From: Leo Yan <leo.yan@linaro.org>
To: James Clark <james.clark@arm.com>
Cc: linux-perf-users@vger.kernel.org, coresight@lists.linaro.org,
shy828301@gmail.com, denik@google.com,
Mathieu Poirier <mathieu.poirier@linaro.org>,
Suzuki K Poulose <suzuki.poulose@arm.com>,
Mike Leach <mike.leach@linaro.org>,
John Garry <john.g.garry@oracle.com>,
Will Deacon <will@kernel.org>,
Peter Zijlstra <peterz@infradead.org>,
Ingo Molnar <mingo@redhat.com>,
Arnaldo Carvalho de Melo <acme@kernel.org>,
Mark Rutland <mark.rutland@arm.com>,
Alexander Shishkin <alexander.shishkin@linux.intel.com>,
Jiri Olsa <jolsa@kernel.org>, Namhyung Kim <namhyung@kernel.org>,
Ian Rogers <irogers@google.com>,
Adrian Hunter <adrian.hunter@intel.com>,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH 4/7] perf: cs-etm: Validate options after applying themperf_pmu__format_bits
Date: Thu, 27 Apr 2023 23:12:28 +0800 [thread overview]
Message-ID: <20230427151228.GA152865@leoy-huanghe> (raw)
In-Reply-To: <20230424134748.228137-5-james.clark@arm.com>
Hi James,
On Mon, Apr 24, 2023 at 02:47:44PM +0100, James Clark wrote:
> Currently the cs_etm_set_option() function both validates and applies
> the config options. Because it's only called when they are added
> automatically, there are some paths where the user can apply the option
> on the command line and skip the validation. By moving it to the end it
> covers both cases.
>
> Also, options don't need to be re-applied anyway, Perf handles parsing
> and applying the config terms automatically.
>
> Signed-off-by: James Clark <james.clark@arm.com>
> ---
> tools/perf/arch/arm/util/cs-etm.c | 152 +++++++++++++-----------------
> 1 file changed, 68 insertions(+), 84 deletions(-)
>
> diff --git a/tools/perf/arch/arm/util/cs-etm.c b/tools/perf/arch/arm/util/cs-etm.c
> index f9b9ebf7fffc..af0a2400c655 100644
> --- a/tools/perf/arch/arm/util/cs-etm.c
> +++ b/tools/perf/arch/arm/util/cs-etm.c
> @@ -69,21 +69,29 @@ static const char * const metadata_ete_ro[] = {
> static bool cs_etm_is_etmv4(struct auxtrace_record *itr, int cpu);
> static bool cs_etm_is_ete(struct auxtrace_record *itr, int cpu);
>
> -static int cs_etm_set_context_id(struct auxtrace_record *itr,
> - struct evsel *evsel, int cpu)
> +static int cs_etm_validate_context_id(struct auxtrace_record *itr,
> + struct evsel *evsel, int cpu)
> {
> - struct cs_etm_recording *ptr;
> - struct perf_pmu *cs_etm_pmu;
> + struct cs_etm_recording *ptr =
> + container_of(itr, struct cs_etm_recording, itr);
> + struct perf_pmu *cs_etm_pmu = ptr->cs_etm_pmu;
> char path[PATH_MAX];
> - int err = -EINVAL;
> + int err;
> u32 val;
> - u64 contextid;
> + u64 contextid =
> + evsel->core.attr.config &
> + (perf_pmu__format_bits(&cs_etm_pmu->format, "contextid1") |
> + perf_pmu__format_bits(&cs_etm_pmu->format, "contextid2"));
Seems to me, this would break backward compability.
The old kernel (before 5.11) doesn't provide 'contextid1' and
'contextid2', so we always check the entry 'contextid' rather than
'contextid1' and 'contextid2'.
With this change, if a kernel doesn't contain 'contextid1' and
'contextid2' formats, will perf tool never trace for contexid?
Thanks,
Leo
>
> - ptr = container_of(itr, struct cs_etm_recording, itr);
> - cs_etm_pmu = ptr->cs_etm_pmu;
> + if (!contextid)
> + return 0;
>
> - if (!cs_etm_is_etmv4(itr, cpu))
> - goto out;
> + /* Not supported in etmv3 */
> + if (!cs_etm_is_etmv4(itr, cpu)) {
> + pr_err("%s: contextid not supported in ETMv3, disable with %s/contextid=0/\n",
> + CORESIGHT_ETM_PMU_NAME, CORESIGHT_ETM_PMU_NAME);
> + return -EINVAL;
> + }
>
> /* Get a handle on TRCIDR2 */
> snprintf(path, PATH_MAX, "cpu%d/%s",
> @@ -92,27 +100,13 @@ static int cs_etm_set_context_id(struct auxtrace_record *itr,
>
> /* There was a problem reading the file, bailing out */
> if (err != 1) {
> - pr_err("%s: can't read file %s\n",
> - CORESIGHT_ETM_PMU_NAME, path);
> - goto out;
> + pr_err("%s: can't read file %s\n", CORESIGHT_ETM_PMU_NAME,
> + path);
> + return err;
> }
>
> - /* User has configured for PID tracing, respects it. */
> - contextid = evsel->core.attr.config &
> - (BIT(ETM_OPT_CTXTID) | BIT(ETM_OPT_CTXTID2));
> -
> - /*
> - * If user doesn't configure the contextid format, parse PMU format and
> - * enable PID tracing according to the "contextid" format bits:
> - *
> - * If bit ETM_OPT_CTXTID is set, trace CONTEXTIDR_EL1;
> - * If bit ETM_OPT_CTXTID2 is set, trace CONTEXTIDR_EL2.
> - */
> - if (!contextid)
> - contextid = perf_pmu__format_bits(&cs_etm_pmu->format,
> - "contextid");
> -
> - if (contextid & BIT(ETM_OPT_CTXTID)) {
> + if (contextid &
> + perf_pmu__format_bits(&cs_etm_pmu->format, "contextid1")) {
> /*
> * TRCIDR2.CIDSIZE, bit [9-5], indicates whether contextID
> * tracing is supported:
> @@ -122,14 +116,14 @@ static int cs_etm_set_context_id(struct auxtrace_record *itr,
> */
> val = BMVAL(val, 5, 9);
> if (!val || val != 0x4) {
> - pr_err("%s: CONTEXTIDR_EL1 isn't supported\n",
> - CORESIGHT_ETM_PMU_NAME);
> - err = -EINVAL;
> - goto out;
> + pr_err("%s: CONTEXTIDR_EL1 isn't supported, disable with %s/contextid1=0/\n",
> + CORESIGHT_ETM_PMU_NAME, CORESIGHT_ETM_PMU_NAME);
> + return -EINVAL;
> }
> }
>
> - if (contextid & BIT(ETM_OPT_CTXTID2)) {
> + if (contextid &
> + perf_pmu__format_bits(&cs_etm_pmu->format, "contextid2")) {
> /*
> * TRCIDR2.VMIDOPT[30:29] != 0 and
> * TRCIDR2.VMIDSIZE[14:10] == 0b00100 (32bit virtual contextid)
> @@ -138,35 +132,34 @@ static int cs_etm_set_context_id(struct auxtrace_record *itr,
> * Any value of VMIDSIZE >= 4 (i.e, > 32bit) is fine for us.
> */
> if (!BMVAL(val, 29, 30) || BMVAL(val, 10, 14) < 4) {
> - pr_err("%s: CONTEXTIDR_EL2 isn't supported\n",
> - CORESIGHT_ETM_PMU_NAME);
> - err = -EINVAL;
> - goto out;
> + pr_err("%s: CONTEXTIDR_EL2 isn't supported, disable with %s/contextid2=0/\n",
> + CORESIGHT_ETM_PMU_NAME, CORESIGHT_ETM_PMU_NAME);
> + return -EINVAL;
> }
> }
>
> - /* All good, let the kernel know */
> - evsel->core.attr.config |= contextid;
> - err = 0;
> -
> -out:
> - return err;
> + return 0;
> }
>
> -static int cs_etm_set_timestamp(struct auxtrace_record *itr,
> - struct evsel *evsel, int cpu)
> +static int cs_etm_validate_timestamp(struct auxtrace_record *itr,
> + struct evsel *evsel, int cpu)
> {
> - struct cs_etm_recording *ptr;
> - struct perf_pmu *cs_etm_pmu;
> + struct cs_etm_recording *ptr =
> + container_of(itr, struct cs_etm_recording, itr);
> + struct perf_pmu *cs_etm_pmu = ptr->cs_etm_pmu;
> char path[PATH_MAX];
> - int err = -EINVAL;
> + int err;
> u32 val;
>
> - ptr = container_of(itr, struct cs_etm_recording, itr);
> - cs_etm_pmu = ptr->cs_etm_pmu;
> + if (!(evsel->core.attr.config &
> + perf_pmu__format_bits(&cs_etm_pmu->format, "timestamp")))
> + return 0;
>
> - if (!cs_etm_is_etmv4(itr, cpu))
> - goto out;
> + if (!cs_etm_is_etmv4(itr, cpu)) {
> + pr_err("%s: timestamp not supported in ETMv3, disable with %s/timestamp=0/\n",
> + CORESIGHT_ETM_PMU_NAME, CORESIGHT_ETM_PMU_NAME);
> + return -EINVAL;
> + }
>
> /* Get a handle on TRCIRD0 */
> snprintf(path, PATH_MAX, "cpu%d/%s",
> @@ -177,7 +170,7 @@ static int cs_etm_set_timestamp(struct auxtrace_record *itr,
> if (err != 1) {
> pr_err("%s: can't read file %s\n",
> CORESIGHT_ETM_PMU_NAME, path);
> - goto out;
> + return err;
> }
>
> /*
> @@ -189,24 +182,21 @@ static int cs_etm_set_timestamp(struct auxtrace_record *itr,
> */
> val &= GENMASK(28, 24);
> if (!val) {
> - err = -EINVAL;
> - goto out;
> + return -EINVAL;
> }
>
> - /* All good, let the kernel know */
> - evsel->core.attr.config |= (1 << ETM_OPT_TS);
> - err = 0;
> -
> -out:
> - return err;
> + return 0;
> }
>
> -#define ETM_SET_OPT_CTXTID (1 << 0)
> -#define ETM_SET_OPT_TS (1 << 1)
> -#define ETM_SET_OPT_MASK (ETM_SET_OPT_CTXTID | ETM_SET_OPT_TS)
> -
> -static int cs_etm_set_option(struct auxtrace_record *itr,
> - struct evsel *evsel, u32 option)
> +/*
> + * Check whether the requested timestamp and contextid options should be
> + * available on all requested CPUs and if not, tell the user how to override.
> + * The kernel will silently disable any unavailable options so a warning here
> + * first is better. In theory the kernel could still disable the option for
> + * some other reason so this is best effort only.
> + */
> +static int cs_etm_validate_config(struct auxtrace_record *itr,
> + struct evsel *evsel)
> {
> int i, err = -EINVAL;
> struct perf_cpu_map *event_cpus = evsel->evlist->core.user_requested_cpus;
> @@ -220,18 +210,11 @@ static int cs_etm_set_option(struct auxtrace_record *itr,
> !perf_cpu_map__has(online_cpus, cpu))
> continue;
>
> - if (option & BIT(ETM_OPT_CTXTID)) {
> - err = cs_etm_set_context_id(itr, evsel, i);
> - if (err)
> - goto out;
> - }
> - if (option & BIT(ETM_OPT_TS)) {
> - err = cs_etm_set_timestamp(itr, evsel, i);
> - if (err)
> - goto out;
> - }
> - if (option & ~(BIT(ETM_OPT_CTXTID) | BIT(ETM_OPT_TS)))
> - /* Nothing else is currently supported */
> + err = cs_etm_validate_context_id(itr, evsel, i);
> + if (err)
> + goto out;
> + err = cs_etm_validate_timestamp(itr, evsel, i);
> + if (err)
> goto out;
> }
>
> @@ -447,10 +430,10 @@ static int cs_etm_recording_options(struct auxtrace_record *itr,
> * when a context switch happened.
> */
> if (!perf_cpu_map__empty(cpus)) {
> - err = cs_etm_set_option(itr, cs_etm_evsel,
> - BIT(ETM_OPT_CTXTID) | BIT(ETM_OPT_TS));
> - if (err)
> - goto out;
> + cs_etm_evsel->core.attr.config |=
> + perf_pmu__format_bits(&cs_etm_pmu->format, "timestamp");
> + cs_etm_evsel->core.attr.config |=
> + perf_pmu__format_bits(&cs_etm_pmu->format, "contextid");
> }
>
> /* Add dummy event to keep tracking */
> @@ -466,6 +449,7 @@ static int cs_etm_recording_options(struct auxtrace_record *itr,
> if (!perf_cpu_map__empty(cpus))
> evsel__set_sample_bit(evsel, TIME);
>
> + err = cs_etm_validate_config(itr, cs_etm_evsel);
> out:
> return err;
> }
> --
> 2.34.1
>
next prev parent reply other threads:[~2023-04-27 15:13 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-04-24 13:47 [PATCH 0/7] perf: cs-etm: Fixes around timestamped and timeless decoding James Clark
2023-04-24 13:47 ` [PATCH 1/7] perf: cs-etm: Fix timeless decode mode detection James Clark
2023-04-24 15:14 ` Suzuki K Poulose
2023-04-26 5:42 ` Denis Nikitin
2023-04-24 13:47 ` [PATCH 2/7] perf tools: Add util function for overriding user set config values James Clark
2023-04-24 15:36 ` Adrian Hunter
2023-04-24 16:43 ` James Clark
2023-04-24 17:45 ` Arnaldo Carvalho de Melo
2023-04-25 11:03 ` James Clark
2023-04-24 13:47 ` [PATCH 3/7] perf: cs-etm: Don't test full_auxtrace because it's always set James Clark
2023-04-24 13:47 ` [PATCH 4/7] perf: cs-etm: Validate options after applying them James Clark
2023-04-27 15:12 ` Leo Yan [this message]
2023-04-27 15:52 ` [PATCH 4/7] perf: cs-etm: Validate options after applying themperf_pmu__format_bits James Clark
2023-04-27 22:10 ` Leo Yan
2023-04-28 12:33 ` James Clark
2023-05-01 7:34 ` Leo Yan
2023-04-24 13:47 ` [PATCH 5/7] perf: cs-etm: Allow user to override timestamp and contextid settings James Clark
2023-04-24 13:47 ` [PATCH 6/7] perf: cs-etm: Use bool type for boolean values James Clark
2023-04-24 13:47 ` [PATCH 7/7] perf: cs-etm: Add separate decode paths for timeless and per-thread modes James Clark
2023-04-26 3:16 ` [PATCH 0/7] perf: cs-etm: Fixes around timestamped and timeless decoding Denis Nikitin
2023-04-26 16:06 ` Yang Shi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20230427151228.GA152865@leoy-huanghe \
--to=leo.yan@linaro.org \
--cc=acme@kernel.org \
--cc=adrian.hunter@intel.com \
--cc=alexander.shishkin@linux.intel.com \
--cc=coresight@lists.linaro.org \
--cc=denik@google.com \
--cc=irogers@google.com \
--cc=james.clark@arm.com \
--cc=john.g.garry@oracle.com \
--cc=jolsa@kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=mark.rutland@arm.com \
--cc=mathieu.poirier@linaro.org \
--cc=mike.leach@linaro.org \
--cc=mingo@redhat.com \
--cc=namhyung@kernel.org \
--cc=peterz@infradead.org \
--cc=shy828301@gmail.com \
--cc=suzuki.poulose@arm.com \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).