Firewire bus going crazy

[petsounds]

At first I thought my Apogee Ensemble was going south, constantly rebooting itself. Then I looked at the system log, and started seeing these Firewire messages:

kernel[0]: FireWire (OHCI) TI ID 823f built-in: 316 bus resets in last 3 minutes.

The only Firewire device I'm running is the Ensemble. Turned it off, unplugged, still get the messages. So I did some searching, could find very little information on this and even less in the way of solutions. I've tried resetting the SMC, PRAM, deleting caches, etc. Nothing has helped. Has anyone run into this?

Thanks.

System: Mac Pro 2008 2.8Ghz dual quad-core, 10GB RAM
OS X 10.8.2

 

[justperry]

There are still quite a few options you can try.

Did you install any drivers lately, sometimes you'll get software conflicts, I had before as well.

Start up in safe mode, is the problem gone?

Try log in to another account, if it's the same there then it is system wide.

Do you have a bootable backup, if so startup from it and see what it does.
Start up in recovery with the disk attached, does it behave the same, this without reinstalling of course, just to check if it does the same here.

Re-applying the last update, you have to manually download 10.8.2 and re-apply it manually.
Edit: download the combo 10.8.2 update. (tip from Bear, poster below)

Look into System Profiler under extensions, are there any 3rd party drivers listed?
If so you might uninstall certain extensions, or use kextunload to unload them.

 

[Bear]

...
Re-applying the last update, you have to manually download 10.8.2 and re-apply it manually.
...

If you are going to do this, I would strongly recommend using the combo update (10.8.x->10.8.2) instead of the point update (10.8.1->10.8.2).

 

[justperry]

If you are going to do this, I would strongly recommend using the combo update (10.8.x->10.8.2) instead of the point update (10.8.1->10.8.2).

Good advise, forgot to say, I'll update my post.

 

[petsounds]

Thanks, those are good suggestions. I do have bootable backups, but just can't spare the time or risk system stability right now to do deep testing, compounded by the fact that the Firewire bus resets are intermittent (looking at the log, haven't had one since 2pm). Hopefully 10.8.3 will come out soon and I'll do the combo update, see if that fixes stuff...

 

[ColdCase]

Believe it or not, I had a bad USB keyboard messing with the motherboard controller that was causing numerous firewire bus resets (at least that was what the log was saying). Enough so that it was difficult to reliably capture video off a DVT camera (firewire connection) or hang more than one or two drives off the FW bus. I spent many hours trying to figure out which FW device was the issue.

The 2008 MacPro was still under warranty at the time. After discussing the issue with Apple support they sent me a replacement keyboard and I haven't noticed a bus reset since.

Don't ask me to explain the interaction between the USB and FW bus and bus controllers.


Edit: Before the video card went completely south recently, there were numerous indications that there was a bad drive, but I think that only affected the SATA bus/devices.

 

[bplein]

Don't ask me to explain the interaction between the USB and FW bus and bus controllers.

In any Intel architecture PC (desktop or server, Mac or other) these controllers hang off the chipset, either of an IOH or off a Southbridge chip. On the Mac Pro this would be (my assumption) an IOH.

If one of the devices is really going wonky, it's possible for it to put the upstream switch (the IOH in this case) into a tailspin. Imagine a bunch of "noise" hitting the switch... it may not have enough time to deal with good packets from another device, so it starts dropping packets.

You were looking in the right direction, looking at all your FW devices, but things like this can be other devices like USB as you discovered, because internally they are all feeding from the same upstream device.