ACK: [SRU][B][PATCH 0/1] revert v4.17 stable backport patch
Colin Ian King
colin.king at canonical.com
Fri Oct 8 10:31:10 UTC 2021
On 08/10/2021 11:00, Krzysztof Kozlowski wrote:
> BugLink: https://bugs.launchpad.net/bugs/1946464
>
> [Impact]
>
> Commit 5047ef5725ed ("x86/mm: Don't free P4D table when it is folded at runtime") should not be backported to v4.15 because:
> 1. It adds pgtable_l5_enabled which does not exist in v4.15,
> 2. It is marked as v4.17+
> 3. It is marked as fixing commit which does not exist in v4.15 and was not backported there.
>
> allyesconfig in x86_64 fails to build:
> ---
> In file included from ../arch/x86/include/asm/mmu_context.h:12,
> from ../include/linux/mmu_context.h:5,
> from ../arch/x86/kvm/../../../virt/kvm/async_pf.c:26:
> ../arch/x86/include/asm/pgalloc.h: In function ‘p4d_free’:
> ../arch/x86/include/asm/pgalloc.h:185:7: error: ‘pgtable_l5_enabled’ undeclared (first use in this function); did you mean ‘movable_node_enabled’?
> 185 | if (!pgtable_l5_enabled)
> | ^~~~~~~~~~~~~~~~~~
> | movable_node_enabled
>
> ---
>
> [Test Plan]
>
> Compile the x86_64 KVM code.
>
> [Where problems could occur]
>
> If backport commit is actually needed, e.g. when we backport 5-level page tables and commit 98219dda2ab5 ("x86/mm: Fold p4d page table layer at runtime"), this could cause double-free and memory corruption.
>
> When backporting x86_64 5-level pages, we need to bring this commit as well.
>
> Best regards,
> Krzysztof
>
> Krzysztof Kozlowski (1):
> UBUNTU: SAUCE: Revert "x86/mm: Don't free P4D table when it is folded
> at runtime"
>
> arch/x86/include/asm/pgalloc.h | 3 ---
> 1 file changed, 3 deletions(-)
>
Yep, makes perfect sense, thanks Krzysztof
Acked-by: Colin Ian King <colin.king at canonical.com>
More information about the kernel-team
mailing list