[PATCH 5/5] drm/amd/pm: Use separate metrics table for smu_v13_0_12
Zhang, Hawking
Hawking.Zhang at amd.com
Fri Feb 21 12:59:45 UTC 2025
[AMD Official Use Only - AMD Internal Distribution Only]
Series is
Reviewed-by: Hawking Zhang <Hawking.Zhang at amd.com>
Regards,
Hawking
-----Original Message-----
From: Kamal, Asad <Asad.Kamal at amd.com>
Sent: Friday, February 21, 2025 20:19
To: amd-gfx at lists.freedesktop.org; Lazar, Lijo <Lijo.Lazar at amd.com>
Cc: Zhang, Hawking <Hawking.Zhang at amd.com>; Ma, Le <Le.Ma at amd.com>; Zhang, Morris <Shiwu.Zhang at amd.com>; Kamal, Asad <Asad.Kamal at amd.com>; Deucher, Alexander <Alexander.Deucher at amd.com>
Subject: [PATCH 5/5] drm/amd/pm: Use separate metrics table for smu_v13_0_12
Use separate metrics table for smu_v13_0_12 and fetch metrics data using that.
v2: Fix jpeg busy indexing (Lijo)
Signed-off-by: Asad Kamal <asad.kamal at amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar at amd.com>
---
drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h | 4 +
.../drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c | 219 +++++++++++++++++-
.../drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c | 9 +-
3 files changed, 230 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h
index 4dc3b37d52b9..cd03caffe317 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h
+++ b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h
@@ -313,6 +313,10 @@ void smu_v13_0_interrupt_work(struct smu_context *smu); bool smu_v13_0_12_is_dpm_running(struct smu_context *smu); int smu_v13_0_12_get_max_metrics_size(void);
int smu_v13_0_12_setup_driver_pptable(struct smu_context *smu);
+int smu_v13_0_12_get_smu_metrics_data(struct smu_context *smu,
+ MetricsMember_t member,
+ uint32_t *value);
+ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void
+**table);
extern const struct cmn2asic_mapping smu_v13_0_12_feature_mask_map[]; extern const struct cmn2asic_msg_mapping smu_v13_0_12_message_map[]; #endif diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c
index e153062adf77..5e7e97d90805 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c
@@ -56,6 +56,10 @@
(FEATURE_MASK(FEATURE_DATA_CALCULATION) | \
FEATURE_MASK(FEATURE_DPM_GFXCLK) | FEATURE_MASK(FEATURE_DPM_FCLK))
+#define NUM_JPEG_RINGS_FW 10
+#define NUM_JPEG_RINGS_GPU_METRICS(gpu_metrics) \
+ (ARRAY_SIZE(gpu_metrics->xcp_stats[0].jpeg_busy) / 4)
+
const struct cmn2asic_mapping smu_v13_0_12_feature_mask_map[SMU_FEATURE_COUNT] = {
SMU_13_0_12_FEA_MAP(SMU_FEATURE_DATA_CALCULATIONS_BIT, FEATURE_DATA_CALCULATION),
SMU_13_0_12_FEA_MAP(SMU_FEATURE_DPM_GFXCLK_BIT, FEATURE_DPM_GFXCLK),
@@ -175,7 +179,7 @@ static int smu_v13_0_12_fru_get_product_info(struct smu_context *smu,
int smu_v13_0_12_get_max_metrics_size(void)
{
- return sizeof(StaticMetricsTable_t);
+ return max(sizeof(StaticMetricsTable_t), sizeof(MetricsTable_t));
}
static int smu_v13_0_12_get_static_metrics_table(struct smu_context *smu) @@ -258,3 +262,216 @@ bool smu_v13_0_12_is_dpm_running(struct smu_context *smu)
return !!(feature_enabled & SMC_DPM_FEATURE); }
+
+int smu_v13_0_12_get_smu_metrics_data(struct smu_context *smu,
+ MetricsMember_t member,
+ uint32_t *value)
+{
+ struct smu_table_context *smu_table = &smu->smu_table;
+ MetricsTable_t *metrics = (MetricsTable_t *)smu_table->metrics_table;
+ struct amdgpu_device *adev = smu->adev;
+ int ret = 0;
+ int xcc_id;
+
+ /* For clocks with multiple instances, only report the first one */
+ switch (member) {
+ case METRICS_CURR_GFXCLK:
+ case METRICS_AVERAGE_GFXCLK:
+ xcc_id = GET_INST(GC, 0);
+ *value = SMUQ10_ROUND(metrics->GfxclkFrequency[xcc_id]);
+ break;
+ case METRICS_CURR_SOCCLK:
+ case METRICS_AVERAGE_SOCCLK:
+ *value = SMUQ10_ROUND(metrics->SocclkFrequency[0]);
+ break;
+ case METRICS_CURR_UCLK:
+ case METRICS_AVERAGE_UCLK:
+ *value = SMUQ10_ROUND(metrics->UclkFrequency);
+ break;
+ case METRICS_CURR_VCLK:
+ *value = SMUQ10_ROUND(metrics->VclkFrequency[0]);
+ break;
+ case METRICS_CURR_DCLK:
+ *value = SMUQ10_ROUND(metrics->DclkFrequency[0]);
+ break;
+ case METRICS_CURR_FCLK:
+ *value = SMUQ10_ROUND(metrics->FclkFrequency);
+ break;
+ case METRICS_AVERAGE_GFXACTIVITY:
+ *value = SMUQ10_ROUND(metrics->SocketGfxBusy);
+ break;
+ case METRICS_AVERAGE_MEMACTIVITY:
+ *value = SMUQ10_ROUND(metrics->DramBandwidthUtilization);
+ break;
+ case METRICS_CURR_SOCKETPOWER:
+ *value = SMUQ10_ROUND(metrics->SocketPower) << 8;
+ break;
+ case METRICS_TEMPERATURE_HOTSPOT:
+ *value = SMUQ10_ROUND(metrics->MaxSocketTemperature) *
+ SMU_TEMPERATURE_UNITS_PER_CENTIGRADES;
+ break;
+ case METRICS_TEMPERATURE_MEM:
+ *value = SMUQ10_ROUND(metrics->MaxHbmTemperature) *
+ SMU_TEMPERATURE_UNITS_PER_CENTIGRADES;
+ break;
+ /* This is the max of all VRs and not just SOC VR.
+ * No need to define another data type for the same.
+ */
+ case METRICS_TEMPERATURE_VRSOC:
+ *value = SMUQ10_ROUND(metrics->MaxVrTemperature) *
+ SMU_TEMPERATURE_UNITS_PER_CENTIGRADES;
+ break;
+ default:
+ *value = UINT_MAX;
+ break;
+ }
+
+ return ret;
+}
+
+ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void
+**table) {
+ struct smu_table_context *smu_table = &smu->smu_table;
+ struct gpu_metrics_v1_7 *gpu_metrics =
+ (struct gpu_metrics_v1_7 *)smu_table->gpu_metrics_table;
+ int ret = 0, xcc_id, inst, i, j, k, idx;
+ struct amdgpu_device *adev = smu->adev;
+ u8 num_jpeg_rings_gpu_metrics;
+ MetricsTable_t *metrics;
+ struct amdgpu_xcp *xcp;
+ u32 inst_mask;
+
+ metrics = kzalloc(sizeof(MetricsTable_t), GFP_KERNEL);
+ memcpy(metrics, smu_table->metrics_table, sizeof(MetricsTable_t));
+
+ smu_cmn_init_soft_gpu_metrics(gpu_metrics, 1, 7);
+
+ gpu_metrics->temperature_hotspot =
+ SMUQ10_ROUND(metrics->MaxSocketTemperature);
+ /* Individual HBM stack temperature is not reported */
+ gpu_metrics->temperature_mem =
+ SMUQ10_ROUND(metrics->MaxHbmTemperature);
+ /* Reports max temperature of all voltage rails */
+ gpu_metrics->temperature_vrsoc =
+ SMUQ10_ROUND(metrics->MaxVrTemperature);
+
+ gpu_metrics->average_gfx_activity =
+ SMUQ10_ROUND(metrics->SocketGfxBusy);
+ gpu_metrics->average_umc_activity =
+ SMUQ10_ROUND(metrics->DramBandwidthUtilization);
+
+ gpu_metrics->mem_max_bandwidth =
+ SMUQ10_ROUND(metrics->MaxDramBandwidth);
+
+ gpu_metrics->curr_socket_power =
+ SMUQ10_ROUND(metrics->SocketPower);
+ /* Energy counter reported in 15.259uJ (2^-16) units */
+ gpu_metrics->energy_accumulator = metrics->SocketEnergyAcc;
+
+ for (i = 0; i < MAX_GFX_CLKS; i++) {
+ xcc_id = GET_INST(GC, i);
+ if (xcc_id >= 0)
+ gpu_metrics->current_gfxclk[i] =
+ SMUQ10_ROUND(metrics->GfxclkFrequency[xcc_id]);
+
+ if (i < MAX_CLKS) {
+ gpu_metrics->current_socclk[i] =
+ SMUQ10_ROUND(metrics->SocclkFrequency[i]);
+ inst = GET_INST(VCN, i);
+ if (inst >= 0) {
+ gpu_metrics->current_vclk0[i] =
+ SMUQ10_ROUND(metrics->VclkFrequency[inst]);
+ gpu_metrics->current_dclk0[i] =
+ SMUQ10_ROUND(metrics->DclkFrequency[inst]);
+ }
+ }
+ }
+
+ gpu_metrics->current_uclk = SMUQ10_ROUND(metrics->UclkFrequency);
+
+ /* Total accumulated cycle counter */
+ gpu_metrics->accumulation_counter = metrics->AccumulationCounter;
+
+ /* Accumulated throttler residencies */
+ gpu_metrics->prochot_residency_acc = metrics->ProchotResidencyAcc;
+ gpu_metrics->ppt_residency_acc = metrics->PptResidencyAcc;
+ gpu_metrics->socket_thm_residency_acc = metrics->SocketThmResidencyAcc;
+ gpu_metrics->vr_thm_residency_acc = metrics->VrThmResidencyAcc;
+ gpu_metrics->hbm_thm_residency_acc = metrics->HbmThmResidencyAcc;
+
+ /* Clock Lock Status. Each bit corresponds to each GFXCLK instance */
+ gpu_metrics->gfxclk_lock_status = metrics->GfxLockXCDMak >>
+GET_INST(GC, 0);
+
+ gpu_metrics->pcie_link_width = metrics->PCIeLinkWidth;
+ gpu_metrics->pcie_link_speed =
+ pcie_gen_to_speed(metrics->PCIeLinkSpeed);
+ gpu_metrics->pcie_bandwidth_acc =
+ SMUQ10_ROUND(metrics->PcieBandwidthAcc[0]);
+ gpu_metrics->pcie_bandwidth_inst =
+ SMUQ10_ROUND(metrics->PcieBandwidth[0]);
+ gpu_metrics->pcie_l0_to_recov_count_acc = metrics->PCIeL0ToRecoveryCountAcc;
+ gpu_metrics->pcie_replay_count_acc = metrics->PCIenReplayAAcc;
+ gpu_metrics->pcie_replay_rover_count_acc =
+ metrics->PCIenReplayARolloverCountAcc;
+ gpu_metrics->pcie_nak_sent_count_acc = metrics->PCIeNAKSentCountAcc;
+ gpu_metrics->pcie_nak_rcvd_count_acc = metrics->PCIeNAKReceivedCountAcc;
+ gpu_metrics->pcie_lc_perf_other_end_recovery =
+metrics->PCIeOtherEndRecoveryAcc;
+
+ gpu_metrics->system_clock_counter = ktime_get_boottime_ns();
+
+ gpu_metrics->gfx_activity_acc = SMUQ10_ROUND(metrics->SocketGfxBusyAcc);
+ gpu_metrics->mem_activity_acc =
+SMUQ10_ROUND(metrics->DramBandwidthUtilizationAcc);
+
+ for (i = 0; i < NUM_XGMI_LINKS; i++) {
+ gpu_metrics->xgmi_read_data_acc[i] =
+ SMUQ10_ROUND(metrics->XgmiReadDataSizeAcc[i]);
+ gpu_metrics->xgmi_write_data_acc[i] =
+ SMUQ10_ROUND(metrics->XgmiWriteDataSizeAcc[i]);
+ ret = amdgpu_get_xgmi_link_status(adev, i);
+ if (ret >= 0)
+ gpu_metrics->xgmi_link_status[i] = ret;
+ }
+
+ gpu_metrics->num_partition = adev->xcp_mgr->num_xcps;
+
+ num_jpeg_rings_gpu_metrics = NUM_JPEG_RINGS_GPU_METRICS(gpu_metrics);
+ for_each_xcp(adev->xcp_mgr, xcp, i) {
+ amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_VCN, &inst_mask);
+ idx = 0;
+ for_each_inst(k, inst_mask) {
+ /* Both JPEG and VCN has same instances */
+ inst = GET_INST(VCN, k);
+
+ for (j = 0; j < num_jpeg_rings_gpu_metrics; ++j) {
+ gpu_metrics->xcp_stats[i].jpeg_busy
+ [(idx * num_jpeg_rings_gpu_metrics) + j] =
+ SMUQ10_ROUND(metrics->JpegBusy
+ [(inst * NUM_JPEG_RINGS_FW) + j]);
+ }
+ gpu_metrics->xcp_stats[i].vcn_busy[idx] =
+ SMUQ10_ROUND(metrics->VcnBusy[inst]);
+ idx++;
+ }
+
+ amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_GFX, &inst_mask);
+ idx = 0;
+ for_each_inst(k, inst_mask) {
+ inst = GET_INST(GC, k);
+ gpu_metrics->xcp_stats[i].gfx_busy_inst[idx] =
+ SMUQ10_ROUND(metrics->GfxBusy[inst]);
+ gpu_metrics->xcp_stats[i].gfx_busy_acc[idx] =
+ SMUQ10_ROUND(metrics->GfxBusyAcc[inst]);
+ idx++;
+ }
+ }
+
+ gpu_metrics->xgmi_link_width = SMUQ10_ROUND(metrics->XgmiWidth);
+ gpu_metrics->xgmi_link_speed = SMUQ10_ROUND(metrics->XgmiBitrate);
+
+ gpu_metrics->firmware_timestamp = metrics->Timestamp;
+
+ *table = (void *)gpu_metrics;
+ kfree(metrics);
+
+ return sizeof(*gpu_metrics);
+}
diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c
index c3e0da680cf8..76cc827f128c 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c
@@ -515,7 +515,8 @@ static int smu_v13_0_6_tables_init(struct smu_context *smu)
PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM);
SMU_TABLE_INIT(tables, SMU_TABLE_SMU_METRICS,
- max(gpu_metrcs_size, smu_v13_0_12_get_max_metrics_size()),
+ max(gpu_metrcs_size,
+ smu_v13_0_12_get_max_metrics_size()),
PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT);
@@ -1145,6 +1146,9 @@ static int smu_v13_0_6_get_smu_metrics_data(struct smu_context *smu,
if (ret)
return ret;
+ if (smu_v13_0_6_cap_supported(smu, SMU_CAP(STATIC_METRICS)))
+ return smu_v13_0_12_get_smu_metrics_data(smu, member, value);
+
/* For clocks with multiple instances, only report the first one */
switch (member) {
case METRICS_CURR_GFXCLK:
@@ -2507,6 +2511,9 @@ static ssize_t smu_v13_0_6_get_gpu_metrics(struct smu_context *smu, void **table
return ret;
}
+ if (smu_v13_0_6_cap_supported(smu, SMU_CAP(STATIC_METRICS)))
+ return smu_v13_0_12_get_gpu_metrics(smu, table);
+
metrics_v1 = (MetricsTableV1_t *)metrics_v0;
metrics_v2 = (MetricsTableV2_t *)metrics_v0;
--
2.46.0
More information about the amd-gfx
mailing list