[PATCH v3] Improve the performance of --num-threads -d 31

["Zhou, Wenjian/周文剑"]

Hi Minoru,

I guess if you change the part of locking page_data_buf to the following,
it will be more correct.
But I'm not sure if you will get the same problem as which in my patch v3.

+		pthread_mutex_lock(&info->page_data_mutex);
+		while (page_data_buf[index].used != FALSE) {
+			index = (index + 1) % info->num_buffers;
+		}
+		page_data_buf[index].used = TRUE;
+		pthread_mutex_unlock(&info->page_data_mutex);

-- 
Thanks
Zhou

On 03/07/2016 04:52 PM, Minoru Usui wrote:
> Hi, Zhou
>
>> -----Original Message-----
>> From: kexec [mailto:kexec-bounces at lists.infradead.org] On Behalf Of "Zhou, Wenjian/周文?"
>> Sent: Monday, March 07, 2016 5:43 PM
>> To: Usui Minoru(碓井 成) <min-usui at ti.jp.nec.com>
>> Cc: Minfei Huang <mhuang at redhat.com>; kexec at lists.infradead.org
>> Subject: Re: [PATCH v3] Improve the performance of --num-threads -d 31
>>
>> Hi Minoru,
>>
>> I got two error dumpfiles by running it less than 100 times.
>
> Oh, I'm sorry.
> Thank you for your test.
>
> I'm waiting for your v4 patch.
>
> Thanks
> Minoru Usui
>
>
>
>> --
>> Thanks
>> Zhou
>>
>> On 03/07/2016 04:22 PM, "Zhou, Wenjian/周文?" wrote:
>>> Hi Minoru,
>>>
>>> I have test your patch since you post.
>>> I can get a failure easily.
>>> If you run it several times, the dumpfile it produces may be different from the correct one.
>>>
>>> +			info->current_pfn = pfn + 1;
>>> +			pthread_mutex_unlock(&info->current_pfn_mutex);
>>> +			page_flag_buf->pfn = pfn;
>>> +			page_flag_buf->ready = FLAG_FILLING;
>>> The above code should be changed to the following:
>>>
>>> +			info->current_pfn = pfn + 1;
>>> +			page_flag_buf->pfn = pfn;
>>> +			page_flag_buf->ready = FLAG_FILLING;
>>> +			pthread_mutex_unlock(&info->current_pfn_mutex);
>>>
>>> I haven't run your code with the bug fixed.
>>> I'll try now.
>>>
>>> By the way, with my patch v3, I can't get an error dumpfile until run it by 5000 times.
>>
>>
>>
>> _______________________________________________
>> kexec mailing list
>> kexec at lists.infradead.org
>> http://lists.infradead.org/mailman/listinfo/kexec