Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs into for-2.6.34-incoming

Resolve merge conflict in fs/xfs/linux-2.6/xfs_export.c.

31 files changed, 622 insertions, 349 deletions

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 ba143972769f..0e9cd1d49130 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bios.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bios.c
@@ -310,63 +310,22 @@ valid_reg(struct nvbios *bios, uint32_t reg)
 	struct drm_device *dev = bios->dev;
 
 	/* C51 has misaligned regs on purpose. Marvellous */
-	if (reg & 0x2 || (reg & 0x1 && dev_priv->VBIOS.pub.chip_version != 0x51)) {
-		NV_ERROR(dev, "========== misaligned reg 0x%08X ==========\n",
-			 reg);
-		return 0;
-	}
-	/*
-	 * Warn on C51 regs that have not been verified accessible in
-	 * mmiotracing
-	 */
+	if (reg & 0x2 ||
+	    (reg & 0x1 && dev_priv->VBIOS.pub.chip_version != 0x51))
+		NV_ERROR(dev, "======= misaligned reg 0x%08X =======\n", reg);
+
+	/* warn on C51 regs that haven't been verified accessible in tracing */
 	if (reg & 0x1 && dev_priv->VBIOS.pub.chip_version == 0x51 &&
 	    reg != 0x130d && reg != 0x1311 && reg != 0x60081d)
 		NV_WARN(dev, "=== C51 misaligned reg 0x%08X not verified ===\n",
 			reg);
 
-	/* Trust the init scripts on G80 */
-	if (dev_priv->card_type >= NV_50)
-		return 1;
-
-	#define WITHIN(x, y, z) ((x >= y) && (x < y + z))
-	if (WITHIN(reg, NV_PMC_OFFSET, NV_PMC_SIZE))
-		return 1;
-	if (WITHIN(reg, NV_PBUS_OFFSET, NV_PBUS_SIZE))
-		return 1;
-	if (WITHIN(reg, NV_PFIFO_OFFSET, NV_PFIFO_SIZE))
-		return 1;
-	if (dev_priv->VBIOS.pub.chip_version >= 0x30 &&
-	    (WITHIN(reg, 0x4000, 0x600) || reg == 0x00004600))
-		return 1;
-	if (dev_priv->VBIOS.pub.chip_version >= 0x40 &&
-						WITHIN(reg, 0xc000, 0x48))
-		return 1;
-	if (dev_priv->VBIOS.pub.chip_version >= 0x17 && reg == 0x0000d204)
-		return 1;
-	if (dev_priv->VBIOS.pub.chip_version >= 0x40) {
-		if (reg == 0x00011014 || reg == 0x00020328)
-			return 1;
-		if (WITHIN(reg, 0x88000, NV_PBUS_SIZE)) /* new PBUS */
-			return 1;
+	if (reg >= (8*1024*1024)) {
+		NV_ERROR(dev, "=== reg 0x%08x out of mapped bounds ===\n", reg);
+		return 0;
 	}
-	if (WITHIN(reg, NV_PFB_OFFSET, NV_PFB_SIZE))
-		return 1;
-	if (WITHIN(reg, NV_PEXTDEV_OFFSET, NV_PEXTDEV_SIZE))
-		return 1;
-	if (WITHIN(reg, NV_PCRTC0_OFFSET, NV_PCRTC0_SIZE * 2))
-		return 1;
-	if (WITHIN(reg, NV_PRAMDAC0_OFFSET, NV_PRAMDAC0_SIZE * 2))
-		return 1;
-	if (dev_priv->VBIOS.pub.chip_version >= 0x17 && reg == 0x0070fff0)
-		return 1;
-	if (dev_priv->VBIOS.pub.chip_version == 0x51 &&
-				WITHIN(reg, NV_PRAMIN_OFFSET, NV_PRAMIN_SIZE))
-		return 1;
-	#undef WITHIN
-
-	NV_ERROR(dev, "========== unknown reg 0x%08X ==========\n", reg);
 
-	return 0;
+	return 1;
 }
 
 static bool
@@ -1906,7 +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;
 
 	/*
@@ -3196,16 +3155,25 @@ static int call_lvds_manufacturer_script(struct drm_device *dev, struct dcb_entr
 	}
 #ifdef __powerpc__
 	/* Powerbook specific quirks */
-	if (script == LVDS_RESET && ((dev->pci_device & 0xffff) == 0x0179 || (dev->pci_device & 0xffff) == 0x0329))
-		nv_write_tmds(dev, dcbent->or, 0, 0x02, 0x72);
-	if ((dev->pci_device & 0xffff) == 0x0179 || (dev->pci_device & 0xffff) == 0x0189 || (dev->pci_device & 0xffff) == 0x0329) {
-		if (script == LVDS_PANEL_ON) {
-			bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL, bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL) | (1 << 31));
-			bios_wr32(bios, NV_PCRTC_GPIO_EXT, bios_rd32(bios, NV_PCRTC_GPIO_EXT) | 1);
-		}
-		if (script == LVDS_PANEL_OFF) {
-			bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL, bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL) & ~(1 << 31));
-			bios_wr32(bios, NV_PCRTC_GPIO_EXT, bios_rd32(bios, NV_PCRTC_GPIO_EXT) & ~3);
+	if ((dev->pci_device & 0xffff) == 0x0179 ||
+	    (dev->pci_device & 0xffff) == 0x0189 ||
+	    (dev->pci_device & 0xffff) == 0x0329) {
+		if (script == LVDS_RESET) {
+			nv_write_tmds(dev, dcbent->or, 0, 0x02, 0x72);
+
+		} else if (script == LVDS_PANEL_ON) {
+			bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL,
+				  bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL)
+				  | (1 << 31));
+			bios_wr32(bios, NV_PCRTC_GPIO_EXT,
+				  bios_rd32(bios, NV_PCRTC_GPIO_EXT) | 1);
+
+		} else if (script == LVDS_PANEL_OFF) {
+			bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL,
+				  bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL)
+				  & ~(1 << 31));
+			bios_wr32(bios, NV_PCRTC_GPIO_EXT,
+				  bios_rd32(bios, NV_PCRTC_GPIO_EXT) & ~3);
 		}
 	}
 #endif
@@ -3797,7 +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;
@@ -3877,8 +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) {
@@ -3887,7 +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]);
@@ -3897,7 +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)
@@ -3910,7 +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]);
@@ -3922,7 +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]);
@@ -3934,7 +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;
@@ -5434,52 +5399,49 @@ static bool
 parse_dcb15_entry(struct drm_device *dev, struct parsed_dcb *dcb,
 		  uint32_t conn, uint32_t conf, struct dcb_entry *entry)
 {
-	if (conn != 0xf0003f00 && conn != 0xf2247f10 && conn != 0xf2204001 &&
-	    conn != 0xf2204301 && conn != 0xf2204311 && conn != 0xf2208001 &&
-	    conn != 0xf2244001 && conn != 0xf2244301 && conn != 0xf2244311 &&
-	    conn != 0xf4204011 && conn != 0xf4208011 && conn != 0xf4248011 &&
-	    conn != 0xf2045ff2 && conn != 0xf2045f14 && conn != 0xf207df14 &&
-	    conn != 0xf2205004 && conn != 0xf2209004) {
-		NV_ERROR(dev, "Unknown DCB 1.5 entry, please report\n");
-
-		/* cause output setting to fail for !TV, so message is seen */
-		if ((conn & 0xf) != 0x1)
-			dcb->entries = 0;
-
-		return false;
-	}
-	/* most of the below is a "best guess" atm */
-	entry->type = conn & 0xf;
-	if (entry->type == 2)
-		/* another way of specifying straps based lvds... */
+	switch (conn & 0x0000000f) {
+	case 0:
+		entry->type = OUTPUT_ANALOG;
+		break;
+	case 1:
+		entry->type = OUTPUT_TV;
+		break;
+	case 2:
+	case 3:
 		entry->type = OUTPUT_LVDS;
-	if (entry->type == 4) { /* digital */
-		if (conn & 0x10)
-			entry->type = OUTPUT_LVDS;
-		else
+		break;
+	case 4:
+		switch ((conn & 0x000000f0) >> 4) {
+		case 0:
 			entry->type = OUTPUT_TMDS;
+			break;
+		case 1:
+			entry->type = OUTPUT_LVDS;
+			break;
+		default:
+			NV_ERROR(dev, "Unknown DCB subtype 4/%d\n",
+				 (conn & 0x000000f0) >> 4);
+			return false;
+		}
+		break;
+	default:
+		NV_ERROR(dev, "Unknown DCB type %d\n", conn & 0x0000000f);
+		return false;
 	}
-	/* what's in bits 5-13? could be some encoder maker thing, in tv case */
-	entry->i2c_index = (conn >> 14) & 0xf;
-	/* raw heads field is in range 0-1, so move to 1-2 */
-	entry->heads = ((conn >> 18) & 0x7) + 1;
-	entry->location = (conn >> 21) & 0xf;
-	/* unused: entry->bus = (conn >> 25) & 0x7; */
-	/* set or to be same as heads -- hopefully safe enough */
-	entry->or = entry->heads;
+
+	entry->i2c_index = (conn & 0x0003c000) >> 14;
+	entry->heads = ((conn & 0x001c0000) >> 18) + 1;
+	entry->or = entry->heads; /* same as heads, hopefully safe enough */
+	entry->location = (conn & 0x01e00000) >> 21;
+	entry->bus = (conn & 0x0e000000) >> 25;
 	entry->duallink_possible = false;
 
 	switch (entry->type) {
 	case OUTPUT_ANALOG:
 		entry->crtconf.maxfreq = (conf & 0xffff) * 10;
 		break;
-	case OUTPUT_LVDS:
-		/*
-		 * This is probably buried in conn's unknown bits.
-		 * This will upset EDID-ful models, if they exist
-		 */
-		entry->lvdsconf.use_straps_for_mode = true;
-		entry->lvdsconf.use_power_scripts = true;
+	case OUTPUT_TV:
+		entry->tvconf.has_component_output = false;
 		break;
 	case OUTPUT_TMDS:
 		/*
@@ -5488,8 +5450,12 @@ parse_dcb15_entry(struct drm_device *dev, struct parsed_dcb *dcb,
 		 */
 		fabricate_vga_output(dcb, entry->i2c_index, entry->heads);
 		break;
-	case OUTPUT_TV:
-		entry->tvconf.has_component_output = false;
+	case OUTPUT_LVDS:
+		if ((conn & 0x00003f00) != 0x10)
+			entry->lvdsconf.use_straps_for_mode = true;
+		entry->lvdsconf.use_power_scripts = true;
+		break;
+	default:
 		break;
 	}
 
@@ -5564,11 +5530,13 @@ void merge_like_dcb_entries(struct drm_device *dev, struct parsed_dcb *dcb)
 	dcb->entries = newentries;
 }
 
-static int parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads)
+static int
+parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads)
 {
+	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct bios_parsed_dcb *bdcb = &bios->bdcb;
 	struct parsed_dcb *dcb;
-	uint16_t dcbptr, i2ctabptr = 0;
+	uint16_t dcbptr = 0, i2ctabptr = 0;
 	uint8_t *dcbtable;
 	uint8_t headerlen = 0x4, entries = DCB_MAX_NUM_ENTRIES;
 	bool configblock = true;
@@ -5579,16 +5547,18 @@ static int parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool two
 	dcb->entries = 0;
 
 	/* get the offset from 0x36 */
-	dcbptr = ROM16(bios->data[0x36]);
+	if (dev_priv->card_type > NV_04) {
+		dcbptr = ROM16(bios->data[0x36]);
+		if (dcbptr == 0x0000)
+			NV_WARN(dev, "No output data (DCB) found in BIOS\n");
+	}
 
+	/* this situation likely means a really old card, pre DCB */
 	if (dcbptr == 0x0) {
-		NV_WARN(dev, "No output data (DCB) found in BIOS, "
-			       "assuming a CRT output exists\n");
-		/* this situation likely means a really old card, pre DCB */
+		NV_INFO(dev, "Assuming a CRT output exists\n");
 		fabricate_vga_output(dcb, LEGACY_I2C_CRT, 1);
 
-		if (nv04_tv_identify(dev,
-				     bios->legacy.i2c_indices.tv) >= 0)
+		if (nv04_tv_identify(dev, bios->legacy.i2c_indices.tv) >= 0)
 			fabricate_tv_output(dcb, twoHeads);
 
 		return 0;
@@ -5892,9 +5862,11 @@ nouveau_bios_run_init_table(struct drm_device *dev, uint16_t table,
 	struct nvbios *bios = &dev_priv->VBIOS;
 	struct init_exec iexec = { true, false };
 
+	mutex_lock(&bios->lock);
 	bios->display.output = dcbent;
 	parse_init_table(bios, table, &iexec);
 	bios->display.output = NULL;
+	mutex_unlock(&bios->lock);
 }
 
 static bool NVInitVBIOS(struct drm_device *dev)
@@ -5903,6 +5875,7 @@ static bool NVInitVBIOS(struct drm_device *dev)
 	struct nvbios *bios = &dev_priv->VBIOS;
 
 	memset(bios, 0, sizeof(struct nvbios));
+	mutex_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..fd94bd6dc264 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;
 
+	struct mutex 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 e342a418d434..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;
@@ -469,6 +469,8 @@ nouveau_bo_move_accel_cleanup(struct nouveau_channel *chan,
 
 	ret = ttm_bo_move_accel_cleanup(&nvbo->bo, fence, NULL,
 					evict, no_wait, new_mem);
+	if (nvbo->channel && nvbo->channel != chan)
+		ret = nouveau_fence_wait(fence, NULL, false, false);
 	nouveau_fence_unref((void *)&fence);
 	return ret;
 }
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 5a10deb8bdbd..d2f63353ea97 100644
--- a/drivers/gpu/drm/nouveau/nouveau_connector.c
+++ b/drivers/gpu/drm/nouveau/nouveau_connector.c
@@ -24,9 +24,12 @@
  *
  */
 
+#include <acpi/button.h>
+
 #include "drmP.h"
 #include "drm_edid.h"
 #include "drm_crtc_helper.h"
+
 #include "nouveau_reg.h"
 #include "nouveau_drv.h"
 #include "nouveau_encoder.h"
@@ -83,14 +86,17 @@ nouveau_encoder_connector_get(struct nouveau_encoder *encoder)
 static void
 nouveau_connector_destroy(struct drm_connector *drm_connector)
 {
-	struct nouveau_connector *connector = nouveau_connector(drm_connector);
-	struct drm_device *dev = connector->base.dev;
+	struct nouveau_connector *nv_connector =
+		nouveau_connector(drm_connector);
+	struct drm_device *dev;
 
-	NV_DEBUG_KMS(dev, "\n");
-
-	if (!connector)
+	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);
 	kfree(drm_connector);
@@ -233,10 +239,21 @@ nouveau_connector_detect(struct drm_connector *connector)
 	if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)
 		nv_encoder = find_encoder_by_type(connector, OUTPUT_LVDS);
 	if (nv_encoder && nv_connector->native_mode) {
+#ifdef CONFIG_ACPI
+		if (!nouveau_ignorelid && !acpi_lid_open())
+			return connector_status_disconnected;
+#endif
 		nouveau_connector_set_encoder(connector, nv_encoder);
 		return connector_status_connected;
 	}
 
+	/* Cleanup the previous EDID block. */
+	if (nv_connector->edid) {
+		drm_mode_connector_update_edid_property(connector, NULL);
+		kfree(nv_connector->edid);
+		nv_connector->edid = NULL;
+	}
+
 	i2c = nouveau_connector_ddc_detect(connector, &nv_encoder);
 	if (i2c) {
 		nouveau_connector_ddc_prepare(connector, &flags);
@@ -247,7 +264,7 @@ nouveau_connector_detect(struct drm_connector *connector)
 		if (!nv_connector->edid) {
 			NV_ERROR(dev, "DDC responded, but no EDID for %s\n",
 				 drm_get_connector_name(connector));
-			return connector_status_disconnected;
+			goto detect_analog;
 		}
 
 		if (nv_encoder->dcb->type == OUTPUT_DP &&
@@ -281,6 +298,7 @@ nouveau_connector_detect(struct drm_connector *connector)
 		return connector_status_connected;
 	}
 
+detect_analog:
 	nv_encoder = find_encoder_by_type(connector, OUTPUT_ANALOG);
 	if (!nv_encoder)
 		nv_encoder = find_encoder_by_type(connector, OUTPUT_TV);
@@ -687,8 +705,12 @@ nouveau_connector_create_lvds(struct drm_device *dev,
 	 */
 	if (!nv_connector->edid && !nv_connector->native_mode &&
 	    !dev_priv->VBIOS.pub.fp_no_ddc) {
-		nv_connector->edid =
+		struct edid *edid =
 			(struct edid *)nouveau_bios_embedded_edid(dev);
+		if (edid) {
+			nv_connector->edid = kmalloc(EDID_LENGTH, GFP_KERNEL);
+			*(nv_connector->edid) = *edid;
+		}
 	}
 
 	if (!nv_connector->edid)
diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.c b/drivers/gpu/drm/nouveau/nouveau_dma.c
index 7afbe8b40d51..50d9e67745af 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dma.c
+++ b/drivers/gpu/drm/nouveau/nouveau_dma.c
@@ -126,47 +126,52 @@ OUT_RINGp(struct nouveau_channel *chan, const void *data, unsigned nr_dwords)
 	chan->dma.cur += nr_dwords;
 }
 
