* [PATCH 01/16] drm/amd/powerplay: add more members for dpm table
@ 2020-07-10 4:47 Evan Quan
2020-07-10 4:47 ` [PATCH 02/16] drm/amd/powerplay: implement a common set dpm table API for smu V11 Evan Quan
` (14 more replies)
0 siblings, 15 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
These members can help to cache the clock frequencies for all
dpm levels. Then simplifying the code for dpm level switching
is possible.
Change-Id: Ic80359adb8c0e018f306782f24e3f8906436f5e2
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h
index 3d746b75396e..289c571d6e4e 100644
--- a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h
+++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h
@@ -48,6 +48,7 @@
#define SMU11_TOOL_SIZE 0x19000
+#define MAX_DPM_LEVELS 16
#define MAX_PCIE_CONF 2
#define CLK_MAP(clk, index) \
@@ -91,9 +92,17 @@ struct smu_11_0_max_sustainable_clocks {
uint32_t soc_clock;
};
+struct smu_11_0_dpm_clk_level {
+ bool enabled;
+ uint32_t value;
+};
+
struct smu_11_0_dpm_table {
- uint32_t min; /* MHz */
- uint32_t max; /* MHz */
+ uint32_t min; /* MHz */
+ uint32_t max; /* MHz */
+ uint32_t count;
+ bool is_fine_grained;
+ struct smu_11_0_dpm_clk_level dpm_levels[MAX_DPM_LEVELS];
};
struct smu_11_0_pcie_table {
@@ -107,7 +116,9 @@ struct smu_11_0_dpm_tables {
struct smu_11_0_dpm_table uclk_table;
struct smu_11_0_dpm_table eclk_table;
struct smu_11_0_dpm_table vclk_table;
+ struct smu_11_0_dpm_table vclk1_table;
struct smu_11_0_dpm_table dclk_table;
+ struct smu_11_0_dpm_table dclk1_table;
struct smu_11_0_dpm_table dcef_table;
struct smu_11_0_dpm_table pixel_table;
struct smu_11_0_dpm_table display_table;
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 02/16] drm/amd/powerplay: implement a common set dpm table API for smu V11
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-13 14:49 ` Alex Deucher
2020-07-10 4:47 ` [PATCH 03/16] drm/amd/powerplay: update Arcturus default dpm table setting Evan Quan
` (13 subsequent siblings)
14 siblings, 1 reply; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
Maximum the code sharing around smu V11.
Change-Id: Ice0a874f3f70457f1012ca566f9f784ff3e9cd94
Signed-off-by: Evan Quan <evan.quan@amd.com>
---
drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h | 4 ++
drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 38 +++++++++++++++++++
2 files changed, 42 insertions(+)
diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h
index 289c571d6e4e..14d6eef8cf17 100644
--- a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h
+++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h
@@ -285,6 +285,10 @@ int smu_v11_0_get_dpm_level_count(struct smu_context *smu,
enum smu_clk_type clk_type,
uint32_t *value);
+int smu_v11_0_set_single_dpm_table(struct smu_context *smu,
+ enum smu_clk_type clk_type,
+ struct smu_11_0_dpm_table *single_dpm_table);
+
int smu_v11_0_get_dpm_level_range(struct smu_context *smu,
enum smu_clk_type clk_type,
uint32_t *min_value,
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index 03be59492af1..7206b9f76042 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
+++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
@@ -1951,6 +1951,44 @@ int smu_v11_0_get_dpm_level_count(struct smu_context *smu,
value);
}
+int smu_v11_0_set_single_dpm_table(struct smu_context *smu,
+ enum smu_clk_type clk_type,
+ struct smu_11_0_dpm_table *single_dpm_table)
+{
+ int ret = 0;
+ uint32_t clk;
+ int i;
+
+ ret = smu_v11_0_get_dpm_level_count(smu,
+ clk_type,
+ &single_dpm_table->count);
+ if (ret) {
+ dev_err(smu->adev->dev, "[%s] failed to get dpm levels!\n", __func__);
+ return ret;
+ }
+
+ for (i = 0; i < single_dpm_table->count; i++) {
+ ret = smu_v11_0_get_dpm_freq_by_index(smu,
+ clk_type,
+ i,
+ &clk);
+ if (ret) {
+ dev_err(smu->adev->dev, "[%s] failed to get dpm freq by index!\n", __func__);
+ return ret;
+ }
+
+ single_dpm_table->dpm_levels[i].value = clk;
+ single_dpm_table->dpm_levels[i].enabled = true;
+
+ if (i == 0)
+ single_dpm_table->min = clk;
+ else if (i == single_dpm_table->count - 1)
+ single_dpm_table->max = clk;
+ }
+
+ return 0;
+}
+
int smu_v11_0_get_dpm_level_range(struct smu_context *smu,
enum smu_clk_type clk_type,
uint32_t *min_value,
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 03/16] drm/amd/powerplay: update Arcturus default dpm table setting
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
2020-07-10 4:47 ` [PATCH 02/16] drm/amd/powerplay: implement a common set dpm table API for smu V11 Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-13 14:49 ` Alex Deucher
2020-07-10 4:47 ` [PATCH 04/16] drm/amd/powerplay: update Navi10 default dpm table setup Evan Quan
` (12 subsequent siblings)
14 siblings, 1 reply; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
Preparing for coming code sharing around performance level
setting.
Change-Id: Iaa77af7a272121503f09ad5fbfbe9dff2d2597b1
Signed-off-by: Evan Quan <evan.quan@amd.com>
---
drivers/gpu/drm/amd/powerplay/arcturus_ppt.c | 297 ++++++++-----------
1 file changed, 119 insertions(+), 178 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
index 5b793e354704..a3747ab4af32 100644
--- a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
@@ -291,7 +291,6 @@ static int arcturus_get_pwr_src_index(struct smu_context *smc, uint32_t index)
return mapping.map_to;
}
-
static int arcturus_get_workload_type(struct smu_context *smu, enum PP_SMC_POWER_PROFILE profile)
{
struct smu_11_0_cmn2aisc_mapping mapping;
@@ -338,23 +337,11 @@ static int arcturus_allocate_dpm_context(struct smu_context *smu)
{
struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
- if (smu_dpm->dpm_context)
- return -EINVAL;
-
- smu_dpm->dpm_context = kzalloc(sizeof(struct arcturus_dpm_table),
+ smu_dpm->dpm_context = kzalloc(sizeof(struct smu_11_0_dpm_context),
GFP_KERNEL);
if (!smu_dpm->dpm_context)
return -ENOMEM;
-
- if (smu_dpm->golden_dpm_context)
- return -EINVAL;
-
- smu_dpm->golden_dpm_context = kzalloc(sizeof(struct arcturus_dpm_table),
- GFP_KERNEL);
- if (!smu_dpm->golden_dpm_context)
- return -ENOMEM;
-
- smu_dpm->dpm_context_size = sizeof(struct arcturus_dpm_table);
+ smu_dpm->dpm_context_size = sizeof(struct smu_11_0_dpm_context);
smu_dpm->dpm_current_power_state = kzalloc(sizeof(struct smu_power_state),
GFP_KERNEL);
@@ -382,119 +369,84 @@ arcturus_get_allowed_feature_mask(struct smu_context *smu,
return 0;
}
-static int
-arcturus_set_single_dpm_table(struct smu_context *smu,
- struct arcturus_single_dpm_table *single_dpm_table,
- PPCLK_e clk_id)
-{
- int ret = 0;
- uint32_t i, num_of_levels = 0, clk;
-
- ret = smu_send_smc_msg_with_param(smu,
- SMU_MSG_GetDpmFreqByIndex,
- (clk_id << 16 | 0xFF),
- &num_of_levels);
- if (ret) {
- dev_err(smu->adev->dev, "[%s] failed to get dpm levels!\n", __func__);
- return ret;
- }
-
- single_dpm_table->count = num_of_levels;
- for (i = 0; i < num_of_levels; i++) {
- ret = smu_send_smc_msg_with_param(smu,
- SMU_MSG_GetDpmFreqByIndex,
- (clk_id << 16 | i),
- &clk);
- if (ret) {
- dev_err(smu->adev->dev, "[%s] failed to get dpm freq by index!\n", __func__);
- return ret;
- }
- single_dpm_table->dpm_levels[i].value = clk;
- single_dpm_table->dpm_levels[i].enabled = true;
- }
- return 0;
-}
-
-static void arcturus_init_single_dpm_state(struct arcturus_dpm_state *dpm_state)
-{
- dpm_state->soft_min_level = 0x0;
- dpm_state->soft_max_level = 0xffff;
- dpm_state->hard_min_level = 0x0;
- dpm_state->hard_max_level = 0xffff;
-}
-
static int arcturus_set_default_dpm_table(struct smu_context *smu)
{
- int ret;
-
- struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
- struct arcturus_dpm_table *dpm_table = NULL;
- struct arcturus_single_dpm_table *single_dpm_table;
-
- dpm_table = smu_dpm->dpm_context;
+ struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
+ PPTable_t *driver_ppt = smu->smu_table.driver_pptable;
+ struct smu_11_0_dpm_table *dpm_table = NULL;
+ int ret = 0;
- /* socclk */
- single_dpm_table = &(dpm_table->soc_table);
+ /* socclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.soc_table;
if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) {
- ret = arcturus_set_single_dpm_table(smu, single_dpm_table,
- PPCLK_SOCCLK);
- if (ret) {
- dev_err(smu->adev->dev, "[%s] failed to get socclk dpm levels!\n", __func__);
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_SOCCLK,
+ dpm_table);
+ if (ret)
return ret;
- }
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_SOCCLK].SnapToDiscrete;
} else {
- single_dpm_table->count = 1;
- single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.socclk / 100;
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.socclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
}
- arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state));
- /* gfxclk */
- single_dpm_table = &(dpm_table->gfx_table);
+ /* gfxclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.gfx_table;
if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT)) {
- ret = arcturus_set_single_dpm_table(smu, single_dpm_table,
- PPCLK_GFXCLK);
- if (ret) {
- dev_err(smu->adev->dev, "[SetupDefaultDpmTable] failed to get gfxclk dpm levels!");
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_GFXCLK,
+ dpm_table);
+ if (ret)
return ret;
- }
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_GFXCLK].SnapToDiscrete;
} else {
- single_dpm_table->count = 1;
- single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100;
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
}
- arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state));
- /* memclk */
- single_dpm_table = &(dpm_table->mem_table);
+ /* memclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.uclk_table;
if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) {
- ret = arcturus_set_single_dpm_table(smu, single_dpm_table,
- PPCLK_UCLK);
- if (ret) {
- dev_err(smu->adev->dev, "[SetupDefaultDpmTable] failed to get memclk dpm levels!");
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_UCLK,
+ dpm_table);
+ if (ret)
return ret;
- }
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_UCLK].SnapToDiscrete;
} else {
- single_dpm_table->count = 1;
- single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.uclk / 100;
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.uclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
}
- arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state));
- /* fclk */
- single_dpm_table = &(dpm_table->fclk_table);
+ /* fclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.fclk_table;
if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_FCLK_BIT)) {
- ret = arcturus_set_single_dpm_table(smu, single_dpm_table,
- PPCLK_FCLK);
- if (ret) {
- dev_err(smu->adev->dev, "[SetupDefaultDpmTable] failed to get fclk dpm levels!");
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_FCLK,
+ dpm_table);
+ if (ret)
return ret;
- }
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_FCLK].SnapToDiscrete;
} else {
- single_dpm_table->count = 1;
- single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.fclk / 100;
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.fclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
}
- arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state));
-
- memcpy(smu_dpm->golden_dpm_context, dpm_table,
- sizeof(struct arcturus_dpm_table));
return 0;
}
@@ -622,7 +574,7 @@ static int arcturus_populate_umd_state_clk(struct smu_context *smu)
static int arcturus_get_clk_table(struct smu_context *smu,
struct pp_clock_levels_with_latency *clocks,
- struct arcturus_single_dpm_table *dpm_table)
+ struct smu_11_0_dpm_table *dpm_table)
{
int i, count;
@@ -824,14 +776,14 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
int i, now, size = 0;
int ret = 0;
struct pp_clock_levels_with_latency clocks;
- struct arcturus_single_dpm_table *single_dpm_table;
+ struct smu_11_0_dpm_table *single_dpm_table;
struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
- struct arcturus_dpm_table *dpm_table = NULL;
+ struct smu_11_0_dpm_context *dpm_context = NULL;
if (amdgpu_ras_intr_triggered())
return snprintf(buf, PAGE_SIZE, "unavailable\n");
- dpm_table = smu_dpm->dpm_context;
+ dpm_context = smu_dpm->dpm_context;
switch (type) {
case SMU_SCLK:
@@ -841,7 +793,7 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
return ret;
}
- single_dpm_table = &(dpm_table->gfx_table);
+ single_dpm_table = &(dpm_context->dpm_tables.gfx_table);
ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table);
if (ret) {
dev_err(smu->adev->dev, "Attempt to get gfx clk levels Failed!");
@@ -868,7 +820,7 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
return ret;
}
- single_dpm_table = &(dpm_table->mem_table);
+ single_dpm_table = &(dpm_context->dpm_tables.uclk_table);
ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table);
if (ret) {
dev_err(smu->adev->dev, "Attempt to get memory clk levels Failed!");
@@ -891,7 +843,7 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
return ret;
}
- single_dpm_table = &(dpm_table->soc_table);
+ single_dpm_table = &(dpm_context->dpm_tables.soc_table);
ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table);
if (ret) {
dev_err(smu->adev->dev, "Attempt to get socclk levels Failed!");
@@ -914,7 +866,7 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
return ret;
}
- single_dpm_table = &(dpm_table->fclk_table);
+ single_dpm_table = &(dpm_context->dpm_tables.fclk_table);
ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table);
if (ret) {
dev_err(smu->adev->dev, "Attempt to get fclk levels Failed!");
@@ -937,20 +889,19 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
return size;
}
-static int arcturus_upload_dpm_level(struct smu_context *smu, bool max,
- uint32_t feature_mask)
+static int arcturus_upload_dpm_level(struct smu_context *smu,
+ bool max,
+ uint32_t feature_mask,
+ uint32_t level)
{
- struct arcturus_single_dpm_table *single_dpm_table;
- struct arcturus_dpm_table *dpm_table =
+ struct smu_11_0_dpm_context *dpm_context =
smu->smu_dpm.dpm_context;
uint32_t freq;
int ret = 0;
if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT) &&
(feature_mask & FEATURE_DPM_GFXCLK_MASK)) {
- single_dpm_table = &(dpm_table->gfx_table);
- freq = max ? single_dpm_table->dpm_state.soft_max_level :
- single_dpm_table->dpm_state.soft_min_level;
+ freq = dpm_context->dpm_tables.gfx_table.dpm_levels[level].value;
ret = smu_send_smc_msg_with_param(smu,
(max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq),
(PPCLK_GFXCLK << 16) | (freq & 0xffff),
@@ -964,9 +915,7 @@ static int arcturus_upload_dpm_level(struct smu_context *smu, bool max,
if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT) &&
(feature_mask & FEATURE_DPM_UCLK_MASK)) {
- single_dpm_table = &(dpm_table->mem_table);
- freq = max ? single_dpm_table->dpm_state.soft_max_level :
- single_dpm_table->dpm_state.soft_min_level;
+ freq = dpm_context->dpm_tables.uclk_table.dpm_levels[level].value;
ret = smu_send_smc_msg_with_param(smu,
(max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq),
(PPCLK_UCLK << 16) | (freq & 0xffff),
@@ -980,9 +929,7 @@ static int arcturus_upload_dpm_level(struct smu_context *smu, bool max,
if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT) &&
(feature_mask & FEATURE_DPM_SOCCLK_MASK)) {
- single_dpm_table = &(dpm_table->soc_table);
- freq = max ? single_dpm_table->dpm_state.soft_max_level :
- single_dpm_table->dpm_state.soft_min_level;
+ freq = dpm_context->dpm_tables.soc_table.dpm_levels[level].value;
ret = smu_send_smc_msg_with_param(smu,
(max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq),
(PPCLK_SOCCLK << 16) | (freq & 0xffff),
@@ -1000,8 +947,8 @@ static int arcturus_upload_dpm_level(struct smu_context *smu, bool max,
static int arcturus_force_clk_levels(struct smu_context *smu,
enum smu_clk_type type, uint32_t mask)
{
- struct arcturus_dpm_table *dpm_table;
- struct arcturus_single_dpm_table *single_dpm_table;
+ struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
+ struct smu_11_0_dpm_table *single_dpm_table = NULL;
uint32_t soft_min_level, soft_max_level;
uint32_t smu_version;
int ret = 0;
@@ -1021,12 +968,9 @@ static int arcturus_force_clk_levels(struct smu_context *smu,
soft_min_level = mask ? (ffs(mask) - 1) : 0;
soft_max_level = mask ? (fls(mask) - 1) : 0;
- dpm_table = smu->smu_dpm.dpm_context;
-
switch (type) {
case SMU_SCLK:
- single_dpm_table = &(dpm_table->gfx_table);
-
+ single_dpm_table = &(dpm_context->dpm_tables.gfx_table);
if (soft_max_level >= single_dpm_table->count) {
dev_err(smu->adev->dev, "Clock level specified %d is over max allowed %d\n",
soft_max_level, single_dpm_table->count - 1);
@@ -1034,18 +978,19 @@ static int arcturus_force_clk_levels(struct smu_context *smu,
break;
}
- single_dpm_table->dpm_state.soft_min_level =
- single_dpm_table->dpm_levels[soft_min_level].value;
- single_dpm_table->dpm_state.soft_max_level =
- single_dpm_table->dpm_levels[soft_max_level].value;
-
- ret = arcturus_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK);
+ ret = arcturus_upload_dpm_level(smu,
+ false,
+ FEATURE_DPM_GFXCLK_MASK,
+ soft_min_level);
if (ret) {
dev_err(smu->adev->dev, "Failed to upload boot level to lowest!\n");
break;
}
- ret = arcturus_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK);
+ ret = arcturus_upload_dpm_level(smu,
+ true,
+ FEATURE_DPM_GFXCLK_MASK,
+ soft_max_level);
if (ret)
dev_err(smu->adev->dev, "Failed to upload dpm max level to highest!\n");
@@ -1256,8 +1201,7 @@ static int arcturus_get_fan_speed_percent(struct smu_context *smu,
return ret;
}
-
-static uint32_t arcturus_find_lowest_dpm_level(struct arcturus_single_dpm_table *table)
+static uint32_t arcturus_find_lowest_dpm_level(struct smu_11_0_dpm_table *table)
{
uint32_t i;
@@ -1274,7 +1218,7 @@ static uint32_t arcturus_find_lowest_dpm_level(struct arcturus_single_dpm_table
}
static uint32_t arcturus_find_highest_dpm_level(struct smu_context *smu,
- struct arcturus_single_dpm_table *table)
+ struct smu_11_0_dpm_table *table)
{
int i = 0;
@@ -1299,34 +1243,33 @@ static uint32_t arcturus_find_highest_dpm_level(struct smu_context *smu,
return i;
}
-
-
static int arcturus_force_dpm_limit_value(struct smu_context *smu, bool highest)
{
- struct arcturus_dpm_table *dpm_table =
- (struct arcturus_dpm_table *)smu->smu_dpm.dpm_context;
+ struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(smu->adev, 0);
uint32_t soft_level;
int ret = 0;
/* gfxclk */
if (highest)
- soft_level = arcturus_find_highest_dpm_level(smu, &(dpm_table->gfx_table));
+ soft_level = arcturus_find_highest_dpm_level(smu, &(dpm_context->dpm_tables.gfx_table));
else
- soft_level = arcturus_find_lowest_dpm_level(&(dpm_table->gfx_table));
+ soft_level = arcturus_find_lowest_dpm_level(&(dpm_context->dpm_tables.gfx_table));
- dpm_table->gfx_table.dpm_state.soft_min_level =
- dpm_table->gfx_table.dpm_state.soft_max_level =
- dpm_table->gfx_table.dpm_levels[soft_level].value;
-
- ret = arcturus_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK);
+ ret = arcturus_upload_dpm_level(smu,
+ false,
+ FEATURE_DPM_GFXCLK_MASK,
+ soft_level);
if (ret) {
dev_err(smu->adev->dev, "Failed to upload boot level to %s!\n",
highest ? "highest" : "lowest");
return ret;
}
- ret = arcturus_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK);
+ ret = arcturus_upload_dpm_level(smu,
+ true,
+ FEATURE_DPM_GFXCLK_MASK,
+ soft_level);
if (ret) {
dev_err(smu->adev->dev, "Failed to upload dpm max level to %s!\n!",
highest ? "highest" : "lowest");
@@ -1345,27 +1288,29 @@ static int arcturus_force_dpm_limit_value(struct smu_context *smu, bool highest)
static int arcturus_unforce_dpm_levels(struct smu_context *smu)
{
- struct arcturus_dpm_table *dpm_table =
- (struct arcturus_dpm_table *)smu->smu_dpm.dpm_context;
+ struct smu_11_0_dpm_context *dpm_context =
+ (struct smu_11_0_dpm_context *)smu->smu_dpm.dpm_context;
struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(smu->adev, 0);
uint32_t soft_min_level, soft_max_level;
int ret = 0;
/* gfxclk */
- soft_min_level = arcturus_find_lowest_dpm_level(&(dpm_table->gfx_table));
- soft_max_level = arcturus_find_highest_dpm_level(smu, &(dpm_table->gfx_table));
- dpm_table->gfx_table.dpm_state.soft_min_level =
- dpm_table->gfx_table.dpm_levels[soft_min_level].value;
- dpm_table->gfx_table.dpm_state.soft_max_level =
- dpm_table->gfx_table.dpm_levels[soft_max_level].value;
-
- ret = arcturus_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK);
+ soft_min_level = arcturus_find_lowest_dpm_level(&(dpm_context->dpm_tables.gfx_table));
+ soft_max_level = arcturus_find_highest_dpm_level(smu, &(dpm_context->dpm_tables.gfx_table));
+
+ ret = arcturus_upload_dpm_level(smu,
+ false,
+ FEATURE_DPM_GFXCLK_MASK,
+ soft_min_level);
if (ret) {
dev_err(smu->adev->dev, "Failed to upload DPM Bootup Levels!");
return ret;
}
- ret = arcturus_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK);
+ ret = arcturus_upload_dpm_level(smu,
+ true,
+ FEATURE_DPM_GFXCLK_MASK,
+ soft_max_level);
if (ret) {
dev_err(smu->adev->dev, "Failed to upload DPM Max Levels!");
return ret;
@@ -1388,18 +1333,15 @@ arcturus_get_profiling_clk_mask(struct smu_context *smu,
uint32_t *mclk_mask,
uint32_t *soc_mask)
{
- struct arcturus_dpm_table *dpm_table =
- (struct arcturus_dpm_table *)smu->smu_dpm.dpm_context;
- struct arcturus_single_dpm_table *gfx_dpm_table;
- struct arcturus_single_dpm_table *mem_dpm_table;
- struct arcturus_single_dpm_table *soc_dpm_table;
+ struct smu_11_0_dpm_context *dpm_context =
+ (struct smu_11_0_dpm_context *)smu->smu_dpm.dpm_context;
+ struct smu_11_0_dpm_table *gfx_dpm_table;
+ struct smu_11_0_dpm_table *mem_dpm_table;
+ struct smu_11_0_dpm_table *soc_dpm_table;
- if (!smu->smu_dpm.dpm_context)
- return -EINVAL;
-
- gfx_dpm_table = &dpm_table->gfx_table;
- mem_dpm_table = &dpm_table->mem_table;
- soc_dpm_table = &dpm_table->soc_table;
+ gfx_dpm_table = &dpm_context->dpm_tables.gfx_table;
+ mem_dpm_table = &dpm_context->dpm_tables.uclk_table;
+ soc_dpm_table = &dpm_context->dpm_tables.soc_table;
*sclk_mask = 0;
*mclk_mask = 0;
@@ -2153,7 +2095,6 @@ static int arcturus_dpm_set_vcn_enable(struct smu_context *smu, bool enable)
return ret;
}
-
static void arcturus_fill_eeprom_i2c_req(SwI2cRequest_t *req, bool write,
uint8_t address, uint32_t numbytes,
uint8_t *data)
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 04/16] drm/amd/powerplay: update Navi10 default dpm table setup
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
2020-07-10 4:47 ` [PATCH 02/16] drm/amd/powerplay: implement a common set dpm table API for smu V11 Evan Quan
2020-07-10 4:47 ` [PATCH 03/16] drm/amd/powerplay: update Arcturus default dpm table setting Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 05/16] drm/amd/powerplay: update Sienna Cichlid " Evan Quan
` (11 subsequent siblings)
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
Cache all clocks levels for every dpm table. They are needed
by other APIs.
Change-Id: I8114cf31e6ec8c9af4578d51749eb213befdcc71
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/powerplay/navi10_ppt.c | 182 ++++++++++++++++++---
1 file changed, 158 insertions(+), 24 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
index d96e8334b5e2..a022e93a487c 100644
--- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
@@ -689,41 +689,175 @@ static int navi10_allocate_dpm_context(struct smu_context *smu)
static int navi10_set_default_dpm_table(struct smu_context *smu)
{
- struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
- struct smu_table_context *table_context = &smu->smu_table;
- struct smu_11_0_dpm_context *dpm_context = smu_dpm->dpm_context;
- PPTable_t *driver_ppt = NULL;
+ struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
+ PPTable_t *driver_ppt = smu->smu_table.driver_pptable;
+ struct smu_11_0_dpm_table *dpm_table;
+ int ret = 0;
int i;
- driver_ppt = table_context->driver_pptable;
-
- dpm_context->dpm_tables.soc_table.min = driver_ppt->FreqTableSocclk[0];
- dpm_context->dpm_tables.soc_table.max = driver_ppt->FreqTableSocclk[NUM_SOCCLK_DPM_LEVELS - 1];
+ /* socclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.soc_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_SOCCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_SOCCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.socclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.gfx_table.min = driver_ppt->FreqTableGfx[0];
- dpm_context->dpm_tables.gfx_table.max = driver_ppt->FreqTableGfx[NUM_GFXCLK_DPM_LEVELS - 1];
+ /* gfxclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.gfx_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_GFXCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_GFXCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.uclk_table.min = driver_ppt->FreqTableUclk[0];
- dpm_context->dpm_tables.uclk_table.max = driver_ppt->FreqTableUclk[NUM_UCLK_DPM_LEVELS - 1];
+ /* uclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.uclk_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_UCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_UCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.uclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.vclk_table.min = driver_ppt->FreqTableVclk[0];
- dpm_context->dpm_tables.vclk_table.max = driver_ppt->FreqTableVclk[NUM_VCLK_DPM_LEVELS - 1];
+ /* vclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.vclk_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_VCN_PG_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_VCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_VCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.vclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.dclk_table.min = driver_ppt->FreqTableDclk[0];
- dpm_context->dpm_tables.dclk_table.max = driver_ppt->FreqTableDclk[NUM_DCLK_DPM_LEVELS - 1];
+ /* dclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.dclk_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_VCN_PG_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_DCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_DCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.dcef_table.min = driver_ppt->FreqTableDcefclk[0];
- dpm_context->dpm_tables.dcef_table.max = driver_ppt->FreqTableDcefclk[NUM_DCEFCLK_DPM_LEVELS - 1];
+ /* dcefclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.dcef_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_DCEFCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_DCEFCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dcefclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.pixel_table.min = driver_ppt->FreqTablePixclk[0];
- dpm_context->dpm_tables.pixel_table.max = driver_ppt->FreqTablePixclk[NUM_PIXCLK_DPM_LEVELS - 1];
+ /* pixelclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.pixel_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_PIXCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_PIXCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dcefclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.display_table.min = driver_ppt->FreqTableDispclk[0];
- dpm_context->dpm_tables.display_table.max = driver_ppt->FreqTableDispclk[NUM_DISPCLK_DPM_LEVELS - 1];
+ /* displayclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.display_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_DISPCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_DISPCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dcefclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.phy_table.min = driver_ppt->FreqTablePhyclk[0];
- dpm_context->dpm_tables.phy_table.max = driver_ppt->FreqTablePhyclk[NUM_PHYCLK_DPM_LEVELS - 1];
+ /* phyclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.phy_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_PHYCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_PHYCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dcefclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
+ /* lclk dpm table setup */
for (i = 0; i < MAX_PCIE_CONF; i++) {
dpm_context->dpm_tables.pcie_table.pcie_gen[i] = driver_ppt->PcieGenSpeed[i];
dpm_context->dpm_tables.pcie_table.pcie_lane[i] = driver_ppt->PcieLaneCount[i];
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 05/16] drm/amd/powerplay: update Sienna Cichlid default dpm table setup
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (2 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 04/16] drm/amd/powerplay: update Navi10 default dpm table setup Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 06/16] drm/amd/powerplay: add new UMD pstate data structure Evan Quan
` (10 subsequent siblings)
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
Cache all clocks levels for every dpm table. They are needed
by other APIs.
Change-Id: Idaa853356720e48ab3279f420ba1ae18bb7de4fd
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
.../drm/amd/powerplay/sienna_cichlid_ppt.c | 234 ++++++++++++++++--
1 file changed, 211 insertions(+), 23 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
index f2bbe56798d7..d750d06378e9 100644
--- a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
@@ -597,41 +597,229 @@ static int sienna_cichlid_allocate_dpm_context(struct smu_context *smu)
static int sienna_cichlid_set_default_dpm_table(struct smu_context *smu)
{
- struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
- struct smu_table_context *table_context = &smu->smu_table;
- struct smu_11_0_dpm_context *dpm_context = smu_dpm->dpm_context;
- PPTable_t *driver_ppt = NULL;
+ struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
+ PPTable_t *driver_ppt = smu->smu_table.driver_pptable;
+ struct smu_11_0_dpm_table *dpm_table;
+ int ret = 0;
int i;
- driver_ppt = table_context->driver_pptable;
+ /* socclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.soc_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_SOCCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_SOCCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.socclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
+
+ /* gfxclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.gfx_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_GFXCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_GFXCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
+
+ /* uclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.uclk_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_UCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_UCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.uclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.soc_table.min = driver_ppt->FreqTableSocclk[0];
- dpm_context->dpm_tables.soc_table.max = driver_ppt->FreqTableSocclk[NUM_SOCCLK_DPM_LEVELS - 1];
+ /* fclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.fclk_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_FCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_FCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_FCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.fclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.gfx_table.min = driver_ppt->FreqTableGfx[0];
- dpm_context->dpm_tables.gfx_table.max = driver_ppt->FreqTableGfx[NUM_GFXCLK_DPM_LEVELS - 1];
+ /* vclk0 dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.vclk_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_MM_DPM_PG_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_VCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_VCLK_0].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.vclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.uclk_table.min = driver_ppt->FreqTableUclk[0];
- dpm_context->dpm_tables.uclk_table.max = driver_ppt->FreqTableUclk[NUM_UCLK_DPM_LEVELS - 1];
+ /* vclk1 dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.vclk1_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_MM_DPM_PG_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_VCLK1,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_VCLK_1].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.vclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.vclk_table.min = driver_ppt->FreqTableVclk[0];
- dpm_context->dpm_tables.vclk_table.max = driver_ppt->FreqTableVclk[NUM_VCLK_DPM_LEVELS - 1];
+ /* dclk0 dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.dclk_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_MM_DPM_PG_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_DCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_DCLK_0].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.dclk_table.min = driver_ppt->FreqTableDclk[0];
- dpm_context->dpm_tables.dclk_table.max = driver_ppt->FreqTableDclk[NUM_DCLK_DPM_LEVELS - 1];
+ /* dclk1 dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.dclk1_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_MM_DPM_PG_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_DCLK1,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_DCLK_1].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.dcef_table.min = driver_ppt->FreqTableDcefclk[0];
- dpm_context->dpm_tables.dcef_table.max = driver_ppt->FreqTableDcefclk[NUM_DCEFCLK_DPM_LEVELS - 1];
+ /* dcefclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.dcef_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_DCEFCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_DCEFCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dcefclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.pixel_table.min = driver_ppt->FreqTablePixclk[0];
- dpm_context->dpm_tables.pixel_table.max = driver_ppt->FreqTablePixclk[NUM_PIXCLK_DPM_LEVELS - 1];
+ /* pixelclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.pixel_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_PIXCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_PIXCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dcefclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.display_table.min = driver_ppt->FreqTableDispclk[0];
- dpm_context->dpm_tables.display_table.max = driver_ppt->FreqTableDispclk[NUM_DISPCLK_DPM_LEVELS - 1];
+ /* displayclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.display_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_DISPCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_DISPCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dcefclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
- dpm_context->dpm_tables.phy_table.min = driver_ppt->FreqTablePhyclk[0];
- dpm_context->dpm_tables.phy_table.max = driver_ppt->FreqTablePhyclk[NUM_PHYCLK_DPM_LEVELS - 1];
+ /* phyclk dpm table setup */
+ dpm_table = &dpm_context->dpm_tables.phy_table;
+ if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_DCEFCLK_BIT)) {
+ ret = smu_v11_0_set_single_dpm_table(smu,
+ SMU_PHYCLK,
+ dpm_table);
+ if (ret)
+ return ret;
+ dpm_table->is_fine_grained =
+ !driver_ppt->DpmDescriptor[PPCLK_PHYCLK].SnapToDiscrete;
+ } else {
+ dpm_table->count = 1;
+ dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.dcefclk / 100;
+ dpm_table->dpm_levels[0].enabled = true;
+ dpm_table->min = dpm_table->dpm_levels[0].value;
+ dpm_table->max = dpm_table->dpm_levels[0].value;
+ }
+ /* lclk dpm table setup */
for (i = 0; i < MAX_PCIE_CONF; i++) {
dpm_context->dpm_tables.pcie_table.pcie_gen[i] = driver_ppt->PcieGenSpeed[i];
dpm_context->dpm_tables.pcie_table.pcie_lane[i] = driver_ppt->PcieLaneCount[i];
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 06/16] drm/amd/powerplay: add new UMD pstate data structure
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (3 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 05/16] drm/amd/powerplay: update Sienna Cichlid " Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 07/16] drm/amd/powerplay: update UMD pstate clock settings Evan Quan
` (9 subsequent siblings)
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
This is used to cache the clock frequencies for all UMD pstates.
So that we do not need to calculate from scratch on every UMD
pstate switch.
Change-Id: I3f2ef5ee2e6e433518f726988bbe5970848b99c8
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h | 15 +++++++++++++++
1 file changed, 15 insertions(+)
diff --git a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h
index 66912884f093..91c8b69da026 100644
--- a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h
+++ b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h
@@ -352,6 +352,20 @@ struct smu_baco_context
bool platform_support;
};
+struct pstates_clk_freq {
+ uint32_t min;
+ uint32_t standard;
+ uint32_t peak;
+};
+
+struct smu_umd_pstate_table {
+ struct pstates_clk_freq gfxclk_pstate;
+ struct pstates_clk_freq socclk_pstate;
+ struct pstates_clk_freq uclk_pstate;
+ struct pstates_clk_freq vclk_pstate;
+ struct pstates_clk_freq dclk_pstate;
+};
+
#define WORKLOAD_POLICY_MAX 7
struct smu_context
{
@@ -376,6 +390,7 @@ struct smu_context
struct dentry *debugfs_sclk;
#endif
+ struct smu_umd_pstate_table pstate_table;
uint32_t pstate_sclk;
uint32_t pstate_mclk;
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 07/16] drm/amd/powerplay: update UMD pstate clock settings
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (4 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 06/16] drm/amd/powerplay: add new UMD pstate data structure Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 08/16] drm/amd/powerplay: update the common API for performance level setting Evan Quan
` (8 subsequent siblings)
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
Preparing for coming code sharing around performance level
setting.
