On Wed, Aug 03, 2016 at 10:05:00PM +0300, Dan Carpenter wrote:
> We changed this around in f13058a93484 ('nfsd: reorganize nfsd_create')
> so "dchild" can't be an error pointer any more. Also we don't need to
> test if dchild is NULL because dput has a check built-in.
Thanks!
> Can it even be NULL?
It was returned from a caller's lookup_one_len, which can only return
pointers or errors, I believe. (And looks like callers also do
fh_verify and would already oops if it was null.)
Anyway, added a brief comment to that effect, fixed up the referenced
commitid (I rewrote some history there, apologies) and committed, will
push out after a few tests....
--b.
>
> Signed-off-by: Dan Carpenter <[email protected]>
>
> diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
> index c844fd6..cd628ce 100644
> --- a/fs/nfsd/vfs.c
> +++ b/fs/nfsd/vfs.c
> @@ -1141,7 +1141,7 @@ nfsd_create_locked(struct svc_rqst *rqstp, struct
> svc_fh *fhp,
> char *fname, int flen, struct iattr *iap,
> int type, dev_t rdev, struct svc_fh *resfhp)
> {
> - struct dentry *dentry, *dchild = NULL;
> + struct dentry *dentry, *dchild;
> struct inode *dirp;
> __be32 err;
> __be32 err2;
> @@ -1208,8 +1208,7 @@ nfsd_create_locked(struct svc_rqst *rqstp, struct
> svc_fh *fhp,
> if (!err)
> err = fh_update(resfhp);
> out:
> - if (dchild && !IS_ERR(dchild))
> - dput(dchild);
> + dput(dchild);
> return err;
>
> out_nfserr:
