|
On Sat June 4 2016, 09:41:51 PM, jim burns wrote:
> On Fri June 3 2016, 09:39:54 PM, jim burns wrote:
> > Pls cc: me in any replies.
> >
> > On Sat May 21 2016, 11:41:27 PM, jim burns wrote:
> > > For kernels 3.19 - 4.2, the system won't boot into xen at all. In all
> > > cases,bare metal boots work fine.
> > >
> > > For kernels 4.3 - 4.6, and 4.7-rc, it gets more interesting, and I am
> > > able
> > > to provide a log (for 4.6). The system refuses to initialize my hard
> > > disk,
> > > and I am dumped into a rescue shell. I was able to initialze the
> > > network,
> > > and mount my Windows machine over cifs, so I could xfer the logs.
> > >
> > > I've played quite a bit with dracut's {add,force}_drivers+= parms, and
> > > the
> > > files in /usr/lib/modules-load.d, and currently load the following
> > > kernel
> > > modules, in the following order, to no effect. (I also don't remember
> > > the
> > > xen version for the last 4 years changing any of my results - 4.2 -
> > > 4.6.)
> >
> > Huh! I just installed kernel 4.5.6 today, and for the first time since
> > 3.18, I can boot into xen! Still can't boot into 4.5.5, or 4.6.0, and the
> > config files for 4.5.5 & 4.5.6 are identical, so I don't know what's
> > changed. I won't be losing the 4.5.6 rpms any time soon!
>
> Just installed 4.7rc1 today, and it boots into xen also (on my old machine).
> Haven't seen a version of 4.6 that will yet. I would love to know what
> changed in 4.7 (and possibly later 4.6 development) that probably was back
> ported between 4.5.5 and 4.5.6 that lets the latter boot into xen also.
Hmm, according to the kernel changelog for 4.5.6 -
https://cdn.kernel.org/pub/linux/kernel/v4.x/ChangeLog-4.5.6
The only xen related patch was:
commit f21a3ff053d11bd9442b6ae7f51503bc5b373feb
Author: Stefano Stabellini <sstabellini@xxxxxxxxxx>
Date: Wed Apr 20 14:15:01 2016 +0100
xen/x86: actually allocate legacy interrupts on PV guests
commit 702f926067d2a4b28c10a3c41a1172dd62d9e735 upstream.
b4ff8389ed14 is incomplete: relies on nr_legacy_irqs() to get the number
of legacy interrupts when actually nr_legacy_irqs() returns 0 after
probe_8259A(). Use NR_IRQS_LEGACY instead.
Signed-off-by: Stefano Stabellini <sstabellini@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
Since 4.5.5 and less, and 4.6.0 don't work for me, and this same patch shows
up in the changelog for 4.6.1, if kernel 4.6.1 works for me, it may have to do with this patch. | 
