Message ID | 20230621023238.87079-8-alexei.starovoitov@gmail.com (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
Series | bpf: Introduce bpf_mem_cache_free_rcu(). | expand |
Hi, On 6/21/2023 10:32 AM, Alexei Starovoitov wrote: > From: Alexei Starovoitov <ast@kernel.org> > > To address OOM issue when one cpu is allocating and another cpu is freeing add > a target bpf_mem_cache hint to allocated objects and when local cpu free_llist > overflows free to that bpf_mem_cache. The hint addresses the OOM while > maintaing the same performance for common case when alloc/free are done on the > same cpu. > > Signed-off-by: Alexei Starovoitov <ast@kernel.org> > --- > kernel/bpf/memalloc.c | 46 ++++++++++++++++++++++++++----------------- > 1 file changed, 28 insertions(+), 18 deletions(-) > > diff --git a/kernel/bpf/memalloc.c b/kernel/bpf/memalloc.c > index 4fd79bd51f5a..8b7645bffd1a 100644 > --- a/kernel/bpf/memalloc.c > +++ b/kernel/bpf/memalloc.c > @@ -98,6 +98,7 @@ struct bpf_mem_cache { > int free_cnt; > int low_watermark, high_watermark, batch; > int percpu_size; > + struct bpf_mem_cache *tgt; > > /* list of objects to be freed after RCU tasks trace GP */ > struct llist_head free_by_rcu_ttrace; > @@ -189,18 +190,11 @@ static void alloc_bulk(struct bpf_mem_cache *c, int cnt, int node) > > for (i = 0; i < cnt; i++) { > /* > - * free_by_rcu_ttrace is only manipulated by irq work refill_work(). > - * IRQ works on the same CPU are called sequentially, so it is > - * safe to use __llist_del_first() here. If alloc_bulk() is > - * invoked by the initial prefill, there will be no running > - * refill_work(), so __llist_del_first() is fine as well. > - * > - * In most cases, objects on free_by_rcu_ttrace are from the same CPU. > - * If some objects come from other CPUs, it doesn't incur any > - * harm because NUMA_NO_NODE means the preference for current > - * numa node and it is not a guarantee. > + * For every 'c' llist_del_first(&c->free_by_rcu_ttrace); is > + * done only by one CPU == current CPU. Other CPUs might > + * llist_add() and llist_del_all() in parallel. > */ > - obj = __llist_del_first(&c->free_by_rcu_ttrace); > + obj = llist_del_first(&c->free_by_rcu_ttrace); According to the comments in llist.h, when there are concurrent llist_del_first() and llist_del_all() operations, locking is needed. > if (!obj) > break; > add_obj_to_free_list(c, obj); > @@ -274,7 +268,7 @@ static void enque_to_free(struct bpf_mem_cache *c, void *obj) > /* bpf_mem_cache is a per-cpu object. Freeing happens in irq_work. > * Nothing races to add to free_by_rcu_ttrace list. > */ > - if (__llist_add(llnode, &c->free_by_rcu_ttrace)) > + if (llist_add(llnode, &c->free_by_rcu_ttrace)) > c->free_by_rcu_ttrace_tail = llnode; > } > > @@ -286,7 +280,7 @@ static void do_call_rcu_ttrace(struct bpf_mem_cache *c) > return; > > WARN_ON_ONCE(!llist_empty(&c->waiting_for_gp_ttrace)); > - llnode = __llist_del_all(&c->free_by_rcu_ttrace); > + llnode = llist_del_all(&c->free_by_rcu_ttrace); > if (llnode) > /* There is no concurrent __llist_add(waiting_for_gp_ttrace) access. > * It doesn't race with llist_del_all either. > @@ -299,16 +293,22 @@ static void do_call_rcu_ttrace(struct bpf_mem_cache *c) > * If RCU Tasks Trace grace period implies RCU grace period, free > * these elements directly, else use call_rcu() to wait for normal > * progs to finish and finally do free_one() on each element. > + * > + * call_rcu_tasks_trace() enqueues to a global queue, so it's ok > + * that current cpu bpf_mem_cache != target bpf_mem_cache. > */ > call_rcu_tasks_trace(&c->rcu_ttrace, __free_rcu_tasks_trace); > } > > static void free_bulk(struct bpf_mem_cache *c) > { > + struct bpf_mem_cache *tgt = c->tgt; > struct llist_node *llnode, *t; > unsigned long flags; > int cnt; > > + WARN_ON_ONCE(tgt->unit_size != c->unit_size); > + > do { > if (IS_ENABLED(CONFIG_PREEMPT_RT)) > local_irq_save(flags); > @@ -322,13 +322,13 @@ static void free_bulk(struct bpf_mem_cache *c) > if (IS_ENABLED(CONFIG_PREEMPT_RT)) > local_irq_restore(flags); > if (llnode) > - enque_to_free(c, llnode); > + enque_to_free(tgt, llnode); > } while (cnt > (c->high_watermark + c->low_watermark) / 2); > > /* and drain free_llist_extra */ > llist_for_each_safe(llnode, t, llist_del_all(&c->free_llist_extra)) > - enque_to_free(c, llnode); > - do_call_rcu_ttrace(c); > + enque_to_free(tgt, llnode); > + do_call_rcu_ttrace(tgt); > } > > static void bpf_mem_refill(struct irq_work *work) > @@ -427,6 +427,7 @@ int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu) > c->unit_size = unit_size; > c->objcg = objcg; > c->percpu_size = percpu_size; > + c->tgt = c; > prefill_mem_cache(c, cpu); > } > ma->cache = pc; > @@ -449,6 +450,7 @@ int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu) > c = &cc->cache[i]; > c->unit_size = sizes[i]; > c->objcg = objcg; > + c->tgt = c; > prefill_mem_cache(c, cpu); > } > } > @@ -467,7 +469,7 @@ static void drain_mem_cache(struct bpf_mem_cache *c) > * Except for waiting_for_gp_ttrace list, there are no concurrent operations > * on these lists, so it is safe to use __llist_del_all(). > */ > - free_all(__llist_del_all(&c->free_by_rcu_ttrace), percpu); > + free_all(llist_del_all(&c->free_by_rcu_ttrace), percpu); > free_all(llist_del_all(&c->waiting_for_gp_ttrace), percpu); > free_all(__llist_del_all(&c->free_llist), percpu); > free_all(__llist_del_all(&c->free_llist_extra), percpu); > @@ -599,8 +601,10 @@ static void notrace *unit_alloc(struct bpf_mem_cache *c) > local_irq_save(flags); > if (local_inc_return(&c->active) == 1) { > llnode = __llist_del_first(&c->free_llist); > - if (llnode) > + if (llnode) { > cnt = --c->free_cnt; > + *(struct bpf_mem_cache **)llnode = c; > + } > } > local_dec(&c->active); > local_irq_restore(flags); > @@ -624,6 +628,12 @@ static void notrace unit_free(struct bpf_mem_cache *c, void *ptr) > > BUILD_BUG_ON(LLIST_NODE_SZ > 8); > > + /* > + * Remember bpf_mem_cache that allocated this object. > + * The hint is not accurate. > + */ > + c->tgt = *(struct bpf_mem_cache **)llnode; > + > local_irq_save(flags); > if (local_inc_return(&c->active) == 1) { > __llist_add(llnode, &c->free_llist);
On Fri, Jun 23, 2023 at 8:28 PM Hou Tao <houtao@huaweicloud.com> wrote: > > > */ > > - obj = __llist_del_first(&c->free_by_rcu_ttrace); > > + obj = llist_del_first(&c->free_by_rcu_ttrace); > According to the comments in llist.h, when there are concurrent > llist_del_first() and llist_del_all() operations, locking is needed. Good question. 1. When only one cpu is doing llist_del_first() locking is not needed. This is the case here. Only this cpu is doing llist_del_first() from this 'c'. 2. The comments doesn't mention it, but llist_del_first() is ok on multiple cpus if ABA problem is addressed by other means. PS please trim your replies.
Hi, On 6/24/2023 11:42 AM, Alexei Starovoitov wrote: > On Fri, Jun 23, 2023 at 8:28 PM Hou Tao <houtao@huaweicloud.com> wrote: >>> */ >>> - obj = __llist_del_first(&c->free_by_rcu_ttrace); >>> + obj = llist_del_first(&c->free_by_rcu_ttrace); >> According to the comments in llist.h, when there are concurrent >> llist_del_first() and llist_del_all() operations, locking is needed. > Good question. > 1. When only one cpu is doing llist_del_first() locking is not needed. > This is the case here. Only this cpu is doing llist_del_first() from this 'c'. > 2. The comments doesn't mention it, but llist_del_first() is ok on > multiple cpus if ABA problem is addressed by other means. Haven't checked the implementation details of lockless list. Will do that later. "by other means" do you mean RCU ? because the reuse will be possible only after one RCU GP. > PS > please trim your replies. Sorry for the inconvenience. Will do next time.
On Fri, Jun 23, 2023 at 8:54 PM Hou Tao <houtao@huaweicloud.com> wrote: > > Hi, > > On 6/24/2023 11:42 AM, Alexei Starovoitov wrote: > > On Fri, Jun 23, 2023 at 8:28 PM Hou Tao <houtao@huaweicloud.com> wrote: > >>> */ > >>> - obj = __llist_del_first(&c->free_by_rcu_ttrace); > >>> + obj = llist_del_first(&c->free_by_rcu_ttrace); > >> According to the comments in llist.h, when there are concurrent > >> llist_del_first() and llist_del_all() operations, locking is needed. > > Good question. > > 1. When only one cpu is doing llist_del_first() locking is not needed. > > This is the case here. Only this cpu is doing llist_del_first() from this 'c'. > > 2. The comments doesn't mention it, but llist_del_first() is ok on > > multiple cpus if ABA problem is addressed by other means. > Haven't checked the implementation details of lockless list. Will do > that later. "by other means" do you mean RCU ? because the reuse will be > possible only after one RCU GP. Right. Like RCU, but that's if we go that route in future patches. We're at 1 == only one cpu is doing llist_del_first. > > PS > > please trim your replies. > Sorry for the inconvenience. Will do next time. > >
diff --git a/kernel/bpf/memalloc.c b/kernel/bpf/memalloc.c index 4fd79bd51f5a..8b7645bffd1a 100644 --- a/kernel/bpf/memalloc.c +++ b/kernel/bpf/memalloc.c @@ -98,6 +98,7 @@ struct bpf_mem_cache { int free_cnt; int low_watermark, high_watermark, batch; int percpu_size; + struct bpf_mem_cache *tgt; /* list of objects to be freed after RCU tasks trace GP */ struct llist_head free_by_rcu_ttrace; @@ -189,18 +190,11 @@ static void alloc_bulk(struct bpf_mem_cache *c, int cnt, int node) for (i = 0; i < cnt; i++) { /* - * free_by_rcu_ttrace is only manipulated by irq work refill_work(). - * IRQ works on the same CPU are called sequentially, so it is - * safe to use __llist_del_first() here. If alloc_bulk() is - * invoked by the initial prefill, there will be no running - * refill_work(), so __llist_del_first() is fine as well. - * - * In most cases, objects on free_by_rcu_ttrace are from the same CPU. - * If some objects come from other CPUs, it doesn't incur any - * harm because NUMA_NO_NODE means the preference for current - * numa node and it is not a guarantee. + * For every 'c' llist_del_first(&c->free_by_rcu_ttrace); is + * done only by one CPU == current CPU. Other CPUs might + * llist_add() and llist_del_all() in parallel. */ - obj = __llist_del_first(&c->free_by_rcu_ttrace); + obj = llist_del_first(&c->free_by_rcu_ttrace); if (!obj) break; add_obj_to_free_list(c, obj); @@ -274,7 +268,7 @@ static void enque_to_free(struct bpf_mem_cache *c, void *obj) /* bpf_mem_cache is a per-cpu object. Freeing happens in irq_work. * Nothing races to add to free_by_rcu_ttrace list. */ - if (__llist_add(llnode, &c->free_by_rcu_ttrace)) + if (llist_add(llnode, &c->free_by_rcu_ttrace)) c->free_by_rcu_ttrace_tail = llnode; } @@ -286,7 +280,7 @@ static void do_call_rcu_ttrace(struct bpf_mem_cache *c) return; WARN_ON_ONCE(!llist_empty(&c->waiting_for_gp_ttrace)); - llnode = __llist_del_all(&c->free_by_rcu_ttrace); + llnode = llist_del_all(&c->free_by_rcu_ttrace); if (llnode) /* There is no concurrent __llist_add(waiting_for_gp_ttrace) access. * It doesn't race with llist_del_all either. @@ -299,16 +293,22 @@ static void do_call_rcu_ttrace(struct bpf_mem_cache *c) * If RCU Tasks Trace grace period implies RCU grace period, free * these elements directly, else use call_rcu() to wait for normal * progs to finish and finally do free_one() on each element. + * + * call_rcu_tasks_trace() enqueues to a global queue, so it's ok + * that current cpu bpf_mem_cache != target bpf_mem_cache. */ call_rcu_tasks_trace(&c->rcu_ttrace, __free_rcu_tasks_trace); } static void free_bulk(struct bpf_mem_cache *c) { + struct bpf_mem_cache *tgt = c->tgt; struct llist_node *llnode, *t; unsigned long flags; int cnt; + WARN_ON_ONCE(tgt->unit_size != c->unit_size); + do { if (IS_ENABLED(CONFIG_PREEMPT_RT)) local_irq_save(flags); @@ -322,13 +322,13 @@ static void free_bulk(struct bpf_mem_cache *c) if (IS_ENABLED(CONFIG_PREEMPT_RT)) local_irq_restore(flags); if (llnode) - enque_to_free(c, llnode); + enque_to_free(tgt, llnode); } while (cnt > (c->high_watermark + c->low_watermark) / 2); /* and drain free_llist_extra */ llist_for_each_safe(llnode, t, llist_del_all(&c->free_llist_extra)) - enque_to_free(c, llnode); - do_call_rcu_ttrace(c); + enque_to_free(tgt, llnode); + do_call_rcu_ttrace(tgt); } static void bpf_mem_refill(struct irq_work *work) @@ -427,6 +427,7 @@ int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu) c->unit_size = unit_size; c->objcg = objcg; c->percpu_size = percpu_size; + c->tgt = c; prefill_mem_cache(c, cpu); } ma->cache = pc; @@ -449,6 +450,7 @@ int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu) c = &cc->cache[i]; c->unit_size = sizes[i]; c->objcg = objcg; + c->tgt = c; prefill_mem_cache(c, cpu); } } @@ -467,7 +469,7 @@ static void drain_mem_cache(struct bpf_mem_cache *c) * Except for waiting_for_gp_ttrace list, there are no concurrent operations * on these lists, so it is safe to use __llist_del_all(). */ - free_all(__llist_del_all(&c->free_by_rcu_ttrace), percpu); + free_all(llist_del_all(&c->free_by_rcu_ttrace), percpu); free_all(llist_del_all(&c->waiting_for_gp_ttrace), percpu); free_all(__llist_del_all(&c->free_llist), percpu); free_all(__llist_del_all(&c->free_llist_extra), percpu); @@ -599,8 +601,10 @@ static void notrace *unit_alloc(struct bpf_mem_cache *c) local_irq_save(flags); if (local_inc_return(&c->active) == 1) { llnode = __llist_del_first(&c->free_llist); - if (llnode) + if (llnode) { cnt = --c->free_cnt; + *(struct bpf_mem_cache **)llnode = c; + } } local_dec(&c->active); local_irq_restore(flags); @@ -624,6 +628,12 @@ static void notrace unit_free(struct bpf_mem_cache *c, void *ptr) BUILD_BUG_ON(LLIST_NODE_SZ > 8); + /* + * Remember bpf_mem_cache that allocated this object. + * The hint is not accurate. + */ + c->tgt = *(struct bpf_mem_cache **)llnode; + local_irq_save(flags); if (local_inc_return(&c->active) == 1) { __llist_add(llnode, &c->free_llist);