* Re: [PATCH] x86/Kconfig: enable ROX also when STRICT_KERNEL_RWX is present
2026-06-25 9:06 [PATCH] x86/Kconfig: enable ROX also when STRICT_KERNEL_RWX is present Richard Weinberger
@ 2026-06-25 9:17 ` Mike Rapoport
2026-06-25 9:19 ` Peter Zijlstra
2026-11-21 15:30 ` Vinny Horgan
2 siblings, 0 replies; 4+ messages in thread
From: Mike Rapoport @ 2026-06-25 9:17 UTC (permalink / raw)
To: Richard Weinberger
Cc: linux-kernel, upstream+x86, peterz, hpa, x86, dave.hansen, bp,
mingo, tglx, stable
On Thu, Jun 25, 2026 at 11:06:27AM +0200, Richard Weinberger wrote:
> Running a kernel with CONFIG_MODULES=n causes the W+X page dectection
> to trigger:
> x86/mm: Found insecure W+X mapping at address 0xffffffffc033a000
>
> The W+X pages come from __its_alloc() with type being EXECMEM_MODULE_TEXT.
> Without ARCH_HAS_EXECMEM_ROX pgprot is PAGE_KERNEL instead of
> PAGE_KERNEL_ROX.
>
> Cc: stable@vger.kernel.org
> Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
> Cc: Mike Rapoport (Microsoft) <rppt@kernel.org>
> Fixes: 47410d839fcda ("x86/Kconfig: only enable ROX cache in execmem when STRICT_MODULE_RWX is set")
> Suggested-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> Signed-off-by: Richard Weinberger <richard@nod.at>
Acked-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> ---
> arch/x86/Kconfig | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> index 0b5f30d769ffb..330ccbf6726ad 100644
> --- a/arch/x86/Kconfig
> +++ b/arch/x86/Kconfig
> @@ -85,7 +85,7 @@ config X86
> select ARCH_HAS_DMA_OPS if GART_IOMMU || XEN
> select ARCH_HAS_EARLY_DEBUG if KGDB
> select ARCH_HAS_ELF_RANDOMIZE
> - select ARCH_HAS_EXECMEM_ROX if X86_64 && STRICT_MODULE_RWX
> + select ARCH_HAS_EXECMEM_ROX if X86_64 && (STRICT_MODULE_RWX || STRICT_KERNEL_RWX)
> select ARCH_HAS_FAST_MULTIPLIER
> select ARCH_HAS_FORTIFY_SOURCE
> select ARCH_HAS_GCOV_PROFILE_ALL
> --
> 2.51.0
>
--
Sincerely yours,
Mike.
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] x86/Kconfig: enable ROX also when STRICT_KERNEL_RWX is present
2026-06-25 9:06 [PATCH] x86/Kconfig: enable ROX also when STRICT_KERNEL_RWX is present Richard Weinberger
2026-06-25 9:17 ` Mike Rapoport
@ 2026-06-25 9:19 ` Peter Zijlstra
2026-11-21 15:30 ` Vinny Horgan
2 siblings, 0 replies; 4+ messages in thread
From: Peter Zijlstra @ 2026-06-25 9:19 UTC (permalink / raw)
To: Richard Weinberger
Cc: linux-kernel, upstream+x86, rppt, hpa, x86, dave.hansen, bp,
mingo, tglx, stable
On Thu, Jun 25, 2026 at 11:06:27AM +0200, Richard Weinberger wrote:
> Running a kernel with CONFIG_MODULES=n causes the W+X page dectection
> to trigger:
> x86/mm: Found insecure W+X mapping at address 0xffffffffc033a000
>
> The W+X pages come from __its_alloc() with type being EXECMEM_MODULE_TEXT.
> Without ARCH_HAS_EXECMEM_ROX pgprot is PAGE_KERNEL instead of
> PAGE_KERNEL_ROX.
>
> Cc: stable@vger.kernel.org
> Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
> Cc: Mike Rapoport (Microsoft) <rppt@kernel.org>
> Fixes: 47410d839fcda ("x86/Kconfig: only enable ROX cache in execmem when STRICT_MODULE_RWX is set")
> Suggested-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> Signed-off-by: Richard Weinberger <richard@nod.at>
> ---
> arch/x86/Kconfig | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> index 0b5f30d769ffb..330ccbf6726ad 100644
> --- a/arch/x86/Kconfig
> +++ b/arch/x86/Kconfig
> @@ -85,7 +85,7 @@ config X86
> select ARCH_HAS_DMA_OPS if GART_IOMMU || XEN
> select ARCH_HAS_EARLY_DEBUG if KGDB
> select ARCH_HAS_ELF_RANDOMIZE
> - select ARCH_HAS_EXECMEM_ROX if X86_64 && STRICT_MODULE_RWX
> + select ARCH_HAS_EXECMEM_ROX if X86_64 && (STRICT_MODULE_RWX || STRICT_KERNEL_RWX)
> select ARCH_HAS_FAST_MULTIPLIER
> select ARCH_HAS_FORTIFY_SOURCE
> select ARCH_HAS_GCOV_PROFILE_ALL
I wonder, is STRICT_KERNEL_RWX=y, MODULE=y, STRICT_MODULE_RWX=n at all
possible? ;-) That seems an eminently insane combo.
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] x86/Kconfig: enable ROX also when STRICT_KERNEL_RWX is present
2026-06-25 9:06 [PATCH] x86/Kconfig: enable ROX also when STRICT_KERNEL_RWX is present Richard Weinberger
2026-06-25 9:17 ` Mike Rapoport
2026-06-25 9:19 ` Peter Zijlstra
@ 2026-11-21 15:30 ` Vinny Horgan
2 siblings, 0 replies; 4+ messages in thread
From: Vinny Horgan @ 2026-11-21 15:30 UTC (permalink / raw)
To: Richard Weinberger
Cc: linux-kernel, upstream+x86, rppt, peterz, hpa, x86, dave.hansen,
bp, mingo, tglx, stable
On Thu, Jun 25, 2026 at 11:06:27AM +0200, Richard Weinberger wrote:
> Running a kernel with CONFIG_MODULES=n causes the W+X page dectection
> to trigger:
> x86/mm: Found insecure W+X mapping at address 0xffffffffc033a000
Same here, on a ThinkPad T490 (i5-8365U; "Spectre V2 : Mitigation:
Enhanced / Automatic IBRS", "ITS: Mitigation: Aligned branch/return
thunks") with a CONFIG_MODULES=n kernel based on 7.3.4:
x86/mm: Found insecure W+X mapping at address 0xffffffffc03f3000
With this patch on top of 7.3.5, the same machine and config boot with
"x86/mm: Checked W+X mappings: passed, no W+X pages found."
Tested-by: Vinny Horgan <vinnyhorgan@gmail.com>
It doesn't seem to have been applied yet: arch/x86/Kconfig in v7.4-rc3,
tip/master and next-20261120 still has the old condition.
QEMU without KVM could not reproduce it: TCG offers no IBRS, so
spectre_v2 falls back to retpolines and no ITS thunk pages are
allocated, with or without indirect_target_selection=force.
I found this, and wrote this reply, with the help of an AI assistant
(Claude) while building a custom kernel for the T490; the results above
are from that machine.
^ permalink raw reply [flat|nested] 4+ messages in thread