Merge tag 'amd-drm-fixes-6.1-2022-11-02' of...

Merge tag 'amd-drm-fixes-6.1-2022-11-02' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.1-2022-11-02:

amdgpu:
- DCN 3.1.4 fixes
- DCN 3.2.x fixes
- GC 11.x fixes
- Virtual display fix
- Fail suspend if resources can't be evicted
- SR-IOV fix
- Display PSR fix

amdkfd:
- Fix possible NULL pointer deref
- GC 11.x trap handler fix
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221103023257.10446-1-alexander.deucher@amd.com

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c

@@ -706,6 +706,13 @@ int amdgpu_amdkfd_submit_ib(struct amdgpu_device *adev,
 
 void amdgpu_amdkfd_set_compute_idle(struct amdgpu_device *adev, bool idle)
 {
+	/* Temporary workaround to fix issues observed in some
+	 * compute applications when GFXOFF is enabled on GFX11.
+	 */
+	if (IP_VERSION_MAJ(adev->ip_versions[GC_HWIP][0]) == 11) {
+		pr_debug("GFXOFF is %s\n", idle ? "enabled" : "disabled");
+		amdgpu_gfx_off_ctrl(adev, idle);
+	}
 	amdgpu_dpm_switch_power_profile(adev,
 					PP_SMC_POWER_PROFILE_COMPUTE,
 					!idle);

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -4060,15 +4060,18 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev)
  * at suspend time.
  *
  */
-static void amdgpu_device_evict_resources(struct amdgpu_device *adev)
+static int amdgpu_device_evict_resources(struct amdgpu_device *adev)
 {
+	int ret;
+
 	/* No need to evict vram on APUs for suspend to ram or s2idle */
 	if ((adev->in_s3 || adev->in_s0ix) && (adev->flags & AMD_IS_APU))
-		return;
+		return 0;
 
-	if (amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM))
+	ret = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM);
+	if (ret)
 		DRM_WARN("evicting device resources failed\n");
-
+	return ret;
 }
 
 /*
@@ -4118,7 +4121,9 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon)
 	if (!adev->in_s0ix)
 		amdgpu_amdkfd_suspend(adev, adev->in_runpm);
 
-	amdgpu_device_evict_resources(adev);
+	r = amdgpu_device_evict_resources(adev);
+	if (r)
+		return r;
 
 	amdgpu_fence_driver_hw_fini(adev);
 

drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c

@@ -2201,7 +2201,8 @@ amdgpu_pci_remove(struct pci_dev *pdev)
 		pm_runtime_forbid(dev->dev);
 	}
 
-	if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2)) {
+	if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2) &&
+	    !amdgpu_sriov_vf(adev)) {
 		bool need_to_reset_gpu = false;
 
 		if (adev->gmc.xgmi.num_physical_nodes > 1) {

drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c

@@ -337,12 +337,14 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 		fw_info->feature = adev->psp.cap_feature_version;
 		break;
 	case AMDGPU_INFO_FW_MES_KIQ:
-		fw_info->ver = adev->mes.ucode_fw_version[0];
-		fw_info->feature = 0;
+		fw_info->ver = adev->mes.kiq_version & AMDGPU_MES_VERSION_MASK;
+		fw_info->feature = (adev->mes.kiq_version & AMDGPU_MES_FEAT_VERSION_MASK)
+					>> AMDGPU_MES_FEAT_VERSION_SHIFT;
 		break;
 	case AMDGPU_INFO_FW_MES:
-		fw_info->ver = adev->mes.ucode_fw_version[1];
-		fw_info->feature = 0;
+		fw_info->ver = adev->mes.sched_version & AMDGPU_MES_VERSION_MASK;
+		fw_info->feature = (adev->mes.sched_version & AMDGPU_MES_FEAT_VERSION_MASK)
+					>> AMDGPU_MES_FEAT_VERSION_SHIFT;
 		break;
 	case AMDGPU_INFO_FW_IMU:
 		fw_info->ver = adev->gfx.imu_fw_version;

drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c

@@ -500,6 +500,8 @@ static int amdgpu_vkms_sw_init(void *handle)
 
 	adev_to_drm(adev)->mode_config.fb_base = adev->gmc.aper_base;
 
+	adev_to_drm(adev)->mode_config.fb_modifiers_not_supported = true;
+
 	r = amdgpu_display_modeset_create_props(adev);
 	if (r)
 		return r;

drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm

@@ -186,6 +186,12 @@ L_SKIP_RESTORE:
 	s_getreg_b32	s_save_trapsts, hwreg(HW_REG_TRAPSTS)
 
 #if SW_SA_TRAP
+	// If ttmp1[30] is set then issue s_barrier to unblock dependent waves.
+	s_bitcmp1_b32	s_save_pc_hi, 30
+	s_cbranch_scc0	L_TRAP_NO_BARRIER
+	s_barrier
+
+L_TRAP_NO_BARRIER:
 	// If ttmp1[31] is set then trap may occur early.
 	// Spin wait until SAVECTX exception is raised.
 	s_bitcmp1_b32	s_save_pc_hi, 31

drivers/gpu/drm/amd/amdkfd/kfd_migrate.c

@@ -973,12 +973,10 @@ static vm_fault_t svm_migrate_to_ram(struct vm_fault *vmf)
 out_unlock_svms:
 	mutex_unlock(&p->svms.lock);
 out_unref_process:
+	pr_debug("CPU fault svms 0x%p address 0x%lx done\n", &p->svms, addr);
 	kfd_unref_process(p);
 out_mmput:
 	mmput(mm);
-
-	pr_debug("CPU fault svms 0x%p address 0x%lx done\n", &p->svms, addr);
-
 	return r ? VM_FAULT_SIGBUS : 0;
 }
 

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c

@@ -1549,6 +1549,9 @@ static int amdgpu_dm_init(struct amdgpu_device *adev)
 
 	adev->dm.dc->debug.visual_confirm = amdgpu_dc_visual_confirm;
 
+	/* TODO: Remove after DP2 receiver gets proper support of Cable ID feature */
+	adev->dm.dc->debug.ignore_cable_id = true;
+
 	r = dm_dmub_hw_init(adev);
 	if (r) {
 		DRM_ERROR("DMUB interface failed to initialize: status=%d\n", r);

drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c

@@ -157,6 +157,7 @@ void dcn32_init_clocks(struct clk_mgr *clk_mgr_base)
 	struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base);
 	unsigned int num_levels;
 	struct clk_limit_num_entries *num_entries_per_clk = &clk_mgr_base->bw_params->clk_table.num_entries_per_clk;
+	unsigned int i;
 
 	memset(&(clk_mgr_base->clks), 0, sizeof(struct dc_clocks));
 	clk_mgr_base->clks.p_state_change_support = true;
@@ -205,18 +206,17 @@ void dcn32_init_clocks(struct clk_mgr *clk_mgr_base)
 		clk_mgr->dpm_present = true;
 
 	if (clk_mgr_base->ctx->dc->debug.min_disp_clk_khz) {
-		unsigned int i;
-
 		for (i = 0; i < num_levels; i++)
 			if (clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz
 					< khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_disp_clk_khz))
 				clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz
 					= khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_disp_clk_khz);
 	}
+	for (i = 0; i < num_levels; i++)
+		if (clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz > 1950)
+			clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz = 1950;
 
 	if (clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz) {
-		unsigned int i;
-
 		for (i = 0; i < num_levels; i++)
 			if (clk_mgr_base->bw_params->clk_table.entries[i].dppclk_mhz
 					< khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz))
@@ -669,6 +669,9 @@ static void dcn32_get_memclk_states_from_smu(struct clk_mgr *clk_mgr_base)
 			&clk_mgr_base->bw_params->clk_table.entries[0].memclk_mhz,
 			&num_entries_per_clk->num_memclk_levels);
 
