Merge branch 'linus' into x86/uv

14 files changed, 192 insertions, 604 deletions

diff --git a/drivers/gpu/drm/drm_drv.c b/drivers/gpu/drm/drm_drv.c
index 96f416afc3f6..3ab1e9cc4692 100644
--- a/drivers/gpu/drm/drm_drv.c
+++ b/drivers/gpu/drm/drm_drv.c
@@ -266,11 +266,19 @@ int drm_init(struct drm_driver *driver)
         for (i = 0; driver->pci_driver.id_table[i].vendor != 0; i++) {
                 pid = (struct pci_device_id *)&driver->pci_driver.id_table[i];
 
+                /* Loop around setting up a DRM device for each PCI device
+                 * matching our ID and device class.  If we had the internal
+                 * function that pci_get_subsys and pci_get_class used, we'd
+                 * be able to just pass pid in instead of doing a two-stage
+                 * thing.
+                 */
                 pdev = NULL;
-                /* pass back in pdev to account for multiple identical cards */
                 while ((pdev =
                         pci_get_subsys(pid->vendor, pid->device, pid->subvendor,
                                        pid->subdevice, pdev)) != NULL) {
+                        if ((pdev->class & pid->class_mask) != pid->class)
+                                continue;
+
                         /* stealth mode requires a manual probe */
                         pci_dev_get(pdev);
                         drm_get_dev(pdev, pid, driver);
diff --git a/drivers/gpu/drm/drm_fops.c b/drivers/gpu/drm/drm_fops.c
index 0d46627663b1..78eeed5caaff 100644
--- a/drivers/gpu/drm/drm_fops.c
+++ b/drivers/gpu/drm/drm_fops.c
@@ -406,8 +406,6 @@ int drm_release(struct inode *inode, struct file *filp)
         if (dev->driver->driver_features & DRIVER_GEM)
                 drm_gem_release(dev, file_priv);
 
-        drm_fasync(-1, filp, 0);
-
         mutex_lock(&dev->ctxlist_mutex);
         if (!list_empty(&dev->ctxlist)) {
                 struct drm_ctx_list *pos, *n;
diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c
index 212a94f715b2..15c8dabc3e97 100644
--- a/drivers/gpu/drm/drm_irq.c
+++ b/drivers/gpu/drm/drm_irq.c
@@ -280,8 +280,6 @@ int drm_irq_uninstall(struct drm_device * dev)
 
         drm_vblank_cleanup(dev);
 
-        dev->locked_tasklet_func = NULL;
-
         return 0;
 }
 EXPORT_SYMBOL(drm_irq_uninstall);
@@ -699,81 +697,3 @@ void drm_handle_vblank(struct drm_device *dev, int crtc)
         drm_vbl_send_signals(dev, crtc);
 }
 EXPORT_SYMBOL(drm_handle_vblank);
-
-/**
- * Tasklet wrapper function.
- *
- * \param data DRM device in disguise.
- *
- * Attempts to grab the HW lock and calls the driver callback on success. On
- * failure, leave the lock marked as contended so the callback can be called
- * from drm_unlock().
- */
-static void drm_locked_tasklet_func(unsigned long data)
-{
-        struct drm_device *dev = (struct drm_device *)data;
-        unsigned long irqflags;
-        void (*tasklet_func)(struct drm_device *);
-        
-        spin_lock_irqsave(&dev->tasklet_lock, irqflags);
-        tasklet_func = dev->locked_tasklet_func;
-        spin_unlock_irqrestore(&dev->tasklet_lock, irqflags);
-
-        if (!tasklet_func ||
-            !drm_lock_take(&dev->lock,
-                           DRM_KERNEL_CONTEXT)) {
-                return;
-        }
-
-        dev->lock.lock_time = jiffies;
-        atomic_inc(&dev->counts[_DRM_STAT_LOCKS]);
-
-        spin_lock_irqsave(&dev->tasklet_lock, irqflags);
-        tasklet_func = dev->locked_tasklet_func;
-        dev->locked_tasklet_func = NULL;
-        spin_unlock_irqrestore(&dev->tasklet_lock, irqflags);
-        
-        if (tasklet_func != NULL)
-                tasklet_func(dev);
-
-        drm_lock_free(&dev->lock,
-                      DRM_KERNEL_CONTEXT);
-}
-
-/**
- * Schedule a tasklet to call back a driver hook with the HW lock held.
- *
- * \param dev DRM device.
- * \param func Driver callback.
- *
- * This is intended for triggering actions that require the HW lock from an
- * interrupt handler. The lock will be grabbed ASAP after the interrupt handler
- * completes. Note that the callback may be called from interrupt or process
- * context, it must not make any assumptions about this. Also, the HW lock will
- * be held with the kernel context or any client context.
- */
-void drm_locked_tasklet(struct drm_device *dev, void (*func)(struct drm_device *))
-{
-        unsigned long irqflags;
-        static DECLARE_TASKLET(drm_tasklet, drm_locked_tasklet_func, 0);
-
-        if (!drm_core_check_feature(dev, DRIVER_HAVE_IRQ) ||
-            test_bit(TASKLET_STATE_SCHED, &drm_tasklet.state))
-                return;
-
-        spin_lock_irqsave(&dev->tasklet_lock, irqflags);
-
-        if (dev->locked_tasklet_func) {
-                spin_unlock_irqrestore(&dev->tasklet_lock, irqflags);
-                return;
-        }
-
-        dev->locked_tasklet_func = func;
-
-        spin_unlock_irqrestore(&dev->tasklet_lock, irqflags);
-
-        drm_tasklet.data = (unsigned long)dev;
-
-        tasklet_hi_schedule(&drm_tasklet);
-}
-EXPORT_SYMBOL(drm_locked_tasklet);
diff --git a/drivers/gpu/drm/drm_lock.c b/drivers/gpu/drm/drm_lock.c
index 888159e03d26..1cfa72031f8f 100644
--- a/drivers/gpu/drm/drm_lock.c
+++ b/drivers/gpu/drm/drm_lock.c
@@ -154,8 +154,6 @@ int drm_lock(struct drm_device *dev, void *data, struct drm_file *file_priv)
 int drm_unlock(struct drm_device *dev, void *data, struct drm_file *file_priv)
 {
         struct drm_lock *lock = data;
-        unsigned long irqflags;
-        void (*tasklet_func)(struct drm_device *);
 
         if (lock->context == DRM_KERNEL_CONTEXT) {
                 DRM_ERROR("Process %d using kernel context %d\n",
@@ -163,13 +161,6 @@ int drm_unlock(struct drm_device *dev, void *data, struct drm_file *file_priv)
                 return -EINVAL;
         }
 
-        spin_lock_irqsave(&dev->tasklet_lock, irqflags);
-        tasklet_func = dev->locked_tasklet_func;
-        dev->locked_tasklet_func = NULL;
-        spin_unlock_irqrestore(&dev->tasklet_lock, irqflags);
-        if (tasklet_func != NULL)
-                tasklet_func(dev);
-
         atomic_inc(&dev->counts[_DRM_STAT_UNLOCKS]);
 
         /* kernel_context_switch isn't used by any of the x86 drm
diff --git a/drivers/gpu/drm/drm_stub.c b/drivers/gpu/drm/drm_stub.c
index 141e33004a76..66c96ec66672 100644
--- a/drivers/gpu/drm/drm_stub.c
+++ b/drivers/gpu/drm/drm_stub.c
@@ -92,7 +92,6 @@ static int drm_fill_in_dev(struct drm_device * dev, struct pci_dev *pdev,
 
         spin_lock_init(&dev->count_lock);
         spin_lock_init(&dev->drw_lock);
-        spin_lock_init(&dev->tasklet_lock);
         spin_lock_init(&dev->lock.spinlock);
         init_timer(&dev->timer);
         mutex_init(&dev->struct_mutex);
diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile
index 5ba78e4fd2b5..d8fb5d8ee7ea 100644
--- a/drivers/gpu/drm/i915/Makefile
+++ b/drivers/gpu/drm/i915/Makefile
@@ -3,13 +3,14 @@
 # Direct Rendering Infrastructure (DRI) in XFree86 4.1.0 and higher.
 
 ccflags-y := -Iinclude/drm
-i915-y := i915_drv.o i915_dma.o i915_irq.o i915_mem.o i915_opregion.o \
+i915-y := i915_drv.o i915_dma.o i915_irq.o i915_mem.o \
           i915_suspend.o \
           i915_gem.o \
           i915_gem_debug.o \
           i915_gem_proc.o \
           i915_gem_tiling.o
 
+i915-$(CONFIG_ACPI)     += i915_opregion.o
 i915-$(CONFIG_COMPAT)   += i915_ioc32.o
 
 obj-$(CONFIG_DRM_I915)  += i915.o
diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 01de536e0211..0d215e38606a 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -154,6 +154,9 @@ static int i915_dma_cleanup(struct drm_device * dev)
         if (I915_NEED_GFX_HWS(dev))
                 i915_free_hws(dev);
 
+        dev_priv->sarea = NULL;
+        dev_priv->sarea_priv = NULL;
+
         return 0;
 }
 
@@ -442,7 +445,7 @@ static void i915_emit_breadcrumb(struct drm_device *dev)
 
         BEGIN_LP_RING(4);
         OUT_RING(MI_STORE_DWORD_INDEX);
-        OUT_RING(5 << MI_STORE_DWORD_INDEX_SHIFT);
+        OUT_RING(I915_BREADCRUMB_INDEX << MI_STORE_DWORD_INDEX_SHIFT);
         OUT_RING(dev_priv->counter);
         OUT_RING(0);
         ADVANCE_LP_RING();
@@ -573,7 +576,7 @@ static int i915_dispatch_flip(struct drm_device * dev)
 
         BEGIN_LP_RING(4);
         OUT_RING(MI_STORE_DWORD_INDEX);
-        OUT_RING(5 << MI_STORE_DWORD_INDEX_SHIFT);
+        OUT_RING(I915_BREADCRUMB_INDEX << MI_STORE_DWORD_INDEX_SHIFT);
         OUT_RING(dev_priv->counter);
         OUT_RING(0);
         ADVANCE_LP_RING();
@@ -608,7 +611,6 @@ static int i915_batchbuffer(struct drm_device *dev, void *data,
                             struct drm_file *file_priv)
 {
         drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
-        u32 *hw_status = dev_priv->hw_status_page;
         drm_i915_sarea_t *sarea_priv = (drm_i915_sarea_t *)
             dev_priv->sarea_priv;
         drm_i915_batchbuffer_t *batch = data;
@@ -634,7 +636,7 @@ static int i915_batchbuffer(struct drm_device *dev, void *data,
         mutex_unlock(&dev->struct_mutex);
 
         if (sarea_priv)
-                sarea_priv->last_dispatch = (int)hw_status[5];
+                sarea_priv->last_dispatch = READ_BREADCRUMB(dev_priv);
         return ret;
 }
 
@@ -642,7 +644,6 @@ static int i915_cmdbuffer(struct drm_device *dev, void *data,
                           struct drm_file *file_priv)
 {
         drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
-        u32 *hw_status = dev_priv->hw_status_page;
         drm_i915_sarea_t *sarea_priv = (drm_i915_sarea_t *)
             dev_priv->sarea_priv;
         drm_i915_cmdbuffer_t *cmdbuf = data;
@@ -670,7 +671,7 @@ static int i915_cmdbuffer(struct drm_device *dev, void *data,
         }
 
         if (sarea_priv)
-                sarea_priv->last_dispatch = (int)hw_status[5];
+                sarea_priv->last_dispatch = READ_BREADCRUMB(dev_priv);
         return 0;
 }
 
@@ -849,8 +850,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
          * be lost or delayed
          */
         if (!IS_I945G(dev) && !IS_I945GM(dev) && !IS_I965GM(dev))
-                if (pci_enable_msi(dev->pdev))
-                        DRM_ERROR("failed to enable MSI\n");
+                pci_enable_msi(dev->pdev);
 
         intel_opregion_init(dev);
 
@@ -960,6 +960,7 @@ struct drm_ioctl_desc i915_ioctls[] = {
         DRM_IOCTL_DEF(DRM_I915_GEM_SW_FINISH, i915_gem_sw_finish_ioctl, 0),
         DRM_IOCTL_DEF(DRM_I915_GEM_SET_TILING, i915_gem_set_tiling, 0),
         DRM_IOCTL_DEF(DRM_I915_GEM_GET_TILING, i915_gem_get_tiling, 0),
+        DRM_IOCTL_DEF(DRM_I915_GEM_GET_APERTURE, i915_gem_get_aperture_ioctl, 0),
 };
 
 int i915_max_ioctl = DRM_ARRAY_SIZE(i915_ioctls);
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index f20ffe17df71..ef1c0b8f8d07 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -31,6 +31,7 @@
 #define _I915_DRV_H_
 
 #include "i915_reg.h"
+#include <linux/io-mapping.h>
 
 /* General customization:
  */
@@ -87,13 +88,6 @@ struct mem_block {
         struct drm_file *file_priv; /* NULL: free, -1: heap, other: real files */
 };
 
-typedef struct _drm_i915_vbl_swap {
-        struct list_head head;
-        drm_drawable_t drw_id;
-        unsigned int pipe;
-        unsigned int sequence;
-} drm_i915_vbl_swap_t;
-
 struct opregion_header;
 struct opregion_acpi;
 struct opregion_swsci;
@@ -145,10 +139,6 @@ typedef struct drm_i915_private {
         unsigned int sr01, adpa, ppcr, dvob, dvoc, lvds;
         int vblank_pipe;
 
