This is the mail archive of the
newlib@sourceware.org
mailing list for the newlib project.
Re: [PATCH, ping1] Make macro checks ARMv8-M baseline proof
- From: Thomas Preud'homme <thomas dot preudhomme at foss dot arm dot com>
- To: "Richard Earnshaw (lists)" <Richard dot Earnshaw at arm dot com>
- Cc: newlib at sourceware dot org, Corinna Vinschen <vinschen at redhat dot com>
- Date: Fri, 05 Feb 2016 14:02:02 +0800
- Subject: Re: [PATCH, ping1] Make macro checks ARMv8-M baseline proof
- Authentication-results: sourceware.org; auth=none
- References: <56719806 dot 40006 at arm dot com> <1469748 dot DKHh6N0Akf at hardin dot shanghai dot arm dot com> <56B36225 dot 5080104 at arm dot com>
On Thursday, February 04, 2016 02:37:25 PM Richard Earnshaw wrote:
> >
> > My apologize, unified tree build works fine for me here. So there is no
> > newlib directory (or symlink) in your build directory? Could you send me
> > <target triplet>/libgloss/arm/Makefile from your build directory?
>
> Ah, light dawns. I don't use libgloss for my builds, relying on the
> semi-hosting support in newlib itself.
Indeed, I didn't realize this part of newlib was not built when libgloss is
used. So my testing there was moot. I could do the same INCLUDES trick in
newlib/libc/sys/arm/Makefile.am, that should work.
On the other hand, there is already some use of __ARM_ARCH_ISA_THUMB and
__ARM_ARCH_ISA_ARM in newlib/libc/machine/arm/ before that commit so I guess
we don't need to include acle-compat.h there?
By the way, what's the reason for passing NEWLIB_CFLAGS explicitely when
invoking libgloss/<target>/Makefile instead of relying on the export mechanism
that would allow target Makefile to extend (with +=) the variable?
Best regards,
Thomas