Paolo Bonzini <pbonz...@redhat.com> writes: > \>> It seems like Coverity Scan doesn't like the new model. Possibly the >>> fault of the third patch. >> >> Works for me with a local 7.0.3 installation. Which I just realized is >> outdated. I'll recheck with 7.6.0. >> >>> Will check (for now I'm still running scans >>> with the old model). >> >> Appreciated. > > Yes, the last patch seems to be the culprit. It failed immediately with > the __coverity_mark_as_afm_freed__(ptr, "g_free") model;
Any diagnostics? Works for me locally with 7.6.0, and matches the reference manual's examples for the ALLOC_FREE_MISMATCH checker. > it's been > running for a while without it even though it's not done yet. > > Perhaps we can do something like this: > > /* Some helpful comment here. */ > #if 1 > #define AFM_g_free AFM_free > #else > #define AFM_g_free "g_free" > #endif Yes. If you want me to write the helpful comment, I need to know what exactly goes wrong. The alternative is to revert the patch for now. Aside: would be nice if Coverity pre-defined a symbol when ALLOC_FREE_MISMATCH is enabled, it would let me enable support for the checker exactly when it's wanted.