-static inline bool
-READ_GET(struct nouveau_channel *chan, uint32_t *get)
+/* Fetch and adjust GPU GET pointer
+ *
+ * Returns:
+ *  value >= 0, the adjusted GET pointer
+ *  -EINVAL if GET pointer currently outside main push buffer
+ *  -EBUSY if timeout exceeded
+ */
+static inline int
+READ_GET(struct nouveau_channel *chan, uint32_t *prev_get, uint32_t *timeout)
 {
 	uint32_t val;
 
 	val = nvchan_rd32(chan, chan->user_get);
-	if (val < chan->pushbuf_base ||
-	    val > chan->pushbuf_base + (chan->dma.max << 2)) {
-		/* meaningless to dma_wait() except to know whether the
-		 * GPU has stalled or not
-		 */
-		*get = val;
-		return false;
+
+	/* reset counter as long as GET is still advancing, this is
+	 * to avoid misdetecting a GPU lockup if the GPU happens to
+	 * just be processing an operation that takes a long time
+	 */
+	if (val != *prev_get) {
+		*prev_get = val;
+		*timeout = 0;
+	}
+
+	if ((++*timeout & 0xff) == 0) {
+		DRM_UDELAY(1);
+		if (*timeout > 100000)
+			return -EBUSY;
 	}
 
-	*get = (val - chan->pushbuf_base) >> 2;
-	return true;
+	if (val < chan->pushbuf_base ||
+	    val > chan->pushbuf_base + (chan->dma.max << 2))
+		return -EINVAL;
+
+	return (val - chan->pushbuf_base) >> 2;
 }
 
 int
 nouveau_dma_wait(struct nouveau_channel *chan, int size)
 {
-	uint32_t get, prev_get = 0, cnt = 0;
-	bool get_valid;
+	uint32_t prev_get = 0, cnt = 0;
+	int get;
 
 	while (chan->dma.free < size) {
-		/* reset counter as long as GET is still advancing, this is
-		 * to avoid misdetecting a GPU lockup if the GPU happens to
-		 * just be processing an operation that takes a long time
-		 */
-		get_valid = READ_GET(chan, &get);
-		if (get != prev_get) {
-			prev_get = get;
-			cnt = 0;
-		}
-
-		if ((++cnt & 0xff) == 0) {
-			DRM_UDELAY(1);
-			if (cnt > 100000)
-				return -EBUSY;
-		}
+		get = READ_GET(chan, &prev_get, &cnt);
+		if (unlikely(get == -EBUSY))
+			return -EBUSY;
 
 		/* loop until we have a usable GET pointer.  the value
 		 * we read from the GPU may be outside the main ring if
@@ -177,7 +182,7 @@ nouveau_dma_wait(struct nouveau_channel *chan, int size)
 		 * from the SKIPS area, so the code below doesn't have to deal
 		 * with some fun corner cases.
 		 */
-		if (!get_valid || get < NOUVEAU_DMA_SKIPS)
+		if (unlikely(get == -EINVAL) || get < NOUVEAU_DMA_SKIPS)
 			continue;
 
 		if (get <= chan->dma.cur) {
@@ -203,6 +208,19 @@ nouveau_dma_wait(struct nouveau_channel *chan, int size)
 			 * after processing the currently pending commands.
 			 */
 			OUT_RING(chan, chan->pushbuf_base | 0x20000000);
+
+			/* wait for GET to depart from the skips area.
+			 * prevents writing GET==PUT and causing a race
+			 * condition that causes us to think the GPU is
+			 * idle when it's not.
+			 */
+			do {
+				get = READ_GET(chan, &prev_get, &cnt);
+				if (unlikely(get == -EBUSY))
+					return -EBUSY;
+				if (unlikely(get == -EINVAL))
+					continue;
+			} while (get <= NOUVEAU_DMA_SKIPS);
 			WRITE_PUT(NOUVEAU_DMA_SKIPS);
 
 			/* we're now submitting commands at the start of
diff --git a/drivers/gpu/drm/nouveau/nouveau_dp.c b/drivers/gpu/drm/nouveau/nouveau_dp.c
index 9e2926c48579..f954ad93e81f 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dp.c
+++ b/drivers/gpu/drm/nouveau/nouveau_dp.c
@@ -490,7 +490,8 @@ nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr,
 		if (!nv_wait(NV50_AUXCH_CTRL(index), 0x00010000, 0x00000000)) {
 			NV_ERROR(dev, "expected bit 16 == 0, got 0x%08x\n",
 				 nv_rd32(dev, NV50_AUXCH_CTRL(index)));
-			return -EBUSY;
+			ret = -EBUSY;
+			goto out;
 		}
 
 		udelay(400);
@@ -502,6 +503,11 @@ nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr,
 	}
 
 	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 06eb993e0883..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)");
@@ -71,6 +71,18 @@ MODULE_PARM_DESC(uscript_tmds, "TMDS output script table ID (>=GeForce 8)");
 int nouveau_uscript_tmds = -1;
 module_param_named(uscript_tmds, nouveau_uscript_tmds, int, 0400);
 
+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 026419fe8791..1c15ef37b71c 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drv.h
+++ b/drivers/gpu/drm/nouveau/nouveau_drv.h
@@ -509,6 +509,8 @@ struct drm_nouveau_private {
 	void __iomem *ramin;
 	uint32_t ramin_size;
 
+	struct nouveau_bo *vga_ram;
+
 	struct workqueue_struct *wq;
 	struct work_struct irq_work;
 
@@ -581,6 +583,7 @@ struct drm_nouveau_private {
 	uint64_t vm_end;
 	struct nouveau_gpuobj *vm_vram_pt[NV50_VM_VRAM_NR];
 	int vm_vram_pt_nr;
+	uint64_t vram_sys_base;
 
