- Merge from trunk into branch IIsarea-1-0-0-20010314sarea-1-0-0-branch

15 files changed, 93 insertions, 148 deletions

diff --git a/libdrm/xf86drmHash.c b/libdrm/xf86drmHash.c
index 24b698cce..70240b950 100644
--- a/libdrm/xf86drmHash.c
+++ b/libdrm/xf86drmHash.c
@@ -10,11 +10,11 @@
  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
  * 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 NONINFRINGEMENT.  IN NO EVENT SHALL
@@ -22,7 +22,7 @@
  * 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.
- * 
+ *
  * Authors: Rickard E. (Rik) Faith <faith@valinux.com>
  *
  * $XFree86: xc/programs/Xserver/hw/xfree86/os-support/linux/drm/xf86drmHash.c,v 1.3 2000/06/17 00:03:34 martin Exp $
@@ -187,7 +187,7 @@ int N(HashDestroy)(void *t)
     int           i;
 
     if (table->magic != HASH_MAGIC) return -1; /* Bad magic */
-    
+
     for (i = 0; i < HASH_SIZE; i++) {
 	for (bucket = table->buckets[i]; bucket;) {
 	    next = bucket->next;
@@ -235,8 +235,8 @@ int N(HashLookup)(void *t, unsigned long key, void **value)
     HashTablePtr  table = (HashTablePtr)t;
     HashBucketPtr bucket;
 
-    if (table->magic != HASH_MAGIC) return -1; /* Bad magic */
-    
+    if (!table || table->magic != HASH_MAGIC) return -1; /* Bad magic */
+
     bucket = HashFind(table, key, NULL);
     if (!bucket) return 1;	/* Not found */
     *value = bucket->value;
@@ -250,7 +250,7 @@ int N(HashInsert)(void *t, unsigned long key, void *value)
     unsigned long hash;
 
     if (table->magic != HASH_MAGIC) return -1; /* Bad magic */
-    
+
     if (HashFind(table, key, &hash)) return 1; /* Already in table */
 
     bucket               = HASH_ALLOC(sizeof(*bucket));
@@ -272,7 +272,7 @@ int N(HashDelete)(void *t, unsigned long key)
     HashBucketPtr bucket;
 
     if (table->magic != HASH_MAGIC) return -1; /* Bad magic */
-    
+
     bucket = HashFind(table, key, &hash);
 
     if (!bucket) return 1;	/* Not found */
@@ -285,7 +285,7 @@ int N(HashDelete)(void *t, unsigned long key)
 int N(HashNext)(void *t, unsigned long *key, void **value)
 {
     HashTablePtr  table = (HashTablePtr)t;
-    
+
     for (; table->p0 < HASH_SIZE;
 	 ++table->p0, table->p1 = table->buckets[table->p0]) {
 	if (table->p1) {
@@ -301,7 +301,7 @@ int N(HashNext)(void *t, unsigned long *key, void **value)
 int N(HashFirst)(void *t, unsigned long *key, void **value)
 {
     HashTablePtr  table = (HashTablePtr)t;
-    
+
     if (table->magic != HASH_MAGIC) return -1; /* Bad magic */
 
     table->p0 = 0;
@@ -337,7 +337,7 @@ static void compute_dist(HashTablePtr table)
 {
     int           i;
     HashBucketPtr bucket;
-    
+
     printf("Entries = %ld, hits = %ld, partials = %ld, misses = %ld\n",
 	   table->entries, table->hits, table->partials, table->misses);
     clear_dist();
@@ -356,7 +356,7 @@ static void check_table(HashTablePtr table,
 {
     unsigned long retval  = 0;
     int           retcode = N(HashLookup)(table, key, &retval);
-    
+
     switch (retcode) {
     case -1:
 	printf("Bad magic = 0x%08lx:"
@@ -391,7 +391,7 @@ int main(void)
     for (i = 256; i >= 0; i--) check_table(table, i, i);
     compute_dist(table);
     N(HashDestroy)(table);
-    
+
     printf("\n***** 1024 consecutive integers ****\n");
     table = N(HashCreate)();
     for (i = 0; i < 1024; i++) N(HashInsert)(table, i, i);
@@ -399,7 +399,7 @@ int main(void)
     for (i = 1024; i >= 0; i--) check_table(table, i, i);
     compute_dist(table);
     N(HashDestroy)(table);
-    
+
     printf("\n***** 1024 consecutive page addresses (4k pages) ****\n");
     table = N(HashCreate)();
     for (i = 0; i < 1024; i++) N(HashInsert)(table, i*4096, i);
@@ -407,7 +407,7 @@ int main(void)
     for (i = 1024; i >= 0; i--) check_table(table, i*4096, i);
     compute_dist(table);
     N(HashDestroy)(table);
-    
+
     printf("\n***** 1024 random integers ****\n");
     table = N(HashCreate)();
     srandom(0xbeefbeef);
@@ -429,7 +429,7 @@ int main(void)
     for (i = 0; i < 5000; i++) check_table(table, random(), i);
     compute_dist(table);
     N(HashDestroy)(table);
-    
+
     return 0;
 }
 #endif
diff --git a/linux-core/r128_drv.c b/linux-core/r128_drv.c
index e42868ed6..daae0f875 100644
--- a/linux-core/r128_drv.c
+++ b/linux-core/r128_drv.c
@@ -38,11 +38,11 @@
 
 #define DRIVER_NAME		"r128"
 #define DRIVER_DESC		"ATI Rage 128"
-#define DRIVER_DATE		"20010216"
+#define DRIVER_DATE		"20010308"
 
 #define DRIVER_MAJOR		2
 #define DRIVER_MINOR		1
-#define DRIVER_PATCHLEVEL	4
+#define DRIVER_PATCHLEVEL	5
 
 #define DRIVER_IOCTLS							    \
    [DRM_IOCTL_NR(DRM_IOCTL_DMA)]             = { r128_cce_buffers,  1, 0 }, \
diff --git a/linux-core/radeon_drv.c b/linux-core/radeon_drv.c
index 1aa889aee..dba2037f9 100644
--- a/linux-core/radeon_drv.c
+++ b/linux-core/radeon_drv.c
@@ -36,11 +36,11 @@
 
 #define DRIVER_NAME		"radeon"
 #define DRIVER_DESC		"ATI Radeon"
-#define DRIVER_DATE		"20010305"
+#define DRIVER_DATE		"20010308"
 
 #define DRIVER_MAJOR		1
 #define DRIVER_MINOR		0
-#define DRIVER_PATCHLEVEL	0
+#define DRIVER_PATCHLEVEL	1
 
 #define DRIVER_IOCTLS							     \
  [DRM_IOCTL_NR(DRM_IOCTL_DMA)]               = { radeon_cp_buffers,  1, 0 }, \
diff --git a/linux/mga_dma.c b/linux/mga_dma.c
index 21e14f376..ec80fb417 100644
--- a/linux/mga_dma.c
+++ b/linux/mga_dma.c
@@ -178,7 +178,7 @@ void mga_do_dma_flush( drm_mga_private_t *dev_priv )
 
 	primary->last_flush = primary->tail;
 
-	head = *primary->head;
+	head = MGA_READ( MGA_PRIMADDRESS );
 
 	if ( head <= tail ) {
 		primary->space = primary->size - primary->tail;
@@ -218,7 +218,7 @@ void mga_do_dma_wrap_start( drm_mga_private_t *dev_priv )
 	primary->last_flush = 0;
 	primary->last_wrap++;
 
-	head = *primary->head;
+	head = MGA_READ( MGA_PRIMADDRESS );
 
 	if ( head == dev_priv->primary->offset ) {
 		primary->space = primary->size;
@@ -240,7 +240,6 @@ void mga_do_dma_wrap_start( drm_mga_private_t *dev_priv )
 
 void mga_do_dma_wrap_end( drm_mga_private_t *dev_priv )
 {
-	drm_mga_primary_buffer_t *primary = &dev_priv->prim;
 	drm_mga_sarea_t *sarea_priv = dev_priv->sarea_priv;
 	u32 head = dev_priv->primary->offset;
 	DRM_DEBUG( "%s:\n", __FUNCTION__ );
@@ -248,8 +247,6 @@ void mga_do_dma_wrap_end( drm_mga_private_t *dev_priv )
 	sarea_priv->last_wrap++;
 	DRM_DEBUG( "   wrap = %d\n", sarea_priv->last_wrap );
 
