| Message ID | 1602152121-240367-10-git-send-email-john.garry@huawei.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | perf pmu-events: Support event aliasing for system PMUs | expand |
On Sun, Oct 18, 2020 at 1:51 AM kernel test robot <rong.a.chen@intel.com> wrote: > > Greeting, > > FYI, we noticed the following commit (built with gcc-9): > > commit: fcc9c5243c478f104014daf4d23db86098d2aef0 ("perf metricgroup: Hack a fix for aliases when covering multiple PMUs") > url: https://github.com/0day-ci/linux/commits/John-Garry/perf-pmu-events-Support-event-aliasing-for-system-PMUs/20201008-182049 > > > in testcase: perf-sanity-tests > version: perf-x86_64-c85fb28b6f99-1_20201008 > with following parameters: > > perf_compiler: gcc > ucode: 0xdc > > > > on test machine: 4 threads Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz with 32G memory > > caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace): I believe this is a Skylake and there is a known bug in the Skylake metric DRAM_Parallel_Reads as described here: https://lore.kernel.org/lkml/CAP-5=fXejVaQa9qfW66cY77qB962+jbe8tT5bsLoOOcFmODnWQ@mail.gmail.com/ Fixing the bug needs more knowledge than what is available in manuals. Hopefully Intel can take a look. Thanks, Ian > > > If you fix the issue, kindly add following tag > Reported-by: kernel test robot <rong.a.chen@intel.com> > > > 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 67 > 67: Parse and process metrics : FAILED! > 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 68 > 68: x86 rdpmc : Ok > 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 69 > 69: Convert perf time to TSC : Ok > 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 70 > 70: DWARF unwind : Ok > 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 71 > 71: x86 instruction decoder - new instructions : Ok > 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 72 > 72: Intel PT packet decoder : Ok > 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 73 > 73: x86 bp modify : Ok > 2020-10-16 19:31:53 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 74 > 74: probe libc's inet_pton & backtrace it with ping : Ok > 2020-10-16 19:31:54 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 75 > 75: Zstd perf.data compression/decompression : Ok > > > > To reproduce: > > git clone https://github.com/intel/lkp-tests.git > cd lkp-tests > bin/lkp install job.yaml # job file is attached in this email > bin/lkp run job.yaml > > > > Thanks, > Rong Chen >
> I believe this is a Skylake and there is a known bug in the Skylake > metric DRAM_Parallel_Reads as described here: > https://lore.kernel.org/lkml/CAP-5=fXejVaQa9qfW66cY77qB962+jbe8tT5bsLoOOcFmODnWQ@mail.gmail.com/ > Fixing the bug needs more knowledge than what is available in manuals. > Hopefully Intel can take a look. Oh I missed the original mail for some reason. Yes it should be cmask instead of thresh for client. I think thresh is used on the server uncore only, not on the client. Jin Yao, can you send a patch please? -Andi
On 10/19/2020 9:52 AM, Andi Kleen wrote: >> I believe this is a Skylake and there is a known bug in the Skylake >> metric DRAM_Parallel_Reads as described here: >> https://lore.kernel.org/lkml/CAP-5=fXejVaQa9qfW66cY77qB962+jbe8tT5bsLoOOcFmODnWQ@mail.gmail.com/ >> Fixing the bug needs more knowledge than what is available in manuals. >> Hopefully Intel can take a look. > > Oh I missed the original mail for some reason. Yes it should be cmask instead of thresh > for client. I think thresh is used on the server uncore only, not on the client. > > Jin Yao, can you send a patch please? > > -Andi > Yes, the DRAM_Parallel_Reads works on server but it's failed on client. I will post a patch to fix that. Thanks Jin Yao
On 19/10/2020 00:30, Ian Rogers wrote: > On Sun, Oct 18, 2020 at 1:51 AM kernel test robot <rong.a.chen@intel.com> wrote: >> >> Greeting, >> >> FYI, we noticed the following commit (built with gcc-9): >> >> commit: fcc9c5243c478f104014daf4d23db86098d2aef0 ("perf metricgroup: Hack a fix for aliases when covering multiple PMUs") >> url: https://github.com/0day-ci/linux/commits/John-Garry/perf-pmu-events-Support-event-aliasing-for-system-PMUs/20201008-182049 >> >> >> in testcase: perf-sanity-tests >> version: perf-x86_64-c85fb28b6f99-1_20201008 >> with following parameters: >> >> perf_compiler: gcc >> ucode: 0xdc >> >> >> >> on test machine: 4 threads Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz with 32G memory >> >> caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace): > > I believe this is a Skylake and there is a known bug in the Skylake > metric DRAM_Parallel_Reads as described here: > https://lore.kernel.org/lkml/CAP-5=fXejVaQa9qfW66cY77qB962+jbe8tT5bsLoOOcFmODnWQ@mail.gmail.com/ > Fixing the bug needs more knowledge than what is available in manuals. > Hopefully Intel can take a look. > > Thanks, > Ian So this named patch ("perf metricgroup: Hack a fix for aliases...") is breaking test #67 on my machine also, which is a broadwell. I will have a look, but I was hoping that Ian would have a proper fix for this on top of ("perf metricgroup: Fix uncore metric expressions"), which now looks to be merged. Thanks! > >> >> >> If you fix the issue, kindly add following tag >> Reported-by: kernel test robot <rong.a.chen@intel.com> >> >> >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 67 >> 67: Parse and process metrics : FAILED! >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 68 >> 68: x86 rdpmc : Ok >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 69 >> 69: Convert perf time to TSC : Ok >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 70 >> 70: DWARF unwind : Ok >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 71 >> 71: x86 instruction decoder - new instructions : Ok >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 72 >> 72: Intel PT packet decoder : Ok >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 73 >> 73: x86 bp modify : Ok >> 2020-10-16 19:31:53 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 74 >> 74: probe libc's inet_pton & backtrace it with ping : Ok >> 2020-10-16 19:31:54 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 75 >> 75: Zstd perf.data compression/decompression : Ok >> >> >> >> To reproduce: >> >> git clone https://github.com/intel/lkp-tests.git >> cd lkp-tests >> bin/lkp install job.yaml # job file is attached in this email >> bin/lkp run job.yaml >> >> >> >> Thanks, >> Rong Chen >> > . >
Hi Garry, Hi Ian, On 10/19/2020 5:48 PM, John Garry wrote: > On 19/10/2020 00:30, Ian Rogers wrote: >> On Sun, Oct 18, 2020 at 1:51 AM kernel test robot <rong.a.chen@intel.com> wrote: >>> >>> Greeting, >>> >>> FYI, we noticed the following commit (built with gcc-9): >>> >>> commit: fcc9c5243c478f104014daf4d23db86098d2aef0 ("perf metricgroup: Hack a fix for aliases when >>> covering multiple PMUs") >>> url: >>> https://github.com/0day-ci/linux/commits/John-Garry/perf-pmu-events-Support-event-aliasing-for-system-PMUs/20201008-182049 >>> >>> >>> >>> in testcase: perf-sanity-tests >>> version: perf-x86_64-c85fb28b6f99-1_20201008 >>> with following parameters: >>> >>> perf_compiler: gcc >>> ucode: 0xdc >>> >>> >>> >>> on test machine: 4 threads Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz with 32G memory >>> >>> caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace): >> >> I believe this is a Skylake and there is a known bug in the Skylake >> metric DRAM_Parallel_Reads as described here: >> https://lore.kernel.org/lkml/CAP-5=fXejVaQa9qfW66cY77qB962+jbe8tT5bsLoOOcFmODnWQ@mail.gmail.com/ >> Fixing the bug needs more knowledge than what is available in manuals. >> Hopefully Intel can take a look. >> >> Thanks, >> Ian > > So this named patch ("perf metricgroup: Hack a fix for aliases...") is breaking test #67 on my > machine also, which is a broadwell. > > I will have a look, but I was hoping that Ian would have a proper fix for this on top of ("perf > metricgroup: Fix uncore metric expressions"), which now looks to be merged. > > Thanks! > I just think they are different issues. On my KBL client, the perf test #67 is passed. But DRAM_Parallel_Reads does have issue. root@kbl-ppc:~# perf stat -M DRAM_Parallel_Reads -- sleep 1 event syntax error: '{arb/event=0x80,umask=0x2/,arb/event=0x80,umask=0x2,thresh=1/}:W' \___ unknown term 'thresh' for pmu 'uncore_arb' valid terms: event,edge,inv,umask,cmask,config,config1,config2,name,period,percore Initial error: event syntax error: '..umask=0x2/,arb/event=0x80,umask=0x2,thresh=1/}:W' \___ Cannot find PMU `arb'. Missing kernel support? Usage: perf stat [<options>] [<command>] -M, --metrics <metric/metric group list> monitor specified metrics or metric groups (separated by ,) I have a patch to fix DRAM_Parallel_Reads. After: root@kbl-ppc:~# perf stat -M MEM_Parallel_Reads -- sleep 1 Performance counter stats for 'system wide': 3,043,952 arb/event=0x80,umask=0x2/ # 1.00 MEM_Parallel_Reads 1.000879932 seconds time elapsed I will post the patch later. Thanks Jin Yao >> >>> >>> >>> If you fix the issue, kindly add following tag >>> Reported-by: kernel test robot <rong.a.chen@intel.com> >>> >>> >>> 2020-10-16 19:31:52 sudo >>> /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf >>> test 67 >>> 67: Parse and process metrics : FAILED! >>> 2020-10-16 19:31:52 sudo >>> /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf >>> test 68 >>> 68: x86 rdpmc : Ok >>> 2020-10-16 19:31:52 sudo >>> /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf >>> test 69 >>> 69: Convert perf time to TSC : Ok >>> 2020-10-16 19:31:52 sudo >>> /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf >>> test 70 >>> 70: DWARF unwind : Ok >>> 2020-10-16 19:31:52 sudo >>> /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf >>> test 71 >>> 71: x86 instruction decoder - new instructions : Ok >>> 2020-10-16 19:31:52 sudo >>> /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf >>> test 72 >>> 72: Intel PT packet decoder : Ok >>> 2020-10-16 19:31:52 sudo >>> /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf >>> test 73 >>> 73: x86 bp modify : Ok >>> 2020-10-16 19:31:53 sudo >>> /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf >>> test 74 >>> 74: probe libc's inet_pton & backtrace it with ping : Ok >>> 2020-10-16 19:31:54 sudo >>> /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf >>> test 75 >>> 75: Zstd perf.data compression/decompression : Ok >>> >>> >>> >>> To reproduce: >>> >>> git clone https://github.com/intel/lkp-tests.git >>> cd lkp-tests >>> bin/lkp install job.yaml # job file is attached in this email >>> bin/lkp run job.yaml >>> >>> >>> >>> Thanks, >>> Rong Chen >>> >> . >> >
On Mon, Oct 19, 2020 at 2:51 AM John Garry <john.garry@huawei.com> wrote: > > On 19/10/2020 00:30, Ian Rogers wrote: > > On Sun, Oct 18, 2020 at 1:51 AM kernel test robot <rong.a.chen@intel.com> wrote: > >> > >> Greeting, > >> > >> FYI, we noticed the following commit (built with gcc-9): > >> > >> commit: fcc9c5243c478f104014daf4d23db86098d2aef0 ("perf metricgroup: Hack a fix for aliases when covering multiple PMUs") > >> url: https://github.com/0day-ci/linux/commits/John-Garry/perf-pmu-events-Support-event-aliasing-for-system-PMUs/20201008-182049 > >> > >> > >> in testcase: perf-sanity-tests > >> version: perf-x86_64-c85fb28b6f99-1_20201008 > >> with following parameters: > >> > >> perf_compiler: gcc > >> ucode: 0xdc > >> > >> > >> > >> on test machine: 4 threads Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz with 32G memory > >> > >> caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace): > > > > I believe this is a Skylake and there is a known bug in the Skylake > > metric DRAM_Parallel_Reads as described here: > > https://lore.kernel.org/lkml/CAP-5=fXejVaQa9qfW66cY77qB962+jbe8tT5bsLoOOcFmODnWQ@mail.gmail.com/ > > Fixing the bug needs more knowledge than what is available in manuals. > > Hopefully Intel can take a look. > > > > Thanks, > > Ian > > So this named patch ("perf metricgroup: Hack a fix for aliases...") is > breaking test #67 on my machine also, which is a broadwell. Thanks for taking a look John. If you want help you can send the output of "perf test 67 -vvv" to me. It is possible Broadwell has similar glitches in the json to Skylake. I tested the original test on server parts as I can access them as cloud machines. > I will have a look, but I was hoping that Ian would have a proper fix > for this on top of ("perf metricgroup: Fix uncore metric expressions"), > which now looks to be merged. I still have these changes to look at in my inbox but I'm assuming they're good :-) Sorry for not getting to them, but it's good they are merged. Thanks, Ian > Thanks! > > > > >> > >> > >> If you fix the issue, kindly add following tag > >> Reported-by: kernel test robot <rong.a.chen@intel.com> > >> > >> > >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 67 > >> 67: Parse and process metrics : FAILED! > >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 68 > >> 68: x86 rdpmc : Ok > >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 69 > >> 69: Convert perf time to TSC : Ok > >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 70 > >> 70: DWARF unwind : Ok > >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 71 > >> 71: x86 instruction decoder - new instructions : Ok > >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 72 > >> 72: Intel PT packet decoder : Ok > >> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 73 > >> 73: x86 bp modify : Ok > >> 2020-10-16 19:31:53 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 74 > >> 74: probe libc's inet_pton & backtrace it with ping : Ok > >> 2020-10-16 19:31:54 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 75 > >> 75: Zstd perf.data compression/decompression : Ok > >> > >> > >> > >> To reproduce: > >> > >> git clone https://github.com/intel/lkp-tests.git > >> cd lkp-tests > >> bin/lkp install job.yaml # job file is attached in this email > >> bin/lkp run job.yaml > >> > >> > >> > >> Thanks, > >> Rong Chen > >> > > . > > >
On 19/10/2020 17:20, Ian Rogers wrote: >>> n >> So this named patch ("perf metricgroup: Hack a fix for aliases...") is >> breaking test #67 on my machine also, which is a broadwell. > Thanks for taking a look John. If you want help you can send the > output of "perf test 67 -vvv" to me. It is possible Broadwell has > similar glitches in the json to Skylake. I tested the original test on > server parts as I can access them as cloud machines. Here it is: john@localhost:~/kernel-dev7/tools/perf> ./perf test -vv 67 Couldn't bump rlimit(MEMLOCK), failures may take place when creating BPF maps, etc 67: Parse and process metrics : --- start --- test child forked, pid 24433 metric expr inst_retired.any / cpu_clk_unhalted.thread for IPC parsing metric: inst_retired.any / cpu_clk_unhalted.thread found event inst_retired.any found event cpu_clk_unhalted.thread adding {inst_retired.any,cpu_clk_unhalted.thread}:W Attempting to add event pmu 'inst_retired.any' with '' that may result in non-fatal errors Attempting to add event pmu 'cpu_clk_unhalted.thread' with '' that may result in non-fatal errors parsing metric: inst_retired.any / cpu_clk_unhalted.thread lookup: is_ref 0, counted 0, val 300.000000: inst_retired.any lookup: is_ref 0, counted 101, val 200.000000: cpu_clk_unhalted.thread metric expr idq_uops_not_delivered.core / (4 * (( ( cpu_clk_unhalted.thread / 2 ) * ( 1 + cpu_clk_unhalted.one_thread_active / cpu_clk_unhalted.ref_xclk ) ))) for Frontend_Bound_SMT parsing metric: idq_uops_not_delivered.core / (4 * (( ( cpu_clk_unhalted.thread / 2 ) * ( 1 + cpu_clk_unhalted.one_thread_active / cpu_clk_unhalted.ref_xclk ) ))) found event cpu_clk_unhalted.one_thread_active found event cpu_clk_unhalted.ref_xclk found event idq_uops_not_delivered.core found event cpu_clk_unhalted.thread adding {cpu_clk_unhalted.one_thread_active,cpu_clk_unhalted.ref_xclk,idq_uops_not_delivered.core,cpu_clk_unhalted.thread}:W Attempting to add event pmu 'cpu_clk_unhalted.one_thread_active' with '' that may result in non-fatal errors Attempting to add event pmu 'cpu_clk_unhalted.ref_xclk' with '' that may result in non-fatal errors Attempting to add event pmu 'idq_uops_not_delivered.core' with '' that may result in non-fatal errors Attempting to add event pmu 'cpu_clk_unhalted.thread' with '' that may result in non-fatal errors parsing metric: idq_uops_not_delivered.core / (4 * (( ( cpu_clk_unhalted.thread / 2 ) * ( 1 + cpu_clk_unhalted.one_thread_active / cpu_clk_unhalted.ref_xclk ) ))) lookup: is_ref 0, counted 46, val 300.000000: idq_uops_not_delivered.core lookup: is_ref 0, counted 0, val 200.000000: cpu_clk_unhalted.thread lookup: is_ref 0, counted 216, val 400.000000: cpu_clk_unhalted.one_thread_active lookup: is_ref 0, counted 46, val 600.000000: cpu_clk_unhalted.ref_xclk metric expr (dcache_miss_cpi + icache_miss_cycles) for cache_miss_cycles parsing metric: (dcache_miss_cpi + icache_miss_cycles) metric expr l1d\-loads\-misses / inst_retired.any for dcache_miss_cpi parsing metric: l1d\-loads\-misses / inst_retired.any metric expr l1i\-loads\-misses / inst_retired.any for icache_miss_cycles parsing metric: l1i\-loads\-misses / inst_retired.any found event inst_retired.any found event l1i-loads-misses found event l1d-loads-misses adding {inst_retired.any,l1i-loads-misses,l1d-loads-misses}:W Attempting to add event pmu 'inst_retired.any' with '' that may result in non-fatal errors adding ref metric icache_miss_cycles: l1i\-loads\-misses / inst_retired.any adding ref metric dcache_miss_cpi: l1d\-loads\-misses / inst_retired.any parsing metric: (dcache_miss_cpi + icache_miss_cycles) lookup: is_ref 1, counted 0, val 0.000000: dcache_miss_cpi processing metric: dcache_miss_cpi ENTRY parsing metric: l1d\-loads\-misses / inst_retired.any lookup: is_ref 0, counted 105, val 300.000000: l1d-loads-misses lookup: is_ref 0, counted 46, val 400.000000: inst_retired.any processing metric: dcache_miss_cpi EXIT: 0.750000 lookup: is_ref 1, counted 0, val 0.000000: icache_miss_cycles processing metric: icache_miss_cycles ENTRY parsing metric: l1i\-loads\-misses / inst_retired.any lookup: is_ref 0, counted 216, val 200.000000: l1i-loads-misses lookup: is_ref 0, counted 46, val 400.000000: inst_retired.any processing metric: icache_miss_cycles EXIT: 0.500000 metric expr d_ratio(dcache_l2_all_hits, dcache_l2_all) for DCache_L2_Hits parsing metric: d_ratio(dcache_l2_all_hits, dcache_l2_all) metric expr l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit for DCache_L2_All_Hits parsing metric: l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit metric expr dcache_l2_all_hits + dcache_l2_all_miss for DCache_L2_All parsing metric: dcache_l2_all_hits + dcache_l2_all_miss metric expr l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit for DCache_L2_All_Hits parsing metric: l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit metric expr max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss for DCache_L2_All_Miss parsing metric: max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss found event l2_rqsts.rfo_miss found event l2_rqsts.pf_miss found event l2_rqsts.rfo_hit found event l2_rqsts.demand_data_rd_hit found event l2_rqsts.all_demand_data_rd found event l2_rqsts.pf_hit adding {l2_rqsts.rfo_miss,l2_rqsts.pf_miss,l2_rqsts.rfo_hit,l2_rqsts.demand_data_rd_hit,l2_rqsts.all_demand_data_rd,l2_rqsts.pf_hit}:W Attempting to add event pmu 'l2_rqsts.rfo_miss' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.pf_miss' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.rfo_hit' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.demand_data_rd_hit' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.all_demand_data_rd' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.pf_hit' with '' that may result in non-fatal errors adding ref metric DCache_L2_All_Miss: max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss adding ref metric DCache_L2_All_Hits: l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit adding ref metric DCache_L2_All: dcache_l2_all_hits + dcache_l2_all_miss adding ref metric DCache_L2_All_Hits: l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit parsing metric: d_ratio(dcache_l2_all_hits, dcache_l2_all) lookup: is_ref 1, counted 0, val 0.000000: dcache_l2_all_hits processing metric: dcache_l2_all_hits ENTRY parsing metric: l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit lookup: is_ref 0, counted 8, val 100.000000: l2_rqsts.demand_data_rd_hit lookup: is_ref 0, counted 224, val 200.000000: l2_rqsts.pf_hit lookup: is_ref 0, counted 46, val 300.000000: l2_rqsts.rfo_hit processing metric: dcache_l2_all_hits EXIT: 600.000000 lookup: is_ref 1, counted 0, val 0.000000: dcache_l2_all processing metric: dcache_l2_all ENTRY parsing metric: dcache_l2_all_hits + dcache_l2_all_miss lookup: is_ref 1, counted 1, val 600.000000: dcache_l2_all_hits lookup: is_ref 1, counted 0, val 0.000000: dcache_l2_all_miss processing metric: dcache_l2_all_miss ENTRY parsing metric: max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss lookup: is_ref 0, counted 112, val 400.000000: l2_rqsts.all_demand_data_rd lookup: is_ref 0, counted 8, val 100.000000: l2_rqsts.demand_data_rd_hit lookup: is_ref 0, counted 100, val 500.000000: l2_rqsts.pf_miss lookup: is_ref 0, counted 120, val 600.000000: l2_rqsts.rfo_miss processing metric: dcache_l2_all_miss EXIT: 1400.000000 processing metric: dcache_l2_all EXIT: 2000.000000 metric expr d_ratio(dcache_l2_all_miss, dcache_l2_all) for DCache_L2_Misses parsing metric: d_ratio(dcache_l2_all_miss, dcache_l2_all) metric expr max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss for DCache_L2_All_Miss parsing metric: max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss metric expr dcache_l2_all_hits + dcache_l2_all_miss for DCache_L2_All parsing metric: dcache_l2_all_hits + dcache_l2_all_miss metric expr l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit for DCache_L2_All_Hits parsing metric: l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit metric expr max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss for DCache_L2_All_Miss parsing metric: max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss found event l2_rqsts.rfo_miss found event l2_rqsts.pf_miss found event l2_rqsts.rfo_hit found event l2_rqsts.demand_data_rd_hit found event l2_rqsts.all_demand_data_rd found event l2_rqsts.pf_hit adding {l2_rqsts.rfo_miss,l2_rqsts.pf_miss,l2_rqsts.rfo_hit,l2_rqsts.demand_data_rd_hit,l2_rqsts.all_demand_data_rd,l2_rqsts.pf_hit}:W Attempting to add event pmu 'l2_rqsts.rfo_miss' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.pf_miss' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.rfo_hit' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.demand_data_rd_hit' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.all_demand_data_rd' with '' that may result in non-fatal errors Attempting to add event pmu 'l2_rqsts.pf_hit' with '' that may result in non-fatal errors adding ref metric DCache_L2_All_Miss: max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss adding ref metric DCache_L2_All_Hits: l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit adding ref metric DCache_L2_All: dcache_l2_all_hits + dcache_l2_all_miss adding ref metric DCache_L2_All_Miss: max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss parsing metric: d_ratio(dcache_l2_all_miss, dcache_l2_all) lookup: is_ref 1, counted 0, val 0.000000: dcache_l2_all_miss processing metric: dcache_l2_all_miss ENTRY parsing metric: max(l2_rqsts.all_demand_data_rd - l2_rqsts.demand_data_rd_hit, 0) + l2_rqsts.pf_miss + l2_rqsts.rfo_miss