Merge remote branch 'origin/master' into modesetting-101

Conflicts:

	linux-core/drm_bo.c
	linux-core/drm_drv.c
	shared-core/drm.h
	shared-core/i915_dma.c
	shared-core/i915_drv.h
	shared-core/i915_irq.c
	shared-core/radeon_irq.c

45 files changed, 1417 insertions, 954 deletions

diff --git a/shared-core/drm.h b/shared-core/drm.h
index f8d44048..7aea3033 100644
--- a/shared-core/drm.h
+++ b/shared-core/drm.h
@@ -555,6 +555,21 @@ union drm_wait_vblank {
 	struct drm_wait_vblank_reply reply;
 };
 
+enum drm_modeset_ctl_cmd {
+	_DRM_PRE_MODESET = 1,
+	_DRM_POST_MODESET = 2,
+};
+
+/**
+ * DRM_IOCTL_MODESET_CTL ioctl argument type
+ *
+ * \sa drmModesetCtl().
+ */
+struct drm_modeset_ctl {
+	unsigned long arg;
+	enum drm_modeset_ctl_cmd cmd;
+};
+
 /**
  * DRM_IOCTL_AGP_ENABLE ioctl argument type.
  *
@@ -662,6 +677,10 @@ struct drm_fence_arg {
 #define DRM_BO_FLAG_EXE         (1ULL << 2)
 
 /*
+ * All of the bits related to access mode
+ */
+#define DRM_BO_MASK_ACCESS	(DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE | DRM_BO_FLAG_EXE)
+/*
  * Status flags. Can be read to determine the actual state of a buffer.
  * Can also be set in the buffer mask before validation.
  */
@@ -701,7 +720,14 @@ struct drm_fence_arg {
  */
 #define DRM_BO_FLAG_NO_MOVE     (1ULL << 8)
 
-/* Mask: Make sure the buffer is in cached memory when mapped
+/* Mask: Make sure the buffer is in cached memory when mapped.  In conjunction
+ * with DRM_BO_FLAG_CACHED it also allows the buffer to be bound into the GART
+ * with unsnooped PTEs instead of snooped, by using chipset-specific cache
+ * flushing at bind time.  A better name might be DRM_BO_FLAG_TT_UNSNOOPED,
+ * as the eviction to local memory (TTM unbind) on map is just a side effect
+ * to prevent aggressive cache prefetch from the GPU disturbing the cache
+ * management that the DRM is doing.
+ *
  * Flags: Acknowledge.
  * Buffers allocated with this flag should not be used for suballocators
  * This type may have issues on CPUs with over-aggressive caching
@@ -741,18 +767,49 @@ struct drm_fence_arg {
 #define DRM_BO_FLAG_MEM_PRIV4  (1ULL << 31)
 /* We can add more of these now with a 64-bit flag type */
 
-/* Memory flag mask */
+/*
+ * This is a mask covering all of the memory type flags; easier to just
+ * use a single constant than a bunch of | values. It covers
+ * DRM_BO_FLAG_MEM_LOCAL through DRM_BO_FLAG_MEM_PRIV4
+ */
 #define DRM_BO_MASK_MEM         0x00000000FF000000ULL
-#define DRM_BO_MASK_MEMTYPE     0x00000000FF0800A0ULL
-
+/*
+ * This adds all of the CPU-mapping options in with the memory
+ * type to label all bits which change how the page gets mapped
+ */
+#define DRM_BO_MASK_MEMTYPE     (DRM_BO_MASK_MEM | \
+				 DRM_BO_FLAG_CACHED_MAPPED | \
+				 DRM_BO_FLAG_CACHED | \
+				 DRM_BO_FLAG_MAPPABLE)
+				 
 /* Driver-private flags */
 #define DRM_BO_MASK_DRIVER      0xFFFF000000000000ULL
 
-/* Don't block on validate and map */
+/*
+ * Don't block on validate and map. Instead, return EBUSY.
+ */
 #define DRM_BO_HINT_DONT_BLOCK  0x00000002
-/* Don't place this buffer on the unfenced list.*/
+/*
+ * Don't place this buffer on the unfenced list. This means
+ * that the buffer will not end up having a fence associated
+ * with it as a result of this operation
+ */
 #define DRM_BO_HINT_DONT_FENCE  0x00000004
+/*
+ * Sleep while waiting for the operation to complete.
+ * Without this flag, the kernel will, instead, spin
+ * until this operation has completed. I'm not sure
+ * why you would ever want this, so please always
+ * provide DRM_BO_HINT_WAIT_LAZY to any operation
+ * which may block
+ */
 #define DRM_BO_HINT_WAIT_LAZY   0x00000008
+/*
+ * The client has compute relocations refering to this buffer using the
+ * offset in the presumed_offset field. If that offset ends up matching
+ * where this buffer lands, the kernel is free to skip executing those
+ * relocations
+ */
 #define DRM_BO_HINT_PRESUMED_OFFSET 0x00000010
 
 #define DRM_BO_INIT_MAGIC 0xfe769812
@@ -774,7 +831,7 @@ struct drm_bo_info_req {
 };
 
 struct drm_bo_create_req {
-	uint64_t mask;
+	uint64_t flags;
 	uint64_t size;
 	uint64_t buffer_start;
 	unsigned int hint;
@@ -790,7 +847,7 @@ struct drm_bo_create_req {
 
 struct drm_bo_info_rep {
 	uint64_t flags;
-	uint64_t mask;
+	uint64_t proposed_flags;
 	uint64_t size;
 	uint64_t offset;
 	uint64_t arg_handle;
@@ -1046,6 +1103,7 @@ struct drm_mode_mode_cmd {
 #define DRM_IOCTL_GET_CLIENT            DRM_IOWR(0x05, struct drm_client)
 #define DRM_IOCTL_GET_STATS             DRM_IOR( 0x06, struct drm_stats)
 #define DRM_IOCTL_SET_VERSION		DRM_IOWR(0x07, struct drm_set_version)
+#define DRM_IOCTL_MODESET_CTL           DRM_IOW(0x08, struct drm_modeset_ctl)
 
 #define DRM_IOCTL_SET_UNIQUE		DRM_IOW( 0x10, struct drm_unique)
 #define DRM_IOCTL_AUTH_MAGIC		DRM_IOW( 0x11, struct drm_auth)
@@ -1118,7 +1176,6 @@ struct drm_mode_mode_cmd {
 #define DRM_IOCTL_BO_WAIT_IDLE          DRM_IOWR(0xd5, struct drm_bo_map_wait_idle_arg)
 #define DRM_IOCTL_BO_VERSION          DRM_IOR(0xd6, struct drm_bo_version_arg)
 
-
 #define DRM_IOCTL_MODE_GETRESOURCES     DRM_IOWR(0xA0, struct drm_mode_card_res)
 #define DRM_IOCTL_MODE_GETCRTC          DRM_IOWR(0xA1, struct drm_mode_crtc)
 #define DRM_IOCTL_MODE_GETOUTPUT        DRM_IOWR(0xA2, struct drm_mode_get_output)
@@ -1133,6 +1190,7 @@ struct drm_mode_mode_cmd {
 #define DRM_IOCTL_MODE_DETACHMODE      DRM_IOWR(0xAA, struct drm_mode_mode_cmd)
 
 #define DRM_IOCTL_MODE_GETPROPERTY     DRM_IOWR(0xAB, struct drm_mode_get_property)
+
 /*@}*/
 
 /**
diff --git a/shared-core/drm_pciids.txt b/shared-core/drm_pciids.txt
index a76413d4..7ef34dfd 100644
--- a/shared-core/drm_pciids.txt
+++ b/shared-core/drm_pciids.txt
@@ -139,6 +139,101 @@
 0x1002 0x5e4c CHIP_RV410|RADEON_NEW_MEMMAP "ATI Radeon RV410 X700 SE"
 0x1002 0x5e4d CHIP_RV410|RADEON_NEW_MEMMAP "ATI Radeon RV410 X700"
 0x1002 0x5e4f CHIP_RV410|RADEON_NEW_MEMMAP "ATI Radeon RV410 X700 SE"
+0x1002 0x7100 CHIP_R520|RADEON_NEW_MEMMAP "ATI Radeon X1800"
+0x1002 0x7101 CHIP_R520|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1800 XT"
+0x1002 0x7102 CHIP_R520|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1800"
+0x1002 0x7103 CHIP_R520|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility FireGL V7200"
+0x1002 0x7104 CHIP_R520|RADEON_NEW_MEMMAP "ATI FireGL V7200"
+0x1002 0x7105 CHIP_R520|RADEON_NEW_MEMMAP "ATI FireGL V5300"
+0x1002 0x7106 CHIP_R520|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility FireGL V7100"
+0x1002 0x7108 CHIP_R520|RADEON_NEW_MEMMAP "ATI Radeon X1800"
+0x1002 0x7109 CHIP_R520|RADEON_NEW_MEMMAP "ATI Radeon X1800"
+0x1002 0x710A CHIP_R520|RADEON_NEW_MEMMAP "ATI Radeon X1800"
+0x1002 0x710B CHIP_R520|RADEON_NEW_MEMMAP "ATI Radeon X1800"
+0x1002 0x710C CHIP_R520|RADEON_NEW_MEMMAP "ATI Radeon X1800"
+0x1002 0x710E CHIP_R520|RADEON_NEW_MEMMAP "ATI FireGL V7300"
+0x1002 0x710F CHIP_R520|RADEON_NEW_MEMMAP "ATI FireGL V7350"
+0x1002 0x7140 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1600"
+0x1002 0x7141 CHIP_RV515|RADEON_NEW_MEMMAP "ATI RV505"
+0x1002 0x7142 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1300/X1550"
+0x1002 0x7143 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1550"
+0x1002 0x7144 CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI M54-GL"
+0x1002 0x7145 CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1400"
+0x1002 0x7146 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1300/X1550"
+0x1002 0x7147 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1550 64-bit"
+0x1002 0x7149 CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1300"
+0x1002 0x714A CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1300"
+0x1002 0x714B CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1300"
+0x1002 0x714C CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1300"
+0x1002 0x714D CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1300"
+0x1002 0x714E CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1300"
+0x1002 0x714F CHIP_RV515|RADEON_NEW_MEMMAP "ATI RV505"
+0x1002 0x7151 CHIP_RV515|RADEON_NEW_MEMMAP "ATI RV505"
+0x1002 0x7152 CHIP_RV515|RADEON_NEW_MEMMAP "ATI FireGL V3300"
+0x1002 0x7153 CHIP_RV515|RADEON_NEW_MEMMAP "ATI FireGL V3350"
+0x1002 0x715E CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1300"
+0x1002 0x715F CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1550 64-bit"
+0x1002 0x7180 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1300/X1550"
+0x1002 0x7181 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1600"
+0x1002 0x7183 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1300/X1550"
+0x1002 0x7186 CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1450"
+0x1002 0x7187 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1300/X1550"
+0x1002 0x7188 CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X2300"
+0x1002 0x718A CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X2300"
+0x1002 0x718B CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1350"
+0x1002 0x718C CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1350"
+0x1002 0x718D CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1450"
+0x1002 0x718F CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1300"
+0x1002 0x7193 CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1550"
+0x1002 0x7196 CHIP_RV515|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1350"
+0x1002 0x719B CHIP_RV515|RADEON_NEW_MEMMAP "ATI FireMV 2250"
+0x1002 0x719F CHIP_RV515|RADEON_NEW_MEMMAP "ATI Radeon X1550 64-bit"
+0x1002 0x71C0 CHIP_RV530|RADEON_NEW_MEMMAP "ATI Radeon X1600"
+0x1002 0x71C1 CHIP_RV530|RADEON_NEW_MEMMAP "ATI Radeon X1650"
+0x1002 0x71C2 CHIP_RV530|RADEON_NEW_MEMMAP "ATI Radeon X1600"
+0x1002 0x71C3 CHIP_RV530|RADEON_NEW_MEMMAP "ATI Radeon X1600"
+0x1002 0x71C4 CHIP_RV530|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility FireGL V5200"
+0x1002 0x71C5 CHIP_RV530|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1600"
+0x1002 0x71C6 CHIP_RV530|RADEON_NEW_MEMMAP "ATI Radeon X1650"
+0x1002 0x71C7 CHIP_RV530|RADEON_NEW_MEMMAP "ATI Radeon X1650"
+0x1002 0x71CD CHIP_RV530|RADEON_NEW_MEMMAP "ATI Radeon X1600"
+0x1002 0x71CE CHIP_RV530|RADEON_NEW_MEMMAP "ATI Radeon X1300 XT/X1600 Pro"
+0x1002 0x71D2 CHIP_RV530|RADEON_NEW_MEMMAP "ATI FireGL V3400"
+0x1002 0x71D4 CHIP_RV530|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility FireGL V5250"
+0x1002 0x71D5 CHIP_RV530|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1700"
+0x1002 0x71D6 CHIP_RV530|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1700 XT"
+0x1002 0x71DA CHIP_RV530|RADEON_NEW_MEMMAP "ATI FireGL V5200"
+0x1002 0x71DE CHIP_RV530|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1700"
+0x1002 0x7200 CHIP_RV530|RADEON_NEW_MEMMAP "ATI  Radeon X2300HD"
+0x1002 0x7210 CHIP_RV530|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon HD 2300"
+0x1002 0x7211 CHIP_RV530|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon HD 2300"
+0x1002 0x7240 CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1950"
+0x1002 0x7243 CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x7244 CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1950"
+0x1002 0x7245 CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x7246 CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x7247 CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x7248 CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x7249 CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x724A CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x724B CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x724C CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x724D CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x724E CHIP_R580|RADEON_NEW_MEMMAP "ATI AMD Stream Processor"
+0x1002 0x724F CHIP_R580|RADEON_NEW_MEMMAP "ATI Radeon X1900"
+0x1002 0x7280 CHIP_RV570|RADEON_NEW_MEMMAP "ATI Radeon X1950"
+0x1002 0x7281 CHIP_RV560|RADEON_NEW_MEMMAP "ATI RV560"
+0x1002 0x7283 CHIP_RV560|RADEON_NEW_MEMMAP "ATI RV560"
+0x1002 0x7284 CHIP_R580|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Mobility Radeon X1900"
+0x1002 0x7287 CHIP_RV560|RADEON_NEW_MEMMAP "ATI RV560"
+0x1002 0x7288 CHIP_RV570|RADEON_NEW_MEMMAP "ATI Radeon X1950 GT"
+0x1002 0x7289 CHIP_RV570|RADEON_NEW_MEMMAP "ATI RV570"
+0x1002 0x728B CHIP_RV570|RADEON_NEW_MEMMAP "ATI RV570"
+0x1002 0x728C CHIP_RV570|RADEON_NEW_MEMMAP "ATI ATI FireGL V7400"
+0x1002 0x7290 CHIP_RV560|RADEON_NEW_MEMMAP "ATI RV560"
+0x1002 0x7291 CHIP_RV560|RADEON_NEW_MEMMAP "ATI Radeon X1650"
+0x1002 0x7293 CHIP_RV560|RADEON_NEW_MEMMAP "ATI Radeon X1650"
+0x1002 0x7297 CHIP_RV560|RADEON_NEW_MEMMAP "ATI RV560"
 0x1002 0x7834 CHIP_RS300|RADEON_IS_IGP|RADEON_NEW_MEMMAP "ATI Radeon RS350 9000/9100 IGP"
 0x1002 0x7835 CHIP_RS300|RADEON_IS_IGP|RADEON_IS_MOBILITY|RADEON_NEW_MEMMAP "ATI Radeon RS350 Mobility IGP"
 
@@ -236,6 +331,7 @@
 0x1106 0x3343 0 "VIA P4M890"
 0x1106 0x3230 VIA_DX9_0 "VIA K8M890"
 0x1106 0x3157 VIA_PRO_GROUP_A "VIA CX700"
+0x1106 0x3371 VIA_DX9_0 "VIA P4M900 / VN896"
 
 [i810]
 0x8086 0x7121 0 "Intel i810 GMCH"
@@ -285,6 +381,7 @@
 0x8086 0x3582 CHIP_I8XX "Intel i852GM/i855GM GMCH"
 0x8086 0x2572 CHIP_I8XX "Intel i865G GMCH"
 0x8086 0x2582 CHIP_I9XX|CHIP_I915 "Intel i915G"
+0x8086 0x258a CHIP_I9XX|CHIP_I915 "Intel E7221 (i915)"
 0x8086 0x2592 CHIP_I9XX|CHIP_I915 "Intel i915GM"
 0x8086 0x2772 CHIP_I9XX|CHIP_I915 "Intel i945G"
 0x8086 0x27A2 CHIP_I9XX|CHIP_I915 "Intel i945GM"
@@ -298,6 +395,7 @@
 0x8086 0x29C2 CHIP_I9XX|CHIP_I915 "Intel G33"
 0x8086 0x29B2 CHIP_I9XX|CHIP_I915 "Intel Q35"
 0x8086 0x29D2 CHIP_I9XX|CHIP_I915 "Intel Q33"
+0x8086 0x2A42 CHIP_I9XX|CHIP_I965 "Intel Integrated Graphics Device"
 
 [imagine]
 0x105d 0x2309 IMAGINE_128 "Imagine 128"
diff --git a/shared-core/i915_dma.c b/shared-core/i915_dma.c
index 4c1a07ac..66d1b13d 100644
--- a/shared-core/i915_dma.c
+++ b/shared-core/i915_dma.c
@@ -718,9 +718,15 @@ int i915_apply_reloc(struct drm_file *file_priv, int num_buffers,
 	unsigned index;
 	unsigned long new_cmd_offset;
 	u32 val;
-	int ret;
+	int ret, i;
+	int buf_index = -1;
+
+	for (i = 0; i <= num_buffers; i++)
+		if (buffers[i].buffer)
+			if (reloc[2] == buffers[i].buffer->base.hash.key)
+				buf_index = i;
 
-	if (reloc[2] >= num_buffers) {
+	if (buf_index == -1) {
 		DRM_ERROR("Illegal relocation buffer %08X\n", reloc[2]);
 		return -EINVAL;
 	}
@@ -729,7 +735,7 @@ int i915_apply_reloc(struct drm_file *file_priv, int num_buffers,
 	 * Short-circuit relocations that were correctly
 	 * guessed by the client
 	 */
-	if (buffers[reloc[2]].presumed_offset_correct && !DRM_DEBUG_RELOCATION)
+	if (buffers[buf_index].presumed_offset_correct && !DRM_DEBUG_RELOCATION)
 		return 0;
 
 	new_cmd_offset = reloc[0];
@@ -756,17 +762,17 @@ int i915_apply_reloc(struct drm_file *file_priv, int num_buffers,
 		relocatee->page_offset = (relocatee->offset & PAGE_MASK);
 	}
 
-	val = buffers[reloc[2]].buffer->offset;
+	val = buffers[buf_index].buffer->offset;
 	index = (reloc[0] - relocatee->page_offset) >> 2;
 
 	/* add in validate */
 	val = val + reloc[1];
 
 	if (DRM_DEBUG_RELOCATION) {
-		if (buffers[reloc[2]].presumed_offset_correct &&
+		if (buffers[buf_index].presumed_offset_correct &&
 		    relocatee->data_page[index] != val) {
 			DRM_DEBUG ("Relocation mismatch source %d target %d buffer %d user %08x kernel %08x\n",
-				   reloc[0], reloc[1], reloc[2], relocatee->data_page[index], val);
+				   reloc[0], reloc[1], buf_index, relocatee->data_page[index], val);
 		}
 	}
 	relocatee->data_page[index] = val;
@@ -775,94 +781,79 @@ int i915_apply_reloc(struct drm_file *file_priv, int num_buffers,
 
 int i915_process_relocs(struct drm_file *file_priv,
 			uint32_t buf_handle,
-			uint32_t *reloc_buf_handle,
+			uint32_t __user **reloc_user_ptr,
 			struct i915_relocatee_info *relocatee,
 			struct drm_i915_validate_buffer *buffers,
 			uint32_t num_buffers)
 {
-	struct drm_device *dev = file_priv->minor->dev;
-	struct drm_buffer_object *reloc_list_object;
-	uint32_t cur_handle = *reloc_buf_handle;
-	uint32_t *reloc_page;
-	int ret, reloc_is_iomem, reloc_stride;
-	uint32_t num_relocs, reloc_offset, reloc_end, reloc_page_offset, next_offset, cur_offset;
-	struct drm_bo_kmap_obj reloc_kmap;
-
-	memset(&reloc_kmap, 0, sizeof(reloc_kmap));
-
-	mutex_lock(&dev->struct_mutex);
-	reloc_list_object = drm_lookup_buffer_object(file_priv, cur_handle, 1);
-	mutex_unlock(&dev->struct_mutex);
-	if (!reloc_list_object)
-		return -EINVAL;
+	int ret, reloc_stride;
+	uint32_t cur_offset;
+	uint32_t reloc_count;
+	uint32_t reloc_type;
+	uint32_t reloc_buf_size;
+	uint32_t *reloc_buf = NULL;
+	int i;
 
-	ret = drm_bo_kmap(reloc_list_object, 0, 1, &reloc_kmap);
+	/* do a copy from user from the user ptr */
+	ret = get_user(reloc_count, *reloc_user_ptr);
 	if (ret) {
 		DRM_ERROR("Could not map relocation buffer.\n");
 		goto out;
 	}
 
-	reloc_page = drm_bmo_virtual(&reloc_kmap, &reloc_is_iomem);
-	num_relocs = reloc_page[0] & 0xffff;
+	ret = get_user(reloc_type, (*reloc_user_ptr)+1);
+	if (ret) {
+		DRM_ERROR("Could not map relocation buffer.\n");
+		goto out;
+	}
 
-	if ((reloc_page[0] >> 16) & 0xffff) {
+	if (reloc_type != 0) {
 		DRM_ERROR("Unsupported relocation type requested\n");
+		ret = -EINVAL;
 		goto out;
 	}
 
-	/* get next relocate buffer handle */
-	*reloc_buf_handle = reloc_page[1];
-	reloc_stride = I915_RELOC0_STRIDE * sizeof(uint32_t); /* may be different for other types of relocs */
-
-	DRM_DEBUG("num relocs is %d, next is %08X\n", num_relocs, reloc_page[1]);
-
-	reloc_page_offset = 0;
-	reloc_offset = I915_RELOC_HEADER * sizeof(uint32_t);
-	reloc_end = reloc_offset + (num_relocs * reloc_stride);
-
-	do {
-		next_offset = drm_bo_offset_end(reloc_offset, reloc_end);
+	reloc_buf_size = (I915_RELOC_HEADER + (reloc_count * I915_RELOC0_STRIDE)) * sizeof(uint32_t);
+	reloc_buf = kmalloc(reloc_buf_size, GFP_KERNEL);
+	if (!reloc_buf) {
+		DRM_ERROR("Out of memory for reloc buffer\n");
+		ret = -ENOMEM;
+		goto out;
+	}
 
-		do {
-			cur_offset = ((reloc_offset + reloc_page_offset) & ~PAGE_MASK) / sizeof(uint32_t);
-			ret = i915_apply_reloc(file_priv, num_buffers,
-					 buffers, relocatee, &reloc_page[cur_offset]);
-			if (ret)
-				goto out;
+	if (copy_from_user(reloc_buf, *reloc_user_ptr, reloc_buf_size)) {
+		ret = -EFAULT;
+		goto out;
+	}
 
-			reloc_offset += reloc_stride;
-		} while (reloc_offset < next_offset);
+	/* get next relocate buffer handle */
+	*reloc_user_ptr = (uint32_t *)*(unsigned long *)&reloc_buf[2];
 
-		drm_bo_kunmap(&reloc_kmap);
+	reloc_stride = I915_RELOC0_STRIDE * sizeof(uint32_t); /* may be different for other types of relocs */
 
-		reloc_offset = next_offset;
-		if (reloc_offset != reloc_end) {
-			ret = drm_bo_kmap(reloc_list_object, reloc_offset >> PAGE_SHIFT, 1, &reloc_kmap);
-			if (ret) {
-				DRM_ERROR("Could not map relocation buffer.\n");
-				goto out;
-			}
+	DRM_DEBUG("num relocs is %d, next is %p\n", reloc_count, *reloc_user_ptr);
 
-			reloc_page = drm_bmo_virtual(&reloc_kmap, &reloc_is_iomem);
-			reloc_page_offset = reloc_offset & ~PAGE_MASK;
-		}
+	for (i = 0; i < reloc_count; i++) {
+		cur_offset = I915_RELOC_HEADER + (i * I915_RELOC0_STRIDE);
+		  
+		ret = i915_apply_reloc(file_priv, num_buffers, buffers,
+				       relocatee, reloc_buf + cur_offset);
+		if (ret)
+			goto out;
+	}
 
-	} while (reloc_offset != reloc_end);
 out:
+
+	if (reloc_buf)
+		kfree(reloc_buf);
 	drm_bo_kunmap(&relocatee->kmap);
 	relocatee->data_page = NULL;
 
-	drm_bo_kunmap(&reloc_kmap);
-
-	mutex_lock(&dev->struct_mutex);
-	drm_bo_usage_deref_locked(&reloc_list_object);
-	mutex_unlock(&dev->struct_mutex);
-
 	return ret;
 }
 
 static int i915_exec_reloc(struct drm_file *file_priv, drm_handle_t buf_handle,
-			   drm_handle_t buf_reloc_handle,
+			   uint32_t __user *reloc_user_ptr,
 			   struct drm_i915_validate_buffer *buffers,
 			   uint32_t buf_count)
 {
@@ -896,8 +887,8 @@ static int i915_exec_reloc(struct drm_file *file_priv, drm_handle_t buf_handle,
 		goto out_err;
 	}
 
-	while (buf_reloc_handle) {
-		ret = i915_process_relocs(file_priv, buf_handle, &buf_reloc_handle, &relocatee, buffers, buf_count);
+	while (reloc_user_ptr) {
+		ret = i915_process_relocs(file_priv, buf_handle, &reloc_user_ptr, &relocatee, buffers, buf_count);
 		if (ret) {
 			DRM_ERROR("process relocs failed\n");
 			break;
@@ -927,8 +918,8 @@ int i915_validate_buffer_list(struct drm_file *file_priv,
 	int ret = 0;
 	unsigned buf_count = 0;
 	struct drm_device *dev = file_priv->minor->dev;
-	uint32_t buf_reloc_handle, buf_handle;
-
+	uint32_t buf_handle;
+	uint32_t __user *reloc_user_ptr;
 
 	do {
 		if (buf_count >= *num_buffers) {
@@ -963,21 +954,19 @@ int i915_validate_buffer_list(struct drm_file *file_priv,
 		}
 
 		buf_handle = req->bo_req.handle;
-		buf_reloc_handle = arg.reloc_handle;
+		reloc_user_ptr = (uint32_t *)(unsigned long)arg.reloc_ptr;
 
-		if (buf_reloc_handle) {
-			ret = i915_exec_reloc(file_priv, buf_handle, buf_reloc_handle, buffers, buf_count);
+		if (reloc_user_ptr) {
+			ret = i915_exec_reloc(file_priv, buf_handle, reloc_user_ptr, buffers, buf_count);
 			if (ret)
 				goto out_err;
 			DRM_MEMORYBARRIER();
 		}
 
 		rep.ret = drm_bo_handle_validate(file_priv, req->bo_req.handle,
-						 req->bo_req.fence_class,
-						 req->bo_req.flags,
-						 req->bo_req.mask,
+						 req->bo_req.flags, req->bo_req.mask,
 						 req->bo_req.hint,
-						 0,
+						 req->bo_req.fence_class, 0,
 						 &rep.bo_info,
 						 &buffers[buf_count].buffer);
 
