linux-kernel - Re: [PATCH] powerpc: remove meaningless KBUILD

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 16 Jul 2019 22:15:47 +1000
From:   Michael Ellerman <mpe@...erman.id.au>
To:     Segher Boessenkool <segher@...nel.crashing.org>
Cc:     Masahiro Yamada <yamada.masahiro@...ionext.com>,
        Stephen Rothwell <sfr@...b.auug.org.au>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Nicholas Piggin <npiggin@...il.com>,
        Paul Mackerras <paulus@...ba.org>,
        linuxppc-dev <linuxppc-dev@...ts.ozlabs.org>
Subject: Re: [PATCH] powerpc: remove meaningless KBUILD_ARFLAGS addition

Segher Boessenkool <segher@...nel.crashing.org> writes:
> On Mon, Jul 15, 2019 at 05:05:34PM +1000, Michael Ellerman wrote:
>> Segher Boessenkool <segher@...nel.crashing.org> writes:
>> > Yes, that is why I used the environment variable, all binutils work
>> > with that.  There was no --target option in GNU ar before 2.22.
>> 
>> Yeah, we're not very good at testing with really old binutils, so I
>> guess we broke that.
>> 
>> I'm inclined to merge this, it doesn't seem to break anything, and it
>> fixes using --target on old binutils that don't have it.
>
> But we don't set the target any other way either.  I don't think this
> will work with a 32-bit toolchain (default target 32 bit) and a 64-bit
> kernel, or the other way around.

I think it does, but maybe I'm misunderstanding.

My test setup is:

  ~/linux$ export PATH=/home/toolchains/ppc/gcc-8-branch/powerpc-linux/bin/:$PATH
  ~/linux$ echo "int test(void) { return 2; }" > test.c
  ~/linux$ powerpc-linux-gcc -c test.c 
  ~/linux$ file test.o 
  test.o: ELF 32-bit MSB relocatable, PowerPC or cisco 4500, version 1 (SYSV), not stripped
  ~/linux$ make CROSS_COMPILE=powerpc-linux- -s ppc64le_defconfig
  ~/linux$ make CROSS_COMPILE=powerpc-linux- -s -j 320
  ~/linux$ echo $?
  0

And it's definitely calling ar with no flags, eg:

  rm -f init/built-in.a; powerpc-linux-ar rcSTPD init/built-in.a init/main.o init/version.o init/do_mounts.o init/do_mounts_rd.o init/do_mounts_initrd.o init/do_mounts_md.o init/initramfs.o init/init_task.o

So presumably at some point ar learnt to cope with objects that don't
match its default? (how do I ask it what its default is?)

> Then again, does that work at *all* nowadays?  Do we even consider that
> important, *should* it work?

Yes and yes. There were a lot of bugs in the kernel makefiles after we
added LE support which prevented a biarch/biendian compiler from working.
But now it does work and we want it to keep working because it means you
can have a single compiler for building 32-bit, 64-bit BE & 64-bit LE.

cheers