On Tue, Jun 23, 2020 at 01:58:15PM -0700, Nick Desaulniers wrote: > My point was more so about avoiding needless cc-option checks when > they're tautological.
No, you're right. The oldest gcc we support is 4.8 and that eats -msse2 just fine. > When Andy says "consider dropping the problematic GCC version" I > wonder if it was meant *just for this selftest* as I suggested, or > outright (which is untenable IMO, as it's a large jump to GCC 7.1+). Yeah, he's been schooled now :) > I'm not sure I agree, but I'll take feature detection any day over > version detection. Yeah, we had another issue recently with a funky compiler on some funky distro: https://lkml.kernel.org/r/caejqkgi3w%2bzvmkrbp4vtaewx1ujxrvnrq03mtrn_yh-pwoo...@mail.gmail.com and I have to say, I'm not at all happy about such subtle issues stemming from how a distro builds its compiler and that having an impact on the kernel build. But I guess this is the new reality now since we're getting more and more married to the compiler. > Guilty, your honor. :P > > Maybe the feature test should be copy+pasta'd to those other places in > the kernel, rather than the version check? "pasta'd"- mmhmhm, tasty noodles. Good idea for lunch tomorrow, thanks! :-) But sure, if you feel bored. Let it soak a while first, though, until all the builders and bots have seen it and haven't triggered moar funky build errors. Thx. -- Regards/Gruss, Boris. https://people.kernel.org/tglx/notes-about-netiquette
