On Thu, May 18, 2017 at 10:05:55 +0200, Michal Privoznik wrote:
https://bugzilla.redhat.com/show_bug.cgi?id=1450349
Problem is, memoryBacking is part of guest ABI. Therefore
changing it on migration/restore from an image can lead
qemu/guest to rejecting the image.
At the same time, move other partial checks of virDomainMemtune
into the same function: virDomainMemtuneCheckABIStability.
Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
---
src/conf/domain_conf.c | 101 ++++++++++++++++++++++++++++++++++++-------------
1 file changed, 75 insertions(+), 26 deletions(-)
diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
index 9eba70a95..89b93f4ad 100644
--- a/src/conf/domain_conf.c
+++ b/src/conf/domain_conf.c
@@ -19786,6 +19786,80 @@ virDomainTPMDefCheckABIStability(virDomainTPMDefPtr src,
return virDomainDeviceInfoCheckABIStability(&src->info, &dst->info);
}
+
+static bool
+virDomainMemtuneCheckABIStability(const virDomainDef *src,
+ const virDomainDef *dst,
+ unsigned int flags)
+{
+ if (virDomainDefGetMemoryInitial(src) != virDomainDefGetMemoryInitial(dst)) {
+ virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
+ _("Target domain max memory %lld "
+ "does not match source %lld"),
+ virDomainDefGetMemoryInitial(dst),
+ virDomainDefGetMemoryInitial(src));
+ return false;
+ }
+
+ if (!(flags & VIR_DOMAIN_DEF_ABI_CHECK_SKIP_VOLATILE) &&
+ src->mem.cur_balloon != dst->mem.cur_balloon) {
+ virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
+ _("Target domain current memory %lld "
+ "does not match source %lld"),
+ dst->mem.cur_balloon,
+ src->mem.cur_balloon);
+ return false;
+ }
+
+ if (src->mem.max_memory != dst->mem.max_memory) {
+ virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
+ _("Target maximum memory size '%llu' "
+ "doesn't match source '%llu'"),
+ dst->mem.max_memory,
+ src->mem.max_memory);
+ return false;
+ }
+
+ if (src->mem.memory_slots != dst->mem.memory_slots) {
+ virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
+ _("Target domain memory slots "
+ "count '%u' doesn't match source
'%u'"),
+ dst->mem.memory_slots,
+ src->mem.memory_slots);
+ return false;
+ }
+
+ if (src->mem.source != dst->mem.source) {
+ virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
+ _("Target memoryBacking source '%s' doesn't
"
+ "match source memoryBacking source'%s'"),
+ virDomainMemorySourceTypeToString(dst->mem.source),
+ virDomainMemorySourceTypeToString(src->mem.source));
+ return false;
+ }
AFAIK this is guest ABI only because qemu uses this to generate
hugepages rather than regular memory, not that it would be guest ABI
per se.
+
+ if (src->mem.access != dst->mem.access) {
+ virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
+ _("Target memoryBacking access '%s' doesn't
"
+ "match access memoryBacking access'%s'"),
+ virDomainMemoryAccessTypeToString(dst->mem.access),
+ virDomainMemoryAccessTypeToString(src->mem.access));
+ return false;
Same here, shared access does not really transform to guest ABI.
+ }
+
+ if (src->mem.allocation != dst->mem.allocation) {
+ virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
+ _("Target memoryBacking allocation '%s' doesn't
"
+ "match allocation memoryBacking
allocation'%s'"),
+ virDomainMemoryAllocationTypeToString(dst->mem.allocation),
+ virDomainMemoryAllocationTypeToString(src->mem.allocation));
+ return false;
This is definitely not guest ABI.
Note that there's a difference between guest ABI and differences in
configurations which are rejected by qemu.
I think we need a mechanism to deal with these in a different way
because it's possible that other hypervisors will have different ideas
on the compatibility of these options.