korg/linux
0
0
Fork 0
mirror of https://mirrors.bfsu.edu.cn/git/linux.git synced 2024-11-18 09:44:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

drm/i915: Pull i915_vma_pin under the vm->mutex

Browse Source 
Replace the struct_mutex requirement for pinning the i915_vma with the
local vm->mutex instead. Note that the vm->mutex is tainted by the
shrinker (we require unbinding from inside fs-reclaim) and so we cannot
allocate while holding that mutex. Instead we have to preallocate
workers to do allocate and apply the PTE updates after we have we
reserved their slot in the drm_mm (using fences to order the PTE writes
with the GPU work and with later unbind).

In adding the asynchronous vma binding, one subtle requirement is to
avoid coupling the binding fence into the backing object->resv. That is
the asynchronous binding only applies to the vma timeline itself and not
to the pages as that is a more global timeline (the binding of one vma
does not need to be ordered with another vma, nor does the implicit GEM
fencing depend on a vma, only on writes to the backing store). Keeping
the vma binding distinct from the backing store timelines is verified by
a number of async gem_exec_fence and gem_exec_schedule tests. The way we
do this is quite simple, we keep the fence for the vma binding separate
and only wait on it as required, and never add it to the obj->resv
itself.

Another consequence in reducing the locking around the vma is the
destruction of the vma is no longer globally serialised by struct_mutex.
A natural solution would be to add a kref to i915_vma, but that requires
decoupling the reference cycles, possibly by introducing a new
i915_mm_pages object that is own by both obj->mm and vma->pages.
However, we have not taken that route due to the overshadowing lmem/ttm
discussions, and instead play a series of complicated games with
trylocks to (hopefully) ensure that only one destruction path is called!

v2: Add some commentary, and some helpers to reduce patch churn.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191004134015.13204-4-chris@chris-wilson.co.uk
This commit is contained in:
Chris Wilson 2019-10-04 14:39:58 +01:00
parent 11331125e1
commit 2850748ef8
40 changed files with 824 additions and 706 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
drivers/gpu/drm/i915/display/intel_display.c
View File

@ -2079,7 +2079,6 @@ intel_pin_and_fence_fb_obj(struct drm_framebuffer *fb,
unsigned int pinctl;
u32 alignment;
WARN_ON(!mutex_is_locked(&dev->struct_mutex));
if (WARN_ON(!i915_gem_object_is_framebuffer(obj)))
return ERR_PTR(-EINVAL);
@ -2163,8 +2162,6 @@ err:
void intel_unpin_fb_vma(struct i915_vma *vma, unsigned long flags)
{
lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);
i915_gem_object_lock(vma->obj);
if (flags & PLANE_HAS_FENCE)
i915_vma_unpin_fence(vma);
@ -3065,12 +3062,10 @@ intel_alloc_initial_plane_obj(struct intel_crtc *crtc,
return false;
}
mutex_lock(&dev->struct_mutex);
obj = i915_gem_object_create_stolen_for_preallocated(dev_priv,
base_aligned,
base_aligned,
size_aligned);
mutex_unlock(&dev->struct_mutex);
if (!obj)
return false;
@ -3232,13 +3227,11 @@ valid_fb:
intel_state->color_plane[0].stride =
intel_fb_pitch(fb, 0, intel_state->base.rotation);
mutex_lock(&dev->struct_mutex);
intel_state->vma =
intel_pin_and_fence_fb_obj(fb,
&intel_state->view,
intel_plane_uses_fence(intel_state),
&intel_state->flags);
mutex_unlock(&dev->struct_mutex);
if (IS_ERR(intel_state->vma)) {
DRM_ERROR("failed to pin boot fb on pipe %d: %li\n",
intel_crtc->pipe, PTR_ERR(intel_state->vma));
@ -14365,8 +14358,6 @@ static void fb_obj_bump_render_priority(struct drm_i915_gem_object *obj)
* bits. Some older platforms need special physical address handling for
* cursor planes.
*
* Must be called with struct_mutex held.
*
* Returns 0 on success, negative error code on failure.
*/
int
@ -14423,15 +14414,8 @@ intel_prepare_plane_fb(struct drm_plane *plane,
if (ret)
return ret;
ret = mutex_lock_interruptible(&dev_priv->drm.struct_mutex);
if (ret) {
i915_gem_object_unpin_pages(obj);
return ret;
}
ret = intel_plane_pin_fb(to_intel_plane_state(new_state));
mutex_unlock(&dev_priv->drm.struct_mutex);
i915_gem_object_unpin_pages(obj);
if (ret)
return ret;
@ -14480,8 +14464,6 @@ intel_prepare_plane_fb(struct drm_plane *plane,
* @old_state: the state from the previous modeset
*
* Cleans up a framebuffer that has just been removed from a plane.
*
* Must be called with struct_mutex held.
*/
void
intel_cleanup_plane_fb(struct drm_plane *plane,
@ -14497,9 +14479,7 @@ intel_cleanup_plane_fb(struct drm_plane *plane,
}
/* Should only be called after a successful intel_prepare_plane_fb()! */
mutex_lock(&dev_priv->drm.struct_mutex);
intel_plane_unpin_fb(to_intel_plane_state(old_state));
mutex_unlock(&dev_priv->drm.struct_mutex);
}
int
@ -14702,7 +14682,6 @@ intel_legacy_cursor_update(struct drm_plane *plane,
u32 src_w, u32 src_h,
struct drm_modeset_acquire_ctx *ctx)
{
struct drm_i915_private *dev_priv = to_i915(crtc->dev);
struct drm_plane_state *old_plane_state, *new_plane_state;
struct intel_plane *intel_plane = to_intel_plane(plane);
struct intel_crtc_state *crtc_state =
@ -14768,13 +14747,9 @@ intel_legacy_cursor_update(struct drm_plane *plane,
if (ret)
goto out_free;
ret = mutex_lock_interruptible(&dev_priv->drm.struct_mutex);
if (ret)
goto out_free;
ret = intel_plane_pin_fb(to_intel_plane_state(new_plane_state));
if (ret)
goto out_unlock;
goto out_free;
intel_frontbuffer_flush(to_intel_frontbuffer(fb), ORIGIN_FLIP);
intel_frontbuffer_track(to_intel_frontbuffer(old_plane_state->fb),
@ -14804,8 +14779,6 @@ intel_legacy_cursor_update(struct drm_plane *plane,
intel_plane_unpin_fb(to_intel_plane_state(old_plane_state));
out_unlock:
mutex_unlock(&dev_priv->drm.struct_mutex);
out_free:
if (new_crtc_state)
intel_crtc_destroy_state(crtc, &new_crtc_state->base);

7
drivers/gpu/drm/i915/display/intel_dsb.c
View File

@ -119,9 +119,7 @@ intel_dsb_get(struct intel_crtc *crtc)
goto err;
}
mutex_lock(&i915->drm.struct_mutex);
vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, PIN_MAPPABLE);
mutex_unlock(&i915->drm.struct_mutex);
if (IS_ERR(vma)) {
DRM_ERROR("Vma creation failed\n");
i915_gem_object_put(obj);
@ -164,10 +162,7 @@ void intel_dsb_put(struct intel_dsb *dsb)
return;
if (atomic_dec_and_test(&dsb->refcount)) {
mutex_lock(&i915->drm.struct_mutex);
i915_gem_object_unpin_map(dsb->vma->obj);
i915_vma_unpin_and_release(&dsb->vma, 0);
mutex_unlock(&i915->drm.struct_mutex);
i915_vma_unpin_and_release(&dsb->vma, I915_VMA_RELEASE_MAP);
dsb->cmd_buf = NULL;
dsb->free_pos = 0;
dsb->ins_start_offset = 0;

8
drivers/gpu/drm/i915/display/intel_fbdev.c
View File

@ -204,7 +204,6 @@ static int intelfb_create(struct drm_fb_helper *helper,
sizes->fb_height = intel_fb->base.height;
}
mutex_lock(&dev->struct_mutex);
wakeref = intel_runtime_pm_get(&dev_priv->runtime_pm);
/* Pin the GGTT vma for our access via info->screen_base.
@ -266,7 +265,6 @@ static int intelfb_create(struct drm_fb_helper *helper,
ifbdev->vma_flags = flags;
intel_runtime_pm_put(&dev_priv->runtime_pm, wakeref);
mutex_unlock(&dev->struct_mutex);
vga_switcheroo_client_fb_set(pdev, info);
return 0;
@ -274,7 +272,6 @@ out_unpin:
intel_unpin_fb_vma(vma, flags);
out_unlock:
intel_runtime_pm_put(&dev_priv->runtime_pm, wakeref);
mutex_unlock(&dev->struct_mutex);
return ret;
}
@ -291,11 +288,8 @@ static void intel_fbdev_destroy(struct intel_fbdev *ifbdev)
drm_fb_helper_fini(&ifbdev->helper);
if (ifbdev->vma) {
mutex_lock(&ifbdev->helper.dev->struct_mutex);
if (ifbdev->vma)
intel_unpin_fb_vma(ifbdev->vma, ifbdev->vma_flags);
mutex_unlock(&ifbdev->helper.dev->struct_mutex);
}
if (ifbdev->fb)
drm_framebuffer_remove(&ifbdev->fb->base);

11
drivers/gpu/drm/i915/display/intel_overlay.c
View File

@ -1303,15 +1303,11 @@ static int get_registers(struct intel_overlay *overlay, bool use_phys)
struct i915_vma *vma;
int err;
mutex_lock(&i915->drm.struct_mutex);
obj = i915_gem_object_create_stolen(i915, PAGE_SIZE);
if (obj == NULL)
obj = i915_gem_object_create_internal(i915, PAGE_SIZE);
if (IS_ERR(obj)) {
err = PTR_ERR(obj);
goto err_unlock;
}
if (IS_ERR(obj))
return PTR_ERR(obj);
vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, PIN_MAPPABLE);
if (IS_ERR(vma)) {
@ -1332,13 +1328,10 @@ static int get_registers(struct intel_overlay *overlay, bool use_phys)
}
overlay->reg_bo = obj;
mutex_unlock(&i915->drm.struct_mutex);
return 0;
err_put_bo:
i915_gem_object_put(obj);
err_unlock:
mutex_unlock(&i915->drm.struct_mutex);
return err;
}

2
drivers/gpu/drm/i915/gem/i915_gem_client_blt.c
View File

@ -211,7 +211,7 @@ static void clear_pages_worker(struct work_struct *work)
* keep track of the GPU activity within this vma/request, and
* propagate the signal from the request to w->dma.
*/
err = i915_active_add_request(&vma->active, rq);
err = __i915_vma_move_to_active(vma, rq);
if (err)
goto out_request;

20
drivers/gpu/drm/i915/gem/i915_gem_context.c
View File

@ -313,8 +313,6 @@ static void i915_gem_context_free(struct i915_gem_context *ctx)
GEM_BUG_ON(!i915_gem_context_is_closed(ctx));
release_hw_id(ctx);
if (ctx->vm)
i915_vm_put(ctx->vm);
free_engines(rcu_access_pointer(ctx->engines));
mutex_destroy(&ctx->engines_mutex);
@ -379,9 +377,13 @@ void i915_gem_context_release(struct kref *ref)
static void context_close(struct i915_gem_context *ctx)
{
i915_gem_context_set_closed(ctx);
if (ctx->vm)
i915_vm_close(ctx->vm);
mutex_lock(&ctx->mutex);
i915_gem_context_set_closed(ctx);
ctx->file_priv = ERR_PTR(-EBADF);
/*
@ -474,7 +476,7 @@ __set_ppgtt(struct i915_gem_context *ctx, struct i915_address_space *vm)
GEM_BUG_ON(old && i915_vm_is_4lvl(vm) != i915_vm_is_4lvl(old));
ctx->vm = i915_vm_get(vm);
ctx->vm = i915_vm_open(vm);
context_apply_all(ctx, __apply_ppgtt, vm);
return old;
@ -488,7 +490,7 @@ static void __assign_ppgtt(struct i915_gem_context *ctx,
vm = __set_ppgtt(ctx, vm);
if (vm)
i915_vm_put(vm);
i915_vm_close(vm);
}
static void __set_timeline(struct intel_timeline **dst,
@ -953,7 +955,7 @@ static int get_ppgtt(struct drm_i915_file_private *file_priv,
if (ret < 0)
goto err_unlock;
i915_vm_get(vm);
i915_vm_open(vm);
args->size = 0;
args->value = ret;
@ -973,7 +975,7 @@ static void set_ppgtt_barrier(void *data)
if (INTEL_GEN(old->i915) < 8)
gen6_ppgtt_unpin_all(i915_vm_to_ppgtt(old));
i915_vm_put(old);
i915_vm_close(old);
}
static int emit_ppgtt_update(struct i915_request *rq, void *data)
@ -1090,8 +1092,8 @@ static int set_ppgtt(struct drm_i915_file_private *file_priv,
set_ppgtt_barrier,
old);
if (err) {
i915_vm_put(__set_ppgtt(ctx, old));
i915_vm_put(old);
i915_vm_close(__set_ppgtt(ctx, old));
i915_vm_close(old);
}
unlock:

13
drivers/gpu/drm/i915/gem/i915_gem_domain.c
View File

@ -288,7 +288,12 @@ restart:
if (!drm_mm_node_allocated(&vma->node))
continue;
ret = i915_vma_bind(vma, cache_level, PIN_UPDATE);
/* Wait for an earlier async bind, need to rewrite it */
ret = i915_vma_sync(vma);
if (ret)
return ret;
ret = i915_vma_bind(vma, cache_level, PIN_UPDATE, NULL);
if (ret)
return ret;
}
@ -391,16 +396,11 @@ int i915_gem_set_caching_ioctl(struct drm_device *dev, void *data,
if (ret)
goto out;
ret = mutex_lock_interruptible(&i915->drm.struct_mutex);
if (ret)
goto out;
ret = i915_gem_object_lock_interruptible(obj);
if (ret == 0) {
ret = i915_gem_object_set_cache_level(obj, level);
i915_gem_object_unlock(obj);
}
mutex_unlock(&i915->drm.struct_mutex);
out:
i915_gem_object_put(obj);
@ -485,6 +485,7 @@ static void i915_gem_object_bump_inactive_ggtt(struct drm_i915_gem_object *obj)
if (!drm_mm_node_allocated(&vma->node))
continue;
GEM_BUG_ON(vma->vm != &i915->ggtt.vm);
list_move_tail(&vma->vm_link, &vma->vm->bound_list);
}
mutex_unlock(&i915->ggtt.vm.mutex);

43
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
View File

@ -698,7 +698,9 @@ static int eb_reserve(struct i915_execbuffer *eb)
case 1:
/* Too fragmented, unbind everything and retry */
mutex_lock(&eb->context->vm->mutex);
err = i915_gem_evict_vm(eb->context->vm);
mutex_unlock(&eb->context->vm->mutex);
if (err)
return err;
break;
@ -972,7 +974,9 @@ static void reloc_cache_reset(struct reloc_cache *cache)
ggtt->vm.clear_range(&ggtt->vm,
cache->node.start,
cache->node.size);
mutex_lock(&ggtt->vm.mutex);
drm_mm_remove_node(&cache->node);
mutex_unlock(&ggtt->vm.mutex);
} else {
i915_vma_unpin((struct i915_vma *)cache->node.mm);
}
@ -1047,11 +1051,13 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj,
PIN_NOEVICT);
if (IS_ERR(vma)) {
memset(&cache->node, 0, sizeof(cache->node));
mutex_lock(&ggtt->vm.mutex);
err = drm_mm_insert_node_in_range
(&ggtt->vm.mm, &cache->node,
PAGE_SIZE, 0, I915_COLOR_UNEVICTABLE,
0, ggtt->mappable_end,
DRM_MM_INSERT_LOW);
mutex_unlock(&ggtt->vm.mutex);
if (err) /* no inactive aperture space, use cpu reloc */
return NULL;
} else {
@ -1416,7 +1422,7 @@ eb_relocate_entry(struct i915_execbuffer *eb,
if (reloc->write_domain == I915_GEM_DOMAIN_INSTRUCTION &&
IS_GEN(eb->i915, 6)) {
err = i915_vma_bind(target, target->obj->cache_level,
PIN_GLOBAL);
PIN_GLOBAL, NULL);
if (WARN_ONCE(err,
"Unexpected failure to bind target VMA!"))
return err;
@ -2142,35 +2148,6 @@ static struct i915_request *eb_throttle(struct intel_context *ce)
return i915_request_get(rq);
}
static int
__eb_pin_context(struct i915_execbuffer *eb, struct intel_context *ce)
{
int err;
if (likely(atomic_inc_not_zero(&ce->pin_count)))
return 0;
err = mutex_lock_interruptible(&eb->i915->drm.struct_mutex);
if (err)
return err;
err = __intel_context_do_pin(ce);
mutex_unlock(&eb->i915->drm.struct_mutex);
return err;
}
static void
__eb_unpin_context(struct i915_execbuffer *eb, struct intel_context *ce)
{
if (likely(atomic_add_unless(&ce->pin_count, -1, 1)))
return;
mutex_lock(&eb->i915->drm.struct_mutex);
intel_context_unpin(ce);
mutex_unlock(&eb->i915->drm.struct_mutex);
}
static int __eb_pin_engine(struct i915_execbuffer *eb, struct intel_context *ce)
{
struct intel_timeline *tl;
@ -2190,7 +2167,7 @@ static int __eb_pin_engine(struct i915_execbuffer *eb, struct intel_context *ce)
* GGTT space, so do this first before we reserve a seqno for
* ourselves.
*/
err = __eb_pin_context(eb, ce);
err = intel_context_pin(ce);
if (err)
return err;
@ -2234,7 +2211,7 @@ err_exit:
intel_context_exit(ce);
intel_context_timeline_unlock(tl);
err_unpin:
__eb_unpin_context(eb, ce);
intel_context_unpin(ce);
return err;
}
@ -2247,7 +2224,7 @@ static void eb_unpin_engine(struct i915_execbuffer *eb)
intel_context_exit(ce);
mutex_unlock(&tl->mutex);
__eb_unpin_context(eb, ce);
intel_context_unpin(ce);
}
static unsigned int

20
drivers/gpu/drm/i915/gem/i915_gem_mman.c
View File

@ -249,16 +249,6 @@ vm_fault_t i915_gem_fault(struct vm_fault *vmf)
if (ret)
goto err_rpm;
ret = i915_mutex_lock_interruptible(dev);
if (ret)
goto err_reset;
/* Access to snoopable pages through the GTT is incoherent. */
if (obj->cache_level != I915_CACHE_NONE && !HAS_LLC(i915)) {
ret = -EFAULT;
goto err_unlock;
}
/* Now pin it into the GTT as needed */
vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0,
PIN_MAPPABLE |
@ -291,7 +281,13 @@ vm_fault_t i915_gem_fault(struct vm_fault *vmf)
}
if (IS_ERR(vma)) {
ret = PTR_ERR(vma);
goto err_unlock;
goto err_reset;
}
/* Access to snoopable pages through the GTT is incoherent. */
if (obj->cache_level != I915_CACHE_NONE && !HAS_LLC(i915)) {
ret = -EFAULT;
goto err_unpin;
}
ret = i915_vma_pin_fence(vma);
@ -329,8 +325,6 @@ err_fence:
i915_vma_unpin_fence(vma);
err_unpin:
__i915_vma_unpin(vma);
err_unlock:
mutex_unlock(&dev->struct_mutex);
err_reset:
intel_gt_reset_unlock(ggtt->vm.gt, srcu);
err_rpm:

31
drivers/gpu/drm/i915/gem/i915_gem_object.c
View File

@ -155,21 +155,30 @@ static void __i915_gem_free_objects(struct drm_i915_private *i915,
wakeref = intel_runtime_pm_get(&i915->runtime_pm);
llist_for_each_entry_safe(obj, on, freed, freed) {
struct i915_vma *vma, *vn;
trace_i915_gem_object_destroy(obj);
mutex_lock(&i915->drm.struct_mutex);
if (!list_empty(&obj->vma.list)) {
struct i915_vma *vma;
list_for_each_entry_safe(vma, vn, &obj->vma.list, obj_link) {
GEM_BUG_ON(i915_vma_is_active(vma));
atomic_and(~I915_VMA_PIN_MASK, &vma->flags);
i915_vma_destroy(vma);
/*
* Note that the vma keeps an object reference while
* it is active, so it *should* not sleep while we
* destroy it. Our debug code errs insits it *might*.
* For the moment, play along.
*/
spin_lock(&obj->vma.lock);
while ((vma = list_first_entry_or_null(&obj->vma.list,
struct i915_vma,
obj_link))) {
GEM_BUG_ON(vma->obj != obj);
spin_unlock(&obj->vma.lock);
i915_vma_destroy(vma);
spin_lock(&obj->vma.lock);
}
spin_unlock(&obj->vma.lock);
}
GEM_BUG_ON(!list_empty(&obj->vma.list));
GEM_BUG_ON(!RB_EMPTY_ROOT(&obj->vma.tree));
mutex_unlock(&i915->drm.struct_mutex);
GEM_BUG_ON(atomic_read(&obj->bind_count));
GEM_BUG_ON(obj->userfault_count);

5
drivers/gpu/drm/i915/gem/i915_gem_object.h
View File

@ -106,6 +106,11 @@ static inline void i915_gem_object_lock(struct drm_i915_gem_object *obj)
dma_resv_lock(obj->base.resv, NULL);
}
static inline bool i915_gem_object_trylock(struct drm_i915_gem_object *obj)
{
return dma_resv_trylock(obj->base.resv);
}
static inline int
i915_gem_object_lock_interruptible(struct drm_i915_gem_object *obj)
{

73
drivers/gpu/drm/i915/gem/i915_gem_shrinker.c
View File

@ -16,40 +16,6 @@
#include "i915_trace.h"
static bool shrinker_lock(struct drm_i915_private *i915,
unsigned int flags,
bool *unlock)
{
struct mutex *m = &i915->drm.struct_mutex;
switch (mutex_trylock_recursive(m)) {
case MUTEX_TRYLOCK_RECURSIVE:
*unlock = false;
return true;
case MUTEX_TRYLOCK_FAILED:
*unlock = false;
if (flags & I915_SHRINK_ACTIVE &&
mutex_lock_killable_nested(m, I915_MM_SHRINKER) == 0)
*unlock = true;
return *unlock;
case MUTEX_TRYLOCK_SUCCESS:
*unlock = true;
return true;
}
BUG();
}
static void shrinker_unlock(struct drm_i915_private *i915, bool unlock)
{
if (!unlock)
return;
mutex_unlock(&i915->drm.struct_mutex);
}
static bool swap_available(void)
{
return get_nr_swap_pages() > 0;
@ -155,10 +121,6 @@ i915_gem_shrink(struct drm_i915_private *i915,
intel_wakeref_t wakeref = 0;
unsigned long count = 0;
unsigned long scanned = 0;
bool unlock;
if (!shrinker_lock(i915, shrink, &unlock))
return 0;
/*
* When shrinking the active list, we should also consider active
@ -268,8 +230,6 @@ i915_gem_shrink(struct drm_i915_private *i915,
if (shrink & I915_SHRINK_BOUND)
intel_runtime_pm_put(&i915->runtime_pm, wakeref);
shrinker_unlock(i915, unlock);
if (nr_scanned)
*nr_scanned += scanned;
return count;
@ -339,19 +299,14 @@ i915_gem_shrinker_scan(struct shrinker *shrinker, struct shrink_control *sc)
struct drm_i915_private *i915 =
container_of(shrinker, struct drm_i915_private, mm.shrinker);
unsigned long freed;
bool unlock;
sc->nr_scanned = 0;
if (!shrinker_lock(i915, 0, &unlock))
return SHRINK_STOP;
freed = i915_gem_shrink(i915,
sc->nr_to_scan,
&sc->nr_scanned,
I915_SHRINK_BOUND |
I915_SHRINK_UNBOUND |
I915_SHRINK_WRITEBACK);
I915_SHRINK_UNBOUND);
if (sc->nr_scanned < sc->nr_to_scan && current_is_kswapd()) {
intel_wakeref_t wakeref;
@ -366,8 +321,6 @@ i915_gem_shrinker_scan(struct shrinker *shrinker, struct shrink_control *sc)
}
}
shrinker_unlock(i915, unlock);
return sc->nr_scanned ? freed : SHRINK_STOP;
}
@ -384,6 +337,7 @@ i915_gem_shrinker_oom(struct notifier_block *nb, unsigned long event, void *ptr)
freed_pages = 0;
with_intel_runtime_pm(&i915->runtime_pm, wakeref)
freed_pages += i915_gem_shrink(i915, -1UL, NULL,
I915_SHRINK_ACTIVE |
I915_SHRINK_BOUND |
I915_SHRINK_UNBOUND |
I915_SHRINK_WRITEBACK);
@ -419,10 +373,6 @@ i915_gem_shrinker_vmap(struct notifier_block *nb, unsigned long event, void *ptr
struct i915_vma *vma, *next;
unsigned long freed_pages = 0;
intel_wakeref_t wakeref;
bool unlock;
if (!shrinker_lock(i915, 0, &unlock))
return NOTIFY_DONE;
with_intel_runtime_pm(&i915->runtime_pm, wakeref)
freed_pages += i915_gem_shrink(i915, -1UL, NULL,
@ -439,15 +389,11 @@ i915_gem_shrinker_vmap(struct notifier_block *nb, unsigned long event, void *ptr
if (!vma->iomap || i915_vma_is_active(vma))
continue;
mutex_unlock(&i915->ggtt.vm.mutex);
if (i915_vma_unbind(vma) == 0)
if (__i915_vma_unbind(vma) == 0)
freed_pages += count;
mutex_lock(&i915->ggtt.vm.mutex);
}
mutex_unlock(&i915->ggtt.vm.mutex);
shrinker_unlock(i915, unlock);
*(unsigned long *)ptr += freed_pages;
return NOTIFY_DONE;
}
@ -490,22 +436,9 @@ void i915_gem_shrinker_taints_mutex(struct drm_i915_private *i915,
fs_reclaim_acquire(GFP_KERNEL);
/*
* As we invariably rely on the struct_mutex within the shrinker,
* but have a complicated recursion dance, taint all the mutexes used
* within the shrinker with the struct_mutex. For completeness, we
* taint with all subclass of struct_mutex, even though we should
* only need tainting by I915_MM_NORMAL to catch possible ABBA
* deadlocks from using struct_mutex inside @mutex.
*/
mutex_acquire(&i915->drm.struct_mutex.dep_map,
I915_MM_SHRINKER, 0, _RET_IP_);
mutex_acquire(&mutex->dep_map, 0, 0, _RET_IP_);
mutex_release(&mutex->dep_map, 0, _RET_IP_);
mutex_release(&i915->drm.struct_mutex.dep_map, 0, _RET_IP_);
fs_reclaim_release(GFP_KERNEL);
if (unlock)

8
drivers/gpu/drm/i915/gem/i915_gem_stolen.c
View File

@ -624,8 +624,6 @@ i915_gem_object_create_stolen_for_preallocated(struct drm_i915_private *dev_priv
if (!drm_mm_initialized(&dev_priv->mm.stolen))
return NULL;
lockdep_assert_held(&dev_priv->drm.struct_mutex);
DRM_DEBUG_DRIVER("creating preallocated stolen object: stolen_offset=%pa, gtt_offset=%pa, size=%pa\n",
&stolen_offset, &gtt_offset, &size);
@ -677,21 +675,25 @@ i915_gem_object_create_stolen_for_preallocated(struct drm_i915_private *dev_priv
* setting up the GTT space. The actual reservation will occur
* later.
*/
mutex_lock(&ggtt->vm.mutex);
ret = i915_gem_gtt_reserve(&ggtt->vm, &vma->node,
size, gtt_offset, obj->cache_level,
0);
if (ret) {
DRM_DEBUG_DRIVER("failed to allocate stolen GTT space\n");
mutex_unlock(&ggtt->vm.mutex);
goto err_pages;
}
GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
GEM_BUG_ON(vma->pages);
vma->pages = obj->mm.pages;
atomic_set(&vma->pages_count, I915_VMA_PAGES_ACTIVE);
set_bit(I915_VMA_GLOBAL_BIND_BIT, __i915_vma_flags(vma));
__i915_vma_set_map_and_fenceable(vma);
mutex_lock(&ggtt->vm.mutex);
list_add_tail(&vma->vm_link, &ggtt->vm.bound_list);
mutex_unlock(&ggtt->vm.mutex);

27
drivers/gpu/drm/i915/gem/i915_gem_tiling.c
View File

@ -181,22 +181,25 @@ static int
i915_gem_object_fence_prepare(struct drm_i915_gem_object *obj,
int tiling_mode, unsigned int stride)
{
struct i915_ggtt *ggtt = &to_i915(obj->base.dev)->ggtt;
struct i915_vma *vma;
int ret;
int ret = 0;
if (tiling_mode == I915_TILING_NONE)
return 0;
mutex_lock(&ggtt->vm.mutex);
for_each_ggtt_vma(vma, obj) {
if (i915_vma_fence_prepare(vma, tiling_mode, stride))
continue;
ret = i915_vma_unbind(vma);
ret = __i915_vma_unbind(vma);
if (ret)
return ret;
break;
}
mutex_unlock(&ggtt->vm.mutex);
return 0;
return ret;
}
int
@ -212,7 +215,6 @@ i915_gem_object_set_tiling(struct drm_i915_gem_object *obj,
GEM_BUG_ON(!i915_tiling_ok(obj, tiling, stride));
GEM_BUG_ON(!stride ^ (tiling == I915_TILING_NONE));
lockdep_assert_held(&i915->drm.struct_mutex);
if ((tiling | stride) == obj->tiling_and_stride)
return 0;
@ -233,16 +235,18 @@ i915_gem_object_set_tiling(struct drm_i915_gem_object *obj,
* whilst executing a fenced command for an untiled object.
*/
err = i915_gem_object_fence_prepare(obj, tiling, stride);
if (err)
return err;
i915_gem_object_lock(obj);
if (i915_gem_object_is_framebuffer(obj)) {
i915_gem_object_unlock(obj);
return -EBUSY;
}
err = i915_gem_object_fence_prepare(obj, tiling, stride);
if (err) {
i915_gem_object_unlock(obj);
return err;
}
/* If the memory has unknown (i.e. varying) swizzling, we pin the
* pages to prevent them being swapped out and causing corruption
* due to the change in swizzling.
@ -368,12 +372,7 @@ i915_gem_set_tiling_ioctl(struct drm_device *dev, void *data,
}
}
err = mutex_lock_interruptible(&dev->struct_mutex);
if (err)
goto err;
err = i915_gem_object_set_tiling(obj, args->tiling_mode, args->stride);
mutex_unlock(&dev->struct_mutex);
/* We have to maintain this existing ABI... */
args->stride = i915_gem_object_get_stride(obj);

27
drivers/gpu/drm/i915/gem/i915_gem_userptr.c
View File

@ -92,7 +92,6 @@ userptr_mn_invalidate_range_start(struct mmu_notifier *_mn,
struct i915_mmu_notifier *mn =
container_of(_mn, struct i915_mmu_notifier, mn);
struct interval_tree_node *it;
struct mutex *unlock = NULL;
unsigned long end;
int ret = 0;
@ -129,33 +128,13 @@ userptr_mn_invalidate_range_start(struct mmu_notifier *_mn,
}
spin_unlock(&mn->lock);
if (!unlock) {
unlock = &mn->mm->i915->drm.struct_mutex;
switch (mutex_trylock_recursive(unlock)) {
default:
case MUTEX_TRYLOCK_FAILED:
if (mutex_lock_killable_nested(unlock, I915_MM_SHRINKER)) {
i915_gem_object_put(obj);
return -EINTR;
}
/* fall through */
case MUTEX_TRYLOCK_SUCCESS:
break;
case MUTEX_TRYLOCK_RECURSIVE:
unlock = ERR_PTR(-EEXIST);
break;
}
}
ret = i915_gem_object_unbind(obj,
I915_GEM_OBJECT_UNBIND_ACTIVE);
if (ret == 0)
ret = __i915_gem_object_put_pages(obj, I915_MM_SHRINKER);
i915_gem_object_put(obj);
if (ret)
goto unlock;
return ret;
spin_lock(&mn->lock);
@ -168,10 +147,6 @@ userptr_mn_invalidate_range_start(struct mmu_notifier *_mn,
}
spin_unlock(&mn->lock);
unlock:
if (!IS_ERR_OR_NULL(unlock))
mutex_unlock(unlock);
return ret;
}

9
drivers/gpu/drm/i915/gem/selftests/huge_pages.c
View File

@ -333,7 +333,12 @@ static int igt_check_page_sizes(struct i915_vma *vma)
struct drm_i915_private *i915 = vma->vm->i915;
unsigned int supported = INTEL_INFO(i915)->page_sizes;
struct drm_i915_gem_object *obj = vma->obj;
int err = 0;
int err;
/* We have to wait for the async bind to complete before our asserts */
err = i915_vma_sync(vma);
if (err)
return err;
if (!HAS_PAGE_SIZES(i915, vma->page_sizes.sg)) {
pr_err("unsupported page_sizes.sg=%u, supported=%u\n",
@ -1390,7 +1395,7 @@ static int igt_ppgtt_pin_update(void *arg)
goto out_unpin;
}
err = i915_vma_bind(vma, I915_CACHE_NONE, PIN_UPDATE);
err = i915_vma_bind(vma, I915_CACHE_NONE, PIN_UPDATE, NULL);
if (err)
goto out_unpin;

12
drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c
View File

@ -971,10 +971,7 @@ emit_rpcs_query(struct drm_i915_gem_object *obj,
if (err)
goto skip_request;
i915_vma_unpin(batch);
i915_vma_close(batch);
i915_vma_put(batch);
i915_vma_unpin_and_release(&batch, 0);
i915_vma_unpin(vma);
*rq_out = i915_request_get(rq);
@ -988,8 +985,7 @@ skip_request:
err_request:
i915_request_add(rq);
err_batch:
i915_vma_unpin(batch);
i915_vma_put(batch);
i915_vma_unpin_and_release(&batch, 0);
err_vma:
i915_vma_unpin(vma);
@ -1533,9 +1529,7 @@ static int write_to_scratch(struct i915_gem_context *ctx,
if (err)
goto skip_request;
i915_vma_unpin(vma);
i915_vma_close(vma);
i915_vma_put(vma);
i915_vma_unpin_and_release(&vma, 0);
i915_request_add(rq);

4
drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c
View File

@ -322,7 +322,6 @@ static int igt_partial_tiling(void *arg)
goto out;
}
mutex_lock(&i915->drm.struct_mutex);
wakeref = intel_runtime_pm_get(&i915->runtime_pm);
if (1) {
@ -415,7 +414,6 @@ next_tiling: ;
out_unlock:
intel_runtime_pm_put(&i915->runtime_pm, wakeref);
mutex_unlock(&i915->drm.struct_mutex);
i915_gem_object_unpin_pages(obj);
out:
i915_gem_object_put(obj);
@ -458,7 +456,6 @@ static int igt_smoke_tiling(void *arg)
goto out;
}
mutex_lock(&i915->drm.struct_mutex);
wakeref = intel_runtime_pm_get(&i915->runtime_pm);
count = 0;
@ -508,7 +505,6 @@ static int igt_smoke_tiling(void *arg)
pr_info("%s: Completed %lu trials\n", __func__, count);
intel_runtime_pm_put(&i915->runtime_pm, wakeref);
mutex_unlock(&i915->drm.struct_mutex);
i915_gem_object_unpin_pages(obj);
out:
i915_gem_object_put(obj);

7
drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.c
View File

@ -154,9 +154,7 @@ int igt_gpu_fill_dw(struct intel_context *ce,
i915_request_add(rq);
i915_vma_unpin(batch);
i915_vma_close(batch);
i915_vma_put(batch);
i915_vma_unpin_and_release(&batch, 0);
return 0;
@ -165,7 +163,6 @@ skip_request:
err_request:
i915_request_add(rq);
err_batch:
i915_vma_unpin(batch);
i915_vma_put(batch);
i915_vma_unpin_and_release(&batch, 0);
return err;
}

5
drivers/gpu/drm/i915/gt/intel_gt.c
View File

@ -302,11 +302,12 @@ void intel_gt_flush_ggtt_writes(struct intel_gt *gt)
with_intel_runtime_pm(&i915->runtime_pm, wakeref) {
struct intel_uncore *uncore = gt->uncore;
unsigned long flags;
spin_lock_irq(&uncore->lock);
spin_lock_irqsave(&uncore->lock, flags);
intel_uncore_posting_read_fw(uncore,
RING_HEAD(RENDER_RING_BASE));
spin_unlock_irq(&uncore->lock);
spin_unlock_irqrestore(&uncore->lock, flags);
}
}

4
drivers/gpu/drm/i915/gt/intel_ringbuffer.c
View File

@ -1338,15 +1338,13 @@ void intel_ring_free(struct kref *ref)
{
struct intel_ring *ring = container_of(ref, typeof(*ring), ref);
i915_vma_close(ring->vma);
i915_vma_put(ring->vma);
kfree(ring);
}
static void __ring_context_fini(struct intel_context *ce)
{
i915_gem_object_put(ce->state->obj);
i915_vma_put(ce->state);
}
static void ring_context_destroy(struct kref *ref)

19
drivers/gpu/drm/i915/gt/selftest_hangcheck.c
View File

@ -1127,15 +1127,14 @@ static int evict_vma(void *data)
{
struct evict_vma *arg = data;
struct i915_address_space *vm = arg->vma->vm;
struct drm_i915_private *i915 = vm->i915;
struct drm_mm_node evict = arg->vma->node;
int err;
complete(&arg->completion);
mutex_lock(&i915->drm.struct_mutex);
mutex_lock(&vm->mutex);
err = i915_gem_evict_for_node(vm, &evict, 0);
mutex_unlock(&i915->drm.struct_mutex);
mutex_unlock(&vm->mutex);
return err;
}
@ -1143,39 +1142,33 @@ static int evict_vma(void *data)
static int evict_fence(void *data)
{
struct evict_vma *arg = data;
struct drm_i915_private *i915 = arg->vma->vm->i915;
int err;
complete(&arg->completion);
mutex_lock(&i915->drm.struct_mutex);
/* Mark the fence register as dirty to force the mmio update. */
err = i915_gem_object_set_tiling(arg->vma->obj, I915_TILING_Y, 512);
if (err) {
pr_err("Invalid Y-tiling settings; err:%d\n", err);
goto out_unlock;
return err;
}
err = i915_vma_pin(arg->vma, 0, 0, PIN_GLOBAL | PIN_MAPPABLE);
if (err) {
pr_err("Unable to pin vma for Y-tiled fence; err:%d\n", err);
goto out_unlock;
return err;
}
err = i915_vma_pin_fence(arg->vma);
i915_vma_unpin(arg->vma);
if (err) {
pr_err("Unable to pin Y-tiled fence; err:%d\n", err);
goto out_unlock;
return err;
}
i915_vma_unpin_fence(arg->vma);
out_unlock:
mutex_unlock(&i915->drm.struct_mutex);
return err;
return 0;
}
static int __igt_reset_evict_vma(struct intel_gt *gt,

12
drivers/gpu/drm/i915/gvt/aperture_gm.c
View File

@ -61,14 +61,14 @@ static int alloc_gm(struct intel_vgpu *vgpu, bool high_gm)
flags = PIN_MAPPABLE;
}
mutex_lock(&dev_priv->drm.struct_mutex);
mutex_lock(&dev_priv->ggtt.vm.mutex);
mmio_hw_access_pre(dev_priv);
ret = i915_gem_gtt_insert(&dev_priv->ggtt.vm, node,
size, I915_GTT_PAGE_SIZE,
I915_COLOR_UNEVICTABLE,
start, end, flags);
mmio_hw_access_post(dev_priv);
mutex_unlock(&dev_priv->drm.struct_mutex);
mutex_unlock(&dev_priv->ggtt.vm.mutex);
if (ret)
gvt_err("fail to alloc %s gm space from host\n",
high_gm ? "high" : "low");
@ -98,9 +98,9 @@ static int alloc_vgpu_gm(struct intel_vgpu *vgpu)
return 0;
out_free_aperture:
mutex_lock(&dev_priv->drm.struct_mutex);
mutex_lock(&dev_priv->ggtt.vm.mutex);
drm_mm_remove_node(&vgpu->gm.low_gm_node);
mutex_unlock(&dev_priv->drm.struct_mutex);
mutex_unlock(&dev_priv->ggtt.vm.mutex);
return ret;
}
@ -108,10 +108,10 @@ static void free_vgpu_gm(struct intel_vgpu *vgpu)
{
struct drm_i915_private *dev_priv = vgpu->gvt->dev_priv;
mutex_lock(&dev_priv->drm.struct_mutex);
mutex_lock(&dev_priv->ggtt.vm.mutex);
drm_mm_remove_node(&vgpu->gm.low_gm_node);
drm_mm_remove_node(&vgpu->gm.high_gm_node);
mutex_unlock(&dev_priv->drm.struct_mutex);
mutex_unlock(&dev_priv->ggtt.vm.mutex);
}
/**

95
drivers/gpu/drm/i915/i915_active.c
View File

@ -146,6 +146,7 @@ __active_retire(struct i915_active *ref)
if (!retire)
return;
GEM_BUG_ON(rcu_access_pointer(ref->excl));
rbtree_postorder_for_each_entry_safe(it, n, &root, node) {
GEM_BUG_ON(i915_active_request_isset(&it->base));
kmem_cache_free(global.slab_cache, it);
@ -245,6 +246,8 @@ void __i915_active_init(struct drm_i915_private *i915,
ref->flags = 0;
ref->active = active;
ref->retire = retire;
ref->excl = NULL;
ref->tree = RB_ROOT;
ref->cache = NULL;
init_llist_head(&ref->preallocated_barriers);
@ -341,6 +344,46 @@ out:
return err;
}
static void excl_cb(struct dma_fence *f, struct dma_fence_cb *cb)
{
struct i915_active *ref = container_of(cb, typeof(*ref), excl_cb);
RCU_INIT_POINTER(ref->excl, NULL);
dma_fence_put(f);
active_retire(ref);
}
void i915_active_set_exclusive(struct i915_active *ref, struct dma_fence *f)
{
/* We expect the caller to manage the exclusive timeline ordering */
GEM_BUG_ON(i915_active_is_idle(ref));
dma_fence_get(f);
rcu_read_lock();
if (rcu_access_pointer(ref->excl)) {
struct dma_fence *old;
old = dma_fence_get_rcu_safe(&ref->excl);
if (old) {
if (dma_fence_remove_callback(old, &ref->excl_cb))
atomic_dec(&ref->count);
dma_fence_put(old);
}
}
rcu_read_unlock();
atomic_inc(&ref->count);
rcu_assign_pointer(ref->excl, f);
if (dma_fence_add_callback(f, &ref->excl_cb, excl_cb)) {
RCU_INIT_POINTER(ref->excl, NULL);
atomic_dec(&ref->count);
dma_fence_put(f);
}
}
int i915_active_acquire(struct i915_active *ref)
{
int err;
@ -399,6 +442,25 @@ void i915_active_ungrab(struct i915_active *ref)
__active_ungrab(ref);
}
static int excl_wait(struct i915_active *ref)
{
struct dma_fence *old;
int err = 0;
if (!rcu_access_pointer(ref->excl))
return 0;
rcu_read_lock();
old = dma_fence_get_rcu_safe(&ref->excl);
rcu_read_unlock();
if (old) {
err = dma_fence_wait(old, true);
dma_fence_put(old);
}
return err;
}
int i915_active_wait(struct i915_active *ref)
{
struct active_node *it, *n;
@ -419,6 +481,10 @@ int i915_active_wait(struct i915_active *ref)
return 0;
}
err = excl_wait(ref);
if (err)
goto out;
rbtree_postorder_for_each_entry_safe(it, n, &ref->tree, node) {
if (is_barrier(&it->base)) { /* unconnected idle-barrier */
err = -EBUSY;
@ -430,6 +496,7 @@ int i915_active_wait(struct i915_active *ref)
break;
}
out:
__active_retire(ref);
if (err)
return err;
@ -454,26 +521,22 @@ int i915_request_await_active_request(struct i915_request *rq,
int i915_request_await_active(struct i915_request *rq, struct i915_active *ref)
{
struct active_node *it, *n;
int err;
int err = 0;
if (RB_EMPTY_ROOT(&ref->tree))
return 0;
if (rcu_access_pointer(ref->excl)) {
struct dma_fence *fence;
/* await allocates and so we need to avoid hitting the shrinker */
err = i915_active_acquire(ref);
if (err)
return err;
mutex_lock(&ref->mutex);
rbtree_postorder_for_each_entry_safe(it, n, &ref->tree, node) {
err = i915_request_await_active_request(rq, &it->base);
if (err)
break;
rcu_read_lock();
fence = dma_fence_get_rcu_safe(&ref->excl);
rcu_read_unlock();
if (fence) {
err = i915_request_await_dma_fence(rq, fence);
dma_fence_put(fence);
}
}
mutex_unlock(&ref->mutex);
i915_active_release(ref);
/* In the future we may choose to await on all fences */
return err;
}

7
drivers/gpu/drm/i915/i915_active.h
View File

@ -379,6 +379,13 @@ i915_active_add_request(struct i915_active *ref, struct i915_request *rq)
return i915_active_ref(ref, i915_request_timeline(rq), rq);
}
void i915_active_set_exclusive(struct i915_active *ref, struct dma_fence *f);
static inline bool i915_active_has_exclusive(struct i915_active *ref)
{
return rcu_access_pointer(ref->excl);
}
int i915_active_wait(struct i915_active *ref);
int i915_request_await_active(struct i915_request *rq,

5
drivers/gpu/drm/i915/i915_active_types.h
View File

@ -8,6 +8,7 @@
#define _I915_ACTIVE_TYPES_H_
#include <linux/atomic.h>
#include <linux/dma-fence.h>
#include <linux/llist.h>
#include <linux/mutex.h>
#include <linux/rbtree.h>
@ -51,6 +52,10 @@ struct i915_active {
struct mutex mutex;
atomic_t count;
/* Preallocated "exclusive" node */
struct dma_fence __rcu *excl;
struct dma_fence_cb excl_cb;
unsigned long flags;
#define I915_ACTIVE_GRAB_BIT 0

2
drivers/gpu/drm/i915/i915_drv.c
View File

@ -1895,10 +1895,8 @@ static int i915_drm_resume(struct drm_device *dev)
if (ret)
DRM_ERROR("failed to re-enable GGTT\n");
mutex_lock(&dev_priv->drm.struct_mutex);
i915_gem_restore_gtt_mappings(dev_priv);
i915_gem_restore_fences(dev_priv);
mutex_unlock(&dev_priv->drm.struct_mutex);
intel_csr_ucode_resume(dev_priv);

