| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <4c2aa7e9-8e4e-aa1a-9c0-23c62f6b15cd@os.amperecomputing.com>
Date: Tue, 5 Nov 2024 16:32:20 -0800 (PST)
From: Ilkka Koskinen <ilkka@...amperecomputing.com>
To: James Clark <james.clark@...aro.org>
cc: Ilkka Koskinen <ilkka@...amperecomputing.com>,
linux-arm-kernel@...ts.infradead.org, linux-perf-users@...r.kernel.org,
linux-kernel@...r.kernel.org, John Garry <john.g.garry@...cle.com>,
Will Deacon <will@...nel.org>, Mike Leach <mike.leach@...aro.org>,
Leo Yan <leo.yan@...ux.dev>, Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...hat.com>, Arnaldo Carvalho de Melo <acme@...nel.org>,
Namhyung Kim <namhyung@...nel.org>, Mark Rutland <mark.rutland@....com>,
Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
Jiri Olsa <jolsa@...nel.org>, Ian Rogers <irogers@...gle.com>,
Adrian Hunter <adrian.hunter@...el.com>,
"Liang, Kan" <kan.liang@...ux.intel.com>,
Graham Woodward <graham.woodward@....com>
Subject: Re: [PATCH v2 2/2] perf arm-spe: Add support for SPE Data Source
packet on AmpereOne
Hi James,
On Mon, 4 Nov 2024, James Clark wrote:
> On 31/10/2024 9:35 pm, Ilkka Koskinen wrote:
>> Decode SPE Data Source packets on AmpereOne. The field is IMPDEF.
>>
>> Signed-off-by: Ilkka Koskinen <ilkka@...amperecomputing.com>
>> ---
>> .../util/arm-spe-decoder/arm-spe-decoder.h | 9 +++
>> tools/perf/util/arm-spe.c | 65 +++++++++++++++++++
>> 2 files changed, 74 insertions(+)
>>
>> diff --git a/tools/perf/util/arm-spe-decoder/arm-spe-decoder.h
>> b/tools/perf/util/arm-spe-decoder/arm-spe-decoder.h
>> index 358c611eeddb..4bcd627e859f 100644
>> --- a/tools/perf/util/arm-spe-decoder/arm-spe-decoder.h
>> +++ b/tools/perf/util/arm-spe-decoder/arm-spe-decoder.h
>> @@ -67,6 +67,15 @@ enum arm_spe_common_data_source {
>> ARM_SPE_COMMON_DS_DRAM = 0xe,
>> };
>> +enum arm_spe_ampereone_data_source {
>> + ARM_SPE_AMPEREONE_LOCAL_CHIP_CACHE_OR_DEVICE = 0x0,
>> + ARM_SPE_AMPEREONE_SLC = 0x3,
>> + ARM_SPE_AMPEREONE_REMOTE_CHIP_CACHE = 0x5,
>> + ARM_SPE_AMPEREONE_DDR = 0x7,
>> + ARM_SPE_AMPEREONE_L1D = 0x8,
>> + ARM_SPE_AMPEREONE_L2D = 0x9,
>> +};
>> +
>> struct arm_spe_record {
>> enum arm_spe_sample_type type;
>> int err;
>> diff --git a/tools/perf/util/arm-spe.c b/tools/perf/util/arm-spe.c
>> index 9586416be30a..700d4bc8d8ec 100644
>> --- a/tools/perf/util/arm-spe.c
>> +++ b/tools/perf/util/arm-spe.c
>> @@ -103,6 +103,30 @@ struct arm_spe_queue {
>> u32 flags;
>> };
>> +struct arm_spe_source_mapping {
>> + u16 source;
>> + enum arm_spe_common_data_source common_src;
>> +};
>> +
>> +#define MAP_SOURCE(src, common) \
>> + { \
>> + .source = ARM_SPE_##src, \
>> + .common_src = ARM_SPE_COMMON_##common, \
>> + }
>> +
>> +static int arm_spe__map_to_common_source(u16 source,
>> + struct arm_spe_source_mapping *tbl,
>> + int nr_sources)
>> +{
>> + while (nr_sources--) {
>> + if (tbl->source == source)
>> + return tbl->common_src;
>> + tbl++;
>> + }
>> +
>
> Hi Ilkka,
>
> I think a simple switch statement here would be easier to follow than the
> loop, custom macro and then having the mappings in some other place:
>
> switch(source)
> case 0x0: /* AMPEREONE_LOCAL_CHIP_CACHE_OR_DEVICE */
> return DS_PEER_CORE;
>
> etc...
>
>> + return -1;
>
> And the default case can return 0xfff directly, which avoids the if else
> later only to convert this -1 back into 0xfff.
I can surely do that.
>
>> +}
>> +
>> static void arm_spe_dump(struct arm_spe *spe __maybe_unused,
>> unsigned char *buf, size_t len)
>> {
>> @@ -443,6 +467,11 @@ static const struct midr_range
>> common_ds_encoding_cpus[] = {
>> {},
>> };
>> +static const struct midr_range ampereone_ds_encoding_cpus[] = {
>> + MIDR_ALL_VERSIONS(MIDR_AMPERE1A),
>> + {},
>> +};
>> +
>> static void arm_spe__sample_flags(struct arm_spe_queue *speq)
>> {
>> const struct arm_spe_record *record = &speq->decoder->record;
>> @@ -532,6 +561,38 @@ static void arm_spe__synth_data_source_common(const
>> struct arm_spe_record *recor
>> }
>> }
>> +static struct arm_spe_source_mapping ampereone_sources[] = {
>> + MAP_SOURCE(AMPEREONE_LOCAL_CHIP_CACHE_OR_DEVICE, DS_PEER_CORE),
>> + MAP_SOURCE(AMPEREONE_SLC, DS_SYS_CACHE),
>> + MAP_SOURCE(AMPEREONE_REMOTE_CHIP_CACHE, DS_REMOTE),
>> + MAP_SOURCE(AMPEREONE_DDR, DS_DRAM),
>> + MAP_SOURCE(AMPEREONE_L1D, DS_L1D),
>> + MAP_SOURCE(AMPEREONE_L2D, DS_L2),
>> +};
>> +
>> +/*
>> + * Source is IMPDEF. Here we convert the source code used on AmpereOne
>> cores
>> + * to the common (Neoverse, Cortex) to avoid duplicating the decoding
>> code.
>> + */
>> +static void arm_spe__synth_data_source_ampereone(const struct
>> arm_spe_record *record,
>> + union perf_mem_data_src
>> *data_src)
>> +{
>> + int common_src;
>> + struct arm_spe_record common_record;
>> +
>> + common_src = arm_spe__map_to_common_source(record->source,
>> + ampereone_sources,
>> +
>> ARRAY_SIZE(ampereone_sources));
>> + if (common_src < 0)
>> + /* Assign a bogus value that's not used for common coding */
>> + common_record.source = 0xfff;
>> + else
>> + common_record.source = common_src;
>> +
>> + common_record.op = record->op;
>> + arm_spe__synth_data_source_common(&common_record, data_src);
>> +}
>> +
>> static void arm_spe__synth_memory_level(const struct arm_spe_record
>> *record,
>> union perf_mem_data_src *data_src)
>> {
>> @@ -606,6 +667,8 @@ static u64 arm_spe__synth_data_source(struct
>> arm_spe_queue *speq,
>> union perf_mem_data_src data_src = { .mem_op = PERF_MEM_OP_NA };
>> bool is_common = arm_spe__is_ds_encoding_supported(speq,
>> common_ds_encoding_cpus);
>> + bool is_ampereone = arm_spe__is_ds_encoding_supported(speq,
>> + ampereone_ds_encoding_cpus);
>
> I know this probably already works, but we don't really need is_common
> is_ampere etc, it will only grow anyway. All we need is a list of midrs and
> function pairs. That also avoids doing is_ampereone even after we already
> know is_common == true.
>
> static const struct data_src[] = {
> ...
> DS(MIDR_ALL_VERSIONS(MIDR_NEOVERSE_V2), common_ds),
> DS(MIDR_ALL_VERSIONS(MIDR_AMPERE1A), ampere_ds),
> {},
> ...
> };
>
> "arm_spe__is_ds_encoding_supported" then becomes a direct call to
> "arm_spe__synth_ds" and we can drop the is_ampereone and is_common vars. Then
> adding new ones doesn't require changing the function anymore.
To be honest, I'm not a big fan of the is_xyz() thing but I just didn't
want to change it. Anyway, I'll change it for the next version.
Cheers, Ilkka
>
>> if (record->op & ARM_SPE_OP_LD)
>> data_src.mem_op = PERF_MEM_OP_LOAD;
>> @@ -616,6 +679,8 @@ static u64 arm_spe__synth_data_source(struct
>> arm_spe_queue *speq,
>> if (is_common)
>> arm_spe__synth_data_source_common(record, &data_src);
>> + else if (is_ampereone)
>> + arm_spe__synth_data_source_ampereone(record, &data_src);
>> else
>> arm_spe__synth_memory_level(record, &data_src);
>>
>
>
Powered by blists - more mailing lists