Merge branch 'drm-patches' of master.kernel.org:/pub/scm/linux/kernel/git/airlied/drm-2.6

* 'drm-patches' of master.kernel.org:/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/i915: Add 965GM pci id update
  drm: just use io_remap_pfn_range on all archs..
  drm: fix DRM_CONSISTENT mapping
  drm: fix up mmap locking in preparation for ttm changes
  drm: fix driver deadlock with AIGLX and reclaim_buffers_locked
  drm: fix warning in drm_fops.c
  drm: allow for more generic drm ioctls
  drm: fix alpha domain handling
  via: fix CX700 pci id
  drm: make drm_io_prot static.
  drm: remove via_mm.h
  drm: add missing NULL assignment
  drm/radeon: Fix u32 overflows when determining AGP base address in card space.
  drm: port over use_vmalloc code from git hashtab
  drm: fix crash with fops lock and fixup sarea/page size locking
  drm: bring bufs code from git tree.
  drm: move protection stuff into separate function
  drm: Use ARRAY_SIZE macro when appropriate
  drm: update README.drm (bugzilla #7933)
  drm: remove unused exports

drivers/char/drm/README.drm

@@ -1,6 +1,6 @@
 ************************************************************
 * For the very latest on DRI development, please see:      *
-*     http://dri.sourceforge.net/                          *
+*     http://dri.freedesktop.org/                          *
 ************************************************************
 
 The Direct Rendering Manager (drm) is a device-independent kernel-level
@@ -26,21 +26,19 @@ ways:
 
 
 Documentation on the DRI is available from:
-    http://precisioninsight.com/piinsights.html
+    http://dri.freedesktop.org/wiki/Documentation
+    http://sourceforge.net/project/showfiles.php?group_id=387
+    http://dri.sourceforge.net/doc/
 
 For specific information about kernel-level support, see:
 
     The Direct Rendering Manager, Kernel Support for the Direct Rendering
     Infrastructure
-    http://precisioninsight.com/dr/drm.html
+    http://dri.sourceforge.net/doc/drm_low_level.html
 
     Hardware Locking for the Direct Rendering Infrastructure
-    http://precisioninsight.com/dr/locking.html
+    http://dri.sourceforge.net/doc/hardware_locking_low_level.html
 
     A Security Analysis of the Direct Rendering Infrastructure
-    http://precisioninsight.com/dr/security.html
+    http://dri.sourceforge.net/doc/security_low_level.html
 
-************************************************************
-* For the very latest on DRI development, please see:      *
-*     http://dri.sourceforge.net/                          *
-************************************************************

drivers/char/drm/drm.h

@@ -654,11 +654,13 @@ typedef struct drm_set_version {
 
 /**
  * Device specific ioctls should only be in their respective headers
- * The device specific ioctl range is from 0x40 to 0x79.
+ * The device specific ioctl range is from 0x40 to 0x99.
+ * Generic IOCTLS restart at 0xA0.
  *
  * \sa drmCommandNone(), drmCommandRead(), drmCommandWrite(), and
  * drmCommandReadWrite().
  */
 #define DRM_COMMAND_BASE                0x40
+#define DRM_COMMAND_END			0xA0
 
 #endif

drivers/char/drm/drmP.h

@@ -414,6 +414,10 @@ typedef struct drm_lock_data {
 	struct file *filp;		/**< File descr of lock holder (0=kernel) */
 	wait_queue_head_t lock_queue;	/**< Queue of blocked processes */
 	unsigned long lock_time;	/**< Time of last lock in jiffies */
+	spinlock_t spinlock;
+	uint32_t kernel_waiters;
+	uint32_t user_waiters;
+	int idle_has_lock;
 } drm_lock_data_t;
 
 /**
@@ -590,6 +594,8 @@ struct drm_driver {
 	void (*reclaim_buffers) (struct drm_device * dev, struct file * filp);
 	void (*reclaim_buffers_locked) (struct drm_device *dev,
 					struct file *filp);
+	void (*reclaim_buffers_idlelocked) (struct drm_device *dev,
+					struct file * filp);
 	unsigned long (*get_map_ofs) (drm_map_t * map);
 	unsigned long (*get_reg_ofs) (struct drm_device * dev);
 	void (*set_version) (struct drm_device * dev, drm_set_version_t * sv);
@@ -764,7 +770,7 @@ static __inline__ int drm_core_check_feature(struct drm_device *dev,
 }
 
 #ifdef __alpha__
-#define drm_get_pci_domain(dev) dev->hose->bus->number
+#define drm_get_pci_domain(dev) dev->hose->index
 #else
 #define drm_get_pci_domain(dev) 0
 #endif
@@ -915,9 +921,18 @@ extern int drm_lock(struct inode *inode, struct file *filp,
 		    unsigned int cmd, unsigned long arg);
 extern int drm_unlock(struct inode *inode, struct file *filp,
 		      unsigned int cmd, unsigned long arg);
-extern int drm_lock_take(__volatile__ unsigned int *lock, unsigned int context);
-extern int drm_lock_free(drm_device_t * dev,
-			 __volatile__ unsigned int *lock, unsigned int context);
+extern int drm_lock_take(drm_lock_data_t *lock_data, unsigned int context);
+extern int drm_lock_free(drm_lock_data_t *lock_data, unsigned int context);
+extern void drm_idlelock_take(drm_lock_data_t *lock_data);
+extern void drm_idlelock_release(drm_lock_data_t *lock_data);
+
+/*
+ * These are exported to drivers so that they can implement fencing using
+ * DMA quiscent + idle. DMA quiescent usually requires the hardware lock.
+ */
+
+extern int drm_i_have_hw_lock(struct file *filp);
+extern int drm_kernel_take_hw_lock(struct file *filp);
 
