linux-perf-users.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Ilkka Koskinen <ilkka@os.amperecomputing.com>
To: Leo Yan <leo.yan@arm.com>
Cc: Ilkka Koskinen <ilkka@os.amperecomputing.com>,
	 John Garry <john.g.garry@oracle.com>,
	Will Deacon <will@kernel.org>,
	 James Clark <james.clark@linaro.org>,
	Mike Leach <mike.leach@linaro.org>,  Leo Yan <leo.yan@linux.dev>,
	Peter Zijlstra <peterz@infradead.org>,
	 Ingo Molnar <mingo@redhat.com>,
	Arnaldo Carvalho de Melo <acme@kernel.org>,
	 Namhyung Kim <namhyung@kernel.org>,
	Mark Rutland <mark.rutland@arm.com>,
	 Alexander Shishkin <alexander.shishkin@linux.intel.com>,
	 Jiri Olsa <jolsa@kernel.org>, Ian Rogers <irogers@google.com>,
	 Adrian Hunter <adrian.hunter@intel.com>,
	 "Liang, Kan" <kan.liang@linux.intel.com>,
	 Graham Woodward <graham.woodward@arm.com>,
	 linux-arm-kernel@lists.infradead.org,
	linux-perf-users@vger.kernel.org,  linux-kernel@vger.kernel.org
Subject: Re: [PATCH v3 1/2] perf arm-spe: Prepare for adding data source packet implementations for other cores
Date: Thu, 7 Nov 2024 15:55:41 -0800 (PST)	[thread overview]
Message-ID: <1c7db0d2-d08c-2d9-2976-30d1e9d8f2f4@os.amperecomputing.com> (raw)
In-Reply-To: <20241107152542.GC47850@e132581.arm.com>


Hi Leo,

On Thu, 7 Nov 2024, Leo Yan wrote:
> Hi Ilkka,
>
> This is a good refactoring for me.  Just several minor comments.
>
> On Wed, Nov 06, 2024 at 07:37:39PM +0000, Ilkka Koskinen wrote:
>>
>> Split Data Source Packet handling to prepare adding support for
>> other implementations.
>>
>> Signed-off-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
>> ---
>>  tools/perf/util/arm-spe.c | 65 ++++++++++++++++++++++++---------------
>>  1 file changed, 40 insertions(+), 25 deletions(-)
>>
>> diff --git a/tools/perf/util/arm-spe.c b/tools/perf/util/arm-spe.c
>> index dbf13f47879c..b222557cc27a 100644
>> --- a/tools/perf/util/arm-spe.c
>> +++ b/tools/perf/util/arm-spe.c
>> @@ -103,6 +103,18 @@ struct arm_spe_queue {
>>         u32                             flags;
>>  };
>>
>> +struct data_src {
>> +       struct midr_range midr_range;
>> +       void (*ds_synth)(const struct arm_spe_record *record,
>> +                        union perf_mem_data_src *data_src);
>> +};
>
> The naming is a bit mess. The data structure and the parameter both
> are called "data_src", though this will not cause building issue.
>
> How about rename the structure "data_src" to "data_source_handle" or
> "data_source_class"?

Yeah, I forgot to revisit the naming part. I like "data"source_handle",
that should clarify it quite a bit.


>
> For the "midr_range" field, I'd like to change it to a pointer:
>
>  struct midr_range *midr_range;
>
> Please see below comments, which will present the reason for defining
> it as a pointer.
>
>> +
>> +#define DS(range, func)                        \
>> +       {                                               \
>> +               .midr_range = range,                    \
>> +               .ds_synth = arm_spe__synth_##func,      \
>> +       }
>> +
>>  static void arm_spe_dump(struct arm_spe *spe __maybe_unused,
>>                          unsigned char *buf, size_t len)
>>  {
>> @@ -430,19 +442,6 @@ static int arm_spe__synth_instruction_sample(struct arm_spe_queue *speq,
>>         return arm_spe_deliver_synth_event(spe, speq, event, &sample);
>>  }
>>
>> -static const struct midr_range common_ds_encoding_cpus[] = {
>> -       MIDR_ALL_VERSIONS(MIDR_CORTEX_A720),
>> -       MIDR_ALL_VERSIONS(MIDR_CORTEX_A725),
>> -       MIDR_ALL_VERSIONS(MIDR_CORTEX_X1C),
>> -       MIDR_ALL_VERSIONS(MIDR_CORTEX_X3),
>> -       MIDR_ALL_VERSIONS(MIDR_CORTEX_X925),
>> -       MIDR_ALL_VERSIONS(MIDR_NEOVERSE_N1),
>> -       MIDR_ALL_VERSIONS(MIDR_NEOVERSE_N2),
>> -       MIDR_ALL_VERSIONS(MIDR_NEOVERSE_V1),
>> -       MIDR_ALL_VERSIONS(MIDR_NEOVERSE_V2),
>> -       {},
>> -};
>
> We can keep this data structure.  For Ampere CPUs, you can add a new
> data structure:
>
>  static const struct midr_range ampereone_ds_encoding_cpus[] = {
>      MIDR_ALL_VERSIONS(MIDR_AMPERE1A),
>      {},
>  };

Sounds good to me. I change all those.

Cheers, Ilkka


>
>> -
>>  static void arm_spe__sample_flags(struct arm_spe_queue *speq)
>>  {
>>         const struct arm_spe_record *record = &speq->decoder->record;
>> @@ -532,6 +531,19 @@ static void arm_spe__synth_data_source_common(const struct arm_spe_record *recor
>>         }
>>  }
>>
>> +static const struct data_src data_sources[] = {
>> +       DS(MIDR_ALL_VERSIONS(MIDR_CORTEX_A720), data_source_common),
>> +       DS(MIDR_ALL_VERSIONS(MIDR_CORTEX_A725), data_source_common),
>> +       DS(MIDR_ALL_VERSIONS(MIDR_CORTEX_X1C), data_source_common),
>> +       DS(MIDR_ALL_VERSIONS(MIDR_CORTEX_X3), data_source_common),
>> +       DS(MIDR_ALL_VERSIONS(MIDR_CORTEX_X925), data_source_common),
>> +       DS(MIDR_ALL_VERSIONS(MIDR_NEOVERSE_N1), data_source_common),
>> +       DS(MIDR_ALL_VERSIONS(MIDR_NEOVERSE_N2), data_source_common),
>> +       DS(MIDR_ALL_VERSIONS(MIDR_NEOVERSE_V1), data_source_common),
>> +       DS(MIDR_ALL_VERSIONS(MIDR_NEOVERSE_V2), data_source_common),
>> +       {},
>> +};
>> +
>
> As a result, we can simplify the structure as:
>
>  static const struct data_src data_sources[] = {
>         DS(common_ds_encoding_cpus, data_source_common),
>         DS(ampereone_ds_encoding_cpus, data_source_ampereone),
>  };
>
>>  static void arm_spe__synth_memory_level(const struct arm_spe_record *record,
>>                                         union perf_mem_data_src *data_src)
>>  {
>> @@ -555,12 +567,14 @@ static void arm_spe__synth_memory_level(const struct arm_spe_record *record,
>>                 data_src->mem_lvl |= PERF_MEM_LVL_REM_CCE1;
>>  }
>>
>> -static bool arm_spe__is_common_ds_encoding(struct arm_spe_queue *speq)
>> +static bool arm_spe__synth_ds(struct arm_spe_queue *speq,
>> +                             const struct arm_spe_record *record,
>> +                             union perf_mem_data_src *data_src)
>>  {
>>         struct arm_spe *spe = speq->spe;
>> -       bool is_in_cpu_list;
>> +       const struct data_src *src = data_sources;
>>         u64 *metadata = NULL;
>> -       u64 midr = 0;
>> +       u64 midr;
>>
>>         /* Metadata version 1 assumes all CPUs are the same (old behavior) */
>>         if (spe->metadata_ver == 1) {
>> @@ -592,18 +606,21 @@ static bool arm_spe__is_common_ds_encoding(struct arm_spe_queue *speq)
>>                 midr = metadata[ARM_SPE_CPU_MIDR];
>>         }
>>
>> -       is_in_cpu_list = is_midr_in_range_list(midr, common_ds_encoding_cpus);
>> -       if (is_in_cpu_list)
>> -               return true;
>> -       else
>> -               return false;
>> +       while (src->midr_range.model) {
>> +               if (is_midr_in_range(midr, &src->midr_range)) {
>> +                       src->ds_synth(record, data_src);
>> +                       return true;
>> +               }
>> +               src++;
>> +       }
>
> Here we can traverse the 'data_sources' array:
>
>          for (i = 0; i < ARRAY_SIZE(data_sources); i++) {
>               if (is_midr_in_range(midr, data_sources[i]->midr_range)) {
>                   ...
>               }
>          }
>
> Thanks,
> Leo
>
>> +
>> +       return false;
>>  }
>>
>>  static u64 arm_spe__synth_data_source(struct arm_spe_queue *speq,
>>                                       const struct arm_spe_record *record)
>>  {
>>         union perf_mem_data_src data_src = { .mem_op = PERF_MEM_OP_NA };
>> -       bool is_common = arm_spe__is_common_ds_encoding(speq);
>>
>>         if (record->op & ARM_SPE_OP_LD)
>>                 data_src.mem_op = PERF_MEM_OP_LOAD;
>> @@ -612,9 +629,7 @@ static u64 arm_spe__synth_data_source(struct arm_spe_queue *speq,
>>         else
>>                 return 0;
>>
>> -       if (is_common)
>> -               arm_spe__synth_data_source_common(record, &data_src);
>> -       else
>> +       if (!arm_spe__synth_ds(speq, record, &data_src))
>>                 arm_spe__synth_memory_level(record, &data_src);
>>
>>         if (record->type & (ARM_SPE_TLB_ACCESS | ARM_SPE_TLB_MISS)) {
>> --
>> 2.47.0
>>
>>
>

