drm/i915: Preallocate stashes for vma page-directories

We need to make the DMA allocations used for page directories to be
performed up front so that we can include those allocations in our
memory reservation pass. The downside is that we have to assume the
worst case, even before we know the final layout, and always allocate
enough page directories for this object, even when there will be overlap.
This unfortunately can be quite expensive, especially as we have to
clear/reset the page directories and DMA pages, but it should only be
required during early phases of a workload when new objects are being
discovered, or after memory/eviction pressure when we need to rebind.
Once we reach steady state, the objects should not be moved and we no
longer need to preallocating the pages tables.

It should be noted that the lifetime for the page directories DMA is
more or less decoupled from individual fences as they will be shared
across objects across timelines.

v2: Only allocate enough PD space for the PTE we may use, we do not need
to allocate PD that will be left as scratch.
v3: Store the shift unto the first PD level to encapsulate the different
PTE counts for gen6/gen8.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200729164219.5737-1-chris@chris-wilson.co.uk
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

1 files changed, 36 insertions, 24 deletions

diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
index 0016ffc7d914..9b8fc990e9ef 100644
--- a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
@@ -172,35 +172,33 @@ static int igt_ppgtt_alloc(void *arg)
 
 	/* Check we can allocate the entire range */
 	for (size = 4096; size <= limit; size <<= 2) {
-		err = ppgtt->vm.allocate_va_range(&ppgtt->vm, 0, size);
-		if (err) {
-			if (err == -ENOMEM) {
-				pr_info("[1] Ran out of memory for va_range [0 + %llx] [bit %d]\n",
-					size, ilog2(size));
-				err = 0; /* virtual space too large! */
-			}
+		struct i915_vm_pt_stash stash = {};
+
+		err = i915_vm_alloc_pt_stash(&ppgtt->vm, &stash, size);
+		if (err)
 			goto err_ppgtt_cleanup;
-		}
 
+		ppgtt->vm.allocate_va_range(&ppgtt->vm, &stash, 0, size);
 		cond_resched();
 
 		ppgtt->vm.clear_range(&ppgtt->vm, 0, size);
+
+		i915_vm_free_pt_stash(&ppgtt->vm, &stash);
 	}
 
 	/* Check we can incrementally allocate the entire range */
 	for (last = 0, size = 4096; size <= limit; last = size, size <<= 2) {
-		err = ppgtt->vm.allocate_va_range(&ppgtt->vm,
-						  last, size - last);
-		if (err) {
-			if (err == -ENOMEM) {
-				pr_info("[2] Ran out of memory for va_range [%llx + %llx] [bit %d]\n",
-					last, size - last, ilog2(size));
-				err = 0; /* virtual space too large! */
-			}
+		struct i915_vm_pt_stash stash = {};
+
+		err = i915_vm_alloc_pt_stash(&ppgtt->vm, &stash, size - last);
+		if (err)
 			goto err_ppgtt_cleanup;
-		}
 
+		ppgtt->vm.allocate_va_range(&ppgtt->vm, &stash,
+					    last, size - last);
 		cond_resched();
+
+		i915_vm_free_pt_stash(&ppgtt->vm, &stash);
 	}
 
 err_ppgtt_cleanup:
@@ -284,9 +282,18 @@ static int lowlevel_hole(struct i915_address_space *vm,
 				break;
 			}
 
-			if (vm->allocate_va_range &&
-			    vm->allocate_va_range(vm, addr, BIT_ULL(size)))
-				break;
+			if (vm->allocate_va_range) {
+				struct i915_vm_pt_stash stash = {};
+
+				if (i915_vm_alloc_pt_stash(vm, &stash,
+							   BIT_ULL(size)))
+					break;
+
+				vm->allocate_va_range(vm, &stash,
+						      addr, BIT_ULL(size));
+
+				i915_vm_free_pt_stash(vm, &stash);
+			}
 
 			mock_vma->pages = obj->mm.pages;
 			mock_vma->node.size = BIT_ULL(size);
@@ -1881,6 +1888,7 @@ static int igt_cs_tlb(void *arg)
 			continue;
 
 		while (!__igt_timeout(end_time, NULL)) {
+			struct i915_vm_pt_stash stash = {};
 			struct i915_request *rq;
 			u64 offset;
 
@@ -1888,10 +1896,6 @@ static int igt_cs_tlb(void *arg)
 						   0, vm->total - PAGE_SIZE,
 						   chunk_size, PAGE_SIZE);
 
-			err = vm->allocate_va_range(vm, offset, chunk_size);
-			if (err)
-				goto end;
-
 			memset32(result, STACK_MAGIC, PAGE_SIZE / sizeof(u32));
 
 			vma = i915_vma_instance(bbe, vm, NULL);
@@ -1904,6 +1908,14 @@ static int igt_cs_tlb(void *arg)
 			if (err)
 				goto end;
 
+			err = i915_vm_alloc_pt_stash(vm, &stash, chunk_size);
+			if (err)
+				goto end;
+
+			vm->allocate_va_range(vm, &stash, offset, chunk_size);
+
+			i915_vm_free_pt_stash(vm, &stash);
+
 			/* Prime the TLB with the dummy pages */
 			for (i = 0; i < count; i++) {
 				vma->node.start = offset + i * PAGE_SIZE;