[PATCH v2] math: Fix x86_64 build for -Os (BZ 33367)

[Adhemerval Zanella Netto]

On 08/09/25 17:09, H.J. Lu wrote:
> On Mon, Sep 8, 2025 at 12:56 PM Adhemerval Zanella
> <adhemerval.zanella@linaro.org> wrote:
>>
>> The compiler might not inline the trunc function call for
>> USE_TRUNC_BUILTIN [1].
>>
>> This patch adds an optimized __trunc/__truncf for x86 used
>> on modf ifunc variant to avoid the trunc libcall.
>>
>> Checked on x86_64, x86_64-v2, x86_64-v3, and x86_64-v4. Used -O2 and
>> -Os options. Performed a full make check on x86_64 with both
>>  optimizations.
>>
>> [1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121861
>> ---
>>  sysdeps/x86/fpu/math_private.h                | 27 +++++++++++++++++++
>>  sysdeps/x86_64/fpu/multiarch/s_modf-avx.c     |  4 +++
>>  sysdeps/x86_64/fpu/multiarch/s_modf-sse4_1.c  |  4 +++
>>  sysdeps/x86_64/fpu/multiarch/s_modf.c         |  3 ++-
>>  sysdeps/x86_64/fpu/multiarch/s_modff-avx.c    |  4 +++
>>  sysdeps/x86_64/fpu/multiarch/s_modff-sse4_1.c |  4 +++
>>  sysdeps/x86_64/fpu/multiarch/s_modff.c        |  3 ++-
>>  7 files changed, 47 insertions(+), 2 deletions(-)
>>
>> diff --git a/sysdeps/x86/fpu/math_private.h b/sysdeps/x86/fpu/math_private.h
>> index 132f011809..d30d580cea 100644
>> --- a/sysdeps/x86/fpu/math_private.h
>> +++ b/sysdeps/x86/fpu/math_private.h
>> @@ -19,6 +19,7 @@
>>  #ifndef X86_MATH_PRIVATE_H
>>  #define X86_MATH_PRIVATE_H 1
>>
>> +#include <math.h>
>>  #include_next <math_private.h>
>>
>>  __extern_always_inline long double
>> @@ -29,4 +30,30 @@ __NTH (__ieee754_atan2l (long double y, long double x))
>>    return ret;
>>  }
>>
>> +__extern_always_inline double
>> +__trunc (double x)
>> +{
>> +#ifdef __AVX__
>> +  asm ("vroundsd $11, %1, %1, %0" : "=v" (x) : "v" (x));
>> +#elif defined __SSE4_1__
>> +  asm ("roundsd $11, %1, %0" : "=x" (x) : "x" (x));
>> +#else
>> +  x = trunc (x);
>> +#endif
>> +  return x;
>> +}
> 
> I will fix GCC 16.   Please make sure that inline asm isn't used for
> GCC 16 or above.

Would it make a lot of difference on modf implementation to add this extra
switch?

> 
>> +__extern_always_inline float
>> +__truncf (float x)
>> +{
>> +#ifdef __AVX__
>> +  asm ("vroundss $11, %1, %1, %0" : "=v" (x) : "v" (x));
>> +#elif defined __SSE4_1__
>> +  asm ("roundss $11, %1, %0" : "=x" (x) : "x" (x));
>> +#else
>> +  x = truncf (x);
>> +#endif
>> +  return x;
>> +}
>> +
>>  #endif
>> diff --git a/sysdeps/x86_64/fpu/multiarch/s_modf-avx.c b/sysdeps/x86_64/fpu/multiarch/s_modf-avx.c
>> index ab4f03db0e..2f84d254a9 100644
>> --- a/sysdeps/x86_64/fpu/multiarch/s_modf-avx.c
>> +++ b/sysdeps/x86_64/fpu/multiarch/s_modf-avx.c
>> @@ -1,3 +1,7 @@
>> +#include <sysdeps/x86/isa-level.h>
>> +#include <math_private.h>
>> +
>>  #define __modf __modf_avx
>> +#define trunc __trunc
>>
>>  #include <sysdeps/ieee754/dbl-64/s_modf.c>
>> diff --git a/sysdeps/x86_64/fpu/multiarch/s_modf-sse4_1.c b/sysdeps/x86_64/fpu/multiarch/s_modf-sse4_1.c
>> index 00aa8cd736..dcf04097f2 100644
>> --- a/sysdeps/x86_64/fpu/multiarch/s_modf-sse4_1.c
>> +++ b/sysdeps/x86_64/fpu/multiarch/s_modf-sse4_1.c
>> @@ -1,3 +1,7 @@
>> +#include <sysdeps/x86/isa-level.h>
>> +#include <math_private.h>
>> +
>>  #define __modf __modf_sse41
>> +#define trunc __trunc
>>
>>  #include <sysdeps/ieee754/dbl-64/s_modf.c>
>> diff --git a/sysdeps/x86_64/fpu/multiarch/s_modf.c b/sysdeps/x86_64/fpu/multiarch/s_modf.c
>> index e365bfcef7..61f77566b1 100644
>> --- a/sysdeps/x86_64/fpu/multiarch/s_modf.c
>> +++ b/sysdeps/x86_64/fpu/multiarch/s_modf.c
>> @@ -18,7 +18,6 @@
>>
>>  #include <sysdeps/x86/isa-level.h>
>>  #if MINIMUM_X86_ISA_LEVEL < AVX_X86_ISA_LEVEL
>> -# define NO_MATH_REDIRECT
>>  # include <libm-alias-double.h>
>>
>>  # define modf __redirect_modf
>> @@ -38,4 +37,6 @@ libm_alias_double (__modf, modf)
>>  #  define __modf __modf_sse2
>>  # endif
>>  #endif
>> +#include <math_private.h>
>> +#define trunc __trunc
>>  #include <sysdeps/ieee754/dbl-64/s_modf.c>
>> diff --git a/sysdeps/x86_64/fpu/multiarch/s_modff-avx.c b/sysdeps/x86_64/fpu/multiarch/s_modff-avx.c
>> index 07cb9c1036..c121ec26e2 100644
>> --- a/sysdeps/x86_64/fpu/multiarch/s_modff-avx.c
>> +++ b/sysdeps/x86_64/fpu/multiarch/s_modff-avx.c
>> @@ -1,3 +1,7 @@
>> +#include <sysdeps/x86/isa-level.h>
>> +#include <math_private.h>
>> +
>>  #define __modff __modff_avx
>> +#define truncf __truncf
>>
>>  #include <sysdeps/ieee754/flt-32/s_modff.c>
>> diff --git a/sysdeps/x86_64/fpu/multiarch/s_modff-sse4_1.c b/sysdeps/x86_64/fpu/multiarch/s_modff-sse4_1.c
>> index 060c5e3979..30a785d385 100644
>> --- a/sysdeps/x86_64/fpu/multiarch/s_modff-sse4_1.c
>> +++ b/sysdeps/x86_64/fpu/multiarch/s_modff-sse4_1.c
>> @@ -1,3 +1,7 @@
>> +#include <sysdeps/x86/isa-level.h>
>> +#include <math_private.h>
>> +
>>  #define __modff __modff_sse41
>> +#define truncf __truncf
>>
>>  #include <sysdeps/ieee754/flt-32/s_modff.c>
>> diff --git a/sysdeps/x86_64/fpu/multiarch/s_modff.c b/sysdeps/x86_64/fpu/multiarch/s_modff.c
>> index a4b5429037..ebf16777fd 100644
>> --- a/sysdeps/x86_64/fpu/multiarch/s_modff.c
>> +++ b/sysdeps/x86_64/fpu/multiarch/s_modff.c
>> @@ -18,7 +18,6 @@
>>
>>  #include <sysdeps/x86/isa-level.h>
>>  #if MINIMUM_X86_ISA_LEVEL < AVX_X86_ISA_LEVEL
>> -# define NO_MATH_REDIRECT
>>  # include <libm-alias-float.h>
>>
>>  # define modff __redirect_modff
>> @@ -38,4 +37,6 @@ libm_alias_float (__modf, modf)
>>  #  define __modff __modff_sse2
>>  # endif
>>  #endif
>> +#include <math_private.h>
>> +#define truncf __truncf
>>  #include <sysdeps/ieee754/flt-32/s_modff.c>
>> --
>> 2.43.0
>>
> 
>