+	/* memclk must have at least one level */
+	num_entries_per_clk->num_memclk_levels = num_entries_per_clk->num_memclk_levels ? num_entries_per_clk->num_memclk_levels : 1;
+
 	dcn32_init_single_clock(clk_mgr, PPCLK_FCLK,
 			&clk_mgr_base->bw_params->clk_table.entries[0].fclk_mhz,
 			&num_entries_per_clk->num_fclk_levels);

drivers/gpu/drm/amd/display/dc/dc.h

@@ -852,6 +852,7 @@ struct dc_debug_options {
 	bool enable_double_buffered_dsc_pg_support;
 	bool enable_dp_dig_pixel_rate_div_policy;
 	enum lttpr_mode lttpr_mode_override;
+	unsigned int dsc_delay_factor_wa_x1000;
 };
 
 struct gpu_info_soc_bounding_box_v1_0;

drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.c

@@ -623,6 +623,10 @@ void hubp2_cursor_set_attributes(
 	hubp->att.size.bits.width    = attr->width;
 	hubp->att.size.bits.height   = attr->height;
 	hubp->att.cur_ctl.bits.mode  = attr->color_format;
+
+	hubp->cur_rect.w = attr->width;
+	hubp->cur_rect.h = attr->height;
+
 	hubp->att.cur_ctl.bits.pitch = hw_pitch;
 	hubp->att.cur_ctl.bits.line_per_chunk = lpc;
 	hubp->att.cur_ctl.bits.cur_2x_magnify = attr->attribute_flags.bits.ENABLE_MAGNIFICATION;

drivers/gpu/drm/amd/display/dc/dcn314/dcn314_resource.c

@@ -847,7 +847,7 @@ static const struct resource_caps res_cap_dcn314 = {
 	.num_ddc = 5,
 	.num_vmid = 16,
 	.num_mpc_3dlut = 2,
-	.num_dsc = 3,
+	.num_dsc = 4,
 };
 
 static const struct dc_plane_cap plane_cap = {

drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c

@@ -1228,6 +1228,7 @@ int dcn20_populate_dml_pipes_from_context(
 		pipes[pipe_cnt].pipe.src.dcc = false;
 		pipes[pipe_cnt].pipe.src.dcc_rate = 1;
 		pipes[pipe_cnt].pipe.dest.synchronized_vblank_all_planes = synchronized_vblank;
+		pipes[pipe_cnt].pipe.dest.synchronize_timings = synchronized_vblank;
 		pipes[pipe_cnt].pipe.dest.hblank_start = timing->h_total - timing->h_front_porch;
 		pipes[pipe_cnt].pipe.dest.hblank_end = pipes[pipe_cnt].pipe.dest.hblank_start
 				- timing->h_addressable

drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c

@@ -2359,9 +2359,11 @@ void dcn32_update_bw_bounding_box_fpu(struct dc *dc, struct clk_bw_params *bw_pa
 
 		if (dc->ctx->dc_bios->vram_info.dram_channel_width_bytes)
 			dcn3_2_soc.dram_channel_width_bytes = dc->ctx->dc_bios->vram_info.dram_channel_width_bytes;
-
 	}
 
+	/* DML DSC delay factor workaround */
+	dcn3_2_ip.dsc_delay_factor_wa = dc->debug.dsc_delay_factor_wa_x1000 / 1000.0;
+
 	/* Override dispclk_dppclk_vco_speed_mhz from Clk Mgr */
 	dcn3_2_soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0;
 	dc->dml.soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0;

drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c

@@ -364,10 +364,11 @@ static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerforman
 	for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k) {
 		v->DSCDelay[k] = dml32_DSCDelayRequirement(mode_lib->vba.DSCEnabled[k],
 				mode_lib->vba.ODMCombineEnabled[k], mode_lib->vba.DSCInputBitPerComponent[k],
-				mode_lib->vba.OutputBpp[k], mode_lib->vba.HActive[k], mode_lib->vba.HTotal[k],
+				mode_lib->vba.OutputBppPerState[mode_lib->vba.VoltageLevel][k],
+				mode_lib->vba.HActive[k], mode_lib->vba.HTotal[k],
 				mode_lib->vba.NumberOfDSCSlices[k], mode_lib->vba.OutputFormat[k],
 				mode_lib->vba.Output[k], mode_lib->vba.PixelClock[k],
-				mode_lib->vba.PixelClockBackEnd[k]);
+				mode_lib->vba.PixelClockBackEnd[k], mode_lib->vba.ip.dsc_delay_factor_wa);
 	}
 
 	for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k)
@@ -1627,7 +1628,7 @@ static void mode_support_configuration(struct vba_vars_st *v,
 				&& !mode_lib->vba.MSOOrODMSplitWithNonDPLink
 				&& !mode_lib->vba.NotEnoughLanesForMSO
 				&& mode_lib->vba.LinkCapacitySupport[i] == true && !mode_lib->vba.P2IWith420
-				&& !mode_lib->vba.DSCOnlyIfNecessaryWithBPP
+				//&& !mode_lib->vba.DSCOnlyIfNecessaryWithBPP
 				&& !mode_lib->vba.DSC422NativeNotSupported
 				&& !mode_lib->vba.MPCCombineMethodIncompatible
 				&& mode_lib->vba.ODMCombine2To1SupportCheckOK[i] == true
@@ -2475,7 +2476,8 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l
 					mode_lib->vba.OutputBppPerState[i][k], mode_lib->vba.HActive[k],
 					mode_lib->vba.HTotal[k], mode_lib->vba.NumberOfDSCSlices[k],
 					mode_lib->vba.OutputFormat[k], mode_lib->vba.Output[k],
-					mode_lib->vba.PixelClock[k], mode_lib->vba.PixelClockBackEnd[k]);
+					mode_lib->vba.PixelClock[k], mode_lib->vba.PixelClockBackEnd[k],
+					mode_lib->vba.ip.dsc_delay_factor_wa);
 		}
 
 		for (k = 0; k <= mode_lib->vba.NumberOfActiveSurfaces - 1; k++) {

drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c

@@ -1726,7 +1726,8 @@ unsigned int dml32_DSCDelayRequirement(bool DSCEnabled,
 		enum output_format_class  OutputFormat,
 		enum output_encoder_class Output,
 		double PixelClock,
-		double PixelClockBackEnd)
+		double PixelClockBackEnd,
+		double dsc_delay_factor_wa)
 {
 	unsigned int DSCDelayRequirement_val;
 
@@ -1746,7 +1747,7 @@ unsigned int dml32_DSCDelayRequirement(bool DSCEnabled,
 		}
 
 		DSCDelayRequirement_val = DSCDelayRequirement_val + (HTotal - HActive) *
-				dml_ceil(DSCDelayRequirement_val / HActive, 1);
+				dml_ceil((double)DSCDelayRequirement_val / HActive, 1);
 
 		DSCDelayRequirement_val = DSCDelayRequirement_val * PixelClock / PixelClockBackEnd;
 
@@ -1764,7 +1765,7 @@ unsigned int dml32_DSCDelayRequirement(bool DSCEnabled,
 	dml_print("DML::%s: DSCDelayRequirement_val = %d\n", __func__, DSCDelayRequirement_val);
 #endif
 
-	return DSCDelayRequirement_val;
+	return dml_ceil(DSCDelayRequirement_val * dsc_delay_factor_wa, 1);
 }
 
 void dml32_CalculateSurfaceSizeInMall(

drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.h

@@ -327,7 +327,8 @@ unsigned int dml32_DSCDelayRequirement(bool DSCEnabled,
 		enum output_format_class  OutputFormat,
 		enum output_encoder_class Output,
 		double PixelClock,
-		double PixelClockBackEnd);
+		double PixelClockBackEnd,
+		double dsc_delay_factor_wa);
 
 void dml32_CalculateSurfaceSizeInMall(
 		unsigned int NumberOfActiveSurfaces,

drivers/gpu/drm/amd/display/dc/dml/dcn32/display_rq_dlg_calc_32.c

@@ -291,8 +291,8 @@ void dml32_rq_dlg_get_dlg_reg(struct display_mode_lib *mode_lib,
 
 	dml_print("DML_DLG: %s: vready_after_vcount0 = %d\n", __func__, dlg_regs->vready_after_vcount0);
 
-	dst_x_after_scaler = get_dst_x_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
-	dst_y_after_scaler = get_dst_y_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx);
+	dst_x_after_scaler = dml_ceil(get_dst_x_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx), 1);
+	dst_y_after_scaler = dml_ceil(get_dst_y_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx), 1);
 
 	// do some adjustment on the dst_after scaler to account for odm combine mode
 	dml_print("DML_DLG: %s: input dst_x_after_scaler   = %d\n", __func__, dst_x_after_scaler);

drivers/gpu/drm/amd/display/dc/dml/dcn321/dcn321_fpu.c

@@ -29,6 +29,7 @@
 #include "dcn321_fpu.h"
 #include "dcn32/dcn32_resource.h"
 #include "dcn321/dcn321_resource.h"
+#include "dml/dcn32/display_mode_vba_util_32.h"
 
 #define DCN3_2_DEFAULT_DET_SIZE 256
 
@@ -119,15 +120,15 @@ struct _vcs_dpi_soc_bounding_box_st dcn3_21_soc = {
 		},
 	},
 	.num_states = 1,
-	.sr_exit_time_us = 12.36,
-	.sr_enter_plus_exit_time_us = 16.72,
+	.sr_exit_time_us = 19.95,
+	.sr_enter_plus_exit_time_us = 24.36,
 	.sr_exit_z8_time_us = 285.0,
 	.sr_enter_plus_exit_z8_time_us = 320,
 	.writeback_latency_us = 12.0,
 	.round_trip_ping_latency_dcfclk_cycles = 263,
-	.urgent_latency_pixel_data_only_us = 4.0,
-	.urgent_latency_pixel_mixed_with_vm_data_us = 4.0,
-	.urgent_latency_vm_data_only_us = 4.0,
+	.urgent_latency_pixel_data_only_us = 9.35,
+	.urgent_latency_pixel_mixed_with_vm_data_us = 9.35,
+	.urgent_latency_vm_data_only_us = 9.35,
 	.fclk_change_latency_us = 20,
 	.usr_retraining_latency_us = 2,
 	.smn_latency_us = 2,
@@ -538,9 +539,11 @@ void dcn321_update_bw_bounding_box_fpu(struct dc *dc, struct clk_bw_params *bw_p
 
 		if (dc->ctx->dc_bios->vram_info.dram_channel_width_bytes)
 			dcn3_21_soc.dram_channel_width_bytes = dc->ctx->dc_bios->vram_info.dram_channel_width_bytes;
-
 	}
 
+	/* DML DSC delay factor workaround */
+	dcn3_21_ip.dsc_delay_factor_wa = dc->debug.dsc_delay_factor_wa_x1000 / 1000.0;
+
 	/* Override dispclk_dppclk_vco_speed_mhz from Clk Mgr */
 	dcn3_21_soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0;
 	dc->dml.soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0;

drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h

@@ -364,6 +364,9 @@ struct _vcs_dpi_ip_params_st {
 	unsigned int max_num_dp2p0_outputs;
 	unsigned int max_num_dp2p0_streams;
 	unsigned int VBlankNomDefaultUS;
+
+	/* DM workarounds */
+	double dsc_delay_factor_wa; // TODO: Remove after implementing root cause fix
 };
 
 struct _vcs_dpi_display_xfc_params_st {

drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.c

@@ -625,7 +625,7 @@ static void fetch_pipe_params(struct display_mode_lib *mode_lib)
 		mode_lib->vba.skip_dio_check[mode_lib->vba.NumberOfActivePlanes] =
 				dout->is_virtual;
 
-		if (!dout->dsc_enable)
+		if (dout->dsc_enable)
 			mode_lib->vba.ForcedOutputLinkBPP[mode_lib->vba.NumberOfActivePlanes] = dout->output_bpp;
 		else
 			mode_lib->vba.ForcedOutputLinkBPP[mode_lib->vba.NumberOfActivePlanes] = 0.0;