diff --git a/drivers/gpu/drm/nouveau/nouveau_acpi.c b/drivers/gpu/drm/nouveau/nouveau_acpi.c
index 1cf488247a16..48227e744753 100644
--- a/drivers/gpu/drm/nouveau/nouveau_acpi.c
+++ b/drivers/gpu/drm/nouveau/nouveau_acpi.c
@@ -90,21 +90,21 @@ int nouveau_hybrid_setup(struct drm_device *dev)
 {
 	int result;
 
-	if (nouveau_dsm(dev, NOUVEAU_DSM_ACTIVE, NOUVEAU_DSM_ACTIVE_QUERY,
+	if (nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_STATE,
 								&result))
 		return -ENODEV;
 
 	NV_INFO(dev, "_DSM hardware status gave 0x%x\n", result);
 
-	if (result & 0x1) {	/* Stamina mode - disable the external GPU */
+	if (result) { /* Ensure that the external GPU is enabled */
+		nouveau_dsm(dev, NOUVEAU_DSM_LED, NOUVEAU_DSM_LED_SPEED, NULL);
+		nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_SPEED,
+									NULL);
+	} else { /* Stamina mode - disable the external GPU */
 		nouveau_dsm(dev, NOUVEAU_DSM_LED, NOUVEAU_DSM_LED_STAMINA,
 									NULL);
 		nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_STAMINA,
 									NULL);
-	} else {		/* Ensure that the external GPU is enabled */
-		nouveau_dsm(dev, NOUVEAU_DSM_LED, NOUVEAU_DSM_LED_SPEED, NULL);
-		nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_SPEED,
-									NULL);
 	}
 
 	return 0;
diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.c b/drivers/gpu/drm/nouveau/nouveau_bios.c
index d7f8d8b4a4b8..2cd0fad17dac 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bios.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bios.c
@@ -1865,7 +1865,7 @@ init_compute_mem(struct nvbios *bios, uint16_t offset, struct init_exec *iexec)
 
 	struct drm_nouveau_private *dev_priv = bios->dev->dev_private;
 
-	if (dev_priv->card_type >= NV_50)
+	if (dev_priv->card_type >= NV_40)
 		return 1;
 
 	/*
@@ -3765,7 +3765,6 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
 	 */
 
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	struct init_exec iexec = {true, false};
 	struct nvbios *bios = &dev_priv->VBIOS;
 	uint8_t *table = &bios->data[bios->display.script_table_ptr];
 	uint8_t *otable = NULL;
@@ -3845,8 +3844,6 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
 		}
 	}
 
-	bios->display.output = dcbent;
-
 	if (pxclk == 0) {
 		script = ROM16(otable[6]);
 		if (!script) {
@@ -3855,7 +3852,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
 		}
 
 		NV_TRACE(dev, "0x%04X: parsing output script 0\n", script);
-		parse_init_table(bios, script, &iexec);
+		nouveau_bios_run_init_table(dev, script, dcbent);
 	} else
 	if (pxclk == -1) {
 		script = ROM16(otable[8]);
@@ -3865,7 +3862,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
 		}
 
 		NV_TRACE(dev, "0x%04X: parsing output script 1\n", script);
-		parse_init_table(bios, script, &iexec);
+		nouveau_bios_run_init_table(dev, script, dcbent);
 	} else
 	if (pxclk == -2) {
 		if (table[4] >= 12)
@@ -3878,7 +3875,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
 		}
 
 		NV_TRACE(dev, "0x%04X: parsing output script 2\n", script);
-		parse_init_table(bios, script, &iexec);
+		nouveau_bios_run_init_table(dev, script, dcbent);
 	} else
 	if (pxclk > 0) {
 		script = ROM16(otable[table[4] + i*6 + 2]);
@@ -3890,7 +3887,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
 		}
 
 		NV_TRACE(dev, "0x%04X: parsing clock script 0\n", script);
-		parse_init_table(bios, script, &iexec);
+		nouveau_bios_run_init_table(dev, script, dcbent);
 	} else
 	if (pxclk < 0) {
 		script = ROM16(otable[table[4] + i*6 + 4]);
@@ -3902,7 +3899,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
 		}
 
 		NV_TRACE(dev, "0x%04X: parsing clock script 1\n", script);
-		parse_init_table(bios, script, &iexec);
+		nouveau_bios_run_init_table(dev, script, dcbent);
 	}
 
 	return 0;
@@ -5864,10 +5861,13 @@ nouveau_bios_run_init_table(struct drm_device *dev, uint16_t table,
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nvbios *bios = &dev_priv->VBIOS;
 	struct init_exec iexec = { true, false };
+	unsigned long flags;
 
+	spin_lock_irqsave(&bios->lock, flags);
 	bios->display.output = dcbent;
 	parse_init_table(bios, table, &iexec);
 	bios->display.output = NULL;
+	spin_unlock_irqrestore(&bios->lock, flags);
 }
 
 static bool NVInitVBIOS(struct drm_device *dev)
@@ -5876,6 +5876,7 @@ static bool NVInitVBIOS(struct drm_device *dev)
 	struct nvbios *bios = &dev_priv->VBIOS;
 
 	memset(bios, 0, sizeof(struct nvbios));
+	spin_lock_init(&bios->lock);
 	bios->dev = dev;
 
 	if (!NVShadowVBIOS(dev, bios->data))
diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.h b/drivers/gpu/drm/nouveau/nouveau_bios.h
index 058e98c76d89..68446fd4146b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bios.h
+++ b/drivers/gpu/drm/nouveau/nouveau_bios.h
@@ -205,6 +205,8 @@ struct nvbios {
 	struct drm_device *dev;
 	struct nouveau_bios_info pub;
 
+	spinlock_t lock;
+
 	uint8_t data[NV_PROM_SIZE];
 	unsigned int length;
 	bool execute;
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c
index db0ed4c13f98..028719fddf76 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo.c
@@ -65,8 +65,10 @@ nouveau_bo_fixup_align(struct drm_device *dev,
 
 	/*
 	 * Some of the tile_flags have a periodic structure of N*4096 bytes,
-	 * align to to that as well as the page size. Overallocate memory to
-	 * avoid corruption of other buffer objects.
+	 * align to to that as well as the page size. Align the size to the
+	 * appropriate boundaries. This does imply that sizes are rounded up
+	 * 3-7 pages, so be aware of this and do not waste memory by allocating
+	 * many small buffers.
 	 */
 	if (dev_priv->card_type == NV_50) {
 		uint32_t block_size = nouveau_mem_fb_amount(dev) >> 15;
@@ -77,22 +79,20 @@ nouveau_bo_fixup_align(struct drm_device *dev,
 		case 0x2800:
 		case 0x4800:
 		case 0x7a00:
-			*size = roundup(*size, block_size);
 			if (is_power_of_2(block_size)) {
-				*size += 3 * block_size;
 				for (i = 1; i < 10; i++) {
 					*align = 12 * i * block_size;
 					if (!(*align % 65536))
 						break;
 				}
 			} else {
-				*size += 6 * block_size;
 				for (i = 1; i < 10; i++) {
 					*align = 8 * i * block_size;
 					if (!(*align % 65536))
 						break;
 				}
 			}
+			*size = roundup(*size, *align);
 			break;
 		default:
 			break;
diff --git a/drivers/gpu/drm/nouveau/nouveau_channel.c b/drivers/gpu/drm/nouveau/nouveau_channel.c
index 343d718a9667..2281f99da7fc 100644
--- a/drivers/gpu/drm/nouveau/nouveau_channel.c
+++ b/drivers/gpu/drm/nouveau/nouveau_channel.c
@@ -278,12 +278,11 @@ nouveau_channel_free(struct nouveau_channel *chan)
 	/* Ensure the channel is no longer active on the GPU */
 	pfifo->reassign(dev, false);
 
-	if (pgraph->channel(dev) == chan) {
-		pgraph->fifo_access(dev, false);
+	pgraph->fifo_access(dev, false);
+	if (pgraph->channel(dev) == chan)
 		pgraph->unload_context(dev);
-		pgraph->fifo_access(dev, true);
-	}
 	pgraph->destroy_context(chan);
+	pgraph->fifo_access(dev, true);
 
 	if (pfifo->channel_id(dev) == chan->id) {
 		pfifo->disable(dev);
diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c
index 7e6d673f3a23..d2f63353ea97 100644
--- a/drivers/gpu/drm/nouveau/nouveau_connector.c
+++ b/drivers/gpu/drm/nouveau/nouveau_connector.c
@@ -88,13 +88,14 @@ nouveau_connector_destroy(struct drm_connector *drm_connector)
 {
 	struct nouveau_connector *nv_connector =
 		nouveau_connector(drm_connector);
-	struct drm_device *dev = nv_connector->base.dev;
-
-	NV_DEBUG_KMS(dev, "\n");
+	struct drm_device *dev;
 
 	if (!nv_connector)
 		return;
 
