Re: [PATCH 0/3] Nouveau fixes for TTM refcount rework

Re: [PATCH 0/3] Nouveau fixes for TTM refcount rework

[Ben Skeggs]

On 19/7/24 02:58, Danilo Krummrich wrote:

> Hi Christian,
>
> Those three patches should unblock your series to use GEM references instead of
> TTM ones.
>
> @Lyude, Dave: Can you please double check?

Hi Danilo,

These look fine to me, and appear to resolve the issues I see with just 
the refcount series applied.

Ben.


Reviewed-by: Ben Skeggs <bskeggs@nvidia.com>

>
> - Danilo
>
> Danilo Krummrich (3):
>    drm/nouveau: prime: fix refcount underflow
>    drm/nouveau: bo: remove unused functions
>    drm/nouveau: use GEM references instead of TTMs
>
>   drivers/gpu/drm/nouveau/dispnv04/crtc.c | 43 +++++++++++++++------
>   drivers/gpu/drm/nouveau/dispnv50/disp.c |  4 +-
>   drivers/gpu/drm/nouveau/nouveau_bo.h    | 50 ++-----------------------
>   drivers/gpu/drm/nouveau/nouveau_chan.c  |  2 +-
>   drivers/gpu/drm/nouveau/nouveau_dmem.c  |  4 +-
>   drivers/gpu/drm/nouveau/nouveau_prime.c |  3 +-
>   drivers/gpu/drm/nouveau/nv10_fence.c    |  2 +-
>   drivers/gpu/drm/nouveau/nv17_fence.c    |  2 +-
>   drivers/gpu/drm/nouveau/nv50_fence.c    |  2 +-
>   drivers/gpu/drm/nouveau/nv84_fence.c    |  4 +-
>   10 files changed, 46 insertions(+), 70 deletions(-)
>
>
> base-commit: 99e0fb8b087120b5a7019f1cff6c5c2b5b925ae5

[PATCH 0/3] Nouveau fixes for TTM refcount rework

[Danilo Krummrich]

Hi Christian,

Those three patches should unblock your series to use GEM references instead of
TTM ones.

@Lyude, Dave: Can you please double check?

- Danilo

Danilo Krummrich (3):
  drm/nouveau: prime: fix refcount underflow
  drm/nouveau: bo: remove unused functions
  drm/nouveau: use GEM references instead of TTMs

 drivers/gpu/drm/nouveau/dispnv04/crtc.c | 43 +++++++++++++++------
 drivers/gpu/drm/nouveau/dispnv50/disp.c |  4 +-
 drivers/gpu/drm/nouveau/nouveau_bo.h    | 50 ++-----------------------
 drivers/gpu/drm/nouveau/nouveau_chan.c  |  2 +-
 drivers/gpu/drm/nouveau/nouveau_dmem.c  |  4 +-
 drivers/gpu/drm/nouveau/nouveau_prime.c |  3 +-
 drivers/gpu/drm/nouveau/nv10_fence.c    |  2 +-
 drivers/gpu/drm/nouveau/nv17_fence.c    |  2 +-
 drivers/gpu/drm/nouveau/nv50_fence.c    |  2 +-
 drivers/gpu/drm/nouveau/nv84_fence.c    |  4 +-
 10 files changed, 46 insertions(+), 70 deletions(-)


base-commit: 99e0fb8b087120b5a7019f1cff6c5c2b5b925ae5
-- 
2.45.2

[PATCH 1/3] drm/nouveau: prime: fix refcount underflow

[Danilo Krummrich]

Calling nouveau_bo_ref() on a nouveau_bo without initializing it (and
hence the backing ttm_bo) leads to a refcount underflow.

Instead of calling nouveau_bo_ref() in the unwind path of
drm_gem_object_init(), clean things up manually.

Fixes: ab9ccb96a6e6 ("drm/nouveau: use prime helpers")
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
---
 drivers/gpu/drm/nouveau/nouveau_prime.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_prime.c b/drivers/gpu/drm/nouveau/nouveau_prime.c
index b58ab595faf8..cd95446d6851 100644
--- a/drivers/gpu/drm/nouveau/nouveau_prime.c
+++ b/drivers/gpu/drm/nouveau/nouveau_prime.c
@@ -64,7 +64,8 @@ struct drm_gem_object *nouveau_gem_prime_import_sg_table(struct drm_device *dev,
 	 * to the caller, instead of a normal nouveau_bo ttm reference. */
 	ret = drm_gem_object_init(dev, &nvbo->bo.base, size);
 	if (ret) {
-		nouveau_bo_ref(NULL, &nvbo);
+		drm_gem_object_release(&nvbo->bo.base);
+		kfree(nvbo);
 		obj = ERR_PTR(-ENOMEM);
 		goto unlock;
 	}
-- 
2.45.2

[PATCH 2/3] drm/nouveau: bo: remove unused functions

[Danilo Krummrich]

nouveau_bo_new_pin_map() and nouveau_bo_unmap_unpin_unref() are unused,
hence remove them.

Signed-off-by: Danilo Krummrich <dakr@kernel.org>
---
 drivers/gpu/drm/nouveau/nouveau_bo.h | 29 ----------------------------
 1 file changed, 29 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.h b/drivers/gpu/drm/nouveau/nouveau_bo.h
index 4e891752c255..3b8dfbb621da 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo.h
+++ b/drivers/gpu/drm/nouveau/nouveau_bo.h
@@ -115,35 +115,6 @@ nvbo_kmap_obj_iovirtual(struct nouveau_bo *nvbo)
 	return ioptr;
 }
 
-static inline void
-nouveau_bo_unmap_unpin_unref(struct nouveau_bo **pnvbo)
-{
-	if (*pnvbo) {
-		nouveau_bo_unmap(*pnvbo);
-		nouveau_bo_unpin(*pnvbo);
-		nouveau_bo_ref(NULL, pnvbo);
-	}
-}
-
-static inline int
-nouveau_bo_new_pin_map(struct nouveau_cli *cli, u64 size, int align, u32 domain,
-		       struct nouveau_bo **pnvbo)
-{
-	int ret = nouveau_bo_new(cli, size, align, domain,
-				 0, 0, NULL, NULL, pnvbo);
-	if (ret == 0) {
-		ret = nouveau_bo_pin(*pnvbo, domain, true);
-		if (ret == 0) {
-			ret = nouveau_bo_map(*pnvbo);
-			if (ret == 0)
-				return ret;
-			nouveau_bo_unpin(*pnvbo);
-		}
-		nouveau_bo_ref(NULL, pnvbo);
-	}
-	return ret;
-}
-
 int nv04_bo_move_init(struct nouveau_channel *, u32);
 int nv04_bo_move_m2mf(struct nouveau_channel *, struct ttm_buffer_object *,
 		      struct ttm_resource *, struct ttm_resource *);
-- 
2.45.2

[PATCH 3/3] drm/nouveau: use GEM references instead of TTMs

[Danilo Krummrich]

TTM wants to get rid of the duplicate refcounting of the embedded GEM
object and its own reference count.

Hence, use of GEM object references where possible.

Also get rid of nouveau_bo_ref() and replace it with nouveau_bo_fini(),
which drops the initial reference we get from initializing a ttm_bo.

Signed-off-by: Danilo Krummrich <dakr@kernel.org>
---
 drivers/gpu/drm/nouveau/dispnv04/crtc.c | 43 ++++++++++++++++++-------
 drivers/gpu/drm/nouveau/dispnv50/disp.c |  4 +--
 drivers/gpu/drm/nouveau/nouveau_bo.h    | 21 ++----------
 drivers/gpu/drm/nouveau/nouveau_chan.c  |  2 +-
 drivers/gpu/drm/nouveau/nouveau_dmem.c  |  4 +--
 drivers/gpu/drm/nouveau/nv10_fence.c    |  2 +-
 drivers/gpu/drm/nouveau/nv17_fence.c    |  2 +-
 drivers/gpu/drm/nouveau/nv50_fence.c    |  2 +-
 drivers/gpu/drm/nouveau/nv84_fence.c    |  4 +--
 9 files changed, 44 insertions(+), 40 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/dispnv04/crtc.c b/drivers/gpu/drm/nouveau/dispnv04/crtc.c
index 4310ad71870b..438e3ec6e1ca 100644
--- a/drivers/gpu/drm/nouveau/dispnv04/crtc.c
+++ b/drivers/gpu/drm/nouveau/dispnv04/crtc.c
@@ -617,9 +617,15 @@ nv_crtc_swap_fbs(struct drm_crtc *crtc, struct drm_framebuffer *old_fb)
 
