https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85574
--- Comment #8 from Richard Biener <rguenth at gcc dot gnu.org> --- To answer myself, no, doing LTO bootstrap twice doesn't get you stage-by-stage equal binaries. Looking at WPA IPA inline dumps (only have them from one build) obviously shows address differences (stage1 and stage2 lto1 are different enough). Besides that the cc1 WPA inline dumps are equal. Because I seem to be "lucky" and this instance of bootstrap ended up with equal binaries :/ Maybe exactly because of dumping. I guess a better chance might be bisecting this. I'll see if as expected the GCC 8 branchpoint is fine. Oh, and for the IPA inline dump I used STAGE1_CFLAGS="-O2", let me try without that. Which means I'll try that again to see if I'm unlucky again.
