Merge tag 'drm-fixes-2019-11-08' of git://anongit.freedesktop.org/drm…

…/drm

Pull drm fixes from Dave Airlie:
 "Weekly fixes for drm: amdgpu has a few but they are pretty scattered
  fixes, the fbdev one is a build regression fix that we didn't want to
  risk leaving out, otherwise a couple of i915, one radeon and a core
  atomic fix.

  core:
   - add missing documentation for GEM shmem madvise helpers
   - Fix for a state dereference in atomic self-refresh helpers

  fbdev:
   - One compilation fix for c2p fbdev helpers

  amdgpu:
   - Fix navi14 display issue root cause and revert workaround
   - GPU reset scheduler interaction fix
   - Fix fan boost on multi-GPU
   - Gfx10 and sdma5 fixes for navi
   - GFXOFF fix for renoir
   - Add navi14 PCI ID
   - GPUVM fix for arcturus

  radeon:
   - Port an SI power fix from amdgpu

  i915:
   - Fix HPD poll to avoid kworker consuming a lot of cpu cycles.
   - Do not use TBT type for non Type-C ports"

* tag 'drm-fixes-2019-11-08' of git://anongit.freedesktop.org/drm/drm:
  drm/radeon: fix si_enable_smc_cac() failed issue
  drm/amdgpu/renoir: move gfxoff handling into gfx9 module
  drm/amdgpu: add warning for GRBM 1-cycle delay issue in gfx9
  drm/amdgpu: add dummy read by engines for some GCVM status registers in gfx10
  drm/amdgpu: register gpu instance before fan boost feature enablment
  drm/amd/swSMU: fix smu workload bit map error
  drm/shmem: Add docbook comments for drm_gem_shmem_object madvise fields
  drm/amdgpu: add navi14 PCI ID
  Revert "drm/amd/display: setting the DIG_MODE to the correct value."
  drm/amd/display: Add ENGINE_ID_DIGD condition check for Navi14
  drm/amdgpu: dont schedule jobs while in reset
  drm/amdgpu/arcturus: properly set BANK_SELECT and FRAGMENT_SIZE
  drm/atomic: fix self-refresh helpers crtc state dereference
  drm/i915/dp: Do not switch aux to TBT mode for non-TC ports
  drm/i915: Avoid HPD poll detect triggering a new detect cycle
  fbdev: c2p: Fix link failure on non-inlining

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

@@ -604,8 +604,11 @@ void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr)
 			continue;
 		}
 
-		for (i = 0; i < num_entities; i++)
+		for (i = 0; i < num_entities; i++) {
+			mutex_lock(&ctx->adev->lock_reset);
 			drm_sched_entity_fini(&ctx->entities[0][i].entity);
+			mutex_unlock(&ctx->adev->lock_reset);
+		}
 	}
 }
 

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

@@ -2885,6 +2885,13 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 			DRM_INFO("amdgpu: acceleration disabled, skipping benchmarks\n");
 	}
 
+	/*
+	 * Register gpu instance before amdgpu_device_enable_mgpu_fan_boost.
+	 * Otherwise the mgpu fan boost feature will be skipped due to the
+	 * gpu instance is counted less.
+	 */
+	amdgpu_register_gpu_instance(adev);
+
 	/* enable clockgating, etc. after ib tests, etc. since some blocks require
 	 * explicit gating rather than handling it automatically.
 	 */

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

@@ -1016,6 +1016,7 @@ static const struct pci_device_id pciidlist[] = {
 	{0x1002, 0x7340, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI14|AMD_EXP_HW_SUPPORT},
 	{0x1002, 0x7341, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI14|AMD_EXP_HW_SUPPORT},
 	{0x1002, 0x7347, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI14|AMD_EXP_HW_SUPPORT},
+	{0x1002, 0x734F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI14|AMD_EXP_HW_SUPPORT},
 
 	/* Renoir */
 	{0x1002, 0x1636, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RENOIR|AMD_IS_APU|AMD_EXP_HW_SUPPORT},

drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h

@@ -289,6 +289,7 @@ struct amdgpu_gfx {
 	uint32_t			mec2_feature_version;
 	bool				mec_fw_write_wait;
 	bool				me_fw_write_wait;
+	bool				cp_fw_write_wait;
 	struct amdgpu_ring		gfx_ring[AMDGPU_MAX_GFX_RINGS];
 	unsigned			num_gfx_rings;
 	struct amdgpu_ring		compute_ring[AMDGPU_MAX_COMPUTE_RINGS];

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

@@ -190,7 +190,6 @@ int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags)
 		pm_runtime_put_autosuspend(dev->dev);
 	}
 
-	amdgpu_register_gpu_instance(adev);
 out:
 	if (r) {
 		/* balance pm_runtime_get_sync in amdgpu_driver_unload_kms */

drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c

@@ -564,6 +564,32 @@ static void gfx_v10_0_free_microcode(struct amdgpu_device *adev)
 	kfree(adev->gfx.rlc.register_list_format);
 }
 
+static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev)
+{
+	adev->gfx.cp_fw_write_wait = false;
+
+	switch (adev->asic_type) {
+	case CHIP_NAVI10:
+	case CHIP_NAVI12:
+	case CHIP_NAVI14:
+		if ((adev->gfx.me_fw_version >= 0x00000046) &&
+		    (adev->gfx.me_feature_version >= 27) &&
+		    (adev->gfx.pfp_fw_version >= 0x00000068) &&
+		    (adev->gfx.pfp_feature_version >= 27) &&
+		    (adev->gfx.mec_fw_version >= 0x0000005b) &&
+		    (adev->gfx.mec_feature_version >= 27))
+			adev->gfx.cp_fw_write_wait = true;
+		break;
+	default:
+		break;
+	}
+
+	if (adev->gfx.cp_fw_write_wait == false)
+		DRM_WARN_ONCE("Warning: check cp_fw_version and update it to realize \
+			      GRBM requires 1-cycle delay in cp firmware\n");
+}
+
+
 static void gfx_v10_0_init_rlc_ext_microcode(struct amdgpu_device *adev)
 {
 	const struct rlc_firmware_header_v2_1 *rlc_hdr;
@@ -832,6 +858,7 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev)
 		}
 	}
 
+	gfx_v10_0_check_fw_write_wait(adev);
 out:
 	if (err) {
 		dev_err(adev->dev,
@@ -4765,6 +4792,24 @@ static void gfx_v10_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
 	gfx_v10_0_wait_reg_mem(ring, 0, 0, 0, reg, 0, val, mask, 0x20);
 }
 
+static void gfx_v10_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						   uint32_t reg0, uint32_t reg1,
+						   uint32_t ref, uint32_t mask)
+{
+	int usepfp = (ring->funcs->type == AMDGPU_RING_TYPE_GFX);
+	struct amdgpu_device *adev = ring->adev;
+	bool fw_version_ok = false;
+
+	fw_version_ok = adev->gfx.cp_fw_write_wait;
+
+	if (fw_version_ok)
+		gfx_v10_0_wait_reg_mem(ring, usepfp, 0, 1, reg0, reg1,
+				       ref, mask, 0x20);
+	else
+		amdgpu_ring_emit_reg_write_reg_wait_helper(ring, reg0, reg1,
+							   ref, mask);
+}
+
 static void
 gfx_v10_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
 				      uint32_t me, uint32_t pipe,
@@ -5155,6 +5200,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_gfx = {
 	.emit_tmz = gfx_v10_0_ring_emit_tmz,
 	.emit_wreg = gfx_v10_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v10_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait,
 };
 
 static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_compute = {
@@ -5188,6 +5234,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_compute = {
 	.pad_ib = amdgpu_ring_generic_pad_ib,
 	.emit_wreg = gfx_v10_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v10_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait,
 };
 
 static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_kiq = {
@@ -5218,6 +5265,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_kiq = {
 	.emit_rreg = gfx_v10_0_ring_emit_rreg,
 	.emit_wreg = gfx_v10_0_ring_emit_wreg,
 	.emit_reg_wait = gfx_v10_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait,
 };
 
 static void gfx_v10_0_set_ring_funcs(struct amdgpu_device *adev)

drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c

@@ -973,6 +973,13 @@ static void gfx_v9_0_check_fw_write_wait(struct amdgpu_device *adev)
 	adev->gfx.me_fw_write_wait = false;
 	adev->gfx.mec_fw_write_wait = false;
 
+	if ((adev->gfx.mec_fw_version < 0x000001a5) ||
+	    (adev->gfx.mec_feature_version < 46) ||
+	    (adev->gfx.pfp_fw_version < 0x000000b7) ||
+	    (adev->gfx.pfp_feature_version < 46))
+		DRM_WARN_ONCE("Warning: check cp_fw_version and update it to realize \
+			      GRBM requires 1-cycle delay in cp firmware\n");
+
 	switch (adev->asic_type) {
 	case CHIP_VEGA10:
 		if ((adev->gfx.me_fw_version >= 0x0000009c) &&
@@ -1039,6 +1046,12 @@ static void gfx_v9_0_check_if_need_gfxoff(struct amdgpu_device *adev)
 			    !adev->gfx.rlc.is_rlc_v2_1))
 			adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
 
+		if (adev->pm.pp_feature & PP_GFXOFF_MASK)
+			adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_RLC_SMU_HS;
+		break;
+	case CHIP_RENOIR:
 		if (adev->pm.pp_feature & PP_GFXOFF_MASK)
 			adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG |
 				AMD_PG_SUPPORT_CP |

drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c

@@ -344,11 +344,9 @@ static uint64_t gmc_v10_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
 	amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 + (2 * vmid),
 			      upper_32_bits(pd_addr));
 
-	amdgpu_ring_emit_wreg(ring, hub->vm_inv_eng0_req + eng, req);
-
-	/* wait for the invalidate to complete */
-	amdgpu_ring_emit_reg_wait(ring, hub->vm_inv_eng0_ack + eng,
-				  1 << vmid, 1 << vmid);
+	amdgpu_ring_emit_reg_write_reg_wait(ring, hub->vm_inv_eng0_req + eng,
+					    hub->vm_inv_eng0_ack + eng,
+					    req, 1 << vmid);
 
 	return pd_addr;
 }

drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c

@@ -219,6 +219,15 @@ static void mmhub_v9_4_init_cache_regs(struct amdgpu_device *adev, int hubid)
 			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
 
 	tmp = mmVML2PF0_VM_L2_CNTL3_DEFAULT;
+	if (adev->gmc.translate_further) {
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL3, BANK_SELECT, 12);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	} else {
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL3, BANK_SELECT, 9);
+		tmp = REG_SET_FIELD(tmp, VML2PF0_VM_L2_CNTL3,
+				    L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
+	}
 	WREG32_SOC15_OFFSET(MMHUB, 0, mmVML2PF0_VM_L2_CNTL3,
 			    hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp);
 

drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c

@@ -1173,6 +1173,16 @@ static void sdma_v5_0_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
 			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10));
 }
 
+static void sdma_v5_0_ring_emit_reg_write_reg_wait(struct amdgpu_ring *ring,
+						   uint32_t reg0, uint32_t reg1,
+						   uint32_t ref, uint32_t mask)
+{
+	amdgpu_ring_emit_wreg(ring, reg0, ref);
+	/* wait for a cycle to reset vm_inv_eng*_ack */
+	amdgpu_ring_emit_reg_wait(ring, reg0, 0, 0);
+	amdgpu_ring_emit_reg_wait(ring, reg1, mask, mask);
+}
+
 static int sdma_v5_0_early_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -1588,7 +1598,7 @@ static const struct amdgpu_ring_funcs sdma_v5_0_ring_funcs = {
 		6 + /* sdma_v5_0_ring_emit_pipeline_sync */
 		/* sdma_v5_0_ring_emit_vm_flush */
 		SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
-		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 +
+		SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 6 * 2 +
 		10 + 10 + 10, /* sdma_v5_0_ring_emit_fence x3 for user fence, vm fence */
 	.emit_ib_size = 7 + 6, /* sdma_v5_0_ring_emit_ib */
 	.emit_ib = sdma_v5_0_ring_emit_ib,
@@ -1602,6 +1612,7 @@ static const struct amdgpu_ring_funcs sdma_v5_0_ring_funcs = {
 	.pad_ib = sdma_v5_0_ring_pad_ib,
 	.emit_wreg = sdma_v5_0_ring_emit_wreg,
 	.emit_reg_wait = sdma_v5_0_ring_emit_reg_wait,
+	.emit_reg_write_reg_wait = sdma_v5_0_ring_emit_reg_write_reg_wait,
 	.init_cond_exec = sdma_v5_0_ring_init_cond_exec,
 	.patch_cond_exec = sdma_v5_0_ring_patch_cond_exec,
 	.preempt_ib = sdma_v5_0_ring_preempt_ib,

drivers/gpu/drm/amd/amdgpu/soc15.c

@@ -1186,11 +1186,6 @@ static int soc15_common_early_init(void *handle)
 				 AMD_PG_SUPPORT_VCN |
 				 AMD_PG_SUPPORT_VCN_DPG;
 		adev->external_rev_id = adev->rev_id + 0x91;
-
-		if (adev->pm.pp_feature & PP_GFXOFF_MASK)
-			adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG |
-				AMD_PG_SUPPORT_CP |
-				AMD_PG_SUPPORT_RLC_SMU_HS;
 		break;
 	default:
 		/* FIXME: not supported yet */

drivers/gpu/drm/amd/display/dc/core/dc_link.c

@@ -2767,15 +2767,6 @@ void core_link_enable_stream(
 					CONTROLLER_DP_TEST_PATTERN_VIDEOMODE,
 					COLOR_DEPTH_UNDEFINED);
 
-		/* This second call is needed to reconfigure the DIG
-		 * as a workaround for the incorrect value being applied
-		 * from transmitter control.
-		 */
-		if (!dc_is_virtual_signal(pipe_ctx->stream->signal))
-			stream->link->link_enc->funcs->setup(
-				stream->link->link_enc,
-				pipe_ctx->stream->signal);
-
 #ifdef CONFIG_DRM_AMD_DC_DSC_SUPPORT
 		if (pipe_ctx->stream->timing.flags.DSC) {
 			if (dc_is_dp_signal(pipe_ctx->stream->signal) ||

drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c

@@ -1107,6 +1107,11 @@ struct stream_encoder *dcn20_stream_encoder_create(
 	if (!enc1)
 		return NULL;
 
+	if (ASICREV_IS_NAVI14_M(ctx->asic_id.hw_internal_rev)) {
+		if (eng_id >= ENGINE_ID_DIGD)
+			eng_id++;
+	}
+
 	dcn20_stream_encoder_construct(enc1, ctx, ctx->dc_bios, eng_id,
 					&stream_enc_regs[eng_id],
 					&se_shift, &se_mask);

drivers/gpu/drm/amd/powerplay/navi10_ppt.c

@@ -205,7 +205,7 @@ static struct smu_11_0_cmn2aisc_mapping navi10_workload_map[PP_SMC_POWER_PROFILE
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_POWERSAVING,		WORKLOAD_PPLIB_POWER_SAVING_BIT),
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VIDEO,		WORKLOAD_PPLIB_VIDEO_BIT),
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VR,			WORKLOAD_PPLIB_VR_BIT),
-	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_COMPUTE,		WORKLOAD_PPLIB_CUSTOM_BIT),
+	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_COMPUTE,		WORKLOAD_PPLIB_COMPUTE_BIT),
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_CUSTOM,		WORKLOAD_PPLIB_CUSTOM_BIT),
 };
 

drivers/gpu/drm/amd/powerplay/vega20_ppt.c

@@ -219,7 +219,7 @@ static struct smu_11_0_cmn2aisc_mapping vega20_workload_map[PP_SMC_POWER_PROFILE
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_POWERSAVING,		WORKLOAD_PPLIB_POWER_SAVING_BIT),
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VIDEO,		WORKLOAD_PPLIB_VIDEO_BIT),
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VR,			WORKLOAD_PPLIB_VR_BIT),
-	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_COMPUTE,		WORKLOAD_PPLIB_CUSTOM_BIT),
+	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_COMPUTE,		WORKLOAD_PPLIB_COMPUTE_BIT),
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_CUSTOM,		WORKLOAD_PPLIB_CUSTOM_BIT),
 };
 

drivers/gpu/drm/drm_atomic_helper.c

@@ -1581,8 +1581,11 @@ static void commit_tail(struct drm_atomic_state *old_state)
 {
 	struct drm_device *dev = old_state->dev;
 	const struct drm_mode_config_helper_funcs *funcs;
+	struct drm_crtc_state *new_crtc_state;
+	struct drm_crtc *crtc;
 	ktime_t start;
 	s64 commit_time_ms;
+	unsigned int i, new_self_refresh_mask = 0;
 
 	funcs = dev->mode_config.helper_private;
 
@@ -1602,6 +1605,15 @@ static void commit_tail(struct drm_atomic_state *old_state)
 
 	drm_atomic_helper_wait_for_dependencies(old_state);
 
+	/*
+	 * We cannot safely access new_crtc_state after
+	 * drm_atomic_helper_commit_hw_done() so figure out which crtc's have
+	 * self-refresh active beforehand:
+	 */
+	for_each_new_crtc_in_state(old_state, crtc, new_crtc_state, i)
+		if (new_crtc_state->self_refresh_active)
+			new_self_refresh_mask |= BIT(i);
+
 	if (funcs && funcs->atomic_commit_tail)
 		funcs->atomic_commit_tail(old_state);
 	else
@@ -1610,7 +1622,8 @@ static void commit_tail(struct drm_atomic_state *old_state)
 	commit_time_ms = ktime_ms_delta(ktime_get(), start);
 	if (commit_time_ms > 0)
 		drm_self_refresh_helper_update_avg_times(old_state,
-						 (unsigned long)commit_time_ms);
+						 (unsigned long)commit_time_ms,
+						 new_self_refresh_mask);
 
 	drm_atomic_helper_commit_cleanup_done(old_state);
 

drivers/gpu/drm/drm_self_refresh_helper.c

@@ -133,29 +133,33 @@ static void drm_self_refresh_helper_entry_work(struct work_struct *work)
  * drm_self_refresh_helper_update_avg_times - Updates a crtc's SR time averages
  * @state: the state which has just been applied to hardware
  * @commit_time_ms: the amount of time in ms that this commit took to complete
+ * @new_self_refresh_mask: bitmask of crtc's that have self_refresh_active in
+ *    new state
  *
  * Called after &drm_mode_config_funcs.atomic_commit_tail, this function will
  * update the average entry/exit self refresh times on self refresh transitions.
  * These averages will be used when calculating how long to delay before
  * entering self refresh mode after activity.
  */
-void drm_self_refresh_helper_update_avg_times(struct drm_atomic_state *state,
-					      unsigned int commit_time_ms)
+void
+drm_self_refresh_helper_update_avg_times(struct drm_atomic_state *state,
+					 unsigned int commit_time_ms,
+					 unsigned int new_self_refresh_mask)
 {
 	struct drm_crtc *crtc;
-	struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+	struct drm_crtc_state *old_crtc_state;
 	int i;
 
-	for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state,
-				      new_crtc_state, i) {
+	for_each_old_crtc_in_state(state, crtc, old_crtc_state, i) {
+		bool new_self_refresh_active = new_self_refresh_mask & BIT(i);
 		struct drm_self_refresh_data *sr_data = crtc->self_refresh_data;
 		struct ewma_psr_time *time;
 
 		if (old_crtc_state->self_refresh_active ==
-		    new_crtc_state->self_refresh_active)
+		    new_self_refresh_active)
 			continue;
 
-		if (new_crtc_state->self_refresh_active)
+		if (new_self_refresh_active)
 			time = &sr_data->entry_avg_ms;
 		else
 			time = &sr_data->exit_avg_ms;

drivers/gpu/drm/i915/display/intel_crt.c

@@ -864,6 +864,13 @@ intel_crt_detect(struct drm_connector *connector,
 
 out:
 	intel_display_power_put(dev_priv, intel_encoder->power_domain, wakeref);
+
+	/*
+	 * Make sure the refs for power wells enabled during detect are
+	 * dropped to avoid a new detect cycle triggered by HPD polling.
+	 */
+	intel_display_power_flush_work(dev_priv);
+
 	return status;
 }