Update.

	* sysdeps/i386/fpu/bits/mathinline.h (atan): Add st(1) to clobbers.
	(log): Likewise.
	(log10): Likewise.
	(log1p): Likewise.
	(log2): Likewise.

diff --git a/ChangeLog b/ChangeLog
index 655c13b..ec11622 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -3,6 +3,12 @@
        * nis/nss_compat/compat-pwd.c (internal_getpwuid_r): Set errno in
        one more case.
 
+       * sysdeps/i386/fpu/bits/mathinline.h (atan): Add st(1) to clobbers.
+       (log): Likewise.
+       (log10): Likewise.
+       (log1p): Likewise.
+       (log2): Likewise.
+
 1999-09-22  Andreas Jaeger  <aj@suse.de>
 
        * stdio-common/tstscanf.c (main): Add test case from PR libc/1313

diff --git a/sysdeps/i386/fpu/bits/mathinline.h b/sysdeps/i386/fpu/bits/mathinline.h
index a45db6e..bf077ba 100644 (file)
--- a/sysdeps/i386/fpu/bits/mathinline.h
+++ b/sysdeps/i386/fpu/bits/mathinline.h
@@ -361,7 +361,7 @@ __inline_mathcode (tan, __x, \
 #define __atan2_code \
   register long double __value;                                                      \
   __asm __volatile__                                                         \
-    ("fpatan\n\t"                                                            \
+    ("fpatan"                                                                \
      : "=t" (__value) : "0" (__x), "u" (__y) : "st(1)");                     \
   return __value
 __inline_mathcode2 (atan2, __y, __x, __atan2_code)
@@ -441,9 +441,9 @@ __inline_mathop (sin, "fsin")
 /* The argument range of this inline version is reduced.  */
 __inline_mathop (cos, "fcos")
 
-__inline_mathop (atan, "fld1; fpatan")
-__inline_mathop (log, "fldln2; fxch; fyl2x")
-__inline_mathop (log10, "fldlg2; fxch; fyl2x")
+__inline_mathop_decl (atan, "fld1; fpatan", "0" (__x) : "st(1)")
+__inline_mathop_decl (log, "fldln2; fxch; fyl2x", "0" (__x) : "st(1)")
+__inline_mathop_decl (log10, "fldlg2; fxch; fyl2x", "0" (__x) : "st(1)")
 
 __inline_mathcode (asin, __x, return __atan2l (__x, __sqrtl (1.0 - __x * __x)))
 __inline_mathcode (acos, __x, return __atan2l (__sqrtl (1.0 - __x * __x), __x))
@@ -520,7 +520,7 @@ __inline_mathcode (log1p, __x, \
       ("fldln2\n\t"                                                          \
        "fxch\n\t"                                                            \
        "fyl2xp1"                                                             \
-       : "=t" (__value) : "0" (__x));                                        \
+       : "=t" (__value) : "0" (__x) : "st(1)");                                      \
   return __value)
 
 
@@ -551,7 +551,7 @@ __inline_mathcode(logb, __x, \
 #endif
 
 #ifdef __USE_ISOC9X
-__inline_mathop(log2, "fld1; fxch; fyl2x")
+__inline_mathop_decl (log2, "fld1; fxch; fyl2x", "0" (__x) : "st(1)")
 
 __MATH_INLINE float ldexpf (float __x, int __y);
 __MATH_INLINE float