Re: Bug#526640: Please try to keep the number of flashes minimal
On Mon, May 04, 2009 at 05:04:52PM +0200, Martin Michlmayr wrote:
> Copying Joey and maks, who initially had the discussion about adding a
> flash-kernel call to update-initramfs. I hope they can comment on
> your proposal.
>
> * Loïc Minier <lool@dooz.org> [2009-05-02 21:14]:
> > On Sat, May 02, 2009, Martin Michlmayr wrote:
> > > This is a long-standing problem (introduced a few months ago). The
> > > problem is that flash-kernel will add a postinst hook for the kernel,
> > > but nowadays update-initramfs will call flash-kernel directly.
> >
> > I was wondering whether we could take steps to move flash-kernel to a
> > trigger; I'm aware of earlier discussion on this topic on -boot. We
> > rediscussed this recently on #ubuntu-arm and the discussion is on:
> > https://bugs.launchpad.net/bugs/365053
> > http://launchpadlibrarian.net/25876740/ubuntu-arm.txt
> > most of the discussion is about moving the logic in update-initramfs'
> > run_bootloader() -- at least the flash-kernel part -- in a new config
> > similar to kenrel-img.conf's postinst_hook.
kernel-img.conf is getting deprecated as it appartains to nobody.
read those discussion some days ago but failed to see an obvious gain.
> > I think it would be possible to make flash-kernel calls trigger a new
> > flash-kernel trigger which would do the real update; the flash-kernel
> > postinst, update-initramfs calls, and kernel installation would all
> > cause this trigger to be activated, and the flash-kernel would only
> > have to ensure that the update-initramfs trigger if any has completed.
> >
> > Am I missing something? Does this make any sense?
did you read latest initramfs-tools git, we pass the version to
flash-kernel does this avoid aboves complications?
Reply to: