1 .. SPDX-License-Identifier: GPL-2.0
2
3 Spectre Side Channels
4 =====================
5
6 Spectre is a class of side channel attacks that exploit branch prediction
7 and speculative execution on modern CPUs to read memory, possibly
8 bypassing access controls. Speculative execution side channel exploits
9 do not modify memory but attempt to infer privileged data in the memory.
10
11 This document covers Spectre variant 1 and Spectre variant 2.
12
13 Affected processors
14 -------------------
15
16 Speculative execution side channel methods affect a wide range of modern
17 high performance processors, since most modern high speed processors
18 use branch prediction and speculative execution.
19
20 The following CPUs are vulnerable:
21
22 - Intel Core, Atom, Pentium, and Xeon processors
23
24 - AMD Phenom, EPYC, and Zen processors
25
26 - IBM POWER and zSeries processors
27
28 - Higher end ARM processors
29
30 - Apple CPUs
31
32 - Higher end MIPS CPUs
33
34 - Likely most other high performance CPUs. Contact your CPU vendor for details.
35
36 Whether a processor is affected or not can be read out from the Spectre
37 vulnerability files in sysfs. See :ref:`spectre_sys_info`.
38
39 Related CVEs
40 ------------
41
42 The following CVE entries describe Spectre variants:
43
44 ============= ======================= ==========================
45 CVE-2017-5753 Bounds check bypass Spectre variant 1
46 CVE-2017-5715 Branch target injection Spectre variant 2
47 CVE-2019-1125 Spectre v1 swapgs Spectre variant 1 (swapgs)
48 ============= ======================= ==========================
49
50 Problem
51 -------
52
53 CPUs use speculative operations to improve performance. That may leave
54 traces of memory accesses or computations in the processor's caches,
55 buffers, and branch predictors. Malicious software may be able to
56 influence the speculative execution paths, and then use the side effects
57 of the speculative execution in the CPUs' caches and buffers to infer
58 privileged data touched during the speculative execution.
59
60 Spectre variant 1 attacks take advantage of speculative execution of
61 conditional branches, while Spectre variant 2 attacks use speculative
62 execution of indirect branches to leak privileged memory.
63 See :ref:`[1] <spec_ref1>` :ref:`[5] <spec_ref5>` :ref:`[6] <spec_ref6>`
64 :ref:`[7] <spec_ref7>` :ref:`[10] <spec_ref10>` :ref:`[11] <spec_ref11>`.
65
66 Spectre variant 1 (Bounds Check Bypass)
67 ---------------------------------------
68
69 The bounds check bypass attack :ref:`[2] <spec_ref2>` takes advantage
70 of speculative execution that bypasses conditional branch instructions
71 used for memory access bounds check (e.g. checking if the index of an
72 array results in memory access within a valid range). This results in
73 memory accesses to invalid memory (with out-of-bound index) that are
74 done speculatively before validation checks resolve. Such speculative
75 memory accesses can leave side effects, creating side channels which
76 leak information to the attacker.
77
78 There are some extensions of Spectre variant 1 attacks for reading data
79 over the network, see :ref:`[12] <spec_ref12>`. However such attacks
80 are difficult, low bandwidth, fragile, and are considered low risk.
81
82 Note that, despite "Bounds Check Bypass" name, Spectre variant 1 is not
83 only about user-controlled array bounds checks. It can affect any
84 conditional checks. The kernel entry code interrupt, exception, and NMI
85 handlers all have conditional swapgs checks. Those may be problematic
86 in the context of Spectre v1, as kernel code can speculatively run with
87 a user GS.
88
89 Spectre variant 2 (Branch Target Injection)
90 -------------------------------------------
91
92 The branch target injection attack takes advantage of speculative
93 execution of indirect branches :ref:`[3] <spec_ref3>`. The indirect
94 branch predictors inside the processor used to guess the target of
95 indirect branches can be influenced by an attacker, causing gadget code
96 to be speculatively executed, thus exposing sensitive data touched by
97 the victim. The side effects left in the CPU's caches during speculative
98 execution can be measured to infer data values.
99
100 .. _poison_btb:
101
102 In Spectre variant 2 attacks, the attacker can steer speculative indirect
103 branches in the victim to gadget code by poisoning the branch target
104 buffer of a CPU used for predicting indirect branch addresses. Such
105 poisoning could be done by indirect branching into existing code,
106 with the address offset of the indirect branch under the attacker's
107 control. Since the branch prediction on impacted hardware does not
108 fully disambiguate branch address and uses the offset for prediction,
109 this could cause privileged code's indirect branch to jump to a gadget
110 code with the same offset.
111
112 The most useful gadgets take an attacker-controlled input parameter (such
113 as a register value) so that the memory read can be controlled. Gadgets
114 without input parameters might be possible, but the attacker would have
115 very little control over what memory can be read, reducing the risk of
116 the attack revealing useful data.
117
118 One other variant 2 attack vector is for the attacker to poison the
119 return stack buffer (RSB) :ref:`[13] <spec_ref13>` to cause speculative
120 subroutine return instruction execution to go to a gadget. An attacker's
121 imbalanced subroutine call instructions might "poison" entries in the
122 return stack buffer which are later consumed by a victim's subroutine
123 return instructions. This attack can be mitigated by flushing the return
124 stack buffer on context switch, or virtual machine (VM) exit.
125
126 On systems with simultaneous multi-threading (SMT), attacks are possible
127 from the sibling thread, as level 1 cache and branch target buffer
128 (BTB) may be shared between hardware threads in a CPU core. A malicious
129 program running on the sibling thread may influence its peer's BTB to
130 steer its indirect branch speculations to gadget code, and measure the
131 speculative execution's side effects left in level 1 cache to infer the
132 victim's data.
133
134 Yet another variant 2 attack vector is for the attacker to poison the
135 Branch History Buffer (BHB) to speculatively steer an indirect branch
136 to a specific Branch Target Buffer (BTB) entry, even if the entry isn't
137 associated with the source address of the indirect branch. Specifically,
138 the BHB might be shared across privilege levels even in the presence of
139 Enhanced IBRS.
140
141 Previously the only known real-world BHB attack vector was via unprivileged
142 eBPF. Further research has found attacks that don't require unprivileged eBPF.
143 For a full mitigation against BHB attacks it is recommended to set BHI_DIS_S or
144 use the BHB clearing sequence.
145
146 Attack scenarios
147 ----------------
148
149 The following list of attack scenarios have been anticipated, but may
150 not cover all possible attack vectors.
151
152 1. A user process attacking the kernel
153 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
154
155 Spectre variant 1
156 ~~~~~~~~~~~~~~~~~
157
158 The attacker passes a parameter to the kernel via a register or
159 via a known address in memory during a syscall. Such parameter may
160 be used later by the kernel as an index to an array or to derive
161 a pointer for a Spectre variant 1 attack. The index or pointer
162 is invalid, but bound checks are bypassed in the code branch taken
163 for speculative execution. This could cause privileged memory to be
164 accessed and leaked.
165
166 For kernel code that has been identified where data pointers could
167 potentially be influenced for Spectre attacks, new "nospec" accessor
168 macros are used to prevent speculative loading of data.
169
170 Spectre variant 1 (swapgs)
171 ~~~~~~~~~~~~~~~~~~~~~~~~~~
172
173 An attacker can train the branch predictor to speculatively skip the
174 swapgs path for an interrupt or exception. If they initialize
175 the GS register to a user-space value, if the swapgs is speculatively
176 skipped, subsequent GS-related percpu accesses in the speculation
177 window will be done with the attacker-controlled GS value. This
178 could cause privileged memory to be accessed and leaked.
