linux-kernel - Re: Kernel 4.17.4 lockup

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:   Wed, 11 Jul 2018 08:40:56 -0700
From:   "H.J. Lu" <hjl.tools@...il.com>
To:     Dave Hansen <dave.hansen@...el.com>
Cc:     "H. Peter Anvin" <hpa@...or.com>,
        LKML <linux-kernel@...r.kernel.org>,
        Andy Lutomirski <luto@...nel.org>,
        Mel Gorman <mgorman@...e.de>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Rik van Riel <riel@...riel.com>,
        Minchan Kim <minchan@...nel.org>
Subject: Re: Kernel 4.17.4 lockup

On Wed, Jul 11, 2018 at 8:13 AM, Dave Hansen <dave.hansen@...el.com> wrote:
> On 07/11/2018 07:56 AM, H.J. Lu wrote:
>> On Mon, Jul 9, 2018 at 8:47 PM, Dave Hansen <dave.hansen@...el.com> wrote:
>>> On 07/09/2018 07:14 PM, H.J. Lu wrote:
>>>>> I'd really want to see this reproduced without KASLR to make the oops
>>>>> easier to read.  It would also be handy to try your workload with all
>>>>> the pedantic debugging: KASAN, slab debugging, DEBUG_PAGE_ALLOC, etc...
>>>>> and see if it still triggers.
>>>> How can I turn them on at boot time?
>>> The only thing you can add at boot time is slab debugging, and it's
>>> probably the most useless of the three that I listed since you're not
>>> actually seeing any slab corruption.
>>>
>>> The rest are compile-time options.
>> I enabled KASAN, slab debugging, DEBUG_PAGE_ALLOC and disabled
>> KASLR.  Machine locked up.  Here is the last kernel message before locking
>> up.
>
> KASAN looks to have caught it, although it scrolled off the screen.  I
> can certainly imagine the oops you saw earlier being caused by stack
> corruption.
>
> Sounds like we need to reproduce this in an environment that can
> actually capture a real oops.  Can you share more about your workload?
> I'll see if I can get it to reproduce in a VM.

This is a quad-core machine with HT and 6 GB RAM.  The workload is
x32 GCC build and test with "make -j8".  The bug is triggered during GCC
test after a couple hours.  I have a script to set up my workload:

https://github.com/hjl-tools/gcc-regression

with a cron job

# It takes about 3 hour to bootstrap x86-64 GCC and 3 hour to run tests,
TIMEOUT=480
# Run it every hour,
30 * * * * /export/gnu/import/git/gcc-test-x32/gcc-build -mx32
--with-pic > /dev/null 2>&1


-- 
H.J.