Re: Identical repeated panics
Re: Identical repeated panics
- Subject: Re: Identical repeated panics
- From: Terry Lambert <email@hidden>
- Date: Fri, 31 Mar 2006 14:39:11 -0800
On Mar 30, 2006, at 9:18 AM, Brian Bechtel wrote:
On 3/30/06, Joaquim Soler i Subils <email@hidden> wrote:
Any explanation/solution? I hope it's something interesting....
One suggestion: You have a USB serial device attached. In some of
the panics, it shows the driver com.mct.driver.mctUSBSerialDriver in
the backtrace. Remove that driver from /System/Library/Extensions and
reboot. See if the panics stop happening.
If you cannot figure out which driver is
com.mct.driver.mctUSBSerialDriver, you can search the Info.plist file
for each driver. Open a terminal window, and type
grep -riIl com.mct.driver.mctUSBSerialDriver /System/Library/
Extensions
I agree that it's probably thee driver. What specifically causes the
panic is an explicit shutdown of the bootstrap thread, which is the
first thread in the kernel task. If it's forced to exit, and then is
woken up, rather than being disassembed, this will be the result.
This is a wakeup of a thread that's already called
"thread_terminate_self".
Most likely, someone is either making assumptions about thread state
for their own driver which starts a thread, or they have an
uninitialized variable that they use to reference what they *think* is
their own thread.
This is basically a "can't happen without third party code
misbehaving" panic. Most likely, the driver is structured
incorrectly, and a callback is being triggered on the thread which was
outstanding at the time the driver was being shut down. This could
happen if at driver shutdown time, the thread is killed without
waiting for all outstanding events to drain out first (common coding
error). Alternately, the thread is being terminated, but its global
reference is being zero'ed - also as a result of not allowing things
to drain out.
-- Terry
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden