[PATCH v3 0/7] virtio-gpu: bugfixes, virgl/3d rendering support

List overview All Threads
Download

newer

older

[Bug 94627] Game Risen on wine...

[PATCH] virtio-gpu: fix output...

Gerd Hoffmann

2 Oct 2015 2 Oct '15

11:58 a.m.

Hi,

Here comes the 3d rendering support patch series for virtio-gpu, along with a few bugfixes and page-flip support. The latter needs some careful review.

Patches 1-3 carry bugfixes. Patches 4-6 bring 3d rendering support. Patch 7 brings page-flip support.

Corresponsing qemu patches have been posted too. Most recent patch series post is here (there will be an update soon, git tree is already more recent than that and is recommended for testing): https://www.mail-archive.com/qemu-devel@nongnu.org/msg319956.html

Thee are also git branches with the kernel and qemu code: https://www.kraxel.org/cgit/linux/log/?h=virtio-gpu https://www.kraxel.org/cgit/qemu/log/?h=rebase/virgl-wip

please review, Gerd

Dave Airlie (2): virtio-gpu: add basic prime support virtio-gpu: mark as a render gpu

Gerd Hoffmann (5): virtio-gpu: add virtio_gpu_queue_ctrl_buffer_locked virtio-gpu: add & use virtio_gpu_queue_fenced_ctrl_buffer virtio-gpu: wait for cursor updates finish virtio-gpu: add 3d/virgl support [wip] virtio-gpu: add page flip support

drivers/gpu/drm/virtio/Makefile | 3 +- drivers/gpu/drm/virtio/virtgpu_display.c | 57 ++- drivers/gpu/drm/virtio/virtgpu_drv.c | 28 +- drivers/gpu/drm/virtio/virtgpu_drv.h | 72 ++++ drivers/gpu/drm/virtio/virtgpu_fence.c | 2 +- drivers/gpu/drm/virtio/virtgpu_gem.c | 41 +++ drivers/gpu/drm/virtio/virtgpu_ioctl.c | 572 +++++++++++++++++++++++++++++++ drivers/gpu/drm/virtio/virtgpu_kms.c | 133 ++++++- drivers/gpu/drm/virtio/virtgpu_prime.c | 71 ++++ drivers/gpu/drm/virtio/virtgpu_ttm.c | 1 + drivers/gpu/drm/virtio/virtgpu_vq.c | 322 ++++++++++++++++- include/uapi/drm/Kbuild | 1 + include/uapi/drm/virtgpu_drm.h | 167 +++++++++ include/uapi/linux/virtio_gpu.h | 112 +++++- 14 files changed, 1564 insertions(+), 18 deletions(-) create mode 100644 drivers/gpu/drm/virtio/virtgpu_ioctl.c create mode 100644 drivers/gpu/drm/virtio/virtgpu_prime.c create mode 100644 include/uapi/drm/virtgpu_drm.h

-- 1.8.3.1

Show replies by date

Gerd Hoffmann

2 Oct 2 Oct

11:58 a.m.

New subject: [PATCH v3 1/7] virtio-gpu: add virtio_gpu_queue_ctrl_buffer_locked

Add virtio_gpu_queue_ctrl_buffer_locked function, which does the same as virtio_gpu_queue_ctrl_buffer but does not take the virtqueue lock. The caller must hold the lock instead.

Signed-off-by: Gerd Hoffmann kraxel@redhat.com --- drivers/gpu/drm/virtio/virtgpu_vq.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c index 1698669f..5b9bc24 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -293,8 +293,8 @@ void virtio_gpu_dequeue_cursor_func(struct work_struct *work) wake_up(&vgdev->cursorq.ack_queue); }

-static int virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev, - struct virtio_gpu_vbuffer *vbuf) +static int virtio_gpu_queue_ctrl_buffer_locked(struct virtio_gpu_device *vgdev, + struct virtio_gpu_vbuffer *vbuf) { struct virtqueue *vq = vgdev->ctrlq.vq; struct scatterlist *sgs[3], vcmd, vout, vresp; @@ -320,7 +320,6 @@ static int virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev, incnt++; }

- spin_lock(&vgdev->ctrlq.qlock); retry: ret = virtqueue_add_sgs(vq, sgs, outcnt, incnt, vbuf, GFP_ATOMIC); if (ret == -ENOSPC) { @@ -331,13 +330,23 @@ retry: } else { virtqueue_kick(vq); } - spin_unlock(&vgdev->ctrlq.qlock);

if (!ret) ret = vq->num_free; return ret; }

+static int virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev, + struct virtio_gpu_vbuffer *vbuf) +{ + int rc; + + spin_lock(&vgdev->ctrlq.qlock); + rc = virtio_gpu_queue_ctrl_buffer_locked(vgdev, vbuf); + spin_unlock(&vgdev->ctrlq.qlock); + return rc; +} + static int virtio_gpu_queue_cursor(struct virtio_gpu_device *vgdev, struct virtio_gpu_vbuffer *vbuf) {

-- 1.8.3.1

Gerd Hoffmann

11:58 a.m.

New subject: [PATCH v3 2/7] virtio-gpu: add & use virtio_gpu_queue_fenced_ctrl_buffer

Add helper function to handle the submission of fenced control requests. Make sure we initialize the fence while holding the virtqueue lock, so requests can't be reordered.

Signed-off-by: Gerd Hoffmann kraxel@redhat.com --- drivers/gpu/drm/virtio/virtgpu_fence.c | 2 +- drivers/gpu/drm/virtio/virtgpu_vq.c | 40 +++++++++++++++++++++++++++++----- 2 files changed, 35 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/virtio/virtgpu_fence.c b/drivers/gpu/drm/virtio/virtgpu_fence.c index 1da6326..793ad9f 100644 --- a/drivers/gpu/drm/virtio/virtgpu_fence.c +++ b/drivers/gpu/drm/virtio/virtgpu_fence.c @@ -81,7 +81,7 @@ int virtio_gpu_fence_emit(struct virtio_gpu_device *vgdev, struct virtio_gpu_fence_driver *drv = &vgdev->fence_drv; unsigned long irq_flags;

- *fence = kmalloc(sizeof(struct virtio_gpu_fence), GFP_KERNEL); + *fence = kmalloc(sizeof(struct virtio_gpu_fence), GFP_ATOMIC); if ((*fence) == NULL) return -ENOMEM;

diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c index 5b9bc24..ee25e9a 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -347,6 +347,38 @@ static int virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev, return rc; }

+static int virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev, + struct virtio_gpu_vbuffer *vbuf, + struct virtio_gpu_ctrl_hdr *hdr, + struct virtio_gpu_fence **fence) +{ + struct virtqueue *vq = vgdev->ctrlq.vq; + int rc; + +again: + spin_lock(&vgdev->ctrlq.qlock); + + /* + * Make sure we have enouth space in the virtqueue. If not + * wait here until we have. + * + * Without that virtio_gpu_queue_ctrl_buffer_nolock might have + * to wait for free space, which can result in fence ids being + * submitted out-of-order. + */ + if (vq->num_free < 3) { + spin_unlock(&vgdev->ctrlq.qlock); + wait_event(vgdev->ctrlq.ack_queue, vq->num_free >= 3); + goto again; + } + + if (fence) + virtio_gpu_fence_emit(vgdev, hdr, fence); + rc = virtio_gpu_queue_ctrl_buffer_locked(vgdev, vbuf); + spin_unlock(&vgdev->ctrlq.qlock); + return rc; +} + static int virtio_gpu_queue_cursor(struct virtio_gpu_device *vgdev, struct virtio_gpu_vbuffer *vbuf) { @@ -499,9 +531,7 @@ void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev, cmd_p->r.x = x; cmd_p->r.y = y;

- if (fence) - virtio_gpu_fence_emit(vgdev, &cmd_p->hdr, fence); - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, &cmd_p->hdr, fence); }

static void @@ -524,9 +554,7 @@ virtio_gpu_cmd_resource_attach_backing(struct virtio_gpu_device *vgdev, vbuf->data_buf = ents; vbuf->data_size = sizeof(*ents) * nents;

- if (fence) - virtio_gpu_fence_emit(vgdev, &cmd_p->hdr, fence); - virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, &cmd_p->hdr, fence); }

static void virtio_gpu_cmd_get_display_info_cb(struct virtio_gpu_device *vgdev,

-- 1.8.3.1

Gerd Hoffmann

11:58 a.m.

New subject: [PATCH v3 3/7] virtio-gpu: wait for cursor updates finish

Signed-off-by: Gerd Hoffmann kraxel@redhat.com --- drivers/gpu/drm/virtio/virtgpu_display.c | 8 ++++++++ 1 file changed, 8 insertions(+)

diff --git a/drivers/gpu/drm/virtio/virtgpu_display.c b/drivers/gpu/drm/virtio/virtgpu_display.c index 4e160ef..c9c1427 100644 --- a/drivers/gpu/drm/virtio/virtgpu_display.c +++ b/drivers/gpu/drm/virtio/virtgpu_display.c @@ -90,6 +90,14 @@ static int virtio_gpu_crtc_cursor_set(struct drm_crtc *crtc, cpu_to_le32(64), cpu_to_le32(64), 0, 0, &fence); + ret = virtio_gpu_object_reserve(qobj, false); + if (!ret) { + reservation_object_add_excl_fence(qobj->tbo.resv, + &fence->f); + fence_put(&fence->f); + virtio_gpu_object_unreserve(qobj); + virtio_gpu_object_wait(qobj, false); + }

output->cursor.hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_UPDATE_CURSOR); output->cursor.resource_id = cpu_to_le32(qobj->hw_res_handle);

-- 1.8.3.1

Gerd Hoffmann

11:58 a.m.

New subject: [PATCH v3 4/7] virtio-gpu: add 3d/virgl support

Add the bits needed for opengl rendering support: query capabilities, new virtio commands, drm ioctls.

Signed-off-by: Dave Airlie airlied@redhat.com Signed-off-by: Gerd Hoffmann kraxel@redhat.com --- drivers/gpu/drm/virtio/Makefile | 3 +- drivers/gpu/drm/virtio/virtgpu_drv.c | 15 + drivers/gpu/drm/virtio/virtgpu_drv.h | 60 ++++ drivers/gpu/drm/virtio/virtgpu_gem.c | 41 +++ drivers/gpu/drm/virtio/virtgpu_ioctl.c | 572 +++++++++++++++++++++++++++++++++ drivers/gpu/drm/virtio/virtgpu_kms.c | 133 +++++++- drivers/gpu/drm/virtio/virtgpu_ttm.c | 1 + drivers/gpu/drm/virtio/virtgpu_vq.c | 265 +++++++++++++++ include/uapi/drm/Kbuild | 1 + include/uapi/drm/virtgpu_drm.h | 167 ++++++++++ include/uapi/linux/virtio_gpu.h | 112 ++++++- 11 files changed, 1367 insertions(+), 3 deletions(-) create mode 100644 drivers/gpu/drm/virtio/virtgpu_ioctl.c create mode 100644 include/uapi/drm/virtgpu_drm.h

diff --git a/drivers/gpu/drm/virtio/Makefile b/drivers/gpu/drm/virtio/Makefile index 2ee1602..da7bf19 100644 --- a/drivers/gpu/drm/virtio/Makefile +++ b/drivers/gpu/drm/virtio/Makefile @@ -6,6 +6,7 @@ ccflags-y := -Iinclude/drm

virtio-gpu-y := virtgpu_drv.o virtgpu_kms.o virtgpu_drm_bus.o virtgpu_gem.o \ virtgpu_fb.o virtgpu_display.o virtgpu_vq.o virtgpu_ttm.o \ - virtgpu_fence.o virtgpu_object.o virtgpu_debugfs.o virtgpu_plane.o + virtgpu_fence.o virtgpu_object.o virtgpu_debugfs.o virtgpu_plane.o \ + virtgpu_ioctl.o

obj-$(CONFIG_DRM_VIRTIO_GPU) += virtio-gpu.o diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.c b/drivers/gpu/drm/virtio/virtgpu_drv.c index 7d9610a..c27d3a3 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.c +++ b/drivers/gpu/drm/virtio/virtgpu_drv.c @@ -73,6 +73,14 @@ static struct virtio_device_id id_table[] = { };

static unsigned int features[] = { +#ifdef __LITTLE_ENDIAN + /* + * Gallium command stream send by virgl is native endian. + * Because of that we only support little endian guests on + * little endian hosts. + */ + VIRTIO_GPU_F_VIRGL, +#endif }; static struct virtio_driver virtio_gpu_driver = { .feature_table = features, @@ -114,6 +122,8 @@ static struct drm_driver driver = { .set_busid = drm_virtio_set_busid, .load = virtio_gpu_driver_load, .unload = virtio_gpu_driver_unload, + .open = virtio_gpu_driver_open, + .postclose = virtio_gpu_driver_postclose,

