Prefer sysfs/JSON events also when no PMU is provided
Ian Rogers
reviewerv8 cover Prefer sysfs/JSON events also when no PMU is provided
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the perf_event_attr?
The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events bus_cycles and cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistent cpu_cycles and avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.
v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change the perf_api_probe to not load all PMUs through scanning, specify a PMU when parsing events.
v7: Expand cover letter, fix a missed core_ok check in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.
v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.
v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.
v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.
Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.
v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perf perf_api_probe: Avoid scanning all PMUs, try software PMU first
tools/perf/builtin-record.c | 63 +++++++++++++++++++--- tools/perf/util/parse-events.c | 47 +++++++++++++---- tools/perf/util/parse-events.h | 3 +- tools/perf/util/parse-events.l | 90 ++++++++++++++++++-------------- tools/perf/util/parse-events.y | 85 ++++++++++++++++++++++-------- tools/perf/util/perf_api_probe.c | 27 +++++++--- tools/perf/util/pmu.c | 9 ++-- 7 files changed, 235 insertions(+), 89 deletions(-)
Signature · 1 line
2.49.0.777.g153de2bbd5-goog
Ian Rogers
reviewerPATCH v8 1/4 perf record: Skip don't fail for events that don't open
Whilst for many tools it is an expected behavior that failure to open a perf event is a failure, ARM decided to name PMU events the same as legacy events and then failed to rename such events on a server uncore SLC PMU. As perf's default behavior when no PMU is specified is to open the event on all PMUs that advertise/"have" the event, this yielded failures when trying to make the priority of legacy and sysfs/json events uniform - something requested by RISC-V and ARM. A legacy event user on ARM hardware may find their event opened on an uncore PMU which for perf record will fail. Arnaldo suggested skipping such events which this patch implements. Rather than have the skipping conditional on running on ARM, the skipping is done on all architectures as such a fundamental behavioral difference could lead to problems with tools built/depending on perf.
An example of perf record failing to open events on x86 is:
```
$ perf record -e data_read,cycles,LLC-prefetch-read -a sleep 0.1
Error:
Failure to open event 'data_read' on PMU 'uncore_imc_free_running_0' which will be removed.
The sys_perf_event_open() syscall returned with 22 (Invalid argument) for event (data_read).
"dmesg | grep -i perf" may provide additional information.
Error:
Failure to open event 'data_read' on PMU 'uncore_imc_free_running_1' which will be removed.
The sys_perf_event_open() syscall returned with 22 (Invalid argument) for event (data_read).
"dmesg | grep -i perf" may provide additional information.
Error:
Failure to open event 'LLC-prefetch-read' on PMU 'cpu' which will be removed.
The LLC-prefetch-read event is not supported.
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 2.188 MB perf.data (87 samples) ]
$ perf report --stats
Aggregated stats:
TOTAL events: 17255
MMAP events: 284 ( 1.6%)
COMM events: 1961 (11.4%)
EXIT events: 1 ( 0.0%)
FORK events: 1960 (11.4%)
SAMPLE events: 87 ( 0.5%)
MMAP2 events: 12836 (74.4%)
KSYMBOL events: 83 ( 0.5%)
BPF_EVENT events: 36 ( 0.2%)
FINISHED_ROUND events: 2 ( 0.0%)
ID_INDEX events: 1 ( 0.0%)
THREAD_MAP events: 1 ( 0.0%)
CPU_MAP events: 1 ( 0.0%)
TIME_CONV events: 1 ( 0.0%)
FINISHED_INIT events: 1 ( 0.0%)
cycles stats:
SAMPLE events: 87
```If all events fail to open then the perf record will fail:
``` $ perf record -e LLC-prefetch-read true Error: Failure to open event 'LLC-prefetch-read' on PMU 'cpu' which will be removed. The LLC-prefetch-read event is not supported. Error: Failure to open any events for recording ```
As an evlist may have dummy events that open when all command line events fail we ignore dummy events when detecting if at least some events open. This still permits the dummy event on its own to be used as a permission check:
``` $ perf record -e dummy true [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.046 MB perf.data ] ``` but allows failure when a dummy event is implicilty inserted or when there are insufficient permissions to open it: ``` $ perf record -e LLC-prefetch-read -a true Error: Failure to open event 'LLC-prefetch-read' on PMU 'cpu' which will be removed. The LLC-prefetch-read event is not supported. Error: Failure to open any events for recording ```
As the first parsed event in an evlist is marked as tracking, removing this event can remove tracking from the evlist, removing mmap events and breaking symbolization. To avoid this, if a tracking event is removed then the next event has tracking added.
The issue with legacy events is that on RISC-V they want the driver to not have mappings from legacy to non-legacy config encodings for each vendor/model due to size, complexity and difficulty to update. It was reported that on ARM Apple-M? CPUs the legacy mapping in the driver was broken and the sysfs/json events should always take precedent, however, it isn't clear this is still the case. It is the case that without working around this issue a legacy event like cycles without a PMU can encode differently than when specified with a PMU - the non-PMU version favoring legacy encodings, the PMU one avoiding legacy encodings. Legacy events are also case sensitive while sysfs/json events are not.
The patch removes events and then adjusts the idx value for each evsel. This is done so that the dense xyarrays used for file descriptors, etc. don't contain broken entries. As event opening happens relatively late in the record process, use of the idx value before the open will have become corrupted, so it is expected there are latent bugs hidden behind this change - the change is best effort. As the only vendor that has broken event names is ARM, this will principally effect ARM users. They will also experience warning messages like those above because of the uncore PMU advertising legacy event names.
Prior versions without adding the tracking data was:
--- tools/perf/builtin-record.c | 63 +++++++++++++++++++++++++++++++++---- 1 file changed, 57 insertions(+), 6 deletions(-)
tools/perf/builtin-record.c +57 -6
@@ -962,7 +962,6 @@ static int record__config_tracking_events(struct record *rec)*/ if (opts->target.initial_delay ||target__has_cpu (&opts->target) ||perf_pmus__num_core_pmus () >1 ) { -/* * User space tasks can migrate between CPUs, so when tracing * selected CPUs, sideband for all CPUs is still needed.
@@ -1367,8 +1366,23 @@ static int record__open(struct record *rec)struct perf_session *session = rec->session;struct record_opts *opts = &rec->opts;int rc =0 ; +bool skipped = false; +bool removed_tracking = false;evlist__for_each_entry (evlist, pos) { +if (removed_tracking) { +/* +* Normally the head of the list has tracking enabled +* for sideband data like mmaps. If this event is +* removed, make sure to add tracking to the next +* processed event. +*/ +if (!pos->tracking) { + pos->tracking = true; +evsel__config (pos, opts, &callchain_param); + } + removed_tracking = false; + } try_again:if (evsel__open (pos, pos->core.cpus, pos->core.threads) <0 ) {if (evsel__fallback (pos, &opts->target, errno, msg,sizeof (msg))) {
@@ -1382,15 +1396,52 @@ static int record__open(struct record *rec) pos =evlist__reset_weak_group (evlist, pos, true); goto try_again; } - rc = -errno;evsel__open_strerror (pos, &opts->target, errno, msg,sizeof (msg)); -ui__error ("%s\n" , msg); - goto out; +ui__error ("Failure to open event '%s' on PMU '%s' which will be removed.\n%s\n" , +evsel__name (pos),evsel__pmu_name (pos), msg); +if (pos->tracking) + removed_tracking = true; + pos->skippable = true; + skipped = true; + }else { + pos->supported = true; } - - pos->supported = true; } +if (skipped) { +struct evsel *tmp; +int idx =0 ; +bool evlist_empty = true; + +/* Remove evsels that failed to open and update indices. */ +evlist__for_each_entry_safe (evlist, tmp, pos) { +if (pos->skippable) { +evlist__remove (evlist, pos); +continue ; + } + +/* +* Note, dummy events may be command line parsed or +* added by the tool. We care about supporting `perf +* record -e dummy` which may be used as a permission +* check. Dummy events that are added to the command +* line and opened along with other events that fail, +* will still fail as if the dummy events were tool +* added events for the sake of code simplicity. +*/ +if (!evsel__is_dummy_event (pos)) + evlist_empty = false; + } +evlist__for_each_entry (evlist, pos) { + pos->core.idx = idx++; + } +/* If list is empty then fail. */ +if (evlist_empty) { +ui__error ("Failure to open any events for recording.\n" ); + rc =-1 ; + goto out; + } + }if (symbol_conf.kptr_restrict && !evlist__exclude_kernel (evlist)) {pr_warning ( "WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted,\n"
Signature · 1 line
2.49.0.777.g153de2bbd5-goog
Ian Rogers
reviewerPATCH v8 2/4 perf parse-events: Reapply "Prefer sysfs/JSON hardware events over legacy"
In reply to Ian RogersOriginally posted and merged from:
https://lore.kernel.org/r/[email protected]
This reverts commit 4f1b067359ac8364cdb7f9fda41085fa85789d0f although the patch is now smaller due to related fixes being applied in commit 22a4db3c3603 ("perf evsel: Add alternate_hw_config and use in evsel__match").
The original commit message was:
It was requested that RISC-V be able to add events to the perf tool so the PMU driver didn't need to map legacy events to config encodings:
https://lore.kernel.org/lkml/[email protected]/
This change makes the priority of events specified without a PMU the same as those specified with a PMU, namely sysfs and JSON events are checked first before using the legacy encoding.
The hw_term is made more generic as a hardware_event that encodes a pair of string and int value, allowing parse_events_multi_pmu_add to fall back on a known encoding when the sysfs/JSON adding fails for core events. As this covers PE_VALUE_SYM_HW, that token is removed and related code simplified.
--- tools/perf/util/parse-events.c | 27 +++++++++--- tools/perf/util/parse-events.l | 76 +++++++++++++++++----------------- tools/perf/util/parse-events.y | 60 ++++++++++++++++++--------- 3 files changed, 99 insertions(+), 64 deletions(-)
tools/perf/util/parse-events.c +21 -6
@@ -1551,8 +1551,8 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state,struct list_head *list =NULL ;struct perf_pmu *pmu =NULL ;YYLTYPE *loc = loc_; -int ok =0 ; -const char *config; +int ok =0 , core_ok =0 ; +const char *tmp;struct parse_events_terms parsed_terms; *listp =NULL ;
@@ -1565,15 +1565,15 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state,return ret; } - config =strdup (event_name); -if (!config) + tmp =strdup (event_name); +if (!tmp) goto out_err;if (parse_events_term__num (&term,PARSE_EVENTS__TERM_TYPE_USER , - config,/*num=*/ 1 ,/*novalue=*/ true, + tmp,/*num=*/ 1 ,/*novalue=*/ true, loc,/*loc_val=*/ NULL ) <0 ) { -zfree (&config); +zfree (&tmp); goto out_err; }list_add_tail (&term->list, &parsed_terms.terms);
@@ -1604,6 +1604,8 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state,pr_debug ("%s -> %s/%s/\n" , event_name, pmu->name, sb.buf);strbuf_release (&sb); ok++; +if (pmu->is_core) + core_ok++; } }
@@ -1617,9 +1619,22 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state,pr_debug ("%s -> fake/%s/\n" , event_name, sb.buf);strbuf_release (&sb); ok++; + core_ok++; } } +if (hw_config !=PERF_COUNT_HW_MAX && !core_ok) { +/* +* The event wasn't found on core PMUs but it has a hardware +* config version to try. +*/ +if (!parse_events_add_numeric (parse_state, list, +PERF_TYPE_HARDWARE , hw_config, + const_parsed_terms, +/*wildcard=*/ true)) + ok++; + } + out_err:parse_events_terms__exit (&parsed_terms);if (ok)
tools/perf/util/parse-events.l +38 -38
107 lines not on this page · raw message
tools/perf/util/parse-events.y +40 -20
130 lines not on this page · raw message
Signature · 1 line
2.49.0.777.g153de2bbd5-goog
Ian Rogers
reviewerPATCH v8 3/4 perf parse-events: Allow software events to be terms
In reply to Ian RogersAllow legacy software events to be specified as, for example, software/cpu-clock/u for consistency in parsing with hardware events. Rename the type for hardware_event to legacy_event given its new dual role.
--- tools/perf/util/parse-events.c | 20 +++++++++++--- tools/perf/util/parse-events.h | 3 ++- tools/perf/util/parse-events.l | 48 ++++++++++++++++++++++------------ tools/perf/util/parse-events.y | 43 ++++++++++++++++++++++-------- tools/perf/util/pmu.c | 9 ++++--- 5 files changed, 86 insertions(+), 37 deletions(-)
tools/perf/util/parse-events.c +16 -4
@@ -805,6 +805,7 @@ const char *parse_events__term_type_str(enum parse_events__term_type term_type) [PARSE_EVENTS__TERM_TYPE_RAW] = "raw", [PARSE_EVENTS__TERM_TYPE_LEGACY_CACHE] = "legacy-cache", [PARSE_EVENTS__TERM_TYPE_HARDWARE] = "hardware", + [PARSE_EVENTS__TERM_TYPE_SOFTWARE] = "software", };if ((unsigned int )term_type >= __PARSE_EVENTS__TERM_TYPE_NR)return "unknown term" ;
@@ -854,6 +855,7 @@ config_term_avail(enum parse_events__term_type term_type, struct parse_events_ercase PARSE_EVENTS__TERM_TYPE_RAW :case PARSE_EVENTS__TERM_TYPE_LEGACY_CACHE :case PARSE_EVENTS__TERM_TYPE_HARDWARE : +case PARSE_EVENTS__TERM_TYPE_SOFTWARE :default :if (!err)return false;
@@ -985,6 +987,7 @@ do { \case PARSE_EVENTS__TERM_TYPE_USER :case PARSE_EVENTS__TERM_TYPE_LEGACY_CACHE :case PARSE_EVENTS__TERM_TYPE_HARDWARE : +case PARSE_EVENTS__TERM_TYPE_SOFTWARE :default : parse_events_error__handle(err, term->err_term, strdup(parse_events__term_type_str(term->type_term)),
@@ -1037,7 +1040,8 @@ static int config_term_pmu(struct perf_event_attr *attr, term->no_value = true; } } -if (term->type_term ==PARSE_EVENTS__TERM_TYPE_HARDWARE ) { +if (term->type_term ==PARSE_EVENTS__TERM_TYPE_HARDWARE || + term->type_term ==PARSE_EVENTS__TERM_TYPE_SOFTWARE ) {struct perf_pmu *pmu =perf_pmus__find_by_type (attr->type);if (!pmu) {
@@ -1057,10 +1061,15 @@ static int config_term_pmu(struct perf_event_attr *attr, term->no_value = true; term->alternate_hw_config = true; }else { - attr->type =PERF_TYPE_HARDWARE ; attr->config = term->val.num; -if (perf_pmus__supports_extended_type ()) - attr->config |= (__u64 )pmu->type <<PERF_PMU_TYPE_SHIFT ; +if (term->type_term ==PARSE_EVENTS__TERM_TYPE_HARDWARE ) { + attr->type =PERF_TYPE_HARDWARE ; +if (perf_pmus__supports_extended_type ()) + attr->config |= (__u64 )pmu->type <<PERF_PMU_TYPE_SHIFT ; + + }else { + attr->type =PERF_TYPE_SOFTWARE ; + } }return 0 ; }
@@ -1108,6 +1117,7 @@ static int config_term_tracepoint(struct perf_event_attr *attr,case PARSE_EVENTS__TERM_TYPE_RAW :case PARSE_EVENTS__TERM_TYPE_LEGACY_CACHE :case PARSE_EVENTS__TERM_TYPE_HARDWARE : +case PARSE_EVENTS__TERM_TYPE_SOFTWARE :default :if (err) { parse_events_error__handle(err, term->err_term,
@@ -1242,6 +1252,7 @@ do { \case PARSE_EVENTS__TERM_TYPE_RAW :case PARSE_EVENTS__TERM_TYPE_LEGACY_CACHE :case PARSE_EVENTS__TERM_TYPE_HARDWARE : +case PARSE_EVENTS__TERM_TYPE_SOFTWARE :default :break ; }
@@ -1296,6 +1307,7 @@ static int get_config_chgs(struct perf_pmu *pmu, struct parse_events_terms *headcase PARSE_EVENTS__TERM_TYPE_RAW :case PARSE_EVENTS__TERM_TYPE_LEGACY_CACHE :case PARSE_EVENTS__TERM_TYPE_HARDWARE : +case PARSE_EVENTS__TERM_TYPE_SOFTWARE :default :break ; }
tools/perf/util/parse-events.h +2 -1
@@ -80,7 +80,8 @@ enum parse_events__term_type {PARSE_EVENTS__TERM_TYPE_RAW ,PARSE_EVENTS__TERM_TYPE_LEGACY_CACHE ,PARSE_EVENTS__TERM_TYPE_HARDWARE , -#define __PARSE_EVENTS__TERM_TYPE_NR (PARSE_EVENTS__TERM_TYPE_HARDWARE + 1) +PARSE_EVENTS__TERM_TYPE_SOFTWARE , +#define __PARSE_EVENTS__TERM_TYPE_NR (PARSE_EVENTS__TERM_TYPE_SOFTWARE + 1) }; struct parse_events_term {
tools/perf/util/parse-events.l +31 -17
75 lines not on this page · raw message
tools/perf/util/parse-events.y +32 -11
110 lines not on this page · raw message
tools/perf/util/pmu.c +5 -4
@@ -1429,7 +1429,7 @@ static int pmu_config_term(const struct perf_pmu *pmu,break ;case PARSE_EVENTS__TERM_TYPE_USER :/* Not hardcoded. */ return -EINVAL ; - case PARSE_EVENTS__TERM_TYPE_NAME ... PARSE_EVENTS__TERM_TYPE_HARDWARE: + case PARSE_EVENTS__TERM_TYPE_NAME ... PARSE_EVENTS__TERM_TYPE_SOFTWARE:/* Skip non-config terms. */ break ;default :