179
180 For example:
181
182 ::
183
184 if (coming from user space)
185 swapgs
186 mov %gs:<percpu_offset>, %reg
187 mov (%reg), %reg1
188
189 When coming from user space, the CPU can speculatively skip the
190 swapgs, and then do a speculative percpu load using the user GS
191 value. So the user can speculatively force a read of any kernel
192 value. If a gadget exists which uses the percpu value as an address
193 in another load/store, then the contents of the kernel value may
194 become visible via an L1 side channel attack.
195
196 A similar attack exists when coming from kernel space. The CPU can
197 speculatively do the swapgs, causing the user GS to get used for the
198 rest of the speculative window.
199
200 Spectre variant 2
201 ~~~~~~~~~~~~~~~~~
202
203 A spectre variant 2 attacker can :ref:`poison <poison_btb>` the branch
204 target buffer (BTB) before issuing syscall to launch an attack.
205 After entering the kernel, the kernel could use the poisoned branch
206 target buffer on indirect jump and jump to gadget code in speculative
207 execution.
208
209 If an attacker tries to control the memory addresses leaked during
210 speculative execution, he would also need to pass a parameter to the
211 gadget, either through a register or a known address in memory. After
212 the gadget has executed, he can measure the side effect.
213
214 The kernel can protect itself against consuming poisoned branch
215 target buffer entries by using return trampolines (also known as
216 "retpoline") :ref:`[3] <spec_ref3>` :ref:`[9] <spec_ref9>` for all
217 indirect branches. Return trampolines trap speculative execution paths
218 to prevent jumping to gadget code during speculative execution.
219 x86 CPUs with Enhanced Indirect Branch Restricted Speculation
220 (Enhanced IBRS) available in hardware should use the feature to
221 mitigate Spectre variant 2 instead of retpoline. Enhanced IBRS is
222 more efficient than retpoline.
223
224 There may be gadget code in firmware which could be exploited with
225 Spectre variant 2 attack by a rogue user process. To mitigate such
226 attacks on x86, Indirect Branch Restricted Speculation (IBRS) feature
227 is turned on before the kernel invokes any firmware code.
228
229 2. A user process attacking another user process
230 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
231
232 A malicious user process can try to attack another user process,
233 either via a context switch on the same hardware thread, or from the
234 sibling hyperthread sharing a physical processor core on simultaneous
235 multi-threading (SMT) system.
236
237 Spectre variant 1 attacks generally require passing parameters
238 between the processes, which needs a data passing relationship, such
239 as remote procedure calls (RPC). Those parameters are used in gadget
240 code to derive invalid data pointers accessing privileged memory in
241 the attacked process.
242
243 Spectre variant 2 attacks can be launched from a rogue process by
244 :ref:`poisoning <poison_btb>` the branch target buffer. This can
245 influence the indirect branch targets for a victim process that either
246 runs later on the same hardware thread, or running concurrently on
247 a sibling hardware thread sharing the same physical core.
248
249 A user process can protect itself against Spectre variant 2 attacks
250 by using the prctl() syscall to disable indirect branch speculation
251 for itself. An administrator can also cordon off an unsafe process
252 from polluting the branch target buffer by disabling the process's
253 indirect branch speculation. This comes with a performance cost
254 from not using indirect branch speculation and clearing the branch
255 target buffer. When SMT is enabled on x86, for a process that has
256 indirect branch speculation disabled, Single Threaded Indirect Branch
257 Predictors (STIBP) :ref:`[4] <spec_ref4>` are turned on to prevent the
258 sibling thread from controlling branch target buffer. In addition,
259 the Indirect Branch Prediction Barrier (IBPB) is issued to clear the
260 branch target buffer when context switching to and from such process.
261
262 On x86, the return stack buffer is stuffed on context switch.
263 This prevents the branch target buffer from being used for branch
264 prediction when the return stack buffer underflows while switching to
265 a deeper call stack. Any poisoned entries in the return stack buffer
266 left by the previous process will also be cleared.
267
268 User programs should use address space randomization to make attacks
269 more difficult (Set /proc/sys/kernel/randomize_va_space = 1 or 2).
270
271 3. A virtualized guest attacking the host
272 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
273
274 The attack mechanism is similar to how user processes attack the
275 kernel. The kernel is entered via hyper-calls or other virtualization
276 exit paths.
277
278 For Spectre variant 1 attacks, rogue guests can pass parameters
279 (e.g. in registers) via hyper-calls to derive invalid pointers to
280 speculate into privileged memory after entering the kernel. For places
281 where such kernel code has been identified, nospec accessor macros
282 are used to stop speculative memory access.
283
284 For Spectre variant 2 attacks, rogue guests can :ref:`poison
285 <poison_btb>` the branch target buffer or return stack buffer, causing
286 the kernel to jump to gadget code in the speculative execution paths.
287
288 To mitigate variant 2, the host kernel can use return trampolines
289 for indirect branches to bypass the poisoned branch target buffer,
290 and flushing the return stack buffer on VM exit. This prevents rogue
291 guests from affecting indirect branching in the host kernel.
292
293 To protect host processes from rogue guests, host processes can have
294 indirect branch speculation disabled via prctl(). The branch target
295 buffer is cleared before context switching to such processes.
296
297 4. A virtualized guest attacking other guest
298 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
299
300 A rogue guest may attack another guest to get data accessible by the
301 other guest.
302
303 Spectre variant 1 attacks are possible if parameters can be passed
304 between guests. This may be done via mechanisms such as shared memory
305 or message passing. Such parameters could be used to derive data
306 pointers to privileged data in guest. The privileged data could be
307 accessed by gadget code in the victim's speculation paths.
308
309 Spectre variant 2 attacks can be launched from a rogue guest by
310 :ref:`poisoning <poison_btb>` the branch target buffer or the return
311 stack buffer. Such poisoned entries could be used to influence
312 speculation execution paths in the victim guest.
313
314 Linux kernel mitigates attacks to other guests running in the same
315 CPU hardware thread by flushing the return stack buffer on VM exit,
316 and clearing the branch target buffer before switching to a new guest.
317
318 If SMT is used, Spectre variant 2 attacks from an untrusted guest
319 in the sibling hyperthread can be mitigated by the administrator,
320 by turning off the unsafe guest's indirect branch speculation via
321 prctl(). A guest can also protect itself by turning on microcode
322 based mitigations (such as IBPB or STIBP on x86) within the guest.
323
324 .. _spectre_sys_info:
325
326 Spectre system information
327 --------------------------
328
329 The Linux kernel provides a sysfs interface to enumerate the current
330 mitigation status of the system for Spectre: whether the system is
331 vulnerable, and which mitigations are active.
332
333 The sysfs file showing Spectre variant 1 mitigation status is:
334
335 /sys/devices/system/cpu/vulnerabilities/spectre_v1
336
337 The possible values in this file are:
338
339 .. list-table::
340
341 * - 'Not affected'
342 - The processor is not vulnerable.
343 * - 'Vulnerable: __user pointer sanitization and usercopy barriers only; no swapgs barriers'
344 - The swapgs protections are disabled; otherwise it has
345 protection in the kernel on a case by case base with explicit
346 pointer sanitation and usercopy LFENCE barriers.
347 * - 'Mitigation: usercopy/swapgs barriers and __user pointer sanitization'
348 - Protection in the kernel on a case by case base with explicit
349 pointer sanitation, usercopy LFENCE barriers, and swapgs LFENCE
350 barriers.
351
352 However, the protections are put in place on a case by case basis,
353 and there is no guarantee that all possible attack vectors for Spectre
354 variant 1 are covered.
355
356 The spectre_v2 kernel file reports if the kernel has been compiled with
357 retpoline mitigation or if the CPU has hardware mitigation, and if the
358 CPU has support for additional process-specific mitigation.
359
360 This file also reports CPU features enabled by microcode to mitigate
361 attack between user processes:
362
363 1. Indirect Branch Prediction Barrier (IBPB) to add additional
364 isolation between processes of different users.
365 2. Single Thread Indirect Branch Predictors (STIBP) to add additional
366 isolation between CPU threads running on the same core.
367
368 These CPU features may impact performance when used and can be enabled
369 per process on a case-by-case base.
370
371 The sysfs file showing Spectre variant 2 mitigation status is:
372
373 /sys/devices/system/cpu/vulnerabilities/spectre_v2
374
375 The possible values in this file are:
376
377 - Kernel status:
378
379 ======================================== =================================
380 'Not affected' The processor is not vulnerable
381 'Mitigation: None' Vulnerable, no mitigation
382 'Mitigation: Retpolines' Use Retpoline thunks
383 'Mitigation: LFENCE' Use LFENCE instructions
384 'Mitigation: Enhanced IBRS' Hardware-focused mitigation
385 'Mitigation: Enhanced IBRS + Retpolines' Hardware-focused + Retpolines
386 'Mitigation: Enhanced IBRS + LFENCE' Hardware-focused + LFENCE
387 ======================================== =================================
388
389 - Firmware status: Show if Indirect Branch Restricted Speculation (IBRS) is
390 used to protect against Spectre variant 2 attacks when calling firmware (x86 only).
391
392 ========== =============================================================
393 'IBRS_FW' Protection against user program attacks when calling firmware
394 ========== =============================================================
395
396 - Indirect branch prediction barrier (IBPB) status for protection between
397 processes of different users. This feature can be controlled through
398 prctl() per process, or through kernel command line options. This is
399 an x86 only feature. For more details see below.
400
401 =================== ========================================================
402 'IBPB: disabled' IBPB unused
403 'IBPB: always-on' Use IBPB on all tasks
404 'IBPB: conditional' Use IBPB on SECCOMP or indirect branch restricted tasks
405 =================== ========================================================
406
407 - Single threaded indirect branch prediction (STIBP) status for protection
408 between different hyper threads. This feature can be controlled through
409 prctl per process, or through kernel command line options. This is x86
410 only feature. For more details see below.
411
412 ==================== ========================================================
413 'STIBP: disabled' STIBP unused
414 'STIBP: forced' Use STIBP on all tasks
415 'STIBP: conditional' Use STIBP on SECCOMP or indirect branch restricted tasks
416 ==================== ========================================================
417
418 - Return stack buffer (RSB) protection status:
419
420 ============= ===========================================
421 'RSB filling' Protection of RSB on context switch enabled
422 ============= ===========================================
423
424 - EIBRS Post-barrier Return Stack Buffer (PBRSB) protection status:
425
426 =========================== =======================================================
427 'PBRSB-eIBRS: SW sequence' CPU is affected and protection of RSB on VMEXIT enabled
428 'PBRSB-eIBRS: Vulnerable' CPU is vulnerable
429 'PBRSB-eIBRS: Not affected' CPU is not affected by PBRSB
430 =========================== =======================================================
431
432 - Branch History Injection (BHI) protection status:
433
434 .. list-table::
435
436 * - BHI: Not affected
437 - System is not affected
438 * - BHI: Retpoline
439 - System is protected by retpoline
440 * - BHI: BHI_DIS_S
441 - System is protected by BHI_DIS_S
442 * - BHI: SW loop, KVM SW loop
443 - System is protected by software clearing sequence
444 * - BHI: Vulnerable
445 - System is vulnerable to BHI
446 * - BHI: Vulnerable, KVM: SW loop
447 - System is vulnerable; KVM is protected by software clearing sequence
448
449 Full mitigation might require a microcode update from the CPU
450 vendor. When the necessary microcode is not available, the kernel will
451 report vulnerability.
452
453 Turning on mitigation for Spectre variant 1 and Spectre variant 2
454 -----------------------------------------------------------------
455
456 1. Kernel mitigation
457 ^^^^^^^^^^^^^^^^^^^^
458
459 Spectre variant 1
460 ~~~~~~~~~~~~~~~~~
461
462 For the Spectre variant 1, vulnerable kernel code (as determined
463 by code audit or scanning tools) is annotated on a case by case
464 basis to use nospec accessor macros for bounds clipping :ref:`[2]
465 <spec_ref2>` to avoid any usable disclosure gadgets. However, it may
466 not cover all attack vectors for Spectre variant 1.
467
468 Copy-from-user code has an LFENCE barrier to prevent the access_ok()
469 check from being mis-speculated. The barrier is done by the
470 barrier_nospec() macro.
471
472 For the swapgs variant of Spectre variant 1, LFENCE barriers are
473 added to interrupt, exception and NMI entry where needed. These
474 barriers are done by the FENCE_SWAPGS_KERNEL_ENTRY and
475 FENCE_SWAPGS_USER_ENTRY macros.
476
477 Spectre variant 2
478 ~~~~~~~~~~~~~~~~~
479
480 For Spectre variant 2 mitigation, the compiler turns indirect calls or
481 jumps in the kernel into equivalent return trampolines (retpolines)
482 :ref:`[3] <spec_ref3>` :ref:`[9] <spec_ref9>` to go to the target
483 addresses. Speculative execution paths under retpolines are trapped
484 in an infinite loop to prevent any speculative execution jumping to
485 a gadget.
486
487 To turn on retpoline mitigation on a vulnerable CPU, the kernel
488 needs to be compiled with a gcc compiler that supports the
489 -mindirect-branch=thunk-extern -mindirect-branch-register options.
490 If the kernel is compiled with a Clang compiler, the compiler needs
491 to support -mretpoline-external-thunk option. The kernel config
492 CONFIG_MITIGATION_RETPOLINE needs to be turned on, and the CPU needs
493 to run with the latest updated microcode.
494
495 On Intel Skylake-era systems the mitigation covers most, but not all,
496 cases. See :ref:`[3] <spec_ref3>` for more details.
497
498 On CPUs with hardware mitigation for Spectre variant 2 (e.g. IBRS
499 or enhanced IBRS on x86), retpoline is automatically disabled at run time.
500
501 Systems which support enhanced IBRS (eIBRS) enable IBRS protection once at
502 boot, by setting the IBRS bit, and they're automatically protected against
503 some Spectre v2 variant attacks. The BHB can still influence the choice of
504 indirect branch predictor entry, and although branch predictor entries are
505 isolated between modes when eIBRS is enabled, the BHB itself is not isolated
506 between modes. Systems which support BHI_DIS_S will set it to protect against
507 BHI attacks.
508
509 On Intel's enhanced IBRS systems, this includes cross-thread branch target
510 injections on SMT systems (STIBP). In other words, Intel eIBRS enables
511 STIBP, too.
512
513 AMD Automatic IBRS does not protect userspace, and Legacy IBRS systems clear
514 the IBRS bit on exit to userspace, therefore both explicitly enable STIBP.
515
516 The retpoline mitigation is turned on by default on vulnerable
517 CPUs. It can be forced on or off by the administrator
518 via the kernel command line and sysfs control files. See
519 :ref:`spectre_mitigation_control_command_line`.
520
521 On x86, indirect branch restricted speculation is turned on by default
522 before invoking any firmware code to prevent Spectre variant 2 exploits
523 using the firmware.
524
525 Using kernel address space randomization (CONFIG_RANDOMIZE_BASE=y
526 and CONFIG_SLAB_FREELIST_RANDOM=y in the kernel configuration) makes
527 attacks on the kernel generally more difficult.
528
529 2. User program mitigation
530 ^^^^^^^^^^^^^^^^^^^^^^^^^^
531
532 User programs can mitigate Spectre variant 1 using LFENCE or "bounds
533 clipping". For more details see :ref:`[2] <spec_ref2>`.
534
535 For Spectre variant 2 mitigation, individual user programs
536 can be compiled with return trampolines for indirect branches.
537 This protects them from consuming poisoned entries in the branch
538 target buffer left by malicious software.
539
540 On legacy IBRS systems, at return to userspace, implicit STIBP is disabled
541 because the kernel clears the IBRS bit. In this case, the userspace programs
542 can disable indirect branch speculation via prctl() (See
543 :ref:`Documentation/userspace-api/spec_ctrl.rst <set_spec_ctrl>`).
544 On x86, this will turn on STIBP to guard against attacks from the
545 sibling thread when the user program is running, and use IBPB to
546 flush the branch target buffer when switching to/from the program.
547
548 Restricting indirect branch speculation on a user program will
549 also prevent the program from launching a variant 2 attack
550 on x86. Administrators can change that behavior via the kernel
551 command line and sysfs control files.
552 See :ref:`spectre_mitigation_control_command_line`.
553
554 Programs that disable their indirect branch speculation will have
555 more overhead and run slower.
556
557 User programs should use address space randomization
558 (/proc/sys/kernel/randomize_va_space = 1 or 2) to make attacks more
559 difficult.
560
561 3. VM mitigation
562 ^^^^^^^^^^^^^^^^
563
564 Within the kernel, Spectre variant 1 attacks from rogue guests are
565 mitigated on a case by case basis in VM exit paths. Vulnerable code
566 uses nospec accessor macros for "bounds clipping", to avoid any
567 usable disclosure gadgets. However, this may not cover all variant
568 1 attack vectors.
569
570 For Spectre variant 2 attacks from rogue guests to the kernel, the
571 Linux kernel uses retpoline or Enhanced IBRS to prevent consumption of
572 poisoned entries in branch target buffer left by rogue guests. It also
573 flushes the return stack buffer on every VM exit to prevent a return
574 stack buffer underflow so poisoned branch target buffer could be used,
575 or attacker guests leaving poisoned entries in the return stack buffer.
576
577 To mitigate guest-to-guest attacks in the same CPU hardware thread,
578 the branch target buffer is sanitized by flushing before switching
579 to a new guest on a CPU.
580
581 The above mitigations are turned on by default on vulnerable CPUs.
582
583 To mitigate guest-to-guest attacks from sibling thread when SMT is
584 in use, an untrusted guest running in the sibling thread can have
585 its indirect branch speculation disabled by administrator via prctl().
586
587 The kernel also allows guests to use any microcode based mitigation
588 they choose to use (such as IBPB or STIBP on x86) to protect themselves.
589
590 .. _spectre_mitigation_control_command_line:
591
592 Mitigation control on the kernel command line
593 ---------------------------------------------
594
595 In general the kernel selects reasonable default mitigations for the
596 current CPU.
597
598 Spectre default mitigations can be disabled or changed at the kernel
599 command line with the following options:
600
601 - nospectre_v1
602 - nospectre_v2
603 - spectre_v2={option}
604 - spectre_v2_user={option}
605 - spectre_bhi={option}
606
607 For more details on the available options, refer to Documentation/admin-guide/kernel-parameters.txt
608
609 Mitigation selection guide
610 --------------------------
611
612 1. Trusted userspace
613 ^^^^^^^^^^^^^^^^^^^^
614
615 If all userspace applications are from trusted sources and do not
616 execute externally supplied untrusted code, then the mitigations can
617 be disabled.
618
619 2. Protect sensitive programs
620 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
621
622 For security-sensitive programs that have secrets (e.g. crypto
623 keys), protection against Spectre variant 2 can be put in place by
624 disabling indirect branch speculation when the program is running
625 (See :ref:`Documentation/userspace-api/spec_ctrl.rst <set_spec_ctrl>`).
626
627 3. Sandbox untrusted programs
628 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
629
630 Untrusted programs that could be a source of attacks can be cordoned
631 off by disabling their indirect branch speculation when they are run
632 (See :ref:`Documentation/userspace-api/spec_ctrl.rst <set_spec_ctrl>`).
633 This prevents untrusted programs from polluting the branch target
634 buffer. This behavior can be changed via the kernel command line
635 and sysfs control files. See
636 :ref:`spectre_mitigation_control_command_line`.
637
638 3. High security mode
639 ^^^^^^^^^^^^^^^^^^^^^
640
641 All Spectre variant 2 mitigations can be forced on
642 at boot time for all programs (See the "on" option in
643 :ref:`spectre_mitigation_control_command_line`). This will add
644 overhead as indirect branch speculations for all programs will be
645 restricted.
646
647 On x86, branch target buffer will be flushed with IBPB when switching
648 to a new program. STIBP is left on all the time to protect programs
649 against variant 2 attacks originating from programs running on
650 sibling threads.
651
652 Alternatively, STIBP can be used only when running programs
653 whose indirect branch speculation is explicitly disabled,
654 while IBPB is still used all the time when switching to a new
655 program to clear the branch target buffer (See "ibpb" option in
656 :ref:`spectre_mitigation_control_command_line`). This "ibpb" option
657 has less performance cost than the "on" option, which leaves STIBP
658 on all the time.
659
660 References on Spectre
661 ---------------------
662
663 Intel white papers:
664
665 .. _spec_ref1:
666
667 [1] `Intel analysis of speculative execution side channels <https://newsroom.intel.com/wp-content/uploads/sites/11/2018/01/Intel-Analysis-of-Speculative-Execution-Side-Channels.pdf>`_.
668
669 .. _spec_ref2:
670
671 [2] `Bounds check bypass <https://software.intel.com/security-software-guidance/software-guidance/bounds-check-bypass>`_.
672
673 .. _spec_ref3:
674
675 [3] `Deep dive: Retpoline: A branch target injection mitigation <https://software.intel.com/security-software-guidance/insights/deep-dive-retpoline-branch-target-injection-mitigation>`_.
676
677 .. _spec_ref4:
678
679 [4] `Deep Dive: Single Thread Indirect Branch Predictors <https://software.intel.com/security-software-guidance/insights/deep-dive-single-thread-indirect-branch-predictors>`_.
680
681 AMD white papers:
682
683 .. _spec_ref5:
684
685 [5] `AMD64 technology indirect branch control extension <https://developer.amd.com/wp-content/resources/Architecture_Guidelines_Update_Indirect_Branch_Control.pdf>`_.
686
687 .. _spec_ref6:
688
689 [6] `Software techniques for managing speculation on AMD processors <https://developer.amd.com/wp-content/resources/Managing-Speculation-on-AMD-Processors.pdf>`_.
690
691 ARM white papers:
692
693 .. _spec_ref7:
694
695 [7] `Cache speculation side-channels <https://developer.arm.com/support/arm-security-updates/speculative-processor-vulnerability/download-the-whitepaper>`_.
696
697 .. _spec_ref8:
698
699 [8] `Cache speculation issues update <https://developer.arm.com/support/arm-security-updates/speculative-processor-vulnerability/latest-updates/cache-speculation-issues-update>`_.
700
701 Google white paper:
702
703 .. _spec_ref9:
704
705 [9] `Retpoline: a software construct for preventing branch-target-injection <https://support.google.com/faqs/answer/7625886>`_.
706
707 MIPS white paper:
708
709 .. _spec_ref10:
710
711 [10] `MIPS: response on speculative execution and side channel vulnerabilities <https://www.mips.com/blog/mips-response-on-speculative-execution-and-side-channel-vulnerabilities/>`_.
712
713 Academic papers:
714
715 .. _spec_ref11:
716
717 [11] `Spectre Attacks: Exploiting Speculative Execution <https://spectreattack.com/spectre.pdf>`_.
718
719 .. _spec_ref12:
720
721 [12] `NetSpectre: Read Arbitrary Memory over Network <https://arxiv.org/abs/1807.10535>`_.
722
723 .. _spec_ref13:
724
725 [13] `Spectre Returns! Speculation Attacks using the Return Stack Buffer <https://www.usenix.org/system/files/conference/woot18/woot18-paper-koruyeh.pdf>`_.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.