 	ret = nouveau_bo_pin(nvbo, NOUVEAU_GEM_DOMAIN_VRAM, false);
 	if (ret == 0) {
-		if (disp->image[nv_crtc->index])
-			nouveau_bo_unpin(disp->image[nv_crtc->index]);
-		nouveau_bo_ref(nvbo, &disp->image[nv_crtc->index]);
+		if (disp->image[nv_crtc->index]) {
+			struct nouveau_bo *bo = disp->image[nv_crtc->index];
+
+			nouveau_bo_unpin(bo);
+			drm_gem_object_put(&bo->bo.base);
+		}
+
+		drm_gem_object_get(&nvbo->bo.base);
+		disp->image[nv_crtc->index] = nvbo;
 	}
 
 	return ret;
@@ -754,13 +760,17 @@ static void nv_crtc_destroy(struct drm_crtc *crtc)
 
 	drm_crtc_cleanup(crtc);
 
-	if (disp->image[nv_crtc->index])
-		nouveau_bo_unpin(disp->image[nv_crtc->index]);
-	nouveau_bo_ref(NULL, &disp->image[nv_crtc->index]);
+	if (disp->image[nv_crtc->index]) {
+		struct nouveau_bo *bo = disp->image[nv_crtc->index];
+
+		nouveau_bo_unpin(bo);
+		drm_gem_object_put(&bo->bo.base);
+		disp->image[nv_crtc->index] = NULL;
+	}
 
 	nouveau_bo_unmap(nv_crtc->cursor.nvbo);
 	nouveau_bo_unpin(nv_crtc->cursor.nvbo);
-	nouveau_bo_ref(NULL, &nv_crtc->cursor.nvbo);
+	nouveau_bo_fini(nv_crtc->cursor.nvbo);
 	nvif_event_dtor(&nv_crtc->vblank);
 	nvif_head_dtor(&nv_crtc->head);
 	kfree(nv_crtc);
@@ -794,9 +804,14 @@ nv_crtc_disable(struct drm_crtc *crtc)
 {
 	struct nv04_display *disp = nv04_display(crtc->dev);
 	struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc);
-	if (disp->image[nv_crtc->index])
-		nouveau_bo_unpin(disp->image[nv_crtc->index]);
-	nouveau_bo_ref(NULL, &disp->image[nv_crtc->index]);
+
+	if (disp->image[nv_crtc->index]) {
+		struct nouveau_bo *bo = disp->image[nv_crtc->index];
+
+		nouveau_bo_unpin(bo);
+		drm_gem_object_put(&bo->bo.base);
+		disp->image[nv_crtc->index] = NULL;
+	}
 }
 
 static int
@@ -1210,7 +1225,11 @@ nv04_crtc_page_flip(struct drm_crtc *crtc, struct drm_framebuffer *fb,
 		PUSH_NVSQ(push, NV05F, 0x0130, 0);
 	}
 
-	nouveau_bo_ref(new_bo, &dispnv04->image[head]);
+	if (dispnv04->image[head])
+		drm_gem_object_put(&dispnv04->image[head]->bo.base);
+
+	drm_gem_object_get(&new_bo->bo.base);
+	dispnv04->image[head] = new_bo;
 
 	ret = nv04_page_flip_emit(chan, old_bo, new_bo, s, &fence);
 	if (ret)
@@ -1329,7 +1348,7 @@ nv04_crtc_create(struct drm_device *dev, int crtc_num)
 				nouveau_bo_unpin(nv_crtc->cursor.nvbo);
 		}
 		if (ret)
-			nouveau_bo_ref(NULL, &nv_crtc->cursor.nvbo);
+			nouveau_bo_fini(nv_crtc->cursor.nvbo);
 	}
 
 	nv04_cursor_init(nv_crtc);
diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c b/drivers/gpu/drm/nouveau/dispnv50/disp.c
index ac9657d7e92d..6d20d3c68fa7 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@@ -2819,7 +2819,7 @@ nv50_display_destroy(struct drm_device *dev)
 	nouveau_bo_unmap(disp->sync);
 	if (disp->sync)
 		nouveau_bo_unpin(disp->sync);
-	nouveau_bo_ref(NULL, &disp->sync);
+	nouveau_bo_fini(disp->sync);
 
 	nouveau_display(dev)->priv = NULL;
 	kfree(disp);
@@ -2862,7 +2862,7 @@ nv50_display_create(struct drm_device *dev)
 				nouveau_bo_unpin(disp->sync);
 		}
 		if (ret)
-			nouveau_bo_ref(NULL, &disp->sync);
+			nouveau_bo_fini(disp->sync);
 	}
 
 	if (ret)
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.h b/drivers/gpu/drm/nouveau/nouveau_bo.h
index 3b8dfbb621da..596a63a50a20 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo.h
+++ b/drivers/gpu/drm/nouveau/nouveau_bo.h
@@ -53,25 +53,10 @@ nouveau_bo(struct ttm_buffer_object *bo)
 	return container_of(bo, struct nouveau_bo, bo);
 }
 
-static inline int
-nouveau_bo_ref(struct nouveau_bo *ref, struct nouveau_bo **pnvbo)
+static inline void
+nouveau_bo_fini(struct nouveau_bo *bo)
 {
-	struct nouveau_bo *prev;
-
-	if (!pnvbo)
-		return -EINVAL;
-	prev = *pnvbo;
-
-	if (ref) {
-		ttm_bo_get(&ref->bo);
-		*pnvbo = nouveau_bo(&ref->bo);
-	} else {
-		*pnvbo = NULL;
-	}
-	if (prev)
-		ttm_bo_put(&prev->bo);
-
-	return 0;
+	ttm_bo_put(&bo->bo);
 }
 
 extern struct ttm_device_funcs nouveau_bo_driver;
diff --git a/drivers/gpu/drm/nouveau/nouveau_chan.c b/drivers/gpu/drm/nouveau/nouveau_chan.c
index 66fca95c10c7..f568ea251e3b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_chan.c
+++ b/drivers/gpu/drm/nouveau/nouveau_chan.c
@@ -110,7 +110,7 @@ nouveau_channel_del(struct nouveau_channel **pchan)
 		nouveau_bo_unmap(chan->push.buffer);
 		if (chan->push.buffer && chan->push.buffer->bo.pin_count)
 			nouveau_bo_unpin(chan->push.buffer);
-		nouveau_bo_ref(NULL, &chan->push.buffer);
+		nouveau_bo_fini(chan->push.buffer);
 		kfree(chan);
 	}
 	*pchan = NULL;
diff --git a/drivers/gpu/drm/nouveau/nouveau_dmem.c b/drivers/gpu/drm/nouveau/nouveau_dmem.c
index 6719353e2e13..7b3b8f4630a2 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dmem.c
+++ b/drivers/gpu/drm/nouveau/nouveau_dmem.c
@@ -294,7 +294,7 @@ nouveau_dmem_chunk_alloc(struct nouveau_drm *drm, struct page **ppage)
 out_bo_unpin:
 	nouveau_bo_unpin(chunk->bo);
 out_bo_free:
-	nouveau_bo_ref(NULL, &chunk->bo);
+	nouveau_bo_fini(chunk->bo);
 out_release:
 	release_mem_region(chunk->pagemap.range.start, range_len(&chunk->pagemap.range));
 out_free:
@@ -426,7 +426,7 @@ nouveau_dmem_fini(struct nouveau_drm *drm)
 	list_for_each_entry_safe(chunk, tmp, &drm->dmem->chunks, list) {
 		nouveau_dmem_evict_chunk(chunk);
 		nouveau_bo_unpin(chunk->bo);
-		nouveau_bo_ref(NULL, &chunk->bo);
+		nouveau_bo_fini(chunk->bo);
 		WARN_ON(chunk->callocated);
 		list_del(&chunk->list);
 		memunmap_pages(&chunk->pagemap);
diff --git a/drivers/gpu/drm/nouveau/nv10_fence.c b/drivers/gpu/drm/nouveau/nv10_fence.c
index c6a0db5b9e21..1a53b8b80467 100644
--- a/drivers/gpu/drm/nouveau/nv10_fence.c
+++ b/drivers/gpu/drm/nouveau/nv10_fence.c
@@ -88,7 +88,7 @@ nv10_fence_destroy(struct nouveau_drm *drm)
 	nouveau_bo_unmap(priv->bo);
 	if (priv->bo)
 		nouveau_bo_unpin(priv->bo);
-	nouveau_bo_ref(NULL, &priv->bo);
+	nouveau_bo_fini(priv->bo);
 	drm->fence = NULL;
 	kfree(priv);
 }
diff --git a/drivers/gpu/drm/nouveau/nv17_fence.c b/drivers/gpu/drm/nouveau/nv17_fence.c
index 07c2e0878c24..2c99f2c1ddcd 100644
--- a/drivers/gpu/drm/nouveau/nv17_fence.c
+++ b/drivers/gpu/drm/nouveau/nv17_fence.c
@@ -141,7 +141,7 @@ nv17_fence_create(struct nouveau_drm *drm)
 				nouveau_bo_unpin(priv->bo);
 		}
 		if (ret)
-			nouveau_bo_ref(NULL, &priv->bo);
+			nouveau_bo_fini(priv->bo);
 	}
 
 	if (ret) {
diff --git a/drivers/gpu/drm/nouveau/nv50_fence.c b/drivers/gpu/drm/nouveau/nv50_fence.c
index ea1e1f480bfe..6fa18f9d26b6 100644
--- a/drivers/gpu/drm/nouveau/nv50_fence.c
+++ b/drivers/gpu/drm/nouveau/nv50_fence.c
@@ -92,7 +92,7 @@ nv50_fence_create(struct nouveau_drm *drm)
 				nouveau_bo_unpin(priv->bo);
 		}
 		if (ret)
-			nouveau_bo_ref(NULL, &priv->bo);
+			nouveau_bo_fini(priv->bo);
 	}
 
 	if (ret) {
diff --git a/drivers/gpu/drm/nouveau/nv84_fence.c b/drivers/gpu/drm/nouveau/nv84_fence.c
index 812b8c62eeba..9ce4c2d60fe3 100644
--- a/drivers/gpu/drm/nouveau/nv84_fence.c
+++ b/drivers/gpu/drm/nouveau/nv84_fence.c
@@ -188,7 +188,7 @@ nv84_fence_destroy(struct nouveau_drm *drm)
 	nouveau_bo_unmap(priv->bo);
 	if (priv->bo)
 		nouveau_bo_unpin(priv->bo);
-	nouveau_bo_ref(NULL, &priv->bo);
+	nouveau_bo_fini(priv->bo);
 	drm->fence = NULL;
 	kfree(priv);
 }
@@ -232,7 +232,7 @@ nv84_fence_create(struct nouveau_drm *drm)
 				nouveau_bo_unpin(priv->bo);
 		}
 		if (ret)
-			nouveau_bo_ref(NULL, &priv->bo);
+			nouveau_bo_fini(priv->bo);
 	}
 
 	if (ret)
-- 
2.45.2

Re: [PATCH 1/3] drm/nouveau: prime: fix refcount underflow

[Christian König]

Am 18.07.24 um 18:58 schrieb Danilo Krummrich:
> Calling nouveau_bo_ref() on a nouveau_bo without initializing it (and
> hence the backing ttm_bo) leads to a refcount underflow.
>
> Instead of calling nouveau_bo_ref() in the unwind path of
> drm_gem_object_init(), clean things up manually.
>
> Fixes: ab9ccb96a6e6 ("drm/nouveau: use prime helpers")
> Signed-off-by: Danilo Krummrich <dakr@kernel.org>

Thanks for looking into this, feel free to add Reviewed-by: Christian 
König <christian.koenig@amd.com> to this patch.

But since especially patch #3 is not something I can fully judge 
correctness on I can only give an acked-by to the other two.

Regards,
Christian.

> ---
>   drivers/gpu/drm/nouveau/nouveau_prime.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/nouveau/nouveau_prime.c b/drivers/gpu/drm/nouveau/nouveau_prime.c
> index b58ab595faf8..cd95446d6851 100644
> --- a/drivers/gpu/drm/nouveau/nouveau_prime.c
> +++ b/drivers/gpu/drm/nouveau/nouveau_prime.c
> @@ -64,7 +64,8 @@ struct drm_gem_object *nouveau_gem_prime_import_sg_table(struct drm_device *dev,
>   	 * to the caller, instead of a normal nouveau_bo ttm reference. */
>   	ret = drm_gem_object_init(dev, &nvbo->bo.base, size);
>   	if (ret) {
> -		nouveau_bo_ref(NULL, &nvbo);
> +		drm_gem_object_release(&nvbo->bo.base);
> +		kfree(nvbo);
>   		obj = ERR_PTR(-ENOMEM);
>   		goto unlock;
>   	}

Re: [PATCH 0/3] Nouveau fixes for TTM refcount rework

[Danilo Krummrich]

On 7/18/24 7:53 AM, Ben Skeggs wrote:
> On 19/7/24 02:58, Danilo Krummrich wrote:
> 
>> Hi Christian,
>>
>> Those three patches should unblock your series to use GEM references instead of
>> TTM ones.
>>
>> @Lyude, Dave: Can you please double check?
> 
> Hi Danilo,
> 
> These look fine to me, and appear to resolve the issues I see with just the refcount series applied.
> 
> Ben.
> 
> 
> Reviewed-by: Ben Skeggs <bskeggs@nvidia.com>

Thanks for reviewing!