@@ -1810,10 +1810,11 @@ int perf_pmu__for_each_format(struct perf_pmu *pmu, void *state, pmu_format_call/* * max-events and driver-config are missing above as are the internal -* types user, metric-id, raw, legacy cache and hardware. Assert against -* the enum parse_events__term_type so they are kept in sync. +* types user, metric-id, raw, legacy cache, hardware and +* software. Assert against the enum parse_events__term_type so they are +* kept in sync. */ -_Static_assert (ARRAY_SIZE (terms) == __PARSE_EVENTS__TERM_TYPE_NR -6 , +_Static_assert (ARRAY_SIZE (terms) == __PARSE_EVENTS__TERM_TYPE_NR -7 ,"perf_pmu__for_each_format()'s terms must be kept in sync with enum parse_events__term_type" );list_for_each_entry (format, &pmu->format, list) {perf_pmu_format__load (pmu, format);
Signature · 1 line
2.49.0.777.g153de2bbd5-goog
Ian Rogers
reviewerPATCH v8 4/4 perf perf_api_probe: Avoid scanning all PMUs, try software PMU first
In reply to Ian RogersAdd PMU names when geting events to probe. The first is only ever used but by not specifying a PMU name perf_api_probe will load all the PMUs.
--- tools/perf/util/perf_api_probe.c | 27 +++++++++++++++++++-------- 1 file changed, 19 insertions(+), 8 deletions(-)
tools/perf/util/perf_api_probe.c +19 -8
@@ -59,10 +59,10 @@ static int perf_do_probe_api(setup_probe_fn_t fn, struct perf_cpu cpu, const chastatic bool perf_probe_api (setup_probe_fn_t fn) { -const char *try[] = {"cycles:u" ,"instructions:u" ,"cpu-clock:u" ,NULL }; +struct perf_pmu *pmu;struct perf_cpu_map *cpus;struct perf_cpu cpu; -int ret, i =0 ; +int ret =0 ; cpus =perf_cpu_map__new_online_cpus ();if (!cpus)
@@ -70,12 +70,23 @@ static bool perf_probe_api(setup_probe_fn_t fn) cpu =perf_cpu_map__cpu (cpus,0 );perf_cpu_map__put (cpus); -do { - ret =perf_do_probe_api (fn, cpu, try[i++]); -if (!ret) -return true; - }while (ret == -EAGAIN && try[i]); - + ret =perf_do_probe_api (fn, cpu,"software/cpu-clock/u" ); +if (!ret) +return true; + + pmu =perf_pmus__scan_core (/*pmu=*/ NULL ); +if (pmu) { +const char *try[] = {"cycles" ,"instructions" ,NULL }; +char buf[256 ]; +int i =0 ; + +while (ret == -EAGAIN && try[i]) { +snprintf (buf,sizeof (buf),"%s/%s/u" , pmu->name, try[i++]); + ret =perf_do_probe_api (fn, cpu, buf); +if (!ret) +return true; + } + }return false; }
Signature · 1 line
2.49.0.777.g153de2bbd5-goog
Ian Rogers
reviewerQuoted 90 lines · Ian Rogers
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the
perf_event_attr?The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events
bus_cyclesand cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistentcpu_cyclesand avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change theperf_api_probeto not load all PMUs through scanning, specify a PMU when parsing events.v7: Expand cover letter, fix a missed
core_okcheck in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
Quoted 7 lines
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.
Ping. Thanks,
Ian
Quoted 19 lines
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perfperf_api_probe: Avoid scanning all PMUs, try software PMU firsttools/perf/builtin-record.c | 63 +++++++++++++++++++--- tools/perf/util/parse-events.c | 47 +++++++++++++---- tools/perf/util/parse-events.h | 3 +- tools/perf/util/parse-events.l | 90 ++++++++++++++++++-------------- tools/perf/util/parse-events.y | 85 ++++++++++++++++++++++-------- tools/perf/util/perf_api_probe.c | 27 +++++++--- tools/perf/util/pmu.c | 9 ++-- 7 files changed, 235 insertions(+), 89 deletions(-)--
2.49.0.777.g153de2bbd5-goog
Namhyung Kim
maintainerHi Ian,
Quoted 95 lines · Ian Rogers
Quoted 94 lines · Ian Rogers
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the
perf_event_attr?The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events
bus_cyclesand cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistentcpu_cyclesand avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change theperf_api_probeto not load all PMUs through scanning, specify a PMU when parsing events.v7: Expand cover letter, fix a missed
core_okcheck in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.
Quoted 13 lines
Quoted 3 lines
v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.Ping. Thanks,
IanQuoted 6 lines
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perfperf_api_probe: Avoid scanning all PMUs, try software PMU first
Sorry for the delay. But I think we wanted to move to this instead:
https://lore.kernel.org/linux-perf-users/[email protected]/
Thanks,
Namhyung
Quoted 13 lines
Quoted 13 lines
tools/perf/builtin-record.c | 63 +++++++++++++++++++--- tools/perf/util/parse-events.c | 47 +++++++++++++---- tools/perf/util/parse-events.h | 3 +- tools/perf/util/parse-events.l | 90 ++++++++++++++++++-------------- tools/perf/util/parse-events.y | 85 ++++++++++++++++++++++-------- tools/perf/util/perf_api_probe.c | 27 +++++++--- tools/perf/util/pmu.c | 9 ++-- 7 files changed, 235 insertions(+), 89 deletions(-)--
2.49.0.777.g153de2bbd5-goog
Ian Rogers
reviewerQuoted 113 lines · Namhyung Kim
Hi Ian,
Quoted 95 lines · Ian Rogers
Quoted 94 lines · Ian Rogers
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the
perf_event_attr?The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events
bus_cyclesand cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistentcpu_cyclesand avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change theperf_api_probeto not load all PMUs through scanning, specify a PMU when parsing events.v7: Expand cover letter, fix a missed
core_okcheck in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.Quoted 13 lines
Quoted 3 lines
v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.Ping. Thanks,
IanQuoted 6 lines
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perfperf_api_probe: Avoid scanning all PMUs, try software PMU first
Quoted 3 lines
Sorry for the delay. But I think we wanted to move to this instead:
Hi Namhyung,
The preference for sysfs/json over legacy was done as a bug fix and because ARM (Mark Rutland) argued strongly that it was the most sensible priority. Intel (Kan Liang) approved the change in priority. RISC-V have wanted this behavior as it enables the migration of event mappings from the driver to the tool. As the primary maintainer of the event parsing and metric code I prefer the priority as legacy events are weird, for example they aren't case insensitive in their naming. For example, on Intel with legacy events as the priority cpu-cycles would be a legacy event, but cpu-Cyles a sysfs one. On ARM cpu_cycles would be a sysfs event, but cpu-cycles a legacy one. A minor character difference and very different and imo surprising event encodings.
