[Qemu-devel] [PATCH v4 0/2] soft-float-fixes for target-s390x

[Qemu-devel] [PATCH v4 0/2] soft-float-fixes for target-s390x

[Richard Henderson]

Changes v3-v4:
  Rebase after source tree rearrangements.

Changes v2-v3:
  Ignore softfloat "style" completely.  Code now formatted per QEMU.

Changes v1-v2:
  Incorporating feedback from Peter Maydell (previously missed in
  the 140+ message thread).


r~


Richard Henderson (2):
  softfloat: Fix uint64_to_float64
  softfloat: Implement uint64_to_float128

 fpu/softfloat.c         | 21 ++++++++++++++++++---
 include/fpu/softfloat.h |  3 +++
 2 files changed, 21 insertions(+), 3 deletions(-)

-- 
1.7.11.7

[Qemu-devel] [PATCH 1/2] softfloat: Fix uint64_to_float64

[Richard Henderson]

The interface to normalizeRoundAndPackFloat64 requires that the
high bit be clear.  Perform one shift-right-and-jam if needed.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
---
 fpu/softfloat.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/fpu/softfloat.c b/fpu/softfloat.c
index 0cfa6b4..20b05d4 100644
--- a/fpu/softfloat.c
+++ b/fpu/softfloat.c
@@ -1271,11 +1271,18 @@ float64 int64_to_float64( int64 a STATUS_PARAM )
 
 }
 
-float64 uint64_to_float64( uint64 a STATUS_PARAM )
+float64 uint64_to_float64(uint64 a STATUS_PARAM)
 {
-    if ( a == 0 ) return float64_zero;
-    return normalizeRoundAndPackFloat64( 0, 0x43C, a STATUS_VAR );
+    int exp =  0x43C;
 
+    if (a == 0) {
+        return float64_zero;
+    }
+    if ((int64_t)a < 0) {
+        shift64RightJamming(a, 1, &a);
+        exp += 1;
+    }
+    return normalizeRoundAndPackFloat64(0, exp, a STATUS_VAR);
 }
 
 /*----------------------------------------------------------------------------
-- 
1.7.11.7

[Qemu-devel] [PATCH 2/2] softfloat: Implement uint64_to_float128

[Richard Henderson]

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
---
 fpu/softfloat.c         | 8 ++++++++
 include/fpu/softfloat.h | 3 +++
 2 files changed, 11 insertions(+)

diff --git a/fpu/softfloat.c b/fpu/softfloat.c
index 20b05d4..ac3d150 100644
--- a/fpu/softfloat.c
+++ b/fpu/softfloat.c
@@ -1339,6 +1339,14 @@ float128 int64_to_float128( int64 a STATUS_PARAM )
 
 }
 
+float128 uint64_to_float128(uint64 a STATUS_PARAM)
+{
+    if (a == 0) {
+        return float128_zero;
+    }
+    return normalizeRoundAndPackFloat128(0, 0x406E, a, 0 STATUS_VAR);
+}
+
 /*----------------------------------------------------------------------------
 | Returns the result of converting the single-precision floating-point value
 | `a' to the 32-bit two's complement integer format.  The conversion is
diff --git a/include/fpu/softfloat.h b/include/fpu/softfloat.h
index 0946f07..f3927e2 100644
--- a/include/fpu/softfloat.h
+++ b/include/fpu/softfloat.h
@@ -237,6 +237,7 @@ float64 int64_to_float64( int64 STATUS_PARAM );
 float64 uint64_to_float64( uint64 STATUS_PARAM );
 floatx80 int64_to_floatx80( int64 STATUS_PARAM );
 float128 int64_to_float128( int64 STATUS_PARAM );
+float128 uint64_to_float128( uint64 STATUS_PARAM );
 
 /*----------------------------------------------------------------------------
 | Software half-precision conversion routines.
@@ -630,6 +631,8 @@ INLINE int float128_is_any_nan(float128 a)
         ((a.low != 0) || ((a.high & 0xffffffffffffLL) != 0));
 }
 
+#define float128_zero make_float128(0, 0)
+
 /*----------------------------------------------------------------------------
 | The pattern for a default generated quadruple-precision NaN.
 *----------------------------------------------------------------------------*/
-- 
1.7.11.7

Re: [Qemu-devel] [PATCH v4 0/2] soft-float-fixes for target-s390x

[Richard Henderson]

Ping.

r~

On 12/31/2012 10:09 AM, Richard Henderson wrote:
> Changes v3-v4:
>   Rebase after source tree rearrangements.
> 
> Changes v2-v3:
>   Ignore softfloat "style" completely.  Code now formatted per QEMU.
> 
> Changes v1-v2:
>   Incorporating feedback from Peter Maydell (previously missed in
>   the 140+ message thread).
> 
> 
> r~
> 
> 
> Richard Henderson (2):
>   softfloat: Fix uint64_to_float64
>   softfloat: Implement uint64_to_float128
> 
>  fpu/softfloat.c         | 21 ++++++++++++++++++---
>  include/fpu/softfloat.h |  3 +++
>  2 files changed, 21 insertions(+), 3 deletions(-)
> 

Re: [Qemu-devel] [PATCH v4 0/2] soft-float-fixes for target-s390x

[Blue Swirl]

Thanks, applied all.

On Mon, Dec 31, 2012 at 6:09 PM, Richard Henderson <rth@twiddle.net> wrote:
> Changes v3-v4:
>   Rebase after source tree rearrangements.
>
> Changes v2-v3:
>   Ignore softfloat "style" completely.  Code now formatted per QEMU.
>
> Changes v1-v2:
>   Incorporating feedback from Peter Maydell (previously missed in
>   the 140+ message thread).
>
>
> r~
>
>
> Richard Henderson (2):
>   softfloat: Fix uint64_to_float64
>   softfloat: Implement uint64_to_float128
>
>  fpu/softfloat.c         | 21 ++++++++++++++++++---
>  include/fpu/softfloat.h |  3 +++
>  2 files changed, 21 insertions(+), 3 deletions(-)
>
> --
> 1.7.11.7
>

[Qemu-devel] [PATCH v3 0/2] soft-float-fixes for target-s390x

[Richard Henderson]

Changes v2-v3:
  Ignore softfloat "style" completely.  Code now formatted per QEMU.

Changes v1-v2:
  Incorporating feedback from Peter Maydell (previously missed in
  the 140+ message thread).


r~


Richard Henderson (2):
  softfloat: Fix uint64_to_float64
  softfloat: Implement uint64_to_float128

 fpu/softfloat.c | 21 ++++++++++++++++++---
 fpu/softfloat.h |  3 +++
 2 files changed, 21 insertions(+), 3 deletions(-)

-- 
1.7.11.7

[Qemu-devel] [PATCH 1/2] softfloat: Fix uint64_to_float64

[Richard Henderson]

The interface to normalizeRoundAndPackFloat64 requires that the
high bit be clear.  Perform one shift-right-and-jam if needed.

Signed-off-by: Richard Henderson <rth@twiddle.net>
---
 fpu/softfloat.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/fpu/softfloat.c b/fpu/softfloat.c
index 8413146..0daad0d 100644
--- a/fpu/softfloat.c
+++ b/fpu/softfloat.c
@@ -1271,11 +1271,18 @@ float64 int64_to_float64( int64 a STATUS_PARAM )
 
 }
 
-float64 uint64_to_float64( uint64 a STATUS_PARAM )
+float64 uint64_to_float64(uint64 a STATUS_PARAM)
 {
-    if ( a == 0 ) return float64_zero;
-    return normalizeRoundAndPackFloat64( 0, 0x43C, a STATUS_VAR );
+    int exp = 0x43C;
 
+    if (a == 0) {
+        return float64_zero;
+    }
+    if ((int64_t)a < 0) {
+        shift64RightJamming(a, 1, &a);
+        exp += 1;
+    }
+    return normalizeRoundAndPackFloat64(0, exp, a STATUS_VAR);
 }
 
 /*----------------------------------------------------------------------------
-- 
1.7.11.7

Re: [Qemu-devel] [PATCH 1/2] softfloat: Fix uint64_to_float64

[Peter Maydell]

On 11 December 2012 17:21, Richard Henderson <rth@twiddle.net> wrote:
> The interface to normalizeRoundAndPackFloat64 requires that the
> high bit be clear.  Perform one shift-right-and-jam if needed.
>
> Signed-off-by: Richard Henderson <rth@twiddle.net>

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

-- PMM

[Qemu-devel] [PATCH v2 0/2] soft-float fixes for target-s390x

[Richard Henderson]

... or for upcoming s390x changes.

The s390x isa has insns for both signed and unsigned conversions
to all three (binary) floating point types.  One of these was
missing and the other producing incorrect results.

Tested with the s390x rewrite on the gcc testsuite.

Changes v1-v2:
  Incorporating feedback from Peter Maydell (previously missed in
  the 140+ message thread).


r~



Richard Henderson (2):
  softfloat: Fix uint64_to_float64
  softfloat: Implement uint64_to_float128

 fpu/softfloat.c | 17 +++++++++++++++--
 fpu/softfloat.h |  3 +++
 2 files changed, 18 insertions(+), 2 deletions(-)

-- 
1.7.11.7

[Qemu-devel] [PATCH 1/2] softfloat: Fix uint64_to_float64

[Richard Henderson]

The interface to normalizeRoundAndPackFloat64 requires that the
high bit be clear.  Perform one shift-right-and-jam if needed.

Signed-off-by: Richard Henderson <rth@twiddle.net>
---
 fpu/softfloat.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/fpu/softfloat.c b/fpu/softfloat.c
index 8413146..62830d7 100644
--- a/fpu/softfloat.c
+++ b/fpu/softfloat.c
@@ -1273,9 +1273,14 @@ float64 int64_to_float64( int64 a STATUS_PARAM )
 
 float64 uint64_to_float64( uint64 a STATUS_PARAM )
 {
-    if ( a == 0 ) return float64_zero;
-    return normalizeRoundAndPackFloat64( 0, 0x43C, a STATUS_VAR );
+    int exp =  0x43C;
 
+    if ( a == 0 ) return float64_zero;
+    if ( (int64_t)a < 0 ) {
+        shift64RightJamming(a, 1, &a);
+        exp += 1;
+    }
+    return normalizeRoundAndPackFloat64( 0, exp, a STATUS_VAR );
 }
 
 /*----------------------------------------------------------------------------
-- 
1.7.11.7

Re: [Qemu-devel] [PATCH 1/2] softfloat: Fix uint64_to_float64

[Peter Maydell]

On 7 December 2012 21:52, Richard Henderson <rth@twiddle.net> wrote:
> The interface to normalizeRoundAndPackFloat64 requires that the
> high bit be clear.  Perform one shift-right-and-jam if needed.
>
> Signed-off-by: Richard Henderson <rth@twiddle.net>

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

Re: [Qemu-devel] [PATCH 1/2] softfloat: Fix uint64_to_float64

[Andreas Färber]

Am 07.12.2012 22:52, schrieb Richard Henderson:
> The interface to normalizeRoundAndPackFloat64 requires that the
> high bit be clear.  Perform one shift-right-and-jam if needed.
> 
> Signed-off-by: Richard Henderson <rth@twiddle.net>
> ---
>  fpu/softfloat.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/fpu/softfloat.c b/fpu/softfloat.c
> index 8413146..62830d7 100644
> --- a/fpu/softfloat.c
> +++ b/fpu/softfloat.c
> @@ -1273,9 +1273,14 @@ float64 int64_to_float64( int64 a STATUS_PARAM )
>  
>  float64 uint64_to_float64( uint64 a STATUS_PARAM )
>  {
> -    if ( a == 0 ) return float64_zero;
> -    return normalizeRoundAndPackFloat64( 0, 0x43C, a STATUS_VAR );
> +    int exp =  0x43C;

Extra space before number.

>  
> +    if ( a == 0 ) return float64_zero;
> +    if ( (int64_t)a < 0 ) {
> +        shift64RightJamming(a, 1, &a);
> +        exp += 1;
> +    }
> +    return normalizeRoundAndPackFloat64( 0, exp, a STATUS_VAR );
>  }
>  
>  /*----------------------------------------------------------------------------

Shouldn't we be updating the Coding Style on all lines we touch? The
shift line matches it but the normalize and ifs don't.

Andreas

-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg

Re: [Qemu-devel] [PATCH 1/2] softfloat: Fix uint64_to_float64

[Peter Maydell]

On 8 December 2012 14:48, Andreas Färber <afaerber@suse.de> wrote:
> Shouldn't we be updating the Coding Style on all lines we touch? The
> shift line matches it but the normalize and ifs don't.

fpu/ is a special case because the softfloat library has such
a bonkers coding style. I tend to update to qemu standard when
I touch bits of it, but following the existing practice is
also OK in my opinion.

-- PMM

Re: [Qemu-devel] [PATCH 1/2] softfloat: Fix uint64_to_float64

[Andreas Färber]

Am 08.12.2012 17:08, schrieb Peter Maydell:
> On 8 December 2012 14:48, Andreas Färber <afaerber@suse.de> wrote:
>> Shouldn't we be updating the Coding Style on all lines we touch? The
>> shift line matches it but the normalize and ifs don't.
> 
> fpu/ is a special case because the softfloat library has such
> a bonkers coding style. I tend to update to qemu standard when
> I touch bits of it, but following the existing practice is
> also OK in my opinion.

The same applies to linux-user and much of ppc code (spaces before
parenthesis, no braces). The only agreed-on way to address this is by
fixing the parts we touch - no veto here, but seems trivial enough to
fix. Otherwise the situation won't ever improve.

Andreas

-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg