[pve-devel] [PATCH qemu-server] Fix races with suspended VMs that can wake up

[Filip Schauer <f.schauer@proxmox.com>]

Fix races with ACPI-suspended VMs which could wake up during migration
or during a suspend-mode backup.

Revert prevention, of ACPI-suspended VMs automatically resuming after
migration, introduced by 7ba974a6828d. The commit introduced a potential
problem that causes a suspended VM that wakes up during migration to
remain paused after the migration finishes.

Furthermore the commit increased the race window during the preparation
of a suspend-mode backup, when a suspended VM wakes up between the
vm_is_paused check in PVE::VZDump::QemuServer::prepare and
PVE::VZDump::QemuServer::qga_fs_freeze. This causes the code to skip
fs-freeze even if the VM has woken up, potentially leaving the file
system in an inconsistent state.

To prevent this, do not treat the suspended runstate as paused when
migrating or archiving a VM.

Signed-off-by: Filip Schauer <f.schauer@proxmox.com>
---
 PVE/API2/Qemu.pm         | 4 ++--
 PVE/QemuMigrate.pm       | 4 +++-
 PVE/QemuServer.pm        | 6 +++---
 PVE/VZDump/QemuServer.pm | 4 +++-
 4 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/PVE/API2/Qemu.pm b/PVE/API2/Qemu.pm
index c8a87f3..9b1d3be 100644
--- a/PVE/API2/Qemu.pm
+++ b/PVE/API2/Qemu.pm
@@ -3083,7 +3083,7 @@ __PACKAGE__->register_method({
 
 	# sending a graceful shutdown command to paused VMs runs into timeouts, and even worse, when
 	# the VM gets resumed later, it still gets the request delivered and powers off
-	if (PVE::QemuServer::vm_is_paused($vmid)) {
+	if (PVE::QemuServer::vm_is_paused($vmid, 1)) {
 	    if ($param->{forceStop}) {
 		warn "VM is paused - stop instead of shutdown\n";
 		$shutdown = 0;
@@ -3159,7 +3159,7 @@ __PACKAGE__->register_method({
 	my $node = extract_param($param, 'node');
 	my $vmid = extract_param($param, 'vmid');
 
-	die "VM is paused - cannot shutdown\n" if PVE::QemuServer::vm_is_paused($vmid);
+	die "VM is paused - cannot shutdown\n" if PVE::QemuServer::vm_is_paused($vmid, 1);
 
 	die "VM $vmid not running\n" if !PVE::QemuServer::check_running($vmid);
 
diff --git a/PVE/QemuMigrate.pm b/PVE/QemuMigrate.pm
index f41c61f..111eeb0 100644
--- a/PVE/QemuMigrate.pm
+++ b/PVE/QemuMigrate.pm
@@ -224,7 +224,9 @@ sub prepare {
 	    }
 	}
 
-	$self->{vm_was_paused} = 1 if PVE::QemuServer::vm_is_paused($vmid);
+	# Do not treat a suspended VM as paused, as it might wake up
+	# during migration and remain paused after migration finishes.
+	$self->{vm_was_paused} = 1 if PVE::QemuServer::vm_is_paused($vmid, 0);
     }
 
     my ($loc_res, $mapped_res, $missing_mappings_by_node) = PVE::QemuServer::check_local_resources($conf, 1);
diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
index 2895675..2cd8948 100644
--- a/PVE/QemuServer.pm
+++ b/PVE/QemuServer.pm
@@ -8497,7 +8497,7 @@ sub complete_migration_storage {
 }
 
 sub vm_is_paused {
-    my ($vmid) = @_;
+    my ($vmid, $include_suspended) = @_;
     my $qmpstatus = eval {
 	PVE::QemuConfig::assert_config_exists_on_node($vmid);
 	mon_cmd($vmid, "query-status");
@@ -8505,8 +8505,8 @@ sub vm_is_paused {
     warn "$@\n" if $@;
     return $qmpstatus && (
 	$qmpstatus->{status} eq "paused" ||
-	$qmpstatus->{status} eq "suspended" ||
-	$qmpstatus->{status} eq "prelaunch"
+	$qmpstatus->{status} eq "prelaunch" ||
+	($include_suspended && $qmpstatus->{status} eq "suspended")
     );
 }
 
diff --git a/PVE/VZDump/QemuServer.pm b/PVE/VZDump/QemuServer.pm
index b38d7e7..f811cbf 100644
--- a/PVE/VZDump/QemuServer.pm
+++ b/PVE/VZDump/QemuServer.pm
@@ -67,7 +67,9 @@ sub prepare {
     $self->{vm_was_paused} = 0;
     if (!PVE::QemuServer::check_running($vmid)) {
 	$self->{vm_was_running} = 0;
-    } elsif (PVE::QemuServer::vm_is_paused($vmid)) {
+    } elsif (PVE::QemuServer::vm_is_paused($vmid, 0)) {
+	# Do not treat a suspended VM as paused, as it would cause us to skip
+	# fs-freeze even if the VM wakes up before we reach qga_fs_freeze.
 	$self->{vm_was_paused} = 1;
     }
 
-- 
2.39.2