Re: [libvirt PATCH 0/3] fix ATTRIBUTE_NONNULL issues

On 3/11/20 8:34 AM, Pavel Hrdina wrote: > On Wed, Mar 11, 2020 at 01:25:50PM +0100, Ján Tomko wrote: >> Missing blurb in the cover letter. >> >> On a Wednesday in 2020, Pavel Hrdina wrote: >>> Pavel Hrdina (3): >>> domain_conf: fix ATTRIBUTE_NONNULL for virDomainDefFormat >>> libvirt_internal: fix ATTRIBUTE_NONNULL for virStateInitialize >>> vircommand: fix ATTRIBUTE_NONNULL for virCommandAddArg* >>> >> >> Do these actually help Coverity do its job or all they're good for >> is finding out where we failed to update the attributes? > > Now that I'm trying to setup coverity scans using scan.coverity.com they > actually help coverity do its job and I managed to run into these issues > while building libvirt using coverity. > > Pavel > Good luck with Coverity issues - if you'd like my list of work-around hacks I don't mind sending you the list of 55 or so local patches in my build environment. There's even more with more recent coverity releases, but I haven't taken the time to work through them.

These types of build issues can also be seen by enabling STATIC_ANALYSIS for a build - IOW: outside of coverity... FWIW: I've tried this for vircommand.h a while ago, see: https://www.redhat.com/archives/libvir-list/2018-December/msg00548.html The only other one I had in my backlog of hacks around things that in general the community hasn't accepted historically is for virnetdevtap.h and the virNetDevTapReattachBridge prototype for argument 2 because in the source code the @brname is used in a STREQ_NULLABLE thus it also gets flagged.