Merge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (40 commits)
  drm/radeon/kms: i2c s/sprintf/snprintf/g for safety
  drm/radeon/kms: fix i2c pad masks on rs4xx
  drm/ttm: Fix up a theoretical deadlock
  drm/radeon/kms: fix tiling info on evergreen
  drm/radeon/kms: fix alignment when allocating buffers
  drm/vmwgfx: Fix up an error path during bo creation
  drm/radeon/kms: register an i2c adapter name for the dp aux bus
  drm/radeon/kms/atom: add proper external encoders support
  drm/radeon/kms/atom: cleanup and unify DVO handling
  drm/radeon/kms: properly power up/down the eDP panel as needed (v4)
  drm/radeon/kms/atom: set sane defaults in atombios_get_encoder_mode()
  drm/radeon/kms: turn the backlight off explicitly for dpms
  drm/radeon/kms: fix typo in r600 cs checker
  drm: radeon: fix error value sign
  drm/radeon/kms: fix and unify tiled buffer alignment checking for r6xx/7xx
  nouveau: Acknowledge HPD irq in handler, not bottom half
  drm/nouveau: Fix a few confusions between "chipset" and "card_type".
  drm/nouveau: don't expose backlight control when available through ACPI
  drm/nouveau/pm: improve memtiming mappings
  drm/nouveau: Make PCIE GART size depend on the available RAMIN space.
  ...

drivers/gpu/drm/nouveau/nouveau_backlight.c

@@ -31,6 +31,7 @@
  */
 
 #include <linux/backlight.h>
+#include <linux/acpi.h>
 
 #include "drmP.h"
 #include "nouveau_drv.h"
@@ -136,6 +137,14 @@ int nouveau_backlight_init(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 
+#ifdef CONFIG_ACPI
+	if (acpi_video_backlight_support()) {
+		NV_INFO(dev, "ACPI backlight interface available, "
+			     "not registering our own\n");
+		return 0;
+	}
+#endif
+
 	switch (dev_priv->card_type) {
 	case NV_40:
 		return nouveau_nv40_backlight_init(dev);

drivers/gpu/drm/nouveau/nouveau_bios.c

@@ -6829,7 +6829,7 @@ nouveau_bios_posted(struct drm_device *dev)
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	unsigned htotal;
 
-	if (dev_priv->chipset >= NV_50) {
+	if (dev_priv->card_type >= NV_50) {
 		if (NVReadVgaCrtc(dev, 0, 0x00) == 0 &&
 		    NVReadVgaCrtc(dev, 0, 0x1a) == 0)
 			return false;

drivers/gpu/drm/nouveau/nouveau_bo.c

@@ -143,8 +143,10 @@ nouveau_bo_new(struct drm_device *dev, struct nouveau_channel *chan,
 	nvbo->no_vm = no_vm;
 	nvbo->tile_mode = tile_mode;
 	nvbo->tile_flags = tile_flags;
+	nvbo->bo.bdev = &dev_priv->ttm.bdev;
 
-	nouveau_bo_fixup_align(dev, tile_mode, tile_flags, &align, &size);
+	nouveau_bo_fixup_align(dev, tile_mode, nouveau_bo_tile_layout(nvbo),
+			       &align, &size);
 	align >>= PAGE_SHIFT;
 
 	nouveau_bo_placement_set(nvbo, flags, 0);
@@ -176,6 +178,31 @@ set_placement_list(uint32_t *pl, unsigned *n, uint32_t type, uint32_t flags)
 		pl[(*n)++] = TTM_PL_FLAG_SYSTEM | flags;
 }
 
+static void
+set_placement_range(struct nouveau_bo *nvbo, uint32_t type)
+{
+	struct drm_nouveau_private *dev_priv = nouveau_bdev(nvbo->bo.bdev);
+
+	if (dev_priv->card_type == NV_10 &&
+	    nvbo->tile_mode && (type & TTM_PL_FLAG_VRAM)) {
+		/*
+		 * Make sure that the color and depth buffers are handled
+		 * by independent memory controller units. Up to a 9x
+		 * speed up when alpha-blending and depth-test are enabled
+		 * at the same time.
+		 */
+		int vram_pages = dev_priv->vram_size >> PAGE_SHIFT;
+
+		if (nvbo->tile_flags & NOUVEAU_GEM_TILE_ZETA) {
+			nvbo->placement.fpfn = vram_pages / 2;
+			nvbo->placement.lpfn = ~0;
+		} else {
+			nvbo->placement.fpfn = 0;
+			nvbo->placement.lpfn = vram_pages / 2;
+		}
+	}
+}
+
 void
 nouveau_bo_placement_set(struct nouveau_bo *nvbo, uint32_t type, uint32_t busy)
 {
@@ -190,6 +217,8 @@ nouveau_bo_placement_set(struct nouveau_bo *nvbo, uint32_t type, uint32_t busy)
 	pl->busy_placement = nvbo->busy_placements;
 	set_placement_list(nvbo->busy_placements, &pl->num_busy_placement,
 			   type | busy, flags);
+
+	set_placement_range(nvbo, type);
 }
 
 int
@@ -525,7 +554,8 @@ nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 		stride  = 16 * 4;
 		height  = amount / stride;
 
-		if (new_mem->mem_type == TTM_PL_VRAM && nvbo->tile_flags) {
+		if (new_mem->mem_type == TTM_PL_VRAM &&
+		    nouveau_bo_tile_layout(nvbo)) {
 			ret = RING_SPACE(chan, 8);
 			if (ret)
 				return ret;
@@ -546,7 +576,8 @@ nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo,
 			BEGIN_RING(chan, NvSubM2MF, 0x0200, 1);
 			OUT_RING  (chan, 1);
 		}
-		if (old_mem->mem_type == TTM_PL_VRAM && nvbo->tile_flags) {
+		if (old_mem->mem_type == TTM_PL_VRAM &&
+		    nouveau_bo_tile_layout(nvbo)) {
 			ret = RING_SPACE(chan, 8);
 			if (ret)
 				return ret;
@@ -753,7 +784,8 @@ nouveau_bo_vm_bind(struct ttm_buffer_object *bo, struct ttm_mem_reg *new_mem,
 	if (dev_priv->card_type == NV_50) {
 		ret = nv50_mem_vm_bind_linear(dev,
 					      offset + dev_priv->vm_vram_base,
-					      new_mem->size, nvbo->tile_flags,
+					      new_mem->size,
+					      nouveau_bo_tile_layout(nvbo),
 					      offset);
 		if (ret)
 			return ret;
@@ -894,7 +926,8 @@ nouveau_ttm_fault_reserve_notify(struct ttm_buffer_object *bo)
 	 * nothing to do here.
 	 */
 	if (bo->mem.mem_type != TTM_PL_VRAM) {
-		if (dev_priv->card_type < NV_50 || !nvbo->tile_flags)
+		if (dev_priv->card_type < NV_50 ||
+		    !nouveau_bo_tile_layout(nvbo))
 			return 0;
 	}
 

drivers/gpu/drm/nouveau/nouveau_connector.c

@@ -281,7 +281,7 @@ detect_analog:
 	nv_encoder = find_encoder_by_type(connector, OUTPUT_ANALOG);
 	if (!nv_encoder && !nouveau_tv_disable)
 		nv_encoder = find_encoder_by_type(connector, OUTPUT_TV);
-	if (nv_encoder) {
+	if (nv_encoder && force) {
 		struct drm_encoder *encoder = to_drm_encoder(nv_encoder);
 		struct drm_encoder_helper_funcs *helper =
 						encoder->helper_private;
@@ -641,11 +641,28 @@ nouveau_connector_get_modes(struct drm_connector *connector)
 	return ret;
 }
 
+static unsigned
+get_tmds_link_bandwidth(struct drm_connector *connector)
+{
+	struct nouveau_connector *nv_connector = nouveau_connector(connector);
+	struct drm_nouveau_private *dev_priv = connector->dev->dev_private;
+	struct dcb_entry *dcb = nv_connector->detected_encoder->dcb;
+
+	if (dcb->location != DCB_LOC_ON_CHIP ||
+	    dev_priv->chipset >= 0x46)
+		return 165000;
+	else if (dev_priv->chipset >= 0x40)
+		return 155000;
+	else if (dev_priv->chipset >= 0x18)
+		return 135000;
+	else
+		return 112000;
+}
+
 static int
 nouveau_connector_mode_valid(struct drm_connector *connector,
 			     struct drm_display_mode *mode)
 {
-	struct drm_nouveau_private *dev_priv = connector->dev->dev_private;
 	struct nouveau_connector *nv_connector = nouveau_connector(connector);
 	struct nouveau_encoder *nv_encoder = nv_connector->detected_encoder;
 	struct drm_encoder *encoder = to_drm_encoder(nv_encoder);
@@ -663,11 +680,9 @@ nouveau_connector_mode_valid(struct drm_connector *connector,
 		max_clock = 400000;
 		break;
 	case OUTPUT_TMDS:
-		if ((dev_priv->card_type >= NV_50 && !nouveau_duallink) ||
-		    !nv_encoder->dcb->duallink_possible)
-			max_clock = 165000;
-		else
-			max_clock = 330000;
+		max_clock = get_tmds_link_bandwidth(connector);
+		if (nouveau_duallink && nv_encoder->dcb->duallink_possible)
+			max_clock *= 2;
 		break;
 	case OUTPUT_ANALOG:
 		max_clock = nv_encoder->dcb->crtconf.maxfreq;
@@ -709,44 +724,6 @@ nouveau_connector_best_encoder(struct drm_connector *connector)
 	return NULL;
 }
 
-void
-nouveau_connector_set_polling(struct drm_connector *connector)
-{
-	struct drm_device *dev = connector->dev;
-	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	struct drm_crtc *crtc;
-	bool spare_crtc = false;
-
-	list_for_each_entry(crtc, &dev->mode_config.crtc_list, head)
-		spare_crtc |= !crtc->enabled;
-
-	connector->polled = 0;
-
-	switch (connector->connector_type) {
-	case DRM_MODE_CONNECTOR_VGA:
-	case DRM_MODE_CONNECTOR_TV:
-		if (dev_priv->card_type >= NV_50 ||
-		    (nv_gf4_disp_arch(dev) && spare_crtc))
-			connector->polled = DRM_CONNECTOR_POLL_CONNECT;
-		break;
-
-	case DRM_MODE_CONNECTOR_DVII:
-	case DRM_MODE_CONNECTOR_DVID:
-	case DRM_MODE_CONNECTOR_HDMIA:
-	case DRM_MODE_CONNECTOR_DisplayPort:
-	case DRM_MODE_CONNECTOR_eDP:
-		if (dev_priv->card_type >= NV_50)
-			connector->polled = DRM_CONNECTOR_POLL_HPD;
-		else if (connector->connector_type == DRM_MODE_CONNECTOR_DVID ||
-			 spare_crtc)
-			connector->polled = DRM_CONNECTOR_POLL_CONNECT;
-		break;
-
-	default:
-		break;
-	}
-}
-
 static const struct drm_connector_helper_funcs
 nouveau_connector_helper_funcs = {
 	.get_modes = nouveau_connector_get_modes,
@@ -872,6 +849,7 @@ nouveau_connector_create(struct drm_device *dev, int index)
 					dev->mode_config.scaling_mode_property,
 					nv_connector->scaling_mode);
 		}
+		connector->polled = DRM_CONNECTOR_POLL_CONNECT;
 		/* fall-through */
 	case DCB_CONNECTOR_TV_0:
 	case DCB_CONNECTOR_TV_1:
@@ -888,11 +866,16 @@ nouveau_connector_create(struct drm_device *dev, int index)
 				dev->mode_config.dithering_mode_property,
 				nv_connector->use_dithering ?
 				DRM_MODE_DITHERING_ON : DRM_MODE_DITHERING_OFF);
+
+		if (dcb->type != DCB_CONNECTOR_LVDS) {
+			if (dev_priv->card_type >= NV_50)
+				connector->polled = DRM_CONNECTOR_POLL_HPD;
+			else
+				connector->polled = DRM_CONNECTOR_POLL_CONNECT;
+		}
 		break;
 	}
 
-	nouveau_connector_set_polling(connector);
-
 	drm_sysfs_connector_add(connector);
 	dcb->drm = connector;
 	return dcb->drm;

drivers/gpu/drm/nouveau/nouveau_connector.h

@@ -52,9 +52,6 @@ static inline struct nouveau_connector *nouveau_connector(
 struct drm_connector *
 nouveau_connector_create(struct drm_device *, int index);
 
-void
-nouveau_connector_set_polling(struct drm_connector *);
-
 int
 nouveau_connector_bpp(struct drm_connector *);
 

drivers/gpu/drm/nouveau/nouveau_drv.h

@@ -100,6 +100,9 @@ struct nouveau_bo {
 	int pin_refcnt;
 };
 
+#define nouveau_bo_tile_layout(nvbo)				\
+	((nvbo)->tile_flags & NOUVEAU_GEM_TILE_LAYOUT_MASK)
+
 static inline struct nouveau_bo *
 nouveau_bo(struct ttm_buffer_object *bo)
 {
@@ -304,6 +307,7 @@ struct nouveau_fifo_engine {
 	void (*destroy_context)(struct nouveau_channel *);
 	int  (*load_context)(struct nouveau_channel *);
 	int  (*unload_context)(struct drm_device *);
+	void (*tlb_flush)(struct drm_device *dev);
 };
 
 struct nouveau_pgraph_object_method {
@@ -336,6 +340,7 @@ struct nouveau_pgraph_engine {
 	void (*destroy_context)(struct nouveau_channel *);
 	int  (*load_context)(struct nouveau_channel *);
 	int  (*unload_context)(struct drm_device *);
+	void (*tlb_flush)(struct drm_device *dev);
 
 	void (*set_region_tiling)(struct drm_device *dev, int i, uint32_t addr,
 				  uint32_t size, uint32_t pitch);
@@ -485,13 +490,13 @@ enum nv04_fp_display_regs {
 };
 
 struct nv04_crtc_reg {
-	unsigned char MiscOutReg;     /* */
+	unsigned char MiscOutReg;
 	uint8_t CRTC[0xa0];
 	uint8_t CR58[0x10];
 	uint8_t Sequencer[5];
 	uint8_t Graphics[9];
 	uint8_t Attribute[21];
-	unsigned char DAC[768];       /* Internal Colorlookuptable */
+	unsigned char DAC[768];
 
 	/* PCRTC regs */
 	uint32_t fb_start;
@@ -539,43 +544,9 @@ struct nv04_output_reg {
 };
 
 struct nv04_mode_state {
-	uint32_t bpp;
-	uint32_t width;
-	uint32_t height;
-	uint32_t interlace;
-	uint32_t repaint0;
-	uint32_t repaint1;
-	uint32_t screen;
-	uint32_t scale;
-	uint32_t dither;
-	uint32_t extra;
-	uint32_t fifo;
-	uint32_t pixel;
-	uint32_t horiz;
-	int arbitration0;
-	int arbitration1;
-	uint32_t pll;
-	uint32_t pllB;
-	uint32_t vpll;
-	uint32_t vpll2;
-	uint32_t vpllB;
-	uint32_t vpll2B;
+	struct nv04_crtc_reg crtc_reg[2];
 	uint32_t pllsel;
 	uint32_t sel_clk;
-	uint32_t general;
-	uint32_t crtcOwner;
-	uint32_t head;
-	uint32_t head2;
-	uint32_t cursorConfig;
-	uint32_t cursor0;
-	uint32_t cursor1;
-	uint32_t cursor2;
-	uint32_t timingH;
-	uint32_t timingV;
-	uint32_t displayV;
-	uint32_t crtcSync;
-
-	struct nv04_crtc_reg crtc_reg[2];
 };
 
 enum nouveau_card_type {
@@ -613,6 +584,12 @@ struct drm_nouveau_private {
 	struct work_struct irq_work;
 	struct work_struct hpd_work;
 