  reply	other threads:[~2024-11-07 23:55 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-11-06 19:37 [PATCH v3 0/2] perf arm-spe: Add support for SPE Data Source packet on AmpereOne Ilkka Koskinen
2024-11-06 19:37 ` [PATCH v3 1/2] perf arm-spe: Prepare for adding data source packet implementations for other cores Ilkka Koskinen
2024-11-07 15:25   ` Leo Yan
2024-11-07 23:55     ` Ilkka Koskinen [this message]
2024-11-06 19:37 ` [PATCH v3 2/2] perf arm-spe: Add support for SPE Data Source packet on AmpereOne Ilkka Koskinen
2024-11-07 15:44   ` Leo Yan
2024-11-07 23:56     ` Ilkka Koskinen

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=1c7db0d2-d08c-2d9-2976-30d1e9d8f2f4@os.amperecomputing.com \
    --to=ilkka@os.amperecomputing.com \
    --cc=acme@kernel.org \
    --cc=adrian.hunter@intel.com \
    --cc=alexander.shishkin@linux.intel.com \
    --cc=graham.woodward@arm.com \
    --cc=irogers@google.com \
    --cc=james.clark@linaro.org \
    --cc=john.g.garry@oracle.com \
    --cc=jolsa@kernel.org \
    --cc=kan.liang@linux.intel.com \
    --cc=leo.yan@arm.com \
    --cc=leo.yan@linux.dev \
    --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=mike.leach@linaro.org \
    --cc=mingo@redhat.com \
    --cc=namhyung@kernel.org \
    --cc=peterz@infradead.org \
    --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).