#include <linux/pci.h>
#include <linux/dma-buf.h>
-static __must_check int i915_gem_object_flush_gpu_write_domain(struct drm_i915_gem_object *obj);
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 __must_check int i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj,
unsigned alignment,
- bool map_and_fenceable);
+ bool map_and_fenceable,
+ bool nonblocking);
static int i915_gem_phys_pwrite(struct drm_device *dev,
struct drm_i915_gem_object *obj,
struct drm_i915_gem_pwrite *args,
static int i915_gem_inactive_shrink(struct shrinker *shrinker,
struct shrink_control *sc);
+static long i915_gem_purge(struct drm_i915_private *dev_priv, long target);
+static void i915_gem_shrink_all(struct drm_i915_private *dev_priv);
static void i915_gem_object_truncate(struct drm_i915_gem_object *obj);
static inline void i915_gem_object_fence_lost(struct drm_i915_gem_object *obj)
static inline bool
i915_gem_object_is_inactive(struct drm_i915_gem_object *obj)
{
- return !obj->active;
+ return obj->gtt_space && !obj->active;
}
int
pinned = 0;
mutex_lock(&dev->struct_mutex);
- list_for_each_entry(obj, &dev_priv->mm.gtt_list, gtt_list)
+ list_for_each_entry(obj, &dev_priv->mm.bound_list, gtt_list)
if (obj->pin_count)
pinned += obj->gtt_space->size;
mutex_unlock(&dev->struct_mutex);
page_length);
kunmap_atomic(vaddr);
- return ret;
+ return ret ? -EFAULT : 0;
}
static void
page_length);
kunmap(page);
- return ret;
+ return ret ? - EFAULT : 0;
}
static int
struct drm_i915_gem_pread *args,
struct drm_file *file)
{
- struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
char __user *user_data;
ssize_t remain;
loff_t offset;
int hit_slowpath = 0;
int prefaulted = 0;
int needs_clflush = 0;
- int release_page;
+ struct scatterlist *sg;
+ int i;
user_data = (char __user *) (uintptr_t) args->data_ptr;
remain = args->size;
* anyway again before the next pread happens. */
if (obj->cache_level == I915_CACHE_NONE)
needs_clflush = 1;
- ret = i915_gem_object_set_to_gtt_domain(obj, false);
- if (ret)
- return ret;
+ if (obj->gtt_space) {
+ ret = i915_gem_object_set_to_gtt_domain(obj, false);
+ if (ret)
+ return ret;
+ }
}
+ ret = i915_gem_object_get_pages(obj);
+ if (ret)
+ return ret;
+
+ i915_gem_object_pin_pages(obj);
+
offset = args->offset;
- while (remain > 0) {
+ for_each_sg(obj->pages->sgl, sg, obj->pages->nents, i) {
struct page *page;
+ if (i < offset >> PAGE_SHIFT)
+ continue;
+
+ if (remain <= 0)
+ break;
+
/* Operation in this page
*
* shmem_page_offset = offset within page in shmem file
if ((shmem_page_offset + page_length) > PAGE_SIZE)
page_length = PAGE_SIZE - shmem_page_offset;
- if (obj->pages) {
- page = obj->pages[offset >> PAGE_SHIFT];
- release_page = 0;
- } else {
- page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT);
- if (IS_ERR(page)) {
- ret = PTR_ERR(page);
- goto out;
- }
- release_page = 1;
- }
-
+ page = sg_page(sg);
page_do_bit17_swizzling = obj_do_bit17_swizzling &&
(page_to_phys(page) & (1 << 17)) != 0;
goto next_page;
hit_slowpath = 1;
- page_cache_get(page);
mutex_unlock(&dev->struct_mutex);
if (!prefaulted) {
needs_clflush);
mutex_lock(&dev->struct_mutex);
- page_cache_release(page);
+
next_page:
mark_page_accessed(page);
- if (release_page)
- page_cache_release(page);
- if (ret) {
- ret = -EFAULT;
+ if (ret)
goto out;
- }
remain -= page_length;
user_data += page_length;
}
out:
+ i915_gem_object_unpin_pages(obj);
+
if (hit_slowpath) {
/* Fixup: Kill any reinstated backing storage pages */
if (obj->madv == __I915_MADV_PURGED)
char __user *user_data;
int page_offset, page_length, ret;
- ret = i915_gem_object_pin(obj, 0, true);
+ ret = i915_gem_object_pin(obj, 0, true, true);
if (ret)
goto out;
page_length);
kunmap_atomic(vaddr);
- return ret;
+ return ret ? -EFAULT : 0;
}
/* Only difference to the fast-path function is that this can handle bit17
page_do_bit17_swizzling);
kunmap(page);
- return ret;
+ return ret ? -EFAULT : 0;
}
static int
struct drm_i915_gem_pwrite *args,
struct drm_file *file)
{
- struct address_space *mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
ssize_t remain;
loff_t offset;
char __user *user_data;
int hit_slowpath = 0;
int needs_clflush_after = 0;
int needs_clflush_before = 0;
- int release_page;
+ int i;
+ struct scatterlist *sg;
user_data = (char __user *) (uintptr_t) args->data_ptr;
remain = args->size;
* right away and we therefore have to clflush anyway. */
if (obj->cache_level == I915_CACHE_NONE)
needs_clflush_after = 1;
- ret = i915_gem_object_set_to_gtt_domain(obj, true);
- if (ret)
- return ret;
+ if (obj->gtt_space) {
+ ret = i915_gem_object_set_to_gtt_domain(obj, true);
+ if (ret)
+ return ret;
+ }
}
/* Same trick applies for invalidate partially written cachelines before
* writing. */
&& obj->cache_level == I915_CACHE_NONE)
needs_clflush_before = 1;
+ ret = i915_gem_object_get_pages(obj);
+ if (ret)
+ return ret;
+
+ i915_gem_object_pin_pages(obj);
+
offset = args->offset;
obj->dirty = 1;
- while (remain > 0) {
+ for_each_sg(obj->pages->sgl, sg, obj->pages->nents, i) {
struct page *page;
int partial_cacheline_write;
+ if (i < offset >> PAGE_SHIFT)
+ continue;
+
+ if (remain <= 0)
+ break;
+
/* Operation in this page
*
* shmem_page_offset = offset within page in shmem file
((shmem_page_offset | page_length)
& (boot_cpu_data.x86_clflush_size - 1));
- if (obj->pages) {
- page = obj->pages[offset >> PAGE_SHIFT];
- release_page = 0;
- } else {
- page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT);
- if (IS_ERR(page)) {
- ret = PTR_ERR(page);
- goto out;
- }
- release_page = 1;
- }
-
+ page = sg_page(sg);
page_do_bit17_swizzling = obj_do_bit17_swizzling &&
(page_to_phys(page) & (1 << 17)) != 0;
goto next_page;
hit_slowpath = 1;
- page_cache_get(page);
mutex_unlock(&dev->struct_mutex);
-
ret = shmem_pwrite_slow(page, shmem_page_offset, page_length,
user_data, page_do_bit17_swizzling,
partial_cacheline_write,
needs_clflush_after);
mutex_lock(&dev->struct_mutex);
- page_cache_release(page);
+
next_page:
set_page_dirty(page);
mark_page_accessed(page);
- if (release_page)
- page_cache_release(page);
- if (ret) {
- ret = -EFAULT;
+ if (ret)
goto out;
- }
remain -= page_length;
user_data += page_length;
}
out:
+ i915_gem_object_unpin_pages(obj);
+
if (hit_slowpath) {
/* Fixup: Kill any reinstated backing storage pages */
if (obj->madv == __I915_MADV_PURGED)
goto out;
}
- if (obj->gtt_space &&
- obj->cache_level == I915_CACHE_NONE &&
+ if (obj->cache_level == I915_CACHE_NONE &&
obj->tiling_mode == I915_TILING_NONE &&
- obj->map_and_fenceable &&
obj->base.write_domain != I915_GEM_DOMAIN_CPU) {
ret = i915_gem_gtt_pwrite_fast(dev, obj, args, file);
/* Note that the gtt paths might fail with non-page-backed user
* textures). Fallback to the shmem path in that case. */
}
- if (ret == -EFAULT)
+ if (ret == -EFAULT || ret == -ENOSPC)
ret = i915_gem_shmem_pwrite(dev, obj, args, file);
out:
return ret;
}
+int
+i915_gem_check_wedge(struct drm_i915_private *dev_priv,
+ bool interruptible)
+{
+ if (atomic_read(&dev_priv->mm.wedged)) {
+ struct completion *x = &dev_priv->error_completion;
+ bool recovery_complete;
+ unsigned long flags;
+
+ /* Give the error handler a chance to run. */
+ spin_lock_irqsave(&x->wait.lock, flags);
+ recovery_complete = x->done > 0;
+ spin_unlock_irqrestore(&x->wait.lock, flags);
+
+ /* Non-interruptible callers can't handle -EAGAIN, hence return
+ * -EIO unconditionally for these. */
+ if (!interruptible)
+ return -EIO;
+
+ /* Recovery complete, but still wedged means reset failure. */
+ if (recovery_complete)
+ return -EIO;
+
+ return -EAGAIN;
+ }
+
+ return 0;
+}
+
+/*
+ * Compare seqno against outstanding lazy request. Emit a request if they are
+ * equal.
