[2/3] softfloat: extract softfloat-target.h

Most definitions in softfloat.h are really target-independent.  Split
the few that stand out as target-dependent, to allow including
softfloat.h from files that are not compiled per-target.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 fpu/softfloat-target.h |  119 ++++++++++++++++++++++++++++++++++++++++++++++++
 fpu/softfloat.c        |   11 +++-
 fpu/softfloat.h        |   70 +----------------------------
 target-arm/helper.c    |    1 +
 4 files changed, 129 insertions(+), 72 deletions(-)
 create mode 100644 fpu/softfloat-target.h

On Thu, May 19, 2011 at 4:47 PM, Paolo Bonzini <pbonzini@redhat.com> wrote:
> Most definitions in softfloat.h are really target-independent.  Split
> the few that stand out as target-dependent, to allow including
> softfloat.h from files that are not compiled per-target.

Maybe the next step should be (or in place of this patch) to move the
various defines to target-xyz/cpu.h?

On 05/20/2011 09:38 PM, Blue Swirl wrote:
> On Thu, May 19, 2011 at 4:47 PM, Paolo Bonzini<pbonzini@redhat.com>  wrote:
> >  Most definitions in softfloat.h are really target-independent.  Split
> >  the few that stand out as target-dependent, to allow including
> >  softfloat.h from files that are not compiled per-target.
>
> Maybe the next step should be (or in place of this patch) to move the
> various defines to target-xyz/cpu.h?

Not sure...  Adding (or increasing, I haven't checked) the dependency of 
softfloat on */cpu.h is bad, but a new file 
target-xyz/softfloat-target.h would be feasible.  You'd miss the 
classification in the two big groups (sNAN sign bit = 0/1), but that's 
not a big deal.

Regarding "in place of this patch" however I disagree, since this patch 
is a naive split of a file in two big pieces, while your proposal 
requires more thought.

Paolo

On Fri, May 20, 2011 at 10:45 PM, Paolo Bonzini <pbonzini@redhat.com> wrote:
> On 05/20/2011 09:38 PM, Blue Swirl wrote:
>>
>> On Thu, May 19, 2011 at 4:47 PM, Paolo Bonzini<pbonzini@redhat.com>
>>  wrote:
>> >  Most definitions in softfloat.h are really target-independent.  Split
>> >  the few that stand out as target-dependent, to allow including
>> >  softfloat.h from files that are not compiled per-target.
>>
>> Maybe the next step should be (or in place of this patch) to move the
>> various defines to target-xyz/cpu.h?
>
> Not sure...  Adding (or increasing, I haven't checked) the dependency of
> softfloat on */cpu.h is bad, but a new file target-xyz/softfloat-target.h
> would be feasible.  You'd miss the classification in the two big groups
> (sNAN sign bit = 0/1), but that's not a big deal.
>
> Regarding "in place of this patch" however I disagree, since this patch is a
> naive split of a file in two big pieces, while your proposal requires more
> thought.

OK. Overall, a nice series.

On Thu, May 19, 2011 at 03:47:00PM +0200, Paolo Bonzini wrote:
> Most definitions in softfloat.h are really target-independent.  Split
> the few that stand out as target-dependent, to allow including
> softfloat.h from files that are not compiled per-target.

Looks ok in principle, but I wonder if we should really create a new
header for that. softfloat-specialize.h is already supposed to contain
the target specific functions, so it's probably a good idea to move them
there instead.

Also please note that this will conflict with the patch series I sent
recently about removing softfloat-native (most of the patches have been
acked, I'll send a new series soon).

> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>  fpu/softfloat-target.h |  119 ++++++++++++++++++++++++++++++++++++++++++++++++
>  fpu/softfloat.c        |   11 +++-
>  fpu/softfloat.h        |   70 +----------------------------
>  target-arm/helper.c    |    1 +
>  4 files changed, 129 insertions(+), 72 deletions(-)
>  create mode 100644 fpu/softfloat-target.h
> 
> diff --git a/fpu/softfloat-target.h b/fpu/softfloat-target.h
> new file mode 100644
> index 0000000..513542c
> --- /dev/null
> +++ b/fpu/softfloat-target.h
> @@ -0,0 +1,119 @@
> +/*
> + * QEMU float support
> + *
> + * Derived from SoftFloat.
> + */
> +
> +/*============================================================================
> +
> +This C header file is part of the SoftFloat IEC/IEEE Floating-point Arithmetic
> +Package, Release 2b.
> +
> +Written by John R. Hauser.  This work was made possible in part by the
> +International Computer Science Institute, located at Suite 600, 1947 Center
> +Street, Berkeley, California 94704.  Funding was partially provided by the
> +National Science Foundation under grant MIP-9311980.  The original version
> +of this code was written as part of a project to build a fixed-point vector
> +processor in collaboration with the University of California at Berkeley,
> +overseen by Profs. Nelson Morgan and John Wawrzynek.  More information
> +is available through the Web page `http://www.cs.berkeley.edu/~jhauser/
> +arithmetic/SoftFloat.html'.
> +
> +THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort has
> +been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT TIMES
> +RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO PERSONS
> +AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ALL LOSSES,
> +COSTS, OR OTHER PROBLEMS THEY INCUR DUE TO THE SOFTWARE, AND WHO FURTHERMORE
> +EFFECTIVELY INDEMNIFY JOHN HAUSER AND THE INTERNATIONAL COMPUTER SCIENCE
> +INSTITUTE (possibly via similar legal warning) AGAINST ALL LOSSES, COSTS, OR
> +OTHER PROBLEMS INCURRED BY THEIR CUSTOMERS AND CLIENTS DUE TO THE SOFTWARE.
> +
> +Derivative works are acceptable, even for commercial purposes, so long as
> +(1) the source code for the derivative work includes prominent notice that
> +the work is derivative, and (2) the source code includes prominent notice with
> +these four paragraphs for those parts of this code that are retained.
> +
> +=============================================================================*/
> +
> +#ifndef SOFTFLOAT_TARGET_H
> +#define SOFTFLOAT_TARGET_H
> +
> +#include "softfloat.h"
> +#include "config-target.h"
> +
> +#if defined(TARGET_MIPS) || defined(TARGET_SH4) || defined(TARGET_UNICORE32)
> +#define SNAN_BIT_IS_ONE		1
> +#else
> +#define SNAN_BIT_IS_ONE		0
> +#endif
> +
> +#ifdef CONFIG_SOFTFLOAT
> +
> +/*----------------------------------------------------------------------------
> +| The pattern for a default generated half-precision NaN.
> +*----------------------------------------------------------------------------*/
> +#if defined(TARGET_ARM)
> +#define float16_default_nan make_float16(0x7E00)
> +#elif SNAN_BIT_IS_ONE
> +#define float16_default_nan make_float16(0x7DFF)
> +#else
> +#define float16_default_nan make_float16(0xFE00)
> +#endif
> +
> +/*----------------------------------------------------------------------------
> +| The pattern for a default generated single-precision NaN.
> +*----------------------------------------------------------------------------*/
> +#if defined(TARGET_SPARC)
> +#define float32_default_nan make_float32(0x7FFFFFFF)
> +#elif defined(TARGET_PPC) || defined(TARGET_ARM) || defined(TARGET_ALPHA)
> +#define float32_default_nan make_float32(0x7FC00000)
> +#elif SNAN_BIT_IS_ONE
> +#define float32_default_nan make_float32(0x7FBFFFFF)
> +#else
> +#define float32_default_nan make_float32(0xFFC00000)
> +#endif
> +
> +/*----------------------------------------------------------------------------
> +| The pattern for a default generated double-precision NaN.
> +*----------------------------------------------------------------------------*/
> +#if defined(TARGET_SPARC)
> +#define float64_default_nan make_float64(LIT64( 0x7FFFFFFFFFFFFFFF ))
> +#elif defined(TARGET_PPC) || defined(TARGET_ARM) || defined(TARGET_ALPHA)
> +#define float64_default_nan make_float64(LIT64( 0x7FF8000000000000 ))
> +#elif SNAN_BIT_IS_ONE
> +#define float64_default_nan make_float64(LIT64( 0x7FF7FFFFFFFFFFFF ))
> +#else
> +#define float64_default_nan make_float64(LIT64( 0xFFF8000000000000 ))
> +#endif
> +
> +#ifdef FLOATX80
> +
> +#if SNAN_BIT_IS_ONE
> +#define floatx80_default_nan_high 0x7FFF
> +#define floatx80_default_nan_low  LIT64( 0xBFFFFFFFFFFFFFFF )
> +#else
> +#define floatx80_default_nan_high 0xFFFF
> +#define floatx80_default_nan_low  LIT64( 0xC000000000000000 )
> +#endif
> +
> +#endif
> +
> +#ifdef FLOAT128
> +
> +/*----------------------------------------------------------------------------
> +| The pattern for a default generated quadruple-precision NaN.  The `high' and
> +| `low' values hold the most- and least-significant bits, respectively.
> +*----------------------------------------------------------------------------*/
> +#if SNAN_BIT_IS_ONE
> +#define float128_default_nan_high LIT64( 0x7FFF7FFFFFFFFFFF )
> +#define float128_default_nan_low  LIT64( 0xFFFFFFFFFFFFFFFF )
> +#else
> +#define float128_default_nan_high LIT64( 0xFFFF800000000000 )
> +#define float128_default_nan_low  LIT64( 0x0000000000000000 )
> +#endif
> +
> +#endif
> +
> +#endif /* !CONFIG_SOFTFLOAT */
> +
> +#endif /* !SOFTFLOAT_TARGET_H */
> diff --git a/fpu/softfloat.c b/fpu/softfloat.c
> index baba1dc..72aefc5 100644
> --- a/fpu/softfloat.c
> +++ b/fpu/softfloat.c
> @@ -45,12 +45,17 @@ these four paragraphs for those parts of this code that are retained.
>  #include "softfloat-macros.h"
>  
>  /*----------------------------------------------------------------------------
> +| Public functions and definitions for the default generated quiet NaNs.
> +| This is target-specific.
> +*----------------------------------------------------------------------------*/
> +#include "softfloat-target.h"
> +
> +/*----------------------------------------------------------------------------
>  | Functions and definitions to determine:  (1) whether tininess for underflow
>  | is detected before or after rounding by default, (2) what (if anything)
>  | happens when exceptions are raised, (3) how signaling NaNs are distinguished
> -| from quiet NaNs, (4) the default generated quiet NaNs, and (5) how NaNs
> -| are propagated from function inputs to output.  These details are target-
> -| specific.
> +| from quiet NaNs, and (4) how NaNs are propagated from function inputs to
> +| output.  These details are target-specific.
>  *----------------------------------------------------------------------------*/
>  #include "softfloat-specialize.h"
>  
> diff --git a/fpu/softfloat.h b/fpu/softfloat.h
> index 5eff085..c9e4446 100644
> --- a/fpu/softfloat.h
> +++ b/fpu/softfloat.h
> @@ -43,7 +43,7 @@ these four paragraphs for those parts of this code that are retained.
>  #endif
>  
>  #include <inttypes.h>
> -#include "config.h"
> +#include "config-host.h"
>  
>  /*----------------------------------------------------------------------------
>  | Each of the following `typedef's defines the most convenient type that holds
> @@ -68,12 +68,6 @@ typedef int64_t int64;
>  #define LIT64( a ) a##LL
>  #define INLINE static inline
>  
> -#if defined(TARGET_MIPS) || defined(TARGET_SH4) || defined(TARGET_UNICORE32)
> -#define SNAN_BIT_IS_ONE		1
> -#else
> -#define SNAN_BIT_IS_ONE		0
> -#endif
> -
>  /*----------------------------------------------------------------------------
>  | The macro `FLOATX80' must be defined to enable the extended double-precision
>  | floating-point format `floatx80'.  If this macro is not defined, the
> @@ -280,17 +274,6 @@ int float16_is_signaling_nan( float16 );
>  float16 float16_maybe_silence_nan( float16 );
>  
>  /*----------------------------------------------------------------------------
> -| The pattern for a default generated half-precision NaN.
> -*----------------------------------------------------------------------------*/
> -#if defined(TARGET_ARM)
> -#define float16_default_nan make_float16(0x7E00)
> -#elif SNAN_BIT_IS_ONE
> -#define float16_default_nan make_float16(0x7DFF)
> -#else
> -#define float16_default_nan make_float16(0xFE00)
> -#endif
> -
> -/*----------------------------------------------------------------------------
>  | Software IEC/IEEE single-precision conversion routines.
>  *----------------------------------------------------------------------------*/
>  int16 float32_to_int16_round_to_zero( float32 STATUS_PARAM );
> @@ -393,19 +376,6 @@ INLINE float32 float32_set_sign(float32 a, int sign)
>  
>  
>  /*----------------------------------------------------------------------------
> -| The pattern for a default generated single-precision NaN.
> -*----------------------------------------------------------------------------*/
