The EFI disk can only be exclusively accessed during cold start, so
skip check_efi_vars() for migration and resume from hibernation.
Also, check_efi_vars() might write the VM configuration, which also
cannot be done in the context of migration at this stage, because the
configuration does not exist on the target yet and it would result in:
> close (rename) atomic file '/etc/pve/nodes/squid176/qemu-server/106.conf'
> failed: File exists
Fixes: c5b3a314 ("fix #6985: ovmf: auto-enroll Microsoft UEFI CA 2023 for
Windows")
Reported-by: Friedrich Weber <[email protected]>
Suggested-by: Thomas Lamprecht <[email protected]>
Signed-off-by: Fiona Ebner <[email protected]>
---
src/PVE/QemuServer.pm | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/src/PVE/QemuServer.pm b/src/PVE/QemuServer.pm
index 9946e445..b39157af 100644
--- a/src/PVE/QemuServer.pm
+++ b/src/PVE/QemuServer.pm
@@ -5610,7 +5610,11 @@ sub vm_start_nolock {
my $storage_hints = generate_storage_hints($conf, 1);
PVE::Storage::activate_volumes($storecfg, $vollist, undef,
$storage_hints);
- check_efi_vars($storecfg, $vmid, $conf) if $conf->{bios} &&
$conf->{bios} eq 'ovmf';
+ # Can only exclusively access EFI disk during cold start. Also,
check_efi_vars() might write
+ # the configuration, which must not be done at this stage of migration
on the target.
+ if (!$statefile && !$resume && $conf->{bios} && $conf->{bios} eq
'ovmf') {
+ check_efi_vars($storecfg, $vmid, $conf);
+ }
# Note that for certain cases like templates, the configuration is
minimized, so need to ensure
# the rest of the function here uses the same configuration that was
used to build the command
--
2.47.3
_______________________________________________
pve-devel mailing list
[email protected]
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
