[PATCH mm-unstable v3] mm/page_alloc: keep track of free highatomic

[PATCH mm-unstable v3] mm/page_alloc: keep track of free highatomic

[Yu Zhao]

OOM kills due to vastly overestimated free highatomic reserves were
observed:

  ... invoked oom-killer: gfp_mask=0x100cca(GFP_HIGHUSER_MOVABLE), order=0 ...
  Node 0 Normal free:1482936kB boost:0kB min:410416kB low:739404kB high:1068392kB reserved_highatomic:1073152KB ...
  Node 0 Normal: 1292*4kB (ME) 1920*8kB (E) 383*16kB (UE) 220*32kB (ME) 340*64kB (E) 2155*128kB (UE) 3243*256kB (UE) 615*512kB (U) 1*1024kB (M) 0*2048kB 0*4096kB = 1477408kB

The second line above shows that the OOM kill was due to the following
condition:

  free (1482936kB) - reserved_highatomic (1073152kB) = 409784KB < min (410416kB)

And the third line shows there were no free pages in any
MIGRATE_HIGHATOMIC pageblocks, which otherwise would show up as type
'H'. Therefore __zone_watermark_unusable_free() underestimated the
usable free memory by over 1GB, which resulted in the unnecessary OOM
kill above.

The comments in __zone_watermark_unusable_free() warns about the
potential risk, i.e.,

  If the caller does not have rights to reserves below the min
  watermark then subtract the high-atomic reserves. This will
  over-estimate the size of the atomic reserve but it avoids a search.

However, it is possible to keep track of free pages in reserved
highatomic pageblocks with a new per-zone counter nr_free_highatomic
protected by the zone lock, to avoid a search when calculating the
usable free memory. And the cost would be minimal, i.e., simple
arithmetics in the highatomic alloc/free/move paths.

Note that since nr_free_highatomic can be relatively small, using a
per-cpu counter might cause too much drift and defeat its purpose,
in addition to the extra memory overhead.

Reported-by: Link Lin <linkl@google.com>
Signed-off-by: Yu Zhao <yuzhao@google.com>
Acked-by: David Rientjes <rientjes@google.com>
---
 include/linux/mmzone.h |  1 +
 mm/page_alloc.c        | 10 +++++++---
 2 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 2e8c4307c728..5e8f567753bd 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -825,6 +825,7 @@ struct zone {
 	unsigned long watermark_boost;
 
 	unsigned long nr_reserved_highatomic;
+	unsigned long nr_free_highatomic;
 
 	/*
 	 * We don't know if the memory that we're going to allocate will be
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index a78acaae6d9c..372a386f34f5 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -635,6 +635,8 @@ compaction_capture(struct capture_control *capc, struct page *page,
 static inline void account_freepages(struct zone *zone, int nr_pages,
 				     int migratetype)
 {
+	lockdep_assert_held(&zone->lock);
+
 	if (is_migrate_isolate(migratetype))
 		return;
 
@@ -642,6 +644,9 @@ static inline void account_freepages(struct zone *zone, int nr_pages,
 
 	if (is_migrate_cma(migratetype))
 		__mod_zone_page_state(zone, NR_FREE_CMA_PAGES, nr_pages);
+
+	if (is_migrate_highatomic(migratetype))
+		WRITE_ONCE(zone->nr_free_highatomic, zone->nr_free_highatomic + nr_pages);
 }
 
 /* Used for pages not on another list */
@@ -3117,11 +3122,10 @@ static inline long __zone_watermark_unusable_free(struct zone *z,
 
 	/*
 	 * If the caller does not have rights to reserves below the min
-	 * watermark then subtract the high-atomic reserves. This will
-	 * over-estimate the size of the atomic reserve but it avoids a search.
+	 * watermark then subtract the free pages reserved for highatomic.
 	 */
 	if (likely(!(alloc_flags & ALLOC_RESERVES)))
-		unusable_free += z->nr_reserved_highatomic;
+		unusable_free += READ_ONCE(z->nr_free_highatomic);
 
 #ifdef CONFIG_CMA
 	/* If allocation can't use CMA areas don't use free CMA pages */
-- 
2.47.0.163.g1226f6d8fa-goog

Re: [PATCH mm-unstable v3] mm/page_alloc: keep track of free highatomic

[Vlastimil Babka]