-        spinlock_t swaps_lock;
-        drm_i915_vbl_swap_t vbl_swaps;
-        unsigned int swaps_pending;
-
         struct intel_opregion opregion;
 
         /* Register state */
@@ -156,6 +146,7 @@ typedef struct drm_i915_private {
         u32 saveDSPACNTR;
         u32 saveDSPBCNTR;
         u32 saveDSPARB;
+        u32 saveRENDERSTANDBY;
         u32 savePIPEACONF;
         u32 savePIPEBCONF;
         u32 savePIPEASRC;
@@ -240,12 +231,11 @@ typedef struct drm_i915_private {
         u8 saveDACDATA[256*3]; /* 256 3-byte colors */
         u8 saveCR[37];
 
-        /** Work task for vblank-related ring access */
-        struct work_struct vblank_work;
-
         struct {
                 struct drm_mm gtt_space;
 
+                struct io_mapping *gtt_mapping;
+
                 /**
                  * List of objects currently involved in rendering from the
                  * ringbuffer.
@@ -441,7 +431,6 @@ extern int i915_irq_wait(struct drm_device *dev, void *data,
 void i915_user_irq_get(struct drm_device *dev);
 void i915_user_irq_put(struct drm_device *dev);
 
-extern void i915_vblank_work_handler(struct work_struct *work);
 extern irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS);
 extern void i915_driver_irq_preinstall(struct drm_device * dev);
 extern int i915_driver_irq_postinstall(struct drm_device *dev);
@@ -502,6 +491,8 @@ int i915_gem_set_tiling(struct drm_device *dev, void *data,
                         struct drm_file *file_priv);
 int i915_gem_get_tiling(struct drm_device *dev, void *data,
                         struct drm_file *file_priv);
+int i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
+                                struct drm_file *file_priv);
 void i915_gem_load(struct drm_device *dev);
 int i915_gem_proc_init(struct drm_minor *minor);
 void i915_gem_proc_cleanup(struct drm_minor *minor);
@@ -539,11 +530,18 @@ extern int i915_restore_state(struct drm_device *dev);
 extern int i915_save_state(struct drm_device *dev);
 extern int i915_restore_state(struct drm_device *dev);
 
+#ifdef CONFIG_ACPI
 /* i915_opregion.c */
 extern int intel_opregion_init(struct drm_device *dev);
 extern void intel_opregion_free(struct drm_device *dev);
 extern void opregion_asle_intr(struct drm_device *dev);
 extern void opregion_enable_asle(struct drm_device *dev);
+#else
+static inline int intel_opregion_init(struct drm_device *dev) { return 0; }
+static inline void intel_opregion_free(struct drm_device *dev) { return; }
+static inline void opregion_asle_intr(struct drm_device *dev) { return; }
+static inline void opregion_enable_asle(struct drm_device *dev) { return; }
+#endif
 
 /**
  * Lock test for when it's just for synchronization of ring access.
@@ -610,8 +608,9 @@ extern void opregion_enable_asle(struct drm_device *dev);
  * The area from dword 0x20 to 0x3ff is available for driver usage.
  */
 #define READ_HWSP(dev_priv, reg)  (((volatile u32*)(dev_priv->hw_status_page))[reg])
-#define READ_BREADCRUMB(dev_priv) READ_HWSP(dev_priv, 5)
+#define READ_BREADCRUMB(dev_priv) READ_HWSP(dev_priv, I915_BREADCRUMB_INDEX)
 #define I915_GEM_HWS_INDEX              0x20
+#define I915_BREADCRUMB_INDEX           0x21
 
 extern int i915_wait_ring(struct drm_device * dev, int n, const char *caller);
 
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 17ae330ff269..6b4a2bd20640 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -79,6 +79,28 @@ i915_gem_init_ioctl(struct drm_device *dev, void *data,
         return 0;
 }
 
+int
+i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
+                            struct drm_file *file_priv)
+{
+        drm_i915_private_t *dev_priv = dev->dev_private;
+        struct drm_i915_gem_get_aperture *args = data;
+        struct drm_i915_gem_object *obj_priv;
+
+        if (!(dev->driver->driver_features & DRIVER_GEM))
+                return -ENODEV;
+
+        args->aper_size = dev->gtt_total;
+        args->aper_available_size = args->aper_size;
+
+        list_for_each_entry(obj_priv, &dev_priv->mm.active_list, list) {
+                if (obj_priv->pin_count > 0)
+                        args->aper_available_size -= obj_priv->obj->size;
+        }
+
+        return 0;
+}
+
 
 /**
  * Creates a new mm object and returns a handle to it.
@@ -171,35 +193,50 @@ i915_gem_pread_ioctl(struct drm_device *dev, void *data,
         return 0;
 }
 
-/*
- * Try to write quickly with an atomic kmap. Return true on success.
- *
- * If this fails (which includes a partial write), we'll redo the whole
- * thing with the slow version.
- *
- * This is a workaround for the low performance of iounmap (approximate
- * 10% cpu cost on normal 3D workloads).  kmap_atomic on HIGHMEM kernels
- * happens to let us map card memory without taking IPIs.  When the vmap
- * rework lands we should be able to dump this hack.
+/* This is the fast write path which cannot handle
+ * page faults in the source data
  */
-static inline int fast_user_write(unsigned long pfn, char __user *user_data,
-                                  int l, int o)
+
+static inline int
+fast_user_write(struct io_mapping *mapping,
+                loff_t page_base, int page_offset,
+                char __user *user_data,
+                int length)
 {
-#ifdef CONFIG_HIGHMEM
-        unsigned long unwritten;
         char *vaddr_atomic;
+        unsigned long unwritten;
 
-        vaddr_atomic = kmap_atomic_pfn(pfn, KM_USER0);
-#if WATCH_PWRITE
-        DRM_INFO("pwrite i %d o %d l %d pfn %ld vaddr %p\n",
-                 i, o, l, pfn, vaddr_atomic);
-#endif
-        unwritten = __copy_from_user_inatomic_nocache(vaddr_atomic + o, user_data, l);
-        kunmap_atomic(vaddr_atomic, KM_USER0);
-        return !unwritten;
-#else
+        vaddr_atomic = io_mapping_map_atomic_wc(mapping, page_base);
+        unwritten = __copy_from_user_inatomic_nocache(vaddr_atomic + page_offset,
+                                                      user_data, length);
+        io_mapping_unmap_atomic(vaddr_atomic);
+        if (unwritten)
+                return -EFAULT;
+        return 0;
+}
+
+/* Here's the write path which can sleep for
+ * page faults
+ */
+
+static inline int
+slow_user_write(struct io_mapping *mapping,
+                loff_t page_base, int page_offset,
+                char __user *user_data,
+                int length)
+{
+        char __iomem *vaddr;
+        unsigned long unwritten;
+
+        vaddr = io_mapping_map_wc(mapping, page_base);
+        if (vaddr == NULL)
+                return -EFAULT;
+        unwritten = __copy_from_user(vaddr + page_offset,
+                                     user_data, length);
+        io_mapping_unmap(vaddr);
+        if (unwritten)
+                return -EFAULT;
         return 0;
-#endif
 }
 
 static int
@@ -208,10 +245,12 @@ i915_gem_gtt_pwrite(struct drm_device *dev, struct drm_gem_object *obj,
                     struct drm_file *file_priv)
 {
         struct drm_i915_gem_object *obj_priv = obj->driver_private;
+        drm_i915_private_t *dev_priv = dev->dev_private;
         ssize_t remain;
-        loff_t offset;
+        loff_t offset, page_base;
         char __user *user_data;
-        int ret = 0;
+        int page_offset, page_length;
+        int ret;
 
         user_data = (char __user *) (uintptr_t) args->data_ptr;
         remain = args->size;
@@ -235,57 +274,37 @@ i915_gem_gtt_pwrite(struct drm_device *dev, struct drm_gem_object *obj,
         obj_priv->dirty = 1;
 
         while (remain > 0) {
-                unsigned long pfn;
-                int i, o, l;
-
                 /* Operation in this page
                  *
-                 * i = page number
-                 * o = offset within page
-                 * l = bytes to copy
+                 * page_base = page offset within aperture
+                 * page_offset = offset within page
+                 * page_length = bytes to copy for this page
                  */
-                i = offset >> PAGE_SHIFT;
-                o = offset & (PAGE_SIZE-1);
-                l = remain;
-                if ((o + l) > PAGE_SIZE)
-                        l = PAGE_SIZE - o;
-
-                pfn = (dev->agp->base >> PAGE_SHIFT) + i;
-
-                if (!fast_user_write(pfn, user_data, l, o)) {
-                        unsigned long unwritten;
-                        char __iomem *vaddr;
-
-                        vaddr = ioremap_wc(pfn << PAGE_SHIFT, PAGE_SIZE);
-#if WATCH_PWRITE
-                        DRM_INFO("pwrite slow i %d o %d l %d "
-                                 "pfn %ld vaddr %p\n",
-                                 i, o, l, pfn, vaddr);
-#endif
-                        if (vaddr == NULL) {
-                                ret = -EFAULT;
-                                goto fail;
-                        }
-                        unwritten = __copy_from_user(vaddr + o, user_data, l);
-#if WATCH_PWRITE
-                        DRM_INFO("unwritten %ld\n", unwritten);
-#endif
-                        iounmap(vaddr);
-                        if (unwritten) {
-                                ret = -EFAULT;
+                page_base = (offset & ~(PAGE_SIZE-1));
+                page_offset = offset & (PAGE_SIZE-1);
+                page_length = remain;
+                if ((page_offset + remain) > PAGE_SIZE)
+                        page_length = PAGE_SIZE - page_offset;
+
+                ret = fast_user_write (dev_priv->mm.gtt_mapping, page_base,
+                                       page_offset, user_data, page_length);
+
+                /* If we get a fault while copying data, then (presumably) our
+                 * source page isn't available. In this case, use the
+                 * non-atomic function
+                 */
+                if (ret) {
+                        ret = slow_user_write (dev_priv->mm.gtt_mapping,
+                                               page_base, page_offset,
+                                               user_data, page_length);
+                        if (ret)
                                 goto fail;
-                        }
                 }
 
-                remain -= l;
-                user_data += l;
-                offset += l;
+                remain -= page_length;
+                user_data += page_length;
+                offset += page_length;
         }
-#if WATCH_PWRITE && 1
-        i915_gem_clflush_object(obj);
-        i915_gem_dump_object(obj, args->offset + args->size, __func__, ~0);
-        i915_gem_clflush_object(obj);
-#endif
 
 fail:
         i915_gem_object_unpin(obj);
@@ -1436,11 +1455,9 @@ i915_gem_object_set_domain_range(struct drm_gem_object *obj,
                                                   read_domains, write_domain);
 
         /* Wait on any GPU rendering to the object to be flushed. */
-        if (obj->write_domain & ~(I915_GEM_DOMAIN_CPU | I915_GEM_DOMAIN_GTT)) {
-                ret = i915_gem_object_wait_rendering(obj);
-                if (ret)
-                        return ret;
-        }
+        ret = i915_gem_object_wait_rendering(obj);
+        if (ret)
+                return ret;
 
         if (obj_priv->page_cpu_valid == NULL) {
                 obj_priv->page_cpu_valid = drm_calloc(1, obj->size / PAGE_SIZE,
@@ -1503,12 +1520,12 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj,
                                  struct drm_i915_gem_exec_object *entry)
 {
         struct drm_device *dev = obj->dev;
+        drm_i915_private_t *dev_priv = dev->dev_private;
         struct drm_i915_gem_relocation_entry reloc;
         struct drm_i915_gem_relocation_entry __user *relocs;
         struct drm_i915_gem_object *obj_priv = obj->driver_private;
         int i, ret;
-        uint32_t last_reloc_offset = -1;
-        void __iomem *reloc_page = NULL;
+        void __iomem *reloc_page;
 
         /* Choose the GTT offset for our buffer and put it there. */
         ret = i915_gem_object_pin(obj, (uint32_t) entry->alignment);
@@ -1631,26 +1648,11 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj,
                  * perform.
                  */
                 reloc_offset = obj_priv->gtt_offset + reloc.offset;
-                if (reloc_page == NULL ||
-                    (last_reloc_offset & ~(PAGE_SIZE - 1)) !=
-                    (reloc_offset & ~(PAGE_SIZE - 1))) {
-                        if (reloc_page != NULL)
-                                iounmap(reloc_page);
-
-                        reloc_page = ioremap_wc(dev->agp->base +
-                                                (reloc_offset &
-                                                 ~(PAGE_SIZE - 1)),
-                                                PAGE_SIZE);
-                        last_reloc_offset = reloc_offset;
-                        if (reloc_page == NULL) {
-                                drm_gem_object_unreference(target_obj);
-                                i915_gem_object_unpin(obj);
-                                return -ENOMEM;
-                        }
-                }
-
+                reloc_page = io_mapping_map_atomic_wc(dev_priv->mm.gtt_mapping,
+                                                      (reloc_offset &
+                                                       ~(PAGE_SIZE - 1)));
                 reloc_entry = (uint32_t __iomem *)(reloc_page +
-                                           (reloc_offset & (PAGE_SIZE - 1)));
+                                                   (reloc_offset & (PAGE_SIZE - 1)));
                 reloc_val = target_obj_priv->gtt_offset + reloc.delta;
 
 #if WATCH_BUF
@@ -1659,6 +1661,7 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj,
                           readl(reloc_entry), reloc_val);
 #endif
                 writel(reloc_val, reloc_entry);
+                io_mapping_unmap_atomic(reloc_page);
 
                 /* Write the updated presumed offset for this entry back out
                  * to the user.
@@ -1674,9 +1677,6 @@ i915_gem_object_pin_and_relocate(struct drm_gem_object *obj,
                 drm_gem_object_unreference(target_obj);
         }
 
-        if (reloc_page != NULL)
-                iounmap(reloc_page);
-
 #if WATCH_BUF
         if (0)
                 i915_gem_dump_object(obj, 128, __func__, ~0);
@@ -2518,6 +2518,10 @@ i915_gem_entervt_ioctl(struct drm_device *dev, void *data,
         if (ret != 0)
                 return ret;
 
+        dev_priv->mm.gtt_mapping = io_mapping_create_wc(dev->agp->base,
+                                                        dev->agp->agp_info.aper_size
+                                                        * 1024 * 1024);
+
         mutex_lock(&dev->struct_mutex);
         BUG_ON(!list_empty(&dev_priv->mm.active_list));
         BUG_ON(!list_empty(&dev_priv->mm.flushing_list));
@@ -2535,11 +2539,13 @@ int
 i915_gem_leavevt_ioctl(struct drm_device *dev, void *data,
                        struct drm_file *file_priv)
 {
+        drm_i915_private_t *dev_priv = dev->dev_private;
         int ret;
 
         ret = i915_gem_idle(dev);
         drm_irq_uninstall(dev);
 
+        io_mapping_free(dev_priv->mm.gtt_mapping);
         return ret;
 }
 
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 26f48932a51e..82752d6177a4 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -80,211 +80,6 @@ i915_pipe_enabled(struct drm_device *dev, int pipe)
         return 0;
 }
 
-/**
- * Emit blits for scheduled buffer swaps.
- *
- * This function will be called with the HW lock held.
- * Because this function must grab the ring mutex (dev->struct_mutex),
- * it can no longer run at soft irq time. We'll fix this when we do
- * the DRI2 swap buffer work.
- */
-static void i915_vblank_tasklet(struct drm_device *dev)
-{
-        drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
-        unsigned long irqflags;
-        struct list_head *list, *tmp, hits, *hit;
-        int nhits, nrects, slice[2], upper[2], lower[2], i;
-        unsigned counter[2];
-        struct drm_drawable_info *drw;
-        drm_i915_sarea_t *sarea_priv = dev_priv->sarea_priv;
-        u32 cpp = dev_priv->cpp;
-        u32 cmd = (cpp == 4) ? (XY_SRC_COPY_BLT_CMD |
-                                XY_SRC_COPY_BLT_WRITE_ALPHA |
-                                XY_SRC_COPY_BLT_WRITE_RGB)
-                             : XY_SRC_COPY_BLT_CMD;
-        u32 src_pitch = sarea_priv->pitch * cpp;
-        u32 dst_pitch = sarea_priv->pitch * cpp;
-        u32 ropcpp = (0xcc << 16) | ((cpp - 1) << 24);
-        RING_LOCALS;
-
-        mutex_lock(&dev->struct_mutex);
-
-        if (IS_I965G(dev) && sarea_priv->front_tiled) {
-                cmd |= XY_SRC_COPY_BLT_DST_TILED;
-                dst_pitch >>= 2;
-        }
-        if (IS_I965G(dev) && sarea_priv->back_tiled) {
-                cmd |= XY_SRC_COPY_BLT_SRC_TILED;
-                src_pitch >>= 2;
-        }
-
-        counter[0] = drm_vblank_count(dev, 0);
-        counter[1] = drm_vblank_count(dev, 1);
-
-        DRM_DEBUG("\n");
-
-        INIT_LIST_HEAD(&hits);
-
-        nhits = nrects = 0;
-
-        spin_lock_irqsave(&dev_priv->swaps_lock, irqflags);
-
-        /* Find buffer swaps scheduled for this vertical blank */
-        list_for_each_safe(list, tmp, &dev_priv->vbl_swaps.head) {
-                drm_i915_vbl_swap_t *vbl_swap =
-                        list_entry(list, drm_i915_vbl_swap_t, head);
-                int pipe = vbl_swap->pipe;
-
-                if ((counter[pipe] - vbl_swap->sequence) > (1<<23))
-                        continue;
-
-                list_del(list);
-                dev_priv->swaps_pending--;
-                drm_vblank_put(dev, pipe);
-
-                spin_unlock(&dev_priv->swaps_lock);
-                spin_lock(&dev->drw_lock);
-
-                drw = drm_get_drawable_info(dev, vbl_swap->drw_id);
-
-                list_for_each(hit, &hits) {
-                        drm_i915_vbl_swap_t *swap_cmp =
-                                list_entry(hit, drm_i915_vbl_swap_t, head);
-                        struct drm_drawable_info *drw_cmp =
-                                drm_get_drawable_info(dev, swap_cmp->drw_id);
-
-                        /* Make sure both drawables are still
-                         * around and have some rectangles before
-                         * we look inside to order them for the
-                         * blts below.
-                         */
-                        if (drw_cmp && drw_cmp->num_rects > 0 &&
-                            drw && drw->num_rects > 0 &&
-                            drw_cmp->rects[0].y1 > drw->rects[0].y1) {
-                                list_add_tail(list, hit);
-                                break;
-                        }
-                }
-
-                spin_unlock(&dev->drw_lock);
-
-                /* List of hits was empty, or we reached the end of it */
-                if (hit == &hits)
-                        list_add_tail(list, hits.prev);
-
-                nhits++;
-
-                spin_lock(&dev_priv->swaps_lock);
-        }
-
-        if (nhits == 0) {
-                spin_unlock_irqrestore(&dev_priv->swaps_lock, irqflags);
-                mutex_unlock(&dev->struct_mutex);
-                return;
-        }
-
-        spin_unlock(&dev_priv->swaps_lock);
-
-        i915_kernel_lost_context(dev);
-
-        if (IS_I965G(dev)) {
-                BEGIN_LP_RING(4);
-
-                OUT_RING(GFX_OP_DRAWRECT_INFO_I965);
-                OUT_RING(0);
-                OUT_RING(((sarea_priv->width - 1) & 0xffff) | ((sarea_priv->height - 1) << 16));
-                OUT_RING(0);
-                ADVANCE_LP_RING();
-        } else {
-                BEGIN_LP_RING(6);
-
-                OUT_RING(GFX_OP_DRAWRECT_INFO);
-                OUT_RING(0);
-                OUT_RING(0);
-                OUT_RING(sarea_priv->width | sarea_priv->height << 16);
-                OUT_RING(sarea_priv->width | sarea_priv->height << 16);
-                OUT_RING(0);
-
-                ADVANCE_LP_RING();
-        }
-
-        sarea_priv->ctxOwner = DRM_KERNEL_CONTEXT;
-
-        upper[0] = upper[1] = 0;
-        slice[0] = max(sarea_priv->pipeA_h / nhits, 1);
-        slice[1] = max(sarea_priv->pipeB_h / nhits, 1);
-        lower[0] = sarea_priv->pipeA_y + slice[0];
-        lower[1] = sarea_priv->pipeB_y + slice[0];
-
-        spin_lock(&dev->drw_lock);
-
-        /* Emit blits for buffer swaps, partitioning both outputs into as many
-         * slices as there are buffer swaps scheduled in order to avoid tearing
-         * (based on the assumption that a single buffer swap would always
-         * complete before scanout starts).
-         */
-        for (i = 0; i++ < nhits;
-             upper[0] = lower[0], lower[0] += slice[0],
-             upper[1] = lower[1], lower[1] += slice[1]) {
-                if (i == nhits)
-                        lower[0] = lower[1] = sarea_priv->height;
-
-                list_for_each(hit, &hits) {
-                        drm_i915_vbl_swap_t *swap_hit =
-                                list_entry(hit, drm_i915_vbl_swap_t, head);
-                        struct drm_clip_rect *rect;
-                        int num_rects, pipe;
-                        unsigned short top, bottom;
-
-                        drw = drm_get_drawable_info(dev, swap_hit->drw_id);
-
-                        /* The drawable may have been destroyed since
-                         * the vblank swap was queued
-                         */
-                        if (!drw)
-                                continue;
-
-                        rect = drw->rects;
-                        pipe = swap_hit->pipe;
-                        top = upper[pipe];
-                        bottom = lower[pipe];
-
-                        for (num_rects = drw->num_rects; num_rects--; rect++) {
-                                int y1 = max(rect->y1, top);
-                                int y2 = min(rect->y2, bottom);
-
-                                if (y1 >= y2)
-                                        continue;
-
-                                BEGIN_LP_RING(8);
-
-                                OUT_RING(cmd);
-                                OUT_RING(ropcpp | dst_pitch);
-                                OUT_RING((y1 << 16) | rect->x1);
-                                OUT_RING((y2 << 16) | rect->x2);
-                                OUT_RING(sarea_priv->front_offset);
-                                OUT_RING((y1 << 16) | rect->x1);
-                                OUT_RING(src_pitch);
-                                OUT_RING(sarea_priv->back_offset);
-
-                                ADVANCE_LP_RING();
-                        }
-                }
-        }
-
-        spin_unlock_irqrestore(&dev->drw_lock, irqflags);
-        mutex_unlock(&dev->struct_mutex);
-
-        list_for_each_safe(hit, tmp, &hits) {
-                drm_i915_vbl_swap_t *swap_hit =
-                        list_entry(hit, drm_i915_vbl_swap_t, head);
-
-                list_del(hit);
-
-                drm_free(swap_hit, sizeof(*swap_hit), DRM_MEM_DRIVER);
-        }
-}
-
 /* Called from drm generic code, passed a 'crtc', which
  * we use as a pipe index
  */
@@ -322,40 +117,6 @@ u32 i915_get_vblank_counter(struct drm_device *dev, int pipe)
         return count;
 }
 
