Merge branch 'sh/stable-updates'

61 files changed, 1208 insertions, 726 deletions

diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig
index 96eddd17e050..305c59003963 100644
--- a/drivers/gpu/drm/Kconfig
+++ b/drivers/gpu/drm/Kconfig
@@ -66,6 +66,8 @@ config DRM_RADEON
 
           If M is selected, the module will be called radeon.
 
+source "drivers/gpu/drm/radeon/Kconfig"
+
 config DRM_I810
         tristate "Intel I810"
         depends on DRM && AGP && AGP_INTEL
diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c
index defcaf108460..f665b05592f3 100644
--- a/drivers/gpu/drm/drm_edid.c
+++ b/drivers/gpu/drm/drm_edid.c
@@ -633,8 +633,7 @@ static struct drm_display_mode *drm_mode_detailed(struct drm_device *dev,
                 return NULL;
         }
         if (!(pt->misc & DRM_EDID_PT_SEPARATE_SYNC)) {
-                printk(KERN_WARNING "integrated sync not supported\n");
-                return NULL;
+                printk(KERN_WARNING "composite sync not supported\n");
         }
 
         /* it is incorrect if hsync/vsync width is zero */
diff --git a/drivers/gpu/drm/drm_fb_helper.c b/drivers/gpu/drm/drm_fb_helper.c
index 1c2b7d44ec05..0f9e90552dc4 100644
--- a/drivers/gpu/drm/drm_fb_helper.c
+++ b/drivers/gpu/drm/drm_fb_helper.c
@@ -389,7 +389,7 @@ int drm_fb_helper_blank(int blank, struct fb_info *info)
                 break;
         /* Display: Off; HSync: On, VSync: On */
         case FB_BLANK_NORMAL:
-                drm_fb_helper_off(info, DRM_MODE_DPMS_ON);
+                drm_fb_helper_off(info, DRM_MODE_DPMS_STANDBY);
                 break;
         /* Display: Off; HSync: Off, VSync: On */
         case FB_BLANK_HSYNC_SUSPEND:
diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c
index e9dbb481c469..8bf3770f294e 100644
--- a/drivers/gpu/drm/drm_gem.c
+++ b/drivers/gpu/drm/drm_gem.c
@@ -142,19 +142,6 @@ drm_gem_object_alloc(struct drm_device *dev, size_t size)
         if (IS_ERR(obj->filp))
                 goto free;
 
-        /* Basically we want to disable the OOM killer and handle ENOMEM
-         * ourselves by sacrificing pages from cached buffers.
-         * XXX shmem_file_[gs]et_gfp_mask()
-         */
-        mapping_set_gfp_mask(obj->filp->f_path.dentry->d_inode->i_mapping,
-                             GFP_HIGHUSER |
-                             __GFP_COLD |
-                             __GFP_FS |
-                             __GFP_RECLAIMABLE |
-                             __GFP_NORETRY |
-                             __GFP_NOWARN |
-                             __GFP_NOMEMALLOC);
-
         kref_init(&obj->refcount);
         kref_init(&obj->handlecount);
         obj->size = size;
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 9c9998c4dceb..a894ade03093 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -290,7 +290,7 @@ static int i915_batchbuffer_info(struct seq_file *m, void *data)
         list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
                 obj = obj_priv->obj;
                 if (obj->read_domains & I915_GEM_DOMAIN_COMMAND) {
-                    ret = i915_gem_object_get_pages(obj);
+                    ret = i915_gem_object_get_pages(obj, 0);
                     if (ret) {
                             DRM_ERROR("Failed to get pages: %d\n", ret);
                             spin_unlock(&dev_priv->mm.active_list_lock);
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 2c1669488b5a..aaf934d96f21 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -872,7 +872,7 @@ int i915_gem_attach_phys_object(struct drm_device *dev,
 void i915_gem_detach_phys_object(struct drm_device *dev,
                                  struct drm_gem_object *obj);
 void i915_gem_free_all_phys_object(struct drm_device *dev);
-int i915_gem_object_get_pages(struct drm_gem_object *obj);
+int i915_gem_object_get_pages(struct drm_gem_object *obj, gfp_t gfpmask);
 void i915_gem_object_put_pages(struct drm_gem_object *obj);
 void i915_gem_release(struct drm_device * dev, struct drm_file *file_priv);
 void i915_gem_object_flush_write_domain(struct drm_gem_object *obj);
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 0c67924ca80c..dda787aafcc6 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -277,7 +277,7 @@ i915_gem_shmem_pread_fast(struct drm_device *dev, struct drm_gem_object *obj,
 
         mutex_lock(&dev->struct_mutex);
 
-        ret = i915_gem_object_get_pages(obj);
+        ret = i915_gem_object_get_pages(obj, 0);
         if (ret != 0)
                 goto fail_unlock;
 
@@ -321,40 +321,24 @@ fail_unlock:
         return ret;
 }
 
-static inline gfp_t
-i915_gem_object_get_page_gfp_mask (struct drm_gem_object *obj)
-{
-        return mapping_gfp_mask(obj->filp->f_path.dentry->d_inode->i_mapping);
-}
-
-static inline void
-i915_gem_object_set_page_gfp_mask (struct drm_gem_object *obj, gfp_t gfp)
-{
-        mapping_set_gfp_mask(obj->filp->f_path.dentry->d_inode->i_mapping, gfp);
-}
-
 static int
 i915_gem_object_get_pages_or_evict(struct drm_gem_object *obj)
 {
         int ret;
 
-        ret = i915_gem_object_get_pages(obj);
+        ret = i915_gem_object_get_pages(obj, __GFP_NORETRY | __GFP_NOWARN);
 
         /* If we've insufficient memory to map in the pages, attempt
          * to make some space by throwing out some old buffers.
          */
         if (ret == -ENOMEM) {
                 struct drm_device *dev = obj->dev;
-                gfp_t gfp;
 
                 ret = i915_gem_evict_something(dev, obj->size);
                 if (ret)
                         return ret;
 
-                gfp = i915_gem_object_get_page_gfp_mask(obj);
-                i915_gem_object_set_page_gfp_mask(obj, gfp & ~__GFP_NORETRY);
-                ret = i915_gem_object_get_pages(obj);
-                i915_gem_object_set_page_gfp_mask (obj, gfp);
+                ret = i915_gem_object_get_pages(obj, 0);
         }
 
         return ret;
@@ -790,7 +774,7 @@ i915_gem_shmem_pwrite_fast(struct drm_device *dev, struct drm_gem_object *obj,
 
         mutex_lock(&dev->struct_mutex);
 
-        ret = i915_gem_object_get_pages(obj);
+        ret = i915_gem_object_get_pages(obj, 0);
         if (ret != 0)
                 goto fail_unlock;
 
@@ -2230,7 +2214,8 @@ i915_gem_evict_something(struct drm_device *dev, int min_size)
 }
 
 int
-i915_gem_object_get_pages(struct drm_gem_object *obj)
+i915_gem_object_get_pages(struct drm_gem_object *obj,
+                          gfp_t gfpmask)
 {
         struct drm_i915_gem_object *obj_priv = obj->driver_private;
         int page_count, i;
@@ -2256,7 +2241,10 @@ i915_gem_object_get_pages(struct drm_gem_object *obj)
         inode = obj->filp->f_path.dentry->d_inode;
         mapping = inode->i_mapping;
         for (i = 0; i < page_count; i++) {
-                page = read_mapping_page(mapping, i, NULL);
+                page = read_cache_page_gfp(mapping, i,
+                                           mapping_gfp_mask (mapping) |
+                                           __GFP_COLD |
+                                           gfpmask);
                 if (IS_ERR(page)) {
                         ret = PTR_ERR(page);
                         i915_gem_object_put_pages(obj);
@@ -2579,7 +2567,7 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment)
         drm_i915_private_t *dev_priv = dev->dev_private;
         struct drm_i915_gem_object *obj_priv = obj->driver_private;
         struct drm_mm_node *free_space;
-        bool retry_alloc = false;
+        gfp_t gfpmask =  __GFP_NORETRY | __GFP_NOWARN;
         int ret;
 
         if (obj_priv->madv != I915_MADV_WILLNEED) {
@@ -2623,15 +2611,7 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment)
         DRM_INFO("Binding object of size %zd at 0x%08x\n",
                  obj->size, obj_priv->gtt_offset);
 #endif
-        if (retry_alloc) {
-                i915_gem_object_set_page_gfp_mask (obj,
-                                                   i915_gem_object_get_page_gfp_mask (obj) & ~__GFP_NORETRY);
-        }
-        ret = i915_gem_object_get_pages(obj);
-        if (retry_alloc) {
-                i915_gem_object_set_page_gfp_mask (obj,
-                                                   i915_gem_object_get_page_gfp_mask (obj) | __GFP_NORETRY);
-        }
+        ret = i915_gem_object_get_pages(obj, gfpmask);
         if (ret) {
                 drm_mm_put_block(obj_priv->gtt_space);
                 obj_priv->gtt_space = NULL;
@@ -2641,9 +2621,9 @@ i915_gem_object_bind_to_gtt(struct drm_gem_object *obj, unsigned alignment)
                         ret = i915_gem_evict_something(dev, obj->size);
                         if (ret) {
                                 /* now try to shrink everyone else */
-                                if (! retry_alloc) {
-                                    retry_alloc = true;
-                                    goto search_free;
+                                if (gfpmask) {
+                                        gfpmask = 0;
+                                        goto search_free;
                                 }
 
                                 return ret;
@@ -4946,7 +4926,7 @@ void i915_gem_detach_phys_object(struct drm_device *dev,
         if (!obj_priv->phys_obj)
                 return;
 
-        ret = i915_gem_object_get_pages(obj);
+        ret = i915_gem_object_get_pages(obj, 0);
         if (ret)
                 goto out;
 
@@ -5004,7 +4984,7 @@ i915_gem_attach_phys_object(struct drm_device *dev,
         obj_priv->phys_obj = dev_priv->mm.phys_objs[id - 1];
         obj_priv->phys_obj->cur_obj = obj;
 
-        ret = i915_gem_object_get_pages(obj);
+        ret = i915_gem_object_get_pages(obj, 0);
         if (ret) {
                 DRM_ERROR("failed to get page list\n");
                 goto out;
diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.c b/drivers/gpu/drm/nouveau/nouveau_bios.c
index ba143972769f..d7f8d8b4a4b8 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bios.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bios.c
@@ -310,63 +310,22 @@ valid_reg(struct nvbios *bios, uint32_t reg)
         struct drm_device *dev = bios->dev;
 
         /* C51 has misaligned regs on purpose. Marvellous */
-        if (reg & 0x2 || (reg & 0x1 && dev_priv->VBIOS.pub.chip_version != 0x51)) {
-                NV_ERROR(dev, "========== misaligned reg 0x%08X ==========\n",
-                         reg);
-                return 0;
-        }
-        /*
-         * Warn on C51 regs that have not been verified accessible in
-         * mmiotracing
-         */
+        if (reg & 0x2 ||
+            (reg & 0x1 && dev_priv->VBIOS.pub.chip_version != 0x51))
+                NV_ERROR(dev, "======= misaligned reg 0x%08X =======\n", reg);
+
+        /* warn on C51 regs that haven't been verified accessible in tracing */
         if (reg & 0x1 && dev_priv->VBIOS.pub.chip_version == 0x51 &&
             reg != 0x130d && reg != 0x1311 && reg != 0x60081d)
                 NV_WARN(dev, "=== C51 misaligned reg 0x%08X not verified ===\n",
                         reg);
 
-        /* Trust the init scripts on G80 */
-        if (dev_priv->card_type >= NV_50)
-                return 1;
-
-        #define WITHIN(x, y, z) ((x >= y) && (x < y + z))
-        if (WITHIN(reg, NV_PMC_OFFSET, NV_PMC_SIZE))
-                return 1;
-        if (WITHIN(reg, NV_PBUS_OFFSET, NV_PBUS_SIZE))
-                return 1;
-        if (WITHIN(reg, NV_PFIFO_OFFSET, NV_PFIFO_SIZE))
-                return 1;
-        if (dev_priv->VBIOS.pub.chip_version >= 0x30 &&
-            (WITHIN(reg, 0x4000, 0x600) || reg == 0x00004600))
-                return 1;
-        if (dev_priv->VBIOS.pub.chip_version >= 0x40 &&
-                                                WITHIN(reg, 0xc000, 0x48))
-                return 1;
-        if (dev_priv->VBIOS.pub.chip_version >= 0x17 && reg == 0x0000d204)
-                return 1;
-        if (dev_priv->VBIOS.pub.chip_version >= 0x40) {
-                if (reg == 0x00011014 || reg == 0x00020328)
-                        return 1;
-                if (WITHIN(reg, 0x88000, NV_PBUS_SIZE)) /* new PBUS */
-                        return 1;
+        if (reg >= (8*1024*1024)) {
+                NV_ERROR(dev, "=== reg 0x%08x out of mapped bounds ===\n", reg);
+                return 0;
         }
-        if (WITHIN(reg, NV_PFB_OFFSET, NV_PFB_SIZE))
-                return 1;
-        if (WITHIN(reg, NV_PEXTDEV_OFFSET, NV_PEXTDEV_SIZE))
-                return 1;
-        if (WITHIN(reg, NV_PCRTC0_OFFSET, NV_PCRTC0_SIZE * 2))
-                return 1;
-        if (WITHIN(reg, NV_PRAMDAC0_OFFSET, NV_PRAMDAC0_SIZE * 2))
-                return 1;
-        if (dev_priv->VBIOS.pub.chip_version >= 0x17 && reg == 0x0070fff0)
-                return 1;
-        if (dev_priv->VBIOS.pub.chip_version == 0x51 &&
-                                WITHIN(reg, NV_PRAMIN_OFFSET, NV_PRAMIN_SIZE))
-                return 1;
-        #undef WITHIN
 
-        NV_ERROR(dev, "========== unknown reg 0x%08X ==========\n", reg);
-
-        return 0;
+        return 1;
 }
 
 static bool
@@ -3196,16 +3155,25 @@ static int call_lvds_manufacturer_script(struct drm_device *dev, struct dcb_entr
         }
 #ifdef __powerpc__
         /* Powerbook specific quirks */
-        if (script == LVDS_RESET && ((dev->pci_device & 0xffff) == 0x0179 || (dev->pci_device & 0xffff) == 0x0329))
-                nv_write_tmds(dev, dcbent->or, 0, 0x02, 0x72);
-        if ((dev->pci_device & 0xffff) == 0x0179 || (dev->pci_device & 0xffff) == 0x0189 || (dev->pci_device & 0xffff) == 0x0329) {
-                if (script == LVDS_PANEL_ON) {
-                        bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL, bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL) | (1 << 31));
-                        bios_wr32(bios, NV_PCRTC_GPIO_EXT, bios_rd32(bios, NV_PCRTC_GPIO_EXT) | 1);
-                }
-                if (script == LVDS_PANEL_OFF) {
-                        bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL, bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL) & ~(1 << 31));
-                        bios_wr32(bios, NV_PCRTC_GPIO_EXT, bios_rd32(bios, NV_PCRTC_GPIO_EXT) & ~3);
+        if ((dev->pci_device & 0xffff) == 0x0179 ||
+            (dev->pci_device & 0xffff) == 0x0189 ||
+            (dev->pci_device & 0xffff) == 0x0329) {
+                if (script == LVDS_RESET) {
+                        nv_write_tmds(dev, dcbent->or, 0, 0x02, 0x72);
+
+                } else if (script == LVDS_PANEL_ON) {
+                        bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL,
+                                  bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL)
+                                  | (1 << 31));
+                        bios_wr32(bios, NV_PCRTC_GPIO_EXT,
+                                  bios_rd32(bios, NV_PCRTC_GPIO_EXT) | 1);
+
+                } else if (script == LVDS_PANEL_OFF) {
+                        bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL,
+                                  bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL)
+                                  & ~(1 << 31));
+                        bios_wr32(bios, NV_PCRTC_GPIO_EXT,
+                                  bios_rd32(bios, NV_PCRTC_GPIO_EXT) & ~3);
                 }
         }
 #endif
@@ -5434,52 +5402,49 @@ static bool
 parse_dcb15_entry(struct drm_device *dev, struct parsed_dcb *dcb,
                   uint32_t conn, uint32_t conf, struct dcb_entry *entry)
 {
-        if (conn != 0xf0003f00 && conn != 0xf2247f10 && conn != 0xf2204001 &&
-            conn != 0xf2204301 && conn != 0xf2204311 && conn != 0xf2208001 &&
-            conn != 0xf2244001 && conn != 0xf2244301 && conn != 0xf2244311 &&
-            conn != 0xf4204011 && conn != 0xf4208011 && conn != 0xf4248011 &&
-            conn != 0xf2045ff2 && conn != 0xf2045f14 && conn != 0xf207df14 &&
-            conn != 0xf2205004 && conn != 0xf2209004) {
-                NV_ERROR(dev, "Unknown DCB 1.5 entry, please report\n");
-
-                /* cause output setting to fail for !TV, so message is seen */
-                if ((conn & 0xf) != 0x1)
-                        dcb->entries = 0;
-
-                return false;
-        }
-        /* most of the below is a "best guess" atm */
-        entry->type = conn & 0xf;
-        if (entry->type == 2)
-                /* another way of specifying straps based lvds... */
+        switch (conn & 0x0000000f) {
+        case 0:
+                entry->type = OUTPUT_ANALOG;
+                break;
+        case 1:
+                entry->type = OUTPUT_TV;
+                break;
+        case 2:
+        case 3:
                 entry->type = OUTPUT_LVDS;
-        if (entry->type == 4) { /* digital */
-                if (conn & 0x10)
-                        entry->type = OUTPUT_LVDS;
-                else
+                break;
+        case 4:
+                switch ((conn & 0x000000f0) >> 4) {
+                case 0:
                         entry->type = OUTPUT_TMDS;
+                        break;
+                case 1:
+                        entry->type = OUTPUT_LVDS;
+                        break;
+                default:
+                        NV_ERROR(dev, "Unknown DCB subtype 4/%d\n",
+                                 (conn & 0x000000f0) >> 4);
+                        return false;
+                }
+                break;
+        default:
+                NV_ERROR(dev, "Unknown DCB type %d\n", conn & 0x0000000f);
+                return false;
         }
-        /* what's in bits 5-13? could be some encoder maker thing, in tv case */
-        entry->i2c_index = (conn >> 14) & 0xf;
-        /* raw heads field is in range 0-1, so move to 1-2 */
-        entry->heads = ((conn >> 18) & 0x7) + 1;
-        entry->location = (conn >> 21) & 0xf;
-        /* unused: entry->bus = (conn >> 25) & 0x7; */
-        /* set or to be same as heads -- hopefully safe enough */
-        entry->or = entry->heads;
+
+        entry->i2c_index = (conn & 0x0003c000) >> 14;
+        entry->heads = ((conn & 0x001c0000) >> 18) + 1;
+        entry->or = entry->heads; /* same as heads, hopefully safe enough */
+        entry->location = (conn & 0x01e00000) >> 21;
+        entry->bus = (conn & 0x0e000000) >> 25;
         entry->duallink_possible = false;
 
         switch (entry->type) {
         case OUTPUT_ANALOG:
                 entry->crtconf.maxfreq = (conf & 0xffff) * 10;
                 break;
-        case OUTPUT_LVDS:
-                /*
-                 * This is probably buried in conn's unknown bits.
-                 * This will upset EDID-ful models, if they exist
-                 */
-                entry->lvdsconf.use_straps_for_mode = true;
-                entry->lvdsconf.use_power_scripts = true;
+        case OUTPUT_TV:
+                entry->tvconf.has_component_output = false;
                 break;
         case OUTPUT_TMDS:
                 /*
@@ -5488,8 +5453,12 @@ parse_dcb15_entry(struct drm_device *dev, struct parsed_dcb *dcb,
                  */
                 fabricate_vga_output(dcb, entry->i2c_index, entry->heads);
                 break;
-        case OUTPUT_TV:
-                entry->tvconf.has_component_output = false;
+        case OUTPUT_LVDS:
+                if ((conn & 0x00003f00) != 0x10)
+                        entry->lvdsconf.use_straps_for_mode = true;
+                entry->lvdsconf.use_power_scripts = true;
+                break;
+        default:
                 break;
         }
 
@@ -5564,11 +5533,13 @@ void merge_like_dcb_entries(struct drm_device *dev, struct parsed_dcb *dcb)
         dcb->entries = newentries;
 }
 
-static int parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads)
+static int
+parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads)
 {
+        struct drm_nouveau_private *dev_priv = dev->dev_private;
         struct bios_parsed_dcb *bdcb = &bios->bdcb;
         struct parsed_dcb *dcb;
-        uint16_t dcbptr, i2ctabptr = 0;
+        uint16_t dcbptr = 0, i2ctabptr = 0;
         uint8_t *dcbtable;
         uint8_t headerlen = 0x4, entries = DCB_MAX_NUM_ENTRIES;
         bool configblock = true;
@@ -5579,16 +5550,18 @@ static int parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool two
         dcb->entries = 0;
 
         /* get the offset from 0x36 */
-        dcbptr = ROM16(bios->data[0x36]);
+        if (dev_priv->card_type > NV_04) {
+                dcbptr = ROM16(bios->data[0x36]);
+                if (dcbptr == 0x0000)
+                        NV_WARN(dev, "No output data (DCB) found in BIOS\n");
+        }
 