+ */
+static int
+i915_gem_check_olr(struct intel_ring_buffer *ring, u32 seqno)
+{
+ int ret;
+
+ BUG_ON(!mutex_is_locked(&ring->dev->struct_mutex));
+
+ ret = 0;
+ if (seqno == ring->outstanding_lazy_request)
+ ret = i915_add_request(ring, NULL, NULL);
+
+ return ret;
+}
+
+/**
+ * __wait_seqno - wait until execution of seqno has finished
+ * @ring: the ring expected to report seqno
+ * @seqno: duh!
+ * @interruptible: do an interruptible wait (normally yes)
+ * @timeout: in - how long to wait (NULL forever); out - how much time remaining
+ *
+ * Returns 0 if the seqno was found within the alloted time. Else returns the
+ * errno with remaining time filled in timeout argument.
+ */
+static int __wait_seqno(struct intel_ring_buffer *ring, u32 seqno,
+ bool interruptible, struct timespec *timeout)
+{
+ drm_i915_private_t *dev_priv = ring->dev->dev_private;
+ struct timespec before, now, wait_time={1,0};
+ unsigned long timeout_jiffies;
+ long end;
+ bool wait_forever = true;
+ int ret;
+
+ if (i915_seqno_passed(ring->get_seqno(ring, true), seqno))
+ return 0;
+
+ trace_i915_gem_request_wait_begin(ring, seqno);
+
+ if (timeout != NULL) {
+ wait_time = *timeout;
+ wait_forever = false;
+ }
+
+ timeout_jiffies = timespec_to_jiffies(&wait_time);
+
+ if (WARN_ON(!ring->irq_get(ring)))
+ return -ENODEV;
+
+ /* Record current time in case interrupted by signal, or wedged * */
+ getrawmonotonic(&before);
+
+#define EXIT_COND \
+ (i915_seqno_passed(ring->get_seqno(ring, false), seqno) || \
+ atomic_read(&dev_priv->mm.wedged))
+ do {
+ if (interruptible)
+ end = wait_event_interruptible_timeout(ring->irq_queue,
+ EXIT_COND,
+ timeout_jiffies);
+ else
+ end = wait_event_timeout(ring->irq_queue, EXIT_COND,
+ timeout_jiffies);
+
+ ret = i915_gem_check_wedge(dev_priv, interruptible);
+ if (ret)
+ end = ret;
+ } while (end == 0 && wait_forever);
+
+ getrawmonotonic(&now);
+
+ ring->irq_put(ring);
+ trace_i915_gem_request_wait_end(ring, seqno);
+#undef EXIT_COND
+
+ if (timeout) {
+ struct timespec sleep_time = timespec_sub(now, before);
+ *timeout = timespec_sub(*timeout, sleep_time);
+ }
+
+ switch (end) {
+ case -EIO:
+ case -EAGAIN: /* Wedged */
+ case -ERESTARTSYS: /* Signal */
+ return (int)end;
+ case 0: /* Timeout */
+ if (timeout)
+ set_normalized_timespec(timeout, 0, 0);
+ return -ETIME;
+ default: /* Completed */
+ WARN_ON(end < 0); /* We're not aware of other errors */
+ return 0;
+ }
+}
+
+/**
+ * Waits for a sequence number to be signaled, and cleans up the
+ * request and object lists appropriately for that event.
+ */
+int
+i915_wait_seqno(struct intel_ring_buffer *ring, uint32_t seqno)
+{
+ struct drm_device *dev = ring->dev;
+ struct drm_i915_private *dev_priv = dev->dev_private;
+ bool interruptible = dev_priv->mm.interruptible;
+ int ret;
+
+ BUG_ON(!mutex_is_locked(&dev->struct_mutex));
+ BUG_ON(seqno == 0);
+
+ ret = i915_gem_check_wedge(dev_priv, interruptible);
+ if (ret)
+ return ret;
+
+ ret = i915_gem_check_olr(ring, seqno);
+ if (ret)
+ return ret;
+
+ return __wait_seqno(ring, seqno, interruptible, NULL);
+}
+
+/**
+ * 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 __must_check int
+i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj,
+ bool readonly)
+{
+ struct intel_ring_buffer *ring = obj->ring;
+ u32 seqno;
+ int ret;
+
+ seqno = readonly ? obj->last_write_seqno : obj->last_read_seqno;
+ if (seqno == 0)
+ return 0;
+
+ ret = i915_wait_seqno(ring, seqno);
+ if (ret)
+ return ret;
+
+ i915_gem_retire_requests_ring(ring);
+
+ /* Manually manage the write flush as we may have not yet
+ * retired the buffer.
+ */
+ if (obj->last_write_seqno &&
+ i915_seqno_passed(seqno, obj->last_write_seqno)) {
+ obj->last_write_seqno = 0;
+ obj->base.write_domain &= ~I915_GEM_GPU_DOMAINS;
+ }
+
+ return 0;
+}
+
+/* A nonblocking variant of the above wait. This is a highly dangerous routine
+ * as the object state may change during this call.
+ */
+static __must_check int
+i915_gem_object_wait_rendering__nonblocking(struct drm_i915_gem_object *obj,
+ bool readonly)
+{
+ struct drm_device *dev = obj->base.dev;
+ struct drm_i915_private *dev_priv = dev->dev_private;
+ struct intel_ring_buffer *ring = obj->ring;
+ u32 seqno;
+ int ret;
+
+ BUG_ON(!mutex_is_locked(&dev->struct_mutex));
+ BUG_ON(!dev_priv->mm.interruptible);
+
+ seqno = readonly ? obj->last_write_seqno : obj->last_read_seqno;
+ if (seqno == 0)
+ return 0;
+
+ ret = i915_gem_check_wedge(dev_priv, true);
+ if (ret)
+ return ret;
+
+ ret = i915_gem_check_olr(ring, seqno);
+ if (ret)
+ return ret;
+
+ mutex_unlock(&dev->struct_mutex);
+ ret = __wait_seqno(ring, seqno, true, NULL);
+ mutex_lock(&dev->struct_mutex);
+
+ i915_gem_retire_requests_ring(ring);
+
+ /* Manually manage the write flush as we may have not yet
+ * retired the buffer.
+ */
+ if (obj->last_write_seqno &&
+ i915_seqno_passed(seqno, obj->last_write_seqno)) {
+ obj->last_write_seqno = 0;
+ obj->base.write_domain &= ~I915_GEM_GPU_DOMAINS;
+ }
+
+ return ret;
+}
+
/**
* Called when user space prepares to use an object with the CPU, either
* through the mmap ioctl's mapping or a GTT mapping.
goto unlock;
}
+ /* Try to flush the object off the GPU without holding the lock.
+ * We will repeat the flush holding the lock in the normal manner
+ * to catch cases where we are gazumped.
+ */
+ ret = i915_gem_object_wait_rendering__nonblocking(obj, !write_domain);
+ if (ret)
+ goto unref;
+
if (read_domains & I915_GEM_DOMAIN_GTT) {
ret = i915_gem_object_set_to_gtt_domain(obj, write_domain != 0);
ret = i915_gem_object_set_to_cpu_domain(obj, write_domain != 0);
}
+unref:
drm_gem_object_unreference(&obj->base);
unlock:
mutex_unlock(&dev->struct_mutex);
goto unlock;
}
if (!obj->gtt_space) {
- ret = i915_gem_object_bind_to_gtt(obj, 0, true);
+ ret = i915_gem_object_bind_to_gtt(obj, 0, true, false);
if (ret)
goto unlock;
return i915_gem_get_gtt_size(dev, size, tiling_mode);
}
-int
-i915_gem_mmap_gtt(struct drm_file *file,
+static int i915_gem_object_create_mmap_offset(struct drm_i915_gem_object *obj)
+{
+ struct drm_i915_private *dev_priv = obj->base.dev->dev_private;
+ int ret;
+
+ if (obj->base.map_list.map)
+ return 0;
+
+ ret = drm_gem_create_mmap_offset(&obj->base);
+ if (ret != -ENOSPC)
+ return ret;
+
+ /* Badly fragmented mmap space? The only way we can recover
+ * space is by destroying unwanted objects. We can't randomly release
+ * mmap_offsets as userspace expects them to be persistent for the
+ * lifetime of the objects. The closest we can is to release the
+ * offsets on purgeable objects by truncating it and marking it purged,
+ * which prevents userspace from ever using that object again.
+ */
+ i915_gem_purge(dev_priv, obj->base.size >> PAGE_SHIFT);
+ ret = drm_gem_create_mmap_offset(&obj->base);
+ if (ret != -ENOSPC)
+ return ret;
+
+ i915_gem_shrink_all(dev_priv);
+ return drm_gem_create_mmap_offset(&obj->base);
+}
+
+static void i915_gem_object_free_mmap_offset(struct drm_i915_gem_object *obj)
+{
+ if (!obj->base.map_list.map)
+ return;
+
+ drm_gem_free_mmap_offset(&obj->base);
+}
+
+int
+i915_gem_mmap_gtt(struct drm_file *file,
struct drm_device *dev,
uint32_t handle,
uint64_t *offset)
goto out;
}
- if (!obj->base.map_list.map) {
- ret = drm_gem_create_mmap_offset(&obj->base);
- if (ret)
- goto out;
- }
+ ret = i915_gem_object_create_mmap_offset(obj);
+ if (ret)
+ goto out;
*offset = (u64)obj->base.map_list.hash.key << PAGE_SHIFT;
return i915_gem_mmap_gtt(file, dev, args->handle, &args->offset);
}
-int
-i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj,
- gfp_t gfpmask)
+/* Immediately discard the backing storage */
+static void
+i915_gem_object_truncate(struct drm_i915_gem_object *obj)
{
- int page_count, i;
- struct address_space *mapping;
struct inode *inode;
- struct page *page;
- if (obj->pages || obj->sg_table)
- return 0;
+ i915_gem_object_free_mmap_offset(obj);
- /* Get the list of pages out of our struct file. They'll be pinned
- * at this point until we release them.
- */
- 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;
+ if (obj->base.filp == NULL)
+ return;
+ /* Our goal here is to return as much of the memory as
+ * is possible back to the system as we are called from OOM.
+ * To do this we must instruct the shmfs to drop all of its
+ * backing pages, *now*.
+ */
inode = obj->base.filp->f_path.dentry->d_inode;
- mapping = inode->i_mapping;
- gfpmask |= mapping_gfp_mask(mapping);
-
- for (i = 0; i < page_count; i++) {
- page = shmem_read_mapping_page_gfp(mapping, i, gfpmask);
- if (IS_ERR(page))
- goto err_pages;
-
- obj->pages[i] = page;
- }
-
- if (i915_gem_object_needs_bit17_swizzle(obj))
- i915_gem_object_do_bit_17_swizzle(obj);
-
- return 0;
+ shmem_truncate_range(inode, 0, (loff_t)-1);
-err_pages:
- while (i--)
- page_cache_release(obj->pages[i]);
+ obj->madv = __I915_MADV_PURGED;
+}
- drm_free_large(obj->pages);
- obj->pages = NULL;
- return PTR_ERR(page);
+static inline int
+i915_gem_object_is_purgeable(struct drm_i915_gem_object *obj)
+{
+ return obj->madv == I915_MADV_DONTNEED;
}
static void
i915_gem_object_put_pages_gtt(struct drm_i915_gem_object *obj)
{
int page_count = obj->base.size / PAGE_SIZE;
- int i;
-
- if (!obj->pages)
- return;
+ struct scatterlist *sg;
+ int ret, i;
BUG_ON(obj->madv == __I915_MADV_PURGED);
+ ret = i915_gem_object_set_to_cpu_domain(obj, true);
+ if (ret) {
+ /* In the event of a disaster, abandon all caches and
+ * hope for the best.
+ */
+ WARN_ON(ret != -EIO);
+ i915_gem_clflush_object(obj);
+ obj->base.read_domains = obj->base.write_domain = I915_GEM_DOMAIN_CPU;
+ }
+
if (i915_gem_object_needs_bit17_swizzle(obj))
i915_gem_object_save_bit_17_swizzle(obj);
if (obj->madv == I915_MADV_DONTNEED)
obj->dirty = 0;
- for (i = 0; i < page_count; i++) {
+ for_each_sg(obj->pages->sgl, sg, page_count, i) {
+ struct page *page = sg_page(sg);
+
if (obj->dirty)
- set_page_dirty(obj->pages[i]);
+ set_page_dirty(page);
if (obj->madv == I915_MADV_WILLNEED)
- mark_page_accessed(obj->pages[i]);
+ mark_page_accessed(page);
- page_cache_release(obj->pages[i]);
+ page_cache_release(page);
}
obj->dirty = 0;
- drm_free_large(obj->pages);
+ sg_free_table(obj->pages);
+ kfree(obj->pages);
+}
+
+static int
+i915_gem_object_put_pages(struct drm_i915_gem_object *obj)
+{
+ const struct drm_i915_gem_object_ops *ops = obj->ops;
+
+ if (obj->pages == NULL)
+ return 0;
+
+ BUG_ON(obj->gtt_space);
+
+ if (obj->pages_pin_count)
+ return -EBUSY;
+
+ ops->put_pages(obj);
obj->pages = NULL;
+
+ list_del(&obj->gtt_list);
+ if (i915_gem_object_is_purgeable(obj))
+ i915_gem_object_truncate(obj);
+
+ return 0;
+}
+
+static long
+i915_gem_purge(struct drm_i915_private *dev_priv, long target)
+{
+ struct drm_i915_gem_object *obj, *next;
+ long count = 0;
+
+ list_for_each_entry_safe(obj, next,
+ &dev_priv->mm.unbound_list,
+ gtt_list) {
+ if (i915_gem_object_is_purgeable(obj) &&
+ i915_gem_object_put_pages(obj) == 0) {
+ count += obj->base.size >> PAGE_SHIFT;
+ if (count >= target)
+ return count;
+ }
+ }
+
+ list_for_each_entry_safe(obj, next,
+ &dev_priv->mm.inactive_list,
+ mm_list) {
+ if (i915_gem_object_is_purgeable(obj) &&
+ i915_gem_object_unbind(obj) == 0 &&
+ i915_gem_object_put_pages(obj) == 0) {
+ count += obj->base.size >> PAGE_SHIFT;
+ if (count >= target)
+ return count;
+ }
+ }
+
+ return count;
+}
+
+static void
+i915_gem_shrink_all(struct drm_i915_private *dev_priv)
+{
+ struct drm_i915_gem_object *obj, *next;
+
+ i915_gem_evict_everything(dev_priv->dev);
+
+ list_for_each_entry_safe(obj, next, &dev_priv->mm.unbound_list, gtt_list)
+ i915_gem_object_put_pages(obj);
+}
+
+static int
+i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
+{
+ struct drm_i915_private *dev_priv = obj->base.dev->dev_private;
+ int page_count, i;
+ struct address_space *mapping;
+ struct sg_table *st;
+ struct scatterlist *sg;
+ struct page *page;
+ gfp_t gfp;
+
+ /* 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->base.read_domains & I915_GEM_GPU_DOMAINS);
+ BUG_ON(obj->base.write_domain & I915_GEM_GPU_DOMAINS);
+
+ st = kmalloc(sizeof(*st), GFP_KERNEL);
+ if (st == NULL)
+ return -ENOMEM;
+
+ page_count = obj->base.size / PAGE_SIZE;
+ if (sg_alloc_table(st, page_count, GFP_KERNEL)) {
+ sg_free_table(st);
+ kfree(st);
+ return -ENOMEM;
+ }
+
+ /* Get the list of pages out of our struct file. They'll be pinned
+ * at this point until we release them.
+ *
+ * Fail silently without starting the shrinker
+ */
+ mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
+ gfp = mapping_gfp_mask(mapping);
+ gfp |= __GFP_NORETRY | __GFP_NOWARN;
+ gfp &= ~(__GFP_IO | __GFP_WAIT);
+ for_each_sg(st->sgl, sg, page_count, i) {
+ page = shmem_read_mapping_page_gfp(mapping, i, gfp);
+ if (IS_ERR(page)) {
+ i915_gem_purge(dev_priv, page_count);
+ page = shmem_read_mapping_page_gfp(mapping, i, gfp);
+ }
+ if (IS_ERR(page)) {
+ /* We've tried hard to allocate the memory by reaping
+ * our own buffer, now let the real VM do its job and
+ * go down in flames if truly OOM.
+ */
+ gfp &= ~(__GFP_NORETRY | __GFP_NOWARN);
+ gfp |= __GFP_IO | __GFP_WAIT;
+
+ i915_gem_shrink_all(dev_priv);
+ page = shmem_read_mapping_page_gfp(mapping, i, gfp);
+ if (IS_ERR(page))
+ goto err_pages;
+
+ gfp |= __GFP_NORETRY | __GFP_NOWARN;
+ gfp &= ~(__GFP_IO | __GFP_WAIT);
+ }
+
+ sg_set_page(sg, page, PAGE_SIZE, 0);
+ }
+
+ if (i915_gem_object_needs_bit17_swizzle(obj))
+ i915_gem_object_do_bit_17_swizzle(obj);
+
+ obj->pages = st;
+ return 0;
+
+err_pages:
+ for_each_sg(st->sgl, sg, i, page_count)
+ page_cache_release(sg_page(sg));
+ sg_free_table(st);
+ kfree(st);
+ return PTR_ERR(page);
+}
+
+/* Ensure that the associated pages are gathered from the backing storage
+ * and pinned into our object. i915_gem_object_get_pages() may be called
+ * multiple times before they are released by a single call to
+ * i915_gem_object_put_pages() - once the pages are no longer referenced
+ * either as a result of memory pressure (reaping pages under the shrinker)
+ * or as the object is itself released.
+ */
+int
+i915_gem_object_get_pages(struct drm_i915_gem_object *obj)
+{
+ struct drm_i915_private *dev_priv = obj->base.dev->dev_private;
+ const struct drm_i915_gem_object_ops *ops = obj->ops;
+ int ret;
+
+ if (obj->pages)
+ return 0;
+
+ BUG_ON(obj->pages_pin_count);
+
+ ret = ops->get_pages(obj);
+ if (ret)
+ return ret;
+
+ list_add_tail(&obj->gtt_list, &dev_priv->mm.unbound_list);
+ return 0;
}
void
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;
+ obj->last_read_seqno = seqno;
if (obj->fenced_gpu_access) {
obj->last_fenced_seqno = seqno;
}
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_i915_gem_object *obj)
+i915_gem_object_move_to_inactive(struct drm_i915_gem_object *obj)
{
struct drm_device *dev = obj->base.dev;
- drm_i915_private_t *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = dev->dev_private;
+ BUG_ON(obj->base.write_domain & ~I915_GEM_GPU_DOMAINS);
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) /* are we a framebuffer? */
+ intel_mark_fb_idle(obj);
list_move_tail(&obj->mm_list, &dev_priv->mm.inactive_list);
- BUG_ON(!list_empty(&obj->gpu_write_list));
- BUG_ON(!obj->active);
+ list_del_init(&obj->ring_list);
obj->ring = NULL;
- i915_gem_object_move_off_active(obj);
+ obj->last_read_seqno = 0;
+ obj->last_write_seqno = 0;
+ obj->base.write_domain = 0;
+
+ obj->last_fenced_seqno = 0;
obj->fenced_gpu_access = false;
obj->active = 0;
- obj->pending_gpu_write = false;
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_i915_gem_object *obj)
-{
- struct inode *inode;
-
- /* Our goal here is to return as much of the memory as
- * is possible back to the system as we are called from OOM.
- * To do this we must instruct the shmfs to drop all of its
- * backing pages, *now*.
- */
- inode = obj->base.filp->f_path.dentry->d_inode;
- shmem_truncate_range(inode, 0, (loff_t)-1);
-
- if (obj->base.map_list.map)
- drm_gem_free_mmap_offset(&obj->base);
-
- obj->madv = __I915_MADV_PURGED;
-}
-
-static inline int
-i915_gem_object_is_purgeable(struct drm_i915_gem_object *obj)
-{
- return obj->madv == I915_MADV_DONTNEED;
-}
-
-static void
-i915_gem_process_flushing_list(struct intel_ring_buffer *ring,
- uint32_t flush_domains)
-{
- struct drm_i915_gem_object *obj, *next;
-
- list_for_each_entry_safe(obj, next,
- &ring->gpu_write_list,
- gpu_write_list) {
- if (obj->base.write_domain & flush_domains) {
- uint32_t old_write_domain = obj->base.write_domain;
-
- obj->base.write_domain = 0;
- list_del_init(&obj->gpu_write_list);
- i915_gem_object_move_to_active(obj, ring,
- i915_gem_next_request_seqno(ring));
-
- trace_i915_gem_object_change_domain(obj,
- obj->base.read_domains,
- old_write_domain);
- }
- }
-}
-
static u32
i915_gem_get_seqno(struct drm_device *dev)
{
* is that the flush _must_ happen before the next request, no matter
* what.
*/
- if (ring->gpu_caches_dirty) {
- ret = i915_gem_flush_ring(ring, 0, I915_GEM_GPU_DOMAINS);
- if (ret)
- return ret;
+ ret = intel_ring_flush_all_caches(ring);
+ if (ret)
+ return ret;
- ring->gpu_caches_dirty = false;
+ if (request == NULL) {
+ request = kmalloc(sizeof(*request), GFP_KERNEL);
+ if (request == NULL)
+ return -ENOMEM;
}
- BUG_ON(request == NULL);
seqno = i915_gem_next_request_seqno(ring);
/* Record the position of the start of the request so that
request_ring_position = intel_ring_get_tail(ring);
ret = ring->add_request(ring, &seqno);
- if (ret)
- return ret;
+ if (ret) {
+ kfree(request);
+ return ret;
+ }
trace_i915_gem_request_add(ring, seqno);
request->emitted_jiffies = jiffies;
was_empty = list_empty(&ring->request_list);
list_add_tail(&request->list, &ring->request_list);
+ request->file_priv = NULL;
if (file) {
struct drm_i915_file_private *file_priv = file->driver_priv;
jiffies +
msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD));
}
- if (was_empty)
+ if (was_empty) {
queue_delayed_work(dev_priv->wq,
&dev_priv->mm.retire_work, HZ);
+ intel_mark_busy(dev_priv->dev);
+ }
}
- WARN_ON(!list_empty(&ring->gpu_write_list));
-
return 0;
}
struct drm_i915_gem_object,
ring_list);
- obj->base.write_domain = 0;
- list_del_init(&obj->gpu_write_list);
i915_gem_object_move_to_inactive(obj);
}
}
for_each_ring(ring, dev_priv, i)
i915_gem_reset_ring_lists(dev_priv, ring);
- /* Remove anything from the flushing lists. The GPU cache is likely
- * to be lost on reset along with the data, so simply move the
- * lost bo to the inactive list.
- */
- while (!list_empty(&dev_priv->mm.flushing_list)) {
- obj = list_first_entry(&dev_priv->mm.flushing_list,
- struct drm_i915_gem_object,
- mm_list);
-
- 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.
*/
WARN_ON(i915_verify_lists(ring->dev));
- seqno = ring->get_seqno(ring);
+ seqno = ring->get_seqno(ring, true);
for (i = 0; i < ARRAY_SIZE(ring->sync_seqno); i++)
if (seqno >= ring->sync_seqno[i])
struct drm_i915_gem_object,
ring_list);
- if (!i915_seqno_passed(seqno, obj->last_rendering_seqno))
+ if (!i915_seqno_passed(seqno, obj->last_read_seqno))
break;
- if (obj->base.write_domain != 0)
- i915_gem_object_move_to_flushing(obj);
- else
- i915_gem_object_move_to_inactive(obj);
+ i915_gem_object_move_to_inactive(obj);
}
if (unlikely(ring->trace_irq_seqno &&
dev = dev_priv->dev;
/* Come back later if the device is busy... */
- if (!mutex_trylock(&dev->struct_mutex)) {
- queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ);
- return;
- }
-
- i915_gem_retire_requests(dev);
-
- /* Send a periodic flush down the ring so we don't hold onto GEM
- * objects indefinitely.
- */
- idle = true;
- for_each_ring(ring, dev_priv, i) {
- if (ring->gpu_caches_dirty) {
- struct drm_i915_gem_request *request;
-
- request = kzalloc(sizeof(*request), GFP_KERNEL);
- if (request == NULL ||
- i915_add_request(ring, NULL, request))
- kfree(request);
- }
-
- idle &= list_empty(&ring->request_list);
- }
-
- if (!dev_priv->mm.suspended && !idle)
- queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ);
-
- mutex_unlock(&dev->struct_mutex);
-}
-
-int
-i915_gem_check_wedge(struct drm_i915_private *dev_priv,
- bool interruptible)
-{
- if (atomic_read(&dev_priv->mm.wedged)) {
- struct completion *x = &dev_priv->error_completion;
- bool recovery_complete;
- unsigned long flags;
-
- /* Give the error handler a chance to run. */
- spin_lock_irqsave(&x->wait.lock, flags);
- recovery_complete = x->done > 0;
- spin_unlock_irqrestore(&x->wait.lock, flags);
-
- /* Non-interruptible callers can't handle -EAGAIN, hence return
- * -EIO unconditionally for these. */
- if (!interruptible)
- return -EIO;
-
- /* Recovery complete, but still wedged means reset failure. */
- if (recovery_complete)
- return -EIO;
-
- return -EAGAIN;
- }
-
- return 0;
-}
-
-/*
- * Compare seqno against outstanding lazy request. Emit a request if they are
- * equal.
- */
-static int
-i915_gem_check_olr(struct intel_ring_buffer *ring, u32 seqno)
-{
- int ret = 0;
-
- BUG_ON(!mutex_is_locked(&ring->dev->struct_mutex));
-
- if (seqno == ring->outstanding_lazy_request) {
- struct drm_i915_gem_request *request;
-
- request = kzalloc(sizeof(*request), GFP_KERNEL);
- if (request == NULL)
- return -ENOMEM;
-
- ret = i915_add_request(ring, NULL, request);
- if (ret) {
- kfree(request);
- return ret;
- }
-
- BUG_ON(seqno != request->seqno);
- }
-
- return ret;
-}
-
-/**
- * __wait_seqno - wait until execution of seqno has finished
- * @ring: the ring expected to report seqno
- * @seqno: duh!
- * @interruptible: do an interruptible wait (normally yes)
- * @timeout: in - how long to wait (NULL forever); out - how much time remaining
- *
- * Returns 0 if the seqno was found within the alloted time. Else returns the
- * errno with remaining time filled in timeout argument.
- */
-static int __wait_seqno(struct intel_ring_buffer *ring, u32 seqno,
- bool interruptible, struct timespec *timeout)
-{
- drm_i915_private_t *dev_priv = ring->dev->dev_private;
- struct timespec before, now, wait_time={1,0};
- unsigned long timeout_jiffies;
- long end;
- bool wait_forever = true;
- int ret;
-
- if (i915_seqno_passed(ring->get_seqno(ring), seqno))
- return 0;
-
- trace_i915_gem_request_wait_begin(ring, seqno);
-
- if (timeout != NULL) {
- wait_time = *timeout;
- wait_forever = false;
- }
-
- timeout_jiffies = timespec_to_jiffies(&wait_time);
-
- if (WARN_ON(!ring->irq_get(ring)))
- return -ENODEV;
-
- /* Record current time in case interrupted by signal, or wedged * */
- getrawmonotonic(&before);
-
-#define EXIT_COND \
- (i915_seqno_passed(ring->get_seqno(ring), seqno) || \
- atomic_read(&dev_priv->mm.wedged))
- do {
- if (interruptible)
- end = wait_event_interruptible_timeout(ring->irq_queue,
- EXIT_COND,
- timeout_jiffies);
- else
- end = wait_event_timeout(ring->irq_queue, EXIT_COND,
- timeout_jiffies);
-
- ret = i915_gem_check_wedge(dev_priv, interruptible);
- if (ret)
- end = ret;
- } while (end == 0 && wait_forever);
-
- getrawmonotonic(&now);
-
- ring->irq_put(ring);
- trace_i915_gem_request_wait_end(ring, seqno);
-#undef EXIT_COND
-
- if (timeout) {
- struct timespec sleep_time = timespec_sub(now, before);
- *timeout = timespec_sub(*timeout, sleep_time);
- }
-
- switch (end) {
- case -EIO:
- case -EAGAIN: /* Wedged */
- case -ERESTARTSYS: /* Signal */
- return (int)end;
- case 0: /* Timeout */
- if (timeout)
- set_normalized_timespec(timeout, 0, 0);
- return -ETIME;
- default: /* Completed */
- WARN_ON(end < 0); /* We're not aware of other errors */
- return 0;
- }
-}
-
-/**
- * Waits for a sequence number to be signaled, and cleans up the
- * request and object lists appropriately for that event.
- */
-int
-i915_wait_seqno(struct intel_ring_buffer *ring, uint32_t seqno)
-{
- drm_i915_private_t *dev_priv = ring->dev->dev_private;
- int ret = 0;
-
- BUG_ON(seqno == 0);
-
- ret = i915_gem_check_wedge(dev_priv, dev_priv->mm.interruptible);
- if (ret)
- return ret;
-
- ret = i915_gem_check_olr(ring, seqno);
- if (ret)
- return ret;
-
- ret = __wait_seqno(ring, seqno, dev_priv->mm.interruptible, NULL);
-
- return ret;
-}
-
-/**
- * Ensures that all rendering to the object has completed and the object is
- * safe to unbind from the GTT or access from the CPU.
- */
-int
-i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj)
-{
- int ret;
+ if (!mutex_trylock(&dev->struct_mutex)) {
+ queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ);
+ return;
+ }
- /* This function only exists to support waiting for existing rendering,
- * not for emitting required flushes.
- */
- BUG_ON((obj->base.write_domain & I915_GEM_GPU_DOMAINS) != 0);
+ i915_gem_retire_requests(dev);
- /* If there is rendering queued on the buffer being evicted, wait for
- * it.
+ /* Send a periodic flush down the ring so we don't hold onto GEM
+ * objects indefinitely.
*/
- if (obj->active) {
- ret = i915_wait_seqno(obj->ring, obj->last_rendering_seqno);
- if (ret)
- return ret;
- i915_gem_retire_requests_ring(obj->ring);
+ idle = true;
+ for_each_ring(ring, dev_priv, i) {
+ if (ring->gpu_caches_dirty)
+ i915_add_request(ring, NULL, NULL);
+
+ idle &= list_empty(&ring->request_list);
}
- return 0;
+ if (!dev_priv->mm.suspended && !idle)
+ queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ);
+ if (idle)
+ intel_mark_idle(dev);
+
+ mutex_unlock(&dev->struct_mutex);
}
/**
int ret;
if (obj->active) {
- ret = i915_gem_object_flush_gpu_write_domain(obj);
+ ret = i915_gem_check_olr(obj->ring, obj->last_read_seqno);
if (ret)
return ret;
- ret = i915_gem_check_olr(obj->ring,
- obj->last_rendering_seqno);
- if (ret)
- return ret;
i915_gem_retire_requests_ring(obj->ring);
}
goto out;
if (obj->active) {
- seqno = obj->last_rendering_seqno;
+ seqno = obj->last_read_seqno;
ring = obj->ring;
}
return 0;
if (to == NULL || !i915_semaphore_is_enabled(obj->base.dev))
- return i915_gem_object_wait_rendering(obj);
+ return i915_gem_object_wait_rendering(obj, false);
idx = intel_ring_sync_index(from, to);
- seqno = obj->last_rendering_seqno;
+ seqno = obj->last_read_seqno;
if (seqno <= from->sync_seqno[idx])
return 0;
if (obj->pin_count)
return -EBUSY;
+ BUG_ON(obj->pages == NULL);
+
ret = i915_gem_object_finish_gpu(obj);
if (ret)
return ret;
i915_gem_object_finish_gtt(obj);
- /* Move the object to the CPU domain to ensure that
- * any possible CPU writes while it's not in the GTT
- * are flushed when we go to remap it.
- */
- if (ret == 0)
- ret = i915_gem_object_set_to_cpu_domain(obj, 1);
- if (ret == -ERESTARTSYS)
- return ret;
- if (ret) {
- /* In the event of a disaster, abandon all caches and
- * hope for the best.
- */
- i915_gem_clflush_object(obj);
- obj->base.read_domains = obj->base.write_domain = I915_GEM_DOMAIN_CPU;
- }
-
/* release the fence reg _after_ flushing */
ret = i915_gem_object_put_fence(obj);
if (ret)
}
i915_gem_gtt_finish_object(obj);
- i915_gem_object_put_pages_gtt(obj);
-
- list_del_init(&obj->gtt_list);
- list_del_init(&obj->mm_list);
+ list_del(&obj->mm_list);
+ list_move_tail(&obj->gtt_list, &dev_priv->mm.unbound_list);
/* Avoid an unnecessary call to unbind on rebind. */
obj->map_and_fenceable = true;
obj->gtt_space = NULL;
obj->gtt_offset = 0;
- if (i915_gem_object_is_purgeable(obj))
- i915_gem_object_truncate(obj);
-
- return ret;
-}
-
-int
-i915_gem_flush_ring(struct intel_ring_buffer *ring,
- uint32_t invalidate_domains,
- uint32_t flush_domains)
-{
- int ret;
-
- if (((invalidate_domains | flush_domains) & I915_GEM_GPU_DOMAINS) == 0)
- return 0;
-
- trace_i915_gem_ring_flush(ring, invalidate_domains, flush_domains);
-
- ret = ring->flush(ring, invalidate_domains, flush_domains);
- if (ret)
- return ret;
-
- if (flush_domains & I915_GEM_GPU_DOMAINS)
- i915_gem_process_flushing_list(ring, flush_domains);
-
return 0;
}
static int i915_ring_idle(struct intel_ring_buffer *ring)
{
- int ret;
-
- if (list_empty(&ring->gpu_write_list) && list_empty(&ring->active_list))
+ if (list_empty(&ring->active_list))
return 0;
- if (!list_empty(&ring->gpu_write_list)) {
- ret = i915_gem_flush_ring(ring,
- I915_GEM_GPU_DOMAINS, I915_GEM_GPU_DOMAINS);
- if (ret)
- return ret;
- }
-
return i915_wait_seqno(ring, i915_gem_next_request_seqno(ring));
}
ret = i915_ring_idle(ring);
if (ret)
return ret;
-
- /* Is the device fubar? */
- if (WARN_ON(!list_empty(&ring->gpu_write_list)))
- return -EBUSY;
}
return 0;
static int
i915_gem_object_flush_fence(struct drm_i915_gem_object *obj)
{
- int ret;
-
- if (obj->fenced_gpu_access) {
- if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) {
- ret = i915_gem_flush_ring(obj->ring,
- 0, obj->base.write_domain);
- if (ret)
- return ret;
- }
-
- obj->fenced_gpu_access = false;
- }
-
if (obj->last_fenced_seqno) {
- ret = i915_wait_seqno(obj->ring, obj->last_fenced_seqno);
+ int ret = i915_wait_seqno(obj->ring, obj->last_fenced_seqno);
if (ret)
return ret;
if (obj->base.read_domains & I915_GEM_DOMAIN_GTT)
mb();
+ obj->fenced_gpu_access = false;
return 0;
}
return 0;
}
+static bool i915_gem_valid_gtt_space(struct drm_device *dev,
+ struct drm_mm_node *gtt_space,
+ unsigned long cache_level)
+{
+ struct drm_mm_node *other;
+
+ /* On non-LLC machines we have to be careful when putting differing
+ * types of snoopable memory together to avoid the prefetcher
+ * crossing memory domains and dieing.
+ */
+ if (HAS_LLC(dev))
+ return true;
+
+ if (gtt_space == NULL)
+ return true;
+
+ if (list_empty(>t_space->node_list))
+ return true;
+
+ other = list_entry(gtt_space->node_list.prev, struct drm_mm_node, node_list);
+ if (other->allocated && !other->hole_follows && other->color != cache_level)
+ return false;
+
+ other = list_entry(gtt_space->node_list.next, struct drm_mm_node, node_list);
+ if (other->allocated && !gtt_space->hole_follows && other->color != cache_level)
+ return false;
+
+ return true;
+}
+
+static void i915_gem_verify_gtt(struct drm_device *dev)
+{
+#if WATCH_GTT
+ struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_gem_object *obj;
+ int err = 0;
+
+ list_for_each_entry(obj, &dev_priv->mm.gtt_list, gtt_list) {
+ if (obj->gtt_space == NULL) {
+ printk(KERN_ERR "object found on GTT list with no space reserved\n");
+ err++;
+ continue;
+ }
+
+ if (obj->cache_level != obj->gtt_space->color) {
+ printk(KERN_ERR "object reserved space [%08lx, %08lx] with wrong color, cache_level=%x, color=%lx\n",
+ obj->gtt_space->start,
+ obj->gtt_space->start + obj->gtt_space->size,
+ obj->cache_level,
+ obj->gtt_space->color);
+ err++;
+ continue;
+ }
+
+ if (!i915_gem_valid_gtt_space(dev,
+ obj->gtt_space,
+ obj->cache_level)) {
+ printk(KERN_ERR "invalid GTT space found at [%08lx, %08lx] - color=%x\n",
+ obj->gtt_space->start,
+ obj->gtt_space->start + obj->gtt_space->size,
+ obj->cache_level);
+ err++;
+ continue;
+ }
+ }
+
+ WARN_ON(err);
+#endif
+}
+
/**
* Finds free space in the GTT aperture and binds the object there.
*/
static int
i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj,
unsigned alignment,
- bool map_and_fenceable)
+ bool map_and_fenceable,
+ bool nonblocking)
{
struct drm_device *dev = obj->base.dev;
drm_i915_private_t *dev_priv = dev->dev_private;
struct drm_mm_node *free_space;
- gfp_t gfpmask = __GFP_NORETRY | __GFP_NOWARN;
u32 size, fence_size, fence_alignment, unfenced_alignment;
bool mappable, fenceable;
int ret;
return -E2BIG;
}
+ ret = i915_gem_object_get_pages(obj);
+ if (ret)
+ return ret;
+
search_free:
if (map_and_fenceable)
free_space =
- drm_mm_search_free_in_range(&dev_priv->mm.gtt_space,
- size, alignment,
- 0, dev_priv->mm.gtt_mappable_end,
- 0);
+ drm_mm_search_free_in_range_color(&dev_priv->mm.gtt_space,
+ size, alignment, obj->cache_level,
+ 0, dev_priv->mm.gtt_mappable_end,
+ false);
else
- free_space = drm_mm_search_free(&dev_priv->mm.gtt_space,
- size, alignment, 0);
+ free_space = drm_mm_search_free_color(&dev_priv->mm.gtt_space,
+ size, alignment, obj->cache_level,
+ false);
if (free_space != NULL) {
if (map_and_fenceable)
obj->gtt_space =
drm_mm_get_block_range_generic(free_space,
- size, alignment, 0,
+ size, alignment, obj->cache_level,
0, dev_priv->mm.gtt_mappable_end,
- 0);
+ false);
else
obj->gtt_space =
- drm_mm_get_block(free_space, size, alignment);
+ drm_mm_get_block_generic(free_space,
+ size, alignment, obj->cache_level,
+ false);
}
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,
- map_and_fenceable);
+ obj->cache_level,
+ map_and_fenceable,
+ nonblocking);
if (ret)
return ret;
goto search_free;
}
-
- ret = i915_gem_object_get_pages_gtt(obj, gfpmask);
- if (ret) {
+ if (WARN_ON(!i915_gem_valid_gtt_space(dev,
+ obj->gtt_space,
+ obj->cache_level))) {
drm_mm_put_block(obj->gtt_space);
obj->gtt_space = NULL;
-
- if (ret == -ENOMEM) {
- /* first try to reclaim some memory by clearing the GTT */
- ret = i915_gem_evict_everything(dev, false);
- if (ret) {
- /* now try to shrink everyone else */
- if (gfpmask) {
- gfpmask = 0;
- goto search_free;
- }
-
- return -ENOMEM;
- }
-
- goto search_free;
- }
-
- return ret;
+ return -EINVAL;
}
+
ret = i915_gem_gtt_prepare_object(obj);
if (ret) {
- i915_gem_object_put_pages_gtt(obj);
drm_mm_put_block(obj->gtt_space);
obj->gtt_space = NULL;
-
- if (i915_gem_evict_everything(dev, false))
- return ret;
-
- goto search_free;
+ return ret;
}
if (!dev_priv->mm.aliasing_ppgtt)
i915_gem_gtt_bind_object(obj, obj->cache_level);
- list_add_tail(&obj->gtt_list, &dev_priv->mm.gtt_list);
+ list_move_tail(&obj->gtt_list, &dev_priv->mm.bound_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->base.read_domains & I915_GEM_GPU_DOMAINS);
- BUG_ON(obj->base.write_domain & I915_GEM_GPU_DOMAINS);
-
obj->gtt_offset = obj->gtt_space->start;
fenceable =
obj->map_and_fenceable = mappable && fenceable;
trace_i915_gem_object_bind(obj, map_and_fenceable);
+ i915_gem_verify_gtt(dev);
return 0;
}
trace_i915_gem_object_clflush(obj);
- 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_i915_gem_object *obj)
-{
- if ((obj->base.write_domain & I915_GEM_GPU_DOMAINS) == 0)
- return 0;
-
- /* Queue the GPU write cache flushing we need. */
- return i915_gem_flush_ring(obj->ring, 0, obj->base.write_domain);
+ drm_clflush_sg(obj->pages);
}
/** Flushes the GTT write domain for the object if it's dirty. */
if (obj->base.write_domain == I915_GEM_DOMAIN_GTT)
return 0;
- ret = i915_gem_object_flush_gpu_write_domain(obj);
+ ret = i915_gem_object_wait_rendering(obj, !write);
if (ret)
return ret;
- if (obj->pending_gpu_write || write) {
- ret = i915_gem_object_wait_rendering(obj);
- if (ret)
- return ret;
- }
-
i915_gem_object_flush_cpu_write_domain(obj);
old_write_domain = obj->base.write_domain;
return -EBUSY;
}
+ if (!i915_gem_valid_gtt_space(dev, obj->gtt_space, cache_level)) {
+ ret = i915_gem_object_unbind(obj);
+ if (ret)
+ return ret;
+ }
+
if (obj->gtt_space) {
ret = i915_gem_object_finish_gpu(obj);
if (ret)
* registers with snooped memory, so relinquish any fences
* currently pointing to our region in the aperture.
*/
- if (INTEL_INFO(obj->base.dev)->gen < 6) {
+ if (INTEL_INFO(dev)->gen < 6) {
ret = i915_gem_object_put_fence(obj);
if (ret)
return ret;
if (obj->has_aliasing_ppgtt_mapping)
i915_ppgtt_bind_object(dev_priv->mm.aliasing_ppgtt,
obj, cache_level);
+
+ obj->gtt_space->color = cache_level;
}
if (cache_level == I915_CACHE_NONE) {
}
obj->cache_level = cache_level;
+ i915_gem_verify_gtt(dev);
return 0;
}
+int i915_gem_get_caching_ioctl(struct drm_device *dev, void *data,
+ struct drm_file *file)
+{
+ struct drm_i915_gem_caching *args = data;
+ struct drm_i915_gem_object *obj;
+ int ret;
+
+ ret = i915_mutex_lock_interruptible(dev);
+ if (ret)
+ return ret;
+
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
+ if (&obj->base == NULL) {
+ ret = -ENOENT;
+ goto unlock;
+ }
+
+ args->caching = obj->cache_level != I915_CACHE_NONE;
+
+ drm_gem_object_unreference(&obj->base);
+unlock:
+ mutex_unlock(&dev->struct_mutex);
+ return ret;
+}
+
+int i915_gem_set_caching_ioctl(struct drm_device *dev, void *data,
+ struct drm_file *file)
+{
+ struct drm_i915_gem_caching *args = data;
+ struct drm_i915_gem_object *obj;
+ enum i915_cache_level level;
+ int ret;
+
+ ret = i915_mutex_lock_interruptible(dev);
+ if (ret)
+ return ret;
+
+ switch (args->caching) {
+ case I915_CACHING_NONE:
+ level = I915_CACHE_NONE;
+ break;
+ case I915_CACHING_CACHED:
+ level = I915_CACHE_LLC;
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ obj = to_intel_bo(drm_gem_object_lookup(dev, file, args->handle));
+ if (&obj->base == NULL) {
+ ret = -ENOENT;
+ goto unlock;
+ }
+
+ ret = i915_gem_object_set_cache_level(obj, level);
+
+ drm_gem_object_unreference(&obj->base);
+unlock:
+ mutex_unlock(&dev->struct_mutex);
+ return ret;
+}
+
/*
* Prepare buffer for display plane (scanout, cursors, etc).
* Can be called from an uninterruptible phase (modesetting) and allows
u32 old_read_domains, old_write_domain;
int ret;
- ret = i915_gem_object_flush_gpu_write_domain(obj);
- if (ret)
- return ret;
-
if (pipelined != obj->ring) {
ret = i915_gem_object_sync(obj, pipelined);
if (ret)
* (e.g. libkms for the bootup splash), we have to ensure that we
* always use map_and_fenceable for all scanout buffers.
*/
- ret = i915_gem_object_pin(obj, alignment, true);
+ ret = i915_gem_object_pin(obj, alignment, true, false);
if (ret)
return ret;
/* It should now be out of any other write domains, and we can update
* the domain values for our changes.
*/
- BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_GTT) != 0);
+ obj->base.write_domain = 0;
obj->base.read_domains |= I915_GEM_DOMAIN_GTT;
trace_i915_gem_object_change_domain(obj,
if ((obj->base.read_domains & I915_GEM_GPU_DOMAINS) == 0)
return 0;
- if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) {
- ret = i915_gem_flush_ring(obj->ring, 0, obj->base.write_domain);
- if (ret)
- return ret;
- }
-
- ret = i915_gem_object_wait_rendering(obj);
+ ret = i915_gem_object_wait_rendering(obj, false);
if (ret)
return ret;
if (obj->base.write_domain == I915_GEM_DOMAIN_CPU)
return 0;
- ret = i915_gem_object_flush_gpu_write_domain(obj);
+ ret = i915_gem_object_wait_rendering(obj, !write);
if (ret)
return ret;
- if (write || obj->pending_gpu_write) {
- ret = i915_gem_object_wait_rendering(obj);
- if (ret)
- return ret;
- }
-
i915_gem_object_flush_gtt_write_domain(obj);
old_write_domain = obj->base.write_domain;
int
i915_gem_object_pin(struct drm_i915_gem_object *obj,
uint32_t alignment,
- bool map_and_fenceable)
+ bool map_and_fenceable,
+ bool nonblocking)
{
int ret;
if (obj->gtt_space == NULL) {
ret = i915_gem_object_bind_to_gtt(obj, alignment,
- map_and_fenceable);
+ map_and_fenceable,
+ nonblocking);
if (ret)
return ret;
}
obj->user_pin_count++;
obj->pin_filp = file;
if (obj->user_pin_count == 1) {
- ret = i915_gem_object_pin(obj, args->alignment, true);
+ ret = i915_gem_object_pin(obj, args->alignment, true, false);
if (ret)
goto out;
}
ret = i915_gem_object_flush_active(obj);
args->busy = obj->active;
+ if (obj->ring) {
+ BUILD_BUG_ON(I915_NUM_RINGS > 16);
+ args->busy |= intel_ring_flag(obj->ring) << 16;
+ }
drm_gem_object_unreference(&obj->base);
unlock:
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) &&
- obj->gtt_space == NULL)
+ /* if the object is no longer attached, discard its backing storage */
+ if (i915_gem_object_is_purgeable(obj) && obj->pages == NULL)
i915_gem_object_truncate(obj);
args->retained = obj->madv != __I915_MADV_PURGED;
return ret;
}
+void i915_gem_object_init(struct drm_i915_gem_object *obj,
+ const struct drm_i915_gem_object_ops *ops)
+{
+ INIT_LIST_HEAD(&obj->mm_list);
+ INIT_LIST_HEAD(&obj->gtt_list);
+ INIT_LIST_HEAD(&obj->ring_list);
+ INIT_LIST_HEAD(&obj->exec_list);
+
+ obj->ops = ops;
+
+ obj->fence_reg = I915_FENCE_REG_NONE;
+ obj->madv = I915_MADV_WILLNEED;
+ /* Avoid an unnecessary call to unbind on the first bind. */
+ obj->map_and_fenceable = true;
+
+ i915_gem_info_add_obj(obj->base.dev->dev_private, obj->base.size);
+}
+
+static const struct drm_i915_gem_object_ops i915_gem_object_ops = {
+ .get_pages = i915_gem_object_get_pages_gtt,
+ .put_pages = i915_gem_object_put_pages_gtt,
+};
+
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;
struct address_space *mapping;
u32 mask;
mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping;
mapping_set_gfp_mask(mapping, mask);
- i915_gem_info_add_obj(dev_priv, size);
+ i915_gem_object_init(obj, &i915_gem_object_ops);
obj->base.write_domain = I915_GEM_DOMAIN_CPU;
obj->base.read_domains = I915_GEM_DOMAIN_CPU;
} else
obj->cache_level = I915_CACHE_NONE;
- 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->exec_list);
- INIT_LIST_HEAD(&obj->gpu_write_list);
- obj->madv = I915_MADV_WILLNEED;
- /* Avoid an unnecessary call to unbind on the first bind. */
- obj->map_and_fenceable = true;
-
return obj;
}
trace_i915_gem_object_destroy(obj);
- if (gem_obj->import_attach)
- drm_prime_gem_destroy(gem_obj, obj->sg_table);
-
if (obj->phys_obj)
i915_gem_detach_phys_object(dev, obj);
dev_priv->mm.interruptible = was_interruptible;
}
- if (obj->base.map_list.map)
- drm_gem_free_mmap_offset(&obj->base);
+ obj->pages_pin_count = 0;
+ i915_gem_object_put_pages(obj);
+ i915_gem_object_free_mmap_offset(obj);
+
+ BUG_ON(obj->pages);
+
+ if (obj->base.import_attach)
+ drm_prime_gem_destroy(&obj->base, NULL);
drm_gem_object_release(&obj->base);
i915_gem_info_remove_obj(dev_priv, obj->base.size);
/* Under UMS, be paranoid and evict. */
if (!drm_core_check_feature(dev, DRIVER_MODESET))
- i915_gem_evict_everything(dev, false);
+ i915_gem_evict_everything(dev);
i915_gem_reset_fences(dev);
}
BUG_ON(!list_empty(&dev_priv->mm.active_list));
- BUG_ON(!list_empty(&dev_priv->mm.flushing_list));
BUG_ON(!list_empty(&dev_priv->mm.inactive_list));
mutex_unlock(&dev->struct_mutex);
{
INIT_LIST_HEAD(&ring->active_list);
INIT_LIST_HEAD(&ring->request_list);
- INIT_LIST_HEAD(&ring->gpu_write_list);
}
void
drm_i915_private_t *dev_priv = dev->dev_private;
INIT_LIST_HEAD(&dev_priv->mm.active_list);
- INIT_LIST_HEAD(&dev_priv->mm.flushing_list);
INIT_LIST_HEAD(&dev_priv->mm.inactive_list);
+ INIT_LIST_HEAD(&dev_priv->mm.unbound_list);
+ INIT_LIST_HEAD(&dev_priv->mm.bound_list);
INIT_LIST_HEAD(&dev_priv->mm.fence_list);
- INIT_LIST_HEAD(&dev_priv->mm.gtt_list);
for (i = 0; i < I915_NUM_RINGS; i++)
init_ring_lists(&dev_priv->ring[i]);
for (i = 0; i < I915_MAX_NUM_FENCES; i++)
spin_unlock(&file_priv->mm.lock);
}
-static int
-i915_gpu_is_active(struct drm_device *dev)
-{
- drm_i915_private_t *dev_priv = dev->dev_private;
- int lists_empty;
-
- lists_empty = list_empty(&dev_priv->mm.flushing_list) &&
- list_empty(&dev_priv->mm.active_list);
-
- return !lists_empty;
-}
-
static int
i915_gem_inactive_shrink(struct shrinker *shrinker, struct shrink_control *sc)
{
struct drm_i915_private,
mm.inactive_shrinker);
struct drm_device *dev = dev_priv->dev;
- struct drm_i915_gem_object *obj, *next;
+ struct drm_i915_gem_object *obj;
int nr_to_scan = sc->nr_to_scan;
int cnt;
if (!mutex_trylock(&dev->struct_mutex))
return 0;
- /* "fast-path" to count number of available objects */
- if (nr_to_scan == 0) {
- cnt = 0;
- list_for_each_entry(obj,
- &dev_priv->mm.inactive_list,
- mm_list)
- cnt++;
- mutex_unlock(&dev->struct_mutex);
- return cnt / 100 * sysctl_vfs_cache_pressure;
- }
-
-rescan:
- /* first scan for clean buffers */
- i915_gem_retire_requests(dev);
-
- list_for_each_entry_safe(obj, next,
- &dev_priv->mm.inactive_list,
- mm_list) {
- if (i915_gem_object_is_purgeable(obj)) {
- if (i915_gem_object_unbind(obj) == 0 &&
- --nr_to_scan == 0)
- break;
- }
+ if (nr_to_scan) {
+ nr_to_scan -= i915_gem_purge(dev_priv, nr_to_scan);
+ if (nr_to_scan > 0)
+ i915_gem_shrink_all(dev_priv);
}
- /* second pass, evict/count anything still on the inactive list */
cnt = 0;
- list_for_each_entry_safe(obj, next,
- &dev_priv->mm.inactive_list,
- mm_list) {
- if (nr_to_scan &&
- i915_gem_object_unbind(obj) == 0)
- nr_to_scan--;
- else
- cnt++;
- }
+ list_for_each_entry(obj, &dev_priv->mm.unbound_list, gtt_list)
+ if (obj->pages_pin_count == 0)
+ cnt += obj->base.size >> PAGE_SHIFT;
+ list_for_each_entry(obj, &dev_priv->mm.bound_list, gtt_list)
+ if (obj->pin_count == 0 && obj->pages_pin_count == 0)
+ cnt += obj->base.size >> PAGE_SHIFT;
- if (nr_to_scan && i915_gpu_is_active(dev)) {
- /*
- * We are desperate for pages, so as a last resort, wait
- * for the GPU to finish and discard whatever we can.
- * This has a dramatic impact to reduce the number of
- * OOM-killer events whilst running the GPU aggressively.
- */
- if (i915_gpu_idle(dev) == 0)
- goto rescan;
- }
mutex_unlock(&dev->struct_mutex);
- return cnt / 100 * sysctl_vfs_cache_pressure;
+ return cnt;
}
projects / linux-2.6-microblaze.git / blobdiff