search for: execbuffer

From: Gustavo Padovan <gustavo.padovan at collabora.co.uk> Instead of wrapping virtio_gpu_execbuffer() to execute the ioctl just execute it directly. Signed-off-by: Gustavo Padovan <gustavo.padovan at collabora.co.uk> --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 24 ++++++++---------------- 1 file changed, 8 insertions(+), 16 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl....

From: Gustavo Padovan <gustavo.padovan at collabora.co.uk> Instead of wrapping virtio_gpu_execbuffer() to execute the ioctl just execute it directly. Signed-off-by: Gustavo Padovan <gustavo.padovan at collabora.co.uk> --- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 24 ++++++++---------------- 1 file changed, 8 insertions(+), 16 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl....

> > Completely different approach: use get_user_pages() and don't copy the > > execbuffer at all. > It would be really nice if execbuffer does not copy. > > The user space owns the buffer and may overwrite the contents > immediately after the ioctl. Oh, right. The exec ioctl doesn't block. So this doesn't work (breaks userspace abi). Scratch the idea then. >...

..._array_lock_resv(struct virtio_gpu_object_array *objs) > > +{ > > + return drm_gem_lock_reservations(objs->objs, objs->nents, > > + &objs->ticket); Unlike in other drivers where an "object array" is only needed in execbuffer, we will use this in several places, and often with only 1 object in the array. Can we special case that and do a quick reservation_object_lock? > > +} > > + > > +void virtio_gpu_array_unlock_resv(struct virtio_gpu_object_array *objs) > > +{ > > + drm_gem_unloc...

...n of > >> dma-bufs. > >> > >> There are two new flags: > >> > >> * VIRTGPU_EXECBUF_FENCE_FD_IN to be used when passing an in-fence fd. > >> * VIRTGPU_EXECBUF_FENCE_FD_OUT to be used when requesting an out-fence fd > >> > >> The execbuffer IOCTL is now read-write to allow the userspace to read the > >> out-fence. > >> > >> On error -1 should be returned in the fence_fd field. > >> > >> Signed-off-by: Gustavo Padovan <gustavo.padovan at collabora.com> > >> Signed-off-by: Robe...

...l at redhat.com> wrote: > > > > There is no need to wait for completion here. > > > > The host will process commands in submit order, so commands can > > reference the new resource just fine even when queued up before > > completion. > > Does virtio_gpu_execbuffer_ioctl also wait for completion for a host response? No. But you pass in a list of objects (drm_virtgpu_execbuffer->bo_handles) used. They will all get reserved, so you can use DRM_IOCTL_VIRTGPU_WAIT on any of these objects to wait for completion. Recently the driver got support for returning...

...t ttm_validate_buffer *buf; - struct ttm_buffer_object *bo; - struct virtio_gpu_object *qobj; - - list_for_each_entry(buf, head, head) { - bo = buf->bo; - qobj = container_of(bo, struct virtio_gpu_object, tbo); - - drm_gem_object_put_unlocked(&qobj->gem_base); - } -} - /* * Usage of execbuffer: * Relocations need to take into account the full VIRTIO_GPUDrawable size. -- 2.18.1

...t ttm_validate_buffer *buf; - struct ttm_buffer_object *bo; - struct virtio_gpu_object *qobj; - - list_for_each_entry(buf, head, head) { - bo = buf->bo; - qobj = container_of(bo, struct virtio_gpu_object, tbo); - - drm_gem_object_put_unlocked(&qobj->gem_base); - } -} - /* * Usage of execbuffer: * Relocations need to take into account the full VIRTIO_GPUDrawable size. -- 2.18.1

...t ttm_validate_buffer *buf; - struct ttm_buffer_object *bo; - struct virtio_gpu_object *qobj; - - list_for_each_entry(buf, head, head) { - bo = buf->bo; - qobj = container_of(bo, struct virtio_gpu_object, tbo); - - drm_gem_object_put_unlocked(&qobj->gem_base); - } -} - /* * Usage of execbuffer: * Relocations need to take into account the full VIRTIO_GPUDrawable size. -- 2.18.1

...t ttm_validate_buffer *buf; - struct ttm_buffer_object *bo; - struct virtio_gpu_object *qobj; - - list_for_each_entry(buf, head, head) { - bo = buf->bo; - qobj = container_of(bo, struct virtio_gpu_object, tbo); - - drm_gem_object_put_unlocked(&qobj->gem_base); - } -} - /* * Usage of execbuffer: * Relocations need to take into account the full VIRTIO_GPUDrawable size. -- 2.18.1

...t ttm_validate_buffer *buf; - struct ttm_buffer_object *bo; - struct virtio_gpu_object *qobj; - - list_for_each_entry(buf, head, head) { - bo = buf->bo; - qobj = container_of(bo, struct virtio_gpu_object, tbo); - - drm_gem_object_put_unlocked(&qobj->gem_base); - } -} - /* * Usage of execbuffer: * Relocations need to take into account the full VIRTIO_GPUDrawable size. -- 2.18.1

...t ttm_validate_buffer *buf; - struct ttm_buffer_object *bo; - struct virtio_gpu_object *qobj; - - list_for_each_entry(buf, head, head) { - bo = buf->bo; - qobj = container_of(bo, struct virtio_gpu_object, tbo); - - drm_gem_object_put_unlocked(&qobj->gem_base); - } -} - /* * Usage of execbuffer: * Relocations need to take into account the full VIRTIO_GPUDrawable size. -- 2.18.1

...xl/qxl_ioctl.c b/drivers/gpu/drm/qxl/qxl_ioctl.c index 5cea6eea72ab..2072ddc9549c 100644 --- a/drivers/gpu/drm/qxl/qxl_ioctl.c +++ b/drivers/gpu/drm/qxl/qxl_ioctl.c @@ -160,7 +160,6 @@ static int qxl_process_single_command(struct qxl_device *qdev, default: DRM_DEBUG("Only draw commands in execbuffers\n"); return -EINVAL; - break; } if (cmd->command_size > PAGE_SIZE - sizeof(union qxl_release_info)) -- 2.18.1

On Sun, Mar 19, 2023 at 9:11?AM Dmitry Osipenko <dmitry.osipenko at collabora.com> wrote: > > Move virtio_gpu_execbuffer_ioctl() into separate virtgpu_submit.c file > and refactor the code along the way to ease addition of new features to > the ioctl. > > Signed-off-by: Dmitry Osipenko <dmitry.osipenko at collabora.com> > --- > drivers/gpu/drm/virtio/Makefile | 2 +- > drivers/gp...

...te_buffer *buf; + struct ttm_buffer_object *bo; + struct virtio_gpu_object *qobj; + list_for_each_entry(buf, head, head) { + bo = buf->bo; + qobj = container_of(bo, struct virtio_gpu_object, tbo); + + drm_gem_object_unreference_unlocked(&qobj->gem_base); + } +} + +static int virtio_gpu_execbuffer(struct drm_device *dev, + struct drm_virtgpu_execbuffer *exbuf, + struct drm_file *drm_file) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct virtio_gpu_fpriv *vfpriv = drm_file->driver_priv; + struct drm_gem_object *gobj; + struct virtio_gpu_fence *fence; + struct vi...

...te_buffer *buf; + struct ttm_buffer_object *bo; + struct virtio_gpu_object *qobj; + list_for_each_entry(buf, head, head) { + bo = buf->bo; + qobj = container_of(bo, struct virtio_gpu_object, tbo); + + drm_gem_object_unreference_unlocked(&qobj->gem_base); + } +} + +static int virtio_gpu_execbuffer(struct drm_device *dev, + struct drm_virtgpu_execbuffer *exbuf, + struct drm_file *drm_file) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct virtio_gpu_fpriv *vfpriv = drm_file->driver_priv; + struct drm_gem_object *gobj; + struct virtio_gpu_fence *fence; + struct vi...

...non-CA compositing state after the CA pass sna/gen6: Tweak programming scanline values sna/gen3+: Remove bogus assertion that the vbo in included before finish sna/gen7: Offset start/end scanlines by one sna/gen7: Place the vsync commands in the same cacheline sna: New execbuffer flags for lut-handle and fast-relocs are upstream sna: Only add bound scanouts to the scanout list sna: Assert that if marked as a scanout it is indeed bound. sna: Use the maximum backlight value if we fail to read the current value sna: Free a non-reusable bo if it expires...

...te_buffer *buf; + struct ttm_buffer_object *bo; + struct virtio_gpu_object *qobj; + list_for_each_entry(buf, head, head) { + bo = buf->bo; + qobj = container_of(bo, struct virtio_gpu_object, tbo); + + drm_gem_object_unreference_unlocked(&qobj->gem_base); + } +} + +static int virtio_gpu_execbuffer(struct drm_device *dev, + struct drm_virtgpu_execbuffer *exbuf, + struct drm_file *drm_file) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct virtio_gpu_fpriv *vfpriv = drm_file->driver_priv; + struct drm_gem_object *gobj; + struct virtio_gpu_fence *fence; + struct vi...

...te_buffer *buf; + struct ttm_buffer_object *bo; + struct virtio_gpu_object *qobj; + list_for_each_entry(buf, head, head) { + bo = buf->bo; + qobj = container_of(bo, struct virtio_gpu_object, tbo); + + drm_gem_object_unreference_unlocked(&qobj->gem_base); + } +} + +static int virtio_gpu_execbuffer(struct drm_device *dev, + struct drm_virtgpu_execbuffer *exbuf, + struct drm_file *drm_file) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct virtio_gpu_fpriv *vfpriv = drm_file->driver_priv; + struct drm_gem_object *gobj; + struct virtio_gpu_fence *fence; + struct vi...

...t; + struct virtio_gpu_object *qobj; > + list_for_each_entry(buf, head, head) { > + bo = buf->bo; > + qobj = container_of(bo, struct virtio_gpu_object, tbo); > + > + drm_gem_object_unreference_unlocked(&qobj->gem_base); > + } > +} > + > +static int virtio_gpu_execbuffer(struct drm_device *dev, > + struct drm_virtgpu_execbuffer *exbuf, > + struct drm_file *drm_file) > +{ > + struct virtio_gpu_device *vgdev = dev->dev_private; > + struct virtio_gpu_fpriv *vfpriv = drm_file->driver_priv; > + struct drm_gem_object *gobj; > + struct v...