigation, you 6942 this mitigation, you need to specify mds=off too.
6943 6943
6944 Not specifying this o 6944 Not specifying this option is equivalent to
6945 tsx_async_abort=full. 6945 tsx_async_abort=full. On CPUs which are MDS affected
6946 and deploy MDS mitiga 6946 and deploy MDS mitigation, TAA mitigation is not
6947 required and doesn't 6947 required and doesn't provide any additional
6948 mitigation. 6948 mitigation.
6949 6949
6950 For details see: 6950 For details see:
6951 Documentation/admin-g 6951 Documentation/admin-guide/hw-vuln/tsx_async_abort.rst
6952 6952
6953 turbografx.map[2|3]= [HW,JOY] 6953 turbografx.map[2|3]= [HW,JOY]
6954 TurboGraFX parallel p 6954 TurboGraFX parallel port interface
6955 Format: 6955 Format:
6956 <port#>,<js1>,<js2>,< 6956 <port#>,<js1>,<js2>,<js3>,<js4>,<js5>,<js6>,<js7>
6957 See also Documentatio 6957 See also Documentation/input/devices/joystick-parport.rst
6958 6958
6959 udbg-immortal [PPC] When debugging 6959 udbg-immortal [PPC] When debugging early kernel crashes that
6960 happen after console_ 6960 happen after console_init() and before a proper
6961 console driver takes 6961 console driver takes over, this boot options might
6962 help "seeing" what's 6962 help "seeing" what's going on.
6963 6963
6964 uhash_entries= [KNL,NET] 6964 uhash_entries= [KNL,NET]
6965 Set number of hash bu 6965 Set number of hash buckets for UDP/UDP-Lite connections
6966 6966
6967 uhci-hcd.ignore_oc= 6967 uhci-hcd.ignore_oc=
6968 [USB] Ignore overcurr 6968 [USB] Ignore overcurrent events (default N).
6969 Some badly-designed m 6969 Some badly-designed motherboards generate lots of
6970 bogus events, for por 6970 bogus events, for ports that aren't wired to
6971 anything. Set this p 6971 anything. Set this parameter to avoid log spamming.
6972 Note that genuine ove 6972 Note that genuine overcurrent events won't be
6973 reported either. 6973 reported either.
6974 6974
6975 unknown_nmi_panic 6975 unknown_nmi_panic
6976 [X86] Cause panic on 6976 [X86] Cause panic on unknown NMI.
6977 6977
6978 unwind_debug [X86-64,EARLY] 6978 unwind_debug [X86-64,EARLY]
6979 Enable unwinder debug 6979 Enable unwinder debug output. This can be
6980 useful for debugging 6980 useful for debugging certain unwinder error
6981 conditions, including 6981 conditions, including corrupt stacks and
6982 bad/missing unwinder 6982 bad/missing unwinder metadata.
6983 6983
6984 usbcore.authorized_default= 6984 usbcore.authorized_default=
6985 [USB] Default USB dev 6985 [USB] Default USB device authorization:
6986 (default -1 = authori 6986 (default -1 = authorized (same as 1),
6987 0 = not authorized, 1 6987 0 = not authorized, 1 = authorized, 2 = authorized
6988 if device connected t 6988 if device connected to internal port)
6989 6989
6990 usbcore.autosuspend= 6990 usbcore.autosuspend=
6991 [USB] The autosuspend 6991 [USB] The autosuspend time delay (in seconds) used
6992 for newly-detected US 6992 for newly-detected USB devices (default 2). This
6993 is the time required 6993 is the time required before an idle device will be
6994 autosuspended. Devic 6994 autosuspended. Devices for which the delay is set
6995 to a negative value w 6995 to a negative value won't be autosuspended at all.
6996 6996
6997 usbcore.usbfs_snoop= 6997 usbcore.usbfs_snoop=
6998 [USB] Set to log all 6998 [USB] Set to log all usbfs traffic (default 0 = off).
6999 6999
7000 usbcore.usbfs_snoop_max= 7000 usbcore.usbfs_snoop_max=
7001 [USB] Maximum number 7001 [USB] Maximum number of bytes to snoop in each URB
7002 (default = 65536). 7002 (default = 65536).
7003 7003
7004 usbcore.blinkenlights= 7004 usbcore.blinkenlights=
7005 [USB] Set to cycle le 7005 [USB] Set to cycle leds on hubs (default 0 = off).
7006 7006
7007 usbcore.old_scheme_first= 7007 usbcore.old_scheme_first=
7008 [USB] Start with the 7008 [USB] Start with the old device initialization
7009 scheme (default 0 = o 7009 scheme (default 0 = off).
7010 7010
7011 usbcore.usbfs_memory_mb= 7011 usbcore.usbfs_memory_mb=
7012 [USB] Memory limit (i 7012 [USB] Memory limit (in MB) for buffers allocated by
7013 usbfs (default = 16, 7013 usbfs (default = 16, 0 = max = 2047).
7014 7014
7015 usbcore.use_both_schemes= 7015 usbcore.use_both_schemes=
7016 [USB] Try the other d 7016 [USB] Try the other device initialization scheme
7017 if the first one fail 7017 if the first one fails (default 1 = enabled).
