RE: [PATCH v2] drm/amd/pm: Vangogh: Add new gpu_metrics_v2_4 to acquire gpu_metrics

From: Quan, Evan
Date: Tue Jul 18 2023 - 00:03:09 EST


[AMD Official Use Only - General]

Hi Wenyou,

I think you already got the greenlight(RB from Mario and ACK from me) to land the change.
Go ahead please.

Evan
> -----Original Message-----
> From: Yang, WenYou <WenYou.Yang@xxxxxxx>
> Sent: Thursday, July 13, 2023 8:56 AM
> To: Yang, WenYou <WenYou.Yang@xxxxxxx>; Deucher, Alexander
> <Alexander.Deucher@xxxxxxx>; Limonciello, Mario
> <Mario.Limonciello@xxxxxxx>; Koenig, Christian
> <Christian.Koenig@xxxxxxx>; Pan, Xinhui <Xinhui.Pan@xxxxxxx>; Quan,
> Evan <Evan.Quan@xxxxxxx>
> Cc: Yuan, Perry <Perry.Yuan@xxxxxxx>; Liang, Richard qi
> <Richardqi.Liang@xxxxxxx>; amd-gfx@xxxxxxxxxxxxxxxxxxxxx; dri-
> devel@xxxxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
> Subject: RE: [PATCH v2] drm/amd/pm: Vangogh: Add new gpu_metrics_v2_4
> to acquire gpu_metrics
>
> [AMD Official Use Only - General]
>
> Any comments? Any advice?
>
> Best Regards,
> Wenyou
>
> > -----Original Message-----
> > From: Wenyou Yang <WenYou.Yang@xxxxxxx>
> > Sent: Wednesday, June 21, 2023 2:32 PM
> > To: Deucher, Alexander <Alexander.Deucher@xxxxxxx>; Limonciello, Mario
> > <Mario.Limonciello@xxxxxxx>; Koenig, Christian
> > <Christian.Koenig@xxxxxxx>; Pan, Xinhui <Xinhui.Pan@xxxxxxx>; Quan,
> > Evan <Evan.Quan@xxxxxxx>
> > Cc: Yuan, Perry <Perry.Yuan@xxxxxxx>; Liang, Richard qi
> > <Richardqi.Liang@xxxxxxx>; amd-gfx@xxxxxxxxxxxxxxxxxxxxx; dri-
> > devel@xxxxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Yang,
> > WenYou <WenYou.Yang@xxxxxxx>
> > Subject: [PATCH v2] drm/amd/pm: Vangogh: Add new gpu_metrics_v2_4 to
> > acquire gpu_metrics
> >
> > To acquire the voltage and current info from gpu_metrics interface,
> > but
> > gpu_metrics_v2_3 doesn't contain them, and to be backward compatible,
> > add new gpu_metrics_v2_4 structure.
> >
> > Reviewed-by: Mario Limonciello <mario.limonciello@xxxxxxx>
> > Acked-by: Evan Quan <evan.quan@xxxxxxx>
> > Signed-off-by: Wenyou Yang <WenYou.Yang@xxxxxxx>
> > ---
> > .../gpu/drm/amd/include/kgd_pp_interface.h | 69 +++++++++++
> > .../gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c | 109
> ++++++++++++++++-
> > -
> > drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c | 3 +
> > 3 files changed, 172 insertions(+), 9 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/amd/include/kgd_pp_interface.h
> > b/drivers/gpu/drm/amd/include/kgd_pp_interface.h
> > index 9f542f6e19ed..90989405eddc 100644
> > --- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h
> > +++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h
> > @@ -892,4 +892,73 @@ struct gpu_metrics_v2_3 {
> > uint16_t average_temperature_core[8]; //
> > average CPU core temperature on APUs
> > uint16_t average_temperature_l3[2];
> > };
> > +
> > +struct gpu_metrics_v2_4 {
> > + struct metrics_table_header common_header;
> > +
> > + /* Temperature (unit: centi-Celsius) */
> > + uint16_t temperature_gfx;
> > + uint16_t temperature_soc;
> > + uint16_t temperature_core[8];
> > + uint16_t temperature_l3[2];
> > +
> > + /* Utilization (unit: centi) */
> > + uint16_t average_gfx_activity;
> > + uint16_t average_mm_activity;
> > +
> > + /* Driver attached timestamp (in ns) */
> > + uint64_t system_clock_counter;
> > +
> > + /* Power/Energy (unit: mW) */
> > + uint16_t average_socket_power;
> > + uint16_t average_cpu_power;
> > + uint16_t average_soc_power;
> > + uint16_t average_gfx_power;
> > + uint16_t average_core_power[8];
> > +
> > + /* Average clocks (unit: MHz) */
> > + uint16_t average_gfxclk_frequency;
> > + uint16_t average_socclk_frequency;
> > + uint16_t average_uclk_frequency;
