Existing VMs with virtio devices and vhost-kernel as the backend
are always started with mem config:
"-m xG"
(with a ram block named "pc.ram")
while new VMs with virtio devices and vhost-user as the backend
are always started with mem config:
"-m xG -numa node,memdev=pc.ram -object memory-backend-file,id=pc.ram,..."
(with a ram block named "/object/pc.ram")
As we migrate from vhost-kernel to vhost-user, it failes as:
Unknown ramblock "pc.ram", cannot accept migration
error while loading state for instance 0x0 of device 'ram'
load of migration failed: Invalid argument
Here are some options to fix this:
1. When we do ram name comparison, we truncate the prefix as this patch shows.
It cannot cover the corner case: the source VM could have two ram blocks
with name of "pc.ram" and "/object/pc.ram".
2. We add an alias name to RAMBlock; when we do name comparison, not only
idstr is compared, but also compared to the alias. But this will add more
complexity to upper layer stack OpenStack/libvirt.
Any thoughts?
Cc: Jason Wang <jasow...@redhat.com>
Cc: Michael S. Tsirkin <m...@redhat.com>
Cc: Maxime Coquelin <maxime.coque...@redhat.com>
Cc: Paolo Bonzini <pbonz...@redhat.com>
Suggested-by: Michael S. Tsirkin <m...@redhat.com>
Signed-off-by: Jianfeng Tan <jianfeng....@intel.com>
---
exec.c | 13 ++++++++++++-
1 file changed, 12 insertions(+), 1 deletion(-)
diff --git a/exec.c b/exec.c
index 4722e52..d294e5c 100644
--- a/exec.c
+++ b/exec.c
@@ -2334,13 +2334,24 @@ found:
RAMBlock *qemu_ram_block_by_name(const char *name)
{
RAMBlock *block;
+ char *name1, *id1;
+ char *name2, *id2;
+
+ name1 = strdup(name);
+ id1 = basename(name1);
RAMBLOCK_FOREACH(block) {
- if (!strcmp(name, block->idstr)) {
+ name2 = strdup(block->idstr);
+ id2 = basename(name2);
+ if (!strcmp(id1, id2)) {
+ free(name1);
+ free(name2);
return block;
}
+ free(name2);
}
+ free(name1);
return NULL;
}
--
2.7.4
