Re: [PATCH 3/6] drm/tegra: Implement buffer object cache

[Thierry Reding <thierry.reding@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1176 bytes --]

On Thu, Sep 16, 2021 at 01:39:49PM +0300, Mikko Perttunen wrote:
> On 9/16/21 12:44 PM, Thierry Reding wrote:
> > From: Thierry Reding <treding@nvidia.com>
> > ...
> > diff --git a/drivers/gpu/drm/tegra/uapi.c b/drivers/gpu/drm/tegra/uapi.c
> > index 794c400c38b1..66fe8717e747 100644
> > --- a/drivers/gpu/drm/tegra/uapi.c
> > +++ b/drivers/gpu/drm/tegra/uapi.c
> > @@ -201,7 +201,8 @@ int tegra_drm_ioctl_channel_map(struct drm_device *drm, void *data, struct drm_f
> >   		goto put_gem;
> >   	}
> > -	mapping->map = host1x_bo_pin(context->client->base.dev, mapping->bo, direction);
> > +	mapping->map = host1x_bo_pin(context->client->base.dev, mapping->bo, direction,
> > +				     &context->client->base.cache);
> 
> Do we need caching here? The map/unmap operation is explicit and should not
> be on the hot path, and this will complicate context isolation support where
> we cannot have an engine-specific cache.

Yeah, we can probably leave it out here. I more or less blindly added
caches for everything, but really they're the most useful for cursors
and large framebuffers that will otherwise get mapped/unmapped once
every frame.

Thierry

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]