Skip to contentSkip to conversation

drm/panfrost: drm_gem_map_offset() helper

PATCH Superseded Steven Price 3 participants 7 messages May 20, 2019 – Jun 10, 2019

Subsystems (5)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Steven Price

author maintainer
#0 lore raw

v3 cover drm/panfrost: drm_gem_map_offset() helper

Panfrost has a re-implementation of drm_gem_dumb_map_offset() with an extra bug regarding the handling of imported buffers. However we don't really want Panfrost calling _dumb functions because it's not a KMS driver.

This series renames drm_gem_dumb_map_offset() to drop the '_dumb' and updates Panfrost to use it rather than it's own implementation.

v2: https://lore.kernel.org/lkml/[email protected]/
Changes since v2:
* Drop the shmem helper

v1: https://lore.kernel.org/lkml/[email protected]/
Changes since v1:
* Rename drm_gem_dumb_map_offset to drop _dumb
* Add a shmem helper

Steven Price (2):
drm/gem: Rename drm_gem_dumb_map_offset() to drm_gem_map_offset() drm/panfrost: Use drm_gem_shmem_map_offset()

 drivers/gpu/drm/drm_dumb_buffers.c      |  4 ++--
 drivers/gpu/drm/drm_gem.c               |  6 +++---
 drivers/gpu/drm/exynos/exynos_drm_gem.c |  3 +--
 drivers/gpu/drm/panfrost/panfrost_drv.c | 16 ++--------------
 include/drm/drm_gem.h                   |  4 ++--
 5 files changed, 10 insertions(+), 23 deletions(-)
Signature · 1 line
2.20.1
author maintainer
#1 lore raw

PATCH v3 1/2 drm/gem: Rename drm_gem_dumb_map_offset() to drm_gem_map_offset()

drm_gem_dumb_map_offset() is a useful helper for non-dumb clients, so rename it to remove the _dumb.

---
 drivers/gpu/drm/drm_dumb_buffers.c      | 4 ++--
 drivers/gpu/drm/drm_gem.c               | 6 +++---
 drivers/gpu/drm/exynos/exynos_drm_gem.c | 3 +--
 include/drm/drm_gem.h                   | 4 ++--
 4 files changed, 8 insertions(+), 9 deletions(-)
drivers/gpu/drm/drm_dumb_buffers.c +2 -2
index 81dfdd33753a..956665464296 100644
@@ -46,7 +46,7 @@
  * To support dumb objects drivers must implement the &drm_driver.dumb_create
  * operation. &drm_driver.dumb_destroy defaults to drm_gem_dumb_destroy() if
  * not set and &drm_driver.dumb_map_offset defaults to
- * drm_gem_dumb_map_offset(). See the callbacks for further details.
+ * drm_gem_map_offset(). See the callbacks for further details.
  *
  * Note that dumb objects may not be used for gpu acceleration, as has been
  * attempted on some ARM embedded platforms. Such drivers really must have
@@ -125,7 +125,7 @@ int drm_mode_mmap_dumb_ioctl(struct drm_device *dev,
 						    args->handle,
 						    &args->offset);
 	else
-		return drm_gem_dumb_map_offset(file_priv, dev, args->handle,
+		return drm_gem_map_offset(file_priv, dev, args->handle,
 					       &args->offset);
 }
 
drivers/gpu/drm/drm_gem.c +3 -3
index 50de138c89e0..99bb7f79a70b 100644
@@ -294,7 +294,7 @@ drm_gem_handle_delete(struct drm_file *filp, u32 handle)
 EXPORT_SYMBOL(drm_gem_handle_delete);
 
 /**
- * drm_gem_dumb_map_offset - return the fake mmap offset for a gem object
+ * drm_gem_map_offset - return the fake mmap offset for a gem object
  * @file: drm file-private structure containing the gem object
  * @dev: corresponding drm_device
  * @handle: gem object handle
@@ -306,7 +306,7 @@ EXPORT_SYMBOL(drm_gem_handle_delete);
  * Returns:
  * 0 on success or a negative error code on failure.
  */
-int drm_gem_dumb_map_offset(struct drm_file *file, struct drm_device *dev,
+int drm_gem_map_offset(struct drm_file *file, struct drm_device *dev,
 			    u32 handle, u64 *offset)
 {
 	struct drm_gem_object *obj;
@@ -332,7 +332,7 @@ int drm_gem_dumb_map_offset(struct drm_file *file, struct drm_device *dev,
 
 	return ret;
 }
-EXPORT_SYMBOL_GPL(drm_gem_dumb_map_offset);
+EXPORT_SYMBOL_GPL(drm_gem_map_offset);
 
 /**
  * drm_gem_dumb_destroy - dumb fb callback helper for gem based drivers
drivers/gpu/drm/exynos/exynos_drm_gem.c +1 -2
index a55f5ac41bf3..5e3aa9e4a096 100644
@@ -276,8 +276,7 @@ int exynos_drm_gem_map_ioctl(struct drm_device *dev, void *data,
 {
 	struct drm_exynos_gem_map *args = data;
 
-	return drm_gem_dumb_map_offset(file_priv, dev, args->handle,
-				       &args->offset);
+	return drm_gem_map_offset(file_priv, dev, args->handle, &args->offset);
 }
 
 struct exynos_drm_gem *exynos_drm_gem_get(struct drm_file *filp,
include/drm/drm_gem.h +2 -2
index 5047c7ee25f5..91b07c2325e9 100644
@@ -395,8 +395,8 @@ int drm_gem_fence_array_add(struct xarray *fence_array,
 int drm_gem_fence_array_add_implicit(struct xarray *fence_array,
 				     struct drm_gem_object *obj,
 				     bool write);
-int drm_gem_dumb_map_offset(struct drm_file *file, struct drm_device *dev,
-			    u32 handle, u64 *offset);
+int drm_gem_map_offset(struct drm_file *file, struct drm_device *dev,
+		       u32 handle, u64 *offset);
 int drm_gem_dumb_destroy(struct drm_file *file,
 			 struct drm_device *dev,
 			 uint32_t handle);
Signature · 1 line
2.20.1
author maintainer
#2 lore raw

PATCH v3 2/2 drm/panfrost: Use drm_gem_shmem_map_offset()

In reply to Steven Price

panfrost_ioctl_mmap_bo() contains a reimplementation of drm_gem_map_offset() but with a bug - it allows mapping imported objects (without going through the exporter). Fix this by switching to use the newly renamed drm_gem_map_offset() function instead which has the bonus of simplifying the code.

---
 drivers/gpu/drm/panfrost/panfrost_drv.c | 16 ++--------------
 1 file changed, 2 insertions(+), 14 deletions(-)
drivers/gpu/drm/panfrost/panfrost_drv.c +2 -14
index d11e2281dde6..8be0cd5d6c7a 100644
@@ -255,26 +255,14 @@ static int panfrost_ioctl_mmap_bo(struct drm_device *dev, void *data,
 		      struct drm_file *file_priv)
 {
 	struct drm_panfrost_mmap_bo *args = data;
-	struct drm_gem_object *gem_obj;
-	int ret;
 
 	if (args->flags != 0) {
 		DRM_INFO("unknown mmap_bo flags: %d\n", args->flags);
 		return -EINVAL;
 	}
 
-	gem_obj = drm_gem_object_lookup(file_priv, args->handle);
-	if (!gem_obj) {
-		DRM_DEBUG("Failed to look up GEM BO %d\n", args->handle);
-		return -ENOENT;
-	}
-
-	ret = drm_gem_create_mmap_offset(gem_obj);
-	if (ret == 0)
-		args->offset = drm_vma_node_offset_addr(&gem_obj->vma_node);
-	drm_gem_object_put_unlocked(gem_obj);
-
-	return ret;
+	return drm_gem_map_offset(file_priv, dev, args->handle,
+				       &args->offset);
 }
 
 static int panfrost_ioctl_get_bo_offset(struct drm_device *dev, void *data,
Signature · 1 line
2.20.1
maintainer
#3 lore raw

Re: [PATCH v3 2/2] drm/panfrost: Use drm_gem_shmem_map_offset()

Quoted 1 line · Steven Price

You forgot to update the subject. I can fixup when applying, but I'd like an ack from Chris on patch 1.

