| Message ID | 20241120030443.2679200-1-lizhi.xu@windriver.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [V3] fs/ntfs3: check if the inode is bad before creating symlink | expand |
On Wed, Nov 20, 2024 at 11:04:43AM +0800, Lizhi Xu wrote: > syzbot reported a null-ptr-deref in pick_link. [1] > > First, i_link and i_dir_seq are in the same union, they share the same memory > address, and i_dir_seq will be updated during the execution of walk_component, > which makes the value of i_link equal to i_dir_seq. > > Secondly, the chmod execution failed, which resulted in setting the mode value > of file0's inode to REG when executing ntfs_bad_inode. > > Third, when creating a symbolic link using the file0 whose inode has been marked > as bad, it is not determined whether its inode is bad, which ultimately leads to > null-ptr-deref when performing a mount operation on the symbolic link bus because > the i_link value is equal to i_dir_seq=2. > > Note: ("file0, bus" are defined in reproducer [2]) > > To avoid null-ptr-deref in pick_link, when creating a symbolic link, first check > whether the inode of file is already bad. I would really like to understand how the hell did that bad inode end up passed to d_splice_alias()/d_instantiate()/whatever it had been. That's the root cause - and it looks like ntfs is too free with make_bad_inode() in general, which might cause other problems.
On Wed, 20 Nov 2024 16:10:45 +0000, Al Viro wrote: > On Wed, Nov 20, 2024 at 11:04:43AM +0800, Lizhi Xu wrote: > > syzbot reported a null-ptr-deref in pick_link. [1] > > > > First, i_link and i_dir_seq are in the same union, they share the same memory > > address, and i_dir_seq will be updated during the execution of walk_component, > > which makes the value of i_link equal to i_dir_seq. > > > > Secondly, the chmod execution failed, which resulted in setting the mode value > > of file0's inode to REG when executing ntfs_bad_inode. > > > > Third, when creating a symbolic link using the file0 whose inode has been marked > > as bad, it is not determined whether its inode is bad, which ultimately leads to > > null-ptr-deref when performing a mount operation on the symbolic link bus because > > the i_link value is equal to i_dir_seq=2. > > > > Note: ("file0, bus" are defined in reproducer [2]) > > > > To avoid null-ptr-deref in pick_link, when creating a symbolic link, first check > > whether the inode of file is already bad. > > I would really like to understand how the hell did that bad inode end up passed > to d_splice_alias()/d_instantiate()/whatever it had been. 1. In the move_mount() process, the inode is created by ntfs_alloc_inode() and enters d_splice_alias() by ntfs_lookup(), at this time inode is good, as shown below: move_mount()-> user_path_at()-> filename_lookup()-> path_lookupat()-> lookup_last()-> walk_component()-> __lookup_slow()-> ntfs_lookup()-> d_splice_alias()-> 2. The subsequent chmod fails, causing the inode to be set to bad. 3. During the link operation, d_instantiate() is executed in ntfs_link() to associate the bad inode with the dentry. 4. During the mount operation, walk_component executes pick_link, triggering null-ptr-deref. Reproducer: move_mount(0xffffffffffffff9c, &(0x7f00000003c0)='./file0\x00', 0xffffffffffffff9c, &(0x7f0000000400)='./file0/file0\x00', 0x140) chmod(&(0x7f0000000080)='./file0\x00', 0x0) link(&(0x7f0000000200)='./file0\x00', &(0x7f0000000240)='./bus\x00') mount$overlay(0x0, &(0x7f00000000c0)='./bus\x00', 0x0, 0x0, 0x0) > > That's the root cause - and it looks like ntfs is too free with make_bad_inode() > in general, which might cause other problems.
On Thu, Nov 21, 2024 at 11:13:29AM +0800, Lizhi Xu wrote: > user_path_at()-> > filename_lookup()-> > path_lookupat()-> > lookup_last()-> > walk_component()-> > __lookup_slow()-> > ntfs_lookup()-> > d_splice_alias()-> > > 2. The subsequent chmod fails, causing the inode to be set to bad. What's wrong with "return an error"? > 3. During the link operation, d_instantiate() is executed in ntfs_link() to associate the bad inode with the dentry. Yecchhh... If nothing else, check for is_bad_inode() should be there for as long as make_bad_inode() is done on live inodes.
On Thu, 21 Nov 2024 03:55:29 +0000, Al Viro wrote: > > user_path_at()-> > > filename_lookup()-> > > path_lookupat()-> > > lookup_last()-> > > walk_component()-> > > __lookup_slow()-> > > ntfs_lookup()-> > > d_splice_alias()-> > > > > 2. The subsequent chmod fails, causing the inode to be set to bad. > > What's wrong with "return an error"? make_bad_inode() is executed in attr_set_size() and the error code -ENOENT is returned; attr_set_size() defined in fs/ntfs3/attrib.c > > > 3. During the link operation, d_instantiate() is executed in ntfs_link() to associate the bad inode with the dentry. > > Yecchhh... If nothing else, check for is_bad_inode() should be there > for as long as make_bad_inode() is done on live inodes. I checked is_bad_inode() in ntfs_link(), see line 146 of ntfs_link(), please see my V3 of the patch [2]. [1] fs/ntfs3/namei.c 19 static int ntfs_link(struct dentry *ode, struct inode *dir, struct dentry *de) 18 { 17 int err; 16 struct inode *inode = d_inode(ode); 15 struct ntfs_inode *ni = ntfs_i(inode); 14 13 if (S_ISDIR(inode->i_mode)) 12 return -EPERM; 11 10 if (inode->i_nlink >= NTFS_LINK_MAX) 9 return -EMLINK; 8 7 ni_lock_dir(ntfs_i(dir)); 6 if (inode != dir) 5 ni_lock(ni); 4 3 inc_nlink(inode); 2 ihold(inode); 1 146 err = ntfs_link_inode(inode, de); 1 2 if (!err) { 3 inode_set_ctime_current(inode); 4 inode_set_mtime_to_ts(dir, inode_set_ctime_current(dir)); 5 mark_inode_dirty(inode); 6 mark_inode_dirty(dir); 7 d_instantiate(de, inode); [2] Reported-by: syzbot+73d8fc29ec7cba8286fa@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=73d8fc29ec7cba8286fa Signed-off-by: Lizhi Xu <lizhi.xu@windriver.com> --- V1 --> V2: add the root cause of the i_link not set issue and imporve the check V2 --> V3: when creating a symbolic link, first check whether the inode of file is bad. fs/ntfs3/inode.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/fs/ntfs3/inode.c b/fs/ntfs3/inode.c index be04d2845bb7..fefbdcf75016 100644 --- a/fs/ntfs3/inode.c +++ b/fs/ntfs3/inode.c @@ -1719,6 +1719,9 @@ int ntfs_link_inode(struct inode *inode, struct dentry *dentry) struct ntfs_sb_info *sbi = inode->i_sb->s_fs_info; struct NTFS_DE *de; + if (is_bad_inode(inode)) + return -EIO; + /* Allocate PATH_MAX bytes. */ de = __getname(); if (!de)
On Wed, 20 Nov 2024 16:10:45 +0000, Al Viro wrote: > > syzbot reported a null-ptr-deref in pick_link. [1] > > > > First, i_link and i_dir_seq are in the same union, they share the same memory > > address, and i_dir_seq will be updated during the execution of walk_component, > > which makes the value of i_link equal to i_dir_seq. > > > > Secondly, the chmod execution failed, which resulted in setting the mode value > > of file0's inode to REG when executing ntfs_bad_inode. > > > > Third, when creating a symbolic link using the file0 whose inode has been marked > > as bad, it is not determined whether its inode is bad, which ultimately leads to > > null-ptr-deref when performing a mount operation on the symbolic link bus because > > the i_link value is equal to i_dir_seq=2. > > > > Note: ("file0, bus" are defined in reproducer [2]) > > > > To avoid null-ptr-deref in pick_link, when creating a symbolic link, first check > > whether the inode of file is already bad. > > I would really like to understand how the hell did that bad inode end up passed > to d_splice_alias()/d_instantiate()/whatever it had been. > > That's the root cause - and it looks like ntfs is too free with make_bad_inode() > in general, which might cause other problems. I will release the patch of the v4 version and add root cause: During the execution of the link command, it sets the inode of the symlink file to the already bad inode of file0 by calling d_instantiate, which ultimately leads to null-ptr-deref when performing a mount operation on the symbolic link bus because it use bad inode's i_link and its value is equal to i_dir_seq=2. BR, Lizhi
