search

PhilippeBekaert / snd-hdspe

New linux driver and tools for RME HDSPe sound cards and extension modules
GNU General Public License v3.0
51 stars 18 forks source link

Compatibility with RME Fireface UFX+ #21

Open JustALawnGnome7 opened 9 months ago

JustALawnGnome7 commented 9 months ago

I noticed on RME's website that the Thunderbolt driver for the Fireface UFX+ is listed as an "HDSP(e) Thunderbolt" driver. Given that Thunderbolt inherently exposes a device as a PCI/PCIe device to the operating system, is there any chance that this driver can be easily modified to make the Fireface UFX+ the first pro-audio Thunderbolt device with Linux support?

I unfortunately don't have access to a Fireface UFX+ interface, but would be interested in owning one if this kind of Linux functionality was available.

Schroedingers-Cat commented 9 months ago

I don't have high hopes for this to happen anytime soon. And I don't know if the UFX+ is close to the HDSPe hardware.

Your best bet in non-PCIe land for such a channel count is the Fireface UFX III with USB class compliance mode for all of its available channels.

Alternatively, stack multiple class-compliant RME USB devices (or Firewire-based using the upstreamed snd-firewire drivers). Note, most other USB class-compliant devices won't support more than 24ins/outs including most RME interfaces. The UFX III is the only exception to this that I know of.

I read somewhere that the RME USB Madiface (not XT) should also work on Linux, but you have to connect some more devices to it to get some kind of UFX functionality and I never tested this myself.

JustALawnGnome7 commented 9 months ago

I guess I'm asking about the UFX+ because the vendor's website explicitly lists the Fireface UFX+ driver as "HDSP(e)".

https://www.rme-audio.de/fireface-ufx.html

Also, being a Thunderbolt interface implicitly makes the UFX+ a PCIe device. For example, I can see my Focusrite Clarett 8PreX Thunderbolt interface listed as a PCIe device when running lspci. I haven't managed to write a PCIe driver for it yet (I'm pretty new to kernel development), but the PCI subsystem definitely sees it.

I guess my real question is: if UFX+ hardware turned out to be as similar to other HDSPe devices as its driver name implies, what kind of information would be required to start probing the device for basic functionality? And given RME's history of opening up their source code to the Linux community, what are the chances of RME granting access to their driver code in order for us to know how similar the UFX+ is to other HDSPe devices?