Quoted 9 lines

panfrost_ioctl_mmap_bo() contains a reimplementation of drm_gem_map_offset() but with a bug - it allows mapping imported objects (without going through the exporter). Fix this by switching to use the newly renamed drm_gem_map_offset() function instead which has the bonus of simplifying the code.

#4 lore raw

Re: [PATCH v3 2/2] drm/panfrost: Use drm_gem_shmem_map_offset()

Quoting Rob Herring (2019-05-21 16:24:27)

Quoted 5 lines · Steven Price
Quoted 1 line

You forgot to update the subject. I can fixup when applying, but I'd like an ack from Chris on patch 1.

I still think it is incorrect as the limitation is purely an issue with the shmem backend and not a generic GEM limitation. It matters if you have a history of passing names and want a consistent api across objects when the apps themselves no longer can tell the difference, and certainly do not have access to the dmabuf fd. i915 provides an indirect mmap interface that uses the dma mapping regardless of source. -Chris

author maintainer
#5 lore raw

Re: [PATCH v3 2/2] drm/panfrost: Use drm_gem_shmem_map_offset()

Quoted 6 lines · Chris Wilson

Quoting Rob Herring (2019-05-21 16:24:27)

Quoted 5 lines · Steven Price
Quoted 1 line

You forgot to update the subject. I can fixup when applying, but I'd like an ack from Chris on patch 1.

Sorry about that - I'll try to be more careful in the future.

Quoted 2 lines

I still think it is incorrect as the limitation is purely an issue with the shmem backend and not a generic GEM limitation. It matters if you

Do you prefer the previous version of this series[1] with the shmem helper?

[1]
https://lore.kernel.org/lkml/[email protected]/

Although this isn't a generic GEM limitation it's currently the same limitation that applies to the 'dumb' drivers as well as shmem backend, so I'd prefer not implementing two identical functions purely because this limitation could be removed in the future.

Quoted 4 lines

have a history of passing names and want a consistent api across objects when the apps themselves no longer can tell the difference, and certainly do not have access to the dmabuf fd. i915 provides an indirect mmap interface that uses the dma mapping regardless of source.

I don't understand the i915 driver, but from a quick look at the source of i915_gem_mmap_ioctl():

	/* prime objects have no backing filp to GEM mmap
	 * pages from.
	 */
	if (!obj->base.filp) {
		addr = -ENXIO;
		goto err;
	}

it looks to me that an attempt to map an imported dmabuf from user space using the ioctl will fail. Am I missing something?

exynos_drm_gem_mmap() is the only (mainline[2]) instance I can see where a transparent mapping of a dma_buf is supported.

[2] mali_kbase does this too - but I'm not convinced it was a good idea.

I could instead add support in shmem/panfrost for transparently passing the request to the exporter (i.e. call dma_buf_mmap()) - but I'm not sure we want to implement this if there isn't going to be a user of the support.

Steve

maintainer
#6 lore raw

Re: [PATCH v3 2/2] drm/panfrost: Use drm_gem_shmem_map_offset()

Quoted 16 lines · Steven Price
Quoted 6 lines · Chris Wilson

Quoting Rob Herring (2019-05-21 16:24:27)

Quoted 5 lines · Steven Price
Quoted 1 line

You forgot to update the subject. I can fixup when applying, but I'd like an ack from Chris on patch 1.

Sorry about that - I'll try to be more careful in the future.

Quoted 2 lines

I still think it is incorrect as the limitation is purely an issue with the shmem backend and not a generic GEM limitation. It matters if you

Do you prefer the previous version of this series[1] with the shmem helper?

Quoted 7 lines

[1]
https://lore.kernel.org/lkml/[email protected]/

Although this isn't a generic GEM limitation it's currently the same limitation that applies to the 'dumb' drivers as well as shmem backend, so I'd prefer not implementing two identical functions purely because this limitation could be removed in the future.

In interest of moving this forward, how about some comments in drm_gem_map_offset() explaining the limitations and when it is appropriate to use the function.

Rob