[PATCH v8 23/70] mm/mmap: change do_brk_flags() to expand existing VMA and add do_brk_munmap()

[Janosch Frank]

On 5/3/22 23:55, Liam Howlett wrote:
> * Heiko Carstens <hca at linux.ibm.com> [220503 15:49]:
>> On Mon, May 02, 2022 at 08:50:04PM +0200, Heiko Carstens wrote:
>>> On Mon, May 02, 2022 at 01:31:00PM +0000, Liam Howlett wrote:
>>>> * Heiko Carstens <hca at linux.ibm.com> [220502 06:18]:
>>>>> On Sun, May 01, 2022 at 05:24:12PM -0700, Andrew Morton wrote:
>>>>>> (cc S390 maintainers)
>>>>>> (cc stable & Greg)
>> ...
>>>>>>> booting.  The last thing I see is:
>>>>>>>
>>>>>>> "[    4.668916] Spectre V2 mitigation: execute trampolines"
>>>>>>>
>>>>>>> I've bisected back to commit e553f62f10d9 (mm, page_alloc: fix
>>>>>>> build_zonerefs_node())
>>>>>>>
>>>>>>> With the this commit, I am unable to boot one out of three times.  When
>>>>>>> using the previous commit I was not able to get it to hang after trying
>>>>>>> 10+ times.  This is a qemu s390 install with KASAN on and I see no error
>>>>>>> messages.  I think it's likely it is this patch, but no guaranteed.
>> ...
>>>>> Liam, could you share your kernel config?
>>>>
>>>> Sure thing.  See attached.
>>>
>>> So, I can reproduce the hanging system now. However this looks like a
>>> qemu problem on s390, since I can reproduce this only with Qemu+TCG.
>>> Qemu with kvm works without any problems (same if I use z/VM as
>>> hypervisor).
>>>
>>> Janosch, Claudio, can you have a look at this please?
>>
>> So, at least for me this problem also exists with plain v5.17.
>> Switching off KASAN, or alternatively switching to KASAN_INLINE
>> "fixes" it for me with Qemu+TCG.
>>
>> Liam, could you please also try to disable KASAN in your config? With
>> that I think we can be almost sure this could be some bug in Qemu.
> 
> With KASAN, my tree fails 100% of the time (mm-stable + my maple tree
> patches)
> 
> Without KASAN, it boots 100% of the time.
> 
> I think this verifies with you say above?
> 
> Thanks,
> Liam

I had a short look yesterday and the boot usually hangs in the raid6 
code. Disabling vector instructions didn't make a difference but a few 
interruptions via GDB solve the problem for some reason.

CCing David and Thomas for TCG