2018-01-01 00:32:25 +00:00
|
|
|
/* Copyright (C) 1992-2018 Free Software Foundation, Inc.
|
1996-11-15 04:08:00 +00:00
|
|
|
This file is part of the GNU C Library.
|
|
|
|
|
|
|
|
|
|
The GNU C Library is free software; you can redistribute it and/or
|
2001-07-06 04:58:11 +00:00
|
|
|
modify it under the terms of the GNU Lesser General Public
|
|
|
|
|
License as published by the Free Software Foundation; either
|
|
|
|
|
version 2.1 of the License, or (at your option) any later version.
|
1996-11-15 04:08:00 +00:00
|
|
|
|
|
|
|
|
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
|
2001-07-06 04:58:11 +00:00
|
|
|
Lesser General Public License for more details.
|
1996-11-15 04:08:00 +00:00
|
|
|
|
2001-07-06 04:58:11 +00:00
|
|
|
You should have received a copy of the GNU Lesser General Public
|
2012-02-09 23:18:22 +00:00
|
|
|
License along with the GNU C Library; if not, see
|
|
|
|
|
<http://www.gnu.org/licenses/>. */
|
1995-02-18 01:27:10 +00:00
|
|
|
|
|
|
|
|
#ifndef _ENDIAN_H
|
|
|
|
|
#define _ENDIAN_H 1
|
1997-06-21 02:59:26 +00:00
|
|
|
|
1995-02-18 01:27:10 +00:00
|
|
|
#include <features.h>
|
|
|
|
|
|
2000-07-24 05:41:25 +00:00
|
|
|
/* Definitions for byte order, according to significance of bytes,
|
|
|
|
|
from low addresses to high addresses. The value is what you get by
|
|
|
|
|
putting '4' in the most significant byte, '3' in the second most
|
|
|
|
|
significant byte, '2' in the second least significant byte, and '1'
|
|
|
|
|
in the least significant byte, and then writing down one digit for
|
|
|
|
|
each byte, starting with the byte at the lowest address at the left,
|
|
|
|
|
and proceeding to the byte with the highest address at the right. */
|
1995-02-18 01:27:10 +00:00
|
|
|
|
|
|
|
|
#define __LITTLE_ENDIAN 1234
|
|
|
|
|
#define __BIG_ENDIAN 4321
|
|
|
|
|
#define __PDP_ENDIAN 3412
|
|
|
|
|
|
|
|
|
|
/* This file defines `__BYTE_ORDER' for the particular machine. */
|
1997-06-21 02:59:26 +00:00
|
|
|
#include <bits/endian.h>
|
1995-02-18 01:27:10 +00:00
|
|
|
|
1998-11-27 11:34:40 +00:00
|
|
|
/* Some machines may need to use a different endianness for floating point
|
|
|
|
|
values. */
|
|
|
|
|
#ifndef __FLOAT_WORD_ORDER
|
|
|
|
|
# define __FLOAT_WORD_ORDER __BYTE_ORDER
|
|
|
|
|
#endif
|
|
|
|
|
|
2014-02-12 23:41:01 +00:00
|
|
|
#ifdef __USE_MISC
|
Update.
1997-10-12 05:09 Ulrich Drepper <drepper@cygnus.com>
* libio/Makefile (routines): Remove iofprintf.
* stdio-common/fprintf.c [USE_IN_LIBIO]: Define _IO_fprintf.
* libio/filedoalloc.c: Use _G_stat64 instead of stat.
* libio/fileops.c (_IO_file_open): Change to take extra argument
indicating whether 32 or 64 bit mode is wanted.
* libio/iofopen.c: Call _IO_file_open with extra argument set to 0.
* libio/iofopen64.c: Call _IO_file_open with extra argument set to 0.
* libio/iolibio.h (_IO_freopen, _IO_freopen64): Likewise.
* libio/iofgetpos.c: Pretty print.
* libio/iofgetpos64.c: Use _IO_fpos64_t for local variable `pos'.
* manual/conf.texi: Document all the _SC_ and _CS_ constants.
* manual/creature.texi: Document _LARGEFILE_SOURCE, _LARGEFILE64_SOURCE
and _FILE_OFFSET_BITS.
* manual/llio.texi: Document truncate and ftruncate.
* manual/stdio.texi: Document positional parameters for printf.
* math/Makefile (headers): Add tgmath.h.
(libm-support): Remove s_lrint, s_llrint, s_lround, and s_llround and
move to ...
(libm-calls): ... here. Add scalbln, s_nextafterx and s_fma.
* math/libm-test.c (lround_test, llround_test): Test for all FP formats
by using FUNC().
* math/libm.map: Add fma, fmaf, fmal, nextafterx, nextafterxf,
nextafterxl, scalbln, scalblnf, scalblnl, lrintf, lrintl, llrintf,
llrintl, lroundf, lroundl, llroundf, and llroundl.
* math/math.h: Document new platform specific macros from mathdef.h.
Remove declaration of lrint, llrint, lround, and llround.
* math/test-double.c: Define TEST_DOUBLE.
* math/test-idouble.c: Likewise.
* math/test-float.c: Define TEST_FLOAT.
* math/test-ifloat.c: Likewise.
* math/tgmath.h: New file.
* math/bits/mathcalls.h: Add nextafterx, scalbln, fma, lrint, llrint,
lround, and llround.
Change second argument of scalbn to `int'.
* sysdeps/libm-ieee754/s_fma.S: New file.
* sysdeps/libm-ieee754/s_fmaf.S: New file.
* sysdeps/libm-ieee754/s_fmal.S: New file.
* sysdeps/libm-i387/s_fma.S: New file.
* sysdeps/libm-i387/s_fmaf.S: New file.
* sysdeps/libm-i387/s_fmal.S: New file.
* sysdeps/libm-i387/s_llrint.S: Change to take double argument.
* sysdeps/libm-i387/s_lrint.S: Likewise.
* sysdeps/libm-i387/s_llrintf.S: New file.
* sysdeps/libm-i387/s_llrintl.S: New file.
* sysdeps/libm-i387/s_lrintf.S: New file.
* sysdeps/libm-i387/s_lrintl.S: New file.
* sysdeps/libm-ieee754/s_llrint.c: Remove version which works on
80bit double.
* sysdeps/libm-ieee754/s_lrint.c: Likewise.
* sysdeps/libm-ieee754/s_llrintf.S: New file.
* sysdeps/libm-ieee754/s_llrintl.S: New file.
* sysdeps/libm-ieee754/s_lrintf.S: New file.
* sysdeps/libm-ieee754/s_lrintl.S: New file.
* sysdeps/libm-i387/s_scalbln.c: New file. Empty file.
* sysdeps/libm-i387/s_scalblnf.c: New file. Empty file.
* sysdeps/libm-i387/s_scalblnl.c: New file. Empty file.
* sysdeps/libm-i387/s_scalbn.c: Add scalbln as alias.
* sysdeps/libm-i387/s_scalbnf.c: Add scalblnf as alias.
* sysdeps/libm-i387/s_scalbnl.c: Add scalblnl as alias.
* sysdeps/libm-ieee754/s_llround.c: Remove version which works on
80bit double.
* sysdeps/libm-ieee754/s_lround.c: Likewise.
* sysdeps/libm-ieee754/s_llroundf.c: Likewise.
* sysdeps/libm-ieee754/s_llroundl.c: Likewise.
* sysdeps/libm-ieee754/s_lroundf.c: Likewise.
* sysdeps/libm-ieee754/s_lroundl.c: Likewise.
* sysdeps/libm-ieee754/s_nextafterl.c: Add alias fo nextafterxl.
* sysdeps/libm-ieee754/s_nextafterx.c: New file.
* sysdeps/libm-ieee754/s_nextafterxf.c: New file.
* sysdeps/libm-ieee754/s_nextafterxl.c: New file.
* sysdeps/libm-ieee754/s_scalbln.c: New file.
* sysdeps/libm-ieee754/s_scalblnf.c: New file.
* sysdeps/libm-ieee754/s_scalblnl.c: New file.
* sysdeps/libm-ieee754/s_scalbn.c: Change to take `int' as second arg.
* sysdeps/libm-ieee754/s_scalbnf.c: Likewise.
* sysdeps/libm-ieee754/s_scalbnl.c: Likewise.
* stdlib/stdlib.h: Protect declarations of __strto*l_internal functions
by #ifdefs since they are duplicated in inttypes.h.
* sysdeps/wordsize-32/inttypes.h: Add definition of strtoimax and
strtoumax plus needed declarations.
* sysdeps/generic/confname.h (_SC_AIO_LISTIO_MAX): Fix typo.
1997-10-09 Andreas Jaeger <aj@arthur.rhein-neckar.de>
* locale/programs/locfile.c (locfile_read): Correct while loop.
* db2/makedb.c (main): Add missing parameter for error output.
(process_input): Likewise.
* resolv/gethnamaddr.c (getanswer): Rewrite a bit to avoid warning.
1997-10-12 05:05 Ulrich Drepper <drepper@cygnus.com>
* libc-map: Add __bzero, __mempcpy.
1997-10-10 18:51 David S. Miller <davem@tanya.rutgers.edu>
* sysdeps/unix/sysv/linux/sparc/bits/ioctls.h: Remove dependencies
on kernel_termios.h
1997-10-09 10:24 Thorsten Kukuk <kukuk@vt.uni-paderborn.de>
Add the changes from the Solaris 2.6 header files, use the new public
defines/functions.
* nis/nis_addmember.c: Updated.
* nis/nis_checkpoint.c: Updated.
* nis/nis_creategroup.c: updated.
* nis/nis_destroygroup.c: Updated.
* nis/nis_getservlist.c: Updated.
* nis/nis_ismember.c: Updated.
* nis/nis_lookup.c: Updated.
* nis/nis_modify.c: Updated.
* nis/nis_ping.c: Updated.
* nis/nis_print.c: Updated.
* nis/nis_print_group_entry.c: Updated.
* nis/nis_remove.c: Updated.
* nis/nis_removemember.c: Updated.
* nis/nis_xdr.c: Updated.
* nis/nss_nisplus/nisplus-alias.c: Updated.
* nis/nss_nisplus/nisplus-ethers.c: Updated.
* nis/nss_nisplus/nisplus-hosts.c: Updated.
* nis/nss_nisplus/nisplus-network.c: Updated.
* nis/nss_nisplus/nisplus-parser.c: Updated.
* nis/nss_nisplus/nisplus-proto.c: Updated.
* nis/nss_nisplus/nisplus-rpc.c: Updated.
* nis/nss_nisplus/nisplus-service.c: Updated.
* nis/rpcsvc/nis.h: Updated.
* nis/rpcsvc/nis.x: Updated.
* nis/rpcsvc/nis_object.x: Updated.
* nis/rpcsvc/nis_tags.h: Updated.
* nis/rpcsvc/nislib.h: Updated.
* nis/lckcache.c: Removed, since Sun has dropped the directory
signatures. The old cache version is now a security risk and not
longer supported by Sun.
* nis/nis_cache.c: Likewise.
* nis/rpcsvc/nis_cache.h: Likewise.
* nis/rpcsvc/nis_cache.x: Likewise.
* nis/nis_call.c: Remove calls to the cache functions.
* nis/libnsl.map: Remove cache and depending functions.
* nis/nis_intern.h: Likewise.
* nis/nis_add.c: Remove #include <rpcsvc/nislib.h>.
* nis/nis_domain_of.c: Likewise.
* nis/nis_domain_of_r.c: Likewise.
* nis/nis_error.c: Likewise.
* nis/nis_file.c: Likewise.
* nis/nis_local_names.c: Likewise.
* nis/nis_mkdir.c: Likewise.
* nis/nis_rmdir.c: Likewise.
* nis/nis_subr.c: Likewise.
* nis/nis_verifygroup.c: Likewise.
* nis/nis_clone.c: Removed, replaced by ...
* nis/nis_clone_dir.c: New.
* nis/nis_clone_obj.c: New.
* nis/nis_clone_res.c: New.
* nis/nis_table.c: Fixed bugs shown through the new clone functions.
* nis/nis_defaults.c: Fixed a lot of race conditions.
* nis/nis_free.c: Rewritten.
* sunrpc/auth_des.c: Fix use of free'ed pointer.
* nis/Makefile (libnsl-routines): Remove nis_clone, nis_cache and
lckcache. Add nis_clone_dir, nis_clone_obj, and nis_clone_res.
1997-10-09 Andreas Jaeger <aj@arthur.rhein-neckar.de>
* wctype/test_wctype.c (TEST): Add parens to avoid ambiguity.
1997-10-08 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
* include/features.h: Don't crash if _XOPEN_SOURCE is defined to
be empty.
1997-10-09 05:54 Ulrich Drepper <drepper@cygnus.com>
* nss/digits_dots.c: Place `result' in resbuf and not in `buffer'.
* nss/getXXbyYY_r.c: Make sure digits_dots.c sees `resbuf' as
struct and not a pointer. Little optimizations.
1997-10-09 05:00 Ulrich Drepper <drepper@cygnus.com>
* sysdeps/stub/getenv.c: Remove unused file.
* sysdeps/stub/lxstat.c: Likewise.
* sysdeps/stub/morecore.c: Likewise.
* sysdeps/stub/putenv.c: Likewise.
* sysdeps/stub/sbrk.c: Likewise.
* sysdeps/stub/setenv.c: Likewise.
* sysdeps/stub/sysd-stdio.c: Likewise.
* sysdeps/stub/sysdep.h: Likewise.
Reported by Zack Weinberg <zack@rabi.phys.columbia.edu>.
1997-10-09 04:58 Ulrich Drepper <drepper@cygnus.com>
* configure.in: Add __bzero definition to DWARF2 unwind test.
Reported by David S. Miller <davem@caip.rutgers.edu>.
1997-10-07 Paul Eggert <eggert@twinsun.com>
* intl/loadmsgcat.c (_nl_load_domain):
Fix &&/|| typo when checking file size.
Check for overflow when stuffing off_t into size_t.
1997-10-07 18:11 Ulrich Drepper <drepper@cygnus.com>
* time/africa: Update from tzdata1997i.
1997-10-07 Andreas Jaeger <aj@arthur.rhein-neckar.de>
* posix/globtest.sh: Add arguments for name of dynamic linker and
call dynamic linker to execute globtest.
* posix/Makefile (tests): Supply arguments to globtest.sh.
1997-10-07 Andreas Jaeger <aj@arthur.rhein-neckar.de>
* nis/rpcsvc/ypupd.h: Add missing __END_DECLS.
1997-10-03 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
* libc.map: Add mempcpy, prctl.
1997-09-30 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
* sysdeps/generic/memcmp.c: Avoid warnings.
* sysdeps/generic/memset.c: Likewise.
* sysdeps/generic/strchr.c: Likewise.
* sysdeps/generic/strlen.c: Likewise.
1997-09-29 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
* malloc/Makefile ($(objpfx)mtrace): Fix typo.
1997-09-29 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
* sysdeps/m68k/dl-machine.h (elf_machine_rela): Fix last change.
The R_68K_GLOB_DAT and R_68K_JMP_SLOT relocations really ignore
the addend, Richard.
(elf_machine_fixup_plt): Don't add the addend.
(elf_machine_plt_value): New function.
* sysdeps/alpha/dl-machine.h (elf_machine_plt_value): New
function.
* sysdeps/sparc/sparc32/dl-machine.h (elf_machine_plt_value): New
function.
* sysdeps/sparc/sparc64/dl-machine.h (elf_machine_plt_value): New
function.
* sysdeps/powerpc/dl-machine.h (elf_machine_plt_value): New
function.
* sysdeps/i386/dl-machine.h (elf_machine_plt_value): New
function.
* elf/dl-runtime.c (fixup, profile_fixup): Don't add in the
addend, instead let the machine dependent setup decide.
1997-09-20 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
* sysdeps/m68k/m68020/bits/string.h: New file.
1997-10-07 04:27 Richard Henderson <rth@cygnus.com>
* Makeconfig (+includes): Add -I$(objpfx).
* stdlib/longlong.h [__sparc__]: Prototype __udiv_qrnnd.
* sysdeps/alpha/setjmp.S: __setjmp is the same as _setjmp. Make
the former a strong symbol and the later a weak alias.
* sysdeps/sparc/sparc32/setjmp.S: Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc64/setjmp.S: Likewise.
1997-10-06 21:01 David S. Miller <davem@tanya.rutgers.edu>
* sysdeps/unix/sysv/linux/sparc/sparc64/bits/types.h: Make ino_t
64-bits.
* sysdeps/unix/sysv/linux/sparc/sparc64/kernel_stat.h: Make st_ino
member 64-bits as well, to match the kernel.
1997-10-06 19:35 Ulrich Drepper <drepper@cygnus.com>
* sysdeps/sparc/sparc64/sub_n.S: Fix typo.
Patch by Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>.
1997-10-06 01:09 Zack Weinberg <zack@rabi.phys.columbia.edu>
* time/README: Correct list of files from tzcode package. Add
contact information for tzcode/tzdata maintainers. Correct
spelling of author's name. Compact lists.
1997-10-06 01:48 Ulrich Drepper <drepper@cygnus.com>
* malloc/malloc.h: Remove hook definition without caller argument.
* malloc/malloc.c: Likewise.
* string/tester.c: Correct strsep test.
* string/bits/string2.h: Define __string2_1bptr_p and use it.
Patch by David S. Miller <davem@tanya.rutgers.edu>.
* math/Makefile (routines): Add s_clog10.
* math/libm-test.c: Add test for clog10.
* math/libm.map: Add clog10{,f,l}.
* math/bits/cmathcalls.h [__USE_GNU]: Add clog10.
* sysdeps/libm-ieee754/s_clog10.c: New file.
* sysdeps/libm-ieee754/s_clog10f.c: New file.
* sysdeps/libm-ieee754/s_clog10l.c: New file.
* manual/math.texi: Describe clog10.
* config.h.in: Add USE_REGPARMS and define internal_function based on
this.
* configure.in: Define USE_REGPARMS for ix86 machines.
* gmon/gmon.c: Mark write_hist, write_call_graph and write_bb_counts
as internal functions.
* inet/getnameinfo.c: Likewise for nrl_domainname.
* inet/getnetgrent_r.c: Likewise for __internal_setnetgrent_reuse.
* inet/rcmd.c: Likewise for __icheckhost.
* intl/dcgettext.c: Likewise for category_to_name and
guess_category_value.
* intl/localealias.c: Likewise for read_alias_file.
* io/fts.c: Likewise for fts_alloc, fts_build, fts_lfree,
fts_maxarglen, fts_padjust, fts_palloc, fts_sort, and fts_stat.
* libio/genops.c: Likewise for save_for_backup.
* malloc/malloc.c (chunk_free, chunk_alloc, chunk_realloc, chunk_align,
main_trim, heap_trim): Likewise.
* malloc/mtrace.c (tr_where): Likewise.
* misc/fstab.c (mnt2fs): Likewise.
* misc/getttyent.c (skip, value): Likewise.
* misc/syslog.c (openlog_internal): Likewise.
* misc/tsearch.c (trecurse, tdestroy_internal): Likewise.
* nss/nsswitch.c (nss_lookup_function, nss_parse_file, nss_getline,
nss_parse_service_list, nss_new_service): Likewise.
* posix/wordexp.c (parse_dollars, parse_backtick, eval_expr): Likewise.
* resolv/inet_ntop.c (inet_ntop4, inet_ntop6): Likewise.
* resolv/inet_pton.c (inet_pton4, inet_pton6): Likewise.
* resolv/res_init.c (res_setoptions): Likewise.
* stdio-common/printf_fp.c (group_number): Likewise.
* stdio-common/vfprintf.c (buffered_vfprintf, group_number): Likewise.
* stdlib/fmtmsg.c (internal_addseverity): Likewise.
* sunrpc/auth_des.c (synchronize): Likewise.
* sunrpc/auth_unix.c (marshal_new_auth): Likewise.
* sunrpc/clnt_perr.c (auth_errmsg): Likewise.
* sunrpc/key_call.c (key_call): Likewise.
* sunprc/pmap_rmt.c (getbroadcastnets): Likewise.
* sunrpc/svc_tcp.c (makefd_xprt): Likewise.
* sunrpc/svcauth_des.c (cache_init, cache_spot, cache_ref, invalidate):
Likewise.
* sunrpc/xdr_rec.c (fix_buf_size, skip_input_bytes, flush_out,
set_input_fragment, get_input_bytes): Likewise.
* sysdeps/unix/sysv/linux/getsysstats.c (get_proc_path,
phys_pages_info): Likewise.
* sysdeps/unix/sysv/linux/if_index.c (opensock): Likewise.
* sysdeps/unix/sysv/linux/poll.c (__emulate_poll): Likewise.
* sysdeps/unix/sysv/linux/readv.c (__atomic_readv_replacement):
Likewise.
* sysdeps/unix/sysv/linux/readv.c (__atomic_writev_replacement):
Likewise.
* time/strptime.c (strptime_internal): Likewise.
* time/tzfile.c (find_transition, compute_tzname_max): Likewise.
* time/tzset.c (compute_change, tz_compute, tzset_internal): Likewise.
* libc.map: Remove _libio_using_thunks, add _fp_hw and _dl_addr.
* ctype/ctype.h: Pretty print.
* grp/grp.h: Likewise.
* include/libc-symbols.h: Likewise.
* include/limits.h: Likewise.
* include/values.h: Likewise.
* io/fcntl.h: Likewise.
* io/sys/stat.h: Likewise.
* libio/stdio.h: Likewise.
* malloc/malloc.h: Likewise.
* misc/err.h: Likewise.
* misc/regexp.h: Likewise.
* misc/sys/cdefs.h: Likewise.
* misc/sys/file.h: Likewise.
* posix/sys/utsname.h: Likewise.
* posix/sys/wait.h: Likewise.
* pwd/pwd.h: Likewise.
* resolv/netdb.h: Likewise.
* signal/signal.h: Likewise.
* stdlib/stdlib.h: Likewise.
* string/endian.h: Likewise.
* string/memory.h: Likewise.
* sysdeps/mach/hurd/bits/fcntl.h: Likewise.
* sysdeps/mach/hurd/sys/param.h: Likewise.
* sysdeps/unix/sysv/linux/sys/param.h: Likewise.
* termios/termios.h: Likewise.
* wcsmbs/wchar.h: Likewise.
* wctype/wctype.h: Likewise.
* sysdeps/unix/bsd/bsd4.4/wait3.c: Use __WAIT_STATUS in definition.
Implement Large File Support API.
* include/features.h: Add suuport for _LARGEFILE_SOURCE,
_LARGEFILE64_SOURCE, and _FILE_OFFSET_BITS.
* libc.map: Add new functions for LFS.
* dirent/Makefile (routines): Add readdir64 and readdir64_r.
* dirent/dirent.h: Update readdir prototype for LFS and add new
prototypes for above functions.
* io/Makefile (routines): Add xstat64, fxstat64, lxstat64,
statfs64, fstatfs64, lstat64, open64, lseek64, creat64, and ftw64.
* io/creat64.c: New file.
* io/fstat64.c: New file.
* io/lstat64.c: New file.
* io/stat64.c: New file.
* io/ftw64.c: New file.
* io/ftw.c: Rewrite to allow easy definition of ftw64.
* io/ftw.h: Add LFS interface.
* io/fcntl.h: Likewise.
* io/sys/stat.h: Likewise.
* io/sys/statfs.h: Likewise.
* libio/Makefile (routines): Add iofgetpos64, iofopen64, iofsetpos64,
freopen64, fseeko64, and ftello64.
* libcio/fseeko64.c: New file.
* libio/ftello64.c: New file.
* libio/iofgetpos64.c: New file.
* libio/iofopen64.c: New file.
* libio/iofsetpos64.c: New file.
* libio/fileops.c (_IO_file_fopen): Change to use _IO_off64_t.
(_IO_file_attach): Likewise.
(_IO_do_write): Likewise.
(_IO_file_sync): Likewise.
(_IO_file_seek): Likewise.
(_IO_file_seekoff): Likewise. Use _G_stat64.
(_IO_file_fopen64): New function.
(_IO_file_jumps): Initialize showmanyc and imbue.
* libio/genops.c (_IO_default_seekpos): Change to use _IO_fpos64_t.
(_IO_default_seekoff): Likewise.
(_IO_default_seek): Likewise.
(_IO_default_showmanyc, _IO_default_imbue): New functions.
* libio/iofopncook.c (_IO_cookie_seek): Change to use _IO_off64_t.
* libio/iolibio.h: Add prototypes for LFS functions.
* libio/ioseekoff.c: Change to use _IO_fpos64_t.
* libio/ioseekpos.c: Likewise.
* libio/libio.h: Define _IO_fpos64_t and _IO_off64_t.
(_IO_FILE): Move _offset field to end and change type to _IO_off64_t.
(_IO_seekoff, _IO_seekpos): Change prototype.
* libio/libioP.h (_IO_seekoff_t, _IO_seekpos_t, _IO_seek_t): Change
to use _IO_off64_t.
Change prototypes for function from the *ops.c files.
* libio/stdio.h: Add LFS interface definition.
* libio/strops.c (_IO_str_seekoff): Change to use _IO_fpos64_t.
* posix/Makefile (routines): Add pread64 and pwrite64.
* posix/confstr.c: Handle _CS_LFS* requests.
* posix/getconf.c: Handle LFS* requests.
* sysdeps/generic/confname.h: Add _CS_LFS* constants.
* posix/unistd.h: Document _LFS64_LARGEFILE and _LFS64_STDIO.
Define off_t and off64_t appropriately. Change prototypes of
LFS functions.
* posix/sys/types.h: Add LFS types.
* resources/Makefile (routines): Add getrlimit64 and setlimit64.
* resource/sys/resource.h: Change prototypes of LFS functions.
* stdio-common/Makefile (routines): Add tmpfile64.
* stdio-common/tmpfile64.c: New file.
* sysdeps/generic/_G_config.h: Define _G_fpos64_t and _G_off64_t.
Define _G_OPEN64, _G_LSEEK64, _G_FSTAT64.
* sysdeps/unix/sysv/linux/_G_config.h: Likewise.
* sysdeps/generic/bits/resource.h: Add LFS definitions.
* sysdeps/unix/bsd/sun/sunos4/bits/resource.h: Likewise.
* sysdeps/unix/sysv/linux/bits/resource.h: Likewise.
* sysdeps/generic/statfs.h: Use __fsblkcnt_t for some of the fields.
* sysdeps/unix/sysv/linux/bits/statfs.h: Likewise.
* sysdeps/unix/sysv/linux/mips/bits/statfs.h: Likewise.
* sysdeps/generic/types.h: Define LFS types.
* sysdeps/unix/sysv/linux/alpha/bits/types.h: Likewise.
* sysdeps/unix/sysv/linux/bits/types.h: Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc64/bits/types.h: Likewise.
* sysdeps/generic/sys/mman.h: Add LFS definitions.
* sysdeps/unix/sysv/linux/sys/mman.h: Likewise.
* sysdeps/generic/mach/hurd/bits/fcntl.h: Add flock LFS extensions.
* sysdeps/unix/bsd/bits/fcntl.h: Likewise.
* sysdeps/unix/sysv/linux/alpha/bits/fcntl.h: Likewise.
* sysdeps/unix/sysv/linux/bits/fcntl.h: Likewise.
* sysdeps/unix/sysv/linux/mips/bits/fcntl.h: Likewise.
* sysdeps/generic/mach/hurd/bits/stat.h: Add stat LFS extensions.
* sysdeps/unix/bsd/bits/stat.h: Likewise.
* sysdeps/unix/bsd/osf/alpha/bits/stat.h: Likewise.
* sysdeps/unix/sysv/linux/alpha/bits/stat.h: Likewise.
* sysdeps/unix/sysv/linux/bits/stat.h: Likewise.
* sysdeps/unix/sysv/linux/mips/bits/stat.h: Likewise.
* sysdeps/unix/sysv/linux/sparc/bits/stat.h: Likewise.
* sysdeps/unix/sysv/sysv4/i386/bits/stat.h: Likewise.
* sysdeps/unix/sysv/sysv4/solaris2/bits/stat.h: Likewise.
* sysdeps/posix/open64.c: New file.
* sysdeps/stub/fstatfs64.c: New file.
* sysdeps/stub/fxstat64.c: New file.
* sysdeps/stub/getrlimit64.c: New file.
* sysdeps/stub/lseek64.c: New file.
* sysdeps/stub/lxstat64.c: New file.
* sysdeps/stub/open64.c: New file.
* sysdeps/stub/pread64.c: New file.
* sysdeps/stub/pwrite64.c: New file.
* sysdeps/stub/readdir64.c: New file.
* sysdeps/stub/readdir64_r.c: New file.
* sysdeps/stub/setrlimit64.c: New file.
* sysdeps/stub/statfs64.c: New file.
* sysdeps/stub/xstat64.c: New file.
* sysdeps/unix/sysv/linux/llseek.c: Define as __llseek and make
llseek and lseek64 weak aliases.
* sysdeps/unix/sysv/linux/lseek64.c: New file. Empty.
* sysdeps/unix/sysv/linux/alpha/bits/dirent.h: New file.
* sysdeps/unix/sysv/linux/bits/dirent.h: Add LFS definitions.
* sysdeps/posix/tempname.c: Add extra argument to trigger use of
open64.
* sysdeps/stub/tempname.c: Likewise.
* stdio-common/tempnam.c: Call __stdio_gen_tempname with extra
argument.
* stdio-common/tmpfile.c: Likewise.
* stdio-common/tmpnam.c: Likewise.
* stdio-common/tmpnam_r.c: Likewise.
* libio/libioP.h: Add definition ofr showmanyc and imbue callbacks.
* libio/fileops.c (_IO_file_jumps): Initialize showmanyc and imbue.
* libio/iofopncook.c (_IO_cookie_jumps): Likewise.
* libio/iopopen.c (_IO_proc_jumps): Likewise.
* libio/memstream.c (_IO_mem_jumps): Likewise.
* libio/obprintf.c (_IO_obstack_jumps): Likewise.
* libio/vsnprintf.c (_IO_strn_jumps): Likewise.
* libio/strops.c (_IO_str_jumps): Likewise.
* manual/arith.texi: Add a few words why cabs should be used.
* manual/llio.texi: Describe sync, fsync, fdatasync.
Tell about cleanup handlers & fcntl,lseek,write,read,close,open.
* manual/process.texi: Tell about cleanup handlers & system,waitpid,
wait.
* manual/signal.texi: Likewise for pause.
* manual/terminal.texi: Likewise for tcdrain.
* manual/time.texi: Document nanosleep.
* posix/exevp.c: Don't use nested function.
* stdlib/ucontext.h: New file.
* sysdeps/i386/sys/ucontext.h: New file. SysV/i386 API definitions.
* sunrpc/xcrypt.c (hexval): Make a macro for efficiency.
* sysdeps/i386/setjmp.h: Make `here` label local.
* sysdeps/i386/elf/start.S: Define _fp_hw "variable".
* sysdeps/stub/fstatfs.c: Correct warning.
* sysdeps/stub/fxstat.c: Likewise.
* sysdeps/stub/lxstat.c: Likewise.
* sysdeps/unix/sysv/i386/i686/time.S: New file.
1997-10-03 20:56 Jason Merrill <jason@yorick.cygnus.com>
* malloc/obstack.h (obstack_empty_p): New macro.
1997-10-04 17:41 Philip Blundell <Philip.Blundell@pobox.com>
* inet/getnameinfo.c (getnameinfo): Remove spurious `#if INET6'.
1997-09-30 Zack Weinberg <zack@rabi.phys.columbia.edu>
* maint.texi: Add copyright terms for libdb (Sleepycat, Harvard).
Document new --with-binutils switch; delete reference to
--with-gnu-as, --with-gnu-ld, --with-gnu-binutils.
Add to description of --without-fp: a kernel FPU emulator
is adequate (from FAQ)
* INSTALL: Regenerated.
1997-09-30 17:29 Richard Henderson <rth@cygnus.com>
* sysdeps/sparc/sparc32/dl-machine.h (elf_machine_rela): Move
_dl_hwcap declaration to ...
(elf_machine_fixup_plt): ... here.
1997-10-12 04:05:44 +00:00
|
|
|
# define LITTLE_ENDIAN __LITTLE_ENDIAN
|
|
|
|
|
# define BIG_ENDIAN __BIG_ENDIAN
|
|
|
|
|
# define PDP_ENDIAN __PDP_ENDIAN
|
|
|
|
|
# define BYTE_ORDER __BYTE_ORDER
|
1995-02-18 01:27:10 +00:00
|
|
|
#endif
|
|
|
|
|
|
2000-07-06 23:42:33 +00:00
|
|
|
#if __BYTE_ORDER == __LITTLE_ENDIAN
|
|
|
|
|
# define __LONG_LONG_PAIR(HI, LO) LO, HI
|
|
|
|
|
#elif __BYTE_ORDER == __BIG_ENDIAN
|
|
|
|
|
# define __LONG_LONG_PAIR(HI, LO) HI, LO
|
|
|
|
|
#endif
|
|
|
|
|
|
2008-05-15 02:54:33 +00:00
|
|
|
|
2014-02-12 23:41:01 +00:00
|
|
|
#if defined __USE_MISC && !defined __ASSEMBLER__
|
2008-05-15 02:54:33 +00:00
|
|
|
/* Conversion interfaces. */
|
2008-05-21 23:39:35 +00:00
|
|
|
# include <bits/byteswap.h>
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# include <bits/uintn-identity.h>
|
2008-05-21 23:39:35 +00:00
|
|
|
|
|
|
|
|
# if __BYTE_ORDER == __LITTLE_ENDIAN
|
|
|
|
|
# define htobe16(x) __bswap_16 (x)
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define htole16(x) __uint16_identity (x)
|
2008-05-21 23:39:35 +00:00
|
|
|
# define be16toh(x) __bswap_16 (x)
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define le16toh(x) __uint16_identity (x)
|
2008-05-21 23:39:35 +00:00
|
|
|
|
|
|
|
|
# define htobe32(x) __bswap_32 (x)
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define htole32(x) __uint32_identity (x)
|
2008-05-21 23:39:35 +00:00
|
|
|
# define be32toh(x) __bswap_32 (x)
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define le32toh(x) __uint32_identity (x)
|
2008-05-21 23:39:35 +00:00
|
|
|
|
2013-01-11 21:13:25 +00:00
|
|
|
# define htobe64(x) __bswap_64 (x)
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define htole64(x) __uint64_identity (x)
|
2013-01-11 21:13:25 +00:00
|
|
|
# define be64toh(x) __bswap_64 (x)
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define le64toh(x) __uint64_identity (x)
|
2012-03-28 14:39:25 +00:00
|
|
|
|
2008-05-21 23:39:35 +00:00
|
|
|
# else
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define htobe16(x) __uint16_identity (x)
|
2008-05-21 23:39:35 +00:00
|
|
|
# define htole16(x) __bswap_16 (x)
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define be16toh(x) __uint16_identity (x)
|
2008-05-21 23:39:35 +00:00
|
|
|
# define le16toh(x) __bswap_16 (x)
|
|
|
|
|
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define htobe32(x) __uint32_identity (x)
|
2008-05-21 23:39:35 +00:00
|
|
|
# define htole32(x) __bswap_32 (x)
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define be32toh(x) __uint32_identity (x)
|
2008-05-21 23:39:35 +00:00
|
|
|
# define le32toh(x) __bswap_32 (x)
|
|
|
|
|
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define htobe64(x) __uint64_identity (x)
|
2013-01-11 21:13:25 +00:00
|
|
|
# define htole64(x) __bswap_64 (x)
|
Make endian-conversion macros always return correct types (bug 16458).
Bug 16458 reports that the endian-conversion macros in <endian.h> and
<netinet/in.h>, in the case where no endian conversion is needed, just
return their arguments without converting to the expected return type,
so failing to act as expected for a macro version of a function. (The
<netinet/in.h> macros, in particular, are described with prototypes in
POSIX so should act like correspondingly prototyped functions.)
Where previously this was a fairly obscure issue, it now results in
glibc build with GCC mainline breaking for big-endian systems:
nss_hesiod/hesiod-service.c: In function '_nss_hesiod_getservbyport_r':
nss_hesiod/hesiod-service.c:142:39: error: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size 6 [-Werror=format-truncation=]
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~
nss_hesiod/hesiod-service.c:142:38: note: using the range [1, -2147483648] for directive argument
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~
nss_hesiod/hesiod-service.c:142:3: note: format output between 2 and 12 bytes into a destination of size 6
snprintf (portstr, sizeof portstr, "%d", ntohs (port));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The port argument is passed as int to this function, so when ntohs
does not convert the compiler cannot tell that the result is within
the range of uint16_t. (I don't know if in fact it's possible for
out-of-range values to reach this function and so get truncated as
strings without this patch or as integers with it.)
This patch arranges for these macros to use identity functions to
ensure appropriate conversions while having warnings for implicit
conversions of function arguments that might not occur with a cast.
Tested for x86_64 and x86; with build-many-glibcs.py with GCC 6; and
with build-many-glibcs.py with GCC mainline for powerpc to test the
build fix.
[BZ #16458]
* bits/uintn-identity.h: New file.
* inet/netinet/in.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (ntohs): Use __uint16_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htonl): Use __uint32_identity.
[__BYTE_ORDER == __BIG_ENDIAN] (htohs): Use __uint16_identity.
* string/endian.h: Include <bits/uintn-identity.h>.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole16): Use
__uint16_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le16toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole32): Use
__uint32_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le32toh): Likewise.
[__BYTE_ORDER == __LITTLE_ENDIAN] (htole64): Use
__uint64_identity.
[__BYTE_ORDER == __LITTLE_ENDIAN] (le64toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe16): Use
__uint16_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be16toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe32): Use
__uint32_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be32toh): Likewise.
[__BYTE_ORDER != __LITTLE_ENDIAN] (htobe64): Use
__uint64_identity.
[__BYTE_ORDER != __LITTLE_ENDIAN] (be64toh): Likewise.
* string/Makefile (headers): Add bits/uintn-identity.h.
(tests): Add test-endian-types.
* string/test-endian-types.c: New file.
* inet/Makefile (tests): Add test-hnto-types.
* inet/test-hnto-types.c: New file.
2017-01-11 15:28:08 +00:00
|
|
|
# define be64toh(x) __uint64_identity (x)
|
2013-01-11 21:13:25 +00:00
|
|
|
# define le64toh(x) __bswap_64 (x)
|
2008-05-21 23:39:35 +00:00
|
|
|
# endif
|
2008-05-15 02:54:33 +00:00
|
|
|
#endif
|
|
|
|
|
|
1995-02-18 01:27:10 +00:00
|
|
|
#endif /* endian.h */
|
