update from main archive 970225cvs/libc-970225

1997-02-24 23:05  Wolfram Gloger  <wmglo@dent.med.uni-muenchen.de>

	* malloc/malloc.c (malloc_get_state): New function.
	Saves global malloc state to an opaque data structure which
	is dynamically allocated in the heap.
	* malloc/malloc.c (malloc_set_state): New function.
	Restore previously obtained state.
	* malloc/malloc.h: Add declaration of malloc_get_state()
	and malloc_set_state().

1997-02-24 23:27  Ulrich Drepper  <drepper@cygnus.com>

	* sysdeps/libm-ieee754/s_cbrtl.c: Shift B1_EXP value to right
	position.

1997-02-24 17:38  Ulrich Drepper  <drepper@cygnus.com>

	* misc/error.c: Make error and error_at_line weak aliases of
	__error and __error_at_line respectively.
	Suggested by David Mosberger-Tang <davidm@AZStarNet.COM>.

	* sysdeps/unix/sysv/linux/i386/socket.S: Update copyright.

1997-02-22 11:30  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* elf/ldd.bash.in: Run the program directly, not as argument
	to the dynamic linker, if it contains an interpreter segment.
	* elf/ldd.sh.in: Likewise.

	* elf/rtld.c (dl_main): In verify mode check whether the dynamic
	object contains an interpreter segment and exit with 2 if not.

1997-02-23 01:23  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* Makefile (distribute): Remove nsswitch.h, netgroup.h, mcheck.h
	and xlocale.h.  Make-dist adds them automagically.

1997-02-22 12:25  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* locale/C-time.c (_nl_C_LC_TIME): Add missing entry for
	time-era-num-entries.

1997-02-06 13:49  Andreas Schwab  <schwab@issan.informatik.uni-dortmund.de>

	* rellns-sh: No need to check for existance of first parameter.

1997-02-24 15:20  Jonathan T. Agnew  <jtagnew@amherst.edu>

	* glibcbug.in: Don't mention destination on MAIL_AGENT command line
	to avoid duplicate mail.

1997-02-24 03:51  Ulrich Drepper  <drepper@cygnus.com>

	* Makefile (distribute): Add isomac.c.
	(tests): Run isomac test.

	* features.h (__USE_ISOC9X): New macro.

	* catgets/catgets.c: Don't use global variable `optind'.  Instead
	use result computed by argp_parse.
	* db/makedb: Likewise.
	* locale/programs/locale.c: Likewise.
	* locale/programs/localedef.c: Likewise.

	* libio/stdio.h: Rewrite.  Make it more readable and add comments.

	* libio/clearerr.c: Remove clearerr_locked alias.
	* libio/feof.c: Remove feof_locked alias.
	* libio/ferror.c: Remove feof_locked alias.
	* libio/fileno.c: Remove fileno_locked alias.
	* libio/fputc.c: Remove fputc_locked alias.
	* libio/getc.c: Remove getc_locked alias.
	* libio/getchar.c: Remove getchar_locked alias.
	* libio/iofflush.c: Remove fflush_locked alias.
	* libio/putc.c: Remove putc_locked alias.
	* libio/putc.c: Remove putchar_locked alias.

	* stdio-common/printf_fp.c: When number is inifinity print INF
	or inf depending on case of specifier.  Same for NaN where NAN
	or nan is printed.  Specified in ISO C 9X.

	* misc/sys/cdefs.h (__restrict): Define to empty string for now.
	* stdio/stdio.h: Add __restrict to prototypes where necessary.
	* libio/stdio.h: Likewise.
	* stdlib/stdlib.h: Likewise.
	* string/string.h: Likewise.
	* time/time.h: Likewise.
	* wcsmbs/wchar.h: Likewise.

	* stdlib/strtod.c: Change to recognize INF, INFINITY, NAN, and
	NAN(...).

	* sysdeps/ieee754/huge_val.h: Define HUGE_VALF and HUGE_VALL instead
	of HUGE_VALf and HUGE_VALL.
	* stdlib/strtof.c (FLOAT_HUGE_VAL): Use standard name HUGE_VALF
	instead of HUGE_VALf.
	* wcsmbs/wcstof.c: Likewise.
	* stdlib/strtold.c (FLOAT_HUGE_VAL): Use standard name HUGE_VALL
	instead of HUGE_VALl.
	* wcsmbs/wcstold.c: Likewise.

	* sysdeps/posix/gai_strerror.c: Use size_t for counter variable to
	avoid warning.

	* wcsmbs/Makefile (routines): Add wcscasecmp and wcsncase.
	* wcsmbs/wchar.h: Add prototypes for wcscasecmp and wcsncase.
	* wcsmbs/wcscasecmp.c: New file.
	* wcsmbs/wcsncase.c: New file.

	* stdlib/strtol.c: Define wide character quad word functions as
	wcstoll and wcstoull and normal versions as strtoll and strtoull.
	* wcsmbs/wchar.h: Add prototypes for wcstoll and wcstoull.
	* wcsmbs/wcstoq: Renamed to wcstoll.c.
	* wcsmbs/wcstouq: Renamed to wcstoull.c.
	* wcsmbs/wcstoll.c: Renamed from wcstoq.c.  Make wcstoq a weak
	alias of wcstoll.
	* wcsmbs/wcstoull.c: Renamed from wcstouq.c.  Make wcstouq a weak
	alias of wcstoull.
	* wcsmbs/Makefile (routines): Replace wcstoq and wcstouq by
	wcstoll and wcstoull respectively.
	* stdlib/strtoq.c: Rename to strtoll.c.
	* stdlib/strtouq.c: Rename to strtoull.c.
	* stdlib/strtoll.c: Renamed from strtoq.c.  Make strtoq a weak
	alias of strtoll.
	* stdlib/strtoll.c: Renamed from strtouq.c.  Make strtouq a weak
	alias of strtoull.
	* stdlib/Makefile (routines): Replace strtoq and strtouq by
	strtoll and strtoull respectively.
	* stdio-common/vfscanf.c: Don't use __strtoq_internal and
	__strtouq_internal but instead __strtoll_internal and
	__strtoull_internal respectively.
	* stdlib/stdlib.h (strtoq): Use __internal_strtoll in inline version.
	(strtouq): Similar with __internal_strtoull.
	* wcsmbs/wchar.h (wcstoq): Use __internal_wcstoll in inline version.
	(wcstouq): Similar with __internal_wcstoull.

1997-02-23 04:38  Ulrich Drepper  <drepper@cygnus.com>

	* stdlib/strtol.c (STRTOL): It is not illegal to parse a minus
	sign in the strtouXX functions.  The results gets simply negated.
	* stdio-common/tstscanf.c: Add testcase for above case.
	* stdlib/tst-strtol.c: Correct tests.

	* manual/stdio-fp.c: New file.  Generate output for example program
	in stdio.texi.

	* stdio-common/Makefile (routines): Add printf_fphex.
	* stdio-common/vfprintf.c: Add handling of %a and %A specifier.
	* stdio-common/printf_fphex.c: New file.  Implement %a and %A
	specifier.

1997-02-22 03:01  Ulrich Drepper  <drepper@cygnus.com>

	* sysdeps/unix/sysv/linux/timebits.h (CLK_TCK): Don't defined if
	__STRICT_ANSI__.

	* math/math.h: Prevent definition of struct exception when using
	C++.

1997-02-22 01:45  Ulrich Drepper  <drepper@cygnus.com>

	* sysdeps/unix/syscalls.list: Dup takes only one argument.
	Reported by Greg McGary.

1997-02-21 00:22  Miles Bader  <miles@gnu.ai.mit.edu>
1997-02-20 01:28  Miles Bader  <miles@gnu.ai.mit.edu>
1997-02-19 13:56  Miles Bader  <miles@gnu.ai.mit.edu>
1997-02-18 15:39  Miles Bader  <miles@gnu.ai.mit.edu>
1997-02-17 10:58  Miles Bader  <miles@gnu.ai.mit.edu>
1997-02-15 10:23  Miles Bader  <miles@gnu.ai.mit.edu>
	(mutex_lock, mutex_unlock, mutex_trylock): Defined in terms of
	__mutex_*.
	(mutex_t): Type removed & replaced by new macro.
	(tsd_key_t): Typedef to int instead of pthread_key_t.
	(tsd_key_create, tsd_setspecific, tsd_getspecific): New macros.
	(__pthread_initialize): New macro, work around assumption of pthreads.

