Re: gcore and core dumps...

[Marco Gerards]

"Alfred M. Szmidt" <ams@kemisten.nu> writes:

> haven't looked into this at all, but thought that I might report it
> anyway and check if it is indeed the intended behavior.
> 
> If /servers/crash is not a core dumping crash server (i.e is either
> --kill or --suspend), then doing `gcore PID' will do whatever the
> crash server is currently.  That is, if the crash server is supposed
> to kill the program, then `gcore PID' will cause the program to die.
> I haven't checked if it will just suspend the processes if crash was
> started with --suspend, but I wouldn't be surprised if it did.  A core
> dump was not produced if crash --kill was used.

Well, I have encountered weirder things.  Perhaps it was my own
fault...

I have /hurd/crash, which is a symlink to crash-suspend (Which is
started as "/hurd/crash --suspend").

Well, I just assume that means that whenever a process segfault it
will wait until I can attach gdb to it.

Well, sometimes I just get coredumps with these settings.  Is this
possible and am I doing something really stupid here or is this a
bug?

Thanks,
Marco