+	dev = nv_connector->base.dev;
+	NV_DEBUG_KMS(dev, "\n");
+
 	kfree(nv_connector->edid);
 	drm_sysfs_connector_remove(drm_connector);
 	drm_connector_cleanup(drm_connector);
diff --git a/drivers/gpu/drm/nouveau/nouveau_dp.c b/drivers/gpu/drm/nouveau/nouveau_dp.c
index dd4937224220..f954ad93e81f 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dp.c
+++ b/drivers/gpu/drm/nouveau/nouveau_dp.c
@@ -502,12 +502,12 @@ nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr,
 			break;
 	}
 
-	if ((stat & NV50_AUXCH_STAT_COUNT) != data_nr) {
-		ret = -EREMOTEIO;
-		goto out;
-	}
-
 	if (cmd & 1) {
+		if ((stat & NV50_AUXCH_STAT_COUNT) != data_nr) {
+			ret = -EREMOTEIO;
+			goto out;
+		}
+
 		for (i = 0; i < 4; i++) {
 			data32[i] = nv_rd32(dev, NV50_AUXCH_DATA_IN(index, i));
 			NV_DEBUG_KMS(dev, "rd %d: 0x%08x\n", i, data32[i]);
diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.c b/drivers/gpu/drm/nouveau/nouveau_drv.c
index 343ab7f17ccc..da3b93b84502 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drv.c
+++ b/drivers/gpu/drm/nouveau/nouveau_drv.c
@@ -56,7 +56,7 @@ int nouveau_vram_pushbuf;
 module_param_named(vram_pushbuf, nouveau_vram_pushbuf, int, 0400);
 
 MODULE_PARM_DESC(vram_notify, "Force DMA notifiers to be in VRAM");
-int nouveau_vram_notify;
+int nouveau_vram_notify = 1;
 module_param_named(vram_notify, nouveau_vram_notify, int, 0400);
 
 MODULE_PARM_DESC(duallink, "Allow dual-link TMDS (>=GeForce 8)");
@@ -75,6 +75,14 @@ MODULE_PARM_DESC(ignorelid, "Ignore ACPI lid status");
 int nouveau_ignorelid = 0;
 module_param_named(ignorelid, nouveau_ignorelid, int, 0400);
 
+MODULE_PARM_DESC(noagp, "Disable all acceleration");
+int nouveau_noaccel = 0;
+module_param_named(noaccel, nouveau_noaccel, int, 0400);
+
+MODULE_PARM_DESC(noagp, "Disable fbcon acceleration");
+int nouveau_nofbaccel = 0;
+module_param_named(nofbaccel, nouveau_nofbaccel, int, 0400);
+
 MODULE_PARM_DESC(tv_norm, "Default TV norm.\n"
 		 "\t\tSupported: PAL, PAL-M, PAL-N, PAL-Nc, NTSC-M, NTSC-J,\n"
 		 "\t\t\thd480i, hd480p, hd576i, hd576p, hd720p, hd1080i.\n"
diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h
index 6b9690418bc7..5445cefdd03e 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drv.h
+++ b/drivers/gpu/drm/nouveau/nouveau_drv.h
@@ -678,6 +678,8 @@ extern int nouveau_reg_debug;
 extern char *nouveau_vbios;
 extern int nouveau_ctxfw;
 extern int nouveau_ignorelid;
+extern int nouveau_nofbaccel;
+extern int nouveau_noaccel;
 
 /* nouveau_state.c */
 extern void nouveau_preclose(struct drm_device *dev, struct drm_file *);
diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.c b/drivers/gpu/drm/nouveau/nouveau_fbcon.c
index 0b05c869e0e7..ea879a2efef3 100644
--- a/drivers/gpu/drm/nouveau/nouveau_fbcon.c
+++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.c
@@ -107,6 +107,34 @@ static struct fb_ops nouveau_fbcon_ops = {
 	.fb_setcmap = drm_fb_helper_setcmap,
 };
 
+static struct fb_ops nv04_fbcon_ops = {
+	.owner = THIS_MODULE,
+	.fb_check_var = drm_fb_helper_check_var,
+	.fb_set_par = drm_fb_helper_set_par,
+	.fb_setcolreg = drm_fb_helper_setcolreg,
+	.fb_fillrect = nv04_fbcon_fillrect,
+	.fb_copyarea = nv04_fbcon_copyarea,
+	.fb_imageblit = nv04_fbcon_imageblit,
+	.fb_sync = nouveau_fbcon_sync,
+	.fb_pan_display = drm_fb_helper_pan_display,
+	.fb_blank = drm_fb_helper_blank,
+	.fb_setcmap = drm_fb_helper_setcmap,
+};
+
+static struct fb_ops nv50_fbcon_ops = {
+	.owner = THIS_MODULE,
+	.fb_check_var = drm_fb_helper_check_var,
+	.fb_set_par = drm_fb_helper_set_par,
+	.fb_setcolreg = drm_fb_helper_setcolreg,
+	.fb_fillrect = nv50_fbcon_fillrect,
+	.fb_copyarea = nv50_fbcon_copyarea,
+	.fb_imageblit = nv50_fbcon_imageblit,
+	.fb_sync = nouveau_fbcon_sync,
+	.fb_pan_display = drm_fb_helper_pan_display,
+	.fb_blank = drm_fb_helper_blank,
+	.fb_setcmap = drm_fb_helper_setcmap,
+};
+
 static void nouveau_fbcon_gamma_set(struct drm_crtc *crtc, u16 red, u16 green,
 				    u16 blue, int regno)
 {
@@ -267,8 +295,12 @@ nouveau_fbcon_create(struct drm_device *dev, uint32_t fb_width,
 	dev_priv->fbdev_info = info;
 
 	strcpy(info->fix.id, "nouveaufb");
-	info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_COPYAREA |
-		      FBINFO_HWACCEL_FILLRECT | FBINFO_HWACCEL_IMAGEBLIT;
+	if (nouveau_nofbaccel)
+		info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_DISABLED;
+	else
+		info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_COPYAREA |
+			      FBINFO_HWACCEL_FILLRECT |
+			      FBINFO_HWACCEL_IMAGEBLIT;
 	info->fbops = &nouveau_fbcon_ops;
 	info->fix.smem_start = dev->mode_config.fb_base + nvbo->bo.offset -
 			       dev_priv->vm_vram_base;
@@ -316,13 +348,15 @@ nouveau_fbcon_create(struct drm_device *dev, uint32_t fb_width,
 	par->nouveau_fb = nouveau_fb;
 	par->dev = dev;
 
-	if (dev_priv->channel) {
+	if (dev_priv->channel && !nouveau_nofbaccel) {
 		switch (dev_priv->card_type) {
 		case NV_50:
 			nv50_fbcon_accel_init(info);
+			info->fbops = &nv50_fbcon_ops;
 			break;
 		default:
 			nv04_fbcon_accel_init(info);
+			info->fbops = &nv04_fbcon_ops;
 			break;
 		};
 	}
diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.h b/drivers/gpu/drm/nouveau/nouveau_fbcon.h
index 462e0b87b4bd..f9c34e1a8c11 100644
--- a/drivers/gpu/drm/nouveau/nouveau_fbcon.h
+++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.h
@@ -40,7 +40,13 @@ int nouveau_fbcon_remove(struct drm_device *dev, struct drm_framebuffer *fb);
 void nouveau_fbcon_restore(void);
 void nouveau_fbcon_zfill(struct drm_device *dev);
 