Change-Id: I51b1536b62995f0fecd51b91f238793f57485aa9
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/powerplay/amdgpu_smu.c | 6 +-
drivers/gpu/drm/amd/powerplay/arcturus_ppt.c | 47 ++++++---
drivers/gpu/drm/amd/powerplay/navi10_ppt.c | 99 ++++++++++++++++---
.../drm/amd/powerplay/sienna_cichlid_ppt.c | 35 ++++---
4 files changed, 141 insertions(+), 46 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
index 6839faaab611..4080b3c792ac 100644
--- a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
+++ b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
@@ -2168,6 +2168,8 @@ int smu_read_sensor(struct smu_context *smu,
enum amd_pp_sensors sensor,
void *data, uint32_t *size)
{
+ struct smu_umd_pstate_table *pstate_table =
+ &smu->pstate_table;
int ret = 0;
if (!smu->pm_enabled || !smu->adev->pm.dpm_enabled)
@@ -2180,11 +2182,11 @@ int smu_read_sensor(struct smu_context *smu,
switch (sensor) {
case AMDGPU_PP_SENSOR_STABLE_PSTATE_SCLK:
- *((uint32_t *)data) = smu->pstate_sclk;
+ *((uint32_t *)data) = pstate_table->gfxclk_pstate.standard * 100;
*size = 4;
break;
case AMDGPU_PP_SENSOR_STABLE_PSTATE_MCLK:
- *((uint32_t *)data) = smu->pstate_mclk;
+ *((uint32_t *)data) = pstate_table->uclk_pstate.standard * 100;
*size = 4;
break;
case AMDGPU_PP_SENSOR_ENABLED_SMC_FEATURES_MASK:
diff --git a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
index a3747ab4af32..33d472ffb2be 100644
--- a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
@@ -548,27 +548,44 @@ static int arcturus_run_btc(struct smu_context *smu)
static int arcturus_populate_umd_state_clk(struct smu_context *smu)
{
- struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
- struct arcturus_dpm_table *dpm_table = NULL;
- struct arcturus_single_dpm_table *gfx_table = NULL;
- struct arcturus_single_dpm_table *mem_table = NULL;
+ struct smu_11_0_dpm_context *dpm_context =
+ smu->smu_dpm.dpm_context;
+ struct smu_11_0_dpm_table *gfx_table =
+ &dpm_context->dpm_tables.gfx_table;
+ struct smu_11_0_dpm_table *mem_table =
+ &dpm_context->dpm_tables.uclk_table;
+ struct smu_11_0_dpm_table *soc_table =
+ &dpm_context->dpm_tables.soc_table;
+ struct smu_umd_pstate_table *pstate_table =
+ &smu->pstate_table;
+
+ pstate_table->gfxclk_pstate.min = gfx_table->min;
+ pstate_table->gfxclk_pstate.peak = gfx_table->max;
- dpm_table = smu_dpm->dpm_context;
- gfx_table = &(dpm_table->gfx_table);
- mem_table = &(dpm_table->mem_table);
+ pstate_table->uclk_pstate.min = mem_table->min;
+ pstate_table->uclk_pstate.peak = mem_table->max;
- smu->pstate_sclk = gfx_table->dpm_levels[0].value;
- smu->pstate_mclk = mem_table->dpm_levels[0].value;
+ pstate_table->socclk_pstate.min = soc_table->min;
+ pstate_table->socclk_pstate.peak = soc_table->max;
if (gfx_table->count > ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL &&
- mem_table->count > ARCTURUS_UMD_PSTATE_MCLK_LEVEL) {
- smu->pstate_sclk = gfx_table->dpm_levels[ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL].value;
- smu->pstate_mclk = mem_table->dpm_levels[ARCTURUS_UMD_PSTATE_MCLK_LEVEL].value;
+ mem_table->count > ARCTURUS_UMD_PSTATE_MCLK_LEVEL &&
+ soc_table->count > ARCTURUS_UMD_PSTATE_SOCCLK_LEVEL) {
+ pstate_table->gfxclk_pstate.standard =
+ gfx_table->dpm_levels[ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL].value;
+ pstate_table->uclk_pstate.standard =
+ mem_table->dpm_levels[ARCTURUS_UMD_PSTATE_MCLK_LEVEL].value;
+ pstate_table->socclk_pstate.standard =
+ soc_table->dpm_levels[ARCTURUS_UMD_PSTATE_SOCCLK_LEVEL].value;
+ } else {
+ pstate_table->gfxclk_pstate.standard =
+ pstate_table->gfxclk_pstate.min;
+ pstate_table->uclk_pstate.standard =
+ pstate_table->uclk_pstate.min;
+ pstate_table->socclk_pstate.standard =
+ pstate_table->socclk_pstate.min;
}
- smu->pstate_sclk = smu->pstate_sclk * 100;
- smu->pstate_mclk = smu->pstate_mclk * 100;
-
return 0;
}
diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
index a022e93a487c..d3e11d81c0ad 100644
--- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
@@ -1213,22 +1213,93 @@ static int navi10_force_clk_levels(struct smu_context *smu,
static int navi10_populate_umd_state_clk(struct smu_context *smu)
{
- int ret = 0;
- uint32_t min_sclk_freq = 0, min_mclk_freq = 0;
-
- ret = smu_v11_0_get_dpm_ultimate_freq(smu, SMU_SCLK, &min_sclk_freq, NULL);
- if (ret)
- return ret;
-
- smu->pstate_sclk = min_sclk_freq * 100;
-
- ret = smu_v11_0_get_dpm_ultimate_freq(smu, SMU_MCLK, &min_mclk_freq, NULL);
- if (ret)
- return ret;
+ struct smu_11_0_dpm_context *dpm_context =
+ smu->smu_dpm.dpm_context;
+ struct smu_11_0_dpm_table *gfx_table =
+ &dpm_context->dpm_tables.gfx_table;
+ struct smu_11_0_dpm_table *mem_table =
+ &dpm_context->dpm_tables.uclk_table;
+ struct smu_11_0_dpm_table *soc_table =
+ &dpm_context->dpm_tables.soc_table;
+ struct smu_umd_pstate_table *pstate_table =
+ &smu->pstate_table;
+ struct amdgpu_device *adev = smu->adev;
+ uint32_t sclk_freq;
- smu->pstate_mclk = min_mclk_freq * 100;
+ pstate_table->gfxclk_pstate.min = gfx_table->min;
+ switch (adev->asic_type) {
+ case CHIP_NAVI10:
+ switch (adev->pdev->revision) {
+ case 0xf0: /* XTX */
+ case 0xc0:
+ sclk_freq = NAVI10_PEAK_SCLK_XTX;
+ break;
+ case 0xf1: /* XT */
+ case 0xc1:
+ sclk_freq = NAVI10_PEAK_SCLK_XT;
+ break;
+ default: /* XL */
+ sclk_freq = NAVI10_PEAK_SCLK_XL;
+ break;
+ }
+ break;
+ case CHIP_NAVI14:
+ switch (adev->pdev->revision) {
+ case 0xc7: /* XT */
+ case 0xf4:
+ sclk_freq = NAVI14_UMD_PSTATE_PEAK_XT_GFXCLK;
+ break;
+ case 0xc1: /* XTM */
+ case 0xf2:
+ sclk_freq = NAVI14_UMD_PSTATE_PEAK_XTM_GFXCLK;
+ break;
+ case 0xc3: /* XLM */
+ case 0xf3:
+ sclk_freq = NAVI14_UMD_PSTATE_PEAK_XLM_GFXCLK;
+ break;
+ case 0xc5: /* XTX */
+ case 0xf6:
+ sclk_freq = NAVI14_UMD_PSTATE_PEAK_XLM_GFXCLK;
+ break;
+ default: /* XL */
+ sclk_freq = NAVI14_UMD_PSTATE_PEAK_XL_GFXCLK;
+ break;
+ }
+ break;
+ case CHIP_NAVI12:
+ sclk_freq = NAVI12_UMD_PSTATE_PEAK_GFXCLK;
+ break;
+ default:
+ sclk_freq = gfx_table->dpm_levels[gfx_table->count - 1].value;
+ break;
+ }
+ pstate_table->gfxclk_pstate.peak = sclk_freq;
+
+ pstate_table->uclk_pstate.min = mem_table->min;
+ pstate_table->uclk_pstate.peak = mem_table->max;
+
+ pstate_table->socclk_pstate.min = soc_table->min;
+ pstate_table->socclk_pstate.peak = soc_table->max;
+
+ if (gfx_table->max > NAVI10_UMD_PSTATE_PROFILING_GFXCLK &&
+ mem_table->max > NAVI10_UMD_PSTATE_PROFILING_MEMCLK &&
+ soc_table->max > NAVI10_UMD_PSTATE_PROFILING_SOCCLK) {
+ pstate_table->gfxclk_pstate.standard =
+ NAVI10_UMD_PSTATE_PROFILING_GFXCLK;
+ pstate_table->uclk_pstate.standard =
+ NAVI10_UMD_PSTATE_PROFILING_MEMCLK;
+ pstate_table->socclk_pstate.standard =
+ NAVI10_UMD_PSTATE_PROFILING_SOCCLK;
+ } else {
+ pstate_table->gfxclk_pstate.standard =
+ pstate_table->gfxclk_pstate.min;
+ pstate_table->uclk_pstate.standard =
+ pstate_table->uclk_pstate.min;
+ pstate_table->socclk_pstate.standard =
+ pstate_table->socclk_pstate.min;
+ }
- return ret;
+ return 0;
}
static int navi10_get_clock_by_type_with_latency(struct smu_context *smu,
diff --git a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
index d750d06378e9..8fae7dd982c7 100644
--- a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
@@ -1114,22 +1114,27 @@ static int sienna_cichlid_force_clk_levels(struct smu_context *smu,
static int sienna_cichlid_populate_umd_state_clk(struct smu_context *smu)
{
- int ret = 0;
- uint32_t min_sclk_freq = 0, min_mclk_freq = 0;
-
- ret = smu_v11_0_get_dpm_ultimate_freq(smu, SMU_SCLK, &min_sclk_freq, NULL);
- if (ret)
- return ret;
-
- smu->pstate_sclk = min_sclk_freq * 100;
+ struct smu_11_0_dpm_context *dpm_context =
+ smu->smu_dpm.dpm_context;
+ struct smu_11_0_dpm_table *gfx_table =
+ &dpm_context->dpm_tables.gfx_table;
+ struct smu_11_0_dpm_table *mem_table =
+ &dpm_context->dpm_tables.uclk_table;
+ struct smu_11_0_dpm_table *soc_table =
+ &dpm_context->dpm_tables.soc_table;
+ struct smu_umd_pstate_table *pstate_table =
+ &smu->pstate_table;
+
+ pstate_table->gfxclk_pstate.min = gfx_table->min;
+ pstate_table->gfxclk_pstate.peak = gfx_table->max;
+
+ pstate_table->uclk_pstate.min = mem_table->min;
+ pstate_table->uclk_pstate.peak = mem_table->max;
+
+ pstate_table->socclk_pstate.min = soc_table->min;
+ pstate_table->socclk_pstate.peak = soc_table->max;
- ret = smu_v11_0_get_dpm_ultimate_freq(smu, SMU_MCLK, &min_mclk_freq, NULL);
- if (ret)
- return ret;
-
- smu->pstate_mclk = min_mclk_freq * 100;
-
- return ret;
+ return 0;
}
static int sienna_cichlid_pre_display_config_changed(struct smu_context *smu)
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 08/16] drm/amd/powerplay: update the common API for performance level setting
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (5 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 07/16] drm/amd/powerplay: update UMD pstate clock settings Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 09/16] drm/amd/powerplay: drop unnecessary Arcturus specific APIs Evan Quan
` (7 subsequent siblings)
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
So that it can be more widely shared around SMU v11 ASICs.
Change-Id: Ie110edf2ec519699448d3ff3215188ba243d2415
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 91 +++++++++++++++++++----
1 file changed, 77 insertions(+), 14 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index 7206b9f76042..f96ff062eb64 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
+++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
@@ -1855,38 +1855,101 @@ int smu_v11_0_override_pcie_parameters(struct smu_context *smu)
int smu_v11_0_set_performance_level(struct smu_context *smu,
enum amd_dpm_forced_level level)
{
+ struct smu_11_0_dpm_context *dpm_context =
+ smu->smu_dpm.dpm_context;
+ struct smu_11_0_dpm_table *gfx_table =
+ &dpm_context->dpm_tables.gfx_table;
+ struct smu_11_0_dpm_table *mem_table =
+ &dpm_context->dpm_tables.uclk_table;
+ struct smu_11_0_dpm_table *soc_table =
+ &dpm_context->dpm_tables.soc_table;
+ struct smu_umd_pstate_table *pstate_table =
+ &smu->pstate_table;
+ struct amdgpu_device *adev = smu->adev;
+ uint32_t sclk_min = 0, sclk_max = 0;
+ uint32_t mclk_min = 0, mclk_max = 0;
+ uint32_t socclk_min = 0, socclk_max = 0;
int ret = 0;
- uint32_t sclk_mask, mclk_mask, soc_mask;
switch (level) {
case AMD_DPM_FORCED_LEVEL_HIGH:
- ret = smu_force_dpm_limit_value(smu, true);
+ sclk_min = sclk_max = gfx_table->max;
+ mclk_min = mclk_max = mem_table->max;
+ socclk_min = socclk_max = soc_table->max;
break;
case AMD_DPM_FORCED_LEVEL_LOW:
- ret = smu_force_dpm_limit_value(smu, false);
+ sclk_min = sclk_max = gfx_table->min;
+ mclk_min = mclk_max = mem_table->min;
+ socclk_min = socclk_max = soc_table->min;
break;
case AMD_DPM_FORCED_LEVEL_AUTO:
+ sclk_min = gfx_table->min;
+ sclk_max = gfx_table->max;
+ mclk_min = mem_table->min;
+ mclk_max = mem_table->max;
+ socclk_min = soc_table->min;
+ socclk_max = soc_table->max;
+ break;
case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD:
- ret = smu_unforce_dpm_levels(smu);
+ sclk_min = sclk_max = pstate_table->gfxclk_pstate.standard;
+ mclk_min = mclk_max = pstate_table->uclk_pstate.standard;
+ socclk_min = socclk_max = pstate_table->socclk_pstate.standard;
break;
case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK:
+ sclk_min = sclk_max = pstate_table->gfxclk_pstate.min;
+ break;
case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK:
+ mclk_min = mclk_max = pstate_table->uclk_pstate.min;
+ break;
case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK:
- ret = smu_get_profiling_clk_mask(smu, level,
- &sclk_mask,
- &mclk_mask,
- &soc_mask);
- if (ret)
- return ret;
- smu_force_clk_levels(smu, SMU_SCLK, 1 << sclk_mask, false);
- smu_force_clk_levels(smu, SMU_MCLK, 1 << mclk_mask, false);
- smu_force_clk_levels(smu, SMU_SOCCLK, 1 << soc_mask, false);
+ sclk_min = sclk_max = pstate_table->gfxclk_pstate.peak;
+ mclk_min = mclk_max = pstate_table->uclk_pstate.peak;
+ socclk_min = socclk_max = pstate_table->socclk_pstate.peak;
break;
case AMD_DPM_FORCED_LEVEL_MANUAL:
case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT:
+ return 0;
default:
- break;
+ dev_err(adev->dev, "Invalid performance level %d\n", level);
+ return -EINVAL;
+ }
+
+ /*
+ * Separate MCLK and SOCCLK soft min/max settings are not allowed
+ * on Arcturus.
+ */
+ if (adev->asic_type == CHIP_ARCTURUS) {
+ mclk_min = mclk_max = 0;
+ socclk_min = socclk_max = 0;
}
+
+ if (sclk_min && sclk_max) {
+ ret = smu_v11_0_set_soft_freq_limited_range(smu,
+ SMU_GFXCLK,
+ sclk_min,
+ sclk_max);
+ if (ret)
+ return ret;
+ }
+
+ if (mclk_min && mclk_max) {
+ ret = smu_v11_0_set_soft_freq_limited_range(smu,
+ SMU_MCLK,
+ mclk_min,
+ mclk_max);
+ if (ret)
+ return ret;
+ }
+
+ if (socclk_min && socclk_max) {
+ ret = smu_v11_0_set_soft_freq_limited_range(smu,
+ SMU_SOCCLK,
+ socclk_min,
+ socclk_max);
+ if (ret)
+ return ret;
+ }
+
return ret;
}
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 09/16] drm/amd/powerplay: drop unnecessary Arcturus specific APIs
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (6 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 08/16] drm/amd/powerplay: update the common API for performance level setting Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 10/16] drm/amd/powerplay: drop unnecessary Navi1x " Evan Quan
` (6 subsequent siblings)
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
As a common performance level setting API is used. Then these
ASIC specific APIs are not needed any more.
Change-Id: Icd96ce42218d78d670dd0c1f88663fd42108b311
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/powerplay/arcturus_ppt.c | 170 -------------------
1 file changed, 170 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
index 33d472ffb2be..afd07c497205 100644
--- a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
@@ -1218,173 +1218,6 @@ static int arcturus_get_fan_speed_percent(struct smu_context *smu,
return ret;
}
-static uint32_t arcturus_find_lowest_dpm_level(struct smu_11_0_dpm_table *table)
-{
- uint32_t i;
-
- for (i = 0; i < table->count; i++) {
- if (table->dpm_levels[i].enabled)
- break;
- }
- if (i >= table->count) {
- i = 0;
- table->dpm_levels[i].enabled = true;
- }
-
- return i;
-}
-
-static uint32_t arcturus_find_highest_dpm_level(struct smu_context *smu,
- struct smu_11_0_dpm_table *table)
-{
- int i = 0;
-
- if (table->count <= 0) {
- dev_err(smu->adev->dev, "[%s] DPM Table has no entry!", __func__);
- return 0;
- }
- if (table->count > MAX_DPM_NUMBER) {
- dev_err(smu->adev->dev, "[%s] DPM Table has too many entries!", __func__);
- return MAX_DPM_NUMBER - 1;
- }
-
- for (i = table->count - 1; i >= 0; i--) {
- if (table->dpm_levels[i].enabled)
- break;
- }
- if (i < 0) {
- i = 0;
- table->dpm_levels[i].enabled = true;
- }
-
- return i;
-}
-
-static int arcturus_force_dpm_limit_value(struct smu_context *smu, bool highest)
-{
- struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
- struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(smu->adev, 0);
- uint32_t soft_level;
- int ret = 0;
-
- /* gfxclk */
- if (highest)
- soft_level = arcturus_find_highest_dpm_level(smu, &(dpm_context->dpm_tables.gfx_table));
- else
- soft_level = arcturus_find_lowest_dpm_level(&(dpm_context->dpm_tables.gfx_table));
-
- ret = arcturus_upload_dpm_level(smu,
- false,
- FEATURE_DPM_GFXCLK_MASK,
- soft_level);
- if (ret) {
- dev_err(smu->adev->dev, "Failed to upload boot level to %s!\n",
- highest ? "highest" : "lowest");
- return ret;
- }
-
- ret = arcturus_upload_dpm_level(smu,
- true,
- FEATURE_DPM_GFXCLK_MASK,
- soft_level);
- if (ret) {
- dev_err(smu->adev->dev, "Failed to upload dpm max level to %s!\n!",
- highest ? "highest" : "lowest");
- return ret;
- }
-
- if (hive)
- /*
- * Force XGMI Pstate to highest or lowest
- * TODO: revise this when xgmi dpm is functional
- */
- ret = smu_v11_0_set_xgmi_pstate(smu, highest ? 1 : 0);
-
- return ret;
-}
-
-static int arcturus_unforce_dpm_levels(struct smu_context *smu)
-{
- struct smu_11_0_dpm_context *dpm_context =
- (struct smu_11_0_dpm_context *)smu->smu_dpm.dpm_context;
- struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(smu->adev, 0);
- uint32_t soft_min_level, soft_max_level;
- int ret = 0;
-
- /* gfxclk */
- soft_min_level = arcturus_find_lowest_dpm_level(&(dpm_context->dpm_tables.gfx_table));
- soft_max_level = arcturus_find_highest_dpm_level(smu, &(dpm_context->dpm_tables.gfx_table));
-
- ret = arcturus_upload_dpm_level(smu,
- false,
- FEATURE_DPM_GFXCLK_MASK,
- soft_min_level);
- if (ret) {
- dev_err(smu->adev->dev, "Failed to upload DPM Bootup Levels!");
- return ret;
- }
-
- ret = arcturus_upload_dpm_level(smu,
- true,
- FEATURE_DPM_GFXCLK_MASK,
- soft_max_level);
- if (ret) {
- dev_err(smu->adev->dev, "Failed to upload DPM Max Levels!");
- return ret;
- }
-
- if (hive)
- /*
- * Reset XGMI Pstate back to default
- * TODO: revise this when xgmi dpm is functional
- */
- ret = smu_v11_0_set_xgmi_pstate(smu, 0);
-
- return ret;
-}
-
-static int
-arcturus_get_profiling_clk_mask(struct smu_context *smu,
- enum amd_dpm_forced_level level,
- uint32_t *sclk_mask,
- uint32_t *mclk_mask,
- uint32_t *soc_mask)
-{
- struct smu_11_0_dpm_context *dpm_context =
- (struct smu_11_0_dpm_context *)smu->smu_dpm.dpm_context;
- struct smu_11_0_dpm_table *gfx_dpm_table;
- struct smu_11_0_dpm_table *mem_dpm_table;
- struct smu_11_0_dpm_table *soc_dpm_table;
-
- gfx_dpm_table = &dpm_context->dpm_tables.gfx_table;
- mem_dpm_table = &dpm_context->dpm_tables.uclk_table;
- soc_dpm_table = &dpm_context->dpm_tables.soc_table;
-
- *sclk_mask = 0;
- *mclk_mask = 0;
- *soc_mask = 0;
-
- if (gfx_dpm_table->count > ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL &&
- mem_dpm_table->count > ARCTURUS_UMD_PSTATE_MCLK_LEVEL &&
- soc_dpm_table->count > ARCTURUS_UMD_PSTATE_SOCCLK_LEVEL) {
- *sclk_mask = ARCTURUS_UMD_PSTATE_GFXCLK_LEVEL;
- *mclk_mask = ARCTURUS_UMD_PSTATE_MCLK_LEVEL;
- *soc_mask = ARCTURUS_UMD_PSTATE_SOCCLK_LEVEL;
- }
-
- if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) {
- *sclk_mask = 0;
- } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) {
- *mclk_mask = 0;
- } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) {
- *sclk_mask = gfx_dpm_table->count - 1;
- *mclk_mask = mem_dpm_table->count - 1;
- *soc_mask = soc_dpm_table->count - 1;
- }
-
- return 0;
-}
-
static int arcturus_get_power_limit(struct smu_context *smu)
{
struct smu_11_0_powerplay_table *powerplay_table =
@@ -2534,9 +2367,6 @@ static const struct pptable_funcs arcturus_ppt_funcs = {
.read_sensor = arcturus_read_sensor,
.get_fan_speed_percent = arcturus_get_fan_speed_percent,
.get_fan_speed_rpm = arcturus_get_fan_speed_rpm,
- .force_dpm_limit_value = arcturus_force_dpm_limit_value,
- .unforce_dpm_levels = arcturus_unforce_dpm_levels,
- .get_profiling_clk_mask = arcturus_get_profiling_clk_mask,
.get_power_profile_mode = arcturus_get_power_profile_mode,
.set_power_profile_mode = arcturus_set_power_profile_mode,
.set_performance_level = arcturus_set_performance_level,
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 10/16] drm/amd/powerplay: drop unnecessary Navi1x specific APIs
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (7 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 09/16] drm/amd/powerplay: drop unnecessary Arcturus specific APIs Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 11/16] drm/amd/powerplay: drop unnecessary Sienna Cichlid " Evan Quan
` (5 subsequent siblings)
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
As a common performance level setting API is used. Then these
ASIC specific APIs are not needed any more.
Change-Id: I2c8831b9d00618c6578ee42b34e26892c5dba515
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/powerplay/navi10_ppt.c | 248 +--------------------
1 file changed, 1 insertion(+), 247 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
index d3e11d81c0ad..6d638a67bc4d 100644
--- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
@@ -1376,59 +1376,6 @@ static int navi10_display_config_changed(struct smu_context *smu)
return ret;
}
-static int navi10_force_dpm_limit_value(struct smu_context *smu, bool highest)
-{
- int ret = 0, i = 0;
- uint32_t min_freq, max_freq, force_freq;
- enum smu_clk_type clk_type;
-
- enum smu_clk_type clks[] = {
- SMU_GFXCLK,
- SMU_MCLK,
- SMU_SOCCLK,
- };
-
- for (i = 0; i < ARRAY_SIZE(clks); i++) {
- clk_type = clks[i];
- ret = smu_v11_0_get_dpm_ultimate_freq(smu, clk_type, &min_freq, &max_freq);
- if (ret)
- return ret;
-
- force_freq = highest ? max_freq : min_freq;
- ret = smu_v11_0_set_soft_freq_limited_range(smu, clk_type, force_freq, force_freq);
- if (ret)
- return ret;
- }
-
- return ret;
-}
-
-static int navi10_unforce_dpm_levels(struct smu_context *smu)
-{
- int ret = 0, i = 0;
- uint32_t min_freq, max_freq;
- enum smu_clk_type clk_type;
-
- enum smu_clk_type clks[] = {
- SMU_GFXCLK,
- SMU_MCLK,
- SMU_SOCCLK,
- };
-
- for (i = 0; i < ARRAY_SIZE(clks); i++) {
- clk_type = clks[i];
- ret = smu_v11_0_get_dpm_ultimate_freq(smu, clk_type, &min_freq, &max_freq);
- if (ret)
- return ret;
-
- ret = smu_v11_0_set_soft_freq_limited_range(smu, clk_type, min_freq, max_freq);
- if (ret)
- return ret;
- }
-
- return ret;
-}
-
static int navi10_get_gpu_power(struct smu_context *smu, uint32_t *value)
{
if (!value)
@@ -1681,47 +1628,6 @@ static int navi10_set_power_profile_mode(struct smu_context *smu, long *input, u
return ret;
}
-static int navi10_get_profiling_clk_mask(struct smu_context *smu,
- enum amd_dpm_forced_level level,
- uint32_t *sclk_mask,
- uint32_t *mclk_mask,
- uint32_t *soc_mask)
-{
- int ret = 0;
- uint32_t level_count = 0;
-
- if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) {
- if (sclk_mask)
- *sclk_mask = 0;
- } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) {
- if (mclk_mask)
- *mclk_mask = 0;
- } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) {
- if(sclk_mask) {
- ret = smu_v11_0_get_dpm_level_count(smu, SMU_SCLK, &level_count);
- if (ret)
- return ret;
- *sclk_mask = level_count - 1;
- }
-
- if(mclk_mask) {
- ret = smu_v11_0_get_dpm_level_count(smu, SMU_MCLK, &level_count);
- if (ret)
- return ret;
- *mclk_mask = level_count - 1;
- }
-
- if(soc_mask) {
- ret = smu_v11_0_get_dpm_level_count(smu, SMU_SOCCLK, &level_count);
- if (ret)
- return ret;
- *soc_mask = level_count - 1;
- }
- }
-
- return ret;
-}
-
static int navi10_notify_smc_display_config(struct smu_context *smu)
{
struct smu_clocks min_clocks = {0};
@@ -1954,155 +1860,6 @@ static int navi10_get_uclk_dpm_states(struct smu_context *smu, uint32_t *clocks_
return 0;
}
-static int navi10_set_performance_level(struct smu_context *smu,
- enum amd_dpm_forced_level level);
-
-static int navi10_set_standard_performance_level(struct smu_context *smu)
-{
- struct amdgpu_device *adev = smu->adev;
- int ret = 0;
- uint32_t sclk_freq = 0, uclk_freq = 0;
-
- switch (adev->asic_type) {
- case CHIP_NAVI10:
- sclk_freq = NAVI10_UMD_PSTATE_PROFILING_GFXCLK;
- uclk_freq = NAVI10_UMD_PSTATE_PROFILING_MEMCLK;
- break;
- case CHIP_NAVI14:
- sclk_freq = NAVI14_UMD_PSTATE_PROFILING_GFXCLK;
- uclk_freq = NAVI14_UMD_PSTATE_PROFILING_MEMCLK;
- break;
- default:
- /* by default, this is same as auto performance level */
- return navi10_set_performance_level(smu, AMD_DPM_FORCED_LEVEL_AUTO);
- }
-
- ret = smu_v11_0_set_soft_freq_limited_range(smu, SMU_SCLK, sclk_freq, sclk_freq);
- if (ret)
- return ret;
- ret = smu_v11_0_set_soft_freq_limited_range(smu, SMU_UCLK, uclk_freq, uclk_freq);
- if (ret)
- return ret;
-
- return ret;
-}
-
-static int navi10_set_peak_performance_level(struct smu_context *smu)
-{
- struct amdgpu_device *adev = smu->adev;
- int ret = 0;
- uint32_t sclk_freq = 0, uclk_freq = 0;
-
- switch (adev->asic_type) {
- case CHIP_NAVI10:
- switch (adev->pdev->revision) {
- case 0xf0: /* XTX */
- case 0xc0:
- sclk_freq = NAVI10_PEAK_SCLK_XTX;
- break;
- case 0xf1: /* XT */
- case 0xc1:
- sclk_freq = NAVI10_PEAK_SCLK_XT;
- break;
- default: /* XL */
- sclk_freq = NAVI10_PEAK_SCLK_XL;
- break;
- }
- break;
- case CHIP_NAVI14:
- switch (adev->pdev->revision) {
- case 0xc7: /* XT */
- case 0xf4:
- sclk_freq = NAVI14_UMD_PSTATE_PEAK_XT_GFXCLK;
- break;
- case 0xc1: /* XTM */
- case 0xf2:
- sclk_freq = NAVI14_UMD_PSTATE_PEAK_XTM_GFXCLK;
- break;
- case 0xc3: /* XLM */
- case 0xf3:
- sclk_freq = NAVI14_UMD_PSTATE_PEAK_XLM_GFXCLK;
- break;
- case 0xc5: /* XTX */
- case 0xf6:
- sclk_freq = NAVI14_UMD_PSTATE_PEAK_XLM_GFXCLK;
- break;
- default: /* XL */
- sclk_freq = NAVI14_UMD_PSTATE_PEAK_XL_GFXCLK;
- break;
- }
- break;
- case CHIP_NAVI12:
- sclk_freq = NAVI12_UMD_PSTATE_PEAK_GFXCLK;
- break;
- default:
- ret = smu_v11_0_get_dpm_level_range(smu,
- SMU_SCLK,
- NULL,
- &sclk_freq);
- if (ret)
- return ret;
- }
-
- ret = smu_v11_0_get_dpm_level_range(smu,
- SMU_UCLK,
- NULL,
- &uclk_freq);
- if (ret)
- return ret;
-
- ret = smu_v11_0_set_soft_freq_limited_range(smu, SMU_SCLK, sclk_freq, sclk_freq);
- if (ret)
- return ret;
- ret = smu_v11_0_set_soft_freq_limited_range(smu, SMU_UCLK, uclk_freq, uclk_freq);
- if (ret)
- return ret;
-
- return ret;
-}
-
-static int navi10_set_performance_level(struct smu_context *smu,
- enum amd_dpm_forced_level level)
-{
- int ret = 0;
- uint32_t sclk_mask, mclk_mask, soc_mask;
-
- switch (level) {
- case AMD_DPM_FORCED_LEVEL_HIGH:
- ret = smu_force_dpm_limit_value(smu, true);
- break;
- case AMD_DPM_FORCED_LEVEL_LOW:
- ret = smu_force_dpm_limit_value(smu, false);
- break;
- case AMD_DPM_FORCED_LEVEL_AUTO:
- ret = smu_unforce_dpm_levels(smu);
- break;
- case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD:
- ret = navi10_set_standard_performance_level(smu);
- break;
- case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK:
- case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK:
- ret = smu_get_profiling_clk_mask(smu, level,
- &sclk_mask,
- &mclk_mask,
- &soc_mask);
- if (ret)
- return ret;
- smu_force_clk_levels(smu, SMU_SCLK, 1 << sclk_mask, false);
- smu_force_clk_levels(smu, SMU_MCLK, 1 << mclk_mask, false);
- smu_force_clk_levels(smu, SMU_SOCCLK, 1 << soc_mask, false);
- break;
- case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK:
- ret = navi10_set_peak_performance_level(smu);
- break;
- case AMD_DPM_FORCED_LEVEL_MANUAL:
- case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT:
- default:
- break;
- }
- return ret;
-}
-
static int navi10_get_thermal_temperature_range(struct smu_context *smu,
struct smu_temperature_range *range)
{
@@ -2622,18 +2379,15 @@ static const struct pptable_funcs navi10_ppt_funcs = {
.pre_display_config_changed = navi10_pre_display_config_changed,
.display_config_changed = navi10_display_config_changed,
.notify_smc_display_config = navi10_notify_smc_display_config,
- .force_dpm_limit_value = navi10_force_dpm_limit_value,
- .unforce_dpm_levels = navi10_unforce_dpm_levels,
.is_dpm_running = navi10_is_dpm_running,
.get_fan_speed_percent = navi10_get_fan_speed_percent,
.get_fan_speed_rpm = navi10_get_fan_speed_rpm,
.get_power_profile_mode = navi10_get_power_profile_mode,
.set_power_profile_mode = navi10_set_power_profile_mode,
- .get_profiling_clk_mask = navi10_get_profiling_clk_mask,
.set_watermarks_table = navi10_set_watermarks_table,
.read_sensor = navi10_read_sensor,
.get_uclk_dpm_states = navi10_get_uclk_dpm_states,
- .set_performance_level = navi10_set_performance_level,
+ .set_performance_level = smu_v11_0_set_performance_level,
.get_thermal_temperature_range = navi10_get_thermal_temperature_range,
.display_disable_memory_clock_switch = navi10_display_disable_memory_clock_switch,
.get_power_limit = navi10_get_power_limit,
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 11/16] drm/amd/powerplay: drop unnecessary Sienna Cichlid specific APIs
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (8 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 10/16] drm/amd/powerplay: drop unnecessary Navi1x " Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 12/16] drm/amd/powerplay: drop Sienna Cichlid specific set_soft_freq_limited_range Evan Quan
` (4 subsequent siblings)
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
As a common performance level setting API is used. Then these
ASIC specific APIs are not needed any more.
Change-Id: I04c810859794b07ce8905a8df797ed6b5ae116a8
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
.../drm/amd/powerplay/sienna_cichlid_ppt.c | 178 +-----------------
1 file changed, 1 insertion(+), 177 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
index 8fae7dd982c7..27f77bde184f 100644
--- a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
@@ -1181,59 +1181,6 @@ static int sienna_cichlid_display_config_changed(struct smu_context *smu)
return ret;
}
-static int sienna_cichlid_force_dpm_limit_value(struct smu_context *smu, bool highest)
-{
- int ret = 0, i = 0;
- uint32_t min_freq, max_freq, force_freq;
- enum smu_clk_type clk_type;
-
- enum smu_clk_type clks[] = {
- SMU_GFXCLK,
- SMU_MCLK,
- SMU_SOCCLK,
- };
-
- for (i = 0; i < ARRAY_SIZE(clks); i++) {
- clk_type = clks[i];
- ret = smu_v11_0_get_dpm_ultimate_freq(smu, clk_type, &min_freq, &max_freq);
- if (ret)
- return ret;
-
- force_freq = highest ? max_freq : min_freq;
- ret = sienna_cichlid_set_soft_freq_limited_range(smu, clk_type, force_freq, force_freq);
- if (ret)
- return ret;
- }
-
- return ret;
-}
-
-static int sienna_cichlid_unforce_dpm_levels(struct smu_context *smu)
-{
- int ret = 0, i = 0;
- uint32_t min_freq, max_freq;
- enum smu_clk_type clk_type;
-
- enum smu_clk_type clks[] = {
- SMU_GFXCLK,
- SMU_MCLK,
- SMU_SOCCLK,
- };
-
- for (i = 0; i < ARRAY_SIZE(clks); i++) {
- clk_type = clks[i];
- ret = smu_v11_0_get_dpm_ultimate_freq(smu, clk_type, &min_freq, &max_freq);
- if (ret)
- return ret;
-
- ret = sienna_cichlid_set_soft_freq_limited_range(smu, clk_type, min_freq, max_freq);
- if (ret)
- return ret;
- }
-
- return ret;
-}
-
static int sienna_cichlid_get_gpu_power(struct smu_context *smu, uint32_t *value)
{
if (!value)
@@ -1486,50 +1433,6 @@ static int sienna_cichlid_set_power_profile_mode(struct smu_context *smu, long *
return ret;
}
-static int sienna_cichlid_get_profiling_clk_mask(struct smu_context *smu,
- enum amd_dpm_forced_level level,
- uint32_t *sclk_mask,
- uint32_t *mclk_mask,
- uint32_t *soc_mask)
-{
- struct amdgpu_device *adev = smu->adev;
- int ret = 0;
- uint32_t level_count = 0;
-
- if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) {
- if (sclk_mask)
- *sclk_mask = 0;
- } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) {
- if (mclk_mask)
- *mclk_mask = 0;
- } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) {
- if(sclk_mask) {
- amdgpu_gfx_off_ctrl(adev, false);
- ret = smu_v11_0_get_dpm_level_count(smu, SMU_SCLK, &level_count);
- amdgpu_gfx_off_ctrl(adev, true);
- if (ret)
- return ret;
- *sclk_mask = level_count - 1;
- }
-
- if(mclk_mask) {
- ret = smu_v11_0_get_dpm_level_count(smu, SMU_MCLK, &level_count);
- if (ret)
- return ret;
- *mclk_mask = level_count - 1;
- }
-
- if(soc_mask) {
- ret = smu_v11_0_get_dpm_level_count(smu, SMU_SOCCLK, &level_count);
- if (ret)
- return ret;
- *soc_mask = level_count - 1;
- }
- }
-
- return ret;
-}
-
static int sienna_cichlid_notify_smc_display_config(struct smu_context *smu)
{
struct smu_clocks min_clocks = {0};
@@ -1761,82 +1664,6 @@ static int sienna_cichlid_get_uclk_dpm_states(struct smu_context *smu, uint32_t
return 0;
}
-static int sienna_cichlid_set_performance_level(struct smu_context *smu,
- enum amd_dpm_forced_level level);
-
-static int sienna_cichlid_set_standard_performance_level(struct smu_context *smu)
-{
- struct amdgpu_device *adev = smu->adev;
- int ret = 0;
- uint32_t sclk_freq = 0, uclk_freq = 0;
-
- switch (adev->asic_type) {
- /* TODO: need to set specify clk value by asic type, not support yet*/
- default:
- /* by default, this is same as auto performance level */
- return sienna_cichlid_set_performance_level(smu, AMD_DPM_FORCED_LEVEL_AUTO);
- }
-
- ret = sienna_cichlid_set_soft_freq_limited_range(smu, SMU_SCLK, sclk_freq, sclk_freq);
- if (ret)
- return ret;
- ret = sienna_cichlid_set_soft_freq_limited_range(smu, SMU_UCLK, uclk_freq, uclk_freq);
- if (ret)
- return ret;
-
- return ret;
-}
-
-static int sienna_cichlid_set_peak_performance_level(struct smu_context *smu)
-{
- int ret = 0;
-
- /* TODO: not support yet*/
- return ret;
-}
-
-static int sienna_cichlid_set_performance_level(struct smu_context *smu,
- enum amd_dpm_forced_level level)
-{
- int ret = 0;
- uint32_t sclk_mask, mclk_mask, soc_mask;
-
- switch (level) {
- case AMD_DPM_FORCED_LEVEL_HIGH:
- ret = smu_force_dpm_limit_value(smu, true);
- break;
- case AMD_DPM_FORCED_LEVEL_LOW:
- ret = smu_force_dpm_limit_value(smu, false);
- break;
- case AMD_DPM_FORCED_LEVEL_AUTO:
- ret = smu_unforce_dpm_levels(smu);
- break;
- case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD:
- ret = sienna_cichlid_set_standard_performance_level(smu);
- break;
- case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK:
- case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK:
- ret = smu_get_profiling_clk_mask(smu, level,
- &sclk_mask,
- &mclk_mask,
- &soc_mask);
- if (ret)
- return ret;
- smu_force_clk_levels(smu, SMU_SCLK, 1 << sclk_mask, false);
- smu_force_clk_levels(smu, SMU_MCLK, 1 << mclk_mask, false);
- smu_force_clk_levels(smu, SMU_SOCCLK, 1 << soc_mask, false);
- break;
- case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK:
- ret = sienna_cichlid_set_peak_performance_level(smu);
- break;
- case AMD_DPM_FORCED_LEVEL_MANUAL:
- case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT:
- default:
- break;
- }
- return ret;
-}
-
static int sienna_cichlid_get_thermal_temperature_range(struct smu_context *smu,
struct smu_temperature_range *range)
{
@@ -2684,18 +2511,15 @@ static const struct pptable_funcs sienna_cichlid_ppt_funcs = {
.pre_display_config_changed = sienna_cichlid_pre_display_config_changed,
.display_config_changed = sienna_cichlid_display_config_changed,
.notify_smc_display_config = sienna_cichlid_notify_smc_display_config,
- .force_dpm_limit_value = sienna_cichlid_force_dpm_limit_value,
- .unforce_dpm_levels = sienna_cichlid_unforce_dpm_levels,
.is_dpm_running = sienna_cichlid_is_dpm_running,
.get_fan_speed_percent = sienna_cichlid_get_fan_speed_percent,
.get_fan_speed_rpm = sienna_cichlid_get_fan_speed_rpm,
.get_power_profile_mode = sienna_cichlid_get_power_profile_mode,
.set_power_profile_mode = sienna_cichlid_set_power_profile_mode,
- .get_profiling_clk_mask = sienna_cichlid_get_profiling_clk_mask,
.set_watermarks_table = sienna_cichlid_set_watermarks_table,
.read_sensor = sienna_cichlid_read_sensor,
.get_uclk_dpm_states = sienna_cichlid_get_uclk_dpm_states,
- .set_performance_level = sienna_cichlid_set_performance_level,
+ .set_performance_level = smu_v11_0_set_performance_level,
.get_thermal_temperature_range = sienna_cichlid_get_thermal_temperature_range,
.display_disable_memory_clock_switch = sienna_cichlid_display_disable_memory_clock_switch,
.get_power_limit = sienna_cichlid_get_power_limit,
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 12/16] drm/amd/powerplay: drop Sienna Cichlid specific set_soft_freq_limited_range
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (9 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 11/16] drm/amd/powerplay: drop unnecessary Sienna Cichlid " Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-13 14:53 ` Alex Deucher
2020-07-10 4:47 ` [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs Evan Quan
` (3 subsequent siblings)
14 siblings, 1 reply; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
Use the common smu_v11_0_set_soft_freq_limited_range.
Change-Id: I9f8772880b324ce9e741291751bb1b8ff4c36ea3
Signed-off-by: Evan Quan <evan.quan@amd.com>
---
.../drm/amd/powerplay/sienna_cichlid_ppt.c | 20 ++-----------------
drivers/gpu/drm/amd/powerplay/smu_internal.h | 1 -
drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 20 +++++++++++++++----
3 files changed, 18 insertions(+), 23 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
index 27f77bde184f..141944df97b0 100644
--- a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
@@ -1046,22 +1046,6 @@ static int sienna_cichlid_print_clk_levels(struct smu_context *smu,
return size;
}
-int sienna_cichlid_set_soft_freq_limited_range(struct smu_context *smu,
- enum smu_clk_type clk_type,
- uint32_t min, uint32_t max)
-{
- struct amdgpu_device *adev = smu->adev;
- int ret;
-
- if (clk_type == SMU_GFXCLK)
- amdgpu_gfx_off_ctrl(adev, false);
- ret = smu_v11_0_set_soft_freq_limited_range(smu, clk_type, min, max);
- if (clk_type == SMU_GFXCLK)
- amdgpu_gfx_off_ctrl(adev, true);
-
- return ret;
-}
-
static int sienna_cichlid_force_clk_levels(struct smu_context *smu,
enum smu_clk_type clk_type, uint32_t mask)
{
@@ -1097,7 +1081,7 @@ static int sienna_cichlid_force_clk_levels(struct smu_context *smu,
if (ret)
goto forec_level_out;
- ret = sienna_cichlid_set_soft_freq_limited_range(smu, clk_type, min_freq, max_freq);
+ ret = smu_v11_0_set_soft_freq_limited_range(smu, clk_type, min_freq, max_freq);
if (ret)
goto forec_level_out;
break;
@@ -2566,7 +2550,7 @@ static const struct pptable_funcs sienna_cichlid_ppt_funcs = {
.baco_enter = smu_v11_0_baco_enter,
.baco_exit = smu_v11_0_baco_exit,
.get_dpm_ultimate_freq = sienna_cichlid_get_dpm_ultimate_freq,
- .set_soft_freq_limited_range = sienna_cichlid_set_soft_freq_limited_range,
+ .set_soft_freq_limited_range = smu_v11_0_set_soft_freq_limited_range,
.override_pcie_parameters = smu_v11_0_override_pcie_parameters,
.set_thermal_range = sienna_cichlid_set_thermal_range,
};
diff --git a/drivers/gpu/drm/amd/powerplay/smu_internal.h b/drivers/gpu/drm/amd/powerplay/smu_internal.h
index 1c808ffe3ab1..91d3965bbe80 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_internal.h
+++ b/drivers/gpu/drm/amd/powerplay/smu_internal.h
@@ -93,7 +93,6 @@
#define smu_asic_set_performance_level(smu, level) smu_ppt_funcs(set_performance_level, -EINVAL, smu, level)
#define smu_dump_pptable(smu) smu_ppt_funcs(dump_pptable, 0, smu)
#define smu_get_dpm_clk_limited(smu, clk_type, dpm_level, freq) smu_ppt_funcs(get_dpm_clk_limited, -EINVAL, smu, clk_type, dpm_level, freq)
-#define smu_set_soft_freq_limited_range(smu, clk_type, min, max) smu_ppt_funcs(set_soft_freq_limited_range, -EINVAL, smu, clk_type, min, max)
#define smu_override_pcie_parameters(smu) smu_ppt_funcs(override_pcie_parameters, 0, smu)
#define smu_update_pcie_parameters(smu, pcie_gen_cap, pcie_width_cap) smu_ppt_funcs(update_pcie_parameters, 0, smu, pcie_gen_cap, pcie_width_cap)
#define smu_set_thermal_range(smu, range) smu_ppt_funcs(set_thermal_range, 0, smu, range)
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index f96ff062eb64..c2779d0b51f6 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
+++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
@@ -1745,9 +1745,12 @@ int smu_v11_0_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type c
return ret;
}
-int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_type clk_type,
- uint32_t min, uint32_t max)
+int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
+ enum smu_clk_type clk_type,
+ uint32_t min,
+ uint32_t max)
{
+ struct amdgpu_device *adev = smu->adev;
int ret = 0, clk_id = 0;
uint32_t param;
@@ -1755,12 +1758,16 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_
if (clk_id < 0)
return clk_id;
+ if (clk_type == SMU_GFXCLK &&
+ adev->asic_type == CHIP_SIENNA_CICHLID)
+ amdgpu_gfx_off_ctrl(adev, false);
+
if (max > 0) {
param = (uint32_t)((clk_id << 16) | (max & 0xffff));
ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxByFreq,
param, NULL);
if (ret)
- return ret;
+ goto out;
}
if (min > 0) {
@@ -1768,9 +1775,14 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_
ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMinByFreq,
param, NULL);
if (ret)
- return ret;
+ goto out;
}
+out:
+ if (clk_type == SMU_GFXCLK &&
+ adev->asic_type == CHIP_SIENNA_CICHLID)
+ amdgpu_gfx_off_ctrl(adev, true);
+
return ret;
}
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (10 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 12/16] drm/amd/powerplay: drop Sienna Cichlid specific set_soft_freq_limited_range Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-13 3:44 ` Quan, Evan
2020-07-10 4:47 ` [PATCH 14/16] drm/amd/powerplay: drop unnecessary wrappers Evan Quan
` (2 subsequent siblings)
14 siblings, 1 reply; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
Before and after setting gfx clock soft max/min frequency.
Change-Id: I6f828da8de096ebc0ae27eaa89f988def2d547ec
Signed-off-by: Evan Quan <evan.quan@amd.com>
---
drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index c2779d0b51f6..33e0718f2635 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
+++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
@@ -1758,8 +1758,7 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
if (clk_id < 0)
return clk_id;
- if (clk_type == SMU_GFXCLK &&
- adev->asic_type == CHIP_SIENNA_CICHLID)
+ if (clk_type == SMU_GFXCLK)
amdgpu_gfx_off_ctrl(adev, false);
if (max > 0) {
@@ -1779,8 +1778,7 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
}
out:
- if (clk_type == SMU_GFXCLK &&
- adev->asic_type == CHIP_SIENNA_CICHLID)
+ if (clk_type == SMU_GFXCLK)
amdgpu_gfx_off_ctrl(adev, true);
return ret;
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 14/16] drm/amd/powerplay: drop unnecessary wrappers
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (11 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 15/16] drm/amd/powerplay: drop smu_v12_0.c unnecessary wrapper Evan Quan
2020-07-10 4:47 ` [PATCH 16/16] drm/amd/powerplay: drop unused APIs and parameters Evan Quan
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
By calling the target APIs directly.
Change-Id: I0f24f603d2fcb94d2078a35c405a1406093ba5e3
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/powerplay/renoir_ppt.c | 22 ++++++++++------------
1 file changed, 10 insertions(+), 12 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/renoir_ppt.c b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c
index d4aa01a05c54..49a8d636ef4d 100644
--- a/drivers/gpu/drm/amd/powerplay/renoir_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c
@@ -743,26 +743,26 @@ static int renoir_set_performance_level(struct smu_context *smu,
switch (level) {
case AMD_DPM_FORCED_LEVEL_HIGH:
- ret = smu_force_dpm_limit_value(smu, true);
+ ret = renoir_force_dpm_limit_value(smu, true);
break;
case AMD_DPM_FORCED_LEVEL_LOW:
- ret = smu_force_dpm_limit_value(smu, false);
+ ret = renoir_force_dpm_limit_value(smu, false);
break;
case AMD_DPM_FORCED_LEVEL_AUTO:
case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD:
- ret = smu_unforce_dpm_levels(smu);
+ ret = renoir_unforce_dpm_levels(smu);
break;
case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK:
case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK:
- ret = smu_get_profiling_clk_mask(smu, level,
- &sclk_mask,
- &mclk_mask,
- &soc_mask);
+ ret = renoir_get_profiling_clk_mask(smu, level,
+ &sclk_mask,
+ &mclk_mask,
+ &soc_mask);
if (ret)
return ret;
- smu_force_clk_levels(smu, SMU_SCLK, 1 << sclk_mask, false);
- smu_force_clk_levels(smu, SMU_MCLK, 1 << mclk_mask, false);
- smu_force_clk_levels(smu, SMU_SOCCLK, 1 << soc_mask, false);
+ renoir_force_clk_levels(smu, SMU_SCLK, 1 << sclk_mask);
+ renoir_force_clk_levels(smu, SMU_MCLK, 1 << mclk_mask);
+ renoir_force_clk_levels(smu, SMU_SOCCLK, 1 << soc_mask);
break;
case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK:
ret = renoir_set_peak_clock_by_device(smu);
@@ -942,8 +942,6 @@ static const struct pptable_funcs renoir_ppt_funcs = {
.get_current_power_state = renoir_get_current_power_state,
.dpm_set_vcn_enable = renoir_dpm_set_vcn_enable,
.dpm_set_jpeg_enable = renoir_dpm_set_jpeg_enable,
- .force_dpm_limit_value = renoir_force_dpm_limit_value,
- .unforce_dpm_levels = renoir_unforce_dpm_levels,
.get_workload_type = renoir_get_workload_type,
.get_profiling_clk_mask = renoir_get_profiling_clk_mask,
.force_clk_levels = renoir_force_clk_levels,
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 15/16] drm/amd/powerplay: drop smu_v12_0.c unnecessary wrapper
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (12 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 14/16] drm/amd/powerplay: drop unnecessary wrappers Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
2020-07-10 4:47 ` [PATCH 16/16] drm/amd/powerplay: drop unused APIs and parameters Evan Quan
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
By moving the implemention to renoir_ppt.c considering
it's really ASIC specific.
Change-Id: I6f7a594235dffdf75b56d1de5b9dc6d49833d7e8
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h | 3 -
drivers/gpu/drm/amd/powerplay/renoir_ppt.c | 172 ++++++++++++++----
drivers/gpu/drm/amd/powerplay/smu_v12_0.c | 100 ----------
3 files changed, 138 insertions(+), 137 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h
index 0c1e1455c68f..fd83a723f32c 100644
--- a/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h
+++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v12_0.h
@@ -73,9 +73,6 @@ int smu_v12_0_set_default_dpm_tables(struct smu_context *smu);
int smu_v12_0_get_enabled_mask(struct smu_context *smu,
uint32_t *feature_mask, uint32_t num);
-int smu_v12_0_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type clk_type,
- uint32_t *min, uint32_t *max);
-
int smu_v12_0_mode2_reset(struct smu_context *smu);
int smu_v12_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_type clk_type,
diff --git a/drivers/gpu/drm/amd/powerplay/renoir_ppt.c b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c
index 49a8d636ef4d..5b76d67d03d7 100644
--- a/drivers/gpu/drm/amd/powerplay/renoir_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c
@@ -241,6 +241,137 @@ static int renoir_get_dpm_clk_limited(struct smu_context *smu, enum smu_clk_type
return 0;
}
+static int renoir_get_profiling_clk_mask(struct smu_context *smu,
+ enum amd_dpm_forced_level level,
+ uint32_t *sclk_mask,
+ uint32_t *mclk_mask,
+ uint32_t *soc_mask)
+{
+
+ if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) {
+ if (sclk_mask)
+ *sclk_mask = 0;
+ } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) {
+ if (mclk_mask)
+ *mclk_mask = 0;
+ } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) {
+ if(sclk_mask)
+ /* The sclk as gfxclk and has three level about max/min/current */
+ *sclk_mask = 3 - 1;
+
+ if(mclk_mask)
+ *mclk_mask = NUM_MEMCLK_DPM_LEVELS - 1;
+
+ if(soc_mask)
+ *soc_mask = NUM_SOCCLK_DPM_LEVELS - 1;
+ }
+
+ return 0;
+}
+
+static int renoir_get_dpm_ultimate_freq(struct smu_context *smu,
+ enum smu_clk_type clk_type,
+ uint32_t *min,
+ uint32_t *max)
+{
+ int ret = 0;
+ uint32_t mclk_mask, soc_mask;
+ uint32_t clock_limit;
+
+ if (!smu_clk_dpm_is_enabled(smu, clk_type)) {
+ switch (clk_type) {
+ case SMU_MCLK:
+ case SMU_UCLK:
+ clock_limit = smu->smu_table.boot_values.uclk;
+ break;
+ case SMU_GFXCLK:
+ case SMU_SCLK:
+ clock_limit = smu->smu_table.boot_values.gfxclk;
+ break;
+ case SMU_SOCCLK:
+ clock_limit = smu->smu_table.boot_values.socclk;
+ break;
+ default:
+ clock_limit = 0;
+ break;
+ }
+
+ /* clock in Mhz unit */
+ if (min)
+ *min = clock_limit / 100;
+ if (max)
+ *max = clock_limit / 100;
+
+ return 0;
+ }
+
+ if (max) {
+ ret = renoir_get_profiling_clk_mask(smu,
+ AMD_DPM_FORCED_LEVEL_PROFILE_PEAK,
+ NULL,
+ &mclk_mask,
+ &soc_mask);
+ if (ret)
+ goto failed;
+
+ switch (clk_type) {
+ case SMU_GFXCLK:
+ case SMU_SCLK:
+ ret = smu_send_smc_msg(smu, SMU_MSG_GetMaxGfxclkFrequency, max);
+ if (ret) {
+ dev_err(smu->adev->dev, "Attempt to get max GX frequency from SMC Failed !\n");
+ goto failed;
+ }
+ break;
+ case SMU_UCLK:
+ case SMU_FCLK:
+ case SMU_MCLK:
+ ret = renoir_get_dpm_clk_limited(smu, clk_type, mclk_mask, max);
+ if (ret)
+ goto failed;
+ break;
+ case SMU_SOCCLK:
+ ret = renoir_get_dpm_clk_limited(smu, clk_type, soc_mask, max);
+ if (ret)
+ goto failed;
+ break;
+ default:
+ ret = -EINVAL;
+ goto failed;
+ }
+ }
+
+ if (min) {
+ switch (clk_type) {
+ case SMU_GFXCLK:
+ case SMU_SCLK:
+ ret = smu_send_smc_msg(smu, SMU_MSG_GetMinGfxclkFrequency, min);
+ if (ret) {
+ dev_err(smu->adev->dev, "Attempt to get min GX frequency from SMC Failed !\n");
+ goto failed;
+ }
+ break;
+ case SMU_UCLK:
+ case SMU_FCLK:
+ case SMU_MCLK:
+ ret = renoir_get_dpm_clk_limited(smu, clk_type, 0, min);
+ if (ret)
+ goto failed;
+ break;
+ case SMU_SOCCLK:
+ ret = renoir_get_dpm_clk_limited(smu, clk_type, 0, min);
+ if (ret)
+ goto failed;
+ break;
+ default:
+ ret = -EINVAL;
+ goto failed;
+ }
+ }
+failed:
+ return ret;
+}
+
static int renoir_print_clk_levels(struct smu_context *smu,
enum smu_clk_type clk_type, char *buf)
{
@@ -264,7 +395,7 @@ static int renoir_print_clk_levels(struct smu_context *smu,
case SMU_SCLK:
/* retirve table returned paramters unit is MHz */
cur_value = metrics.ClockFrequency[CLOCK_GFXCLK];
- ret = smu_v12_0_get_dpm_ultimate_freq(smu, SMU_GFXCLK, &min, &max);
+ ret = renoir_get_dpm_ultimate_freq(smu, SMU_GFXCLK, &min, &max);
if (!ret) {
/* driver only know min/max gfx_clk, Add level 1 for all other gfx clks */
if (cur_value == max)
@@ -434,7 +565,7 @@ static int renoir_force_dpm_limit_value(struct smu_context *smu, bool highest)
for (i = 0; i < ARRAY_SIZE(clks); i++) {
clk_type = clks[i];
- ret = smu_v12_0_get_dpm_ultimate_freq(smu, clk_type, &min_freq, &max_freq);
+ ret = renoir_get_dpm_ultimate_freq(smu, clk_type, &min_freq, &max_freq);
if (ret)
return ret;
@@ -468,7 +599,7 @@ static int renoir_unforce_dpm_levels(struct smu_context *smu) {
clk_type = clk_feature_map[i].clk_type;
- ret = smu_v12_0_get_dpm_ultimate_freq(smu, clk_type, &min_freq, &max_freq);
+ ret = renoir_get_dpm_ultimate_freq(smu, clk_type, &min_freq, &max_freq);
if (ret)
return ret;
@@ -552,33 +683,6 @@ static int renoir_get_workload_type(struct smu_context *smu, uint32_t profile)
return pplib_workload;
}
-static int renoir_get_profiling_clk_mask(struct smu_context *smu,
- enum amd_dpm_forced_level level,
- uint32_t *sclk_mask,
- uint32_t *mclk_mask,
- uint32_t *soc_mask)
-{
-
- if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK) {
- if (sclk_mask)
- *sclk_mask = 0;
- } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK) {
- if (mclk_mask)
- *mclk_mask = 0;
- } else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK) {
- if(sclk_mask)
- /* The sclk as gfxclk and has three level about max/min/current */
- *sclk_mask = 3 - 1;
-
- if(mclk_mask)
- *mclk_mask = NUM_MEMCLK_DPM_LEVELS - 1;
-
- if(soc_mask)
- *soc_mask = NUM_SOCCLK_DPM_LEVELS - 1;
- }
-
- return 0;
-}
/**
* This interface get dpm clock table for dc
@@ -633,7 +737,7 @@ static int renoir_force_clk_levels(struct smu_context *smu,
return -EINVAL;
}
- ret = smu_v12_0_get_dpm_ultimate_freq(smu, SMU_GFXCLK, &min_freq, &max_freq);
+ ret = renoir_get_dpm_ultimate_freq(smu, SMU_GFXCLK, &min_freq, &max_freq);
if (ret)
return ret;
ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxGfxClk,
@@ -716,7 +820,7 @@ static int renoir_set_peak_clock_by_device(struct smu_context *smu)
int ret = 0;
uint32_t sclk_freq = 0, uclk_freq = 0;
- ret = smu_v12_0_get_dpm_ultimate_freq(smu, SMU_SCLK, NULL, &sclk_freq);
+ ret = renoir_get_dpm_ultimate_freq(smu, SMU_SCLK, NULL, &sclk_freq);
if (ret)
return ret;
@@ -724,7 +828,7 @@ static int renoir_set_peak_clock_by_device(struct smu_context *smu)
if (ret)
return ret;
- ret = smu_v12_0_get_dpm_ultimate_freq(smu, SMU_UCLK, NULL, &uclk_freq);
+ ret = renoir_get_dpm_ultimate_freq(smu, SMU_UCLK, NULL, &uclk_freq);
if (ret)
return ret;
@@ -961,7 +1065,7 @@ static const struct pptable_funcs renoir_ppt_funcs = {
.fini_smc_tables = smu_v12_0_fini_smc_tables,
.set_default_dpm_table = smu_v12_0_set_default_dpm_tables,
.get_enabled_mask = smu_v12_0_get_enabled_mask,
- .get_dpm_ultimate_freq = smu_v12_0_get_dpm_ultimate_freq,
+ .get_dpm_ultimate_freq = renoir_get_dpm_ultimate_freq,
.mode2_reset = smu_v12_0_mode2_reset,
.set_soft_freq_limited_range = smu_v12_0_set_soft_freq_limited_range,
.set_driver_table_location = smu_v12_0_set_driver_table_location,
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v12_0.c b/drivers/gpu/drm/amd/powerplay/smu_v12_0.c
index 6400a0acad63..4e1b11d07438 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_v12_0.c
+++ b/drivers/gpu/drm/amd/powerplay/smu_v12_0.c
@@ -316,106 +316,6 @@ int smu_v12_0_get_enabled_mask(struct smu_context *smu,
return ret;
}
-int smu_v12_0_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type clk_type,
- uint32_t *min, uint32_t *max)
-{
- int ret = 0;
- uint32_t mclk_mask, soc_mask;
- uint32_t clock_limit;
-
- if (!smu_clk_dpm_is_enabled(smu, clk_type)) {
- switch (clk_type) {
- case SMU_MCLK:
- case SMU_UCLK:
- clock_limit = smu->smu_table.boot_values.uclk;
- break;
- case SMU_GFXCLK:
- case SMU_SCLK:
- clock_limit = smu->smu_table.boot_values.gfxclk;
- break;
- case SMU_SOCCLK:
- clock_limit = smu->smu_table.boot_values.socclk;
- break;
- default:
- clock_limit = 0;
- break;
- }
-
- /* clock in Mhz unit */
- if (min)
- *min = clock_limit / 100;
- if (max)
- *max = clock_limit / 100;
-
- return 0;
- }
-
- if (max) {
- ret = smu_get_profiling_clk_mask(smu, AMD_DPM_FORCED_LEVEL_PROFILE_PEAK,
- NULL,
- &mclk_mask,
- &soc_mask);
- if (ret)
- goto failed;
-
- switch (clk_type) {
- case SMU_GFXCLK:
- case SMU_SCLK:
- ret = smu_send_smc_msg(smu, SMU_MSG_GetMaxGfxclkFrequency, max);
- if (ret) {
- dev_err(smu->adev->dev, "Attempt to get max GX frequency from SMC Failed !\n");
- goto failed;
- }
- break;
- case SMU_UCLK:
- case SMU_FCLK:
- case SMU_MCLK:
- ret = smu_get_dpm_clk_limited(smu, clk_type, mclk_mask, max);
- if (ret)
- goto failed;
- break;
- case SMU_SOCCLK:
- ret = smu_get_dpm_clk_limited(smu, clk_type, soc_mask, max);
- if (ret)
- goto failed;
- break;
- default:
- ret = -EINVAL;
- goto failed;
- }
- }
-
- if (min) {
- switch (clk_type) {
- case SMU_GFXCLK:
- case SMU_SCLK:
- ret = smu_send_smc_msg(smu, SMU_MSG_GetMinGfxclkFrequency, min);
- if (ret) {
- dev_err(smu->adev->dev, "Attempt to get min GX frequency from SMC Failed !\n");
- goto failed;
- }
- break;
- case SMU_UCLK:
- case SMU_FCLK:
- case SMU_MCLK:
- ret = smu_get_dpm_clk_limited(smu, clk_type, 0, min);
- if (ret)
- goto failed;
- break;
- case SMU_SOCCLK:
- ret = smu_get_dpm_clk_limited(smu, clk_type, 0, min);
- if (ret)
- goto failed;
- break;
- default:
- ret = -EINVAL;
- goto failed;
- }
- }
-failed:
- return ret;
-}
-
int smu_v12_0_mode2_reset(struct smu_context *smu){
return smu_v12_0_send_msg_with_param(smu, SMU_MSG_GfxDeviceDriverReset, SMU_RESET_MODE_2, NULL);
}
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 16/16] drm/amd/powerplay: drop unused APIs and parameters
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
` (13 preceding siblings ...)
2020-07-10 4:47 ` [PATCH 15/16] drm/amd/powerplay: drop smu_v12_0.c unnecessary wrapper Evan Quan
@ 2020-07-10 4:47 ` Evan Quan
14 siblings, 0 replies; 22+ messages in thread
From: Evan Quan @ 2020-07-10 4:47 UTC (permalink / raw)
To: amd-gfx; +Cc: alexander.deucher, Evan Quan
Leftover of previous performance level setting cleanups.
Change-Id: Idddc4adce365b34eacbc13f75cc0629859c6d412
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 12 ++++++------
drivers/gpu/drm/amd/powerplay/amdgpu_smu.c | 9 +++------
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h | 12 +-----------
drivers/gpu/drm/amd/powerplay/renoir_ppt.c | 2 --
drivers/gpu/drm/amd/powerplay/smu_internal.h | 5 -----
5 files changed, 10 insertions(+), 30 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
index d2401379bd33..20f39aa04fb9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
@@ -1105,7 +1105,7 @@ static ssize_t amdgpu_set_pp_dpm_sclk(struct device *dev,
}
if (is_support_sw_smu(adev))
- ret = smu_force_clk_levels(&adev->smu, SMU_SCLK, mask, true);
+ ret = smu_force_clk_levels(&adev->smu, SMU_SCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_SCLK, mask);
@@ -1173,7 +1173,7 @@ static ssize_t amdgpu_set_pp_dpm_mclk(struct device *dev,
}
if (is_support_sw_smu(adev))
- ret = smu_force_clk_levels(&adev->smu, SMU_MCLK, mask, true);
+ ret = smu_force_clk_levels(&adev->smu, SMU_MCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_MCLK, mask);
@@ -1241,7 +1241,7 @@ static ssize_t amdgpu_set_pp_dpm_socclk(struct device *dev,
}
if (is_support_sw_smu(adev))
- ret = smu_force_clk_levels(&adev->smu, SMU_SOCCLK, mask, true);
+ ret = smu_force_clk_levels(&adev->smu, SMU_SOCCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_SOCCLK, mask);
else
@@ -1311,7 +1311,7 @@ static ssize_t amdgpu_set_pp_dpm_fclk(struct device *dev,
}
if (is_support_sw_smu(adev))
- ret = smu_force_clk_levels(&adev->smu, SMU_FCLK, mask, true);
+ ret = smu_force_clk_levels(&adev->smu, SMU_FCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_FCLK, mask);
else
@@ -1381,7 +1381,7 @@ static ssize_t amdgpu_set_pp_dpm_dcefclk(struct device *dev,
}
if (is_support_sw_smu(adev))
- ret = smu_force_clk_levels(&adev->smu, SMU_DCEFCLK, mask, true);
+ ret = smu_force_clk_levels(&adev->smu, SMU_DCEFCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_DCEFCLK, mask);
else
@@ -1451,7 +1451,7 @@ static ssize_t amdgpu_set_pp_dpm_pcie(struct device *dev,
}
if (is_support_sw_smu(adev))
- ret = smu_force_clk_levels(&adev->smu, SMU_PCIE, mask, true);
+ ret = smu_force_clk_levels(&adev->smu, SMU_PCIE, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_PCIE, mask);
else
diff --git a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
index 4080b3c792ac..38b3b47d12b7 100644
--- a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
+++ b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
@@ -1764,8 +1764,7 @@ int smu_set_display_count(struct smu_context *smu, uint32_t count)
int smu_force_clk_levels(struct smu_context *smu,
enum smu_clk_type clk_type,
- uint32_t mask,
- bool lock_needed)
+ uint32_t mask)
{
struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm);
int ret = 0;
@@ -1778,14 +1777,12 @@ int smu_force_clk_levels(struct smu_context *smu,
return -EINVAL;
}
- if (lock_needed)
- mutex_lock(&smu->mutex);
+ mutex_lock(&smu->mutex);
if (smu->ppt_funcs && smu->ppt_funcs->force_clk_levels)
ret = smu->ppt_funcs->force_clk_levels(smu, clk_type, mask);
- if (lock_needed)
- mutex_unlock(&smu->mutex);
+ mutex_unlock(&smu->mutex);
return ret;
}
diff --git a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h
index 91c8b69da026..470b0377a860 100644
--- a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h
+++ b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h
@@ -476,13 +476,6 @@ struct pptable_funcs {
int (*display_config_changed)(struct smu_context *smu);
int (*apply_clocks_adjust_rules)(struct smu_context *smu);
int (*notify_smc_display_config)(struct smu_context *smu);
- int (*force_dpm_limit_value)(struct smu_context *smu, bool highest);
- int (*unforce_dpm_levels)(struct smu_context *smu);
- int (*get_profiling_clk_mask)(struct smu_context *smu,
- enum amd_dpm_forced_level level,
- uint32_t *sclk_mask,
- uint32_t *mclk_mask,
- uint32_t *soc_mask);
int (*set_cpu_power_state)(struct smu_context *smu);
bool (*is_dpm_running)(struct smu_context *smu);
int (*tables_init)(struct smu_context *smu, struct smu_table *tables);
@@ -498,8 +491,6 @@ struct pptable_funcs {
int (*display_disable_memory_clock_switch)(struct smu_context *smu, bool disable_memory_clock_switch);
void (*dump_pptable)(struct smu_context *smu);
int (*get_power_limit)(struct smu_context *smu);
- int (*get_dpm_clk_limited)(struct smu_context *smu, enum smu_clk_type clk_type,
- uint32_t dpm_level, uint32_t *freq);
int (*set_df_cstate)(struct smu_context *smu, enum pp_df_cstate state);
int (*allow_xgmi_power_down)(struct smu_context *smu, bool en);
int (*update_pcie_parameters)(struct smu_context *smu, uint32_t pcie_gen_cap, uint32_t pcie_width_cap);
@@ -745,8 +736,7 @@ size_t smu_sys_get_pp_feature_mask(struct smu_context *smu, char *buf);
int smu_sys_set_pp_feature_mask(struct smu_context *smu, uint64_t new_mask);
int smu_force_clk_levels(struct smu_context *smu,
enum smu_clk_type clk_type,
- uint32_t mask,
- bool lock_needed);
+ uint32_t mask);
int smu_set_mp1_state(struct smu_context *smu,
enum pp_mp1_state mp1_state);
int smu_set_df_cstate(struct smu_context *smu,
diff --git a/drivers/gpu/drm/amd/powerplay/renoir_ppt.c b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c
index 5b76d67d03d7..866ae014eda2 100644
--- a/drivers/gpu/drm/amd/powerplay/renoir_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/renoir_ppt.c
@@ -1041,13 +1041,11 @@ static const struct pptable_funcs renoir_ppt_funcs = {
.get_smu_table_index = renoir_get_smu_table_index,
.tables_init = renoir_tables_init,
.set_power_state = NULL,
- .get_dpm_clk_limited = renoir_get_dpm_clk_limited,
.print_clk_levels = renoir_print_clk_levels,
.get_current_power_state = renoir_get_current_power_state,
.dpm_set_vcn_enable = renoir_dpm_set_vcn_enable,
.dpm_set_jpeg_enable = renoir_dpm_set_jpeg_enable,
.get_workload_type = renoir_get_workload_type,
- .get_profiling_clk_mask = renoir_get_profiling_clk_mask,
.force_clk_levels = renoir_force_clk_levels,
.set_power_profile_mode = renoir_set_power_profile_mode,
.set_performance_level = renoir_set_performance_level,
diff --git a/drivers/gpu/drm/amd/powerplay/smu_internal.h b/drivers/gpu/drm/amd/powerplay/smu_internal.h
index 91d3965bbe80..727498a6e76d 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_internal.h
+++ b/drivers/gpu/drm/amd/powerplay/smu_internal.h
@@ -68,8 +68,6 @@
#define smu_display_config_changed(smu) smu_ppt_funcs(display_config_changed, 0 , smu)
#define smu_apply_clocks_adjust_rules(smu) smu_ppt_funcs(apply_clocks_adjust_rules, 0, smu)
#define smu_notify_smc_display_config(smu) smu_ppt_funcs(notify_smc_display_config, 0, smu)
-#define smu_force_dpm_limit_value(smu, highest) smu_ppt_funcs(force_dpm_limit_value, 0, smu, highest)
-#define smu_unforce_dpm_levels(smu) smu_ppt_funcs(unforce_dpm_levels, 0, smu)
#define smu_set_cpu_power_state(smu) smu_ppt_funcs(set_cpu_power_state, 0, smu)
#define smu_msg_get_index(smu, msg) smu_ppt_funcs(get_smu_msg_index, -EINVAL, smu, msg)
#define smu_clk_get_index(smu, clk) smu_ppt_funcs(get_smu_clk_index, -EINVAL, smu, clk)
@@ -92,7 +90,6 @@
#define smu_get_dpm_ultimate_freq(smu, param, min, max) smu_ppt_funcs(get_dpm_ultimate_freq, 0, smu, param, min, max)
#define smu_asic_set_performance_level(smu, level) smu_ppt_funcs(set_performance_level, -EINVAL, smu, level)
#define smu_dump_pptable(smu) smu_ppt_funcs(dump_pptable, 0, smu)
-#define smu_get_dpm_clk_limited(smu, clk_type, dpm_level, freq) smu_ppt_funcs(get_dpm_clk_limited, -EINVAL, smu, clk_type, dpm_level, freq)
#define smu_override_pcie_parameters(smu) smu_ppt_funcs(override_pcie_parameters, 0, smu)
#define smu_update_pcie_parameters(smu, pcie_gen_cap, pcie_width_cap) smu_ppt_funcs(update_pcie_parameters, 0, smu, pcie_gen_cap, pcie_width_cap)
#define smu_set_thermal_range(smu, range) smu_ppt_funcs(set_thermal_range, 0, smu, range)
@@ -103,7 +100,5 @@
#define smu_get_unique_id(smu) smu_ppt_funcs(get_unique_id, 0, smu)
#define smu_log_thermal_throttling(smu) smu_ppt_funcs(log_thermal_throttling_event, 0, smu)
#define smu_get_asic_power_limits(smu) smu_ppt_funcs(get_power_limit, 0, smu)
-#define smu_get_profiling_clk_mask(smu, level, sclk_mask, mclk_mask, soc_mask) \
- smu_ppt_funcs(get_profiling_clk_mask, 0, smu, level, sclk_mask, mclk_mask, soc_mask)
#endif
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* RE: [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
2020-07-10 4:47 ` [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs Evan Quan
@ 2020-07-13 3:44 ` Quan, Evan
2020-07-14 6:55 ` Quan, Evan
0 siblings, 1 reply; 22+ messages in thread
From: Quan, Evan @ 2020-07-13 3:44 UTC (permalink / raw)
To: amd-gfx; +Cc: Deucher, Alexander
[AMD Official Use Only - Internal Distribution Only]
Ping for this one and patch12, patch2 and patch3
BR
Evan
-----Original Message-----
From: Quan, Evan <Evan.Quan@amd.com>
Sent: Friday, July 10, 2020 12:48 PM
To: amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Quan, Evan <Evan.Quan@amd.com>
Subject: [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
Before and after setting gfx clock soft max/min frequency.
Change-Id: I6f828da8de096ebc0ae27eaa89f988def2d547ec
Signed-off-by: Evan Quan <evan.quan@amd.com>
---
drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index c2779d0b51f6..33e0718f2635 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
+++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
@@ -1758,8 +1758,7 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
if (clk_id < 0)
return clk_id;
-if (clk_type == SMU_GFXCLK &&
- adev->asic_type == CHIP_SIENNA_CICHLID)
+if (clk_type == SMU_GFXCLK)
amdgpu_gfx_off_ctrl(adev, false);
if (max > 0) {
@@ -1779,8 +1778,7 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
}
out:
-if (clk_type == SMU_GFXCLK &&
- adev->asic_type == CHIP_SIENNA_CICHLID)
+if (clk_type == SMU_GFXCLK)
amdgpu_gfx_off_ctrl(adev, true);
return ret;
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* Re: [PATCH 02/16] drm/amd/powerplay: implement a common set dpm table API for smu V11
2020-07-10 4:47 ` [PATCH 02/16] drm/amd/powerplay: implement a common set dpm table API for smu V11 Evan Quan
@ 2020-07-13 14:49 ` Alex Deucher
0 siblings, 0 replies; 22+ messages in thread
From: Alex Deucher @ 2020-07-13 14:49 UTC (permalink / raw)
To: Evan Quan; +Cc: Deucher, Alexander, amd-gfx list
On Fri, Jul 10, 2020 at 12:48 AM Evan Quan <evan.quan@amd.com> wrote:
>
> Maximum the code sharing around smu V11.
>
> Change-Id: Ice0a874f3f70457f1012ca566f9f784ff3e9cd94
> Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
> ---
> drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h | 4 ++
> drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 38 +++++++++++++++++++
> 2 files changed, 42 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h
> index 289c571d6e4e..14d6eef8cf17 100644
> --- a/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h
> +++ b/drivers/gpu/drm/amd/powerplay/inc/smu_v11_0.h
> @@ -285,6 +285,10 @@ int smu_v11_0_get_dpm_level_count(struct smu_context *smu,
> enum smu_clk_type clk_type,
> uint32_t *value);
>
> +int smu_v11_0_set_single_dpm_table(struct smu_context *smu,
> + enum smu_clk_type clk_type,
> + struct smu_11_0_dpm_table *single_dpm_table);
> +
> int smu_v11_0_get_dpm_level_range(struct smu_context *smu,
> enum smu_clk_type clk_type,
> uint32_t *min_value,
> diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
> index 03be59492af1..7206b9f76042 100644
> --- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
> +++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
> @@ -1951,6 +1951,44 @@ int smu_v11_0_get_dpm_level_count(struct smu_context *smu,
> value);
> }
>
> +int smu_v11_0_set_single_dpm_table(struct smu_context *smu,
> + enum smu_clk_type clk_type,
> + struct smu_11_0_dpm_table *single_dpm_table)
> +{
> + int ret = 0;
> + uint32_t clk;
> + int i;
> +
> + ret = smu_v11_0_get_dpm_level_count(smu,
> + clk_type,
> + &single_dpm_table->count);
> + if (ret) {
> + dev_err(smu->adev->dev, "[%s] failed to get dpm levels!\n", __func__);
> + return ret;
> + }
> +
> + for (i = 0; i < single_dpm_table->count; i++) {
> + ret = smu_v11_0_get_dpm_freq_by_index(smu,
> + clk_type,
> + i,
> + &clk);
> + if (ret) {
> + dev_err(smu->adev->dev, "[%s] failed to get dpm freq by index!\n", __func__);
> + return ret;
> + }
> +
> + single_dpm_table->dpm_levels[i].value = clk;
> + single_dpm_table->dpm_levels[i].enabled = true;
> +
> + if (i == 0)
> + single_dpm_table->min = clk;
> + else if (i == single_dpm_table->count - 1)
> + single_dpm_table->max = clk;
> + }
> +
> + return 0;
> +}
> +
> int smu_v11_0_get_dpm_level_range(struct smu_context *smu,
> enum smu_clk_type clk_type,
> uint32_t *min_value,
> --
> 2.27.0
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 03/16] drm/amd/powerplay: update Arcturus default dpm table setting
2020-07-10 4:47 ` [PATCH 03/16] drm/amd/powerplay: update Arcturus default dpm table setting Evan Quan
@ 2020-07-13 14:49 ` Alex Deucher
0 siblings, 0 replies; 22+ messages in thread
From: Alex Deucher @ 2020-07-13 14:49 UTC (permalink / raw)
To: Evan Quan; +Cc: Deucher, Alexander, amd-gfx list
On Fri, Jul 10, 2020 at 12:48 AM Evan Quan <evan.quan@amd.com> wrote:
>
> Preparing for coming code sharing around performance level
> setting.
>
> Change-Id: Iaa77af7a272121503f09ad5fbfbe9dff2d2597b1
> Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
> ---
> drivers/gpu/drm/amd/powerplay/arcturus_ppt.c | 297 ++++++++-----------
> 1 file changed, 119 insertions(+), 178 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
> index 5b793e354704..a3747ab4af32 100644
> --- a/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
> +++ b/drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
> @@ -291,7 +291,6 @@ static int arcturus_get_pwr_src_index(struct smu_context *smc, uint32_t index)
> return mapping.map_to;
> }
>
> -
> static int arcturus_get_workload_type(struct smu_context *smu, enum PP_SMC_POWER_PROFILE profile)
> {
> struct smu_11_0_cmn2aisc_mapping mapping;
> @@ -338,23 +337,11 @@ static int arcturus_allocate_dpm_context(struct smu_context *smu)
> {
> struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
>
> - if (smu_dpm->dpm_context)
> - return -EINVAL;
> -
> - smu_dpm->dpm_context = kzalloc(sizeof(struct arcturus_dpm_table),
> + smu_dpm->dpm_context = kzalloc(sizeof(struct smu_11_0_dpm_context),
> GFP_KERNEL);
> if (!smu_dpm->dpm_context)
> return -ENOMEM;
> -
> - if (smu_dpm->golden_dpm_context)
> - return -EINVAL;
> -
> - smu_dpm->golden_dpm_context = kzalloc(sizeof(struct arcturus_dpm_table),
> - GFP_KERNEL);
> - if (!smu_dpm->golden_dpm_context)
> - return -ENOMEM;
> -
> - smu_dpm->dpm_context_size = sizeof(struct arcturus_dpm_table);
> + smu_dpm->dpm_context_size = sizeof(struct smu_11_0_dpm_context);
>
> smu_dpm->dpm_current_power_state = kzalloc(sizeof(struct smu_power_state),
> GFP_KERNEL);
> @@ -382,119 +369,84 @@ arcturus_get_allowed_feature_mask(struct smu_context *smu,
> return 0;
> }
>
> -static int
> -arcturus_set_single_dpm_table(struct smu_context *smu,
> - struct arcturus_single_dpm_table *single_dpm_table,
> - PPCLK_e clk_id)
> -{
> - int ret = 0;
> - uint32_t i, num_of_levels = 0, clk;
> -
> - ret = smu_send_smc_msg_with_param(smu,
> - SMU_MSG_GetDpmFreqByIndex,
> - (clk_id << 16 | 0xFF),
> - &num_of_levels);
> - if (ret) {
> - dev_err(smu->adev->dev, "[%s] failed to get dpm levels!\n", __func__);
> - return ret;
> - }
> -
> - single_dpm_table->count = num_of_levels;
> - for (i = 0; i < num_of_levels; i++) {
> - ret = smu_send_smc_msg_with_param(smu,
> - SMU_MSG_GetDpmFreqByIndex,
> - (clk_id << 16 | i),
> - &clk);
> - if (ret) {
> - dev_err(smu->adev->dev, "[%s] failed to get dpm freq by index!\n", __func__);
> - return ret;
> - }
> - single_dpm_table->dpm_levels[i].value = clk;
> - single_dpm_table->dpm_levels[i].enabled = true;
> - }
> - return 0;
> -}
> -
> -static void arcturus_init_single_dpm_state(struct arcturus_dpm_state *dpm_state)
> -{
> - dpm_state->soft_min_level = 0x0;
> - dpm_state->soft_max_level = 0xffff;
> - dpm_state->hard_min_level = 0x0;
> - dpm_state->hard_max_level = 0xffff;
> -}
> -
> static int arcturus_set_default_dpm_table(struct smu_context *smu)
> {
> - int ret;
> -
> - struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
> - struct arcturus_dpm_table *dpm_table = NULL;
> - struct arcturus_single_dpm_table *single_dpm_table;
> -
> - dpm_table = smu_dpm->dpm_context;
> + struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
> + PPTable_t *driver_ppt = smu->smu_table.driver_pptable;
> + struct smu_11_0_dpm_table *dpm_table = NULL;
> + int ret = 0;
>
> - /* socclk */
> - single_dpm_table = &(dpm_table->soc_table);
> + /* socclk dpm table setup */
> + dpm_table = &dpm_context->dpm_tables.soc_table;
> if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT)) {
> - ret = arcturus_set_single_dpm_table(smu, single_dpm_table,
> - PPCLK_SOCCLK);
> - if (ret) {
> - dev_err(smu->adev->dev, "[%s] failed to get socclk dpm levels!\n", __func__);
> + ret = smu_v11_0_set_single_dpm_table(smu,
> + SMU_SOCCLK,
> + dpm_table);
> + if (ret)
> return ret;
> - }
> + dpm_table->is_fine_grained =
> + !driver_ppt->DpmDescriptor[PPCLK_SOCCLK].SnapToDiscrete;
> } else {
> - single_dpm_table->count = 1;
> - single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.socclk / 100;
> + dpm_table->count = 1;
> + dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.socclk / 100;
> + dpm_table->dpm_levels[0].enabled = true;
> + dpm_table->min = dpm_table->dpm_levels[0].value;
> + dpm_table->max = dpm_table->dpm_levels[0].value;
> }
> - arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state));
>
> - /* gfxclk */
> - single_dpm_table = &(dpm_table->gfx_table);
> + /* gfxclk dpm table setup */
> + dpm_table = &dpm_context->dpm_tables.gfx_table;
> if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT)) {
> - ret = arcturus_set_single_dpm_table(smu, single_dpm_table,
> - PPCLK_GFXCLK);
> - if (ret) {
> - dev_err(smu->adev->dev, "[SetupDefaultDpmTable] failed to get gfxclk dpm levels!");
> + ret = smu_v11_0_set_single_dpm_table(smu,
> + SMU_GFXCLK,
> + dpm_table);
> + if (ret)
> return ret;
> - }
> + dpm_table->is_fine_grained =
> + !driver_ppt->DpmDescriptor[PPCLK_GFXCLK].SnapToDiscrete;
> } else {
> - single_dpm_table->count = 1;
> - single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100;
> + dpm_table->count = 1;
> + dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.gfxclk / 100;
> + dpm_table->dpm_levels[0].enabled = true;
> + dpm_table->min = dpm_table->dpm_levels[0].value;
> + dpm_table->max = dpm_table->dpm_levels[0].value;
> }
> - arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state));
>
> - /* memclk */
> - single_dpm_table = &(dpm_table->mem_table);
> + /* memclk dpm table setup */
> + dpm_table = &dpm_context->dpm_tables.uclk_table;
> if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) {
> - ret = arcturus_set_single_dpm_table(smu, single_dpm_table,
> - PPCLK_UCLK);
> - if (ret) {
> - dev_err(smu->adev->dev, "[SetupDefaultDpmTable] failed to get memclk dpm levels!");
> + ret = smu_v11_0_set_single_dpm_table(smu,
> + SMU_UCLK,
> + dpm_table);
> + if (ret)
> return ret;
> - }
> + dpm_table->is_fine_grained =
> + !driver_ppt->DpmDescriptor[PPCLK_UCLK].SnapToDiscrete;
> } else {
> - single_dpm_table->count = 1;
> - single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.uclk / 100;
> + dpm_table->count = 1;
> + dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.uclk / 100;
> + dpm_table->dpm_levels[0].enabled = true;
> + dpm_table->min = dpm_table->dpm_levels[0].value;
> + dpm_table->max = dpm_table->dpm_levels[0].value;
> }
> - arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state));
>
> - /* fclk */
> - single_dpm_table = &(dpm_table->fclk_table);
> + /* fclk dpm table setup */
> + dpm_table = &dpm_context->dpm_tables.fclk_table;
> if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_FCLK_BIT)) {
> - ret = arcturus_set_single_dpm_table(smu, single_dpm_table,
> - PPCLK_FCLK);
> - if (ret) {
> - dev_err(smu->adev->dev, "[SetupDefaultDpmTable] failed to get fclk dpm levels!");
> + ret = smu_v11_0_set_single_dpm_table(smu,
> + SMU_FCLK,
> + dpm_table);
> + if (ret)
> return ret;
> - }
> + dpm_table->is_fine_grained =
> + !driver_ppt->DpmDescriptor[PPCLK_FCLK].SnapToDiscrete;
> } else {
> - single_dpm_table->count = 1;
> - single_dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.fclk / 100;
> + dpm_table->count = 1;
> + dpm_table->dpm_levels[0].value = smu->smu_table.boot_values.fclk / 100;
> + dpm_table->dpm_levels[0].enabled = true;
> + dpm_table->min = dpm_table->dpm_levels[0].value;
> + dpm_table->max = dpm_table->dpm_levels[0].value;
> }
> - arcturus_init_single_dpm_state(&(single_dpm_table->dpm_state));
> -
> - memcpy(smu_dpm->golden_dpm_context, dpm_table,
> - sizeof(struct arcturus_dpm_table));
>
> return 0;
> }
> @@ -622,7 +574,7 @@ static int arcturus_populate_umd_state_clk(struct smu_context *smu)
>
> static int arcturus_get_clk_table(struct smu_context *smu,
> struct pp_clock_levels_with_latency *clocks,
> - struct arcturus_single_dpm_table *dpm_table)
> + struct smu_11_0_dpm_table *dpm_table)
> {
> int i, count;
>
> @@ -824,14 +776,14 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
> int i, now, size = 0;
> int ret = 0;
> struct pp_clock_levels_with_latency clocks;
> - struct arcturus_single_dpm_table *single_dpm_table;
> + struct smu_11_0_dpm_table *single_dpm_table;
> struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
> - struct arcturus_dpm_table *dpm_table = NULL;
> + struct smu_11_0_dpm_context *dpm_context = NULL;
>
> if (amdgpu_ras_intr_triggered())
> return snprintf(buf, PAGE_SIZE, "unavailable\n");
>
> - dpm_table = smu_dpm->dpm_context;
> + dpm_context = smu_dpm->dpm_context;
>
> switch (type) {
> case SMU_SCLK:
> @@ -841,7 +793,7 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
> return ret;
> }
>
> - single_dpm_table = &(dpm_table->gfx_table);
> + single_dpm_table = &(dpm_context->dpm_tables.gfx_table);
> ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table);
> if (ret) {
> dev_err(smu->adev->dev, "Attempt to get gfx clk levels Failed!");
> @@ -868,7 +820,7 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
> return ret;
> }
>
> - single_dpm_table = &(dpm_table->mem_table);
> + single_dpm_table = &(dpm_context->dpm_tables.uclk_table);
> ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table);
> if (ret) {
> dev_err(smu->adev->dev, "Attempt to get memory clk levels Failed!");
> @@ -891,7 +843,7 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
> return ret;
> }
>
> - single_dpm_table = &(dpm_table->soc_table);
> + single_dpm_table = &(dpm_context->dpm_tables.soc_table);
> ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table);
> if (ret) {
> dev_err(smu->adev->dev, "Attempt to get socclk levels Failed!");
> @@ -914,7 +866,7 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
> return ret;
> }
>
> - single_dpm_table = &(dpm_table->fclk_table);
> + single_dpm_table = &(dpm_context->dpm_tables.fclk_table);
> ret = arcturus_get_clk_table(smu, &clocks, single_dpm_table);
> if (ret) {
> dev_err(smu->adev->dev, "Attempt to get fclk levels Failed!");
> @@ -937,20 +889,19 @@ static int arcturus_print_clk_levels(struct smu_context *smu,
> return size;
> }
>
> -static int arcturus_upload_dpm_level(struct smu_context *smu, bool max,
> - uint32_t feature_mask)
> +static int arcturus_upload_dpm_level(struct smu_context *smu,
> + bool max,
> + uint32_t feature_mask,
> + uint32_t level)
> {
> - struct arcturus_single_dpm_table *single_dpm_table;
> - struct arcturus_dpm_table *dpm_table =
> + struct smu_11_0_dpm_context *dpm_context =
> smu->smu_dpm.dpm_context;
> uint32_t freq;
> int ret = 0;
>
> if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT) &&
> (feature_mask & FEATURE_DPM_GFXCLK_MASK)) {
> - single_dpm_table = &(dpm_table->gfx_table);
> - freq = max ? single_dpm_table->dpm_state.soft_max_level :
> - single_dpm_table->dpm_state.soft_min_level;
> + freq = dpm_context->dpm_tables.gfx_table.dpm_levels[level].value;
> ret = smu_send_smc_msg_with_param(smu,
> (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq),
> (PPCLK_GFXCLK << 16) | (freq & 0xffff),
> @@ -964,9 +915,7 @@ static int arcturus_upload_dpm_level(struct smu_context *smu, bool max,
>
> if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT) &&
> (feature_mask & FEATURE_DPM_UCLK_MASK)) {
> - single_dpm_table = &(dpm_table->mem_table);
> - freq = max ? single_dpm_table->dpm_state.soft_max_level :
> - single_dpm_table->dpm_state.soft_min_level;
> + freq = dpm_context->dpm_tables.uclk_table.dpm_levels[level].value;
> ret = smu_send_smc_msg_with_param(smu,
> (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq),
> (PPCLK_UCLK << 16) | (freq & 0xffff),
> @@ -980,9 +929,7 @@ static int arcturus_upload_dpm_level(struct smu_context *smu, bool max,
>
> if (smu_feature_is_enabled(smu, SMU_FEATURE_DPM_SOCCLK_BIT) &&
> (feature_mask & FEATURE_DPM_SOCCLK_MASK)) {
> - single_dpm_table = &(dpm_table->soc_table);
> - freq = max ? single_dpm_table->dpm_state.soft_max_level :
> - single_dpm_table->dpm_state.soft_min_level;
> + freq = dpm_context->dpm_tables.soc_table.dpm_levels[level].value;
> ret = smu_send_smc_msg_with_param(smu,
> (max ? SMU_MSG_SetSoftMaxByFreq : SMU_MSG_SetSoftMinByFreq),
> (PPCLK_SOCCLK << 16) | (freq & 0xffff),
> @@ -1000,8 +947,8 @@ static int arcturus_upload_dpm_level(struct smu_context *smu, bool max,
> static int arcturus_force_clk_levels(struct smu_context *smu,
> enum smu_clk_type type, uint32_t mask)
> {
> - struct arcturus_dpm_table *dpm_table;
> - struct arcturus_single_dpm_table *single_dpm_table;
> + struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
> + struct smu_11_0_dpm_table *single_dpm_table = NULL;
> uint32_t soft_min_level, soft_max_level;
> uint32_t smu_version;
> int ret = 0;
> @@ -1021,12 +968,9 @@ static int arcturus_force_clk_levels(struct smu_context *smu,
> soft_min_level = mask ? (ffs(mask) - 1) : 0;
> soft_max_level = mask ? (fls(mask) - 1) : 0;
>
> - dpm_table = smu->smu_dpm.dpm_context;
> -
> switch (type) {
> case SMU_SCLK:
> - single_dpm_table = &(dpm_table->gfx_table);
> -
> + single_dpm_table = &(dpm_context->dpm_tables.gfx_table);
> if (soft_max_level >= single_dpm_table->count) {
> dev_err(smu->adev->dev, "Clock level specified %d is over max allowed %d\n",
> soft_max_level, single_dpm_table->count - 1);
> @@ -1034,18 +978,19 @@ static int arcturus_force_clk_levels(struct smu_context *smu,
> break;
> }
>
> - single_dpm_table->dpm_state.soft_min_level =
> - single_dpm_table->dpm_levels[soft_min_level].value;
> - single_dpm_table->dpm_state.soft_max_level =
> - single_dpm_table->dpm_levels[soft_max_level].value;
> -
> - ret = arcturus_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK);
> + ret = arcturus_upload_dpm_level(smu,
> + false,
> + FEATURE_DPM_GFXCLK_MASK,
> + soft_min_level);
> if (ret) {
> dev_err(smu->adev->dev, "Failed to upload boot level to lowest!\n");
> break;
> }
>
> - ret = arcturus_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK);
> + ret = arcturus_upload_dpm_level(smu,
> + true,
> + FEATURE_DPM_GFXCLK_MASK,
> + soft_max_level);
> if (ret)
> dev_err(smu->adev->dev, "Failed to upload dpm max level to highest!\n");
