drm/i915: Move phys objects to its own file

Continuing the decluttering of i915_gem.c, this time the legacy physical
object.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190528092956.14910-5-chris@chris-wilson.co.uk

drivers/gpu/drm/i915/Makefile

@@ -88,6 +88,8 @@ i915-y += $(gt-y)
 obj-y += gem/
 gem-y += \
 	gem/i915_gem_object.o \
+	gem/i915_gem_pages.o \
+	gem/i915_gem_phys.o \
 	gem/i915_gem_shmem.o
 i915-y += \
 	  $(gem-y) \

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

@@ -33,11 +33,17 @@ void __i915_gem_object_release_shmem(struct drm_i915_gem_object *obj,
 				     struct sg_table *pages,
 				     bool needs_clflush);
 
+int i915_gem_object_attach_phys(struct drm_i915_gem_object *obj, int align);
+
 void i915_gem_close_object(struct drm_gem_object *gem, struct drm_file *file);
 void i915_gem_free_object(struct drm_gem_object *obj);
 
 void i915_gem_flush_free_objects(struct drm_i915_private *i915);
 
+struct sg_table *
+__i915_gem_object_unset_pages(struct drm_i915_gem_object *obj);
+void i915_gem_object_truncate(struct drm_i915_gem_object *obj);
+
 /**
  * i915_gem_object_lookup_rcu - look up a temporary GEM object from its handle
  * @filp: DRM file private date
@@ -236,6 +242,8 @@ i915_gem_object_get_dma_address(struct drm_i915_gem_object *obj,
 void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj,
 				 struct sg_table *pages,
 				 unsigned int sg_page_sizes);
+
+int ____i915_gem_object_get_pages(struct drm_i915_gem_object *obj);
 int __i915_gem_object_get_pages(struct drm_i915_gem_object *obj);
 
 static inline int __must_check
@@ -291,7 +299,8 @@ enum i915_mm_subclass { /* lockdep subclass for obj->mm.lock/struct_mutex */
 
 int __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
 				enum i915_mm_subclass subclass);
-void __i915_gem_object_truncate(struct drm_i915_gem_object *obj);
+void i915_gem_object_truncate(struct drm_i915_gem_object *obj);
+void i915_gem_object_writeback(struct drm_i915_gem_object *obj);
 
 enum i915_map_type {
 	I915_MAP_WB = 0,

drivers/gpu/drm/i915/gem/i915_gem_object_types.h

@@ -52,6 +52,8 @@ struct drm_i915_gem_object_ops {
 	int (*get_pages)(struct drm_i915_gem_object *obj);
 	void (*put_pages)(struct drm_i915_gem_object *obj,
 			  struct sg_table *pages);
+	void (*truncate)(struct drm_i915_gem_object *obj);
+	void (*writeback)(struct drm_i915_gem_object *obj);
 
 	int (*pwrite)(struct drm_i915_gem_object *obj,
 		      const struct drm_i915_gem_pwrite *arg);

drivers/gpu/drm/i915/gem/i915_gem_phys.c

+/*
+ * SPDX-License-Identifier: MIT
+ *
+ * Copyright © 2014-2016 Intel Corporation
+ */
+
+#include <linux/highmem.h>
+#include <linux/shmem_fs.h>
+#include <linux/swap.h>
+
+#include <drm/drm.h> /* for drm_legacy.h! */
+#include <drm/drm_cache.h>
+#include <drm/drm_legacy.h> /* for drm_pci.h! */
+#include <drm/drm_pci.h>
+
+#include "i915_drv.h"
+#include "i915_gem_object.h"
+
+static int i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj)
+{
+	struct address_space *mapping = obj->base.filp->f_mapping;
+	struct drm_dma_handle *phys;
+	struct sg_table *st;
+	struct scatterlist *sg;
+	char *vaddr;
+	int i;
+	int err;
+
+	if (WARN_ON(i915_gem_object_needs_bit17_swizzle(obj)))
+		return -EINVAL;
+
+	/* Always aligning to the object size, allows a single allocation
+	 * to handle all possible callers, and given typical object sizes,
+	 * the alignment of the buddy allocation will naturally match.
+	 */
+	phys = drm_pci_alloc(obj->base.dev,
+			     roundup_pow_of_two(obj->base.size),
+			     roundup_pow_of_two(obj->base.size));
+	if (!phys)
+		return -ENOMEM;
+
+	vaddr = phys->vaddr;
+	for (i = 0; i < obj->base.size / PAGE_SIZE; i++) {
+		struct page *page;
+		char *src;
+
+		page = shmem_read_mapping_page(mapping, i);
+		if (IS_ERR(page)) {
+			err = PTR_ERR(page);
+			goto err_phys;
+		}
+
+		src = kmap_atomic(page);
+		memcpy(vaddr, src, PAGE_SIZE);
+		drm_clflush_virt_range(vaddr, PAGE_SIZE);
+		kunmap_atomic(src);
+
+		put_page(page);
+		vaddr += PAGE_SIZE;
+	}
+
+	i915_gem_chipset_flush(to_i915(obj->base.dev));
+
+	st = kmalloc(sizeof(*st), GFP_KERNEL);
+	if (!st) {
+		err = -ENOMEM;
+		goto err_phys;
+	}
+
+	if (sg_alloc_table(st, 1, GFP_KERNEL)) {
+		kfree(st);
+		err = -ENOMEM;
+		goto err_phys;
+	}
+
+	sg = st->sgl;
+	sg->offset = 0;
+	sg->length = obj->base.size;
+
+	sg_dma_address(sg) = phys->busaddr;
+	sg_dma_len(sg) = obj->base.size;
+
+	obj->phys_handle = phys;
+
+	__i915_gem_object_set_pages(obj, st, sg->length);
+
+	return 0;
+
+err_phys:
+	drm_pci_free(obj->base.dev, phys);
+
+	return err;
+}
+
+static void
+i915_gem_object_put_pages_phys(struct drm_i915_gem_object *obj,
+			       struct sg_table *pages)
+{
+	__i915_gem_object_release_shmem(obj, pages, false);
+
+	if (obj->mm.dirty) {
+		struct address_space *mapping = obj->base.filp->f_mapping;
+		char *vaddr = obj->phys_handle->vaddr;
+		int i;
+
+		for (i = 0; i < obj->base.size / PAGE_SIZE; i++) {
+			struct page *page;
+			char *dst;
+
+			page = shmem_read_mapping_page(mapping, i);
+			if (IS_ERR(page))
+				continue;
+
+			dst = kmap_atomic(page);
+			drm_clflush_virt_range(vaddr, PAGE_SIZE);
+			memcpy(dst, vaddr, PAGE_SIZE);
+			kunmap_atomic(dst);
+
+			set_page_dirty(page);
+			if (obj->mm.madv == I915_MADV_WILLNEED)
+				mark_page_accessed(page);
+			put_page(page);
+			vaddr += PAGE_SIZE;
+		}
+		obj->mm.dirty = false;
+	}
+
+	sg_free_table(pages);
+	kfree(pages);
+
+	drm_pci_free(obj->base.dev, obj->phys_handle);
+}
+
+static void
+i915_gem_object_release_phys(struct drm_i915_gem_object *obj)
+{
+	i915_gem_object_unpin_pages(obj);
+}
+
+static const struct drm_i915_gem_object_ops i915_gem_phys_ops = {
+	.get_pages = i915_gem_object_get_pages_phys,
+	.put_pages = i915_gem_object_put_pages_phys,
+	.release = i915_gem_object_release_phys,
+};
+
+int i915_gem_object_attach_phys(struct drm_i915_gem_object *obj, int align)
+{
+	struct sg_table *pages;
+	int err;
+
+	if (align > obj->base.size)
+		return -EINVAL;
+
+	if (obj->ops == &i915_gem_phys_ops)
+		return 0;
+
+	if (obj->ops != &i915_gem_shmem_ops)
+		return -EINVAL;
+
+	err = i915_gem_object_unbind(obj);
+	if (err)
+		return err;
+
+	mutex_lock(&obj->mm.lock);
+
+	if (obj->mm.madv != I915_MADV_WILLNEED) {
+		err = -EFAULT;
+		goto err_unlock;
+	}
+
+	if (obj->mm.quirked) {
+		err = -EFAULT;
+		goto err_unlock;
+	}
+
+	if (obj->mm.mapping) {
+		err = -EBUSY;
+		goto err_unlock;
+	}
+
+	pages = __i915_gem_object_unset_pages(obj);
+
+	obj->ops = &i915_gem_phys_ops;
+
+	err = ____i915_gem_object_get_pages(obj);
+	if (err)
+		goto err_xfer;
+
+	/* Perma-pin (until release) the physical set of pages */
+	__i915_gem_object_pin_pages(obj);
+
+	if (!IS_ERR_OR_NULL(pages))
+		i915_gem_shmem_ops.put_pages(obj, pages);
+	mutex_unlock(&obj->mm.lock);
+	return 0;
+
+err_xfer:
+	obj->ops = &i915_gem_shmem_ops;
+	if (!IS_ERR_OR_NULL(pages)) {
+		unsigned int sg_page_sizes = i915_sg_page_sizes(pages->sgl);
+
+		__i915_gem_object_set_pages(obj, pages, sg_page_sizes);
+	}
+err_unlock:
+	mutex_unlock(&obj->mm.lock);
+	return err;
+}
+
+#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
+#include "selftests/i915_gem_phys.c"
+#endif

drivers/gpu/drm/i915/gem/i915_gem_shmem.c

@@ -213,6 +213,65 @@ static int shmem_get_pages(struct drm_i915_gem_object *obj)
 	return ret;
 }
 
+static void
+shmem_truncate(struct drm_i915_gem_object *obj)
+{
+	/*
+	 * Our goal here is to return as much of the memory as
+	 * is possible back to the system as we are called from OOM.
+	 * To do this we must instruct the shmfs to drop all of its
+	 * backing pages, *now*.
+	 */
+	shmem_truncate_range(file_inode(obj->base.filp), 0, (loff_t)-1);
+	obj->mm.madv = __I915_MADV_PURGED;
+	obj->mm.pages = ERR_PTR(-EFAULT);
+}
+
+static void
+shmem_writeback(struct drm_i915_gem_object *obj)
+{
+	struct address_space *mapping;
+	struct writeback_control wbc = {
+		.sync_mode = WB_SYNC_NONE,
+		.nr_to_write = SWAP_CLUSTER_MAX,
+		.range_start = 0,
+		.range_end = LLONG_MAX,
+		.for_reclaim = 1,
+	};
+	unsigned long i;
+
+	/*
+	 * Leave mmapings intact (GTT will have been revoked on unbinding,
+	 * leaving only CPU mmapings around) and add those pages to the LRU
+	 * instead of invoking writeback so they are aged and paged out
+	 * as normal.
+	 */
+	mapping = obj->base.filp->f_mapping;
+
+	/* Begin writeback on each dirty page */
+	for (i = 0; i < obj->base.size >> PAGE_SHIFT; i++) {
+		struct page *page;
+
+		page = find_lock_entry(mapping, i);
+		if (!page || xa_is_value(page))
+			continue;
+
+		if (!page_mapped(page) && clear_page_dirty_for_io(page)) {
+			int ret;
+
+			SetPageReclaim(page);
+			ret = mapping->a_ops->writepage(page, &wbc);
+			if (!PageWriteback(page))
+				ClearPageReclaim(page);
+			if (!ret)
+				goto put;
+		}
+		unlock_page(page);
+put:
+		put_page(page);
+	}
+}
+
 void
 __i915_gem_object_release_shmem(struct drm_i915_gem_object *obj,
 				struct sg_table *pages,
@@ -362,6 +421,8 @@ const struct drm_i915_gem_object_ops i915_gem_shmem_ops = {
 
 	.get_pages = shmem_get_pages,
 	.put_pages = shmem_put_pages,
+	.truncate = shmem_truncate,
+	.writeback = shmem_writeback,
 
 	.pwrite = shmem_pwrite,
 };

drivers/gpu/drm/i915/gem/selftests/i915_gem_phys.c

+/*
+ * SPDX-License-Identifier: MIT
+ *
+ * Copyright © 2016 Intel Corporation
+ */
+
+#include "i915_selftest.h"
+
+#include "selftests/mock_gem_device.h"
+
+static int mock_phys_object(void *arg)
+{
+	struct drm_i915_private *i915 = arg;
+	struct drm_i915_gem_object *obj;
+	int err;
+
+	/* Create an object and bind it to a contiguous set of physical pages,
+	 * i.e. exercise the i915_gem_object_phys API.
+	 */
+
+	obj = i915_gem_object_create_shmem(i915, PAGE_SIZE);
+	if (IS_ERR(obj)) {
+		err = PTR_ERR(obj);
+		pr_err("i915_gem_object_create failed, err=%d\n", err);
+		goto out;
+	}
+
+	mutex_lock(&i915->drm.struct_mutex);
+	err = i915_gem_object_attach_phys(obj, PAGE_SIZE);
+	mutex_unlock(&i915->drm.struct_mutex);
+	if (err) {
+		pr_err("i915_gem_object_attach_phys failed, err=%d\n", err);
+		goto out_obj;
+	}
+
+	if (obj->ops != &i915_gem_phys_ops) {
+		pr_err("i915_gem_object_attach_phys did not create a phys object\n");
+		err = -EINVAL;
+		goto out_obj;
+	}
+
+	if (!atomic_read(&obj->mm.pages_pin_count)) {
+		pr_err("i915_gem_object_attach_phys did not pin its phys pages\n");
+		err = -EINVAL;
+		goto out_obj;
+	}
+
+	/* Make the object dirty so that put_pages must do copy back the data */
+	mutex_lock(&i915->drm.struct_mutex);
+	err = i915_gem_object_set_to_gtt_domain(obj, true);
+	mutex_unlock(&i915->drm.struct_mutex);
+	if (err) {
+		pr_err("i915_gem_object_set_to_gtt_domain failed with err=%d\n",
+		       err);
+		goto out_obj;
+	}
+
+out_obj:
+	i915_gem_object_put(obj);
+out:
+	return err;
+}
+
+int i915_gem_phys_mock_selftests(void)
+{
+	static const struct i915_subtest tests[] = {
+		SUBTEST(mock_phys_object),
+	};
+	struct drm_i915_private *i915;
+	int err;
+
+	i915 = mock_gem_device();
+	if (!i915)
+		return -ENOMEM;
+
+	err = i915_subtests(tests, i915);
+
+	drm_dev_put(&i915->drm);
+	return err;
+}

drivers/gpu/drm/i915/i915_drv.h

@@ -2903,8 +2903,6 @@ i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj,
 				     const struct i915_ggtt_view *view,
 				     unsigned int flags);
 void i915_gem_object_unpin_from_display_plane(struct i915_vma *vma);
-int i915_gem_object_attach_phys(struct drm_i915_gem_object *obj,
-				int align);
 int i915_gem_open(struct drm_i915_private *i915, struct drm_file *file);
 void i915_gem_release(struct drm_device *dev, struct drm_file *file);
 

drivers/gpu/drm/i915/i915_gem_shrinker.c

@@ -114,65 +114,18 @@ static bool unsafe_drop_pages(struct drm_i915_gem_object *obj)
 	return !i915_gem_object_has_pages(obj);
 }
 
-static void __start_writeback(struct drm_i915_gem_object *obj,
+static void try_to_writeback(struct drm_i915_gem_object *obj,
 			     unsigned int flags)
 {
-	struct address_space *mapping;
-	struct writeback_control wbc = {
-		.sync_mode = WB_SYNC_NONE,
-		.nr_to_write = SWAP_CLUSTER_MAX,
-		.range_start = 0,
-		.range_end = LLONG_MAX,
-		.for_reclaim = 1,
-	};
-	unsigned long i;
-
-	lockdep_assert_held(&obj->mm.lock);
-	GEM_BUG_ON(i915_gem_object_has_pages(obj));
-
 	switch (obj->mm.madv) {
 	case I915_MADV_DONTNEED:
-		__i915_gem_object_truncate(obj);
+		i915_gem_object_truncate(obj);
 	case __I915_MADV_PURGED:
 		return;
 	}
 
-	if (!obj->base.filp)
-		return;
-
-	if (!(flags & I915_SHRINK_WRITEBACK))
-		return;
-
-	/*
-	 * Leave mmapings intact (GTT will have been revoked on unbinding,
-	 * leaving only CPU mmapings around) and add those pages to the LRU
-	 * instead of invoking writeback so they are aged and paged out
-	 * as normal.
-	 */
-	mapping = obj->base.filp->f_mapping;
-
-	/* Begin writeback on each dirty page */
-	for (i = 0; i < obj->base.size >> PAGE_SHIFT; i++) {
-		struct page *page;
-
-		page = find_lock_entry(mapping, i);
-		if (!page || xa_is_value(page))
-			continue;
-
-		if (!page_mapped(page) && clear_page_dirty_for_io(page)) {
-			int ret;
-
-			SetPageReclaim(page);
-			ret = mapping->a_ops->writepage(page, &wbc);
-			if (!PageWriteback(page))
-				ClearPageReclaim(page);
-			if (!ret)
-				goto put;
-		}
-		unlock_page(page);
-put:
-		put_page(page);
-	}
+	if (flags & I915_SHRINK_WRITEBACK)
+		i915_gem_object_writeback(obj);
 }
 
 /**
@@ -315,7 +268,7 @@ i915_gem_shrink(struct drm_i915_private *i915,
 				mutex_lock_nested(&obj->mm.lock,
 						  I915_MM_SHRINKER);
 				if (!i915_gem_object_has_pages(obj)) {
-					__start_writeback(obj, flags);
+					try_to_writeback(obj, flags);
 					count += obj->base.size >> PAGE_SHIFT;
 				}
 				mutex_unlock(&obj->mm.lock);

drivers/gpu/drm/i915/selftests/i915_gem_object.c

@@ -49,59 +49,6 @@ static int igt_gem_object(void *arg)
 	return err;
 }
 
-static int igt_phys_object(void *arg)
-{
-	struct drm_i915_private *i915 = arg;
-	struct drm_i915_gem_object *obj;
-	int err;
-
-	/* Create an object and bind it to a contiguous set of physical pages,
-	 * i.e. exercise the i915_gem_object_phys API.
-	 */
-
-	obj = i915_gem_object_create_shmem(i915, PAGE_SIZE);
-	if (IS_ERR(obj)) {
-		err = PTR_ERR(obj);
-		pr_err("i915_gem_object_create failed, err=%d\n", err);
-		goto out;
-	}
-
-	mutex_lock(&i915->drm.struct_mutex);
-	err = i915_gem_object_attach_phys(obj, PAGE_SIZE);
-	mutex_unlock(&i915->drm.struct_mutex);
-	if (err) {
-		pr_err("i915_gem_object_attach_phys failed, err=%d\n", err);
-		goto out_obj;
-	}
-
-	if (obj->ops != &i915_gem_phys_ops) {
-		pr_err("i915_gem_object_attach_phys did not create a phys object\n");
-		err = -EINVAL;
-		goto out_obj;
-	}
-
-	if (!atomic_read(&obj->mm.pages_pin_count)) {
-		pr_err("i915_gem_object_attach_phys did not pin its phys pages\n");
-		err = -EINVAL;
-		goto out_obj;
-	}
-
-	/* Make the object dirty so that put_pages must do copy back the data */
-	mutex_lock(&i915->drm.struct_mutex);
-	err = i915_gem_object_set_to_gtt_domain(obj, true);
-	mutex_unlock(&i915->drm.struct_mutex);
-	if (err) {
-		pr_err("i915_gem_object_set_to_gtt_domain failed with err=%d\n",
-		       err);
-		goto out_obj;
-	}
-
-out_obj:
-	i915_gem_object_put(obj);
-out:
-	return err;
-}
-
 static int igt_gem_huge(void *arg)
 {
 	const unsigned int nreal = 509; /* just to be awkward */
@@ -631,7 +578,6 @@ int i915_gem_object_mock_selftests(void)
 {
 	static const struct i915_subtest tests[] = {
 		SUBTEST(igt_gem_object),
-		SUBTEST(igt_phys_object),
 	};
 	struct drm_i915_private *i915;
 	int err;

drivers/gpu/drm/i915/selftests/i915_mock_selftests.h

@@ -18,6 +18,7 @@ selftest(engine, intel_engine_cs_mock_selftests)
 selftest(timelines, i915_timeline_mock_selftests)
 selftest(requests, i915_request_mock_selftests)
 selftest(objects, i915_gem_object_mock_selftests)
+selftest(phys, i915_gem_phys_mock_selftests)
 selftest(dmabuf, i915_gem_dmabuf_mock_selftests)
 selftest(vma, i915_vma_mock_selftests)
 selftest(evict, i915_gem_evict_mock_selftests)