[Intel-gfx] [PATCH v4 3/3] drm/vgem: Keep a reference to the dmabuf when mmaped

Chris Wilson chris at chris-wilson.co.uk
Wed Oct 5 12:21:45 UTC 2016


In order to keep the dmabuf alive whilst the mmap is, we need to hold a
reference to the dmabuf and not the backing object. This is important as
the dmabuf not only keeps the object alive, but also the device so that

	dmabuf = vgem_create_dmabuf();
	ptr = mmap(... dmabuf ...);
	close(dmabuf);

persists across module-unload as well as device closure.

Testcase: igt/vgem_basic/unload
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Petri Latvala <petri.latvala at intel.com>
---
 drivers/gpu/drm/vgem/vgem_drv.c | 17 +++--------------
 1 file changed, 3 insertions(+), 14 deletions(-)

diff --git a/drivers/gpu/drm/vgem/vgem_drv.c b/drivers/gpu/drm/vgem/vgem_drv.c
index f36c14729b55..74a83e41efa9 100644
--- a/drivers/gpu/drm/vgem/vgem_drv.c
+++ b/drivers/gpu/drm/vgem/vgem_drv.c
@@ -198,7 +198,6 @@ static struct drm_ioctl_desc vgem_ioctls[] = {
 
 static int vgem_mmap(struct file *filp, struct vm_area_struct *vma)
 {
-	unsigned long flags = vma->vm_flags;
 	int ret;
 
 	ret = drm_gem_mmap(filp, vma);
@@ -208,7 +207,7 @@ static int vgem_mmap(struct file *filp, struct vm_area_struct *vma)
 	/* Keep the WC mmaping set by drm_gem_mmap() but our pages
 	 * are ordinary and not special.
 	 */
-	vma->vm_flags = flags | VM_DONTEXPAND | VM_DONTDUMP;
+	vma->vm_flags &= ~(VM_IO | VM_PFNMAP);
 	return 0;
 }
 
@@ -281,21 +280,11 @@ static int vgem_prime_mmap(struct drm_gem_object *obj,
 {
 	int ret;
 
-	if (obj->size < vma->vm_end - vma->vm_start)
-		return -EINVAL;
-
-	if (!obj->filp)
-		return -ENODEV;
-
-	ret = obj->filp->f_op->mmap(obj->filp, vma);
+	ret = drm_gem_mmap_obj(obj, obj->size, vma);
 	if (ret)
 		return ret;
 
-	fput(vma->vm_file);
-	vma->vm_file = get_file(obj->filp);
-	vma->vm_flags |= VM_DONTEXPAND | VM_DONTDUMP;
-	vma->vm_page_prot = pgprot_writecombine(vm_get_page_prot(vma->vm_flags));
-
+	vma->vm_flags &= ~(VM_IO | VM_PFNMAP);
 	return 0;
 }
 
-- 
2.9.3



More information about the Intel-gfx mailing list