-	*primary->head = head;
-
 	mga_flush_write_combine();
 	MGA_WRITE( MGA_PRIMADDRESS, head | MGA_DMA_GENERAL );
 
@@ -272,7 +269,7 @@ static void mga_freelist_print( drm_device_t *dev )
 	DRM_INFO( "\n" );
 	DRM_INFO( "current dispatch: last=0x%x done=0x%x\n",
 		  dev_priv->sarea_priv->last_dispatch,
-		  (unsigned int)(*dev_priv->prim.head - 
+		  (unsigned int)(MGA_READ( MGA_PRIMADDRESS ) -
 				 dev_priv->primary->offset) );
 	DRM_INFO( "current freelist:\n" );
 
@@ -375,7 +372,7 @@ static drm_buf_t *mga_freelist_get( drm_device_t *dev )
 	u32 head, wrap;
 	DRM_DEBUG( "%s:\n", __FUNCTION__ );
 
-	head = *dev_priv->prim.head;
+	head = MGA_READ( MGA_PRIMADDRESS );
 	wrap = dev_priv->sarea_priv->last_wrap;
 
 	DRM_DEBUG( "   tail=0x%06lx %d\n",
@@ -403,7 +400,7 @@ int mga_freelist_put( drm_device_t *dev, drm_buf_t *buf )
 {
 	drm_mga_private_t *dev_priv = dev->dev_private;
 	drm_mga_buf_priv_t *buf_priv = buf->dev_private;
-	drm_mga_freelist_t *head, *next, *prev;
+	drm_mga_freelist_t *head, *entry, *prev;
 
 	DRM_DEBUG( "%s: age=0x%06lx wrap=%d\n",
 		   __FUNCTION__,
@@ -411,22 +408,23 @@ int mga_freelist_put( drm_device_t *dev, drm_buf_t *buf )
 		   dev_priv->primary->offset,
 		   buf_priv->list_entry->age.wrap );
 
-	/* Put buffer on the head + 1, as the head is a sentinal.
-	 */
-
-	next = buf_priv->list_entry;
+	entry = buf_priv->list_entry;
 	head = dev_priv->head;
-	prev = head->next;
 
 	if ( buf_priv->list_entry->age.head == MGA_BUFFER_USED ) {
-		SET_AGE( &next->age, MGA_BUFFER_FREE, 0 );
+		SET_AGE( &entry->age, MGA_BUFFER_FREE, 0 );
+		prev = dev_priv->tail;
+		prev->next = entry;
+		entry->prev = prev;
+		entry->next = NULL;
+	} else {
+		prev = head->next;
+		head->next = entry;
+		prev->prev = entry;
+		entry->prev = head;
+		entry->next = prev;
 	}
 
-	head->next = next;
-	prev->prev = next;
-	next->prev = head;
-	next->next = prev;
-
 	return 0;
 }
 
@@ -529,7 +527,6 @@ static int mga_do_init_dma( drm_device_t *dev, drm_mga_init_t *init )
 			      + dev_priv->primary->size);
 	dev_priv->prim.size = dev_priv->primary->size;
 
-	dev_priv->prim.head = &dev_priv->prim.status[0];
 	dev_priv->prim.tail = 0;
 	dev_priv->prim.space = dev_priv->prim.size;
 
@@ -625,10 +622,11 @@ int mga_dma_flush( struct inode *inode, struct file *filp,
 
 	WRAP_TEST_WITH_RETURN( dev_priv );
 
+#if 0
 	if ( lock.flags & (_DRM_LOCK_FLUSH | _DRM_LOCK_FLUSH_ALL) ) {
 		mga_do_dma_flush( dev_priv );
 	}
-
+#endif
 	if ( lock.flags & _DRM_LOCK_QUIESCENT ) {
 		return mga_do_wait_for_idle( dev_priv );
 	} else {
diff --git a/linux/mga_drm.h b/linux/mga_drm.h
index c2fe4a188..5cf863185 100644
--- a/linux/mga_drm.h
+++ b/linux/mga_drm.h
@@ -97,7 +97,7 @@
 /* 32 buffers of 64k each, total 2 meg.
  */
 #define MGA_BUFFER_SIZE		(1 << 16)
-#define MGA_NUM_BUFFERS		32
+#define MGA_NUM_BUFFERS		128
 
 /* Keep these small for testing.
  */
diff --git a/linux/mga_drv.h b/linux/mga_drv.h
index f56186e1e..bd33d9d1e 100644
--- a/linux/mga_drv.h
+++ b/linux/mga_drv.h
@@ -36,7 +36,6 @@ typedef struct drm_mga_primary_buffer {
 	u8 *end;
 	int size;
 
-	volatile u32 *head;
 	u32 tail;
 	int space;
 
@@ -246,6 +245,13 @@ do {									\
 
 #define FLUSH_DMA()							\
 do {									\
+	if ( 0 ) {							\
+		DRM_INFO( __FUNCTION__ ":\n" );				\
+		DRM_INFO( "   tail=0x%06x head=0x%06lx\n",		\
+			  dev_priv->prim.tail,				\
+			  MGA_READ( MGA_PRIMADDRESS ) -			\
+			  dev_priv->primary->offset );			\
+	}								\
 	if ( dev_priv->prim.space < dev_priv->prim.high_mark ) {	\
 		mga_do_dma_wrap_start( dev_priv );			\
 	} else {							\
@@ -278,7 +284,7 @@ do {									\
 } while (0)
 
 
-/* Buffer ageing via primary DMA stream head pointer.
+/* Buffer aging via primary DMA stream head pointer.
  */
 
 #define SET_AGE( age, h, w )						\
diff --git a/linux/mga_state.c b/linux/mga_state.c
index dc2c8c301..1e1c2827a 100644
--- a/linux/mga_state.c
+++ b/linux/mga_state.c
@@ -591,20 +591,9 @@ static void mga_dma_dispatch_swap( drm_device_t *dev )
 	int i;
 	DMA_LOCALS;
 	DRM_DEBUG( __FUNCTION__ ":\n" );
-	DRM_DEBUG( "   head = 0x%06x\n", *dev_priv->prim.head );
+	DRM_DEBUG( "   head = 0x%06x\n", MGA_READ( MGA_PRIMADDRESS ) );
 
-	sarea_priv->last_frame.head = dev_priv->prim.tail;
-	sarea_priv->last_frame.wrap = dev_priv->prim.last_wrap;
-
-	DRM_DEBUG( "   tail = 0x%06x\n", dev_priv->prim.tail );
-	DRM_DEBUG( "   wrap = 0x%06x\n", dev_priv->prim.last_wrap );
-
-	BEGIN_DMA( 4 + nbox );
-
-	DMA_BLOCK( MGA_DMAPAD,	0x00000000,
-		   MGA_DMAPAD,	0x00000000,
-		   MGA_DWGSYNC,	0x00007100,
-		   MGA_DWGSYNC,	0x00007000 );
+	BEGIN_DMA( 3 + nbox );
 
 	DMA_BLOCK( MGA_DSTORG,	dev_priv->front_offset,
 		   MGA_MACCESS,	dev_priv->maccess,
@@ -635,6 +624,12 @@ static void mga_dma_dispatch_swap( drm_device_t *dev )
 
 	ADVANCE_DMA();
 
+	DRM_DEBUG( "   tail = 0x%06x\n", dev_priv->prim.tail );
+	DRM_DEBUG( "   wrap = 0x%06x\n", dev_priv->prim.last_wrap );
+
+	sarea_priv->last_frame.head = dev_priv->prim.tail;
+	sarea_priv->last_frame.wrap = dev_priv->prim.last_wrap;
+
 	FLUSH_DMA();
 
 	DRM_DEBUG( "%s... done.\n", __FUNCTION__ );
diff --git a/linux/r128_cce.c b/linux/r128_cce.c
index f4b4f0c4a..c0e20c213 100644
--- a/linux/r128_cce.c
+++ b/linux/r128_cce.c
@@ -839,13 +839,9 @@ int r128_wait_ring( drm_r128_private_t *dev_priv, int n )
 	int i;
 
 	for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) {
-		ring->space = *ring->head - ring->tail;
-		if ( ring->space <= 0 )
-			ring->space += ring->size;
-
+		r128_update_ring_snapshot( ring );
 		if ( ring->space >= n )
 			return 0;
-
 		udelay( 1 );
 	}
 
