| Message ID | 1684737363-31554-1-git-send-email-zhaoyang.huang@unisoc.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [PATCHv4] mm: skip CMA pages when they are not available | expand |
On Mon, May 22, 2023 at 02:36:03PM +0800, zhaoyang.huang wrote: > +#ifdef CONFIG_CMA > +/* > + * It is waste of effort to scan and reclaim CMA pages if it is not available > + * for current allocation context > + */ > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > +{ > + if (!current_is_kswapd() && > + gfp_migratetype(sc->gfp_mask) != MIGRATE_MOVABLE && > + get_pageblock_migratetype(&folio->page) == MIGRATE_CMA) > + return true; > + return false; > +} > +#else > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > +{ > + return false; > +} > +#endif > + > /* > * Isolating page from the lruvec to fill in @dst list by nr_to_scan times. > * > @@ -2239,7 +2259,8 @@ static unsigned long isolate_lru_folios(unsigned long nr_to_scan, > nr_pages = folio_nr_pages(folio); > total_scan += nr_pages; > > - if (folio_zonenum(folio) > sc->reclaim_idx) { > + if (folio_zonenum(folio) > sc->reclaim_idx || > + skip_cma(folio, sc)) { > nr_skipped[folio_zonenum(folio)] += nr_pages; > move_to = &folios_skipped; > goto move; I have no idea if what this patch is trying to accomplish is correct, but I no longer object to how it is doing it.
On Fri, May 26, 2023 at 4:03 AM Matthew Wilcox <willy@infradead.org> wrote: > > On Mon, May 22, 2023 at 02:36:03PM +0800, zhaoyang.huang wrote: > > +#ifdef CONFIG_CMA > > +/* > > + * It is waste of effort to scan and reclaim CMA pages if it is not available > > + * for current allocation context > > + */ > > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > > +{ > > + if (!current_is_kswapd() && > > + gfp_migratetype(sc->gfp_mask) != MIGRATE_MOVABLE && > > + get_pageblock_migratetype(&folio->page) == MIGRATE_CMA) > > + return true; > > + return false; > > +} > > +#else > > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > > +{ > > + return false; > > +} > > +#endif > > + > > /* > > * Isolating page from the lruvec to fill in @dst list by nr_to_scan times. > > * > > @@ -2239,7 +2259,8 @@ static unsigned long isolate_lru_folios(unsigned long nr_to_scan, > > nr_pages = folio_nr_pages(folio); > > total_scan += nr_pages; > > > > - if (folio_zonenum(folio) > sc->reclaim_idx) { > > + if (folio_zonenum(folio) > sc->reclaim_idx || > > + skip_cma(folio, sc)) { > > nr_skipped[folio_zonenum(folio)] += nr_pages; > > move_to = &folios_skipped; > > goto move; > > I have no idea if what this patch is trying to accomplish is correct, > but I no longer object to how it is doing it. IMO, this is necessary as there could be such weird scenario, that is an GFP_KERNEL allocation might get 32 MIGRATE_CMA pages via direct_reclaim which lead to a low PSI_MEM/vmpressure value but return a NULL pointer
On 22.05.23 08:36, zhaoyang.huang wrote: > From: Zhaoyang Huang <zhaoyang.huang@unisoc.com> > > This patch fixes unproductive reclaiming of CMA pages by skipping them when they > are not available for current context. It is arise from bellowing OOM issue, which > caused by large proportion of MIGRATE_CMA pages among free pages. > > [ 36.172486] [03-19 10:05:52.172] ActivityManager: page allocation failure: order:0, mode:0xc00(GFP_NOIO), nodemask=(null),cpuset=foreground,mems_allowed=0 > [ 36.189447] [03-19 10:05:52.189] DMA32: 0*4kB 447*8kB (C) 217*16kB (C) 124*32kB (C) 136*64kB (C) 70*128kB (C) 22*256kB (C) 3*512kB (C) 0*1024kB 0*2048kB 0*4096kB = 35848kB > [ 36.193125] [03-19 10:05:52.193] Normal: 231*4kB (UMEH) 49*8kB (MEH) 14*16kB (H) 13*32kB (H) 8*64kB (H) 2*128kB (H) 0*256kB 1*512kB (H) 0*1024kB 0*2048kB 0*4096kB = 3236kB > ... > [ 36.234447] [03-19 10:05:52.234] SLUB: Unable to allocate memory on node -1, gfp=0xa20(GFP_ATOMIC) > [ 36.234455] [03-19 10:05:52.234] cache: ext4_io_end, object size: 64, buffer size: 64, default order: 0, min order: 0 > [ 36.234459] [03-19 10:05:52.234] node 0: slabs: 53,objs: 3392, free: 0 > > Signed-off-by: Zhaoyang Huang <zhaoyang.huang@unisoc.com> > --- > v2: update commit message and fix build error when CONFIG_CMA is not set > v3,v4: update code and comments > --- > --- > mm/vmscan.c | 23 ++++++++++++++++++++++- > 1 file changed, 22 insertions(+), 1 deletion(-) > > diff --git a/mm/vmscan.c b/mm/vmscan.c > index bd6637f..20facec 100644 > --- a/mm/vmscan.c > +++ b/mm/vmscan.c > @@ -2193,6 +2193,26 @@ static __always_inline void update_lru_sizes(struct lruvec *lruvec, > > } > > +#ifdef CONFIG_CMA > +/* > + * It is waste of effort to scan and reclaim CMA pages if it is not available > + * for current allocation context > + */ /* * Only movable allocations may end up on MIGRATE_CMA pageblocks. If * we're not dealing with a movable allocation, it doesn't make sense to * reclaim from these pageblocks: the reclaimed memory is unusable for * this allocation. */ Did I get it right? > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > +{ > + if (!current_is_kswapd() && > + gfp_migratetype(sc->gfp_mask) != MIGRATE_MOVABLE && > + get_pageblock_migratetype(&folio->page) == MIGRATE_CMA) > + return true; > + return false; return !current_is_kswapd() && gfp_migratetype(sc->gfp_mask) != MIGRATE_MOVABLE && get_pageblock_migratetype(&folio->page) == MIGRATE_CMA;
On Mon, May 22, 2023 at 02:36:03PM +0800, zhaoyang.huang wrote: > From: Zhaoyang Huang <zhaoyang.huang@unisoc.com> > > This patch fixes unproductive reclaiming of CMA pages by skipping them when they > are not available for current context. It is arise from bellowing OOM issue, which > caused by large proportion of MIGRATE_CMA pages among free pages. > > [ 36.172486] [03-19 10:05:52.172] ActivityManager: page allocation failure: order:0, mode:0xc00(GFP_NOIO), nodemask=(null),cpuset=foreground,mems_allowed=0 > [ 36.189447] [03-19 10:05:52.189] DMA32: 0*4kB 447*8kB (C) 217*16kB (C) 124*32kB (C) 136*64kB (C) 70*128kB (C) 22*256kB (C) 3*512kB (C) 0*1024kB 0*2048kB 0*4096kB = 35848kB > [ 36.193125] [03-19 10:05:52.193] Normal: 231*4kB (UMEH) 49*8kB (MEH) 14*16kB (H) 13*32kB (H) 8*64kB (H) 2*128kB (H) 0*256kB 1*512kB (H) 0*1024kB 0*2048kB 0*4096kB = 3236kB > ... > [ 36.234447] [03-19 10:05:52.234] SLUB: Unable to allocate memory on node -1, gfp=0xa20(GFP_ATOMIC) > [ 36.234455] [03-19 10:05:52.234] cache: ext4_io_end, object size: 64, buffer size: 64, default order: 0, min order: 0 > [ 36.234459] [03-19 10:05:52.234] node 0: slabs: 53,objs: 3392, free: 0 > > Signed-off-by: Zhaoyang Huang <zhaoyang.huang@unisoc.com> > --- > v2: update commit message and fix build error when CONFIG_CMA is not set > v3,v4: update code and comments > --- > --- > mm/vmscan.c | 23 ++++++++++++++++++++++- > 1 file changed, 22 insertions(+), 1 deletion(-) > > diff --git a/mm/vmscan.c b/mm/vmscan.c > index bd6637f..20facec 100644 > --- a/mm/vmscan.c > +++ b/mm/vmscan.c > @@ -2193,6 +2193,26 @@ static __always_inline void update_lru_sizes(struct lruvec *lruvec, > > } > > +#ifdef CONFIG_CMA > +/* > + * It is waste of effort to scan and reclaim CMA pages if it is not available > + * for current allocation context > + */ > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > +{ > + if (!current_is_kswapd() && The function is called by isolate_lru_folios which is used by both background and direct reclaims at the same time. And sc->reclaim_idx below to filter unproductive reclaim out is used for both cases but why does the cma is considering only direct reclaim path? > + gfp_migratetype(sc->gfp_mask) != MIGRATE_MOVABLE && > + get_pageblock_migratetype(&folio->page) == MIGRATE_CMA) > + return true; > + return false; > +} > +#else > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > +{ > + return false; > +} > +#endif > + > /* > * Isolating page from the lruvec to fill in @dst list by nr_to_scan times. > * > @@ -2239,7 +2259,8 @@ static unsigned long isolate_lru_folios(unsigned long nr_to_scan, > nr_pages = folio_nr_pages(folio); > total_scan += nr_pages; > > - if (folio_zonenum(folio) > sc->reclaim_idx) { > + if (folio_zonenum(folio) > sc->reclaim_idx || > + skip_cma(folio, sc)) { > nr_skipped[folio_zonenum(folio)] += nr_pages; > move_to = &folios_skipped; > goto move; > -- > 1.9.1 >
