Update. cvs/libc-ud-970307 cvs/libc-970316 cvs/libc-970315 cvs/libc-970314 cvs/libc-970313 cvs/libc-970312 cvs/libc-970311 cvs/libc-970310 cvs/libc-970309 cvs/libc-970308

1997-03-08 05:30  Ulrich Drepper  <drepper@cygnus.com>

	* argp/Makefile (routines): Add argp-eexst.
	* argp/argp-ba.c (argp_program_bug_address): Make `const'.
	* argp/argp-eexst.c: New file.  Define exit status variable.
	* argp/argp-help.c (__argp_state_help): Use argp_err_exit_status
	variable instead of always exiting with value 1.
	* argp/argp-pv.c (argp_program_version): Make `const'.
	* argp/argp.h: Declare argp_err_exit_status.
	Patches by Miles Bader <miles@gnu.ai.mit.edu>.

	* locale/programs/localedef.c: Use argp_err_exit_status to
	terminate with correct value.

	* inet/rcmd.c (iruserok): Use euidaccess instead of half-hearted
	switching of UID before opening .rhosts.

	* libio/vsnprintf.c: Change implementation to follow ISO C 9X
	proposal.  The return value now is always the number of characters
	which would be written if enough space is available.
	* manual/stdio.texi: Update description for new behaviour.

	* locale/locale.h (__locale_t): Don't use __locale_t for struct
	name and pointer to struct since old gccs cannot keep the namespaces
	apart.  Rename struct to __locale_struct.
	* locale/duplocale.h: Likewise.
	* locale/newlocale.h: Likewise.

	* math/Makefile (headers): Add complex.h and cmathcalls.h.
	(routines): Remove w_cabs.  Add conj, cimag, creal, and cabs.
	* math/math.h: Undefine __MATHDECL_1, __MATHDECL, and __MATHCALL
	after use.
	(signgam): Move declaration to here from mathcalls.h.
	* math/mathcalls.h (cabs, __cabs_complex, signgam): Remove definition.
	Correct comment for fmin and fmax.
	Start implementing complex math function from ISO C 9X.
	* math/complex.h: New file.
	* math/cmathcalls.h: New file.
	* math/cabs.c: New file.
	* math/cabsf.c: New file.
	* math/cabsl.c: New file.
	* math/cimag.c: New file.
	* math/cimagf.c: New file.
	* math/cimagl.c: New file.
	* math/conj.c: New file.
	* math/conjf.c: New file.
	* math/conjl.c: New file.
	* math/creal.c: New file.
	* math/crealf.c: New file.
	* math/creall.c: New file.
	* sysdeps/libm-ieee754/w_cabs.c: Removed.
	* sysdeps/libm-ieee754/w_cabsf.c: Removed.
	* sysdeps/libm-ieee754/w_cabsl.c: Removed.

	* posix/regex.c: Merge with regex sources from Arnold Robbins'
	version in GNU awk.
	* posix/regex.h: Likewise.

	Add regex test suite by Tom Lord.
	* posix/Makefile (distribute): Add TESTS, TESTS2C.sed, and testcases.h.
	(tests): Add runtests.
	(before-compile): Add testcases.h.
	(testcases.h): New rule to generate header with tests.
	* posix/TESTS: New file.
	* posix/TESTS2C.sed.: New file.
	* posix/runtests.c: New file.
	* posix/testcases.h: New file.

	* sysdeps/unix/sysv/linux/poll.c: Test whether poll syscall is
	available and use fall-back implementation if not.

	* sysdeps/unix/sysv/linux/syscalls.list: Sort entries.

	* time/ialloc.c: Update from tzcode1997c.
	* time/private.h: Likewise.
	* time/scheck.c: Likewise.
	* time/tzselect.ksh: Likewise.
	* time/zdump.c: Likewise.
	* time/zic.c: Likewise.

	* time/tzfile.c: Pretty print.

1997-03-06 07:37  Geoff Keating  <geoffk@ozemail.com.au>

	Port to powerpc-*-linux-gnu.  Slightly tested, under MkLinux,
	on a 601.

	* sysdeps/powerpc/Implies: Added.
	* sysdeps/powerpc/__longjmp.S: Added.
	* sysdeps/powerpc/__math.h: Added.
	* sysdeps/powerpc/bsd-_setjmp.S: Added.
	* sysdeps/powerpc/bsd-setjmp.S: Added.
	* sysdeps/powerpc/dl-machine.h: Added.
	* sysdeps/powerpc/ffs.c: Added.
	* sysdeps/powerpc/fpu_control.h: Added.
	* sysdeps/powerpc/jmp_buf.h: Added.
	* sysdeps/powerpc/setjmp.S: Added.
	* sysdeps/powerpc/strlen.s: Added.
	* sysdeps/powerpc/elf/start.c: Added.
	* sysdeps/powerpc/fpu_control.h: Added.
	* sysdeps/powerpc/jmp_buf.h: Added.

	* sysdeps/unix/sysv/linux/powerpc/Dist: Added.
	* sysdeps/unix/sysv/linux/powerpc/_exit.S: Added.
	* sysdeps/unix/sysv/linux/powerpc/brk.c: Added.
	* sysdeps/unix/sysv/linux/powerpc/clone.S: Added.
	* sysdeps/unix/sysv/linux/powerpc/init-first.h: Added.
	* sysdeps/unix/sysv/linux/powerpc/ioctl-types.h: Added.
	* sysdeps/unix/sysv/linux/powerpc/profil.c: Added.
	* sysdeps/unix/sysv/linux/powerpc/sigreturn.S: Added.
	* sysdeps/unix/sysv/linux/powerpc/socket.S: Added.
	* sysdeps/unix/sysv/linux/powerpc/syscall.S: Added.
	* sysdeps/unix/sysv/linux/powerpc/syscalls.list: Added.
	* sysdeps/unix/sysv/linux/powerpc/sysdep.c: Added.
	* sysdeps/unix/sysv/linux/powerpc/sysdep.h: Added.
	* sysdeps/unix/sysv/linux/powerpc/termbits.h: Added.
	* sysdeps/unix/sysv/linux/powerpc/sys/syscall.h: Added.

1997-03-05 05:24  Geoff Keating  <geoffk@ozemail.com.au>

	* elf/dl-runtime.c (fixup): Add ELF_FIXUP_RETURNS_ADDRESS switch,
	because knowing the first instruction of a PowerPC PLT trampoline
	is not very helpful.

1997-03-04 08:04  Geoff Keating  <geoffk@ozemail.com.au>

	* elf/dl-load.c (ELF_PREFERRED_ADDRESS, ELF_PREFERRED_ADDRESS_DATA,
	ELF_FIXED_ADDRESS): Added. These are for dl-machine.h to indicate
	a preference as to where executables should be loaded.

1997-02-28 08:50  Geoff Keating  <geoffk@ozemail.com.au>

	* elf/elf.h: Add in all those PowerPC reloc types.

1997-02-24 07:12  Geoff Keating  <geoffk@ozemail.com.au>

	* stdio-common/vfscanf.c: Use __va_copy if available.

1997-03-06 13:50  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* malloc/malloc.c (mprotect) [_LIBC]: Define as __mprotect to
	clean up namespace.

1997-03-07 14:27  Ulrich Drepper  <drepper@cygnus.com>

	* sysdeps/libm-ieee754/s_fpclassify.c (__fpclassify): Don't use
	non-existing GET_WORDS macro, use EXTRACT_WORDS.
	Reported by Andres Schwab <schwab@issan.informatik.uni-dortmund.de>.

1997-03-07 05:27  Ulrich Drepper  <drepper@cygnus.com>

	* sysdeps/unix/sysv/linux/i386/sigaction.c: The user- and
	kernel-level sigaction structure is different.  handle this
	correctly.
	Bug report by Andres Jaeger <jaeger@informatik.uni-kl.de>.

1997-03-06 05:55  David Engel  <david@sw.ods.com>

	* sysdeps/generic/dl-cache.c (_dl_load_cache_lookup): Also
	recognize cache entries with flag == 3.

1997-03-06 01:05  Ulrich Drepper  <drepper@cygnus.com>

	* sysdeps/generic/netinet/in.h (INADDR_NONE): Cast to unsigned int.
	(INADDR_LOOPBACK): Likewise.
	* sysdeps/unix/sysv/linux/netinet/in.h: Likewise.

	* manual/socket.texi: Correct types of INADDR_* constants from
	unsigned long to unsigned.

1997-03-05 23:14  Ulrich Drepper  <drepper@cygnus.com>

	* time/antarctica: Update from tzdata1997c.
	* time/asia: Likewise.
	* time/zone.tab: Likewise.

1997-03-05 00:43  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* configure.in (libc_cv_asm_symver_directive,
	libc_cv_ld_version_script_option): Remove unknown instruction from
	assembler test file.

1997-03-04 19:14  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* sysdeps/m68k/fpu/__math.h: Update feature tests to use
	__USE_ISOC9X.
	[__USE_ISOC9X]: Define __log2 and __exp2 inlines.
	(fabs): Remove defininition, it is a gcc builtin.
	(sqrt, __sqrt): Remove definition, they have wrappers in libm.

1997-03-04 10:11  H.J. Lu  <hjl@lucon.org>

	* sunrpc/Makefile ($(objpfx)rpc-proto.c): Create subdir before
	generating file.

1997-03-05 03:59  Ulrich Drepper  <drepper@cygnus.com>

	* inet/rcmd.c (__ivaliduser): Don't use getline, but __getline
	instead to avoid namespace problems.
	Reported by David Mosberger-Tang <davidm@AZStarNet.COM>.

1997-03-03 19:01  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* login/utmp_file.c: Fix previous change.  Make portable by
	checking the _HAVE_UT_* feature macros.

	* libio/fileops.c (_IO_file_finish): Likewise.

17 files changed, 625 insertions, 16 deletions

diff --git a/math/Makefile b/math/Makefile
index b8964baa88..c4ed312c2b 100644
--- a/math/Makefile
+++ b/math/Makefile
@@ -21,7 +21,8 @@
 subdir		:= math
 
 # Installed header files.
-headers		:= math.h mathcalls.h __math.h huge_val.h nan.h fpu_control.h
+headers		:= math.h mathcalls.h __math.h huge_val.h nan.h		\
+		   fpu_control.h complex.h cmathcalls.h
 
