Re: [PATCH v4 1/2] perf: uncore: Adding documentation for ThunderX2 pmu uncore driver

[Ganapatrao Kulkarni]

On Fri, Apr 27, 2018 at 2:25 AM, Randy Dunlap <rdunlap@xxxxxxxxxxxxx> wrote:
> Hi,
>
> Just a few typo corrections...
>
> On 04/25/2018 02:00 AM, Ganapatrao Kulkarni wrote:
>> Documentation for the UNCORE PMUs on Cavium's ThunderX2 SoC.
>> The SoC has PMU support in its L3 cache controller (L3C) and in the
>> DDR4 Memory Controller (DMC).
>>
>> Signed-off-by: Ganapatrao Kulkarni <ganapatrao.kulkarni@xxxxxxxxxx>
>> ---
>>  Documentation/perf/thunderx2-pmu.txt | 66 ++++++++++++++++++++++++++++++++++++
>>  1 file changed, 66 insertions(+)
>>  create mode 100644 Documentation/perf/thunderx2-pmu.txt
>>
>> diff --git a/Documentation/perf/thunderx2-pmu.txt b/Documentation/perf/thunderx2-pmu.txt
>> new file mode 100644
>> index 0000000..9e9f535
>> --- /dev/null
>> +++ b/Documentation/perf/thunderx2-pmu.txt
>> @@ -0,0 +1,66 @@
>> +
>> +Cavium ThunderX2 SoC Performance Monitoring Unit (PMU UNCORE)
>> +==========================================================================
>> +
>> +ThunderX2 SoC PMU consists of independent system wide per Socket PMUs such
>> +as Level 3 Cache(L3C) and DDR4 Memory Controller(DMC).
>> +
>> +It has 8 independent DMC PMUs to capture performance events corresponding
>> +to 8 channels of DDR4 Memory Controller. There are 16 independent L3C PMUs
>> +to capture events corresponding to 16 tiles of L3 cache. Each PMU supports
>> +up to 4 counters.
>> +
>> +Counters are independent programmable and can be started and stopped
>
>                 independently

thanks, will update.
>
>> +individually. Each counter can be set to sample specific perf events.
>> +Counters are 32 bit and does not support overflow interrupt, they are
>
>                            do not                    interrupt; they are
>
>
>> +sampled at every 2 seconds. The Counters register access are multiplexed
>> +across channels of DMC and L3C. The muxing(select channel) is done through
>> +write to a Secure register using smcc calls.
>> +
>> +PMU UNCORE (perf) driver:
>> +
>> +The thunderx2-pmu driver registers several perf PMUs for DMC and L3C devices.
>> +Each of the PMU provides description of its available events
>
>         of the PMUs
>
>> +and configuration options in sysfs.
>> +     see /sys/devices/uncore_<l3c_S_X/dmc_S_X/>
>> +
>> +S is socket id and X represents channel number.
>> +Each PMU can be used to sample up to 4 events simultaneously.
>> +
>> +The "format" directory describes format of the config (event ID).
>> +The "events" directory provides configuration templates for all
>> +supported event types that can be used with perf tool.
>> +
>> +For example, "uncore_dmc_0_0/cnt_cycles/" is an
>> +equivalent of "uncore_dmc_0_0/config=0x1/".
>> +
>> +Each perf driver also provides a "cpumask" sysfs attribute, which contains a
>> +single CPU ID of the processor which is likely to be used to handle all the
>> +PMU events. It will be the first online CPU from the NUMA node of PMU device.
>> +
>> +Example for perf tool use:
>> +
>> +perf stat -a -e \
>> +uncore_dmc_0_0/cnt_cycles/,\
>> +uncore_dmc_0_1/cnt_cycles/,\
>> +uncore_dmc_0_2/cnt_cycles/,\
>> +uncore_dmc_0_3/cnt_cycles/,\
>> +uncore_dmc_0_4/cnt_cycles/,\
>> +uncore_dmc_0_5/cnt_cycles/,\
>> +uncore_dmc_0_6/cnt_cycles/,\
>> +uncore_dmc_0_7/cnt_cycles/ sleep 1
>> +
>> +perf stat -a -e \
>> +uncore_dmc_0_0/cancelled_read_txns/,\
>> +uncore_dmc_0_0/cnt_cycles/,\
>> +uncore_dmc_0_0/consumed_read_txns/,\
>> +uncore_dmc_0_0/data_transfers/ sleep 1
>> +
>> +perf stat -a -e \
>> +uncore_l3c_0_0/l3_retry/,\
>> +uncore_l3c_0_0/read_hit/,\
>> +uncore_l3c_0_0/read_request/,\
>> +uncore_l3c_0_0/inv_request/ sleep 1
>> +
>> +The driver does not support sampling, therefore "perf record" will
>> +not work. Per-task (without "-a") perf sessions are not supported.
>>
>
>
> --
> ~Randy

thanks
Ganapat