Re: 5.4pre panic

From: Rene Ladan (r.c.ladan_at_student.tue.nl)
Date: 03/30/05

  • Next message: Rene Ladan: "Re: new LORs on 5.4 pre" 
    Date: Wed, 30 Mar 2005 16:17:08 +0200
To: Kris Kennaway <kris@obsecurity.org>

    On Wed, Mar 30, 2005 at 05:51:04AM -0800, Kris Kennaway wrote:
    > On Wed, Mar 30, 2005 at 12:11:23PM +0200, Rene Ladan wrote:
    > > Got this panic with auto-reboot (no dump :( ), saved from dmesg on
    > > 5.4pre 2003-05-29 09:49 UTC
    > >
    > > processor eflags = IOPL = 0
    > > current process = 29 (swi1: net)
    > > trap number = 3
    > > panic: breakpoint instruction fault
    >
    > I'd have expected this panic to be the result of setting a breakpoint,
    > i.e. requesting the kernel to panic at this location. Are you
    > absolutely certain this is not the case?
    I did not set a breakpoint, maybe some application was doing it for me.
    I was working in X at the moment.

    > If so, you should try to use
    > DDB to obtain a traceback to find out what it was doing.
    I have DDB/KDB in my kernelconfig, but auto-reboot without dump doesn't
    help much, does it?

    >
    > Kris

    Regards,
    Rene 

    -- 
"It won't fit on the line."
		-- me, 2001

  • Next message: Rene Ladan: "Re: new LORs on 5.4 pre"

    Relevant Pages

    • Re: [RFC] hwbkpt: Hardware breakpoints (was Kwatch)
      ... Entry n is the mask for when there are n kernel ... * Install the kernel breakpoints in their debug registers. ... * Store a thread breakpoint array entry's address ...
      (Linux-Kernel)
    • Re: [RFC] hwbkpt: Hardware breakpoints (was Kwatch)
      ... PC is changing to other than the breakpoint location hit by the handler ... EFLAGS image stored on the stack by the debug exception. ... If register with no callback followed by unregister ... It's no less meaningful than for a kernel allocation. ...
      (Linux-Kernel)
    • Re: [RFC] hwbkpt: Hardware breakpoints (was Kwatch)
      ... There's a minimal amount of additional activity needed to deal with kernel ... breakpoint updates that might arrive in the middle of a context switch. ... * Install the kernel breakpoints in their debug registers. ...
      (Linux-Kernel)
    • Re: [RFC] hwbkpt: Hardware breakpoints (was Kwatch)
      ... Debug exceptions don't occur very often, and when they do it won't ... The global kernel allocations will already be set in the processor from the ... CPU 1 goes in and alters the user breakpoint data, ... * Hardware breakpoints are implemented using the CPU's debug registers, ...
      (Linux-Kernel)
    • [RFC] hwbkpt: Hardware breakpoints (was Kwatch)
      ... When CPUs go on- or off-line, their debug registers need to be ... callback invoked when the breakpoint is installed ... * target address can be located in either kernel space or user space. ... * Install the debug register values for a new thread. ...
      (Linux-Kernel)