On 5/3/19 1:04 PM, Andrea Bolognani wrote:
On Fri, 2019-05-03 at 10:46 +0200, Boris Fiuczynski wrote:
In addition adjusting *s390x-latest.args files to qemu deprecation changes made in commit e8c2c8bd078 (Prefer '-overcommit mem-lock' over -realtime mlock').
Signed-off-by: Boris Fiuczynski <fiuczy@linux.ibm.com> --- .../caps_3.1.0.s390x.replies | 20893 ++++++++++++++++ .../qemucapabilitiesdata/caps_3.1.0.s390x.xml | 2728 ++ ...othreads-virtio-scsi-ccw.s390x-latest.args | 2 +- .../s390x-ccw-graphics.s390x-latest.args | 2 +- .../s390x-ccw-headless.s390x-latest.args | 2 +- .../vhost-vsock-ccw-auto.s390x-latest.args | 2 +- .../vhost-vsock-ccw.s390x-latest.args | 2 +- 7 files changed, 23626 insertions(+), 5 deletions(-) create mode 100644 tests/qemucapabilitiesdata/caps_3.1.0.s390x.replies create mode 100644 tests/qemucapabilitiesdata/caps_3.1.0.s390x.xml
A few things.
First of all, while this message reached my inbox just fine, it didn't reach the list yet because of the sheer size, which is way over the configured limit and thus got stuck in the moderation, needing manual intervention to get through. In the future, please consider sending a snipped version to the mailing list and making the full changes available somewhere, like I've done for example with
https://www.redhat.com/archives/libvir-list/2019-April/msg01490.html
You also mention adjusting some .args files: I hope you haven't done so manually, but rather executed
VIR_TEST_REGENERATE_OUTPUT=1 make check
and let it fix the output files for you. If not, TYL ;) Thanks, I used VIR_TEST_REGENERATE_OUTPUT but not with make check.
As for the patch itself: while it looks fine, personally I don't feel we need to have capabilities for every single QEMU version on every single architecture, so considering that you're adding 4.0.0 in the same series and there aren't any 3.1.0-specific tests, I would say let's just skip this one version and add capabilitie for 4.0.0 only. Does that sound reasonable?
In general I kind of disagree because if the s390x capabilities for every qemu version would have exited the commits e8c2c8bd078 and a1dce96236f mentioned would not have missed the changes in the s390x related args files. In this case it is already history and therefor skipping 3.1.0 by moving the *.args files from the 3.1.0 patch 1 into 3 and removing patch 2 feels OK. -- Mit freundlichen Grüßen/Kind regards Boris Fiuczynski IBM Deutschland Research & Development GmbH Vorsitzender des Aufsichtsrats: Matthias Hartmann Geschäftsführung: Dirk Wittkopp Sitz der Gesellschaft: Böblingen Registergericht: Amtsgericht Stuttgart, HRB 243294