Mon Mar 4 20:54:40 1996 Andreas Schwab <schwab@issan.informatik.uni-d…

…ortmund.de>

	* Makeconfig ($(common-objpfx)config.make): Depend on config.h.in.


Mon Mar  4 17:35:09 1996  Roland McGrath  <roland@charlie-brown.gnu.ai.mit.edu>

	* hurd/catch-signal.c (hurd_safe_memmove): New function.
	(hurd_safe_copyin, hurd_safe_copyout): New functions.
	* hurd/hurd/sigpreempt.h: Declare them.

Sun Mar  3 08:43:44 1996  Roland McGrath  <roland@charlie-brown.gnu.ai.mit.edu>

	Replace math code with fdlibm from Sun as modified for netbsd by
	JT Conklin and Ian Taylor, including x86 FPU support.
	* sysdeps/libm-ieee754, sysdeps/libm-i387: New directories.
	* math/math_private.h: New file.
	* sysdeps/i386/fpu/Implies: New file.
	* sysdeps/ieee754/Implies: New file.
	* math/machine/asm.h, math/machine/endian.h: New files.
	* math/Makefile, math/math.h: Rewritten.
	* mathcalls.h, math/mathcalls.h: New file, broken out of math.h.
	* math/finite.c: File removed.
	* sysdeps/generic/Makefile [$(subdir)=math]: Frobnication removed.

	* math/test-math.c: Include errno.h and string.h.

	* sysdeps/unix/bsd/dirstream.h: File removed.
	* sysdeps/unix/bsd/readdir.c: File removed.

ChangeLog

@@ -1,3 +1,7 @@
+Mon Mar  4 20:54:40 1996  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>
+
+	* Makeconfig ($(common-objpfx)config.make): Depend on config.h.in.
+
 Tue Mar  5 12:14:57 1996  Miles Bader  <miles@gnu.ai.mit.edu>
 
 	* sysdeps/mach/hurd/select.c (__select): Also don't fault just
@@ -9,7 +13,32 @@ Mon Mar  4 17:35:35 1996  Miles Bader  <miles@gnu.ai.mit.edu>
 	Don't increment GOT only because READ/WRITE/EXCEPTFDS is 0!
 	Don't return without frobbing the bitmasks after a timeout.
 	When clearing the bitmasks, only loop from FIRSTFD to LASTFD.
-
+
+Mon Mar  4 17:35:09 1996  Roland McGrath  <roland@charlie-brown.gnu.ai.mit.edu>
+
+	* hurd/catch-signal.c (hurd_safe_memmove): New function.
+	(hurd_safe_copyin, hurd_safe_copyout): New functions.
+	* hurd/hurd/sigpreempt.h: Declare them.
+
+Sun Mar  3 08:43:44 1996  Roland McGrath  <roland@charlie-brown.gnu.ai.mit.edu>
+
+	Replace math code with fdlibm from Sun as modified for netbsd by
+	JT Conklin and Ian Taylor, including x86 FPU support.
+	* sysdeps/libm-ieee754, sysdeps/libm-i387: New directories.
+	* math/math_private.h: New file.
+	* sysdeps/i386/fpu/Implies: New file.
+	* sysdeps/ieee754/Implies: New file.
+	* math/machine/asm.h, math/machine/endian.h: New files.
+	* math/Makefile, math/math.h: Rewritten.
+	* mathcalls.h, math/mathcalls.h: New file, broken out of math.h.
+	* math/finite.c: File removed.
+	* sysdeps/generic/Makefile [$(subdir)=math]: Frobnication removed.
+
+	* math/test-math.c: Include errno.h and string.h.
+
+	* sysdeps/unix/bsd/dirstream.h: File removed.
+	* sysdeps/unix/bsd/readdir.c: File removed.
+
 Sat Mar  2 16:35:40 1996  Roland McGrath  <roland@charlie-brown.gnu.ai.mit.edu>
 
 	* sysdeps/unix/sysv/linux/m68k/profil-counter.h: File removed.

NEWS

@@ -37,6 +37,11 @@ Version 1.10
   through the interface in <sys/gmon.h>.  The profiling code was adapted
   from 4.4 BSD-Lite.
 
+* The math code has been replaced with a math library based on fdlibm from
+  Sun, and modified by JT Conklin with i387 support and by Ian Taylor with
+  `float' functions.  The math functions now reside in a separate library,
+  so programs using them will need to use `-lm' their linking commands.
+
 * The new functions `strtoq' and `strtouq' parse integer values from
   strings, like `strtol' and `strtoul', but they return `long long int' and
   `unsigned long long int' values, respectively (64-bit quantities).
@@ -53,6 +58,10 @@ Version 1.10
 * The new header <langinfo.h> defines an interface for accessing
   various locale-dependent data (using the locale chosen with `setlocale').
 
+* Ulrich Drepper has contributed a new suite of functions for operation on
+  wide-character and multibyte-character strings; see <wcstr.h> and <mbstr.h>.
+  These new functions are intended to conform to the ISO C specification.
+
 * You can now use positional parameter specifications in format strings
   for the `printf' and `scanf' families of functions.  For example,
   `printf ("Number %2$d, Mr %1$s\n", "Jones", 6);'' prints
@@ -163,10 +172,13 @@ Version 1.10
 
 * New function `getsid' returns session ID number on systems that support it.
 
-* The 4.4 BSD `db' library has been incorporated into the GNU C library.
-  New header files <db.h> and <mpool.h> provide a rich set of functions for
-  several types of simple databases stored in memory and in files, and
-  <ndbm.h> is an old `ndbm'-compatbile interface using the `db' functions.
+* We have incorporated the 4.4 BSD `db' library (version 1.85).  New header
+  files <db.h> and <mpool.h> provide a rich set of functions for several
+  types of simple databases stored in memory and in files, and <ndbm.h> is
+  an old `ndbm'-compatbile interface using the `db' functions.
+
+* New macro `strdupa' copies a string like `strdup', but uses local stack
+  space from `alloca' instead of dynamic heap space from `malloc'.
 
 Version 1.09
 

math/Makefile

@@ -1,4 +1,6 @@
-# Copyright (C) 1991, 1992, 1994, 1995 Free Software Foundation, Inc.
+# Makefile for the math library.
+
+# Copyright (C) 1996 Free Software Foundation, Inc.
 # This file is part of the GNU C Library.
 
 # The GNU C Library is free software; you can redistribute it and/or
@@ -16,24 +18,58 @@
 # not, write to the Free Software Foundation, Inc., 675 Mass Ave,
 # Cambridge, MA 02139, USA.
 
-#
-#	Makefile for math.
-#
-subdir	:= math
+subdir		:= math
+
+# Installed header files.
+headers		:= math.h mathcalls.h __math.h huge_val.h nan.h
+
+# Internal header files.
+distribute	:= math_private.h machine/asm.h machine/endian.h
 
-headers	:= math.h __math.h huge_val.h nan.h
 
+# Build the -lm library.
 
-routines := acos asin atan cos sin tan cosh sinh tanh exp fabs ldexp	\
-	    log log10 floor sqrt fmod frexp pow atan2 ceil modf		\
-	    isinf isnan finite infnan copysign drem logb		\
-	    rint hypot cabs cbrt expm1 log1p acosh asinh atanh		\
-	    isinfl isnanl
-tests := # test-math
-install-lib := libm.a
+extra-libs	:= libm
+libm.so-version	:= .0
+libm-routines	:= e_acos e_acosf e_acosh e_acoshf e_asin e_asinf	\
+		   e_atan2 e_atan2f e_atanh e_atanhf e_cosh e_coshf	\
+		   e_exp e_expf e_fmod e_fmodf e_hypot e_hypotf e_j0	\
+		   e_j0f e_j1 e_j1f e_jn e_jnf e_lgamma_r e_lgammaf_r	\
+		   e_log e_log10 e_log10f e_logf e_pow e_powf		\
+		   e_rem_pio2 e_rem_pio2f e_remainder e_remainderf	\
+		   e_scalb e_scalbf e_sinh e_sinhf e_sqrt e_sqrtf k_cos	\
+		   k_cosf k_rem_pio2 k_rem_pio2f k_sin k_sinf		\
+		   k_standard k_tan k_tanf s_asinh s_asinhf s_atan	\
+		   s_atanf s_cbrt s_cbrtf s_ceil s_ceilf s_copysign	\
+		   s_copysignf s_cos s_cosf s_erf s_erff s_expm1	\
+		   s_expm1f s_fabs s_fabsf s_finite s_finitef s_floor	\
+		   s_floorf s_frexp s_frexpf s_ilogb s_ilogbf		\
+		   s_ldexp s_ldexpf s_lib_version s_log1p s_log1pf	\
+		   s_logb s_logbf s_matherr s_modf s_modff s_nextafter	\
+		   s_nextafterf s_rint s_rintf s_scalbn s_scalbnf	\
+		   s_signgam s_significand s_significandf s_sin s_sinf	\
+		   s_tan s_tanf s_tanh s_tanhf w_acos w_acosf w_acosh	\
+		   w_acoshf w_asin w_asinf w_atan2 w_atan2f w_atanh	\
+		   w_atanhf w_cabs w_cabsf w_cosh w_coshf w_drem	\
+		   w_dremf w_exp w_expf w_fmod w_fmodf w_gamma		\
+		   w_gamma_r w_gammaf w_gammaf_r w_hypot w_hypotf w_j0	\
+		   w_j0f w_j1 w_j1f w_jn w_jnf w_lgamma w_lgamma_r	\
+		   w_lgammaf w_lgammaf_r w_log w_log10 w_log10f w_logf	\
+		   w_pow w_powf w_remainder w_remainderf w_scalb	\
+		   w_scalbf w_sinh w_sinhf w_sqrt w_sqrtf
+
+# These functions are in libc instead of libm because __printf_fp
+# calls them, so any program using printf will need them linked in,
+# and we don't want to have to link every program with -lm.
+routines := s_isinf s_isinff s_isnan s_isnanf isinfl isnanl
 
 
 include ../Rules
 
-$(objpfx)libm.a: $(dep-dummy-lib); $(make-dummy-lib)
-lib: $(objpfx)libm.a
+
+# This file defines the default _LIB_VERSION variable that controls
+# the error return conventions for the math functions.
+CPPFLAGS-s_lib_version.c := -D_POSIX_MODE
+
+# The fdlibm code generates a lot of these warnings but is otherwise clean.
+override CFLAGS += -Wno-uninitialized -Wno-write-strings

math/machine/asm.h

@@ -0,0 +1,10 @@
+/* The libm assembly code wants to include <machine/asm.h> to define the
+   ENTRY macro.  We define assembly-related macros in sysdep.h and
+   asm-syntax.h.  */
+
+#include <sysdep.h>
+#include <asm-syntax.h>
+
+/* The libm assembly code uses this macro for RCSid strings.
+   We don't put RCSid strings into object files.  */
+#define RCSID(id) /* ignore them */

math/machine/endian.h

@@ -0,0 +1,4 @@
+/* math_private.h wants to include <machine/endian.h>; we provide this
+   file so it can, leaving math_private.h unmodified from the original.  */
+
+#include <endian.h>