 				/* Buffer management support (drm_bufs.h) */
 extern int drm_addbufs_agp(drm_device_t * dev, drm_buf_desc_t * request);

drivers/char/drm/drm_bufs.c

@@ -57,7 +57,8 @@ static drm_map_list_t *drm_find_matching_map(drm_device_t *dev,
 	list_for_each(list, &dev->maplist->head) {
 		drm_map_list_t *entry = list_entry(list, drm_map_list_t, head);
 		if (entry->map && map->type == entry->map->type &&
-		    entry->map->offset == map->offset) {
+		    ((entry->map->offset == map->offset) ||
+		     (map->type == _DRM_SHM && map->flags==_DRM_CONTAINS_LOCK))) {
 			return entry;
 		}
 	}
@@ -180,8 +181,20 @@ static int drm_addmap_core(drm_device_t * dev, unsigned int offset,
 		if (map->type == _DRM_REGISTERS)
 			map->handle = ioremap(map->offset, map->size);
 		break;
-
 	case _DRM_SHM:
+		list = drm_find_matching_map(dev, map);
+		if (list != NULL) {
+			if(list->map->size != map->size) {
+				DRM_DEBUG("Matching maps of type %d with "
+					  "mismatched sizes, (%ld vs %ld)\n",
+					  map->type, map->size, list->map->size);
+				list->map->size = map->size;
+			}
+
+			drm_free(map, sizeof(*map), DRM_MEM_MAPS);
+			*maplist = list;
+			return 0;
+		}
 		map->handle = vmalloc_user(map->size);
 		DRM_DEBUG("%lu %d %p\n",
 			  map->size, drm_order(map->size), map->handle);
@@ -200,15 +213,45 @@ static int drm_addmap_core(drm_device_t * dev, unsigned int offset,
 			dev->sigdata.lock = dev->lock.hw_lock = map->handle;	/* Pointer to lock */
 		}
 		break;
-	case _DRM_AGP:
-		if (drm_core_has_AGP(dev)) {
+	case _DRM_AGP: {
+		drm_agp_mem_t *entry;
+		int valid = 0;
+
+		if (!drm_core_has_AGP(dev)) {
+			drm_free(map, sizeof(*map), DRM_MEM_MAPS);
+			return -EINVAL;
+		}
 #ifdef __alpha__
-			map->offset += dev->hose->mem_space->start;
+		map->offset += dev->hose->mem_space->start;
 #endif
-			map->offset += dev->agp->base;
-			map->mtrr = dev->agp->agp_mtrr;	/* for getmap */
+		/* Note: dev->agp->base may actually be 0 when the DRM
+		 * is not in control of AGP space. But if user space is
+		 * it should already have added the AGP base itself.
+		 */
+		map->offset += dev->agp->base;
+		map->mtrr = dev->agp->agp_mtrr;	/* for getmap */
+
+		/* This assumes the DRM is in total control of AGP space.
+		 * It's not always the case as AGP can be in the control
+		 * of user space (i.e. i810 driver). So this loop will get
+		 * skipped and we double check that dev->agp->memory is
+		 * actually set as well as being invalid before EPERM'ing
+		 */
+		for (entry = dev->agp->memory; entry; entry = entry->next) {
+			if ((map->offset >= entry->bound) &&
+			    (map->offset + map->size <= entry->bound + entry->pages * PAGE_SIZE)) {
+				valid = 1;
+				break;
+			}
 		}
+		if (dev->agp->memory && !valid) {
+			drm_free(map, sizeof(*map), DRM_MEM_MAPS);
+			return -EPERM;
+		}
+		DRM_DEBUG("AGP offset = 0x%08lx, size = 0x%08lx\n", map->offset, map->size);
+
 		break;
+	}
 	case _DRM_SCATTER_GATHER:
 		if (!dev->sg) {
 			drm_free(map, sizeof(*map), DRM_MEM_MAPS);
@@ -267,7 +310,7 @@ static int drm_addmap_core(drm_device_t * dev, unsigned int offset,
 
 	*maplist = list;
 	return 0;
-}
+	}
 
 int drm_addmap(drm_device_t * dev, unsigned int offset,
 	       unsigned int size, drm_map_type_t type,
@@ -519,6 +562,7 @@ int drm_addbufs_agp(drm_device_t * dev, drm_buf_desc_t * request)
 {
 	drm_device_dma_t *dma = dev->dma;
 	drm_buf_entry_t *entry;
+	drm_agp_mem_t *agp_entry;
 	drm_buf_t *buf;
 	unsigned long offset;
 	unsigned long agp_offset;
@@ -529,7 +573,7 @@ int drm_addbufs_agp(drm_device_t * dev, drm_buf_desc_t * request)
 	int page_order;
 	int total;
 	int byte_count;
-	int i;
+	int i, valid;
 	drm_buf_t **temp_buflist;
 
 	if (!dma)
@@ -560,6 +604,19 @@ int drm_addbufs_agp(drm_device_t * dev, drm_buf_desc_t * request)
 	if (dev->queue_count)
 		return -EBUSY;	/* Not while in use */
 
+	/* Make sure buffers are located in AGP memory that we own */
+	valid = 0;
+	for (agp_entry = dev->agp->memory; agp_entry; agp_entry = agp_entry->next) {
+		if ((agp_offset >= agp_entry->bound) &&
+		    (agp_offset + total * count <= agp_entry->bound + agp_entry->pages * PAGE_SIZE)) {
+			valid = 1;
+			break;
+		}
+	}
+	if (dev->agp->memory && !valid) {
+		DRM_DEBUG("zone invalid\n");
+		return -EINVAL;
+	}
 	spin_lock(&dev->count_lock);
 	if (dev->buf_use) {
 		spin_unlock(&dev->count_lock);

drivers/char/drm/drm_drv.c

@@ -496,11 +496,14 @@ int drm_ioctl(struct inode *inode, struct file *filp,
 		  (long)old_encode_dev(priv->head->device),
 		  priv->authenticated);
 
-	if (nr < DRIVER_IOCTL_COUNT)
-		ioctl = &drm_ioctls[nr];
-	else if ((nr >= DRM_COMMAND_BASE)
+	if ((nr >= DRIVER_IOCTL_COUNT) &&
+	    ((nr < DRM_COMMAND_BASE) || (nr >= DRM_COMMAND_END)))
+		goto err_i1;
+	if ((nr >= DRM_COMMAND_BASE) && (nr < DRM_COMMAND_END)
 		 && (nr < DRM_COMMAND_BASE + dev->driver->num_ioctls))
 		ioctl = &dev->driver->ioctls[nr - DRM_COMMAND_BASE];
+	else if ((nr >= DRM_COMMAND_END) || (nr < DRM_COMMAND_BASE))
+		ioctl = &drm_ioctls[nr];
 	else
 		goto err_i1;
 

drivers/char/drm/drm_fops.c

@@ -46,6 +46,7 @@ static int drm_setup(drm_device_t * dev)
 	drm_local_map_t *map;
 	int i;
 	int ret;
+	u32 sareapage;
 
 	if (dev->driver->firstopen) {
 		ret = dev->driver->firstopen(dev);
@@ -56,7 +57,8 @@ static int drm_setup(drm_device_t * dev)
 	dev->magicfree.next = NULL;
 
 	/* prebuild the SAREA */
-	i = drm_addmap(dev, 0, SAREA_MAX, _DRM_SHM, _DRM_CONTAINS_LOCK, &map);
+	sareapage = max_t(unsigned, SAREA_MAX, PAGE_SIZE);
+	i = drm_addmap(dev, 0, sareapage, _DRM_SHM, _DRM_CONTAINS_LOCK, &map);
 	if (i != 0)
 		return i;
 
@@ -84,7 +86,7 @@ static int drm_setup(drm_device_t * dev)
 	INIT_LIST_HEAD(&dev->ctxlist->head);
 
 	dev->vmalist = NULL;
-	dev->sigdata.lock = dev->lock.hw_lock = NULL;
+	dev->sigdata.lock = NULL;
 	init_waitqueue_head(&dev->lock.lock_queue);
 	dev->queue_count = 0;
 	dev->queue_reserved = 0;
@@ -354,58 +356,56 @@ int drm_release(struct inode *inode, struct file *filp)
 		  current->pid, (long)old_encode_dev(priv->head->device),
 		  dev->open_count);
 
-	if (priv->lock_count && dev->lock.hw_lock &&
-	    _DRM_LOCK_IS_HELD(dev->lock.hw_lock->lock) &&
-	    dev->lock.filp == filp) {
-		DRM_DEBUG("File %p released, freeing lock for context %d\n",
-			  filp, _DRM_LOCKING_CONTEXT(dev->lock.hw_lock->lock));
-
-		if (dev->driver->reclaim_buffers_locked)
+	if (dev->driver->reclaim_buffers_locked && dev->lock.hw_lock) {
+		if (drm_i_have_hw_lock(filp)) {
 			dev->driver->reclaim_buffers_locked(dev, filp);
-
-		drm_lock_free(dev, &dev->lock.hw_lock->lock,
-			      _DRM_LOCKING_CONTEXT(dev->lock.hw_lock->lock));
-
-		/* FIXME: may require heavy-handed reset of
-		   hardware at this point, possibly
-		   processed via a callback to the X
-		   server. */
-	} else if (dev->driver->reclaim_buffers_locked && priv->lock_count
-		   && dev->lock.hw_lock) {
-		/* The lock is required to reclaim buffers */
-		DECLARE_WAITQUEUE(entry, current);
-
-		add_wait_queue(&dev->lock.lock_queue, &entry);
-		for (;;) {
-			__set_current_state(TASK_INTERRUPTIBLE);
-			if (!dev->lock.hw_lock) {
-				/* Device has been unregistered */
-				retcode = -EINTR;
-				break;
+		} else {
+			unsigned long _end=jiffies + 3*DRM_HZ;
+			int locked = 0;
+
+			drm_idlelock_take(&dev->lock);
+
+			/*
+			 * Wait for a while.
+			 */
+
+			do{
+				spin_lock(&dev->lock.spinlock);
+				locked = dev->lock.idle_has_lock;
+				spin_unlock(&dev->lock.spinlock);
+				if (locked)
+					break;
+				schedule();
+			} while (!time_after_eq(jiffies, _end));
+
+			if (!locked) {
+				DRM_ERROR("reclaim_buffers_locked() deadlock. Please rework this\n"
+					  "\tdriver to use reclaim_buffers_idlelocked() instead.\n"
+					  "\tI will go on reclaiming the buffers anyway.\n");
 			}
-			if (drm_lock_take(&dev->lock.hw_lock->lock,
-					  DRM_KERNEL_CONTEXT)) {
-				dev->lock.filp = filp;
-				dev->lock.lock_time = jiffies;
-				atomic_inc(&dev->counts[_DRM_STAT_LOCKS]);
-				break;	/* Got lock */
-			}
-			/* Contention */
-			schedule();
-			if (signal_pending(current)) {
-				retcode = -ERESTARTSYS;
-				break;
-			}
-		}
-		__set_current_state(TASK_RUNNING);
-		remove_wait_queue(&dev->lock.lock_queue, &entry);
-		if (!retcode) {
+
 			dev->driver->reclaim_buffers_locked(dev, filp);
-			drm_lock_free(dev, &dev->lock.hw_lock->lock,
-				      DRM_KERNEL_CONTEXT);
+			drm_idlelock_release(&dev->lock);
 		}
 	}
 
+	if (dev->driver->reclaim_buffers_idlelocked && dev->lock.hw_lock) {
+
+		drm_idlelock_take(&dev->lock);
+		dev->driver->reclaim_buffers_idlelocked(dev, filp);
+		drm_idlelock_release(&dev->lock);
+
+	}
+
+	if (drm_i_have_hw_lock(filp)) {
+		DRM_DEBUG("File %p released, freeing lock for context %d\n",
+			  filp, _DRM_LOCKING_CONTEXT(dev->lock.hw_lock->lock));
+
+		drm_lock_free(&dev->lock,
+			      _DRM_LOCKING_CONTEXT(dev->lock.hw_lock->lock));
+	}
+
+
 	if (drm_core_check_feature(dev, DRIVER_HAVE_DMA) &&
 	    !dev->driver->reclaim_buffers_locked) {
 		dev->driver->reclaim_buffers(dev, filp);

drivers/char/drm/drm_hashtab.c

@@ -43,7 +43,16 @@ int drm_ht_create(drm_open_hash_t *ht, unsigned int order)
 	ht->size = 1 << order;
 	ht->order = order;
 	ht->fill = 0;
-	ht->table = vmalloc(ht->size*sizeof(*ht->table));
+	ht->table = NULL;
+	ht->use_vmalloc = ((ht->size * sizeof(*ht->table)) > PAGE_SIZE);
+	if (!ht->use_vmalloc) {
+		ht->table = drm_calloc(ht->size, sizeof(*ht->table),
+				       DRM_MEM_HASHTAB);
+	}
+	if (!ht->table) {
+		ht->use_vmalloc = 1;
+		ht->table = vmalloc(ht->size*sizeof(*ht->table));
+	}
 	if (!ht->table) {
 		DRM_ERROR("Out of memory for hash table\n");
 		return -ENOMEM;
@@ -183,7 +192,11 @@ int drm_ht_remove_item(drm_open_hash_t *ht, drm_hash_item_t *item)
 void drm_ht_remove(drm_open_hash_t *ht)
 {
 	if (ht->table) {
-		vfree(ht->table);
+		if (ht->use_vmalloc)
+			vfree(ht->table);
+		else
+			drm_free(ht->table, ht->size * sizeof(*ht->table),
+				 DRM_MEM_HASHTAB);
 		ht->table = NULL;
 	}
 }

drivers/char/drm/drm_hashtab.h

@@ -47,6 +47,7 @@ typedef struct drm_open_hash{
 	unsigned int order;
 	unsigned int fill;
 	struct hlist_head *table;
+	int use_vmalloc;
 } drm_open_hash_t;
 
 

drivers/char/drm/drm_irq.c

@@ -424,7 +424,7 @@ static void drm_locked_tasklet_func(unsigned long data)
 	spin_lock_irqsave(&dev->tasklet_lock, irqflags);
 
 	if (!dev->locked_tasklet_func ||
-	    !drm_lock_take(&dev->lock.hw_lock->lock,
+	    !drm_lock_take(&dev->lock,
 			   DRM_KERNEL_CONTEXT)) {
 		spin_unlock_irqrestore(&dev->tasklet_lock, irqflags);
 		return;
@@ -435,7 +435,7 @@ static void drm_locked_tasklet_func(unsigned long data)
 
 	dev->locked_tasklet_func(dev);
 
-	drm_lock_free(dev, &dev->lock.hw_lock->lock,
+	drm_lock_free(&dev->lock,
 		      DRM_KERNEL_CONTEXT);
 
 	dev->locked_tasklet_func = NULL;

drivers/char/drm/drm_lock.c

@@ -35,9 +35,6 @@
 
 #include "drmP.h"
 
-static int drm_lock_transfer(drm_device_t * dev,
-			     __volatile__ unsigned int *lock,
-			     unsigned int context);
 static int drm_notifier(void *priv);
 
 /**
@@ -80,6 +77,9 @@ int drm_lock(struct inode *inode, struct file *filp,
 			return -EINVAL;
 
 	add_wait_queue(&dev->lock.lock_queue, &entry);
+	spin_lock(&dev->lock.spinlock);
+	dev->lock.user_waiters++;
+	spin_unlock(&dev->lock.spinlock);
 	for (;;) {
 		__set_current_state(TASK_INTERRUPTIBLE);
 		if (!dev->lock.hw_lock) {
@@ -87,7 +87,7 @@ int drm_lock(struct inode *inode, struct file *filp,
 			ret = -EINTR;
 			break;
 		}
-		if (drm_lock_take(&dev->lock.hw_lock->lock, lock.context)) {
+		if (drm_lock_take(&dev->lock, lock.context)) {
 			dev->lock.filp = filp;
 			dev->lock.lock_time = jiffies;
 			atomic_inc(&dev->counts[_DRM_STAT_LOCKS]);
@@ -101,12 +101,14 @@ int drm_lock(struct inode *inode, struct file *filp,
 			break;
 		}
 	}
+	spin_lock(&dev->lock.spinlock);
+	dev->lock.user_waiters--;
+	spin_unlock(&dev->lock.spinlock);
 	__set_current_state(TASK_RUNNING);
 	remove_wait_queue(&dev->lock.lock_queue, &entry);
 
-	DRM_DEBUG("%d %s\n", lock.context, ret ? "interrupted" : "has lock");
-	if (ret)
-		return ret;
+	DRM_DEBUG( "%d %s\n", lock.context, ret ? "interrupted" : "has lock" );
+	if (ret) return ret;
 
 	sigemptyset(&dev->sigmask);
 	sigaddset(&dev->sigmask, SIGSTOP);
@@ -127,14 +129,12 @@ int drm_lock(struct inode *inode, struct file *filp,
 		}
 	}
 
-	/* dev->driver->kernel_context_switch isn't used by any of the x86
-	 *  drivers but is used by the Sparc driver.
-	 */
 	if (dev->driver->kernel_context_switch &&
 	    dev->last_context != lock.context) {
 		dev->driver->kernel_context_switch(dev, dev->last_context,
 						   lock.context);
 	}
+
 	return 0;
 }
 
@@ -184,12 +184,8 @@ int drm_unlock(struct inode *inode, struct file *filp,
 	if (dev->driver->kernel_context_switch_unlock)
 		dev->driver->kernel_context_switch_unlock(dev);
 	else {
-		drm_lock_transfer(dev, &dev->lock.hw_lock->lock,
-				  DRM_KERNEL_CONTEXT);
-
-		if (drm_lock_free(dev, &dev->lock.hw_lock->lock,
-				  DRM_KERNEL_CONTEXT)) {
-			DRM_ERROR("\n");
+		if (drm_lock_free(&dev->lock,lock.context)) {
+			/* FIXME: Should really bail out here. */
 		}
 	}
 
@@ -206,18 +202,26 @@ int drm_unlock(struct inode *inode, struct file *filp,
  *
  * Attempt to mark the lock as held by the given context, via the \p cmpxchg instruction.
  */
-int drm_lock_take(__volatile__ unsigned int *lock, unsigned int context)
+int drm_lock_take(drm_lock_data_t *lock_data,
+		  unsigned int context)
 {
 	unsigned int old, new, prev;
+	volatile unsigned int *lock = &lock_data->hw_lock->lock;
 
+	spin_lock(&lock_data->spinlock);
 	do {
 		old = *lock;
 		if (old & _DRM_LOCK_HELD)
 			new = old | _DRM_LOCK_CONT;
-		else
-			new = context | _DRM_LOCK_HELD;
+		else {
+			new = context | _DRM_LOCK_HELD |
+				((lock_data->user_waiters + lock_data->kernel_waiters > 1) ?
+				 _DRM_LOCK_CONT : 0);
+		}
 		prev = cmpxchg(lock, old, new);
 	} while (prev != old);
+	spin_unlock(&lock_data->spinlock);
+
 	if (_DRM_LOCKING_CONTEXT(old) == context) {
 		if (old & _DRM_LOCK_HELD) {
 			if (context != DRM_KERNEL_CONTEXT) {
@@ -227,7 +231,8 @@ int drm_lock_take(__volatile__ unsigned int *lock, unsigned int context)
 			return 0;
 		}
 	}
-	if (new == (context | _DRM_LOCK_HELD)) {
+
+	if ((_DRM_LOCKING_CONTEXT(new)) == context && (new & _DRM_LOCK_HELD)) {
 		/* Have lock */
 		return 1;
 	}
@@ -246,13 +251,13 @@ int drm_lock_take(__volatile__ unsigned int *lock, unsigned int context)
  * Resets the lock file pointer.
  * Marks the lock as held by the given context, via the \p cmpxchg instruction.
  */
-static int drm_lock_transfer(drm_device_t * dev,
-			     __volatile__ unsigned int *lock,
+static int drm_lock_transfer(drm_lock_data_t *lock_data,
 			     unsigned int context)
 {
 	unsigned int old, new, prev;
+	volatile unsigned int *lock = &lock_data->hw_lock->lock;
 
-	dev->lock.filp = NULL;
+	lock_data->filp = NULL;
 	do {
 		old = *lock;
 		new = context | _DRM_LOCK_HELD;
@@ -272,23 +277,32 @@ static int drm_lock_transfer(drm_device_t * dev,
  * Marks the lock as not held, via the \p cmpxchg instruction. Wakes any task
  * waiting on the lock queue.
  */
-int drm_lock_free(drm_device_t * dev,
-		  __volatile__ unsigned int *lock, unsigned int context)
+int drm_lock_free(drm_lock_data_t *lock_data, unsigned int context)
 {
 	unsigned int old, new, prev;
+	volatile unsigned int *lock = &lock_data->hw_lock->lock;
+
+	spin_lock(&lock_data->spinlock);
+	if (lock_data->kernel_waiters != 0) {
+		drm_lock_transfer(lock_data, 0);
+		lock_data->idle_has_lock = 1;
+		spin_unlock(&lock_data->spinlock);
+		return 1;
+	}
+	spin_unlock(&lock_data->spinlock);
 
-	dev->lock.filp = NULL;
 	do {
 		old = *lock;
-		new = 0;
+		new = _DRM_LOCKING_CONTEXT(old);
 		prev = cmpxchg(lock, old, new);
 	} while (prev != old);
+
 	if (_DRM_LOCK_IS_HELD(old) && _DRM_LOCKING_CONTEXT(old) != context) {
 		DRM_ERROR("%d freed heavyweight lock held by %d\n",
 			  context, _DRM_LOCKING_CONTEXT(old));
 		return 1;
 	}
-	wake_up_interruptible(&dev->lock.lock_queue);
+	wake_up_interruptible(&lock_data->lock_queue);
 	return 0;
 }
 
@@ -322,3 +336,67 @@ static int drm_notifier(void *priv)
 	} while (prev != old);
 	return 0;
 }
+
+/**
+ * This function returns immediately and takes the hw lock
+ * with the kernel context if it is free, otherwise it gets the highest priority when and if
+ * it is eventually released.
+ *
+ * This guarantees that the kernel will _eventually_ have the lock _unless_ it is held
+ * by a blocked process. (In the latter case an explicit wait for the hardware lock would cause
+ * a deadlock, which is why the "idlelock" was invented).
+ *
+ * This should be sufficient to wait for GPU idle without
+ * having to worry about starvation.
+ */
+
+void drm_idlelock_take(drm_lock_data_t *lock_data)
+{
+	int ret = 0;
+
+	spin_lock(&lock_data->spinlock);
+	lock_data->kernel_waiters++;
+	if (!lock_data->idle_has_lock) {
+
+		spin_unlock(&lock_data->spinlock);
+		ret = drm_lock_take(lock_data, DRM_KERNEL_CONTEXT);
+		spin_lock(&lock_data->spinlock);
+
+		if (ret == 1)
+			lock_data->idle_has_lock = 1;
+	}
+	spin_unlock(&lock_data->spinlock);
+}
+EXPORT_SYMBOL(drm_idlelock_take);
+
+void drm_idlelock_release(drm_lock_data_t *lock_data)
+{
+	unsigned int old, prev;
+	volatile unsigned int *lock = &lock_data->hw_lock->lock;
+
+	spin_lock(&lock_data->spinlock);
+	if (--lock_data->kernel_waiters == 0) {
+		if (lock_data->idle_has_lock) {
+			do {
+				old = *lock;
+				prev = cmpxchg(lock, old, DRM_KERNEL_CONTEXT);
+			} while (prev != old);
+			wake_up_interruptible(&lock_data->lock_queue);
+			lock_data->idle_has_lock = 0;
+		}
+	}
+	spin_unlock(&lock_data->spinlock);
+}
+EXPORT_SYMBOL(drm_idlelock_release);
+
+
+int drm_i_have_hw_lock(struct file *filp)
+{
+	DRM_DEVICE;
+
+	return (priv->lock_count && dev->lock.hw_lock &&
+		_DRM_LOCK_IS_HELD(dev->lock.hw_lock->lock) &&
+		dev->lock.filp == filp);
+}
+
+EXPORT_SYMBOL(drm_i_have_hw_lock);

drivers/char/drm/drm_mm.c

@@ -274,7 +274,6 @@ int drm_mm_init(drm_mm_t * mm, unsigned long start, unsigned long size)
 	return drm_mm_create_tail_node(mm, start, size);
 }
 
-EXPORT_SYMBOL(drm_mm_init);
 
 void drm_mm_takedown(drm_mm_t * mm)
 {
@@ -295,4 +294,3 @@ void drm_mm_takedown(drm_mm_t * mm)
 	drm_free(entry, sizeof(*entry), DRM_MEM_MM);
 }
 
-EXPORT_SYMBOL(drm_mm_takedown);

drivers/char/drm/drm_pciids.h

@@ -230,10 +230,10 @@
 	{0x1106, 0x7205, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
 	{0x1106, 0x3108, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
 	{0x1106, 0x3304, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
-	{0x1106, 0x3157, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
 	{0x1106, 0x3344, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
 	{0x1106, 0x3343, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
 	{0x1106, 0x3230, PCI_ANY_ID, PCI_ANY_ID, 0, 0, VIA_DX9_0}, \
+	{0x1106, 0x3157, PCI_ANY_ID, PCI_ANY_ID, 0, 0, VIA_PRO_GROUP_A}, \
 	{0, 0, 0}
 
 #define i810_PCI_IDS \
@@ -296,5 +296,6 @@
 	{0x8086, 0x2982, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
 	{0x8086, 0x2992, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
 	{0x8086, 0x29a2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
+	{0x8086, 0x2a02, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, \
 	{0, 0, 0}
 

drivers/char/drm/drm_proc.c

@@ -72,7 +72,7 @@ static struct drm_proc_list {
 #endif
 };
 
-#define DRM_PROC_ENTRIES (sizeof(drm_proc_list)/sizeof(drm_proc_list[0]))
+#define DRM_PROC_ENTRIES ARRAY_SIZE(drm_proc_list)
 
 /**
  * Initialize the DRI proc filesystem for a device.

drivers/char/drm/drm_stub.c

@@ -62,6 +62,7 @@ static int drm_fill_in_dev(drm_device_t * 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);
 	mutex_init(&dev->ctxlist_mutex);

drivers/char/drm/drm_vm.c

@@ -41,6 +41,30 @@
 static void drm_vm_open(struct vm_area_struct *vma);
 static void drm_vm_close(struct vm_area_struct *vma);
 
+static pgprot_t drm_io_prot(uint32_t map_type, struct vm_area_struct *vma)
+{
+	pgprot_t tmp = vm_get_page_prot(vma->vm_flags);
+
+#if defined(__i386__) || defined(__x86_64__)
+	if (boot_cpu_data.x86 > 3 && map_type != _DRM_AGP) {
+		pgprot_val(tmp) |= _PAGE_PCD;
+		pgprot_val(tmp) &= ~_PAGE_PWT;
+	}
+#elif defined(__powerpc__)
+	pgprot_val(tmp) |= _PAGE_NO_CACHE;
+	if (map_type == _DRM_REGISTERS)
+		pgprot_val(tmp) |= _PAGE_GUARDED;
+#endif
+#if defined(__ia64__)
+	if (efi_range_is_wc(vma->vm_start, vma->vm_end -
+				    vma->vm_start))
+		tmp = pgprot_writecombine(tmp);
+	else
+		tmp = pgprot_noncached(tmp);
+#endif
+	return tmp;
+}
+
 /**
  * \c nopage method for AGP virtual memory.
  *
@@ -151,8 +175,7 @@ static __inline__ struct page *drm_do_vm_shm_nopage(struct vm_area_struct *vma,
 
 	offset = address - vma->vm_start;
 	i = (unsigned long)map->handle + offset;
-	page = (map->type == _DRM_CONSISTENT) ?
-		virt_to_page((void *)i) : vmalloc_to_page((void *)i);
+	page = vmalloc_to_page((void *)i);
 	if (!page)
 		return NOPAGE_SIGBUS;
 	get_page(page);
@@ -389,7 +412,7 @@ static struct vm_operations_struct drm_vm_sg_ops = {
  * Create a new drm_vma_entry structure as the \p vma private data entry and
  * add it to drm_device::vmalist.
  */
-static void drm_vm_open(struct vm_area_struct *vma)
+static void drm_vm_open_locked(struct vm_area_struct *vma)
 {
 	drm_file_t *priv = vma->vm_file->private_data;
 	drm_device_t *dev = priv->head->dev;
@@ -401,15 +424,23 @@ static void drm_vm_open(struct vm_area_struct *vma)
 
 	vma_entry = drm_alloc(sizeof(*vma_entry), DRM_MEM_VMAS);
 	if (vma_entry) {
-		mutex_lock(&dev->struct_mutex);
 		vma_entry->vma = vma;
 		vma_entry->next = dev->vmalist;
 		vma_entry->pid = current->pid;
 		dev->vmalist = vma_entry;
-		mutex_unlock(&dev->struct_mutex);
 	}
 }
 
+static void drm_vm_open(struct vm_area_struct *vma)
+{
+	drm_file_t *priv = vma->vm_file->private_data;
+	drm_device_t *dev = priv->head->dev;
+
+	mutex_lock(&dev->struct_mutex);
+	drm_vm_open_locked(vma);
+	mutex_unlock(&dev->struct_mutex);
+}
+
 /**
  * \c close method for all virtual memory types.
  *
@@ -460,7 +491,6 @@ static int drm_mmap_dma(struct file *filp, struct vm_area_struct *vma)
 	drm_device_dma_t *dma;
 	unsigned long length = vma->vm_end - vma->vm_start;
 
-	lock_kernel();
 	dev = priv->head->dev;
 	dma = dev->dma;
 	DRM_DEBUG("start = 0x%lx, end = 0x%lx, page offset = 0x%lx\n",
@@ -468,10 +498,8 @@ static int drm_mmap_dma(struct file *filp, struct vm_area_struct *vma)
 
 	/* Length must match exact page count */
 	if (!dma || (length >> PAGE_SHIFT) != dma->page_count) {
-		unlock_kernel();
 		return -EINVAL;
 	}
-	unlock_kernel();
 
 	if (!capable(CAP_SYS_ADMIN) &&
 	    (dma->flags & _DRM_DMA_USE_PCI_RO)) {
@@ -494,7 +522,7 @@ static int drm_mmap_dma(struct file *filp, struct vm_area_struct *vma)
 	vma->vm_flags |= VM_RESERVED;	/* Don't swap */
 
 	vma->vm_file = filp;	/* Needed for drm_vm_open() */
-	drm_vm_open(vma);
+	drm_vm_open_locked(vma);
 	return 0;
 }
 
@@ -529,7 +557,7 @@ EXPORT_SYMBOL(drm_core_get_reg_ofs);
  * according to the mapping type and remaps the pages. Finally sets the file
  * pointer and calls vm_open().
  */
-int drm_mmap(struct file *filp, struct vm_area_struct *vma)
+static int drm_mmap_locked(struct file *filp, struct vm_area_struct *vma)
 {
 	drm_file_t *priv = filp->private_data;
 	drm_device_t *dev = priv->head->dev;
@@ -565,7 +593,7 @@ int drm_mmap(struct file *filp, struct vm_area_struct *vma)
 		return -EPERM;
 
 	/* Check for valid size. */
-	if (map->size != vma->vm_end - vma->vm_start)
+	if (map->size < vma->vm_end - vma->vm_start)
 		return -EINVAL;
 
 	if (!capable(CAP_SYS_ADMIN) && (map->flags & _DRM_READ_ONLY)) {
@@ -600,37 +628,16 @@ int drm_mmap(struct file *filp, struct vm_area_struct *vma)
 		/* fall through to _DRM_FRAME_BUFFER... */
 	case _DRM_FRAME_BUFFER:
 	case _DRM_REGISTERS:
-#if defined(__i386__) || defined(__x86_64__)
-		if (boot_cpu_data.x86 > 3 && map->type != _DRM_AGP) {
-			pgprot_val(vma->vm_page_prot) |= _PAGE_PCD;
-			pgprot_val(vma->vm_page_prot) &= ~_PAGE_PWT;
-		}
-#elif defined(__powerpc__)
-		pgprot_val(vma->vm_page_prot) |= _PAGE_NO_CACHE;
-		if (map->type == _DRM_REGISTERS)
-			pgprot_val(vma->vm_page_prot) |= _PAGE_GUARDED;
-#endif
-		vma->vm_flags |= VM_IO;	/* not in core dump */
-#if defined(__ia64__)
-		if (efi_range_is_wc(vma->vm_start, vma->vm_end - vma->vm_start))
-			vma->vm_page_prot =
-			    pgprot_writecombine(vma->vm_page_prot);
-		else
-			vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
-#endif
 		offset = dev->driver->get_reg_ofs(dev);
+		vma->vm_flags |= VM_IO;	/* not in core dump */
+		vma->vm_page_prot = drm_io_prot(map->type, vma);
 #ifdef __sparc__
 		vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+#endif
 		if (io_remap_pfn_range(vma, vma->vm_start,
 				       (map->offset + offset) >> PAGE_SHIFT,
 				       vma->vm_end - vma->vm_start,
 				       vma->vm_page_prot))
-#else
-		if (io_remap_pfn_range(vma, vma->vm_start,
-				       (map->offset + offset) >> PAGE_SHIFT,
-				       vma->vm_end - vma->vm_start,
-				       vma->vm_page_prot))
-#endif
 			return -EAGAIN;
 		DRM_DEBUG("   Type = %d; start = 0x%lx, end = 0x%lx,"
 			  " offset = 0x%lx\n",
@@ -638,10 +645,15 @@ int drm_mmap(struct file *filp, struct vm_area_struct *vma)
 			  vma->vm_start, vma->vm_end, map->offset + offset);
 		vma->vm_ops = &drm_vm_ops;
 		break;
-	case _DRM_SHM:
 	case _DRM_CONSISTENT:
-		/* Consistent memory is really like shared memory. It's only
-		 * allocate in a different way */
+		/* Consistent memory is really like shared memory. But
+		 * it's allocated in a different way, so avoid nopage */
+		if (remap_pfn_range(vma, vma->vm_start,
+		    page_to_pfn(virt_to_page(map->handle)),
+		    vma->vm_end - vma->vm_start, vma->vm_page_prot))
+			return -EAGAIN;
+	/* fall through to _DRM_SHM */
+	case _DRM_SHM:
 		vma->vm_ops = &drm_vm_shm_ops;
 		vma->vm_private_data = (void *)map;
 		/* Don't let this area swap.  Change when
@@ -659,8 +671,20 @@ int drm_mmap(struct file *filp, struct vm_area_struct *vma)
 	vma->vm_flags |= VM_RESERVED;	/* Don't swap */
 
