| Message ID | 20201127041405.3459198-1-guro@fb.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | mm: memcg/slab: fix obj_cgroup_charge() return value handling | expand |
On Thu, Nov 26, 2020 at 8:14 PM Roman Gushchin <guro@fb.com> wrote: > > Commit 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches > for all allocations") introduced a regression into the handling of the > obj_cgroup_charge() return value. If a non-zero value is returned > (indicating of exceeding one of memory.max limits), the allocation > should fail, instead of falling back to non-accounted mode. > > To make the code more readable, move memcg_slab_pre_alloc_hook() > and memcg_slab_post_alloc_hook() calling conditions into bodies > of these hooks. > > Fixes: 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches for all allocations") > Signed-off-by: Roman Gushchin <guro@fb.com> > Cc: stable@vger.kernel.org > --- > mm/slab.h | 40 ++++++++++++++++++++++++---------------- > 1 file changed, 24 insertions(+), 16 deletions(-) > > diff --git a/mm/slab.h b/mm/slab.h > index 59aeb0d9f11b..5dc89d8fb05e 100644 > --- a/mm/slab.h > +++ b/mm/slab.h > @@ -257,22 +257,32 @@ static inline size_t obj_full_size(struct kmem_cache *s) > return s->size + sizeof(struct obj_cgroup *); > } > > -static inline struct obj_cgroup *memcg_slab_pre_alloc_hook(struct kmem_cache *s, > - size_t objects, > - gfp_t flags) > +/* > + * Returns true if the allocation should fail. IMO returning false if the allocation should fail makes this more clear. Otherwise the patch looks good to me. > + */ > +static inline bool memcg_slab_pre_alloc_hook(struct kmem_cache *s, > + struct obj_cgroup **objcgp, > + size_t objects, gfp_t flags) > { > struct obj_cgroup *objcg; > > + if (!memcg_kmem_enabled()) > + return false; > + > + if (!(flags & __GFP_ACCOUNT) && !(s->flags & SLAB_ACCOUNT)) > + return false; > + > objcg = get_obj_cgroup_from_current(); > if (!objcg) > - return NULL; > + return false; > > if (obj_cgroup_charge(objcg, flags, objects * obj_full_size(s))) { > obj_cgroup_put(objcg); > - return NULL; > + return true; > } > > - return objcg; > + *objcgp = objcg; > + return false; > }
On Thu, Nov 26, 2020 at 09:55:24PM -0800, Shakeel Butt wrote: > On Thu, Nov 26, 2020 at 8:14 PM Roman Gushchin <guro@fb.com> wrote: > > > > Commit 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches > > for all allocations") introduced a regression into the handling of the > > obj_cgroup_charge() return value. If a non-zero value is returned > > (indicating of exceeding one of memory.max limits), the allocation > > should fail, instead of falling back to non-accounted mode. > > > > To make the code more readable, move memcg_slab_pre_alloc_hook() > > and memcg_slab_post_alloc_hook() calling conditions into bodies > > of these hooks. > > > > Fixes: 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches for all allocations") > > Signed-off-by: Roman Gushchin <guro@fb.com> > > Cc: stable@vger.kernel.org > > --- > > mm/slab.h | 40 ++++++++++++++++++++++++---------------- > > 1 file changed, 24 insertions(+), 16 deletions(-) > > > > diff --git a/mm/slab.h b/mm/slab.h > > index 59aeb0d9f11b..5dc89d8fb05e 100644 > > --- a/mm/slab.h > > +++ b/mm/slab.h > > @@ -257,22 +257,32 @@ static inline size_t obj_full_size(struct kmem_cache *s) > > return s->size + sizeof(struct obj_cgroup *); > > } > > > > -static inline struct obj_cgroup *memcg_slab_pre_alloc_hook(struct kmem_cache *s, > > - size_t objects, > > - gfp_t flags) > > +/* > > + * Returns true if the allocation should fail. > > IMO returning false if the allocation should fail makes this more > clear. Otherwise the patch looks good to me. Ok, I agree. Here is an updated version. Thank you for looking in! -- From 456ce03f1c91baf5e2441dce0649e09617437fe4 Mon Sep 17 00:00:00 2001 From: Roman Gushchin <guro@fb.com> Date: Thu, 26 Nov 2020 07:39:57 -0800 Subject: [PATCH v2] mm: memcg/slab: fix obj_cgroup_charge() return value handling Commit 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches for all allocations") introduced a regression into the handling of the obj_cgroup_charge() return value. If a non-zero value is returned (indicating of exceeding one of memory.max limits), the