This is the mail archive of the
mailing list for the Cygwin project.
Re: setup hangs during postinstall
On Mon, 13 Oct 2003, Christopher Faylor wrote:
> On Mon, Oct 13, 2003 at 12:22:12PM -0500, Brian Ford wrote:
> >Since setup must be launched from explorer to hang, I set
> >CYGWIN_DEBUG=cygpath in my system environment variables and rebooted. To
> >make sure it worked, I launched a bash under rxvt and did a "cygpath -S".
> >Up popped the gdb cmd shell. But, under setup, no dice.
> Are you saying that there was no popup? That would imply that CYGWIN_DEBUG
> is not in the environment. You could just hack on the code in dcrt0.cc to
> automatically pop up a gdb when command line contains cygpath, as a temporary
> measure, of course.
Yes, but I'm sure it *was* in the environment. In fact, after terminating
my manually attached gdb session, it popped up (this is NT4, so cypath
had to exit at gdb detach time).
Ok about hacking dcrt0.cc. Will try as time permits.
> >cygpath hangs just like before. The stack trace looks a little different.
> >Note that I have not had time to hand decode the holes in the trace yet,
> >but it looks to me like a startup problem before the CYGWIN_DEBUG
> >handling code. A "fork exec copy" problem maybe? I know that's vague,
> >but I think you know what I mean.
> The stack trace below is from a process that is running cygpath not
> from cygpath itself.
Ok. That is what it looked like to me too, but I don't know how the
Cygwin "fork exec" stuff works yet (spawn_guts is huge), so I wasn't
Maybe I was just too rushed, but I'm almost positive I attached to
the only thing ps -ef said was cygpath. Why does the spawner (bash?) say
> So, there should be another cygpath process which is stuck, as you say,
> prior to the point where it has "acquired the pid". You should be able
> to see it with a "ps -W".
Ok, I'll look.
Senior Realtime Software Engineer
VITAL - Visual Simulation Systems
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html