Merge branch 'drm-next-3.10' of git://people.freedesktop.org/~agd5f/linux into drm-next

Alex writes:
This is the initial 3.10 pull request for radeon.  The big changes here
are UVD support and proper tiling support for SI.  The rest is
bug fixes.  I hope to have another pull request later in the week with
some new things we've been working on internally.

* 'drm-next-3.10' of git://people.freedesktop.org/~agd5f/linux: (28 commits)
  drm/radeon: Always flush the VM
  drm/radeon: re-enable PTE/PDE packet for set_page on cayman/TN
  drm/radeon: cleanup properly if mmio mapping fails
  drm/radeon/evergreen+: don't enable HPD interrupts on eDP/LVDS
  drm/radeon: add si tile mode array query v3
  drm/radeon: add ring working query
  drm/radeon: handle broken disabled rb mask gracefully
  drm/radeon: add pcie set/get lanes callbacks for newer asics
  drm/radeon: update r600 set/get pcie lane config
  drm/radeon/kms: replace *REG32_PCIE_P with *REG32_PCIE_PORT
  drm/radeon: remove unused blit remnants from si.c
  drm/radeon: add UVD tiling addr config v2
  drm/radeon: init UVD clocks to sane defaults
  drm/radeon: add set_uvd_clocks callback for r7xx v3
  drm/radeon: add set_uvd_clocks callback for SI
  drm/radeon: add set_uvd_clocks callback for evergreen
  drm/radeon: add set_uvd_clocks callback for ON/LN/TN (v4)
  drm/radeon: add radeon_atom_get_clock_dividers helper
  drm/radeon: add pm callback for setting uvd clocks
  drm/radeon: UVD bringup v8
  ...

drivers/gpu/drm/radeon/Makefile

@@ -76,7 +76,7 @@ radeon-y += radeon_device.o radeon_asic.o radeon_kms.o \
 	evergreen.o evergreen_cs.o evergreen_blit_shaders.o evergreen_blit_kms.o \
 	evergreen_hdmi.o radeon_trace_points.o ni.o cayman_blit_shaders.o \
 	atombios_encoders.o radeon_semaphore.o radeon_sa.o atombios_i2c.o si.o \
-	si_blit_shaders.o radeon_prime.o
+	si_blit_shaders.o radeon_prime.o radeon_uvd.o
 
 radeon-$(CONFIG_COMPAT) += radeon_ioc32.o
 radeon-$(CONFIG_VGA_SWITCHEROO) += radeon_atpx_handler.o

drivers/gpu/drm/radeon/atombios_crtc.c

@@ -557,6 +557,9 @@ static u32 atombios_adjust_pll(struct drm_crtc *crtc,
 		/* use frac fb div on APUs */
 		if (ASIC_IS_DCE41(rdev) || ASIC_IS_DCE61(rdev))
 			radeon_crtc->pll_flags |= RADEON_PLL_USE_FRAC_FB_DIV;
+		/* use frac fb div on RS780/RS880 */
+		if ((rdev->family == CHIP_RS780) || (rdev->family == CHIP_RS880))
+			radeon_crtc->pll_flags |= RADEON_PLL_USE_FRAC_FB_DIV;
 		if (ASIC_IS_DCE32(rdev) && mode->clock > 165000)
 			radeon_crtc->pll_flags |= RADEON_PLL_USE_FRAC_FB_DIV;
 	} else {

drivers/gpu/drm/radeon/evergreen.c

@@ -84,6 +84,217 @@ void evergreen_tiling_fields(unsigned tiling_flags, unsigned *bankw,
 	}
 }
 
+static int sumo_set_uvd_clock(struct radeon_device *rdev, u32 clock,
+			      u32 cntl_reg, u32 status_reg)
+{
+	int r, i;
+	struct atom_clock_dividers dividers;
+
+        r = radeon_atom_get_clock_dividers(rdev, COMPUTE_ENGINE_PLL_PARAM,
+					   clock, false, &dividers);
+	if (r)
+		return r;
+
+	WREG32_P(cntl_reg, dividers.post_div, ~(DCLK_DIR_CNTL_EN|DCLK_DIVIDER_MASK));
+
+	for (i = 0; i < 100; i++) {
+		if (RREG32(status_reg) & DCLK_STATUS)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+int sumo_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk)
+{
+	int r = 0;
+	u32 cg_scratch = RREG32(CG_SCRATCH1);
+
+	r = sumo_set_uvd_clock(rdev, vclk, CG_VCLK_CNTL, CG_VCLK_STATUS);
+	if (r)
+		goto done;
+	cg_scratch &= 0xffff0000;
+	cg_scratch |= vclk / 100; /* Mhz */
+
+	r = sumo_set_uvd_clock(rdev, dclk, CG_DCLK_CNTL, CG_DCLK_STATUS);
+	if (r)
+		goto done;
+	cg_scratch &= 0x0000ffff;
+	cg_scratch |= (dclk / 100) << 16; /* Mhz */
+
+done:
+	WREG32(CG_SCRATCH1, cg_scratch);
+
+	return r;
+}
+
+static int evergreen_uvd_calc_post_div(unsigned target_freq,
+				       unsigned vco_freq,
+				       unsigned *div)
+{
+	/* target larger than vco frequency ? */
+	if (vco_freq < target_freq)
+		return -1; /* forget it */
+
+	/* Fclk = Fvco / PDIV */
+	*div = vco_freq / target_freq;
+
+	/* we alway need a frequency less than or equal the target */
+	if ((vco_freq / *div) > target_freq)
+		*div += 1;
+
+	/* dividers above 5 must be even */
+	if (*div > 5 && *div % 2)
+		*div += 1;
+
+	/* out of range ? */
+	if (*div >= 128)
+		return -1; /* forget it */
+
+	return vco_freq / *div;
+}
+
+static int evergreen_uvd_send_upll_ctlreq(struct radeon_device *rdev)
+{
+	unsigned i;
+
+	/* assert UPLL_CTLREQ */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_CTLREQ_MASK, ~UPLL_CTLREQ_MASK);
+
+	/* wait for CTLACK and CTLACK2 to get asserted */
+	for (i = 0; i < 100; ++i) {
+		uint32_t mask = UPLL_CTLACK_MASK | UPLL_CTLACK2_MASK;
+		if ((RREG32(CG_UPLL_FUNC_CNTL) & mask) == mask)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	/* deassert UPLL_CTLREQ */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_CTLREQ_MASK);
+
+	return 0;
+}
+
+int evergreen_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk)
+{
+	/* start off with something large */
+	int optimal_diff_score = 0x7FFFFFF;
+	unsigned optimal_fb_div = 0, optimal_vclk_div = 0;
+	unsigned optimal_dclk_div = 0, optimal_vco_freq = 0;
+	unsigned vco_freq;
+	int r;
+
+	/* loop through vco from low to high */
+	for (vco_freq = 125000; vco_freq <= 250000; vco_freq += 100) {
+		unsigned fb_div = vco_freq / rdev->clock.spll.reference_freq * 16384;
+		int calc_clk, diff_score, diff_vclk, diff_dclk;
+		unsigned vclk_div, dclk_div;
+
+		/* fb div out of range ? */
+		if (fb_div > 0x03FFFFFF)
+			break; /* it can oly get worse */
+
+		/* calc vclk with current vco freq. */
+		calc_clk = evergreen_uvd_calc_post_div(vclk, vco_freq, &vclk_div);
+		if (calc_clk == -1)
+			break; /* vco is too big, it has to stop. */
+		diff_vclk = vclk - calc_clk;
+
+		/* calc dclk with current vco freq. */
+		calc_clk = evergreen_uvd_calc_post_div(dclk, vco_freq, &dclk_div);
+		if (calc_clk == -1)
+			break; /* vco is too big, it has to stop. */
+		diff_dclk = dclk - calc_clk;
+
+		/* determine if this vco setting is better than current optimal settings */
+		diff_score = abs(diff_vclk) + abs(diff_dclk);
+		if (diff_score < optimal_diff_score) {
+			optimal_fb_div = fb_div;
+			optimal_vclk_div = vclk_div;
+			optimal_dclk_div = dclk_div;
+			optimal_vco_freq = vco_freq;
+			optimal_diff_score = diff_score;
+			if (optimal_diff_score == 0)
+				break; /* it can't get better than this */
+		}
+	}
+
+	/* set VCO_MODE to 1 */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_VCO_MODE_MASK, ~UPLL_VCO_MODE_MASK);
+
+	/* toggle UPLL_SLEEP to 1 then back to 0 */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_SLEEP_MASK, ~UPLL_SLEEP_MASK);
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_SLEEP_MASK);
+
+	/* deassert UPLL_RESET */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_RESET_MASK);
+
+	mdelay(1);
+
+	/* bypass vclk and dclk with bclk */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		VCLK_SRC_SEL(1) | DCLK_SRC_SEL(1),
+		~(VCLK_SRC_SEL_MASK | DCLK_SRC_SEL_MASK));
+
+	/* put PLL in bypass mode */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_BYPASS_EN_MASK, ~UPLL_BYPASS_EN_MASK);
+
+	r = evergreen_uvd_send_upll_ctlreq(rdev);
+	if (r)
+		return r;
+
+	/* assert UPLL_RESET again */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_RESET_MASK, ~UPLL_RESET_MASK);
+
+	/* disable spread spectrum. */
+	WREG32_P(CG_UPLL_SPREAD_SPECTRUM, 0, ~SSEN_MASK);
+
+	/* set feedback divider */
+	WREG32_P(CG_UPLL_FUNC_CNTL_3, UPLL_FB_DIV(optimal_fb_div), ~UPLL_FB_DIV_MASK);
+
+	/* set ref divider to 0 */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_REF_DIV_MASK);
+
+	if (optimal_vco_freq < 187500)
+		WREG32_P(CG_UPLL_FUNC_CNTL_4, 0, ~UPLL_SPARE_ISPARE9);
+	else
+		WREG32_P(CG_UPLL_FUNC_CNTL_4, UPLL_SPARE_ISPARE9, ~UPLL_SPARE_ISPARE9);
+
+	/* set PDIV_A and PDIV_B */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		UPLL_PDIV_A(optimal_vclk_div) | UPLL_PDIV_B(optimal_dclk_div),
+		~(UPLL_PDIV_A_MASK | UPLL_PDIV_B_MASK));
+
+	/* give the PLL some time to settle */
+	mdelay(15);
+
+	/* deassert PLL_RESET */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_RESET_MASK);
+
+	mdelay(15);
+
+	/* switch from bypass mode to normal mode */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_BYPASS_EN_MASK);
+
+	r = evergreen_uvd_send_upll_ctlreq(rdev);
+	if (r)
+		return r;
+
+	/* switch VCLK and DCLK selection */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		VCLK_SRC_SEL(2) | DCLK_SRC_SEL(2),
+		~(VCLK_SRC_SEL_MASK | DCLK_SRC_SEL_MASK));
+
+	mdelay(100);
+
+	return 0;
+}
+
 void evergreen_fix_pci_max_read_req_size(struct radeon_device *rdev)
 {
 	u16 ctl, v;
@@ -608,6 +819,16 @@ void evergreen_hpd_init(struct radeon_device *rdev)
 
 	list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
 		struct radeon_connector *radeon_connector = to_radeon_connector(connector);
+
+		if (connector->connector_type == DRM_MODE_CONNECTOR_eDP ||
+		    connector->connector_type == DRM_MODE_CONNECTOR_LVDS) {
+			/* don't try to enable hpd on eDP or LVDS avoid breaking the
+			 * aux dp channel on imac and help (but not completely fix)
+			 * https://bugzilla.redhat.com/show_bug.cgi?id=726143
+			 * also avoid interrupt storms during dpms.
+			 */
+			continue;
+		}
 		switch (radeon_connector->hpd.hpd) {
 		case RADEON_HPD_1:
 			WREG32(DC_HPD1_CONTROL, tmp);
@@ -2050,6 +2271,14 @@ static void evergreen_gpu_init(struct radeon_device *rdev)
 	}
 	/* enabled rb are just the one not disabled :) */
 	disabled_rb_mask = tmp;
+	tmp = 0;
+	for (i = 0; i < rdev->config.evergreen.max_backends; i++)
+		tmp |= (1 << i);
+	/* if all the backends are disabled, fix it up here */
+	if ((disabled_rb_mask & tmp) == tmp) {
+		for (i = 0; i < rdev->config.evergreen.max_backends; i++)
+			disabled_rb_mask &= ~(1 << i);
+	}
 
 	WREG32(GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES | SE_BROADCAST_WRITES);
 	WREG32(RLC_GFX_INDEX, INSTANCE_BROADCAST_WRITES | SE_BROADCAST_WRITES);
@@ -2058,6 +2287,9 @@ static void evergreen_gpu_init(struct radeon_device *rdev)
 	WREG32(DMIF_ADDR_CONFIG, gb_addr_config);
 	WREG32(HDP_ADDR_CONFIG, gb_addr_config);
 	WREG32(DMA_TILING_CONFIG, gb_addr_config);
+	WREG32(UVD_UDEC_ADDR_CONFIG, gb_addr_config);
+	WREG32(UVD_UDEC_DB_ADDR_CONFIG, gb_addr_config);
+	WREG32(UVD_UDEC_DBW_ADDR_CONFIG, gb_addr_config);
 
 	if ((rdev->config.evergreen.max_backends == 1) &&
 	    (rdev->flags & RADEON_IS_IGP)) {
@@ -3360,6 +3592,9 @@ restart_ih:
 				DRM_ERROR("Unhandled interrupt: %d %d\n", src_id, src_data);
 				break;
 			}
+		case 124: /* UVD */
+			DRM_DEBUG("IH: UVD int: 0x%08x\n", src_data);
+			radeon_fence_process(rdev, R600_RING_TYPE_UVD_INDEX);
 			break;
 		case 146:
 		case 147:
@@ -3571,7 +3806,7 @@ int evergreen_copy_dma(struct radeon_device *rdev,
 
 static int evergreen_startup(struct radeon_device *rdev)
 {
-	struct radeon_ring *ring = &rdev->ring[RADEON_RING_TYPE_GFX_INDEX];
+	struct radeon_ring *ring;
 	int r;
 
 	/* enable pcie gen2 link */
@@ -3638,6 +3873,17 @@ static int evergreen_startup(struct radeon_device *rdev)
 		return r;
 	}
 
+	r = rv770_uvd_resume(rdev);
+	if (!r) {
+		r = radeon_fence_driver_start_ring(rdev,
+						   R600_RING_TYPE_UVD_INDEX);
+		if (r)
+			dev_err(rdev->dev, "UVD fences init error (%d).\n", r);
+	}
+
+	if (r)
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_size = 0;
+
 	/* Enable IRQ */
 	r = r600_irq_init(rdev);
 	if (r) {
@@ -3647,6 +3893,7 @@ static int evergreen_startup(struct radeon_device *rdev)
 	}
 	evergreen_irq_set(rdev);
 
+	ring = &rdev->ring[RADEON_RING_TYPE_GFX_INDEX];
 	r = radeon_ring_init(rdev, ring, ring->ring_size, RADEON_WB_CP_RPTR_OFFSET,
 			     R600_CP_RB_RPTR, R600_CP_RB_WPTR,
 			     0, 0xfffff, RADEON_CP_PACKET2);
@@ -3670,6 +3917,19 @@ static int evergreen_startup(struct radeon_device *rdev)
 	if (r)
 		return r;
 
+	ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	if (ring->ring_size) {
+		r = radeon_ring_init(rdev, ring, ring->ring_size,
+				     R600_WB_UVD_RPTR_OFFSET,
+				     UVD_RBC_RB_RPTR, UVD_RBC_RB_WPTR,
+				     0, 0xfffff, RADEON_CP_PACKET2);
+		if (!r)
+			r = r600_uvd_init(rdev);
+
+		if (r)
+			DRM_ERROR("radeon: error initializing UVD (%d).\n", r);
+	}
+
 	r = radeon_ib_pool_init(rdev);
 	if (r) {
 		dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
@@ -3716,8 +3976,10 @@ int evergreen_resume(struct radeon_device *rdev)
 int evergreen_suspend(struct radeon_device *rdev)
 {
 	r600_audio_fini(rdev);
+	radeon_uvd_suspend(rdev);
 	r700_cp_stop(rdev);
 	r600_dma_stop(rdev);
+	r600_uvd_rbc_stop(rdev);
 	evergreen_irq_suspend(rdev);
 	radeon_wb_disable(rdev);
 	evergreen_pcie_gart_disable(rdev);
@@ -3797,6 +4059,13 @@ int evergreen_init(struct radeon_device *rdev)
 	rdev->ring[R600_RING_TYPE_DMA_INDEX].ring_obj = NULL;
 	r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX], 64 * 1024);
 
+	r = radeon_uvd_init(rdev);
+	if (!r) {
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_obj = NULL;
+		r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_UVD_INDEX],
+			       4096);
+	}
+
 	rdev->ih.ring_obj = NULL;
 	r600_ih_ring_init(rdev, 64 * 1024);
 
@@ -3843,6 +4112,7 @@ void evergreen_fini(struct radeon_device *rdev)
 	radeon_ib_pool_fini(rdev);
 	radeon_irq_kms_fini(rdev);
 	evergreen_pcie_gart_fini(rdev);
+	radeon_uvd_fini(rdev);
 	r600_vram_scratch_fini(rdev);
 	radeon_gem_fini(rdev);
 	radeon_fence_driver_fini(rdev);
@@ -3878,7 +4148,7 @@ void evergreen_pcie_gen2_enable(struct radeon_device *rdev)
 	if (!(mask & DRM_PCIE_SPEED_50))
 		return;
 
-	speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+	speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 	if (speed_cntl & LC_CURRENT_DATA_RATE) {
 		DRM_INFO("PCIE gen 2 link speeds already enabled\n");
 		return;
@@ -3889,33 +4159,33 @@ void evergreen_pcie_gen2_enable(struct radeon_device *rdev)
 	if ((speed_cntl & LC_OTHER_SIDE_EVER_SENT_GEN2) ||
 	    (speed_cntl & LC_OTHER_SIDE_SUPPORTS_GEN2)) {
 
-		link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
+		link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
 		link_width_cntl &= ~LC_UPCONFIGURE_DIS;
-		WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
 
-		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 		speed_cntl &= ~LC_TARGET_LINK_SPEED_OVERRIDE_EN;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
-		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 		speed_cntl |= LC_CLR_FAILED_SPD_CHANGE_CNT;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
-		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 		speed_cntl &= ~LC_CLR_FAILED_SPD_CHANGE_CNT;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
-		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 		speed_cntl |= LC_GEN2_EN_STRAP;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
 	} else {
-		link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
+		link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
 		/* XXX: only disable it if gen1 bridge vendor == 0x111d or 0x1106 */
 		if (1)
 			link_width_cntl |= LC_UPCONFIGURE_DIS;
 		else
 			link_width_cntl &= ~LC_UPCONFIGURE_DIS;
-		WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
 	}
 }

drivers/gpu/drm/radeon/evergreend.h

@@ -53,6 +53,43 @@
 #define RCU_IND_INDEX           			0x100
 #define RCU_IND_DATA            			0x104
 
+/* discrete uvd clocks */
+#define CG_UPLL_FUNC_CNTL				0x718
+#	define UPLL_RESET_MASK				0x00000001
+#	define UPLL_SLEEP_MASK				0x00000002
+#	define UPLL_BYPASS_EN_MASK			0x00000004
+#	define UPLL_CTLREQ_MASK				0x00000008
+#	define UPLL_REF_DIV_MASK			0x001F0000
+#	define UPLL_VCO_MODE_MASK			0x00000200
+#	define UPLL_CTLACK_MASK				0x40000000
+#	define UPLL_CTLACK2_MASK			0x80000000
+#define CG_UPLL_FUNC_CNTL_2				0x71c
+#	define UPLL_PDIV_A(x)				((x) << 0)
+#	define UPLL_PDIV_A_MASK				0x0000007F
+#	define UPLL_PDIV_B(x)				((x) << 8)
+#	define UPLL_PDIV_B_MASK				0x00007F00
+#	define VCLK_SRC_SEL(x)				((x) << 20)
+#	define VCLK_SRC_SEL_MASK			0x01F00000
+#	define DCLK_SRC_SEL(x)				((x) << 25)
+#	define DCLK_SRC_SEL_MASK			0x3E000000
+#define CG_UPLL_FUNC_CNTL_3				0x720
+#	define UPLL_FB_DIV(x)				((x) << 0)
+#	define UPLL_FB_DIV_MASK				0x01FFFFFF
+#define CG_UPLL_FUNC_CNTL_4				0x854
+#	define UPLL_SPARE_ISPARE9			0x00020000
+#define CG_UPLL_SPREAD_SPECTRUM				0x79c
+#	define SSEN_MASK				0x00000001
+
+/* fusion uvd clocks */
+#define CG_DCLK_CNTL                                    0x610
+#       define DCLK_DIVIDER_MASK                        0x7f
+#       define DCLK_DIR_CNTL_EN                         (1 << 8)
+#define CG_DCLK_STATUS                                  0x614
+#       define DCLK_STATUS                              (1 << 0)
+#define CG_VCLK_CNTL                                    0x618
+#define CG_VCLK_STATUS                                  0x61c
+#define	CG_SCRATCH1					0x820
+
 #define GRBM_GFX_INDEX          			0x802C
 #define		INSTANCE_INDEX(x)			((x) << 0)
 #define		SE_INDEX(x)     			((x) << 16)
@@ -992,6 +1029,16 @@
 #       define TARGET_LINK_SPEED_MASK                     (0xf << 0)
 #       define SELECTABLE_DEEMPHASIS                      (1 << 6)
 
+
+/*
+ * UVD
+ */
+#define UVD_UDEC_ADDR_CONFIG				0xef4c
+#define UVD_UDEC_DB_ADDR_CONFIG				0xef50
+#define UVD_UDEC_DBW_ADDR_CONFIG			0xef54
+#define UVD_RBC_RB_RPTR					0xf690
+#define UVD_RBC_RB_WPTR					0xf694
+
 /*
  * PM4
  */

drivers/gpu/drm/radeon/ni.c

@@ -615,15 +615,28 @@ static void cayman_gpu_init(struct radeon_device *rdev)
 	}
 	/* enabled rb are just the one not disabled :) */
 	disabled_rb_mask = tmp;
+	tmp = 0;
+	for (i = 0; i < (rdev->config.cayman.max_backends_per_se * rdev->config.cayman.max_shader_engines); i++)
+		tmp |= (1 << i);
+	/* if all the backends are disabled, fix it up here */
+	if ((disabled_rb_mask & tmp) == tmp) {
+		for (i = 0; i < (rdev->config.cayman.max_backends_per_se * rdev->config.cayman.max_shader_engines); i++)
+			disabled_rb_mask &= ~(1 << i);
+	}
 
