On Wed, Oct 28, 2020 at 4:00 PM H.J. Lu <hjl.to...@gmail.com> wrote: > On Wed, Oct 28, 2020 at 11:40 AM Nick Bowler <nbow...@draconx.ca> wrote: > > On 2020-10-28, Zack Weinberg <za...@panix.com> wrote: > > > On Wed, Oct 28, 2020 at 2:16 PM Nick Bowler <nbow...@draconx.ca> wrote: > > >> On 2020-10-28, H.J. Lu <hjl.to...@gmail.com> wrote: > > >> > GCC introduced some time ago option -flto=jobserver in order to use the > > >> > GNU Make jobserver when parallelising LTO builds. It is actually a > > >> > similar "recursive make". When doing a recursive make, you need to > > >> > place a '+' character at the beginning of the recipe line in order to > > >> > let GNU Make pass the jobserver file descriptors to the child > > >> > processes. > > >> > > > >> > Add the --jobserver option to add a '+' character to the recipe line in > > >> > program.am and ltlibrary.am. > > > ... > > >> Surely this needs to be a configure-time option, perhaps combined > > >> with some sort of configure test, since otherwise users won't get > > >> this choice, right? As an automake option the choice made by whomever > > >> prepares the distribution will get baked into distributed Makefile.in > > >> files... > > > > > > I was going to say something very similar: there shouldn't be an > > > option at all. The decision of whether or not to put + at the > > > beginning of the recipe line should be made _when make is run_, based > > > on whether -flto=jobserver actually appears in $(LDFLAGS) or wherever. > > > > Since this check cannot be done inside a rule, testing LDFLAGS at make > > time is probably impractical to do portably.
LTO_PLUS = $(if $(findstring -flto=jobserver,$(LDFLAGS)),+) I think it's okay if this only works with GNU Make. > "make -n" will execute the recipe with the "+" prefix. But --jobserver > is off by default. People who use --jobserver prefer a working GCC > -flto=jobserver over a broken "make -n". What Nick is saying - and I agree with him - is that it's not okay for the package maintainer to make that decision. zw
