Re: [Qemu-devel] Status of my hacks on the MTTCG WIP branch

[Alex Bennée]

alvise rigo <address@hidden> writes:

> On Fri, Jan 15, 2016 at 4:25 PM, Alex Bennée <address@hidden> wrote:
>>
>> alvise rigo <address@hidden> writes:
>>
>>> On Fri, Jan 15, 2016 at 3:51 PM, Alex Bennée <address@hidden> wrote:
>>>>
>>>> alvise rigo <address@hidden> writes:
>>>>
>>>>> This problem could be related to a missing multi-threaded aware
>>>>> translation of the atomic instructions.
>>>>> I'm working on this missing piece, probably the next week I will
>>>>> publish something.
>>>>
>>>> Maybe. We still have Fred's:
>>>>
>>>>   Use atomic cmpxchg to atomically check the exclusive value in a STREX
>>>>
>>>> Which I think papers over the cracks for both arm and aarch64 in MTTCG
>>>> while not being as correct as your work.
>>>
>>> Keep in mind that Linux on arm64 uses the LDXP/STXP instructions that
>>> exist solely in aarch64.
>>> These instructions are purely emulated now and can potentially write
>>> 128 bits of data in a non-atomic fashion.
>>
>> Sure, but I doubt they are the reason for this hang as the kernel
>> doesn't use them.
>
> The kernel does use them for __cmpxchg_double in
> arch/arm64/include/asm/atomic_ll_sc.h.

I take it back, if I'd have grepped for "ldxp" instead of "stxp" I would
have seen it, sorry about that ;-)

> In any case, the normal exclusive instructions are also emulated in
> target-arm/translate-a64.c.

I'll check on them on Monday. I'd assumed all the stuff was in the
helpers as I scanned through and missed the translate.c changes Fred
made. Hopefully that will be the last hurdle.

In the meantime if I'm not booting Jessie I can get MTTCG aarch64
working with a initrd based rootfs. Once I've gone through those I'm
planning on giving it a good stress test with -fsantize=threads.

>
> alvise
>
>>
>>>
>>> Regards,
>>> alvise
>>>
>>>>
>>>>>
>>>>> Regards,
>>>>> alvise
>>>>>
>>>>> On Fri, Jan 15, 2016 at 3:24 PM, Pranith Kumar <address@hidden> wrote:
>>>>>> Hi Alex,
>>>>>>
>>>>>> On Fri, Jan 15, 2016 at 8:53 AM, Alex Bennée <address@hidden> wrote:
>>>>>>> Can you try this branch:
>>>>>>>
>>>>>>>
>>>>>>> https://github.com/stsquad/qemu/tree/mttcg/multi_tcg_v8_wip_ajb_fix_locks-r1
>>>>>>>
>>>>>>> I think I've caught all the things likely to screw up addressing.
>>>>>>>
>>>>>>
>>>>>> I tried this branch and the boot hangs like follows:
>>>>>>
>>>>>> [    2.001083] random: systemd-udevd urandom read with 1 bits of entropy
>>>>>> available
>>>>>> main-loop: WARNING: I/O thread spun for 1000 iterations
>>>>>> [   23.778970] INFO: rcu_sched detected stalls on CPUs/tasks: {} 
>>>>>> (detected
>>>>>> by 0, t=2102 jiffies, g=-165, c=-166, q=83)
>>>>
>>>> This is just saying the kernel has been waiting for a while and nothing
>>>> has happened.
>>>>
>>>>>> I will try to debug and see where it is hanging.
>>>>
>>>> If we knew what the kernel was waiting for that would be useful to know.
>>>>
>>>>>>
>>>>>> Thanks!
>>>>>> --
>>>>>> Pranith
>>>>
>>>>
>>>> --
>>>> Alex Bennée
>>
>>
>> --
>> Alex Bennée


--
Alex Bennée