[pve-devel] [PATCH qemu-server v3 3/3] fix #3258: block vm start when pci device is already in use

[Dominik Csapak]

on vm start, we reserve all pciids that we use, and
remove the reservation again in vm_stop_cleanup

first with only a time-based reservation but after the vm is started,
we reserve again but with the pid.

for this, we have to move the start_timeout calculation above the
hostpci handling.

also moved the pci initialization out of the conf parsing loop
so that we can reserve all ids before we actually touch any of them

while touching the lines, fix the indentation

this way, when a vm starts with a pci device that is already configured
for a different running vm, will not be started and the user gets
the error that the device is already in use

Signed-off-by: Dominik Csapak <d.csapak at proxmox.com>
---
 PVE/QemuServer.pm | 50 +++++++++++++++++++++++++++++++++++++++--------
 1 file changed, 42 insertions(+), 8 deletions(-)

diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
index f78b2cc..e504e9a 100644
--- a/PVE/QemuServer.pm
+++ b/PVE/QemuServer.pm
@@ -5381,16 +5381,40 @@ sub vm_start_nolock {
 	push @$cmd, '-S';
     }
 
+    my $start_timeout = $params->{timeout} // config_aware_timeout($conf, $resume);
+    my $pciids = [];
+    my $pci_devices = {};
+
     # host pci devices
     for (my $i = 0; $i < $PVE::QemuServer::PCI::MAX_HOSTPCI_DEVICES; $i++)  {
-      my $d = parse_hostpci($conf->{"hostpci$i"});
-      next if !$d;
-      my $pcidevices = $d->{pciid};
-      foreach my $pcidevice (@$pcidevices) {
-	    my $pciid = $pcidevice->{id};
+	my $d = parse_hostpci($conf->{"hostpci$i"});
+	next if !$d;
+	$pci_devices->{$i} = $d;
 
-	    PVE::QemuServer::PCI::prepare_pci_device($vmid, $pciid, $i, $d->{mdev});
-      }
+	my $pcidevices = $d->{pciid};
+
+	my $ids = [map { $_->{id} } @$pcidevices];
+	push @$pciids, @$ids;
+    }
+
+    # reserve all pci ids before actually doing anything with them
+    PVE::QemuServer::PCI::reserve_pci_usage($pciids, $vmid, $start_timeout);
+
+    eval {
+	for my $i (sort keys %$pci_devices) {
+	    my $d = $pci_devices->{$i};
+	    my $pcidevices = $d->{pciid};
+	    foreach my $pcidevice (@$pcidevices) {
+		my $pciid = $pcidevice->{id};
+		PVE::QemuServer::PCI::prepare_pci_device($vmid, $pciid, $i, $d->{mdev});
+	    }
+	}
+    };
+
+    if (my $err = $@) {
+	eval { PVE::QemuServer::PCI::remove_pci_reservation($pciids) };
+	warn $@ if $@;
+	die $err;
     }
 
     PVE::Storage::activate_volumes($storecfg, $vollist);
@@ -5405,7 +5429,6 @@ sub vm_start_nolock {
 
     my $cpuunits = get_cpuunits($conf);
 
-    my $start_timeout = $params->{timeout} // config_aware_timeout($conf, $resume);
     my %run_params = (
 	timeout => $statefile ? undef : $start_timeout,
 	umask => 0077,
@@ -5485,9 +5508,17 @@ sub vm_start_nolock {
     if (my $err = $@) {
 	# deactivate volumes if start fails
 	eval { PVE::Storage::deactivate_volumes($storecfg, $vollist); };
+	eval { PVE::QemuServer::PCI::remove_pci_reservation($pciids) };
+
 	die "start failed: $err";
     }
 
+    # reserve all pciids again with the pid
+    # the vm is already started, we can only warn on error here
+    my $pid = PVE::QemuServer::Helpers::vm_running_locally($vmid);
+    eval { PVE::QemuServer::PCI::reserve_pci_usage($pciids, $vmid, undef, $pid) };
+    warn $@ if $@;
+
     print "migration listens on $migrate_uri\n" if $migrate_uri;
     $res->{migrate_uri} = $migrate_uri;
 
@@ -5676,6 +5707,7 @@ sub vm_stop_cleanup {
 	    unlink '/dev/shm/pve-shm-' . ($ivshmem->{name} // $vmid);
 	}
 
+	my $ids = [];
 	foreach my $key (keys %$conf) {
 	    next if $key !~ m/^hostpci(\d+)$/;
 	    my $hostpciindex = $1;
@@ -5684,9 +5716,11 @@ sub vm_stop_cleanup {
 
 	    foreach my $pci (@{$d->{pciid}}) {
 		my $pciid = $pci->{id};
+		push @$ids, $pci->{id};
 		PVE::SysFSTools::pci_cleanup_mdev_device($pciid, $uuid);
 	    }
 	}
+	PVE::QemuServer::PCI::remove_pci_reservation($ids);
 
 	vmconfig_apply_pending($vmid, $conf, $storecfg) if $apply_pending_changes;
     };
-- 
2.30.2