Re: [PATCH] Makefile: Set CC/AR variable only if it doesn't have a value

[Henrik Austad <henrik@austad.us>]

[-- Attachment #1: Type: text/plain, Size: 1517 bytes --]

On Mon, Jan 04, 2016 at 05:25:07PM +0100, Uwe Kleine-König wrote:
> On Mon, Jan 04, 2016 at 08:20:01AM -0800, Khem Raj wrote:
> > 
> > > On Jan 4, 2016, at 5:39 AM, Jeff Epler <jepler@unpythonic.net> wrote:
> > > 
> > > I believe the patch does not have the desired effect, because make(1)
> > > always sets CC automatically:
> > > 
> > > $ cat raj.mk
> > > CROSS_COMPILE=x86_64-linux-gnu-
> > > CC?=$(CROSS_COMPILE)gcc
> > > .PHONY: default
> > > default:
> > > 	@echo CC=$(CC)
> > > $ make -f raj.mk
> > > CC=cc
> > > 
> > > With your patch, those who specify CROSS_COMPILE, or for whom gcc and cc
> > > are different, will get different (worse) behavior.
> > 
> > Right. I think if CC was used as such with out constructing out of CROSS_COMPILE
> > in Makefile then my issue would be fixed too. However this would mean that cross compiling users
> > now have to pass CC = <cross-compiler> themselves instead of CROSS_COMPILE prefix. Is that
> > acceptable ?
> 
> You could play with something like:
> 
> ifeq($(origin CC),default)
> CC=$(CROSS_COMPILE)gcc
> endif
> 
> but not sure I'd like the result. If you asked me, I'd drop all that
> CROSS_COMPILE stuff.

Why? Isn't this the normal way of triggering cross-compile? AFIAK, that's 
how the kernel does it, and that seems like a pretty sane way.

And yes, I compile rt-tests to several architectures, so I'm quite fond of 
the CROSS_COMPILE switch. In fact, I might get cross if you cross it out :)

-- 
Henrik Austad

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]