On 08/20/2012 07:49 AM, Daniel P. Berrange wrote:
From: "Daniel P. Berrange" <berrange@redhat.com>
Take advantage of the previously added monitor helpers to create a test suite for the QEMU JSON monitor impl. As a proof of concept, this tests the 'qemuMonitorGetStatus' implementation --- .gitignore | 1 + tests/Makefile.am | 11 +++- tests/qemumonitorjsontest.c | 150 ++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 161 insertions(+), 1 deletion(-) create mode 100644 tests/qemumonitorjsontest.c
+ if (!test) + return -1; + + if (qemuMonitorTestAddItem(test, "query-status", + "{ " + " \"return\": { " + " \"status\": \"running\", " + " \"singlestep\": false, " + " \"running\": true " + " } " + "}") < 0) + goto cleanup; + if (qemuMonitorTestAddItem(test, "query-status", + "{ " + " \"return\": { " + " \"singlestep\": false, " + " \"running\": false " + " } " + "}") < 0) + goto cleanup;
So the trick is to batch up several replies (even if the same command will be queried more than once),
+ if (qemuMonitorGetStatus(qemuMonitorTestGetMonitor(test), + &running, &reason) < 0) + goto cleanup;
then as each call triggers a monitor command, our batched replies start getting fed back, and we can see that the rest of the qemu driver handled the reply as expected. Looks good; should be easy enough to copy to add more tests of a similar nature. ACK. -- Eric Blake eblake@redhat.com +1-919-301-3266 Libvirt virtualization library http://libvirt.org