Adding support to specify raw event with 'r0<HEX>' syntax within
pmu term syntax like:
-e cpu/r0xdead/
It will be used to specify raw events in cases where they conflict
with real pmu terms, like 'read', which is valid raw event syntax,
but also a possible pmu term name as reported by Jin Yao.
Reported-by: Jin Yao <[email protected]>
Signed-off-by: Jiri Olsa <[email protected]>
---
tools/perf/Documentation/perf-list.txt | 1 +
tools/perf/tests/parse-events.c | 5 +++++
tools/perf/util/parse-events.l | 1 +
3 files changed, 7 insertions(+)
diff --git a/tools/perf/Documentation/perf-list.txt b/tools/perf/Documentation/perf-list.txt
index 376a50b3452d..10ed539a8859 100644
--- a/tools/perf/Documentation/perf-list.txt
+++ b/tools/perf/Documentation/perf-list.txt
@@ -119,6 +119,7 @@ It's also possible to use pmu syntax:
perf record -e r1a8 -a sleep 1
perf record -e cpu/r1a8/ ...
+ perf record -e cpu/r0x1a8/ ...
You should refer to the processor specific documentation for getting these
details. Some of them are referenced in the SEE ALSO section below.
diff --git a/tools/perf/tests/parse-events.c b/tools/perf/tests/parse-events.c
index 895188b63f96..5aaddcb0058a 100644
--- a/tools/perf/tests/parse-events.c
+++ b/tools/perf/tests/parse-events.c
@@ -1766,6 +1766,11 @@ static struct evlist_test test__events_pmu[] = {
.check = test__checkevent_raw_pmu,
.id = 4,
},
+ {
+ .name = "software/r0x1a/",
+ .check = test__checkevent_raw_pmu,
+ .id = 4,
+ },
};
struct terms_test {
diff --git a/tools/perf/util/parse-events.l b/tools/perf/util/parse-events.l
index 56912c9641f5..44c85fea5d00 100644
--- a/tools/perf/util/parse-events.l
+++ b/tools/perf/util/parse-events.l
@@ -293,6 +293,7 @@ percore { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_PERCORE); }
aux-output { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_AUX_OUTPUT); }
aux-sample-size { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_AUX_SAMPLE_SIZE); }
r{num_raw_hex} { return raw(yyscanner); }
+r0x{num_raw_hex} { return raw(yyscanner); }
, { return ','; }
"/" { BEGIN(INITIAL); return '/'; }
{name_minus} { return str(yyscanner, PE_NAME); }
--
2.25.4
Jin Yao reported issue with possible conflict between raw
events and term values in pmu event syntax.
Currently following syntax is resolved as raw event with
0xead value:
uncore_imc_free_running/read/
instead of using 'read' term from uncore_imc_free_running pmu,
because 'read' is correct raw event syntax with 0xead value.
To solve this issue we do following:
- check existing terms during rXXXX syntax processing
and make them priority in case of conflict
- allow pmu/r0x1234/ syntax to be able to specify conflicting
raw event (implemented in previous patch)
Also adding automated tests for this and perf_pmu__parse_cleanup
call to parse_events_terms, so the test gets properly cleaned up.
Reported-by: Jin Yao <[email protected]>
Signed-off-by: Jiri Olsa <[email protected]>
---
tools/perf/tests/parse-events.c | 37 ++++++++++++++++++++++++++++++++-
tools/perf/util/parse-events.c | 23 ++++++++++++++++++++
tools/perf/util/parse-events.h | 2 ++
tools/perf/util/parse-events.l | 19 ++++++++++-------
4 files changed, 72 insertions(+), 9 deletions(-)
diff --git a/tools/perf/tests/parse-events.c b/tools/perf/tests/parse-events.c
index 5aaddcb0058a..7f9f87a470c3 100644
--- a/tools/perf/tests/parse-events.c
+++ b/tools/perf/tests/parse-events.c
@@ -631,6 +631,34 @@ static int test__checkterms_simple(struct list_head *terms)
TEST_ASSERT_VAL("wrong val", term->val.num == 1);
TEST_ASSERT_VAL("wrong config", !strcmp(term->config, "umask"));
+ /*
+ * read
+ *
+ * The perf_pmu__test_parse_init injects 'read' term into
+ * perf_pmu_events_list, so 'read' is evaluated as read term
+ * and not as raw event with 'ead' hex value.
+ */
+ term = list_entry(term->list.next, struct parse_events_term, list);
+ TEST_ASSERT_VAL("wrong type term",
+ term->type_term == PARSE_EVENTS__TERM_TYPE_USER);
+ TEST_ASSERT_VAL("wrong type val",
+ term->type_val == PARSE_EVENTS__TERM_TYPE_NUM);
+ TEST_ASSERT_VAL("wrong val", term->val.num == 1);
+ TEST_ASSERT_VAL("wrong config", !strcmp(term->config, "read"));
+
+ /*
+ * r0xead
+ *
+ * To be still able to pass 'ead' value with 'r' syntax,
+ * we added support to parse 'r0xHEX' event.
+ */
+ term = list_entry(term->list.next, struct parse_events_term, list);
+ TEST_ASSERT_VAL("wrong type term",
+ term->type_term == PARSE_EVENTS__TERM_TYPE_CONFIG);
+ TEST_ASSERT_VAL("wrong type val",
+ term->type_val == PARSE_EVENTS__TERM_TYPE_NUM);
+ TEST_ASSERT_VAL("wrong val", term->val.num == 0xead);
+ TEST_ASSERT_VAL("wrong config", !term->config);
return 0;
}
@@ -1781,7 +1809,7 @@ struct terms_test {
static struct terms_test test__terms[] = {
[0] = {
- .str = "config=10,config1,config2=3,umask=1",
+ .str = "config=10,config1,config2=3,umask=1,read,r0xead",
.check = test__checkterms_simple,
},
};
@@ -1841,6 +1869,13 @@ static int test_term(struct terms_test *t)
INIT_LIST_HEAD(&terms);
+ /*
+ * The perf_pmu__test_parse_init prepares perf_pmu_events_list
+ * which gets freed in parse_events_terms.
+ */
+ if (perf_pmu__test_parse_init())
+ return -1;
+
ret = parse_events_terms(&terms, t->str);
if (ret) {
pr_debug("failed to parse terms '%s', err %d\n",
diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
index e88e4c7a2a9a..254f02a7fb0d 100644
--- a/tools/perf/util/parse-events.c
+++ b/tools/perf/util/parse-events.c
@@ -2019,6 +2019,27 @@ static void perf_pmu__parse_init(void)
perf_pmu__parse_cleanup();
}
+int perf_pmu__test_parse_init(void)
+{
+ struct perf_pmu_event_symbol *list;
+
+ list = malloc(sizeof(*list) * 1);
+ if (!list)
+ return -ENOMEM;
+
+ list->type = PMU_EVENT_SYMBOL;
+ list->symbol = strdup("read");
+
+ if (!list->symbol) {
+ free(list);
+ return -ENOMEM;
+ }
+
+ perf_pmu_events_list = list;
+ perf_pmu_events_list_num = 1;
+ return 0;
+}
+
enum perf_pmu_event_symbol_type
perf_pmu__parse_check(const char *name)
{
@@ -2080,6 +2101,8 @@ int parse_events_terms(struct list_head *terms, const char *str)
int ret;
ret = parse_events__scanner(str, &parse_state);
+ perf_pmu__parse_cleanup();
+
if (!ret) {
list_splice(parse_state.terms, terms);
zfree(&parse_state.terms);
diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h
index f0010095fc8c..00cde7d2e30c 100644
--- a/tools/perf/util/parse-events.h
+++ b/tools/perf/util/parse-events.h
@@ -261,4 +261,6 @@ static inline bool is_sdt_event(char *str __maybe_unused)
}
#endif /* HAVE_LIBELF_SUPPORT */
+int perf_pmu__test_parse_init(void);
+
#endif /* __PERF_PARSE_EVENTS_H */
diff --git a/tools/perf/util/parse-events.l b/tools/perf/util/parse-events.l
index 44c85fea5d00..3ca5fd2829ca 100644
--- a/tools/perf/util/parse-events.l
+++ b/tools/perf/util/parse-events.l
@@ -41,14 +41,6 @@ static int value(yyscan_t scanner, int base)
return __value(yylval, text, base, PE_VALUE);
}
-static int raw(yyscan_t scanner)
-{
- YYSTYPE *yylval = parse_events_get_lval(scanner);
- char *text = parse_events_get_text(scanner);
-
- return __value(yylval, text + 1, 16, PE_RAW);
-}
-
static int str(yyscan_t scanner, int token)
{
YYSTYPE *yylval = parse_events_get_lval(scanner);
@@ -72,6 +64,17 @@ static int str(yyscan_t scanner, int token)
return token;
}
+static int raw(yyscan_t scanner)
+{
+ YYSTYPE *yylval = parse_events_get_lval(scanner);
+ char *text = parse_events_get_text(scanner);
+
+ if (perf_pmu__parse_check(text) == PMU_EVENT_SYMBOL)
+ return str(scanner, PE_NAME);
+
+ return __value(yylval, text + 1, 16, PE_RAW);
+}
+
static bool isbpf_suffix(char *text)
{
int len = strlen(text);
--
2.25.4
On Sat, Jul 25, 2020 at 5:20 AM Jiri Olsa <[email protected]> wrote:
>
> Adding support to specify raw event with 'r0<HEX>' syntax within
> pmu term syntax like:
>
> -e cpu/r0xdead/
>
> It will be used to specify raw events in cases where they conflict
> with real pmu terms, like 'read', which is valid raw event syntax,
> but also a possible pmu term name as reported by Jin Yao.
