#include <linux/slab.h>
#include <linux/swap.h>
#include <linux/pci.h>
-#include <linux/intel-gtt.h>
-
-static uint32_t i915_gem_get_gtt_alignment(struct drm_i915_gem_object *obj_priv);
-static uint32_t i915_gem_get_gtt_size(struct drm_i915_gem_object *obj_priv);
-
-static int i915_gem_object_flush_gpu_write_domain(struct drm_gem_object *obj,
- bool pipelined);
-static void i915_gem_object_flush_gtt_write_domain(struct drm_gem_object *obj);
-static void i915_gem_object_flush_cpu_write_domain(struct drm_gem_object *obj);
-static int i915_gem_object_set_to_cpu_domain(struct drm_gem_object *obj,
- int write);
-static int i915_gem_object_set_cpu_read_domain_range(struct drm_gem_object *obj,
+
+static int i915_gem_object_flush_gpu_write_domain(struct drm_i915_gem_object *obj,
+ struct intel_ring_buffer *pipelined);
+static void i915_gem_object_flush_gtt_write_domain(struct drm_i915_gem_object *obj);
+static void i915_gem_object_flush_cpu_write_domain(struct drm_i915_gem_object *obj);
+static int i915_gem_object_set_to_cpu_domain(struct drm_i915_gem_object *obj,
+ bool write);
+static int i915_gem_object_set_cpu_read_domain_range(struct drm_i915_gem_object *obj,
uint64_t offset,
uint64_t size);
-static void i915_gem_object_set_to_full_cpu_read_domain(struct drm_gem_object *obj);
-static int i915_gem_object_wait_rendering(struct drm_gem_object *obj,
- bool interruptible);
-static int i915_gem_object_bind_to_gtt(struct drm_gem_object *obj,
+static void i915_gem_object_set_to_full_cpu_read_domain(struct drm_i915_gem_object *obj);
+static int i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj,
unsigned alignment,
- bool mappable,
- bool need_fence);
-static void i915_gem_clear_fence_reg(struct drm_gem_object *obj);
-static int i915_gem_phys_pwrite(struct drm_device *dev, struct drm_gem_object *obj,
+ bool map_and_fenceable);
+static void i915_gem_clear_fence_reg(struct drm_i915_gem_object *obj);
+static int i915_gem_phys_pwrite(struct drm_device *dev,
+ struct drm_i915_gem_object *obj,
struct drm_i915_gem_pwrite *args,
- struct drm_file *file_priv);
-static void i915_gem_free_object_tail(struct drm_gem_object *obj);
+ struct drm_file *file);
+static void i915_gem_free_object_tail(struct drm_i915_gem_object *obj);
static int i915_gem_inactive_shrink(struct shrinker *shrinker,
int nr_to_scan,
dev_priv->mm.object_memory -= size;
}
-static void i915_gem_info_add_gtt(struct drm_i915_private *dev_priv,
- struct drm_i915_gem_object *obj)
-{
- dev_priv->mm.gtt_count++;
- dev_priv->mm.gtt_memory += obj->gtt_space->size;
- if (obj->gtt_offset < dev_priv->mm.gtt_mappable_end) {
- dev_priv->mm.mappable_gtt_used +=
- min_t(size_t, obj->gtt_space->size,
- dev_priv->mm.gtt_mappable_end - obj->gtt_offset);
- }
-}
-
-static void i915_gem_info_remove_gtt(struct drm_i915_private *dev_priv,
- struct drm_i915_gem_object *obj)
-{
- dev_priv->mm.gtt_count--;
- dev_priv->mm.gtt_memory -= obj->gtt_space->size;
- if (obj->gtt_offset < dev_priv->mm.gtt_mappable_end) {
- dev_priv->mm.mappable_gtt_used -=
- min_t(size_t, obj->gtt_space->size,
- dev_priv->mm.gtt_mappable_end - obj->gtt_offset);
- }
-}
-
-/**
- * Update the mappable working set counters. Call _only_ when there is a change
- * in one of (pin|fault)_mappable and update *_mappable _before_ calling.
- * @mappable: new state the changed mappable flag (either pin_ or fault_).
- */
-static void
-i915_gem_info_update_mappable(struct drm_i915_private *dev_priv,
- struct drm_i915_gem_object *obj,
- bool mappable)
-{
- if (mappable) {
- if (obj->pin_mappable && obj->fault_mappable)
- /* Combined state was already mappable. */
- return;
- dev_priv->mm.gtt_mappable_count++;
- dev_priv->mm.gtt_mappable_memory += obj->gtt_space->size;
- } else {
- if (obj->pin_mappable || obj->fault_mappable)
- /* Combined state still mappable. */
- return;
- dev_priv->mm.gtt_mappable_count--;
- dev_priv->mm.gtt_mappable_memory -= obj->gtt_space->size;
- }
-}
-
-static void i915_gem_info_add_pin(struct drm_i915_private *dev_priv,
- struct drm_i915_gem_object *obj,
- bool mappable)
-{
- dev_priv->mm.pin_count++;
- dev_priv->mm.pin_memory += obj->gtt_space->size;
- if (mappable) {
- obj->pin_mappable = true;
- i915_gem_info_update_mappable(dev_priv, obj, true);
- }
-}
-
-static void i915_gem_info_remove_pin(struct drm_i915_private *dev_priv,
- struct drm_i915_gem_object *obj)
-{
- dev_priv->mm.pin_count--;
- dev_priv->mm.pin_memory -= obj->gtt_space->size;
- if (obj->pin_mappable) {
- obj->pin_mappable = false;
- i915_gem_info_update_mappable(dev_priv, obj, false);
- }
-}
-
int
i915_gem_check_is_wedged(struct drm_device *dev)
{
return -EIO;
}
-static int i915_mutex_lock_interruptible(struct drm_device *dev)
+int i915_mutex_lock_interruptible(struct drm_device *dev)
{
struct drm_i915_private *dev_priv = dev->dev_private;
int ret;
}
static inline bool
-i915_gem_object_is_inactive(struct drm_i915_gem_object *obj_priv)
+i915_gem_object_is_inactive(struct drm_i915_gem_object *obj)
{
- return obj_priv->gtt_space &&
- !obj_priv->active &&
- obj_priv->pin_count == 0;
+ return obj->gtt_space && !obj->active && obj->pin_count == 0;
}
-int i915_gem_do_init(struct drm_device *dev,
- unsigned long start,
- unsigned long mappable_end,
- unsigned long end)
+void i915_gem_do_init(struct drm_device *dev,
+ unsigned long start,
+ unsigned long mappable_end,
+ unsigned long end)
{
drm_i915_private_t *dev_priv = dev->dev_private;
- if (start >= end ||
- (start & (PAGE_SIZE - 1)) != 0 ||
- (end & (PAGE_SIZE - 1)) != 0) {
- return -EINVAL;
- }
-
drm_mm_init(&dev_priv->mm.gtt_space, start,
end - start);
dev_priv->mm.gtt_total = end - start;
dev_priv->mm.mappable_gtt_total = min(end, mappable_end) - start;
dev_priv->mm.gtt_mappable_end = mappable_end;
-
- return 0;
}
int
i915_gem_init_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_gem_init *args = data;
- int ret;
+
+ if (args->gtt_start >= args->gtt_end ||
+ (args->gtt_end | args->gtt_start) & (PAGE_SIZE - 1))
+ return -EINVAL;
mutex_lock(&dev->struct_mutex);
- ret = i915_gem_do_init(dev, args->gtt_start, args->gtt_end, args->gtt_end);
+ i915_gem_do_init(dev, args->gtt_start, args->gtt_end, args->gtt_end);
mutex_unlock(&dev->struct_mutex);
- return ret;
+ return 0;
}
int
i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_private *dev_priv = dev->dev_private;
struct drm_i915_gem_get_aperture *args = data;
+ struct drm_i915_gem_object *obj;
+ size_t pinned;
if (!(dev->driver->driver_features & DRIVER_GEM))
return -ENODEV;
+ pinned = 0;
mutex_lock(&dev->struct_mutex);
- args->aper_size = dev_priv->mm.gtt_total;
- args->aper_available_size = args->aper_size - dev_priv->mm.pin_memory;
+ list_for_each_entry(obj, &dev_priv->mm.pinned_list, mm_list)
+ pinned += obj->gtt_space->size;
mutex_unlock(&dev->struct_mutex);
+ args->aper_size = dev_priv->mm.gtt_total;
+ args->aper_available_size = args->aper_size -pinned;
+
return 0;
}
-
/**
* Creates a new mm object and returns a handle to it.
*/
int
i915_gem_create_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_gem_create *args = data;
- struct drm_gem_object *obj;
+ struct drm_i915_gem_object *obj;
int ret;
u32 handle;
if (obj == NULL)
return -ENOMEM;
- ret = drm_gem_handle_create(file_priv, obj, &handle);
+ ret = drm_gem_handle_create(file, &obj->base, &handle);
if (ret) {
- drm_gem_object_release(obj);
- i915_gem_info_remove_obj(dev->dev_private, obj->size);
+ drm_gem_object_release(&obj->base);
+ i915_gem_info_remove_obj(dev->dev_private, obj->base.size);
kfree(obj);
return ret;
}
/* drop reference from allocate - handle holds it now */
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
trace_i915_gem_object_create(obj);
args->handle = handle;
return 0;
}
-static int i915_gem_object_needs_bit17_swizzle(struct drm_gem_object *obj)
+static int i915_gem_object_needs_bit17_swizzle(struct drm_i915_gem_object *obj)
{
- drm_i915_private_t *dev_priv = obj->dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+ drm_i915_private_t *dev_priv = obj->base.dev->dev_private;
return dev_priv->mm.bit_6_swizzle_x == I915_BIT_6_SWIZZLE_9_10_17 &&
- obj_priv->tiling_mode != I915_TILING_NONE;
+ obj->tiling_mode != I915_TILING_NONE;
}
static inline void
* fault, it fails so we can fall back to i915_gem_shmem_pwrite_slow().
*/
static int
-i915_gem_shmem_pread_fast(struct drm_device *dev, struct drm_gem_object *obj,
+i915_gem_shmem_pread_fast(struct drm_device *dev,
+ struct drm_i915_gem_object *obj,
struct drm_i915_gem_pread *args,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- struct address_space *mapping = obj->filp->f_path.dentry->d_inode->i_mapping;
+ struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
ssize_t remain;
loff_t offset;
char __user *user_data;
user_data = (char __user *) (uintptr_t) args->data_ptr;
remain = args->size;
- obj_priv = to_intel_bo(obj);
offset = args->offset;
while (remain > 0) {
* and not take page faults.
*/
static int
-i915_gem_shmem_pread_slow(struct drm_device *dev, struct drm_gem_object *obj,
+i915_gem_shmem_pread_slow(struct drm_device *dev,
+ struct drm_i915_gem_object *obj,
struct drm_i915_gem_pread *args,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
- struct address_space *mapping = obj->filp->f_path.dentry->d_inode->i_mapping;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+ struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
struct mm_struct *mm = current->mm;
struct page **user_pages;
ssize_t remain;
do_bit17_swizzling = i915_gem_object_needs_bit17_swizzle(obj);
- obj_priv = to_intel_bo(obj);
offset = args->offset;
while (remain > 0) {
*/
int
i915_gem_pread_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_gem_pread *args = data;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
+ struct drm_i915_gem_object *obj;
int ret = 0;
+ if (args->size == 0)
+ return 0;
+
+ if (!access_ok(VERIFY_WRITE,
+ (char __user *)(uintptr_t)args->data_ptr,
+ args->size))
+ return -EFAULT;
+
+ ret = fault_in_pages_writeable((char __user *)(uintptr_t)args->data_ptr,
+ args->size);
+ if (ret)
+ return -EFAULT;
+
ret = i915_mutex_lock_interruptible(dev);
if (ret)
return ret;
- obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
if (obj == NULL) {
ret = -ENOENT;
goto unlock;
}
- obj_priv = to_intel_bo(obj);
/* Bounds check source. */
- if (args->offset > obj->size || args->size > obj->size - args->offset) {
+ if (args->offset > obj->base.size ||
+ args->size > obj->base.size - args->offset) {
ret = -EINVAL;
goto out;
}
- if (args->size == 0)
- goto out;
-
- if (!access_ok(VERIFY_WRITE,
- (char __user *)(uintptr_t)args->data_ptr,
- args->size)) {
- ret = -EFAULT;
- goto out;
- }
-
- ret = fault_in_pages_writeable((char __user *)(uintptr_t)args->data_ptr,
- args->size);
- if (ret) {
- ret = -EFAULT;
- goto out;
- }
-
ret = i915_gem_object_set_cpu_read_domain_range(obj,
args->offset,
args->size);
ret = -EFAULT;
if (!i915_gem_object_needs_bit17_swizzle(obj))
- ret = i915_gem_shmem_pread_fast(dev, obj, args, file_priv);
+ ret = i915_gem_shmem_pread_fast(dev, obj, args, file);
if (ret == -EFAULT)
- ret = i915_gem_shmem_pread_slow(dev, obj, args, file_priv);
+ ret = i915_gem_shmem_pread_slow(dev, obj, args, file);
out:
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
return ret;
* user into the GTT, uncached.
*/
static int
-i915_gem_gtt_pwrite_fast(struct drm_device *dev, struct drm_gem_object *obj,
+i915_gem_gtt_pwrite_fast(struct drm_device *dev,
+ struct drm_i915_gem_object *obj,
struct drm_i915_gem_pwrite *args,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
drm_i915_private_t *dev_priv = dev->dev_private;
ssize_t remain;
loff_t offset, page_base;
user_data = (char __user *) (uintptr_t) args->data_ptr;
remain = args->size;
- obj_priv = to_intel_bo(obj);
- offset = obj_priv->gtt_offset + args->offset;
+ offset = obj->gtt_offset + args->offset;
while (remain > 0) {
/* Operation in this page
* than using i915_gem_gtt_pwrite_fast on a G45 (32-bit).
*/
static int
-i915_gem_gtt_pwrite_slow(struct drm_device *dev, struct drm_gem_object *obj,
+i915_gem_gtt_pwrite_slow(struct drm_device *dev,
+ struct drm_i915_gem_object *obj,
struct drm_i915_gem_pwrite *args,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
drm_i915_private_t *dev_priv = dev->dev_private;
ssize_t remain;
loff_t gtt_page_base, offset;
if (ret)
goto out_unpin_pages;
- obj_priv = to_intel_bo(obj);
- offset = obj_priv->gtt_offset + args->offset;
+ offset = obj->gtt_offset + args->offset;
while (remain > 0) {
/* Operation in this page
* copy_from_user into the kmapped pages backing the object.
*/
static int
-i915_gem_shmem_pwrite_fast(struct drm_device *dev, struct drm_gem_object *obj,
+i915_gem_shmem_pwrite_fast(struct drm_device *dev,
+ struct drm_i915_gem_object *obj,
struct drm_i915_gem_pwrite *args,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
- struct address_space *mapping = obj->filp->f_path.dentry->d_inode->i_mapping;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+ struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
ssize_t remain;
loff_t offset;
char __user *user_data;
user_data = (char __user *) (uintptr_t) args->data_ptr;
remain = args->size;
- obj_priv = to_intel_bo(obj);
offset = args->offset;
- obj_priv->dirty = 1;
+ obj->dirty = 1;
while (remain > 0) {
struct page *page;
* struct_mutex is held.
*/
static int
-i915_gem_shmem_pwrite_slow(struct drm_device *dev, struct drm_gem_object *obj,
+i915_gem_shmem_pwrite_slow(struct drm_device *dev,
+ struct drm_i915_gem_object *obj,
struct drm_i915_gem_pwrite *args,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
- struct address_space *mapping = obj->filp->f_path.dentry->d_inode->i_mapping;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+ struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
struct mm_struct *mm = current->mm;
struct page **user_pages;
ssize_t remain;
do_bit17_swizzling = i915_gem_object_needs_bit17_swizzle(obj);
- obj_priv = to_intel_bo(obj);
offset = args->offset;
- obj_priv->dirty = 1;
+ obj->dirty = 1;
while (remain > 0) {
struct page *page;
struct drm_file *file)
{
struct drm_i915_gem_pwrite *args = data;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
- int ret = 0;
+ struct drm_i915_gem_object *obj;
+ int ret;
+
+ if (args->size == 0)
+ return 0;
+
+ if (!access_ok(VERIFY_READ,
+ (char __user *)(uintptr_t)args->data_ptr,
+ args->size))
+ return -EFAULT;
+
+ ret = fault_in_pages_readable((char __user *)(uintptr_t)args->data_ptr,
+ args->size);
+ if (ret)
+ return -EFAULT;
ret = i915_mutex_lock_interruptible(dev);
if (ret)
return ret;
- obj = drm_gem_object_lookup(dev, file, args->handle);
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
if (obj == NULL) {
ret = -ENOENT;
goto unlock;
}
- obj_priv = to_intel_bo(obj);
-
/* Bounds check destination. */
- if (args->offset > obj->size || args->size > obj->size - args->offset) {
+ if (args->offset > obj->base.size ||
+ args->size > obj->base.size - args->offset) {
ret = -EINVAL;
goto out;
}
- if (args->size == 0)
- goto out;
-
- if (!access_ok(VERIFY_READ,
- (char __user *)(uintptr_t)args->data_ptr,
- args->size)) {
- ret = -EFAULT;
- goto out;
- }
-
- ret = fault_in_pages_readable((char __user *)(uintptr_t)args->data_ptr,
- args->size);
- if (ret) {
- ret = -EFAULT;
- goto out;
- }
-
/* We can only do the GTT pwrite on untiled buffers, as otherwise
* it would end up going through the fenced access, and we'll get
* different detiling behavior between reading and writing.
* pread/pwrite currently are reading and writing from the CPU
* perspective, requiring manual detiling by the client.
*/
- if (obj_priv->phys_obj)
+ if (obj->phys_obj)
ret = i915_gem_phys_pwrite(dev, obj, args, file);
- else if (obj_priv->tiling_mode == I915_TILING_NONE &&
- obj_priv->gtt_space &&
- obj->write_domain != I915_GEM_DOMAIN_CPU) {
- ret = i915_gem_object_pin(obj, 0, true, false);
+ else if (obj->tiling_mode == I915_TILING_NONE &&
+ obj->gtt_space &&
+ obj->base.write_domain != I915_GEM_DOMAIN_CPU) {
+ ret = i915_gem_object_pin(obj, 0, true);
if (ret)
goto out;
}
out:
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
return ret;
*/
int
i915_gem_set_domain_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_private *dev_priv = dev->dev_private;
struct drm_i915_gem_set_domain *args = data;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
+ struct drm_i915_gem_object *obj;
uint32_t read_domains = args->read_domains;
uint32_t write_domain = args->write_domain;
int ret;
if (ret)
return ret;
- obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
if (obj == NULL) {
ret = -ENOENT;
goto unlock;
}
- obj_priv = to_intel_bo(obj);
intel_mark_busy(dev, obj);
/* Update the LRU on the fence for the CPU access that's
* about to occur.
*/
- if (obj_priv->fence_reg != I915_FENCE_REG_NONE) {
+ if (obj->fence_reg != I915_FENCE_REG_NONE) {
struct drm_i915_fence_reg *reg =
- &dev_priv->fence_regs[obj_priv->fence_reg];
+ &dev_priv->fence_regs[obj->fence_reg];
list_move_tail(®->lru_list,
&dev_priv->mm.fence_list);
}
}
/* Maintain LRU order of "inactive" objects */
- if (ret == 0 && i915_gem_object_is_inactive(obj_priv))
- list_move_tail(&obj_priv->mm_list, &dev_priv->mm.inactive_list);
+ if (ret == 0 && i915_gem_object_is_inactive(obj))
+ list_move_tail(&obj->mm_list, &dev_priv->mm.inactive_list);
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
return ret;
*/
int
i915_gem_sw_finish_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_gem_sw_finish *args = data;
- struct drm_gem_object *obj;
+ struct drm_i915_gem_object *obj;
int ret = 0;
if (!(dev->driver->driver_features & DRIVER_GEM))
if (ret)
return ret;
- obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
if (obj == NULL) {
ret = -ENOENT;
goto unlock;
}
/* Pinned buffers may be scanout, so flush the cache */
- if (to_intel_bo(obj)->pin_count)
+ if (obj->pin_count)
i915_gem_object_flush_cpu_write_domain(obj);
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
return ret;
*/
int
i915_gem_mmap_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_private *dev_priv = dev->dev_private;
struct drm_i915_gem_mmap *args = data;
if (!(dev->driver->driver_features & DRIVER_GEM))
return -ENODEV;
- obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+ obj = drm_gem_object_lookup(dev, file, args->handle);
if (obj == NULL)
return -ENOENT;
*/
int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
- struct drm_gem_object *obj = vma->vm_private_data;
- struct drm_device *dev = obj->dev;
+ struct drm_i915_gem_object *obj = to_intel_bo(vma->vm_private_data);
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
pgoff_t page_offset;
unsigned long pfn;
int ret = 0;
/* Now bind it into the GTT if needed */
mutex_lock(&dev->struct_mutex);
- BUG_ON(obj_priv->pin_count && !obj_priv->pin_mappable);
- if (obj_priv->gtt_space) {
- if (!obj_priv->mappable ||
- (obj_priv->tiling_mode && !obj_priv->fenceable)) {
- ret = i915_gem_object_unbind(obj);
- if (ret)
- goto unlock;
- }
+ if (!obj->map_and_fenceable) {
+ ret = i915_gem_object_unbind(obj);
+ if (ret)
+ goto unlock;
}
-
- if (!obj_priv->gtt_space) {
- ret = i915_gem_object_bind_to_gtt(obj, 0,
- true, obj_priv->tiling_mode);
+ if (!obj->gtt_space) {
+ ret = i915_gem_object_bind_to_gtt(obj, 0, true);
if (ret)
goto unlock;
}
if (ret)
goto unlock;
- if (!obj_priv->fault_mappable) {
- obj_priv->fault_mappable = true;
- i915_gem_info_update_mappable(dev_priv, obj_priv, true);
- }
-
/* Need a new fence register? */
- if (obj_priv->tiling_mode != I915_TILING_NONE) {
+ if (obj->tiling_mode != I915_TILING_NONE) {
ret = i915_gem_object_get_fence_reg(obj, true);
if (ret)
goto unlock;
}
- if (i915_gem_object_is_inactive(obj_priv))
- list_move_tail(&obj_priv->mm_list, &dev_priv->mm.inactive_list);
+ if (i915_gem_object_is_inactive(obj))
+ list_move_tail(&obj->mm_list, &dev_priv->mm.inactive_list);
+
+ obj->fault_mappable = true;
- pfn = ((dev->agp->base + obj_priv->gtt_offset) >> PAGE_SHIFT) +
+ pfn = ((dev->agp->base + obj->gtt_offset) >> PAGE_SHIFT) +
page_offset;
/* Finally, remap it using the new GTT offset */
mutex_unlock(&dev->struct_mutex);
switch (ret) {
+ case -EAGAIN:
+ set_need_resched();
case 0:
case -ERESTARTSYS:
return VM_FAULT_NOPAGE;
case -ENOMEM:
- case -EAGAIN:
return VM_FAULT_OOM;
default:
return VM_FAULT_SIGBUS;
* This routine allocates and attaches a fake offset for @obj.
*/
static int
-i915_gem_create_mmap_offset(struct drm_gem_object *obj)
+i915_gem_create_mmap_offset(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
struct drm_gem_mm *mm = dev->mm_private;
struct drm_map_list *list;
struct drm_local_map *map;
int ret = 0;
/* Set the object up for mmap'ing */
- list = &obj->map_list;
+ list = &obj->base.map_list;
list->map = kzalloc(sizeof(struct drm_map_list), GFP_KERNEL);
if (!list->map)
return -ENOMEM;
map = list->map;
map->type = _DRM_GEM;
- map->size = obj->size;
+ map->size = obj->base.size;
map->handle = obj;
/* Get a DRM GEM mmap offset allocated... */
list->file_offset_node = drm_mm_search_free(&mm->offset_manager,
- obj->size / PAGE_SIZE, 0, 0);
+ obj->base.size / PAGE_SIZE,
+ 0, 0);
if (!list->file_offset_node) {
- DRM_ERROR("failed to allocate offset for bo %d\n", obj->name);
+ DRM_ERROR("failed to allocate offset for bo %d\n",
+ obj->base.name);
ret = -ENOSPC;
goto out_free_list;
}
list->file_offset_node = drm_mm_get_block(list->file_offset_node,
- obj->size / PAGE_SIZE, 0);
+ obj->base.size / PAGE_SIZE,
+ 0);
if (!list->file_offset_node) {
ret = -ENOMEM;
goto out_free_list;
* fixup by i915_gem_fault().
*/
void
-i915_gem_release_mmap(struct drm_gem_object *obj)
+i915_gem_release_mmap(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+ if (!obj->fault_mappable)
+ return;
- if (unlikely(obj->map_list.map && dev->dev_mapping))
- unmap_mapping_range(dev->dev_mapping,
- (loff_t)obj->map_list.hash.key<<PAGE_SHIFT,
- obj->size, 1);
+ unmap_mapping_range(obj->base.dev->dev_mapping,
+ (loff_t)obj->base.map_list.hash.key<<PAGE_SHIFT,
+ obj->base.size, 1);
- if (obj_priv->fault_mappable) {
- obj_priv->fault_mappable = false;
- i915_gem_info_update_mappable(dev_priv, obj_priv, false);
- }
+ obj->fault_mappable = false;
}
static void
-i915_gem_free_mmap_offset(struct drm_gem_object *obj)
+i915_gem_free_mmap_offset(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
struct drm_gem_mm *mm = dev->mm_private;
- struct drm_map_list *list = &obj->map_list;
+ struct drm_map_list *list = &obj->base.map_list;
drm_ht_remove_item(&mm->offset_hash, &list->hash);
drm_mm_put_block(list->file_offset_node);
list->map = NULL;
}
+static uint32_t
+i915_gem_get_gtt_size(struct drm_i915_gem_object *obj)
+{
+ struct drm_device *dev = obj->base.dev;
+ uint32_t size;
+
+ if (INTEL_INFO(dev)->gen >= 4 ||
+ obj->tiling_mode == I915_TILING_NONE)
+ return obj->base.size;
+
+ /* Previous chips need a power-of-two fence region when tiling */
+ if (INTEL_INFO(dev)->gen == 3)
+ size = 1024*1024;
+ else
+ size = 512*1024;
+
+ while (size < obj->base.size)
+ size <<= 1;
+
+ return size;
+}
+
/**
* i915_gem_get_gtt_alignment - return required GTT alignment for an object
* @obj: object to check
*
* Return the required GTT alignment for an object, taking into account
- * potential fence register mapping if needed.
+ * potential fence register mapping.
*/
static uint32_t
-i915_gem_get_gtt_alignment(struct drm_i915_gem_object *obj_priv)
+i915_gem_get_gtt_alignment(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj_priv->base.dev;
+ struct drm_device *dev = obj->base.dev;
/*
* Minimum alignment is 4k (GTT page size), but might be greater
* if a fence register is needed for the object.
*/
if (INTEL_INFO(dev)->gen >= 4 ||
- obj_priv->tiling_mode == I915_TILING_NONE)
+ obj->tiling_mode == I915_TILING_NONE)
return 4096;
/*
* Previous chips need to be aligned to the size of the smallest
* fence register that can contain the object.
*/
- return i915_gem_get_gtt_size(obj_priv);
+ return i915_gem_get_gtt_size(obj);
}
+/**
+ * i915_gem_get_unfenced_gtt_alignment - return required GTT alignment for an
+ * unfenced object
+ * @obj: object to check
+ *
+ * Return the required GTT alignment for an object, only taking into account
+ * unfenced tiled surface requirements.
+ */
static uint32_t
-i915_gem_get_gtt_size(struct drm_i915_gem_object *obj_priv)
+i915_gem_get_unfenced_gtt_alignment(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj_priv->base.dev;
- uint32_t size;
+ struct drm_device *dev = obj->base.dev;
+ int tile_height;
/*
- * Minimum alignment is 4k (GTT page size), but might be greater
- * if a fence register is needed for the object.
+ * Minimum alignment is 4k (GTT page size) for sane hw.
*/
- if (INTEL_INFO(dev)->gen >= 4)
- return obj_priv->base.size;
+ if (INTEL_INFO(dev)->gen >= 4 || IS_G33(dev) ||
+ obj->tiling_mode == I915_TILING_NONE)
+ return 4096;
/*
- * Previous chips need to be aligned to the size of the smallest
- * fence register that can contain the object.
+ * Older chips need unfenced tiled buffers to be aligned to the left
+ * edge of an even tile row (where tile rows are counted as if the bo is
+ * placed in a fenced gtt region).
*/
- if (INTEL_INFO(dev)->gen == 3)
- size = 1024*1024;
+ if (IS_GEN2(dev) ||
+ (obj->tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev)))
+ tile_height = 32;
else
- size = 512*1024;
-
- while (size < obj_priv->base.size)
- size <<= 1;
+ tile_height = 8;
- return size;
+ return tile_height * obj->stride * 2;
}
/**
* i915_gem_mmap_gtt_ioctl - prepare an object for GTT mmap'ing
* @dev: DRM device
* @data: GTT mapping ioctl data
- * @file_priv: GEM object info
+ * @file: GEM object info
*
* Simply returns the fake offset to userspace so it can mmap it.
* The mmap call will end up in drm_gem_mmap(), which will set things
*/
int
i915_gem_mmap_gtt_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_private *dev_priv = dev->dev_private;
struct drm_i915_gem_mmap_gtt *args = data;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
+ struct drm_i915_gem_object *obj;
int ret;
if (!(dev->driver->driver_features & DRIVER_GEM))
if (ret)
return ret;
- obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
if (obj == NULL) {
ret = -ENOENT;
goto unlock;
}
- obj_priv = to_intel_bo(obj);
- if (obj->size > dev_priv->mm.gtt_mappable_end) {
+ if (obj->base.size > dev_priv->mm.gtt_mappable_end) {
ret = -E2BIG;
goto unlock;
}
- if (obj_priv->madv != I915_MADV_WILLNEED) {
+ if (obj->madv != I915_MADV_WILLNEED) {
DRM_ERROR("Attempting to mmap a purgeable buffer\n");
ret = -EINVAL;
goto out;
}
- if (!obj->map_list.map) {
+ if (!obj->base.map_list.map) {
ret = i915_gem_create_mmap_offset(obj);
if (ret)
goto out;
}
- args->offset = (u64)obj->map_list.hash.key << PAGE_SHIFT;
+ args->offset = (u64)obj->base.map_list.hash.key << PAGE_SHIFT;
out:
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
return ret;
}
static int
-i915_gem_object_get_pages_gtt(struct drm_gem_object *obj,
+i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj,
gfp_t gfpmask)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
int page_count, i;
struct address_space *mapping;
struct inode *inode;
/* Get the list of pages out of our struct file. They'll be pinned
* at this point until we release them.
*/
- page_count = obj->size / PAGE_SIZE;
- BUG_ON(obj_priv->pages != NULL);
- obj_priv->pages = drm_malloc_ab(page_count, sizeof(struct page *));
- if (obj_priv->pages == NULL)
+ page_count = obj->base.size / PAGE_SIZE;
+ BUG_ON(obj->pages != NULL);
+ obj->pages = drm_malloc_ab(page_count, sizeof(struct page *));
+ if (obj->pages == NULL)
return -ENOMEM;
- inode = obj->filp->f_path.dentry->d_inode;
+ inode = obj->base.filp->f_path.dentry->d_inode;
mapping = inode->i_mapping;
for (i = 0; i < page_count; i++) {
page = read_cache_page_gfp(mapping, i,
if (IS_ERR(page))
goto err_pages;
- obj_priv->pages[i] = page;
+ obj->pages[i] = page;
}
- if (obj_priv->tiling_mode != I915_TILING_NONE)
+ if (obj->tiling_mode != I915_TILING_NONE)
i915_gem_object_do_bit_17_swizzle(obj);
return 0;
err_pages:
while (i--)
- page_cache_release(obj_priv->pages[i]);
+ page_cache_release(obj->pages[i]);
- drm_free_large(obj_priv->pages);
- obj_priv->pages = NULL;
+ drm_free_large(obj->pages);
+ obj->pages = NULL;
return PTR_ERR(page);
}
static void
-i915_gem_object_put_pages_gtt(struct drm_gem_object *obj)
+i915_gem_object_put_pages_gtt(struct drm_i915_gem_object *obj)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- int page_count = obj->size / PAGE_SIZE;
+ int page_count = obj->base.size / PAGE_SIZE;
int i;
- BUG_ON(obj_priv->madv == __I915_MADV_PURGED);
+ BUG_ON(obj->madv == __I915_MADV_PURGED);
- if (obj_priv->tiling_mode != I915_TILING_NONE)
+ if (obj->tiling_mode != I915_TILING_NONE)
i915_gem_object_save_bit_17_swizzle(obj);
- if (obj_priv->madv == I915_MADV_DONTNEED)
- obj_priv->dirty = 0;
+ if (obj->madv == I915_MADV_DONTNEED)
+ obj->dirty = 0;
for (i = 0; i < page_count; i++) {
- if (obj_priv->dirty)
- set_page_dirty(obj_priv->pages[i]);
+ if (obj->dirty)
+ set_page_dirty(obj->pages[i]);
- if (obj_priv->madv == I915_MADV_WILLNEED)
- mark_page_accessed(obj_priv->pages[i]);
+ if (obj->madv == I915_MADV_WILLNEED)
+ mark_page_accessed(obj->pages[i]);
- page_cache_release(obj_priv->pages[i]);
+ page_cache_release(obj->pages[i]);
}
- obj_priv->dirty = 0;
-
- drm_free_large(obj_priv->pages);
- obj_priv->pages = NULL;
-}
-
-static uint32_t
-i915_gem_next_request_seqno(struct drm_device *dev,
- struct intel_ring_buffer *ring)
-{
- drm_i915_private_t *dev_priv = dev->dev_private;
+ obj->dirty = 0;
- ring->outstanding_lazy_request = true;
- return dev_priv->next_seqno;
+ drm_free_large(obj->pages);
+ obj->pages = NULL;
}
-static void
-i915_gem_object_move_to_active(struct drm_gem_object *obj,
+void
+i915_gem_object_move_to_active(struct drm_i915_gem_object *obj,
struct intel_ring_buffer *ring)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
struct drm_i915_private *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
uint32_t seqno = i915_gem_next_request_seqno(dev, ring);
BUG_ON(ring == NULL);
- obj_priv->ring = ring;
+ obj->ring = ring;
/* Add a reference if we're newly entering the active list. */
- if (!obj_priv->active) {
- drm_gem_object_reference(obj);
- obj_priv->active = 1;
+ if (!obj->active) {
+ drm_gem_object_reference(&obj->base);
+ obj->active = 1;
}
/* Move from whatever list we were on to the tail of execution. */
- list_move_tail(&obj_priv->mm_list, &dev_priv->mm.active_list);
- list_move_tail(&obj_priv->ring_list, &ring->active_list);
- obj_priv->last_rendering_seqno = seqno;
+ list_move_tail(&obj->mm_list, &dev_priv->mm.active_list);
+ list_move_tail(&obj->ring_list, &ring->active_list);
+
+ obj->last_rendering_seqno = seqno;
+ if (obj->fenced_gpu_access) {
+ struct drm_i915_fence_reg *reg;
+
+ BUG_ON(obj->fence_reg == I915_FENCE_REG_NONE);
+
+ obj->last_fenced_seqno = seqno;
+ obj->last_fenced_ring = ring;
+
+ reg = &dev_priv->fence_regs[obj->fence_reg];
+ list_move_tail(®->lru_list, &dev_priv->mm.fence_list);
+ }
+}
+
+static void
+i915_gem_object_move_off_active(struct drm_i915_gem_object *obj)
+{
+ list_del_init(&obj->ring_list);
+ obj->last_rendering_seqno = 0;
+ obj->last_fenced_seqno = 0;
}
static void
-i915_gem_object_move_to_flushing(struct drm_gem_object *obj)
+i915_gem_object_move_to_flushing(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- BUG_ON(!obj_priv->active);
- list_move_tail(&obj_priv->mm_list, &dev_priv->mm.flushing_list);
- list_del_init(&obj_priv->ring_list);
- obj_priv->last_rendering_seqno = 0;
+ BUG_ON(!obj->active);
+ list_move_tail(&obj->mm_list, &dev_priv->mm.flushing_list);
+
+ i915_gem_object_move_off_active(obj);
+}
+
+static void
+i915_gem_object_move_to_inactive(struct drm_i915_gem_object *obj)
+{
+ struct drm_device *dev = obj->base.dev;
+ struct drm_i915_private *dev_priv = dev->dev_private;
+
+ if (obj->pin_count != 0)
+ list_move_tail(&obj->mm_list, &dev_priv->mm.pinned_list);
+ else
+ list_move_tail(&obj->mm_list, &dev_priv->mm.inactive_list);
+
+ BUG_ON(!list_empty(&obj->gpu_write_list));
+ BUG_ON(!obj->active);
+ obj->ring = NULL;
+
+ i915_gem_object_move_off_active(obj);
+ obj->fenced_gpu_access = false;
+ obj->last_fenced_ring = NULL;
+
+ obj->active = 0;
+ drm_gem_object_unreference(&obj->base);
+
+ WARN_ON(i915_verify_lists(dev));
}
/* Immediately discard the backing storage */
static void
-i915_gem_object_truncate(struct drm_gem_object *obj)
+i915_gem_object_truncate(struct drm_i915_gem_object *obj)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
struct inode *inode;
/* Our goal here is to return as much of the memory as
* backing pages, *now*. Here we mirror the actions taken
* when by shmem_delete_inode() to release the backing store.
*/
- inode = obj->filp->f_path.dentry->d_inode;
+ inode = obj->base.filp->f_path.dentry->d_inode;
truncate_inode_pages(inode->i_mapping, 0);
if (inode->i_op->truncate_range)
inode->i_op->truncate_range(inode, 0, (loff_t)-1);
- obj_priv->madv = __I915_MADV_PURGED;
+ obj->madv = __I915_MADV_PURGED;
}
static inline int
-i915_gem_object_is_purgeable(struct drm_i915_gem_object *obj_priv)
-{
- return obj_priv->madv == I915_MADV_DONTNEED;
-}
-
-static void
-i915_gem_object_move_to_inactive(struct drm_gem_object *obj)
+i915_gem_object_is_purgeable(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
- drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
-
- if (obj_priv->pin_count != 0)
- list_move_tail(&obj_priv->mm_list, &dev_priv->mm.pinned_list);
- else
- list_move_tail(&obj_priv->mm_list, &dev_priv->mm.inactive_list);
- list_del_init(&obj_priv->ring_list);
-
- BUG_ON(!list_empty(&obj_priv->gpu_write_list));
-
- obj_priv->last_rendering_seqno = 0;
- obj_priv->ring = NULL;
- if (obj_priv->active) {
- obj_priv->active = 0;
- drm_gem_object_unreference(obj);
- }
- WARN_ON(i915_verify_lists(dev));
+ return obj->madv == I915_MADV_DONTNEED;
}
static void
uint32_t flush_domains,
struct intel_ring_buffer *ring)
{
- drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv, *next;
+ struct drm_i915_gem_object *obj, *next;
- list_for_each_entry_safe(obj_priv, next,
+ list_for_each_entry_safe(obj, next,
&ring->gpu_write_list,
gpu_write_list) {
- struct drm_gem_object *obj = &obj_priv->base;
-
- if (obj->write_domain & flush_domains) {
- uint32_t old_write_domain = obj->write_domain;
+ if (obj->base.write_domain & flush_domains) {
+ uint32_t old_write_domain = obj->base.write_domain;
- obj->write_domain = 0;
- list_del_init(&obj_priv->gpu_write_list);
+ obj->base.write_domain = 0;
+ list_del_init(&obj->gpu_write_list);
i915_gem_object_move_to_active(obj, ring);
- /* update the fence lru list */
- if (obj_priv->fence_reg != I915_FENCE_REG_NONE) {
- struct drm_i915_fence_reg *reg =
- &dev_priv->fence_regs[obj_priv->fence_reg];
- list_move_tail(®->lru_list,
- &dev_priv->mm.fence_list);
- }
-
trace_i915_gem_object_change_domain(obj,
- obj->read_domains,
+ obj->base.read_domains,
old_write_domain);
}
}
return 0;
}
-/**
- * Command execution barrier
- *
- * Ensures that all commands in the ring are finished
- * before signalling the CPU
- */
-static void
-i915_retire_commands(struct drm_device *dev, struct intel_ring_buffer *ring)
-{
- uint32_t flush_domains = 0;
-
- /* The sampler always gets flushed on i965 (sigh) */
- if (INTEL_INFO(dev)->gen >= 4)
- flush_domains |= I915_GEM_DOMAIN_SAMPLER;
-
- ring->flush(ring, I915_GEM_DOMAIN_COMMAND, flush_domains);
-}
-
static inline void
i915_gem_request_remove_from_client(struct drm_i915_gem_request *request)
{
}
while (!list_empty(&ring->active_list)) {
- struct drm_i915_gem_object *obj_priv;
+ struct drm_i915_gem_object *obj;
- obj_priv = list_first_entry(&ring->active_list,
- struct drm_i915_gem_object,
- ring_list);
+ obj = list_first_entry(&ring->active_list,
+ struct drm_i915_gem_object,
+ ring_list);
- obj_priv->base.write_domain = 0;
- list_del_init(&obj_priv->gpu_write_list);
- i915_gem_object_move_to_inactive(&obj_priv->base);
+ obj->base.write_domain = 0;
+ list_del_init(&obj->gpu_write_list);
+ i915_gem_object_move_to_inactive(obj);
}
}
-void i915_gem_reset(struct drm_device *dev)
+static void i915_gem_reset_fences(struct drm_device *dev)
{
struct drm_i915_private *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv;
int i;
+ for (i = 0; i < 16; i++) {
+ struct drm_i915_fence_reg *reg = &dev_priv->fence_regs[i];
+ if (reg->obj)
+ i915_gem_clear_fence_reg(reg->obj);
+ }
+}
+
+void i915_gem_reset(struct drm_device *dev)
+{
+ struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_gem_object *obj;
+
i915_gem_reset_ring_lists(dev_priv, &dev_priv->render_ring);
i915_gem_reset_ring_lists(dev_priv, &dev_priv->bsd_ring);
i915_gem_reset_ring_lists(dev_priv, &dev_priv->blt_ring);
* lost bo to the inactive list.
*/
while (!list_empty(&dev_priv->mm.flushing_list)) {
- obj_priv = list_first_entry(&dev_priv->mm.flushing_list,
- struct drm_i915_gem_object,
- mm_list);
+ obj= list_first_entry(&dev_priv->mm.flushing_list,
+ struct drm_i915_gem_object,
+ mm_list);
- obj_priv->base.write_domain = 0;
- list_del_init(&obj_priv->gpu_write_list);
- i915_gem_object_move_to_inactive(&obj_priv->base);
+ obj->base.write_domain = 0;
+ list_del_init(&obj->gpu_write_list);
+ i915_gem_object_move_to_inactive(obj);
}
/* Move everything out of the GPU domains to ensure we do any
* necessary invalidation upon reuse.
*/
- list_for_each_entry(obj_priv,
+ list_for_each_entry(obj,
&dev_priv->mm.inactive_list,
mm_list)
{
- obj_priv->base.read_domains &= ~I915_GEM_GPU_DOMAINS;
+ obj->base.read_domains &= ~I915_GEM_GPU_DOMAINS;
}
/* The fence registers are invalidated so clear them out */
- for (i = 0; i < 16; i++) {
- struct drm_i915_fence_reg *reg;
-
- reg = &dev_priv->fence_regs[i];
- if (!reg->obj)
- continue;
-
- i915_gem_clear_fence_reg(reg->obj);
- }
+ i915_gem_reset_fences(dev);
}
/**
* by the ringbuffer to the flushing/inactive lists as appropriate.
*/
while (!list_empty(&ring->active_list)) {
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
+ struct drm_i915_gem_object *obj;
- obj_priv = list_first_entry(&ring->active_list,
- struct drm_i915_gem_object,
- ring_list);
+ obj= list_first_entry(&ring->active_list,
+ struct drm_i915_gem_object,
+ ring_list);
- if (!i915_seqno_passed(seqno, obj_priv->last_rendering_seqno))
+ if (!i915_seqno_passed(seqno, obj->last_rendering_seqno))
break;
- obj = &obj_priv->base;
- if (obj->write_domain != 0)
+ if (obj->base.write_domain != 0)
i915_gem_object_move_to_flushing(obj);
else
i915_gem_object_move_to_inactive(obj);
drm_i915_private_t *dev_priv = dev->dev_private;
if (!list_empty(&dev_priv->mm.deferred_free_list)) {
- struct drm_i915_gem_object *obj_priv, *tmp;
+ struct drm_i915_gem_object *obj, *next;
/* We must be careful that during unbind() we do not
* accidentally infinitely recurse into retire requests.
* Currently:
* retire -> free -> unbind -> wait -> retire_ring
*/
- list_for_each_entry_safe(obj_priv, tmp,
+ list_for_each_entry_safe(obj, next,
&dev_priv->mm.deferred_free_list,
mm_list)
- i915_gem_free_object_tail(&obj_priv->base);
+ i915_gem_free_object_tail(obj);
}
i915_gem_retire_requests_ring(dev, &dev_priv->render_ring);
if (atomic_read(&dev_priv->mm.wedged))
return -EAGAIN;
- if (ring->outstanding_lazy_request) {
+ if (seqno == ring->outstanding_lazy_request) {
struct drm_i915_gem_request *request;
request = kzalloc(sizeof(*request), GFP_KERNEL);
seqno = request->seqno;
}
- BUG_ON(seqno == dev_priv->next_seqno);
if (!i915_seqno_passed(ring->get_seqno(ring), seqno)) {
if (HAS_PCH_SPLIT(dev))
return i915_do_wait_request(dev, seqno, 1, ring);
}
-static void
-i915_gem_flush_ring(struct drm_device *dev,
- struct drm_file *file_priv,
- struct intel_ring_buffer *ring,
- uint32_t invalidate_domains,
- uint32_t flush_domains)
-{
- ring->flush(ring, invalidate_domains, flush_domains);
- i915_gem_process_flushing_list(dev, flush_domains, ring);
-}
-
-static void
-i915_gem_flush(struct drm_device *dev,
- struct drm_file *file_priv,
- uint32_t invalidate_domains,
- uint32_t flush_domains,
- uint32_t flush_rings)
-{
- drm_i915_private_t *dev_priv = dev->dev_private;
-
- if (flush_domains & I915_GEM_DOMAIN_CPU)
- drm_agp_chipset_flush(dev);
-
- if ((flush_domains | invalidate_domains) & I915_GEM_GPU_DOMAINS) {
- if (flush_rings & RING_RENDER)
- i915_gem_flush_ring(dev, file_priv,
- &dev_priv->render_ring,
- invalidate_domains, flush_domains);
- if (flush_rings & RING_BSD)
- i915_gem_flush_ring(dev, file_priv,
- &dev_priv->bsd_ring,
- invalidate_domains, flush_domains);
- if (flush_rings & RING_BLT)
- i915_gem_flush_ring(dev, file_priv,
- &dev_priv->blt_ring,
- invalidate_domains, flush_domains);
- }
-}
-
/**
* Ensures that all rendering to the object has completed and the object is
* safe to unbind from the GTT or access from the CPU.
*/
-static int
-i915_gem_object_wait_rendering(struct drm_gem_object *obj,
+int
+i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj,
bool interruptible)
{
- struct drm_device *dev = obj->dev;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+ struct drm_device *dev = obj->base.dev;
int ret;
/* This function only exists to support waiting for existing rendering,
* not for emitting required flushes.
*/
- BUG_ON((obj->write_domain & I915_GEM_GPU_DOMAINS) != 0);
+ BUG_ON((obj->base.write_domain & I915_GEM_GPU_DOMAINS) != 0);
/* If there is rendering queued on the buffer being evicted, wait for
* it.
*/
- if (obj_priv->active) {
+ if (obj->active) {
ret = i915_do_wait_request(dev,
- obj_priv->last_rendering_seqno,
+ obj->last_rendering_seqno,
interruptible,
- obj_priv->ring);
+ obj->ring);
if (ret)
return ret;
}
* Unbinds an object from the GTT aperture.
*/
int
-i915_gem_object_unbind(struct drm_gem_object *obj)
+i915_gem_object_unbind(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
int ret = 0;
- if (obj_priv->gtt_space == NULL)
+ if (obj->gtt_space == NULL)
return 0;
- if (obj_priv->pin_count != 0) {
+ if (obj->pin_count != 0) {
DRM_ERROR("Attempting to unbind pinned buffer\n");
return -EINVAL;
}
*/
if (ret) {
i915_gem_clflush_object(obj);
- obj->read_domains = obj->write_domain = I915_GEM_DOMAIN_CPU;
+ obj->base.read_domains = obj->base.write_domain = I915_GEM_DOMAIN_CPU;
}
/* release the fence reg _after_ flushing */
- if (obj_priv->fence_reg != I915_FENCE_REG_NONE)
+ if (obj->fence_reg != I915_FENCE_REG_NONE)
i915_gem_clear_fence_reg(obj);
- drm_unbind_agp(obj_priv->agp_mem);
- drm_free_agp(obj_priv->agp_mem, obj->size / PAGE_SIZE);
-
+ i915_gem_gtt_unbind_object(obj);
i915_gem_object_put_pages_gtt(obj);
- i915_gem_info_remove_gtt(dev_priv, obj_priv);
- list_del_init(&obj_priv->mm_list);
- obj_priv->fenceable = true;
- obj_priv->mappable = true;
+ list_del_init(&obj->gtt_list);
+ list_del_init(&obj->mm_list);
+ /* Avoid an unnecessary call to unbind on rebind. */
+ obj->map_and_fenceable = true;
- drm_mm_put_block(obj_priv->gtt_space);
- obj_priv->gtt_space = NULL;
- obj_priv->gtt_offset = 0;
+ drm_mm_put_block(obj->gtt_space);
+ obj->gtt_space = NULL;
+ obj->gtt_offset = 0;
- if (i915_gem_object_is_purgeable(obj_priv))
+ if (i915_gem_object_is_purgeable(obj))
i915_gem_object_truncate(obj);
trace_i915_gem_object_unbind(obj);
return ret;
}
+void
+i915_gem_flush_ring(struct drm_device *dev,
+ struct intel_ring_buffer *ring,
+ uint32_t invalidate_domains,
+ uint32_t flush_domains)
+{
+ ring->flush(ring, invalidate_domains, flush_domains);
+ i915_gem_process_flushing_list(dev, flush_domains, ring);
+}
+
static int i915_ring_idle(struct drm_device *dev,
struct intel_ring_buffer *ring)
{
- if (list_empty(&ring->gpu_write_list))
+ if (list_empty(&ring->gpu_write_list) && list_empty(&ring->active_list))
return 0;
- i915_gem_flush_ring(dev, NULL, ring,
+ i915_gem_flush_ring(dev, ring,
I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS);
return i915_wait_request(dev,
i915_gem_next_request_seqno(dev, ring),
int ret;
lists_empty = (list_empty(&dev_priv->mm.flushing_list) &&
- list_empty(&dev_priv->render_ring.active_list) &&
- list_empty(&dev_priv->bsd_ring.active_list) &&
- list_empty(&dev_priv->blt_ring.active_list));
+ list_empty(&dev_priv->mm.active_list));
if (lists_empty)
return 0;
return 0;
}
-static void sandybridge_write_fence_reg(struct drm_gem_object *obj)
+static int sandybridge_write_fence_reg(struct drm_i915_gem_object *obj,
+ struct intel_ring_buffer *pipelined)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- u32 size = i915_gem_get_gtt_size(obj_priv);
- int regnum = obj_priv->fence_reg;
+ u32 size = obj->gtt_space->size;
+ int regnum = obj->fence_reg;
uint64_t val;
- val = (uint64_t)((obj_priv->gtt_offset + size - 4096) &
- 0xfffff000) << 32;
- val |= obj_priv->gtt_offset & 0xfffff000;
- val |= (uint64_t)((obj_priv->stride / 128) - 1) <<
+ val = (uint64_t)((obj->gtt_offset + size - 4096) &
+ 0xfffff000) << 32;
+ val |= obj->gtt_offset & 0xfffff000;
+ val |= (uint64_t)((obj->stride / 128) - 1) <<
SANDYBRIDGE_FENCE_PITCH_SHIFT;
- if (obj_priv->tiling_mode == I915_TILING_Y)
+ if (obj->tiling_mode == I915_TILING_Y)
val |= 1 << I965_FENCE_TILING_Y_SHIFT;
val |= I965_FENCE_REG_VALID;
- I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 + (regnum * 8), val);
+ if (pipelined) {
+ int ret = intel_ring_begin(pipelined, 6);
+ if (ret)
+ return ret;
+
+ intel_ring_emit(pipelined, MI_NOOP);
+ intel_ring_emit(pipelined, MI_LOAD_REGISTER_IMM(2));
+ intel_ring_emit(pipelined, FENCE_REG_SANDYBRIDGE_0 + regnum*8);
+ intel_ring_emit(pipelined, (u32)val);
+ intel_ring_emit(pipelined, FENCE_REG_SANDYBRIDGE_0 + regnum*8 + 4);
+ intel_ring_emit(pipelined, (u32)(val >> 32));
+ intel_ring_advance(pipelined);
+ } else
+ I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 + regnum * 8, val);
+
+ return 0;
}
-static void i965_write_fence_reg(struct drm_gem_object *obj)
+static int i965_write_fence_reg(struct drm_i915_gem_object *obj,
+ struct intel_ring_buffer *pipelined)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- u32 size = i915_gem_get_gtt_size(obj_priv);
- int regnum = obj_priv->fence_reg;
+ u32 size = obj->gtt_space->size;
+ int regnum = obj->fence_reg;
uint64_t val;
- val = (uint64_t)((obj_priv->gtt_offset + size - 4096) &
+ val = (uint64_t)((obj->gtt_offset + size - 4096) &
0xfffff000) << 32;
- val |= obj_priv->gtt_offset & 0xfffff000;
- val |= ((obj_priv->stride / 128) - 1) << I965_FENCE_PITCH_SHIFT;
- if (obj_priv->tiling_mode == I915_TILING_Y)
+ val |= obj->gtt_offset & 0xfffff000;
+ val |= ((obj->stride / 128) - 1) << I965_FENCE_PITCH_SHIFT;
+ if (obj->tiling_mode == I915_TILING_Y)
val |= 1 << I965_FENCE_TILING_Y_SHIFT;
val |= I965_FENCE_REG_VALID;
- I915_WRITE64(FENCE_REG_965_0 + (regnum * 8), val);
+ if (pipelined) {
+ int ret = intel_ring_begin(pipelined, 6);
+ if (ret)
+ return ret;
+
+ intel_ring_emit(pipelined, MI_NOOP);
+ intel_ring_emit(pipelined, MI_LOAD_REGISTER_IMM(2));
+ intel_ring_emit(pipelined, FENCE_REG_965_0 + regnum*8);
+ intel_ring_emit(pipelined, (u32)val);
+ intel_ring_emit(pipelined, FENCE_REG_965_0 + regnum*8 + 4);
+ intel_ring_emit(pipelined, (u32)(val >> 32));
+ intel_ring_advance(pipelined);
+ } else
+ I915_WRITE64(FENCE_REG_965_0 + regnum * 8, val);
+
+ return 0;
}
-static void i915_write_fence_reg(struct drm_gem_object *obj)
+static int i915_write_fence_reg(struct drm_i915_gem_object *obj,
+ struct intel_ring_buffer *pipelined)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- u32 size = i915_gem_get_gtt_size(obj_priv);
- uint32_t fence_reg, val, pitch_val;
+ u32 size = obj->gtt_space->size;
+ u32 fence_reg, val, pitch_val;
int tile_width;
- if ((obj_priv->gtt_offset & ~I915_FENCE_START_MASK) ||
- (obj_priv->gtt_offset & (size - 1))) {
- WARN(1, "%s: object 0x%08x [fenceable? %d] not 1M or size (0x%08x) aligned [gtt_space offset=%lx, size=%lx]\n",
- __func__, obj_priv->gtt_offset, obj_priv->fenceable, size,
- obj_priv->gtt_space->start, obj_priv->gtt_space->size);
- return;
- }
+ if (WARN((obj->gtt_offset & ~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))
+ return -EINVAL;
- if (obj_priv->tiling_mode == I915_TILING_Y &&
- HAS_128_BYTE_Y_TILING(dev))
+ if (obj->tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev))
tile_width = 128;
else
tile_width = 512;
/* Note: pitch better be a power of two tile widths */
- pitch_val = obj_priv->stride / tile_width;
+ pitch_val = obj->stride / tile_width;
pitch_val = ffs(pitch_val) - 1;
- if (obj_priv->tiling_mode == I915_TILING_Y &&
- HAS_128_BYTE_Y_TILING(dev))
- WARN_ON(pitch_val > I830_FENCE_MAX_PITCH_VAL);
- else
- WARN_ON(pitch_val > I915_FENCE_MAX_PITCH_VAL);
-
- val = obj_priv->gtt_offset;
- if (obj_priv->tiling_mode == I915_TILING_Y)
+ val = obj->gtt_offset;
+ if (obj->tiling_mode == I915_TILING_Y)
val |= 1 << I830_FENCE_TILING_Y_SHIFT;
val |= I915_FENCE_SIZE_BITS(size);
val |= pitch_val << I830_FENCE_PITCH_SHIFT;
val |= I830_FENCE_REG_VALID;
- fence_reg = obj_priv->fence_reg;
+ fence_reg = obj->fence_reg;
if (fence_reg < 8)
fence_reg = FENCE_REG_830_0 + fence_reg * 4;
else
fence_reg = FENCE_REG_945_8 + (fence_reg - 8) * 4;
- I915_WRITE(fence_reg, val);
+
+ if (pipelined) {
+ int ret = intel_ring_begin(pipelined, 4);
+ if (ret)
+ return ret;
+
+ intel_ring_emit(pipelined, MI_NOOP);
+ intel_ring_emit(pipelined, MI_LOAD_REGISTER_IMM(1));
+ intel_ring_emit(pipelined, fence_reg);
+ intel_ring_emit(pipelined, val);
+ intel_ring_advance(pipelined);
+ } else
+ I915_WRITE(fence_reg, val);
+
+ return 0;
}
-static void i830_write_fence_reg(struct drm_gem_object *obj)
+static int i830_write_fence_reg(struct drm_i915_gem_object *obj,
+ struct intel_ring_buffer *pipelined)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- u32 size = i915_gem_get_gtt_size(obj_priv);
- int regnum = obj_priv->fence_reg;
+ u32 size = obj->gtt_space->size;
+ int regnum = obj->fence_reg;
uint32_t val;
uint32_t pitch_val;
- uint32_t fence_size_bits;
- if ((obj_priv->gtt_offset & ~I830_FENCE_START_MASK) ||
- (obj_priv->gtt_offset & (obj->size - 1))) {
- WARN(1, "%s: object 0x%08x not 512K or size aligned\n",
- __func__, obj_priv->gtt_offset);
- return;
- }
+ if (WARN((obj->gtt_offset & ~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))
+ return -EINVAL;
- pitch_val = obj_priv->stride / 128;
+ pitch_val = obj->stride / 128;
pitch_val = ffs(pitch_val) - 1;
- WARN_ON(pitch_val > I830_FENCE_MAX_PITCH_VAL);
- val = obj_priv->gtt_offset;
- if (obj_priv->tiling_mode == I915_TILING_Y)
+ val = obj->gtt_offset;
+ if (obj->tiling_mode == I915_TILING_Y)
val |= 1 << I830_FENCE_TILING_Y_SHIFT;
- fence_size_bits = I830_FENCE_SIZE_BITS(size);
- WARN_ON(fence_size_bits & ~0x00000f00);
- val |= fence_size_bits;
+ val |= I830_FENCE_SIZE_BITS(size);
val |= pitch_val << I830_FENCE_PITCH_SHIFT;
val |= I830_FENCE_REG_VALID;
- I915_WRITE(FENCE_REG_830_0 + (regnum * 4), val);
+ if (pipelined) {
+ int ret = intel_ring_begin(pipelined, 4);
+ if (ret)
+ return ret;
+
+ intel_ring_emit(pipelined, MI_NOOP);
+ intel_ring_emit(pipelined, MI_LOAD_REGISTER_IMM(1));
+ intel_ring_emit(pipelined, FENCE_REG_830_0 + regnum*4);
+ intel_ring_emit(pipelined, val);
+ intel_ring_advance(pipelined);
+ } else
+ I915_WRITE(FENCE_REG_830_0 + regnum * 4, val);
+
+ return 0;
}
static int i915_find_fence_reg(struct drm_device *dev,
{
struct drm_i915_private *dev_priv = dev->dev_private;
struct drm_i915_fence_reg *reg;
- struct drm_i915_gem_object *obj_priv = NULL;
+ struct drm_i915_gem_object *obj = NULL;
int i, avail, ret;
/* First try to find a free reg */
if (!reg->obj)
return i;
- obj_priv = to_intel_bo(reg->obj);
- if (!obj_priv->pin_count)
- avail++;
+ if (!reg->obj->pin_count)
+ avail++;
}
if (avail == 0)
avail = I915_FENCE_REG_NONE;
list_for_each_entry(reg, &dev_priv->mm.fence_list,
lru_list) {
- obj_priv = to_intel_bo(reg->obj);
- if (obj_priv->pin_count)
+ obj = reg->obj;
+ if (obj->pin_count)
continue;
/* found one! */
- avail = obj_priv->fence_reg;
+ avail = obj->fence_reg;
break;
}
* might drop that one, causing a use-after-free in it. So hold a
* private reference to obj like the other callers of put_fence_reg
* (set_tiling ioctl) do. */
- drm_gem_object_reference(&obj_priv->base);
- ret = i915_gem_object_put_fence_reg(&obj_priv->base, interruptible);
- drm_gem_object_unreference(&obj_priv->base);
+ drm_gem_object_reference(&obj->base);
+ ret = i915_gem_object_put_fence_reg(obj, interruptible);
+ drm_gem_object_unreference(&obj->base);
if (ret != 0)
return ret;
* and tiling format.
*/
int
-i915_gem_object_get_fence_reg(struct drm_gem_object *obj,
+i915_gem_object_get_fence_reg(struct drm_i915_gem_object *obj,
bool interruptible)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
struct drm_i915_private *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
struct drm_i915_fence_reg *reg = NULL;
+ struct intel_ring_buffer *pipelined = NULL;
int ret;
/* Just update our place in the LRU if our fence is getting used. */
- if (obj_priv->fence_reg != I915_FENCE_REG_NONE) {
- reg = &dev_priv->fence_regs[obj_priv->fence_reg];
+ if (obj->fence_reg != I915_FENCE_REG_NONE) {
+ reg = &dev_priv->fence_regs[obj->fence_reg];
list_move_tail(®->lru_list, &dev_priv->mm.fence_list);
return 0;
}
- switch (obj_priv->tiling_mode) {
+ switch (obj->tiling_mode) {
case I915_TILING_NONE:
WARN(1, "allocating a fence for non-tiled object?\n");
break;
case I915_TILING_X:
- if (!obj_priv->stride)
+ if (!obj->stride)
return -EINVAL;
- WARN((obj_priv->stride & (512 - 1)),
+ WARN((obj->stride & (512 - 1)),
"object 0x%08x is X tiled but has non-512B pitch\n",
- obj_priv->gtt_offset);
+ obj->gtt_offset);
break;
case I915_TILING_Y:
- if (!obj_priv->stride)
+ if (!obj->stride)
return -EINVAL;
- WARN((obj_priv->stride & (128 - 1)),
+ WARN((obj->stride & (128 - 1)),
"object 0x%08x is Y tiled but has non-128B pitch\n",
- obj_priv->gtt_offset);
+ obj->gtt_offset);
break;
}
if (ret < 0)
return ret;
- obj_priv->fence_reg = ret;
- reg = &dev_priv->fence_regs[obj_priv->fence_reg];
+ obj->fence_reg = ret;
+ reg = &dev_priv->fence_regs[obj->fence_reg];
list_add_tail(®->lru_list, &dev_priv->mm.fence_list);
reg->obj = obj;
switch (INTEL_INFO(dev)->gen) {
case 6:
- sandybridge_write_fence_reg(obj);
+ ret = sandybridge_write_fence_reg(obj, pipelined);
break;
case 5:
case 4:
- i965_write_fence_reg(obj);
+ ret = i965_write_fence_reg(obj, pipelined);
break;
case 3:
- i915_write_fence_reg(obj);
+ ret = i915_write_fence_reg(obj, pipelined);
break;
case 2:
- i830_write_fence_reg(obj);
+ ret = i830_write_fence_reg(obj, pipelined);
break;
}
trace_i915_gem_object_get_fence(obj,
- obj_priv->fence_reg,
- obj_priv->tiling_mode);
-
- return 0;
+ obj->fence_reg,
+ obj->tiling_mode);
+ return ret;
}
/**
* @obj: object to clear
*
* Zeroes out the fence register itself and clears out the associated
- * data structures in dev_priv and obj_priv.
+ * data structures in dev_priv and obj.
*/
static void
-i915_gem_clear_fence_reg(struct drm_gem_object *obj)
+i915_gem_clear_fence_reg(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- struct drm_i915_fence_reg *reg =
- &dev_priv->fence_regs[obj_priv->fence_reg];
+ struct drm_i915_fence_reg *reg = &dev_priv->fence_regs[obj->fence_reg];
uint32_t fence_reg;
switch (INTEL_INFO(dev)->gen) {
case 6:
I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 +
- (obj_priv->fence_reg * 8), 0);
+ (obj->fence_reg * 8), 0);
break;
case 5:
case 4:
- I915_WRITE64(FENCE_REG_965_0 + (obj_priv->fence_reg * 8), 0);
+ I915_WRITE64(FENCE_REG_965_0 + (obj->fence_reg * 8), 0);
break;
case 3:
- if (obj_priv->fence_reg >= 8)
- fence_reg = FENCE_REG_945_8 + (obj_priv->fence_reg - 8) * 4;
+ if (obj->fence_reg >= 8)
+ fence_reg = FENCE_REG_945_8 + (obj->fence_reg - 8) * 4;
else
case 2:
- fence_reg = FENCE_REG_830_0 + obj_priv->fence_reg * 4;
+ fence_reg = FENCE_REG_830_0 + obj->fence_reg * 4;
I915_WRITE(fence_reg, 0);
break;
}
reg->obj = NULL;
- obj_priv->fence_reg = I915_FENCE_REG_NONE;
+ obj->fence_reg = I915_FENCE_REG_NONE;
list_del_init(®->lru_list);
}
* @bool: whether the wait upon the fence is interruptible
*
* Zeroes out the fence register itself and clears out the associated
- * data structures in dev_priv and obj_priv.
+ * data structures in dev_priv and obj.
*/
int
-i915_gem_object_put_fence_reg(struct drm_gem_object *obj,
+i915_gem_object_put_fence_reg(struct drm_i915_gem_object *obj,
bool interruptible)
{
- struct drm_device *dev = obj->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- struct drm_i915_fence_reg *reg;
+ struct drm_device *dev = obj->base.dev;
+ int ret;
- if (obj_priv->fence_reg == I915_FENCE_REG_NONE)
+ if (obj->fence_reg == I915_FENCE_REG_NONE)
return 0;
/* If we've changed tiling, GTT-mappings of the object
* therefore we must wait for any outstanding access to complete
* before clearing the fence.
*/
- reg = &dev_priv->fence_regs[obj_priv->fence_reg];
- if (reg->gpu) {
- int ret;
-
- ret = i915_gem_object_flush_gpu_write_domain(obj, true);
+ if (obj->fenced_gpu_access) {
+ ret = i915_gem_object_flush_gpu_write_domain(obj, NULL);
if (ret)
return ret;
- ret = i915_gem_object_wait_rendering(obj, interruptible);
+ obj->fenced_gpu_access = false;
+ }
+
+ if (obj->last_fenced_seqno) {
+ ret = i915_do_wait_request(dev,
+ obj->last_fenced_seqno,
+ interruptible,
+ obj->last_fenced_ring);
if (ret)
return ret;
- reg->gpu = false;
+ obj->last_fenced_seqno = false;
}
i915_gem_object_flush_gtt_write_domain(obj);
* Finds free space in the GTT aperture and binds the object there.
*/
static int
-i915_gem_object_bind_to_gtt(struct drm_gem_object *obj,
+i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj,
unsigned alignment,
- bool mappable,
- bool need_fence)
+ bool map_and_fenceable)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
struct drm_mm_node *free_space;
gfp_t gfpmask = __GFP_NORETRY | __GFP_NOWARN;
- u32 size, fence_size, fence_alignment;
+ u32 size, fence_size, fence_alignment, unfenced_alignment;
+ bool mappable, fenceable;
int ret;
- if (obj_priv->madv != I915_MADV_WILLNEED) {
+ if (obj->madv != I915_MADV_WILLNEED) {
DRM_ERROR("Attempting to bind a purgeable object\n");
return -EINVAL;
}
- fence_size = i915_gem_get_gtt_size(obj_priv);
- fence_alignment = i915_gem_get_gtt_alignment(obj_priv);
+ fence_size = i915_gem_get_gtt_size(obj);
+ fence_alignment = i915_gem_get_gtt_alignment(obj);
+ unfenced_alignment = i915_gem_get_unfenced_gtt_alignment(obj);
if (alignment == 0)
- alignment = need_fence ? fence_alignment : 4096;
- if (need_fence && alignment & (fence_alignment - 1)) {
+ alignment = map_and_fenceable ? fence_alignment :
+ unfenced_alignment;
+ if (map_and_fenceable && alignment & (fence_alignment - 1)) {
DRM_ERROR("Invalid object alignment requested %u\n", alignment);
return -EINVAL;
}
- size = need_fence ? fence_size : obj->size;
+ size = map_and_fenceable ? fence_size : obj->base.size;
/* If the object is bigger than the entire aperture, reject it early
* before evicting everything in a vain attempt to find space.
*/
- if (obj->size >
- (mappable ? dev_priv->mm.gtt_mappable_end : dev_priv->mm.gtt_total)) {
+ if (obj->base.size >
+ (map_and_fenceable ? dev_priv->mm.gtt_mappable_end : dev_priv->mm.gtt_total)) {
DRM_ERROR("Attempting to bind an object larger than the aperture\n");
return -E2BIG;
}
search_free:
- if (mappable)
+ if (map_and_fenceable)
free_space =
drm_mm_search_free_in_range(&dev_priv->mm.gtt_space,
size, alignment, 0,
size, alignment, 0);
if (free_space != NULL) {
- if (mappable)
- obj_priv->gtt_space =
+ if (map_and_fenceable)
+ obj->gtt_space =
drm_mm_get_block_range_generic(free_space,
size, alignment, 0,
dev_priv->mm.gtt_mappable_end,
0);
else
- obj_priv->gtt_space =
+ obj->gtt_space =
drm_mm_get_block(free_space, size, alignment);
}
- if (obj_priv->gtt_space == NULL) {
+ if (obj->gtt_space == NULL) {
/* If the gtt is empty and we're still having trouble
* fitting our object in, we're out of memory.
*/
- ret = i915_gem_evict_something(dev, size, alignment, mappable);
+ ret = i915_gem_evict_something(dev, size, alignment,
+ map_and_fenceable);
if (ret)
return ret;
ret = i915_gem_object_get_pages_gtt(obj, gfpmask);
if (ret) {
- drm_mm_put_block(obj_priv->gtt_space);
- obj_priv->gtt_space = NULL;
+ drm_mm_put_block(obj->gtt_space);
+ obj->gtt_space = NULL;
if (ret == -ENOMEM) {
/* first try to clear up some space from the GTT */
ret = i915_gem_evict_something(dev, size,
- alignment, mappable);
+ alignment,
+ map_and_fenceable);
if (ret) {
/* now try to shrink everyone else */
if (gfpmask) {
return ret;
}
- /* Create an AGP memory structure pointing at our pages, and bind it
- * into the GTT.
- */
- obj_priv->agp_mem = drm_agp_bind_pages(dev,
- obj_priv->pages,
- obj->size >> PAGE_SHIFT,
- obj_priv->gtt_space->start,
- obj_priv->agp_type);
- if (obj_priv->agp_mem == NULL) {
+ ret = i915_gem_gtt_bind_object(obj);
+ if (ret) {
i915_gem_object_put_pages_gtt(obj);
- drm_mm_put_block(obj_priv->gtt_space);
- obj_priv->gtt_space = NULL;
+ drm_mm_put_block(obj->gtt_space);
+ obj->gtt_space = NULL;
ret = i915_gem_evict_something(dev, size,
- alignment, mappable);
+ alignment, map_and_fenceable);
if (ret)
return ret;
goto search_free;
}
- obj_priv->gtt_offset = obj_priv->gtt_space->start;
-
- /* keep track of bounds object by adding it to the inactive list */
- list_add_tail(&obj_priv->mm_list, &dev_priv->mm.inactive_list);
- i915_gem_info_add_gtt(dev_priv, obj_priv);
+ list_add_tail(&obj->gtt_list, &dev_priv->mm.gtt_list);
+ list_add_tail(&obj->mm_list, &dev_priv->mm.inactive_list);
/* Assert that the object is not currently in any GPU domain. As it
* wasn't in the GTT, there shouldn't be any way it could have been in
* a GPU cache
*/
- BUG_ON(obj->read_domains & I915_GEM_GPU_DOMAINS);
- BUG_ON(obj->write_domain & I915_GEM_GPU_DOMAINS);
+ BUG_ON(obj->base.read_domains & I915_GEM_GPU_DOMAINS);
+ BUG_ON(obj->base.write_domain & I915_GEM_GPU_DOMAINS);
+
+ obj->gtt_offset = obj->gtt_space->start;
- trace_i915_gem_object_bind(obj, obj_priv->gtt_offset, mappable);
+ fenceable =
+ obj->gtt_space->size == fence_size &&
+ (obj->gtt_space->start & (fence_alignment -1)) == 0;
- obj_priv->fenceable =
- obj_priv->gtt_space->size == fence_size &&
- (obj_priv->gtt_space->start & (fence_alignment -1)) == 0;
+ mappable =
+ obj->gtt_offset + obj->base.size <= dev_priv->mm.gtt_mappable_end;
- obj_priv->mappable =
- obj_priv->gtt_offset + obj->size <= dev_priv->mm.gtt_mappable_end;
+ obj->map_and_fenceable = mappable && fenceable;
+ trace_i915_gem_object_bind(obj, obj->gtt_offset, map_and_fenceable);
return 0;
}
void
-i915_gem_clflush_object(struct drm_gem_object *obj)
+i915_gem_clflush_object(struct drm_i915_gem_object *obj)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
-
/* If we don't have a page list set up, then we're not pinned
* to GPU, and we can ignore the cache flush because it'll happen
* again at bind time.
*/
- if (obj_priv->pages == NULL)
+ if (obj->pages == NULL)
return;
trace_i915_gem_object_clflush(obj);
- drm_clflush_pages(obj_priv->pages, obj->size / PAGE_SIZE);
+ drm_clflush_pages(obj->pages, obj->base.size / PAGE_SIZE);
}
/** Flushes any GPU write domain for the object if it's dirty. */
static int
-i915_gem_object_flush_gpu_write_domain(struct drm_gem_object *obj,
- bool pipelined)
+i915_gem_object_flush_gpu_write_domain(struct drm_i915_gem_object *obj,
+ struct intel_ring_buffer *pipelined)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
- if ((obj->write_domain & I915_GEM_GPU_DOMAINS) == 0)
+ if ((obj->base.write_domain & I915_GEM_GPU_DOMAINS) == 0)
return 0;
/* Queue the GPU write cache flushing we need. */
- i915_gem_flush_ring(dev, NULL,
- to_intel_bo(obj)->ring,
- 0, obj->write_domain);
- BUG_ON(obj->write_domain);
+ i915_gem_flush_ring(dev, obj->ring, 0, obj->base.write_domain);
+ BUG_ON(obj->base.write_domain);
- if (pipelined)
+ if (pipelined && pipelined == obj->ring)
return 0;
return i915_gem_object_wait_rendering(obj, true);
/** Flushes the GTT write domain for the object if it's dirty. */
static void
-i915_gem_object_flush_gtt_write_domain(struct drm_gem_object *obj)
+i915_gem_object_flush_gtt_write_domain(struct drm_i915_gem_object *obj)
{
uint32_t old_write_domain;
- if (obj->write_domain != I915_GEM_DOMAIN_GTT)
+ if (obj->base.write_domain != I915_GEM_DOMAIN_GTT)
return;
/* No actual flushing is required for the GTT write domain. Writes
*/
i915_gem_release_mmap(obj);
- old_write_domain = obj->write_domain;
- obj->write_domain = 0;
+ old_write_domain = obj->base.write_domain;
+ obj->base.write_domain = 0;
trace_i915_gem_object_change_domain(obj,
- obj->read_domains,
+ obj->base.read_domains,
old_write_domain);
}
/** Flushes the CPU write domain for the object if it's dirty. */
static void
-i915_gem_object_flush_cpu_write_domain(struct drm_gem_object *obj)
+i915_gem_object_flush_cpu_write_domain(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
uint32_t old_write_domain;
- if (obj->write_domain != I915_GEM_DOMAIN_CPU)
+ if (obj->base.write_domain != I915_GEM_DOMAIN_CPU)
return;
i915_gem_clflush_object(obj);
- drm_agp_chipset_flush(dev);
- old_write_domain = obj->write_domain;
- obj->write_domain = 0;
+ intel_gtt_chipset_flush();
+ old_write_domain = obj->base.write_domain;
+ obj->base.write_domain = 0;
trace_i915_gem_object_change_domain(obj,
- obj->read_domains,
+ obj->base.read_domains,
old_write_domain);
}
* flushes to occur.
*/
int
-i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj, int write)
+i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj, bool write)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
uint32_t old_write_domain, old_read_domains;
int ret;
/* Not valid to be called on unbound objects. */
- if (obj_priv->gtt_space == NULL)
+ if (obj->gtt_space == NULL)
return -EINVAL;
- ret = i915_gem_object_flush_gpu_write_domain(obj, false);
+ ret = i915_gem_object_flush_gpu_write_domain(obj, NULL);
if (ret != 0)
return ret;
return ret;
}
- old_write_domain = obj->write_domain;
- old_read_domains = obj->read_domains;
+ old_write_domain = obj->base.write_domain;
+ old_read_domains = obj->base.read_domains;
/* It should now be out of any other write domains, and we can update
* the domain values for our changes.
*/
- BUG_ON((obj->write_domain & ~I915_GEM_DOMAIN_GTT) != 0);
- obj->read_domains |= I915_GEM_DOMAIN_GTT;
+ BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_GTT) != 0);
+ obj->base.read_domains |= I915_GEM_DOMAIN_GTT;
if (write) {
- obj->read_domains = I915_GEM_DOMAIN_GTT;
- obj->write_domain = I915_GEM_DOMAIN_GTT;
- obj_priv->dirty = 1;
+ obj->base.read_domains = I915_GEM_DOMAIN_GTT;
+ obj->base.write_domain = I915_GEM_DOMAIN_GTT;
+ obj->dirty = 1;
}
trace_i915_gem_object_change_domain(obj,
* wait, as in modesetting process we're not supposed to be interrupted.
*/
int
-i915_gem_object_set_to_display_plane(struct drm_gem_object *obj,
- bool pipelined)
+i915_gem_object_set_to_display_plane(struct drm_i915_gem_object *obj,
+ struct intel_ring_buffer *pipelined)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
uint32_t old_read_domains;
int ret;
/* Not valid to be called on unbound objects. */
- if (obj_priv->gtt_space == NULL)
+ if (obj->gtt_space == NULL)
return -EINVAL;
- ret = i915_gem_object_flush_gpu_write_domain(obj, true);
+ ret = i915_gem_object_flush_gpu_write_domain(obj, pipelined);
if (ret)
return ret;
i915_gem_object_flush_cpu_write_domain(obj);
- old_read_domains = obj->read_domains;
- obj->read_domains |= I915_GEM_DOMAIN_GTT;
+ old_read_domains = obj->base.read_domains;
+ obj->base.read_domains |= I915_GEM_DOMAIN_GTT;
trace_i915_gem_object_change_domain(obj,
old_read_domains,
- obj->write_domain);
+ obj->base.write_domain);
return 0;
}
+int
+i915_gem_object_flush_gpu(struct drm_i915_gem_object *obj,
+ bool interruptible)
+{
+ if (!obj->active)
+ return 0;
+
+ if (obj->base.write_domain & I915_GEM_GPU_DOMAINS)
+ i915_gem_flush_ring(obj->base.dev, obj->ring,
+ 0, obj->base.write_domain);
+
+ return i915_gem_object_wait_rendering(obj, interruptible);
+}
+
/**
* Moves a single object to the CPU read, and possibly write domain.
*
* flushes to occur.
*/
static int
-i915_gem_object_set_to_cpu_domain(struct drm_gem_object *obj, int write)
+i915_gem_object_set_to_cpu_domain(struct drm_i915_gem_object *obj, bool write)
{
uint32_t old_write_domain, old_read_domains;
int ret;
return ret;
}
- old_write_domain = obj->write_domain;
- old_read_domains = obj->read_domains;
+ old_write_domain = obj->base.write_domain;
+ old_read_domains = obj->base.read_domains;
/* Flush the CPU cache if it's still invalid. */
- if ((obj->read_domains & I915_GEM_DOMAIN_CPU) == 0) {
+ if ((obj->base.read_domains & I915_GEM_DOMAIN_CPU) == 0) {
i915_gem_clflush_object(obj);
- obj->read_domains |= I915_GEM_DOMAIN_CPU;
+ obj->base.read_domains |= I915_GEM_DOMAIN_CPU;
}
/* It should now be out of any other write domains, and we can update
* the domain values for our changes.
*/
- BUG_ON((obj->write_domain & ~I915_GEM_DOMAIN_CPU) != 0);
+ BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_CPU) != 0);
/* If we're writing through the CPU, then the GPU read domains will
* need to be invalidated at next use.
*/
if (write) {
- obj->read_domains = I915_GEM_DOMAIN_CPU;
- obj->write_domain = I915_GEM_DOMAIN_CPU;
+ obj->base.read_domains = I915_GEM_DOMAIN_CPU;
+ obj->base.write_domain = I915_GEM_DOMAIN_CPU;
}
trace_i915_gem_object_change_domain(obj,
return 0;
}
-/*
- * Set the next domain for the specified object. This
- * may not actually perform the necessary flushing/invaliding though,
- * as that may want to be batched with other set_domain operations
- *
- * This is (we hope) the only really tricky part of gem. The goal
- * is fairly simple -- track which caches hold bits of the object
- * and make sure they remain coherent. A few concrete examples may
- * help to explain how it works. For shorthand, we use the notation
- * (read_domains, write_domain), e.g. (CPU, CPU) to indicate the
- * a pair of read and write domain masks.
- *
- * Case 1: the batch buffer
- *
- * 1. Allocated
- * 2. Written by CPU
- * 3. Mapped to GTT
- * 4. Read by GPU
- * 5. Unmapped from GTT
- * 6. Freed
- *
- * Let's take these a step at a time
- *
- * 1. Allocated
- * Pages allocated from the kernel may still have
- * cache contents, so we set them to (CPU, CPU) always.
- * 2. Written by CPU (using pwrite)
- * The pwrite function calls set_domain (CPU, CPU) and
- * this function does nothing (as nothing changes)
- * 3. Mapped by GTT
- * This function asserts that the object is not
- * currently in any GPU-based read or write domains
- * 4. Read by GPU
- * i915_gem_execbuffer calls set_domain (COMMAND, 0).
- * As write_domain is zero, this function adds in the
- * current read domains (CPU+COMMAND, 0).
- * flush_domains is set to CPU.
- * invalidate_domains is set to COMMAND
- * clflush is run to get data out of the CPU caches
- * then i915_dev_set_domain calls i915_gem_flush to
- * emit an MI_FLUSH and drm_agp_chipset_flush
- * 5. Unmapped from GTT
- * i915_gem_object_unbind calls set_domain (CPU, CPU)
- * flush_domains and invalidate_domains end up both zero
- * so no flushing/invalidating happens
- * 6. Freed
- * yay, done
- *
- * Case 2: The shared render buffer
- *
- * 1. Allocated
- * 2. Mapped to GTT
- * 3. Read/written by GPU
- * 4. set_domain to (CPU,CPU)
- * 5. Read/written by CPU
- * 6. Read/written by GPU
- *
- * 1. Allocated
- * Same as last example, (CPU, CPU)
- * 2. Mapped to GTT
- * Nothing changes (assertions find that it is not in the GPU)
- * 3. Read/written by GPU
- * execbuffer calls set_domain (RENDER, RENDER)
- * flush_domains gets CPU
- * invalidate_domains gets GPU
- * clflush (obj)
- * MI_FLUSH and drm_agp_chipset_flush
- * 4. set_domain (CPU, CPU)
- * flush_domains gets GPU
- * invalidate_domains gets CPU
- * wait_rendering (obj) to make sure all drawing is complete.
- * This will include an MI_FLUSH to get the data from GPU
- * to memory
- * clflush (obj) to invalidate the CPU cache
- * Another MI_FLUSH in i915_gem_flush (eliminate this somehow?)
- * 5. Read/written by CPU
- * cache lines are loaded and dirtied
- * 6. Read written by GPU
- * Same as last GPU access
- *
- * Case 3: The constant buffer
- *
- * 1. Allocated
- * 2. Written by CPU
- * 3. Read by GPU
- * 4. Updated (written) by CPU again
- * 5. Read by GPU
- *
- * 1. Allocated
- * (CPU, CPU)
- * 2. Written by CPU
- * (CPU, CPU)
- * 3. Read by GPU
- * (CPU+RENDER, 0)
- * flush_domains = CPU
- * invalidate_domains = RENDER
- * clflush (obj)
- * MI_FLUSH
- * drm_agp_chipset_flush
- * 4. Updated (written) by CPU again
- * (CPU, CPU)
- * flush_domains = 0 (no previous write domain)
- * invalidate_domains = 0 (no new read domains)
- * 5. Read by GPU
- * (CPU+RENDER, 0)
- * flush_domains = CPU
- * invalidate_domains = RENDER
- * clflush (obj)
- * MI_FLUSH
- * drm_agp_chipset_flush
- */
-static void
-i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj,
- struct intel_ring_buffer *ring)
-{
- struct drm_device *dev = obj->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- uint32_t invalidate_domains = 0;
- uint32_t flush_domains = 0;
-
- /*
- * If the object isn't moving to a new write domain,
- * let the object stay in multiple read domains
- */
- if (obj->pending_write_domain == 0)
- obj->pending_read_domains |= obj->read_domains;
-
- /*
- * Flush the current write domain if
- * the new read domains don't match. Invalidate
- * any read domains which differ from the old
- * write domain
- */
- if (obj->write_domain &&
- (obj->write_domain != obj->pending_read_domains ||
- obj_priv->ring != ring)) {
- flush_domains |= obj->write_domain;
- invalidate_domains |=
- obj->pending_read_domains & ~obj->write_domain;
- }
- /*
- * Invalidate any read caches which may have
- * stale data. That is, any new read domains.
- */
- invalidate_domains |= obj->pending_read_domains & ~obj->read_domains;
- if ((flush_domains | invalidate_domains) & I915_GEM_DOMAIN_CPU)
- i915_gem_clflush_object(obj);
-
- /* blow away mappings if mapped through GTT */
- if ((flush_domains | invalidate_domains) & I915_GEM_DOMAIN_GTT)
- i915_gem_release_mmap(obj);
-
- /* The actual obj->write_domain will be updated with
- * pending_write_domain after we emit the accumulated flush for all
- * of our domain changes in execbuffers (which clears objects'
- * write_domains). So if we have a current write domain that we
- * aren't changing, set pending_write_domain to that.
- */
- if (flush_domains == 0 && obj->pending_write_domain == 0)
- obj->pending_write_domain = obj->write_domain;
-
- dev->invalidate_domains |= invalidate_domains;
- dev->flush_domains |= flush_domains;
- if (flush_domains & I915_GEM_GPU_DOMAINS)
- dev_priv->mm.flush_rings |= obj_priv->ring->id;
- if (invalidate_domains & I915_GEM_GPU_DOMAINS)
- dev_priv->mm.flush_rings |= ring->id;
-}
-
/**
* Moves the object from a partially CPU read to a full one.
*
* and doesn't handle transitioning from !(read_domains & I915_GEM_DOMAIN_CPU).
*/
static void
-i915_gem_object_set_to_full_cpu_read_domain(struct drm_gem_object *obj)
+i915_gem_object_set_to_full_cpu_read_domain(struct drm_i915_gem_object *obj)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
-
- if (!obj_priv->page_cpu_valid)
+ if (!obj->page_cpu_valid)
return;
/* If we're partially in the CPU read domain, finish moving it in.
*/
- if (obj->read_domains & I915_GEM_DOMAIN_CPU) {
+ if (obj->base.read_domains & I915_GEM_DOMAIN_CPU) {
int i;
- for (i = 0; i <= (obj->size - 1) / PAGE_SIZE; i++) {
- if (obj_priv->page_cpu_valid[i])
+ for (i = 0; i <= (obj->base.size - 1) / PAGE_SIZE; i++) {
+ if (obj->page_cpu_valid[i])
continue;
- drm_clflush_pages(obj_priv->pages + i, 1);
+ drm_clflush_pages(obj->pages + i, 1);
}
}
/* Free the page_cpu_valid mappings which are now stale, whether
* or not we've got I915_GEM_DOMAIN_CPU.
*/
- kfree(obj_priv->page_cpu_valid);
- obj_priv->page_cpu_valid = NULL;
+ kfree(obj->page_cpu_valid);
+ obj->page_cpu_valid = NULL;
}
/**
* flushes to occur.
*/
static int
-i915_gem_object_set_cpu_read_domain_range(struct drm_gem_object *obj,
+i915_gem_object_set_cpu_read_domain_range(struct drm_i915_gem_object *obj,
uint64_t offset, uint64_t size)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
uint32_t old_read_domains;
int i, ret;
- if (offset == 0 && size == obj->size)
+ if (offset == 0 && size == obj->base.size)
return i915_gem_object_set_to_cpu_domain(obj, 0);
ret = i915_gem_object_flush_gpu_write_domain(obj, false);
i915_gem_object_flush_gtt_write_domain(obj);
/* If we're already fully in the CPU read domain, we're done. */
- if (obj_priv->page_cpu_valid == NULL &&
- (obj->read_domains & I915_GEM_DOMAIN_CPU) != 0)
+ if (obj->page_cpu_valid == NULL &&
+ (obj->base.read_domains & I915_GEM_DOMAIN_CPU) != 0)
return 0;
/* Otherwise, create/clear the per-page CPU read domain flag if we're
* newly adding I915_GEM_DOMAIN_CPU
*/
- if (obj_priv->page_cpu_valid == NULL) {
- obj_priv->page_cpu_valid = kzalloc(obj->size / PAGE_SIZE,
- GFP_KERNEL);
- if (obj_priv->page_cpu_valid == NULL)
+ if (obj->page_cpu_valid == NULL) {
+ obj->page_cpu_valid = kzalloc(obj->base.size / PAGE_SIZE,
+ GFP_KERNEL);
+ if (obj->page_cpu_valid == NULL)
return -ENOMEM;
- } else if ((obj->read_domains & I915_GEM_DOMAIN_CPU) == 0)
- memset(obj_priv->page_cpu_valid, 0, obj->size / PAGE_SIZE);
+ } else if ((obj->base.read_domains & I915_GEM_DOMAIN_CPU) == 0)
+ memset(obj->page_cpu_valid, 0, obj->base.size / PAGE_SIZE);
/* Flush the cache on any pages that are still invalid from the CPU's
* perspective.
*/
for (i = offset / PAGE_SIZE; i <= (offset + size - 1) / PAGE_SIZE;
i++) {
- if (obj_priv->page_cpu_valid[i])
+ if (obj->page_cpu_valid[i])
continue;
- drm_clflush_pages(obj_priv->pages + i, 1);
+ drm_clflush_pages(obj->pages + i, 1);
- obj_priv->page_cpu_valid[i] = 1;
+ obj->page_cpu_valid[i] = 1;
}
/* It should now be out of any other write domains, and we can update
* the domain values for our changes.
*/
- BUG_ON((obj->write_domain & ~I915_GEM_DOMAIN_CPU) != 0);
+ BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_CPU) != 0);
- old_read_domains = obj->read_domains;
- obj->read_domains |= I915_GEM_DOMAIN_CPU;
+ old_read_domains = obj->base.read_domains;
+ obj->base.read_domains |= I915_GEM_DOMAIN_CPU;
trace_i915_gem_object_change_domain(obj,
old_read_domains,
- obj->write_domain);
-
- return 0;
-}
-
-/**
- * Pin an object to the GTT and evaluate the relocations landing in it.
- */
-static int
-i915_gem_execbuffer_relocate(struct drm_i915_gem_object *obj,
- struct drm_file *file_priv,
- struct drm_i915_gem_exec_object2 *entry)
-{
- struct drm_device *dev = obj->base.dev;
- drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_relocation_entry __user *user_relocs;
- struct drm_gem_object *target_obj = NULL;
- uint32_t target_handle = 0;
- int i, ret = 0;
-
- user_relocs = (void __user *)(uintptr_t)entry->relocs_ptr;
- for (i = 0; i < entry->relocation_count; i++) {
- struct drm_i915_gem_relocation_entry reloc;
- uint32_t target_offset;
-
- if (__copy_from_user_inatomic(&reloc,
- user_relocs+i,
- sizeof(reloc))) {
- ret = -EFAULT;
- break;
- }
-
- if (reloc.target_handle != target_handle) {
- drm_gem_object_unreference(target_obj);
-
- target_obj = drm_gem_object_lookup(dev, file_priv,
- reloc.target_handle);
- if (target_obj == NULL) {
- ret = -ENOENT;
- break;
- }
-
- target_handle = reloc.target_handle;
- }
- target_offset = to_intel_bo(target_obj)->gtt_offset;
-
-#if WATCH_RELOC
- DRM_INFO("%s: obj %p offset %08x target %d "
- "read %08x write %08x gtt %08x "
- "presumed %08x delta %08x\n",
- __func__,
- obj,
- (int) reloc.offset,
- (int) reloc.target_handle,
- (int) reloc.read_domains,
- (int) reloc.write_domain,
- (int) target_offset,
- (int) reloc.presumed_offset,
- reloc.delta);
-#endif
-
- /* The target buffer should have appeared before us in the
- * exec_object list, so it should have a GTT space bound by now.
- */
- if (target_offset == 0) {
- DRM_ERROR("No GTT space found for object %d\n",
- reloc.target_handle);
- ret = -EINVAL;
- break;
- }
-
- /* Validate that the target is in a valid r/w GPU domain */
- if (reloc.write_domain & (reloc.write_domain - 1)) {
- DRM_ERROR("reloc with multiple write domains: "
- "obj %p target %d offset %d "
- "read %08x write %08x",
- obj, reloc.target_handle,
- (int) reloc.offset,
- reloc.read_domains,
- reloc.write_domain);
- ret = -EINVAL;
- break;
- }
- if (reloc.write_domain & I915_GEM_DOMAIN_CPU ||
- reloc.read_domains & I915_GEM_DOMAIN_CPU) {
- DRM_ERROR("reloc with read/write CPU domains: "
- "obj %p target %d offset %d "
- "read %08x write %08x",
- obj, reloc.target_handle,
- (int) reloc.offset,
- reloc.read_domains,
- reloc.write_domain);
- ret = -EINVAL;
- break;
- }
- if (reloc.write_domain && target_obj->pending_write_domain &&
- reloc.write_domain != target_obj->pending_write_domain) {
- DRM_ERROR("Write domain conflict: "
- "obj %p target %d offset %d "
- "new %08x old %08x\n",
- obj, reloc.target_handle,
- (int) reloc.offset,
- reloc.write_domain,
- target_obj->pending_write_domain);
- ret = -EINVAL;
- break;
- }
-
- target_obj->pending_read_domains |= reloc.read_domains;
- target_obj->pending_write_domain |= reloc.write_domain;
-
- /* If the relocation already has the right value in it, no
- * more work needs to be done.
- */
- if (target_offset == reloc.presumed_offset)
- continue;
-
- /* Check that the relocation address is valid... */
- if (reloc.offset > obj->base.size - 4) {
- DRM_ERROR("Relocation beyond object bounds: "
- "obj %p target %d offset %d size %d.\n",
- obj, reloc.target_handle,
- (int) reloc.offset, (int) obj->base.size);
- ret = -EINVAL;
- break;
- }
- if (reloc.offset & 3) {
- DRM_ERROR("Relocation not 4-byte aligned: "
- "obj %p target %d offset %d.\n",
- obj, reloc.target_handle,
- (int) reloc.offset);
- ret = -EINVAL;
- break;
- }
-
- /* and points to somewhere within the target object. */
- if (reloc.delta >= target_obj->size) {
- DRM_ERROR("Relocation beyond target object bounds: "
- "obj %p target %d delta %d size %d.\n",
- obj, reloc.target_handle,
- (int) reloc.delta, (int) target_obj->size);
- ret = -EINVAL;
- break;
- }
-
- reloc.delta += target_offset;
- if (obj->base.write_domain == I915_GEM_DOMAIN_CPU) {
- uint32_t page_offset = reloc.offset & ~PAGE_MASK;
- char *vaddr;
-
- vaddr = kmap_atomic(obj->pages[reloc.offset >> PAGE_SHIFT]);
- *(uint32_t *)(vaddr + page_offset) = reloc.delta;
- kunmap_atomic(vaddr);
- } else {
- uint32_t __iomem *reloc_entry;
- void __iomem *reloc_page;
-
- ret = i915_gem_object_set_to_gtt_domain(&obj->base, 1);
- if (ret)
- break;
-
- /* Map the page containing the relocation we're going to perform. */
- reloc.offset += obj->gtt_offset;
- reloc_page = io_mapping_map_atomic_wc(dev_priv->mm.gtt_mapping,
- reloc.offset & PAGE_MASK);
- reloc_entry = (uint32_t __iomem *)
- (reloc_page + (reloc.offset & ~PAGE_MASK));
- iowrite32(reloc.delta, reloc_entry);
- io_mapping_unmap_atomic(reloc_page);
- }
-
- /* and update the user's relocation entry */
- reloc.presumed_offset = target_offset;
- if (__copy_to_user_inatomic(&user_relocs[i].presumed_offset,
- &reloc.presumed_offset,
- sizeof(reloc.presumed_offset))) {
- ret = -EFAULT;
- break;
- }
- }
-
- drm_gem_object_unreference(target_obj);
- return ret;
-}
-
-static int
-i915_gem_execbuffer_pin(struct drm_device *dev,
- struct drm_file *file,
- struct drm_gem_object **object_list,
- struct drm_i915_gem_exec_object2 *exec_list,
- int count)
-{
- struct drm_i915_private *dev_priv = dev->dev_private;
- int ret, i, retry;
-
- /* attempt to pin all of the buffers into the GTT */
- retry = 0;
- do {
- ret = 0;
- for (i = 0; i < count; i++) {
- struct drm_i915_gem_exec_object2 *entry = &exec_list[i];
- struct drm_i915_gem_object *obj = to_intel_bo(object_list[i]);
- bool need_fence =
- entry->flags & EXEC_OBJECT_NEEDS_FENCE &&
- obj->tiling_mode != I915_TILING_NONE;
-
- /* g33/pnv can't fence buffers in the unmappable part */
- bool need_mappable =
- entry->relocation_count ? true : need_fence;
-
- /* Check fence reg constraints and rebind if necessary */
- if ((need_fence && !obj->fenceable) ||
- (need_mappable && !obj->mappable)) {
- ret = i915_gem_object_unbind(&obj->base);
- if (ret)
- break;
- }
-
- ret = i915_gem_object_pin(&obj->base,
- entry->alignment,
- need_mappable,
- need_fence);
- if (ret)
- break;
-
- /*
- * Pre-965 chips need a fence register set up in order
- * to properly handle blits to/from tiled surfaces.
- */
- if (need_fence) {
- ret = i915_gem_object_get_fence_reg(&obj->base, true);
- if (ret) {
- i915_gem_object_unpin(&obj->base);
- break;
- }
-
- dev_priv->fence_regs[obj->fence_reg].gpu = true;
- }
-
- entry->offset = obj->gtt_offset;
- }
-
- while (i--)
- i915_gem_object_unpin(object_list[i]);
-
- if (ret != -ENOSPC || retry > 1)
- return ret;
-
- /* First attempt, just clear anything that is purgeable.
- * Second attempt, clear the entire GTT.
- */
- ret = i915_gem_evict_everything(dev, retry == 0);
- if (ret)
- return ret;
-
- retry++;
- } while (1);
-}
-
-static int
-i915_gem_execbuffer_move_to_gpu(struct drm_device *dev,
- struct drm_file *file,
- struct intel_ring_buffer *ring,
- struct drm_gem_object **objects,
- int count)
-{
- struct drm_i915_private *dev_priv = dev->dev_private;
- int ret, i;
-
- /* Zero the global flush/invalidate flags. These
- * will be modified as new domains are computed
- * for each object
- */
- dev->invalidate_domains = 0;
- dev->flush_domains = 0;
- dev_priv->mm.flush_rings = 0;
- for (i = 0; i < count; i++)
- i915_gem_object_set_to_gpu_domain(objects[i], ring);
-
- if (dev->invalidate_domains | dev->flush_domains) {
-#if WATCH_EXEC
- DRM_INFO("%s: invalidate_domains %08x flush_domains %08x\n",
- __func__,
- dev->invalidate_domains,
- dev->flush_domains);
-#endif
- i915_gem_flush(dev, file,
- dev->invalidate_domains,
- dev->flush_domains,
- dev_priv->mm.flush_rings);
- }
-
- for (i = 0; i < count; i++) {
- struct drm_i915_gem_object *obj = to_intel_bo(objects[i]);
- /* XXX replace with semaphores */
- if (obj->ring && ring != obj->ring) {
- ret = i915_gem_object_wait_rendering(&obj->base, true);
- if (ret)
- return ret;
- }
- }
+ obj->base.write_domain);
return 0;
}
return ret;
}
-static int
-i915_gem_check_execbuffer(struct drm_i915_gem_execbuffer2 *exec,
- uint64_t exec_offset)
-{
- uint32_t exec_start, exec_len;
-
- exec_start = (uint32_t) exec_offset + exec->batch_start_offset;
- exec_len = (uint32_t) exec->batch_len;
-
- if ((exec_start | exec_len) & 0x7)
- return -EINVAL;
-
- if (!exec_start)
- return -EINVAL;
-
- return 0;
-}
-
-static int
-validate_exec_list(struct drm_i915_gem_exec_object2 *exec,
- int count)
-{
- int i;
-
- for (i = 0; i < count; i++) {
- char __user *ptr = (char __user *)(uintptr_t)exec[i].relocs_ptr;
- size_t length = exec[i].relocation_count * sizeof(struct drm_i915_gem_relocation_entry);
-
- if (!access_ok(VERIFY_READ, ptr, length))
- return -EFAULT;
-
- /* we may also need to update the presumed offsets */
- if (!access_ok(VERIFY_WRITE, ptr, length))
- return -EFAULT;
-
- if (fault_in_pages_readable(ptr, length))
- return -EFAULT;
- }
-
- return 0;
-}
-
-static int
-i915_gem_do_execbuffer(struct drm_device *dev, void *data,
- struct drm_file *file,
- struct drm_i915_gem_execbuffer2 *args,
- struct drm_i915_gem_exec_object2 *exec_list)
-{
- drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_gem_object **object_list = NULL;
- struct drm_gem_object *batch_obj;
- struct drm_clip_rect *cliprects = NULL;
- struct drm_i915_gem_request *request = NULL;
- int ret, i, flips;
- uint64_t exec_offset;
-
- struct intel_ring_buffer *ring = NULL;
-
- ret = i915_gem_check_is_wedged(dev);
- if (ret)
- return ret;
-
- ret = validate_exec_list(exec_list, args->buffer_count);
- if (ret)
- return ret;
-
-#if WATCH_EXEC
- DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n",
- (int) args->buffers_ptr, args->buffer_count, args->batch_len);
-#endif
- switch (args->flags & I915_EXEC_RING_MASK) {
- case I915_EXEC_DEFAULT:
- case I915_EXEC_RENDER:
- ring = &dev_priv->render_ring;
- break;
- case I915_EXEC_BSD:
- if (!HAS_BSD(dev)) {
- DRM_ERROR("execbuf with invalid ring (BSD)\n");
- return -EINVAL;
- }
- ring = &dev_priv->bsd_ring;
- break;
- case I915_EXEC_BLT:
- if (!HAS_BLT(dev)) {
- DRM_ERROR("execbuf with invalid ring (BLT)\n");
- return -EINVAL;
- }
- ring = &dev_priv->blt_ring;
- break;
- default:
- DRM_ERROR("execbuf with unknown ring: %d\n",
- (int)(args->flags & I915_EXEC_RING_MASK));
- return -EINVAL;
- }
-
- if (args->buffer_count < 1) {
- DRM_ERROR("execbuf with %d buffers\n", args->buffer_count);
- return -EINVAL;
- }
- object_list = drm_malloc_ab(sizeof(*object_list), args->buffer_count);
- if (object_list == NULL) {
- DRM_ERROR("Failed to allocate object list for %d buffers\n",
- args->buffer_count);
- ret = -ENOMEM;
- goto pre_mutex_err;
- }
-
- if (args->num_cliprects != 0) {
- cliprects = kcalloc(args->num_cliprects, sizeof(*cliprects),
- GFP_KERNEL);
- if (cliprects == NULL) {
- ret = -ENOMEM;
- goto pre_mutex_err;
- }
-
- ret = copy_from_user(cliprects,
- (struct drm_clip_rect __user *)
- (uintptr_t) args->cliprects_ptr,
- sizeof(*cliprects) * args->num_cliprects);
- if (ret != 0) {
- DRM_ERROR("copy %d cliprects failed: %d\n",
- args->num_cliprects, ret);
- ret = -EFAULT;
- goto pre_mutex_err;
- }
- }
-
- request = kzalloc(sizeof(*request), GFP_KERNEL);
- if (request == NULL) {
- ret = -ENOMEM;
- goto pre_mutex_err;
- }
-
- ret = i915_mutex_lock_interruptible(dev);
- if (ret)
- goto pre_mutex_err;
-
- if (dev_priv->mm.suspended) {
- mutex_unlock(&dev->struct_mutex);
- ret = -EBUSY;
- goto pre_mutex_err;
- }
-
- /* Look up object handles */
- for (i = 0; i < args->buffer_count; i++) {
- struct drm_i915_gem_object *obj_priv;
-
- object_list[i] = drm_gem_object_lookup(dev, file,
- exec_list[i].handle);
- if (object_list[i] == NULL) {
- DRM_ERROR("Invalid object handle %d at index %d\n",
- exec_list[i].handle, i);
- /* prevent error path from reading uninitialized data */
- args->buffer_count = i + 1;
- ret = -ENOENT;
- goto err;
- }
-
- obj_priv = to_intel_bo(object_list[i]);
- if (obj_priv->in_execbuffer) {
- DRM_ERROR("Object %p appears more than once in object list\n",
- object_list[i]);
- /* prevent error path from reading uninitialized data */
- args->buffer_count = i + 1;
- ret = -EINVAL;
- goto err;
- }
- obj_priv->in_execbuffer = true;
- }
-
- /* Move the objects en-masse into the GTT, evicting if necessary. */
- ret = i915_gem_execbuffer_pin(dev, file,
- object_list, exec_list,
- args->buffer_count);
- if (ret)
- goto err;
-
- /* The objects are in their final locations, apply the relocations. */
- for (i = 0; i < args->buffer_count; i++) {
- struct drm_i915_gem_object *obj = to_intel_bo(object_list[i]);
- obj->base.pending_read_domains = 0;
- obj->base.pending_write_domain = 0;
- ret = i915_gem_execbuffer_relocate(obj, file, &exec_list[i]);
- if (ret)
- goto err;
- }
-
- /* Set the pending read domains for the batch buffer to COMMAND */
- batch_obj = object_list[args->buffer_count-1];
- if (batch_obj->pending_write_domain) {
- DRM_ERROR("Attempting to use self-modifying batch buffer\n");
- ret = -EINVAL;
- goto err;
- }
- batch_obj->pending_read_domains |= I915_GEM_DOMAIN_COMMAND;
-
- /* Sanity check the batch buffer */
- exec_offset = to_intel_bo(batch_obj)->gtt_offset;
- ret = i915_gem_check_execbuffer(args, exec_offset);
- if (ret != 0) {
- DRM_ERROR("execbuf with invalid offset/length\n");
- goto err;
- }
-
- ret = i915_gem_execbuffer_move_to_gpu(dev, file, ring,
- object_list, args->buffer_count);
- if (ret)
- goto err;
-
-#if WATCH_COHERENCY
- for (i = 0; i < args->buffer_count; i++) {
- i915_gem_object_check_coherency(object_list[i],
- exec_list[i].handle);
- }
-#endif
-
-#if WATCH_EXEC
- i915_gem_dump_object(batch_obj,
- args->batch_len,
- __func__,
- ~0);
-#endif
-
- /* Check for any pending flips. As we only maintain a flip queue depth
- * of 1, we can simply insert a WAIT for the next display flip prior
- * to executing the batch and avoid stalling the CPU.
- */
- flips = 0;
- for (i = 0; i < args->buffer_count; i++) {
- if (object_list[i]->write_domain)
- flips |= atomic_read(&to_intel_bo(object_list[i])->pending_flip);
- }
- if (flips) {
- int plane, flip_mask;
-
- for (plane = 0; flips >> plane; plane++) {
- if (((flips >> plane) & 1) == 0)
- continue;
-
- if (plane)
- flip_mask = MI_WAIT_FOR_PLANE_B_FLIP;
- else
- flip_mask = MI_WAIT_FOR_PLANE_A_FLIP;
-
- ret = intel_ring_begin(ring, 2);
- if (ret)
- goto err;
-
- intel_ring_emit(ring, MI_WAIT_FOR_EVENT | flip_mask);
- intel_ring_emit(ring, MI_NOOP);
- intel_ring_advance(ring);
- }
- }
-
- /* Exec the batchbuffer */
- ret = ring->dispatch_execbuffer(ring, args, cliprects, exec_offset);
- if (ret) {
- DRM_ERROR("dispatch failed %d\n", ret);
- goto err;
- }
-
- for (i = 0; i < args->buffer_count; i++) {
- struct drm_gem_object *obj = object_list[i];
-
- obj->read_domains = obj->pending_read_domains;
- obj->write_domain = obj->pending_write_domain;
-
- i915_gem_object_move_to_active(obj, ring);
- if (obj->write_domain) {
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- obj_priv->dirty = 1;
- list_move_tail(&obj_priv->gpu_write_list,
- &ring->gpu_write_list);
- intel_mark_busy(dev, obj);
- }
-
- trace_i915_gem_object_change_domain(obj,
- obj->read_domains,
- obj->write_domain);
- }
-
- /*
- * Ensure that the commands in the batch buffer are
- * finished before the interrupt fires
- */
- i915_retire_commands(dev, ring);
-
- if (i915_add_request(dev, file, request, ring))
- ring->outstanding_lazy_request = true;
- else
- request = NULL;
-
-err:
- for (i = 0; i < args->buffer_count; i++) {
- if (object_list[i] == NULL)
- break;
-
- to_intel_bo(object_list[i])->in_execbuffer = false;
- drm_gem_object_unreference(object_list[i]);
- }
-
- mutex_unlock(&dev->struct_mutex);
-
-pre_mutex_err:
- drm_free_large(object_list);
- kfree(cliprects);
- kfree(request);
-
- return ret;
-}
-
-/*
- * Legacy execbuffer just creates an exec2 list from the original exec object
- * list array and passes it to the real function.
- */
-int
-i915_gem_execbuffer(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
-{
- struct drm_i915_gem_execbuffer *args = data;
- struct drm_i915_gem_execbuffer2 exec2;
- struct drm_i915_gem_exec_object *exec_list = NULL;
- struct drm_i915_gem_exec_object2 *exec2_list = NULL;
- int ret, i;
-
-#if WATCH_EXEC
- DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n",
- (int) args->buffers_ptr, args->buffer_count, args->batch_len);
-#endif
-
- if (args->buffer_count < 1) {
- DRM_ERROR("execbuf with %d buffers\n", args->buffer_count);
- return -EINVAL;
- }
-
- /* Copy in the exec list from userland */
- exec_list = drm_malloc_ab(sizeof(*exec_list), args->buffer_count);
- exec2_list = drm_malloc_ab(sizeof(*exec2_list), args->buffer_count);
- if (exec_list == NULL || exec2_list == NULL) {
- DRM_ERROR("Failed to allocate exec list for %d buffers\n",
- args->buffer_count);
- drm_free_large(exec_list);
- drm_free_large(exec2_list);
- return -ENOMEM;
- }
- ret = copy_from_user(exec_list,
- (struct drm_i915_relocation_entry __user *)
- (uintptr_t) args->buffers_ptr,
- sizeof(*exec_list) * args->buffer_count);
- if (ret != 0) {
- DRM_ERROR("copy %d exec entries failed %d\n",
- args->buffer_count, ret);
- drm_free_large(exec_list);
- drm_free_large(exec2_list);
- return -EFAULT;
- }
-
- for (i = 0; i < args->buffer_count; i++) {
- exec2_list[i].handle = exec_list[i].handle;
- exec2_list[i].relocation_count = exec_list[i].relocation_count;
- exec2_list[i].relocs_ptr = exec_list[i].relocs_ptr;
- exec2_list[i].alignment = exec_list[i].alignment;
- exec2_list[i].offset = exec_list[i].offset;
- if (INTEL_INFO(dev)->gen < 4)
- exec2_list[i].flags = EXEC_OBJECT_NEEDS_FENCE;
- else
- exec2_list[i].flags = 0;
- }
-
- exec2.buffers_ptr = args->buffers_ptr;
- exec2.buffer_count = args->buffer_count;
- exec2.batch_start_offset = args->batch_start_offset;
- exec2.batch_len = args->batch_len;
- exec2.DR1 = args->DR1;
- exec2.DR4 = args->DR4;
- exec2.num_cliprects = args->num_cliprects;
- exec2.cliprects_ptr = args->cliprects_ptr;
- exec2.flags = I915_EXEC_RENDER;
-
- ret = i915_gem_do_execbuffer(dev, data, file_priv, &exec2, exec2_list);
- if (!ret) {
- /* Copy the new buffer offsets back to the user's exec list. */
- for (i = 0; i < args->buffer_count; i++)
- exec_list[i].offset = exec2_list[i].offset;
- /* ... and back out to userspace */
- ret = copy_to_user((struct drm_i915_relocation_entry __user *)
- (uintptr_t) args->buffers_ptr,
- exec_list,
- sizeof(*exec_list) * args->buffer_count);
- if (ret) {
- ret = -EFAULT;
- DRM_ERROR("failed to copy %d exec entries "
- "back to user (%d)\n",
- args->buffer_count, ret);
- }
- }
-
- drm_free_large(exec_list);
- drm_free_large(exec2_list);
- return ret;
-}
-
-int
-i915_gem_execbuffer2(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
-{
- struct drm_i915_gem_execbuffer2 *args = data;
- struct drm_i915_gem_exec_object2 *exec2_list = NULL;
- int ret;
-
-#if WATCH_EXEC
- DRM_INFO("buffers_ptr %d buffer_count %d len %08x\n",
- (int) args->buffers_ptr, args->buffer_count, args->batch_len);
-#endif
-
- if (args->buffer_count < 1) {
- DRM_ERROR("execbuf2 with %d buffers\n", args->buffer_count);
- return -EINVAL;
- }
-
- exec2_list = drm_malloc_ab(sizeof(*exec2_list), args->buffer_count);
- if (exec2_list == NULL) {
- DRM_ERROR("Failed to allocate exec list for %d buffers\n",
- args->buffer_count);
- return -ENOMEM;
- }
- ret = copy_from_user(exec2_list,
- (struct drm_i915_relocation_entry __user *)
- (uintptr_t) args->buffers_ptr,
- sizeof(*exec2_list) * args->buffer_count);
- if (ret != 0) {
- DRM_ERROR("copy %d exec entries failed %d\n",
- args->buffer_count, ret);
- drm_free_large(exec2_list);
- return -EFAULT;
- }
-
- ret = i915_gem_do_execbuffer(dev, data, file_priv, args, exec2_list);
- if (!ret) {
- /* Copy the new buffer offsets back to the user's exec list. */
- ret = copy_to_user((struct drm_i915_relocation_entry __user *)
- (uintptr_t) args->buffers_ptr,
- exec2_list,
- sizeof(*exec2_list) * args->buffer_count);
- if (ret) {
- ret = -EFAULT;
- DRM_ERROR("failed to copy %d exec entries "
- "back to user (%d)\n",
- args->buffer_count, ret);
- }
- }
-
- drm_free_large(exec2_list);
- return ret;
-}
-
int
-i915_gem_object_pin(struct drm_gem_object *obj, uint32_t alignment,
- bool mappable, bool need_fence)
+i915_gem_object_pin(struct drm_i915_gem_object *obj,
+ uint32_t alignment,
+ bool map_and_fenceable)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
struct drm_i915_private *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
int ret;
- BUG_ON(obj_priv->pin_count == DRM_I915_GEM_OBJECT_MAX_PIN_COUNT);
+ BUG_ON(obj->pin_count == DRM_I915_GEM_OBJECT_MAX_PIN_COUNT);
WARN_ON(i915_verify_lists(dev));
- if (obj_priv->gtt_space != NULL) {
- if ((alignment && obj_priv->gtt_offset & (alignment - 1)) ||
- (need_fence && !obj_priv->fenceable) ||
- (mappable && !obj_priv->mappable)) {
- WARN(obj_priv->pin_count,
+ if (obj->gtt_space != NULL) {
+ if ((alignment && obj->gtt_offset & (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, need_fence=%d, fenceable=%d, mappable=%d, cpu_accessible=%d\n",
- obj_priv->gtt_offset, alignment,
- need_fence, obj_priv->fenceable,
- mappable, obj_priv->mappable);
+ " offset=%x, req.alignment=%x, req.map_and_fenceable=%d,"
+ " obj->map_and_fenceable=%d\n",
+ obj->gtt_offset, alignment,
+ map_and_fenceable,
+ obj->map_and_fenceable);
ret = i915_gem_object_unbind(obj);
if (ret)
return ret;
}
}
- if (obj_priv->gtt_space == NULL) {
+ if (obj->gtt_space == NULL) {
ret = i915_gem_object_bind_to_gtt(obj, alignment,
- mappable, need_fence);
+ map_and_fenceable);
if (ret)
return ret;
}
- if (obj_priv->pin_count++ == 0) {
- i915_gem_info_add_pin(dev_priv, obj_priv, mappable);
- if (!obj_priv->active)
- list_move_tail(&obj_priv->mm_list,
+ if (obj->pin_count++ == 0) {
+ if (!obj->active)
+ list_move_tail(&obj->mm_list,
&dev_priv->mm.pinned_list);
}
- BUG_ON(!obj_priv->pin_mappable && mappable);
+ obj->pin_mappable |= map_and_fenceable;
WARN_ON(i915_verify_lists(dev));
return 0;
}
void
-i915_gem_object_unpin(struct drm_gem_object *obj)
+i915_gem_object_unpin(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
WARN_ON(i915_verify_lists(dev));
- BUG_ON(obj_priv->pin_count == 0);
- BUG_ON(obj_priv->gtt_space == NULL);
+ BUG_ON(obj->pin_count == 0);
+ BUG_ON(obj->gtt_space == NULL);
- if (--obj_priv->pin_count == 0) {
- if (!obj_priv->active)
- list_move_tail(&obj_priv->mm_list,
+ if (--obj->pin_count == 0) {
+ if (!obj->active)
+ list_move_tail(&obj->mm_list,
&dev_priv->mm.inactive_list);
- i915_gem_info_remove_pin(dev_priv, obj_priv);
+ obj->pin_mappable = false;
}
WARN_ON(i915_verify_lists(dev));
}
int
i915_gem_pin_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_gem_pin *args = data;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
+ struct drm_i915_gem_object *obj;
int ret;
ret = i915_mutex_lock_interruptible(dev);
if (ret)
return ret;
- obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
if (obj == NULL) {
ret = -ENOENT;
goto unlock;
}
- obj_priv = to_intel_bo(obj);
- if (obj_priv->madv != I915_MADV_WILLNEED) {
+ if (obj->madv != I915_MADV_WILLNEED) {
DRM_ERROR("Attempting to pin a purgeable buffer\n");
ret = -EINVAL;
goto out;
}
- if (obj_priv->pin_filp != NULL && obj_priv->pin_filp != file_priv) {
+ if (obj->pin_filp != NULL && obj->pin_filp != file) {
DRM_ERROR("Already pinned in i915_gem_pin_ioctl(): %d\n",
args->handle);
ret = -EINVAL;
goto out;
}
- obj_priv->user_pin_count++;
- obj_priv->pin_filp = file_priv;
- if (obj_priv->user_pin_count == 1) {
- ret = i915_gem_object_pin(obj, args->alignment,
- true, obj_priv->tiling_mode);
+ obj->user_pin_count++;
+ obj->pin_filp = file;
+ if (obj->user_pin_count == 1) {
+ ret = i915_gem_object_pin(obj, args->alignment, true);
if (ret)
goto out;
}
* as the X server doesn't manage domains yet
*/
i915_gem_object_flush_cpu_write_domain(obj);
- args->offset = obj_priv->gtt_offset;
+ args->offset = obj->gtt_offset;
out:
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
return ret;
int
i915_gem_unpin_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_gem_pin *args = data;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
+ struct drm_i915_gem_object *obj;
int ret;
ret = i915_mutex_lock_interruptible(dev);
if (ret)
return ret;
- obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
if (obj == NULL) {
ret = -ENOENT;
goto unlock;
}
- obj_priv = to_intel_bo(obj);
- if (obj_priv->pin_filp != file_priv) {
+ if (obj->pin_filp != file) {
DRM_ERROR("Not pinned by caller in i915_gem_pin_ioctl(): %d\n",
args->handle);
ret = -EINVAL;
goto out;
}
- obj_priv->user_pin_count--;
- if (obj_priv->user_pin_count == 0) {
- obj_priv->pin_filp = NULL;
+ obj->user_pin_count--;
+ if (obj->user_pin_count == 0) {
+ obj->pin_filp = NULL;
i915_gem_object_unpin(obj);
}
out:
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
return ret;
int
i915_gem_busy_ioctl(struct drm_device *dev, void *data,
- struct drm_file *file_priv)
+ struct drm_file *file)
{
struct drm_i915_gem_busy *args = data;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
+ struct drm_i915_gem_object *obj;
int ret;
ret = i915_mutex_lock_interruptible(dev);
if (ret)
return ret;
- obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
if (obj == NULL) {
ret = -ENOENT;
goto unlock;
}
- obj_priv = to_intel_bo(obj);
/* Count all active objects as busy, even if they are currently not used
* by the gpu. Users of this interface expect objects to eventually
* become non-busy without any further actions, therefore emit any
* necessary flushes here.
*/
- args->busy = obj_priv->active;
+ args->busy = obj->active;
if (args->busy) {
/* Unconditionally flush objects, even when the gpu still uses this
* object. Userspace calling this function indicates that it wants to
* use this buffer rather sooner than later, so issuing the required
* flush earlier is beneficial.
*/
- if (obj->write_domain & I915_GEM_GPU_DOMAINS)
- i915_gem_flush_ring(dev, file_priv,
- obj_priv->ring,
- 0, obj->write_domain);
+ if (obj->base.write_domain & I915_GEM_GPU_DOMAINS)
+ i915_gem_flush_ring(dev, obj->ring,
+ 0, obj->base.write_domain);
/* Update the active list for the hardware's current position.
* Otherwise this only updates on a delayed timer or when irqs
* are actually unmasked, and our working set ends up being
* larger than required.
*/
- i915_gem_retire_requests_ring(dev, obj_priv->ring);
+ i915_gem_retire_requests_ring(dev, obj->ring);
- args->busy = obj_priv->active;
+ args->busy = obj->active;
}
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
return ret;
struct drm_file *file_priv)
{
struct drm_i915_gem_madvise *args = data;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
+ struct drm_i915_gem_object *obj;
int ret;
switch (args->madv) {
if (ret)
return ret;
- obj = drm_gem_object_lookup(dev, file_priv, args->handle);
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file_priv, args->handle));
if (obj == NULL) {
ret = -ENOENT;
goto unlock;
}
- obj_priv = to_intel_bo(obj);
- if (obj_priv->pin_count) {
+ if (obj->pin_count) {
ret = -EINVAL;
goto out;
}
- if (obj_priv->madv != __I915_MADV_PURGED)
- obj_priv->madv = args->madv;
+ if (obj->madv != __I915_MADV_PURGED)
+ obj->madv = args->madv;
/* if the object is no longer bound, discard its backing storage */
- if (i915_gem_object_is_purgeable(obj_priv) &&
- obj_priv->gtt_space == NULL)
+ if (i915_gem_object_is_purgeable(obj) &&
+ obj->gtt_space == NULL)
i915_gem_object_truncate(obj);
- args->retained = obj_priv->madv != __I915_MADV_PURGED;
+ args->retained = obj->madv != __I915_MADV_PURGED;
out:
- drm_gem_object_unreference(obj);
+ drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
return ret;
}
-struct drm_gem_object * i915_gem_alloc_object(struct drm_device *dev,
- size_t size)
+struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev,
+ size_t size)
{
struct drm_i915_private *dev_priv = dev->dev_private;
struct drm_i915_gem_object *obj;
obj->base.driver_private = NULL;
obj->fence_reg = I915_FENCE_REG_NONE;
INIT_LIST_HEAD(&obj->mm_list);
+ INIT_LIST_HEAD(&obj->gtt_list);
INIT_LIST_HEAD(&obj->ring_list);
INIT_LIST_HEAD(&obj->gpu_write_list);
obj->madv = I915_MADV_WILLNEED;
- obj->fenceable = true;
- obj->mappable = true;
+ /* Avoid an unnecessary call to unbind on the first bind. */
+ obj->map_and_fenceable = true;
- return &obj->base;
+ return obj;
}
int i915_gem_init_object(struct drm_gem_object *obj)
return 0;
}
-static void i915_gem_free_object_tail(struct drm_gem_object *obj)
+static void i915_gem_free_object_tail(struct drm_i915_gem_object *obj)
{
- struct drm_device *dev = obj->dev;
+ struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
int ret;
ret = i915_gem_object_unbind(obj);
if (ret == -ERESTARTSYS) {
- list_move(&obj_priv->mm_list,
+ list_move(&obj->mm_list,
&dev_priv->mm.deferred_free_list);
return;
}
- if (obj->map_list.map)
+ if (obj->base.map_list.map)
i915_gem_free_mmap_offset(obj);
- drm_gem_object_release(obj);
- i915_gem_info_remove_obj(dev_priv, obj->size);
+ drm_gem_object_release(&obj->base);
+ i915_gem_info_remove_obj(dev_priv, obj->base.size);
- kfree(obj_priv->page_cpu_valid);
- kfree(obj_priv->bit_17);
- kfree(obj_priv);
+ kfree(obj->page_cpu_valid);
+ kfree(obj->bit_17);
+ kfree(obj);
}
-void i915_gem_free_object(struct drm_gem_object *obj)
+void i915_gem_free_object(struct drm_gem_object *gem_obj)
{
- struct drm_device *dev = obj->dev;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+ struct drm_i915_gem_object *obj = to_intel_bo(gem_obj);
+ struct drm_device *dev = obj->base.dev;
trace_i915_gem_object_destroy(obj);
- while (obj_priv->pin_count > 0)
+ while (obj->pin_count > 0)
i915_gem_object_unpin(obj);
- if (obj_priv->phys_obj)
+ if (obj->phys_obj)
i915_gem_detach_phys_object(dev, obj);
i915_gem_free_object_tail(obj);
}
}
+ i915_gem_reset_fences(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!
return 0;
}
-/*
- * 965+ support PIPE_CONTROL commands, which provide finer grained control
- * over cache flushing.
- */
-static int
-i915_gem_init_pipe_control(struct drm_device *dev)
-{
- drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
- int ret;
-
- obj = i915_gem_alloc_object(dev, 4096);
- if (obj == NULL) {
- DRM_ERROR("Failed to allocate seqno page\n");
- ret = -ENOMEM;
- goto err;
- }
- obj_priv = to_intel_bo(obj);
- obj_priv->agp_type = AGP_USER_CACHED_MEMORY;
-
- ret = i915_gem_object_pin(obj, 4096, true, false);
- if (ret)
- goto err_unref;
-
- dev_priv->seqno_gfx_addr = obj_priv->gtt_offset;
- dev_priv->seqno_page = kmap(obj_priv->pages[0]);
- if (dev_priv->seqno_page == NULL)
- goto err_unpin;
-
- dev_priv->seqno_obj = obj;
- memset(dev_priv->seqno_page, 0, PAGE_SIZE);
-
- return 0;
-
-err_unpin:
- i915_gem_object_unpin(obj);
-err_unref:
- drm_gem_object_unreference(obj);
-err:
- return ret;
-}
-
-
-static void
-i915_gem_cleanup_pipe_control(struct drm_device *dev)
-{
- drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_gem_object *obj;
- struct drm_i915_gem_object *obj_priv;
-
- obj = dev_priv->seqno_obj;
- obj_priv = to_intel_bo(obj);
- kunmap(obj_priv->pages[0]);
- i915_gem_object_unpin(obj);
- drm_gem_object_unreference(obj);
- dev_priv->seqno_obj = NULL;
-
- dev_priv->seqno_page = NULL;
-}
-
int
i915_gem_init_ringbuffer(struct drm_device *dev)
{
drm_i915_private_t *dev_priv = dev->dev_private;
int ret;
- if (HAS_PIPE_CONTROL(dev)) {
- ret = i915_gem_init_pipe_control(dev);
- if (ret)
- return ret;
- }
-
ret = intel_init_render_ring_buffer(dev);
if (ret)
- goto cleanup_pipe_control;
+ return ret;
if (HAS_BSD(dev)) {
ret = intel_init_bsd_ring_buffer(dev);
intel_cleanup_ring_buffer(&dev_priv->bsd_ring);
cleanup_render_ring:
intel_cleanup_ring_buffer(&dev_priv->render_ring);
-cleanup_pipe_control:
- if (HAS_PIPE_CONTROL(dev))
- i915_gem_cleanup_pipe_control(dev);
return ret;
}
intel_cleanup_ring_buffer(&dev_priv->render_ring);
intel_cleanup_ring_buffer(&dev_priv->bsd_ring);
intel_cleanup_ring_buffer(&dev_priv->blt_ring);
- if (HAS_PIPE_CONTROL(dev))
- i915_gem_cleanup_pipe_control(dev);
}
int
INIT_LIST_HEAD(&dev_priv->mm.pinned_list);
INIT_LIST_HEAD(&dev_priv->mm.fence_list);
INIT_LIST_HEAD(&dev_priv->mm.deferred_free_list);
+ INIT_LIST_HEAD(&dev_priv->mm.gtt_list);
init_ring_lists(&dev_priv->render_ring);
init_ring_lists(&dev_priv->bsd_ring);
init_ring_lists(&dev_priv->blt_ring);
}
void i915_gem_detach_phys_object(struct drm_device *dev,
- struct drm_gem_object *obj)
+ struct drm_i915_gem_object *obj)
{
- struct address_space *mapping = obj->filp->f_path.dentry->d_inode->i_mapping;
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
+ struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
char *vaddr;
int i;
int page_count;
- if (!obj_priv->phys_obj)
+ if (!obj->phys_obj)
return;
- vaddr = obj_priv->phys_obj->handle->vaddr;
-
- page_count = obj->size / PAGE_SIZE;
+ vaddr = obj->phys_obj->handle->vaddr;
+ page_count = obj->base.size / PAGE_SIZE;
for (i = 0; i < page_count; i++) {
struct page *page = read_cache_page_gfp(mapping, i,
GFP_HIGHUSER | __GFP_RECLAIMABLE);
page_cache_release(page);
}
}
- drm_agp_chipset_flush(dev);
+ intel_gtt_chipset_flush();
- obj_priv->phys_obj->cur_obj = NULL;
- obj_priv->phys_obj = NULL;
+ obj->phys_obj->cur_obj = NULL;
+ obj->phys_obj = NULL;
}
int
i915_gem_attach_phys_object(struct drm_device *dev,
- struct drm_gem_object *obj,
+ struct drm_i915_gem_object *obj,
int id,
int align)
{
- struct address_space *mapping = obj->filp->f_path.dentry->d_inode->i_mapping;
+ struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
drm_i915_private_t *dev_priv = dev->dev_private;
- struct drm_i915_gem_object *obj_priv;
int ret = 0;
int page_count;
int i;
if (id > I915_MAX_PHYS_OBJECT)
return -EINVAL;
- obj_priv = to_intel_bo(obj);
-
- if (obj_priv->phys_obj) {
- if (obj_priv->phys_obj->id == id)
+ if (obj->phys_obj) {
+ if (obj->phys_obj->id == id)
return 0;
i915_gem_detach_phys_object(dev, obj);
}
/* create a new object */
if (!dev_priv->mm.phys_objs[id - 1]) {
ret = i915_gem_init_phys_object(dev, id,
- obj->size, align);
+ obj->base.size, align);
if (ret) {
- DRM_ERROR("failed to init phys object %d size: %zu\n", id, obj->size);
+ DRM_ERROR("failed to init phys object %d size: %zu\n",
+ id, obj->base.size);
return ret;
}
}
/* bind to the object */
- obj_priv->phys_obj = dev_priv->mm.phys_objs[id - 1];
- obj_priv->phys_obj->cur_obj = obj;
+ obj->phys_obj = dev_priv->mm.phys_objs[id - 1];
+ obj->phys_obj->cur_obj = obj;
- page_count = obj->size / PAGE_SIZE;
+ page_count = obj->base.size / PAGE_SIZE;
for (i = 0; i < page_count; i++) {
struct page *page;
return PTR_ERR(page);
src = kmap_atomic(page);
- dst = obj_priv->phys_obj->handle->vaddr + (i * PAGE_SIZE);
+ dst = obj->phys_obj->handle->vaddr + (i * PAGE_SIZE);
memcpy(dst, src, PAGE_SIZE);
kunmap_atomic(src);
}
static int
-i915_gem_phys_pwrite(struct drm_device *dev, struct drm_gem_object *obj,
+i915_gem_phys_pwrite(struct drm_device *dev,
+ struct drm_i915_gem_object *obj,
struct drm_i915_gem_pwrite *args,
struct drm_file *file_priv)
{
- struct drm_i915_gem_object *obj_priv = to_intel_bo(obj);
- void *obj_addr;
- int ret;
- char __user *user_data;
+ void *vaddr = obj->phys_obj->handle->vaddr + args->offset;
+ char __user *user_data = (char __user *) (uintptr_t) args->data_ptr;
- user_data = (char __user *) (uintptr_t) args->data_ptr;
- obj_addr = obj_priv->phys_obj->handle->vaddr + args->offset;
+ if (__copy_from_user_inatomic_nocache(vaddr, user_data, args->size)) {
+ unsigned long unwritten;
- DRM_DEBUG_DRIVER("obj_addr %p, %lld\n", obj_addr, args->size);
- ret = copy_from_user(obj_addr, user_data, args->size);
- if (ret)
- return -EFAULT;
+ /* The physical object once assigned is fixed for the lifetime
+ * of the obj, so we can safely drop the lock and continue
+ * to access vaddr.
+ */
+ mutex_unlock(&dev->struct_mutex);
+ unwritten = copy_from_user(vaddr, user_data, args->size);
+ mutex_lock(&dev->struct_mutex);
+ if (unwritten)
+ return -EFAULT;
+ }
- drm_agp_chipset_flush(dev);
+ intel_gtt_chipset_flush();
return 0;
}
&dev_priv->mm.inactive_list,
mm_list) {
if (i915_gem_object_is_purgeable(obj)) {
- i915_gem_object_unbind(&obj->base);
- if (--nr_to_scan == 0)
+ if (i915_gem_object_unbind(obj) == 0 &&
+ --nr_to_scan == 0)
break;
}
}
list_for_each_entry_safe(obj, next,
&dev_priv->mm.inactive_list,
mm_list) {
- if (nr_to_scan) {
- i915_gem_object_unbind(&obj->base);
+ if (nr_to_scan &&
+ i915_gem_object_unbind(obj) == 0)
nr_to_scan--;
- } else
+ else
cnt++;
}