 	vma->vm_file = filp;	/* Needed for drm_vm_open() */
-	drm_vm_open(vma);
+	drm_vm_open_locked(vma);
 	return 0;
 }
 
+int drm_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+	drm_file_t *priv = filp->private_data;
+	drm_device_t *dev = priv->head->dev;
+	int ret;
+
+	mutex_lock(&dev->struct_mutex);
+	ret = drm_mmap_locked(filp, vma);
+	mutex_unlock(&dev->struct_mutex);
+
+	return ret;
+}
 EXPORT_SYMBOL(drm_mmap);

drivers/char/drm/i915_dma.c

@@ -34,7 +34,8 @@
 #define IS_I965G(dev) (dev->pci_device == 0x2972 || \
 		       dev->pci_device == 0x2982 || \
 		       dev->pci_device == 0x2992 || \
-		       dev->pci_device == 0x29A2)
+		       dev->pci_device == 0x29A2 || \
+		       dev->pci_device == 0x2A02)
 
 /* Really want an OS-independent resettable timer.  Would like to have
  * this loop run for (eg) 3 sec, but have the timer reset every time

drivers/char/drm/radeon_cp.c

@@ -1560,8 +1560,8 @@ static int radeon_do_init_cp(drm_device_t * dev, drm_radeon_init_t * init)
 		if (dev_priv->flags & RADEON_IS_AGP) {
 			base = dev->agp->base;
 			/* Check if valid */
-			if ((base + dev_priv->gart_size) > dev_priv->fb_location &&
-			    base < (dev_priv->fb_location + dev_priv->fb_size)) {
+			if ((base + dev_priv->gart_size - 1) >= dev_priv->fb_location &&
+			    base < (dev_priv->fb_location + dev_priv->fb_size - 1)) {
 				DRM_INFO("Can't use AGP base @0x%08lx, won't fit\n",
 					 dev->agp->base);
 				base = 0;
@@ -1571,8 +1571,8 @@ static int radeon_do_init_cp(drm_device_t * dev, drm_radeon_init_t * init)
 		/* If not or if AGP is at 0 (Macs), try to put it elsewhere */
 		if (base == 0) {
 			base = dev_priv->fb_location + dev_priv->fb_size;
-			if (((base + dev_priv->gart_size) & 0xfffffffful)
-			    < base)
+			if (base < dev_priv->fb_location ||
+			    ((base + dev_priv->gart_size) & 0xfffffffful) < base)
 				base = dev_priv->fb_location
 					- dev_priv->gart_size;
 		}		

drivers/char/drm/sis_drv.c

@@ -71,7 +71,7 @@ static struct drm_driver driver = {
 	.context_dtor = NULL,
 	.dma_quiescent = sis_idle,
 	.reclaim_buffers = NULL,
-	.reclaim_buffers_locked = sis_reclaim_buffers_locked,
+	.reclaim_buffers_idlelocked = sis_reclaim_buffers_locked,
 	.lastclose = sis_lastclose,
 	.get_map_ofs = drm_core_get_map_ofs,
 	.get_reg_ofs = drm_core_get_reg_ofs,

drivers/char/drm/via_drv.c

@@ -52,7 +52,8 @@ static struct drm_driver driver = {
 	.dma_quiescent = via_driver_dma_quiescent,
 	.dri_library_name = dri_library_name,
 	.reclaim_buffers = drm_core_reclaim_buffers,
-	.reclaim_buffers_locked = via_reclaim_buffers_locked,
+	.reclaim_buffers_locked = NULL,
+	.reclaim_buffers_idlelocked = via_reclaim_buffers_locked,
 	.lastclose = via_lastclose,
 	.get_map_ofs = drm_core_get_map_ofs,
 	.get_reg_ofs = drm_core_get_reg_ofs,

drivers/char/drm/via_mm.h

@@ -1,40 +0,0 @@
-/*
- * Copyright 1998-2003 VIA Technologies, Inc. All Rights Reserved.
- * Copyright 2001-2003 S3 Graphics, Inc. All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sub license,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * VIA, S3 GRAPHICS, AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
- * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
- * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
- * DEALINGS IN THE SOFTWARE.
- */
-#ifndef _via_drm_mm_h_
-#define _via_drm_mm_h_
-
-typedef struct {
-	unsigned int context;
-	unsigned int size;
-	unsigned long offset;
-	unsigned long free;
-} drm_via_mm_t;
-
-typedef struct {
-	unsigned int size;
-	unsigned long handle;
-	void *virtual;
-} drm_via_dma_t;
-
-#endif