@@ -1092,7 +1081,8 @@ static int i915_execbuffer(struct drm_device *dev, void *data,
 	sarea_priv->last_dispatch = READ_BREADCRUMB(dev_priv);
 
 	/* fence */
-	ret = drm_fence_buffer_objects(dev, NULL, 0, NULL, &fence);
+	ret = drm_fence_buffer_objects(dev, NULL, fence_arg->flags, 
+				       NULL, &fence);
 	if (ret)
 		goto out_err0;
 
@@ -1174,7 +1164,7 @@ static int i915_getparam(struct drm_device *dev, void *data,
 	int value;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -1208,7 +1198,7 @@ static int i915_setparam(struct drm_device *dev, void *data,
 	drm_i915_setparam_t *param = data;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -1252,7 +1242,7 @@ static int i915_mmio(struct drm_device *dev, void *data,
 	int i;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -1295,7 +1285,7 @@ static int i915_set_status_page(struct drm_device *dev, void *data,
 	drm_i915_hws_addr_t *hws = data;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 	DRM_DEBUG("set status page addr 0x%08x\n", (u32)hws->addr);
diff --git a/shared-core/i915_drm.h b/shared-core/i915_drm.h
index c5819953..d48d7665 100644
--- a/shared-core/i915_drm.h
+++ b/shared-core/i915_drm.h
@@ -329,9 +329,9 @@ typedef struct drm_i915_hws_addr {
 
 /*
  * Relocation header is 4 uint32_ts
- * 0 - (16-bit relocation type << 16)| 16 bit reloc count
- * 1 - buffer handle for another list of relocs
- * 2-3 - spare.
+ * 0 - 32 bit reloc count
+ * 1 - 32-bit relocation type
+ * 2-3 - 64-bit user buffer handle ptr for another list of relocs.
  */
 #define I915_RELOC_HEADER 4
 
@@ -339,7 +339,7 @@ typedef struct drm_i915_hws_addr {
  * type 0 relocation has 4-uint32_t stride
  * 0 - offset into buffer
  * 1 - delta to add in
- * 2 - index into buffer list
+ * 2 - buffer handle
  * 3 - reserved (for optimisations later).
  */
 #define I915_RELOC_TYPE_0 0
@@ -347,7 +347,7 @@ typedef struct drm_i915_hws_addr {
 
 struct drm_i915_op_arg {
 	uint64_t next;
-	uint32_t reloc_handle;
+	uint64_t reloc_ptr;
 	int handled;
 	union {
 		struct drm_bo_op_req req;
diff --git a/shared-core/i915_drv.h b/shared-core/i915_drv.h
index 45b7786a..49e23ac3 100644
--- a/shared-core/i915_drv.h
+++ b/shared-core/i915_drv.h
@@ -271,23 +271,23 @@ extern int i915_irq_emit(struct drm_device *dev, void *data,
 extern int i915_irq_wait(struct drm_device *dev, void *data,
 			 struct drm_file *file_priv);
 
-extern void i915_driver_wait_next_vblank(struct drm_device *dev, int pipe);
-extern int i915_driver_vblank_wait(struct drm_device *dev, unsigned int *sequence);
-extern int i915_driver_vblank_wait2(struct drm_device *dev, unsigned int *sequence);
 extern irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS);
 extern void i915_driver_irq_preinstall(struct drm_device * dev);
-extern void i915_driver_irq_postinstall(struct drm_device * dev);
+extern int i915_driver_irq_postinstall(struct drm_device * dev);
 extern void i915_driver_irq_uninstall(struct drm_device * dev);
 extern int i915_vblank_pipe_set(struct drm_device *dev, void *data,
 				struct drm_file *file_priv);
 extern int i915_vblank_pipe_get(struct drm_device *dev, void *data,
 				struct drm_file *file_priv);
 extern int i915_emit_irq(struct drm_device * dev);
-extern void i915_user_irq_on(struct drm_i915_private *dev_priv);
-extern void i915_user_irq_off(struct drm_i915_private *dev_priv);
 extern void i915_enable_interrupt (struct drm_device *dev);
+extern int i915_enable_vblank(struct drm_device *dev, int crtc);
+extern void i915_disable_vblank(struct drm_device *dev, int crtc);
+extern u32 i915_get_vblank_counter(struct drm_device *dev, int crtc);
 extern int i915_vblank_swap(struct drm_device *dev, void *data,
 			    struct drm_file *file_priv);
+extern void i915_user_irq_on(struct drm_i915_private *dev_priv);
+extern void i915_user_irq_off(struct drm_i915_private *dev_priv);
 
 /* i915_mem.c */
 extern int i915_mem_alloc(struct drm_device *dev, void *data,
@@ -318,12 +318,12 @@ extern int i915_fence_has_irq(struct drm_device *dev, uint32_t class, uint32_t f
 #ifdef I915_HAVE_BUFFER
 /* i915_buffer.c */
 extern struct drm_ttm_backend *i915_create_ttm_backend_entry(struct drm_device *dev);
-extern int i915_fence_types(struct drm_buffer_object *bo, uint32_t *fclass,
-			    uint32_t *type);
+extern int i915_fence_type(struct drm_buffer_object *bo, uint32_t *fclass,
+			   uint32_t *type);
 extern int i915_invalidate_caches(struct drm_device *dev, uint64_t buffer_flags);
 extern int i915_init_mem_type(struct drm_device *dev, uint32_t type,
 			       struct drm_mem_type_manager *man);
-extern uint32_t i915_evict_mask(struct drm_buffer_object *bo);
+extern uint64_t i915_evict_flags(struct drm_buffer_object *bo);
 extern int i915_move(struct drm_buffer_object *bo, int evict,
 		int no_wait, struct drm_bo_mem_reg *new_mem);
 void i915_flush_ttm(struct drm_ttm *ttm);
@@ -354,8 +354,8 @@ extern void intel_modeset_cleanup(struct drm_device *dev);
 
 #define BEGIN_LP_RING(n) do {				\
 	if (I915_VERBOSE)				\
-		DRM_DEBUG("BEGIN_LP_RING(%d) in %s\n",	\
-	                         (n), __FUNCTION__);           \
+		DRM_DEBUG("BEGIN_LP_RING(%d)\n",	\
+	                         (n));		        \
 	if (dev_priv->ring.space < (n)*4)                      \
 		i915_wait_ring(dev, (n)*4, __FUNCTION__);      \
 	outcount = 0;					\
@@ -516,8 +516,41 @@ extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller);
 #define I915REG_INT_ENABLE_R	0x020a0
 #define I915REG_INSTPM	        0x020c0
 
+#define PIPEADSL		0x70000
+#define PIPEBDSL		0x71000
+
 #define I915REG_PIPEASTAT	0x70024
 #define I915REG_PIPEBSTAT	0x71024
+/*
+ * The two pipe frame counter registers are not synchronized, so
+ * reading a stable value is somewhat tricky. The following code 
+ * should work:
+ *
+ *  do {
+ *    high1 = ((INREG(PIPEAFRAMEHIGH) & PIPE_FRAME_HIGH_MASK) >>
+ *             PIPE_FRAME_HIGH_SHIFT;
+ *    low1 =  ((INREG(PIPEAFRAMEPIXEL) & PIPE_FRAME_LOW_MASK) >>
+ *             PIPE_FRAME_LOW_SHIFT);
+ *    high2 = ((INREG(PIPEAFRAMEHIGH) & PIPE_FRAME_HIGH_MASK) >>
+ *             PIPE_FRAME_HIGH_SHIFT);
+ *  } while (high1 != high2);
+ *  frame = (high1 << 8) | low1;
+ */
+#define PIPEAFRAMEHIGH          0x70040
+#define PIPEBFRAMEHIGH		0x71040
+#define PIPE_FRAME_HIGH_MASK    0x0000ffff
+#define PIPE_FRAME_HIGH_SHIFT   0
+#define PIPEAFRAMEPIXEL         0x70044
+#define PIPEBFRAMEPIXEL		0x71044
+
+#define PIPE_FRAME_LOW_MASK     0xff000000
+#define PIPE_FRAME_LOW_SHIFT    24
+/*
+ * Pixel within the current frame is counted in the PIPEAFRAMEPIXEL register
+ * and is 24 bits wide.
+ */
+#define PIPE_PIXEL_MASK         0x00ffffff
+#define PIPE_PIXEL_SHIFT        0
 
 #define I915_VBLANK_INTERRUPT_ENABLE	(1UL<<17)
 #define I915_VBLANK_CLEAR		(1UL<<1)
@@ -871,6 +904,14 @@ extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller);
 #define BCLRPAT_B	0x61020
 #define VSYNCSHIFT_B	0x61028
 
+#define HACTIVE_MASK	0x00000fff
+#define VTOTAL_MASK	0x00001fff
+#define VTOTAL_SHIFT	16
+#define VACTIVE_MASK	0x00000fff
+#define VBLANK_END_MASK	0x00001fff
+#define VBLANK_END_SHIFT 16
+#define VBLANK_START_MASK 0x00001fff
+
 #define PP_STATUS	0x61200
 # define PP_ON					(1 << 31)
 /**
@@ -1274,7 +1315,7 @@ extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller);
 #define IS_I855(dev) ((dev)->pci_device == 0x3582)
 #define IS_I865G(dev) ((dev)->pci_device == 0x2572)
 
-#define IS_I915G(dev) (dev->pci_device == 0x2582)/* || dev->pci_device == PCI_DEVICE_ID_INTELPCI_CHIP_E7221_G)*/
+#define IS_I915G(dev) ((dev)->pci_device == 0x2582 || (dev)->pci_device == 0x258a)
 #define IS_I915GM(dev) ((dev)->pci_device == 0x2592)
 #define IS_I945G(dev) ((dev)->pci_device == 0x2772)
 #define IS_I945GM(dev) ((dev)->pci_device == 0x27A2)
@@ -1284,10 +1325,13 @@ extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller);
 		       (dev)->pci_device == 0x2992 || \
 		       (dev)->pci_device == 0x29A2 || \
 		       (dev)->pci_device == 0x2A02 || \
-		       (dev)->pci_device == 0x2A12)
+		       (dev)->pci_device == 0x2A12 || \
+		       (dev)->pci_device == 0x2A42)
 
 #define IS_I965GM(dev) ((dev)->pci_device == 0x2A02)
 
+#define IS_IGD_GM(dev) ((dev)->pci_device == 0x2A42)
+
 #define IS_G33(dev)    ((dev)->pci_device == 0x29C2 ||	\
 			(dev)->pci_device == 0x29B2 ||	\
 			(dev)->pci_device == 0x29D2)
@@ -1296,7 +1340,7 @@ extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller);
 		      IS_I945GM(dev) || IS_I965G(dev) || IS_G33(dev))
 
 #define IS_MOBILE(dev) (IS_I830(dev) || IS_I85X(dev) || IS_I915GM(dev) || \
-			IS_I945GM(dev) || IS_I965GM(dev))
+			IS_I945GM(dev) || IS_I965GM(dev) || IS_IGD_GM(dev))
 
 #define PRIMARY_RINGBUFFER_SIZE         (128*1024)
 
diff --git a/shared-core/i915_irq.c b/shared-core/i915_irq.c
index ac5361f2..9b391b75 100644
--- a/shared-core/i915_irq.c
+++ b/shared-core/i915_irq.c
@@ -45,9 +45,9 @@
  * @dev: DRM device
  * @plane: plane to look for
  *
- * We need to get the pipe associated with a given plane to correctly perform
- * vblank driven swapping, and they may not always be equal.  So look up the
- * pipe associated with @plane here.
+ * The Intel Mesa & 2D drivers call the vblank routines with a plane number
+ * rather than a pipe number, since they may not always be equal.  This routine
+ * maps the given @plane back to a pipe number.
  */
 static int
 i915_get_pipe(struct drm_device *dev, int plane)
@@ -61,6 +61,44 @@ i915_get_pipe(struct drm_device *dev, int plane)
 }
 
 /**
+ * i915_get_plane - return the the plane associated with a given pipe
+ * @dev: DRM device
+ * @pipe: pipe to look for
+ *
+ * The Intel Mesa & 2D drivers call the vblank routines with a plane number
+ * rather than a plane number, since they may not always be equal.  This routine
+ * maps the given @pipe back to a plane number.
+ */
+static int
+i915_get_plane(struct drm_device *dev, int pipe)
+{
+	if (i915_get_pipe(dev, 0) == pipe)
+		return 0;
+	return 1;
+}
+
+/**
+ * i915_pipe_enabled - check if a pipe is enabled
+ * @dev: DRM device
+ * @pipe: pipe to check
+ *
+ * Reading certain registers when the pipe is disabled can hang the chip.
+ * Use this routine to make sure the PLL is running and the pipe is active
+ * before reading such registers if unsure.
+ */
+static int
+i915_pipe_enabled(struct drm_device *dev, int pipe)
+{
+	struct drm_i915_private *dev_priv = (struct drm_i915_private *) dev->dev_private;
+	unsigned long pipeconf = pipe ? PIPEBCONF : PIPEACONF;
+
+	if (I915_READ(pipeconf) & PIPEACONF_ENABLE)
+		return 1;
+
+	return 0;
+}
+
+/**
  * Emit a synchronous flip.
  *
  * This function must be called with the drawable spinlock held.
@@ -117,8 +155,7 @@ static void i915_vblank_tasklet(struct drm_device *dev)
 	struct drm_i915_private *dev_priv = (struct drm_i915_private *) dev->dev_private;
 	struct list_head *list, *tmp, hits, *hit;
 	int nhits, nrects, slice[2], upper[2], lower[2], i, num_pages;
-	unsigned counter[2] = { atomic_read(&dev->vbl_received),
-				atomic_read(&dev->vbl_received2) };
+	unsigned counter[2];
 	struct drm_drawable_info *drw;
 	struct drm_i915_sarea *sarea_priv = dev_priv->sarea_priv;
 	u32 cpp = dev_priv->cpp,  offsets[3];
@@ -130,6 +167,9 @@ static void i915_vblank_tasklet(struct drm_device *dev)
 			  (cpp << 23) | (1 << 24);
 	RING_LOCALS;
 
+	counter[0] = drm_vblank_count(dev, 0);
+	counter[1] = drm_vblank_count(dev, 1);
+
 	DRM_DEBUG("\n");
 
 	INIT_LIST_HEAD(&hits);
@@ -154,6 +194,7 @@ static void i915_vblank_tasklet(struct drm_device *dev)
 
 		list_del(list);
 		dev_priv->swaps_pending--;
+		drm_vblank_put(dev, pipe);
 
 		DRM_SPINUNLOCK(&dev_priv->swaps_lock);
 		DRM_SPINLOCK(&dev->drw_lock);
@@ -303,6 +344,73 @@ static void i915_vblank_tasklet(struct drm_device *dev)
 		drm_free(swap_hit, sizeof(*swap_hit), DRM_MEM_DRIVER);
 	}
 }
+#if 0
+static int i915_in_vblank(struct drm_device *dev, int pipe)
+{
+	struct drm_i915_private *dev_priv = (struct drm_i915_private *) dev->dev_private;
+	unsigned long pipedsl, vblank, vtotal;
+	unsigned long vbl_start, vbl_end, cur_line;
+
+	pipedsl = pipe ? PIPEBDSL : PIPEADSL;
+	vblank = pipe ? VBLANK_B : VBLANK_A;
+	vtotal = pipe ? VTOTAL_B : VTOTAL_A;
+
+	vbl_start = I915_READ(vblank) & VBLANK_START_MASK;
+	vbl_end = (I915_READ(vblank) >> VBLANK_END_SHIFT) & VBLANK_END_MASK;
+
+	cur_line = I915_READ(pipedsl);
+
+	if (cur_line >= vbl_start)
+		return 1;
+
+	return 0;
+}
+#endif
+u32 i915_get_vblank_counter(struct drm_device *dev, int plane)
+{
+	struct drm_i915_private *dev_priv = (struct drm_i915_private *) dev->dev_private;
+	unsigned long high_frame;
+	unsigned long low_frame;
+	u32 high1, high2, low, count;
+	int pipe;
+
+	pipe = i915_get_pipe(dev, plane);
+	high_frame = pipe ? PIPEBFRAMEHIGH : PIPEAFRAMEHIGH;
+	low_frame = pipe ? PIPEBFRAMEPIXEL : PIPEAFRAMEPIXEL;
+
+	if (!i915_pipe_enabled(dev, pipe)) {
+	    printk(KERN_ERR "trying to get vblank count for disabled "
+		   "pipe %d\n", pipe);
+	    return 0;
+	}
+
+	/*
+	 * High & low register fields aren't synchronized, so make sure
+	 * we get a low value that's stable across two reads of the high
+	 * register.
+	 */
+	do {
+		high1 = ((I915_READ(high_frame) & PIPE_FRAME_HIGH_MASK) >>
+			 PIPE_FRAME_HIGH_SHIFT);
+		low =  ((I915_READ(low_frame) & PIPE_FRAME_LOW_MASK) >>
+			PIPE_FRAME_LOW_SHIFT);
+		high2 = ((I915_READ(high_frame) & PIPE_FRAME_HIGH_MASK) >>
+			 PIPE_FRAME_HIGH_SHIFT);
+	} while (high1 != high2);
+
+	count = (high1 << 8) | low;
+
+	/*
+	 * If we're in the middle of the vblank period, the
+	 * above regs won't have been updated yet, so return
+	 * an incremented count to stay accurate
+	 */
+#if 0
+	if (i915_in_vblank(dev, pipe))
+		count++;
+#endif
+	return count;
+}
 
 #define HOTPLUG_CMD_CRT 1
 #define HOTPLUG_CMD_SDVOB 4
@@ -445,11 +553,10 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 	pipeb_stats = I915_READ(I915REG_PIPEBSTAT);
 
 	/* On i8xx hw the IIR and IER are 16bit on i9xx its 32bit */
-	if (IS_I9XX(dev)) {
+	if (IS_I9XX(dev))
 		temp = I915_READ(I915REG_INT_IDENTITY_R);
-	} else {
+	else
 		temp = I915_READ16(I915REG_INT_IDENTITY_R);
-	}
 
 	temp2 = temp;
 	temp &= (dev_priv->irq_enable_reg | USER_INT_FLAG);
@@ -464,7 +571,7 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 	}
 #else
 #if 0
-	DRM_DEBUG("%s flag=%08x\n", __FUNCTION__, temp);
+	DRM_DEBUG("flag=%08x\n", temp);
 #endif
 #endif
 
@@ -479,8 +586,32 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 		(void) I915_READ16(I915REG_INT_IDENTITY_R);
 	}
 
+	/*
+	 * Clear the PIPE(A|B)STAT regs before the IIR otherwise
+	 * we may get extra interrupts.
+	 */
+	if (temp & VSYNC_PIPEA_FLAG) {
+		drm_handle_vblank(dev, i915_get_plane(dev, 0));
+		I915_WRITE(I915REG_PIPEASTAT,
+			   pipea_stats | I915_VBLANK_INTERRUPT_ENABLE |
+			   I915_VBLANK_CLEAR);
+	}
+
+	if (temp & VSYNC_PIPEB_FLAG) {
+		drm_handle_vblank(dev, i915_get_plane(dev, 1));
+		I915_WRITE(I915REG_PIPEBSTAT,
+			   pipeb_stats | I915_VBLANK_INTERRUPT_ENABLE |
+			   I915_VBLANK_CLEAR);
+	}
+
+	I915_WRITE16(I915REG_INT_IDENTITY_R, temp);
+	(void) I915_READ16(I915REG_INT_IDENTITY_R); /* Flush posted write */
+
 	DRM_READMEMORYBARRIER();
 
+	temp &= (dev_priv->irq_enable_reg | USER_INT_FLAG | VSYNC_PIPEA_FLAG |
+		 VSYNC_PIPEB_FLAG);
+
 	dev_priv->sarea_priv->last_dispatch = READ_BREADCRUMB(dev_priv);
 
 	if (temp & USER_INT_FLAG) {
@@ -491,32 +622,8 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 	}
 
 	if (temp & (VSYNC_PIPEA_FLAG | VSYNC_PIPEB_FLAG)) {
-		int vblank_pipe = dev_priv->vblank_pipe;
-
-		if ((vblank_pipe &
-		     (DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B))
-		    == (DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B)) {
-			if (temp & VSYNC_PIPEA_FLAG)
-				atomic_inc(&dev->vbl_received);
-			if (temp & VSYNC_PIPEB_FLAG)
-				atomic_inc(&dev->vbl_received2);
-		} else if (((temp & VSYNC_PIPEA_FLAG) &&
-			    (vblank_pipe & DRM_I915_VBLANK_PIPE_A)) ||
-			   ((temp & VSYNC_PIPEB_FLAG) &&
-			    (vblank_pipe & DRM_I915_VBLANK_PIPE_B)))
-			atomic_inc(&dev->vbl_received);
-
-		DRM_WAKEUP(&dev->vbl_queue);
-		drm_vbl_send_signals(dev);
-
 		if (dev_priv->swaps_pending > 0)
 			drm_locked_tasklet(dev, i915_vblank_tasklet);
-		I915_WRITE(I915REG_PIPEASTAT,
-			pipea_stats|I915_VBLANK_INTERRUPT_ENABLE|
-			I915_VBLANK_CLEAR);
-		I915_WRITE(I915REG_PIPEBSTAT,
-			pipeb_stats|I915_VBLANK_INTERRUPT_ENABLE|
-			I915_VBLANK_CLEAR);
 	}
 
 	/* for now lest just ack it */
@@ -535,13 +642,12 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 
 int i915_emit_irq(struct drm_device *dev)
 {
-	
 	struct drm_i915_private *dev_priv = dev->dev_private;
 	RING_LOCALS;
 
 	i915_kernel_lost_context(dev);
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	i915_emit_breadcrumb(dev);
 
@@ -580,7 +686,7 @@ static int i915_wait_irq(struct drm_device * dev, int irq_nr)
 	struct drm_i915_private *dev_priv = (struct drm_i915_private *) dev->dev_private;
 	int ret = 0;
 
-	DRM_DEBUG("%s irq_nr=%d breadcrumb=%d\n", __FUNCTION__, irq_nr,
+	DRM_DEBUG("irq_nr=%d breadcrumb=%d\n", irq_nr,
 		  READ_BREADCRUMB(dev_priv));
 
 	if (READ_BREADCRUMB(dev_priv) >= irq_nr)
@@ -592,8 +698,7 @@ static int i915_wait_irq(struct drm_device * dev, int irq_nr)
 	i915_user_irq_off(dev_priv);
 
 	if (ret == -EBUSY) {
-		DRM_ERROR("%s: EBUSY -- rec: %d emitted: %d\n",
-			  __FUNCTION__,
+		DRM_ERROR("EBUSY -- rec: %d emitted: %d\n",
 			  READ_BREADCRUMB(dev_priv), (int)dev_priv->counter);
 	}
 
@@ -601,64 +706,6 @@ static int i915_wait_irq(struct drm_device * dev, int irq_nr)
 	return ret;
 }
 
-static int i915_driver_vblank_do_wait(struct drm_device *dev,
-				      unsigned int *sequence,
-				      atomic_t *counter)
-{
-	struct drm_i915_private *dev_priv = dev->dev_private;
-	unsigned int cur_vblank;
-	int ret = 0;
-
-	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
-		return -EINVAL;
-	}
-
-	DRM_WAIT_ON(ret, dev->vbl_queue, 3 * DRM_HZ,
-		    (((cur_vblank = atomic_read(counter))
-			- *sequence) <= (1<<23)));
-
-	*sequence = cur_vblank;
-
-	return ret;
-}
-
-void i915_driver_wait_next_vblank(struct drm_device *dev, int pipe)
-{
-	unsigned int seq;
-
-	seq = pipe ? atomic_read(&dev->vbl_received2) + 1 :
-		atomic_read(&dev->vbl_received) + 1;
-
-	if (!pipe)
-		i915_driver_vblank_do_wait(dev, &seq, &dev->vbl_received);
-	else
-		i915_driver_vblank_do_wait(dev, &seq, &dev->vbl_received2);
-}
-
-int i915_driver_vblank_wait(struct drm_device *dev, unsigned int *sequence)
-{
-	atomic_t *counter;
-
-	if (i915_get_pipe(dev, 0) == 0)
-		counter = &dev->vbl_received;
-	else
-		counter = &dev->vbl_received2;
-	return i915_driver_vblank_do_wait(dev, sequence, counter);
-}
-
-int i915_driver_vblank_wait2(struct drm_device *dev, unsigned int *sequence)
-{
-	atomic_t *counter;
-
-	if (i915_get_pipe(dev, 1) == 0)
-		counter = &dev->vbl_received;
-	else
-		counter = &dev->vbl_received2;
-
-	return i915_driver_vblank_do_wait(dev, sequence, counter);
-}
-
 /* Needs the lock as it touches the ring.
  */
 int i915_irq_emit(struct drm_device *dev, void *data,
@@ -671,7 +718,7 @@ int i915_irq_emit(struct drm_device *dev, void *data,
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -694,24 +741,63 @@ int i915_irq_wait(struct drm_device *dev, void *data,
 	struct drm_i915_irq_wait *irqwait = data;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
 	return i915_wait_irq(dev, irqwait->irq_seq);
 }
 
-void i915_enable_interrupt (struct drm_device *dev)
+int i915_enable_vblank(struct drm_device *dev, int plane)
 {
 	struct drm_i915_private *dev_priv = (struct drm_i915_private *) dev->dev_private;
-	struct drm_output *o;
+	int pipe = i915_get_pipe(dev, plane);
 	
-	dev_priv->irq_enable_reg = USER_INT_FLAG; 
-
-	if (dev_priv->vblank_pipe & DRM_I915_VBLANK_PIPE_A)
+	switch (pipe) {
+	case 0:
 		dev_priv->irq_enable_reg |= VSYNC_PIPEA_FLAG;
-	if (dev_priv->vblank_pipe & DRM_I915_VBLANK_PIPE_B)
+		break;
+	case 1:
 		dev_priv->irq_enable_reg |= VSYNC_PIPEB_FLAG;
+		break;
+	default:
+		DRM_ERROR("tried to enable vblank on non-existent pipe %d\n",
+			  pipe);
+		break;
+	}
+
+	I915_WRITE16(I915REG_INT_ENABLE_R, dev_priv->irq_enable_reg);
+
+	return 0;
+}
+
+void i915_disable_vblank(struct drm_device *dev, int plane)
+{
+	struct drm_i915_private *dev_priv = (struct drm_i915_private *) dev->dev_private;
+	int pipe = i915_get_pipe(dev, plane);
+
+	switch (pipe) {
+	case 0:
+		dev_priv->irq_enable_reg &= ~VSYNC_PIPEA_FLAG;
+		break;
+	case 1:
+		dev_priv->irq_enable_reg &= ~VSYNC_PIPEB_FLAG;
+		break;
+	default:
+		DRM_ERROR("tried to disable vblank on non-existent pipe %d\n",
+			  pipe);
+		break;
+	}
+
+	I915_WRITE16(I915REG_INT_ENABLE_R, dev_priv->irq_enable_reg);
+}
+
+void i915_enable_interrupt (struct drm_device *dev)
+{
+	struct drm_i915_private *dev_priv = (struct drm_i915_private *) dev->dev_private;
+	struct drm_output *o;
+
+	dev_priv->irq_enable_reg |= USER_INT_FLAG;
 
 	if (IS_I9XX(dev) && dev->mode_config.num_output) {
 		dev_priv->irq_enable_reg |= HOTPLUG_FLAG;
@@ -760,20 +846,17 @@ int i915_vblank_pipe_set(struct drm_device *dev, void *data,
 	struct drm_i915_vblank_pipe *pipe = data;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
 	if (pipe->pipe & ~(DRM_I915_VBLANK_PIPE_A|DRM_I915_VBLANK_PIPE_B)) {
-		DRM_ERROR("%s called with invalid pipe 0x%x\n",
-			  __FUNCTION__, pipe->pipe);
+		DRM_ERROR("called with invalid pipe 0x%x\n", pipe->pipe);
 		return -EINVAL;
 	}
 
 	dev_priv->vblank_pipe = pipe->pipe;
 
