linux-kernel - Re: blktrace: NULL ptr deref in sysfs_blk_trace_attr

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: <536A54C5.9070803@oracle.com>
Date:	Wed, 07 May 2014 11:44:05 -0400
From:	Sasha Levin <sasha.levin@...cle.com>
To:	Jan Kara <jack@...e.cz>
CC:	axboe@...nel.dk, LKML <linux-kernel@...r.kernel.org>,
	Dave Jones <davej@...hat.com>
Subject: Re: blktrace: NULL ptr deref in sysfs_blk_trace_attr_store

On 05/07/2014 11:41 AM, Jan Kara wrote:
>   Hello,
> 
> On Tue 06-05-14 22:45:19, Sasha Levin wrote:
>> While fuzzing with trinity inside a KVM tools guest running the latest -next
>> kernel I've stumbled on the following spew:
>>
>> [ 6230.740626] BUG: unable to handle kernel NULL pointer dereference at           (null)
>> [ 6230.740626] IP: __list_del_entry (lib/list_debug.c:57)
>> [ 6230.740626] PGD 44d661067 PUD 44d660067 PMD 0
>> [ 6230.740626] Oops: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC
>> [ 6230.740626] Dumping ftrace buffer:
>> [ 6230.740626]    (ftrace buffer empty)
>> [ 6230.740626] Modules linked in:
>> [ 6230.740626] CPU: 2 PID: 23998 Comm: trinity-c257 Not tainted 3.15.0-rc4-next-20140506-sasha-00021-gc164334-dirty #447
>> [ 6230.740626] task: ffff8804716db000 ti: ffff88044ca9e000 task.ti: ffff88044ca9e000
>> [ 6230.740626] RIP: __list_del_entry (lib/list_debug.c:57)
>> [ 6230.740626] RSP: 0018:ffff88044ca9fdd8  EFLAGS: 00010007
>> [ 6230.740626] RAX: 0000000000000000 RBX: ffff88005a455948 RCX: dead000000200200
>> [ 6230.740626] RDX: 0000000000000000 RSI: ffffffff938fdf78 RDI: ffff88005a455948
>> [ 6230.740626] RBP: ffff88044ca9fdd8 R08: 0000000000000001 R09: ffff8804716dbdd0
>> [ 6230.740626] R10: ffff8804716db000 R11: 0000000000000000 R12: 00000000000007f5
>> [ 6230.740626] R13: ffff88017b880000 R14: ffff88005a4558f0 R15: ffff88036a188758
>> [ 6230.740626] FS:  00007fec10273700(0000) GS:ffff8800a6c00000(0000) knlGS:0000000000000000
>> [ 6230.740626] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
>> [ 6230.740626] CR2: 0000000000000000 CR3: 000000044d662000 CR4: 00000000000006a0
>> [ 6230.740626] DR0: 00000000006df000 DR1: 0000000000000000 DR2: 0000000000000000
>> [ 6230.740626] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000600
>> [ 6230.740626] Stack:
>> [ 6230.740626]  ffff88044ca9fdf8 ffffffff8fb40441 ffff88036a188740 ffff88036a188740
>> [ 6230.740626]  ffff88044ca9fe68 ffffffff8f25d75d 0000000000000000 ffff8804716db000
>> [ 6230.740626]  0000000000000286 ffff880333f61330 0000000000000000 0000000000000286
>> [ 6230.740626] Call Trace:
>> [ 6230.740626] list_del (lib/list_debug.c:78)
>> [ 6230.740626] sysfs_blk_trace_attr_store (include/linux/spinlock.h:353 kernel/trace/blktrace.c:1498 kernel/trace/blktrace.c:1740)
>> [ 6230.740626] dev_attr_store (drivers/base/core.c:138)
>> [ 6230.740626] sysfs_kf_write (fs/sysfs/file.c:114)
>> [ 6230.740626] kernfs_fop_write (fs/kernfs/file.c:295)
>> [ 6230.740626] vfs_write (fs/read_write.c:532)
>> [ 6230.740626] SyS_write (fs/read_write.c:584 fs/read_write.c:576)
>> [ 6230.740626] ia32_do_call (arch/x86/ia32/ia32entry.S:430)
>> [ 6230.740626] Code: 7d 93 be 3e 00 00 00 48 c7 c7 d2 46 7d 93 31 c0 e8 6f fb 61 ff eb 2c 0f 1f 44 00 00 48 b9 00 02 20 00 00 00 ad de 48 39 c8 74 8c <4c> 8b 00 4c 39 c7 75 a6 4c 8b 42 08 4c 39 c7 75 bc 48 89 42 08
>> [ 6230.740626] RIP __list_del_entry (lib/list_debug.c:57)
>> [ 6230.740626]  RSP <ffff88044ca9fdd8>
>> [ 6230.740626] CR2: 0000000000000000
>   Hum, interesting. So it seems that blk_trace_remove_queue() got struct
> blk_trace which isn't healthy - list_head has been zeroed out and since
> this is the first thing we touch in the structure we don't really know what
> else is in there. But I don't see how such structure could get there since
> all places that modify q->blk_trace seem to hold bdev->bd_mutex.
> 
> OTOH the code in blk_trace_setup_queue() does:
>         old_bt = xchg(&q->blk_trace, bt);
>         if (old_bt != NULL) {
>                 (void)xchg(&q->blk_trace, old_bt);
>                 ret = -EBUSY;
>                 goto free_bt;
>         }
> 
> Which looks somewhat strange if modifications of q->blk_trace are really
> guarded by bd_mutex. Jens?

I probably should have mentioned that initially, but I've added some scripting to
stress CPU and memory hotplugging recently, so there's a good chance that this somehow
occurred as a result of a CPU going away/coming back.


Thanks,
Sasha

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/