BUILD_BUG() in smaps_account() (was: Re: [PATCHv12 01/37] mm, proc: adjust PSS calculation)

BUILD_BUG() in smaps_account() (was: Re: [PATCHv12 01/37] mm, proc: adjust PSS calculation)

[Geert Uytterhoeven]

Hi Kirill,

On Tue, Oct 6, 2015 at 5:23 PM, Kirill A. Shutemov
<kirill.shutemov@linux.intel.com> wrote:
> With new refcounting all subpages of the compound page are not necessary
> have the same mapcount. We need to take into account mapcount of every
> sub-page.
>
> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> Tested-by: Sasha Levin <sasha.levin@oracle.com>
> Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
> Acked-by: Jerome Marchand <jmarchan@redhat.com>
> Acked-by: Vlastimil Babka <vbabka@suse.cz>
> ---
>  fs/proc/task_mmu.c | 47 +++++++++++++++++++++++++++++++----------------
>  1 file changed, 31 insertions(+), 16 deletions(-)
>
> diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
> index bd167675a06f..ace02a4a07db 100644
> --- a/fs/proc/task_mmu.c
> +++ b/fs/proc/task_mmu.c
> @@ -454,9 +454,10 @@ struct mem_size_stats {
>  };
>
>  static void smaps_account(struct mem_size_stats *mss, struct page *page,
> -               unsigned long size, bool young, bool dirty)
> +               bool compound, bool young, bool dirty)
>  {
> -       int mapcount;
> +       int i, nr = compound ? HPAGE_PMD_NR : 1;

If CONFIG_TRANSPARENT_HUGEPAGE is not set, we have:

    #define HPAGE_PMD_NR (1<<HPAGE_PMD_ORDER)
    #define HPAGE_PMD_ORDER (HPAGE_PMD_SHIFT-PAGE_SHIFT)
    #define HPAGE_PMD_SHIFT ({ BUILD_BUG(); 0; })

Depending on compiler version and optimization level, the BUILD_BUG() may be
optimized away (smaps_account() is always called with compound = false if
CONFIG_TRANSPARENT_HUGEPAGE=n), or lead to a build failure:

    fs/built-in.o: In function `smaps_account':
    task_mmu.c:(.text+0x4f8fa): undefined reference to
`__compiletime_assert_471'

Seen with m68k/allmodconfig or allyesconfig and gcc version 4.1.2 20061115
(prerelease) (Ubuntu 4.1.1-21).
Not seen when compiling the affected file with gcc 4.6.3 or 4.9.0, or with the
m68k defconfigs.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: BUILD_BUG() in smaps_account() (was: Re: [PATCHv12 01/37] mm, proc: adjust PSS calculation)

[Kirill A. Shutemov]

On Mon, Jan 18, 2016 at 11:09:00AM +0100, Geert Uytterhoeven wrote:
> Hi Kirill,
> 
> On Tue, Oct 6, 2015 at 5:23 PM, Kirill A. Shutemov
> <kirill.shutemov@linux.intel.com> wrote:
> > With new refcounting all subpages of the compound page are not necessary
> > have the same mapcount. We need to take into account mapcount of every
> > sub-page.
> >
> > Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> > Tested-by: Sasha Levin <sasha.levin@oracle.com>
> > Tested-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
> > Acked-by: Jerome Marchand <jmarchan@redhat.com>
> > Acked-by: Vlastimil Babka <vbabka@suse.cz>
> > ---
> >  fs/proc/task_mmu.c | 47 +++++++++++++++++++++++++++++++----------------
> >  1 file changed, 31 insertions(+), 16 deletions(-)
> >
> > diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
> > index bd167675a06f..ace02a4a07db 100644
> > --- a/fs/proc/task_mmu.c
> > +++ b/fs/proc/task_mmu.c
> > @@ -454,9 +454,10 @@ struct mem_size_stats {
> >  };
> >
> >  static void smaps_account(struct mem_size_stats *mss, struct page *page,
> > -               unsigned long size, bool young, bool dirty)
> > +               bool compound, bool young, bool dirty)
> >  {
> > -       int mapcount;
> > +       int i, nr = compound ? HPAGE_PMD_NR : 1;
> 
> If CONFIG_TRANSPARENT_HUGEPAGE is not set, we have:
> 
>     #define HPAGE_PMD_NR (1<<HPAGE_PMD_ORDER)
>     #define HPAGE_PMD_ORDER (HPAGE_PMD_SHIFT-PAGE_SHIFT)
>     #define HPAGE_PMD_SHIFT ({ BUILD_BUG(); 0; })
> 
> Depending on compiler version and optimization level, the BUILD_BUG() may be
> optimized away (smaps_account() is always called with compound = false if
> CONFIG_TRANSPARENT_HUGEPAGE=n), or lead to a build failure:
> 
>     fs/built-in.o: In function `smaps_account':
>     task_mmu.c:(.text+0x4f8fa): undefined reference to
> `__compiletime_assert_471'
> 
> Seen with m68k/allmodconfig or allyesconfig and gcc version 4.1.2 20061115
> (prerelease) (Ubuntu 4.1.1-21).
> Not seen when compiling the affected file with gcc 4.6.3 or 4.9.0, or with the
> m68k defconfigs.

Ughh.

Please, test this:

Re: BUILD_BUG() in smaps_account() (was: Re: [PATCHv12 01/37] mm, proc: adjust PSS calculation)

[Geert Uytterhoeven]

Hi Kirill,

On Mon, Jan 18, 2016 at 12:40 PM, Kirill A. Shutemov
<kirill@shutemov.name> wrote:
> On Mon, Jan 18, 2016 at 11:09:00AM +0100, Geert Uytterhoeven wrote:
>>     fs/built-in.o: In function `smaps_account':
>>     task_mmu.c:(.text+0x4f8fa): undefined reference to
>> `__compiletime_assert_471'
>>
>> Seen with m68k/allmodconfig or allyesconfig and gcc version 4.1.2 20061115
>> (prerelease) (Ubuntu 4.1.1-21).
>> Not seen when compiling the affected file with gcc 4.6.3 or 4.9.0, or with the
>> m68k defconfigs.
>
> Ughh.
>
> Please, test this:
>
> From 5ac27019f886eef033e84c9579e09099469ccbf9 Mon Sep 17 00:00:00 2001
> From: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
> Date: Mon, 18 Jan 2016 14:32:49 +0300
> Subject: [PATCH] mm, proc: add workaround for old compilers
>
> For THP=n, HPAGE_PMD_NR in smaps_account() expands to BUILD_BUG().
> That's fine since this codepath is eliminated by modern compilers.
>
> But older compilers have not that efficient dead code elimination.
> It causes problem at least with gcc 4.1.2 on m68k.
>
> Let's replace HPAGE_PMD_NR with 1 << compound_order(page).
>
> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>

Thanks, that fixes it!

Tested-by: Geert Uytterhoeven <geert@linux-m68k.org>

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: BUILD_BUG() in smaps_account() (was: Re: [PATCHv12 01/37] mm, proc: adjust PSS calculation)

[Tony Luck]

On Mon, Jan 18, 2016 at 6:56 AM, Geert Uytterhoeven
<geert@linux-m68k.org> wrote:
> Hi Kirill,
>
> On Mon, Jan 18, 2016 at 12:40 PM, Kirill A. Shutemov
> <kirill@shutemov.name> wrote:
>> On Mon, Jan 18, 2016 at 11:09:00AM +0100, Geert Uytterhoeven wrote:
>>>     fs/built-in.o: In function `smaps_account':
>>>     task_mmu.c:(.text+0x4f8fa): undefined reference to
>>> `__compiletime_assert_471'
>>>
>>> Seen with m68k/allmodconfig or allyesconfig and gcc version 4.1.2 20061115
>>> (prerelease) (Ubuntu 4.1.1-21).
>>> Not seen when compiling the affected file with gcc 4.6.3 or 4.9.0, or with the
>>> m68k defconfigs.
>>
>> Ughh.
>>
>> Please, test this:
>>
>> From 5ac27019f886eef033e84c9579e09099469ccbf9 Mon Sep 17 00:00:00 2001
>> From: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
>> Date: Mon, 18 Jan 2016 14:32:49 +0300
>> Subject: [PATCH] mm, proc: add workaround for old compilers
>>
>> For THP=n, HPAGE_PMD_NR in smaps_account() expands to BUILD_BUG().
>> That's fine since this codepath is eliminated by modern compilers.
>>
>> But older compilers have not that efficient dead code elimination.
>> It causes problem at least with gcc 4.1.2 on m68k.
>>
>> Let's replace HPAGE_PMD_NR with 1 << compound_order(page).
>>
>> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
>> Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
>
> Thanks, that fixes it!
>
> Tested-by: Geert Uytterhoeven <geert@linux-m68k.org>

Same breakage on ia64 (with gcc 4.3.4).  Same fix works for me.

Tested-by: Tony Luck <tony.luck@intel.com>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>