lkml.org 
[lkml]   [2021]   [Jan]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH v6] perf stat: Fix wrong skipping for per-die aggregation
From
Date
Hi Arnaldo, Jiri,

On 1/16/2021 4:31 AM, Arnaldo Carvalho de Melo wrote:
> Em Fri, Jan 15, 2021 at 05:28:14PM -0300, Arnaldo Carvalho de Melo escreveu:
>> Em Thu, Jan 14, 2021 at 08:00:32PM +0100, Jiri Olsa escreveu:
>>> On Thu, Jan 14, 2021 at 09:27:55AM +0800, Jin Yao wrote:
>>>
>>> SNIP
>>>
>>>> 2.003776312 S1-D0 1 855616 Bytes llc_misses.mem_read
>>>> 2.003776312 S1-D1 1 949376 Bytes llc_misses.mem_read
>>>> 3.006512788 S0-D0 1 1338880 Bytes llc_misses.mem_read
>>>> 3.006512788 S0-D1 1 920064 Bytes llc_misses.mem_read
>>>> 3.006512788 S1-D0 1 877184 Bytes llc_misses.mem_read
>>>> 3.006512788 S1-D1 1 1020736 Bytes llc_misses.mem_read
>>>> 4.008895291 S0-D0 1 926592 Bytes llc_misses.mem_read
>>>> 4.008895291 S0-D1 1 906368 Bytes llc_misses.mem_read
>>>> 4.008895291 S1-D0 1 892224 Bytes llc_misses.mem_read
>>>> 4.008895291 S1-D1 1 987712 Bytes llc_misses.mem_read
>>>> 5.001590993 S0-D0 1 962624 Bytes llc_misses.mem_read
>>>> 5.001590993 S0-D1 1 912512 Bytes llc_misses.mem_read
>>>> 5.001590993 S1-D0 1 891200 Bytes llc_misses.mem_read
>>>> 5.001590993 S1-D1 1 978432 Bytes llc_misses.mem_read
>>>>
>>>> On no-die system, die_id is 0, actually it's hashmap(socket,0), original behavior
>>>> is not changed.
>>>>
>>>> Reported-by: Huang Ying <ying.huang@intel.com>
>>>> Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
>>>> ---
>>>> v6:
>>>> Fix the perf test python failure by adding hashmap.c to python-ext-sources.
>>>>
>>>> root@kbl-ppc:~# ./perf test python
>>>> 19: 'import perf' in python : Ok
>>>
>>> Acked-by: Jiri Olsa <jolsa@redhat.com>
>>
>> Jin, this is breaking the build in some 32-bit system, can you please
>> take a look to validate these warnings?
>
> One such system:
>
> 28 13.75 debian:experimental-x-mipsel : FAIL mipsel-linux-gnu-gcc (Debian 10.2.1-3) 10.2.1 20201224
>
>
>> CC /tmp/build/perf/util/srcline.o
>> util/stat.c: In function 'pkg_id_hash':
>> util/stat.c:285:9: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
>> return (int64_t)key & 0xffffffff;
>> ^
>> util/stat.c: In function 'pkg_id_equal':
>> util/stat.c:291:9: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
>> return (int64_t)key1 == (int64_t)key2;
>> ^
>> util/stat.c:291:26: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
>> return (int64_t)key1 == (int64_t)key2;
>> ^
>> util/stat.c: In function 'check_per_pkg':
>> util/stat.c:342:26: error: cast to pointer from integer of different size [-Werror=int-to-pointer-cast]
>> if (hashmap__find(mask, (void *)key, NULL))
>> ^
>> util/stat.c:345:28: error: cast to pointer from integer of different size [-Werror=int-to-pointer-cast]
>> ret = hashmap__add(mask, (void *)key, (void *)1);
>> ^
>> CC /tmp/build/perf/tests/expand-cgroup.o
>

Thanks for reporting this build issue on 32 bit system.

In v7, I use size_t to replace uint64_t and change the hash key to 'die_id << 16 | socket_id'. I
assume that 16 bits is enough for socket id, is it true? But if it's not true, we have to use a more
complicated way such as allocate a structure which has die_id and socket_id and add it to hashmap.

But I'm not sure if that's necessary because I can't imagine a system which has socket id > 65535.

Thanks
Jin Yao

\
 
 \ /
  Last update: 2021-01-18 05:19    [W:0.096 / U:0.064 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site