On 23/07/2019 15.01, Peter Maydell wrote: > On Tue, 23 Jul 2019 at 13:58, Alex Bennée <alex.ben...@linaro.org> wrote: >> >> >> Peter Maydell <peter.mayd...@linaro.org> writes: >> >>> On Tue, 23 Jul 2019 at 13:16, Philippe Mathieu-Daudé <phi...@redhat.com> >>> wrote: >>>> >>>> On 7/23/19 1:46 PM, Alex Bennée wrote: >>>>>> I see Thomas Huth has a patch on-list for that, but this >>>>>> didn't manifest as a problem before this pullreq. >>>>> >>>>> OK, I'll add it and rebuild the PR. >>>> >>>> But Thomas got another error later... >>> >>> If we're not sure what all the needed fixes are we >>> should probably just drop the change that starts >>> running the iotests under 'make check'. Otherwise this >>> pullreq is going to miss rc2, and it's too big to go into rc3.
I think OpenBSD was the final thing that is not working. I've checked freebsd and macOS and they were working fine for me there. The netbsd image does not have "bash", so the tests are skipped there. >> I'm just double checking now - it does seem OpenBSD is very slow even >> with 18 cores assigned. > > We should avoid enabling "very slow" tests in make check too... It's not the iotests that are slow here, it's the whole openbsd image - even compiling is way slower than natively... it feels like it is ignoring the multiple cores and only running single-threaded. Thomas