> -#if defined(TARGET_SPARC)
> -#define float32_default_nan make_float32(0x7FFFFFFF)
> -#elif defined(TARGET_PPC) || defined(TARGET_ARM) || defined(TARGET_ALPHA)
> -#define float32_default_nan make_float32(0x7FC00000)
> -#elif SNAN_BIT_IS_ONE
> -#define float32_default_nan make_float32(0x7FBFFFFF)
> -#else
> -#define float32_default_nan make_float32(0xFFC00000)
> -#endif
> -
> -/*----------------------------------------------------------------------------
>  | Software IEC/IEEE double-precision conversion routines.
>  *----------------------------------------------------------------------------*/
>  int16 float64_to_int16_round_to_zero( float64 STATUS_PARAM );
> @@ -504,19 +474,6 @@ INLINE float64 float64_set_sign(float64 a, int sign)
>  #define float64_half make_float64(0x3fe0000000000000LL)
>  #define float64_infinity make_float64(0x7ff0000000000000LL)
>  
> -/*----------------------------------------------------------------------------
> -| The pattern for a default generated double-precision NaN.
> -*----------------------------------------------------------------------------*/
> -#if defined(TARGET_SPARC)
> -#define float64_default_nan make_float64(LIT64( 0x7FFFFFFFFFFFFFFF ))
> -#elif defined(TARGET_PPC) || defined(TARGET_ARM) || defined(TARGET_ALPHA)
> -#define float64_default_nan make_float64(LIT64( 0x7FF8000000000000 ))
> -#elif SNAN_BIT_IS_ONE
> -#define float64_default_nan make_float64(LIT64( 0x7FF7FFFFFFFFFFFF ))
> -#else
> -#define float64_default_nan make_float64(LIT64( 0xFFF8000000000000 ))
> -#endif
> -
>  #ifdef FLOATX80
>  
>  /*----------------------------------------------------------------------------
> @@ -596,19 +553,6 @@ INLINE int floatx80_is_any_nan(floatx80 a)
>  #define floatx80_half make_floatx80(0x3ffe, 0x8000000000000000LL)
>  #define floatx80_infinity make_floatx80(0x7fff, 0x8000000000000000LL)
>  
> -/*----------------------------------------------------------------------------
> -| The pattern for a default generated extended double-precision NaN.  The
> -| `high' and `low' values hold the most- and least-significant bits,
> -| respectively.
> -*----------------------------------------------------------------------------*/
> -#if SNAN_BIT_IS_ONE
> -#define floatx80_default_nan_high 0x7FFF
> -#define floatx80_default_nan_low  LIT64( 0xBFFFFFFFFFFFFFFF )
> -#else
> -#define floatx80_default_nan_high 0xFFFF
> -#define floatx80_default_nan_low  LIT64( 0xC000000000000000 )
> -#endif
> -
>  #endif
>  
>  #ifdef FLOAT128
> @@ -684,18 +628,6 @@ INLINE int float128_is_any_nan(float128 a)
>          ((a.low != 0) || ((a.high & 0xffffffffffffLL) != 0));
>  }
>  
> -/*----------------------------------------------------------------------------
> -| The pattern for a default generated quadruple-precision NaN.  The `high' and
> -| `low' values hold the most- and least-significant bits, respectively.
> -*----------------------------------------------------------------------------*/
> -#if SNAN_BIT_IS_ONE
> -#define float128_default_nan_high LIT64( 0x7FFF7FFFFFFFFFFF )
> -#define float128_default_nan_low  LIT64( 0xFFFFFFFFFFFFFFFF )
> -#else
> -#define float128_default_nan_high LIT64( 0xFFFF800000000000 )
> -#define float128_default_nan_low  LIT64( 0x0000000000000000 )
> -#endif
> -
>  #endif
>  
>  #else /* CONFIG_SOFTFLOAT */
> diff --git a/target-arm/helper.c b/target-arm/helper.c
> index 62ae72e..0f15cf3 100644
> --- a/target-arm/helper.c
> +++ b/target-arm/helper.c
> @@ -8,6 +8,7 @@
>  #include "helper.h"
>  #include "qemu-common.h"
>  #include "host-utils.h"
> +#include "softfloat-target.h"
>  #if !defined(CONFIG_USER_ONLY)
>  #include "hw/loader.h"
>  #endif
> -- 
> 1.7.4.4
> 
> 
> 
>

On 05/23/2011 11:11 PM, Aurelien Jarno wrote:
> Looks ok in principle, but I wonder if we should really create a new
> header for that. softfloat-specialize.h is already supposed to contain
> the target specific functions, so it's probably a good idea to move them
> there instead.

I was wondering about that too... I was thinking of separating _private_ 
target-specific functions and a public interface that is target-specific 
just because we go for macros rather than extern variables.

In fact, using const variables is another way to solve the problem, 
without introducing a new header.  I'll probably do that for v2.

> Also please note that this will conflict with the patch series I sent
> recently about removing softfloat-native (most of the patches have been
> acked, I'll send a new series soon).

Yes, no problem.

Paolo