 	/* the mtrr covering the FB */
 	int fb_mtrr;
@@ -675,6 +678,9 @@ extern char *nouveau_tv_norm;
 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 2009db2426c3..70cc30803e3b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_gem.c
+++ b/drivers/gpu/drm/nouveau/nouveau_gem.c
@@ -321,6 +321,7 @@ retry:
 		else {
 			NV_ERROR(dev, "invalid valid domains: 0x%08x\n",
 				 b->valid_domains);
+			list_add_tail(&nvbo->entry, &op->both_list);
 			validate_fini(op, NULL);
 			return -EINVAL;
 		}
@@ -466,13 +467,14 @@ u_memcpya(uint64_t user, unsigned nmemb, unsigned size)
 static int
 nouveau_gem_pushbuf_reloc_apply(struct nouveau_channel *chan, int nr_bo,
 				struct drm_nouveau_gem_pushbuf_bo *bo,
-				int nr_relocs, uint64_t ptr_relocs,
-				int nr_dwords, int first_dword,
+				unsigned nr_relocs, uint64_t ptr_relocs,
+				unsigned nr_dwords, unsigned first_dword,
 				uint32_t *pushbuf, bool is_iomem)
 {
 	struct drm_nouveau_gem_pushbuf_reloc *reloc = NULL;
 	struct drm_device *dev = chan->dev;
-	int ret = 0, i;
+	int ret = 0;
+	unsigned i;
 
 	reloc = u_memcpya(ptr_relocs, nr_relocs, sizeof(*reloc));
 	if (IS_ERR(reloc))
@@ -667,6 +669,18 @@ nouveau_gem_ioctl_pushbuf_call(struct drm_device *dev, void *data,
 	}
 	pbbo = nouveau_gem_object(gem);
 
+	if ((req->offset & 3) || req->nr_dwords < 2 ||
+	    (unsigned long)req->offset > (unsigned long)pbbo->bo.mem.size ||
+	    (unsigned long)req->nr_dwords >
+	     ((unsigned long)(pbbo->bo.mem.size - req->offset ) >> 2)) {
+		NV_ERROR(dev, "pb call misaligned or out of bounds: "
+			      "%d + %d * 4 > %ld\n",
+			 req->offset, req->nr_dwords, pbbo->bo.mem.size);
+		ret = -EINVAL;
+		drm_gem_object_unreference(gem);
+		goto out;
+	}
+
 	ret = ttm_bo_reserve(&pbbo->bo, false, false, true,
 			     chan->fence.sequence);
 	if (ret) {
@@ -911,7 +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 919a619ca7fa..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);
@@ -483,6 +497,13 @@ nouveau_pgraph_intr_error(struct drm_device *dev, uint32_t nsource)
 	if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) {
 		if (nouveau_pgraph_intr_swmthd(dev, &trap))
 			unhandled = 1;
+	} else if (nsource & NV03_PGRAPH_NSOURCE_DMA_VTX_PROTECTION) {
+		uint32_t v = nv_rd32(dev, 0x402000);
+		nv_wr32(dev, 0x402000, v);
+
+		/* dump the error anyway for now: it's useful for
+		   Gallium development */
+		unhandled = 1;
 	} else {
 		unhandled = 1;
 	}
@@ -559,86 +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;
-
-		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 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);
-	}
+		if (status & 0x00200000) {
+			int r;
+
+			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 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);
+		}
 
-	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_mem.c b/drivers/gpu/drm/nouveau/nouveau_mem.c
index fb9bdd6edf1f..2dc09dbd817d 100644
--- a/drivers/gpu/drm/nouveau/nouveau_mem.c
+++ b/drivers/gpu/drm/nouveau/nouveau_mem.c
@@ -285,53 +285,50 @@ nv50_mem_vm_bind_linear(struct drm_device *dev, uint64_t virt, uint32_t size,
 			uint32_t flags, uint64_t phys)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	struct nouveau_gpuobj **pgt;
-	unsigned psz, pfl, pages;
-
-	if (virt >= dev_priv->vm_gart_base &&
-	    (virt + size) < (dev_priv->vm_gart_base + dev_priv->vm_gart_size)) {
-		psz = 12;
-		pgt = &dev_priv->gart_info.sg_ctxdma;
-		pfl = 0x21;
-		virt -= dev_priv->vm_gart_base;
-	} else
-	if (virt >= dev_priv->vm_vram_base &&
-	    (virt + size) < (dev_priv->vm_vram_base + dev_priv->vm_vram_size)) {
-		psz = 16;
-		pgt = dev_priv->vm_vram_pt;
-		pfl = 0x01;
-		virt -= dev_priv->vm_vram_base;
-	} else {
-		NV_ERROR(dev, "Invalid address: 0x%16llx-0x%16llx\n",
-			 virt, virt + size - 1);
-		return -EINVAL;
-	}
+	struct nouveau_gpuobj *pgt;
+	unsigned block;
+	int i;
 
-	pages = size >> psz;
+	virt = ((virt - dev_priv->vm_vram_base) >> 16) << 1;
+	size = (size >> 16) << 1;
+
+	phys |= ((uint64_t)flags << 32);
+	phys |= 1;
+	if (dev_priv->vram_sys_base) {
+		phys += dev_priv->vram_sys_base;
+		phys |= 0x30;
+	}
 
 	dev_priv->engine.instmem.prepare_access(dev, true);
-	if (flags & 0x80000000) {
-		while (pages--) {
-			struct nouveau_gpuobj *pt = pgt[virt >> 29];
-			unsigned pte = ((virt & 0x1fffffffULL) >> psz) << 1;
+	while (size) {
+		unsigned offset_h = upper_32_bits(phys);
+		unsigned offset_l = lower_32_bits(phys);
+		unsigned pte, end;
+
+		for (i = 7; i >= 0; i--) {
+			block = 1 << (i + 1);
+			if (size >= block && !(virt & (block - 1)))
+				break;
+		}
+		offset_l |= (i << 7);
 
-			nv_wo32(dev, pt, pte++, 0x00000000);
-			nv_wo32(dev, pt, pte++, 0x00000000);
+		phys += block << 15;
+		size -= block;
 
-			virt += (1 << psz);
-		}
-	} else {
-		while (pages--) {
-			struct nouveau_gpuobj *pt = pgt[virt >> 29];
-			unsigned pte = ((virt & 0x1fffffffULL) >> psz) << 1;
-			unsigned offset_h = upper_32_bits(phys) & 0xff;
-			unsigned offset_l = lower_32_bits(phys);
+		while (block) {
+			pgt = dev_priv->vm_vram_pt[virt >> 14];
+			pte = virt & 0x3ffe;
 
-			nv_wo32(dev, pt, pte++, offset_l | pfl);
-			nv_wo32(dev, pt, pte++, offset_h | flags);
+			end = pte + block;
+			if (end > 16384)
+				end = 16384;
+			block -= (end - pte);
+			virt  += (end - pte);
 
-			phys += (1 << psz);
-			virt += (1 << psz);
+			while (pte < end) {
+				nv_wo32(dev, pgt, pte++, offset_l);
+				nv_wo32(dev, pgt, pte++, offset_h);
+			}
 		}
 	}
 	dev_priv->engine.instmem.finish_access(dev);
@@ -356,7 +353,41 @@ nv50_mem_vm_bind_linear(struct drm_device *dev, uint64_t virt, uint32_t size,
 void
 nv50_mem_vm_unbind(struct drm_device *dev, uint64_t virt, uint32_t size)
 {
-	nv50_mem_vm_bind_linear(dev, virt, size, 0x80000000, 0);
+	struct drm_nouveau_private *dev_priv = dev->dev_private;
+	struct nouveau_gpuobj *pgt;
+	unsigned pages, pte, end;
+
+	virt -= dev_priv->vm_vram_base;
+	pages = (size >> 16) << 1;
+
+	dev_priv->engine.instmem.prepare_access(dev, true);
+	while (pages) {
+		pgt = dev_priv->vm_vram_pt[virt >> 29];
+		pte = (virt & 0x1ffe0000ULL) >> 15;
+
+		end = pte + pages;
+		if (end > 16384)
+			end = 16384;
+		pages -= (end - pte);
+		virt  += (end - pte) << 15;
+
+		while (pte < end)
+			nv_wo32(dev, pgt, pte++, 0);
+	}
+	dev_priv->engine.instmem.finish_access(dev);
+
+	nv_wr32(dev, 0x100c80, 0x00050001);
+	if (!nv_wait(0x100c80, 0x00000001, 0x00000000)) {
+		NV_ERROR(dev, "timeout: (0x100c80 & 1) == 0 (2)\n");
+		NV_ERROR(dev, "0x100c80 = 0x%08x\n", nv_rd32(dev, 0x100c80));
+		return;
+	}
+
+	nv_wr32(dev, 0x100c80, 0x00000001);
+	if (!nv_wait(0x100c80, 0x00000001, 0x00000000)) {
+		NV_ERROR(dev, "timeout: (0x100c80 & 1) == 0 (2)\n");
+		NV_ERROR(dev, "0x100c80 = 0x%08x\n", nv_rd32(dev, 0x100c80));
+	}
 }
 
 /*
@@ -383,9 +414,8 @@ void nouveau_mem_close(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 
-	if (dev_priv->ttm.bdev.man[TTM_PL_PRIV0].has_type)
-		ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_PRIV0);
-	ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM);
+	nouveau_bo_unpin(dev_priv->vga_ram);
+	nouveau_bo_ref(NULL, &dev_priv->vga_ram);
 
 	ttm_bo_device_release(&dev_priv->ttm.bdev);
 
@@ -622,6 +652,15 @@ nouveau_mem_init(struct drm_device *dev)
 		return ret;
 	}
 
+	ret = nouveau_bo_new(dev, NULL, 256*1024, 0, TTM_PL_FLAG_VRAM,
+			     0, 0, true, true, &dev_priv->vga_ram);
+	if (ret == 0)
+		ret = nouveau_bo_pin(dev_priv->vga_ram, TTM_PL_FLAG_VRAM);
+	if (ret) {
+		NV_WARN(dev, "failed to reserve VGA memory\n");
+		nouveau_bo_ref(NULL, &dev_priv->vga_ram);
+	}
+
 	/* GART */
 #if !defined(__powerpc__) && !defined(__ia64__)
 	if (drm_device_is_agp(dev) && dev->agp) {
@@ -653,6 +692,7 @@ nouveau_mem_init(struct drm_device *dev)
 	dev_priv->fb_mtrr = drm_mtrr_add(drm_get_resource_start(dev, 1),
 					 drm_get_resource_len(dev, 1),
 					 DRM_MTRR_WC);
+
 	return 0;
 }
 
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 09b9a46dfc0e..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:
@@ -518,13 +532,16 @@ 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);
 
 		mutex_lock(&dev->struct_mutex);
+		ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM);
 		ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_TT);
 		mutex_unlock(&dev->struct_mutex);
 		nouveau_sgdma_takedown(dev);
@@ -816,6 +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_dac.c b/drivers/gpu/drm/nouveau/nv04_dac.c
index d0e038d28948..1d73b15d70da 100644
--- a/drivers/gpu/drm/nouveau/nv04_dac.c
+++ b/drivers/gpu/drm/nouveau/nv04_dac.c
@@ -119,7 +119,7 @@ static enum drm_connector_status nv04_dac_detect(struct drm_encoder *encoder,
 						 struct drm_connector *connector)
 {
 	struct drm_device *dev = encoder->dev;
-	uint8_t saved_seq1, saved_pi, saved_rpc1;
+	uint8_t saved_seq1, saved_pi, saved_rpc1, saved_cr_mode;
 	uint8_t saved_palette0[3], saved_palette_mask;
 	uint32_t saved_rtest_ctrl, saved_rgen_ctrl;
 	int i;
@@ -135,6 +135,9 @@ static enum drm_connector_status nv04_dac_detect(struct drm_encoder *encoder,
 		/* only implemented for head A for now */
 		NVSetOwner(dev, 0);
 
+	saved_cr_mode = NVReadVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX);
+	NVWriteVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX, saved_cr_mode | 0x80);
+
 	saved_seq1 = NVReadVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX);
 	NVWriteVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX, saved_seq1 & ~0x20);
 
@@ -203,6 +206,7 @@ out:
 	NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_PIXEL_INDEX, saved_pi);
 	NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_RPC1_INDEX, saved_rpc1);
 	NVWriteVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX, saved_seq1);
+	NVWriteVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX, saved_cr_mode);
 
 	if (blue == 0x18) {
 		NV_INFO(dev, "Load detected on head A\n");
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/nv04_instmem.c b/drivers/gpu/drm/nouveau/nv04_instmem.c
index a20c206625a2..a3b9563a6f60 100644
--- a/drivers/gpu/drm/nouveau/nv04_instmem.c
+++ b/drivers/gpu/drm/nouveau/nv04_instmem.c
@@ -30,7 +30,7 @@ nv04_instmem_determine_amount(struct drm_device *dev)
 		 * of vram.  For now, only reserve a small piece until we know
 		 * more about what each chipset requires.
 		 */
-		switch (dev_priv->chipset & 0xf0) {
+		switch (dev_priv->chipset) {
 		case 0x40:
 		case 0x47:
 		case 0x49:
diff --git a/drivers/gpu/drm/nouveau/nv17_tv.c b/drivers/gpu/drm/nouveau/nv17_tv.c
index 58b917c3341b..21ac6e49b6ee 100644
--- a/drivers/gpu/drm/nouveau/nv17_tv.c
+++ b/drivers/gpu/drm/nouveau/nv17_tv.c
@@ -579,6 +579,8 @@ static void nv17_tv_restore(struct drm_encoder *encoder)
 				nouveau_encoder(encoder)->restore.output);
 
 	nv17_tv_state_load(dev, &to_tv_enc(encoder)->saved_state);
+
+	nouveau_encoder(encoder)->last_dpms = NV_DPMS_CLEARED;
 }
 
 static int nv17_tv_create_resources(struct drm_encoder *encoder,
diff --git a/drivers/gpu/drm/nouveau/nv50_crtc.c b/drivers/gpu/drm/nouveau/nv50_crtc.c
index 118d3285fd8c..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);
@@ -432,6 +435,7 @@ nv50_crtc_prepare(struct drm_crtc *crtc)
 	struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc);
 	struct drm_device *dev = crtc->dev;
 	struct drm_encoder *encoder;
+	uint32_t dac = 0, sor = 0;
 
 	NV_DEBUG_KMS(dev, "index %d\n", nv_crtc->index);
 
@@ -439,9 +443,28 @@ nv50_crtc_prepare(struct drm_crtc *crtc)
 	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
 		struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
 
-		if (drm_helper_encoder_in_use(encoder))
+		if (!drm_helper_encoder_in_use(encoder))
 			continue;
 
+		if (nv_encoder->dcb->type == OUTPUT_ANALOG ||
+		    nv_encoder->dcb->type == OUTPUT_TV)
+			dac |= (1 << nv_encoder->or);
+		else
+			sor |= (1 << nv_encoder->or);
+	}
+
+	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+		struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
+
+		if (nv_encoder->dcb->type == OUTPUT_ANALOG ||
+		    nv_encoder->dcb->type == OUTPUT_TV) {
+			if (dac & (1 << nv_encoder->or))
+				continue;
+		} else {
+			if (sor & (1 << nv_encoder->or))
+				continue;
+		}
+
 		nv_encoder->disconnect(nv_encoder);
 	}
 
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 39caf167587d..204a79ff10f4 100644
--- a/drivers/gpu/drm/nouveau/nv50_fifo.c
+++ b/drivers/gpu/drm/nouveau/nv50_fifo.c
@@ -272,7 +272,7 @@ nv50_fifo_create_context(struct nouveau_channel *chan)
 			return ret;
 		ramfc = chan->ramfc->gpuobj;
 
-		ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, 4096, 256,
+		ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, 4096, 1024,
 					     0, &chan->cache);
 		if (ret)
 			return ret;
@@ -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 ca79f32be44c..6d504801b514 100644
--- a/drivers/gpu/drm/nouveau/nv50_graph.c
+++ b/drivers/gpu/drm/nouveau/nv50_graph.c
@@ -84,7 +84,7 @@ nv50_graph_init_regs__nv(struct drm_device *dev)
 	nv_wr32(dev, 0x400804, 0xc0000000);
 	nv_wr32(dev, 0x406800, 0xc0000000);
 	nv_wr32(dev, 0x400c04, 0xc0000000);
-	nv_wr32(dev, 0x401804, 0xc0000000);
+	nv_wr32(dev, 0x401800, 0xc0000000);
 	nv_wr32(dev, 0x405018, 0xc0000000);
 	nv_wr32(dev, 0x402000, 0xc0000000);
 
@@ -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;
@@ -275,19 +281,18 @@ 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))
 		return 0;
 	inst &= NV50_PGRAPH_CTXCTL_CUR_INSTANCE;
 
-	nv_wr32(dev, 0x400500, fifo & ~1);
+	nouveau_wait_for_idle(dev);
 	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_instmem.c b/drivers/gpu/drm/nouveau/nv50_instmem.c
index 94400f777e7f..f0dc4e36ef05 100644
--- a/drivers/gpu/drm/nouveau/nv50_instmem.c
+++ b/drivers/gpu/drm/nouveau/nv50_instmem.c
@@ -76,6 +76,11 @@ nv50_instmem_init(struct drm_device *dev)
 	for (i = 0x1700; i <= 0x1710; i += 4)
 		priv->save1700[(i-0x1700)/4] = nv_rd32(dev, i);
 
+	if (dev_priv->chipset == 0xaa || dev_priv->chipset == 0xac)
+		dev_priv->vram_sys_base = nv_rd32(dev, 0x100e10) << 12;
+	else
+		dev_priv->vram_sys_base = 0;
+
 	/* Reserve the last MiB of VRAM, we should probably try to avoid
 	 * setting up the below tables over the top of the VBIOS image at
 	 * some point.
@@ -172,16 +177,28 @@ nv50_instmem_init(struct drm_device *dev)
 	 * We map the entire fake channel into the start of the PRAMIN BAR
 	 */
 	ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, pt_size, 0x1000,
-							0, &priv->pramin_pt);
+				     0, &priv->pramin_pt);
 	if (ret)
 		return ret;
 
-	for (i = 0, v = c_offset; i < pt_size; i += 8, v += 0x1000) {
-		if (v < (c_offset + c_size))
-			BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, v | 1);
-		else
-			BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, 0x00000009);
+	v = c_offset | 1;
+	if (dev_priv->vram_sys_base) {
+		v += dev_priv->vram_sys_base;
+		v |= 0x30;
+	}
+
+	i = 0;
+	while (v < dev_priv->vram_sys_base + c_offset + c_size) {
+		BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, v);
+		BAR0_WI32(priv->pramin_pt->gpuobj, i + 4, 0x00000000);
+		v += 0x1000;
+		i += 8;
+	}
+
+	while (i < pt_size) {
+		BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, 0x00000000);
 		BAR0_WI32(priv->pramin_pt->gpuobj, i + 4, 0x00000000);
+		i += 8;
 	}
 
 	BAR0_WI32(chan->vm_pd, 0x00, priv->pramin_pt->instance | 0x63);
@@ -416,7 +433,9 @@ nv50_instmem_bind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nv50_instmem_priv *priv = dev_priv->engine.instmem.priv;
-	uint32_t pte, pte_end, vram;
+	struct nouveau_gpuobj *pramin_pt = priv->pramin_pt->gpuobj;
+	uint32_t pte, pte_end;
+	uint64_t vram;
 
 	if (!gpuobj->im_backing || !gpuobj->im_pramin || gpuobj->im_bound)
 		return -EINVAL;
@@ -424,20 +443,24 @@ nv50_instmem_bind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj)
 	NV_DEBUG(dev, "st=0x%0llx sz=0x%0llx\n",
 		 gpuobj->im_pramin->start, gpuobj->im_pramin->size);
 
-	pte     = (gpuobj->im_pramin->start >> 12) << 3;
-	pte_end = ((gpuobj->im_pramin->size >> 12) << 3) + pte;
+	pte     = (gpuobj->im_pramin->start >> 12) << 1;
+	pte_end = ((gpuobj->im_pramin->size >> 12) << 1) + pte;
 	vram    = gpuobj->im_backing_start;
 
 	NV_DEBUG(dev, "pramin=0x%llx, pte=%d, pte_end=%d\n",
 		 gpuobj->im_pramin->start, pte, pte_end);
 	NV_DEBUG(dev, "first vram page: 0x%08x\n", gpuobj->im_backing_start);
 
+	vram |= 1;
+	if (dev_priv->vram_sys_base) {
+		vram += dev_priv->vram_sys_base;
+		vram |= 0x30;
+	}
+
 	dev_priv->engine.instmem.prepare_access(dev, true);
 	while (pte < pte_end) {
-		nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 0)/4, vram | 1);
-		nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 4)/4, 0x00000000);
-
-		pte += 8;
+		nv_wo32(dev, pramin_pt, pte++, lower_32_bits(vram));
+		nv_wo32(dev, pramin_pt, pte++, upper_32_bits(vram));
 		vram += NV50_INSTMEM_PAGE_SIZE;
 	}
 	dev_priv->engine.instmem.finish_access(dev);
@@ -470,14 +493,13 @@ nv50_instmem_unbind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj)
 	if (gpuobj->im_bound == 0)
 		return -EINVAL;
 
-	pte     = (gpuobj->im_pramin->start >> 12) << 3;
-	pte_end = ((gpuobj->im_pramin->size >> 12) << 3) + pte;
+	pte     = (gpuobj->im_pramin->start >> 12) << 1;
+	pte_end = ((gpuobj->im_pramin->size >> 12) << 1) + pte;
 
 	dev_priv->engine.instmem.prepare_access(dev, true);
 	while (pte < pte_end) {
-		nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 0)/4, 0x00000009);
-		nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 4)/4, 0x00000000);
-		pte += 8;
+		nv_wo32(dev, priv->pramin_pt->gpuobj, pte++, 0x00000000);
+		nv_wo32(dev, priv->pramin_pt->gpuobj, pte++, 0x00000000);
 	}
 	dev_priv->engine.instmem.finish_access(dev);
 
diff --git a/drivers/gpu/drm/nouveau/nv50_sor.c b/drivers/gpu/drm/nouveau/nv50_sor.c
index e395c16d30f5..c2fff543b06f 100644
--- a/drivers/gpu/drm/nouveau/nv50_sor.c
+++ b/drivers/gpu/drm/nouveau/nv50_sor.c
@@ -90,11 +90,25 @@ nv50_sor_dpms(struct drm_encoder *encoder, int mode)
 {
 	struct drm_device *dev = encoder->dev;
 	struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
+	struct drm_encoder *enc;
 	uint32_t val;
 	int or = nv_encoder->or;
 
 	NV_DEBUG_KMS(dev, "or %d mode %d\n", or, mode);
 
+	nv_encoder->last_dpms = mode;
+	list_for_each_entry(enc, &dev->mode_config.encoder_list, head) {
+		struct nouveau_encoder *nvenc = nouveau_encoder(enc);
+
+		if (nvenc == nv_encoder ||
+		    nvenc->disconnect != nv50_sor_disconnect ||
+		    nvenc->dcb->or != nv_encoder->dcb->or)
+			continue;
+
+		if (nvenc->last_dpms == DRM_MODE_DPMS_ON)
+			return;
+	}
+
 	/* wait for it to be done */
 	if (!nv_wait(NV50_PDISPLAY_SOR_DPMS_CTRL(or),
 		     NV50_PDISPLAY_SOR_DPMS_CTRL_PENDING, 0)) {