83
drivers/gpu/drm/i915/i915_gem.c
View File

@ -62,20 +62,31 @@
#include "intel_pm.h"
static int
insert_mappable_node(struct i915_ggtt *ggtt,
struct drm_mm_node *node, u32 size)
insert_mappable_node(struct i915_ggtt *ggtt, struct drm_mm_node *node, u32 size)
{
int err;
err = mutex_lock_interruptible(&ggtt->vm.mutex);
if (err)
return err;
memset(node, 0, sizeof(*node));
return drm_mm_insert_node_in_range(&ggtt->vm.mm, node,
size, 0, I915_COLOR_UNEVICTABLE,
0, ggtt->mappable_end,
DRM_MM_INSERT_LOW);
err = drm_mm_insert_node_in_range(&ggtt->vm.mm, node,
size, 0, I915_COLOR_UNEVICTABLE,
0, ggtt->mappable_end,
DRM_MM_INSERT_LOW);
mutex_unlock(&ggtt->vm.mutex);
return err;
}
static void
remove_mappable_node(struct drm_mm_node *node)
remove_mappable_node(struct i915_ggtt *ggtt, struct drm_mm_node *node)
{
mutex_lock(&ggtt->vm.mutex);
drm_mm_remove_node(node);
mutex_unlock(&ggtt->vm.mutex);
}
int
@ -87,7 +98,8 @@ i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
struct i915_vma *vma;
u64 pinned;
mutex_lock(&ggtt->vm.mutex);
if (mutex_lock_interruptible(&ggtt->vm.mutex))
return -EINTR;
pinned = ggtt->vm.reserved;
list_for_each_entry(vma, &ggtt->vm.bound_list, vm_link)
@ -109,20 +121,24 @@ int i915_gem_object_unbind(struct drm_i915_gem_object *obj,
LIST_HEAD(still_in_list);
int ret = 0;
lockdep_assert_held(&obj->base.dev->struct_mutex);
spin_lock(&obj->vma.lock);
while (!ret && (vma = list_first_entry_or_null(&obj->vma.list,
struct i915_vma,
obj_link))) {
struct i915_address_space *vm = vma->vm;
ret = -EBUSY;
if (!i915_vm_tryopen(vm))
break;
list_move_tail(&vma->obj_link, &still_in_list);
spin_unlock(&obj->vma.lock);
ret = -EBUSY;
if (flags & I915_GEM_OBJECT_UNBIND_ACTIVE ||
!i915_vma_is_active(vma))
ret = i915_vma_unbind(vma);
i915_vm_close(vm);
spin_lock(&obj->vma.lock);
}
list_splice(&still_in_list, &obj->vma.list);
@ -338,10 +354,6 @@ i915_gem_gtt_pread(struct drm_i915_gem_object *obj,
u64 remain, offset;
int ret;
ret = mutex_lock_interruptible(&i915->drm.struct_mutex);
if (ret)
return ret;
wakeref = intel_runtime_pm_get(&i915->runtime_pm);
vma = ERR_PTR(-ENODEV);
if (!i915_gem_object_is_tiled(obj))
@ -355,12 +367,10 @@ i915_gem_gtt_pread(struct drm_i915_gem_object *obj,
} else {
ret = insert_mappable_node(ggtt, &node, PAGE_SIZE);
if (ret)
goto out_unlock;
goto out_rpm;
GEM_BUG_ON(!drm_mm_node_allocated(&node));
}
mutex_unlock(&i915->drm.struct_mutex);
ret = i915_gem_object_lock_interruptible(obj);
if (ret)
goto out_unpin;
@ -414,17 +424,14 @@ i915_gem_gtt_pread(struct drm_i915_gem_object *obj,
i915_gem_object_unlock_fence(obj, fence);
out_unpin:
mutex_lock(&i915->drm.struct_mutex);
if (drm_mm_node_allocated(&node)) {
ggtt->vm.clear_range(&ggtt->vm, node.start, node.size);
remove_mappable_node(&node);
remove_mappable_node(ggtt, &node);
} else {
i915_vma_unpin(vma);
}
out_unlock:
out_rpm:
intel_runtime_pm_put(&i915->runtime_pm, wakeref);
mutex_unlock(&i915->drm.struct_mutex);
return ret;
}
@ -531,10 +538,6 @@ i915_gem_gtt_pwrite_fast(struct drm_i915_gem_object *obj,
void __user *user_data;
int ret;
ret = mutex_lock_interruptible(&i915->drm.struct_mutex);
if (ret)
return ret;
if (i915_gem_object_has_struct_page(obj)) {
/*
* Avoid waking the device up if we can fallback, as
@ -544,10 +547,8 @@ i915_gem_gtt_pwrite_fast(struct drm_i915_gem_object *obj,
* using the cache bypass of indirect GGTT access.
*/
wakeref = intel_runtime_pm_get_if_in_use(rpm);
if (!wakeref) {
ret = -EFAULT;
goto out_unlock;
}
if (!wakeref)
return -EFAULT;
} else {
/* No backing pages, no fallback, we must force GGTT access */
wakeref = intel_runtime_pm_get(rpm);
@ -569,8 +570,6 @@ i915_gem_gtt_pwrite_fast(struct drm_i915_gem_object *obj,
GEM_BUG_ON(!drm_mm_node_allocated(&node));
}
mutex_unlock(&i915->drm.struct_mutex);
ret = i915_gem_object_lock_interruptible(obj);
if (ret)
goto out_unpin;
@ -634,18 +633,15 @@ i915_gem_gtt_pwrite_fast(struct drm_i915_gem_object *obj,
i915_gem_object_unlock_fence(obj, fence);
out_unpin:
mutex_lock(&i915->drm.struct_mutex);
intel_gt_flush_ggtt_writes(ggtt->vm.gt);
if (drm_mm_node_allocated(&node)) {
ggtt->vm.clear_range(&ggtt->vm, node.start, node.size);
remove_mappable_node(&node);
remove_mappable_node(ggtt, &node);
} else {
i915_vma_unpin(vma);
}
out_rpm:
intel_runtime_pm_put(rpm, wakeref);
out_unlock:
mutex_unlock(&i915->drm.struct_mutex);
return ret;
}
@ -968,8 +964,6 @@ i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj,
struct i915_vma *vma;
int ret;
lockdep_assert_held(&obj->base.dev->struct_mutex);
if (i915_gem_object_never_bind_ggtt(obj))
return ERR_PTR(-ENODEV);
@ -1019,13 +1013,6 @@ i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj,
return ERR_PTR(-ENOSPC);
}
WARN(i915_vma_is_pinned(vma),
"bo is already pinned in ggtt with incorrect alignment:"
" offset=%08x, req.alignment=%llx,"
" req.map_and_fenceable=%d, vma->map_and_fenceable=%d\n",
i915_ggtt_offset(vma), alignment,
!!(flags & PIN_MAPPABLE),
i915_vma_is_map_and_fenceable(vma));
ret = i915_vma_unbind(vma);
if (ret)
return ERR_PTR(ret);
@ -1444,8 +1431,6 @@ err_unlock:
}
if (ret == -EIO) {
mutex_lock(&dev_priv->drm.struct_mutex);
/*
* Allow engines or uC initialisation to fail by marking the GPU
* as wedged. But we only want to do this when the GPU is angry,
@ -1462,8 +1447,6 @@ err_unlock:
i915_gem_restore_gtt_mappings(dev_priv);
i915_gem_restore_fences(dev_priv);
intel_init_clock_gating(dev_priv);
mutex_unlock(&dev_priv->drm.struct_mutex);
}
i915_gem_drain_freed_objects(dev_priv);

28
drivers/gpu/drm/i915/i915_gem_evict.c
View File

@ -47,8 +47,7 @@ static int ggtt_flush(struct drm_i915_private *i915)
* bound by their active reference.
*/
return i915_gem_wait_for_idle(i915,
I915_WAIT_INTERRUPTIBLE |
I915_WAIT_LOCKED,
I915_WAIT_INTERRUPTIBLE,
MAX_SCHEDULE_TIMEOUT);
}
@ -104,7 +103,7 @@ i915_gem_evict_something(struct i915_address_space *vm,
struct i915_vma *active;
int ret;
lockdep_assert_held(&vm->i915->drm.struct_mutex);
lockdep_assert_held(&vm->mutex);
trace_i915_gem_evict(vm, min_size, alignment, flags);
/*
@ -127,15 +126,6 @@ i915_gem_evict_something(struct i915_address_space *vm,
min_size, alignment, color,
start, end, mode);
/*
* Retire before we search the active list. Although we have
* reasonable accuracy in our retirement lists, we may have
* a stray pin (preventing eviction) that can only be resolved by
* retiring.
*/
if (!(flags & PIN_NONBLOCK))
i915_retire_requests(dev_priv);
search_again:
active = NULL;
INIT_LIST_HEAD(&eviction_list);
@ -235,12 +225,12 @@ found:
list_for_each_entry_safe(vma, next, &eviction_list, evict_link) {
__i915_vma_unpin(vma);
if (ret == 0)
ret = i915_vma_unbind(vma);
ret = __i915_vma_unbind(vma);
}
while (ret == 0 && (node = drm_mm_scan_color_evict(&scan))) {
vma = container_of(node, struct i915_vma, node);
ret = i915_vma_unbind(vma);
ret = __i915_vma_unbind(vma);
}
return ret;
@ -268,7 +258,7 @@ int i915_gem_evict_for_node(struct i915_address_space *vm,
struct i915_vma *vma, *next;
int ret = 0;
lockdep_assert_held(&vm->i915->drm.struct_mutex);
lockdep_assert_held(&vm->mutex);
GEM_BUG_ON(!IS_ALIGNED(start, I915_GTT_PAGE_SIZE));
GEM_BUG_ON(!IS_ALIGNED(end, I915_GTT_PAGE_SIZE));
@ -349,7 +339,7 @@ int i915_gem_evict_for_node(struct i915_address_space *vm,
list_for_each_entry_safe(vma, next, &eviction_list, evict_link) {
__i915_vma_unpin(vma);
if (ret == 0)
ret = i915_vma_unbind(vma);
ret = __i915_vma_unbind(vma);
}
return ret;
@ -373,7 +363,7 @@ int i915_gem_evict_vm(struct i915_address_space *vm)
struct i915_vma *vma, *next;
int ret;
lockdep_assert_held(&vm->i915->drm.struct_mutex);
lockdep_assert_held(&vm->mutex);
trace_i915_gem_evict_vm(vm);
/* Switch back to the default context in order to unpin
@ -388,7 +378,6 @@ int i915_gem_evict_vm(struct i915_address_space *vm)
}
INIT_LIST_HEAD(&eviction_list);
mutex_lock(&vm->mutex);
list_for_each_entry(vma, &vm->bound_list, vm_link) {
if (i915_vma_is_pinned(vma))
continue;
@ -396,13 +385,12 @@ int i915_gem_evict_vm(struct i915_address_space *vm)
__i915_vma_pin(vma);
list_add(&vma->evict_link, &eviction_list);
}
mutex_unlock(&vm->mutex);
ret = 0;
list_for_each_entry_safe(vma, next, &eviction_list, evict_link) {
__i915_vma_unpin(vma);
if (ret == 0)
ret = i915_vma_unbind(vma);
ret = __i915_vma_unbind(vma);
}
return ret;
}

9
drivers/gpu/drm/i915/i915_gem_fence_reg.c
View File

@ -230,14 +230,15 @@ static int fence_update(struct i915_fence_reg *fence,
i915_gem_object_get_tiling(vma->obj)))
return -EINVAL;
ret = i915_active_wait(&vma->active);
ret = i915_vma_sync(vma);
if (ret)
return ret;
}
old = xchg(&fence->vma, NULL);
if (old) {
ret = i915_active_wait(&old->active);
/* XXX Ideally we would move the waiting to outside the mutex */
ret = i915_vma_sync(old);
if (ret) {
fence->vma = old;
return ret;
@ -331,13 +332,15 @@ static struct i915_fence_reg *fence_find(struct drm_i915_private *i915)
return ERR_PTR(-EDEADLK);
}
static int __i915_vma_pin_fence(struct i915_vma *vma)
int __i915_vma_pin_fence(struct i915_vma *vma)
{
struct i915_ggtt *ggtt = i915_vm_to_ggtt(vma->vm);
struct i915_fence_reg *fence;
struct i915_vma *set = i915_gem_object_is_tiled(vma->obj) ? vma : NULL;
int err;
lockdep_assert_held(&vma->vm->mutex);
/* Just update our place in the LRU if our fence is getting reused. */
if (vma->fence) {
fence = vma->fence;

109
drivers/gpu/drm/i915/i915_gem_gtt.c
View File

@ -150,16 +150,18 @@ static void gmch_ggtt_invalidate(struct i915_ggtt *ggtt)
static int ppgtt_bind_vma(struct i915_vma *vma,
enum i915_cache_level cache_level,
u32 unused)
u32 flags)
{
u32 pte_flags;
int err;
if (!i915_vma_is_bound(vma, I915_VMA_LOCAL_BIND)) {
if (flags & I915_VMA_ALLOC) {
err = vma->vm->allocate_va_range(vma->vm,
vma->node.start, vma->size);
if (err)
return err;
set_bit(I915_VMA_ALLOC_BIT, __i915_vma_flags(vma));
}
/* Applicable to VLV, and gen8+ */
@ -167,6 +169,7 @@ static int ppgtt_bind_vma(struct i915_vma *vma,
if (i915_gem_object_is_readonly(vma->obj))
pte_flags |= PTE_READ_ONLY;
GEM_BUG_ON(!test_bit(I915_VMA_ALLOC_BIT, __i915_vma_flags(vma)));
vma->vm->insert_entries(vma->vm, vma, cache_level, pte_flags);
wmb();
@ -175,7 +178,8 @@ static int ppgtt_bind_vma(struct i915_vma *vma,
static void ppgtt_unbind_vma(struct i915_vma *vma)
{
vma->vm->clear_range(vma->vm, vma->node.start, vma->size);
if (test_and_clear_bit(I915_VMA_ALLOC_BIT, __i915_vma_flags(vma)))
vma->vm->clear_range(vma->vm, vma->node.start, vma->size);
}
static int ppgtt_set_pages(struct i915_vma *vma)
@ -503,15 +507,26 @@ static void i915_address_space_fini(struct i915_address_space *vm)
mutex_destroy(&vm->mutex);
}
static void ppgtt_destroy_vma(struct i915_address_space *vm)
void __i915_vm_close(struct i915_address_space *vm)
{
struct i915_vma *vma, *vn;
mutex_lock(&vm->i915->drm.struct_mutex);
list_for_each_entry_safe(vma, vn, &vm->bound_list, vm_link)
mutex_lock(&vm->mutex);
list_for_each_entry_safe(vma, vn, &vm->bound_list, vm_link) {
struct drm_i915_gem_object *obj = vma->obj;
/* Keep the obj (and hence the vma) alive as _we_ destroy it */
if (!kref_get_unless_zero(&obj->base.refcount))
continue;
atomic_and(~I915_VMA_PIN_MASK, &vma->flags);
WARN_ON(__i915_vma_unbind(vma));
i915_vma_destroy(vma);
i915_gem_object_put(obj);
}
GEM_BUG_ON(!list_empty(&vm->bound_list));
mutex_unlock(&vm->i915->drm.struct_mutex);
mutex_unlock(&vm->mutex);
}
static void __i915_vm_release(struct work_struct *work)
@ -519,8 +534,6 @@ static void __i915_vm_release(struct work_struct *work)
struct i915_address_space *vm =
container_of(work, struct i915_address_space, rcu.work);
ppgtt_destroy_vma(vm);
vm->cleanup(vm);
i915_address_space_fini(vm);
@ -535,7 +548,6 @@ void i915_vm_release(struct kref *kref)
GEM_BUG_ON(i915_is_ggtt(vm));
trace_i915_ppgtt_release(vm);
vm->closed = true;
queue_rcu_work(vm->i915->wq, &vm->rcu);
}
@ -543,6 +555,7 @@ static void i915_address_space_init(struct i915_address_space *vm, int subclass)
{
kref_init(&vm->ref);
INIT_RCU_WORK(&vm->rcu, __i915_vm_release);
atomic_set(&vm->open, 1);
/*
* The vm->mutex must be reclaim safe (for use in the shrinker).
@ -1771,12 +1784,8 @@ static void gen6_ppgtt_free_pd(struct gen6_ppgtt *ppgtt)
static void gen6_ppgtt_cleanup(struct i915_address_space *vm)
{
struct gen6_ppgtt *ppgtt = to_gen6_ppgtt(i915_vm_to_ppgtt(vm));
struct drm_i915_private *i915 = vm->i915;
/* FIXME remove the struct_mutex to bring the locking under control */
mutex_lock(&i915->drm.struct_mutex);
i915_vma_destroy(ppgtt->vma);
mutex_unlock(&i915->drm.struct_mutex);
gen6_ppgtt_free_pd(ppgtt);
free_scratch(vm);
@ -1865,7 +1874,8 @@ static struct i915_vma *pd_vma_create(struct gen6_ppgtt *ppgtt, int size)
i915_active_init(i915, &vma->active, NULL, NULL);
vma->vm = &ggtt->vm;
mutex_init(&vma->pages_mutex);
vma->vm = i915_vm_get(&ggtt->vm);
vma->ops = &pd_vma_ops;
vma->private = ppgtt;
@ -1885,7 +1895,7 @@ int gen6_ppgtt_pin(struct i915_ppgtt *base)
struct gen6_ppgtt *ppgtt = to_gen6_ppgtt(base);
int err = 0;
GEM_BUG_ON(ppgtt->base.vm.closed);
GEM_BUG_ON(!atomic_read(&ppgtt->base.vm.open));
/*
* Workaround the limited maximum vma->pin_count and the aliasing_ppgtt
@ -2463,14 +2473,18 @@ static int aliasing_gtt_bind_vma(struct i915_vma *vma,
if (flags & I915_VMA_LOCAL_BIND) {
struct i915_ppgtt *alias = i915_vm_to_ggtt(vma->vm)->alias;
if (!i915_vma_is_bound(vma, I915_VMA_LOCAL_BIND)) {
if (flags & I915_VMA_ALLOC) {
ret = alias->vm.allocate_va_range(&alias->vm,
vma->node.start,
vma->size);
if (ret)
return ret;
set_bit(I915_VMA_ALLOC_BIT, __i915_vma_flags(vma));
}
GEM_BUG_ON(!test_bit(I915_VMA_ALLOC_BIT,
__i915_vma_flags(vma)));
alias->vm.insert_entries(&alias->vm, vma,
cache_level, pte_flags);
}
@ -2499,7 +2513,7 @@ static void aliasing_gtt_unbind_vma(struct i915_vma *vma)
vm->clear_range(vm, vma->node.start, vma->size);
}
if (i915_vma_is_bound(vma, I915_VMA_LOCAL_BIND)) {
if (test_and_clear_bit(I915_VMA_ALLOC_BIT, __i915_vma_flags(vma))) {
struct i915_address_space *vm =
&i915_vm_to_ggtt(vma->vm)->alias->vm;
@ -2602,22 +2616,16 @@ err_ppgtt:
static void fini_aliasing_ppgtt(struct i915_ggtt *ggtt)
{
struct drm_i915_private *i915 = ggtt->vm.i915;
struct i915_ppgtt *ppgtt;
mutex_lock(&i915->drm.struct_mutex);
ppgtt = fetch_and_zero(&ggtt->alias);
if (!ppgtt)
goto out;
return;
i915_vm_put(&ppgtt->vm);
ggtt->vm.vma_ops.bind_vma = ggtt_bind_vma;
ggtt->vm.vma_ops.unbind_vma = ggtt_unbind_vma;
out:
mutex_unlock(&i915->drm.struct_mutex);
}
static int ggtt_reserve_guc_top(struct i915_ggtt *ggtt)
@ -2734,32 +2742,28 @@ int i915_init_ggtt(struct drm_i915_private *i915)
static void ggtt_cleanup_hw(struct i915_ggtt *ggtt)
{
struct drm_i915_private *i915 = ggtt->vm.i915;
struct i915_vma *vma, *vn;
ggtt->vm.closed = true;
atomic_set(&ggtt->vm.open, 0);
rcu_barrier(); /* flush the RCU'ed__i915_vm_release */
flush_workqueue(i915->wq);
flush_workqueue(ggtt->vm.i915->wq);
mutex_lock(&i915->drm.struct_mutex);
mutex_lock(&ggtt->vm.mutex);
list_for_each_entry_safe(vma, vn, &ggtt->vm.bound_list, vm_link)
WARN_ON(i915_vma_unbind(vma));
WARN_ON(__i915_vma_unbind(vma));
if (drm_mm_node_allocated(&ggtt->error_capture))
drm_mm_remove_node(&ggtt->error_capture);
ggtt_release_guc_top(ggtt);
if (drm_mm_initialized(&ggtt->vm.mm)) {
intel_vgt_deballoon(ggtt);
i915_address_space_fini(&ggtt->vm);
}
intel_vgt_deballoon(ggtt);
ggtt->vm.cleanup(&ggtt->vm);
mutex_unlock(&i915->drm.struct_mutex);
mutex_unlock(&ggtt->vm.mutex);
i915_address_space_fini(&ggtt->vm);
arch_phys_wc_del(ggtt->mtrr);
io_mapping_fini(&ggtt->iomap);
@ -3188,9 +3192,6 @@ int i915_ggtt_probe_hw(struct drm_i915_private *i915)
static int ggtt_init_hw(struct i915_ggtt *ggtt)
{
struct drm_i915_private *i915 = ggtt->vm.i915;
int ret = 0;
mutex_lock(&i915->drm.struct_mutex);
i915_address_space_init(&ggtt->vm, VM_CLASS_GGTT);
@ -3206,18 +3207,14 @@ static int ggtt_init_hw(struct i915_ggtt *ggtt)
ggtt->gmadr.start,
ggtt->mappable_end)) {
ggtt->vm.cleanup(&ggtt->vm);
ret = -EIO;
goto out;
return -EIO;
}
ggtt->mtrr = arch_phys_wc_add(ggtt->gmadr.start, ggtt->mappable_end);
i915_ggtt_init_fences(ggtt);
out:
mutex_unlock(&i915->drm.struct_mutex);
return ret;
return 0;
}
/**
@ -3289,6 +3286,7 @@ static void ggtt_restore_mappings(struct i915_ggtt *ggtt)
{
struct i915_vma *vma, *vn;
bool flush = false;
int open;
intel_gt_check_and_clear_faults(ggtt->vm.gt);
@ -3296,7 +3294,9 @@ static void ggtt_restore_mappings(struct i915_ggtt *ggtt)
/* First fill our portion of the GTT with scratch pages */
ggtt->vm.clear_range(&ggtt->vm, 0, ggtt->vm.total);
ggtt->vm.closed = true; /* skip rewriting PTE on VMA unbind */
/* Skip rewriting PTE on VMA unbind. */
open = atomic_xchg(&ggtt->vm.open, 0);
/* clflush objects bound into the GGTT and rebind them. */
list_for_each_entry_safe(vma, vn, &ggtt->vm.bound_list, vm_link) {
@ -3305,24 +3305,20 @@ static void ggtt_restore_mappings(struct i915_ggtt *ggtt)
if (!i915_vma_is_bound(vma, I915_VMA_GLOBAL_BIND))
continue;
mutex_unlock(&ggtt->vm.mutex);
if (!i915_vma_unbind(vma))
goto lock;
if (!__i915_vma_unbind(vma))
continue;
clear_bit(I915_VMA_GLOBAL_BIND_BIT, __i915_vma_flags(vma));
WARN_ON(i915_vma_bind(vma,
obj ? obj->cache_level : 0,
PIN_UPDATE));
PIN_GLOBAL, NULL));
if (obj) { /* only used during resume => exclusive access */
flush |= fetch_and_zero(&obj->write_domain);
obj->read_domains |= I915_GEM_DOMAIN_GTT;
}
lock:
mutex_lock(&ggtt->vm.mutex);
}
ggtt->vm.closed = false;
atomic_set(&ggtt->vm.open, open);
ggtt->invalidate(ggtt);
mutex_unlock(&ggtt->vm.mutex);
@ -3714,7 +3710,8 @@ int i915_gem_gtt_insert(struct i915_address_space *vm,
u64 offset;
int err;
lockdep_assert_held(&vm->i915->drm.struct_mutex);
lockdep_assert_held(&vm->mutex);
GEM_BUG_ON(!size);
GEM_BUG_ON(!IS_ALIGNED(size, I915_GTT_PAGE_SIZE));
GEM_BUG_ON(alignment && !is_power_of_2(alignment));

45
drivers/gpu/drm/i915/i915_gem_gtt.h
View File

@ -307,7 +307,14 @@ struct i915_address_space {
unsigned int bind_async_flags;
bool closed;
/*
* Each active user context has its own address space (in full-ppgtt).
* Since the vm may be shared between multiple contexts, we count how
* many contexts keep us "open". Once open hits zero, we are closed
* and do not allow any new attachments, and proceed to shutdown our
* vma and page directories.
*/
atomic_t open;
struct mutex mutex; /* protects vma and our lists */
#define VM_CLASS_GGTT 0
@ -581,6 +588,35 @@ static inline void i915_vm_put(struct i915_address_space *vm)
kref_put(&vm->ref, i915_vm_release);
}
static inline struct i915_address_space *
i915_vm_open(struct i915_address_space *vm)
{
GEM_BUG_ON(!atomic_read(&vm->open));
atomic_inc(&vm->open);
return i915_vm_get(vm);
}
static inline bool
i915_vm_tryopen(struct i915_address_space *vm)
{
if (atomic_add_unless(&vm->open, 1, 0))
return i915_vm_get(vm);
return false;
}
void __i915_vm_close(struct i915_address_space *vm);
static inline void
i915_vm_close(struct i915_address_space *vm)
{
GEM_BUG_ON(!atomic_read(&vm->open));
if (atomic_dec_and_test(&vm->open))
__i915_vm_close(vm);
i915_vm_put(vm);
}
int gen6_ppgtt_pin(struct i915_ppgtt *base);
void gen6_ppgtt_unpin(struct i915_ppgtt *base);
void gen6_ppgtt_unpin_all(struct i915_ppgtt *base);
@ -613,10 +649,9 @@ int i915_gem_gtt_insert(struct i915_address_space *vm,
#define PIN_OFFSET_BIAS BIT_ULL(6)
#define PIN_OFFSET_FIXED BIT_ULL(7)
#define PIN_MBZ BIT_ULL(8) /* I915_VMA_PIN_OVERFLOW */
#define PIN_GLOBAL BIT_ULL(9) /* I915_VMA_GLOBAL_BIND */
#define PIN_USER BIT_ULL(10) /* I915_VMA_LOCAL_BIND */
#define PIN_UPDATE BIT_ULL(11)
#define PIN_UPDATE BIT_ULL(9)
#define PIN_GLOBAL BIT_ULL(10) /* I915_VMA_GLOBAL_BIND */
#define PIN_USER BIT_ULL(11) /* I915_VMA_LOCAL_BIND */
#define PIN_OFFSET_MASK (-I915_GTT_PAGE_SIZE)

32
drivers/gpu/drm/i915/i915_perf.c
View File

@ -1204,15 +1204,10 @@ static int i915_oa_read(struct i915_perf_stream *stream,
static struct intel_context *oa_pin_context(struct i915_perf_stream *stream)
{
struct i915_gem_engines_iter it;
struct drm_i915_private *i915 = stream->dev_priv;
struct i915_gem_context *ctx = stream->ctx;
struct intel_context *ce;
int err;
err = i915_mutex_lock_interruptible(&i915->drm);
if (err)
return ERR_PTR(err);
for_each_gem_engine(ce, i915_gem_context_lock_engines(ctx), it) {
if (ce->engine->class != RENDER_CLASS)
continue;
@ -1229,10 +1224,6 @@ static struct intel_context *oa_pin_context(struct i915_perf_stream *stream)
}
i915_gem_context_unlock_engines(ctx);
mutex_unlock(&i915->drm.struct_mutex);
if (err)
return ERR_PTR(err);
return stream->pinned_ctx;
}
@ -1331,32 +1322,22 @@ static int oa_get_render_ctx_id(struct i915_perf_stream *stream)
*/
static void oa_put_render_ctx_id(struct i915_perf_stream *stream)
{
struct drm_i915_private *dev_priv = stream->dev_priv;
struct intel_context *ce;
stream->specific_ctx_id = INVALID_CTX_ID;
stream->specific_ctx_id_mask = 0;
ce = fetch_and_zero(&stream->pinned_ctx);
if (ce) {
mutex_lock(&dev_priv->drm.struct_mutex);
if (ce)
intel_context_unpin(ce);
mutex_unlock(&dev_priv->drm.struct_mutex);
}
}
static void
free_oa_buffer(struct i915_perf_stream *stream)
{
struct drm_i915_private *i915 = stream->dev_priv;
mutex_lock(&i915->drm.struct_mutex);
i915_vma_unpin_and_release(&stream->oa_buffer.vma,
I915_VMA_RELEASE_MAP);
mutex_unlock(&i915->drm.struct_mutex);
stream->oa_buffer.vaddr = NULL;
}
@ -1511,18 +1492,13 @@ static int alloc_oa_buffer(struct i915_perf_stream *stream)
if (WARN_ON(stream->oa_buffer.vma))
return -ENODEV;
ret = i915_mutex_lock_interruptible(&dev_priv->drm);
if (ret)
return ret;
BUILD_BUG_ON_NOT_POWER_OF_2(OA_BUFFER_SIZE);
BUILD_BUG_ON(OA_BUFFER_SIZE < SZ_128K || OA_BUFFER_SIZE > SZ_16M);
bo = i915_gem_object_create_shmem(dev_priv, OA_BUFFER_SIZE);
if (IS_ERR(bo)) {
DRM_ERROR("Failed to allocate OA buffer\n");
ret = PTR_ERR(bo);
goto unlock;
return PTR_ERR(bo);
}
i915_gem_object_set_cache_coherency(bo, I915_CACHE_LLC);
@ -1546,7 +1522,7 @@ static int alloc_oa_buffer(struct i915_perf_stream *stream)
i915_ggtt_offset(stream->oa_buffer.vma),
stream->oa_buffer.vaddr);
goto unlock;
return 0;
err_unpin:
__i915_vma_unpin(vma);
@ -1557,8 +1533,6 @@ err_unref:
stream->oa_buffer.vaddr = NULL;
stream->oa_buffer.vma = NULL;
unlock:
mutex_unlock(&dev_priv->drm.struct_mutex);
return ret;
}

