kernel/linux.git/arch/x86/include/asm/mwait.h, branch v7.1

x86/idle: Use MONITORX and MWAITX mnemonics in

2025-08-22T11:52:21+00:00

Current minimum required version of binutils is 2.30, which supports MONITORX and MWAITX instruction mnemonics. Replace the byte-wise specification of MONITORX and MWAITX with these proper mnemonics. No functional change intended. Signed-off-by: Uros Bizjak Signed-off-by: Borislav Petkov (AMD) Link: https://lore.kernel.org/20250616083056.157460-1-ubizjak@gmail.com

x86/process: Move the buffer clearing before MONITOR

2025-06-17T15:17:12+00:00

Move the VERW clearing before the MONITOR so that VERW doesn't disarm it and the machine never enters C1. Original idea by Kim Phillips . Suggested-by: Andrew Cooper Signed-off-by: Borislav Petkov (AMD)

x86/bugs: Add a Transient Scheduler Attacks mitigation

2025-06-17T15:17:02+00:00

Add the required features detection glue to bugs.c et all in order to support the TSA mitigation. Co-developed-by: Kim Phillips Signed-off-by: Kim Phillips Signed-off-by: Borislav Petkov (AMD) Reviewed-by: Pawan Gupta

x86/bugs: Rename MDS machinery to something more generic

2025-06-16T16:45:18+00:00

It will be used by other x86 mitigations. No functional changes. Signed-off-by: Borislav Petkov (AMD) Reviewed-by: Pawan Gupta

x86/idle: Use MONITOR and MWAIT mnemonics in

2025-04-03T14:28:38+00:00

Current minimum required version of binutils is 2.25, which supports MONITOR and MWAIT instruction mnemonics. Replace the byte-wise specification of MONITOR and MWAIT with these proper mnemonics. No functional change intended. Note: LLVM assembler is not able to assemble correct forms of MONITOR and MWAIT instructions with explicit operands and reports: error: invalid operand for instruction monitor %rax,%ecx,%edx ^~~~ # https://lore.kernel.org/oe-kbuild-all/202504030802.2lEVBSpN-lkp@intel.com/ Use instruction mnemonics with implicit operands to work around this issue. Reported-by: kernel test robot Signed-off-by: Uros Bizjak Signed-off-by: Ingo Molnar Acked-by: Borislav Petkov (AMD) Link: https://lore.kernel.org/r/20250403125111.429805-1-ubizjak@gmail.com

x86/idle: Change arguments of mwait_idle_with_hints() to u32

2025-04-03T14:27:42+00:00

All functions in mwait_idle_with_hints() cast eax and ecx arguments to u32. Propagate argument type to the enclosing function. Suggested-by: Andrew Cooper Signed-off-by: Uros Bizjak Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20250403073105.245987-1-ubizjak@gmail.com

x86/idle: Remove CONFIG_AS_TPAUSE

2025-04-03T11:19:18+00:00

There is not much point in CONFIG_AS_TPAUSE at all when the emitted assembly is always the same - it only obfuscates the __tpause() code in essence. Remove the TPAUSE insn mnemonic from __tpause() and leave only the equivalent byte-wise definition. This can then be changed back to insn mnemonic once binutils 2.31.1 is the minimum version to build the kernel. (Right now it's 2.25.) Signed-off-by: Uros Bizjak Signed-off-by: Ingo Molnar Cc: Andy Lutomirski Cc: Brian Gerst Cc: Juergen Gross Cc: Andrew Cooper Cc: Rik van Riel Cc: "H. Peter Anvin" Cc: Peter Zijlstra Cc: Rafael J. Wysocki Cc: Linus Torvalds Link: https://lore.kernel.org/r/20250402180827.3762-4-ubizjak@gmail.com

x86/idle: Remove .s output beautifying delimiters from simpler asm() templates

2025-04-03T11:19:18+00:00

Delimiters in asm() templates such as ';', '\t' or '\n' are not required syntactically, they were used historically in the Linux kernel to prettify the compiler's .s output for people who were looking at compiler generated .s output. Most x86 developers these days are primarily looking at: 1) objdump --disassemble-all .o 2) perf top's live kernel function annotation and disassembler feature that uses /dev/mem. ... because: - this kind of assembler output is standardized regardless of compiler used, - it's generally less messy looking, - it gives ground-truth instead of being some intermediate layer in the toolchain that might or might not be the real deal, - and on a live kernel it also sees through the kernel's various layers of runtime patching code obfuscation facilities, also known as: alternative-instructions, tracepoints and jump labels. There are some cases where the .s output is the most useful tool, such as alternatives() code generation, but other than that these delimiters used in simple asm() statements mostly add noise to the source code side, which isn't desirable for assembly code that is fragile enough already. Remove the delimiters for , which also happens to make the GCC inliner's asm() instruction length heuristics more accurate... [ mingo: Wrote a new changelog to give historic context and to give people a chance to object. :-) ] Signed-off-by: Uros Bizjak Signed-off-by: Ingo Molnar Cc: Andy Lutomirski Cc: Brian Gerst Cc: Juergen Gross Cc: Andrew Cooper Cc: Rik van Riel Cc: "H. Peter Anvin" Cc: Peter Zijlstra Cc: Rafael J. Wysocki Cc: Linus Torvalds Link: https://lore.kernel.org/r/20250402180827.3762-3-ubizjak@gmail.com

x86/idle: Standardize argument types for MONITOR{,X} and MWAIT{,X} instruction wrappers on 'u32'

2025-04-02T20:26:17+00:00

MONITOR and MONITORX expect 32-bit unsigned integer arguments in the %ecx and %edx registers. MWAIT and MWAITX expect 32-bit usigned int argument in %eax and %ecx registers. Some of the helpers around these instructions in are using too wide types (long), standardize on u32 instead that makes it clear that this is a hardware ABI. [ mingo: Cleaned up the changelog. ] Signed-off-by: Uros Bizjak Signed-off-by: Ingo Molnar Cc: Andy Lutomirski Cc: Brian Gerst Cc: Juergen Gross Cc: Andrew Cooper Cc: Rik van Riel Cc: "H. Peter Anvin" Cc: Peter Zijlstra Cc: Rafael J. Wysocki Cc: Linus Torvalds Link: https://lore.kernel.org/r/20250402180827.3762-1-ubizjak@gmail.com

x86/idle: Remove MFENCEs for X86_BUG_CLFLUSH_MONITOR in mwait_idle_with_hints() and prefer_mwait_c1_over_halt()

2025-04-02T20:02:26+00:00

The following commit, 12 years ago: 7e98b7192046 ("x86, idle: Use static_cpu_has() for CLFLUSH workaround, add barriers") added barriers around the CLFLUSH in mwait_idle_with_hints(), justified with: ... and add memory barriers around it since the documentation is explicit that CLFLUSH is only ordered with respect to MFENCE. This also triggered, 11 years ago, the same adjustment in: f8e617f45829 ("sched/idle/x86: Optimize unnecessary mwait_idle() resched IPIs") during development, although it failed to get the static_cpu_has_bug() treatment. X86_BUG_CLFLUSH_MONITOR (a.k.a the AAI65 errata) is specific to Intel CPUs, and the SDM currently states: Executions of the CLFLUSH instruction are ordered with respect to each other and with respect to writes, locked read-modify-write instructions, and fence instructions[1]. With footnote 1 reading: Earlier versions of this manual specified that executions of the CLFLUSH instruction were ordered only by the MFENCE instruction. All processors implementing the CLFLUSH instruction also order it relative to the other operations enumerated above. i.e. The SDM was incorrect at the time, and barriers should not have been inserted. Double checking the original AAI65 errata (not available from intel.com any more) shows no mention of barriers either. Note: If this were a general codepath, the MFENCEs would be needed, because AMD CPUs of the same vintage do sport otherwise-unordered CLFLUSHs. Remove the unnecessary barriers. Furthermore, use a plain alternative(), rather than static_cpu_has_bug() and/or no optimisation. The workaround is a single instruction. Use an explicit %rax pointer rather than a general memory operand, because MONITOR takes the pointer implicitly in the same way. [ mingo: Cleaned up the commit a bit. ] Fixes: 7e98b7192046 ("x86, idle: Use static_cpu_has() for CLFLUSH workaround, add barriers") Signed-off-by: Andrew Cooper Signed-off-by: Ingo Molnar Acked-by: Dave Hansen Acked-by: Borislav Petkov (AMD) Cc: "H. Peter Anvin" Cc: Peter Zijlstra Cc: Rik van Riel Cc: Linus Torvalds Cc: Andy Lutomirski Cc: Brian Gerst Cc: Juergen Gross Cc: Rafael J. Wysocki Link: https://lore.kernel.org/r/20250402172458.1378112-1-andrew.cooper3@citrix.com