[v4,5/5] math: Remove the error handling wrapper from fmod and fmodf

Message ID 20230320160118.352206-6-adhemerval.zanella@linaro.org
State Committed
Headers
Series Improve fmod and fmodf |

Checks

Context Check Description
dj/TryBot-32bit success Build for i686

Commit Message

Adhemerval Zanella Netto March 20, 2023, 4:01 p.m. UTC
  The error handling is moved to sysdeps/ieee754 version with no SVID
support.  The compatibility symbol versions still use the wrapper
with SVID error handling around the new code.  There is no new symbol
version nor compatibility code on !LIBM_SVID_COMPAT targets
(e.g. riscv).

The ia64 is unchanged, since it still uses the arch specific
__libm_error_region on its implementation.  For both i686 and m68k,
which provive arch specific implementation, wrappers are added so
no new symbol are added (which would require to change the
implementations).

It shows an small improvement, the results for fmod:

  Architecture     | Input           | master   | patch
  -----------------|-----------------|----------|--------
  x86_64 (Ryzen 9) | subnormals      | 12.5049  | 9.40992
  x86_64 (Ryzen 9) | normal          | 296.939  | 296.738
  x86_64 (Ryzen 9) | close-exponents | 16.0244  | 13.119
  aarch64 (N1)     | subnormal       | 6.81778  | 4.33313
  aarch64 (N1)     | normal          | 155.620  | 152.915
  aarch64 (N1)     | close-exponents | 8.21306  | 5.76138
  armhf (N1)       | subnormal       | 15.1083  | 14.5746
  armhf (N1)       | normal          | 244.833  | 241.738
  armhf (N1)       | close-exponents | 21.8182  | 22.457

Checked on x86_64-linux-gnu, i686-linux-gnu, and aarch64-linux-gnu.
---
 math/Versions                                 |  4 ++++
 math/w_fmod_compat.c                          | 13 ++++++++---
 math/w_fmodf_compat.c                         |  6 ++---
 sysdeps/i386/fpu/w_fmod_compat.c              | 14 ++++++++++++
 sysdeps/i386/fpu/w_fmodf_compat.c             | 14 ++++++++++++
 sysdeps/ieee754/dbl-64/e_fmod.c               | 22 +++++++++++++++----
 sysdeps/ieee754/dbl-64/math_config.h          |  9 ++++++++
 sysdeps/ieee754/dbl-64/math_err.c             |  6 +++++
 sysdeps/ieee754/dbl-64/w_fmod.c               |  1 +
 sysdeps/ieee754/flt-32/e_fmodf.c              | 20 ++++++++++++++---
 sysdeps/ieee754/flt-32/math_config.h          |  7 ++++++
 sysdeps/ieee754/flt-32/math_errf.c            |  6 +++++
 sysdeps/ieee754/flt-32/w_fmodf.c              |  1 +
 sysdeps/m68k/m680x0/fpu/w_fmod_compat.c       | 14 ++++++++++++
 sysdeps/m68k/m680x0/fpu/w_fmodf_compat.c      | 14 ++++++++++++
 sysdeps/unix/sysv/linux/aarch64/libm.abilist  |  2 ++
 sysdeps/unix/sysv/linux/alpha/libm.abilist    |  2 ++
 sysdeps/unix/sysv/linux/arm/be/libm.abilist   |  2 ++
 sysdeps/unix/sysv/linux/arm/le/libm.abilist   |  2 ++
 sysdeps/unix/sysv/linux/hppa/libm.abilist     |  2 ++
 .../sysv/linux/m68k/coldfire/libm.abilist     |  2 ++
 .../sysv/linux/microblaze/be/libm.abilist     |  2 ++
 .../sysv/linux/microblaze/le/libm.abilist     |  2 ++
 .../unix/sysv/linux/mips/mips32/libm.abilist  |  2 ++
 .../unix/sysv/linux/mips/mips64/libm.abilist  |  2 ++
 sysdeps/unix/sysv/linux/nios2/libm.abilist    |  2 ++
 .../linux/powerpc/powerpc32/fpu/libm.abilist  |  2 ++
 .../powerpc/powerpc32/nofpu/libm.abilist      |  2 ++
 .../linux/powerpc/powerpc64/be/libm.abilist   |  2 ++
 .../linux/powerpc/powerpc64/le/libm.abilist   |  2 ++
 .../unix/sysv/linux/s390/s390-32/libm.abilist |  2 ++
 .../unix/sysv/linux/s390/s390-64/libm.abilist |  2 ++
 sysdeps/unix/sysv/linux/sh/be/libm.abilist    |  2 ++
 sysdeps/unix/sysv/linux/sh/le/libm.abilist    |  2 ++
 .../sysv/linux/sparc/sparc32/libm.abilist     |  2 ++
 .../sysv/linux/sparc/sparc64/libm.abilist     |  2 ++
 .../unix/sysv/linux/x86_64/64/libm.abilist    |  2 ++
 .../unix/sysv/linux/x86_64/x32/libm.abilist   |  2 ++
 38 files changed, 184 insertions(+), 13 deletions(-)
 create mode 100644 sysdeps/i386/fpu/w_fmod_compat.c
 create mode 100644 sysdeps/i386/fpu/w_fmodf_compat.c
 create mode 100644 sysdeps/ieee754/dbl-64/w_fmod.c
 create mode 100644 sysdeps/ieee754/flt-32/w_fmodf.c
 create mode 100644 sysdeps/m68k/m680x0/fpu/w_fmod_compat.c
 create mode 100644 sysdeps/m68k/m680x0/fpu/w_fmodf_compat.c
  

