Remove selection for ARM variants from menuconfig, v2

[Carmelo AMOROSO]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 1/10/2011 7:55 AM, Khem Raj wrote:
> On 1/8/2011 4:45 PM, Yann E. MORIN wrote:
>  > All,
>  >
>  > This patch series removes the variants selection for the ARM architecture.
>  > The initial idea is that the user is responsible for properly configuring
>  > the compiler in the first place, or passing appropriate EXTRA_CFLAGS to
>  > use settings that are not the compiler's default.
>  >
>  >
>  > I split the process in small steps, so that each step contains a single,
>  > semantically self-contained change that is easy to understand and review,
>  > and that does not break.
>  >
>  > For example, before we can remove the variants selection, we need a way to
>  > know if use of BX is possible, so we need to know if the CPU is Thumb-aware
>  > (and even Thumb-only). Then we can remove the CPUs selection, remove the
>  > forcing blind options, and add a prompt to the Thumb option.
>  >
>  > This means that some stuff added in one patch (mostly in patch #2) will get
>  > removed in a subsequent patch. Up to, and including patch #2, there is no
>  > change in behavior, from the perspective of the user.
>  >
>  > Last patch, "remove EABI/OABI selection", might seem dubious in the first
>  > place, as Khem pointed out. Indeed, this means that to detect whether we
>  > are EABI or OABI, we need a cross-compiler. This poses no problem for
>  > NPTL, as a cross-compiler is required even to install the headers. But
>  > for LinuxThreads, we currently do not require a cross-compiler when
>  > installing the headers. Having to would be a serious regression. But
>  > from my experiments, headers do not depend on EABI or OABI. The
>  > installed headers only differ in bits/uClibc_config.h:
>  > - build date
>  > - #define and #undef for OABI/EABI
>  >
>  > So I believe that removing the EABI selection is perfectly valid.
>  >
>  > For info, I produced a reference uClibc.so from origin, built each steps
>  > with corresponding options, objdumped the resulting uClibc.so and compared
>  > to the reference that the build would produce the same library.
>  >
>  >
>  > Changes since v1:
>  > - keep the "Compile n Thumb mode" option (thus it's a new feature)
>  > - fix variable names as per Khem's suggestion on IRC
>  > - reorder series, EABI selection removal now last in series
>  >
>  > [PATCH 1/7] ARM: reorder "Use BX" option
>  > [PATCH 2/7] ARM: introduce blind options to select& force THUMB mode
>  > [PATCH 3/7] ARM: remove sub-arch/variants selection from menuconfig
>  > [PATCH 4/7] ARM: transform the EABI/OABI choice into a boolean
>  > [PATCH 5/7] ARM: #include<bits/arm_asm.h> where __USE_BX__ is used
>  > [PATCH 6/7] ARM: detect BX availibility at build time
>  > [PATCH 7/7] ARM: remove EABI/OABI selection
>  >
>  > Rules.mak | 20 +-----
>  > extra/Configs/Config.arm | 122 ++----------------------------
>  > ldso/ldso/arm/dl-startup.h | 1 +
>  > libc/sysdeps/linux/arm/Makefile.arch | 9 ++-
>  > libc/sysdeps/linux/arm/bits/arm_asm.h | 9 ++-
>  > libc/sysdeps/linux/arm/bits/huge_val.h | 4 +-
>  > libc/sysdeps/linux/arm/sysdep.h | 1 +
>  > 7 files changed, 28 insertions(+), 138 deletions(-)
>  >
> 
> 
> 
> this patchset looks good to me. I am currently testing it out in
> different builds and it has been holding good. I will try it out
> in different combination and run the testsuite then if all
> is ok consider it in.
> 

I'm testing for an ARM926EJ-S (that one I've originally included the
- --mtune option) and let you know.

I assume that we should do the same clean-up for other archs.

Carmelo

>  > Regards,
>  > Yann E. MORIN.
>  > _______________________________________________
>  > uClibc mailing list
>  > uClibc at uclibc.org
>  > http://lists.busybox.net/mailman/listinfo/uclibc
> 
> _______________________________________________
> uClibc mailing list
> uClibc at uclibc.org
> http://lists.busybox.net/mailman/listinfo/uclibc
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk0qs+YACgkQoRq/3BrK1s9GBgCgwj/L9UFolv+FWmvHNwWjvaJ+
MEYAoKqErfgZcMXyjKUXDsTnd2RDkBPg
=zVaa
-----END PGP SIGNATURE-----