perf test pmu-events case
v2 cover perf test pmu-events case
This series introduces a basic pmu-events test case.
The test works in two steps:
- Verifies events in generated pmu-events.c are as expected for some test events
- Verifies correctly created aliases for those test events for the HW PMUs in the host system
The test events are invented, and are (mostly) arch agnostic (even though I copied these events from pre-existing x86 and arm64 JSONs).
In the test, core and uncore events are treated slightly differently for testing aliases. For core test events, all events are matched. However, for uncore test events, these have "Unit" property, so can only be matched when the corresponding HW PMU exists in the host sytem.
A test run looks like this on my x86 dev box:
$ tools/perf/perf test -vv 10
10: PMU events :
--- start ---
test child forked, pid 26610
testing event table bp_l1_btb_correct: pass
testing event table bp_l2_btb_correct: pass
testing event table segment_reg_loads.any: pass
testing event table dispatch_blocked.any: pass
testing event table eist_trans: pass
testing event table uncore_hisi_ddrc.flux_wcmd: pass
testing event table unc_cbo_xsnp_response.miss_eviction: pass
Using CPUID GenuineIntel-6-3D-4intel_pt default config: tsc,pt,branch
skipping testing PMU breakpoint
testing aliases PMU uncore_cbox_1: matched event unc_cbo_xsnp_response.miss_eviction testing PMU uncore_cbox_1 aliases: pass
testing aliases PMU cpu: matched event bp_l1_btb_correct
testing aliases PMU cpu: matched event bp_l2_btb_correct
testing aliases PMU cpu: matched event segment_reg_loads.any
testing aliases PMU cpu: matched event dispatch_blocked.any
testing aliases PMU cpu: matched event eist_trans
testing PMU cpu aliases: pass
skipping testing PMU software
skipping testing PMU intel_bts
testing aliases PMU uncore_cbox_0: matched event unc_cbo_xsnp_response.miss_eviction testing PMU uncore_cbox_0 aliases: pass
skipping testing PMU tracepoint
test child finished with 0
---- end ----
PMU events: Ok
Changes since v1:
- Verify event table in generated pmu-events.c, in addition to verifying aliases
- Don't create separate table for test events, but include as a "testcpu"
John Garry (7):
perf jevents: Add some test events
perf jevents: Support test events folder
perf pmu: Refactor pmu_add_cpu_aliases()
perf test: Add pmu-events test
perf pmu: Add is_pmu_core()
perf pmu: Make pmu_uncore_alias_match() public
perf test: Test pmu-events aliases
.../perf/pmu-events/arch/test/test_cpu/branch.json | 12 + .../perf/pmu-events/arch/test/test_cpu/other.json | 26 ++ .../perf/pmu-events/arch/test/test_cpu/uncore.json | 21 ++ tools/perf/pmu-events/jevents.c | 30 ++ tools/perf/tests/Build | 1 + tools/perf/tests/builtin-test.c | 4 + tools/perf/tests/pmu-events.c | 373 +++++++++++++++++++++ tools/perf/tests/tests.h | 1 + tools/perf/util/pmu.c | 28 +- tools/perf/util/pmu.h | 5 + 10 files changed, 492 insertions(+), 9 deletions(-) create mode 100644 tools/perf/pmu-events/arch/test/test_cpu/branch.json create mode 100644 tools/perf/pmu-events/arch/test/test_cpu/other.json create mode 100644 tools/perf/pmu-events/arch/test/test_cpu/uncore.json create mode 100644 tools/perf/tests/pmu-events.c
Signature · 1 line
2.12.3
PATCH v2 1/7 perf jevents: Add some test events
Add some test PMU events. The events are randomly chosen from x86 and arm64 JSONs. The events include CPU and uncore events.
--- .../perf/pmu-events/arch/test/test_cpu/branch.json | 12 ++++++++++ .../perf/pmu-events/arch/test/test_cpu/other.json | 26 ++++++++++++++++++++++ .../perf/pmu-events/arch/test/test_cpu/uncore.json | 21 +++++++++++++++++ 3 files changed, 59 insertions(+) create mode 100644 tools/perf/pmu-events/arch/test/test_cpu/branch.json create mode 100644 tools/perf/pmu-events/arch/test/test_cpu/other.json create mode 100644 tools/perf/pmu-events/arch/test/test_cpu/uncore.json
tools/perf/pmu-events/arch/test/test_cpu/branch.json new file +12 -0
@@ -0,0 +1,12 @@ +[ + { + "EventName": "bp_l1_btb_correct", + "EventCode": "0x8a", + "BriefDescription": "L1 BTB Correction." + }, + { + "EventName": "bp_l2_btb_correct", + "EventCode": "0x8b", + "BriefDescription": "L2 BTB Correction." + } +]
tools/perf/pmu-events/arch/test/test_cpu/other.json new file +26 -0
@@ -0,0 +1,26 @@ +[ + { + "EventCode": "0x6", + "Counter": "0,1", + "UMask": "0x80", + "EventName": "SEGMENT_REG_LOADS.ANY", + "SampleAfterValue": "200000", + "BriefDescription": "Number of segment register loads." + }, + { + "EventCode": "0x9", + "Counter": "0,1", + "UMask": "0x20", + "EventName": "DISPATCH_BLOCKED.ANY", + "SampleAfterValue": "200000", + "BriefDescription": "Memory cluster signals to block micro-op dispatch for any reason" + }, + { + "EventCode": "0x3A", + "Counter": "0,1", + "UMask": "0x0", + "EventName": "EIST_TRANS", + "SampleAfterValue": "200000", + "BriefDescription": "Number of Enhanced Intel SpeedStep(R) Technology (EIST) transitions" + } +]
tools/perf/pmu-events/arch/test/test_cpu/uncore.json new file +21 -0
@@ -0,0 +1,21 @@ +[ + { + "EventCode": "0x02", + "EventName": "uncore_hisi_ddrc.flux_wcmd", + "BriefDescription": "DDRC write commands", + "PublicDescription": "DDRC write commands", + "Unit": "hisi_sccl,ddrc" + }, + { + "Unit": "CBO", + "EventCode": "0x22", + "UMask": "0x81", + "EventName": "UNC_CBO_XSNP_RESPONSE.MISS_EVICTION", + "BriefDescription": "A cross-core snoop resulted from L3 Eviction which misses in some processor core.", + "PublicDescription": "A cross-core snoop resulted from L3 Eviction which misses in some processor core.", + "Counter": "0,1", + "CounterMask": "0", + "Invert": "0", + "EdgeDetect": "0" + } +]
Signature · 1 line
2.12.3
[tip: perf/urgent] perf jevents: Add some test events
The following commit has been merged into the perf/urgent branch of tip:
The following commit has been merged into the perf/urgent branch of tip:
Commit-ID: c52db67a74b3a9b834af4588489cfea22ec280a3 Gitweb: https://git.kernel.org/tip/c52db67a74b3a9b834af4588489cfea22ec280a3 Author: John Garry <[email protected]> AuthorDate: Tue, 17 Mar 2020 19:02:13 +08:00 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitterDate: Tue, 24 Mar 2020 10:35:59 -03:00
perf jevents: Add some test events
Add some test PMU events. The events are randomly chosen from x86 and arm64 JSONs. The events include CPU and uncore events.
--- tools/perf/pmu-events/arch/test/test_cpu/branch.json | 12 +++++- tools/perf/pmu-events/arch/test/test_cpu/other.json | 26 +++++++++++- tools/perf/pmu-events/arch/test/test_cpu/uncore.json | 21 +++++++++- 3 files changed, 59 insertions(+) create mode 100644 tools/perf/pmu-events/arch/test/test_cpu/branch.json create mode 100644 tools/perf/pmu-events/arch/test/test_cpu/other.json create mode 100644 tools/perf/pmu-events/arch/test/test_cpu/uncore.json
tools/perf/pmu-events/arch/test/test_cpu/branch.json new file +12 -0
@@ -0,0 +1,12 @@ +[ + { + "EventName": "bp_l1_btb_correct", + "EventCode": "0x8a", + "BriefDescription": "L1 BTB Correction." + }, + { + "EventName": "bp_l2_btb_correct", + "EventCode": "0x8b", + "BriefDescription": "L2 BTB Correction." + } +]
tools/perf/pmu-events/arch/test/test_cpu/other.json new file +26 -0
@@ -0,0 +1,26 @@ +[ + { + "EventCode": "0x6", + "Counter": "0,1", + "UMask": "0x80", + "EventName": "SEGMENT_REG_LOADS.ANY", + "SampleAfterValue": "200000", + "BriefDescription": "Number of segment register loads." + }, + { + "EventCode": "0x9", + "Counter": "0,1", + "UMask": "0x20", + "EventName": "DISPATCH_BLOCKED.ANY", + "SampleAfterValue": "200000", + "BriefDescription": "Memory cluster signals to block micro-op dispatch for any reason" + }, + { + "EventCode": "0x3A", + "Counter": "0,1", + "UMask": "0x0", + "EventName": "EIST_TRANS", + "SampleAfterValue": "200000", + "BriefDescription": "Number of Enhanced Intel SpeedStep(R) Technology (EIST) transitions" + } +]
tools/perf/pmu-events/arch/test/test_cpu/uncore.json new file +21 -0
@@ -0,0 +1,21 @@ +[ + { + "EventCode": "0x02", + "EventName": "uncore_hisi_ddrc.flux_wcmd", + "BriefDescription": "DDRC write commands", + "PublicDescription": "DDRC write commands", + "Unit": "hisi_sccl,ddrc" + }, + { + "Unit": "CBO", + "EventCode": "0x22", + "UMask": "0x81", + "EventName": "UNC_CBO_XSNP_RESPONSE.MISS_EVICTION", + "BriefDescription": "A cross-core snoop resulted from L3 Eviction which misses in some processor core.", + "PublicDescription": "A cross-core snoop resulted from L3 Eviction which misses in some processor core.", + "Counter": "0,1", + "CounterMask": "0", + "Invert": "0", + "EdgeDetect": "0" + } +]
PATCH v2 2/7 perf jevents: Support test events folder
In reply to John GarryWith the goal of supporting pmu-events test case, introduce support for a test events folder.
These test events can be used for testing generation of pmu-event tables and alias creation for any arch.
When running the pmu-events test case, these test events will be used as the platform-agnostic events, so aliases can be created per-PMU and validated against known expected values.
To support the test events, add a "testcpu" entry in pmu_events_map[]. The pmu-events test will be able to lookup the events map for "testcpu", to verify the generated tables against expected values.
The resultant generated pmu-events.c will now look like the following:
struct pmu_event pme_ampere_emag[] = {
{
.name = "ldrex_spec",
.event = "event=0x6c",
.desc = "Exclusive operation spe...",
.topic = "intrinsic",
.long_desc = "Exclusive operation ...",
},
...
};
struct pmu_event pme_test_cpu[] = {
{
.name = "uncore_hisi_ddrc.flux_wcmd",
.event = "event=0x2",
.desc = "DDRC write commands. Unit: hisi_sccl,ddrc ",
.topic = "uncore",
.long_desc = "DDRC write commands",
.pmu = "hisi_sccl,ddrc",
},
{
.name = "unc_cbo_xsnp_response.miss_eviction",
.event = "umask=0x81,event=0x22",
.desc = "Unit: uncore_cbox A cross-core snoop resulted ...",
.topic = "uncore",
.long_desc = "A cross-core snoop resulted from L3 ...",
.pmu = "uncore_cbox",
},
{
.name = "eist_trans",
.event = "umask=0x0,period=200000,event=0x3a",
.desc = "Number of Enhanced Intel SpeedStep(R) ...",
.topic = "other",
},
{
.name = 0,
},};
struct pmu_events_map pmu_events_map[] = {
...
{
.cpuid = "0x00000000500f0000",
.version = "v1",
.type = "core",
.table = pme_ampere_emag
},
...
{
.cpuid = "testcpu",
.version = "v1",
.type = "core",
.table = pme_test_cpu,
},
{
.cpuid = 0,
.version = 0,
.type = 0,
.table = 0,
},};
--- tools/perf/pmu-events/jevents.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+)
tools/perf/pmu-events/jevents.c +30 -0
@@ -764,6 +764,19 @@ static void print_mapping_table_suffix(FILE *outfp)fprintf (outfp,"};\n" ); } +static void print_mapping_test_table (FILE *outfp) +{ +/* +* Print the terminating, NULL entry. +*/ +fprintf (outfp,"{\n" ); +fprintf (outfp,"\t.cpuid = \"testcpu\",\n" ); +fprintf (outfp,"\t.version = \"v1\",\n" ); +fprintf (outfp,"\t.type = \"core\",\n" ); +fprintf (outfp,"\t.table = pme_test_cpu,\n" ); +fprintf (outfp,"},\n" ); +} +static int process_mapfile (FILE *outfp,char *fpath) {int n =16384 ;
@@ -841,6 +854,7 @@ static int process_mapfile(FILE *outfp, char *fpath) } out: +print_mapping_test_table (outfp);print_mapping_table_suffix (outfp);fclose (mapfp);free (line);
@@ -1161,6 +1175,22 @@ int main(int argc, char *argv[]) goto empty_map; } +sprintf (ldirname,"%s/test" , start_dirname); + + rc =nftw (ldirname, process_one_file, maxfds,0 ); +if (rc && verbose) { +pr_info ("%s: Error walking file tree %s rc=%d for test\n" , + prog, ldirname, rc); + goto empty_map; + }else if (rc <0 ) { +/* Make build fail */ +free_arch_std_events (); + ret =1 ; + goto out_free_mapfile; + }else if (rc) { + goto empty_map; + } +if (close_table)print_events_table_suffix (eventsfp);
Signature · 1 line
2.12.3
Re: [PATCH v2 2/7] perf jevents: Support test events folder
Quoted 10 lines · John Garry
With the goal of supporting pmu-events test case, introduce support for a test events folder.
These test events can be used for testing generation of pmu-event tables and alias creation for any arch.
When running the pmu-events test case, these test events will be used as the platform-agnostic events, so aliases can be created per-PMU and validated against known expected values.
Quoted 5 lines
To support the test events, add a "testcpu" entry in pmu_events_map[]. The pmu-events test will be able to lookup the events map for "testcpu", to verify the generated tables against expected values.
The resultant generated
pmu-events.cwill now look like the following:
can't compile this one:
HOSTCC pmu-events/jevents.opmu-events/jevents.c: In function ‘main’:pmu-events/jevents.c:1195:3: error: ‘ret’ undeclared (first use in this function)
1195 | ret = 1;
| ^~~
pmu-events/jevents.c:1195:3: note: each undeclared identifier is reported only once for each function it appears in
pmu-events/jevents.c:1196:3: error: label ‘out_free_mapfile’ used but not defined
1196 | goto out_free_mapfile;
| ^~~~mv: cannot stat 'pmu-events/.jevents.o.tmp': No such file or directory
make[3]: *** [/home/jolsa/kernel/linux-perf/tools/build/Makefile.build:97: pmu-events/jevents.o] Error 1 make[2]: *** [Makefile.perf:619: pmu-events/jevents-in.o] Error 2
make[1]: *** [Makefile.perf:225: sub-make] Error 2
make: *** [Makefile:70: all] Error 2
jirka
Re: [PATCH v2 2/7] perf jevents: Support test events folder
Quoted 28 lines · Jiri Olsa
Quoted 10 lines · John Garry
With the goal of supporting pmu-events test case, introduce support for a test events folder.
These test events can be used for testing generation of pmu-event tables and alias creation for any arch.
When running the pmu-events test case, these test events will be used as the platform-agnostic events, so aliases can be created per-PMU and validated against known expected values.
Quoted 5 lines
To support the test events, add a "testcpu" entry in pmu_events_map[]. The pmu-events test will be able to lookup the events map for "testcpu", to verify the generated tables against expected values.
The resultant generated
pmu-events.cwill now look like the following:can't compile this one:
HOSTCC
pmu-events/jevents.opmu-events/jevents.c: In function ‘main’:pmu-events/jevents.c:1195:3: error: ‘ret’ undeclared (first use in this function)1195 | ret = 1; | ^~~ pmu-events/jevents.c:1195:3: note: each undeclared identifier is reported only once for each function it appears in pmu-events/jevents.c:1196:3: error: label ‘out_free_mapfile’ used but not defined 1196 | goto out_free_mapfile; | ^~~~
Quoted 5 lines
mv: cannot stat 'pmu-events/.jevents.o.tmp': No such file or directory
make[3]: *** [/home/jolsa/kernel/linux-perf/tools/build/Makefile.build:97:pmu-events/jevents.o] Error 1 make[2]: *** [Makefile.perf:619:pmu-events/jevents-in.o] Error 2
make[1]: *** [Makefile.perf:225: sub-make] Error 2
make: *** [Makefile:70: all] Error 2
Hi jirka,
What baseline are you using? I used v5.6-rc6. The patches are here:
https://github.com/hisilicon/kernel-dev/commits/private-topic-perf-5.6-pmu-events-test-upstream-v2
thanks,
John
Re: [PATCH v2 2/7] perf jevents: Support test events folder
Quoted 35 lines · John Garry
Quoted 28 lines · Jiri Olsa
Quoted 10 lines · John Garry
With the goal of supporting pmu-events test case, introduce support for a test events folder.
These test events can be used for testing generation of pmu-event tables and alias creation for any arch.
When running the pmu-events test case, these test events will be used as the platform-agnostic events, so aliases can be created per-PMU and validated against known expected values.
Quoted 5 lines
To support the test events, add a "testcpu" entry in pmu_events_map[]. The pmu-events test will be able to lookup the events map for "testcpu", to verify the generated tables against expected values.
The resultant generated
pmu-events.cwill now look like the following:can't compile this one:
HOSTCC
pmu-events/jevents.opmu-events/jevents.c: In function ‘main’:pmu-events/jevents.c:1195:3: error: ‘ret’ undeclared (first use in this function)1195 | ret = 1; | ^~~ pmu-events/jevents.c:1195:3: note: each undeclared identifier is reported only once for each function it appears in pmu-events/jevents.c:1196:3: error: label ‘out_free_mapfile’ used but not defined 1196 | goto out_free_mapfile; | ^~~~Quoted 5 lines
mv: cannot stat 'pmu-events/.jevents.o.tmp': No such file or directory
make[3]: *** [/home/jolsa/kernel/linux-perf/tools/build/Makefile.build:97:pmu-events/jevents.o] Error 1 make[2]: *** [Makefile.perf:619:pmu-events/jevents-in.o] Error 2
make[1]: *** [Makefile.perf:225: sub-make] Error 2
make: *** [Makefile:70: all] Error 2
Quoted 3 lines
Hi jirka,
What baseline are you using? I used v5.6-rc6. The patches are here:
I applied your patches on Arnaldo's perf/core
Quoted 2 lines
https://github.com/hisilicon/kernel-dev/commits/private-topic-perf-5.6-pmu-events-test-upstream-v2
ok, will check
jirka
Re: [PATCH v2 2/7] perf jevents: Support test events folder
Quoted 36 lines · Jiri Olsa
Quoted 35 lines · John Garry
Quoted 33 lines · Jiri Olsa
Quoted 10 lines · John Garry
With the goal of supporting pmu-events test case, introduce support for a test events folder.
These test events can be used for testing generation of pmu-event tables and alias creation for any arch.
When running the pmu-events test case, these test events will be used as the platform-agnostic events, so aliases can be created per-PMU and validated against known expected values.
Quoted 5 lines
To support the test events, add a "testcpu" entry in pmu_events_map[]. The pmu-events test will be able to lookup the events map for "testcpu", to verify the generated tables against expected values.
The resultant generated
pmu-events.cwill now look like the following:can't compile this one:
HOSTCC
pmu-events/jevents.opmu-events/jevents.c: In function ‘main’:pmu-events/jevents.c:1195:3: error: ‘ret’ undeclared (first use in this function)1195 | ret = 1; | ^~~ pmu-events/jevents.c:1195:3: note: each undeclared identifier is reported only once for each function it appears in pmu-events/jevents.c:1196:3: error: label ‘out_free_mapfile’ used but not defined 1196 | goto out_free_mapfile; | ^~~~mv: cannot stat 'pmu-events/.jevents.o.tmp': No such file or directory
make[3]: *** [/home/jolsa/kernel/linux-perf/tools/build/Makefile.build:97:pmu-events/jevents.o] Error 1 make[2]: *** [Makefile.perf:619:pmu-events/jevents-in.o] Error 2
make[1]: *** [Makefile.perf:225: sub-make] Error 2
make: *** [Makefile:70: all] Error 2
Quoted 5 lines
Quoted 3 lines
Hi jirka,
What baseline are you using? I used v5.6-rc6. The patches are here:
I applied your patches on Arnaldo's perf/core
My recent fix on jevents.c does not seem to be on that branch, but it is on perf/urgent and also included in v5.6-rc6
Thanks,
John
Quoted 10 lines
Quoted 3 lines
https://github.com/hisilicon/kernel-dev/commits/private-topic-perf-5.6-pmu-events-test-upstream-v2
ok, will check
jirka
.
Re: [PATCH v2 2/7] perf jevents: Support test events folder
Em Tue, Mar 17, 2020 at 05:42:46PM +0000, John Garry escreveu:
Quoted 37 lines · Jiri Olsa
Quoted 36 lines · John Garry
Quoted 35 lines · Jiri Olsa
Quoted 33 lines · John Garry
Quoted 10 lines
With the goal of supporting pmu-events test case, introduce support for a test events folder.
These test events can be used for testing generation of pmu-event tables and alias creation for any arch.
When running the pmu-events test case, these test events will be used as the platform-agnostic events, so aliases can be created per-PMU and validated against known expected values.
Quoted 5 lines
To support the test events, add a "testcpu" entry in pmu_events_map[]. The pmu-events test will be able to lookup the events map for "testcpu", to verify the generated tables against expected values.
The resultant generated
pmu-events.cwill now look like the following:can't compile this one:
HOSTCC
pmu-events/jevents.opmu-events/jevents.c: In function ‘main’:pmu-events/jevents.c:1195:3: error: ‘ret’ undeclared (first use in this function)1195 | ret = 1; | ^~~ pmu-events/jevents.c:1195:3: note: each undeclared identifier is reported only once for each function it appears in pmu-events/jevents.c:1196:3: error: label ‘out_free_mapfile’ used but not defined 1196 | goto out_free_mapfile; | ^~~~mv: cannot stat 'pmu-events/.jevents.o.tmp': No such file or directory
make[3]: *** [/home/jolsa/kernel/linux-perf/tools/build/Makefile.build:97:pmu-events/jevents.o] Error 1 make[2]: *** [Makefile.perf:619:pmu-events/jevents-in.o] Error 2
make[1]: *** [Makefile.perf:225: sub-make] Error 2
make: *** [Makefile:70: all] Error 2
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Hi jirka,
What baseline are you using? I used v5.6-rc6. The patches are here:
I applied your patches on Arnaldo's perf/core
My recent fix on
jevents.cdoes not seem to be on that branch, but it is on perf/urgent and also included in v5.6-rc6
I'll merge perf/urgent into perf/core soon,
- Arnaldo
Quoted 14 lines
Thanks,
JohnQuoted 10 lines
Quoted 3 lines
https://github.com/hisilicon/kernel-dev/commits/private-topic-perf-5.6-pmu-events-test-upstream-v2
ok, will check
jirka
.
Signature · 1 line
- Arnaldo
[tip: perf/urgent] perf jevents: Support test events folder
In reply to John GarryThe following commit has been merged into the perf/urgent branch of tip:
The following commit has been merged into the perf/urgent branch of tip:
Commit-ID: d84478088780acdecfcf50a0e52b71cc4ab7c520 Gitweb: https://git.kernel.org/tip/d84478088780acdecfcf50a0e52b71cc4ab7c520 Author: John Garry <[email protected]> AuthorDate: Tue, 17 Mar 2020 19:02:14 +08:00 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitterDate: Tue, 24 Mar 2020 10:35:59 -03:00
perf jevents: Support test events folder
With the goal of supporting pmu-events test case, introduce support for a test events folder.
These test events can be used for testing generation of pmu-event tables and alias creation for any arch.
When running the pmu-events test case, these test events will be used as the platform-agnostic events, so aliases can be created per-PMU and validated against known expected values.
To support the test events, add a "testcpu" entry in pmu_events_map[]. The pmu-events test will be able to lookup the events map for "testcpu", to verify the generated tables against expected values.
The resultant generated pmu-events.c will now look like the following:
struct pmu_event pme_ampere_emag[] = {
{
.name = "ldrex_spec",
.event = "event=0x6c",
.desc = "Exclusive operation spe...",
.topic = "intrinsic",
.long_desc = "Exclusive operation ...",
},
...
};
struct pmu_event pme_test_cpu[] = {
{
.name = "uncore_hisi_ddrc.flux_wcmd",
.event = "event=0x2",
.desc = "DDRC write commands. Unit: hisi_sccl,ddrc ",
.topic = "uncore",
.long_desc = "DDRC write commands",
.pmu = "hisi_sccl,ddrc",
},
{
.name = "unc_cbo_xsnp_response.miss_eviction",
.event = "umask=0x81,event=0x22",
.desc = "Unit: uncore_cbox A cross-core snoop resulted ...",
.topic = "uncore",
.long_desc = "A cross-core snoop resulted from L3 ...",
.pmu = "uncore_cbox",
},
{
.name = "eist_trans",
.event = "umask=0x0,period=200000,event=0x3a",
.desc = "Number of Enhanced Intel SpeedStep(R) ...",
.topic = "other",
},
{
.name = 0,
},
};
struct pmu_events_map pmu_events_map[] = {
...
{
.cpuid = "0x00000000500f0000",
.version = "v1",
.type = "core",
.table = pme_ampere_emag
},
...
{
.cpuid = "testcpu",
.version = "v1",
.type = "core",
.table = pme_test_cpu,
},
{
.cpuid = 0,
.version = 0,
.type = 0,
.table = 0,
},
};--- tools/perf/pmu-events/jevents.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+)
tools/perf/pmu-events/jevents.c +30 -0
@@ -771,6 +771,19 @@ static void print_mapping_table_suffix(FILE *outfp)fprintf (outfp,"};\n" ); } +static void print_mapping_test_table (FILE *outfp) +{ +/* +* Print the terminating, NULL entry. +*/ +fprintf (outfp,"{\n" ); +fprintf (outfp,"\t.cpuid = \"testcpu\",\n" ); +fprintf (outfp,"\t.version = \"v1\",\n" ); +fprintf (outfp,"\t.type = \"core\",\n" ); +fprintf (outfp,"\t.table = pme_test_cpu,\n" ); +fprintf (outfp,"},\n" ); +} +static int process_mapfile (FILE *outfp,char *fpath) {int n =16384 ;
@@ -848,6 +861,7 @@ static int process_mapfile(FILE *outfp, char *fpath) } out: +print_mapping_test_table (outfp);print_mapping_table_suffix (outfp);fclose (mapfp);free (line);
@@ -1168,6 +1182,22 @@ int main(int argc, char *argv[]) goto empty_map; } +sprintf (ldirname,"%s/test" , start_dirname); + + rc =nftw (ldirname, process_one_file, maxfds,0 ); +if (rc && verbose) { +pr_info ("%s: Error walking file tree %s rc=%d for test\n" , + prog, ldirname, rc); + goto empty_map; + }else if (rc <0 ) { +/* Make build fail */ +free_arch_std_events (); + ret =1 ; + goto out_free_mapfile; + }else if (rc) { + goto empty_map; + } +if (close_table)print_events_table_suffix (eventsfp);
PATCH v2 3/7 perf pmu: Refactor pmu_add_cpu_aliases()
In reply to John GarryCreate pmu_add_cpu_aliases_map() from pmu_add_cpu_aliases(), so the caller can pass the map; the pmu-events test would use this since there would be no CPUID matching to a mapfile there.
--- tools/perf/util/pmu.c | 21 +++++++++++++-------- tools/perf/util/pmu.h | 3 +++ 2 files changed, 16 insertions(+), 8 deletions(-)
tools/perf/util/pmu.c +13 -8
@@ -21,7 +21,6 @@#include "pmu.h" #include "parse-events.h" #include "header.h" -#include "pmu-events/pmu-events.h" #include "string2.h" #include "strbuf.h" #include "fncache.h"
@@ -744,16 +743,11 @@ static bool pmu_uncore_alias_match(const char *pmu_name, const char *name)* to the current running CPU. Then, add all PMU events from that table * as aliases. */ -static void pmu_add_cpu_aliases (struct list_head *head,struct perf_pmu *pmu) +void pmu_add_cpu_aliases_map (struct list_head *head,struct perf_pmu *pmu, +struct pmu_events_map *map) {int i; -struct pmu_events_map *map;const char *name = pmu->name; - - map =perf_pmu__find_map (pmu); -if (!map) -return ; -/* * Found a matching PMU events table. Create aliases */
@@ -788,6 +782,17 @@ static void pmu_add_cpu_aliases(struct list_head *head, struct perf_pmu *pmu) } } +static void pmu_add_cpu_aliases (struct list_head *head,struct perf_pmu *pmu) +{ +struct pmu_events_map *map; + + map =perf_pmu__find_map (pmu); +if (!map) +return ; + +pmu_add_cpu_aliases_map (head, pmu, map); +} +struct perf_event_attr *__weak perf_pmu__get_default_config (struct perf_pmu *pmu__maybe_unused ) {
tools/perf/util/pmu.h +3 -0
@@ -7,6 +7,7 @@#include <linux/perf_event.h> #include <stdbool.h> #include "parse-events.h" +#include "pmu-events/pmu-events.h" struct perf_evsel_config_term ;
@@ -97,6 +98,8 @@ int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt,int perf_pmu__test (void );struct perf_event_attr *perf_pmu__get_default_config (struct perf_pmu *pmu); +void pmu_add_cpu_aliases_map (struct list_head *head,struct perf_pmu *pmu, +struct pmu_events_map *map);struct pmu_events_map *perf_pmu__find_map (struct perf_pmu *pmu);
Signature · 1 line
2.12.3
[tip: perf/urgent] perf pmu: Refactor pmu_add_cpu_aliases()
The following commit has been merged into the perf/urgent branch of tip:
The following commit has been merged into the perf/urgent branch of tip:
Commit-ID: e45ad701e784e0eed8a07b537b47afb302c59dab Gitweb: https://git.kernel.org/tip/e45ad701e784e0eed8a07b537b47afb302c59dab Author: John Garry <[email protected]> AuthorDate: Tue, 17 Mar 2020 19:02:15 +08:00 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitterDate: Tue, 24 Mar 2020 10:35:59 -03:00
perf pmu: Refactor pmu_add_cpu_aliases()
Create pmu_add_cpu_aliases_map() from pmu_add_cpu_aliases(), so the caller can pass the map; the pmu-events test would use this since there would be no CPUID matching to a mapfile there.
--- tools/perf/util/pmu.c | 21 +++++++++++++-------- tools/perf/util/pmu.h | 3 +++ 2 files changed, 16 insertions(+), 8 deletions(-)
tools/perf/util/pmu.c +13 -8
@@ -21,7 +21,6 @@#include "pmu.h" #include "parse-events.h" #include "header.h" -#include "pmu-events/pmu-events.h" #include "string2.h" #include "strbuf.h" #include "fncache.h"
@@ -744,16 +743,11 @@ out:* to the current running CPU. Then, add all PMU events from that table * as aliases. */ -static void pmu_add_cpu_aliases (struct list_head *head,struct perf_pmu *pmu) +void pmu_add_cpu_aliases_map (struct list_head *head,struct perf_pmu *pmu, +struct pmu_events_map *map) {int i; -struct pmu_events_map *map;const char *name = pmu->name; - - map =perf_pmu__find_map (pmu); -if (!map) -return ; -/* * Found a matching PMU events table. Create aliases */
@@ -788,6 +782,17 @@ new_alias: } } +static void pmu_add_cpu_aliases (struct list_head *head,struct perf_pmu *pmu) +{ +struct pmu_events_map *map; + + map =perf_pmu__find_map (pmu); +if (!map) +return ; + +pmu_add_cpu_aliases_map (head, pmu, map); +} +struct perf_event_attr *__weak perf_pmu__get_default_config (struct perf_pmu *pmu__maybe_unused ) {
tools/perf/util/pmu.h +3 -0
@@ -7,6 +7,7 @@#include <linux/perf_event.h> #include <stdbool.h> #include "parse-events.h" +#include "pmu-events/pmu-events.h" struct perf_evsel_config_term ;
@@ -97,6 +98,8 @@ int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt,int perf_pmu__test (void );struct perf_event_attr *perf_pmu__get_default_config (struct perf_pmu *pmu); +void pmu_add_cpu_aliases_map (struct list_head *head,struct perf_pmu *pmu, +struct pmu_events_map *map);struct pmu_events_map *perf_pmu__find_map (struct perf_pmu *pmu);
PATCH v2 4/7 perf test: Add pmu-events test
In reply to John GarryAdd a pmu-events test.
The initial test will verify that the test tables in generated pmu-events.c match against known, expected values.
For known events added in pmu-events/arch/test, we need to add an entry in test_cpu_aliases_events[] or test_uncore_events[].
A sample run is as follows for x86:
john@linux-3c19:~/linux> tools/perf/perf test -vv 10
10: PMU event aliases :
--- start ---
test child forked, pid 5316
testing event table bp_l1_btb_correct: pass
testing event table bp_l2_btb_correct: pass
testing event table segment_reg_loads.any: pass testing event table dispatch_blocked.any: pass testing event table eist_trans: pass
testing event table uncore_hisi_ddrc.flux_wcmd: pass testing event table unc_cbo_xsnp_response.miss_eviction: pass test child finished with 0
---- end ----
PMU event aliases: Ok
--- tools/perf/tests/Build | 1 + tools/perf/tests/builtin-test.c | 4 + tools/perf/tests/pmu-events.c | 227 ++++++++++++++++++++++++++++++++++++++++ tools/perf/tests/tests.h | 1 + 4 files changed, 233 insertions(+) create mode 100644 tools/perf/tests/pmu-events.c
tools/perf/tests/Build +1 -0
@@ -14,6 +14,7 @@ perf-y += evsel-roundtrip-name.o perf-y += evsel-tp-sched.o perf-y += fdarray.o perf-y += pmu.o +perf-y += pmu-events.o perf-y += hists_common.o perf-y += hists_link.o perf-y += hists_filter.o
tools/perf/tests/builtin-test.c +4 -0
@@ -73,6 +73,10 @@ static struct test generic_tests[] = { .func = test__pmu, }, { + .desc = "PMU events", + .func = test__pmu_events, + }, + { .desc = "DSO data read", .func = test__dso_data, },
tools/perf/tests/pmu-events.c new file +227 -0
227 lines not on this page · raw message
tools/perf/tests/tests.h +1 -0
@@ -49,6 +49,7 @@ int test__perf_evsel__roundtrip_name_test(struct test *test, int subtest);int test__perf_evsel__tp_sched_test (struct test *test,int subtest);int test__syscall_openat_tp_fields (struct test *test,int subtest);int test__pmu (struct test *test,int subtest); +int test__pmu_events (struct test *test,int subtest);int test__attr (struct test *test,int subtest);int test__dso_data (struct test *test,int subtest);int test__dso_data_cache (struct test *test,int subtest);
Signature · 1 line
2.12.3
[tip: perf/urgent] perf test: Add pmu-events test
The following commit has been merged into the perf/urgent branch of tip:
The following commit has been merged into the perf/urgent branch of tip:
Commit-ID: a6c925fd3aa2aba043a73b0ad57ffd296e1c42c9 Gitweb: https://git.kernel.org/tip/a6c925fd3aa2aba043a73b0ad57ffd296e1c42c9 Author: John Garry <[email protected]> AuthorDate: Tue, 17 Mar 2020 19:02:16 +08:00 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitterDate: Tue, 24 Mar 2020 10:35:59 -03:00
perf test: Add pmu-events test
The initial test will verify that the test tables in generated pmu-events.c match against known, expected values.
For known events added in pmu-events/arch/test, we need to add an entry in test_cpu_aliases_events[] or test_uncore_events[].
A sample run is as follows for x86:
john@linux-3c19:~/linux> tools/perf/perf test -vv 10
10: PMU event aliases :
--- start ---
test child forked, pid 5316
testing event table bp_l1_btb_correct: pass
testing event table bp_l2_btb_correct: pass
testing event table segment_reg_loads.any: pass testing event table dispatch_blocked.any: pass testing event table eist_trans: pass
testing event table uncore_hisi_ddrc.flux_wcmd: pass testing event table unc_cbo_xsnp_response.miss_eviction: pass test child finished with 0
---- end ----
PMU event aliases: Ok
--- tools/perf/tests/Build | 1 +- tools/perf/tests/builtin-test.c | 4 +- tools/perf/tests/pmu-events.c | 233 +++++++++++++++++++++++++++++++- tools/perf/tests/tests.h | 1 +- 4 files changed, 239 insertions(+) create mode 100644 tools/perf/tests/pmu-events.c
tools/perf/tests/Build +1 -0
@@ -14,6 +14,7 @@ perf-y += evsel-roundtrip-name.o perf-y += evsel-tp-sched.o perf-y += fdarray.o perf-y += pmu.o +perf-y += pmu-events.o perf-y += hists_common.o perf-y += hists_link.o perf-y += hists_filter.o
tools/perf/tests/builtin-test.c +4 -0
@@ -73,6 +73,10 @@ static struct test generic_tests[] = { .func = test__pmu, }, { + .desc = "PMU events", + .func = test__pmu_events, + }, + { .desc = "DSO data read", .func = test__dso_data, },
tools/perf/tests/pmu-events.c new file +233 -0
233 lines not on this page · raw message
tools/perf/tests/tests.h +1 -0
@@ -49,6 +49,7 @@ int test__perf_evsel__roundtrip_name_test(struct test *test, int subtest);int test__perf_evsel__tp_sched_test (struct test *test,int subtest);int test__syscall_openat_tp_fields (struct test *test,int subtest);int test__pmu (struct test *test,int subtest); +int test__pmu_events (struct test *test,int subtest);int test__attr (struct test *test,int subtest);int test__dso_data (struct test *test,int subtest);int test__dso_data_cache (struct test *test,int subtest);
PATCH v2 5/7 perf pmu: Add is_pmu_core()
In reply to John GarryAdd a function to decide whether a PMU is a core PMU.
--- tools/perf/util/pmu.c | 5 +++++ tools/perf/util/pmu.h | 1 + 2 files changed, 6 insertions(+)
tools/perf/util/pmu.c +5 -0
@@ -1400,6 +1400,11 @@ static void wordwrap(char *s, int start, int max, int corr) } } +bool is_pmu_core (const char *name) +{ +return !strcmp (name,"cpu" ) ||is_arm_pmu_core (name); +} +void print_pmu_events (const char *event_glob,bool name_only,bool quiet_flag,bool long_desc,bool details_flag,bool deprecated) {
tools/perf/util/pmu.h +1 -0
@@ -88,6 +88,7 @@ int perf_pmu__format_parse(char *dir, struct list_head *head);struct perf_pmu *perf_pmu__scan (struct perf_pmu *pmu); +bool is_pmu_core (const char *name);void print_pmu_events (const char *event_glob,bool name_only,bool quiet,bool long_desc,bool details_flag,bool deprecated);
Signature · 1 line
2.12.3
[tip: perf/urgent] perf pmu: Add is_pmu_core()
The following commit has been merged into the perf/urgent branch of tip:
The following commit has been merged into the perf/urgent branch of tip:
Commit-ID: d504fae93dd61b734aefc403c7653d958aef655a Gitweb: https://git.kernel.org/tip/d504fae93dd61b734aefc403c7653d958aef655a Author: John Garry <[email protected]> AuthorDate: Tue, 17 Mar 2020 19:02:17 +08:00 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitterDate: Tue, 24 Mar 2020 10:35:59 -03:00
perf pmu: Add is_pmu_core()
Add a function to decide whether a PMU is a core PMU.
--- tools/perf/util/pmu.c | 5 +++++ tools/perf/util/pmu.h | 1 + 2 files changed, 6 insertions(+)
tools/perf/util/pmu.c +5 -0
@@ -1400,6 +1400,11 @@ static void wordwrap(char *s, int start, int max, int corr) } } +bool is_pmu_core (const char *name) +{ +return !strcmp (name,"cpu" ) ||is_arm_pmu_core (name); +} +void print_pmu_events (const char *event_glob,bool name_only,bool quiet_flag,bool long_desc,bool details_flag,bool deprecated) {
tools/perf/util/pmu.h +1 -0
@@ -88,6 +88,7 @@ int perf_pmu__format_parse(char *dir, struct list_head *head);struct perf_pmu *perf_pmu__scan (struct perf_pmu *pmu); +bool is_pmu_core (const char *name);void print_pmu_events (const char *event_glob,bool name_only,bool quiet,bool long_desc,bool details_flag,bool deprecated);
PATCH v2 6/7 perf pmu: Make pmu_uncore_alias_match() public
In reply to John GarryThe perf pmu-events test will want to use pmu_uncore_alias_match(), so make it public.
--- tools/perf/util/pmu.c | 2 +- tools/perf/util/pmu.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-)
tools/perf/util/pmu.c +1 -1
@@ -698,7 +698,7 @@ struct pmu_events_map *perf_pmu__find_map(struct perf_pmu *pmu)return map; } -static bool pmu_uncore_alias_match (const char *pmu_name,const char *name) +bool pmu_uncore_alias_match (const char *pmu_name,const char *name) {char *tmp =NULL , *tok, *str;bool res;
tools/perf/util/pmu.h +1 -0
@@ -103,6 +103,7 @@ void pmu_add_cpu_aliases_map(struct list_head *head, struct perf_pmu *pmu, struct pmu_events_map *map);struct pmu_events_map *perf_pmu__find_map (struct perf_pmu *pmu); +bool pmu_uncore_alias_match (const char *pmu_name,const char *name);int perf_pmu__convert_scale (const char *scale,char **end,double *sval);
Signature · 1 line
2.12.3
[tip: perf/urgent] perf pmu: Make pmu_uncore_alias_match() public
The following commit has been merged into the perf/urgent branch of tip:
The following commit has been merged into the perf/urgent branch of tip:
Commit-ID: 5b9a50001b2c23f90f2a21db4763f3a9599588c4 Gitweb: https://git.kernel.org/tip/5b9a50001b2c23f90f2a21db4763f3a9599588c4 Author: John Garry <[email protected]> AuthorDate: Tue, 17 Mar 2020 19:02:18 +08:00 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitterDate: Tue, 24 Mar 2020 10:35:59 -03:00
perf pmu: Make pmu_uncore_alias_match() public
The perf pmu-events test will want to use pmu_uncore_alias_match(), so make it public.
--- tools/perf/util/pmu.c | 2 +- tools/perf/util/pmu.h | 1 + 2 files changed, 2 insertions(+), 1 deletion(-)
tools/perf/util/pmu.c +1 -1
@@ -698,7 +698,7 @@ struct pmu_events_map *perf_pmu__find_map(struct perf_pmu *pmu)return map; } -static bool pmu_uncore_alias_match (const char *pmu_name,const char *name) +bool pmu_uncore_alias_match (const char *pmu_name,const char *name) {char *tmp =NULL , *tok, *str;bool res;
tools/perf/util/pmu.h +1 -0
@@ -103,6 +103,7 @@ void pmu_add_cpu_aliases_map(struct list_head *head, struct perf_pmu *pmu, struct pmu_events_map *map);struct pmu_events_map *perf_pmu__find_map (struct perf_pmu *pmu); +bool pmu_uncore_alias_match (const char *pmu_name,const char *name);int perf_pmu__convert_scale (const char *scale,char **end,double *sval);
PATCH v2 7/7 perf test: Test pmu-events aliases
In reply to John GarryAdd creating event aliases to the pmu-events test.
So currently we verify that the generated pmu-events.c is as expected for some test events. Now test that we generate aliases as expected for those events during normal operation.
For that, we cycle through each HW PMU in the system, and use the test events to create aliases, and verify those against known, expected values.
For core PMUs, we should create an alias for every event in test_cpu_events[].
However, for uncore PMUs, they need to be matched by the pmu_event.pmu member, so use test_uncore_events[]; so check the match beforehand with pmu_uncore_alias_match().
A sample run is as follows for my x86 machine:
john@linux-3c19:~/linux> tools/perf/perf test -vv 10
10: PMU events :
--- start ---
...
testing PMU uncore_arb aliases: no events to match
testing PMU cstate_pkg aliases: no events to match
skipping testing PMU breakpoint
testing aliases PMU uncore_cbox_1: matched event unc_cbo_xsnp_response.miss_eviction testing PMU uncore_cbox_1 aliases: pass
testing PMU power aliases: no events to match
testing aliases PMU cpu: matched event bp_l1_btb_correct
testing aliases PMU cpu: matched event bp_l2_btb_correct
testing aliases PMU cpu: matched event segment_reg_loads.any
testing aliases PMU cpu: matched event dispatch_blocked.any
testing aliases PMU cpu: matched event eist_trans
testing PMU cpu aliases: pass
testing PMU intel_pt aliases: no events to match
skipping testing PMU software
skipping testing PMU intel_bts
testing PMU uncore_imc aliases: no events to match
testing aliases PMU uncore_cbox_0: matched event unc_cbo_xsnp_response.miss_eviction testing PMU uncore_cbox_0 aliases: pass
testing PMU cstate_core aliases: no events to match
skipping testing PMU tracepoint
testing PMU msr aliases: no events to match
test child finished with 0
--- tools/perf/tests/pmu-events.c | 148 +++++++++++++++++++++++++++++++++++++++++- 1 file changed, 147 insertions(+), 1 deletion(-)
tools/perf/tests/pmu-events.c +147 -1
@@ -5,13 +5,24 @@#include <errno.h> #include <stdio.h> #include <linux/kernel.h> - +#include <linux/zalloc.h> #include "debug.h" #include "../pmu-events/pmu-events.h" struct perf_pmu_test_event {struct pmu_event event; + +/* extra events for aliases */ +const char *alias_str; + +/* +* Note: For when PublicDescription does not exist in the JSON, we +* will have no long_desc in pmu_event.long_desc, but long_desc may +* be set in the alias. +*/ +const char *alias_long_desc; }; +static struct perf_pmu_test_event test_cpu_events[] = { { .event = {
@@ -20,6 +31,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L1 BTB Correction", .topic = "branch", }, + .alias_str = "event=0x8a", + .alias_long_desc = "L1 BTB Correction", }, { .event = {
@@ -28,6 +41,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L2 BTB Correction", .topic = "branch", }, + .alias_str = "event=0x8b", + .alias_long_desc = "L2 BTB Correction", }, { .event = {
@@ -36,6 +51,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of segment register loads", .topic = "other", }, + .alias_str = "umask=0x80,(null)=0x30d40,event=0x6", + .alias_long_desc = "Number of segment register loads", }, { .event = {
@@ -44,6 +61,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Memory cluster signals to block micro-op dispatch for any reason", .topic = "other", }, + .alias_str = "umask=0x20,(null)=0x30d40,event=0x9", + .alias_long_desc = "Memory cluster signals to block micro-op dispatch for any reason", }, { .event = {
@@ -52,6 +71,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of Enhanced Intel SpeedStep(R) Technology (EIST) transitions", .topic = "other", }, + .alias_str = "umask=0,(null)=0x30d40,event=0x3a", + .alias_long_desc = "Number of Enhanced Intel SpeedStep(R) Technology (EIST) transitions", }, {/* sentinel */ }
@@ -67,6 +88,8 @@ static struct perf_pmu_test_event test_uncore_events[] = { .long_desc = "DDRC write commands", .pmu = "hisi_sccl,ddrc", }, + .alias_str = "event=0x2", + .alias_long_desc = "DDRC write commands", }, { .event = {
@@ -77,6 +100,8 @@ static struct perf_pmu_test_event test_uncore_events[] = { .long_desc = "A cross-core snoop resulted from L3 Eviction which misses in some processor core", .pmu = "uncore_cbox", }, + .alias_str = "umask=0x81,event=0x22", + .alias_long_desc = "A cross-core snoop resulted from L3 Eviction which misses in some processor core", }, {/* sentinel */ }
@@ -217,11 +242,132 @@ static int __test_pmu_event_table(void)return 0 ; } + +static struct perf_pmu_alias *find_alias (const char *test_event,struct list_head *aliases) +{ +struct perf_pmu_alias *alias; + +list_for_each_entry (alias, aliases, list) +if (!strcmp (test_event, alias->name)) +return alias; + +return NULL ; +} + +/* Verify aliases are as expected */ +static int __test__pmu_event_aliases (char *pmu_name,int *count) +{ +struct perf_pmu_test_event *test; +struct pmu_event *te; +struct perf_pmu *pmu; +LIST_HEAD (aliases); +int res =0 ; +bool use_uncore_table; +struct pmu_events_map *map =__test_pmu_get_events_map (); + +if (!map) +return -1 ; + +if (is_pmu_core (pmu_name)) { + test = &test_cpu_events[0 ]; + use_uncore_table = false; + }else { + test = &test_uncore_events[0 ]; + use_uncore_table = true; + } + + pmu =zalloc (sizeof (*pmu)); +if (!pmu) +return -1 ; + + pmu->name = pmu_name; + +pmu_add_cpu_aliases_map (&aliases, pmu, map); + +for (te = &test->event; te->name; test++, te = &test->event) { +struct perf_pmu_alias *alias =find_alias (te->name, &aliases); + +if (!alias) { +bool uncore_match =pmu_uncore_alias_match (pmu_name, + te->pmu); + +if (use_uncore_table && !uncore_match) { +pr_debug3 ("testing aliases PMU %s: skip matching alias %s\n" , + pmu_name, te->name); +continue ; + } + +pr_debug2 ("testing aliases PMU %s: no alias, alias_table->name=%s\n" , + pmu_name, te->name); + res =-1 ; +break ; + } + +if (!is_same (alias->desc, te->desc)) { +pr_debug2 ("testing aliases PMU %s: mismatched desc, %s vs %s\n" , + pmu_name, alias->desc, te->desc); + res =-1 ; +break ; + } + +if (!is_same (alias->long_desc, test->alias_long_desc)) { +pr_debug2 ("testing aliases PMU %s: mismatched long_desc, %s vs %s\n" , + pmu_name, alias->long_desc, + test->alias_long_desc); + res =-1 ; +break ; + } + +if (!is_same (alias->str, test->alias_str)) { +pr_debug2 ("testing aliases PMU %s: mismatched str, %s vs %s\n" , + pmu_name, alias->str, test->alias_str); + res =-1 ; +break ; + } + +if (!is_same (alias->topic, te->topic)) { +pr_debug2 ("testing aliases PMU %s: mismatched topic, %s vs %s\n" , + pmu_name, alias->topic, te->topic); + res =-1 ; +break ; + } + + (*count)++; +pr_debug2 ("testing aliases PMU %s: matched event %s\n" , + pmu_name, alias->name); + } + +free (pmu); +return res; +} +int test__pmu_events (struct test *test__maybe_unused ,int subtest__maybe_unused ) { +struct perf_pmu *pmu =NULL ; +if (__test_pmu_event_table ())return -1 ; +while ((pmu =perf_pmu__scan (pmu)) !=NULL ) { +int count =0 ; + +if (list_empty (&pmu->format)) { +pr_debug2 ("skipping testing PMU %s\n" , pmu->name); +continue ; + } + +if (__test__pmu_event_aliases (pmu->name, &count)) { +pr_debug ("testing PMU %s aliases: failed\n" , pmu->name); +return -1 ; + } + +if (count ==0 ) +pr_debug3 ("testing PMU %s aliases: no events to match\n" , + pmu->name); +else +pr_debug ("testing PMU %s aliases: pass\n" , pmu->name); + } +return 0 ; }
Signature · 1 line
2.12.3
Re: [PATCH v2 7/7] perf test: Test pmu-events aliases
On Tue, Mar 17, 2020 at 07:02:19PM +0800, John Garry wrote:
SNIP
Quoted 36 lines
struct
perf_pmu_test_event{tools/perf/tests/pmu-events.c +15
-0struct pmu_event event; + +/* extra events for aliases */ +const char *alias_str; + +/* +* Note: For when PublicDescription does not exist in the JSON, we +* will have no long_desc in pmu_event.long_desc, but long_desc may +* be set in the alias. +*/ +const char *alias_long_desc; }; +static struct perf_pmu_test_event test_cpu_events[] = { { .event = {@@ -20,6 +31,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L1 BTB Correction", .topic = "branch", }, + .alias_str = "event=0x8a", + .alias_long_desc = "L1 BTB Correction", }, { .event = {@@ -28,6 +41,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L2 BTB Correction", .topic = "branch", }, + .alias_str = "event=0x8b", + .alias_long_desc = "L2 BTB Correction", }, { .event = {
tools/perf/tests/pmu-events.c static struct perf_pmu_test_event test_cpu_events[] = { · line 54
Quoted 4 lines
tools/perf/tests/pmu-events.c +1
-0@@ -36,6 +51,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of segment register loads", .topic = "other", }, + .alias_str = "umask=0x80,(null)=0x30d40,event=0x6",
ah so we are using other pmus because of the format definitions
why is there the '(null)' in there?
jirka
Re: [PATCH v2 7/7] perf test: Test pmu-events aliases
Quoted 38 lines · Jiri Olsa
On Tue, Mar 17, 2020 at 07:02:19PM +0800, John Garry wrote:
SNIP
Quoted 34 lines
struct perf_pmu_test_event { struct pmu_event event;tools/perf/tests/pmu-events.c +15
-0+ +/* extra events for aliases */ +const char *alias_str; + +/* +* Note: For when PublicDescription does not exist in the JSON, we +* will have no long_desc in pmu_event.long_desc, but long_desc may +* be set in the alias. +*/ +const char *alias_long_desc; }; +static struct perf_pmu_test_event test_cpu_events[] = { { .event = {@@ -20,6 +31,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L1 BTB Correction", .topic ="branch" , }, + .alias_str = "event=0x8a", + .alias_long_desc = "L1 BTB Correction", }, { .event = {@@ -28,6 +41,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L2 BTB Correction", .topic ="branch" , }, + .alias_str = "event=0x8b", + .alias_long_desc = "L2 BTB Correction", }, {
Quoted 9 lines
tools/perf/tests/pmu-events.c static struct perf_pmu_test_event test_cpu_events[] = { · line 54
Quoted 6 lines
tools/perf/tests/pmu-events.c +1
-0.event = {@@ -36,6 +51,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of segment register loads", .topic ="other" , }, + .alias_str = "umask=0x80,(null)=0x30d40,event=0x6",ah so we are using other pmus because of the format definitions
Hi jirka,
Quoted 2 lines
why is there the '(null)' in there?
Well this is just coming from the generated alias string in the pmu code, and it does not seem to be handling "period" argument properly. It needs to be checked. Thanks,
John
Re: [PATCH v2 7/7] perf test: Test pmu-events aliases
Quoted 49 lines · John Garry
Quoted 38 lines · Jiri Olsa
On Tue, Mar 17, 2020 at 07:02:19PM +0800, John Garry wrote:
SNIP
Quoted 34 lines
struct perf_pmu_test_event { struct pmu_event event;tools/perf/tests/pmu-events.c +15
-0+ +/* extra events for aliases */ +const char *alias_str; + +/* +* Note: For when PublicDescription does not exist in the JSON, we +* will have no long_desc in pmu_event.long_desc, but long_desc may +* be set in the alias. +*/ +const char *alias_long_desc; }; +static struct perf_pmu_test_event test_cpu_events[] = { { .event = {@@ -20,6 +31,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L1 BTB Correction", .topic ="branch" , }, + .alias_str = "event=0x8a", + .alias_long_desc = "L1 BTB Correction", }, { .event = {@@ -28,6 +41,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L2 BTB Correction", .topic ="branch" , }, + .alias_str = "event=0x8b", + .alias_long_desc = "L2 BTB Correction", }, {Quoted 9 lines
tools/perf/tests/pmu-events.c static struct perf_pmu_test_event test_cpu_events[] = { · line 54
Quoted 6 lines
tools/perf/tests/pmu-events.c +1
-0.event = {@@ -36,6 +51,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of segment register loads", .topic ="other" , }, + .alias_str = "umask=0x80,(null)=0x30d40,event=0x6",ah so we are using other pmus because of the format definitions
Quoted 8 lines
Hi jirka,
Quoted 2 lines
why is there the '(null)' in there?
Well this is just coming from the generated alias string in the pmu code, and it does not seem to be handling "period" argument properly. It needs to be checked.
nice, it found first issue already ;-)
jirka
Arnaldo Carvalho de Melo
AppliedRe: [PATCH v2 7/7] perf test: Test pmu-events aliases
Em Tue, Mar 17, 2020 at 06:07:30PM +0100, Jiri Olsa escreveu:
Quoted 11 lines · John Garry
Quoted 9 lines · Jiri Olsa
Quoted 8 lines · John Garry
tools/perf/tests/pmu-events.c static struct perf_pmu_test_event test_cpu_events[] = { · line 54
Quoted 5 lines
tools/perf/tests/pmu-events.c +1
-0@@ -36,6 +51,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of segment register loads", .topic ="other" , }, + .alias_str = "umask=0x80,(null)=0x30d40,event=0x6",ah so we are using other pmus because of the format definitions
Quoted 7 lines
Quoted 5 lines
Quoted 1 line
why is there the '(null)' in there?
Well this is just coming from the generated alias string in the pmu code, and it does not seem to be handling "period" argument properly. It needs to be checked.
nice, it found first issue already ;-)
Applied the series to perf/core, good job! What about the fix for the above (null) problem?
Cheers,
- Arnaldo
Re: [PATCH v2 7/7] perf test: Test pmu-events aliases
Quoted 19 lines · Arnaldo Carvalho de Melo
Em Tue, Mar 17, 2020 at 06:07:30PM +0100, Jiri Olsa escreveu:
Quoted 18 lines · John Garry
Quoted 10 lines · Jiri Olsa
Quoted 8 lines · John Garry
tools/perf/tests/pmu-events.c static struct perf_pmu_test_event test_cpu_events[] = { · line 54
Quoted 5 lines
tools/perf/tests/pmu-events.c +1
-0@@ -36,6 +51,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of segment register loads", .topic ="other" , }, + .alias_str = "umask=0x80,(null)=0x30d40,event=0x6",ah so we are using other pmus because of the format definitions
Quoted 5 lines
Quoted 1 line
why is there the '(null)' in there?
Well this is just coming from the generated alias string in the pmu code, and it does not seem to be handling "period" argument properly. It needs to be checked.
nice, it found first issue already ;-)
thanks
Quoted 3 lines
Applied the series to perf/core, good job! What about the fix for the above (null) problem?
So I had started to look at that, but then the codepath lead into the lex parsing, which I am not familiar with.
So from when we parse the event terms in parse_events_terms(), we get 3x terms:
config=umask, then newval=umask=0x80
confg=(null), then newval=umask=0x80,(null)=x030d40
config=event, then newval=umask=0x80,(null)=x030d40,event=0x6
I can continue to look. Maybe jirka has an idea on this and what happens in the lex parsing.
Cheers,
John
Re: [PATCH v2 7/7] perf test: Test pmu-events aliases
Quoted 29 lines · John Garry
Quoted 18 lines · Arnaldo Carvalho de Melo
Em Tue, Mar 17, 2020 at 06:07:30PM +0100, Jiri Olsa escreveu:
Quoted 17 lines · John Garry
Quoted 10 lines · Jiri Olsa
Quoted 8 lines · John Garry
tools/perf/tests/pmu-events.c static struct perf_pmu_test_event test_cpu_events[] = { · line 54
Quoted 5 lines
tools/perf/tests/pmu-events.c +1
-0@@ -36,6 +51,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of segment register loads", .topic ="other" , }, + .alias_str = "umask=0x80,(null)=0x30d40,event=0x6",ah so we are using other pmus because of the format definitions
Quoted 5 lines
Quoted 1 line
why is there the '(null)' in there?
Well this is just coming from the generated alias string in the pmu code, and it does not seem to be handling "period" argument properly. It needs to be checked.
nice, it found first issue already ;-)
thanks
Quoted 3 lines
Applied the series to perf/core, good job! What about the fix for the above (null) problem?
So I had started to look at that, but then the codepath lead into the lex parsing, which I am not familiar with.
Quoted 8 lines
So from when we parse the event terms in
parse_events_terms(), we get 3x terms:
config=umask, then newval=umask=0x80
confg=(null), then newval=umask=0x80,(null)=x030d40 config=event, then newval=umask=0x80,(null)=x030d40,event=0x6I can continue to look. Maybe jirka has an idea on this and what happens in the lex parsing.
yep, I plan to check on it
jirka
[tip: perf/urgent] perf test: Test pmu-events aliases
In reply to John GarryThe following commit has been merged into the perf/urgent branch of tip:
The following commit has been merged into the perf/urgent branch of tip:
Commit-ID: 956a78356c24c1ac3becf854135dc065b6b74265 Gitweb: https://git.kernel.org/tip/956a78356c24c1ac3becf854135dc065b6b74265 Author: John Garry <[email protected]> AuthorDate: Tue, 17 Mar 2020 19:02:19 +08:00 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitterDate: Tue, 24 Mar 2020 10:36:00 -03:00
perf test: Test pmu-events aliases
Add creating event aliases to the pmu-events test.
So currently we verify that the generated pmu-events.c is as expected for some test events. Now test that we generate aliases as expected for those events during normal operation.
For that, we cycle through each HW PMU in the system, and use the test events to create aliases, and verify those against known, expected values.
For core PMUs, we should create an alias for every event in test_cpu_events[].
However, for uncore PMUs, they need to be matched by the pmu_event.pmu member, so use test_uncore_events[]; so check the match beforehand with pmu_uncore_alias_match().
A sample run is as follows for my x86 machine:
john@linux-3c19:~/linux> tools/perf/perf test -vv 10
10: PMU events :
--- start ---
...
testing PMU uncore_arb aliases: no events to match
testing PMU cstate_pkg aliases: no events to match
skipping testing PMU breakpoint
testing aliases PMU uncore_cbox_1: matched event unc_cbo_xsnp_response.miss_eviction testing PMU uncore_cbox_1 aliases: pass
testing PMU power aliases: no events to match
testing aliases PMU cpu: matched event bp_l1_btb_correct
testing aliases PMU cpu: matched event bp_l2_btb_correct
testing aliases PMU cpu: matched event segment_reg_loads.any
testing aliases PMU cpu: matched event dispatch_blocked.any
testing aliases PMU cpu: matched event eist_trans
testing PMU cpu aliases: pass
testing PMU intel_pt aliases: no events to match
skipping testing PMU software
skipping testing PMU intel_bts
testing PMU uncore_imc aliases: no events to match
testing aliases PMU uncore_cbox_0: matched event unc_cbo_xsnp_response.miss_eviction testing PMU uncore_cbox_0 aliases: pass
testing PMU cstate_core aliases: no events to match
skipping testing PMU tracepoint
testing PMU msr aliases: no events to match
test child finished with 0
--- tools/perf/tests/pmu-events.c | 148 ++++++++++++++++++++++++++++++++- 1 file changed, 147 insertions(+), 1 deletion(-)
tools/perf/tests/pmu-events.c +147 -1
@@ -5,13 +5,24 @@#include <errno.h> #include <stdio.h> #include <linux/kernel.h> - +#include <linux/zalloc.h> #include "debug.h" #include "../pmu-events/pmu-events.h" struct perf_pmu_test_event {struct pmu_event event; + +/* extra events for aliases */ +const char *alias_str; + +/* +* Note: For when PublicDescription does not exist in the JSON, we +* will have no long_desc in pmu_event.long_desc, but long_desc may +* be set in the alias. +*/ +const char *alias_long_desc; }; +static struct perf_pmu_test_event test_cpu_events[] = { { .event = {
@@ -20,6 +31,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L1 BTB Correction", .topic = "branch", }, + .alias_str = "event=0x8a", + .alias_long_desc = "L1 BTB Correction", }, { .event = {
@@ -28,6 +41,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "L2 BTB Correction", .topic = "branch", }, + .alias_str = "event=0x8b", + .alias_long_desc = "L2 BTB Correction", }, { .event = {
@@ -36,6 +51,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of segment register loads", .topic = "other", }, + .alias_str = "umask=0x80,(null)=0x30d40,event=0x6", + .alias_long_desc = "Number of segment register loads", }, { .event = {
@@ -44,6 +61,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Memory cluster signals to block micro-op dispatch for any reason", .topic = "other", }, + .alias_str = "umask=0x20,(null)=0x30d40,event=0x9", + .alias_long_desc = "Memory cluster signals to block micro-op dispatch for any reason", }, { .event = {
@@ -52,6 +71,8 @@ static struct perf_pmu_test_event test_cpu_events[] = { .desc = "Number of Enhanced Intel SpeedStep(R) Technology (EIST) transitions", .topic = "other", }, + .alias_str = "umask=0,(null)=0x30d40,event=0x3a", + .alias_long_desc = "Number of Enhanced Intel SpeedStep(R) Technology (EIST) transitions", }, {/* sentinel */ .event = {
@@ -70,6 +91,8 @@ static struct perf_pmu_test_event test_uncore_events[] = { .long_desc = "DDRC write commands", .pmu = "hisi_sccl,ddrc", }, + .alias_str = "event=0x2", + .alias_long_desc = "DDRC write commands", }, { .event = {
@@ -80,6 +103,8 @@ static struct perf_pmu_test_event test_uncore_events[] = { .long_desc = "A cross-core snoop resulted from L3 Eviction which misses in some processor core", .pmu = "uncore_cbox", }, + .alias_str = "umask=0x81,event=0x22", + .alias_long_desc = "A cross-core snoop resulted from L3 Eviction which misses in some processor core", }, {/* sentinel */ .event = {
@@ -223,11 +248,132 @@ static int __test_pmu_event_table(void)return 0 ; } + +static struct perf_pmu_alias *find_alias (const char *test_event,struct list_head *aliases) +{ +struct perf_pmu_alias *alias; + +list_for_each_entry (alias, aliases, list) +if (!strcmp (test_event, alias->name)) +return alias; + +return NULL ; +} + +/* Verify aliases are as expected */ +static int __test__pmu_event_aliases (char *pmu_name,int *count) +{ +struct perf_pmu_test_event *test; +struct pmu_event *te; +struct perf_pmu *pmu; +LIST_HEAD (aliases); +int res =0 ; +bool use_uncore_table; +struct pmu_events_map *map =__test_pmu_get_events_map (); + +if (!map) +return -1 ; + +if (is_pmu_core (pmu_name)) { + test = &test_cpu_events[0 ]; + use_uncore_table = false; + }else { + test = &test_uncore_events[0 ]; + use_uncore_table = true; + } + + pmu =zalloc (sizeof (*pmu)); +if (!pmu) +return -1 ; + + pmu->name = pmu_name; + +pmu_add_cpu_aliases_map (&aliases, pmu, map); + +for (te = &test->event; te->name; test++, te = &test->event) { +struct perf_pmu_alias *alias =find_alias (te->name, &aliases); + +if (!alias) { +bool uncore_match =pmu_uncore_alias_match (pmu_name, + te->pmu); + +if (use_uncore_table && !uncore_match) { +pr_debug3 ("testing aliases PMU %s: skip matching alias %s\n" , + pmu_name, te->name); +continue ; + } + +pr_debug2 ("testing aliases PMU %s: no alias, alias_table->name=%s\n" , + pmu_name, te->name); + res =-1 ; +break ; + } + +if (!is_same (alias->desc, te->desc)) { +pr_debug2 ("testing aliases PMU %s: mismatched desc, %s vs %s\n" , + pmu_name, alias->desc, te->desc); + res =-1 ; +break ; + } + +if (!is_same (alias->long_desc, test->alias_long_desc)) { +pr_debug2 ("testing aliases PMU %s: mismatched long_desc, %s vs %s\n" , + pmu_name, alias->long_desc, + test->alias_long_desc); + res =-1 ; +break ; + } + +if (!is_same (alias->str, test->alias_str)) { +pr_debug2 ("testing aliases PMU %s: mismatched str, %s vs %s\n" , + pmu_name, alias->str, test->alias_str); + res =-1 ; +break ; + } + +if (!is_same (alias->topic, te->topic)) { +pr_debug2 ("testing aliases PMU %s: mismatched topic, %s vs %s\n" , + pmu_name, alias->topic, te->topic); + res =-1 ; +break ; + } + + (*count)++; +pr_debug2 ("testing aliases PMU %s: matched event %s\n" , + pmu_name, alias->name); + } + +free (pmu); +return res; +} +int test__pmu_events (struct test *test__maybe_unused ,int subtest__maybe_unused ) { +struct perf_pmu *pmu =NULL ; +if (__test_pmu_event_table ())return -1 ; +while ((pmu =perf_pmu__scan (pmu)) !=NULL ) { +int count =0 ; + +if (list_empty (&pmu->format)) { +pr_debug2 ("skipping testing PMU %s\n" , pmu->name); +continue ; + } + +if (__test__pmu_event_aliases (pmu->name, &count)) { +pr_debug ("testing PMU %s aliases: failed\n" , pmu->name); +return -1 ; + } + +if (count ==0 ) +pr_debug3 ("testing PMU %s aliases: no events to match\n" , + pmu->name); +else +pr_debug ("testing PMU %s aliases: pass\n" , pmu->name); + } +return 0 ; }
On Tue, Mar 17, 2020 at 07:02:12PM +0800, John Garry wrote:
SNIP
Quoted 38 lines
$ tools/perf/perf test -vv 1010: PMU events :
--- start ---
test child forked, pid 26610
testing event tablebp_l1_btb_correct: pass
testing event tablebp_l2_btb_correct: pass
testing event table segment_reg_loads.any: pass
testing event table dispatch_blocked.any: pass
testing event tableeist_trans: pass
testing event table uncore_hisi_ddrc.flux_wcmd: pass
testing event table unc_cbo_xsnp_response.miss_eviction: pass
Using CPUID GenuineIntel-6-3D-4intel_ptdefault config: tsc,pt,branch
skipping testing PMU breakpoint
testing aliases PMUuncore_cbox_1: matched event unc_cbo_xsnp_response.miss_eviction testing PMUuncore_cbox_1aliases: pass
testing aliases PMU cpu: matched eventbp_l1_btb_correct
testing aliases PMU cpu: matched eventbp_l2_btb_correct
testing aliases PMU cpu: matched event segment_reg_loads.any
testing aliases PMU cpu: matched event dispatch_blocked.any
testing aliases PMU cpu: matched eventeist_trans
testing PMU cpu aliases: pass
skipping testing PMU software
skipping testing PMUintel_bts
testing aliases PMUuncore_cbox_0: matched event unc_cbo_xsnp_response.miss_eviction testing PMUuncore_cbox_0aliases: pass
skipping testing PMU tracepoint
test child finished with 0
---- end ----
PMU events: OkChanges since v1:
- Verify event table in generatedpmu-events.c, in addition to verifying aliases
- Don't create separate table for test events, but include as a "testcpu"
Quoted 8 lines
John Garry (7):
perf jevents: Add some test events
perf jevents: Support test events folder
perf pmu: Refactorpmu_add_cpu_aliases()
perf test: Add pmu-events test
perf pmu: Addis_pmu_core()
perf pmu: Makepmu_uncore_alias_match()public
perf test: Test pmu-events aliases
thanks a lot!
jirka