lookup: is_ref 0, counted 128, val 400.000000: l2_rqsts.all_demand_data_rd lookup: is_ref 0, counted 224, val 100.000000: l2_rqsts.demand_data_rd_hit lookup: is_ref 0, counted 112, val 500.000000: l2_rqsts.pf_miss lookup: is_ref 0, counted 0, val 600.000000: l2_rqsts.rfo_miss processing metric: dcache_l2_all_miss EXIT: 1400.000000 lookup: is_ref 1, counted 0, val 0.000000: dcache_l2_all processing metric: dcache_l2_all ENTRY parsing metric: dcache_l2_all_hits + dcache_l2_all_miss lookup: is_ref 1, counted 0, val 0.000000: dcache_l2_all_hits processing metric: dcache_l2_all_hits ENTRY parsing metric: l2_rqsts.demand_data_rd_hit + l2_rqsts.pf_hit + l2_rqsts.rfo_hit lookup: is_ref 0, counted 224, val 100.000000: l2_rqsts.demand_data_rd_hit lookup: is_ref 0, counted 0, val 200.000000: l2_rqsts.pf_hit lookup: is_ref 0, counted 152, val 300.000000: l2_rqsts.rfo_hit processing metric: dcache_l2_all_hits EXIT: 600.000000 lookup: is_ref 1, counted 1, val 1400.000000: dcache_l2_all_miss processing metric: dcache_l2_all EXIT: 2000.000000 metric expr ipc + m2 for M1 parsing metric: ipc + m2 metric expr ipc + m1 for M2 parsing metric: ipc + m1 metric expr ipc + m2 for M1 parsing metric: ipc + m2 failed: recursion detected for m2 metric expr 1/m3 for M3 parsing metric: 1/m3 metric expr 1/m3 for M3 parsing metric: 1/m3 failed: recursion detected for m3 metric expr inst_retired.any / cpu_clk_unhalted.thread for IPC parsing metric: inst_retired.any / cpu_clk_unhalted.thread metric expr (dcache_miss_cpi + icache_miss_cycles) for cache_miss_cycles parsing metric: (dcache_miss_cpi + icache_miss_cycles) metric expr l1d\-loads\-misses / inst_retired.any for dcache_miss_cpi parsing metric: l1d\-loads\-misses / inst_retired.any metric expr l1i\-loads\-misses / inst_retired.any for icache_miss_cycles parsing metric: l1i\-loads\-misses / inst_retired.any found event inst_retired.any found event cpu_clk_unhalted.thread found event inst_retired.any found event l1i-loads-misses found event l1d-loads-misses adding {inst_retired.any,cpu_clk_unhalted.thread}:W,{inst_retired.any,l1i-loads-misses,l1d-loads-misses}:W Attempting to add event pmu 'inst_retired.any' with '' that may result in non-fatal errors Attempting to add event pmu 'cpu_clk_unhalted.thread' with '' that may result in non-fatal errors Attempting to add event pmu 'inst_retired.any' with '' that may result in non-fatal errors parsing metric: inst_retired.any / cpu_clk_unhalted.thread lookup: is_ref 0, counted 64, val 800.000000: inst_retired.any lookup: is_ref 0, counted 32, val 200.000000: cpu_clk_unhalted.thread adding ref metric icache_miss_cycles: l1i\-loads\-misses / inst_retired.any adding ref metric dcache_miss_cpi: l1d\-loads\-misses / inst_retired.any parsing metric: (dcache_miss_cpi + icache_miss_cycles) lookup: is_ref 1, counted 0, val 0.000000: dcache_miss_cpi processing metric: dcache_miss_cpi ENTRY parsing metric: l1d\-loads\-misses / inst_retired.any lookup: is_ref 0, counted 0, val 300.000000: l1d-loads-misses lookup: is_ref 0, counted 64, val 800.000000: inst_retired.any processing metric: dcache_miss_cpi EXIT: 0.375000 lookup: is_ref 1, counted 0, val 0.000000: icache_miss_cycles processing metric: icache_miss_cycles ENTRY parsing metric: l1i\-loads\-misses / inst_retired.any lookup: is_ref 0, counted 32, val 200.000000: l1i-loads-misses lookup: is_ref 0, counted 64, val 800.000000: inst_retired.any processing metric: icache_miss_cycles EXIT: 0.250000 FAILED tests/parse-metric.c:343 group IPC failed, wrong ratio FAILED tests/parse-metric.c:357 test metric group test child finished with -1 ---- end ---- Parse and process metrics: FAILED! john@localhost:~/kernel-dev7/tools/perf> > >> I will have a look, but I was hoping that Ian would have a proper fix >> for this on top of ("perf metricgroup: Fix uncore metric expressions"), >> which now looks to be merged. > I still have these changes to look at in my inbox but I'm assuming > they're good:-) really? I wouldn't assume my hacky patch is good :) > Sorry for not getting to them, but it's good they are > merged. cheers
On 10/19/20 9:50 PM, Ian Rogers wrote: > On Mon, Oct 19, 2020 at 2:51 AM John Garry <john.garry@huawei.com> wrote: >> >> On 19/10/2020 00:30, Ian Rogers wrote: >>> On Sun, Oct 18, 2020 at 1:51 AM kernel test robot <rong.a.chen@intel.com> wrote: >>>> >>>> Greeting, >>>> >>>> FYI, we noticed the following commit (built with gcc-9): >>>> >>>> commit: fcc9c5243c478f104014daf4d23db86098d2aef0 ("perf metricgroup: Hack a fix for aliases when covering multiple PMUs") >>>> url: https://github.com/0day-ci/linux/commits/John-Garry/perf-pmu-events-Support-event-aliasing-for-system-PMUs/20201008-182049 >>>> >>>> >>>> in testcase: perf-sanity-tests >>>> version: perf-x86_64-c85fb28b6f99-1_20201008 >>>> with following parameters: >>>> >>>> perf_compiler: gcc >>>> ucode: 0xdc >>>> >>>> >>>> >>>> on test machine: 4 threads Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz with 32G memory >>>> >>>> caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace): >>> >>> I believe this is a Skylake and there is a known bug in the Skylake >>> metric DRAM_Parallel_Reads as described here: >>> https://lore.kernel.org/lkml/CAP-5=fXejVaQa9qfW66cY77qB962+jbe8tT5bsLoOOcFmODnWQ@mail.gmail.com/ >>> Fixing the bug needs more knowledge than what is available in manuals. >>> Hopefully Intel can take a look. >>> >>> Thanks, >>> Ian >> >> So this named patch ("perf metricgroup: Hack a fix for aliases...") is >> breaking test #67 on my machine also, which is a broadwell. > > Thanks for taking a look John. If you want help you can send the > output of "perf test 67 -vvv" to me. It is possible Broadwell has > similar glitches in the json to Skylake. I tested the original test on > server parts as I can access them as cloud machines. > >> I will have a look, but I was hoping that Ian would have a proper fix >> for this on top of ("perf metricgroup: Fix uncore metric expressions"), >> which now looks to be merged. > > I still have these changes to look at in my inbox but I'm assuming > they're good :-) Sorry for not getting to them, but it's good they are > merged. Hi Ian, Checked in upstream kernel with your fix patch, in powerpc also test case 67 is passing. But I am getting issue in test 10 for powerpc [command]# ./perf test 10 10: PMU events : 10.1: PMU event table sanity : Ok 10.2: PMU event map aliases : Ok 10.3: Parsing of PMU event table metrics : Skip (some metrics failed) 10.4: Parsing of PMU event table metrics with fake PMUs : FAILED! Was debugging it, issue is with commit e1c92a7fbbc5 perf tests: Add another metric parsing test. So, there we are passing different runtime parameter value in "expr__find_other and expr__parse" in function `metric_parse_fake`. I believe we need to send same value. I will send fix patch for the same. Thanks, Kajol Jain > > Thanks, > Ian > >> Thanks! >> >>> >>>> >>>> >>>> If you fix the issue, kindly add following tag >>>> Reported-by: kernel test robot <rong.a.chen@intel.com> >>>> >>>> >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 67 >>>> 67: Parse and process metrics : FAILED! >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 68 >>>> 68: x86 rdpmc : Ok >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 69 >>>> 69: Convert perf time to TSC : Ok >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 70 >>>> 70: DWARF unwind : Ok >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 71 >>>> 71: x86 instruction decoder - new instructions : Ok >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 72 >>>> 72: Intel PT packet decoder : Ok >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 73 >>>> 73: x86 bp modify : Ok >>>> 2020-10-16 19:31:53 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 74 >>>> 74: probe libc's inet_pton & backtrace it with ping : Ok >>>> 2020-10-16 19:31:54 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 75 >>>> 75: Zstd perf.data compression/decompression : Ok >>>> >>>> >>>> >>>> To reproduce: >>>> >>>> git clone https://github.com/intel/lkp-tests.git >>>> cd lkp-tests >>>> bin/lkp install job.yaml # job file is attached in this email >>>> bin/lkp run job.yaml >>>> >>>> >>>> >>>> Thanks, >>>> Rong Chen >>>> >>> . >>> >>
On Tue, Oct 20, 2020 at 1:56 AM kajoljain <kjain@linux.ibm.com> wrote: > > > > On 10/19/20 9:50 PM, Ian Rogers wrote: > > On Mon, Oct 19, 2020 at 2:51 AM John Garry <john.garry@huawei.com> wrote: > >> > >> On 19/10/2020 00:30, Ian Rogers wrote: > >>> On Sun, Oct 18, 2020 at 1:51 AM kernel test robot <rong.a.chen@intel.com> wrote: > >>>> > >>>> Greeting, > >>>> > >>>> FYI, we noticed the following commit (built with gcc-9): > >>>> > >>>> commit: fcc9c5243c478f104014daf4d23db86098d2aef0 ("perf metricgroup: Hack a fix for aliases when covering multiple PMUs") > >>>> url: https://github.com/0day-ci/linux/commits/John-Garry/perf-pmu-events-Support-event-aliasing-for-system-PMUs/20201008-182049 > >>>> > >>>> > >>>> in testcase: perf-sanity-tests > >>>> version: perf-x86_64-c85fb28b6f99-1_20201008 > >>>> with following parameters: > >>>> > >>>> perf_compiler: gcc > >>>> ucode: 0xdc > >>>> > >>>> > >>>> > >>>> on test machine: 4 threads Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz with 32G memory > >>>> > >>>> caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace): > >>> > >>> I believe this is a Skylake and there is a known bug in the Skylake > >>> metric DRAM_Parallel_Reads as described here: > >>> https://lore.kernel.org/lkml/CAP-5=fXejVaQa9qfW66cY77qB962+jbe8tT5bsLoOOcFmODnWQ@mail.gmail.com/ > >>> Fixing the bug needs more knowledge than what is available in manuals. > >>> Hopefully Intel can take a look. > >>> > >>> Thanks, > >>> Ian > >> > >> So this named patch ("perf metricgroup: Hack a fix for aliases...") is > >> breaking test #67 on my machine also, which is a broadwell. > > > > Thanks for taking a look John. If you want help you can send the > > output of "perf test 67 -vvv" to me. It is possible Broadwell has > > similar glitches in the json to Skylake. I tested the original test on > > server parts as I can access them as cloud machines. > > > >> I will have a look, but I was hoping that Ian would have a proper fix > >> for this on top of ("perf metricgroup: Fix uncore metric expressions"), > >> which now looks to be merged. > > > > I still have these changes to look at in my inbox but I'm assuming > > they're good :-) Sorry for not getting to them, but it's good they are > > merged. > > Hi Ian, > Checked in upstream kernel with your fix patch, in powerpc also test case 67 is passing. > But I am getting issue in test 10 for powerpc > > [command]# ./perf test 10 > 10: PMU events : > 10.1: PMU event table sanity : Ok > 10.2: PMU event map aliases : Ok > 10.3: Parsing of PMU event table metrics : Skip (some metrics failed) > 10.4: Parsing of PMU event table metrics with fake PMUs : FAILED! > > Was debugging it, issue is with commit e1c92a7fbbc5 perf tests: Add another metric parsing test. > > So, there we are passing different runtime parameter value in "expr__find_other and expr__parse" > in function `metric_parse_fake`. I believe we need to send same value. > I will send fix patch for the same. > > Thanks, > Kajol Jain Thanks, the fake support was done by Jiri. I do try to test on Power 8. The awesome thing, aside from the testing nit fixes, is that the metrics will actually work once the test is passing :-). They may of course report junk. Thanks, Ian > > > > Thanks, > > Ian > > > >> Thanks! > >> > >>> > >>>> > >>>> > >>>> If you fix the issue, kindly add following tag > >>>> Reported-by: kernel test robot <rong.a.chen@intel.com> > >>>> > >>>> > >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 67 > >>>> 67: Parse and process metrics : FAILED! > >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 68 > >>>> 68: x86 rdpmc : Ok > >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 69 > >>>> 69: Convert perf time to TSC : Ok > >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 70 > >>>> 70: DWARF unwind : Ok > >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 71 > >>>> 71: x86 instruction decoder - new instructions : Ok > >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 72 > >>>> 72: Intel PT packet decoder : Ok > >>>> 2020-10-16 19:31:52 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 73 > >>>> 73: x86 bp modify : Ok > >>>> 2020-10-16 19:31:53 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 74 > >>>> 74: probe libc's inet_pton & backtrace it with ping : Ok > >>>> 2020-10-16 19:31:54 sudo /usr/src/perf_selftests-x86_64-rhel-8.3-fcc9c5243c478f104014daf4d23db86098d2aef0/tools/perf/perf test 75 > >>>> 75: Zstd perf.data compression/decompression : Ok > >>>> > >>>> > >>>> > >>>> To reproduce: > >>>> > >>>> git clone https://github.com/intel/lkp-tests.git > >>>> cd lkp-tests > >>>> bin/lkp install job.yaml # job file is attached in this email > >>>> bin/lkp run job.yaml > >>>> > >>>> > >>>> > >>>> Thanks, > >>>> Rong Chen > >>>> > >>> . > >>> > >>
On 20/10/2020 17:53, Ian Rogers wrote: >>> Thanks for taking a look John. If you want help you can send the >>> output of "perf test 67 -vvv" to me. It is possible Broadwell has >>> similar glitches in the json to Skylake. I tested the original test on >>> server parts as I can access them as cloud machines. >>> >>>> I will have a look, but I was hoping that Ian would have a proper fix >>>> for this on top of ("perf metricgroup: Fix uncore metric expressions"), >>>> which now looks to be merged. >>> I still have these changes to look at in my inbox but I'm assuming >>> they're good:-) Sorry for not getting to them, but it's good they are >>> merged. >> Hi Ian, >> Checked in upstream kernel with your fix patch, in powerpc also test case 67 is passing. >> But I am getting issue in test 10 for powerpc >> >> [command]# ./perf test 10 >> 10: PMU events : >> 10.1: PMU event table sanity : Ok >> 10.2: PMU event map aliases : Ok >> 10.3: Parsing of PMU event table metrics : Skip (some metrics failed) >> 10.4: Parsing of PMU event table metrics with fake PMUs : FAILED! >> >> Was debugging it, issue is with commit e1c92a7fbbc5 perf tests: Add another metric parsing test. >> >> So, there we are passing different runtime parameter value in "expr__find_other and expr__parse" >> in function `metric_parse_fake`. I believe we need to send same value. >> I will send fix patch for the same. Just wondering, was a patch ever submitted for this? Something still broken? I can't see any recent relevant changes to tests/pmu-events.c Thanks, John
On Tue, Nov 3, 2020 at 6:43 AM John Garry <john.garry@huawei.com> wrote: > > On 20/10/2020 17:53, Ian Rogers wrote: > >>> Thanks for taking a look John. If you want help you can send the > >>> output of "perf test 67 -vvv" to me. It is possible Broadwell has > >>> similar glitches in the json to Skylake. I tested the original test on > >>> server parts as I can access them as cloud machines. > >>> > >>>> I will have a look, but I was hoping that Ian would have a proper fix > >>>> for this on top of ("perf metricgroup: Fix uncore metric expressions"), > >>>> which now looks to be merged. > >>> I still have these changes to look at in my inbox but I'm assuming > >>> they're good:-) Sorry for not getting to them, but it's good they are > >>> merged. > >> Hi Ian, > >> Checked in upstream kernel with your fix patch, in powerpc also test case 67 is passing. > >> But I am getting issue in test 10 for powerpc > >> > >> [command]# ./perf test 10 > >> 10: PMU events : > >> 10.1: PMU event table sanity : Ok > >> 10.2: PMU event map aliases : Ok > >> 10.3: Parsing of PMU event table metrics : Skip (some metrics failed) > >> 10.4: Parsing of PMU event table metrics with fake PMUs : FAILED! > >> > >> Was debugging it, issue is with commit e1c92a7fbbc5 perf tests: Add another metric parsing test. > >> > >> So, there we are passing different runtime parameter value in "expr__find_other and expr__parse" > >> in function `metric_parse_fake`. I believe we need to send same value. > >> I will send fix patch for the same. > > Just wondering, was a patch ever submitted for this? Something still > broken? I can't see any recent relevant changes to tests/pmu-events.c The test itself shouldn't have changed, but the json files parsed by jevents and turned into C code that the test exercises should have changed. Jin Yao has sent two patch sets fixing a metric issue on SKL (Skylake non-server) that should hopefully fix the issue there - I'll check the status on these. Are you testing on Skylake? Thanks, Ian > Thanks, > John
On 03/11/2020 16:05, Ian Rogers wrote: > On Tue, Nov 3, 2020 at 6:43 AM John Garry<john.garry@huawei.com> wrote: >> On 20/10/2020 17:53, Ian Rogers wrote: >>>>> Thanks for taking a look John. If you want help you can send the >>>>> output of "perf test 67 -vvv" to me. It is possible Broadwell has >>>>> similar glitches in the json to Skylake. I tested the original test on >>>>> server parts as I can access them as cloud machines. >>>>> >>>>>> I will have a look, but I was hoping that Ian would have a proper fix >>>>>> for this on top of ("perf metricgroup: Fix uncore metric expressions"), >>>>>> which now looks to be merged. >>>>> I still have these changes to look at in my inbox but I'm assuming >>>>> they're good:-) Sorry for not getting to them, but it's good they are >>>>> merged. >>>> Hi Ian, >>>> Checked in upstream kernel with your fix patch, in powerpc also test case 67 is passing. >>>> But I am getting issue in test 10 for powerpc >>>> >>>> [command]# ./perf test 10 >>>> 10: PMU events : >>>> 10.1: PMU event table sanity : Ok >>>> 10.2: PMU event map aliases : Ok >>>> 10.3: Parsing of PMU event table metrics : Skip (some metrics failed) >>>> 10.4: Parsing of PMU event table metrics with fake PMUs : FAILED! >>>> >>>> Was debugging it, issue is with commit e1c92a7fbbc5 perf tests: Add another metric parsing test. >>>> >>>> So, there we are passing different runtime parameter value in "expr__find_other and expr__parse" >>>> in function `metric_parse_fake`. I believe we need to send same value. >>>> I will send fix patch for the same. >> Just wondering, was a patch ever submitted for this? Something still >> broken? I can't see any recent relevant changes to tests/pmu-events.c > The test itself shouldn't have changed, but the json files parsed by > jevents and turned into C code that the test exercises should have > changed. Jin Yao has sent two patch sets fixing a metric issue on SKL > (Skylake non-server) that should hopefully fix the issue there - I'll > check the status on these. Are you testing on Skylake? So I have re-read this thread, and it seems that 2x different things are being discussed: a. some breakage for test #10 on skylake b. test #67 being broken It seems that a. has been addressed. That's what I was asking about just now. So about b., which I thought may be broken for some other reason apart from my hacky patch. But it seems not the case, and a proper patch is needed there. Ian, have you had a chance to consider this issue in b.? That is, we have breakage for metrics using uncore alias expressions for when multiple uncore PMUs associated exist in the system? As before, looks broken by ded80bda8bc9 ("perf expr: Migrate expr ids table to a hashmap") Thanks, John
On 11/3/20 10:24 PM, John Garry wrote: > On 03/11/2020 16:05, Ian Rogers wrote: >> On Tue, Nov 3, 2020 at 6:43 AM John Garry<john.garry@huawei.com> wrote: >>> On 20/10/2020 17:53, Ian Rogers wrote: >>>>>> Thanks for taking a look John. If you want help you can send the >>>>>> output of "perf test 67 -vvv" to me. It is possible Broadwell has >>>>>> similar glitches in the json to Skylake. I tested the original test on >>>>>> server parts as I can access them as cloud machines. >>>>>> >>>>>>> I will have a look, but I was hoping that Ian would have a proper fix >>>>>>> for this on top of ("perf metricgroup: Fix uncore metric expressions"), >>>>>>> which now looks to be merged. >>>>>> I still have these changes to look at in my inbox but I'm assuming >>>>>> they're good:-) Sorry for not getting to them, but it's good they are >>>>>> merged. >>>>> Hi Ian, >>>>> Checked in upstream kernel with your fix patch, in powerpc also test case 67 is passing. >>>>> But I am getting issue in test 10 for powerpc >>>>> >>>>> [command]# ./perf test 10 >>>>> 10: PMU events : >>>>> 10.1: PMU event table sanity : Ok >>>>> 10.2: PMU event map aliases : Ok >>>>> 10.3: Parsing of PMU event table metrics : Skip (some metrics failed) >>>>> 10.4: Parsing of PMU event table metrics with fake PMUs : FAILED! >>>>> >>>>> Was debugging it, issue is with commit e1c92a7fbbc5 perf tests: Add another metric parsing test. >>>>> >>>>> So, there we are passing different runtime parameter value in "expr__find_other and expr__parse" >>>>> in function `metric_parse_fake`. I believe we need to send same value. >>>>> I will send fix patch for the same. >>> Just wondering, was a patch ever submitted for this? Something still >>> broken? I can't see any recent relevant changes to tests/pmu-events.c >> The test itself shouldn't have changed, but the json files parsed by >> jevents and turned into C code that the test exercises should have >> changed. Jin Yao has sent two patch sets fixing a metric issue on SKL >> (Skylake non-server) that should hopefully fix the issue there - I'll >> check the status on these. Are you testing on Skylake? > > So I have re-read this thread, and it seems that 2x different things are being discussed: > a. some breakage for test #10 on skylake > b. test #67 being broken > > It seems that a. has been addressed. That's what I was asking about just now. Hi Ian/John, The breakage for test #10 which I mentioned is for power9 machine, if that you were asking. I still need to send fix patch out. I will send it soon. Thanks, Kajol Jain > > So about b., which I thought may be broken for some other reason apart from my hacky patch. But it seems not the case, and a proper patch is needed there. > > Ian, have you had a chance to consider this issue in b.? That is, we have breakage for metrics using uncore alias expressions for when multiple uncore PMUs associated exist in the system? As before, looks broken by ded80bda8bc9 ("perf expr: Migrate expr ids table to a hashmap") > > Thanks, > John > >
diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c index d948a7f910cf..82ba3638f48c 100644 --- a/tools/perf/util/metricgroup.c +++ b/tools/perf/util/metricgroup.c @@ -213,7 +213,7 @@ static struct evsel *find_evsel_group(struct evlist *perf_evlist, /* Ignore event if already used and merging is disabled. */ if (metric_no_merge && test_bit(ev->idx, evlist_used)) continue; - if (!has_constraint && ev->leader != current_leader) { + if (!has_constraint && (!current_leader || strcmp(current_leader->name, ev->leader->name))) { /* * Start of a new group, discard the whole match and * start again. @@ -279,7 +279,7 @@ static struct evsel *find_evsel_group(struct evlist *perf_evlist, * when then group is left. */ if (!has_constraint && - ev->leader != metric_events[i]->leader) + strcmp(ev->leader->name, metric_events[i]->leader->name)) break; if (!strcmp(metric_events[i]->name, ev->name)) { set_bit(ev->idx, evlist_used);
Function find_evsel_group() seems broken for aliases covering multiple PMUs, as dicussed at: https://lore.kernel.org/lkml/CAP-5=fUy6FOszNRwJF6ZNpqQSSyrnLPV6GbkEcZMqAhUp3X0ZA@mail.gmail.com/ For now, hack a fix which I only know works for me. Signed-off-by: John Garry <john.garry@huawei.com> --- tools/perf/util/metricgroup.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)