Comments

Wilco Dijkstra April 3, 2023, 1:43 p.m. UTC | #1
Hi Adhemerval,

LGTM - one minor nit about the is_inf functions, are they actually needed?
I'm happy for this to go in either way.

Reviewed-by: Wilco Dijkstra  <Wilco.Dijkstra@arm.com>

Cheers,
Wilco


diff --git a/sysdeps/ieee754/dbl-64/e_fmod.c b/sysdeps/ieee754/dbl-64/e_fmod.c
index d20466fd5d..ff506c40b4 100644
--- a/sysdeps/ieee754/dbl-64/e_fmod.c
+++ b/sysdeps/ieee754/dbl-64/e_fmod.c
@@ -16,7 +16,9 @@
    License along with the GNU C Library; if not, see
    <https://www.gnu.org/licenses/>.  */
 
+#include <libm-alias-double.h>
 #include <libm-alias-finite.h>
+#include <math-svid-compat.h>
 #include <math.h>
 #include "math_config.h"
 
@@ -44,7 +46,7 @@
      }  */
 
 double
-__ieee754_fmod (double x, double y)
+__fmod (double x, double y)
 {
   uint64_t hx = asuint64 (x);
   uint64_t hy = asuint64 (y);
@@ -56,11 +58,16 @@ __ieee754_fmod (double x, double y)
 
   /* Special cases:
      - If x or y is a Nan, NaN is returned.
-     - If x is an inifinity, a NaN is returned.
+     - If x is an inifinity, a NaN is returned and EDOM is set.
      - If y is zero, Nan is returned.
      - If x is +0/-0, and y is not zero, +0/-0 is returned.  */
-  if (__glibc_unlikely (hy == 0        || hx >= EXPONENT_MASK || hy > EXPONENT_MASK))
-    return (x * y) / (x * y);
+  if (__glibc_unlikely (hy == 0
+                       || hx >= EXPONENT_MASK || hy > EXPONENT_MASK))
+    {
+      if (is_nan (hx) || is_nan (hy))
+       return (x * y) / (x * y);
+      return __math_edom ((x * y) / (x * y));
+    }

OK

   if (__glibc_unlikely (hx <= hy))
     {
@@ -142,4 +149,11 @@ __ieee754_fmod (double x, double y)
 
   return make_double (mx, ey, sx);
 }
+strong_alias (__fmod, __ieee754_fmod)
 libm_alias_finite (__ieee754_fmod, __fmod)
+#if LIBM_SVID_COMPAT
+versioned_symbol (libm, __fmod, fmod, GLIBC_2_38);
+libm_alias_double_other (__fmod, fmod)
+#else
+libm_alias_double (__fmod, fmod)
+#endif
diff --git a/sysdeps/ieee754/dbl-64/math_config.h b/sysdeps/ieee754/dbl-64/math_config.h
index 2049cea3f7..5173b99e4a 100644
--- a/sysdeps/ieee754/dbl-64/math_config.h
+++ b/sysdeps/ieee754/dbl-64/math_config.h
@@ -121,6 +121,12 @@ is_nan (uint64_t x)
   return (x & EXP_MANT_MASK) > EXPONENT_MASK;
 }
 
+static inline bool
+is_inf (uint64_t x)
+{
+  return (x & EXP_MANT_MASK) == EXPONENT_MASK;
+}

Is this used?

 static inline uint64_t
 get_mantissa (uint64_t x)
 {
@@ -170,6 +176,9 @@ attribute_hidden double __math_invalid (double);
 
 /* Error handling using output checking, only for errno setting.  */
 
+/* Check if the result generated a demain error.  */
+attribute_hidden double __math_edom (double x);
+
 /* Check if the result overflowed to infinity.  */
 attribute_hidden double __math_check_oflow (double);
 /* Check if the result underflowed to 0.  */
diff --git a/sysdeps/ieee754/dbl-64/math_err.c b/sysdeps/ieee754/dbl-64/math_err.c
index 5f5f965d3e..377a4caad2 100644
--- a/sysdeps/ieee754/dbl-64/math_err.c
+++ b/sysdeps/ieee754/dbl-64/math_err.c
@@ -33,6 +33,12 @@ with_errno (double y, int e)
 #define with_errno(x, e) (x)
 #endif
 
+attribute_hidden double
+__math_edom (double y)
+{
+  return with_errno (y, EDOM);
+}
+
 /* NOINLINE reduces code size.  */
 NOINLINE static double
 xflow (uint32_t sign, double y)
diff --git a/sysdeps/ieee754/dbl-64/w_fmod.c b/sysdeps/ieee754/dbl-64/w_fmod.c
new file mode 100644
index 0000000000..1cc8931700
--- /dev/null
+++ b/sysdeps/ieee754/dbl-64/w_fmod.c
@@ -0,0 +1 @@
+/* Not needed.  */
diff --git a/sysdeps/ieee754/flt-32/e_fmodf.c b/sysdeps/ieee754/flt-32/e_fmodf.c
index 53db32cbe2..c0b5dccc23 100644
--- a/sysdeps/ieee754/flt-32/e_fmodf.c
+++ b/sysdeps/ieee754/flt-32/e_fmodf.c
@@ -17,6 +17,8 @@
    <https://www.gnu.org/licenses/>.  */
 
 #include <libm-alias-finite.h>
+#include <libm-alias-float.h>
+#include <math-svid-compat.h>
 #include <math.h>
 #include "math_config.h"
 
@@ -44,7 +46,7 @@
      }  */
 
 float
-__ieee754_fmodf (float x, float y)
+__fmodf (float x, float y)
 {
   uint32_t hx = asuint (x);
   uint32_t hy = asuint (y);
@@ -59,8 +61,13 @@ __ieee754_fmodf (float x, float y)
      - If x is an inifinity, a NaN is returned.
      - If y is zero, Nan is returned.
      - If x is +0/-0, and y is not zero, +0/-0 is returned.  */
-  if (__glibc_unlikely (hy == 0        || hx >= EXPONENT_MASK || hy > EXPONENT_MASK))
-    return (x * y) / (x * y);
+  if (__glibc_unlikely (hy == 0
+                       || hx >= EXPONENT_MASK || hy > EXPONENT_MASK))
+    {
+      if (is_nan (hx) || is_nan (hy))
+       return (x * y) / (x * y);
+      return __math_edomf ((x * y) / (x * y));
+    }

OK

   if (__glibc_unlikely (hx <= hy))
     {
@@ -141,4 +148,11 @@ __ieee754_fmodf (float x, float y)
 
   return make_float (mx, ey, sx);
 }
+strong_alias (__fmodf, __ieee754_fmodf)
+#if LIBM_SVID_COMPAT
+versioned_symbol (libm, __fmodf, fmodf, GLIBC_2_38);
+libm_alias_float_other (__fmod, fmod)
+#else
+libm_alias_float (__fmod, fmod)
+#endif
 libm_alias_finite (__ieee754_fmodf, __fmodf)
diff --git a/sysdeps/ieee754/flt-32/math_config.h b/sysdeps/ieee754/flt-32/math_config.h
index 829430ea28..f97cd39df1 100644
--- a/sysdeps/ieee754/flt-32/math_config.h
+++ b/sysdeps/ieee754/flt-32/math_config.h
@@ -125,6 +125,12 @@ is_nan (uint32_t x)
   return (x & EXP_MANT_MASK) > EXPONENT_MASK;
 }
 
+static inline bool
+is_inf (uint32_t x)
+{
+  return (x & EXP_MANT_MASK) == EXPONENT_MASK;
+}

Used?
  
Adhemerval Zanella Netto April 3, 2023, 6:33 p.m. UTC | #2
On 03/04/23 10:43, Wilco Dijkstra wrote:
> Hi Adhemerval,
> 
> LGTM - one minor nit about the is_inf functions, are they actually needed?
> I'm happy for this to go in either way.
> 
> Reviewed-by: Wilco Dijkstra  <Wilco.Dijkstra@arm.com>
> 
> Cheers,
> Wilco
> 

> diff --git a/sysdeps/ieee754/flt-32/math_config.h b/sysdeps/ieee754/flt-32/math_config.h
> index 829430ea28..f97cd39df1 100644
> --- a/sysdeps/ieee754/flt-32/math_config.h
> +++ b/sysdeps/ieee754/flt-32/math_config.h
> @@ -125,6 +125,12 @@ is_nan (uint32_t x)
>    return (x & EXP_MANT_MASK) > EXPONENT_MASK;
>  }
>  
> +static inline bool
> +is_inf (uint32_t x)
> +{
> +  return (x & EXP_MANT_MASK) == EXPONENT_MASK;
> +}
> 
> Used?

It is a left-over from previous version, I will remove it.
  

Patch

diff --git a/math/Versions b/math/Versions
index fef7a08c3b..759b5fac7f 100644
--- a/math/Versions
+++ b/math/Versions
@@ -631,4 +631,8 @@  libm {
     # No SVID compatible error handling.
     hypotf; hypot;
   }
+  GLIBC_2.38 {
+    # No SVID compatible error handling.
+    fmod; fmodf;
+  }
 }
diff --git a/math/w_fmod_compat.c b/math/w_fmod_compat.c
index c6947f3042..aef98885a3 100644
--- a/math/w_fmod_compat.c
+++ b/math/w_fmod_compat.c
@@ -20,10 +20,10 @@ 
 #include <math-svid-compat.h>
 #include <libm-alias-double.h>
 
-#if LIBM_SVID_COMPAT
+#if LIBM_SVID_COMPAT && SHLIB_COMPAT (libm, GLIBC_2_0, GLIBC_2_38)
 /* wrapper fmod */
 double
-__fmod (double x, double y)
+__fmod_compat (double x, double y)
 {
   if (__builtin_expect (isinf (x) || y == 0.0, 0)
       && _LIB_VERSION != _IEEE_ && !isnan (y) && !isnan (x))
@@ -32,5 +32,12 @@  __fmod (double x, double y)
 
   return __ieee754_fmod (x, y);
 }
-libm_alias_double (__fmod, fmod)
+compat_symbol (libm, __fmod_compat, fmod, GLIBC_2_0);
+# ifdef NO_LONG_DOUBLE
+weak_alias (__fmod_compat, fmodl)
+# endif
+# ifdef LONG_DOUBLE_COMPAT
+LONG_DOUBLE_COMPAT_CHOOSE_libm_fmodl (
+  compat_symbol (libm, __fmod_compat, fmodl, FIRST_VERSION_libm_fmodl), );
+# endif
 #endif
diff --git a/math/w_fmodf_compat.c b/math/w_fmodf_compat.c
index 5025b06684..55823961f4 100644
--- a/math/w_fmodf_compat.c
+++ b/math/w_fmodf_compat.c
@@ -20,10 +20,10 @@ 
 #include <math-svid-compat.h>
 #include <libm-alias-float.h>
 
-#if LIBM_SVID_COMPAT
+#if LIBM_SVID_COMPAT && SHLIB_COMPAT (libm, GLIBC_2_0, GLIBC_2_38)
 /* wrapper fmodf */
 float
-__fmodf (float x, float y)
+__fmod_compatf (float x, float y)
 {
   if (__builtin_expect (isinf (x) || y == 0.0f, 0)
       && _LIB_VERSION != _IEEE_ && !isnan (y) && !isnan (x))
@@ -32,5 +32,5 @@  __fmodf (float x, float y)
 
   return __ieee754_fmodf (x, y);
 }
-libm_alias_float (__fmod, fmod)
+compat_symbol (libm, __fmod_compatf, fmodf, GLIBC_2_0);
 #endif
diff --git a/sysdeps/i386/fpu/w_fmod_compat.c b/sysdeps/i386/fpu/w_fmod_compat.c
new file mode 100644
index 0000000000..5ac9995ffd
--- /dev/null
+++ b/sysdeps/i386/fpu/w_fmod_compat.c
@@ -0,0 +1,14 @@ 
+/* i386 provides an optimized __ieee752_fmod.  */
+#include <math-svid-compat.h>
+#ifdef SHARED
+# undef SHLIB_COMPAT
+# define SHLIB_COMPAT(a, b, c) 1
+# undef LIBM_SVID_COMPAT
+# define LIBM_SVID_COMPAT 1
+# undef compat_symbol
+# define compat_symbol(a, b, c, d)
+#endif
+#include <math/w_fmod_compat.c>
+#ifdef SHARED
+libm_alias_double (__fmod_compat, fmod)
+#endif
diff --git a/sysdeps/i386/fpu/w_fmodf_compat.c b/sysdeps/i386/fpu/w_fmodf_compat.c
new file mode 100644
index 0000000000..cc417e07d3
--- /dev/null
+++ b/sysdeps/i386/fpu/w_fmodf_compat.c
@@ -0,0 +1,14 @@ 
+/* i386 provides an optimized __ieee752_fmodf.  */
+#include <math-svid-compat.h>
+#ifdef SHARED
+# undef SHLIB_COMPAT
+# define SHLIB_COMPAT(a, b, c) 1
+# undef LIBM_SVID_COMPAT
+# define LIBM_SVID_COMPAT 1
+# undef compat_symbol
+# define compat_symbol(a, b, c, d)
+#endif
+#include <math/w_fmodf_compat.c>
+#ifdef SHARED
+libm_alias_float (__fmod_compat, fmod)
+#endif
diff --git a/sysdeps/ieee754/dbl-64/e_fmod.c b/sysdeps/ieee754/dbl-64/e_fmod.c
index d20466fd5d..ff506c40b4 100644
--- a/sysdeps/ieee754/dbl-64/e_fmod.c
+++ b/sysdeps/ieee754/dbl-64/e_fmod.c
@@ -16,7 +16,9 @@ 
    License along with the GNU C Library; if not, see
    <https://www.gnu.org/licenses/>.  */
 
+#include <libm-alias-double.h>
 #include <libm-alias-finite.h>
+#include <math-svid-compat.h>
 #include <math.h>
 #include "math_config.h"
 
@@ -44,7 +46,7 @@ 
      }  */
 
 double
-__ieee754_fmod (double x, double y)
+__fmod (double x, double y)
 {
   uint64_t hx = asuint64 (x);
   uint64_t hy = asuint64 (y);
@@ -56,11 +58,16 @@  __ieee754_fmod (double x, double y)
 
   /* Special cases:
      - If x or y is a Nan, NaN is returned.
-     - If x is an inifinity, a NaN is returned.
+     - If x is an inifinity, a NaN is returned and EDOM is set.
      - If y is zero, Nan is returned.
      - If x is +0/-0, and y is not zero, +0/-0 is returned.  */
-  if (__glibc_unlikely (hy == 0	|| hx >= EXPONENT_MASK || hy > EXPONENT_MASK))
-    return (x * y) / (x * y);
+  if (__glibc_unlikely (hy == 0
+			|| hx >= EXPONENT_MASK || hy > EXPONENT_MASK))
+    {
+      if (is_nan (hx) || is_nan (hy))
+	return (x * y) / (x * y);
+      return __math_edom ((x * y) / (x * y));
+    }
 
   if (__glibc_unlikely (hx <= hy))
     {
@@ -142,4 +149,11 @@  __ieee754_fmod (double x, double y)
 
   return make_double (mx, ey, sx);
 }
+strong_alias (__fmod, __ieee754_fmod)
 libm_alias_finite (__ieee754_fmod, __fmod)
+#if LIBM_SVID_COMPAT
+versioned_symbol (libm, __fmod, fmod, GLIBC_2_38);
+libm_alias_double_other (__fmod, fmod)
+#else
+libm_alias_double (__fmod, fmod)
+#endif
diff --git a/sysdeps/ieee754/dbl-64/math_config.h b/sysdeps/ieee754/dbl-64/math_config.h
index 2049cea3f7..5173b99e4a 100644
--- a/sysdeps/ieee754/dbl-64/math_config.h
+++ b/sysdeps/ieee754/dbl-64/math_config.h
@@ -121,6 +121,12 @@  is_nan (uint64_t x)
   return (x & EXP_MANT_MASK) > EXPONENT_MASK;
 }
 
+static inline bool
+is_inf (uint64_t x)
+{
+  return (x & EXP_MANT_MASK) == EXPONENT_MASK;
+}
+
 static inline uint64_t
 get_mantissa (uint64_t x)
 {
@@ -170,6 +176,9 @@  attribute_hidden double __math_invalid (double);
 
 /* Error handling using output checking, only for errno setting.  */
 
+/* Check if the result generated a demain error.  */
+attribute_hidden double __math_edom (double x);
+
 /* Check if the result overflowed to infinity.  */
 attribute_hidden double __math_check_oflow (double);
 /* Check if the result underflowed to 0.  */
diff --git a/sysdeps/ieee754/dbl-64/math_err.c b/sysdeps/ieee754/dbl-64/math_err.c
index 5f5f965d3e..377a4caad2 100644
--- a/sysdeps/ieee754/dbl-64/math_err.c
+++ b/sysdeps/ieee754/dbl-64/math_err.c
@@ -33,6 +33,12 @@  with_errno (double y, int e)
 #define with_errno(x, e) (x)
 #endif
 
+attribute_hidden double
+__math_edom (double y)
+{
+  return with_errno (y, EDOM);
+}
+
 /* NOINLINE reduces code size.  */
 NOINLINE static double
 xflow (uint32_t sign, double y)
diff --git a/sysdeps/ieee754/dbl-64/w_fmod.c b/sysdeps/ieee754/dbl-64/w_fmod.c
new file mode 100644
index 0000000000..1cc8931700
--- /dev/null
+++ b/sysdeps/ieee754/dbl-64/w_fmod.c
@@ -0,0 +1 @@ 
+/* Not needed.  */
diff --git a/sysdeps/ieee754/flt-32/e_fmodf.c b/sysdeps/ieee754/flt-32/e_fmodf.c
index 53db32cbe2..c0b5dccc23 100644
--- a/sysdeps/ieee754/flt-32/e_fmodf.c
+++ b/sysdeps/ieee754/flt-32/e_fmodf.c
@@ -17,6 +17,8 @@ 
    <https://www.gnu.org/licenses/>.  */
 
 #include <libm-alias-finite.h>
+#include <libm-alias-float.h>
+#include <math-svid-compat.h>
 #include <math.h>
 #include "math_config.h"
 
@@ -44,7 +46,7 @@ 
      }  */
 
 float
-__ieee754_fmodf (float x, float y)
+__fmodf (float x, float y)
 {
   uint32_t hx = asuint (x);
   uint32_t hy = asuint (y);
@@ -59,8 +61,13 @@  __ieee754_fmodf (float x, float y)
      - If x is an inifinity, a NaN is returned.
      - If y is zero, Nan is returned.
      - If x is +0/-0, and y is not zero, +0/-0 is returned.  */
-  if (__glibc_unlikely (hy == 0	|| hx >= EXPONENT_MASK || hy > EXPONENT_MASK))
-    return (x * y) / (x * y);
+  if (__glibc_unlikely (hy == 0
+			|| hx >= EXPONENT_MASK || hy > EXPONENT_MASK))
+    {
+      if (is_nan (hx) || is_nan (hy))
+	return (x * y) / (x * y);
+      return __math_edomf ((x * y) / (x * y));
+    }
 
   if (__glibc_unlikely (hx <= hy))
     {
@@ -141,4 +148,11 @@  __ieee754_fmodf (float x, float y)
 
   return make_float (mx, ey, sx);
 }
+strong_alias (__fmodf, __ieee754_fmodf)
+#if LIBM_SVID_COMPAT
+versioned_symbol (libm, __fmodf, fmodf, GLIBC_2_38);
+libm_alias_float_other (__fmod, fmod)
+#else
+libm_alias_float (__fmod, fmod)
+#endif
 libm_alias_finite (__ieee754_fmodf, __fmodf)
diff --git a/sysdeps/ieee754/flt-32/math_config.h b/sysdeps/ieee754/flt-32/math_config.h
index 829430ea28..f97cd39df1 100644
--- a/sysdeps/ieee754/flt-32/math_config.h
+++ b/sysdeps/ieee754/flt-32/math_config.h
@@ -125,6 +125,12 @@  is_nan (uint32_t x)
   return (x & EXP_MANT_MASK) > EXPONENT_MASK;
 }
 
+static inline bool
+is_inf (uint32_t x)
+{
+  return (x & EXP_MANT_MASK) == EXPONENT_MASK;
+}
+
 static inline uint32_t
 get_mantissa (uint32_t x)
 {
@@ -158,6 +164,7 @@  attribute_hidden float __math_uflowf (uint32_t);
 attribute_hidden float __math_may_uflowf (uint32_t);
 attribute_hidden float __math_divzerof (uint32_t);
 attribute_hidden float __math_invalidf (float);
+attribute_hidden float __math_edomf (float x);
 
 /* Shared between expf, exp2f, exp10f, and powf.  */
 #define EXP2F_TABLE_BITS 5
diff --git a/sysdeps/ieee754/flt-32/math_errf.c b/sysdeps/ieee754/flt-32/math_errf.c
index 0389ce0a81..7245ab8d96 100644
--- a/sysdeps/ieee754/flt-32/math_errf.c
+++ b/sysdeps/ieee754/flt-32/math_errf.c
@@ -31,6 +31,12 @@  with_errnof (float y, int e)
 # define with_errnof(x, e) (x)
 #endif
 
+attribute_hidden float
+__math_edomf (float y)
+{
+  return with_errnof (y, EDOM);
+}
+
 /* NOINLINE prevents fenv semantics breaking optimizations.  */
 NOINLINE static float
 xflowf (uint32_t sign, float y)
diff --git a/sysdeps/ieee754/flt-32/w_fmodf.c b/sysdeps/ieee754/flt-32/w_fmodf.c
new file mode 100644
index 0000000000..1cc8931700
--- /dev/null
+++ b/sysdeps/ieee754/flt-32/w_fmodf.c
@@ -0,0 +1 @@ 
+/* Not needed.  */
diff --git a/sysdeps/m68k/m680x0/fpu/w_fmod_compat.c b/sysdeps/m68k/m680x0/fpu/w_fmod_compat.c
new file mode 100644
index 0000000000..527d4fbed2
--- /dev/null
+++ b/sysdeps/m68k/m680x0/fpu/w_fmod_compat.c
@@ -0,0 +1,14 @@ 
+/* m68k provides an optimized __ieee752_fmod.  */
+#include <math-svid-compat.h>
+#ifdef SHARED
+# undef SHLIB_COMPAT
+# define SHLIB_COMPAT(a, b, c) 1
+# undef LIBM_SVID_COMPAT
+# define LIBM_SVID_COMPAT 1
+# undef compat_symbol
+# define compat_symbol(a, b, c, d)
+#endif
+#include <math/w_fmod_compat.c>
+#ifdef SHARED
+libm_alias_double (__fmod_compat, fmod)
+#endif
diff --git a/sysdeps/m68k/m680x0/fpu/w_fmodf_compat.c b/sysdeps/m68k/m680x0/fpu/w_fmodf_compat.c
new file mode 100644
index 0000000000..5043586b91
--- /dev/null
+++ b/sysdeps/m68k/m680x0/fpu/w_fmodf_compat.c
@@ -0,0 +1,14 @@ 
+/* m68k provides an optimized __ieee752_fmodf.  */
+#include <math-svid-compat.h>
+#ifdef SHARED
+# undef SHLIB_COMPAT
+# define SHLIB_COMPAT(a, b, c) 1
+# undef LIBM_SVID_COMPAT
+# define LIBM_SVID_COMPAT 1
+# undef compat_symbol
+# define compat_symbol(a, b, c, d)
+#endif
+#include <math/w_fmodf_compat.c>
+#ifdef SHARED
+libm_alias_float (__fmod_compat, fmod)
+#endif
diff --git a/sysdeps/unix/sysv/linux/aarch64/libm.abilist b/sysdeps/unix/sysv/linux/aarch64/libm.abilist
index c2e3c6453e..e3924c3499 100644
--- a/sysdeps/unix/sysv/linux/aarch64/libm.abilist
+++ b/sysdeps/unix/sysv/linux/aarch64/libm.abilist
@@ -1146,3 +1146,5 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
diff --git a/sysdeps/unix/sysv/linux/alpha/libm.abilist b/sysdeps/unix/sysv/linux/alpha/libm.abilist
index 4f85b6180f..066dd1a6b3 100644
--- a/sysdeps/unix/sysv/linux/alpha/libm.abilist
+++ b/sysdeps/unix/sysv/linux/alpha/libm.abilist
@@ -1203,6 +1203,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 __clog10l F
 GLIBC_2.4 __finitel F
 GLIBC_2.4 __fpclassifyl F
diff --git a/sysdeps/unix/sysv/linux/arm/be/libm.abilist b/sysdeps/unix/sysv/linux/arm/be/libm.abilist
index 36190add84..4adfed072b 100644
--- a/sysdeps/unix/sysv/linux/arm/be/libm.abilist
+++ b/sysdeps/unix/sysv/linux/arm/be/libm.abilist
@@ -533,6 +533,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 _LIB_VERSION D 0x4
 GLIBC_2.4 __clog10 F
 GLIBC_2.4 __clog10f F
diff --git a/sysdeps/unix/sysv/linux/arm/le/libm.abilist b/sysdeps/unix/sysv/linux/arm/le/libm.abilist
index 36190add84..4adfed072b 100644
--- a/sysdeps/unix/sysv/linux/arm/le/libm.abilist
+++ b/sysdeps/unix/sysv/linux/arm/le/libm.abilist
@@ -533,6 +533,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 _LIB_VERSION D 0x4
 GLIBC_2.4 __clog10 F
 GLIBC_2.4 __clog10f F
diff --git a/sysdeps/unix/sysv/linux/hppa/libm.abilist b/sysdeps/unix/sysv/linux/hppa/libm.abilist
index b5dd4e851f..5c3f47caef 100644
--- a/sysdeps/unix/sysv/linux/hppa/libm.abilist
+++ b/sysdeps/unix/sysv/linux/hppa/libm.abilist
@@ -844,4 +844,6 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 exp2l F
diff --git a/sysdeps/unix/sysv/linux/m68k/coldfire/libm.abilist b/sysdeps/unix/sysv/linux/m68k/coldfire/libm.abilist
index 36190add84..4adfed072b 100644
--- a/sysdeps/unix/sysv/linux/m68k/coldfire/libm.abilist
+++ b/sysdeps/unix/sysv/linux/m68k/coldfire/libm.abilist
@@ -533,6 +533,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 _LIB_VERSION D 0x4
 GLIBC_2.4 __clog10 F
 GLIBC_2.4 __clog10f F
diff --git a/sysdeps/unix/sysv/linux/microblaze/be/libm.abilist b/sysdeps/unix/sysv/linux/microblaze/be/libm.abilist
index 274ecff630..ea0dbd4b72 100644
--- a/sysdeps/unix/sysv/linux/microblaze/be/libm.abilist
+++ b/sysdeps/unix/sysv/linux/microblaze/be/libm.abilist
@@ -845,3 +845,5 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
diff --git a/sysdeps/unix/sysv/linux/microblaze/le/libm.abilist b/sysdeps/unix/sysv/linux/microblaze/le/libm.abilist
index 274ecff630..ea0dbd4b72 100644
--- a/sysdeps/unix/sysv/linux/microblaze/le/libm.abilist
+++ b/sysdeps/unix/sysv/linux/microblaze/le/libm.abilist
@@ -845,3 +845,5 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
diff --git a/sysdeps/unix/sysv/linux/mips/mips32/libm.abilist b/sysdeps/unix/sysv/linux/mips/mips32/libm.abilist
index 08b902118d..7c94a8db70 100644
--- a/sysdeps/unix/sysv/linux/mips/mips32/libm.abilist
+++ b/sysdeps/unix/sysv/linux/mips/mips32/libm.abilist
@@ -844,4 +844,6 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 exp2l F
diff --git a/sysdeps/unix/sysv/linux/mips/mips64/libm.abilist b/sysdeps/unix/sysv/linux/mips/mips64/libm.abilist
index 09bb3bd75b..0d08245adf 100644
--- a/sysdeps/unix/sysv/linux/mips/mips64/libm.abilist
+++ b/sysdeps/unix/sysv/linux/mips/mips64/libm.abilist
@@ -1146,3 +1146,5 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
diff --git a/sysdeps/unix/sysv/linux/nios2/libm.abilist b/sysdeps/unix/sysv/linux/nios2/libm.abilist
index 11abbb5668..a3b776fece 100644
--- a/sysdeps/unix/sysv/linux/nios2/libm.abilist
+++ b/sysdeps/unix/sysv/linux/nios2/libm.abilist
@@ -845,3 +845,5 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist
index 1688809c36..c696215739 100644
--- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist
+++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libm.abilist
@@ -890,6 +890,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 __clog10l F
 GLIBC_2.4 __finitel F
 GLIBC_2.4 __fpclassifyl F
diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist
index e880cebd78..a5ce9b1997 100644
--- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist
+++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libm.abilist
@@ -889,6 +889,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 __clog10l F
 GLIBC_2.4 __finitel F
 GLIBC_2.4 __fpclassifyl F
diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc64/be/libm.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc64/be/libm.abilist
index 033385dfc1..8e41056d0d 100644
--- a/sysdeps/unix/sysv/linux/powerpc/powerpc64/be/libm.abilist
+++ b/sysdeps/unix/sysv/linux/powerpc/powerpc64/be/libm.abilist
@@ -883,6 +883,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 __clog10l F
 GLIBC_2.4 __finitel F
 GLIBC_2.4 __fpclassifyl F
diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc64/le/libm.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc64/le/libm.abilist
index 7923d428bc..939f29a0f2 100644
--- a/sysdeps/unix/sysv/linux/powerpc/powerpc64/le/libm.abilist
+++ b/sysdeps/unix/sysv/linux/powerpc/powerpc64/le/libm.abilist
@@ -1318,3 +1318,5 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
diff --git a/sysdeps/unix/sysv/linux/s390/s390-32/libm.abilist b/sysdeps/unix/sysv/linux/s390/s390-32/libm.abilist
index 9a84163089..4baefb1217 100644
--- a/sysdeps/unix/sysv/linux/s390/s390-32/libm.abilist
+++ b/sysdeps/unix/sysv/linux/s390/s390-32/libm.abilist
@@ -1147,6 +1147,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 __clog10l F
 GLIBC_2.4 __finitel F
 GLIBC_2.4 __fpclassifyl F
diff --git a/sysdeps/unix/sysv/linux/s390/s390-64/libm.abilist b/sysdeps/unix/sysv/linux/s390/s390-64/libm.abilist
index 174bde4fa0..debb767575 100644
--- a/sysdeps/unix/sysv/linux/s390/s390-64/libm.abilist
+++ b/sysdeps/unix/sysv/linux/s390/s390-64/libm.abilist
@@ -1147,6 +1147,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 __clog10l F
 GLIBC_2.4 __finitel F
 GLIBC_2.4 __fpclassifyl F
diff --git a/sysdeps/unix/sysv/linux/sh/be/libm.abilist b/sysdeps/unix/sysv/linux/sh/be/libm.abilist
index 1e1324d667..fb94386fff 100644
--- a/sysdeps/unix/sysv/linux/sh/be/libm.abilist
+++ b/sysdeps/unix/sysv/linux/sh/be/libm.abilist
@@ -844,4 +844,6 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 exp2l F
diff --git a/sysdeps/unix/sysv/linux/sh/le/libm.abilist b/sysdeps/unix/sysv/linux/sh/le/libm.abilist
index 1e1324d667..fb94386fff 100644
--- a/sysdeps/unix/sysv/linux/sh/le/libm.abilist
+++ b/sysdeps/unix/sysv/linux/sh/le/libm.abilist
@@ -844,4 +844,6 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 exp2l F
diff --git a/sysdeps/unix/sysv/linux/sparc/sparc32/libm.abilist b/sysdeps/unix/sysv/linux/sparc/sparc32/libm.abilist
index 217e6eff7f..2fdb5ff145 100644
--- a/sysdeps/unix/sysv/linux/sparc/sparc32/libm.abilist
+++ b/sysdeps/unix/sysv/linux/sparc/sparc32/libm.abilist
@@ -1154,6 +1154,8 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
 GLIBC_2.4 __clog10l F
 GLIBC_2.4 __finitel F
 GLIBC_2.4 __fpclassifyl F
diff --git a/sysdeps/unix/sysv/linux/sparc/sparc64/libm.abilist b/sysdeps/unix/sysv/linux/sparc/sparc64/libm.abilist
index 6b53b0c59f..02d4c3a754 100644
--- a/sysdeps/unix/sysv/linux/sparc/sparc64/libm.abilist
+++ b/sysdeps/unix/sysv/linux/sparc/sparc64/libm.abilist
@@ -1146,3 +1146,5 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
diff --git a/sysdeps/unix/sysv/linux/x86_64/64/libm.abilist b/sysdeps/unix/sysv/linux/x86_64/64/libm.abilist
index dbefbc3a1a..cc8be5b510 100644
--- a/sysdeps/unix/sysv/linux/x86_64/64/libm.abilist
+++ b/sysdeps/unix/sysv/linux/x86_64/64/libm.abilist
@@ -1179,3 +1179,5 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F
diff --git a/sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist b/sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist
index 8001d0f219..fda4df2653 100644
--- a/sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist
+++ b/sysdeps/unix/sysv/linux/x86_64/x32/libm.abilist
@@ -1179,3 +1179,5 @@  GLIBC_2.35 fsqrt F
 GLIBC_2.35 fsqrtl F
 GLIBC_2.35 hypot F
 GLIBC_2.35 hypotf F
+GLIBC_2.38 fmod F
+GLIBC_2.38 fmodf F