> > + uint16_t average_fclk_frequency;
> > + uint16_t average_vclk_frequency;
> > + uint16_t average_dclk_frequency;
> > +
> > + /* Current clocks (unit: MHz) */
> > + uint16_t current_gfxclk;
> > + uint16_t current_socclk;
> > + uint16_t current_uclk;
> > + uint16_t current_fclk;
> > + uint16_t current_vclk;
> > + uint16_t current_dclk;
> > + uint16_t current_coreclk[8];
> > + uint16_t current_l3clk[2];
> > +
> > + /* Throttle status (ASIC dependent) */
> > + uint32_t throttle_status;
> > +
> > + /* Fans */
> > + uint16_t fan_pwm;
> > +
> > + uint16_t padding[3];
> > +
> > + /* Throttle status (ASIC independent) */
> > + uint64_t indep_throttle_status;
> > +
> > + /* Average Temperature (unit: centi-Celsius) */
> > + uint16_t average_temperature_gfx;
> > + uint16_t average_temperature_soc;
> > + uint16_t average_temperature_core[8];
> > + uint16_t average_temperature_l3[2];
> > +
> > + /* Power/Voltage (unit: mV) */
> > + uint16_t average_cpu_voltage;
> > + uint16_t average_soc_voltage;
> > + uint16_t average_gfx_voltage;
> > +
> > + /* Power/Current (unit: mA) */
> > + uint16_t average_cpu_current;
> > + uint16_t average_soc_current;
> > + uint16_t average_gfx_current;
> > +};
> > #endif
> > diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c
> > b/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c
> > index 067b4e0b026c..185d0b50ee8e 100644
> > --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c
> > +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c
> > @@ -1854,6 +1854,86 @@ static ssize_t
> > vangogh_get_gpu_metrics_v2_3(struct
> > smu_context *smu,
> > return sizeof(struct gpu_metrics_v2_3); }
> >
> > +static ssize_t vangogh_get_gpu_metrics_v2_4(struct smu_context *smu,
> > + void **table) {
> > + SmuMetrics_t metrics;
> > + struct smu_table_context *smu_table = &smu->smu_table;
> > + struct gpu_metrics_v2_4 *gpu_metrics =
> > + (struct gpu_metrics_v2_4 *)smu_table-
> > >gpu_metrics_table;
> > + int ret = 0;
> > +
> > + ret = smu_cmn_get_metrics_table(smu, &metrics, true);
> > + if (ret)
> > + return ret;
> > +
> > + smu_cmn_init_soft_gpu_metrics(gpu_metrics, 2, 4);
> > +
> > + gpu_metrics->temperature_gfx = metrics.Current.GfxTemperature;
> > + gpu_metrics->temperature_soc = metrics.Current.SocTemperature;
> > + memcpy(&gpu_metrics->temperature_core[0],
> > + &metrics.Current.CoreTemperature[0],
> > + sizeof(uint16_t) * 4);
> > + gpu_metrics->temperature_l3[0] =
> > + metrics.Current.L3Temperature[0];
> > +
> > + gpu_metrics->average_temperature_gfx =
> > metrics.Average.GfxTemperature;
> > + gpu_metrics->average_temperature_soc =
> > metrics.Average.SocTemperature;
> > + memcpy(&gpu_metrics->average_temperature_core[0],
> > + &metrics.Average.CoreTemperature[0],
> > + sizeof(uint16_t) * 4);
> > + gpu_metrics->average_temperature_l3[0] =
> > +metrics.Average.L3Temperature[0];
> > +
> > + gpu_metrics->average_gfx_activity = metrics.Current.GfxActivity;
> > + gpu_metrics->average_mm_activity = metrics.Current.UvdActivity;
> > +
> > + gpu_metrics->average_socket_power =
> > metrics.Current.CurrentSocketPower;
> > + gpu_metrics->average_cpu_power = metrics.Current.Power[0];
> > + gpu_metrics->average_soc_power = metrics.Current.Power[1];
> > + gpu_metrics->average_gfx_power = metrics.Current.Power[2];
> > +
> > + gpu_metrics->average_cpu_voltage = metrics.Current.Voltage[0];
> > + gpu_metrics->average_soc_voltage = metrics.Current.Voltage[1];
> > + gpu_metrics->average_gfx_voltage = metrics.Current.Voltage[2];
> > +
> > + gpu_metrics->average_cpu_current = metrics.Current.Current[0];
> > + gpu_metrics->average_soc_current = metrics.Current.Current[1];
> > + gpu_metrics->average_gfx_current = metrics.Current.Current[2];
