On sexta-feira, 4 de março de 2016 09:58:53 PST Marc Mutz wrote: > On Friday 04 March 2016 07:52:15 Thiago Macieira wrote: > > Found in GCC 6's changelog (http://gcc.gnu.org/gcc-6/changes.html): > > > Value range propagation now assumes that the this pointer of C++ member > > > functions is non-null. This eliminates common null pointer checks but > > > also breaks some non-conforming code-bases (such as Qt-5, Chromium, > > > KDevelop). As a temporary work-around -fno-delete-null-pointer-checks > > > can be used. Wrong code can be identified by using -fsanitize=undefined. > > > > Are we free of such mistakes? Or do we need to enable -fno-delete-null- > > pointer-checks? > > This comes to mind; > > QPointer<QObject> that = this; > // ... > if (!that) return; > > Not sure it's affected, though.
Shouldn't be because that !that call isn't checking the this pointer, but instead is checking the strong refcount inside the QSharedPointer. > Don't know off-handedly anything else that where this == nullptr would be > valid. Only way to find out is to enable -fno-delete-null-pointer-checks > except for -developer-build and trying to fix the ubsan issues. It's not valid. The this pointer is never null, according to the standard. And yet we had code like that in V4 (qtdeclarative). Clang started complaining about it and I tried to fix it by just removing the checks, but they were apparently in use. Since I don't remember seeing that warning anymore, it's likely to be fixed. But I am asking to be sure. > /me upgrades to GCC 6. I might still have a patch or two that aren't merged that fix GCC 6 warnings. But when you compile with it, can you make sure it doesn't print any misleading-identation warnings? There were a couple of false positives that I reported and GCC fixed. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center _______________________________________________ Development mailing list [email protected] http://lists.qt-project.org/mailman/listinfo/development