@Christian, I applied the series to drm-misc-next for you to go ahead and
cherry-picked the bug fix to drm-misc-fixes.

- Danilo

> 
>>
>> - Danilo
>>
>> Danilo Krummrich (3):
>>    drm/nouveau: prime: fix refcount underflow
>>    drm/nouveau: bo: remove unused functions
>>    drm/nouveau: use GEM references instead of TTMs
>>
>>   drivers/gpu/drm/nouveau/dispnv04/crtc.c | 43 +++++++++++++++------
>>   drivers/gpu/drm/nouveau/dispnv50/disp.c |  4 +-
>>   drivers/gpu/drm/nouveau/nouveau_bo.h    | 50 ++-----------------------
>>   drivers/gpu/drm/nouveau/nouveau_chan.c  |  2 +-
>>   drivers/gpu/drm/nouveau/nouveau_dmem.c  |  4 +-
>>   drivers/gpu/drm/nouveau/nouveau_prime.c |  3 +-
>>   drivers/gpu/drm/nouveau/nv10_fence.c    |  2 +-
>>   drivers/gpu/drm/nouveau/nv17_fence.c    |  2 +-
>>   drivers/gpu/drm/nouveau/nv50_fence.c    |  2 +-
>>   drivers/gpu/drm/nouveau/nv84_fence.c    |  4 +-
>>   10 files changed, 46 insertions(+), 70 deletions(-)
>>
>>
>> base-commit: 99e0fb8b087120b5a7019f1cff6c5c2b5b925ae5

Re: [PATCH 0/3] Nouveau fixes for TTM refcount rework

[Christian König]

Am 22.07.24 um 23:01 schrieb Danilo Krummrich:
> On 7/18/24 7:53 AM, Ben Skeggs wrote:
>> On 19/7/24 02:58, Danilo Krummrich wrote:
>>
>>> Hi Christian,
>>>
>>> Those three patches should unblock your series to use GEM references 
>>> instead of
>>> TTM ones.
>>>
>>> @Lyude, Dave: Can you please double check?
>>
>> Hi Danilo,
>>
>> These look fine to me, and appear to resolve the issues I see with 
>> just the refcount series applied.
>>
>> Ben.
>>
>>
>> Reviewed-by: Ben Skeggs <bskeggs@nvidia.com>
>
> Thanks for reviewing!
>
> @Christian, I applied the series to drm-misc-next for you to go ahead and
> cherry-picked the bug fix to drm-misc-fixes.

Thanks, I was already about to re-base my stuff on it.

Going to send out a new series today.

Christian.

>
> - Danilo
>
>>
>>>
>>> - Danilo
>>>
>>> Danilo Krummrich (3):
>>>    drm/nouveau: prime: fix refcount underflow
>>>    drm/nouveau: bo: remove unused functions
>>>    drm/nouveau: use GEM references instead of TTMs
>>>
>>>   drivers/gpu/drm/nouveau/dispnv04/crtc.c | 43 +++++++++++++++------
>>>   drivers/gpu/drm/nouveau/dispnv50/disp.c |  4 +-
>>>   drivers/gpu/drm/nouveau/nouveau_bo.h    | 50 
>>> ++-----------------------
>>>   drivers/gpu/drm/nouveau/nouveau_chan.c  |  2 +-
>>>   drivers/gpu/drm/nouveau/nouveau_dmem.c  |  4 +-
>>>   drivers/gpu/drm/nouveau/nouveau_prime.c |  3 +-
>>>   drivers/gpu/drm/nouveau/nv10_fence.c    |  2 +-
>>>   drivers/gpu/drm/nouveau/nv17_fence.c    |  2 +-
>>>   drivers/gpu/drm/nouveau/nv50_fence.c    |  2 +-
>>>   drivers/gpu/drm/nouveau/nv84_fence.c    |  4 +-
>>>   10 files changed, 46 insertions(+), 70 deletions(-)
>>>
>>>
>>> base-commit: 99e0fb8b087120b5a7019f1cff6c5c2b5b925ae5