On 10/23/2010 04:12 PM, Kevin O'Connor wrote:
On Thu, Oct 21, 2010 at 12:07:17PM +0200, Avi Kivity wrote:
How do we manage the stable series wrt this issue?
qemu-kvm-0.12.5 has a regression within the stable series that this patch fixes. qemu 0.12.5 does not, but only because it does not emulate polarity in the I/O APIC correctly.
There are several paths we could take:
- do nothing, bug is fixed in mainline
- release a seabios 0.x.1 for qemu 0.13.1 with this patch
- same, plus seabios 0.y.1 for qemu 0.12.6 with this patch
- skip qemu (which is not truly affected), patch qemu-kvm's copy of
seabios for both 0.12.z and 0.13.z
The third option is the most "correct" from a release engineering point of view, but involves more work for everyone.
I'm okay with making tags and branches of seabios for bug fixes. So far qemu/kvm has just grabbed various builds of seabios - is it worthwhile to branch off of the seabios-0.6.1 version - which would mean qemu/kvm would pull in additional changes beyond the bug fix above?
qemu 0.12 is based on 0.5.1-stable, appears to be an untagged commit qemu 0.13 is based on 17d3e46511, doesn't appear to be a part of a branch or a tag?
git-wise, tags are more important than branches. You can always retrofit a branch to a tag (and you can always retrofit a tag to a commit hash). For the qemu git repositories, neither matter so much since the commit is recorded in git; but the distro people really like nice stable tags with lots of digits and dots in them.