+        /* this situation likely means a really old card, pre DCB */
         if (dcbptr == 0x0) {
-                NV_WARN(dev, "No output data (DCB) found in BIOS, "
-                               "assuming a CRT output exists\n");
-                /* this situation likely means a really old card, pre DCB */
+                NV_INFO(dev, "Assuming a CRT output exists\n");
                 fabricate_vga_output(dcb, LEGACY_I2C_CRT, 1);
 
-                if (nv04_tv_identify(dev,
-                                     bios->legacy.i2c_indices.tv) >= 0)
+                if (nv04_tv_identify(dev, bios->legacy.i2c_indices.tv) >= 0)
                         fabricate_tv_output(dcb, twoHeads);
 
                 return 0;
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c
index e342a418d434..db0ed4c13f98 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo.c
@@ -469,6 +469,8 @@ nouveau_bo_move_accel_cleanup(struct nouveau_channel *chan,
 
         ret = ttm_bo_move_accel_cleanup(&nvbo->bo, fence, NULL,
                                         evict, no_wait, new_mem);
+        if (nvbo->channel && nvbo->channel != chan)
+                ret = nouveau_fence_wait(fence, NULL, false, false);
         nouveau_fence_unref((void *)&fence);
         return ret;
 }
diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c
index 5a10deb8bdbd..7e6d673f3a23 100644
--- a/drivers/gpu/drm/nouveau/nouveau_connector.c
+++ b/drivers/gpu/drm/nouveau/nouveau_connector.c
@@ -24,9 +24,12 @@
  *
  */
 
+#include <acpi/button.h>
+
 #include "drmP.h"
 #include "drm_edid.h"
 #include "drm_crtc_helper.h"
+
 #include "nouveau_reg.h"
 #include "nouveau_drv.h"
 #include "nouveau_encoder.h"
@@ -83,14 +86,16 @@ nouveau_encoder_connector_get(struct nouveau_encoder *encoder)
 static void
 nouveau_connector_destroy(struct drm_connector *drm_connector)
 {
-        struct nouveau_connector *connector = nouveau_connector(drm_connector);
-        struct drm_device *dev = connector->base.dev;
+        struct nouveau_connector *nv_connector =
+                nouveau_connector(drm_connector);
+        struct drm_device *dev = nv_connector->base.dev;
 
         NV_DEBUG_KMS(dev, "\n");
 
-        if (!connector)
+        if (!nv_connector)
                 return;
 
+        kfree(nv_connector->edid);
         drm_sysfs_connector_remove(drm_connector);
         drm_connector_cleanup(drm_connector);
         kfree(drm_connector);
@@ -233,10 +238,21 @@ nouveau_connector_detect(struct drm_connector *connector)
         if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)
                 nv_encoder = find_encoder_by_type(connector, OUTPUT_LVDS);
         if (nv_encoder && nv_connector->native_mode) {
+#ifdef CONFIG_ACPI
+                if (!nouveau_ignorelid && !acpi_lid_open())
+                        return connector_status_disconnected;
+#endif
                 nouveau_connector_set_encoder(connector, nv_encoder);
                 return connector_status_connected;
         }
 
+        /* Cleanup the previous EDID block. */
+        if (nv_connector->edid) {
+                drm_mode_connector_update_edid_property(connector, NULL);
+                kfree(nv_connector->edid);
+                nv_connector->edid = NULL;
+        }
+
         i2c = nouveau_connector_ddc_detect(connector, &nv_encoder);
         if (i2c) {
                 nouveau_connector_ddc_prepare(connector, &flags);
@@ -247,7 +263,7 @@ nouveau_connector_detect(struct drm_connector *connector)
                 if (!nv_connector->edid) {
                         NV_ERROR(dev, "DDC responded, but no EDID for %s\n",
                                  drm_get_connector_name(connector));
-                        return connector_status_disconnected;
+                        goto detect_analog;
                 }
 
                 if (nv_encoder->dcb->type == OUTPUT_DP &&
@@ -281,6 +297,7 @@ nouveau_connector_detect(struct drm_connector *connector)
                 return connector_status_connected;
         }
 
+detect_analog:
         nv_encoder = find_encoder_by_type(connector, OUTPUT_ANALOG);
         if (!nv_encoder)
                 nv_encoder = find_encoder_by_type(connector, OUTPUT_TV);
@@ -687,8 +704,12 @@ nouveau_connector_create_lvds(struct drm_device *dev,
          */
         if (!nv_connector->edid && !nv_connector->native_mode &&
             !dev_priv->VBIOS.pub.fp_no_ddc) {
-                nv_connector->edid =
+                struct edid *edid =
                         (struct edid *)nouveau_bios_embedded_edid(dev);
+                if (edid) {
+                        nv_connector->edid = kmalloc(EDID_LENGTH, GFP_KERNEL);
+                        *(nv_connector->edid) = *edid;
+                }
         }
 
         if (!nv_connector->edid)
diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.c b/drivers/gpu/drm/nouveau/nouveau_dma.c
index 7afbe8b40d51..50d9e67745af 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dma.c
+++ b/drivers/gpu/drm/nouveau/nouveau_dma.c
@@ -126,47 +126,52 @@ OUT_RINGp(struct nouveau_channel *chan, const void *data, unsigned nr_dwords)
         chan->dma.cur += nr_dwords;
 }
 
-static inline bool
-READ_GET(struct nouveau_channel *chan, uint32_t *get)
+/* Fetch and adjust GPU GET pointer
+ *
+ * Returns:
+ *  value >= 0, the adjusted GET pointer
+ *  -EINVAL if GET pointer currently outside main push buffer
+ *  -EBUSY if timeout exceeded
+ */
+static inline int
+READ_GET(struct nouveau_channel *chan, uint32_t *prev_get, uint32_t *timeout)
 {
         uint32_t val;
 
         val = nvchan_rd32(chan, chan->user_get);
-        if (val < chan->pushbuf_base ||
-            val > chan->pushbuf_base + (chan->dma.max << 2)) {
-                /* meaningless to dma_wait() except to know whether the
-                 * GPU has stalled or not
-                 */
-                *get = val;
-                return false;
+
+        /* reset counter as long as GET is still advancing, this is
+         * to avoid misdetecting a GPU lockup if the GPU happens to
+         * just be processing an operation that takes a long time
+         */
+        if (val != *prev_get) {
+                *prev_get = val;
+                *timeout = 0;
+        }
+
+        if ((++*timeout & 0xff) == 0) {
+                DRM_UDELAY(1);
+                if (*timeout > 100000)
+                        return -EBUSY;
         }
 
-        *get = (val - chan->pushbuf_base) >> 2;
-        return true;
+        if (val < chan->pushbuf_base ||
+            val > chan->pushbuf_base + (chan->dma.max << 2))
+                return -EINVAL;
+
+        return (val - chan->pushbuf_base) >> 2;
 }
 
 int
 nouveau_dma_wait(struct nouveau_channel *chan, int size)
 {
-        uint32_t get, prev_get = 0, cnt = 0;
-        bool get_valid;
+        uint32_t prev_get = 0, cnt = 0;
+        int get;
 
         while (chan->dma.free < size) {
-                /* reset counter as long as GET is still advancing, this is
-                 * to avoid misdetecting a GPU lockup if the GPU happens to
-                 * just be processing an operation that takes a long time
-                 */
-                get_valid = READ_GET(chan, &get);
-                if (get != prev_get) {
-                        prev_get = get;
-                        cnt = 0;
-                }
-
-                if ((++cnt & 0xff) == 0) {
-                        DRM_UDELAY(1);
-                        if (cnt > 100000)
-                                return -EBUSY;
-                }
+                get = READ_GET(chan, &prev_get, &cnt);
+                if (unlikely(get == -EBUSY))
+                        return -EBUSY;
 
                 /* loop until we have a usable GET pointer.  the value
                  * we read from the GPU may be outside the main ring if
@@ -177,7 +182,7 @@ nouveau_dma_wait(struct nouveau_channel *chan, int size)
                  * from the SKIPS area, so the code below doesn't have to deal
                  * with some fun corner cases.
                  */
-                if (!get_valid || get < NOUVEAU_DMA_SKIPS)
+                if (unlikely(get == -EINVAL) || get < NOUVEAU_DMA_SKIPS)
                         continue;
 
                 if (get <= chan->dma.cur) {
@@ -203,6 +208,19 @@ nouveau_dma_wait(struct nouveau_channel *chan, int size)
                          * after processing the currently pending commands.
                          */
                         OUT_RING(chan, chan->pushbuf_base | 0x20000000);
+
+                        /* wait for GET to depart from the skips area.
+                         * prevents writing GET==PUT and causing a race
+                         * condition that causes us to think the GPU is
+                         * idle when it's not.
+                         */
+                        do {
+                                get = READ_GET(chan, &prev_get, &cnt);
+                                if (unlikely(get == -EBUSY))
+                                        return -EBUSY;
+                                if (unlikely(get == -EINVAL))
+                                        continue;
+                        } while (get <= NOUVEAU_DMA_SKIPS);
                         WRITE_PUT(NOUVEAU_DMA_SKIPS);
 
                         /* we're now submitting commands at the start of
diff --git a/drivers/gpu/drm/nouveau/nouveau_dp.c b/drivers/gpu/drm/nouveau/nouveau_dp.c
index 9e2926c48579..dd4937224220 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dp.c
+++ b/drivers/gpu/drm/nouveau/nouveau_dp.c
@@ -490,7 +490,8 @@ nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr,
                 if (!nv_wait(NV50_AUXCH_CTRL(index), 0x00010000, 0x00000000)) {
                         NV_ERROR(dev, "expected bit 16 == 0, got 0x%08x\n",
                                  nv_rd32(dev, NV50_AUXCH_CTRL(index)));
-                        return -EBUSY;
+                        ret = -EBUSY;
+                        goto out;
                 }
 
                 udelay(400);
@@ -501,6 +502,11 @@ nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr,
                         break;
         }
 
+        if ((stat & NV50_AUXCH_STAT_COUNT) != data_nr) {
+                ret = -EREMOTEIO;
+                goto out;
+        }
+
         if (cmd & 1) {
                 for (i = 0; i < 4; i++) {
                         data32[i] = nv_rd32(dev, NV50_AUXCH_DATA_IN(index, i));
diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.c b/drivers/gpu/drm/nouveau/nouveau_drv.c
index 06eb993e0883..343ab7f17ccc 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drv.c
+++ b/drivers/gpu/drm/nouveau/nouveau_drv.c
@@ -71,6 +71,10 @@ MODULE_PARM_DESC(uscript_tmds, "TMDS output script table ID (>=GeForce 8)");
 int nouveau_uscript_tmds = -1;
 module_param_named(uscript_tmds, nouveau_uscript_tmds, int, 0400);
 
+MODULE_PARM_DESC(ignorelid, "Ignore ACPI lid status");
+int nouveau_ignorelid = 0;
+module_param_named(ignorelid, nouveau_ignorelid, int, 0400);
+
 MODULE_PARM_DESC(tv_norm, "Default TV norm.\n"
                  "\t\tSupported: PAL, PAL-M, PAL-N, PAL-Nc, NTSC-M, NTSC-J,\n"
                  "\t\t\thd480i, hd480p, hd576i, hd576p, hd720p, hd1080i.\n"
diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h
index 026419fe8791..6b9690418bc7 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drv.h
+++ b/drivers/gpu/drm/nouveau/nouveau_drv.h
@@ -509,6 +509,8 @@ struct drm_nouveau_private {
         void __iomem *ramin;
         uint32_t ramin_size;
 
+        struct nouveau_bo *vga_ram;
+
         struct workqueue_struct *wq;
         struct work_struct irq_work;
 
@@ -675,6 +677,7 @@ extern char *nouveau_tv_norm;
 extern int nouveau_reg_debug;
 extern char *nouveau_vbios;
 extern int nouveau_ctxfw;
+extern int nouveau_ignorelid;
 
 /* nouveau_state.c */
 extern void nouveau_preclose(struct drm_device *dev, struct drm_file *);
diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c
index 2009db2426c3..6ac804b0c9f9 100644
--- a/drivers/gpu/drm/nouveau/nouveau_gem.c
+++ b/drivers/gpu/drm/nouveau/nouveau_gem.c
@@ -321,6 +321,7 @@ retry:
                 else {
                         NV_ERROR(dev, "invalid valid domains: 0x%08x\n",
                                  b->valid_domains);
+                        list_add_tail(&nvbo->entry, &op->both_list);
                         validate_fini(op, NULL);
                         return -EINVAL;
                 }
@@ -466,13 +467,14 @@ u_memcpya(uint64_t user, unsigned nmemb, unsigned size)
 static int
 nouveau_gem_pushbuf_reloc_apply(struct nouveau_channel *chan, int nr_bo,
                                 struct drm_nouveau_gem_pushbuf_bo *bo,
-                                int nr_relocs, uint64_t ptr_relocs,
-                                int nr_dwords, int first_dword,
+                                unsigned nr_relocs, uint64_t ptr_relocs,
+                                unsigned nr_dwords, unsigned first_dword,
                                 uint32_t *pushbuf, bool is_iomem)
 {
         struct drm_nouveau_gem_pushbuf_reloc *reloc = NULL;
         struct drm_device *dev = chan->dev;
-        int ret = 0, i;
+        int ret = 0;
+        unsigned i;
 
         reloc = u_memcpya(ptr_relocs, nr_relocs, sizeof(*reloc));
         if (IS_ERR(reloc))
@@ -667,6 +669,18 @@ nouveau_gem_ioctl_pushbuf_call(struct drm_device *dev, void *data,
         }
         pbbo = nouveau_gem_object(gem);
 
+        if ((req->offset & 3) || req->nr_dwords < 2 ||
+            (unsigned long)req->offset > (unsigned long)pbbo->bo.mem.size ||
+            (unsigned long)req->nr_dwords >
+             ((unsigned long)(pbbo->bo.mem.size - req->offset ) >> 2)) {
+                NV_ERROR(dev, "pb call misaligned or out of bounds: "
+                              "%d + %d * 4 > %ld\n",
+                         req->offset, req->nr_dwords, pbbo->bo.mem.size);
+                ret = -EINVAL;
+                drm_gem_object_unreference(gem);
+                goto out;
+        }
+
         ret = ttm_bo_reserve(&pbbo->bo, false, false, true,
                              chan->fence.sequence);
         if (ret) {
diff --git a/drivers/gpu/drm/nouveau/nouveau_irq.c b/drivers/gpu/drm/nouveau/nouveau_irq.c
index 919a619ca7fa..3b9bad66162a 100644
--- a/drivers/gpu/drm/nouveau/nouveau_irq.c
+++ b/drivers/gpu/drm/nouveau/nouveau_irq.c
@@ -483,6 +483,13 @@ nouveau_pgraph_intr_error(struct drm_device *dev, uint32_t nsource)
         if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) {
                 if (nouveau_pgraph_intr_swmthd(dev, &trap))
                         unhandled = 1;
+        } else if (nsource & NV03_PGRAPH_NSOURCE_DMA_VTX_PROTECTION) {
+                uint32_t v = nv_rd32(dev, 0x402000);
+                nv_wr32(dev, 0x402000, v);
+
+                /* dump the error anyway for now: it's useful for
+                   Gallium development */
+                unhandled = 1;
         } else {
                 unhandled = 1;
         }
diff --git a/drivers/gpu/drm/nouveau/nouveau_mem.c b/drivers/gpu/drm/nouveau/nouveau_mem.c
index fb9bdd6edf1f..8f3a12f614ed 100644
--- a/drivers/gpu/drm/nouveau/nouveau_mem.c
+++ b/drivers/gpu/drm/nouveau/nouveau_mem.c
@@ -383,9 +383,8 @@ void nouveau_mem_close(struct drm_device *dev)
 {
         struct drm_nouveau_private *dev_priv = dev->dev_private;
 
-        if (dev_priv->ttm.bdev.man[TTM_PL_PRIV0].has_type)
-                ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_PRIV0);
-        ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM);
+        nouveau_bo_unpin(dev_priv->vga_ram);
+        nouveau_bo_ref(NULL, &dev_priv->vga_ram);
 
         ttm_bo_device_release(&dev_priv->ttm.bdev);
 
@@ -622,6 +621,15 @@ nouveau_mem_init(struct drm_device *dev)
                 return ret;
         }
 
+        ret = nouveau_bo_new(dev, NULL, 256*1024, 0, TTM_PL_FLAG_VRAM,
+                             0, 0, true, true, &dev_priv->vga_ram);
+        if (ret == 0)
+                ret = nouveau_bo_pin(dev_priv->vga_ram, TTM_PL_FLAG_VRAM);
+        if (ret) {
+                NV_WARN(dev, "failed to reserve VGA memory\n");
+                nouveau_bo_ref(NULL, &dev_priv->vga_ram);
+        }
+
         /* GART */
 #if !defined(__powerpc__) && !defined(__ia64__)
         if (drm_device_is_agp(dev) && dev->agp) {
@@ -653,6 +661,7 @@ nouveau_mem_init(struct drm_device *dev)
         dev_priv->fb_mtrr = drm_mtrr_add(drm_get_resource_start(dev, 1),
                                          drm_get_resource_len(dev, 1),
                                          DRM_MTRR_WC);
+
         return 0;
 }
 
diff --git a/drivers/gpu/drm/nouveau/nouveau_state.c b/drivers/gpu/drm/nouveau/nouveau_state.c
index 09b9a46dfc0e..f2d0187ba152 100644
--- a/drivers/gpu/drm/nouveau/nouveau_state.c
+++ b/drivers/gpu/drm/nouveau/nouveau_state.c
@@ -525,6 +525,7 @@ static void nouveau_card_takedown(struct drm_device *dev)
                 engine->mc.takedown(dev);
 
                 mutex_lock(&dev->struct_mutex);
+                ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM);
                 ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_TT);
                 mutex_unlock(&dev->struct_mutex);
                 nouveau_sgdma_takedown(dev);
diff --git a/drivers/gpu/drm/nouveau/nv04_instmem.c b/drivers/gpu/drm/nouveau/nv04_instmem.c
index a20c206625a2..a3b9563a6f60 100644
--- a/drivers/gpu/drm/nouveau/nv04_instmem.c
+++ b/drivers/gpu/drm/nouveau/nv04_instmem.c
@@ -30,7 +30,7 @@ nv04_instmem_determine_amount(struct drm_device *dev)
                  * of vram.  For now, only reserve a small piece until we know
                  * more about what each chipset requires.
                  */
-                switch (dev_priv->chipset & 0xf0) {
+                switch (dev_priv->chipset) {
                 case 0x40:
                 case 0x47:
                 case 0x49:
diff --git a/drivers/gpu/drm/nouveau/nv50_crtc.c b/drivers/gpu/drm/nouveau/nv50_crtc.c
index 118d3285fd8c..40b7360841f8 100644
--- a/drivers/gpu/drm/nouveau/nv50_crtc.c
+++ b/drivers/gpu/drm/nouveau/nv50_crtc.c
@@ -432,6 +432,7 @@ nv50_crtc_prepare(struct drm_crtc *crtc)
         struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc);
         struct drm_device *dev = crtc->dev;
         struct drm_encoder *encoder;
+        uint32_t dac = 0, sor = 0;
 
         NV_DEBUG_KMS(dev, "index %d\n", nv_crtc->index);
 
@@ -439,9 +440,28 @@ nv50_crtc_prepare(struct drm_crtc *crtc)
         list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
                 struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
 
-                if (drm_helper_encoder_in_use(encoder))
+                if (!drm_helper_encoder_in_use(encoder))
                         continue;
 
+                if (nv_encoder->dcb->type == OUTPUT_ANALOG ||
+                    nv_encoder->dcb->type == OUTPUT_TV)
+                        dac |= (1 << nv_encoder->or);
+                else
+                        sor |= (1 << nv_encoder->or);
+        }
+
+        list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+                struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
+
+                if (nv_encoder->dcb->type == OUTPUT_ANALOG ||
+                    nv_encoder->dcb->type == OUTPUT_TV) {
+                        if (dac & (1 << nv_encoder->or))
+                                continue;
+                } else {
+                        if (sor & (1 << nv_encoder->or))
+                                continue;
+                }
+
                 nv_encoder->disconnect(nv_encoder);
         }
 
diff --git a/drivers/gpu/drm/nouveau/nv50_fifo.c b/drivers/gpu/drm/nouveau/nv50_fifo.c
index 39caf167587d..32b244bcb482 100644
--- a/drivers/gpu/drm/nouveau/nv50_fifo.c
+++ b/drivers/gpu/drm/nouveau/nv50_fifo.c
@@ -272,7 +272,7 @@ nv50_fifo_create_context(struct nouveau_channel *chan)
                         return ret;
                 ramfc = chan->ramfc->gpuobj;
 
-                ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, 4096, 256,
+                ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, 4096, 1024,
                                              0, &chan->cache);
                 if (ret)
                         return ret;
diff --git a/drivers/gpu/drm/nouveau/nv50_graph.c b/drivers/gpu/drm/nouveau/nv50_graph.c
index ca79f32be44c..20319e59d368 100644
--- a/drivers/gpu/drm/nouveau/nv50_graph.c
+++ b/drivers/gpu/drm/nouveau/nv50_graph.c
@@ -84,7 +84,7 @@ nv50_graph_init_regs__nv(struct drm_device *dev)
         nv_wr32(dev, 0x400804, 0xc0000000);
         nv_wr32(dev, 0x406800, 0xc0000000);
         nv_wr32(dev, 0x400c04, 0xc0000000);
-        nv_wr32(dev, 0x401804, 0xc0000000);
+        nv_wr32(dev, 0x401800, 0xc0000000);
         nv_wr32(dev, 0x405018, 0xc0000000);
         nv_wr32(dev, 0x402000, 0xc0000000);
 
@@ -282,6 +282,7 @@ nv50_graph_unload_context(struct drm_device *dev)
                 return 0;
         inst &= NV50_PGRAPH_CTXCTL_CUR_INSTANCE;
 
+        nouveau_wait_for_idle(dev);
         nv_wr32(dev, 0x400500, fifo & ~1);
         nv_wr32(dev, 0x400784, inst);
         nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x20);
diff --git a/drivers/gpu/drm/nouveau/nv50_sor.c b/drivers/gpu/drm/nouveau/nv50_sor.c
index e395c16d30f5..ecf1936b8224 100644
--- a/drivers/gpu/drm/nouveau/nv50_sor.c
+++ b/drivers/gpu/drm/nouveau/nv50_sor.c
@@ -90,11 +90,24 @@ nv50_sor_dpms(struct drm_encoder *encoder, int mode)
 {
         struct drm_device *dev = encoder->dev;
         struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
+        struct drm_encoder *enc;
         uint32_t val;
         int or = nv_encoder->or;
 
         NV_DEBUG_KMS(dev, "or %d mode %d\n", or, mode);
 
+        nv_encoder->last_dpms = mode;
+        list_for_each_entry(enc, &dev->mode_config.encoder_list, head) {
+                struct nouveau_encoder *nvenc = nouveau_encoder(enc);
+
+                if (nvenc == nv_encoder ||
+                    nvenc->dcb->or != nv_encoder->dcb->or)
+                        continue;
+
+                if (nvenc->last_dpms == DRM_MODE_DPMS_ON)
+                        return;
+        }
+
         /* wait for it to be done */
         if (!nv_wait(NV50_PDISPLAY_SOR_DPMS_CTRL(or),
                      NV50_PDISPLAY_SOR_DPMS_CTRL_PENDING, 0)) {
diff --git a/drivers/gpu/drm/radeon/atom.c b/drivers/gpu/drm/radeon/atom.c
index 388140a7e651..e3b44562d265 100644
--- a/drivers/gpu/drm/radeon/atom.c
+++ b/drivers/gpu/drm/radeon/atom.c
@@ -246,6 +246,9 @@ static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
                 case ATOM_WS_ATTRIBUTES:
                         val = gctx->io_attr;
                         break;
+                case ATOM_WS_REGPTR:
+                        val = gctx->reg_block;
+                        break;
                 default:
                         val = ctx->ws[idx];
                 }
@@ -385,6 +388,32 @@ static uint32_t atom_get_src(atom_exec_context *ctx, uint8_t attr, int *ptr)
         return atom_get_src_int(ctx, attr, ptr, NULL, 1);
 }
 
+static uint32_t atom_get_src_direct(atom_exec_context *ctx, uint8_t align, int *ptr)
+{
+        uint32_t val = 0xCDCDCDCD;
+
+        switch (align) {
+        case ATOM_SRC_DWORD:
+                val = U32(*ptr);
+                (*ptr) += 4;
+                break;
+        case ATOM_SRC_WORD0:
+        case ATOM_SRC_WORD8:
+        case ATOM_SRC_WORD16:
+                val = U16(*ptr);
+                (*ptr) += 2;
+                break;
+        case ATOM_SRC_BYTE0:
+        case ATOM_SRC_BYTE8:
+        case ATOM_SRC_BYTE16:
+        case ATOM_SRC_BYTE24:
+                val = U8(*ptr);
+                (*ptr)++;
+                break;
+        }
+        return val;
+}
+
 static uint32_t atom_get_dst(atom_exec_context *ctx, int arg, uint8_t attr,
                              int *ptr, uint32_t *saved, int print)
 {
@@ -482,6 +511,9 @@ static void atom_put_dst(atom_exec_context *ctx, int arg, uint8_t attr,
                 case ATOM_WS_ATTRIBUTES:
                         gctx->io_attr = val;
                         break;
+                case ATOM_WS_REGPTR:
+                        gctx->reg_block = val;
+                        break;
                 default:
                         ctx->ws[idx] = val;
                 }
@@ -677,7 +709,7 @@ static void atom_op_mask(atom_exec_context *ctx, int *ptr, int arg)
         SDEBUG("   dst: ");
         dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
         SDEBUG("   src1: ");
-        src1 = atom_get_src(ctx, attr, ptr);
+        src1 = atom_get_src_direct(ctx, ((attr >> 3) & 7), ptr);
         SDEBUG("   src2: ");
         src2 = atom_get_src(ctx, attr, ptr);
         dst &= src1;
@@ -809,6 +841,38 @@ static void atom_op_setregblock(atom_exec_context *ctx, int *ptr, int arg)
         SDEBUG("   base: 0x%04X\n", ctx->ctx->reg_block);
 }
 
+static void atom_op_shift_left(atom_exec_context *ctx, int *ptr, int arg)
+{
+        uint8_t attr = U8((*ptr)++), shift;
+        uint32_t saved, dst;
+        int dptr = *ptr;
+        attr &= 0x38;
+        attr |= atom_def_dst[attr >> 3] << 6;
+        SDEBUG("   dst: ");
+        dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+        shift = atom_get_src_direct(ctx, ATOM_SRC_BYTE0, ptr);
+        SDEBUG("   shift: %d\n", shift);
+        dst <<= shift;
+        SDEBUG("   dst: ");
+        atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
+static void atom_op_shift_right(atom_exec_context *ctx, int *ptr, int arg)
+{
+        uint8_t attr = U8((*ptr)++), shift;
+        uint32_t saved, dst;
+        int dptr = *ptr;
+        attr &= 0x38;
+        attr |= atom_def_dst[attr >> 3] << 6;
+        SDEBUG("   dst: ");
+        dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
+        shift = atom_get_src_direct(ctx, ATOM_SRC_BYTE0, ptr);
+        SDEBUG("   shift: %d\n", shift);
+        dst >>= shift;
+        SDEBUG("   dst: ");
+        atom_put_dst(ctx, arg, attr, &dptr, dst, saved);
+}
+
 static void atom_op_shl(atom_exec_context *ctx, int *ptr, int arg)
 {
         uint8_t attr = U8((*ptr)++), shift;
@@ -818,7 +882,7 @@ static void atom_op_shl(atom_exec_context *ctx, int *ptr, int arg)
         attr |= atom_def_dst[attr >> 3] << 6;
         SDEBUG("   dst: ");
         dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
-        shift = U8((*ptr)++);
+        shift = atom_get_src(ctx, attr, ptr);
         SDEBUG("   shift: %d\n", shift);
         dst <<= shift;
         SDEBUG("   dst: ");
@@ -834,7 +898,7 @@ static void atom_op_shr(atom_exec_context *ctx, int *ptr, int arg)
         attr |= atom_def_dst[attr >> 3] << 6;
         SDEBUG("   dst: ");
         dst = atom_get_dst(ctx, arg, attr, ptr, &saved, 1);
-        shift = U8((*ptr)++);
+        shift = atom_get_src(ctx, attr, ptr);
         SDEBUG("   shift: %d\n", shift);
         dst >>= shift;
         SDEBUG("   dst: ");
@@ -937,18 +1001,18 @@ static struct {
         atom_op_or, ATOM_ARG_FB}, {
         atom_op_or, ATOM_ARG_PLL}, {
         atom_op_or, ATOM_ARG_MC}, {
-        atom_op_shl, ATOM_ARG_REG}, {
-        atom_op_shl, ATOM_ARG_PS}, {
-        atom_op_shl, ATOM_ARG_WS}, {
-        atom_op_shl, ATOM_ARG_FB}, {
-        atom_op_shl, ATOM_ARG_PLL}, {
-        atom_op_shl, ATOM_ARG_MC}, {
-        atom_op_shr, ATOM_ARG_REG}, {
-        atom_op_shr, ATOM_ARG_PS}, {
-        atom_op_shr, ATOM_ARG_WS}, {
-        atom_op_shr, ATOM_ARG_FB}, {
-        atom_op_shr, ATOM_ARG_PLL}, {
-        atom_op_shr, ATOM_ARG_MC}, {
+        atom_op_shift_left, ATOM_ARG_REG}, {
+        atom_op_shift_left, ATOM_ARG_PS}, {
+        atom_op_shift_left, ATOM_ARG_WS}, {
+        atom_op_shift_left, ATOM_ARG_FB}, {
+        atom_op_shift_left, ATOM_ARG_PLL}, {
+        atom_op_shift_left, ATOM_ARG_MC}, {
+        atom_op_shift_right, ATOM_ARG_REG}, {
+        atom_op_shift_right, ATOM_ARG_PS}, {
+        atom_op_shift_right, ATOM_ARG_WS}, {
+        atom_op_shift_right, ATOM_ARG_FB}, {
+        atom_op_shift_right, ATOM_ARG_PLL}, {
+        atom_op_shift_right, ATOM_ARG_MC}, {
         atom_op_mul, ATOM_ARG_REG}, {
         atom_op_mul, ATOM_ARG_PS}, {
         atom_op_mul, ATOM_ARG_WS}, {
@@ -1058,8 +1122,6 @@ static void atom_execute_table_locked(struct atom_context *ctx, int index, uint3
 
         SDEBUG(">> execute %04X (len %d, WS %d, PS %d)\n", base, len, ws, ps);
 
-        /* reset reg block */
-        ctx->reg_block = 0;
         ectx.ctx = ctx;
         ectx.ps_shift = ps / 4;
         ectx.start = base;
@@ -1096,6 +1158,12 @@ static void atom_execute_table_locked(struct atom_context *ctx, int index, uint3
 void atom_execute_table(struct atom_context *ctx, int index, uint32_t * params)
 {
         mutex_lock(&ctx->mutex);
+        /* reset reg block */
+        ctx->reg_block = 0;
+        /* reset fb window */
+        ctx->fb_base = 0;
+        /* reset io mode */
+        ctx->io_mode = ATOM_IO_MM;
         atom_execute_table_locked(ctx, index, params);
         mutex_unlock(&ctx->mutex);
 }
diff --git a/drivers/gpu/drm/radeon/atom.h b/drivers/gpu/drm/radeon/atom.h
index 47fd943f6d14..bc73781423a1 100644
--- a/drivers/gpu/drm/radeon/atom.h
+++ b/drivers/gpu/drm/radeon/atom.h
@@ -91,6 +91,7 @@
 #define ATOM_WS_AND_MASK        0x45
 #define ATOM_WS_FB_WINDOW       0x46
 #define ATOM_WS_ATTRIBUTES      0x47
+#define ATOM_WS_REGPTR          0x48
 
 #define ATOM_IIO_NOP            0
 #define ATOM_IIO_START          1
diff --git a/drivers/gpu/drm/radeon/atombios_crtc.c b/drivers/gpu/drm/radeon/atombios_crtc.c
index 260fcf59f00c..af464e351fbd 100644
--- a/drivers/gpu/drm/radeon/atombios_crtc.c
+++ b/drivers/gpu/drm/radeon/atombios_crtc.c
@@ -307,7 +307,6 @@ atombios_set_crtc_dtd_timing(struct drm_crtc *crtc,
         args.susModeMiscInfo.usAccess = cpu_to_le16(misc);
         args.ucCRTC = radeon_crtc->crtc_id;
 
-        printk("executing set crtc dtd timing\n");
         atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 }
 
@@ -347,7 +346,6 @@ static void atombios_crtc_set_timing(struct drm_crtc *crtc,
         args.susModeMiscInfo.usAccess = cpu_to_le16(misc);
         args.ucCRTC = radeon_crtc->crtc_id;
 
-        printk("executing set crtc timing\n");
         atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 }
 
@@ -409,59 +407,57 @@ static void atombios_set_ss(struct drm_crtc *crtc, int enable)
         }
 }
 
-void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
+union adjust_pixel_clock {
+        ADJUST_DISPLAY_PLL_PS_ALLOCATION v1;
+};
+
+static u32 atombios_adjust_pll(struct drm_crtc *crtc,
+                               struct drm_display_mode *mode,
+                               struct radeon_pll *pll)
 {
-        struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc);
         struct drm_device *dev = crtc->dev;
         struct radeon_device *rdev = dev->dev_private;
         struct drm_encoder *encoder = NULL;
         struct radeon_encoder *radeon_encoder = NULL;
-        uint8_t frev, crev;
-        int index;
-        SET_PIXEL_CLOCK_PS_ALLOCATION args;
-        PIXEL_CLOCK_PARAMETERS *spc1_ptr;
-        PIXEL_CLOCK_PARAMETERS_V2 *spc2_ptr;
-        PIXEL_CLOCK_PARAMETERS_V3 *spc3_ptr;
-        uint32_t pll_clock = mode->clock;
-        uint32_t adjusted_clock;
-        uint32_t ref_div = 0, fb_div = 0, frac_fb_div = 0, post_div = 0;
-        struct radeon_pll *pll;
-        int pll_flags = 0;
+        u32 adjusted_clock = mode->clock;
 
-        memset(&args, 0, sizeof(args));
+        /* reset the pll flags */
+        pll->flags = 0;
 
         if (ASIC_IS_AVIVO(rdev)) {
                 if ((rdev->family == CHIP_RS600) ||
                     (rdev->family == CHIP_RS690) ||
                     (rdev->family == CHIP_RS740))
-                        pll_flags |= (RADEON_PLL_USE_FRAC_FB_DIV |
-                                      RADEON_PLL_PREFER_CLOSEST_LOWER);
+                        pll->flags |= (RADEON_PLL_USE_FRAC_FB_DIV |
+                                       RADEON_PLL_PREFER_CLOSEST_LOWER);
 
                 if (ASIC_IS_DCE32(rdev) && mode->clock > 200000)        /* range limits??? */
-                        pll_flags |= RADEON_PLL_PREFER_HIGH_FB_DIV;
+                        pll->flags |= RADEON_PLL_PREFER_HIGH_FB_DIV;
                 else
-                        pll_flags |= RADEON_PLL_PREFER_LOW_REF_DIV;
+                        pll->flags |= RADEON_PLL_PREFER_LOW_REF_DIV;
         } else {
-                pll_flags |= RADEON_PLL_LEGACY;
+                pll->flags |= RADEON_PLL_LEGACY;
 
                 if (mode->clock > 200000)       /* range limits??? */
-                        pll_flags |= RADEON_PLL_PREFER_HIGH_FB_DIV;
+                        pll->flags |= RADEON_PLL_PREFER_HIGH_FB_DIV;
                 else
-                        pll_flags |= RADEON_PLL_PREFER_LOW_REF_DIV;
+                        pll->flags |= RADEON_PLL_PREFER_LOW_REF_DIV;
 
         }
 
         list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
                 if (encoder->crtc == crtc) {
-                        if (!ASIC_IS_AVIVO(rdev)) {
-                                if (encoder->encoder_type !=
-                                    DRM_MODE_ENCODER_DAC)
-                                        pll_flags |= RADEON_PLL_NO_ODD_POST_DIV;
-                                if (encoder->encoder_type ==
-                                        DRM_MODE_ENCODER_LVDS)
-                                        pll_flags |= RADEON_PLL_USE_REF_DIV;
-                        }
                         radeon_encoder = to_radeon_encoder(encoder);
+                        if (ASIC_IS_AVIVO(rdev)) {
+                                /* DVO wants 2x pixel clock if the DVO chip is in 12 bit mode */
+                                if (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1)
+                                        adjusted_clock = mode->clock * 2;
+                        } else {
+                                if (encoder->encoder_type != DRM_MODE_ENCODER_DAC)
+                                        pll->flags |= RADEON_PLL_NO_ODD_POST_DIV;
+                                if (encoder->encoder_type == DRM_MODE_ENCODER_LVDS)
+                                        pll->flags |= RADEON_PLL_USE_REF_DIV;
+                        }
                         break;
                 }
         }
@@ -471,46 +467,101 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
          * special hw requirements.
          */
         if (ASIC_IS_DCE3(rdev)) {
-                ADJUST_DISPLAY_PLL_PS_ALLOCATION adjust_pll_args;
+                union adjust_pixel_clock args;
+                struct radeon_encoder_atom_dig *dig;
+                u8 frev, crev;
+                int index;
 
-                if (!encoder)
-                        return;
-
-                memset(&adjust_pll_args, 0, sizeof(adjust_pll_args));
-                adjust_pll_args.usPixelClock = cpu_to_le16(mode->clock / 10);
-                adjust_pll_args.ucTransmitterID = radeon_encoder->encoder_id;
-                adjust_pll_args.ucEncodeMode = atombios_get_encoder_mode(encoder);
+                if (!radeon_encoder->enc_priv)
+                        return adjusted_clock;
+                dig = radeon_encoder->enc_priv;
 
                 index = GetIndexIntoMasterTable(COMMAND, AdjustDisplayPll);
-                atom_execute_table(rdev->mode_info.atom_context,
-                                   index, (uint32_t *)&adjust_pll_args);
-                adjusted_clock = le16_to_cpu(adjust_pll_args.usPixelClock) * 10;
-        } else {
-                /* DVO wants 2x pixel clock if the DVO chip is in 12 bit mode */
-                if (ASIC_IS_AVIVO(rdev) &&
-                    (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1))
-                        adjusted_clock = mode->clock * 2;
-                else
-                        adjusted_clock = mode->clock;
+                atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev,
+                                      &crev);
+
+                memset(&args, 0, sizeof(args));
+
+                switch (frev) {
+                case 1:
+                        switch (crev) {
+                        case 1:
+                        case 2:
+                                args.v1.usPixelClock = cpu_to_le16(mode->clock / 10);
+                                args.v1.ucTransmitterID = radeon_encoder->encoder_id;
+                                args.v1.ucEncodeMode = atombios_get_encoder_mode(encoder);
+
+                                atom_execute_table(rdev->mode_info.atom_context,
+                                                   index, (uint32_t *)&args);
+                                adjusted_clock = le16_to_cpu(args.v1.usPixelClock) * 10;
+                                break;
+                        default:
+                                DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+                                return adjusted_clock;
+                        }
+                        break;
+                default:
+                        DRM_ERROR("Unknown table version %d %d\n", frev, crev);
+                        return adjusted_clock;
+                }
         }
+        return adjusted_clock;
+}
+
+union set_pixel_clock {
+        SET_PIXEL_CLOCK_PS_ALLOCATION base;
+        PIXEL_CLOCK_PARAMETERS v1;
+        PIXEL_CLOCK_PARAMETERS_V2 v2;
+        PIXEL_CLOCK_PARAMETERS_V3 v3;
+};
+
+void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
+{
+        struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc);
+        struct drm_device *dev = crtc->dev;
+        struct radeon_device *rdev = dev->dev_private;
+        struct drm_encoder *encoder = NULL;
+        struct radeon_encoder *radeon_encoder = NULL;
+        u8 frev, crev;
+        int index;
+        union set_pixel_clock args;
+        u32 pll_clock = mode->clock;
+        u32 ref_div = 0, fb_div = 0, frac_fb_div = 0, post_div = 0;
+        struct radeon_pll *pll;
+        u32 adjusted_clock;
+
+        memset(&args, 0, sizeof(args));
+
+        list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+                if (encoder->crtc == crtc) {
+                        radeon_encoder = to_radeon_encoder(encoder);
+                        break;
+                }
+        }
+
+        if (!radeon_encoder)
+                return;
 
         if (radeon_crtc->crtc_id == 0)
                 pll = &rdev->clock.p1pll;
         else
                 pll = &rdev->clock.p2pll;
 
+        /* adjust pixel clock as needed */
+        adjusted_clock = atombios_adjust_pll(crtc, mode, pll);
+
         if (ASIC_IS_AVIVO(rdev)) {
                 if (radeon_new_pll)
                         radeon_compute_pll_avivo(pll, adjusted_clock, &pll_clock,
                                                  &fb_div, &frac_fb_div,
-                                                 &ref_div, &post_div, pll_flags);
+                                                 &ref_div, &post_div);
                 else
                         radeon_compute_pll(pll, adjusted_clock, &pll_clock,
                                            &fb_div, &frac_fb_div,
-                                           &ref_div, &post_div, pll_flags);
+                                           &ref_div, &post_div);
         } else
                 radeon_compute_pll(pll, adjusted_clock, &pll_clock, &fb_div, &frac_fb_div,
-                                   &ref_div, &post_div, pll_flags);
+                                   &ref_div, &post_div);
 
         index = GetIndexIntoMasterTable(COMMAND, SetPixelClock);
         atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev,
@@ -520,45 +571,38 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
         case 1:
                 switch (crev) {
                 case 1:
-                        spc1_ptr = (PIXEL_CLOCK_PARAMETERS *) & args.sPCLKInput;
-                        spc1_ptr->usPixelClock = cpu_to_le16(mode->clock / 10);
-                        spc1_ptr->usRefDiv = cpu_to_le16(ref_div);
-                        spc1_ptr->usFbDiv = cpu_to_le16(fb_div);
-                        spc1_ptr->ucFracFbDiv = frac_fb_div;
-                        spc1_ptr->ucPostDiv = post_div;
-                        spc1_ptr->ucPpll =
+                        args.v1.usPixelClock = cpu_to_le16(mode->clock / 10);
+                        args.v1.usRefDiv = cpu_to_le16(ref_div);
+                        args.v1.usFbDiv = cpu_to_le16(fb_div);
+                        args.v1.ucFracFbDiv = frac_fb_div;
+                        args.v1.ucPostDiv = post_div;
+                        args.v1.ucPpll =
                             radeon_crtc->crtc_id ? ATOM_PPLL2 : ATOM_PPLL1;
-                        spc1_ptr->ucCRTC = radeon_crtc->crtc_id;
-                        spc1_ptr->ucRefDivSrc = 1;
+                        args.v1.ucCRTC = radeon_crtc->crtc_id;
+                        args.v1.ucRefDivSrc = 1;
                         break;
                 case 2:
-                        spc2_ptr =
-                            (PIXEL_CLOCK_PARAMETERS_V2 *) & args.sPCLKInput;
-                        spc2_ptr->usPixelClock = cpu_to_le16(mode->clock / 10);
-                        spc2_ptr->usRefDiv = cpu_to_le16(ref_div);
-                        spc2_ptr->usFbDiv = cpu_to_le16(fb_div);
-                        spc2_ptr->ucFracFbDiv = frac_fb_div;
-                        spc2_ptr->ucPostDiv = post_div;
-                        spc2_ptr->ucPpll =
+                        args.v2.usPixelClock = cpu_to_le16(mode->clock / 10);
+                        args.v2.usRefDiv = cpu_to_le16(ref_div);
+                        args.v2.usFbDiv = cpu_to_le16(fb_div);
+                        args.v2.ucFracFbDiv = frac_fb_div;
+                        args.v2.ucPostDiv = post_div;
+                        args.v2.ucPpll =
                             radeon_crtc->crtc_id ? ATOM_PPLL2 : ATOM_PPLL1;
-                        spc2_ptr->ucCRTC = radeon_crtc->crtc_id;
-                        spc2_ptr->ucRefDivSrc = 1;
+                        args.v2.ucCRTC = radeon_crtc->crtc_id;
+                        args.v2.ucRefDivSrc = 1;
                         break;
                 case 3:
-                        if (!encoder)
-                                return;
-                        spc3_ptr =
-                            (PIXEL_CLOCK_PARAMETERS_V3 *) & args.sPCLKInput;
-                        spc3_ptr->usPixelClock = cpu_to_le16(mode->clock / 10);
-                        spc3_ptr->usRefDiv = cpu_to_le16(ref_div);
-                        spc3_ptr->usFbDiv = cpu_to_le16(fb_div);
-                        spc3_ptr->ucFracFbDiv = frac_fb_div;
-                        spc3_ptr->ucPostDiv = post_div;
-                        spc3_ptr->ucPpll =
+                        args.v3.usPixelClock = cpu_to_le16(mode->clock / 10);
+                        args.v3.usRefDiv = cpu_to_le16(ref_div);
+                        args.v3.usFbDiv = cpu_to_le16(fb_div);
+                        args.v3.ucFracFbDiv = frac_fb_div;
+                        args.v3.ucPostDiv = post_div;
+                        args.v3.ucPpll =
                             radeon_crtc->crtc_id ? ATOM_PPLL2 : ATOM_PPLL1;
-                        spc3_ptr->ucMiscInfo = (radeon_crtc->crtc_id << 2);
-                        spc3_ptr->ucTransmitterId = radeon_encoder->encoder_id;
-                        spc3_ptr->ucEncoderMode =
+                        args.v3.ucMiscInfo = (radeon_crtc->crtc_id << 2);
+                        args.v3.ucTransmitterId = radeon_encoder->encoder_id;
+                        args.v3.ucEncoderMode =
                             atombios_get_encoder_mode(encoder);
                         break;
                 default:
@@ -571,12 +615,11 @@ void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
                 return;
         }
 
-        printk("executing set pll\n");
         atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args);
 }
 
-int atombios_crtc_set_base(struct drm_crtc *crtc, int x, int y,
-                           struct drm_framebuffer *old_fb)
+static int avivo_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+                               struct drm_framebuffer *old_fb)
 {
         struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc);
         struct drm_device *dev = crtc->dev;
@@ -706,6 +749,42 @@ int atombios_crtc_set_base(struct drm_crtc *crtc, int x, int y,
         return 0;
 }
 
+int atombios_crtc_set_base(struct drm_crtc *crtc, int x, int y,
+                           struct drm_framebuffer *old_fb)
+{
+        struct drm_device *dev = crtc->dev;
+        struct radeon_device *rdev = dev->dev_private;
+
+        if (ASIC_IS_AVIVO(rdev))
+                return avivo_crtc_set_base(crtc, x, y, old_fb);
+        else
+                return radeon_crtc_set_base(crtc, x, y, old_fb);
+}
+
+/* properly set additional regs when using atombios */
+static void radeon_legacy_atom_fixup(struct drm_crtc *crtc)
+{
+        struct drm_device *dev = crtc->dev;
+        struct radeon_device *rdev = dev->dev_private;
+        struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc);
+        u32 disp_merge_cntl;
+
+        switch (radeon_crtc->crtc_id) {
+        case 0:
+                disp_merge_cntl = RREG32(RADEON_DISP_MERGE_CNTL);
+                disp_merge_cntl &= ~RADEON_DISP_RGB_OFFSET_EN;
+                WREG32(RADEON_DISP_MERGE_CNTL, disp_merge_cntl);
+                break;
+        case 1:
+                disp_merge_cntl = RREG32(RADEON_DISP2_MERGE_CNTL);
+                disp_merge_cntl &= ~RADEON_DISP2_RGB_OFFSET_EN;
+                WREG32(RADEON_DISP2_MERGE_CNTL, disp_merge_cntl);
+                WREG32(RADEON_FP_H2_SYNC_STRT_WID,   RREG32(RADEON_CRTC2_H_SYNC_STRT_WID));
+                WREG32(RADEON_FP_V2_SYNC_STRT_WID,   RREG32(RADEON_CRTC2_V_SYNC_STRT_WID));
+                break;
+        }
+}
+
 int atombios_crtc_mode_set(struct drm_crtc *crtc,
                            struct drm_display_mode *mode,
                            struct drm_display_mode *adjusted_mode,
@@ -727,8 +806,8 @@ int atombios_crtc_mode_set(struct drm_crtc *crtc,
         else {
                 if (radeon_crtc->crtc_id == 0)
                         atombios_set_crtc_dtd_timing(crtc, adjusted_mode);
-                radeon_crtc_set_base(crtc, x, y, old_fb);
-                radeon_legacy_atom_set_surface(crtc);
+                atombios_crtc_set_base(crtc, x, y, old_fb);
+                radeon_legacy_atom_fixup(crtc);
         }
         atombios_overscan_setup(crtc, mode, adjusted_mode);
         atombios_scaler_setup(crtc);
@@ -746,8 +825,8 @@ static bool atombios_crtc_mode_fixup(struct drm_crtc *crtc,
 
 static void atombios_crtc_prepare(struct drm_crtc *crtc)
 {
-        atombios_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
         atombios_lock_crtc(crtc, 1);
+        atombios_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
 }
 
 static void atombios_crtc_commit(struct drm_crtc *crtc)
diff --git a/drivers/gpu/drm/radeon/atombios_dp.c b/drivers/gpu/drm/radeon/atombios_dp.c
index 3eb0ca5b3d73..71060114d5de 100644
--- a/drivers/gpu/drm/radeon/atombios_dp.c
+++ b/drivers/gpu/drm/radeon/atombios_dp.c
@@ -468,7 +468,7 @@ void radeon_dp_set_link_config(struct drm_connector *connector,
         struct radeon_connector *radeon_connector;
         struct radeon_connector_atom_dig *dig_connector;
 
-        if ((connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort) ||
+        if ((connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort) &&
             (connector->connector_type != DRM_MODE_CONNECTOR_eDP))
                 return;
 
@@ -583,7 +583,7 @@ void dp_link_train(struct drm_encoder *encoder,
         u8 train_set[4];
         int i;
 
-        if ((connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort) ||
+        if ((connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort) &&
             (connector->connector_type != DRM_MODE_CONNECTOR_eDP))
                 return;
 
@@ -596,21 +596,14 @@ void dp_link_train(struct drm_encoder *encoder,
                 return;
         dig_connector = radeon_connector->con_priv;
 
-        if (ASIC_IS_DCE32(rdev)) {
-                if (dig->dig_block)
-                        enc_id |= ATOM_DP_CONFIG_DIG2_ENCODER;
-                else
-                        enc_id |= ATOM_DP_CONFIG_DIG1_ENCODER;
-                if (dig_connector->linkb)
-                        enc_id |= ATOM_DP_CONFIG_LINK_B;
-                else
-                        enc_id |= ATOM_DP_CONFIG_LINK_A;
-        } else {
-                if (dig_connector->linkb)
-                        enc_id |= ATOM_DP_CONFIG_DIG2_ENCODER | ATOM_DP_CONFIG_LINK_B;
-                else
-                        enc_id |= ATOM_DP_CONFIG_DIG1_ENCODER | ATOM_DP_CONFIG_LINK_A;
-        }
+        if (dig->dig_encoder)
+                enc_id |= ATOM_DP_CONFIG_DIG2_ENCODER;
+        else
+                enc_id |= ATOM_DP_CONFIG_DIG1_ENCODER;
+        if (dig_connector->linkb)
+                enc_id |= ATOM_DP_CONFIG_LINK_B;
+        else
+                enc_id |= ATOM_DP_CONFIG_LINK_A;
 
         memset(link_configuration, 0, DP_LINK_CONFIGURATION_SIZE);
         if (dig_connector->dp_clock == 270000)
diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c
index 8760d66e058a..11c9a3fe6810 100644
--- a/drivers/gpu/drm/radeon/r100.c
+++ b/drivers/gpu/drm/radeon/r100.c
@@ -1504,6 +1504,7 @@ static int r100_packet3_check(struct radeon_cs_parser *p,
                         DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n");
                         return -EINVAL;
                 }
+                track->vtx_size = r100_get_vtx_size(radeon_get_ib_value(p, idx + 0));
                 track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1);
                 track->immd_dwords = pkt->count - 1;
                 r = r100_cs_track_check(p->rdev, track);
@@ -3399,9 +3400,7 @@ int r100_mc_init(struct radeon_device *rdev)
         if (rdev->flags & RADEON_IS_AGP) {
                 r = radeon_agp_init(rdev);
                 if (r) {
-                        printk(KERN_WARNING "[drm] Disabling AGP\n");
-                        rdev->flags &= ~RADEON_IS_AGP;
-                        rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024;
+                        radeon_agp_disable(rdev);
                 } else {
                         rdev->mc.gtt_location = rdev->mc.agp_base;
                 }
diff --git a/drivers/gpu/drm/radeon/r200.c b/drivers/gpu/drm/radeon/r200.c
index 20942127c46b..ff1e0cd608bf 100644
--- a/drivers/gpu/drm/radeon/r200.c
+++ b/drivers/gpu/drm/radeon/r200.c
@@ -371,13 +371,16 @@ int r200_packet0_check(struct radeon_cs_parser *p,
                 case 5:
                 case 6:
                 case 7:
+                        /* 1D/2D */
                         track->textures[i].tex_coord_type = 0;
                         break;
                 case 1:
-                        track->textures[i].tex_coord_type = 1;
+                        /* CUBE */
+                        track->textures[i].tex_coord_type = 2;
                         break;
                 case 2:
-                        track->textures[i].tex_coord_type = 2;
+                        /* 3D */
+                        track->textures[i].tex_coord_type = 1;
                         break;
                 }
                 break;
diff --git a/drivers/gpu/drm/radeon/r420.c b/drivers/gpu/drm/radeon/r420.c
index 053404e71a9d..4526faaacca8 100644
--- a/drivers/gpu/drm/radeon/r420.c
+++ b/drivers/gpu/drm/radeon/r420.c
@@ -50,9 +50,7 @@ int r420_mc_init(struct radeon_device *rdev)
         if (rdev->flags & RADEON_IS_AGP) {
                 r = radeon_agp_init(rdev);
                 if (r) {
-                        printk(KERN_WARNING "[drm] Disabling AGP\n");
-                        rdev->flags &= ~RADEON_IS_AGP;
-                        rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024;
+                        radeon_agp_disable(rdev);
                 } else {
                         rdev->mc.gtt_location = rdev->mc.agp_base;
                 }
diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c
index f5ff3490929f..1b6d0001b20e 100644
--- a/drivers/gpu/drm/radeon/r600.c
+++ b/drivers/gpu/drm/radeon/r600.c
@@ -624,7 +624,6 @@ int r600_mc_init(struct radeon_device *rdev)
         fixed20_12 a;
         u32 tmp;
         int chansize, numchan;
-        int r;
 
         /* Get VRAM informations */
         rdev->mc.vram_is_ddr = true;
@@ -667,9 +666,6 @@ int r600_mc_init(struct radeon_device *rdev)
                 rdev->mc.real_vram_size = rdev->mc.aper_size;
 
         if (rdev->flags & RADEON_IS_AGP) {
-                r = radeon_agp_init(rdev);
-                if (r)
-                        return r;
                 /* gtt_size is setup by radeon_agp_init */
                 rdev->mc.gtt_location = rdev->mc.agp_base;
                 tmp = 0xFFFFFFFFUL - rdev->mc.agp_base - rdev->mc.gtt_size;
@@ -1792,23 +1788,24 @@ void r600_fence_ring_emit(struct radeon_device *rdev,
         radeon_ring_write(rdev, RB_INT_STAT);
 }
 
-int r600_copy_dma(struct radeon_device *rdev,
-                  uint64_t src_offset,
-                  uint64_t dst_offset,
-                  unsigned num_pages,
-                  struct radeon_fence *fence)
-{
-        /* FIXME: implement */
-        return 0;
-}
-
 int r600_copy_blit(struct radeon_device *rdev,
                    uint64_t src_offset, uint64_t dst_offset,
                    unsigned num_pages, struct radeon_fence *fence)
 {
-        r600_blit_prepare_copy(rdev, num_pages * RADEON_GPU_PAGE_SIZE);
+        int r;
+
+        mutex_lock(&rdev->r600_blit.mutex);
+        rdev->r600_blit.vb_ib = NULL;
+        r = r600_blit_prepare_copy(rdev, num_pages * RADEON_GPU_PAGE_SIZE);
+        if (r) {
+                if (rdev->r600_blit.vb_ib)
+                        radeon_ib_free(rdev, &rdev->r600_blit.vb_ib);
+                mutex_unlock(&rdev->r600_blit.mutex);
+                return r;
+        }
         r600_kms_blit_copy(rdev, src_offset, dst_offset, num_pages * RADEON_GPU_PAGE_SIZE);
         r600_blit_done_copy(rdev, fence);
+        mutex_unlock(&rdev->r600_blit.mutex);
         return 0;
 }
 
@@ -1864,26 +1861,19 @@ int r600_startup(struct radeon_device *rdev)
                         return r;
         }
         r600_gpu_init(rdev);
-
-        if (!rdev->r600_blit.shader_obj) {
-                r = r600_blit_init(rdev);
+        /* pin copy shader into vram */
+        if (rdev->r600_blit.shader_obj) {
+                r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
+                if (unlikely(r != 0))
+                        return r;
+                r = radeon_bo_pin(rdev->r600_blit.shader_obj, RADEON_GEM_DOMAIN_VRAM,
+                                &rdev->r600_blit.shader_gpu_addr);
+                radeon_bo_unreserve(rdev->r600_blit.shader_obj);
                 if (r) {
-                        DRM_ERROR("radeon: failed blitter (%d).\n", r);
+                        dev_err(rdev->dev, "(%d) pin blit object failed\n", r);
                         return r;
                 }
         }
-
-        r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
-        if (unlikely(r != 0))
-                return r;
-        r = radeon_bo_pin(rdev->r600_blit.shader_obj, RADEON_GEM_DOMAIN_VRAM,
-                        &rdev->r600_blit.shader_gpu_addr);
-        radeon_bo_unreserve(rdev->r600_blit.shader_obj);
-        if (r) {
-                dev_err(rdev->dev, "(%d) pin blit object failed\n", r);
-                return r;
-        }
-
         /* Enable IRQ */
         r = r600_irq_init(rdev);
         if (r) {
@@ -1958,14 +1948,17 @@ int r600_suspend(struct radeon_device *rdev)
         /* FIXME: we should wait for ring to be empty */
         r600_cp_stop(rdev);
         rdev->cp.ready = false;
+        r600_irq_suspend(rdev);
         r600_wb_disable(rdev);
         r600_pcie_gart_disable(rdev);
         /* unpin shaders bo */
-        r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
-        if (unlikely(r != 0))
-                return r;
-        radeon_bo_unpin(rdev->r600_blit.shader_obj);
-        radeon_bo_unreserve(rdev->r600_blit.shader_obj);
+        if (rdev->r600_blit.shader_obj) {
+                r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
+                if (!r) {
+                        radeon_bo_unpin(rdev->r600_blit.shader_obj);
+                        radeon_bo_unreserve(rdev->r600_blit.shader_obj);
+                }
+        }
         return 0;
 }
 
@@ -2026,6 +2019,11 @@ int r600_init(struct radeon_device *rdev)
         r = radeon_fence_driver_init(rdev);
         if (r)
                 return r;
+        if (rdev->flags & RADEON_IS_AGP) {
+                r = radeon_agp_init(rdev);
+                if (r)
+                        radeon_agp_disable(rdev);
+        }
         r = r600_mc_init(rdev);
         if (r)
                 return r;
@@ -2047,6 +2045,12 @@ int r600_init(struct radeon_device *rdev)
         r = r600_pcie_gart_init(rdev);
         if (r)
                 return r;
+        r = r600_blit_init(rdev);
+        if (r) {
+                r600_blit_fini(rdev);
+                rdev->asic->copy = NULL;
+                dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r);
+        }
 
         rdev->accel_working = true;
         r = r600_startup(rdev);
@@ -2060,13 +2064,14 @@ int r600_init(struct radeon_device *rdev)
         if (rdev->accel_working) {
                 r = radeon_ib_pool_init(rdev);
                 if (r) {
-                        DRM_ERROR("radeon: failed initializing IB pool (%d).\n", r);
-                        rdev->accel_working = false;
-                }
-                r = r600_ib_test(rdev);
-                if (r) {
-                        DRM_ERROR("radeon: failed testing IB (%d).\n", r);
+                        dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
                         rdev->accel_working = false;
+                } else {
+                        r = r600_ib_test(rdev);
+                        if (r) {
+                                dev_err(rdev->dev, "IB test failed (%d).\n", r);
+                                rdev->accel_working = false;
+                        }
                 }
         }
 
@@ -2197,14 +2202,14 @@ void r600_ih_ring_init(struct radeon_device *rdev, unsigned ring_size)
         rb_bufsz = drm_order(ring_size / 4);
         ring_size = (1 << rb_bufsz) * 4;
         rdev->ih.ring_size = ring_size;
-        rdev->ih.align_mask = 4 - 1;
+        rdev->ih.ptr_mask = rdev->ih.ring_size - 1;
+        rdev->ih.rptr = 0;
 }
 
-static int r600_ih_ring_alloc(struct radeon_device *rdev, unsigned ring_size)
+static int r600_ih_ring_alloc(struct radeon_device *rdev)
 {
         int r;
 
-        rdev->ih.ring_size = ring_size;
         /* Allocate ring buffer */
         if (rdev->ih.ring_obj == NULL) {
                 r = radeon_bo_create(rdev, NULL, rdev->ih.ring_size,
@@ -2234,9 +2239,6 @@ static int r600_ih_ring_alloc(struct radeon_device *rdev, unsigned ring_size)
                         return r;
                 }
         }
-        rdev->ih.ptr_mask = (rdev->cp.ring_size / 4) - 1;
-        rdev->ih.rptr = 0;
-
         return 0;
 }
 
@@ -2386,7 +2388,7 @@ int r600_irq_init(struct radeon_device *rdev)
         u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
 
         /* allocate ring */
-        ret = r600_ih_ring_alloc(rdev, rdev->ih.ring_size);
+        ret = r600_ih_ring_alloc(rdev);
         if (ret)
                 return ret;
 
@@ -2449,10 +2451,15 @@ int r600_irq_init(struct radeon_device *rdev)
         return ret;
 }
 
-void r600_irq_fini(struct radeon_device *rdev)
+void r600_irq_suspend(struct radeon_device *rdev)
 {
         r600_disable_interrupts(rdev);
         r600_rlc_stop(rdev);
+}
+
+void r600_irq_fini(struct radeon_device *rdev)
+{
+        r600_irq_suspend(rdev);
         r600_ih_ring_fini(rdev);
 }
 
