linux-stable/tools/perf
Jin Yao 0dfbe4c646 perf vendor events: Fix DRAM_BW_Use 0 issue for CLX/SKX
Ian reports an issue that the metric DRAM_BW_Use often remains 0.

The metric expression for DRAM_BW_Use on CLX/SKX:

"( 64 * ( uncore_imc@cas_count_read@ + uncore_imc@cas_count_write@ ) / 1000000000 ) / duration_time"

The counts of uncore_imc/cas_count_read/ and uncore_imc/cas_count_write/
are scaled up by 64, that is to turn a count of cache lines into bytes,
the count is then divided by 1000000000 to give GB.

However, the counts of uncore_imc/cas_count_read/ and
uncore_imc/cas_count_write/ have been scaled yet.

The scale values are from sysfs, such as
/sys/devices/uncore_imc_0/events/cas_count_read.scale.
It's 6.103515625e-5 (64 / 1024.0 / 1024.0).

So if we use original metric expression, the result is not correct.

But the difficulty is, for SKL client, the counts are not scaled.

The metric expression for DRAM_BW_Use on SKL:

"64 * ( arb@event\\=0x81\\,umask\\=0x1@ + arb@event\\=0x84\\,umask\\=0x1@ ) / 1000000 / duration_time / 1000"

root@kbl-ppc:~# perf stat -M DRAM_BW_Use -a -- sleep 1

 Performance counter stats for 'system wide':

               190      arb/event=0x84,umask=0x1/ #     1.86 DRAM_BW_Use
        29,093,178      arb/event=0x81,umask=0x1/
     1,000,703,287 ns   duration_time

       1.000703287 seconds time elapsed

The result is expected.

So the easy way is just change the metric expression for CLX/SKX.
This patch changes the metric expression to:

"( ( ( uncore_imc@cas_count_read@ + uncore_imc@cas_count_write@ ) * 1048576 ) / 1000000000 ) / duration_time"

1048576 = 1024 * 1024.

Before (tested on CLX):

root@lkp-csl-2sp5 ~# perf stat -M DRAM_BW_Use -a -- sleep 1

 Performance counter stats for 'system wide':

            765.35 MiB  uncore_imc/cas_count_read/ #     0.00 DRAM_BW_Use
              5.42 MiB  uncore_imc/cas_count_write/
        1001515088 ns   duration_time

       1.001515088 seconds time elapsed

After:

root@lkp-csl-2sp5 ~# perf stat -M DRAM_BW_Use -a -- sleep 1

 Performance counter stats for 'system wide':

            767.95 MiB  uncore_imc/cas_count_read/ #     0.80 DRAM_BW_Use
              5.02 MiB  uncore_imc/cas_count_write/
        1001900010 ns   duration_time

       1.001900010 seconds time elapsed

Fixes: 038d3b53c2 ("perf vendor events intel: Update CascadelakeX events to v1.08")
Fixes: b5ff7f2799 ("perf vendor events: Update SkylakeX events to v1.21")
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20201023005334.7869-1-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2020-11-03 08:31:31 -03:00
..
arch tools headers UAPI: Update process_madvise affected files 2020-11-03 08:29:30 -03:00
bench perf tools changes for v5.10: 1st batch 2020-10-17 11:47:46 -07:00
Documentation perf c2c: Update documentation for metrics reorganization 2020-10-15 12:02:12 -03:00
examples/bpf Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next 2020-01-28 16:02:33 -08:00
include/bpf perf bpf: Remove bpf/ subdir from bpf.h headers used to build bpf events 2020-02-18 10:13:28 -03:00
jvmti perf jvmti: Remove redundant jitdump line table entries 2020-05-29 16:51:38 -03:00
pmu-events perf vendor events: Fix DRAM_BW_Use 0 issue for CLX/SKX 2020-11-03 08:31:31 -03:00
python
scripts perf script: Add min, max to futex-contention output, in addition to avg 2020-09-23 12:58:53 -03:00
tests perf tools: Add support for exclusive groups/events 2020-10-14 12:24:28 -03:00
trace perf trace: Use the autogenerated mmap 'prot' string/id table 2020-10-01 11:35:01 -03:00
ui perf top/report: Fix infinite loop in the TUI for grouped events 2020-09-01 12:15:52 -03:00
util perf tools: Fix crash with non-jited bpf progs 2020-11-03 08:30:34 -03:00
.gitignore .gitignore: add SPDX License Identifier 2020-03-25 11:50:48 +01:00
Build
builtin-annotate.c perf evsel: Rename perf_evsel__resort*() to evsel__resort*() 2020-05-28 10:03:24 -03:00
builtin-bench.c perf bench: Add build-id injection benchmark 2020-10-13 10:59:42 -03:00
builtin-buildid-cache.c perf tools: Pass build_id object to build_id__sprintf() 2020-10-14 08:46:22 -03:00
builtin-buildid-list.c
builtin-c2c.c perf c2c: Add metrics "RMT Load Hit" 2020-10-15 09:34:51 -03:00
builtin-config.c
builtin-data.c perf data: Add support to store time of day in CTF data conversion 2020-08-06 09:43:37 -03:00
builtin-diff.c perf diff: Support hot streams comparison 2020-10-14 13:34:48 -03:00
builtin-evlist.c perf evsel: Rename perf_evsel__fprintf() to evsel__fprintf() 2020-05-28 10:03:24 -03:00
builtin-ftrace.c perf: ftrace: Add filter support for option -F/--funcs 2020-09-04 16:11:16 -03:00
builtin-help.c
builtin-inject.c perf tools: Pass build_id object to filename__read_build_id() 2020-10-14 08:45:16 -03:00
builtin-kallsyms.c
builtin-kmem.c perf kmem: Pass additional arguments to 'perf record' 2020-07-10 09:37:55 -03:00
builtin-kvm.c perf evlist: Fix the class prefix for 'struct evlist' 'add' evsel methods 2020-06-22 16:28:09 -03:00
builtin-list.c perf list: Remove dead code in argument check 2020-09-09 11:12:10 -03:00
builtin-lock.c perf lock: Rename perf_evsel__*() operating on 'struct evsel *' to evsel__*() 2020-05-05 16:35:31 -03:00
builtin-mem.c perf c2c: Fix 'perf c2c record -e list' to show the default events used 2020-05-28 10:03:25 -03:00
builtin-probe.c perf probe: Do not show the skipped events 2020-05-28 10:03:24 -03:00
builtin-record.c perf tools: Consolidate close_control_option()'s into one function 2020-09-04 16:11:16 -03:00
builtin-report.c perf report: Disable ordered_events for raw dump 2020-09-01 12:20:25 -03:00
builtin-sched.c perf sched: Show start of latency as well 2020-10-13 11:01:42 -03:00
builtin-script.c perf script: Display negative tid in non-sample events 2020-09-17 16:06:22 -03:00
builtin-stat.c perf tools: Allow creation of cgroup without open 2020-09-28 09:18:06 -03:00
builtin-timechart.c perf tools: Replace zero-length array with flexible-array 2020-05-28 10:03:27 -03:00
builtin-top.c perf top: Skip side-band event setup if HAVE_LIBBPF_SUPPORT is not set 2020-08-21 10:22:23 -03:00
builtin-trace.c perf trace: Fix segfault when trying to trace events by cgroup 2020-11-03 08:31:03 -03:00
builtin-version.c perf tools: Make GTK2 support opt-in 2020-09-04 17:11:59 -03:00
builtin.h
check-headers.sh perf tools: Separate the checking of headers only used to build beautification tables 2020-09-29 08:56:38 -03:00
command-list.txt
CREDITS
design.txt perf tools: Support CAP_PERFMON capability 2020-04-16 12:19:08 -03:00
Makefile tools: Let O= makes handle a relative path with -C option 2020-03-06 17:08:28 -03:00
Makefile.config perf tools: Remove LTO compiler options when building perl support 2020-11-03 08:24:54 -03:00
Makefile.perf perf trace: Use the autogenerated mmap 'prot' string/id table 2020-10-01 11:35:01 -03:00
MANIFEST libperf: Move to tools/lib/perf 2020-01-06 11:46:09 -03:00
perf-archive.sh
perf-completion.sh
perf-read-vdso.c
perf-sys.h perf tests: Call test_attr__open() directly 2020-09-10 11:55:37 -03:00
perf-with-kcore.sh
perf.c
perf.h