 # Internal header files.
 distribute	:= math_private.h machine/asm.h machine/endian.h
@@ -43,10 +44,11 @@ libm-calls = e_acos e_acosh e_asin e_atan2 e_atanh e_cosh e_exp e_fmod	\
 	     s_floor s_frexp s_ilogb s_ldexp s_log1p s_logb 		\
 	     s_modf s_nextafter s_rint s_scalbn s_significand		\
 	     s_sin s_tan s_tanh w_acos w_acosh w_asin			\
-	     w_atan2 w_atanh w_cabs w_cosh w_drem w_exp w_fmod w_gamma	\
+	     w_atan2 w_atanh w_cosh w_drem w_exp w_fmod w_gamma		\
 	     w_gamma_r w_hypot w_j0 w_j1 w_jn w_lgamma w_lgamma_r	\
 	     w_log w_log10 w_pow w_remainder w_scalb w_sinh w_sqrt	\
-	     s_signbit s_fpclassify s_fmax s_fmin s_fdim
+	     s_signbit s_fpclassify s_fmax s_fmin s_fdim		\
+	     conj cimag creal cabs
 libm-routines = $(libm-support) $(libm-calls) \
 		$(patsubst %_rf,%f_r,$(libm-calls:=f))	\
 		$(long-m-$(long-double-fcts))
diff --git a/math/cabs.c b/math/cabs.c
new file mode 100644
index 0000000000..cf775ded95
--- /dev/null
+++ b/math/cabs.c
@@ -0,0 +1,33 @@
+/* Return the complex absolute value of double complex value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+#include <math.h>
+
+double
+__cabs (double complex z)
+{
+  return __hypot (__real__ z, __imag__ z);
+}
+weak_alias (__cabs, cabs)
+#ifdef NO_LONG_DOUBLE
+strong_alias (__cabs, __cabsl)
+weak_alias (__cabs, cabsl)
+#endif
diff --git a/math/cabsf.c b/math/cabsf.c
new file mode 100644
index 0000000000..2ea2660d97
--- /dev/null
+++ b/math/cabsf.c
@@ -0,0 +1,29 @@
+/* Return the complex absolute value of float complex value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+#include <math.h>
+
+float
+__cabsf (float complex z)
+{
+  return __hypotf (__real__ z, __imag__ z);
+}
+weak_alias (__cabsf, cabsf)
diff --git a/math/cabsl.c b/math/cabsl.c
new file mode 100644
index 0000000000..1f3554cef2
--- /dev/null
+++ b/math/cabsl.c
@@ -0,0 +1,29 @@
+/* Return the complex absolute value of long double complex value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+#include <math.h>
+
+long double
+__cabsl (long double complex z)
+{
+  return __hypotl (__real__ z, __imag__ z);
+}
+weak_alias (__cabsl, cabsl)
diff --git a/math/cimag.c b/math/cimag.c
new file mode 100644
index 0000000000..13efa1476e
--- /dev/null
+++ b/math/cimag.c
@@ -0,0 +1,32 @@
+/* Return imaginary part of complex double value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+
+double
+__cimag (double complex z)
+{
+  return __imag__ z;
+}
+weak_alias (__cimag, cimag)
+#ifdef NO_LONG_DOUBLE
+strong_alias (__cimag, __cimagl)
+weak_alias (__cimag, cimagl)
+#endif
diff --git a/math/cimagf.c b/math/cimagf.c
new file mode 100644
index 0000000000..8edb4e613e
--- /dev/null
+++ b/math/cimagf.c
@@ -0,0 +1,28 @@
+/* Return imaginary part of complex float value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+
+float
+__cimagf (float complex z)
+{
+  return __imag__ z;
+}
+weak_alias (__cimagf, cimagf)
diff --git a/math/cimagl.c b/math/cimagl.c
new file mode 100644
index 0000000000..9c58899274
--- /dev/null
+++ b/math/cimagl.c
@@ -0,0 +1,28 @@
+/* Return imaginary part of complex long double value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+
+long double
+__cimagl (long double complex z)
+{
+  return __imag__ z;
+}
+weak_alias (__cimagl, cimagl)
diff --git a/math/cmathcalls.h b/math/cmathcalls.h
new file mode 100644
index 0000000000..658c3a729d
--- /dev/null
+++ b/math/cmathcalls.h
@@ -0,0 +1,152 @@
+/* Prototype declarations for complex math functions;
+   helper file for <complex.h>.
+   Copyright (C) 1997 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
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+/* NOTE: Because of the special way this file is used by <math.h>, this
+   file must NOT be protected from multiple inclusion as header files
+   usually are.
+
+   This file provides prototype declarations for the math functions.
+   Most functions are declared using the macro:
+
+   __MATHCALL (NAME, (ARGS...));
+
+   This means there is a function `NAME' returning `double' and a function
+   `NAMEf' returning `float'.  Each place `_Mdouble_' appears in the
+   prototype, that is actually `double' in the prototype for `NAME' and
+   `float' in the prototype for `NAMEf'.  Reentrant variant functions are
+   called `NAME_r' and `NAMEf_r'.
+
+   Functions returning other types like `int' are declared using the macro:
+
+   __MATHDECL (TYPE, NAME, (ARGS...));
+
+   This is just like __MATHCALL but for a function returning `TYPE'
+   instead of `_Mdouble_'.  In all of these cases, there is still
+   both a `NAME' and a `NAMEf' that takes `float' arguments.  */
+
+#ifndef _COMPLEX_H
+ #error "Never include cmathcalls.h directly; include <complex.h> instead."
+#endif
+
+#define _Mdouble_complex_ _Mdouble_ complex
+
+
+/* Trigonometric functions.  */
+
+/* Arc cosine of Z.  */
+__MATHCALL (cacos, (_Mdouble_complex_ __z));
+/* Arc sine of Z.  */
+__MATHCALL (casin, (_Mdouble_complex_ __z));
+/* Arc tangent of Z.  */
+__MATHCALL (catan, (_Mdouble_complex_ __z));
+
+/* Cosine of Z.  */
+__MATHCALL (ccos, (_Mdouble_complex_ __z));
+/* Sine of Z.  */
+__MATHCALL (csin, (_Mdouble_complex_ __z));
+/* Tangent of Z.  */
+__MATHCALL (ctan, (_Mdouble_complex_ __z));
+
+
+/* Hyperbolic functions.  */
+
+/* Hyperbolic arc cosine of Z.  */
+__MATHCALL (cacosh, (_Mdouble_complex_ __z));
+/* Hyperbolic arc sine of Z.  */
+__MATHCALL (casinh, (_Mdouble_complex_ __z));
+/* Hyperbolic arc tangent of Z.  */
+__MATHCALL (catanh, (_Mdouble_complex_ __z));
+
+/* Hyperbolic cosine of Z.  */
+__MATHCALL (ccosh, (_Mdouble_complex_ __z));
+/* Hyperbolic sine of Z.  */
+__MATHCALL (csinh, (_Mdouble_complex_ __z));
+/* Hyperbolic tangent of Z.  */
+__MATHCALL (ctanh, (_Mdouble_complex_ __z));
+
+
+/* Exponential and logarithmic functions.  */
+
+/* Exponential function of Z.  */
+__MATHCALL (cexp, (_Mdouble_complex_ __z));
+
+/* Natural logarithm of Z.  */
+__MATHCALL (clog, (_Mdouble_complex_ __z));
+
+
+/* Power functions.  */
+
+/* Return X to the Y power.  */
+__MATHCALL (cpow, (_Mdouble_complex_ __x, _Mdouble_complex_ __y));
+
+/* Return the square root of Z.  */
+__MATHCALL (csqrt, (_Mdouble_complex_ __z));
+
+
+/* Absolute value, projections, conjugates, and projection.  */
+
+/* Absolute value of Z.  */
+__MATHDECL (_Mdouble_, cabs, (_Mdouble_complex_ __z));
+
+/* Argument value of Z.  */
+__MATHDECL (_Mdouble_, carg, (_Mdouble_complex_ __z));
+
+/* Complex conjugate of Z.  */
+__MATHCALL (conj, (_Mdouble_complex_ __z));
+
+/* Projection of Z onto the Riemann sphere.  */
+__MATHCALL (cproj, (_Mdouble_complex_ __z));
+
+
+/* Decomposing complex values.  */
+
+/* Imaginary part of Z.  */
+__MATHDECL (_Mdouble_, cimag, (_Mdouble_complex_ __z));
+
+/* Real part of Z.  */
+__MATHDECL (_Mdouble_, creal, (_Mdouble_complex_ __z));
+
+
+/* Now some optimized versions.  GCC has handy notations for these
+   functions.  */
+#if defined __GNUC__ && defined __OPTIMIZE__
+
+/* Imaginary part of Z.  */
+extern __inline _Mdouble_
+__MATH_PRECNAME(cimag) (_Mdouble_complex_ __z)
+{
+  return __imag__ __z;
+}
+
+/* Real part of Z.  */
+extern __inline _Mdouble_
+__MATH_PRECNAME(creal) (_Mdouble_complex_ __z)
+{
+  return __real__ __z;
+}
+
+/* Complex conjugate of Z.  */
+extern __inline _Mdouble_complex_
+__MATH_PRECNAME(conj) (_Mdouble_complex_ __z)
+{
+  return ~__z;
+}
+
+#endif
diff --git a/math/complex.h b/math/complex.h
new file mode 100644
index 0000000000..6fc5c1975f
--- /dev/null
+++ b/math/complex.h
@@ -0,0 +1,102 @@
+/* Copyright (C) 1997 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
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+/*
+ *	ISO C 9X:  7.8 Complex arithmetic	<complex.h>
+ */
+
+#ifndef _COMPLEX_H
+#define _COMPLEX_H	1
+
+/* We might need to add support for more compilers here.  But once ISO
+   C 9X is out hopefully all maintained compilers will provide the data
+   types `float complex' and `double complex'.  */
+#if (__GNUC__ == 2 && __GNUC_MINOR__ >= 7) || __GNUC__ > 2
+# define complex __complex__
+#endif
+
+
+/* Narrowest imaginary unit.  This depends on the floating-point
+   evaluation method.
+   XXX This probably has to go into a gcc related file.  */
+#define _Imaginary_I	(DBL_EPSISON * 1.0i)
+
+/* Another more descriptive name is `I'.  */
+#undef I
+#define I _Imaginary_I
+
+
+/* Optimization aids.  This is not yet implemented in gcc and yonce it
+   is this will probably be available in a gcc header.  */
+#define CX_LIMITED_RANGE_ON
+#define CX_LIMITED_RANGE_OFF
+#define CX_LIMITED_RANGE_DEFAULT
+
+
+/* The file <cmathcalls.h> contains the prototypes for all the actual
+   math functions.  These macros are used for those prototypes, so
+   we can easily declare each function as both `name' and `__name',
+   and can declare the float versions `namef' and `__namef'.  */
+
+#define __MATHCALL(function, args)	\
+  __MATHDECL (_Mdouble_complex_,function, args)
+#define __MATHDECL(type, function, args) \
+  __MATHDECL_1(type, function, args); \
+  __MATHDECL_1(type, __CONCAT(__,function), args)
+#define __MATHDECL_1(type, function, args) \
+  extern type __MATH_PRECNAME(function) args
+
+#define _Mdouble_ 		double
+#define __MATH_PRECNAME(name)	name
+#include <cmathcalls.h>
+#undef	_Mdouble_
+#undef	__MATH_PRECNAME
+
+/* Now the float versions.  */
+#ifndef _Mfloat_
+#define _Mfloat_		float
+#endif
+#define _Mdouble_ 		_Mfloat_
+#ifdef __STDC__
+#define __MATH_PRECNAME(name)	name##f
+#else
+#define __MATH_PRECNAME(name)	name/**/f
+#endif
+#include <cmathcalls.h>
+#undef	_Mdouble_
+#undef	__MATH_PRECNAME
+
+/* And the long double versions.  It is non-critical to define them
+   here unconditionally since `long double' is required in ISO C 9X.  */
+#ifndef _Mlong_double_
+#define _Mlong_double_		long double
+#endif
+#define _Mdouble_ 		_Mlong_double_
+#ifdef __STDC__
+#define __MATH_PRECNAME(name)	name##l
+#else
+#define __MATH_PRECNAME(name)	name/**/l
+#endif
+#include <cmathcalls.h>
+#undef	_Mdouble_
+#undef	__MATH_PRECNAME
+#undef	__MATHDECL_1
+#undef	__MATHDECL
+#undef	__MATHCALL
+
+#endif /* complex.h */
diff --git a/math/conj.c b/math/conj.c
new file mode 100644
index 0000000000..c15c417e4c
--- /dev/null
+++ b/math/conj.c
@@ -0,0 +1,32 @@
+/* Return complex conjugate of complex double value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+
+double complex
+__conj (double complex z)
+{
+  return ~z;
+}
+weak_alias (__conj, conj)
+#ifdef NO_LONG_DOUBLE
+strong_alias (__conj, __conjl)
+weak_alias (__conj, conjl)
+#endif
diff --git a/math/conjf.c b/math/conjf.c
new file mode 100644
index 0000000000..828edaf2b7
--- /dev/null
+++ b/math/conjf.c
@@ -0,0 +1,28 @@
+/* Return complex conjugate of complex float value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+
+float complex
+__conjf (float complex z)
+{
+  return ~z;
+}
+weak_alias (__conjf, conjf)
diff --git a/math/conjl.c b/math/conjl.c
new file mode 100644
index 0000000000..9e44dc32a1
--- /dev/null
+++ b/math/conjl.c
@@ -0,0 +1,28 @@
+/* Return complex conjugate of complex long double value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+
+long double complex
+__conjl (long double complex z)
+{
+  return ~z;
+}
+weak_alias (__conjl, conjl)
diff --git a/math/creal.c b/math/creal.c
new file mode 100644
index 0000000000..c3ec4f7802
--- /dev/null
+++ b/math/creal.c
@@ -0,0 +1,32 @@
+/* Return real part of complex double value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+
+double
+__creal (double complex z)
+{
+  return __real__ z;
+}
+weak_alias (__creal, creal)
+#ifdef NO_LONG_DOUBLE
+strong_alias (__creal, __creall)
+weak_alias (__creal, creall)
+#endif
diff --git a/math/crealf.c b/math/crealf.c
new file mode 100644
index 0000000000..bb9c2c1418
--- /dev/null
+++ b/math/crealf.c
@@ -0,0 +1,28 @@
+/* Return real part of complex float value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+
+float
+__crealf (float complex z)
+{
+  return __real__ z;
+}
+weak_alias (__crealf, crealf)
diff --git a/math/creall.c b/math/creall.c
new file mode 100644
index 0000000000..0ab1aab2cb
--- /dev/null
+++ b/math/creall.c
@@ -0,0 +1,28 @@
+/* Return real part of complex long double value.
+   Copyright (C) 1997 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#include <complex.h>
+
+long double
+__creall (long double complex z)
+{
+  return __real__ z;
+}
+weak_alias (__creall, creall)
diff --git a/math/math.h b/math/math.h
index e08d210be8..22274f023e 100644
--- a/math/math.h
+++ b/math/math.h
@@ -96,6 +96,15 @@ __BEGIN_DECLS
 #endif /* __STDC__ || __GNUC__ */
 
 #endif	/* Use misc or ISO C 9X.  */
+#undef	__MATHDECL_1
+#undef	__MATHDECL
+#undef	__MATHCALL
+
+
+#if defined __USE_MISC || defined __USE_XOPEN || defined __USE_ISOC9X
+/* This variable is used by `gamma' and `lgamma'.  */
+extern int signgam;
+#endif
 
 
 /* ISO C 9X defines some generic macros which work on any data type.  */
diff --git a/math/mathcalls.h b/math/mathcalls.h
index fd8ed86e81..d529c1a8e2 100644
--- a/math/mathcalls.h
+++ b/math/mathcalls.h
@@ -178,14 +178,6 @@ __MATHCALL (scalbn,, (_Mdouble_ __x, int __n));
 /* Return the remainder of X/Y.  */
 __MATHCALL (drem,, (_Mdouble_ __x, _Mdouble_ __y));
 
-struct __MATH_PRECNAME(__cabs_complex,)
-{
-  _Mdouble_ x, y;
-};
-
-/* Return `sqrt(X*X + Y*Y)'.  */
-__MATHCALL (cabs,, (struct __MATH_PRECNAME(__cabs_complex,)));
-
 
 /* Return the fractional part of X after dividing out `ilogb (X)'.  */
 __MATHCALL (significand,, (_Mdouble_ __x));
@@ -226,9 +218,6 @@ __MATHCALL (erfc,, (_Mdouble_));
 __MATHCALL (gamma,, (_Mdouble_));
 __MATHCALL (lgamma,, (_Mdouble_));
 
-/* This variable is used by `gamma' and `lgamma'.  */
-extern int signgam;
-
 #ifdef __USE_MISC
 /* Reentrant versions of gamma and lgamma.  Those functions use the global
    variable `signgam'.  The reentrant versions instead take a pointer and
@@ -279,10 +268,10 @@ __MATHCALL (remquo,, (_Mdouble_ __x, _Mdouble_ __y, int *__quo));
 /* Return positive difference between X and Y.  */
 __MATHCALL (fdim,, (_Mdouble_ __x, _Mdouble_ __y));
 
-/* Return minimum numeric value from X and Y.  */
+/* Return maximum numeric value from X and Y.  */
 __MATHCALL (fmax,, (_Mdouble_ __x, _Mdouble_ __y));
 
-/* Return maximum numeric value from X and Y.  */
+/* Return minimum numeric value from X and Y.  */
 __MATHCALL (fmin,, (_Mdouble_ __x, _Mdouble_ __y));