Please be more specific:
1- What System are you on? Leopard (which version)? Snow Leopard?
2- Your Mac specs: model? RAM? Processor?
3- What FF? 400 or 800? Driver version? Firmware?
The Fireface in OSX doesn't actually mount or un-mount. It just auto-launch its two applications when turned on: TotalMix and FireFace Settings.
To remove the unit, just quit the latter apps, turn down speakers volume and switch off FF.
As to Core Audio hanging forever forcing users to restart the computer:
I'm one of those iMac users who had to lean on workarounds to get their audio system working properly (2 years now), but STILL I believe that the one and only culprit for this sometimes unreliable FF/Mac relationship is : Apple FIREWIRE BUS along with its kernel extensions.
I don't want to spread mis-infos, this is just my experience and my point of view, consolidated after many different "geeky", "risky" system intrusions.
My story (briefly): been working flawlessly (Logic 8.0.2) on Tiger 10.4.11 and on Leopard (till version 10.5.4 !!!) using the FW 6pin/9pin cable workaround since first posted on this forum. Not a problem whatsoever.
Since Leopard 10.5.5 on, including Snow Leopard: FF400 and external FW drives no longer recognised and mounted after waking from sleep-mode; Core Audio freezing, System Profiler unable to detect its own firewire ports. Full immersion into reading and decoding logs, result: weird behaviour as to the system FW .kexts (5 files).
Analogic solution (don't try this at home if you don't know what you're doing!):
Installed FW.kext "package" from Tiger 10.4.11 Installer through Pacifist, in order not to come across permissions problems.
Everything went back to normality: FF400 and Ext. FW HD regularly seen and launched/mounted.
More details to come.
I'd like to read replies from RME guys first, as to this Frankenstein-like workaround.
Still I have to rely on the iMac/FF400 workaround, but hey, this has always worked fine actually.
Apple FW hardware-chipset caused and still causes problems to many RME users, now I'm pretty sure that the whole interaction between chipset and FW kernel extensions has to get involved to give a more in-deep analysis spectrum.
As already stated: just my experience, don't take it as a solid-solution or even semi-solid-solution.
Regards
Davide