Re: [PATCH] mm: shmem: fix the shmem large folio allocation for the i915 driver

[Patryk Kowalczyk <patryk@kowalczyk.ws>]

[-- Attachment #1: Type: text/plain, Size: 2182 bytes --]

Hi,
In my tests, the performance drop ranges from a few percent up to 13% in
Unigine Superposition
under heavy memory usage on the CPU Core Ultra 155H with the Xe 128 EU GPU.
Other users have reported performance impact up to 30% on certain workloads.
Please find more  in the regressions reports:
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/14645
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/13845

I believe the change should be backported to all active kernel branches
after version 6.12.

best regards,
Patryk


pon., 28 lip 2025 o 23:44 Andrew Morton <akpm@linux-foundation.org>
napisał(a):

> On Mon, 28 Jul 2025 16:03:53 +0800 Baolin Wang <
> baolin.wang@linux.alibaba.com> wrote:
>
> > After commit acd7ccb284b8 ("mm: shmem: add large folio support for
> tmpfs"),
> > we extend the 'huge=' option to allow any sized large folios for tmpfs,
> > which means tmpfs will allow getting a highest order hint based on the
> size
> > of write() and fallocate() paths, and then will try each allowable large
> order.
> >
> > However, when the i915 driver allocates shmem memory, it doesn't provide
> hint
> > information about the size of the large folio to be allocated, resulting
> in
> > the inability to allocate PMD-sized shmem, which in turn affects GPU
> performance.
> >
> > To fix this issue, add the 'end' information for shmem_read_folio_gfp()
> to help
> > allocate PMD-sized large folios. Additionally, use the maximum
> allocation chunk
> > (via mapping_max_folio_size()) to determine the size of the large folios
> to
> > allocate in the i915 driver.
>
> What is the magnitude of the performance change?
>
> > Fixes: acd7ccb284b8 ("mm: shmem: add large folio support for tmpfs")
> > Reported-by: Patryk Kowalczyk <patryk@kowalczyk.ws>
> > Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > Tested-by: Patryk Kowalczyk <patryk@kowalczyk.ws>
>
> This isn't a regression fix, is it?  acd7ccb284b8 adds a new feature
> and we have now found a flaw in it.
>
> Still, we could bend the rules a little bit and backport this, depends
> on how significant the runtime effect is.
>

[-- Attachment #2: Type: text/html, Size: 3163 bytes --]