Message ID | 20190610165806.24854-1-robh@kernel.org |
---|---|
State | Accepted |
Commit | a0e93c41eb75ab05bd07d4bd8027f30d484b3966 |
Headers | show |
Series | drm/panfrost: Align GEM objects GPU VA to 2MB | expand |
On Mon, 10 Jun 2019 at 18:58, Rob Herring <robh@kernel.org> wrote: > > In order to increase the chances of using 2MB pages, we need to align the > GPU VA mapping to 2MB. Only do this if the object size is 2MB or more. > > Cc: Robin Murphy <robin.murphy@arm.com> > Cc: Steven Price <steven.price@arm.com> > Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> > Signed-off-by: Rob Herring <robh@kernel.org> Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com> Thanks! Tomeu > --- > drivers/gpu/drm/panfrost/panfrost_gem.c | 8 ++++++-- > 1 file changed, 6 insertions(+), 2 deletions(-) > > diff --git a/drivers/gpu/drm/panfrost/panfrost_gem.c b/drivers/gpu/drm/panfrost/panfrost_gem.c > index a5528a360ef4..886875ae31d3 100644 > --- a/drivers/gpu/drm/panfrost/panfrost_gem.c > +++ b/drivers/gpu/drm/panfrost/panfrost_gem.c > @@ -52,6 +52,7 @@ struct drm_gem_object *panfrost_gem_create_object(struct drm_device *dev, size_t > int ret; > struct panfrost_device *pfdev = dev->dev_private; > struct panfrost_gem_object *obj; > + u64 align; > > obj = kzalloc(sizeof(*obj), GFP_KERNEL); > if (!obj) > @@ -59,9 +60,12 @@ struct drm_gem_object *panfrost_gem_create_object(struct drm_device *dev, size_t > > obj->base.base.funcs = &panfrost_gem_funcs; > > + size = roundup(size, PAGE_SIZE); > + align = size >= SZ_2M ? SZ_2M >> PAGE_SHIFT : 0; > + > spin_lock(&pfdev->mm_lock); > - ret = drm_mm_insert_node(&pfdev->mm, &obj->node, > - roundup(size, PAGE_SIZE) >> PAGE_SHIFT); > + ret = drm_mm_insert_node_generic(&pfdev->mm, &obj->node, > + size >> PAGE_SHIFT, align, 0, 0); > spin_unlock(&pfdev->mm_lock); > if (ret) > goto free_obj; > -- > 2.20.1 > > _______________________________________________ > dri-devel mailing list > dri-devel@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/dri-devel
diff --git a/drivers/gpu/drm/panfrost/panfrost_gem.c b/drivers/gpu/drm/panfrost/panfrost_gem.c index a5528a360ef4..886875ae31d3 100644 --- a/drivers/gpu/drm/panfrost/panfrost_gem.c +++ b/drivers/gpu/drm/panfrost/panfrost_gem.c @@ -52,6 +52,7 @@ struct drm_gem_object *panfrost_gem_create_object(struct drm_device *dev, size_t int ret; struct panfrost_device *pfdev = dev->dev_private; struct panfrost_gem_object *obj; + u64 align; obj = kzalloc(sizeof(*obj), GFP_KERNEL); if (!obj) @@ -59,9 +60,12 @@ struct drm_gem_object *panfrost_gem_create_object(struct drm_device *dev, size_t obj->base.base.funcs = &panfrost_gem_funcs; + size = roundup(size, PAGE_SIZE); + align = size >= SZ_2M ? SZ_2M >> PAGE_SHIFT : 0; + spin_lock(&pfdev->mm_lock); - ret = drm_mm_insert_node(&pfdev->mm, &obj->node, - roundup(size, PAGE_SIZE) >> PAGE_SHIFT); + ret = drm_mm_insert_node_generic(&pfdev->mm, &obj->node, + size >> PAGE_SHIFT, align, 0, 0); spin_unlock(&pfdev->mm_lock); if (ret) goto free_obj;
In order to increase the chances of using 2MB pages, we need to align the GPU VA mapping to 2MB. Only do this if the object size is 2MB or more. Cc: Robin Murphy <robin.murphy@arm.com> Cc: Steven Price <steven.price@arm.com> Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Signed-off-by: Rob Herring <robh@kernel.org> --- drivers/gpu/drm/panfrost/panfrost_gem.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)