@@ -854,19 +850,6 @@ int r128_wait_ring( drm_r128_private_t *dev_priv, int n )
 	return -EBUSY;
 }
 
-void r128_update_ring_snapshot( drm_r128_private_t *dev_priv )
-{
-	drm_r128_ring_buffer_t *ring = &dev_priv->ring;
-
-	ring->space = *ring->head - ring->tail;
-#if R128_PERFORMANCE_BOXES
-	if ( ring->space == 0 )
-		atomic_inc( &dev_priv->idle_count );
-#endif
-	if ( ring->space <= 0 )
-		ring->space += ring->size;
-}
-
 static int r128_cce_get_buffers( drm_device_t *dev, drm_dma_t *d )
 {
 	int i;
diff --git a/linux/r128_drv.c b/linux/r128_drv.c
index e42868ed6..daae0f875 100644
--- a/linux/r128_drv.c
+++ b/linux/r128_drv.c
@@ -38,11 +38,11 @@
 
 #define DRIVER_NAME		"r128"
 #define DRIVER_DESC		"ATI Rage 128"
-#define DRIVER_DATE		"20010216"
+#define DRIVER_DATE		"20010308"
 
 #define DRIVER_MAJOR		2
 #define DRIVER_MINOR		1
-#define DRIVER_PATCHLEVEL	4
+#define DRIVER_PATCHLEVEL	5
 
 #define DRIVER_IOCTLS							    \
    [DRM_IOCTL_NR(DRM_IOCTL_DMA)]             = { r128_cce_buffers,  1, 0 }, \
diff --git a/linux/r128_drv.h b/linux/r128_drv.h
index 9f53746f8..6216b9a1b 100644
--- a/linux/r128_drv.h
+++ b/linux/r128_drv.h
@@ -130,7 +130,14 @@ extern void r128_freelist_reset( drm_device_t *dev );
 extern drm_buf_t *r128_freelist_get( drm_device_t *dev );
 
 extern int r128_wait_ring( drm_r128_private_t *dev_priv, int n );
-extern void r128_update_ring_snapshot( drm_r128_private_t *dev_priv );
+
+static inline void
+r128_update_ring_snapshot( drm_r128_ring_buffer_t *ring )
+{
+	ring->space = (*(volatile int *)ring->head - ring->tail) * sizeof(u32);
+	if ( ring->space <= 0 )
+		ring->space += ring->size;
+}
 
 extern int r128_do_cce_idle( drm_r128_private_t *dev_priv );
 extern int r128_do_cleanup_cce( drm_device_t *dev );
@@ -415,9 +422,7 @@ do {									\
 	drm_r128_ring_buffer_t *ring = &dev_priv->ring; int i;		\
 	if ( ring->space < ring->high_mark ) {				\
 		for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) {	\
-			ring->space = *ring->head - ring->tail;		\
-			if ( ring->space <= 0 )				\
-				ring->space += ring->size;		\
+			r128_update_ring_snapshot( ring );		\
 			if ( ring->space >= ring->high_mark )		\
 				goto __ring_space_done;			\
 			udelay( 1 );					\
@@ -462,7 +467,7 @@ do {									\
 		DRM_INFO( "BEGIN_RING( %d ) in %s\n",			\
 			   (n), __FUNCTION__ );				\
 	}								\
-	if ( dev_priv->ring.space < (n) * sizeof(u32) ) {		\
+	if ( dev_priv->ring.space <= (n) * sizeof(u32) ) {		\
 		r128_wait_ring( dev_priv, (n) * sizeof(u32) );		\
 	}								\
 	dev_priv->ring.space -= (n) * sizeof(u32);			\
diff --git a/linux/r128_state.c b/linux/r128_state.c
index fa16f3ced..9fc6b485c 100644
--- a/linux/r128_state.c
+++ b/linux/r128_state.c
@@ -372,8 +372,6 @@ static void r128_cce_dispatch_clear( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	if ( dev_priv->page_flipping && dev_priv->current_page == 1 ) {
 		unsigned int tmp = flags;
 