On 10/28/24 19:26, Yu Zhao wrote:
> OOM kills due to vastly overestimated free highatomic reserves were
> observed:
> 
>   ... invoked oom-killer: gfp_mask=0x100cca(GFP_HIGHUSER_MOVABLE), order=0 ...
>   Node 0 Normal free:1482936kB boost:0kB min:410416kB low:739404kB high:1068392kB reserved_highatomic:1073152KB ...
>   Node 0 Normal: 1292*4kB (ME) 1920*8kB (E) 383*16kB (UE) 220*32kB (ME) 340*64kB (E) 2155*128kB (UE) 3243*256kB (UE) 615*512kB (U) 1*1024kB (M) 0*2048kB 0*4096kB = 1477408kB
> 
> The second line above shows that the OOM kill was due to the following
> condition:
> 
>   free (1482936kB) - reserved_highatomic (1073152kB) = 409784KB < min (410416kB)
> 
> And the third line shows there were no free pages in any
> MIGRATE_HIGHATOMIC pageblocks, which otherwise would show up as type
> 'H'. Therefore __zone_watermark_unusable_free() underestimated the
> usable free memory by over 1GB, which resulted in the unnecessary OOM
> kill above.
> 
> The comments in __zone_watermark_unusable_free() warns about the
> potential risk, i.e.,
> 
>   If the caller does not have rights to reserves below the min
>   watermark then subtract the high-atomic reserves. This will
>   over-estimate the size of the atomic reserve but it avoids a search.
> 
> However, it is possible to keep track of free pages in reserved
> highatomic pageblocks with a new per-zone counter nr_free_highatomic
> protected by the zone lock, to avoid a search when calculating the
> usable free memory. And the cost would be minimal, i.e., simple
> arithmetics in the highatomic alloc/free/move paths.
> 
> Note that since nr_free_highatomic can be relatively small, using a
> per-cpu counter might cause too much drift and defeat its purpose,
> in addition to the extra memory overhead.
> 
> Reported-by: Link Lin <linkl@google.com>
> Signed-off-by: Yu Zhao <yuzhao@google.com>
> Acked-by: David Rientjes <rientjes@google.com>

Acked-by: Vlastimil Babka <vbabka@suse.cz>

For LTS benefit I'd also add:

Cc: <stable@vger.kernel.org> # v6.12+

> ---
>  include/linux/mmzone.h |  1 +
>  mm/page_alloc.c        | 10 +++++++---
>  2 files changed, 8 insertions(+), 3 deletions(-)
> 
> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> index 2e8c4307c728..5e8f567753bd 100644
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -825,6 +825,7 @@ struct zone {
>  	unsigned long watermark_boost;
>  
>  	unsigned long nr_reserved_highatomic;
> +	unsigned long nr_free_highatomic;
>  
>  	/*
>  	 * We don't know if the memory that we're going to allocate will be
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index a78acaae6d9c..372a386f34f5 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -635,6 +635,8 @@ compaction_capture(struct capture_control *capc, struct page *page,
>  static inline void account_freepages(struct zone *zone, int nr_pages,
>  				     int migratetype)
>  {
> +	lockdep_assert_held(&zone->lock);
> +
>  	if (is_migrate_isolate(migratetype))
>  		return;
>  
> @@ -642,6 +644,9 @@ static inline void account_freepages(struct zone *zone, int nr_pages,
>  
>  	if (is_migrate_cma(migratetype))
>  		__mod_zone_page_state(zone, NR_FREE_CMA_PAGES, nr_pages);
> +
> +	if (is_migrate_highatomic(migratetype))
> +		WRITE_ONCE(zone->nr_free_highatomic, zone->nr_free_highatomic + nr_pages);
>  }
>  
>  /* Used for pages not on another list */
> @@ -3117,11 +3122,10 @@ static inline long __zone_watermark_unusable_free(struct zone *z,
>  
>  	/*
>  	 * If the caller does not have rights to reserves below the min
> -	 * watermark then subtract the high-atomic reserves. This will
> -	 * over-estimate the size of the atomic reserve but it avoids a search.
> +	 * watermark then subtract the free pages reserved for highatomic.
>  	 */
>  	if (likely(!(alloc_flags & ALLOC_RESERVES)))
> -		unusable_free += z->nr_reserved_highatomic;
> +		unusable_free += READ_ONCE(z->nr_free_highatomic);
>  
>  #ifdef CONFIG_CMA
>  	/* If allocation can't use CMA areas don't use free CMA pages */

Re: [PATCH mm-unstable v3] mm/page_alloc: keep track of free highatomic

[Johannes Weiner]

On Mon, Oct 28, 2024 at 12:26:53PM -0600, Yu Zhao wrote:
> OOM kills due to vastly overestimated free highatomic reserves were
> observed:
> 
>   ... invoked oom-killer: gfp_mask=0x100cca(GFP_HIGHUSER_MOVABLE), order=0 ...
>   Node 0 Normal free:1482936kB boost:0kB min:410416kB low:739404kB high:1068392kB reserved_highatomic:1073152KB ...
>   Node 0 Normal: 1292*4kB (ME) 1920*8kB (E) 383*16kB (UE) 220*32kB (ME) 340*64kB (E) 2155*128kB (UE) 3243*256kB (UE) 615*512kB (U) 1*1024kB (M) 0*2048kB 0*4096kB = 1477408kB
> 
> The second line above shows that the OOM kill was due to the following
> condition:
> 
>   free (1482936kB) - reserved_highatomic (1073152kB) = 409784KB < min (410416kB)
> 
> And the third line shows there were no free pages in any
> MIGRATE_HIGHATOMIC pageblocks, which otherwise would show up as type
> 'H'. Therefore __zone_watermark_unusable_free() underestimated the
> usable free memory by over 1GB, which resulted in the unnecessary OOM
> kill above.
> 
> The comments in __zone_watermark_unusable_free() warns about the
> potential risk, i.e.,
> 
>   If the caller does not have rights to reserves below the min
>   watermark then subtract the high-atomic reserves. This will
>   over-estimate the size of the atomic reserve but it avoids a search.
> 
> However, it is possible to keep track of free pages in reserved
> highatomic pageblocks with a new per-zone counter nr_free_highatomic
> protected by the zone lock, to avoid a search when calculating the
> usable free memory. And the cost would be minimal, i.e., simple
> arithmetics in the highatomic alloc/free/move paths.
> 
> Note that since nr_free_highatomic can be relatively small, using a
> per-cpu counter might cause too much drift and defeat its purpose,
> in addition to the extra memory overhead.
> 
> Reported-by: Link Lin <linkl@google.com>
> Signed-off-by: Yu Zhao <yuzhao@google.com>
> Acked-by: David Rientjes <rientjes@google.com>

Acked-by: Johannes Weiner <hannes@cmpxchg.org>

> @@ -642,6 +644,9 @@ static inline void account_freepages(struct zone *zone, int nr_pages,
>  
>  	if (is_migrate_cma(migratetype))
>  		__mod_zone_page_state(zone, NR_FREE_CMA_PAGES, nr_pages);
> +
> +	if (is_migrate_highatomic(migratetype))
> +		WRITE_ONCE(zone->nr_free_highatomic, zone->nr_free_highatomic + nr_pages);

Minor nit, the page can only be of one migratetype, so `else if' would
be better.

Re: [PATCH mm-unstable v3] mm/page_alloc: keep track of free highatomic

[Yu Zhao]

On Tue, Oct 29, 2024 at 10:46 AM Johannes Weiner <hannes@cmpxchg.org> wrote:
>
> On Mon, Oct 28, 2024 at 12:26:53PM -0600, Yu Zhao wrote:
> > OOM kills due to vastly overestimated free highatomic reserves were
> > observed:
> >
> >   ... invoked oom-killer: gfp_mask=0x100cca(GFP_HIGHUSER_MOVABLE), order=0 ...
> >   Node 0 Normal free:1482936kB boost:0kB min:410416kB low:739404kB high:1068392kB reserved_highatomic:1073152KB ...
> >   Node 0 Normal: 1292*4kB (ME) 1920*8kB (E) 383*16kB (UE) 220*32kB (ME) 340*64kB (E) 2155*128kB (UE) 3243*256kB (UE) 615*512kB (U) 1*1024kB (M) 0*2048kB 0*4096kB = 1477408kB
> >
> > The second line above shows that the OOM kill was due to the following
> > condition:
> >
> >   free (1482936kB) - reserved_highatomic (1073152kB) = 409784KB < min (410416kB)
> >
> > And the third line shows there were no free pages in any
> > MIGRATE_HIGHATOMIC pageblocks, which otherwise would show up as type
> > 'H'. Therefore __zone_watermark_unusable_free() underestimated the
> > usable free memory by over 1GB, which resulted in the unnecessary OOM
> > kill above.
> >
> > The comments in __zone_watermark_unusable_free() warns about the
> > potential risk, i.e.,
> >
> >   If the caller does not have rights to reserves below the min
> >   watermark then subtract the high-atomic reserves. This will
> >   over-estimate the size of the atomic reserve but it avoids a search.
> >
> > However, it is possible to keep track of free pages in reserved
> > highatomic pageblocks with a new per-zone counter nr_free_highatomic
> > protected by the zone lock, to avoid a search when calculating the
> > usable free memory. And the cost would be minimal, i.e., simple
> > arithmetics in the highatomic alloc/free/move paths.
> >
> > Note that since nr_free_highatomic can be relatively small, using a
> > per-cpu counter might cause too much drift and defeat its purpose,
> > in addition to the extra memory overhead.
> >
> > Reported-by: Link Lin <linkl@google.com>
> > Signed-off-by: Yu Zhao <yuzhao@google.com>
> > Acked-by: David Rientjes <rientjes@google.com>
>
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>
>
> > @@ -642,6 +644,9 @@ static inline void account_freepages(struct zone *zone, int nr_pages,
> >
> >       if (is_migrate_cma(migratetype))
> >               __mod_zone_page_state(zone, NR_FREE_CMA_PAGES, nr_pages);
> > +
> > +     if (is_migrate_highatomic(migratetype))
> > +             WRITE_ONCE(zone->nr_free_highatomic, zone->nr_free_highatomic + nr_pages);
>
> Minor nit, the page can only be of one migratetype, so `else if' would
> be better.

Right (copied and pasted without thinking).

Andrew, could you please fix this up in place? Thank you!