On 08/08/2017 10:16 AM, Kevin Wolf wrote: >>>> is sleep for ms portable? >>> Sadly, sub-second sleep is a GNU coreutils feature; I suspect the BSD >>> machines may fail to parse it. (Of course, we could do some sort of >>> 'sleep $SMALL', where $SMALL is 0.5 if sleep supports it, and 1 otherwise). >>> >> sleep for 1 second may lead to more then one request done before qemu quite. > > _supported_os Linux > > So do we really care about portability? And are all the other test cases > working on the BSDs?
You've got valid points there - I suspect that a run of qemu-iotests on BSD would turn up quite a few failures, some of which would be easy to address; but it's not on my personal priority list. I'm fine with an approach of committing something that works where it was first tested, then adding followup patches to improve portability, especially as long as we are not currently getting anyone complaining about qemu-iotests failures on BSD. -- Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3266 Virtualization: qemu.org | libvirt.org
Description: OpenPGP digital signature