The previous commit 06f81c63ebc19cb0e51f9b397991f6d6ae56d090 masked
a flaw with handling of errors for NFS disk labels. If you had multiple
devices that required labelling, and an disk on root squashing NFS, then
when the failure occurred for the NFS file, all further devices would
be skipped. The caller would then ignore the error and try to launch
the guest anyway, unless there was a stdin_path on NFS. Since NFS
error handling was not dealt with in the correct place, this in turn
meant that disk hotplug failed with root squashing NFS. All in all a
bit of a mess. This tries to clean up the code handling errors
closer to the point of occurrance
libvirt_private.syms | 1 +
qemu/qemu_driver.c | 3 +--
qemu/qemu_monitor_text.c | 28 +++++++++++++++++++---------
security/security_selinux.c | 11 +++++++----
util/storage_file.c | 32 +++++++++++++++++++++++++-------
util/storage_file.h | 9 +++++++++
6 files changed, 62 insertions(+), 22 deletions(-)
Daniel