linux-kernel - Re: PROBLEM: syzkaller found / pool corruption-overwrite / page in user-area or NULL

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]

Message-ID: <3b3184e0-d913-6519-0f9d-2f01ef795650@lca.pw>
Date:   Thu, 10 Jan 2019 23:52:58 -0500
From:   Qian Cai <cai@....pw>
To:     Esme <esploit@...tonmail.ch>
Cc:     James Bottomley <jejb@...ux.ibm.com>,
        "dgilbert@...erlog.com" <dgilbert@...erlog.com>,
        "martin.petersen@...cle.com" <martin.petersen@...cle.com>,
        "linux-scsi@...r.kernel.org" <linux-scsi@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-mm@...ck.org" <linux-mm@...ck.org>
Subject: Re: PROBLEM: syzkaller found / pool corruption-overwrite / page in
 user-area or NULL



On 1/10/19 10:15 PM, Esme wrote:
>>> [ 75.793150] RIP: 0010:rb_insert_color+0x189/0x1480
>>
>> What's in that line? Try,
>>
>> $ ./scripts/faddr2line vmlinux rb_insert_color+0x189/0x1480
> 
> rb_insert_color+0x189/0x1480:
> __rb_insert at /home/files/git/linux/lib/rbtree.c:131
> (inlined by) rb_insert_color at /home/files/git/linux/lib/rbtree.c:452
> 

gparent = rb_red_parent(parent);

tmp = gparent->rb_right; <-- GFP triggered here.

It suggests gparent is NULL. Looks like it misses a check there because parent
is the top node.

>>
>> What's steps to reproduce this?
> 
> The steps is the kernel config provided (proc.config) and I double checked the attached C code from the qemu image (attached here).  If the kernel does not immediately crash, a ^C will cause the fault to be noticed.  The report from earlier is the report from the same code, my assumption was that the possible pool/redzone corruption is making it a bit tricky to pin down.
> 
> If you would like alternative kernel settings please let me know, I can do that, also, my current test-bench has about 256 core's on x64, 64 of them are bare metal and 32 are arm64.  Any possible preferred configuration tweaks I'm all ears, I'll be including some of these steps you suggested to me in any/additional upcoming threads (Thank you for that so far and future suggestions).
> 
> Also, there is some occasionally varying stacks depending on the corruption, so this stack just now (another execution of test3.c);

I am unable to reproduce any of those here. What's is the output of
/proc/cmdline in your guest when this happens?