On 8/20/26 18:38, Roman Bogorodskiy wrote:
> I was doing pre-merge testing of the
>
> "bhyve: implement virDomainSetLifecycleAction() API"
>
> series and spotted a few issues related to the live/transient
> domain definitions handling (and in the original series as well, so
> will likely send a v2 later).
>
> This area has quite a few nuances, so I extended the TCK
> test:
>
> https://gitlab.com/libvirt/libvirt-tck/-/merge_requests/92
>
> TCK also helped to catch the issue with the firmware data filling
> with the scripts/domain/405-ovmf-nvram-efi.t test. It is fixed in the
> final patch of the series.
>
> I think I plan to add more TCK tests for that, but decided to send
> this series first to make sure I'm on the right track.
>
> Roman Bogorodskiy (4):
> bhyve: process: do not leak live changes to persistent XML
> bhyve: process: discard stale live definition on reconnect failure
> bhyve: honor VIR_DOMAIN_XML_INACTIVE in virDomainGetXMLDesc()
> bhyve: domain: fix filling of firmware data
>
> src/bhyve/bhyve_domain.c | 11 ++++++++++-
> src/bhyve/bhyve_driver.c | 18 ++++++++++++++----
> src/bhyve/bhyve_firmware.c | 12 ++++++------
> src/bhyve/bhyve_firmware.h | 2 +-
> src/bhyve/bhyve_process.c | 9 ++++++++-
> .../three_firmwares/BHYVE_UEFI_VARS.fd | 0
> .../x86_64/bhyvexml2xmlout-numa.xml | 2 ++
> tests/bhyvexml2xmltest.c | 10 +++++++++-
> 8 files changed, 50 insertions(+), 14 deletions(-)
> create mode 100644 tests/bhyvefirmwaredata/three_firmwares/BHYVE_UEFI_VARS.fd
>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Michal