perf tests: Fix false TEST_OK result for 'perf test hist'
PATCH 1/1 perf tests: Fix false TEST_OK result for 'perf test hist'
Commit 71d6de64feddd4b455555326fba2111b3006d9e0 ('perf test: Fix hist testcases when kptr_restrict is on') solves a double free problem when 'perf test hist' calling setup_fake_machine(). However, the result is still incorrect. For example:
$ ./perf test -v 'filtering hist entries' 25: Test filtering hist entries : --- start --- test child forked, pid 4186 Cannot create kernel maps test child finished with 0 ---- end ---- Test filtering hist entries: Ok
In this case the body of this test is not get executed at all, but the result is 'Ok'. I think 'Skip' should be better.
The reason is that *ALL* hists test cases forget to reset err after using it to hold an error code.
This patch ensure err is reset.
In case when kptr_restrict prevent normal user get kernel address, this test should be skipped, not fail. This patch uses linux/err.h to store error code in return value of setup_fake_machine(), and let 'EACCES' to indicate this problem.
--- tools/perf/tests/hists_common.c | 12 +++++++++--- tools/perf/tests/hists_common.h | 1 + tools/perf/tests/hists_cumulate.c | 6 +++++- tools/perf/tests/hists_filter.c | 6 +++++- tools/perf/tests/hists_link.c | 6 +++++- tools/perf/tests/hists_output.c | 6 +++++- 6 files changed, 30 insertions(+), 7 deletions(-)
tools/perf/tests/hists_common.c +9 -3
@@ -81,15 +81,20 @@ struct machine *setup_fake_machine(struct machines *machines) {struct machine *machine =machines__find (machines,HOST_KERNEL_ID );size_t i; +int err = -ENOMEM ;if (machine ==NULL ) {pr_debug ("Not enough memory for machine setup\n" ); -return NULL ; +return ERR_PTR (-ENOMEM ); }if (machine__create_kernel_maps (machine)) {pr_debug ("Cannot create kernel maps\n" ); -return NULL ; +if (symbol_conf.kptr_restrict) { +pr_debug ("Hint: Check /proc/sys/kernel/kptr_restrict.\n" ); + err = -EACCES ; + } +return ERR_PTR (err); }for (i =0 ; i <ARRAY_SIZE (fake_threads); i++) {
@@ -155,7 +160,8 @@ struct machine *setup_fake_machine(struct machines *machines) out:pr_debug ("Not enough memory for machine setup\n" );machine__delete_threads (machine); -return NULL ; +machine__exit (machine); +return ERR_PTR (err); }void print_hists_in (struct hists *hists)
tools/perf/tests/hists_common.h +1 -0
@@ -1,5 +1,6 @@#ifndef __PERF_TESTS__HISTS_COMMON_H__#define __PERF_TESTS__HISTS_COMMON_H__ +#include <linux/err.h> struct machine ;struct machines ;
tools/perf/tests/hists_cumulate.c +5 -1
@@ -706,13 +706,17 @@ int test__hists_cumulate(int subtest __maybe_unused) err =parse_events (evlist,"cpu-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;machines__init (&machines);/* setup threads/dso/map/symbols also */ machine =setup_fake_machine (&machines); -if (!machine) +if (IS_ERR (machine)) { +if (PTR_ERR (machine) == -EACCES ) + err =TEST_SKIP ; goto out; + }if (verbose >1 )machine__fprintf (machine, stderr);
tools/perf/tests/hists_filter.c +5 -1
@@ -120,6 +120,7 @@ int test__hists_filter(int subtest __maybe_unused) err =parse_events (evlist,"task-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;/* default sort order (comm,dso,sym) will be used */ if (setup_sorting () <0 )
@@ -129,8 +130,11 @@ int test__hists_filter(int subtest __maybe_unused)/* setup threads/dso/map/symbols also */ machine =setup_fake_machine (&machines); -if (!machine) +if (IS_ERR (machine)) { +if (PTR_ERR (machine) == -EACCES ) + err =TEST_SKIP ; goto out; + }if (verbose >1 )machine__fprintf (machine, stderr);
tools/perf/tests/hists_link.c +5 -1
@@ -293,6 +293,7 @@ int test__hists_link(int subtest __maybe_unused)if (err) goto out; + err =TEST_FAIL ;/* default sort order (comm,dso,sym) will be used */ if (setup_sorting () <0 ) goto out;
@@ -301,8 +302,11 @@ int test__hists_link(int subtest __maybe_unused)/* setup threads/dso/map/symbols also */ machine =setup_fake_machine (&machines); -if (!machine) +if (IS_ERR (machine)) { +if (PTR_ERR (machine) == -EACCES ) + err =TEST_SKIP ; goto out; + }if (verbose >1 )machine__fprintf (machine, stderr);
tools/perf/tests/hists_output.c +5 -1
@@ -597,13 +597,17 @@ int test__hists_output(int subtest __maybe_unused) err =parse_events (evlist,"cpu-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;machines__init (&machines);/* setup threads/dso/map/symbols also */ machine =setup_fake_machine (&machines); -if (!machine) +if (IS_ERR (machine)) { +if (PTR_ERR (machine) == -EACCES ) + err =TEST_SKIP ; goto out; + }if (verbose >1 )machine__fprintf (machine, stderr);
Signature · 1 line
1.8.3.4
Namhyung Kim
maintainerHi,
Quoted 14 lines · Wang Nan
Commit
71d6de64feddd4b455555326fba2111b3006d9e0('perf test: Fix hist testcases whenkptr_restrictis on') solves a double free problem when 'perf test hist' callingsetup_fake_machine(). However, the result is still incorrect. For example:$ ./perf test -v 'filtering hist entries' 25: Test filtering hist entries : --- start --- test child forked, pid 4186 Cannot create kernel maps test child finished with 0 ---- end ---- Test filtering hist entries: Ok
Quoted 2 lines
In this case the body of this test is not get executed at all, but the result is 'Ok'. I think 'Skip' should be better.
Right, I missed this part.
But on second thought, I think it should not create real kernel maps. What setup_fake_machine() does is to create/insert fake dso/mmap/symbols for testing hist functions. So it has to use the predefined info not real ones.
Quoted 10 lines
The reason is that *ALL* hists test cases forget to reset err after using it to hold an error code.
This patch ensure err is reset.
In case when
kptr_restrictprevent normal user get kernel address, this test should be skipped, not fail. This patch useslinux/err.hto store error code in return value ofsetup_fake_machine(), and let 'EACCES' to indicate this problem.
Please just remove the machine__create_kernel_maps() in setup_fake_machine() and simply return TEST_FAIL if there's an error.
Thanks,
Namhyung
Quoted 162 lines
Signed-off-by Wang Nan <[email protected]>Cc Arnaldo Carvalho de Melo <[email protected]>Cc Jiri Olsa <[email protected]>Cc Masami Hiramatsu <[email protected]>Cc Namhyung Kim <[email protected]>--- tools/perf/tests/hists_common.c | 12 +++++++++--- tools/perf/tests/hists_common.h | 1 + tools/perf/tests/hists_cumulate.c | 6 +++++- tools/perf/tests/hists_filter.c | 6 +++++- tools/perf/tests/hists_link.c | 6 +++++- tools/perf/tests/hists_output.c | 6 +++++- 6 files changed, 30 insertions(+), 7 deletions(-)tools/perf/tests/hists_common.c +9