@@ -477,8 +475,6 @@ static void r128_cce_dispatch_swap( drm_device_t *dev )
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	r128_update_ring_snapshot( dev_priv );
-
 #if R128_PERFORMANCE_BOXES
 	/* Do some trivial performance monitoring...
 	 */
@@ -535,8 +531,6 @@ static void r128_cce_dispatch_flip( drm_device_t *dev )
 	RING_LOCALS;
 	DRM_DEBUG( "%s: page=%d\n", __FUNCTION__, dev_priv->current_page );
 
-	r128_update_ring_snapshot( dev_priv );
-
 #if R128_PERFORMANCE_BOXES
 	/* Do some trivial performance monitoring...
 	 */
@@ -587,8 +581,6 @@ static void r128_cce_dispatch_vertex( drm_device_t *dev,
 	DRM_DEBUG( "%s: buf=%d nbox=%d\n",
 		   __FUNCTION__, buf->idx, sarea_priv->nbox );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	if ( 0 )
 		r128_print_dirty( "dispatch_vertex", sarea_priv->dirty );
 
@@ -656,8 +648,6 @@ static void r128_cce_dispatch_indirect( drm_device_t *dev,
 	DRM_DEBUG( "indirect: buf=%d s=0x%x e=0x%x\n",
 		   buf->idx, start, end );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	if ( start != end ) {
 		int offset = buf->bus_address + start;
 		int dwords = (end - start + 3) / sizeof(u32);
@@ -722,8 +712,6 @@ static void r128_cce_dispatch_indices( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "indices: s=%d e=%d c=%d\n", start, end, count );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	if ( 0 )
 		r128_print_dirty( "dispatch_indices", sarea_priv->dirty );
 
@@ -799,8 +787,6 @@ static int r128_cce_dispatch_blit( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	/* The compiler won't optimize away a division by a variable,
 	 * even if the only legal values are powers of two.  Thus, we'll
 	 * use a shift instead.
@@ -911,8 +897,6 @@ static int r128_cce_dispatch_write_span( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	count = depth->n;
 	if ( copy_from_user( &x, depth->x, sizeof(x) ) ) {
 		return -EFAULT;
@@ -1006,8 +990,6 @@ static int r128_cce_dispatch_write_pixels( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	count = depth->n;
 
 	x = kmalloc( count * sizeof(*x), 0 );
@@ -1123,8 +1105,6 @@ static int r128_cce_dispatch_read_span( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	count = depth->n;
 	if ( copy_from_user( &x, depth->x, sizeof(x) ) ) {
 		return -EFAULT;
@@ -1167,8 +1147,6 @@ static int r128_cce_dispatch_read_pixels( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	count = depth->n;
 	if ( count > dev_priv->depth_pitch ) {
 		count = dev_priv->depth_pitch;
@@ -1236,8 +1214,6 @@ static void r128_cce_dispatch_stipple( drm_device_t *dev, u32 *stipple )
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	r128_update_ring_snapshot( dev_priv );
-
 	BEGIN_RING( 33 );
 
 	OUT_RING( CCE_PACKET0( R128_BRUSH_DATA0, 31 ) );
diff --git a/linux/radeon_cp.c b/linux/radeon_cp.c
index c80665caf..ed8b1bbca 100644
--- a/linux/radeon_cp.c
+++ b/linux/radeon_cp.c
@@ -1166,13 +1166,9 @@ int radeon_wait_ring( drm_radeon_private_t *dev_priv, int n )
 	int i;
 
 	for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) {
-		ring->space = *ring->head - ring->tail;
-		if ( ring->space <= 0 )
-			ring->space += ring->size;
-
-		if ( ring->space >= n )
+		radeon_update_ring_snapshot( ring );
+		if ( ring->space > n )
 			return 0;
-
 		udelay( 1 );
 	}
 
@@ -1181,17 +1177,6 @@ int radeon_wait_ring( drm_radeon_private_t *dev_priv, int n )
 	return -EBUSY;
 }
 
-void radeon_update_ring_snapshot( drm_radeon_private_t *dev_priv )
-{
-	drm_radeon_ring_buffer_t *ring = &dev_priv->ring;
-
-	ring->space = *ring->head - ring->tail;
-	if ( ring->space == 0 )
-		atomic_inc( &dev_priv->idle_count );
-	if ( ring->space <= 0 )
-		ring->space += ring->size;
-}
-
 static int radeon_cp_get_buffers( drm_device_t *dev, drm_dma_t *d )
 {
 	int i;
diff --git a/linux/radeon_drv.c b/linux/radeon_drv.c
index 1aa889aee..dba2037f9 100644
--- a/linux/radeon_drv.c
+++ b/linux/radeon_drv.c
@@ -36,11 +36,11 @@
 
 #define DRIVER_NAME		"radeon"
 #define DRIVER_DESC		"ATI Radeon"
-#define DRIVER_DATE		"20010305"
+#define DRIVER_DATE		"20010308"
 
 #define DRIVER_MAJOR		1
 #define DRIVER_MINOR		0
-#define DRIVER_PATCHLEVEL	0
+#define DRIVER_PATCHLEVEL	1
 
 #define DRIVER_IOCTLS							     \
  [DRM_IOCTL_NR(DRM_IOCTL_DMA)]               = { radeon_cp_buffers,  1, 0 }, \
diff --git a/linux/radeon_drv.h b/linux/radeon_drv.h
index d279b0467..f176bb560 100644
--- a/linux/radeon_drv.h
+++ b/linux/radeon_drv.h
@@ -146,7 +146,14 @@ extern void radeon_freelist_reset( drm_device_t *dev );
 extern drm_buf_t *radeon_freelist_get( drm_device_t *dev );
 
 extern int radeon_wait_ring( drm_radeon_private_t *dev_priv, int n );
-extern void radeon_update_ring_snapshot( drm_radeon_private_t *dev_priv );
+
+static inline void
+radeon_update_ring_snapshot( drm_radeon_ring_buffer_t *ring )
+{
+	ring->space = (*(volatile int *)ring->head - ring->tail) * sizeof(u32);
+	if ( ring->space <= 0 )
+		ring->space += ring->size;
+}
 
 extern int radeon_do_cp_idle( drm_radeon_private_t *dev_priv );
 extern int radeon_do_cleanup_cp( drm_device_t *dev );
@@ -601,9 +608,7 @@ do {									\
 	drm_radeon_ring_buffer_t *ring = &dev_priv->ring; int i;	\
 	if ( ring->space < ring->high_mark ) {				\
 		for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) {	\
-			ring->space = *ring->head - ring->tail;		\
-			if ( ring->space <= 0 )				\
-				ring->space += ring->size;		\
+			radeon_update_ring_snapshot( ring );		\
 			if ( ring->space >= ring->high_mark )		\
 				goto __ring_space_done;			\
 			udelay( 1 );					\
@@ -657,7 +662,7 @@ do {									\
 		DRM_INFO( "BEGIN_RING( %d ) in %s\n",			\
 			   n, __FUNCTION__ );				\
 	}								\
-	if ( dev_priv->ring.space < (n) * sizeof(u32) ) {		\
+	if ( dev_priv->ring.space <= (n) * sizeof(u32) ) {		\
 		radeon_wait_ring( dev_priv, (n) * sizeof(u32) );	\
 	}								\
 	dev_priv->ring.space -= (n) * sizeof(u32);			\
diff --git a/linux/radeon_state.c b/linux/radeon_state.c
index ff1b3512f..9360c43b0 100644
--- a/linux/radeon_state.c
+++ b/linux/radeon_state.c
@@ -498,8 +498,6 @@ static void radeon_cp_dispatch_clear( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	radeon_update_ring_snapshot( dev_priv );
-
 	if ( dev_priv->page_flipping && dev_priv->current_page == 1 ) {
 		unsigned int tmp = flags;
 
@@ -656,8 +654,6 @@ static void radeon_cp_dispatch_swap( drm_device_t *dev )
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	radeon_update_ring_snapshot( dev_priv );
-
 #if RADEON_PERFORMANCE_BOXES
 	/* Do some trivial performance monitoring...
 	 */
@@ -725,8 +721,6 @@ static void radeon_cp_dispatch_flip( drm_device_t *dev )
 	RING_LOCALS;
 	DRM_DEBUG( "%s: page=%d\n", __FUNCTION__, dev_priv->current_page );
 
-	radeon_update_ring_snapshot( dev_priv );
-
 #if RADEON_PERFORMANCE_BOXES
 	/* Do some trivial performance monitoring...
 	 */
@@ -777,8 +771,6 @@ static void radeon_cp_dispatch_vertex( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "%s: nbox=%d\n", __FUNCTION__, sarea_priv->nbox );
 
-	radeon_update_ring_snapshot( dev_priv );
-
 	if ( 0 )
 		radeon_print_dirty( "dispatch_vertex", sarea_priv->dirty );
 
@@ -845,8 +837,6 @@ static void radeon_cp_dispatch_indirect( drm_device_t *dev,
 	DRM_DEBUG( "indirect: buf=%d s=0x%x e=0x%x\n",
 		   buf->idx, start, end );
 
-	radeon_update_ring_snapshot( dev_priv );
-
 	if ( start != end ) {
 		int offset = (dev_priv->agp_buffers_offset
 			      + buf->offset + start);
@@ -909,8 +899,6 @@ static void radeon_cp_dispatch_indices( drm_device_t *dev,
 	RING_LOCALS;
 	DRM_DEBUG( "indices: s=%d e=%d c=%d\n", start, end, count );
 
-	radeon_update_ring_snapshot( dev_priv );
-
 	if ( 0 )
 		radeon_print_dirty( "dispatch_indices", sarea_priv->dirty );
 
@@ -1066,8 +1054,10 @@ static int radeon_cp_dispatch_texture( drm_device_t *dev,
 		image->height -= height;
 		image->data = (char *)image->data + size;
 
-		if ( copy_to_user( tex->image, image, sizeof(*image) ) )
+		if ( copy_to_user( tex->image, image, sizeof(*image) ) ) {
+			DRM_ERROR( "EFAULT on tex->image\n" );
 			return -EFAULT;
+		}
 	} else if ( size < 4 ) {
 		size = 4;
 	}
@@ -1101,16 +1091,21 @@ static int radeon_cp_dispatch_texture( drm_device_t *dev,
 		/* Texture image width is larger than the minimum, so we
 		 * can upload it directly.
 		 */
-		if ( copy_from_user( buffer, data, dwords * sizeof(u32) ) )
+		if ( copy_from_user( buffer, data, dwords * sizeof(u32) ) ) {
+			DRM_ERROR( "EFAULT on data, %d dwords\n", dwords );
 			return -EFAULT;
+		}
 	} else {
 		/* Texture image width is less than the minimum, so we
 		 * need to pad out each image scanline to the minimum
 		 * width.
 		 */
 		for ( i = 0 ; i < tex->height ; i++ ) {
-			if ( copy_from_user( buffer, data, tex_width ) )
+			if ( copy_from_user( buffer, data, tex_width ) ) {
+				DRM_ERROR( "EFAULT on pad, %d bytes\n",
+					   tex_width );
 				return -EFAULT;
+			}
 			buffer += 8;
 			data += tex_width;
 		}
@@ -1143,8 +1138,6 @@ static void radeon_cp_dispatch_stipple( drm_device_t *dev, u32 *stipple )
 	RING_LOCALS;
 	DRM_DEBUG( "%s\n", __FUNCTION__ );
 
-	radeon_update_ring_snapshot( dev_priv );
-
 	BEGIN_RING( 35 );
 
 	OUT_RING( CP_PACKET0( RADEON_RE_STIPPLE_ADDR, 0 ) );
@@ -1180,7 +1173,6 @@ int radeon_cp_clear( struct inode *inode, struct file *filp,
 			     sizeof(clear) ) )
 		return -EFAULT;
 
-
 	RING_SPACE_TEST_WITH_RETURN( dev_priv );
 
 	if ( sarea_priv->nbox > RADEON_NR_SAREA_CLIPRECTS )