On Thu, Mar 01, 2018 at 08:10:36AM +0000, Jason McIntyre wrote: > On Thu, Mar 01, 2018 at 07:05:08PM +1100, Jonathan Gray wrote: > > On Thu, Mar 01, 2018 at 07:51:44AM +0000, Jason McIntyre wrote: > > > On Thu, Mar 01, 2018 at 06:15:59PM +1100, Jonathan Gray wrote: > > > > From lld svn rev 324739 with -z muldefs and -z retpolineplt text removed > > > > to match lld 5.0.1. > > > > > > > > > > morning. > > > > > > this will replace the ld page for archs building llvm, right? haven;t > > > the stomach to read it yet, but no objections. you should probably add > > > an Nm entry to NAME for "ld" though: > > > > > > .Sh NAME > > > .Nm ld.lld , > > > .Nm ld > > > > lld is ld only on arm64 but is built/installed on all archs that build > > llvm as ld.lld. So we can't do that at the moment as ld is binutils/bfd > > ld everywhere else. > > > > ah ok. > > it will still make sense no? if it gets installed for arm64 only it will > be correct for that arch and the situation unchanged for others. or have i > misunderstood?
The binary is installed on all architectures that build llvm as /usr/bin/ld.lld and is also installed as /usr/bin/ld on arm64. With ld.1 from binutils and ld.lld.1 with ld in .Nm 'man ld' will pick the binutils page but the ld.lld.1 page will show up in 'man -a ld' even on architectures where ld is not lld. > > > > > > > also don;t know if you tag this stuff $OpenBSD$ or not. > > > > Adding $OpenBSD$/$Mdocdate$ would be local diffs. I'm inclined > > to not add them. > > > > fair enough. > > jmc >