This series replaces the global disable_notify state by command-level bools to control vq kicks. When command batching is applied to more places, this prevents one process from affecting another process.
Call virtqueue_kick_prepare once in virtio_gpu_enable_notify, not whenever a command is added. This should be more efficient since the intention is to batch commands.
Signed-off-by: Chia-I Wu olvaffe@gmail.com --- drivers/gpu/drm/virtio/virtgpu_drv.h | 1 - drivers/gpu/drm/virtio/virtgpu_vq.c | 28 +++++++++++++++++----------- 2 files changed, 17 insertions(+), 12 deletions(-)
diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.h b/drivers/gpu/drm/virtio/virtgpu_drv.h index 7e69c06e168ea..0af1724bda390 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.h +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h @@ -178,7 +178,6 @@ struct virtio_gpu_device { bool vqs_ready;
bool disable_notify; - bool pending_notify;
struct ida resource_ida;
diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c index 41e475fbd67bf..204b1577a1873 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -364,16 +364,13 @@ static void virtio_gpu_queue_ctrl_sgs(struct virtio_gpu_device *vgdev,
trace_virtio_gpu_cmd_queue(vq, virtio_gpu_vbuf_ctrl_hdr(vbuf));
- notify = virtqueue_kick_prepare(vq); + if (!vgdev->disable_notify) + notify = virtqueue_kick_prepare(vq);
spin_unlock(&vgdev->ctrlq.qlock);
- if (notify) { - if (vgdev->disable_notify) - vgdev->pending_notify = true; - else - virtqueue_notify(vq); - } + if (notify) + virtqueue_notify(vq); }
static void virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev, @@ -436,12 +433,21 @@ void virtio_gpu_disable_notify(struct virtio_gpu_device *vgdev)
void virtio_gpu_enable_notify(struct virtio_gpu_device *vgdev) { + struct virtqueue *vq = vgdev->ctrlq.vq; + bool notify; + vgdev->disable_notify = false;
- if (!vgdev->pending_notify) - return; - vgdev->pending_notify = false; - virtqueue_notify(vgdev->ctrlq.vq); + spin_lock(&vgdev->ctrlq.qlock); + notify = virtqueue_kick_prepare(vq); + spin_unlock(&vgdev->ctrlq.qlock); + + /* Do not call virtqueue_notify with the lock held because + * virtio_gpu_dequeue_ctrl_func may contend for the lock if an irq is + * generated while we are in virtqueue_notify. + */ + if (notify) + virtqueue_notify(vq); }
static void virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev,
The global disable_notify state does not scale well when we start using it in more places and when there are multiple threads. Use command-level bools to control whether to notify or not.
The naming conventions are
virtio_gpu_cmd_foo -> add foo and commit is implied virtio_gpu_ctrl_bar -> add bar to ctrlq and commit is caller-controlled
virtio_gpu_{disable,enable}_notify is also replaced by virtio_gpu_commit_ctrl.
Signed-off-by: Chia-I Wu olvaffe@gmail.com --- drivers/gpu/drm/virtio/virtgpu_display.c | 9 ++- drivers/gpu/drm/virtio/virtgpu_drv.h | 38 +++++----- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 4 +- drivers/gpu/drm/virtio/virtgpu_plane.c | 42 +++++------ drivers/gpu/drm/virtio/virtgpu_vq.c | 94 ++++++++++++------------ 5 files changed, 92 insertions(+), 95 deletions(-)
diff --git a/drivers/gpu/drm/virtio/virtgpu_display.c b/drivers/gpu/drm/virtio/virtgpu_display.c index 7b0f0643bb2dd..34b1136b05120 100644 --- a/drivers/gpu/drm/virtio/virtgpu_display.c +++ b/drivers/gpu/drm/virtio/virtgpu_display.c @@ -87,9 +87,9 @@ static void virtio_gpu_crtc_mode_set_nofb(struct drm_crtc *crtc) struct virtio_gpu_device *vgdev = dev->dev_private; struct virtio_gpu_output *output = drm_crtc_to_virtio_gpu_output(crtc);
- virtio_gpu_cmd_set_scanout(vgdev, output->index, 0, - crtc->mode.hdisplay, - crtc->mode.vdisplay, 0, 0); + virtio_gpu_ctrl_set_scanout(vgdev, output->index, 0, + crtc->mode.hdisplay, + crtc->mode.vdisplay, 0, 0, true); }
static void virtio_gpu_crtc_atomic_enable(struct drm_crtc *crtc, @@ -107,7 +107,8 @@ static void virtio_gpu_crtc_atomic_disable(struct drm_crtc *crtc, struct virtio_gpu_device *vgdev = dev->dev_private; struct virtio_gpu_output *output = drm_crtc_to_virtio_gpu_output(crtc);
- virtio_gpu_cmd_set_scanout(vgdev, output->index, 0, 0, 0, 0, 0); + virtio_gpu_ctrl_set_scanout(vgdev, output->index, 0, 0, 0, 0, 0, + true); output->enabled = false; }
diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.h b/drivers/gpu/drm/virtio/virtgpu_drv.h index 0af1724bda390..b9a948a6d6cf7 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.h +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h @@ -177,8 +177,6 @@ struct virtio_gpu_device { struct kmem_cache *vbufs; bool vqs_ready;
- bool disable_notify; - struct ida resource_ida;