+void nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region);
+void nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect);
+void nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image);
 int nv04_fbcon_accel_init(struct fb_info *info);
+void nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect);
+void nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region);
+void nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image);
 int nv50_fbcon_accel_init(struct fb_info *info);
 
 void nouveau_fbcon_gpu_lockup(struct fb_info *info);
diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c
index 6ac804b0c9f9..70cc30803e3b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_gem.c
+++ b/drivers/gpu/drm/nouveau/nouveau_gem.c
@@ -925,7 +925,9 @@ nouveau_gem_ioctl_cpu_prep(struct drm_device *dev, void *data,
 	}
 
 	if (req->flags & NOUVEAU_GEM_CPU_PREP_NOBLOCK) {
+		spin_lock(&nvbo->bo.lock);
 		ret = ttm_bo_wait(&nvbo->bo, false, false, no_wait);
+		spin_unlock(&nvbo->bo.lock);
 	} else {
 		ret = ttm_bo_synccpu_write_grab(&nvbo->bo, no_wait);
 		if (ret == 0)
diff --git a/drivers/gpu/drm/nouveau/nouveau_grctx.c b/drivers/gpu/drm/nouveau/nouveau_grctx.c
index 419f4c2b3b89..c7ebec696747 100644
--- a/drivers/gpu/drm/nouveau/nouveau_grctx.c
+++ b/drivers/gpu/drm/nouveau/nouveau_grctx.c
@@ -97,8 +97,8 @@ nouveau_grctx_prog_load(struct drm_device *dev)
 		}
 
 		pgraph->ctxvals = kmalloc(fw->size, GFP_KERNEL);
-		if (!pgraph->ctxprog) {
-			NV_ERROR(dev, "OOM copying ctxprog\n");
+		if (!pgraph->ctxvals) {
+			NV_ERROR(dev, "OOM copying ctxvals\n");
 			release_firmware(fw);
 			nouveau_grctx_fini(dev);
 			return -ENOMEM;
diff --git a/drivers/gpu/drm/nouveau/nouveau_irq.c b/drivers/gpu/drm/nouveau/nouveau_irq.c
index 3b9bad66162a..447f9f69d6b1 100644
--- a/drivers/gpu/drm/nouveau/nouveau_irq.c
+++ b/drivers/gpu/drm/nouveau/nouveau_irq.c
@@ -211,6 +211,20 @@ nouveau_fifo_irq_handler(struct drm_device *dev)
 								get + 4);
 		}
 
+		if (status & NV_PFIFO_INTR_SEMAPHORE) {
+			uint32_t sem;
+
+			status &= ~NV_PFIFO_INTR_SEMAPHORE;
+			nv_wr32(dev, NV03_PFIFO_INTR_0,
+				NV_PFIFO_INTR_SEMAPHORE);
+
+			sem = nv_rd32(dev, NV10_PFIFO_CACHE1_SEMAPHORE);
+			nv_wr32(dev, NV10_PFIFO_CACHE1_SEMAPHORE, sem | 0x1);
+
+			nv_wr32(dev, NV03_PFIFO_CACHE1_GET, get + 4);
+			nv_wr32(dev, NV04_PFIFO_CACHE1_PULL0, 1);
+		}
+
 		if (status) {
 			NV_INFO(dev, "PFIFO_INTR 0x%08x - Ch %d\n",
 				status, chid);
@@ -566,86 +580,99 @@ nouveau_pgraph_irq_handler(struct drm_device *dev)
 static void
 nv50_pgraph_irq_handler(struct drm_device *dev)
 {
-	uint32_t status, nsource;
+	uint32_t status;
 
-	status = nv_rd32(dev, NV03_PGRAPH_INTR);
-	nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE);
+	while ((status = nv_rd32(dev, NV03_PGRAPH_INTR))) {
+		uint32_t nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE);
 
-	if (status & 0x00000001) {
-		nouveau_pgraph_intr_notify(dev, nsource);
-		status &= ~0x00000001;
-		nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000001);
-	}
+		if (status & 0x00000001) {
+			nouveau_pgraph_intr_notify(dev, nsource);
+			status &= ~0x00000001;
+			nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000001);
+		}
 
-	if (status & 0x00000010) {
-		nouveau_pgraph_intr_error(dev, nsource |
-					  NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD);
+		if (status & 0x00000010) {
+			nouveau_pgraph_intr_error(dev, nsource |
+					NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD);
 
-		status &= ~0x00000010;
-		nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000010);
-	}
+			status &= ~0x00000010;
+			nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000010);
+		}
 
-	if (status & 0x00001000) {
-		nv_wr32(dev, 0x400500, 0x00000000);
-		nv_wr32(dev, NV03_PGRAPH_INTR, NV_PGRAPH_INTR_CONTEXT_SWITCH);
-		nv_wr32(dev, NV40_PGRAPH_INTR_EN, nv_rd32(dev,
-			NV40_PGRAPH_INTR_EN) & ~NV_PGRAPH_INTR_CONTEXT_SWITCH);
-		nv_wr32(dev, 0x400500, 0x00010001);
+		if (status & 0x00001000) {
+			nv_wr32(dev, 0x400500, 0x00000000);
+			nv_wr32(dev, NV03_PGRAPH_INTR,
+				NV_PGRAPH_INTR_CONTEXT_SWITCH);
+			nv_wr32(dev, NV40_PGRAPH_INTR_EN, nv_rd32(dev,
+				NV40_PGRAPH_INTR_EN) &
+				~NV_PGRAPH_INTR_CONTEXT_SWITCH);
+			nv_wr32(dev, 0x400500, 0x00010001);
 
-		nv50_graph_context_switch(dev);
+			nv50_graph_context_switch(dev);
 
-		status &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH;
-	}
+			status &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH;
+		}
 
-	if (status & 0x00100000) {
-		nouveau_pgraph_intr_error(dev, nsource |
-					  NV03_PGRAPH_NSOURCE_DATA_ERROR);
+		if (status & 0x00100000) {
+			nouveau_pgraph_intr_error(dev, nsource |
+					NV03_PGRAPH_NSOURCE_DATA_ERROR);
 
-		status &= ~0x00100000;
-		nv_wr32(dev, NV03_PGRAPH_INTR, 0x00100000);
-	}
+			status &= ~0x00100000;
+			nv_wr32(dev, NV03_PGRAPH_INTR, 0x00100000);
+		}
 
-	if (status & 0x00200000) {
-		int r;
+		if (status & 0x00200000) {
+			int r;
 
-		nouveau_pgraph_intr_error(dev, nsource |
-					  NV03_PGRAPH_NSOURCE_PROTECTION_ERROR);
+			nouveau_pgraph_intr_error(dev, nsource |
+					NV03_PGRAPH_NSOURCE_PROTECTION_ERROR);
 
-		NV_ERROR(dev, "magic set 1:\n");
-		for (r = 0x408900; r <= 0x408910; r += 4)
-			NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r));
-		nv_wr32(dev, 0x408900, nv_rd32(dev, 0x408904) | 0xc0000000);
-		for (r = 0x408e08; r <= 0x408e24; r += 4)
-			NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r));
-		nv_wr32(dev, 0x408e08, nv_rd32(dev, 0x408e08) | 0xc0000000);
+			NV_ERROR(dev, "magic set 1:\n");
+			for (r = 0x408900; r <= 0x408910; r += 4)
+				NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r,
+					nv_rd32(dev, r));
+			nv_wr32(dev, 0x408900,
+				nv_rd32(dev, 0x408904) | 0xc0000000);
+			for (r = 0x408e08; r <= 0x408e24; r += 4)
+				NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r,
+							nv_rd32(dev, r));
+			nv_wr32(dev, 0x408e08,
+				nv_rd32(dev, 0x408e08) | 0xc0000000);
 
-		NV_ERROR(dev, "magic set 2:\n");
-		for (r = 0x409900; r <= 0x409910; r += 4)
-			NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r));
-		nv_wr32(dev, 0x409900, nv_rd32(dev, 0x409904) | 0xc0000000);
-		for (r = 0x409e08; r <= 0x409e24; r += 4)
-			NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r));
-		nv_wr32(dev, 0x409e08, nv_rd32(dev, 0x409e08) | 0xc0000000);
+			NV_ERROR(dev, "magic set 2:\n");
+			for (r = 0x409900; r <= 0x409910; r += 4)
+				NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r,
+					nv_rd32(dev, r));
+			nv_wr32(dev, 0x409900,
+				nv_rd32(dev, 0x409904) | 0xc0000000);
+			for (r = 0x409e08; r <= 0x409e24; r += 4)
+				NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r,
+					nv_rd32(dev, r));
+			nv_wr32(dev, 0x409e08,
+				nv_rd32(dev, 0x409e08) | 0xc0000000);
 
-		status &= ~0x00200000;
-		nv_wr32(dev, NV03_PGRAPH_NSOURCE, nsource);
-		nv_wr32(dev, NV03_PGRAPH_INTR, 0x00200000);
-	}
+			status &= ~0x00200000;
+			nv_wr32(dev, NV03_PGRAPH_NSOURCE, nsource);
+			nv_wr32(dev, NV03_PGRAPH_INTR, 0x00200000);
+		}
 
-	if (status) {
-		NV_INFO(dev, "Unhandled PGRAPH_INTR - 0x%08x\n", status);
-		nv_wr32(dev, NV03_PGRAPH_INTR, status);
-	}
+		if (status) {
+			NV_INFO(dev, "Unhandled PGRAPH_INTR - 0x%08x\n",
+				status);
+			nv_wr32(dev, NV03_PGRAPH_INTR, status);
+		}
 
