[PATCH 1/1] Revert "ovl: set I_CREATING on inode being created"
Aaron Ma
aaron.ma at canonical.com
Mon Aug 5 17:20:27 UTC 2019
This reverts commit b7fe2730d202b8e8fab1884af4f954d793082fdd.
BugLink: https://bugs.launchpad.net/bugs/1838982
This upstream commit depends on the following commits:
e950564b97fd ("vfs: don't evict uninitialized inode")
80ea09a002bf ("vfs: factor out inode_insert5()")
With this commit only,
Crash happens during the setup (dpkg) phase of sbuild using a overlayfs
based chroot with the current bionic-proposed kernel (4.15.0-56-generic
[25553.379381] ? ovl_get_origin_fh+0x23/0x150 [overlay]
[25553.379386] ? ovl_inode_test+0x20/0x20 [overlay]
[25553.379390] ? ovl_lock_rename_workdir+0x50/0x50 [overlay]
[25553.379396] ovl_get_inode+0xa2/0x450 [overlay]
[25553.379401] ovl_lookup+0x275/0x760 [overlay]
[25553.379406] lookup_slow+0xab/0x170
[25553.379409] ? lookup_slow+0xab/0x170
[25553.379413] walk_component+0x1c3/0x470
[25553.379416] ? path_init+0x177/0x2f0
[25553.379419] path_lookupat+0x84/0x1f0
[25553.379423] ? __put_cred+0x3d/0x50
[25553.379426] ? revert_creds+0x2f/0x40
[25553.379429] filename_lookup+0xb6/0x190
[25553.379434] ? __check_object_size+0xaf/0x1b0
[25553.379438] user_path_at_empty+0x36/0x40
[25553.379441] ? user_path_at_empty+0x36/0x40
[25553.379444] SyS_chown+0x4d/0xe0
[25553.379449] do_syscall_64+0x73/0x130
[25553.379453] entry_SYSCALL_64_after_hwframe+0x3d/0xa2
commit 80ea09a002bf is not a fix, no need to backport.
Signed-off-by: Aaron Ma <aaron.ma at canonical.com>
---
fs/overlayfs/dir.c | 4 ----
1 file changed, 4 deletions(-)
diff --git a/fs/overlayfs/dir.c b/fs/overlayfs/dir.c
index 4afa892d0da6..fde17ea1282d 100644
--- a/fs/overlayfs/dir.c
+++ b/fs/overlayfs/dir.c
@@ -548,10 +548,6 @@ static int ovl_create_object(struct dentry *dentry, int mode, dev_t rdev,
if (!inode)
goto out_drop_write;
- spin_lock(&inode->i_lock);
- inode->i_state |= I_CREATING;
- spin_unlock(&inode->i_lock);
-
inode_init_owner(inode, dentry->d_parent->d_inode, mode);
attr.mode = inode->i_mode;
--
2.17.1
More information about the kernel-team
mailing list