 	WREG32(GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES | SE_BROADCAST_WRITES);
 	WREG32(RLC_GFX_INDEX, INSTANCE_BROADCAST_WRITES | SE_BROADCAST_WRITES);
 
 	WREG32(GB_ADDR_CONFIG, gb_addr_config);
 	WREG32(DMIF_ADDR_CONFIG, gb_addr_config);
+	if (ASIC_IS_DCE6(rdev))
+		WREG32(DMIF_ADDR_CALC, gb_addr_config);
 	WREG32(HDP_ADDR_CONFIG, gb_addr_config);
 	WREG32(DMA_TILING_CONFIG + DMA0_REGISTER_OFFSET, gb_addr_config);
 	WREG32(DMA_TILING_CONFIG + DMA1_REGISTER_OFFSET, gb_addr_config);
+	WREG32(UVD_UDEC_ADDR_CONFIG, gb_addr_config);
+	WREG32(UVD_UDEC_DB_ADDR_CONFIG, gb_addr_config);
+	WREG32(UVD_UDEC_DBW_ADDR_CONFIG, gb_addr_config);
 
 	if ((rdev->config.cayman.max_backends_per_se == 1) &&
 	    (rdev->flags & RADEON_IS_IGP)) {
@@ -931,6 +944,23 @@ void cayman_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
 	radeon_ring_write(ring, 10); /* poll interval */
 }
 
+void cayman_uvd_semaphore_emit(struct radeon_device *rdev,
+			       struct radeon_ring *ring,
+			       struct radeon_semaphore *semaphore,
+			       bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_ADDR_LOW, 0));
+	radeon_ring_write(ring, (addr >> 3) & 0x000FFFFF);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_ADDR_HIGH, 0));
+	radeon_ring_write(ring, (addr >> 23) & 0x000FFFFF);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_CMD, 0));
+	radeon_ring_write(ring, 0x80 | (emit_wait ? 1 : 0));
+}
+
 static void cayman_cp_enable(struct radeon_device *rdev, bool enable)
 {
 	if (enable)
@@ -1682,6 +1712,16 @@ static int cayman_startup(struct radeon_device *rdev)
 		return r;
 	}
 
+	r = rv770_uvd_resume(rdev);
+	if (!r) {
+		r = radeon_fence_driver_start_ring(rdev,
+						   R600_RING_TYPE_UVD_INDEX);
+		if (r)
+			dev_err(rdev->dev, "UVD fences init error (%d).\n", r);
+	}
+	if (r)
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_size = 0;
+
 	r = radeon_fence_driver_start_ring(rdev, CAYMAN_RING_TYPE_CP1_INDEX);
 	if (r) {
 		dev_err(rdev->dev, "failed initializing CP fences (%d).\n", r);
@@ -1748,6 +1788,18 @@ static int cayman_startup(struct radeon_device *rdev)
 	if (r)
 		return r;
 
+	ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	if (ring->ring_size) {
+		r = radeon_ring_init(rdev, ring, ring->ring_size,
+				     R600_WB_UVD_RPTR_OFFSET,
+				     UVD_RBC_RB_RPTR, UVD_RBC_RB_WPTR,
+				     0, 0xfffff, RADEON_CP_PACKET2);
+		if (!r)
+			r = r600_uvd_init(rdev);
+		if (r)
+			DRM_ERROR("radeon: failed initializing UVD (%d).\n", r);
+	}
+
 	r = radeon_ib_pool_init(rdev);
 	if (r) {
 		dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
@@ -1794,6 +1846,8 @@ int cayman_suspend(struct radeon_device *rdev)
 	radeon_vm_manager_fini(rdev);
 	cayman_cp_enable(rdev, false);
 	cayman_dma_stop(rdev);
+	r600_uvd_rbc_stop(rdev);
+	radeon_uvd_suspend(rdev);
 	evergreen_irq_suspend(rdev);
 	radeon_wb_disable(rdev);
 	cayman_pcie_gart_disable(rdev);
@@ -1868,6 +1922,13 @@ int cayman_init(struct radeon_device *rdev)
 	ring->ring_obj = NULL;
 	r600_ring_init(rdev, ring, 64 * 1024);
 
+	r = radeon_uvd_init(rdev);
+	if (!r) {
+		ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+		ring->ring_obj = NULL;
+		r600_ring_init(rdev, ring, 4096);
+	}
+
 	rdev->ih.ring_obj = NULL;
 	r600_ih_ring_init(rdev, 64 * 1024);
 
@@ -1919,6 +1980,7 @@ void cayman_fini(struct radeon_device *rdev)
 	radeon_vm_manager_fini(rdev);
 	radeon_ib_pool_fini(rdev);
 	radeon_irq_kms_fini(rdev);
+	radeon_uvd_fini(rdev);
 	cayman_pcie_gart_fini(rdev);
 	r600_vram_scratch_fini(rdev);
 	radeon_gem_fini(rdev);
@@ -2017,28 +2079,57 @@ void cayman_vm_set_page(struct radeon_device *rdev,
 			}
 		}
 	} else {
-		while (count) {
-			ndw = count * 2;
-			if (ndw > 0xFFFFE)
-				ndw = 0xFFFFE;
+		if ((flags & RADEON_VM_PAGE_SYSTEM) ||
+		    (count == 1)) {
+			while (count) {
+				ndw = count * 2;
+				if (ndw > 0xFFFFE)
+					ndw = 0xFFFFE;
+
+				/* for non-physically contiguous pages (system) */
+				ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, ndw);
+				ib->ptr[ib->length_dw++] = pe;
+				ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
+				for (; ndw > 0; ndw -= 2, --count, pe += 8) {
+					if (flags & RADEON_VM_PAGE_SYSTEM) {
+						value = radeon_vm_map_gart(rdev, addr);
+						value &= 0xFFFFFFFFFFFFF000ULL;
+					} else if (flags & RADEON_VM_PAGE_VALID) {
+						value = addr;
+					} else {
+						value = 0;
+					}
+					addr += incr;
+					value |= r600_flags;
+					ib->ptr[ib->length_dw++] = value;
+					ib->ptr[ib->length_dw++] = upper_32_bits(value);
+				}
+			}
+			while (ib->length_dw & 0x7)
+				ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0);
+		} else {
+			while (count) {
+				ndw = count * 2;
+				if (ndw > 0xFFFFE)
+					ndw = 0xFFFFE;
 
-			/* for non-physically contiguous pages (system) */
-			ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, ndw);
-			ib->ptr[ib->length_dw++] = pe;
-			ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
-			for (; ndw > 0; ndw -= 2, --count, pe += 8) {
-				if (flags & RADEON_VM_PAGE_SYSTEM) {
-					value = radeon_vm_map_gart(rdev, addr);
-					value &= 0xFFFFFFFFFFFFF000ULL;
-				} else if (flags & RADEON_VM_PAGE_VALID) {
+				if (flags & RADEON_VM_PAGE_VALID)
 					value = addr;
-				} else {
+				else
 					value = 0;
-				}
-				addr += incr;
-				value |= r600_flags;
-				ib->ptr[ib->length_dw++] = value;
+				/* for physically contiguous pages (vram) */
+				ib->ptr[ib->length_dw++] = DMA_PTE_PDE_PACKET(ndw);
+				ib->ptr[ib->length_dw++] = pe; /* dst addr */
+				ib->ptr[ib->length_dw++] = upper_32_bits(pe) & 0xff;
+				ib->ptr[ib->length_dw++] = r600_flags; /* mask */
+				ib->ptr[ib->length_dw++] = 0;
+				ib->ptr[ib->length_dw++] = value; /* value */
 				ib->ptr[ib->length_dw++] = upper_32_bits(value);
+				ib->ptr[ib->length_dw++] = incr; /* increment size */
+				ib->ptr[ib->length_dw++] = 0;
+				pe += ndw * 4;
+				addr += (ndw / 2) * incr;
+				count -= ndw / 2;
 			}
 		}
 		while (ib->length_dw & 0x7)

drivers/gpu/drm/radeon/nid.h

@@ -45,6 +45,10 @@
 #define ARUBA_GB_ADDR_CONFIG_GOLDEN        0x12010001
 
 #define DMIF_ADDR_CONFIG  				0xBD4
+
+/* DCE6 only */
+#define DMIF_ADDR_CALC  				0xC00
+
 #define	SRBM_GFX_CNTL				        0x0E44
 #define		RINGID(x)					(((x) & 0x3) << 0)
 #define		VMID(x)						(((x) & 0x7) << 0)
@@ -485,6 +489,18 @@
 #       define CACHE_FLUSH_AND_INV_EVENT_TS                     (0x14 << 0)
 #       define CACHE_FLUSH_AND_INV_EVENT                        (0x16 << 0)
 
+/*
+ * UVD
+ */
+#define UVD_SEMA_ADDR_LOW				0xEF00
+#define UVD_SEMA_ADDR_HIGH				0xEF04
+#define UVD_SEMA_CMD					0xEF08
+#define UVD_UDEC_ADDR_CONFIG				0xEF4C
+#define UVD_UDEC_DB_ADDR_CONFIG				0xEF50
+#define UVD_UDEC_DBW_ADDR_CONFIG			0xEF54
+#define UVD_RBC_RB_RPTR					0xF690
+#define UVD_RBC_RB_WPTR					0xF694
+
 /*
  * PM4
  */
@@ -668,6 +684,11 @@
 					 (((vmid) & 0xF) << 20) |	\
 					 (((n) & 0xFFFFF) << 0))
 
+#define DMA_PTE_PDE_PACKET(n)		((2 << 28) |			\
+					 (1 << 26) |			\
+					 (1 << 21) |			\
+					 (((n) & 0xFFFFF) << 0))
+
 /* async DMA Packet types */
 #define	DMA_PACKET_WRITE				  0x2
 #define	DMA_PACKET_COPY					  0x3

drivers/gpu/drm/radeon/r600.c

@@ -1145,7 +1145,7 @@ static void r600_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc
 	}
 	if (rdev->flags & RADEON_IS_AGP) {
 		size_bf = mc->gtt_start;
-		size_af = 0xFFFFFFFF - mc->gtt_end;
+		size_af = mc->mc_mask - mc->gtt_end;
 		if (size_bf > size_af) {
 			if (mc->mc_vram_size > size_bf) {
 				dev_warn(rdev->dev, "limiting VRAM\n");
@@ -2551,6 +2551,185 @@ void r600_dma_fini(struct radeon_device *rdev)
 	radeon_ring_fini(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX]);
 }
 
+/*
+ * UVD
+ */
+int r600_uvd_rbc_start(struct radeon_device *rdev)
+{
+	struct radeon_ring *ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	uint64_t rptr_addr;
+	uint32_t rb_bufsz, tmp;
+	int r;
+
+	rptr_addr = rdev->wb.gpu_addr + R600_WB_UVD_RPTR_OFFSET;
+
+	if (upper_32_bits(rptr_addr) != upper_32_bits(ring->gpu_addr)) {
+		DRM_ERROR("UVD ring and rptr not in the same 4GB segment!\n");
+		return -EINVAL;
+	}
+
+	/* force RBC into idle state */
+	WREG32(UVD_RBC_RB_CNTL, 0x11010101);
+
+	/* Set the write pointer delay */
+	WREG32(UVD_RBC_RB_WPTR_CNTL, 0);
+
+	/* set the wb address */
+	WREG32(UVD_RBC_RB_RPTR_ADDR, rptr_addr >> 2);
+
+	/* programm the 4GB memory segment for rptr and ring buffer */
+	WREG32(UVD_LMI_EXT40_ADDR, upper_32_bits(rptr_addr) |
+				   (0x7 << 16) | (0x1 << 31));
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(UVD_RBC_RB_RPTR, 0x0);
+
+	ring->wptr = ring->rptr = RREG32(UVD_RBC_RB_RPTR);
+	WREG32(UVD_RBC_RB_WPTR, ring->wptr);
+
+	/* set the ring address */
+	WREG32(UVD_RBC_RB_BASE, ring->gpu_addr);
+
+	/* Set ring buffer size */
+	rb_bufsz = drm_order(ring->ring_size);
+	rb_bufsz = (0x1 << 8) | rb_bufsz;
+	WREG32(UVD_RBC_RB_CNTL, rb_bufsz);
+
+	ring->ready = true;
+	r = radeon_ring_test(rdev, R600_RING_TYPE_UVD_INDEX, ring);
+	if (r) {
+		ring->ready = false;
+		return r;
+	}
+
+	r = radeon_ring_lock(rdev, ring, 10);
+	if (r) {
+		DRM_ERROR("radeon: ring failed to lock UVD ring (%d).\n", r);
+		return r;
+	}
+
+	tmp = PACKET0(UVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL, 0);
+	radeon_ring_write(ring, tmp);
+	radeon_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(UVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL, 0);
+	radeon_ring_write(ring, tmp);
+	radeon_ring_write(ring, 0xFFFFF);
+
+	tmp = PACKET0(UVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL, 0);
+	radeon_ring_write(ring, tmp);
+	radeon_ring_write(ring, 0xFFFFF);
+
+	/* Clear timeout status bits */
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_TIMEOUT_STATUS, 0));
+	radeon_ring_write(ring, 0x8);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_CNTL, 0));
+	radeon_ring_write(ring, 1);
+
+	radeon_ring_unlock_commit(rdev, ring);
+
+	return 0;
+}
+
+void r600_uvd_rbc_stop(struct radeon_device *rdev)
+{
+	struct radeon_ring *ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+
+	/* force RBC into idle state */
+	WREG32(UVD_RBC_RB_CNTL, 0x11010101);
+	ring->ready = false;
+}
+
+int r600_uvd_init(struct radeon_device *rdev)
+{
+	int i, j, r;
+
+	/* disable clock gating */
+	WREG32(UVD_CGC_GATE, 0);
+
+	/* disable interupt */
+	WREG32_P(UVD_MASTINT_EN, 0, ~(1 << 1));
+
+	/* put LMI, VCPU, RBC etc... into reset */
+	WREG32(UVD_SOFT_RESET, LMI_SOFT_RESET | VCPU_SOFT_RESET |
+	       LBSI_SOFT_RESET | RBC_SOFT_RESET | CSM_SOFT_RESET |
+	       CXW_SOFT_RESET | TAP_SOFT_RESET | LMI_UMC_SOFT_RESET);
+	mdelay(5);
+
+	/* take UVD block out of reset */
+	WREG32_P(SRBM_SOFT_RESET, 0, ~SOFT_RESET_UVD);
+	mdelay(5);
+
+	/* initialize UVD memory controller */
+	WREG32(UVD_LMI_CTRL, 0x40 | (1 << 8) | (1 << 13) |
+			     (1 << 21) | (1 << 9) | (1 << 20));
+
+	/* disable byte swapping */
+	WREG32(UVD_LMI_SWAP_CNTL, 0);
+	WREG32(UVD_MP_SWAP_CNTL, 0);
+
+	WREG32(UVD_MPC_SET_MUXA0, 0x40c2040);
+	WREG32(UVD_MPC_SET_MUXA1, 0x0);
+	WREG32(UVD_MPC_SET_MUXB0, 0x40c2040);
+	WREG32(UVD_MPC_SET_MUXB1, 0x0);
+	WREG32(UVD_MPC_SET_ALU, 0);
+	WREG32(UVD_MPC_SET_MUX, 0x88);
+
+	/* Stall UMC */
+	WREG32_P(UVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	WREG32_P(UVD_RB_ARB_CTRL, 1 << 3, ~(1 << 3));
+
+	/* take all subblocks out of reset, except VCPU */
+	WREG32(UVD_SOFT_RESET, VCPU_SOFT_RESET);
+	mdelay(5);
+
+	/* enable VCPU clock */
+	WREG32(UVD_VCPU_CNTL,  1 << 9);
+
+	/* enable UMC */
+	WREG32_P(UVD_LMI_CTRL2, 0, ~(1 << 8));
+
+	/* boot up the VCPU */
+	WREG32(UVD_SOFT_RESET, 0);
+	mdelay(10);
+
+	WREG32_P(UVD_RB_ARB_CTRL, 0, ~(1 << 3));
+
+	for (i = 0; i < 10; ++i) {
+		uint32_t status;
+		for (j = 0; j < 100; ++j) {
+			status = RREG32(UVD_STATUS);
+			if (status & 2)
+				break;
+			mdelay(10);
+		}
+		r = 0;
+		if (status & 2)
+			break;
+
+		DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n");
+		WREG32_P(UVD_SOFT_RESET, VCPU_SOFT_RESET, ~VCPU_SOFT_RESET);
+		mdelay(10);
+		WREG32_P(UVD_SOFT_RESET, 0, ~VCPU_SOFT_RESET);
+		mdelay(10);
+		r = -1;
+	}
+	if (r) {
+		DRM_ERROR("UVD not responding, giving up!!!\n");
+		return r;
+	}
+	/* enable interupt */
+	WREG32_P(UVD_MASTINT_EN, 3<<1, ~(3 << 1));
+
+	r = r600_uvd_rbc_start(rdev);
+	if (r)
+		return r;
+
+	DRM_INFO("UVD initialized successfully.\n");
+	return 0;
+}
+
 /*
  * GPU scratch registers helpers function.
  */
@@ -2660,6 +2839,40 @@ int r600_dma_ring_test(struct radeon_device *rdev,
 	return r;
 }
 
+int r600_uvd_ring_test(struct radeon_device *rdev, struct radeon_ring *ring)
+{
+	uint32_t tmp = 0;
+	unsigned i;
+	int r;
+
+	WREG32(UVD_CONTEXT_ID, 0xCAFEDEAD);
+	r = radeon_ring_lock(rdev, ring, 3);
+	if (r) {
+		DRM_ERROR("radeon: cp failed to lock ring %d (%d).\n",
+			  ring->idx, r);
+		return r;
+	}
+	radeon_ring_write(ring, PACKET0(UVD_CONTEXT_ID, 0));
+	radeon_ring_write(ring, 0xDEADBEEF);
+	radeon_ring_unlock_commit(rdev, ring);
+	for (i = 0; i < rdev->usec_timeout; i++) {
+		tmp = RREG32(UVD_CONTEXT_ID);
+		if (tmp == 0xDEADBEEF)
+			break;
+		DRM_UDELAY(1);
+	}
+
+	if (i < rdev->usec_timeout) {
+		DRM_INFO("ring test on %d succeeded in %d usecs\n",
+			 ring->idx, i);
+	} else {
+		DRM_ERROR("radeon: ring %d test failed (0x%08X)\n",
+			  ring->idx, tmp);
+		r = -EINVAL;
+	}
+	return r;
+}
+
 /*
  * CP fences/semaphores
  */
@@ -2711,6 +2924,30 @@ void r600_fence_ring_emit(struct radeon_device *rdev,
 	}
 }
 
+void r600_uvd_fence_emit(struct radeon_device *rdev,
+			 struct radeon_fence *fence)
+{
+	struct radeon_ring *ring = &rdev->ring[fence->ring];
+	uint32_t addr = rdev->fence_drv[fence->ring].gpu_addr;
+
+	radeon_ring_write(ring, PACKET0(UVD_CONTEXT_ID, 0));
+	radeon_ring_write(ring, fence->seq);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_DATA0, 0));
+	radeon_ring_write(ring, addr & 0xffffffff);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_DATA1, 0));
+	radeon_ring_write(ring, upper_32_bits(addr) & 0xff);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_CMD, 0));
+	radeon_ring_write(ring, 0);
+
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_DATA0, 0));
+	radeon_ring_write(ring, 0);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_DATA1, 0));
+	radeon_ring_write(ring, 0);
+	radeon_ring_write(ring, PACKET0(UVD_GPCOM_VCPU_CMD, 0));
+	radeon_ring_write(ring, 2);
+	return;
+}
+
 void r600_semaphore_ring_emit(struct radeon_device *rdev,
 			      struct radeon_ring *ring,
 			      struct radeon_semaphore *semaphore,
@@ -2780,6 +3017,23 @@ void r600_dma_semaphore_ring_emit(struct radeon_device *rdev,
 	radeon_ring_write(ring, upper_32_bits(addr) & 0xff);
 }
 
+void r600_uvd_semaphore_emit(struct radeon_device *rdev,
+			     struct radeon_ring *ring,
+			     struct radeon_semaphore *semaphore,
+			     bool emit_wait)
+{
+	uint64_t addr = semaphore->gpu_addr;
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_ADDR_LOW, 0));
+	radeon_ring_write(ring, (addr >> 3) & 0x000FFFFF);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_ADDR_HIGH, 0));
+	radeon_ring_write(ring, (addr >> 23) & 0x000FFFFF);
+
+	radeon_ring_write(ring, PACKET0(UVD_SEMA_CMD, 0));
+	radeon_ring_write(ring, emit_wait ? 1 : 0);
+}
+
 int r600_copy_blit(struct radeon_device *rdev,
 		   uint64_t src_offset,
 		   uint64_t dst_offset,
@@ -3183,6 +3437,16 @@ void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
 	radeon_ring_write(ring, ib->length_dw);
 }
 
+void r600_uvd_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
+{
+	struct radeon_ring *ring = &rdev->ring[ib->ring];
+
+	radeon_ring_write(ring, PACKET0(UVD_RBC_IB_BASE, 0));
+	radeon_ring_write(ring, ib->gpu_addr);
+	radeon_ring_write(ring, PACKET0(UVD_RBC_IB_SIZE, 0));
+	radeon_ring_write(ring, ib->length_dw);
+}
+
 int r600_ib_test(struct radeon_device *rdev, struct radeon_ring *ring)
 {
 	struct radeon_ib ib;
@@ -3300,6 +3564,33 @@ int r600_dma_ib_test(struct radeon_device *rdev, struct radeon_ring *ring)
 	return r;
 }
 
+int r600_uvd_ib_test(struct radeon_device *rdev, struct radeon_ring *ring)
+{
+	struct radeon_fence *fence;
+	int r;
+
+	r = radeon_uvd_get_create_msg(rdev, ring->idx, 1, NULL);
+	if (r) {
+		DRM_ERROR("radeon: failed to get create msg (%d).\n", r);
+		return r;
+	}
+
+	r = radeon_uvd_get_destroy_msg(rdev, ring->idx, 1, &fence);
+	if (r) {
+		DRM_ERROR("radeon: failed to get destroy ib (%d).\n", r);
+		return r;
+	}
+
+	r = radeon_fence_wait(fence, false);
+	if (r) {
+		DRM_ERROR("radeon: fence wait failed (%d).\n", r);
+		return r;
+	}
+	DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
+	radeon_fence_unref(&fence);
+	return r;
+}
+
 /**
  * r600_dma_ring_ib_execute - Schedule an IB on the DMA engine
  *
@@ -4232,7 +4523,7 @@ void r600_ioctl_wait_idle(struct radeon_device *rdev, struct radeon_bo *bo)
 
 void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes)
 {
-	u32 link_width_cntl, mask, target_reg;
+	u32 link_width_cntl, mask;
 
 	if (rdev->flags & RADEON_IS_IGP)
 		return;
@@ -4244,7 +4535,7 @@ void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes)
 	if (ASIC_IS_X2(rdev))
 		return;
 
-	/* FIXME wait for idle */
+	radeon_gui_idle(rdev);
 
 	switch (lanes) {
 	case 0:
@@ -4263,53 +4554,24 @@ void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes)
 		mask = RADEON_PCIE_LC_LINK_WIDTH_X8;
 		break;
 	case 12:
+		/* not actually supported */
 		mask = RADEON_PCIE_LC_LINK_WIDTH_X12;
 		break;
 	case 16:
-	default:
 		mask = RADEON_PCIE_LC_LINK_WIDTH_X16;
 		break;
-	}
-
-	link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL);
-
-	if ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) ==
-	    (mask << RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT))
-		return;
-
-	if (link_width_cntl & R600_PCIE_LC_UPCONFIGURE_DIS)
+	default:
+		DRM_ERROR("invalid pcie lane request: %d\n", lanes);
 		return;
+	}
 
-	link_width_cntl &= ~(RADEON_PCIE_LC_LINK_WIDTH_MASK |
-			     RADEON_PCIE_LC_RECONFIG_NOW |
-			     R600_PCIE_LC_RENEGOTIATE_EN |
-			     R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE);
-	link_width_cntl |= mask;
-
-	WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
-
-        /* some northbridges can renegotiate the link rather than requiring                                  
-         * a complete re-config.                                                                             
-         * e.g., AMD 780/790 northbridges (pci ids: 0x5956, 0x5957, 0x5958, etc.)                            
-         */
-        if (link_width_cntl & R600_PCIE_LC_RENEGOTIATION_SUPPORT)
-		link_width_cntl |= R600_PCIE_LC_RENEGOTIATE_EN | R600_PCIE_LC_UPCONFIGURE_SUPPORT;
-        else
-		link_width_cntl |= R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE;
-
-	WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, (link_width_cntl |
-						       RADEON_PCIE_LC_RECONFIG_NOW));
-
-        if (rdev->family >= CHIP_RV770)
-		target_reg = R700_TARGET_AND_CURRENT_PROFILE_INDEX;
-        else
-		target_reg = R600_TARGET_AND_CURRENT_PROFILE_INDEX;
-
-        /* wait for lane set to complete */
-        link_width_cntl = RREG32(target_reg);
-        while (link_width_cntl == 0xffffffff)
-		link_width_cntl = RREG32(target_reg);
+	link_width_cntl = RREG32_PCIE_PORT(RADEON_PCIE_LC_LINK_WIDTH_CNTL);
+	link_width_cntl &= ~RADEON_PCIE_LC_LINK_WIDTH_MASK;
+	link_width_cntl |= mask << RADEON_PCIE_LC_LINK_WIDTH_SHIFT;
+	link_width_cntl |= (RADEON_PCIE_LC_RECONFIG_NOW |
+			    R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE);
 
+	WREG32_PCIE_PORT(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
 }
 
 int r600_get_pcie_lanes(struct radeon_device *rdev)
@@ -4326,13 +4588,11 @@ int r600_get_pcie_lanes(struct radeon_device *rdev)
 	if (ASIC_IS_X2(rdev))
 		return 0;
 
-	/* FIXME wait for idle */
+	radeon_gui_idle(rdev);
 
-	link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL);
+	link_width_cntl = RREG32_PCIE_PORT(RADEON_PCIE_LC_LINK_WIDTH_CNTL);
 
 	switch ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) >> RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT) {
-	case RADEON_PCIE_LC_LINK_WIDTH_X0:
-		return 0;
 	case RADEON_PCIE_LC_LINK_WIDTH_X1:
 		return 1;
 	case RADEON_PCIE_LC_LINK_WIDTH_X2:
@@ -4341,6 +4601,10 @@ int r600_get_pcie_lanes(struct radeon_device *rdev)
 		return 4;
 	case RADEON_PCIE_LC_LINK_WIDTH_X8:
 		return 8;
+	case RADEON_PCIE_LC_LINK_WIDTH_X12:
+		/* not actually supported */
+		return 12;
+	case RADEON_PCIE_LC_LINK_WIDTH_X0:
 	case RADEON_PCIE_LC_LINK_WIDTH_X16:
 	default:
 		return 16;
@@ -4378,7 +4642,7 @@ static void r600_pcie_gen2_enable(struct radeon_device *rdev)
 	if (!(mask & DRM_PCIE_SPEED_50))
 		return;
 
-	speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+	speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 	if (speed_cntl & LC_CURRENT_DATA_RATE) {
 		DRM_INFO("PCIE gen 2 link speeds already enabled\n");
 		return;
@@ -4391,23 +4655,23 @@ static void r600_pcie_gen2_enable(struct radeon_device *rdev)
 	    (rdev->family == CHIP_RV620) ||
 	    (rdev->family == CHIP_RV635)) {
 		/* advertise upconfig capability */
-		link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
+		link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
 		link_width_cntl &= ~LC_UPCONFIGURE_DIS;
-		WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
-		link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
+		WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+		link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
 		if (link_width_cntl & LC_RENEGOTIATION_SUPPORT) {
 			lanes = (link_width_cntl & LC_LINK_WIDTH_RD_MASK) >> LC_LINK_WIDTH_RD_SHIFT;
 			link_width_cntl &= ~(LC_LINK_WIDTH_MASK |
 					     LC_RECONFIG_ARC_MISSING_ESCAPE);
 			link_width_cntl |= lanes | LC_RECONFIG_NOW | LC_RENEGOTIATE_EN;
-			WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+			WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
 		} else {
 			link_width_cntl |= LC_UPCONFIGURE_DIS;
-			WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+			WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
 		}
 	}
 
-	speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+	speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 	if ((speed_cntl & LC_OTHER_SIDE_EVER_SENT_GEN2) &&
 	    (speed_cntl & LC_OTHER_SIDE_SUPPORTS_GEN2)) {
 
@@ -4428,7 +4692,7 @@ static void r600_pcie_gen2_enable(struct radeon_device *rdev)
 		speed_cntl &= ~LC_VOLTAGE_TIMER_SEL_MASK;
 		speed_cntl &= ~LC_FORCE_DIS_HW_SPEED_CHANGE;
 		speed_cntl |= LC_FORCE_EN_HW_SPEED_CHANGE;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
 		tmp = RREG32(0x541c);
 		WREG32(0x541c, tmp | 0x8);
@@ -4442,27 +4706,27 @@ static void r600_pcie_gen2_enable(struct radeon_device *rdev)
 		if ((rdev->family == CHIP_RV670) ||
 		    (rdev->family == CHIP_RV620) ||
 		    (rdev->family == CHIP_RV635)) {
-			training_cntl = RREG32_PCIE_P(PCIE_LC_TRAINING_CNTL);
+			training_cntl = RREG32_PCIE_PORT(PCIE_LC_TRAINING_CNTL);
 			training_cntl &= ~LC_POINT_7_PLUS_EN;
-			WREG32_PCIE_P(PCIE_LC_TRAINING_CNTL, training_cntl);
+			WREG32_PCIE_PORT(PCIE_LC_TRAINING_CNTL, training_cntl);
 		} else {
-			speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+			speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 			speed_cntl &= ~LC_TARGET_LINK_SPEED_OVERRIDE_EN;
-			WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+			WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 		}
 
-		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 		speed_cntl |= LC_GEN2_EN_STRAP;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
 	} else {
-		link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
+		link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
 		/* XXX: only disable it if gen1 bridge vendor == 0x111d or 0x1106 */
 		if (1)
 			link_width_cntl |= LC_UPCONFIGURE_DIS;
 		else
 			link_width_cntl &= ~LC_UPCONFIGURE_DIS;
-		WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
 	}
 }
 

drivers/gpu/drm/radeon/r600d.h

@@ -691,6 +691,7 @@
 #define SRBM_SOFT_RESET                                   0xe60
 #       define SOFT_RESET_DMA                             (1 << 12)
 #       define SOFT_RESET_RLC                             (1 << 13)
+#       define SOFT_RESET_UVD                             (1 << 18)
 #       define RV770_SOFT_RESET_DMA                       (1 << 20)
 
 #define CP_INT_CNTL                                       0xc124
@@ -1142,6 +1143,66 @@
 #       define AFMT_AZ_FORMAT_WTRIG_ACK      (1 << 29)
 #       define AFMT_AZ_AUDIO_ENABLE_CHG_ACK  (1 << 30)
 
+/*
+ * UVD
+ */
+#define UVD_SEMA_ADDR_LOW				0xef00
+#define UVD_SEMA_ADDR_HIGH				0xef04
+#define UVD_SEMA_CMD					0xef08
+
+#define UVD_GPCOM_VCPU_CMD				0xef0c
+#define UVD_GPCOM_VCPU_DATA0				0xef10
+#define UVD_GPCOM_VCPU_DATA1				0xef14
+#define UVD_ENGINE_CNTL					0xef18
+
+#define UVD_SEMA_CNTL					0xf400
+#define UVD_RB_ARB_CTRL					0xf480
+
+#define UVD_LMI_EXT40_ADDR				0xf498
+#define UVD_CGC_GATE					0xf4a8
+#define UVD_LMI_CTRL2					0xf4f4
+#define UVD_MASTINT_EN					0xf500
+#define UVD_LMI_ADDR_EXT				0xf594
+#define UVD_LMI_CTRL					0xf598
+#define UVD_LMI_SWAP_CNTL				0xf5b4
+#define UVD_MP_SWAP_CNTL				0xf5bC
+#define UVD_MPC_CNTL					0xf5dC
+#define UVD_MPC_SET_MUXA0				0xf5e4
+#define UVD_MPC_SET_MUXA1				0xf5e8
+#define UVD_MPC_SET_MUXB0				0xf5eC
+#define UVD_MPC_SET_MUXB1				0xf5f0
+#define UVD_MPC_SET_MUX					0xf5f4
+#define UVD_MPC_SET_ALU					0xf5f8
+
+#define UVD_VCPU_CNTL					0xf660
+#define UVD_SOFT_RESET					0xf680
+#define		RBC_SOFT_RESET					(1<<0)
+#define		LBSI_SOFT_RESET					(1<<1)
+#define		LMI_SOFT_RESET					(1<<2)
+#define		VCPU_SOFT_RESET					(1<<3)
+#define		CSM_SOFT_RESET					(1<<5)
+#define		CXW_SOFT_RESET					(1<<6)
+#define		TAP_SOFT_RESET					(1<<7)
+#define		LMI_UMC_SOFT_RESET				(1<<13)
+#define UVD_RBC_IB_BASE					0xf684
+#define UVD_RBC_IB_SIZE					0xf688
+#define UVD_RBC_RB_BASE					0xf68c
+#define UVD_RBC_RB_RPTR					0xf690
+#define UVD_RBC_RB_WPTR					0xf694
+#define UVD_RBC_RB_WPTR_CNTL				0xf698
+
+#define UVD_STATUS					0xf6bc
+
+#define UVD_SEMA_TIMEOUT_STATUS				0xf6c0
+#define UVD_SEMA_WAIT_INCOMPLETE_TIMEOUT_CNTL		0xf6c4
+#define UVD_SEMA_WAIT_FAULT_TIMEOUT_CNTL		0xf6c8
+#define UVD_SEMA_SIGNAL_INCOMPLETE_TIMEOUT_CNTL		0xf6cc
+
+#define UVD_RBC_RB_CNTL					0xf6a4
+#define UVD_RBC_RB_RPTR_ADDR				0xf6a8
+
+#define UVD_CONTEXT_ID					0xf6f4
+
 /*
  * PM4
  */

drivers/gpu/drm/radeon/radeon.h

@@ -95,6 +95,7 @@ extern int radeon_hw_i2c;
 extern int radeon_pcie_gen2;
 extern int radeon_msi;
 extern int radeon_lockup_timeout;
+extern int radeon_fastfb;
 
 /*
  * Copy from radeon_drv.h so we don't have to include both and have conflicting
@@ -109,24 +110,27 @@ extern int radeon_lockup_timeout;
 #define RADEON_BIOS_NUM_SCRATCH			8
 
 /* max number of rings */
-#define RADEON_NUM_RINGS			5
+#define RADEON_NUM_RINGS			6
 
 /* fence seq are set to this number when signaled */
 #define RADEON_FENCE_SIGNALED_SEQ		0LL
 
 /* internal ring indices */
 /* r1xx+ has gfx CP ring */
-#define RADEON_RING_TYPE_GFX_INDEX		0
+#define RADEON_RING_TYPE_GFX_INDEX	0
 
 /* cayman has 2 compute CP rings */
-#define CAYMAN_RING_TYPE_CP1_INDEX		1
-#define CAYMAN_RING_TYPE_CP2_INDEX		2
+#define CAYMAN_RING_TYPE_CP1_INDEX	1
+#define CAYMAN_RING_TYPE_CP2_INDEX	2
 
 /* R600+ has an async dma ring */
 #define R600_RING_TYPE_DMA_INDEX		3
 /* cayman add a second async dma ring */
 #define CAYMAN_RING_TYPE_DMA1_INDEX		4
 
+/* R600+ */
+#define R600_RING_TYPE_UVD_INDEX	5
+
 /* hardcode those limit for now */
 #define RADEON_VA_IB_OFFSET			(1 << 20)
 #define RADEON_VA_RESERVED_SIZE			(8 << 20)
@@ -202,6 +206,11 @@ void radeon_pm_suspend(struct radeon_device *rdev);
 void radeon_pm_resume(struct radeon_device *rdev);
 void radeon_combios_get_power_modes(struct radeon_device *rdev);
 void radeon_atombios_get_power_modes(struct radeon_device *rdev);
+int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
+				   u8 clock_type,
+				   u32 clock,
+				   bool strobe_mode,
+				   struct atom_clock_dividers *dividers);
 void radeon_atom_set_voltage(struct radeon_device *rdev, u16 voltage_level, u8 voltage_type);
 void rs690_pm_info(struct radeon_device *rdev);
 extern int rv6xx_get_temp(struct radeon_device *rdev);
@@ -357,8 +366,9 @@ struct radeon_bo_list {
 	struct ttm_validate_buffer tv;
 	struct radeon_bo	*bo;
 	uint64_t		gpu_offset;
-	unsigned		rdomain;
-	unsigned		wdomain;
+	bool			written;
+	unsigned		domain;
+	unsigned		alt_domain;
 	u32			tiling_flags;
 };
 
@@ -517,6 +527,7 @@ struct radeon_mc {
 	bool			vram_is_ddr;
 	bool			igp_sideport_enabled;
 	u64                     gtt_base_align;
+	u64                     mc_mask;
 };
 
 bool radeon_combios_sideport_present(struct radeon_device *rdev);
@@ -918,6 +929,7 @@ struct radeon_wb {
 #define R600_WB_DMA_RPTR_OFFSET   1792
 #define R600_WB_IH_WPTR_OFFSET   2048
 #define CAYMAN_WB_DMA1_RPTR_OFFSET   2304
+#define R600_WB_UVD_RPTR_OFFSET  2560
 #define R600_WB_EVENT_OFFSET     3072
 
 /**
@@ -1118,6 +1130,33 @@ struct radeon_pm {
 int radeon_pm_get_type_index(struct radeon_device *rdev,
 			     enum radeon_pm_state_type ps_type,
 			     int instance);
+/*
+ * UVD
+ */
+#define RADEON_MAX_UVD_HANDLES	10
+#define RADEON_UVD_STACK_SIZE	(1024*1024)
+#define RADEON_UVD_HEAP_SIZE	(1024*1024)
+
+struct radeon_uvd {
+	struct radeon_bo	*vcpu_bo;
+	void			*cpu_addr;
+	uint64_t		gpu_addr;
+	atomic_t		handles[RADEON_MAX_UVD_HANDLES];
+	struct drm_file		*filp[RADEON_MAX_UVD_HANDLES];
+};
+
+int radeon_uvd_init(struct radeon_device *rdev);
+void radeon_uvd_fini(struct radeon_device *rdev);
+int radeon_uvd_suspend(struct radeon_device *rdev);
+int radeon_uvd_resume(struct radeon_device *rdev);
+int radeon_uvd_get_create_msg(struct radeon_device *rdev, int ring,
+			      uint32_t handle, struct radeon_fence **fence);
+int radeon_uvd_get_destroy_msg(struct radeon_device *rdev, int ring,
+			       uint32_t handle, struct radeon_fence **fence);
+void radeon_uvd_force_into_uvd_segment(struct radeon_bo *rbo);
+void radeon_uvd_free_handles(struct radeon_device *rdev,
+			     struct drm_file *filp);
+int radeon_uvd_cs_parse(struct radeon_cs_parser *parser);
 
 struct r600_audio {
 	int			channels;
@@ -1281,6 +1320,7 @@ struct radeon_asic {
 		int (*get_pcie_lanes)(struct radeon_device *rdev);
 		void (*set_pcie_lanes)(struct radeon_device *rdev, int lanes);
 		void (*set_clock_gating)(struct radeon_device *rdev, int enable);
+		int (*set_uvd_clocks)(struct radeon_device *rdev, u32 vclk, u32 dclk);
 	} pm;
 	/* pageflipping */
 	struct {
@@ -1443,6 +1483,7 @@ struct si_asic {
 	unsigned multi_gpu_tile_size;
 
 	unsigned tile_config;
+	uint32_t tile_mode_array[32];
 };
 
 union radeon_asic_config {
@@ -1608,6 +1649,7 @@ struct radeon_device {
 	struct radeon_asic		*asic;
 	struct radeon_gem		gem;
 	struct radeon_pm		pm;
+	struct radeon_uvd		uvd;
 	uint32_t			bios_scratch[RADEON_BIOS_NUM_SCRATCH];
 	struct radeon_wb		wb;
 	struct radeon_dummy_page	dummy_page;
@@ -1615,12 +1657,14 @@ struct radeon_device {
 	bool				suspend;
 	bool				need_dma32;
 	bool				accel_working;
+	bool				fastfb_working; /* IGP feature*/
 	struct radeon_surface_reg surface_regs[RADEON_GEM_MAX_SURFACES];
 	const struct firmware *me_fw;	/* all family ME firmware */
 	const struct firmware *pfp_fw;	/* r6/700 PFP firmware */
 	const struct firmware *rlc_fw;	/* r6/700 RLC firmware */
 	const struct firmware *mc_fw;	/* NI MC firmware */
 	const struct firmware *ce_fw;	/* SI CE firmware */
+	const struct firmware *uvd_fw;	/* UVD firmware */
 	struct r600_blit r600_blit;
 	struct r600_vram_scratch vram_scratch;
 	int msi_enabled; /* msi enabled */
@@ -1688,8 +1732,8 @@ void r100_io_wreg(struct radeon_device *rdev, u32 reg, u32 v);
 #define WREG32_MC(reg, v) rdev->mc_wreg(rdev, (reg), (v))
 #define RREG32_PCIE(reg) rv370_pcie_rreg(rdev, (reg))
 #define WREG32_PCIE(reg, v) rv370_pcie_wreg(rdev, (reg), (v))
-#define RREG32_PCIE_P(reg) rdev->pciep_rreg(rdev, (reg))
-#define WREG32_PCIE_P(reg, v) rdev->pciep_wreg(rdev, (reg), (v))
+#define RREG32_PCIE_PORT(reg) rdev->pciep_rreg(rdev, (reg))
+#define WREG32_PCIE_PORT(reg, v) rdev->pciep_wreg(rdev, (reg), (v))
 #define WREG32_P(reg, val, mask)				\
 	do {							\
 		uint32_t tmp_ = RREG32(reg);			\
@@ -1845,6 +1889,7 @@ void radeon_ring_write(struct radeon_ring *ring, uint32_t v);
 #define radeon_get_pcie_lanes(rdev) (rdev)->asic->pm.get_pcie_lanes((rdev))
 #define radeon_set_pcie_lanes(rdev, l) (rdev)->asic->pm.set_pcie_lanes((rdev), (l))
 #define radeon_set_clock_gating(rdev, e) (rdev)->asic->pm.set_clock_gating((rdev), (e))
+#define radeon_set_uvd_clocks(rdev, v, d) (rdev)->asic->pm.set_uvd_clocks((rdev), (v), (d))
 #define radeon_set_surface_reg(rdev, r, f, p, o, s) ((rdev)->asic->surface.set_reg((rdev), (r), (f), (p), (o), (s)))
 #define radeon_clear_surface_reg(rdev, r) ((rdev)->asic->surface.clear_reg((rdev), (r)))
 #define radeon_bandwidth_update(rdev) (rdev)->asic->display.bandwidth_update((rdev))

drivers/gpu/drm/radeon/radeon_asic.c

@@ -1130,6 +1130,15 @@ static struct radeon_asic rv770_asic = {
 			.ring_test = &r600_dma_ring_test,
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &r600_dma_is_lockup,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &r600_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1174,6 +1183,7 @@ static struct radeon_asic rv770_asic = {
 		.get_pcie_lanes = &r600_get_pcie_lanes,
 		.set_pcie_lanes = &r600_set_pcie_lanes,
 		.set_clock_gating = &radeon_atom_set_clock_gating,
+		.set_uvd_clocks = &rv770_set_uvd_clocks,
 	},
 	.pflip = {
 		.pre_page_flip = &rs600_pre_page_flip,
@@ -1216,6 +1226,15 @@ static struct radeon_asic evergreen_asic = {
 			.ring_test = &r600_dma_ring_test,
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &evergreen_dma_is_lockup,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &r600_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1260,6 +1279,7 @@ static struct radeon_asic evergreen_asic = {
 		.get_pcie_lanes = &r600_get_pcie_lanes,
 		.set_pcie_lanes = &r600_set_pcie_lanes,
 		.set_clock_gating = NULL,
+		.set_uvd_clocks = &evergreen_set_uvd_clocks,
 	},
 	.pflip = {
 		.pre_page_flip = &evergreen_pre_page_flip,
@@ -1302,6 +1322,15 @@ static struct radeon_asic sumo_asic = {
 			.ring_test = &r600_dma_ring_test,
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &evergreen_dma_is_lockup,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &r600_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1346,6 +1375,7 @@ static struct radeon_asic sumo_asic = {
 		.get_pcie_lanes = NULL,
 		.set_pcie_lanes = NULL,
 		.set_clock_gating = NULL,
+		.set_uvd_clocks = &sumo_set_uvd_clocks,
 	},
 	.pflip = {
 		.pre_page_flip = &evergreen_pre_page_flip,
@@ -1388,6 +1418,15 @@ static struct radeon_asic btc_asic = {
 			.ring_test = &r600_dma_ring_test,
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &evergreen_dma_is_lockup,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &r600_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1429,9 +1468,10 @@ static struct radeon_asic btc_asic = {
 		.set_engine_clock = &radeon_atom_set_engine_clock,
 		.get_memory_clock = &radeon_atom_get_memory_clock,
 		.set_memory_clock = &radeon_atom_set_memory_clock,
-		.get_pcie_lanes = NULL,
-		.set_pcie_lanes = NULL,
+		.get_pcie_lanes = &r600_get_pcie_lanes,
+		.set_pcie_lanes = &r600_set_pcie_lanes,
 		.set_clock_gating = NULL,
+		.set_uvd_clocks = &evergreen_set_uvd_clocks,
 	},
 	.pflip = {
 		.pre_page_flip = &evergreen_pre_page_flip,
@@ -1517,6 +1557,15 @@ static struct radeon_asic cayman_asic = {
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &cayman_dma_is_lockup,
 			.vm_flush = &cayman_dma_vm_flush,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &cayman_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1558,9 +1607,10 @@ static struct radeon_asic cayman_asic = {
 		.set_engine_clock = &radeon_atom_set_engine_clock,
 		.get_memory_clock = &radeon_atom_get_memory_clock,
 		.set_memory_clock = &radeon_atom_set_memory_clock,
-		.get_pcie_lanes = NULL,
-		.set_pcie_lanes = NULL,
+		.get_pcie_lanes = &r600_get_pcie_lanes,
+		.set_pcie_lanes = &r600_set_pcie_lanes,
 		.set_clock_gating = NULL,
+		.set_uvd_clocks = &evergreen_set_uvd_clocks,
 	},
 	.pflip = {
 		.pre_page_flip = &evergreen_pre_page_flip,
@@ -1646,6 +1696,15 @@ static struct radeon_asic trinity_asic = {
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &cayman_dma_is_lockup,
 			.vm_flush = &cayman_dma_vm_flush,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &cayman_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1690,6 +1749,7 @@ static struct radeon_asic trinity_asic = {
 		.get_pcie_lanes = NULL,
 		.set_pcie_lanes = NULL,
 		.set_clock_gating = NULL,
+		.set_uvd_clocks = &sumo_set_uvd_clocks,
 	},
 	.pflip = {
 		.pre_page_flip = &evergreen_pre_page_flip,
@@ -1775,6 +1835,15 @@ static struct radeon_asic si_asic = {
 			.ib_test = &r600_dma_ib_test,
 			.is_lockup = &si_dma_is_lockup,
 			.vm_flush = &si_dma_vm_flush,
+		},
+		[R600_RING_TYPE_UVD_INDEX] = {
+			.ib_execute = &r600_uvd_ib_execute,
+			.emit_fence = &r600_uvd_fence_emit,
+			.emit_semaphore = &cayman_uvd_semaphore_emit,
+			.cs_parse = &radeon_uvd_cs_parse,
+			.ring_test = &r600_uvd_ring_test,
+			.ib_test = &r600_uvd_ib_test,
+			.is_lockup = &radeon_ring_test_lockup,
 		}
 	},
 	.irq = {
@@ -1816,9 +1885,10 @@ static struct radeon_asic si_asic = {
 		.set_engine_clock = &radeon_atom_set_engine_clock,
 		.get_memory_clock = &radeon_atom_get_memory_clock,
 		.set_memory_clock = &radeon_atom_set_memory_clock,
-		.get_pcie_lanes = NULL,
-		.set_pcie_lanes = NULL,
+		.get_pcie_lanes = &r600_get_pcie_lanes,
+		.set_pcie_lanes = &r600_set_pcie_lanes,
 		.set_clock_gating = NULL,
+		.set_uvd_clocks = &si_set_uvd_clocks,
 	},
 	.pflip = {
 		.pre_page_flip = &evergreen_pre_page_flip,

drivers/gpu/drm/radeon/radeon_asic.h

@@ -330,6 +330,7 @@ int r600_dma_ib_test(struct radeon_device *rdev, struct radeon_ring *ring);
 void r600_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib);
 int r600_ring_test(struct radeon_device *rdev, struct radeon_ring *cp);
 int r600_dma_ring_test(struct radeon_device *rdev, struct radeon_ring *cp);
+int r600_uvd_ring_test(struct radeon_device *rdev, struct radeon_ring *ring);
 int r600_copy_blit(struct radeon_device *rdev,
 		   uint64_t src_offset, uint64_t dst_offset,
 		   unsigned num_gpu_pages, struct radeon_fence **fence);
@@ -392,6 +393,19 @@ int r600_mc_wait_for_idle(struct radeon_device *rdev);
 u32 r600_get_xclk(struct radeon_device *rdev);
 uint64_t r600_get_gpu_clock_counter(struct radeon_device *rdev);
 
+/* uvd */
+int r600_uvd_init(struct radeon_device *rdev);
+int r600_uvd_rbc_start(struct radeon_device *rdev);
+void r600_uvd_rbc_stop(struct radeon_device *rdev);
+int r600_uvd_ib_test(struct radeon_device *rdev, struct radeon_ring *ring);
+void r600_uvd_fence_emit(struct radeon_device *rdev,
+			 struct radeon_fence *fence);
+void r600_uvd_semaphore_emit(struct radeon_device *rdev,
+			     struct radeon_ring *ring,
+			     struct radeon_semaphore *semaphore,
+			     bool emit_wait);
+void r600_uvd_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib);
+
 /*
  * rv770,rv730,rv710,rv740
  */
@@ -409,6 +423,8 @@ int rv770_copy_dma(struct radeon_device *rdev,
 		  unsigned num_gpu_pages,
 		   struct radeon_fence **fence);
 u32 rv770_get_xclk(struct radeon_device *rdev);
+int rv770_uvd_resume(struct radeon_device *rdev);
+int rv770_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk);
 
 /*
  * evergreen
@@ -444,6 +460,8 @@ extern void evergreen_pm_prepare(struct radeon_device *rdev);
 extern void evergreen_pm_finish(struct radeon_device *rdev);
 extern void sumo_pm_init_profile(struct radeon_device *rdev);
 extern void btc_pm_init_profile(struct radeon_device *rdev);
+int sumo_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk);
+int evergreen_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk);
 extern void evergreen_pre_page_flip(struct radeon_device *rdev, int crtc);
 extern u32 evergreen_page_flip(struct radeon_device *rdev, int crtc, u64 crtc_base);
 extern void evergreen_post_page_flip(struct radeon_device *rdev, int crtc);
@@ -465,6 +483,10 @@ int evergreen_copy_dma(struct radeon_device *rdev,
  */
 void cayman_fence_ring_emit(struct radeon_device *rdev,
 			    struct radeon_fence *fence);
+void cayman_uvd_semaphore_emit(struct radeon_device *rdev,
+			       struct radeon_ring *ring,
+			       struct radeon_semaphore *semaphore,
+			       bool emit_wait);
 void cayman_pcie_gart_tlb_flush(struct radeon_device *rdev);
 int cayman_init(struct radeon_device *rdev);
 void cayman_fini(struct radeon_device *rdev);
@@ -524,5 +546,6 @@ int si_copy_dma(struct radeon_device *rdev,
 void si_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm);
 u32 si_get_xclk(struct radeon_device *rdev);
 uint64_t si_get_gpu_clock_counter(struct radeon_device *rdev);
+int si_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk);
 
 #endif

drivers/gpu/drm/radeon/radeon_atombios.c

@@ -2654,6 +2654,113 @@ void radeon_atombios_get_power_modes(struct radeon_device *rdev)
 		rdev->pm.current_vddc = 0;
 }
 
+union get_clock_dividers {
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS v1;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V2 v2;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V3 v3;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V4 v4;
+	struct _COMPUTE_MEMORY_ENGINE_PLL_PARAMETERS_V5 v5;
+};
+
+int radeon_atom_get_clock_dividers(struct radeon_device *rdev,
+				   u8 clock_type,
+				   u32 clock,
+				   bool strobe_mode,
+				   struct atom_clock_dividers *dividers)
+{
+	union get_clock_dividers args;
+	int index = GetIndexIntoMasterTable(COMMAND, ComputeMemoryEnginePLL);
+	u8 frev, crev;
+
+	memset(&args, 0, sizeof(args));
+	memset(dividers, 0, sizeof(struct atom_clock_dividers));
+
+	if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev))
+		return -EINVAL;
+
+	switch (crev) {
+	case 1:
+		/* r4xx, r5xx */
+		args.v1.ucAction = clock_type;
+		args.v1.ulClock = cpu_to_le32(clock);	/* 10 khz */
+
+		atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		dividers->post_div = args.v1.ucPostDiv;
+		dividers->fb_div = args.v1.ucFbDiv;
+		dividers->enable_post_div = true;
+		break;
+	case 2:
+	case 3:
+		/* r6xx, r7xx, evergreen, ni */
+		if (rdev->family <= CHIP_RV770) {
+			args.v2.ucAction = clock_type;
+			args.v2.ulClock = cpu_to_le32(clock);	/* 10 khz */
+
+			atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+
+			dividers->post_div = args.v2.ucPostDiv;
+			dividers->fb_div = le16_to_cpu(args.v2.usFbDiv);
+			dividers->ref_div = args.v2.ucAction;
+			if (rdev->family == CHIP_RV770) {
+				dividers->enable_post_div = (le32_to_cpu(args.v2.ulClock) & (1 << 24)) ?
+					true : false;
+				dividers->vco_mode = (le32_to_cpu(args.v2.ulClock) & (1 << 25)) ? 1 : 0;
+			} else
+				dividers->enable_post_div = (dividers->fb_div & 1) ? true : false;
+		} else {
+			if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
+				args.v3.ulClock.ulComputeClockFlag = clock_type;
+				args.v3.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+
+				atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+
+				dividers->post_div = args.v3.ucPostDiv;
+				dividers->enable_post_div = (args.v3.ucCntlFlag &
+							     ATOM_PLL_CNTL_FLAG_PLL_POST_DIV_EN) ? true : false;
+				dividers->enable_dithen = (args.v3.ucCntlFlag &
+							   ATOM_PLL_CNTL_FLAG_FRACTION_DISABLE) ? false : true;
+				dividers->fb_div = le16_to_cpu(args.v3.ulFbDiv.usFbDiv);
+				dividers->frac_fb_div = le16_to_cpu(args.v3.ulFbDiv.usFbDivFrac);
+				dividers->ref_div = args.v3.ucRefDiv;
+				dividers->vco_mode = (args.v3.ucCntlFlag &
+						      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
+			} else {
+				args.v5.ulClock.ulComputeClockFlag = clock_type;
+				args.v5.ulClock.ulClockFreq = cpu_to_le32(clock);	/* 10 khz */
+				if (strobe_mode)
+					args.v5.ucInputFlag = ATOM_PLL_INPUT_FLAG_PLL_STROBE_MODE_EN;
+
+				atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+
+				dividers->post_div = args.v5.ucPostDiv;
+				dividers->enable_post_div = (args.v5.ucCntlFlag &
+							     ATOM_PLL_CNTL_FLAG_PLL_POST_DIV_EN) ? true : false;
+				dividers->enable_dithen = (args.v5.ucCntlFlag &
+							   ATOM_PLL_CNTL_FLAG_FRACTION_DISABLE) ? false : true;
+				dividers->whole_fb_div = le16_to_cpu(args.v5.ulFbDiv.usFbDiv);
+				dividers->frac_fb_div = le16_to_cpu(args.v5.ulFbDiv.usFbDivFrac);
+				dividers->ref_div = args.v5.ucRefDiv;
+				dividers->vco_mode = (args.v5.ucCntlFlag &
+						      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
+			}
+		}
+		break;
+	case 4:
+		/* fusion */
+		args.v4.ulClock = cpu_to_le32(clock);	/* 10 khz */
+
+		atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		dividers->post_div = args.v4.ucPostDiv;
+		dividers->real_clock = le32_to_cpu(args.v4.ulClock);
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
 void radeon_atom_set_clock_gating(struct radeon_device *rdev, int enable)
 {
 	DYNAMIC_CLOCK_GATING_PS_ALLOCATION args;

drivers/gpu/drm/radeon/radeon_cs.c

@@ -63,30 +63,50 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p)
 				break;
 			}
 		}
-		if (!duplicate) {
-			p->relocs[i].gobj = drm_gem_object_lookup(ddev,
-								  p->filp,
-								  r->handle);
-			if (p->relocs[i].gobj == NULL) {
-				DRM_ERROR("gem object lookup failed 0x%x\n",
-					  r->handle);
-				return -ENOENT;
-			}
-			p->relocs_ptr[i] = &p->relocs[i];
-			p->relocs[i].robj = gem_to_radeon_bo(p->relocs[i].gobj);
-			p->relocs[i].lobj.bo = p->relocs[i].robj;
-			p->relocs[i].lobj.wdomain = r->write_domain;
-			p->relocs[i].lobj.rdomain = r->read_domains;
-			p->relocs[i].lobj.tv.bo = &p->relocs[i].robj->tbo;
-			p->relocs[i].handle = r->handle;
-			p->relocs[i].flags = r->flags;
-			radeon_bo_list_add_object(&p->relocs[i].lobj,
-						  &p->validated);
-
-		} else
+		if (duplicate) {
 			p->relocs[i].handle = 0;
+			continue;
+		}
+
+		p->relocs[i].gobj = drm_gem_object_lookup(ddev, p->filp,
+							  r->handle);
+		if (p->relocs[i].gobj == NULL) {
+			DRM_ERROR("gem object lookup failed 0x%x\n",
+				  r->handle);
+			return -ENOENT;
+		}
+		p->relocs_ptr[i] = &p->relocs[i];
+		p->relocs[i].robj = gem_to_radeon_bo(p->relocs[i].gobj);
+		p->relocs[i].lobj.bo = p->relocs[i].robj;
+		p->relocs[i].lobj.written = !!r->write_domain;
+
+		/* the first reloc of an UVD job is the
+		   msg and that must be in VRAM */
+		if (p->ring == R600_RING_TYPE_UVD_INDEX && i == 0) {
+			/* TODO: is this still needed for NI+ ? */
+			p->relocs[i].lobj.domain =
+				RADEON_GEM_DOMAIN_VRAM;
+
+			p->relocs[i].lobj.alt_domain =
+				RADEON_GEM_DOMAIN_VRAM;
+
+		} else {
+			uint32_t domain = r->write_domain ?
+				r->write_domain : r->read_domains;
+
+			p->relocs[i].lobj.domain = domain;
+			if (domain == RADEON_GEM_DOMAIN_VRAM)
+				domain |= RADEON_GEM_DOMAIN_GTT;
+			p->relocs[i].lobj.alt_domain = domain;
+		}
+
+		p->relocs[i].lobj.tv.bo = &p->relocs[i].robj->tbo;
+		p->relocs[i].handle = r->handle;
+
+		radeon_bo_list_add_object(&p->relocs[i].lobj,
+					  &p->validated);
 	}
-	return radeon_bo_list_validate(&p->validated);
+	return radeon_bo_list_validate(&p->validated, p->ring);
 }
 
 static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority)
@@ -121,6 +141,9 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority
 			return -EINVAL;
 		}
 		break;
+	case RADEON_CS_RING_UVD:
+		p->ring = R600_RING_TYPE_UVD_INDEX;
+		break;
 	}
 	return 0;
 }
@@ -241,15 +264,15 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data)
 			return -EINVAL;
 		}
 
-		/* we only support VM on SI+ */
-		if ((p->rdev->family >= CHIP_TAHITI) &&
-		    ((p->cs_flags & RADEON_CS_USE_VM) == 0)) {
-			DRM_ERROR("VM required on SI+!\n");
+		if (radeon_cs_get_ring(p, ring, priority))
 			return -EINVAL;
-		}
 
-		if (radeon_cs_get_ring(p, ring, priority))
+		/* we only support VM on some SI+ rings */
+		if ((p->rdev->asic->ring[p->ring].cs_parse == NULL) &&
+		   ((p->cs_flags & RADEON_CS_USE_VM) == 0)) {
+			DRM_ERROR("Ring %d requires VM!\n", p->ring);
 			return -EINVAL;
+		}
 	}
 
 	/* deal with non-vm */

drivers/gpu/drm/radeon/radeon_device.c

@@ -359,7 +359,7 @@ void radeon_vram_location(struct radeon_device *rdev, struct radeon_mc *mc, u64
 	uint64_t limit = (uint64_t)radeon_vram_limit << 20;
 
 	mc->vram_start = base;
-	if (mc->mc_vram_size > (0xFFFFFFFF - base + 1)) {
+	if (mc->mc_vram_size > (rdev->mc.mc_mask - base + 1)) {
 		dev_warn(rdev->dev, "limiting VRAM to PCI aperture size\n");
 		mc->real_vram_size = mc->aper_size;
 		mc->mc_vram_size = mc->aper_size;
@@ -394,7 +394,7 @@ void radeon_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc)
 {
 	u64 size_af, size_bf;
 
-	size_af = ((0xFFFFFFFF - mc->vram_end) + mc->gtt_base_align) & ~mc->gtt_base_align;
+	size_af = ((rdev->mc.mc_mask - mc->vram_end) + mc->gtt_base_align) & ~mc->gtt_base_align;
 	size_bf = mc->vram_start & ~mc->gtt_base_align;
 	if (size_bf > size_af) {
 		if (mc->gtt_size > size_bf) {
@@ -1068,6 +1068,17 @@ int radeon_device_init(struct radeon_device *rdev,
 		radeon_agp_disable(rdev);
 	}
 
+	/* Set the internal MC address mask
+	 * This is the max address of the GPU's
+	 * internal address space.
+	 */
+	if (rdev->family >= CHIP_CAYMAN)
+		rdev->mc.mc_mask = 0xffffffffffULL; /* 40 bit MC */
+	else if (rdev->family >= CHIP_CEDAR)
+		rdev->mc.mc_mask = 0xfffffffffULL; /* 36 bit MC */
+	else
+		rdev->mc.mc_mask = 0xffffffffULL; /* 32 bit MC */
+
 	/* set DMA mask + need_dma32 flags.
 	 * PCIE - can handle 40-bits.
 	 * IGP - can handle 40-bits

drivers/gpu/drm/radeon/radeon_drv.c

@@ -71,9 +71,12 @@
  *   2.28.0 - r600-eg: Add MEM_WRITE packet support
  *   2.29.0 - R500 FP16 color clear registers
  *   2.30.0 - fix for FMASK texturing
+ *   2.31.0 - Add fastfb support for rs690
+ *   2.32.0 - new info request for rings working
+ *   2.33.0 - Add SI tiling mode array query
  */
 #define KMS_DRIVER_MAJOR	2
-#define KMS_DRIVER_MINOR	30
+#define KMS_DRIVER_MINOR	33
 #define KMS_DRIVER_PATCHLEVEL	0
 int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags);
 int radeon_driver_unload_kms(struct drm_device *dev);
@@ -160,6 +163,7 @@ int radeon_hw_i2c = 0;
 int radeon_pcie_gen2 = -1;
 int radeon_msi = -1;
 int radeon_lockup_timeout = 10000;
+int radeon_fastfb = 0;
 
 MODULE_PARM_DESC(no_wb, "Disable AGP writeback for scratch registers");
 module_param_named(no_wb, radeon_no_wb, int, 0444);
@@ -212,6 +216,9 @@ module_param_named(msi, radeon_msi, int, 0444);
 MODULE_PARM_DESC(lockup_timeout, "GPU lockup timeout in ms (defaul 10000 = 10 seconds, 0 = disable)");
 module_param_named(lockup_timeout, radeon_lockup_timeout, int, 0444);
 
+MODULE_PARM_DESC(fastfb, "Direct FB access for IGP chips (0 = disable, 1 = enable)");
+module_param_named(fastfb, radeon_fastfb, int, 0444);
+
 static struct pci_device_id pciidlist[] = {
 	radeon_PCI_IDS
 };

drivers/gpu/drm/radeon/radeon_fence.c

@@ -31,9 +31,9 @@
 #include <linux/seq_file.h>
 #include <linux/atomic.h>
 #include <linux/wait.h>
-#include <linux/list.h>
 #include <linux/kref.h>
 #include <linux/slab.h>
+#include <linux/firmware.h>
 #include <drm/drmP.h>
 #include "radeon_reg.h"
 #include "radeon.h"
@@ -767,8 +767,21 @@ int radeon_fence_driver_start_ring(struct radeon_device *rdev, int ring)
 
 	radeon_scratch_free(rdev, rdev->fence_drv[ring].scratch_reg);
 	if (rdev->wb.use_event || !radeon_ring_supports_scratch_reg(rdev, &rdev->ring[ring])) {
-		rdev->fence_drv[ring].scratch_reg = 0;
-		index = R600_WB_EVENT_OFFSET + ring * 4;
+		if (ring != R600_RING_TYPE_UVD_INDEX) {
+			rdev->fence_drv[ring].scratch_reg = 0;
+			index = R600_WB_EVENT_OFFSET + ring * 4;
+			rdev->fence_drv[ring].cpu_addr = &rdev->wb.wb[index/4];
+			rdev->fence_drv[ring].gpu_addr = rdev->wb.gpu_addr +
+							 index;
+
+		} else {
+			/* put fence directly behind firmware */
+			rdev->fence_drv[ring].cpu_addr = rdev->uvd.cpu_addr +
+							 rdev->uvd_fw->size;
+			rdev->fence_drv[ring].gpu_addr = rdev->uvd.gpu_addr +
+							 rdev->uvd_fw->size;
+		}
+
 	} else {
 		r = radeon_scratch_get(rdev, &rdev->fence_drv[ring].scratch_reg);
 		if (r) {
@@ -778,9 +791,9 @@ int radeon_fence_driver_start_ring(struct radeon_device *rdev, int ring)
 		index = RADEON_WB_SCRATCH_OFFSET +
 			rdev->fence_drv[ring].scratch_reg -
 			rdev->scratch.reg_base;
+		rdev->fence_drv[ring].cpu_addr = &rdev->wb.wb[index/4];
+		rdev->fence_drv[ring].gpu_addr = rdev->wb.gpu_addr + index;
 	}
-	rdev->fence_drv[ring].cpu_addr = &rdev->wb.wb[index/4];
-	rdev->fence_drv[ring].gpu_addr = rdev->wb.gpu_addr + index;
 	radeon_fence_write(rdev, atomic64_read(&rdev->fence_drv[ring].last_seq), ring);
 	rdev->fence_drv[ring].initialized = true;
 	dev_info(rdev->dev, "fence driver on ring %d use gpu addr 0x%016llx and cpu addr 0x%p\n",

drivers/gpu/drm/radeon/radeon_kms.c

@@ -50,9 +50,13 @@ int radeon_driver_unload_kms(struct drm_device *dev)
 
 	if (rdev == NULL)
 		return 0;
+	if (rdev->rmmio == NULL)
+		goto done_free;
 	radeon_acpi_fini(rdev);
 	radeon_modeset_fini(rdev);
 	radeon_device_fini(rdev);
+
+done_free:
 	kfree(rdev);
 	dev->dev_private = NULL;
 	return 0;
@@ -176,80 +180,65 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 	struct radeon_device *rdev = dev->dev_private;
 	struct drm_radeon_info *info = data;
 	struct radeon_mode_info *minfo = &rdev->mode_info;
-	uint32_t value, *value_ptr;
-	uint64_t value64, *value_ptr64;
+	uint32_t *value, value_tmp, *value_ptr, value_size;
+	uint64_t value64;
 	struct drm_crtc *crtc;
 	int i, found;
 
-	/* TIMESTAMP is a 64-bit value, needs special handling. */
-	if (info->request == RADEON_INFO_TIMESTAMP) {
-		if (rdev->family >= CHIP_R600) {
-			value_ptr64 = (uint64_t*)((unsigned long)info->value);
-			value64 = radeon_get_gpu_clock_counter(rdev);
-
-			if (DRM_COPY_TO_USER(value_ptr64, &value64, sizeof(value64))) {
-				DRM_ERROR("copy_to_user %s:%u\n", __func__, __LINE__);
-				return -EFAULT;
-			}
-			return 0;
-		} else {
-			DRM_DEBUG_KMS("timestamp is r6xx+ only!\n");
-			return -EINVAL;
-		}
-	}
-
 	value_ptr = (uint32_t *)((unsigned long)info->value);
-	if (DRM_COPY_FROM_USER(&value, value_ptr, sizeof(value))) {
-		DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
-		return -EFAULT;
-	}
+	value = &value_tmp;
+	value_size = sizeof(uint32_t);
 
 	switch (info->request) {
 	case RADEON_INFO_DEVICE_ID:
-		value = dev->pci_device;
+		*value = dev->pci_device;
 		break;
 	case RADEON_INFO_NUM_GB_PIPES:
-		value = rdev->num_gb_pipes;
+		*value = rdev->num_gb_pipes;
 		break;
 	case RADEON_INFO_NUM_Z_PIPES:
-		value = rdev->num_z_pipes;
+		*value = rdev->num_z_pipes;
 		break;
 	case RADEON_INFO_ACCEL_WORKING:
 		/* xf86-video-ati 6.13.0 relies on this being false for evergreen */
 		if ((rdev->family >= CHIP_CEDAR) && (rdev->family <= CHIP_HEMLOCK))
-			value = false;
+			*value = false;
 		else
-			value = rdev->accel_working;
+			*value = rdev->accel_working;
 		break;
 	case RADEON_INFO_CRTC_FROM_ID:
+		if (DRM_COPY_FROM_USER(value, value_ptr, sizeof(uint32_t))) {
+			DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
+			return -EFAULT;
+		}
 		for (i = 0, found = 0; i < rdev->num_crtc; i++) {
 			crtc = (struct drm_crtc *)minfo->crtcs[i];
-			if (crtc && crtc->base.id == value) {
+			if (crtc && crtc->base.id == *value) {
 				struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc);
-				value = radeon_crtc->crtc_id;
+				*value = radeon_crtc->crtc_id;
 				found = 1;
 				break;
 			}
 		}
 		if (!found) {
-			DRM_DEBUG_KMS("unknown crtc id %d\n", value);
+			DRM_DEBUG_KMS("unknown crtc id %d\n", *value);
 			return -EINVAL;
 		}
 		break;
 	case RADEON_INFO_ACCEL_WORKING2:
-		value = rdev->accel_working;
+		*value = rdev->accel_working;
 		break;
 	case RADEON_INFO_TILING_CONFIG:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.tile_config;
+			*value = rdev->config.si.tile_config;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.tile_config;
+			*value = rdev->config.cayman.tile_config;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.tile_config;
+			*value = rdev->config.evergreen.tile_config;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.tile_config;
+			*value = rdev->config.rv770.tile_config;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.tile_config;
+			*value = rdev->config.r600.tile_config;
 		else {
 			DRM_DEBUG_KMS("tiling config is r6xx+ only!\n");
 			return -EINVAL;
@@ -262,73 +251,81 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		 *
 		 * When returning, the value is 1 if filp owns hyper-z access,
 		 * 0 otherwise. */
-		if (value >= 2) {
-			DRM_DEBUG_KMS("WANT_HYPERZ: invalid value %d\n", value);
+		if (DRM_COPY_FROM_USER(value, value_ptr, sizeof(uint32_t))) {
+			DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
+			return -EFAULT;
+		}
+		if (*value >= 2) {
+			DRM_DEBUG_KMS("WANT_HYPERZ: invalid value %d\n", *value);
 			return -EINVAL;
 		}
-		radeon_set_filp_rights(dev, &rdev->hyperz_filp, filp, &value);
+		radeon_set_filp_rights(dev, &rdev->hyperz_filp, filp, value);
 		break;
 	case RADEON_INFO_WANT_CMASK:
 		/* The same logic as Hyper-Z. */
-		if (value >= 2) {
-			DRM_DEBUG_KMS("WANT_CMASK: invalid value %d\n", value);
+		if (DRM_COPY_FROM_USER(value, value_ptr, sizeof(uint32_t))) {
+			DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
+			return -EFAULT;
+		}
+		if (*value >= 2) {
+			DRM_DEBUG_KMS("WANT_CMASK: invalid value %d\n", *value);
 			return -EINVAL;
 		}
-		radeon_set_filp_rights(dev, &rdev->cmask_filp, filp, &value);
+		radeon_set_filp_rights(dev, &rdev->cmask_filp, filp, value);
 		break;
 	case RADEON_INFO_CLOCK_CRYSTAL_FREQ:
 		/* return clock value in KHz */
 		if (rdev->asic->get_xclk)
-			value = radeon_get_xclk(rdev) * 10;
+			*value = radeon_get_xclk(rdev) * 10;
 		else
-			value = rdev->clock.spll.reference_freq * 10;
+			*value = rdev->clock.spll.reference_freq * 10;
 		break;
 	case RADEON_INFO_NUM_BACKENDS:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_backends_per_se *
+			*value = rdev->config.si.max_backends_per_se *
 				rdev->config.si.max_shader_engines;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.max_backends_per_se *
+			*value = rdev->config.cayman.max_backends_per_se *
 				rdev->config.cayman.max_shader_engines;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.max_backends;
+			*value = rdev->config.evergreen.max_backends;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.max_backends;
+			*value = rdev->config.rv770.max_backends;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.max_backends;
+			*value = rdev->config.r600.max_backends;
 		else {
 			return -EINVAL;
 		}
 		break;
 	case RADEON_INFO_NUM_TILE_PIPES:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_tile_pipes;
+			*value = rdev->config.si.max_tile_pipes;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.max_tile_pipes;
+			*value = rdev->config.cayman.max_tile_pipes;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.max_tile_pipes;
+			*value = rdev->config.evergreen.max_tile_pipes;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.max_tile_pipes;
+			*value = rdev->config.rv770.max_tile_pipes;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.max_tile_pipes;
+			*value = rdev->config.r600.max_tile_pipes;
 		else {
 			return -EINVAL;
 		}
 		break;
 	case RADEON_INFO_FUSION_GART_WORKING:
-		value = 1;
+		*value = 1;
 		break;
 	case RADEON_INFO_BACKEND_MAP:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.backend_map;
+			*value = rdev->config.si.backend_map;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.backend_map;
+			*value = rdev->config.cayman.backend_map;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.backend_map;
+			*value = rdev->config.evergreen.backend_map;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.backend_map;
+			*value = rdev->config.rv770.backend_map;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.backend_map;
+			*value = rdev->config.r600.backend_map;
 		else {
 			return -EINVAL;
 		}
@@ -337,50 +334,91 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		/* this is where we report if vm is supported or not */
 		if (rdev->family < CHIP_CAYMAN)
 			return -EINVAL;
-		value = RADEON_VA_RESERVED_SIZE;
+		*value = RADEON_VA_RESERVED_SIZE;
 		break;
 	case RADEON_INFO_IB_VM_MAX_SIZE:
 		/* this is where we report if vm is supported or not */
 		if (rdev->family < CHIP_CAYMAN)
 			return -EINVAL;
-		value = RADEON_IB_VM_MAX_SIZE;
+		*value = RADEON_IB_VM_MAX_SIZE;
 		break;
 	case RADEON_INFO_MAX_PIPES:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_cu_per_sh;
+			*value = rdev->config.si.max_cu_per_sh;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.max_pipes_per_simd;
+			*value = rdev->config.cayman.max_pipes_per_simd;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.max_pipes;
+			*value = rdev->config.evergreen.max_pipes;
 		else if (rdev->family >= CHIP_RV770)
-			value = rdev->config.rv770.max_pipes;
+			*value = rdev->config.rv770.max_pipes;
 		else if (rdev->family >= CHIP_R600)
-			value = rdev->config.r600.max_pipes;
+			*value = rdev->config.r600.max_pipes;
 		else {
 			return -EINVAL;
 		}
 		break;
+	case RADEON_INFO_TIMESTAMP:
+		if (rdev->family < CHIP_R600) {
+			DRM_DEBUG_KMS("timestamp is r6xx+ only!\n");
+			return -EINVAL;
+		}
+		value = (uint32_t*)&value64;
+		value_size = sizeof(uint64_t);
+		value64 = radeon_get_gpu_clock_counter(rdev);
+		break;
 	case RADEON_INFO_MAX_SE:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_shader_engines;
+			*value = rdev->config.si.max_shader_engines;
 		else if (rdev->family >= CHIP_CAYMAN)
-			value = rdev->config.cayman.max_shader_engines;
+			*value = rdev->config.cayman.max_shader_engines;
 		else if (rdev->family >= CHIP_CEDAR)
-			value = rdev->config.evergreen.num_ses;
+			*value = rdev->config.evergreen.num_ses;
 		else
-			value = 1;
+			*value = 1;
 		break;
 	case RADEON_INFO_MAX_SH_PER_SE:
 		if (rdev->family >= CHIP_TAHITI)
-			value = rdev->config.si.max_sh_per_se;
+			*value = rdev->config.si.max_sh_per_se;
 		else
 			return -EINVAL;
 		break;
+	case RADEON_INFO_FASTFB_WORKING:
+		*value = rdev->fastfb_working;
+		break;
+	case RADEON_INFO_RING_WORKING:
+		if (DRM_COPY_FROM_USER(value, value_ptr, sizeof(uint32_t))) {
+			DRM_ERROR("copy_from_user %s:%u\n", __func__, __LINE__);
+			return -EFAULT;
+		}
+		switch (*value) {
+		case RADEON_CS_RING_GFX:
+		case RADEON_CS_RING_COMPUTE:
+			*value = rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ready;
+			break;
+		case RADEON_CS_RING_DMA:
+			*value = rdev->ring[R600_RING_TYPE_DMA_INDEX].ready;
+			*value |= rdev->ring[CAYMAN_RING_TYPE_DMA1_INDEX].ready;
+			break;
+		case RADEON_CS_RING_UVD:
+			*value = rdev->ring[R600_RING_TYPE_UVD_INDEX].ready;
+			break;
+		default:
+			return -EINVAL;
+		}
+		break;
+	case RADEON_INFO_SI_TILE_MODE_ARRAY:
+		if (rdev->family < CHIP_TAHITI) {
+			DRM_DEBUG_KMS("tile mode array is si only!\n");
+			return -EINVAL;
+		}
+		value = rdev->config.si.tile_mode_array;
+		value_size = sizeof(uint32_t)*32;
+		break;
 	default:
 		DRM_DEBUG_KMS("Invalid request %d\n", info->request);
 		return -EINVAL;
 	}
-	if (DRM_COPY_TO_USER(value_ptr, &value, sizeof(uint32_t))) {
+	if (DRM_COPY_TO_USER(value_ptr, (char*)value, value_size)) {
 		DRM_ERROR("copy_to_user %s:%u\n", __func__, __LINE__);
 		return -EFAULT;
 	}
@@ -513,6 +551,7 @@ void radeon_driver_preclose_kms(struct drm_device *dev,
 		rdev->hyperz_filp = NULL;
 	if (rdev->cmask_filp == file_priv)
 		rdev->cmask_filp = NULL;
+	radeon_uvd_free_handles(rdev, file_priv);
 }
 
 /*

drivers/gpu/drm/radeon/radeon_mode.h

@@ -492,6 +492,29 @@ struct radeon_framebuffer {
 #define ENCODER_MODE_IS_DP(em) (((em) == ATOM_ENCODER_MODE_DP) || \
 				((em) == ATOM_ENCODER_MODE_DP_MST))
 
+struct atom_clock_dividers {
+	u32 post_div;
+	union {
+		struct {
+#ifdef __BIG_ENDIAN
+			u32 reserved : 6;
+			u32 whole_fb_div : 12;
+			u32 frac_fb_div : 14;
+#else
+			u32 frac_fb_div : 14;
+			u32 whole_fb_div : 12;
+			u32 reserved : 6;
+#endif
+		};
+		u32 fb_div;
+	};
+	u32 ref_div;
+	bool enable_post_div;
+	bool enable_dithen;
+	u32 vco_mode;
+	u32 real_clock;
+};
+
 extern enum radeon_tv_std
 radeon_combios_get_tv_info(struct radeon_device *rdev);
 extern enum radeon_tv_std

drivers/gpu/drm/radeon/radeon_object.c

@@ -321,8 +321,10 @@ void radeon_bo_force_delete(struct radeon_device *rdev)
 int radeon_bo_init(struct radeon_device *rdev)
 {
 	/* Add an MTRR for the VRAM */
-	rdev->mc.vram_mtrr = mtrr_add(rdev->mc.aper_base, rdev->mc.aper_size,
+	if (!rdev->fastfb_working) {
+		rdev->mc.vram_mtrr = mtrr_add(rdev->mc.aper_base, rdev->mc.aper_size,
 			MTRR_TYPE_WRCOMB, 1);
+	}
 	DRM_INFO("Detected VRAM RAM=%lluM, BAR=%lluM\n",
 		rdev->mc.mc_vram_size >> 20,
 		(unsigned long long)rdev->mc.aper_size >> 20);
@@ -339,14 +341,14 @@ void radeon_bo_fini(struct radeon_device *rdev)
 void radeon_bo_list_add_object(struct radeon_bo_list *lobj,
 				struct list_head *head)
 {
-	if (lobj->wdomain) {
+	if (lobj->written) {
 		list_add(&lobj->tv.head, head);
 	} else {
 		list_add_tail(&lobj->tv.head, head);
 	}
 }
 
-int radeon_bo_list_validate(struct list_head *head)
+int radeon_bo_list_validate(struct list_head *head, int ring)
 {
 	struct radeon_bo_list *lobj;
 	struct radeon_bo *bo;
@@ -360,15 +362,17 @@ int radeon_bo_list_validate(struct list_head *head)
 	list_for_each_entry(lobj, head, tv.head) {
 		bo = lobj->bo;
 		if (!bo->pin_count) {
-			domain = lobj->wdomain ? lobj->wdomain : lobj->rdomain;
+			domain = lobj->domain;
 			
 		retry:
 			radeon_ttm_placement_from_domain(bo, domain);
+			if (ring == R600_RING_TYPE_UVD_INDEX)
+				radeon_uvd_force_into_uvd_segment(bo);
 			r = ttm_bo_validate(&bo->tbo, &bo->placement,
 						true, false);
 			if (unlikely(r)) {
-				if (r != -ERESTARTSYS && domain == RADEON_GEM_DOMAIN_VRAM) {
-					domain |= RADEON_GEM_DOMAIN_GTT;
+				if (r != -ERESTARTSYS && domain != lobj->alt_domain) {
+					domain = lobj->alt_domain;
 					goto retry;
 				}
 				return r;

drivers/gpu/drm/radeon/radeon_object.h

@@ -128,7 +128,7 @@ extern int radeon_bo_init(struct radeon_device *rdev);
 extern void radeon_bo_fini(struct radeon_device *rdev);
 extern void radeon_bo_list_add_object(struct radeon_bo_list *lobj,
 				struct list_head *head);
-extern int radeon_bo_list_validate(struct list_head *head);
+extern int radeon_bo_list_validate(struct list_head *head, int ring);
 extern int radeon_bo_fbdev_mmap(struct radeon_bo *bo,
 				struct vm_area_struct *vma);
 extern int radeon_bo_set_tiling_flags(struct radeon_bo *bo,

drivers/gpu/drm/radeon/radeon_pm.c

@@ -843,7 +843,11 @@ static int radeon_debugfs_pm_info(struct seq_file *m, void *data)
 	struct radeon_device *rdev = dev->dev_private;
 
 	seq_printf(m, "default engine clock: %u0 kHz\n", rdev->pm.default_sclk);
-	seq_printf(m, "current engine clock: %u0 kHz\n", radeon_get_engine_clock(rdev));
+	/* radeon_get_engine_clock is not reliable on APUs so just print the current clock */
+	if ((rdev->family >= CHIP_PALM) && (rdev->flags & RADEON_IS_IGP))
+		seq_printf(m, "current engine clock: %u0 kHz\n", rdev->pm.current_sclk);
+	else
+		seq_printf(m, "current engine clock: %u0 kHz\n", radeon_get_engine_clock(rdev));
 	seq_printf(m, "default memory clock: %u0 kHz\n", rdev->pm.default_mclk);
 	if (rdev->asic->pm.get_memory_clock)
 		seq_printf(m, "current memory clock: %u0 kHz\n", radeon_get_memory_clock(rdev));

drivers/gpu/drm/radeon/radeon_ring.c

@@ -180,7 +180,8 @@ int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib,
 		radeon_semaphore_free(rdev, &ib->semaphore, NULL);
 	}
 	/* if we can't remember our last VM flush then flush now! */
-	if (ib->vm && !ib->vm->last_flush) {
+	/* XXX figure out why we have to flush for every IB */
+	if (ib->vm /*&& !ib->vm->last_flush*/) {
 		radeon_ring_vm_flush(rdev, ib->ring, ib->vm);
 	}
 	if (const_ib) {
@@ -368,7 +369,7 @@ void radeon_ring_free_size(struct radeon_device *rdev, struct radeon_ring *ring)
 {
 	u32 rptr;
 
-	if (rdev->wb.enabled)
+	if (rdev->wb.enabled && ring != &rdev->ring[R600_RING_TYPE_UVD_INDEX])
 		rptr = le32_to_cpu(rdev->wb.wb[ring->rptr_offs/4]);
 	else
 		rptr = RREG32(ring->rptr_reg);
@@ -821,18 +822,20 @@ static int radeon_debugfs_ring_info(struct seq_file *m, void *data)
 	return 0;
 }
 
-static int radeon_ring_type_gfx_index = RADEON_RING_TYPE_GFX_INDEX;
-static int cayman_ring_type_cp1_index = CAYMAN_RING_TYPE_CP1_INDEX;
-static int cayman_ring_type_cp2_index = CAYMAN_RING_TYPE_CP2_INDEX;
-static int radeon_ring_type_dma1_index = R600_RING_TYPE_DMA_INDEX;
-static int radeon_ring_type_dma2_index = CAYMAN_RING_TYPE_DMA1_INDEX;
+static int radeon_gfx_index = RADEON_RING_TYPE_GFX_INDEX;
+static int cayman_cp1_index = CAYMAN_RING_TYPE_CP1_INDEX;
+static int cayman_cp2_index = CAYMAN_RING_TYPE_CP2_INDEX;
+static int radeon_dma1_index = R600_RING_TYPE_DMA_INDEX;
+static int radeon_dma2_index = CAYMAN_RING_TYPE_DMA1_INDEX;
+static int r600_uvd_index = R600_RING_TYPE_UVD_INDEX;
 
 static struct drm_info_list radeon_debugfs_ring_info_list[] = {
-	{"radeon_ring_gfx", radeon_debugfs_ring_info, 0, &radeon_ring_type_gfx_index},
-	{"radeon_ring_cp1", radeon_debugfs_ring_info, 0, &cayman_ring_type_cp1_index},
-	{"radeon_ring_cp2", radeon_debugfs_ring_info, 0, &cayman_ring_type_cp2_index},
-	{"radeon_ring_dma1", radeon_debugfs_ring_info, 0, &radeon_ring_type_dma1_index},
-	{"radeon_ring_dma2", radeon_debugfs_ring_info, 0, &radeon_ring_type_dma2_index},
+	{"radeon_ring_gfx", radeon_debugfs_ring_info, 0, &radeon_gfx_index},
+	{"radeon_ring_cp1", radeon_debugfs_ring_info, 0, &cayman_cp1_index},
+	{"radeon_ring_cp2", radeon_debugfs_ring_info, 0, &cayman_cp2_index},
+	{"radeon_ring_dma1", radeon_debugfs_ring_info, 0, &radeon_dma1_index},
+	{"radeon_ring_dma2", radeon_debugfs_ring_info, 0, &radeon_dma2_index},
+	{"radeon_ring_uvd", radeon_debugfs_ring_info, 0, &r600_uvd_index},
 };
 
 static int radeon_debugfs_sa_info(struct seq_file *m, void *data)

drivers/gpu/drm/radeon/radeon_test.c

@@ -252,6 +252,36 @@ void radeon_test_moves(struct radeon_device *rdev)
 		radeon_do_test_moves(rdev, RADEON_TEST_COPY_BLIT);
 }
 
+static int radeon_test_create_and_emit_fence(struct radeon_device *rdev,
+					     struct radeon_ring *ring,
+					     struct radeon_fence **fence)
+{
+	int r;
+
+	if (ring->idx == R600_RING_TYPE_UVD_INDEX) {
+		r = radeon_uvd_get_create_msg(rdev, ring->idx, 1, NULL);
+		if (r) {
+			DRM_ERROR("Failed to get dummy create msg\n");
+			return r;
+		}
+
+		r = radeon_uvd_get_destroy_msg(rdev, ring->idx, 1, fence);
+		if (r) {
+			DRM_ERROR("Failed to get dummy destroy msg\n");
+			return r;
+		}
+	} else {
+		r = radeon_ring_lock(rdev, ring, 64);
+		if (r) {
+			DRM_ERROR("Failed to lock ring A %d\n", ring->idx);
+			return r;
+		}
+		radeon_fence_emit(rdev, fence, ring->idx);
+		radeon_ring_unlock_commit(rdev, ring);
+	}
+	return 0;
+}
+
 void radeon_test_ring_sync(struct radeon_device *rdev,
 			   struct radeon_ring *ringA,
 			   struct radeon_ring *ringB)
@@ -272,21 +302,24 @@ void radeon_test_ring_sync(struct radeon_device *rdev,
 		goto out_cleanup;
 	}
 	radeon_semaphore_emit_wait(rdev, ringA->idx, semaphore);
-	r = radeon_fence_emit(rdev, &fence1, ringA->idx);
-	if (r) {
-		DRM_ERROR("Failed to emit fence 1\n");
-		radeon_ring_unlock_undo(rdev, ringA);
+	radeon_ring_unlock_commit(rdev, ringA);
+
+	r = radeon_test_create_and_emit_fence(rdev, ringA, &fence1);
+	if (r)
 		goto out_cleanup;
-	}
-	radeon_semaphore_emit_wait(rdev, ringA->idx, semaphore);
-	r = radeon_fence_emit(rdev, &fence2, ringA->idx);
+
+	r = radeon_ring_lock(rdev, ringA, 64);
 	if (r) {
-		DRM_ERROR("Failed to emit fence 2\n");
-		radeon_ring_unlock_undo(rdev, ringA);
+		DRM_ERROR("Failed to lock ring A %d\n", ringA->idx);
 		goto out_cleanup;
 	}
+	radeon_semaphore_emit_wait(rdev, ringA->idx, semaphore);
 	radeon_ring_unlock_commit(rdev, ringA);
 
+	r = radeon_test_create_and_emit_fence(rdev, ringA, &fence2);
+	if (r)
+		goto out_cleanup;
+
 	mdelay(1000);
 
 	if (radeon_fence_signaled(fence1)) {
@@ -364,27 +397,22 @@ static void radeon_test_ring_sync2(struct radeon_device *rdev,
 		goto out_cleanup;
 	}
 	radeon_semaphore_emit_wait(rdev, ringA->idx, semaphore);
-	r = radeon_fence_emit(rdev, &fenceA, ringA->idx);
-	if (r) {
-		DRM_ERROR("Failed to emit sync fence 1\n");
-		radeon_ring_unlock_undo(rdev, ringA);
-		goto out_cleanup;
-	}
 	radeon_ring_unlock_commit(rdev, ringA);
 
+	r = radeon_test_create_and_emit_fence(rdev, ringA, &fenceA);
+	if (r)
+		goto out_cleanup;
+
 	r = radeon_ring_lock(rdev, ringB, 64);
 	if (r) {
 		DRM_ERROR("Failed to lock ring B %d\n", ringB->idx);
 		goto out_cleanup;
 	}
 	radeon_semaphore_emit_wait(rdev, ringB->idx, semaphore);
-	r = radeon_fence_emit(rdev, &fenceB, ringB->idx);
-	if (r) {
-		DRM_ERROR("Failed to create sync fence 2\n");
-		radeon_ring_unlock_undo(rdev, ringB);
-		goto out_cleanup;
-	}
 	radeon_ring_unlock_commit(rdev, ringB);
+	r = radeon_test_create_and_emit_fence(rdev, ringB, &fenceB);
+	if (r)
+		goto out_cleanup;
 
 	mdelay(1000);
 
@@ -393,7 +421,7 @@ static void radeon_test_ring_sync2(struct radeon_device *rdev,
 		goto out_cleanup;
 	}
 	if (radeon_fence_signaled(fenceB)) {
-		DRM_ERROR("Fence A signaled without waiting for semaphore.\n");
+		DRM_ERROR("Fence B signaled without waiting for semaphore.\n");
 		goto out_cleanup;
 	}
 

drivers/gpu/drm/radeon/radeon_uvd.c

@@ -0,0 +1,666 @@
+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ */
+/*
+ * Authors:
+ *    Christian König <deathsimple@vodafone.de>
+ */
+
+#include <linux/firmware.h>
+#include <linux/module.h>
+#include <drm/drmP.h>
+#include <drm/drm.h>
+
+#include "radeon.h"
+#include "r600d.h"
+
+/* Firmware Names */
+#define FIRMWARE_RV710		"radeon/RV710_uvd.bin"
+#define FIRMWARE_CYPRESS	"radeon/CYPRESS_uvd.bin"
+#define FIRMWARE_SUMO		"radeon/SUMO_uvd.bin"
+#define FIRMWARE_TAHITI		"radeon/TAHITI_uvd.bin"
+
+MODULE_FIRMWARE(FIRMWARE_RV710);
+MODULE_FIRMWARE(FIRMWARE_CYPRESS);
+MODULE_FIRMWARE(FIRMWARE_SUMO);
+MODULE_FIRMWARE(FIRMWARE_TAHITI);
+
+int radeon_uvd_init(struct radeon_device *rdev)
+{
+	struct platform_device *pdev;
+	unsigned long bo_size;
+	const char *fw_name;
+	int i, r;
+
+	pdev = platform_device_register_simple("radeon_uvd", 0, NULL, 0);
+	r = IS_ERR(pdev);
+	if (r) {
+		dev_err(rdev->dev, "radeon_uvd: Failed to register firmware\n");
+		return -EINVAL;
+	}
+
+	switch (rdev->family) {
+	case CHIP_RV710:
+	case CHIP_RV730:
+	case CHIP_RV740:
+		fw_name = FIRMWARE_RV710;
+		break;
+
+	case CHIP_CYPRESS:
+	case CHIP_HEMLOCK:
+	case CHIP_JUNIPER:
+	case CHIP_REDWOOD:
+	case CHIP_CEDAR:
+		fw_name = FIRMWARE_CYPRESS;
+		break;
+
+	case CHIP_SUMO:
+	case CHIP_SUMO2:
+	case CHIP_PALM:
+	case CHIP_CAYMAN:
+	case CHIP_BARTS:
+	case CHIP_TURKS:
+	case CHIP_CAICOS:
+		fw_name = FIRMWARE_SUMO;
+		break;
+
+	case CHIP_TAHITI:
+	case CHIP_VERDE:
+	case CHIP_PITCAIRN:
+	case CHIP_ARUBA:
+		fw_name = FIRMWARE_TAHITI;
+		break;
+
+	default:
+		return -EINVAL;
+	}
+
+	r = request_firmware(&rdev->uvd_fw, fw_name, &pdev->dev);
+	if (r) {
+		dev_err(rdev->dev, "radeon_uvd: Can't load firmware \"%s\"\n",
+			fw_name);
+		platform_device_unregister(pdev);
+		return r;
+	}
+
+	platform_device_unregister(pdev);
+
+	bo_size = RADEON_GPU_PAGE_ALIGN(rdev->uvd_fw->size + 4) +
+		  RADEON_UVD_STACK_SIZE + RADEON_UVD_HEAP_SIZE;
+	r = radeon_bo_create(rdev, bo_size, PAGE_SIZE, true,
+			     RADEON_GEM_DOMAIN_VRAM, NULL, &rdev->uvd.vcpu_bo);
+	if (r) {
+		dev_err(rdev->dev, "(%d) failed to allocate UVD bo\n", r);
+		return r;
+	}
+
+	r = radeon_uvd_resume(rdev);
+	if (r)
+		return r;
+
+	memset(rdev->uvd.cpu_addr, 0, bo_size);
+	memcpy(rdev->uvd.cpu_addr, rdev->uvd_fw->data, rdev->uvd_fw->size);
+
+	r = radeon_uvd_suspend(rdev);
+	if (r)
+		return r;
+
+	for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i) {
+		atomic_set(&rdev->uvd.handles[i], 0);
+		rdev->uvd.filp[i] = NULL;
+	}
+
+	return 0;
+}
+
+void radeon_uvd_fini(struct radeon_device *rdev)
+{
+	radeon_uvd_suspend(rdev);
+	radeon_bo_unref(&rdev->uvd.vcpu_bo);
+}
+
+int radeon_uvd_suspend(struct radeon_device *rdev)
+{
+	int r;
+
+	if (rdev->uvd.vcpu_bo == NULL)
+		return 0;
+
+	r = radeon_bo_reserve(rdev->uvd.vcpu_bo, false);
+	if (!r) {
+		radeon_bo_kunmap(rdev->uvd.vcpu_bo);
+		radeon_bo_unpin(rdev->uvd.vcpu_bo);
+		radeon_bo_unreserve(rdev->uvd.vcpu_bo);
+	}
+	return r;
+}
+
+int radeon_uvd_resume(struct radeon_device *rdev)
+{
+	int r;
+
+	if (rdev->uvd.vcpu_bo == NULL)
+		return -EINVAL;
+
+	r = radeon_bo_reserve(rdev->uvd.vcpu_bo, false);
+	if (r) {
+		radeon_bo_unref(&rdev->uvd.vcpu_bo);
+		dev_err(rdev->dev, "(%d) failed to reserve UVD bo\n", r);
+		return r;
+	}
+
+	r = radeon_bo_pin(rdev->uvd.vcpu_bo, RADEON_GEM_DOMAIN_VRAM,
+			  &rdev->uvd.gpu_addr);
+	if (r) {
+		radeon_bo_unreserve(rdev->uvd.vcpu_bo);
+		radeon_bo_unref(&rdev->uvd.vcpu_bo);
+		dev_err(rdev->dev, "(%d) UVD bo pin failed\n", r);
+		return r;
+	}
+
+	r = radeon_bo_kmap(rdev->uvd.vcpu_bo, &rdev->uvd.cpu_addr);
+	if (r) {
+		dev_err(rdev->dev, "(%d) UVD map failed\n", r);
+		return r;
+	}
+
+	radeon_bo_unreserve(rdev->uvd.vcpu_bo);
+
+	radeon_set_uvd_clocks(rdev, 53300, 40000);
+
+	return 0;
+}
+
+void radeon_uvd_force_into_uvd_segment(struct radeon_bo *rbo)
+{
+	rbo->placement.fpfn = 0 >> PAGE_SHIFT;
+	rbo->placement.lpfn = (256 * 1024 * 1024) >> PAGE_SHIFT;
+}
+
+void radeon_uvd_free_handles(struct radeon_device *rdev, struct drm_file *filp)
+{
+	int i, r;
+	for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i) {
+		if (rdev->uvd.filp[i] == filp) {
+			uint32_t handle = atomic_read(&rdev->uvd.handles[i]);
+			struct radeon_fence *fence;
+
+			r = radeon_uvd_get_destroy_msg(rdev,
+				R600_RING_TYPE_UVD_INDEX, handle, &fence);
+			if (r) {
+				DRM_ERROR("Error destroying UVD (%d)!\n", r);
+				continue;
+			}
+
+			radeon_fence_wait(fence, false);
+			radeon_fence_unref(&fence);
+
+			rdev->uvd.filp[i] = NULL;
+			atomic_set(&rdev->uvd.handles[i], 0);
+		}
+	}
+}
+
+static int radeon_uvd_cs_msg_decode(uint32_t *msg, unsigned buf_sizes[])
+{
+	unsigned stream_type = msg[4];
+	unsigned width = msg[6];
+	unsigned height = msg[7];
+	unsigned dpb_size = msg[9];
+	unsigned pitch = msg[28];
+
+	unsigned width_in_mb = width / 16;
+	unsigned height_in_mb = ALIGN(height / 16, 2);
+
+	unsigned image_size, tmp, min_dpb_size;
+
+	image_size = width * height;
+	image_size += image_size / 2;
+	image_size = ALIGN(image_size, 1024);
+
+	switch (stream_type) {
+	case 0: /* H264 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 17;
+
+		/* macroblock context buffer */
+		min_dpb_size += width_in_mb * height_in_mb * 17 * 192;
+
+		/* IT surface buffer */
+		min_dpb_size += width_in_mb * height_in_mb * 32;
+		break;
+
+	case 1: /* VC1 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+
+		/* CONTEXT_BUFFER */
+		min_dpb_size += width_in_mb * height_in_mb * 128;
+
+		/* IT surface buffer */
+		min_dpb_size += width_in_mb * 64;
+
+		/* DB surface buffer */
+		min_dpb_size += width_in_mb * 128;
+
+		/* BP */
+		tmp = max(width_in_mb, height_in_mb);
+		min_dpb_size += ALIGN(tmp * 7 * 16, 64);
+		break;
+
+	case 3: /* MPEG2 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+		break;
+
+	case 4: /* MPEG4 */
+
+		/* reference picture buffer */
+		min_dpb_size = image_size * 3;
+
+		/* CM */
+		min_dpb_size += width_in_mb * height_in_mb * 64;
+
+		/* IT surface buffer */
+		min_dpb_size += ALIGN(width_in_mb * height_in_mb * 32, 64);
+		break;
+
+	default:
+		DRM_ERROR("UVD codec not handled %d!\n", stream_type);
+		return -EINVAL;
+	}
+
+	if (width > pitch) {
+		DRM_ERROR("Invalid UVD decoding target pitch!\n");
+		return -EINVAL;
+	}
+
+	if (dpb_size < min_dpb_size) {
+		DRM_ERROR("Invalid dpb_size in UVD message (%d / %d)!\n",
+			  dpb_size, min_dpb_size);
+		return -EINVAL;
+	}
+
+	buf_sizes[0x1] = dpb_size;
+	buf_sizes[0x2] = image_size;
+	return 0;
+}
+
+static int radeon_uvd_cs_msg(struct radeon_cs_parser *p, struct radeon_bo *bo,
+			     unsigned offset, unsigned buf_sizes[])
+{
+	int32_t *msg, msg_type, handle;
+	void *ptr;
+
+	int i, r;
+
+	if (offset & 0x3F) {
+		DRM_ERROR("UVD messages must be 64 byte aligned!\n");
+		return -EINVAL;
+	}
+
+	r = radeon_bo_kmap(bo, &ptr);
+	if (r)
+		return r;
+
+	msg = ptr + offset;
+
+	msg_type = msg[1];
+	handle = msg[2];
+
+	if (handle == 0) {
+		DRM_ERROR("Invalid UVD handle!\n");
+		return -EINVAL;
+	}
+
+	if (msg_type == 1) {
+		/* it's a decode msg, calc buffer sizes */
+		r = radeon_uvd_cs_msg_decode(msg, buf_sizes);
+		radeon_bo_kunmap(bo);
+		if (r)
+			return r;
+
+	} else if (msg_type == 2) {
+		/* it's a destroy msg, free the handle */
+		for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i)
+			atomic_cmpxchg(&p->rdev->uvd.handles[i], handle, 0);
+		radeon_bo_kunmap(bo);
+		return 0;
+	} else {
+		/* it's a create msg, no special handling needed */
+		radeon_bo_kunmap(bo);
+	}
+
+	/* create or decode, validate the handle */
+	for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i) {
+		if (atomic_read(&p->rdev->uvd.handles[i]) == handle)
+			return 0;
+	}
+
+	/* handle not found try to alloc a new one */
+	for (i = 0; i < RADEON_MAX_UVD_HANDLES; ++i) {
+		if (!atomic_cmpxchg(&p->rdev->uvd.handles[i], 0, handle)) {
+			p->rdev->uvd.filp[i] = p->filp;
+			return 0;
+		}
+	}
+
+	DRM_ERROR("No more free UVD handles!\n");
+	return -EINVAL;
+}
+
+static int radeon_uvd_cs_reloc(struct radeon_cs_parser *p,
+			       int data0, int data1,
+			       unsigned buf_sizes[])
+{
+	struct radeon_cs_chunk *relocs_chunk;
+	struct radeon_cs_reloc *reloc;
+	unsigned idx, cmd, offset;
+	uint64_t start, end;
+	int r;
+
+	relocs_chunk = &p->chunks[p->chunk_relocs_idx];
+	offset = radeon_get_ib_value(p, data0);
+	idx = radeon_get_ib_value(p, data1);
+	if (idx >= relocs_chunk->length_dw) {
+		DRM_ERROR("Relocs at %d after relocations chunk end %d !\n",
+			  idx, relocs_chunk->length_dw);
+		return -EINVAL;
+	}
+
+	reloc = p->relocs_ptr[(idx / 4)];
+	start = reloc->lobj.gpu_offset;
+	end = start + radeon_bo_size(reloc->robj);
+	start += offset;
+
+	p->ib.ptr[data0] = start & 0xFFFFFFFF;
+	p->ib.ptr[data1] = start >> 32;
+
+	cmd = radeon_get_ib_value(p, p->idx) >> 1;
+
+	if (cmd < 0x4) {
+		if ((end - start) < buf_sizes[cmd]) {
+			DRM_ERROR("buffer to small (%d / %d)!\n",
+				  (unsigned)(end - start), buf_sizes[cmd]);
+			return -EINVAL;
+		}
+
+	} else if (cmd != 0x100) {
+		DRM_ERROR("invalid UVD command %X!\n", cmd);
+		return -EINVAL;
+	}
+
+	if (cmd == 0) {
+		if (end & 0xFFFFFFFFF0000000) {
+			DRM_ERROR("msg buffer %LX-%LX out of 256MB segment!\n",
+				  start, end);
+			return -EINVAL;
+		}
+
+		r = radeon_uvd_cs_msg(p, reloc->robj, offset, buf_sizes);
+		if (r)
+			return r;
+	}
+
+	if ((start & 0xFFFFFFFFF0000000) != (end & 0xFFFFFFFFF0000000)) {
+		DRM_ERROR("reloc %LX-%LX crossing 256MB boundary!\n",
+			  start, end);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int radeon_uvd_cs_reg(struct radeon_cs_parser *p,
+			     struct radeon_cs_packet *pkt,
+			     int *data0, int *data1,
+			     unsigned buf_sizes[])
+{
+	int i, r;
+
+	p->idx++;
+	for (i = 0; i <= pkt->count; ++i) {
+		switch (pkt->reg + i*4) {
+		case UVD_GPCOM_VCPU_DATA0:
+			*data0 = p->idx;
+			break;
+		case UVD_GPCOM_VCPU_DATA1:
+			*data1 = p->idx;
+			break;
+		case UVD_GPCOM_VCPU_CMD:
+			r = radeon_uvd_cs_reloc(p, *data0, *data1, buf_sizes);
+			if (r)
+				return r;
+			break;
+		case UVD_ENGINE_CNTL:
+			break;
+		default:
+			DRM_ERROR("Invalid reg 0x%X!\n",
+				  pkt->reg + i*4);
+			return -EINVAL;
+		}
+		p->idx++;
+	}
+	return 0;
+}
+
+int radeon_uvd_cs_parse(struct radeon_cs_parser *p)
+{
+	struct radeon_cs_packet pkt;
+	int r, data0 = 0, data1 = 0;
+
+	/* minimum buffer sizes */
+	unsigned buf_sizes[] = {
+		[0x00000000]	=	2048,
+		[0x00000001]	=	32 * 1024 * 1024,
+		[0x00000002]	=	2048 * 1152 * 3,
+		[0x00000003]	=	2048,
+	};
+
+	if (p->chunks[p->chunk_ib_idx].length_dw % 16) {
+		DRM_ERROR("UVD IB length (%d) not 16 dwords aligned!\n",
+			  p->chunks[p->chunk_ib_idx].length_dw);
+		return -EINVAL;
+	}
+
+	if (p->chunk_relocs_idx == -1) {
+		DRM_ERROR("No relocation chunk !\n");
+		return -EINVAL;
+	}
+
+
+	do {
+		r = radeon_cs_packet_parse(p, &pkt, p->idx);
+		if (r)
+			return r;
+		switch (pkt.type) {
+		case RADEON_PACKET_TYPE0:
+			r = radeon_uvd_cs_reg(p, &pkt, &data0,
+					      &data1, buf_sizes);
+			if (r)
+				return r;
+			break;
+		case RADEON_PACKET_TYPE2:
+			p->idx += pkt.count + 2;
+			break;
+		default:
+			DRM_ERROR("Unknown packet type %d !\n", pkt.type);
+			return -EINVAL;
+		}
+	} while (p->idx < p->chunks[p->chunk_ib_idx].length_dw);
+	return 0;
+}
+
+static int radeon_uvd_send_msg(struct radeon_device *rdev,
+			       int ring, struct radeon_bo *bo,
+			       struct radeon_fence **fence)
+{
+	struct ttm_validate_buffer tv;
+	struct list_head head;
+	struct radeon_ib ib;
+	uint64_t addr;
+	int i, r;
+
+	memset(&tv, 0, sizeof(tv));
+	tv.bo = &bo->tbo;
+
+	INIT_LIST_HEAD(&head);
+	list_add(&tv.head, &head);
+
+	r = ttm_eu_reserve_buffers(&head);
+	if (r)
+		return r;
+
+	radeon_ttm_placement_from_domain(bo, RADEON_GEM_DOMAIN_VRAM);
+	radeon_uvd_force_into_uvd_segment(bo);
+
+	r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
+	if (r) {
+		ttm_eu_backoff_reservation(&head);
+		return r;
+	}
+
+	r = radeon_ib_get(rdev, ring, &ib, NULL, 16);
+	if (r) {
+		ttm_eu_backoff_reservation(&head);
+		return r;
+	}
+
+	addr = radeon_bo_gpu_offset(bo);
+	ib.ptr[0] = PACKET0(UVD_GPCOM_VCPU_DATA0, 0);
+	ib.ptr[1] = addr;
+	ib.ptr[2] = PACKET0(UVD_GPCOM_VCPU_DATA1, 0);
+	ib.ptr[3] = addr >> 32;
+	ib.ptr[4] = PACKET0(UVD_GPCOM_VCPU_CMD, 0);
+	ib.ptr[5] = 0;
+	for (i = 6; i < 16; ++i)
+		ib.ptr[i] = PACKET2(0);
+	ib.length_dw = 16;
+
+	r = radeon_ib_schedule(rdev, &ib, NULL);
+	if (r) {
+		ttm_eu_backoff_reservation(&head);
+		return r;
+	}
+	ttm_eu_fence_buffer_objects(&head, ib.fence);
+
+	if (fence)
+		*fence = radeon_fence_ref(ib.fence);
+
+	radeon_ib_free(rdev, &ib);
+	radeon_bo_unref(&bo);
+	return 0;
+}
+
+/* multiple fence commands without any stream commands in between can
+   crash the vcpu so just try to emmit a dummy create/destroy msg to
+   avoid this */
+int radeon_uvd_get_create_msg(struct radeon_device *rdev, int ring,
+			      uint32_t handle, struct radeon_fence **fence)
+{
+	struct radeon_bo *bo;
+	uint32_t *msg;
+	int r, i;
+
+	r = radeon_bo_create(rdev, 1024, PAGE_SIZE, true,
+			     RADEON_GEM_DOMAIN_VRAM, NULL, &bo);
+	if (r)
+		return r;
+
+	r = radeon_bo_reserve(bo, false);
+	if (r) {
+		radeon_bo_unref(&bo);
+		return r;
+	}
+
+	r = radeon_bo_kmap(bo, (void **)&msg);
+	if (r) {
+		radeon_bo_unreserve(bo);
+		radeon_bo_unref(&bo);
+		return r;
+	}
+
+	/* stitch together an UVD create msg */
+	msg[0] = 0x00000de4;
+	msg[1] = 0x00000000;
+	msg[2] = handle;
+	msg[3] = 0x00000000;
+	msg[4] = 0x00000000;
+	msg[5] = 0x00000000;
+	msg[6] = 0x00000000;
+	msg[7] = 0x00000780;
+	msg[8] = 0x00000440;
+	msg[9] = 0x00000000;
+	msg[10] = 0x01b37000;
+	for (i = 11; i < 1024; ++i)
+		msg[i] = 0x0;
+
+	radeon_bo_kunmap(bo);
+	radeon_bo_unreserve(bo);
+
+	return radeon_uvd_send_msg(rdev, ring, bo, fence);
+}
+
+int radeon_uvd_get_destroy_msg(struct radeon_device *rdev, int ring,
+			       uint32_t handle, struct radeon_fence **fence)
+{
+	struct radeon_bo *bo;
+	uint32_t *msg;
+	int r, i;
+
+	r = radeon_bo_create(rdev, 1024, PAGE_SIZE, true,
+			     RADEON_GEM_DOMAIN_VRAM, NULL, &bo);
+	if (r)
+		return r;
+
+	r = radeon_bo_reserve(bo, false);
+	if (r) {
+		radeon_bo_unref(&bo);
+		return r;
+	}
+
+	r = radeon_bo_kmap(bo, (void **)&msg);
+	if (r) {
+		radeon_bo_unreserve(bo);
+		radeon_bo_unref(&bo);
+		return r;
+	}
+
+	/* stitch together an UVD destroy msg */
+	msg[0] = 0x00000de4;
+	msg[1] = 0x00000002;
+	msg[2] = handle;
+	msg[3] = 0x00000000;
+	for (i = 4; i < 1024; ++i)
+		msg[i] = 0x0;
+
+	radeon_bo_kunmap(bo);
+	radeon_bo_unreserve(bo);
+
+	return radeon_uvd_send_msg(rdev, ring, bo, fence);
+}

