This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] powerpc: Remove uses of operand modifier (%s) in inline asm


On Wed, 27 Jan 2016 17:19:12 +0000
Joseph Myers <joseph@codesourcery.com> wrote:

> On Wed, 27 Jan 2016, Gabriel F. T. Gomes wrote:
> 
> > Thus, we can replace __builtin_ffs with __builtin_clz and remove
> > the %s operand modifier.  
> 
> Doesn't that mean you need to make these definitions conditional on 
> __GNUC_PREREQ (3, 4) (which I think should be fine to do - we
> shouldn't need to care about these optimizations for pre-3.4
> compilers), as that was the version where __builtin_clz was
> introduced?
> 

I agree.

The attached patch addresses your concerns.

>From 978a2e3f70d60f68171c97b7c5e0cdcf0436d6b5 Mon Sep 17 00:00:00 2001
From: "Gabriel F. T. Gomes" <gftg@linux.vnet.ibm.com>
Date: Fri, 22 Jan 2016 18:05:05 -0200
Subject: [PATCH] powerpc: Remove uses of operand modifier (%s) in inline asm

The operand modifier %s on powerpc is an undocumented internal implementation
detail of GCC.  Besides that, the GCC community wants to remove it.  This patch
rewrites the expressions that use this modifier with logically equivalent
expressions that don't require it.

Explanation for the substitution:

The %s modifier takes an immediate operand and prints 32 less such immediate.
Thus, in the previous code, the expression resulted in:

  32 - __builtin_ffs(e)

where e was guaranteed to have exactly a single bit set, by the following
expressions:

  (e & (e-1) == 0) : e has at most one bit set.
  (e != 0)         : e is not zero, thus it has at least one bit set.

Since we guarantee that there is exactly only one bit set, the following
statement is true:

  32 - __builtin_ffs(e) == __builtin_clz(e)

Thus, we can replace __builtin_ffs with __builtin_clz and remove the %s operand
modifier.

2016-01-25  Gabriel F. T. Gomes  <gftg@linux.vnet.ibm.com>

	* sysdeps/powerpc/bits/fenvinline.h (feraiseexcept): Remove use of %s
	operand modifier.
	(feclearexcept): Likewise.
---
 sysdeps/powerpc/bits/fenvinline.h | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/sysdeps/powerpc/bits/fenvinline.h b/sysdeps/powerpc/bits/fenvinline.h
index 4a7b2af..c283ede 100644
--- a/sysdeps/powerpc/bits/fenvinline.h
+++ b/sysdeps/powerpc/bits/fenvinline.h
@@ -32,8 +32,10 @@
    warning when __excepts is not a constant.  Otherwise, they mean the
    same as just plain 'i'.  */
 
+#  if __GNUC_PREREQ(3, 4)
+
 /* Inline definition for feraiseexcept.  */
-#  define feraiseexcept(__excepts) \
+#   define feraiseexcept(__excepts) \
   (__extension__  ({ 							      \
     int __e = __excepts;						      \
     int __ret;								      \
@@ -42,8 +44,8 @@
         && __e != FE_INVALID)						      \
       {									      \
 	if (__e != 0)							      \
-	  __asm__ __volatile__ ("mtfsb1 %s0"				      \
-				: : "i#*X" (__builtin_ffs (__e)));	      \
+	  __asm__ __volatile__ ("mtfsb1 %0"				      \
+				: : "i#*X" (__builtin_clz (__e)));	      \
         __ret = 0;							      \
       }									      \
     else								      \
@@ -52,7 +54,7 @@
   }))
 
 /* Inline definition for feclearexcept.  */
-#  define feclearexcept(__excepts) \
+#   define feclearexcept(__excepts) \
   (__extension__  ({ 							      \
     int __e = __excepts;						      \
     int __ret;								      \
@@ -61,8 +63,8 @@
         && __e != FE_INVALID)						      \
       {									      \
 	if (__e != 0)							      \
-	  __asm__ __volatile__ ("mtfsb0 %s0"				      \
-				: : "i#*X" (__builtin_ffs (__e)));	      \
+	  __asm__ __volatile__ ("mtfsb0 %0"				      \
+				: : "i#*X" (__builtin_clz (__e)));	      \
         __ret = 0;							      \
       }									      \
     else								      \
@@ -70,6 +72,8 @@
     __ret;								      \
   }))
 
+#  endif /* __GNUC_PREREQ(3, 4).  */
+
 # endif /* !__NO_MATH_INLINES.  */
 
 #endif /* __GNUC__ && !_SOFT_FLOAT && !__NO_FPRS__ */
-- 
2.4.3


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]