Merge branch 'bpf-use-gfp_kernel-in-bpf_event_entry_gen'

Hou Tao says:

====================
The simple patch set aims to replace GFP_ATOMIC by GFP_KERNEL in
bpf_event_entry_gen(). These two patches in the patch set were
preparatory patches in "Fix the release of inner map" patchset [1] and
are not needed for v2, so re-post it to bpf-next tree.

Patch #1 reduces the scope of rcu_read_lock when updating fd map and
patch #2 replaces GFP_ATOMIC by GFP_KERNEL. Please see individual
patches for more details.

Change Log:
v3:
 * patch #1: fallback to patch #1 in v1. Update comments in
             bpf_fd_htab_map_update_elem() to explain the reason for
	     rcu_read_lock() (Alexei)

v2: https://lore.kernel.org/bpf/20231211073843.1888058-1-houtao@huaweicloud.com/
 * patch #1: add rcu_read_lock/unlock() for bpf_fd_array_map_update_elem
             as well to make it consistent with
	     bpf_fd_htab_map_update_elem and update commit message
             accordingly (Alexei)
 * patch #1/#2: collects ack tags from Yonghong

v1: https://lore.kernel.org/bpf/20231208103357.2637299-1-houtao@huaweicloud.com/

[1]: https://lore.kernel.org/bpf/20231107140702.1891778-1-houtao@huaweicloud.com/
====================

Link: https://lore.kernel.org/r/20231214043010.3458072-1-houtao@huaweicloud.comSigned-off-by: Alexei Starovoitov <ast@kernel.org>

kernel/bpf/arraymap.c

@@ -1195,7 +1195,7 @@ static struct bpf_event_entry *bpf_event_entry_gen(struct file *perf_file,
 {
 	struct bpf_event_entry *ee;
 
-	ee = kzalloc(sizeof(*ee), GFP_ATOMIC);
+	ee = kzalloc(sizeof(*ee), GFP_KERNEL);
 	if (ee) {
 		ee->event = perf_file->private_data;
 		ee->perf_file = perf_file;

kernel/bpf/hashtab.c

@@ -2523,7 +2523,13 @@ int bpf_fd_htab_map_update_elem(struct bpf_map *map, struct file *map_file,
 	if (IS_ERR(ptr))
 		return PTR_ERR(ptr);
 
+	/* The htab bucket lock is always held during update operations in fd
+	 * htab map, and the following rcu_read_lock() is only used to avoid
+	 * the WARN_ON_ONCE in htab_map_update_elem().
+	 */
+	rcu_read_lock();
 	ret = htab_map_update_elem(map, key, &ptr, map_flags);
+	rcu_read_unlock();
 	if (ret)
 		map->ops->map_fd_put_ptr(map, ptr, false);
 

kernel/bpf/syscall.c

@@ -184,15 +184,11 @@ static int bpf_map_update_value(struct bpf_map *map, struct file *map_file,
 		err = bpf_percpu_cgroup_storage_update(map, key, value,
 						       flags);
 	} else if (IS_FD_ARRAY(map)) {
-		rcu_read_lock();
 		err = bpf_fd_array_map_update_elem(map, map_file, key, value,
 						   flags);
-		rcu_read_unlock();
 	} else if (map->map_type == BPF_MAP_TYPE_HASH_OF_MAPS) {
-		rcu_read_lock();
 		err = bpf_fd_htab_map_update_elem(map, map_file, key, value,
 						  flags);
-		rcu_read_unlock();
 	} else if (map->map_type == BPF_MAP_TYPE_REUSEPORT_SOCKARRAY) {
 		/* rcu_read_lock() is not needed */
 		err = bpf_fd_reuseport_array_update_elem(map, key, value,