static inline bool
i915_gem_object_is_inactive(struct drm_i915_gem_object *obj)
{
- return obj->gtt_space && !obj->active;
+ return i915_gem_obj_ggtt_bound(obj) && !obj->active;
}
int
mutex_lock(&dev->struct_mutex);
list_for_each_entry(obj, &dev_priv->mm.bound_list, global_list)
if (obj->pin_count)
- pinned += obj->gtt_space->size;
+ pinned += i915_gem_obj_ggtt_size(obj);
mutex_unlock(&dev->struct_mutex);
args->aper_size = dev_priv->gtt.total;
* anyway again before the next pread happens. */
if (obj->cache_level == I915_CACHE_NONE)
needs_clflush = 1;
- if (obj->gtt_space) {
+ if (i915_gem_obj_ggtt_bound(obj)) {
ret = i915_gem_object_set_to_gtt_domain(obj, false);
if (ret)
return ret;
user_data = to_user_ptr(args->data_ptr);
remain = args->size;
- offset = obj->gtt_offset + args->offset;
+ offset = i915_gem_obj_ggtt_offset(obj) + args->offset;
while (remain > 0) {
/* Operation in this page
* right away and we therefore have to clflush anyway. */
if (obj->cache_level == I915_CACHE_NONE)
needs_clflush_after = 1;
- if (obj->gtt_space) {
+ if (i915_gem_obj_ggtt_bound(obj)) {
ret = i915_gem_object_set_to_gtt_domain(obj, true);
if (ret)
return ret;
obj->fault_mappable = true;
- pfn = ((dev_priv->gtt.mappable_base + obj->gtt_offset) >> PAGE_SHIFT) +
- page_offset;
+ pfn = dev_priv->gtt.mappable_base + i915_gem_obj_ggtt_offset(obj);
+ pfn >>= PAGE_SHIFT;
+ pfn += page_offset;
/* Finally, remap it using the new GTT offset */
ret = vm_insert_pfn(vma, (unsigned long)vmf->virtual_address, pfn);
if (obj->pages == NULL)
return 0;
- BUG_ON(obj->gtt_space);
+ BUG_ON(i915_gem_obj_ggtt_bound(obj));
if (obj->pages_pin_count)
return -EBUSY;
u32 seqno = intel_ring_get_seqno(ring);
BUG_ON(ring == NULL);
+ if (obj->ring != ring && obj->last_write_seqno) {
+ /* Keep the seqno relative to the current ring */
+ obj->last_write_seqno = seqno;
+ }
obj->ring = ring;
/* Add a reference if we're newly entering the active list. */
trace_i915_gem_request_add(ring, request->seqno);
ring->outstanding_lazy_request = 0;
- if (!dev_priv->mm.suspended) {
+ if (!dev_priv->ums.mm_suspended) {
if (i915_enable_hangcheck) {
mod_timer(&dev_priv->gpu_error.hangcheck_timer,
round_jiffies_up(jiffies + DRM_I915_HANGCHECK_JIFFIES));
static bool i915_head_inside_object(u32 acthd, struct drm_i915_gem_object *obj)
{
- if (acthd >= obj->gtt_offset &&
- acthd < obj->gtt_offset + obj->base.size)
+ if (acthd >= i915_gem_obj_ggtt_offset(obj) &&
+ acthd < i915_gem_obj_ggtt_offset(obj) + obj->base.size)
return true;
return false;
if (ring->hangcheck.action != wait &&
i915_request_guilty(request, acthd, &inside)) {
- DRM_ERROR("%s hung %s bo (0x%x ctx %d) at 0x%x\n",
+ DRM_ERROR("%s hung %s bo (0x%lx ctx %d) at 0x%x\n",
ring->name,
inside ? "inside" : "flushing",
request->batch_obj ?
