[PATCH v1] fs: Fix error checking for d_hash_and_lookup()

Wang Ming posted 1 patch 2 years, 6 months ago
fs/dcache.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
[PATCH v1] fs: Fix error checking for d_hash_and_lookup()
Posted by Wang Ming 2 years, 6 months ago
In case of failure, debugfs_create_dir() returns NULL or an error
pointer. Most incorrect error checks were fixed, but the one in
d_add_ci() was forgotten.

Fixes: d9171b934526 ("parallel lookups machinery, part 4 (and last)")
Signed-off-by: Wang Ming <machel@vivo.com>
---
 fs/dcache.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/dcache.c b/fs/dcache.c
index 52e6d5fdab6b..2f03e275d2e0 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -2220,7 +2220,7 @@ struct dentry *d_add_ci(struct dentry *dentry, struct inode *inode,
 	 * if not go ahead and create it now.
 	 */
 	found = d_hash_and_lookup(dentry->d_parent, name);
-	if (found) {
+	if (!IS_ERR_OR_NULL(found)) {
 		iput(inode);
 		return found;
 	}
-- 
2.25.1
Re: [PATCH v1] fs: Fix error checking for d_hash_and_lookup()
Posted by Christian Brauner 2 years, 6 months ago
On Thu, Jul 13, 2023 at 07:05:00PM +0800, Wang Ming wrote:
> In case of failure, debugfs_create_dir() returns NULL or an error

What on earth does debugfs_create_dir() have to do with this?

> pointer. Most incorrect error checks were fixed, but the one in
> d_add_ci() was forgotten.
> 
> Fixes: d9171b934526 ("parallel lookups machinery, part 4 (and last)")
> Signed-off-by: Wang Ming <machel@vivo.com>
> ---
>  fs/dcache.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/fs/dcache.c b/fs/dcache.c
> index 52e6d5fdab6b..2f03e275d2e0 100644
> --- a/fs/dcache.c
> +++ b/fs/dcache.c
> @@ -2220,7 +2220,7 @@ struct dentry *d_add_ci(struct dentry *dentry, struct inode *inode,
>  	 * if not go ahead and create it now.
>  	 */
>  	found = d_hash_and_lookup(dentry->d_parent, name);
> -	if (found) {
> +	if (!IS_ERR_OR_NULL(found)) {

I don't understand.

If d_hash_and_lookup() fails due to custom hash function failure then
the old code bubbles that upwards. You're changing that so that it now
adds a new dentry under the generic hash. That can't possibly be correct.