Because our fuzzer has a problem, I don't have a C reproducer so far.
I reported the crash becasue I saw the crash repeatedly in our fuzzer and I hoped the report is helpful. But it seems not enough.
If I was wrong and I made you confused, I am really sorry for that.
Could you give me a second?
I am trying to fix our fuzzer and to make a C reproducer.
I think the C reproducer is necessary here.
On 24 Jul 2018, 2:29 PM +0900, Al Viro <viro@xxxxxxxxxxxxxxxxxx>, wrote:
On Tue, Jul 24, 2018 at 06:17:26AM +0100, Al Viro wrote:
On Tue, Jul 24, 2018 at 12:45:42PM +0900, Dae R. Jeong wrote:
Diagnosis:
We think that it is possible that link_path_walk() dereferences a
freed pointer when cleanup_mnt() is executed between path_init() and
link_path_walk().
Since I'm not an expert on a file system and don't fully understand
the crash, please see a executed program and a crash log below in
case that my understanding is wrong.
Executed Program:
Thread0 Thread1
mkdir("./file0")
|--------------------------|
| mount("./file0", "./file0", "devpts", 0x0, "")
| |
openat(AT_FDCWD, chroot("./file0")
"/dev/vcs", 0x200, 0x0) umount("./file0", 0x2)
openat(), chroot(), umount() syscalls are executed after mount() syscall.
We think a race occurs between openat() and chroot() because RaceFuzzer
executed openat() and chroot() concurrently.
(Possible) Thread interleaving:
CPU0 (path_openat) CPU1 (cleanup_mnt)
Wait a bloody minute. Where does cleanup_mnt() come from in that thing?
You are doing lazy-umount of the thing you've chrooted into; if it ends
up with zero refcount on that mount, we are already in deep, deep trouble,
races with open() on not. Simply following that with stat / (in thread 1,
without thread0 at all) would end up accessing the same vfsmount. And
if it's been freed, we are well and truly fucked, race or no race.
I really want details. *Is* cleanup_mnt() called by thread 1 in your
reproducer before the use-after-free hits? And what's the root of
thread 0 at that point?