io_uring: add abstraction around apoll cache

In preparation for adding limits, and one more user, abstract out the
core bits of the allocation+free cache.

Signed-off-by: Jens Axboe <axboe@kernel.dk>

include/linux/io_uring_types.h

@@ -158,6 +158,10 @@ struct io_ev_fd {
 	struct rcu_head		rcu;
 };
 
+struct io_alloc_cache {
+	struct hlist_head	list;
+};
+
 struct io_ring_ctx {
 	/* const or read-mostly hot data */
 	struct {
@@ -216,7 +220,7 @@ struct io_ring_ctx {
 
 		struct io_hash_table	cancel_table_locked;
 		struct list_head	cq_overflow_list;
-		struct list_head	apoll_cache;
+		struct io_alloc_cache	apoll_cache;
 		struct xarray		personalities;
 		u32			pers_next;
 	} ____cacheline_aligned_in_smp;

io_uring/alloc_cache.h

@@ -0,0 +1,41 @@
+#ifndef IOU_ALLOC_CACHE_H
+#define IOU_ALLOC_CACHE_H
+
+struct io_cache_entry {
+	struct hlist_node	node;
+};
+
+static inline void io_alloc_cache_put(struct io_alloc_cache *cache,
+				      struct io_cache_entry *entry)
+{
+	hlist_add_head(&entry->node, &cache->list);
+}
+
+static inline struct io_cache_entry *io_alloc_cache_get(struct io_alloc_cache *cache)
+{
+	if (!hlist_empty(&cache->list)) {
+		struct hlist_node *node = cache->list.first;
+
+		hlist_del(node);
+		return container_of(node, struct io_cache_entry, node);
+	}
+
+	return NULL;
+}
+
+static inline void io_alloc_cache_init(struct io_alloc_cache *cache)
+{
+	INIT_HLIST_HEAD(&cache->list);
+}
+
+static inline void io_alloc_cache_free(struct io_alloc_cache *cache,
+					void (*free)(struct io_cache_entry *))
+{
+	while (!hlist_empty(&cache->list)) {
+		struct hlist_node *node = cache->list.first;
+
+		hlist_del(node);
+		free(container_of(node, struct io_cache_entry, node));
+	}
+}
+#endif

io_uring/io_uring.c

@@ -92,6 +92,7 @@
 
 #include "timeout.h"
 #include "poll.h"
+#include "alloc_cache.h"
 
 #define IORING_MAX_ENTRIES	32768
 #define IORING_MAX_CQ_ENTRIES	(2 * IORING_MAX_ENTRIES)
@@ -295,7 +296,7 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	INIT_LIST_HEAD(&ctx->sqd_list);
 	INIT_LIST_HEAD(&ctx->cq_overflow_list);
 	INIT_LIST_HEAD(&ctx->io_buffers_cache);
-	INIT_LIST_HEAD(&ctx->apoll_cache);
+	io_alloc_cache_init(&ctx->apoll_cache);
 	init_completion(&ctx->ref_comp);
 	xa_init_flags(&ctx->personalities, XA_FLAGS_ALLOC1);
 	mutex_init(&ctx->uring_lock);
@@ -1180,8 +1181,7 @@ void io_free_batch_list(struct io_ring_ctx *ctx, struct io_wq_work_node *node)
 
 				if (apoll->double_poll)
 					kfree(apoll->double_poll);
-				list_add(&apoll->poll.wait.entry,
-						&ctx->apoll_cache);
+				io_alloc_cache_put(&ctx->apoll_cache, &apoll->cache);
 				req->flags &= ~REQ_F_POLLED;
 			}
 			if (req->flags & IO_REQ_LINK_FLAGS)
@@ -2467,7 +2467,7 @@ static __cold void io_ring_ctx_free(struct io_ring_ctx *ctx)
 	if (ctx->rings)
 		__io_cqring_overflow_flush(ctx, true);
 	io_eventfd_unregister(ctx);
-	io_flush_apoll_cache(ctx);
+	io_alloc_cache_free(&ctx->apoll_cache, io_apoll_cache_free);
 	mutex_unlock(&ctx->uring_lock);
 	io_destroy_buffers(ctx);
 	if (ctx->sq_creds)

io_uring/poll.c

@@ -590,16 +590,15 @@ static struct async_poll *io_req_alloc_apoll(struct io_kiocb *req,
 					     unsigned issue_flags)
 {
 	struct io_ring_ctx *ctx = req->ctx;
+	struct io_cache_entry *entry;
 	struct async_poll *apoll;
 
 	if (req->flags & REQ_F_POLLED) {
 		apoll = req->apoll;
 		kfree(apoll->double_poll);
 	} else if (!(issue_flags & IO_URING_F_UNLOCKED) &&
-		   !list_empty(&ctx->apoll_cache)) {
-		apoll = list_first_entry(&ctx->apoll_cache, struct async_poll,
-						poll.wait.entry);
-		list_del_init(&apoll->poll.wait.entry);
+		   (entry = io_alloc_cache_get(&ctx->apoll_cache)) != NULL) {
+		apoll = container_of(entry, struct async_poll, cache);
 	} else {
 		apoll = kmalloc(sizeof(*apoll), GFP_ATOMIC);
 		if (unlikely(!apoll))
@@ -960,14 +959,7 @@ int io_poll_remove(struct io_kiocb *req, unsigned int issue_flags)
 	return IOU_OK;
 }
 
-void io_flush_apoll_cache(struct io_ring_ctx *ctx)
+void io_apoll_cache_free(struct io_cache_entry *entry)
 {
-	struct async_poll *apoll;
-
-	while (!list_empty(&ctx->apoll_cache)) {
-		apoll = list_first_entry(&ctx->apoll_cache, struct async_poll,
-						poll.wait.entry);
-		list_del(&apoll->poll.wait.entry);
-		kfree(apoll);
-	}
+	kfree(container_of(entry, struct async_poll, cache));
 }

io_uring/poll.h

@@ -1,5 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 
+#include "alloc_cache.h"
+
 enum {
 	IO_APOLL_OK,
 	IO_APOLL_ABORTED,
@@ -14,7 +16,10 @@ struct io_poll {
 };
 
 struct async_poll {
-	struct io_poll		poll;
+	union {
+		struct io_poll		poll;
+		struct io_cache_entry	cache;
+	};
 	struct io_poll		*double_poll;
 };
 
@@ -31,4 +36,4 @@ int io_arm_poll_handler(struct io_kiocb *req, unsigned issue_flags);
 bool io_poll_remove_all(struct io_ring_ctx *ctx, struct task_struct *tsk,
 			bool cancel_all);
 
-void io_flush_apoll_cache(struct io_ring_ctx *ctx);
+void io_apoll_cache_free(struct io_cache_entry *entry);