7018 7018
7019 usbcore.initial_descriptor_timeout= 7019 usbcore.initial_descriptor_timeout=
7020 [USB] Specifies timeo 7020 [USB] Specifies timeout for the initial 64-byte
7021 USB_REQ_GET_DESCRIPTO 7021 USB_REQ_GET_DESCRIPTOR request in milliseconds
7022 (default 5000 = 5.0 s 7022 (default 5000 = 5.0 seconds).
7023 7023
7024 usbcore.nousb [USB] Disable the USB 7024 usbcore.nousb [USB] Disable the USB subsystem
7025 7025
7026 usbcore.quirks= 7026 usbcore.quirks=
7027 [USB] A list of quirk 7027 [USB] A list of quirk entries to augment the built-in
7028 usb core quirk list. 7028 usb core quirk list. List entries are separated by
7029 commas. Each entry ha 7029 commas. Each entry has the form
7030 VendorID:ProductID:Fl 7030 VendorID:ProductID:Flags. The IDs are 4-digit hex
7031 numbers and Flags is 7031 numbers and Flags is a set of letters. Each letter
7032 will change the built 7032 will change the built-in quirk; setting it if it is
7033 clear and clearing it 7033 clear and clearing it if it is set. The letters have
7034 the following meaning 7034 the following meanings:
7035 a = USB_QUIRK 7035 a = USB_QUIRK_STRING_FETCH_255 (string
7036 descr 7036 descriptors must not be fetched using
7037 a 255 7037 a 255-byte read);
7038 b = USB_QUIRK 7038 b = USB_QUIRK_RESET_RESUME (device can't resume
7039 corre 7039 correctly so reset it instead);
7040 c = USB_QUIRK 7040 c = USB_QUIRK_NO_SET_INTF (device can't handle
7041 Set-I 7041 Set-Interface requests);
7042 d = USB_QUIRK 7042 d = USB_QUIRK_CONFIG_INTF_STRINGS (device can't
7043 handl 7043 handle its Configuration or Interface
7044 strin 7044 strings);
7045 e = USB_QUIRK 7045 e = USB_QUIRK_RESET (device can't be reset
7046 (e.g 7046 (e.g morph devices), don't use reset);
7047 f = USB_QUIRK 7047 f = USB_QUIRK_HONOR_BNUMINTERFACES (device has
7048 more 7048 more interface descriptions than the
7049 bNumI 7049 bNumInterfaces count, and can't handle
7050 talki 7050 talking to these interfaces);
7051 g = USB_QUIRK 7051 g = USB_QUIRK_DELAY_INIT (device needs a pause
7052 durin 7052 during initialization, after we read
7053 the d 7053 the device descriptor);
7054 h = USB_QUIRK 7054 h = USB_QUIRK_LINEAR_UFRAME_INTR_BINTERVAL (For
7055 high 7055 high speed and super speed interrupt
7056 endpo 7056 endpoints, the USB 2.0 and USB 3.0 spec
7057 requi 7057 require the interval in microframes (1
7058 micro 7058 microframe = 125 microseconds) to be
7059 calcu 7059 calculated as interval = 2 ^
7060 (bInt 7060 (bInterval-1).
7061 Devic 7061 Devices with this quirk report their
7062 bInte 7062 bInterval as the result of this
7063 calcu 7063 calculation instead of the exponent
7064 varia 7064 variable used in the calculation);
7065 i = USB_QUIRK 7065 i = USB_QUIRK_DEVICE_QUALIFIER (device can't
7066 handl 7066 handle device_qualifier descriptor
7067 reque 7067 requests);
7068 j = USB_QUIRK 7068 j = USB_QUIRK_IGNORE_REMOTE_WAKEUP (device
7069 gener 7069 generates spurious wakeup, ignore
7070 remot 7070 remote wakeup capability);
7071 k = USB_QUIRK 7071 k = USB_QUIRK_NO_LPM (device can't handle Link
7072 Power 7072 Power Management);
7073 l = USB_QUIRK 7073 l = USB_QUIRK_LINEAR_FRAME_INTR_BINTERVAL
7074 (Devi 7074 (Device reports its bInterval as linear
7075 frame 7075 frames instead of the USB 2.0
7076 calcu 7076 calculation);
7077 m = USB_QUIRK 7077 m = USB_QUIRK_DISCONNECT_SUSPEND (Device needs
7078 to be 7078 to be disconnected before suspend to
7079 preve 7079 prevent spurious wakeup);
7080 n = USB_QUIRK 7080 n = USB_QUIRK_DELAY_CTRL_MSG (Device needs a
7081 pause 7081 pause after every control message);
7082 o = USB_QUIRK 7082 o = USB_QUIRK_HUB_SLOW_RESET (Hub needs extra
7083 delay 7083 delay after resetting its port);
7084 p = USB_QUIRK 7084 p = USB_QUIRK_SHORT_SET_ADDRESS_REQ_TIMEOUT
7085 (Redu 7085 (Reduce timeout of the SET_ADDRESS
7086 reque 7086 request from 5000 ms to 500 ms);
7087 Example: quirks=0781: 7087 Example: quirks=0781:5580:bk,0a5c:5834:gij
7088 7088
7089 usbhid.mousepoll= 7089 usbhid.mousepoll=
7090 [USBHID] The interval 7090 [USBHID] The interval which mice are to be polled at.
7091 7091
7092 usbhid.jspoll= 7092 usbhid.jspoll=
7093 [USBHID] The interval 7093 [USBHID] The interval which joysticks are to be polled at.
7094 7094
7095 usbhid.kbpoll= 7095 usbhid.kbpoll=
7096 [USBHID] The interval 7096 [USBHID] The interval which keyboards are to be polled at.
7097 7097
7098 usb-storage.delay_use= 7098 usb-storage.delay_use=
7099 [UMS] The delay in se 7099 [UMS] The delay in seconds before a new device is
7100 scanned for Logical U 7100 scanned for Logical Units (default 1).
7101 Optionally the delay 7101 Optionally the delay in milliseconds if the value has
7102 suffix with "ms". 7102 suffix with "ms".
7103 Example: delay_use=25 7103 Example: delay_use=2567ms
7104 7104
7105 usb-storage.quirks= 7105 usb-storage.quirks=
7106 [UMS] A list of quirk 7106 [UMS] A list of quirks entries to supplement or
7107 override the built-in 7107 override the built-in unusual_devs list. List
7108 entries are separated 7108 entries are separated by commas. Each entry has
7109 the form VID:PID:Flag 7109 the form VID:PID:Flags where VID and PID are Vendor
7110 and Product ID values 7110 and Product ID values (4-digit hex numbers) and
7111 Flags is a set of cha 7111 Flags is a set of characters, each corresponding
7112 to a common usb-stora 7112 to a common usb-storage quirk flag as follows:
7113 a = SANE_SENS 7113 a = SANE_SENSE (collect more than 18 bytes
7114 of se 7114 of sense data, not on uas);
7115 b = BAD_SENSE 7115 b = BAD_SENSE (don't collect more than 18
7116 bytes 7116 bytes of sense data, not on uas);
7117 c = FIX_CAPAC 7117 c = FIX_CAPACITY (decrease the reported
7118 devic 7118 device capacity by one sector);
7119 d = NO_READ_D 7119 d = NO_READ_DISC_INFO (don't use
7120 READ_ 7120 READ_DISC_INFO command, not on uas);
7121 e = NO_READ_C 7121 e = NO_READ_CAPACITY_16 (don't use
7122 READ_ 7122 READ_CAPACITY_16 command);
7123 f = NO_REPORT 7123 f = NO_REPORT_OPCODES (don't use report opcodes
7124 comma 7124 command, uas only);
7125 g = MAX_SECTO 7125 g = MAX_SECTORS_240 (don't transfer more than
7126 240 s 7126 240 sectors at a time, uas only);
7127 h = CAPACITY_ 7127 h = CAPACITY_HEURISTICS (decrease the
7128 repor 7128 reported device capacity by one
7129 secto 7129 sector if the number is odd);
7130 i = IGNORE_DE 7130 i = IGNORE_DEVICE (don't bind to this
7131 devic 7131 device);
7132 j = NO_REPORT 7132 j = NO_REPORT_LUNS (don't use report luns
7133 comma 7133 command, uas only);
7134 k = NO_SAME ( 7134 k = NO_SAME (do not use WRITE_SAME, uas only)
7135 l = NOT_LOCKA 7135 l = NOT_LOCKABLE (don't try to lock and
7136 unloc 7136 unlock ejectable media, not on uas);
7137 m = MAX_SECTO 7137 m = MAX_SECTORS_64 (don't transfer more
7138 than 7138 than 64 sectors = 32 KB at a time,
7139 not o 7139 not on uas);
7140 n = INITIAL_R 7140 n = INITIAL_READ10 (force a retry of the
7141 initi 7141 initial READ(10) command, not on uas);
7142 o = CAPACITY_ 7142 o = CAPACITY_OK (accept the capacity
7143 repor 7143 reported by the device, not on uas);
7144 p = WRITE_CAC 7144 p = WRITE_CACHE (the device cache is ON
7145 by de 7145 by default, not on uas);
7146 r = IGNORE_RE 7146 r = IGNORE_RESIDUE (the device reports
7147 bogus 7147 bogus residue values, not on uas);
7148 s = SINGLE_LU 7148 s = SINGLE_LUN (the device has only one
7149 Logic 7149 Logical Unit);
7150 t = NO_ATA_1X 7150 t = NO_ATA_1X (don't allow ATA(12) and ATA(16)
7151 comma 7151 commands, uas only);
7152 u = IGNORE_UA 7152 u = IGNORE_UAS (don't bind to the uas driver);
7153 w = NO_WP_DET 7153 w = NO_WP_DETECT (don't test whether the
7154 mediu 7154 medium is write-protected).
7155 y = ALWAYS_SY 7155 y = ALWAYS_SYNC (issue a SYNCHRONIZE_CACHE
7156 even 7156 even if the device claims no cache,
7157 not o 7157 not on uas)
7158 Example: quirks=0419: 7158 Example: quirks=0419:aaf5:rl,0421:0433:rc
7159 7159
7160 user_debug= [KNL,ARM] 7160 user_debug= [KNL,ARM]
7161 Format: <int> 7161 Format: <int>
7162 See arch/arm/Kconfig. 7162 See arch/arm/Kconfig.debug help text.
7163 1 - undefine 7163 1 - undefined instruction events
7164 2 - system c 7164 2 - system calls
7165 4 - invalid 7165 4 - invalid data aborts
7166 8 - SIGSEGV 7166 8 - SIGSEGV faults
7167 16 - SIGBUS f 7167 16 - SIGBUS faults
7168 Example: user_debug=3 7168 Example: user_debug=31
7169 7169
7170 userpte= 7170 userpte=
7171 [X86,EARLY] Flags con 7171 [X86,EARLY] Flags controlling user PTE allocations.
7172 7172
7173 nohigh = do n 7173 nohigh = do not allocate PTE pages in
7174 HIGHM 7174 HIGHMEM regardless of setting
7175 of CO 7175 of CONFIG_HIGHPTE.
7176 7176
7177 vdso= [X86,SH,SPARC] 7177 vdso= [X86,SH,SPARC]
7178 On X86_32, this is an 7178 On X86_32, this is an alias for vdso32=. Otherwise:
7179 7179
7180 vdso=1: enable VDSO ( 7180 vdso=1: enable VDSO (the default)
7181 vdso=0: disable VDSO 7181 vdso=0: disable VDSO mapping
7182 7182
7183 vdso32= [X86] Control the 32- 7183 vdso32= [X86] Control the 32-bit vDSO
7184 vdso32=1: enable 32-b 7184 vdso32=1: enable 32-bit VDSO
7185 vdso32=0 or vdso32=2: 7185 vdso32=0 or vdso32=2: disable 32-bit VDSO
7186 7186
7187 See the help text for 7187 See the help text for CONFIG_COMPAT_VDSO for more
7188 details. If CONFIG_C 7188 details. If CONFIG_COMPAT_VDSO is set, the default is
7189 vdso32=0; otherwise, 7189 vdso32=0; otherwise, the default is vdso32=1.
7190 7190
7191 For compatibility wit 7191 For compatibility with older kernels, vdso32=2 is an
7192 alias for vdso32=0. 7192 alias for vdso32=0.
7193 7193
7194 Try vdso32=0 if you e 7194 Try vdso32=0 if you encounter an error that says:
7195 dl_main: Assertion `( 7195 dl_main: Assertion `(void *) ph->p_vaddr == _rtld_local._dl_sysinfo_dso' failed!
7196 7196
7197 video= [FB,EARLY] Frame buff 7197 video= [FB,EARLY] Frame buffer configuration
7198 See Documentation/fb/ 7198 See Documentation/fb/modedb.rst.
7199 7199
7200 video.brightness_switch_enabled= [ACP 7200 video.brightness_switch_enabled= [ACPI]
7201 Format: [0|1] 7201 Format: [0|1]
7202 If set to 1, on recei 7202 If set to 1, on receiving an ACPI notify event
7203 generated by hotkey, 7203 generated by hotkey, video driver will adjust brightness
7204 level and then send o 7204 level and then send out the event to user space through
7205 the allocated input d 7205 the allocated input device. If set to 0, video driver
7206 will only send out th 7206 will only send out the event without touching backlight
7207 brightness level. 7207 brightness level.
7208 default: 1 7208 default: 1
7209 7209
7210 virtio_mmio.device= 7210 virtio_mmio.device=
7211 [VMMIO] Memory mapped 7211 [VMMIO] Memory mapped virtio (platform) device.
7212 7212
7213 <size>@<basead 7213 <size>@<baseaddr>:<irq>[:<id>]
7214 where: 7214 where:
7215 <size> := 7215 <size> := size (can use standard suffixes
7216 7216 like K, M and G)
7217 <baseaddr> := 7217 <baseaddr> := physical base address
7218 <irq> := 7218 <irq> := interrupt number (as passed to
7219 7219 request_irq())
7220 <id> := 7220 <id> := (optional) platform device id
7221 example: 7221 example:
7222 virtio_mmio.d 7222 virtio_mmio.device=1K@0x100b0000:48:7
7223 7223
7224 Can be used multiple 7224 Can be used multiple times for multiple devices.
7225 7225
7226 vga= [BOOT,X86-32] Select 7226 vga= [BOOT,X86-32] Select a particular video mode
7227 See Documentation/arc 7227 See Documentation/arch/x86/boot.rst and
7228 Documentation/admin-g 7228 Documentation/admin-guide/svga.rst.
7229 Use vga=ask for menu. 7229 Use vga=ask for menu.
7230 This is actually a bo 7230 This is actually a boot loader parameter; the value is
7231 passed to the kernel 7231 passed to the kernel using a special protocol.
7232 7232
7233 vm_debug[=options] [KNL] Availab 7233 vm_debug[=options] [KNL] Available with CONFIG_DEBUG_VM=y.
7234 May slow down system 7234 May slow down system boot speed, especially when
7235 enabled on systems wi 7235 enabled on systems with a large amount of memory.
7236 All options are enabl 7236 All options are enabled by default, and this
7237 interface is meant to 7237 interface is meant to allow for selectively
7238 enabling or disabling 7238 enabling or disabling specific virtual memory
7239 debugging features. 7239 debugging features.
7240 7240
7241 Available options are 7241 Available options are:
7242 P Enable page s 7242 P Enable page structure init time poisoning
7243 - Disable all o 7243 - Disable all of the above options
7244 7244
7245 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forc 7245 vmalloc=nn[KMG] [KNL,BOOT,EARLY] Forces the vmalloc area to have an
7246 exact size of <nn>. T 7246 exact size of <nn>. This can be used to increase
7247 the minimum size (128 7247 the minimum size (128MB on x86, arm32 platforms).