+	struct {
+		spinlock_t lock;
+		uint32_t hpd0_bits;
+		uint32_t hpd1_bits;
+	} hpd_state;
+
 	struct list_head vbl_waiting;
 
 	struct {
@@ -1045,6 +1022,7 @@ extern int  nv50_fifo_create_context(struct nouveau_channel *);
 extern void nv50_fifo_destroy_context(struct nouveau_channel *);
 extern int  nv50_fifo_load_context(struct nouveau_channel *);
 extern int  nv50_fifo_unload_context(struct drm_device *);
+extern void nv50_fifo_tlb_flush(struct drm_device *dev);
 
 /* nvc0_fifo.c */
 extern int  nvc0_fifo_init(struct drm_device *);
@@ -1122,6 +1100,8 @@ extern int  nv50_graph_load_context(struct nouveau_channel *);
 extern int  nv50_graph_unload_context(struct drm_device *);
 extern void nv50_graph_context_switch(struct drm_device *);
 extern int  nv50_grctx_init(struct nouveau_grctx *);
+extern void nv50_graph_tlb_flush(struct drm_device *dev);
+extern void nv86_graph_tlb_flush(struct drm_device *dev);
 
 /* nvc0_graph.c */
 extern int  nvc0_graph_init(struct drm_device *);
@@ -1239,7 +1219,6 @@ extern u16 nouveau_bo_rd16(struct nouveau_bo *nvbo, unsigned index);
 extern void nouveau_bo_wr16(struct nouveau_bo *nvbo, unsigned index, u16 val);
 extern u32 nouveau_bo_rd32(struct nouveau_bo *nvbo, unsigned index);
 extern void nouveau_bo_wr32(struct nouveau_bo *nvbo, unsigned index, u32 val);
-extern int nouveau_bo_sync_gpu(struct nouveau_bo *, struct nouveau_channel *);
 
 /* nouveau_fence.c */
 struct nouveau_fence;

drivers/gpu/drm/nouveau/nouveau_fence.c

@@ -249,6 +249,7 @@ alloc_semaphore(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_semaphore *sema;
+	int ret;
 
 	if (!USE_SEMA(dev))
 		return NULL;
@@ -257,10 +258,14 @@ alloc_semaphore(struct drm_device *dev)
 	if (!sema)
 		goto fail;
 
+	ret = drm_mm_pre_get(&dev_priv->fence.heap);
+	if (ret)
+		goto fail;
+
 	spin_lock(&dev_priv->fence.lock);
 	sema->mem = drm_mm_search_free(&dev_priv->fence.heap, 4, 0, 0);
 	if (sema->mem)
-		sema->mem = drm_mm_get_block(sema->mem, 4, 0);
+		sema->mem = drm_mm_get_block_atomic(sema->mem, 4, 0);
 	spin_unlock(&dev_priv->fence.lock);
 
 	if (!sema->mem)

drivers/gpu/drm/nouveau/nouveau_gem.c

@@ -107,23 +107,29 @@ nouveau_gem_info(struct drm_gem_object *gem, struct drm_nouveau_gem_info *rep)
 }
 
 static bool
-nouveau_gem_tile_flags_valid(struct drm_device *dev, uint32_t tile_flags) {
-	switch (tile_flags) {
-	case 0x0000:
-	case 0x1800:
-	case 0x2800:
-	case 0x4800:
-	case 0x7000:
-	case 0x7400:
-	case 0x7a00:
-	case 0xe000:
-		break;
-	default:
-		NV_ERROR(dev, "bad page flags: 0x%08x\n", tile_flags);
-		return false;
+nouveau_gem_tile_flags_valid(struct drm_device *dev, uint32_t tile_flags)
+{
+	struct drm_nouveau_private *dev_priv = dev->dev_private;
+
+	if (dev_priv->card_type >= NV_50) {
+		switch (tile_flags & NOUVEAU_GEM_TILE_LAYOUT_MASK) {
+		case 0x0000:
+		case 0x1800:
+		case 0x2800:
+		case 0x4800:
+		case 0x7000:
+		case 0x7400:
+		case 0x7a00:
+		case 0xe000:
+			return true;
+		}
+	} else {
+		if (!(tile_flags & NOUVEAU_GEM_TILE_LAYOUT_MASK))
+			return true;
 	}
 
-	return true;
+	NV_ERROR(dev, "bad page flags: 0x%08x\n", tile_flags);
+	return false;
 }
 
 int

drivers/gpu/drm/nouveau/nouveau_hw.c

@@ -519,11 +519,11 @@ nouveau_hw_fix_bad_vpll(struct drm_device *dev, int head)
 
 	struct pll_lims pll_lim;
 	struct nouveau_pll_vals pv;
-	uint32_t pllreg = head ? NV_RAMDAC_VPLL2 : NV_PRAMDAC_VPLL_COEFF;
+	enum pll_types pll = head ? PLL_VPLL1 : PLL_VPLL0;
 
-	if (get_pll_limits(dev, pllreg, &pll_lim))
+	if (get_pll_limits(dev, pll, &pll_lim))
 		return;
-	nouveau_hw_get_pllvals(dev, pllreg, &pv);
+	nouveau_hw_get_pllvals(dev, pll, &pv);
 
 	if (pv.M1 >= pll_lim.vco1.min_m && pv.M1 <= pll_lim.vco1.max_m &&
 	    pv.N1 >= pll_lim.vco1.min_n && pv.N1 <= pll_lim.vco1.max_n &&
@@ -536,7 +536,7 @@ nouveau_hw_fix_bad_vpll(struct drm_device *dev, int head)
 	pv.M1 = pll_lim.vco1.max_m;
 	pv.N1 = pll_lim.vco1.min_n;
 	pv.log2P = pll_lim.max_usable_log2p;
-	nouveau_hw_setpll(dev, pllreg, &pv);
+	nouveau_hw_setpll(dev, pll_lim.reg, &pv);
 }
 
 /*

drivers/gpu/drm/nouveau/nouveau_hw.h

@@ -415,6 +415,25 @@ nv_fix_nv40_hw_cursor(struct drm_device *dev, int head)
 	NVWriteRAMDAC(dev, head, NV_PRAMDAC_CU_START_POS, curpos);
 }
 
+static inline void
+nv_set_crtc_base(struct drm_device *dev, int head, uint32_t offset)
+{
+	struct drm_nouveau_private *dev_priv = dev->dev_private;
+
+	NVWriteCRTC(dev, head, NV_PCRTC_START, offset);
+
+	if (dev_priv->card_type == NV_04) {
+		/*
+		 * Hilarious, the 24th bit doesn't want to stick to
+		 * PCRTC_START...
+		 */
+		int cre_heb = NVReadVgaCrtc(dev, head, NV_CIO_CRE_HEB__INDEX);
+
+		NVWriteVgaCrtc(dev, head, NV_CIO_CRE_HEB__INDEX,
+			       (cre_heb & ~0x40) | ((offset >> 18) & 0x40));
+	}
+}
+
 static inline void
 nv_show_cursor(struct drm_device *dev, int head, bool show)
 {

drivers/gpu/drm/nouveau/nouveau_i2c.c

@@ -256,7 +256,7 @@ nouveau_i2c_find(struct drm_device *dev, int index)
 	if (index >= DCB_MAX_NUM_I2C_ENTRIES)
 		return NULL;
 
-	if (dev_priv->chipset >= NV_50 && (i2c->entry & 0x00000100)) {
+	if (dev_priv->card_type >= NV_50 && (i2c->entry & 0x00000100)) {
 		uint32_t reg = 0xe500, val;
 
 		if (i2c->port_type == 6) {

drivers/gpu/drm/nouveau/nouveau_irq.c

@@ -42,6 +42,13 @@
 #include "nouveau_connector.h"
 #include "nv50_display.h"
 
+static DEFINE_RATELIMIT_STATE(nouveau_ratelimit_state, 3 * HZ, 20);
+
+static int nouveau_ratelimit(void)
+{
+	return __ratelimit(&nouveau_ratelimit_state);
+}
+
 void
 nouveau_irq_preinstall(struct drm_device *dev)
 {
@@ -53,6 +60,7 @@ nouveau_irq_preinstall(struct drm_device *dev)
 	if (dev_priv->card_type >= NV_50) {
 		INIT_WORK(&dev_priv->irq_work, nv50_display_irq_handler_bh);
 		INIT_WORK(&dev_priv->hpd_work, nv50_display_irq_hotplug_bh);
+		spin_lock_init(&dev_priv->hpd_state.lock);
 		INIT_LIST_HEAD(&dev_priv->vbl_waiting);
 	}
 }
@@ -202,8 +210,8 @@ nouveau_fifo_irq_handler(struct drm_device *dev)
 		}
 
 		if (status & NV_PFIFO_INTR_DMA_PUSHER) {
-			u32 get = nv_rd32(dev, 0x003244);
-			u32 put = nv_rd32(dev, 0x003240);
+			u32 dma_get = nv_rd32(dev, 0x003244);
+			u32 dma_put = nv_rd32(dev, 0x003240);
 			u32 push = nv_rd32(dev, 0x003220);
 			u32 state = nv_rd32(dev, 0x003228);
 
@@ -213,16 +221,18 @@ nouveau_fifo_irq_handler(struct drm_device *dev)
 				u32 ib_get = nv_rd32(dev, 0x003334);
 				u32 ib_put = nv_rd32(dev, 0x003330);
 
-				NV_INFO(dev, "PFIFO_DMA_PUSHER - Ch %d Get 0x%02x%08x "
+				if (nouveau_ratelimit())
+					NV_INFO(dev, "PFIFO_DMA_PUSHER - Ch %d Get 0x%02x%08x "
 					     "Put 0x%02x%08x IbGet 0x%08x IbPut 0x%08x "
 					     "State 0x%08x Push 0x%08x\n",
-					chid, ho_get, get, ho_put, put, ib_get, ib_put,
-					state, push);
+						chid, ho_get, dma_get, ho_put,
+						dma_put, ib_get, ib_put, state,
+						push);
 
 				/* METHOD_COUNT, in DMA_STATE on earlier chipsets */
 				nv_wr32(dev, 0x003364, 0x00000000);
-				if (get != put || ho_get != ho_put) {
-					nv_wr32(dev, 0x003244, put);
+				if (dma_get != dma_put || ho_get != ho_put) {
+					nv_wr32(dev, 0x003244, dma_put);
 					nv_wr32(dev, 0x003328, ho_put);
 				} else
 				if (ib_get != ib_put) {
@@ -231,10 +241,10 @@ nouveau_fifo_irq_handler(struct drm_device *dev)
 			} else {
 				NV_INFO(dev, "PFIFO_DMA_PUSHER - Ch %d Get 0x%08x "
 					     "Put 0x%08x State 0x%08x Push 0x%08x\n",
-					chid, get, put, state, push);
+					chid, dma_get, dma_put, state, push);
 
-				if (get != put)
-					nv_wr32(dev, 0x003244, put);
+				if (dma_get != dma_put)
+					nv_wr32(dev, 0x003244, dma_put);
 			}
 
 			nv_wr32(dev, 0x003228, 0x00000000);
@@ -266,8 +276,9 @@ nouveau_fifo_irq_handler(struct drm_device *dev)
 		}
 
 		if (status) {
-			NV_INFO(dev, "PFIFO_INTR 0x%08x - Ch %d\n",
-				status, chid);
+			if (nouveau_ratelimit())
+				NV_INFO(dev, "PFIFO_INTR 0x%08x - Ch %d\n",
+					status, chid);
 			nv_wr32(dev, NV03_PFIFO_INTR_0, status);
 			status = 0;
 		}
@@ -544,13 +555,6 @@ nouveau_pgraph_intr_notify(struct drm_device *dev, uint32_t nsource)
 		nouveau_graph_dump_trap_info(dev, "PGRAPH_NOTIFY", &trap);
 }
 
-static DEFINE_RATELIMIT_STATE(nouveau_ratelimit_state, 3 * HZ, 20);
-
-static int nouveau_ratelimit(void)
-{
-	return __ratelimit(&nouveau_ratelimit_state);
-}
-
 
 static inline void
 nouveau_pgraph_intr_error(struct drm_device *dev, uint32_t nsource)

drivers/gpu/drm/nouveau/nouveau_mem.c

@@ -33,9 +33,9 @@
 #include "drmP.h"
 #include "drm.h"
 #include "drm_sarea.h"
-#include "nouveau_drv.h"
 
-#define MIN(a,b) a < b ? a : b
+#include "nouveau_drv.h"
+#include "nouveau_pm.h"
 
 /*
  * NV10-NV40 tiling helpers
@@ -175,11 +175,10 @@ nv50_mem_vm_bind_linear(struct drm_device *dev, uint64_t virt, uint32_t size,
 			}
 		}
 	}
-	dev_priv->engine.instmem.flush(dev);
 
-	nv50_vm_flush(dev, 5);
-	nv50_vm_flush(dev, 0);
-	nv50_vm_flush(dev, 4);
+	dev_priv->engine.instmem.flush(dev);
+	dev_priv->engine.fifo.tlb_flush(dev);
+	dev_priv->engine.graph.tlb_flush(dev);
 	nv50_vm_flush(dev, 6);
 	return 0;
 }
@@ -209,11 +208,10 @@ nv50_mem_vm_unbind(struct drm_device *dev, uint64_t virt, uint32_t size)
 			pte++;
 		}
 	}
-	dev_priv->engine.instmem.flush(dev);
 
-	nv50_vm_flush(dev, 5);
-	nv50_vm_flush(dev, 0);
-	nv50_vm_flush(dev, 4);
+	dev_priv->engine.instmem.flush(dev);
+	dev_priv->engine.fifo.tlb_flush(dev);
+	dev_priv->engine.graph.tlb_flush(dev);
 	nv50_vm_flush(dev, 6);
 }
 
@@ -653,6 +651,7 @@ nouveau_mem_gart_init(struct drm_device *dev)
 void
 nouveau_mem_timing_init(struct drm_device *dev)
 {
+	/* cards < NVC0 only */
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_pm_engine *pm = &dev_priv->engine.pm;
 	struct nouveau_pm_memtimings *memtimings = &pm->memtimings;
@@ -719,14 +718,14 @@ nouveau_mem_timing_init(struct drm_device *dev)
 		tUNK_19 = 1;
 		tUNK_20 = 0;
 		tUNK_21 = 0;
-		switch (MIN(recordlen,21)) {
-		case 21:
+		switch (min(recordlen, 22)) {
+		case 22:
 			tUNK_21 = entry[21];
-		case 20:
+		case 21:
 			tUNK_20 = entry[20];
-		case 19:
+		case 20:
 			tUNK_19 = entry[19];
-		case 18:
+		case 19:
 			tUNK_18 = entry[18];
 		default:
 			tUNK_0  = entry[0];
@@ -756,24 +755,30 @@ nouveau_mem_timing_init(struct drm_device *dev)
 		timing->reg_100228 = (tUNK_12 << 16 | tUNK_11 << 8 | tUNK_10);
 		if(recordlen > 19) {
 			timing->reg_100228 += (tUNK_19 - 1) << 24;
-		} else {
+		}/* I cannot back-up this else-statement right now
+			 else {
 			timing->reg_100228 += tUNK_12 << 24;
-		}
+		}*/
 
 		/* XXX: reg_10022c */
+		timing->reg_10022c = tUNK_2 - 1;
 
 		timing->reg_100230 = (tUNK_20 << 24 | tUNK_21 << 16 |
 				      tUNK_13 << 8  | tUNK_13);
 
 		/* XXX: +6? */
 		timing->reg_100234 = (tRAS << 24 | (tUNK_19 + 6) << 8 | tRC);
-		if(tUNK_10 > tUNK_11) {
-			timing->reg_100234 += tUNK_10 << 16;
-		} else {
-			timing->reg_100234 += tUNK_11 << 16;
+		timing->reg_100234 += max(tUNK_10,tUNK_11) << 16;
+
+		/* XXX; reg_100238, reg_10023c
+		 * reg: 0x00??????
+		 * reg_10023c:
+		 *      0 for pre-NV50 cards
+		 *      0x????0202 for NV50+ cards (empirical evidence) */
+		if(dev_priv->card_type >= NV_50) {
+			timing->reg_10023c = 0x202;
 		}
 
-		/* XXX; reg_100238, reg_10023c */
 		NV_DEBUG(dev, "Entry %d: 220: %08x %08x %08x %08x\n", i,
 			 timing->reg_100220, timing->reg_100224,
 			 timing->reg_100228, timing->reg_10022c);

drivers/gpu/drm/nouveau/nouveau_object.c

@@ -129,7 +129,7 @@ nouveau_gpuobj_new(struct drm_device *dev, struct nouveau_channel *chan,
 			if (ramin == NULL) {
 				spin_unlock(&dev_priv->ramin_lock);
 				nouveau_gpuobj_ref(NULL, &gpuobj);
-				return ret;
+				return -ENOMEM;
 			}
 
 			ramin = drm_mm_get_block_atomic(ramin, size, align);

drivers/gpu/drm/nouveau/nouveau_pm.c

@@ -284,6 +284,7 @@ nouveau_sysfs_fini(struct drm_device *dev)
 	}
 }
 
+#ifdef CONFIG_HWMON
 static ssize_t
 nouveau_hwmon_show_temp(struct device *d, struct device_attribute *a, char *buf)
 {
@@ -395,10 +396,12 @@ static struct attribute *hwmon_attributes[] = {
 static const struct attribute_group hwmon_attrgroup = {
 	.attrs = hwmon_attributes,
 };
+#endif
 
 static int
 nouveau_hwmon_init(struct drm_device *dev)
 {
+#ifdef CONFIG_HWMON
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_pm_engine *pm = &dev_priv->engine.pm;
 	struct device *hwmon_dev;
@@ -425,13 +428,14 @@ nouveau_hwmon_init(struct drm_device *dev)
 	}
 
 	pm->hwmon = hwmon_dev;
-
+#endif
 	return 0;
 }
 
 static void
 nouveau_hwmon_fini(struct drm_device *dev)
 {
+#ifdef CONFIG_HWMON
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_pm_engine *pm = &dev_priv->engine.pm;
 
@@ -439,6 +443,7 @@ nouveau_hwmon_fini(struct drm_device *dev)
 		sysfs_remove_group(&pm->hwmon->kobj, &hwmon_attrgroup);
 		hwmon_device_unregister(pm->hwmon);
 	}
+#endif
 }
 
 int

drivers/gpu/drm/nouveau/nouveau_ramht.c

@@ -153,26 +153,42 @@ nouveau_ramht_insert(struct nouveau_channel *chan, u32 handle,
 	return -ENOMEM;
 }
 
+static struct nouveau_ramht_entry *
+nouveau_ramht_remove_entry(struct nouveau_channel *chan, u32 handle)
+{
+	struct nouveau_ramht *ramht = chan ? chan->ramht : NULL;
+	struct nouveau_ramht_entry *entry;
+	unsigned long flags;
+
+	if (!ramht)
+		return NULL;
+
+	spin_lock_irqsave(&ramht->lock, flags);
+	list_for_each_entry(entry, &ramht->entries, head) {
+		if (entry->channel == chan &&
+		    (!handle || entry->handle == handle)) {
+			list_del(&entry->head);
+			spin_unlock_irqrestore(&ramht->lock, flags);
+
+			return entry;
+		}
+	}
+	spin_unlock_irqrestore(&ramht->lock, flags);
+
+	return NULL;
+}
+
 static void
-nouveau_ramht_remove_locked(struct nouveau_channel *chan, u32 handle)
+nouveau_ramht_remove_hash(struct nouveau_channel *chan, u32 handle)
 {
 	struct drm_device *dev = chan->dev;
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_instmem_engine *instmem = &dev_priv->engine.instmem;
 	struct nouveau_gpuobj *ramht = chan->ramht->gpuobj;
-	struct nouveau_ramht_entry *entry, *tmp;
+	unsigned long flags;
 	u32 co, ho;
 
-	list_for_each_entry_safe(entry, tmp, &chan->ramht->entries, head) {
-		if (entry->channel != chan || entry->handle != handle)
-			continue;
-
-		nouveau_gpuobj_ref(NULL, &entry->gpuobj);
-		list_del(&entry->head);
-		kfree(entry);
-		break;
-	}
-
+	spin_lock_irqsave(&chan->ramht->lock, flags);
 	co = ho = nouveau_ramht_hash_handle(chan, handle);
 	do {
 		if (nouveau_ramht_entry_valid(dev, ramht, co) &&
@@ -184,7 +200,7 @@ nouveau_ramht_remove_locked(struct nouveau_channel *chan, u32 handle)
 			nv_wo32(ramht, co + 0, 0x00000000);
 			nv_wo32(ramht, co + 4, 0x00000000);
 			instmem->flush(dev);
-			return;
+			goto out;
 		}
 
 		co += 8;
@@ -194,17 +210,22 @@ nouveau_ramht_remove_locked(struct nouveau_channel *chan, u32 handle)
 
 	NV_ERROR(dev, "RAMHT entry not found. ch=%d, handle=0x%08x\n",
 		 chan->id, handle);
+out:
+	spin_unlock_irqrestore(&chan->ramht->lock, flags);
 }
 
 void
 nouveau_ramht_remove(struct nouveau_channel *chan, u32 handle)
 {
-	struct nouveau_ramht *ramht = chan->ramht;
-	unsigned long flags;
+	struct nouveau_ramht_entry *entry;
 
-	spin_lock_irqsave(&ramht->lock, flags);
-	nouveau_ramht_remove_locked(chan, handle);
-	spin_unlock_irqrestore(&ramht->lock, flags);
+	entry = nouveau_ramht_remove_entry(chan, handle);
+	if (!entry)
+		return;
+
+	nouveau_ramht_remove_hash(chan, entry->handle);
+	nouveau_gpuobj_ref(NULL, &entry->gpuobj);
+	kfree(entry);
 }
 
 struct nouveau_gpuobj *
@@ -265,23 +286,19 @@ void
 nouveau_ramht_ref(struct nouveau_ramht *ref, struct nouveau_ramht **ptr,
 		  struct nouveau_channel *chan)
 {
-	struct nouveau_ramht_entry *entry, *tmp;
+	struct nouveau_ramht_entry *entry;
 	struct nouveau_ramht *ramht;
-	unsigned long flags;
 
 	if (ref)
 		kref_get(&ref->refcount);
 
 	ramht = *ptr;
 	if (ramht) {
-		spin_lock_irqsave(&ramht->lock, flags);
-		list_for_each_entry_safe(entry, tmp, &ramht->entries, head) {
-			if (entry->channel != chan)
-				continue;
-
-			nouveau_ramht_remove_locked(chan, entry->handle);
+		while ((entry = nouveau_ramht_remove_entry(chan, 0))) {
+			nouveau_ramht_remove_hash(chan, entry->handle);
+			nouveau_gpuobj_ref(NULL, &entry->gpuobj);
+			kfree(entry);
 		}
-		spin_unlock_irqrestore(&ramht->lock, flags);
 
 		kref_put(&ramht->refcount, nouveau_ramht_del);
 	}

drivers/gpu/drm/nouveau/nouveau_sgdma.c

@@ -120,8 +120,8 @@ nouveau_sgdma_bind(struct ttm_backend *be, struct ttm_mem_reg *mem)
 	dev_priv->engine.instmem.flush(nvbe->dev);
 
 	if (dev_priv->card_type == NV_50) {
-		nv50_vm_flush(dev, 5); /* PGRAPH */
-		nv50_vm_flush(dev, 0); /* PFIFO */
+		dev_priv->engine.fifo.tlb_flush(dev);
+		dev_priv->engine.graph.tlb_flush(dev);
 	}
 
 	nvbe->bound = true;
@@ -162,8 +162,8 @@ nouveau_sgdma_unbind(struct ttm_backend *be)
 	dev_priv->engine.instmem.flush(nvbe->dev);
 
 	if (dev_priv->card_type == NV_50) {
-		nv50_vm_flush(dev, 5);
-		nv50_vm_flush(dev, 0);
+		dev_priv->engine.fifo.tlb_flush(dev);
+		dev_priv->engine.graph.tlb_flush(dev);
 	}
 
 	nvbe->bound = false;
@@ -224,7 +224,11 @@ nouveau_sgdma_init(struct drm_device *dev)
 	int i, ret;
 
 	if (dev_priv->card_type < NV_50) {
-		aper_size = (64 * 1024 * 1024);
+		if(dev_priv->ramin_rsvd_vram < 2 * 1024 * 1024)
+			aper_size = 64 * 1024 * 1024;
+		else
+			aper_size = 512 * 1024 * 1024;
+
 		obj_size  = (aper_size >> NV_CTXDMA_PAGE_SHIFT) * 4;
 		obj_size += 8; /* ctxdma header */
 	} else {

drivers/gpu/drm/nouveau/nouveau_state.c

@@ -354,6 +354,15 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
 		engine->graph.destroy_context	= nv50_graph_destroy_context;
 		engine->graph.load_context	= nv50_graph_load_context;
 		engine->graph.unload_context	= nv50_graph_unload_context;
+		if (dev_priv->chipset != 0x86)
+			engine->graph.tlb_flush	= nv50_graph_tlb_flush;
+		else {
+			/* from what i can see nvidia do this on every
+			 * pre-NVA3 board except NVAC, but, we've only
+			 * ever seen problems on NV86
+			 */
+			engine->graph.tlb_flush	= nv86_graph_tlb_flush;
+		}
 		engine->fifo.channels		= 128;
 		engine->fifo.init		= nv50_fifo_init;
 		engine->fifo.takedown		= nv50_fifo_takedown;
@@ -365,6 +374,7 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
 		engine->fifo.destroy_context	= nv50_fifo_destroy_context;
 		engine->fifo.load_context	= nv50_fifo_load_context;
 		engine->fifo.unload_context	= nv50_fifo_unload_context;
+		engine->fifo.tlb_flush		= nv50_fifo_tlb_flush;
 		engine->display.early_init	= nv50_display_early_init;
 		engine->display.late_takedown	= nv50_display_late_takedown;
 		engine->display.create		= nv50_display_create;
@@ -1041,6 +1051,9 @@ int nouveau_ioctl_getparam(struct drm_device *dev, void *data,
 	case NOUVEAU_GETPARAM_PTIMER_TIME:
 		getparam->value = dev_priv->engine.timer.read(dev);
 		break;
+	case NOUVEAU_GETPARAM_HAS_BO_USAGE:
+		getparam->value = 1;
+		break;
 	case NOUVEAU_GETPARAM_GRAPH_UNITS:
 		/* NV40 and NV50 versions are quite different, but register
 		 * address is the same. User is supposed to know the card
@@ -1051,7 +1064,7 @@ int nouveau_ioctl_getparam(struct drm_device *dev, void *data,
 		}
 		/* FALLTHRU */
 	default:
-		NV_ERROR(dev, "unknown parameter %lld\n", getparam->param);
+		NV_DEBUG(dev, "unknown parameter %lld\n", getparam->param);
 		return -EINVAL;
 	}
 
@@ -1066,7 +1079,7 @@ nouveau_ioctl_setparam(struct drm_device *dev, void *data,
 
 	switch (setparam->param) {
 	default:
-		NV_ERROR(dev, "unknown parameter %lld\n", setparam->param);
+		NV_DEBUG(dev, "unknown parameter %lld\n", setparam->param);
 		return -EINVAL;
 	}
 

drivers/gpu/drm/nouveau/nouveau_temp.c

@@ -191,7 +191,7 @@ nv40_temp_get(struct drm_device *dev)
 	int offset = sensor->offset_mult / sensor->offset_div;
 	int core_temp;
 
-	if (dev_priv->chipset >= 0x50) {
+	if (dev_priv->card_type >= NV_50) {
 		core_temp = nv_rd32(dev, 0x20008);
 	} else {
 		core_temp = nv_rd32(dev, 0x0015b4) & 0x1fff;

drivers/gpu/drm/nouveau/nv04_crtc.c

@@ -158,7 +158,6 @@ nv_crtc_dpms(struct drm_crtc *crtc, int mode)
 {
 	struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
-	struct drm_connector *connector;
 	unsigned char seq1 = 0, crtc17 = 0;
 	unsigned char crtc1A;
 
@@ -213,10 +212,6 @@ nv_crtc_dpms(struct drm_crtc *crtc, int mode)
 	NVVgaSeqReset(dev, nv_crtc->index, false);
 
 	NVWriteVgaCrtc(dev, nv_crtc->index, NV_CIO_CRE_RPC1_INDEX, crtc1A);
-
-	/* Update connector polling modes */
-	list_for_each_entry(connector, &dev->mode_config.connector_list, head)
-		nouveau_connector_set_polling(connector);
 }
 
 static bool
@@ -831,7 +826,7 @@ nv04_crtc_do_mode_set_base(struct drm_crtc *crtc,
 	/* Update the framebuffer location. */
 	regp->fb_start = nv_crtc->fb.offset & ~3;
 	regp->fb_start += (y * drm_fb->pitch) + (x * drm_fb->bits_per_pixel / 8);
-	NVWriteCRTC(dev, nv_crtc->index, NV_PCRTC_START, regp->fb_start);
+	nv_set_crtc_base(dev, nv_crtc->index, regp->fb_start);
 
 	/* Update the arbitration parameters. */
 	nouveau_calc_arb(dev, crtc->mode.clock, drm_fb->bits_per_pixel,

drivers/gpu/drm/nouveau/nv04_dfp.c

@@ -185,14 +185,15 @@ static bool nv04_dfp_mode_fixup(struct drm_encoder *encoder,
 	struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
 	struct nouveau_connector *nv_connector = nouveau_encoder_connector_get(nv_encoder);
 
-	/* For internal panels and gpu scaling on DVI we need the native mode */
-	if (nv_connector->scaling_mode != DRM_MODE_SCALE_NONE) {
-		if (!nv_connector->native_mode)
-			return false;
+	if (!nv_connector->native_mode ||
+	    nv_connector->scaling_mode == DRM_MODE_SCALE_NONE ||
+	    mode->hdisplay > nv_connector->native_mode->hdisplay ||
+	    mode->vdisplay > nv_connector->native_mode->vdisplay) {
+		nv_encoder->mode = *adjusted_mode;
+
+	} else {
 		nv_encoder->mode = *nv_connector->native_mode;
 		adjusted_mode->clock = nv_connector->native_mode->clock;
-	} else {
-		nv_encoder->mode = *adjusted_mode;
 	}
 
 	return true;

drivers/gpu/drm/nouveau/nv04_pm.c

@@ -76,6 +76,15 @@ nv04_pm_clock_set(struct drm_device *dev, void *pre_state)
 		reg += 4;
 
 	nouveau_hw_setpll(dev, reg, &state->calc);
+
+	if (dev_priv->card_type < NV_30 && reg == NV_PRAMDAC_MPLL_COEFF) {
+		if (dev_priv->card_type == NV_20)
+			nv_mask(dev, 0x1002c4, 0, 1 << 20);
+
+		/* Reset the DLLs */
+		nv_mask(dev, 0x1002c0, 0, 1 << 8);
+	}
+
 	kfree(state);
 }
 

drivers/gpu/drm/nouveau/nv50_calc.c

@@ -51,24 +51,28 @@ nv50_calc_pll2(struct drm_device *dev, struct pll_lims *pll, int clk,
 	       int *N, int *fN, int *M, int *P)
 {
 	fixed20_12 fb_div, a, b;
+	u32 refclk = pll->refclk / 10;
+	u32 max_vco_freq = pll->vco1.maxfreq / 10;
+	u32 max_vco_inputfreq = pll->vco1.max_inputfreq / 10;
+	clk /= 10;
 
-	*P = pll->vco1.maxfreq / clk;
+	*P = max_vco_freq / clk;
 	if (*P > pll->max_p)
 		*P = pll->max_p;
 	if (*P < pll->min_p)
 		*P = pll->min_p;
 
-	/* *M = ceil(refclk / pll->vco.max_inputfreq); */
-	a.full = dfixed_const(pll->refclk);
-	b.full = dfixed_const(pll->vco1.max_inputfreq);
+	/* *M = floor((refclk + max_vco_inputfreq) / max_vco_inputfreq); */
+	a.full = dfixed_const(refclk + max_vco_inputfreq);
+	b.full = dfixed_const(max_vco_inputfreq);
 	a.full = dfixed_div(a, b);
-	a.full = dfixed_ceil(a);
+	a.full = dfixed_floor(a);
 	*M = dfixed_trunc(a);
 
 	/* fb_div = (vco * *M) / refclk; */
 	fb_div.full = dfixed_const(clk * *P);
 	fb_div.full = dfixed_mul(fb_div, a);
-	a.full = dfixed_const(pll->refclk);
+	a.full = dfixed_const(refclk);
 	fb_div.full = dfixed_div(fb_div, a);
 
 	/* *N = floor(fb_div); */

drivers/gpu/drm/nouveau/nv50_crtc.c

@@ -546,7 +546,7 @@ nv50_crtc_do_mode_set_base(struct drm_crtc *crtc,
 	}
 
 	nv_crtc->fb.offset = fb->nvbo->bo.offset - dev_priv->vm_vram_base;
-	nv_crtc->fb.tile_flags = fb->nvbo->tile_flags;
+	nv_crtc->fb.tile_flags = nouveau_bo_tile_layout(fb->nvbo);
 	nv_crtc->fb.cpp = drm_fb->bits_per_pixel / 8;
 	if (!nv_crtc->fb.blanked && dev_priv->chipset != 0x50) {
 		ret = RING_SPACE(evo, 2);
@@ -578,7 +578,7 @@ nv50_crtc_do_mode_set_base(struct drm_crtc *crtc,
 				  fb->nvbo->tile_mode);
 	}
 	if (dev_priv->chipset == 0x50)
-		OUT_RING(evo, (fb->nvbo->tile_flags << 8) | format);
+		OUT_RING(evo, (nv_crtc->fb.tile_flags << 8) | format);
 	else
 		OUT_RING(evo, format);
 

drivers/gpu/drm/nouveau/nv50_display.c

@@ -1032,11 +1032,18 @@ nv50_display_irq_hotplug_bh(struct work_struct *work)
 	struct drm_connector *connector;
 	const uint32_t gpio_reg[4] = { 0xe104, 0xe108, 0xe280, 0xe284 };
 	uint32_t unplug_mask, plug_mask, change_mask;
-	uint32_t hpd0, hpd1 = 0;
+	uint32_t hpd0, hpd1;
 
-	hpd0 = nv_rd32(dev, 0xe054) & nv_rd32(dev, 0xe050);
+	spin_lock_irq(&dev_priv->hpd_state.lock);
+	hpd0 = dev_priv->hpd_state.hpd0_bits;
+	dev_priv->hpd_state.hpd0_bits = 0;
+	hpd1 = dev_priv->hpd_state.hpd1_bits;
+	dev_priv->hpd_state.hpd1_bits = 0;
+	spin_unlock_irq(&dev_priv->hpd_state.lock);
+
+	hpd0 &= nv_rd32(dev, 0xe050);
 	if (dev_priv->chipset >= 0x90)
-		hpd1 = nv_rd32(dev, 0xe074) & nv_rd32(dev, 0xe070);
+		hpd1 &= nv_rd32(dev, 0xe070);
 
 	plug_mask   = (hpd0 & 0x0000ffff) | (hpd1 << 16);
 	unplug_mask = (hpd0 >> 16) | (hpd1 & 0xffff0000);
@@ -1078,10 +1085,6 @@ nv50_display_irq_hotplug_bh(struct work_struct *work)
 			helper->dpms(connector->encoder, DRM_MODE_DPMS_OFF);
 	}
 
-	nv_wr32(dev, 0xe054, nv_rd32(dev, 0xe054));
-	if (dev_priv->chipset >= 0x90)
-		nv_wr32(dev, 0xe074, nv_rd32(dev, 0xe074));
-
 	drm_helper_hpd_irq_event(dev);
 }
 
@@ -1092,8 +1095,22 @@ nv50_display_irq_handler(struct drm_device *dev)
 	uint32_t delayed = 0;
 
 	if (nv_rd32(dev, NV50_PMC_INTR_0) & NV50_PMC_INTR_0_HOTPLUG) {
-		if (!work_pending(&dev_priv->hpd_work))
-			queue_work(dev_priv->wq, &dev_priv->hpd_work);
+		uint32_t hpd0_bits, hpd1_bits = 0;
+
+		hpd0_bits = nv_rd32(dev, 0xe054);
+		nv_wr32(dev, 0xe054, hpd0_bits);
+
+		if (dev_priv->chipset >= 0x90) {
+			hpd1_bits = nv_rd32(dev, 0xe074);
+			nv_wr32(dev, 0xe074, hpd1_bits);
+		}
+
+		spin_lock(&dev_priv->hpd_state.lock);
+		dev_priv->hpd_state.hpd0_bits |= hpd0_bits;
+		dev_priv->hpd_state.hpd1_bits |= hpd1_bits;
+		spin_unlock(&dev_priv->hpd_state.lock);
+
+		queue_work(dev_priv->wq, &dev_priv->hpd_work);
 	}
 
 	while (nv_rd32(dev, NV50_PMC_INTR_0) & NV50_PMC_INTR_0_DISPLAY) {

drivers/gpu/drm/nouveau/nv50_fifo.c

@@ -464,3 +464,8 @@ nv50_fifo_unload_context(struct drm_device *dev)
 	return 0;
 }
 
+void
+nv50_fifo_tlb_flush(struct drm_device *dev)
+{
+	nv50_vm_flush(dev, 5);
+}

drivers/gpu/drm/nouveau/nv50_graph.c

@@ -402,3 +402,55 @@ struct nouveau_pgraph_object_class nv50_graph_grclass[] = {
 	{ 0x8597, false, NULL }, /* tesla (nva3, nva5, nva8) */
 	{}
 };
+
+void
+nv50_graph_tlb_flush(struct drm_device *dev)
+{
+	nv50_vm_flush(dev, 0);
+}
+
+void
+nv86_graph_tlb_flush(struct drm_device *dev)
+{
+	struct drm_nouveau_private *dev_priv = dev->dev_private;
+	struct nouveau_timer_engine *ptimer = &dev_priv->engine.timer;
+	bool idle, timeout = false;
+	unsigned long flags;
+	u64 start;
+	u32 tmp;
+
+	spin_lock_irqsave(&dev_priv->context_switch_lock, flags);
+	nv_mask(dev, 0x400500, 0x00000001, 0x00000000);
+
+	start = ptimer->read(dev);
+	do {
+		idle = true;
+
+		for (tmp = nv_rd32(dev, 0x400380); tmp && idle; tmp >>= 3) {
+			if ((tmp & 7) == 1)
+				idle = false;
+		}
+
+		for (tmp = nv_rd32(dev, 0x400384); tmp && idle; tmp >>= 3) {
+			if ((tmp & 7) == 1)
+				idle = false;
+		}
+
+		for (tmp = nv_rd32(dev, 0x400388); tmp && idle; tmp >>= 3) {
+			if ((tmp & 7) == 1)
+				idle = false;
+		}
+	} while (!idle && !(timeout = ptimer->read(dev) - start > 2000000000));
+
+	if (timeout) {
+		NV_ERROR(dev, "PGRAPH TLB flush idle timeout fail: "
+			      "0x%08x 0x%08x 0x%08x 0x%08x\n",
+			 nv_rd32(dev, 0x400700), nv_rd32(dev, 0x400380),
+			 nv_rd32(dev, 0x400384), nv_rd32(dev, 0x400388));
+	}
+
+	nv50_vm_flush(dev, 0);
+
+	nv_mask(dev, 0x400500, 0x00000001, 0x00000001);
+	spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags);
+}

drivers/gpu/drm/nouveau/nv50_instmem.c

@@ -402,7 +402,6 @@ nv50_instmem_bind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj)
 	}
 	dev_priv->engine.instmem.flush(dev);
 
-	nv50_vm_flush(dev, 4);
 	nv50_vm_flush(dev, 6);
 
 	gpuobj->im_bound = 1;

drivers/gpu/drm/radeon/evergreen.c

@@ -1650,7 +1650,36 @@ static void evergreen_gpu_init(struct radeon_device *rdev)
 		}
 	}
 
-	rdev->config.evergreen.tile_config = gb_addr_config;
+	/* setup tiling info dword.  gb_addr_config is not adequate since it does
+	 * not have bank info, so create a custom tiling dword.
+	 * bits 3:0   num_pipes
+	 * bits 7:4   num_banks
+	 * bits 11:8  group_size
+	 * bits 15:12 row_size
+	 */
+	rdev->config.evergreen.tile_config = 0;
+	switch (rdev->config.evergreen.max_tile_pipes) {
+	case 1:
+	default:
+		rdev->config.evergreen.tile_config |= (0 << 0);
+		break;
+	case 2:
+		rdev->config.evergreen.tile_config |= (1 << 0);
+		break;
+	case 4:
+		rdev->config.evergreen.tile_config |= (2 << 0);
+		break;
+	case 8:
+		rdev->config.evergreen.tile_config |= (3 << 0);
+		break;
+	}
+	rdev->config.evergreen.tile_config |=
+		((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) << 4;
+	rdev->config.evergreen.tile_config |=
+		((mc_arb_ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT) << 8;
+	rdev->config.evergreen.tile_config |=
+		((gb_addr_config & 0x30000000) >> 28) << 12;
+
 	WREG32(GB_BACKEND_MAP, gb_backend_map);
 	WREG32(GB_ADDR_CONFIG, gb_addr_config);
 	WREG32(DMIF_ADDR_CONFIG, gb_addr_config);

drivers/gpu/drm/radeon/evergreen_blit_kms.c

@@ -459,7 +459,7 @@ int evergreen_blit_init(struct radeon_device *rdev)
 	obj_size += evergreen_ps_size * 4;
 	obj_size = ALIGN(obj_size, 256);
 
-	r = radeon_bo_create(rdev, NULL, obj_size, true, RADEON_GEM_DOMAIN_VRAM,
+	r = radeon_bo_create(rdev, NULL, obj_size, PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM,
 				&rdev->r600_blit.shader_obj);
 	if (r) {
 		DRM_ERROR("evergreen failed to allocate shader\n");

drivers/gpu/drm/radeon/r600.c

@@ -2718,7 +2718,7 @@ static int r600_ih_ring_alloc(struct radeon_device *rdev)
 	/* Allocate ring buffer */
 	if (rdev->ih.ring_obj == NULL) {
 		r = radeon_bo_create(rdev, NULL, rdev->ih.ring_size,
-				     true,
+				     PAGE_SIZE, true,
 				     RADEON_GEM_DOMAIN_GTT,
 				     &rdev->ih.ring_obj);
 		if (r) {

drivers/gpu/drm/radeon/r600_blit_kms.c

@@ -501,7 +501,7 @@ int r600_blit_init(struct radeon_device *rdev)
 	obj_size += r6xx_ps_size * 4;
 	obj_size = ALIGN(obj_size, 256);
 
-	r = radeon_bo_create(rdev, NULL, obj_size, true, RADEON_GEM_DOMAIN_VRAM,
+	r = radeon_bo_create(rdev, NULL, obj_size, PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM,
 				&rdev->r600_blit.shader_obj);
 	if (r) {
 		DRM_ERROR("r600 failed to allocate shader\n");

drivers/gpu/drm/radeon/r600_cs.c

@@ -50,6 +50,7 @@ struct r600_cs_track {
 	u32			nsamples;
 	u32			cb_color_base_last[8];
 	struct radeon_bo	*cb_color_bo[8];
+	u64			cb_color_bo_mc[8];
 	u32			cb_color_bo_offset[8];
 	struct radeon_bo	*cb_color_frag_bo[8];
 	struct radeon_bo	*cb_color_tile_bo[8];
@@ -67,6 +68,7 @@ struct r600_cs_track {
 	u32			db_depth_size;
 	u32			db_offset;
 	struct radeon_bo	*db_bo;
+	u64			db_bo_mc;
 };
 
 static inline int r600_bpe_from_format(u32 *bpe, u32 format)
@@ -140,6 +142,68 @@ static inline int r600_bpe_from_format(u32 *bpe, u32 format)
 	return 0;
 }
 
+struct array_mode_checker {
+	int array_mode;
+	u32 group_size;
+	u32 nbanks;
+	u32 npipes;
+	u32 nsamples;
+	u32 bpe;
+};
+
+/* returns alignment in pixels for pitch/height/depth and bytes for base */
+static inline int r600_get_array_mode_alignment(struct array_mode_checker *values,
+						u32 *pitch_align,
+						u32 *height_align,
+						u32 *depth_align,
+						u64 *base_align)
+{
+	u32 tile_width = 8;
+	u32 tile_height = 8;
+	u32 macro_tile_width = values->nbanks;
+	u32 macro_tile_height = values->npipes;
+	u32 tile_bytes = tile_width * tile_height * values->bpe * values->nsamples;
+	u32 macro_tile_bytes = macro_tile_width * macro_tile_height * tile_bytes;
+
+	switch (values->array_mode) {
+	case ARRAY_LINEAR_GENERAL:
+		/* technically tile_width/_height for pitch/height */
+		*pitch_align = 1; /* tile_width */
+		*height_align = 1; /* tile_height */
+		*depth_align = 1;
+		*base_align = 1;
+		break;
+	case ARRAY_LINEAR_ALIGNED:
+		*pitch_align = max((u32)64, (u32)(values->group_size / values->bpe));
+		*height_align = tile_height;
+		*depth_align = 1;
+		*base_align = values->group_size;
+		break;
+	case ARRAY_1D_TILED_THIN1:
+		*pitch_align = max((u32)tile_width,
+				   (u32)(values->group_size /
+					 (tile_height * values->bpe * values->nsamples)));
+		*height_align = tile_height;
+		*depth_align = 1;
+		*base_align = values->group_size;
+		break;
+	case ARRAY_2D_TILED_THIN1:
+		*pitch_align = max((u32)macro_tile_width,
+				  (u32)(((values->group_size / tile_height) /
+					 (values->bpe * values->nsamples)) *
+					values->nbanks)) * tile_width;
+		*height_align = macro_tile_height * tile_height;
+		*depth_align = 1;
+		*base_align = max(macro_tile_bytes,
+				  (*pitch_align) * values->bpe * (*height_align) * values->nsamples);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
 static void r600_cs_track_init(struct r600_cs_track *track)
 {
 	int i;
@@ -153,10 +217,12 @@ static void r600_cs_track_init(struct r600_cs_track *track)
 		track->cb_color_info[i] = 0;
 		track->cb_color_bo[i] = NULL;
 		track->cb_color_bo_offset[i] = 0xFFFFFFFF;
+		track->cb_color_bo_mc[i] = 0xFFFFFFFF;
 	}
 	track->cb_target_mask = 0xFFFFFFFF;
 	track->cb_shader_mask = 0xFFFFFFFF;
 	track->db_bo = NULL;
+	track->db_bo_mc = 0xFFFFFFFF;
 	/* assume the biggest format and that htile is enabled */
 	track->db_depth_info = 7 | (1 << 25);
 	track->db_depth_view = 0xFFFFC000;
@@ -168,7 +234,10 @@ static void r600_cs_track_init(struct r600_cs_track *track)
 static inline int r600_cs_track_validate_cb(struct radeon_cs_parser *p, int i)
 {
 	struct r600_cs_track *track = p->track;
-	u32 bpe = 0, pitch, slice_tile_max, size, tmp, height, pitch_align;
+	u32 bpe = 0, slice_tile_max, size, tmp;
+	u32 height, height_align, pitch, pitch_align, depth_align;
+	u64 base_offset, base_align;
+	struct array_mode_checker array_check;
 	volatile u32 *ib = p->ib->ptr;
 	unsigned array_mode;
 
@@ -183,60 +252,40 @@ static inline int r600_cs_track_validate_cb(struct radeon_cs_parser *p, int i)
 			i, track->cb_color_info[i]);
 		return -EINVAL;
 	}
-	/* pitch is the number of 8x8 tiles per row */
-	pitch = G_028060_PITCH_TILE_MAX(track->cb_color_size[i]) + 1;
+	/* pitch in pixels */
+	pitch = (G_028060_PITCH_TILE_MAX(track->cb_color_size[i]) + 1) * 8;
 	slice_tile_max = G_028060_SLICE_TILE_MAX(track->cb_color_size[i]) + 1;
 	slice_tile_max *= 64;
-	height = slice_tile_max / (pitch * 8);
+	height = slice_tile_max / pitch;
 	if (height > 8192)
 		height = 8192;
 	array_mode = G_0280A0_ARRAY_MODE(track->cb_color_info[i]);
+
+	base_offset = track->cb_color_bo_mc[i] + track->cb_color_bo_offset[i];
+	array_check.array_mode = array_mode;
+	array_check.group_size = track->group_size;
+	array_check.nbanks = track->nbanks;
+	array_check.npipes = track->npipes;
+	array_check.nsamples = track->nsamples;
+	array_check.bpe = bpe;
+	if (r600_get_array_mode_alignment(&array_check,
+					  &pitch_align, &height_align, &depth_align, &base_align)) {
+		dev_warn(p->dev, "%s invalid tiling %d for %d (0x%08X)\n", __func__,
+			 G_0280A0_ARRAY_MODE(track->cb_color_info[i]), i,
+			 track->cb_color_info[i]);
+		return -EINVAL;
+	}
 	switch (array_mode) {
 	case V_0280A0_ARRAY_LINEAR_GENERAL:
-		/* technically height & 0x7 */
 		break;
 	case V_0280A0_ARRAY_LINEAR_ALIGNED:
-		pitch_align = max((u32)64, (u32)(track->group_size / bpe)) / 8;
-		if (!IS_ALIGNED(pitch, pitch_align)) {
-			dev_warn(p->dev, "%s:%d cb pitch (%d) invalid\n",
-				 __func__, __LINE__, pitch);
-			return -EINVAL;
-		}
-		if (!IS_ALIGNED(height, 8)) {
-			dev_warn(p->dev, "%s:%d cb height (%d) invalid\n",
-				 __func__, __LINE__, height);
-			return -EINVAL;
-		}
 		break;
 	case V_0280A0_ARRAY_1D_TILED_THIN1:
-		pitch_align = max((u32)8, (u32)(track->group_size / (8 * bpe * track->nsamples))) / 8;
-		if (!IS_ALIGNED(pitch, pitch_align)) {
-			dev_warn(p->dev, "%s:%d cb pitch (%d) invalid\n",
-				 __func__, __LINE__, pitch);
-			return -EINVAL;
-		}
 		/* avoid breaking userspace */
 		if (height > 7)
 			height &= ~0x7;
-		if (!IS_ALIGNED(height, 8)) {
-			dev_warn(p->dev, "%s:%d cb height (%d) invalid\n",
-				 __func__, __LINE__, height);
-			return -EINVAL;
-		}
 		break;
 	case V_0280A0_ARRAY_2D_TILED_THIN1:
-		pitch_align = max((u32)track->nbanks,
-				  (u32)(((track->group_size / 8) / (bpe * track->nsamples)) * track->nbanks)) / 8;
-		if (!IS_ALIGNED(pitch, pitch_align)) {
-			dev_warn(p->dev, "%s:%d cb pitch (%d) invalid\n",
-				__func__, __LINE__, pitch);
-			return -EINVAL;
-		}
-		if (!IS_ALIGNED((height / 8), track->npipes)) {
-			dev_warn(p->dev, "%s:%d cb height (%d) invalid\n",
-				 __func__, __LINE__, height);
-			return -EINVAL;
-		}
 		break;
 	default:
 		dev_warn(p->dev, "%s invalid tiling %d for %d (0x%08X)\n", __func__,
@@ -244,13 +293,29 @@ static inline int r600_cs_track_validate_cb(struct radeon_cs_parser *p, int i)
 			track->cb_color_info[i]);
 		return -EINVAL;
 	}
+
+	if (!IS_ALIGNED(pitch, pitch_align)) {
+		dev_warn(p->dev, "%s:%d cb pitch (%d) invalid\n",
+			 __func__, __LINE__, pitch);
+		return -EINVAL;
+	}
+	if (!IS_ALIGNED(height, height_align)) {
+		dev_warn(p->dev, "%s:%d cb height (%d) invalid\n",
+			 __func__, __LINE__, height);
+		return -EINVAL;
+	}
+	if (!IS_ALIGNED(base_offset, base_align)) {
+		dev_warn(p->dev, "%s offset[%d] 0x%llx not aligned\n", __func__, i, base_offset);
+		return -EINVAL;
+	}
+
 	/* check offset */
-	tmp = height * pitch * 8 * bpe;
+	tmp = height * pitch * bpe;
 	if ((tmp + track->cb_color_bo_offset[i]) > radeon_bo_size(track->cb_color_bo[i])) {
 		if (array_mode == V_0280A0_ARRAY_LINEAR_GENERAL) {
 			/* the initial DDX does bad things with the CB size occasionally */
 			/* it rounds up height too far for slice tile max but the BO is smaller */
-			tmp = (height - 7) * 8 * bpe;
+			tmp = (height - 7) * pitch * bpe;
 			if ((tmp + track->cb_color_bo_offset[i]) > radeon_bo_size(track->cb_color_bo[i])) {
 				dev_warn(p->dev, "%s offset[%d] %d %d %lu too big\n", __func__, i, track->cb_color_bo_offset[i], tmp, radeon_bo_size(track->cb_color_bo[i]));
 				return -EINVAL;
@@ -260,15 +325,11 @@ static inline int r600_cs_track_validate_cb(struct radeon_cs_parser *p, int i)
 			return -EINVAL;
 		}
 	}
-	if (!IS_ALIGNED(track->cb_color_bo_offset[i], track->group_size)) {
-		dev_warn(p->dev, "%s offset[%d] %d not aligned\n", __func__, i, track->cb_color_bo_offset[i]);
-		return -EINVAL;
-	}
 	/* limit max tile */
-	tmp = (height * pitch * 8) >> 6;
+	tmp = (height * pitch) >> 6;
 	if (tmp < slice_tile_max)
 		slice_tile_max = tmp;
-	tmp = S_028060_PITCH_TILE_MAX(pitch - 1) |
+	tmp = S_028060_PITCH_TILE_MAX((pitch / 8) - 1) |
 		S_028060_SLICE_TILE_MAX(slice_tile_max - 1);
 	ib[track->cb_color_size_idx[i]] = tmp;
 	return 0;
@@ -310,7 +371,12 @@ static int r600_cs_track_check(struct radeon_cs_parser *p)
 	/* Check depth buffer */
 	if (G_028800_STENCIL_ENABLE(track->db_depth_control) ||
 		G_028800_Z_ENABLE(track->db_depth_control)) {
-		u32 nviews, bpe, ntiles, pitch, pitch_align, height, size, slice_tile_max;
+		u32 nviews, bpe, ntiles, size, slice_tile_max;
+		u32 height, height_align, pitch, pitch_align, depth_align;
+		u64 base_offset, base_align;
+		struct array_mode_checker array_check;
+		int array_mode;
+
 		if (track->db_bo == NULL) {
 			dev_warn(p->dev, "z/stencil with no depth buffer\n");
 			return -EINVAL;
@@ -353,41 +419,34 @@ static int r600_cs_track_check(struct radeon_cs_parser *p)
 			ib[track->db_depth_size_idx] = S_028000_SLICE_TILE_MAX(tmp - 1) | (track->db_depth_size & 0x3FF);
 		} else {
 			size = radeon_bo_size(track->db_bo);
-			pitch = G_028000_PITCH_TILE_MAX(track->db_depth_size) + 1;
+			/* pitch in pixels */
+			pitch = (G_028000_PITCH_TILE_MAX(track->db_depth_size) + 1) * 8;
 			slice_tile_max = G_028000_SLICE_TILE_MAX(track->db_depth_size) + 1;
 			slice_tile_max *= 64;
-			height = slice_tile_max / (pitch * 8);
+			height = slice_tile_max / pitch;
 			if (height > 8192)
 				height = 8192;
-			switch (G_028010_ARRAY_MODE(track->db_depth_info)) {
+			base_offset = track->db_bo_mc + track->db_offset;
+			array_mode = G_028010_ARRAY_MODE(track->db_depth_info);
+			array_check.array_mode = array_mode;
+			array_check.group_size = track->group_size;
+			array_check.nbanks = track->nbanks;
+			array_check.npipes = track->npipes;
+			array_check.nsamples = track->nsamples;
+			array_check.bpe = bpe;
+			if (r600_get_array_mode_alignment(&array_check,
+							  &pitch_align, &height_align, &depth_align, &base_align)) {
+				dev_warn(p->dev, "%s invalid tiling %d (0x%08X)\n", __func__,
+					 G_028010_ARRAY_MODE(track->db_depth_info),
+					 track->db_depth_info);
+				return -EINVAL;
+			}
+			switch (array_mode) {
 			case V_028010_ARRAY_1D_TILED_THIN1:
-				pitch_align = (max((u32)8, (u32)(track->group_size / (8 * bpe))) / 8);
-				if (!IS_ALIGNED(pitch, pitch_align)) {
-					dev_warn(p->dev, "%s:%d db pitch (%d) invalid\n",
-						 __func__, __LINE__, pitch);
-					return -EINVAL;
-				}
 				/* don't break userspace */
 				height &= ~0x7;
-				if (!IS_ALIGNED(height, 8)) {
-					dev_warn(p->dev, "%s:%d db height (%d) invalid\n",
-						 __func__, __LINE__, height);
-					return -EINVAL;
-				}
 				break;
 			case V_028010_ARRAY_2D_TILED_THIN1:
-				pitch_align = max((u32)track->nbanks,
-						  (u32)(((track->group_size / 8) / bpe) * track->nbanks)) / 8;
-				if (!IS_ALIGNED(pitch, pitch_align)) {
-					dev_warn(p->dev, "%s:%d db pitch (%d) invalid\n",
-						 __func__, __LINE__, pitch);
-					return -EINVAL;
-				}
-				if (!IS_ALIGNED((height / 8), track->npipes)) {
-					dev_warn(p->dev, "%s:%d db height (%d) invalid\n",
-						 __func__, __LINE__, height);
-					return -EINVAL;
-				}
 				break;
 			default:
 				dev_warn(p->dev, "%s invalid tiling %d (0x%08X)\n", __func__,
@@ -395,15 +454,27 @@ static int r600_cs_track_check(struct radeon_cs_parser *p)
 					 track->db_depth_info);
 				return -EINVAL;
 			}
-			if (!IS_ALIGNED(track->db_offset, track->group_size)) {
-				dev_warn(p->dev, "%s offset[%d] %d not aligned\n", __func__, i, track->db_offset);
+
+			if (!IS_ALIGNED(pitch, pitch_align)) {
+				dev_warn(p->dev, "%s:%d db pitch (%d) invalid\n",
+					 __func__, __LINE__, pitch);
+				return -EINVAL;
+			}
+			if (!IS_ALIGNED(height, height_align)) {
+				dev_warn(p->dev, "%s:%d db height (%d) invalid\n",
+					 __func__, __LINE__, height);
 				return -EINVAL;
 			}
+			if (!IS_ALIGNED(base_offset, base_align)) {
+				dev_warn(p->dev, "%s offset[%d] 0x%llx not aligned\n", __func__, i, base_offset);
+				return -EINVAL;
+			}
+
 			ntiles = G_028000_SLICE_TILE_MAX(track->db_depth_size) + 1;
 			nviews = G_028004_SLICE_MAX(track->db_depth_view) + 1;
 			tmp = ntiles * bpe * 64 * nviews;
 			if ((tmp + track->db_offset) > radeon_bo_size(track->db_bo)) {
-				dev_warn(p->dev, "z/stencil buffer too small (0x%08X %d %d %d -> %d have %ld)\n",
+				dev_warn(p->dev, "z/stencil buffer too small (0x%08X %d %d %d -> %u have %lu)\n",
 						track->db_depth_size, ntiles, nviews, bpe, tmp + track->db_offset,
 						radeon_bo_size(track->db_bo));
 				return -EINVAL;
@@ -954,6 +1025,7 @@ static inline int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx
 		ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
 		track->cb_color_base_last[tmp] = ib[idx];
 		track->cb_color_bo[tmp] = reloc->robj;
+		track->cb_color_bo_mc[tmp] = reloc->lobj.gpu_offset;
 		break;
 	case DB_DEPTH_BASE:
 		r = r600_cs_packet_next_reloc(p, &reloc);
@@ -965,6 +1037,7 @@ static inline int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx
 		track->db_offset = radeon_get_ib_value(p, idx) << 8;
 		ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
 		track->db_bo = reloc->robj;
+		track->db_bo_mc = reloc->lobj.gpu_offset;
 		break;
 	case DB_HTILE_DATA_BASE:
 	case SQ_PGM_START_FS:
@@ -1086,16 +1159,25 @@ static void r600_texture_size(unsigned nfaces, unsigned blevel, unsigned nlevels
 static inline int r600_check_texture_resource(struct radeon_cs_parser *p,  u32 idx,
 					      struct radeon_bo *texture,
 					      struct radeon_bo *mipmap,
+					      u64 base_offset,
+					      u64 mip_offset,
 					      u32 tiling_flags)
 {
 	struct r600_cs_track *track = p->track;
 	u32 nfaces, nlevels, blevel, w0, h0, d0, bpe = 0;
-	u32 word0, word1, l0_size, mipmap_size, pitch, pitch_align;
+	u32 word0, word1, l0_size, mipmap_size;
+	u32 height_align, pitch, pitch_align, depth_align;
+	u64 base_align;
+	struct array_mode_checker array_check;
 
 	/* on legacy kernel we don't perform advanced check */
 	if (p->rdev == NULL)
 		return 0;
 
+	/* convert to bytes */
+	base_offset <<= 8;
+	mip_offset <<= 8;
+
 	word0 = radeon_get_ib_value(p, idx + 0);
 	if (tiling_flags & RADEON_TILING_MACRO)
 		word0 |= S_038000_TILE_MODE(V_038000_ARRAY_2D_TILED_THIN1);
@@ -1128,46 +1210,38 @@ static inline int r600_check_texture_resource(struct radeon_cs_parser *p,  u32 i
 		return -EINVAL;
 	}
 
-	pitch = G_038000_PITCH(word0) + 1;
-	switch (G_038000_TILE_MODE(word0)) {
-	case V_038000_ARRAY_LINEAR_GENERAL:
-		pitch_align = 1;
-		/* XXX check height align */
-		break;
-	case V_038000_ARRAY_LINEAR_ALIGNED:
-		pitch_align = max((u32)64, (u32)(track->group_size / bpe)) / 8;
-		if (!IS_ALIGNED(pitch, pitch_align)) {
-			dev_warn(p->dev, "%s:%d tex pitch (%d) invalid\n",
-				 __func__, __LINE__, pitch);
-			return -EINVAL;
-		}
-		/* XXX check height align */
-		break;
-	case V_038000_ARRAY_1D_TILED_THIN1:
-		pitch_align = max((u32)8, (u32)(track->group_size / (8 * bpe))) / 8;
-		if (!IS_ALIGNED(pitch, pitch_align)) {
-			dev_warn(p->dev, "%s:%d tex pitch (%d) invalid\n",
-				 __func__, __LINE__, pitch);
-			return -EINVAL;
-		}
-		/* XXX check height align */
-		break;
-	case V_038000_ARRAY_2D_TILED_THIN1:
-		pitch_align = max((u32)track->nbanks,
-				  (u32)(((track->group_size / 8) / bpe) * track->nbanks)) / 8;
-		if (!IS_ALIGNED(pitch, pitch_align)) {
-			dev_warn(p->dev, "%s:%d tex pitch (%d) invalid\n",
-				__func__, __LINE__, pitch);
-			return -EINVAL;
-		}
-		/* XXX check height align */
-		break;
-	default:
-		dev_warn(p->dev, "%s invalid tiling %d (0x%08X)\n", __func__,
-			 G_038000_TILE_MODE(word0), word0);
+	/* pitch in texels */
+	pitch = (G_038000_PITCH(word0) + 1) * 8;
+	array_check.array_mode = G_038000_TILE_MODE(word0);
+	array_check.group_size = track->group_size;
+	array_check.nbanks = track->nbanks;
+	array_check.npipes = track->npipes;
+	array_check.nsamples = 1;
+	array_check.bpe = bpe;
+	if (r600_get_array_mode_alignment(&array_check,
+					  &pitch_align, &height_align, &depth_align, &base_align)) {
+		dev_warn(p->dev, "%s:%d tex array mode (%d) invalid\n",
+			 __func__, __LINE__, G_038000_TILE_MODE(word0));
+		return -EINVAL;
+	}
+
+	/* XXX check height as well... */
+
+	if (!IS_ALIGNED(pitch, pitch_align)) {
+		dev_warn(p->dev, "%s:%d tex pitch (%d) invalid\n",
+			 __func__, __LINE__, pitch);
+		return -EINVAL;
+	}
+	if (!IS_ALIGNED(base_offset, base_align)) {
+		dev_warn(p->dev, "%s:%d tex base offset (0x%llx) invalid\n",
+			 __func__, __LINE__, base_offset);
+		return -EINVAL;
+	}
+	if (!IS_ALIGNED(mip_offset, base_align)) {
+		dev_warn(p->dev, "%s:%d tex mip offset (0x%llx) invalid\n",
+			 __func__, __LINE__, mip_offset);
 		return -EINVAL;
 	}
-	/* XXX check offset align */
 
 	word0 = radeon_get_ib_value(p, idx + 4);
 	word1 = radeon_get_ib_value(p, idx + 5);
@@ -1402,7 +1476,10 @@ static int r600_packet3_check(struct radeon_cs_parser *p,
 				mip_offset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
 				mipmap = reloc->robj;
 				r = r600_check_texture_resource(p,  idx+(i*7)+1,
-								texture, mipmap, reloc->lobj.tiling_flags);
+								texture, mipmap,
+								base_offset + radeon_get_ib_value(p, idx+1+(i*7)+2),
+								mip_offset + radeon_get_ib_value(p, idx+1+(i*7)+3),
+								reloc->lobj.tiling_flags);
 				if (r)
 					return r;
 				ib[idx+1+(i*7)+2] += base_offset;

drivers/gpu/drm/radeon/r600d.h

@@ -51,6 +51,12 @@
 #define PTE_READABLE				(1 << 5)
 #define PTE_WRITEABLE				(1 << 6)
 
+/* tiling bits */
+#define     ARRAY_LINEAR_GENERAL              0x00000000
+#define     ARRAY_LINEAR_ALIGNED              0x00000001
+#define     ARRAY_1D_TILED_THIN1              0x00000002
+#define     ARRAY_2D_TILED_THIN1              0x00000004
+
 /* Registers */
 #define	ARB_POP						0x2418
 #define 	ENABLE_TC128					(1 << 30)

drivers/gpu/drm/radeon/radeon.h

@@ -1262,6 +1262,10 @@ void r100_pll_errata_after_index(struct radeon_device *rdev);
 		(rdev->family == CHIP_RS400) ||			\
 		(rdev->family == CHIP_RS480))
 #define ASIC_IS_AVIVO(rdev) ((rdev->family >= CHIP_RS600))
+#define ASIC_IS_DCE2(rdev) ((rdev->family == CHIP_RS600)  ||	\
+			    (rdev->family == CHIP_RS690)  ||	\
+			    (rdev->family == CHIP_RS740)  ||	\
+			    (rdev->family >= CHIP_R600))
 #define ASIC_IS_DCE3(rdev) ((rdev->family >= CHIP_RV620))
 #define ASIC_IS_DCE32(rdev) ((rdev->family >= CHIP_RV730))
 #define ASIC_IS_DCE4(rdev) ((rdev->family >= CHIP_CEDAR))

drivers/gpu/drm/radeon/radeon_benchmark.c

@@ -41,7 +41,7 @@ void radeon_benchmark_move(struct radeon_device *rdev, unsigned bsize,
 
 	size = bsize;
 	n = 1024;
-	r = radeon_bo_create(rdev, NULL, size, true, sdomain, &sobj);
+	r = radeon_bo_create(rdev, NULL, size, PAGE_SIZE, true, sdomain, &sobj);
 	if (r) {
 		goto out_cleanup;
 	}
@@ -53,7 +53,7 @@ void radeon_benchmark_move(struct radeon_device *rdev, unsigned bsize,
 	if (r) {
 		goto out_cleanup;
 	}
-	r = radeon_bo_create(rdev, NULL, size, true, ddomain, &dobj);
+	r = radeon_bo_create(rdev, NULL, size, PAGE_SIZE, true, ddomain, &dobj);
 	if (r) {
 		goto out_cleanup;
 	}

drivers/gpu/drm/radeon/radeon_combios.c

@@ -571,6 +571,7 @@ static struct radeon_i2c_bus_rec combios_setup_i2c_bus(struct radeon_device *rde
 	}
 
 	if (clk_mask && data_mask) {
+		/* system specific masks */
 		i2c.mask_clk_mask = clk_mask;
 		i2c.mask_data_mask = data_mask;
 		i2c.a_clk_mask = clk_mask;
@@ -579,7 +580,19 @@ static struct radeon_i2c_bus_rec combios_setup_i2c_bus(struct radeon_device *rde
 		i2c.en_data_mask = data_mask;
 		i2c.y_clk_mask = clk_mask;
 		i2c.y_data_mask = data_mask;
+	} else if ((ddc_line == RADEON_GPIOPAD_MASK) ||
+		   (ddc_line == RADEON_MDGPIO_MASK)) {
+		/* default gpiopad masks */
+		i2c.mask_clk_mask = (0x20 << 8);
+		i2c.mask_data_mask = 0x80;
+		i2c.a_clk_mask = (0x20 << 8);
+		i2c.a_data_mask = 0x80;
+		i2c.en_clk_mask = (0x20 << 8);
+		i2c.en_data_mask = 0x80;
+		i2c.y_clk_mask = (0x20 << 8);
+		i2c.y_data_mask = 0x80;
 	} else {
+		/* default masks for ddc pads */
 		i2c.mask_clk_mask = RADEON_GPIO_EN_1;
 		i2c.mask_data_mask = RADEON_GPIO_EN_0;
 		i2c.a_clk_mask = RADEON_GPIO_A_1;

drivers/gpu/drm/radeon/radeon_connectors.c

@@ -1008,9 +1008,21 @@ static void radeon_dp_connector_destroy(struct drm_connector *connector)
 static int radeon_dp_get_modes(struct drm_connector *connector)
 {
 	struct radeon_connector *radeon_connector = to_radeon_connector(connector);
+	struct radeon_connector_atom_dig *radeon_dig_connector = radeon_connector->con_priv;
 	int ret;
 
+	if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+		if (!radeon_dig_connector->edp_on)
+			atombios_set_edp_panel_power(connector,
+						     ATOM_TRANSMITTER_ACTION_POWER_ON);
+	}
 	ret = radeon_ddc_get_modes(radeon_connector);
+	if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
+		if (!radeon_dig_connector->edp_on)
+			atombios_set_edp_panel_power(connector,
+						     ATOM_TRANSMITTER_ACTION_POWER_OFF);
+	}
+
 	return ret;
 }
 
@@ -1029,8 +1041,14 @@ radeon_dp_detect(struct drm_connector *connector, bool force)
 	if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) {
 		/* eDP is always DP */
 		radeon_dig_connector->dp_sink_type = CONNECTOR_OBJECT_ID_DISPLAYPORT;
+		if (!radeon_dig_connector->edp_on)
+			atombios_set_edp_panel_power(connector,
+						     ATOM_TRANSMITTER_ACTION_POWER_ON);
 		if (radeon_dp_getdpcd(radeon_connector))
 			ret = connector_status_connected;
+		if (!radeon_dig_connector->edp_on)
+			atombios_set_edp_panel_power(connector,
+						     ATOM_TRANSMITTER_ACTION_POWER_OFF);
 	} else {
 		radeon_dig_connector->dp_sink_type = radeon_dp_getsinktype(radeon_connector);
 		if (radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) {

drivers/gpu/drm/radeon/radeon_device.c

@@ -180,7 +180,7 @@ int radeon_wb_init(struct radeon_device *rdev)
 	int r;
 
 	if (rdev->wb.wb_obj == NULL) {
-		r = radeon_bo_create(rdev, NULL, RADEON_GPU_PAGE_SIZE, true,
+		r = radeon_bo_create(rdev, NULL, RADEON_GPU_PAGE_SIZE, PAGE_SIZE, true,
 				RADEON_GEM_DOMAIN_GTT, &rdev->wb.wb_obj);
 		if (r) {
 			dev_warn(rdev->dev, "(%d) create WB bo failed\n", r);

drivers/gpu/drm/radeon/radeon_encoders.c

@@ -176,6 +176,7 @@ static inline bool radeon_encoder_is_digital(struct drm_encoder *encoder)
 		return false;
 	}
 }
+
 void
 radeon_link_encoder_connector(struct drm_device *dev)
 {
@@ -228,6 +229,27 @@ radeon_get_connector_for_encoder(struct drm_encoder *encoder)
 	return NULL;
 }
 
+struct drm_encoder *radeon_atom_get_external_encoder(struct drm_encoder *encoder)
+{
+	struct drm_device *dev = encoder->dev;
+	struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
+	struct drm_encoder *other_encoder;
+	struct radeon_encoder *other_radeon_encoder;
+
+	if (radeon_encoder->is_ext_encoder)
+		return NULL;
+
+	list_for_each_entry(other_encoder, &dev->mode_config.encoder_list, head) {
+		if (other_encoder == encoder)
+			continue;
+		other_radeon_encoder = to_radeon_encoder(other_encoder);
+		if (other_radeon_encoder->is_ext_encoder &&
+		    (radeon_encoder->devices & other_radeon_encoder->devices))
+			return other_encoder;
+	}
+	return NULL;
+}
+
 void radeon_panel_mode_fixup(struct drm_encoder *encoder,
 			     struct drm_display_mode *adjusted_mode)
 {
@@ -426,52 +448,49 @@ atombios_tv_setup(struct drm_encoder *encoder, int action)
 
 }
 
-void
-atombios_external_tmds_setup(struct drm_encoder *encoder, int action)
-{
-	struct drm_device *dev = encoder->dev;
-	struct radeon_device *rdev = dev->dev_private;
-	struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
-	ENABLE_EXTERNAL_TMDS_ENCODER_PS_ALLOCATION args;
-	int index = 0;
-
-	memset(&args, 0, sizeof(args));
-
-	index = GetIndexIntoMasterTable(COMMAND, DVOEncoderControl);
-
-	args.sXTmdsEncoder.ucEnable = action;
-
-	if (radeon_encoder->pixel_clock > 165000)
-		args.sXTmdsEncoder.ucMisc = PANEL_ENCODER_MISC_DUAL;
-
-	/*if (pScrn->rgbBits == 8)*/
-	args.sXTmdsEncoder.ucMisc |= (1 << 1);
-
-	atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
-
-}
+union dvo_encoder_control {
+	ENABLE_EXTERNAL_TMDS_ENCODER_PS_ALLOCATION ext_tmds;
+	DVO_ENCODER_CONTROL_PS_ALLOCATION dvo;
+	DVO_ENCODER_CONTROL_PS_ALLOCATION_V3 dvo_v3;
+};
 
-static void
-atombios_ddia_setup(struct drm_encoder *encoder, int action)
+void
+atombios_dvo_setup(struct drm_encoder *encoder, int action)
 {
 	struct drm_device *dev = encoder->dev;
 	struct radeon_device *rdev = dev->dev_private;
 	struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
-	DVO_ENCODER_CONTROL_PS_ALLOCATION args;
-	int index = 0;
+	union dvo_encoder_control args;
+	int index = GetIndexIntoMasterTable(COMMAND, DVOEncoderControl);
 
 	memset(&args, 0, sizeof(args));
 
-	index = GetIndexIntoMasterTable(COMMAND, DVOEncoderControl);
+	if (ASIC_IS_DCE3(rdev)) {
+		/* DCE3+ */
+		args.dvo_v3.ucAction = action;
+		args.dvo_v3.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10);
+		args.dvo_v3.ucDVOConfig = 0; /* XXX */
+	} else if (ASIC_IS_DCE2(rdev)) {
+		/* DCE2 (pre-DCE3 R6xx, RS600/690/740 */
+		args.dvo.sDVOEncoder.ucAction = action;
+		args.dvo.sDVOEncoder.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10);
+		/* DFP1, CRT1, TV1 depending on the type of port */
+		args.dvo.sDVOEncoder.ucDeviceType = ATOM_DEVICE_DFP1_INDEX;
+
+		if (radeon_encoder->pixel_clock > 165000)
+			args.dvo.sDVOEncoder.usDevAttr.sDigAttrib.ucAttribute |= PANEL_ENCODER_MISC_DUAL;
+	} else {
+		/* R4xx, R5xx */
+		args.ext_tmds.sXTmdsEncoder.ucEnable = action;
 
-	args.sDVOEncoder.ucAction = action;
-	args.sDVOEncoder.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10);
+		if (radeon_encoder->pixel_clock > 165000)
+			args.ext_tmds.sXTmdsEncoder.ucMisc |= PANEL_ENCODER_MISC_DUAL;
 
