drm/i915/gtt: split up i915_gem_gtt

Attempt to split i915_gem_gtt.[ch] into more manageable chunks.
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20200107134009.3255354-1-chris@chris-wilson.co.uk

drivers/gpu/drm/i915/Makefile

@@ -74,6 +74,8 @@ gt-y += \
 	gt/debugfs_engines.o \
 	gt/debugfs_gt.o \
 	gt/debugfs_gt_pm.o \
+	gt/gen6_ppgtt.o \
+	gt/gen8_ppgtt.o \
 	gt/intel_breadcrumbs.o \
 	gt/intel_context.o \
 	gt/intel_engine_cs.o \
@@ -81,14 +83,17 @@ gt-y += \
 	gt/intel_engine_pm.o \
 	gt/intel_engine_pool.o \
 	gt/intel_engine_user.o \
+	gt/intel_ggtt.o \
 	gt/intel_gt.o \
 	gt/intel_gt_irq.o \
 	gt/intel_gt_pm.o \
 	gt/intel_gt_pm_irq.o \
 	gt/intel_gt_requests.o \
+	gt/intel_gtt.o \
 	gt/intel_llc.o \
 	gt/intel_lrc.o \
 	gt/intel_mocs.o \
+	gt/intel_ppgtt.o \
 	gt/intel_rc6.o \
 	gt/intel_renderstate.o \
 	gt/intel_reset.o \

drivers/gpu/drm/i915/gem/i915_gem_context.c

@@ -69,6 +69,7 @@
 
 #include <drm/i915_drm.h>
 
+#include "gt/gen6_ppgtt.h"
 #include "gt/intel_context.h"
 #include "gt/intel_engine_heartbeat.h"
 #include "gt/intel_engine_pm.h"
@@ -705,7 +706,7 @@ i915_gem_create_context(struct drm_i915_private *i915, unsigned int flags)
 	if (HAS_FULL_PPGTT(i915)) {
 		struct i915_ppgtt *ppgtt;
 
-		ppgtt = i915_ppgtt_create(i915);
+		ppgtt = i915_ppgtt_create(&i915->gt);
 		if (IS_ERR(ppgtt)) {
 			DRM_DEBUG_DRIVER("PPGTT setup failed (%ld)\n",
 					 PTR_ERR(ppgtt));
@@ -861,7 +862,7 @@ int i915_gem_vm_create_ioctl(struct drm_device *dev, void *data,
 	if (args->flags)
 		return -EINVAL;
 
-	ppgtt = i915_ppgtt_create(i915);
+	ppgtt = i915_ppgtt_create(&i915->gt);
 	if (IS_ERR(ppgtt))
 		return PTR_ERR(ppgtt);
 

drivers/gpu/drm/i915/gem/i915_gem_context.h

@@ -13,7 +13,6 @@
 
 #include "i915_drv.h"
 #include "i915_gem.h"
-#include "i915_gem_gtt.h"
 #include "i915_scheduler.h"
 #include "intel_device_info.h"
 

drivers/gpu/drm/i915/gem/i915_gem_object.h

@@ -16,6 +16,7 @@
 #include "display/intel_frontbuffer.h"
 #include "i915_gem_object_types.h"
 #include "i915_gem_gtt.h"
+#include "i915_vma_types.h"
 
 void i915_gem_init__objects(struct drm_i915_private *i915);
 

drivers/gpu/drm/i915/gem/selftests/huge_pages.c

@@ -1865,7 +1865,7 @@ int i915_gem_huge_page_mock_selftests(void)
 	mkwrite_device_info(dev_priv)->ppgtt_type = INTEL_PPGTT_FULL;
 	mkwrite_device_info(dev_priv)->ppgtt_size = 48;
 
-	ppgtt = i915_ppgtt_create(dev_priv);
+	ppgtt = i915_ppgtt_create(&dev_priv->gt);
 	if (IS_ERR(ppgtt)) {
 		err = PTR_ERR(ppgtt);
 		goto out_unlock;

drivers/gpu/drm/i915/gt/gen6_ppgtt.h

+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright © 2020 Intel Corporation
+ */
+
+#ifndef __GEN6_PPGTT_H__
+#define __GEN6_PPGTT_H__
+
+#include "intel_gtt.h"
+
+struct gen6_ppgtt {
+	struct i915_ppgtt base;
+
+	struct mutex flush;
+	struct i915_vma *vma;
+	gen6_pte_t __iomem *pd_addr;
+
+	atomic_t pin_count;
+	struct mutex pin_mutex;
+
+	bool scan_for_unused_pt;
+};
+
+static inline u32 gen6_pte_index(u32 addr)
+{
+	return i915_pte_index(addr, GEN6_PDE_SHIFT);
+}
+
+static inline u32 gen6_pte_count(u32 addr, u32 length)
+{
+	return i915_pte_count(addr, length, GEN6_PDE_SHIFT);
+}
+
+static inline u32 gen6_pde_index(u32 addr)
+{
+	return i915_pde_index(addr, GEN6_PDE_SHIFT);
+}
+
+#define __to_gen6_ppgtt(base) container_of(base, struct gen6_ppgtt, base)
+
+static inline struct gen6_ppgtt *to_gen6_ppgtt(struct i915_ppgtt *base)
+{
+	BUILD_BUG_ON(offsetof(struct gen6_ppgtt, base));
+	return __to_gen6_ppgtt(base);
+}
+
+/*
+ * gen6_for_each_pde() iterates over every pde from start until start+length.
+ * If start and start+length are not perfectly divisible, the macro will round
+ * down and up as needed. Start=0 and length=2G effectively iterates over
+ * every PDE in the system. The macro modifies ALL its parameters except 'pd',
+ * so each of the other parameters should preferably be a simple variable, or
+ * at most an lvalue with no side-effects!
+ */
+#define gen6_for_each_pde(pt, pd, start, length, iter)			\
+	for (iter = gen6_pde_index(start);				\
+	     length > 0 && iter < I915_PDES &&				\
+		     (pt = i915_pt_entry(pd, iter), true);		\
+	     ({ u32 temp = ALIGN(start+1, 1 << GEN6_PDE_SHIFT);		\
+		    temp = min(temp - start, length);			\
+		    start += temp, length -= temp; }), ++iter)
+
+#define gen6_for_all_pdes(pt, pd, iter)					\
+	for (iter = 0;							\
+	     iter < I915_PDES &&					\
+		     (pt = i915_pt_entry(pd, iter), true);		\
+	     ++iter)
+
+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);
+void gen6_ppgtt_enable(struct intel_gt *gt);
+void gen7_ppgtt_enable(struct intel_gt *gt);
+struct i915_ppgtt *gen6_ppgtt_create(struct intel_gt *gt);
+
+#endif

drivers/gpu/drm/i915/gt/gen8_ppgtt.h

+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright © 2020 Intel Corporation
+ */
+
+#ifndef __GEN8_PPGTT_H__
+#define __GEN8_PPGTT_H__
+
+struct intel_gt;
+
+struct i915_ppgtt *gen8_ppgtt_create(struct intel_gt *gt);
+
+#endif

drivers/gpu/drm/i915/gt/intel_gt.c

@@ -366,7 +366,7 @@ static void intel_gt_fini_scratch(struct intel_gt *gt)
 static struct i915_address_space *kernel_vm(struct intel_gt *gt)
 {
 	if (INTEL_PPGTT(gt->i915) > INTEL_PPGTT_ALIASING)
-		return &i915_ppgtt_create(gt->i915)->vm;
+		return &i915_ppgtt_create(gt)->vm;
 	else
 		return i915_vm_get(&gt->ggtt->vm);
 }

drivers/gpu/drm/i915/gt/intel_ppgtt.c

+// SPDX-License-Identifier: MIT
+/*
+ * Copyright © 2020 Intel Corporation
+ */
+
+#include <linux/slab.h>
+
+#include "i915_trace.h"
+#include "intel_gtt.h"
+#include "gen6_ppgtt.h"
+#include "gen8_ppgtt.h"
+
+struct i915_page_table *alloc_pt(struct i915_address_space *vm)
+{
+	struct i915_page_table *pt;
+
+	pt = kmalloc(sizeof(*pt), I915_GFP_ALLOW_FAIL);
+	if (unlikely(!pt))
+		return ERR_PTR(-ENOMEM);
+
+	if (unlikely(setup_page_dma(vm, &pt->base))) {
+		kfree(pt);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	atomic_set(&pt->used, 0);
+	return pt;
+}
+
+struct i915_page_directory *__alloc_pd(size_t sz)
+{
+	struct i915_page_directory *pd;
+
+	pd = kzalloc(sz, I915_GFP_ALLOW_FAIL);
+	if (unlikely(!pd))
+		return NULL;
+
+	spin_lock_init(&pd->lock);
+	return pd;
+}
+
+struct i915_page_directory *alloc_pd(struct i915_address_space *vm)
+{
+	struct i915_page_directory *pd;
+
+	pd = __alloc_pd(sizeof(*pd));
+	if (unlikely(!pd))
+		return ERR_PTR(-ENOMEM);
+
+	if (unlikely(setup_page_dma(vm, px_base(pd)))) {
+		kfree(pd);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	return pd;
+}
+
+void free_pd(struct i915_address_space *vm, struct i915_page_dma *pd)
+{
+	cleanup_page_dma(vm, pd);
+	kfree(pd);
+}
+
+static inline void
+write_dma_entry(struct i915_page_dma * const pdma,
+		const unsigned short idx,
+		const u64 encoded_entry)
+{
+	u64 * const vaddr = kmap_atomic(pdma->page);
+
+	vaddr[idx] = encoded_entry;
+	kunmap_atomic(vaddr);
+}
+
+void
+__set_pd_entry(struct i915_page_directory * const pd,
+	       const unsigned short idx,
+	       struct i915_page_dma * const to,
+	       u64 (*encode)(const dma_addr_t, const enum i915_cache_level))
+{
+	/* Each thread pre-pins the pd, and we may have a thread per pde. */
+	GEM_BUG_ON(atomic_read(px_used(pd)) > NALLOC * ARRAY_SIZE(pd->entry));
+
+	atomic_inc(px_used(pd));
+	pd->entry[idx] = to;
+	write_dma_entry(px_base(pd), idx, encode(to->daddr, I915_CACHE_LLC));
+}
+
+void
+clear_pd_entry(struct i915_page_directory * const pd,
+	       const unsigned short idx,
+	       const struct i915_page_scratch * const scratch)
+{
+	GEM_BUG_ON(atomic_read(px_used(pd)) == 0);
+
+	write_dma_entry(px_base(pd), idx, scratch->encode);
+	pd->entry[idx] = NULL;
+	atomic_dec(px_used(pd));
+}
+
+bool
+release_pd_entry(struct i915_page_directory * const pd,
+		 const unsigned short idx,
+		 struct i915_page_table * const pt,
+		 const struct i915_page_scratch * const scratch)
+{
+	bool free = false;
+
+	if (atomic_add_unless(&pt->used, -1, 1))
+		return false;
+
+	spin_lock(&pd->lock);
+	if (atomic_dec_and_test(&pt->used)) {
+		clear_pd_entry(pd, idx, scratch);
+		free = true;
+	}
+	spin_unlock(&pd->lock);
+
+	return free;
+}
+
+int i915_ppgtt_init_hw(struct intel_gt *gt)
+{
+	struct drm_i915_private *i915 = gt->i915;
+
+	gtt_write_workarounds(gt);
+
+	if (IS_GEN(i915, 6))
+		gen6_ppgtt_enable(gt);
+	else if (IS_GEN(i915, 7))
+		gen7_ppgtt_enable(gt);
+
+	return 0;
+}
+
+static struct i915_ppgtt *
+__ppgtt_create(struct intel_gt *gt)
+{
+	if (INTEL_GEN(gt->i915) < 8)
+		return gen6_ppgtt_create(gt);
+	else
+		return gen8_ppgtt_create(gt);
+}
+
+struct i915_ppgtt *i915_ppgtt_create(struct intel_gt *gt)
+{
+	struct i915_ppgtt *ppgtt;
+
+	ppgtt = __ppgtt_create(gt);
+	if (IS_ERR(ppgtt))
+		return ppgtt;
+
+	trace_i915_ppgtt_create(&ppgtt->vm);
+
+	return ppgtt;
+}
+
+static int ppgtt_bind_vma(struct i915_vma *vma,
+			  enum i915_cache_level cache_level,
+			  u32 flags)
+{
+	u32 pte_flags;
+	int err;
+
+	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+ */
+	pte_flags = 0;
+	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();
+
+	return 0;
+}
+
+static void ppgtt_unbind_vma(struct i915_vma *vma)
+{
+	if (test_and_clear_bit(I915_VMA_ALLOC_BIT, __i915_vma_flags(vma)))
+		vma->vm->clear_range(vma->vm, vma->node.start, vma->size);
+}
+
+int ppgtt_set_pages(struct i915_vma *vma)
+{
+	GEM_BUG_ON(vma->pages);
+
+	vma->pages = vma->obj->mm.pages;
+
+	vma->page_sizes = vma->obj->mm.page_sizes;
+
+	return 0;
+}
+
+void ppgtt_init(struct i915_ppgtt *ppgtt, struct intel_gt *gt)
+{
+	struct drm_i915_private *i915 = gt->i915;
+
+	ppgtt->vm.gt = gt;
+	ppgtt->vm.i915 = i915;
+	ppgtt->vm.dma = &i915->drm.pdev->dev;
+	ppgtt->vm.total = BIT_ULL(INTEL_INFO(i915)->ppgtt_size);
+
+	i915_address_space_init(&ppgtt->vm, VM_CLASS_PPGTT);
+
+	ppgtt->vm.vma_ops.bind_vma    = ppgtt_bind_vma;
+	ppgtt->vm.vma_ops.unbind_vma  = ppgtt_unbind_vma;
+	ppgtt->vm.vma_ops.set_pages   = ppgtt_set_pages;
+	ppgtt->vm.vma_ops.clear_pages = clear_pages;
+}

drivers/gpu/drm/i915/gt/intel_ring_submission.c

@@ -33,6 +33,7 @@
 
 #include "gem/i915_gem_context.h"
 
+#include "gen6_ppgtt.h"
 #include "i915_drv.h"
 #include "i915_trace.h"
 #include "intel_context.h"

drivers/gpu/drm/i915/gt/selftest_hangcheck.c

@@ -1312,7 +1312,7 @@ static int igt_reset_evict_ppgtt(void *arg)
 	if (INTEL_PPGTT(gt->i915) < INTEL_PPGTT_FULL)
 		return 0;
 
-	ppgtt = i915_ppgtt_create(gt->i915);
+	ppgtt = i915_ppgtt_create(gt);
 	if (IS_ERR(ppgtt))
 		return PTR_ERR(ppgtt);
 

drivers/gpu/drm/i915/gvt/scheduler.c

@@ -1224,7 +1224,7 @@ int intel_vgpu_setup_submission(struct intel_vgpu *vgpu)
 	enum intel_engine_id i;
 	int ret;
 
-	ppgtt = i915_ppgtt_create(i915);
+	ppgtt = i915_ppgtt_create(&i915->gt);
 	if (IS_ERR(ppgtt))
 		return PTR_ERR(ppgtt);
 

drivers/gpu/drm/i915/selftests/mock_gtt.c

@@ -55,6 +55,11 @@ static void mock_cleanup(struct i915_address_space *vm)
 {
 }
 
+static void mock_clear_range(struct i915_address_space *vm,
+			     u64 start, u64 length)
+{
+}
+
 struct i915_ppgtt *mock_ppgtt(struct drm_i915_private *i915, const char *name)
 {
 	struct i915_ppgtt *ppgtt;
@@ -70,7 +75,7 @@ struct i915_ppgtt *mock_ppgtt(struct drm_i915_private *i915, const char *name)
 
 	i915_address_space_init(&ppgtt->vm, VM_CLASS_PPGTT);
 
-	ppgtt->vm.clear_range = nop_clear_range;
+	ppgtt->vm.clear_range = mock_clear_range;
 	ppgtt->vm.insert_page = mock_insert_page;
 	ppgtt->vm.insert_entries = mock_insert_entries;
 	ppgtt->vm.cleanup = mock_cleanup;
@@ -107,7 +112,7 @@ void mock_init_ggtt(struct drm_i915_private *i915, struct i915_ggtt *ggtt)
 	ggtt->mappable_end = resource_size(&ggtt->gmadr);
 	ggtt->vm.total = 4096 * PAGE_SIZE;
 
-	ggtt->vm.clear_range = nop_clear_range;
+	ggtt->vm.clear_range = mock_clear_range;
 	ggtt->vm.insert_page = mock_insert_page;
 	ggtt->vm.insert_entries = mock_insert_entries;
 	ggtt->vm.cleanup = mock_cleanup;