Merge tag 'drm-fixes-2023-05-12' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "About the usual for this stage, bunch of amdgpu, a few i915 and a
  scattering of fixes across the board"

  dsc:
   - macro fixes

  simplefb:
   - fix VESA format

  scheduler:
   - timeout handling fix

  fbdev:
   - avoid potential out-of-bounds access in generic fbdev emulation

  ast:
   - improve AST2500+ compat on ARM

  mipi-dsi:
   - small mipi-dsi fix

  amdgpu:
   - VCN3 fixes
   - APUs always support PCI atomics
   - legacy power management fixes
   - DCN 3.1.4 fix
   - DCFCLK fix
   - fix several RAS irq refcount mismatches
   - GPU Reset fix
   - GFX 11.0.4 fix

  i915:
   - taint kernel when force_probe is used
   - NULL deref and div-by-zero fixes for display
   - GuC error capture fix for Xe devices"

* tag 'drm-fixes-2023-05-12' of git://anongit.freedesktop.org/drm/drm: (24 commits)
  drm/amdgpu: change gfx 11.0.4 external_id range
  drm/amdgpu/jpeg: Remove harvest checking for JPEG3
  drm/amdgpu/gfx: disable gfx9 cp_ecc_error_irq only when enabling legacy gfx ras
  drm/amd/pm: avoid potential UBSAN issue on legacy asics
  drm/i915: taint kernel when force probing unsupported devices
  drm/i915/dp: prevent potential div-by-zero
  drm/i915: Fix NULL ptr deref by checking new_crtc_state
  drm/i915/guc: Don't capture Gen8 regs on Xe devices
  drm/amdgpu: disable sdma ecc irq only when sdma RAS is enabled in suspend
  drm/amdgpu: Fix vram recover doesn't work after whole GPU reset (v2)
  drm/amdgpu: drop gfx_v11_0_cp_ecc_error_irq_funcs
  drm/amd/display: Enforce 60us prefetch for 200Mhz DCFCLK modes
  drm/amd/display: Add symclk workaround during disable link output
  drm/amd/pm: parse pp_handle under appropriate conditions
  drm/amdgpu: set gfx9 onwards APU atomics support to be true
  drm/amdgpu/nv: update VCN 3 max HEVC encoding resolution
  drm/sched: Check scheduler work queue before calling timeout handling
  drm/mipi-dsi: Set the fwnode for mipi_dsi_device
  drm/nouveau/disp: More DP_RECEIVER_CAP_SIZE array fixes
  drm/dsc: fix DP_DSC_MAX_BPP_DELTA_* macro values
  ...

Author: torvalds

drivers/firmware/sysfb_simplefb.c

@@ -51,7 +51,8 @@ __init bool sysfb_parse_mode(const struct screen_info *si,
 	 *
 	 * It's not easily possible to fix this in struct screen_info,
 	 * as this could break UAPI. The best solution is to compute
-	 * bits_per_pixel here and ignore lfb_depth. In the loop below,
+	 * bits_per_pixel from the color bits, reserved bits and
+	 * reported lfb_depth, whichever is highest.  In the loop below,
 	 * ignore simplefb formats with alpha bits, as EFI and VESA
 	 * don't specify alpha channels.
 	 */
@@ -60,6 +61,7 @@ __init bool sysfb_parse_mode(const struct screen_info *si,
 					  si->green_size + si->green_pos,
 					  si->blue_size + si->blue_pos),
 				     si->rsvd_size + si->rsvd_pos);
+		bits_per_pixel = max_t(u32, bits_per_pixel, si->lfb_depth);
 	} else {
 		bits_per_pixel = si->lfb_depth;
 	}

