On Mon, 08/03 20:45, Max Reitz wrote:
> >+static void test_single_job(int expected)
> >+{
> >+ BlockJob *job;
> >+ BlockJobTxn *txn;
> >+ int result = -EINPROGRESS;
> >+
> >+ txn = block_job_txn_new();
> >+ job = test_block_job_start(1, true, expected, &result);
>
> I think I'd like it better if 0 was passed if expected == -ECANCELED. This
> way, we don't know whether the test really was cancelled or whether it just
> exited normally, returning -ECANCELED because that was what was passed here.
The job will yield right after starting, so I think it is OK.
> >+ /* Now make job2 finish before the main loop kicks jobs. This means
> >+ * simulates the race between a pending kick and another job completing.
>
> Perhaps the "means" should be removed?
Yes.
>
> Even if you don't decide to address any of my comments:
>
> Reviewed-by: Max Reitz <mre...@redhat.com>
Thanks,
Fam
