Re: [PATCH 3/9] mm, page_alloc: remove setup_pageset()
On 10/5/20 2:59 PM, Michal Hocko wrote: On Tue 22-09-20 16:37:06, Vlastimil Babka wrote: We initialize boot-time pagesets with setup_pageset(), which sets high and batch values that effectively disable pcplists. We can remove this wrapper if we just set these values for all pagesets in pageset_init(). Non-boot pagesets then subsequently update them to the proper values. No functional change. Signed-off-by: Vlastimil Babka --- mm/page_alloc.c | 17 ++--- 1 file changed, 10 insertions(+), 7 deletions(-) diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 26069c8d1b19..76c2b4578723 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -5823,7 +5823,7 @@ static void build_zonelists(pg_data_t *pgdat) * not check if the processor is online before following the pageset pointer. * Other parts of the kernel may not check if the zone is available. */ -static void setup_pageset(struct per_cpu_pageset *p); +static void pageset_init(struct per_cpu_pageset *p); static DEFINE_PER_CPU(struct per_cpu_pageset, boot_pageset); static DEFINE_PER_CPU(struct per_cpu_nodestat, boot_nodestats); @@ -5891,7 +5891,7 @@ build_all_zonelists_init(void) * (a chicken-egg dilemma). */ for_each_possible_cpu(cpu) - setup_pageset(_cpu(boot_pageset, cpu)); + pageset_init(_cpu(boot_pageset, cpu)); mminit_verify_zonelist(); cpuset_init_current_mems_allowed(); @@ -6210,12 +6210,15 @@ static void pageset_init(struct per_cpu_pageset *p) pcp = >pcp; for (migratetype = 0; migratetype < MIGRATE_PCPTYPES; migratetype++) INIT_LIST_HEAD(>lists[migratetype]); -} -static void setup_pageset(struct per_cpu_pageset *p) -{ - pageset_init(p); - pageset_update(>pcp, 0, 1); + /* +* Set batch and high values safe for a boot pageset. A true percpu +* pageset's initialization will update them subsequently. Here we don't +* need to be as careful as pageset_update() as nobody can access the +* pageset yet. Isn't this slightly misleading? pageset_init is called from setup_zone_pageset which is called from the memory hotplug as well. Isn't this more about It is. early zone initialization rather than boot pagesets? Or am I misreading the patch? It's for boot pagesets (call from build_all_zonelists_init()) that we need these particular values (well, just high = 0). For early zone initialization during hotplug the values don't matter as the caller is setup_zone_pageset() who immediately replaces the values by zone_set_pageset_high_and_batch(). Which is what the comment tries to say. +*/ + pcp->high = 0; + pcp->batch = 1; } /* -- 2.28.0
Re: [PATCH 3/9] mm, page_alloc: remove setup_pageset()
On Tue 22-09-20 16:37:06, Vlastimil Babka wrote: > We initialize boot-time pagesets with setup_pageset(), which sets high and > batch values that effectively disable pcplists. > > We can remove this wrapper if we just set these values for all pagesets in > pageset_init(). Non-boot pagesets then subsequently update them to the proper > values. > > No functional change. > > Signed-off-by: Vlastimil Babka > --- > mm/page_alloc.c | 17 ++--- > 1 file changed, 10 insertions(+), 7 deletions(-) > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > index 26069c8d1b19..76c2b4578723 100644 > --- a/mm/page_alloc.c > +++ b/mm/page_alloc.c > @@ -5823,7 +5823,7 @@ static void build_zonelists(pg_data_t *pgdat) > * not check if the processor is online before following the pageset pointer. > * Other parts of the kernel may not check if the zone is available. > */ > -static void