On Mon, Mar 30, 2015 at 04:53:38PM +0100, Thomas Wood wrote: > diff --git a/tests/eviction_common.c b/tests/eviction_common.c > index 4fa5c04..9e06cbb 100644 > --- a/tests/eviction_common.c > +++ b/tests/eviction_common.c > @@ -200,18 +200,19 @@ static void mlocked_evictions(int fd, struct > igt_eviction_test_ops *ops, > ops->close(fd, bo[n]); > > out: > - write(result[1], &ret, sizeof(ret)); > + igt_assert_eq(write(result[1], &ret, sizeof(ret)), > + sizeof(ret)); > } > > igt_waitchildren(); > > fcntl(result[0], F_SETFL, fcntl(result[0], F_GETFL) | O_NONBLOCK); > - read(result[0], &ret, sizeof(ret)); > + igt_assert_eq(read(result[0], &ret, sizeof(ret)), sizeof(ret));
No. This read() may return -1 (with ret then being == -1 due to the earlier initialisation) if the child is killed by a sigbus.... Hah, and now I realise my overcomplication, if we just get the status back via igt_waitchildren passing the result back via a pipe is silly. -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
