Subject: Re: kernel debugger
To: der Mouse <mouse@Rodents.Montreal.QC.CA>
From: Guenther Grau <Guenther.Grau@marconicomms.com>
Date: 03/14/2000 22:32:05
der Mouse wrote:
> > If this is a "can't happen" kind of condition, rather than a
> > temporary debugging aid, you should probably call panic() instead,
> > since that will work whether or not DDB is compiled in, and will
> > cause the right thing to happen either way (panic/crash dump for
> > post-mortem, or dropping into the debugger).
> This brings up a question: if panic() drops into ddb, is there any way
> (short of really ugly hacks like bashing return addresses on the stack)
> to "ignore" the panic, to cause panic() to simply return?
I don't think panic() is designed to work like this.
If you don't like this, you can always compile in
your own panic implementation, which just returns.
panic() should be used rarely, IMHO, and only in
these places where the kernel cannot ensure running
properly anymore, due to a severy failure. Any other
uses of panic are a bug, IMHO :-) But I am sure you
knew that already, so I probably misunderstood your