From: Jeff Dike <jdike@ka...> - 2001-05-04 04:38:04
> o it got upset in wait_for_stop(). In wait_for_stop(), `status' is
> 0xa7f - so WSTOPSIG(status) is in fact SIGUSR1. wait_for_stop()
> doesn't seem to be expecting that.
For this one, I need to see the stack of whatever process is being traced at
this point. And figure out why you couldn't. When this happens again there
are a couple things you can try:
first, give that process a SIGSTOP before calling detach on it. detach() is
supposed to do that anyway, so this might just be superstition.
second, look at its /proc/<pid>/stat and get its ip and sp. With the sp, you
can reconstruct the stack by getting the stack page from
current_task.thread.kernel_stack, getting a hex dump from the sp on up, and
doing 'i sym 0xwhatever' on anything that looks like a text address.
> #0 panic (fmt=0x101633c0 "outer trampoline didn't exit 0") at panic.c:52
On this, I want the exit status that caused it to panic.