>
> @@ -1256,8 +1201,7 @@ static int arcturus_get_fan_speed_percent(struct smu_context *smu,
> return ret;
> }
>
> -
> -static uint32_t arcturus_find_lowest_dpm_level(struct arcturus_single_dpm_table *table)
> +static uint32_t arcturus_find_lowest_dpm_level(struct smu_11_0_dpm_table *table)
> {
> uint32_t i;
>
> @@ -1274,7 +1218,7 @@ static uint32_t arcturus_find_lowest_dpm_level(struct arcturus_single_dpm_table
> }
>
> static uint32_t arcturus_find_highest_dpm_level(struct smu_context *smu,
> - struct arcturus_single_dpm_table *table)
> + struct smu_11_0_dpm_table *table)
> {
> int i = 0;
>
> @@ -1299,34 +1243,33 @@ static uint32_t arcturus_find_highest_dpm_level(struct smu_context *smu,
> return i;
> }
>
> -
> -
> static int arcturus_force_dpm_limit_value(struct smu_context *smu, bool highest)
> {
> - struct arcturus_dpm_table *dpm_table =
> - (struct arcturus_dpm_table *)smu->smu_dpm.dpm_context;
> + struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
> struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(smu->adev, 0);
> uint32_t soft_level;
> int ret = 0;
>
> /* gfxclk */
> if (highest)
> - soft_level = arcturus_find_highest_dpm_level(smu, &(dpm_table->gfx_table));
> + soft_level = arcturus_find_highest_dpm_level(smu, &(dpm_context->dpm_tables.gfx_table));
> else
> - soft_level = arcturus_find_lowest_dpm_level(&(dpm_table->gfx_table));
> + soft_level = arcturus_find_lowest_dpm_level(&(dpm_context->dpm_tables.gfx_table));
>
> - dpm_table->gfx_table.dpm_state.soft_min_level =
> - dpm_table->gfx_table.dpm_state.soft_max_level =
> - dpm_table->gfx_table.dpm_levels[soft_level].value;
> -
> - ret = arcturus_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK);
> + ret = arcturus_upload_dpm_level(smu,
> + false,
> + FEATURE_DPM_GFXCLK_MASK,
> + soft_level);
> if (ret) {
> dev_err(smu->adev->dev, "Failed to upload boot level to %s!\n",
> highest ? "highest" : "lowest");
> return ret;
> }
>
> - ret = arcturus_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK);
> + ret = arcturus_upload_dpm_level(smu,
> + true,
> + FEATURE_DPM_GFXCLK_MASK,
> + soft_level);
> if (ret) {
> dev_err(smu->adev->dev, "Failed to upload dpm max level to %s!\n!",
> highest ? "highest" : "lowest");
> @@ -1345,27 +1288,29 @@ static int arcturus_force_dpm_limit_value(struct smu_context *smu, bool highest)
>
> static int arcturus_unforce_dpm_levels(struct smu_context *smu)
> {
> - struct arcturus_dpm_table *dpm_table =
> - (struct arcturus_dpm_table *)smu->smu_dpm.dpm_context;
> + struct smu_11_0_dpm_context *dpm_context =
> + (struct smu_11_0_dpm_context *)smu->smu_dpm.dpm_context;
> struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(smu->adev, 0);
> uint32_t soft_min_level, soft_max_level;
> int ret = 0;
>
> /* gfxclk */
> - soft_min_level = arcturus_find_lowest_dpm_level(&(dpm_table->gfx_table));
> - soft_max_level = arcturus_find_highest_dpm_level(smu, &(dpm_table->gfx_table));
> - dpm_table->gfx_table.dpm_state.soft_min_level =
> - dpm_table->gfx_table.dpm_levels[soft_min_level].value;
> - dpm_table->gfx_table.dpm_state.soft_max_level =
> - dpm_table->gfx_table.dpm_levels[soft_max_level].value;
> -
> - ret = arcturus_upload_dpm_level(smu, false, FEATURE_DPM_GFXCLK_MASK);
> + soft_min_level = arcturus_find_lowest_dpm_level(&(dpm_context->dpm_tables.gfx_table));
> + soft_max_level = arcturus_find_highest_dpm_level(smu, &(dpm_context->dpm_tables.gfx_table));
> +
> + ret = arcturus_upload_dpm_level(smu,
> + false,
> + FEATURE_DPM_GFXCLK_MASK,
> + soft_min_level);
> if (ret) {
> dev_err(smu->adev->dev, "Failed to upload DPM Bootup Levels!");
> return ret;
> }
>
> - ret = arcturus_upload_dpm_level(smu, true, FEATURE_DPM_GFXCLK_MASK);
> + ret = arcturus_upload_dpm_level(smu,
> + true,
> + FEATURE_DPM_GFXCLK_MASK,
> + soft_max_level);
> if (ret) {
> dev_err(smu->adev->dev, "Failed to upload DPM Max Levels!");
> return ret;
> @@ -1388,18 +1333,15 @@ arcturus_get_profiling_clk_mask(struct smu_context *smu,
> uint32_t *mclk_mask,
> uint32_t *soc_mask)
> {
> - struct arcturus_dpm_table *dpm_table =
> - (struct arcturus_dpm_table *)smu->smu_dpm.dpm_context;
> - struct arcturus_single_dpm_table *gfx_dpm_table;
> - struct arcturus_single_dpm_table *mem_dpm_table;
> - struct arcturus_single_dpm_table *soc_dpm_table;
> + struct smu_11_0_dpm_context *dpm_context =
> + (struct smu_11_0_dpm_context *)smu->smu_dpm.dpm_context;
> + struct smu_11_0_dpm_table *gfx_dpm_table;
> + struct smu_11_0_dpm_table *mem_dpm_table;
> + struct smu_11_0_dpm_table *soc_dpm_table;
>
> - if (!smu->smu_dpm.dpm_context)
> - return -EINVAL;
> -
> - gfx_dpm_table = &dpm_table->gfx_table;
> - mem_dpm_table = &dpm_table->mem_table;
> - soc_dpm_table = &dpm_table->soc_table;
> + gfx_dpm_table = &dpm_context->dpm_tables.gfx_table;
> + mem_dpm_table = &dpm_context->dpm_tables.uclk_table;
> + soc_dpm_table = &dpm_context->dpm_tables.soc_table;
>
> *sclk_mask = 0;
> *mclk_mask = 0;
> @@ -2153,7 +2095,6 @@ static int arcturus_dpm_set_vcn_enable(struct smu_context *smu, bool enable)
> return ret;
> }
>
> -
> static void arcturus_fill_eeprom_i2c_req(SwI2cRequest_t *req, bool write,
> uint8_t address, uint32_t numbytes,
> uint8_t *data)
> --
> 2.27.0
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 12/16] drm/amd/powerplay: drop Sienna Cichlid specific set_soft_freq_limited_range
2020-07-10 4:47 ` [PATCH 12/16] drm/amd/powerplay: drop Sienna Cichlid specific set_soft_freq_limited_range Evan Quan
@ 2020-07-13 14:53 ` Alex Deucher
0 siblings, 0 replies; 22+ messages in thread
From: Alex Deucher @ 2020-07-13 14:53 UTC (permalink / raw)
To: Evan Quan; +Cc: Deucher, Alexander, amd-gfx list
On Fri, Jul 10, 2020 at 12:48 AM Evan Quan <evan.quan@amd.com> wrote:
>
> Use the common smu_v11_0_set_soft_freq_limited_range.
>
> Change-Id: I9f8772880b324ce9e741291751bb1b8ff4c36ea3
> Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
> ---
> .../drm/amd/powerplay/sienna_cichlid_ppt.c | 20 ++-----------------
> drivers/gpu/drm/amd/powerplay/smu_internal.h | 1 -
> drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 20 +++++++++++++++----
> 3 files changed, 18 insertions(+), 23 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
> index 27f77bde184f..141944df97b0 100644
> --- a/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
> +++ b/drivers/gpu/drm/amd/powerplay/sienna_cichlid_ppt.c
> @@ -1046,22 +1046,6 @@ static int sienna_cichlid_print_clk_levels(struct smu_context *smu,
> return size;
> }
>
> -int sienna_cichlid_set_soft_freq_limited_range(struct smu_context *smu,
> - enum smu_clk_type clk_type,
> - uint32_t min, uint32_t max)
> -{
> - struct amdgpu_device *adev = smu->adev;
> - int ret;
> -
> - if (clk_type == SMU_GFXCLK)
> - amdgpu_gfx_off_ctrl(adev, false);
> - ret = smu_v11_0_set_soft_freq_limited_range(smu, clk_type, min, max);
> - if (clk_type == SMU_GFXCLK)
> - amdgpu_gfx_off_ctrl(adev, true);
> -
> - return ret;
> -}
> -
> static int sienna_cichlid_force_clk_levels(struct smu_context *smu,
> enum smu_clk_type clk_type, uint32_t mask)
> {
> @@ -1097,7 +1081,7 @@ static int sienna_cichlid_force_clk_levels(struct smu_context *smu,
> if (ret)
> goto forec_level_out;
>
> - ret = sienna_cichlid_set_soft_freq_limited_range(smu, clk_type, min_freq, max_freq);
> + ret = smu_v11_0_set_soft_freq_limited_range(smu, clk_type, min_freq, max_freq);
> if (ret)
> goto forec_level_out;
> break;
> @@ -2566,7 +2550,7 @@ static const struct pptable_funcs sienna_cichlid_ppt_funcs = {
> .baco_enter = smu_v11_0_baco_enter,
> .baco_exit = smu_v11_0_baco_exit,
> .get_dpm_ultimate_freq = sienna_cichlid_get_dpm_ultimate_freq,
> - .set_soft_freq_limited_range = sienna_cichlid_set_soft_freq_limited_range,
> + .set_soft_freq_limited_range = smu_v11_0_set_soft_freq_limited_range,
> .override_pcie_parameters = smu_v11_0_override_pcie_parameters,
> .set_thermal_range = sienna_cichlid_set_thermal_range,
> };
> diff --git a/drivers/gpu/drm/amd/powerplay/smu_internal.h b/drivers/gpu/drm/amd/powerplay/smu_internal.h
> index 1c808ffe3ab1..91d3965bbe80 100644
> --- a/drivers/gpu/drm/amd/powerplay/smu_internal.h
> +++ b/drivers/gpu/drm/amd/powerplay/smu_internal.h
> @@ -93,7 +93,6 @@
> #define smu_asic_set_performance_level(smu, level) smu_ppt_funcs(set_performance_level, -EINVAL, smu, level)
> #define smu_dump_pptable(smu) smu_ppt_funcs(dump_pptable, 0, smu)
> #define smu_get_dpm_clk_limited(smu, clk_type, dpm_level, freq) smu_ppt_funcs(get_dpm_clk_limited, -EINVAL, smu, clk_type, dpm_level, freq)
> -#define smu_set_soft_freq_limited_range(smu, clk_type, min, max) smu_ppt_funcs(set_soft_freq_limited_range, -EINVAL, smu, clk_type, min, max)
> #define smu_override_pcie_parameters(smu) smu_ppt_funcs(override_pcie_parameters, 0, smu)
> #define smu_update_pcie_parameters(smu, pcie_gen_cap, pcie_width_cap) smu_ppt_funcs(update_pcie_parameters, 0, smu, pcie_gen_cap, pcie_width_cap)
> #define smu_set_thermal_range(smu, range) smu_ppt_funcs(set_thermal_range, 0, smu, range)
> diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
> index f96ff062eb64..c2779d0b51f6 100644
> --- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
> +++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
> @@ -1745,9 +1745,12 @@ int smu_v11_0_get_dpm_ultimate_freq(struct smu_context *smu, enum smu_clk_type c
> return ret;
> }
>
> -int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_type clk_type,
> - uint32_t min, uint32_t max)
> +int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
> + enum smu_clk_type clk_type,
> + uint32_t min,
> + uint32_t max)
> {
> + struct amdgpu_device *adev = smu->adev;
> int ret = 0, clk_id = 0;
> uint32_t param;
>
> @@ -1755,12 +1758,16 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_
> if (clk_id < 0)
> return clk_id;
>
> + if (clk_type == SMU_GFXCLK &&
> + adev->asic_type == CHIP_SIENNA_CICHLID)
> + amdgpu_gfx_off_ctrl(adev, false);
> +
> if (max > 0) {
> param = (uint32_t)((clk_id << 16) | (max & 0xffff));
> ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMaxByFreq,
> param, NULL);
> if (ret)
> - return ret;
> + goto out;
> }
>
> if (min > 0) {
> @@ -1768,9 +1775,14 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu, enum smu_clk_
> ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetSoftMinByFreq,
> param, NULL);
> if (ret)
> - return ret;
> + goto out;
> }
>
> +out:
> + if (clk_type == SMU_GFXCLK &&
> + adev->asic_type == CHIP_SIENNA_CICHLID)
> + amdgpu_gfx_off_ctrl(adev, true);
> +
> return ret;
> }
>
> --
> 2.27.0
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply [flat|nested] 22+ messages in thread
* RE: [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
2020-07-13 3:44 ` Quan, Evan
@ 2020-07-14 6:55 ` Quan, Evan
2020-07-14 13:48 ` Deucher, Alexander
0 siblings, 1 reply; 22+ messages in thread
From: Quan, Evan @ 2020-07-14 6:55 UTC (permalink / raw)
To: amd-gfx; +Cc: Deucher, Alexander
[AMD Official Use Only - Internal Distribution Only]
Hi Alex,
Can I have a RB for this patch also?
BR
Evan
-----Original Message-----
From: Quan, Evan
Sent: Monday, July 13, 2020 11:45 AM
To: amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander <Alexander.Deucher@amd.com>
Subject: RE: [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
Ping for this one and patch12, patch2 and patch3
BR
Evan
-----Original Message-----
From: Quan, Evan <Evan.Quan@amd.com>
Sent: Friday, July 10, 2020 12:48 PM
To: amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Quan, Evan <Evan.Quan@amd.com>
Subject: [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
Before and after setting gfx clock soft max/min frequency.
Change-Id: I6f828da8de096ebc0ae27eaa89f988def2d547ec
Signed-off-by: Evan Quan <evan.quan@amd.com>
---
drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index c2779d0b51f6..33e0718f2635 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
+++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
@@ -1758,8 +1758,7 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
if (clk_id < 0)
return clk_id;
-if (clk_type == SMU_GFXCLK &&
- adev->asic_type == CHIP_SIENNA_CICHLID)
+if (clk_type == SMU_GFXCLK)
amdgpu_gfx_off_ctrl(adev, false);
if (max > 0) {
@@ -1779,8 +1778,7 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
}
out:
-if (clk_type == SMU_GFXCLK &&
- adev->asic_type == CHIP_SIENNA_CICHLID)
+if (clk_type == SMU_GFXCLK)
amdgpu_gfx_off_ctrl(adev, true);
return ret;
--
2.27.0
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
* Re: [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
2020-07-14 6:55 ` Quan, Evan
@ 2020-07-14 13:48 ` Deucher, Alexander
0 siblings, 0 replies; 22+ messages in thread
From: Deucher, Alexander @ 2020-07-14 13:48 UTC (permalink / raw)
To: Quan, Evan, amd-gfx
[-- Attachment #1.1: Type: text/plain, Size: 2268 bytes --]
[AMD Public Use]
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
________________________________
From: Quan, Evan <Evan.Quan@amd.com>
Sent: Tuesday, July 14, 2020 2:55 AM
To: amd-gfx@lists.freedesktop.org <amd-gfx@lists.freedesktop.org>
Cc: Deucher, Alexander <Alexander.Deucher@amd.com>
Subject: RE: [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
[AMD Official Use Only - Internal Distribution Only]
Hi Alex,
Can I have a RB for this patch also?
BR
Evan
-----Original Message-----
From: Quan, Evan
Sent: Monday, July 13, 2020 11:45 AM
To: amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander <Alexander.Deucher@amd.com>
Subject: RE: [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
Ping for this one and patch12, patch2 and patch3
BR
Evan
-----Original Message-----
From: Quan, Evan <Evan.Quan@amd.com>
Sent: Friday, July 10, 2020 12:48 PM
To: amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Quan, Evan <Evan.Quan@amd.com>
Subject: [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs
Before and after setting gfx clock soft max/min frequency.
Change-Id: I6f828da8de096ebc0ae27eaa89f988def2d547ec
Signed-off-by: Evan Quan <evan.quan@amd.com>
---
drivers/gpu/drm/amd/powerplay/smu_v11_0.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index c2779d0b51f6..33e0718f2635 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
+++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
@@ -1758,8 +1758,7 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
if (clk_id < 0)
return clk_id;
-if (clk_type == SMU_GFXCLK &&
- adev->asic_type == CHIP_SIENNA_CICHLID)
+if (clk_type == SMU_GFXCLK)
amdgpu_gfx_off_ctrl(adev, false);
if (max > 0) {
@@ -1779,8 +1778,7 @@ int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu,
}
out:
-if (clk_type == SMU_GFXCLK &&
- adev->asic_type == CHIP_SIENNA_CICHLID)
+if (clk_type == SMU_GFXCLK)
amdgpu_gfx_off_ctrl(adev, true);
return ret;
--
2.27.0
[-- Attachment #1.2: Type: text/html, Size: 3627 bytes --]
[-- Attachment #2: Type: text/plain, Size: 154 bytes --]
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
^ permalink raw reply related [flat|nested] 22+ messages in thread
end of thread, other threads:[~2020-07-14 13:48 UTC | newest]
Thread overview: 22+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-07-10 4:47 [PATCH 01/16] drm/amd/powerplay: add more members for dpm table Evan Quan
2020-07-10 4:47 ` [PATCH 02/16] drm/amd/powerplay: implement a common set dpm table API for smu V11 Evan Quan
2020-07-13 14:49 ` Alex Deucher
2020-07-10 4:47 ` [PATCH 03/16] drm/amd/powerplay: update Arcturus default dpm table setting Evan Quan
2020-07-13 14:49 ` Alex Deucher
2020-07-10 4:47 ` [PATCH 04/16] drm/amd/powerplay: update Navi10 default dpm table setup Evan Quan
2020-07-10 4:47 ` [PATCH 05/16] drm/amd/powerplay: update Sienna Cichlid " Evan Quan
2020-07-10 4:47 ` [PATCH 06/16] drm/amd/powerplay: add new UMD pstate data structure Evan Quan
2020-07-10 4:47 ` [PATCH 07/16] drm/amd/powerplay: update UMD pstate clock settings Evan Quan
2020-07-10 4:47 ` [PATCH 08/16] drm/amd/powerplay: update the common API for performance level setting Evan Quan
2020-07-10 4:47 ` [PATCH 09/16] drm/amd/powerplay: drop unnecessary Arcturus specific APIs Evan Quan
2020-07-10 4:47 ` [PATCH 10/16] drm/amd/powerplay: drop unnecessary Navi1x " Evan Quan
2020-07-10 4:47 ` [PATCH 11/16] drm/amd/powerplay: drop unnecessary Sienna Cichlid " Evan Quan
2020-07-10 4:47 ` [PATCH 12/16] drm/amd/powerplay: drop Sienna Cichlid specific set_soft_freq_limited_range Evan Quan
2020-07-13 14:53 ` Alex Deucher
2020-07-10 4:47 ` [PATCH 13/16] drm/amd/powerplay: apply gfxoff disablement/enablement for all SMU11 ASICs Evan Quan
2020-07-13 3:44 ` Quan, Evan
2020-07-14 6:55 ` Quan, Evan
2020-07-14 13:48 ` Deucher, Alexander
2020-07-10 4:47 ` [PATCH 14/16] drm/amd/powerplay: drop unnecessary wrappers Evan Quan
2020-07-10 4:47 ` [PATCH 15/16] drm/amd/powerplay: drop smu_v12_0.c unnecessary wrapper Evan Quan
2020-07-10 4:47 ` [PATCH 16/16] drm/amd/powerplay: drop unused APIs and parameters Evan Quan
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.