| Message ID | 20201008114201.18824-4-vbabka@suse.cz (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | disable pcplists during memory offline | expand |
On Thu 08-10-20 13:41:57, 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 <vbabka@suse.cz> > Reviewed-by: David Hildenbrand <david@redhat.com> Acked-by: Michal Hocko <mhocko@suse.com> Btw. where do we initialize pcp->count? I thought that pcp allocator zeroes out the allocated memory but alloc_percpu is GFP_KERNEL like. > --- > 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 463f40b12aca..f827b42a2475 100644 > --- a/mm/page_alloc.c > +++ b/mm/page_alloc.c > @@ -5899,7 +5899,7 @@ 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); > > @@ -5967,7 +5967,7 @@ build_all_zonelists_init(void) > * (a chicken-egg dilemma). > */ > for_each_possible_cpu(cpu) > - setup_pageset(&per_cpu(boot_pageset, cpu)); > + pageset_init(&per_cpu(boot_pageset, cpu)); > > mminit_verify_zonelist(); > cpuset_init_current_mems_allowed(); > @@ -6286,12 +6286,15 @@ static void pageset_init(struct per_cpu_pageset *p) > pcp = &p->pcp; > for (migratetype = 0; migratetype < MIGRATE_PCPTYPES; migratetype++) > INIT_LIST_HEAD(&pcp->lists[migratetype]); > -} > > -static void setup_pageset(struct per_cpu_pageset *p) > -{ > - pageset_init(p); > - pageset_update(&p->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; > } > > /* > -- > 2.28.0
On 10/8/20 2:23 PM, Michal Hocko wrote: > On Thu 08-10-20 13:41:57, 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 <vbabka@suse.cz> >> Reviewed-by: David Hildenbrand <david@redhat.com> > > Acked-by: Michal Hocko <mhocko@suse.com> Thanks! > Btw. where do we initialize pcp->count? I thought that pcp allocator > zeroes out the allocated memory but alloc_percpu is GFP_KERNEL like. pageset_init() does: memset(p, 0, sizeof(*p))
On Thu 08-10-20 14:56:13, Vlastimil Babka wrote: > On 10/8/20 2:23 PM, Michal Hocko wrote: > > On Thu 08-10-20 13:41:57, 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 <vbabka@suse.cz> > > > Reviewed-by: David Hildenbrand <david@redhat.com> > > > > Acked-by: Michal Hocko <mhocko@suse.com> > > Thanks! > > > Btw. where do we initialize pcp->count? I thought that pcp allocator > > zeroes out the allocated memory but alloc_percpu is GFP_KERNEL like. > > pageset_init() does: > memset(p, 0, sizeof(*p)) Ohh, I have missed pcp is embeded into per_cpu_pageset. Using it as s pointer here confused me. Sorry about the noise.
On Thu, Oct 08, 2020 at 01:41:57PM +0200, 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 <vbabka@suse.cz> > Reviewed-by: David Hildenbrand <david@redhat.com> I think I already reviewed this one, but: Reviewed-by: Oscar Salvador <osalvador@suse.de>
diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 463f40b12aca..f827b42a2475 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -5899,7 +5899,7 @@ 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); @@ -5967,7 +5967,7 @@ build_all_zonelists_init(void) * (a chicken-egg dilemma). */ for_each_possible_cpu(cpu) - setup_pageset(&per_cpu(boot_pageset, cpu)); + pageset_init(&per_cpu(boot_pageset, cpu)); mminit_verify_zonelist(); cpuset_init_current_mems_allowed(); @@ -6286,12 +6286,15 @@ static void pageset_init(struct per_cpu_pageset *p) pcp = &p->pcp; for (migratetype = 0; migratetype < MIGRATE_PCPTYPES; migratetype++) INIT_LIST_HEAD(&pcp->lists[migratetype]); -} -static void setup_pageset(struct per_cpu_pageset *p) -{ - pageset_init(p); - pageset_update(&p->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; } /*