[lug] insight as gdb frontend?

[D. Stimits]

Tom Tromey wrote:
> 
> >>>>> "D" == D Stimits <stimits at idcomm.com> writes:
> 
> D> But as far as the reboot thing, it does on two different chipsets,
> D> 3 different 2.2.x kernels, and 2 different 2.4.x kernels. I think
> D> it has a weakness in SMP, that isn't entirely the kernel's fault,
> D> unless there is some long-standing and unknown bug that has
> D> traversed through all the kernels since several years back.
> 
> All I'm saying is that if the kernel crashes due to the actions of
> *any* userland program, then that is a kernel bug.  Period.  Sure, you
> might be able to work around the kernel bug by changing the userland
> program (in this case gdb).  But maybe you can't.  And anyway, it is
> still a bug in the kernel.
> 
> Look at it this way: gdb isn't doing anything special.  Unusual, maybe
> (not many programs use ptrace or whatever), but not special.  gdb
> doesn't run as root.  gdb doesn't install kernel modules.  If gdb can
> make a sequence of system calls which crashes your system, then so can
> any program.  This is a security hole.

I'm not sure reboot applies as a kernel crash. It shuts down and
restarts the system; my users have authority to do that, and it isn't
dying. I think it is possible that if it sends the right interrupt with
the proper value in a register that this is expected. It is a security
hole, but the exact sequence of events that leads to this under
different kernels and different chipsets and different hardware makes me
wonder, it is probably both a kernel issue and a program issue.

D. Stimits, stimits at idcomm.com

> 
> Tom
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug