drm/amd/powerplay: update Arcturus smu-driver headers
Subsystems (2)
PATCH 1/4 drm/amd/powerplay: update Arcturus smu-driver headers
To fit the latest 54.24.0 PMFW.
--- .../drm/amd/powerplay/inc/smu11_driver_if_arcturus.h | 12 ++++++++++-- drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h | 2 +- 2 files changed, 11 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/powerplay/inc/smu11_driver_if_arcturus.h +10 -2
@@ -200,6 +200,8 @@#define THROTTLER_PPM_BIT 13#define THROTTLER_FIT_BIT 14#define THROTTLER_APCC_BIT 15 +#define THROTTLER_VRHOT0_BIT 16 +#define THROTTLER_VRHOT1_BIT 17// Table transfer status #define TABLE_TRANSFER_OK 0x0
@@ -742,6 +744,9 @@ typedef struct {uint16_t SocketPowerLpfTau; +uint16_t VcnClkAverageLpfTau; +uint16_t padding16; +// Padding - ignore uint32_t MmHubPadding[8 ];// SMU internal use } DriverSmuConfig_t;
@@ -767,9 +772,12 @@ typedef struct {uint32_t ThrottlerStatus ;uint16_t CurrFanSpeed ; -uint16_t Padding16; +uint16_t AverageVclkFrequency ; +uint16_t AverageDclkFrequency ; +uint16_t VcnActivityPercentage ; +uint32_t EnergyAccumulator ; -uint32_t Padding[4 ]; +uint32_t Padding[2 ];// Padding - ignore uint32_t MmHubPadding[8 ];// SMU internal use
drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h +1 -1
@@ -26,7 +26,7 @@#include "amdgpu_smu.h" #define SMU11_DRIVER_IF_VERSION_INV 0xFFFFFFFF -#define SMU11_DRIVER_IF_VERSION_ARCT 0x14 +#define SMU11_DRIVER_IF_VERSION_ARCT 0x17#define SMU11_DRIVER_IF_VERSION_NV10 0x36#define SMU11_DRIVER_IF_VERSION_NV12 0x33#define SMU11_DRIVER_IF_VERSION_NV14 0x36
Signature · 6 lines
2.26.2 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 2/4 drm/amd/powerplay: implement ASIC specific thermal throttling logging
Enable this for Arcturus only for now.
--- drivers/gpu/drm/amd/powerplay/arcturus_ppt.c | 44 +++++++++++++++++++ .../gpu/drm/amd/powerplay/inc/amdgpu_smu.h | 1 + drivers/gpu/drm/amd/powerplay/smu_internal.h | 2 + 3 files changed, 47 insertions(+)
drivers/gpu/drm/amd/powerplay/arcturus_ppt.c +44 -0
@@ -2329,6 +2329,49 @@ static int arcturus_allow_xgmi_power_down(struct smu_context *smu, bool en) NULL); } +static const struct throttling_logging_label { +uint32_t feature_mask; +const char *label; +} logging_label[] = { + {(1U <<THROTTLER_TEMP_HOTSPOT_BIT ),"GPU" }, + {(1U <<THROTTLER_TEMP_MEM_BIT ),"HBM" }, + {(1U <<THROTTLER_TEMP_VR_GFX_BIT ),"VR of GFX rail" }, + {(1U <<THROTTLER_TEMP_VR_MEM_BIT ),"VR of HBM rail" }, + {(1U <<THROTTLER_TEMP_VR_SOC_BIT ),"VR of SOC rail" }, + {(1U <<THROTTLER_VRHOT0_BIT ),"VR0 HOT" }, + {(1U <<THROTTLER_VRHOT1_BIT ),"VR1 HOT" }, +}; +static void arcturus_log_thermal_throttling_event (struct smu_context *smu) +{ +int throttler_idx, throtting_events =0 , buf_idx =0 ; +struct amdgpu_device *adev = smu->adev; +SmuMetrics_t metrics; +char log_buf[256 ]; + +arcturus_get_metrics_table (smu, &metrics); + +memset (log_buf,0 ,sizeof (log_buf)); +for (throttler_idx =0 ; throttler_idx <ARRAY_SIZE (logging_label); + throttler_idx++) { +if (metrics.ThrottlerStatus & logging_label[throttler_idx].feature_mask) { + throtting_events++; + buf_idx +=snprintf (log_buf + buf_idx, +sizeof (log_buf) - buf_idx, +"%s%s" , + throtting_events >1 ?" and " :"" , + logging_label[throttler_idx].label); +if (buf_idx >=sizeof (log_buf)) { +pr_err ("buffer overflow!\n" ); + log_buf[sizeof (log_buf) -1 ] ='\0' ; +break ; + } + } + } + +dev_warn (adev->dev,"WARN: GPU thermal throttling temperature reached, expect performance decrease. %s.\n" , + log_buf); +} +static const struct pptable_funcs arcturus_ppt_funcs = {/* translate smu index into arcturus specific index */ .get_smu_msg_index = arcturus_get_smu_msg_index,
@@ -2423,6 +2466,7 @@ static const struct pptable_funcs arcturus_ppt_funcs = { .get_pptable_power_limit = arcturus_get_pptable_power_limit, .set_df_cstate = arcturus_set_df_cstate, .allow_xgmi_power_down = arcturus_allow_xgmi_power_down, + .log_thermal_throttling_event = arcturus_log_thermal_throttling_event, };void arcturus_set_ppt_funcs (struct smu_context *smu)
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +1 -0
@@ -575,6 +575,7 @@ struct pptable_funcs {uint32_t (*get_pptable_power_limit)(struct smu_context *smu);int (*disable_umc_cdr_12gbps_workaround)(struct smu_context *smu);int (*set_power_source)(struct smu_context *smu,enum smu_power_src_type power_src); +void (*log_thermal_throttling_event)(struct smu_context *smu); };int smu_load_microcode (struct smu_context *smu);
drivers/gpu/drm/amd/powerplay/smu_internal.h +2 -0
@@ -219,4 +219,6 @@ static inline int smu_send_smc_msg(struct smu_context *smu, enum smu_message_typ#define smu_i2c_eeprom_fini (smu, control) \ ((smu)->ppt_funcs->i2c_eeprom_fini ? (smu)->ppt_funcs->i2c_eeprom_fini((control)) : 0) +#define smu_log_thermal_throttling (smu) \ + ((smu)->ppt_funcs->log_thermal_throttling_event ? (smu)->ppt_funcs->log_thermal_throttling_event((smu)) : 0)#endif
Signature · 6 lines
2.26.2 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 3/4 drm/amd/powerplay: enable thermal throttling logging support V2
In reply to Evan QuanCurrently this feature is supported on Arcturus only. PMFW will interrupt driver the first time when thermal throttling happened and every one second afterwards if the throttling continuing. On receiving the 1st interrupt, driver logs it the first time. However, if the throttling continues, the logging will be performed every minute to avoid log flooding.
V2: simplify the implemention by ratelimited printk
--- drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 11 +++++++++++ 1 file changed, 11 insertions(+)
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +11 -0
@@ -1524,6 +1524,7 @@ static int smu_v11_0_irq_process(struct amdgpu_device *adev,struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { + struct smu_context *smu = &adev->smu;uint32_t client_id = entry->client_id;uint32_t src_id = entry->src_id;/*
@@ -1532,6 +1533,11 @@ static int smu_v11_0_irq_process(struct amdgpu_device *adev,*/ uint32_t ctxid = entry->src_data[0 ];uint32_t data; +/* +* if the throttling continues, the logging will be performed every +* minute to avoid log flooding. +*/ + static DEFINE_RATELIMIT_STATE(ratelimit_state, 60 * HZ, 1);if (client_id ==SOC15_IH_CLIENTID_THM ) {switch (src_id) {
@@ -1575,6 +1581,11 @@ static int smu_v11_0_irq_process(struct amdgpu_device *adev,dev_dbg (adev->dev,"Switched to DC mode!\n" );smu_v11_0_ack_ac_dc_interrupt (&adev->smu);break ; +case 0x7 : +if (__ratelimit (&ratelimit_state)) +smu_log_thermal_throttling (smu); + +break ; } } }
Signature · 6 lines
2.26.2 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
PATCH 4/4 drm/amdgpu: added a sysfs interface for thermal throttling related V3
In reply to Evan QuanUser can check and set the enablement of throttling logging and the interval between each logging.
V2: simplify the sysfs interface(no string parsing)
V3: add proper lock protection on updating throttling_logging_rs.interval
--- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 3 + drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 11 ++++ drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 71 ++++++++++++++++++++++ drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 10 ++- 4 files changed, 89 insertions(+), 6 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu.h +3 -0
@@ -996,6 +996,9 @@ struct amdgpu_device {char serial[16 ];struct amdgpu_autodump autodump; + +atomic_t throttling_logging_enabled; +struct ratelimit_state throttling_logging_rs; };static inline struct amdgpu_device *amdgpu_ttm_adev (struct ttm_bo_device *bdev)
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +11 -0
@@ -3035,6 +3035,17 @@ int amdgpu_device_init(struct amdgpu_device *adev, adev->gfx.gfx_off_req_count =1 ; adev->pm.ac_power =power_supply_is_system_supplied () >0 ; +atomic_set (&adev->throttling_logging_enabled,1 ); +/* +* if the throttling continues, the logging will be performed every +* minute to avoid log flooding. "-1" applied here due to the thermal +* throttling interrupt comes every 1 second. So, the total logging +* interval is 59 seconds(retelimited printk interval) + 1(waiting +* for throttling interrup) = 60 seconds. +*/ +ratelimit_state_init (&adev->throttling_logging_rs, (60 -1 ) *HZ ,1 ); +ratelimit_set_flags (&adev->throttling_logging_rs,RATELIMIT_MSG_ON_RELEASE ); +/* Registers mapping */ /* TODO: block userspace mapping of io register */ if (adev->asic_type >=CHIP_BONAIRE ) {
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +71 -0
83 lines not on this page · raw message
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +4 -6
@@ -1533,11 +1533,6 @@ static int smu_v11_0_irq_process(struct amdgpu_device *adev,*/ uint32_t ctxid = entry->src_data[0 ];uint32_t data; -/* -* if the throttling continues, the logging will be performed every -* minute to avoid log flooding. -*/ - static DEFINE_RATELIMIT_STATE(ratelimit_state, 60 * HZ, 1);if (client_id ==SOC15_IH_CLIENTID_THM ) {switch (src_id) {
@@ -1582,7 +1577,10 @@ static int smu_v11_0_irq_process(struct amdgpu_device *adev,smu_v11_0_ack_ac_dc_interrupt (&adev->smu);break ;case 0x7 : -if (__ratelimit (&ratelimit_state)) +if (!atomic_read (&adev->throttling_logging_enabled)) +return 0 ; + +if (__ratelimit (&adev->throttling_logging_rs))smu_log_thermal_throttling (smu);break ;
Signature · 6 lines
2.26.2 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx