diff options
author | Ulrich Drepper <drepper@redhat.com> | 1997-03-25 01:45:04 +0000 |
---|---|---|
committer | Ulrich Drepper <drepper@redhat.com> | 1997-03-25 01:45:04 +0000 |
commit | 3e5f55575217bd22a301a77eafe00f2f3a727107 (patch) | |
tree | 21f840213d3720c9fa8c8fdad17917897b5ee386 /stdio-common | |
parent | bc9f6000f6752153e5e1902259d5f491a88a1ae5 (diff) | |
download | glibc-3e5f55575217bd22a301a77eafe00f2f3a727107.tar.gz glibc-3e5f55575217bd22a301a77eafe00f2f3a727107.tar.xz glibc-3e5f55575217bd22a301a77eafe00f2f3a727107.zip |
Update.
1997-03-25 02:15 Ulrich Drepper <drepper@cygnus.com> * gmon/gmon.c: Optimize a bit by using more sets of records to write in a single writev call. * math/math.h: Add definitions of macros __MATHCALLX and __MATHDECLX. * math/mathcalls.h: Use __MATHCALLX for fabs, infnan, copysign, nan, isnan, nextafter, trunc, __fpclassify and signbit to mark as `const'. Use __MATHDECLX for isinf and finite. * sysdeps/generic/setenv.c [_LIBC]: Define __clearenv and make clearenv a weak alias. Implement complex exponential function. * sysdeps/libm-ieee754/s_cexp.c: New file. * sysdeps/libm-ieee754/s_cexpf.c: New file. * sysdeps/libm-ieee754/s_cexpl.c: New file. * sysdeps/libm-ieee754/s_nan.c: Define function as __nan and make nan a weak alias. * sysdeps/libm-ieee754/s_nanf.c: Likewise for nanf. * sysdeps/libm-ieee754/s_nanl.c: Likewise for nanl. * sysdeps/unix/sysv/linux/iovec.h: Don't use kernel header because of type clashes. Add all definitions here. * sysdeps/unix/sysv/linux/shmat.c: Likewise. Correct types according to XPG4.2. * sysdeps/unix/sysv/linux/shmdt.c: Likewise. * sysdeps/unix/sysv/linux/msgrcv.c: Likewise. * sysvipc/sys/shm.h (shmat, shmdt): Correct types. * sysdeps/unix/sysv/linux/sys/kd.h: Define _LINUX_TYPES_H to avoid use of kernel types. 1997-03-25 00:00 Ulrich Drepper <drepper@cygnus.com> * sysdeps/unix/sysv/linux/netinet/ip.h (struct ip_timestamp): Correct typos. Reported by a sun <asun@zoology.washington.edu>. 1997-03-20 21:58 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * sysdeps/libm-ieee754/s_asinhl.c: Fix sign of result. 1997-03-20 16:20 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * sysdeps/libm-ieee754/e_atan2l.c: Fix typo. 1997-03-20 14:23 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * sysdeps/m68k/fpu/__math.h (__logb, logb): Don't define. * sysdeps/m68k/fpu/s_logb.c: Removed. * sysdeps/m68k/fpu/s_logbf.c: Removed. * sysdeps/m68k/fpu/s_logbl.c: Removed. * sysdeps/m68k/fpu/__math.h (__ieee754_atan2, __ieee754_pow): Don't define here. * sysdeps/m68k/fpu/e_atan2.c: Rewritten. * sysdeps/m68k/fpu/e_pow.c: Likewise. * sysdeps/m68k/fpu/e_atan2f.c: Based on e_atan2.c. * sysdeps/m68k/fpu/e_atan2l.c: Likewise. * sysdeps/m68k/fpu/e_powf.c: Based on e_pow.c. * sysdeps/m68k/fpu/e_powl.c: Likewise. * sysdeps/m68k/fpu/s_log2.c: New file. * sysdeps/m68k/fpu/s_log2f.c: New file. * sysdeps/m68k/fpu/s_log2l.c: New file. * sysdeps/m68k/fpu/s_exp2.c: New file. * sysdeps/m68k/fpu/s_exp2f.c: New file. * sysdeps/m68k/fpu/s_exp2l.c: New file. 1997-03-20 14:46 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * Makeconfig (all): Make sure this is always the default goal. 1997-03-20 11:09 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * math/libm-test.c: Increase epsilon in many places. Fix many typos. Allow optinal argument also for short option. 1997-03-20 11:09 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * math/Makefile ($(objpfx)$(tests)): Link against libm. 1997-03-24 23:14 Ulrich Drepper <drepper@cygnus.com> * sysdeps/libm-ieee754/s_fpclassifyl.c: Correct recognition of denormalized numbers. * sysdeps/i386/huge_val.h: Remove references to byte order macros. Don't include <endian.h>. 1997-03-19 15:18 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * sysdeps/m68k/huge_val.h: Remove references to byte order macros. Don't include <endian.h>. 1997-03-24 23:09 Ulrich Drepper <drepper@cygnus.com> * inet/rcmd.c (iruserok): Use access instead of euidaccess. 1997-03-15 18:08 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * math/cmathcalls.h: Remove whitespace before function name in use of __MATHDECL macro, to make it compilable with a traditional preprocessor. 1997-03-24 15:31 Ulrich Drepper <drepper@cygnus.com> * configure.in: Use AC_PROG_CC instead of AC_PROG_TOOL to find compiler. 1997-03-24 02:34 Ulrich Drepper <drepper@cygnus.com> * sysdeps/libm-ieee754/s_finitel.c (__finitel): Shift return value by 31 positions to get 0/1 result. Patch by Joe Keane <jgk@jgk.org>. 1997-03-23 12:15 H.J. Lu <hjl@gnu.ai.mit.edu> * posix/getopt.c (__getopt_nonoption_flags): Make it extern to prevent from ld linking in getopt.o even if there is another incompatible one. * posix/getopt_init.c (__getopt_nonoption_flags): Remove extern. 1997-03-23 23:30 Thorsten Kukuk <kukuk@vt.uni-paderborn.de> * nis/nis_call.c (__do_niscall): Print message if cold start file does not exist. * nis/nis_file.c: Don't print error messages. * nis/nis_local_name.c (nis_local_host): Fix pointer errors. * nis/rpcsvc/ypclnt.h: Fix copyright and prototypes. * nis/rpcsvc/nis.h: Likewise. * nis/ypclnt.c: Fix prototypes. 1997-03-24 01:36 Ulrich Drepper <drepper@cygnus.com> * sysdeps/libm-i387/e_scalb.S: Pop additional result before returning. * sysdeps/libm-i387/e_scalbf.S: Likewise. * sysdeps/libm-i387/e_scalbl.S: Likewise. Reported by Andreas Jaeger <aj@arthur.pfalz.de>. * elf/rtld.c (dl_main): Set main_map->l_opencount to 1 also if dynamic linker was called implicitly by the kernel. Reported by Eric Youngdale <eric@andante.jic.com>. * math/Makefile (CFLAGS-test-float.c, CFLAGS-test-double.c, CFLAGS-test-longdouble.c): New variables. Set to -fno-inline to prevent clever optimizations which corrupt the tests. 1997-03-23 21:33 Andreas Jaeger <aj@arthur.pfalz.de> * math/libm-test.c (scalb_test): New function. (sqrt_test): New function. (scalbn_test): New function. (ilogb_test): New function. (main): Added calls for new test functions. (ldexp_test): Add another test for ldexp(x,0). 1997-03-23 12:35 Andreas Jaeger <aj@arthur.pfalz.de> * math/libm-test.c (remquo_test): Correct messages so that they match the tests. (copysign_test): Likewise. 1997-03-23 16:28 Ulrich Drepper <drepper@cygnus.com> * stdio-common/Makefile (routines): Add printf_size. * stdio-common/printf.h: Add declarations for printf_size and printf_size_info. * stdio-common/printf-parse.h (__printf_arginfo_table): Correct declaration. (parse_one_spec): Test whether __printf_function_table is not NULL before using registered handlers and don't test __printf_arginfo_table. Update nargs also when handler is registered. * stdio-common/vfprintf.c (vfprintf): Count number of specifiers processed in fast loop. * stdio-common/printf_size: New file. 1997-03-22 04:53 Ulgmon_out_readrich Drepper <drepper@cygnus.com>
Diffstat (limited to 'stdio-common')
-rw-r--r-- | stdio-common/Makefile | 2 | ||||
-rw-r--r-- | stdio-common/printf-parse.h | 27 | ||||
-rw-r--r-- | stdio-common/printf.h | 17 | ||||
-rw-r--r-- | stdio-common/printf_size.c | 250 | ||||
-rw-r--r-- | stdio-common/vfprintf.c | 4 |
5 files changed, 284 insertions, 16 deletions
diff --git a/stdio-common/Makefile b/stdio-common/Makefile index d1d922ed36..b09a3a2eeb 100644 --- a/stdio-common/Makefile +++ b/stdio-common/Makefile @@ -26,7 +26,7 @@ headers := stdio_lim.h printf.h routines := \ ctermid cuserid \ vfprintf vprintf printf_fp reg-printf printf-prs _itoa printf_fphex \ - fprintf printf snprintf sprintf asprintf dprintf \ + printf_size fprintf printf snprintf sprintf asprintf dprintf \ vfscanf \ fscanf scanf sscanf \ perror psignal \ diff --git a/stdio-common/printf-parse.h b/stdio-common/printf-parse.h index 141164b04b..f1fecdb275 100644 --- a/stdio-common/printf-parse.h +++ b/stdio-common/printf-parse.h @@ -1,5 +1,5 @@ /* Internal header for parsing printf format strings. - Copyright (C) 1995, 1996 Free Software Foundation, Inc. + Copyright (C) 1995, 1996, 1997 Free Software Foundation, Inc. This file is part of th GNU C Library. The GNU C Library is free software; you can redistribute it and/or @@ -106,8 +106,9 @@ find_spec (const char *format, mbstate_t *ps) } -/* This is defined in reg-printf.c. */ -extern printf_arginfo_function **__printf_arginfo_table; +/* These are defined in reg-printf.c. */ +extern printf_arginfo_function *__printf_arginfo_table[]; +extern printf_function **__printf_function_table; /* FORMAT must point to a '%' at the beginning of a spec. Fills in *SPEC @@ -299,8 +300,9 @@ parse_one_spec (const UCHAR_T *format, size_t posn, struct printf_spec *spec, /* Get the format specification. */ spec->info.spec = (wchar_t) *format++; - if (__printf_arginfo_table != NULL && - __printf_arginfo_table[spec->info.spec] != NULL) + if (__printf_function_table != NULL + && spec->info.spec <= UCHAR_MAX + && __printf_arginfo_table[spec->info.spec] != NULL) /* We don't try to get the types for all arguments if the format uses more than one. The normal case is covered though. */ spec->ndata_args = (*__printf_arginfo_table[spec->info.spec]) @@ -362,15 +364,14 @@ parse_one_spec (const UCHAR_T *format, size_t posn, struct printf_spec *spec, spec->ndata_args = 0; break; } + } - if (spec->data_arg == -1 && spec->ndata_args > 0) - { - /* There are args consumed, but no positional spec. - Use the next sequential arg position. */ - spec->data_arg = posn; - posn += spec->ndata_args; - nargs += spec->ndata_args; - } + if (spec->data_arg == -1 && spec->ndata_args > 0) + { + /* There are args consumed, but no positional spec. Use the + next sequential arg position. */ + spec->data_arg = posn; + nargs += spec->ndata_args; } if (spec->info.spec == L'\0') diff --git a/stdio-common/printf.h b/stdio-common/printf.h index 6d9e8a455f..c49172b88c 100644 --- a/stdio-common/printf.h +++ b/stdio-common/printf.h @@ -1,4 +1,4 @@ -/* Copyright (C) 1991, 92, 93, 95, 96 Free Software Foundation, Inc. +/* Copyright (C) 1991, 92, 93, 95, 96, 97 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 @@ -122,6 +122,21 @@ enum #define PA_FLAG_PTR (1 << 11) + +/* Function which can be registered as `printf'-handlers. */ + +/* Print floating point value using using abbreviations for the orders + of magnitude used for numbers ('k' for kilo, 'm' for mega etc). If + the format specifier is a uppercase character powers of 1000 are + used. Otherwise powers of 1024. */ +extern int printf_size __P ((FILE *__fp, __const struct printf_info *__info, + __const void *__const *args)); + +/* This is the appropriate argument information function for `printf_size'. */ +extern int printf_size_info __P ((__const struct printf_info *__info, + size_t __n, int *__argtypes)); + + __END_DECLS #endif /* printf.h */ diff --git a/stdio-common/printf_size.c b/stdio-common/printf_size.c new file mode 100644 index 0000000000..0184269181 --- /dev/null +++ b/stdio-common/printf_size.c @@ -0,0 +1,250 @@ +/* Print size value using units for orders of magnitude. + Copyright (C) 1997 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997. + Based on a proposal by Larry McVoy <lm@sgi.com>. + + 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> +#ifdef USE_IN_LIBIO +# include <libioP.h> +#else +# include <stdio.h> +#endif + + +/* This defines make it possible to use the same code for GNU C library and + the GNU I/O library. */ +#ifdef USE_IN_LIBIO +# 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 \ + { \ + register size_t outlen = (len); \ + if (len > 20) \ + { \ + if (PUT (fp, ptr, outlen) != outlen) \ + return -1; \ + ptr += outlen; \ + done += outlen; \ + } \ + else \ + { \ + while (outlen-- > 0) \ + outchar (*ptr++); \ + } \ + } while (0) + +#define PADN(ch, len) \ + do \ + { \ + if (PAD (fp, ch, len) != len) \ + return -1; \ + done += len; \ + } \ + while (0) + +/* Prototype for helper functions. */ +extern int __printf_fp (FILE *fp, const struct printf_info *info, + const void *const *args); + + +int +printf_size (FILE *fp, const struct printf_info *info, const void *const *args) +{ + /* Units for the both formats. */ + static const char units[2][8] = + { + " kmgtps", /* For binary format. */ + " KMGTPS" /* For decimal format. */ + }; + const char *tag = units[isupper (info->spec) != 0]; + int divisor = isupper (info->spec) ? 1000 : 1024; + + /* The floating-point value to output. */ + union + { + union ieee754_double dbl; + union ieee854_long_double ldbl; + } + fpnum; + const void *ptr = &fpnum; + + int negative = 0; + + /* "NaN" or "Inf" for the special cases. */ + const char *special = NULL; + + struct printf_info fp_info; + int done = 0; + + + /* 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 = "nan"; + negative = 0; + } + else if (__isinfl (fpnum.ldbl.d)) + { + special = "inf"; + + negative = fpnum.ldbl.d < 0; + } + else + while (fpnum.ldbl.d >= divisor && tag[1] != '\0') + { + fpnum.ldbl.d /= divisor; + ++tag; + } + } + else + { + fpnum.dbl.d = *(const double *) args[0]; + + /* Check for special values: not a number or infinity. */ + if (__isnan (fpnum.dbl.d)) + { + special = "nan"; + negative = 0; + } + else if (__isinf (fpnum.dbl.d)) + { + special = "inf"; + + negative = fpnum.dbl.d < 0; + } + else + while (fpnum.dbl.d >= divisor && tag[1] != '\0') + { + fpnum.dbl.d /= divisor; + ++tag; + } + } + + 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; + } + + /* Prepare to print the number. We want to use `__printf_fp' so we + have to prepare a `printf_info' structure. */ + fp_info.spec = 'f'; + fp_info.prec = info->prec < 0 ? 3 : info->prec; + fp_info.is_long_double = info->is_long_double; + fp_info.is_short = info->is_short; + fp_info.is_long = info->is_long; + fp_info.alt = info->alt; + fp_info.space = info->space; + fp_info.left = info->left; + fp_info.showsign = info->showsign; + fp_info.group = info->group; + fp_info.extra = info->extra; + fp_info.pad = info->pad; + + if (fp_info.left && fp_info.pad == L' ') + { + /* We must do the padding ourself since the unit character must + be placed before the padding spaces. */ + fp_info.width = 0; + + done = __printf_fp (fp, &fp_info, &ptr); + if (done > 0) + { + outchar (*tag); + if (info->width > done) + PADN (' ', info->width - done); + } + } + else + { + /* We can let __printf_fp do all the printing and just add our + unit character afterwards. */ + fp_info.width = info->width - 1; + + done = __printf_fp (fp, &fp_info, &ptr); + if (done > 0) + outchar (*tag); + } + + return done; +} + +/* This is the function used by `vfprintf' to determine number and + type of the arguments. */ +int +printf_size_info (const struct printf_info *info, size_t n, int *argtypes) +{ + /* We need only one double or long double argument. */ + if (n >= 1) + argtypes[0] = PA_DOUBLE | (info->is_long_double ? PA_FLAG_LONG_DOUBLE : 0); + + return 1; +} diff --git a/stdio-common/vfprintf.c b/stdio-common/vfprintf.c index c206c2f0ac..856c65dbb5 100644 --- a/stdio-common/vfprintf.c +++ b/stdio-common/vfprintf.c @@ -1117,7 +1117,6 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap) is_long = sizeof (size_t) > sizeof (unsigned int); JUMP (*++f, step4_jumps); - /* Process current format. */ while (1) { @@ -1136,6 +1135,9 @@ vfprintf (FILE *s, const CHAR_T *format, va_list ap) goto do_positional; } + /* The format is correctly handled. */ + ++nspecs_done; + /* Look for next format specifier. */ f = find_spec ((end_of_spec = ++f), &mbstate); |