[PATCH] ovl: fix regression caused for lookup helpers API changes

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



The lookup helpers API was changed by merge of vfs-6.16-rc1.async.dir to
pass a non-const qstr pointer argument to lookup_one*() helpers.

All of the callers of this API were changed to pass a pointer to temp
copy of qstr, except overlays that was passing a const pointer to
dentry->d_name that was changed to pass a non-const copy instead
when doing a lookup in lower layer which is not the fs of said dentry.

This wrong use of the API caused a regression in fstest overlay/012.

Fix the regression by making a non-const copy of dentry->d_name prior
to calling the lookup API, but the API should be fixed to not allow this
class of bugs.

Cc: NeilBrown <neilb@xxxxxxx>
Fixes: 5741909697a3 ("VFS: improve interface for lookup_one functions")
Fixes: 390e34bc1490 ("VFS: change lookup_one_common and lookup_noperm_common to take a qstr")
Signed-off-by: Amir Goldstein <amir73il@xxxxxxxxx>
---
 fs/overlayfs/namei.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

Christian,

Please fast track this fix to Linus before 6.16-rc1 and
consider how you would like to fix the lookup_one*() API.

This change is independent of the ovl changes staged on overlayfs-next,
which I am assuming Miklos is going to send to Linus.

Thanks,
Amir.

diff --git a/fs/overlayfs/namei.c b/fs/overlayfs/namei.c
index bf722daf19a9..00979555223d 100644
--- a/fs/overlayfs/namei.c
+++ b/fs/overlayfs/namei.c
@@ -1371,7 +1371,7 @@ struct dentry *ovl_lookup(struct inode *dir, struct dentry *dentry,
 bool ovl_lower_positive(struct dentry *dentry)
 {
 	struct ovl_entry *poe = OVL_E(dentry->d_parent);
-	struct qstr *name = &dentry->d_name;
+	const struct qstr *name = &dentry->d_name;
 	const struct cred *old_cred;
 	unsigned int i;
 	bool positive = false;
@@ -1394,9 +1394,16 @@ bool ovl_lower_positive(struct dentry *dentry)
 		struct dentry *this;
 		struct ovl_path *parentpath = &ovl_lowerstack(poe)[i];
 
+		/*
+		 * We need to make a non-const copy of dentry->d_name,
+		 * becuase lookup_one_positive_unlocked() will hash name
+		 * with parentpath base, which is on another (lwoer fs).
+		 * TODO: the lookup_* API should be changed not to allow this.
+		 */
 		this = lookup_one_positive_unlocked(
 				mnt_idmap(parentpath->layer->mnt),
-				name, parentpath->dentry);
+				&QSTR_LEN(name->name, name->len),
+				parentpath->dentry);
 		if (IS_ERR(this)) {
 			switch (PTR_ERR(this)) {
 			case -ENOENT:
-- 
2.34.1





[Index of Archives]     [Linux Ext4 Filesystem]     [Union Filesystem]     [Filesystem Testing]     [Ceph Users]     [Ecryptfs]     [NTFS 3]     [AutoFS]     [Kernel Newbies]     [Share Photos]     [Security]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux Cachefs]     [Reiser Filesystem]     [Linux RAID]     [NTFS 3]     [Samba]     [Device Mapper]     [CEPH Development]

  Powered by Linux