-	{
-		const int isb = (1 << 16) | (1 << 0);
+		{
+			const int isb = (1 << 16) | (1 << 0);
 
-		if ((nv_rd32(dev, 0x400500) & isb) != isb)
-			nv_wr32(dev, 0x400500, nv_rd32(dev, 0x400500) | isb);
-		nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) & ~(1 << 31));
+			if ((nv_rd32(dev, 0x400500) & isb) != isb)
+				nv_wr32(dev, 0x400500,
+					nv_rd32(dev, 0x400500) | isb);
+		}
 	}
 
 	nv_wr32(dev, NV03_PMC_INTR_0, NV_PMC_INTR_0_PGRAPH_PENDING);
+	nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) & ~(1 << 31));
 }
 
 static void
diff --git a/drivers/gpu/drm/nouveau/nouveau_notifier.c b/drivers/gpu/drm/nouveau/nouveau_notifier.c
index 6c66a34b6345..d99dc087f9b1 100644
--- a/drivers/gpu/drm/nouveau/nouveau_notifier.c
+++ b/drivers/gpu/drm/nouveau/nouveau_notifier.c
@@ -34,15 +34,20 @@ nouveau_notifier_init_channel(struct nouveau_channel *chan)
 {
 	struct drm_device *dev = chan->dev;
 	struct nouveau_bo *ntfy = NULL;
+	uint32_t flags;
 	int ret;
 
-	ret = nouveau_gem_new(dev, NULL, PAGE_SIZE, 0, nouveau_vram_notify ?
-			      TTM_PL_FLAG_VRAM : TTM_PL_FLAG_TT,
+	if (nouveau_vram_notify)
+		flags = TTM_PL_FLAG_VRAM;
+	else
+		flags = TTM_PL_FLAG_TT;
+
+	ret = nouveau_gem_new(dev, NULL, PAGE_SIZE, 0, flags,
 			      0, 0x0000, false, true, &ntfy);
 	if (ret)
 		return ret;
 
-	ret = nouveau_bo_pin(ntfy, TTM_PL_FLAG_VRAM);
+	ret = nouveau_bo_pin(ntfy, flags);
 	if (ret)
 		goto out_err;
 
@@ -128,6 +133,8 @@ nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle,
 			target = NV_DMA_TARGET_PCI;
 		} else {
 			target = NV_DMA_TARGET_AGP;
+			if (dev_priv->card_type >= NV_50)
+				offset += dev_priv->vm_gart_base;
 		}
 	} else {
 		NV_ERROR(dev, "Bad DMA target, mem_type %d!\n",
diff --git a/drivers/gpu/drm/nouveau/nouveau_object.c b/drivers/gpu/drm/nouveau/nouveau_object.c
index 6c2cf81716df..e7c100ba63a1 100644
--- a/drivers/gpu/drm/nouveau/nouveau_object.c
+++ b/drivers/gpu/drm/nouveau/nouveau_object.c
@@ -885,11 +885,12 @@ int
 nouveau_gpuobj_sw_new(struct nouveau_channel *chan, int class,
 		      struct nouveau_gpuobj **gpuobj_ret)
 {
-	struct drm_nouveau_private *dev_priv = chan->dev->dev_private;
+	struct drm_nouveau_private *dev_priv;
 	struct nouveau_gpuobj *gpuobj;
 
 	if (!chan || !gpuobj_ret || *gpuobj_ret != NULL)
 		return -EINVAL;
+	dev_priv = chan->dev->dev_private;
 
 	gpuobj = kzalloc(sizeof(*gpuobj), GFP_KERNEL);
 	if (!gpuobj)
diff --git a/drivers/gpu/drm/nouveau/nouveau_reg.h b/drivers/gpu/drm/nouveau/nouveau_reg.h
index 251f1b3b38b9..aa9b310e41be 100644
--- a/drivers/gpu/drm/nouveau/nouveau_reg.h
+++ b/drivers/gpu/drm/nouveau/nouveau_reg.h
@@ -99,6 +99,7 @@
  * the card will hang early on in the X init process.
  */
 #    define NV_PMC_ENABLE_UNK13                               (1<<13)
+#define NV40_PMC_GRAPH_UNITS				   0x00001540
 #define NV40_PMC_BACKLIGHT				   0x000015f0
 #	define NV40_PMC_BACKLIGHT_MASK			   0x001f0000
 #define NV40_PMC_1700                                      0x00001700
diff --git a/drivers/gpu/drm/nouveau/nouveau_sgdma.c b/drivers/gpu/drm/nouveau/nouveau_sgdma.c
index 4c7f1e403e80..ed1590577b6c 100644
--- a/drivers/gpu/drm/nouveau/nouveau_sgdma.c
+++ b/drivers/gpu/drm/nouveau/nouveau_sgdma.c
@@ -54,11 +54,12 @@ static void
 nouveau_sgdma_clear(struct ttm_backend *be)
 {
 	struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
-	struct drm_device *dev = nvbe->dev;
-
-	NV_DEBUG(nvbe->dev, "\n");
+	struct drm_device *dev;
 
 	if (nvbe && nvbe->pages) {
+		dev = nvbe->dev;
+		NV_DEBUG(dev, "\n");
+
 		if (nvbe->bound)
 			be->func->unbind(be);
 
diff --git a/drivers/gpu/drm/nouveau/nouveau_state.c b/drivers/gpu/drm/nouveau/nouveau_state.c
index f2d0187ba152..a4851af5b05e 100644
--- a/drivers/gpu/drm/nouveau/nouveau_state.c
+++ b/drivers/gpu/drm/nouveau/nouveau_state.c
@@ -310,6 +310,14 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
 static unsigned int
 nouveau_vga_set_decode(void *priv, bool state)
 {
+	struct drm_device *dev = priv;
+	struct drm_nouveau_private *dev_priv = dev->dev_private;
+
+	if (dev_priv->chipset >= 0x40)
+		nv_wr32(dev, 0x88054, state);
+	else
+		nv_wr32(dev, 0x1854, state);
+
 	if (state)
 		return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM |
 		       VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM;
@@ -427,15 +435,19 @@ nouveau_card_init(struct drm_device *dev)
 	if (ret)
 		goto out_timer;
 
-	/* PGRAPH */
-	ret = engine->graph.init(dev);
-	if (ret)
-		goto out_fb;
+	if (nouveau_noaccel)
+		engine->graph.accel_blocked = true;
+	else {
+		/* PGRAPH */
+		ret = engine->graph.init(dev);
+		if (ret)
+			goto out_fb;
 
-	/* PFIFO */
-	ret = engine->fifo.init(dev);
-	if (ret)
-		goto out_graph;
+		/* PFIFO */
+		ret = engine->fifo.init(dev);
+		if (ret)
+			goto out_graph;
+	}
 
 	/* this call irq_preinstall, register irq handler and
 	 * call irq_postinstall
@@ -479,9 +491,11 @@ nouveau_card_init(struct drm_device *dev)
 out_irq:
 	drm_irq_uninstall(dev);
 out_fifo:
-	engine->fifo.takedown(dev);
+	if (!nouveau_noaccel)
+		engine->fifo.takedown(dev);
 out_graph:
-	engine->graph.takedown(dev);
+	if (!nouveau_noaccel)
+		engine->graph.takedown(dev);
 out_fb:
 	engine->fb.takedown(dev);
 out_timer:
@@ -518,8 +532,10 @@ static void nouveau_card_takedown(struct drm_device *dev)
 			dev_priv->channel = NULL;
 		}
 
-		engine->fifo.takedown(dev);
-		engine->graph.takedown(dev);
+		if (!nouveau_noaccel) {
+			engine->fifo.takedown(dev);
+			engine->graph.takedown(dev);
+		}
 		engine->fb.takedown(dev);
 		engine->timer.takedown(dev);
 		engine->mc.takedown(dev);
@@ -817,6 +833,15 @@ int nouveau_ioctl_getparam(struct drm_device *dev, void *data,
 	case NOUVEAU_GETPARAM_VM_VRAM_BASE:
 		getparam->value = dev_priv->vm_vram_base;
 		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
+		 * family anyway... */
+		if (dev_priv->chipset >= 0x40) {
+			getparam->value = nv_rd32(dev, NV40_PMC_GRAPH_UNITS);
+			break;
+		}
+		/* FALLTHRU */
 	default:
 		NV_ERROR(dev, "unknown parameter %lld\n", getparam->param);
 		return -EINVAL;
diff --git a/drivers/gpu/drm/nouveau/nv04_fbcon.c b/drivers/gpu/drm/nouveau/nv04_fbcon.c
index d910873c1368..fd01caabd5c3 100644
--- a/drivers/gpu/drm/nouveau/nv04_fbcon.c
+++ b/drivers/gpu/drm/nouveau/nv04_fbcon.c
@@ -27,7 +27,7 @@
 #include "nouveau_dma.h"
 #include "nouveau_fbcon.h"
 
