On Tue, Jun 21, 2016 at 12:17:48PM +0000, Wang, Zhi A wrote: > To me, atexit is not reliable. As if something wrong happened during the test > and an unexpected signal was received, the prog would be aborted. atexit() > would not work on that case. Better do failsafe work in parent process and > run the test case in child process I think.
Welcome to igt. atexit here handles signals, but we don't yet have universal setup/cleanup. Just a wishlist. -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
