[dpdk-dev] pthread_barrier_deadlock in -rc1

[Tan, Jianfeng]

On 5/2/2018 6:08 PM, Olivier Matz wrote:
> On Wed, May 02, 2018 at 06:01:20PM +0800, Tan, Jianfeng wrote:
>> Hi Olivier and Anatoly,
>>
>> [...]
>>
>>>>> Below patch can fix another strange sigsegv issue in my VM. Please check
>>>>> if it works for you. I doubt it's use-after-free problem which could
>>>>> lead to different issues in different env. Please have a try.
>>>>>
>>>>>
>>>>> diff --git a/lib/librte_eal/common/eal_common_thread.c
>>>>> b/lib/librte_eal/common/eal_common_thread.c
>>>>> index de69452..d91b67d 100644
>>>>> --- a/lib/librte_eal/common/eal_common_thread.c
>>>>> +++ b/lib/librte_eal/common/eal_common_thread.c
>>>>> @@ -205,6 +205,7 @@ rte_ctrl_thread_create(pthread_t *thread, const char
>>>>> *name,
>>>>>                    goto fail;
>>>>>
>>>>>            pthread_barrier_wait(&params->configured);
>>>>> +       pthread_barrier_destroy(&params->configured);
>>>>>            free(params);
>>>> Should destroy also be called in fail case?
>>> Yes, and also pthread_barrier_wait().
>>>
>>> I did a quick test simulating a failure of rte_thread_setname(), and
>>> without the wait, the barrier in the child thread blocks forever.
>>>
>> v1 has been sent without seeing this email. You are right, I misunderstood
>> it. Will send a v2 based on this result.
>>
>> Thanks,
>> Jianfeng
> Thanks.
>
> Not sure it should be in the same patch, however there is also
> a memory leak here:
>
>          ret = pthread_create(thread, attr, rte_thread_init, (void *)params);
>          if (ret != 0)
>                  return ret;

My bad, will fix it in a separate patch.

Thanks,
Jianfeng