Merge remote branch 'nouveau/for-airlied' into drm-next-stage

* nouveau/for-airlied: (25 commits)
  drm/nouveau: use ALIGN instead of open coding it
  drm/nouveau: report unknown connector state if lid closed
  drm/nouveau: support version 0x20 displayport tables
  drm/nouveau: Fix noaccel/nofbaccel option descriptions.
  drm/nv50: Implement ctxprog/state generation.
  drm/nouveau: use dcb connector types throughout the driver
  drm/nv50: enable hpd on any connector we know the gpio line for
  drm/nouveau: use dcb connector table for creating drm connectors
  drm/nouveau: construct a connector table for cards that lack a real one
  drm/nouveau: check for known dcb connector types
  drm/nouveau: parse dcb gpio/connector tables after encoders
  drm/nouveau: reorganise bios header, add dcb connector type enums
  drm/nouveau: merge nvbios and nouveau_bios_info
  drm/nouveau: merge parsed_dcb and bios_parsed_dcb into dcb_table
  drm/nouveau: rename parsed_dcb_gpio to dcb_gpio_table
  drm/nouveau: allow retrieval of vbios image from debugfs
  drm/nouveau: fix missing spin_unlock in failure path
  drm/nouveau: fix i2ctable bounds checking
  drm/nouveau: fix nouveau_i2c_find bounds checking
  drm/nouveau: fix pramdac_table range checking
  ...

Conflicts:
	drivers/gpu/drm/nouveau/nouveau_gem.c

drivers/gpu/drm/nouveau/Makefile

@@ -16,7 +16,7 @@ nouveau-y := nouveau_drv.o nouveau_state.o nouveau_channel.o nouveau_mem.o \
              nv04_fifo.o nv10_fifo.o nv40_fifo.o nv50_fifo.o \
              nv04_graph.o nv10_graph.o nv20_graph.o \
              nv40_graph.o nv50_graph.o \
-             nv40_grctx.o \
+             nv40_grctx.o nv50_grctx.o \
              nv04_instmem.o nv50_instmem.o \
              nv50_crtc.o nv50_dac.o nv50_sor.o \
              nv50_cursor.o nv50_display.o nv50_fbcon.o \

drivers/gpu/drm/nouveau/nouveau_bios.h

@@ -34,9 +34,67 @@
 
 #define DCB_LOC_ON_CHIP 0
 
+struct dcb_i2c_entry {
+	uint8_t port_type;
+	uint8_t read, write;
+	struct nouveau_i2c_chan *chan;
+};
+
+enum dcb_gpio_tag {
+	DCB_GPIO_TVDAC0 = 0xc,
+	DCB_GPIO_TVDAC1 = 0x2d,
+};
+
+struct dcb_gpio_entry {
+	enum dcb_gpio_tag tag;
+	int line;
+	bool invert;
+};
+
+struct dcb_gpio_table {
+	int entries;
+	struct dcb_gpio_entry entry[DCB_MAX_NUM_GPIO_ENTRIES];
+};
+
+enum dcb_connector_type {
+	DCB_CONNECTOR_VGA = 0x00,
+	DCB_CONNECTOR_TV_0 = 0x10,
+	DCB_CONNECTOR_TV_1 = 0x11,
+	DCB_CONNECTOR_TV_3 = 0x13,
+	DCB_CONNECTOR_DVI_I = 0x30,
+	DCB_CONNECTOR_DVI_D = 0x31,
+	DCB_CONNECTOR_LVDS = 0x40,
+	DCB_CONNECTOR_DP = 0x46,
+	DCB_CONNECTOR_eDP = 0x47,
+	DCB_CONNECTOR_HDMI_0 = 0x60,
+	DCB_CONNECTOR_HDMI_1 = 0x61,
+	DCB_CONNECTOR_NONE = 0xff
+};
+
+struct dcb_connector_table_entry {
+	uint32_t entry;
+	enum dcb_connector_type type;
+	uint8_t index;
+	uint8_t gpio_tag;
+};
+
+struct dcb_connector_table {
+	int entries;
+	struct dcb_connector_table_entry entry[DCB_MAX_NUM_CONNECTOR_ENTRIES];
+};
+
+enum dcb_type {
+	OUTPUT_ANALOG = 0,
+	OUTPUT_TV = 1,
+	OUTPUT_TMDS = 2,
+	OUTPUT_LVDS = 3,
+	OUTPUT_DP = 6,
+	OUTPUT_ANY = -1
+};
+
 struct dcb_entry {
 	int index;	/* may not be raw dcb index if merging has happened */
-	uint8_t type;
+	enum dcb_type type;
 	uint8_t i2c_index;
 	uint8_t heads;
 	uint8_t connector;
@@ -71,69 +129,22 @@ struct dcb_entry {
 	bool i2c_upper_default;
 };
 
-struct dcb_i2c_entry {
-	uint8_t port_type;
-	uint8_t read, write;
-	struct nouveau_i2c_chan *chan;
-};
+struct dcb_table {
+	uint8_t version;
 
-struct parsed_dcb {
 	int entries;
 	struct dcb_entry entry[DCB_MAX_NUM_ENTRIES];
-	struct dcb_i2c_entry i2c[DCB_MAX_NUM_I2C_ENTRIES];
-};
-
-enum dcb_gpio_tag {
-	DCB_GPIO_TVDAC0 = 0xc,
-	DCB_GPIO_TVDAC1 = 0x2d,
-};
-
-struct dcb_gpio_entry {
-	enum dcb_gpio_tag tag;
-	int line;
-	bool invert;
-};
-
-struct parsed_dcb_gpio {
-	int entries;
-	struct dcb_gpio_entry entry[DCB_MAX_NUM_GPIO_ENTRIES];
-};
-
-struct dcb_connector_table_entry {
-	uint32_t entry;
-	uint8_t type;
-	uint8_t index;
-	uint8_t gpio_tag;
-};
-
-struct dcb_connector_table {
-	int entries;
-	struct dcb_connector_table_entry entry[DCB_MAX_NUM_CONNECTOR_ENTRIES];
-};
-
-struct bios_parsed_dcb {
-	uint8_t version;
-
-	struct parsed_dcb dcb;
 
 	uint8_t *i2c_table;
 	uint8_t i2c_default_indices;
+	struct dcb_i2c_entry i2c[DCB_MAX_NUM_I2C_ENTRIES];
 
 	uint16_t gpio_table_ptr;
-	struct parsed_dcb_gpio gpio;
+	struct dcb_gpio_table gpio;
 	uint16_t connector_table_ptr;
 	struct dcb_connector_table connector;
 };
 
-enum nouveau_encoder_type {
-	OUTPUT_ANALOG = 0,
-	OUTPUT_TV = 1,
-	OUTPUT_TMDS = 2,
-	OUTPUT_LVDS = 3,
-	OUTPUT_DP = 6,
-	OUTPUT_ANY = -1
-};
-
 enum nouveau_or {
 	OUTPUT_A = (1 << 0),
 	OUTPUT_B = (1 << 1),
@@ -190,8 +201,8 @@ struct pll_lims {
 	int refclk;
 };
 
-struct nouveau_bios_info {
-	struct parsed_dcb *dcb;
+struct nvbios {
+	struct drm_device *dev;
 
 	uint8_t chip_version;
 
@@ -199,11 +210,6 @@ struct nouveau_bios_info {
 	uint32_t tvdactestval;
 	uint8_t digital_min_front_porch;
 	bool fp_no_ddc;
-};
-
-struct nvbios {
-	struct drm_device *dev;
-	struct nouveau_bios_info pub;
 
 	struct mutex lock;
 
@@ -234,7 +240,7 @@ struct nvbios {
 	uint16_t some_script_ptr; /* BIT I + 14 */
 	uint16_t init96_tbl_ptr; /* BIT I + 16 */
 
-	struct bios_parsed_dcb bdcb;
+	struct dcb_table dcb;
 
 	struct {
 		int crtchead;

drivers/gpu/drm/nouveau/nouveau_calc.c

@@ -274,7 +274,7 @@ getMNP_single(struct drm_device *dev, struct pll_lims *pll_lim, int clk,
 	 * returns calculated clock
 	 */
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	int cv = dev_priv->vbios->chip_version;
+	int cv = dev_priv->vbios.chip_version;
 	int minvco = pll_lim->vco1.minfreq, maxvco = pll_lim->vco1.maxfreq;
 	int minM = pll_lim->vco1.min_m, maxM = pll_lim->vco1.max_m;
 	int minN = pll_lim->vco1.min_n, maxN = pll_lim->vco1.max_n;
@@ -373,7 +373,7 @@ getMNP_double(struct drm_device *dev, struct pll_lims *pll_lim, int clk,
 	 * returns calculated clock
 	 */
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	int chip_version = dev_priv->vbios->chip_version;
+	int chip_version = dev_priv->vbios.chip_version;
 	int minvco1 = pll_lim->vco1.minfreq, maxvco1 = pll_lim->vco1.maxfreq;
 	int minvco2 = pll_lim->vco2.minfreq, maxvco2 = pll_lim->vco2.maxfreq;
 	int minU1 = pll_lim->vco1.min_inputfreq, minU2 = pll_lim->vco2.min_inputfreq;

drivers/gpu/drm/nouveau/nouveau_channel.c

@@ -35,22 +35,27 @@ nouveau_channel_pushbuf_ctxdma_init(struct nouveau_channel *chan)
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_bo *pb = chan->pushbuf_bo;
 	struct nouveau_gpuobj *pushbuf = NULL;
-	uint32_t start = pb->bo.mem.mm_node->start << PAGE_SHIFT;
 	int ret;
 
+	if (dev_priv->card_type >= NV_50) {
+		ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, 0,
+					     dev_priv->vm_end, NV_DMA_ACCESS_RO,
+					     NV_DMA_TARGET_AGP, &pushbuf);
+		chan->pushbuf_base = pb->bo.offset;
+	} else
 	if (pb->bo.mem.mem_type == TTM_PL_TT) {
 		ret = nouveau_gpuobj_gart_dma_new(chan, 0,
 						  dev_priv->gart_info.aper_size,
 						  NV_DMA_ACCESS_RO, &pushbuf,
 						  NULL);
-		chan->pushbuf_base = start;
+		chan->pushbuf_base = pb->bo.mem.mm_node->start << PAGE_SHIFT;
 	} else
 	if (dev_priv->card_type != NV_04) {
 		ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, 0,
 					     dev_priv->fb_available_size,
 					     NV_DMA_ACCESS_RO,
 					     NV_DMA_TARGET_VIDMEM, &pushbuf);
-		chan->pushbuf_base = start;
+		chan->pushbuf_base = pb->bo.mem.mm_node->start << PAGE_SHIFT;
 	} else {
 		/* NV04 cmdbuf hack, from original ddx.. not sure of it's
 		 * exact reason for existing :)  PCI access to cmdbuf in
@@ -61,7 +66,7 @@ nouveau_channel_pushbuf_ctxdma_init(struct nouveau_channel *chan)
 					     dev_priv->fb_available_size,
 					     NV_DMA_ACCESS_RO,
 					     NV_DMA_TARGET_PCI, &pushbuf);
-		chan->pushbuf_base = start;
+		chan->pushbuf_base = pb->bo.mem.mm_node->start << PAGE_SHIFT;
 	}
 
 	ret = nouveau_gpuobj_ref_add(dev, chan, 0, pushbuf, &chan->pushbuf);
@@ -275,9 +280,18 @@ nouveau_channel_free(struct nouveau_channel *chan)
 	 */
 	nouveau_fence_fini(chan);
 
-	/* Ensure the channel is no longer active on the GPU */
+	/* This will prevent pfifo from switching channels. */
 	pfifo->reassign(dev, false);
 
+	/* We want to give pgraph a chance to idle and get rid of all potential
+	 * errors. We need to do this before the lock, otherwise the irq handler
+	 * is unable to process them.
+	 */
+	if (pgraph->channel(dev) == chan)
+		nouveau_wait_for_idle(dev);
+
+	spin_lock_irqsave(&dev_priv->context_switch_lock, flags);
+
 	pgraph->fifo_access(dev, false);
 	if (pgraph->channel(dev) == chan)
 		pgraph->unload_context(dev);
@@ -293,6 +307,8 @@ nouveau_channel_free(struct nouveau_channel *chan)
 
 	pfifo->reassign(dev, true);
 
+	spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags);
+
 	/* Release the channel's resources */
 	nouveau_gpuobj_ref_del(dev, &chan->pushbuf);
 	if (chan->pushbuf_bo) {
@@ -369,6 +385,14 @@ nouveau_ioctl_fifo_alloc(struct drm_device *dev, void *data,
 		return ret;
 	init->channel  = chan->id;
 
+	if (chan->dma.ib_max)
+		init->pushbuf_domains = NOUVEAU_GEM_DOMAIN_VRAM |
+					NOUVEAU_GEM_DOMAIN_GART;
+	else if (chan->pushbuf_bo->bo.mem.mem_type == TTM_PL_VRAM)
+		init->pushbuf_domains = NOUVEAU_GEM_DOMAIN_VRAM;
+	else
+		init->pushbuf_domains = NOUVEAU_GEM_DOMAIN_GART;
+
 	init->subchan[0].handle = NvM2MF;
 	if (dev_priv->card_type < NV_50)
 		init->subchan[0].grclass = 0x0039;
@@ -408,7 +432,6 @@ nouveau_ioctl_fifo_free(struct drm_device *dev, void *data,
  ***********************************/
 
 struct drm_ioctl_desc nouveau_ioctls[] = {
-	DRM_IOCTL_DEF(DRM_NOUVEAU_CARD_INIT, nouveau_ioctl_card_init, DRM_AUTH),
 	DRM_IOCTL_DEF(DRM_NOUVEAU_GETPARAM, nouveau_ioctl_getparam, DRM_AUTH),
 	DRM_IOCTL_DEF(DRM_NOUVEAU_SETPARAM, nouveau_ioctl_setparam, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
 	DRM_IOCTL_DEF(DRM_NOUVEAU_CHANNEL_ALLOC, nouveau_ioctl_fifo_alloc, DRM_AUTH),
@@ -418,13 +441,9 @@ struct drm_ioctl_desc nouveau_ioctls[] = {
 	DRM_IOCTL_DEF(DRM_NOUVEAU_GPUOBJ_FREE, nouveau_ioctl_gpuobj_free, DRM_AUTH),
 	DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_NEW, nouveau_gem_ioctl_new, DRM_AUTH),
 	DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_PUSHBUF, nouveau_gem_ioctl_pushbuf, DRM_AUTH),
-	DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_PUSHBUF_CALL, nouveau_gem_ioctl_pushbuf_call, DRM_AUTH),
-	DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_PIN, nouveau_gem_ioctl_pin, DRM_AUTH),
-	DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_UNPIN, nouveau_gem_ioctl_unpin, DRM_AUTH),
 	DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_CPU_PREP, nouveau_gem_ioctl_cpu_prep, DRM_AUTH),
 	DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_CPU_FINI, nouveau_gem_ioctl_cpu_fini, DRM_AUTH),
 	DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_INFO, nouveau_gem_ioctl_info, DRM_AUTH),
-	DRM_IOCTL_DEF(DRM_NOUVEAU_GEM_PUSHBUF_CALL2, nouveau_gem_ioctl_pushbuf_call2, DRM_AUTH),
 };
 
 int nouveau_max_ioctl = DRM_ARRAY_SIZE(nouveau_ioctls);

drivers/gpu/drm/nouveau/nouveau_connector.h

@@ -49,6 +49,7 @@ static inline struct nouveau_connector *nouveau_connector(
 	return container_of(con, struct nouveau_connector, base);
 }
 
-int nouveau_connector_create(struct drm_device *dev, int i2c_index, int type);
+int nouveau_connector_create(struct drm_device *,
+			     struct dcb_connector_table_entry *);
 
 #endif /* __NOUVEAU_CONNECTOR_H__ */

drivers/gpu/drm/nouveau/nouveau_debugfs.c

@@ -47,12 +47,23 @@ nouveau_debugfs_channel_info(struct seq_file *m, void *data)
 	seq_printf(m, "           cur: 0x%08x\n", chan->dma.cur << 2);
 	seq_printf(m, "           put: 0x%08x\n", chan->dma.put << 2);
 	seq_printf(m, "          free: 0x%08x\n", chan->dma.free << 2);
+	if (chan->dma.ib_max) {
+		seq_printf(m, "        ib max: 0x%08x\n", chan->dma.ib_max);
+		seq_printf(m, "        ib put: 0x%08x\n", chan->dma.ib_put);
+		seq_printf(m, "       ib free: 0x%08x\n", chan->dma.ib_free);
+	}
 
 	seq_printf(m, "gpu fifo state:\n");
 	seq_printf(m, "           get: 0x%08x\n",
 					nvchan_rd32(chan, chan->user_get));
 	seq_printf(m, "           put: 0x%08x\n",
 					nvchan_rd32(chan, chan->user_put));
+	if (chan->dma.ib_max) {
+		seq_printf(m, "        ib get: 0x%08x\n",
+			   nvchan_rd32(chan, 0x88));
+		seq_printf(m, "        ib put: 0x%08x\n",
+			   nvchan_rd32(chan, 0x8c));
+	}
 
 	seq_printf(m, "last fence    : %d\n", chan->fence.sequence);
 	seq_printf(m, "last signalled: %d\n", chan->fence.sequence_ack);
@@ -133,9 +144,22 @@ nouveau_debugfs_memory_info(struct seq_file *m, void *data)
 	return 0;
 }
 
+static int
+nouveau_debugfs_vbios_image(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_nouveau_private *dev_priv = node->minor->dev->dev_private;
+	int i;
+
+	for (i = 0; i < dev_priv->vbios.length; i++)
+		seq_printf(m, "%c", dev_priv->vbios.data[i]);
+	return 0;
+}
+
 static struct drm_info_list nouveau_debugfs_list[] = {
 	{ "chipset", nouveau_debugfs_chipset_info, 0, NULL },
 	{ "memory", nouveau_debugfs_memory_info, 0, NULL },
+	{ "vbios.rom", nouveau_debugfs_vbios_image, 0, NULL },
 };
 #define NOUVEAU_DEBUGFS_ENTRIES ARRAY_SIZE(nouveau_debugfs_list)
 

drivers/gpu/drm/nouveau/nouveau_dma.c

@@ -32,7 +32,22 @@
 void
 nouveau_dma_pre_init(struct nouveau_channel *chan)
 {
-	chan->dma.max  = (chan->pushbuf_bo->bo.mem.size >> 2) - 2;
+	struct drm_nouveau_private *dev_priv = chan->dev->dev_private;
+	struct nouveau_bo *pushbuf = chan->pushbuf_bo;
+
+	if (dev_priv->card_type == NV_50) {
+		const int ib_size = pushbuf->bo.mem.size / 2;
+
+		chan->dma.ib_base = (pushbuf->bo.mem.size - ib_size) >> 2;
+		chan->dma.ib_max = (ib_size / 8) - 1;
+		chan->dma.ib_put = 0;
+		chan->dma.ib_free = chan->dma.ib_max - chan->dma.ib_put;
+
+		chan->dma.max = (pushbuf->bo.mem.size - ib_size) >> 2;
+	} else {
+		chan->dma.max  = (pushbuf->bo.mem.size >> 2) - 2;
+	}
+
 	chan->dma.put  = 0;
 	chan->dma.cur  = chan->dma.put;
 	chan->dma.free = chan->dma.max - chan->dma.cur;
@@ -162,12 +177,101 @@ READ_GET(struct nouveau_channel *chan, uint32_t *prev_get, uint32_t *timeout)
 	return (val - chan->pushbuf_base) >> 2;
 }
 
+void
+nv50_dma_push(struct nouveau_channel *chan, struct nouveau_bo *bo,
+	      int delta, int length)
+{
+	struct nouveau_bo *pb = chan->pushbuf_bo;
+	uint64_t offset = bo->bo.offset + delta;
+	int ip = (chan->dma.ib_put * 2) + chan->dma.ib_base;
+
+	BUG_ON(chan->dma.ib_free < 1);
+	nouveau_bo_wr32(pb, ip++, lower_32_bits(offset));
+	nouveau_bo_wr32(pb, ip++, upper_32_bits(offset) | length << 8);
+
+	chan->dma.ib_put = (chan->dma.ib_put + 1) & chan->dma.ib_max;
+	nvchan_wr32(chan, 0x8c, chan->dma.ib_put);
+	chan->dma.ib_free--;
+}
+
+static int
+nv50_dma_push_wait(struct nouveau_channel *chan, int count)
+{
+	uint32_t cnt = 0, prev_get = 0;
+
+	while (chan->dma.ib_free < count) {
+		uint32_t get = nvchan_rd32(chan, 0x88);
+		if (get != prev_get) {
+			prev_get = get;
+			cnt = 0;
+		}
+
+		if ((++cnt & 0xff) == 0) {
+			DRM_UDELAY(1);
+			if (cnt > 100000)
+				return -EBUSY;
+		}
+
+		chan->dma.ib_free = get - chan->dma.ib_put;
+		if (chan->dma.ib_free <= 0)
+			chan->dma.ib_free += chan->dma.ib_max + 1;
+	}
+
+	return 0;
+}
+
+static int
+nv50_dma_wait(struct nouveau_channel *chan, int slots, int count)
+{
+	uint32_t cnt = 0, prev_get = 0;
+	int ret;
+
+	ret = nv50_dma_push_wait(chan, slots + 1);
+	if (unlikely(ret))
+		return ret;
+
+	while (chan->dma.free < count) {
+		int get = READ_GET(chan, &prev_get, &cnt);
+		if (unlikely(get < 0)) {
+			if (get == -EINVAL)
+				continue;
+
+			return get;
+		}
+
+		if (get <= chan->dma.cur) {
+			chan->dma.free = chan->dma.max - chan->dma.cur;
+			if (chan->dma.free >= count)
+				break;
+
+			FIRE_RING(chan);
+			do {
+				get = READ_GET(chan, &prev_get, &cnt);
+				if (unlikely(get < 0)) {
+					if (get == -EINVAL)
+						continue;
+					return get;
+				}
+			} while (get == 0);
+			chan->dma.cur = 0;
+			chan->dma.put = 0;
+		}
+
+		chan->dma.free = get - chan->dma.cur - 1;
+	}
+
+	return 0;
+}
+
 int
-nouveau_dma_wait(struct nouveau_channel *chan, int size)
+nouveau_dma_wait(struct nouveau_channel *chan, int slots, int size)
 {
 	uint32_t prev_get = 0, cnt = 0;
 	int get;
 
+	if (chan->dma.ib_max)
+		return nv50_dma_wait(chan, slots, size);
+
 	while (chan->dma.free < size) {
 		get = READ_GET(chan, &prev_get, &cnt);
 		if (unlikely(get == -EBUSY))

drivers/gpu/drm/nouveau/nouveau_dma.h

@@ -31,6 +31,9 @@
 #define NOUVEAU_DMA_DEBUG 0
 #endif
 
+void nv50_dma_push(struct nouveau_channel *, struct nouveau_bo *,
+		   int delta, int length);
+
 /*
  * There's a hw race condition where you can't jump to your PUT offset,
  * to avoid this we jump to offset + SKIPS and fill the difference with
@@ -96,13 +99,11 @@ enum {
 static __must_check inline int
 RING_SPACE(struct nouveau_channel *chan, int size)
 {
-	if (chan->dma.free < size) {
-		int ret;
+	int ret;
 
-		ret = nouveau_dma_wait(chan, size);
-		if (ret)
-			return ret;
-	}
+	ret = nouveau_dma_wait(chan, 1, size);
+	if (ret)
+		return ret;
 
 	chan->dma.free -= size;
 	return 0;
@@ -146,7 +147,13 @@ FIRE_RING(struct nouveau_channel *chan)
 		return;
 	chan->accel_done = true;
 
-	WRITE_PUT(chan->dma.cur);
+	if (chan->dma.ib_max) {
+		nv50_dma_push(chan, chan->pushbuf_bo, chan->dma.put << 2,
+			      (chan->dma.cur - chan->dma.put) << 2);
+	} else {
+		WRITE_PUT(chan->dma.cur);
+	}
+
 	chan->dma.put = chan->dma.cur;
 }
 

drivers/gpu/drm/nouveau/nouveau_drv.c

@@ -75,11 +75,11 @@ 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");
+MODULE_PARM_DESC(noaccel, "Disable all acceleration");
 int nouveau_noaccel = 0;
 module_param_named(noaccel, nouveau_noaccel, int, 0400);
 
-MODULE_PARM_DESC(noagp, "Disable fbcon acceleration");
+MODULE_PARM_DESC(nofbaccel, "Disable fbcon acceleration");
 int nouveau_nofbaccel = 0;
 module_param_named(nofbaccel, nouveau_nofbaccel, int, 0400);
 

drivers/gpu/drm/nouveau/nouveau_drv.h

@@ -34,7 +34,7 @@
 
 #define DRIVER_MAJOR		0
 #define DRIVER_MINOR		0
-#define DRIVER_PATCHLEVEL	15
+#define DRIVER_PATCHLEVEL	16
 
 #define NOUVEAU_FAMILY   0x0000FFFF
 #define NOUVEAU_FLAGS    0xFFFF0000
@@ -83,6 +83,7 @@ struct nouveau_bo {
 	struct drm_file *reserved_by;
 	struct list_head entry;
 	int pbbo_index;
+	bool validate_mapped;
 
 	struct nouveau_channel *channel;
 
@@ -239,6 +240,11 @@ struct nouveau_channel {
 		int cur;
 		int put;
 		/* access via pushbuf_bo */
+
+		int ib_base;
+		int ib_max;
+		int ib_free;
+		int ib_put;
 	} dma;
 
 	uint32_t sw_subchannel[8];
@@ -533,6 +539,9 @@ struct drm_nouveau_private {
 	struct nouveau_engine engine;
 	struct nouveau_channel *channel;
 
+	/* For PFIFO and PGRAPH. */
+	spinlock_t context_switch_lock;
+
 	/* RAMIN configuration, RAMFC, RAMHT and RAMRO offsets */
 	struct nouveau_gpuobj *ramht;
 	uint32_t ramin_rsvd_vram;
@@ -596,8 +605,7 @@ struct drm_nouveau_private {
 
 	struct list_head gpuobj_list;
 
-	struct nvbios VBIOS;
-	struct nouveau_bios_info *vbios;
+	struct nvbios vbios;
 
 	struct nv04_mode_state mode_reg;
 	struct nv04_mode_state saved_reg;
@@ -696,12 +704,6 @@ extern bool nouveau_wait_until(struct drm_device *, uint64_t timeout,
 			       uint32_t reg, uint32_t mask, uint32_t val);
 extern bool nouveau_wait_for_idle(struct drm_device *);
 extern int  nouveau_card_init(struct drm_device *);
-extern int  nouveau_ioctl_card_init(struct drm_device *, void *data,
-				    struct drm_file *);
-extern int  nouveau_ioctl_suspend(struct drm_device *, void *data,
-				  struct drm_file *);
-extern int  nouveau_ioctl_resume(struct drm_device *, void *data,
-				 struct drm_file *);
 
 /* nouveau_mem.c */
 extern int  nouveau_mem_init_heap(struct mem_block **, uint64_t start,
@@ -845,7 +847,7 @@ nouveau_debugfs_channel_fini(struct nouveau_channel *chan)
 /* nouveau_dma.c */
 extern void nouveau_dma_pre_init(struct nouveau_channel *);
 extern int  nouveau_dma_init(struct nouveau_channel *);
-extern int  nouveau_dma_wait(struct nouveau_channel *, int size);
+extern int  nouveau_dma_wait(struct nouveau_channel *, int slots, int size);
 
 /* nouveau_acpi.c */
 #ifdef CONFIG_ACPI
@@ -1027,6 +1029,7 @@ extern void nv50_graph_destroy_context(struct nouveau_channel *);
 extern int  nv50_graph_load_context(struct nouveau_channel *);
 extern int  nv50_graph_unload_context(struct drm_device *);
 extern void nv50_graph_context_switch(struct drm_device *);
+extern int  nv50_grctx_init(struct nouveau_grctx *);
 
 /* nouveau_grctx.c */
 extern int  nouveau_grctx_prog_load(struct drm_device *);
@@ -1152,16 +1155,6 @@ extern int nouveau_gem_ioctl_new(struct drm_device *, void *,
 				 struct drm_file *);
 extern int nouveau_gem_ioctl_pushbuf(struct drm_device *, void *,
 				     struct drm_file *);
-extern int nouveau_gem_ioctl_pushbuf_call(struct drm_device *, void *,
-					  struct drm_file *);
-extern int nouveau_gem_ioctl_pushbuf_call2(struct drm_device *, void *,
-					   struct drm_file *);
-extern int nouveau_gem_ioctl_pin(struct drm_device *, void *,
-				 struct drm_file *);
-extern int nouveau_gem_ioctl_unpin(struct drm_device *, void *,
-				   struct drm_file *);
-extern int nouveau_gem_ioctl_tile(struct drm_device *, void *,
-				  struct drm_file *);
 extern int nouveau_gem_ioctl_cpu_prep(struct drm_device *, void *,
 				      struct drm_file *);
 extern int nouveau_gem_ioctl_cpu_fini(struct drm_device *, void *,

drivers/gpu/drm/nouveau/nouveau_hw.c

@@ -160,7 +160,7 @@ static void
 setPLL_single(struct drm_device *dev, uint32_t reg, struct nouveau_pll_vals *pv)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	int chip_version = dev_priv->vbios->chip_version;
+	int chip_version = dev_priv->vbios.chip_version;
 	uint32_t oldpll = NVReadRAMDAC(dev, 0, reg);
 	int oldN = (oldpll >> 8) & 0xff, oldM = oldpll & 0xff;
 	uint32_t pll = (oldpll & 0xfff80000) | pv->log2P << 16 | pv->NM1;
@@ -216,7 +216,7 @@ setPLL_double_highregs(struct drm_device *dev, uint32_t reg1,
 		       struct nouveau_pll_vals *pv)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	int chip_version = dev_priv->vbios->chip_version;
+	int chip_version = dev_priv->vbios.chip_version;
 	bool nv3035 = chip_version == 0x30 || chip_version == 0x35;
 	uint32_t reg2 = reg1 + ((reg1 == NV_RAMDAC_VPLL2) ? 0x5c : 0x70);
 	uint32_t oldpll1 = NVReadRAMDAC(dev, 0, reg1);
@@ -374,7 +374,7 @@ nouveau_hw_setpll(struct drm_device *dev, uint32_t reg1,
 		  struct nouveau_pll_vals *pv)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	int cv = dev_priv->vbios->chip_version;
+	int cv = dev_priv->vbios.chip_version;
 
 	if (cv == 0x30 || cv == 0x31 || cv == 0x35 || cv == 0x36 ||
 	    cv >= 0x40) {

drivers/gpu/drm/nouveau/nouveau_i2c.c

@@ -254,16 +254,16 @@ struct nouveau_i2c_chan *
 nouveau_i2c_find(struct drm_device *dev, int index)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	struct nvbios *bios = &dev_priv->VBIOS;
+	struct nvbios *bios = &dev_priv->vbios;
 
-	if (index > DCB_MAX_NUM_I2C_ENTRIES)
+	if (index >= DCB_MAX_NUM_I2C_ENTRIES)
 		return NULL;
 
-	if (!bios->bdcb.dcb.i2c[index].chan) {
-		if (nouveau_i2c_init(dev, &bios->bdcb.dcb.i2c[index], index))
+	if (!bios->dcb.i2c[index].chan) {
+		if (nouveau_i2c_init(dev, &bios->dcb.i2c[index], index))
 			return NULL;
 	}
 
-	return bios->bdcb.dcb.i2c[index].chan;
+	return bios->dcb.i2c[index].chan;
 }
 

drivers/gpu/drm/nouveau/nouveau_irq.c

@@ -691,11 +691,14 @@ nouveau_irq_handler(DRM_IRQ_ARGS)
 	struct drm_device *dev = (struct drm_device *)arg;
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	uint32_t status, fbdev_flags = 0;
+	unsigned long flags;
 
 	status = nv_rd32(dev, NV03_PMC_INTR_0);
 	if (!status)
 		return IRQ_NONE;
 
+	spin_lock_irqsave(&dev_priv->context_switch_lock, flags);
+
 	if (dev_priv->fbdev_info) {
 		fbdev_flags = dev_priv->fbdev_info->flags;
 		dev_priv->fbdev_info->flags |= FBINFO_HWACCEL_DISABLED;
@@ -733,5 +736,7 @@ nouveau_irq_handler(DRM_IRQ_ARGS)
 	if (dev_priv->fbdev_info)
 		dev_priv->fbdev_info->flags = fbdev_flags;
 
+	spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags);
+
 	return IRQ_HANDLED;
 }

drivers/gpu/drm/nouveau/nouveau_state.c

@@ -391,6 +391,7 @@ nouveau_card_init(struct drm_device *dev)
 		goto out;
 	engine = &dev_priv->engine;
 	dev_priv->init_state = NOUVEAU_CARD_INIT_FAILED;
+	spin_lock_init(&dev_priv->context_switch_lock);
 
 	/* Parse BIOS tables / Run init tables if card not POSTed */
 	if (drm_core_check_feature(dev, DRIVER_MODESET)) {
@@ -776,13 +777,6 @@ int nouveau_unload(struct drm_device *dev)
 	return 0;
 }
 
-int
-nouveau_ioctl_card_init(struct drm_device *dev, void *data,
-			struct drm_file *file_priv)
-{
-	return nouveau_card_init(dev);
-}
-
 int nouveau_ioctl_getparam(struct drm_device *dev, void *data,
 						struct drm_file *file_priv)
 {

drivers/gpu/drm/nouveau/nv04_dac.c

@@ -230,13 +230,13 @@ uint32_t nv17_dac_sample_load(struct drm_encoder *encoder)
 	if (dcb->type == OUTPUT_TV) {
 		testval = RGB_TEST_DATA(0xa0, 0xa0, 0xa0);
 
-		if (dev_priv->vbios->tvdactestval)
-			testval = dev_priv->vbios->tvdactestval;
+		if (dev_priv->vbios.tvdactestval)
+			testval = dev_priv->vbios.tvdactestval;
 	} else {
 		testval = RGB_TEST_DATA(0x140, 0x140, 0x140); /* 0x94050140 */
 
-		if (dev_priv->vbios->dactestval)
-			testval = dev_priv->vbios->dactestval;
+		if (dev_priv->vbios.dactestval)
+			testval = dev_priv->vbios.dactestval;
 	}
 
 	saved_rtest_ctrl = NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset);

drivers/gpu/drm/nouveau/nv04_dfp.c

@@ -269,10 +269,10 @@ static void nv04_dfp_mode_set(struct drm_encoder *encoder,
 	regp->fp_horiz_regs[FP_TOTAL] = output_mode->htotal - 1;
 	if (!nv_gf4_disp_arch(dev) ||
 	    (output_mode->hsync_start - output_mode->hdisplay) >=
-					dev_priv->vbios->digital_min_front_porch)
+					dev_priv->vbios.digital_min_front_porch)
 		regp->fp_horiz_regs[FP_CRTC] = output_mode->hdisplay;
 	else
-		regp->fp_horiz_regs[FP_CRTC] = output_mode->hsync_start - dev_priv->vbios->digital_min_front_porch - 1;
+		regp->fp_horiz_regs[FP_CRTC] = output_mode->hsync_start - dev_priv->vbios.digital_min_front_porch - 1;
 	regp->fp_horiz_regs[FP_SYNC_START] = output_mode->hsync_start - 1;
 	regp->fp_horiz_regs[FP_SYNC_END] = output_mode->hsync_end - 1;
 	regp->fp_horiz_regs[FP_VALID_START] = output_mode->hskew;

drivers/gpu/drm/nouveau/nv04_display.c

@@ -93,10 +93,9 @@ int
 nv04_display_create(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	struct parsed_dcb *dcb = dev_priv->vbios->dcb;
+	struct dcb_table *dcb = &dev_priv->vbios.dcb;
 	struct drm_encoder *encoder;
 	struct drm_crtc *crtc;
-	uint16_t connector[16] = { 0 };
 	int i, ret;
 
 	NV_DEBUG_KMS(dev, "\n");
@@ -154,52 +153,10 @@ nv04_display_create(struct drm_device *dev)
 
 		if (ret)
 			continue;
-
-		connector[dcbent->connector] |= (1 << dcbent->type);
 	}
 
-	for (i = 0; i < dcb->entries; i++) {
-		struct dcb_entry *dcbent = &dcb->entry[i];
-		uint16_t encoders;
-		int type;
-
-		encoders = connector[dcbent->connector];
-		if (!(encoders & (1 << dcbent->type)))
-			continue;
-		connector[dcbent->connector] = 0;
-
-		switch (dcbent->type) {
-		case OUTPUT_ANALOG:
-			if (!MULTIPLE_ENCODERS(encoders))
-				type = DRM_MODE_CONNECTOR_VGA;
-			else
-				type = DRM_MODE_CONNECTOR_DVII;
-			break;
-		case OUTPUT_TMDS:
-			if (!MULTIPLE_ENCODERS(encoders))
-				type = DRM_MODE_CONNECTOR_DVID;
-			else
-				type = DRM_MODE_CONNECTOR_DVII;
-			break;
-		case OUTPUT_LVDS:
-			type = DRM_MODE_CONNECTOR_LVDS;
-#if 0
-			/* don't create i2c adapter when lvds ddc not allowed */
-			if (dcbent->lvdsconf.use_straps_for_mode ||
-			    dev_priv->vbios->fp_no_ddc)
-				i2c_index = 0xf;
-#endif
-			break;
-		case OUTPUT_TV:
-			type = DRM_MODE_CONNECTOR_TV;
-			break;
-		default:
-			type = DRM_MODE_CONNECTOR_Unknown;
-			continue;
-		}
-
-		nouveau_connector_create(dev, dcbent->connector, type);
-	}
+	for (i = 0; i < dcb->connector.entries; i++)
+		nouveau_connector_create(dev, &dcb->connector.entry[i]);
 
 	/* Save previous state */
 	NVLockVgaCrtcs(dev, false);

drivers/gpu/drm/nouveau/nv04_fbcon.c

@@ -118,7 +118,7 @@ nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
 		return;
 	}
 
-	width = (image->width + 31) & ~31;
+	width = ALIGN(image->width, 32);
 	dsize = (width * image->height) >> 5;
 
 	if (info->fix.visual == FB_VISUAL_TRUECOLOR ||

drivers/gpu/drm/nouveau/nv04_fifo.c

@@ -117,6 +117,7 @@ nv04_fifo_create_context(struct nouveau_channel *chan)
 {
 	struct drm_device *dev = chan->dev;
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
+	unsigned long flags;
 	int ret;
 
 	ret = nouveau_gpuobj_new_fake(dev, NV04_RAMFC(chan->id), ~0,
@@ -127,6 +128,8 @@ nv04_fifo_create_context(struct nouveau_channel *chan)
 	if (ret)
 		return ret;
 
+	spin_lock_irqsave(&dev_priv->context_switch_lock, flags);
+
 	/* Setup initial state */
 	dev_priv->engine.instmem.prepare_access(dev, true);
 	RAMFC_WR(DMA_PUT, chan->pushbuf_base);
@@ -144,6 +147,8 @@ nv04_fifo_create_context(struct nouveau_channel *chan)
 	/* enable the fifo dma operation */
 	nv_wr32(dev, NV04_PFIFO_MODE,
 		nv_rd32(dev, NV04_PFIFO_MODE) | (1 << chan->id));
+
+	spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags);
 	return 0;
 }
 

drivers/gpu/drm/nouveau/nv04_tv.c

@@ -262,7 +262,7 @@ int nv04_tv_create(struct drm_device *dev, struct dcb_entry *entry)
 	nv_encoder->or = ffs(entry->or) - 1;
 
 	/* Run the slave-specific initialization */
-	adap = &dev_priv->vbios->dcb->i2c[i2c_index].chan->adapter;
+	adap = &dev_priv->vbios.dcb.i2c[i2c_index].chan->adapter;
 
 	was_locked = NVLockVgaCrtcs(dev, false);
 

drivers/gpu/drm/nouveau/nv17_tv.c

@@ -45,8 +45,8 @@ static uint32_t nv42_tv_sample_load(struct drm_encoder *encoder)
 
 #define RGB_TEST_DATA(r, g, b) (r << 0 | g << 10 | b << 20)
 	testval = RGB_TEST_DATA(0x82, 0xeb, 0x82);
-	if (dev_priv->vbios->tvdactestval)
-		testval = dev_priv->vbios->tvdactestval;
+	if (dev_priv->vbios.tvdactestval)
+		testval = dev_priv->vbios.tvdactestval;
 
 	dacclk = NVReadRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset);
 	head = (dacclk & 0x100) >> 8;
@@ -367,7 +367,7 @@ static void nv17_tv_prepare(struct drm_encoder *encoder)
 			     !enc->crtc &&
 			     nv04_dfp_get_bound_head(dev, dcb) == head) {
 				nv04_dfp_bind_head(dev, dcb, head ^ 1,
-						dev_priv->VBIOS.fp.dual_link);
+						dev_priv->vbios.fp.dual_link);
 			}
 		}
 

drivers/gpu/drm/nouveau/nv40_fifo.c

@@ -37,6 +37,7 @@ nv40_fifo_create_context(struct nouveau_channel *chan)
 	struct drm_device *dev = chan->dev;
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	uint32_t fc = NV40_RAMFC(chan->id);
+	unsigned long flags;
 	int ret;
 
 	ret = nouveau_gpuobj_new_fake(dev, NV40_RAMFC(chan->id), ~0,
@@ -45,6 +46,8 @@ nv40_fifo_create_context(struct nouveau_channel *chan)
 	if (ret)
 		return ret;
 
+	spin_lock_irqsave(&dev_priv->context_switch_lock, flags);
+
 	dev_priv->engine.instmem.prepare_access(dev, true);
 	nv_wi32(dev, fc +  0, chan->pushbuf_base);
 	nv_wi32(dev, fc +  4, chan->pushbuf_base);
@@ -63,6 +66,8 @@ nv40_fifo_create_context(struct nouveau_channel *chan)
 	/* enable the fifo dma operation */
 	nv_wr32(dev, NV04_PFIFO_MODE,
 		nv_rd32(dev, NV04_PFIFO_MODE) | (1 << chan->id));
+
+	spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags);
 	return 0;
 }
 

drivers/gpu/drm/nouveau/nv50_dac.c

@@ -79,8 +79,8 @@ nv50_dac_detect(struct drm_encoder *encoder, struct drm_connector *connector)
 	}
 
 	/* Use bios provided value if possible. */
-	if (dev_priv->vbios->dactestval) {
-		load_pattern = dev_priv->vbios->dactestval;
+	if (dev_priv->vbios.dactestval) {
+		load_pattern = dev_priv->vbios.dactestval;
 		NV_DEBUG_KMS(dev, "Using bios provided load_pattern of %d\n",
 			  load_pattern);
 	} else {

drivers/gpu/drm/nouveau/nv50_display.c

@@ -370,9 +370,7 @@ nv50_display_init(struct drm_device *dev)
 		struct nouveau_connector *conn = nouveau_connector(connector);
 		struct dcb_gpio_entry *gpio;
 
-		if (connector->connector_type != DRM_MODE_CONNECTOR_DVII &&
-		    connector->connector_type != DRM_MODE_CONNECTOR_DVID &&
-		    connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort)
+		if (conn->dcb->gpio_tag == 0xff)
 			continue;
 
 		gpio = nouveau_bios_gpio_entry(dev, conn->dcb->gpio_tag);
@@ -465,8 +463,7 @@ static int nv50_display_disable(struct drm_device *dev)
 int nv50_display_create(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	struct parsed_dcb *dcb = dev_priv->vbios->dcb;
-	uint32_t connector[16] = {};
+	struct dcb_table *dcb = &dev_priv->vbios.dcb;
 	int ret, i;
 
 	NV_DEBUG_KMS(dev, "\n");
@@ -522,44 +519,13 @@ int nv50_display_create(struct drm_device *dev)
 			NV_WARN(dev, "DCB encoder %d unknown\n", entry->type);
 			continue;
 		}
-
-		connector[entry->connector] |= (1 << entry->type);
 	}
 
-	/* It appears that DCB 3.0+ VBIOS has a connector table, however,
-	 * I'm not 100% certain how to decode it correctly yet so just
-	 * look at what encoders are present on each connector index and
-	 * attempt to derive the connector type from that.
-	 */
-	for (i = 0 ; i < dcb->entries; i++) {
-		struct dcb_entry *entry = &dcb->entry[i];
-		uint16_t encoders;
-		int type;
-
-		encoders = connector[entry->connector];
-		if (!(encoders & (1 << entry->type)))
+	for (i = 0 ; i < dcb->connector.entries; i++) {
+		if (i != 0 && dcb->connector.entry[i].index ==
+			      dcb->connector.entry[i - 1].index)
 			continue;
-		connector[entry->connector] = 0;
-
-		if (encoders & (1 << OUTPUT_DP)) {
-			type = DRM_MODE_CONNECTOR_DisplayPort;
-		} else if (encoders & (1 << OUTPUT_TMDS)) {
-			if (encoders & (1 << OUTPUT_ANALOG))
-				type = DRM_MODE_CONNECTOR_DVII;
-			else
-				type = DRM_MODE_CONNECTOR_DVID;
-		} else if (encoders & (1 << OUTPUT_ANALOG)) {
-			type = DRM_MODE_CONNECTOR_VGA;
-		} else if (encoders & (1 << OUTPUT_LVDS)) {
-			type = DRM_MODE_CONNECTOR_LVDS;
-		} else {
-			type = DRM_MODE_CONNECTOR_Unknown;
-		}
-
-		if (type == DRM_MODE_CONNECTOR_Unknown)
-			continue;
-
-		nouveau_connector_create(dev, entry->connector, type);
+		nouveau_connector_create(dev, &dcb->connector.entry[i]);
 	}
 
 	ret = nv50_display_init(dev);
@@ -667,8 +633,8 @@ nv50_display_irq_head(struct drm_device *dev, int *phead,
 		return -1;
 	}
 
-	for (i = 0; i < dev_priv->vbios->dcb->entries; i++) {
-		struct dcb_entry *dcbent = &dev_priv->vbios->dcb->entry[i];
+	for (i = 0; i < dev_priv->vbios.dcb.entries; i++) {
+		struct dcb_entry *dcbent = &dev_priv->vbios.dcb.entry[i];
 
 		if (dcbent->type != type)
 			continue;
@@ -692,7 +658,7 @@ nv50_display_script_select(struct drm_device *dev, struct dcb_entry *dcbent,
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_connector *nv_connector = NULL;
 	struct drm_encoder *encoder;
-	struct nvbios *bios = &dev_priv->VBIOS;
+	struct nvbios *bios = &dev_priv->vbios;
 	uint32_t mc, script = 0, or;
 
 	list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
@@ -710,7 +676,7 @@ nv50_display_script_select(struct drm_device *dev, struct dcb_entry *dcbent,
 	switch (dcbent->type) {
 	case OUTPUT_LVDS:
 		script = (mc >> 8) & 0xf;
-		if (bios->pub.fp_no_ddc) {
+		if (bios->fp_no_ddc) {
 			if (bios->fp.dual_link)
 				script |= 0x0100;
 			if (bios->fp.if_is_24bit)

drivers/gpu/drm/nouveau/nv50_fbcon.c

@@ -109,7 +109,7 @@ nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
 		return;
 	}
 
-	width = (image->width + 31) & ~31;
+	width = ALIGN(image->width, 32);
 	dwords = (width * image->height) >> 5;
 
 	BEGIN_RING(chan, NvSub2D, 0x0814, 2);

drivers/gpu/drm/nouveau/nv50_fifo.c

@@ -243,6 +243,7 @@ nv50_fifo_create_context(struct nouveau_channel *chan)
 	struct drm_device *dev = chan->dev;
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_gpuobj *ramfc = NULL;
+	unsigned long flags;
 	int ret;
 
 	NV_DEBUG(dev, "ch%d\n", chan->id);
@@ -278,19 +279,21 @@ nv50_fifo_create_context(struct nouveau_channel *chan)
 			return ret;
 	}
 
+	spin_lock_irqsave(&dev_priv->context_switch_lock, flags);
+
 	dev_priv->engine.instmem.prepare_access(dev, true);
 
-	nv_wo32(dev, ramfc, 0x08/4, chan->pushbuf_base);
-	nv_wo32(dev, ramfc, 0x10/4, chan->pushbuf_base);
 	nv_wo32(dev, ramfc, 0x48/4, chan->pushbuf->instance >> 4);
 	nv_wo32(dev, ramfc, 0x80/4, (0xc << 24) | (chan->ramht->instance >> 4));
-	nv_wo32(dev, ramfc, 0x3c/4, 0x00086078);
 	nv_wo32(dev, ramfc, 0x44/4, 0x2101ffff);
 	nv_wo32(dev, ramfc, 0x60/4, 0x7fffffff);
 	nv_wo32(dev, ramfc, 0x40/4, 0x00000000);
 	nv_wo32(dev, ramfc, 0x7c/4, 0x30000001);
 	nv_wo32(dev, ramfc, 0x78/4, 0x00000000);
-	nv_wo32(dev, ramfc, 0x4c/4, 0xffffffff);
+	nv_wo32(dev, ramfc, 0x3c/4, 0x403f6078);
+	nv_wo32(dev, ramfc, 0x50/4, chan->pushbuf_base +
+				    chan->dma.ib_base * 4);
+	nv_wo32(dev, ramfc, 0x54/4, drm_order(chan->dma.ib_max + 1) << 16);
 
 	if (!IS_G80) {
 		nv_wo32(dev, chan->ramin->gpuobj, 0, chan->id);
@@ -306,10 +309,12 @@ nv50_fifo_create_context(struct nouveau_channel *chan)
 	ret = nv50_fifo_channel_enable(dev, chan->id, false);
 	if (ret) {
 		NV_ERROR(dev, "error enabling ch%d: %d\n", chan->id, ret);
+		spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags);
 		nouveau_gpuobj_ref_del(dev, &chan->ramfc);
 		return ret;
 	}
 
+	spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags);
 	return 0;
 }
 

drivers/gpu/drm/nouveau/nv50_graph.c

@@ -28,30 +28,7 @@
 #include "drm.h"
 #include "nouveau_drv.h"
 
-MODULE_FIRMWARE("nouveau/nv50.ctxprog");
-MODULE_FIRMWARE("nouveau/nv50.ctxvals");
-MODULE_FIRMWARE("nouveau/nv84.ctxprog");
-MODULE_FIRMWARE("nouveau/nv84.ctxvals");
-MODULE_FIRMWARE("nouveau/nv86.ctxprog");
-MODULE_FIRMWARE("nouveau/nv86.ctxvals");
-MODULE_FIRMWARE("nouveau/nv92.ctxprog");
-MODULE_FIRMWARE("nouveau/nv92.ctxvals");
-MODULE_FIRMWARE("nouveau/nv94.ctxprog");
-MODULE_FIRMWARE("nouveau/nv94.ctxvals");
-MODULE_FIRMWARE("nouveau/nv96.ctxprog");
-MODULE_FIRMWARE("nouveau/nv96.ctxvals");
-MODULE_FIRMWARE("nouveau/nv98.ctxprog");
-MODULE_FIRMWARE("nouveau/nv98.ctxvals");
-MODULE_FIRMWARE("nouveau/nva0.ctxprog");
-MODULE_FIRMWARE("nouveau/nva0.ctxvals");
-MODULE_FIRMWARE("nouveau/nva5.ctxprog");
-MODULE_FIRMWARE("nouveau/nva5.ctxvals");
-MODULE_FIRMWARE("nouveau/nva8.ctxprog");
-MODULE_FIRMWARE("nouveau/nva8.ctxvals");
-MODULE_FIRMWARE("nouveau/nvaa.ctxprog");
-MODULE_FIRMWARE("nouveau/nvaa.ctxvals");
-MODULE_FIRMWARE("nouveau/nvac.ctxprog");
-MODULE_FIRMWARE("nouveau/nvac.ctxvals");
+#include "nouveau_grctx.h"
 
 #define IS_G80 ((dev_priv->chipset & 0xf0) == 0x50)
 
@@ -111,9 +88,34 @@ nv50_graph_init_ctxctl(struct drm_device *dev)
 
 	NV_DEBUG(dev, "\n");
 
-	nouveau_grctx_prog_load(dev);
-	if (!dev_priv->engine.graph.ctxprog)
-		dev_priv->engine.graph.accel_blocked = true;
+	if (nouveau_ctxfw) {
+		nouveau_grctx_prog_load(dev);
+		dev_priv->engine.graph.grctx_size = 0x70000;
+	}
+	if (!dev_priv->engine.graph.ctxprog) {
+		struct nouveau_grctx ctx = {};
+		uint32_t *cp = kmalloc(512 * 4, GFP_KERNEL);
+		int i;
+		if (!cp) {
+			NV_ERROR(dev, "Couldn't alloc ctxprog! Disabling acceleration.\n");
+			dev_priv->engine.graph.accel_blocked = true;
+			return 0;
+		}
+		ctx.dev = dev;
+		ctx.mode = NOUVEAU_GRCTX_PROG;
+		ctx.data = cp;
+		ctx.ctxprog_max = 512;
+		if (!nv50_grctx_init(&ctx)) {
+			dev_priv->engine.graph.grctx_size = ctx.ctxvals_pos * 4;
+
+			nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_INDEX, 0);
+			for (i = 0; i < ctx.ctxprog_len; i++)
+				nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_DATA, cp[i]);
+		} else {
+			dev_priv->engine.graph.accel_blocked = true;
+		}
+		kfree(cp);
+	}
 
 	nv_wr32(dev, 0x400320, 4);
 	nv_wr32(dev, NV40_PGRAPH_CTXCTL_CUR, 0);
@@ -193,13 +195,13 @@ nv50_graph_create_context(struct nouveau_channel *chan)
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_gpuobj *ramin = chan->ramin->gpuobj;
 	struct nouveau_gpuobj *ctx;
-	uint32_t grctx_size = 0x70000;
+	struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph;
 	int hdr, ret;
 
 	NV_DEBUG(dev, "ch%d\n", chan->id);
 
-	ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, grctx_size, 0x1000,
-				     NVOBJ_FLAG_ZERO_ALLOC |
+	ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, pgraph->grctx_size,
+				     0x1000, NVOBJ_FLAG_ZERO_ALLOC |
 				     NVOBJ_FLAG_ZERO_FREE, &chan->ramin_grctx);
 	if (ret)
 		return ret;
@@ -209,7 +211,7 @@ nv50_graph_create_context(struct nouveau_channel *chan)
 	dev_priv->engine.instmem.prepare_access(dev, true);
 	nv_wo32(dev, ramin, (hdr + 0x00)/4, 0x00190002);
 	nv_wo32(dev, ramin, (hdr + 0x04)/4, chan->ramin_grctx->instance +
-					   grctx_size - 1);
+					   pgraph->grctx_size - 1);
 	nv_wo32(dev, ramin, (hdr + 0x08)/4, chan->ramin_grctx->instance);
 	nv_wo32(dev, ramin, (hdr + 0x0c)/4, 0);
 	nv_wo32(dev, ramin, (hdr + 0x10)/4, 0);
@@ -217,7 +219,15 @@ nv50_graph_create_context(struct nouveau_channel *chan)
 	dev_priv->engine.instmem.finish_access(dev);
 
 	dev_priv->engine.instmem.prepare_access(dev, true);
-	nouveau_grctx_vals_load(dev, ctx);
+	if (!pgraph->ctxprog) {
+		struct nouveau_grctx ctx = {};
+		ctx.dev = chan->dev;
+		ctx.mode = NOUVEAU_GRCTX_VALS;
+		ctx.data = chan->ramin_grctx->gpuobj;
+		nv50_grctx_init(&ctx);
+	} else {
+		nouveau_grctx_vals_load(dev, ctx);
+	}
 	nv_wo32(dev, ctx, 0x00000/4, chan->ramin->instance >> 12);
 	if ((dev_priv->chipset & 0xf0) == 0xa0)
 		nv_wo32(dev, ctx, 0x00004/4, 0x00000000);

drivers/gpu/drm/nouveau/nv50_instmem.c

@@ -390,7 +390,7 @@ nv50_instmem_populate(struct drm_device *dev, struct nouveau_gpuobj *gpuobj,
 	if (gpuobj->im_backing)
 		return -EINVAL;
 
-	*sz = (*sz + (NV50_INSTMEM_PAGE_SIZE-1)) & ~(NV50_INSTMEM_PAGE_SIZE-1);
+	*sz = ALIGN(*sz, NV50_INSTMEM_PAGE_SIZE);
 	if (*sz == 0)
 		return -EINVAL;
 

include/drm/nouveau_drm.h

@@ -25,13 +25,14 @@
 #ifndef __NOUVEAU_DRM_H__
 #define __NOUVEAU_DRM_H__
 
-#define NOUVEAU_DRM_HEADER_PATCHLEVEL 15
+#define NOUVEAU_DRM_HEADER_PATCHLEVEL 16
 
 struct drm_nouveau_channel_alloc {
 	uint32_t     fb_ctxdma_handle;
 	uint32_t     tt_ctxdma_handle;
 
 	int          channel;
+	uint32_t     pushbuf_domains;
 
 	/* Notifier memory */
 	uint32_t     notifier_handle;
@@ -109,68 +110,58 @@ struct drm_nouveau_gem_new {
 	uint32_t align;
 };
 
+#define NOUVEAU_GEM_MAX_BUFFERS 1024
+struct drm_nouveau_gem_pushbuf_bo_presumed {
+	uint32_t valid;
+	uint32_t domain;
+	uint64_t offset;
+};
+
 struct drm_nouveau_gem_pushbuf_bo {
 	uint64_t user_priv;
 	uint32_t handle;
 	uint32_t read_domains;
 	uint32_t write_domains;
 	uint32_t valid_domains;
-	uint32_t presumed_ok;
-	uint32_t presumed_domain;
-	uint64_t presumed_offset;
+	struct drm_nouveau_gem_pushbuf_bo_presumed presumed;
 };
 
 #define NOUVEAU_GEM_RELOC_LOW  (1 << 0)
 #define NOUVEAU_GEM_RELOC_HIGH (1 << 1)
 #define NOUVEAU_GEM_RELOC_OR   (1 << 2)
+#define NOUVEAU_GEM_MAX_RELOCS 1024
 struct drm_nouveau_gem_pushbuf_reloc {
+	uint32_t reloc_bo_index;
+	uint32_t reloc_bo_offset;
 	uint32_t bo_index;
-	uint32_t reloc_index;
 	uint32_t flags;
 	uint32_t data;
 	uint32_t vor;
 	uint32_t tor;
 };
 
-#define NOUVEAU_GEM_MAX_BUFFERS 1024
-#define NOUVEAU_GEM_MAX_RELOCS 1024
+#define NOUVEAU_GEM_MAX_PUSH 512
+struct drm_nouveau_gem_pushbuf_push {
+	uint32_t bo_index;
+	uint32_t pad;
+	uint64_t offset;
+	uint64_t length;
+};
 
 struct drm_nouveau_gem_pushbuf {
 	uint32_t channel;
-	uint32_t nr_dwords;
 	uint32_t nr_buffers;
-	uint32_t nr_relocs;
-	uint64_t dwords;
 	uint64_t buffers;
-	uint64_t relocs;
-};
-
-struct drm_nouveau_gem_pushbuf_call {
-	uint32_t channel;
-	uint32_t handle;
-	uint32_t offset;
-	uint32_t nr_buffers;
 	uint32_t nr_relocs;
-	uint32_t nr_dwords;
-	uint64_t buffers;
+	uint32_t nr_push;
 	uint64_t relocs;
+	uint64_t push;
 	uint32_t suffix0;
 	uint32_t suffix1;
-	/* below only accessed for CALL2 */
 	uint64_t vram_available;
 	uint64_t gart_available;
 };
 
-struct drm_nouveau_gem_pin {
-	uint32_t handle;
-	uint32_t domain;
-	uint64_t offset;
-};
-
-struct drm_nouveau_gem_unpin {
-	uint32_t handle;
-};
-
 #define NOUVEAU_GEM_CPU_PREP_NOWAIT                                  0x00000001
 #define NOUVEAU_GEM_CPU_PREP_NOBLOCK                                 0x00000002
 #define NOUVEAU_GEM_CPU_PREP_WRITE                                   0x00000004
@@ -183,14 +174,6 @@ struct drm_nouveau_gem_cpu_fini {
 	uint32_t handle;
 };
 
-struct drm_nouveau_gem_tile {
-	uint32_t handle;
-	uint32_t offset;
-	uint32_t size;
-	uint32_t tile_mode;
-	uint32_t tile_flags;
-};
-
 enum nouveau_bus_type {
 	NV_AGP     = 0,
 	NV_PCI     = 1,
@@ -200,22 +183,17 @@ enum nouveau_bus_type {
 struct drm_nouveau_sarea {
 };
 
-#define DRM_NOUVEAU_CARD_INIT          0x00
-#define DRM_NOUVEAU_GETPARAM           0x01
-#define DRM_NOUVEAU_SETPARAM           0x02
-#define DRM_NOUVEAU_CHANNEL_ALLOC      0x03
-#define DRM_NOUVEAU_CHANNEL_FREE       0x04
-#define DRM_NOUVEAU_GROBJ_ALLOC        0x05
-#define DRM_NOUVEAU_NOTIFIEROBJ_ALLOC  0x06
-#define DRM_NOUVEAU_GPUOBJ_FREE        0x07
+#define DRM_NOUVEAU_GETPARAM           0x00
+#define DRM_NOUVEAU_SETPARAM           0x01
+#define DRM_NOUVEAU_CHANNEL_ALLOC      0x02
+#define DRM_NOUVEAU_CHANNEL_FREE       0x03
+#define DRM_NOUVEAU_GROBJ_ALLOC        0x04
+#define DRM_NOUVEAU_NOTIFIEROBJ_ALLOC  0x05
+#define DRM_NOUVEAU_GPUOBJ_FREE        0x06
 #define DRM_NOUVEAU_GEM_NEW            0x40
 #define DRM_NOUVEAU_GEM_PUSHBUF        0x41
-#define DRM_NOUVEAU_GEM_PUSHBUF_CALL   0x42
-#define DRM_NOUVEAU_GEM_PIN            0x43 /* !KMS only */
-#define DRM_NOUVEAU_GEM_UNPIN          0x44 /* !KMS only */
-#define DRM_NOUVEAU_GEM_CPU_PREP       0x45
-#define DRM_NOUVEAU_GEM_CPU_FINI       0x46
-#define DRM_NOUVEAU_GEM_INFO           0x47
-#define DRM_NOUVEAU_GEM_PUSHBUF_CALL2  0x48
+#define DRM_NOUVEAU_GEM_CPU_PREP       0x42
+#define DRM_NOUVEAU_GEM_CPU_FINI       0x43
+#define DRM_NOUVEAU_GEM_INFO           0x44
 
 #endif /* __NOUVEAU_DRM_H__ */