Skip to contentSkip to conversation

drm/i915/gtt: Allow >= 4GB offsets in X86_32

PATCH In review Michel Thierry 2 participants 7 messages Aug 7, 2015

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/4 drm/i915/gtt: Allow >= 4GB offsets in X86_32

Similar to commit c44ef60e437019b8ca1dab8b4d2e8761fd4ce1e9 ("drm/i915/gtt:
Allow >= 4GB sizes for vm"), i915_gem_obj_offset and i915_gem_obj_ggtt_offset return an unsigned long, which in only 4-bytes long in 32-bit kernels.

Change return type (and other related offset variables) to u64.

Since Global GTT is always limited to 4GB, this change would not be required in i915_gem_obj_ggtt_offset, but this is done for consistency.

v2: Remove unnecessary offset variable in do_pin, as we already have
vma->node.start (Chris).
Update GGTT offset too (Tvrtko).

Cc Chris Wilson <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Tvrtko Ursulin <[email protected]>
---
 drivers/gpu/drm/i915/i915_drv.h       | 12 +++++-------
 drivers/gpu/drm/i915/i915_gem.c       | 18 +++++++-----------
 drivers/gpu/drm/i915/i915_gem_fence.c |  4 ++--
 drivers/gpu/drm/i915/i915_gem_gtt.c   |  8 ++++----
 drivers/gpu/drm/i915/i915_gem_gtt.h   |  2 +-
 drivers/gpu/drm/i915/intel_fbdev.c    |  2 +-
 6 files changed, 20 insertions(+), 26 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +5 -7
index bfd6291..4768a93 100644
@@ -2968,13 +2968,11 @@ struct drm_gem_object *i915_gem_prime_import(struct drm_device *dev,
 struct dma_buf *i915_gem_prime_export(struct drm_device *dev,
 				struct drm_gem_object *gem_obj, int flags);
 
-unsigned long
-i915_gem_obj_ggtt_offset_view(struct drm_i915_gem_object *o,
-			      const struct i915_ggtt_view *view);
-unsigned long
-i915_gem_obj_offset(struct drm_i915_gem_object *o,
-		    struct i915_address_space *vm);
-static inline unsigned long
+u64 i915_gem_obj_ggtt_offset_view(struct drm_i915_gem_object *o,
+				  const struct i915_ggtt_view *view);
+u64 i915_gem_obj_offset(struct drm_i915_gem_object *o,
+			struct i915_address_space *vm);
+static inline u64
 i915_gem_obj_ggtt_offset(struct drm_i915_gem_object *o)
 {
 	return i915_gem_obj_ggtt_offset_view(o, &i915_ggtt_view_normal);
drivers/gpu/drm/i915/i915_gem.c +7 -11
index a86227e..b9dde7a 100644
@@ -4225,15 +4225,13 @@ i915_gem_object_do_pin(struct drm_i915_gem_object *obj,
 			return -EBUSY;
 
 		if (i915_vma_misplaced(vma, alignment, flags)) {
-			unsigned long offset;
-			offset = ggtt_view ? i915_gem_obj_ggtt_offset_view(obj, ggtt_view) :
-					     i915_gem_obj_offset(obj, vm);
 			WARN(vma->pin_count,
 			     "bo is already pinned in %s with incorrect alignment:"
-			     " offset=%lx, req.alignment=%x, req.map_and_fenceable=%d,"
+			     " offset=%08x %08x, req.alignment=%x, req.map_and_fenceable=%d,"
 			     " obj->map_and_fenceable=%d\n",
 			     ggtt_view ? "ggtt" : "ppgtt",
-			     offset,
+			     upper_32_bits(vma->node.start),
+			     lower_32_bits(vma->node.start),
 			     alignment,
 			     !!(flags & PIN_MAPPABLE),
 			     obj->map_and_fenceable);
@@ -5188,9 +5186,8 @@ void i915_gem_track_fb(struct drm_i915_gem_object *old,
 }
 
 /* All the new VM stuff */
-unsigned long
-i915_gem_obj_offset(struct drm_i915_gem_object *o,
-		    struct i915_address_space *vm)
+u64 i915_gem_obj_offset(struct drm_i915_gem_object *o,
+			struct i915_address_space *vm)
 {
 	struct drm_i915_private *dev_priv = o->base.dev->dev_private;
 	struct i915_vma *vma;
@@ -5210,9 +5207,8 @@ i915_gem_obj_offset(struct drm_i915_gem_object *o,
 	return -1;
 }
 
-unsigned long
-i915_gem_obj_ggtt_offset_view(struct drm_i915_gem_object *o,
-			      const struct i915_ggtt_view *view)
+u64 i915_gem_obj_ggtt_offset_view(struct drm_i915_gem_object *o,
+				  const struct i915_ggtt_view *view)
 {
 	struct i915_address_space *ggtt = i915_obj_to_ggtt(o);
 	struct i915_vma *vma;
drivers/gpu/drm/i915/i915_gem_fence.c +2 -2
index af1f8c4..6077dff 100644
@@ -128,7 +128,7 @@ static void i915_write_fence_reg(struct drm_device *dev, int reg,
 		WARN((i915_gem_obj_ggtt_offset(obj) & ~I915_FENCE_START_MASK) ||
 		     (size & -size) != size ||
 		     (i915_gem_obj_ggtt_offset(obj) & (size - 1)),
-		     "object 0x%08lx [fenceable? %d] not 1M or pot-size (0x%08x) aligned\n",
+		     "object 0x%08llx [fenceable? %d] not 1M or pot-size (0x%08x) aligned\n",
 		     i915_gem_obj_ggtt_offset(obj), obj->map_and_fenceable, size);
 
 		if (obj->tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev))
@@ -171,7 +171,7 @@ static void i830_write_fence_reg(struct drm_device *dev, int reg,
 		WARN((i915_gem_obj_ggtt_offset(obj) & ~I830_FENCE_START_MASK) ||
 		     (size & -size) != size ||
 		     (i915_gem_obj_ggtt_offset(obj) & (size - 1)),
-		     "object 0x%08lx not 512K or pot-size 0x%08x aligned\n",
+		     "object 0x%08llx not 512K or pot-size 0x%08x aligned\n",
 		     i915_gem_obj_ggtt_offset(obj), size);
 
 		pitch_val = obj->stride / 128;
drivers/gpu/drm/i915/i915_gem_gtt.c +4 -4
index a7dbc20..dee3b39 100644
@@ -2558,9 +2558,9 @@ static void i915_gtt_color_adjust(struct drm_mm_node *node,
 }
 
 static int i915_gem_setup_global_gtt(struct drm_device *dev,
-				     unsigned long start,
-				     unsigned long mappable_end,
-				     unsigned long end)
+				     u64 start,
+				     u64 mappable_end,
+				     u64 end)
 {
 	/* Let GEM Manage all of the aperture.
 	 *
@@ -2599,7 +2599,7 @@ static int i915_gem_setup_global_gtt(struct drm_device *dev,
 	list_for_each_entry(obj, &dev_priv->mm.bound_list, global_list) {
 		struct i915_vma *vma = i915_gem_obj_to_vma(obj, ggtt_vm);
 
-		DRM_DEBUG_KMS("reserving preallocated space: %lx + %zx\n",
+		DRM_DEBUG_KMS("reserving preallocated space: %llx + %zx\n",
 			      i915_gem_obj_ggtt_offset(obj), obj->base.size);
 
 		WARN_ON(i915_gem_obj_ggtt_bound(obj));
drivers/gpu/drm/i915/i915_gem_gtt.h +1 -1
index 8412b15..b6c9990 100644
@@ -149,7 +149,7 @@ struct i915_ggtt_view {
 
 	union {
 		struct {
-			unsigned long offset;
+			u64 offset;
 			unsigned int size;
 		} partial;
 	} params;
drivers/gpu/drm/i915/intel_fbdev.c +1 -1
index 8c6a6fa..96476d7 100644
@@ -263,7 +263,7 @@ static int intelfb_create(struct drm_fb_helper *helper,
 
 	/* Use default scratch pixmap (info->pixmap.flags = FB_PIXMAP_SYSTEM) */
 
-	DRM_DEBUG_KMS("allocated %dx%d fb: 0x%08lx, bo %p\n",
+	DRM_DEBUG_KMS("allocated %dx%d fb: 0x%08llx, bo %p\n",
 		      fb->width, fb->height,
 		      i915_gem_obj_ggtt_offset(obj), obj);
 
Signature · 6 lines
2.5.0

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

PATCH 2/4 drm/i915: Use complete virtual address range on 32-bit platforms

With the offset length being taken care of in ("drm/i915/gtt: Allow >= 4GB offsets in X86_32"), the code should be finally safe in 32-bit kernels.

This reverts commit 501fd70fcaebc911b6b96a7b331e6960e5af67e7

Author: Michel Thierry <[email protected]>
Date:   Fri May 29 14:15:05 2015 +0100

drm/i915: limit PPGTT size to 2GB in 32-bit platforms

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 8 --------
 1 file changed, 8 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -8
index dee3b39..451913a 100644
@@ -1495,14 +1495,6 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
 			goto free_scratch;
 
 		ppgtt->base.total = 1ULL << 32;
-		if (IS_ENABLED(CONFIG_X86_32))
-			/* While we have a proliferation of size_t variables
-			 * we cannot represent the full ppgtt size on 32bit,
-			 * so limit it to the same size as the GGTT (currently
-			 * 2GiB).
-			 */
-			ppgtt->base.total = to_i915(ppgtt->base.dev)->gtt.base.total;
-
 		ppgtt->switch_mm = gen8_legacy_mm_switch;
 		trace_i915_page_directory_pointer_entry_alloc(&ppgtt->base,
 							      0, 0,
Signature · 6 lines
2.5.0

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v1 2/4
#2 lore raw

Re: [PATCH 2/4] drm/i915: Use complete virtual address range on 32-bit platforms

Quoted 11 lines · Michel Thierry

With the offset length being taken care of in ("drm/i915/gtt: Allow >= 4GB offsets in X86_32"), the code should be finally safe in 32-bit kernels.

This reverts commit 501fd70fcaebc911b6b96a7b331e6960e5af67e7

Author: Michel Thierry <[email protected]>
Date:   Fri May 29 14:15:05 2015 +0100

drm/i915: limit PPGTT size to 2GB in 32-bit platforms

I think we should be safe....

The biggest problem is that who do have testing 32bit gen8+? -Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw

PATCH 3/4 drm/i915: Always pass dev pointer in pdp_init

In reply to Michel Thierry

And fix 0-DAY kernel test infrastructure warning.

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
index 451913a..d2910a8 100644
@@ -1490,7 +1490,7 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
 		ppgtt->base.total = 1ULL << 48;
 		ppgtt->switch_mm = gen8_48b_mm_switch;
 	} else {
-		ret = __pdp_init(false, &ppgtt->pdp);
+		ret = __pdp_init(ppgtt->base.dev, &ppgtt->pdp);
 		if (ret)
 			goto free_scratch;
 
Signature · 6 lines
2.5.0

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v1 3/4
#4 lore raw

Re: [PATCH 3/4] drm/i915: Always pass dev pointer in pdp_init

Quoted 4 lines · Michel Thierry

And fix 0-DAY kernel test infrastructure warning.

Nah, for extra credit, pass NULL!

Not that this needs a r-b,

-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#5 lore raw

PATCH 4/4 drm/i915/ppgtt: Abstract 4lvl and legacy functions

In reply to Michel Thierry

Cleanup, allocate, insert, clear and dump functions were already vfuncs, this patch now point them to the 4-level (48-bit) or legacy (32-bit) versions.

This removes unnecessary checks of which ppgtt is in use, as it now only happens once, in ppgtt_init.

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 253 ++++++++++++++++++++++--------------
 1 file changed, 153 insertions(+), 100 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +153 -100
index d2910a8..e0d9ae0 100644
@@ -755,28 +755,37 @@ static void gen8_ppgtt_clear_pte_range(struct i915_address_space *vm,
 	}
 }
 
-static void gen8_ppgtt_clear_range(struct i915_address_space *vm,
-				   uint64_t start,
-				   uint64_t length,
-				   bool use_scratch)
+static void gen8_legacy_ppgtt_clear_range(struct i915_address_space *vm,
+					  uint64_t start,
+					  uint64_t length,
+					  bool use_scratch)
 {
 	struct i915_hw_ppgtt *ppgtt =
 		container_of(vm, struct i915_hw_ppgtt, base);
 	gen8_pte_t scratch_pte = gen8_pte_encode(px_dma(vm->scratch_page),
 						 I915_CACHE_LLC, use_scratch);
 
-	if (!USES_FULL_48BIT_PPGTT(vm->dev)) {
-		gen8_ppgtt_clear_pte_range(vm, &ppgtt->pdp, start, length,
-					   scratch_pte);
-	} else {
-		uint64_t templ4, pml4e;
-		struct i915_page_directory_pointer *pdp;
+	gen8_ppgtt_clear_pte_range(vm, &ppgtt->pdp, start, length,
+				   scratch_pte);
+}
 
-		gen8_for_each_pml4e(pdp, &ppgtt->pml4, start, length, templ4, pml4e) {
-			gen8_ppgtt_clear_pte_range(vm, pdp, start, length,
-						   scratch_pte);
-		}
+static void gen8_4lvl_ppgtt_clear_range(struct i915_address_space *vm,
+					uint64_t start,
+					uint64_t length,
+					bool use_scratch)
+{
+	struct i915_hw_ppgtt *ppgtt =
+		container_of(vm, struct i915_hw_ppgtt, base);
+	struct i915_page_directory_pointer *pdp;
+	uint64_t templ4, pml4e;
+	gen8_pte_t scratch_pte = gen8_pte_encode(px_dma(vm->scratch_page),
+						 I915_CACHE_LLC, use_scratch);
+
+	gen8_for_each_pml4e(pdp, &ppgtt->pml4, start, length, templ4, pml4e) {
+		gen8_ppgtt_clear_pte_range(vm, pdp, start, length,
+					   scratch_pte);
 	}
+
 }
 
 static void
@@ -821,11 +830,11 @@ gen8_ppgtt_insert_pte_entries(struct i915_address_space *vm,
 		kunmap_px(ppgtt, pt_vaddr);
 }
 
-static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,
-				      struct sg_table *pages,
-				      uint64_t start,
-				      enum i915_cache_level cache_level,
-				      u32 unused)
+static void gen8_legacy_ppgtt_insert_entries(struct i915_address_space *vm,
+					     struct sg_table *pages,
+					     uint64_t start,
+					     enum i915_cache_level cache_level,
+					     u32 unused)
 {
 	struct i915_hw_ppgtt *ppgtt =
 		container_of(vm, struct i915_hw_ppgtt, base);
@@ -833,18 +842,28 @@ static void gen8_ppgtt_insert_entries(struct i915_address_space *vm,
 
 	__sg_page_iter_start(&sg_iter, pages->sgl, sg_nents(pages->sgl), 0);
 
-	if (!USES_FULL_48BIT_PPGTT(vm->dev)) {
-		gen8_ppgtt_insert_pte_entries(vm, &ppgtt->pdp, &sg_iter, start,
-					      cache_level);
-	} else {
-		struct i915_page_directory_pointer *pdp;
-		uint64_t templ4, pml4e;
-		uint64_t length = (uint64_t)pages->orig_nents << PAGE_SHIFT;
+	gen8_ppgtt_insert_pte_entries(vm, &ppgtt->pdp, &sg_iter, start,
+				      cache_level);
+}
 
-		gen8_for_each_pml4e(pdp, &ppgtt->pml4, start, length, templ4, pml4e) {
-			gen8_ppgtt_insert_pte_entries(vm, pdp, &sg_iter,
-						      start, cache_level);
-		}
+static void gen8_4lvl_ppgtt_insert_entries(struct i915_address_space *vm,
+					   struct sg_table *pages,
+					   uint64_t start,
+					   enum i915_cache_level cache_level,
+					   u32 unused)
+{
+	struct i915_hw_ppgtt *ppgtt =
+		container_of(vm, struct i915_hw_ppgtt, base);
+	struct i915_page_directory_pointer *pdp;
+	struct sg_page_iter sg_iter;
+	uint64_t templ4, pml4e;
+	uint64_t length = (uint64_t)pages->orig_nents << PAGE_SHIFT;
+
+	__sg_page_iter_start(&sg_iter, pages->sgl, sg_nents(pages->sgl), 0);
+
+	gen8_for_each_pml4e(pdp, &ppgtt->pml4, start, length, templ4, pml4e) {
+		gen8_ppgtt_insert_pte_entries(vm, pdp, &sg_iter,
+					      start, cache_level);
 	}
 }
 
@@ -915,7 +934,7 @@ static void gen8_free_scratch(struct i915_address_space *vm)
 	free_scratch_page(dev, vm->scratch_page);
 }
 
-static void gen8_ppgtt_cleanup_3lvl(struct drm_device *dev,
+static void gen8_3lvl_ppgtt_cleanup(struct drm_device *dev,
 				    struct i915_page_directory_pointer *pdp)
 {
 	int i;
@@ -931,30 +950,29 @@ static void gen8_ppgtt_cleanup_3lvl(struct drm_device *dev,
 	free_pdp(dev, pdp);
 }
 
-static void gen8_ppgtt_cleanup_4lvl(struct i915_hw_ppgtt *ppgtt)
+static void gen8_legacy_ppgtt_cleanup(struct i915_address_space *vm)
 {
+	struct i915_hw_ppgtt *ppgtt =
+		container_of(vm, struct i915_hw_ppgtt, base);
+
+	gen8_3lvl_ppgtt_cleanup(ppgtt->base.dev, &ppgtt->pdp);
+	gen8_free_scratch(vm);
+}
+
+static void gen8_4lvl_ppgtt_cleanup(struct i915_address_space *vm)
+{
+	struct i915_hw_ppgtt *ppgtt =
+		container_of(vm, struct i915_hw_ppgtt, base);
 	int i;
 
 	for_each_set_bit(i, ppgtt->pml4.used_pml4es, GEN8_PML4ES_PER_PML4) {
 		if (WARN_ON(!ppgtt->pml4.pdps[i]))
 			continue;
 
-		gen8_ppgtt_cleanup_3lvl(ppgtt->base.dev, ppgtt->pml4.pdps[i]);
+		gen8_3lvl_ppgtt_cleanup(ppgtt->base.dev, ppgtt->pml4.pdps[i]);
 	}
 
 	cleanup_px(ppgtt->base.dev, &ppgtt->pml4);
-}
-
-static void gen8_ppgtt_cleanup(struct i915_address_space *vm)
-{
-	struct i915_hw_ppgtt *ppgtt =
-		container_of(vm, struct i915_hw_ppgtt, base);
-
-	if (!USES_FULL_48BIT_PPGTT(ppgtt->base.dev))
-		gen8_ppgtt_cleanup_3lvl(ppgtt->base.dev, &ppgtt->pdp);
-	else
-		gen8_ppgtt_cleanup_4lvl(ppgtt);
-
 	gen8_free_scratch(vm);
 }
 
@@ -1191,7 +1209,7 @@ static void mark_tlbs_dirty(struct i915_hw_ppgtt *ppgtt)
 	ppgtt->pd_dirty_rings = INTEL_INFO(ppgtt->base.dev)->ring_mask;
 }
 
-static int gen8_alloc_va_range_3lvl(struct i915_address_space *vm,
+static int gen8_3lvl_alloc_va_range(struct i915_address_space *vm,
 				    struct i915_page_directory_pointer *pdp,
 				    uint64_t start,
 				    uint64_t length)
@@ -1311,14 +1329,23 @@ err_out:
 	return ret;
 }
 
-static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm,
-				    struct i915_pml4 *pml4,
+static int gen8_legacy_alloc_va_range(struct i915_address_space *vm,
+				      uint64_t start, uint64_t length)
+{
+	struct i915_hw_ppgtt *ppgtt =
+		container_of(vm, struct i915_hw_ppgtt, base);
+
+	return gen8_3lvl_alloc_va_range(vm, &ppgtt->pdp, start, length);
+}
+
+static int gen8_4lvl_alloc_va_range(struct i915_address_space *vm,
 				    uint64_t start,
 				    uint64_t length)
 {
 	DECLARE_BITMAP(new_pdps, GEN8_PML4ES_PER_PML4);
 	struct i915_hw_ppgtt *ppgtt =
 			container_of(vm, struct i915_hw_ppgtt, base);
+	struct i915_pml4 *pml4 = &ppgtt->pml4;
 	struct i915_page_directory_pointer *pdp;
 	uint64_t temp, pml4e;
 	int ret = 0;
@@ -1342,7 +1369,7 @@ static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm,
 	gen8_for_each_pml4e(pdp, pml4, start, length, temp, pml4e) {
 		WARN_ON(!pdp);
 
-		ret = gen8_alloc_va_range_3lvl(vm, pdp, start, length);
+		ret = gen8_3lvl_alloc_va_range(vm, pdp, start, length);
 		if (ret)
 			goto err_out;
 
@@ -1356,24 +1383,12 @@ static int gen8_alloc_va_range_4lvl(struct i915_address_space *vm,
 
 err_out:
 	for_each_set_bit(pml4e, new_pdps, GEN8_PML4ES_PER_PML4)
-		gen8_ppgtt_cleanup_3lvl(vm->dev, pml4->pdps[pml4e]);
+		gen8_3lvl_ppgtt_cleanup(vm->dev, pml4->pdps[pml4e]);
 
 	return ret;
 }
 
-static int gen8_alloc_va_range(struct i915_address_space *vm,
-			       uint64_t start, uint64_t length)
-{
-	struct i915_hw_ppgtt *ppgtt =
-		container_of(vm, struct i915_hw_ppgtt, base);
-
-	if (USES_FULL_48BIT_PPGTT(vm->dev))
-		return gen8_alloc_va_range_4lvl(vm, &ppgtt->pml4, start, length);
-	else
-		return gen8_alloc_va_range_3lvl(vm, &ppgtt->pdp, start, length);
-}
-
-static void gen8_dump_pdp(struct i915_page_directory_pointer *pdp,
+static void gen8_pdp_dump(struct i915_page_directory_pointer *pdp,
 			  uint64_t start, uint64_t length,
 			  gen8_pte_t scratch_pte,
 			  struct seq_file *m)
@@ -1431,7 +1446,8 @@ static void gen8_dump_pdp(struct i915_page_directory_pointer *pdp,
 	}
 }
 
-static void gen8_dump_ppgtt(struct i915_hw_ppgtt *ppgtt, struct seq_file *m)
+static void gen8_legacy_ppgtt_dump(struct i915_hw_ppgtt *ppgtt,
+				  struct seq_file *m)
 {
 	struct i915_address_space *vm = &ppgtt->base;
 	uint64_t start = ppgtt->base.start;
@@ -1439,20 +1455,27 @@ static void gen8_dump_ppgtt(struct i915_hw_ppgtt *ppgtt, struct seq_file *m)
 	gen8_pte_t scratch_pte = gen8_pte_encode(px_dma(vm->scratch_page),
 						 I915_CACHE_LLC, true);
 
-	if (!USES_FULL_48BIT_PPGTT(vm->dev)) {
-		gen8_dump_pdp(&ppgtt->pdp, start, length, scratch_pte, m);
-	} else {
-		uint64_t templ4, pml4e;
-		struct i915_pml4 *pml4 = &ppgtt->pml4;
-		struct i915_page_directory_pointer *pdp;
+	gen8_pdp_dump(&ppgtt->pdp, start, length, scratch_pte, m);
+}
 
-		gen8_for_each_pml4e(pdp, pml4, start, length, templ4, pml4e) {
-			if (!test_bit(pml4e, pml4->used_pml4es))
-				continue;
+static void gen8_4lvl_ppgtt_dump(struct i915_hw_ppgtt *ppgtt,
+				 struct seq_file *m)
+{
+	struct i915_address_space *vm = &ppgtt->base;
+	uint64_t start = ppgtt->base.start;
+	uint64_t length = ppgtt->base.total;
+	uint64_t templ4, pml4e;
+	struct i915_pml4 *pml4 = &ppgtt->pml4;
+	struct i915_page_directory_pointer *pdp;
+	gen8_pte_t scratch_pte = gen8_pte_encode(px_dma(vm->scratch_page),
+						 I915_CACHE_LLC, true);
 
-			seq_printf(m, "    PML4E #%llu\n", pml4e);
-			gen8_dump_pdp(pdp, start, length, scratch_pte, m);
-		}
+	gen8_for_each_pml4e(pdp, pml4, start, length, templ4, pml4e) {
+		if (!test_bit(pml4e, pml4->used_pml4es))
+			continue;
+
+		seq_printf(m, "    PML4E #%llu\n", pml4e);
+		gen8_pdp_dump(pdp, start, length, scratch_pte, m);
 	}
 }
 
@@ -1463,7 +1486,7 @@ static void gen8_dump_ppgtt(struct i915_hw_ppgtt *ppgtt, struct seq_file *m)
  * space.
  *
  */
-static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
+static int gen8_legacy_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
 {
 	int ret;
 
@@ -1472,34 +1495,62 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
 		return ret;
 
 	ppgtt->base.start = 0;
-	ppgtt->base.cleanup = gen8_ppgtt_cleanup;
-	ppgtt->base.allocate_va_range = gen8_alloc_va_range;
-	ppgtt->base.insert_entries = gen8_ppgtt_insert_entries;
-	ppgtt->base.clear_range = gen8_ppgtt_clear_range;
+	ppgtt->base.total = 1ULL << 32;
+	ppgtt->switch_mm = gen8_legacy_mm_switch;
+	ppgtt->base.cleanup = gen8_legacy_ppgtt_cleanup;
+	ppgtt->base.allocate_va_range = gen8_legacy_alloc_va_range;
+	ppgtt->base.insert_entries = gen8_legacy_ppgtt_insert_entries;
+	ppgtt->base.clear_range = gen8_legacy_ppgtt_clear_range;
+	ppgtt->debug_dump = gen8_legacy_ppgtt_dump;
 	ppgtt->base.unbind_vma = ppgtt_unbind_vma;
 	ppgtt->base.bind_vma = ppgtt_bind_vma;
-	ppgtt->debug_dump = gen8_dump_ppgtt;
 
-	if (USES_FULL_48BIT_PPGTT(ppgtt->base.dev)) {
-		ret = setup_px(ppgtt->base.dev, &ppgtt->pml4);
-		if (ret)
-			goto free_scratch;
+	ret = __pdp_init(ppgtt->base.dev, &ppgtt->pdp);
+	if (ret)
+		goto free_scratch;
 
-		gen8_initialize_pml4(&ppgtt->base, &ppgtt->pml4);
+	trace_i915_page_directory_pointer_entry_alloc(&ppgtt->base,
+						      0, 0,
+						      GEN8_PML4E_SHIFT);
 
-		ppgtt->base.total = 1ULL << 48;
-		ppgtt->switch_mm = gen8_48b_mm_switch;
-	} else {
-		ret = __pdp_init(ppgtt->base.dev, &ppgtt->pdp);
-		if (ret)
-			goto free_scratch;
+	return 0;
 
-		ppgtt->base.total = 1ULL << 32;
-		ppgtt->switch_mm = gen8_legacy_mm_switch;
-		trace_i915_page_directory_pointer_entry_alloc(&ppgtt->base,
-							      0, 0,
-							      GEN8_PML4E_SHIFT);
-	}
+free_scratch:
+	gen8_free_scratch(&ppgtt->base);
+	return ret;
+}
+
+/*
+ * GEN8 48-bit ppgtt programming is accomplished through 512 PDP registers
+ * per PML4, with a net effect resembling a 4-level page table in normal
+ * x86-64 terms. Each PDP represents 1GB of memory 512 * 512 * 512 * 512 * 4096
+ * = 256TB 48b address space.
+ *
+ */
+static int gen8_4lvl_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
+{
+	int ret;
+
+	ret = gen8_init_scratch(&ppgtt->base);
+	if (ret)
+		return ret;
+
+	ppgtt->base.start = 0;
+	ppgtt->base.total = 1ULL << 48;
+	ppgtt->switch_mm = gen8_48b_mm_switch;
+	ppgtt->base.cleanup = gen8_4lvl_ppgtt_cleanup;
+	ppgtt->base.allocate_va_range = gen8_4lvl_alloc_va_range;
+	ppgtt->base.insert_entries = gen8_4lvl_ppgtt_insert_entries;
+	ppgtt->base.clear_range = gen8_4lvl_ppgtt_clear_range;
+	ppgtt->debug_dump = gen8_4lvl_ppgtt_dump;
+	ppgtt->base.unbind_vma = ppgtt_unbind_vma;
+	ppgtt->base.bind_vma = ppgtt_bind_vma;
+
+	ret = setup_px(ppgtt->base.dev, &ppgtt->pml4);
+	if (ret)
+		goto free_scratch;
+
+	gen8_initialize_pml4(&ppgtt->base, &ppgtt->pml4);
 
 	return 0;
 
@@ -2071,8 +2122,10 @@ static int __hw_ppgtt_init(struct drm_device *dev, struct i915_hw_ppgtt *ppgtt)
 
 	if (INTEL_INFO(dev)->gen < 8)
 		return gen6_ppgtt_init(ppgtt);
+	else if (!USES_FULL_48BIT_PPGTT(ppgtt->base.dev))
+		return gen8_legacy_ppgtt_init(ppgtt);
 	else
-		return gen8_ppgtt_init(ppgtt);
+		return gen8_4lvl_ppgtt_init(ppgtt);
 }
 
 int i915_ppgtt_init(struct drm_device *dev, struct i915_hw_ppgtt *ppgtt)
Signature · 6 lines
2.5.0

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v1 1/4
#6 lore raw
In reply to Michel Thierry
Quoted 9 lines · Michel Thierry

Similar to commit c44ef60e437019b8ca1dab8b4d2e8761fd4ce1e9 ("drm/i915/gtt:
Allow >= 4GB sizes for vm"), i915_gem_obj_offset and i915_gem_obj_ggtt_offset return an unsigned long, which in only 4-bytes long in 32-bit kernels.

Change return type (and other related offset variables) to u64.

Since Global GTT is always limited to 4GB, this change would not be required in i915_gem_obj_ggtt_offset, but this is done for consistency.

Quoted 8 lines

v2: Remove unnecessary offset variable in do_pin, as we already have
vma->node.start (Chris).
Update GGTT offset too (Tvrtko).

Cc Chris Wilson <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Tvrtko Ursulin <[email protected]>

There are a few more places where we use ggtt_offset() and ggtt_bound() when we already have the vma, but they are only in the context of this patch and not the meat. They just make me cringe.

-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx