Re: [stable] aboot - support recent kernels, fix build issues
On Tue, May 27, 2008 at 01:18:52PM -0600, dann frazier wrote:
> On Tue, May 27, 2008 at 08:17:09PM +0200, Uwe Schindler wrote:
> > The problem with updating the aboot package is a little bit more
> > complicated:
> > It is not enough to update the aboot package alone! You must also write the
> > new boot record to your HDD. So if you would add a dependency to the kernel
> > image, that needs a newer aboot, it would not help: the people to make their
> > systems unbootable.
> I agree, and ideally the boot block would be written out with every
> update of aboot. I don't feel comfortable making such a large
> behavior change in etch, but I think it'd be great to see this happen
> for lenny.
I believe that would make us the only bootloader package in the archive to
routinely make such a high-risk change at package install time. I'm not
aware of any method for reliably detecting the boot device from inside the
running system on alpha.
I think the right change is for the kernel package to detect in preinst that
a wrong boot block is installed, and bail. Unfortunately, I haven't had any
time yet to work out how to detect such a wrong boot block.
> For etchnhalf, I think the real benefit of this update is at
> installtime. New installs w/ 2.6.24 will automatically get the updated
> aboot. The case of existing 2.6.18 users updating to 2.6.24
> should be explained in the release notes - but for those who don't
> read the release notes, they can still revert to their existing 2.6.18
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer http://www.debian.org/