	* sysdeps/mach/hurd/i386/init-first.c (__libc_argv, __libc_argc):
	__hurd_sigthread_stack_end, __hurd_sigthread_stack_variables,
	__hurd_threadvar_max, __hurd_threadvar_stack_offset,
	__hurd_threadvar_stack_mask): Variables removed.
1997-02-14 14:07  Miles Bader  <miles@gnu.ai.mit.edu>
	* hurd/hurd.h (_hurd_pids_changed_stamp, _hurd_pids_changed_sync):
1997-02-24 17:06  Geoffrey Keating  <geoffk@discus.anu.edu.au>

	* sysdeps/unix/sysv/linux/accept.S (NARGS): Describe number of
	arguments taken, for sysdeps/unix/sysv/linux/powerpc/socket.S.
	* sysdeps/unix/sysv/linux/bind.S: Likewise.
	* sysdeps/unix/sysv/linux/connect.S: Likewise.
	* sysdeps/unix/sysv/linux/getpeername.S: Likewise.
	* sysdeps/unix/sysv/linux/getsockname.S: Likewise.
	* sysdeps/unix/sysv/linux/getsockopt.S: Likewise.
	* sysdeps/unix/sysv/linux/listen.S: Likewise.
	* sysdeps/unix/sysv/linux/recv.S: Likewise.
	* sysdeps/unix/sysv/linux/recvfrom.S: Likewise.
	* sysdeps/unix/sysv/linux/recvmsg.S: Likewise.
	* sysdeps/unix/sysv/linux/send.S: Likewise.
	* sysdeps/unix/sysv/linux/sendmsg.S: Likewise.
	* sysdeps/unix/sysv/linux/sendto.S: Likewise.
	* sysdeps/unix/sysv/linux/setsockopt.S: Likewise.
	* sysdeps/unix/sysv/linux/shutdown.S: Likewise.
	* sysdeps/unix/sysv/linux/socketpair.S: Likewise.

1997-02-15 04:51  Ulrich Drepper  <drepper@cygnus.com>

6 files changed, 527 insertions, 25 deletions

diff --git a/stdio-common/Makefile b/stdio-common/Makefile
index 814374b..d1d922e 100644
--- a/stdio-common/Makefile
+++ b/stdio-common/Makefile
@@ -25,7 +25,7 @@ headers	:= stdio_lim.h printf.h
 
 routines	:=							      \
 	ctermid cuserid							      \
-	vfprintf vprintf printf_fp reg-printf printf-prs _itoa		      \
+	vfprintf vprintf printf_fp reg-printf printf-prs _itoa printf_fphex   \
 	fprintf printf snprintf sprintf asprintf dprintf		      \
 	vfscanf 							      \
 	fscanf scanf sscanf						      \
diff --git a/stdio-common/printf_fp.c b/stdio-common/printf_fp.c
index dede144..8cff7a4 100644
--- a/stdio-common/printf_fp.c
+++ b/stdio-common/printf_fp.c
@@ -1,5 +1,5 @@
 /* Floating point output for `printf'.
-   Copyright (C) 1995, 1996 Free Software Foundation, Inc.
+   Copyright (C) 1995, 1996, 1997 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
    Written by Ulrich Drepper <drepper@gnu.ai.mit.edu>, 1995.
 
@@ -291,12 +291,12 @@ __printf_fp (FILE *fp,
       /* Check for special values: not a number or infinity.  */
       if (__isnanl (fpnum.ldbl))
 	{
-	  special = "NaN";
+	  special = isupper (info->spec) ? "NAN" : "nan";
 	  is_neg = 0;
 	}
       else if (__isinfl (fpnum.ldbl))
 	{
-	  special = "Inf";
+	  special = isupper (info->spec) ? "INF" : "inf";
 	  is_neg = fpnum.ldbl < 0;
 	}
       else
@@ -316,12 +316,12 @@ __printf_fp (FILE *fp,
       /* Check for special values: not a number or infinity.  */
       if (__isnan (fpnum.dbl))
 	{
-	  special = "NaN";
+	  special = isupper (info->spec) ? "NAN" : "nan";
 	  is_neg = 0;
 	}
       else if (__isinf (fpnum.dbl))
 	{
-	  special = "Inf";
+	  special = isupper (info->spec) ? "INF" : "inf";
 	  is_neg = fpnum.dbl < 0;
 	}
       else
diff --git a/stdio-common/printf_fphex.c b/stdio-common/printf_fphex.c
new file mode 100644
index 0000000..d442b56
--- /dev/null
+++ b/stdio-common/printf_fphex.c
@@ -0,0 +1,440 @@
+/* Print floating point number in hexadecimal notation according to
+   ISO C 9X.
+   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 <ctype.h>
+#include <ieee754.h>
+#include <math.h>
+#include <printf.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include "_itoa.h"
+#include "../locale/localeinfo.h"
+
+/* #define NDEBUG 1*/		/* Undefine this for debugging assertions.  */
+#include <assert.h>
+
+/* This defines make it possible to use the same code for GNU C library and
+   the GNU I/O library.	 */
+#ifdef USE_IN_LIBIO
+# include <libioP.h>
+# define PUT(f, s, n) _IO_sputn (f, s, n)
+# define PAD(f, c, n) _IO_padn (f, c, n)
+/* We use this file GNU C library and GNU I/O library.	So make
+   names equal.	 */
+# undef putc
+# define putc(c, f) _IO_putc_unlocked (c, f)
+# define size_t     _IO_size_t
+# define FILE	     _IO_FILE
+#else	/* ! USE_IN_LIBIO */
+# define PUT(f, s, n) fwrite (s, 1, n, f)
+# define PAD(f, c, n) __printf_pad (f, c, n)
+ssize_t __printf_pad __P ((FILE *, char pad, int n)); /* In vfprintf.c.  */
+#endif	/* USE_IN_LIBIO */
+
+/* Macros for doing the actual output.  */
+
+#define outchar(ch)							      \
+  do									      \
+    {									      \
+      register const int outc = (ch);					      \
+      if (putc (outc, fp) == EOF)					      \
+	return -1;							      \
+      ++done;								      \
+    } while (0)
+
+#define PRINT(ptr, len)							      \
+  do									      \
+    {									      \
+       int outlen = (len);						      \
+       const char *cp = (ptr);						      \
+       while (outlen-- > 0)						      \
+	 outchar (*cp++);						      \
+    } while (0)
+
+#define PADN(ch, len)							      \
+  do									      \
+    {									      \
+      if (PAD (fp, ch, len) != len)					      \
+	return -1;							      \
+      done += len;							      \
+    }									      \
+  while (0)
+
+
+int
+__printf_fphex (FILE *fp,
+		const struct printf_info *info,
+		const void *const *args)
+{
+  /* The floating-point value to output.  */
+  union
+    {
+      union ieee754_double dbl;
+      union ieee854_long_double ldbl;
+    }
+  fpnum;
+
+  /* Locale-dependent representation of decimal point.	*/
+  wchar_t decimal;
+
+  /* "NaN" or "Inf" for the special cases.  */
+  const char *special = NULL;
+
+  /* Buffer for the generated number string for the mantissa.  The
+     maximal size for the mantissa is 64 bits.  */
+  char numbuf[16];
+  char *numstr;
+  char *numend;
+  int negative;
+
+  /* The maximal exponent of two in decimal notation has 5 digits.  */
+  char expbuf[5];
+  char *expstr;
+  int expnegative;
+  int exponent;
+
+  /* Non-zero is mantissa is zero.  */
+  int zero_mantissa;
+
+  /* The leading digit before the decimal point.  */
+  char leading;
+
+  /* Precision.  */
+  int precision = info->prec;
+
+  /* Width.  */
+  int width = info->width;
+
+  /* Number of characters written.  */
+  int done = 0;
+
+
+  /* Figure out the decimal point character.  */
+  if (info->extra == 0)
+    {
+      if (mbtowc (&decimal, _NL_CURRENT (LC_NUMERIC, DECIMAL_POINT),
+		  strlen (_NL_CURRENT (LC_NUMERIC, DECIMAL_POINT))) <= 0)
+	decimal = (wchar_t) *_NL_CURRENT (LC_NUMERIC, DECIMAL_POINT);
+    }
+  else
+    {
+      if (mbtowc (&decimal, _NL_CURRENT (LC_MONETARY, MON_DECIMAL_POINT),
+		  strlen (_NL_CURRENT (LC_MONETARY, MON_DECIMAL_POINT))) <= 0)
+	decimal = (wchar_t) *_NL_CURRENT (LC_MONETARY, MON_DECIMAL_POINT);
+    }
+  /* Give default value.  */
+  if (decimal == L'\0')
+    decimal = L'.';
+
+
+  /* Fetch the argument value.	*/
+  if (info->is_long_double && sizeof (long double) > sizeof (double))
+    {
+      fpnum.ldbl.d = *(const long double *) args[0];
+
+      /* Check for special values: not a number or infinity.  */
+      if (__isnanl (fpnum.ldbl.d))
+	{
+	  special = isupper (info->spec) ? "NAN" : "nan";
+	  negative = 0;
+	}
+      else
+	{
+	  if (__isinfl (fpnum.ldbl.d))
+	    special = isupper (info->spec) ? "INF" : "inf";
+
+	  negative = fpnum.ldbl.d < 0;
+	}
+    }
+  else
+    {
+      fpnum.dbl.d = *(const double *) args[0];
+
+      /* Check for special values: not a number or infinity.  */
+      if (__isnan (fpnum.dbl.d))
+	{
+	  special = isupper (info->spec) ? "NAN" : "nan";
+	  negative = 0;
+	}
+      else
+	{
+	  if (__isinf (fpnum.dbl.d))
+	    special = isupper (info->spec) ? "INF" : "inf";
+
+	  negative = fpnum.dbl.d < 0;
+	}
+    }
+
+  if (special)
+    {
+      int width = info->prec > info->width ? info->prec : info->width;
+
+      if (negative || info->showsign || info->space)
+	--width;
+      width -= 3;
+
+      if (!info->left && width > 0)
+	PADN (' ', width);
+
+      if (negative)
+	outchar ('-');
+      else if (info->showsign)
+	outchar ('+');
+      else if (info->space)
+	outchar (' ');
+
+      PRINT (special, 3);
+
+      if (info->left && width > 0)
+	PADN (' ', width);
+
+      return done;
+    }
+
+  /* We are handling here only 64 and 80 bit IEEE foating point
+     numbers.  */
+  if (info->is_long_double == 0 || sizeof (double) == sizeof (long double))
+    {
+      /* We have 52 bits of mantissa plus one implicit digit.  Since
+	 52 bits are representable without rest using hexadecimal
+	 digits we use only the implicit digits for the number before
+	 the decimal point.  */
+      if (sizeof (unsigned long int) > 6)
+	{
+	  unsigned long int num;
+
+	  num = (((unsigned long int) fpnum.dbl.ieee.mantissa0) << 32
+		 | fpnum.dbl.ieee.mantissa1);
+
+	  zero_mantissa = num == 0;
+
+	  numstr = _itoa_word (num, numbuf + sizeof numbuf, 16,
+			       info->spec == 'A');
+	}
+      else
+	{
+	  unsigned long long int num;
+
+	  num = (((unsigned long long int) fpnum.dbl.ieee.mantissa0) << 32
+		 | fpnum.dbl.ieee.mantissa1);
+
+	  zero_mantissa = num == 0;
+
+	  numstr = _itoa (num, numbuf + sizeof numbuf, 16, info->spec == 'A');
+	}
+
+      /* Fill with zeroes.  */
+      while (numstr > numbuf + (sizeof numbuf - 13))	/* 52 ÷ 4 = 13 */
+	*--numstr = '0';
+
+      leading = fpnum.dbl.ieee.exponent == 0 ? '0' : '1';
+
+      exponent = fpnum.dbl.ieee.exponent;
+
+      if (exponent != 0 && exponent < IEEE754_DOUBLE_BIAS)
+	{
+	  expnegative = 1;
+	  exponent = abs (exponent - IEEE754_DOUBLE_BIAS);
+	}
+      else
+	{
+	  expnegative = 0;
+	  if (exponent != 0)
+	    exponent -= IEEE754_DOUBLE_BIAS;
+	}
+    }
+  else
+    {
+      /* The "strange" 80 bit format on ix86 and m68k has an explicit
+	 leading digit in the 64 bit mantissa.  */
+      assert (sizeof (long double) == 12);
+
+      if (sizeof (unsigned long int) > 6)
+	{
+	  unsigned long int num;
+
+	  num = (((unsigned long int) fpnum.dbl.ieee.mantissa0) << 32
+		 | fpnum.dbl.ieee.mantissa1);
+
+	  zero_mantissa = num == 0;
+
+	  numstr = _itoa_word (num, numbuf + sizeof numbuf, 16,
+			       info->spec == 'A');
+	}
+      else
+	{
+	  unsigned long long int num;
+
+	  num = (((unsigned long long int) fpnum.ldbl.ieee.mantissa0) << 32
+		 | fpnum.ldbl.ieee.mantissa1);
+
+	  zero_mantissa = num == 0;
+
+	  numstr = _itoa (num, numbuf + sizeof numbuf, 16, info->spec == 'A');
+	}
+
+      /* We use a full nibble for the leading digit.  */
+      leading = *numstr++;
+
+      /* Fill with zeroes.  */
+      while (numstr > numbuf + (sizeof numbuf - 15))	/* 60 ÷ 4 = 15 */
+	*--numstr = '0';
+
+      /* We have 3 bits from the mantissa in the leading nibble.  */
+      exponent = fpnum.ldbl.ieee.exponent - 3;
+
+      if (exponent != 0 && exponent < IEEE854_LONG_DOUBLE_BIAS)
+	{
+	  expnegative = 1;
+	  exponent = abs (exponent - IEEE854_LONG_DOUBLE_BIAS);
+	}
+      else
+	{
+	  expnegative = 0;
+	  if (exponent != 0)
+	    exponent -= IEEE854_LONG_DOUBLE_BIAS;
+	}
+    }
+
+  /* Look for trailing zeroes.  */
+  if (! zero_mantissa)
+    {
+      numend = numbuf + sizeof numbuf;
+      while (numend[-1] == '0')
+	--numend;
+
+      if (precision == -1)
+	precision = numend - numstr;
+      else if (precision < numend - numstr
+	       && (numstr[precision] > 5
+		   || (numstr[precision] == 5
+		       && (precision + 1 < numend - numstr
+			   /* Round to even.  */
+			   || (precision > 0
+			       && ((numstr[precision - 1] & 1)
+				   ^ (isdigit (numstr[precision - 1]) == 0)))
+			   || (precision == 0
+			       && ((leading & 1)
+				   ^ (isdigit (leading) == 0)))))))
+	{
+	  /* Round up.  */
+	  int cnt = precision;
+	  while (--cnt >= 0)
+	    {
+	      char ch = numstr[cnt];
+	      /* We assume that the digits and the letters are ordered
+		 like in ASCII.  This is true for the rest of GNU, too.  */
+	      if (ch == '9')
+		{
+		  numstr[cnt] = info->spec;	/* This is tricky,
+						   think about it!  */
+		  break;
+		}
+	      else if (tolower (ch) < 'f')
+		{
+		  ++numstr[cnt];
+		  break;
+		}
+	      else
+		numstr[cnt] = '0';
+	    }
+	  if (cnt < 0)
+	    {
+	      /* The mantissa so far was fff...f  Now increment the
+		 leading digit.  Here it is again possible that we
+		 get an overflow.  */
+	      if (leading == '9')
+		leading = info->spec;
+	      else if (tolower (leading) < 'f')
+		++leading;
+	      else
+		{
+		  leading = 1;
+		  if (expnegative)
+		    {
+		      exponent += 4;
+		      if (exponent >= 0)
+			expnegative = 0;
+		    }
+		  else
+		    exponent += 4;
+		}
+	    }
+	}
+    }
+  else
+    numend = numstr;
+
+  /* Now we can compute the exponent string.  */
+  expstr = _itoa_word (exponent, expbuf + sizeof expbuf, 10, 0);
+
+  /* Now we have all information to compute the size.  */
+  width -= ((negative || info->showsign || info->space)
+	    /* Sign.  */
+	    + 2    + 1 + 1 + precision + 1 + 1
+	    /* 0x    h   .   hhh         P   ExpoSign.  */
+	    + ((expbuf + sizeof expbuf) - expstr));
+	    /* Exponent.  */
+
+  /* A special case if when the mantissa is zero and the `#' is not
+     given.  In this case we must not print the decimal point.  */
+  if (zero_mantissa && precision == 0 && !info->alt)
+    ++width;		/* This nihilates the +1 for the decimal-point
+			   character in the following equation.  */
+
+  if (!info->left && width > 0)
+    PADN (' ', width);
+
+  if (negative)
+    outchar ('-');
+  else if (info->showsign)
+    outchar ('+');
+  else if (info->space)
+    outchar (' ');
+
+  outchar ('0');
+  outchar (info->spec == 'A' ? 'X' : 'x');
+  outchar (leading);
+
+  if (!zero_mantissa || precision != 0 || info->alt)
+    outchar (decimal);
+
+  if (!zero_mantissa || precision != 0)
+    {
+      PRINT (numstr, MIN (numend - numstr, precision));
+      if (precision > numend - numstr)
+	PADN ('0', precision - (numend - numstr));
+    }
+
+  if (info->left && info->pad == '0' && width > 0)
+    PADN ('0', width);
+
+  outchar (info->spec == 'A' ? 'P' : 'p');
+
+  outchar (expnegative ? '-' : '+');
+
+  PRINT (expstr, (expbuf + sizeof expbuf) - expstr);
+
+  if (info->left && info->pad != '0' && width > 0)
+    PADN (info->pad, width);
+
+  return done;
+}
diff --git a/stdio-common/tstscanf.c b/stdio-common/tstscanf.c
index d76d95b..85be441 100644
--- a/stdio-common/tstscanf.c
+++ b/stdio-common/tstscanf.c
@@ -140,5 +140,15 @@ main (int argc, char **argv)
   if (out != stdout)
     pclose (out);
 
+  fputs ("Test 3:\n", out);
+  {
+    int res, val, n;
+
+    res = sscanf ("-242", "%3o%n", &val, &n);
+    printf ("res = %d, val = %d, n = %d\n", res, val, n);
+    if (res != 1 || val != -20 || n != 3)
+      return 1;
+  }
+
   exit(EXIT_SUCCESS);
 }
diff --git a/stdio-common/vfprintf.c b/stdio-common/vfprintf.c
index eda0d7b..c206c2f 100644
--- a/stdio-common/vfprintf.c
+++ b/stdio-common/vfprintf.c
@@ -238,7 +238,7 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
     /* '4' */  8, /* '5' */  8, /* '6' */  8, /* '7' */  8,
     /* '8' */  8, /* '9' */  8,            0,            0,
 	       0,            0,            0,            0,
-	       0,            0,            0, /* 'C' */ 25,
+	       0, /* 'A' */ 26,            0, /* 'C' */ 25,
 	       0, /* 'E' */ 19,            0, /* 'G' */ 19,
 	       0,            0,            0,            0,
     /* 'L' */ 12,            0,            0,            0,
@@ -246,7 +246,7 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
 	       0,            0,            0,            0,
     /* 'X' */ 18,            0, /* 'Z' */ 13,            0,
 	       0,            0,            0,            0,
-	       0,            0,            0, /* 'c' */ 20,
+	       0, /* 'a' */ 26,            0, /* 'c' */ 20,
     /* 'd' */ 15, /* 'e' */ 19, /* 'f' */ 19, /* 'g' */ 19,
     /* 'h' */ 10, /* 'i' */ 15,            0,            0,
     /* 'l' */ 11, /* 'm' */ 24, /* 'n' */ 23, /* 'o' */ 17,
@@ -270,7 +270,7 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
 
 #define STEP0_3_TABLE							      \
     /* Step 0: at the beginning.  */					      \
-    static const void *step0_jumps[26] =				      \
+    static const void *step0_jumps[27] =				      \
     {									      \
       REF (form_unknown),						      \
       REF (flag_space),		/* for ' ' */				      \
@@ -297,10 +297,11 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
       REF (form_pointer),	/* for 'p' */				      \
       REF (form_number),	/* for 'n' */				      \
       REF (form_strerror),	/* for 'm' */				      \
-      REF (form_wcharacter)	/* for 'C' */				      \
+      REF (form_wcharacter),	/* for 'C' */				      \
+      REF (form_floathex)	/* for 'A', 'a' */			      \
     };									      \
     /* Step 1: after processing width.  */				      \
-    static const void *step1_jumps[26] =				      \
+    static const void *step1_jumps[27] =				      \
     {									      \
       REF (form_unknown),						      \
       REF (form_unknown),	/* for ' ' */				      \
@@ -327,10 +328,11 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
       REF (form_pointer),	/* for 'p' */				      \
       REF (form_number),	/* for 'n' */				      \
       REF (form_strerror),	/* for 'm' */				      \
-      REF (form_wcharacter)	/* for 'C' */				      \
+      REF (form_wcharacter),	/* for 'C' */				      \
+      REF (form_floathex)	/* for 'A', 'a' */			      \
     };									      \
     /* Step 2: after processing precision.  */				      \
-    static const void *step2_jumps[26] =				      \
+    static const void *step2_jumps[27] =				      \
     {									      \
       REF (form_unknown),						      \
       REF (form_unknown),	/* for ' ' */				      \
@@ -357,10 +359,11 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
       REF (form_pointer),	/* for 'p' */				      \
       REF (form_number),	/* for 'n' */				      \
       REF (form_strerror),	/* for 'm' */				      \
-      REF (form_wcharacter)	/* for 'C' */				      \
+      REF (form_wcharacter),	/* for 'C' */				      \
+      REF (form_floathex)	/* for 'A', 'a' */			      \
     };									      \
     /* Step 3: after processing first 'l' modifier.  */			      \
-    static const void *step3_jumps[26] =				      \
+    static const void *step3_jumps[27] =				      \
     {									      \
       REF (form_unknown),						      \
       REF (form_unknown),	/* for ' ' */				      \
@@ -387,12 +390,13 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
       REF (form_pointer),	/* for 'p' */				      \
       REF (form_number),	/* for 'n' */				      \
       REF (form_strerror),	/* for 'm' */				      \
-      REF (form_wcharacter)	/* for 'C' */				      \
+      REF (form_wcharacter),	/* for 'C' */				      \
+      REF (form_floathex)	/* for 'A', 'a' */			      \
     }
 
 #define STEP4_TABLE							      \
     /* Step 4: processing format specifier.  */				      \
-    static const void *step4_jumps[26] =				      \
+    static const void *step4_jumps[27] =				      \
     {									      \
       REF (form_unknown),						      \
       REF (form_unknown),	/* for ' ' */				      \
@@ -419,7 +423,8 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
       REF (form_pointer),	/* for 'p' */				      \
       REF (form_number),	/* for 'n' */				      \
       REF (form_strerror),	/* for 'm' */				      \
-      REF (form_wcharacter)	/* for 'C' */				      \
+      REF (form_wcharacter),	/* for 'C' */				      \
+      REF (form_floathex)	/* for 'A', 'a' */			      \
     }
 
 
@@ -702,6 +707,53 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
       }									      \
       break;								      \
 									      \
+    LABEL (form_floathex):						      \
+      {									      \
+        /* FLoating point number printed as hexadecimal number.  */	      \
+	extern int __printf_fphex __P ((FILE *, const struct printf_info *,   \
+					const void **const));		      \
+	const void *ptr;						      \
+	int function_done;						      \
+									      \
+	if (fspec == NULL)						      \
+	  {								      \
+	    struct printf_info info = { prec: prec,			      \
+					width: width,			      \
+					spec: spec,			      \
+					is_long_double: is_long_double,	      \
+					is_short: is_short,		      \
+					is_long: is_long,		      \
+					alt: alt,			      \
+					space: space,			      \
+					left: left,			      \
+					showsign: showsign,		      \
+					group: group,			      \
+					pad: pad,			      \
+					extra: 0 };			      \
+									      \
+	    if (is_long_double)						      \
+	      the_arg.pa_long_double = va_arg (ap, long double);	      \
+	    else							      \
+	      the_arg.pa_double = va_arg (ap, double);			      \
+	    ptr = (const void *) &the_arg;				      \
+									      \
+	    function_done = __printf_fphex (s, &info, &ptr);		      \
+	  }								      \
+	else								      \
+	  {								      \
+	    ptr = (const void *) &args_value[fspec->data_arg];		      \
+									      \
+	    function_done = __printf_fphex (s, &fspec->info, &ptr);	      \
+	  }								      \
+									      \
+	if (function_done < 0)						      \
+	  /* Error in print handler.  */				      \
+	  return -1;							      \
+									      \
+	done += function_done;						      \
+      }									      \
+      break;								      \
+									      \
     LABEL (form_character):						      \
       /* Character.  */							      \
       if (is_long)							      \
@@ -881,7 +933,7 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap)
   done = 0;
   grouping = (const char *) -1;
 #ifdef __va_copy
-  /* This macro will be available soon in gcc's <stdarg.h>.  Me need it
+  /* This macro will be available soon in gcc's <stdarg.h>.  We need it
      since on some systems `va_list' is not an integral type.  */
   __va_copy (ap_save, ap);
 #else
@@ -1405,16 +1457,16 @@ printf_unknown (FILE *s, const struct printf_info *info,
   if (info->width != 0)
     {
       w = _itoa_word (info->width, workend + 1, 10, 0);
-      while (++w <= workend)
-	outchar (*w);
+      while (w <= workend)
+	outchar (*w++);
     }
 
   if (info->prec != -1)
     {
       outchar ('.');
       w = _itoa_word (info->prec, workend + 1, 10, 0);
-      while (++w <= workend)
-	outchar (*w);
+      while (w <= workend)
+	outchar (*w++);
     }
 
   if (info->spec != '\0')
diff --git a/stdio-common/vfscanf.c b/stdio-common/vfscanf.c
index f201c46..8653088 100644
--- a/stdio-common/vfscanf.c
+++ b/stdio-common/vfscanf.c
@@ -841,9 +841,9 @@ __vfscanf (FILE *s, const char *format, va_list argptr)
 	  if (flags & LONGDBL)
 	    {
 	      if (number_signed)
-		num.q = __strtoq_internal (wp, &tw, base, flags & GROUP);
+		num.q = __strtoll_internal (wp, &tw, base, flags & GROUP);
 	      else
-		num.uq = __strtouq_internal (wp, &tw, base, flags & GROUP);
+		num.uq = __strtoull_internal (wp, &tw, base, flags & GROUP);
 	    }
 	  else
 	    {