[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAJuCfpEnLGi8754DQeumLz0t+d6aXmEPEV64BMC0oRRqr9UX1A@mail.gmail.com>
Date: Wed, 17 Jul 2024 14:20:30 -0700
From: Suren Baghdasaryan <surenb@...gle.com>
To: Vlastimil Babka <vbabka@...e.cz>
Cc: akpm@...ux-foundation.org, kent.overstreet@...ux.dev, hch@...radead.org,
pasha.tatashin@...een.com, souravpanda@...gle.com, keescook@...omium.org,
linux-kernel@...r.kernel.org, linux-mm@...ck.org,
kernel test robot <lkp@...el.com>
Subject: Re: [PATCH v2 2/2] alloc_tag: outline and export free_reserved_page()
On Wed, Jul 17, 2024 at 1:19 PM Vlastimil Babka <vbabka@...e.cz> wrote:
>
> On 7/17/24 10:04 PM, Suren Baghdasaryan wrote:
> > On Wed, Jul 17, 2024 at 12:36 PM Vlastimil Babka <vbabka@...e.cz> wrote:
> >>
> >> On 7/17/24 8:12 PM, Suren Baghdasaryan wrote:
> >> > Outline and export free_reserved_page() because modules use it and it
> >> > in turn uses page_ext_{get|put} which should not be exported. The same
> >> > result could be obtained by outlining {get|put}_page_tag_ref() but that
> >> > would have higher performance impact as these functions are used in
> >> > more performance critical paths.
> >> >
> >> > Fixes: dcfe378c81f7 ("lib: introduce support for page allocation tagging")
> >> > Reported-by: kernel test robot <lkp@...el.com>
> >> > Closes: https://lore.kernel.org/oe-kbuild-all/202407080044.DWMC9N9I-lkp@intel.com/
> >> > Suggested-by: Christoph Hellwig <hch@...radead.org>
> >> > Suggested-by: Vlastimil Babka <vbabka@...e.cz>
> >> > Signed-off-by: Suren Baghdasaryan <surenb@...gle.com>
> >>
> >> Acked-by: Vlastimil Babka <vbabka@...e.cz>
> >>
> >> Are these two patches now stable@ material as 6.10 build is broken on ppc64
> >> with alloc taging enabled?
> >
> > I tested them with that specific configuration mentioned in the bug
> > report and with several usual ones I use.
> > Yeah, I guess from now on all such fixes should have
> >
> > Cc: stable@...r.kernel.org # v6.10
>
> Right. BTW I have just realized that the way you did Patch 2/2 and outlined
> the whole free_reserved_page() (which is fair, it's an init-time function),
> mem_alloc_profiling_enabled() didn't stay inlined so Patch 1/2 is in fact
> not necessary anymore?
Yeah, I think you are right, currently no module has reasons to use
mem_alloc_profiling_enabled() directly. That might change in the
future but we can add the export at the time it's needed.
I checked and ppc64 build passes with just this patch. Let me post v3
with just this patch and Cc stable@.
>
>
>
Powered by blists - more mailing lists