Re: [gentoo-dev] [PATCH v2 2/4] kernel-build.eclass: Build logic for dist-kernels

[Michał Górny]

On Mon, 2020-01-06 at 13:07 -0500, Mike Gilbert wrote:
> On Mon, Jan 6, 2020 at 12:54 PM Michał Górny <mgorny@g.o> wrote:
> > On Mon, 2020-01-06 at 12:43 -0500, Mike Gilbert wrote:
> > > On Sun, Jan 5, 2020 at 1:27 AM Michał Górny <mgorny@g.o> wrote:
> > > > +# @FUNCTION: kernel-build_src_configure
> > > > +# @DESCRIPTION:
> > > > +# Prepare the toolchain for building the kernel, get the default .config
> > > > +# or restore savedconfig, and get build tree configured for modprep.
> > > > +kernel-build_src_configure() {
> > > > +       debug-print-function ${FUNCNAME} "${@}"
> > > > +
> > > > +       # force ld.bfd if we can find it easily
> > > > +       local LD="$(tc-getLD)"
> > > > +       if type -P "${LD}.bfd" &>/dev/null; then
> > > > +               LD+=.bfd
> > > > +       fi
> > >
> > > Is there some reason not to use the tc-ld-disable-gold function?
> > >
> >
> > Yes.  As the name says, it covers gold usage only and not lld.
>
> It might be a good idea to copy this logic to handle multi-word LD values.
>
> https://gitweb.gentoo.org/repo/gentoo.git/tree/eclass/toolchain-funcs.eclass#n498
>
> Or, a nicer alternative would be to refactor tc-ld-disable-gold into 2
> functions: move most of the logic into a new function
> "tc-ld-force-bfd", and update tc-ld-disable-gold to call the former if
> tc-ld-is-gold is true. I can write a patch for that if you agree.
>

Sure, that sounds like a good idea.  Please also replace which(1) with
'type' while at it.

-- 
Best regards,
Michał Górny