On Fri Oct 16, 2026 at 9:00 AM CEST, Michal Privoznik via Devel wrote:
From: Michal Privoznik <mprivozn@redhat.com>
Inside of qemuxml2argvmock.c there's an alternative implementation of qemuInterfaceOpenVhostNet(). Fair enough. But the original function (living in qemu_interface.c) has a check which exits early from the function. This check is lacking in the mock which then leads qemuxmlconftest generate a different cmd line than the production code.
The is particularly important, when the check is changed and we
s/The/The check/ ??
want to verify through tests that it doesn't break anything.
Well, ideally we should extract the part of the function that opens the vhostnet_path and mock that which would lead to more deduplication and it would also be more error-prone in case any future change is done. But that's not a hard blocker.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> --- tests/qemuxml2argvmock.c | 7 ++++++- tests/qemuxmlconfdata/downscript.x86_64-latest.args | 2 +- tests/qemuxmlconfdata/net-coalesce.x86_64-latest.args | 6 +++--- .../net-eth-unmanaged-tap.x86_64-latest.args | 2 +- tests/qemuxmlconfdata/net-isolated-port.x86_64-latest.args | 2 +- tests/qemuxmlconfdata/net-midonet.x86_64-latest.args | 2 +- tests/qemuxmlconfdata/net-mtu.x86_64-latest.args | 4 ++-- .../net-virtio-network-portgroup.x86_64-latest.args | 2 +- .../net-virtio-teaming-network.x86_64-latest.args | 2 +- tests/qemuxmlconfdata/net-virtio-vhost.x86_64-latest.args | 6 +++--- .../qemuxmlconfdata/tap-vhost-incorrect.x86_64-latest.args | 2 +- tests/qemuxmlconfdata/tap-vhost.x86_64-latest.args | 2 +- 12 files changed, 22 insertions(+), 17 deletions(-)
diff --git a/tests/qemuxml2argvmock.c b/tests/qemuxml2argvmock.c index 4f47b2598e..7b211a191d 100644 --- a/tests/qemuxml2argvmock.c +++ b/tests/qemuxml2argvmock.c @@ -222,13 +222,18 @@ virNetDevOpenvswitchGetVhostuserIfname(const char *path G_GNUC_UNUSED, }
int -qemuInterfaceOpenVhostNet(virDomainObj *vm G_GNUC_UNUSED, +qemuInterfaceOpenVhostNet(virDomainObj *vm, virDomainNetDef *net) { qemuDomainNetworkPrivate *netpriv = QEMU_DOMAIN_NETWORK_PRIVATE(net); size_t vhostfdSize = net->driver.virtio.queues; size_t i;
+ if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM || + net->driver.virtio.name == VIR_DOMAIN_NET_DRIVER_TYPE_QEMU) { + return 0; + } + if (!vhostfdSize) vhostfdSize = 1;
One nitpick, just to make it "look" like the mocked function, I'd reorder these two conditions. But since the changes (apart from the commit message fix) are optionals, then Reviewed-by: Martin Kletzander <mkletzan@redhat.com> with the commit message fixed.