On Wed, Oct 3, 2018 at 6:29 PM Michal Hocko <mhocko@xxxxxxxxxx> wrote:
>
> On Sat 29-09-18 03:36:11, Jann Horn wrote:
> > commit 7a9cdebdcc17 ("mm: get rid of vmacache_flush_all() entirely")
> > removed the VMACACHE_FULL_FLUSHES statistics, but didn't remove the
> > corresponding entry in vmstat_text. This causes an out-of-bounds access in
> > vmstat_show().
> >
> > Luckily this only affects kernels with CONFIG_DEBUG_VM_VMACACHE=y, which is
> > probably very rare.
> >
> > Having two gigantic arrays that must be kept in sync isn't exactly robust.
> > To make it easier to catch such issues in the future, add a BUILD_BUG_ON().
> >
> > Fixes: 7a9cdebdcc17 ("mm: get rid of vmacache_flush_all() entirely")
> > Cc: stable@xxxxxxxxxxxxxxx
> > Signed-off-by: Jann Horn <jannh@xxxxxxxxxx>
>
> Those could be two separate patches but anyway
> Acked-by: Michal Hocko <mhocko@xxxxxxxx>
>
> to both changes. I have burned myself on this in the past as well. Build
> bugon would save me a lot of debugging.
I actually sent a v2 that splits this into two patches, and adds
another fix for nr_tlb_remote_flush and nr_tlb_remote_flush_received
for systems with CONFIG_VM_EVENT_COUNTERS=y && CONFIG_DEBUG_TLBFLUSH=y
&& CONFIG_SMP=n. akpm has already added the v2 patches to the mm tree.