wait_queue_head_t resp_wq; @@ -262,20 +260,22 @@ void virtio_gpu_cmd_create_resource(struct virtio_gpu_device *vgdev, struct virtio_gpu_fence *fence); void virtio_gpu_cmd_unref_resource(struct virtio_gpu_device *vgdev, uint32_t resource_id); -void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, - uint64_t offset, - uint32_t width, uint32_t height, - uint32_t x, uint32_t y, - struct virtio_gpu_object_array *objs, - struct virtio_gpu_fence *fence); -void virtio_gpu_cmd_resource_flush(struct virtio_gpu_device *vgdev, - uint32_t resource_id, - uint32_t x, uint32_t y, - uint32_t width, uint32_t height); -void virtio_gpu_cmd_set_scanout(struct virtio_gpu_device *vgdev, - uint32_t scanout_id, uint32_t resource_id, - uint32_t width, uint32_t height, - uint32_t x, uint32_t y); +void virtio_gpu_ctrl_transfer_to_host_2d(struct virtio_gpu_device *vgdev, + uint64_t offset, + uint32_t width, uint32_t height, + uint32_t x, uint32_t y, + struct virtio_gpu_object_array *objs, + struct virtio_gpu_fence *fence, + bool commit); +void virtio_gpu_ctrl_resource_flush(struct virtio_gpu_device *vgdev, + uint32_t resource_id, + uint32_t x, uint32_t y, + uint32_t width, uint32_t height, + bool commit); +void virtio_gpu_ctrl_set_scanout(struct virtio_gpu_device *vgdev, + uint32_t scanout_id, uint32_t resource_id, + uint32_t width, uint32_t height, + uint32_t x, uint32_t y, bool commit); int virtio_gpu_object_attach(struct virtio_gpu_device *vgdev, struct virtio_gpu_object *obj, struct virtio_gpu_fence *fence); @@ -324,6 +324,9 @@ virtio_gpu_cmd_resource_create_3d(struct virtio_gpu_device *vgdev, struct virtio_gpu_object_params *params, struct virtio_gpu_object_array *objs, struct virtio_gpu_fence *fence); + +void virtio_gpu_commit_ctrl(struct virtio_gpu_device *vgdev); + void virtio_gpu_ctrl_ack(struct virtqueue *vq); void virtio_gpu_cursor_ack(struct virtqueue *vq); void virtio_gpu_fence_ack(struct virtqueue *vq); @@ -331,9 +334,6 @@ void virtio_gpu_dequeue_ctrl_func(struct work_struct *work); void virtio_gpu_dequeue_cursor_func(struct work_struct *work); void virtio_gpu_dequeue_fence_func(struct work_struct *work);
-void virtio_gpu_disable_notify(struct virtio_gpu_device *vgdev); -void virtio_gpu_enable_notify(struct virtio_gpu_device *vgdev); - /* virtio_gpu_display.c */ void virtio_gpu_modeset_init(struct virtio_gpu_device *vgdev); void virtio_gpu_modeset_fini(struct virtio_gpu_device *vgdev); diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index 205ec4abae2b9..2bb50dec5c43d 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ioctl.c +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c @@ -339,10 +339,10 @@ static int virtio_gpu_transfer_to_host_ioctl(struct drm_device *dev, void *data, return -ENOENT;
if (!vgdev->has_virgl_3d) { - virtio_gpu_cmd_transfer_to_host_2d + virtio_gpu_ctrl_transfer_to_host_2d (vgdev, offset, args->box.w, args->box.h, args->box.x, args->box.y, - objs, NULL); + objs, NULL, true); } else { ret = virtio_gpu_array_lock_resv(objs); if (ret != 0) diff --git a/drivers/gpu/drm/virtio/virtgpu_plane.c b/drivers/gpu/drm/virtio/virtgpu_plane.c index ac42c84d2d7f2..7300823e80520 100644 --- a/drivers/gpu/drm/virtio/virtgpu_plane.c +++ b/drivers/gpu/drm/virtio/virtgpu_plane.c @@ -122,8 +122,8 @@ static void virtio_gpu_update_dumb_bo(struct virtio_gpu_device *vgdev, return; virtio_gpu_array_add_obj(objs, &bo->base.base);
- virtio_gpu_cmd_transfer_to_host_2d(vgdev, off, w, h, x, y, - objs, NULL); + virtio_gpu_ctrl_transfer_to_host_2d(vgdev, off, w, h, x, y, + objs, NULL, false); }
static void virtio_gpu_primary_plane_update(struct drm_plane *plane, @@ -144,18 +144,16 @@ static void virtio_gpu_primary_plane_update(struct drm_plane *plane,
if (!plane->state->fb || !output->enabled) { DRM_DEBUG("nofb\n"); - virtio_gpu_cmd_set_scanout(vgdev, output->index, 0, - plane->state->src_w >> 16, - plane->state->src_h >> 16, - 0, 0); + virtio_gpu_ctrl_set_scanout(vgdev, output->index, 0, + plane->state->src_w >> 16, + plane->state->src_h >> 16, + 0, 0, true); return; }
if (!drm_atomic_helper_damage_merged(old_state, plane->state, &rect)) return;
- virtio_gpu_disable_notify(vgdev); - bo = gem_to_virtio_gpu_obj(plane->state->fb->obj[0]); if (bo->dumb) virtio_gpu_update_dumb_bo(vgdev, plane->state, &rect); @@ -173,21 +171,21 @@ static void virtio_gpu_primary_plane_update(struct drm_plane *plane, plane->state->src_h >> 16, plane->state->src_x >> 16, plane->state->src_y >> 16); - virtio_gpu_cmd_set_scanout(vgdev, output->index, - bo->hw_res_handle, - plane->state->src_w >> 16, - plane->state->src_h >> 16, - plane->state->src_x >> 16, - plane->state->src_y >> 16); + virtio_gpu_ctrl_set_scanout(vgdev, output->index, + bo->hw_res_handle, + plane->state->src_w >> 16, + plane->state->src_h >> 16, + plane->state->src_x >> 16, + plane->state->src_y >> 16, false); }
- virtio_gpu_cmd_resource_flush(vgdev, bo->hw_res_handle, - rect.x1, - rect.y1, - rect.x2 - rect.x1, - rect.y2 - rect.y1); + virtio_gpu_ctrl_resource_flush(vgdev, bo->hw_res_handle, + rect.x1, + rect.y1, + rect.x2 - rect.x1, + rect.y2 - rect.y1, false);
- virtio_gpu_enable_notify(vgdev); + virtio_gpu_commit_ctrl(vgdev); }
static int virtio_gpu_cursor_prepare_fb(struct drm_plane *plane, @@ -260,11 +258,11 @@ static void virtio_gpu_cursor_plane_update(struct drm_plane *plane, if (!objs) return; virtio_gpu_array_add_obj(objs, vgfb->base.obj[0]); - virtio_gpu_cmd_transfer_to_host_2d + virtio_gpu_ctrl_transfer_to_host_2d (vgdev, 0, plane->state->crtc_w, plane->state->crtc_h, - 0, 0, objs, vgfb->fence); + 0, 0, objs, vgfb->fence, true); dma_fence_wait(&vgfb->fence->f, true); dma_fence_put(&vgfb->fence->f); vgfb->fence = NULL; diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c index 204b1577a1873..f66e1372b9a7f 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -324,7 +324,8 @@ static void virtio_gpu_queue_ctrl_sgs(struct virtio_gpu_device *vgdev, int elemcnt, struct scatterlist **sgs, int outcnt, - int incnt) + int incnt, + bool commit) { struct virtqueue *vq = vgdev->ctrlq.vq; bool notify = false; @@ -364,7 +365,7 @@ static void virtio_gpu_queue_ctrl_sgs(struct virtio_gpu_device *vgdev,
trace_virtio_gpu_cmd_queue(vq, virtio_gpu_vbuf_ctrl_hdr(vbuf));
- if (!vgdev->disable_notify) + if (commit) notify = virtqueue_kick_prepare(vq);
spin_unlock(&vgdev->ctrlq.qlock); @@ -375,7 +376,8 @@ static void virtio_gpu_queue_ctrl_sgs(struct virtio_gpu_device *vgdev,
static void virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev, struct virtio_gpu_vbuffer *vbuf, - struct virtio_gpu_fence *fence) + struct virtio_gpu_fence *fence, + bool commit) { struct scatterlist *sgs[3], vcmd, vout, vresp; struct sg_table *sgt = NULL; @@ -418,7 +420,7 @@ static void virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev, }
virtio_gpu_queue_ctrl_sgs(vgdev, vbuf, fence, elemcnt, sgs, outcnt, - incnt); + incnt, commit);
if (sgt) { sg_free_table(sgt); @@ -426,18 +428,18 @@ static void virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev, } }
-void virtio_gpu_disable_notify(struct virtio_gpu_device *vgdev) +static void virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev, + struct virtio_gpu_vbuffer *vbuf, + bool commit) { - vgdev->disable_notify = true; + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, NULL, commit); }
-void virtio_gpu_enable_notify(struct virtio_gpu_device *vgdev) +void virtio_gpu_commit_ctrl(struct virtio_gpu_device *vgdev) { struct virtqueue *vq = vgdev->ctrlq.vq; bool notify;
- vgdev->disable_notify = false; - spin_lock(&vgdev->ctrlq.qlock); notify = virtqueue_kick_prepare(vq); spin_unlock(&vgdev->ctrlq.qlock); @@ -450,12 +452,6 @@ void virtio_gpu_enable_notify(struct virtio_gpu_device *vgdev) virtqueue_notify(vq); }
-static void virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev, - struct virtio_gpu_vbuffer *vbuf) -{ - virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, NULL); -} - static void virtio_gpu_queue_cursor(struct virtio_gpu_device *vgdev, struct virtio_gpu_vbuffer *vbuf) { @@ -517,7 +513,7 @@ void virtio_gpu_cmd_create_resource(struct virtio_gpu_device *vgdev, cmd_p->width = cpu_to_le32(params->width); cmd_p->height = cpu_to_le32(params->height);
- virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence, true); bo->created = true; }
@@ -533,7 +529,7 @@ void virtio_gpu_cmd_unref_resource(struct virtio_gpu_device *vgdev, cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_RESOURCE_UNREF); cmd_p->resource_id = cpu_to_le32(resource_id);
- virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, true); }
static void virtio_gpu_cmd_resource_inval_backing(struct virtio_gpu_device *vgdev, @@ -549,13 +545,13 @@ static void virtio_gpu_cmd_resource_inval_backing(struct virtio_gpu_device *vgde cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_RESOURCE_DETACH_BACKING); cmd_p->resource_id = cpu_to_le32(resource_id);
- virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence, true); }
-void virtio_gpu_cmd_set_scanout(struct virtio_gpu_device *vgdev, - uint32_t scanout_id, uint32_t resource_id, - uint32_t width, uint32_t height, - uint32_t x, uint32_t y) +void virtio_gpu_ctrl_set_scanout(struct virtio_gpu_device *vgdev, + uint32_t scanout_id, uint32_t resource_id, + uint32_t width, uint32_t height, + uint32_t x, uint32_t y, bool commit) { struct virtio_gpu_set_scanout *cmd_p; struct virtio_gpu_vbuffer *vbuf; @@ -571,13 +567,14 @@ void virtio_gpu_cmd_set_scanout(struct virtio_gpu_device *vgdev, cmd_p->r.x = cpu_to_le32(x); cmd_p->r.y = cpu_to_le32(y);
- virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, commit); }
-void virtio_gpu_cmd_resource_flush(struct virtio_gpu_device *vgdev, - uint32_t resource_id, - uint32_t x, uint32_t y, - uint32_t width, uint32_t height) +void virtio_gpu_ctrl_resource_flush(struct virtio_gpu_device *vgdev, + uint32_t resource_id, + uint32_t x, uint32_t y, + uint32_t width, uint32_t height, + bool commit) { struct virtio_gpu_resource_flush *cmd_p; struct virtio_gpu_vbuffer *vbuf; @@ -592,15 +589,16 @@ void virtio_gpu_cmd_resource_flush(struct virtio_gpu_device *vgdev, cmd_p->r.x = cpu_to_le32(x); cmd_p->r.y = cpu_to_le32(y);
- virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, commit); }
