drm/exynos: add userptr feature for g2d module

This patch adds userptr feautre for G2D module.

The userptr means user space address allocated by malloc().
And the purpose of this feature is to make G2D's dma able
to access the user space region.

To user this feature, user should flag G2D_BUF_USRPTR to
offset variable of struct drm_exynos_g2d_cmd and fill
struct drm_exynos_g2d_userptr with user space address
and size for it and then should set a pointer to
drm_exynos_g2d_userptr object to data variable of struct
drm_exynos_g2d_cmd. The last bit of offset variable is used
to check if the cmdlist's buffer type is userptr or not.
If userptr, the g2d driver gets user space address and size
and then gets pages through get_user_pages().
(another case is counted as gem handle)

Below is sample codes:

static void set_cmd(struct drm_exynos_g2d_cmd *cmd,
		unsigned long offset, unsigned long data)
{
	cmd->offset = offset;
	cmd->data = data;
}

static int solid_fill_test(int x, int y, unsigned long userptr)
{
	struct drm_exynos_g2d_cmd cmd_gem[5];
	struct drm_exynos_g2d_userptr g2d_userptr;
	unsigned int gem_nr = 0;
	...

	g2d_userptr.userptr = userptr;
	g2d_userptr.size = x * y * 4;

	set_cmd(&cmd_gem[gem_nr++], DST_BASE_ADDR_REG |
					G2D_BUF_USERPTR,
			(unsigned long)&g2d_userptr);
	...
}

int main(int argc, char **argv)
{
	unsigned long addr;
	...

	addr = malloc(x * y * 4);
	...

	solid_fill_test(x, y, addr);
	...
}

And next, the pages are mapped with iommu table and the device
address is set to cmdlist so that G2D's dma can access it.
As you may know, the pages from get_user_pages() are pinned.
In other words, they CAN NOT be migrated and also swapped out.
So the dma access would be safe.

But the use of userptr feature has performance overhead so
this patch also has memory pool to the userptr feature.
Please, assume that user sends cmdlist filled with userptr
and size every time to g2d driver, and the get_user_pages
funcion will be called every time.

The memory pool has maximum 64MB size and the userptr that
user had ever sent, is holded in the memory pool.
This meaning is that if the userptr from user is same as one
in the memory pool, device address to the userptr in the memory
pool is set to cmdlist.

And last, the pages from get_user_pages() will be freed once
user calls free() and the dma access is completed. Actually,
get_user_pages() takes 2 reference counts if the user process
has never accessed user region allocated by malloc(). Then, if
the user calls free(), the page reference count becomes 1 and
becomes 0 with put_page() call. And the reverse holds as well.
This means how the pages backed are used by dma and freed.

This patch is based on "drm/exynos: add iommu support for g2d",
	https://patchwork.kernel.org/patch/1629481/Signed-off-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>

drivers/gpu/drm/exynos/exynos_drm_drv.h

@@ -231,8 +231,7 @@ struct exynos_drm_g2d_private {
 	struct device		*dev;
 	struct list_head	inuse_cmdlist;
 	struct list_head	event_list;
-	struct list_head	gem_list;
-	unsigned int		gem_nr;
+	struct list_head	userptr_list;
 };
 
 struct drm_exynos_file_private {

drivers/gpu/drm/exynos/exynos_drm_gem.c

@@ -448,6 +448,129 @@ int exynos_drm_gem_get_ioctl(struct drm_device *dev, void *data,
 	return 0;
 }
 
+struct vm_area_struct *exynos_gem_get_vma(struct vm_area_struct *vma)
+{
+	struct vm_area_struct *vma_copy;
+
+	vma_copy = kmalloc(sizeof(*vma_copy), GFP_KERNEL);
+	if (!vma_copy)
+		return NULL;
+
+	if (vma->vm_ops && vma->vm_ops->open)
+		vma->vm_ops->open(vma);
+
+	if (vma->vm_file)
+		get_file(vma->vm_file);
+
+	memcpy(vma_copy, vma, sizeof(*vma));
+
+	vma_copy->vm_mm = NULL;
+	vma_copy->vm_next = NULL;
+	vma_copy->vm_prev = NULL;
+
+	return vma_copy;
+}
+
+void exynos_gem_put_vma(struct vm_area_struct *vma)
+{
+	if (!vma)
+		return;
+
+	if (vma->vm_ops && vma->vm_ops->close)
+		vma->vm_ops->close(vma);
+
+	if (vma->vm_file)
+		fput(vma->vm_file);
+
+	kfree(vma);
+}
+
+int exynos_gem_get_pages_from_userptr(unsigned long start,
+						unsigned int npages,
+						struct page **pages,
+						struct vm_area_struct *vma)
+{
+	int get_npages;
+
+	/* the memory region mmaped with VM_PFNMAP. */
+	if (vma_is_io(vma)) {
+		unsigned int i;
+
+		for (i = 0; i < npages; ++i, start += PAGE_SIZE) {
+			unsigned long pfn;
+			int ret = follow_pfn(vma, start, &pfn);
+			if (ret)
+				return ret;
+
+			pages[i] = pfn_to_page(pfn);
+		}
+
+		if (i != npages) {
+			DRM_ERROR("failed to get user_pages.\n");
+			return -EINVAL;
+		}
+
+		return 0;
+	}
+
+	get_npages = get_user_pages(current, current->mm, start,
+					npages, 1, 1, pages, NULL);
+	get_npages = max(get_npages, 0);
+	if (get_npages != npages) {
+		DRM_ERROR("failed to get user_pages.\n");
+		while (get_npages)
+			put_page(pages[--get_npages]);
+		return -EFAULT;
+	}
+
+	return 0;
+}
+
+void exynos_gem_put_pages_to_userptr(struct page **pages,
+					unsigned int npages,
+					struct vm_area_struct *vma)
+{
+	if (!vma_is_io(vma)) {
+		unsigned int i;
+
+		for (i = 0; i < npages; i++) {
+			set_page_dirty_lock(pages[i]);
+
+			/*
+			 * undo the reference we took when populating
+			 * the table.
+			 */
+			put_page(pages[i]);
+		}
+	}
+}
+
+int exynos_gem_map_sgt_with_dma(struct drm_device *drm_dev,
+				struct sg_table *sgt,
+				enum dma_data_direction dir)
+{
+	int nents;
+
+	mutex_lock(&drm_dev->struct_mutex);
+
+	nents = dma_map_sg(drm_dev->dev, sgt->sgl, sgt->nents, dir);
+	if (!nents) {
+		DRM_ERROR("failed to map sgl with dma.\n");
+		mutex_unlock(&drm_dev->struct_mutex);
+		return nents;
+	}
+
+	mutex_unlock(&drm_dev->struct_mutex);
+	return 0;
+}
+
+void exynos_gem_unmap_sgt_from_dma(struct drm_device *drm_dev,
+				struct sg_table *sgt,
+				enum dma_data_direction dir)
+{
+	dma_unmap_sg(drm_dev->dev, sgt->sgl, sgt->nents, dir);
+}
+
 int exynos_drm_gem_init_object(struct drm_gem_object *obj)
 {
 	DRM_DEBUG_KMS("%s\n", __FILE__);

drivers/gpu/drm/exynos/exynos_drm_gem.h

@@ -35,22 +35,29 @@
  * exynos drm gem buffer structure.
  *
  * @kvaddr: kernel virtual address to allocated memory region.
+ * *userptr: user space address.
  * @dma_addr: bus address(accessed by dma) to allocated memory region.
  *	- this address could be physical address without IOMMU and
  *	device address with IOMMU.
+ * @write: whether pages will be written to by the caller.
  * @sgt: sg table to transfer page data.
  * @pages: contain all pages to allocated memory region.
  * @page_size: could be 4K, 64K or 1MB.
  * @size: size of allocated memory region.
+ * @pfnmap: indicate whether memory region from userptr is mmaped with
+ *	VM_PFNMAP or not.
  */
 struct exynos_drm_gem_buf {
 	void __iomem		*kvaddr;
+	unsigned long		userptr;
 	dma_addr_t		dma_addr;
 	struct dma_attrs	dma_attrs;
+	unsigned int		write;
 	struct sg_table		*sgt;
 	struct page		**pages;
 	unsigned long		page_size;
 	unsigned long		size;
+	bool			pfnmap;
 };
 
 /*
@@ -66,6 +73,7 @@ struct exynos_drm_gem_buf {
  *	or at framebuffer creation.
  * @size: size requested from user, in bytes and this size is aligned
  *	in page unit.
+ * @vma: a pointer to vm_area.
  * @flags: indicate memory type to allocated buffer and cache attruibute.
  *
  * P.S. this object would be transfered to user as kms_bo.handle so
@@ -75,6 +83,7 @@ struct exynos_drm_gem_obj {
 	struct drm_gem_object		base;
 	struct exynos_drm_gem_buf	*buffer;
 	unsigned long			size;
+	struct vm_area_struct		*vma;
 	unsigned int			flags;
 };
 
@@ -129,6 +138,10 @@ int exynos_drm_gem_map_offset_ioctl(struct drm_device *dev, void *data,
 int exynos_drm_gem_mmap_ioctl(struct drm_device *dev, void *data,
 			      struct drm_file *file_priv);
 
+/* map user space allocated by malloc to pages. */
+int exynos_drm_gem_userptr_ioctl(struct drm_device *dev, void *data,
+				      struct drm_file *file_priv);
+
 /* get buffer information to memory region allocated by gem. */
 int exynos_drm_gem_get_ioctl(struct drm_device *dev, void *data,
 				      struct drm_file *file_priv);
@@ -164,4 +177,36 @@ int exynos_drm_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
 /* set vm_flags and we can change the vm attribute to other one at here. */
 int exynos_drm_gem_mmap(struct file *filp, struct vm_area_struct *vma);
 
+static inline int vma_is_io(struct vm_area_struct *vma)
+{
+	return !!(vma->vm_flags & (VM_IO | VM_PFNMAP));
+}
+
+/* get a copy of a virtual memory region. */
+struct vm_area_struct *exynos_gem_get_vma(struct vm_area_struct *vma);
+
+/* release a userspace virtual memory area. */
+void exynos_gem_put_vma(struct vm_area_struct *vma);
+
+/* get pages from user space. */
+int exynos_gem_get_pages_from_userptr(unsigned long start,
+						unsigned int npages,
+						struct page **pages,
+						struct vm_area_struct *vma);
+
+/* drop the reference to pages. */
+void exynos_gem_put_pages_to_userptr(struct page **pages,
+					unsigned int npages,
+					struct vm_area_struct *vma);
+
+/* map sgt with dma region. */
+int exynos_gem_map_sgt_with_dma(struct drm_device *drm_dev,
+				struct sg_table *sgt,
+				enum dma_data_direction dir);
+
+/* unmap sgt from dma region. */
+void exynos_gem_unmap_sgt_from_dma(struct drm_device *drm_dev,
+				struct sg_table *sgt,
+				enum dma_data_direction dir);
+
 #endif

include/uapi/drm/exynos_drm.h

@@ -133,17 +133,26 @@ struct drm_exynos_g2d_cmd {
 	__u32	data;
 };
 
+enum drm_exynos_g2d_buf_type {
+	G2D_BUF_USERPTR = 1 << 31,
+};
+
 enum drm_exynos_g2d_event_type {
 	G2D_EVENT_NOT,
 	G2D_EVENT_NONSTOP,
 	G2D_EVENT_STOP,		/* not yet */
 };
 
+struct drm_exynos_g2d_userptr {
+	unsigned long userptr;
+	unsigned long size;
+};
+
 struct drm_exynos_g2d_set_cmdlist {
 	__u64					cmd;
-	__u64					cmd_gem;
+	__u64					cmd_buf;
 	__u32					cmd_nr;
-	__u32					cmd_gem_nr;
+	__u32					cmd_buf_nr;
 
 	/* for g2d event */
 	__u64					event_type;