在 2022/10/17 09:55, Dawei Li 写道:
s_inodes is superblock-specific resource, which should be
protected by sb's specific lock s_inode_list_lock.
v2: update the locking mechanisim to protect mutual-exclusive access
both for s_inode_list_lock & erofs_fscache_domain_init_cookie(), as the
reviewing comments from Jia Zhu.
v1:
https://lore.kernel.org/all/tycp286mb23237a9993e0ffcfe5c2bdbeca...@tycp286mb2323.jpnp286.prod.outlook.com/
base-commit: 8436c4a57bd147b0bd2943ab499bb8368981b9e1
Signed-off-by: Dawei Li <[email protected]>
---
fs/erofs/fscache.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/fs/erofs/fscache.c b/fs/erofs/fscache.c
index 998cd26a1b3b..fe05bc51f9f2 100644
--- a/fs/erofs/fscache.c
+++ b/fs/erofs/fscache.c
@@ -590,14 +590,17 @@ struct erofs_fscache *erofs_domain_register_cookie(struct
super_block *sb,
struct super_block *psb = erofs_pseudo_mnt->mnt_sb;
mutex_lock(&erofs_domain_cookies_lock);
+ spin_lock(&psb->s_inode_list_lock);
list_for_each_entry(inode, &psb->s_inodes, i_sb_list) {
ctx = inode->i_private;
if (!ctx || ctx->domain != domain || strcmp(ctx->name, name))
continue;
igrab(inode);
+ spin_unlock(&psb->s_inode_list_lock);
mutex_unlock(&erofs_domain_cookies_lock);
return ctx;
}
+ spin_unlock(&psb->s_inode_list_lock);
ctx = erofs_fscache_domain_init_cookie(sb, name, need_inode);
mutex_unlock(&erofs_domain_cookies_lock);
return ctx;
LGTM
Reviewed-by: Jia Zhu <[email protected]>