diff --git a/fs/ntfs3/inode.c b/fs/ntfs3/inode.c index be04d2845bb7..fefbdcf75016 100644 --- a/fs/ntfs3/inode.c +++ b/fs/ntfs3/inode.c @@ -1719,6 +1719,9 @@ int ntfs_link_inode(struct inode *inode, struct dentry *dentry) struct ntfs_sb_info *sbi = inode->i_sb->s_fs_info; struct NTFS_DE *de; + if (is_bad_inode(inode)) + return -EIO; + /* Allocate PATH_MAX bytes. */ de = __getname(); if (!de)
syzbot reported a null-ptr-deref in pick_link. [1] First, i_link and i_dir_seq are in the same union, they share the same memory address, and i_dir_seq will be updated during the execution of walk_component, which makes the value of i_link equal to i_dir_seq. Secondly, the chmod execution failed, which resulted in setting the mode value of file0's inode to REG when executing ntfs_bad_inode. Third, when creating a symbolic link using the file0 whose inode has been marked as bad, it is not determined whether its inode is bad, which ultimately leads to null-ptr-deref when performing a mount operation on the symbolic link bus because the i_link value is equal to i_dir_seq=2. Note: ("file0, bus" are defined in reproducer [2]) To avoid null-ptr-deref in pick_link, when creating a symbolic link, first check whether the inode of file is already bad. [1] KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007] CPU: 0 UID: 0 PID: 5310 Comm: syz-executor255 Not tainted 6.12.0-rc6-syzkaller-00318-ga9cda7c0ffed #0 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2~bpo12+1 04/01/2014 RIP: 0010:pick_link+0x51c/0xd50 fs/namei.c:1864 Code: c1 e8 03 42 80 3c 38 00 74 08 48 89 df e8 fc 00 e9 ff 48 8b 2b 48 85 ed 0f 84 92 00 00 00 e8 7b 36 7f ff 48 89 e8 48 c1 e8 03 <42> 0f b6 04 38 84 c0 0f 85 a2 05 00 00 0f b6 5d 00 bf 2f 00 00 00 RSP: 0018:ffffc9000d147998 EFLAGS: 00010246 RAX: 0000000000000000 RBX: ffff88804558dec8 RCX: ffff88801ec7a440 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000 RBP: 0000000000000002 R08: ffffffff8215a35f R09: 1ffffffff203a13d R10: dffffc0000000000 R11: fffffbfff203a13e R12: 1ffff92001a28f93 R13: ffffc9000d147af8 R14: 1ffff92001a28f5f R15: dffffc0000000000 FS: 0000555577611380(0000) GS:ffff88801fc00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007fcc0a595ed8 CR3: 0000000035760000 CR4: 0000000000352ef0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: <TASK> step_into+0xca9/0x1080 fs/namei.c:1923 lookup_last fs/namei.c:2556 [inline] path_lookupat+0x16f/0x450 fs/namei.c:2580 filename_lookup+0x256/0x610 fs/namei.c:2609 user_path_at+0x3a/0x60 fs/namei.c:3016 do_mount fs/namespace.c:3844 [inline] __do_sys_mount fs/namespace.c:4057 [inline] __se_sys_mount+0x297/0x3c0 fs/namespace.c:4034 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7f4b18ad5b19 Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 f1 17 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007ffc2e486c48 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 RAX: ffffffffffffffda RBX: 0030656c69662f2e RCX: 00007f4b18ad5b19 RDX: 0000000000000000 RSI: 00000000200000c0 RDI: 0000000000000000 RBP: 00007f4b18b685f0 R08: 0000000000000000 R09: 00005555776124c0 R10: 0000000000000000 R11: 0000000000000246 R12: 00007ffc2e486c70 R13: 00007ffc2e486e98 R14: 431bde82d7b634db R15: 00007f4b18b1e03b </TASK> [2] move_mount(0xffffffffffffff9c, &(0x7f00000003c0)='./file0\x00', 0xffffffffffffff9c, &(0x7f0000000400)='./file0/file0\x00', 0x140) chmod(&(0x7f0000000080)='./file0\x00', 0x0) link(&(0x7f0000000200)='./file0\x00', &(0x7f0000000240)='./bus\x00') mount$overlay(0x0, &(0x7f00000000c0)='./bus\x00', 0x0, 0x0, 0x0) Reported-by: syzbot+73d8fc29ec7cba8286fa@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=73d8fc29ec7cba8286fa Signed-off-by: Lizhi Xu <lizhi.xu@windriver.com> --- V1 --> V2: add the root cause of the i_link not set issue and imporve the check V2 --> V3: when creating a symbolic link, first check whether the inode of file is bad. fs/ntfs3/inode.c | 3 +++ 1 file changed, 3 insertions(+)