diff options
author | 2016-12-18 16:53:26 +0100 | |
---|---|---|
committer | 2017-01-21 18:07:28 +0100 | |
commit | e42a9b3011f96ad26f4a19db77ac44cad31a4290 (patch) | |
tree | b6a1c7f918138dd3e4705d170dde85eb2e5aaa6a /patches/fuel-nailgun-agent/0004-AArch64-Read-hugepage-size-from-proc-meminfo.patch | |
parent | 5c258a9ae96fd2a0c6fd56e41c4368467e931782 (diff) |
Uplift Armband to Fuel Newton
[ Dan Andresan ]
- puppet: database: Fix Percona XtraBackup sync
[ Alexandru Avadanii ]
- disable all plugins;
- re-enable remote tracking;
- remove "Revert: Point to specific snapshot ...";
- patch context adjustments;
- obsolete "kernel-bump", linux-image-lts-xenial is now the default;
- network-checker iface state check now fixed upstream [1];
- fuel-nailgun-agent hugepage size should also check /proc/meminfo;
- fuel-nailgun-agent CPU detection for AArch64;
- nova AArch64 hugepage support is now upstream;
- obsolete Cirros Test VM direct kernel boot (by switch to AAVMF):
* f-l/0001-upload_cirros-Add-direct-kernel-boot-support.patch
* f-w/0001-direct-kernel-boot-for-cirros.patch
- rework m1.micro RAM size patch after puppet manifest split upstream;
- re-enable arch-agnostic plugins which were rebased in Fuel@OPNFV:
* f_yardstick-pluginbuild
* f_congress-pluginbuild
- do NOT retire MySQL SST provider patch series (nack: ARMBAND-186),
rebase (and keep for now) MySQL SST provider patches, as trying to
use xtrabackup-v2 revelead a regression since Colorado.3.0, and
these patches simplify troubleshooting a lot;
- AArch64: nova: libvirt: Use host-model cpu (ARMBAND-193);
- AArch64: nova: libvirt: Use pointer_model instead of use_usb_tablet;
- m1.micro RAM size insufficient for TestVM with AAVMF (s/128/256/)
- switch Cirros TestVM to AAVMF from direct kernel boot;
- backport nova libvirt driver fix for deleting instances booted with
AAVMF firmware from [2];
TODO (later):
- Include ISO build time fixes for cirros_testvm in Armband package;
TODO (ODL, later):
- test & revise leveldb patching;
- bring back Qugga patching for arm64;
- configure systemd service to automatically respawn;
[1] https://review.openstack.org/#/c/417373/
[2] https://review.openstack.org/#/c/357190/
JIRA: ARMBAND-29
JIRA: ARMBAND-32
JIRA: ARMBAND-63
JIRA: ARMBAND-88
JIRA: ARMBAND-116
JIRA: ARMBAND-118
JIRA: ARMBAND-186
JIRA: ARMBAND-193
JIRA: ARMBAND-194
JIRA: ARMBAND-195
JIRA: ARMBAND-196
JIRA: ARMBAND-197
Change-Id: Ia99022e364e61245d109cabab9d0ed7157b4d2f5
Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com>
Signed-off-by: Dan Andresan <dan.andresan@enea.com>
Diffstat (limited to 'patches/fuel-nailgun-agent/0004-AArch64-Read-hugepage-size-from-proc-meminfo.patch')
-rw-r--r-- | patches/fuel-nailgun-agent/0004-AArch64-Read-hugepage-size-from-proc-meminfo.patch | 30 |
1 files changed, 30 insertions, 0 deletions
diff --git a/patches/fuel-nailgun-agent/0004-AArch64-Read-hugepage-size-from-proc-meminfo.patch b/patches/fuel-nailgun-agent/0004-AArch64-Read-hugepage-size-from-proc-meminfo.patch new file mode 100644 index 00000000..de0ca78a --- /dev/null +++ b/patches/fuel-nailgun-agent/0004-AArch64-Read-hugepage-size-from-proc-meminfo.patch @@ -0,0 +1,30 @@ +From: Alexandru Avadanii <Alexandru.Avadanii@enea.com> +Date: Sun, 15 Jan 2017 08:14:12 +0100 +Subject: [PATCH] AArch64: Read hugepage size from /proc/meminfo + +This method should be arch-indepedent, provided /proc/meminfo +reports the correct information. + +Signed-off-by: Alexandru Avadanii <Alexandru.Avadanii@enea.com> +--- + agent | 7 +++++++ + 1 file changed, 7 insertions(+) + +diff --git a/agent b/agent +index eeb4d26..d514de6 100755 +--- a/agent ++++ b/agent +@@ -1399,6 +1399,13 @@ class NodeAgent + def supported_hugepages + return [2048, 1048576] if _get_detailed_cpuinfo['0'][:flags].include?('pdpe1gb') + return [2048] if _get_detailed_cpuinfo['0'][:flags].include?('pse') ++ # AArch64 does not expose CPU flags, but we can rely on /proc/meminfo ++ File.open('/proc/meminfo').each do |l| ++ case l.strip ++ when /Hugepagesize:\s+(\d+)\s+kB/ ++ return [$1.to_i()] ++ end ++ end + [] + end + |