linux-kernel - Re: [PATCH] cgroup: never call do_group

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] [day] [month] [year] [list]

Message-ID: <20190509145949.GU374014@devbig004.ftw2.facebook.com>
Date:   Thu, 9 May 2019 07:59:49 -0700
From:   Tejun Heo <tj@...nel.org>
To:     Roman Gushchin <guro@...com>
Cc:     Oleg Nesterov <oleg@...hat.com>, kernel-team@...com,
        cgroups@...r.kernel.org, linux-kernel@...r.kernel.org,
        Qian Cai <cai@....pw>
Subject: Re: [PATCH] cgroup: never call do_group_exit() with task->frozen bit
 set

On Wed, May 08, 2019 at 01:34:20PM -0700, Roman Gushchin wrote:
> I've got two independent reports that cgroup_task_frozen() check
> in cgroup_exit() has been triggered by lkp libhugetlbfs-test and
> LTP ptrace01 tests.
> 
> For example:
> [   44.576072] WARNING: CPU: 1 PID: 3028 at kernel/cgroup/cgroup.c:5932 cgroup_exit+0x148/0x160
> [   44.577724] Modules linked in: crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel sr_mod cdrom
> bochs_drm sg ttm ata_generic pata_acpi ppdev drm_kms_helper snd_pcm syscopyarea aesni_intel snd_timer
> sysfillrect sysimgblt snd crypto_simd cryptd glue_helper soundcore fb_sys_fops joydev drm serio_raw pcspkr
> ata_piix libata i2c_piix4 floppy parport_pc parport ip_tables
> [   44.583106] CPU: 1 PID: 3028 Comm: ptrace-write-hu Not tainted 5.1.0-rc3-00053-g9262503 #5
> [   44.584600] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014
> [   44.586116] RIP: 0010:cgroup_exit+0x148/0x160
> [   44.587135] Code: 0f 84 50 ff ff ff 48 8b 85 c8 0c 00 00 48 8b 78 70 e8 ec 2e 00 00 e9 3b ff ff ff f0 ff 43 60
> 0f 88 72 21 89 00 e9 48 ff ff ff <0f> 0b e9 1b ff ff ff e8 3c 73 f4 ff 66 90 66 2e 0f 1f 84 00 00 00
> [   44.590113] RSP: 0018:ffffb25702dcfd30 EFLAGS: 00010002
> [   44.591167] RAX: ffff96a7fee32410 RBX: ffff96a7ff1d6000 RCX: dead000000000200
> [   44.592446] RDX: ffff96a7ff1d6080 RSI: ffff96a7fec75290 RDI: ffff96a7fec75290
> [   44.593715] RBP: ffff96a7fec745c0 R08: ffff96a7fec74658 R09: 0000000000000000
> [   44.594985] R10: 0000000000000000 R11: 0000000000000001 R12: ffff96a7fec75101
> [   44.596266] R13: ffff96a7fec745c0 R14: ffff96a7ff3bde30 R15: ffff96a7fec75130
> [   44.597550] FS:  0000000000000000(0000) GS:ffff96a7dd700000(0000) knlGS:0000000000000000
> [   44.598950] CS:  0010 DS: 002b ES: 002b CR0: 0000000080050033
> [   44.600098] CR2: 00000000f7a00000 CR3: 000000000d20e000 CR4: 00000000000406e0
> [   44.601417] Call Trace:
> [   44.602777]  do_exit+0x337/0xc40
> [   44.603677]  do_group_exit+0x3a/0xa0
> [   44.604610]  get_signal+0x12e/0x8d0
> [   44.605533]  ? __switch_to_asm+0x40/0x70
> [   44.606503]  do_signal+0x36/0x650
> [   44.607409]  ? __switch_to_asm+0x40/0x70
> [   44.608383]  ? __schedule+0x267/0x860
> [   44.609329]  exit_to_usermode_loop+0x89/0xf0
> [   44.610349]  do_fast_syscall_32+0x251/0x2e3
> [   44.611357]  entry_SYSENTER_compat+0x7f/0x91
> [   44.612376] ---[ end trace e4ca5cfc4b7f7964 ]---
> 
> The problem is caused by the ptrace_signal() call in the for loop
> in get_signal(). There is a cgroup_enter_frozen() call inside
> ptrace_signal(), so after exit from ptrace_signal() the task->frozen
> bit might be set. In this case do_group_exit() can be called with the
> task->frozen bit set and trigger the warning. This is only place where
> we can leave the loop with the task->frozen bit set and without
> setting JOBCTL_TRAP_FREEZE and TIF_SIGPENDING.
> 
> To resolve this problem, let's move cgroup_leave_frozen(true) call to
> just after the fatal label. If the task is going to die, the frozen
> bit must be cleared no matter how we get into this point.
> 
> Reported-by: kernel test robot <rong.a.chen@...el.com>
> Reported-by: Qian Cai <cai@....pw>
> Cc: Oleg Nesterov <oleg@...hat.com>
> Cc: Tejun Heo <tj@...nel.org>
> Signed-off-by: Roman Gushchin <guro@...com>

Applied to cgroup/for-5.2 for now.

Thanks.

-- 
tejun