https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60078
--- Comment #15 from Eric Botcazou <ebotcazou at gcc dot gnu.org> --- > that happened only once. and the problem did never ever repeat. > But my gut feeling is still that there is a race conditition. Yes, I agree that the usage of Side_Effect_Finger looks suspicious here. > However I have been recently working on TSAN a bit, and I have an > experimental tsan version that I played with. Initially > it ICE'd on Ada's VIEW_CONVERT_EXPRs, but when I fixed that I got: "data races" (unprotected concurrent accesses) are allowed in Ada if the object is declared as atomic, in which case GNAT generates synchronization directives.
