On Fri, Jan 6, 2023 at 12:45 PM Daniel Vetter <daniel@xxxxxxxx> wrote:
>
> On Fri, 6 Jan 2023 at 10:56, Stanislaw Gruszka
> <stanislaw.gruszka@xxxxxxxxxxxxxxx> wrote:
> >
> > On Fri, Jan 06, 2023 at 10:28:15AM +0100, Daniel Vetter wrote:
> > > On Thu, Jan 05, 2023 at 07:38:26PM +0200, Oded Gabbay wrote:
> > > > On Thu, Jan 5, 2023 at 6:25 PM Jeffrey Hugo <quic_jhugo@xxxxxxxxxxx> wrote:
> > > > >
> > > > > On 1/5/2023 5:57 AM, Daniel Vetter wrote:
> > > > > > On Thu, Dec 08, 2022 at 12:07:27PM +0100, Jacek Lawrynowicz wrote:
> > > > > >> +static const struct drm_driver driver = {
> > > > > >> + .driver_features = DRIVER_GEM | DRIVER_COMPUTE_ACCEL,
> > > > > >
> > > > > > So I was wondering whether this is a bright idea, and whether we shouldn't
> > > > > > just go ahead and infuse more meaning into accel vs render nodes.
> > > > > >
> > > > > > The uapi relevant part of render nodes is that they're multi-user safe, at
> > > > > > least as much as feasible. Every new open() gives you a new private
> > > > > > accelerator. This also has implications on how userspace drivers iterate
> > > > > > them, they just open them all in turn and check whether it's the right
> > > > > > one - because userspace apis allow applications to enumerate them all.
> > > > > > Which also means that any devicie initialization at open() time is a
> > > > > > really bad idea.
> > > > > >
> > > > > > A lot of the compute accelerators otoh (well habanalabs) are single user,
> > > > > > init can be done at open() time because you only open this when you
> > > > > > actually know you're going to use it.
> > > > > >
> > > > > > So given this, shouldn't multi-user inference engines be more like render
> > > > > > drivers, and less like accel? So DRIVER_RENDER, but still under
> > > > > > drivers/accel.
> > > > > >
> > > > > > This way that entire separate /dev node would actually become meaningful
> > > > > > beyond just the basic bikeshed:
> > > > > > - render nodes are multi user, safe to iterate and open() just for
> > > > > > iteration
> > > > > > - accel nodes are single user, you really should not ever open them unless
> > > > > > you want to use them
> > > > > >
> > > > > > Of course would need a doc patch :-)
> > > > > >
> > > > > > Thoughts?
> > > > > > -Daniel
> > > > >
> > > > > Hmm.
> > > > >
> > > > > I admit, I thought DRIVER_ACCEL was the same as DRIVER_RENDER, except
> > > > > that DRIVER_ACCEL dropped the "legacy" dual node setup and also avoided
> > > > > "legacy" userspace.
> > > > >
> > > > > qaic is multi-user. I thought habana was the same, at-least for
> > > > > inference. Oded, am I wrong?
> > > > Habana's devices support a single user at a time acquiring the device
> > > > and working on it.
> > > > Both for training and inference.
> > > > >
> > > > > So, if DRIVER_ACCEL is for single-user (training?), and multi-user ends
> > > > > up in DRIVER_RENDER, that would seem to mean qaic ends up using
> > > > > DRIVER_RENDER and not DRIVER_ACCEL. Then qaic ends up over under
> > > > > /dev/dri with both a card node (never used) and a render node. That
> > > > > would seem to mean that the "legacy" userspace would open qaic nodes by
> > > > > default - something I understood Oded was trying to avoid.
> > > > >
> > > > > If there really a usecase for DRIVER_ACCEL to support single-user? I
> > > > > wonder why we can't default to multi-user, and if a particular
> > > > > user/driver has a single-user usecase, it enforces that in a driver
> > > > > specific manner?
> > > > >
> > > > > -Jeff
> > > >
> > > > Honestly, Daniel, I don't like this suggestion. I don't understand why
> > > > you make a connection between render/accel to single/multi user.
> > > >
> > > > As Jeff has said, one of the goals was to expose accelerator devices
> > > > to userspace with new device char nodes so we won't be bothered by
> > > > legacy userspace graphics software. This is something we all agreed on
> > > > and I don't see why we should change it now, even if you think it's
> > > > bike-shedding (which I disagree with).
> > > >
> > > > But in any case, creating a new device char nodes had nothing to do
> > > > with whether the device supports single or multi user. I can
> > > > definitely see in the future training devices that support multiple
> > > > users.
> > > >
> > > > The common drm/accel ioctls should of course not be limited to a
> > > > single user, and I agree with Jeff here, if a specific driver has such
> > > > a limitation (e.g. Habana), then that driver should handle it on its
> > > > own.
> > > > Maybe if there will be multiple drivers with such a limitation, we can
> > > > make that "handling" to be common code.
> > > >
> > > > Bottom line, I prefer we keep things as we all agreed upon in LPC.
> > >
> > > The problem is going to happen as soon as you have cross-vendor userspace.
> > > Which I'm kinda hoping is at least still the aspiration. Because with
> > > cross-vendor userspace you generally iterate & open all devices before you
> > > select the one you're going to use. And so we do kinda need a distinction,
> > > or we need that the single-user drivers also guarantee that open() is
> > > cheap.
> >
> > FWIW we had good support in ivpu for probe open's in form of lazy context
> > allocation. It was removed recently due to review feedback that this is
> > unnecessary, but we can add it back.
>
> Yeah once you have more than 1 multi-user accel chip in the system you
> need to do that. Which is really the reason why I think smashing
> multi-user client accel things into render is good, it forces drivers
> to suck less.
I still don't think it is a good idea, nor that it was the original
intention of this endeavour.
I don't want to repeat my previous email, so I'll just say again I
don't agree with the connection you make between accel devices and
single users.
Today Habana supports single-user, tomorrow it might change. But that
doesn't mean I will want to present my device as a GPU because I don't
want some app to try and render stuff on my device.
Moreover, accel devices can actually be specific IPs inside a more
general ASIC (e.g. GPU) which supports multi-user. In that case, one
might want to have an accel driver that supports multi-user on those
IPs but exposes /dev/accel and not render nodes.
Oded
>
> On that topic, does your userspace still do the drmIoctl() wrapper?
> -Daniel
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch
