Re: [syzbot] [overlayfs?] general protection fault in d_path

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

 



On Wed, Jul 5, 2023 at 4:06 PM Christian Brauner <brauner@xxxxxxxxxx> wrote:
>
> On Wed, Jul 05, 2023 at 05:00:45AM -0700, syzbot wrote:
> > Hello,
> >
> > syzbot found the following issue on:
> >
> > HEAD commit:    d528014517f2 Revert ".gitignore: ignore *.cover and *.mbx"
> > git tree:       upstream
> > console output: https://syzkaller.appspot.com/x/log.txt?x=14fad002a80000
> > kernel config:  https://syzkaller.appspot.com/x/.config?x=1085b4238c9eb6ba
> > dashboard link: https://syzkaller.appspot.com/bug?extid=a67fc5321ffb4b311c98
> > compiler:       Debian clang version 15.0.7, GNU ld (GNU Binutils for Debian) 2.35.2
> >
> > Unfortunately, I don't have any reproducer for this issue yet.
> >
> > Downloadable assets:
> > disk image: https://storage.googleapis.com/syzbot-assets/fef94e788067/disk-d5280145.raw.xz
> > vmlinux: https://storage.googleapis.com/syzbot-assets/576412ea518b/vmlinux-d5280145.xz
> > kernel image: https://storage.googleapis.com/syzbot-assets/685a0e4be06b/bzImage-d5280145.xz
> >
> > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > Reported-by: syzbot+a67fc5321ffb4b311c98@xxxxxxxxxxxxxxxxxxxxxxxxx
> >
> > general protection fault, probably for non-canonical address 0xdffffc000000000a: 0000 [#1] PREEMPT SMP KASAN
> > KASAN: null-ptr-deref in range [0x0000000000000050-0x0000000000000057]
> > CPU: 1 PID: 10127 Comm: syz-executor.3 Not tainted 6.4.0-syzkaller-11478-gd528014517f2 #0
> > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 05/27/2023
> > RIP: 0010:__lock_acquire+0x10d/0x7f70 kernel/locking/lockdep.c:5012
> > Code: 85 75 18 00 00 83 3d 15 c8 2c 0d 00 48 89 9c 24 10 01 00 00 0f 84 f8 0f 00 00 83 3d 5c de b3 0b 00 74 34 48 89 d0 48 c1 e8 03 <42> 80 3c 00 00 74 1a 48 89 d7 e8 b4 51 79 00 48 8b 94 24 80 00 00
> > RSP: 0018:ffffc900169be9e0 EFLAGS: 00010006
> > RAX: 000000000000000a RBX: 1ffff92002d37d60 RCX: 0000000000000002
> > RDX: 0000000000000050 RSI: 0000000000000000 RDI: 0000000000000050
> > RBP: ffffc900169beca8 R08: dffffc0000000000 R09: 0000000000000001
> > R10: dffffc0000000000 R11: fffffbfff1d2fe76 R12: 0000000000000000
> > R13: 0000000000000001 R14: 0000000000000002 R15: ffff88802091d940
> > FS:  0000000000000000(0000) GS:ffff8880b9900000(0000) knlGS:0000000000000000
> > CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> > CR2: 00007fa22a3fe000 CR3: 000000004b5e1000 CR4: 00000000003506e0
> > DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> > DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> > Call Trace:
> >  <TASK>
> >  lock_acquire+0x1e3/0x520 kernel/locking/lockdep.c:5761
> >  seqcount_lockdep_reader_access+0x139/0x220 include/linux/seqlock.h:102
> >  get_fs_root_rcu fs/d_path.c:244 [inline]
> >  d_path+0x2f0/0x6e0 fs/d_path.c:286
> >  audit_log_d_path+0xd3/0x310 kernel/audit.c:2139
> >  dump_common_audit_data security/lsm_audit.c:224 [inline]
> >  common_lsm_audit+0x7cf/0x1a90 security/lsm_audit.c:458
> >  smack_log+0x421/0x540 security/smack/smack_access.c:383
> >  smk_tskacc+0x2ff/0x360 security/smack/smack_access.c:253
> >  smack_inode_getattr+0x203/0x270 security/smack/smack_lsm.c:1202
> >  security_inode_getattr+0xd3/0x120 security/security.c:2114
> >  vfs_getattr+0x25/0x70 fs/stat.c:167
> >  ovl_getattr+0x1b1/0xf70 fs/overlayfs/inode.c:174
> >  ima_check_last_writer security/integrity/ima/ima_main.c:171 [inline]
> >  ima_file_free+0x26e/0x4b0 security/integrity/ima/ima_main.c:203
>
> Ugh, I think the root of this might all be the call back into
> vfs_getattr() that happens on overlayfs:
>
> __fput()
> -> ima_file_free()
>    -> mutex_lock()
>    -> vfs_getattr_nosec()
>       -> i_op->getattr() == ovl_getattr()
>          -> vfs_getattr()
>             -> security_inode_getattr()
>    -> mutex_unlock()
>
> So either overlayfs needs to call vfs_getattr_nosec() when the request
> comes from vfs_getattr_nosec() or this needs to use
> backing_file_real_path() to operate on the real underlying path.
>
> Thoughts?

The latter.

IMA code cannot operate on a mixture of real inode (file_inode())
real dentry (file_dentry()) and ovl path, especially for reading
stat.change_cookie which is not really well defined in ovl.

At least those direct f_path references need to be fixed:

security/integrity/ima/ima_main.c:
vfs_getattr_nosec(&file->f_path, &stat,
security/integrity/ima/ima_api.c:       result =
vfs_getattr_nosec(&file->f_path, &stat, STATX_CHANGE_COOKIE,
security/integrity/ima/ima_crypto.c:            f =
dentry_open(&file->f_path, flags, file->f_cred);

and then all the places that format full path for audit logs:
security/integrity/ima/ima_main.c:      *pathname =
ima_d_path(&file->f_path, pathbuf, filename);

Need to decide if it is prefered to log the full ovl path or the
relative real path (relative to the private mount clone of the ovl layer).

Thanks,
Amir.




[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