526
drivers/gpu/drm/i915/i915_vma.c
View File

@ -32,6 +32,7 @@
#include "i915_drv.h"
#include "i915_globals.h"
#include "i915_sw_fence_work.h"
#include "i915_trace.h"
#include "i915_vma.h"
@ -110,7 +111,8 @@ vma_create(struct drm_i915_gem_object *obj,
if (vma == NULL)
return ERR_PTR(-ENOMEM);
vma->vm = vm;
mutex_init(&vma->pages_mutex);
vma->vm = i915_vm_get(vm);
vma->ops = &vm->vma_ops;
vma->obj = obj;
vma->resv = obj->base.resv;
@ -261,8 +263,6 @@ vma_lookup(struct drm_i915_gem_object *obj,
* Once created, the VMA is kept until either the object is freed, or the
* address space is closed.
*
* Must be called with struct_mutex held.
*
* Returns the vma, or an error pointer.
*/
struct i915_vma *
@ -273,7 +273,7 @@ i915_vma_instance(struct drm_i915_gem_object *obj,
struct i915_vma *vma;
GEM_BUG_ON(view && !i915_is_ggtt(vm));
GEM_BUG_ON(vm->closed);
GEM_BUG_ON(!atomic_read(&vm->open));
spin_lock(&obj->vma.lock);
vma = vma_lookup(obj, vm, view);
@ -287,18 +287,63 @@ i915_vma_instance(struct drm_i915_gem_object *obj,
return vma;
}
struct i915_vma_work {
struct dma_fence_work base;
struct i915_vma *vma;
enum i915_cache_level cache_level;
unsigned int flags;
};
static int __vma_bind(struct dma_fence_work *work)
{
struct i915_vma_work *vw = container_of(work, typeof(*vw), base);
struct i915_vma *vma = vw->vma;
int err;
err = vma->ops->bind_vma(vma, vw->cache_level, vw->flags);
if (err)
atomic_or(I915_VMA_ERROR, &vma->flags);
if (vma->obj)
__i915_gem_object_unpin_pages(vma->obj);
return err;
}
static const struct dma_fence_work_ops bind_ops = {
.name = "bind",
.work = __vma_bind,
};
struct i915_vma_work *i915_vma_work(void)
{
struct i915_vma_work *vw;
vw = kzalloc(sizeof(*vw), GFP_KERNEL);
if (!vw)
return NULL;
dma_fence_work_init(&vw->base, &bind_ops);
vw->base.dma.error = -EAGAIN; /* disable the worker by default */
return vw;
}
/**
* i915_vma_bind - Sets up PTEs for an VMA in it's corresponding address space.
* @vma: VMA to map
* @cache_level: mapping cache level
* @flags: flags like global or local mapping
* @work: preallocated worker for allocating and binding the PTE
*
* DMA addresses are taken from the scatter-gather table of this object (or of
* this VMA in case of non-default GGTT views) and PTE entries set up.
* Note that DMA addresses are also the only part of the SG table we care about.
*/
int i915_vma_bind(struct i915_vma *vma, enum i915_cache_level cache_level,
u32 flags)
int i915_vma_bind(struct i915_vma *vma,
enum i915_cache_level cache_level,
u32 flags,
struct i915_vma_work *work)
{
u32 bind_flags;
u32 vma_flags;
@ -315,11 +360,8 @@ int i915_vma_bind(struct i915_vma *vma, enum i915_cache_level cache_level,
if (GEM_DEBUG_WARN_ON(!flags))
return -EINVAL;
bind_flags = 0;
if (flags & PIN_GLOBAL)
bind_flags |= I915_VMA_GLOBAL_BIND;
if (flags & PIN_USER)
bind_flags |= I915_VMA_LOCAL_BIND;
bind_flags = flags;
bind_flags &= I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND;
vma_flags = atomic_read(&vma->flags);
vma_flags &= I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND;
@ -333,9 +375,32 @@ int i915_vma_bind(struct i915_vma *vma, enum i915_cache_level cache_level,
GEM_BUG_ON(!vma->pages);
trace_i915_vma_bind(vma, bind_flags);
ret = vma->ops->bind_vma(vma, cache_level, bind_flags);
if (ret)
return ret;
if (work && (bind_flags & ~vma_flags) & vma->vm->bind_async_flags) {
work->vma = vma;
work->cache_level = cache_level;
work->flags = bind_flags | I915_VMA_ALLOC;
/*
* Note we only want to chain up to the migration fence on
* the pages (not the object itself). As we don't track that,
* yet, we have to use the exclusive fence instead.
*
* Also note that we do not want to track the async vma as
* part of the obj->resv->excl_fence as it only affects
* execution and not content or object's backing store lifetime.
*/
GEM_BUG_ON(i915_active_has_exclusive(&vma->active));
i915_active_set_exclusive(&vma->active, &work->base.dma);
work->base.dma.error = 0; /* enable the queue_work() */
if (vma->obj)
__i915_gem_object_pin_pages(vma->obj);
} else {
GEM_BUG_ON((bind_flags & ~vma_flags) & vma->vm->bind_async_flags);
ret = vma->ops->bind_vma(vma, cache_level, bind_flags);
if (ret)
return ret;
}
atomic_or(bind_flags, &vma->flags);
return 0;
@ -348,9 +413,7 @@ void __iomem *i915_vma_pin_iomap(struct i915_vma *vma)
/* Access through the GTT requires the device to be awake. */
assert_rpm_wakelock_held(&vma->vm->i915->runtime_pm);
lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);
if (WARN_ON(!i915_vma_is_map_and_fenceable(vma))) {
if (GEM_WARN_ON(!i915_vma_is_map_and_fenceable(vma))) {
err = -ENODEV;
goto err;
}
@ -358,7 +421,7 @@ void __iomem *i915_vma_pin_iomap(struct i915_vma *vma)
GEM_BUG_ON(!i915_vma_is_ggtt(vma));
GEM_BUG_ON(!i915_vma_is_bound(vma, I915_VMA_GLOBAL_BIND));
ptr = vma->iomap;
ptr = READ_ONCE(vma->iomap);
if (ptr == NULL) {
ptr = io_mapping_map_wc(&i915_vm_to_ggtt(vma->vm)->iomap,
vma->node.start,
@ -368,7 +431,10 @@ void __iomem *i915_vma_pin_iomap(struct i915_vma *vma)
goto err;
}
vma->iomap = ptr;
if (unlikely(cmpxchg(&vma->iomap, NULL, ptr))) {
io_mapping_unmap(ptr);
ptr = vma->iomap;
}
}
__i915_vma_pin(vma);
@ -388,18 +454,12 @@ err:
void i915_vma_flush_writes(struct i915_vma *vma)
{
if (!i915_vma_has_ggtt_write(vma))
return;
intel_gt_flush_ggtt_writes(vma->vm->gt);
i915_vma_unset_ggtt_write(vma);
if (i915_vma_unset_ggtt_write(vma))
intel_gt_flush_ggtt_writes(vma->vm->gt);
}
void i915_vma_unpin_iomap(struct i915_vma *vma)
{
lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);
GEM_BUG_ON(vma->iomap == NULL);
i915_vma_flush_writes(vma);
@ -435,6 +495,9 @@ bool i915_vma_misplaced(const struct i915_vma *vma,
if (!drm_mm_node_allocated(&vma->node))
return false;
if (test_bit(I915_VMA_ERROR_BIT, __i915_vma_flags(vma)))
return true;
if (vma->node.size < size)
return true;
@ -535,7 +598,6 @@ static void assert_bind_count(const struct drm_i915_gem_object *obj)
static int
i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
{
struct drm_i915_private *dev_priv = vma->vm->i915;
unsigned long color;
u64 start, end;
int ret;
@ -561,7 +623,7 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
end = vma->vm->total;
if (flags & PIN_MAPPABLE)
end = min_t(u64, end, dev_priv->ggtt.mappable_end);
end = min_t(u64, end, i915_vm_to_ggtt(vma->vm)->mappable_end);
if (flags & PIN_ZONE_4G)
end = min_t(u64, end, (1ULL << 32) - I915_GTT_PAGE_SIZE);
GEM_BUG_ON(!IS_ALIGNED(end, I915_GTT_PAGE_SIZE));
@ -578,34 +640,20 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
}
color = 0;
if (vma->obj) {
ret = i915_gem_object_pin_pages(vma->obj);
if (ret)
return ret;
if (i915_vm_has_cache_coloring(vma->vm))
color = vma->obj->cache_level;
}
GEM_BUG_ON(vma->pages);
ret = vma->ops->set_pages(vma);
if (ret)
goto err_unpin;
if (vma->obj && i915_vm_has_cache_coloring(vma->vm))
color = vma->obj->cache_level;
if (flags & PIN_OFFSET_FIXED) {
u64 offset = flags & PIN_OFFSET_MASK;
if (!IS_ALIGNED(offset, alignment) ||
range_overflows(offset, size, end)) {
ret = -EINVAL;
goto err_clear;
}
range_overflows(offset, size, end))
return -EINVAL;
ret = i915_gem_gtt_reserve(vma->vm, &vma->node,
size, offset, color,
flags);
if (ret)
goto err_clear;
return ret;
} else {
/*
* We only support huge gtt pages through the 48b PPGTT,
@ -644,7 +692,7 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
size, alignment, color,
start, end, flags);
if (ret)
goto err_clear;
return ret;
GEM_BUG_ON(vma->node.start < start);
GEM_BUG_ON(vma->node.start + vma->node.size > end);
@ -652,23 +700,15 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
GEM_BUG_ON(!i915_gem_valid_gtt_space(vma, color));
mutex_lock(&vma->vm->mutex);
list_add_tail(&vma->vm_link, &vma->vm->bound_list);
mutex_unlock(&vma->vm->mutex);
if (vma->obj) {
atomic_inc(&vma->obj->mm.pages_pin_count);
atomic_inc(&vma->obj->bind_count);
assert_bind_count(vma->obj);
}
return 0;
err_clear:
vma->ops->clear_pages(vma);
err_unpin:
if (vma->obj)
i915_gem_object_unpin_pages(vma->obj);
return ret;
}
static void
@ -677,12 +717,7 @@ i915_vma_remove(struct i915_vma *vma)
GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
GEM_BUG_ON(i915_vma_is_bound(vma, I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND));
vma->ops->clear_pages(vma);
mutex_lock(&vma->vm->mutex);
drm_mm_remove_node(&vma->node);
list_del(&vma->vm_link);
mutex_unlock(&vma->vm->mutex);
/*
* Since the unbound list is global, only move to that list if
@ -701,51 +736,211 @@ i915_vma_remove(struct i915_vma *vma)
i915_gem_object_unpin_pages(obj);
assert_bind_count(obj);
}
drm_mm_remove_node(&vma->node);
}
int __i915_vma_do_pin(struct i915_vma *vma,
u64 size, u64 alignment, u64 flags)
static bool try_qad_pin(struct i915_vma *vma, unsigned int flags)
{
const unsigned int bound = atomic_read(&vma->flags);
int ret;
unsigned int bound;
bool pinned = true;
lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);
GEM_BUG_ON((flags & (PIN_GLOBAL | PIN_USER)) == 0);
GEM_BUG_ON((flags & PIN_GLOBAL) && !i915_vma_is_ggtt(vma));
bound = atomic_read(&vma->flags);
do {
if (unlikely(flags & ~bound))
return false;
if (WARN_ON(bound & I915_VMA_PIN_OVERFLOW)) {
ret = -EBUSY;
goto err_unpin;
if (unlikely(bound & (I915_VMA_OVERFLOW | I915_VMA_ERROR)))
return false;
if (!(bound & I915_VMA_PIN_MASK))
goto unpinned;
GEM_BUG_ON(((bound + 1) & I915_VMA_PIN_MASK) == 0);
} while (!atomic_try_cmpxchg(&vma->flags, &bound, bound + 1));
return true;
unpinned:
/*
* If pin_count==0, but we are bound, check under the lock to avoid
* racing with a concurrent i915_vma_unbind().
*/
mutex_lock(&vma->vm->mutex);
do {
if (unlikely(bound & (I915_VMA_OVERFLOW | I915_VMA_ERROR))) {
pinned = false;
break;
}
if (unlikely(flags & ~bound)) {
pinned = false;
break;
}
} while (!atomic_try_cmpxchg(&vma->flags, &bound, bound + 1));
mutex_unlock(&vma->vm->mutex);
return pinned;
}
static int vma_get_pages(struct i915_vma *vma)
{
int err = 0;
if (atomic_add_unless(&vma->pages_count, 1, 0))
return 0;
/* Allocations ahoy! */
if (mutex_lock_interruptible(&vma->pages_mutex))
return -EINTR;
if (!atomic_read(&vma->pages_count)) {
if (vma->obj) {
err = i915_gem_object_pin_pages(vma->obj);
if (err)
goto unlock;
}
err = vma->ops->set_pages(vma);
if (err)
goto unlock;
}
atomic_inc(&vma->pages_count);
unlock:
mutex_unlock(&vma->pages_mutex);
return err;
}
static void __vma_put_pages(struct i915_vma *vma, unsigned int count)
{
/* We allocate under vma_get_pages, so beware the shrinker */
mutex_lock_nested(&vma->pages_mutex, SINGLE_DEPTH_NESTING);
GEM_BUG_ON(atomic_read(&vma->pages_count) < count);
if (atomic_sub_return(count, &vma->pages_count) == 0) {
vma->ops->clear_pages(vma);
GEM_BUG_ON(vma->pages);
if (vma->obj)
i915_gem_object_unpin_pages(vma->obj);
}
mutex_unlock(&vma->pages_mutex);
}
static void vma_put_pages(struct i915_vma *vma)
{
if (atomic_add_unless(&vma->pages_count, -1, 1))
return;
__vma_put_pages(vma, 1);
}
static void vma_unbind_pages(struct i915_vma *vma)
{
unsigned int count;
lockdep_assert_held(&vma->vm->mutex);
/* The upper portion of pages_count is the number of bindings */
count = atomic_read(&vma->pages_count);
count >>= I915_VMA_PAGES_BIAS;
GEM_BUG_ON(!count);
__vma_put_pages(vma, count | count << I915_VMA_PAGES_BIAS);
}
int i915_vma_pin(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
{
struct i915_vma_work *work = NULL;
unsigned int bound;
int err;
BUILD_BUG_ON(PIN_GLOBAL != I915_VMA_GLOBAL_BIND);
BUILD_BUG_ON(PIN_USER != I915_VMA_LOCAL_BIND);
GEM_BUG_ON(flags & PIN_UPDATE);
GEM_BUG_ON(!(flags & (PIN_USER | PIN_GLOBAL)));
/* First try and grab the pin without rebinding the vma */
if (try_qad_pin(vma, flags & I915_VMA_BIND_MASK))
return 0;
err = vma_get_pages(vma);
if (err)
return err;
if (flags & vma->vm->bind_async_flags) {
work = i915_vma_work();
if (!work) {
err = -ENOMEM;
goto err_pages;
}
}
if ((bound & I915_VMA_BIND_MASK) == 0) {
ret = i915_vma_insert(vma, size, alignment, flags);
if (ret)
goto err_unpin;
}
GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
/* No more allocations allowed once we hold vm->mutex */
err = mutex_lock_interruptible(&vma->vm->mutex);
if (err)
goto err_fence;
ret = i915_vma_bind(vma, vma->obj ? vma->obj->cache_level : 0, flags);
if (ret)
bound = atomic_read(&vma->flags);
if (unlikely(bound & I915_VMA_ERROR)) {
err = -ENOMEM;
goto err_unlock;
}
if (unlikely(!((bound + 1) & I915_VMA_PIN_MASK))) {
err = -EAGAIN; /* pins are meant to be fairly temporary */
goto err_unlock;
}
if (unlikely(!(flags & ~bound & I915_VMA_BIND_MASK))) {
__i915_vma_pin(vma);
goto err_unlock;
}
err = i915_active_acquire(&vma->active);
if (err)
goto err_unlock;
if (!(bound & I915_VMA_BIND_MASK)) {
err = i915_vma_insert(vma, size, alignment, flags);
if (err)
goto err_active;
if (i915_is_ggtt(vma->vm))
__i915_vma_set_map_and_fenceable(vma);
}
GEM_BUG_ON(!vma->pages);
err = i915_vma_bind(vma,
vma->obj ? vma->obj->cache_level : 0,
flags, work);
if (err)
goto err_remove;
GEM_BUG_ON(!i915_vma_is_bound(vma, I915_VMA_BIND_MASK));
if ((bound ^ atomic_read(&vma->flags)) & I915_VMA_GLOBAL_BIND)
__i915_vma_set_map_and_fenceable(vma);
/* There should only be at most 2 active bindings (user, global) */
GEM_BUG_ON(bound + I915_VMA_PAGES_ACTIVE < bound);
atomic_add(I915_VMA_PAGES_ACTIVE, &vma->pages_count);
list_move_tail(&vma->vm_link, &vma->vm->bound_list);
__i915_vma_pin(vma);
GEM_BUG_ON(!i915_vma_is_pinned(vma));
GEM_BUG_ON(!i915_vma_is_bound(vma, flags));
GEM_BUG_ON(i915_vma_misplaced(vma, size, alignment, flags));
return 0;
err_remove:
if ((bound & I915_VMA_BIND_MASK) == 0) {
if (!i915_vma_is_bound(vma, I915_VMA_BIND_MASK))
i915_vma_remove(vma);
GEM_BUG_ON(vma->pages);
GEM_BUG_ON(atomic_read(&vma->flags) & I915_VMA_BIND_MASK);
}
err_unpin:
__i915_vma_unpin(vma);
return ret;
err_active:
i915_active_release(&vma->active);
err_unlock:
mutex_unlock(&vma->vm->mutex);
err_fence:
if (work)
dma_fence_work_commit(&work->base);
err_pages:
vma_put_pages(vma);
return err;
}
void i915_vma_close(struct i915_vma *vma)
@ -776,9 +971,6 @@ static void __i915_vma_remove_closed(struct i915_vma *vma)
{
struct drm_i915_private *i915 = vma->vm->i915;
if (!i915_vma_is_closed(vma))
return;
spin_lock_irq(&i915->gt.closed_lock);
list_del_init(&vma->closed_link);
spin_unlock_irq(&i915->gt.closed_lock);
@ -786,40 +978,35 @@ static void __i915_vma_remove_closed(struct i915_vma *vma)
void i915_vma_reopen(struct i915_vma *vma)
{
__i915_vma_remove_closed(vma);
if (i915_vma_is_closed(vma))
__i915_vma_remove_closed(vma);
}
static void __i915_vma_destroy(struct i915_vma *vma)
void i915_vma_destroy(struct i915_vma *vma)
{
GEM_BUG_ON(drm_mm_node_allocated(&vma->node));
GEM_BUG_ON(vma->fence);
if (drm_mm_node_allocated(&vma->node)) {
mutex_lock(&vma->vm->mutex);
atomic_and(~I915_VMA_PIN_MASK, &vma->flags);
WARN_ON(__i915_vma_unbind(vma));
mutex_unlock(&vma->vm->mutex);
GEM_BUG_ON(drm_mm_node_allocated(&vma->node));
}
GEM_BUG_ON(i915_vma_is_active(vma));
if (vma->obj) {
struct drm_i915_gem_object *obj = vma->obj;
spin_lock(&obj->vma.lock);
list_del(&vma->obj_link);
rb_erase(&vma->obj_node, &vma->obj->vma.tree);
rb_erase(&vma->obj_node, &obj->vma.tree);
spin_unlock(&obj->vma.lock);
}
i915_active_fini(&vma->active);
i915_vma_free(vma);
}
void i915_vma_destroy(struct i915_vma *vma)
{
lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);
GEM_BUG_ON(i915_vma_is_pinned(vma));
__i915_vma_remove_closed(vma);
i915_vm_put(vma->vm);
WARN_ON(i915_vma_unbind(vma));
GEM_BUG_ON(i915_vma_is_active(vma));
__i915_vma_destroy(vma);
i915_active_fini(&vma->active);
i915_vma_free(vma);
}
void i915_vma_parked(struct drm_i915_private *i915)
@ -828,12 +1015,32 @@ void i915_vma_parked(struct drm_i915_private *i915)
spin_lock_irq(&i915->gt.closed_lock);
list_for_each_entry_safe(vma, next, &i915->gt.closed_vma, closed_link) {
list_del_init(&vma->closed_link);
struct drm_i915_gem_object *obj = vma->obj;
struct i915_address_space *vm = vma->vm;
/* XXX All to avoid keeping a reference on i915_vma itself */
if (!kref_get_unless_zero(&obj->base.refcount))
continue;
if (!i915_vm_tryopen(vm)) {
i915_gem_object_put(obj);
obj = NULL;
}
spin_unlock_irq(&i915->gt.closed_lock);
i915_vma_destroy(vma);
if (obj) {
i915_vma_destroy(vma);
i915_gem_object_put(obj);
}
i915_vm_close(vm);
/* Restart after dropping lock */
spin_lock_irq(&i915->gt.closed_lock);
next = list_first_entry(&i915->gt.closed_vma,
typeof(*next), closed_link);
}
spin_unlock_irq(&i915->gt.closed_lock);
}
@ -873,6 +1080,20 @@ void i915_vma_revoke_mmap(struct i915_vma *vma)
list_del(&vma->obj->userfault_link);
}
int __i915_vma_move_to_active(struct i915_vma *vma, struct i915_request *rq)
{
int err;
GEM_BUG_ON(!i915_vma_is_pinned(vma));
/* Wait for the vma to be bound before we start! */
err = i915_request_await_active(rq, &vma->active);
if (err)
return err;
return i915_active_add_request(&vma->active, rq);
}
int i915_vma_move_to_active(struct i915_vma *vma,
struct i915_request *rq,
unsigned int flags)
@ -880,19 +1101,9 @@ int i915_vma_move_to_active(struct i915_vma *vma,
struct drm_i915_gem_object *obj = vma->obj;
int err;
assert_vma_held(vma);
assert_object_held(obj);
GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
/*
* Add a reference if we're newly entering the active list.
* The order in which we add operations to the retirement queue is
* vital here: mark_active adds to the start of the callback list,
* such that subsequent callbacks are called first. Therefore we
* add the active reference first and queue for it to be dropped
* *last*.
*/
err = i915_active_add_request(&vma->active, rq);
err = __i915_vma_move_to_active(vma, rq);
if (unlikely(err))
return err;
@ -918,38 +1129,23 @@ int i915_vma_move_to_active(struct i915_vma *vma,
return 0;
}
int i915_vma_unbind(struct i915_vma *vma)
int __i915_vma_unbind(struct i915_vma *vma)
{
int ret;
lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);
lockdep_assert_held(&vma->vm->mutex);
/*
* First wait upon any activity as retiring the request may
* have side-effects such as unpinning or even unbinding this vma.
*
* XXX Actually waiting under the vm->mutex is a hinderance and
* should be pipelined wherever possible. In cases where that is
* unavoidable, we should lift the wait to before the mutex.
*/
might_sleep();
if (i915_vma_is_active(vma)) {
/*
* When a closed VMA is retired, it is unbound - eek.
* In order to prevent it from being recursively closed,
* take a pin on the vma so that the second unbind is
* aborted.
*
* Even more scary is that the retire callback may free
* the object (last active vma). To prevent the explosion
* we defer the actual object free to a worker that can
* only proceed once it acquires the struct_mutex (which
* we currently hold, therefore it cannot free this object
* before we are finished).
*/
__i915_vma_pin(vma);
ret = i915_active_wait(&vma->active);
__i915_vma_unpin(vma);
if (ret)
return ret;
}
GEM_BUG_ON(i915_vma_is_active(vma));
ret = i915_vma_sync(vma);
if (ret)
return ret;
if (i915_vma_is_pinned(vma)) {
vma_print_allocator(vma, "is pinned");
@ -970,16 +1166,12 @@ int i915_vma_unbind(struct i915_vma *vma)
GEM_BUG_ON(i915_vma_has_ggtt_write(vma));
/* release the fence reg _after_ flushing */
mutex_lock(&vma->vm->mutex);
ret = i915_vma_revoke_fence(vma);
mutex_unlock(&vma->vm->mutex);
if (ret)
return ret;
/* Force a pagefault for domain tracking on next user access */
mutex_lock(&vma->vm->mutex);
i915_vma_revoke_mmap(vma);
mutex_unlock(&vma->vm->mutex);
__i915_vma_iounmap(vma);
clear_bit(I915_VMA_CAN_FENCE_BIT, __i915_vma_flags(vma));
@ -987,17 +1179,33 @@ int i915_vma_unbind(struct i915_vma *vma)
GEM_BUG_ON(vma->fence);
GEM_BUG_ON(i915_vma_has_userfault(vma));
if (likely(!vma->vm->closed)) {
if (likely(atomic_read(&vma->vm->open))) {
trace_i915_vma_unbind(vma);
vma->ops->unbind_vma(vma);
}
atomic_and(~I915_VMA_BIND_MASK, &vma->flags);
atomic_and(~(I915_VMA_BIND_MASK | I915_VMA_ERROR), &vma->flags);
vma_unbind_pages(vma);
i915_vma_remove(vma);
return 0;
}
int i915_vma_unbind(struct i915_vma *vma)
{
struct i915_address_space *vm = vma->vm;
int err;
err = mutex_lock_interruptible(&vm->mutex);
if (err)
return err;
err = __i915_vma_unbind(vma);
mutex_unlock(&vm->mutex);
return err;
}
struct i915_vma *i915_vma_make_unshrinkable(struct i915_vma *vma)
{
i915_gem_object_make_unshrinkable(vma->obj);

84
drivers/gpu/drm/i915/i915_vma.h
View File

@ -96,25 +96,28 @@ struct i915_vma {
* exclusive cachelines of a single page, so a maximum of 64 possible
* users.
*/
#define I915_VMA_PIN_MASK 0xff
#define I915_VMA_PIN_OVERFLOW_BIT 8
#define I915_VMA_PIN_OVERFLOW ((int)BIT(I915_VMA_PIN_OVERFLOW_BIT))
#define I915_VMA_PIN_MASK 0x3ff
#define I915_VMA_OVERFLOW 0x200
/** Flags and address space this VMA is bound to */
#define I915_VMA_GLOBAL_BIND_BIT 9
#define I915_VMA_LOCAL_BIND_BIT 10
#define I915_VMA_GLOBAL_BIND_BIT 10
#define I915_VMA_LOCAL_BIND_BIT 11
#define I915_VMA_GLOBAL_BIND ((int)BIT(I915_VMA_GLOBAL_BIND_BIT))
#define I915_VMA_LOCAL_BIND ((int)BIT(I915_VMA_LOCAL_BIND_BIT))
#define I915_VMA_BIND_MASK (I915_VMA_GLOBAL_BIND | \
I915_VMA_LOCAL_BIND | \
I915_VMA_PIN_OVERFLOW)
#define I915_VMA_BIND_MASK (I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND)
#define I915_VMA_GGTT_BIT 11
#define I915_VMA_CAN_FENCE_BIT 12
#define I915_VMA_USERFAULT_BIT 13
#define I915_VMA_GGTT_WRITE_BIT 14
#define I915_VMA_ALLOC_BIT 12
#define I915_VMA_ALLOC ((int)BIT(I915_VMA_ALLOC_BIT))
#define I915_VMA_ERROR_BIT 13
#define I915_VMA_ERROR ((int)BIT(I915_VMA_ERROR_BIT))
#define I915_VMA_GGTT_BIT 14
#define I915_VMA_CAN_FENCE_BIT 15
#define I915_VMA_USERFAULT_BIT 16
#define I915_VMA_GGTT_WRITE_BIT 17
#define I915_VMA_GGTT ((int)BIT(I915_VMA_GGTT_BIT))
#define I915_VMA_CAN_FENCE ((int)BIT(I915_VMA_CAN_FENCE_BIT))
@ -123,6 +126,11 @@ struct i915_vma {
struct i915_active active;
#define I915_VMA_PAGES_BIAS 24
#define I915_VMA_PAGES_ACTIVE (BIT(24) | 1)
atomic_t pages_count; /* number of active binds to the pages */
struct mutex pages_mutex; /* protect acquire/release of backing pages */
/**
* Support different GGTT views into the same object.
* This means there can be multiple VMA mappings per object and per VM.
@ -169,6 +177,8 @@ static inline bool i915_vma_is_active(const struct i915_vma *vma)
return !i915_active_is_idle(&vma->active);
}
int __must_check __i915_vma_move_to_active(struct i915_vma *vma,
struct i915_request *rq);
int __must_check i915_vma_move_to_active(struct i915_vma *vma,
struct i915_request *rq,
unsigned int flags);
@ -307,13 +317,18 @@ i915_vma_compare(struct i915_vma *vma,
return memcmp(&vma->ggtt_view.partial, &view->partial, view->type);
}
int i915_vma_bind(struct i915_vma *vma, enum i915_cache_level cache_level,
u32 flags);
struct i915_vma_work *i915_vma_work(void);
int i915_vma_bind(struct i915_vma *vma,
enum i915_cache_level cache_level,
u32 flags,
struct i915_vma_work *work);
bool i915_gem_valid_gtt_space(struct i915_vma *vma, unsigned long color);
bool i915_vma_misplaced(const struct i915_vma *vma,
u64 size, u64 alignment, u64 flags);
void __i915_vma_set_map_and_fenceable(struct i915_vma *vma);
void i915_vma_revoke_mmap(struct i915_vma *vma);
int __i915_vma_unbind(struct i915_vma *vma);
int __must_check i915_vma_unbind(struct i915_vma *vma);
void i915_vma_unlink_ctx(struct i915_vma *vma);
void i915_vma_close(struct i915_vma *vma);
@ -332,26 +347,8 @@ static inline void i915_vma_unlock(struct i915_vma *vma)
dma_resv_unlock(vma->resv);
}
int __i915_vma_do_pin(struct i915_vma *vma,
u64 size, u64 alignment, u64 flags);
static inline int __must_check
i915_vma_pin(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
{
BUILD_BUG_ON(PIN_MBZ != I915_VMA_PIN_OVERFLOW);
BUILD_BUG_ON(PIN_GLOBAL != I915_VMA_GLOBAL_BIND);
BUILD_BUG_ON(PIN_USER != I915_VMA_LOCAL_BIND);
/* Pin early to prevent the shrinker/eviction logic from destroying
* our vma as we insert and bind.
*/
if (likely(((atomic_inc_return(&vma->flags) ^ flags) & I915_VMA_BIND_MASK) == 0)) {
GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
GEM_BUG_ON(i915_vma_misplaced(vma, size, alignment, flags));
return 0;
}
return __i915_vma_do_pin(vma, size, alignment, flags);
}
int __must_check
i915_vma_pin(struct i915_vma *vma, u64 size, u64 alignment, u64 flags);
static inline int i915_vma_pin_count(const struct i915_vma *vma)
{
@ -366,17 +363,17 @@ static inline bool i915_vma_is_pinned(const struct i915_vma *vma)
static inline void __i915_vma_pin(struct i915_vma *vma)
{
atomic_inc(&vma->flags);
GEM_BUG_ON(atomic_read(&vma->flags) & I915_VMA_PIN_OVERFLOW);
GEM_BUG_ON(!i915_vma_is_pinned(vma));
}
static inline void __i915_vma_unpin(struct i915_vma *vma)
{
GEM_BUG_ON(!i915_vma_is_pinned(vma));
atomic_dec(&vma->flags);
}
static inline void i915_vma_unpin(struct i915_vma *vma)
{
GEM_BUG_ON(!i915_vma_is_pinned(vma));
GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
__i915_vma_unpin(vma);
}
@ -402,8 +399,6 @@ static inline bool i915_node_color_differs(const struct drm_mm_node *node,
* the caller must call i915_vma_unpin_iomap to relinquish the pinning
* after the iomapping is no longer required.
*
* Callers must hold the struct_mutex.
*
* Returns a valid iomapped pointer or ERR_PTR.
*/
void __iomem *i915_vma_pin_iomap(struct i915_vma *vma);
@ -415,8 +410,8 @@ void __iomem *i915_vma_pin_iomap(struct i915_vma *vma);
*
* Unpins the previously iomapped VMA from i915_vma_pin_iomap().
*
* Callers must hold the struct_mutex. This function is only valid to be
* called on a VMA previously iomapped by the caller with i915_vma_pin_iomap().
* This function is only valid to be called on a VMA previously
* iomapped by the caller with i915_vma_pin_iomap().
*/
void i915_vma_unpin_iomap(struct i915_vma *vma);
@ -444,6 +439,8 @@ static inline struct page *i915_vma_first_page(struct i915_vma *vma)
int __must_check i915_vma_pin_fence(struct i915_vma *vma);
int __must_check i915_vma_revoke_fence(struct i915_vma *vma);
int __i915_vma_pin_fence(struct i915_vma *vma);
static inline void __i915_vma_unpin_fence(struct i915_vma *vma)
{
GEM_BUG_ON(atomic_read(&vma->fence->pin_count) <= 0);
@ -461,7 +458,6 @@ static inline void __i915_vma_unpin_fence(struct i915_vma *vma)
static inline void
i915_vma_unpin_fence(struct i915_vma *vma)
{
/* lockdep_assert_held(&vma->vm->i915->drm.struct_mutex); */
if (vma->fence)
__i915_vma_unpin_fence(vma);
}
@ -490,4 +486,10 @@ struct i915_vma *i915_vma_make_unshrinkable(struct i915_vma *vma);
void i915_vma_make_shrinkable(struct i915_vma *vma);
void i915_vma_make_purgeable(struct i915_vma *vma);
static inline int i915_vma_sync(struct i915_vma *vma)
{
/* Wait for the asynchronous bindings and pending GPU reads */
return i915_active_wait(&vma->active);
}
#endif

2
drivers/gpu/drm/i915/selftests/i915_gem.c
View File

@ -119,10 +119,8 @@ static void pm_resume(struct drm_i915_private *i915)
intel_gt_sanitize(&i915->gt, false);
i915_gem_sanitize(i915);
mutex_lock(&i915->drm.struct_mutex);
i915_gem_restore_gtt_mappings(i915);
i915_gem_restore_fences(i915);
mutex_unlock(&i915->drm.struct_mutex);
i915_gem_resume(i915);
}

36
drivers/gpu/drm/i915/selftests/i915_gem_evict.c
View File

@ -106,14 +106,11 @@ static int populate_ggtt(struct drm_i915_private *i915,
static void unpin_ggtt(struct drm_i915_private *i915)
{
struct i915_ggtt *ggtt = &i915->ggtt;
struct i915_vma *vma;
mutex_lock(&ggtt->vm.mutex);
list_for_each_entry(vma, &i915->ggtt.vm.bound_list, vm_link)
if (vma->obj->mm.quirked)
i915_vma_unpin(vma);
mutex_unlock(&ggtt->vm.mutex);
}
static void cleanup_objects(struct drm_i915_private *i915,
@ -127,11 +124,7 @@ static void cleanup_objects(struct drm_i915_private *i915,
i915_gem_object_put(obj);
}
mutex_unlock(&i915->drm.struct_mutex);
i915_gem_drain_freed_objects(i915);
mutex_lock(&i915->drm.struct_mutex);
}
static int igt_evict_something(void *arg)
@ -148,10 +141,12 @@ static int igt_evict_something(void *arg)
goto cleanup;
/* Everything is pinned, nothing should happen */
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_evict_something(&ggtt->vm,
I915_GTT_PAGE_SIZE, 0, 0,
0, U64_MAX,
0);
mutex_unlock(&ggtt->vm.mutex);
if (err != -ENOSPC) {
pr_err("i915_gem_evict_something failed on a full GGTT with err=%d\n",
err);
@ -161,10 +156,12 @@ static int igt_evict_something(void *arg)
unpin_ggtt(i915);
/* Everything is unpinned, we should be able to evict something */
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_evict_something(&ggtt->vm,
I915_GTT_PAGE_SIZE, 0, 0,
0, U64_MAX,
0);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("i915_gem_evict_something failed on a full GGTT with err=%d\n",
err);
@ -230,7 +227,9 @@ static int igt_evict_for_vma(void *arg)
goto cleanup;
/* Everything is pinned, nothing should happen */
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_evict_for_node(&ggtt->vm, &target, 0);
mutex_unlock(&ggtt->vm.mutex);
if (err != -ENOSPC) {
pr_err("i915_gem_evict_for_node on a full GGTT returned err=%d\n",
err);
@ -240,7 +239,9 @@ static int igt_evict_for_vma(void *arg)
unpin_ggtt(i915);
/* Everything is unpinned, we should be able to evict the node */
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_evict_for_node(&ggtt->vm, &target, 0);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("i915_gem_evict_for_node returned err=%d\n",
err);
@ -319,7 +320,9 @@ static int igt_evict_for_cache_color(void *arg)
i915_vma_unpin(vma);
/* Remove just the second vma */
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_evict_for_node(&ggtt->vm, &target, 0);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("[0]i915_gem_evict_for_node returned err=%d\n", err);
goto cleanup;
@ -330,7 +333,9 @@ static int igt_evict_for_cache_color(void *arg)
*/
target.color = I915_CACHE_L3_LLC;
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_evict_for_node(&ggtt->vm, &target, 0);
mutex_unlock(&ggtt->vm.mutex);
if (!err) {
pr_err("[1]i915_gem_evict_for_node returned err=%d\n", err);
err = -EINVAL;
@ -360,7 +365,9 @@ static int igt_evict_vm(void *arg)
goto cleanup;
/* Everything is pinned, nothing should happen */
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_evict_vm(&ggtt->vm);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("i915_gem_evict_vm on a full GGTT returned err=%d]\n",
err);
@ -369,7 +376,9 @@ static int igt_evict_vm(void *arg)
unpin_ggtt(i915);
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_evict_vm(&ggtt->vm);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("i915_gem_evict_vm on a full GGTT returned err=%d]\n",
err);
@ -410,11 +419,11 @@ static int igt_evict_contexts(void *arg)
if (!HAS_FULL_PPGTT(i915))
return 0;
mutex_lock(&i915->drm.struct_mutex);
wakeref = intel_runtime_pm_get(&i915->runtime_pm);
/* Reserve a block so that we know we have enough to fit a few rq */
memset(&hole, 0, sizeof(hole));
mutex_lock(&i915->ggtt.vm.mutex);
err = i915_gem_gtt_insert(&i915->ggtt.vm, &hole,
PRETEND_GGTT_SIZE, 0, I915_COLOR_UNEVICTABLE,
0, i915->ggtt.vm.total,
@ -427,7 +436,9 @@ static int igt_evict_contexts(void *arg)
do {
struct reserved *r;
mutex_unlock(&i915->ggtt.vm.mutex);
r = kcalloc(1, sizeof(*r), GFP_KERNEL);
mutex_lock(&i915->ggtt.vm.mutex);
if (!r) {
err = -ENOMEM;
goto out_locked;
@ -447,7 +458,7 @@ static int igt_evict_contexts(void *arg)
count++;
} while (1);
drm_mm_remove_node(&hole);
mutex_unlock(&i915->drm.struct_mutex);
mutex_unlock(&i915->ggtt.vm.mutex);
pr_info("Filled GGTT with %lu 1MiB nodes\n", count);
/* Overfill the GGTT with context objects and so try to evict one. */
@ -510,7 +521,7 @@ static int igt_evict_contexts(void *arg)
break;
}
mutex_lock(&i915->drm.struct_mutex);
mutex_lock(&i915->ggtt.vm.mutex);
out_locked:
if (igt_flush_test(i915, I915_WAIT_LOCKED))
err = -EIO;
@ -524,8 +535,8 @@ out_locked:
}
if (drm_mm_node_allocated(&hole))
drm_mm_remove_node(&hole);
mutex_unlock(&i915->ggtt.vm.mutex);
intel_runtime_pm_put(&i915->runtime_pm, wakeref);
mutex_unlock(&i915->drm.struct_mutex);
return err;
}
@ -547,12 +558,9 @@ int i915_gem_evict_mock_selftests(void)
if (!i915)
return -ENOMEM;
mutex_lock(&i915->drm.struct_mutex);
with_intel_runtime_pm(&i915->runtime_pm, wakeref)
err = i915_subtests(tests, i915);
mutex_unlock(&i915->drm.struct_mutex);
drm_dev_put(&i915->drm);
return err;
}

58
drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
View File

@ -38,16 +38,7 @@
static void cleanup_freed_objects(struct drm_i915_private *i915)
{
/*
* As we may hold onto the struct_mutex for inordinate lengths of
* time, the NMI khungtaskd detector may fire for the free objects
* worker.
*/
mutex_unlock(&i915->drm.struct_mutex);
i915_gem_drain_freed_objects(i915);
mutex_lock(&i915->drm.struct_mutex);
}
static void fake_free_pages(struct drm_i915_gem_object *obj,
@ -880,6 +871,15 @@ static int __shrink_hole(struct drm_i915_private *i915,
i915_vma_unpin(vma);
addr += size;
/*
* Since we are injecting allocation faults at random intervals,
* wait for this allocation to complete before we change the
* faultinjection.
*/
err = i915_vma_sync(vma);
if (err)
break;
if (igt_timeout(end_time,
"%s timed out at ofset %llx [%llx - %llx]\n",
__func__, addr, hole_start, hole_end)) {
@ -1013,21 +1013,19 @@ static int exercise_ppgtt(struct drm_i915_private *dev_priv,
if (IS_ERR(file))
return PTR_ERR(file);
mutex_lock(&dev_priv->drm.struct_mutex);
ppgtt = i915_ppgtt_create(dev_priv);
if (IS_ERR(ppgtt)) {
err = PTR_ERR(ppgtt);
goto out_unlock;
goto out_free;
}
GEM_BUG_ON(offset_in_page(ppgtt->vm.total));
GEM_BUG_ON(ppgtt->vm.closed);
GEM_BUG_ON(!atomic_read(&ppgtt->vm.open));
err = func(dev_priv, &ppgtt->vm, 0, ppgtt->vm.total, end_time);
i915_vm_put(&ppgtt->vm);
out_unlock:
mutex_unlock(&dev_priv->drm.struct_mutex);
out_free:
mock_file_free(dev_priv, file);
return err;
}
@ -1090,7 +1088,6 @@ static int exercise_ggtt(struct drm_i915_private *i915,
IGT_TIMEOUT(end_time);
int err = 0;
mutex_lock(&i915->drm.struct_mutex);
restart:
list_sort(NULL, &ggtt->vm.mm.hole_stack, sort_holes);
drm_mm_for_each_hole(node, &ggtt->vm.mm, hole_start, hole_end) {
@ -1111,7 +1108,6 @@ restart:
last = hole_end;
goto restart;
}
mutex_unlock(&i915->drm.struct_mutex);
return err;
}
@ -1153,13 +1149,9 @@ static int igt_ggtt_page(void *arg)
unsigned int *order, n;
int err;
mutex_lock(&i915->drm.struct_mutex);
obj = i915_gem_object_create_internal(i915, PAGE_SIZE);
if (IS_ERR(obj)) {
err = PTR_ERR(obj);
goto out_unlock;
}
if (IS_ERR(obj))
return PTR_ERR(obj);
err = i915_gem_object_pin_pages(obj);
if (err)
@ -1227,8 +1219,6 @@ out_unpin:
i915_gem_object_unpin_pages(obj);
out_free:
i915_gem_object_put(obj);
out_unlock:
mutex_unlock(&i915->drm.struct_mutex);
return err;
}
@ -1239,6 +1229,9 @@ static void track_vma_bind(struct i915_vma *vma)
atomic_inc(&obj->bind_count); /* track for eviction later */
__i915_gem_object_pin_pages(obj);
GEM_BUG_ON(vma->pages);
atomic_set(&vma->pages_count, I915_VMA_PAGES_ACTIVE);
__i915_gem_object_pin_pages(obj);
vma->pages = obj->mm.pages;
mutex_lock(&vma->vm->mutex);
@ -1336,11 +1329,13 @@ static int igt_gtt_reserve(void *arg)
goto out;
}
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_gtt_reserve(&ggtt->vm, &vma->node,
obj->base.size,
total,
obj->cache_level,
0);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("i915_gem_gtt_reserve (pass 1) failed at %llu/%llu with err=%d\n",
total, ggtt->vm.total, err);
@ -1386,11 +1381,13 @@ static int igt_gtt_reserve(void *arg)
goto out;
}
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_gtt_reserve(&ggtt->vm, &vma->node,
obj->base.size,
total,
obj->cache_level,
0);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("i915_gem_gtt_reserve (pass 2) failed at %llu/%llu with err=%d\n",
total, ggtt->vm.total, err);
@ -1431,11 +1428,13 @@ static int igt_gtt_reserve(void *arg)
2 * I915_GTT_PAGE_SIZE,
I915_GTT_MIN_ALIGNMENT);
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_gtt_reserve(&ggtt->vm, &vma->node,
obj->base.size,
offset,
obj->cache_level,
0);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("i915_gem_gtt_reserve (pass 3) failed at %llu/%llu with err=%d\n",
total, ggtt->vm.total, err);
@ -1504,11 +1503,13 @@ static int igt_gtt_insert(void *arg)
/* Check a couple of obviously invalid requests */
for (ii = invalid_insert; ii->size; ii++) {
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_gtt_insert(&ggtt->vm, &tmp,
ii->size, ii->alignment,
I915_COLOR_UNEVICTABLE,
ii->start, ii->end,
0);
mutex_unlock(&ggtt->vm.mutex);
if (err != -ENOSPC) {
pr_err("Invalid i915_gem_gtt_insert(.size=%llx, .alignment=%llx, .start=%llx, .end=%llx) succeeded (err=%d)\n",
ii->size, ii->alignment, ii->start, ii->end,
@ -1544,10 +1545,12 @@ static int igt_gtt_insert(void *arg)
goto out;
}
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_gtt_insert(&ggtt->vm, &vma->node,
obj->base.size, 0, obj->cache_level,
0, ggtt->vm.total,
0);
mutex_unlock(&ggtt->vm.mutex);
if (err == -ENOSPC) {
/* maxed out the GGTT space */
i915_gem_object_put(obj);
@ -1602,10 +1605,12 @@ static int igt_gtt_insert(void *arg)
goto out;
}
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_gtt_insert(&ggtt->vm, &vma->node,
obj->base.size, 0, obj->cache_level,
0, ggtt->vm.total,
0);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("i915_gem_gtt_insert (pass 2) failed at %llu/%llu with err=%d\n",
total, ggtt->vm.total, err);
@ -1649,10 +1654,12 @@ static int igt_gtt_insert(void *arg)
goto out;
}
mutex_lock(&ggtt->vm.mutex);
err = i915_gem_gtt_insert(&ggtt->vm, &vma->node,
obj->base.size, 0, obj->cache_level,
0, ggtt->vm.total,
0);
mutex_unlock(&ggtt->vm.mutex);
if (err) {
pr_err("i915_gem_gtt_insert (pass 3) failed at %llu/%llu with err=%d\n",
total, ggtt->vm.total, err);
@ -1696,8 +1703,9 @@ int i915_gem_gtt_mock_selftests(void)
}
mock_init_ggtt(i915, ggtt);
mutex_lock(&i915->drm.struct_mutex);
err = i915_subtests(tests, ggtt);
mutex_lock(&i915->drm.struct_mutex);
mock_device_flush(i915);
mutex_unlock(&i915->drm.struct_mutex);

7
drivers/gpu/drm/i915/selftests/i915_request.c
View File

@ -647,8 +647,15 @@ static struct i915_vma *empty_batch(struct drm_i915_private *i915)
if (err)
goto err;
/* Force the wait wait now to avoid including it in the benchmark */
err = i915_vma_sync(vma);
if (err)
goto err_pin;
return vma;
err_pin:
i915_vma_unpin(vma);
err:
i915_gem_object_put(obj);
return ERR_PTR(err);

6
drivers/gpu/drm/i915/selftests/i915_vma.c
View File

@ -831,8 +831,9 @@ int i915_vma_mock_selftests(void)
}
mock_init_ggtt(i915, ggtt);
mutex_lock(&i915->drm.struct_mutex);
err = i915_subtests(tests, ggtt);
mutex_lock(&i915->drm.struct_mutex);
mock_device_flush(i915);
mutex_unlock(&i915->drm.struct_mutex);
@ -879,8 +880,6 @@ static int igt_vma_remapped_gtt(void *arg)
if (IS_ERR(obj))
return PTR_ERR(obj);
mutex_lock(&i915->drm.struct_mutex);
wakeref = intel_runtime_pm_get(&i915->runtime_pm);
for (t = types; *t; t++) {
@ -976,7 +975,6 @@ static int igt_vma_remapped_gtt(void *arg)
out:
intel_runtime_pm_put(&i915->runtime_pm, wakeref);
mutex_unlock(&i915->drm.struct_mutex);
i915_gem_object_put(obj);
return err;