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

@@ -3757,6 +3757,12 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 		adev->have_atomics_support = ((struct amd_sriov_msg_pf2vf_info *)
 			adev->virt.fw_reserve.p_pf2vf)->pcie_atomic_ops_support_flags ==
 			(PCI_EXP_DEVCAP2_ATOMIC_COMP32 | PCI_EXP_DEVCAP2_ATOMIC_COMP64);
+	/* APUs w/ gfx9 onwards doesn't reply on PCIe atomics, rather it is a
+	 * internal path natively support atomics, set have_atomics_support to true.
+	 */
+	else if ((adev->flags & AMD_IS_APU) &&
+		(adev->ip_versions[GC_HWIP][0] > IP_VERSION(9, 0, 0)))
+		adev->have_atomics_support = true;
 	else
 		adev->have_atomics_support =
 			!pci_enable_atomic_ops_to_root(adev->pdev,
@@ -4506,7 +4512,11 @@ static int amdgpu_device_recover_vram(struct amdgpu_device *adev)
 	dev_info(adev->dev, "recover vram bo from shadow start\n");
 	mutex_lock(&adev->shadow_list_lock);
 	list_for_each_entry(vmbo, &adev->shadow_list, shadow_list) {
-		shadow = &vmbo->bo;
+		/* If vm is compute context or adev is APU, shadow will be NULL */
+		if (!vmbo->shadow)
+			continue;
+		shadow = vmbo->shadow;
+
 		/* No need to recover an evicted BO */
 		if (shadow->tbo.resource->mem_type != TTM_PL_TT ||
 		    shadow->tbo.resource->start == AMDGPU_BO_INVALID_OFFSET ||

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

@@ -687,9 +687,11 @@ int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *r
 		if (r)
 			return r;
 
-		r = amdgpu_irq_get(adev, &adev->gfx.cp_ecc_error_irq, 0);
-		if (r)
-			goto late_fini;
+		if (adev->gfx.cp_ecc_error_irq.funcs) {
+			r = amdgpu_irq_get(adev, &adev->gfx.cp_ecc_error_irq, 0);
+			if (r)
+				goto late_fini;
+		}
 	} else {
 		amdgpu_ras_feature_enable_on_boot(adev, ras_block, 0);
 	}

drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c

@@ -1315,13 +1315,6 @@ static int gfx_v11_0_sw_init(void *handle)
 	if (r)
 		return r;
 
-	/* ECC error */
-	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP,
-				  GFX_11_0_0__SRCID__CP_ECC_ERROR,
-				  &adev->gfx.cp_ecc_error_irq);
-	if (r)
-		return r;
-
 	/* FED error */
 	r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX,
 				  GFX_11_0_0__SRCID__RLC_GC_FED_INTERRUPT,
@@ -4444,7 +4437,6 @@ static int gfx_v11_0_hw_fini(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 	int r;
 
-	amdgpu_irq_put(adev, &adev->gfx.cp_ecc_error_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
 
@@ -5897,36 +5889,6 @@ static void gfx_v11_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev
 	}
 }
 
-#define CP_ME1_PIPE_INST_ADDR_INTERVAL  0x1
-#define SET_ECC_ME_PIPE_STATE(reg_addr, state) \
-	do { \
-		uint32_t tmp = RREG32_SOC15_IP(GC, reg_addr); \
-		tmp = REG_SET_FIELD(tmp, CP_ME1_PIPE0_INT_CNTL, CP_ECC_ERROR_INT_ENABLE, state); \
-		WREG32_SOC15_IP(GC, reg_addr, tmp); \
-	} while (0)
-
-static int gfx_v11_0_set_cp_ecc_error_state(struct amdgpu_device *adev,
-							struct amdgpu_irq_src *source,
-							unsigned type,
-							enum amdgpu_interrupt_state state)
-{
-	uint32_t ecc_irq_state = 0;
-	uint32_t pipe0_int_cntl_addr = 0;
-	int i = 0;
-
-	ecc_irq_state = (state == AMDGPU_IRQ_STATE_ENABLE) ? 1 : 0;
-
-	pipe0_int_cntl_addr = SOC15_REG_OFFSET(GC, 0, regCP_ME1_PIPE0_INT_CNTL);
-
-	WREG32_FIELD15_PREREG(GC, 0, CP_INT_CNTL_RING0, CP_ECC_ERROR_INT_ENABLE, ecc_irq_state);
-
-	for (i = 0; i < adev->gfx.mec.num_pipe_per_mec; i++)
-		SET_ECC_ME_PIPE_STATE(pipe0_int_cntl_addr + i * CP_ME1_PIPE_INST_ADDR_INTERVAL,
-					ecc_irq_state);
-
-	return 0;
-}
-
 static int gfx_v11_0_set_eop_interrupt_state(struct amdgpu_device *adev,
 					    struct amdgpu_irq_src *src,
 					    unsigned type,
@@ -6341,11 +6303,6 @@ static const struct amdgpu_irq_src_funcs gfx_v11_0_priv_inst_irq_funcs = {
 	.process = gfx_v11_0_priv_inst_irq,
 };
 
-static const struct amdgpu_irq_src_funcs gfx_v11_0_cp_ecc_error_irq_funcs = {
-	.set = gfx_v11_0_set_cp_ecc_error_state,
-	.process = amdgpu_gfx_cp_ecc_error_irq,
-};
-
 static const struct amdgpu_irq_src_funcs gfx_v11_0_rlc_gc_fed_irq_funcs = {
 	.process = gfx_v11_0_rlc_gc_fed_irq,
 };
@@ -6361,9 +6318,6 @@ static void gfx_v11_0_set_irq_funcs(struct amdgpu_device *adev)
 	adev->gfx.priv_inst_irq.num_types = 1;
 	adev->gfx.priv_inst_irq.funcs = &gfx_v11_0_priv_inst_irq_funcs;
 
-	adev->gfx.cp_ecc_error_irq.num_types = 1; /* CP ECC error */
-	adev->gfx.cp_ecc_error_irq.funcs = &gfx_v11_0_cp_ecc_error_irq_funcs;
-
 	adev->gfx.rlc_gc_fed_irq.num_types = 1; /* 0x80 FED error */
 	adev->gfx.rlc_gc_fed_irq.funcs = &gfx_v11_0_rlc_gc_fed_irq_funcs;
 

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

@@ -3764,7 +3764,8 @@ static int gfx_v9_0_hw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	amdgpu_irq_put(adev, &adev->gfx.cp_ecc_error_irq, 0);
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
+		amdgpu_irq_put(adev, &adev->gfx.cp_ecc_error_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
 

drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c

@@ -54,6 +54,7 @@ static int jpeg_v3_0_early_init(void *handle)
 
 	switch (adev->ip_versions[UVD_HWIP][0]) {
 	case IP_VERSION(3, 1, 1):
+	case IP_VERSION(3, 1, 2):
 		break;
 	default:
 		harvest = RREG32_SOC15(JPEG, 0, mmCC_UVD_HARVESTING);

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

@@ -98,6 +98,16 @@ static const struct amdgpu_video_codecs nv_video_codecs_decode =
 };
 
 /* Sienna Cichlid */
+static const struct amdgpu_video_codec_info sc_video_codecs_encode_array[] = {
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2160, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 7680, 4352, 0)},
+};
+
+static const struct amdgpu_video_codecs sc_video_codecs_encode = {
+	.codec_count = ARRAY_SIZE(sc_video_codecs_encode_array),
+	.codec_array = sc_video_codecs_encode_array,
+};
+
 static const struct amdgpu_video_codec_info sc_video_codecs_decode_array_vcn0[] =
 {
 	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 4096, 4096, 3)},
@@ -136,8 +146,8 @@ static const struct amdgpu_video_codecs sc_video_codecs_decode_vcn1 =
 /* SRIOV Sienna Cichlid, not const since data is controlled by host */
 static struct amdgpu_video_codec_info sriov_sc_video_codecs_encode_array[] =
 {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 2304, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2160, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 7680, 4352, 0)},
 };
 
 static struct amdgpu_video_codec_info sriov_sc_video_codecs_decode_array_vcn0[] =
@@ -237,12 +247,12 @@ static int nv_query_video_codecs(struct amdgpu_device *adev, bool encode,
 		} else {
 			if (adev->vcn.harvest_config & AMDGPU_VCN_HARVEST_VCN0) {
 				if (encode)
-					*codecs = &nv_video_codecs_encode;
+					*codecs = &sc_video_codecs_encode;
 				else
 					*codecs = &sc_video_codecs_decode_vcn1;
 			} else {
 				if (encode)
-					*codecs = &nv_video_codecs_encode;
+					*codecs = &sc_video_codecs_encode;
 				else
 					*codecs = &sc_video_codecs_decode_vcn0;
 			}
@@ -251,14 +261,14 @@ static int nv_query_video_codecs(struct amdgpu_device *adev, bool encode,
 	case IP_VERSION(3, 0, 16):
 	case IP_VERSION(3, 0, 2):
 		if (encode)
-			*codecs = &nv_video_codecs_encode;
+			*codecs = &sc_video_codecs_encode;
 		else
 			*codecs = &sc_video_codecs_decode_vcn0;
 		return 0;
 	case IP_VERSION(3, 1, 1):
 	case IP_VERSION(3, 1, 2):
 		if (encode)
-			*codecs = &nv_video_codecs_encode;
+			*codecs = &sc_video_codecs_encode;
 		else
 			*codecs = &yc_video_codecs_decode;
 		return 0;

drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c

@@ -1917,9 +1917,11 @@ static int sdma_v4_0_hw_fini(void *handle)
 		return 0;
 	}
 
-	for (i = 0; i < adev->sdma.num_instances; i++) {
-		amdgpu_irq_put(adev, &adev->sdma.ecc_irq,
-			       AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			amdgpu_irq_put(adev, &adev->sdma.ecc_irq,
+				       AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+		}
 	}
 
 	sdma_v4_0_ctx_switch_enable(adev, false);

drivers/gpu/drm/amd/amdgpu/soc21.c

@@ -711,7 +711,7 @@ static int soc21_common_early_init(void *handle)
 			AMD_PG_SUPPORT_VCN_DPG |
 			AMD_PG_SUPPORT_GFX_PG |
 			AMD_PG_SUPPORT_JPEG;
-		adev->external_rev_id = adev->rev_id + 0x1;
+		adev->external_rev_id = adev->rev_id + 0x80;
 		break;
 
 	default:

drivers/gpu/drm/amd/display/dc/dcn314/dcn314_hwseq.c

@@ -423,3 +423,68 @@ void dcn314_hubp_pg_control(struct dce_hwseq *hws, unsigned int hubp_inst, bool
 
 	PERF_TRACE();
 }
+static void apply_symclk_on_tx_off_wa(struct dc_link *link)
+{
+	/* There are use cases where SYMCLK is referenced by OTG. For instance
+	 * for TMDS signal, OTG relies SYMCLK even if TX video output is off.
+	 * However current link interface will power off PHY when disabling link
+	 * output. This will turn off SYMCLK generated by PHY. The workaround is
+	 * to identify such case where SYMCLK is still in use by OTG when we
+	 * power off PHY. When this is detected, we will temporarily power PHY
+	 * back on and move PHY's SYMCLK state to SYMCLK_ON_TX_OFF by calling
+	 * program_pix_clk interface. When OTG is disabled, we will then power
+	 * off PHY by calling disable link output again.
+	 *
+	 * In future dcn generations, we plan to rework transmitter control
+	 * interface so that we could have an option to set SYMCLK ON TX OFF
+	 * state in one step without this workaround
+	 */
+
+	struct dc *dc = link->ctx->dc;
+	struct pipe_ctx *pipe_ctx = NULL;
+	uint8_t i;
+
+	if (link->phy_state.symclk_ref_cnts.otg > 0) {
+		for (i = 0; i < MAX_PIPES; i++) {
+			pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i];
+			if (pipe_ctx->stream && pipe_ctx->stream->link == link && pipe_ctx->top_pipe == NULL) {
+				pipe_ctx->clock_source->funcs->program_pix_clk(
+						pipe_ctx->clock_source,
+						&pipe_ctx->stream_res.pix_clk_params,
+						dc->link_srv->dp_get_encoding_format(
+								&pipe_ctx->link_config.dp_link_settings),
+						&pipe_ctx->pll_settings);
+				link->phy_state.symclk_state = SYMCLK_ON_TX_OFF;
+				break;
+			}
+		}
+	}
+}
+
+void dcn314_disable_link_output(struct dc_link *link,
+		const struct link_resource *link_res,
+		enum signal_type signal)
+{
+	struct dc *dc = link->ctx->dc;
+	const struct link_hwss *link_hwss = get_link_hwss(link, link_res);
+	struct dmcu *dmcu = dc->res_pool->dmcu;
+
+	if (signal == SIGNAL_TYPE_EDP &&
+			link->dc->hwss.edp_backlight_control)
+		link->dc->hwss.edp_backlight_control(link, false);
+	else if (dmcu != NULL && dmcu->funcs->lock_phy)
+		dmcu->funcs->lock_phy(dmcu);
+
+	link_hwss->disable_link_output(link, link_res, signal);
+	link->phy_state.symclk_state = SYMCLK_OFF_TX_OFF;
+	/*
+	 * Add the logic to extract BOTH power up and power down sequences
+	 * from enable/disable link output and only call edp panel control
+	 * in enable_link_dp and disable_link_dp once.
+	 */
+	if (dmcu != NULL && dmcu->funcs->lock_phy)
+		dmcu->funcs->unlock_phy(dmcu);
+	dc->link_srv->dp_trace_source_sequence(link, DPCD_SOURCE_SEQ_AFTER_DISABLE_LINK_PHY);
+
+	apply_symclk_on_tx_off_wa(link);
+}