On 11.09.2017 19:20, Eric Blake wrote: > We already have another qtest_init() in the tree (namely, as part > of the device initialization of the qtest device at the top level > qtest.c), with a different signature; having two different > qtest_init() is confusing, so an upcoming patch will consolidate > all testsuite callers onto a unified spelling. But the > consolidation is easier if qtest_start() is further down the call > chain, rather than an intermediate wrapper, so swap the call chain > around. This includes renaming qtest_init_without_qmp_handshake() > into qtest_start_without_qmp_handshake(), as it remains the lowest > point in the call stack. > > Note that qtest_init() now asserts that global_qtest was not set > on entry, and clears it on exit, to preserve the behavior of > existing tests that assert the same (and also proving that we > fixed all tests that had parallel connections); but later patches > will eventually simplify things by getting rid of qtest_init() > and global_qtest altogether. > > Signed-off-by: Eric Blake <ebl...@redhat.com>
Reviewed-by: Thomas Huth <th...@redhat.com>