[patch] Fixes for SH-4 without an FPU

[Carmelo AMOROSO]

On 6/3/2010 3:34 PM, Andrew Stubbs wrote:
> This patch disables SH-4 optimizations that rely on the FPU when
> building for variants that don't have an FPU, such as SH-4AL.
> 
> Signed-off-by: Andrew Stubbs <ams at codesourcery.com>
> ---
> diff --git a/libc/string/sh/sh4/memcpy.S b/libc/string/sh/sh4/memcpy.S
> index efdaf8b..83d40d4 100644
> --- a/libc/string/sh/sh4/memcpy.S
> +++ b/libc/string/sh/sh4/memcpy.S
> @@ -21,7 +21,7 @@
>  #include <sysdep.h>
>  #include <endian.h>
> 
> -#ifdef __LITTLE_ENDIAN__
> +#if defined(__LITTLE_ENDIAN__) && defined(__FPU_ANY__)
>  #define    MEMCPY_USES_FPU
>  /* Use paired single precision load or store mode for 64-bit tranfering.
>   * FPSCR.SZ=1,FPSCR.SZ=0 is well defined on both SH4-200 and SH4-300.
> diff --git a/libc/string/sh/sh4/memmove.c b/libc/string/sh/sh4/memmove.c
> index 4d52db2..67f9a1e 100644
> --- a/libc/string/sh/sh4/memmove.c
> +++ b/libc/string/sh/sh4/memmove.c
> @@ -7,8 +7,12 @@
>   * Licensed under the LGPL v2.1, see the file COPYING.LIB in this tarball.
>   */
> 
> -#include <string.h>
> 
> +#ifndef __FPU_ANY__
> +#include "../../memmove.c"
> +#else
> +
> +#include <string.h>
> 
>  #define FPSCR_SR    (1 << 20)
>  #define STORE_FPSCR(x)    __asm__ volatile("sts fpscr, %0" : "=r"(x))
> @@ -115,3 +119,5 @@ void *memmove(void *dest, const void *src, size_t len)
>  }
> 
>  libc_hidden_def(memmove)
> +
> +#endif
> diff --git a/libc/string/sh/sh4/memset.S b/libc/string/sh/sh4/memset.S
> index 1a57cb9..5cef0db 100644
> --- a/libc/string/sh/sh4/memset.S
> +++ b/libc/string/sh/sh4/memset.S
> @@ -17,7 +17,7 @@
> 
>  #include <sysdep.h>
> 
> -#ifdef __LITTLE_ENDIAN__
> +#if defined(__LITTLE_ENDIAN__) && defined(__FPU_ANY__)
>  #define MEMSET_USES_FPU
>  /* Use paired single precision load or store mode for 64-bit tranfering.
>   * FPSCR.SZ=1,FPSCR.SZ=0 is well defined on both SH4-200 and SH4-300.
> diff --git a/libc/sysdeps/linux/sh/Makefile.arch
> b/libc/sysdeps/linux/sh/Makefile.arch
> index 727a3f4..62f3d47 100644
> --- a/libc/sysdeps/linux/sh/Makefile.arch
> +++ b/libc/sysdeps/linux/sh/Makefile.arch
> @@ -12,5 +12,5 @@ CSRC := \
>  SSRC := setjmp.S __longjmp.S ___fpscr_values.S
> 
>  ifneq ($(UCLIBC_HAS_THREADS_NATIVE),y)
> -SSRC += vfork.S
> +SSRC += vfork.S clone.S
>  endif


Hi Andrew,
I'll include partially your changes excluding the Makefiles (you did not
clarified the relationship with the FPU fixes, I guess your patch
contains some spurious changes, but I'm not sure).
Anyway the guard to use should be __SH_FPU_ANY__ instead of FPU_ANY (at
least using the compiler I have (gcc 4.3.4) the built-in macro defined
is actually __SH_FPU_ANY__)

Regards,
Carmelo

> _______________________________________________
> uClibc mailing list
> uClibc at uclibc.org
> http://lists.busybox.net/mailman/listinfo/uclibc
>