>
> Reported-by: Jin Yao <[email protected]>
> Signed-off-by: Jiri Olsa <[email protected]>
Acked-by: Ian Rogers <[email protected]>
> ---
> tools/perf/Documentation/perf-list.txt | 1 +
> tools/perf/tests/parse-events.c | 5 +++++
> tools/perf/util/parse-events.l | 1 +
> 3 files changed, 7 insertions(+)
>
> diff --git a/tools/perf/Documentation/perf-list.txt b/tools/perf/Documentation/perf-list.txt
> index 376a50b3452d..10ed539a8859 100644
> --- a/tools/perf/Documentation/perf-list.txt
> +++ b/tools/perf/Documentation/perf-list.txt
> @@ -119,6 +119,7 @@ It's also possible to use pmu syntax:
>
> perf record -e r1a8 -a sleep 1
> perf record -e cpu/r1a8/ ...
> + perf record -e cpu/r0x1a8/ ...
>
> You should refer to the processor specific documentation for getting these
> details. Some of them are referenced in the SEE ALSO section below.
> diff --git a/tools/perf/tests/parse-events.c b/tools/perf/tests/parse-events.c
> index 895188b63f96..5aaddcb0058a 100644
> --- a/tools/perf/tests/parse-events.c
> +++ b/tools/perf/tests/parse-events.c
> @@ -1766,6 +1766,11 @@ static struct evlist_test test__events_pmu[] = {
> .check = test__checkevent_raw_pmu,
> .id = 4,
> },
> + {
> + .name = "software/r0x1a/",
> + .check = test__checkevent_raw_pmu,
> + .id = 4,
> + },
> };
>
> struct terms_test {
> diff --git a/tools/perf/util/parse-events.l b/tools/perf/util/parse-events.l
> index 56912c9641f5..44c85fea5d00 100644
> --- a/tools/perf/util/parse-events.l
> +++ b/tools/perf/util/parse-events.l
> @@ -293,6 +293,7 @@ percore { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_PERCORE); }
> aux-output { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_AUX_OUTPUT); }
> aux-sample-size { return term(yyscanner, PARSE_EVENTS__TERM_TYPE_AUX_SAMPLE_SIZE); }
> r{num_raw_hex} { return raw(yyscanner); }
> +r0x{num_raw_hex} { return raw(yyscanner); }
> , { return ','; }
> "/" { BEGIN(INITIAL); return '/'; }
> {name_minus} { return str(yyscanner, PE_NAME); }
> --
> 2.25.4
>
On Sat, Jul 25, 2020 at 5:20 AM Jiri Olsa <[email protected]> wrote:
>
> Jin Yao reported issue with possible conflict between raw
> events and term values in pmu event syntax.
>
> Currently following syntax is resolved as raw event with
> 0xead value:
> uncore_imc_free_running/read/
>
> instead of using 'read' term from uncore_imc_free_running pmu,
> because 'read' is correct raw event syntax with 0xead value.
>
> To solve this issue we do following:
> - check existing terms during rXXXX syntax processing
> and make them priority in case of conflict
> - allow pmu/r0x1234/ syntax to be able to specify conflicting
> raw event (implemented in previous patch)
>
> Also adding automated tests for this and perf_pmu__parse_cleanup
> call to parse_events_terms, so the test gets properly cleaned up.
>
> Reported-by: Jin Yao <[email protected]>
> Signed-off-by: Jiri Olsa <[email protected]>
> ---
> tools/perf/tests/parse-events.c | 37 ++++++++++++++++++++++++++++++++-
> tools/perf/util/parse-events.c | 23 ++++++++++++++++++++
> tools/perf/util/parse-events.h | 2 ++
> tools/perf/util/parse-events.l | 19 ++++++++++-------
> 4 files changed, 72 insertions(+), 9 deletions(-)
>
> diff --git a/tools/perf/tests/parse-events.c b/tools/perf/tests/parse-events.c
> index 5aaddcb0058a..7f9f87a470c3 100644
> --- a/tools/perf/tests/parse-events.c
> +++ b/tools/perf/tests/parse-events.c
> @@ -631,6 +631,34 @@ static int test__checkterms_simple(struct list_head *terms)
> TEST_ASSERT_VAL("wrong val", term->val.num == 1);
> TEST_ASSERT_VAL("wrong config", !strcmp(term->config, "umask"));
>
> + /*
> + * read
> + *
> + * The perf_pmu__test_parse_init injects 'read' term into
> + * perf_pmu_events_list, so 'read' is evaluated as read term
> + * and not as raw event with 'ead' hex value.
> + */
> + term = list_entry(term->list.next, struct parse_events_term, list);
> + TEST_ASSERT_VAL("wrong type term",
> + term->type_term == PARSE_EVENTS__TERM_TYPE_USER);
> + TEST_ASSERT_VAL("wrong type val",
> + term->type_val == PARSE_EVENTS__TERM_TYPE_NUM);
> + TEST_ASSERT_VAL("wrong val", term->val.num == 1);
> + TEST_ASSERT_VAL("wrong config", !strcmp(term->config, "read"));
> +
> + /*
> + * r0xead
> + *
> + * To be still able to pass 'ead' value with 'r' syntax,
> + * we added support to parse 'r0xHEX' event.
> + */
> + term = list_entry(term->list.next, struct parse_events_term, list);
> + TEST_ASSERT_VAL("wrong type term",
> + term->type_term == PARSE_EVENTS__TERM_TYPE_CONFIG);
> + TEST_ASSERT_VAL("wrong type val",
> + term->type_val == PARSE_EVENTS__TERM_TYPE_NUM);
> + TEST_ASSERT_VAL("wrong val", term->val.num == 0xead);
> + TEST_ASSERT_VAL("wrong config", !term->config);
> return 0;
> }
>
> @@ -1781,7 +1809,7 @@ struct terms_test {
>
> static struct terms_test test__terms[] = {
> [0] = {
> - .str = "config=10,config1,config2=3,umask=1",
> + .str = "config=10,config1,config2=3,umask=1,read,r0xead",
> .check = test__checkterms_simple,
> },
> };
> @@ -1841,6 +1869,13 @@ static int test_term(struct terms_test *t)
>
> INIT_LIST_HEAD(&terms);
>
> + /*
> + * The perf_pmu__test_parse_init prepares perf_pmu_events_list
> + * which gets freed in parse_events_terms.
> + */
> + if (perf_pmu__test_parse_init())
> + return -1;
> +
> ret = parse_events_terms(&terms, t->str);
> if (ret) {
> pr_debug("failed to parse terms '%s', err %d\n",
> diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
> index e88e4c7a2a9a..254f02a7fb0d 100644
> --- a/tools/perf/util/parse-events.c
> +++ b/tools/perf/util/parse-events.c
> @@ -2019,6 +2019,27 @@ static void perf_pmu__parse_init(void)
> perf_pmu__parse_cleanup();
> }
>
> +int perf_pmu__test_parse_init(void)
> +{
> + struct perf_pmu_event_symbol *list;
> +
> + list = malloc(sizeof(*list) * 1);
> + if (!list)
> + return -ENOMEM;
> +
> + list->type = PMU_EVENT_SYMBOL;
> + list->symbol = strdup("read");
> +
> + if (!list->symbol) {
> + free(list);
> + return -ENOMEM;
> + }
> +
> + perf_pmu_events_list = list;
> + perf_pmu_events_list_num = 1;
> + return 0;
> +}
nit: It's easy to see in the test code why this is necessary, could
the function be moved there? If not perhaps add a function comment?
The test in the function name is quite load bearing.
> +
> enum perf_pmu_event_symbol_type
> perf_pmu__parse_check(const char *name)
> {
> @@ -2080,6 +2101,8 @@ int parse_events_terms(struct list_head *terms, const char *str)
> int ret;
>
> ret = parse_events__scanner(str, &parse_state);
> + perf_pmu__parse_cleanup();
> +
> if (!ret) {
> list_splice(parse_state.terms, terms);
> zfree(&parse_state.terms);
> diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h
> index f0010095fc8c..00cde7d2e30c 100644
> --- a/tools/perf/util/parse-events.h
> +++ b/tools/perf/util/parse-events.h
> @@ -261,4 +261,6 @@ static inline bool is_sdt_event(char *str __maybe_unused)
> }
> #endif /* HAVE_LIBELF_SUPPORT */
>
> +int perf_pmu__test_parse_init(void);
> +
> #endif /* __PERF_PARSE_EVENTS_H */
> diff --git a/tools/perf/util/parse-events.l b/tools/perf/util/parse-events.l
> index 44c85fea5d00..3ca5fd2829ca 100644
> --- a/tools/perf/util/parse-events.l
> +++ b/tools/perf/util/parse-events.l
> @@ -41,14 +41,6 @@ static int value(yyscan_t scanner, int base)
> return __value(yylval, text, base, PE_VALUE);
> }
>
> -static int raw(yyscan_t scanner)
> -{
> - YYSTYPE *yylval = parse_events_get_lval(scanner);
> - char *text = parse_events_get_text(scanner);
> -
> - return __value(yylval, text + 1, 16, PE_RAW);
> -}
> -
> static int str(yyscan_t scanner, int token)
> {
> YYSTYPE *yylval = parse_events_get_lval(scanner);
> @@ -72,6 +64,17 @@ static int str(yyscan_t scanner, int token)
> return token;
> }
>
> +static int raw(yyscan_t scanner)
> +{
> + YYSTYPE *yylval = parse_events_get_lval(scanner);
> + char *text = parse_events_get_text(scanner);
> +
> + if (perf_pmu__parse_check(text) == PMU_EVENT_SYMBOL)
> + return str(scanner, PE_NAME);
> +
> + return __value(yylval, text + 1, 16, PE_RAW);
> +}
> +
> static bool isbpf_suffix(char *text)
> {
> int len = strlen(text);
> --
> 2.25.4
>
On Sat, Jul 25, 2020 at 09:58:13AM -0700, Ian Rogers wrote:
SNIP
> > ret = parse_events_terms(&terms, t->str);
> > if (ret) {
> > pr_debug("failed to parse terms '%s', err %d\n",
> > diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
> > index e88e4c7a2a9a..254f02a7fb0d 100644
> > --- a/tools/perf/util/parse-events.c
> > +++ b/tools/perf/util/parse-events.c
> > @@ -2019,6 +2019,27 @@ static void perf_pmu__parse_init(void)
> > perf_pmu__parse_cleanup();
> > }
> >
> > +int perf_pmu__test_parse_init(void)
> > +{
> > + struct perf_pmu_event_symbol *list;
> > +
> > + list = malloc(sizeof(*list) * 1);
> > + if (!list)
> > + return -ENOMEM;
> > +
> > + list->type = PMU_EVENT_SYMBOL;
> > + list->symbol = strdup("read");
> > +
> > + if (!list->symbol) {
> > + free(list);
> > + return -ENOMEM;
> > + }
> > +
> > + perf_pmu_events_list = list;
> > + perf_pmu_events_list_num = 1;
> > + return 0;
> > +}
>
> nit: It's easy to see in the test code why this is necessary, could
> the function be moved there? If not perhaps add a function comment?
> The test in the function name is quite load bearing.
both perf_pmu_events_list/cnt are static and I'd like to keep it that
way, so the function needs to be in here.. I'll add comment explaining
this
jirka
Em Sat, Jul 25, 2020 at 09:40:55AM -0700, Ian Rogers escreveu:
> On Sat, Jul 25, 2020 at 5:20 AM Jiri Olsa <[email protected]> wrote:
> >
> > Adding support to specify raw event with 'r0<HEX>' syntax within
> > pmu term syntax like:
> >
> > -e cpu/r0xdead/
> >
> > It will be used to specify raw events in cases where they conflict
> > with real pmu terms, like 'read', which is valid raw event syntax,
> > but also a possible pmu term name as reported by Jin Yao.
> >
> > Reported-by: Jin Yao <[email protected]>
> > Signed-off-by: Jiri Olsa <[email protected]>
>
> Acked-by: Ian Rogers <[email protected]>
Thanks, applied.
- Arnaldo