[AMD Official Use Only - AMD Internal Distribution Only]
Reviewed-by: Yang Wang <kevinyang.wang@xxxxxxx>
It is better to change the patch tile prefix to 'drm/amd/pm' .
Best Regards,
Kevin
-----Original Message-----
From: Kenneth Feng <kenneth.feng@xxxxxxx>
Sent: Thursday, July 4, 2024 8:17 AM
To: amd-gfx@xxxxxxxxxxxxxxxxxxxxx
Cc: Wang, Yang(Kevin) <KevinYang.Wang@xxxxxxx>; Feng, Kenneth <Kenneth.Feng@xxxxxxx>
Subject: [PATCH] amdgpu/pm: support gpu_metrics sysfs interface for smu v14.0.2/3
support gpu_metrics sysfs interface for smu v14.0.2/3
Signed-off-by: Kenneth Feng <kenneth.feng@xxxxxxx>
---
.../drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c | 86 ++++++++++++++++++-
1 file changed, 84 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
index 98ea58d792ca..e1a27903c80a 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
@@ -66,6 +66,7 @@
#define MP0_MP1_DATA_REGION_SIZE_COMBOPPTABLE 0x4000
#define DEBUGSMC_MSG_Mode1Reset 2
+#define LINK_SPEED_MAX 3
static struct cmn2asic_msg_mapping smu_v14_0_2_message_map[SMU_MSG_MAX_COUNT] = {
MSG_MAP(TestMessage, PPSMC_MSG_TestMessage, 1),
@@ -221,7 +222,6 @@ static struct cmn2asic_mapping smu_v14_0_2_workload_map[PP_SMC_POWER_PROFILE_COU
WORKLOAD_MAP(PP_SMC_POWER_PROFILE_WINDOW3D, WORKLOAD_PPLIB_WINDOW_3D_BIT),
};
-#if 0
static const uint8_t smu_v14_0_2_throttler_map[] = {
[THROTTLER_PPT0_BIT] = (SMU_THROTTLER_PPT0_BIT),
[THROTTLER_PPT1_BIT] = (SMU_THROTTLER_PPT1_BIT),
@@ -241,7 +241,6 @@ static const uint8_t smu_v14_0_2_throttler_map[] = {
[THROTTLER_GFX_APCC_PLUS_BIT] = (SMU_THROTTLER_APCC_BIT),
[THROTTLER_FIT_BIT] = (SMU_THROTTLER_FIT_BIT),
};
-#endif
static int
smu_v14_0_2_get_allowed_feature_mask(struct smu_context *smu, @@ -1869,6 +1868,88 @@ static ssize_t smu_v14_0_2_get_ecc_info(struct smu_context *smu,
return ret;
}
+static ssize_t smu_v14_0_2_get_gpu_metrics(struct smu_context *smu,
+ void **table)
+{
+ struct smu_table_context *smu_table = &smu->smu_table;
+ struct gpu_metrics_v1_3 *gpu_metrics =
+ (struct gpu_metrics_v1_3 *)smu_table->gpu_metrics_table;
+ SmuMetricsExternal_t metrics_ext;
+ SmuMetrics_t *metrics = &metrics_ext.SmuMetrics;
+ int ret = 0;
+
+ ret = smu_cmn_get_metrics_table(smu,
+ &metrics_ext,
+ true);
+ if (ret)
+ return ret;
+
+ smu_cmn_init_soft_gpu_metrics(gpu_metrics, 1, 3);
+
+ gpu_metrics->temperature_edge = metrics->AvgTemperature[TEMP_EDGE];
+ gpu_metrics->temperature_hotspot = metrics->AvgTemperature[TEMP_HOTSPOT];
+ gpu_metrics->temperature_mem = metrics->AvgTemperature[TEMP_MEM];
+ gpu_metrics->temperature_vrgfx = metrics->AvgTemperature[TEMP_VR_GFX];
+ gpu_metrics->temperature_vrsoc = metrics->AvgTemperature[TEMP_VR_SOC];
+ gpu_metrics->temperature_vrmem = max(metrics->AvgTemperature[TEMP_VR_MEM0],
+ metrics->AvgTemperature[TEMP_VR_MEM1]);
+
+ gpu_metrics->average_gfx_activity = metrics->AverageGfxActivity;
+ gpu_metrics->average_umc_activity = metrics->AverageUclkActivity;
+ gpu_metrics->average_mm_activity = max(metrics->Vcn0ActivityPercentage,
+ metrics->Vcn1ActivityPercentage);
+
+ gpu_metrics->average_socket_power = metrics->AverageSocketPower;
+ gpu_metrics->energy_accumulator = metrics->EnergyAccumulator;
+
+ if (metrics->AverageGfxActivity <= SMU_14_0_2_BUSY_THRESHOLD)
+ gpu_metrics->average_gfxclk_frequency = metrics->AverageGfxclkFrequencyPostDs;
+ else
+ gpu_metrics->average_gfxclk_frequency =
+metrics->AverageGfxclkFrequencyPreDs;
+
+ if (metrics->AverageUclkActivity <= SMU_14_0_2_BUSY_THRESHOLD)
+ gpu_metrics->average_uclk_frequency = metrics->AverageMemclkFrequencyPostDs;
+ else
+ gpu_metrics->average_uclk_frequency =
+metrics->AverageMemclkFrequencyPreDs;
+
+ gpu_metrics->average_vclk0_frequency = metrics->AverageVclk0Frequency;
+ gpu_metrics->average_dclk0_frequency = metrics->AverageDclk0Frequency;
+ gpu_metrics->average_vclk1_frequency = metrics->AverageVclk1Frequency;
+ gpu_metrics->average_dclk1_frequency = metrics->AverageDclk1Frequency;
+
+ gpu_metrics->current_gfxclk = gpu_metrics->average_gfxclk_frequency;
+ gpu_metrics->current_socclk = metrics->CurrClock[PPCLK_SOCCLK];
+ gpu_metrics->current_uclk = metrics->CurrClock[PPCLK_UCLK];
+ gpu_metrics->current_vclk0 = metrics->CurrClock[PPCLK_VCLK_0];
+ gpu_metrics->current_dclk0 = metrics->CurrClock[PPCLK_DCLK_0];
+ gpu_metrics->current_vclk1 = metrics->CurrClock[PPCLK_VCLK_0];
+ gpu_metrics->current_dclk1 = metrics->CurrClock[PPCLK_DCLK_0];
+
+ gpu_metrics->throttle_status =
+ smu_v14_0_2_get_throttler_status(metrics);
+ gpu_metrics->indep_throttle_status =
+ smu_cmn_get_indep_throttler_status(gpu_metrics->throttle_status,
+ smu_v14_0_2_throttler_map);
+
+ gpu_metrics->current_fan_speed = metrics->AvgFanRpm;
+
+ gpu_metrics->pcie_link_width = metrics->PcieWidth;
+ if ((metrics->PcieRate - 1) > LINK_SPEED_MAX)
+ gpu_metrics->pcie_link_speed = pcie_gen_to_speed(1);
+ else
+ gpu_metrics->pcie_link_speed = pcie_gen_to_speed(metrics->PcieRate);
+
+ gpu_metrics->system_clock_counter = ktime_get_boottime_ns();
+
+ gpu_metrics->voltage_gfx = metrics->AvgVoltage[SVI_PLANE_VDD_GFX];
+ gpu_metrics->voltage_soc = metrics->AvgVoltage[SVI_PLANE_VDD_SOC];
+ gpu_metrics->voltage_mem = metrics->AvgVoltage[SVI_PLANE_VDDIO_MEM];
+
+ *table = (void *)gpu_metrics;
+
+ return sizeof(struct gpu_metrics_v1_3); }
+
static const struct pptable_funcs smu_v14_0_2_ppt_funcs = {
.get_allowed_feature_mask = smu_v14_0_2_get_allowed_feature_mask,
.set_default_dpm_table = smu_v14_0_2_set_default_dpm_table,
@@ -1905,6 +1986,7 @@ static const struct pptable_funcs smu_v14_0_2_ppt_funcs = {
.enable_thermal_alert = smu_v14_0_enable_thermal_alert,
.disable_thermal_alert = smu_v14_0_disable_thermal_alert,
.notify_memory_pool_location = smu_v14_0_notify_memory_pool_location,
+ .get_gpu_metrics = smu_v14_0_2_get_gpu_metrics,
.set_soft_freq_limited_range = smu_v14_0_set_soft_freq_limited_range,
.init_pptable_microcode = smu_v14_0_init_pptable_microcode,
.populate_umd_state_clk = smu_v14_0_2_populate_umd_state_clk,
--
2.34.1