-	i915_enable_interrupt (dev);
-
 	return 0;
 }
 
@@ -785,7 +868,7 @@ int i915_vblank_pipe_get(struct drm_device *dev, void *data,
 	u16 flag;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -811,6 +894,7 @@ int i915_vblank_swap(struct drm_device *dev, void *data,
 	unsigned int pipe, seqtype, curseq, plane;
 	unsigned long irqflags;
 	struct list_head *list;
+	int ret;
 
 	if (!dev_priv) {
 		DRM_ERROR("%s called with no initialization\n", __func__);
@@ -854,7 +938,8 @@ int i915_vblank_swap(struct drm_device *dev, void *data,
 
 	DRM_SPINUNLOCK_IRQRESTORE(&dev->drw_lock, irqflags);
 
-	curseq = atomic_read(pipe ? &dev->vbl_received2 : &dev->vbl_received);
+	drm_update_vblank_count(dev, pipe);
+	curseq = drm_vblank_count(dev, pipe);
 
 	if (seqtype == _DRM_VBLANK_RELATIVE)
 		swap->sequence += curseq;
@@ -927,6 +1012,12 @@ int i915_vblank_swap(struct drm_device *dev, void *data,
 
 	DRM_DEBUG("\n");
 
+	ret = drm_vblank_get(dev, pipe);
+	if (ret) {
+		drm_free(vbl_swap, sizeof(*vbl_swap), DRM_MEM_DRIVER);
+		return ret;
+	}
+
 	vbl_swap->drw_id = swap->drawable;
 	vbl_swap->plane = plane;
 	vbl_swap->sequence = swap->sequence;
@@ -937,7 +1028,7 @@ int i915_vblank_swap(struct drm_device *dev, void *data,
 
 	DRM_SPINLOCK_IRQSAVE(&dev_priv->swaps_lock, irqflags);
 
-	list_add_tail((struct list_head *)vbl_swap, &dev_priv->vbl_swaps.head);
+	list_add_tail(&vbl_swap->head, &dev_priv->vbl_swaps.head);
 	dev_priv->swaps_pending++;
 
 	DRM_SPINUNLOCK_IRQRESTORE(&dev_priv->swaps_lock, irqflags);
@@ -962,9 +1053,10 @@ void i915_driver_irq_preinstall(struct drm_device * dev)
 
 }
 
-void i915_driver_irq_postinstall(struct drm_device * dev)
+int i915_driver_irq_postinstall(struct drm_device * dev)
 {
 	struct drm_i915_private *dev_priv = (struct drm_i915_private *) dev->dev_private;
+	int ret, num_pipes = 2;
 
 	DRM_SPININIT(&dev_priv->swaps_lock, "swap");
 	INIT_LIST_HEAD(&dev_priv->vbl_swaps.head);
@@ -972,6 +1064,13 @@ void i915_driver_irq_postinstall(struct drm_device * dev)
 
 	DRM_SPININIT(&dev_priv->user_irq_lock, "userirq");
 	dev_priv->user_irq_refcount = 0;
+	dev_priv->irq_enable_reg = 0;
+
+	ret = drm_vblank_init(dev, num_pipes);
+	if (ret)
+		return ret;
+
+	dev->max_vblank_count = 0xffffff; /* only 24 bits of frame count */
 
 	i915_enable_interrupt(dev);
 	DRM_INIT_WAITQUEUE(&dev_priv->irq_queue);
@@ -981,6 +1080,7 @@ void i915_driver_irq_postinstall(struct drm_device * dev)
 	 */
 
 	I915_WRITE(I915REG_INSTPM, (1 << 5) | (1 << 21));
+	return 0;
 }
 
 void i915_driver_irq_uninstall(struct drm_device * dev)
diff --git a/shared-core/i915_mem.c b/shared-core/i915_mem.c
index 4c7ced24..f24547c7 100644
--- a/shared-core/i915_mem.c
+++ b/shared-core/i915_mem.c
@@ -276,7 +276,7 @@ int i915_mem_alloc(struct drm_device *dev, void *data,
 	struct mem_block *block, **heap;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -314,7 +314,7 @@ int i915_mem_free(struct drm_device *dev, void *data,
 	struct mem_block *block, **heap;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -342,7 +342,7 @@ int i915_mem_init_heap(struct drm_device *dev, void *data,
 	struct mem_block **heap;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -366,7 +366,7 @@ int i915_mem_destroy_heap( struct drm_device *dev, void *data,
 	struct mem_block **heap;
 
 	if ( !dev_priv ) {
-		DRM_ERROR( "%s called with no initialization\n", __FUNCTION__ );
+		DRM_ERROR( "called with no initialization\n" );
 		return -EINVAL;
 	}
 
diff --git a/shared-core/mach64_dma.c b/shared-core/mach64_dma.c
index 411b98d5..339234fa 100644
--- a/shared-core/mach64_dma.c
+++ b/shared-core/mach64_dma.c
@@ -57,7 +57,7 @@
  * \returns zero on success, or -EBUSY if the timeout (specificed by
  * drm_mach64_private::usec_timeout) occurs.
  */
-int mach64_do_wait_for_fifo(drm_mach64_private_t * dev_priv, int entries)
+int mach64_do_wait_for_fifo(drm_mach64_private_t *dev_priv, int entries)
 {
 	int slots = 0, i;
 
@@ -68,15 +68,14 @@ int mach64_do_wait_for_fifo(drm_mach64_private_t * dev_priv, int entries)
 		DRM_UDELAY(1);
 	}
 
-	DRM_INFO("%s failed! slots=%d entries=%d\n", __FUNCTION__, slots,
-		 entries);
+	DRM_INFO("failed! slots=%d entries=%d\n", slots, entries);
 	return -EBUSY;
 }
 
 /**
  * Wait for the draw engine to be idle.
  */
-int mach64_do_wait_for_idle(drm_mach64_private_t * dev_priv)
+int mach64_do_wait_for_idle(drm_mach64_private_t *dev_priv)
 {
 	int i, ret;
 
@@ -85,14 +84,12 @@ int mach64_do_wait_for_idle(drm_mach64_private_t * dev_priv)
 		return ret;
 
 	for (i = 0; i < dev_priv->usec_timeout; i++) {
-		if (!(MACH64_READ(MACH64_GUI_STAT) & MACH64_GUI_ACTIVE)) {
+		if (!(MACH64_READ(MACH64_GUI_STAT) & MACH64_GUI_ACTIVE))
 			return 0;
-		}
 		DRM_UDELAY(1);
 	}
 
-	DRM_INFO("%s failed! GUI_STAT=0x%08x\n", __FUNCTION__,
-		 MACH64_READ(MACH64_GUI_STAT));
+	DRM_INFO("failed! GUI_STAT=0x%08x\n", MACH64_READ(MACH64_GUI_STAT));
 	mach64_dump_ring_info(dev_priv);
 	return -EBUSY;
 }
@@ -116,7 +113,7 @@ int mach64_do_wait_for_idle(drm_mach64_private_t * dev_priv)
  *
  * \sa mach64_dump_ring_info()
  */
-int mach64_wait_ring(drm_mach64_private_t * dev_priv, int n)
+int mach64_wait_ring(drm_mach64_private_t *dev_priv, int n)
 {
 	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
 	int i;
@@ -124,9 +121,8 @@ int mach64_wait_ring(drm_mach64_private_t * dev_priv, int n)
 	for (i = 0; i < dev_priv->usec_timeout; i++) {
 		mach64_update_ring_snapshot(dev_priv);
 		if (ring->space >= n) {
-			if (i > 0) {
-				DRM_DEBUG("%s: %d usecs\n", __FUNCTION__, i);
-			}
+			if (i > 0)
+				DRM_DEBUG("%d usecs\n", i);
 			return 0;
 		}
 		DRM_UDELAY(1);
@@ -143,7 +139,7 @@ int mach64_wait_ring(drm_mach64_private_t * dev_priv, int n)
  *
  * \sa mach64_wait_ring()
  */
-static int mach64_ring_idle(drm_mach64_private_t * dev_priv)
+static int mach64_ring_idle(drm_mach64_private_t *dev_priv)
 {
 	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
 	u32 head;
@@ -155,9 +151,8 @@ static int mach64_ring_idle(drm_mach64_private_t * dev_priv)
 		mach64_update_ring_snapshot(dev_priv);
 		if (ring->head == ring->tail &&
 		    !(MACH64_READ(MACH64_GUI_STAT) & MACH64_GUI_ACTIVE)) {
-			if (i > 0) {
-				DRM_DEBUG("%s: %d usecs\n", __FUNCTION__, i);
-			}
+			if (i > 0)
+				DRM_DEBUG("%d usecs\n", i);
 			return 0;
 		}
 		if (ring->head == head) {
@@ -169,8 +164,7 @@ static int mach64_ring_idle(drm_mach64_private_t * dev_priv)
 		DRM_UDELAY(1);
 	}
 
-	DRM_INFO("%s failed! GUI_STAT=0x%08x\n", __FUNCTION__,
-		 MACH64_READ(MACH64_GUI_STAT));
+	DRM_INFO("failed! GUI_STAT=0x%08x\n", MACH64_READ(MACH64_GUI_STAT));
 	mach64_dump_ring_info(dev_priv);
 	return -EBUSY;
 }
@@ -180,7 +174,7 @@ static int mach64_ring_idle(drm_mach64_private_t * dev_priv)
  *
  * \sa mach64_do_engine_reset()
  */
-static void mach64_ring_reset(drm_mach64_private_t * dev_priv)
+static void mach64_ring_reset(drm_mach64_private_t *dev_priv)
 {
 	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
 
@@ -198,7 +192,7 @@ static void mach64_ring_reset(drm_mach64_private_t * dev_priv)
 /**
  * Ensure the all the queued commands will be processed.
  */
-int mach64_do_dma_flush(drm_mach64_private_t * dev_priv)
+int mach64_do_dma_flush(drm_mach64_private_t *dev_priv)
 {
 	/* FIXME: It's not necessary to wait for idle when flushing
 	 * we just need to ensure the ring will be completely processed
@@ -210,14 +204,14 @@ int mach64_do_dma_flush(drm_mach64_private_t * dev_priv)
 /**
  * Stop all DMA activity.
  */
-int mach64_do_dma_idle(drm_mach64_private_t * dev_priv)
+int mach64_do_dma_idle(drm_mach64_private_t *dev_priv)
 {
 	int ret;
 
 	/* wait for completion */
 	if ((ret = mach64_ring_idle(dev_priv)) < 0) {
-		DRM_ERROR("%s failed BM_GUI_TABLE=0x%08x tail: %u\n",
-			  __FUNCTION__, MACH64_READ(MACH64_BM_GUI_TABLE),
+		DRM_ERROR("failed BM_GUI_TABLE=0x%08x tail: %u\n",
+			  MACH64_READ(MACH64_BM_GUI_TABLE),
 			  dev_priv->ring.tail);
 		return ret;
 	}
@@ -232,11 +226,11 @@ int mach64_do_dma_idle(drm_mach64_private_t * dev_priv)
 /**
  * Reset the engine.  This will stop the DMA if it is running.
  */
-int mach64_do_engine_reset(drm_mach64_private_t * dev_priv)
+int mach64_do_engine_reset(drm_mach64_private_t *dev_priv)
 {
 	u32 tmp;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	/* Kill off any outstanding DMA transfers.
 	 */
@@ -276,7 +270,7 @@ int mach64_do_engine_reset(drm_mach64_private_t * dev_priv)
 /**
  * Dump engine registers values.
  */
-void mach64_dump_engine_info(drm_mach64_private_t * dev_priv)
+void mach64_dump_engine_info(drm_mach64_private_t *dev_priv)
 {
 	DRM_INFO("\n");
 	if (!dev_priv->is_pci) {
@@ -417,8 +411,8 @@ void mach64_dump_engine_info(drm_mach64_private_t * dev_priv)
  * Used by mach64_dump_ring_info() to dump the contents of the current buffer
  * pointed by the ring head.
  */
-static void mach64_dump_buf_info(drm_mach64_private_t * dev_priv,
-				 struct drm_buf * buf)
+static void mach64_dump_buf_info(drm_mach64_private_t *dev_priv,
+				 struct drm_buf *buf)
 {
 	u32 addr = GETBUFADDR(buf);
 	u32 used = buf->used >> 2;
@@ -477,7 +471,7 @@ static void mach64_dump_buf_info(drm_mach64_private_t * dev_priv,
  * Dump the ring state and contents, including the contents of the buffer being
  * processed by the graphics engine.
  */
-void mach64_dump_ring_info(drm_mach64_private_t * dev_priv)
+void mach64_dump_ring_info(drm_mach64_private_t *dev_priv)
 {
 	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
 	int i, skipped;
@@ -526,9 +520,8 @@ void mach64_dump_ring_info(drm_mach64_private_t * dev_priv)
 
 			u32 buf_addr = GETBUFADDR(buf);
 
-			if (buf_addr <= addr && addr < buf_addr + buf->used) {
+			if (buf_addr <= addr && addr < buf_addr + buf->used)
 				mach64_dump_buf_info(dev_priv, buf);
-			}
 		}
 	}
 
@@ -564,13 +557,13 @@ void mach64_dump_ring_info(drm_mach64_private_t * dev_priv)
 
 /**
  * Add the end mark to the ring's new tail position.
- * 
+ *
  * The bus master engine will keep processing the DMA buffers listed in the ring
  * until it finds this mark, making it stop.
- * 
+ *
  * \sa mach64_clear_dma_eol
  */ 
-static __inline__ void mach64_set_dma_eol(volatile u32 * addr)
+static __inline__ void mach64_set_dma_eol(volatile u32 *addr)
 {
 #if defined(__i386__)
 	int nr = 31;
@@ -611,17 +604,17 @@ static __inline__ void mach64_set_dma_eol(volatile u32 * addr)
 }
 
 /**
- * Remove the end mark from the ring's old tail position.  
- * 
+ * Remove the end mark from the ring's old tail position.
+ *
  * It should be called after calling mach64_set_dma_eol to mark the ring's new
  * tail position.
- * 
- * We update the end marks while the bus master engine is in operation. Since  
+ *
+ * We update the end marks while the bus master engine is in operation. Since
  * the bus master engine may potentially be reading from the same position
- * that we write, we must change atomically to avoid having intermediary bad 
+ * that we write, we must change atomically to avoid having intermediary bad
  * data.
  */
-static __inline__ void mach64_clear_dma_eol(volatile u32 * addr)
+static __inline__ void mach64_clear_dma_eol(volatile u32 *addr)
 {
 #if defined(__i386__)
 	int nr = 31;
@@ -661,35 +654,35 @@ static __inline__ void mach64_clear_dma_eol(volatile u32 * addr)
 #endif
 }
 
-#define RING_LOCALS									\
+#define RING_LOCALS							\
 	int _ring_tail, _ring_write; unsigned int _ring_mask; volatile u32 *_ring
 
 #define RING_WRITE_OFS  _ring_write
 
-#define BEGIN_RING( n )									\
-do {											\
-	if ( MACH64_VERBOSE ) {								\
-		DRM_INFO( "BEGIN_RING( %d ) in %s\n",					\
-			   (n), __FUNCTION__ );						\
-	}										\
-	if ( dev_priv->ring.space <= (n) * sizeof(u32) ) {				\
-		int ret;								\
-		if ((ret=mach64_wait_ring( dev_priv, (n) * sizeof(u32))) < 0 ) {	\
-			DRM_ERROR( "wait_ring failed, resetting engine\n");		\
-			mach64_dump_engine_info( dev_priv );				\
-			mach64_do_engine_reset( dev_priv );				\
-			return ret;							\
-		}									\
-	}										\
-	dev_priv->ring.space -= (n) * sizeof(u32);					\
-	_ring = (u32 *) dev_priv->ring.start;						\
-	_ring_tail = _ring_write = dev_priv->ring.tail;					\
-	_ring_mask = dev_priv->ring.tail_mask;						\
-} while (0)
+#define BEGIN_RING(n)							\
+	do {								\
+		if (MACH64_VERBOSE) {					\
+			DRM_INFO( "BEGIN_RING( %d ) \n",		\
+				  (n) );				\
+		}							\
+		if (dev_priv->ring.space <= (n) * sizeof(u32)) {	\
+			int ret;					\
+			if ((ret = mach64_wait_ring( dev_priv, (n) * sizeof(u32))) < 0 ) { \
+				DRM_ERROR( "wait_ring failed, resetting engine\n"); \
+				mach64_dump_engine_info( dev_priv );	\
+				mach64_do_engine_reset( dev_priv );	\
+				return ret;				\
+			}						\
+		}							\
+		dev_priv->ring.space -= (n) * sizeof(u32);		\
+		_ring = (u32 *) dev_priv->ring.start;			\
+		_ring_tail = _ring_write = dev_priv->ring.tail;		\
+		_ring_mask = dev_priv->ring.tail_mask;			\
+	} while (0)
 
 #define OUT_RING( x )						\
 do {								\
-	if ( MACH64_VERBOSE ) {					\
+	if (MACH64_VERBOSE) {					\
 		DRM_INFO( "   OUT_RING( 0x%08x ) at 0x%x\n",	\
 			   (unsigned int)(x), _ring_write );	\
 	}							\
@@ -699,7 +692,7 @@ do {								\
 
 #define ADVANCE_RING()							\
 do {									\
-	if ( MACH64_VERBOSE ) {						\
+	if (MACH64_VERBOSE) {						\
 		DRM_INFO( "ADVANCE_RING() wr=0x%06x tail=0x%06x\n",	\
 			  _ring_write, _ring_tail );			\
 	}								\
@@ -835,7 +828,7 @@ static int mach64_bm_dma_test(struct drm_device * dev)
 	u32 src_cntl, pat_reg0, pat_reg1;
 	int i, count, failed;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	table = (u32 *) dev_priv->ring.start;
 
@@ -1011,7 +1004,7 @@ static int mach64_do_dma_init(struct drm_device * dev, drm_mach64_init_t * init)
 	u32 tmp;
 	int i, ret;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	dev_priv = drm_alloc(sizeof(drm_mach64_private_t), DRM_MEM_DRIVER);
 	if (dev_priv == NULL)
@@ -1221,7 +1214,7 @@ static int mach64_do_dma_init(struct drm_device * dev, drm_mach64_init_t * init)
 /** MMIO Pseudo-DMA (intended primarily for debugging, not performance)
  */
 
-int mach64_do_dispatch_pseudo_dma(drm_mach64_private_t * dev_priv)
+int mach64_do_dispatch_pseudo_dma(drm_mach64_private_t *dev_priv)
 {
 	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
 	volatile u32 *ring_read;
@@ -1236,9 +1229,7 @@ int mach64_do_dispatch_pseudo_dma(drm_mach64_private_t * dev_priv)
 	target = MACH64_BM_ADDR;
 
 	if ((ret = mach64_do_wait_for_idle(dev_priv)) < 0) {
-		DRM_INFO
-		    ("%s: idle failed before pseudo-dma dispatch, resetting engine\n",
-		     __FUNCTION__);
+		DRM_INFO("idle failed before pseudo-dma dispatch, resetting engine\n");
 		mach64_dump_engine_info(dev_priv);
 		mach64_do_engine_reset(dev_priv);
 		return ret;
@@ -1359,7 +1350,7 @@ int mach64_do_dispatch_pseudo_dma(drm_mach64_private_t * dev_priv)
 	MACH64_WRITE(MACH64_BM_GUI_TABLE_CMD,
 		     ring->head_addr | MACH64_CIRCULAR_BUF_SIZE_16KB);
 
-	DRM_DEBUG("%s completed\n", __FUNCTION__);
+	DRM_DEBUG("completed\n");
 	return 0;
 }
 
@@ -1372,7 +1363,7 @@ int mach64_do_dispatch_pseudo_dma(drm_mach64_private_t * dev_priv)
 
 int mach64_do_cleanup_dma(struct drm_device * dev)
 {
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	/* Make sure interrupts are disabled here because the uninstall ioctl
 	 * may not have been called from userspace and after dev_private
@@ -1416,7 +1407,7 @@ int mach64_dma_init(struct drm_device *dev, void *data,
 {
 	drm_mach64_init_t *init = data;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
@@ -1435,7 +1426,7 @@ int mach64_dma_idle(struct drm_device *dev, void *data,
 {
 	drm_mach64_private_t *dev_priv = dev->dev_private;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
@@ -1447,7 +1438,7 @@ int mach64_dma_flush(struct drm_device *dev, void *data,
 {
 	drm_mach64_private_t *dev_priv = dev->dev_private;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
@@ -1459,7 +1450,7 @@ int mach64_engine_reset(struct drm_device *dev, void *data,
 {
 	drm_mach64_private_t *dev_priv = dev->dev_private;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
@@ -1481,8 +1472,7 @@ int mach64_init_freelist(struct drm_device * dev)
 	struct list_head *ptr;
 	int i;
 
-	DRM_DEBUG("%s: adding %d buffers to freelist\n", __FUNCTION__,
-		  dma->buf_count);
+	DRM_DEBUG("adding %d buffers to freelist\n", dma->buf_count);
 
 	for (i = 0; i < dma->buf_count; i++) {
 		if ((entry =
@@ -1506,7 +1496,7 @@ void mach64_destroy_freelist(struct drm_device * dev)
 	struct list_head *ptr;
 	struct list_head *tmp;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	list_for_each_safe(ptr, tmp, &dev_priv->pending) {
 		list_del(ptr);
@@ -1529,7 +1519,7 @@ void mach64_destroy_freelist(struct drm_device * dev)
 /* IMPORTANT: This function should only be called when the engine is idle or locked up,
  * as it assumes all buffers in the pending list have been completed by the hardware.
  */
-int mach64_do_release_used_buffers(drm_mach64_private_t * dev_priv)
+int mach64_do_release_used_buffers(drm_mach64_private_t *dev_priv)
 {
 	struct list_head *ptr;
 	struct list_head *tmp;
@@ -1551,13 +1541,12 @@ int mach64_do_release_used_buffers(drm_mach64_private_t * dev_priv)
 		}
 	}
 
-	DRM_DEBUG("%s: released %d buffers from pending list\n", __FUNCTION__,
-		  i);
+	DRM_DEBUG("released %d buffers from pending list\n", i);
 
 	return 0;
 }
 
-static int mach64_do_reclaim_completed(drm_mach64_private_t * dev_priv)
+static int mach64_do_reclaim_completed(drm_mach64_private_t *dev_priv)
 {
 	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
 	struct list_head *ptr;
@@ -1579,8 +1568,7 @@ static int mach64_do_reclaim_completed(drm_mach64_private_t * dev_priv)
 #endif
 		/* last pass is complete, so release everything */
 		mach64_do_release_used_buffers(dev_priv);
-		DRM_DEBUG("%s: idle engine, freed all buffers.\n",
-		     __FUNCTION__);
+		DRM_DEBUG("idle engine, freed all buffers.\n");
 		if (list_empty(&dev_priv->free_list)) {
 			DRM_ERROR("Freelist empty with idle engine\n");
 			return -1;
@@ -1621,9 +1609,9 @@ static int mach64_do_reclaim_completed(drm_mach64_private_t * dev_priv)
 			list_del(ptr);
 			list_add_tail(ptr, &dev_priv->free_list);
 			DRM_DEBUG
-			    ("%s: freed processed buffer (head=%d tail=%d "
+			    ("freed processed buffer (head=%d tail=%d "
 			     "buf ring ofs=%d).\n",
-			     __FUNCTION__, head, tail, ofs);
+			     head, tail, ofs);
 			return 0;
 		}
 	}
@@ -1631,7 +1619,7 @@ static int mach64_do_reclaim_completed(drm_mach64_private_t * dev_priv)
 	return 1;
 }
 
-struct drm_buf *mach64_freelist_get(drm_mach64_private_t * dev_priv)
+struct drm_buf *mach64_freelist_get(drm_mach64_private_t *dev_priv)
 {
 	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
 	drm_mach64_freelist_t *entry;
@@ -1677,7 +1665,7 @@ struct drm_buf *mach64_freelist_get(drm_mach64_private_t * dev_priv)
 	return entry->buf;
 }
 
-int mach64_freelist_put(drm_mach64_private_t * dev_priv, struct drm_buf * copy_buf)
+int mach64_freelist_put(drm_mach64_private_t *dev_priv, struct drm_buf *copy_buf)
 {
 	struct list_head *ptr;
 	drm_mach64_freelist_t *entry;
@@ -1686,8 +1674,7 @@ int mach64_freelist_put(drm_mach64_private_t * dev_priv, struct drm_buf * copy_b
 	list_for_each(ptr, &dev_priv->pending) {
 		entry = list_entry(ptr, drm_mach64_freelist_t, list);
 		if (copy_buf == entry->buf) {
-			DRM_ERROR("%s: Trying to release a pending buf\n",
-			     __FUNCTION__);
+			DRM_ERROR("Trying to release a pending buf\n");
 			return -EFAULT;
 		}
 	}
diff --git a/shared-core/mach64_drv.h b/shared-core/mach64_drv.h
index 1768a2a4..fb8a7724 100644
--- a/shared-core/mach64_drv.h
+++ b/shared-core/mach64_drv.h
@@ -96,6 +96,8 @@ typedef struct drm_mach64_private {
 	unsigned int depth_bpp;
 	unsigned int depth_offset, depth_pitch;
 
+	atomic_t vbl_received;          /**< Number of vblanks received. */
+
 	u32 front_offset_pitch;
 	u32 back_offset_pitch;
 	u32 depth_offset_pitch;
@@ -160,13 +162,14 @@ extern int mach64_dma_blit(struct drm_device *dev, void *data,
 			   struct drm_file *file_priv);
 extern int mach64_get_param(struct drm_device *dev, void *data,
 			    struct drm_file *file_priv);
-extern int mach64_driver_vblank_wait(struct drm_device * dev,
-				     unsigned int *sequence);
 
+extern u32 mach64_get_vblank_counter(struct drm_device *dev, int crtc);
+extern int mach64_enable_vblank(struct drm_device *dev, int crtc);
+extern void mach64_disable_vblank(struct drm_device *dev, int crtc);
 extern irqreturn_t mach64_driver_irq_handler(DRM_IRQ_ARGS);
-extern void mach64_driver_irq_preinstall(struct drm_device * dev);
-extern void mach64_driver_irq_postinstall(struct drm_device * dev);
-extern void mach64_driver_irq_uninstall(struct drm_device * dev);
+extern void mach64_driver_irq_preinstall(struct drm_device *dev);
+extern int mach64_driver_irq_postinstall(struct drm_device *dev);
+extern void mach64_driver_irq_uninstall(struct drm_device *dev);
 
 /* ================================================================
  * Registers
@@ -536,8 +539,7 @@ static __inline__ void mach64_ring_start(drm_mach64_private_t * dev_priv)
 {
 	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
 
-	DRM_DEBUG("%s: head_addr: 0x%08x head: %d tail: %d space: %d\n",
-		  __FUNCTION__,
+	DRM_DEBUG("head_addr: 0x%08x head: %d tail: %d space: %d\n",
 		  ring->head_addr, ring->head, ring->tail, ring->space);
 
 	if (mach64_do_wait_for_idle(dev_priv) < 0) {
@@ -563,8 +565,7 @@ static __inline__ void mach64_ring_start(drm_mach64_private_t * dev_priv)
 static __inline__ void mach64_ring_resume(drm_mach64_private_t * dev_priv,
 					  drm_mach64_descriptor_ring_t * ring)
 {
-	DRM_DEBUG("%s: head_addr: 0x%08x head: %d tail: %d space: %d\n",
-		  __FUNCTION__,
+	DRM_DEBUG("head_addr: 0x%08x head: %d tail: %d space: %d\n",
 		  ring->head_addr, ring->head, ring->tail, ring->space);
 
 	/* reset descriptor table ring head */
@@ -583,8 +584,7 @@ static __inline__ void mach64_ring_resume(drm_mach64_private_t * dev_priv,
 		MACH64_WRITE(MACH64_DST_HEIGHT_WIDTH, 0);
 		if (dev_priv->driver_mode == MACH64_MODE_DMA_SYNC) {
 			if ((mach64_do_wait_for_idle(dev_priv)) < 0) {
-				DRM_ERROR("%s: idle failed, resetting engine\n",
-					  __FUNCTION__);
+				DRM_ERROR("idle failed, resetting engine\n");
 				mach64_dump_engine_info(dev_priv);
 				mach64_do_engine_reset(dev_priv);
 				return;
@@ -609,8 +609,7 @@ static __inline__ void mach64_ring_resume(drm_mach64_private_t * dev_priv,
 static __inline__ void mach64_ring_tick(drm_mach64_private_t * dev_priv,
 					drm_mach64_descriptor_ring_t * ring)
 {
-	DRM_DEBUG("%s: head_addr: 0x%08x head: %d tail: %d space: %d\n",
-		  __FUNCTION__,
+	DRM_DEBUG("head_addr: 0x%08x head: %d tail: %d space: %d\n",
 		  ring->head_addr, ring->head, ring->tail, ring->space);
 
 	if (!dev_priv->ring_running) {
@@ -657,8 +656,7 @@ static __inline__ void mach64_ring_tick(drm_mach64_private_t * dev_priv,
 
 static __inline__ void mach64_ring_stop(drm_mach64_private_t * dev_priv)
 {
-	DRM_DEBUG("%s: head_addr: 0x%08x head: %d tail: %d space: %d\n",
-		  __FUNCTION__,
+	DRM_DEBUG("head_addr: 0x%08x head: %d tail: %d space: %d\n",
 		  dev_priv->ring.head_addr, dev_priv->ring.head,
 		  dev_priv->ring.tail, dev_priv->ring.space);
 
@@ -679,7 +677,7 @@ mach64_update_ring_snapshot(drm_mach64_private_t * dev_priv)
 {
 	drm_mach64_descriptor_ring_t *ring = &dev_priv->ring;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	mach64_ring_tick(dev_priv, ring);
 
@@ -720,7 +718,7 @@ static __inline__ int mach64_find_pending_buf_entry(drm_mach64_private_t *
 	struct list_head *ptr;
 #if MACH64_EXTRA_CHECKING
 	if (list_empty(&dev_priv->pending)) {
-		DRM_ERROR("Empty pending list in %s\n", __FUNCTION__);
+		DRM_ERROR("Empty pending list in \n");
 		return -EINVAL;
 	}
 #endif
@@ -747,18 +745,15 @@ do {						\
 #define DMAGETPTR( file_priv, dev_priv, n )				\
 do {									\
 	if ( MACH64_VERBOSE ) {						\
-		DRM_INFO( "DMAGETPTR( %d ) in %s\n",			\
-			  n, __FUNCTION__ );				\
+		DRM_INFO( "DMAGETPTR( %d )\n", (n) );			\
 	}								\
 	_buf = mach64_freelist_get( dev_priv );				\
 	if (_buf == NULL) {						\
-		DRM_ERROR("%s: couldn't get buffer in DMAGETPTR\n",	\
-			   __FUNCTION__ );				\
+		DRM_ERROR("couldn't get buffer in DMAGETPTR\n");	\
 		return -EAGAIN;					\
 	}								\
 	if (_buf->pending) {						\
-	        DRM_ERROR("%s: pending buf in DMAGETPTR\n",		\
-			   __FUNCTION__ );				\
+	        DRM_ERROR("pending buf in DMAGETPTR\n");		\
 		return -EFAULT;					\
 	}								\
 	_buf->file_priv = file_priv;					\
@@ -778,92 +773,87 @@ do {								\
 	_buf->used += 8;					\
 } while (0)
 
-#define DMAADVANCE( dev_priv, _discard )						     \
-do {											     \
-	struct list_head *ptr;								     \
-	int ret;									     \
-											     \
-	if ( MACH64_VERBOSE ) {								     \
-		DRM_INFO( "DMAADVANCE() in %s\n", __FUNCTION__ );			     \
-	}										     \
-											     \
-	if (_buf->used <= 0) {								     \
-		DRM_ERROR( "DMAADVANCE() in %s: sending empty buf %d\n",		     \
-				   __FUNCTION__, _buf->idx );				     \
-		return -EFAULT;							     \
-	}										     \
-	if (_buf->pending) {								     \
-                /* This is a resued buffer, so we need to find it in the pending list */     \
-		if ( (ret=mach64_find_pending_buf_entry(dev_priv, &_entry, _buf)) ) {	     \
-			DRM_ERROR( "DMAADVANCE() in %s: couldn't find pending buf %d\n",     \
-				   __FUNCTION__, _buf->idx );				     \
-			return ret;							     \
-		}									     \
-		if (_entry->discard) {							     \
-			DRM_ERROR( "DMAADVANCE() in %s: sending discarded pending buf %d\n", \
-				   __FUNCTION__, _buf->idx );				     \
-			return -EFAULT;						     \
-		}									     \
-	} else {									     \
-		if (list_empty(&dev_priv->placeholders)) {				     \
-			DRM_ERROR( "DMAADVANCE() in %s: empty placeholder list\n",	     \
-				__FUNCTION__ );						     \
-			return -EFAULT;						     \
-		}									     \
-		ptr = dev_priv->placeholders.next;					     \
-		list_del(ptr);								     \
-		_entry = list_entry(ptr, drm_mach64_freelist_t, list);			     \
-		_buf->pending = 1;							     \
-		_entry->buf = _buf;							     \
-		list_add_tail(ptr, &dev_priv->pending);					     \
-	}										     \
-	_entry->discard = (_discard);							     \
-	if ( (ret = mach64_add_buf_to_ring( dev_priv, _entry )) )			     \
-		return ret;								     \
-} while (0)
-
-#define DMADISCARDBUF()									\
-do {											\
-	if (_entry == NULL) {								\
-		int ret;								\
-		if ( (ret=mach64_find_pending_buf_entry(dev_priv, &_entry, _buf)) ) {	\
-			DRM_ERROR( "%s: couldn't find pending buf %d\n",		\
-				   __FUNCTION__, _buf->idx );				\
-			return ret;							\
-		}									\
-	}										\
-	_entry->discard = 1;								\
-} while(0)
-
-#define DMAADVANCEHOSTDATA( dev_priv )							\
-do {											\
-	struct list_head *ptr;								\
-	int ret;									\
-											\
-	if ( MACH64_VERBOSE ) {								\
-		DRM_INFO( "DMAADVANCEHOSTDATA() in %s\n", __FUNCTION__ );		\
-	}										\
-											\
-	if (_buf->used <= 0) {								\
-		DRM_ERROR( "DMAADVANCEHOSTDATA() in %s: sending empty buf %d\n",	\
-				   __FUNCTION__, _buf->idx );				\
-		return -EFAULT;							\
-	}										\
-	if (list_empty(&dev_priv->placeholders)) {					\
-		DRM_ERROR( "%s: empty placeholder list in DMAADVANCEHOSTDATA()\n",	\
-			   __FUNCTION__ );						\
-		return -EFAULT;							\
-	}										\
-											\
-        ptr = dev_priv->placeholders.next;						\
-	list_del(ptr);									\
-	_entry = list_entry(ptr, drm_mach64_freelist_t, list);				\
-	_entry->buf = _buf;								\
-	_entry->buf->pending = 1;							\
-	list_add_tail(ptr, &dev_priv->pending);						\
-	_entry->discard = 1;								\
-	if ( (ret = mach64_add_hostdata_buf_to_ring( dev_priv, _entry )) )		\
-		return ret;								\
-} while (0)
+#define DMAADVANCE( dev_priv, _discard )				\
+	do {								\
+		struct list_head *ptr;					\
+		int ret;						\
+									\
+		if ( MACH64_VERBOSE ) {					\
+			DRM_INFO( "DMAADVANCE() in \n" );		\
+		}							\
+									\
+		if (_buf->used <= 0) {					\
+			DRM_ERROR( "DMAADVANCE(): sending empty buf %d\n", \
+				   _buf->idx );				\
+			return -EFAULT;					\
+		}							\
+		if (_buf->pending) {					\
+			/* This is a resued buffer, so we need to find it in the pending list */ \
+			if ((ret = mach64_find_pending_buf_entry(dev_priv, &_entry, _buf))) { \
+				DRM_ERROR( "DMAADVANCE(): couldn't find pending buf %d\n", _buf->idx );	\
+				return ret;				\
+			}						\
+			if (_entry->discard) {				\
+				DRM_ERROR( "DMAADVANCE(): sending discarded pending buf %d\n", _buf->idx ); \
+				return -EFAULT;				\
+			}						\
+		} else {						\
+			if (list_empty(&dev_priv->placeholders)) {	\
+				DRM_ERROR( "DMAADVANCE(): empty placeholder list\n"); \
+				return -EFAULT;				\
+			}						\
+			ptr = dev_priv->placeholders.next;		\
+			list_del(ptr);					\
+			_entry = list_entry(ptr, drm_mach64_freelist_t, list); \
+			_buf->pending = 1;				\
+			_entry->buf = _buf;				\
+			list_add_tail(ptr, &dev_priv->pending);		\
+		}							\
+		_entry->discard = (_discard);				\
+		if ((ret = mach64_add_buf_to_ring( dev_priv, _entry ))) \
+			return ret;					\
+	} while (0)
+
+#define DMADISCARDBUF()							\
+	do {								\
+		if (_entry == NULL) {					\
+			int ret;					\
+			if ((ret = mach64_find_pending_buf_entry(dev_priv, &_entry, _buf))) { \
+				DRM_ERROR( "couldn't find pending buf %d\n", \
+					   _buf->idx );			\
+				return ret;				\
+			}						\
+		}							\
+		_entry->discard = 1;					\
+	} while(0)
+
+#define DMAADVANCEHOSTDATA( dev_priv )					\
+	do {								\
+		struct list_head *ptr;					\
+		int ret;						\
+									\
+		if ( MACH64_VERBOSE ) {					\
+			DRM_INFO( "DMAADVANCEHOSTDATA() in \n" );	\
+		}							\
+									\
+		if (_buf->used <= 0) {					\
+			DRM_ERROR( "DMAADVANCEHOSTDATA(): sending empty buf %d\n", _buf->idx );	\
+			return -EFAULT;					\
+		}							\
+		if (list_empty(&dev_priv->placeholders)) {		\
+			DRM_ERROR( "empty placeholder list in DMAADVANCEHOSTDATA()\n" ); \
+			return -EFAULT;					\
+		}							\
+									\
+		ptr = dev_priv->placeholders.next;			\
+		list_del(ptr);						\
+		_entry = list_entry(ptr, drm_mach64_freelist_t, list);	\
+		_entry->buf = _buf;					\
+		_entry->buf->pending = 1;				\
+		list_add_tail(ptr, &dev_priv->pending);			\
+		_entry->discard = 1;					\
+		if ((ret = mach64_add_hostdata_buf_to_ring( dev_priv, _entry ))) \
+			return ret;					\
+	} while (0)
 
 #endif				/* __MACH64_DRV_H__ */
diff --git a/shared-core/mach64_irq.c b/shared-core/mach64_irq.c
index 4122dd91..2d522a6c 100644
--- a/shared-core/mach64_irq.c
+++ b/shared-core/mach64_irq.c
@@ -42,9 +42,8 @@
 
 irqreturn_t mach64_driver_irq_handler(DRM_IRQ_ARGS)
 {
-	struct drm_device *dev = (struct drm_device *) arg;
-	drm_mach64_private_t *dev_priv =
-	    (drm_mach64_private_t *) dev->dev_private;
+	struct drm_device *dev = arg;
+	drm_mach64_private_t *dev_priv = dev->dev_private;
 	int status;
 
 	status = MACH64_READ(MACH64_CRTC_INT_CNTL);
@@ -62,74 +61,81 @@ irqreturn_t mach64_driver_irq_handler(DRM_IRQ_ARGS)
 			     (status & ~MACH64_CRTC_INT_ACKS)
 			     | MACH64_CRTC_VBLANK_INT);
 
-		atomic_inc(&dev->vbl_received);
-		DRM_WAKEUP(&dev->vbl_queue);
-		drm_vbl_send_signals(dev);
+		atomic_inc(&dev_priv->vbl_received);
+		drm_handle_vblank(dev, 0);
 		return IRQ_HANDLED;
 	}
 	return IRQ_NONE;
 }
 
-int mach64_driver_vblank_wait(struct drm_device * dev, unsigned int *sequence)
+u32 mach64_get_vblank_counter(struct drm_device * dev, int crtc)
 {
-	unsigned int cur_vblank;
-	int ret = 0;
-
-	/* Assume that the user has missed the current sequence number
-	 * by about a day rather than she wants to wait for years
-	 * using vertical blanks...
-	 */
-	DRM_WAIT_ON(ret, dev->vbl_queue, 3 * DRM_HZ,
-		    (((cur_vblank = atomic_read(&dev->vbl_received))
-		      - *sequence) <= (1 << 23)));
+	const drm_mach64_private_t *const dev_priv = dev->dev_private;
+		
+	if (crtc != 0) {
+		return 0;
+	}
+	
+	return atomic_read(&dev_priv->vbl_received);
+}
 
-	*sequence = cur_vblank;
+int mach64_enable_vblank(struct drm_device * dev, int crtc)
+{
+	drm_mach64_private_t *dev_priv = dev->dev_private;
+	u32 status = MACH64_READ(MACH64_CRTC_INT_CNTL);
+	
+	if (crtc != 0) {
+		DRM_ERROR("tried to enable vblank on non-existent crtc %d\n", crtc);
+		return 0;
+	}
+	
+	DRM_DEBUG("before enable vblank CRTC_INT_CTNL: 0x%08x\n", status);
+	
+	/* Turn on VBLANK interrupt */
+	MACH64_WRITE(MACH64_CRTC_INT_CNTL, MACH64_READ(MACH64_CRTC_INT_CNTL)
+		     | MACH64_CRTC_VBLANK_INT_EN);
 
-	return ret;
+	return 0;
 }
 
-/* drm_dma.h hooks
-*/
-void mach64_driver_irq_preinstall(struct drm_device * dev)
-{
-	drm_mach64_private_t *dev_priv =
-	    (drm_mach64_private_t *) dev->dev_private;
 
+void mach64_disable_vblank(struct drm_device * dev, int crtc)
+{
+	drm_mach64_private_t *dev_priv = dev->dev_private;
 	u32 status = MACH64_READ(MACH64_CRTC_INT_CNTL);
 
-	DRM_DEBUG("before install CRTC_INT_CTNL: 0x%08x\n", status);
+	DRM_DEBUG("before disable vblank CRTC_INT_CTNL: 0x%08x\n", status);
 
 	/* Disable and clear VBLANK interrupt */
 	MACH64_WRITE(MACH64_CRTC_INT_CNTL, (status & ~MACH64_CRTC_VBLANK_INT_EN)
 		     | MACH64_CRTC_VBLANK_INT);
 }
 
-void mach64_driver_irq_postinstall(struct drm_device * dev)
+/* drm_dma.h hooks
+*/
+void mach64_driver_irq_preinstall(struct drm_device * dev)
 {
-	drm_mach64_private_t *dev_priv =
-	    (drm_mach64_private_t *) dev->dev_private;
+	drm_mach64_private_t *dev_priv = dev->dev_private;
 
-	/* Turn on VBLANK interrupt */
-	MACH64_WRITE(MACH64_CRTC_INT_CNTL, MACH64_READ(MACH64_CRTC_INT_CNTL)
-		     | MACH64_CRTC_VBLANK_INT_EN);
+	u32 status = MACH64_READ(MACH64_CRTC_INT_CNTL);
 
-	DRM_DEBUG("after install CRTC_INT_CTNL: 0x%08x\n",
-		  MACH64_READ(MACH64_CRTC_INT_CNTL));
+	DRM_DEBUG("before install CRTC_INT_CTNL: 0x%08x\n", status);
 
+	mach64_disable_vblank(dev,0);
+}
+
+int mach64_driver_irq_postinstall(struct drm_device * dev)
+{
+	return drm_vblank_init(dev, 1);
 }
 
 void mach64_driver_irq_uninstall(struct drm_device * dev)
 {
-	drm_mach64_private_t *dev_priv =
-	    (drm_mach64_private_t *) dev->dev_private;
+	drm_mach64_private_t *dev_priv = dev->dev_private;
 	if (!dev_priv)
 		return;
 
-	/* Disable and clear VBLANK interrupt */
-	MACH64_WRITE(MACH64_CRTC_INT_CNTL,
-		     (MACH64_READ(MACH64_CRTC_INT_CNTL) &
-		      ~MACH64_CRTC_VBLANK_INT_EN)
-		     | MACH64_CRTC_VBLANK_INT);
+	mach64_disable_vblank(dev, 0);
 
 	DRM_DEBUG("after uninstall CRTC_INT_CTNL: 0x%08x\n",
 		  MACH64_READ(MACH64_CRTC_INT_CNTL));
diff --git a/shared-core/mach64_state.c b/shared-core/mach64_state.c
index 88ff4843..c82f38bb 100644
--- a/shared-core/mach64_state.c
+++ b/shared-core/mach64_state.c
@@ -95,7 +95,7 @@ static int mach64_emit_cliprect(struct drm_file *file_priv,
 	drm_mach64_context_regs_t *regs = &sarea_priv->context_state;
 	DMALOCALS;
 
-	DRM_DEBUG("%s: box=%p\n", __FUNCTION__, box);
+	DRM_DEBUG("box=%p\n", box);
 
 	/* Get GL scissor */
 	/* FIXME: store scissor in SAREA as a cliprect instead of in
@@ -146,7 +146,7 @@ static __inline__ int mach64_emit_state(struct drm_file *file_priv,
 	if (MACH64_VERBOSE) {
 		mach64_print_dirty(__FUNCTION__, dirty);
 	} else {
-		DRM_DEBUG("%s: dirty=0x%08x\n", __FUNCTION__, dirty);
+		DRM_DEBUG("dirty=0x%08x\n", dirty);
 	}
 
 	DMAGETPTR(file_priv, dev_priv, 17);	/* returns on failure to get buffer */
@@ -229,7 +229,7 @@ static int mach64_dma_dispatch_clear(struct drm_device * dev,
 	int i;
 	DMALOCALS;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	switch (dev_priv->fb_bpp) {
 	case 16:
@@ -368,7 +368,7 @@ static int mach64_dma_dispatch_swap(struct drm_device * dev,
 	int i;
 	DMALOCALS;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	switch (dev_priv->fb_bpp) {
 	case 16:
@@ -445,7 +445,7 @@ static int mach64_do_get_frames_queued(drm_mach64_private_t * dev_priv)
 	int i, start;
 	u32 head, tail, ofs;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	if (sarea_priv->frames_queued == 0)
 		return 0;
@@ -525,15 +525,14 @@ static __inline__ int copy_from_user_vertex(u32 *to,
 				from += count;
 				to += count;
 			} else {
-				DRM_ERROR("%s: Got bad command: 0x%04x\n",
-					  __FUNCTION__, reg);
+				DRM_ERROR("Got bad command: 0x%04x\n", reg);
 				drm_free(orig_from, bytes, DRM_MEM_DRIVER);
 				return -EACCES;
 			}
 		} else {
 			DRM_ERROR
-			    ("%s: Got bad command count(=%u) dwords remaining=%lu\n",
-			     __FUNCTION__, count, n);
+			    ("Got bad command count(=%u) dwords remaining=%lu\n",
+			     count, n);
 			drm_free(orig_from, bytes, DRM_MEM_DRIVER);
 			return -EINVAL;
 		}
@@ -543,7 +542,7 @@ static __inline__ int copy_from_user_vertex(u32 *to,
 	if (n == 0)
 		return 0;
 	else {
-		DRM_ERROR("%s: Bad buf->used(=%lu)\n", __FUNCTION__, bytes);
+		DRM_ERROR("Bad buf->used(=%lu)\n", bytes);
 		return -EINVAL;
 	}
 }
@@ -563,15 +562,15 @@ static int mach64_dma_dispatch_vertex(struct drm_device * dev,
 	int verify_ret = 0;
 	DMALOCALS;
 
-	DRM_DEBUG("%s: buf=%p used=%lu nbox=%d\n",
-		  __FUNCTION__, buf, used, sarea_priv->nbox);
+	DRM_DEBUG("buf=%p used=%lu nbox=%d\n",
+		  buf, used, sarea_priv->nbox);
 
 	if (!used)
 		goto _vertex_done;
 
 	copy_buf = mach64_freelist_get(dev_priv);
 	if (copy_buf == NULL) {
-		DRM_ERROR("%s: couldn't get buffer\n", __FUNCTION__);
+		DRM_ERROR("couldn't get buffer\n");
 		return -EAGAIN;
 	}
 
@@ -698,7 +697,7 @@ static int mach64_dma_dispatch_blit(struct drm_device * dev,
 
 	copy_buf = mach64_freelist_get(dev_priv);
 	if (copy_buf == NULL) {
-		DRM_ERROR("%s: couldn't get buffer\n", __FUNCTION__);
+		DRM_ERROR("couldn't get buffer\n");
 		return -EAGAIN;
 	}
 
@@ -759,7 +758,7 @@ static int mach64_dma_dispatch_blit(struct drm_device * dev,
 	DMAOUTREG(MACH64_DST_X_Y, (blit->y << 16) | blit->x);
 	DMAOUTREG(MACH64_DST_WIDTH_HEIGHT, (blit->height << 16) | blit->width);
 
-	DRM_DEBUG("%s: %lu bytes\n", __FUNCTION__, used);
+	DRM_DEBUG("%lu bytes\n", used);
 
 	/* Add the buffer to the queue */
 	DMAADVANCEHOSTDATA(dev_priv);
@@ -780,7 +779,7 @@ int mach64_dma_clear(struct drm_device *dev, void *data,
 	drm_mach64_clear_t *clear = data;
 	int ret;
 
-	DRM_DEBUG("%s: pid=%d\n", __FUNCTION__, DRM_CURRENTPID);
+	DRM_DEBUG("pid=%d\n", DRM_CURRENTPID);
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
@@ -805,7 +804,7 @@ int mach64_dma_swap(struct drm_device *dev, void *data,
 	drm_mach64_sarea_t *sarea_priv = dev_priv->sarea_priv;
 	int ret;
 
-	DRM_DEBUG("%s: pid=%d\n", __FUNCTION__, DRM_CURRENTPID);
+	DRM_DEBUG("pid=%d\n", DRM_CURRENTPID);
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
@@ -830,12 +829,12 @@ int mach64_dma_vertex(struct drm_device *dev, void *data,
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
-	DRM_DEBUG("%s: pid=%d buf=%p used=%lu discard=%d\n",
-		  __FUNCTION__, DRM_CURRENTPID,
+	DRM_DEBUG("pid=%d buf=%p used=%lu discard=%d\n",
+		  DRM_CURRENTPID,
 		  vertex->buf, vertex->used, vertex->discard);
 
 	if (vertex->prim < 0 || vertex->prim > MACH64_PRIM_POLYGON) {
@@ -882,10 +881,10 @@ int mach64_get_param(struct drm_device *dev, void *data,
 	drm_mach64_getparam_t *param = data;
 	int value;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
diff --git a/shared-core/mga_dma.c b/shared-core/mga_dma.c
index 67236b2d..d56f4d7a 100644
--- a/shared-core/mga_dma.c
+++ b/shared-core/mga_dma.c
@@ -997,7 +997,7 @@ static int mga_do_cleanup_dma(struct drm_device *dev, int full_cleanup)
 		}
 	}
 
-	return 0;
+	return err;
 }
 
 int mga_dma_init(struct drm_device *dev, void *data,
@@ -1049,7 +1049,7 @@ int mga_dma_flush(struct drm_device *dev, void *data,
 #if MGA_DMA_DEBUG
 		int ret = mga_do_wait_for_idle(dev_priv);
 		if (ret < 0)
-			DRM_INFO("%s: -EBUSY\n", __FUNCTION__);
+			DRM_INFO("-EBUSY\n");
 		return ret;
 #else
 		return mga_do_wait_for_idle(dev_priv);
diff --git a/shared-core/mga_drv.h b/shared-core/mga_drv.h
index 0c7fab5b..bf3be808 100644
--- a/shared-core/mga_drv.h
+++ b/shared-core/mga_drv.h
@@ -113,13 +113,14 @@ typedef struct drm_mga_private {
 	 * \sa drm_mga_private_t::mmio
 	 */
 	/*@{*/
-	u32 mmio_base;             /**< Bus address of base of MMIO. */
-	u32 mmio_size;             /**< Size of the MMIO region. */
+	u32 mmio_base;			/**< Bus address of base of MMIO. */
+	u32 mmio_size;			/**< Size of the MMIO region. */
 	/*@}*/
 
 	u32 clear_cmd;
 	u32 maccess;
 
+	atomic_t vbl_received;		/**< Number of vblanks received. */
 	wait_queue_head_t fence_queue;
 	atomic_t last_fence_retired;
 	u32 next_fence_to_post;
@@ -181,11 +182,14 @@ extern int mga_warp_install_microcode(drm_mga_private_t * dev_priv);
 extern int mga_warp_init(drm_mga_private_t * dev_priv);
 
 				/* mga_irq.c */
+extern int mga_enable_vblank(struct drm_device *dev, int crtc);
+extern void mga_disable_vblank(struct drm_device *dev, int crtc);
+extern u32 mga_get_vblank_counter(struct drm_device *dev, int crtc);
 extern int mga_driver_fence_wait(struct drm_device * dev, unsigned int *sequence);
 extern int mga_driver_vblank_wait(struct drm_device * dev, unsigned int *sequence);
 extern irqreturn_t mga_driver_irq_handler(DRM_IRQ_ARGS);
 extern void mga_driver_irq_preinstall(struct drm_device * dev);
-extern void mga_driver_irq_postinstall(struct drm_device * dev);
+extern int mga_driver_irq_postinstall(struct drm_device * dev);
 extern void mga_driver_irq_uninstall(struct drm_device * dev);
 extern long mga_compat_ioctl(struct file *filp, unsigned int cmd,
 			     unsigned long arg);
@@ -249,7 +253,7 @@ do {									\
 		} else if ( dev_priv->prim.space <			\
 			    dev_priv->prim.high_mark ) {		\
 			if ( MGA_DMA_DEBUG )				\
-				DRM_INFO( "%s: wrap...\n", __FUNCTION__ );	\
+				DRM_INFO( "wrap...\n");		\
 			return -EBUSY;			\
 		}							\
 	}								\
@@ -260,7 +264,7 @@ do {									\
 	if ( test_bit( 0, &dev_priv->prim.wrapped ) ) {			\
 		if ( mga_do_wait_for_idle( dev_priv ) < 0 ) {		\
 			if ( MGA_DMA_DEBUG )				\
-				DRM_INFO( "%s: wrap...\n", __FUNCTION__ );	\
+				DRM_INFO( "wrap...\n");		\
 			return -EBUSY;			\
 		}							\
 		mga_do_dma_wrap_end( dev_priv );			\
@@ -280,8 +284,7 @@ do {									\
 #define BEGIN_DMA( n )							\
 do {									\
 	if ( MGA_VERBOSE ) {						\
-		DRM_INFO( "BEGIN_DMA( %d ) in %s\n",			\
-			  (n), __FUNCTION__ );				\
+		DRM_INFO( "BEGIN_DMA( %d )\n", (n) );		\
 		DRM_INFO( "   space=0x%x req=0x%Zx\n",			\
 			  dev_priv->prim.space, (n) * DMA_BLOCK_SIZE );	\
 	}								\
@@ -292,7 +295,7 @@ do {									\
 #define BEGIN_DMA_WRAP()						\
 do {									\
 	if ( MGA_VERBOSE ) {						\
-		DRM_INFO( "BEGIN_DMA() in %s\n", __FUNCTION__ );		\
+		DRM_INFO( "BEGIN_DMA()\n" );				\
 		DRM_INFO( "   space=0x%x\n", dev_priv->prim.space );	\
 	}								\
 	prim = dev_priv->prim.start;					\
@@ -311,7 +314,7 @@ do {									\
 #define FLUSH_DMA()							\
 do {									\
 	if ( 0 ) {							\
-		DRM_INFO( "%s:\n", __FUNCTION__ );				\
+		DRM_INFO( "\n" );					\
 		DRM_INFO( "   tail=0x%06x head=0x%06lx\n",		\
 			  dev_priv->prim.tail,				\
 			  MGA_READ( MGA_PRIMADDRESS ) -			\
diff --git a/shared-core/mga_irq.c b/shared-core/mga_irq.c
index 0f83577f..4fe86322 100644
--- a/shared-core/mga_irq.c
+++ b/shared-core/mga_irq.c
@@ -36,6 +36,20 @@
 #include "mga_drm.h"
 #include "mga_drv.h"
 
+u32 mga_get_vblank_counter(struct drm_device *dev, int crtc)
+{
+	const drm_mga_private_t *const dev_priv = 
+		(drm_mga_private_t *) dev->dev_private;
+
+	if (crtc != 0) {
+		return 0;
+	}
+
+
+	return atomic_read(&dev_priv->vbl_received);
+}
+
+
 irqreturn_t mga_driver_irq_handler(DRM_IRQ_ARGS)
 {
 	struct drm_device *dev = (struct drm_device *) arg;
@@ -48,9 +62,8 @@ irqreturn_t mga_driver_irq_handler(DRM_IRQ_ARGS)
 	/* VBLANK interrupt */
 	if (status & MGA_VLINEPEN) {
 		MGA_WRITE(MGA_ICLEAR, MGA_VLINEICLR);
-		atomic_inc(&dev->vbl_received);
-		DRM_WAKEUP(&dev->vbl_queue);
-		drm_vbl_send_signals(dev);
+		atomic_inc(&dev_priv->vbl_received);
+		drm_handle_vblank(dev, 0);
 		handled = 1;
 	}
 
@@ -79,22 +92,34 @@ irqreturn_t mga_driver_irq_handler(DRM_IRQ_ARGS)
 	return IRQ_NONE;
 }
 
-int mga_driver_vblank_wait(struct drm_device * dev, unsigned int *sequence)
+int mga_enable_vblank(struct drm_device *dev, int crtc)
 {
-	unsigned int cur_vblank;
-	int ret = 0;
+	drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private;
 
-	/* Assume that the user has missed the current sequence number
-	 * by about a day rather than she wants to wait for years
-	 * using vertical blanks...
-	 */
-	DRM_WAIT_ON(ret, dev->vbl_queue, 3 * DRM_HZ,
-		    (((cur_vblank = atomic_read(&dev->vbl_received))
-		      - *sequence) <= (1 << 23)));
+	if (crtc != 0) {
+		DRM_ERROR("tried to enable vblank on non-existent crtc %d\n",
+			  crtc);
+		return 0;
+	}
 
-	*sequence = cur_vblank;
+	MGA_WRITE(MGA_IEN, MGA_VLINEIEN | MGA_SOFTRAPEN);
+	return 0;
+}
 
-	return ret;
+
+void mga_disable_vblank(struct drm_device *dev, int crtc)
+{
+	if (crtc != 0) {
+		DRM_ERROR("tried to disable vblank on non-existent crtc %d\n",
+			  crtc);
+	}
+
+	/* Do *NOT* disable the vertical refresh interrupt.  MGA doesn't have
+	 * a nice hardware counter that tracks the number of refreshes when
+	 * the interrupt is disabled, and the kernel doesn't know the refresh
+	 * rate to calculate an estimate.
+	 */
+	/* MGA_WRITE(MGA_IEN, MGA_VLINEIEN | MGA_SOFTRAPEN); */
 }
 
 int mga_driver_fence_wait(struct drm_device * dev, unsigned int *sequence)
@@ -126,14 +151,22 @@ void mga_driver_irq_preinstall(struct drm_device * dev)
 	MGA_WRITE(MGA_ICLEAR, ~0);
 }
 
-void mga_driver_irq_postinstall(struct drm_device * dev)
+int mga_driver_irq_postinstall(struct drm_device * dev)
 {
 	drm_mga_private_t *dev_priv = (drm_mga_private_t *) dev->dev_private;
+	int ret;
+
+	ret = drm_vblank_init(dev, 1);
+	if (ret)
+		return ret;
 
 	DRM_INIT_WAITQUEUE(&dev_priv->fence_queue);
 
-	/* Turn on vertical blank interrupt and soft trap interrupt. */
-	MGA_WRITE(MGA_IEN, MGA_VLINEIEN | MGA_SOFTRAPEN);
+	/* Turn on soft trap interrupt.  Vertical blank interrupts are enabled
+	 * in mga_enable_vblank.
+	 */
+	MGA_WRITE(MGA_IEN, MGA_SOFTRAPEN);
+	return 0;
 }
 
 void mga_driver_irq_uninstall(struct drm_device * dev)
diff --git a/shared-core/mga_state.c b/shared-core/mga_state.c
index d9cc9d7e..2da108be 100644
--- a/shared-core/mga_state.c
+++ b/shared-core/mga_state.c
@@ -644,7 +644,7 @@ static void mga_dma_dispatch_swap(struct drm_device * dev)
 
 	FLUSH_DMA();
 
-	DRM_DEBUG("%s... done.\n", __FUNCTION__);
+	DRM_DEBUG("... done.\n");
 }
 
 static void mga_dma_dispatch_vertex(struct drm_device * dev, struct drm_buf * buf)
@@ -656,7 +656,7 @@ static void mga_dma_dispatch_vertex(struct drm_device * dev, struct drm_buf * bu
 	u32 length = (u32) buf->used;
 	int i = 0;
 	DMA_LOCALS;
-	DRM_DEBUG("vertex: buf=%d used=%d\n", buf->idx, buf->used);
+	DRM_DEBUG("buf=%d used=%d\n", buf->idx, buf->used);
 
 	if (buf->used) {
 		buf_priv->dispatched = 1;
@@ -703,7 +703,7 @@ static void mga_dma_dispatch_indices(struct drm_device * dev, struct drm_buf * b
 	u32 address = (u32) buf->bus_address;
 	int i = 0;
 	DMA_LOCALS;
-	DRM_DEBUG("indices: buf=%d start=%d end=%d\n", buf->idx, start, end);
+	DRM_DEBUG("buf=%d start=%d end=%d\n", buf->idx, start, end);
 
 	if (start != end) {
 		buf_priv->dispatched = 1;
@@ -988,7 +988,7 @@ static int mga_dma_iload(struct drm_device *dev, void *data, struct drm_file *fi
 #if 0
 	if (mga_do_wait_for_idle(dev_priv) < 0) {
 		if (MGA_DMA_DEBUG)
-			DRM_INFO("%s: -EBUSY\n", __FUNCTION__);
+			DRM_INFO("-EBUSY\n");
 		return -EBUSY;
 	}
 #endif
@@ -1047,7 +1047,7 @@ static int mga_getparam(struct drm_device *dev, void *data, struct drm_file *fil
 	int value;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -1079,7 +1079,7 @@ static int mga_set_fence(struct drm_device *dev, void *data, struct drm_file *fi
 	DMA_LOCALS;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -1108,7 +1108,7 @@ static int mga_wait_fence(struct drm_device *dev, void *data, struct drm_file *f
 	u32 *fence = data;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
diff --git a/shared-core/nouveau_drv.h b/shared-core/nouveau_drv.h
index 85a0d0b5..4184aa5b 100644
--- a/shared-core/nouveau_drv.h
+++ b/shared-core/nouveau_drv.h
@@ -432,7 +432,7 @@ extern int nouveau_ioctl_gpuobj_free(struct drm_device *, void *data,
 /* nouveau_irq.c */
 extern irqreturn_t nouveau_irq_handler(DRM_IRQ_ARGS);
 extern void        nouveau_irq_preinstall(struct drm_device *);
-extern void        nouveau_irq_postinstall(struct drm_device *);
+extern int         nouveau_irq_postinstall(struct drm_device *);
 extern void        nouveau_irq_uninstall(struct drm_device *);
 
 /* nouveau_sgdma.c */
diff --git a/shared-core/nouveau_irq.c b/shared-core/nouveau_irq.c
index 500fda2f..bceb81ab 100644
--- a/shared-core/nouveau_irq.c
+++ b/shared-core/nouveau_irq.c
@@ -46,13 +46,15 @@ nouveau_irq_preinstall(struct drm_device *dev)
 	NV_WRITE(NV03_PMC_INTR_EN_0, 0);
 }
 
-void
+int
 nouveau_irq_postinstall(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 
 	/* Master enable */
 	NV_WRITE(NV03_PMC_INTR_EN_0, NV_PMC_INTR_EN_0_MASTER_ENABLE);
+
+	return 0;
 }
 
 void
diff --git a/shared-core/nouveau_mem.c b/shared-core/nouveau_mem.c
index 5259ea27..f4e641b9 100644
--- a/shared-core/nouveau_mem.c
+++ b/shared-core/nouveau_mem.c
@@ -619,7 +619,7 @@ alloc_ok:
 		block->map_handle = entry->user_token;
 	}
 
-	DRM_DEBUG("allocated 0x%llx type=0x%08x\n", block->start, block->flags);
+	DRM_DEBUG("allocated %lld bytes at 0x%llx type=0x%08x\n", block->size, block->start, block->flags);
 	return block;
 }
 
diff --git a/shared-core/nouveau_reg.h b/shared-core/nouveau_reg.h
index 7ac02f4a..a2506146 100644
--- a/shared-core/nouveau_reg.h
+++ b/shared-core/nouveau_reg.h
@@ -112,6 +112,9 @@
 #define NV50_PUNK_BAR3_CTXDMA_VALID                           (1<<31)
 #define NV50_PUNK_UNK1710                                  0x00001710
 
+#define NV04_PBUS_PCI_NV_1                                 0x00001804
+#define NV04_PBUS_PCI_NV_19                                0x0000184C
+
 #define NV04_PTIMER_INTR_0                                 0x00009100
 #define NV04_PTIMER_INTR_EN_0                              0x00009140
 #define NV04_PTIMER_NUMERATOR                              0x00009200
diff --git a/shared-core/nouveau_state.c b/shared-core/nouveau_state.c
index 16c86494..7086a0ab 100644
--- a/shared-core/nouveau_state.c
+++ b/shared-core/nouveau_state.c
@@ -88,7 +88,6 @@ static int nouveau_init_card_mappings(struct drm_device *dev)
 
 static int nouveau_stub_init(struct drm_device *dev) { return 0; }
 static void nouveau_stub_takedown(struct drm_device *dev) {}
-static uint64_t nouveau_stub_timer_read(struct drm_device *dev) { return 0; }
 
 static int nouveau_init_engine_ptrs(struct drm_device *dev)
 {
@@ -251,9 +250,9 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
 		engine->instmem.unbind		= nv50_instmem_unbind;
 		engine->mc.init		= nv50_mc_init;
 		engine->mc.takedown	= nv50_mc_takedown;
-		engine->timer.init	= nouveau_stub_init;
-		engine->timer.read	= nouveau_stub_timer_read;
-		engine->timer.takedown	= nouveau_stub_takedown;
+		engine->timer.init	= nv04_timer_init;
+		engine->timer.read	= nv04_timer_read;
+		engine->timer.takedown	= nv04_timer_takedown;
 		engine->fb.init		= nouveau_stub_init;
 		engine->fb.takedown	= nouveau_stub_takedown;
 		engine->graph.init	= nv50_graph_init;
@@ -285,9 +284,6 @@ nouveau_card_init(struct drm_device *dev)
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
 	struct nouveau_engine *engine;
 	int ret;
-#if defined(__powerpc__)
-	struct device_node *dn;
-#endif
 
 	DRM_DEBUG("prev state = %d\n", dev_priv->init_state);
 
@@ -295,43 +291,6 @@ nouveau_card_init(struct drm_device *dev)
 		return 0;
 	dev_priv->ttm = 0;
 
-	/* Map any PCI resources we need on the card */
-	ret = nouveau_init_card_mappings(dev);
-	if (ret) return ret;
-
-#if defined(__powerpc__)
-	/* Put the card in BE mode if it's not */
-	if (NV_READ(NV03_PMC_BOOT_1))
-		NV_WRITE(NV03_PMC_BOOT_1,0x00000001);
-
-	DRM_MEMORYBARRIER();
-#endif
-
-#if defined(__linux__) && defined(__powerpc__)
-	/* if we have an OF card, copy vbios to RAMIN */
-	dn = pci_device_to_OF_node(dev->pdev);
-	if (dn)
-	{
-		int size;
-#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22))
-		const uint32_t *bios = of_get_property(dn, "NVDA,BMP", &size);
-#else
-		const uint32_t *bios = get_property(dn, "NVDA,BMP", &size);
-#endif
-		if (bios)
-		{
-			int i;
-			for(i=0;i<size;i+=4)
-				NV_WI32(i, bios[i/4]);
-			DRM_INFO("OF bios successfully copied (%d bytes)\n",size);
-		}
-		else
-			DRM_INFO("Unable to get the OF bios\n");
-	}
-	else
-		DRM_INFO("Unable to get the OF node\n");
-#endif
-
 	/* Determine exact chipset we're running on */
 	if (dev_priv->card_type < NV_10)
 		dev_priv->chipset = dev_priv->card_type;
@@ -451,6 +410,47 @@ void nouveau_preclose(struct drm_device *dev, struct drm_file *file_priv)
 /* first module load, setup the mmio/fb mapping */
 int nouveau_firstopen(struct drm_device *dev)
 {
+#if defined(__powerpc__)
+	struct drm_nouveau_private *dev_priv = dev->dev_private;
+	struct device_node *dn;
+#endif
+	int ret;
+	/* Map any PCI resources we need on the card */
+	ret = nouveau_init_card_mappings(dev);
+	if (ret) return ret;
+
+#if defined(__powerpc__)
+	/* Put the card in BE mode if it's not */
+	if (NV_READ(NV03_PMC_BOOT_1))
+		NV_WRITE(NV03_PMC_BOOT_1,0x00000001);
+
+	DRM_MEMORYBARRIER();
+#endif
+
+#if defined(__linux__) && defined(__powerpc__)
+	/* if we have an OF card, copy vbios to RAMIN */
+	dn = pci_device_to_OF_node(dev->pdev);
+	if (dn)
+	{
+		int size;
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22))
+		const uint32_t *bios = of_get_property(dn, "NVDA,BMP", &size);
+#else
+		const uint32_t *bios = get_property(dn, "NVDA,BMP", &size);
+#endif
+		if (bios)
+		{
+			int i;
+			for(i=0;i<size;i+=4)
+				NV_WI32(i, bios[i/4]);
+			DRM_INFO("OF bios successfully copied (%d bytes)\n",size);
+		}
+		else
+			DRM_INFO("Unable to get the OF bios\n");
+	}
+	else
+		DRM_INFO("Unable to get the OF node\n");
+#endif
 	return 0;
 }
 
@@ -498,7 +498,12 @@ int nouveau_load(struct drm_device *dev, unsigned long flags)
 
 	iounmap(regs);
 
-	if (architecture >= 0x50) {
+	if (architecture >= 0x80) {
+		dev_priv->card_type = NV_50;
+	} else if (architecture >= 0x60) {
+		/* FIXME we need to figure out who's who for NV6x */
+		dev_priv->card_type = NV_44;
+	} else if (architecture >= 0x50) {
 		dev_priv->card_type = NV_50;
 	} else if (architecture >= 0x40) {
 		uint8_t subarch = architecture & 0xf;
diff --git a/shared-core/nv04_graph.c b/shared-core/nv04_graph.c
index 81a6d5c8..6caae257 100644
--- a/shared-core/nv04_graph.c
+++ b/shared-core/nv04_graph.c
@@ -497,7 +497,7 @@ int nv04_graph_init(struct drm_device *dev) {
 	/*haiku same*/
 
 	/*NV_WRITE(NV04_PGRAPH_DEBUG_3, 0xfad4ff31);*/
-	NV_WRITE(NV04_PGRAPH_DEBUG_3, 0x10d4ff31);
+	NV_WRITE(NV04_PGRAPH_DEBUG_3, 0xf0d4ff31);
 	/*haiku and blob 10d4*/
 
 	NV_WRITE(NV04_PGRAPH_STATE        , 0xFFFFFFFF);
diff --git a/shared-core/nv04_mc.c b/shared-core/nv04_mc.c
index 24c1f7b3..766f3a33 100644
--- a/shared-core/nv04_mc.c
+++ b/shared-core/nv04_mc.c
@@ -7,12 +7,25 @@ int
 nv04_mc_init(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
+	uint32_t saved_pci_nv_1, saved_pci_nv_19;
+
+	saved_pci_nv_1 = NV_READ(NV04_PBUS_PCI_NV_1);
+	saved_pci_nv_19 = NV_READ(NV04_PBUS_PCI_NV_19);
+
+	/* clear busmaster bit */
+	NV_WRITE(NV04_PBUS_PCI_NV_1, saved_pci_nv_1 & ~(0x00000001 << 2));
+	/* clear SBA and AGP bits */
+	NV_WRITE(NV04_PBUS_PCI_NV_19, saved_pci_nv_19 & 0xfffff0ff);
 
 	/* Power up everything, resetting each individual unit will
 	 * be done later if needed.
 	 */
 	NV_WRITE(NV03_PMC_ENABLE, 0xFFFFFFFF);
 
+	/* and restore (gives effect of resetting AGP) */
+	NV_WRITE(NV04_PBUS_PCI_NV_19, saved_pci_nv_19);
+	NV_WRITE(NV04_PBUS_PCI_NV_1, saved_pci_nv_1);
+
 	return 0;
 }
 
diff --git a/shared-core/nv20_graph.c b/shared-core/nv20_graph.c
index 81501ed7..37a147b5 100644
--- a/shared-core/nv20_graph.c
+++ b/shared-core/nv20_graph.c
@@ -432,19 +432,6 @@ static void nv30_31_graph_context_init(struct drm_device *dev,
 	INSTANCE_WR(ctx, 0x385c/4, 0x3f800000);
 	INSTANCE_WR(ctx, 0x3864/4, 0xbf800000);
 	INSTANCE_WR(ctx, 0x386c/4, 0xbf800000);
-
-	/* nv30gl stuff */
-	for (i=0; i<8; i++) {
-		INSTANCE_WR(ctx, (0x4dfc/4)+i, 0x001c527d);
-	}
-	INSTANCE_WR(ctx, 0x4e3c/4, 0x001c527c);
-/* these ones make dma fifo hang
-	INSTANCE_WR(ctx, 0x567c/4, 0x000a0000);
-	INSTANCE_WR(ctx, 0x0878/4, 0x01000000);
-	INSTANCE_WR(ctx, 0x02f4/4, 0x0001ffff);
-
-	INSTANCE_WR(ctx, 0x0028/4, INSTANCE_RD(ctx, 0x0028/4) | 1);
-*/
 }
 
 static void nv34_graph_context_init(struct drm_device *dev,
@@ -655,6 +642,7 @@ int nv20_graph_load_context(struct nouveau_channel *chan)
 	NV_WRITE(NV20_PGRAPH_CHANNEL_CTX_POINTER, inst);
 	NV_WRITE(NV20_PGRAPH_CHANNEL_CTX_XFER,
 		 NV20_PGRAPH_CHANNEL_CTX_XFER_LOAD);
+	NV_WRITE(NV10_PGRAPH_CTX_CONTROL, 0x10010100);
 
 	nouveau_wait_for_idle(dev);
 	return 0;
@@ -772,7 +760,7 @@ int nv20_graph_init(struct drm_device *dev) {
 	NV_WRITE(NV10_PGRAPH_RDI_INDEX, 0x00EA000C);
 	NV_WRITE(NV10_PGRAPH_RDI_DATA, NV_READ(0x100324));
 
-	NV_WRITE(NV10_PGRAPH_CTX_CONTROL, 0x10010100);
+	NV_WRITE(NV10_PGRAPH_CTX_CONTROL, 0x10000100);
 	NV_WRITE(NV10_PGRAPH_STATE      , 0xFFFFFFFF);
 	NV_WRITE(NV04_PGRAPH_FIFO       , 0x00000001);
 
@@ -867,7 +855,7 @@ int nv30_graph_init(struct drm_device *dev)
 		NV_WRITE(NV10_PGRAPH_TSTATUS(i), NV_READ(NV10_PFB_TSTATUS(i)));
 	}
 
-	NV_WRITE(NV10_PGRAPH_CTX_CONTROL, 0x10010100);
+	NV_WRITE(NV10_PGRAPH_CTX_CONTROL, 0x10000100);
 	NV_WRITE(NV10_PGRAPH_STATE      , 0xFFFFFFFF);
 	NV_WRITE(NV04_PGRAPH_FIFO       , 0x00000001);
 
diff --git a/shared-core/nv50_fifo.c b/shared-core/nv50_fifo.c
index c5cde913..edf4edbf 100644
--- a/shared-core/nv50_fifo.c
+++ b/shared-core/nv50_fifo.c
@@ -28,9 +28,10 @@
 #include "drm.h"
 #include "nouveau_drv.h"
 
-typedef struct {
-	struct nouveau_gpuobj_ref *thingo;
-} nv50_fifo_priv;
+struct nv50_fifo_priv {
+	struct nouveau_gpuobj_ref *thingo[2];
+	int cur_thingo;
+};
 
 #define IS_G80 ((dev_priv->chipset & 0xf0) == 0x50)
 
@@ -38,23 +39,23 @@ static void
 nv50_fifo_init_thingo(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	nv50_fifo_priv *priv = dev_priv->Engine.fifo.priv;
-	struct nouveau_gpuobj_ref *thingo = priv->thingo;
-	int i, fi=2;
+	struct nv50_fifo_priv *priv = dev_priv->Engine.fifo.priv;
+	struct nouveau_gpuobj_ref *cur;
+	int i, nr;
 
 	DRM_DEBUG("\n");
 
-	INSTANCE_WR(thingo->gpuobj, 0, 0x7e);
-	INSTANCE_WR(thingo->gpuobj, 1, 0x7e);
-	for (i = 1; i < 127; i++, fi) {
+	cur = priv->thingo[priv->cur_thingo];
+	priv->cur_thingo = !priv->cur_thingo;
+
+	/* We never schedule channel 0 or 127 */
+	for (i = 1, nr = 0; i < 127; i++) {
 		if (dev_priv->fifos[i]) {
-			INSTANCE_WR(thingo->gpuobj, fi, i);
-			fi++;
+			INSTANCE_WR(cur->gpuobj, nr++, i);
 		}
 	}
-
-	NV_WRITE(0x32f4, thingo->instance >> 12);
-	NV_WRITE(0x32ec, fi);
+	NV_WRITE(0x32f4, cur->instance >> 12);
+	NV_WRITE(0x32ec, nr);
 	NV_WRITE(0x2500, 0x101);
 }
 
@@ -98,14 +99,12 @@ static void
 nv50_fifo_init_reset(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	uint32_t pmc_e;
+	uint32_t pmc_e = NV_PMC_ENABLE_PFIFO;
 
 	DRM_DEBUG("\n");
 
-	pmc_e = NV_READ(NV03_PMC_ENABLE);
-	NV_WRITE(NV03_PMC_ENABLE, pmc_e & ~NV_PMC_ENABLE_PFIFO);
-	pmc_e = NV_READ(NV03_PMC_ENABLE);
-	NV_WRITE(NV03_PMC_ENABLE, pmc_e |  NV_PMC_ENABLE_PFIFO);
+	NV_WRITE(NV03_PMC_ENABLE, NV_READ(NV03_PMC_ENABLE) & ~pmc_e);
+	NV_WRITE(NV03_PMC_ENABLE, NV_READ(NV03_PMC_ENABLE) |  pmc_e);
 }
 
 static void
@@ -141,7 +140,7 @@ nv50_fifo_init_regs__nv(struct drm_device *dev)
 	NV_WRITE(0x250c, 0x6f3cfc34);
 }
 
-static int
+static void
 nv50_fifo_init_regs(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
@@ -158,15 +157,13 @@ nv50_fifo_init_regs(struct drm_device *dev)
 	/* Enable dummy channels setup by nv50_instmem.c */
 	nv50_fifo_channel_enable(dev, 0, 1);
 	nv50_fifo_channel_enable(dev, 127, 1);
-
-	return 0;
 }
 
 int
 nv50_fifo_init(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	nv50_fifo_priv *priv;
+	struct nv50_fifo_priv *priv;
 	int ret;
 
 	DRM_DEBUG("\n");
@@ -179,18 +176,23 @@ nv50_fifo_init(struct drm_device *dev)
 	nv50_fifo_init_reset(dev);
 	nv50_fifo_init_intr(dev);
 
-	if ((ret = nouveau_gpuobj_new_ref(dev, NULL, NULL, 0, (128+2)*4, 0x1000,
-				   NVOBJ_FLAG_ZERO_ALLOC,
-				   &priv->thingo))) {
-		DRM_ERROR("error creating thingo: %d\n", ret);
+	ret = nouveau_gpuobj_new_ref(dev, NULL, NULL, 0, 128*4, 0x1000,
+				     NVOBJ_FLAG_ZERO_ALLOC, &priv->thingo[0]);
+	if (ret) {
+		DRM_ERROR("error creating thingo0: %d\n", ret);
 		return ret;
 	}
 
-	nv50_fifo_init_context_table(dev);
+	ret = nouveau_gpuobj_new_ref(dev, NULL, NULL, 0, 128*4, 0x1000,
+				     NVOBJ_FLAG_ZERO_ALLOC, &priv->thingo[1]);
+	if (ret) {
+		DRM_ERROR("error creating thingo1: %d\n", ret);
+		return ret;
+	}
 
+	nv50_fifo_init_context_table(dev);
 	nv50_fifo_init_regs__nv(dev);
-	if ((ret = nv50_fifo_init_regs(dev)))
-		return ret;
+	nv50_fifo_init_regs(dev);
 
 	return 0;
 }
@@ -199,14 +201,15 @@ void
 nv50_fifo_takedown(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	nv50_fifo_priv *priv = dev_priv->Engine.fifo.priv;
+	struct nv50_fifo_priv *priv = dev_priv->Engine.fifo.priv;
 
 	DRM_DEBUG("\n");
 
 	if (!priv)
 		return;
 
-	nouveau_gpuobj_ref_del(dev, &priv->thingo);
+	nouveau_gpuobj_ref_del(dev, &priv->thingo[0]);
+	nouveau_gpuobj_ref_del(dev, &priv->thingo[1]);
 
 	dev_priv->Engine.fifo.priv = NULL;
 	drm_free(priv, sizeof(*priv), DRM_MEM_DRIVER);
@@ -234,18 +237,18 @@ nv50_fifo_create_context(struct nouveau_channel *chan)
 	if (IS_G80) {
 		uint32_t ramfc_offset = chan->ramin->gpuobj->im_pramin->start;
 		uint32_t vram_offset = chan->ramin->gpuobj->im_backing->start;
-		if ((ret = nouveau_gpuobj_new_fake(dev, ramfc_offset,
-						   vram_offset, 0x100,
-						   NVOBJ_FLAG_ZERO_ALLOC |
-						   NVOBJ_FLAG_ZERO_FREE,
-						   &ramfc, &chan->ramfc)))
-				return ret;
+		ret = nouveau_gpuobj_new_fake(dev, ramfc_offset, vram_offset,
+					      0x100, NVOBJ_FLAG_ZERO_ALLOC |
+					      NVOBJ_FLAG_ZERO_FREE, &ramfc,
+					      &chan->ramfc);
+		if (ret)
+			return ret;
 	} else {
-		if ((ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, 0x100,
-						  256,
-						  NVOBJ_FLAG_ZERO_ALLOC |
-						  NVOBJ_FLAG_ZERO_FREE,
-						  &chan->ramfc)))
+		ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, 0x100, 256,
+					     NVOBJ_FLAG_ZERO_ALLOC |
+					     NVOBJ_FLAG_ZERO_FREE,
+					     &chan->ramfc);
+		if (ret)
 			return ret;
 		ramfc = chan->ramfc->gpuobj;
 	}
@@ -272,7 +275,8 @@ nv50_fifo_create_context(struct nouveau_channel *chan)
 		INSTANCE_WR(ramfc, 0x98/4, chan->ramin->instance >> 12);
 	}
 
-	if ((ret = nv50_fifo_channel_enable(dev, chan->id, 0))) {
+	ret = nv50_fifo_channel_enable(dev, chan->id, 0);
+	if (ret) {
 		DRM_ERROR("error enabling ch%d: %d\n", chan->id, ret);
 		nouveau_gpuobj_ref_del(dev, &chan->ramfc);
 		return ret;
diff --git a/shared-core/nv50_graph.c b/shared-core/nv50_graph.c
index 5814abd4..503f45dd 100644
--- a/shared-core/nv50_graph.c
+++ b/shared-core/nv50_graph.c
@@ -34,14 +34,12 @@ static void
 nv50_graph_init_reset(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	uint32_t pmc_e;
+	uint32_t pmc_e = NV_PMC_ENABLE_PGRAPH | (1 << 21);
 
 	DRM_DEBUG("\n");
 
-	pmc_e = NV_READ(NV03_PMC_ENABLE);
-	NV_WRITE(NV03_PMC_ENABLE, pmc_e & ~NV_PMC_ENABLE_PGRAPH);
-	pmc_e = NV_READ(NV03_PMC_ENABLE);
-	NV_WRITE(NV03_PMC_ENABLE, pmc_e |  NV_PMC_ENABLE_PGRAPH);
+	NV_WRITE(NV03_PMC_ENABLE, NV_READ(NV03_PMC_ENABLE) & ~pmc_e);
+	NV_WRITE(NV03_PMC_ENABLE, NV_READ(NV03_PMC_ENABLE) |  pmc_e);
 }
 
 static void
@@ -51,6 +49,7 @@ nv50_graph_init_intr(struct drm_device *dev)
 
 	DRM_DEBUG("\n");
 	NV_WRITE(NV03_PGRAPH_INTR, 0xffffffff);
+	NV_WRITE(0x400138, 0xffffffff);
 	NV_WRITE(NV40_PGRAPH_INTR_EN, 0xffffffff);
 }
 
@@ -146,12 +145,53 @@ static uint32_t nv84_ctx_voodoo[] = {
 	0x00415e06, 0x00415f05, 0x0060000d, 0x00700005, 0x0070000d, 0x00700006,
 	0x0070000b, 0x0070000e, 0x0070001c, 0x0060000c, ~0
 };
+ 
+static uint32_t nv86_ctx_voodoo[] = {
+	0x0070008e, 0x0070009c, 0x00200020, 0x00600008, 0x0050004c, 0x00400e89, 
+	0x00200000, 0x00600007, 0x00300000, 0x00c000ff, 0x00200000, 0x008000ff, 
+	0x00700009, 0x0040dd4d, 0x00402944, 0x00402905, 0x0040290d, 0x0040b906, 
+	0x00600005, 0x004015c5, 0x00600011, 0x0040270b, 0x004021c5, 0x00700000, 
+	0x00700081, 0x00600004, 0x0050004a, 0x00216d80, 0x00600007, 0x00c02801, 
+	0x0020002e, 0x00800001, 0x005000cb, 0x0090ffff, 0x0091ffff, 0x00200020, 
+	0x00600008, 0x0050004c, 0x00600009, 0x0040b945, 0x0040d44d, 0x0070009d, 
+	0x00402dcf, 0x0070009f, 0x0050009f, 0x00402ac0, 0x00200200, 0x00600008, 
+	0x00402a4f, 0x00402ac0, 0x004030cc, 0x00700081, 0x00200000, 0x00600006, 
+	0x00700000, 0x00111bfc, 0x00700083, 0x00300000, 0x00216d80, 0x00600007, 
+	0x00c00b01, 0x0020001e, 0x00800001, 0x005000cb, 0x00c000ff, 0x00700080, 
+	0x00700083, 0x00200047, 0x00600006, 0x0011020a, 0x00200280, 0x00600007, 
+	0x00300000, 0x00c000ff, 0x00c800ff, 0x0040c407, 0x00202916, 0x008000ff, 
+	0x0040508c, 0x005000cb, 0x00a0023f, 0x00200040, 0x00600006, 0x0070000f, 
+	0x00170202, 0x0011020a, 0x00200032, 0x0010020d, 0x001c0242, 0x00120302, 
+	0x00140402, 0x00180500, 0x00130509, 0x00150550, 0x00110605, 0x0020000f, 
+	0x00100607, 0x00110700, 0x00110900, 0x00120902, 0x00110a00, 0x00160b02, 
+	0x00120b28, 0x00140b2b, 0x00110c01, 0x00111400, 0x00111405, 0x00111407, 
+	0x00111409, 0x0011140b, 0x002000cb, 0x00101500, 0x0040790f, 0x0040794b, 
+	0x00214b40, 0x00600007, 0x00200442, 0x008800ff, 0x0070008f, 0x0040798c, 
+	0x005000cb, 0x00000000, 0x0020002b, 0x00101a05, 0x00131c00, 0x00121c04, 
+	0x00141c20, 0x00111c25, 0x00131c40, 0x00121c44, 0x00141c60, 0x00111c65, 
+	0x00131f00, 0x00191f40, 0x004099e0, 0x002001d9, 0x00600006, 0x00200044, 
+	0x00102080, 0x001120c6, 0x001520c9, 0x001920d0, 0x00122100, 0x00122103, 
+	0x00162200, 0x00122207, 0x00112280, 0x00112300, 0x00112302, 0x00122380, 
+	0x0011238b, 0x00112394, 0x0011239c, 0x00000000, 0x0040a00f, 0x005000cb, 
+	0x00214b40, 0x00600007, 0x00200442, 0x008800ff, 0x005000cb, 0x0040a387, 
+	0x0060000a, 0x00000000, 0x0040b200, 0x007000a0, 0x00700080, 0x00200280, 
+	0x00600007, 0x00200004, 0x00c000ff, 0x008000ff, 0x005000cb, 0x00700000, 
+	0x00200000, 0x00600006, 0x00111bfe, 0x0040d44d, 0x00700000, 0x00200000, 
+	0x00600006, 0x00111bfe, 0x00700080, 0x0070001d, 0x0040114d, 0x00700081, 
+	0x00600004, 0x0050004a, 0x0040be88, 0x0060000b, 0x00200000, 0x00600006, 
+	0x00700000, 0x0040d40b, 0x00111bfd, 0x0040424d, 0x00202916, 0x008000fd, 
+	0x005000cb, 0x00c00002, 0x00200280, 0x00600007, 0x00200160, 0x00800002, 
+	0x005000cb, 0x00c01802, 0x002027b6, 0x00800002, 0x005000cb, 0x00404e4d, 
+	0x0060000b, 0x0040d24d, 0x00700001, 0x00700003, 0x0040d806, 0x0040d905, 
+	0x0060000d, 0x00700005, 0x0070000d, 0x00700006, 0x0070000b, 0x0070000e, 
+	0x0070001c, 0x0060000c, ~0
+};
 
-static void
+static int
 nv50_graph_init_ctxctl(struct drm_device *dev)
 {
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	uint32_t *voodoo;
+	uint32_t *voodoo = NULL;
 
 	DRM_DEBUG("\n");
 
@@ -159,34 +199,42 @@ nv50_graph_init_ctxctl(struct drm_device *dev)
 	case 0x84:
 		voodoo = nv84_ctx_voodoo;
 		break;
+	case 0x86:
+		voodoo = nv86_ctx_voodoo;
+		break;
 	default:
 		DRM_ERROR("no voodoo for chipset NV%02x\n", dev_priv->chipset);
-		break;
+		return -EINVAL;
 	}
 
-	if (voodoo) {
-		NV_WRITE(NV40_PGRAPH_CTXCTL_UCODE_INDEX, 0);
-		while (*voodoo != ~0) {
-			NV_WRITE(NV40_PGRAPH_CTXCTL_UCODE_DATA, *voodoo);
-			voodoo++;
-		}
+	NV_WRITE(NV40_PGRAPH_CTXCTL_UCODE_INDEX, 0);
+	while (*voodoo != ~0) {
+		NV_WRITE(NV40_PGRAPH_CTXCTL_UCODE_DATA, *voodoo);
+		voodoo++;
 	}
 
 	NV_WRITE(0x400320, 4);
 	NV_WRITE(NV40_PGRAPH_CTXCTL_CUR, 0);
 	NV_WRITE(NV20_PGRAPH_CHANNEL_CTX_POINTER, 0);
+
+	return 0;
 }
 
 int
 nv50_graph_init(struct drm_device *dev)
 {
+	int ret;
+
 	DRM_DEBUG("\n");
 
 	nv50_graph_init_reset(dev);
 	nv50_graph_init_intr(dev);
 	nv50_graph_init_regs__nv(dev);
 	nv50_graph_init_regs(dev);
-	nv50_graph_init_ctxctl(dev);
+
+	ret = nv50_graph_init_ctxctl(dev);
+	if (ret)
+		return ret;
 
 	return 0;
 }
@@ -209,11 +257,10 @@ nv50_graph_create_context(struct nouveau_channel *chan)
 
 	DRM_DEBUG("ch%d\n", chan->id);
 
-	if ((ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0,
-					  grctx_size, 0x1000,
-					  NVOBJ_FLAG_ZERO_ALLOC |
-					  NVOBJ_FLAG_ZERO_FREE,
-					  &chan->ramin_grctx)))
+	ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, grctx_size, 0x1000,
+				     NVOBJ_FLAG_ZERO_ALLOC |
+				     NVOBJ_FLAG_ZERO_FREE, &chan->ramin_grctx);
+	if (ret)
 		return ret;
 
 	hdr = IS_G80 ? 0x200 : 0x20;
@@ -225,11 +272,16 @@ nv50_graph_create_context(struct nouveau_channel *chan)
 	INSTANCE_WR(ramin, (hdr + 0x10)/4, 0);
 	INSTANCE_WR(ramin, (hdr + 0x14)/4, 0x00010000);
 
-	if ((ret = engine->graph.load_context(chan))) {
+	ret = engine->graph.load_context(chan);
+	if (ret) {
 		DRM_ERROR("Error hacking up initial context: %d\n", ret);
 		return ret;
 	}
 
+	INSTANCE_WR(chan->ramin_grctx->gpuobj, 0x00000/4,
+		    chan->ramin->instance >> 12);
+	INSTANCE_WR(chan->ramin_grctx->gpuobj, 0x0011c/4, 0x00000002);
+
 	return 0;
 }
 
@@ -259,7 +311,7 @@ nv50_graph_transfer_context(struct drm_device *dev, uint32_t inst, int save)
 	DRM_DEBUG("inst=0x%08x, save=%d\n", inst, save);
 
 	old_cp = NV_READ(NV20_PGRAPH_CHANNEL_CTX_POINTER);
-	NV_WRITE(NV20_PGRAPH_CHANNEL_CTX_POINTER, inst | (1<<31));
+	NV_WRITE(NV20_PGRAPH_CHANNEL_CTX_POINTER, inst);
 	NV_WRITE(0x400824, NV_READ(0x400824) |
 		 (save ? NV40_PGRAPH_CTXCTL_0310_XFER_SAVE :
 			 NV40_PGRAPH_CTXCTL_0310_XFER_LOAD));
@@ -286,7 +338,7 @@ nv50_graph_load_context(struct nouveau_channel *chan)
 {
 	struct drm_device *dev = chan->dev;
 	struct drm_nouveau_private *dev_priv = dev->dev_private;
-	uint32_t inst = ((chan->ramin->instance >> 12) | (1<<31));
+	uint32_t inst = chan->ramin->instance >> 12;
 	int ret; (void)ret;
 
 	DRM_DEBUG("ch%d\n", chan->id);
@@ -298,7 +350,7 @@ nv50_graph_load_context(struct nouveau_channel *chan)
 
 	NV_WRITE(NV20_PGRAPH_CHANNEL_CTX_POINTER, inst);
 	NV_WRITE(0x400320, 4);
-	NV_WRITE(NV40_PGRAPH_CTXCTL_CUR, inst);
+	NV_WRITE(NV40_PGRAPH_CTXCTL_CUR, inst | (1<<31));
 
 	return 0;
 }
@@ -307,7 +359,7 @@ int
 nv50_graph_save_context(struct nouveau_channel *chan)
 {
 	struct drm_device *dev = chan->dev;
-	uint32_t inst = ((chan->ramin->instance >> 12) | (1<<31));
+	uint32_t inst = chan->ramin->instance >> 12;
 
 	DRM_DEBUG("ch%d\n", chan->id);
 
diff --git a/shared-core/nv50_instmem.c b/shared-core/nv50_instmem.c
index c80574a9..9687ecbb 100644
--- a/shared-core/nv50_instmem.c
+++ b/shared-core/nv50_instmem.c
@@ -69,6 +69,10 @@ nv50_instmem_init(struct drm_device *dev)
 		return -ENOMEM;
 	dev_priv->Engine.instmem.priv = priv;
 
+	/* Save state, will restore at takedown. */
+	for (i = 0x1700; i <= 0x1710; i+=4)
+		priv->save1700[(i-0x1700)/4] = NV_READ(i);
+
 	/* 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.
diff --git a/shared-core/r128_cce.c b/shared-core/r128_cce.c
index 8ce128bd..204ea37d 100644
--- a/shared-core/r128_cce.c
+++ b/shared-core/r128_cce.c
@@ -649,7 +649,7 @@ int r128_cce_start(struct drm_device *dev, void *data, struct drm_file *file_pri
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (dev_priv->cce_running || dev_priv->cce_mode == R128_PM4_NONPM4) {
-		DRM_DEBUG("%s while CCE running\n", __FUNCTION__);
+		DRM_DEBUG("while CCE running\n");
 		return 0;
 	}
 
@@ -708,7 +708,7 @@ int r128_cce_reset(struct drm_device *dev, void *data, struct drm_file *file_pri
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_DEBUG("%s called before init done\n", __FUNCTION__);
+		DRM_DEBUG("called before init done\n");
 		return -EINVAL;
 	}
 
diff --git a/shared-core/r128_drv.h b/shared-core/r128_drv.h
index b0558389..ab8b6297 100644
--- a/shared-core/r128_drv.h
+++ b/shared-core/r128_drv.h
@@ -97,6 +97,8 @@ typedef struct drm_r128_private {
 	u32 crtc_offset;
 	u32 crtc_offset_cntl;
 
+	atomic_t vbl_received;
+
 	u32 color_fmt;
 	unsigned int front_offset;
 	unsigned int front_pitch;
@@ -149,11 +151,12 @@ extern int r128_wait_ring(drm_r128_private_t * dev_priv, int n);
 extern int r128_do_cce_idle(drm_r128_private_t * dev_priv);
 extern int r128_do_cleanup_cce(struct drm_device * dev);
 
-extern int r128_driver_vblank_wait(struct drm_device * dev, unsigned int *sequence);
-
+extern int r128_enable_vblank(struct drm_device *dev, int crtc);
+extern void r128_disable_vblank(struct drm_device *dev, int crtc);
+extern u32 r128_get_vblank_counter(struct drm_device *dev, int crtc);
 extern irqreturn_t r128_driver_irq_handler(DRM_IRQ_ARGS);
 extern void r128_driver_irq_preinstall(struct drm_device * dev);
-extern void r128_driver_irq_postinstall(struct drm_device * dev);
+extern int r128_driver_irq_postinstall(struct drm_device * dev);
 extern void r128_driver_irq_uninstall(struct drm_device * dev);
 extern void r128_driver_lastclose(struct drm_device * dev);
 extern void r128_driver_preclose(struct drm_device * dev,
@@ -462,8 +465,7 @@ do {									\
 
 #define BEGIN_RING( n ) do {						\
 	if ( R128_VERBOSE ) {						\
-		DRM_INFO( "BEGIN_RING( %d ) in %s\n",			\
-			   (n), __FUNCTION__ );				\
+		DRM_INFO( "BEGIN_RING( %d )\n", (n));			\
 	}								\
 	if ( dev_priv->ring.space <= (n) * sizeof(u32) ) {		\
 		COMMIT_RING();						\
diff --git a/shared-core/r128_irq.c b/shared-core/r128_irq.c
index c76fdca7..5b95bd89 100644
--- a/shared-core/r128_irq.c
+++ b/shared-core/r128_irq.c
@@ -35,6 +35,16 @@
 #include "r128_drm.h"
 #include "r128_drv.h"
 
+u32 r128_get_vblank_counter(struct drm_device *dev, int crtc)
+{
+	const drm_r128_private_t *dev_priv = dev->dev_private;
+
+	if (crtc != 0)
+		return 0;
+
+	return atomic_read(&dev_priv->vbl_received);
+}
+
 irqreturn_t r128_driver_irq_handler(DRM_IRQ_ARGS)
 {
 	struct drm_device *dev = (struct drm_device *) arg;
@@ -46,30 +56,38 @@ irqreturn_t r128_driver_irq_handler(DRM_IRQ_ARGS)
 	/* VBLANK interrupt */
 	if (status & R128_CRTC_VBLANK_INT) {
 		R128_WRITE(R128_GEN_INT_STATUS, R128_CRTC_VBLANK_INT_AK);
-		atomic_inc(&dev->vbl_received);
-		DRM_WAKEUP(&dev->vbl_queue);
-		drm_vbl_send_signals(dev);
+		atomic_inc(&dev_priv->vbl_received);
+		drm_handle_vblank(dev, 0);
 		return IRQ_HANDLED;
 	}
 	return IRQ_NONE;
 }
 
-int r128_driver_vblank_wait(struct drm_device * dev, unsigned int *sequence)
+int r128_enable_vblank(struct drm_device *dev, int crtc)
 {
-	unsigned int cur_vblank;
-	int ret = 0;
+	drm_r128_private_t *dev_priv = dev->dev_private;
 
-	/* Assume that the user has missed the current sequence number
-	 * by about a day rather than she wants to wait for years
-	 * using vertical blanks...
-	 */
-	DRM_WAIT_ON(ret, dev->vbl_queue, 3 * DRM_HZ,
-		    (((cur_vblank = atomic_read(&dev->vbl_received))
-		      - *sequence) <= (1 << 23)));
+	if (crtc != 0) {
+		DRM_ERROR("%s:  bad crtc %d\n", __FUNCTION__, crtc);
+		return -EINVAL;
+	}
 
-	*sequence = cur_vblank;
+	R128_WRITE(R128_GEN_INT_CNTL, R128_CRTC_VBLANK_INT_EN);
+	return 0;
+}
+
+void r128_disable_vblank(struct drm_device *dev, int crtc)
+{
+	if (crtc != 0)
+		DRM_ERROR("%s:  bad crtc %d\n", __FUNCTION__, crtc);
 
-	return ret;
+	/*
+	 * FIXME: implement proper interrupt disable by using the vblank
+	 * counter register (if available)
+	 *
+	 * R128_WRITE(R128_GEN_INT_CNTL,
+	 *            R128_READ(R128_GEN_INT_CNTL) & ~R128_CRTC_VBLANK_INT_EN);
+	 */
 }
 
 void r128_driver_irq_preinstall(struct drm_device * dev)
@@ -82,12 +100,9 @@ void r128_driver_irq_preinstall(struct drm_device * dev)
 	R128_WRITE(R128_GEN_INT_STATUS, R128_CRTC_VBLANK_INT_AK);
 }
 
-void r128_driver_irq_postinstall(struct drm_device * dev)
+int r128_driver_irq_postinstall(struct drm_device * dev)
 {
-	drm_r128_private_t *dev_priv = (drm_r128_private_t *) dev->dev_private;
-
-	/* Turn on VBL interrupt */
-	R128_WRITE(R128_GEN_INT_CNTL, R128_CRTC_VBLANK_INT_EN);
+	return drm_vblank_init(dev, 1);
 }
 
 void r128_driver_irq_uninstall(struct drm_device * dev)
diff --git a/shared-core/r128_state.c b/shared-core/r128_state.c
index b7f483ca..51a9afce 100644
--- a/shared-core/r128_state.c
+++ b/shared-core/r128_state.c
@@ -42,7 +42,7 @@ static void r128_emit_clip_rects(drm_r128_private_t * dev_priv,
 {
 	u32 aux_sc_cntl = 0x00000000;
 	RING_LOCALS;
-	DRM_DEBUG("    %s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	BEGIN_RING((count < 3 ? count : 3) * 5 + 2);
 
@@ -85,7 +85,7 @@ static __inline__ void r128_emit_core(drm_r128_private_t * dev_priv)
 	drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv;
 	drm_r128_context_regs_t *ctx = &sarea_priv->context_state;
 	RING_LOCALS;
-	DRM_DEBUG("    %s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	BEGIN_RING(2);
 
@@ -100,7 +100,7 @@ static __inline__ void r128_emit_context(drm_r128_private_t * dev_priv)
 	drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv;
 	drm_r128_context_regs_t *ctx = &sarea_priv->context_state;
 	RING_LOCALS;
-	DRM_DEBUG("    %s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	BEGIN_RING(13);
 
@@ -126,7 +126,7 @@ static __inline__ void r128_emit_setup(drm_r128_private_t * dev_priv)
 	drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv;
 	drm_r128_context_regs_t *ctx = &sarea_priv->context_state;
 	RING_LOCALS;
-	DRM_DEBUG("    %s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	BEGIN_RING(3);
 
@@ -142,7 +142,7 @@ static __inline__ void r128_emit_masks(drm_r128_private_t * dev_priv)
 	drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv;
 	drm_r128_context_regs_t *ctx = &sarea_priv->context_state;
 	RING_LOCALS;
-	DRM_DEBUG("    %s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	BEGIN_RING(5);
 
@@ -161,7 +161,7 @@ static __inline__ void r128_emit_window(drm_r128_private_t * dev_priv)
 	drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv;
 	drm_r128_context_regs_t *ctx = &sarea_priv->context_state;
 	RING_LOCALS;
-	DRM_DEBUG("    %s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	BEGIN_RING(2);
 
@@ -178,7 +178,7 @@ static __inline__ void r128_emit_tex0(drm_r128_private_t * dev_priv)
 	drm_r128_texture_regs_t *tex = &sarea_priv->tex_state[0];
 	int i;
 	RING_LOCALS;
-	DRM_DEBUG("    %s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	BEGIN_RING(7 + R128_MAX_TEXTURE_LEVELS);
 
@@ -204,7 +204,7 @@ static __inline__ void r128_emit_tex1(drm_r128_private_t * dev_priv)
 	drm_r128_texture_regs_t *tex = &sarea_priv->tex_state[1];
 	int i;
 	RING_LOCALS;
-	DRM_DEBUG("    %s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	BEGIN_RING(5 + R128_MAX_TEXTURE_LEVELS);
 
@@ -226,7 +226,7 @@ static void r128_emit_state(drm_r128_private_t * dev_priv)
 	drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv;
 	unsigned int dirty = sarea_priv->dirty;
 
-	DRM_DEBUG("%s: dirty=0x%08x\n", __FUNCTION__, dirty);
+	DRM_DEBUG("dirty=0x%08x\n", dirty);
 
 	if (dirty & R128_UPLOAD_CORE) {
 		r128_emit_core(dev_priv);
@@ -362,7 +362,7 @@ static void r128_cce_dispatch_clear(struct drm_device * dev,
 	unsigned int flags = clear->flags;
 	int i;
 	RING_LOCALS;
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	if (dev_priv->page_flipping && dev_priv->current_page == 1) {
 		unsigned int tmp = flags;
@@ -466,7 +466,7 @@ static void r128_cce_dispatch_swap(struct drm_device * dev)
 	struct drm_clip_rect *pbox = sarea_priv->boxes;
 	int i;
 	RING_LOCALS;
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 #if R128_PERFORMANCE_BOXES
 	/* Do some trivial performance monitoring...
@@ -528,8 +528,7 @@ static void r128_cce_dispatch_flip(struct drm_device * dev)
 {
 	drm_r128_private_t *dev_priv = dev->dev_private;
 	RING_LOCALS;
-	DRM_DEBUG("%s: page=%d pfCurrentPage=%d\n",
-		  __FUNCTION__,
+	DRM_DEBUG("page=%d pfCurrentPage=%d\n",
 		  dev_priv->current_page, dev_priv->sarea_priv->pfCurrentPage);
 
 #if R128_PERFORMANCE_BOXES
@@ -1156,7 +1155,7 @@ static int r128_cce_dispatch_read_pixels(struct drm_device * dev,
 	int count, *x, *y;
 	int i, xbuf_size, ybuf_size;
 	RING_LOCALS;
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	count = depth->n;
 	if (count > 4096 || count <= 0)
@@ -1226,7 +1225,7 @@ static void r128_cce_dispatch_stipple(struct drm_device * dev, u32 * stipple)
 	drm_r128_private_t *dev_priv = dev->dev_private;
 	int i;
 	RING_LOCALS;
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	BEGIN_RING(33);
 
