On Fri, May 18, 2018 at 03:21:11PM +0300, Mikko Perttunen wrote:
> On 05/17/2018 06:34 PM, Thierry Reding wrote:
> > From: Thierry Reding <treding@xxxxxxxxxx>
> >
> > Userspace needs to know the version of the interface implemented by a
> > client so it can create the proper command streams. Allow individual
> > drivers to store this version along with the client so that it can be
> > returned to userspace upon opening a channel.
> >
> > Signed-off-by: Thierry Reding <treding@xxxxxxxxxx>
> > ---
> > include/linux/host1x.h | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/include/linux/host1x.h b/include/linux/host1x.h
> > index 89110d896d72..57d26406bdfd 100644
> > --- a/include/linux/host1x.h
> > +++ b/include/linux/host1x.h
> > @@ -49,6 +49,7 @@ struct host1x_client_ops {
> > * @dev: pointer to struct device backing this host1x client
> > * @ops: host1x client operations
> > * @class: host1x class represented by this client
> > + * @version: interface version implemented by this client
> > * @channel: host1x channel associated with this client
> > * @syncpts: array of syncpoints requested for this client
> > * @num_syncpts: number of syncpoints requested for this client
> > @@ -61,6 +62,8 @@ struct host1x_client {
> > const struct host1x_client_ops *ops;
> > enum host1x_class class;
> > + unsigned int version;
> > +
>
> It doesn't seem to me that this fits here - Host1x doesn't provide any
> userspace interface, TegraDRM does. We will (hopefully) have clients in the
> future that use a different userspace interface, or don't have one at all.
> So this property should be on TegraDRM side instead.
I think the line is somewhat blurry. Technically it is the host1x
clients that define the version of the interface that they provide. By
interface I mean the "class" of the command stream supported. So gr2d
will define what methods are valid to use on a specific version of the
module. The same goes for gr3d and VIC. The hardware module is where
that specification lives.
So if we ever had a client that didn't provide a userspace interface via
Tegra DRM we'd still want to represent what version of the (command
stream) interface is expected. By moving this to drm/tegra, we move the
hardware specific bits into the userspace related parts, so I'm not sure
it is a really good fit.
However, I do understand where you're coming from. Ultimately where
userspace gets involved is at the ABI level that drm/tegra exposes. And
for drivers that don't expose a userspace ABI, they'd likely need to
deal with the version number internally, so storing this in host1x
clients wouldn't fit very well either.
I don't really care all that much how exactly we store the version
information so that it can be reported back to userspace, and moving it
to drm/tegra makes the code somewhat more elegant. I came up with the
below, which is completely untested but should work.
Thierry
--- >8 ---
diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c
index 4330d5d48801..5dc02340007e 100644
--- a/drivers/gpu/drm/tegra/drm.c
+++ b/drivers/gpu/drm/tegra/drm.c
@@ -1389,7 +1389,7 @@ static int tegra_open_channel(struct drm_device *drm, void *data,
break;
args->syncpts = client->base.num_syncpts;
- args->version = client->base.version;
+ args->version = client->version;
args->context = context->id;
break;
}
diff --git a/drivers/gpu/drm/tegra/drm.h b/drivers/gpu/drm/tegra/drm.h
index 13660f6225a2..f493e0bf2172 100644
--- a/drivers/gpu/drm/tegra/drm.h
+++ b/drivers/gpu/drm/tegra/drm.h
@@ -105,6 +105,7 @@ struct tegra_drm_client {
struct host1x_client base;
struct list_head list;
+ unsigned int version;
const struct tegra_drm_client_ops *ops;
};
diff --git a/drivers/gpu/drm/tegra/gr2d.c b/drivers/gpu/drm/tegra/gr2d.c
index 7cae8252f0fa..165a1cece039 100644
--- a/drivers/gpu/drm/tegra/gr2d.c
+++ b/drivers/gpu/drm/tegra/gr2d.c
@@ -196,11 +196,11 @@ static int gr2d_probe(struct platform_device *pdev)
gr2d->client.base.ops = &gr2d_client_ops;
gr2d->client.base.dev = dev;
gr2d->client.base.class = HOST1X_CLASS_GR2D;
- gr2d->client.base.version = gr2d->soc->version;
gr2d->client.base.syncpts = syncpts;
gr2d->client.base.num_syncpts = 1;
INIT_LIST_HEAD(&gr2d->client.list);
+ gr2d->client.version = gr2d->soc->version;
gr2d->client.ops = &gr2d_ops;
err = host1x_client_register(&gr2d->client.base);
diff --git a/drivers/gpu/drm/tegra/gr3d.c b/drivers/gpu/drm/tegra/gr3d.c
index 9bd89916d2ef..0d72e78bd6e8 100644
--- a/drivers/gpu/drm/tegra/gr3d.c
+++ b/drivers/gpu/drm/tegra/gr3d.c
@@ -325,11 +325,11 @@ static int gr3d_probe(struct platform_device *pdev)
gr3d->client.base.ops = &gr3d_client_ops;
gr3d->client.base.dev = &pdev->dev;
gr3d->client.base.class = HOST1X_CLASS_GR3D;
- gr3d->client.base.version = gr3d->soc->version;
gr3d->client.base.syncpts = syncpts;
gr3d->client.base.num_syncpts = 1;
INIT_LIST_HEAD(&gr3d->client.list);
+ gr3d->client.version = gr3d->soc->version;
gr3d->client.ops = &gr3d_ops;
err = host1x_client_register(&gr3d->client.base);
diff --git a/drivers/gpu/drm/tegra/vic.c b/drivers/gpu/drm/tegra/vic.c
index b9df467ca71e..9fa77405db01 100644
--- a/drivers/gpu/drm/tegra/vic.c
+++ b/drivers/gpu/drm/tegra/vic.c
@@ -342,12 +342,12 @@ static int vic_probe(struct platform_device *pdev)
vic->client.base.ops = &vic_client_ops;
vic->client.base.dev = dev;
vic->client.base.class = HOST1X_CLASS_VIC;
- vic->client.base.version = vic->config->version;
vic->client.base.syncpts = syncpts;
vic->client.base.num_syncpts = 1;
vic->dev = dev;
INIT_LIST_HEAD(&vic->client.list);
+ vic->client.version = vic->config->version;
vic->client.ops = &vic_ops;
err = host1x_client_register(&vic->client.base);
diff --git a/include/linux/host1x.h b/include/linux/host1x.h
index 6dc819b1ab64..518864cfbd16 100644
--- a/include/linux/host1x.h
+++ b/include/linux/host1x.h
@@ -62,7 +62,6 @@ struct host1x_client {
const struct host1x_client_ops *ops;
enum host1x_class class;
- unsigned int version;
struct host1x_channel *channel;
Attachment:
signature.asc
Description: PGP signature
_______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel
