f2fs: avoid double lock for cp_rwsem

The __f2fs_add_link is covered by cp_rwsem all the time.
This calls init_inode_metadata, which conducts some acl operations including
memory allocation with GFP_KERNEL previously.
But, under memory pressure, f2fs_write_data_page can be called, which also
grabs cp_rwsem too.

In this case, this incurs a deadlock pointed by Chao.
Thread #1        Thread #2
 down_read
                 down_write
  down_read
 -> here down_read should wait forever.

Reviewed-by: Chao Yu <chao2.yu@samsung.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

diff --git a/fs/f2fs/acl.c b/fs/f2fs/acl.c
index 1ccb26bc2a0b5cc709806715e8eec7fcdcb88854..7f12d28ad94cce4f1cdb746dde1e702239e381ee 100644 (file)
--- a/fs/f2fs/acl.c
+++ b/fs/f2fs/acl.c
@@ -62,7 +62,7 @@ static struct posix_acl *f2fs_acl_from_disk(const char *value, size_t size)
        if (count == 0)
                return NULL;
 
-       acl = posix_acl_alloc(count, GFP_KERNEL);
+       acl = posix_acl_alloc(count, GFP_NOFS);
        if (!acl)
                return ERR_PTR(-ENOMEM);
 
@@ -116,7 +116,7 @@ static void *f2fs_acl_to_disk(const struct posix_acl *acl, size_t *size)
        int i;
 
        f2fs_acl = kmalloc(sizeof(struct f2fs_acl_header) + acl->a_count *
-                       sizeof(struct f2fs_acl_entry), GFP_KERNEL);
+                       sizeof(struct f2fs_acl_entry), GFP_NOFS);
        if (!f2fs_acl)
                return ERR_PTR(-ENOMEM);