- request->batch_obj->gtt_offset : 0,
+ i915_gem_obj_ggtt_offset(request->batch_obj) : 0,
request->ctx ? request->ctx->id : 0,
acthd);
idle &= list_empty(&ring->request_list);
}
- if (!dev_priv->mm.suspended && !idle)
+ if (!dev_priv->ums.mm_suspended && !idle)
queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work,
round_jiffies_up_relative(HZ));
if (idle)
drm_i915_private_t *dev_priv = obj->base.dev->dev_private;
int ret;
- if (obj->gtt_space == NULL)
+ if (!i915_gem_obj_ggtt_bound(obj))
return 0;
if (obj->pin_count)
/* Avoid an unnecessary call to unbind on rebind. */
obj->map_and_fenceable = true;
- drm_mm_put_block(obj->gtt_space);
- obj->gtt_space = NULL;
- obj->gtt_offset = 0;
+ drm_mm_remove_node(&obj->gtt_space);
return 0;
}
drm_i915_private_t *dev_priv = dev->dev_private;
int fence_reg;
int fence_pitch_shift;
- uint64_t val;
if (INTEL_INFO(dev)->gen >= 6) {
fence_reg = FENCE_REG_SANDYBRIDGE_0;
fence_pitch_shift = I965_FENCE_PITCH_SHIFT;
}
+ fence_reg += reg * 8;
+
+ /* To w/a incoherency with non-atomic 64-bit register updates,
+ * we split the 64-bit update into two 32-bit writes. In order
+ * for a partial fence not to be evaluated between writes, we
+ * precede the update with write to turn off the fence register,
+ * and only enable the fence as the last step.
+ *
+ * For extra levels of paranoia, we make sure each step lands
+ * before applying the next step.
+ */
+ I915_WRITE(fence_reg, 0);
+ POSTING_READ(fence_reg);
+
if (obj) {
- u32 size = obj->gtt_space->size;
+ u32 size = i915_gem_obj_ggtt_size(obj);
+ uint64_t val;
- val = (uint64_t)((obj->gtt_offset + size - 4096) &
+ val = (uint64_t)((i915_gem_obj_ggtt_offset(obj) + size - 4096) &
0xfffff000) << 32;
- val |= obj->gtt_offset & 0xfffff000;
+ val |= i915_gem_obj_ggtt_offset(obj) & 0xfffff000;
val |= (uint64_t)((obj->stride / 128) - 1) << fence_pitch_shift;
if (obj->tiling_mode == I915_TILING_Y)
val |= 1 << I965_FENCE_TILING_Y_SHIFT;
val |= I965_FENCE_REG_VALID;
- } else
- val = 0;
- fence_reg += reg * 8;
- I915_WRITE64(fence_reg, val);
- POSTING_READ(fence_reg);
+ I915_WRITE(fence_reg + 4, val >> 32);
+ POSTING_READ(fence_reg + 4);
+
+ I915_WRITE(fence_reg + 0, val);
+ POSTING_READ(fence_reg);
+ } else {
+ I915_WRITE(fence_reg + 4, 0);
+ POSTING_READ(fence_reg + 4);
+ }
}
static void i915_write_fence_reg(struct drm_device *dev, int reg,
u32 val;
if (obj) {
- u32 size = obj->gtt_space->size;
+ u32 size = i915_gem_obj_ggtt_size(obj);
int pitch_val;
int tile_width;
- WARN((obj->gtt_offset & ~I915_FENCE_START_MASK) ||
+ WARN((i915_gem_obj_ggtt_offset(obj) & ~I915_FENCE_START_MASK) ||
(size & -size) != size ||
- (obj->gtt_offset & (size - 1)),
- "object 0x%08x [fenceable? %d] not 1M or pot-size (0x%08x) aligned\n",
- obj->gtt_offset, obj->map_and_fenceable, size);
+ (i915_gem_obj_ggtt_offset(obj) & (size - 1)),
+ "object 0x%08lx [fenceable? %d] not 1M or pot-size (0x%08x) aligned\n",
+ i915_gem_obj_ggtt_offset(obj), obj->map_and_fenceable, size);
if (obj->tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev))
tile_width = 128;
pitch_val = obj->stride / tile_width;
pitch_val = ffs(pitch_val) - 1;
- val = obj->gtt_offset;
+ val = i915_gem_obj_ggtt_offset(obj);
if (obj->tiling_mode == I915_TILING_Y)
val |= 1 << I830_FENCE_TILING_Y_SHIFT;
val |= I915_FENCE_SIZE_BITS(size);
uint32_t val;
if (obj) {
- u32 size = obj->gtt_space->size;
+ u32 size = i915_gem_obj_ggtt_size(obj);
uint32_t pitch_val;
- WARN((obj->gtt_offset & ~I830_FENCE_START_MASK) ||
+ WARN((i915_gem_obj_ggtt_offset(obj) & ~I830_FENCE_START_MASK) ||
(size & -size) != size ||
- (obj->gtt_offset & (size - 1)),
- "object 0x%08x not 512K or pot-size 0x%08x aligned\n",
- obj->gtt_offset, size);
+ (i915_gem_obj_ggtt_offset(obj) & (size - 1)),
+ "object 0x%08lx not 512K or pot-size 0x%08x aligned\n",
+ i915_gem_obj_ggtt_offset(obj), size);
pitch_val = obj->stride / 128;
pitch_val = ffs(pitch_val) - 1;
- val = obj->gtt_offset;
+ val = i915_gem_obj_ggtt_offset(obj);
if (obj->tiling_mode == I915_TILING_Y)
val |= 1 << I830_FENCE_TILING_Y_SHIFT;
val |= I830_FENCE_SIZE_BITS(size);
return fence - dev_priv->fence_regs;
}
-struct write_fence {
- struct drm_device *dev;
- struct drm_i915_gem_object *obj;
- int fence;
-};
-
-static void i915_gem_write_fence__ipi(void *data)
-{
- struct write_fence *args = data;
-
- /* Required for SNB+ with LLC */
- wbinvd();
-
- /* Required for VLV */
- i915_gem_write_fence(args->dev, args->fence, args->obj);
-}
-
static void i915_gem_object_update_fence(struct drm_i915_gem_object *obj,
struct drm_i915_fence_reg *fence,
bool enable)
{
struct drm_i915_private *dev_priv = obj->base.dev->dev_private;
- struct write_fence args = {
- .dev = obj->base.dev,
- .fence = fence_number(dev_priv, fence),
- .obj = enable ? obj : NULL,
- };
-
- /* In order to fully serialize access to the fenced region and
- * the update to the fence register we need to take extreme
- * measures on SNB+. In theory, the write to the fence register
- * flushes all memory transactions before, and coupled with the
- * mb() placed around the register write we serialise all memory
- * operations with respect to the changes in the tiler. Yet, on
- * SNB+ we need to take a step further and emit an explicit wbinvd()
- * on each processor in order to manually flush all memory
- * transactions before updating the fence register.
- *
- * However, Valleyview complicates matter. There the wbinvd is
- * insufficient and unlike SNB/IVB requires the serialising
- * register write. (Note that that register write by itself is
- * conversely not sufficient for SNB+.) To compromise, we do both.
- */
- if (INTEL_INFO(args.dev)->gen >= 6)
- on_each_cpu(i915_gem_write_fence__ipi, &args, 1);
- else
- i915_gem_write_fence(args.dev, args.fence, args.obj);
+ int reg = fence_number(dev_priv, fence);
+
+ i915_gem_write_fence(obj->base.dev, reg, enable ? obj : NULL);
if (enable) {
- obj->fence_reg = args.fence;
+ obj->fence_reg = reg;
fence->obj = obj;
list_move_tail(&fence->lru_list, &dev_priv->mm.fence_list);
} else {
if (HAS_LLC(dev))
return true;
- if (gtt_space == NULL)
+ if (!drm_mm_node_allocated(gtt_space))
return true;
if (list_empty(>t_space->node_list))
if (obj->cache_level != obj->gtt_space->color) {
printk(KERN_ERR "object reserved space [%08lx, %08lx] with wrong color, cache_level=%x, color=%lx\n",
- obj->gtt_space->start,
- obj->gtt_space->start + obj->gtt_space->size,
+ i915_gem_obj_ggtt_offset(obj),
+ i915_gem_obj_ggtt_offset(obj) + i915_gem_obj_ggtt_size(obj),
obj->cache_level,
obj->gtt_space->color);
err++;
obj->gtt_space,
obj->cache_level)) {
printk(KERN_ERR "invalid GTT space found at [%08lx, %08lx] - color=%x\n",
- obj->gtt_space->start,
- obj->gtt_space->start + obj->gtt_space->size,
+ i915_gem_obj_ggtt_offset(obj),
+ i915_gem_obj_ggtt_offset(obj) + i915_gem_obj_ggtt_size(obj),
obj->cache_level);
err++;
continue;
{
struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_mm_node *node;
u32 size, fence_size, fence_alignment, unfenced_alignment;
bool mappable, fenceable;
size_t gtt_max = map_and_fenceable ?
