| Message ID | 20210709000509.2618345-3-surenb@google.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | mm, memcg: Optimizations to minimize overhead when memcgs are disabled | expand |
On Thu, Jul 08, 2021 at 05:05:08PM -0700, Suren Baghdasaryan wrote: > Inline mem_cgroup_{charge/uncharge} and mem_cgroup_uncharge_list functions > functions to perform mem_cgroup_disabled static key check inline before > calling the main body of the function. This minimizes the memcg overhead > in the pagefault and exit_mmap paths when memcgs are disabled using > cgroup_disable=memory command-line option. > This change results in ~0.4% overhead reduction when running PFT test > comparing {CONFIG_MEMCG=n} against {CONFIG_MEMCG=y, cgroup_disable=memory} > configurationon on an 8-core ARM64 Android device. > > Signed-off-by: Suren Baghdasaryan <surenb@google.com> Sounds reasonable to me as well. One comment: > @@ -693,13 +693,59 @@ static inline bool mem_cgroup_below_min(struct mem_cgroup *memcg) > page_counter_read(&memcg->memory); > } > > -int mem_cgroup_charge(struct page *page, struct mm_struct *mm, gfp_t gfp_mask); > +struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm); > + > +int __mem_cgroup_charge(struct page *page, struct mem_cgroup *memcg, > + gfp_t gfp); > +/** > + * mem_cgroup_charge - charge a newly allocated page to a cgroup > + * @page: page to charge > + * @mm: mm context of the victim > + * @gfp_mask: reclaim mode > + * > + * Try to charge @page to the memcg that @mm belongs to, reclaiming > + * pages according to @gfp_mask if necessary. if @mm is NULL, try to > + * charge to the active memcg. > + * > + * Do not use this for pages allocated for swapin. > + * > + * Returns 0 on success. Otherwise, an error code is returned. > + */ > +static inline int mem_cgroup_charge(struct page *page, struct mm_struct *mm, > + gfp_t gfp_mask) > +{ > + struct mem_cgroup *memcg; > + int ret; > + > + if (mem_cgroup_disabled()) > + return 0; > + > + memcg = get_mem_cgroup_from_mm(mm); > + ret = __mem_cgroup_charge(page, memcg, gfp_mask); > + css_put(&memcg->css); > + > + return ret; Why not do int __mem_cgroup_charge(struct page *page, struct mm_struct *mm, gfp_t gfp_mask); static inline int mem_cgroup_charge(struct page *page, struct mm_struct *mm, gfp_t gfp_mask) { if (mem_cgroup_disabled()) return 0; return __mem_cgroup_charge(page, memcg, gfp_mask); } like in the other cases as well? That would avoid inlining two separate function calls into all the callsites... There is an (internal) __mem_cgroup_charge() already, but you can rename it charge_memcg().
On Fri, Jul 9, 2021 at 7:48 AM Johannes Weiner <hannes@cmpxchg.org> wrote: > > On Thu, Jul 08, 2021 at 05:05:08PM -0700, Suren Baghdasaryan wrote: > > Inline mem_cgroup_{charge/uncharge} and mem_cgroup_uncharge_list functions > > functions to perform mem_cgroup_disabled static key check inline before > > calling the main body of the function. This minimizes the memcg overhead > > in the pagefault and exit_mmap paths when memcgs are disabled using > > cgroup_disable=memory command-line option. > > This change results in ~0.4% overhead reduction when running PFT test > > comparing {CONFIG_MEMCG=n} against {CONFIG_MEMCG=y, cgroup_disable=memory} > > configurationon on an 8-core ARM64 Android device. > > > > Signed-off-by: Suren Baghdasaryan <surenb@google.com> > > Sounds reasonable to me as well. One comment: > > > @@ -693,13 +693,59 @@ static inline bool mem_cgroup_below_min(struct mem_cgroup *memcg) > > page_counter_read(&memcg->memory); > > } > > > > -int mem_cgroup_charge(struct page *page, struct mm_struct *mm, gfp_t gfp_mask); > > +struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm); > > + > > +int __mem_cgroup_charge(struct page *page, struct mem_cgroup *memcg, > > + gfp_t gfp); > > +/** > > + * mem_cgroup_charge - charge a newly allocated page to a cgroup > > + * @page: page to charge > > + * @mm: mm context of the victim > > + * @gfp_mask: reclaim mode > > + * > > + * Try to charge @page to the memcg that @mm belongs to, reclaiming > > + * pages according to @gfp_mask if necessary. if @mm is NULL, try to > > + * charge to the active memcg. > > + * > > + * Do not use this for pages allocated for swapin. > > + * > > + * Returns 0 on success. Otherwise, an error code is returned. > > + */ > > +static inline int mem_cgroup_charge(struct page *page, struct mm_struct *mm, > > + gfp_t gfp_mask) > > +{ > > + struct mem_cgroup *memcg; > > + int ret; > > + > > + if (mem_cgroup_disabled()) > > + return 0; > > + > > + memcg = get_mem_cgroup_from_mm(mm); > > + ret = __mem_cgroup_charge(page, memcg, gfp_mask); > > + css_put(&memcg->css); > > + > > + return ret; > > Why not do > > int __mem_cgroup_charge(struct page *page, struct mm_struct *mm, > gfp_t gfp_mask); > > static inline int mem_cgroup_charge(struct page *page, struct mm_struct *mm, > gfp_t gfp_mask) > { > if (mem_cgroup_disabled()) > return 0; > > return __mem_cgroup_charge(page, memcg, gfp_mask); > } > > like in the other cases as well? > > That would avoid inlining two separate function calls into all the > callsites... > > There is an (internal) __mem_cgroup_charge() already, but you can > rename it charge_memcg(). Sounds good. I'll post an updated version with your suggestion. Thanks for the review, Johannes! >
On Fri, Jul 9, 2021 at 8:18 AM Suren Baghdasaryan <surenb@google.com> wrote: > > On Fri, Jul 9, 2021 at 7:48 AM Johannes Weiner <hannes@cmpxchg.org> wrote: > > > > On Thu, Jul 08, 2021 at 05:05:08PM -0700, Suren Baghdasaryan wrote: > > > Inline mem_cgroup_{charge/uncharge} and mem_cgroup_uncharge_list functions > > > functions to perform mem_cgroup_disabled static key check inline before > > > calling the main body of the function. This minimizes the memcg overhead > > > in the pagefault and exit_mmap paths when memcgs are disabled using > > > cgroup_disable=memory command-line option. > > > This change results in ~0.4% overhead reduction when running PFT test > > > comparing {CONFIG_MEMCG=n} against {CONFIG_MEMCG=y, cgroup_disable=memory} > > > configurationon on an 8-core ARM64 Android device. > > > > > > Signed-off-by: Suren Baghdasaryan <surenb@google.com> > > > > Sounds reasonable to me as well. One comment: > > > > > @@ -693,13 +693,59 @@ static inline bool mem_cgroup_below_min(struct mem_cgroup *memcg) > > > page_counter_read(&memcg->memory); > > > } > > > > > > -int mem_cgroup_charge(struct page *page, struct mm_struct *mm, gfp_t gfp_mask); > > > +struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm); > > > + > > > +int __mem_cgroup_charge(struct page *page, struct mem_cgroup *memcg, > > > + gfp_t gfp); > > > +/** > > > + * mem_cgroup_charge - charge a newly allocated page to a cgroup > > > + * @page: page to charge > > > + * @mm: mm context of the victim > > > + * @gfp_mask: reclaim mode > > > + * > > > + * Try to charge @page to the memcg that @mm belongs to, reclaiming > > > + * pages according to @gfp_mask if necessary. if @mm is NULL, try to > > > + * charge to the active memcg. > > > + * > > > + * Do not use this for pages allocated for swapin. > > > + * > > > + * Returns 0 on success. Otherwise, an error code is returned. > > > + */ > > > +static inline int mem_cgroup_charge(struct page *page, struct mm_struct *mm, > > > + gfp_t gfp_mask) > > > +{ > > > + struct mem_cgroup *memcg; > > > + int ret; > > > + > > > + if (mem_cgroup_disabled()) > > > + return 0; > > > + > > > + memcg = get_mem_cgroup_from_mm(mm); > > > + ret = __mem_cgroup_charge(page, memcg, gfp_mask); > > > + css_put(&memcg->css); > > > + > > > + return ret; > > > > Why not do > > > > int __mem_cgroup_charge(struct page *page, struct mm_struct *mm, > > gfp_t gfp_mask); > > > > static inline int mem_cgroup_charge(struct page *page, struct mm_struct *mm, > > gfp_t gfp_mask) > > { > > if (mem_cgroup_disabled()) > > return 0; > > > > return __mem_cgroup_charge(page, memcg, gfp_mask); > > } > > > > like in the other cases as well? > > > > That would avoid inlining two separate function calls into all the > > callsites... > > > > There is an (internal) __mem_cgroup_charge() already, but you can > > rename it charge_memcg(). > > Sounds good. I'll post an updated version with your suggestion. > Thanks for the review, Johannes! Posted v2 just for this patch at https://lore.kernel.org/patchwork/patch/1455550 . Please let me know if you want me to resent the whole patchset instead of just this patch. > > >
diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index bfe5c486f4ad..480815feb116 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -693,13 +693,59 @@ static inline bool mem_cgroup_below_min(struct mem_cgroup *memcg) page_counter_read(&memcg->memory); } -int mem_cgroup_charge(struct page *page, struct mm_struct *mm, gfp_t gfp_mask); +struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm); + +int __mem_cgroup_charge(struct page *page, struct mem_cgroup *memcg, + gfp_t gfp); +/** + * mem_cgroup_charge - charge a newly allocated page to a cgroup + * @page: page to charge + * @mm: mm context of the victim + * @gfp_mask: reclaim mode + * + * Try to charge @page to the memcg that @mm belongs to, reclaiming + * pages according to @gfp_mask if necessary. if @mm is NULL, try to + * charge to the active memcg. + * + * Do not use this for