Merge tag 'drm-fixes-2024-08-02' of https://gitlab.freedesktop.org/drm/kernel

Pull drm fixes from Dave Airlie:
 "Regular weekly fixes. This is a bit larger than usual but doesn't seem
  too crazy.

  Most of it is vmwgfx changes that fix a bunch of issues with wayland
  userspaces with dma-buf/external buffers and modesetting fixes.

  Otherwise it's kinda spread out, v3d fixes some new ioctls, nouveau
  has regression revert and fixes, amdgpu, i915 and ast have some small
  fixes, and some core fixes spread about.

  client:
   - fix error code

  atomic:
   - allow damage clips with async flips
   - allow explicit sync with async flips

  kselftests:
   - fix dmabuf-heaps test

  panic:
   - fix schedule_work in panic paths

  panel:
   - fix OrangePi Neo orientation

  gpuvm:
   - fix missing dependency

  amdgpu:
   - SMU 14.x update
   - Fix contiguous VRAM handling for IB parsing
   - GFX 12 fix
   - Regression fix for old APUs

  i915:
   - Static analysis fix for int overflow
   - Fix for HDCP2_STREAM_STATUS macro and removal of PWR_CLK_STATE for gen12

  nouveau:
   - revert busy wait change that caused a resume regression
   - fix buffer placement fault on dynamic pm s/r
   - fix refcount underflow

  ast:
   - fix black screen on resume
   - wake during connector status detect

  v3d:
   - fix issues with perf/timestamp ioctls

  vmwgfx:
   - fix deadlock in dma-buf fence polling
   - fix screen surface refcounting
   - fix dumb buffer handling
   - fix support for external buffers
   - fix overlay with screen targets
   - trigger modeset on screen moves"

* tag 'drm-fixes-2024-08-02' of https://gitlab.freedesktop.org/drm/kernel: (31 commits)
  Revert "nouveau: rip out busy fence waits"
  nouveau: set placement to original placement on uvmm validate.
  drm/atomic: Allow userspace to use damage clips with async flips
  drm/atomic: Allow userspace to use explicit sync with atomic async flips
  drm/i915: Fix possible int overflow in skl_ddi_calculate_wrpll()
  drm/i915/hdcp: Fix HDCP2_STREAM_STATUS macro
  drm/ast: astdp: Wake up during connector status detection
  i915/perf: Remove code to update PWR_CLK_STATE for gen12
  kselftests: dmabuf-heaps: Ensure the driver name is null-terminated
  drm/client: Fix error code in drm_client_buffer_vmap_local()
  drm/amdgpu: Fix APU handling in amdgpu_pm_load_smu_firmware()
  drm/amdgpu: increase mes log buffer size for gfx12
  drm/amdgpu: fix contiguous handling for IB parsing v2
  drm/amdgpu/pm: support gpu_metrics sysfs interface for smu v14.0.2/3
  drm/vmwgfx: Trigger a modeset when the screen moves
  drm/vmwgfx: Fix overlay when using Screen Targets
  drm/vmwgfx: Add basic support for external buffers
  drm/vmwgfx: Fix handling of dumb buffers
  drm/vmwgfx: Make sure the screen surface is ref counted
  drm/vmwgfx: Fix a deadlock in dma buf fence polling
  ...

2 files changed, 86 insertions, 3 deletions

diff --git a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
index a1b8a82d77cf..8b7d6ed7e2ed 100644
--- a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
+++ b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c
@@ -618,7 +618,8 @@ int amdgpu_pm_load_smu_firmware(struct amdgpu_device *adev, uint32_t *smu_versio
 	const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
 	int r = 0;
 
-	if (!pp_funcs || !pp_funcs->load_firmware || adev->flags & AMD_IS_APU)
+	if (!pp_funcs || !pp_funcs->load_firmware ||
+	    (is_support_sw_smu(adev) && (adev->flags & AMD_IS_APU)))
 		return 0;
 
 	mutex_lock(&adev->pm.mutex);
diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
index 98ea58d792ca..e1a27903c80a 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
@@ -66,6 +66,7 @@
 
 #define MP0_MP1_DATA_REGION_SIZE_COMBOPPTABLE	0x4000
 #define DEBUGSMC_MSG_Mode1Reset        2
+#define LINK_SPEED_MAX					3
 
 static struct cmn2asic_msg_mapping smu_v14_0_2_message_map[SMU_MSG_MAX_COUNT] = {
 	MSG_MAP(TestMessage,			PPSMC_MSG_TestMessage,                 1),
@@ -221,7 +222,6 @@ static struct cmn2asic_mapping smu_v14_0_2_workload_map[PP_SMC_POWER_PROFILE_COU
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_WINDOW3D,		WORKLOAD_PPLIB_WINDOW_3D_BIT),
 };
 
-#if 0
 static const uint8_t smu_v14_0_2_throttler_map[] = {
 	[THROTTLER_PPT0_BIT]		= (SMU_THROTTLER_PPT0_BIT),
 	[THROTTLER_PPT1_BIT]		= (SMU_THROTTLER_PPT1_BIT),
@@ -241,7 +241,6 @@ static const uint8_t smu_v14_0_2_throttler_map[] = {
 	[THROTTLER_GFX_APCC_PLUS_BIT]	= (SMU_THROTTLER_APCC_BIT),
 	[THROTTLER_FIT_BIT]		= (SMU_THROTTLER_FIT_BIT),
 };
-#endif
 
 static int
 smu_v14_0_2_get_allowed_feature_mask(struct smu_context *smu,
@@ -1869,6 +1868,88 @@ static ssize_t smu_v14_0_2_get_ecc_info(struct smu_context *smu,
 	return ret;
 }
 
+static ssize_t smu_v14_0_2_get_gpu_metrics(struct smu_context *smu,
+					   void **table)
+{
+	struct smu_table_context *smu_table = &smu->smu_table;
+	struct gpu_metrics_v1_3 *gpu_metrics =
+		(struct gpu_metrics_v1_3 *)smu_table->gpu_metrics_table;
+	SmuMetricsExternal_t metrics_ext;
+	SmuMetrics_t *metrics = &metrics_ext.SmuMetrics;
+	int ret = 0;
+
+	ret = smu_cmn_get_metrics_table(smu,
+					&metrics_ext,
+					true);
+	if (ret)
+		return ret;
+
+	smu_cmn_init_soft_gpu_metrics(gpu_metrics, 1, 3);
+
+	gpu_metrics->temperature_edge = metrics->AvgTemperature[TEMP_EDGE];
+	gpu_metrics->temperature_hotspot = metrics->AvgTemperature[TEMP_HOTSPOT];
+	gpu_metrics->temperature_mem = metrics->AvgTemperature[TEMP_MEM];
+	gpu_metrics->temperature_vrgfx = metrics->AvgTemperature[TEMP_VR_GFX];
+	gpu_metrics->temperature_vrsoc = metrics->AvgTemperature[TEMP_VR_SOC];
+	gpu_metrics->temperature_vrmem = max(metrics->AvgTemperature[TEMP_VR_MEM0],
+					     metrics->AvgTemperature[TEMP_VR_MEM1]);
+
+	gpu_metrics->average_gfx_activity = metrics->AverageGfxActivity;
+	gpu_metrics->average_umc_activity = metrics->AverageUclkActivity;
+	gpu_metrics->average_mm_activity = max(metrics->Vcn0ActivityPercentage,
+					       metrics->Vcn1ActivityPercentage);
+
+	gpu_metrics->average_socket_power = metrics->AverageSocketPower;
+	gpu_metrics->energy_accumulator = metrics->EnergyAccumulator;
+
+	if (metrics->AverageGfxActivity <= SMU_14_0_2_BUSY_THRESHOLD)
+		gpu_metrics->average_gfxclk_frequency = metrics->AverageGfxclkFrequencyPostDs;
+	else
+		gpu_metrics->average_gfxclk_frequency = metrics->AverageGfxclkFrequencyPreDs;
+
+	if (metrics->AverageUclkActivity <= SMU_14_0_2_BUSY_THRESHOLD)
+		gpu_metrics->average_uclk_frequency = metrics->AverageMemclkFrequencyPostDs;
+	else
+		gpu_metrics->average_uclk_frequency = metrics->AverageMemclkFrequencyPreDs;
+
+	gpu_metrics->average_vclk0_frequency = metrics->AverageVclk0Frequency;
+	gpu_metrics->average_dclk0_frequency = metrics->AverageDclk0Frequency;
+	gpu_metrics->average_vclk1_frequency = metrics->AverageVclk1Frequency;
+	gpu_metrics->average_dclk1_frequency = metrics->AverageDclk1Frequency;
+
+	gpu_metrics->current_gfxclk = gpu_metrics->average_gfxclk_frequency;
+	gpu_metrics->current_socclk = metrics->CurrClock[PPCLK_SOCCLK];
+	gpu_metrics->current_uclk = metrics->CurrClock[PPCLK_UCLK];
+	gpu_metrics->current_vclk0 = metrics->CurrClock[PPCLK_VCLK_0];
+	gpu_metrics->current_dclk0 = metrics->CurrClock[PPCLK_DCLK_0];
+	gpu_metrics->current_vclk1 = metrics->CurrClock[PPCLK_VCLK_0];
+	gpu_metrics->current_dclk1 = metrics->CurrClock[PPCLK_DCLK_0];
+
+	gpu_metrics->throttle_status =
+			smu_v14_0_2_get_throttler_status(metrics);
+	gpu_metrics->indep_throttle_status =
+			smu_cmn_get_indep_throttler_status(gpu_metrics->throttle_status,
+							   smu_v14_0_2_throttler_map);
+
+	gpu_metrics->current_fan_speed = metrics->AvgFanRpm;
+
+	gpu_metrics->pcie_link_width = metrics->PcieWidth;
+	if ((metrics->PcieRate - 1) > LINK_SPEED_MAX)
+		gpu_metrics->pcie_link_speed = pcie_gen_to_speed(1);
+	else
+		gpu_metrics->pcie_link_speed = pcie_gen_to_speed(metrics->PcieRate);
+
+	gpu_metrics->system_clock_counter = ktime_get_boottime_ns();
+
+	gpu_metrics->voltage_gfx = metrics->AvgVoltage[SVI_PLANE_VDD_GFX];
+	gpu_metrics->voltage_soc = metrics->AvgVoltage[SVI_PLANE_VDD_SOC];
+	gpu_metrics->voltage_mem = metrics->AvgVoltage[SVI_PLANE_VDDIO_MEM];
+
+	*table = (void *)gpu_metrics;
+
+	return sizeof(struct gpu_metrics_v1_3);
+}
+
 static const struct pptable_funcs smu_v14_0_2_ppt_funcs = {
 	.get_allowed_feature_mask = smu_v14_0_2_get_allowed_feature_mask,
 	.set_default_dpm_table = smu_v14_0_2_set_default_dpm_table,
@@ -1905,6 +1986,7 @@ static const struct pptable_funcs smu_v14_0_2_ppt_funcs = {
 	.enable_thermal_alert = smu_v14_0_enable_thermal_alert,
 	.disable_thermal_alert = smu_v14_0_disable_thermal_alert,
 	.notify_memory_pool_location = smu_v14_0_notify_memory_pool_location,
+	.get_gpu_metrics = smu_v14_0_2_get_gpu_metrics,
 	.set_soft_freq_limited_range = smu_v14_0_set_soft_freq_limited_range,
 	.init_pptable_microcode = smu_v14_0_init_pptable_microcode,
 	.populate_umd_state_clk = smu_v14_0_2_populate_umd_state_clk,