> > +
> > + memcpy(&gpu_metrics->average_core_power[0],
> > + &metrics.Average.CorePower[0],
> > + sizeof(uint16_t) * 4);
> > +
> > + gpu_metrics->average_gfxclk_frequency =
> > metrics.Average.GfxclkFrequency;
> > + gpu_metrics->average_socclk_frequency =
> > metrics.Average.SocclkFrequency;
> > + gpu_metrics->average_uclk_frequency =
> > metrics.Average.MemclkFrequency;
> > + gpu_metrics->average_fclk_frequency =
> > metrics.Average.MemclkFrequency;
> > + gpu_metrics->average_vclk_frequency = metrics.Average.VclkFrequency;
> > + gpu_metrics->average_dclk_frequency =
> > metrics.Average.DclkFrequency;
> > +
> > + gpu_metrics->current_gfxclk = metrics.Current.GfxclkFrequency;
> > + gpu_metrics->current_socclk = metrics.Current.SocclkFrequency;
> > + gpu_metrics->current_uclk = metrics.Current.MemclkFrequency;
> > + gpu_metrics->current_fclk = metrics.Current.MemclkFrequency;
> > + gpu_metrics->current_vclk = metrics.Current.VclkFrequency;
> > + gpu_metrics->current_dclk = metrics.Current.DclkFrequency;
> > +
> > + memcpy(&gpu_metrics->current_coreclk[0],
> > + &metrics.Current.CoreFrequency[0],
> > + sizeof(uint16_t) * 4);
> > + gpu_metrics->current_l3clk[0] = metrics.Current.L3Frequency[0];
> > +
> > + gpu_metrics->throttle_status = metrics.Current.ThrottlerStatus;
> > + gpu_metrics->indep_throttle_status =
> > +
> >
> > smu_cmn_get_indep_throttler_status(metrics.Current.ThrottlerStatus,
> > +
> > vangogh_throttler_map);
> > +
> > + gpu_metrics->system_clock_counter = ktime_get_boottime_ns();
> > +
> > + *table = (void *)gpu_metrics;
> > +
> > + return sizeof(struct gpu_metrics_v2_4); }
> > +
> > static ssize_t vangogh_get_gpu_metrics(struct smu_context *smu,
> > void **table) { @@ -1923,23
> > +2003,34 @@ static ssize_t vangogh_common_get_gpu_metrics(struct
> > smu_context *smu, {
> > uint32_t if_version;
> > uint32_t smu_version;
> > + uint32_t smu_program;
> > + uint32_t fw_version;
> > int ret = 0;
> >
> > ret = smu_cmn_get_smc_version(smu, &if_version, &smu_version);
> > - if (ret) {
> > + if (ret)
> > return ret;
> > - }
> >
> > - if (smu_version >= 0x043F3E00) {
> > - if (if_version < 0x3)
> > - ret = vangogh_get_legacy_gpu_metrics_v2_3(smu,
> > table);
> > + smu_program = (smu_version >> 24) & 0xff;
> > + fw_version = smu_version & 0xffffff;
> > + if (smu_program == 6) {
> > + if (fw_version >= 0x3F0800)
> > + ret = vangogh_get_gpu_metrics_v2_4(smu, table);
> > else
> > ret = vangogh_get_gpu_metrics_v2_3(smu, table);
> > +
> > } else {
> > - if (if_version < 0x3)
> > - ret = vangogh_get_legacy_gpu_metrics(smu, table);
> > - else
> > - ret = vangogh_get_gpu_metrics(smu, table);
> > + if (smu_version >= 0x043F3E00) {
> > + if (if_version < 0x3)
> > + ret =
> > vangogh_get_legacy_gpu_metrics_v2_3(smu, table);
> > + else
> > + ret = vangogh_get_gpu_metrics_v2_3(smu,
> > table);
> > + } else {
> > + if (if_version < 0x3)
> > + ret =
> > + vangogh_get_legacy_gpu_metrics(smu,
> > table);
> > + else
> > + ret = vangogh_get_gpu_metrics(smu, table);
> > + }
> > }
> >
> > return ret;
> > diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c
> > b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c
> > index 3ecb900e6ecd..cee5e32b4ff9 100644
> > --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c
> > +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c
> > @@ -996,6 +996,9 @@ void smu_cmn_init_soft_gpu_metrics(void *table,
> > uint8_t frev, uint8_t crev)
> > case METRICS_VERSION(2, 3):
> > structure_size = sizeof(struct gpu_metrics_v2_3);
> > break;
> > + case METRICS_VERSION(2, 4):
> > + structure_size = sizeof(struct gpu_metrics_v2_4);
> > + break;
> > default:
> > return;
> > }
> > --
> > 2.39.2
>