Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Oleksandr Andrushchenko <andr2000@gmail.com>
To:     xen-devel@lists.xenproject.org, linux-kernel@vger.kernel.org,
        dri-devel@lists.freedesktop.org, linux-media@vger.kernel.org,
        jgross@suse.com, boris.ostrovsky@oracle.com, konrad.wilk@oracle.com
Cc:     daniel.vetter@intel.com, andr2000@gmail.com, dongwon.kim@intel.com,
        matthew.d.roper@intel.com,
        Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Subject: [Xen-devel][RFC 3/3] xen/gntdev: Add support for Linux dma buffers
Date:   Thu, 17 May 2018 11:26:04 +0300
Message-Id: <20180517082604.14828-4-andr2000@gmail.com>
In-Reply-To: <20180517082604.14828-1-andr2000@gmail.com>
References: <20180517082604.14828-1-andr2000@gmail.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

From: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>

Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
---
 drivers/xen/gntdev.c      | 954 +++++++++++++++++++++++++++++++++++++-
 include/uapi/xen/gntdev.h | 101 ++++
 include/xen/gntdev_exp.h  |  23 +
 3 files changed, 1066 insertions(+), 12 deletions(-)
 create mode 100644 include/xen/gntdev_exp.h

diff --git a/drivers/xen/gntdev.c b/drivers/xen/gntdev.c
index 9510f228efe9..0ee88e193362 100644
--- a/drivers/xen/gntdev.c
+++ b/drivers/xen/gntdev.c
@@ -4,6 +4,8 @@
  * Device for accessing (in user-space) pages that have been granted by other
  * domains.
  *
+ * DMA buffer implementation is based on drivers/gpu/drm/drm_prime.c.
+ *
  * Copyright (c) 2006-2007, D G Murray.
  *           (c) 2009 Gerd Hoffmann <kraxel@redhat.com>
  *
@@ -37,6 +39,9 @@
 #include <linux/highmem.h>
 #include <linux/refcount.h>
 
+#include <linux/dma-buf.h>
+#include <linux/of_device.h>
+
 #include <xen/xen.h>
 #include <xen/grant_table.h>
 #include <xen/balloon.h>
@@ -61,16 +66,39 @@ static atomic_t pages_mapped = ATOMIC_INIT(0);
 static int use_ptemod;
 #define populate_freeable_maps use_ptemod
 
+#ifndef GRANT_INVALID_REF
+/*
+ * Note on usage of grant reference 0 as invalid grant reference:
+ * grant reference 0 is valid, but never exposed to a driver,
+ * because of the fact it is already in use/reserved by the PV console.
+ */
+#define GRANT_INVALID_REF	0
+#endif
+
 struct gntdev_priv {
 	/* maps with visible offsets in the file descriptor */
 	struct list_head maps;
 	/* maps that are not visible; will be freed on munmap.
 	 * Only populated if populate_freeable_maps == 1 */
 	struct list_head freeable_maps;
+	/* List of dma-bufs. */
+	struct list_head dma_bufs;
 	/* lock protects maps and freeable_maps */
 	struct mutex lock;
 	struct mm_struct *mm;
 	struct mmu_notifier mn;
+
+	/* Private data of the hyper DMA buffers. */
+
+	struct device *dev;
+	/* List of exported DMA buffers. */
+	struct list_head dmabuf_exp_list;
+	/* List of wait objects. */
+	struct list_head dmabuf_exp_wait_list;
+	/* List of imported DMA buffers. */
+	struct list_head dmabuf_imp_list;
+	/* This is the lock which protects dma_buf_xxx lists. */
+	struct mutex dmabuf_lock;
 };
 
 struct unmap_notify {
@@ -95,10 +123,65 @@ struct grant_map {
 	struct gnttab_unmap_grant_ref *kunmap_ops;
 	struct page **pages;
 	unsigned long pages_vm_start;
+
+	/*
+	 * All the fields starting with dmabuf_ are only valid if this
+	 * mapping is used for exporting a DMA buffer.
+	 * If dmabuf_vaddr is not NULL then this mapping is backed by DMA
+	 * capable memory.
+	 */
+
+	/* Flags used to create this DMA buffer: GNTDEV_DMABUF_FLAG_XXX. */
+	bool dmabuf_flags;
+	/* Virtual/CPU address of the DMA buffer. */
+	void *dmabuf_vaddr;
+	/* Bus address of the DMA buffer. */
+	dma_addr_t dmabuf_bus_addr;
+};
+
+struct hyper_dmabuf {
+	struct gntdev_priv *priv;
+	struct dma_buf *dmabuf;
+	struct list_head next;
+	int fd;
+
+	union {
+		struct {
+			/* Exported buffers are reference counted. */
+			struct kref refcount;
+			struct grant_map *map;
+		} exp;
+		struct {
+			/* Granted references of the imported buffer. */
+			grant_ref_t *refs;
+			/* Scatter-gather table of the imported buffer. */
+			struct sg_table *sgt;
+			/* dma-buf attachment of the imported buffer. */
+			struct dma_buf_attachment *attach;
+		} imp;
+	} u;
+
+	/* Number of pages this buffer has. */
+	int nr_pages;
+	/* Pages of this buffer. */
+	struct page **pages;
+};
+
+struct hyper_dmabuf_wait_obj {
+	struct list_head next;
+	struct hyper_dmabuf *hyper_dmabuf;
+	struct completion completion;
+};
+
+struct hyper_dambuf_attachment {
+	struct sg_table *sgt;
+	enum dma_data_direction dir;
 };
 
 static int unmap_grant_pages(struct grant_map *map, int offset, int pages);
 
+static struct miscdevice gntdev_miscdev;
+
 /* ------------------------------------------------------------------ */
 
 static void gntdev_print_maps(struct gntdev_priv *priv,
@@ -120,8 +203,17 @@ static void gntdev_free_map(struct grant_map *map)
 	if (map == NULL)
 		return;
 
