On Thu, Mar 7, 2013 at 2:18 PM, Dave Jones <da...@redhat.com> wrote:
>
> Ok, that didn't enlightenment me so much..
Ooh, I'm good.
> [ 304.559707] WARNING: at fs/namei.c:696 nd_jump_link+0x85/0xa0()
> [ 304.692736] [<ffffffff811c1795>] nd_jump_link+0x85/0xa0
> [ 304.693541] [<ffffffff81224aec>] proc_pid_follow_link+0x6c/0x70
> [ 304.694445] [<ffffffff811be661>] path_lookupat+0x2d1/0x740
> [ 304.695270] [<ffffffff811beb04>] filename_lookup+0x34/0xc0
> [ 304.696112] [<ffffffff811c193e>] user_path_at_empty+0x8e/0x110
> [ 304.697007] [<ffffffff811c19d1>] user_path_at+0x11/0x20
> [ 304.697800] [<ffffffff811d944f>] sys_setxattr+0x3f/0xe0
> [ 304.698609] [<ffffffff816cdbc2>] system_call_fastpath+0x16/0x1b
> [ 304.722423] ---[ end trace 469ba8f58f5994e6 ]---
> [ 304.723132] old=fd new=1:9
>
> thoughts ?
Yes, I can reproduce this, and I know what's going on. I don't have a
patch, though, and I think it's in Al's court on what we should do.
Al?
Anyway, so the "old=fd" thing implies that it's one of the symlinks in
/proc/xyz/fd/, so that's something. It could have been the
cwd/exe/mmap ones. But that was kind of the expectation to begin with.
The "new=1:9" is the path that the file descriptor contained, and
since you're playing games with /proc, /sys and /dev, *and* since you
hit the "it shouldn't be a symlink" test, I'm *guessing* that it might
be something like
/sys/dev/char/1:9
which is a symlink to ../../devices/virtual/mem/urandom/.
And here is how you reproduce this:
Do:
int main(int argc, char **argv)
{
int fd = open(argv[1], O_PATH | O_NOFOLLOW);
struct stat st;
fstat(fd, &st);
sleep(100);
}
and then run
./a.out /sys/dev/char/1:9 &
which will look up that symlink using O_NOFOLLOW. Take note of the
<pid>, and then do
ls -l /proc/<pid>/fd/3/
and you get the oops, because when it follows that pseudo-symlink for
fd3, it will hit that symlink we just looked up. And trigger the
BUG_ON() you see.
Al, I think the BUG_ON() is simply bogus. Should we just remove it? Or
is there some conceptual reason why we can't handle a symlink there?
Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