setup_pageset(struct per_cpu_pageset *p); > +static void pageset_init(struct per_cpu_pageset *p); > static DEFINE_PER_CPU(struct per_cpu_pageset, boot_pageset); > static DEFINE_PER_CPU(struct per_cpu_nodestat, boot_nodestats); > > @@ -5891,7 +5891,7 @@ build_all_zonelists_init(void) >* (a chicken-egg dilemma). >*/ > for_each_possible_cpu(cpu) > - setup_pageset(_cpu(boot_pageset, cpu)); > + pageset_init(_cpu(boot_pageset, cpu)); > > mminit_verify_zonelist(); > cpuset_init_current_mems_allowed(); > @@ -6210,12 +6210,15 @@ static void pageset_init(struct per_cpu_pageset *p) > pcp = >pcp; > for (migratetype = 0; migratetype < MIGRATE_PCPTYPES; migratetype++) > INIT_LIST_HEAD(>lists[migratetype]); > -} > > -static void setup_pageset(struct per_cpu_pageset *p) > -{ > - pageset_init(p); > - pageset_update(>pcp, 0, 1); > + /* > + * Set batch and high values safe for a boot pageset. A true percpu > + * pageset's initialization will update them subsequently. Here we don't > + * need to be as careful as pageset_update() as nobody can access the > + * pageset yet. Isn't this slightly misleading? pageset_init is called from setup_zone_pageset which is called from the memory hotplug as well. Isn't this more about early zone initialization rather than boot pagesets? Or am I misreading the patch? > + */ > + pcp->high = 0; > + pcp->batch = 1; > } > > /* > -- > 2.28.0 -- Michal Hocko SUSE Labs
Re: [PATCH 3/9] mm, page_alloc: remove setup_pageset()
On 22.09.20 16:37, Vlastimil Babka wrote: > We initialize boot-time pagesets with setup_pageset(), which sets high and > batch values that effectively disable pcplists. > > We can remove this wrapper if we just set these values for all pagesets in > pageset_init(). Non-boot pagesets then subsequently update them to the proper > values. > > No functional change. > > Signed-off-by: Vlastimil Babka > --- > mm/page_alloc.c | 17 ++--- > 1 file changed, 10 insertions(+), 7 deletions(-) > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > index 26069c8d1b19..76c2b4578723 100644 > --- a/mm/page_alloc.c > +++ b/mm/page_alloc.c > @@ -5823,7 +5823,7 @@ static void build_zonelists(pg_data_t *pgdat) > * not check if the processor is online before following the pageset pointer. > * Other parts of the kernel may not check if the zone is available. > */ > -static void setup_pageset(struct per_cpu_pageset *p); > +static void pageset_init(struct per_cpu_pageset *p); > static DEFINE_PER_CPU(struct per_cpu_pageset, boot_pageset); > static DEFINE_PER_CPU(struct per_cpu_nodestat, boot_nodestats); > > @@ -5891,7 +5891,7 @@ build_all_zonelists_init(void) >* (a chicken-egg dilemma). >*/ > for_each_possible_cpu(cpu) > - setup_pageset(_cpu(boot_pageset, cpu)); > + pageset_init(_cpu(boot_pageset, cpu)); > > mminit_verify_zonelist(); > cpuset_init_current_mems_allowed(); > @@ -6210,12 +6210,15 @@ static void pageset_init(struct per_cpu_pageset *p) > pcp = >pcp; > for (migratetype = 0; migratetype < MIGRATE_PCPTYPES; migratetype++) > INIT_LIST_HEAD(>lists[migratetype]); > -} > > -static void setup_pageset(struct per_cpu_pageset *p) > -{ > - pageset_init(p); > - pageset_update(>pcp, 0, 1); > + /* > + * Set batch and high values safe for a boot pageset. A true percpu > + * pageset's initialization will update them subsequently. Here we don't > + * need to be as careful as pageset_update() as nobody can access the > + * pageset yet. > + */ > + pcp->high = 0; > + pcp->batch = 1; > } > > /* > Reviewed-by: David Hildenbrand -- Thanks, David / dhildenb