On your RFC thread Arnaldo and James said that legacy events somehow enabled a form of drill down. As event parsing is mapping a name to a perf_event_attr I completely don't see this as the mapping is opaque.
I strongly believe we need consistency. If perf stat -e cycles .. prints cpu_core/cycles/ as the event name on a hybrid Intel, then perf stat -e cpu_core/cycles/ .. should have the same perf_event_attr. Both patch series achieve this but this one does it with consistency, and from what I see it, the support of 3 vendors.
Thanks,
Ian
Quoted 16 lines
Thanks,
NamhyungQuoted 13 lines
Quoted 13 lines
tools/perf/builtin-record.c | 63 +++++++++++++++++++--- tools/perf/util/parse-events.c | 47 +++++++++++++---- tools/perf/util/parse-events.h | 3 +- tools/perf/util/parse-events.l | 90 ++++++++++++++++++-------------- tools/perf/util/parse-events.y | 85 ++++++++++++++++++++++-------- tools/perf/util/perf_api_probe.c | 27 +++++++--- tools/perf/util/pmu.c | 9 ++-- 7 files changed, 235 insertions(+), 89 deletions(-)--
2.49.0.777.g153de2bbd5-goog
Namhyung Kim
maintainerQuoted 120 lines · Ian Rogers
Quoted 113 lines · Namhyung Kim
Hi Ian,
Quoted 95 lines · Ian Rogers
Quoted 94 lines · Ian Rogers
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the
perf_event_attr?The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events
bus_cyclesand cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistentcpu_cyclesand avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change theperf_api_probeto not load all PMUs through scanning, specify a PMU when parsing events.v7: Expand cover letter, fix a missed
core_okcheck in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.Quoted 13 lines
Quoted 3 lines
v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.Ping. Thanks,
IanQuoted 6 lines
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perfperf_api_probe: Avoid scanning all PMUs, try software PMU firstQuoted 3 lines
Sorry for the delay. But I think we wanted to move to this instead:
Hi Namhyung,
Quoted 11 lines
The preference for sysfs/json over legacy was done as a bug fix and because ARM (Mark Rutland) argued strongly that it was the most sensible priority. Intel (Kan Liang) approved the change in priority. RISC-V have wanted this behavior as it enables the migration of event mappings from the driver to the tool. As the primary maintainer of the event parsing and metric code I prefer the priority as legacy events are weird, for example they aren't case insensitive in their naming. For example, on Intel with legacy events as the priority cpu-cycles would be a legacy event, but cpu-Cyles a sysfs one. On ARM
cpu_cycleswould be a sysfs event, but cpu-cycles a legacy one. A minor character difference and very different and imo surprising event encodings.
Yeah, but it has worked like that for a long time.
Quoted 4 lines
On your RFC thread Arnaldo and James said that legacy events somehow enabled a form of drill down. As event parsing is mapping a name to a
perf_event_attrI completely don't see this as the mapping is opaque.
Is it opaque? (I'd say it standard event rather than legacy event.) I think the mapping for the standard events are clearly defined.
Quoted 6 lines
I strongly believe we need consistency. If
perf stat -e cycles ..prints cpu_core/cycles/ as the event name on a hybrid Intel, thenperf stat -e cpu_core/cycles/ ..should have the sameperf_event_attr. Both patch series achieve this but this one does it with consistency, and from what I see it, the support of 3 vendors.
Right, it's not consistent. Maybe we need a different uniq event name for extended (standard) events. How about "cycles(cpu_core)"? I guess we don't want to add a space between the PMU and event names to avoid potential user impact when they parse the output.
Thanks,
Namhyung
Ian Rogers
reviewerQuoted 137 lines · Namhyung Kim
Quoted 120 lines · Ian Rogers
Quoted 113 lines · Namhyung Kim
Hi Ian,
Quoted 95 lines · Ian Rogers
Quoted 94 lines · Ian Rogers
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the
perf_event_attr?The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events
bus_cyclesand cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistentcpu_cyclesand avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change theperf_api_probeto not load all PMUs through scanning, specify a PMU when parsing events.v7: Expand cover letter, fix a missed
core_okcheck in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.Quoted 13 lines
Quoted 3 lines
v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.Ping. Thanks,
IanQuoted 6 lines
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perfperf_api_probe: Avoid scanning all PMUs, try software PMU firstQuoted 3 lines
Sorry for the delay. But I think we wanted to move to this instead:
Hi Namhyung,
Quoted 11 lines
The preference for sysfs/json over legacy was done as a bug fix and because ARM (Mark Rutland) argued strongly that it was the most sensible priority. Intel (Kan Liang) approved the change in priority. RISC-V have wanted this behavior as it enables the migration of event mappings from the driver to the tool. As the primary maintainer of the event parsing and metric code I prefer the priority as legacy events are weird, for example they aren't case insensitive in their naming. For example, on Intel with legacy events as the priority cpu-cycles would be a legacy event, but cpu-Cyles a sysfs one. On ARM
cpu_cycleswould be a sysfs event, but cpu-cycles a legacy one. A minor character difference and very different and imo surprising event encodings.Yeah, but it has worked like that for a long time.
Quoted 1 line
Quoted 6 lines
Quoted 3 lines
On your RFC thread Arnaldo and James said that legacy events somehow enabled a form of drill down. As event parsing is mapping a name to a
perf_event_attrI completely don't see this as the mapping is opaque.Is it opaque? (I'd say it standard event rather than legacy event.) I think the mapping for the standard events are clearly defined.
Which standard events? Going through them (abbreviated to avoid repetition):
- PERF_COUNT_HW_CPU_CYCLES, ok.
- PERF_COUNT_HW_INSTRUCTIONS, well does that include speculatively
executed instructions or not?
- ...
- PERF_COUNT_HW_STALLED_CYCLES_FRONTEND, what does this count on an
in order CPU?
- ...
The hardware cache events are far worse as things like LLC mean the L2 cache, however, it is far more typical for this to mean L3 these days. Standard and clearly defined, sorry absolutely not. They are a minefield of well intentioned event name components waiting to explode when a vendor inadvertently creates a combination that happens to match a combination perf thinks is significant.
There was a similar attempt for raw events where you can go r123 for the hex 123 event config, it was missed that rEAD is a valid hex raw event as well as a useful event name. The event parsing now has a lot of special handling to avoid exploding on this - and yes the priority is that sysfs/json has priority over the raw event encoding.
Quoted 11 lines
Quoted 6 lines
I strongly believe we need consistency. If
perf stat -e cycles ..prints cpu_core/cycles/ as the event name on a hybrid Intel, thenperf stat -e cpu_core/cycles/ ..should have the sameperf_event_attr. Both patch series achieve this but this one does it with consistency, and from what I see it, the support of 3 vendors.Right, it's not consistent. Maybe we need a different uniq event name for extended (standard) events. How about "
cycles(cpu_core)"? I guess we don't want to add a space between the PMU and event names to avoid potential user impact when they parse the output.
