From owner-freebsd-hackers Mon Jun 5 21:45:04 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id VAA26777 for hackers-outgoing; Mon, 5 Jun 1995 21:45:04 -0700 Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.34]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id VAA26769 for ; Mon, 5 Jun 1995 21:45:01 -0700 Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id OAA23875 for freebsd-hackers@freebsd.org; Tue, 6 Jun 1995 14:40:28 +1000 Date: Tue, 6 Jun 1995 14:40:28 +1000 From: Bruce Evans Message-Id: <199506060440.OAA23875@godzilla.zeta.org.au> To: freebsd-hackers@freebsd.org Subject: Re: Kernels, panics & the debugger Sender: hackers-owner@freebsd.org Precedence: bulk >[debugger, but not in case of panic] >> A kernel variable or something like this would be great; I would really like >> ... >There's already such a variable available, it's currently used by the >console to disable DDB fallback while the console is in graphics mode. >I'm not sure if Søren has already implemented the logic in syscons >however; i've done it for pcvt. This variable stops all debugger entries and crashes the system (or worse) if a breakpoint inserted by the debugger is hit (inserting breakpoints corrupts the code and the debugger doesn't get a chance to clean up). Better handling of trace traps is also required. Single stepping of syscalls starting from user mode currently causes debugger entry if ddb is configured. This should be disabled if ddb-on-panic is disabled and when the screen is in graphics mode and the console uses the screen. Single stepping of lcalls under X/pcvt must be verry sloow now :-). Bruce