Re: BUG: unable to handle kernel paging request in hmac_init_tfm
From: Dmitry Vyukov
Date: Fri Dec 22 2017 - 04:01:36 EST
On Fri, Dec 22, 2017 at 3:27 AM, Eric Biggers <ebiggers3@xxxxxxxxx> wrote:
> On Thu, Dec 21, 2017 at 08:44:03AM +0100, 'Dmitry Vyukov' via syzkaller-bugs wrote:
>> On Thu, Dec 21, 2017 at 12:09 AM, Eric Biggers <ebiggers3@xxxxxxxxx> wrote:
>> > On Mon, Dec 18, 2017 at 11:36:01AM -0800, syzbot wrote:
>> >> Hello,
>> >>
>> >> syzkaller hit the following crash on
>> >> 6084b576dca2e898f5c101baef151f7bfdbb606d
>> >> git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/master
>> >> compiler: gcc (GCC) 7.1.1 20170620
>> >> .config is attached
>> >
>> > FYI, in linux-next KASAN and other memory debugging options are now behind
>> > CONFIG_DEBUG_MEMORY. So, I think KASAN isn't getting turned on anymore, despite
>> > the CONFIG_KASAN=y. (Patch was "lib/: make "Memory Debugging" a menuconfig to
>> > ease disabling it all".)
>>
>> Ouch! That would be pretty bad.
>>
>> But I've tried both linux-next HEAD at:
>>
>> commit 0e08c463db387a2adcb0243b15ab868a73f87807 (HEAD, tag:
>> next-20171221, linux-next/master)
>> Author: Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx>
>> Date: Thu Dec 21 15:37:39 2017 +1100
>> Add linux-next specific files for 20171221
>>
>> and mmots HEAD at:
>>
>> commit 75aa5540627fdb3d8f86229776ea87f995275351 (HEAD, tag:
>> v4.15-rc4-mmots-2017-12-20-19-10, mmots/master)
>> Author: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
>> Date: Thu Dec 21 04:02:17 2017 +0000
>> pci: test for unexpectedly disabled bridges
>>
>> and after running make olddefconfig I still see CONFIG_KASAN=y in
>> .config, nor I can find CONFIG_DEBUG_MEMORY in menuconfig.
>>
>> What am I missing? What commit has added the config?
>>
>
> Ah, I see the patch was added to -mm on 2017-12-12 but removed two days later:
>
> https://www.spinics.net/lists/mm-commits/msg129685.html
> https://www.spinics.net/lists/mm-commits/msg129696.html
>
> So it was in linux-next temporarily (including the kernel version used for this
> particular bug report) but now it's not. Just keep an eye out for it coming
> back, I guess.
I see. Then I will add CONFIG_DEBUG_MEMORY=y to our config
proactively. Without that patch olddefconfig will just wipe that
config, so should work.