Fix NULL point deref in v9fs_dir_release
Subsystems (2)
PATCH 1/1 Fix NULL point deref in v9fs_dir_release
There are situations in VFS where we endup calling v9fs_dir_release() before even we instantiate the filp. Hence the check.
Signed-off-by Venkateswararao Jujjuri <[email protected]>
--- fs/9p/vfs_dir.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-)
fs/9p/vfs_dir.c +2 -0
@@ -290,6 +290,8 @@ int v9fs_dir_release(struct inode *inode, struct file *filp) {struct p9_fid *fid; +if (!filp) +return 0 ; fid = filp->private_data;P9_DPRINTK (P9_DEBUG_VFS ,"inode: %p filp: %p fid: %d\n" , inode, filp, fid->fid);
Signature · 1 line
1.6.5.2
Quoted 2 lines · Venkateswararao Jujjuri (JV)
There are situations in VFS where we endup calling
v9fs_dir_release()before even we instantiate the filp. Hence the check.
Err, what callchain would that be? This really should not happen.
Quoted 6 lines · Christoph Hellwig
Quoted 2 lines · Venkateswararao Jujjuri (JV)
There are situations in VFS where we endup calling
v9fs_dir_release()before even we instantiate the filp. Hence the check.Err, what callchain would that be? This really should not happen.
Call Trace: [<ffffffff81209cb1>] v9fs_dir_release+0x29/0x2f [<ffffffff81105985>] fput+0x13a/0x1ec [<ffffffff81207e97>] ? v9fs_open_created+0x0/0xd [<ffffffff811031bd>] __dentry_open+0x1d3/0x29e [<ffffffff811033d3>] lookup_instantiate_filp+0x6b/0x8c [<ffffffff81208fc6>] v9fs_vfs_create_dotl+0x1a7/0x20d [<ffffffff8110e30a>] vfs_create+0x70/0x92 [<ffffffff8110eddc>] do_last+0x2e0/0x605 [<ffffffff8110f553>] do_filp_open+0x1f8/0x5f8 [<ffffffff811019e7>] ? mem_cgroup_charge_common+0x6a/0x7a [<ffffffff8125e933>] ? might_fault+0x21/0x23 [<ffffffff8125ea32>] ? __strncpy_from_user+0x1e/0x49 [<ffffffff81118ec2>] ? alloc_fd+0x7b/0x124 [<ffffffff81102ef2>] do_sys_open+0x63/0x10f [<ffffffff81102fd1>] sys_open+0x20/0x22 [<ffffffff81009c32>] system_call_fastpath+0x16/0x1b