-void
-i915_vblank_work_handler(struct work_struct *work)
-{
-        drm_i915_private_t *dev_priv = container_of(work, drm_i915_private_t,
-                                                    vblank_work);
-        struct drm_device *dev = dev_priv->dev;
-        unsigned long irqflags;
-
-        if (dev->lock.hw_lock == NULL) {
-                i915_vblank_tasklet(dev);
-                return;
-        }
-
-        spin_lock_irqsave(&dev->tasklet_lock, irqflags);
-        dev->locked_tasklet_func = i915_vblank_tasklet;
-        spin_unlock_irqrestore(&dev->tasklet_lock, irqflags);
-
-        /* Try to get the lock now, if this fails, the lock
-         * holder will execute the tasklet during unlock
-         */
-        if (!drm_lock_take(&dev->lock, DRM_KERNEL_CONTEXT))
-                return;
-
-        dev->lock.lock_time = jiffies;
-        atomic_inc(&dev->counts[_DRM_STAT_LOCKS]);
-
-        spin_lock_irqsave(&dev->tasklet_lock, irqflags);
-        dev->locked_tasklet_func = NULL;
-        spin_unlock_irqrestore(&dev->tasklet_lock, irqflags);
-
-        i915_vblank_tasklet(dev);
-        drm_lock_free(&dev->lock, DRM_KERNEL_CONTEXT);
-}
-
 irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
 {
         struct drm_device *dev = (struct drm_device *) arg;
@@ -433,9 +194,6 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
         if (iir & I915_ASLE_INTERRUPT)
                 opregion_asle_intr(dev);
 
-        if (vblank && dev_priv->swaps_pending > 0)
-                schedule_work(&dev_priv->vblank_work);
-
         return IRQ_HANDLED;
 }
 
@@ -454,12 +212,10 @@ static int i915_emit_irq(struct drm_device * dev)
         if (dev_priv->sarea_priv)
                 dev_priv->sarea_priv->last_enqueue = dev_priv->counter;
 
-        BEGIN_LP_RING(6);
+        BEGIN_LP_RING(4);
         OUT_RING(MI_STORE_DWORD_INDEX);
-        OUT_RING(5 << MI_STORE_DWORD_INDEX_SHIFT);
+        OUT_RING(I915_BREADCRUMB_INDEX << MI_STORE_DWORD_INDEX_SHIFT);
         OUT_RING(dev_priv->counter);
-        OUT_RING(0);
-        OUT_RING(0);
         OUT_RING(MI_USER_INTERRUPT);
         ADVANCE_LP_RING();
 
@@ -696,123 +452,21 @@ int i915_vblank_pipe_get(struct drm_device *dev, void *data,
 int i915_vblank_swap(struct drm_device *dev, void *data,
                      struct drm_file *file_priv)
 {
-        drm_i915_private_t *dev_priv = dev->dev_private;
-        drm_i915_vblank_swap_t *swap = data;
-        drm_i915_vbl_swap_t *vbl_swap, *vbl_old;
-        unsigned int pipe, seqtype, curseq;
-        unsigned long irqflags;
-        struct list_head *list;
-        int ret;
-
-        if (!dev_priv || !dev_priv->sarea_priv) {
-                DRM_ERROR("%s called with no initialization\n", __func__);
-                return -EINVAL;
-        }
-
-        if (dev_priv->sarea_priv->rotation) {
-                DRM_DEBUG("Rotation not supported\n");
-                return -EINVAL;
-        }
-
-        if (swap->seqtype & ~(_DRM_VBLANK_RELATIVE | _DRM_VBLANK_ABSOLUTE |
-                             _DRM_VBLANK_SECONDARY | _DRM_VBLANK_NEXTONMISS)) {
-                DRM_ERROR("Invalid sequence type 0x%x\n", swap->seqtype);
-                return -EINVAL;
-        }
-
-        pipe = (swap->seqtype & _DRM_VBLANK_SECONDARY) ? 1 : 0;
-
-        seqtype = swap->seqtype & (_DRM_VBLANK_RELATIVE | _DRM_VBLANK_ABSOLUTE);
-
-        if (!(dev_priv->vblank_pipe & (1 << pipe))) {
-                DRM_ERROR("Invalid pipe %d\n", pipe);
-                return -EINVAL;
-        }
-
-        spin_lock_irqsave(&dev->drw_lock, irqflags);
-
-        if (!drm_get_drawable_info(dev, swap->drawable)) {
-                spin_unlock_irqrestore(&dev->drw_lock, irqflags);
-                DRM_DEBUG("Invalid drawable ID %d\n", swap->drawable);
-                return -EINVAL;
-        }
-
-        spin_unlock_irqrestore(&dev->drw_lock, irqflags);
-
-        /*
-         * We take the ref here and put it when the swap actually completes
-         * in the tasklet.
+        /* The delayed swap mechanism was fundamentally racy, and has been
+         * removed.  The model was that the client requested a delayed flip/swap
+         * from the kernel, then waited for vblank before continuing to perform
+         * rendering.  The problem was that the kernel might wake the client
+         * up before it dispatched the vblank swap (since the lock has to be
+         * held while touching the ringbuffer), in which case the client would
+         * clear and start the next frame before the swap occurred, and
+         * flicker would occur in addition to likely missing the vblank.
+         *
+         * In the absence of this ioctl, userland falls back to a correct path
+         * of waiting for a vblank, then dispatching the swap on its own.
+         * Context switching to userland and back is plenty fast enough for
+         * meeting the requirements of vblank swapping.
          */
-        ret = drm_vblank_get(dev, pipe);
-        if (ret)
-                return ret;
-        curseq = drm_vblank_count(dev, pipe);
-
-        if (seqtype == _DRM_VBLANK_RELATIVE)
-                swap->sequence += curseq;
-
-        if ((curseq - swap->sequence) <= (1<<23)) {
-                if (swap->seqtype & _DRM_VBLANK_NEXTONMISS) {
-                        swap->sequence = curseq + 1;
-                } else {
-                        DRM_DEBUG("Missed target sequence\n");
-                        drm_vblank_put(dev, pipe);
-                        return -EINVAL;
-                }
-        }
-
-        vbl_swap = drm_calloc(1, sizeof(*vbl_swap), DRM_MEM_DRIVER);
-
-        if (!vbl_swap) {
-                DRM_ERROR("Failed to allocate memory to queue swap\n");
-                drm_vblank_put(dev, pipe);
-                return -ENOMEM;
-        }
-
-        vbl_swap->drw_id = swap->drawable;
-        vbl_swap->pipe = pipe;
-        vbl_swap->sequence = swap->sequence;
-
-        spin_lock_irqsave(&dev_priv->swaps_lock, irqflags);
-
-        list_for_each(list, &dev_priv->vbl_swaps.head) {
-                vbl_old = list_entry(list, drm_i915_vbl_swap_t, head);
-
-                if (vbl_old->drw_id == swap->drawable &&
-                    vbl_old->pipe == pipe &&
-                    vbl_old->sequence == swap->sequence) {
-                        spin_unlock_irqrestore(&dev_priv->swaps_lock, irqflags);
-                        drm_vblank_put(dev, pipe);
-                        drm_free(vbl_swap, sizeof(*vbl_swap), DRM_MEM_DRIVER);
-                        DRM_DEBUG("Already scheduled\n");
-                        return 0;
-                }
-        }
-
-        if (dev_priv->swaps_pending >= 10) {
-                DRM_DEBUG("Too many swaps queued\n");
-                DRM_DEBUG(" pipe 0: %d pipe 1: %d\n",
-                          drm_vblank_count(dev, 0),
-                          drm_vblank_count(dev, 1));
-
-                list_for_each(list, &dev_priv->vbl_swaps.head) {
-                        vbl_old = list_entry(list, drm_i915_vbl_swap_t, head);
-                        DRM_DEBUG("\tdrw %x pipe %d seq %x\n",
-                                  vbl_old->drw_id, vbl_old->pipe,
-                                  vbl_old->sequence);
-                }
-                spin_unlock_irqrestore(&dev_priv->swaps_lock, irqflags);
-                drm_vblank_put(dev, pipe);
-                drm_free(vbl_swap, sizeof(*vbl_swap), DRM_MEM_DRIVER);
-                return -EBUSY;
-        }
-
-        list_add_tail(&vbl_swap->head, &dev_priv->vbl_swaps.head);
-        dev_priv->swaps_pending++;
-
-        spin_unlock_irqrestore(&dev_priv->swaps_lock, irqflags);
-
-        return 0;
+        return -EINVAL;
 }
 
 /* drm_dma.h hooks
@@ -831,11 +485,6 @@ int i915_driver_irq_postinstall(struct drm_device *dev)
         drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private;
         int ret, num_pipes = 2;
 
-        spin_lock_init(&dev_priv->swaps_lock);
-        INIT_LIST_HEAD(&dev_priv->vbl_swaps.head);
-        INIT_WORK(&dev_priv->vblank_work, i915_vblank_work_handler);
-        dev_priv->swaps_pending = 0;
-
         /* Set initial unmasked IRQs to just the selected vblank pipes. */
         dev_priv->irq_mask_reg = ~0;
 
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 5c2d9f206d05..0e476eba36e6 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -527,6 +527,9 @@
 #define C0DRB3                  0x10206
 #define C1DRB3                  0x10606
 
+/** GM965 GM45 render standby register */
+#define MCHBAR_RENDER_STANDBY   0x111B8
+
 /*
  * Overlay regs
  */
diff --git a/drivers/gpu/drm/i915/i915_suspend.c b/drivers/gpu/drm/i915/i915_suspend.c
index 603fe742ccd4..5ddc6e595c0c 100644
--- a/drivers/gpu/drm/i915/i915_suspend.c
+++ b/drivers/gpu/drm/i915/i915_suspend.c
@@ -240,6 +240,10 @@ int i915_save_state(struct drm_device *dev)
 
         pci_read_config_byte(dev->pdev, LBB, &dev_priv->saveLBB);
 
+        /* Render Standby */
+        if (IS_I965G(dev) && IS_MOBILE(dev))
+                dev_priv->saveRENDERSTANDBY = I915_READ(MCHBAR_RENDER_STANDBY);
+
         /* Display arbitration control */
         dev_priv->saveDSPARB = I915_READ(DSPARB);
 
@@ -365,6 +369,11 @@ int i915_restore_state(struct drm_device *dev)
 
         pci_write_config_byte(dev->pdev, LBB, dev_priv->saveLBB);
 
+        /* Render Standby */
+        if (IS_I965G(dev) && IS_MOBILE(dev))
+                I915_WRITE(MCHBAR_RENDER_STANDBY, dev_priv->saveRENDERSTANDBY);
+
+        /* Display arbitration */
         I915_WRITE(DSPARB, dev_priv->saveDSPARB);
 
         /* Pipe & plane A info */
diff --git a/drivers/gpu/drm/radeon/radeon_cp.c b/drivers/gpu/drm/radeon/radeon_cp.c
index 59a2132a8f57..abdc1ae38467 100644
--- a/drivers/gpu/drm/radeon/radeon_cp.c
+++ b/drivers/gpu/drm/radeon/radeon_cp.c
@@ -653,15 +653,16 @@ static void radeon_cp_init_ring_buffer(struct drm_device * dev,
         RADEON_WRITE(RADEON_SCRATCH_UMSK, 0x7);
 
         /* Turn on bus mastering */
-        if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS400) ||
-            ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) ||
+        if (((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS690) ||
             ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS740)) {
-                /* rs400, rs690/rs740 */
-                tmp = RADEON_READ(RADEON_BUS_CNTL) & ~RS400_BUS_MASTER_DIS;
+                /* rs600/rs690/rs740 */
+                tmp = RADEON_READ(RADEON_BUS_CNTL) & ~RS600_BUS_MASTER_DIS;
                 RADEON_WRITE(RADEON_BUS_CNTL, tmp);
-        } else if (!(((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RV380) ||
-                    ((dev_priv->flags & RADEON_FAMILY_MASK) >= CHIP_R423))) {
-                /* r1xx, r2xx, r300, r(v)350, r420/r481, rs480 */
+        } else if (((dev_priv->flags & RADEON_FAMILY_MASK) <= CHIP_RV350) ||
+                   ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_R420) ||
+                   ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS400) ||
+                   ((dev_priv->flags & RADEON_FAMILY_MASK) == CHIP_RS480)) {
+                /* r1xx, r2xx, r300, r(v)350, r420/r481, rs400/rs480 */
                 tmp = RADEON_READ(RADEON_BUS_CNTL) & ~RADEON_BUS_MASTER_DIS;
                 RADEON_WRITE(RADEON_BUS_CNTL, tmp);
         } /* PCIE cards appears to not need this */
@@ -1750,6 +1751,12 @@ int radeon_driver_load(struct drm_device *dev, unsigned long flags)
         else
                 dev_priv->flags |= RADEON_IS_PCI;
 
+        ret = drm_addmap(dev, drm_get_resource_start(dev, 2),
+                         drm_get_resource_len(dev, 2), _DRM_REGISTERS,
+                         _DRM_READ_ONLY | _DRM_DRIVER, &dev_priv->mmio);
+        if (ret != 0)
+                return ret;
+
         DRM_DEBUG("%s card detected\n",
                   ((dev_priv->flags & RADEON_IS_AGP) ? "AGP" : (((dev_priv->flags & RADEON_IS_PCIE) ? "PCIE" : "PCI"))));
         return ret;
@@ -1766,12 +1773,6 @@ int radeon_driver_firstopen(struct drm_device *dev)
 
         dev_priv->gart_info.table_size = RADEON_PCIGART_TABLE_SIZE;
 
-        ret = drm_addmap(dev, drm_get_resource_start(dev, 2),
-                         drm_get_resource_len(dev, 2), _DRM_REGISTERS,
-                         _DRM_READ_ONLY, &dev_priv->mmio);
-        if (ret != 0)
-                return ret;
-
         dev_priv->fb_aper_offset = drm_get_resource_start(dev, 0);
         ret = drm_addmap(dev, dev_priv->fb_aper_offset,
                          drm_get_resource_len(dev, 0), _DRM_FRAME_BUFFER,
@@ -1787,6 +1788,9 @@ int radeon_driver_unload(struct drm_device *dev)
         drm_radeon_private_t *dev_priv = dev->dev_private;
 
         DRM_DEBUG("\n");
+
+        drm_rmmap(dev, dev_priv->mmio);
+
         drm_free(dev_priv, sizeof(*dev_priv), DRM_MEM_DRIVER);
 
         dev->dev_private = NULL;
diff --git a/drivers/gpu/drm/radeon/radeon_drv.h b/drivers/gpu/drm/radeon/radeon_drv.h
index 4dbb813910c3..7a183789be97 100644
--- a/drivers/gpu/drm/radeon/radeon_drv.h
+++ b/drivers/gpu/drm/radeon/radeon_drv.h
@@ -287,7 +287,6 @@ typedef struct drm_radeon_private {
         unsigned long gart_textures_offset;
 
         drm_local_map_t *sarea;
-        drm_local_map_t *mmio;
         drm_local_map_t *cp_ring;
         drm_local_map_t *ring_rptr;
         drm_local_map_t *gart_textures;
@@ -318,6 +317,7 @@ typedef struct drm_radeon_private {
 
         int num_gb_pipes;
         int track_flush;
+        drm_local_map_t *mmio;
 } drm_radeon_private_t;
 
 typedef struct drm_radeon_buf_priv {
@@ -447,12 +447,12 @@ extern int r300_do_cp_cmdbuf(struct drm_device *dev,
  * handling, not bus mastering itself.
  */
 #define RADEON_BUS_CNTL                 0x0030
-/* r1xx, r2xx, r300, r(v)350, r420/r481, rs480 */
+/* r1xx, r2xx, r300, r(v)350, r420/r481, rs400/rs480 */
 #       define RADEON_BUS_MASTER_DIS            (1 << 6)
-/* rs400, rs690/rs740 */
-#       define RS400_BUS_MASTER_DIS             (1 << 14)
-#       define RS400_MSI_REARM                  (1 << 20)
-/* see RS480_MSI_REARM in AIC_CNTL for rs480 */
+/* rs600/rs690/rs740 */
+#       define RS600_BUS_MASTER_DIS             (1 << 14)
+#       define RS600_MSI_REARM                  (1 << 20)
+/* see RS400_MSI_REARM in AIC_CNTL for rs480 */
 
 #define RADEON_BUS_CNTL1                0x0034
 #       define RADEON_PMI_BM_DIS                (1 << 2)
@@ -937,7 +937,7 @@ extern int r300_do_cp_cmdbuf(struct drm_device *dev,
 
 #define RADEON_AIC_CNTL                 0x01d0
 #       define RADEON_PCIGART_TRANSLATE_EN      (1 << 0)
-#       define RS480_MSI_REARM                  (1 << 3)
+#       define RS400_MSI_REARM                  (1 << 3)
 #define RADEON_AIC_STAT                 0x01d4
 #define RADEON_AIC_PT_BASE              0x01d8
 #define RADEON_AIC_LO_ADDR              0x01dc