-void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, - uint64_t offset, - uint32_t width, uint32_t height, - uint32_t x, uint32_t y, - struct virtio_gpu_object_array *objs, - struct virtio_gpu_fence *fence) +void virtio_gpu_ctrl_transfer_to_host_2d(struct virtio_gpu_device *vgdev, + uint64_t offset, + uint32_t width, uint32_t height, + uint32_t x, uint32_t y, + struct virtio_gpu_object_array *objs, + struct virtio_gpu_fence *fence, + bool commit) { struct virtio_gpu_object *bo = gem_to_virtio_gpu_obj(objs->objs[0]); struct virtio_gpu_transfer_to_host_2d *cmd_p; @@ -624,7 +622,7 @@ void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, cmd_p->r.x = cpu_to_le32(x); cmd_p->r.y = cpu_to_le32(y);
- virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence, commit); }
static void @@ -647,7 +645,7 @@ virtio_gpu_cmd_resource_attach_backing(struct virtio_gpu_device *vgdev, vbuf->data_buf = ents; vbuf->data_size = sizeof(*ents) * nents;
- virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence, true); }
static void virtio_gpu_cmd_get_display_info_cb(struct virtio_gpu_device *vgdev, @@ -779,7 +777,7 @@ int virtio_gpu_cmd_get_display_info(struct virtio_gpu_device *vgdev)
vgdev->display_info_pending = true; cmd_p->type = cpu_to_le32(VIRTIO_GPU_CMD_GET_DISPLAY_INFO); - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, true); return 0; }
@@ -802,7 +800,7 @@ int virtio_gpu_cmd_get_capset_info(struct virtio_gpu_device *vgdev, int idx)
cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_GET_CAPSET_INFO); cmd_p->capset_index = cpu_to_le32(idx); - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, true); return 0; }
@@ -877,7 +875,7 @@ int virtio_gpu_cmd_get_capset(struct virtio_gpu_device *vgdev, cmd_p->capset_id = cpu_to_le32(vgdev->capsets[idx].id); cmd_p->capset_version = cpu_to_le32(version); *cache_p = cache_ent; - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, true);
return 0; } @@ -904,7 +902,7 @@ int virtio_gpu_cmd_get_edids(struct virtio_gpu_device *vgdev) resp_buf); cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_GET_EDID); cmd_p->scanout = cpu_to_le32(scanout); - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, true); }
return 0; @@ -924,7 +922,7 @@ void virtio_gpu_cmd_context_create(struct virtio_gpu_device *vgdev, uint32_t id, cmd_p->nlen = cpu_to_le32(nlen); strncpy(cmd_p->debug_name, name, sizeof(cmd_p->debug_name) - 1); cmd_p->debug_name[sizeof(cmd_p->debug_name) - 1] = 0; - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, true); }
void virtio_gpu_cmd_context_destroy(struct virtio_gpu_device *vgdev, @@ -938,7 +936,7 @@ void virtio_gpu_cmd_context_destroy(struct virtio_gpu_device *vgdev,
cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_CTX_DESTROY); cmd_p->hdr.ctx_id = cpu_to_le32(id); - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, true); }
void virtio_gpu_cmd_context_attach_resource(struct virtio_gpu_device *vgdev, @@ -956,7 +954,7 @@ void virtio_gpu_cmd_context_attach_resource(struct virtio_gpu_device *vgdev, cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_CTX_ATTACH_RESOURCE); cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id); cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle); - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, true);
}
@@ -975,7 +973,7 @@ void virtio_gpu_cmd_context_detach_resource(struct virtio_gpu_device *vgdev, cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_CTX_DETACH_RESOURCE); cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id); cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle); - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf, true); }
void @@ -1006,7 +1004,7 @@ virtio_gpu_cmd_resource_create_3d(struct virtio_gpu_device *vgdev, cmd_p->nr_samples = cpu_to_le32(params->nr_samples); cmd_p->flags = cpu_to_le32(params->flags);
- virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence, true); bo->created = true; }
@@ -1039,7 +1037,7 @@ void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev, cmd_p->offset = cpu_to_le64(offset); cmd_p->level = cpu_to_le32(level);
- virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence, true); }
void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev, @@ -1065,7 +1063,7 @@ void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev, cmd_p->offset = cpu_to_le64(offset); cmd_p->level = cpu_to_le32(level);
- virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence, true); }
void virtio_gpu_cmd_submit(struct virtio_gpu_device *vgdev, @@ -1088,7 +1086,7 @@ void virtio_gpu_cmd_submit(struct virtio_gpu_device *vgdev, cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id); cmd_p->size = cpu_to_le32(data_size);
- virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence, true); }
int virtio_gpu_object_attach(struct virtio_gpu_device *vgdev,
On Thu, Feb 06, 2020 at 11:22:14AM -0800, Chia-I Wu wrote:
The global disable_notify state does not scale well when we start using it in more places and when there are multiple threads. Use command-level bools to control whether to notify or not.
Hmm, I don't like passing around the bool everywhere that much.
How about this alternative approach: Flip the default to NOT kick commands. Have an explicit kick function to request a kick when needed. Maybe automatically kick fenced commands.
cheers, Gerd
Hi,
This series replaces the global disable_notify state by command-level bools to control vq kicks. When command batching is applied to more places, this prevents one process from affecting another process.
v2: update to this convention
virtio_gpu_cmd_foo: add foo and commit virtio_gpu_add_foo: add foo but do not commit
However, the spinlock is now grabbed twice in both add and commit.
Call virtqueue_kick_prepare once in virtio_gpu_enable_notify, not whenever a command is added. This should be more efficient since the intention is to batch commands.
Signed-off-by: Chia-I Wu olvaffe@gmail.com --- drivers/gpu/drm/virtio/virtgpu_drv.h | 1 - drivers/gpu/drm/virtio/virtgpu_vq.c | 34 ++++++++++++++++++---------- 2 files changed, 22 insertions(+), 13 deletions(-)
diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.h b/drivers/gpu/drm/virtio/virtgpu_drv.h index d278c8c50f394..d3fcc25068132 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.h +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h @@ -178,7 +178,6 @@ struct virtio_gpu_device { bool vqs_ready;
bool disable_notify; - bool pending_notify;
struct ida resource_ida;
diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c index cc02fc4bab2a5..0c9be2142ba1a 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -367,16 +367,13 @@ static void virtio_gpu_queue_ctrl_sgs(struct virtio_gpu_device *vgdev,
trace_virtio_gpu_cmd_queue(vq, virtio_gpu_vbuf_ctrl_hdr(vbuf));
- notify = virtqueue_kick_prepare(vq); + if (!vgdev->disable_notify) + notify = virtqueue_kick_prepare(vq);
spin_unlock(&vgdev->ctrlq.qlock);
- if (notify) { - if (vgdev->disable_notify) - vgdev->pending_notify = true; - else - virtqueue_notify(vq); - } + if (notify) + virtqueue_notify(vq); }
static void virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev, @@ -432,6 +429,23 @@ static void virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev, } }
+static void virtio_gpu_commit_ctrl(struct virtio_gpu_device *vgdev) +{ + struct virtqueue *vq = vgdev->ctrlq.vq; + bool notify; + + spin_lock(&vgdev->ctrlq.qlock); + notify = virtqueue_kick_prepare(vq); + spin_unlock(&vgdev->ctrlq.qlock); + + /* Do not call virtqueue_notify with the lock held because + * virtio_gpu_dequeue_ctrl_func may contend for the lock if an irq is + * generated while we are in virtqueue_notify. + */ + if (notify) + virtqueue_notify(vq); +} + void virtio_gpu_disable_notify(struct virtio_gpu_device *vgdev) { vgdev->disable_notify = true; @@ -440,11 +454,7 @@ void virtio_gpu_disable_notify(struct virtio_gpu_device *vgdev) void virtio_gpu_enable_notify(struct virtio_gpu_device *vgdev) { vgdev->disable_notify = false; - - if (!vgdev->pending_notify) - return; - vgdev->pending_notify = false; - virtqueue_notify(vgdev->ctrlq.vq); + virtio_gpu_commit_ctrl(vgdev); }
static void virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev,
virtio_gpu_queue_ctrl_sgs queues only. virtio_gpu_commit_ctrl must be explicitly called. This however means that we need to grab the spinlock twice.
Signed-off-by: Chia-I Wu olvaffe@gmail.com --- drivers/gpu/drm/virtio/virtgpu_vq.c | 29 ++++++++++++++++++++++------- 1 file changed, 22 insertions(+), 7 deletions(-)
diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c index 0c9be2142ba1a..d76b24f2b2bff 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -367,13 +367,7 @@ static void virtio_gpu_queue_ctrl_sgs(struct virtio_gpu_device *vgdev,
trace_virtio_gpu_cmd_queue(vq, virtio_gpu_vbuf_ctrl_hdr(vbuf));
- if (!vgdev->disable_notify) - notify = virtqueue_kick_prepare(vq); - spin_unlock(&vgdev->ctrlq.qlock); - - if (notify) - virtqueue_notify(vq); }
static void virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev, @@ -434,6 +428,9 @@ static void virtio_gpu_commit_ctrl(struct virtio_gpu_device *vgdev) struct virtqueue *vq = vgdev->ctrlq.vq; bool notify;
+ if (vgdev->disable_notify) + return; + spin_lock(&vgdev->ctrlq.qlock); notify = virtqueue_kick_prepare(vq); spin_unlock(&vgdev->ctrlq.qlock); @@ -525,6 +522,7 @@ void virtio_gpu_cmd_create_resource(struct virtio_gpu_device *vgdev, cmd_p->height = cpu_to_le32(params->height);
virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_commit_ctrl(vgdev); bo->created = true; }
@@ -541,6 +539,7 @@ void virtio_gpu_cmd_unref_resource(struct virtio_gpu_device *vgdev, cmd_p->resource_id = cpu_to_le32(resource_id);
virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev); }
static void virtio_gpu_cmd_resource_inval_backing(struct virtio_gpu_device *vgdev, @@ -557,6 +556,7 @@ static void virtio_gpu_cmd_resource_inval_backing(struct virtio_gpu_device *vgde cmd_p->resource_id = cpu_to_le32(resource_id);
virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_commit_ctrl(vgdev); }
void virtio_gpu_cmd_set_scanout(struct virtio_gpu_device *vgdev, @@ -579,6 +579,7 @@ void virtio_gpu_cmd_set_scanout(struct virtio_gpu_device *vgdev, cmd_p->r.y = cpu_to_le32(y);
virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev); }
void virtio_gpu_cmd_resource_flush(struct virtio_gpu_device *vgdev, @@ -600,6 +601,7 @@ void virtio_gpu_cmd_resource_flush(struct virtio_gpu_device *vgdev, cmd_p->r.y = cpu_to_le32(y);
virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev); }
void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, @@ -632,6 +634,7 @@ void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, cmd_p->r.y = cpu_to_le32(y);
virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_commit_ctrl(vgdev); }
static void @@ -655,6 +658,7 @@ virtio_gpu_cmd_resource_attach_backing(struct virtio_gpu_device *vgdev, vbuf->data_size = sizeof(*ents) * nents;
virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_commit_ctrl(vgdev); }
static void virtio_gpu_cmd_get_display_info_cb(struct virtio_gpu_device *vgdev, @@ -787,6 +791,7 @@ int virtio_gpu_cmd_get_display_info(struct virtio_gpu_device *vgdev) vgdev->display_info_pending = true; cmd_p->type = cpu_to_le32(VIRTIO_GPU_CMD_GET_DISPLAY_INFO); virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev); return 0; }
@@ -810,6 +815,7 @@ int virtio_gpu_cmd_get_capset_info(struct virtio_gpu_device *vgdev, int idx) cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_GET_CAPSET_INFO); cmd_p->capset_index = cpu_to_le32(idx); virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev); return 0; }
@@ -885,6 +891,7 @@ int virtio_gpu_cmd_get_capset(struct virtio_gpu_device *vgdev, cmd_p->capset_version = cpu_to_le32(version); *cache_p = cache_ent; virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev);
return 0; } @@ -912,6 +919,7 @@ int virtio_gpu_cmd_get_edids(struct virtio_gpu_device *vgdev) cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_GET_EDID); cmd_p->scanout = cpu_to_le32(scanout); virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev); }
return 0; @@ -932,6 +940,7 @@ void virtio_gpu_cmd_context_create(struct virtio_gpu_device *vgdev, uint32_t id, strncpy(cmd_p->debug_name, name, sizeof(cmd_p->debug_name) - 1); cmd_p->debug_name[sizeof(cmd_p->debug_name) - 1] = 0; virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev); }
void virtio_gpu_cmd_context_destroy(struct virtio_gpu_device *vgdev, @@ -946,6 +955,7 @@ void virtio_gpu_cmd_context_destroy(struct virtio_gpu_device *vgdev, cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_CTX_DESTROY); cmd_p->hdr.ctx_id = cpu_to_le32(id); virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev); }
void virtio_gpu_cmd_context_attach_resource(struct virtio_gpu_device *vgdev, @@ -964,7 +974,7 @@ void virtio_gpu_cmd_context_attach_resource(struct virtio_gpu_device *vgdev, cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id); cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle); virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); - + virtio_gpu_commit_ctrl(vgdev); }
void virtio_gpu_cmd_context_detach_resource(struct virtio_gpu_device *vgdev, @@ -983,6 +993,7 @@ void virtio_gpu_cmd_context_detach_resource(struct virtio_gpu_device *vgdev, cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id); cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle); virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_commit_ctrl(vgdev); }
void @@ -1014,6 +1025,7 @@ virtio_gpu_cmd_resource_create_3d(struct virtio_gpu_device *vgdev, cmd_p->flags = cpu_to_le32(params->flags);
virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_commit_ctrl(vgdev); bo->created = true; }
@@ -1047,6 +1059,7 @@ void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev, cmd_p->level = cpu_to_le32(level);
virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_commit_ctrl(vgdev); }
void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev, @@ -1073,6 +1086,7 @@ void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev, cmd_p->level = cpu_to_le32(level);
virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_commit_ctrl(vgdev); }
void virtio_gpu_cmd_submit(struct virtio_gpu_device *vgdev, @@ -1096,6 +1110,7 @@ void virtio_gpu_cmd_submit(struct virtio_gpu_device *vgdev, cmd_p->size = cpu_to_le32(data_size);
virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); + virtio_gpu_commit_ctrl(vgdev); }
int virtio_gpu_object_attach(struct virtio_gpu_device *vgdev,
Hi,
@@ -541,6 +539,7 @@ void virtio_gpu_cmd_unref_resource(struct virtio_gpu_device *vgdev, cmd_p->resource_id = cpu_to_le32(resource_id);
virtio_gpu_queue_ctrl_buffer(vgdev, vbuf);
- virtio_gpu_commit_ctrl(vgdev);
}
Well, I was more thinking about adding that to higher-level functions so we get more actual batching out of it ...