allocation should fail, instead of falling back to non-accounted mode. To make the code more readable, move memcg_slab_pre_alloc_hook() and memcg_slab_post_alloc_hook() calling conditions into bodies of these hooks. Fixes: 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches for all allocations") Signed-off-by: Roman Gushchin <guro@fb.com> Cc: stable@vger.kernel.org --- mm/slab.h | 40 ++++++++++++++++++++++++---------------- 1 file changed, 24 insertions(+), 16 deletions(-) diff --git a/mm/slab.h b/mm/slab.h index 59aeb0d9f11b..0698a3c0a9da 100644 --- a/mm/slab.h +++ b/mm/slab.h @@ -257,22 +257,32 @@ static inline size_t obj_full_size(struct kmem_cache *s) return s->size + sizeof(struct obj_cgroup *); } -static inline struct obj_cgroup *memcg_slab_pre_alloc_hook(struct kmem_cache *s, - size_t objects, - gfp_t flags) +/* + * Returns false if the allocation should fail. + */ +static inline bool memcg_slab_pre_alloc_hook(struct kmem_cache *s, + struct obj_cgroup **objcgp, + size_t objects, gfp_t flags) { struct obj_cgroup *objcg; + if (!memcg_kmem_enabled()) + return true; + + if (!(flags & __GFP_ACCOUNT) && !(s->flags & SLAB_ACCOUNT)) + return true; + objcg = get_obj_cgroup_from_current(); if (!objcg) - return NULL; + return true; if (obj_cgroup_charge(objcg, flags, objects * obj_full_size(s))) { obj_cgroup_put(objcg); - return NULL; + return false; } - return objcg; + *objcgp = objcg; + return true; } static inline void mod_objcg_state(struct obj_cgroup *objcg, @@ -298,7 +308,7 @@ static inline void memcg_slab_post_alloc_hook(struct kmem_cache *s, unsigned long off; size_t i; - if (!objcg) + if (!memcg_kmem_enabled() || !objcg) return; flags &= ~__GFP_ACCOUNT; @@ -382,11 +392,11 @@ static inline void memcg_free_page_obj_cgroups(struct page *page) { } -static inline struct obj_cgroup *memcg_slab_pre_alloc_hook(struct kmem_cache *s, - size_t objects, - gfp_t flags) +static inline bool memcg_slab_pre_alloc_hook(struct kmem_cache *s, + struct obj_cgroup **objcgp, + size_t objects, gfp_t flags) { - return NULL; + return true; } static inline void memcg_slab_post_alloc_hook(struct kmem_cache *s, @@ -494,9 +504,8 @@ static inline struct kmem_cache *slab_pre_alloc_hook(struct kmem_cache *s, if (should_failslab(s, flags)) return NULL; - if (memcg_kmem_enabled() && - ((flags & __GFP_ACCOUNT) || (s->flags & SLAB_ACCOUNT))) - *objcgp = memcg_slab_pre_alloc_hook(s, size, flags); + if (!memcg_slab_pre_alloc_hook(s, objcgp, size, flags)) + return NULL; return s; } @@ -515,8 +524,7 @@ static inline void slab_post_alloc_hook(struct kmem_cache *s, s->flags, flags); } - if (memcg_kmem_enabled()) - memcg_slab_post_alloc_hook(s, objcg, flags, size, p); + memcg_slab_post_alloc_hook(s, objcg, flags, size, p); } #ifndef CONFIG_SLOB
On Fri, Nov 27, 2020 at 8:18 AM Roman Gushchin <guro@fb.com> wrote: > > On Thu, Nov 26, 2020 at 09:55:24PM -0800, Shakeel Butt wrote: > > On Thu, Nov 26, 2020 at 8:14 PM Roman Gushchin <guro@fb.com> wrote: > > > > > > Commit 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches > > > for all allocations") introduced a regression into the handling of the > > > obj_cgroup_charge() return value. If a non-zero value is returned > > > (indicating of exceeding one of memory.max limits), the allocation > > > should fail, instead of falling back to non-accounted mode. > > > > > > To make the code more readable, move memcg_slab_pre_alloc_hook() > > > and memcg_slab_post_alloc_hook() calling conditions into bodies > > > of these hooks. > > > > > > Fixes: 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches for all allocations") > > > Signed-off-by: Roman Gushchin <guro@fb.com> > > > Cc: stable@vger.kernel.org > > > --- > > > mm/slab.h | 40 ++++++++++++++++++++++++---------------- > > > 1 file changed, 24 insertions(+), 16 deletions(-) > > > > > > diff --git a/mm/slab.h b/mm/slab.h > > > index 59aeb0d9f11b..5dc89d8fb05e 100644 > > > --- a/mm/slab.h > > > +++ b/mm/slab.h > > > @@ -257,22 +257,32 @@ static inline size_t obj_full_size(struct kmem_cache *s) > > > return s->size + sizeof(struct obj_cgroup *); > > > } > > > > > > -static inline struct obj_cgroup *memcg_slab_pre_alloc_hook(struct kmem_cache *s, > > > - size_t objects, > > > - gfp_t flags) > > > +/* > > > + * Returns true if the allocation should fail. > > > > IMO returning false if the allocation should fail makes this more > > clear. Otherwise the patch looks good to me. > > Ok, I agree. Here is an updated version. > > Thank you for looking in! > > -- > > From 456ce03f1c91baf5e2441dce0649e09617437fe4 Mon Sep 17 00:00:00 2001 > From: Roman Gushchin <guro@fb.com> > Date: Thu, 26 Nov 2020 07:39:57 -0800 > Subject: [PATCH v2] mm: memcg/slab: fix obj_cgroup_charge() return value > handling > > Commit 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches > for all allocations") introduced a regression into the handling of the > obj_cgroup_charge() return value. If a non-zero value is returned > (indicating of exceeding one of memory.max limits), the allocation > should fail, instead of falling back to non-accounted mode. > > To make the code more readable, move memcg_slab_pre_alloc_hook() > and memcg_slab_post_alloc_hook() calling conditions into bodies > of these hooks. > > Fixes: 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches for all allocations") > Signed-off-by: Roman Gushchin <guro@fb.com> > Cc: stable@vger.kernel.org Reviewed-by: Shakeel Butt <shakeelb@google.com>
diff --git a/mm/slab.h b/mm/slab.h index 59aeb0d9f11b..5dc89d8fb05e 100644 --- a/mm/slab.h +++ b/mm/slab.h @@ -257,22 +257,32 @@ static inline size_t obj_full_size(struct kmem_cache *s) return s->size + sizeof(struct obj_cgroup *); } -static inline struct obj_cgroup *memcg_slab_pre_alloc_hook(struct kmem_cache *s, - size_t objects, - gfp_t flags) +/* + * Returns true if the allocation should fail. + */ +static inline bool memcg_slab_pre_alloc_hook(struct kmem_cache *s, + struct obj_cgroup **objcgp, + size_t objects, gfp_t flags) { struct obj_cgroup *objcg; + if (!memcg_kmem_enabled()) + return false; + + if (!(flags & __GFP_ACCOUNT) && !(s->flags & SLAB_ACCOUNT)) + return false; + objcg = get_obj_cgroup_from_current(); if (!objcg) - return NULL; + return false; if (obj_cgroup_charge(objcg, flags, objects * obj_full_size(s))) { obj_cgroup_put(objcg); - return NULL; + return true; } - return objcg; + *objcgp = objcg; + return false; } static inline void mod_objcg_state(struct obj_cgroup *objcg, @@ -298,7 +308,7 @@ static inline void memcg_slab_post_alloc_hook(struct kmem_cache *s, unsigned long off; size_t i; - if (!objcg) + if (!memcg_kmem_enabled() || !objcg) return; flags &= ~__GFP_ACCOUNT; @@ -382,11 +392,11 @@ static inline void memcg_free_page_obj_cgroups(struct page *page) { } -static inline struct obj_cgroup *memcg_slab_pre_alloc_hook(struct kmem_cache *s, - size_t objects, - gfp_t flags) +static inline bool memcg_slab_pre_alloc_hook(struct kmem_cache *s, + struct obj_cgroup **objcgp, + size_t objects, gfp_t flags) { - return NULL; + return false; } static inline void memcg_slab_post_alloc_hook(struct kmem_cache *s, @@ -494,9 +504,8 @@ static inline struct kmem_cache *slab_pre_alloc_hook(struct kmem_cache *s, if (should_failslab(s, flags)) return NULL; - if (memcg_kmem_enabled() && - ((flags & __GFP_ACCOUNT) || (s->flags & SLAB_ACCOUNT))) - *objcgp = memcg_slab_pre_alloc_hook(s, size, flags); + if (memcg_slab_pre_alloc_hook(s, objcgp, size, flags)) + return NULL; return s; } @@ -515,8 +524,7 @@ static inline void slab_post_alloc_hook(struct kmem_cache *s, s->flags, flags); } - if (memcg_kmem_enabled()) - memcg_slab_post_alloc_hook(s, objcg, flags, size, p); + memcg_slab_post_alloc_hook(s, objcg, flags, size, p); } #ifndef CONFIG_SLOB
Commit 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches for all allocations") introduced a regression into the handling of the obj_cgroup_charge() return value. If a non-zero value is returned (indicating of exceeding one of memory.max limits), the allocation should fail, instead of falling back to non-accounted mode. To make the code more readable, move memcg_slab_pre_alloc_hook() and memcg_slab_post_alloc_hook() calling conditions into bodies of these hooks. Fixes: 10befea91b61 ("mm: memcg/slab: use a single set of kmem_caches for all allocations") Signed-off-by: Roman Gushchin <guro@fb.com> Cc: stable@vger.kernel.org --- mm/slab.h | 40 ++++++++++++++++++++++++---------------- 1 file changed, 24 insertions(+), 16 deletions(-)