-	if (radeon_encoder->pixel_clock > 165000)
-		args.sDVOEncoder.usDevAttr.sDigAttrib.ucAttribute = PANEL_ENCODER_MISC_DUAL;
+		/*if (pScrn->rgbBits == 8)*/
+		args.ext_tmds.sXTmdsEncoder.ucMisc |= ATOM_PANEL_MISC_888RGB;
+	}
 
 	atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
-
 }
 
 union lvds_encoder_control {
@@ -532,14 +551,14 @@ atombios_digital_setup(struct drm_encoder *encoder, int action)
 				if (dig->lcd_misc & ATOM_PANEL_MISC_DUAL)
 					args.v1.ucMisc |= PANEL_ENCODER_MISC_DUAL;
 				if (dig->lcd_misc & ATOM_PANEL_MISC_888RGB)
-					args.v1.ucMisc |= (1 << 1);
+					args.v1.ucMisc |= ATOM_PANEL_MISC_888RGB;
 			} else {
 				if (dig->linkb)
 					args.v1.ucMisc |= PANEL_ENCODER_MISC_TMDS_LINKB;
 				if (radeon_encoder->pixel_clock > 165000)
 					args.v1.ucMisc |= PANEL_ENCODER_MISC_DUAL;
 				/*if (pScrn->rgbBits == 8) */
-				args.v1.ucMisc |= (1 << 1);
+				args.v1.ucMisc |= ATOM_PANEL_MISC_888RGB;
 			}
 			break;
 		case 2:
@@ -595,6 +614,7 @@ atombios_digital_setup(struct drm_encoder *encoder, int action)
 int
 atombios_get_encoder_mode(struct drm_encoder *encoder)
 {
+	struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
 	struct drm_device *dev = encoder->dev;
 	struct radeon_device *rdev = dev->dev_private;
 	struct drm_connector *connector;
@@ -602,9 +622,20 @@ atombios_get_encoder_mode(struct drm_encoder *encoder)
 	struct radeon_connector_atom_dig *dig_connector;
 
 	connector = radeon_get_connector_for_encoder(encoder);
-	if (!connector)
-		return 0;
-
+	if (!connector) {
+		switch (radeon_encoder->encoder_id) {
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+		case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+		case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+		case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+			return ATOM_ENCODER_MODE_DVI;
+		case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
+		case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2:
+		default:
+			return ATOM_ENCODER_MODE_CRT;
+		}
+	}
 	radeon_connector = to_radeon_connector(connector);
 
 	switch (connector->connector_type) {
@@ -834,6 +865,9 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t
 	memset(&args, 0, sizeof(args));
 
 	switch (radeon_encoder->encoder_id) {
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+		index = GetIndexIntoMasterTable(COMMAND, DVOOutputControl);
+		break;
 	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
 	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
 	case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
@@ -978,6 +1012,105 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t
 	atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 }
 
+void
+atombios_set_edp_panel_power(struct drm_connector *connector, int action)
+{
+	struct radeon_connector *radeon_connector = to_radeon_connector(connector);
+	struct drm_device *dev = radeon_connector->base.dev;
+	struct radeon_device *rdev = dev->dev_private;
+	union dig_transmitter_control args;
+	int index = GetIndexIntoMasterTable(COMMAND, UNIPHYTransmitterControl);
+	uint8_t frev, crev;
+
+	if (connector->connector_type != DRM_MODE_CONNECTOR_eDP)
+		return;
+
+	if (!ASIC_IS_DCE4(rdev))
+		return;
+
+	if ((action != ATOM_TRANSMITTER_ACTION_POWER_ON) ||
+	    (action != ATOM_TRANSMITTER_ACTION_POWER_OFF))
+		return;
+
+	if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	memset(&args, 0, sizeof(args));
+
+	args.v1.ucAction = action;
+
+	atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
+union external_encoder_control {
+	EXTERNAL_ENCODER_CONTROL_PS_ALLOCATION v1;
+};
+
+static void
+atombios_external_encoder_setup(struct drm_encoder *encoder,
+				struct drm_encoder *ext_encoder,
+				int action)
+{
+	struct drm_device *dev = encoder->dev;
+	struct radeon_device *rdev = dev->dev_private;
+	struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
+	union external_encoder_control args;
+	struct drm_connector *connector = radeon_get_connector_for_encoder(encoder);
+	int index = GetIndexIntoMasterTable(COMMAND, ExternalEncoderControl);
+	u8 frev, crev;
+	int dp_clock = 0;
+	int dp_lane_count = 0;
+	int connector_object_id = 0;
+
+	if (connector) {
+		struct radeon_connector *radeon_connector = to_radeon_connector(connector);
+		struct radeon_connector_atom_dig *dig_connector =
+			radeon_connector->con_priv;
+
+		dp_clock = dig_connector->dp_clock;
+		dp_lane_count = dig_connector->dp_lane_count;
+		connector_object_id =
+			(radeon_connector->connector_object_id & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT;
+	}
+
+	memset(&args, 0, sizeof(args));
+
+	if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev))
+		return;
+
+	switch (frev) {
+	case 1:
+		/* no params on frev 1 */
+		break;
+	case 2:
+		switch (crev) {
+		case 1:
+		case 2:
+			args.v1.sDigEncoder.ucAction = action;
+			args.v1.sDigEncoder.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10);
+			args.v1.sDigEncoder.ucEncoderMode = atombios_get_encoder_mode(encoder);
+
+			if (args.v1.sDigEncoder.ucEncoderMode == ATOM_ENCODER_MODE_DP) {
+				if (dp_clock == 270000)
+					args.v1.sDigEncoder.ucConfig |= ATOM_ENCODER_CONFIG_DPLINKRATE_2_70GHZ;
+				args.v1.sDigEncoder.ucLaneNum = dp_lane_count;
+			} else if (radeon_encoder->pixel_clock > 165000)
+				args.v1.sDigEncoder.ucLaneNum = 8;
+			else
+				args.v1.sDigEncoder.ucLaneNum = 4;
+			break;
+		default:
+			DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
+			return;
+		}
+		break;
+	default:
+		DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
+		return;
+	}
+	atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+}
+
 static void
 atombios_yuv_setup(struct drm_encoder *encoder, bool enable)
 {
@@ -1021,6 +1154,7 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode)
 	struct drm_device *dev = encoder->dev;
 	struct radeon_device *rdev = dev->dev_private;
 	struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
+	struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder);
 	DISPLAY_DEVICE_OUTPUT_CONTROL_PS_ALLOCATION args;
 	int index = 0;
 	bool is_dig = false;
@@ -1043,9 +1177,14 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode)
 		break;
 	case ENCODER_OBJECT_ID_INTERNAL_DVO1:
 	case ENCODER_OBJECT_ID_INTERNAL_DDI:
-	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
 		index = GetIndexIntoMasterTable(COMMAND, DVOOutputControl);
 		break;
+	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+		if (ASIC_IS_DCE3(rdev))
+			is_dig = true;
+		else
+			index = GetIndexIntoMasterTable(COMMAND, DVOOutputControl);
+		break;
 	case ENCODER_OBJECT_ID_INTERNAL_LVDS:
 		index = GetIndexIntoMasterTable(COMMAND, LCD1OutputControl);
 		break;
@@ -1082,34 +1221,85 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode)
 			if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_DP) {
 				struct drm_connector *connector = radeon_get_connector_for_encoder(encoder);
 
+				if (connector &&
+				    (connector->connector_type == DRM_MODE_CONNECTOR_eDP)) {
+					struct radeon_connector *radeon_connector = to_radeon_connector(connector);
+					struct radeon_connector_atom_dig *radeon_dig_connector =
+						radeon_connector->con_priv;
+					atombios_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_ON);
+					radeon_dig_connector->edp_on = true;
+				}
 				dp_link_train(encoder, connector);
 				if (ASIC_IS_DCE4(rdev))
 					atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_ON);
 			}
+			if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+				atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_LCD_BLON, 0, 0);
 			break;
 		case DRM_MODE_DPMS_STANDBY:
 		case DRM_MODE_DPMS_SUSPEND:
 		case DRM_MODE_DPMS_OFF:
 			atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE_OUTPUT, 0, 0);
 			if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_DP) {
+				struct drm_connector *connector = radeon_get_connector_for_encoder(encoder);
+
 				if (ASIC_IS_DCE4(rdev))
 					atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_OFF);
+				if (connector &&
+				    (connector->connector_type == DRM_MODE_CONNECTOR_eDP)) {
+					struct radeon_connector *radeon_connector = to_radeon_connector(connector);
+					struct radeon_connector_atom_dig *radeon_dig_connector =
+						radeon_connector->con_priv;
+					atombios_set_edp_panel_power(connector,
+								     ATOM_TRANSMITTER_ACTION_POWER_OFF);
+					radeon_dig_connector->edp_on = false;
+				}
 			}
+			if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+				atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_LCD_BLOFF, 0, 0);
 			break;
 		}
 	} else {
 		switch (mode) {
 		case DRM_MODE_DPMS_ON:
 			args.ucAction = ATOM_ENABLE;
+			atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+			if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+				args.ucAction = ATOM_LCD_BLON;
+				atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+			}
 			break;
 		case DRM_MODE_DPMS_STANDBY:
 		case DRM_MODE_DPMS_SUSPEND:
 		case DRM_MODE_DPMS_OFF:
 			args.ucAction = ATOM_DISABLE;
+			atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+			if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) {
+				args.ucAction = ATOM_LCD_BLOFF;
+				atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
+			}
 			break;
 		}
-		atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 	}
+
+	if (ext_encoder) {
+		int action;
+
+		switch (mode) {
+		case DRM_MODE_DPMS_ON:
+		default:
+			action = ATOM_ENABLE;
+			break;
+		case DRM_MODE_DPMS_STANDBY:
+		case DRM_MODE_DPMS_SUSPEND:
+		case DRM_MODE_DPMS_OFF:
+			action = ATOM_DISABLE;
+			break;
+		}
+		atombios_external_encoder_setup(encoder, ext_encoder, action);
+	}
+
 	radeon_atombios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false);
 
 }
@@ -1242,7 +1432,7 @@ atombios_set_encoder_crtc_source(struct drm_encoder *encoder)
 		break;
 	default:
 		DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
-		break;
+		return;
 	}
 
 	atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
@@ -1357,6 +1547,7 @@ radeon_atom_encoder_mode_set(struct drm_encoder *encoder,
 	struct drm_device *dev = encoder->dev;
 	struct radeon_device *rdev = dev->dev_private;
 	struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
+	struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder);
 
 	radeon_encoder->pixel_clock = adjusted_mode->clock;
 
@@ -1400,11 +1591,9 @@ radeon_atom_encoder_mode_set(struct drm_encoder *encoder,
 		}
 		break;
 	case ENCODER_OBJECT_ID_INTERNAL_DDI:
-		atombios_ddia_setup(encoder, ATOM_ENABLE);
-		break;
 	case ENCODER_OBJECT_ID_INTERNAL_DVO1:
 	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
-		atombios_external_tmds_setup(encoder, ATOM_ENABLE);
+		atombios_dvo_setup(encoder, ATOM_ENABLE);
 		break;
 	case ENCODER_OBJECT_ID_INTERNAL_DAC1:
 	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
@@ -1419,6 +1608,11 @@ radeon_atom_encoder_mode_set(struct drm_encoder *encoder,
 		}
 		break;
 	}
+
+	if (ext_encoder) {
+		atombios_external_encoder_setup(encoder, ext_encoder, ATOM_ENABLE);
+	}
+
 	atombios_apply_encoder_quirks(encoder, adjusted_mode);
 
 	if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) {
@@ -1595,11 +1789,9 @@ static void radeon_atom_encoder_disable(struct drm_encoder *encoder)
 		}
 		break;
 	case ENCODER_OBJECT_ID_INTERNAL_DDI:
-		atombios_ddia_setup(encoder, ATOM_DISABLE);
-		break;
 	case ENCODER_OBJECT_ID_INTERNAL_DVO1:
 	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
-		atombios_external_tmds_setup(encoder, ATOM_DISABLE);
+		atombios_dvo_setup(encoder, ATOM_DISABLE);
 		break;
 	case ENCODER_OBJECT_ID_INTERNAL_DAC1:
 	case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
@@ -1621,6 +1813,53 @@ disable_done:
 	radeon_encoder->active_device = 0;
 }
 
+/* these are handled by the primary encoders */
+static void radeon_atom_ext_prepare(struct drm_encoder *encoder)
+{
+
+}
+
+static void radeon_atom_ext_commit(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+radeon_atom_ext_mode_set(struct drm_encoder *encoder,
+			 struct drm_display_mode *mode,
+			 struct drm_display_mode *adjusted_mode)
+{
+
+}
+
+static void radeon_atom_ext_disable(struct drm_encoder *encoder)
+{
+
+}
+
+static void
+radeon_atom_ext_dpms(struct drm_encoder *encoder, int mode)
+{
+
+}
+
+static bool radeon_atom_ext_mode_fixup(struct drm_encoder *encoder,
+				       struct drm_display_mode *mode,
+				       struct drm_display_mode *adjusted_mode)
+{
+	return true;
+}
+
+static const struct drm_encoder_helper_funcs radeon_atom_ext_helper_funcs = {
+	.dpms = radeon_atom_ext_dpms,
+	.mode_fixup = radeon_atom_ext_mode_fixup,
+	.prepare = radeon_atom_ext_prepare,
+	.mode_set = radeon_atom_ext_mode_set,
+	.commit = radeon_atom_ext_commit,
+	.disable = radeon_atom_ext_disable,
+	/* no detect for TMDS/LVDS yet */
+};
+
 static const struct drm_encoder_helper_funcs radeon_atom_dig_helper_funcs = {
 	.dpms = radeon_atom_encoder_dpms,
 	.mode_fixup = radeon_atom_mode_fixup,
@@ -1730,6 +1969,7 @@ radeon_add_atom_encoder(struct drm_device *dev, uint32_t encoder_enum, uint32_t
 	radeon_encoder->devices = supported_device;
 	radeon_encoder->rmx_type = RMX_OFF;
 	radeon_encoder->underscan_type = UNDERSCAN_OFF;
+	radeon_encoder->is_ext_encoder = false;
 
 	switch (radeon_encoder->encoder_id) {
 	case ENCODER_OBJECT_ID_INTERNAL_LVDS:
@@ -1771,6 +2011,9 @@ radeon_add_atom_encoder(struct drm_device *dev, uint32_t encoder_enum, uint32_t
 			radeon_encoder->rmx_type = RMX_FULL;
 			drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_LVDS);
 			radeon_encoder->enc_priv = radeon_atombios_get_lvds_info(radeon_encoder);
+		} else if (radeon_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT)) {
+			drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_DAC);
+			radeon_encoder->enc_priv = radeon_atombios_set_dig_info(radeon_encoder);
 		} else {
 			drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_TMDS);
 			radeon_encoder->enc_priv = radeon_atombios_set_dig_info(radeon_encoder);
@@ -1779,5 +2022,22 @@ radeon_add_atom_encoder(struct drm_device *dev, uint32_t encoder_enum, uint32_t
 		}
 		drm_encoder_helper_add(encoder, &radeon_atom_dig_helper_funcs);
 		break;
+	case ENCODER_OBJECT_ID_SI170B:
+	case ENCODER_OBJECT_ID_CH7303:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOA:
+	case ENCODER_OBJECT_ID_EXTERNAL_SDVOB:
+	case ENCODER_OBJECT_ID_TITFP513:
+	case ENCODER_OBJECT_ID_VT1623:
+	case ENCODER_OBJECT_ID_HDMI_SI1930:
+		/* these are handled by the primary encoders */
+		radeon_encoder->is_ext_encoder = true;
+		if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT))
+			drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_LVDS);
+		else if (radeon_encoder->devices & (ATOM_DEVICE_CRT_SUPPORT))
+			drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_DAC);
+		else
+			drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_TMDS);
+		drm_encoder_helper_add(encoder, &radeon_atom_ext_helper_funcs);
+		break;
 	}
 }

drivers/gpu/drm/radeon/radeon_gart.c

@@ -79,8 +79,8 @@ int radeon_gart_table_vram_alloc(struct radeon_device *rdev)
 
 	if (rdev->gart.table.vram.robj == NULL) {
 		r = radeon_bo_create(rdev, NULL, rdev->gart.table_size,
-					true, RADEON_GEM_DOMAIN_VRAM,
-					&rdev->gart.table.vram.robj);
+				     PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM,
+				     &rdev->gart.table.vram.robj);
 		if (r) {
 			return r;
 		}

drivers/gpu/drm/radeon/radeon_gem.c

@@ -67,7 +67,7 @@ int radeon_gem_object_create(struct radeon_device *rdev, int size,
 	if (alignment < PAGE_SIZE) {
 		alignment = PAGE_SIZE;
 	}
-	r = radeon_bo_create(rdev, gobj, size, kernel, initial_domain, &robj);
+	r = radeon_bo_create(rdev, gobj, size, alignment, kernel, initial_domain, &robj);
 	if (r) {
 		if (r != -ERESTARTSYS)
 			DRM_ERROR("Failed to allocate GEM object (%d, %d, %u, %d)\n",

drivers/gpu/drm/radeon/radeon_i2c.c

@@ -896,7 +896,8 @@ struct radeon_i2c_chan *radeon_i2c_create(struct drm_device *dev,
 	     ((rdev->family <= CHIP_RS480) ||
 	      ((rdev->family >= CHIP_RV515) && (rdev->family <= CHIP_R580))))) {
 		/* set the radeon hw i2c adapter */
-		sprintf(i2c->adapter.name, "Radeon i2c hw bus %s", name);
+		snprintf(i2c->adapter.name, sizeof(i2c->adapter.name),
+			 "Radeon i2c hw bus %s", name);
 		i2c->adapter.algo = &radeon_i2c_algo;
 		ret = i2c_add_adapter(&i2c->adapter);
 		if (ret) {
@@ -905,7 +906,8 @@ struct radeon_i2c_chan *radeon_i2c_create(struct drm_device *dev,
 		}
 	} else {
 		/* set the radeon bit adapter */
-		sprintf(i2c->adapter.name, "Radeon i2c bit bus %s", name);
+		snprintf(i2c->adapter.name, sizeof(i2c->adapter.name),
+			 "Radeon i2c bit bus %s", name);
 		i2c->adapter.algo_data = &i2c->algo.bit;
 		i2c->algo.bit.pre_xfer = pre_xfer;
 		i2c->algo.bit.post_xfer = post_xfer;
@@ -946,6 +948,8 @@ struct radeon_i2c_chan *radeon_i2c_create_dp(struct drm_device *dev,
 	i2c->rec = *rec;
 	i2c->adapter.owner = THIS_MODULE;
 	i2c->dev = dev;
+	snprintf(i2c->adapter.name, sizeof(i2c->adapter.name),
+		 "Radeon aux bus %s", name);
 	i2c_set_adapdata(&i2c->adapter, i2c);
 	i2c->adapter.algo_data = &i2c->algo.dp;
 	i2c->algo.dp.aux_ch = radeon_dp_i2c_aux_ch;

drivers/gpu/drm/radeon/radeon_irq.c

@@ -76,7 +76,7 @@ int radeon_enable_vblank(struct drm_device *dev, int crtc)
 		default:
 			DRM_ERROR("tried to enable vblank on non-existent crtc %d\n",
 				  crtc);
-			return EINVAL;
+			return -EINVAL;
 		}
 	} else {
 		switch (crtc) {
@@ -89,7 +89,7 @@ int radeon_enable_vblank(struct drm_device *dev, int crtc)
 		default:
 			DRM_ERROR("tried to enable vblank on non-existent crtc %d\n",
 				  crtc);
-			return EINVAL;
+			return -EINVAL;
 		}
 	}
 

drivers/gpu/drm/radeon/radeon_legacy_encoders.c

@@ -670,7 +670,7 @@ static void radeon_legacy_tmds_ext_mode_set(struct drm_encoder *encoder,
 
 	if (rdev->is_atom_bios) {
 		radeon_encoder->pixel_clock = adjusted_mode->clock;
-		atombios_external_tmds_setup(encoder, ATOM_ENABLE);
+		atombios_dvo_setup(encoder, ATOM_ENABLE);
 		fp2_gen_cntl = RREG32(RADEON_FP2_GEN_CNTL);
 	} else {
 		fp2_gen_cntl = RREG32(RADEON_FP2_GEN_CNTL);

drivers/gpu/drm/radeon/radeon_mode.h

@@ -375,6 +375,7 @@ struct radeon_encoder {
 	int hdmi_config_offset;
 	int hdmi_audio_workaround;
 	int hdmi_buffer_status;
+	bool is_ext_encoder;
 };
 
 struct radeon_connector_atom_dig {
@@ -385,6 +386,7 @@ struct radeon_connector_atom_dig {
 	u8 dp_sink_type;
 	int dp_clock;
 	int dp_lane_count;
+	bool edp_on;
 };
 
 struct radeon_gpio_rec {
@@ -523,9 +525,10 @@ struct drm_encoder *radeon_encoder_legacy_primary_dac_add(struct drm_device *dev
 struct drm_encoder *radeon_encoder_legacy_tv_dac_add(struct drm_device *dev, int bios_index, int with_tv);
 struct drm_encoder *radeon_encoder_legacy_tmds_int_add(struct drm_device *dev, int bios_index);
 struct drm_encoder *radeon_encoder_legacy_tmds_ext_add(struct drm_device *dev, int bios_index);
-extern void atombios_external_tmds_setup(struct drm_encoder *encoder, int action);
+extern void atombios_dvo_setup(struct drm_encoder *encoder, int action);
 extern void atombios_digital_setup(struct drm_encoder *encoder, int action);
 extern int atombios_get_encoder_mode(struct drm_encoder *encoder);
+extern void atombios_set_edp_panel_power(struct drm_connector *connector, int action);
 extern void radeon_encoder_set_active_device(struct drm_encoder *encoder);
 
 extern void radeon_crtc_load_lut(struct drm_crtc *crtc);

drivers/gpu/drm/radeon/radeon_object.c

@@ -86,11 +86,12 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain)
 }
 
 int radeon_bo_create(struct radeon_device *rdev, struct drm_gem_object *gobj,
-			unsigned long size, bool kernel, u32 domain,
-			struct radeon_bo **bo_ptr)
+		     unsigned long size, int byte_align, bool kernel, u32 domain,
+		     struct radeon_bo **bo_ptr)
 {
 	struct radeon_bo *bo;
 	enum ttm_bo_type type;
+	int page_align = roundup(byte_align, PAGE_SIZE) >> PAGE_SHIFT;
 	int r;
 
 	if (unlikely(rdev->mman.bdev.dev_mapping == NULL)) {
@@ -115,7 +116,7 @@ retry:
 	/* Kernel allocation are uninterruptible */
 	mutex_lock(&rdev->vram_mutex);
 	r = ttm_bo_init(&rdev->mman.bdev, &bo->tbo, size, type,
-			&bo->placement, 0, 0, !kernel, NULL, size,
+			&bo->placement, page_align, 0, !kernel, NULL, size,
 			&radeon_ttm_bo_destroy);
 	mutex_unlock(&rdev->vram_mutex);
 	if (unlikely(r != 0)) {

drivers/gpu/drm/radeon/radeon_object.h

@@ -137,9 +137,10 @@ static inline int radeon_bo_wait(struct radeon_bo *bo, u32 *mem_type,
 }
 
 extern int radeon_bo_create(struct radeon_device *rdev,
-				struct drm_gem_object *gobj, unsigned long size,
-				bool kernel, u32 domain,
-				struct radeon_bo **bo_ptr);
+			    struct drm_gem_object *gobj, unsigned long size,
+			    int byte_align,
+			    bool kernel, u32 domain,
+			    struct radeon_bo **bo_ptr);
 extern int radeon_bo_kmap(struct radeon_bo *bo, void **ptr);
 extern void radeon_bo_kunmap(struct radeon_bo *bo);
 extern void radeon_bo_unref(struct radeon_bo **bo);

drivers/gpu/drm/radeon/radeon_ring.c

@@ -176,8 +176,8 @@ int radeon_ib_pool_init(struct radeon_device *rdev)
 	INIT_LIST_HEAD(&rdev->ib_pool.bogus_ib);
 	/* Allocate 1M object buffer */
 	r = radeon_bo_create(rdev, NULL,  RADEON_IB_POOL_SIZE*64*1024,
-				true, RADEON_GEM_DOMAIN_GTT,
-				&rdev->ib_pool.robj);
+			     PAGE_SIZE, true, RADEON_GEM_DOMAIN_GTT,
+			     &rdev->ib_pool.robj);
 	if (r) {
 		DRM_ERROR("radeon: failed to ib pool (%d).\n", r);
 		return r;
@@ -332,7 +332,7 @@ int radeon_ring_init(struct radeon_device *rdev, unsigned ring_size)
 	rdev->cp.ring_size = ring_size;
 	/* Allocate ring buffer */
 	if (rdev->cp.ring_obj == NULL) {
-		r = radeon_bo_create(rdev, NULL, rdev->cp.ring_size, true,
+		r = radeon_bo_create(rdev, NULL, rdev->cp.ring_size, PAGE_SIZE, true,
 					RADEON_GEM_DOMAIN_GTT,
 					&rdev->cp.ring_obj);
 		if (r) {

drivers/gpu/drm/radeon/radeon_test.c

@@ -52,7 +52,7 @@ void radeon_test_moves(struct radeon_device *rdev)
 		goto out_cleanup;
 	}
 
-	r = radeon_bo_create(rdev, NULL, size, true, RADEON_GEM_DOMAIN_VRAM,
+	r = radeon_bo_create(rdev, NULL, size, PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM,
 				&vram_obj);
 	if (r) {
 		DRM_ERROR("Failed to create VRAM object\n");
@@ -71,7 +71,7 @@ void radeon_test_moves(struct radeon_device *rdev)
 		void **gtt_start, **gtt_end;
 		void **vram_start, **vram_end;
 
-		r = radeon_bo_create(rdev, NULL, size, true,
+		r = radeon_bo_create(rdev, NULL, size, PAGE_SIZE, true,
 					 RADEON_GEM_DOMAIN_GTT, gtt_obj + i);
 		if (r) {
 			DRM_ERROR("Failed to create GTT object %d\n", i);

drivers/gpu/drm/radeon/radeon_ttm.c

@@ -529,7 +529,7 @@ int radeon_ttm_init(struct radeon_device *rdev)
 		DRM_ERROR("Failed initializing VRAM heap.\n");
 		return r;
 	}
-	r = radeon_bo_create(rdev, NULL, 256 * 1024, true,
+	r = radeon_bo_create(rdev, NULL, 256 * 1024, PAGE_SIZE, true,
 				RADEON_GEM_DOMAIN_VRAM,
 				&rdev->stollen_vga_memory);
 	if (r) {

drivers/gpu/drm/radeon/rv770.c

@@ -915,8 +915,8 @@ static int rv770_vram_scratch_init(struct radeon_device *rdev)
 
 	if (rdev->vram_scratch.robj == NULL) {
 		r = radeon_bo_create(rdev, NULL, RADEON_GPU_PAGE_SIZE,
-					true, RADEON_GEM_DOMAIN_VRAM,
-					&rdev->vram_scratch.robj);
+				     PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM,
+				     &rdev->vram_scratch.robj);
 		if (r) {
 			return r;
 		}

drivers/gpu/drm/ttm/ttm_bo.c

@@ -224,6 +224,9 @@ int ttm_bo_reserve_locked(struct ttm_buffer_object *bo,
 	int ret;
 
 	while (unlikely(atomic_cmpxchg(&bo->reserved, 0, 1) != 0)) {
+		/**
+		 * Deadlock avoidance for multi-bo reserving.
+		 */
 		if (use_sequence && bo->seq_valid &&
 			(sequence - bo->val_seq < (1 << 31))) {
 			return -EAGAIN;
@@ -241,6 +244,14 @@ int ttm_bo_reserve_locked(struct ttm_buffer_object *bo,
 	}
 
 	if (use_sequence) {
+		/**
+		 * Wake up waiters that may need to recheck for deadlock,
+		 * if we decreased the sequence number.
+		 */
+		if (unlikely((bo->val_seq - sequence < (1 << 31))
+			     || !bo->seq_valid))
+			wake_up_all(&bo->event_queue);
+
 		bo->val_seq = sequence;
 		bo->seq_valid = true;
 	} else {

drivers/gpu/drm/vmwgfx/vmwgfx_resource.c

@@ -862,7 +862,7 @@ int vmw_dmabuf_alloc_ioctl(struct drm_device *dev, void *data,
 			      &vmw_vram_sys_placement, true,
 			      &vmw_user_dmabuf_destroy);
 	if (unlikely(ret != 0))
-		return ret;
+		goto out_no_dmabuf;
 
 	tmp = ttm_bo_reference(&vmw_user_bo->dma.base);
 	ret = ttm_base_object_init(vmw_fpriv(file_priv)->tfile,
@@ -870,19 +870,21 @@ int vmw_dmabuf_alloc_ioctl(struct drm_device *dev, void *data,
 				   false,
 				   ttm_buffer_type,
 				   &vmw_user_dmabuf_release, NULL);
-	if (unlikely(ret != 0)) {
-		ttm_bo_unref(&tmp);
-	} else {
+	if (unlikely(ret != 0))
+		goto out_no_base_object;
+	else {
 		rep->handle = vmw_user_bo->base.hash.key;
 		rep->map_handle = vmw_user_bo->dma.base.addr_space_offset;
 		rep->cur_gmr_id = vmw_user_bo->base.hash.key;
 		rep->cur_gmr_offset = 0;
 	}
-	ttm_bo_unref(&tmp);
 
+out_no_base_object:
+	ttm_bo_unref(&tmp);
+out_no_dmabuf:
 	ttm_read_unlock(&vmaster->lock);
 
-	return 0;
+	return ret;
 }
 
 int vmw_dmabuf_unref_ioctl(struct drm_device *dev, void *data,

include/drm/nouveau_drm.h

@@ -80,6 +80,7 @@ struct drm_nouveau_gpuobj_free {
 #define NOUVEAU_GETPARAM_VM_VRAM_BASE    12
 #define NOUVEAU_GETPARAM_GRAPH_UNITS     13
 #define NOUVEAU_GETPARAM_PTIMER_TIME     14
+#define NOUVEAU_GETPARAM_HAS_BO_USAGE    15
 struct drm_nouveau_getparam {
 	uint64_t param;
 	uint64_t value;
@@ -95,6 +96,12 @@ struct drm_nouveau_setparam {
 #define NOUVEAU_GEM_DOMAIN_GART      (1 << 2)
 #define NOUVEAU_GEM_DOMAIN_MAPPABLE  (1 << 3)
 
+#define NOUVEAU_GEM_TILE_LAYOUT_MASK 0x0000ff00
+#define NOUVEAU_GEM_TILE_16BPP       0x00000001
+#define NOUVEAU_GEM_TILE_32BPP       0x00000002
+#define NOUVEAU_GEM_TILE_ZETA        0x00000004
+#define NOUVEAU_GEM_TILE_NONCONTIG   0x00000008
+
 struct drm_nouveau_gem_info {
 	uint32_t handle;
 	uint32_t domain;