Re: [HEADSUP] Re: [PATCH] newlib: libm: merge build up a directory

Mike Frysinger <[email protected]> Wed, 16 Feb 2022 23:38:43 -0500

Newsgroups	gmane.comp.lib.newlib
Message-ID	<Yg3RUwXnmoeOIVKp@vapier>

On 16 Feb 2022 09:50, Corinna Vinschen wrote:
> On Feb 12 15:34, Mike Frysinger wrote:
> > Convert all the libm/ subdir makes into the top-level Makefile.  This
> > allows us to build all of libm from the top Makefile without using any
> > recursive make calls.  This is faster and avoids the funky lib.a logic
> > where we unpack subdir archives to repack into a single libm.a.  The
> > machine override logic is maintained though by way of Makefile include
> > ordering, and source file accumulation in libm_a_SOURCES.
> 
> What targets did you test this with?

testing these targets:
aarch64-elf arc-elf arm-eabi
bfin-elf
cr16-elf cris-elf csky-elf
epiphany-elf
frv-elf
h8300-elf
i386-elf i686-pc-cygwin iq2000-elf
m68k-elf microblaze-elf mips-elf mn10300-elf moxie-elf
nds32le-elf
or1k-elf
powerpc-elf
sparc-elf sparc-rtems
v850e-elf visium-elf
xstormy16-elf

the actual testing:
- build w/out the patch
make install DESTDIR=$PWD/old.root
ar t every library
objdump -d every object
- build w/the patch
make install DESTDIR=$PWD/new.root
ar t every library
objdump -d every object

run `diff -ur` on the 2 roots.  they install the same set of files, and
the disassembly of the objects is the same.

so should be good to go.  only needed about 30G to determine this :p.

> > One thing to note is that this will require GNU Make because of:
>   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> 
> Anybody here having a problem with that?

while i couldn't find any prior usage in newlib specifically (i think all
the code i looked through is portable make), libgloss def has GNUisms in
at least 12 of its ports:
aarch64 arm bfin cris csky m68k msp430 nios2 or1k riscv visium xstormy16

no one seems to have complained so far :).
-mike

signature.asc (application/pgp-signature, 833 B)