-	if (map->pages)
+	if (map->dmabuf_vaddr) {
+		bool coherent = map->dmabuf_flags &
+			GNTDEV_DMABUF_FLAG_DMA_COHERENT;
+
+		gnttab_dma_free_pages(gntdev_miscdev.this_device,
+				      coherent, map->count, map->pages,
+				      map->dmabuf_vaddr, map->dmabuf_bus_addr);
+	} else if (map->pages) {
 		gnttab_free_pages(map->count, map->pages);
+	}
+
 	kfree(map->pages);
 	kfree(map->grants);
 	kfree(map->map_ops);
@@ -131,7 +223,7 @@ static void gntdev_free_map(struct grant_map *map)
 	kfree(map);
 }
 
-static struct grant_map *gntdev_alloc_map(struct gntdev_priv *priv, int count)
+static struct grant_map *gntdev_alloc_map(int count, int dmabuf_flags)
 {
 	struct grant_map *add;
 	int i;
@@ -154,8 +246,26 @@ static struct grant_map *gntdev_alloc_map(struct gntdev_priv *priv, int count)
 	    NULL == add->pages)
 		goto err;
 
-	if (gnttab_alloc_pages(count, add->pages))
-		goto err;
+	add->dmabuf_flags = dmabuf_flags;
+
+	/*
+	 * Check if this mapping is requested to be backed
+	 * by a DMA buffer.
+	 */
+	if (dmabuf_flags & (GNTDEV_DMABUF_FLAG_DMA_WC |
+		     GNTDEV_DMABUF_FLAG_DMA_COHERENT)) {
+		bool coherent = dmabuf_flags & GNTDEV_DMABUF_FLAG_DMA_COHERENT;
+
+		if (gnttab_dma_alloc_pages(gntdev_miscdev.this_device,
+					   coherent,
+					   count, add->pages,
+					   &add->dmabuf_vaddr,
+					   &add->dmabuf_bus_addr))
+			goto err;
+	} else {
+		if (gnttab_alloc_pages(count, add->pages))
+			goto err;
+	}
 
 	for (i = 0; i < count; i++) {
 		add->map_ops[i].handle = -1;
@@ -233,6 +343,15 @@ static void gntdev_put_map(struct gntdev_priv *priv, struct grant_map *map)
 	gntdev_free_map(map);
 }
 
+static void gntdev_put_map_unlink(struct gntdev_priv *priv,
+				  struct grant_map *map)
+{
+	mutex_lock(&priv->lock);
+	list_del(&map->next);
+	gntdev_put_map(NULL /* already removed */, map);
+	mutex_unlock(&priv->lock);
+}
+
 /* ------------------------------------------------------------------ */
 
 static int find_grant_ptes(pte_t *pte, pgtable_t token,
@@ -324,6 +443,12 @@ static int map_grant_pages(struct grant_map *map)
 		map->unmap_ops[i].handle = map->map_ops[i].handle;
 		if (use_ptemod)
 			map->kunmap_ops[i].handle = map->kmap_ops[i].handle;
+		else if (map->dmabuf_vaddr) {
+			unsigned long mfn;
+
+			mfn = __pfn_to_mfn(page_to_pfn(map->pages[i]));
+			map->unmap_ops[i].dev_bus_addr = __pfn_to_phys(mfn);
+		}
 	}
 	return err;
 }
@@ -527,19 +652,48 @@ static const struct mmu_notifier_ops gntdev_mmu_ops = {
 
 /* ------------------------------------------------------------------ */
 
-static int gntdev_open(struct inode *inode, struct file *flip)
+struct gntdev_priv *gntdev_alloc_context(struct device *dev)
 {
 	struct gntdev_priv *priv;
-	int ret = 0;
 
 	priv = kzalloc(sizeof(*priv), GFP_KERNEL);
 	if (!priv)
-		return -ENOMEM;
+		return ERR_PTR(-ENOMEM);
 
 	INIT_LIST_HEAD(&priv->maps);
 	INIT_LIST_HEAD(&priv->freeable_maps);
 	mutex_init(&priv->lock);
 
+	priv->dev = dev;
+	mutex_init(&priv->dmabuf_lock);
+	INIT_LIST_HEAD(&priv->dmabuf_exp_list);
+	INIT_LIST_HEAD(&priv->dmabuf_exp_wait_list);
+	INIT_LIST_HEAD(&priv->dmabuf_imp_list);
+
+	/*
+	 * The device is not spawn from a device tree, so arch_setup_dma_ops
+	 * is not called, thus leaving the device with dummy DMA ops.
+	 * This makes the device return error on PRIME buffer import, which
+	 * is not correct: to fix this call of_dma_configure() with a NULL
+	 * node to set default DMA ops.
+	 */
+	of_dma_configure(dev, NULL);
+
+	pr_debug("priv %p\n", priv);
+
+	return priv;
+}
+EXPORT_SYMBOL(gntdev_alloc_context);
+
+static int gntdev_open(struct inode *inode, struct file *flip)
+{
+	struct gntdev_priv *priv;
+	int ret = 0;
+
+	priv = gntdev_alloc_context(gntdev_miscdev.this_device);
+	if (IS_ERR(priv))
+		return PTR_ERR(priv);
+
 	if (use_ptemod) {
 		priv->mm = get_task_mm(current);
 		if (!priv->mm) {
@@ -557,14 +711,12 @@ static int gntdev_open(struct inode *inode, struct file *flip)
 	}
 
 	flip->private_data = priv;
-	pr_debug("priv %p\n", priv);
 
 	return 0;
 }
 
-static int gntdev_release(struct inode *inode, struct file *flip)
+void gntdev_free_context(struct gntdev_priv *priv)
 {
-	struct gntdev_priv *priv = flip->private_data;
 	struct grant_map *map;
 
 	pr_debug("priv %p\n", priv);
@@ -576,11 +728,28 @@ static int gntdev_release(struct inode *inode, struct file *flip)
 		gntdev_put_map(NULL /* already removed */, map);
 	}
 	WARN_ON(!list_empty(&priv->freeable_maps));
+
 	mutex_unlock(&priv->lock);
 