@@ -2467,8 +2474,12 @@ int r600_irq_set(struct radeon_device *rdev)
                 return -EINVAL;
         }
         /* don't enable anything if the ih is disabled */
-        if (!rdev->ih.enabled)
+        if (!rdev->ih.enabled) {
+                r600_disable_interrupts(rdev);
+                /* force the active interrupt state to all disabled */
+                r600_disable_interrupt_state(rdev);
                 return 0;
+        }
 
         if (ASIC_IS_DCE3(rdev)) {
                 hpd1 = RREG32(DC_HPD1_INT_CONTROL) & ~DC_HPDx_INT_EN;
@@ -2638,16 +2649,18 @@ static inline u32 r600_get_ih_wptr(struct radeon_device *rdev)
         wptr = RREG32(IH_RB_WPTR);
 
         if (wptr & RB_OVERFLOW) {
-                WARN_ON(1);
-                /* XXX deal with overflow */
-                DRM_ERROR("IH RB overflow\n");
+                /* When a ring buffer overflow happen start parsing interrupt
+                 * from the last not overwritten vector (wptr + 16). Hopefully
+                 * this should allow us to catchup.
+                 */
+                dev_warn(rdev->dev, "IH ring buffer overflow (0x%08X, %d, %d)\n",
+                        wptr, rdev->ih.rptr, (wptr + 16) + rdev->ih.ptr_mask);
+                rdev->ih.rptr = (wptr + 16) & rdev->ih.ptr_mask;
                 tmp = RREG32(IH_RB_CNTL);
                 tmp |= IH_WPTR_OVERFLOW_CLEAR;
                 WREG32(IH_RB_CNTL, tmp);
         }
-        wptr = wptr & WPTR_OFFSET_MASK;
-
-        return wptr;
+        return (wptr & rdev->ih.ptr_mask);
 }
 
 /*        r600 IV Ring
@@ -2683,12 +2696,13 @@ int r600_irq_process(struct radeon_device *rdev)
         u32 wptr = r600_get_ih_wptr(rdev);
         u32 rptr = rdev->ih.rptr;
         u32 src_id, src_data;
-        u32 last_entry = rdev->ih.ring_size - 16;
         u32 ring_index, disp_int, disp_int_cont, disp_int_cont2;
         unsigned long flags;
         bool queue_hotplug = false;
 
         DRM_DEBUG("r600_irq_process start: rptr %d, wptr %d\n", rptr, wptr);
+        if (!rdev->ih.enabled)
+                return IRQ_NONE;
 
         spin_lock_irqsave(&rdev->ih.lock, flags);
 
@@ -2817,10 +2831,8 @@ restart_ih:
                 }
 
                 /* wptr/rptr are in bytes! */
-                if (rptr == last_entry)
-                        rptr = 0;
-                else
-                        rptr += 16;
+                rptr += 16;
+                rptr &= rdev->ih.ptr_mask;
         }
         /* make sure wptr hasn't changed while processing */
         wptr = r600_get_ih_wptr(rdev);
diff --git a/drivers/gpu/drm/radeon/r600_blit_kms.c b/drivers/gpu/drm/radeon/r600_blit_kms.c
index 8787ea89dc6e..af1c3ca8a4cb 100644
--- a/drivers/gpu/drm/radeon/r600_blit_kms.c
+++ b/drivers/gpu/drm/radeon/r600_blit_kms.c
@@ -449,6 +449,7 @@ int r600_blit_init(struct radeon_device *rdev)
         u32 packet2s[16];
         int num_packet2s = 0;
 
+        mutex_init(&rdev->r600_blit.mutex);
         rdev->r600_blit.state_offset = 0;
 
         if (rdev->family >= CHIP_RV770)
@@ -512,14 +513,16 @@ void r600_blit_fini(struct radeon_device *rdev)
 {
         int r;
 
+        if (rdev->r600_blit.shader_obj == NULL)
+                return;
+        /* If we can't reserve the bo, unref should be enough to destroy
+         * it when it becomes idle.
+         */
         r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
-        if (unlikely(r != 0)) {
-                dev_err(rdev->dev, "(%d) can't finish r600 blit\n", r);
-                goto out_unref;
+        if (!r) {
+                radeon_bo_unpin(rdev->r600_blit.shader_obj);
+                radeon_bo_unreserve(rdev->r600_blit.shader_obj);
         }
-        radeon_bo_unpin(rdev->r600_blit.shader_obj);
-        radeon_bo_unreserve(rdev->r600_blit.shader_obj);
-out_unref:
         radeon_bo_unref(&rdev->r600_blit.shader_obj);
 }
 
@@ -555,7 +558,8 @@ int r600_blit_prepare_copy(struct radeon_device *rdev, int size_bytes)
         int dwords_per_loop = 76, num_loops;
 
         r = r600_vb_ib_get(rdev);
-        WARN_ON(r);
+        if (r)
+                return r;
 
         /* set_render_target emits 2 extra dwords on rv6xx */
         if (rdev->family > CHIP_R600 && rdev->family < CHIP_RV770)
@@ -581,7 +585,8 @@ int r600_blit_prepare_copy(struct radeon_device *rdev, int size_bytes)
         ring_size += 5; /* done copy */
         ring_size += 7; /* fence emit for done copy */
         r = radeon_ring_lock(rdev, ring_size);
-        WARN_ON(r);
+        if (r)
+                return r;
 
         set_default_state(rdev); /* 14 */
         set_shaders(rdev); /* 26 */
diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c
index 44060b92d9e6..e4c45ec16507 100644
--- a/drivers/gpu/drm/radeon/r600_cs.c
+++ b/drivers/gpu/drm/radeon/r600_cs.c
@@ -36,6 +36,10 @@ static int r600_cs_packet_next_reloc_nomm(struct radeon_cs_parser *p,
 typedef int (*next_reloc_t)(struct radeon_cs_parser*, struct radeon_cs_reloc**);
 static next_reloc_t r600_cs_packet_next_reloc = &r600_cs_packet_next_reloc_mm;
 
+struct r600_cs_track {
+        u32     cb_color0_base_last;
+};
+
 /**
  * r600_cs_packet_parse() - parse cp packet and point ib index to next packet
  * @parser:     parser structure holding parsing context.
@@ -177,6 +181,28 @@ static int r600_cs_packet_next_reloc_nomm(struct radeon_cs_parser *p,
 }
 
 /**
+ * r600_cs_packet_next_is_pkt3_nop() - test if next packet is packet3 nop for reloc
+ * @parser:             parser structure holding parsing context.
+ *
+ * Check next packet is relocation packet3, do bo validation and compute
+ * GPU offset using the provided start.
+ **/
+static inline int r600_cs_packet_next_is_pkt3_nop(struct radeon_cs_parser *p)
+{
+        struct radeon_cs_packet p3reloc;
+        int r;
+
+        r = r600_cs_packet_parse(p, &p3reloc, p->idx);
+        if (r) {
+                return 0;
+        }
+        if (p3reloc.type != PACKET_TYPE3 || p3reloc.opcode != PACKET3_NOP) {
+                return 0;
+        }
+        return 1;
+}
+
+/**
  * r600_cs_packet_next_vline() - parse userspace VLINE packet
  * @parser:             parser structure holding parsing context.
  *
@@ -337,6 +363,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p,
                                 struct radeon_cs_packet *pkt)
 {
         struct radeon_cs_reloc *reloc;
+        struct r600_cs_track *track;
         volatile u32 *ib;
         unsigned idx;
         unsigned i;
@@ -344,6 +371,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p,
         int r;
         u32 idx_value;
 
+        track = (struct r600_cs_track *)p->track;
         ib = p->ib->ptr;
         idx = pkt->idx + 1;
         idx_value = radeon_get_ib_value(p, idx);
@@ -503,9 +531,60 @@ static int r600_packet3_check(struct radeon_cs_parser *p,
                 for (i = 0; i < pkt->count; i++) {
                         reg = start_reg + (4 * i);
                         switch (reg) {
+                        /* This register were added late, there is userspace
+                         * which does provide relocation for those but set
+                         * 0 offset. In order to avoid breaking old userspace
+                         * we detect this and set address to point to last
+                         * CB_COLOR0_BASE, note that if userspace doesn't set
+                         * CB_COLOR0_BASE before this register we will report
+                         * error. Old userspace always set CB_COLOR0_BASE
+                         * before any of this.
+                         */
+                        case R_0280E0_CB_COLOR0_FRAG:
+                        case R_0280E4_CB_COLOR1_FRAG:
+                        case R_0280E8_CB_COLOR2_FRAG:
+                        case R_0280EC_CB_COLOR3_FRAG:
+                        case R_0280F0_CB_COLOR4_FRAG:
+                        case R_0280F4_CB_COLOR5_FRAG:
+                        case R_0280F8_CB_COLOR6_FRAG:
+                        case R_0280FC_CB_COLOR7_FRAG:
+                        case R_0280C0_CB_COLOR0_TILE:
+                        case R_0280C4_CB_COLOR1_TILE:
+                        case R_0280C8_CB_COLOR2_TILE:
+                        case R_0280CC_CB_COLOR3_TILE:
+                        case R_0280D0_CB_COLOR4_TILE:
+                        case R_0280D4_CB_COLOR5_TILE:
+                        case R_0280D8_CB_COLOR6_TILE:
+                        case R_0280DC_CB_COLOR7_TILE:
+                                if (!r600_cs_packet_next_is_pkt3_nop(p)) {
+                                        if (!track->cb_color0_base_last) {
+                                                dev_err(p->dev, "Broken old userspace ? no cb_color0_base supplied before trying to write 0x%08X\n", reg);
+                                                return -EINVAL;
+                                        }
+                                        ib[idx+1+i] = track->cb_color0_base_last;
+                                        printk_once(KERN_WARNING "radeon: You have old & broken userspace "
+                                                "please consider updating mesa & xf86-video-ati\n");
+                                } else {
+                                        r = r600_cs_packet_next_reloc(p, &reloc);
+                                        if (r) {
+                                                dev_err(p->dev, "bad SET_CONTEXT_REG 0x%04X\n", reg);
+                                                return -EINVAL;
+                                        }
+                                        ib[idx+1+i] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
+                                }
+                                break;
                         case DB_DEPTH_BASE:
                         case DB_HTILE_DATA_BASE:
                         case CB_COLOR0_BASE:
+                                r = r600_cs_packet_next_reloc(p, &reloc);
+                                if (r) {
+                                        DRM_ERROR("bad SET_CONTEXT_REG "
+                                                        "0x%04X\n", reg);
+                                        return -EINVAL;
+                                }
+                                ib[idx+1+i] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
+                                track->cb_color0_base_last = ib[idx+1+i];
+                                break;
                         case CB_COLOR1_BASE:
                         case CB_COLOR2_BASE:
                         case CB_COLOR3_BASE:
@@ -678,8 +757,11 @@ static int r600_packet3_check(struct radeon_cs_parser *p,
 int r600_cs_parse(struct radeon_cs_parser *p)
 {
         struct radeon_cs_packet pkt;
+        struct r600_cs_track *track;
         int r;
 
+        track = kzalloc(sizeof(*track), GFP_KERNEL);
+        p->track = track;
         do {
                 r = r600_cs_packet_parse(p, &pkt, p->idx);
                 if (r) {
@@ -757,6 +839,7 @@ int r600_cs_legacy(struct drm_device *dev, void *data, struct drm_file *filp,
         /* initialize parser */
         memset(&parser, 0, sizeof(struct radeon_cs_parser));
         parser.filp = filp;
+        parser.dev = &dev->pdev->dev;
         parser.rdev = NULL;
         parser.family = family;
         parser.ib = &fake_ib;
diff --git a/drivers/gpu/drm/radeon/r600d.h b/drivers/gpu/drm/radeon/r600d.h
index 05894edadab4..30480881aed1 100644
--- a/drivers/gpu/drm/radeon/r600d.h
+++ b/drivers/gpu/drm/radeon/r600d.h
@@ -882,4 +882,29 @@
 #define         S_000E60_SOFT_RESET_VMC(x)              (((x) & 1) << 17)
 
 #define R_005480_HDP_MEM_COHERENCY_FLUSH_CNTL           0x5480
+
+#define R_0280E0_CB_COLOR0_FRAG                      0x0280E0
+#define   S_0280E0_BASE_256B(x)                        (((x) & 0xFFFFFFFF) << 0)
+#define   G_0280E0_BASE_256B(x)                        (((x) >> 0) & 0xFFFFFFFF)
+#define   C_0280E0_BASE_256B                           0x00000000
+#define R_0280E4_CB_COLOR1_FRAG                      0x0280E4
+#define R_0280E8_CB_COLOR2_FRAG                      0x0280E8
+#define R_0280EC_CB_COLOR3_FRAG                      0x0280EC
+#define R_0280F0_CB_COLOR4_FRAG                      0x0280F0
+#define R_0280F4_CB_COLOR5_FRAG                      0x0280F4
+#define R_0280F8_CB_COLOR6_FRAG                      0x0280F8
+#define R_0280FC_CB_COLOR7_FRAG                      0x0280FC
+#define R_0280C0_CB_COLOR0_TILE                      0x0280C0
+#define   S_0280C0_BASE_256B(x)                        (((x) & 0xFFFFFFFF) << 0)
+#define   G_0280C0_BASE_256B(x)                        (((x) >> 0) & 0xFFFFFFFF)
+#define   C_0280C0_BASE_256B                           0x00000000
+#define R_0280C4_CB_COLOR1_TILE                      0x0280C4
+#define R_0280C8_CB_COLOR2_TILE                      0x0280C8
+#define R_0280CC_CB_COLOR3_TILE                      0x0280CC
+#define R_0280D0_CB_COLOR4_TILE                      0x0280D0
+#define R_0280D4_CB_COLOR5_TILE                      0x0280D4
+#define R_0280D8_CB_COLOR6_TILE                      0x0280D8
+#define R_0280DC_CB_COLOR7_TILE                      0x0280DC
+
+
 #endif
diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index eb5f99b9469d..2d5f2bfa7201 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -410,13 +410,13 @@ struct r600_ih {
         unsigned                wptr_old;
         unsigned                ring_size;
         uint64_t                gpu_addr;
-        uint32_t                align_mask;
         uint32_t                ptr_mask;
         spinlock_t              lock;
         bool                    enabled;
 };
 
 struct r600_blit {
+        struct mutex            mutex;
         struct radeon_bo        *shader_obj;
         u64 shader_gpu_addr;
         u32 vs_offset, ps_offset;
@@ -465,6 +465,7 @@ struct radeon_cs_chunk {
 };
 
 struct radeon_cs_parser {
+        struct device           *dev;
         struct radeon_device    *rdev;
         struct drm_file         *filp;
         /* chunks */
@@ -847,7 +848,7 @@ void r600_kms_blit_copy(struct radeon_device *rdev,
 
 static inline uint32_t r100_mm_rreg(struct radeon_device *rdev, uint32_t reg)
 {
-        if (reg < 0x10000)
+        if (reg < rdev->rmmio_size)
                 return readl(((void __iomem *)rdev->rmmio) + reg);
         else {
                 writel(reg, ((void __iomem *)rdev->rmmio) + RADEON_MM_INDEX);
@@ -857,7 +858,7 @@ static inline uint32_t r100_mm_rreg(struct radeon_device *rdev, uint32_t reg)
 
 static inline void r100_mm_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v)
 {
-        if (reg < 0x10000)
+        if (reg < rdev->rmmio_size)
                 writel(v, ((void __iomem *)rdev->rmmio) + reg);
         else {
                 writel(reg, ((void __iomem *)rdev->rmmio) + RADEON_MM_INDEX);
@@ -1017,6 +1018,8 @@ static inline void radeon_ring_write(struct radeon_device *rdev, uint32_t v)
 #define radeon_hpd_set_polarity(rdev, hpd) (rdev)->asic->hpd_set_polarity((rdev), (hpd))
 
 /* Common functions */
+/* AGP */
+extern void radeon_agp_disable(struct radeon_device *rdev);
 extern int radeon_gart_table_vram_pin(struct radeon_device *rdev);
 extern int radeon_modeset_init(struct radeon_device *rdev);
 extern void radeon_modeset_fini(struct radeon_device *rdev);
@@ -1160,7 +1163,8 @@ extern int r600_irq_init(struct radeon_device *rdev);
 extern void r600_irq_fini(struct radeon_device *rdev);
 extern void r600_ih_ring_init(struct radeon_device *rdev, unsigned ring_size);
 extern int r600_irq_set(struct radeon_device *rdev);
-
+extern void r600_irq_suspend(struct radeon_device *rdev);
+/* r600 audio */
 extern int r600_audio_init(struct radeon_device *rdev);
 extern int r600_audio_tmds_index(struct drm_encoder *encoder);
 extern void r600_audio_set_clock(struct drm_encoder *encoder, int clock);
diff --git a/drivers/gpu/drm/radeon/radeon_agp.c b/drivers/gpu/drm/radeon/radeon_agp.c
index 220f454ea9fa..c0681a5556dc 100644
--- a/drivers/gpu/drm/radeon/radeon_agp.c
+++ b/drivers/gpu/drm/radeon/radeon_agp.c
@@ -144,9 +144,19 @@ int radeon_agp_init(struct radeon_device *rdev)
 
         ret = drm_agp_info(rdev->ddev, &info);
         if (ret) {
+                drm_agp_release(rdev->ddev);
                 DRM_ERROR("Unable to get AGP info: %d\n", ret);
                 return ret;
         }
+
+        if (rdev->ddev->agp->agp_info.aper_size < 32) {
+                drm_agp_release(rdev->ddev);
+                dev_warn(rdev->dev, "AGP aperture too small (%zuM) "
+                        "need at least 32M, disabling AGP\n",
+                        rdev->ddev->agp->agp_info.aper_size);
+                return -EINVAL;
+        }
+
         mode.mode = info.mode;
         agp_status = (RREG32(RADEON_AGP_STATUS) | RADEON_AGPv3_MODE) & mode.mode;
         is_v3 = !!(agp_status & RADEON_AGPv3_MODE);
@@ -221,6 +231,7 @@ int radeon_agp_init(struct radeon_device *rdev)
         ret = drm_agp_enable(rdev->ddev, mode);
         if (ret) {
                 DRM_ERROR("Unable to enable AGP (mode = 0x%lx)\n", mode.mode);
+                drm_agp_release(rdev->ddev);
                 return ret;
         }
 
diff --git a/drivers/gpu/drm/radeon/radeon_clocks.c b/drivers/gpu/drm/radeon/radeon_clocks.c
index 812f24dbc2a8..73c4405bf42f 100644
--- a/drivers/gpu/drm/radeon/radeon_clocks.c
+++ b/drivers/gpu/drm/radeon/radeon_clocks.c
@@ -56,7 +56,7 @@ uint32_t radeon_legacy_get_engine_clock(struct radeon_device *rdev)
         else if (post_div == 3)
                 sclk >>= 2;
         else if (post_div == 4)
-                sclk >>= 4;
+                sclk >>= 3;
 
         return sclk;
 }
@@ -86,7 +86,7 @@ uint32_t radeon_legacy_get_memory_clock(struct radeon_device *rdev)
         else if (post_div == 3)
                 mclk >>= 2;
         else if (post_div == 4)
-                mclk >>= 4;
+                mclk >>= 3;
 
         return mclk;
 }
diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c
index 65590a0f1d93..1190148cf5e6 100644
--- a/drivers/gpu/drm/radeon/radeon_cs.c
+++ b/drivers/gpu/drm/radeon/radeon_cs.c
@@ -189,7 +189,7 @@ static void radeon_cs_parser_fini(struct radeon_cs_parser *parser, int error)
 {
         unsigned i;
 
-        if (error) {
+        if (error && parser->ib) {
                 radeon_bo_list_unvalidate(&parser->validated,
                                                 parser->ib->fence);
         } else {
@@ -231,6 +231,7 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
         memset(&parser, 0, sizeof(struct radeon_cs_parser));
         parser.filp = filp;
         parser.rdev = rdev;
+        parser.dev = rdev->dev;
         r = radeon_cs_parser_init(&parser, data);
         if (r) {
                 DRM_ERROR("Failed to initialize parser !\n");
diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index 0c51f8e46613..768b1509fa03 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -544,6 +544,7 @@ void radeon_agp_disable(struct radeon_device *rdev)
                 rdev->asic->gart_tlb_flush = &r100_pci_gart_tlb_flush;
                 rdev->asic->gart_set_page = &r100_pci_gart_set_page;
         }
+        rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024;
 }
 
 void radeon_check_arguments(struct radeon_device *rdev)
diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c
index 0ec491ead2ff..6a92f994cc26 100644
--- a/drivers/gpu/drm/radeon/radeon_display.c
+++ b/drivers/gpu/drm/radeon/radeon_display.c
@@ -357,7 +357,8 @@ int radeon_ddc_get_modes(struct radeon_connector *radeon_connector)
         if ((radeon_connector->base.connector_type == DRM_MODE_CONNECTOR_DisplayPort) ||
             (radeon_connector->base.connector_type == DRM_MODE_CONNECTOR_eDP)) {
                 struct radeon_connector_atom_dig *dig = radeon_connector->con_priv;
-                if (dig->dp_i2c_bus)
+                if ((dig->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT ||
+                     dig->dp_sink_type == CONNECTOR_OBJECT_ID_eDP) && dig->dp_i2c_bus)
                         radeon_connector->edid = drm_get_edid(&radeon_connector->base, &dig->dp_i2c_bus->adapter);
         }
         if (!radeon_connector->ddc_bus)
@@ -410,11 +411,12 @@ void radeon_compute_pll(struct radeon_pll *pll,
                         uint32_t *fb_div_p,
                         uint32_t *frac_fb_div_p,
                         uint32_t *ref_div_p,
-                        uint32_t *post_div_p,
-                        int flags)
+                        uint32_t *post_div_p)
 {
         uint32_t min_ref_div = pll->min_ref_div;
         uint32_t max_ref_div = pll->max_ref_div;
+        uint32_t min_post_div = pll->min_post_div;
+        uint32_t max_post_div = pll->max_post_div;
         uint32_t min_fractional_feed_div = 0;
         uint32_t max_fractional_feed_div = 0;
         uint32_t best_vco = pll->best_vco;
@@ -430,7 +432,7 @@ void radeon_compute_pll(struct radeon_pll *pll,
         DRM_DEBUG("PLL freq %llu %u %u\n", freq, pll->min_ref_div, pll->max_ref_div);
         freq = freq * 1000;
 
-        if (flags & RADEON_PLL_USE_REF_DIV)
+        if (pll->flags & RADEON_PLL_USE_REF_DIV)
                 min_ref_div = max_ref_div = pll->reference_div;
         else {
                 while (min_ref_div < max_ref_div-1) {
@@ -445,19 +447,22 @@ void radeon_compute_pll(struct radeon_pll *pll,
                 }
         }
 
-        if (flags & RADEON_PLL_USE_FRAC_FB_DIV) {
+        if (pll->flags & RADEON_PLL_USE_POST_DIV)
+                min_post_div = max_post_div = pll->post_div;
+
+        if (pll->flags & RADEON_PLL_USE_FRAC_FB_DIV) {
                 min_fractional_feed_div = pll->min_frac_feedback_div;
                 max_fractional_feed_div = pll->max_frac_feedback_div;
         }
 
-        for (post_div = pll->min_post_div; post_div <= pll->max_post_div; ++post_div) {
+        for (post_div = min_post_div; post_div <= max_post_div; ++post_div) {
                 uint32_t ref_div;
 
-                if ((flags & RADEON_PLL_NO_ODD_POST_DIV) && (post_div & 1))
+                if ((pll->flags & RADEON_PLL_NO_ODD_POST_DIV) && (post_div & 1))
                         continue;
 
                 /* legacy radeons only have a few post_divs */
-                if (flags & RADEON_PLL_LEGACY) {
+                if (pll->flags & RADEON_PLL_LEGACY) {
                         if ((post_div == 5) ||
                             (post_div == 7) ||
                             (post_div == 9) ||
@@ -504,7 +509,7 @@ void radeon_compute_pll(struct radeon_pll *pll,
                                         tmp += (uint64_t)pll->reference_freq * 1000 * frac_feedback_div;
                                         current_freq = radeon_div(tmp, ref_div * post_div);
 
-                                        if (flags & RADEON_PLL_PREFER_CLOSEST_LOWER) {
+                                        if (pll->flags & RADEON_PLL_PREFER_CLOSEST_LOWER) {
                                                 error = freq - current_freq;
                                                 error = error < 0 ? 0xffffffff : error;
                                         } else
@@ -531,12 +536,12 @@ void radeon_compute_pll(struct radeon_pll *pll,
                                                         best_freq = current_freq;
                                                         best_error = error;
                                                         best_vco_diff = vco_diff;
-                                                } else if (((flags & RADEON_PLL_PREFER_LOW_REF_DIV) && (ref_div < best_ref_div)) ||
-                                                           ((flags & RADEON_PLL_PREFER_HIGH_REF_DIV) && (ref_div > best_ref_div)) ||
-                                                           ((flags & RADEON_PLL_PREFER_LOW_FB_DIV) && (feedback_div < best_feedback_div)) ||
-                                                           ((flags & RADEON_PLL_PREFER_HIGH_FB_DIV) && (feedback_div > best_feedback_div)) ||
-                                                           ((flags & RADEON_PLL_PREFER_LOW_POST_DIV) && (post_div < best_post_div)) ||
-                                                           ((flags & RADEON_PLL_PREFER_HIGH_POST_DIV) && (post_div > best_post_div))) {
+                                                } else if (((pll->flags & RADEON_PLL_PREFER_LOW_REF_DIV) && (ref_div < best_ref_div)) ||
+                                                           ((pll->flags & RADEON_PLL_PREFER_HIGH_REF_DIV) && (ref_div > best_ref_div)) ||
+                                                           ((pll->flags & RADEON_PLL_PREFER_LOW_FB_DIV) && (feedback_div < best_feedback_div)) ||
+                                                           ((pll->flags & RADEON_PLL_PREFER_HIGH_FB_DIV) && (feedback_div > best_feedback_div)) ||
+                                                           ((pll->flags & RADEON_PLL_PREFER_LOW_POST_DIV) && (post_div < best_post_div)) ||
+                                                           ((pll->flags & RADEON_PLL_PREFER_HIGH_POST_DIV) && (post_div > best_post_div))) {
                                                         best_post_div = post_div;
                                                         best_ref_div = ref_div;
                                                         best_feedback_div = feedback_div;
@@ -572,8 +577,7 @@ void radeon_compute_pll_avivo(struct radeon_pll *pll,
                               uint32_t *fb_div_p,
                               uint32_t *frac_fb_div_p,
                               uint32_t *ref_div_p,
-                              uint32_t *post_div_p,
-                              int flags)
+                              uint32_t *post_div_p)
 {
         fixed20_12 m, n, frac_n, p, f_vco, f_pclk, best_freq;
         fixed20_12 pll_out_max, pll_out_min;
@@ -667,7 +671,6 @@ static void radeon_user_framebuffer_destroy(struct drm_framebuffer *fb)
                 radeonfb_remove(dev, fb);
 
         if (radeon_fb->obj) {
-                radeon_gem_object_unpin(radeon_fb->obj);
                 mutex_lock(&dev->struct_mutex);
                 drm_gem_object_unreference(radeon_fb->obj);
                 mutex_unlock(&dev->struct_mutex);
@@ -715,7 +718,11 @@ radeon_user_framebuffer_create(struct drm_device *dev,
         struct drm_gem_object *obj;
 
         obj = drm_gem_object_lookup(dev, file_priv, mode_cmd->handle);
-
+        if (obj ==  NULL) {
+                dev_err(&dev->pdev->dev, "No GEM object associated to handle 0x%08X, "
+                        "can't create framebuffer\n", mode_cmd->handle);
+                return NULL;
+        }
         return radeon_framebuffer_create(dev, mode_cmd, obj);
 }
 
diff --git a/drivers/gpu/drm/radeon/radeon_encoders.c b/drivers/gpu/drm/radeon/radeon_encoders.c
index 82eb551970b9..3c91724457ca 100644
--- a/drivers/gpu/drm/radeon/radeon_encoders.c
+++ b/drivers/gpu/drm/radeon/radeon_encoders.c
@@ -156,6 +156,26 @@ radeon_get_encoder_id(struct drm_device *dev, uint32_t supported_device, uint8_t
         return ret;
 }
 
+static inline bool radeon_encoder_is_digital(struct drm_encoder *encoder)
+{
+        struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
+        switch (radeon_encoder->encoder_id) {
+        case ENCODER_OBJECT_ID_INTERNAL_LVDS:
+        case ENCODER_OBJECT_ID_INTERNAL_TMDS1:
+        case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_TMDS1:
+        case ENCODER_OBJECT_ID_INTERNAL_LVTM1:
+        case ENCODER_OBJECT_ID_INTERNAL_DVO1:
+        case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
+        case ENCODER_OBJECT_ID_INTERNAL_DDI:
+        case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
+        case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+        case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
+        case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
+                return true;
+        default:
+                return false;
+        }
+}
 void
 radeon_link_encoder_connector(struct drm_device *dev)
 {
@@ -202,7 +222,7 @@ radeon_get_connector_for_encoder(struct drm_encoder *encoder)
 
         list_for_each_entry(connector, &dev->mode_config.connector_list, head) {
                 radeon_connector = to_radeon_connector(connector);
-                if (radeon_encoder->devices & radeon_connector->devices)
+                if (radeon_encoder->active_device & radeon_connector->devices)
                         return connector;
         }
         return NULL;
@@ -676,31 +696,11 @@ atombios_dig_encoder_setup(struct drm_encoder *encoder, int action)
 
         memset(&args, 0, sizeof(args));
 
-        if (ASIC_IS_DCE32(rdev)) {
-                if (dig->dig_block)
-                        index = GetIndexIntoMasterTable(COMMAND, DIG2EncoderControl);
-                else
-                        index = GetIndexIntoMasterTable(COMMAND, DIG1EncoderControl);
-                num = dig->dig_block + 1;
-        } else {
-                switch (radeon_encoder->encoder_id) {
-                case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
-                        /* XXX doesn't really matter which dig encoder we pick as long as it's
-                         * not already in use
-                         */
-                        if (dig_connector->linkb)
-                                index = GetIndexIntoMasterTable(COMMAND, DIG2EncoderControl);
-                        else
-                                index = GetIndexIntoMasterTable(COMMAND, DIG1EncoderControl);
-                        num = 1;
-                        break;
-                case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
-                        /* Only dig2 encoder can drive LVTMA */
-                        index = GetIndexIntoMasterTable(COMMAND, DIG2EncoderControl);
-                        num = 2;
-                        break;
-                }
-        }
+        if (dig->dig_encoder)
+                index = GetIndexIntoMasterTable(COMMAND, DIG2EncoderControl);
+        else
+                index = GetIndexIntoMasterTable(COMMAND, DIG1EncoderControl);
+        num = dig->dig_encoder + 1;
 
         atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev);
 
@@ -822,7 +822,7 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t
                         args.v1.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10);
         }
         if (ASIC_IS_DCE32(rdev)) {
-                if (dig->dig_block)
+                if (dig->dig_encoder == 1)
                         args.v2.acConfig.ucEncoderSel = 1;
                 if (dig_connector->linkb)
                         args.v2.acConfig.ucLinkSel = 1;
@@ -849,17 +849,16 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t
                                 args.v2.acConfig.fCoherentMode = 1;
                 }
         } else {
+
                 args.v1.ucConfig = ATOM_TRANSMITTER_CONFIG_CLKSRC_PPLL;
 
+                if (dig->dig_encoder)
+                        args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG2_ENCODER;
+                else
+                        args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG1_ENCODER;
+
                 switch (radeon_encoder->encoder_id) {
                 case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
-                        /* XXX doesn't really matter which dig encoder we pick as long as it's
-                         * not already in use
-                         */
-                        if (dig_connector->linkb)
-                                args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG2_ENCODER;
-                        else
-                                args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG1_ENCODER;
                         if (rdev->flags & RADEON_IS_IGP) {
                                 if (radeon_encoder->pixel_clock > 165000) {
                                         if (dig_connector->igp_lane_info & 0x3)
@@ -878,10 +877,6 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t
                                 }
                         }
                         break;
-                case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
-                        /* Only dig2 encoder can drive LVTMA */
-                        args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG2_ENCODER;
-                        break;
                 }
 
                 if (radeon_encoder->pixel_clock > 165000)
@@ -1046,6 +1041,7 @@ atombios_set_encoder_crtc_source(struct drm_encoder *encoder)
         union crtc_sourc_param args;
         int index = GetIndexIntoMasterTable(COMMAND, SelectCRTC_Source);
         uint8_t frev, crev;
+        struct radeon_encoder_atom_dig *dig;
 
         memset(&args, 0, sizeof(args));
 
@@ -1109,40 +1105,16 @@ atombios_set_encoder_crtc_source(struct drm_encoder *encoder)
                         case ENCODER_OBJECT_ID_INTERNAL_UNIPHY:
                         case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1:
                         case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2:
-                                if (ASIC_IS_DCE32(rdev)) {
-                                        if (radeon_crtc->crtc_id)
-                                                args.v2.ucEncoderID = ASIC_INT_DIG2_ENCODER_ID;
-                                        else
-                                                args.v2.ucEncoderID = ASIC_INT_DIG1_ENCODER_ID;
-                                } else {
-                                        struct drm_connector *connector;
-                                        struct radeon_connector *radeon_connector;
-                                        struct radeon_connector_atom_dig *dig_connector;
-
-                                        connector = radeon_get_connector_for_encoder(encoder);
-                                        if (!connector)
-                                                return;
-                                        radeon_connector = to_radeon_connector(connector);
-                                        if (!radeon_connector->con_priv)
-                                                return;
-                                        dig_connector = radeon_connector->con_priv;
-
-                                        /* XXX doesn't really matter which dig encoder we pick as long as it's
-                                         * not already in use
-                                         */
-                                        if (dig_connector->linkb)
-                                                args.v2.ucEncoderID = ASIC_INT_DIG2_ENCODER_ID;
-                                        else
-                                                args.v2.ucEncoderID = ASIC_INT_DIG1_ENCODER_ID;
-                                }
+                        case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
+                                dig = radeon_encoder->enc_priv;
+                                if (dig->dig_encoder)
+                                        args.v2.ucEncoderID = ASIC_INT_DIG2_ENCODER_ID;
+                                else
+                                        args.v2.ucEncoderID = ASIC_INT_DIG1_ENCODER_ID;
                                 break;
                         case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1:
                                 args.v2.ucEncoderID = ASIC_INT_DVO_ENCODER_ID;
                                 break;
-                        case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA:
-                                /* Only dig2 encoder can drive LVTMA */
-                                args.v2.ucEncoderID = ASIC_INT_DIG2_ENCODER_ID;
-                                break;
                         case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1:
                                 if (radeon_encoder->active_device & (ATOM_DEVICE_TV_SUPPORT))
                                         args.v2.ucEncoderID = ASIC_INT_TV_ENCODER_ID;
@@ -1202,6 +1174,47 @@ atombios_apply_encoder_quirks(struct drm_encoder *encoder,
         }
 }
 
+static int radeon_atom_pick_dig_encoder(struct drm_encoder *encoder)
+{
+        struct drm_device *dev = encoder->dev;
+        struct radeon_device *rdev = dev->dev_private;
+        struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc);
+        struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
+        struct drm_encoder *test_encoder;
+        struct radeon_encoder_atom_dig *dig;
+        uint32_t dig_enc_in_use = 0;
+        /* on DCE32 and encoder can driver any block so just crtc id */
+        if (ASIC_IS_DCE32(rdev)) {
+                return radeon_crtc->crtc_id;
+        }
+
+        /* on DCE3 - LVTMA can only be driven by DIGB */
+        list_for_each_entry(test_encoder, &dev->mode_config.encoder_list, head) {
+                struct radeon_encoder *radeon_test_encoder;
+
+                if (encoder == test_encoder)
+                        continue;
+
+                if (!radeon_encoder_is_digital(test_encoder))
+                        continue;
+
+                radeon_test_encoder = to_radeon_encoder(test_encoder);
+                dig = radeon_test_encoder->enc_priv;
+
+                if (dig->dig_encoder >= 0)
+                        dig_enc_in_use |= (1 << dig->dig_encoder);
+        }
+
+        if (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA) {
+                if (dig_enc_in_use & 0x2)
+                        DRM_ERROR("LVDS required digital encoder 2 but it was in use - stealing\n");
+                return 1;
+        }
+        if (!(dig_enc_in_use & 1))
+                return 0;
+        return 1;
+}
+
 static void
 radeon_atom_encoder_mode_set(struct drm_encoder *encoder,
                              struct drm_display_mode *mode,
@@ -1214,12 +1227,9 @@ radeon_atom_encoder_mode_set(struct drm_encoder *encoder,
 
         if (radeon_encoder->active_device &
             (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) {
-                if (radeon_encoder->enc_priv) {
-                        struct radeon_encoder_atom_dig *dig;
-
-                        dig = radeon_encoder->enc_priv;
-                        dig->dig_block = radeon_crtc->crtc_id;
-                }
+                struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv;
+                if (dig)
+                        dig->dig_encoder = radeon_atom_pick_dig_encoder(encoder);
         }
         radeon_encoder->pixel_clock = adjusted_mode->clock;
 
@@ -1379,7 +1389,13 @@ static void radeon_atom_encoder_commit(struct drm_encoder *encoder)
 static void radeon_atom_encoder_disable(struct drm_encoder *encoder)
 {
         struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
+        struct radeon_encoder_atom_dig *dig;
         radeon_atom_encoder_dpms(encoder, DRM_MODE_DPMS_OFF);
+
+        if (radeon_encoder_is_digital(encoder)) {
+                dig = radeon_encoder->enc_priv;
+                dig->dig_encoder = -1;
+        }
         radeon_encoder->active_device = 0;
 }
 
@@ -1436,6 +1452,7 @@ radeon_atombios_set_dig_info(struct radeon_encoder *radeon_encoder)
 
         /* coherent mode by default */
         dig->coherent_mode = true;
+        dig->dig_encoder = -1;
 
         return dig;
 }
diff --git a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c
index cc27485a07ad..b6d8081e1246 100644
--- a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c
+++ b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c
@@ -339,69 +339,6 @@ void radeon_crtc_dpms(struct drm_crtc *crtc, int mode)
         }
 }
 
-/* properly set crtc bpp when using atombios */
-void radeon_legacy_atom_set_surface(struct drm_crtc *crtc)
-{
-        struct drm_device *dev = crtc->dev;
-        struct radeon_device *rdev = dev->dev_private;
-        struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc);
-        int format;
-        uint32_t crtc_gen_cntl;
-        uint32_t disp_merge_cntl;
-        uint32_t crtc_pitch;
-
-        switch (crtc->fb->bits_per_pixel) {
-        case 8:
-                format = 2;
-                break;
-        case 15:      /*  555 */
-                format = 3;
-                break;
-        case 16:      /*  565 */
-                format = 4;
-                break;
-        case 24:      /*  RGB */
-                format = 5;
-                break;
-        case 32:      /* xRGB */
-                format = 6;
-                break;
-        default:
-                return;
-        }
-
-        crtc_pitch  = ((((crtc->fb->pitch / (crtc->fb->bits_per_pixel / 8)) * crtc->fb->bits_per_pixel) +
-                        ((crtc->fb->bits_per_pixel * 8) - 1)) /
-                       (crtc->fb->bits_per_pixel * 8));
-        crtc_pitch |= crtc_pitch << 16;
-
-        WREG32(RADEON_CRTC_PITCH + radeon_crtc->crtc_offset, crtc_pitch);
-
-        switch (radeon_crtc->crtc_id) {
-        case 0:
-                disp_merge_cntl = RREG32(RADEON_DISP_MERGE_CNTL);
-                disp_merge_cntl &= ~RADEON_DISP_RGB_OFFSET_EN;
-                WREG32(RADEON_DISP_MERGE_CNTL, disp_merge_cntl);
-
-                crtc_gen_cntl = RREG32(RADEON_CRTC_GEN_CNTL) & 0xfffff0ff;
-                crtc_gen_cntl |= (format << 8);
-                crtc_gen_cntl |= RADEON_CRTC_EXT_DISP_EN;
-                WREG32(RADEON_CRTC_GEN_CNTL, crtc_gen_cntl);
-                break;
-        case 1:
-                disp_merge_cntl = RREG32(RADEON_DISP2_MERGE_CNTL);
-                disp_merge_cntl &= ~RADEON_DISP2_RGB_OFFSET_EN;
-                WREG32(RADEON_DISP2_MERGE_CNTL, disp_merge_cntl);
-
-                crtc_gen_cntl = RREG32(RADEON_CRTC2_GEN_CNTL) & 0xfffff0ff;
-                crtc_gen_cntl |= (format << 8);
-                WREG32(RADEON_CRTC2_GEN_CNTL, crtc_gen_cntl);
-                WREG32(RADEON_FP_H2_SYNC_STRT_WID,   RREG32(RADEON_CRTC2_H_SYNC_STRT_WID));
-                WREG32(RADEON_FP_V2_SYNC_STRT_WID,   RREG32(RADEON_CRTC2_V_SYNC_STRT_WID));
-                break;
-        }
-}
-
 int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y,
                          struct drm_framebuffer *old_fb)
 {
@@ -755,7 +692,6 @@ static void radeon_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
         uint32_t post_divider = 0;
         uint32_t freq = 0;
         uint8_t pll_gain;
-        int pll_flags = RADEON_PLL_LEGACY;
         bool use_bios_divs = false;
         /* PLL registers */
         uint32_t pll_ref_div = 0;
@@ -789,10 +725,12 @@ static void radeon_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
         else
                 pll = &rdev->clock.p1pll;
 
+        pll->flags = RADEON_PLL_LEGACY;
+
         if (mode->clock > 200000) /* range limits??? */
-                pll_flags |= RADEON_PLL_PREFER_HIGH_FB_DIV;
+                pll->flags |= RADEON_PLL_PREFER_HIGH_FB_DIV;
         else
-                pll_flags |= RADEON_PLL_PREFER_LOW_REF_DIV;
+                pll->flags |= RADEON_PLL_PREFER_LOW_REF_DIV;
 
         list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
                 if (encoder->crtc == crtc) {
@@ -804,7 +742,7 @@ static void radeon_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
                         }
 
                         if (encoder->encoder_type != DRM_MODE_ENCODER_DAC)
-                                pll_flags |= RADEON_PLL_NO_ODD_POST_DIV;
+                                pll->flags |= RADEON_PLL_NO_ODD_POST_DIV;
                         if (encoder->encoder_type == DRM_MODE_ENCODER_LVDS) {
                                 if (!rdev->is_atom_bios) {
                                         struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder);
@@ -819,7 +757,7 @@ static void radeon_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
                                                 }
                                         }
                                 }
-                                pll_flags |= RADEON_PLL_USE_REF_DIV;
+                                pll->flags |= RADEON_PLL_USE_REF_DIV;
                         }
                 }
         }
@@ -829,8 +767,7 @@ static void radeon_set_pll(struct drm_crtc *crtc, struct drm_display_mode *mode)
         if (!use_bios_divs) {
                 radeon_compute_pll(pll, mode->clock,
                                    &freq, &feedback_div, &frac_fb_div,
-                                   &reference_div, &post_divider,
-                                   pll_flags);
+                                   &reference_div, &post_divider);
 
                 for (post_div = &post_divs[0]; post_div->divider; ++post_div) {
                         if (post_div->divider == post_divider)
diff --git a/drivers/gpu/drm/radeon/radeon_mode.h b/drivers/gpu/drm/radeon/radeon_mode.h
index 91cb041cb40d..e81b2aeb6a8f 100644
--- a/drivers/gpu/drm/radeon/radeon_mode.h
+++ b/drivers/gpu/drm/radeon/radeon_mode.h
@@ -125,16 +125,24 @@ struct radeon_tmds_pll {
 #define RADEON_PLL_PREFER_HIGH_POST_DIV (1 << 9)
 #define RADEON_PLL_USE_FRAC_FB_DIV      (1 << 10)
 #define RADEON_PLL_PREFER_CLOSEST_LOWER (1 << 11)
+#define RADEON_PLL_USE_POST_DIV         (1 << 12)
 
 struct radeon_pll {
-        uint16_t reference_freq;
-        uint16_t reference_div;
+        /* reference frequency */
+        uint32_t reference_freq;
+
+        /* fixed dividers */
+        uint32_t reference_div;
+        uint32_t post_div;
+
+        /* pll in/out limits */
         uint32_t pll_in_min;
         uint32_t pll_in_max;
         uint32_t pll_out_min;
         uint32_t pll_out_max;
-        uint16_t xclk;
+        uint32_t best_vco;
 
+        /* divider limits */
         uint32_t min_ref_div;
         uint32_t max_ref_div;
         uint32_t min_post_div;
@@ -143,7 +151,12 @@ struct radeon_pll {
         uint32_t max_feedback_div;
         uint32_t min_frac_feedback_div;
         uint32_t max_frac_feedback_div;
-        uint32_t best_vco;
+
+        /* flags for the current clock */
+        uint32_t flags;
+
+        /* pll id */
+        uint32_t id;
 };
 
 struct radeon_i2c_chan {
@@ -286,7 +299,7 @@ struct radeon_atom_ss {
 struct radeon_encoder_atom_dig {
         /* atom dig */
         bool coherent_mode;
-        int dig_block;
+        int dig_encoder; /* -1 disabled, 0 DIGA, 1 DIGB */
         /* atom lvds */
         uint32_t lvds_misc;
         uint16_t panel_pwr_delay;
@@ -417,8 +430,7 @@ extern void radeon_compute_pll(struct radeon_pll *pll,
                                uint32_t *fb_div_p,
                                uint32_t *frac_fb_div_p,
                                uint32_t *ref_div_p,
-                               uint32_t *post_div_p,
-                               int flags);
+                               uint32_t *post_div_p);
 
 extern void radeon_compute_pll_avivo(struct radeon_pll *pll,
                                      uint64_t freq,
@@ -426,8 +438,7 @@ extern void radeon_compute_pll_avivo(struct radeon_pll *pll,
                                      uint32_t *fb_div_p,
                                      uint32_t *frac_fb_div_p,
                                      uint32_t *ref_div_p,
-                                     uint32_t *post_div_p,
-                                     int flags);
+                                     uint32_t *post_div_p);
 
 extern void radeon_setup_encoder_clones(struct drm_device *dev);
 
@@ -453,7 +464,6 @@ extern void atombios_crtc_dpms(struct drm_crtc *crtc, int mode);
 
 extern int radeon_crtc_set_base(struct drm_crtc *crtc, int x, int y,
                                  struct drm_framebuffer *old_fb);
-extern void radeon_legacy_atom_set_surface(struct drm_crtc *crtc);
 
 extern int radeon_crtc_cursor_set(struct drm_crtc *crtc,
                                   struct drm_file *file_priv,
diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c
index 4e636de877b2..d72a71bff218 100644
--- a/drivers/gpu/drm/radeon/radeon_object.c
+++ b/drivers/gpu/drm/radeon/radeon_object.c
@@ -220,7 +220,8 @@ int radeon_bo_unpin(struct radeon_bo *bo)
 
 int radeon_bo_evict_vram(struct radeon_device *rdev)
 {
-        if (rdev->flags & RADEON_IS_IGP) {
+        /* late 2.6.33 fix IGP hibernate - we need pm ops to do this correct */
+        if (0 && (rdev->flags & RADEON_IS_IGP)) {
                 if (rdev->mc.igp_sideport_enabled == false)
                         /* Useless to evict on IGP chips */
                         return 0;
diff --git a/drivers/gpu/drm/radeon/reg_srcs/r200 b/drivers/gpu/drm/radeon/reg_srcs/r200
index 6021c8849a16..c29ac434ac9c 100644
--- a/drivers/gpu/drm/radeon/reg_srcs/r200
+++ b/drivers/gpu/drm/radeon/reg_srcs/r200
@@ -91,6 +91,8 @@ r200 0x3294
 0x22b8 SE_TCL_TEX_CYL_WRAP_CTL
 0x22c0 SE_TCL_UCP_VERT_BLEND_CNTL
 0x22c4 SE_TCL_POINT_SPRITE_CNTL
+0x22d0 SE_PVS_CNTL
+0x22d4 SE_PVS_CONST_CNTL
 0x2648 RE_POINTSIZE
 0x26c0 RE_TOP_LEFT
 0x26c4 RE_MISC
diff --git a/drivers/gpu/drm/radeon/rv770.c b/drivers/gpu/drm/radeon/rv770.c
index 59c71245fb91..afd9e8213c29 100644
--- a/drivers/gpu/drm/radeon/rv770.c
+++ b/drivers/gpu/drm/radeon/rv770.c
@@ -779,7 +779,6 @@ int rv770_mc_init(struct radeon_device *rdev)
         fixed20_12 a;
         u32 tmp;
         int chansize, numchan;
-        int r;
 
         /* Get VRAM informations */
         rdev->mc.vram_is_ddr = true;
@@ -822,9 +821,6 @@ int rv770_mc_init(struct radeon_device *rdev)
                 rdev->mc.real_vram_size = rdev->mc.aper_size;
 
         if (rdev->flags & RADEON_IS_AGP) {
-                r = radeon_agp_init(rdev);
-                if (r)
-                        return r;
                 /* gtt_size is setup by radeon_agp_init */
                 rdev->mc.gtt_location = rdev->mc.agp_base;
                 tmp = 0xFFFFFFFFUL - rdev->mc.agp_base - rdev->mc.gtt_size;
@@ -891,26 +887,19 @@ static int rv770_startup(struct radeon_device *rdev)
                         return r;
         }
         rv770_gpu_init(rdev);
-
-        if (!rdev->r600_blit.shader_obj) {
-                r = r600_blit_init(rdev);
+        /* pin copy shader into vram */
+        if (rdev->r600_blit.shader_obj) {
+                r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
+                if (unlikely(r != 0))
+                        return r;
+                r = radeon_bo_pin(rdev->r600_blit.shader_obj, RADEON_GEM_DOMAIN_VRAM,
+                                &rdev->r600_blit.shader_gpu_addr);
+                radeon_bo_unreserve(rdev->r600_blit.shader_obj);
                 if (r) {
-                        DRM_ERROR("radeon: failed blitter (%d).\n", r);
+                        DRM_ERROR("failed to pin blit object %d\n", r);
                         return r;
                 }
         }
-
-        r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
-        if (unlikely(r != 0))
-                return r;
-        r = radeon_bo_pin(rdev->r600_blit.shader_obj, RADEON_GEM_DOMAIN_VRAM,
-                        &rdev->r600_blit.shader_gpu_addr);
-        radeon_bo_unreserve(rdev->r600_blit.shader_obj);
-        if (r) {
-                DRM_ERROR("failed to pin blit object %d\n", r);
-                return r;
-        }
-
         /* Enable IRQ */
         r = r600_irq_init(rdev);
         if (r) {
@@ -972,13 +961,16 @@ int rv770_suspend(struct radeon_device *rdev)
         /* FIXME: we should wait for ring to be empty */
         r700_cp_stop(rdev);
         rdev->cp.ready = false;
+        r600_irq_suspend(rdev);
         r600_wb_disable(rdev);
         rv770_pcie_gart_disable(rdev);
         /* unpin shaders bo */
-        r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
-        if (likely(r == 0)) {
-                radeon_bo_unpin(rdev->r600_blit.shader_obj);
-                radeon_bo_unreserve(rdev->r600_blit.shader_obj);
+        if (rdev->r600_blit.shader_obj) {
+                r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false);
+                if (likely(r == 0)) {
+                        radeon_bo_unpin(rdev->r600_blit.shader_obj);
+                        radeon_bo_unreserve(rdev->r600_blit.shader_obj);
+                }
         }
         return 0;
 }
@@ -1037,6 +1029,11 @@ int rv770_init(struct radeon_device *rdev)
         r = radeon_fence_driver_init(rdev);
         if (r)
                 return r;
+        if (rdev->flags & RADEON_IS_AGP) {
+                r = radeon_agp_init(rdev);
+                if (r)
+                        radeon_agp_disable(rdev);
+        }
         r = rv770_mc_init(rdev);
         if (r)
                 return r;
@@ -1058,6 +1055,12 @@ int rv770_init(struct radeon_device *rdev)
         r = r600_pcie_gart_init(rdev);
         if (r)
                 return r;
+        r = r600_blit_init(rdev);
+        if (r) {
+                r600_blit_fini(rdev);
+                rdev->asic->copy = NULL;
+                dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r);
+        }
 
         rdev->accel_working = true;
         r = rv770_startup(rdev);
@@ -1071,13 +1074,14 @@ int rv770_init(struct radeon_device *rdev)
         if (rdev->accel_working) {
                 r = radeon_ib_pool_init(rdev);
                 if (r) {
-                        DRM_ERROR("radeon: failed initializing IB pool (%d).\n", r);
-                        rdev->accel_working = false;
-                }
-                r = r600_ib_test(rdev);
-                if (r) {
-                        DRM_ERROR("radeon: failed testing IB (%d).\n", r);
+                        dev_err(rdev->dev, "IB initialization failed (%d).\n", r);
                         rdev->accel_working = false;
+                } else {
+                        r = r600_ib_test(rdev);
+                        if (r) {
+                                dev_err(rdev->dev, "IB test failed (%d).\n", r);
+                                rdev->accel_working = false;
+                        }
                 }
         }
         return 0;
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index 2920f9a279e1..1a3e909b7bba 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -426,7 +426,8 @@ moved:
                     bdev->man[bo->mem.mem_type].gpu_offset;
                 bo->cur_placement = bo->mem.placement;
                 spin_unlock(&bo->lock);
-        }
+        } else
+                bo->offset = 0;
 
         return 0;
 
@@ -523,52 +524,44 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, bool remove_all)
 static int ttm_bo_delayed_delete(struct ttm_bo_device *bdev, bool remove_all)
 {
         struct ttm_bo_global *glob = bdev->glob;
-        struct ttm_buffer_object *entry, *nentry;
-        struct list_head *list, *next;
-        int ret;
+        struct ttm_buffer_object *entry = NULL;
+        int ret = 0;
 
         spin_lock(&glob->lru_lock);
-        list_for_each_safe(list, next, &bdev->ddestroy) {
-                entry = list_entry(list, struct ttm_buffer_object, ddestroy);
-                nentry = NULL;
+        if (list_empty(&bdev->ddestroy))
+                goto out_unlock;
 
-                /*
-                 * Protect the next list entry from destruction while we
-                 * unlock the lru_lock.
-                 */
+        entry = list_first_entry(&bdev->ddestroy,
+                struct ttm_buffer_object, ddestroy);
+        kref_get(&entry->list_kref);
 
-                if (next != &bdev->ddestroy) {
-                        nentry = list_entry(next, struct ttm_buffer_object,
-                                            ddestroy);
+        for (;;) {
+                struct ttm_buffer_object *nentry = NULL;
+
+                if (entry->ddestroy.next != &bdev->ddestroy) {
+                        nentry = list_first_entry(&entry->ddestroy,
+                                struct ttm_buffer_object, ddestroy);
                         kref_get(&nentry->list_kref);
                 }
-                kref_get(&entry->list_kref);
 
                 spin_unlock(&glob->lru_lock);
                 ret = ttm_bo_cleanup_refs(entry, remove_all);
                 kref_put(&entry->list_kref, ttm_bo_release_list);
+                entry = nentry;
+
+                if (ret || !entry)
+                        goto out;
 
                 spin_lock(&glob->lru_lock);
-                if (nentry) {
-                        bool next_onlist = !list_empty(next);
-                        spin_unlock(&glob->lru_lock);
-                        kref_put(&nentry->list_kref, ttm_bo_release_list);
-                        spin_lock(&glob->lru_lock);
-                        /*
-                         * Someone might have raced us and removed the
-                         * next entry from the list. We don't bother restarting
-                         * list traversal.
-                         */
-
-                        if (!next_onlist)
-                                break;
-                }
-                if (ret)
+                if (list_empty(&entry->ddestroy))
                         break;
         }
-        ret = !list_empty(&bdev->ddestroy);
-        spin_unlock(&glob->lru_lock);
 
+out_unlock:
+        spin_unlock(&glob->lru_lock);
+out:
+        if (entry)
+                kref_put(&entry->list_kref, ttm_bo_release_list);
         return ret;
 }
 
@@ -950,6 +943,14 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo,
                 ttm_flag_masked(&cur_flags, placement->busy_placement[i],
                                 ~TTM_PL_MASK_MEMTYPE);
 
+
+                if (mem_type == TTM_PL_SYSTEM) {
+                        mem->mem_type = mem_type;
+                        mem->placement = cur_flags;
+                        mem->mm_node = NULL;
+                        return 0;
+                }
+
                 ret = ttm_bo_mem_force_space(bo, mem_type, placement, mem,
                                                 interruptible, no_wait);
                 if (ret == 0 && mem->mm_node) {
@@ -1844,6 +1845,9 @@ static int ttm_bo_swapout(struct ttm_mem_shrink *shrink)
          * anyone tries to access a ttm page.
          */
 
+        if (bo->bdev->driver->swap_notify)
+                bo->bdev->driver->swap_notify(bo);
+
         ret = ttm_tt_swapout(bo->ttm, bo->persistant_swap_storage);
 out:
 
@@ -1864,3 +1868,4 @@ void ttm_bo_swapout_all(struct ttm_bo_device *bdev)
         while (ttm_bo_swapout(&bdev->glob->shrink) == 0)
                 ;
 }
+EXPORT_SYMBOL(ttm_bo_swapout_all);
diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c
index 2ecf7d0c64f6..5ca37a58a98c 100644
--- a/drivers/gpu/drm/ttm/ttm_bo_util.c
+++ b/drivers/gpu/drm/ttm/ttm_bo_util.c
@@ -53,7 +53,6 @@ int ttm_bo_move_ttm(struct ttm_buffer_object *bo,
 {
         struct ttm_tt *ttm = bo->ttm;
         struct ttm_mem_reg *old_mem = &bo->mem;
-        uint32_t save_flags = old_mem->placement;
         int ret;
 
         if (old_mem->mem_type != TTM_PL_SYSTEM) {
@@ -62,7 +61,6 @@ int ttm_bo_move_ttm(struct ttm_buffer_object *bo,
                 ttm_flag_masked(&old_mem->placement, TTM_PL_FLAG_SYSTEM,
                                 TTM_PL_MASK_MEM);
                 old_mem->mem_type = TTM_PL_SYSTEM;
-                save_flags = old_mem->placement;
         }
 
         ret = ttm_tt_set_placement_caching(ttm, new_mem->placement);
@@ -77,7 +75,7 @@ int ttm_bo_move_ttm(struct ttm_buffer_object *bo,
 
         *old_mem = *new_mem;
         new_mem->mm_node = NULL;
-        ttm_flag_masked(&save_flags, new_mem->placement, TTM_PL_MASK_MEMTYPE);
+
         return 0;
 }
 EXPORT_SYMBOL(ttm_bo_move_ttm);
@@ -219,7 +217,6 @@ int ttm_bo_move_memcpy(struct ttm_buffer_object *bo,
         void *old_iomap;
         void *new_iomap;
         int ret;
-        uint32_t save_flags = old_mem->placement;
         unsigned long i;
         unsigned long page;
         unsigned long add = 0;
@@ -270,7 +267,6 @@ out2:
 
         *old_mem = *new_mem;
         new_mem->mm_node = NULL;
-        ttm_flag_masked(&save_flags, new_mem->placement, TTM_PL_MASK_MEMTYPE);
 
         if ((man->flags & TTM_MEMTYPE_FLAG_FIXED) && (ttm != NULL)) {
                 ttm_tt_unbind(ttm);
@@ -537,7 +533,6 @@ int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo,
         struct ttm_mem_type_manager *man = &bdev->man[new_mem->mem_type];
         struct ttm_mem_reg *old_mem = &bo->mem;
         int ret;
-        uint32_t save_flags = old_mem->placement;
         struct ttm_buffer_object *ghost_obj;
         void *tmp_obj = NULL;
 
@@ -598,7 +593,7 @@ int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo,
 
         *old_mem = *new_mem;
         new_mem->mm_node = NULL;
-        ttm_flag_masked(&save_flags, new_mem->placement, TTM_PL_MASK_MEMTYPE);
+
         return 0;
 }
 EXPORT_SYMBOL(ttm_bo_move_accel_cleanup);
diff --git a/drivers/gpu/drm/ttm/ttm_lock.c b/drivers/gpu/drm/ttm/ttm_lock.c
index f619ebcaa4ec..3d172ef04ee1 100644
--- a/drivers/gpu/drm/ttm/ttm_lock.c
+++ b/drivers/gpu/drm/ttm/ttm_lock.c
@@ -288,6 +288,7 @@ void ttm_suspend_unlock(struct ttm_lock *lock)
         wake_up_all(&lock->queue);
         spin_unlock(&lock->lock);
 }
+EXPORT_SYMBOL(ttm_suspend_unlock);
 
 static bool __ttm_suspend_lock(struct ttm_lock *lock)
 {
@@ -309,3 +310,4 @@ void ttm_suspend_lock(struct ttm_lock *lock)
 {
         wait_event(lock->queue, __ttm_suspend_lock(lock));
 }
+EXPORT_SYMBOL(ttm_suspend_lock);
diff --git a/drivers/gpu/drm/ttm/ttm_object.c b/drivers/gpu/drm/ttm/ttm_object.c
index 1099abac824b..75e9d6f86ba4 100644
--- a/drivers/gpu/drm/ttm/ttm_object.c
+++ b/drivers/gpu/drm/ttm/ttm_object.c
@@ -109,8 +109,8 @@ struct ttm_ref_object {
         struct drm_hash_item hash;
         struct list_head head;
         struct kref kref;
-        struct ttm_base_object *obj;
         enum ttm_ref_type ref_type;
+        struct ttm_base_object *obj;
         struct ttm_object_file *tfile;
 };
 
diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c
index 9c2b1cc5dba5..e2123af7775a 100644
--- a/drivers/gpu/drm/ttm/ttm_tt.c
+++ b/drivers/gpu/drm/ttm/ttm_tt.c
@@ -198,17 +198,26 @@ EXPORT_SYMBOL(ttm_tt_populate);
 static inline int ttm_tt_set_page_caching(struct page *p,
                                           enum ttm_caching_state c_state)
 {
+        int ret = 0;
+
         if (PageHighMem(p))
                 return 0;
 
-        switch (c_state) {
-        case tt_cached:
-                return set_pages_wb(p, 1);
-        case tt_wc:
-            return set_memory_wc((unsigned long) page_address(p), 1);
-        default:
-                return set_pages_uc(p, 1);
+        if (get_page_memtype(p) != -1) {
+                /* p isn't in the default caching state, set it to
+                 * writeback first to free its current memtype. */
+
+                ret = set_pages_wb(p, 1);
+                if (ret)
+                        return ret;
         }
+
+        if (c_state == tt_wc)
+                ret = set_memory_wc((unsigned long) page_address(p), 1);
+        else if (c_state == tt_uncached)
+                ret = set_pages_uc(p, 1);
+
+        return ret;
 }
 #else /* CONFIG_X86 */
 static inline int ttm_tt_set_page_caching(struct page *p,
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_buffer.c b/drivers/gpu/drm/vmwgfx/vmwgfx_buffer.c
index d6f2d2b882e9..825ebe3d89d5 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_buffer.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_buffer.c
@@ -48,6 +48,15 @@ struct ttm_placement vmw_vram_placement = {
         .busy_placement = &vram_placement_flags
 };
 
+struct ttm_placement vmw_vram_sys_placement = {
+        .fpfn = 0,
+        .lpfn = 0,
+        .num_placement = 1,
+        .placement = &vram_placement_flags,
+        .num_busy_placement = 1,
+        .busy_placement = &sys_placement_flags
+};
+
 struct ttm_placement vmw_vram_ne_placement = {
         .fpfn = 0,
         .lpfn = 0,
@@ -172,6 +181,18 @@ static int vmw_verify_access(struct ttm_buffer_object *bo, struct file *filp)
         return 0;
 }
 