i915_gem_object_pin_pages(obj);
- node = kzalloc(sizeof(*node), GFP_KERNEL);
- if (node == NULL) {
- i915_gem_object_unpin_pages(obj);
- return -ENOMEM;
- }
-
search_free:
- ret = drm_mm_insert_node_in_range_generic(&dev_priv->mm.gtt_space, node,
+ ret = drm_mm_insert_node_in_range_generic(&dev_priv->mm.gtt_space,
+ &obj->gtt_space,
size, alignment,
obj->cache_level, 0, gtt_max);
if (ret) {
goto search_free;
i915_gem_object_unpin_pages(obj);
- kfree(node);
return ret;
}
- if (WARN_ON(!i915_gem_valid_gtt_space(dev, node, obj->cache_level))) {
+ if (WARN_ON(!i915_gem_valid_gtt_space(dev, &obj->gtt_space,
+ obj->cache_level))) {
i915_gem_object_unpin_pages(obj);
- drm_mm_put_block(node);
+ drm_mm_remove_node(&obj->gtt_space);
return -EINVAL;
}
ret = i915_gem_gtt_prepare_object(obj);
if (ret) {
i915_gem_object_unpin_pages(obj);
- drm_mm_put_block(node);
+ drm_mm_remove_node(&obj->gtt_space);
return ret;
}
list_move_tail(&obj->global_list, &dev_priv->mm.bound_list);
list_add_tail(&obj->mm_list, &dev_priv->mm.inactive_list);
- obj->gtt_space = node;
- obj->gtt_offset = node->start;
-
fenceable =
- node->size == fence_size &&
- (node->start & (fence_alignment - 1)) == 0;
+ i915_gem_obj_ggtt_size(obj) == fence_size &&
+ (i915_gem_obj_ggtt_offset(obj) & (fence_alignment - 1)) == 0;
- mappable =
- obj->gtt_offset + obj->base.size <= dev_priv->gtt.mappable_end;
+ mappable = i915_gem_obj_ggtt_offset(obj) + obj->base.size <=
+ dev_priv->gtt.mappable_end;
obj->map_and_fenceable = mappable && fenceable;
int ret;
/* Not valid to be called on unbound objects. */
- if (obj->gtt_space == NULL)
+ if (!i915_gem_obj_ggtt_bound(obj))
return -EINVAL;
if (obj->base.write_domain == I915_GEM_DOMAIN_GTT)
return -EBUSY;
}
- if (!i915_gem_valid_gtt_space(dev, obj->gtt_space, cache_level)) {
+ if (!i915_gem_valid_gtt_space(dev, &obj->gtt_space, cache_level)) {
ret = i915_gem_object_unbind(obj);
if (ret)
return ret;
}
- if (obj->gtt_space) {
+ if (i915_gem_obj_ggtt_bound(obj)) {
ret = i915_gem_object_finish_gpu(obj);
if (ret)
return ret;
i915_ppgtt_bind_object(dev_priv->mm.aliasing_ppgtt,
obj, cache_level);
- obj->gtt_space->color = cache_level;
+ i915_gem_obj_ggtt_set_color(obj, cache_level);
}
if (cache_level == I915_CACHE_NONE) {
if (WARN_ON(obj->pin_count == DRM_I915_GEM_OBJECT_MAX_PIN_COUNT))
return -EBUSY;
- if (obj->gtt_space != NULL) {
- if ((alignment && obj->gtt_offset & (alignment - 1)) ||
+ if (i915_gem_obj_ggtt_bound(obj)) {
+ if ((alignment && i915_gem_obj_ggtt_offset(obj) & (alignment - 1)) ||
(map_and_fenceable && !obj->map_and_fenceable)) {
WARN(obj->pin_count,
"bo is already pinned with incorrect alignment:"
- " offset=%x, req.alignment=%x, req.map_and_fenceable=%d,"
+ " offset=%lx, req.alignment=%x, req.map_and_fenceable=%d,"
" obj->map_and_fenceable=%d\n",
- obj->gtt_offset, alignment,
+ i915_gem_obj_ggtt_offset(obj), alignment,
map_and_fenceable,
obj->map_and_fenceable);
ret = i915_gem_object_unbind(obj);
}
}
- if (obj->gtt_space == NULL) {
+ if (!i915_gem_obj_ggtt_bound(obj)) {
struct drm_i915_private *dev_priv = obj->base.dev->dev_private;
ret = i915_gem_object_bind_to_gtt(obj, alignment,
i915_gem_object_unpin(struct drm_i915_gem_object *obj)
{
BUG_ON(obj->pin_count == 0);
- BUG_ON(obj->gtt_space == NULL);
+ BUG_ON(!i915_gem_obj_ggtt_bound(obj));
if (--obj->pin_count == 0)
obj->pin_mappable = false;
* as the X server doesn't manage domains yet
*/
i915_gem_object_flush_cpu_write_domain(obj);
- args->offset = obj->gtt_offset;
+ args->offset = i915_gem_obj_ggtt_offset(obj);
out:
drm_gem_object_unreference(&obj->base);
unlock:
drm_i915_private_t *dev_priv = dev->dev_private;
int ret;
- mutex_lock(&dev->struct_mutex);
-
- if (dev_priv->mm.suspended) {
+ if (dev_priv->ums.mm_suspended) {
mutex_unlock(&dev->struct_mutex);
return 0;
}
if (!drm_core_check_feature(dev, DRIVER_MODESET))
i915_gem_evict_everything(dev);
- /* Hack! Don't let anybody do execbuf while we don't control the chip.
- * We need to replace this with a semaphore, or something.
- * And not confound mm.suspended!
- */
- dev_priv->mm.suspended = 1;
del_timer_sync(&dev_priv->gpu_error.hangcheck_timer);
i915_kernel_lost_context(dev);
i915_gem_cleanup_ringbuffer(dev);
- mutex_unlock(&dev->struct_mutex);
-
/* Cancel the retire work handler, which should be idle now. */
cancel_delayed_work_sync(&dev_priv->mm.retire_work);
i915_gem_entervt_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv)
{
- drm_i915_private_t *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = dev->dev_private;
int ret;
if (drm_core_check_feature(dev, DRIVER_MODESET))
}
mutex_lock(&dev->struct_mutex);
- dev_priv->mm.suspended = 0;
+ dev_priv->ums.mm_suspended = 0;
ret = i915_gem_init_hw(dev);
if (ret != 0) {
cleanup_ringbuffer:
mutex_lock(&dev->struct_mutex);
i915_gem_cleanup_ringbuffer(dev);
- dev_priv->mm.suspended = 1;
+ dev_priv->ums.mm_suspended = 1;
mutex_unlock(&dev->struct_mutex);
return ret;
i915_gem_leavevt_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv)
{
+ struct drm_i915_private *dev_priv = dev->dev_private;
+ int ret;
+
if (drm_core_check_feature(dev, DRIVER_MODESET))
return 0;
drm_irq_uninstall(dev);
- return i915_gem_idle(dev);
+
+ mutex_lock(&dev->struct_mutex);
+ ret = i915_gem_idle(dev);
+
+ /* Hack! Don't let anybody do execbuf while we don't control the chip.
+ * We need to replace this with a semaphore, or something.
+ * And not confound ums.mm_suspended!
+ */
+ if (ret != 0)
+ dev_priv->ums.mm_suspended = 1;
+ mutex_unlock(&dev->struct_mutex);
+
+ return ret;
}
void
if (drm_core_check_feature(dev, DRIVER_MODESET))
return;
+ mutex_lock(&dev->struct_mutex);
ret = i915_gem_idle(dev);
if (ret)
DRM_ERROR("failed to idle hardware: %d\n", ret);
+ mutex_unlock(&dev->struct_mutex);
}
static void
list_for_each_entry(obj, &dev_priv->mm.unbound_list, global_list)
if (obj->pages_pin_count == 0)
cnt += obj->base.size >> PAGE_SHIFT;
- list_for_each_entry(obj, &dev_priv->mm.inactive_list, global_list)
+ list_for_each_entry(obj, &dev_priv->mm.inactive_list, mm_list)
if (obj->pin_count == 0 && obj->pages_pin_count == 0)
cnt += obj->base.size >> PAGE_SHIFT;