-3@@ -81,15 +81,20 @@ struct machine *setup_fake_machine(struct machines *machines) {struct machine *machine =machines__find (machines,HOST_KERNEL_ID );size_t i; +int err = -ENOMEM ;if (machine ==NULL ) {pr_debug ("Not enough memory for machine setup\n" ); -return NULL ; +return ERR_PTR (-ENOMEM ); }if (machine__create_kernel_maps (machine)) {pr_debug ("Cannot create kernel maps\n" ); -return NULL ; +if (symbol_conf.kptr_restrict) { +pr_debug ("Hint: Check /proc/sys/kernel/kptr_restrict.\n" ); + err = -EACCES ; + } +return ERR_PTR (err); }for (i =0 ; i <ARRAY_SIZE (fake_threads); i++) {@@ -155,7 +160,8 @@ struct machine *setup_fake_machine(struct machines *machines) out:pr_debug ("Not enough memory for machine setup\n" );machine__delete_threads (machine); -return NULL ; +machine__exit (machine); +return ERR_PTR (err); }void print_hists_in (struct hists *hists)tools/perf/tests/hists_common.h +1
-0@@ -1,5 +1,6 @@#ifndef __PERF_TESTS__HISTS_COMMON_H__#define __PERF_TESTS__HISTS_COMMON_H__ +#include <linux/err.h> struct machine ;struct machines ;tools/perf/tests/hists_cumulate.c +5
-1@@ -706,13 +706,17 @@ int test__hists_cumulate(int subtest __maybe_unused) err =parse_events (evlist,"cpu-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;machines__init (&machines);/* setup threads/dso/map/symbols also */ machine =setup_fake_machine (&machines); -if (!machine) +if (IS_ERR (machine)) { +if (PTR_ERR (machine) == -EACCES ) + err =TEST_SKIP ; goto out; + }if (verbose >1 )machine__fprintf (machine, stderr);tools/perf/tests/hists_filter.c +5
-1@@ -120,6 +120,7 @@ int test__hists_filter(int subtest __maybe_unused) err =parse_events (evlist,"task-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;/* default sort order (comm,dso,sym) will be used */ if (setup_sorting () <0 )@@ -129,8 +130,11 @@ int test__hists_filter(int subtest __maybe_unused)/* setup threads/dso/map/symbols also */ machine =setup_fake_machine (&machines); -if (!machine) +if (IS_ERR (machine)) { +if (PTR_ERR (machine) == -EACCES ) + err =TEST_SKIP ; goto out; + }if (verbose >1 )machine__fprintf (machine, stderr);tools/perf/tests/hists_link.c +5
-1@@ -293,6 +293,7 @@ int test__hists_link(int subtest __maybe_unused)if (err) goto out; + err =TEST_FAIL ;/* default sort order (comm,dso,sym) will be used */ if (setup_sorting () <0 ) goto out;@@ -301,8 +302,11 @@ int test__hists_link(int subtest __maybe_unused)/* setup threads/dso/map/symbols also */ machine =setup_fake_machine (&machines); -if (!machine) +if (IS_ERR (machine)) { +if (PTR_ERR (machine) == -EACCES ) + err =TEST_SKIP ; goto out; + }if (verbose >1 )machine__fprintf (machine, stderr);tools/perf/tests/hists_output.c +5
-1@@ -597,13 +597,17 @@ int test__hists_output(int subtest __maybe_unused) err =parse_events (evlist,"cpu-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;machines__init (&machines);/* setup threads/dso/map/symbols also */ machine =setup_fake_machine (&machines); -if (!machine) +if (IS_ERR (machine)) { +if (PTR_ERR (machine) == -EACCES ) + err =TEST_SKIP ; goto out; + }if (verbose >1 )machine__fprintf (machine, stderr);Signature · 1 line
1.8.3.4
PATCH v2 1/1 perf test: Fix false TEST_OK result for 'perf test hist'
Commit 71d6de64feddd4b455555326fba2111b3006d9e0 ('perf test: Fix hist testcases when kptr_restrict is on') solves a double free problem when 'perf test hist' calling setup_fake_machine(). However, the result is still incorrect. For example:
$ ./perf test -v 'filtering hist entries' 25: Test filtering hist entries : --- start --- test child forked, pid 4186 Cannot create kernel maps test child finished with 0 ---- end ---- Test filtering hist entries: Ok
In this case the body of this test is not get executed at all, but the result is 'Ok'.
Actually, in setup_fake_machine() there's no need to create real kernel maps. What we want is the fake maps. This patch removes the machine__create_kernel_maps() in setup_fake_machine(), so it won't be affected by kptr_restrict setting.
Test result:
$ cat /proc/sys/kernel/kptr_restrict 1 $ ~/perf test -v hist 15: Test matching and linking multiple hists : --- start --- test child forked, pid 24031 test child finished with 0 ---- end ---- Test matching and linking multiple hists: Ok [SNIP]
--- tools/perf/tests/hists_common.c | 5 ----- 1 file changed, 5 deletions(-)
tools/perf/tests/hists_common.c +0 -5
@@ -87,11 +87,6 @@ struct machine *setup_fake_machine(struct machines *machines)return NULL ; } -if (machine__create_kernel_maps (machine)) { -pr_debug ("Cannot create kernel maps\n" ); -return NULL ; - } -for (i =0 ; i <ARRAY_SIZE (fake_threads); i++) {struct thread *thread;
Signature · 1 line
1.8.3.4
PATCH v2 1/1 perf test: Reset err after using it hold errcode in hist testcases
All hists test cases forget to reset err after using it to hold an error code. If error occure in setup_fake_machine() it incorrectly return TEST_OK.
This patch fixes it.
--- tools/perf/tests/hists_cumulate.c | 1 + tools/perf/tests/hists_filter.c | 1 + tools/perf/tests/hists_link.c | 1 + tools/perf/tests/hists_output.c | 1 + 4 files changed, 4 insertions(+)
tools/perf/tests/hists_cumulate.c +1 -0
@@ -706,6 +706,7 @@ int test__hists_cumulate(int subtest __maybe_unused) err =parse_events (evlist,"cpu-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;machines__init (&machines);
tools/perf/tests/hists_filter.c +1 -0
@@ -120,6 +120,7 @@ int test__hists_filter(int subtest __maybe_unused) err =parse_events (evlist,"task-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;/* default sort order (comm,dso,sym) will be used */ if (setup_sorting () <0 )
tools/perf/tests/hists_link.c +1 -0
@@ -293,6 +293,7 @@ int test__hists_link(int subtest __maybe_unused)if (err) goto out; + err =TEST_FAIL ;/* default sort order (comm,dso,sym) will be used */ if (setup_sorting () <0 ) goto out;
tools/perf/tests/hists_output.c +1 -0
@@ -597,6 +597,7 @@ int test__hists_output(int subtest __maybe_unused) err =parse_events (evlist,"cpu-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;machines__init (&machines);
Signature · 1 line
1.8.3.4
Re: [PATCH v2] perf test: Reset err after using it hold errcode in hist testcases
Quoted 11 lines · Wang Nan
All hists test cases forget to reset err after using it to hold an error code. If error occure in
setup_fake_machine()it incorrectly returnTEST_OK.This patch fixes it.
Signed-off-by Wang Nan <[email protected]>Suggested-by Namhyung Kim <[email protected]>Cc Arnaldo Carvalho de Melo <[email protected]>Cc Jiri Olsa <[email protected]>Cc Masami Hiramatsu <[email protected]>
Thanks,
Namhyung
Quoted 58 lines
--- tools/perf/tests/hists_cumulate.c | 1 + tools/perf/tests/hists_filter.c | 1 + tools/perf/tests/hists_link.c | 1 + tools/perf/tests/hists_output.c | 1 + 4 files changed, 4 insertions(+)tools/perf/tests/hists_cumulate.c +1
-0@@ -706,6 +706,7 @@ int test__hists_cumulate(int subtest __maybe_unused) err =parse_events (evlist,"cpu-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;machines__init (&machines);tools/perf/tests/hists_filter.c +1
-0@@ -120,6 +120,7 @@ int test__hists_filter(int subtest __maybe_unused) err =parse_events (evlist,"task-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;/* default sort order (comm,dso,sym) will be used */ if (setup_sorting () <0 )tools/perf/tests/hists_link.c +1
-0@@ -293,6 +293,7 @@ int test__hists_link(int subtest __maybe_unused)if (err) goto out; + err =TEST_FAIL ;/* default sort order (comm,dso,sym) will be used */ if (setup_sorting () <0 ) goto out;tools/perf/tests/hists_output.c +1
-0@@ -597,6 +597,7 @@ int test__hists_output(int subtest __maybe_unused) err =parse_events (evlist,"cpu-clock" ,NULL );if (err) goto out; + err =TEST_FAIL ;machines__init (&machines);Signature · 1 line
1.8.3.4
Re: [PATCH v2] perf test: Fix false TEST_OK result for 'perf test hist'
In reply to Wang NanQuoted 35 lines · Wang Nan
Commit
71d6de64feddd4b455555326fba2111b3006d9e0('perf test: Fix hist testcases whenkptr_restrictis on') solves a double free problem when 'perf test hist' callingsetup_fake_machine(). However, the result is still incorrect. For example:$ ./perf test -v 'filtering hist entries' 25: Test filtering hist entries : --- start --- test child forked, pid 4186 Cannot create kernel maps test child finished with 0 ---- end ---- Test filtering hist entries: OkIn this case the body of this test is not get executed at all, but the result is 'Ok'.
Actually, in
setup_fake_machine()there's no need to create real kernel maps. What we want is the fake maps. This patch removes themachine__create_kernel_maps()insetup_fake_machine(), so it won't be affected bykptr_restrictsetting.Test result:
$ cat /proc/sys/kernel/kptr_restrict 1 $ ~/perf test -v hist 15: Test matching and linking multiple hists : --- start --- test child forked, pid 24031 test child finished with 0 ---- end ---- Test matching and linking multiple hists: Ok [SNIP]
Quoted 5 lines
Signed-off-by Wang Nan <[email protected]>Suggested-by Namhyung Kim <[email protected]>Cc Arnaldo Carvalho de Melo <[email protected]>Cc Jiri Olsa <[email protected]>Cc Masami Hiramatsu <[email protected]>
Thanks,
Namhyung
Quoted 23 lines
--- tools/perf/tests/hists_common.c | 5 ----- 1 file changed, 5 deletions(-)tools/perf/tests/hists_common.c +0
-5@@ -87,11 +87,6 @@ struct machine *setup_fake_machine(struct machines *machines)return NULL ; } -if (machine__create_kernel_maps (machine)) { -pr_debug ("Cannot create kernel maps\n" ); -return NULL ; - } -for (i =0 ; i <ARRAY_SIZE (fake_threads); i++) {struct thread *thread;Signature · 1 line
1.8.3.4
Quoted 27 lines · Wang Nan
Commit
71d6de64feddd4b455555326fba2111b3006d9e0('perf test: Fix hist testcases whenkptr_restrictis on') solves a double free problem when 'perf test hist' callingsetup_fake_machine(). However, the result is still incorrect. For example:$ ./perf test -v 'filtering hist entries' 25: Test filtering hist entries : --- start --- test child forked, pid 4186 Cannot create kernel maps test child finished with 0 ---- end ---- Test filtering hist entries: OkIn this case the body of this test is not get executed at all, but the result is 'Ok'. I think 'Skip' should be better.
The reason is that *ALL* hists test cases forget to reset err after using it to hold an error code.
This patch ensure err is reset.
In case when
kptr_restrictprevent normal user get kernel address, this test should be skipped, not fail. This patch useslinux/err.hto store error code in return value ofsetup_fake_machine(), and let 'EACCES' to indicate this problem.
Quoted 5 lines
Signed-off-by Wang Nan <[email protected]>Cc Arnaldo Carvalho de Melo <[email protected]>Cc Jiri Olsa <[email protected]>Cc Masami Hiramatsu <[email protected]>Cc Namhyung Kim <[email protected]>
thanks,
jirka