+static void vmw_move_notify(struct ttm_buffer_object *bo,
+                     struct ttm_mem_reg *new_mem)
+{
+        if (new_mem->mem_type != TTM_PL_SYSTEM)
+                vmw_dmabuf_gmr_unbind(bo);
+}
+
+static void vmw_swap_notify(struct ttm_buffer_object *bo)
+{
+        vmw_dmabuf_gmr_unbind(bo);
+}
+
 /**
  * FIXME: We're using the old vmware polling method to sync.
  * Do this with fences instead.
@@ -225,5 +246,7 @@ struct ttm_bo_driver vmw_bo_driver = {
         .sync_obj_wait = vmw_sync_obj_wait,
         .sync_obj_flush = vmw_sync_obj_flush,
         .sync_obj_unref = vmw_sync_obj_unref,
-        .sync_obj_ref = vmw_sync_obj_ref
+        .sync_obj_ref = vmw_sync_obj_ref,
+        .move_notify = vmw_move_notify,
+        .swap_notify = vmw_swap_notify
 };
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
index 1db1ef30be2b..a6e8f687fa64 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
@@ -147,6 +147,8 @@ static char *vmw_devname = "vmwgfx";
 
 static int vmw_probe(struct pci_dev *, const struct pci_device_id *);
 static void vmw_master_init(struct vmw_master *);
+static int vmwgfx_pm_notifier(struct notifier_block *nb, unsigned long val,
+                              void *ptr);
 
 static void vmw_print_capabilities(uint32_t capabilities)
 {
@@ -207,6 +209,7 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset)
 {
         struct vmw_private *dev_priv;
         int ret;
+        uint32_t svga_id;
 
         dev_priv = kzalloc(sizeof(*dev_priv), GFP_KERNEL);
         if (unlikely(dev_priv == NULL)) {
@@ -217,6 +220,7 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset)
 
         dev_priv->dev = dev;
         dev_priv->vmw_chipset = chipset;
+        dev_priv->last_read_sequence = (uint32_t) -100;
         mutex_init(&dev_priv->hw_mutex);
         mutex_init(&dev_priv->cmdbuf_mutex);
         rwlock_init(&dev_priv->resource_lock);
@@ -236,6 +240,16 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset)
         dev_priv->mmio_start = pci_resource_start(dev->pdev, 2);
 
         mutex_lock(&dev_priv->hw_mutex);
+
+        vmw_write(dev_priv, SVGA_REG_ID, SVGA_ID_2);
+        svga_id = vmw_read(dev_priv, SVGA_REG_ID);
+        if (svga_id != SVGA_ID_2) {
+                ret = -ENOSYS;
+                DRM_ERROR("Unsuported SVGA ID 0x%x\n", svga_id);
+                mutex_unlock(&dev_priv->hw_mutex);
+                goto out_err0;
+        }
+
         dev_priv->capabilities = vmw_read(dev_priv, SVGA_REG_CAPABILITIES);
 
         if (dev_priv->capabilities & SVGA_CAP_GMR) {
@@ -351,6 +365,11 @@ static int vmw_driver_load(struct drm_device *dev, unsigned long chipset)
                 vmw_fb_init(dev_priv);
         }
 
+        dev_priv->pm_nb.notifier_call = vmwgfx_pm_notifier;
+        register_pm_notifier(&dev_priv->pm_nb);
+
+        DRM_INFO("%s", vmw_fifo_have_3d(dev_priv) ? "Have 3D\n" : "No 3D\n");
+
         return 0;
 
 out_no_device:
@@ -385,6 +404,8 @@ static int vmw_driver_unload(struct drm_device *dev)
 
         DRM_INFO(VMWGFX_DRIVER_NAME " unload.\n");
 
+        unregister_pm_notifier(&dev_priv->pm_nb);
+
         if (!dev_priv->stealth) {
                 vmw_fb_close(dev_priv);
                 vmw_kms_close(dev_priv);
@@ -650,6 +671,57 @@ static void vmw_remove(struct pci_dev *pdev)
         drm_put_dev(dev);
 }
 
+static int vmwgfx_pm_notifier(struct notifier_block *nb, unsigned long val,
+                              void *ptr)
+{
+        struct vmw_private *dev_priv =
+                container_of(nb, struct vmw_private, pm_nb);
+        struct vmw_master *vmaster = dev_priv->active_master;
+
+        switch (val) {
+        case PM_HIBERNATION_PREPARE:
+        case PM_SUSPEND_PREPARE:
+                ttm_suspend_lock(&vmaster->lock);
+
+                /**
+                 * This empties VRAM and unbinds all GMR bindings.
+                 * Buffer contents is moved to swappable memory.
+                 */
+                ttm_bo_swapout_all(&dev_priv->bdev);
+                break;
+        case PM_POST_HIBERNATION:
+        case PM_POST_SUSPEND:
+                ttm_suspend_unlock(&vmaster->lock);
+                break;
+        case PM_RESTORE_PREPARE:
+                break;
+        case PM_POST_RESTORE:
+                break;
+        default:
+                break;
+        }
+        return 0;
+}
+
+/**
+ * These might not be needed with the virtual SVGA device.
+ */
+
+int vmw_pci_suspend(struct pci_dev *pdev, pm_message_t state)
+{
+        pci_save_state(pdev);
+        pci_disable_device(pdev);
+        pci_set_power_state(pdev, PCI_D3hot);
+        return 0;
+}
+
+int vmw_pci_resume(struct pci_dev *pdev)
+{
+        pci_set_power_state(pdev, PCI_D0);
+        pci_restore_state(pdev);
+        return pci_enable_device(pdev);
+}
+
 static struct drm_driver driver = {
         .driver_features = DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED |
         DRIVER_MODESET,
@@ -689,7 +761,9 @@ static struct drm_driver driver = {
                        .name = VMWGFX_DRIVER_NAME,
                        .id_table = vmw_pci_id_list,
                        .probe = vmw_probe,
-                       .remove = vmw_remove
+                       .remove = vmw_remove,
+                       .suspend = vmw_pci_suspend,
+                       .resume = vmw_pci_resume
                        },
         .name = VMWGFX_DRIVER_NAME,
         .desc = VMWGFX_DRIVER_DESC,
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
index e61bd85b6975..135be9688c90 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
@@ -32,6 +32,7 @@
 #include "drmP.h"
 #include "vmwgfx_drm.h"
 #include "drm_hashtab.h"
+#include "linux/suspend.h"
 #include "ttm/ttm_bo_driver.h"
 #include "ttm/ttm_object.h"
 #include "ttm/ttm_lock.h"
@@ -95,6 +96,8 @@ struct vmw_surface {
         struct drm_vmw_size *sizes;
         uint32_t num_sizes;
 
+        bool scanout;
+
         /* TODO so far just a extra pointer */
         struct vmw_cursor_snooper snooper;
 };
@@ -258,6 +261,7 @@ struct vmw_private {
 
         struct vmw_master *active_master;
         struct vmw_master fbdev_master;
+        struct notifier_block pm_nb;
 };
 
 static inline struct vmw_private *vmw_priv(struct drm_device *dev)
@@ -353,6 +357,7 @@ extern int vmw_dmabuf_to_start_of_vram(struct vmw_private *vmw_priv,
                                        struct vmw_dma_buffer *bo);
 extern int vmw_dmabuf_from_vram(struct vmw_private *vmw_priv,
                                 struct vmw_dma_buffer *bo);
+extern void vmw_dmabuf_gmr_unbind(struct ttm_buffer_object *bo);
 extern int vmw_stream_claim_ioctl(struct drm_device *dev, void *data,
                                   struct drm_file *file_priv);
 extern int vmw_stream_unref_ioctl(struct drm_device *dev, void *data,
@@ -386,6 +391,7 @@ extern int vmw_fifo_send_fence(struct vmw_private *dev_priv,
                                uint32_t *sequence);
 extern void vmw_fifo_ping_host(struct vmw_private *dev_priv, uint32_t reason);
 extern int vmw_fifo_mmap(struct file *filp, struct vm_area_struct *vma);
+extern bool vmw_fifo_have_3d(struct vmw_private *dev_priv);
 
 /**
  * TTM glue - vmwgfx_ttm_glue.c
@@ -401,6 +407,7 @@ extern int vmw_mmap(struct file *filp, struct vm_area_struct *vma);
 
 extern struct ttm_placement vmw_vram_placement;
 extern struct ttm_placement vmw_vram_ne_placement;
+extern struct ttm_placement vmw_vram_sys_placement;
 extern struct ttm_placement vmw_sys_placement;
 extern struct ttm_bo_driver vmw_bo_driver;
 extern int vmw_dma_quiescent(struct drm_device *dev);
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
index 2e92da567403..d69caf92ffe7 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
@@ -490,10 +490,29 @@ static int vmw_validate_single_buffer(struct vmw_private *dev_priv,
         if (vmw_dmabuf_gmr(bo) != SVGA_GMR_NULL)
                 return 0;
 
+        /**
+         * Put BO in VRAM, only if there is space.
+         */
+
+        ret = ttm_bo_validate(bo, &vmw_vram_sys_placement, true, false);
+        if (unlikely(ret == -ERESTARTSYS))
+                return ret;
+
+        /**
+         * Otherwise, set it up as GMR.
+         */
+
+        if (vmw_dmabuf_gmr(bo) != SVGA_GMR_NULL)
+                return 0;
+
         ret = vmw_gmr_bind(dev_priv, bo);
         if (likely(ret == 0 || ret == -ERESTARTSYS))
                 return ret;
 
+        /**
+         * If that failed, try VRAM again, this time evicting
+         * previous contents.
+         */
 
         ret = ttm_bo_validate(bo, &vmw_vram_placement, true, false);
         return ret;
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c
index 641dde76ada1..4f4f6432be8b 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c
@@ -649,14 +649,6 @@ int vmw_dmabuf_to_start_of_vram(struct vmw_private *vmw_priv,
         if (unlikely(ret != 0))
                 goto err_unlock;
 
-        if (vmw_bo->gmr_bound) {
-                vmw_gmr_unbind(vmw_priv, vmw_bo->gmr_id);
-                spin_lock(&bo->glob->lru_lock);
-                ida_remove(&vmw_priv->gmr_ida, vmw_bo->gmr_id);
-                spin_unlock(&bo->glob->lru_lock);
-                vmw_bo->gmr_bound = NULL;
-        }
-
         ret = ttm_bo_validate(bo, &ne_placement, false, false);
         ttm_bo_unreserve(bo);
 err_unlock:
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c
index 01feb48af333..4157547cc6e4 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fifo.c
@@ -29,6 +29,25 @@
 #include "drmP.h"
 #include "ttm/ttm_placement.h"
 
+bool vmw_fifo_have_3d(struct vmw_private *dev_priv)
+{
+        __le32 __iomem *fifo_mem = dev_priv->mmio_virt;
+        uint32_t fifo_min, hwversion;
+
+        fifo_min = ioread32(fifo_mem  + SVGA_FIFO_MIN);
+        if (fifo_min <= SVGA_FIFO_3D_HWVERSION * sizeof(unsigned int))
+                return false;
+
+        hwversion = ioread32(fifo_mem + SVGA_FIFO_3D_HWVERSION);
+        if (hwversion == 0)
+                return false;
+
+        if (hwversion < SVGA3D_HWVERSION_WS65_B1)
+                return false;
+
+        return true;
+}
+
 int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
 {
         __le32 __iomem *fifo_mem = dev_priv->mmio_virt;
@@ -98,8 +117,7 @@ int vmw_fifo_init(struct vmw_private *dev_priv, struct vmw_fifo_state *fifo)
                  (unsigned int) min,
                  (unsigned int) fifo->capabilities);
 
-        dev_priv->fence_seq = (uint32_t) -100;
-        dev_priv->last_read_sequence = (uint32_t) -100;
+        dev_priv->fence_seq = dev_priv->last_read_sequence;
         iowrite32(dev_priv->last_read_sequence, fifo_mem + SVGA_FIFO_FENCE);
 
         return vmw_fifo_send_fence(dev_priv, &dummy);
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
index 5fa6a4ed238a..778851f9f1d6 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_ioctl.c
@@ -43,7 +43,7 @@ int vmw_getparam_ioctl(struct drm_device *dev, void *data,
                 param->value = vmw_overlay_num_free_overlays(dev_priv);
                 break;
         case DRM_VMW_PARAM_3D:
-                param->value = dev_priv->capabilities & SVGA_CAP_3D ? 1 : 0;
+                param->value = vmw_fifo_have_3d(dev_priv) ? 1 : 0;
                 break;
         case DRM_VMW_PARAM_FIFO_OFFSET:
                 param->value = dev_priv->mmio_start;
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
index b1af76e371c3..eeba6d1d06e4 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
@@ -553,9 +553,7 @@ int vmw_framebuffer_dmabuf_dirty(struct drm_framebuffer *framebuffer,
         } *cmd;
         int i, increment = 1;
 
-        if (!num_clips ||
-            !(dev_priv->fifo.capabilities &
-              SVGA_FIFO_CAP_SCREEN_OBJECT)) {
+        if (!num_clips) {
                 num_clips = 1;
                 clips = &norect;
                 norect.x1 = norect.y1 = 0;
@@ -574,10 +572,10 @@ int vmw_framebuffer_dmabuf_dirty(struct drm_framebuffer *framebuffer,
 
         for (i = 0; i < num_clips; i++, clips += increment) {
                 cmd[i].header = cpu_to_le32(SVGA_CMD_UPDATE);
-                cmd[i].body.x = cpu_to_le32(clips[i].x1);
-                cmd[i].body.y = cpu_to_le32(clips[i].y1);
-                cmd[i].body.width = cpu_to_le32(clips[i].x2 - clips[i].x1);
-                cmd[i].body.height = cpu_to_le32(clips[i].y2 - clips[i].y1);
+                cmd[i].body.x = cpu_to_le32(clips->x1);
+                cmd[i].body.y = cpu_to_le32(clips->y1);
+                cmd[i].body.width = cpu_to_le32(clips->x2 - clips->x1);
+                cmd[i].body.height = cpu_to_le32(clips->y2 - clips->y1);
         }
 
         vmw_fifo_commit(dev_priv, sizeof(*cmd) * num_clips);
@@ -709,6 +707,9 @@ static struct drm_framebuffer *vmw_kms_fb_create(struct drm_device *dev,
         if (ret)
                 goto try_dmabuf;
 
+        if (!surface->scanout)
+                goto err_not_scanout;
+
         ret = vmw_kms_new_framebuffer_surface(dev_priv, surface, &vfb,
                                               mode_cmd->width, mode_cmd->height);
 
@@ -742,6 +743,13 @@ try_dmabuf:
         }
 
         return &vfb->base;
+
+err_not_scanout:
+        DRM_ERROR("surface not marked as scanout\n");
+        /* vmw_user_surface_lookup takes one ref */
+        vmw_surface_unreference(&surface);
+
+        return NULL;
 }
 
 static int vmw_kms_fb_changed(struct drm_device *dev)
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c b/drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c
index bb6e6a096d25..5b6eabeb7f51 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c
@@ -104,7 +104,6 @@ static int vmw_dmabuf_pin_in_vram(struct vmw_private *dev_priv,
                                   bool pin, bool interruptible)
 {
         struct ttm_buffer_object *bo = &buf->base;
-        struct ttm_bo_global *glob = bo->glob;
         struct ttm_placement *overlay_placement = &vmw_vram_placement;
         int ret;
 
@@ -116,14 +115,6 @@ static int vmw_dmabuf_pin_in_vram(struct vmw_private *dev_priv,
         if (unlikely(ret != 0))
                 goto err;
 
-        if (buf->gmr_bound) {
-                vmw_gmr_unbind(dev_priv, buf->gmr_id);
-                spin_lock(&glob->lru_lock);
-                ida_remove(&dev_priv->gmr_ida, buf->gmr_id);
-                spin_unlock(&glob->lru_lock);
-                buf->gmr_bound = NULL;
-        }
-
         if (pin)
                 overlay_placement = &vmw_vram_ne_placement;
 
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
index c012d5927f65..c7efbd47ab84 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
@@ -35,6 +35,11 @@
 #define VMW_RES_SURFACE ttm_driver_type1
 #define VMW_RES_STREAM ttm_driver_type2
 
+/* XXX: This isn't a real hardware flag, but just a hack for kernel to
+ * know about primary surfaces. Find a better way to accomplish this.
+ */
+#define SVGA3D_SURFACE_HINT_SCANOUT (1 << 9)
+
 struct vmw_user_context {
         struct ttm_base_object base;
         struct vmw_resource res;
@@ -599,6 +604,36 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
         if (unlikely(ret != 0))
                 goto out_err1;
 
+        if (srf->flags & SVGA3D_SURFACE_HINT_SCANOUT) {
+                /* we should not send this flag down to hardware since
+                 * its not a official one
+                 */
+                srf->flags &= ~SVGA3D_SURFACE_HINT_SCANOUT;
+                srf->scanout = true;
+        } else {
+                srf->scanout = false;
+        }
+
+        if (srf->scanout &&
+            srf->num_sizes == 1 &&
+            srf->sizes[0].width == 64 &&
+            srf->sizes[0].height == 64 &&
+            srf->format == SVGA3D_A8R8G8B8) {
+
+                srf->snooper.image = kmalloc(64 * 64 * 4, GFP_KERNEL);
+                /* clear the image */
+                if (srf->snooper.image) {
+                        memset(srf->snooper.image, 0x00, 64 * 64 * 4);
+                } else {
+                        DRM_ERROR("Failed to allocate cursor_image\n");
+                        ret = -ENOMEM;
+                        goto out_err1;
+                }
+        } else {
+                srf->snooper.image = NULL;
+        }
+        srf->snooper.crtc = NULL;
+
         user_srf->base.shareable = false;
         user_srf->base.tfile = NULL;
 
@@ -622,24 +657,6 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
                 return ret;
         }
 
-        if (srf->flags & (1 << 9) &&
-            srf->num_sizes == 1 &&
-            srf->sizes[0].width == 64 &&
-            srf->sizes[0].height == 64 &&
-            srf->format == SVGA3D_A8R8G8B8) {
-
-                srf->snooper.image = kmalloc(64 * 64 * 4, GFP_KERNEL);
-                /* clear the image */
-                if (srf->snooper.image)
-                        memset(srf->snooper.image, 0x00, 64 * 64 * 4);
-                else
-                        DRM_ERROR("Failed to allocate cursor_image\n");
-
-        } else {
-                srf->snooper.image = NULL;
-        }
-        srf->snooper.crtc = NULL;
-
         rep->sid = user_srf->base.hash.key;
         if (rep->sid == SVGA3D_INVALID_ID)
                 DRM_ERROR("Created bad Surface ID.\n");
@@ -754,20 +771,29 @@ static size_t vmw_dmabuf_acc_size(struct ttm_bo_global *glob,
         return bo_user_size + page_array_size;
 }
 
-void vmw_dmabuf_bo_free(struct ttm_buffer_object *bo)
+void vmw_dmabuf_gmr_unbind(struct ttm_buffer_object *bo)
 {
         struct vmw_dma_buffer *vmw_bo = vmw_dma_buffer(bo);
         struct ttm_bo_global *glob = bo->glob;
         struct vmw_private *dev_priv =
                 container_of(bo->bdev, struct vmw_private, bdev);
 
-        ttm_mem_global_free(glob->mem_glob, bo->acc_size);
         if (vmw_bo->gmr_bound) {
                 vmw_gmr_unbind(dev_priv, vmw_bo->gmr_id);
                 spin_lock(&glob->lru_lock);
                 ida_remove(&dev_priv->gmr_ida, vmw_bo->gmr_id);
                 spin_unlock(&glob->lru_lock);
+                vmw_bo->gmr_bound = false;
         }
+}
+
+void vmw_dmabuf_bo_free(struct ttm_buffer_object *bo)
+{
+        struct vmw_dma_buffer *vmw_bo = vmw_dma_buffer(bo);
+        struct ttm_bo_global *glob = bo->glob;
+
+        vmw_dmabuf_gmr_unbind(bo);
+        ttm_mem_global_free(glob->mem_glob, bo->acc_size);
         kfree(vmw_bo);
 }
 
@@ -813,18 +839,10 @@ int vmw_dmabuf_init(struct vmw_private *dev_priv,
 static void vmw_user_dmabuf_destroy(struct ttm_buffer_object *bo)
 {
         struct vmw_user_dma_buffer *vmw_user_bo = vmw_user_dma_buffer(bo);
-        struct vmw_dma_buffer *vmw_bo = &vmw_user_bo->dma;
         struct ttm_bo_global *glob = bo->glob;
-        struct vmw_private *dev_priv =
-                container_of(bo->bdev, struct vmw_private, bdev);
 
+        vmw_dmabuf_gmr_unbind(bo);
         ttm_mem_global_free(glob->mem_glob, bo->acc_size);
-        if (vmw_bo->gmr_bound) {
-                vmw_gmr_unbind(dev_priv, vmw_bo->gmr_id);
-                spin_lock(&glob->lru_lock);
-                ida_remove(&dev_priv->gmr_ida, vmw_bo->gmr_id);
-                spin_unlock(&glob->lru_lock);
-        }
         kfree(vmw_user_bo);
 }
 
@@ -868,7 +886,7 @@ int vmw_dmabuf_alloc_ioctl(struct drm_device *dev, void *data,
         }
 
         ret = vmw_dmabuf_init(dev_priv, &vmw_user_bo->dma, req->size,
-                              &vmw_vram_placement, true,
+                              &vmw_vram_sys_placement, true,
                               &vmw_user_dmabuf_destroy);
         if (unlikely(ret != 0))
                 return ret;