linux-kernel - Re: KCSAN: data-race in task_dump_owner / task_dump

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 17 Oct 2019 14:56:47 +0200
From:   Marco Elver <elver@...gle.com>
To:     syzbot <syzbot+e392f8008a294fdf8891@...kaller.appspotmail.com>
Cc:     adobriyan@...il.com, Andrew Morton <akpm@...ux-foundation.org>,
        casey@...aufler-ca.com, christian@...uner.io,
        Kees Cook <keescook@...omium.org>, kent.overstreet@...il.com,
        khlebnikov@...dex-team.ru, linux-fsdevel@...r.kernel.org,
        LKML <linux-kernel@...r.kernel.org>, mhocko@...e.com,
        shakeelb@...gle.com, syzkaller-bugs@...glegroups.com,
        Thomas Gleixner <tglx@...utronix.de>
Subject: Re: KCSAN: data-race in task_dump_owner / task_dump_owner

Hi,

On Thu, 17 Oct 2019 at 14:36, syzbot
<syzbot+e392f8008a294fdf8891@...kaller.appspotmail.com> wrote:
>
> Hello,
>
> syzbot found the following crash on:
>
> HEAD commit:    d724f94f x86, kcsan: Enable KCSAN for x86
> git tree:       https://github.com/google/ktsan.git kcsan
> console output: https://syzkaller.appspot.com/x/log.txt?x=17884db3600000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=c0906aa620713d80
> dashboard link: https://syzkaller.appspot.com/bug?extid=e392f8008a294fdf8891
> compiler:       gcc (GCC) 9.0.0 20181231 (experimental)
>
> Unfortunately, I don't have any reproducer for this crash yet.
>
> IMPORTANT: if you fix the bug, please add the following tag to the commit:
> Reported-by: syzbot+e392f8008a294fdf8891@...kaller.appspotmail.com
>
> ==================================================================
> BUG: KCSAN: data-race in task_dump_owner / task_dump_owner
>
> write to 0xffff8881255bb7fc of 4 bytes by task 7804 on cpu 0:
>   task_dump_owner+0xd8/0x260 fs/proc/base.c:1742
>   pid_update_inode+0x3c/0x70 fs/proc/base.c:1818
>   pid_revalidate+0x91/0xd0 fs/proc/base.c:1841
>   d_revalidate fs/namei.c:765 [inline]
>   d_revalidate fs/namei.c:762 [inline]
>   lookup_fast+0x7cb/0x7e0 fs/namei.c:1613
>   walk_component+0x6d/0xe80 fs/namei.c:1804
>   link_path_walk.part.0+0x5d3/0xa90 fs/namei.c:2139
>   link_path_walk fs/namei.c:2070 [inline]
>   path_openat+0x14f/0x3530 fs/namei.c:3532
>   do_filp_open+0x11e/0x1b0 fs/namei.c:3563
>   do_sys_open+0x3b3/0x4f0 fs/open.c:1089
>   __do_sys_open fs/open.c:1107 [inline]
>   __se_sys_open fs/open.c:1102 [inline]
>   __x64_sys_open+0x55/0x70 fs/open.c:1102
>   do_syscall_64+0xcf/0x2f0 arch/x86/entry/common.c:296
>   entry_SYSCALL_64_after_hwframe+0x44/0xa9
>
> write to 0xffff8881255bb7fc of 4 bytes by task 7813 on cpu 1:
>   task_dump_owner+0xd8/0x260 fs/proc/base.c:1742
>   pid_update_inode+0x3c/0x70 fs/proc/base.c:1818
>   pid_revalidate+0x91/0xd0 fs/proc/base.c:1841
>   d_revalidate fs/namei.c:765 [inline]
>   d_revalidate fs/namei.c:762 [inline]
>   lookup_fast+0x7cb/0x7e0 fs/namei.c:1613
>   walk_component+0x6d/0xe80 fs/namei.c:1804
>   lookup_last fs/namei.c:2271 [inline]
>   path_lookupat.isra.0+0x13a/0x5a0 fs/namei.c:2316
>   filename_lookup+0x145/0x2d0 fs/namei.c:2346
>   user_path_at_empty+0x4c/0x70 fs/namei.c:2606
>   user_path_at include/linux/namei.h:60 [inline]
>   vfs_statx+0xd9/0x190 fs/stat.c:187
>   vfs_stat include/linux/fs.h:3188 [inline]
>   __do_sys_newstat+0x51/0xb0 fs/stat.c:341
>   __se_sys_newstat fs/stat.c:337 [inline]
>   __x64_sys_newstat+0x3a/0x50 fs/stat.c:337
>   do_syscall_64+0xcf/0x2f0 arch/x86/entry/common.c:296
>   entry_SYSCALL_64_after_hwframe+0x44/0xa9
>
> Reported by Kernel Concurrency Sanitizer on:
> CPU: 1 PID: 7813 Comm: ps Not tainted 5.3.0+ #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
> Google 01/01/2011
> ==================================================================

My understanding is, that for every access to /proc/<pid>,
d_revalidate is called, and /proc-fs implementation simply says that
pid_revalidate always revalidates by rewriting uid/gid because "owning
task may have performed a setuid(), etc." presumably so every access
to a /proc/<pid> entry always has the right uid/gid (in effect
updating /proc/<pid> lazily via d_revalidate).

Is it possible that one of the tasks above could be preempted after
doing its writes to *ruid/*rgid, another thread writing some other
values (after setuid / seteuid), and then the preempted thread seeing
the other values? Assertion here should never fail:
=== TASK 1 ===
| seteuid(1000);
| seteuid(0);
| stat("/proc/<pid-of-task-1>", &fstat);
| assert(fstat.st_uid == 0);
=== TASK 2 ===
| stat("/proc/<pid-of-task-1>", ...);


Best Wishes,
-- Marco