On Sat, May 27, 2023 at 3:36 AM David Hildenbrand <david@redhat.com> wrote: > > On 22.05.23 08:36, zhaoyang.huang wrote: > > From: Zhaoyang Huang <zhaoyang.huang@unisoc.com> > > > > This patch fixes unproductive reclaiming of CMA pages by skipping them when they > > are not available for current context. It is arise from bellowing OOM issue, which > > caused by large proportion of MIGRATE_CMA pages among free pages. > > > > [ 36.172486] [03-19 10:05:52.172] ActivityManager: page allocation failure: order:0, mode:0xc00(GFP_NOIO), nodemask=(null),cpuset=foreground,mems_allowed=0 > > [ 36.189447] [03-19 10:05:52.189] DMA32: 0*4kB 447*8kB (C) 217*16kB (C) 124*32kB (C) 136*64kB (C) 70*128kB (C) 22*256kB (C) 3*512kB (C) 0*1024kB 0*2048kB 0*4096kB = 35848kB > > [ 36.193125] [03-19 10:05:52.193] Normal: 231*4kB (UMEH) 49*8kB (MEH) 14*16kB (H) 13*32kB (H) 8*64kB (H) 2*128kB (H) 0*256kB 1*512kB (H) 0*1024kB 0*2048kB 0*4096kB = 3236kB > > ... > > [ 36.234447] [03-19 10:05:52.234] SLUB: Unable to allocate memory on node -1, gfp=0xa20(GFP_ATOMIC) > > [ 36.234455] [03-19 10:05:52.234] cache: ext4_io_end, object size: 64, buffer size: 64, default order: 0, min order: 0 > > [ 36.234459] [03-19 10:05:52.234] node 0: slabs: 53,objs: 3392, free: 0 > > > > Signed-off-by: Zhaoyang Huang <zhaoyang.huang@unisoc.com> > > --- > > v2: update commit message and fix build error when CONFIG_CMA is not set > > v3,v4: update code and comments > > --- > > --- > > mm/vmscan.c | 23 ++++++++++++++++++++++- > > 1 file changed, 22 insertions(+), 1 deletion(-) > > > > diff --git a/mm/vmscan.c b/mm/vmscan.c > > index bd6637f..20facec 100644 > > --- a/mm/vmscan.c > > +++ b/mm/vmscan.c > > @@ -2193,6 +2193,26 @@ static __always_inline void update_lru_sizes(struct lruvec *lruvec, > > > > } > > > > +#ifdef CONFIG_CMA > > +/* > > + * It is waste of effort to scan and reclaim CMA pages if it is not available > > + * for current allocation context > > + */ > > /* > * Only movable allocations may end up on MIGRATE_CMA pageblocks. If > * we're not dealing with a movable allocation, it doesn't make sense to > * reclaim from these pageblocks: the reclaimed memory is unusable for > * this allocation. > */ > > Did I get it right? Yes, it is right. > > > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > > +{ > > + if (!current_is_kswapd() && > > + gfp_migratetype(sc->gfp_mask) != MIGRATE_MOVABLE && > > + get_pageblock_migratetype(&folio->page) == MIGRATE_CMA) > > + return true; > > + return false; > > return !current_is_kswapd() && > gfp_migratetype(sc->gfp_mask) != MIGRATE_MOVABLE && > get_pageblock_migratetype(&folio->page) == MIGRATE_CMA; ok, thanks > > > -- > Thanks, > > David / dhildenb >
On Sat, May 27, 2023 at 7:03 AM Minchan Kim <minchan@kernel.org> wrote: > > On Mon, May 22, 2023 at 02:36:03PM +0800, zhaoyang.huang wrote: > > From: Zhaoyang Huang <zhaoyang.huang@unisoc.com> > > > > This patch fixes unproductive reclaiming of CMA pages by skipping them when they > > are not available for current context. It is arise from bellowing OOM issue, which > > caused by large proportion of MIGRATE_CMA pages among free pages. > > > > [ 36.172486] [03-19 10:05:52.172] ActivityManager: page allocation failure: order:0, mode:0xc00(GFP_NOIO), nodemask=(null),cpuset=foreground,mems_allowed=0 > > [ 36.189447] [03-19 10:05:52.189] DMA32: 0*4kB 447*8kB (C) 217*16kB (C) 124*32kB (C) 136*64kB (C) 70*128kB (C) 22*256kB (C) 3*512kB (C) 0*1024kB 0*2048kB 0*4096kB = 35848kB > > [ 36.193125] [03-19 10:05:52.193] Normal: 231*4kB (UMEH) 49*8kB (MEH) 14*16kB (H) 13*32kB (H) 8*64kB (H) 2*128kB (H) 0*256kB 1*512kB (H) 0*1024kB 0*2048kB 0*4096kB = 3236kB > > ... > > [ 36.234447] [03-19 10:05:52.234] SLUB: Unable to allocate memory on node -1, gfp=0xa20(GFP_ATOMIC) > > [ 36.234455] [03-19 10:05:52.234] cache: ext4_io_end, object size: 64, buffer size: 64, default order: 0, min order: 0 > > [ 36.234459] [03-19 10:05:52.234] node 0: slabs: 53,objs: 3392, free: 0 > > > > Signed-off-by: Zhaoyang Huang <zhaoyang.huang@unisoc.com> > > --- > > v2: update commit message and fix build error when CONFIG_CMA is not set > > v3,v4: update code and comments > > --- > > --- > > mm/vmscan.c | 23 ++++++++++++++++++++++- > > 1 file changed, 22 insertions(+), 1 deletion(-) > > > > diff --git a/mm/vmscan.c b/mm/vmscan.c > > index bd6637f..20facec 100644 > > --- a/mm/vmscan.c > > +++ b/mm/vmscan.c > > @@ -2193,6 +2193,26 @@ static __always_inline void update_lru_sizes(struct lruvec *lruvec, > > > > } > > > > +#ifdef CONFIG_CMA > > +/* > > + * It is waste of effort to scan and reclaim CMA pages if it is not available > > + * for current allocation context > > + */ > > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > > +{ > > + if (!current_is_kswapd() && > > The function is called by isolate_lru_folios which is used by both background > and direct reclaims at the same time. And sc->reclaim_idx below to filter > unproductive reclaim out is used for both cases but why does the cma is considering > only direct reclaim path? Because kswapd's sc->gfp_mask = GFP_KERNEL which can not distinguish this scenario > > > > + gfp_migratetype(sc->gfp_mask) != MIGRATE_MOVABLE && > > + get_pageblock_migratetype(&folio->page) == MIGRATE_CMA) > > + return true; > > + return false; > > +} > > +#else > > +static bool skip_cma(struct folio *folio, struct scan_control *sc) > > +{ > > + return false; > > +} > > +#endif > > + > > /* > > * Isolating page from the lruvec to fill in @dst list by nr_to_scan times. > > * > > @@ -2239,7 +2259,8 @@ static unsigned long isolate_lru_folios(unsigned long nr_to_scan, > > nr_pages = folio_nr_pages(folio); > > total_scan += nr_pages; > > > > - if (folio_zonenum(folio) > sc->reclaim_idx) { > > + if (folio_zonenum(folio) > sc->reclaim_idx || > > + skip_cma(folio, sc)) { > > nr_skipped[folio_zonenum(folio)] += nr_pages; > > move_to = &folios_skipped; > > goto move; > > -- > > 1.9.1 > >
diff --git a/mm/vmscan.c b/mm/vmscan.c index bd6637f..20facec 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -2193,6 +2193,26 @@ static __always_inline void update_lru_sizes(struct lruvec *lruvec, } +#ifdef CONFIG_CMA +/* + * It is waste of effort to scan and reclaim CMA pages if it is not available + * for current allocation context + */ +static bool skip_cma(struct folio *folio, struct scan_control *sc) +{ + if (!current_is_kswapd() && + gfp_migratetype(sc->gfp_mask) != MIGRATE_MOVABLE && + get_pageblock_migratetype(&folio->page) == MIGRATE_CMA) + return true; + return false; +} +#else +static bool skip_cma(struct folio *folio, struct scan_control *sc) +{ + return false; +} +#endif + /* * Isolating page from the lruvec to fill in @dst list by nr_to_scan times. * @@ -2239,7 +2259,8 @@ static unsigned long isolate_lru_folios(unsigned long nr_to_scan, nr_pages = folio_nr_pages(folio); total_scan += nr_pages; - if (folio_zonenum(folio) > sc->reclaim_idx) { + if (folio_zonenum(folio) > sc->reclaim_idx || + skip_cma(folio, sc)) { nr_skipped[folio_zonenum(folio)] += nr_pages; move_to = &folios_skipped; goto move;