[PATCH hotfix 6.12 v2 4/8] mm: resolve faulty mmap_region() error path behaviour

[Lorenzo Stoakes]

On Tue, Oct 29, 2024 at 08:50:11AM +0100, Vlastimil Babka wrote:
> On 10/28/24 23:14, Lorenzo Stoakes wrote:

<snip>

> > Unfortunately we can't do this and have everything work entirely
> > consistently.
>
> Consistently with the current implementation, which seems inconsiscent wrt
> hugetlbfs.

Yeah 100% I think this is an oversight.

<snip>

> > This is likely an oversight but we want to try to keep behaviour identical
> > to before in this patch.
>
> If it's confirmed to be oversight and MAP_SHARED hugetlbfs should be
> allowed, we could add another is_file_hugepages() condition

Yeah I mean I think this _is_ an oversight but I think best to delay a fix for
that to a follow-up series and just do this for the hotfix. Might throw in a
comment.

Andrew requested a comment on the shmem_file() check anyway so can sling this in
there!

>
> > So continue to check VM_MTE_ALLOWED which arch_calc_vm_flag_bits() sets if
> > MAP_ANON.
> >
> > Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes at oracle.com>
>
> Acked-by: Vlastimil Babka <vbabka at suse.cz>

Thanks!

<snip>