[PATCH 3/7] drm/ttm: allow reserve more than one shared slot
Michel Dänzer
michel at daenzer.net
Mon Sep 24 14:11:12 UTC 2018
The shortlog should say "allow reserving more than one shared fence
slot", same as patch 2.
On 2018-09-24 1:58 p.m., Christian König wrote:
> Let's support simultaneous submissions to multiple engines.
>
> Signed-off-by: Christian König <christian.koenig at amd.com>
> ---
> drivers/gpu/drm/ttm/ttm_execbuf_util.c | 6 ++++--
> include/drm/ttm/ttm_execbuf_util.h | 2 +-
> 2 files changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/ttm/ttm_execbuf_util.c b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
> index e493edb0d3e7..9b842884530a 100644
> --- a/drivers/gpu/drm/ttm/ttm_execbuf_util.c
> +++ b/drivers/gpu/drm/ttm/ttm_execbuf_util.c
> @@ -129,7 +129,8 @@ int ttm_eu_reserve_buffers(struct ww_acquire_ctx *ticket,
> if (!entry->shared)
> continue;
>
> - ret = reservation_object_reserve_shared(bo->resv, 1);
> + ret = reservation_object_reserve_shared(bo->resv,
> + entry->shared);
> if (!ret)
> continue;
> }
> @@ -151,7 +152,8 @@ int ttm_eu_reserve_buffers(struct ww_acquire_ctx *ticket,
> }
>
> if (!ret && entry->shared)
> - ret = reservation_object_reserve_shared(bo->resv, 1);
> + ret = reservation_object_reserve_shared(bo->resv,
> + entry->shared);
>
> if (unlikely(ret != 0)) {
> if (ret == -EINTR)
> diff --git a/include/drm/ttm/ttm_execbuf_util.h b/include/drm/ttm/ttm_execbuf_util.h
> index b0fdd1980034..3b16dda9bd08 100644
> --- a/include/drm/ttm/ttm_execbuf_util.h
> +++ b/include/drm/ttm/ttm_execbuf_util.h
> @@ -46,7 +46,7 @@
> struct ttm_validate_buffer {
> struct list_head head;
> struct ttm_buffer_object *bo;
> - bool shared;
> + unsigned int shared;
> };
>
> /**
>
It would be better to change the name of the field to something more
descriptive, e.g. "num_shared_fence_contexts", and fix up all users
accordingly in this patch (which will get rid of one hunk in patch 4).
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
More information about the amd-gfx
mailing list