We could get -fwrapv back for our debug builds if we switched to GCC 8.x (see: #27597 (moved)). As a data point: I've been compiling (and using) some custom Tor Browser with GCC 8.x for a while now and did not run into any issues. Thus, we might consider 8.x sufficiently stable to switch to it in particular when preparing for eser68.
I think it's good to be conservative here and not switching to GCC 9.1 right away. It seems using GCC 8.3.0 is a decent solution for our Tor Browser 9. bug_25930_v3(https://gitweb.torproject.org/user/gk/tor-browser-build.git/commit/?h=bug_25930_v3&id=e8bf1330d8e5e5ccc48d69d47ab424ce746c30ee) has the proposed fix. Note: I am leaving mingw-w64/gcc untouched for now. All things seem to build fine, just our spec hack for Firefox is breaking. But that should get fixed by mingw-w64/clang anyway, so we bump the GCC part for mingw-w64/gcc after we switched Firefox to mingw-w64/clang (#28238 (moved)).
Trac: Status: new to needs_review Summary: Update gcc to 7.X to Update gcc to 8.X Keywords: N/Adeleted, GeorgKoppen201905, TorBrowserTeam201905R added