Planning to merge this through the drm/vmwgfx tree soon, so if there
are any objections, please speak up.
Graphics APIs like OpenGL 4.4 and Vulkan require the graphics driver
to provide coherent graphics memory, meaning that the GPU sees any
content written to the coherent memory on the next GPU operation that
touches that memory, and the CPU sees any content written by the GPU
to that memory immediately after any fence object trailing the GPU
operation has signaled.
Paravirtual drivers that otherwise require explicit synchronization
needs to do this by hooking up dirty tracking to pagefault handlers
and buffer object validation. This is a first attempt to do that for
the vmwgfx driver.
The mm patches has been out for RFC. I think I have addressed all the
feedback I got, except a possible softdirty breakage. But although the
dirty-tracking and softdirty may write-protect PTEs both care about,
that shouldn't really cause any operation interference. In particular
since we use the hardware dirty PTE bits and softdirty uses other PTE bits.
For the TTM changes they are hopefully in line with the long-term
strategy of making helpers out of what's left of TTM.
The code has been tested and excercised by a tailored version of mesa
where we disable all explicit synchronization and assume graphics memory
is coherent. The performance loss varies of course; a typical number is
around 5%.
Changes v1-v2:
- Addressed a number of typos and formatting issues.
- Added a usage warning for apply_to_pfn_range() and apply_to_page_range()
- Re-evaluated the decision to use apply_to_pfn_range() rather than
modifying the pagewalk.c. It still looks like generically handling the
transparent huge page cases requires the mmap_sem to be held at least
in read mode, so sticking with apply_to_pfn_range() for now.
- The TTM page-fault helper vma copy argument was scratched in favour of
a pageprot_t argument.
Changes v3:
- Adapted to upstream API changes.
Changes v4:
- Adapted to upstream mmu_notifier changes. (Jerome?)
- Fixed a couple of warnings on 32-bit x86
- Fixed image offset computation on multisample images.
Cc: Andrew Morton <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Minchan Kim <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Huang Ying <[email protected]>
Cc: Souptick Joarder <[email protected]>
Cc: "Jérôme Glisse" <[email protected]>
Cc: "Christian König" <[email protected]>
Cc: [email protected]
From: Thomas Hellstrom <[email protected]>
Driver fault callbacks are allowed to drop the mmap_sem when expecting
long hardware waits to avoid blocking other mm users. Allow the mkwrite
callbacks to do the same by returning early on VM_FAULT_RETRY.
In particular we want to be able to drop the mmap_sem when waiting for
a reservation object lock on a GPU buffer object. These locks may be
held while waiting for the GPU.
Cc: Andrew Morton <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Minchan Kim <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Huang Ying <[email protected]>
Cc: Souptick Joarder <[email protected]>
Cc: "Jérôme Glisse" <[email protected]>
Cc: [email protected]
Cc: [email protected]
Signed-off-by: Thomas Hellstrom <[email protected]>
Reviewed-by: Ralph Campbell <[email protected]>
---
mm/memory.c | 10 ++++++----
1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/mm/memory.c b/mm/memory.c
index ddf20bd0c317..168f546af1ad 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -2238,7 +2238,7 @@ static vm_fault_t do_page_mkwrite(struct vm_fault *vmf)
ret = vmf->vma->vm_ops->page_mkwrite(vmf);
/* Restore original flags so that caller is not surprised */
vmf->flags = old_flags;
- if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))
+ if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE | VM_FAULT_RETRY)))
return ret;
if (unlikely(!(ret & VM_FAULT_LOCKED))) {
lock_page(page);
@@ -2515,7 +2515,7 @@ static vm_fault_t wp_pfn_shared(struct vm_fault *vmf)
pte_unmap_unlock(vmf->pte, vmf->ptl);
vmf->flags |= FAULT_FLAG_MKWRITE;
ret = vma->vm_ops->pfn_mkwrite(vmf);
- if (ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE))
+ if (ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE | VM_FAULT_RETRY))
return ret;
return finish_mkwrite_fault(vmf);
}
@@ -2536,7 +2536,8 @@ static vm_fault_t wp_page_shared(struct vm_fault *vmf)
pte_unmap_unlock(vmf->pte, vmf->ptl);
tmp = do_page_mkwrite(vmf);
if (unlikely(!tmp || (tmp &
- (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))) {
+ (VM_FAULT_ERROR | VM_FAULT_NOPAGE |
+ VM_FAULT_RETRY)))) {
put_page(vmf->page);
return tmp;
}
@@ -3601,7 +3602,8 @@ static vm_fault_t do_shared_fault(struct vm_fault *vmf)
unlock_page(vmf->page);
tmp = do_page_mkwrite(vmf);
if (unlikely(!tmp ||
- (tmp & (VM_FAULT_ERROR | VM_FAULT_NOPAGE)))) {
+ (tmp & (VM_FAULT_ERROR | VM_FAULT_NOPAGE |
+ VM_FAULT_RETRY)))) {
put_page(vmf->page);
return tmp;
}
--
2.20.1
From: Thomas Hellstrom <[email protected]>
Similar to write-coherent resources, make sure that from the user-space
point of view, GPU rendered contents is automatically available for
reading by the CPU.
Signed-off-by: Thomas Hellstrom <[email protected]>
Reviewed-by: Deepak Rawat <[email protected]>
---
drivers/gpu/drm/vmwgfx/vmwgfx_drv.h | 7 +-
drivers/gpu/drm/vmwgfx/vmwgfx_page_dirty.c | 73 ++++++++++++-
drivers/gpu/drm/vmwgfx/vmwgfx_resource.c | 103 +++++++++++++++++-
drivers/gpu/drm/vmwgfx/vmwgfx_resource_priv.h | 2 +
drivers/gpu/drm/vmwgfx/vmwgfx_validation.c | 3 +-
5 files changed, 177 insertions(+), 11 deletions(-)
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
index 9b347923196f..dae3a39bf402 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
@@ -689,7 +689,8 @@ extern void vmw_resource_unreference(struct vmw_resource **p_res);
extern struct vmw_resource *vmw_resource_reference(struct vmw_resource *res);
extern struct vmw_resource *
vmw_resource_reference_unless_doomed(struct vmw_resource *res);
-extern int vmw_resource_validate(struct vmw_resource *res, bool intr);
+extern int vmw_resource_validate(struct vmw_resource *res, bool intr,
+ bool dirtying);
extern int vmw_resource_reserve(struct vmw_resource *res, bool interruptible,
bool no_backup);
extern bool vmw_resource_needs_backup(const struct vmw_resource *res);
@@ -733,6 +734,8 @@ void vmw_resource_mob_attach(struct vmw_resource *res);
void vmw_resource_mob_detach(struct vmw_resource *res);
void vmw_resource_dirty_update(struct vmw_resource *res, pgoff_t start,
pgoff_t end);
+int vmw_resources_clean(struct vmw_buffer_object *vbo, pgoff_t start,
+ pgoff_t end, pgoff_t *num_prefault);
/**
* vmw_resource_mob_attached - Whether a resource currently has a mob attached
@@ -1427,6 +1430,8 @@ int vmw_bo_dirty_add(struct vmw_buffer_object *vbo);
void vmw_bo_dirty_transfer_to_res(struct vmw_resource *res);
void vmw_bo_dirty_clear_res(struct vmw_resource *res);
void vmw_bo_dirty_release(struct vmw_buffer_object *vbo);
+void vmw_bo_dirty_unmap(struct vmw_buffer_object *vbo,
+ pgoff_t start, pgoff_t end);
vm_fault_t vmw_bo_vm_fault(struct vm_fault *vmf);
vm_fault_t vmw_bo_vm_mkwrite(struct vm_fault *vmf);
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_page_dirty.c b/drivers/gpu/drm/vmwgfx/vmwgfx_page_dirty.c
index 8d154f90bdc0..730c51e397dd 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_page_dirty.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_page_dirty.c
@@ -153,7 +153,6 @@ static void vmw_bo_dirty_scan_mkwrite(struct vmw_buffer_object *vbo)
}
}
-
/**
* vmw_bo_dirty_scan - Scan for dirty pages and add them to the dirty
* tracking structure
@@ -171,6 +170,51 @@ void vmw_bo_dirty_scan(struct vmw_buffer_object *vbo)
vmw_bo_dirty_scan_mkwrite(vbo);
}
+/**
+ * vmw_bo_dirty_pre_unmap - write-protect and pick up dirty pages before
+ * an unmap_mapping_range operation.
+ * @vbo: The buffer object,
+ * @start: First page of the range within the buffer object.
+ * @end: Last page of the range within the buffer object + 1.
+ *
+ * If we're using the _PAGETABLE scan method, we may leak dirty pages
+ * when calling unmap_mapping_range(). This function makes sure we pick
+ * up all dirty pages.
+ */
+static void vmw_bo_dirty_pre_unmap(struct vmw_buffer_object *vbo,
+ pgoff_t start, pgoff_t end)
+{
+ struct vmw_bo_dirty *dirty = vbo->dirty;
+ unsigned long offset = drm_vma_node_start(&vbo->base.vma_node);
+ struct address_space *mapping = vbo->base.bdev->dev_mapping;
+
+ if (dirty->method != VMW_BO_DIRTY_PAGETABLE || start >= end)
+ return;
+
+ apply_as_wrprotect(mapping, start + offset, end - start);
+ apply_as_clean(mapping, start + offset, end - start, offset,
+ &dirty->bitmap[0], &dirty->start, &dirty->end);
+}
+
+/**
+ * vmw_bo_dirty_unmap - Clear all ptes pointing to a range within a bo
+ * @vbo: The buffer object,
+ * @start: First page of the range within the buffer object.
+ * @end: Last page of the range within the buffer object + 1.
+ *
+ * This is similar to ttm_bo_unmap_virtual_locked() except it takes a subrange.
+ */
+void vmw_bo_dirty_unmap(struct vmw_buffer_object *vbo,
+ pgoff_t start, pgoff_t end)
+{
+ unsigned long offset = drm_vma_node_start(&vbo->base.vma_node);
+ struct address_space *mapping = vbo->base.bdev->dev_mapping;
+
+ vmw_bo_dirty_pre_unmap(vbo, start, end);
+ unmap_shared_mapping_range(mapping, (offset + start) << PAGE_SHIFT,
+ (loff_t) (end - start) << PAGE_SHIFT);
+}
+
/**
* vmw_bo_dirty_add - Add a dirty-tracking user to a buffer object
* @vbo: The buffer object
@@ -389,21 +433,40 @@ vm_fault_t vmw_bo_vm_fault(struct vm_fault *vmf)
if (ret)
return ret;
+ num_prefault = (vma->vm_flags & VM_RAND_READ) ? 1 :
+ TTM_BO_VM_NUM_PREFAULT;
+
+ if (vbo->dirty) {
+ pgoff_t allowed_prefault;
+ unsigned long page_offset;
+
+ page_offset = vmf->pgoff - drm_vma_node_start(&bo->vma_node);
+ if (page_offset >= bo->num_pages ||
+ vmw_resources_clean(vbo, page_offset,
+ page_offset + PAGE_SIZE,
+ &allowed_prefault)) {
+ ret = VM_FAULT_SIGBUS;
+ goto out_unlock;
+ }
+
+ num_prefault = min(num_prefault, allowed_prefault);
+ }
+
/*
- * This will cause mkwrite() to be called for each pte on
- * write-enable vmas.
+ * If we don't track dirty using the MKWRITE method, make sure
+ * sure the page protection is write-enabled so we don't get
+ * a lot of unnecessary write faults.
*/
if (vbo->dirty && vbo->dirty->method == VMW_BO_DIRTY_MKWRITE)
prot = vma->vm_page_prot;
else
prot = vm_get_page_prot(vma->vm_flags);
- num_prefault = (vma->vm_flags & VM_RAND_READ) ? 0 :
- TTM_BO_VM_NUM_PREFAULT;
ret = ttm_bo_vm_fault_reserved(vmf, prot, num_prefault);
if (ret == VM_FAULT_RETRY && !(vmf->flags & FAULT_FLAG_RETRY_NOWAIT))
return ret;
+out_unlock:
reservation_object_unlock(bo->resv);
return ret;
}
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
index b84dd5953886..d70ee0df5c13 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
@@ -395,7 +395,8 @@ static int vmw_resource_buf_alloc(struct vmw_resource *res,
* should be retried once resources have been freed up.
*/
static int vmw_resource_do_validate(struct vmw_resource *res,
- struct ttm_validate_buffer *val_buf)
+ struct ttm_validate_buffer *val_buf,
+ bool dirtying)
{
int ret = 0;
const struct vmw_res_func *func = res->func;
@@ -437,6 +438,15 @@ static int vmw_resource_do_validate(struct vmw_resource *res,
* the resource.
*/
if (res->dirty) {
+ if (dirtying && !res->res_dirty) {
+ pgoff_t start = res->backup_offset >> PAGE_SHIFT;
+ pgoff_t end = __KERNEL_DIV_ROUND_UP
+ (res->backup_offset + res->backup_size,
+ PAGE_SIZE);
+
+ vmw_bo_dirty_unmap(res->backup, start, end);
+ }
+
vmw_bo_dirty_transfer_to_res(res);
return func->dirty_sync(res);
}
@@ -681,6 +691,7 @@ static int vmw_resource_do_evict(struct ww_acquire_ctx *ticket,
* to the device.
* @res: The resource to make visible to the device.
* @intr: Perform waits interruptible if possible.
+ * @dirtying: Pending GPU operation will dirty the resource
*
* On succesful return, any backup DMA buffer pointed to by @res->backup will
* be reserved and validated.
@@ -690,7 +701,8 @@ static int vmw_resource_do_evict(struct ww_acquire_ctx *ticket,
* Return: Zero on success, -ERESTARTSYS if interrupted, negative error code
* on failure.
*/
-int vmw_resource_validate(struct vmw_resource *res, bool intr)
+int vmw_resource_validate(struct vmw_resource *res, bool intr,
+ bool dirtying)
{
int ret;
struct vmw_resource *evict_res;
@@ -707,7 +719,7 @@ int vmw_resource_validate(struct vmw_resource *res, bool intr)
if (res->backup)
val_buf.bo = &res->backup->base;
do {
- ret = vmw_resource_do_validate(res, &val_buf);
+ ret = vmw_resource_do_validate(res, &val_buf, dirtying);
if (likely(ret != -EBUSY))
break;
@@ -1007,7 +1019,7 @@ int vmw_resource_pin(struct vmw_resource *res, bool interruptible)
/* Do we really need to pin the MOB as well? */
vmw_bo_pin_reserved(vbo, true);
}
- ret = vmw_resource_validate(res, interruptible);
+ ret = vmw_resource_validate(res, interruptible, true);
if (vbo)
ttm_bo_unreserve(&vbo->base);
if (ret)
@@ -1082,3 +1094,86 @@ void vmw_resource_dirty_update(struct vmw_resource *res, pgoff_t start,
res->func->dirty_range_add(res, start << PAGE_SHIFT,
end << PAGE_SHIFT);
}
+
+/**
+ * vmw_resources_clean - Clean resources intersecting a mob range
+ * @vbo: The mob buffer object
+ * @start: The mob page offset starting the range
+ * @end: The mob page offset ending the range
+ * @num_prefault: Returns how many pages including the first have been
+ * cleaned and are ok to prefault
+ */
+int vmw_resources_clean(struct vmw_buffer_object *vbo, pgoff_t start,
+ pgoff_t end, pgoff_t *num_prefault)
+{
+ struct rb_node *cur = vbo->res_tree.rb_node;
+ struct vmw_resource *found = NULL;
+ unsigned long res_start = start << PAGE_SHIFT;
+ unsigned long res_end = end << PAGE_SHIFT;
+ unsigned long last_cleaned = 0;
+
+ /*
+ * Find the resource with lowest backup_offset that intersects the
+ * range.
+ */
+ while (cur) {
+ struct vmw_resource *cur_res =
+ container_of(cur, struct vmw_resource, mob_node);
+
+ if (cur_res->backup_offset >= res_end) {
+ cur = cur->rb_left;
+ } else if (cur_res->backup_offset + cur_res->backup_size <=
+ res_start) {
+ cur = cur->rb_right;
+ } else {
+ found = cur_res;
+ cur = cur->rb_left;
+ /* Continue to look for resources with lower offsets */
+ }
+ }
+
+ /*
+ * In order of increasing backup_offset, clean dirty resorces
+ * intersecting the range.
+ */
+ while (found) {
+ if (found->res_dirty) {
+ int ret;
+
+ if (!found->func->clean)
+ return -EINVAL;
+
+ ret = found->func->clean(found);
+ if (ret)
+ return ret;
+
+ found->res_dirty = false;
+ }
+ last_cleaned = found->backup_offset + found->backup_size;
+ cur = rb_next(&found->mob_node);
+ if (!cur)
+ break;
+
+ found = container_of(cur, struct vmw_resource, mob_node);
+ if (found->backup_offset >= res_end)
+ break;
+ }
+
+ /*
+ * Set number of pages allowed prefaulting and fence the buffer object
+ */
+ *num_prefault = 1;
+ if (last_cleaned > res_start) {
+ struct ttm_buffer_object *bo = &vbo->base;
+
+ *num_prefault = __KERNEL_DIV_ROUND_UP(last_cleaned - res_start,
+ PAGE_SIZE);
+ vmw_bo_fence_single(bo, NULL);
+ if (bo->moving)
+ dma_fence_put(bo->moving);
+ bo->moving = dma_fence_get
+ (reservation_object_get_excl(bo->resv));
+ }
+
+ return 0;
+}
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource_priv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_resource_priv.h
index c85144286cfe..3b7438b2d289 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource_priv.h
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource_priv.h
@@ -77,6 +77,7 @@ struct vmw_user_resource_conv {
* @dirty_sync: Upload the dirty mob contents to the resource.
* @dirty_add_range: Add a sequential dirty range to the resource
* dirty tracker.
+ * @clean: Clean the resource.
*/
struct vmw_res_func {
enum vmw_res_type res_type;
@@ -101,6 +102,7 @@ struct vmw_res_func {
int (*dirty_sync)(struct vmw_resource *res);
void (*dirty_range_add)(struct vmw_resource *res, size_t start,
size_t end);
+ int (*clean)(struct vmw_resource *res);
};
/**
diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_validation.c b/drivers/gpu/drm/vmwgfx/vmwgfx_validation.c
index 71349a7bae90..9aaf807ed73c 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_validation.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_validation.c
@@ -641,7 +641,8 @@ int vmw_validation_res_validate(struct vmw_validation_context *ctx, bool intr)
struct vmw_resource *res = val->res;
struct vmw_buffer_object *backup = res->backup;
- ret = vmw_resource_validate(res, intr);
+ ret = vmw_resource_validate(res, intr, val->dirty_set &&
+ val->dirty);
if (ret) {
if (ret != -ERESTARTSYS)
DRM_ERROR("Failed to validate resource.\n");
--
2.20.1
From: Thomas Hellstrom <[email protected]>
Add a pointer to the struct vm_operations_struct in the bo_device, and
assign that pointer to the default value currently used.
The driver can then optionally modify that pointer and the new value
can be used for each new vma created.
Cc: "Christian König" <[email protected]>
Signed-off-by: Thomas Hellstrom <[email protected]>
Reviewed-by: Christian König <[email protected]>
---
drivers/gpu/drm/ttm/ttm_bo.c | 1 +
drivers/gpu/drm/ttm/ttm_bo_vm.c | 6 +++---
include/drm/ttm/ttm_bo_driver.h | 6 ++++++
3 files changed, 10 insertions(+), 3 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index c7de667d482a..6953dd264172 100644
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -1739,6 +1739,7 @@ int ttm_bo_device_init(struct ttm_bo_device *bdev,
mutex_lock(&ttm_global_mutex);
list_add_tail(&bdev->device_list, &glob->device_list);
mutex_unlock(&ttm_global_mutex);
+ bdev->vm_ops = &ttm_bo_vm_ops;
return 0;
out_no_sys:
diff --git a/drivers/gpu/drm/ttm/ttm_bo_vm.c b/drivers/gpu/drm/ttm/ttm_bo_vm.c
index 6dacff49c1cc..196e13a0adad 100644
--- a/drivers/gpu/drm/ttm/ttm_bo_vm.c
+++ b/drivers/gpu/drm/ttm/ttm_bo_vm.c
@@ -395,7 +395,7 @@ static int ttm_bo_vm_access(struct vm_area_struct *vma, unsigned long addr,
return ret;
}
-static const struct vm_operations_struct ttm_bo_vm_ops = {
+const struct vm_operations_struct ttm_bo_vm_ops = {
.fault = ttm_bo_vm_fault,
.open = ttm_bo_vm_open,
.close = ttm_bo_vm_close,
@@ -448,7 +448,7 @@ int ttm_bo_mmap(struct file *filp, struct vm_area_struct *vma,
if (unlikely(ret != 0))
goto out_unref;
- vma->vm_ops = &ttm_bo_vm_ops;
+ vma->vm_ops = bdev->vm_ops;
/*
* Note: We're transferring the bo reference to
@@ -480,7 +480,7 @@ int ttm_fbdev_mmap(struct vm_area_struct *vma, struct ttm_buffer_object *bo)
ttm_bo_get(bo);
- vma->vm_ops = &ttm_bo_vm_ops;
+ vma->vm_ops = bo->bdev->vm_ops;
vma->vm_private_data = bo;
vma->vm_flags |= VM_MIXEDMAP;
vma->vm_flags |= VM_IO | VM_DONTEXPAND;
diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h
index c9b8ba492f24..a2d810a2504d 100644
--- a/include/drm/ttm/ttm_bo_driver.h
+++ b/include/drm/ttm/ttm_bo_driver.h
@@ -442,6 +442,9 @@ extern struct ttm_bo_global {
* @driver: Pointer to a struct ttm_bo_driver struct setup by the driver.
* @man: An array of mem_type_managers.
* @vma_manager: Address space manager
+ * @vm_ops: Pointer to the struct vm_operations_struct used for this
+ * device's VM operations. The driver may override this before the first
+ * mmap() call.
* lru_lock: Spinlock that protects the buffer+device lru lists and
* ddestroy lists.
* @dev_mapping: A pointer to the struct address_space representing the
@@ -460,6 +463,7 @@ struct ttm_bo_device {
struct ttm_bo_global *glob;
struct ttm_bo_driver *driver;
struct ttm_mem_type_manager man[TTM_NUM_MEM_TYPES];
+ const struct vm_operations_struct *vm_ops;
/*
* Protected by internal locks.
@@ -488,6 +492,8 @@ struct ttm_bo_device {
bool no_retry;
};
+extern const struct vm_operations_struct ttm_bo_vm_ops;
+
/**
* struct ttm_lru_bulk_move_pos
*
--
2.20.1
From: Thomas Hellstrom <[email protected]>
With the vmwgfx dirty tracking, the default TTM fault handler is not
completely sufficient (vmwgfx need to modify the vma->vm_flags member,
and also needs to restrict the number of prefaults).
We also want to replicate the new ttm_bo_vm_reserve() functionality
So start turning the TTM vm code into helpers: ttm_bo_vm_fault_reserved()
and ttm_bo_vm_reserve(), and provide a default TTM fault handler for other
drivers to use.
Cc: "Christian König" <[email protected]>
Signed-off-by: Thomas Hellstrom <[email protected]>
Reviewed-by: "Christian König" <[email protected]> #v1
---
drivers/gpu/drm/ttm/ttm_bo_vm.c | 175 +++++++++++++++++++-------------
include/drm/ttm/ttm_bo_api.h | 10 ++
2 files changed, 113 insertions(+), 72 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_bo_vm.c b/drivers/gpu/drm/ttm/ttm_bo_vm.c
index 196e13a0adad..2d9862fcf6fd 100644
--- a/drivers/gpu/drm/ttm/ttm_bo_vm.c
+++ b/drivers/gpu/drm/ttm/ttm_bo_vm.c
@@ -42,8 +42,6 @@
#include <linux/uaccess.h>
#include <linux/mem_encrypt.h>
-#define TTM_BO_VM_NUM_PREFAULT 16
-
static vm_fault_t ttm_bo_vm_fault_idle(struct ttm_buffer_object *bo,
struct vm_fault *vmf)
{
@@ -106,31 +104,30 @@ static unsigned long ttm_bo_io_mem_pfn(struct ttm_buffer_object *bo,
+ page_offset;
}
-static vm_fault_t ttm_bo_vm_fault(struct vm_fault *vmf)
+/**
+ * ttm_bo_vm_reserve - Reserve a buffer object in a retryable vm callback
+ * @bo: The buffer object
+ * @vmf: The fault structure handed to the callback
+ *
+ * vm callbacks like fault() and *_mkwrite() allow for the mm_sem to be dropped
+ * during long waits, and after the wait the callback will be restarted. This
+ * is to allow other threads using the same virtual memory space concurrent
+ * access to map(), unmap() completely unrelated buffer objects. TTM buffer
+ * object reservations sometimes wait for GPU and should therefore be
+ * considered long waits. This function reserves the buffer object interruptibly
+ * taking this into account. Starvation is avoided by the vm system not
+ * allowing too many repeated restarts.
+ * This function is intended to be used in customized fault() and _mkwrite()
+ * handlers.
+ *
+ * Return:
+ * 0 on success and the bo was reserved.
+ * VM_FAULT_RETRY if blocking wait.
+ * VM_FAULT_NOPAGE if blocking wait and retrying was not allowed.
+ */
+vm_fault_t ttm_bo_vm_reserve(struct ttm_buffer_object *bo,
+ struct vm_fault *vmf)
{
- struct vm_area_struct *vma = vmf->vma;
- struct ttm_buffer_object *bo = (struct ttm_buffer_object *)
- vma->vm_private_data;
- struct ttm_bo_device *bdev = bo->bdev;
- unsigned long page_offset;
- unsigned long page_last;
- unsigned long pfn;
- struct ttm_tt *ttm = NULL;
- struct page *page;
- int err;
- int i;
- vm_fault_t ret = VM_FAULT_NOPAGE;
- unsigned long address = vmf->address;
- struct ttm_mem_type_manager *man =
- &bdev->man[bo->mem.mem_type];
- struct vm_area_struct cvma;
-
- /*
- * Work around locking order reversal in fault / nopfn
- * between mmap_sem and bo_reserve: Perform a trylock operation
- * for reserve, and if it fails, retry the fault after waiting
- * for the buffer to become unreserved.
- */
if (unlikely(!reservation_object_trylock(bo->resv))) {
if (vmf->flags & FAULT_FLAG_ALLOW_RETRY) {
if (!(vmf->flags & FAULT_FLAG_RETRY_NOWAIT)) {
@@ -151,14 +148,55 @@ static vm_fault_t ttm_bo_vm_fault(struct vm_fault *vmf)
return VM_FAULT_NOPAGE;
}
+ return 0;
+}
+EXPORT_SYMBOL(ttm_bo_vm_reserve);
+
+/**
+ * ttm_bo_vm_fault_reserved - TTM fault helper
+ * @vmf: The struct vm_fault given as argument to the fault callback
+ * @prot: The page protection to be used for this memory area.
+ * @num_prefault: Maximum number of prefault pages. The caller may want to
+ * specify this based on madvice settings and the size of the GPU object
+ * backed by the memory.
+ *
+ * This function inserts one or more page table entries pointing to the
+ * memory backing the buffer object, and then returns a return code
+ * instructing the caller to retry the page access.
+ *
+ * Return:
+ * VM_FAULT_NOPAGE on success or pending signal
+ * VM_FAULT_SIGBUS on unspecified error
+ * VM_FAULT_OOM on out-of-memory
+ * VM_FAULT_RETRY if retryable wait
+ */
+vm_fault_t ttm_bo_vm_fault_reserved(struct vm_fault *vmf,
+ pgprot_t prot,
+ pgoff_t num_prefault)
+{
+ struct vm_area_struct *vma = vmf->vma;
+ struct vm_area_struct cvma = *vma;
+ struct ttm_buffer_object *bo = (struct ttm_buffer_object *)
+ vma->vm_private_data;
+ struct ttm_bo_device *bdev = bo->bdev;
+ unsigned long page_offset;
+ unsigned long page_last;
+ unsigned long pfn;
+ struct ttm_tt *ttm = NULL;
+ struct page *page;
+ int err;
+ pgoff_t i;
+ vm_fault_t ret = VM_FAULT_NOPAGE;
+ unsigned long address = vmf->address;
+ struct ttm_mem_type_manager *man =
+ &bdev->man[bo->mem.mem_type];
+
/*
* Refuse to fault imported pages. This should be handled
* (if at all) by redirecting mmap to the exporter.
*/
- if (bo->ttm && (bo->ttm->page_flags & TTM_PAGE_FLAG_SG)) {
- ret = VM_FAULT_SIGBUS;
- goto out_unlock;
- }
+ if (bo->ttm && (bo->ttm->page_flags & TTM_PAGE_FLAG_SG))
+ return VM_FAULT_SIGBUS;
if (bdev->driver->fault_reserve_notify) {
struct dma_fence *moving = dma_fence_get(bo->moving);
@@ -169,11 +207,9 @@ static vm_fault_t ttm_bo_vm_fault(struct vm_fault *vmf)
break;
case -EBUSY:
case -ERESTARTSYS:
- ret = VM_FAULT_NOPAGE;
- goto out_unlock;
+ return VM_FAULT_NOPAGE;
default:
- ret = VM_FAULT_SIGBUS;
- goto out_unlock;
+ return VM_FAULT_SIGBUS;
}
if (bo->moving != moving) {
@@ -189,26 +225,15 @@ static vm_fault_t ttm_bo_vm_fault(struct vm_fault *vmf)
* move.
*/
ret = ttm_bo_vm_fault_idle(bo, vmf);
- if (unlikely(ret != 0)) {
- if (ret == VM_FAULT_RETRY &&
- !(vmf->flags & FAULT_FLAG_RETRY_NOWAIT)) {
- /* The BO has already been unreserved. */
- return ret;
- }
-
- goto out_unlock;
- }
+ if (unlikely(ret != 0))
+ return ret;
err = ttm_mem_io_lock(man, true);
- if (unlikely(err != 0)) {
- ret = VM_FAULT_NOPAGE;
- goto out_unlock;
- }
+ if (unlikely(err != 0))
+ return VM_FAULT_NOPAGE;
err = ttm_mem_io_reserve_vm(bo);
- if (unlikely(err != 0)) {
- ret = VM_FAULT_SIGBUS;
- goto out_io_unlock;
- }
+ if (unlikely(err != 0))
+ return VM_FAULT_SIGBUS;
page_offset = ((address - vma->vm_start) >> PAGE_SHIFT) +
vma->vm_pgoff - drm_vma_node_start(&bo->vma_node);
@@ -220,18 +245,8 @@ static vm_fault_t ttm_bo_vm_fault(struct vm_fault *vmf)
goto out_io_unlock;
}
- /*
- * Make a local vma copy to modify the page_prot member
- * and vm_flags if necessary. The vma parameter is protected
- * by mmap_sem in write mode.
- */
- cvma = *vma;
- cvma.vm_page_prot = vm_get_page_prot(cvma.vm_flags);
-
- if (bo->mem.bus.is_iomem) {
- cvma.vm_page_prot = ttm_io_prot(bo->mem.placement,
- cvma.vm_page_prot);
- } else {
+ cvma.vm_page_prot = ttm_io_prot(bo->mem.placement, prot);
+ if (!bo->mem.bus.is_iomem) {
struct ttm_operation_ctx ctx = {
.interruptible = false,
.no_wait_gpu = false,
@@ -240,24 +255,21 @@ static vm_fault_t ttm_bo_vm_fault(struct vm_fault *vmf)
};
ttm = bo->ttm;
- cvma.vm_page_prot = ttm_io_prot(bo->mem.placement,
- cvma.vm_page_prot);
-
- /* Allocate all page at once, most common usage */
- if (ttm_tt_populate(ttm, &ctx)) {
+ if (ttm_tt_populate(bo->ttm, &ctx)) {
ret = VM_FAULT_OOM;
goto out_io_unlock;
}
+ } else {
+ /* Iomem should not be marked encrypted */
+ cvma.vm_page_prot = pgprot_decrypted(cvma.vm_page_prot);
}
/*
* Speculatively prefault a number of pages. Only error on
* first page.
*/
- for (i = 0; i < TTM_BO_VM_NUM_PREFAULT; ++i) {
+ for (i = 0; i < num_prefault; ++i) {
if (bo->mem.bus.is_iomem) {
- /* Iomem should not be marked encrypted */
- cvma.vm_page_prot = pgprot_decrypted(cvma.vm_page_prot);
pfn = ttm_bo_io_mem_pfn(bo, page_offset);
} else {
page = ttm->pages[page_offset];
@@ -295,7 +307,26 @@ static vm_fault_t ttm_bo_vm_fault(struct vm_fault *vmf)
ret = VM_FAULT_NOPAGE;
out_io_unlock:
ttm_mem_io_unlock(man);
-out_unlock:
+ return ret;
+}
+EXPORT_SYMBOL(ttm_bo_vm_fault_reserved);
+
+static vm_fault_t ttm_bo_vm_fault(struct vm_fault *vmf)
+{
+ struct vm_area_struct *vma = vmf->vma;
+ pgprot_t prot;
+ struct ttm_buffer_object *bo = vma->vm_private_data;
+ vm_fault_t ret;
+
+ ret = ttm_bo_vm_reserve(bo, vmf);
+ if (ret)
+ return ret;
+
+ prot = vm_get_page_prot(vma->vm_flags);
+ ret = ttm_bo_vm_fault_reserved(vmf, prot, TTM_BO_VM_NUM_PREFAULT);
+ if (ret == VM_FAULT_RETRY && !(vmf->flags & FAULT_FLAG_RETRY_NOWAIT))
+ return ret;
+
reservation_object_unlock(bo->resv);
return ret;
}
diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h
index 49d9cdfc58f2..435d02f719a8 100644
--- a/include/drm/ttm/ttm_bo_api.h
+++ b/include/drm/ttm/ttm_bo_api.h
@@ -768,4 +768,14 @@ int ttm_bo_swapout(struct ttm_bo_global *glob,
struct ttm_operation_ctx *ctx);
void ttm_bo_swapout_all(struct ttm_bo_device *bdev);
int ttm_bo_wait_unreserved(struct ttm_buffer_object *bo);
+
+/* Default number of pre-faulted pages in the TTM fault handler */
+#define TTM_BO_VM_NUM_PREFAULT 16
+
+vm_fault_t ttm_bo_vm_reserve(struct ttm_buffer_object *bo,
+ struct vm_fault *vmf);
+
+vm_fault_t ttm_bo_vm_fault_reserved(struct vm_fault *vmf,
+ pgprot_t prot,
+ pgoff_t num_prefault);
#endif
--
2.20.1