[PATCH] drm/nouveau: fix nouveau_mm/nouveau_mm_node leak

[PATCH] drm/nouveau: fix nouveau_mm/nouveau_mm_node leak

[Marcin Slusarz]

Signed-off-by: Marcin Slusarz <marcin.slusarz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
---
 drivers/gpu/drm/nouveau/core/core/gpuobj.c         | 6 +++++-
 drivers/gpu/drm/nouveau/core/include/core/gpuobj.h | 3 +++
 2 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/core/core/gpuobj.c b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
index c2a7608..48121d2 100644
--- a/drivers/gpu/drm/nouveau/core/core/gpuobj.c
+++ b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
@@ -39,8 +39,11 @@ nouveau_gpuobj_destroy(struct nouveau_gpuobj *gpuobj)
 			nv_wo32(gpuobj, i, 0x00000000);
 	}
 
+	if (gpuobj->node)
+		nouveau_mm_free(gpuobj->node_heap, &gpuobj->node);
+
 	if (gpuobj->heap.block_size)
-		nouveau_mm_fini(&gpuobj->heap);
+		WARN_ON(nouveau_mm_fini(&gpuobj->heap));
 
 	nouveau_object_destroy(&gpuobj->base);
 }
@@ -114,6 +117,7 @@ nouveau_gpuobj_create_(struct nouveau_object *parent,
 				      max(align, (u32)1), &gpuobj->node);
 		if (ret)
 			return ret;
+		gpuobj->node_heap = heap;
 
 		gpuobj->addr += gpuobj->node->offset;
 	}
diff --git a/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h b/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h
index d09adf1..f65bf5b 100644
--- a/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h
+++ b/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h
@@ -16,7 +16,10 @@ struct nouveau_vm;
 struct nouveau_gpuobj {
 	struct nouveau_object base;
 	struct nouveau_object *parent;
+
+	struct nouveau_mm *node_heap;
 	struct nouveau_mm_node *node;
+
 	struct nouveau_mm heap;
 
 	u32 flags;
-- 
1.7.12

Re: [PATCH] drm/nouveau: fix nouveau_mm/nouveau_mm_node leak

[Marcin Slusarz]

On Thu, Oct 11, 2012 at 11:53:09PM +0200, Marcin Slusarz wrote:
> Signed-off-by: Marcin Slusarz <marcin.slusarz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> ---
>  drivers/gpu/drm/nouveau/core/core/gpuobj.c         | 6 +++++-
>  drivers/gpu/drm/nouveau/core/include/core/gpuobj.h | 3 +++
>  2 files changed, 8 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/nouveau/core/core/gpuobj.c b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> index c2a7608..48121d2 100644
> --- a/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> +++ b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> @@ -39,8 +39,11 @@ nouveau_gpuobj_destroy(struct nouveau_gpuobj *gpuobj)
>  			nv_wo32(gpuobj, i, 0x00000000);
>  	}
>  
> +	if (gpuobj->node)
> +		nouveau_mm_free(gpuobj->node_heap, &gpuobj->node);
> +
>  	if (gpuobj->heap.block_size)
> -		nouveau_mm_fini(&gpuobj->heap);
> +		WARN_ON(nouveau_mm_fini(&gpuobj->heap));
>  
>  	nouveau_object_destroy(&gpuobj->base);
>  }
> @@ -114,6 +117,7 @@ nouveau_gpuobj_create_(struct nouveau_object *parent,
>  				      max(align, (u32)1), &gpuobj->node);
>  		if (ret)
>  			return ret;
> +		gpuobj->node_heap = heap;
>  
>  		gpuobj->addr += gpuobj->node->offset;
>  	}
> diff --git a/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h b/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h
> index d09adf1..f65bf5b 100644
> --- a/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h
> +++ b/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h
> @@ -16,7 +16,10 @@ struct nouveau_vm;
>  struct nouveau_gpuobj {
>  	struct nouveau_object base;
>  	struct nouveau_object *parent;
> +
> +	struct nouveau_mm *node_heap;
>  	struct nouveau_mm_node *node;
> +
>  	struct nouveau_mm heap;
>  
>  	u32 flags;
> -- 

What's wrong with this patch?

Re: [PATCH] drm/nouveau: fix nouveau_mm/nouveau_mm_node leak

[Ben Skeggs]

On Thu, Oct 11, 2012 at 11:53:09PM +0200, Marcin Slusarz wrote:
> Signed-off-by: Marcin Slusarz <marcin.slusarz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> ---
>  drivers/gpu/drm/nouveau/core/core/gpuobj.c         | 6 +++++-
>  drivers/gpu/drm/nouveau/core/include/core/gpuobj.h | 3 +++
>  2 files changed, 8 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/nouveau/core/core/gpuobj.c b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> index c2a7608..48121d2 100644
> --- a/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> +++ b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> @@ -39,8 +39,11 @@ nouveau_gpuobj_destroy(struct nouveau_gpuobj *gpuobj)
>  			nv_wo32(gpuobj, i, 0x00000000);
>  	}
>  
> +	if (gpuobj->node)
> +		nouveau_mm_free(gpuobj->node_heap, &gpuobj->node);
> +
if (gpuobj->node) {
	nouveau_mm_free(&nv_gpuobj(gpuobj->parent)->heap,
			&gpuobj->node);
}

Or something to that effect, instead of having to store the heap
pointer again.

>  	if (gpuobj->heap.block_size)
> -		nouveau_mm_fini(&gpuobj->heap);
> +		WARN_ON(nouveau_mm_fini(&gpuobj->heap));
Alright, I get this.  However, perhaps we should go the full hog here
and make nouveau_mm_fini() directly do the WARN_ON() in this situation?

There was, once upon a time, reasons for it not doing this, I don't
believe they're valid anymore though.

If you want to do this, that'd be great.  Bonus points for being in a
separate patch :)

>  
>  	nouveau_object_destroy(&gpuobj->base);
>  }
> @@ -114,6 +117,7 @@ nouveau_gpuobj_create_(struct nouveau_object *parent,
>  				      max(align, (u32)1), &gpuobj->node);
>  		if (ret)
>  			return ret;
> +		gpuobj->node_heap = heap;
>  
>  		gpuobj->addr += gpuobj->node->offset;
>  	}
> diff --git a/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h b/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h
> index d09adf1..f65bf5b 100644
> --- a/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h
> +++ b/drivers/gpu/drm/nouveau/core/include/core/gpuobj.h
> @@ -16,7 +16,10 @@ struct nouveau_vm;
>  struct nouveau_gpuobj {
>  	struct nouveau_object base;
>  	struct nouveau_object *parent;
> +
> +	struct nouveau_mm *node_heap;
>  	struct nouveau_mm_node *node;
> +
>  	struct nouveau_mm heap;
>  
>  	u32 flags;
> -- 
> 1.7.12
> 
> _______________________________________________
> Nouveau mailing list
> Nouveau-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
> http://lists.freedesktop.org/mailman/listinfo/nouveau

Re: [PATCH] drm/nouveau: fix nouveau_mm/nouveau_mm_node leak

[Marcin Slusarz]

On Fri, Oct 19, 2012 at 04:05:14PM +1000, Ben Skeggs wrote:
> On Thu, Oct 11, 2012 at 11:53:09PM +0200, Marcin Slusarz wrote:
> > Signed-off-by: Marcin Slusarz <marcin.slusarz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> > ---
> >  drivers/gpu/drm/nouveau/core/core/gpuobj.c         | 6 +++++-
> >  drivers/gpu/drm/nouveau/core/include/core/gpuobj.h | 3 +++
> >  2 files changed, 8 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/gpu/drm/nouveau/core/core/gpuobj.c b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> > index c2a7608..48121d2 100644
> > --- a/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> > +++ b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> > @@ -39,8 +39,11 @@ nouveau_gpuobj_destroy(struct nouveau_gpuobj *gpuobj)
> >  			nv_wo32(gpuobj, i, 0x00000000);
> >  	}
> >  
> > +	if (gpuobj->node)
> > +		nouveau_mm_free(gpuobj->node_heap, &gpuobj->node);
> > +
> if (gpuobj->node) {
> 	nouveau_mm_free(&nv_gpuobj(gpuobj->parent)->heap,
> 			&gpuobj->node);
> }
> 
> Or something to that effect, instead of having to store the heap
> pointer again.

Oh, right.
(Somehow I assumed pargpu to be different object than parent when
I first read it.)

> >  	if (gpuobj->heap.block_size)
> > -		nouveau_mm_fini(&gpuobj->heap);
> > +		WARN_ON(nouveau_mm_fini(&gpuobj->heap));
> Alright, I get this.  However, perhaps we should go the full hog here
> and make nouveau_mm_fini() directly do the WARN_ON() in this situation?
> 
> There was, once upon a time, reasons for it not doing this, I don't
> believe they're valid anymore though.

OK.

Marcin

[PATCH v2] drm/nouveau: fix nouveau_mm/nouveau_mm_node leak

[Marcin Slusarz]

v2: use already existing parent

Signed-off-by: Marcin Slusarz <marcin.slusarz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
---
 drivers/gpu/drm/nouveau/core/core/gpuobj.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/gpu/drm/nouveau/core/core/gpuobj.c b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
index c2a7608..18b3f5c 100644
--- a/drivers/gpu/drm/nouveau/core/core/gpuobj.c
+++ b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
@@ -39,6 +39,11 @@ nouveau_gpuobj_destroy(struct nouveau_gpuobj *gpuobj)
 			nv_wo32(gpuobj, i, 0x00000000);
 	}
 
+	if (gpuobj->node) {
+		nouveau_mm_free(&nv_gpuobj(gpuobj->parent)->heap,
+				&gpuobj->node);
+	}
+
 	if (gpuobj->heap.block_size)
 		nouveau_mm_fini(&gpuobj->heap);
 
-- 
1.7.12

[PATCH] drm/nouveau: warn when trying to free mm which is still in use

[Marcin Slusarz]

Signed-off-by: Marcin Slusarz <marcin.slusarz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
---
 drivers/gpu/drm/nouveau/core/core/mm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/nouveau/core/core/mm.c b/drivers/gpu/drm/nouveau/core/core/mm.c
index bfddf87..4d62064 100644
--- a/drivers/gpu/drm/nouveau/core/core/mm.c
+++ b/drivers/gpu/drm/nouveau/core/core/mm.c
@@ -236,7 +236,7 @@ nouveau_mm_fini(struct nouveau_mm *mm)
 	int nodes = 0;
 
 	list_for_each_entry(node, &mm->nodes, nl_entry) {
-		if (nodes++ == mm->heap_nodes)
+		if (WARN_ON(nodes++ == mm->heap_nodes))
 			return -EBUSY;
 	}
 
-- 
1.7.12

Re: [PATCH] drm/nouveau: fix nouveau_mm/nouveau_mm_node leak

[Ben Skeggs]

On Mon, Oct 22, 2012 at 12:19:35AM +0200, Marcin Slusarz wrote:
> On Fri, Oct 19, 2012 at 04:05:14PM +1000, Ben Skeggs wrote:
> > On Thu, Oct 11, 2012 at 11:53:09PM +0200, Marcin Slusarz wrote:
> > > Signed-off-by: Marcin Slusarz <marcin.slusarz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> > > ---
> > >  drivers/gpu/drm/nouveau/core/core/gpuobj.c         | 6 +++++-
> > >  drivers/gpu/drm/nouveau/core/include/core/gpuobj.h | 3 +++
> > >  2 files changed, 8 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/gpu/drm/nouveau/core/core/gpuobj.c b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> > > index c2a7608..48121d2 100644
> > > --- a/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> > > +++ b/drivers/gpu/drm/nouveau/core/core/gpuobj.c
> > > @@ -39,8 +39,11 @@ nouveau_gpuobj_destroy(struct nouveau_gpuobj *gpuobj)
> > >  			nv_wo32(gpuobj, i, 0x00000000);
> > >  	}
> > >  
> > > +	if (gpuobj->node)
> > > +		nouveau_mm_free(gpuobj->node_heap, &gpuobj->node);
> > > +
> > if (gpuobj->node) {
> > 	nouveau_mm_free(&nv_gpuobj(gpuobj->parent)->heap,
> > 			&gpuobj->node);
> > }
> > 
> > Or something to that effect, instead of having to store the heap
> > pointer again.
Thanks, merged both patches :)

> 
> Oh, right.
> (Somehow I assumed pargpu to be different object than parent when
> I first read it.)
> 
> > >  	if (gpuobj->heap.block_size)
> > > -		nouveau_mm_fini(&gpuobj->heap);
> > > +		WARN_ON(nouveau_mm_fini(&gpuobj->heap));
> > Alright, I get this.  However, perhaps we should go the full hog here
> > and make nouveau_mm_fini() directly do the WARN_ON() in this situation?
> > 
> > There was, once upon a time, reasons for it not doing this, I don't
> > believe they're valid anymore though.
> 
> OK.
> 
> Marcin