glibc/SOURCES/glibc-rh1783303-4.patch
2021-09-16 08:47:37 +00:00

201 lines
7.0 KiB
Diff

commit 3db85a9814784a74536a1f0e7b7ddbfef7dc84bb
Author: Paul A. Clarke <pc@us.ibm.com>
Date: Thu Jun 20 11:57:18 2019 -0500
powerpc: Use faster means to access FPSCR when possible in some cases
Using 'mffs' instruction to read the Floating Point Status Control Register
(FPSCR) can force a processor flush in some cases, with undesirable
performance impact. If the values of the bits in the FPSCR which force the
flush are not needed, an instruction that is new to POWER9 (ISA version 3.0),
'mffsl' can be used instead.
Cases included: get_rounding_mode, fegetround, fegetmode, fegetexcept.
* sysdeps/powerpc/bits/fenvinline.h (__fegetround): Use
__fegetround_ISA300() or __fegetround_ISA2() as appropriate.
(__fegetround_ISA300) New.
(__fegetround_ISA2) New.
* sysdeps/powerpc/fpu_control.h (IS_ISA300): New.
(_FPU_MFFS): Move implementation...
(_FPU_GETCW): Here.
(_FPU_MFFSL): Move implementation....
(_FPU_GET_RC_ISA300): Here. New.
(_FPU_GET_RC): Use _FPU_GET_RC_ISA300() or _FPU_GETCW() as appropriate.
* sysdeps/powerpc/fpu/fenv_libc.h (fegetenv_status_ISA300): New.
(fegetenv_status): New.
* sysdeps/powerpc/fpu/fegetmode.c (fegetmode): Use fegetenv_status()
instead of fegetenv_register().
* sysdeps/powerpc/fpu/fegetexcept.c (__fegetexcept): Likewise.
Reviewed-by: Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
diff --git a/sysdeps/powerpc/bits/fenvinline.h b/sysdeps/powerpc/bits/fenvinline.h
index 41316386ba75e903..caec8ead6e17219d 100644
--- a/sysdeps/powerpc/bits/fenvinline.h
+++ b/sysdeps/powerpc/bits/fenvinline.h
@@ -18,13 +18,36 @@
#if defined __GNUC__ && !defined _SOFT_FLOAT && !defined __NO_FPRS__
-/* Inline definition for fegetround. */
-# define __fegetround() \
- (__extension__ ({ int __fegetround_result; \
- __asm__ __volatile__ \
- ("mcrfs 7,7 ; mfcr %0" \
- : "=r"(__fegetround_result) : : "cr7"); \
- __fegetround_result & 3; }))
+/* Inline definitions for fegetround. */
+# define __fegetround_ISA300() \
+ (__extension__ ({ \
+ union { double __d; unsigned long long __ll; } __u; \
+ __asm__ __volatile__ ( \
+ ".machine push; .machine \"power9\"; mffsl %0; .machine pop" \
+ : "=f" (__u.__d)); \
+ __u.__ll & 0x0000000000000003LL; \
+ }))
+
+# define __fegetround_ISA2() \
+ (__extension__ ({ \
+ int __fegetround_result; \
+ __asm__ __volatile__ ("mcrfs 7,7 ; mfcr %0" \
+ : "=r"(__fegetround_result) : : "cr7"); \
+ __fegetround_result & 3; \
+ }))
+
+# ifdef _ARCH_PWR9
+# define __fegetround() __fegetround_ISA300()
+# elif defined __BUILTIN_CPU_SUPPORTS__
+# define __fegetround() \
+ (__glibc_likely (__builtin_cpu_supports ("arch_3_00")) \
+ ? __fegetround_ISA300() \
+ : __fegetround_ISA2() \
+ )
+# else
+# define __fegetround() __fegetround_ISA2()
+# endif
+
# define fegetround() __fegetround ()
# ifndef __NO_MATH_INLINES
diff --git a/sysdeps/powerpc/fpu/fegetexcept.c b/sysdeps/powerpc/fpu/fegetexcept.c
index a053a32bfe11c0d4..9d77adea59939ece 100644
--- a/sysdeps/powerpc/fpu/fegetexcept.c
+++ b/sysdeps/powerpc/fpu/fegetexcept.c
@@ -25,7 +25,7 @@ __fegetexcept (void)
fenv_union_t fe;
int result = 0;
- fe.fenv = fegetenv_register ();
+ fe.fenv = fegetenv_status ();
if (fe.l & (1 << (31 - FPSCR_XE)))
result |= FE_INEXACT;
diff --git a/sysdeps/powerpc/fpu/fegetmode.c b/sysdeps/powerpc/fpu/fegetmode.c
index b83dc9f625d2248a..75493e5f24c8b05b 100644
--- a/sysdeps/powerpc/fpu/fegetmode.c
+++ b/sysdeps/powerpc/fpu/fegetmode.c
@@ -21,6 +21,6 @@
int
fegetmode (femode_t *modep)
{
- *modep = fegetenv_register ();
+ *modep = fegetenv_status ();
return 0;
}
diff --git a/sysdeps/powerpc/fpu/fenv_libc.h b/sysdeps/powerpc/fpu/fenv_libc.h
index d6945903b525748e..cc00df033da47c1a 100644
--- a/sysdeps/powerpc/fpu/fenv_libc.h
+++ b/sysdeps/powerpc/fpu/fenv_libc.h
@@ -35,6 +35,27 @@ extern const fenv_t *__fe_mask_env (void) attribute_hidden;
#define fegetenv_register() \
({ fenv_t env; asm volatile ("mffs %0" : "=f" (env)); env; })
+/* Equivalent to fegetenv_register, but only returns bits for
+ status, exception enables, and mode. */
+
+#define fegetenv_status_ISA300() \
+ ({register double __fr; \
+ __asm__ __volatile__ ( \
+ ".machine push; .machine \"power9\"; mffsl %0; .machine pop" \
+ : "=f" (__fr)); \
+ __fr; \
+ })
+
+#ifdef _ARCH_PWR9
+# define fegetenv_status() fegetenv_status_ISA300()
+#else
+# define fegetenv_status() \
+ (__glibc_likely (__builtin_cpu_supports ("arch_3_00")) \
+ ? fegetenv_status_ISA300() \
+ : fegetenv_register() \
+ )
+#endif
+
/* Equivalent to fesetenv, but takes a fenv_t instead of a pointer. */
#define fesetenv_register(env) \
do { \
diff --git a/sysdeps/powerpc/fpu_control.h b/sysdeps/powerpc/fpu_control.h
index 90063d77bbbf794f..e0ee622e246c0d61 100644
--- a/sysdeps/powerpc/fpu_control.h
+++ b/sysdeps/powerpc/fpu_control.h
@@ -96,35 +96,37 @@ extern fpu_control_t __fpu_control;
typedef unsigned int fpu_control_t;
/* Macros for accessing the hardware control word. */
-# define __FPU_MFFS() \
- ({register double __fr; \
- __asm__ __volatile__("mffs %0" : "=f" (__fr)); \
- __fr; \
- })
-
# define _FPU_GETCW(cw) \
({union { double __d; unsigned long long __ll; } __u; \
- __u.__d = __FPU_MFFS(); \
+ __asm__ __volatile__("mffs %0" : "=f" (__u.__d)); \
(cw) = (fpu_control_t) __u.__ll; \
(fpu_control_t) __u.__ll; \
})
-#ifdef _ARCH_PWR9
-# define __FPU_MFFSL() \
- ({register double __fr; \
- __asm__ __volatile__("mffsl %0" : "=f" (__fr)); \
- __fr; \
+# define _FPU_GET_RC_ISA300() \
+ ({union { double __d; unsigned long long __ll; } __u; \
+ __asm__ __volatile__( \
+ ".machine push; .machine \"power9\"; mffsl %0; .machine pop" \
+ : "=f" (__u.__d)); \
+ (fpu_control_t) (__u.__ll & _FPU_MASK_RC); \
})
-#else
-# define __FPU_MFFSL() __FPU_MFFS()
-#endif
-
-# define _FPU_GET_RC() \
- ({union { double __d; unsigned long long __ll; } __u; \
- __u.__d = __FPU_MFFSL(); \
- __u.__ll &= _FPU_MASK_RC; \
- (fpu_control_t) __u.__ll; \
+
+# ifdef _ARCH_PWR9
+# define _FPU_GET_RC() _FPU_GET_RC_ISA300()
+# elif defined __BUILTIN_CPU_SUPPORTS__
+# define _FPU_GET_RC() \
+ ({fpu_control_t __rc; \
+ __rc = __glibc_likely (__builtin_cpu_supports ("arch_3_00")) \
+ ? _FPU_GET_RC_ISA300 () \
+ : _FPU_GETCW (__rc) & _FPU_MASK_RC; \
+ __rc; \
+ })
+# else
+# define _FPU_GET_RC() \
+ ({fpu_control_t __rc = _FPU_GETCW (__rc) & _FPU_MASK_RC; \
+ __rc; \
})
+# endif
# define _FPU_SETCW(cw) \
{ union { double __d; unsigned long long __ll; } __u; \