cheers, Gerd
The global disable_notify state does not scale well when we start using it in more places and when there are multiple threads. Allow individual commands to be added to the virtqueue but not committed.
The naming conventions are
virtio_gpu_cmd_foo -> add foo and commit virtio_gpu_add_bar -> add bar but do not commit
virtio_gpu_{disable,enable}_notify is replaced by virtio_gpu_commit_ctrl.
Signed-off-by: Chia-I Wu olvaffe@gmail.com --- drivers/gpu/drm/virtio/virtgpu_drv.h | 42 +++++++++++++++++++++----- drivers/gpu/drm/virtio/virtgpu_plane.c | 10 +++--- drivers/gpu/drm/virtio/virtgpu_vq.c | 25 +++------------ 3 files changed, 42 insertions(+), 35 deletions(-)
diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.h b/drivers/gpu/drm/virtio/virtgpu_drv.h index d3fcc25068132..2365f196f3169 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.h +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h @@ -177,8 +177,6 @@ struct virtio_gpu_device { struct kmem_cache *vbufs; bool vqs_ready;
- bool disable_notify; - struct ida resource_ida;
wait_queue_head_t resp_wq; @@ -256,6 +254,9 @@ void virtio_gpu_array_put_free_work(struct work_struct *work); /* virtio vg */ int virtio_gpu_alloc_vbufs(struct virtio_gpu_device *vgdev); void virtio_gpu_free_vbufs(struct virtio_gpu_device *vgdev); + +void virtio_gpu_commit_ctrl(struct virtio_gpu_device *vgdev); + void virtio_gpu_cmd_create_resource(struct virtio_gpu_device *vgdev, struct virtio_gpu_object *bo, struct virtio_gpu_object_params *params, @@ -263,20 +264,48 @@ void virtio_gpu_cmd_create_resource(struct virtio_gpu_device *vgdev, struct virtio_gpu_fence *fence); void virtio_gpu_cmd_unref_resource(struct virtio_gpu_device *vgdev, uint32_t resource_id); -void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, + +void virtio_gpu_add_transfer_to_host_2d(struct virtio_gpu_device *vgdev, uint64_t offset, uint32_t width, uint32_t height, uint32_t x, uint32_t y, struct virtio_gpu_object_array *objs, struct virtio_gpu_fence *fence); -void virtio_gpu_cmd_resource_flush(struct virtio_gpu_device *vgdev, + +static inline void +virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, + uint64_t offset, + uint32_t width, uint32_t height, + uint32_t x, uint32_t y, + struct virtio_gpu_object_array *objs, + struct virtio_gpu_fence *fence) +{ + virtio_gpu_add_transfer_to_host_2d(vgdev, offset, width, height, x, y, + objs, fence); + virtio_gpu_commit_ctrl(vgdev); +} + +void virtio_gpu_add_resource_flush(struct virtio_gpu_device *vgdev, uint32_t resource_id, uint32_t x, uint32_t y, uint32_t width, uint32_t height); -void virtio_gpu_cmd_set_scanout(struct virtio_gpu_device *vgdev, + +void virtio_gpu_add_set_scanout(struct virtio_gpu_device *vgdev, uint32_t scanout_id, uint32_t resource_id, uint32_t width, uint32_t height, uint32_t x, uint32_t y); + +static inline void +virtio_gpu_cmd_set_scanout(struct virtio_gpu_device *vgdev, + uint32_t scanout_id, uint32_t resource_id, + uint32_t width, uint32_t height, + uint32_t x, uint32_t y) +{ + virtio_gpu_add_set_scanout(vgdev, scanout_id, resource_id, + width, height, x, y); + virtio_gpu_commit_ctrl(vgdev); +} + int virtio_gpu_object_attach(struct virtio_gpu_device *vgdev, struct virtio_gpu_object *obj, struct virtio_gpu_fence *fence); @@ -332,9 +361,6 @@ void virtio_gpu_dequeue_ctrl_func(struct work_struct *work); void virtio_gpu_dequeue_cursor_func(struct work_struct *work); void virtio_gpu_dequeue_fence_func(struct work_struct *work);
-void virtio_gpu_disable_notify(struct virtio_gpu_device *vgdev); -void virtio_gpu_enable_notify(struct virtio_gpu_device *vgdev); - /* virtio_gpu_display.c */ void virtio_gpu_modeset_init(struct virtio_gpu_device *vgdev); void virtio_gpu_modeset_fini(struct virtio_gpu_device *vgdev); diff --git a/drivers/gpu/drm/virtio/virtgpu_plane.c b/drivers/gpu/drm/virtio/virtgpu_plane.c index ac42c84d2d7f2..448ee01cce026 100644 --- a/drivers/gpu/drm/virtio/virtgpu_plane.c +++ b/drivers/gpu/drm/virtio/virtgpu_plane.c @@ -122,7 +122,7 @@ static void virtio_gpu_update_dumb_bo(struct virtio_gpu_device *vgdev, return; virtio_gpu_array_add_obj(objs, &bo->base.base);