.dumb_create = virtio_gpu_mode_dumb_create, .dumb_map_offset = virtio_gpu_mode_dumb_mmap, @@ -125,8 +135,13 @@ static struct drm_driver driver = { #endif

.gem_free_object = virtio_gpu_gem_free_object, + .gem_open_object = virtio_gpu_gem_object_open, + .gem_close_object = virtio_gpu_gem_object_close, .fops = &virtio_gpu_driver_fops,

+ .ioctls = virtio_gpu_ioctls, + .num_ioctls = DRM_VIRTIO_NUM_IOCTLS, + .name = DRIVER_NAME, .desc = DRIVER_DESC, .date = DRIVER_DATE, diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.h b/drivers/gpu/drm/virtio/virtgpu_drv.h index 6d4db2d..2719108 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.h +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h @@ -146,6 +146,21 @@ struct virtio_gpu_queue { struct work_struct dequeue_work; };

+struct virtio_gpu_drv_capset { + uint32_t id; + uint32_t max_version; + uint32_t max_size; +}; + +struct virtio_gpu_drv_cap_cache { + struct list_head head; + void *caps_cache; + uint32_t id; + uint32_t version; + uint32_t size; + atomic_t is_valid; +}; + struct virtio_gpu_device { struct device *dev; struct drm_device *ddev; @@ -179,7 +194,13 @@ struct virtio_gpu_device { struct idr ctx_id_idr; spinlock_t ctx_id_idr_lock;

+ bool has_virgl_3d; + struct work_struct config_changed_work; + + struct virtio_gpu_drv_capset *capsets; + uint32_t num_capsets; + struct list_head cap_cache; };

struct virtio_gpu_fpriv { @@ -193,6 +214,8 @@ extern struct drm_ioctl_desc virtio_gpu_ioctls[DRM_VIRTIO_NUM_IOCTLS]; /* virtio_kms.c */ int virtio_gpu_driver_load(struct drm_device *dev, unsigned long flags); int virtio_gpu_driver_unload(struct drm_device *dev); +int virtio_gpu_driver_open(struct drm_device *dev, struct drm_file *file); +void virtio_gpu_driver_postclose(struct drm_device *dev, struct drm_file *file);

/* virtio_gem.c */ void virtio_gpu_gem_free_object(struct drm_gem_object *gem_obj); @@ -203,6 +226,10 @@ int virtio_gpu_gem_create(struct drm_file *file, uint64_t size, struct drm_gem_object **obj_p, uint32_t *handle_p); +int virtio_gpu_gem_object_open(struct drm_gem_object *obj, + struct drm_file *file); +void virtio_gpu_gem_object_close(struct drm_gem_object *obj, + struct drm_file *file); struct virtio_gpu_object *virtio_gpu_alloc_object(struct drm_device *dev, size_t size, bool kernel, bool pinned); @@ -260,10 +287,43 @@ void virtio_gpu_cursor_ping(struct virtio_gpu_device *vgdev, int virtio_gpu_cmd_get_display_info(struct virtio_gpu_device *vgdev); void virtio_gpu_cmd_resource_inval_backing(struct virtio_gpu_device *vgdev, uint32_t resource_id); +int virtio_gpu_cmd_get_capset_info(struct virtio_gpu_device *vgdev, int idx); +int virtio_gpu_cmd_get_capset(struct virtio_gpu_device *vgdev, + int idx, int version, + struct virtio_gpu_drv_cap_cache **cache_p); +void virtio_gpu_cmd_context_create(struct virtio_gpu_device *vgdev, uint32_t id, + uint32_t nlen, const char *name); +void virtio_gpu_cmd_context_destroy(struct virtio_gpu_device *vgdev, + uint32_t id); +void virtio_gpu_cmd_context_attach_resource(struct virtio_gpu_device *vgdev, + uint32_t ctx_id, + uint32_t resource_id); +void virtio_gpu_cmd_context_detach_resource(struct virtio_gpu_device *vgdev, + uint32_t ctx_id, + uint32_t resource_id); +void virtio_gpu_cmd_submit(struct virtio_gpu_device *vgdev, + void *data, uint32_t data_size, + uint32_t ctx_id, struct virtio_gpu_fence **fence); +void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev, + uint32_t resource_id, uint32_t ctx_id, + uint64_t offset, uint32_t level, + struct virtio_gpu_box *box, + struct virtio_gpu_fence **fence); +void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev, + uint32_t resource_id, uint32_t ctx_id, + uint64_t offset, uint32_t level, + struct virtio_gpu_box *box, + struct virtio_gpu_fence **fence); +void +virtio_gpu_cmd_resource_create_3d(struct virtio_gpu_device *vgdev, + struct virtio_gpu_resource_create_3d *rc_3d, + struct virtio_gpu_fence **fence); void virtio_gpu_ctrl_ack(struct virtqueue *vq); void virtio_gpu_cursor_ack(struct virtqueue *vq); +void virtio_gpu_fence_ack(struct virtqueue *vq); 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);

/* virtio_gpu_display.c */ int virtio_gpu_framebuffer_init(struct drm_device *dev, diff --git a/drivers/gpu/drm/virtio/virtgpu_gem.c b/drivers/gpu/drm/virtio/virtgpu_gem.c index cfa0d27..1feb7ce 100644 --- a/drivers/gpu/drm/virtio/virtgpu_gem.c +++ b/drivers/gpu/drm/virtio/virtgpu_gem.c @@ -138,3 +138,44 @@ int virtio_gpu_mode_dumb_mmap(struct drm_file *file_priv, drm_gem_object_unreference_unlocked(gobj); return 0; } + +int virtio_gpu_gem_object_open(struct drm_gem_object *obj, + struct drm_file *file) +{ + struct virtio_gpu_device *vgdev = obj->dev->dev_private; + struct virtio_gpu_fpriv *vfpriv = file->driver_priv; + struct virtio_gpu_object *qobj = gem_to_virtio_gpu_obj(obj); + int r; + + if (!vgdev->has_virgl_3d) + return 0; + + r = virtio_gpu_object_reserve(qobj, false); + if (r) + return r; + + virtio_gpu_cmd_context_attach_resource(vgdev, vfpriv->ctx_id, + qobj->hw_res_handle); + virtio_gpu_object_unreserve(qobj); + return 0; +} + +void virtio_gpu_gem_object_close(struct drm_gem_object *obj, + struct drm_file *file) +{ + struct virtio_gpu_device *vgdev = obj->dev->dev_private; + struct virtio_gpu_fpriv *vfpriv = file->driver_priv; + struct virtio_gpu_object *qobj = gem_to_virtio_gpu_obj(obj); + int r; + + if (!vgdev->has_virgl_3d) + return; + + r = virtio_gpu_object_reserve(qobj, false); + if (r) + return; + + virtio_gpu_cmd_context_detach_resource(vgdev, vfpriv->ctx_id, + qobj->hw_res_handle); + virtio_gpu_object_unreserve(qobj); +} diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c new file mode 100644 index 0000000..d5c9e02 --- /dev/null +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c @@ -0,0 +1,572 @@ +/* + * Copyright (C) 2015 Red Hat, Inc. + * All Rights Reserved. + * + * Authors: + * Dave Airlie + * Alon Levy + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include <drm/drmP.h> +#include "virtgpu_drv.h" +#include <drm/virtgpu_drm.h> +#include "ttm/ttm_execbuf_util.h" + +static void convert_to_hw_box(struct virtio_gpu_box *dst, + const struct drm_virtgpu_3d_box *src) +{ + dst->x = cpu_to_le32(src->x); + dst->y = cpu_to_le32(src->y); + dst->z = cpu_to_le32(src->z); + dst->w = cpu_to_le32(src->w); + dst->h = cpu_to_le32(src->h); + dst->d = cpu_to_le32(src->d); +} + +static int virtio_gpu_map_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct drm_virtgpu_map *virtio_gpu_map = data; + + return virtio_gpu_mode_dumb_mmap(file_priv, vgdev->ddev, + virtio_gpu_map->handle, + &virtio_gpu_map->offset); +} + +static int virtio_gpu_object_list_validate(struct ww_acquire_ctx *ticket, + struct list_head *head) +{ + struct ttm_validate_buffer *buf; + struct ttm_buffer_object *bo; + struct virtio_gpu_object *qobj; + int ret; + + ret = ttm_eu_reserve_buffers(ticket, head, true, NULL); + if (ret != 0) + return ret; + + list_for_each_entry(buf, head, head) { + bo = buf->bo; + qobj = container_of(bo, struct virtio_gpu_object, tbo); + ret = ttm_bo_validate(bo, &qobj->placement, false, false); + if (ret) { + ttm_eu_backoff_reservation(ticket, head); + return ret; + } + } + return 0; +} + +static void virtio_gpu_unref_list(struct list_head *head) +{ + struct 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_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 virtio_gpu_object *qobj; + int ret; + uint32_t *bo_handles = NULL; + void __user *user_bo_handles = NULL; + struct list_head validate_list; + struct ttm_validate_buffer *buflist = NULL; + struct ttm_validate_buffer cmdbuffer; + int i; + struct ww_acquire_ctx ticket; + void *buf; + + if (vgdev->has_virgl_3d == false) + return -ENOSYS; + + memset(&cmdbuffer, 0, sizeof(struct ttm_validate_buffer)); + INIT_LIST_HEAD(&validate_list); + if (exbuf->num_bo_handles) { + + bo_handles = drm_malloc_ab(exbuf->num_bo_handles, + sizeof(uint32_t)); + buflist = drm_calloc_large(exbuf->num_bo_handles, + sizeof(struct ttm_validate_buffer)); + if (!bo_handles || !buflist) { + drm_free_large(bo_handles); + drm_free_large(buflist); + return -ENOMEM; + } + + user_bo_handles = (void __user *)(uintptr_t)exbuf->bo_handles; + if (copy_from_user(bo_handles, user_bo_handles, + exbuf->num_bo_handles * sizeof(uint32_t))) { + ret = -EFAULT; + drm_free_large(bo_handles); + return ret; + } + + for (i = 0; i < exbuf->num_bo_handles; i++) { + gobj = drm_gem_object_lookup(dev, + drm_file, bo_handles[i]); + if (!gobj) { + drm_free_large(bo_handles); + drm_free_large(buflist); + return -ENOENT; + } + + qobj = gem_to_virtio_gpu_obj(gobj); + buflist[i].bo = &qobj->tbo; + + list_add(&buflist[i].head, &validate_list); + } + drm_free_large(bo_handles); + } + + ret = virtio_gpu_object_list_validate(&ticket, &validate_list); + if (ret) + goto out_free; + + buf = kmalloc(exbuf->size, GFP_KERNEL); + if (!buf) { + ret = -ENOMEM; + goto out_unresv; + } + if (copy_from_user(buf, (void __user *)(uintptr_t)exbuf->command, + exbuf->size)) { + kfree(buf); + ret = -EFAULT; + goto out_unresv; + } + virtio_gpu_cmd_submit(vgdev, buf, exbuf->size, + vfpriv->ctx_id, &fence); + + ttm_eu_fence_buffer_objects(&ticket, &validate_list, &fence->f); + + /* fence the command bo */ + virtio_gpu_unref_list(&validate_list); + drm_free_large(buflist); + fence_put(&fence->f); + return 0; + +out_unresv: + ttm_eu_backoff_reservation(&ticket, &validate_list); +out_free: + virtio_gpu_unref_list(&validate_list); + drm_free_large(buflist); + return ret; +} + +/* + * Usage of execbuffer: + * Relocations need to take into account the full VIRTIO_GPUDrawable size. + * However, the command as passed from user space must *not* contain the initial + * VIRTIO_GPUReleaseInfo struct (first XXX bytes) + */ +static int virtio_gpu_execbuffer_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_virtgpu_execbuffer *execbuffer = data; + return virtio_gpu_execbuffer(dev, execbuffer, file_priv); +} + + +static int virtio_gpu_getparam_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct drm_virtgpu_getparam *param = data; + int value; + + switch (param->param) { + case VIRTGPU_PARAM_3D_FEATURES: + value = vgdev->has_virgl_3d == true ? 1 : 0; + break; + default: + return -EINVAL; + } + if (copy_to_user((void __user *)(unsigned long)param->value, + &value, sizeof(int))) { + return -EFAULT; + } + return 0; +} + +static int virtio_gpu_resource_create_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct drm_virtgpu_resource_create *rc = data; + int ret; + uint32_t res_id; + struct virtio_gpu_object *qobj; + struct drm_gem_object *obj; + uint32_t handle = 0; + uint32_t size; + struct list_head validate_list; + struct ttm_validate_buffer mainbuf; + struct virtio_gpu_fence *fence = NULL; + struct ww_acquire_ctx ticket; + struct virtio_gpu_resource_create_3d rc_3d; + + if (vgdev->has_virgl_3d == false) { + if (rc->depth > 1) + return -EINVAL; + if (rc->nr_samples > 1) + return -EINVAL; + if (rc->last_level > 1) + return -EINVAL; + if (rc->target != 2) + return -EINVAL; + if (rc->array_size > 1) + return -EINVAL; + } + + INIT_LIST_HEAD(&validate_list); + memset(&mainbuf, 0, sizeof(struct ttm_validate_buffer)); + + virtio_gpu_resource_id_get(vgdev, &res_id); + + size = rc->size; + + /* allocate a single page size object */ + if (size == 0) + size = PAGE_SIZE; + + qobj = virtio_gpu_alloc_object(dev, size, false, false); + if (IS_ERR(qobj)) { + ret = PTR_ERR(qobj); + goto fail_id; + } + obj = &qobj->gem_base; + + if (!vgdev->has_virgl_3d) { + virtio_gpu_cmd_create_resource(vgdev, res_id, rc->format, + rc->width, rc->height); + + ret = virtio_gpu_object_attach(vgdev, qobj, res_id, NULL); + } else { + /* use a gem reference since unref list undoes them */ + drm_gem_object_reference(&qobj->gem_base); + mainbuf.bo = &qobj->tbo; + list_add(&mainbuf.head, &validate_list); + + ret = virtio_gpu_object_list_validate(&ticket, &validate_list); + if (ret) { + DRM_DEBUG("failed to validate\n"); + goto fail_unref; + } + + rc_3d.resource_id = cpu_to_le32(res_id); + rc_3d.target = cpu_to_le32(rc->target); + rc_3d.format = cpu_to_le32(rc->format); + rc_3d.bind = cpu_to_le32(rc->bind); + rc_3d.width = cpu_to_le32(rc->width); + rc_3d.height = cpu_to_le32(rc->height); + rc_3d.depth = cpu_to_le32(rc->depth); + rc_3d.array_size = cpu_to_le32(rc->array_size); + rc_3d.last_level = cpu_to_le32(rc->last_level); + rc_3d.nr_samples = cpu_to_le32(rc->nr_samples); + rc_3d.flags = cpu_to_le32(rc->flags); + + virtio_gpu_cmd_resource_create_3d(vgdev, &rc_3d, NULL); + ret = virtio_gpu_object_attach(vgdev, qobj, res_id, &fence); + if (ret) + goto fail_unref; + ttm_eu_fence_buffer_objects(&ticket, &validate_list, &fence->f); + } + + qobj->hw_res_handle = res_id; + + ret = drm_gem_handle_create(file_priv, obj, &handle); + if (ret) { + + drm_gem_object_release(obj); + if (vgdev->has_virgl_3d) { + virtio_gpu_unref_list(&validate_list); + fence_put(&fence->f); + } + return ret; + } + drm_gem_object_unreference_unlocked(obj); + + rc->res_handle = res_id; /* similiar to a VM address */ + rc->bo_handle = handle; + + if (vgdev->has_virgl_3d) { + virtio_gpu_unref_list(&validate_list); + fence_put(&fence->f); + } + return 0; +fail_unref: + if (vgdev->has_virgl_3d) { + virtio_gpu_unref_list(&validate_list); + fence_put(&fence->f); + } +//fail_obj: +// drm_gem_object_handle_unreference_unlocked(obj); +fail_id: + virtio_gpu_resource_id_put(vgdev, res_id); + return ret; +} + +static int virtio_gpu_resource_info_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_virtgpu_resource_info *ri = data; + struct drm_gem_object *gobj = NULL; + struct virtio_gpu_object *qobj = NULL; + + gobj = drm_gem_object_lookup(dev, file_priv, ri->bo_handle); + if (gobj == NULL) + return -ENOENT; + + qobj = gem_to_virtio_gpu_obj(gobj); + + ri->size = qobj->gem_base.size; + ri->res_handle = qobj->hw_res_handle; + drm_gem_object_unreference_unlocked(gobj); + return 0; +} + +static int virtio_gpu_transfer_from_host_ioctl(struct drm_device *dev, + void *data, + struct drm_file *file) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct virtio_gpu_fpriv *vfpriv = file->driver_priv; + struct drm_virtgpu_3d_transfer_from_host *args = data; + struct drm_gem_object *gobj = NULL; + struct virtio_gpu_object *qobj = NULL; + struct virtio_gpu_fence *fence; + int ret; + u32 offset = args->offset; + struct virtio_gpu_box box; + + if (vgdev->has_virgl_3d == false) + return -ENOSYS; + + gobj = drm_gem_object_lookup(dev, file, args->bo_handle); + if (gobj == NULL) + return -ENOENT; + + qobj = gem_to_virtio_gpu_obj(gobj); + + ret = virtio_gpu_object_reserve(qobj, false); + if (ret) + goto out; + + ret = ttm_bo_validate(&qobj->tbo, &qobj->placement, + true, false); + if (unlikely(ret)) + goto out_unres; + + convert_to_hw_box(&box, &args->box); + virtio_gpu_cmd_transfer_from_host_3d + (vgdev, qobj->hw_res_handle, + vfpriv->ctx_id, offset, args->level, + &box, &fence); + reservation_object_add_excl_fence(qobj->tbo.resv, + &fence->f); + + fence_put(&fence->f); +out_unres: + virtio_gpu_object_unreserve(qobj); +out: + drm_gem_object_unreference_unlocked(gobj); + return ret; +} + +static int virtio_gpu_transfer_to_host_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct virtio_gpu_fpriv *vfpriv = file->driver_priv; + struct drm_virtgpu_3d_transfer_to_host *args = data; + struct drm_gem_object *gobj = NULL; + struct virtio_gpu_object *qobj = NULL; + struct virtio_gpu_fence *fence; + struct virtio_gpu_box box; + int ret; + u32 offset = args->offset; + + gobj = drm_gem_object_lookup(dev, file, args->bo_handle); + if (gobj == NULL) + return -ENOENT; + + qobj = gem_to_virtio_gpu_obj(gobj); + + ret = virtio_gpu_object_reserve(qobj, false); + if (ret) + goto out; + + ret = ttm_bo_validate(&qobj->tbo, &qobj->placement, + true, false); + if (unlikely(ret)) + goto out_unres; + + convert_to_hw_box(&box, &args->box); + if (!vgdev->has_virgl_3d) { + virtio_gpu_cmd_transfer_to_host_2d + (vgdev, qobj->hw_res_handle, offset, + box.w, box.h, box.x, box.y, NULL); + } else { + virtio_gpu_cmd_transfer_to_host_3d + (vgdev, qobj->hw_res_handle, + vfpriv ? vfpriv->ctx_id : 0, offset, + args->level, &box, &fence); + reservation_object_add_excl_fence(qobj->tbo.resv, + &fence->f); + fence_put(&fence->f); + } + +out_unres: + virtio_gpu_object_unreserve(qobj); +out: + drm_gem_object_unreference_unlocked(gobj); + return ret; +} + +static int virtio_gpu_wait_ioctl(struct drm_device *dev, void *data, + struct drm_file *file) +{ + struct drm_virtgpu_3d_wait *args = data; + struct drm_gem_object *gobj = NULL; + struct virtio_gpu_object *qobj = NULL; + int ret; + bool nowait = false; + + gobj = drm_gem_object_lookup(dev, file, args->handle); + if (gobj == NULL) + return -ENOENT; + + qobj = gem_to_virtio_gpu_obj(gobj); + + if (args->flags & VIRTGPU_WAIT_NOWAIT) + nowait = true; + ret = virtio_gpu_object_wait(qobj, nowait); + + drm_gem_object_unreference_unlocked(gobj); + return ret; +} + +static int virtio_gpu_get_caps_ioctl(struct drm_device *dev, + void *data, struct drm_file *file) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct drm_virtgpu_get_caps *args = data; + int size; + int i; + int found_valid = -1; + int ret; + struct virtio_gpu_drv_cap_cache *cache_ent; + void *ptr; + if (vgdev->num_capsets == 0) + return -ENOSYS; + + spin_lock(&vgdev->display_info_lock); + for (i = 0; i < vgdev->num_capsets; i++) { + if (vgdev->capsets[i].id == args->cap_set_id) { + if (vgdev->capsets[i].max_version >= args->cap_set_ver) { + found_valid = i; + break; + } + } + } + + if (found_valid == -1) { + spin_unlock(&vgdev->display_info_lock); + return -EINVAL; + } + + size = vgdev->capsets[found_valid].max_size; + if (args->size > size) { + spin_unlock(&vgdev->display_info_lock); + return -EINVAL; + } + + list_for_each_entry(cache_ent, &vgdev->cap_cache, head) { + if (cache_ent->id == args->cap_set_id && + cache_ent->version == args->cap_set_ver) { + ptr = cache_ent->caps_cache; + spin_unlock(&vgdev->display_info_lock); + goto copy_exit; + } + } + spin_unlock(&vgdev->display_info_lock); + + /* not in cache - need to talk to hw */ + virtio_gpu_cmd_get_capset(vgdev, found_valid, args->cap_set_ver, + &cache_ent); + + ret = wait_event_timeout(vgdev->resp_wq, + atomic_read(&cache_ent->is_valid), 5 * HZ); + + ptr = cache_ent->caps_cache; + +copy_exit: + if (copy_to_user((void __user *)(unsigned long)args->addr, ptr, size)) + return -EFAULT; + + return 0; +} + +struct drm_ioctl_desc virtio_gpu_ioctls[DRM_VIRTIO_NUM_IOCTLS] = { + DRM_IOCTL_DEF_DRV(VIRTGPU_MAP, virtio_gpu_map_ioctl, + DRM_AUTH|DRM_UNLOCKED), + + DRM_IOCTL_DEF_DRV(VIRTGPU_EXECBUFFER, virtio_gpu_execbuffer_ioctl, + DRM_AUTH|DRM_UNLOCKED), + + DRM_IOCTL_DEF_DRV(VIRTGPU_GETPARAM, virtio_gpu_getparam_ioctl, + DRM_AUTH|DRM_UNLOCKED), + + DRM_IOCTL_DEF_DRV(VIRTGPU_RESOURCE_CREATE, + virtio_gpu_resource_create_ioctl, + DRM_AUTH|DRM_UNLOCKED), + + DRM_IOCTL_DEF_DRV(VIRTGPU_RESOURCE_INFO, virtio_gpu_resource_info_ioctl, + DRM_AUTH|DRM_UNLOCKED), + + /* make transfer async to the main ring? - no sure, can we + thread these in the underlying GL */ + DRM_IOCTL_DEF_DRV(VIRTGPU_TRANSFER_FROM_HOST, + virtio_gpu_transfer_from_host_ioctl, + DRM_AUTH|DRM_UNLOCKED), + DRM_IOCTL_DEF_DRV(VIRTGPU_TRANSFER_TO_HOST, + virtio_gpu_transfer_to_host_ioctl, + DRM_AUTH|DRM_UNLOCKED), + + DRM_IOCTL_DEF_DRV(VIRTGPU_WAIT, virtio_gpu_wait_ioctl, + DRM_AUTH|DRM_UNLOCKED), + + DRM_IOCTL_DEF_DRV(VIRTGPU_GET_CAPS, virtio_gpu_get_caps_ioctl, + DRM_AUTH|DRM_UNLOCKED), +}; diff --git a/drivers/gpu/drm/virtio/virtgpu_kms.c b/drivers/gpu/drm/virtio/virtgpu_kms.c index 782766c..06496a1 100644 --- a/drivers/gpu/drm/virtio/virtgpu_kms.c +++ b/drivers/gpu/drm/virtio/virtgpu_kms.c @@ -52,6 +52,41 @@ static void virtio_gpu_config_changed_work_func(struct work_struct *work) events_clear, &events_clear); }