-static void
+void
 nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
 {
 	struct nouveau_fbcon_par *par = info->par;
@@ -54,7 +54,7 @@ nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
 	FIRE_RING(chan);
 }
 
-static void
+void
 nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
 {
 	struct nouveau_fbcon_par *par = info->par;
@@ -88,7 +88,7 @@ nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
 	FIRE_RING(chan);
 }
 
-static void
+void
 nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
 {
 	struct nouveau_fbcon_par *par = info->par;
@@ -307,9 +307,6 @@ nv04_fbcon_accel_init(struct fb_info *info)
 
 	FIRE_RING(chan);
 
-	info->fbops->fb_fillrect = nv04_fbcon_fillrect;
-	info->fbops->fb_copyarea = nv04_fbcon_copyarea;
-	info->fbops->fb_imageblit = nv04_fbcon_imageblit;
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/nouveau/nv50_crtc.c b/drivers/gpu/drm/nouveau/nv50_crtc.c
index 40b7360841f8..d1a651e3400c 100644
--- a/drivers/gpu/drm/nouveau/nv50_crtc.c
+++ b/drivers/gpu/drm/nouveau/nv50_crtc.c
@@ -298,14 +298,17 @@ nv50_crtc_set_clock(struct drm_device *dev, int head, int pclk)
 static void
 nv50_crtc_destroy(struct drm_crtc *crtc)
 {
-	struct drm_device *dev = crtc->dev;
-	struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc);
-
-	NV_DEBUG_KMS(dev, "\n");
+	struct drm_device *dev;
+	struct nouveau_crtc *nv_crtc;
 
 	if (!crtc)
 		return;
 
+	dev = crtc->dev;
+	nv_crtc = nouveau_crtc(crtc);
+
+	NV_DEBUG_KMS(dev, "\n");
+
 	drm_crtc_cleanup(&nv_crtc->base);
 
 	nv50_cursor_fini(nv_crtc);
diff --git a/drivers/gpu/drm/nouveau/nv50_fbcon.c b/drivers/gpu/drm/nouveau/nv50_fbcon.c
index e4f279ee61cf..0f57cdf7ccb2 100644
--- a/drivers/gpu/drm/nouveau/nv50_fbcon.c
+++ b/drivers/gpu/drm/nouveau/nv50_fbcon.c
@@ -3,7 +3,7 @@
 #include "nouveau_dma.h"
 #include "nouveau_fbcon.h"
 
-static void
+void
 nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
 {
 	struct nouveau_fbcon_par *par = info->par;
@@ -46,7 +46,7 @@ nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
 	FIRE_RING(chan);
 }
 
-static void
+void
 nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
 {
 	struct nouveau_fbcon_par *par = info->par;
@@ -81,7 +81,7 @@ nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
 	FIRE_RING(chan);
 }
 
-static void
+void
 nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
 {
 	struct nouveau_fbcon_par *par = info->par;
@@ -262,9 +262,6 @@ nv50_fbcon_accel_init(struct fb_info *info)
 	OUT_RING(chan, info->fix.smem_start - dev_priv->fb_phys +
 			 dev_priv->vm_vram_base);
 
-	info->fbops->fb_fillrect = nv50_fbcon_fillrect;
-	info->fbops->fb_copyarea = nv50_fbcon_copyarea;
-	info->fbops->fb_imageblit = nv50_fbcon_imageblit;
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/nouveau/nv50_fifo.c b/drivers/gpu/drm/nouveau/nv50_fifo.c
index 32b244bcb482..204a79ff10f4 100644
--- a/drivers/gpu/drm/nouveau/nv50_fifo.c
+++ b/drivers/gpu/drm/nouveau/nv50_fifo.c
@@ -317,17 +317,20 @@ void
 nv50_fifo_destroy_context(struct nouveau_channel *chan)
 {
 	struct drm_device *dev = chan->dev;
+	struct nouveau_gpuobj_ref *ramfc = chan->ramfc;
 
 	NV_DEBUG(dev, "ch%d\n", chan->id);
 
-	nouveau_gpuobj_ref_del(dev, &chan->ramfc);
-	nouveau_gpuobj_ref_del(dev, &chan->cache);
-
+	/* This will ensure the channel is seen as disabled. */
+	chan->ramfc = NULL;
 	nv50_fifo_channel_disable(dev, chan->id, false);
 
 	/* Dummy channel, also used on ch 127 */
 	if (chan->id == 0)
 		nv50_fifo_channel_disable(dev, 127, false);
+
+	nouveau_gpuobj_ref_del(dev, &ramfc);
+	nouveau_gpuobj_ref_del(dev, &chan->cache);
 }
 
 int
diff --git a/drivers/gpu/drm/nouveau/nv50_graph.c b/drivers/gpu/drm/nouveau/nv50_graph.c
index 20319e59d368..6d504801b514 100644
--- a/drivers/gpu/drm/nouveau/nv50_graph.c
+++ b/drivers/gpu/drm/nouveau/nv50_graph.c
@@ -165,6 +165,12 @@ nv50_graph_channel(struct drm_device *dev)
 	uint32_t inst;
 	int i;
 
+	/* Be sure we're not in the middle of a context switch or bad things
+	 * will happen, such as unloading the wrong pgraph context.
+	 */
+	if (!nv_wait(0x400300, 0x00000001, 0x00000000))
+		NV_ERROR(dev, "Ctxprog is still running\n");
+
 	inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR);
 	if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED))
 		return NULL;
@@ -275,7 +281,7 @@ nv50_graph_load_context(struct nouveau_channel *chan)
 int
 nv50_graph_unload_context(struct drm_device *dev)
 {
-	uint32_t inst, fifo = nv_rd32(dev, 0x400500);
+	uint32_t inst;
 
 	inst  = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR);
 	if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED))
@@ -283,12 +289,10 @@ nv50_graph_unload_context(struct drm_device *dev)
 	inst &= NV50_PGRAPH_CTXCTL_CUR_INSTANCE;
 
 	nouveau_wait_for_idle(dev);
-	nv_wr32(dev, 0x400500, fifo & ~1);
 	nv_wr32(dev, 0x400784, inst);
 	nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x20);
 	nv_wr32(dev, 0x400304, nv_rd32(dev, 0x400304) | 0x01);
 	nouveau_wait_for_idle(dev);
-	nv_wr32(dev, 0x400500, fifo);
 
 	nv_wr32(dev, NV50_PGRAPH_CTXCTL_CUR, inst);
 	return 0;
diff --git a/drivers/gpu/drm/nouveau/nv50_sor.c b/drivers/gpu/drm/nouveau/nv50_sor.c
index ecf1936b8224..c2fff543b06f 100644
--- a/drivers/gpu/drm/nouveau/nv50_sor.c
+++ b/drivers/gpu/drm/nouveau/nv50_sor.c
@@ -101,6 +101,7 @@ nv50_sor_dpms(struct drm_encoder *encoder, int mode)
 		struct nouveau_encoder *nvenc = nouveau_encoder(enc);
 
 		if (nvenc == nv_encoder ||
+		    nvenc->disconnect != nv50_sor_disconnect ||
 		    nvenc->dcb->or != nv_encoder->dcb->or)
 			continue;
 
diff --git a/drivers/gpu/drm/radeon/Kconfig b/drivers/gpu/drm/radeon/Kconfig
index 5982321be4d5..1c02d23f6fcc 100644
--- a/drivers/gpu/drm/radeon/Kconfig
+++ b/drivers/gpu/drm/radeon/Kconfig
@@ -1,10 +1,14 @@
 config DRM_RADEON_KMS
-	bool "Enable modesetting on radeon by default"
+	bool "Enable modesetting on radeon by default - NEW DRIVER"
 	depends on DRM_RADEON
 	help
-	  Choose this option if you want kernel modesetting enabled by default,
-	  and you have a new enough userspace to support this. Running old
-	  userspaces with this enabled will cause pain.
+	  Choose this option if you want kernel modesetting enabled by default.
+
+	  This is a completely new driver. It's only part of the existing drm
+	  for compatibility reasons. It requires an entirely different graphics
+	  stack above it and works very differently from the old drm stack.
+	  i.e. don't enable this unless you know what you are doing it may
+	  cause issues or bugs compared to the previous userspace driver stack.
 
 	  When kernel modesetting is enabled the IOCTL of radeon/drm
 	  driver are considered as invalid and an error message is printed
diff --git a/drivers/gpu/drm/radeon/atombios_dp.c b/drivers/gpu/drm/radeon/atombios_dp.c
index 71060114d5de..b32eeea5bb8b 100644
--- a/drivers/gpu/drm/radeon/atombios_dp.c
+++ b/drivers/gpu/drm/radeon/atombios_dp.c
@@ -332,11 +332,13 @@ bool radeon_process_aux_ch(struct radeon_i2c_chan *chan, u8 *req_bytes,
 	PROCESS_AUX_CHANNEL_TRANSACTION_PS_ALLOCATION args;
 	int index = GetIndexIntoMasterTable(COMMAND, ProcessAuxChannelTransaction);
 	unsigned char *base;
+	int retry_count = 0;
 
 	memset(&args, 0, sizeof(args));
 
 	base = (unsigned char *)rdev->mode_info.atom_context->scratch;
 
+retry:
 	memcpy(base, req_bytes, num_bytes);
 
 	args.lpAuxRequest = 0;
@@ -347,10 +349,12 @@ bool radeon_process_aux_ch(struct radeon_i2c_chan *chan, u8 *req_bytes,
 
 	atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 
-	if (args.ucReplyStatus) {
-		DRM_DEBUG("failed to get auxch %02x%02x %02x %02x 0x%02x %02x\n",
+	if (args.ucReplyStatus && !args.ucDataOutLen) {
+		if (args.ucReplyStatus == 0x20 && retry_count < 10)
+			goto retry;
+		DRM_DEBUG("failed to get auxch %02x%02x %02x %02x 0x%02x %02x after %d retries\n",
 			  req_bytes[1], req_bytes[0], req_bytes[2], req_bytes[3],
-			  chan->rec.i2c_id, args.ucReplyStatus);
+			  chan->rec.i2c_id, args.ucReplyStatus, retry_count);
 		return false;
 	}
 
diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c
index a1198d99cdf9..2ffcf5a03551 100644
--- a/drivers/gpu/drm/radeon/r600.c
+++ b/drivers/gpu/drm/radeon/r600.c
@@ -1950,6 +1950,13 @@ int r600_resume(struct radeon_device *rdev)
 		DRM_ERROR("radeon: failled testing IB (%d).\n", r);
 		return r;
 	}
+
+	r = r600_audio_init(rdev);
+	if (r) {
+		DRM_ERROR("radeon: audio resume failed\n");
+		return r;
+	}
+
 	return r;
 }
 
@@ -1957,6 +1964,7 @@ int r600_suspend(struct radeon_device *rdev)
 {
 	int r;
 
+	r600_audio_fini(rdev);
 	/* FIXME: we should wait for ring to be empty */
 	r600_cp_stop(rdev);
 	rdev->cp.ready = false;
diff --git a/drivers/gpu/drm/radeon/r600_audio.c b/drivers/gpu/drm/radeon/r600_audio.c
index b1c1d3433454..0dcb6904c4ff 100644
--- a/drivers/gpu/drm/radeon/r600_audio.c
+++ b/drivers/gpu/drm/radeon/r600_audio.c
@@ -261,7 +261,6 @@ void r600_audio_fini(struct radeon_device *rdev)
 	if (!r600_audio_chipset_supported(rdev))
 		return;
 
-	WREG32_P(R600_AUDIO_ENABLE, 0x0, ~0x81000000);
-
 	del_timer(&rdev->audio_timer);
+	WREG32_P(R600_AUDIO_ENABLE, 0x0, ~0x81000000);
 }
diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c
index fa82ca74324e..2dcda6115874 100644
--- a/drivers/gpu/drm/radeon/radeon_atombios.c
+++ b/drivers/gpu/drm/radeon/radeon_atombios.c
@@ -287,6 +287,15 @@ static bool radeon_atom_apply_quirks(struct drm_device *dev,
 			*connector_type = DRM_MODE_CONNECTOR_DVID;
 	}
 
+	/* XFX Pine Group device rv730 reports no VGA DDC lines
+	 * even though they are wired up to record 0x93
+	 */
+	if ((dev->pdev->device == 0x9498) &&
+	    (dev->pdev->subsystem_vendor == 0x1682) &&
+	    (dev->pdev->subsystem_device == 0x2452)) {
+		struct radeon_device *rdev = dev->dev_private;
+		*i2c_bus = radeon_lookup_i2c_gpio(rdev, 0x93);
+	}
 	return true;
 }
 
diff --git a/drivers/gpu/drm/radeon/radeon_benchmark.c b/drivers/gpu/drm/radeon/radeon_benchmark.c
index 4ddfd4b5bc51..7932dc4d6b90 100644
--- a/drivers/gpu/drm/radeon/radeon_benchmark.c
+++ b/drivers/gpu/drm/radeon/radeon_benchmark.c
@@ -65,31 +65,42 @@ void radeon_benchmark_move(struct radeon_device *rdev, unsigned bsize,
 	if (r) {
 		goto out_cleanup;
 	}
-	start_jiffies = jiffies;
-	for (i = 0; i < n; i++) {
-		r = radeon_fence_create(rdev, &fence);
-		if (r) {
-			goto out_cleanup;
+
+	/* r100 doesn't have dma engine so skip the test */
+	if (rdev->asic->copy_dma) {
+
+		start_jiffies = jiffies;
+		for (i = 0; i < n; i++) {
+			r = radeon_fence_create(rdev, &fence);
+			if (r) {
+				goto out_cleanup;
+			}
+
+			r = radeon_copy_dma(rdev, saddr, daddr,
+					size / RADEON_GPU_PAGE_SIZE, fence);
+
+			if (r) {
+				goto out_cleanup;
+			}
+			r = radeon_fence_wait(fence, false);
+			if (r) {
+				goto out_cleanup;
+			}
+			radeon_fence_unref(&fence);
 		}
-		r = radeon_copy_dma(rdev, saddr, daddr, size / RADEON_GPU_PAGE_SIZE, fence);
-		if (r) {
-			goto out_cleanup;
+		end_jiffies = jiffies;
+		time = end_jiffies - start_jiffies;
+		time = jiffies_to_msecs(time);
+		if (time > 0) {
+			i = ((n * size) >> 10) / time;
+			printk(KERN_INFO "radeon: dma %u bo moves of %ukb from"
+					" %d to %d in %lums (%ukb/ms %ukb/s %uM/s)\n",
+					n, size >> 10,
+					sdomain, ddomain, time,
+					i, i * 1000, (i * 1000) / 1024);
 		}
-		r = radeon_fence_wait(fence, false);
-		if (r) {
-			goto out_cleanup;
-		}
-		radeon_fence_unref(&fence);
-	}
-	end_jiffies = jiffies;
-	time = end_jiffies - start_jiffies;
-	time = jiffies_to_msecs(time);
-	if (time > 0) {
-		i = ((n * size) >> 10) / time;
-		printk(KERN_INFO "radeon: dma %u bo moves of %ukb from %d to %d"
-		       " in %lums (%ukb/ms %ukb/s %uM/s)\n", n, size >> 10,
-		       sdomain, ddomain, time, i, i * 1000, (i * 1000) / 1024);
 	}
+
 	start_jiffies = jiffies;
 	for (i = 0; i < n; i++) {
 		r = radeon_fence_create(rdev, &fence);
diff --git a/drivers/gpu/drm/radeon/radeon_connectors.c b/drivers/gpu/drm/radeon/radeon_connectors.c
index 2d8e5a70f284..238188540017 100644
--- a/drivers/gpu/drm/radeon/radeon_connectors.c
+++ b/drivers/gpu/drm/radeon/radeon_connectors.c
@@ -580,16 +580,18 @@ static enum drm_connector_status radeon_vga_detect(struct drm_connector *connect
 	struct radeon_connector *radeon_connector = to_radeon_connector(connector);
 	struct drm_encoder *encoder;
 	struct drm_encoder_helper_funcs *encoder_funcs;
-	bool dret;
+	bool dret = false;
 	enum drm_connector_status ret = connector_status_disconnected;
 
 	encoder = radeon_best_single_encoder(connector);
 	if (!encoder)
 		ret = connector_status_disconnected;
 
-	radeon_i2c_do_lock(radeon_connector->ddc_bus, 1);
-	dret = radeon_ddc_probe(radeon_connector);
-	radeon_i2c_do_lock(radeon_connector->ddc_bus, 0);
+	if (radeon_connector->ddc_bus) {
+		radeon_i2c_do_lock(radeon_connector->ddc_bus, 1);
+		dret = radeon_ddc_probe(radeon_connector);
+		radeon_i2c_do_lock(radeon_connector->ddc_bus, 0);
+	}
 	if (dret) {
 		if (radeon_connector->edid) {
 			kfree(radeon_connector->edid);
@@ -740,11 +742,13 @@ static enum drm_connector_status radeon_dvi_detect(struct drm_connector *connect
 	struct drm_mode_object *obj;
 	int i;
 	enum drm_connector_status ret = connector_status_disconnected;
-	bool dret;
+	bool dret = false;
 
-	radeon_i2c_do_lock(radeon_connector->ddc_bus, 1);
-	dret = radeon_ddc_probe(radeon_connector);
-	radeon_i2c_do_lock(radeon_connector->ddc_bus, 0);
+	if (radeon_connector->ddc_bus) {
+		radeon_i2c_do_lock(radeon_connector->ddc_bus, 1);
+		dret = radeon_ddc_probe(radeon_connector);
+		radeon_i2c_do_lock(radeon_connector->ddc_bus, 0);
+	}
 	if (dret) {
 		if (radeon_connector->edid) {
 			kfree(radeon_connector->edid);
diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c
index 6a92f994cc26..7e17a362b54b 100644
--- a/drivers/gpu/drm/radeon/radeon_display.c
+++ b/drivers/gpu/drm/radeon/radeon_display.c
@@ -278,7 +278,7 @@ static void radeon_print_display_setup(struct drm_device *dev)
 		DRM_INFO("  %s\n", connector_names[connector->connector_type]);
 		if (radeon_connector->hpd.hpd != RADEON_HPD_NONE)
 			DRM_INFO("  %s\n", hpd_names[radeon_connector->hpd.hpd]);
-		if (radeon_connector->ddc_bus)
+		if (radeon_connector->ddc_bus) {
 			DRM_INFO("  DDC: 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x 0x%x\n",
 				 radeon_connector->ddc_bus->rec.mask_clk_reg,
 				 radeon_connector->ddc_bus->rec.mask_data_reg,
@@ -288,6 +288,15 @@ static void radeon_print_display_setup(struct drm_device *dev)
 				 radeon_connector->ddc_bus->rec.en_data_reg,
 				 radeon_connector->ddc_bus->rec.y_clk_reg,
 				 radeon_connector->ddc_bus->rec.y_data_reg);
+		} else {
+			if (connector->connector_type == DRM_MODE_CONNECTOR_VGA ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_DVII ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_DVID ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_DVIA ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_HDMIA ||
+			    connector->connector_type == DRM_MODE_CONNECTOR_HDMIB)
+				DRM_INFO("  DDC: no ddc bus - possible BIOS bug - please report to xorg-driver-ati@lists.x.org\n");
+		}
 		DRM_INFO("  Encoders:\n");
 		list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
 			radeon_encoder = to_radeon_encoder(encoder);
diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c
index 3ba213d1b06c..d71e346e9ab5 100644
--- a/drivers/gpu/drm/radeon/radeon_fb.c
+++ b/drivers/gpu/drm/radeon/radeon_fb.c
@@ -248,7 +248,7 @@ int radeonfb_create(struct drm_device *dev,
 	if (ret)
 		goto out_unref;
 
-	memset_io(fbptr, 0xff, aligned_size);
+	memset_io(fbptr, 0x0, aligned_size);
 
 	strcpy(info->fix.id, "radeondrmfb");
 
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
index 135be9688c90..356dc935ec13 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
@@ -39,10 +39,10 @@
 #include "ttm/ttm_execbuf_util.h"
 #include "ttm/ttm_module.h"
 
-#define VMWGFX_DRIVER_DATE "20090724"
-#define VMWGFX_DRIVER_MAJOR 0
-#define VMWGFX_DRIVER_MINOR 1
-#define VMWGFX_DRIVER_PATCHLEVEL 2
+#define VMWGFX_DRIVER_DATE "20100209"
+#define VMWGFX_DRIVER_MAJOR 1
+#define VMWGFX_DRIVER_MINOR 0
+#define VMWGFX_DRIVER_PATCHLEVEL 0
 #define VMWGFX_FILE_PAGE_OFFSET 0x00100000
 #define VMWGFX_FIFO_STATIC_SIZE (1024*1024)
 #define VMWGFX_MAX_RELOCATIONS 2048
@@ -113,6 +113,7 @@ struct vmw_fifo_state {
 	unsigned long static_buffer_size;
 	bool using_bounce_buffer;
 	uint32_t capabilities;
+	struct mutex fifo_mutex;
 	struct rw_semaphore rwsem;
 };
 
@@ -213,7 +214,7 @@ struct vmw_private {
 	 * Fencing and IRQs.
 	 */
 
-	uint32_t fence_seq;
+	atomic_t fence_seq;
 	wait_queue_head_t fence_queue;
 	wait_queue_head_t fifo_queue;
 	atomic_t fence_queue_waiters;
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c
index 4157547cc6e4..39d43a01d846 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c
@@ -74,6 +74,7 @@ int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
 	fifo->reserved_size = 0;
 	fifo->using_bounce_buffer = false;
 
+	mutex_init(&fifo->fifo_mutex);
 	init_rwsem(&fifo->rwsem);
 
 	/*
@@ -117,7 +118,7 @@ int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
 		 (unsigned int) min,
 		 (unsigned int) fifo->capabilities);
 
-	dev_priv->fence_seq = dev_priv->last_read_sequence;
+	atomic_set(&dev_priv->fence_seq, dev_priv->last_read_sequence);
 	iowrite32(dev_priv->last_read_sequence, fifo_mem + SVGA_FIFO_FENCE);
 
 	return vmw_fifo_send_fence(dev_priv, &dummy);
@@ -283,7 +284,7 @@ void *vmw_fifo_reserve(struct vmw_private *dev_priv, uint32_t bytes)
 	uint32_t reserveable = fifo_state->capabilities & SVGA_FIFO_CAP_RESERVE;
 	int ret;
 
-	down_write(&fifo_state->rwsem);
+	mutex_lock(&fifo_state->fifo_mutex);
 	max = ioread32(fifo_mem + SVGA_FIFO_MAX);
 	min = ioread32(fifo_mem + SVGA_FIFO_MIN);
 	next_cmd = ioread32(fifo_mem + SVGA_FIFO_NEXT_CMD);
@@ -351,7 +352,7 @@ void *vmw_fifo_reserve(struct vmw_private *dev_priv, uint32_t bytes)
 	}
 out_err:
 	fifo_state->reserved_size = 0;
-	up_write(&fifo_state->rwsem);
+	mutex_unlock(&fifo_state->fifo_mutex);
 	return NULL;
 }
 
@@ -426,6 +427,7 @@ void vmw_fifo_commit(struct vmw_private *dev_priv, uint32_t bytes)
 
 	}
 
+	down_write(&fifo_state->rwsem);
 	if (fifo_state->using_bounce_buffer || reserveable) {
 		next_cmd += bytes;
 		if (next_cmd >= max)
@@ -437,8 +439,9 @@ void vmw_fifo_commit(struct vmw_private *dev_priv, uint32_t bytes)
 	if (reserveable)
 		iowrite32(0, fifo_mem + SVGA_FIFO_RESERVED);
 	mb();
-	vmw_fifo_ping_host(dev_priv, SVGA_SYNC_GENERIC);
 	up_write(&fifo_state->rwsem);
+	vmw_fifo_ping_host(dev_priv, SVGA_SYNC_GENERIC);
+	mutex_unlock(&fifo_state->fifo_mutex);
 }
 
 int vmw_fifo_send_fence(struct vmw_private *dev_priv, uint32_t *sequence)
@@ -451,9 +454,7 @@ int vmw_fifo_send_fence(struct vmw_private *dev_priv, uint32_t *sequence)
 
 	fm = vmw_fifo_reserve(dev_priv, bytes);
 	if (unlikely(fm == NULL)) {
-		down_write(&fifo_state->rwsem);
-		*sequence = dev_priv->fence_seq;
-		up_write(&fifo_state->rwsem);
+		*sequence = atomic_read(&dev_priv->fence_seq);
 		ret = -ENOMEM;
 		(void)vmw_fallback_wait(dev_priv, false, true, *sequence,
 					false, 3*HZ);
@@ -461,7 +462,7 @@ int vmw_fifo_send_fence(struct vmw_private *dev_priv, uint32_t *sequence)
 	}
 
 	do {
-		*sequence = dev_priv->fence_seq++;
+		*sequence = atomic_add_return(1, &dev_priv->fence_seq);
 	} while (*sequence == 0);
 
 	if (!(fifo_state->capabilities & SVGA_FIFO_CAP_FENCE)) {
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
index 778851f9f1d6..1c7a316454d8 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
@@ -48,6 +48,12 @@ int vmw_getparam_ioctl(struct drm_device *dev, void *data,
 	case DRM_VMW_PARAM_FIFO_OFFSET:
 		param->value = dev_priv->mmio_start;
 		break;
+	case DRM_VMW_PARAM_HW_CAPS:
+		param->value = dev_priv->capabilities;
+		break;
+	case DRM_VMW_PARAM_FIFO_CAPS:
+		param->value = dev_priv->fifo.capabilities;
+		break;
 	default:
 		DRM_ERROR("Illegal vmwgfx get param request: %d\n",
 			  param->param);
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c b/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c
index d40086fc8647..4d7cb5393860 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_irq.c
@@ -84,20 +84,13 @@ bool vmw_fence_signaled(struct vmw_private *dev_priv,
 	    vmw_fifo_idle(dev_priv, sequence))
 		return true;
 
-	/**
-	 * Below is to signal stale fences that have wrapped.
-	 * First, block fence submission.
-	 */
-
-	down_read(&fifo_state->rwsem);
-
 	/**
 	 * Then check if the sequence is higher than what we've actually
 	 * emitted. Then the fence is stale and signaled.
 	 */
 
-	ret = ((dev_priv->fence_seq - sequence) > VMW_FENCE_WRAP);
-	up_read(&fifo_state->rwsem);
+	ret = ((atomic_read(&dev_priv->fence_seq) - sequence)
+	       > VMW_FENCE_WRAP);
 
 	return ret;
 }
@@ -127,7 +120,7 @@ int vmw_fallback_wait(struct vmw_private *dev_priv,
 
 	if (fifo_idle)
 		down_read(&fifo_state->rwsem);
-	signal_seq = dev_priv->fence_seq;
+	signal_seq = atomic_read(&dev_priv->fence_seq);
 	ret = 0;
 
 	for (;;) {
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
index eeba6d1d06e4..31f9afed0a63 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
@@ -769,10 +769,10 @@ int vmw_kms_init(struct vmw_private *dev_priv)
 
 	drm_mode_config_init(dev);
 	dev->mode_config.funcs = &vmw_kms_funcs;
-	dev->mode_config.min_width = 640;
-	dev->mode_config.min_height = 480;
-	dev->mode_config.max_width = 2048;
-	dev->mode_config.max_height = 2048;
+	dev->mode_config.min_width = 1;
+	dev->mode_config.min_height = 1;
+	dev->mode_config.max_width = dev_priv->fb_max_width;
+	dev->mode_config.max_height = dev_priv->fb_max_height;
 
 	ret = vmw_kms_init_legacy_display_system(dev_priv);
 
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
index c7efbd47ab84..f8fbbc67a406 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
@@ -35,11 +35,6 @@
 #define VMW_RES_SURFACE ttm_driver_type1
 #define VMW_RES_STREAM ttm_driver_type2
 
-/* XXX: This isn't a real hardware flag, but just a hack for kernel to
- * know about primary surfaces. Find a better way to accomplish this.
- */
-#define SVGA3D_SURFACE_HINT_SCANOUT (1 << 9)
-
 struct vmw_user_context {
 	struct ttm_base_object base;
 	struct vmw_resource res;
@@ -579,6 +574,7 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
 
 	srf->flags = req->flags;
 	srf->format = req->format;
+	srf->scanout = req->scanout;
 	memcpy(srf->mip_levels, req->mip_levels, sizeof(srf->mip_levels));
 	srf->num_sizes = 0;
 	for (i = 0; i < DRM_VMW_MAX_SURFACE_FACES; ++i)
@@ -604,16 +600,6 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
 	if (unlikely(ret != 0))
 		goto out_err1;
 
-	if (srf->flags & SVGA3D_SURFACE_HINT_SCANOUT) {
-		/* we should not send this flag down to hardware since
-		 * its not a official one
-		 */
-		srf->flags &= ~SVGA3D_SURFACE_HINT_SCANOUT;
-		srf->scanout = true;
-	} else {
-		srf->scanout = false;
-	}
-
 	if (srf->scanout &&
 	    srf->num_sizes == 1 &&
 	    srf->sizes[0].width == 64 &&
diff --git a/drivers/gpu/vga/vgaarb.c b/drivers/gpu/vga/vgaarb.c
index 1ac0c93603c9..24b56dc54597 100644
--- a/drivers/gpu/vga/vgaarb.c
+++ b/drivers/gpu/vga/vgaarb.c
@@ -961,7 +961,7 @@ static ssize_t vga_arb_write(struct file *file, const char __user * buf,
 		remaining -= 7;
 		pr_devel("client 0x%p called 'target'\n", priv);
 		/* if target is default */
-		if (!strncmp(buf, "default", 7))
+		if (!strncmp(kbuf, "default", 7))
 			pdev = pci_dev_get(vga_default_device());
 		else {
 			if (!vga_pci_str_to_vars(curr_pos, remaining,
diff --git a/include/drm/nouveau_drm.h b/include/drm/nouveau_drm.h
index 1e67c441ea82..f745948b61e4 100644
--- a/include/drm/nouveau_drm.h
+++ b/include/drm/nouveau_drm.h
@@ -77,6 +77,7 @@ struct drm_nouveau_gpuobj_free {
 #define NOUVEAU_GETPARAM_PCI_PHYSICAL    10
 #define NOUVEAU_GETPARAM_CHIPSET_ID      11
 #define NOUVEAU_GETPARAM_VM_VRAM_BASE    12
+#define NOUVEAU_GETPARAM_GRAPH_UNITS     13
 struct drm_nouveau_getparam {
 	uint64_t param;
 	uint64_t value;
diff --git a/include/drm/vmwgfx_drm.h b/include/drm/vmwgfx_drm.h
index 2be7e1249b6f..c7645f480d12 100644
--- a/include/drm/vmwgfx_drm.h
+++ b/include/drm/vmwgfx_drm.h
@@ -68,7 +68,8 @@
 #define DRM_VMW_PARAM_NUM_FREE_STREAMS 1
 #define DRM_VMW_PARAM_3D               2
 #define DRM_VMW_PARAM_FIFO_OFFSET      3
-
+#define DRM_VMW_PARAM_HW_CAPS          4
+#define DRM_VMW_PARAM_FIFO_CAPS        5
 
 /**
  * struct drm_vmw_getparam_arg
@@ -181,6 +182,8 @@ struct drm_vmw_context_arg {
  * The size of the array should equal the total number of mipmap levels.
  * @shareable: Boolean whether other clients (as identified by file descriptors)
  * may reference this surface.
+ * @scanout: Boolean whether the surface is intended to be used as a
+ * scanout.
  *
  * Input data to the DRM_VMW_CREATE_SURFACE Ioctl.
  * Output data from the DRM_VMW_REF_SURFACE Ioctl.
@@ -192,7 +195,7 @@ struct drm_vmw_surface_create_req {
 	uint32_t mip_levels[DRM_VMW_MAX_SURFACE_FACES];
 	uint64_t size_addr;
 	int32_t shareable;
-	uint32_t pad64;
+	int32_t scanout;
 };
 
 /**
@@ -295,17 +298,28 @@ union drm_vmw_surface_reference_arg {
  *
  * @commands: User-space address of a command buffer cast to an uint64_t.
  * @command-size: Size in bytes of the command buffer.
+ * @throttle-us: Sleep until software is less than @throttle_us
+ * microseconds ahead of hardware. The driver may round this value
+ * to the nearest kernel tick.
  * @fence_rep: User-space address of a struct drm_vmw_fence_rep cast to an
  * uint64_t.
+ * @version: Allows expanding the execbuf ioctl parameters without breaking
+ * backwards compatibility, since user-space will always tell the kernel
+ * which version it uses.
+ * @flags: Execbuf flags. None currently.
  *
  * Argument to the DRM_VMW_EXECBUF Ioctl.
  */
 
+#define DRM_VMW_EXECBUF_VERSION 0
+
 struct drm_vmw_execbuf_arg {
 	uint64_t commands;
 	uint32_t command_size;
-	uint32_t pad64;
+	uint32_t throttle_us;
 	uint64_t fence_rep;
+	 uint32_t version;
+	 uint32_t flags;
 };
 
 /**