>>> On 11.06.13 at 16:47, ar16 <[email protected]> wrote: > Can you comment re: if/when the *suse `version` of the xen kernel will > move from the suse-modified&maintained, backported kernel to the pv-ops > upstream version, no longer requiring a separate kernel?
There's no clear plan here yet. To date the list of missing features is still too long for me to drop my resistance to that step, and the rate of regressions introduced upstream is still too high. Nevertheless I'm considering the move more actively than I did sys a year back. But that move would - for both above named reasons - require quite a bit of resources being available, which I can't currently foresee to happen. In any event - tying the move to pv-ops to the dropping of a separate Xen kernel is questionable: As long as the PV spin lock infrastructure can't be uniformly enabled due to having performance effects on the kernel when run on raw hardware, we will be in need of a separate kernel anyway. The good news here is that PV spin lock work has been resumed upstream recently, after having been stalled for a couple of years. Jan -- To unsubscribe, e-mail: [email protected] To contact the owner, e-mail: [email protected]
