forked from pool/libvirt
88 lines
3.8 KiB
Diff
88 lines
3.8 KiB
Diff
|
commit 2552752f0b6504a80f6306e5aae2c7063d24f1ab
|
||
|
|
Author: Jim Fehlig <jfehlig@suse.com>
|
||
|
|
Date: Mon Oct 14 14:01:00 2019 -0600
|
||
|
|
|
||
|
|
libxl: Fix lock manager lock ordering
|
||
|
|
|
||
|
|
The ordering of lock manager locks in the libxl driver has a flaw that was
|
||
|
|
uncovered by a migration error path. In the perform phase of migration, the
|
||
|
|
source host calls virDomainLockProcessPause to release the lock before
|
||
|
|
sending the VM to the destination host. If the send fails an attempt is made
|
||
|
|
to reacquire the lock with virDomainLockProcessResume, but that too can fail
|
||
|
|
if the destination host has not finished cleaning up the failed VM and
|
||
|
|
releasing the lock it acquired when starting to receive the VM.
|
||
|
|
|
||
|
|
This change delays calling virDomainLockProcessResume in libxlDomainStart
|
||
|
|
until the VM is successfully created, but before it is unpaused. A similar
|
||
|
|
approach is used by the qemu driver, avoiding the need to release the lock
|
||
|
|
if VM creation fails. In the migration perform phase, releasing the lock
|
||
|
|
with virDomainLockProcessPause is delayed until the VM is successfully
|
||
|
|
sent to the destination, which avoids reacquiring the lock if the send
|
||
|
|
fails.
|
||
|
|
|
||
|
|
Signed-off-by: Jim Fehlig <jfehlig@suse.com>
|
||
|
|
Reviewed-by: Cole Robinson <crobinso@redhat.com>
|
||
|
|
|
||
|
|
Index: libvirt-5.9.0/src/libxl/libxl_domain.c
|
||
|
|
===================================================================
|
||
|
|
--- libvirt-5.9.0.orig/src/libxl/libxl_domain.c
|
||
|
|
+++ libvirt-5.9.0/src/libxl/libxl_domain.c
|
||
|
|
@@ -1347,13 +1347,6 @@ libxlDomainStart(libxlDriverPrivatePtr d
|
||
|
|
NULL) < 0)
|
||
|
|
goto cleanup;
|
||
|
|
|
||
|
|
- if (virDomainLockProcessResume(driver->lockManager,
|
||
|
|
- "xen:///system",
|
||
|
|
- vm,
|
||
|
|
- priv->lockState) < 0)
|
||
|
|
- goto cleanup;
|
||
|
|
- VIR_FREE(priv->lockState);
|
||
|
|
-
|
||
|
|
if (libxlNetworkPrepareDevices(vm->def) < 0)
|
||
|
|
goto cleanup_dom;
|
||
|
|
|
||
|
|
@@ -1436,6 +1429,13 @@ libxlDomainStart(libxlDriverPrivatePtr d
|
||
|
|
|
||
|
|
libxlLoggerOpenFile(cfg->logger, domid, vm->def->name, config_json);
|
||
|
|
|
||
|
|
+ if (virDomainLockProcessResume(driver->lockManager,
|
||
|
|
+ "xen:///system",
|
||
|
|
+ vm,
|
||
|
|
+ priv->lockState) < 0)
|
||
|
|
+ goto destroy_dom;
|
||
|
|
+ VIR_FREE(priv->lockState);
|
||
|
|
+
|
||
|
|
/* Always enable domain death events */
|
||
|
|
if (libxl_evenable_domain_death(cfg->ctx, vm->def->id, 0, &priv->deathW))
|
||
|
|
goto destroy_dom;
|
||
|
|
Index: libvirt-5.9.0/src/libxl/libxl_migration.c
|
||
|
|
===================================================================
|
||
|
|
--- libvirt-5.9.0.orig/src/libxl/libxl_migration.c
|
||
|
|
+++ libvirt-5.9.0/src/libxl/libxl_migration.c
|
||
|
|
@@ -1240,20 +1240,16 @@ libxlDomainMigrationSrcPerform(libxlDriv
|
||
|
|
sockfd = virNetSocketDupFD(sock, true);
|
||
|
|
virObjectUnref(sock);
|
||
|
|
|
||
|
|
- if (virDomainLockProcessPause(driver->lockManager, vm, &priv->lockState) < 0)
|
||
|
|
- VIR_WARN("Unable to release lease on %s", vm->def->name);
|
||
|
|
- VIR_DEBUG("Preserving lock state '%s'", NULLSTR(priv->lockState));
|
||
|
|
-
|
||
|
|
/* suspend vm and send saved data to dst through socket fd */
|
||
|
|
virObjectUnlock(vm);
|
||
|
|
ret = libxlDoMigrateSrcSend(driver, vm, flags, sockfd);
|
||
|
|
virObjectLock(vm);
|
||
|
|
|
||
|
|
- if (ret < 0) {
|
||
|
|
- virDomainLockProcessResume(driver->lockManager,
|
||
|
|
- "xen:///system",
|
||
|
|
- vm,
|
||
|
|
- priv->lockState);
|
||
|
|
+ if (ret == 0) {
|
||
|
|
+ if (virDomainLockProcessPause(driver->lockManager, vm, &priv->lockState) < 0)
|
||
|
|
+ VIR_WARN("Unable to release lease on %s", vm->def->name);
|
||
|
|
+ VIR_DEBUG("Preserving lock state '%s'", NULLSTR(priv->lockState));
|
||
|
|
+ } else {
|
||
|
|
/*
|
||
|
|
* Confirm phase will not be executed if perform fails. End the
|
||
|
|
* job started in begin phase.
|