- virtio_gpu_cmd_transfer_to_host_2d(vgdev, off, w, h, x, y, + virtio_gpu_add_transfer_to_host_2d(vgdev, off, w, h, x, y, objs, NULL); }
@@ -154,8 +154,6 @@ static void virtio_gpu_primary_plane_update(struct drm_plane *plane, if (!drm_atomic_helper_damage_merged(old_state, plane->state, &rect)) return;
- virtio_gpu_disable_notify(vgdev); - bo = gem_to_virtio_gpu_obj(plane->state->fb->obj[0]); if (bo->dumb) virtio_gpu_update_dumb_bo(vgdev, plane->state, &rect); @@ -173,7 +171,7 @@ static void virtio_gpu_primary_plane_update(struct drm_plane *plane, plane->state->src_h >> 16, plane->state->src_x >> 16, plane->state->src_y >> 16); - virtio_gpu_cmd_set_scanout(vgdev, output->index, + virtio_gpu_add_set_scanout(vgdev, output->index, bo->hw_res_handle, plane->state->src_w >> 16, plane->state->src_h >> 16, @@ -181,13 +179,13 @@ static void virtio_gpu_primary_plane_update(struct drm_plane *plane, plane->state->src_y >> 16); }
- virtio_gpu_cmd_resource_flush(vgdev, bo->hw_res_handle, + virtio_gpu_add_resource_flush(vgdev, bo->hw_res_handle, rect.x1, rect.y1, rect.x2 - rect.x1, rect.y2 - rect.y1);
- virtio_gpu_enable_notify(vgdev); + virtio_gpu_commit_ctrl(vgdev); }
static int virtio_gpu_cursor_prepare_fb(struct drm_plane *plane, diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c index d76b24f2b2bff..4f2c24c4ddc3f 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -423,14 +423,11 @@ static void virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev, } }
-static void virtio_gpu_commit_ctrl(struct virtio_gpu_device *vgdev) +void virtio_gpu_commit_ctrl(struct virtio_gpu_device *vgdev) { struct virtqueue *vq = vgdev->ctrlq.vq; bool notify;
- if (vgdev->disable_notify) - return; - spin_lock(&vgdev->ctrlq.qlock); notify = virtqueue_kick_prepare(vq); spin_unlock(&vgdev->ctrlq.qlock); @@ -443,17 +440,6 @@ static void virtio_gpu_commit_ctrl(struct virtio_gpu_device *vgdev) virtqueue_notify(vq); }
-void virtio_gpu_disable_notify(struct virtio_gpu_device *vgdev) -{ - vgdev->disable_notify = true; -} - -void virtio_gpu_enable_notify(struct virtio_gpu_device *vgdev) -{ - vgdev->disable_notify = false; - virtio_gpu_commit_ctrl(vgdev); -} - static void virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev, struct virtio_gpu_vbuffer *vbuf) { @@ -559,7 +545,7 @@ static void virtio_gpu_cmd_resource_inval_backing(struct virtio_gpu_device *vgde virtio_gpu_commit_ctrl(vgdev); }
-void virtio_gpu_cmd_set_scanout(struct virtio_gpu_device *vgdev, +void virtio_gpu_add_set_scanout(struct virtio_gpu_device *vgdev, uint32_t scanout_id, uint32_t resource_id, uint32_t width, uint32_t height, uint32_t x, uint32_t y) @@ -579,10 +565,9 @@ void virtio_gpu_cmd_set_scanout(struct virtio_gpu_device *vgdev, cmd_p->r.y = cpu_to_le32(y);
virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); - virtio_gpu_commit_ctrl(vgdev); }
-void virtio_gpu_cmd_resource_flush(struct virtio_gpu_device *vgdev, +void virtio_gpu_add_resource_flush(struct virtio_gpu_device *vgdev, uint32_t resource_id, uint32_t x, uint32_t y, uint32_t width, uint32_t height) @@ -601,10 +586,9 @@ void virtio_gpu_cmd_resource_flush(struct virtio_gpu_device *vgdev, cmd_p->r.y = cpu_to_le32(y);
virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); - virtio_gpu_commit_ctrl(vgdev); }
-void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, +void virtio_gpu_add_transfer_to_host_2d(struct virtio_gpu_device *vgdev, uint64_t offset, uint32_t width, uint32_t height, uint32_t x, uint32_t y, @@ -634,7 +618,6 @@ void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, cmd_p->r.y = cpu_to_le32(y);
virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, fence); - virtio_gpu_commit_ctrl(vgdev); }
static void
dri-devel@lists.freedesktop.org
-
Chia-I Wu -
Gerd Hoffmann