We could and it would very likely break tooling. The intent is that cpu-cycles matches cpu_core/cpu-cycles/ and cpu_atom/cpu-cycles/ and they are expected to all be the same event. Currently with the PMU they are sysfs encoded but without a PMU they are legacy encoded but printed (uniquified) as if they were with a PMU and sysfs encoded. This is misleading.
Thanks,
Ian
Quoted 3 lines
Thanks,
Namhyung
Namhyung Kim
maintainerQuoted 145 lines · Ian Rogers
Quoted 137 lines · Namhyung Kim
Quoted 120 lines · Ian Rogers
Quoted 113 lines · Namhyung Kim
Hi Ian,
Quoted 95 lines · Ian Rogers
Quoted 94 lines · Ian Rogers
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the
perf_event_attr?The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events
bus_cyclesand cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistentcpu_cyclesand avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change theperf_api_probeto not load all PMUs through scanning, specify a PMU when parsing events.v7: Expand cover letter, fix a missed
core_okcheck in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.Quoted 13 lines
Quoted 3 lines
v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.Ping. Thanks,
IanQuoted 6 lines
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perfperf_api_probe: Avoid scanning all PMUs, try software PMU firstQuoted 3 lines
Sorry for the delay. But I think we wanted to move to this instead:
Hi Namhyung,
Quoted 11 lines
The preference for sysfs/json over legacy was done as a bug fix and because ARM (Mark Rutland) argued strongly that it was the most sensible priority. Intel (Kan Liang) approved the change in priority. RISC-V have wanted this behavior as it enables the migration of event mappings from the driver to the tool. As the primary maintainer of the event parsing and metric code I prefer the priority as legacy events are weird, for example they aren't case insensitive in their naming. For example, on Intel with legacy events as the priority cpu-cycles would be a legacy event, but cpu-Cyles a sysfs one. On ARM
cpu_cycleswould be a sysfs event, but cpu-cycles a legacy one. A minor character difference and very different and imo surprising event encodings.Yeah, but it has worked like that for a long time.
Quoted 1 line
Quoted 6 lines
Quoted 3 lines
On your RFC thread Arnaldo and James said that legacy events somehow enabled a form of drill down. As event parsing is mapping a name to a
perf_event_attrI completely don't see this as the mapping is opaque.Is it opaque? (I'd say it standard event rather than legacy event.) I think the mapping for the standard events are clearly defined.
Quoted 8 lines
Which standard events? Going through them (abbreviated to avoid repetition):
-PERF_COUNT_HW_CPU_CYCLES, ok.
-PERF_COUNT_HW_INSTRUCTIONS, well does that include speculatively
executed instructions or not?
- ...
-PERF_COUNT_HW_STALLED_CYCLES_FRONTEND, what does this count on an
in order CPU?
- ...
I mean the mapping from event name to event encoding (PERF_COUNT_HW_...). I think the internal event mapping is the driver's business.
Quoted 7 lines
The hardware cache events are far worse as things like LLC mean the L2 cache, however, it is far more typical for this to mean L3 these days. Standard and clearly defined, sorry absolutely not. They are a minefield of well intentioned event name components waiting to explode when a vendor inadvertently creates a combination that happens to match a combination perf thinks is significant.
Again, it belongs to the driver.
Quoted 6 lines
There was a similar attempt for raw events where you can go r123 for the hex 123 event config, it was missed that rEAD is a valid hex raw event as well as a useful event name. The event parsing now has a lot of special handling to avoid exploding on this - and yes the priority is that sysfs/json has priority over the raw event encoding.
Agreed, the raw encoding can be a problem and it makes sense the sysfs/ JSON has the priority.
Quoted 13 lines
Quoted 8 lines
Quoted 7 lines
I strongly believe we need consistency. If
perf stat -e cycles ..prints cpu_core/cycles/ as the event name on a hybrid Intel, thenperf stat -e cpu_core/cycles/ ..should have the sameperf_event_attr. Both patch series achieve this but this one does it with consistency, and from what I see it, the support of 3 vendors.Quoted 4 lines
Right, it's not consistent. Maybe we need a different uniq event name for extended (standard) events. How about "
cycles(cpu_core)"? I guess we don't want to add a space between the PMU and event names to avoid potential user impact when they parse the output.
Quoted 6 lines
We could and it would very likely break tooling. The intent is that cpu-cycles matches cpu_core/cpu-cycles/ and cpu_atom/cpu-cycles/ and they are expected to all be the same event. Currently with the PMU they are sysfs encoded but without a PMU they are legacy encoded but printed (uniquified) as if they were with a PMU and sysfs encoded. This is misleading.
Hmm.. I don't know what's the correct way to handle this. Can we change it not to use extended standard events and to convert to sysfs events then?
Thanks,
Namhyung
Ian Rogers
reviewerQuoted 160 lines · Namhyung Kim
Quoted 145 lines · Ian Rogers
Quoted 137 lines · Namhyung Kim
Quoted 120 lines · Ian Rogers
Quoted 113 lines · Namhyung Kim
Hi Ian,
Quoted 95 lines · Ian Rogers
Quoted 94 lines · Ian Rogers
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the
perf_event_attr?The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events
bus_cyclesand cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistentcpu_cyclesand avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change theperf_api_probeto not load all PMUs through scanning, specify a PMU when parsing events.v7: Expand cover letter, fix a missed
core_okcheck in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.Quoted 13 lines
Quoted 3 lines
v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.Ping. Thanks,
IanQuoted 6 lines
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perfperf_api_probe: Avoid scanning all PMUs, try software PMU firstQuoted 3 lines
Sorry for the delay. But I think we wanted to move to this instead:
Hi Namhyung,
Quoted 11 lines
The preference for sysfs/json over legacy was done as a bug fix and because ARM (Mark Rutland) argued strongly that it was the most sensible priority. Intel (Kan Liang) approved the change in priority. RISC-V have wanted this behavior as it enables the migration of event mappings from the driver to the tool. As the primary maintainer of the event parsing and metric code I prefer the priority as legacy events are weird, for example they aren't case insensitive in their naming. For example, on Intel with legacy events as the priority cpu-cycles would be a legacy event, but cpu-Cyles a sysfs one. On ARM
cpu_cycleswould be a sysfs event, but cpu-cycles a legacy one. A minor character difference and very different and imo surprising event encodings.Yeah, but it has worked like that for a long time.
Quoted 1 line
Quoted 6 lines
Quoted 3 lines
On your RFC thread Arnaldo and James said that legacy events somehow enabled a form of drill down. As event parsing is mapping a name to a
perf_event_attrI completely don't see this as the mapping is opaque.Is it opaque? (I'd say it standard event rather than legacy event.) I think the mapping for the standard events are clearly defined.
Quoted 8 lines
Which standard events? Going through them (abbreviated to avoid repetition):
-PERF_COUNT_HW_CPU_CYCLES, ok.
-PERF_COUNT_HW_INSTRUCTIONS, well does that include speculatively
executed instructions or not?
- ...
-PERF_COUNT_HW_STALLED_CYCLES_FRONTEND, what does this count on an
in order CPU?
- ...I mean the mapping from event name to event encoding (
PERF_COUNT_HW_...). I think the internal event mapping is the driver's business.Quoted 1 line
Quoted 8 lines
Quoted 6 lines
The hardware cache events are far worse as things like LLC mean the L2 cache, however, it is far more typical for this to mean L3 these days. Standard and clearly defined, sorry absolutely not. They are a minefield of well intentioned event name components waiting to explode when a vendor inadvertently creates a combination that happens to match a combination perf thinks is significant.
Again, it belongs to the driver.
But we know there are broken drivers. That is why this started. The Apple-M core PMU had broken mappings and was reliant on being treated as an uncore PMU so that legacy encodings weren't used. By forcing the use of legacy encodings you'd break new perf tools on v6.6 kernels on Apple-M IIRC.
When an event is encoded in json it can have a description. The description can identify what the exact behaviors wrt speculation, what LLC means, etc. are. The legacy event names are a minefield and ambiguous.
Why are they a minefield? We import metrics from sources like the TMA spreadsheet:
https://github.com/intel/perfmon/blob/main/TMA_Metrics-full.csv
The events named in here are intended for use across a range of tools like vtune and emon. When we import event names in the perf build we assume they are case insensitive and match them by lower-casing them:
https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/pmu-events/jevents.py?h=perf-tools-next#n326
Legacy cache events are made up of components:
https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/util/parse-events.l?h=perf-tools-next#n258
``` lc_type (L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node) lc_op_result (load|loads|read|store|stores|write|prefetch|prefetches|speculative-read|speculative-load|refs|Reference|ops|access|misses|miss) ```
If legacy events have the priority then an event that accidentally combines these components can switch from using the event json that comes with the metric to using a legacy event that has a different behavior. It may also result in parse errors.
We are avoiding stepping on mines by luck currently and the approach of using legacy first is essentially error prone. (It also breaks Apple-M and is the opposite of what RISC-V have requested)
Quoted 30 lines
Quoted 6 lines
There was a similar attempt for raw events where you can go r123 for the hex 123 event config, it was missed that rEAD is a valid hex raw event as well as a useful event name. The event parsing now has a lot of special handling to avoid exploding on this - and yes the priority is that sysfs/json has priority over the raw event encoding.
Agreed, the raw encoding can be a problem and it makes sense the sysfs/ JSON has the priority.
Quoted 13 lines
Quoted 8 lines
Quoted 7 lines
I strongly believe we need consistency. If
perf stat -e cycles ..prints cpu_core/cycles/ as the event name on a hybrid Intel, thenperf stat -e cpu_core/cycles/ ..should have the sameperf_event_attr. Both patch series achieve this but this one does it with consistency, and from what I see it, the support of 3 vendors.Quoted 4 lines
Right, it's not consistent. Maybe we need a different uniq event name for extended (standard) events. How about "
cycles(cpu_core)"? I guess we don't want to add a space between the PMU and event names to avoid potential user impact when they parse the output.Quoted 6 lines
We could and it would very likely break tooling. The intent is that cpu-cycles matches cpu_core/cpu-cycles/ and cpu_atom/cpu-cycles/ and they are expected to all be the same event. Currently with the PMU they are sysfs encoded but without a PMU they are legacy encoded but printed (uniquified) as if they were with a PMU and sysfs encoded. This is misleading.
Quoted 3 lines
Hmm.. I don't know what's the correct way to handle this. Can we change it not to use extended standard events and to convert to sysfs events then?
Tbh, I think a lot of this is coming down to 1 event, cycles. It is the only unambiguous legacy event. If we make cycles only mean core PMU cycles (ie cpu_core, cpu_atom on hybrid, cpu on non-hybrid) then I think this is matching what you are expecting to happen. I mean we only match cycles as legacy first and everything else is sysfs/json first and legacy last. I think other legacy events cpu-cycles aren't special and don't get legacy first treatment. Having just 1 special legacy only event cycles is a lot less offensive to me than having say branch-prefetch-read also be legacy. We accept cycles is a minefield and special, but it is unique in this.
Thanks,
Ian
Quoted 3 lines
Thanks,
Namhyung
Namhyung Kim
maintainerQuoted 168 lines · Ian Rogers
Quoted 167 lines · Namhyung Kim
Quoted 145 lines · Ian Rogers
Quoted 137 lines · Namhyung Kim
Quoted 120 lines · Ian Rogers
Quoted 113 lines · Namhyung Kim
Hi Ian,
Quoted 95 lines · Ian Rogers
Quoted 94 lines · Ian Rogers
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the
perf_event_attr?The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events
bus_cyclesand cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistentcpu_cyclesand avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change theperf_api_probeto not load all PMUs through scanning, specify a PMU when parsing events.v7: Expand cover letter, fix a missed
core_okcheck in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.Quoted 13 lines
Quoted 3 lines
v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.Ping. Thanks,
IanQuoted 6 lines
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perfperf_api_probe: Avoid scanning all PMUs, try software PMU firstQuoted 3 lines
Sorry for the delay. But I think we wanted to move to this instead:
Hi Namhyung,
Quoted 11 lines
The preference for sysfs/json over legacy was done as a bug fix and because ARM (Mark Rutland) argued strongly that it was the most sensible priority. Intel (Kan Liang) approved the change in priority. RISC-V have wanted this behavior as it enables the migration of event mappings from the driver to the tool. As the primary maintainer of the event parsing and metric code I prefer the priority as legacy events are weird, for example they aren't case insensitive in their naming. For example, on Intel with legacy events as the priority cpu-cycles would be a legacy event, but cpu-Cyles a sysfs one. On ARM
cpu_cycleswould be a sysfs event, but cpu-cycles a legacy one. A minor character difference and very different and imo surprising event encodings.Yeah, but it has worked like that for a long time.
Quoted 1 line
Quoted 6 lines
Quoted 3 lines
On your RFC thread Arnaldo and James said that legacy events somehow enabled a form of drill down. As event parsing is mapping a name to a
perf_event_attrI completely don't see this as the mapping is opaque.Is it opaque? (I'd say it standard event rather than legacy event.) I think the mapping for the standard events are clearly defined.
Quoted 8 lines
Which standard events? Going through them (abbreviated to avoid repetition):
-PERF_COUNT_HW_CPU_CYCLES, ok.
-PERF_COUNT_HW_INSTRUCTIONS, well does that include speculatively
executed instructions or not?
- ...
-PERF_COUNT_HW_STALLED_CYCLES_FRONTEND, what does this count on an
in order CPU?
- ...I mean the mapping from event name to event encoding (
PERF_COUNT_HW_...). I think the internal event mapping is the driver's business.Quoted 7 lines
The hardware cache events are far worse as things like LLC mean the L2 cache, however, it is far more typical for this to mean L3 these days. Standard and clearly defined, sorry absolutely not. They are a minefield of well intentioned event name components waiting to explode when a vendor inadvertently creates a combination that happens to match a combination perf thinks is significant.
Quoted 7 lines
Quoted 1 line
Again, it belongs to the driver.
But we know there are broken drivers. That is why this started. The Apple-M core PMU had broken mappings and was reliant on being treated as an uncore PMU so that legacy encodings weren't used. By forcing the use of legacy encodings you'd break new perf tools on v6.6 kernels on Apple-M IIRC.
Didn't they use sysfs events (i.e. '-e PMU/EVENT/' form) always?
Quoted 5 lines
When an event is encoded in json it can have a description. The description can identify what the exact behaviors wrt speculation, what LLC means, etc. are. The legacy event names are a minefield and ambiguous.
Right, sometimes it's not 100% clear and descriptions would be nice.
Quoted 10 lines
Why are they a minefield? We import metrics from sources like the TMA spreadsheet:
https://github.com/intel/perfmon/blob/main/TMA_Metrics-full.csv
The events named in here are intended for use across a range of tools like vtune and emon. When we import event names in the perf build we assume they are case insensitive and match them by lower-casing them:
https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/pmu-events/jevents.py?h=perf-tools-next#n326
Legacy cache events are made up of components:
https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/util/parse-events.l?h=perf-tools-next#n258
Quoted 4 lines
``` lc_type (L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node) lc_op_result (load|loads|read|store|stores|write|prefetch|prefetches|speculative-read|speculative-load|refs|Reference|ops|access|misses|miss) ```
Oh, looks like it's too permissive. Do we have speculative-load in the standard cache events?
Quoted 8 lines
If legacy events have the priority then an event that accidentally combines these components can switch from using the event json that comes with the metric to using a legacy event that has a different behavior. It may also result in parse errors.
We are avoiding stepping on mines by luck currently and the approach of using legacy first is essentially error prone. (It also breaks Apple-M and is the opposite of what RISC-V have requested)
Hmm.. ok. I feel like cache events should have lower priority at least. If it's all about metrics, maybe we can add something like a new parse option like parse_evens_option_args.prefer_json to have different priority settings and use it for metrics?
Quoted 35 lines
Quoted 31 lines
Quoted 7 lines
There was a similar attempt for raw events where you can go r123 for the hex 123 event config, it was missed that rEAD is a valid hex raw event as well as a useful event name. The event parsing now has a lot of special handling to avoid exploding on this - and yes the priority is that sysfs/json has priority over the raw event encoding.
Agreed, the raw encoding can be a problem and it makes sense the sysfs/ JSON has the priority.
Quoted 13 lines
Quoted 8 lines
Quoted 7 lines
I strongly believe we need consistency. If
perf stat -e cycles ..prints cpu_core/cycles/ as the event name on a hybrid Intel, thenperf stat -e cpu_core/cycles/ ..should have the sameperf_event_attr. Both patch series achieve this but this one does it with consistency, and from what I see it, the support of 3 vendors.Quoted 4 lines
Right, it's not consistent. Maybe we need a different uniq event name for extended (standard) events. How about "
cycles(cpu_core)"? I guess we don't want to add a space between the PMU and event names to avoid potential user impact when they parse the output.Quoted 6 lines
We could and it would very likely break tooling. The intent is that cpu-cycles matches cpu_core/cpu-cycles/ and cpu_atom/cpu-cycles/ and they are expected to all be the same event. Currently with the PMU they are sysfs encoded but without a PMU they are legacy encoded but printed (uniquified) as if they were with a PMU and sysfs encoded. This is misleading.
Quoted 3 lines
Hmm.. I don't know what's the correct way to handle this. Can we change it not to use extended standard events and to convert to sysfs events then?
Quoted 10 lines
Tbh, I think a lot of this is coming down to 1 event, cycles. It is the only unambiguous legacy event. If we make cycles only mean core PMU cycles (ie
cpu_core,cpu_atomon hybrid, cpu on non-hybrid) then I think this is matching what you are expecting to happen. I mean we only match cycles as legacy first and everything else is sysfs/json first and legacy last. I think other legacy events cpu-cycles aren't special and don't get legacy first treatment. Having just 1 special legacy only event cycles is a lot less offensive to me than having say branch-prefetch-read also be legacy. We accept cycles is a minefield and special, but it is unique in this.
Well.. I think cycles is an alias to cpu-cycles so they are the same. So it's not about special casing 'cycles'. Any PMU may add conflicting event names to sysfs (which is discouraging though).
Thanks,
Namhyung
Ian Rogers
reviewerQuoted 212 lines · Namhyung Kim
Quoted 168 lines · Ian Rogers
Quoted 167 lines · Namhyung Kim
Quoted 145 lines · Ian Rogers
Quoted 137 lines · Namhyung Kim
Quoted 120 lines · Ian Rogers
Quoted 113 lines · Namhyung Kim
Hi Ian,
Quoted 95 lines · Ian Rogers
Quoted 94 lines · Ian Rogers
At the RISC-V summit the topic of avoiding event data being in the RISC-V PMU kernel driver came up. There is a preference for sysfs/JSON events being the priority when no PMU is provided so that legacy events maybe supported via json. Originally Mark Rutland also expressed at LPC 2023 that doing this would resolve bugs on ARM Apple M? processors, but James Clark more recently tested this and believes the driver issues there may not have existed or have been resolved. In any case, it is inconsistent that with a PMU event names avoid legacy encodings, but when wildcarding PMUs (ie without a PMU with the event name) the legacy encodings have priority.
The situation is further inconsistent as legacy events are case sensitive, so on Intel that provides a sysfs instructions event, the instructions event without a PMU and lowercase is legacy while with uppercase letters it matches with sysfs which is case insensitive. Are there legacy events with upper case letters? Yes there are, the cache ones mix case freely:
L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node
meaning LLC that means L2 (which is wrong) both match as part of a legacy cache name but llc and l2 would only match sysfs/json events. The whole thing just points at the ridiculous nature of legacy events and why we'd want them to be preffered I don't know. Why should case of a letter or having a PMU prefix impact the encoding in the
perf_event_attr?The patch doing this work was reverted in a v6.10 release candidate as, even though the patch was posted for weeks and had been on linux-next for weeks without issue, Linus was in the habit of using explicit legacy events with unsupported precision options on his Neoverse-N1. This machine has SLC PMU events for bus and CPU cycles where ARM decided to call the events
bus_cyclesand cycles, the latter being also a legacy event name. ARM haven't renamed the cycles event to a more consistentcpu_cyclesand avoided the problem. With these changes the problematic event will now be skipped, a large warning produced, and perf record will continue for the other PMU events. This solution was proposed by Arnaldo.v8: Change removing of failed to open events that are tracking so that
the tracking moves to the next event. Make software events able to specified with a PMU. Change theperf_api_probeto not load all PMUs through scanning, specify a PMU when parsing events.v7: Expand cover letter, fix a missed
core_okcheck in the v6
rebase. Note, as with v6 there is an alternate series that prioritizes legacy events but that is silly and I'd prefer we didn't do it.v6: Rebase of v5 (dropping already merged patches):
https://lore.kernel.org/lkml/[email protected]/
that unusually had an RFC posted for it:
https://lore.kernel.org/lkml/[email protected]/
Note, this patch conflicts/contradicts:
https://lore.kernel.org/lkml/[email protected]/
that I posted so that we could either consistently prioritize sysfs/json (these patches) or legacy events (the other patches). That lack of event printing and encoding inconsistency is most prominent in the encoding of events like "instructions" which on hybrid are reported as "cpu_core/instructions/" but "instructions" before these patches gets a legacy encoding while "cpu_core/instructions/" gets a sysfs/json encoding. These patches make "instructions" always get a sysfs/json encoding while the alternate patches make it always get a legacy encoding.v5: Follow Namhyung's suggestion and ignore the case where command
line dummy events fail to open alongside other events that all fail to open. Note, the Tested-by tags are left on the series as v4 and v5 were changing an error case that doesn't occur in testing but was manually tested by myself.v4: Rework the no events opening change from v3 to make it handle
multiple dummy events. Sadly an evlist isn't empty if it just contains dummy events as the dummy event may be used with "perf record -e dummy .." as a way to determine whether permission issues exist. Other software events like cpu-clock would suffice for this, but the using dummy genie has left the bottle.Another problem is that we appear to have an excessive number of dummy events added, for example, we can likely avoid a dummy event and add sideband data to the original event. For auxtrace more dummy events may be opened too. Anyway, this has led to the approach taken in patch 3 where the number of dummy parsed events is computed. If the number of removed/failing-to-open non-dummy events matches the number of non-dummy events then we want to fail, but only if there are no parsed dummy events or if there was one then it must have opened. The math here is hard to read, but passes my manual testing.
v3: Make no events opening for perf record a failure as suggested by
James Clark and Aditya Bodkhe <[email protected]>. Also, rebase.Quoted 13 lines
Quoted 3 lines
v2: Rebase and add tested-by tags from James Clark, Leo Yan and Atish
Patra who have tested on RISC-V and ARM CPUs, including the problem case from before.Ping. Thanks,
IanQuoted 6 lines
Ian Rogers (4):
perf record: Skip don't fail for events that don't open perf parse-events: Reapply "Prefer sysfs/JSON hardware events over
legacy"
perf parse-events: Allow software events to be terms
perfperf_api_probe: Avoid scanning all PMUs, try software PMU firstQuoted 3 lines
Sorry for the delay. But I think we wanted to move to this instead:
Hi Namhyung,
Quoted 11 lines
The preference for sysfs/json over legacy was done as a bug fix and because ARM (Mark Rutland) argued strongly that it was the most sensible priority. Intel (Kan Liang) approved the change in priority. RISC-V have wanted this behavior as it enables the migration of event mappings from the driver to the tool. As the primary maintainer of the event parsing and metric code I prefer the priority as legacy events are weird, for example they aren't case insensitive in their naming. For example, on Intel with legacy events as the priority cpu-cycles would be a legacy event, but cpu-Cyles a sysfs one. On ARM
cpu_cycleswould be a sysfs event, but cpu-cycles a legacy one. A minor character difference and very different and imo surprising event encodings.Yeah, but it has worked like that for a long time.
Quoted 1 line
Quoted 6 lines
Quoted 3 lines
On your RFC thread Arnaldo and James said that legacy events somehow enabled a form of drill down. As event parsing is mapping a name to a
perf_event_attrI completely don't see this as the mapping is opaque.Is it opaque? (I'd say it standard event rather than legacy event.) I think the mapping for the standard events are clearly defined.
Quoted 8 lines
Which standard events? Going through them (abbreviated to avoid repetition):
-PERF_COUNT_HW_CPU_CYCLES, ok.
-PERF_COUNT_HW_INSTRUCTIONS, well does that include speculatively
executed instructions or not?
- ...
-PERF_COUNT_HW_STALLED_CYCLES_FRONTEND, what does this count on an
in order CPU?
- ...I mean the mapping from event name to event encoding (
PERF_COUNT_HW_...). I think the internal event mapping is the driver's business.Quoted 7 lines
The hardware cache events are far worse as things like LLC mean the L2 cache, however, it is far more typical for this to mean L3 these days. Standard and clearly defined, sorry absolutely not. They are a minefield of well intentioned event name components waiting to explode when a vendor inadvertently creates a combination that happens to match a combination perf thinks is significant.
Quoted 7 lines
Quoted 1 line
Again, it belongs to the driver.
But we know there are broken drivers. That is why this started. The Apple-M core PMU had broken mappings and was reliant on being treated as an uncore PMU so that legacy encodings weren't used. By forcing the use of legacy encodings you'd break new perf tools on v6.6 kernels on Apple-M IIRC.
Didn't they use sysfs events (i.e. '-e PMU/EVENT/' form) always?
Quoted 5 lines
When an event is encoded in json it can have a description. The description can identify what the exact behaviors wrt speculation, what LLC means, etc. are. The legacy event names are a minefield and ambiguous.
Right, sometimes it's not 100% clear and descriptions would be nice.
Quoted 10 lines
Why are they a minefield? We import metrics from sources like the TMA spreadsheet:
https://github.com/intel/perfmon/blob/main/TMA_Metrics-full.csv
The events named in here are intended for use across a range of tools like vtune and emon. When we import event names in the perf build we assume they are case insensitive and match them by lower-casing them:
https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/pmu-events/jevents.py?h=perf-tools-next#n326
Legacy cache events are made up of components:
https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/util/parse-events.l?h=perf-tools-next#n258Quoted 4 lines
``` lc_type (L1-dcache|l1-d|l1d|L1-data|L1-icache|l1-i|l1i|L1-instruction|LLC|L2|dTLB|d-tlb|Data-TLB|iTLB|i-tlb|Instruction-TLB|branch|branches|bpu|btb|bpc|node) lc_op_result (load|loads|read|store|stores|write|prefetch|prefetches|speculative-read|speculative-load|refs|Reference|ops|access|misses|miss) ```Oh, looks like it's too permissive. Do we have speculative-load in the standard cache events?
Quoted 5 lines
If legacy events have the priority then an event that accidentally combines these components can switch from using the event json that comes with the metric to using a legacy event that has a different behavior. It may also result in parse errors.
Quoted 8 lines
Quoted 3 lines
We are avoiding stepping on mines by luck currently and the approach of using legacy first is essentially error prone. (It also breaks Apple-M and is the opposite of what RISC-V have requested)
Hmm.. ok. I feel like cache events should have lower priority at least. If it's all about metrics, maybe we can add something like a new parse option like parse_evens_option_args.prefer_json to have different priority settings and use it for metrics?
There are only 10 legacy non-cache events, but after cycles the meaning is ambiguous and likely often just not implemented. Saying the meaning is known because of the driver, I'm not aware of any driver giving descriptions.
For a flag, the issue is that we prefer json when there is a PMU but not when it is wildcarded. So when is the flag used or not?
My eventual hope is that we can standardize on sysfs encoding. So we translate json to sysfs, then compress the files/directories and
access them built into the perf tool this way. Then, as with say
addr2line or objdump, we have a priority list like:
"sysfs(/sys),sysfs(inbuilt),legacy". Then we could allow "sysfs(/home/ian/myevents),sysfs(/sys),sysfs(inbuilt),legacy" so users
can manually add events and metrics. A priority list for the behavior today would be something like "legacy(no-pmu),sysfs(/sys),sysfs(inbuilt),legacy(with-pmu)". If we
have hardware cache events at a low priority may it is "legacy(no-pmu+type_hardware),sysfs(/sys),sysfs(inbuilt),legacy(with-pmu+type_hw_cache)" what this series is trying to achieve is just "sysfs(/sys),sysfs(inbuilt),legacy" and the alternate series is "legacy,sysfs(/sys),sysfs(inbuilt)".
On x86 events are 70% of the binary size but just compressing the json shows that this could be halved.
Making the lookup use a list and user supplied directories may allow us to write tests that create a hybrid looking sysfs to test on non-hybrid, etc.
Quoted 46 lines
Quoted 35 lines
Quoted 31 lines
Quoted 7 lines
There was a similar attempt for raw events where you can go r123 for the hex 123 event config, it was missed that rEAD is a valid hex raw event as well as a useful event name. The event parsing now has a lot of special handling to avoid exploding on this - and yes the priority is that sysfs/json has priority over the raw event encoding.
Agreed, the raw encoding can be a problem and it makes sense the sysfs/ JSON has the priority.
Quoted 13 lines
Quoted 8 lines
Quoted 7 lines
I strongly believe we need consistency. If
perf stat -e cycles ..prints cpu_core/cycles/ as the event name on a hybrid Intel, thenperf stat -e cpu_core/cycles/ ..should have the sameperf_event_attr. Both patch series achieve this but this one does it with consistency, and from what I see it, the support of 3 vendors.Quoted 4 lines
Right, it's not consistent. Maybe we need a different uniq event name for extended (standard) events. How about "
cycles(cpu_core)"? I guess we don't want to add a space between the PMU and event names to avoid potential user impact when they parse the output.Quoted 6 lines
We could and it would very likely break tooling. The intent is that cpu-cycles matches cpu_core/cpu-cycles/ and cpu_atom/cpu-cycles/ and they are expected to all be the same event. Currently with the PMU they are sysfs encoded but without a PMU they are legacy encoded but printed (uniquified) as if they were with a PMU and sysfs encoded. This is misleading.
Quoted 3 lines
Hmm.. I don't know what's the correct way to handle this. Can we change it not to use extended standard events and to convert to sysfs events then?
Quoted 10 lines
Tbh, I think a lot of this is coming down to 1 event, cycles. It is the only unambiguous legacy event. If we make cycles only mean core PMU cycles (ie
cpu_core,cpu_atomon hybrid, cpu on non-hybrid) then I think this is matching what you are expecting to happen. I mean we only match cycles as legacy first and everything else is sysfs/json first and legacy last. I think other legacy events cpu-cycles aren't special and don't get legacy first treatment. Having just 1 special legacy only event cycles is a lot less offensive to me than having say branch-prefetch-read also be legacy. We accept cycles is a minefield and special, but it is unique in this.
Quoted 3 lines
Well.. I think cycles is an alias to cpu-cycles so they are the same. So it's not about special casing 'cycles'. Any PMU may add conflicting event names to sysfs (which is discouraging though).
So it has been standard for Intel to have a cpu-cycles sysfs event:
``` $ cat /sys/bus/event_source/devices/cpu/events/cpu-cycles event=0x3c ```
I'm only aware of ARM using cycles as an event name and they also use cpu_cycles, I wish they'd pick a lane (preferably not the one that conflicts with a legacy event name).
I'm not particularly keen on having cycles not be wildcard (other than for core PMUs) and meaning the legacy encoding. The issue is that the legacy encoding will break broken PMUs like on older Apple-M linux, and what to do in the RISC-V case?
I can understand not wanting to change what users see and I think that may be the only expedient way to do it. I don't think anyone cares about the cpu-cycles event name and on Intel it seems they'd prefer the sysfs encoding, hence advertising the event.
Thanks,
Ian
Quoted 3 lines
Thanks,
Namhyung