drivers/gpu/drm/radeon/rs690.c

@@ -148,6 +148,8 @@ void rs690_pm_info(struct radeon_device *rdev)
 static void rs690_mc_init(struct radeon_device *rdev)
 {
 	u64 base;
+	uint32_t h_addr, l_addr;
+	unsigned long long k8_addr;
 
 	rs400_gart_adjust_size(rdev);
 	rdev->mc.vram_is_ddr = true;
@@ -160,6 +162,27 @@ static void rs690_mc_init(struct radeon_device *rdev)
 	base = RREG32_MC(R_000100_MCCFG_FB_LOCATION);
 	base = G_000100_MC_FB_START(base) << 16;
 	rdev->mc.igp_sideport_enabled = radeon_atombios_sideport_present(rdev);
+
+	/* Use K8 direct mapping for fast fb access. */ 
+	rdev->fastfb_working = false;
+	h_addr = G_00005F_K8_ADDR_EXT(RREG32_MC(R_00005F_MC_MISC_UMA_CNTL));
+	l_addr = RREG32_MC(R_00001E_K8_FB_LOCATION);
+	k8_addr = ((unsigned long long)h_addr) << 32 | l_addr;
+#if defined(CONFIG_X86_32) && !defined(CONFIG_X86_PAE)
+	if (k8_addr + rdev->mc.visible_vram_size < 0x100000000ULL)	
+#endif
+	{
+		/* FastFB shall be used with UMA memory. Here it is simply disabled when sideport 
+		 * memory is present.
+		 */
+		if (rdev->mc.igp_sideport_enabled == false && radeon_fastfb == 1) {
+			DRM_INFO("Direct mapping: aper base at 0x%llx, replaced by direct mapping base 0x%llx.\n", 
+					(unsigned long long)rdev->mc.aper_base, k8_addr);
+			rdev->mc.aper_base = (resource_size_t)k8_addr;
+			rdev->fastfb_working = true;
+		}
+	}  
+
 	rs690_pm_info(rdev);
 	radeon_vram_location(rdev, &rdev->mc, base);
 	rdev->mc.gtt_base_align = rdev->mc.gtt_size - 1;

drivers/gpu/drm/radeon/rs690d.h

@@ -29,6 +29,9 @@
 #define __RS690D_H__
 
 /* Registers */
+#define R_00001E_K8_FB_LOCATION                      0x00001E
+#define R_00005F_MC_MISC_UMA_CNTL                    0x00005F
+#define   G_00005F_K8_ADDR_EXT(x)                      (((x) >> 0) & 0xFF)
 #define R_000078_MC_INDEX                            0x000078
 #define   S_000078_MC_IND_ADDR(x)                      (((x) & 0x1FF) << 0)
 #define   G_000078_MC_IND_ADDR(x)                      (((x) >> 0) & 0x1FF)

drivers/gpu/drm/radeon/rv515.c

@@ -348,7 +348,7 @@ void rv515_mc_resume(struct radeon_device *rdev, struct rv515_mc_save *save)
 	/* update crtc base addresses */
 	for (i = 0; i < rdev->num_crtc; i++) {
 		if (rdev->family >= CHIP_RV770) {
-			if (i == 1) {
+			if (i == 0) {
 				WREG32(R700_D1GRPH_PRIMARY_SURFACE_ADDRESS_HIGH,
 				       upper_32_bits(rdev->mc.vram_start));
 				WREG32(R700_D1GRPH_SECONDARY_SURFACE_ADDRESS_HIGH,

drivers/gpu/drm/radeon/rv770.c

@@ -42,6 +42,162 @@
 static void rv770_gpu_init(struct radeon_device *rdev);
 void rv770_fini(struct radeon_device *rdev);
 static void rv770_pcie_gen2_enable(struct radeon_device *rdev);
+int evergreen_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk);
+
+static int rv770_uvd_calc_post_div(unsigned target_freq,
+				   unsigned vco_freq,
+				   unsigned *div)
+{
+	/* Fclk = Fvco / PDIV */
+	*div = vco_freq / target_freq;
+
+	/* we alway need a frequency less than or equal the target */
+	if ((vco_freq / *div) > target_freq)
+		*div += 1;
+
+	/* out of range ? */
+	if (*div > 30)
+		return -1; /* forget it */
+
+	*div -= 1;
+	return vco_freq / (*div + 1);
+}
+
+static int rv770_uvd_send_upll_ctlreq(struct radeon_device *rdev)
+{
+	unsigned i;
+
+	/* assert UPLL_CTLREQ */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_CTLREQ_MASK, ~UPLL_CTLREQ_MASK);
+
+	/* wait for CTLACK and CTLACK2 to get asserted */
+	for (i = 0; i < 100; ++i) {
+		uint32_t mask = UPLL_CTLACK_MASK | UPLL_CTLACK2_MASK;
+		if ((RREG32(CG_UPLL_FUNC_CNTL) & mask) == mask)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	/* deassert UPLL_CTLREQ */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_CTLREQ_MASK);
+
+	return 0;
+}
+
+int rv770_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk)
+{
+	/* start off with something large */
+	int optimal_diff_score = 0x7FFFFFF;
+	unsigned optimal_fb_div = 0, optimal_vclk_div = 0;
+	unsigned optimal_dclk_div = 0, optimal_vco_freq = 0;
+	unsigned vco_freq, vco_min = 50000, vco_max = 160000;
+	unsigned ref_freq = rdev->clock.spll.reference_freq;
+	int r;
+
+	/* RV740 uses evergreen uvd clk programming */
+	if (rdev->family == CHIP_RV740)
+		return evergreen_set_uvd_clocks(rdev, vclk, dclk);
+
+	/* loop through vco from low to high */
+	vco_min = max(max(vco_min, vclk), dclk);
+	for (vco_freq = vco_min; vco_freq <= vco_max; vco_freq += 500) {
+		uint64_t fb_div = (uint64_t)vco_freq * 43663;
+		int calc_clk, diff_score, diff_vclk, diff_dclk;
+		unsigned vclk_div, dclk_div;
+
+		do_div(fb_div, ref_freq);
+		fb_div |= 1;
+
+		/* fb div out of range ? */
+		if (fb_div > 0x03FFFFFF)
+			break; /* it can oly get worse */
+
+		/* calc vclk with current vco freq. */
+		calc_clk = rv770_uvd_calc_post_div(vclk, vco_freq, &vclk_div);
+		if (calc_clk == -1)
+			break; /* vco is too big, it has to stop. */
+		diff_vclk = vclk - calc_clk;
+
+		/* calc dclk with current vco freq. */
+		calc_clk = rv770_uvd_calc_post_div(dclk, vco_freq, &dclk_div);
+		if (calc_clk == -1)
+			break; /* vco is too big, it has to stop. */
+		diff_dclk = dclk - calc_clk;
+
+		/* determine if this vco setting is better than current optimal settings */
+		diff_score = abs(diff_vclk) + abs(diff_dclk);
+		if (diff_score < optimal_diff_score) {
+			optimal_fb_div = fb_div;
+			optimal_vclk_div = vclk_div;
+			optimal_dclk_div = dclk_div;
+			optimal_vco_freq = vco_freq;
+			optimal_diff_score = diff_score;
+			if (optimal_diff_score == 0)
+				break; /* it can't get better than this */
+		}
+	}
+
+	/* bypass vclk and dclk with bclk */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		 VCLK_SRC_SEL(1) | DCLK_SRC_SEL(1),
+		 ~(VCLK_SRC_SEL_MASK | DCLK_SRC_SEL_MASK));
+
+	/* set UPLL_FB_DIV to 0x50000 */
+	WREG32_P(CG_UPLL_FUNC_CNTL_3, UPLL_FB_DIV(0x50000), ~UPLL_FB_DIV_MASK);
+
+	/* deassert UPLL_RESET */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_RESET_MASK);
+
+	/* assert BYPASS EN and FB_DIV[0] <- ??? why? */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_BYPASS_EN_MASK, ~UPLL_BYPASS_EN_MASK);
+	WREG32_P(CG_UPLL_FUNC_CNTL_3, UPLL_FB_DIV(1), ~UPLL_FB_DIV(1));
+
+	r = rv770_uvd_send_upll_ctlreq(rdev);
+	if (r)
+		return r;
+
+	/* assert PLL_RESET */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_RESET_MASK, ~UPLL_RESET_MASK);
+
+	/* set the required FB_DIV, REF_DIV, Post divder values */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_REF_DIV(1), ~UPLL_REF_DIV_MASK);
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		 UPLL_SW_HILEN(optimal_vclk_div >> 1) |
+		 UPLL_SW_LOLEN((optimal_vclk_div >> 1) + (optimal_vclk_div & 1)) |
+		 UPLL_SW_HILEN2(optimal_dclk_div >> 1) |
+		 UPLL_SW_LOLEN2((optimal_dclk_div >> 1) + (optimal_dclk_div & 1)),
+		 ~UPLL_SW_MASK);
+
+	WREG32_P(CG_UPLL_FUNC_CNTL_3, UPLL_FB_DIV(optimal_fb_div),
+		 ~UPLL_FB_DIV_MASK);
+
+	/* give the PLL some time to settle */
+	mdelay(15);
+
+	/* deassert PLL_RESET */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_RESET_MASK);
+
+	mdelay(15);
+
+	/* deassert BYPASS EN and FB_DIV[0] <- ??? why? */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_BYPASS_EN_MASK);
+	WREG32_P(CG_UPLL_FUNC_CNTL_3, 0, ~UPLL_FB_DIV(1));
+
+	r = rv770_uvd_send_upll_ctlreq(rdev);
+	if (r)
+		return r;
+
+	/* switch VCLK and DCLK selection */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		 VCLK_SRC_SEL(2) | DCLK_SRC_SEL(2),
+		 ~(VCLK_SRC_SEL_MASK | DCLK_SRC_SEL_MASK));
+
+	mdelay(100);
+
+	return 0;
+}
 
 #define PCIE_BUS_CLK                10000
 #define TCLK                        (PCIE_BUS_CLK / 10)
@@ -68,6 +224,105 @@ u32 rv770_get_xclk(struct radeon_device *rdev)
 	return reference_clock;
 }
 
+int rv770_uvd_resume(struct radeon_device *rdev)
+{
+	uint64_t addr;
+	uint32_t chip_id, size;
+	int r;
+
+	r = radeon_uvd_resume(rdev);
+	if (r)
+		return r;
+
+	/* programm the VCPU memory controller bits 0-27 */
+	addr = rdev->uvd.gpu_addr >> 3;
+	size = RADEON_GPU_PAGE_ALIGN(rdev->uvd_fw->size + 4) >> 3;
+	WREG32(UVD_VCPU_CACHE_OFFSET0, addr);
+	WREG32(UVD_VCPU_CACHE_SIZE0, size);
+
+	addr += size;
+	size = RADEON_UVD_STACK_SIZE >> 3;
+	WREG32(UVD_VCPU_CACHE_OFFSET1, addr);
+	WREG32(UVD_VCPU_CACHE_SIZE1, size);
+
+	addr += size;
+	size = RADEON_UVD_HEAP_SIZE >> 3;
+	WREG32(UVD_VCPU_CACHE_OFFSET2, addr);
+	WREG32(UVD_VCPU_CACHE_SIZE2, size);
+
+	/* bits 28-31 */
+	addr = (rdev->uvd.gpu_addr >> 28) & 0xF;
+	WREG32(UVD_LMI_ADDR_EXT, (addr << 12) | (addr << 0));
+
+	/* bits 32-39 */
+	addr = (rdev->uvd.gpu_addr >> 32) & 0xFF;
+	WREG32(UVD_LMI_EXT40_ADDR, addr | (0x9 << 16) | (0x1 << 31));
+
+	/* tell firmware which hardware it is running on */
+	switch (rdev->family) {
+	default:
+		return -EINVAL;
+	case CHIP_RV710:
+		chip_id = 0x01000005;
+		break;
+	case CHIP_RV730:
+		chip_id = 0x01000006;
+		break;
+	case CHIP_RV740:
+		chip_id = 0x01000007;
+		break;
+	case CHIP_CYPRESS:
+	case CHIP_HEMLOCK:
+		chip_id = 0x01000008;
+		break;
+	case CHIP_JUNIPER:
+		chip_id = 0x01000009;
+		break;
+	case CHIP_REDWOOD:
+		chip_id = 0x0100000a;
+		break;
+	case CHIP_CEDAR:
+		chip_id = 0x0100000b;
+		break;
+	case CHIP_SUMO:
+		chip_id = 0x0100000c;
+		break;
+	case CHIP_SUMO2:
+		chip_id = 0x0100000d;
+		break;
+	case CHIP_PALM:
+		chip_id = 0x0100000e;
+		break;
+	case CHIP_CAYMAN:
+		chip_id = 0x0100000f;
+		break;
+	case CHIP_BARTS:
+		chip_id = 0x01000010;
+		break;
+	case CHIP_TURKS:
+		chip_id = 0x01000011;
+		break;
+	case CHIP_CAICOS:
+		chip_id = 0x01000012;
+		break;
+	case CHIP_TAHITI:
+		chip_id = 0x01000014;
+		break;
+	case CHIP_VERDE:
+		chip_id = 0x01000015;
+		break;
+	case CHIP_PITCAIRN:
+		chip_id = 0x01000016;
+		break;
+	case CHIP_ARUBA:
+		chip_id = 0x01000017;
+		break;
+	}
+	WREG32(UVD_VCPU_CHIP_ID, chip_id);
+
+	return 0;
+}
+
 u32 rv770_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base)
 {
 	struct radeon_crtc *radeon_crtc = rdev->mode_info.crtcs[crtc_id];
@@ -611,6 +866,11 @@ static void rv770_gpu_init(struct radeon_device *rdev)
 	WREG32(HDP_TILING_CONFIG, (gb_tiling_config & 0xffff));
 	WREG32(DMA_TILING_CONFIG, (gb_tiling_config & 0xffff));
 	WREG32(DMA_TILING_CONFIG2, (gb_tiling_config & 0xffff));
+	if (rdev->family == CHIP_RV730) {
+		WREG32(UVD_UDEC_DB_TILING_CONFIG, (gb_tiling_config & 0xffff));
+		WREG32(UVD_UDEC_DBW_TILING_CONFIG, (gb_tiling_config & 0xffff));
+		WREG32(UVD_UDEC_TILING_CONFIG, (gb_tiling_config & 0xffff));
+	}
 
 	WREG32(CGTS_SYS_TCC_DISABLE, 0);
 	WREG32(CGTS_TCC_DISABLE, 0);
@@ -840,7 +1100,7 @@ void r700_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc)
 	}
 	if (rdev->flags & RADEON_IS_AGP) {
 		size_bf = mc->gtt_start;
-		size_af = 0xFFFFFFFF - mc->gtt_end;
+		size_af = mc->mc_mask - mc->gtt_end;
 		if (size_bf > size_af) {
 			if (mc->mc_vram_size > size_bf) {
 				dev_warn(rdev->dev, "limiting VRAM\n");
@@ -1040,6 +1300,17 @@ static int rv770_startup(struct radeon_device *rdev)
 		return r;
 	}
 
+	r = rv770_uvd_resume(rdev);
+	if (!r) {
+		r = radeon_fence_driver_start_ring(rdev,
+						   R600_RING_TYPE_UVD_INDEX);
+		if (r)
+			dev_err(rdev->dev, "UVD fences init error (%d).\n", r);
+	}
+
+	if (r)
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_size = 0;
+
 	/* Enable IRQ */
 	r = r600_irq_init(rdev);
 	if (r) {
@@ -1074,6 +1345,19 @@ static int rv770_startup(struct radeon_device *rdev)
 	if (r)
 		return r;
 
+	ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	if (ring->ring_size) {
+		r = radeon_ring_init(rdev, ring, ring->ring_size,
+				     R600_WB_UVD_RPTR_OFFSET,
+				     UVD_RBC_RB_RPTR, UVD_RBC_RB_WPTR,
+				     0, 0xfffff, RADEON_CP_PACKET2);
+		if (!r)
+			r = r600_uvd_init(rdev);
+
+		if (r)
+			DRM_ERROR("radeon: failed initializing UVD (%d).\n", r);
+	}
+
 	r = radeon_ib_pool_init(rdev);
 	if (r) {
 		dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
@@ -1115,6 +1399,7 @@ int rv770_resume(struct radeon_device *rdev)
 int rv770_suspend(struct radeon_device *rdev)
 {
 	r600_audio_fini(rdev);
+	radeon_uvd_suspend(rdev);
 	r700_cp_stop(rdev);
 	r600_dma_stop(rdev);
 	r600_irq_suspend(rdev);
@@ -1190,6 +1475,13 @@ int rv770_init(struct radeon_device *rdev)
 	rdev->ring[R600_RING_TYPE_DMA_INDEX].ring_obj = NULL;
 	r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX], 64 * 1024);
 
+	r = radeon_uvd_init(rdev);
+	if (!r) {
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_obj = NULL;
+		r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_UVD_INDEX],
+			       4096);
+	}
+
 	rdev->ih.ring_obj = NULL;
 	r600_ih_ring_init(rdev, 64 * 1024);
 
@@ -1224,6 +1516,7 @@ void rv770_fini(struct radeon_device *rdev)
 	radeon_ib_pool_fini(rdev);
 	radeon_irq_kms_fini(rdev);
 	rv770_pcie_gart_fini(rdev);
+	radeon_uvd_fini(rdev);
 	r600_vram_scratch_fini(rdev);
 	radeon_gem_fini(rdev);
 	radeon_fence_driver_fini(rdev);
@@ -1264,23 +1557,23 @@ static void rv770_pcie_gen2_enable(struct radeon_device *rdev)
 	DRM_INFO("enabling PCIE gen 2 link speeds, disable with radeon.pcie_gen2=0\n");
 
 	/* advertise upconfig capability */
-	link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
+	link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
 	link_width_cntl &= ~LC_UPCONFIGURE_DIS;
-	WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
-	link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
+	WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+	link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
 	if (link_width_cntl & LC_RENEGOTIATION_SUPPORT) {
 		lanes = (link_width_cntl & LC_LINK_WIDTH_RD_MASK) >> LC_LINK_WIDTH_RD_SHIFT;
 		link_width_cntl &= ~(LC_LINK_WIDTH_MASK |
 				     LC_RECONFIG_ARC_MISSING_ESCAPE);
 		link_width_cntl |= lanes | LC_RECONFIG_NOW |
 			LC_RENEGOTIATE_EN | LC_UPCONFIGURE_SUPPORT;
-		WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
 	} else {
 		link_width_cntl |= LC_UPCONFIGURE_DIS;
-		WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
 	}
 
-	speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+	speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 	if ((speed_cntl & LC_OTHER_SIDE_EVER_SENT_GEN2) &&
 	    (speed_cntl & LC_OTHER_SIDE_SUPPORTS_GEN2)) {
 
@@ -1293,29 +1586,29 @@ static void rv770_pcie_gen2_enable(struct radeon_device *rdev)
 		WREG16(0x4088, link_cntl2);
 		WREG32(MM_CFGREGS_CNTL, 0);
 
-		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 		speed_cntl &= ~LC_TARGET_LINK_SPEED_OVERRIDE_EN;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
-		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 		speed_cntl |= LC_CLR_FAILED_SPD_CHANGE_CNT;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
-		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 		speed_cntl &= ~LC_CLR_FAILED_SPD_CHANGE_CNT;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
-		speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL);
+		speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL);
 		speed_cntl |= LC_GEN2_EN_STRAP;
-		WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl);
 
 	} else {
-		link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL);
+		link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL);
 		/* XXX: only disable it if gen1 bridge vendor == 0x111d or 0x1106 */
 		if (1)
 			link_width_cntl |= LC_UPCONFIGURE_DIS;
 		else
 			link_width_cntl &= ~LC_UPCONFIGURE_DIS;
-		WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
+		WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
 	}
 }

drivers/gpu/drm/radeon/rv770d.h

@@ -38,6 +38,30 @@
 #define R7XX_MAX_PIPES             8
 #define R7XX_MAX_PIPES_MASK        0xff
 
+/* discrete uvd clocks */
+#define CG_UPLL_FUNC_CNTL				0x718
+#	define UPLL_RESET_MASK				0x00000001
+#	define UPLL_SLEEP_MASK				0x00000002
+#	define UPLL_BYPASS_EN_MASK			0x00000004
+#	define UPLL_CTLREQ_MASK				0x00000008
+#	define UPLL_REF_DIV(x)				((x) << 16)
+#	define UPLL_REF_DIV_MASK			0x001F0000
+#	define UPLL_CTLACK_MASK				0x40000000
+#	define UPLL_CTLACK2_MASK			0x80000000
+#define CG_UPLL_FUNC_CNTL_2				0x71c
+#	define UPLL_SW_HILEN(x)				((x) << 0)
+#	define UPLL_SW_LOLEN(x)				((x) << 4)
+#	define UPLL_SW_HILEN2(x)			((x) << 8)
+#	define UPLL_SW_LOLEN2(x)			((x) << 12)
+#	define UPLL_SW_MASK				0x0000FFFF
+#	define VCLK_SRC_SEL(x)				((x) << 20)
+#	define VCLK_SRC_SEL_MASK			0x01F00000
+#	define DCLK_SRC_SEL(x)				((x) << 25)
+#	define DCLK_SRC_SEL_MASK			0x3E000000
+#define CG_UPLL_FUNC_CNTL_3				0x720
+#	define UPLL_FB_DIV(x)				((x) << 0)
+#	define UPLL_FB_DIV_MASK				0x01FFFFFF
+
 /* Registers */
 #define	CB_COLOR0_BASE					0x28040
 #define	CB_COLOR1_BASE					0x28044
@@ -112,6 +136,11 @@
 #define DMA_TILING_CONFIG                               0x3ec8
 #define DMA_TILING_CONFIG2                              0xd0b8
 
+/* RV730 only */
+#define UVD_UDEC_TILING_CONFIG                          0xef40
+#define UVD_UDEC_DB_TILING_CONFIG                       0xef44
+#define UVD_UDEC_DBW_TILING_CONFIG                      0xef48
+
 #define	GC_USER_SHADER_PIPE_CONFIG			0x8954
 #define		INACTIVE_QD_PIPES(x)				((x) << 8)
 #define		INACTIVE_QD_PIPES_MASK				0x0000FF00
@@ -671,4 +700,18 @@
 #       define TARGET_LINK_SPEED_MASK                     (0xf << 0)
 #       define SELECTABLE_DEEMPHASIS                      (1 << 6)
 
+/* UVD */
+#define UVD_LMI_EXT40_ADDR				0xf498
+#define UVD_VCPU_CHIP_ID				0xf4d4
+#define UVD_VCPU_CACHE_OFFSET0				0xf4d8
+#define UVD_VCPU_CACHE_SIZE0				0xf4dc
+#define UVD_VCPU_CACHE_OFFSET1				0xf4e0
+#define UVD_VCPU_CACHE_SIZE1				0xf4e4
+#define UVD_VCPU_CACHE_OFFSET2				0xf4e8
+#define UVD_VCPU_CACHE_SIZE2				0xf4ec
+#define UVD_LMI_ADDR_EXT				0xf594
+
+#define UVD_RBC_RB_RPTR					0xf690
+#define UVD_RBC_RB_WPTR					0xf694
+
 #endif

drivers/gpu/drm/radeon/si.c

@@ -1211,6 +1211,7 @@ static void si_tiling_mode_table_init(struct radeon_device *rdev)
 				gb_tile_moden = 0;
 				break;
 			}
+			rdev->config.si.tile_mode_array[reg_offset] = gb_tile_moden;
 			WREG32(GB_TILE_MODE0 + (reg_offset * 4), gb_tile_moden);
 		}
 	} else if ((rdev->family == CHIP_VERDE) ||
@@ -1451,6 +1452,7 @@ static void si_tiling_mode_table_init(struct radeon_device *rdev)
 				gb_tile_moden = 0;
 				break;
 			}
+			rdev->config.si.tile_mode_array[reg_offset] = gb_tile_moden;
 			WREG32(GB_TILE_MODE0 + (reg_offset * 4), gb_tile_moden);
 		}
 	} else
@@ -1765,9 +1767,13 @@ static void si_gpu_init(struct radeon_device *rdev)
 
 	WREG32(GB_ADDR_CONFIG, gb_addr_config);
 	WREG32(DMIF_ADDR_CONFIG, gb_addr_config);
+	WREG32(DMIF_ADDR_CALC, gb_addr_config);
 	WREG32(HDP_ADDR_CONFIG, gb_addr_config);
 	WREG32(DMA_TILING_CONFIG + DMA0_REGISTER_OFFSET, gb_addr_config);
 	WREG32(DMA_TILING_CONFIG + DMA1_REGISTER_OFFSET, gb_addr_config);
+	WREG32(UVD_UDEC_ADDR_CONFIG, gb_addr_config);
+	WREG32(UVD_UDEC_DB_ADDR_CONFIG, gb_addr_config);
+	WREG32(UVD_UDEC_DBW_ADDR_CONFIG, gb_addr_config);
 
 	si_tiling_mode_table_init(rdev);
 
@@ -2538,46 +2544,6 @@ static void si_mc_program(struct radeon_device *rdev)
 	rv515_vga_render_disable(rdev);
 }
 
-/* SI MC address space is 40 bits */
-static void si_vram_location(struct radeon_device *rdev,
-			     struct radeon_mc *mc, u64 base)
-{
-	mc->vram_start = base;
-	if (mc->mc_vram_size > (0xFFFFFFFFFFULL - base + 1)) {
-		dev_warn(rdev->dev, "limiting VRAM to PCI aperture size\n");
-		mc->real_vram_size = mc->aper_size;
-		mc->mc_vram_size = mc->aper_size;
-	}
-	mc->vram_end = mc->vram_start + mc->mc_vram_size - 1;
-	dev_info(rdev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n",
-			mc->mc_vram_size >> 20, mc->vram_start,
-			mc->vram_end, mc->real_vram_size >> 20);
-}
-
-static void si_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc)
-{
-	u64 size_af, size_bf;
-
-	size_af = ((0xFFFFFFFFFFULL - mc->vram_end) + mc->gtt_base_align) & ~mc->gtt_base_align;
-	size_bf = mc->vram_start & ~mc->gtt_base_align;
-	if (size_bf > size_af) {
-		if (mc->gtt_size > size_bf) {
-			dev_warn(rdev->dev, "limiting GTT\n");
-			mc->gtt_size = size_bf;
-		}
-		mc->gtt_start = (mc->vram_start & ~mc->gtt_base_align) - mc->gtt_size;
-	} else {
-		if (mc->gtt_size > size_af) {
-			dev_warn(rdev->dev, "limiting GTT\n");
-			mc->gtt_size = size_af;
-		}
-		mc->gtt_start = (mc->vram_end + 1 + mc->gtt_base_align) & ~mc->gtt_base_align;
-	}
-	mc->gtt_end = mc->gtt_start + mc->gtt_size - 1;
-	dev_info(rdev->dev, "GTT: %lluM 0x%016llX - 0x%016llX\n",
-			mc->gtt_size >> 20, mc->gtt_start, mc->gtt_end);
-}
-
 static void si_vram_gtt_location(struct radeon_device *rdev,
 				 struct radeon_mc *mc)
 {
@@ -2587,9 +2553,9 @@ static void si_vram_gtt_location(struct radeon_device *rdev,
 		mc->real_vram_size = 0xFFC0000000ULL;
 		mc->mc_vram_size = 0xFFC0000000ULL;
 	}
-	si_vram_location(rdev, &rdev->mc, 0);
+	radeon_vram_location(rdev, &rdev->mc, 0);
 	rdev->mc.gtt_base_align = 0;
-	si_gtt_location(rdev, mc);
+	radeon_gtt_location(rdev, mc);
 }
 
 static int si_mc_init(struct radeon_device *rdev)
@@ -4322,14 +4288,6 @@ static int si_startup(struct radeon_device *rdev)
 		return r;
 	si_gpu_init(rdev);
 
-#if 0
-	r = evergreen_blit_init(rdev);
-	if (r) {
-		r600_blit_fini(rdev);
-		rdev->asic->copy = NULL;
-		dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r);
-	}
-#endif
 	/* allocate rlc buffers */
 	r = si_rlc_init(rdev);
 	if (r) {
@@ -4372,6 +4330,16 @@ static int si_startup(struct radeon_device *rdev)
 		return r;
 	}
 
+	r = rv770_uvd_resume(rdev);
+	if (!r) {
+		r = radeon_fence_driver_start_ring(rdev,
+						   R600_RING_TYPE_UVD_INDEX);
+		if (r)
+			dev_err(rdev->dev, "UVD fences init error (%d).\n", r);
+	}
+	if (r)
+		rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_size = 0;
+
 	/* Enable IRQ */
 	r = si_irq_init(rdev);
 	if (r) {
@@ -4429,6 +4397,18 @@ static int si_startup(struct radeon_device *rdev)
 	if (r)
 		return r;
 
+	ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+	if (ring->ring_size) {
+		r = radeon_ring_init(rdev, ring, ring->ring_size,
+				     R600_WB_UVD_RPTR_OFFSET,
+				     UVD_RBC_RB_RPTR, UVD_RBC_RB_WPTR,
+				     0, 0xfffff, RADEON_CP_PACKET2);
+		if (!r)
+			r = r600_uvd_init(rdev);
+		if (r)
+			DRM_ERROR("radeon: failed initializing UVD (%d).\n", r);
+	}
+
 	r = radeon_ib_pool_init(rdev);
 	if (r) {
 		dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
@@ -4472,6 +4452,8 @@ int si_suspend(struct radeon_device *rdev)
 	radeon_vm_manager_fini(rdev);
 	si_cp_enable(rdev, false);
 	cayman_dma_stop(rdev);
+	r600_uvd_rbc_stop(rdev);
+	radeon_uvd_suspend(rdev);
 	si_irq_suspend(rdev);
 	radeon_wb_disable(rdev);
 	si_pcie_gart_disable(rdev);
@@ -4557,6 +4539,13 @@ int si_init(struct radeon_device *rdev)
 	ring->ring_obj = NULL;
 	r600_ring_init(rdev, ring, 64 * 1024);
 
+	r = radeon_uvd_init(rdev);
+	if (!r) {
+		ring = &rdev->ring[R600_RING_TYPE_UVD_INDEX];
+		ring->ring_obj = NULL;
+		r600_ring_init(rdev, ring, 4096);
+	}
+
 	rdev->ih.ring_obj = NULL;
 	r600_ih_ring_init(rdev, 64 * 1024);
 
@@ -4594,9 +4583,6 @@ int si_init(struct radeon_device *rdev)
 
 void si_fini(struct radeon_device *rdev)
 {
-#if 0
-	r600_blit_fini(rdev);
-#endif
 	si_cp_fini(rdev);
 	cayman_dma_fini(rdev);
 	si_irq_fini(rdev);
@@ -4605,6 +4591,7 @@ void si_fini(struct radeon_device *rdev)
 	radeon_vm_manager_fini(rdev);
 	radeon_ib_pool_fini(rdev);
 	radeon_irq_kms_fini(rdev);
+	radeon_uvd_fini(rdev);
 	si_pcie_gart_fini(rdev);
 	r600_vram_scratch_fini(rdev);
 	radeon_gem_fini(rdev);
@@ -4634,3 +4621,170 @@ uint64_t si_get_gpu_clock_counter(struct radeon_device *rdev)
 	mutex_unlock(&rdev->gpu_clock_mutex);
 	return clock;
 }
+
+static int si_uvd_calc_post_div(unsigned target_freq,
+				unsigned vco_freq,
+				unsigned *div)
+{
+	/* target larger than vco frequency ? */
+	if (vco_freq < target_freq)
+		return -1; /* forget it */
+
+	/* Fclk = Fvco / PDIV */
+	*div = vco_freq / target_freq;
+
+	/* we alway need a frequency less than or equal the target */
+	if ((vco_freq / *div) > target_freq)
+		*div += 1;
+
+	/* dividers above 5 must be even */
+	if (*div > 5 && *div % 2)
+		*div += 1;
+
+	/* out of range ? */
+	if (*div >= 128)
+		return -1; /* forget it */
+
+	return vco_freq / *div;
+}
+
+static int si_uvd_send_upll_ctlreq(struct radeon_device *rdev)
+{
+	unsigned i;
+
+	/* assert UPLL_CTLREQ */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_CTLREQ_MASK, ~UPLL_CTLREQ_MASK);
+
+	/* wait for CTLACK and CTLACK2 to get asserted */
+	for (i = 0; i < 100; ++i) {
+		uint32_t mask = UPLL_CTLACK_MASK | UPLL_CTLACK2_MASK;
+		if ((RREG32(CG_UPLL_FUNC_CNTL) & mask) == mask)
+			break;
+		mdelay(10);
+	}
+	if (i == 100)
+		return -ETIMEDOUT;
+
+	/* deassert UPLL_CTLREQ */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_CTLREQ_MASK);
+
+	return 0;
+}
+
+int si_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk)
+{
+	/* start off with something large */
+	int optimal_diff_score = 0x7FFFFFF;
+	unsigned optimal_fb_div = 0, optimal_vclk_div = 0;
+	unsigned optimal_dclk_div = 0, optimal_vco_freq = 0;
+	unsigned vco_freq;
+	int r;
+
+	/* loop through vco from low to high */
+	for (vco_freq = 125000; vco_freq <= 250000; vco_freq += 100) {
+		unsigned fb_div = vco_freq / rdev->clock.spll.reference_freq * 16384;
+		int calc_clk, diff_score, diff_vclk, diff_dclk;
+		unsigned vclk_div, dclk_div;
+
+		/* fb div out of range ? */
+		if (fb_div > 0x03FFFFFF)
+			break; /* it can oly get worse */
+
+		/* calc vclk with current vco freq. */
+		calc_clk = si_uvd_calc_post_div(vclk, vco_freq, &vclk_div);
+		if (calc_clk == -1)
+			break; /* vco is too big, it has to stop. */
+		diff_vclk = vclk - calc_clk;
+
+		/* calc dclk with current vco freq. */
+		calc_clk = si_uvd_calc_post_div(dclk, vco_freq, &dclk_div);
+		if (calc_clk == -1)
+			break; /* vco is too big, it has to stop. */
+		diff_dclk = dclk - calc_clk;
+
+		/* determine if this vco setting is better than current optimal settings */
+		diff_score = abs(diff_vclk) + abs(diff_dclk);
+		if (diff_score < optimal_diff_score) {
+			optimal_fb_div = fb_div;
+			optimal_vclk_div = vclk_div;
+			optimal_dclk_div = dclk_div;
+			optimal_vco_freq = vco_freq;
+			optimal_diff_score = diff_score;
+			if (optimal_diff_score == 0)
+				break; /* it can't get better than this */
+		}
+	}
+
+	/* set RESET_ANTI_MUX to 0 */
+	WREG32_P(CG_UPLL_FUNC_CNTL_5, 0, ~RESET_ANTI_MUX_MASK);
+
+	/* set VCO_MODE to 1 */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_VCO_MODE_MASK, ~UPLL_VCO_MODE_MASK);
+
+	/* toggle UPLL_SLEEP to 1 then back to 0 */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_SLEEP_MASK, ~UPLL_SLEEP_MASK);
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_SLEEP_MASK);
+
+	/* deassert UPLL_RESET */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_RESET_MASK);
+
+	mdelay(1);
+
+	/* bypass vclk and dclk with bclk */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		VCLK_SRC_SEL(1) | DCLK_SRC_SEL(1),
+		~(VCLK_SRC_SEL_MASK | DCLK_SRC_SEL_MASK));
+
+	/* put PLL in bypass mode */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_BYPASS_EN_MASK, ~UPLL_BYPASS_EN_MASK);
+
+	r = si_uvd_send_upll_ctlreq(rdev);
+	if (r)
+		return r;
+
+	/* assert UPLL_RESET again */
+	WREG32_P(CG_UPLL_FUNC_CNTL, UPLL_RESET_MASK, ~UPLL_RESET_MASK);
+
+	/* disable spread spectrum. */
+	WREG32_P(CG_UPLL_SPREAD_SPECTRUM, 0, ~SSEN_MASK);
+
+	/* set feedback divider */
+	WREG32_P(CG_UPLL_FUNC_CNTL_3, UPLL_FB_DIV(optimal_fb_div), ~UPLL_FB_DIV_MASK);
+
+	/* set ref divider to 0 */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_REF_DIV_MASK);
+
+	if (optimal_vco_freq < 187500)
+		WREG32_P(CG_UPLL_FUNC_CNTL_4, 0, ~UPLL_SPARE_ISPARE9);
+	else
+		WREG32_P(CG_UPLL_FUNC_CNTL_4, UPLL_SPARE_ISPARE9, ~UPLL_SPARE_ISPARE9);
+
+	/* set PDIV_A and PDIV_B */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		UPLL_PDIV_A(optimal_vclk_div) | UPLL_PDIV_B(optimal_dclk_div),
+		~(UPLL_PDIV_A_MASK | UPLL_PDIV_B_MASK));
+
+	/* give the PLL some time to settle */
+	mdelay(15);
+
+	/* deassert PLL_RESET */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_RESET_MASK);
+
+	mdelay(15);
+
+	/* switch from bypass mode to normal mode */
+	WREG32_P(CG_UPLL_FUNC_CNTL, 0, ~UPLL_BYPASS_EN_MASK);
+
+	r = si_uvd_send_upll_ctlreq(rdev);
+	if (r)
+		return r;
+
+	/* switch VCLK and DCLK selection */
+	WREG32_P(CG_UPLL_FUNC_CNTL_2,
+		VCLK_SRC_SEL(2) | DCLK_SRC_SEL(2),
+		~(VCLK_SRC_SEL_MASK | DCLK_SRC_SEL_MASK));
+
+	mdelay(100);
+
+	return 0;
+}

drivers/gpu/drm/radeon/sid.h

@@ -29,6 +29,35 @@
 #define TAHITI_GB_ADDR_CONFIG_GOLDEN        0x12011003
 #define VERDE_GB_ADDR_CONFIG_GOLDEN         0x12010002
 
+/* discrete uvd clocks */
+#define	CG_UPLL_FUNC_CNTL				0x634
+#	define UPLL_RESET_MASK				0x00000001
+#	define UPLL_SLEEP_MASK				0x00000002
+#	define UPLL_BYPASS_EN_MASK			0x00000004
+#	define UPLL_CTLREQ_MASK				0x00000008
+#	define UPLL_VCO_MODE_MASK			0x00000600
+#	define UPLL_REF_DIV_MASK			0x001F0000
+#	define UPLL_CTLACK_MASK				0x40000000
+#	define UPLL_CTLACK2_MASK			0x80000000
+#define	CG_UPLL_FUNC_CNTL_2				0x638
+#	define UPLL_PDIV_A(x)				((x) << 0)
+#	define UPLL_PDIV_A_MASK				0x0000007F
+#	define UPLL_PDIV_B(x)				((x) << 8)
+#	define UPLL_PDIV_B_MASK				0x00007F00
+#	define VCLK_SRC_SEL(x)				((x) << 20)
+#	define VCLK_SRC_SEL_MASK			0x01F00000
+#	define DCLK_SRC_SEL(x)				((x) << 25)
+#	define DCLK_SRC_SEL_MASK			0x3E000000
+#define	CG_UPLL_FUNC_CNTL_3				0x63C
+#	define UPLL_FB_DIV(x)				((x) << 0)
+#	define UPLL_FB_DIV_MASK				0x01FFFFFF
+#define	CG_UPLL_FUNC_CNTL_4                             0x644
+#	define UPLL_SPARE_ISPARE9			0x00020000
+#define	CG_UPLL_FUNC_CNTL_5				0x648
+#	define RESET_ANTI_MUX_MASK			0x00000200
+#define	CG_UPLL_SPREAD_SPECTRUM				0x650
+#	define SSEN_MASK				0x00000001
+
 #define	CG_MULT_THERMAL_STATUS					0x714
 #define		ASIC_MAX_TEMP(x)				((x) << 0)
 #define		ASIC_MAX_TEMP_MASK				0x000001ff
@@ -65,6 +94,8 @@
 
 #define DMIF_ADDR_CONFIG  				0xBD4
 
+#define DMIF_ADDR_CALC  				0xC00
+
 #define	SRBM_STATUS				        0xE50
 #define		GRBM_RQ_PENDING 			(1 << 5)
 #define		VMC_BUSY 				(1 << 8)
@@ -797,6 +828,15 @@
 #       define THREAD_TRACE_FLUSH                       (54 << 0)
 #       define THREAD_TRACE_FINISH                      (55 << 0)
 
+/*
+ * UVD
+ */
+#define UVD_UDEC_ADDR_CONFIG				0xEF4C
+#define UVD_UDEC_DB_ADDR_CONFIG				0xEF50
+#define UVD_UDEC_DBW_ADDR_CONFIG			0xEF54
+#define UVD_RBC_RB_RPTR					0xF690
+#define UVD_RBC_RB_WPTR					0xF694
+
 /*
  * PM4
  */

include/uapi/drm/radeon_drm.h

@@ -918,6 +918,7 @@ struct drm_radeon_gem_va {
 #define RADEON_CS_RING_GFX          0
 #define RADEON_CS_RING_COMPUTE      1
 #define RADEON_CS_RING_DMA          2
+#define RADEON_CS_RING_UVD          3
 /* The third dword of RADEON_CHUNK_ID_FLAGS is a sint32 that sets the priority */
 /* 0 = normal, + = higher priority, - = lower priority */
 
@@ -972,6 +973,13 @@ struct drm_radeon_cs {
 #define RADEON_INFO_MAX_SE		0x12
 /* max SH per SE */
 #define RADEON_INFO_MAX_SH_PER_SE	0x13
+/* fast fb access is enabled */
+#define RADEON_INFO_FASTFB_WORKING	0x14
+/* query if a RADEON_CS_RING_* submission is supported */
+#define RADEON_INFO_RING_WORKING	0x15
+/* SI tile mode array */
+#define RADEON_INFO_SI_TILE_MODE_ARRAY	0x16
+
 
 struct drm_radeon_info {
 	uint32_t		request;
@@ -979,4 +987,22 @@ struct drm_radeon_info {
 	uint64_t		value;
 };
 
+/* Those correspond to the tile index to use, this is to explicitly state
+ * the API that is implicitly defined by the tile mode array.
+ */
+#define SI_TILE_MODE_COLOR_LINEAR_ALIGNED	8
+#define SI_TILE_MODE_COLOR_1D			13
+#define SI_TILE_MODE_COLOR_1D_SCANOUT		9
+#define SI_TILE_MODE_COLOR_2D_8BPP		14
+#define SI_TILE_MODE_COLOR_2D_16BPP		15
+#define SI_TILE_MODE_COLOR_2D_32BPP		16
+#define SI_TILE_MODE_COLOR_2D_64BPP		17
+#define SI_TILE_MODE_COLOR_2D_SCANOUT_16BPP	11
+#define SI_TILE_MODE_COLOR_2D_SCANOUT_32BPP	12
+#define SI_TILE_MODE_DEPTH_STENCIL_1D		4
+#define SI_TILE_MODE_DEPTH_STENCIL_2D		0
+#define SI_TILE_MODE_DEPTH_STENCIL_2D_2AA	3
+#define SI_TILE_MODE_DEPTH_STENCIL_2D_4AA	3
+#define SI_TILE_MODE_DEPTH_STENCIL_2D_8AA	2
+
 #endif