PT22 + MacStudio + RME MadiFX - not functioning

[Hugh-H]

Hello,

I'm having an issue running PT22 on a MacStudio M1Max with an RME HDSPe MadiFX interface in an expansion chassis. I'll check on the RME forum but I'm posting here because the system works fine with the RME MadiFX and other DAWs such as Nuendo, hoping someone has some information.

A - MacStudio M1Max
B - OS12.6.1 (not officially approved but.....)
C - PT22.7
D - RME HDSPe MadiFX in a Sonnet expansion chassis, driver v1.31 (latest M1 native driver).
E - This system with MadiFX card works fine with other daws such as Nuendo.
F - This system with PT22.7 works fine with other audio interfaces including RME UFX+.
G - This system with PT22.7 works fine with HDX cards in an expansion chassis, HDX cards and MadiFX card are not present simultaneously.
H - Clearing preferences and reinstalling PT does not help.

Symptoms -
A - Launching another daw on the system, all works as advertised and all 194x196 I/O work fine.
B - Launching PT22.7 on the system, the RME driver and PT both lock up and become unresponsive. Force quitting PT does not bring back the RME driver, only a reboot restores operation.
C - Launching another daw to put the MadiFX into operation then concurrently attempting to run PT (the MadiFX is multi-client) still stalls the card driver and PT. Nuendo then stalls because the card driver is non-responsive.

I've tested this multiple times with multiple reboots and it occurs every time. Looking at the crash logs the RME driver thread crashes with a KERN_INVALID_ADDRESS error and a possible pointer authentication failure. No other crashed threads.

From the testing it appears -
- Card works fine in the system.
- CoreAudio works fine in the system.
- PT22.7 works fine in the system with other interfaces.
- PT22.7 with the MadiFX calls a kernel address that isn't a good idea.

Any brilliant suggestions from the esteemed crowd that has read this far?
Thanks,
Hugh

[Darryl Ramm]

I would be contacting RME support.

Pro Tools does not get to write to kernel memory, but it could be doing something that upsets the RME driver and it writes to an invalid address. The Apple Silicon drivers have to deal with crypto based pointer authentication, and this may have some of the symptoms of that. Unfortunately you can't disable that, or kernel address randomization either, something else that maybe might be a factor here. Thunderbolt connectivity might also introduce some problems for Apple Silicon drivers that might not be be present in pure PCIe environments, and driver vendors may need to update the internal of their drivers/what kernel software support they are using, I expect RME to fully on top of that, but they need to investigate what you are seeing

[audiogamble]

At what point do things lock up. Is it as soon as you launch pro tools, or just when you open a session?

What interface(s) are downstream of your rme madi card?

How is clocking set up in your system? Is the RME the master, or is something else?


Sent from my iPhone using Tapatalk

[its2loud]

Was this working fine before a particular update or is this a new system?

How is the expansion chassis plugged into the MacStudio?

[Hugh-H]

Hello,
Thank you for the follow up -

@audiogamble -
The RME driver locks up when PT attempts to use it as it's interface and as mentioned previously PT works fine with other interfaces.
It's not clocking or anything outside the RME card, as mentioned previously the card works fine with other daws on the same MacStudio.

@its2loud -
The MacStudio is a new system and works fine with other interfaces and with HDX cards.
The MadiFX card in the expansion chassis works fine outside PT with other daws.

@Darryl Ramm -
I will contact RME, I am asking here because PT is doing something that other daws are not doing and it's causing a kernel thread crash. The card otherwise works fine.
I'll update if I find anything.

Hugh

[bradch00]

One thought, would it be worth trying to run one of your other DAWs under Rosetta and see if something similar happens?
Doesn’t pinpoint the problem, but may narrow the scope of investigation.

[weezul]

>194x196 I/O work fine.

My guess is it might be something to do with this? Can you decrease the amount of channels actually being used or shown in Totalmix or HDSPe to the CoreAudio driver? If so I'd try it and see if Pro Tools works. Maybe something is wrong when it tries to limit you to 64io?

[its2loud]

Originally Posted by Hugh-H

@its2loud -
The MacStudio is a new system and works fine with other interfaces and with HDX cards.
The MadiFX card in the expansion chassis works fine outside PT with other daws.

Hugh

I PM’d you

[Hugh-H]

Hello,
I'm writing this in case anyone else needs assistance and is searching this issue, we've seemingly found the cause and workaround.

RME was unable to assist, we sent crash reports but Matthias said they were unable to test a similar system so no resolution would be found. The only possibility of working is to attempt the new RME DriverKit driver which for the MadiFX requires a minimum of OS11. We use our audio interfaces and software on a lot of daws with various OS versions. The fact that we must reflash MadiFX card firmware to move a MadiFX between an OS12 daw and an OS10.x daw, which we do a lot, is a hardship for us so we've been looking for a solution.

Testing with 7 different Sonnet expansion chassis versions, it appears certain older Sonnet chassis firmware is not compatible with whatever location/pointer calls the MacStudio is making across the TB bus causing the RME driver thread to crash. Interestingly we previously thought it was a PT issue only as other audio software worked fine with the older chassis after testing for 10 hours of heavy dubber use. Testing with another even older chassis rev caused other software to crash the RME driver as well. The same older chassis firmware works fine with an Intel system running the same OS and PT and other software versions.

Swapping the older TB interface card in a chassis with a compatible newer TB card, that chassis appears to now work fine. We have one older firmware chassis that we don't have a compatible newer TB card for so it's not going to be able to be used in this scenario. Note that the newer cards are not necessarily TB3, TB2 works fine, but certain older interface cards do not like what PT/RME are doing.

Hope this helps someone,
Hugh

[Darryl Ramm]

Wow thanks for posting that info. What a pain. It's interesting to see RME have issues here, maybe helps Avid look better with problems they have been having with HD Native. Is there any way to identify minimum rev Sonnet boxes that work OK or not?

I really wish Apple would get going with a real expandable mac Pro with PCIe slots... that's hopefully only slightly unaffordable