Bug#776450: Xen PVH support for grub-xen in Buster

[Hans van Kranenburg]

On 1/7/19 10:17 PM, Colin Watson wrote:
> On Mon, Jan 07, 2019 at 09:53:36PM +0100, Hans van Kranenburg wrote:
>> Note that:
>>
>> Installing for x86_64-xen platform.
>> grub-install: warning: no hints available for your platform. Expect
>> reduced performance.
>> Installation finished. No error reported.
>>
>> ...which means that it only has /boot/grub/x86_64-xen with a copy of the
>> modules. Maybe that's something to think about for the part of the story
>> that actually expects those modules to be there? Maybe just always throw
>> all of the three different types in there?
> 
> Oh right, I indeed forgot that bit.  I've pushed a fix (though no need
> to retest).  i386-xen can only boot a 32-bit kernel and so isn't usable
> with an amd64 userspace, but i386-xen_pvh can be used either way.
> 
>> Anyway, the first thing we're testing doesn't rely on that, so let's
>> continue. In the xen config file, I use...
>>
>> ---- >8 ----
>> kernel = "/usr/lib/grub-xen/grub-i386-xen_pvh.bin"
>> type = "pvh"
>> ---- >8 ----
>>
>> ...and yes, blue grub screen and it boots.
> 
> Excellent; thanks a lot for the test!

And thanks for picking the whole issue up so quickly. ;] I'm really
happy the completed puzzle is going to be in Buster now.

>> Suggestion:
>>
>> From usability point of view, it would actually also be nice if
>> console=hvc0 would be added by default in the GRUB_CMDLINE_LINUX_DEFAULT
>> for inside the Xen domU.
>>
>> The default is now "quiet", but if you remove "quiet", it stays as
>> quiet, and for a new user, it doesn't have to be super obvious what to
>> do in this case.
> 
> It would be helpful not to collect too many not-very-closely-related
> issues into this single bug.  In any case, if you install via the Debian
> installer then it arranges for this to happen; we could look into doing
> it somewhere else if there were a compelling reason to do so, but I'm a
> little wary of shuffling all that fairly delicate machinery around as we
> approach release, and d-i already sets up a number of other similar
> things so is in a good position to do this.

Yes, you're right.

Hans