@@ -1309,7 +1308,7 @@ static int r128_do_cleanup_pageflip(struct drm_device * dev)
 static int r128_cce_flip(struct drm_device *dev, void *data, struct drm_file *file_priv)
 {
 	drm_r128_private_t *dev_priv = dev->dev_private;
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
@@ -1328,7 +1327,7 @@ static int r128_cce_swap(struct drm_device *dev, void *data, struct drm_file *fi
 {
 	drm_r128_private_t *dev_priv = dev->dev_private;
 	drm_r128_sarea_t *sarea_priv = dev_priv->sarea_priv;
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
@@ -1356,7 +1355,7 @@ static int r128_cce_vertex(struct drm_device *dev, void *data, struct drm_file *
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -1412,7 +1411,7 @@ static int r128_cce_indices(struct drm_device *dev, void *data, struct drm_file
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -1557,11 +1556,11 @@ static int r128_cce_indirect(struct drm_device *dev, void *data, struct drm_file
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
-	DRM_DEBUG("indirect: idx=%d s=%d e=%d d=%d\n",
+	DRM_DEBUG("idx=%d s=%d e=%d d=%d\n",
 		  indirect->idx, indirect->start, indirect->end,
 		  indirect->discard);
 
@@ -1622,7 +1621,7 @@ static int r128_getparam(struct drm_device *dev, void *data, struct drm_file *fi
 	int value;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
diff --git a/shared-core/r300_cmdbuf.c b/shared-core/r300_cmdbuf.c
index 6ab907c6..a26a71d5 100644
--- a/shared-core/r300_cmdbuf.c
+++ b/shared-core/r300_cmdbuf.c
@@ -77,23 +77,31 @@ static int r300_emit_cliprects(drm_radeon_private_t *dev_priv,
 				return -EFAULT;
 			}
 
-			box.x1 =
-			    (box.x1 +
-			     R300_CLIPRECT_OFFSET) & R300_CLIPRECT_MASK;
-			box.y1 =
-			    (box.y1 +
-			     R300_CLIPRECT_OFFSET) & R300_CLIPRECT_MASK;
-			box.x2 =
-			    (box.x2 +
-			     R300_CLIPRECT_OFFSET) & R300_CLIPRECT_MASK;
-			box.y2 =
-			    (box.y2 +
-			     R300_CLIPRECT_OFFSET) & R300_CLIPRECT_MASK;
+			if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV515) {
+				box.x1 = (box.x1) &
+					R300_CLIPRECT_MASK;
+				box.y1 = (box.y1) &
+					R300_CLIPRECT_MASK;
+				box.x2 = (box.x2) &
+					R300_CLIPRECT_MASK;
+				box.y2 = (box.y2) &
+					R300_CLIPRECT_MASK;
+			} else {
+				box.x1 = (box.x1 + R300_CLIPRECT_OFFSET) &
+					R300_CLIPRECT_MASK;
+				box.y1 = (box.y1 + R300_CLIPRECT_OFFSET) &
+					R300_CLIPRECT_MASK;
+				box.x2 = (box.x2 + R300_CLIPRECT_OFFSET) &
+					R300_CLIPRECT_MASK;
+				box.y2 = (box.y2 + R300_CLIPRECT_OFFSET) &
+					R300_CLIPRECT_MASK;
 
+			}
 			OUT_RING((box.x1 << R300_CLIPRECT_X_SHIFT) |
 				 (box.y1 << R300_CLIPRECT_Y_SHIFT));
 			OUT_RING((box.x2 << R300_CLIPRECT_X_SHIFT) |
 				 (box.y2 << R300_CLIPRECT_Y_SHIFT));
+
 		}
 
 		OUT_RING_REG(R300_RE_CLIPRECT_CNTL, r300_cliprect_cntl[nr - 1]);
@@ -133,9 +141,11 @@ static int r300_emit_cliprects(drm_radeon_private_t *dev_priv,
 
 static u8 r300_reg_flags[0x10000 >> 2];
 
-void r300_init_reg_flags(void)
+void r300_init_reg_flags(struct drm_device *dev)
 {
 	int i;
+	drm_radeon_private_t *dev_priv = dev->dev_private;
+
 	memset(r300_reg_flags, 0, 0x10000 >> 2);
 #define ADD_RANGE_MARK(reg, count,mark) \
 		for(i=((reg)>>2);i<((reg)>>2)+(count);i++)\
@@ -230,6 +240,9 @@ void r300_init_reg_flags(void)
 	ADD_RANGE(R300_VAP_INPUT_ROUTE_0_0, 8);
 	ADD_RANGE(R300_VAP_INPUT_ROUTE_1_0, 8);
 
+	if ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_RV515) {
+		ADD_RANGE(0x4074, 16);
+	}
 }
 
 static __inline__ int r300_check_range(unsigned reg, int count)
diff --git a/shared-core/radeon_cp.c b/shared-core/radeon_cp.c
index c2863346..f0e55105 100644
--- a/shared-core/radeon_cp.c
+++ b/shared-core/radeon_cp.c
@@ -816,19 +816,44 @@ static const u32 R300_cp_microcode[][2] = {
 	{ 0000000000, 0000000000 },
 };
 
+static u32 RADEON_READ_MCIND(drm_radeon_private_t *dev_priv, int addr)
+{
+	u32 ret;
+	RADEON_WRITE(R520_MC_IND_INDEX, 0x7f0000 | (addr & 0xff));
+	ret = RADEON_READ(R520_MC_IND_DATA);
+	RADEON_WRITE(R520_MC_IND_INDEX, 0);
+	return ret;
+}
+
 u32 radeon_read_fb_location(drm_radeon_private_t *dev_priv)
 {
-	return RADEON_READ(RADEON_MC_FB_LOCATION);
+	
+	if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV515)
+		return RADEON_READ_MCIND(dev_priv, RV515_MC_FB_LOCATION);
+	else if ((dev_priv->flags & RADEON_FAMILY_MASK) > CHIP_RV515)
+		return RADEON_READ_MCIND(dev_priv, R520_MC_FB_LOCATION);
+	else
+		return RADEON_READ(RADEON_MC_FB_LOCATION);
 }
 
 static void radeon_write_fb_location(drm_radeon_private_t *dev_priv, u32 fb_loc)
 {
-	RADEON_WRITE(RADEON_MC_FB_LOCATION, fb_loc);
+	if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV515)
+		RADEON_WRITE_MCIND(RV515_MC_FB_LOCATION, fb_loc);
+	else if ((dev_priv->flags & RADEON_FAMILY_MASK) > CHIP_RV515)
+		RADEON_WRITE_MCIND(R520_MC_FB_LOCATION, fb_loc);
+	else
+		RADEON_WRITE(RADEON_MC_FB_LOCATION, fb_loc);
 }
 
 static void radeon_write_agp_location(drm_radeon_private_t *dev_priv, u32 agp_loc)
 {
-	RADEON_WRITE(RADEON_MC_AGP_LOCATION, agp_loc);
+	if ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV515)
+		RADEON_WRITE_MCIND(RV515_MC_AGP_LOCATION, agp_loc);
+	else if ((dev_priv->flags & RADEON_FAMILY_MASK) > CHIP_RV515)
+		RADEON_WRITE_MCIND(R520_MC_AGP_LOCATION, agp_loc);
+	else
+		RADEON_WRITE(RADEON_MC_AGP_LOCATION, agp_loc);
 }
 
 static int RADEON_READ_PLL(struct drm_device * dev, int addr)
@@ -1089,41 +1114,43 @@ static int radeon_do_engine_reset(struct drm_device * dev)
 
 	radeon_do_pixcache_flush(dev_priv);
 
-	clock_cntl_index = RADEON_READ(RADEON_CLOCK_CNTL_INDEX);
-	mclk_cntl = RADEON_READ_PLL(dev, RADEON_MCLK_CNTL);
-
-	RADEON_WRITE_PLL(RADEON_MCLK_CNTL, (mclk_cntl |
-					    RADEON_FORCEON_MCLKA |
-					    RADEON_FORCEON_MCLKB |
-					    RADEON_FORCEON_YCLKA |
-					    RADEON_FORCEON_YCLKB |
-					    RADEON_FORCEON_MC |
-					    RADEON_FORCEON_AIC));
-
-	rbbm_soft_reset = RADEON_READ(RADEON_RBBM_SOFT_RESET);
-
-	RADEON_WRITE(RADEON_RBBM_SOFT_RESET, (rbbm_soft_reset |
-					      RADEON_SOFT_RESET_CP |
-					      RADEON_SOFT_RESET_HI |
-					      RADEON_SOFT_RESET_SE |
-					      RADEON_SOFT_RESET_RE |
-					      RADEON_SOFT_RESET_PP |
-					      RADEON_SOFT_RESET_E2 |
-					      RADEON_SOFT_RESET_RB));
-	RADEON_READ(RADEON_RBBM_SOFT_RESET);
-	RADEON_WRITE(RADEON_RBBM_SOFT_RESET, (rbbm_soft_reset &
-					      ~(RADEON_SOFT_RESET_CP |
-						RADEON_SOFT_RESET_HI |
-						RADEON_SOFT_RESET_SE |
-						RADEON_SOFT_RESET_RE |
-						RADEON_SOFT_RESET_PP |
-						RADEON_SOFT_RESET_E2 |
-						RADEON_SOFT_RESET_RB)));
-	RADEON_READ(RADEON_RBBM_SOFT_RESET);
-
-	RADEON_WRITE_PLL(RADEON_MCLK_CNTL, mclk_cntl);
-	RADEON_WRITE(RADEON_CLOCK_CNTL_INDEX, clock_cntl_index);
-	RADEON_WRITE(RADEON_RBBM_SOFT_RESET, rbbm_soft_reset);
+	if ((dev_priv->flags & RADEON_FAMILY_MASK) < CHIP_RV515) {
+		clock_cntl_index = RADEON_READ(RADEON_CLOCK_CNTL_INDEX);
+		mclk_cntl = RADEON_READ_PLL(dev, RADEON_MCLK_CNTL);
+		
+		RADEON_WRITE_PLL(RADEON_MCLK_CNTL, (mclk_cntl |
+						    RADEON_FORCEON_MCLKA |
+						    RADEON_FORCEON_MCLKB |
+						    RADEON_FORCEON_YCLKA |
+						    RADEON_FORCEON_YCLKB |
+						    RADEON_FORCEON_MC |
+						    RADEON_FORCEON_AIC));
+		
+		rbbm_soft_reset = RADEON_READ(RADEON_RBBM_SOFT_RESET);
+		
+		RADEON_WRITE(RADEON_RBBM_SOFT_RESET, (rbbm_soft_reset |
+						      RADEON_SOFT_RESET_CP |
+						      RADEON_SOFT_RESET_HI |
+						      RADEON_SOFT_RESET_SE |
+						      RADEON_SOFT_RESET_RE |
+						      RADEON_SOFT_RESET_PP |
+						      RADEON_SOFT_RESET_E2 |
+						      RADEON_SOFT_RESET_RB));
+		RADEON_READ(RADEON_RBBM_SOFT_RESET);
+		RADEON_WRITE(RADEON_RBBM_SOFT_RESET, (rbbm_soft_reset &
+						      ~(RADEON_SOFT_RESET_CP |
+							RADEON_SOFT_RESET_HI |
+							RADEON_SOFT_RESET_SE |
+							RADEON_SOFT_RESET_RE |
+							RADEON_SOFT_RESET_PP |
+							RADEON_SOFT_RESET_E2 |
+							RADEON_SOFT_RESET_RB)));
+		RADEON_READ(RADEON_RBBM_SOFT_RESET);
+		
+		RADEON_WRITE_PLL(RADEON_MCLK_CNTL, mclk_cntl);
+		RADEON_WRITE(RADEON_CLOCK_CNTL_INDEX, clock_cntl_index);
+		RADEON_WRITE(RADEON_RBBM_SOFT_RESET, rbbm_soft_reset);
+	}
 
 	/* Reset the CP ring */
 	radeon_do_cp_reset(dev_priv);
@@ -1881,7 +1908,7 @@ int radeon_cp_init(struct drm_device *dev, void *data, struct drm_file *file_pri
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (init->func == RADEON_INIT_R300_CP)
-		r300_init_reg_flags();
+		r300_init_reg_flags(dev);
 
 	switch (init->func) {
 	case RADEON_INIT_CP:
@@ -1903,12 +1930,12 @@ int radeon_cp_start(struct drm_device *dev, void *data, struct drm_file *file_pr
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (dev_priv->cp_running) {
-		DRM_DEBUG("%s while CP running\n", __FUNCTION__);
+		DRM_DEBUG("while CP running\n");
 		return 0;
 	}
 	if (dev_priv->cp_mode == RADEON_CSQ_PRIDIS_INDDIS) {
-		DRM_DEBUG("%s called with bogus CP mode (%d)\n",
-			  __FUNCTION__, dev_priv->cp_mode);
+		DRM_DEBUG("called with bogus CP mode (%d)\n",
+			  dev_priv->cp_mode);
 		return 0;
 	}
 
@@ -2018,7 +2045,7 @@ int radeon_cp_reset(struct drm_device *dev, void *data, struct drm_file *file_pr
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_DEBUG("%s called before init done\n", __FUNCTION__);
+		DRM_DEBUG("called before init done\n");
 		return -EINVAL;
 	}
 
@@ -2295,6 +2322,10 @@ int radeon_driver_load(struct drm_device *dev, unsigned long flags)
 	case CHIP_R350:
 	case CHIP_R420:
 	case CHIP_RV410:
+	case CHIP_RV515:
+	case CHIP_R520:
+	case CHIP_RV570:
+	case CHIP_R580:
 		dev_priv->flags |= RADEON_HAS_HIERZ;
 		break;
 	default:
diff --git a/shared-core/radeon_drv.h b/shared-core/radeon_drv.h
index 37b9c43f..63877008 100644
--- a/shared-core/radeon_drv.h
+++ b/shared-core/radeon_drv.h
@@ -129,6 +129,12 @@ enum radeon_family {
 	CHIP_R420,
 	CHIP_RV410,
 	CHIP_RS400,
+	CHIP_RV515,
+	CHIP_R520,
+	CHIP_RV530,
+	CHIP_RV560,
+	CHIP_RV570,
+	CHIP_R580,
 	CHIP_LAST,
 };
 
@@ -305,6 +311,9 @@ typedef struct drm_radeon_private {
 
 	u32 scratch_ages[5];
 
+	unsigned int crtc_last_cnt;
+	unsigned int crtc2_last_cnt;
+
 	/* starting from here on, data is preserved accross an open */
 	uint32_t flags;		/* see radeon_chip_flags */
 	unsigned long fb_aper_offset;
@@ -374,13 +383,13 @@ extern int radeon_irq_wait(struct drm_device *dev, void *data, struct drm_file *
 extern int radeon_emit_irq(struct drm_device * dev);
 
 extern void radeon_do_release(struct drm_device * dev);
-extern int radeon_driver_vblank_wait(struct drm_device * dev,
-				     unsigned int *sequence);
-extern int radeon_driver_vblank_wait2(struct drm_device * dev,
-				      unsigned int *sequence);
+extern u32 radeon_get_vblank_counter(struct drm_device *dev, int crtc);
+extern int radeon_enable_vblank(struct drm_device *dev, int crtc);
+extern void radeon_disable_vblank(struct drm_device *dev, int crtc);
+extern void radeon_do_release(struct drm_device * dev);
 extern irqreturn_t radeon_driver_irq_handler(DRM_IRQ_ARGS);
 extern void radeon_driver_irq_preinstall(struct drm_device * dev);
-extern void radeon_driver_irq_postinstall(struct drm_device * dev);
+extern int radeon_driver_irq_postinstall(struct drm_device * dev);
 extern void radeon_driver_irq_uninstall(struct drm_device * dev);
 extern int radeon_vblank_crtc_get(struct drm_device *dev);
 extern int radeon_vblank_crtc_set(struct drm_device *dev, int64_t value);
@@ -399,7 +408,7 @@ extern long radeon_compat_ioctl(struct file *filp, unsigned int cmd,
 					 unsigned long arg);
 
 /* r300_cmdbuf.c */
-extern void r300_init_reg_flags(void);
+extern void r300_init_reg_flags(struct drm_device *dev);
 
 extern int r300_do_cp_cmdbuf(struct drm_device *dev,
 			     struct drm_file *file_priv,
@@ -423,7 +432,7 @@ extern int radeon_fence_has_irq(struct drm_device *dev, uint32_t class, uint32_t
 extern struct drm_ttm_backend *radeon_create_ttm_backend_entry(struct drm_device *dev);
 extern int radeon_fence_types(struct drm_buffer_object *bo, uint32_t *class, uint32_t *type);
 extern int radeon_invalidate_caches(struct drm_device *dev, uint64_t buffer_flags);
-extern uint32_t radeon_evict_mask(struct drm_buffer_object *bo);
+extern uint64_t radeon_evict_flags(struct drm_buffer_object *bo);
 extern int radeon_init_mem_type(struct drm_device * dev, uint32_t type,
 				struct drm_mem_type_manager * man);
 extern int radeon_move(struct drm_buffer_object * bo,
@@ -494,6 +503,15 @@ extern int radeon_move(struct drm_buffer_object * bo,
 #define RADEON_IGPGART_ENABLE           0x38
 #define RADEON_IGPGART_UNK_39           0x39
 
+#define R520_MC_IND_INDEX 0x70
+#define R520_MC_IND_WR_EN (1<<24)
+#define R520_MC_IND_DATA  0x74
+
+#define RV515_MC_FB_LOCATION 0x01
+#define RV515_MC_AGP_LOCATION 0x02
+
+#define R520_MC_FB_LOCATION 0x04
+#define R520_MC_AGP_LOCATION 0x05
 
 #define RADEON_MPP_TB_CONFIG		0x01c0
 #define RADEON_MEM_CNTL			0x0140
@@ -545,6 +563,12 @@ extern int radeon_move(struct drm_buffer_object * bo,
 				? DRM_READ32( dev_priv->ring_rptr, RADEON_SCRATCHOFF(x) ) \
 				: RADEON_READ( RADEON_SCRATCH_REG0 + 4*(x) ) )
 
+#define RADEON_CRTC_CRNT_FRAME 0x0214
+#define RADEON_CRTC2_CRNT_FRAME 0x0314
+
+#define RADEON_CRTC_STATUS		0x005c
+#define RADEON_CRTC2_STATUS		0x03fc
+
 #define RADEON_GEN_INT_CNTL		0x0040
 #	define RADEON_CRTC_VBLANK_MASK		(1 << 0)
 #	define RADEON_CRTC2_VBLANK_MASK		(1 << 9)
@@ -1117,6 +1141,13 @@ do {									\
 	RADEON_WRITE( RADEON_PCIE_DATA, (val) );			\
 } while (0)
 
+#define RADEON_WRITE_MCIND( addr, val )					\
+	do {								\
+		RADEON_WRITE(R520_MC_IND_INDEX, 0xff0000 | ((addr) & 0xff));	\
+		RADEON_WRITE(R520_MC_IND_DATA, (val));			\
+		RADEON_WRITE(R520_MC_IND_INDEX, 0);	\
+	} while (0)
+
 #define CP_PACKET0( reg, n )						\
 	(RADEON_CP_PACKET0 | ((n) << 16) | ((reg) >> 2))
 #define CP_PACKET0_TABLE( reg, n )					\
@@ -1227,8 +1258,7 @@ do {									\
 
 #define BEGIN_RING( n ) do {						\
 	if ( RADEON_VERBOSE ) {						\
-		DRM_INFO( "BEGIN_RING( %d ) in %s\n",			\
-			   n, __FUNCTION__ );				\
+		DRM_INFO( "BEGIN_RING( %d )\n", (n));			\
 	}								\
 	if ( dev_priv->ring.space <= (n) * sizeof(u32) ) {		\
 		COMMIT_RING();						\
diff --git a/shared-core/radeon_irq.c b/shared-core/radeon_irq.c
index 544e3a0f..3f6ace88 100644
--- a/shared-core/radeon_irq.c
+++ b/shared-core/radeon_irq.c
@@ -35,12 +35,61 @@
 #include "radeon_drm.h"
 #include "radeon_drv.h"
 
-static __inline__ u32 radeon_acknowledge_irqs(drm_radeon_private_t * dev_priv,
-					      u32 mask)
+static void radeon_irq_set_state(struct drm_device *dev, u32 mask, int state)
 {
-	u32 irqs = RADEON_READ(RADEON_GEN_INT_STATUS) & mask;
+	drm_radeon_private_t *dev_priv = dev->dev_private;
+
+	if (state)
+		dev_priv->irq_enable_reg |= mask;
+	else
+		dev_priv->irq_enable_reg &= ~mask;
+
+	RADEON_WRITE(RADEON_GEN_INT_CNTL, dev_priv->irq_enable_reg);
+}
+
+int radeon_enable_vblank(struct drm_device *dev, int crtc)
+{
+	switch (crtc) {
+	case 0:
+		radeon_irq_set_state(dev, RADEON_CRTC_VBLANK_MASK, 1);
+		break;
+	case 1:
+		radeon_irq_set_state(dev, RADEON_CRTC2_VBLANK_MASK, 1);
+		break;
+	default:
+		DRM_ERROR("tried to enable vblank on non-existent crtc %d\n",
+			  crtc);
+		return EINVAL;
+	}
+
+	return 0;
+}
+
+void radeon_disable_vblank(struct drm_device *dev, int crtc)
+{
+	switch (crtc) {
+	case 0:
+		radeon_irq_set_state(dev, RADEON_CRTC_VBLANK_MASK, 0);
+		break;
+	case 1:
+		radeon_irq_set_state(dev, RADEON_CRTC2_VBLANK_MASK, 0);
+		break;
+	default:
+		DRM_ERROR("tried to enable vblank on non-existent crtc %d\n",
+			  crtc);
+		break;
+	}
+}
+
+static __inline__ u32 radeon_acknowledge_irqs(drm_radeon_private_t * dev_priv)
+{
+	u32 irqs = RADEON_READ(RADEON_GEN_INT_STATUS) &
+		(RADEON_SW_INT_TEST | RADEON_CRTC_VBLANK_STAT |
+		 RADEON_CRTC2_VBLANK_STAT);
+
 	if (irqs)
 		RADEON_WRITE(RADEON_GEN_INT_STATUS, irqs);
+
 	return irqs;
 }
 
@@ -72,9 +121,7 @@ irqreturn_t radeon_driver_irq_handler(DRM_IRQ_ARGS)
 	/* Only consider the bits we're interested in - others could be used
 	 * outside the DRM
 	 */
-	stat = radeon_acknowledge_irqs(dev_priv, (RADEON_SW_INT_TEST_ACK |
-						  RADEON_CRTC_VBLANK_STAT |
-						  RADEON_CRTC2_VBLANK_STAT));
+	stat = radeon_acknowledge_irqs(dev_priv);
 	if (!stat)
 		return IRQ_NONE;
 
@@ -87,27 +134,11 @@ irqreturn_t radeon_driver_irq_handler(DRM_IRQ_ARGS)
 		radeon_fence_handler(dev);
 #endif
 	}
-
 	/* VBLANK interrupt */
-	if (stat & (RADEON_CRTC_VBLANK_STAT|RADEON_CRTC2_VBLANK_STAT)) {
-		int vblank_crtc = dev_priv->vblank_crtc;
-
-		if ((vblank_crtc &
-		     (DRM_RADEON_VBLANK_CRTC1 | DRM_RADEON_VBLANK_CRTC2)) ==
-		    (DRM_RADEON_VBLANK_CRTC1 | DRM_RADEON_VBLANK_CRTC2)) {
-			if (stat & RADEON_CRTC_VBLANK_STAT)
-				atomic_inc(&dev->vbl_received);
-			if (stat & RADEON_CRTC2_VBLANK_STAT)
-				atomic_inc(&dev->vbl_received2);
-		} else if (((stat & RADEON_CRTC_VBLANK_STAT) &&
-			   (vblank_crtc & DRM_RADEON_VBLANK_CRTC1)) ||
-			   ((stat & RADEON_CRTC2_VBLANK_STAT) &&
-			    (vblank_crtc & DRM_RADEON_VBLANK_CRTC2)))
-			atomic_inc(&dev->vbl_received);
-
-		DRM_WAKEUP(&dev->vbl_queue);
-		drm_vbl_send_signals(dev);
-	}
+	if (stat & RADEON_CRTC_VBLANK_STAT)
+		drm_handle_vblank(dev, 0);
+	if (stat & RADEON_CRTC2_VBLANK_STAT)
+		drm_handle_vblank(dev, 1);
 
 	return IRQ_HANDLED;
 }
@@ -146,54 +177,27 @@ static int radeon_wait_irq(struct drm_device * dev, int swi_nr)
 	return ret;
 }
 
-static int radeon_driver_vblank_do_wait(struct drm_device * dev,
-					unsigned int *sequence, int crtc)
+u32 radeon_get_vblank_counter(struct drm_device *dev, int crtc)
 {
-	drm_radeon_private_t *dev_priv =
-	    (drm_radeon_private_t *) dev->dev_private;
-	unsigned int cur_vblank;
-	int ret = 0;
-	int ack = 0;
-	atomic_t *counter;
+	drm_radeon_private_t *dev_priv = dev->dev_private;
+	u32 crtc_cnt_reg, crtc_status_reg;
+
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
-	if (crtc == DRM_RADEON_VBLANK_CRTC1) {
-		counter = &dev->vbl_received;
-		ack |= RADEON_CRTC_VBLANK_STAT;
-	} else if (crtc == DRM_RADEON_VBLANK_CRTC2) {
-		counter = &dev->vbl_received2;
-		ack |= RADEON_CRTC2_VBLANK_STAT;
-	} else
+	if (crtc == 0) {
+		crtc_cnt_reg = RADEON_CRTC_CRNT_FRAME;
+		crtc_status_reg = RADEON_CRTC_STATUS;
+	} else if (crtc == 1) {
+		crtc_cnt_reg = RADEON_CRTC2_CRNT_FRAME;
+		crtc_status_reg = RADEON_CRTC2_STATUS;
+	} else {
 		return -EINVAL;
+	}
 
-	radeon_acknowledge_irqs(dev_priv, ack);
-
-	dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE;
-
-	/* Assume that the user has missed the current sequence number
-	 * by about a day rather than she wants to wait for years
-	 * using vertical blanks...
-	 */
-	DRM_WAIT_ON(ret, dev->vbl_queue, 3 * DRM_HZ,
-		    (((cur_vblank = atomic_read(counter))
-		      - *sequence) <= (1 << 23)));
-
-	*sequence = cur_vblank;
-
-	return ret;
-}
-
-int radeon_driver_vblank_wait(struct drm_device *dev, unsigned int *sequence)
-{
-	return radeon_driver_vblank_do_wait(dev, sequence, DRM_RADEON_VBLANK_CRTC1);
-}
-
-int radeon_driver_vblank_wait2(struct drm_device *dev, unsigned int *sequence)
-{
-	return radeon_driver_vblank_do_wait(dev, sequence, DRM_RADEON_VBLANK_CRTC2);
+	return RADEON_READ(crtc_cnt_reg) + (RADEON_READ(crtc_status_reg) & 1);
 }
 
 /* Needs the lock as it touches the ring.
@@ -207,7 +211,7 @@ int radeon_irq_emit(struct drm_device *dev, void *data, struct drm_file *file_pr
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -229,28 +233,13 @@ int radeon_irq_wait(struct drm_device *dev, void *data, struct drm_file *file_pr
 	drm_radeon_irq_wait_t *irqwait = data;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
 	return radeon_wait_irq(dev, irqwait->irq_seq);
 }
 
-static void radeon_enable_interrupt(struct drm_device *dev)
-{
-	drm_radeon_private_t *dev_priv = (drm_radeon_private_t *) dev->dev_private;
-
-	dev_priv->irq_enable_reg = RADEON_SW_INT_ENABLE;
-	if (dev_priv->vblank_crtc & DRM_RADEON_VBLANK_CRTC1)
-		dev_priv->irq_enable_reg |= RADEON_CRTC_VBLANK_MASK;
-
-	if (dev_priv->vblank_crtc & DRM_RADEON_VBLANK_CRTC2)
-		dev_priv->irq_enable_reg |= RADEON_CRTC2_VBLANK_MASK;
-
-	RADEON_WRITE(RADEON_GEN_INT_CNTL, dev_priv->irq_enable_reg);
-	dev_priv->irq_enabled = 1;
-}
-
 /* drm_dma.h hooks
 */
 void radeon_driver_irq_preinstall(struct drm_device * dev)
@@ -262,20 +251,27 @@ void radeon_driver_irq_preinstall(struct drm_device * dev)
 	RADEON_WRITE(RADEON_GEN_INT_CNTL, 0);
 
 	/* Clear bits if they're already high */
-	radeon_acknowledge_irqs(dev_priv, (RADEON_SW_INT_TEST_ACK |
-					   RADEON_CRTC_VBLANK_STAT |
-					   RADEON_CRTC2_VBLANK_STAT));
+	radeon_acknowledge_irqs(dev_priv);
 }
 
-void radeon_driver_irq_postinstall(struct drm_device * dev)
+int radeon_driver_irq_postinstall(struct drm_device * dev)
 {
 	drm_radeon_private_t *dev_priv =
 	    (drm_radeon_private_t *) dev->dev_private;
+	int ret;
 
 	dev_priv->counter = 0;
 	DRM_INIT_WAITQUEUE(&dev_priv->irq_queue);
 
-	radeon_enable_interrupt(dev);
+	ret = drm_vblank_init(dev, 2);
+	if (ret)
+		return ret;
+
+	dev->max_vblank_count = 0x001fffff;
+
+	radeon_irq_set_state(dev, RADEON_SW_INT_ENABLE, 1);
+
+	return 0;
 }
 
 void radeon_driver_irq_uninstall(struct drm_device * dev)
@@ -317,6 +313,5 @@ int radeon_vblank_crtc_set(struct drm_device *dev, int64_t value)
 		return -EINVAL;
 	}
 	dev_priv->vblank_crtc = (unsigned int)value;
-	radeon_enable_interrupt(dev);
 	return 0;
 }
diff --git a/shared-core/radeon_mem.c b/shared-core/radeon_mem.c
index 9947e940..1e582ee0 100644
--- a/shared-core/radeon_mem.c
+++ b/shared-core/radeon_mem.c
@@ -224,7 +224,7 @@ int radeon_mem_alloc(struct drm_device *dev, void *data, struct drm_file *file_p
 	struct mem_block *block, **heap;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -259,7 +259,7 @@ int radeon_mem_free(struct drm_device *dev, void *data, struct drm_file *file_pr
 	struct mem_block *block, **heap;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -285,7 +285,7 @@ int radeon_mem_init_heap(struct drm_device *dev, void *data, struct drm_file *fi
 	struct mem_block **heap;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
diff --git a/shared-core/radeon_ms.h b/shared-core/radeon_ms.h
index a784882b..66533837 100644
--- a/shared-core/radeon_ms.h
+++ b/shared-core/radeon_ms.h
@@ -338,7 +338,7 @@ int radeon_ms_bo_get_gpu_addr(struct drm_device *dev,
 int radeon_ms_bo_move(struct drm_buffer_object * bo, int evict,
 		      int no_wait, struct drm_bo_mem_reg * new_mem);
 struct drm_ttm_backend *radeon_ms_create_ttm_backend(struct drm_device * dev);
-uint32_t radeon_ms_evict_mask(struct drm_buffer_object *bo);
+uint64_t radeon_ms_evict_flags(struct drm_buffer_object *bo);
 int radeon_ms_init_mem_type(struct drm_device * dev, uint32_t type,
 			    struct drm_mem_type_manager * man);
 int radeon_ms_invalidate_caches(struct drm_device * dev, uint64_t flags);
diff --git a/shared-core/radeon_ms_bo.c b/shared-core/radeon_ms_bo.c
index cedad689..6d9a97c1 100644
--- a/shared-core/radeon_ms_bo.c
+++ b/shared-core/radeon_ms_bo.c
@@ -141,7 +141,7 @@ static int radeon_ms_bo_move_flip(struct drm_buffer_object *bo,
 
 	tmp_mem = *new_mem;
 	tmp_mem.mm_node = NULL;
-	tmp_mem.mask = DRM_BO_FLAG_MEM_TT |
+	tmp_mem.flags = DRM_BO_FLAG_MEM_TT |
 		       DRM_BO_FLAG_CACHED |
 		       DRM_BO_FLAG_FORCE_CACHING;
 	ret = drm_bo_mem_space(bo, &tmp_mem, no_wait);
@@ -149,7 +149,7 @@ static int radeon_ms_bo_move_flip(struct drm_buffer_object *bo,
 		return ret;
 	}
 
-	ret = drm_bind_ttm(bo->ttm, &tmp_mem);
+	ret = drm_ttm_bind(bo->ttm, &tmp_mem);
 	if (ret) {
 		goto out_cleanup;
 	}
@@ -217,7 +217,7 @@ struct drm_ttm_backend *radeon_ms_create_ttm_backend(struct drm_device * dev)
 	return NULL;
 }
 
-uint32_t radeon_ms_evict_mask(struct drm_buffer_object *bo)
+uint64_t radeon_ms_evict_flags(struct drm_buffer_object *bo)
 {
 	switch (bo->mem.mem_type) {
 	case DRM_BO_MEM_LOCAL:
diff --git a/shared-core/radeon_ms_drm.c b/shared-core/radeon_ms_drm.c
index 91ca4a32..bf76b45d 100644
--- a/shared-core/radeon_ms_drm.c
+++ b/shared-core/radeon_ms_drm.c
@@ -63,7 +63,7 @@ struct drm_bo_driver radeon_ms_bo_driver = {
 	.fence_type = radeon_ms_fence_types,
 	.invalidate_caches = radeon_ms_invalidate_caches,
 	.init_mem_type = radeon_ms_init_mem_type,
-	.evict_mask = radeon_ms_evict_mask,
+	.evict_flags = radeon_ms_evict_flags,
 	.move = radeon_ms_bo_move,
 	.ttm_cache_flush = radeon_ms_ttm_flush,
 };
diff --git a/shared-core/radeon_state.c b/shared-core/radeon_state.c
index 623576ad..6f2e05b3 100644
--- a/shared-core/radeon_state.c
+++ b/shared-core/radeon_state.c
@@ -898,7 +898,7 @@ static void radeon_cp_dispatch_clear(struct drm_device * dev,
 			int w = pbox[i].x2 - x;
 			int h = pbox[i].y2 - y;
 
-			DRM_DEBUG("dispatch clear %d,%d-%d,%d flags 0x%x\n",
+			DRM_DEBUG("%d,%d-%d,%d flags 0x%x\n",
 				  x, y, w, h, flags);
 
 			if (flags & RADEON_FRONT) {
@@ -1368,7 +1368,7 @@ static void radeon_cp_dispatch_swap(struct drm_device * dev)
 		int w = pbox[i].x2 - x;
 		int h = pbox[i].y2 - y;
 
-		DRM_DEBUG("dispatch swap %d,%d-%d,%d\n", x, y, w, h);
+		DRM_DEBUG("%d,%d-%d,%d\n", x, y, w, h);
 
 		BEGIN_RING(9);
 
@@ -1422,8 +1422,7 @@ static void radeon_cp_dispatch_flip(struct drm_device * dev)
 	int offset = (dev_priv->sarea_priv->pfCurrentPage == 1)
 	    ? dev_priv->front_offset : dev_priv->back_offset;
 	RING_LOCALS;
-	DRM_DEBUG("%s: pfCurrentPage=%d\n",
-		  __FUNCTION__,
+	DRM_DEBUG("pfCurrentPage=%d\n",
 		  dev_priv->sarea_priv->pfCurrentPage);
 
 	/* Do some trivial performance monitoring...
@@ -1562,7 +1561,7 @@ static void radeon_cp_dispatch_indirect(struct drm_device * dev,
 {
 	drm_radeon_private_t *dev_priv = dev->dev_private;
 	RING_LOCALS;
-	DRM_DEBUG("indirect: buf=%d s=0x%x e=0x%x\n", buf->idx, start, end);
+	DRM_DEBUG("buf=%d s=0x%x e=0x%x\n", buf->idx, start, end);
 
 	if (start != end) {
 		int offset = (dev_priv->gart_buffers_offset
@@ -1758,7 +1757,7 @@ static int radeon_cp_dispatch_texture(struct drm_device * dev,
 			buf = radeon_freelist_get(dev);
 		}
 		if (!buf) {
-			DRM_DEBUG("radeon_cp_dispatch_texture: EAGAIN\n");
+			DRM_DEBUG("EAGAIN\n");
 			if (DRM_COPY_TO_USER(tex->image, image, sizeof(*image)))
 				return -EFAULT;
 			return -EAGAIN;
@@ -2084,7 +2083,7 @@ static int radeon_surface_alloc(struct drm_device *dev, void *data, struct drm_f
 	drm_radeon_surface_alloc_t *alloc = data;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -2100,7 +2099,7 @@ static int radeon_surface_free(struct drm_device *dev, void *data, struct drm_fi
 	drm_radeon_surface_free_t *memfree = data;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -2215,7 +2214,7 @@ static int radeon_cp_vertex(struct drm_device *dev, void *data, struct drm_file
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -2299,7 +2298,7 @@ static int radeon_cp_indices(struct drm_device *dev, void *data, struct drm_file
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 	sarea_priv = dev_priv->sarea_priv;
@@ -2437,11 +2436,11 @@ static int radeon_cp_indirect(struct drm_device *dev, void *data, struct drm_fil
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
-	DRM_DEBUG("indirect: idx=%d s=%d e=%d d=%d\n",
+	DRM_DEBUG("idx=%d s=%d e=%d d=%d\n",
 		  indirect->idx, indirect->start, indirect->end,
 		  indirect->discard);
 
@@ -2509,7 +2508,7 @@ static int radeon_cp_vertex2(struct drm_device *dev, void *data, struct drm_file
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -2814,7 +2813,7 @@ static int radeon_emit_wait(struct drm_device * dev, int flags)
 	drm_radeon_private_t *dev_priv = dev->dev_private;
 	RING_LOCALS;
 
-	DRM_DEBUG("%s: %x\n", __FUNCTION__, flags);
+	DRM_DEBUG("%x\n", flags);
 	switch (flags) {
 	case RADEON_WAIT_2D:
 		BEGIN_RING(2);
@@ -2852,7 +2851,7 @@ static int radeon_cp_cmdbuf(struct drm_device *dev, void *data, struct drm_file
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -3013,7 +3012,7 @@ static int radeon_cp_getparam(struct drm_device *dev, void *data, struct drm_fil
 	int value;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
@@ -3104,7 +3103,7 @@ static int radeon_cp_setparam(struct drm_device *dev, void *data, struct drm_fil
 	struct drm_radeon_driver_file_fields *radeon_priv;
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
diff --git a/shared-core/via_dma.c b/shared-core/via_dma.c
index ab39f53c..431738a9 100644
--- a/shared-core/via_dma.c
+++ b/shared-core/via_dma.c
@@ -167,14 +167,12 @@ static int via_initialize(struct drm_device * dev,
 	}
 
 	if (dev_priv->ring.virtual_start != NULL) {
-		DRM_ERROR("%s called again without calling cleanup\n",
-			  __FUNCTION__);
+		DRM_ERROR("called again without calling cleanup\n");
 		return -EFAULT;
 	}
 
 	if (!dev->agp || !dev->agp->base) {
-		DRM_ERROR("%s called with no agp memory available\n",
-			  __FUNCTION__);
+		DRM_ERROR("called with no agp memory available\n");
 		return -EFAULT;
 	}
 
@@ -257,8 +255,7 @@ static int via_dispatch_cmdbuffer(struct drm_device * dev, drm_via_cmdbuffer_t *
 	dev_priv = (drm_via_private_t *) dev->dev_private;
 
 	if (dev_priv->ring.virtual_start == NULL) {
-		DRM_ERROR("%s called without initializing AGP ring buffer.\n",
-			  __FUNCTION__);
+		DRM_ERROR("called without initializing AGP ring buffer.\n");
 		return -EFAULT;
 	}
 
@@ -327,8 +324,7 @@ static int via_cmdbuffer(struct drm_device *dev, void *data, struct drm_file *fi
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
-	DRM_DEBUG("via cmdbuffer, buf %p size %lu\n", cmdbuf->buf,
-		  cmdbuf->size);
+	DRM_DEBUG("buf %p size %lu\n", cmdbuf->buf, cmdbuf->size);
 
 	ret = via_dispatch_cmdbuffer(dev, cmdbuf);
 	if (ret) {
@@ -369,8 +365,7 @@ static int via_pci_cmdbuffer(struct drm_device *dev, void *data, struct drm_file
 
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
-	DRM_DEBUG("via_pci_cmdbuffer, buf %p size %lu\n", cmdbuf->buf,
-		  cmdbuf->size);
+	DRM_DEBUG("buf %p size %lu\n", cmdbuf->buf, cmdbuf->size);
 
 	ret = via_dispatch_pci_cmdbuffer(dev, cmdbuf);
 	if (ret) {
@@ -640,14 +635,13 @@ static int via_cmdbuf_size(struct drm_device *dev, void *data, struct drm_file *
 	uint32_t tmp_size, count;
 	drm_via_private_t *dev_priv;
 
-	DRM_DEBUG("via cmdbuf_size\n");
+	DRM_DEBUG("\n");
 	LOCK_TEST_WITH_RETURN(dev, file_priv);
 
 	dev_priv = (drm_via_private_t *) dev->dev_private;
 
 	if (dev_priv->ring.virtual_start == NULL) {
-		DRM_ERROR("%s called without initializing AGP ring buffer.\n",
-			  __FUNCTION__);
+		DRM_ERROR("called without initializing AGP ring buffer.\n");
 		return -EFAULT;
 	}
 
diff --git a/shared-core/via_drv.c b/shared-core/via_drv.c
index 9f099555..a802e4ae 100644
--- a/shared-core/via_drv.c
+++ b/shared-core/via_drv.c
@@ -74,7 +74,7 @@ static struct drm_bo_driver via_bo_driver = {
 	.fence_type = via_fence_types,
 	.invalidate_caches = via_invalidate_caches,
 	.init_mem_type = via_init_mem_type,
-	.evict_mask = via_evict_mask,
+	.evict_flags = via_evict_flags,
 	.move = NULL,
 };
 #endif
@@ -83,14 +83,16 @@ static int probe(struct pci_dev *pdev, const struct pci_device_id *ent);
 static struct drm_driver driver = {
 	.driver_features =
 	    DRIVER_USE_AGP | DRIVER_USE_MTRR | DRIVER_HAVE_IRQ |
-	    DRIVER_IRQ_SHARED | DRIVER_IRQ_VBL,
+	    DRIVER_IRQ_SHARED,
 	.load = via_driver_load,
 	.unload = via_driver_unload,
 #ifndef VIA_HAVE_CORE_MM
 	.context_ctor = via_init_context,
 #endif
 	.context_dtor = via_final_context,
-	.vblank_wait = via_driver_vblank_wait,
+	.get_vblank_counter = via_get_vblank_counter,
+	.enable_vblank = via_enable_vblank,
+	.disable_vblank = via_disable_vblank,
 	.irq_preinstall = via_driver_irq_preinstall,
 	.irq_postinstall = via_driver_irq_postinstall,
 	.irq_uninstall = via_driver_irq_uninstall,
diff --git a/shared-core/via_drv.h b/shared-core/via_drv.h
index d6da8bd9..8dd4a727 100644
--- a/shared-core/via_drv.h
+++ b/shared-core/via_drv.h
@@ -102,6 +102,7 @@ typedef struct drm_via_private {
 	struct timeval last_vblank;
 	int last_vblank_valid;
 	unsigned usec_per_vblank;
+	atomic_t vbl_received;
 	drm_via_state_t hc_state;
 	char pci_buf[VIA_PCI_BUF_SIZE];
 	const uint32_t *fire_offsets[VIA_FIRE_BUF_SIZE];
@@ -166,11 +167,13 @@ extern int via_driver_unload(struct drm_device *dev);
 extern int via_final_context(struct drm_device * dev, int context);
 
 extern int via_do_cleanup_map(struct drm_device * dev);
-extern int via_driver_vblank_wait(struct drm_device * dev, unsigned int *sequence);
+extern u32 via_get_vblank_counter(struct drm_device *dev, int crtc);
+extern int via_enable_vblank(struct drm_device *dev, int crtc);
+extern void via_disable_vblank(struct drm_device *dev, int crtc);
 
 extern irqreturn_t via_driver_irq_handler(DRM_IRQ_ARGS);
 extern void via_driver_irq_preinstall(struct drm_device * dev);
-extern void via_driver_irq_postinstall(struct drm_device * dev);
+extern int via_driver_irq_postinstall(struct drm_device * dev);
 extern void via_driver_irq_uninstall(struct drm_device * dev);
 
 extern int via_dma_cleanup(struct drm_device * dev);
@@ -211,7 +214,7 @@ extern int via_fence_types(struct drm_buffer_object *bo, uint32_t *fclass,
 extern int via_invalidate_caches(struct drm_device *dev, uint64_t buffer_flags);
 extern int via_init_mem_type(struct drm_device *dev, uint32_t type,
 			       struct drm_mem_type_manager *man);
-extern uint32_t via_evict_mask(struct drm_buffer_object *bo);
+extern uint64_t via_evict_flags(struct drm_buffer_object *bo);
 extern int via_move(struct drm_buffer_object *bo, int evict,
 		int no_wait, struct drm_bo_mem_reg *new_mem);
 #endif
diff --git a/shared-core/via_irq.c b/shared-core/via_irq.c
index ec3f50bd..b8e652e6 100644
--- a/shared-core/via_irq.c
+++ b/shared-core/via_irq.c
@@ -96,6 +96,15 @@ static unsigned time_diff(struct timeval *now,struct timeval *then)
 		1000000 - (then->tv_usec - now->tv_usec);
 }
 
+u32 via_get_vblank_counter(struct drm_device *dev, int crtc)
+{
+	drm_via_private_t *dev_priv = dev->dev_private;
+	if (crtc != 0)
+		return 0;
+
+	return atomic_read(&dev_priv->vbl_received);
+}
+
 irqreturn_t via_driver_irq_handler(DRM_IRQ_ARGS)
 {
 	struct drm_device *dev = (struct drm_device *) arg;
@@ -108,8 +117,8 @@ irqreturn_t via_driver_irq_handler(DRM_IRQ_ARGS)
 
 	status = VIA_READ(VIA_REG_INTERRUPT);
 	if (status & VIA_IRQ_VBLANK_PENDING) {
-		atomic_inc(&dev->vbl_received);
-		if (!(atomic_read(&dev->vbl_received) & 0x0F)) {
+		atomic_inc(&dev_priv->vbl_received);
+		if (!(atomic_read(&dev_priv->vbl_received) & 0x0F)) {
 #ifdef __linux__
 			do_gettimeofday(&cur_vblank);
 #else
@@ -123,12 +132,11 @@ irqreturn_t via_driver_irq_handler(DRM_IRQ_ARGS)
 			dev_priv->last_vblank = cur_vblank;
 			dev_priv->last_vblank_valid = 1;
 		}
-		if (!(atomic_read(&dev->vbl_received) & 0xFF)) {
+		if (!(atomic_read(&dev_priv->vbl_received) & 0xFF)) {
 			DRM_DEBUG("US per vblank is: %u\n",
 				  dev_priv->usec_per_vblank);
 		}
-		DRM_WAKEUP(&dev->vbl_queue);
-		drm_vbl_send_signals(dev);
+		drm_handle_vblank(dev, 0);
 		handled = 1;
 	}
 
@@ -170,31 +178,25 @@ static __inline__ void viadrv_acknowledge_irqs(drm_via_private_t * dev_priv)
 	}
 }
 
-int via_driver_vblank_wait(struct drm_device * dev, unsigned int *sequence)
+int via_enable_vblank(struct drm_device *dev, int crtc)
 {
-	drm_via_private_t *dev_priv = (drm_via_private_t *) dev->dev_private;
-	unsigned int cur_vblank;
-	int ret = 0;
+	drm_via_private_t *dev_priv = dev->dev_private;
+	u32 status;
 
-	DRM_DEBUG("viadrv_vblank_wait\n");
-	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+	if (crtc != 0) {
+		DRM_ERROR("%s:  bad crtc %d\n", __FUNCTION__, crtc);
 		return -EINVAL;
 	}
 
-	viadrv_acknowledge_irqs(dev_priv);
-
-	/* Assume that the user has missed the current sequence number
-	 * by about a day rather than she wants to wait for years
-	 * using vertical blanks...
-	 */
-
-	DRM_WAIT_ON(ret, dev->vbl_queue, 3 * DRM_HZ,
-		    (((cur_vblank = atomic_read(&dev->vbl_received)) -
-		      *sequence) <= (1 << 23)));
+	status = VIA_READ(VIA_REG_INTERRUPT);
+	VIA_WRITE(VIA_REG_INTERRUPT, status & VIA_IRQ_VBLANK_ENABLE);
+	return 0;
+}
 
-	*sequence = cur_vblank;
-	return ret;
+void via_disable_vblank(struct drm_device *dev, int crtc)
+{
+	if (crtc != 0)
+		DRM_ERROR("%s:  bad crtc %d\n", __FUNCTION__, crtc);
 }
 
 static int
@@ -208,24 +210,23 @@ via_driver_irq_wait(struct drm_device * dev, unsigned int irq, int force_sequenc
 	maskarray_t *masks;
 	int real_irq;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	if (!dev_priv) {
-		DRM_ERROR("%s called with no initialization\n", __FUNCTION__);
+		DRM_ERROR("called with no initialization\n");
 		return -EINVAL;
 	}
 
 	if (irq >= drm_via_irq_num) {
-		DRM_ERROR("%s Trying to wait on unknown irq %d\n", __FUNCTION__,
-			  irq);
+		DRM_ERROR("Trying to wait on unknown irq %d\n", irq);
 		return -EINVAL;
 	}
 
 	real_irq = dev_priv->irq_map[irq];
 
 	if (real_irq < 0) {
-		DRM_ERROR("%s Video IRQ %d not available on this hardware.\n",
-			  __FUNCTION__, irq);
+		DRM_ERROR("Video IRQ %d not available on this hardware.\n",
+			  irq);
 		return -EINVAL;
 	}
 
@@ -259,7 +260,7 @@ void via_driver_irq_preinstall(struct drm_device * dev)
 	drm_via_irq_t *cur_irq;
 	int i;
 
-	DRM_DEBUG("driver_irq_preinstall: dev_priv: %p\n", dev_priv);
+	DRM_DEBUG("dev_priv: %p\n", dev_priv);
 	if (dev_priv) {
 		cur_irq = dev_priv->via_irqs;
 
@@ -301,23 +302,24 @@ void via_driver_irq_preinstall(struct drm_device * dev)
 	}
 }
 
-void via_driver_irq_postinstall(struct drm_device * dev)
+int via_driver_irq_postinstall(struct drm_device * dev)
 {
 	drm_via_private_t *dev_priv = (drm_via_private_t *) dev->dev_private;
 	u32 status;
 
 	DRM_DEBUG("via_driver_irq_postinstall\n");
-	if (dev_priv) {
-		status = VIA_READ(VIA_REG_INTERRUPT);
-		VIA_WRITE(VIA_REG_INTERRUPT, status | VIA_IRQ_GLOBAL
-			  | dev_priv->irq_enable_mask);
+	if (!dev_priv)
+		return -EINVAL;
 
-		/* Some magic, oh for some data sheets ! */
+	status = VIA_READ(VIA_REG_INTERRUPT);
+	VIA_WRITE(VIA_REG_INTERRUPT, status | VIA_IRQ_GLOBAL
+		  | dev_priv->irq_enable_mask);
 
-		VIA_WRITE8(0x83d4, 0x11);
-		VIA_WRITE8(0x83d5, VIA_READ8(0x83d5) | 0x30);
+	/* Some magic, oh for some data sheets ! */
+	VIA_WRITE8(0x83d4, 0x11);
+	VIA_WRITE8(0x83d5, VIA_READ8(0x83d5) | 0x30);
 
-	}
+	return 0;
 }
 
 void via_driver_irq_uninstall(struct drm_device * dev)
@@ -325,7 +327,7 @@ void via_driver_irq_uninstall(struct drm_device * dev)
 	drm_via_private_t *dev_priv = (drm_via_private_t *) dev->dev_private;
 	u32 status;
 
-	DRM_DEBUG("driver_irq_uninstall)\n");
+	DRM_DEBUG("\n");
 	if (dev_priv) {
 
 		/* Some more magic, oh for some data sheets ! */
@@ -352,7 +354,7 @@ int via_wait_irq(struct drm_device *dev, void *data, struct drm_file *file_priv)
 		return -EINVAL;
 
 	if (irqwait->request.irq >= dev_priv->num_irqs) {
-		DRM_ERROR("%s Trying to wait on unknown irq %d\n", __FUNCTION__,
+		DRM_ERROR("Trying to wait on unknown irq %d\n",
 			  irqwait->request.irq);
 		return -EINVAL;
 	}
@@ -371,8 +373,7 @@ int via_wait_irq(struct drm_device *dev, void *data, struct drm_file *file_priv)
 	}
 
 	if (irqwait->request.type & VIA_IRQ_SIGNAL) {
-		DRM_ERROR("%s Signals on Via IRQs not implemented yet.\n",
-			  __FUNCTION__);
+		DRM_ERROR("Signals on Via IRQs not implemented yet.\n");
 		return -EINVAL;
 	}
 
diff --git a/shared-core/via_map.c b/shared-core/via_map.c
index b8f61673..11bfa551 100644
--- a/shared-core/via_map.c
+++ b/shared-core/via_map.c
@@ -30,7 +30,7 @@ static int via_do_init_map(struct drm_device * dev, drm_via_init_t * init)
 	drm_via_private_t *dev_priv = dev->dev_private;
 	int ret = 0;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	dev_priv->sarea = drm_getsarea(dev);
 	if (!dev_priv->sarea) {
@@ -95,7 +95,7 @@ int via_map_init(struct drm_device *dev, void *data, struct drm_file *file_priv)
 {
 	drm_via_init_t *init = data;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	switch (init->func) {
 	case VIA_INIT_MAP:
diff --git a/shared-core/via_video.c b/shared-core/via_video.c
index c15e75b5..6ec04ac1 100644
--- a/shared-core/via_video.c
+++ b/shared-core/via_video.c
@@ -33,7 +33,7 @@ void via_init_futex(drm_via_private_t * dev_priv)
 {
 	unsigned int i;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	for (i = 0; i < VIA_NR_XVMC_LOCKS; ++i) {
 		DRM_INIT_WAITQUEUE(&(dev_priv->decoder_queue[i]));
@@ -73,7 +73,7 @@ int via_decoder_futex(struct drm_device *dev, void *data, struct drm_file *file_
 	drm_via_sarea_t *sAPriv = dev_priv->sarea_priv;
 	int ret = 0;
 
-	DRM_DEBUG("%s\n", __FUNCTION__);
+	DRM_DEBUG("\n");
 
 	if (fx->lock > VIA_NR_XVMC_LOCKS)
 		return -EFAULT;