7248 It can also be used t 7248 It can also be used to decrease the size and leave more room
7249 for directly mapped k 7249 for directly mapped kernel RAM. Note that this parameter does
7250 not exist on many oth 7250 not exist on many other platforms (including arm64, alpha,
7251 loongarch, arc, csky, 7251 loongarch, arc, csky, hexagon, microblaze, mips, nios2, openrisc,
7252 parisc, m64k, powerpc 7252 parisc, m64k, powerpc, riscv, sh, um, xtensa, s390, sparc).
7253 7253
7254 vmcp_cma=nn[MG] [KNL,S390,EARLY] 7254 vmcp_cma=nn[MG] [KNL,S390,EARLY]
7255 Sets the memory size 7255 Sets the memory size reserved for contiguous memory
7256 allocations for the v 7256 allocations for the vmcp device driver.
7257 7257
7258 vmhalt= [KNL,S390] Perform z/ 7258 vmhalt= [KNL,S390] Perform z/VM CP command after system halt.
7259 Format: <command> 7259 Format: <command>
7260 7260
7261 vmpanic= [KNL,S390] Perform z/ 7261 vmpanic= [KNL,S390] Perform z/VM CP command after kernel panic.
7262 Format: <command> 7262 Format: <command>
7263 7263
7264 vmpoff= [KNL,S390] Perform z/ 7264 vmpoff= [KNL,S390] Perform z/VM CP command after power off.
7265 Format: <command> 7265 Format: <command>
7266 7266
7267 vsyscall= [X86-64,EARLY] 7267 vsyscall= [X86-64,EARLY]
7268 Controls the behavior 7268 Controls the behavior of vsyscalls (i.e. calls to
7269 fixed addresses of 0x 7269 fixed addresses of 0xffffffffff600x00 from legacy
7270 code). Most statical 7270 code). Most statically-linked binaries and older
7271 versions of glibc use 7271 versions of glibc use these calls. Because these
7272 functions are at fixe 7272 functions are at fixed addresses, they make nice
7273 targets for exploits 7273 targets for exploits that can control RIP.
7274 7274
7275 emulate Vsyscalls 7275 emulate Vsyscalls turn into traps and are emulated
7276 reasonabl 7276 reasonably safely. The vsyscall page is
7277 readable. 7277 readable.
7278 7278
7279 xonly [default] 7279 xonly [default] Vsyscalls turn into traps and are
7280 emulated 7280 emulated reasonably safely. The vsyscall
7281 page is n 7281 page is not readable.
7282 7282
7283 none Vsyscalls 7283 none Vsyscalls don't work at all. This makes
7284 them quit 7284 them quite hard to use for exploits but
7285 might bre 7285 might break your system.
7286 7286
7287 vt.color= [VT] Default text col 7287 vt.color= [VT] Default text color.
7288 Format: 0xYX, X = for 7288 Format: 0xYX, X = foreground, Y = background.
7289 Default: 0x07 = light 7289 Default: 0x07 = light gray on black.
7290 7290
7291 vt.cur_default= [VT] Default cursor s 7291 vt.cur_default= [VT] Default cursor shape.
7292 Format: 0xCCBBAA, whe 7292 Format: 0xCCBBAA, where AA, BB, and CC are the same as
7293 the parameters of the 7293 the parameters of the <Esc>[?A;B;Cc escape sequence;
7294 see VGA-softcursor.tx 7294 see VGA-softcursor.txt. Default: 2 = underline.
7295 7295
7296 vt.default_blu= [VT] 7296 vt.default_blu= [VT]
7297 Format: <blue0>,<blue 7297 Format: <blue0>,<blue1>,<blue2>,...,<blue15>
7298 Change the default bl 7298 Change the default blue palette of the console.
7299 This is a 16-member a 7299 This is a 16-member array composed of values
7300 ranging from 0-255. 7300 ranging from 0-255.
7301 7301
7302 vt.default_grn= [VT] 7302 vt.default_grn= [VT]
7303 Format: <green0>,<gre 7303 Format: <green0>,<green1>,<green2>,...,<green15>
7304 Change the default gr 7304 Change the default green palette of the console.
7305 This is a 16-member a 7305 This is a 16-member array composed of values
7306 ranging from 0-255. 7306 ranging from 0-255.
7307 7307
7308 vt.default_red= [VT] 7308 vt.default_red= [VT]
7309 Format: <red0>,<red1> 7309 Format: <red0>,<red1>,<red2>,...,<red15>
7310 Change the default re 7310 Change the default red palette of the console.
7311 This is a 16-member a 7311 This is a 16-member array composed of values
7312 ranging from 0-255. 7312 ranging from 0-255.
7313 7313
7314 vt.default_utf8= 7314 vt.default_utf8=
7315 [VT] 7315 [VT]
7316 Format=<0|1> 7316 Format=<0|1>
7317 Set system-wide defau 7317 Set system-wide default UTF-8 mode for all tty's.
7318 Default is 1, i.e. UT 7318 Default is 1, i.e. UTF-8 mode is enabled for all
7319 newly opened terminal 7319 newly opened terminals.
7320 7320
7321 vt.global_cursor_default= 7321 vt.global_cursor_default=
7322 [VT] 7322 [VT]
7323 Format=<-1|0|1> 7323 Format=<-1|0|1>
7324 Set system-wide defau 7324 Set system-wide default for whether a cursor
7325 is shown on new VTs. 7325 is shown on new VTs. Default is -1,
7326 i.e. cursors will be 7326 i.e. cursors will be created by default unless
7327 overridden by individ 7327 overridden by individual drivers. 0 will hide
7328 cursors, 1 will displ 7328 cursors, 1 will display them.
7329 7329
7330 vt.italic= [VT] Default color fo 7330 vt.italic= [VT] Default color for italic text; 0-15.
7331 Default: 2 = green. 7331 Default: 2 = green.
7332 7332
7333 vt.underline= [VT] Default color fo 7333 vt.underline= [VT] Default color for underlined text; 0-15.
7334 Default: 3 = cyan. 7334 Default: 3 = cyan.
7335 7335
7336 watchdog timers [HW,WDT] For informat 7336 watchdog timers [HW,WDT] For information on watchdog timers,
7337 see Documentation/wat 7337 see Documentation/watchdog/watchdog-parameters.rst
7338 or other driver-speci 7338 or other driver-specific files in the
7339 Documentation/watchdo 7339 Documentation/watchdog/ directory.
7340 7340
7341 watchdog_thresh= 7341 watchdog_thresh=
7342 [KNL] 7342 [KNL]
7343 Set the hard lockup d 7343 Set the hard lockup detector stall duration
7344 threshold in seconds. 7344 threshold in seconds. The soft lockup detector
7345 threshold is set to t 7345 threshold is set to twice the value. A value of 0
7346 disables both lockup 7346 disables both lockup detectors. Default is 10
7347 seconds. 7347 seconds.
7348 7348
7349 workqueue.unbound_cpus= 7349 workqueue.unbound_cpus=
7350 [KNL,SMP] Specify to 7350 [KNL,SMP] Specify to constrain one or some CPUs
7351 to use in unbound wor 7351 to use in unbound workqueues.
7352 Format: <cpu-list> 7352 Format: <cpu-list>
7353 By default, all onlin 7353 By default, all online CPUs are available for
7354 unbound workqueues. 7354 unbound workqueues.
7355 7355
7356 workqueue.watchdog_thresh= 7356 workqueue.watchdog_thresh=
7357 If CONFIG_WQ_WATCHDOG 7357 If CONFIG_WQ_WATCHDOG is configured, workqueue can
7358 warn stall conditions 7358 warn stall conditions and dump internal state to
7359 help debugging. 0 di 7359 help debugging. 0 disables workqueue stall
7360 detection; otherwise, 7360 detection; otherwise, it's the stall threshold
7361 duration in seconds. 7361 duration in seconds. The default value is 30 and
7362 it can be updated at 7362 it can be updated at runtime by writing to the
7363 corresponding sysfs f 7363 corresponding sysfs file.
7364 7364
7365 workqueue.cpu_intensive_thresh_us= 7365 workqueue.cpu_intensive_thresh_us=
7366 Per-cpu work items wh 7366 Per-cpu work items which run for longer than this
7367 threshold are automat 7367 threshold are automatically considered CPU intensive
7368 and excluded from con 7368 and excluded from concurrency management to prevent
7369 them from noticeably 7369 them from noticeably delaying other per-cpu work
7370 items. Default is 100 7370 items. Default is 10000 (10ms).
7371 7371
7372 If CONFIG_WQ_CPU_INTE 7372 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel
7373 will report the work 7373 will report the work functions which violate this
7374 threshold repeatedly. 7374 threshold repeatedly. They are likely good
7375 candidates for using 7375 candidates for using WQ_UNBOUND workqueues instead.
7376 7376
7377 workqueue.cpu_intensive_warning_thres 7377 workqueue.cpu_intensive_warning_thresh=<uint>
7378 If CONFIG_WQ_CPU_INTE 7378 If CONFIG_WQ_CPU_INTENSIVE_REPORT is set, the kernel
7379 will report the work 7379 will report the work functions which violate the
7380 intensive_threshold_u 7380 intensive_threshold_us repeatedly. In order to prevent
7381 spurious warnings, st 7381 spurious warnings, start printing only after a work
7382 function has violated 7382 function has violated this threshold number of times.
7383 7383
7384 The default is 4 time 7384 The default is 4 times. 0 disables the warning.
7385 7385
7386 workqueue.power_efficient 7386 workqueue.power_efficient
7387 Per-cpu workqueues ar 7387 Per-cpu workqueues are generally preferred because
7388 they show better perf 7388 they show better performance thanks to cache
7389 locality; unfortunate 7389 locality; unfortunately, per-cpu workqueues tend to
7390 be more power hungry 7390 be more power hungry than unbound workqueues.
7391 7391
7392 Enabling this makes t 7392 Enabling this makes the per-cpu workqueues which
7393 were observed to cont 7393 were observed to contribute significantly to power
7394 consumption unbound, 7394 consumption unbound, leading to measurably lower
7395 power usage at the co 7395 power usage at the cost of small performance
7396 overhead. 7396 overhead.
7397 7397
7398 The default value of 7398 The default value of this parameter is determined by
7399 the config option CON 7399 the config option CONFIG_WQ_POWER_EFFICIENT_DEFAULT.
7400 7400
7401 workqueue.default_affinity_scope= 7401 workqueue.default_affinity_scope=
7402 Select the default af 7402 Select the default affinity scope to use for unbound
7403 workqueues. Can be on 7403 workqueues. Can be one of "cpu", "smt", "cache",
7404 "numa" and "system". 7404 "numa" and "system". Default is "cache". For more
7405 information, see the 7405 information, see the Affinity Scopes section in
7406 Documentation/core-ap 7406 Documentation/core-api/workqueue.rst.
7407 7407
7408 This can be changed a 7408 This can be changed after boot by writing to the
7409 matching /sys/module/ 7409 matching /sys/module/workqueue/parameters file. All
7410 workqueues with the " 7410 workqueues with the "default" affinity scope will be
7411 updated accordingly. 7411 updated accordingly.
7412 7412
7413 workqueue.debug_force_rr_cpu 7413 workqueue.debug_force_rr_cpu
7414 Workqueue used to imp 7414 Workqueue used to implicitly guarantee that work
7415 items queued without 7415 items queued without explicit CPU specified are put
7416 on the local CPU. Th 7416 on the local CPU. This guarantee is no longer true
7417 and while local CPU i 7417 and while local CPU is still preferred work items
7418 may be put on foreign 7418 may be put on foreign CPUs. This debug option
7419 forces round-robin CP 7419 forces round-robin CPU selection to flush out
7420 usages which depend o 7420 usages which depend on the now broken guarantee.
7421 When enabled, memory 7421 When enabled, memory and cache locality will be
7422 impacted. 7422 impacted.
7423 7423
7424 writecombine= [LOONGARCH,EARLY] Con 7424 writecombine= [LOONGARCH,EARLY] Control the MAT (Memory Access
7425 Type) of ioremap_wc() 7425 Type) of ioremap_wc().
7426 7426
7427 on - Enable writeco 7427 on - Enable writecombine, use WUC for ioremap_wc()
7428 off - Disable writec 7428 off - Disable writecombine, use SUC for ioremap_wc()
7429 7429
7430 x2apic_phys [X86-64,APIC,EARLY] U 7430 x2apic_phys [X86-64,APIC,EARLY] Use x2apic physical mode instead of
7431 default x2apic cluste 7431 default x2apic cluster mode on platforms
7432 supporting x2apic. 7432 supporting x2apic.
7433 7433
7434 xen_512gb_limit [KNL,X86-64,X 7434 xen_512gb_limit [KNL,X86-64,XEN]
7435 Restricts the kernel 7435 Restricts the kernel running paravirtualized under Xen
7436 to use only up to 512 7436 to use only up to 512 GB of RAM. The reason to do so is
7437 crash analysis tools 7437 crash analysis tools and Xen tools for doing domain
7438 save/restore/migratio 7438 save/restore/migration must be enabled to handle larger
7439 domains. 7439 domains.
7440 7440
7441 xen_emul_unplug= [HW,X 7441 xen_emul_unplug= [HW,X86,XEN,EARLY]
7442 Unplug Xen emulated d 7442 Unplug Xen emulated devices
7443 Format: [unplug0,][un 7443 Format: [unplug0,][unplug1]
7444 ide-disks -- unplug p 7444 ide-disks -- unplug primary master IDE devices
7445 aux-ide-disks -- unpl 7445 aux-ide-disks -- unplug non-primary-master IDE devices
7446 nics -- unplug networ 7446 nics -- unplug network devices
7447 all -- unplug all emu 7447 all -- unplug all emulated devices (NICs and IDE disks)
7448 unnecessary -- unplug 7448 unnecessary -- unplugging emulated devices is
7449 unnecessary e 7449 unnecessary even if the host did not respond to
7450 the unplug pr 7450 the unplug protocol
7451 never -- do not unplu 7451 never -- do not unplug even if version check succeeds
7452 7452
7453 xen_legacy_crash [X86,XEN,EARL 7453 xen_legacy_crash [X86,XEN,EARLY]
7454 Crash from Xen panic 7454 Crash from Xen panic notifier, without executing late
7455 panic() code such as 7455 panic() code such as dumping handler.
7456 7456
7457 xen_mc_debug [X86,XEN,EARLY] 7457 xen_mc_debug [X86,XEN,EARLY]
7458 Enable multicall debu 7458 Enable multicall debugging when running as a Xen PV guest.
7459 Enabling this feature 7459 Enabling this feature will reduce performance a little
7460 bit, so it should onl 7460 bit, so it should only be enabled for obtaining extended
7461 debug data in case of 7461 debug data in case of multicall errors.
7462 7462
7463 xen_msr_safe= [X86,XEN,EARLY] 7463 xen_msr_safe= [X86,XEN,EARLY]
7464 Format: <bool> 7464 Format: <bool>
7465 Select whether to alw 7465 Select whether to always use non-faulting (safe) MSR
7466 access functions when 7466 access functions when running as Xen PV guest. The
7467 default value is cont 7467 default value is controlled by CONFIG_XEN_PV_MSR_SAFE.
7468 7468
7469 xen_nopv [X86] 7469 xen_nopv [X86]
7470 Disables the PV optim 7470 Disables the PV optimizations forcing the HVM guest to
7471 run as generic HVM gu 7471 run as generic HVM guest with no PV drivers.
7472 This option is obsole 7472 This option is obsoleted by the "nopv" option, which
7473 has equivalent effect 7473 has equivalent effect for XEN platform.
7474 7474
7475 xen_no_vector_callback 7475 xen_no_vector_callback
7476 [KNL,X86,XEN,EARLY] D 7476 [KNL,X86,XEN,EARLY] Disable the vector callback for Xen
7477 event channel interru 7477 event channel interrupts.
7478 7478
7479 xen_scrub_pages= [XEN] 7479 xen_scrub_pages= [XEN]
7480 Boolean option to con 7480 Boolean option to control scrubbing pages before giving them back
7481 to Xen, for use by ot 7481 to Xen, for use by other domains. Can be also changed at runtime
7482 with /sys/devices/sys 7482 with /sys/devices/system/xen_memory/xen_memory0/scrub_pages.
7483 Default value control 7483 Default value controlled with CONFIG_XEN_SCRUB_PAGES_DEFAULT.
7484 7484
7485 xen_timer_slop= [X86-64,XEN,EARLY] 7485 xen_timer_slop= [X86-64,XEN,EARLY]
7486 Set the timer slop (i 7486 Set the timer slop (in nanoseconds) for the virtual Xen
7487 timers (default is 10 7487 timers (default is 100000). This adjusts the minimum
7488 delta of virtualized 7488 delta of virtualized Xen timers, where lower values
7489 improve timer resolut 7489 improve timer resolution at the expense of processing
7490 more timer interrupts 7490 more timer interrupts.
7491 7491
7492 xen.balloon_boot_timeout= [XEN] 7492 xen.balloon_boot_timeout= [XEN]
7493 The time (in seconds) 7493 The time (in seconds) to wait before giving up to boot
7494 in case initial ballo 7494 in case initial ballooning fails to free enough memory.
7495 Applies only when run 7495 Applies only when running as HVM or PVH guest and
7496 started with less mem 7496 started with less memory configured than allowed at
7497 max. Default is 180. 7497 max. Default is 180.
7498 7498
7499 xen.event_eoi_delay= [XEN] 7499 xen.event_eoi_delay= [XEN]
7500 How long to delay EOI 7500 How long to delay EOI handling in case of event
7501 storms (jiffies). Def 7501 storms (jiffies). Default is 10.
7502 7502
7503 xen.event_loop_timeout= [XEN] 7503 xen.event_loop_timeout= [XEN]
7504 After which time (jif 7504 After which time (jiffies) the event handling loop
7505 should start to delay 7505 should start to delay EOI handling. Default is 2.
7506 7506
7507 xen.fifo_events= [XEN] 7507 xen.fifo_events= [XEN]
7508 Boolean parameter to 7508 Boolean parameter to disable using fifo event handling
7509 even if available. No 7509 even if available. Normally fifo event handling is
7510 preferred over the 2- 7510 preferred over the 2-level event handling, as it is
7511 fairer and the number 7511 fairer and the number of possible event channels is
7512 much higher. Default 7512 much higher. Default is on (use fifo events).
7513 7513
7514 xirc2ps_cs= [NET,PCMCIA] 7514 xirc2ps_cs= [NET,PCMCIA]
7515 Format: 7515 Format:
7516 <irq>,<irq_mask>,<io> 7516 <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
7517 7517
7518 xive= [PPC] 7518 xive= [PPC]
7519 By default on POWER9 7519 By default on POWER9 and above, the kernel will
7520 natively use the XIVE 7520 natively use the XIVE interrupt controller. This option
7521 allows the fallback f 7521 allows the fallback firmware mode to be used:
7522 7522
7523 off Fallback to 7523 off Fallback to firmware control of XIVE interrupt
7524 controller 7524 controller on both pseries and powernv
7525 platforms. 7525 platforms. Only useful on POWER9 and above.
7526 7526
7527 xive.store-eoi=off [PPC] 7527 xive.store-eoi=off [PPC]
7528 By default on POWER10 7528 By default on POWER10 and above, the kernel will use
7529 stores for EOI handli 7529 stores for EOI handling when the XIVE interrupt mode
7530 is active. This optio 7530 is active. This option allows the XIVE driver to use
7531 loads instead, as on 7531 loads instead, as on POWER9.
7532 7532
7533 xhci-hcd.quirks [USB,KNL] 7533 xhci-hcd.quirks [USB,KNL]
7534 A hex value specifyin 7534 A hex value specifying bitmask with supplemental xhci
7535 host controller quirk 7535 host controller quirks. Meaning of each bit can be
7536 consulted in header d 7536 consulted in header drivers/usb/host/xhci.h.
7537 7537
7538 xmon [PPC,EARLY] 7538 xmon [PPC,EARLY]
7539 Format: { early | on 7539 Format: { early | on | rw | ro | off }
7540 Controls if xmon debu 7540 Controls if xmon debugger is enabled. Default is off.
7541 Passing only "xmon" i 7541 Passing only "xmon" is equivalent to "xmon=early".
7542 early Call xmon as 7542 early Call xmon as early as possible on boot; xmon
7543 debugger is c 7543 debugger is called from setup_arch().
7544 on xmon debugger 7544 on xmon debugger hooks will be installed so xmon
7545 is only calle 7545 is only called on a kernel crash. Default mode,
7546 i.e. either " 7546 i.e. either "ro" or "rw" mode, is controlled
7547 with CONFIG_X 7547 with CONFIG_XMON_DEFAULT_RO_MODE.
7548 rw xmon debugger 7548 rw xmon debugger hooks will be installed so xmon
7549 is called onl 7549 is called only on a kernel crash, mode is write,
7550 meaning SPR r 7550 meaning SPR registers, memory and, other data
7551 can be writte 7551 can be written using xmon commands.
7552 ro same as "rw" 7552 ro same as "rw" option above but SPR registers,
7553 memory, and o 7553 memory, and other data can't be written using
7554 xmon commands 7554 xmon commands.
7555 off xmon is disab 7555 off xmon is disabled.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.