Re: [PATCH] x86/mm: Define the zapped-entry checks only with CONFIG_DEBUG_VM

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] x86/mm: commit title contradicts code and changelog has unstripped scissors
--

commit 1651219a7e4a74c995baa4203be8528bd7cfe1d0
Author: Kiryl Shutsemau <kas@kernel.org>

x86/mm: Define the zapped-entry checks only with CONFIG_DEBUG_VM

This commit removes the arch_check_zapped_pte(), arch_check_zapped_pmd(),
and arch_check_zapped_pud() helpers along with their callers and
pud_shstk() to avoid return mispredictions during page table zapping.

>     x86/mm: Define the zapped-entry checks only with CONFIG_DEBUG_VM
>     
>     On Mon, Nov 02, 2026 at 01:12:47PM -0800, Dave Hansen wrote:
>     > Or, heck, I'm not that attached to these checks. Let's just kill them.
>     
>     Works for me. x86 is the only architecture with them, although arm64 and
>     riscv have shadow stacks too, and the x86 helpers already keep
>     Write=0,Dirty=1 out of other VMAs. Patch below, without the Fixes: tag.
>     
>     Rick, any objection?
>     
>     -- >8 --
>     From: "Kiryl Shutsemau (Meta)" <kas@kernel.org>
>     Subject: [PATCH] x86/mm: Remove the zapped-entry checks
>     
>     Zapping a range calls arch_check_zapped_pte() for every batch of PTEs,
>     which is every page for small folios, and arch_check_zapped_pmd() or
>     arch_check_zapped_pud() for every huge leaf. Only x86 implements them, as
>     out-of-line functions that warn about a shadow stack entry (Write=0,
>     Dirty=1) in a VMA without VM_SHADOW_STACK.
>     
>     The warning is a VM_WARN_ON_ONCE(), so without CONFIG_DEBUG_VM the
>     functions are empty, but the calls remain. With the SRSO "Safe RET"
>     mitigation the return of every such call mispredicts.
>     
>     The x86 helpers already keep such entries out of other VMAs.
>     pte_wrprotect(), pte_mkdirty(), pte_modify() and their PMD versions go
>     through SavedDirty, and only pte_mkwrite() and pmd_mkwrite() on a
>     VM_SHADOW_STACK VMA create the combination.
>     
>     Remove the checks, their callers and pud_shstk(), which has no other
>     user.
>     
>     In a KVM guest on Zen 4 with Safe RET, zapping 4K pages takes one branch
>     miss per page fewer. MADV_DONTNEED of 1 GiB gets 6% faster on anonymous
>     memory and 8% on shmem, munmap() 5%. THP is unchanged. Without Safe RET
>     the difference is within noise.
>     
>     Suggested-by: Dave Hansen <dave.hansen@linux.intel.com>
>     Signed-off-by: Kiryl Shutsemau (Meta) <kas@kernel.org>

[Severity: Medium]
Was this patch applied without passing --scissors to git am?

The commit title still reads "x86/mm: Define the zapped-entry checks only
with CONFIG_DEBUG_VM", whereas the patch unconditionally deletes
arch_check_zapped_pte(), arch_check_zapped_pmd(), and
arch_check_zapped_pud() in arch/x86/mm/pgtable.c, arch/x86/include/asm/pgtable.h,
and include/linux/pgtable.h, along with pud_shstk() in
arch/x86/include/asm/pgtable.h and the callers in zap_present_folio_ptes()
and zap_present_ptes() in mm/memory.c and zap_huge_pmd() and zap_huge_pud()
in mm/huge_memory.c.

In addition, the changelog retains the mailing list reply, the "-- >8 --"
scissors line, and the embedded From and Subject headers ("x86/mm: Remove
the zapped-entry checks"). Should the preceding reply and scissors headers
be stripped and the commit title updated to "x86/mm: Remove the
zapped-entry checks"?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/aumzjHyJsshPJBUI@thinkstation?part=1