+static void virtio_gpu_ctx_id_get(struct virtio_gpu_device *vgdev, + uint32_t *resid) +{ + int handle; + + idr_preload(GFP_KERNEL); + spin_lock(&vgdev->ctx_id_idr_lock); + handle = idr_alloc(&vgdev->ctx_id_idr, NULL, 1, 0, 0); + spin_unlock(&vgdev->ctx_id_idr_lock); + idr_preload_end(); + *resid = handle; +} + +static void virtio_gpu_ctx_id_put(struct virtio_gpu_device *vgdev, uint32_t id) +{ + spin_lock(&vgdev->ctx_id_idr_lock); + idr_remove(&vgdev->ctx_id_idr, id); + spin_unlock(&vgdev->ctx_id_idr_lock); +} + +static void virtio_gpu_context_create(struct virtio_gpu_device *vgdev, + uint32_t nlen, const char *name, + uint32_t *ctx_id) +{ + virtio_gpu_ctx_id_get(vgdev, ctx_id); + virtio_gpu_cmd_context_create(vgdev, *ctx_id, nlen, name); +} + +static void virtio_gpu_context_destroy(struct virtio_gpu_device *vgdev, + uint32_t ctx_id) +{ + virtio_gpu_cmd_context_destroy(vgdev, ctx_id); + virtio_gpu_ctx_id_put(vgdev, ctx_id); +} + static void virtio_gpu_init_vq(struct virtio_gpu_queue *vgvq, void (*work_func)(struct work_struct *work)) { @@ -60,6 +95,36 @@ static void virtio_gpu_init_vq(struct virtio_gpu_queue *vgvq, INIT_WORK(&vgvq->dequeue_work, work_func); }

+static void virtio_gpu_get_capsets(struct virtio_gpu_device *vgdev, + int num_capsets) +{ + int i, ret; + + vgdev->capsets = kcalloc(num_capsets, + sizeof(struct virtio_gpu_drv_capset), + GFP_KERNEL); + if (!vgdev->capsets) { + DRM_ERROR("failed to allocate cap sets\n"); + return; + } + for (i = 0; i < num_capsets; i++) { + virtio_gpu_cmd_get_capset_info(vgdev, i); + ret = wait_event_timeout(vgdev->resp_wq, + vgdev->capsets[i].id > 0, 5 * HZ); + if (ret == 0) { + DRM_ERROR("timed out waiting for cap set %d\n", i); + kfree(vgdev->capsets); + vgdev->capsets = NULL; + return; + } + DRM_INFO("cap set %d: id %d, max-version %d, max-size %d\n", + i, vgdev->capsets[i].id, + vgdev->capsets[i].max_version, + vgdev->capsets[i].max_size); + } + vgdev->num_capsets = num_capsets; +} + int virtio_gpu_driver_load(struct drm_device *dev, unsigned long flags) { static vq_callback_t *callbacks[] = { @@ -70,7 +135,7 @@ int virtio_gpu_driver_load(struct drm_device *dev, unsigned long flags) struct virtio_gpu_device *vgdev; /* this will expand later */ struct virtqueue *vqs[2]; - u32 num_scanouts; + u32 num_scanouts, num_capsets; int ret;

if (!virtio_has_feature(dev->virtdev, VIRTIO_F_VERSION_1)) @@ -96,9 +161,15 @@ int virtio_gpu_driver_load(struct drm_device *dev, unsigned long flags)

spin_lock_init(&vgdev->fence_drv.lock); INIT_LIST_HEAD(&vgdev->fence_drv.fences); + INIT_LIST_HEAD(&vgdev->cap_cache); INIT_WORK(&vgdev->config_changed_work, virtio_gpu_config_changed_work_func);

+ if (virtio_has_feature(vgdev->vdev, VIRTIO_GPU_F_VIRGL)) + vgdev->has_virgl_3d = true; + DRM_INFO("virgl 3d acceleration %s\n", + vgdev->has_virgl_3d ? "enabled" : "not available"); + ret = vgdev->vdev->config->find_vqs(vgdev->vdev, 2, vqs, callbacks, names); if (ret) { @@ -129,6 +200,11 @@ int virtio_gpu_driver_load(struct drm_device *dev, unsigned long flags) ret = -EINVAL; goto err_scanouts; } + DRM_INFO("number of scanouts: %d\n", num_scanouts); + + virtio_cread(vgdev->vdev, struct virtio_gpu_config, + num_capsets, &num_capsets); + DRM_INFO("number of cap sets: %d\n", num_capsets);

ret = virtio_gpu_modeset_init(vgdev); if (ret) @@ -137,6 +213,8 @@ int virtio_gpu_driver_load(struct drm_device *dev, unsigned long flags) virtio_device_ready(vgdev->vdev); vgdev->vqs_ready = true;

+ if (num_capsets) + virtio_gpu_get_capsets(vgdev, num_capsets); virtio_gpu_cmd_get_display_info(vgdev); wait_event_timeout(vgdev->resp_wq, !vgdev->display_info_pending, 5 * HZ); @@ -157,6 +235,16 @@ err_vqs: return ret; }

+static void virtio_gpu_cleanup_cap_cache(struct virtio_gpu_device *vgdev) +{ + struct virtio_gpu_drv_cap_cache *cache_ent, *tmp; + + list_for_each_entry_safe(cache_ent, tmp, &vgdev->cap_cache, head) { + kfree(cache_ent->caps_cache); + kfree(cache_ent); + } +} + int virtio_gpu_driver_unload(struct drm_device *dev) { struct virtio_gpu_device *vgdev = dev->dev_private; @@ -170,6 +258,49 @@ int virtio_gpu_driver_unload(struct drm_device *dev) virtio_gpu_modeset_fini(vgdev); virtio_gpu_ttm_fini(vgdev); virtio_gpu_free_vbufs(vgdev); + virtio_gpu_cleanup_cap_cache(vgdev); + kfree(vgdev->capsets); kfree(vgdev); return 0; } + +int virtio_gpu_driver_open(struct drm_device *dev, struct drm_file *file) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct virtio_gpu_fpriv *vfpriv; + uint32_t id; + char dbgname[64], tmpname[TASK_COMM_LEN]; + + /* can't create contexts without 3d renderer */ + if (!vgdev->has_virgl_3d) + return 0; + + get_task_comm(tmpname, current); + snprintf(dbgname, sizeof(dbgname), "%s", tmpname); + dbgname[63] = 0; + /* allocate a virt GPU context for this opener */ + vfpriv = kzalloc(sizeof(*vfpriv), GFP_KERNEL); + if (!vfpriv) + return -ENOMEM; + + virtio_gpu_context_create(vgdev, strlen(dbgname), dbgname, &id); + + vfpriv->ctx_id = id; + file->driver_priv = vfpriv; + return 0; +} + +void virtio_gpu_driver_postclose(struct drm_device *dev, struct drm_file *file) +{ + struct virtio_gpu_device *vgdev = dev->dev_private; + struct virtio_gpu_fpriv *vfpriv; + + if (!vgdev->has_virgl_3d) + return; + + vfpriv = file->driver_priv; + + virtio_gpu_context_destroy(vgdev, vfpriv->ctx_id); + kfree(vfpriv); + file->driver_priv = NULL; +} diff --git a/drivers/gpu/drm/virtio/virtgpu_ttm.c b/drivers/gpu/drm/virtio/virtgpu_ttm.c index b092d7b..9fd924c 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ttm.c +++ b/drivers/gpu/drm/virtio/virtgpu_ttm.c @@ -32,6 +32,7 @@ #include <ttm/ttm_module.h> #include <drm/drmP.h> #include <drm/drm.h> +#include <drm/virtgpu_drm.h> #include "virtgpu_drv.h"

#include <linux/delay.h> diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c index ee25e9a..5a0f8a7 100644 --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -586,6 +586,47 @@ static void virtio_gpu_cmd_get_display_info_cb(struct virtio_gpu_device *vgdev, drm_kms_helper_hotplug_event(vgdev->ddev); }

+static void virtio_gpu_cmd_get_capset_info_cb(struct virtio_gpu_device *vgdev, + struct virtio_gpu_vbuffer *vbuf) +{ + struct virtio_gpu_get_capset_info *cmd = + (struct virtio_gpu_get_capset_info *)vbuf->buf; + struct virtio_gpu_resp_capset_info *resp = + (struct virtio_gpu_resp_capset_info *)vbuf->resp_buf; + int i = le32_to_cpu(cmd->capset_index); + + spin_lock(&vgdev->display_info_lock); + vgdev->capsets[i].id = le32_to_cpu(resp->capset_id); + vgdev->capsets[i].max_version = le32_to_cpu(resp->capset_max_version); + vgdev->capsets[i].max_size = le32_to_cpu(resp->capset_max_size); + spin_unlock(&vgdev->display_info_lock); + wake_up(&vgdev->resp_wq); +} + +static void virtio_gpu_cmd_capset_cb(struct virtio_gpu_device *vgdev, + struct virtio_gpu_vbuffer *vbuf) +{ + struct virtio_gpu_get_capset *cmd = + (struct virtio_gpu_get_capset *)vbuf->buf; + struct virtio_gpu_resp_capset *resp = + (struct virtio_gpu_resp_capset *)vbuf->resp_buf; + struct virtio_gpu_drv_cap_cache *cache_ent; + + spin_lock(&vgdev->display_info_lock); + list_for_each_entry(cache_ent, &vgdev->cap_cache, head) { + if (cache_ent->version == le32_to_cpu(cmd->capset_version) && + cache_ent->id == le32_to_cpu(cmd->capset_id)) { + memcpy(cache_ent->caps_cache, resp->capset_data, + cache_ent->size); + atomic_set(&cache_ent->is_valid, 1); + break; + } + } + spin_unlock(&vgdev->display_info_lock); + wake_up(&vgdev->resp_wq); +} + + int virtio_gpu_cmd_get_display_info(struct virtio_gpu_device *vgdev) { struct virtio_gpu_ctrl_hdr *cmd_p; @@ -609,6 +650,230 @@ int virtio_gpu_cmd_get_display_info(struct virtio_gpu_device *vgdev) return 0; }

+int virtio_gpu_cmd_get_capset_info(struct virtio_gpu_device *vgdev, int idx) +{ + struct virtio_gpu_get_capset_info *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + void *resp_buf; + + resp_buf = kzalloc(sizeof(struct virtio_gpu_resp_capset_info), + GFP_KERNEL); + if (!resp_buf) + return -ENOMEM; + + cmd_p = virtio_gpu_alloc_cmd_resp + (vgdev, &virtio_gpu_cmd_get_capset_info_cb, &vbuf, + sizeof(*cmd_p), sizeof(struct virtio_gpu_resp_capset_info), + resp_buf); + memset(cmd_p, 0, sizeof(*cmd_p)); + + 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); + return 0; +} + +int virtio_gpu_cmd_get_capset(struct virtio_gpu_device *vgdev, + int idx, int version, + struct virtio_gpu_drv_cap_cache **cache_p) +{ + struct virtio_gpu_get_capset *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + int max_size = vgdev->capsets[idx].max_size; + struct virtio_gpu_drv_cap_cache *cache_ent; + void *resp_buf; + + if (idx > vgdev->num_capsets) + return -EINVAL; + + if (version > vgdev->capsets[idx].max_version) + return -EINVAL; + + cache_ent = kzalloc(sizeof(*cache_ent), GFP_KERNEL); + if (!cache_ent) + return -ENOMEM; + + cache_ent->caps_cache = kmalloc(max_size, GFP_KERNEL); + if (!cache_ent->caps_cache) { + kfree(cache_ent); + return -ENOMEM; + } + + resp_buf = kzalloc(sizeof(struct virtio_gpu_resp_capset) + max_size, + GFP_KERNEL); + if (!resp_buf) { + kfree(cache_ent->caps_cache); + kfree(cache_ent); + return -ENOMEM; + } + + cache_ent->version = version; + cache_ent->id = vgdev->capsets[idx].id; + atomic_set(&cache_ent->is_valid, 0); + cache_ent->size = max_size; + spin_lock(&vgdev->display_info_lock); + list_add_tail(&cache_ent->head, &vgdev->cap_cache); + spin_unlock(&vgdev->display_info_lock); + + cmd_p = virtio_gpu_alloc_cmd_resp + (vgdev, &virtio_gpu_cmd_capset_cb, &vbuf, sizeof(*cmd_p), + sizeof(struct virtio_gpu_resp_capset) + max_size, + resp_buf); + cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_GET_CAPSET); + 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); + + return 0; +} + +void virtio_gpu_cmd_context_create(struct virtio_gpu_device *vgdev, uint32_t id, + uint32_t nlen, const char *name) +{ + struct virtio_gpu_ctx_create *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + + cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p)); + memset(cmd_p, 0, sizeof(*cmd_p)); + + cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_CTX_CREATE); + cmd_p->hdr.ctx_id = cpu_to_le32(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); +} + +void virtio_gpu_cmd_context_destroy(struct virtio_gpu_device *vgdev, + uint32_t id) +{ + struct virtio_gpu_ctx_destroy *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + + cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p)); + memset(cmd_p, 0, sizeof(*cmd_p)); + + 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); +} + +void virtio_gpu_cmd_context_attach_resource(struct virtio_gpu_device *vgdev, + uint32_t ctx_id, + uint32_t resource_id) +{ + struct virtio_gpu_ctx_resource *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + + cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p)); + memset(cmd_p, 0, sizeof(*cmd_p)); + + 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(resource_id); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); + +} + +void virtio_gpu_cmd_context_detach_resource(struct virtio_gpu_device *vgdev, + uint32_t ctx_id, + uint32_t resource_id) +{ + struct virtio_gpu_ctx_resource *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + + cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p)); + memset(cmd_p, 0, sizeof(*cmd_p)); + + 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(resource_id); + virtio_gpu_queue_ctrl_buffer(vgdev, vbuf); +} + +void +virtio_gpu_cmd_resource_create_3d(struct virtio_gpu_device *vgdev, + struct virtio_gpu_resource_create_3d *rc_3d, + struct virtio_gpu_fence **fence) +{ + struct virtio_gpu_resource_create_3d *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + + cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p)); + memset(cmd_p, 0, sizeof(*cmd_p)); + + *cmd_p = *rc_3d; + cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_RESOURCE_CREATE_3D); + cmd_p->hdr.flags = 0; + + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, &cmd_p->hdr, fence); +} + +void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev, + uint32_t resource_id, uint32_t ctx_id, + uint64_t offset, uint32_t level, + struct virtio_gpu_box *box, + struct virtio_gpu_fence **fence) +{ + struct virtio_gpu_transfer_host_3d *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + + cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p)); + memset(cmd_p, 0, sizeof(*cmd_p)); + + cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_TRANSFER_TO_HOST_3D); + cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id); + cmd_p->resource_id = cpu_to_le32(resource_id); + cmd_p->box = *box; + cmd_p->offset = cpu_to_le64(offset); + cmd_p->level = cpu_to_le32(level); + + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, &cmd_p->hdr, fence); +} + +void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev, + uint32_t resource_id, uint32_t ctx_id, + uint64_t offset, uint32_t level, + struct virtio_gpu_box *box, + struct virtio_gpu_fence **fence) +{ + struct virtio_gpu_transfer_host_3d *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + + cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p)); + memset(cmd_p, 0, sizeof(*cmd_p)); + + cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_TRANSFER_FROM_HOST_3D); + cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id); + cmd_p->resource_id = cpu_to_le32(resource_id); + cmd_p->box = *box; + cmd_p->offset = cpu_to_le64(offset); + cmd_p->level = cpu_to_le32(level); + + virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, &cmd_p->hdr, fence); +} + +void virtio_gpu_cmd_submit(struct virtio_gpu_device *vgdev, + void *data, uint32_t data_size, + uint32_t ctx_id, struct virtio_gpu_fence **fence) +{ + struct virtio_gpu_cmd_submit *cmd_p; + struct virtio_gpu_vbuffer *vbuf; + + cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p)); + memset(cmd_p, 0, sizeof(*cmd_p)); + + vbuf->data_buf = data; + vbuf->data_size = data_size; + + cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_SUBMIT_3D); + 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, &cmd_p->hdr, fence); +} + int virtio_gpu_object_attach(struct virtio_gpu_device *vgdev, struct virtio_gpu_object *obj, uint32_t resource_id, diff --git a/include/uapi/drm/Kbuild b/include/uapi/drm/Kbuild index 2d9a25d..38d4370 100644 --- a/include/uapi/drm/Kbuild +++ b/include/uapi/drm/Kbuild @@ -17,3 +17,4 @@ header-y += tegra_drm.h header-y += via_drm.h header-y += vmwgfx_drm.h header-y += msm_drm.h +header-y += virtgpu_drm.h diff --git a/include/uapi/drm/virtgpu_drm.h b/include/uapi/drm/virtgpu_drm.h new file mode 100644 index 0000000..fc9e2d6 --- /dev/null +++ b/include/uapi/drm/virtgpu_drm.h @@ -0,0 +1,167 @@ +/* + * Copyright 2013 Red Hat + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#ifndef VIRTGPU_DRM_H +#define VIRTGPU_DRM_H + +#include <stddef.h> +#include "drm/drm.h" + +/* Please note that modifications to all structs defined here are + * subject to backwards-compatibility constraints. + * + * Do not use pointers, use uint64_t instead for 32 bit / 64 bit user/kernel + * compatibility Keep fields aligned to their size + */ + +#define DRM_VIRTGPU_MAP 0x01 +#define DRM_VIRTGPU_EXECBUFFER 0x02 +#define DRM_VIRTGPU_GETPARAM 0x03 +#define DRM_VIRTGPU_RESOURCE_CREATE 0x04 +#define DRM_VIRTGPU_RESOURCE_INFO 0x05 +#define DRM_VIRTGPU_TRANSFER_FROM_HOST 0x06 +#define DRM_VIRTGPU_TRANSFER_TO_HOST 0x07 +#define DRM_VIRTGPU_WAIT 0x08 +#define DRM_VIRTGPU_GET_CAPS 0x09 + +struct drm_virtgpu_map { + uint64_t offset; /* use for mmap system call */ + uint32_t handle; + uint32_t pad; +}; + +struct drm_virtgpu_execbuffer { + uint32_t flags; /* for future use */ + uint32_t size; + uint64_t command; /* void* */ + uint64_t bo_handles; + uint32_t num_bo_handles; + uint32_t pad; +}; + +#define VIRTGPU_PARAM_3D_FEATURES 1 /* do we have 3D features in the hw */ + +struct drm_virtgpu_getparam { + uint64_t param; + uint64_t value; +}; + +/* NO_BO flags? NO resource flag? */ +/* resource flag for y_0_top */ +struct drm_virtgpu_resource_create { + uint32_t target; + uint32_t format; + uint32_t bind; + uint32_t width; + uint32_t height; + uint32_t depth; + uint32_t array_size; + uint32_t last_level; + uint32_t nr_samples; + uint32_t flags; + uint32_t bo_handle; /* if this is set - recreate a new resource attached to this bo ? */ + uint32_t res_handle; /* returned by kernel */ + uint32_t size; /* validate transfer in the host */ + uint32_t stride; /* validate transfer in the host */ +}; + +struct drm_virtgpu_resource_info { + uint32_t bo_handle; + uint32_t res_handle; + uint32_t size; + uint32_t stride; +}; + +struct drm_virtgpu_3d_box { + uint32_t x; + uint32_t y; + uint32_t z; + uint32_t w; + uint32_t h; + uint32_t d; +}; + +struct drm_virtgpu_3d_transfer_to_host { + uint32_t bo_handle; + struct drm_virtgpu_3d_box box; + uint32_t level; + uint32_t offset; +}; + +struct drm_virtgpu_3d_transfer_from_host { + uint32_t bo_handle; + struct drm_virtgpu_3d_box box; + uint32_t level; + uint32_t offset; +}; + +#define VIRTGPU_WAIT_NOWAIT 1 /* like it */ +struct drm_virtgpu_3d_wait { + uint32_t handle; /* 0 is an invalid handle */ + uint32_t flags; +}; + +struct drm_virtgpu_get_caps { + uint32_t cap_set_id; + uint32_t cap_set_ver; + uint64_t addr; + uint32_t size; + uint32_t pad; +}; + +#define DRM_IOCTL_VIRTGPU_MAP \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_VIRTGPU_MAP, struct drm_virtgpu_map) + +#define DRM_IOCTL_VIRTGPU_EXECBUFFER \ + DRM_IOW(DRM_COMMAND_BASE + DRM_VIRTGPU_EXECBUFFER,\ + struct drm_virtgpu_execbuffer) + +#define DRM_IOCTL_VIRTGPU_GETPARAM \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_VIRTGPU_GETPARAM,\ + struct drm_virtgpu_getparam) + +#define DRM_IOCTL_VIRTGPU_RESOURCE_CREATE \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_VIRTGPU_RESOURCE_CREATE, \ + struct drm_virtgpu_resource_create) + +#define DRM_IOCTL_VIRTGPU_RESOURCE_INFO \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_VIRTGPU_RESOURCE_INFO, \ + struct drm_virtgpu_resource_info) + +#define DRM_IOCTL_VIRTGPU_TRANSFER_FROM_HOST \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_VIRTGPU_TRANSFER_FROM_HOST, \ + struct drm_virtgpu_3d_transfer_from_host) + +#define DRM_IOCTL_VIRTGPU_TRANSFER_TO_HOST \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_VIRTGPU_TRANSFER_TO_HOST, \ + struct drm_virtgpu_3d_transfer_to_host) + +#define DRM_IOCTL_VIRTGPU_WAIT \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_VIRTGPU_WAIT, \ + struct drm_virtgpu_3d_wait) + +#define DRM_IOCTL_VIRTGPU_GET_CAPS \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_VIRTGPU_GET_CAPS, \ + struct drm_virtgpu_get_caps) + +#endif diff --git a/include/uapi/linux/virtio_gpu.h b/include/uapi/linux/virtio_gpu.h index 478be52..7a63faa 100644 --- a/include/uapi/linux/virtio_gpu.h +++ b/include/uapi/linux/virtio_gpu.h @@ -40,6 +40,8 @@

#include <linux/types.h>

+#define VIRTIO_GPU_F_VIRGL 0 + enum virtio_gpu_ctrl_type { VIRTIO_GPU_UNDEFINED = 0,

@@ -52,6 +54,18 @@ enum virtio_gpu_ctrl_type { VIRTIO_GPU_CMD_TRANSFER_TO_HOST_2D, VIRTIO_GPU_CMD_RESOURCE_ATTACH_BACKING, VIRTIO_GPU_CMD_RESOURCE_DETACH_BACKING, + VIRTIO_GPU_CMD_GET_CAPSET_INFO, + VIRTIO_GPU_CMD_GET_CAPSET, + + /* 3d commands */ + VIRTIO_GPU_CMD_CTX_CREATE = 0x0200, + VIRTIO_GPU_CMD_CTX_DESTROY, + VIRTIO_GPU_CMD_CTX_ATTACH_RESOURCE, + VIRTIO_GPU_CMD_CTX_DETACH_RESOURCE, + VIRTIO_GPU_CMD_RESOURCE_CREATE_3D, + VIRTIO_GPU_CMD_TRANSFER_TO_HOST_3D, + VIRTIO_GPU_CMD_TRANSFER_FROM_HOST_3D, + VIRTIO_GPU_CMD_SUBMIT_3D,

/* cursor commands */ VIRTIO_GPU_CMD_UPDATE_CURSOR = 0x0300, @@ -60,6 +74,8 @@ enum virtio_gpu_ctrl_type { /* success responses */ VIRTIO_GPU_RESP_OK_NODATA = 0x1100, VIRTIO_GPU_RESP_OK_DISPLAY_INFO, + VIRTIO_GPU_RESP_OK_CAPSET_INFO, + VIRTIO_GPU_RESP_OK_CAPSET,

/* error responses */ VIRTIO_GPU_RESP_ERR_UNSPEC = 0x1200, @@ -180,13 +196,107 @@ struct virtio_gpu_resp_display_info { } pmodes[VIRTIO_GPU_MAX_SCANOUTS]; };

+/* data passed in the control vq, 3d related */ + +struct virtio_gpu_box { + __le32 x, y, z; + __le32 w, h, d; +}; + +/* VIRTIO_GPU_CMD_TRANSFER_TO_HOST_3D, VIRTIO_GPU_CMD_TRANSFER_FROM_HOST_3D */ +struct virtio_gpu_transfer_host_3d { + struct virtio_gpu_ctrl_hdr hdr; + struct virtio_gpu_box box; + __le64 offset; + __le32 resource_id; + __le32 level; + __le32 stride; + __le32 layer_stride; +}; + +/* VIRTIO_GPU_CMD_RESOURCE_CREATE_3D */ +#define VIRTIO_GPU_RESOURCE_FLAG_Y_0_TOP (1 << 0) +struct virtio_gpu_resource_create_3d { + struct virtio_gpu_ctrl_hdr hdr; + __le32 resource_id; + __le32 target; + __le32 format; + __le32 bind; + __le32 width; + __le32 height; + __le32 depth; + __le32 array_size; + __le32 last_level; + __le32 nr_samples; + __le32 flags; + __le32 padding; +}; + +/* VIRTIO_GPU_CMD_CTX_CREATE */ +struct virtio_gpu_ctx_create { + struct virtio_gpu_ctrl_hdr hdr; + __le32 nlen; + __le32 padding; + char debug_name[64]; +}; + +/* VIRTIO_GPU_CMD_CTX_DESTROY */ +struct virtio_gpu_ctx_destroy { + struct virtio_gpu_ctrl_hdr hdr; +}; + +/* VIRTIO_GPU_CMD_CTX_ATTACH_RESOURCE, VIRTIO_GPU_CMD_CTX_DETACH_RESOURCE */ +struct virtio_gpu_ctx_resource { + struct virtio_gpu_ctrl_hdr hdr; + __le32 resource_id; + __le32 padding; +}; + +/* VIRTIO_GPU_CMD_SUBMIT_3D */ +struct virtio_gpu_cmd_submit { + struct virtio_gpu_ctrl_hdr hdr; + __le32 size; + __le32 padding; +}; + +#define VIRTIO_GPU_CAPSET_VIRGL 1 + +/* VIRTIO_GPU_CMD_GET_CAPSET_INFO */ +struct virtio_gpu_get_capset_info { + struct virtio_gpu_ctrl_hdr hdr; + __le32 capset_index; + __le32 padding; +}; + +/* VIRTIO_GPU_RESP_OK_CAPSET_INFO */ +struct virtio_gpu_resp_capset_info { + struct virtio_gpu_ctrl_hdr hdr; + __le32 capset_id; + __le32 capset_max_version; + __le32 capset_max_size; + __le32 padding; +}; + +/* VIRTIO_GPU_CMD_GET_CAPSET */ +struct virtio_gpu_get_capset { + struct virtio_gpu_ctrl_hdr hdr; + __le32 capset_id; + __le32 capset_version; +}; + +/* VIRTIO_GPU_RESP_OK_CAPSET */ +struct virtio_gpu_resp_capset { + struct virtio_gpu_ctrl_hdr hdr; + uint8_t capset_data[]; +}; + #define VIRTIO_GPU_EVENT_DISPLAY (1 << 0)

struct virtio_gpu_config { __u32 events_read; __u32 events_clear; __u32 num_scanouts; - __u32 reserved; + __u32 num_capsets; };

/* simple formats for fbcon/X use */

-- 1.8.3.1

Gerd Hoffmann

11:58 a.m.

New subject: [PATCH v3 5/7] virtio-gpu: add basic prime support

From: Dave Airlie airlied@redhat.com

This is enough to enable DRI3.

Signed-off-by: Dave Airlie airlied@redhat.com Signed-off-by: Gerd Hoffmann kraxel@redhat.com --- drivers/gpu/drm/virtio/Makefile | 2 +- drivers/gpu/drm/virtio/virtgpu_drv.c | 13 ++++++- drivers/gpu/drm/virtio/virtgpu_drv.h | 12 ++++++ drivers/gpu/drm/virtio/virtgpu_prime.c | 71 ++++++++++++++++++++++++++++++++++ 4 files changed, 96 insertions(+), 2 deletions(-) create mode 100644 drivers/gpu/drm/virtio/virtgpu_prime.c

diff --git a/drivers/gpu/drm/virtio/Makefile b/drivers/gpu/drm/virtio/Makefile index da7bf19..3fb8eac 100644 --- a/drivers/gpu/drm/virtio/Makefile +++ b/drivers/gpu/drm/virtio/Makefile @@ -7,6 +7,6 @@ ccflags-y := -Iinclude/drm virtio-gpu-y := virtgpu_drv.o virtgpu_kms.o virtgpu_drm_bus.o virtgpu_gem.o \ virtgpu_fb.o virtgpu_display.o virtgpu_vq.o virtgpu_ttm.o \ virtgpu_fence.o virtgpu_object.o virtgpu_debugfs.o virtgpu_plane.o \ - virtgpu_ioctl.o + virtgpu_ioctl.o virtgpu_prime.o

obj-$(CONFIG_DRM_VIRTIO_GPU) += virtio-gpu.o diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.c b/drivers/gpu/drm/virtio/virtgpu_drv.c index c27d3a3..1f15ef8 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.c +++ b/drivers/gpu/drm/virtio/virtgpu_drv.c @@ -118,7 +118,7 @@ static const struct file_operations virtio_gpu_driver_fops = {

static struct drm_driver driver = { - .driver_features = DRIVER_MODESET | DRIVER_GEM, + .driver_features = DRIVER_MODESET | DRIVER_GEM | DRIVER_PRIME, .set_busid = drm_virtio_set_busid, .load = virtio_gpu_driver_load, .unload = virtio_gpu_driver_unload, @@ -133,6 +133,17 @@ static struct drm_driver driver = { .debugfs_init = virtio_gpu_debugfs_init, .debugfs_cleanup = virtio_gpu_debugfs_takedown, #endif + .prime_handle_to_fd = drm_gem_prime_handle_to_fd, + .prime_fd_to_handle = drm_gem_prime_fd_to_handle, + .gem_prime_export = drm_gem_prime_export, + .gem_prime_import = drm_gem_prime_import, + .gem_prime_pin = virtgpu_gem_prime_pin, + .gem_prime_unpin = virtgpu_gem_prime_unpin, + .gem_prime_get_sg_table = virtgpu_gem_prime_get_sg_table, + .gem_prime_import_sg_table = virtgpu_gem_prime_import_sg_table, + .gem_prime_vmap = virtgpu_gem_prime_vmap, + .gem_prime_vunmap = virtgpu_gem_prime_vunmap, + .gem_prime_mmap = virtgpu_gem_prime_mmap,

.gem_free_object = virtio_gpu_gem_free_object, .gem_open_object = virtio_gpu_gem_object_open, diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.h b/drivers/gpu/drm/virtio/virtgpu_drv.h index 2719108..79f0abe 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.h +++ b/drivers/gpu/drm/virtio/virtgpu_drv.h @@ -359,6 +359,18 @@ int virtio_gpu_object_get_sg_table(struct virtio_gpu_device *qdev, void virtio_gpu_object_free_sg_table(struct virtio_gpu_object *bo); int virtio_gpu_object_wait(struct virtio_gpu_object *bo, bool no_wait);

+/* virtgpu_prime.c */ +int virtgpu_gem_prime_pin(struct drm_gem_object *obj); +void virtgpu_gem_prime_unpin(struct drm_gem_object *obj); +struct sg_table *virtgpu_gem_prime_get_sg_table(struct drm_gem_object *obj); +struct drm_gem_object *virtgpu_gem_prime_import_sg_table( + struct drm_device *dev, struct dma_buf_attachment *attach, + struct sg_table *sgt); +void *virtgpu_gem_prime_vmap(struct drm_gem_object *obj); +void virtgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr); +int virtgpu_gem_prime_mmap(struct drm_gem_object *obj, + struct vm_area_struct *vma); + static inline struct virtio_gpu_object* virtio_gpu_object_ref(struct virtio_gpu_object *bo) { diff --git a/drivers/gpu/drm/virtio/virtgpu_prime.c b/drivers/gpu/drm/virtio/virtgpu_prime.c new file mode 100644 index 0000000..385e0eb --- /dev/null +++ b/drivers/gpu/drm/virtio/virtgpu_prime.c @@ -0,0 +1,71 @@ +/* + * Copyright 2014 Canonical + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Andreas Pokorny + */ + +#include "virtgpu_drv.h" + +/* Empty Implementations as there should not be any other driver for a virtual + * device that might share buffers with virtgpu */ + +int virtgpu_gem_prime_pin(struct drm_gem_object *obj) +{ + WARN_ONCE(1, "not implemented"); + return -ENODEV; +} + +void virtgpu_gem_prime_unpin(struct drm_gem_object *obj) +{ + WARN_ONCE(1, "not implemented"); +} + + +struct sg_table *virtgpu_gem_prime_get_sg_table(struct drm_gem_object *obj) +{ + WARN_ONCE(1, "not implemented"); + return ERR_PTR(-ENODEV); +} + +struct drm_gem_object *virtgpu_gem_prime_import_sg_table( + struct drm_device *dev, struct dma_buf_attachment *attach, + struct sg_table *table) +{ + WARN_ONCE(1, "not implemented"); + return ERR_PTR(-ENODEV); +} + +void *virtgpu_gem_prime_vmap(struct drm_gem_object *obj) +{ + WARN_ONCE(1, "not implemented"); + return ERR_PTR(-ENODEV); +} + +void virtgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr) +{ + WARN_ONCE(1, "not implemented"); +} + +int virtgpu_gem_prime_mmap(struct drm_gem_object *obj, + struct vm_area_struct *area) +{ + return -ENODEV; +}

-- 1.8.3.1

Gerd Hoffmann

11:58 a.m.

