Re: bug tracking
On Tue, 2002-11-19 at 22:17, Rob Weir wrote:
> On Wed, Nov 20, 2002 at 03:03:59PM +1300, Richard Hector wrote:
> > On Tue, 2002-11-19 at 21:36, Rob Weir wrote:
> > >
> > > When the kernel crashes, there's no way for it to be able to know that
> > > it's state is consistent. Because of this, it's not safe for it to try
> > > to write to disks (since it could easily destroy everything on the
> > > disks).
> > >
> > > The best it can manage is to write an 'oops' to the screen. You';; have
> > > to either write this down manually off the screen, or plug in a serial
> > > console and tell the kernel to dump oopses onto the serial port.
> >
> > Other unixes seem to manage to dump to the swap partition - is there
> > some significant difference that make this impractical/more dangerous
> > for Linux?
>
> I believe that 2.5 has this capability now too. I'm not sure what's
> changed though to make it safe.
With a patch I believe. Some believe its too dangerous to have a kernel
in the middle of an oops trying to write to the disks. And I also
belive Linus said it was unnessecary and vendors could patch the kernel
if they wanted it. I think I remember seeing it on lwn or just maybe
picked it up reading lkm, either way dont expect it in 2.6.
--
Scott Henson <debian-list@silvercoin.dyndns.org>
Reply to: