Merge tag 'drm-msm-next-2023-08-20' of https://gitlab.freedesktop.org/drm/msm into drm-next

Updates for v6.6, which includes a backmerge of msm-fixes to avoid conficts.

Core:
- SM6125 MDSS support

DPU:
- SM6125 DPU support
- Added subblocks to display snapshot
- Use UBWC data from MDSS driver rather than duplicating it
- dpu_core_perf cleanup

DSI:
- Enabled burst mode to fix CMD mode panels
- Runtime PM support
- refgen regulator support

DSI PHY:
- SM6125 support in 14nm DSI PHY driver

GPU:
- Rework GPU identification to prepare for a7xx, and other a7xx prep
- Cleanups and fixes
- Disallow legacy relocs on a6xx and newer
- a690: switch to using a660_gmu.bin fw as this is what we have in
  linux-firmware and we see no evidence that it should be different
  from other a660 family (a6xx subgen 4) devices
- Submit overhead opts, 1.6x faster for NO_IMPLICIT_SYNC commits with
  100 BOs to 2.5x faster for 1000 BOs

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rob Clark <robdclark@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAF6AEGv_01g-edjdfKLWWcb-rO5aSyLsv5FpbKrTkXVL9+ngTQ@mail.gmail.com

1 files changed, 8 insertions, 7 deletions

diff --git a/drivers/gpu/drm/msm/msm_ringbuffer.c b/drivers/gpu/drm/msm/msm_ringbuffer.c
index b60199184409..40c0bc35a44c 100644
--- a/drivers/gpu/drm/msm/msm_ringbuffer.c
+++ b/drivers/gpu/drm/msm/msm_ringbuffer.c
@@ -16,25 +16,26 @@ static struct dma_fence *msm_job_run(struct drm_sched_job *job)
 	struct msm_gem_submit *submit = to_msm_submit(job);
 	struct msm_fence_context *fctx = submit->ring->fctx;
 	struct msm_gpu *gpu = submit->gpu;
+	struct msm_drm_private *priv = gpu->dev->dev_private;
 	int i;
 
 	msm_fence_init(submit->hw_fence, fctx);
 
+	submit->seqno = submit->hw_fence->seqno;
+
+	mutex_lock(&priv->lru.lock);
+
 	for (i = 0; i < submit->nr_bos; i++) {
-		struct drm_gem_object *obj = &submit->bos[i].obj->base;
+		struct drm_gem_object *obj = submit->bos[i].obj;
 
-		msm_gem_vma_unpin_fenced(submit->bos[i].vma, fctx);
 		msm_gem_unpin_active(obj);
-		submit->bos[i].flags &= ~(BO_VMA_PINNED | BO_OBJ_PINNED);
+		submit->bos[i].flags &= ~BO_PINNED;
 	}
 
-	/* TODO move submit path over to using a per-ring lock.. */
-	mutex_lock(&gpu->lock);
+	mutex_unlock(&priv->lru.lock);
 
 	msm_gpu_submit(gpu, submit);
 
-	mutex_unlock(&gpu->lock);
-
 	return dma_fence_get(submit->hw_fence);
 }