New subject: [PATCH v3 6/7] virtio-gpu: mark as a render gpu

From: Dave Airlie airlied@redhat.com

Also add DRM_RENDER_ALLOW to the ioctls.

Signed-off-by: Gerd Hoffmann kraxel@redhat.com --- drivers/gpu/drm/virtio/virtgpu_drv.c | 2 +- drivers/gpu/drm/virtio/virtgpu_ioctl.c | 18 +++++++++--------- 2 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.c b/drivers/gpu/drm/virtio/virtgpu_drv.c index 1f15ef8..b40ed60 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.c +++ b/drivers/gpu/drm/virtio/virtgpu_drv.c @@ -118,7 +118,7 @@ static const struct file_operations virtio_gpu_driver_fops = {

static struct drm_driver driver = { - .driver_features = DRIVER_MODESET | DRIVER_GEM | DRIVER_PRIME, + .driver_features = DRIVER_MODESET | DRIVER_GEM | DRIVER_PRIME | DRIVER_RENDER, .set_busid = drm_virtio_set_busid, .load = virtio_gpu_driver_load, .unload = virtio_gpu_driver_unload, diff --git a/drivers/gpu/drm/virtio/virtgpu_ioctl.c b/drivers/gpu/drm/virtio/virtgpu_ioctl.c index d5c9e02..d297f38 100644 --- a/drivers/gpu/drm/virtio/virtgpu_ioctl.c +++ b/drivers/gpu/drm/virtio/virtgpu_ioctl.c @@ -540,33 +540,33 @@ copy_exit:

struct drm_ioctl_desc virtio_gpu_ioctls[DRM_VIRTIO_NUM_IOCTLS] = { DRM_IOCTL_DEF_DRV(VIRTGPU_MAP, virtio_gpu_map_ioctl, - DRM_AUTH|DRM_UNLOCKED), + DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW),

DRM_IOCTL_DEF_DRV(VIRTGPU_EXECBUFFER, virtio_gpu_execbuffer_ioctl, - DRM_AUTH|DRM_UNLOCKED), + DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW),

DRM_IOCTL_DEF_DRV(VIRTGPU_GETPARAM, virtio_gpu_getparam_ioctl, - DRM_AUTH|DRM_UNLOCKED), + DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW),

DRM_IOCTL_DEF_DRV(VIRTGPU_RESOURCE_CREATE, virtio_gpu_resource_create_ioctl, - DRM_AUTH|DRM_UNLOCKED), + DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW),

DRM_IOCTL_DEF_DRV(VIRTGPU_RESOURCE_INFO, virtio_gpu_resource_info_ioctl, - DRM_AUTH|DRM_UNLOCKED), + DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW),

/* make transfer async to the main ring? - no sure, can we thread these in the underlying GL */ DRM_IOCTL_DEF_DRV(VIRTGPU_TRANSFER_FROM_HOST, virtio_gpu_transfer_from_host_ioctl, - DRM_AUTH|DRM_UNLOCKED), + DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(VIRTGPU_TRANSFER_TO_HOST, virtio_gpu_transfer_to_host_ioctl, - DRM_AUTH|DRM_UNLOCKED), + DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW),

DRM_IOCTL_DEF_DRV(VIRTGPU_WAIT, virtio_gpu_wait_ioctl, - DRM_AUTH|DRM_UNLOCKED), + DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW),

DRM_IOCTL_DEF_DRV(VIRTGPU_GET_CAPS, virtio_gpu_get_caps_ioctl, - DRM_AUTH|DRM_UNLOCKED), + DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW), };

-- 1.8.3.1

Gerd Hoffmann

11:58 a.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

Signed-off-by: Gerd Hoffmann kraxel@redhat.com --- drivers/gpu/drm/virtio/virtgpu_display.c | 49 ++++++++++++++++++++++++++++++-- 1 file changed, 46 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/virtio/virtgpu_display.c b/drivers/gpu/drm/virtio/virtgpu_display.c index c9c1427..f545913 100644 --- a/drivers/gpu/drm/virtio/virtgpu_display.c +++ b/drivers/gpu/drm/virtio/virtgpu_display.c @@ -125,6 +125,51 @@ static int virtio_gpu_crtc_cursor_move(struct drm_crtc *crtc, return 0; }

+static int virtio_gpu_page_flip(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_pending_vblank_event *event, + uint32_t flags) +{ + struct virtio_gpu_device *vgdev = crtc->dev->dev_private; + struct virtio_gpu_output *output = + container_of(crtc, struct virtio_gpu_output, crtc); + struct drm_plane *plane = crtc->primary; + struct virtio_gpu_framebuffer *vgfb; + struct virtio_gpu_object *bo; + unsigned long irqflags; + uint32_t handle; + + plane->fb = fb; + vgfb = to_virtio_gpu_framebuffer(plane->fb); + bo = gem_to_virtio_gpu_obj(vgfb->obj); + handle = bo->hw_res_handle; + + DRM_DEBUG("handle 0x%x%s, crtc %dx%d\n", handle, + bo->dumb ? ", dumb" : "", + crtc->mode.hdisplay, crtc->mode.vdisplay); + if (bo->dumb) { + virtio_gpu_cmd_transfer_to_host_2d + (vgdev, handle, 0, + cpu_to_le32(crtc->mode.hdisplay), + cpu_to_le32(crtc->mode.vdisplay), + 0, 0, NULL); + } + virtio_gpu_cmd_set_scanout(vgdev, output->index, handle, + crtc->mode.hdisplay, + crtc->mode.vdisplay, 0, 0); + virtio_gpu_cmd_resource_flush(vgdev, handle, 0, 0, + crtc->mode.hdisplay, + crtc->mode.vdisplay); + + if (event) { + spin_lock_irqsave(&crtc->dev->event_lock, irqflags); + drm_send_vblank_event(crtc->dev, -1, event); + spin_unlock_irqrestore(&crtc->dev->event_lock, irqflags); + } + + return 0; +} + static const struct drm_crtc_funcs virtio_gpu_crtc_funcs = { .cursor_set2 = virtio_gpu_crtc_cursor_set, .cursor_move = virtio_gpu_crtc_cursor_move, @@ -132,9 +177,7 @@ static const struct drm_crtc_funcs virtio_gpu_crtc_funcs = { .set_config = drm_atomic_helper_set_config, .destroy = drm_crtc_cleanup,

-#if 0 /* not (yet) working without vblank support according to docs */ - .page_flip = drm_atomic_helper_page_flip, -#endif + .page_flip = virtio_gpu_page_flip, .reset = drm_atomic_helper_crtc_reset, .atomic_duplicate_state = drm_atomic_helper_crtc_duplicate_state, .atomic_destroy_state = drm_atomic_helper_crtc_destroy_state,

-- 1.8.3.1

Daniel Vetter

25 May 25 May

4:37 p.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

On Fri, Oct 2, 2015 at 1:58 PM, Gerd Hoffmann kraxel@redhat.com wrote:

...

Signed-off-by: Gerd Hoffmann kraxel@redhat.com

So I entirely missed this, but this isn't really how to implement page_flip for an atomic driver. Working on some stuff and will hack up a likely totally broken patch, but should be enough as guideline. -Daniel

...

drivers/gpu/drm/virtio/virtgpu_display.c | 49 ++++++++++++++++++++++++++++++-- 1 file changed, 46 insertions(+), 3 deletions(-)

+static int virtio_gpu_page_flip(struct drm_crtc *crtc,

                          struct drm_framebuffer *fb,

                          struct drm_pending_vblank_event *event,

                          uint32_t flags)

  struct virtio_gpu_device *vgdev = crtc->dev->dev_private;

```
  struct virtio_gpu_output *output =
```

          container_of(crtc, struct virtio_gpu_output, crtc);

  struct drm_plane *plane = crtc->primary;

```
  struct virtio_gpu_framebuffer *vgfb;
```
```
  struct virtio_gpu_object *bo;
```
```
  unsigned long irqflags;
```
```
  uint32_t handle;
```
```
  plane->fb = fb;
```

  vgfb = to_virtio_gpu_framebuffer(plane->fb);

  bo = gem_to_virtio_gpu_obj(vgfb->obj);

```
  handle = bo->hw_res_handle;
```

  DRM_DEBUG("handle 0x%x%s, crtc %dx%d\n", handle,

```
            bo->dumb ? ", dumb" : "",
```

            crtc->mode.hdisplay, crtc->mode.vdisplay);

```
  if (bo->dumb) {
```

          virtio_gpu_cmd_transfer_to_host_2d

```
                  (vgdev, handle, 0,
```

                   cpu_to_le32(crtc->mode.hdisplay),

                   cpu_to_le32(crtc->mode.vdisplay),

```
                   0, 0, NULL);
```
```
  }
```

  virtio_gpu_cmd_set_scanout(vgdev, output->index, handle,

                             crtc->mode.hdisplay,

                             crtc->mode.vdisplay, 0, 0);

  virtio_gpu_cmd_resource_flush(vgdev, handle, 0, 0,

                                crtc->mode.hdisplay,

                                crtc->mode.vdisplay);

```
  if (event) {
```

          spin_lock_irqsave(&crtc->dev->event_lock, irqflags);

          drm_send_vblank_event(crtc->dev, -1, event);

          spin_unlock_irqrestore(&crtc->dev->event_lock, irqflags);

```
  }
```
```
  return 0;
```

static const struct drm_crtc_funcs virtio_gpu_crtc_funcs = { .cursor_set2 = virtio_gpu_crtc_cursor_set, .cursor_move = virtio_gpu_crtc_cursor_move, @@ -132,9 +177,7 @@ static const struct drm_crtc_funcs virtio_gpu_crtc_funcs = { .set_config = drm_atomic_helper_set_config, .destroy = drm_crtc_cleanup,

-#if 0 /* not (yet) working without vblank support according to docs */

  .page_flip              = drm_atomic_helper_page_flip,

-#endif

  .page_flip              = virtio_gpu_page_flip,
  .reset                  = drm_atomic_helper_crtc_reset,
  .atomic_duplicate_state = drm_atomic_helper_crtc_duplicate_state,
  .atomic_destroy_state   = drm_atomic_helper_crtc_destroy_state,

-- 1.8.3.1

dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel

-- Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch

Gerd Hoffmann

27 May 27 May

7:46 a.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

On Mi, 2016-05-25 at 18:37 +0200, Daniel Vetter wrote:

...

On Fri, Oct 2, 2015 at 1:58 PM, Gerd Hoffmann kraxel@redhat.com wrote:

...
Signed-off-by: Gerd Hoffmann kraxel@redhat.com

So I entirely missed this, but this isn't really how to implement page_flip for an atomic driver. Working on some stuff and will hack up a likely totally broken patch, but should be enough as guideline. -Daniel

Hmm, no patch in my inbox yet. Care to send it over or give some hints what is wrong with this?

thanks, Gerd

Daniel Vetter

7:50 a.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

On Fri, May 27, 2016 at 09:46:03AM +0200, Gerd Hoffmann wrote:

...

On Mi, 2016-05-25 at 18:37 +0200, Daniel Vetter wrote:

...
On Fri, Oct 2, 2015 at 1:58 PM, Gerd Hoffmann kraxel@redhat.com wrote:

...
Signed-off-by: Gerd Hoffmann kraxel@redhat.com

So I entirely missed this, but this isn't really how to implement page_flip for an atomic driver. Working on some stuff and will hack up a likely totally broken patch, but should be enough as guideline. -Daniel

Hmm, no patch in my inbox yet. Care to send it over or give some hints what is wrong with this?

You should use drm_atomic_helper_page_flip as .page_flip hook instead of rolling your own, when you have a brand-new atomic driver. If that helper doesn't work that means you have a bug in your driver somewhere.

And indeed virtio just plugs in drm_atomic_helper_commit, which thus far doesnt' do nonblocking commits, which means the page flip stuff won't work. I'm working on generic nonblocking atomic to fix this up. Currently still wip and buggy though, hence no patches.

But I'll take you up on the implied offer to help out and test ;-) -Daniel

-- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch

Daniel Vetter

30 May 30 May

8:42 a.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

On Fri, May 27, 2016 at 09:50:27AM +0200, Daniel Vetter wrote:

...

On Fri, May 27, 2016 at 09:46:03AM +0200, Gerd Hoffmann wrote:

...
On Mi, 2016-05-25 at 18:37 +0200, Daniel Vetter wrote:

...
On Fri, Oct 2, 2015 at 1:58 PM, Gerd Hoffmann kraxel@redhat.com wrote:

...
Signed-off-by: Gerd Hoffmann kraxel@redhat.com

So I entirely missed this, but this isn't really how to implement page_flip for an atomic driver. Working on some stuff and will hack up a likely totally broken patch, but should be enough as guideline. -Daniel

Hmm, no patch in my inbox yet. Care to send it over or give some hints what is wrong with this?

You should use drm_atomic_helper_page_flip as .page_flip hook instead of rolling your own, when you have a brand-new atomic driver. If that helper doesn't work that means you have a bug in your driver somewhere.

And indeed virtio just plugs in drm_atomic_helper_commit, which thus far doesnt' do nonblocking commits, which means the page flip stuff won't work. I'm working on generic nonblocking atomic to fix this up. Currently still wip and buggy though, hence no patches.

But I'll take you up on the implied offer to help out and test ;-)

Ok, patches are now on the list and also in my fdo repo at:

https://cgit.freedesktop.org/~danvet/drm/log/

git://people.freedesktop.org/~danvet/drm stuff

Would be really awesome if you could test this on virtio. Note that the new nonblocking helpers require that your atomic backend gets the drm event handling right. So if there's a bug in that logic then you'll see lots of dmesg noise about waits timing out (after 10s of waiting). From a quick inspection it should work though.

Upshot is that once you get normal modeset to work with those new helpers, then nonblocking should Just Work, and there's no need anymore for additional testing. Or at least shouldn't. -Daniel

-- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch

Gerd Hoffmann

12:06 p.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

Hi,

...

...
But I'll take you up on the implied offer to help out and test ;-)

git://people.freedesktop.org/~danvet/drm stuff

Tried that branch.

...

Would be really awesome if you could test this on virtio. Note that the new nonblocking helpers require that your atomic backend gets the drm event handling right. So if there's a bug in that logic then you'll see lots of dmesg noise about waits timing out (after 10s of waiting). From a quick inspection it should work though.

No timeouts. Yay!

But it seems crtcs can be (temporarely) disabled now, so we might have to pick up the crtc from old_state in virtio_gpu_plane_atomic_update to figure which virtual output needs to be turned off. Ran into this last week already. Happened with multihead setups only, but the same patch fixes this one too ;)

https://lists.freedesktop.org/archives/dri-devel/2016-May/108772.html

cheers, Gerd

Daniel Vetter

2:43 p.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

On Mon, May 30, 2016 at 02:06:50PM +0200, Gerd Hoffmann wrote:

...

Hi,

...
...
But I'll take you up on the implied offer to help out and test ;-)

git://people.freedesktop.org/~danvet/drm stuff

Tried that branch.

...
Would be really awesome if you could test this on virtio. Note that the new nonblocking helpers require that your atomic backend gets the drm event handling right. So if there's a bug in that logic then you'll see lots of dmesg noise about waits timing out (after 10s of waiting). From a quick inspection it should work though.

No timeouts. Yay!

But it seems crtcs can be (temporarely) disabled now, so we might have to pick up the crtc from old_state in virtio_gpu_plane_atomic_update to figure which virtual output needs to be turned off. Ran into this last week already. Happened with multihead setups only, but the same patch fixes this one too ;)

https://lists.freedesktop.org/archives/dri-devel/2016-May/108772.html

Hm, smells more like virtio isn't too happy with the default ordering of the commit operation. The default is:

- Disable any crtc/encoders that need to be disabled/change. - Bash new plane setup into hw. - Enable all crtcs/encoders that need to be enabled/have changed.

There's two problems: - some hw gets real grumpy if you bash in plane state without the crtc state yet matching. - if you do runtime pm nothing is enabled and the writes get lost at best, or hang your interconnect at worst.

That's why you can overwrite atomic_commit_tail, and use something more sensible. See for example what it looks like for rockchip. I have a gut feeling that should also take care of your troubles. Using the old crtc is definitely not what you want.

Another option is that virtio isn't happy about bashing in plane state for disabled crtc. Again helpers have you covered, look at the active_only parameter for drm_atomic_helper_commit_planes().

Aside, if you wonder why these defaults: They match what the crtc helpers are doing, but they are a bit surprising indeed.

Cheers, Daniel

-- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch

Gerd Hoffmann

31 May 31 May

6:18 a.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

Hi,

...

...
https://lists.freedesktop.org/archives/dri-devel/2016-May/108772.html

Hm, smells more like virtio isn't too happy with the default ordering of the commit operation. The default is:

Disable any crtc/encoders that need to be disabled/change.

Bash new plane setup into hw.

Enable all crtcs/encoders that need to be enabled/have changed.

There's two problems:

some hw gets real grumpy if you bash in plane state without the crtc state yet matching.

if you do runtime pm nothing is enabled and the writes get lost at best, or hang your interconnect at worst.

That's why you can overwrite atomic_commit_tail, and use something more sensible. See for example what it looks like for rockchip. I have a gut feeling that should also take care of your troubles. Using the old crtc is definitely not what you want.

...

Another option is that virtio isn't happy about bashing in plane state for disabled crtc. Again helpers have you covered, look at the active_only parameter for drm_atomic_helper_commit_planes().

virtio-gpu is a bit simplified compared to real hardware, so there isn't really separate plane/crtc state.

Right now the virtual outputs are linked to drm_crtc. To apply any changes I need to lookup the crtc to figure which virtual output should be updated.

So, setting active_only should make sure I have a valid crtc pointer on plane updates, right? It probably also skips the disable + enable crtc steps on commit? What happens when outputs are disabled?

Maybe it makes sense to link our virtual outputs to (primary) planes not crtcs?

cheers, Gerd

Daniel Vetter

6:37 a.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

On Tue, May 31, 2016 at 8:18 AM, Gerd Hoffmann kraxel@redhat.com wrote:

...

...
...
https://lists.freedesktop.org/archives/dri-devel/2016-May/108772.html

Hm, smells more like virtio isn't too happy with the default ordering of the commit operation. The default is:

Disable any crtc/encoders that need to be disabled/change.

Bash new plane setup into hw.

Enable all crtcs/encoders that need to be enabled/have changed.

There's two problems:

some hw gets real grumpy if you bash in plane state without the crtc state yet matching.

if you do runtime pm nothing is enabled and the writes get lost at best, or hang your interconnect at worst.

That's why you can overwrite atomic_commit_tail, and use something more sensible. See for example what it looks like for rockchip. I have a gut feeling that should also take care of your troubles. Using the old crtc is definitely not what you want.

...
Another option is that virtio isn't happy about bashing in plane state for disabled crtc. Again helpers have you covered, look at the active_only parameter for drm_atomic_helper_commit_planes().

virtio-gpu is a bit simplified compared to real hardware, so there isn't really separate plane/crtc state.

Right now the virtual outputs are linked to drm_crtc. To apply any changes I need to lookup the crtc to figure which virtual output should be updated.

So, setting active_only should make sure I have a valid crtc pointer on plane updates, right? It probably also skips the disable + enable crtc steps on commit? What happens when outputs are disabled?

Maybe it makes sense to link our virtual outputs to (primary) planes not crtcs?

Nah, I just misunderstood your patch. If it's all about finding the corresponding crtc, then you're all good. I thought there was some other reason (like the virtual hw getting upset about certain things). I'll pick up your patch into my nonblocking atomic branch to make sure virtio isn't accidentally broken. btw can you pls drop an ack or r-b onto my virtio conversion? I already added your tested-by.

Thanks, Daniel

-- Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch

Gerd Hoffmann

7:34 a.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

Hi,

...

...
Right now the virtual outputs are linked to drm_crtc. To apply any changes I need to lookup the crtc to figure which virtual output should be updated.

...

...
So, setting active_only should make sure I have a valid crtc pointer on plane updates, right? It probably also skips the disable + enable crtc steps on commit? What happens when outputs are disabled?

...

Nah, I just misunderstood your patch. If it's all about finding the corresponding crtc, then you're all good.

Yes, it's all about finding the crtc.

...

I thought there was some other reason (like the virtual hw getting upset about certain things).

virtio wouldn't be upset.

It's a pointless exercise though to first disable the output, just to re-enable it the next moment with the new page-flipped framebuffer. So I guess I should look at the active_only thing nevertheless.

...

btw can you pls drop an ack or r-b onto my virtio conversion? I already added your tested-by.

Grr, mail is not in my dri-devel folder. Guess that is the <censored> "avoid-duplicates" mailman option at work.

Feel free to just add the r-b too. Or I'll send it for the next version of the series.

cheers, Gerd

Daniel Vetter

7:39 a.m.

New subject: [PATCH v3 7/7] [wip] virtio-gpu: add page flip support

On Tue, May 31, 2016 at 9:34 AM, Gerd Hoffmann kraxel@redhat.com wrote:

...

Hi,

...
...
Right now the virtual outputs are linked to drm_crtc. To apply any changes I need to lookup the crtc to figure which virtual output should be updated.

...
...
So, setting active_only should make sure I have a valid crtc pointer on plane updates, right? It probably also skips the disable + enable crtc steps on commit? What happens when outputs are disabled?

...
Nah, I just misunderstood your patch. If it's all about finding the corresponding crtc, then you're all good.

Yes, it's all about finding the crtc.

...
I thought there was some other reason (like the virtual hw getting upset about certain things).

virtio wouldn't be upset.

It's a pointless exercise though to first disable the output, just to re-enable it the next moment with the new page-flipped framebuffer. So I guess I should look at the active_only thing nevertheless.

It's still possible that the plane can get disabled without it getting enabled. Userspace is allowed to do this. But since this suprises a lot of driver writers there's a special atomic_plane_disable hook you can use for the disable path. Instead of hand-rolling that check in your own code.

...

...
btw can you pls drop an ack or r-b onto my virtio conversion? I already added your tested-by.

Grr, mail is not in my dri-devel folder. Guess that is the <censored> "avoid-duplicates" mailman option at work.

Feel free to just add the r-b too. Or I'll send it for the next version of the series.

Added your r-b locally so it won't get lost on the next round.

Thanks, Daniel

-- Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch

3269

Age (days ago)

3511

Last active (days ago)

dri-devel@lists.freedesktop.org

17 comments

2 participants

tags (0)

participants (2)

Daniel Vetter
Gerd Hoffmann