[PATCH qemu-server] fix #8251: migrate: nbd: align check for absolute path on target side

[PATCH qemu-server] fix #8251: migrate: nbd: align check for absolute path on target side

[Fiona Ebner]

A migration with both NBD storage migration and a passed-through
device (with shared=1, otherwise migration fails on the source side
already) would fail with "unable to parse volume ID 'XYZ'" on the
target side. The problematic call is parse_volume_id() in
vm_migrate_get_nbd_disks(). Without NBD storage migration, this is
not reached. Fix the bug by adding the same check for absolute paths
that scan_local_volumes() does on the source side.

Note that remote migration does not use vm_migrate_get_nbd_disks(),
but does its own NBD handling and absolute paths are limited to
'root@pam' only, so a configuration with absolute paths is denied for
the API token used for remote migration, when the volume access is
checked.

Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
---
 src/PVE/QemuServer.pm | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/src/PVE/QemuServer.pm b/src/PVE/QemuServer.pm
index 8149184e..9f8c7b13 100644
--- a/src/PVE/QemuServer.pm
+++ b/src/PVE/QemuServer.pm
@@ -5225,6 +5225,12 @@ sub vm_migrate_get_nbd_disks {
 
             return if !$volid;
 
+            # Matching the check in scan_local_volumes().
+            if ($volid =~ m|^/|) {
+                return if $drive->{shared};
+                die "$ds: unexpected non-shared local file/device '$volid'\n";
+            }
+
             my ($storeid, $volname) = PVE::Storage::parse_volume_id($volid);
 
             my $scfg = PVE::Storage::storage_config($storecfg, $storeid);
-- 
2.47.3