Re: [libvirt] [PATCH] qemuBuildMemoryBackendStr: Handle one more corner case

On Tue, Aug 08, 2017 at 05:04:15PM +0200, Michal Privoznik wrote: s/to not specify/not to specify/ reads a bit nicely, or even s/to not specify any/to specify no/ or "allowed to omit the page size" Choice is yours Reviewed-by: Martin Kletzander <mkletzan(a)redhat.com&gt; One more note that I told you in person, but I can't resist the urge to share it more widely. The code is gross because it can't be made much better due to the fact that we keep the parsed data the same way they are in the XML. I think this could be made nicer if we were to fill all the information during parsing. When parsing /memory/hugepages we would fill in missing information right away. Then, when parsing /cpu/numa/cell we could copy that information to the specific cells (if missing) and because the whole definition is already available from parsing /memory/hugepages there's less lookups. Finally, when parsing /devices/memory/source we can directly get missing info from the numa cell if nodemask is present or global hugepages if it is not, and we don't have to lookup anything, because it is already guaranteed to be filled in for the previous parts. I hope it's understandable. What I hope even more is that someone refactors the code this way =)