+	WARN(!list_empty(&priv->dmabuf_exp_list),
+	     "Removing with non-empty exported DMA buffer list!\n");
+	WARN(!list_empty(&priv->dmabuf_exp_wait_list),
+	     "Removing with pending wait objects!\n");
+	WARN(!list_empty(&priv->dmabuf_imp_list),
+	     "Removing with non-empty imported DMA buffer list!\n");
+
+	kfree(priv);
+}
+EXPORT_SYMBOL(gntdev_free_context);
+
+static int gntdev_release(struct inode *inode, struct file *flip)
+{
+	struct gntdev_priv *priv = flip->private_data;
+
 	if (use_ptemod)
 		mmu_notifier_unregister(&priv->mn, priv->mm);
-	kfree(priv);
+
+	gntdev_free_context(priv);
 	return 0;
 }
 
@@ -598,7 +767,7 @@ static long gntdev_ioctl_map_grant_ref(struct gntdev_priv *priv,
 		return -EINVAL;
 
 	err = -ENOMEM;
-	map = gntdev_alloc_map(priv, op.count);
+	map = gntdev_alloc_map(op.count, 0 /* this is not a dma-buf */);
 	if (!map)
 		return err;
 
@@ -949,6 +1118,755 @@ static long gntdev_ioctl_grant_copy(struct gntdev_priv *priv, void __user *u)
 	return ret;
 }
 
+/* ------------------------------------------------------------------ */
+/* DMA buffer support.                                                */
+/* ------------------------------------------------------------------ */
+
+/* ------------------------------------------------------------------ */
+/* Implementation of wait for exported DMA buffer released.           */
+/* ------------------------------------------------------------------ */
+
+static void dmabuf_exp_release(struct kref *kref);
+
+static struct hyper_dmabuf_wait_obj *
+dmabuf_exp_wait_obj_new(struct gntdev_priv *priv,
+			struct hyper_dmabuf *hyper_dmabuf)
+{
+	struct hyper_dmabuf_wait_obj *obj;
+
+	obj = kzalloc(sizeof(*obj), GFP_KERNEL);
+	if (!obj)
+		return ERR_PTR(-ENOMEM);
+
+	init_completion(&obj->completion);
+	obj->hyper_dmabuf = hyper_dmabuf;
+
+	mutex_lock(&priv->dmabuf_lock);
+	list_add(&obj->next, &priv->dmabuf_exp_wait_list);
+	/* Put our reference and wait for hyper_dmabuf's release to fire. */
+	kref_put(&hyper_dmabuf->u.exp.refcount, dmabuf_exp_release);
+	mutex_unlock(&priv->dmabuf_lock);
+	return obj;
+}
+
+static void dmabuf_exp_wait_obj_free(struct gntdev_priv *priv,
+				     struct hyper_dmabuf_wait_obj *obj)
+{
+	struct hyper_dmabuf_wait_obj *cur_obj, *q;
+
+	mutex_lock(&priv->dmabuf_lock);
+	list_for_each_entry_safe(cur_obj, q, &priv->dmabuf_exp_wait_list, next)
+		if (cur_obj == obj) {
+			list_del(&obj->next);
+			kfree(obj);
+			break;
+		}
+	mutex_unlock(&priv->dmabuf_lock);
+}
+
+static int dmabuf_exp_wait_obj_wait(struct hyper_dmabuf_wait_obj *obj,
+				    u32 wait_to_ms)
+{
+	if (wait_for_completion_timeout(&obj->completion,
+			msecs_to_jiffies(wait_to_ms)) <= 0)
+		return -ETIMEDOUT;
+
+	return 0;
+}
+
+static void dmabuf_exp_wait_obj_signal(struct gntdev_priv *priv,
+				       struct hyper_dmabuf *hyper_dmabuf)
+{
+	struct hyper_dmabuf_wait_obj *obj, *q;
+
+	list_for_each_entry_safe(obj, q, &priv->dmabuf_exp_wait_list, next)
+		if (obj->hyper_dmabuf == hyper_dmabuf) {
+			pr_debug("Found hyper_dmabuf in the wait list, wake\n");
+			complete_all(&obj->completion);
+		}
+}
+
+static struct hyper_dmabuf *
+dmabuf_exp_wait_obj_get_by_fd(struct gntdev_priv *priv, int fd)
+{
+	struct hyper_dmabuf *q, *hyper_dmabuf, *ret = ERR_PTR(-ENOENT);
+
+	mutex_lock(&priv->dmabuf_lock);
+	list_for_each_entry_safe(hyper_dmabuf, q, &priv->dmabuf_exp_list, next)
+		if (hyper_dmabuf->fd == fd) {
+			pr_debug("Found hyper_dmabuf in the wait list\n");
+			kref_get(&hyper_dmabuf->u.exp.refcount);
+			ret = hyper_dmabuf;
+			break;
+		}
+	mutex_unlock(&priv->dmabuf_lock);
+	return ret;
+}
+
+static int dmabuf_exp_wait_released(struct gntdev_priv *priv, int fd,
+				    int wait_to_ms)
+{
+	struct hyper_dmabuf *hyper_dmabuf;
+	struct hyper_dmabuf_wait_obj *obj;
+	int ret;
+
+	pr_debug("Will wait for dma-buf with fd %d\n", fd);
+	/*
+	 * Try to find the DMA buffer: if not found means that
+	 * either the buffer has already been released or file descriptor
+	 * provided is wrong.
+	 */
+	hyper_dmabuf = dmabuf_exp_wait_obj_get_by_fd(priv, fd);
+	if (IS_ERR(hyper_dmabuf))
+		return PTR_ERR(hyper_dmabuf);
+
+	/*
+	 * hyper_dmabuf still exists and is reference count locked by us now,
+	 * so prepare to wait: allocate wait object and add it to the wait list,
+	 * so we can find it on release.
+	 */
+	obj = dmabuf_exp_wait_obj_new(priv, hyper_dmabuf);
+	if (IS_ERR(obj)) {
+		pr_err("Failed to setup wait object, ret %ld\n", PTR_ERR(obj));
+		return PTR_ERR(obj);
+	}
+
+	ret = dmabuf_exp_wait_obj_wait(obj, wait_to_ms);
+	dmabuf_exp_wait_obj_free(priv, obj);
+	return ret;
+}
+
+/* ------------------------------------------------------------------ */
+/* DMA buffer export support.                                         */
+/* ------------------------------------------------------------------ */
+
+static struct sg_table *
+dmabuf_pages_to_sgt(struct page **pages, unsigned int nr_pages)
+{
+	struct sg_table *sgt;
+	int ret;
+
+	sgt = kmalloc(sizeof(struct sg_table), GFP_KERNEL);
+	if (!sgt) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = sg_alloc_table_from_pages(sgt, pages, nr_pages, 0,
+					nr_pages << PAGE_SHIFT,
+					GFP_KERNEL);
+	if (ret)
+		goto out;
+
+	return sgt;
+
+out:
+	kfree(sgt);
+	return ERR_PTR(ret);
+}
+
+static int dmabuf_exp_ops_attach(struct dma_buf *dma_buf,
+				 struct device *target_dev,
+				 struct dma_buf_attachment *attach)
+{
+	struct hyper_dambuf_attachment *hyper_dambuf_attach;
+
+	hyper_dambuf_attach = kzalloc(sizeof(*hyper_dambuf_attach), GFP_KERNEL);
+	if (!hyper_dambuf_attach)
+		return -ENOMEM;
+
+	hyper_dambuf_attach->dir = DMA_NONE;
+	attach->priv = hyper_dambuf_attach;
+	/* Might need to pin the pages of the buffer now. */
+	return 0;
+}
+
+static void dmabuf_exp_ops_detach(struct dma_buf *dma_buf,
+				  struct dma_buf_attachment *attach)
+{
+	struct hyper_dambuf_attachment *hyper_dambuf_attach = attach->priv;
+
+	if (hyper_dambuf_attach) {
+		struct sg_table *sgt = hyper_dambuf_attach->sgt;
+
+		if (sgt) {
+			if (hyper_dambuf_attach->dir != DMA_NONE)
+				dma_unmap_sg_attrs(attach->dev, sgt->sgl,
+						   sgt->nents,
+						   hyper_dambuf_attach->dir,
+						   DMA_ATTR_SKIP_CPU_SYNC);
+			sg_free_table(sgt);
+		}
+
+		kfree(sgt);
+		kfree(hyper_dambuf_attach);
+		attach->priv = NULL;
+	}
+	/* Might need to unpin the pages of the buffer now. */
+}
+
+static struct sg_table *
+dmabuf_exp_ops_map_dma_buf(struct dma_buf_attachment *attach,
+			   enum dma_data_direction dir)
+{
+	struct hyper_dambuf_attachment *hyper_dambuf_attach = attach->priv;
+	struct hyper_dmabuf *hyper_dmabuf = attach->dmabuf->priv;
+	struct sg_table *sgt;
+
+	pr_debug("Mapping %d pages for dev %p\n", hyper_dmabuf->nr_pages,
+		 attach->dev);
+
+	if (WARN_ON(dir == DMA_NONE || !hyper_dambuf_attach))
+		return ERR_PTR(-EINVAL);
+
+	/* Return the cached mapping when possible. */
+	if (hyper_dambuf_attach->dir == dir)
+		return hyper_dambuf_attach->sgt;
+
+	/*
+	 * Two mappings with different directions for the same attachment are
+	 * not allowed.
+	 */
+	if (WARN_ON(hyper_dambuf_attach->dir != DMA_NONE))
+		return ERR_PTR(-EBUSY);
+
+	sgt = dmabuf_pages_to_sgt(hyper_dmabuf->pages, hyper_dmabuf->nr_pages);
+	if (!IS_ERR(sgt)) {
+		if (!dma_map_sg_attrs(attach->dev, sgt->sgl, sgt->nents, dir,
+				      DMA_ATTR_SKIP_CPU_SYNC)) {
+			sg_free_table(sgt);
+			kfree(sgt);
+			sgt = ERR_PTR(-ENOMEM);
+		} else {
+			hyper_dambuf_attach->sgt = sgt;
+			hyper_dambuf_attach->dir = dir;
+		}
+	}
+	if (IS_ERR(sgt)
+		pr_err("Failed to map sg table for dev %p\n", attach->dev);
+	return sgt;
+}
+
+static void dmabuf_exp_ops_unmap_dma_buf(struct dma_buf_attachment *attach,
+					 struct sg_table *sgt,
+					 enum dma_data_direction dir)
+{
+	/* Not implemented. The unmap is done at dmabuf_exp_ops_detach(). */
+}
+
+static void dmabuf_exp_release(struct kref *kref)
+{
+	struct hyper_dmabuf *hyper_dmabuf =
+		container_of(kref, struct hyper_dmabuf,
+			     u.exp.refcount);
+
+	dmabuf_exp_wait_obj_signal(hyper_dmabuf->priv, hyper_dmabuf);
+	list_del(&hyper_dmabuf->next);
+	kfree(hyper_dmabuf);
+}
+
+static void dmabuf_exp_ops_release(struct dma_buf *dma_buf)
+{
+	struct hyper_dmabuf *hyper_dmabuf = dma_buf->priv;
+	struct gntdev_priv *priv = hyper_dmabuf->priv;
+
+	gntdev_put_map_unlink(priv, hyper_dmabuf->u.exp.map);
+	mutex_lock(&priv->dmabuf_lock);
+	kref_put(&hyper_dmabuf->u.exp.refcount, dmabuf_exp_release);
+	mutex_unlock(&priv->dmabuf_lock);
+}
+
+static void *dmabuf_exp_ops_kmap_atomic(struct dma_buf *dma_buf,
+					unsigned long page_num)
+{
+	/* Not implemented. */
+	return NULL;
+}
+
+static void *dmabuf_exp_ops_kmap(struct dma_buf *dma_buf,
+				 unsigned long page_num)
+{
+	/* Not implemented. */
+	return NULL;
+}
+
+static int dmabuf_exp_ops_mmap(struct dma_buf *dma_buf,
+			       struct vm_area_struct *vma)
+{
+	/* Not implemented. */
+	return 0;
+}
+
+static const struct dma_buf_ops dmabuf_exp_ops =  {
+	.attach = dmabuf_exp_ops_attach,
+	.detach = dmabuf_exp_ops_detach,
+	.map_dma_buf = dmabuf_exp_ops_map_dma_buf,
+	.unmap_dma_buf = dmabuf_exp_ops_unmap_dma_buf,
+	.release = dmabuf_exp_ops_release,
+	.kmap = dmabuf_exp_ops_kmap,
+	.kmap_atomic = dmabuf_exp_ops_kmap_atomic,
+	.mmap = dmabuf_exp_ops_mmap,
+};
+
+static int dmabuf_export(struct gntdev_priv *priv, struct grant_map *map,
+			 int *fd)
+{
+	DEFINE_DMA_BUF_EXPORT_INFO(exp_info);
+	struct hyper_dmabuf *hyper_dmabuf;
+	int ret = 0;
+
+	hyper_dmabuf = kzalloc(sizeof(*hyper_dmabuf), GFP_KERNEL);
+	if (!hyper_dmabuf)
+		return -ENOMEM;
+
+	kref_init(&hyper_dmabuf->u.exp.refcount);
+
+	hyper_dmabuf->priv = priv;
+	hyper_dmabuf->nr_pages = map->count;
+	hyper_dmabuf->pages = map->pages;
+	hyper_dmabuf->u.exp.map = map;
+
+	exp_info.exp_name = KBUILD_MODNAME;
+	exp_info.ops = &dmabuf_exp_ops;
+	exp_info.size = map->count << PAGE_SHIFT;
+	exp_info.flags = O_RDWR;
+	exp_info.priv = hyper_dmabuf;
+
+	hyper_dmabuf->dmabuf = dma_buf_export(&exp_info);
+	if (IS_ERR(hyper_dmabuf->dmabuf)) {
+		ret = PTR_ERR(hyper_dmabuf->dmabuf);
+		hyper_dmabuf->dmabuf = NULL;
+		goto fail;
+	}
+
+	ret = dma_buf_fd(hyper_dmabuf->dmabuf, O_CLOEXEC);
+	if (ret < 0)
+		goto fail;
+
+	hyper_dmabuf->fd = ret;
+	*fd = ret;
+
+	pr_debug("Exporting DMA buffer with fd %d\n", ret);
+
+	mutex_lock(&priv->dmabuf_lock);
+	list_add(&hyper_dmabuf->next, &priv->dmabuf_exp_list);
+	mutex_unlock(&priv->dmabuf_lock);
+	return 0;
+
+fail:
+	if (hyper_dmabuf->dmabuf)
+		dma_buf_put(hyper_dmabuf->dmabuf);
+	kfree(hyper_dmabuf);
+	return ret;
+}
+
+static struct grant_map *
+dmabuf_exp_alloc_backing_storage(int dmabuf_flags, int count)
+{
+	struct grant_map *map;
+
+	if (unlikely(count <= 0))
+		return ERR_PTR(-EINVAL);
+
+	if ((dmabuf_flags & GNTDEV_DMABUF_FLAG_DMA_WC) &&
+	    (dmabuf_flags & GNTDEV_DMABUF_FLAG_DMA_COHERENT)) {
+		pr_err("Wrong dma-buf flags: either WC or coherent, not both\n");
+		return ERR_PTR(-EINVAL);
+	}
+
+	map = gntdev_alloc_map(count, dmabuf_flags);
+	if (!map)
+		return ERR_PTR(-ENOMEM);
+
+	if (unlikely(atomic_add_return(count, &pages_mapped) > limit)) {
+		pr_err("can't map: over limit\n");
+		gntdev_put_map(NULL, map);
+		return ERR_PTR(-ENOMEM);
+	}
+	return map;
+}
+
+int gntdev_dmabuf_exp_from_refs(struct gntdev_priv *priv, int flags,
+				int count, u32 domid, u32 *refs, u32 *fd)
+{
+	struct grant_map *map;
+	int i, ret;
+
+	*fd = -1;
+
+	if (use_ptemod) {
+		pr_err("Cannot provide dma-buf: use_ptemode %d\n",
+		       use_ptemod);
+		return -EINVAL;
+	}
+
+	map = dmabuf_exp_alloc_backing_storage(flags, count);
+	if (IS_ERR(map))
+		return PTR_ERR(map);
+
+	for (i = 0; i < count; i++) {
+		map->grants[i].domid = domid;
+		map->grants[i].ref = refs[i];
+	}
+
+	mutex_lock(&priv->lock);
+	gntdev_add_map(priv, map);
+	mutex_unlock(&priv->lock);
+
+	map->flags |= GNTMAP_host_map;
+#if defined(CONFIG_X86)
+	map->flags |= GNTMAP_device_map;
+#endif
+
+	ret = map_grant_pages(map);
+	if (ret < 0)
+		goto out;
+
+	ret = dmabuf_export(priv, map, fd);
+	if (ret < 0)
+		goto out;
+
+	return 0;
+
+out:
+	gntdev_put_map_unlink(priv, map);
+	return ret;
+}
+EXPORT_SYMBOL(gntdev_dmabuf_exp_from_refs);
+
+static long
+gntdev_ioctl_dmabuf_exp_from_refs(struct gntdev_priv *priv,
+				  struct ioctl_gntdev_dmabuf_exp_from_refs __user *u)
+{
+	struct ioctl_gntdev_dmabuf_exp_from_refs op;
+	u32 *refs;
+	long ret;
+
+	if (copy_from_user(&op, u, sizeof(op)) != 0)
+		return -EFAULT;
+
+	refs = kcalloc(op.count, sizeof(*refs), GFP_KERNEL);
+	if (!refs)
+		return -ENOMEM;
+
+	if (copy_from_user(refs, u->refs, sizeof(*refs) * op.count) != 0) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+	ret = gntdev_dmabuf_exp_from_refs(priv, op.flags, op.count,
+					  op.domid, refs, &op.fd);
+	if (ret)
+		goto out;
+
+	if (copy_to_user(u, &op, sizeof(op)) != 0)
+		ret = -EFAULT;
+
+out:
+	kfree(refs);
+	return ret;
+}
+
+static long
+gntdev_ioctl_dmabuf_exp_wait_released(struct gntdev_priv *priv,
+				      struct ioctl_gntdev_dmabuf_exp_wait_released __user *u)
+{
+	struct ioctl_gntdev_dmabuf_exp_wait_released op;
+
+	if (copy_from_user(&op, u, sizeof(op)) != 0)
+		return -EFAULT;
+
+	return dmabuf_exp_wait_released(priv, op.fd, op.wait_to_ms);
+}
+
+int gntdev_dmabuf_exp_wait_released(struct gntdev_priv *priv, u32 fd,
+				    int wait_to_ms)
+{
+	return dmabuf_exp_wait_released(priv, fd, wait_to_ms);
+}
+EXPORT_SYMBOL(gntdev_dmabuf_exp_wait_released);
+
+/* ------------------------------------------------------------------ */
+/* DMA buffer import support.                                         */
+/* ------------------------------------------------------------------ */
+
+static int
+dmabuf_imp_grant_foreign_access(struct page **pages, u32 *refs,
+				int count, int domid)
+{
+	grant_ref_t priv_gref_head;
+	int i, ret;
+
+	ret = gnttab_alloc_grant_references(count, &priv_gref_head);
+	if (ret < 0) {
+		pr_err("Cannot allocate grant references, ret %d\n", ret);
+		return ret;
+	}
+
+	for (i = 0; i < count; i++) {
+		int cur_ref;
+
+		cur_ref = gnttab_claim_grant_reference(&priv_gref_head);
+		if (cur_ref < 0) {
+			ret = cur_ref;
+			pr_err("Cannot claim grant reference, ret %d\n", ret);
+			goto out;
+		}
+
+		gnttab_grant_foreign_access_ref(cur_ref, domid,
+						xen_page_to_gfn(pages[i]), 0);
+		refs[i] = cur_ref;
+	}
+
+	ret = 0;
+
+out:
+	gnttab_free_grant_references(priv_gref_head);
+	return ret;
+}
+
+static void dmabuf_imp_end_foreign_access(u32 *refs, int count)
+{
+	int i;
+
+	for (i = 0; i < count; i++)
+		if (refs[i] != GRANT_INVALID_REF)
+			gnttab_end_foreign_access(refs[i], 0, 0UL);
+}
+
+static void dmabuf_imp_free_storage(struct hyper_dmabuf *hyper_dmabuf)
+{
+	kfree(hyper_dmabuf->pages);
+	kfree(hyper_dmabuf->u.imp.refs);
+	kfree(hyper_dmabuf);
+}
+
+static struct hyper_dmabuf *dmabuf_imp_alloc_storage(int count)
+{
+	struct hyper_dmabuf *hyper_dmabuf;
+	int i;
+
+	hyper_dmabuf = kzalloc(sizeof(*hyper_dmabuf), GFP_KERNEL);
+	if (!hyper_dmabuf)
+		goto fail;
+
+	hyper_dmabuf->u.imp.refs = kcalloc(count,
+					   sizeof(hyper_dmabuf->u.imp.refs[0]),
+					   GFP_KERNEL);
+	if (!hyper_dmabuf->u.imp.refs)
+		goto fail;
+
+	hyper_dmabuf->pages = kcalloc(count,
+				      sizeof(hyper_dmabuf->pages[0]),
+				      GFP_KERNEL);
+	if (!hyper_dmabuf->pages)
+		goto fail;
+
+	hyper_dmabuf->nr_pages = count;
+
+	for (i = 0; i < count; i++)
+	     hyper_dmabuf->u.imp.refs[i] = GRANT_INVALID_REF;
+
+	return hyper_dmabuf;
+
+fail:
+	dmabuf_imp_free_storage(hyper_dmabuf);
+	return ERR_PTR(-ENOMEM);
+}
+
+static struct hyper_dmabuf *
+dmabuf_import(struct gntdev_priv *priv, int fd, int count, int domid)
+{
+	struct hyper_dmabuf *hyper_dmabuf, *ret;
+	struct dma_buf *dma_buf;
+	struct dma_buf_attachment *attach;
+	struct sg_table *sgt;
+	struct sg_page_iter sg_iter;
+	int i;
+
+	dma_buf = dma_buf_get(fd);
+	if (IS_ERR(dma_buf))
+		return ERR_CAST(dma_buf);
+
+	hyper_dmabuf = dmabuf_imp_alloc_storage(count);
+	if (IS_ERR(hyper_dmabuf)) {
+		ret = hyper_dmabuf;
+		goto fail_put;
+	}
+
+	hyper_dmabuf->priv = priv;
+	hyper_dmabuf->fd = fd;
+
+	attach = dma_buf_attach(dma_buf, priv->dev);
+	if (IS_ERR(attach)) {
+		ret = ERR_CAST(attach);
+		goto fail_free_obj;
+	}
+
+	hyper_dmabuf->u.imp.attach = attach;
+
+	sgt = dma_buf_map_attachment(attach, DMA_BIDIRECTIONAL);
+	if (IS_ERR(sgt)) {
+		ret = ERR_CAST(sgt);
+		goto fail_detach;
+	}
+
+	/* Check number of pages that imported buffer has. */
+	if (attach->dmabuf->size != hyper_dmabuf->nr_pages << PAGE_SHIFT) {
+		ret = ERR_PTR(-EINVAL);
+		pr_err("DMA buffer has %zu pages, user-space expects %d\n",
+		       attach->dmabuf->size, hyper_dmabuf->nr_pages);
+		goto fail_unmap;
+	}
+
+	hyper_dmabuf->u.imp.sgt = sgt;
+
+	/* Now convert sgt to array of pages and check for page validity. */
+	i = 0;
+	for_each_sg_page(sgt->sgl, &sg_iter, sgt->nents, 0) {
+		struct page *page = sg_page_iter_page(&sg_iter);
+		/*
+		 * Check if page is valid: this can happen if we are given
+		 * a page from VRAM or other resources which are not backed
+		 * by a struct page.
+		 */
+		if (!pfn_valid(page_to_pfn(page))) {
+			ret = ERR_PTR(-EINVAL);
+			goto fail_unmap;
+		}
+
+		hyper_dmabuf->pages[i++] = page;
+	}
+
+	ret = ERR_PTR(dmabuf_imp_grant_foreign_access(hyper_dmabuf->pages,
+						      hyper_dmabuf->u.imp.refs,
+						      count, domid));
+	if (IS_ERR(ret))
+		goto fail_end_access;
+
+	pr_debug("Imported DMA buffer with fd %d\n", fd);
+
+	mutex_lock(&priv->dmabuf_lock);
+	list_add(&hyper_dmabuf->next, &priv->dmabuf_imp_list);
+	mutex_unlock(&priv->dmabuf_lock);
+
+	return hyper_dmabuf;
+
+fail_end_access:
+	dmabuf_imp_end_foreign_access(hyper_dmabuf->u.imp.refs, count);
+fail_unmap:
+	dma_buf_unmap_attachment(attach, sgt, DMA_BIDIRECTIONAL);
+fail_detach:
+	dma_buf_detach(dma_buf, attach);
+fail_free_obj:
+	dmabuf_imp_free_storage(hyper_dmabuf);
+fail_put:
+	dma_buf_put(dma_buf);
+	return ret;
+}
+
+/*
+ * Find the hyper dma-buf by its file descriptor and remove
+ * it from the buffer's list.
+ */
+static struct hyper_dmabuf *
+dmabuf_imp_find_unlink(struct gntdev_priv *priv, int fd)
+{
+	struct hyper_dmabuf *q, *hyper_dmabuf, *ret = ERR_PTR(-ENOENT);
+
+	mutex_lock(&priv->dmabuf_lock);
+	list_for_each_entry_safe(hyper_dmabuf, q, &priv->dmabuf_imp_list, next) {
+		if (hyper_dmabuf->fd == fd) {
+			pr_debug("Found hyper_dmabuf in the import list\n");
+			ret = hyper_dmabuf;
+			list_del(&hyper_dmabuf->next);
+			break;
+		}
+	}
+	mutex_unlock(&priv->dmabuf_lock);
+	return ret;
+}
+
+int gntdev_dmabuf_imp_release(struct gntdev_priv *priv, u32 fd)
+{
+	struct hyper_dmabuf *hyper_dmabuf;
+	struct dma_buf_attachment *attach;
+	struct dma_buf *dma_buf;
+
+	hyper_dmabuf = dmabuf_imp_find_unlink(priv, fd);
+	if (IS_ERR(hyper_dmabuf))
+		return PTR_ERR(hyper_dmabuf);
+
+	pr_debug("Releasing DMA buffer with fd %d\n", fd);
+
+	attach = hyper_dmabuf->u.imp.attach;
+
+	if (hyper_dmabuf->u.imp.sgt)
+		dma_buf_unmap_attachment(attach, hyper_dmabuf->u.imp.sgt,
+					 DMA_BIDIRECTIONAL);
+	dma_buf = attach->dmabuf;
+	dma_buf_detach(attach->dmabuf, attach);
+	dma_buf_put(dma_buf);
+
+	dmabuf_imp_end_foreign_access(hyper_dmabuf->u.imp.refs,
+				      hyper_dmabuf->nr_pages);
+	dmabuf_imp_free_storage(hyper_dmabuf);
+	return 0;
+}
+EXPORT_SYMBOL(gntdev_dmabuf_imp_release);
+
+static long
+gntdev_ioctl_dmabuf_imp_release(struct gntdev_priv *priv,
+				struct ioctl_gntdev_dmabuf_imp_release __user *u)
+{
+	struct ioctl_gntdev_dmabuf_imp_release op;
+
+	if (copy_from_user(&op, u, sizeof(op)) != 0)
+		return -EFAULT;
+
+	return gntdev_dmabuf_imp_release(priv, op.fd);
+}
+
+static long
+gntdev_ioctl_dmabuf_imp_to_refs(struct gntdev_priv *priv,
+				struct ioctl_gntdev_dmabuf_imp_to_refs __user *u)
+{
+	struct ioctl_gntdev_dmabuf_imp_to_refs op;
+	struct hyper_dmabuf *hyper_dmabuf;
+	long ret;
+
+	if (copy_from_user(&op, u, sizeof(op)) != 0)
+		return -EFAULT;
+
+	hyper_dmabuf = dmabuf_import(priv, op.fd, op.count, op.domid);
+	if (IS_ERR(hyper_dmabuf))
+		return PTR_ERR(hyper_dmabuf);
+
+	if (copy_to_user(u->refs, hyper_dmabuf->u.imp.refs,
+			 sizeof(*u->refs) * op.count) != 0) {
+		ret = -EFAULT;
+		goto out_release;
+	}
+	return 0;
+
+out_release:
+	gntdev_dmabuf_imp_release(priv, op.fd);
+	return ret;
+}
+
+u32 *gntdev_dmabuf_imp_to_refs(struct gntdev_priv *priv, int fd,
+			      int count, int domid, u32 **refs)
+{
+	struct hyper_dmabuf *hyper_dmabuf;
+
+	hyper_dmabuf = dmabuf_import(priv, fd, count, domid);
+	if (IS_ERR(hyper_dmabuf))
+		return ERR_CAST(hyper_dmabuf);
+
+	return hyper_dmabuf->u.imp.refs;
+}
+EXPORT_SYMBOL(gntdev_dmabuf_imp_to_refs);
+
 static long gntdev_ioctl(struct file *flip,
 			 unsigned int cmd, unsigned long arg)
 {
@@ -971,6 +1889,18 @@ static long gntdev_ioctl(struct file *flip,
 	case IOCTL_GNTDEV_GRANT_COPY:
 		return gntdev_ioctl_grant_copy(priv, ptr);
 
