Re: [PATCH perf/core 00/22] perf refcnt debugger API and fixes

["Wangnan (F)" <wangnan0@huawei.com>]

On 2015/12/11 10:08, 平松雅巳 / HIRAMATU，MASAMI wrote:
> From: Wangnan (F) [mailto:wangnan0@huawei.com]
>> On 2015/12/10 23:12, 'Arnaldo Carvalho de Melo' wrote:
>>
>> [SNIP]
>>> But this requires having these special refcnt__ routines, that will make
>>> tools/perf/ code patterns for reference counts look different that the
>>> refcount patterns in the kernel :-\
>>>
>>> And would be a requirement to change the observed workload :-\
>>>
>>> Is this _strictly_ required?
>> No. The requirement should be:
>>
>>   1. The create/get/put/delete functions are non-inline (because dwarf info
>>      is not as reliable as symbol);
>>   2. From their argument list, we can always get the variable we need (the
>>      pointer of objects, the value of refcnt, etc.)
> However, we have to customize it for each application. Perf itself might be OK
> but others might have different implementation.

If limited to pairwise operations ({{m,c}alloc,strdup} vs free, open vs 
close),
I think it is possible to abstract a uniformed pattern.

Thank you.