Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1189894

Summary: Ugly hacks to Nova for supporting aarch64
Product: Red Hat OpenStack Reporter: Richard W.M. Jones <rjones>
Component: openstack-novaAssignee: Eoghan Glynn <eglynn>
Status: CLOSED DEFERRED QA Contact: nlevinki <nlevinki>
Severity: medium Docs Contact:
Priority: medium    
Version: 7.0 (Kilo)CC: berrange, dasmith, drjones, eglynn, kchamart, rjones, sbauza, sclewis, sferdjao, sgordon, srevivo, vromanso
Target Milestone: ---   
Target Release: 9.0 (Mitaka)   
Hardware: aarch64   
OS: Linux   
Whiteboard:
Fixed In Version: openstack-nova-2014.2.2-3.aarch64.1.el7ost Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2016-04-26 14:14:32 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On:    
Bug Blocks: 1175340    
Attachments:
Description Flags
0001-Massive-hacks-to-get-things-working-on-aarch64-RHBZ-.patch
none
0001-HACK-nova-Workaround-for-libvirt-RHBZ-1184411.patch
none
0002-HACK-nova-loader-element-for-AAVMF.patch none

Description Richard W.M. Jones 2015-02-05 18:33:39 UTC
Description of problem:

As $subject says, these are very ugly and not upstreamable hacks
which let us boot guests on OpenStack on aarch64 (ARM 64 bit).

This bug is just to track the hacks and also discuss what can
be done to turn them into something upstreamable.

Version-Release number of selected component (if applicable):

openstack-nova-2014.2.1-15.aa7a

Additional information:

At the moment I have added these hacks to rhos-6.0-rhel-7-aarch64
in dist-git.

Comment 3 Richard W.M. Jones 2015-02-05 18:43:11 UTC
Created attachment 988596 [details]
0001-Massive-hacks-to-get-things-working-on-aarch64-RHBZ-.patch

Comment 4 Daniel Berrangé 2015-02-06 09:27:22 UTC
The attached patch doesn't change any code except the specfile ?!?!

Comment 5 Richard W.M. Jones 2015-02-06 10:08:24 UTC
Bugzilla is broken and is known to corrupt patches.  Use this link:

https://bugzilla.redhat.com/attachment.cgi?id=988596

Comment 6 Richard W.M. Jones 2015-02-06 10:10:16 UTC
Bugzilla PatchReader bug 554878.

Comment 7 Richard W.M. Jones 2015-02-19 14:52:18 UTC
Created attachment 993650 [details]
0001-HACK-nova-Workaround-for-libvirt-RHBZ-1184411.patch

Updated patch 1/2.

Comment 8 Richard W.M. Jones 2015-02-19 14:52:47 UTC
Created attachment 993651 [details]
0002-HACK-nova-loader-element-for-AAVMF.patch

Updated patch 2/2.

Comment 9 Richard W.M. Jones 2015-02-19 14:57:43 UTC
Setting to MODIFIED as there is a new build which includes
these hacks:

http://brewweb.devel.redhat.com/brew/taskinfo?taskID=8754987

Comment 10 Richard W.M. Jones 2015-03-04 15:29:58 UTC
Setting back to ASSIGNED, since we could add support for virtio-scsi
CDROMs.  See Dan's comment:

http://post-office.corp.redhat.com/archives/virt-arm/2015-March/msg00025.html

in this thread:

http://post-office.corp.redhat.com/archives/virt-arm/2015-March/thread.html#00012

Comment 11 Daniel Berrangé 2015-03-04 15:38:05 UTC
NB, since we don't have to worry about backwards compatibility with legacy guests that only support virtio-blk, we could probably just make AArch64 default to virtio-scsi for both disks & cdrom, and thus ignore virtio-blk.

Comment 12 Richard W.M. Jones 2015-03-04 15:44:22 UTC
(In reply to Daniel Berrange from comment #11)
> NB, since we don't have to worry about backwards compatibility with legacy
> guests that only support virtio-blk, we could probably just make AArch64
> default to virtio-scsi for both disks & cdrom, and thus ignore virtio-blk.

Note that virtio-scsi is buggy upstream on aarch64 (bug 1184405).

Comment 15 Andrew Jones 2016-04-26 13:42:45 UTC
Are there still some hacks being tracked? Or does upstream work without them now, and this bug can be closed?

Thanks,
drew

Comment 16 Richard W.M. Jones 2016-04-26 14:14:32 UTC
Upstream is supposed to support it out of the box now, although
I've not tested that in a while.

In any case it's sufficient that we have this bug with the old
patches attached in case we need to refer to them, but it can be
closed.