Date: Wed, 5 Jul 1995 15:05:26 +1000 From: Bruce Evans <bde@zeta.org.au> To: bde@zeta.org.au, karl@bagpuss.demon.co.uk Cc: hackers@freebsd.org, matt@lkg.dec.com Subject: Re: Best way to diagnose system lockups? Message-ID: <199507050505.PAA17190@godzilla.zeta.org.au>
next in thread | raw e-mail | index | archive | help
>> I usually remove the keyboard bit (1<<1) from tty_imask to debug wedging >> at spltty. >Does this make it possible to enter ddb at splclock/splimp too, or just >spltty? At spltty and splimp, but not at splclock. You could remove the bit from the clock imask to enter ddb at splclock too. This is not so easy since the clock imask is a compile-time constant. Alternatively, if ddb is enabled on the serial console, then it can be entered at any spl. I don't fully trust ddb to interrupt at any point (especially inside context switches). >> This is perfectly safe if you don't touch the keyboard and >> fairly safe (perhaps perfectly safe) even if you do (the keyboard intr >> handler sets the bit although spltty() doesn't). >hmm dont touch the keyboard at what time? After turning off the keyboard bit. I usually turn it off inside ddb but it would obviously be safer to change it in a program while you'r not touching the keyboard. >I've missed the lack of a 'halt button' for some time. Is it possible to >attach DDB to some NMI, even if this involves some extra hardware to trigger >that NMI? Seems to me that would be more useful than a PC-reset card :-) More useful than a PC reset button too :-). Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199507050505.PAA17190>