Re: [PATCH 1/1] mm: handle profiling for fake memory allocations during compaction

[Suren Baghdasaryan <surenb@google.com>]

On Fri, Jun 14, 2024 at 6:19 PM Andrew Morton <akpm@linux-foundation.org> wrote:
>
> On Fri, 14 Jun 2024 16:05:04 -0700 Suren Baghdasaryan <surenb@google.com> wrote:
>
> > During compaction isolated free pages are marked allocated so that they
> > can be split and/or freed. For that, post_alloc_hook() is used inside
> > split_map_pages() and release_free_list(). split_map_pages() marks free
> > pages allocated, splits the pages and then lets alloc_contig_range_noprof()
> > free those pages. release_free_list() marks free pages and immediately
> > frees them. This usage of post_alloc_hook() affect memory allocation
> > profiling because these functions might not be called from an instrumented
> > allocator, therefore current->alloc_tag is NULL and when debugging is
> > enabled (CONFIG_MEM_ALLOC_PROFILING_DEBUG=y) that causes warnings.
>
> It would be helpful to quote the warnings for the changelog.  And a
> Reported-by:/Closes: if appropriate.

This was not really reported anywhere but if someone hit this
condition (it requires compaction to be running) then the warning
would be generated.

>
> I'm assuming we want this in 6.10-rcX?

Yes please. Otherwise someone will report that they are getting this
warning when the system is under memory pressure and
CONFIG_MEM_ALLOC_PROFILING_DEBUG is enabled.

>
> Please help in identifying the Fixes:, for anyone who might be
> backporting allocation profiling.

I think we can mark it as

Fixes: b951aaff5035 ("mm: enable page allocation tagging")

but it's really a separate patch which covers a corner case.
Please let me know if you want me to send a v2 with this tag added.
Thanks,
Suren.


>