pages allocated for swapin. + * + * Returns 0 on success. Otherwise, an error code is returned. + */ +static inline int mem_cgroup_charge(struct page *page, struct mm_struct *mm, + gfp_t gfp_mask) +{ + struct mem_cgroup *memcg; + int ret; + + if (mem_cgroup_disabled()) + return 0; + + memcg = get_mem_cgroup_from_mm(mm); + ret = __mem_cgroup_charge(page, memcg, gfp_mask); + css_put(&memcg->css); + + return ret; +} + int mem_cgroup_swapin_charge_page(struct page *page, struct mm_struct *mm, gfp_t gfp, swp_entry_t entry); void mem_cgroup_swapin_uncharge_swap(swp_entry_t entry); -void mem_cgroup_uncharge(struct page *page); -void mem_cgroup_uncharge_list(struct list_head *page_list); +void __mem_cgroup_uncharge(struct page *page); +static inline void mem_cgroup_uncharge(struct page *page) +{ + if (mem_cgroup_disabled()) + return; + __mem_cgroup_uncharge(page); +} + +void __mem_cgroup_uncharge_list(struct list_head *page_list); +static inline void mem_cgroup_uncharge_list(struct list_head *page_list) +{ + if (mem_cgroup_disabled()) + return; + __mem_cgroup_uncharge_list(page_list); +} void mem_cgroup_migrate(struct page *oldpage, struct page *newpage); @@ -756,8 +802,6 @@ static inline struct lruvec *mem_cgroup_page_lruvec(struct page *page) struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p); -struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm); - struct lruvec *lock_page_lruvec(struct page *page); struct lruvec *lock_page_lruvec_irq(struct page *page); struct lruvec *lock_page_lruvec_irqsave(struct page *page, diff --git a/mm/memcontrol.c b/mm/memcontrol.c index a228cd51c4bd..da677b55b2fe 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -6701,8 +6701,8 @@ void mem_cgroup_calculate_protection(struct mem_cgroup *root, atomic_long_read(&parent->memory.children_low_usage))); } -static int __mem_cgroup_charge(struct page *page, struct mem_cgroup *memcg, - gfp_t gfp) +int __mem_cgroup_charge(struct page *page, struct mem_cgroup *memcg, + gfp_t gfp) { unsigned int nr_pages = thp_nr_pages(page); int ret; @@ -6722,35 +6722,6 @@ static int __mem_cgroup_charge(struct page *page, struct mem_cgroup *memcg, return ret; } -/** - * mem_cgroup_charge - charge a newly allocated page to a cgroup - * @page: page to charge - * @mm: mm context of the victim - * @gfp_mask: reclaim mode - * - * Try to charge @page to the memcg that @mm belongs to, reclaiming - * pages according to @gfp_mask if necessary. if @mm is NULL, try to - * charge to the active memcg. - * - * Do not use this for pages allocated for swapin. - * - * Returns 0 on success. Otherwise, an error code is returned. - */ -int mem_cgroup_charge(struct page *page, struct mm_struct *mm, gfp_t gfp_mask) -{ - struct mem_cgroup *memcg; - int ret; - - if (mem_cgroup_disabled()) - return 0; - - memcg = get_mem_cgroup_from_mm(mm); - ret = __mem_cgroup_charge(page, memcg, gfp_mask); - css_put(&memcg->css); - - return ret; -} - /** * mem_cgroup_swapin_charge_page - charge a newly allocated page for swapin * @page: page to charge @@ -6921,13 +6892,10 @@ static void uncharge_page(struct page *page, struct uncharge_gather *ug) * * Uncharge a page previously charged with mem_cgroup_charge(). */ -void mem_cgroup_uncharge(struct page *page) +void __mem_cgroup_uncharge(struct page *page) { struct uncharge_gather ug; - if (mem_cgroup_disabled()) - return; - /* Don't touch page->lru of any random page, pre-check: */ if (!page_memcg(page)) return; @@ -6944,14 +6912,11 @@ void mem_cgroup_uncharge(struct page *page) * Uncharge a list of pages previously charged with * mem_cgroup_charge(). */ -void mem_cgroup_uncharge_list(struct list_head *page_list) +void __mem_cgroup_uncharge_list(struct list_head *page_list) { struct uncharge_gather ug; struct page *page; - if (mem_cgroup_disabled()) - return; - uncharge_gather_clear(&ug); list_for_each_entry(page, page_list, lru) uncharge_page(page, &ug);
Inline mem_cgroup_{charge/uncharge} and mem_cgroup_uncharge_list functions functions to perform mem_cgroup_disabled static key check inline before calling the main body of the function. This minimizes the memcg overhead in the pagefault and exit_mmap paths when memcgs are disabled using cgroup_disable=memory command-line option. This change results in ~0.4% overhead reduction when running PFT test comparing {CONFIG_MEMCG=n} against {CONFIG_MEMCG=y, cgroup_disable=memory} configurationon on an 8-core ARM64 Android device. Signed-off-by: Suren Baghdasaryan <surenb@google.com> --- include/linux/memcontrol.h | 54 ++++++++++++++++++++++++++++++++++---- mm/memcontrol.c | 43 +++--------------------------- 2 files changed, 53 insertions(+), 44 deletions(-)