Andrew,
We received your email last Friday and we have been looking into what the cause may be.
Thanks Phil, it's very good to know.
> the very first non-OS software I put on was
> Dev Studio. It also Kernel panics within 3 days every single time if I
> leave it running.
This has puzzled us as the Dev Studio is actually a carbon app and is several layers away from the kernel and drivers driver and is completely different code to the server. It should crash before it can tear down the kernel. We are also puzzled as to how the Dev Studio can cause a kernal panic when it is not doing anything at all. It is not consuming CPU or memory. It is just resident in memory.
I hesitate to guess as well, and it still seems odd to me that Apple can so easily blame a software package that isn't even running as root to cause a kernel panic... and what you say does just make it even more puzzling.
All your kernal panics are occuring in the MPIC driver which is the multiprocessor interrupt controller which interfaces the Multi processors to the other sub systems.
Kernel loadable modules in backtrace (with dependencies): com.apple.driver.AppleMPIC(1.1)@0x1d51c000
It is also always happening on CPU 0 and it is always a cpu_signal_handler: Lock pass timed out. These subsystems are invisible to us and are not called directly by the Witango Dev Studio or Server but through calls to APIs. Can you ask your apple engineer what can cause these types of panics in the MPIC driver. He should also be able to give you a detailed description of the meaning of the panics.
He's an Apple Technician, not an engineer. (unless the terms are interchangeable in this context?) However, I'll definitely be asking him when he gets back on Monday.
Can you pass this information on to me as it may lead us to what is causing the panics. Can he explain how a carbon app and a mach-o daemon can cause the same panic when sitting idle in memory not consuming CPU or other system resources?
That does sound very strange. I didn't know that, and assumed you were using some similar libraries or something. (although I don't even know enough about Mac programming to know if that's possible or not between a mach-o and carbon app) I almost think to check it again, but I was very careful to add a single app at a time and wait at least more than 4 days... and again after I simply rebooted and didn't load the Dev Studio, it didn't crash.
The MPIC driver is part of the opendarwin project. We have been trying to check bugzilla on the opendarwin site to see if there are any known bugs, but the bugzilla system has been down. Actually a lot of the Open Darwin site is not working.
E.g
[...]
We are endeavouring to resolve this issue but until we can reproduce it or find another user having the same issue it is like finding a needle in a haystack.
I very much appreciate that.. your efforts, and your situation. It's quite frustrating over on my end, and Apple's brush off didn't help.
Are you able to switch processors around or disable CPU0? Do you have another Dual processor G4 you can test the apps on.
I found someone down the hall with a dual G4. Slightly older (putting it in the hardware version range of Jeff's on the list here who's works) ... but she's willing to let me install witango and let it run for a few days. Not an ideal test with her running all her other apps, but at least if it kernel panics in a few days, we'll know something.. (I think)
I also dug up the OS X Server disk, so I'm going to try to install that on this dual, which is also the OS that Jeff is running successfully.
I see Jon just posted that he is running successfully. Now I wonder if it might be somewhat related to the speed. The panic log does say "time out" so perhaps these fastest(?) G4s have an issue. This is the 1.4Ghz G4 dual, which I think (not sure) was the fastest / last dual G4 to come out before the G5 was released.
Also what Jeff just posted about one of hardware, firmware, or OS needing to change, I personally tend to agree with.. though the Apple guy didn't really right come out and say it, he very much insinuated that the 3rd party developer would have to figure out how to work around any OS issues. (or work it out with Apple through developer channels.. but if they're as much help as he was...) and from that I think I can extrapolate firmware and hardware as well, unless there still is an actual hardware fault that the hardware test is not picking up. I believe the only way to be sure of that is to get another 1.4Ghz dual G4 (as in, virtually identical) and test it.
I will try the OS X Server and let you know how it goes. I'll also speak with our Apple Tech on Monday and let you know if anything comes of that.
Thanks again Phil, you always seem to do it right.
Cheers
- Andrew
________________________________________________________________________ TO UNSUBSCRIBE: Go to http://www.witango.com/maillist.taf