+	case IOCTL_GNTDEV_DMABUF_EXP_FROM_REFS:
+		return gntdev_ioctl_dmabuf_exp_from_refs(priv, ptr);
+
+	case IOCTL_GNTDEV_DMABUF_EXP_WAIT_RELEASED:
+		return gntdev_ioctl_dmabuf_exp_wait_released(priv, ptr);
+
+	case IOCTL_GNTDEV_DMABUF_IMP_TO_REFS:
+		return gntdev_ioctl_dmabuf_imp_to_refs(priv, ptr);
+
+	case IOCTL_GNTDEV_DMABUF_IMP_RELEASE:
+		return gntdev_ioctl_dmabuf_imp_release(priv, ptr);
+
 	default:
 		pr_debug("priv %p, unknown cmd %x\n", priv, cmd);
 		return -ENOIOCTLCMD;
diff --git a/include/uapi/xen/gntdev.h b/include/uapi/xen/gntdev.h
index d0661977667e..7cf7ca31db21 100644
--- a/include/uapi/xen/gntdev.h
+++ b/include/uapi/xen/gntdev.h
@@ -199,4 +199,105 @@ struct ioctl_gntdev_grant_copy {
 /* Send an interrupt on the indicated event channel */
 #define UNMAP_NOTIFY_SEND_EVENT 0x2
 
+/*
+ * Create a dma-buf [1] from grant references @refs of count @count provided
+ * by the foreign domain @domid with flags @flags.
+ *
+ * By default dma-buf is backed by system memory pages, but by providing
+ * GNTDEV_DMABUF_FLAG_DMA flag it can also be created as a DMA write-combine
+ * buffer, e.g. allocated with dma_alloc_wc.
+ *
+ * Returns 0 if dma-buf was successfully created and the corresponding
+ * dma-buf's file descriptor is returned in @fd.
+ *
+ * [1] https://elixir.bootlin.com/linux/latest/source/Documentation/driver-api/dma-buf.rst
+ */
+
+/*
+ * Request dma-buf backing storage to be allocated with DMA API:
+ * the buffer is backed with memory allocated with dma_alloc_wc.
+ */
+#define GNTDEV_DMABUF_FLAG_DMA_WC	(1 << 1)
+
+/*
+ * Request dma-buf backing storage to be allocated with DMA API:
+ * the buffer is backed with memory allocated with dma_alloc_coherent.
+ */
+#define GNTDEV_DMABUF_FLAG_DMA_COHERENT	(1 << 2)
+
+#define IOCTL_GNTDEV_DMABUF_EXP_FROM_REFS \
+	_IOC(_IOC_NONE, 'G', 9, \
+	     sizeof(struct ioctl_gntdev_dmabuf_exp_from_refs))
+struct ioctl_gntdev_dmabuf_exp_from_refs {
+	/* IN parameters. */
+	/* Specific options for this dma-buf: see GNTDEV_DMABUF_FLAG_XXX. */
+	__u32 flags;
+	/* Number of grant references in @refs array. */
+	__u32 count;
+	/* OUT parameters. */
+	/* File descriptor of the dma-buf. */
+	__u32 fd;
+	/* The domain ID of the grant references to be mapped. */
+	__u32 domid;
+	/* Variable IN parameter. */
+	/* Array of grant references of size @count. */
+	__u32 refs[1];
+};
+
+/*
+ * This will block until the dma-buf with the file descriptor @fd is
+ * released. This is only valid for buffers created with
+ * IOCTL_GNTDEV_DMABUF_EXP_FROM_REFS.
+ *
+ * If withing @wait_to_ms milliseconds the buffer is not released
+ * then -ETIMEDOUT error is returned.
+ * If the buffer with file descriptor @fd does not exist or has already
+ * been released, then -ENOENT is returned. For valid file descriptors
+ * this must not be treated as error.
+ */
+#define IOCTL_GNTDEV_DMABUF_EXP_WAIT_RELEASED \
+	_IOC(_IOC_NONE, 'G', 10, \
+	     sizeof(struct ioctl_gntdev_dmabuf_exp_wait_released))
+struct ioctl_gntdev_dmabuf_exp_wait_released {
+	/* IN parameters */
+	__u32 fd;
+	__u32 wait_to_ms;
+};
+
+/*
+ * Import a dma-buf with file descriptor @fd and export granted references
+ * to the pages of that dma-buf into array @refs of size @count.
+ */
+#define IOCTL_GNTDEV_DMABUF_IMP_TO_REFS \
+	_IOC(_IOC_NONE, 'G', 11, \
+	     sizeof(struct ioctl_gntdev_dmabuf_imp_to_refs))
+struct ioctl_gntdev_dmabuf_imp_to_refs {
+	/* IN parameters. */
+	/* File descriptor of the dma-buf. */
+	__u32 fd;
+	/* Number of grant references in @refs array. */
+	__u32 count;
+	/* The domain ID for which references to be granted. */
+	__u32 domid;
+	/* Reserved - must be zero. */
+	__u32 reserved;
+	/* OUT parameters. */
+	/* Array of grant references of size @count. */
+	__u32 refs[1];
+};
+
+/*
+ * This will close all references to an imported buffer, so it can be
+ * released by the owner. This is only valid for buffers created with
+ * IOCTL_GNTDEV_DMABUF_IMP_TO_REFS.
+ */
+#define IOCTL_GNTDEV_DMABUF_IMP_RELEASE \
+	_IOC(_IOC_NONE, 'G', 12, \
+	     sizeof(struct ioctl_gntdev_dmabuf_imp_release))
+struct ioctl_gntdev_dmabuf_imp_release {
+	/* IN parameters */
+	__u32 fd;
+	__u32 reserved;
+};
+
 #endif /* __LINUX_PUBLIC_GNTDEV_H__ */
diff --git a/include/xen/gntdev_exp.h b/include/xen/gntdev_exp.h
new file mode 100644
index 000000000000..aaf45bda30ac
--- /dev/null
+++ b/include/xen/gntdev_exp.h
@@ -0,0 +1,23 @@
+/******************************************************************************
+ * Xen grant device exported functionality
+ */
+
+#ifndef _XEN_GNTDEV_EXP_H
+#define _XEN_GNTDEV_EXP_H
+
+struct gntdev_priv *gntdev_alloc_context(struct device *dev);
+
+void gntdev_free_context(struct gntdev_priv *priv);
+
+int gntdev_dmabuf_exp_from_refs(struct gntdev_priv *priv, int flags,
+				int count, u32 domid, u32 *refs, u32 *fd);
+
+int gntdev_dmabuf_exp_wait_released(struct gntdev_priv *priv, u32 fd,
+				    int wait_to_ms);
+
+int gntdev_dmabuf_imp_release(struct gntdev_priv *priv, u32 fd);
+
+u32 *gntdev_dmabuf_imp_to_refs(struct gntdev_priv *priv, int fd,
+			      int count, int domid, u32 **refs);
+
+#endif
-- 
2.17.0