1 GNU C Library NEWS -- history of user-visible changes.
2 Copyright (C) 1992-2019 Free Software Foundation, Inc.
3 See the end for copying conditions.
5 Please send GNU C library bug reports via <https://sourceware.org/bugzilla/>
6 using `glibc' in the "product" field.
12 * Unicode 12.1.0 Support: Character encoding, character type info, and
13 transliteration tables are all updated to Unicode 12.1.0, using
14 generator scripts contributed by Mike FABIAN (Red Hat).
16 * The dynamic linker accepts the --preload argument to preload shared
17 objects, in addition to the LD_PRELOAD environment variable.
19 * The twalk_r function has been added. It is similar to the existing
20 twalk function, but it passes an additional caller-supplied argument
21 to the callback function.
23 * On Linux, the getdents64, gettid, and tgkill functions have been added.
25 * Minguo (Republic of China) calendar support has been added as an
26 alternative calendar for the following locales: zh_TW, cmn_TW, hak_TW,
29 * The entry for the new Japanese era has been added for ja_JP locale.
31 * Memory allocation functions malloc, calloc, realloc, reallocarray, valloc,
32 pvalloc, memalign, and posix_memalign fail now with total object size
33 larger than PTRDIFF_MAX. This is to avoid potential undefined behavior with
34 pointer subtraction within the allocated object, where results might
35 overflow the ptrdiff_t type.
37 * The dynamic linker no longer refuses to load objects which reference
38 versioned symbols whose implementation has moved to a different soname
39 since the object has been linked. The old error message, symbol
40 FUNCTION-NAME, version SYMBOL-VERSION not defined in file DSO-NAME with
41 link time reference, is gone.
43 * Add new POSIX-proposed pthread_cond_clockwait, pthread_mutex_clocklock,
44 pthread_rwlock_clockrdlock, pthread_rwlock_clockwrlock and sem_clockwait
45 functions. These behave similarly to their "timed" equivalents, but also
46 accept a clockid_t parameter to determine which clock their timeout should
47 be measured against. All functions allow waiting against CLOCK_MONOTONIC
48 and CLOCK_REALTIME. The decision of which clock to be used is made at the
49 time of the wait (unlike with pthread_condattr_setclock, which requires
50 the clock choice at initialization time).
52 Deprecated and removed features, and other changes affecting compatibility:
54 * The copy_file_range function fails with ENOSYS if the kernel does not
55 support the system call of the same name. Previously, user space
56 emulation was performed, but its behavior did not match the kernel
57 behavior, which was deemed too confusing. Applications which use the
58 copy_file_range function can no longer rely on glibc to provide a fallback
59 on kernels that do not support the copy_file_range system call, and if
60 this function returns ENOSYS, they will need to use their own fallback.
61 Support for copy_file_range for most architectures was added in version
62 4.5 of the mainline Linux kernel.
64 * The functions clock_gettime, clock_getres, clock_settime,
65 clock_getcpuclockid, clock_nanosleep were removed from the librt library
66 for new applications (on architectures which had them). Instead, the
67 definitions in libc will be used automatically, which have been available
70 * The obsolete and never-implemented XSI STREAMS header files <stropts.h>
71 and <sys/stropts.h> have been removed.
73 * Support for the "inet6" option in /etc/resolv.conf and the RES_USE_INET6
74 resolver flag (deprecated in glibc 2.25) have been removed.
76 * The obsolete RES_INSECURE1 and RES_INSECURE2 option flags for the DNS stub
77 resolver have been removed from <resolv.h>.
79 * With --enable-bind-now, installed programs are now linked with the
82 * Support for the PowerPC SPE ISA extension (powerpc-*-*gnuspe*
83 configurations) has been removed, following the deprecation of this
84 subarchitecture in version 8 of GCC, and its removal in version 9.
86 * On 32-bit Arm, support for the port-based I/O emulation and the <sys/io.h>
87 header have been removed.
89 * The Linux-specific <sys/sysctl.h> header and the sysctl function have been
90 deprecated and will be removed from a future version of glibc.
91 Application should directly access /proc instead. For obtaining random
92 bits, the getentropy function can be used.
94 Changes to build and runtime requirements:
96 * GCC 6.2 or later is required to build the GNU C Library.
98 Older GCC versions and non-GNU compilers are still supported when
99 compiling programs that use the GNU C Library.
101 Security related changes:
103 CVE-2019-7309: x86-64 memcmp used signed Jcc instructions to check
104 size. For x86-64, memcmp on an object size larger than SSIZE_MAX
105 has undefined behavior. On x32, the size_t argument may be passed
106 in the lower 32 bits of the 64-bit RDX register with non-zero upper
107 32 bits. When it happened with the sign bit of RDX register set,
108 memcmp gave the wrong result since it treated the size argument as
109 zero. Reported by H.J. Lu.
111 CVE-2019-9169: Attempted case-insensitive regular-expression match
112 via proceed_next_node in posix/regexec.c leads to heap-based buffer
113 over-read. Reported by Hongxu Chen.
115 The following bugs are resolved with this release:
117 [The release manager will add the list generated by
118 scripts/list-fixed-bugs.py just before the release.]
125 * The getcpu wrapper function has been added, which returns the currently
126 used CPU and NUMA node. This function is Linux-specific.
128 * A new convenience target has been added for distribution maintainers
129 to build and install all locales as directories with files. The new
130 target is run by issuing the following command in your build tree:
131 'make localedata/install-locale-files', with an optional DESTDIR
132 to set the install root if you wish to install into a non-default
135 * Optimized generic exp, exp2, log, log2, pow, sinf, cosf, sincosf and tanf.
137 * The reallocarray function is now declared under _DEFAULT_SOURCE, not just
138 for _GNU_SOURCE, to match BSD environments.
140 * For powercp64le ABI, Transactional Lock Elision is now enabled iff kernel
141 indicates that it will abort the transaction prior to entering the kernel
142 (PPC_FEATURE2_HTM_NOSC on hwcap2). On older kernels the transaction is
143 suspended, and this caused some undefined side-effects issues by aborting
144 transactions manually. Glibc avoided it by abort transactions manually on
145 each syscall, but it lead to performance issues on newer kernels where the
146 HTM state is saved and restore lazily (the state being saved even when the
147 process actually does not use HTM).
149 * The functions posix_spawn_file_actions_addchdir_np and
150 posix_spawn_file_actions_addfchdir_np have been added, enabling
151 posix_spawn and posix_spawnp to run the new process in a different
152 directory. These functions are GNU extensions. The function
153 posix_spawn_file_actions_addchdir_np is similar to the Solaris function
156 * The popen and system do not run atfork handlers anymore (BZ#17490).
157 Although it is a possible POSIX violation, the POSIX rationale in
158 pthread_atfork documentation regarding atfork handlers is to handle
159 inconsistent mutex state after a fork call in a multi-threaded process.
160 In both popen and system there is no direct access to user-defined mutexes.
162 * Support for the C-SKY ABIV2 running on Linux has been added. This port
163 requires at least binutils-2.32, gcc-9.0, and linux-4.20. Two ABIs are
165 - C-SKY ABIV2 soft-float little-endian
166 - C-SKY ABIV2 hard-float little-endian
168 * strftime's default formatting of a locale's alternative year (%Ey)
169 has been changed to zero-pad the year to a minimum of two digits,
170 like "%y". This improves the display of Japanese era years during
171 the first nine years of a new era, and is expected to be harmless
172 for all other locales (only Japanese locales regularly have
173 alternative year numbers less than 10). Zero-padding can be
174 overridden with the '_' or '-' flags (which are GNU extensions).
176 * As a GNU extension, the '_' and '-' flags can now be applied to
177 "%EY" to control how the year number is formatted; they have the
178 same effect that they would on "%Ey".
180 Deprecated and removed features, and other changes affecting compatibility:
182 * The glibc.tune tunable namespace has been renamed to glibc.cpu and the
183 tunable glibc.tune.cpu has been renamed to glibc.cpu.name.
185 * The type of the pr_uid and pr_gid members of struct elf_prpsinfo, defined
186 in <sys/procfs.h>, has been corrected to match the type actually used by
187 the Linux kernel. This affects the size and layout of that structure on
188 MicroBlaze, MIPS (n64 ABI only), Nios II and RISC-V.
190 * For the MIPS n32 ABI, the type of the pr_sigpend and pr_sighold members of
191 struct elf_prstatus, and the pr_flag member of struct elf_prpsinfo,
192 defined in <sys/procfs.h>, has been corrected to match the type actually
193 used by the Linux kernel. This affects the size and layout of those
196 * An archaic GNU extension to scanf, under which '%as', '%aS', and '%a[...]'
197 meant to scan a string and allocate space for it with malloc, is now
198 restricted to programs compiled in C89 or C++98 mode with _GNU_SOURCE
199 defined. This extension conflicts with C99's use of '%a' to scan a
200 hexadecimal floating-point number, which is now available to programs
201 compiled as C99 or C++11 or higher, regardless of _GNU_SOURCE.
203 POSIX.1-2008 includes the feature of allocating a buffer for string input
204 with malloc, using the modifier letter 'm' instead. Programs using
205 '%as', '%aS', or '%a[...]' with the old GNU meaning should change to
206 '%ms', '%mS', or '%m[...]' respectively. Programs that wish to use the
207 C99 '%a' no longer need to avoid _GNU_SOURCE.
209 GCC's -Wformat warnings can detect most uses of this extension, as long
210 as all functions that call vscanf, vfscanf, or vsscanf are annotated with
211 __attribute__ ((format (scanf, ...))).
213 Changes to build and runtime requirements:
215 * Python 3.4 or later is required to build the GNU C Library.
217 * On most architectures, GCC 5 or later is required to build the GNU C
218 Library. (On powerpc64le, GCC 6.2 or later is still required, as before.)
220 Older GCC versions and non-GNU compilers are still supported when
221 compiling programs that use the GNU C Library.
223 Security related changes:
225 CVE-2018-19591: A file descriptor leak in if_nametoindex can lead to a
226 denial of service due to resource exhaustion when processing getaddrinfo
227 calls with crafted host names. Reported by Guido Vranken.
229 CVE-2019-6488: On x32, the size_t parameter may be passed in the lower
230 32 bits of a 64-bit register with with non-zero upper 32 bit. When it
231 happened, accessing the 32-bit size_t value as the full 64-bit register
232 in the assembly string/memory functions would cause a buffer overflow.
235 CVE-2016-10739: The getaddrinfo function could successfully parse IPv4
236 addresses with arbitrary trailing characters, potentially leading to data
237 or command injection issues in applications.
239 The following bugs are resolved with this release:
241 [10425] localedata: it_IT/it_CH: LC_TIME format is wrong
242 [10496] localedata: 12h time representation in multiple locales faulty
243 [10797] localedata: it_IT locale numeric does not have a separator for
245 [11319] libc: dprintf doesn't handle errors properly
246 [16346] time: mktime: potentially unsafe use of localtime_offset
247 [17248] build: glibc should not sort CFLAGS (support gcc plugins and
249 [17405] libc: Implement posix_spawn_file_actions_addchdir_np,
250 posix_spawn_file_actions_addfchdir_np
251 [17426] localedata: Indian locales: set the correct date format
252 [17490] stdio: popen should not invoke atfork handlers
253 [17783] libc: TIOCSER_TEMT conditions inconsistent
254 [18040] regex: use-after-free in regexec/get_subexp
255 [18093] libc: Corrupted aux-cache causes ldconfig to segfault
256 [20018] network: getaddrinfo should reject IP addresses with trailing
257 characters (CVE-2016-10739)
258 [20209] localedata: Spelling mistake for Sunday in Greenlandic kl_GL
259 [20271] libc: Missing "\n" in __libc_fatal calls
260 [20480] dynamic-link: Patch: ifunc not executable, crashes sudo qemu
261 [20544] libc: RFE: atexit, __cxa_atexit, on_exit should assert function
262 pointer argument is non-NULL
263 [21037] stdio: open_memstream and freopen
264 [21286] libc: bits/siginfo.h is missing enum definition for TRAP_HWBKPT
265 [21716] time: Crash in glibc's mktime in low-memory situations
266 [22834] stdio: Subprocess forked by popen may crash in Linux when
267 multithreads call popen
268 [22927] network: crash in vn_gai_enqueue_request if requests_tail was NULL
269 and pthread_create fails.
270 [23032] hurd: sysdeps/htl/pt-barrier-init.c:39: bad call to memcmp ?
271 [23125] libc: riscv64: endless loop when throwing an exception from a
273 [23275] nptl: Race in pthread_mutex_lock while promoting to
274 PTHREAD_MUTEX_ELISION_NP.
275 [23400] libc: stdlib/test-bz22786.c creates temporary files in glibc
277 [23479] math: [mips] bits/fenv.h should not define some macros for soft-
279 [23490] libc: sysdeps/unix/sysv/linux/x86/tst-cet-property-2.c:49: off by
281 [23497] libc: readdir64@GLIBC_2.1 cannot parse the kernel directory stream
282 [23509] dynamic-link: CET enabled glibc is incompatible with the older
284 [23520] nscd: nscd: Use-after-free in addgetnetgrentX and its callers
285 [23521] nss: get_next_alias nss_files file stream leak
286 [23538] nptl: Hang in pthread_cond_broadcast
287 [23562] libc: Wrong type for si_band in Linux-specific siginfo_t
288 [23578] regex: Invalid memory access if regex pattern contains NUL byte
289 [23579] libc: Errors misreported in preadv2
290 [23597] build: support/test-container.c doesn't work with different
292 [23603] time: mktime signed integer overflow on large timestamps
293 [23606] libc: Missing ENDBR32 in sysdeps/i386/start.S
294 [23614] libc: powerpc: missing CFI register information in __mpn_*
296 [23637] string: Generic strstr/strcasestr fails with huge needles
297 [23640] libc: no way to easily clear FD_CLOEXEC in
298 posix_spawn_file_actions_adddup2()
299 [23649] libc: [microblaze/mips/nios2/riscv] sys/procfs.h pr_uid, pr_gid
301 [23656] libc: [mips n32] sys/procfs.h pr_sigpend, pr_sighold, pr_flag have
303 [23679] libc: gethostid: Missing NULL check for gethostbyname_r result
304 [23689] libc: Bug in documentation for rusage.ru_ixrss in
305 bits/types/struct_rusage.h
306 [23690] dynamic-link: Segfault in _dl_profile_fixup with a high number of
308 [23707] dynamic-link: Missing unwind info in sysdeps/powerpc/powerpc32/dl-
310 [23709] string: glibc 2.25 lacks sse2 optimized strstr()
311 [23716] dynamic-link: _dl_runtime_resolve_shstk isn't selected properly
312 [23717] libc: glibc: stdlib/tst-setcontext9 test suite failure on
314 [23724] localedata: Albanian date formats are incorrect
315 [23735] math: libnldbl_nonshared.a references internal libm symbols
316 [23740] localedata: kl_GL: Month names and date formats need update
317 [23744] regex: regex refactorings to remove BE, avoid duplication
318 [23745] time: mktime fix for Gnulib + coreutils
319 [23758] time: Improve the width of alternate representation for year in
321 [23783] libc: [mips] Missing CMSPAR bits/termios.h
322 [23789] time: mktime does not set errno on failure
323 [23791] localedata: Wrong monetary format for ca_ES locale
324 [23793] locale: c32rtomb and mbrtoc32 should not alias wcrtomb and mbrtowc
325 [23794] locale: c16rtomb does not handle surrogate pairs
326 [23821] libc: si_band in siginfo_t has wrong type long int on sparc64
327 [23822] math: ia64 static libm.a is missing exp2f, log2f and powf symbols
328 [23836] time: time/tst-mktime2 test failure on Arm (32-bit)
329 [23848] libc: [sparc] Some socket syscalls wrongly assumed to be present
330 [23861] nptl: rdlock stalls indefinitely on an unlocked pthread rwlock
331 [23862] libc: [sh] missing kernel-features.h undefines
332 [23864] libc: [riscv] missing kernel-features.h undefines
333 [23867] libc: [arm/microblaze] __ASSUME_MLOCK2 incorrect
334 [23907] malloc: Incorrect double-free malloc tcache check disregards
336 [23913] libc: off-by-one in function maybe_script_execute in
337 sysdeps/posix/spawni.c
338 [23915] libc: [arm] __ASSUME_COPY_FILE_RANGE incorrect
339 [23923] locale: Add --no-hard-links option to localedef
340 [23927] network: Linux if_nametoindex() does not close descriptor
342 [23961] math: powf can overflow to inf without setting errno in non-
343 nearest rounding mode
344 [23967] libc: [2.28 Regression]: New sigaction implementation breaks m68k
345 [23972] libc: __old_getdents64 uses wrong d_off value on overflow
346 [23993] libc: glibc 2.29 doesn't build with gcc 4.9
347 [23995] localedata: Remove execution flags from localedata/locales/bi_VU
348 [24011] localedata: Fixed small type in comment for locale bs_BA
349 [24018] libc: gettext() may return NULL
350 [24022] build: riscv build failure with Linux kernel 4.20-rc7
351 [24023] build: [2.29 Regression] FAIL: elf/check-localplt
352 [24024] string: strerror() might set errno to ENOMEM due to -fno-math-
354 [24027] malloc: glibc: realloc() ncopies 32-bit integer overflow
355 [24034] libc: tst-cancel21-static fails with SIGBUS on pre-ARMv7 when
357 [24046] localedata: en_US locale doesn't define date_fmt
358 [24063] manual: @var{errno} should be @code{errno}
359 [24066] soft-fp: Inconsistent _FP_W_TYPE_SIZE check
360 [24088] libc: VSCR field is not being correctly read in ucontext_t on
362 [24096] time: Specifying '_' or '-' flag for "%EY" does not produce the
364 [24097] string: Can't use 64-bit register for size_t in assembly codes for
366 [24110] hurd: SS_DISABLE never set in stack_t value returned by
368 [24112] network: Do not send DNS queries for non-host names (where all
369 answers will be rejected)
370 [24130] libc: alpha __remqu corrupts $f3 register
377 * The localization data for ISO 14651 is updated to match the 2016
378 Edition 4 release of the standard, this matches data provided by
379 Unicode 9.0.0. This update introduces significant improvements to the
380 collation of Unicode characters. This release deviates slightly from
381 the standard in that the collation element ordering for lowercase and
382 uppercase LATIN script characters is adjusted to ensure that regular
383 expressions with ranges like [a-z] and [A-Z] don't interleave e.g. A
384 is not matched by [a-z]. With the update many locales have been
385 updated to take advantage of the new collation information. The new
386 collation information has increased the size of the compiled locale
387 archive or binary locales.
389 * The GNU C Library can now be compiled with support for Intel CET, AKA
390 Intel Control-flow Enforcement Technology. When the library is built
391 with --enable-cet, the resulting glibc is protected with indirect
392 branch tracking (IBT) and shadow stack (SHSTK). CET-enabled glibc is
393 compatible with all existing executables and shared libraries. This
394 feature is currently supported on i386, x86_64 and x32 with GCC 8 and
395 binutils 2.29 or later. Note that CET-enabled glibc requires CPUs
396 capable of multi-byte NOPs, like x86-64 processors as well as Intel
397 Pentium Pro or newer. NOTE: --enable-cet has been tested for i686,
398 x86_64 and x32 on non-CET processors. --enable-cet has been tested
399 for x86_64 and x32 on CET SDVs, but Intel CET support hasn't been
402 * The GNU C Library now has correct support for ABSOLUTE symbols
403 (SHN_ABS-relative symbols). Previously such ABSOLUTE symbols were
404 relocated incorrectly or in some cases discarded. The GNU linker can
405 make use of the newer semantics, but it must communicate it to the
406 dynamic loader by setting the ELF file's identification (EI_ABIVERSION
407 field) to indicate such support is required.
409 * Unicode 11.0.0 Support: Character encoding, character type info, and
410 transliteration tables are all updated to Unicode 11.0.0, using
411 generator scripts contributed by Mike FABIAN (Red Hat).
413 * <math.h> functions that round their results to a narrower type are added
414 from TS 18661-1:2014 and TS 18661-3:2015:
416 - fadd, faddl, daddl and corresponding fMaddfN, fMaddfNx, fMxaddfN and
419 - fsub, fsubl, dsubl and corresponding fMsubfN, fMsubfNx, fMxsubfN and
422 - fmul, fmull, dmull and corresponding fMmulfN, fMmulfNx, fMxmulfN and
425 - fdiv, fdivl, ddivl and corresponding fMdivfN, fMdivfNx, fMxdivfN and
428 * Two grammatical forms of month names are now supported for the following
429 languages: Armenian, Asturian, Catalan, Czech, Kashubian, Occitan, Ossetian,
430 Scottish Gaelic, Upper Sorbian, and Walloon. The following languages now
431 support two grammatical forms in abbreviated month names: Catalan, Greek,
434 * Newly added locales: Lower Sorbian (dsb_DE) and Yakut (sah_RU) also
435 include the support for two grammatical forms of month names.
437 * Building and running on GNU/Hurd systems now works without out-of-tree
440 * The renameat2 function has been added, a variant of the renameat function
441 which has a flags argument. If the flags are zero, the renameat2 function
442 acts like renameat. If the flag is not zero and there is no kernel
443 support for renameat2, the function will fail with an errno value of
444 EINVAL. This is different from the existing gnulib function renameatu,
445 which performs a plain rename operation in case of a RENAME_NOREPLACE
446 flags and a non-existing destination (and therefore has a race condition
447 that can clobber the destination inadvertently).
449 * The statx function has been added, a variant of the fstatat64
450 function with an additional flags argument. If there is no direct
451 kernel support for statx, glibc provides basic stat support based on
452 the fstatat64 function.
454 * IDN domain names in getaddrinfo and getnameinfo now use the system libidn2
455 library if installed. libidn2 version 2.0.5 or later is recommended. If
456 libidn2 is not available, internationalized domain names are not encoded
457 or decoded even if the AI_IDN or NI_IDN flags are passed to getaddrinfo or
458 getnameinfo. (getaddrinfo calls with non-ASCII names and AI_IDN will fail
459 with an encoding error.) Flags which used to change the IDN encoding and
460 decoding behavior (AI_IDN_ALLOW_UNASSIGNED, AI_IDN_USE_STD3_ASCII_RULES,
461 NI_IDN_ALLOW_UNASSIGNED, NI_IDN_USE_STD3_ASCII_RULES) have been
462 deprecated. They no longer have any effect.
464 * Parsing of dynamic string tokens in DT_RPATH, DT_RUNPATH, DT_NEEDED,
465 DT_AUXILIARY, and DT_FILTER has been expanded to support the full
466 range of ELF gABI expressions including such constructs as
467 '$ORIGIN$ORIGIN' (if valid). For SUID/GUID applications the rules
468 have been further restricted, and where in the past a dynamic string
469 token sequence may have been interpreted as a literal string it will
470 now cause a load failure. These load failures were always considered
471 unspecified behaviour from the perspective of the dynamic loader, and
472 for safety are now load errors e.g. /foo/${ORIGIN}.so in DT_NEEDED
473 results in a load failure now.
475 * Support for ISO C threads (ISO/IEC 9899:2011) has been added. The
476 implementation includes all the standard functions provided by
479 - thrd_current, thrd_equal, thrd_sleep, thrd_yield, thrd_create,
480 thrd_detach, thrd_exit, and thrd_join for thread management.
482 - mtx_init, mtx_lock, mtx_timedlock, mtx_trylock, mtx_unlock, and
483 mtx_destroy for mutual exclusion.
485 - call_once for function call synchronization.
487 - cnd_broadcast, cnd_destroy, cnd_init, cnd_signal, cnd_timedwait, and
488 cnd_wait for conditional variables.
490 - tss_create, tss_delete, tss_get, and tss_set for thread-local storage.
492 Application developers must link against libpthread to use ISO C threads.
494 Deprecated and removed features, and other changes affecting compatibility:
496 * The nonstandard header files <libio.h> and <_G_config.h> are no longer
497 installed. Software that was using either header should be updated to
498 use standard <stdio.h> interfaces instead.
500 * The stdio functions 'getc' and 'putc' are no longer defined as macros.
501 This was never required by the C standard, and the macros just expanded
502 to call alternative names for the same functions. If you hoped getc and
503 putc would provide performance improvements over fgetc and fputc, instead
504 investigate using (f)getc_unlocked and (f)putc_unlocked, and, if
505 necessary, flockfile and funlockfile.
507 * All stdio functions now treat end-of-file as a sticky condition. If you
508 read from a file until EOF, and then the file is enlarged by another
509 process, you must call clearerr or another function with the same effect
510 (e.g. fseek, rewind) before you can read the additional data. This
511 corrects a longstanding C99 conformance bug. It is most likely to affect
512 programs that use stdio to read interactive input from a terminal.
515 * The macros 'major', 'minor', and 'makedev' are now only available from
516 the header <sys/sysmacros.h>; not from <sys/types.h> or various other
517 headers that happen to include <sys/types.h>. These macros are rarely
518 used, not part of POSIX nor XSI, and their names frequently collide with
519 user code; see https://sourceware.org/bugzilla/show_bug.cgi?id=19239 for
522 <sys/sysmacros.h> is a GNU extension. Portable programs that require
523 these macros should first include <sys/types.h>, and then include
524 <sys/sysmacros.h> if __GNU_LIBRARY__ is defined.
526 * The tilegx*-*-linux-gnu configurations are no longer supported.
528 * The obsolete function ustat is no longer available to newly linked
529 binaries; the headers <ustat.h> and <sys/ustat.h> have been removed. This
530 function has been deprecated in favor of fstatfs and statfs.
532 * The obsolete function nfsservctl is no longer available to newly linked
533 binaries. This function was specific to systems using the Linux kernel
534 and could not usefully be used with the GNU C Library on systems with
535 version 3.1 or later of the Linux kernel.
537 * The obsolete function name llseek is no longer available to newly linked
538 binaries. This function was specific to systems using the Linux kernel
539 and was not declared in a header. Programs should use the lseek64 name
540 for this function instead.
542 * The AI_IDN_ALLOW_UNASSIGNED and NI_IDN_ALLOW_UNASSIGNED flags for the
543 getaddrinfo and getnameinfo functions have been deprecated. The behavior
544 previously selected by them is now always enabled.
546 * The AI_IDN_USE_STD3_ASCII_RULES and NI_IDN_USE_STD3_ASCII_RULES flags for
547 the getaddrinfo and getnameinfo functions have been deprecated. The STD3
548 restriction (rejecting '_' in host names, among other things) has been
549 removed, for increased compatibility with non-IDN name resolution.
551 * The fcntl function now have a Long File Support variant named fcntl64. It
552 is added to fix some Linux Open File Description (OFD) locks usage on non
553 LFS mode. As for others *64 functions, fcntl64 semantics are analogous with
554 fcntl and LFS support is handled transparently. Also for Linux, the OFD
555 locks act as a cancellation entrypoint.
557 * The obsolete functions encrypt, encrypt_r, setkey, setkey_r, cbc_crypt,
558 ecb_crypt, and des_setparity are no longer available to newly linked
559 binaries, and the headers <rpc/des_crypt.h> and <rpc/rpc_des.h> are no
560 longer installed. These functions encrypted and decrypted data with the
561 DES block cipher, which is no longer considered secure. Software that
562 still uses these functions should switch to a modern cryptography library,
565 * Reflecting the removal of the encrypt and setkey functions above, the
566 macro _XOPEN_CRYPT is no longer defined. As a consequence, the crypt
567 function is no longer declared unless _DEFAULT_SOURCE or _GNU_SOURCE is
570 * The obsolete function fcrypt is no longer available to newly linked
571 binaries. It was just another name for the standard function crypt,
572 and it has not appeared in any header file in many years.
574 * We have tentative plans to hand off maintenance of the passphrase-hashing
575 library, libcrypt, to a separate development project that will, we hope,
576 keep up better with new passphrase-hashing algorithms. We will continue
577 to declare 'crypt' in <unistd.h>, and programs that use 'crypt' or
578 'crypt_r' should not need to change at all; however, distributions will
579 need to install <crypt.h> and libcrypt from a separate project.
581 In this release, if the configure option --disable-crypt is used, glibc
582 will not install <crypt.h> or libcrypt, making room for the separate
583 project's versions of these files. The plan is to make this the default
584 behavior in a future release.
586 Changes to build and runtime requirements:
588 GNU make 4.0 or later is now required to build glibc.
590 Security related changes:
592 CVE-2016-6261, CVE-2016-6263, CVE-2017-14062: Various vulnerabilities have
593 been fixed by removing the glibc-internal IDNA implementation and using
594 the system-provided libidn2 library instead. Originally reported by Hanno
595 Böck and Christian Weisgerber.
597 CVE-2017-18269: An SSE2-based memmove implementation for the i386
598 architecture could corrupt memory. Reported by Max Horn.
600 CVE-2018-11236: Very long pathname arguments to realpath function could
601 result in an integer overflow and buffer overflow. Reported by Alexey
604 CVE-2018-11237: The mempcpy implementation for the Intel Xeon Phi
605 architecture could write beyond the target buffer, resulting in a buffer
606 overflow. Reported by Andreas Schwab.
608 The following bugs are resolved with this release:
610 [1190] stdio: fgetc()/fread() behaviour is not POSIX compliant
611 [6889] manual: 'PWD' mentioned but not specified
612 [13575] libc: SSIZE_MAX defined as LONG_MAX is inconsistent with ssize_t,
613 when __WORDSIZE != 64
614 [13762] regex: re_search etc. should return -2 on memory exhaustion
615 [13888] build: /tmp usage during testing
616 [13932] math: dbl-64 pow unexpectedly slow for some inputs
617 [14092] nptl: Support C11 threads
618 [14095] localedata: Review / update collation data from Unicode / ISO
620 [14508] libc: -Wformat warnings
621 [14553] libc: Namespace pollution loff_t in sys/types.h
622 [14890] libc: Make NT_PRFPREG canonical.
623 [15105] libc: Extra PLT references with -Os
624 [15512] libc: __bswap_constant_16 not compiled when -Werror -Wsign-
626 [16335] manual: Feature test macro documentation incomplete and out of
628 [16552] libc: Unify umount implementations in terms of umount2
629 [17082] libc: htons et al.: statement-expressions prevent use on global
630 scope with -O1 and higher
631 [17343] libc: Signed integer overflow in /stdlib/random_r.c
632 [17438] localedata: pt_BR: wrong d_fmt delimiter
633 [17662] libc: please implement binding for the new renameat2 syscall
634 [17721] libc: __restrict defined as /* Ignore */ even in c11
635 [17979] libc: inconsistency between uchar.h and stdint.h
636 [18018] dynamic-link: Additional $ORIGIN handling issues (CVE-2011-0536)
637 [18023] libc: extend_alloca is broken (questionable pointer comparison,
638 horrible machine code)
639 [18124] libc: hppa: setcontext erroneously returns -1 as exit code for
641 [18471] libc: llseek should be a compat symbol
642 [18473] soft-fp: [powerpc-nofpu] __sqrtsf2, __sqrtdf2 should be compat
644 [18991] nss: nss_files skips large entry in database
645 [19239] libc: Including stdlib.h ends up with macros major and minor being
647 [19463] libc: linknamespace failures when compiled with -Os
648 [19485] localedata: csb_PL: Update month translations + add yesstr/nostr
649 [19527] locale: Normalized charset name not recognized by setlocale
650 [19667] string: Missing Sanity Check for malloc calls in file 'testcopy.c'
651 [19668] libc: Missing Sanity Check for malloc() in file 'tst-setcontext-
653 [19728] network: out of bounds stack read in libidn function
654 idna_to_ascii_4i (CVE-2016-6261)
655 [19729] network: out of bounds heap read on invalid utf-8 inputs in
656 stringprep_utf8_nfkc_normalize (CVE-2016-6263)
657 [19818] dynamic-link: Absolute (SHN_ABS) symbols incorrectly relocated by
659 [20079] libc: Add SHT_X86_64_UNWIND to elf.h
660 [20251] libc: 32bit programs pass garbage in struct flock for OFD locks
661 [20419] dynamic-link: files with large allocated notes crash in
663 [20530] libc: bswap_16 should use __builtin_bswap16() when available
664 [20890] dynamic-link: ldconfig: fsync the files before atomic rename
665 [20980] manual: CFLAGS environment variable replaces vital options
666 [21163] regex: Assertion failure in pop_fail_stack when executing a
667 malformed regexp (CVE-2015-8985)
668 [21234] manual: use of CFLAGS makes glibc detect no optimization
669 [21269] dynamic-link: i386 sigaction sa_restorer handling is wrong
670 [21313] build: Compile Error GCC 5.4.0 MIPS with -0S
671 [21314] build: Compile Error GCC 5.2.0 MIPS with -0s
672 [21508] locale: intl/tst-gettext failure with latest msgfmt
673 [21547] localedata: Tibetan script collation broken (Dzongkha and Tibetan)
674 [21812] network: getifaddrs() returns entries with ifa_name == NULL
675 [21895] libc: ppc64 setjmp/longjmp not fully interoperable with static
677 [21942] dynamic-link: _dl_dst_substitute incorrectly handles $ORIGIN: with
679 [22241] localedata: New locale: Yakut (Sakha) locale for Russia (sah_RU)
680 [22247] network: Integer overflow in the decode_digit function in
681 puny_decode.c in libidn (CVE-2017-14062)
682 [22342] nscd: NSCD not properly caching netgroup
683 [22391] nptl: Signal function clear NPTL internal symbols inconsistently
684 [22550] localedata: es_ES locale (and other es_* locales): collation
685 should treat ñ as a primary different character, sync the collation
686 for Spanish with CLDR
687 [22638] dynamic-link: sparc: static binaries are broken if glibc is built
688 by gcc configured with --enable-default-pie
689 [22639] time: year 2039 bug for localtime etc. on 64-bit platforms
690 [22644] string: memmove-sse2-unaligned on 32bit x86 produces garbage when
691 crossing 2GB threshold (CVE-2017-18269)
692 [22646] localedata: redundant data (LC_TIME) for es_CL, es_CU, es_EC and
694 [22735] time: Misleading typo in time.h source comment regarding
696 [22753] libc: preadv2/pwritev2 fallback code should handle offset=-1
697 [22761] libc: No trailing `%n' conversion specifier in FMT passed from
698 `__assert_perror_fail ()' to `__assert_fail_base ()'
699 [22766] libc: all glibc internal dlopen should use RTLD_NOW for robust
701 [22786] libc: Stack buffer overflow in realpath() if input size is close
702 to SSIZE_MAX (CVE-2018-11236)
703 [22787] dynamic-link: _dl_check_caller returns false when libc is linked
704 through an absolute DT_NEEDED path
705 [22792] build: tcb-offsets.h dependency dropped
706 [22797] libc: pkey_get() uses non-reserved name of argument
707 [22807] libc: PTRACE_* constants missing for powerpc
708 [22818] glob: posix/tst-glob_lstat_compat failure on alpha
709 [22827] dynamic-link: RISC-V ELF64 parser mis-reads flag in ldconfig
710 [22830] malloc: malloc_stats doesn't restore cancellation state on stderr
711 [22848] localedata: ca_ES: update date definitions from CLDR
712 [22862] build: _DEFAULT_SOURCE is defined even when _ISOC11_SOURCE is
713 [22884] math: RISCV fmax/fmin handle signalling NANs incorrectly
714 [22896] localedata: Update locale data for an_ES
715 [22902] math: float128 test failures with GCC 8
716 [22918] libc: multiple common of `__nss_shadow_database'
717 [22919] libc: sparc32: backtrace yields infinite backtrace with
719 [22926] libc: FTBFS on powerpcspe
720 [22932] localedata: lt_LT: Update of abbreviated month names from CLDR
722 [22937] localedata: Greek (el_GR, el_CY) locales actually need ab_alt_mon
723 [22947] libc: FAIL: misc/tst-preadvwritev2
724 [22963] localedata: cs_CZ: Add alternative month names
725 [22987] math: [powerpc/sparc] fdim inlines errno, exceptions handling
726 [22996] localedata: change LC_PAPER to en_US in es_BO locale
727 [22998] dynamic-link: execstack tests are disabled when SELinux is
729 [23005] network: Crash in __res_context_send after memory allocation
731 [23007] math: strtod cannot handle -nan
732 [23024] nss: getlogin_r is performing NSS lookups when loginid isn't set
733 [23036] regex: regex equivalence class regression
734 [23037] libc: initialize msg_flags to zero for sendmmsg() calls
735 [23069] libc: sigaction broken on riscv64-linux-gnu
736 [23094] localedata: hr_HR: wrong thousands_sep and mon_thousands_sep
737 [23102] dynamic-link: Incorrect parsing of multiple consecutive $variable
738 patterns in runpath entries (e.g. $ORIGIN$ORIGIN)
739 [23137] nptl: s390: pthread_join sometimes block indefinitely (on 31bit
740 and libc build with -Os)
741 [23140] localedata: More languages need two forms of month names
742 [23145] libc: _init/_fini aren't marked as hidden
743 [23152] localedata: gd_GB: Fix typo in "May" (abbreviated)
744 [23171] math: C++ iseqsig for long double converts arguments to double
745 [23178] nscd: sudo will fail when it is run in concurrent with commands
746 that changes /etc/passwd
747 [23196] string: __mempcpy_avx512_no_vzeroupper mishandles large copies
749 [23206] dynamic-link: static-pie + dlopen breaks debugger interaction
750 [23208] localedata: New locale - Lower Sorbian (dsb)
751 [23233] regex: Memory leak in build_charclass_op function in file
753 [23236] stdio: Harden function pointers in _IO_str_fields
754 [23250] nptl: Offset of __private_ss differs from GCC
755 [23253] math: tgamma test suite failures on i686 with -march=x86-64
756 -mtune=generic -mfpmath=sse
757 [23259] dynamic-link: Unsubstituted ${ORIGIN} remains in DT_NEEDED for
759 [23264] libc: posix_spawnp wrongly executes ENOEXEC in non compat mode
760 [23266] nis: stringop-truncation warning with new gcc8.1 in nisplus-
762 [23272] math: fma(INFINITY,INFIITY,0.0) should be INFINITY
763 [23277] math: nan function should not have const attribute
764 [23279] math: scanf and strtod wrong for some hex floating-point
765 [23280] math: wscanf rounds wrong; wcstod is ok for negative numbers and
767 [23290] localedata: IBM273 is not equivalent to ISO-8859-1
768 [23303] build: undefined reference to symbol
769 '__parse_hwcap_and_convert_at_platform@@GLIBC_2.23'
770 [23307] dynamic-link: Absolute symbols whose value is zero ignored in
772 [23313] stdio: libio vtables validation and standard file object
774 [23329] libc: The __libc_freeres infrastructure is not properly run across
776 [23349] libc: Various glibc headers no longer compatible with
778 [23351] malloc: Remove unused code related to heap dumps and malloc
780 [23363] stdio: stdio-common/tst-printf.c has non-free license
781 [23396] regex: Regex equivalence regression in single-byte locales
782 [23422] localedata: oc_FR: More updates of locale data
783 [23442] build: New warning with GCC 8
784 [23448] libc: Out of bounds access in IBM-1390 converter
785 [23456] libc: Wrong index_cpu_LZCNT
786 [23458] build: tst-get-cpu-features-static isn't added to tests
787 [23459] libc: COMMON_CPUID_INDEX_80000001 isn't populated for Intel
789 [23467] dynamic-link: x86/CET: A property note parser bug
796 * The GNU C Library can now be compiled with support for building static
797 PIE executables (See --enable-static-pie in INSTALL). These static PIE
798 executables are like static executables but can be loaded at any address
799 and provide additional security hardening benefits at the cost of some
800 memory and performance. When the library is built with --enable-static-pie
801 the resulting libc.a is usable with GCC 8 and above to create static PIE
802 executables using the GCC option '-static-pie'. This feature is currently
803 supported on i386, x86_64 and x32 with binutils 2.29 or later, and on
804 aarch64 with binutils 2.30 or later.
806 * Optimized x86-64 asin, atan2, exp, expf, log, pow, atan, sin, cosf,
807 sinf, sincosf and tan with FMA, contributed by Arjan van de Ven and
810 * Optimized x86-64 trunc and truncf for processors with SSE4.1.
812 * Optimized generic expf, exp2f, logf, log2f, powf, sinf, cosf and sincosf.
814 * In order to support faster and safer process termination the malloc API
815 family of functions will no longer print a failure address and stack
816 backtrace after detecting heap corruption. The goal is to minimize the
817 amount of work done after corruption is detected and to avoid potential
818 security issues in continued process execution. Reducing shutdown time
819 leads to lower overall process restart latency, so there is benefit both
820 from a security and performance perspective.
822 * The abort function terminates the process immediately, without flushing
823 stdio streams. Previous glibc versions used to flush streams, resulting
824 in deadlocks and further data corruption. This change also affects
825 process aborts as the result of assertion failures.
827 * On platforms where long double has the IEEE binary128 format (aarch64,
828 alpha, mips64, riscv, s390 and sparc), the math library now implements
829 _Float128 interfaces for that type, as defined by ISO/IEC TS 18661-3:2015.
830 These are the same interfaces added in version 2.26 for some platforms where
831 this format is supported but is not the format of long double.
833 * On platforms with support for _Float64x (aarch64, alpha, i386, ia64,
834 mips64, powerpc64le, riscv, s390, sparc and x86_64), the math library now
835 implements interfaces for that type, as defined by ISO/IEC TS
836 18661-3:2015. These are corresponding interfaces to those supported for
839 * The math library now implements interfaces for the _Float32, _Float64 and
840 _Float32x types, as defined by ISO/IEC TS 18661-3:2015. These are
841 corresponding interfaces to those supported for _Float128.
843 * glibc now implements the memfd_create and mlock2 functions on Linux.
845 * Support for memory protection keys was added. The <sys/mman.h> header now
846 declares the functions pkey_alloc, pkey_free, pkey_mprotect, pkey_set,
849 * The copy_file_range function was added.
851 * Optimized memcpy, mempcpy, memmove, and memset for sparc M7.
853 * The ldconfig utility now processes `include' directives using the C/POSIX
854 collation ordering. Previous glibc versions used locale-specific
855 ordering, the change might break systems that relied on that.
857 * Support for two grammatical forms of month names has been added.
858 In a call to strftime, the "%B" and "%b" format specifiers will now
859 produce the grammatical form required when the month is used as part
860 of a complete date. New "%OB" and "%Ob" specifiers produce the form
861 required when the month is named by itself. For instance, in Greek
862 and in many Slavic and Baltic languages, "%B" will produce the month
863 in genitive case, and "%OB" will produce the month in nominative case.
865 In a call to strptime, "%B", "%b", "%h", "%OB", "%Ob", and "%Oh"
866 are all valid and will all accept any known form of month
867 name---standalone or complete, abbreviated or full. In a call to
868 nl_langinfo, the query constants MON_1..12 and ABMON_1..12 return
869 the strings used by "%B" and "%b", respectively. New query
870 constants ALTMON_1..12 and _NL_ABALTMON_1..12 return the strings
871 used by "%OB" and "%Ob", respectively.
873 In a locale definition file, use "alt_mon" and "ab_alt_mon" to
874 define the strings for %OB and %Ob, respectively; these have the
875 same syntax as "mon" and "abmon". These arrays are optional; if they
876 are not provided then they have the same content as "mon" and "abmon",
879 These features are provided for locales which define "alt_mon" and/or
880 "ab_alt_mon" in their locale source data. This release includes such
881 alternative month name data for the following languages: Belarusian,
882 Croatian, Greek, Lithuanian, Polish, Russian, and Ukrainian.
884 This feature is currently a GNU extension, but it is expected to
885 be added to the next revision of POSIX, and it is also already
886 available on some BSD-derived operating systems.
888 This feature will cause existing statically compiled applications
889 to fail to load locales and fall back to the builtin C/POSIX locales.
890 See notes below for other changes affecting compatibility.
892 * Support for the RISC-V ISA running on Linux has been added. This port
893 requires at least binutils-2.30, gcc-7.3.0, and linux-4.15; and is supported
894 for the following ISA and ABI pairs:
900 Deprecated and removed features, and other changes affecting compatibility:
902 * Statically compiled applications attempting to load locales compiled for the
903 GNU C Library version 2.27 will fail and fall back to the builtin C/POSIX
904 locale. The reason for this is that the addition of the new "%OB" and "%Ob",
905 support for two grammatical forms of the month names, also extends the locale
906 data binary format. Static applications needing locale support must be
907 recompiled to match the runtime and data they are deployed with. In some
908 distributions there is an upgrade window where dynamically linked applications
909 may use a new library but the old locale data and also fall back to the
910 builtin C/POSIX locales; restarting the application process is sufficient to
913 * Support for statically linked applications which call dlopen is deprecated
914 and will be removed in a future version of glibc. Applications which call
915 dlopen need to be linked dynamically instead.
917 * Support for old programs which use internal stdio data structures and
918 functions is deprecated. This includes programs which use the C++ streams
919 provided by libstdc++ in GCC 2.95. Programs which use the internal
920 symbols _IO_adjust_wcolumn, _IO_default_doallocate, _IO_default_finish,
921 _IO_default_pbackfail, _IO_default_uflow, _IO_default_xsgetn,
922 _IO_default_xsputn, _IO_doallocbuf, _IO_do_write, _IO_file_attach,
923 _IO_file_close, _IO_file_close_it, _IO_file_doallocate, _IO_file_fopen,
924 _IO_file_init, _IO_file_jumps, _IO_fileno, _IO_file_open,
925 _IO_file_overflow, _IO_file_read, _IO_file_seek, _IO_file_seekoff,
926 _IO_file_setbuf, _IO_file_stat, _IO_file_sync, _IO_file_underflow,
927 _IO_file_write, _IO_file_xsputn, _IO_flockfile, _IO_flush_all,
928 _IO_flush_all_linebuffered, _IO_free_backup_area, _IO_free_wbackup_area,
929 _IO_init, _IO_init_marker, _IO_init_wmarker, _IO_iter_begin, _IO_iter_end,
930 _IO_iter_file, _IO_iter_next, _IO_least_wmarker, _IO_link_in,
931 _IO_list_all, _IO_list_lock, _IO_list_resetlock, _IO_list_unlock,
932 _IO_marker_delta, _IO_marker_difference, _IO_remove_marker, _IO_seekmark,
933 _IO_seekwmark, _IO_str_init_readonly, _IO_str_init_static,
934 _IO_str_overflow, _IO_str_pbackfail, _IO_str_seekoff, _IO_str_underflow,
935 _IO_switch_to_main_wget_area, _IO_switch_to_wget_mode,
936 _IO_unsave_wmarkers, _IO_wdefault_doallocate, _IO_wdefault_finish,
937 _IO_wdefault_pbackfail, _IO_wdefault_setbuf, _IO_wdefault_uflow,
938 _IO_wdefault_xsgetn, _IO_wdefault_xsputn, _IO_wdoallocbuf, _IO_wdo_write,
939 _IO_wfile_jumps, _IO_wfile_overflow, _IO_wfile_sync, _IO_wfile_underflow,
940 _IO_wfile_xsputn, _IO_wmarker_delta, or _IO_wsetb may stop working with a
941 future version of glibc. Unlike other symbol removals, these old
942 applications will not be supported using compatibility symbols.
944 * On GNU/Linux, the obsolete Linux constant PTRACE_SEIZE_DEVEL is no longer
945 defined by <sys/ptrace.h>.
947 * libm no longer supports SVID error handling (calling a user-provided
948 matherr function on error) or the _LIB_VERSION variable to control error
949 handling. (SVID error handling and the _LIB_VERSION variable still work
950 for binaries linked against older versions of the GNU C Library.) The
951 libieee.a library is no longer provided. math.h no longer defines struct
952 exception, or the macros X_TLOSS, DOMAIN, SING, OVERFLOW, UNDERFLOW,
953 TLOSS, PLOSS and HUGE.
955 * The libm functions pow10, pow10f and pow10l are no longer supported for
956 new programs. Programs should use the standard names exp10, exp10f and
957 exp10l for these functions instead.
959 * The mcontext_t type is no longer the same as struct sigcontext. On
960 platforms where it was previously the same, this changes the C++ name
961 mangling for interfaces involving this type.
963 * The add-ons mechanism for building additional packages at the same time as
964 glibc has been removed. The --enable-add-ons configure option is now
967 * The --without-fp configure option is now ignored. Whether hardware
968 floating-point instructions are used is now configured based on whether
969 the compiler used at configure time (without any options implied by a
970 --with-cpu= configure option) uses such instructions.
972 * The res_hnok, res_dnok, res_mailok and res_ownok functions now check that
973 the specified string can be parsed as a domain name.
975 * In the malloc_info output, the <heap> element may contain another <aspace>
976 element, "subheaps", which contains the number of sub-heaps.
978 * The libresolv function p_secstodate is no longer supported for new
981 * The tilepro-*-linux-gnu configuration is no longer supported.
983 * The nonstandard header files <libio.h> and <_G_config.h> are deprecated
984 and will be removed in a future release. Software that is still using
985 either header should be updated to use standard <stdio.h> interfaces
988 libio.h was originally the header for a set of supported GNU extensions,
989 but they have not been maintained as such in many years, they are now
990 standing in the way of improvements to stdio, and we don't think there are
991 any remaining external users. _G_config.h was never intended for public
992 use, but predates the bits convention.
994 Changes to build and runtime requirements:
996 * bison version 2.7 or later is required to generate code in the 'intl'
999 Security related changes:
1001 CVE-2009-5064: The ldd script would sometimes run the program under
1002 examination directly, without preventing code execution through the
1003 dynamic linker. (The glibc project disputes that this is a security
1004 vulnerability; only trusted binaries must be examined using the ldd
1007 CVE-2017-15670: The glob function, when invoked with GLOB_TILDE,
1008 suffered from a one-byte overflow during ~ operator processing (either
1009 on the stack or the heap, depending on the length of the user name).
1010 Reported by Tim Rühsen.
1012 CVE-2017-15671: The glob function, when invoked with GLOB_TILDE,
1013 would sometimes fail to free memory allocated during ~ operator
1014 processing, leading to a memory leak and, potentially, to a denial
1017 CVE-2017-15804: The glob function, when invoked with GLOB_TILDE and
1018 without GLOB_NOESCAPE, could write past the end of a buffer while
1019 unescaping user names. Reported by Tim Rühsen.
1021 CVE-2017-17426: The malloc function, when called with an object size near
1022 the value SIZE_MAX, would return a pointer to a buffer which is too small,
1023 instead of NULL. This was a regression introduced with the new malloc
1024 thread cache in glibc 2.26. Reported by Iain Buclaw.
1026 CVE-2017-1000408: Incorrect array size computation in _dl_init_paths leads
1027 to the allocation of too much memory. (This is not a security bug per se,
1028 it is mentioned here only because of the CVE assignment.) Reported by
1031 CVE-2017-1000409: Buffer overflow in _dl_init_paths due to miscomputation
1032 of the number of search path components. (This is not a security
1033 vulnerability per se because no trust boundary is crossed if the fix for
1034 CVE-2017-1000366 has been applied, but it is mentioned here only because
1035 of the CVE assignment.) Reported by Qualys.
1037 CVE-2017-16997: Incorrect handling of RPATH or RUNPATH containing $ORIGIN
1038 for AT_SECURE or SUID binaries could be used to load libraries from the
1041 CVE-2018-1000001: Buffer underflow in realpath function when getcwd function
1042 succeeds without returning an absolute path due to unexpected behaviour
1043 of the Linux kernel getcwd syscall. Reported by halfdog.
1045 CVE-2018-6485: The posix_memalign and memalign functions, when called with
1046 an object size near the value of SIZE_MAX, would return a pointer to a
1047 buffer which is too small, instead of NULL. Reported by Jakub Wilk.
1049 CVE-2018-6551: The malloc function, when called with an object size near
1050 the value of SIZE_MAX, would return a pointer to a buffer which is too
1051 small, instead of NULL.
1053 The following bugs are resolved with this release:
1055 [866] glob: glob should match dangling symlinks
1056 [1062] glob: sysdeps/generic/glob.c merge from gnulib (part 3 of 3)
1057 [2522] localedata: ca_ES@valencia: new Valencian (meridional Catalan)
1059 [5997] math: Very slow execution of sinf function
1060 [10580] localedata: hr_HR: updated locale
1061 [10871] locale: 'mon' array should contain both nominative and genitive
1063 [12349] localedata: eu_ES: incorrect thousands separator
1064 [13605] localedata: shn_MM: new Shan locale
1065 [13805] localedata: ru_RU: currency should use ',' as radix point
1066 [13953] localedata: km_KH: locale update
1067 [13994] localedata: mjw_IN: new locale
1068 [14121] build: make writes .mo files in po directory
1069 [14333] libc: Fix the race between atexit() and exit()
1070 [14681] dynamic-link: _dl_get_origin leaks memory via executable link map.
1071 [14925] localedata: bn_*: LC_IDENTIFICATION.language key should be
1073 [15260] localedata: LC_MESSAGES.{yes,no}{str,expr}: various errors
1074 [15261] localedata: LC_MESSAGES.yesexpr/noexpr: inconsistent use of full-
1075 width Latin characters
1076 [15332] localedata: es_CU: locale update
1077 [15436] stdio: Don't close or flush stdio streams on abort
1078 [15537] localedata: lv_LV: invalid collation for Latvian diacritical
1080 [16148] localedata: ca_ES: incorrect thousands separator
1081 [16750] dynamic-link: ldd should not try to execute the binaries
1083 [16777] localedata: pl_PL: incorrect thousands separator in locale
1084 [16905] localedata: hanzi: new collation
1085 [17563] localedata: cmn_TW: add hanzi collation
1086 [17750] localedata: wrong collation order of diacritics in most locales
1087 [17804] libc: scandirat fails with ENOMEM because it checks for errno even
1089 [17956] build: Build fails on missing definitions from header file
1090 nss/nss.h when Mozilla NSS is used for cryptography
1091 [18203] libc: realpath() does not handle unreachable paths correctly
1092 [18572] dynamic-link: [arm] Lazy TLSDESC relocation has data race
1093 [18812] localedata: kab_DZ: new Kabyle Algeria locale
1094 [18822] libc: Internal functions are called via PLT
1095 [18858] string: _HAVE_STRING_ARCH_xxx aren't defined for i386 nor x86_64
1096 [19170] libc: __gmon_start__ defined in hppa in crtn.S
1097 [19574] libc: glibc should support building static PIE binaries
1098 [19852] localedata: charmaps/UTF-8: incorrect wcwidth for U+3099 and
1100 [19971] glob: glob: Do not skip entries with zero d_ino values
1101 [19982] localedata: fr.po: spelling mistake for error code EXDEV
1102 [20008] localedata: km_KH: convert to translit_neutral
1103 [20009] localedata: tr_TR: convert LC_CTYPE to i18n
1104 [20142] math: [x86_64] Add SSE4.1 trunc, truncf
1105 [20204] dynamic-link: _dl_open_hook and _dlfcn_hook hardening
1106 [20482] localedata: de_CH: abbreviated weekdays should be two letters
1107 [20498] localedata: miq_NI: new Mískitu / Miskito (miq) language locale
1109 [20532] nss: getaddrinfo uses errno and h_errno without guaranteeing
1110 they're set, wrong errors returned by gaih_inet when lookup functions
1112 [20756] localedata: [PATCH] Use Unicode wise thousands separator
1113 [20826] network: posix/tst-getaddrinfo5 fails on hosts without network
1115 [20952] localedata: yuw_PG: new locale
1116 [21084] localedata: charmaps/IBM858: new codepage
1117 [21161] manual: [PATCH] fix typo in manual/arith.texi on strtoul prototype
1118 [21242] libc: assert gives pedantic warning in old gcc versions
1119 [21265] dynamic-link: _dl_runtime_resolve isn't compatible with Intel C++
1120 __regcall calling convention
1121 [21309] math: signed integer overflow in sysdeps/ieee754/dbl-64/e_pow.c
1122 [21326] libc: C99 functions are not declared for C++11 and later if
1123 _GNU_SOURCE is not predefined by g++
1124 [21457] libc: sys/ucontext.h namespace
1125 [21530] libc: tmpfile() should be implemented using O_TMPFILE
1126 [21660] math: GCC fails to compile a formula with tgmath.h
1127 [21672] nptl: sys-libs/glibc on ia64 crashes on thread exit: signal
1128 SIGSEGV, Segmentation fault: pthread_create.c:432: __madvise
1129 (pd->stackblock, freesize - PTHREAD_STACK_MIN, MADV_DONTNEED);
1130 [21684] math: tgmath.h handling of complex integers
1131 [21685] math: tgmath.h handling of bit-fields
1132 [21686] math: tgmath.h handling of __int128
1133 [21706] localedata: yesstr and nostr are missing for Breton [LC_MESSAGES]
1135 [21745] libc: [powerpc64le] Extra PLT reference with --enable-stack-
1137 [21750] localedata: column width of characters incompatible with classical
1139 [21754] malloc: malloc: Perform as little work as possible after heap
1140 consistency check failures
1141 [21780] libc: hppa: p{read,write}v2 does not set ENOSUP on invalid flag
1142 [21790] libc: Missing __memset_zero_constant_len_parameter in libc.so
1143 [21791] string: Unused XXX_chk_XXX functions in libc.a
1144 [21815] dynamic-link: FAIL: elf/tst-prelink-cmp with GCC is defaulted to
1146 [21836] localedata: Removed redundant data (LC_MONETARY) in various Indian
1148 [21845] localedata: Added new Locale bho_NP
1149 [21853] localedata: Fix abday Which looks same as day in zh_SG
1150 [21854] localedata: Added New Locale en_SC
1151 [21864] libc: xmalloc.o is compiled with -DMODULE_NAME=libc
1152 [21871] dynamic-link: _dl_runtime_resolve_avx_opt is slower than
1153 _dl_runtime_resolve_avx_slow
1154 [21885] network: getaddrinfo: gethosts does not release resolver context
1155 on memory allocation failure
1156 [21899] libc: XPG4.2 sigaction namespace
1157 [21908] dynamic-link: dynamic linker broke on ia64 (mmap2 consolidation is
1159 [21913] libc: static binaries SIGSEGV in __brk when host's gcc is pie-by-
1161 [21915] nss: nss_files can return with NSS_STATUS_SUCCESS and a clobbered
1162 errno value, causing getaddrinfo to fail
1163 [21920] localedata: Fix p_cs_precedes/n_cs_precedes for mt_MT
1164 [21922] network: getaddrinfo with AF_INET/AF_INET6 returns EAI_NONAME
1165 instead of EAI_NODATA
1166 [21928] libc: sys/ptrace.h: remove obsolete temporary development Linux
1167 constant PTRACE_SEIZE_DEVEL
1168 [21930] math: C-only gcc builtins used in <math.h> isinf
1169 [21932] network: Unpaired __resolv_context_get in generic get*_r
1171 [21941] math: powerpc: Wrong register constraint for xssqrtqp in sqrtf128
1172 [21944] libc: sigval namespace
1173 [21951] localedata: Update hanzi collation by stroke
1174 [21955] math: Wrong alignment of L(SP_RANGE)/L(SP_INF_0) in
1175 sysdeps/x86_64/fpu/e_expf.S
1176 [21956] libc: Stack allocation in MIPS syscall impl (ubounded stack
1177 allocation in syscall loops)
1178 [21959] localedata: Fix Country name for xh_ZA
1179 [21960] localedata: Fix abmon for bem_ZM
1180 [21966] math: AVX2 mathvec functions use FMA without checking
1181 [21967] math: When 512-bit AVX2 wrapper functions in mathvec are used?
1182 [21971] localedata: Added New Locale for mfe_MU
1183 [21972] libc: assert macro requires operator== (int) for its argument type
1184 [21973] math: [sparc] libm missing sqrtl compat symbol
1185 [21974] libc: Remove __bb_init_func and __bb_exit_func
1186 [21982] string: stratcliff.c: error: assuming signed overflow does not
1188 [21986] stdio: __guess_grouping is called incorrectly
1189 [21987] math: [sparc32] wrong bits/long-double.h installed
1190 [22019] localedata: Wrong placement of monetary symbol in el_GR (negative
1192 [22022] localedata: Missing country_name for mni_IN
1193 [22023] localedata: Removed redundant data (LC_TIME and LC_MESSAGES) for
1195 [22025] locale: iconv: Inconsistency between pointer mangling and NULL
1197 [22026] locale: iconv_open: heap overflow on gconv_init failure
1198 [22028] math: bits/math-finite.h _MSUF_ expansion namespace
1199 [22035] math: [m68k] bits/math-inline.h macro namespace
1200 [22038] localedata: Fix abbreviated weeks and months for Somali
1201 [22044] localedata: Remove redundant data for Limburgish Language
1202 [22050] malloc: Linking with -lmcheck does not hook
1203 __malloc_initialize_hook correctly
1204 [22051] libc: zero terminator in the middle of glibc's .eh_frame
1205 [22052] malloc: malloc failed to compile with GCC 7 and -O3
1206 [22070] localedata: charmaps/UTF-8: wcwidth for
1207 Prepended_Concatenation_Mark codepoints set to 0 (should be 1)
1208 [22074] localedata: charmaps/UTF-8: wcwidth for U+1160-U+11FF (Hangul
1209 Jungseong and Jongseong) should be 0
1210 [22078] nss: nss_files performance issue in multi mode
1211 [22082] math: bits/math-finite.h exp10 condition
1212 [22086] libc: pcprofiledump incorrect cross-endian condition
1213 [22093] dynamic-link: ld.so no longer searches in .../x86_64
1214 [22095] network: Name server address allocation memory leak in resolv.conf
1216 [22096] network: __resolv_conf_attach can incorrectly free passed conf
1218 [22100] localedata: om_KE: LC_TIME: copy redundant data from om_ET
1219 [22101] dynamic-link: Dynamic loader must ignore "debug" shared objects
1221 [22111] malloc: malloc: per thread cache is not returned when thread exits
1222 [22112] localedata: Fix LC_TELEPHONE/LC_NAME for az_AZ
1223 [22134] libc: [linux] implement fexecve with execveat
1224 [22142] libc: [powerpc] printf oupts a wrong value of DBL_MAX on ppc64 and
1226 [22145] libc: ttyname() gives up too early in the face of namespaces
1227 [22146] math: C++ build issue with float128 on x86_64
1228 [22153] nptl: nptl: save error code before process termination
1229 [22156] libc: [hppa,ia64,microblaze] Executable stack default
1230 [22159] malloc: malloc: MALLOC_CHECK_ broken with --enable-tunables=no
1231 [22161] nscd: nscd cache prune for netgroups hangs after timeout bump
1232 [22165] libc: [hppa] Text relocations in libc.so
1233 [22180] libc: destructor registered via __cxa_atexit is called twice
1234 [22183] glob: commit 5554304f0ddd ("posix: Allow glob to match dangling
1235 symlinks") cause "make" segfaults
1236 [22189] math: [powerpc] math_private.h definitions of math_opt_barrier and
1238 [22207] libc: FAIL: stdlib/test-atexit-race
1239 [22225] math: nearbyint arithmetic moved before feholdexcept
1240 [22229] math: [sparc32] missing copysignl, fabsl, fmal compat symbols
1241 [22235] math: iscanonical in C++ and float128
1242 [22243] math: log2(0) and log10(0) are wrong in downward rounding without
1243 the svid compat wrapper
1244 [22244] math: ynf and yn are wrong without the svid compat wrapper
1245 [22273] libc: Improper assert in Linux posix_spawn implementation
1246 [22284] libc: -pg -pie doesn't work
1247 [22292] locale: localedef exits with error 4 when it should be error 1
1248 [22294] locale: Allow "" for int_currency_symbol definition in locales.
1249 [22295] locale: Don't warn on non-symbolic characters in locale sources in
1251 [22296] math: glibc 2.26: signbit build issue with Gcc 5.5.0 on x86_64
1252 [22298] nptl: x32: lockups on recursive pthread_mutex_lock after upgrade
1254 [22299] dynamic-link: Problem with $PLATFORM on x86_64 platform
1255 [22320] glob: Buffer overflow in glob with GLOB_TILDE (CVE-2017-15670)
1256 [22321] libc: sysconf(_SC_IOV_MAX) returns -1 on Linux
1257 [22322] libc: [mips64] wrong bits/long-double.h installed
1258 [22325] glob: Memory leak in glob with GLOB_TILDE (CVE-2017-15671)
1259 [22332] glob: Buffer overflow in glob with GLOB_TILDE in unescaping
1261 [22336] localedata: cs_CZ LC_COLLATE does not use i18n
1262 [22343] malloc: Integer overflow in posix_memalign
1263 [22347] libc: getrandom() returns the number of bytes that were copied to
1264 the buffer even though the comments say "Return 0 on success and -1 on
1266 [22353] string: sysdeps/i386/i586/strcpy.S isn't maintainable
1267 [22362] libc: Installed crt1.o, crti,.o and crtn.o files are used with
1269 [22370] dynamic-link: Incorrect note padding check
1270 [22375] libc: malloc returns pointer from tcache_get when should return
1271 NULL (CVE-2017-17426)
1272 [22377] math: iseqsig, float128 and C++
1273 [22382] localedata: Error in tpi_PG locale
1274 [22387] localedata: Replace unicode sequences <Uxxxx> for characters
1275 inside the ASCII printable range
1276 [22402] math: [powerpc64le] __MATH_TG does not support _Float128 for
1278 [22403] localedata: Slash needs escaping in some locales
1279 [22408] malloc: malloc_info access heaps without arena lock, ignores heaps
1280 [22409] network: res_hnok does not accept some host names used on the
1282 [22412] network: res_dnok, res_hnok should perform syntax checks
1283 [22413] network: ns_name_pton ignores syntactically invalid trailing
1285 [22415] stdio: setvbuf can lead to invalid free/segfault
1286 [22432] build: Non-deterministic build
1287 [22439] malloc: malloc_info should compute summary statistics for all sub-
1289 [22442] network: if_nametoindex could report index for the wrong
1290 networking interface
1291 [22446] build: aliasing violation calling readlink in handle_request
1292 [22447] build: unsafe call to strlen with a non-string in getlogin_r.c
1293 [22457] libc: Generic preadv/pwritev incorrectly calls __posix_memalign
1294 [22459] libc: FAIL: elf/check-localplt with __stack_chk_fail related to
1295 __nscd_hash/__nss_hash
1296 [22463] network: p_secstodate overflow handling
1297 [22469] localedata: pl_PL LC_COLLATE does not use i18n
1298 [22478] libc: sigwait can fail with EINTR
1299 [22505] libc: ldconfig processes include directive in locale-specific
1301 [22515] localedata: hsb_DE LC_COLLATE does not use copy "iso14651_t1"
1302 [22517] localedata: et_EE LC_COLLATE does not use copy "iso14651_t1"
1303 [22519] localedata: is_IS LC_COLLATE does not use copy "iso14651_t1"
1304 [22524] localedata: lt_LT LC_COLLATE does not use copy "iso14651_t1"
1305 [22527] localedata: tr_TR LC_COLLATE does not use copy "iso14651_t1"
1306 [22534] localedata: Collation rules for Serbian and Bosnian should be the
1307 same as for Croatian
1308 [22561] math: [DR#471] cacosh (0 + iNaN) should return NaN +/- i pi/2
1309 [22568] math: [DR#471] ctanh (0 + iNaN), ctanh (0 + i Inf)
1310 [22577] libc: missing newline after "cannot allocate TLS data structures
1312 [22588] manual: manual/conf.texi: missing underscore in front of
1314 [22593] math: nextafter and nexttoward are declared with const attribute
1315 [22596] manual: manual: finite(nan) wrongly described as returning nonzero
1316 [22603] string: ia64 memchr overflows internal pointer check
1317 [22605] libc: SH clone does not set the exit code correctly
1318 [22606] dynamic-link: Incorrect array size computation in _dl_init_paths
1320 [22607] dynamic-link: Buffer Overflow in _dl_init_paths (CVE-2017-1000409)
1321 [22611] malloc: malloc/tst-realloc wrongly assumes that errno must not be
1322 modified in case of success
1323 [22614] build: gcc: error: unrecognized command line option ‘-no-pie’
1324 [22615] manual: manual: ambiguous wording about errno value in case of
1326 [22624] libc: MIPS setjmp() saves incorrect 'o0' register in --enable-
1328 [22625] dynamic-link: RPATH $ORIGIN replaced by PWD for AT_SECURE/SUID
1329 binaries or if /proc is not mounted (CVE-2017-16997)
1330 [22627] dynamic-link: $ORIGIN in $LD_LIBRARY_PATH is substituted twice
1331 [22630] build: $(no-pie-ldflag) is no longer effective
1332 [22631] math: [m68k] Bad const attributes in bits/mathinline.h
1333 [22635] nptl: pthread_self returns NULL before libpthread is loaded
1334 [22636] nptl: PTHREAD_STACK_MIN is too small on x86-64
1335 [22637] nptl: guard size is subtracted from thread stack size instead of
1337 [22648] libc: getrlimit/setrlimit with RLIM_INFINITY broken on alpha
1338 [22657] localedata: hu_HU: Avoid double space in date
1339 [22660] math: fmax, fmin sNaN handling on alpha
1340 [22664] libc: New warning of GCC8
1341 [22665] math: alpha: ceil and floor raise inexact exceptions
1342 [22666] math: alpha: trunc raise inexact exceptions
1343 [22667] libc: makecontext lacks stack alignment on i386
1344 [22678] libc: prlimit fails for RLIM_INFINITY values on 32-bit machines
1345 [22679] libc: getcwd(3) can succeed without returning an absolute path
1347 [22685] libc: PowerPC: Static AT_SECURE binaries segfault with lock-
1348 elision and tunables
1349 [22687] math: [powerpc-nofpu] complex long double functions spurious
1351 [22688] math: [powerpc-nofpu] remainderl wrong sign of zero result
1352 [22690] math: [ldbl-128ibm] lrintl, lroundl missing "invalid" exceptions
1353 [22691] math: [powerpc-nofpu] fmaxmagl, fminmagl spurious "invalid"
1355 [22693] math: [ldbl-128ibm] log1pl (-qNaN) spurious "invalid" exception
1356 [22697] math: [powerpc] llround spurious "inexact" exceptions on 32-bit
1358 [22701] nis: Incomplete removal of libnsl
1359 [22702] math: [powerpc-nofpu] nearbyintl traps with trapping "inexact"
1360 [22707] libc: Missing defines in elf.h for DF_1_STUB and DF_1_PIE.
1361 [22715] dynamic-link: FAIL: elf/tst-audit10
1362 [22719] libc: Backtrace tests fail on hppa
1363 [22742] libc: [aarch64] mcontext_t __reserved field got renamed
1364 [22743] nptl: __pthread_register_cancel corrupts stack after f81ddabffd
1365 [22765] crypt: (struct crypt_data *data)->initialized is not set to zero
1366 before the first call to crypt_r () in crypt/badsalttest.c
1373 * A per-thread cache has been added to malloc. Access to the cache requires
1374 no locks and therefore significantly accelerates the fast path to allocate
1375 and free small amounts of memory. Refilling an empty cache requires locking
1376 the underlying arena. Performance measurements show significant gains in a
1377 wide variety of user workloads. Workloads were captured using a special
1378 instrumented malloc and analyzed with a malloc simulator. Contributed by
1379 DJ Delorie with the help of Florian Weimer, and Carlos O'Donell.
1381 * Unicode 10.0.0 Support: Character encoding, character type info, and
1382 transliteration tables are all updated to Unicode 10.0.0, using
1383 generator scripts contributed by Mike FABIAN (Red Hat).
1384 These updates cause user visible changes, especially the changes in
1385 wcwidth for many emoji characters cause problems when emoji sequences
1386 are rendered with pango, see for example:
1387 https://bugzilla.gnome.org/show_bug.cgi?id=780669#c5
1389 * Collation of Hungarian has been overhauled and is now consistent with "The
1390 Rules of Hungarian Orthography, 12th edition" (Bug 18934). Contributed by
1393 * Improvements to the DNS stub resolver, contributed by Florian Weimer:
1395 - The GNU C Library will now detect when /etc/resolv.conf has been
1396 modified and reload the changed configuration. The new resolver option
1397 “no-reload” (RES_NORELOAD) disables this behavior.
1399 - The GNU C Library now supports an arbitrary number of search domains
1400 (configured using the “search” directive in /etc/resolv.conf);
1401 previously, there was a hard limit of six domains. For backward
1402 compatibility, applications that directly modify the ‘_res’ global
1403 object are still limited to six search domains.
1405 - When the “rotate” (RES_ROTATE) resolver option is active, the GNU C
1406 Library will now randomly pick a name server from the configuration as a
1407 starting point. (Previously, the second name server was always used.)
1409 * The tunables feature is now enabled by default. This allows users to tweak
1410 behavior of the GNU C Library using the GLIBC_TUNABLES environment variable.
1412 * New function reallocarray, which resizes an allocated block (like realloc)
1413 to the product of two sizes, with a guaranteed clean failure upon integer
1414 overflow in the multiplication. Originally from OpenBSD, contributed by
1415 Dennis Wölfing and Rüdiger Sonderfeld.
1417 * New wrappers for the Linux-specific system calls preadv2 and pwritev2.
1418 These are extended versions of preadv and pwritev, respectively, taking an
1419 additional flags argument. The set of supported flags depends on the
1420 running kernel; full support currently requires kernel 4.7 or later.
1422 * posix_spawnattr_setflags now supports the flag POSIX_SPAWN_SETSID, to
1423 create a new session ID for the spawned process. This feature is
1424 scheduled to be added to the next major revision of POSIX; for the time
1425 being, it is available under _GNU_SOURCE.
1427 * errno.h is now safe to use from C-preprocessed assembly language on all
1428 supported operating systems. In this context, it will only define the
1429 Exxxx constants, as preprocessor macros expanding to integer literals.
1431 * On ia64, powerpc64le, x86-32, and x86-64, the math library now implements
1432 128-bit floating point as defined by ISO/IEC/IEEE 60559:2011 (IEEE
1433 754-2008) and ISO/IEC TS 18661-3:2015. Contributed by Paul E. Murphy,
1434 Gabriel F. T. Gomes, Tulio Magno Quites Machado Filho, and Joseph Myers.
1436 To compile programs that use this feature, the compiler must support
1437 128-bit floating point with the type name _Float128 (as defined by TS
1438 18661-3) or __float128 (the nonstandard name used by GCC for C++, and for
1439 C prior to version 7). _GNU_SOURCE or __STDC_WANT_IEC_60559_TYPES_EXT__
1440 must be defined to make the new interfaces visible.
1442 The new functions and macros correspond to those present for other
1443 floating-point types (except for a few obsolescent interfaces not
1444 supported for the new type), with F128 or f128 suffixes; for example,
1445 strtof128, HUGE_VAL_F128 and cosf128. Following TS 18661-3, there are no
1446 printf or scanf formats for the new type; the strfromf128 and strtof128
1447 interfaces should be used instead.
1449 Deprecated and removed features, and other changes affecting compatibility:
1451 * The synchronization that pthread_spin_unlock performs has been changed to
1452 now be equivalent to a C11 atomic store with release memory order to the
1453 spin lock's memory location. Previously, several (but not all)
1454 architectures used stronger synchronization (e.g., containing what is
1455 often called a full barrier). This change can improve performance, but
1456 may affect odd fringe uses of spin locks that depend on the previous
1457 behavior (e.g., using spin locks as atomic variables to try to implement
1458 Dekker's mutual exclusion algorithm).
1460 * The port to Native Client running on ARMv7-A (--host=arm-nacl) has been
1463 * Sun RPC is deprecated. The rpcgen program, librpcsvc, and Sun RPC headers
1464 will only be built and installed when the GNU C Library is configured with
1465 --enable-obsolete-rpc. This allows alternative RPC implementations, such
1466 as TIRPC or rpcsvc-proto, to be used.
1468 * The NIS(+) name service modules, libnss_nis, libnss_nisplus, and
1469 libnss_compat, are deprecated, and will not be built or installed by
1472 The NIS(+) support library, libnsl, is also deprecated. By default, a
1473 compatibility shared library will be built and installed, but not headers
1474 or development libraries. Only a few NIS-related programs require this
1475 library. (In particular, the GNU C Library has never required programs
1476 that use 'gethostbyname' to be linked with libnsl.)
1478 Replacement implementations based on TIRPC, which additionally support
1479 IPv6, are available from <https://github.com/thkukuk/>. The configure
1480 option --enable-obsolete-nsl will cause libnsl's headers, and the NIS(+)
1481 name service modules, to be built and installed.
1483 * The DNS stub resolver no longer performs EDNS fallback. If EDNS or DNSSEC
1484 support is enabled, the configured recursive resolver must support EDNS.
1485 (Responding to EDNS-enabled queries with responses which are not
1486 EDNS-enabled is fine, but FORMERR responses are not.)
1488 * res_mkquery and res_nmkquery no longer support the IQUERY opcode. DNS
1489 servers have not supported this opcode for a long time.
1491 * The _res_opcodes variable has been removed from libresolv. It had been
1492 exported by accident.
1494 * <string.h> no longer includes inline versions of any string functions,
1495 as this kind of optimization is better done by the compiler. The macros
1496 __USE_STRING_INLINES and __NO_STRING_INLINES no longer have any effect.
1498 * The nonstandard header <xlocale.h> has been removed. Most programs should
1499 use <locale.h> instead. If you have a specific need for the definition of
1500 locale_t with no other declarations, please contact
1501 libc-alpha@sourceware.org and explain.
1503 * The obsolete header <sys/ultrasound.h> has been removed.
1505 * The obsolete signal constant SIGUNUSED is no longer defined by <signal.h>.
1507 * The obsolete function cfree has been removed. Applications should use
1510 * The stack_t type no longer has the name struct sigaltstack. This changes
1511 the C++ name mangling for interfaces involving this type.
1513 * The ucontext_t type no longer has the name struct ucontext. This changes
1514 the C++ name mangling for interfaces involving this type.
1516 * On M68k GNU/Linux and MIPS GNU/Linux, the fpregset_t type no longer has
1517 the name struct fpregset. On Nios II GNU/Linux, the mcontext_t type no
1518 longer has the name struct mcontext. On SPARC GNU/Linux, the struct
1519 mc_fq, struct rwindow, struct fpq and struct fq types are no longer
1520 defined in sys/ucontext.h, the mc_fpu_t type no longer has the name struct
1521 mc_fpu, the gwindows_t type no longer has the name struct gwindows and the
1522 fpregset_t type no longer has the name struct fpu. This changes the C++
1523 name mangling for interfaces involving those types.
1525 * On S/390 GNU/Linux, the constants defined by <sys/ptrace.h> have been
1526 synced with the kernel:
1528 - PTRACE_GETREGS, PTRACE_SETREGS, PTRACE_GETFPREGS and PTRACE_SETFPREGS
1529 are not supported on this architecture and have been removed.
1531 - PTRACE_SINGLEBLOCK, PTRACE_SECCOMP_GET_FILTER, PTRACE_PEEKUSR_AREA,
1532 PTRACE_POKEUSR_AREA, PTRACE_GET_LAST_BREAK, PTRACE_ENABLE_TE,
1533 PTRACE_DISABLE_TE and PTRACE_TE_ABORT_RAND have been added.
1535 Programs that assume the GET/SETREGS ptrace requests are universally
1536 available will now fail to build, instead of malfunctioning at runtime.
1538 Changes to build and runtime requirements:
1540 * Linux kernel 3.2 or later is required at runtime, on all architectures
1541 supported by that kernel. (This is a change from version 2.25 only for
1544 * GNU Binutils 2.25 or later is now required to build the GNU C Library.
1546 * On most architectures, GCC 4.9 or later is required to build the GNU C
1547 Library. On powerpc64le, GCC 6.2 or later is required.
1549 Older GCC versions and non-GNU compilers are still supported when
1550 compiling programs that use the GNU C Library. (We do not know exactly
1551 how old, and some GNU extensions to C may be _de facto_ required. If you
1552 are interested in helping us make this statement less vague, please
1553 contact libc-alpha@sourceware.org.)
1555 Security related changes:
1557 * The DNS stub resolver limits the advertised UDP buffer size to 1200 bytes,
1558 to avoid fragmentation-based spoofing attacks (CVE-2017-12132).
1560 * LD_LIBRARY_PATH is now ignored in binaries running in privileged AT_SECURE
1561 mode to guard against local privilege escalation attacks (CVE-2017-1000366).
1563 * Avoid printing a backtrace from the __stack_chk_fail function since it is
1564 called on a corrupt stack and a backtrace is unreliable on a corrupt stack
1567 * A use-after-free vulnerability in clntudp_call in the Sun RPC system has been
1568 fixed (CVE-2017-12133).
1570 The following bugs are resolved with this release:
1572 [984] network: Respond to changed resolv.conf in gethostbyname
1573 [5010] network: sunrpc service cleanup causes unwanted port mapper traffic
1574 [12068] localedata: sc_IT: misspelled yesexpr/day/abday/mon/abmon/date_fmt
1576 [12189] libc: __stack_chk_fail should not attempt a backtrace
1578 [14096] time: Race condition on timezone/tst-timezone.out
1579 [14172] localedata: az_IR: new locale
1580 [14995] build: glibc fails to build if gold is the default linker, even if
1582 [15998] build: [powerpc] Set arch_minimum_kernel for powerpc LE
1583 [16637] network: inet_pton function is accepting IPv6 with bad format
1584 [16640] string: string/strtok.c: undefined behaviour inconsistent between
1585 x86 and other generic code
1586 [16875] localedata: ko_KR: fix lang_name
1587 [17225] localedata: ar_SY: localized month names for May and June are
1589 [17297] localedata: da_DK: wrong date_fmt string
1590 [18907] stdio: Incorrect order of __wur __THROW in <printf.h>
1591 [18934] localedata: hu_HU: collate: fix multiple bugs and add tests
1592 [18988] nptl: pthread wastes memory with mlockall(MCL_FUTURE)
1593 [19066] localedata: ar_SA abbreviated day and month names are in English
1594 [19569] network: resolv: Support an arbitrary number of search domains
1595 [19570] network: Implement random DNS server selection in the stub
1597 [19838] locale: localedef fails on PA-RISC
1598 [19919] localedata: iso14651_t1_common: Correct the Malayalam sorting
1599 order of 0D36 and 0D37
1600 [19922] localedata: iso14651_t1_common: Define collation for Malayalam
1602 [20098] libc: FAIL: debug/backtrace-tst on hppa
1603 [20257] network: sunrpc: clntudp_call does not enforce timeout when
1605 [20275] localedata: locale day/abday/mon/abmon should not have trailing
1607 [20313] localedata: Update locale data to Unicode 9.0
1608 [20424] manual: Document how to provide a malloc replacement
1609 [20496] localedata: agr_PE: new language locale Awajún / Aguaruna (agr)
1611 [20686] locale: Add el_GR@euro to SUPPORTED.
1612 [20831] dynamic-link: _dl_map_segments does not test for __mprotect
1613 failures consistently
1614 [21015] dynamic-link: Document and fix --enable-bind-now
1615 [21016] nptl: pthread_cond support is broken on hppa
1616 [21029] libc: glibc-2.23 (and later) fails to compile with -fno-omit-
1617 frame-pointer on i386
1618 [21049] libc: segfault in longjmp_chk() due to clobbered processor
1620 [21075] libc: unused assigment to %g4 in sparc/sparc{64,32}/clone.S
1621 [21088] libc: Build fails with --enable-static-nss
1622 [21094] math: cosf(1.57079697) has 3 ulp error on targets where the
1623 generic c code is used
1624 [21109] libc: Tunables broken on big-endian
1625 [21112] math: powf has large ulp errors with base close to 1 and exponent
1627 [21115] network: sunrpc: Use-after-free in error path in clntudp_call
1629 [21120] malloc: glibc malloc is incompatible with GCC 7
1630 [21130] math: Incorrect return from y0l (-inf) and y1l (-inf) when linking
1632 [21134] math: Exception (divide by zero) not set for y0/y1 (0.0) and y0/y1
1633 (-0.0) when linking with -lieee
1634 [21171] math: log10, log2 and lgamma return incorrect results
1635 [21179] libc: handle R_SPARC_DISP64 and R_SPARC_REGISTER relocs
1636 [21182] libc: __memchr_sse2: regression in glibc-2.25 on i686
1637 [21207] localedata: ce_RU: update weekdays from CLDR
1638 [21209] dynamic-link: LD_HWCAP_MASK read in setuid binaries
1639 [21217] localedata: Update months from CLDR-31
1640 [21232] libc: miss posix_fadvise64 on MIPS64 when static linking
1641 [21243] libc: support_delete_temp_file should issue warning for failed
1643 [21244] libc: support resolv_test_start() socket fd close should be
1645 [21253] libc: localedef randomly segfaults when using -fstack-check due to
1646 new posix_spawn implementation
1647 [21258] dynamic-link: Branch predication in _dl_runtime_resolve_avx512_opt
1648 leads to lower CPU frequency
1649 [21259] libc: [alpha] termios.h missing IXANY for POSIX
1650 [21261] libc: [sparc64] bits/setjmp.h namespace
1651 [21267] network: [mips] bits/socket.h IOC* namespace
1652 [21268] libc: [alpha] termios.h NL2, NL3 namespace
1653 [21270] libc: mmap64 silently truncates large offset values
1654 [21275] libc: posix_spawn always crashes on ia64 now
1655 [21277] libc: [alpha] termios.h missing IUCLC for UNIX98 and older
1656 [21280] math: [powerpc] logbl for POWER7 return incorrect results
1657 [21289] libc: Incorrect declaration for 32-bit platforms with
1658 _FILE_OFFSET_BITS=64 causes build error
1659 [21295] network: GETAI(AF_UNSPEC) drops IPv6 addresses if nss module does
1660 not support gethostbyname4_r
1661 [21298] nptl: rwlock can deadlock on frequent reader/writer phase
1663 [21338] malloc: mallopt M_ARENA_MAX doesn't set the maximum number of
1665 [21340] libc: Support POSIX_SPAWN_SETSID
1666 [21357] libc: unwind-dw2-fde deadlock when using AddressSanitizer
1667 [21359] network: ns_name_pack needs additional byte in destination buffer
1668 [21361] network: resolv: Reduce advertised EDNS0 buffer size to guard
1669 against fragmentation attacks (CVE-2017-12132)
1670 [21369] network: resolv: Remove EDNS fallback
1671 [21371] libc: Missing timespec definition when compiled with _XOPEN_SOURCE
1673 [21386] nptl: Assertion in fork for distinct parent PID is incorrect
1674 [21391] dynamic-link: x86: Set dl_platform and dl_hwcap from CPU features
1675 [21393] stdio: Missing dup3 error check in freopen, freopen64
1676 [21396] libc: Use AVX2 memcpy/memset on Skylake server
1677 [21399] localedata: Bad description for U00EC in
1678 localedata/charmaps/CP1254
1679 [21411] malloc: realloc documentation error
1680 [21426] network: sys/socket.h uio.h namespace
1681 [21428] libc: [aarch64] tst-backtrace5 testsuite failure
1682 [21445] libc: signal.h bsd_signal namespace
1683 [21455] network: Network headers stdint.h namespace
1684 [21474] network: resolv: res_init does not use RES_DFLRETRY (2) but 4 for
1686 [21475] network: resolv: Overlong search path is truncated mid-label
1687 [21511] libc: sigstack namespace
1688 [21512] libc: clone() ends up calling exit_group() through _exit() wrapper
1689 [21514] libc: sysdeps/unix/sysv/linux/sys/syscall.h:31:27: fatal error:
1690 bits/syscall.h: No such file or directory
1691 [21517] libc: struct sigaltstack namespace
1692 [21528] dynamic-link: Duplicated minimal strtoul implementations in ld.so
1693 [21533] localedata: Update locale data to Unicode 10.0
1695 ../sysdeps/unix/sysv/linux/s390/s390-32/__makecontext_ret.S:44: Error:
1696 junk at end of line, first unrecognized character is `@'
1697 [21538] libc: SIG_HOLD missing for XPG4
1698 [21539] libc: S390: Mismatch between kernel and glibc ptrace.h with
1699 request 12: PTRACE_SINGLEBLOCK vs PTRACE_GETREGS.
1700 [21542] libc: Use conservative default for sysconf (_SC_NPROCESSORS_ONLN)
1701 [21543] libc: sigevent namespace
1702 [21548] libc: [mips] get/set/make/swap context for MIPS O32 assume wrong
1703 size for general purpose registers in mcontext_t structure
1704 [21550] libc: sigwait namespace
1705 [21552] libc: XPG4 bsd_signal namespace
1706 [21554] libc: sigpause namespace
1707 [21560] libc: sys/wait.h signal.h namespace
1708 [21561] libc: waitid namespace
1709 [21573] nptl: GCC 7: /usr/bin/install: cannot remove
1710 '/usr/include/stdlib.h': Permission denied
1711 [21575] libc: sys/wait.h missing struct rusage definition
1712 [21584] libc: sigaltstack etc namespace
1713 [21597] libc: siginterrupt namespace
1714 [21607] math: hppa: FAIL: math/test-tgmath
1715 [21609] dynamic-link: Incomplete workaround for GCC __tls_get_addr ABI
1717 [21622] libc: [tile] missing SA_* for POSIX.1:2008
1718 [21624] dynamic-link: ld.so: Unsafe alloca allows local attackers to alias
1719 stack and heap (CVE-2017-1000366)
1720 [21625] libc: wait3 namespace
1721 [21654] nss: Incorrect pointer alignment in NSS group merge result
1723 [21657] network: Parse interface zone id for node-local multicast
1724 [21662] string: memcmp-avx2-movbe.S lacks saturating subtraction for
1726 [21666] libc: .symver is used on common symbol
1727 [21668] network: resolv: res_init cross-thread broadcast introduces race
1729 [21687] math: tgmath.h totalorder, totalordermag return type
1730 [21694] locale: Current Glibc Locale Does Not Support Tok-Pisin and Fiji
1732 [21696] libc: Incorrect assumption of of __cpu_mask in
1733 posix/sched_cpucount.c
1734 [21697] libc: sysdeps/posix/spawni.c: 2 * suspicious condition ?
1735 [21706] localedata: yesstr and nostr are missing for Breton [LC_MESSAGES]
1737 [21707] math: ppc64le: Invalid IFUNC resolver from libgcc calls getauxval,
1738 leading to relocation crash
1739 [21709] libc: resolv_conf.c:552: update_from_conf: Assertion
1740 `resolv_conf_matches (resp, conf)' failed.
1741 [21710] localedata: Added Samoan language locale for Samoa
1742 [21711] localedata: Pashto yesstr/nostr locale are missing
1743 [21715] nptl: sysdeps/nptl/bits/pthreadtypes.h: typedef guard
1744 __have_pthread_attr_t can cause redefinition of typedef ‘pthread_attr_t’
1745 [21721] localedata: Incorrect Full Weekday names for ks_IN@devanagari
1746 [21723] localedata: yesstr/nostr missing for Chinese language locale
1747 [21724] localedata: yesstr and nostr are missing for Xhosa [LC_MESSAGES]
1749 [21727] localedata: yesstr and nostr are missing for Tsonga [LC_MESSAGES]
1751 [21728] localedata: New Locale for Tongan language
1752 [21729] localedata: incorrect LC_NAME fields for hi_IN
1753 [21733] localedata: yesstr and nostr are missing for zh_HK
1754 [21734] localedata: Missing yesstr and nostr are for kw_GB
1755 [21738] libc: misc/tst-preadvwritev2 and misc/tst-preadvwritev64v2 fail
1756 [21741] libc: Undefined __memmove_chk_XXX and __memset_chk_XXX in libc.a
1757 [21742] libc: _dl_num_cache_relocations is undefined in libc.a
1758 [21743] localedata: ks_IN@devanagari: abday strings mismatch the day
1760 [21744] libc: Tests failing on --enable-tunables --enable-stack-
1762 [21749] localedata: Wrong abbreviated day name (“abday”) for
1764 [21756] localedata: missing yesstr, nostr for nds_DE and nds_NL
1765 [21757] localedata: missing yesstr, nostr for pap_AW and pap_CW
1766 [21759] localedata: missing yesstr and nostr for Tigrinya
1767 [21760] localedata: Fix LC_MESSAGES and LC_ADDRESS for anp_IN
1768 [21766] localedata: Wrong LC_MESSAGES for om_ET Locale
1769 [21767] localedata: Missing Bislama locales
1770 [21768] localedata: Missing yesstr and nostr for aa_ET
1771 [21770] localedata: Missing Field in li_NL
1772 [21778] nptl: Robust mutex may deadlock
1773 [21779] libc: MicroBlaze segfaults when loading libpthread
1774 [21783] localedata: Fix int_select international_call_prefixes
1775 [21784] localedata: Inconsistency in country_isbn
1776 [21788] localedata: Missing Country Postal Abbreviations
1777 [21794] localedata: Added-country_isbn-for-Italy
1778 [21795] localedata: Add/Fix country_isbn for France
1779 [21796] localedata: Added country_isbn for Republic of Korea
1780 [21797] localedata: Fix inconsistency in country_isbn and missing prefixes
1781 [21799] localedata: Added int_select international_call_prefixes
1782 [21801] localedata: Added int_select international_call_prefixes
1783 [21804] nptl: Double semicolon in thread-shared-types.h
1784 [21807] localedata: LC_ADDRESS fix for pap_CW
1785 [21808] localedata: Fix LC_ADDRESS for pap_AW
1786 [21821] localedata: Added country_name in mai_IN
1787 [21822] localedata: Fix LC_TIME for mai_IN
1788 [21823] localedata: missing yesstr, nostr for sa_IN
1789 [21825] localedata: Fix name_mrs for mag_IN
1790 [21828] localedata: 2.26 changelog should mention user visible changes
1792 [21835] localedata: Added Maithili language locale for Nepal
1793 [21838] localedata: Removed redundant data for the_NP
1794 [21839] localedata: Fix LC_MONETARY for ta_LK
1795 [21844] localedata: Fix Latin characters and Months Sequence.
1796 [21848] localedata: Fix mai_NP Title Name
1801 * The feature test macro __STDC_WANT_LIB_EXT2__, from ISO/IEC TR
1802 24731-2:2010, is supported to enable declarations of functions from that
1803 TR. Note that not all functions from that TR are supported by the GNU C
1806 * The feature test macro __STDC_WANT_IEC_60559_BFP_EXT__, from ISO/IEC TS
1807 18661-1:2014, is supported to enable declarations of functions and macros
1808 from that TS. Note that not all features from that TS are supported by
1811 * The feature test macro __STDC_WANT_IEC_60559_FUNCS_EXT__, from ISO/IEC TS
1812 18661-4:2015, is supported to enable declarations of functions and macros
1813 from that TS. Note that most features from that TS are not supported by
1816 * The nonstandard feature selection macros _REENTRANT and _THREAD_SAFE are
1817 now treated as compatibility synonyms for _POSIX_C_SOURCE=199506L.
1818 Since the GNU C Library defaults to a much newer revision of POSIX, this
1819 will only affect programs that specifically request an old conformance
1820 mode. For instance, a program compiled with -std=c89 -D_REENTRANT will
1821 see a change in the visible declarations, but a program compiled with
1822 just -D_REENTRANT, or -std=c99 -D_POSIX_C_SOURCE=200809L -D_REENTRANT,
1825 Some C libraries once required _REENTRANT and/or _THREAD_SAFE to be
1826 defined by all multithreaded code, but glibc has not required this for
1829 * The inclusion of <sys/sysmacros.h> by <sys/types.h> is deprecated. This
1830 means that in a future release, the macros “major”, “minor”, and “makedev”
1831 will only be available from <sys/sysmacros.h>.
1833 These macros are not part of POSIX nor XSI, and their names frequently
1834 collide with user code; see for instance glibc bug 19239 and Red Hat bug
1835 130601. <stdlib.h> includes <sys/types.h> under _GNU_SOURCE, and C++ code
1836 presently cannot avoid being compiled under _GNU_SOURCE, exacerbating the
1839 * New <fenv.h> features from TS 18661-1:2014 are added to libm: the
1840 fesetexcept, fetestexceptflag, fegetmode and fesetmode functions, the
1841 femode_t type and the FE_DFL_MODE and FE_SNANS_ALWAYS_SIGNAL macros.
1843 * Integer width macros from TS 18661-1:2014 are added to <limits.h>:
1844 CHAR_WIDTH, SCHAR_WIDTH, UCHAR_WIDTH, SHRT_WIDTH, USHRT_WIDTH, INT_WIDTH,
1845 UINT_WIDTH, LONG_WIDTH, ULONG_WIDTH, LLONG_WIDTH, ULLONG_WIDTH; and to
1846 <stdint.h>: INT8_WIDTH, UINT8_WIDTH, INT16_WIDTH, UINT16_WIDTH,
1847 INT32_WIDTH, UINT32_WIDTH, INT64_WIDTH, UINT64_WIDTH, INT_LEAST8_WIDTH,
1848 UINT_LEAST8_WIDTH, INT_LEAST16_WIDTH, UINT_LEAST16_WIDTH,
1849 INT_LEAST32_WIDTH, UINT_LEAST32_WIDTH, INT_LEAST64_WIDTH,
1850 UINT_LEAST64_WIDTH, INT_FAST8_WIDTH, UINT_FAST8_WIDTH, INT_FAST16_WIDTH,
1851 UINT_FAST16_WIDTH, INT_FAST32_WIDTH, UINT_FAST32_WIDTH, INT_FAST64_WIDTH,
1852 UINT_FAST64_WIDTH, INTPTR_WIDTH, UINTPTR_WIDTH, INTMAX_WIDTH,
1853 UINTMAX_WIDTH, PTRDIFF_WIDTH, SIG_ATOMIC_WIDTH, SIZE_WIDTH, WCHAR_WIDTH,
1856 * New <math.h> features are added from TS 18661-1:2014:
1858 - Signaling NaN macros: SNANF, SNAN, SNANL.
1860 - Nearest integer functions: roundeven, roundevenf, roundevenl, fromfp,
1861 fromfpf, fromfpl, ufromfp, ufromfpf, ufromfpl, fromfpx, fromfpxf,
1862 fromfpxl, ufromfpx, ufromfpxf, ufromfpxl.
1864 - llogb functions: the llogb, llogbf and llogbl functions, and the
1865 FP_LLOGB0 and FP_LLOGBNAN macros.
1867 - Max-min magnitude functions: fmaxmag, fmaxmagf, fmaxmagl, fminmag,
1870 - Comparison macros: iseqsig.
1872 - Classification macros: iscanonical, issubnormal, iszero.
1874 - Total order functions: totalorder, totalorderf, totalorderl,
1875 totalordermag, totalordermagf, totalordermagl.
1877 - Canonicalize functions: canonicalize, canonicalizef, canonicalizel.
1879 - NaN functions: getpayload, getpayloadf, getpayloadl, setpayload,
1880 setpayloadf, setpayloadl, setpayloadsig, setpayloadsigf, setpayloadsigl.
1882 * The functions strfromd, strfromf, and strfroml, from ISO/IEC TS 18661-1:2014,
1883 are added to libc. They convert a floating-point number into string.
1885 * Most of glibc can now be built with the stack smashing protector enabled.
1886 It is recommended to build glibc with --enable-stack-protector=strong.
1887 Implemented by Nick Alcock (Oracle).
1889 * The function explicit_bzero, from OpenBSD, has been added to libc. It is
1890 intended to be used instead of memset() to erase sensitive data after use;
1891 the compiler will not optimize out calls to explicit_bzero even if they
1892 are "unnecessary" (in the sense that no _correct_ program can observe the
1893 effects of the memory clear).
1895 * On ColdFire, MicroBlaze, Nios II and SH3, the float_t type is now defined
1896 to float instead of double. This does not affect the ABI of any libraries
1897 that are part of the GNU C Library, but may affect the ABI of other
1898 libraries that use this type in their interfaces.
1900 * On x86_64, when compiling with -mfpmath=387 or -mfpmath=sse+387, the
1901 float_t and double_t types are now defined to long double instead of float
1902 and double. These options are not the default, and this does not affect
1903 the ABI of any libraries that are part of the GNU C Library, but it may
1904 affect the ABI of other libraries that use this type in their interfaces,
1905 if they are compiled or used with those options.
1907 * The getentropy and getrandom functions, and the <sys/random.h> header file
1910 * The buffer size for byte-oriented stdio streams is now limited to 8192
1911 bytes by default. Previously, on Linux, the default buffer size on most
1912 file systems was 4096 bytes (and thus remains unchanged), except on
1913 network file systems, where the buffer size was unpredictable and could be
1914 as large as several megabytes.
1916 * The <sys/quota.h> header now includes the <linux/quota.h> header. Support
1917 for the Linux quota interface which predates kernel version 2.4.22 has
1920 * The malloc_get_state and malloc_set_state functions have been removed.
1921 Already-existing binaries that dynamically link to these functions will
1922 get a hidden implementation in which malloc_get_state is a stub. As far
1923 as we know, these functions are used only by GNU Emacs and this change
1924 will not adversely affect already-built Emacs executables. Any undumped
1925 Emacs executables, which normally exist only during an Emacs build, should
1926 be rebuilt by re-running “./configure; make” in the Emacs build tree.
1928 * The “ip6-dotint” and “no-ip6-dotint” resolver options, and the
1929 corresponding RES_NOIP6DOTINT flag from <resolv.h> have been removed.
1930 “no-ip6-dotint” had already been the default, and support for the
1931 “ip6-dotint” option was removed from the Internet in 2006.
1933 * The "ip6-bytestring" resolver option and the corresponding RES_USEBSTRING
1934 flag from <resolv.h> have been removed. The option relied on a
1935 backwards-incompatible DNS extension which was never deployed on the
1938 * The flags RES_AAONLY, RES_PRIMARY, RES_NOCHECKNAME, RES_KEEPTSIG,
1939 RES_BLAST defined in the <resolv.h> header file have been deprecated.
1940 They were already unimplemented.
1942 * The "inet6" option in /etc/resolv.conf and the RES_USE_INET6 flag for
1943 _res.flags are deprecated. The flag was standardized in RFC 2133, but
1944 removed again from the IETF name lookup interface specification in RFC
1945 2553. Applications should use getaddrinfo instead.
1947 * DNSSEC-related declarations and definitions have been removed from the
1948 <arpa/nameser.h> header file, and libresolv will no longer attempt to
1949 decode the data part of DNSSEC record types. Previous versions of glibc
1950 only implemented minimal support for the previous version of DNSSEC, which
1951 is incompatible with the currently deployed version.
1953 * The resource record type classification macros ns_t_qt_p, ns_t_mrr_p,
1954 ns_t_rr_p, ns_t_udp_p, ns_t_xfr_p have been removed from the
1955 <arpa/nameser.h> header file because the distinction between RR types and
1956 meta-RR types is not officially standardized, subject to revision, and
1957 thus not suitable for encoding in a macro.
1959 * The types res_sendhookact, res_send_qhook, re_send_rhook, and the qhook
1960 and rhook members of the res_state type in <resolv.h> have been removed.
1961 The glibc stub resolver did not support these hooks, but the header file
1962 did not reflect that.
1964 * For multi-arch support it is recommended to use a GCC which has
1965 been built with support for GNU indirect functions. This ensures
1966 that correct debugging information is generated for functions
1967 selected by IFUNC resolvers. This support can either be enabled by
1968 configuring GCC with '--enable-gnu-indirect-function', or by
1969 enabling it by default by setting 'default_gnu_indirect_function'
1970 variable for a particular architecture in the GCC source file
1973 * GDB pretty printers have been added for mutex and condition variable
1974 structures in POSIX Threads. When installed and loaded in gdb these pretty
1975 printers show various pthread variables in human-readable form when read
1976 using the 'print' or 'display' commands in gdb.
1978 * Tunables feature added to allow tweaking of the runtime for an application
1979 program. This feature can be enabled with the '--enable-tunables' configure
1980 flag. The GNU C Library manual has details on usage and README.tunables has
1981 instructions on adding new tunables to the library.
1983 * A new version of condition variables functions have been implemented in
1984 the NPTL implementation of POSIX Threads to provide stronger ordering
1987 * A new version of pthread_rwlock functions have been implemented to use a more
1988 scalable algorithm primarily through not using a critical section anymore to
1991 Security related changes:
1993 * On ARM EABI (32-bit), generating a backtrace for execution contexts which
1994 have been created with makecontext could fail to terminate due to a
1995 missing .cantunwind annotation. This has been observed to lead to a hang
1996 (denial of service) in some Go applications compiled with gccgo. Reported
1997 by Andreas Schwab. (CVE-2016-6323)
1999 * The DNS stub resolver functions would crash due to a NULL pointer
2000 dereference when processing a query with a valid DNS question type which
2001 was used internally in the implementation. The stub resolver now uses a
2002 question type which is outside the range of valid question type values.
2005 The following bugs are resolved with this release:
2007 [4099] stdio: Overly agressive caching by stream i/o functions.
2008 [7065] build: Support building glibc with -fstack-protector or -fstack-
2010 [9842] localedata: en_CA: incorrect date format
2011 [13165] nptl: pthread_cond_wait() can consume a signal that was sent
2012 before it started waiting
2013 [14139] manual: Do not hardcode platform names in manual/libm-err-tab.pl
2014 [15765] nptl: sem_open is wrongly a cancellation point
2015 [16421] network: IN6_IS_ADDR_UNSPECIFIED can use undefined s6_addr32
2016 [16458] libc: endian.h and netinet/in.h byte order macros return values of
2018 [16628] dynamic-link: Segfault after a binary without pthread dlopen()s a
2019 library linked with pthread
2020 [16630] nptl: Use SYSENTER for pthread_cond_broadcast/signal() (i.e. fix
2021 "FIXME: Ingo" issue)
2022 [16907] libc: <argp.h> compiled with --std=cXX disables __attribute__
2023 [17252] libc: getrandom and getentropy syscall
2024 [17730] malloc: thread-local storage is sometimes improperly free()'d
2025 after being __libc_memalign()'d
2026 [18241] stdio: failed fseek on memstream does not set errno and can fail
2028 [18243] nptl: sem_wait, sem_timedwait are cancellation points shm_open is
2030 [18463] nptl: pthread_cond_broadcast issue when surrounded by
2031 PTHREAD_PRIO_INHERIT mutex on ARM
2032 [18784] network: res_query and related function crash for special record
2033 type queries (CVE-2015-5180)
2034 [19380] math: strtod does not raise "inexact"
2035 [19387] string: Integer overflow in memchr
2036 [19390] string: Integer overflow in strncat
2037 [19398] build: linknamespace tests fail with massively parallel build
2038 [19402] nptl: Deadlock with robust shared mutex and asynchronous
2040 [19469] malloc: M_PERTURB in test-skeleton.c invalidates malloc tests
2041 [19473] malloc: Turn malloc_get_state etc. in compatibility symbols
2042 [19514] libc: [PATCH] Fix spelling errors in spelling
2043 "implement"/"implementation" in several places
2044 [19582] network: Deprecate RES_USE_INET6
2045 [19673] manual: clog10 docs appear to be erroneous
2046 [19810] dynamic-link: dlopen with both RTLD_NOLOAD and RTLD_NODELETE
2047 causes a segmentation fault
2048 [19826] libc: invalid pointer returned from __tls_get_addr with static
2050 [20016] network: resolv: Remove hooks support from the API
2051 [20019] dynamic-link: NULL pointer dereference in libc.so.6 IFUNC due to
2053 [20033] math: [x86_64] vectorized math function don't call the __finite
2055 [20116] nptl: use after free in pthread_create
2056 [20181] stdio: open_memstream(): writes not at end of stream corrupt data
2057 [20292] dynamic-link: Comparison in elf/dl-open.c _dl_addr_inside_object
2059 [20311] nptl: please install proc_service.h
2060 [20366] libc: Compilation errors in installed headers in strict-compliance
2062 [20370] malloc: malloc: Arena free list management is still racy
2063 (incorrect fix in bug 19243)
2064 [20386] libc: assert (X = 0) does not result in GCC warning
2065 [20432] malloc: malloc: Minimize interface required for interposition
2066 [20435] libc: Missing unwind info in __startcontext causes infinite loop
2067 in _Unwind_Backtrace (CVE-2016-6323)
2068 [20444] hurd: recvmsg: PF_LOCAL sockets and msg_name lead to SIGLOST
2069 [20452] nptl: Addition of sysdep.o to libpthread.a breaks relinking
2071 [20455] math: [powerpc] fesetexceptflag fails to clear FE_INVALID
2072 [20459] localedata: et_EE: locale has wrong {p,n}_cs_precedes value
2073 [20477] network: resolv: incorrect double-checked locking related to
2075 [20478] libc: libc_ifunc macro and similar usages leads to false debug-
2077 [20495] math: x86_64 performance degradation due to AVX/SSE transition
2079 [20497] localedata: lt_LT: LC_TIME d_fmt used is obsolete
2080 [20508] dynamic-link: _dl_runtime_resolve_avx/_dl_runtime_profile_avx512
2081 cause transition penalty
2082 [20517] math: sparcv9 missing fdiml compat symbol
2083 [20524] manual: strverscmp is inconsistent
2084 [20525] libc: <sys/quota.h> should be based on kernel headers
2085 [20539] math: GCC 7 -static -lm fails to link at -O3
2086 [20554] libc: ld: bss-plt forced due to /usr/lib/libc_nonshared.a(ppc-
2088 [20558] string: POSIX bcopy/bzero decls do not implement Object Size
2090 [20591] network: Remove obsolete DNSSEC support
2091 [20592] network: DNS resource record type classification macros in
2092 <arpa/nameser.h> are incorrect
2093 [20593] network: Update DNS RR type definitions
2094 [20611] network: getaddrinfo accepts invalid numeric scope IDs
2095 [20615] build: glibc build fails when using --with-cpu=power9 --enable-
2097 [20629] network: libresolv: Remove support for bitstring labels
2099 [20647] libc: GLIBC quitting every program - glibc on Pentium-S leads to
2100 assertion: "maxidx >= 2"
2101 [20660] math: [arm] Use VSQRT
2102 [20662] libc: checking whether x86_64-pc-linux-gnu-gcc implicitly enables
2103 -fstack-protector no (32bit gcc 6.2.0 pie and ssp enable)
2104 [20689] libc: Test for FMA should also check for AVX.
2105 [20707] glob: gl_pathv entries not set to NULL with GLOB_DOOFFS
2106 [20715] math: iszero macro breaks existing code
2107 [20718] math: [powerpc] copysignl raises "invalid" for sNaN
2108 [20728] libc: powerpc: Missing TOC stub in clone
2109 [20729] build: glibc-2.24 fails to build for i486 with -Os
2110 [20750] build: Build fails with default PIE enabled gcc-6.x
2111 [20768] math: [alpha] sqrt fegetenv namespace
2112 [20785] libc: binutils 2.28 fails to assemble power6/memset.S file in
2114 [20787] math: float_t is defined as float incorrectly on x86_64 even if
2115 __FLT_EVAL_METHOD__ is 2
2116 [20790] network: rpcgen buffer overrun in get_prog_declaration
2117 [20822] nptl: race condition in __lll_unlock_elision on powerpc
2118 [20829] libc: crypt snprintf namespace
2119 [20847] libc: tst-vfork3 failure
2120 [20855] math: Default bits/mathdef.h has inappropriate float_t
2121 [20859] math: [sh4] FP_ILOGB0 invalid
2122 [20864] localedata: iconv: cp936 missing single-byte euro sign (0x80,
2123 U+20AC), not same as GBK
2124 [20915] dynamic-link: global-dynamic TLS broken on aarch64
2125 [20916] math: pow handling of sNaN arguments
2126 [20918] build: Building with --enable-nss-crypt fails tst-linkall-static.
2127 [20919] math: Bad pow (qNaN, 0) result with -lieee
2128 [20940] math: hypot sNaN handling
2129 [20947] math: fmax, fmin sNaN handling
2130 [20956] libc: debug/tst-backtrace3-6 don't work with -O3 anymore
2131 [20964] network: sunrpc: Stack-based buffer overflow in getrpcport with
2133 [20971] string: powerpc64/power7 memchr overflows internal pointer check
2134 [20973] nptl: robust mutexes: Lost wake-ups
2135 [20974] locale: bs_BA: yesexpr/noexpr regexes accept any character
2136 [20978] nis: strlen on null pointer in nss_nisplus
2137 [20985] nptl: robust mutexes: lowlevelrobustlock assembly on x86 blocks on
2139 [21014] string: i686 memchr overflows internal pointer check
2140 [21019] libc: [mips] n32 lseek incorrect on overflow
2141 [21022] libc: [microblaze] __backtrace get_frame_size namespace
2142 [21026] libc: [MIPS] readahead syscall is broken on n64
2143 [21028] math: Fallback fesetexceptflag should always succeed
2144 [21045] libc: [powerpc-nofpu] swapcontext does not restore signal mask
2145 [21047] math: arm: fpu_control.h: _FPU_GETCW/_FPU_SETCW is rejected by
2147 [21053] libc: [SH] Namespace pollution from sys/ucontext.h
2148 [21061] librt: [microblaze] librt lost clock_* exports
2149 [21073] libc: tunables: insecure environment variables passed to
2150 subprocesses with AT_SECURE
2151 [21081] string: Missing vzeroupper in memset-vec-unaligned-erms.S
2155 * The minimum Linux kernel version that this version of the GNU C Library
2156 can be used with is 3.2, except on i[4567]86 and x86_64, where Linux
2157 kernel version 2.6.32 or later suffices (on architectures that already
2158 required kernel versions more recent than 3.2, those requirements remain
2159 unchanged). Linux 3.2 or later kernel headers are required on all
2162 * The pap_AN locale has been deleted. This has been deprecated for a long
2163 time. It has been replaced by pap_AW & pap_CW, both of which have long
2164 been included in previous releases.
2166 * The readdir_r and readdir64_r functions have been deprecated. It is
2167 recommended to use readdir and readdir64 instead.
2169 * The type “union wait” has been removed. It was deprecated in the early
2170 1990s and never part of POSIX. Application code should use the int type
2171 instead of “union wait”.
2173 * A new NSS action is added to facilitate large distributed system
2174 administration. The action, MERGE, allows remote user stores like LDAP
2175 to be merged into local user stores like /etc/groups in order to provide
2176 easy to use, updated, and managed sets of merged credentials. The new
2177 action can be used by configuring it in /etc/nsswitch.conf:
2178 group: files [SUCCESS=merge] nis
2179 Implemented by Stephen Gallagher (Red Hat).
2181 * The deprecated __malloc_initialize_hook variable has been removed from the
2184 * The long unused localedef --old-style option has been removed. It hasn't
2185 done anything in over 16 years. Scripts using this option can safely
2188 * nextupl, nextup, nextupf, nextdownl, nextdown and nextdownf are added to
2189 libm. They are defined by TS 18661 and IEEE754-2008. The nextup functions
2190 return the next representable value in the direction of positive infinity
2191 and the nextdown functions return the next representable value in the
2192 direction of negative infinity. These are currently enabled as GNU
2195 Security related changes:
2197 * An unnecessary stack copy in _nss_dns_getnetbyname_r was removed. It
2198 could result in a stack overflow when getnetbyname was called with an
2199 overly long name. (CVE-2016-3075)
2201 * Previously, getaddrinfo copied large amounts of address data to the stack,
2202 even after the fix for CVE-2013-4458 has been applied, potentially
2203 resulting in a stack overflow. getaddrinfo now uses a heap allocation
2204 instead. Reported by Michael Petlan. (CVE-2016-3706)
2206 * The glob function suffered from a stack-based buffer overflow when it was
2207 called with the GLOB_ALTDIRFUNC flag and encountered a long file name.
2208 Reported by Alexander Cherepanov. (CVE-2016-1234)
2210 * The Sun RPC UDP client could exhaust all available stack space when
2211 flooded with crafted ICMP and UDP messages. Reported by Aldy Hernandez'
2212 alloca plugin for GCC. (CVE-2016-4429)
2214 * The IPv6 name server management code in libresolv could result in a memory
2215 leak for each thread which is created, performs a failing naming lookup,
2216 and exits. Over time, this could result in a denial of service due to
2217 memory exhaustion. Reported by Matthias Schiffer. (CVE-2016-5417)
2219 The following bugs are resolved with this release:
2221 [1170] localedata: ne_NP: update Nepali locale definition file
2222 [3629] manual: stpcpy description in string.texi refers to MS-DOG instead
2224 [6527] malloc: [powerpc] Malloc alignment insufficient for PowerPC
2225 [6796] math: fdim() does not set errno on overflow
2226 [10354] libc: posix_spawn should use vfork() in more cases than presently
2227 [11213] localedata: localedata: add copyright disclaimer to locale files
2228 [12143] localedata: chr_US: new Cherokee locale
2229 [12450] localedata: sgs_LT: new locale
2230 [12676] localedata: ln_CD: new locale
2231 [13237] localedata: LC_ADDRESS.country_name: update all locales w/latest
2233 [13304] math: fma, fmaf, fmal produce wrong results
2234 [14259] build: --localedir arg to configure is ignored
2235 [14499] nptl: Does posix_spawn invoke atfork handlers / use vfork?
2236 [14750] libc: Race condition in posix_spawn vfork usage vs signal handlers
2237 [14934] localedata: es_CL: wrong first weekday chilean locale
2238 [15262] localedata: LC_MESSAGES.yesexpr/noexpr: inconsistent use of
2240 [15263] localedata: LC_MESSAGES.yesexpr/noexpr: inconsistent use of 1/0
2242 [15264] localedata: LC_MESSAGES.yesstr/nostr: lacking in many locales
2243 [15368] nptl: raise() is not async-signal-safe
2244 [15479] math: ceil, floor, round and trunc raise inexact exception
2245 [15578] localedata: kk_KZ: various updates
2246 [16003] localedata: pap_AN: punt old locale
2247 [16137] localedata: iw_IL: punt old locale
2248 [16190] localedata: eo: new esperanto locale
2249 [16374] localedata: lv_LV: change currency symbol in LC_MONETARY to euro
2250 [16742] malloc: race condition: pthread_atfork() called before first
2251 malloc() results in unexpected locking behaviour/deadlocks
2252 [16975] localedata: LC_MESSAGES.yesexpr/noexpr: revisit capitalization in
2254 [16983] localedata: postal_fmt does not allow %l and %n modifiers
2255 [17565] localedata: pt_PT: wrong (work-)week start
2256 [17899] math: [powerpc] floorl returns negative zero with FE_DOWNWARD
2257 [17950] build: Build fails with -msse
2258 [18205] localedata: be_BY*: wrong first_weekday and first_workday
2259 [18433] libc: posix_spawn does not return correctly upon failure to
2261 [18453] localedata: charmaps/IBM875: incorrect codes
2262 [18712] string: bits/string2.h incompatible with -O2 -Werror=packed
2264 [18896] localedata: he_IL: improvements for currency
2265 [18911] localedata: ro_RO: Correcting week day name for "Tuesday" in
2266 Romanian locale data
2267 [18960] locale: s390: _nl_locale_subfreeres uses larl opcode on misaligned
2269 [19056] libc: Deprecate readdir_r
2270 [19133] localedata: pt_*: days & months should be lowercase in Portuguese
2272 [19198] localedata: nl_NL: small improvements for Dutch locales
2273 [19257] network: Per-thread memory leak in __res_vinit with IPv6
2274 nameservers (CVE-2016-5417)
2275 [19269] build: tst-audit4 and tst-audit10 failures with gcc-6 on non avx
2277 [19400] locale: Language missing in "iso-639.def", trivial fix in
2279 [19431] malloc: Deadlock between fflush, getdelim, and fork
2280 [19505] libc: Incorrect file descriptor validity checks in
2281 posix_spawn_file_actions_add{open,close,dup2}
2282 [19509] dynamic-link: dlsym, dlvsym do not report errors through dlerror
2283 when using RTLD_NEXT
2284 [19512] locale: Stale `#ifndef HAVE_BUILTIN_EXPECT' in
2285 `intl/{gettextP,loadinfo}.h'
2286 [19534] libc: execle, execlp may use malloc
2287 [19568] localedata: *_CH: Swiss locales have inconsistent start of week
2288 [19573] network: res_nclose and __res_maybe_init disagree about name
2289 server initialization, breaking Hesiod
2290 [19575] localedata: Status of GB18030 tables
2291 [19581] localedata: sr_* date_fmt string contains additional newline
2292 [19583] string: SSSE3_Fast_Copy_Backward flag needs to be enabled for AMD
2294 [19592] math: [ldbl-128ibm] ceill incorrect in non-default rounding modes
2295 [19593] math: [ldbl-128ibm] truncl incorrect in non-default rounding modes
2296 [19594] math: [ldbl-128ibm] roundl incorrect in non-default rounding modes
2297 [19595] math: [ldbl-128ibm] fmodl incorrect for results in subnormal
2299 [19602] math: [ldbl-128ibm] fmodl handling of equal arguments with low
2301 [19603] math: [ldbl-128ibm] remainderl, remquol incorrect sign handling in
2303 [19610] dynamic-link: ldconfig -X removes stale symbolic links
2304 [19613] libc: s390x (64 bit) macro expansion WCOREDUMP and others
2305 [19633] locale: strfmon_l applies global locale to number formatting
2306 [19642] network: Memory leak in getnameinfo
2307 [19648] libc: test-skeleton.c: Do not set RLIMIT_DATA
2308 [19653] libc: Potential for NULL pointer dereference (CWE-476) in
2310 [19654] math: [x86_64] Need testcase for BZ #19590 fix
2311 [19671] localedata: Missing Sanity Check for malloc() in 'tst-fmon.c' &
2313 [19674] math: [ldbl-128ibm] powl incorrect overflow handling
2314 [19677] math: [ldbl-128ibm] remainderl equality test incorrect for zero
2316 [19678] math: [ldbl-128ibm] nextafterl, nexttowardl incorrect sign of zero
2318 [19679] dynamic-link: gcc-4.9.3 C++ exception handling broken due to
2320 [19726] locale: Converting UCS4LE to INTERNAL with iconv() does not update
2321 pointers and lengths in error-case.
2322 [19727] locale: Converting from/to UTF-xx with iconv() does not always
2323 report errors on UTF-16 surrogates values.
2324 [19755] nscd: nscd assertion failure in gc
2325 [19758] dynamic-link: Typo in EXTRA_LD_ENVVARS for x86-64
2326 [19759] libc: mempcpy shouldn't be inlined
2327 [19762] dynamic-link: HAS_CPU_FEATURE/HAS_ARCH_FEATURE are easy to misuse
2328 [19765] libc: s390 needs an optimized mempcpy
2329 [19779] glob: glob: buffer overflow with GLOB_ALTDIRFUNC due to incorrect
2330 NAME_MAX limit assumption (CVE-2016-1234)
2331 [19783] build: benchtests don't support --enable-hardcoded-path-in-tests
2332 [19787] network: Missing and incorrect truncation checks in getnameinfo
2333 [19790] math: [ldbl-128ibm] nearbyintl incorrect in non-default rounding
2335 [19791] network: Assertion failure in res_query.c with un-connectable name
2337 [19792] libc: MIPS: backtrace yields infinite backtrace with makecontext
2338 [19822] math: libm.so install clobbers old version
2339 [19825] network: resolv: send_vc can return uninitialized data in second
2340 response to getaddrinfo
2341 [19830] network: nss_dns: should check RDATA length against buffer length
2342 [19831] network: nss_dns: getaddrinfo returns uninitialized data when
2343 confronted with A/AAAA records of invalid size
2344 [19837] nss: nss_db: No retries for some long lines with a larger buffer
2345 [19848] math: powl(10,n) for n=-4,-5,-6,-7 is off by more than 1 ULP
2346 [19853] stdio: Printing IBM long double in decimal with high precision is
2348 [19860] build: x86_64: compile errors for tst-audit10 and tst-auditmod10b
2349 [19861] nptl: libpthread IFUNC resolver for fork can lead to crash
2350 [19862] network: resolv, nss_dns: Remove remaining logging of unexpected
2352 [19865] network: Assertion failure or memory leak in
2353 _nss_dns_getcanonname_r
2354 [19868] network: nss_dns: netent code does not skip over non-PTR records
2355 [19879] network: nss_dns: Stack overflow in getnetbyname implementation
2357 [19881] string: Improve x86-64 memset
2358 [19907] string: Incorrect memcpy tests
2359 [19916] dynamic-link: S390: fprs/vrs are not saved/restored while
2361 [19925] libc: termios.h XCASE namespace
2362 [19928] string: memmove-vec-unaligned-erms.S is slow with large data size
2363 [19929] libc: limits.h NL_NMAX namespace
2364 [19931] stdio: Memory leak in vfprintf
2365 [19957] libc: clone(CLONE_VM) access invalid parent memory
2366 [19963] localedata: en_IL: New locale
2367 [19989] stdio: stdio.h cuserid namespace
2368 [19994] network: getaddrinfo does not restore RES_USE_INET6 flag in
2370 [19996] locale: langinfo.h nl_langinfo_l namespace
2371 [20005] stdio: fflush on a file opened with fmemopen resets position to 0
2372 [20010] network: getaddrinfo: Stack overflow in hostent translation
2374 [20012] stdio: libio: fmemopen append mode failure
2375 [20014] stdio: stdio.h namespace for pre-threads POSIX
2376 [20017] network: resolv: Use gmtime_r instead of gmtime in p_secstodate
2377 [20023] libc: fcntl.h timespec namespace
2378 [20024] math: [x86_64] vectorized sincos trashes the stack
2379 [20031] network: nss_hesiod: Heap overflow in get_txt_records
2380 [20041] time: sys/time.h timespec namespace
2381 [20043] libc: unistd.h missing cuserid for UNIX98 and before
2382 [20044] libc: unistd.h missing pthread_atfork for UNIX98
2383 [20051] libc: ttyslot in wrong header under wrong conditions
2384 [20054] libc: gethostname not declared for XPG4
2385 [20055] libc: termios.h missing tcgetsid for XPG4
2386 [20072] dynamic-link: x86 init_cpu_features is called twice in static
2388 [20073] libc: sys/stat.h fchmod namespace
2389 [20074] libc: stdlib.h rand_r namespace
2390 [20076] libc: sys/stat.h missing S_IFSOCK, S_ISSOCK for XPG4
2391 [20094] libc: stdlib.h should not declare grantpt, ptsname, unlockpt for
2393 [20111] libc: struct sockaddr_storage cannot be aggregate-copied
2394 [20112] network: sunrpc: stack (frame) overflow in Sun RPC clntudp_call
2396 [20115] string: Extra alignment in memset-vec-unaligned-erms.S
2397 [20119] libc: Wrong mask for processors level type from CPUID
2398 [20139] dynamic-link: Upper part of zmm is zeroed if Glibc is built with
2399 AS not supporting AVX512
2400 [20151] math: [ldbl-128/ldbl-128ibm] j0l, j1l, y0l, y1l return sNaN for
2402 [20153] math: [ldbl-128ibm] sqrtl (sNaN) returns sNaN
2403 [20156] math: [ldbl-128ibm] ceill, rintl etc. return sNaN for sNaN
2405 [20157] math: [powerpc] fabsl (sNaN) wrongly raises "invalid"
2406 [20160] math: [powerpc] ceil, rint etc. return sNaN for sNaN input
2407 [20178] libc: posix_spawn{p} should not call exit
2408 [20191] stdio: libio: vtables hardening
2409 [20195] string: FMA4 detection requires CPUID execution with register
2411 [20198] libc: quick_exit incorrectly destroys C++11 thread objects.
2412 [20205] math: [i386/x86_64] nextafterl incorrect incrementing negative
2414 [20212] math: acos (sNaN) returns sNaN
2415 [20213] math: asin (sNaN) returns sNaN
2416 [20214] network: Linux header sync with linux/in6.h and ipv6.h again.
2417 [20218] math: [i386] asinhl (sNaN) returns sNaN
2418 [20219] math: [i386] atanhl (sNaN) returns sNaN
2419 [20222] stdio: fopencookie: Mangle function pointers
2420 [20224] math: [i386] cbrtl (sNaN) returns sNaN
2421 [20225] math: ldexp, scalbn, scalbln return sNaN for sNaN input
2422 [20226] math: [i386/x86_64] expl, exp10l, expm1l return sNaN for sNaN
2424 [20227] math: [i386/x86_64] logl (sNaN) returns sNaN
2425 [20228] math: [i386/x86_64] log10l (sNaN) returns sNaN
2426 [20229] math: [i386/x86_64] log1pl (sNaN) returns sNaN
2427 [20232] math: [ldbl-128] expm1l (sNaN) returns sNaN
2428 [20233] math: [ldbl-128ibm] expm1l (sNaN) returns sNaN
2429 [20234] math: [ldbl-128ibm] log1pl (sNaN) returns sNaN
2430 [20235] math: [i386/x86_64] log2l (sNaN) returns sNaN
2431 [20237] nss: nss_db: get*ent segfaults without preceding set*ent
2432 [20240] math: modf (sNaN) returns sNaN
2433 [20248] libc: debug/tst-longjump_chk2 calls printf from a signal handler
2434 [20250] math: frexp (sNaN) returns sNaN
2435 [20252] math: atan2 (sNaN, qNaN) fails to raise "invalid"
2436 [20255] math: [i386] fdim, fdimf return with excess range and precision /
2438 [20256] math: [i386/x86_64] fdiml returns sNaN for sNaN input
2439 [20260] string: ../sysdeps/x86/bits/string.h:1092:3: error: array
2440 subscript is below array bounds [-Werror=array-bounds]
2441 [20262] nis: _nss_nis_initgroups_dyn always returns NSS_STATUS_NOTFOUND
2442 [20263] nptl: robust mutex deadlocks if other thread requests timedlock
2444 [20277] libc: $dp is not initialized correctly in sysdeps/hppa/start.S
2445 [20284] malloc: malloc: Corrupt arena avoidance causes unnecessary mmap
2447 [20296] math: [i386/x86_64] scalbl returns sNaN for sNaN input, missing
2448 "invalid" exceptions
2449 [20314] nptl: make[4]: *** [/usr/include/stdlib.h] Error 1
2450 [20316] localedata: id_ID: Februari instead of Pebruari
2451 [20327] string: POWER8 strcasecmp returns incorrect result
2452 [20347] math: Failure: Test: j0_downward (0xap+0)
2453 [20348] libc: FAIL: misc/tst-preadvwritev64
2454 [20349] libc: 64-bit value is passed differently in p{readv,writev}{64}
2455 [20350] libc: There is no test for p{read,write}64
2456 [20357] math: Incorrect cos result for 1.5174239687223976
2457 [20384] build: Don't run libmvec-sincos-avx* tests on non avx machines
2461 * Unicode 8.0.0 Support: Character encoding, character type info, and
2462 transliteration tables are all updated to Unicode 8.0.0, using new
2463 and/or improved generator scripts contributed by Mike FABIAN (Red Hat).
2464 These updates cause user visible changes, such as the fixes for bugs
2465 89, 16061, and 18568.
2467 * sched_setaffinity, pthread_setaffinity_np no longer attempt to guess the
2468 kernel-internal CPU set size. This means that requests that change the
2469 CPU affinity which failed before (for example, an all-ones CPU mask) will
2470 now succeed. Applications that need to determine the effective CPU
2471 affinities need to call sched_getaffinity or pthread_getaffinity_np after
2472 setting it because the kernel can adjust it (and the previous size check
2473 would not detect this in the majority of cases).
2475 * The fts.h header can now be used with -D_FILE_OFFSET_BITS=64. With LFS
2476 the following new symbols are used: fts64_children, fts64_close,
2477 fts64_open, fts64_read and fts64_set.
2479 * getaddrinfo now detects certain invalid responses on an internal netlink
2480 socket. If such responses are received, an affected process will
2481 terminate with an error message of "Unexpected error <number> on netlink
2482 descriptor <number>" or "Unexpected netlink response of size <number> on
2483 descriptor <number>". The most likely cause for these errors is a
2484 multi-threaded application which erroneously closes and reuses the netlink
2485 file descriptor while it is used by getaddrinfo.
2487 * A defect in the malloc implementation, present since glibc 2.15 (2012) or
2488 glibc 2.10 via --enable-experimental-malloc (2009), could result in the
2489 unnecessary serialization of memory allocation requests across threads.
2490 The defect is now corrected. Users should see a substantial increase in
2491 the concurent throughput of allocation requests for applications which
2492 trigger this bug. Affected applications typically create and
2493 destroy threads frequently. (Bug 19048 was reported and analyzed by
2496 * There is now a --disable-timezone-tools configure option for disabling the
2497 building and installing of the timezone related utilities (zic, zdump, and
2498 tzselect). This is useful for people who build the timezone data and code
2499 independent of the GNU C Library.
2501 * The obsolete header <regexp.h> has been removed. Programs that require
2502 this header must be updated to use <regex.h> instead.
2504 * The obsolete functions bdflush, create_module, get_kernel_syms,
2505 query_module and uselib are no longer available to newly linked binaries;
2506 the header <sys/kdaemon.h> has been removed. These functions and header
2507 were specific to systems using the Linux kernel and could not usefully be
2508 used with the GNU C Library on systems with version 2.6 or later of the
2511 * Optimized string, wcsmbs and memory functions for IBM z13.
2512 Implemented by Stefan Liebler.
2514 * Newly linked programs that define a variable called signgam will no longer
2515 have it set by the lgamma, lgammaf and lgammal functions. Programs that
2516 require signgam to be set by those functions must ensure that they use the
2517 variable provided by the GNU C Library and declared in <math.h>, without
2518 defining their own copy.
2520 * The minimum GCC version that can be used to build this version of the GNU
2521 C Library is GCC 4.7. Older GCC versions, and non-GNU compilers, can
2522 still be used to compile programs using the GNU C Library.
2524 Security related changes:
2526 * An out-of-bounds value in a broken-out struct tm argument to strftime no
2527 longer causes a crash. Reported by Adam Nielsen. (CVE-2015-8776)
2529 * The LD_POINTER_GUARD environment variable can no longer be used to disable
2530 the pointer guard feature. It is always enabled. Previously,
2531 LD_POINTER_GUARD could be used to disable security hardening in binaries
2532 running in privileged AT_SECURE mode. Reported by Hector Marco-Gisbert.
2535 * An integer overflow in hcreate and hcreate_r could lead to an
2536 out-of-bounds memory access. Reported by Szabolcs Nagy. (CVE-2015-8778)
2538 * The catopen function no longer has unbounded stack usage. Reported by
2539 Max. (CVE-2015-8779)
2541 * The nan, nanf and nanl functions no longer have unbounded stack usage
2542 depending on the length of the string passed as an argument to the
2543 functions. Reported by Joseph Myers. (CVE-2014-9761)
2545 * A stack-based buffer overflow was found in libresolv when invoked from
2546 libnss_dns, allowing specially crafted DNS responses to seize control
2547 of execution flow in the DNS client. The buffer overflow occurs in
2548 the functions send_dg (send datagram) and send_vc (send TCP) for the
2549 NSS module libnss_dns.so.2 when calling getaddrinfo with AF_UNSPEC
2550 family. The use of AF_UNSPEC triggers the low-level resolver code to
2551 send out two parallel queries for A and AAAA. A mismanagement of the
2552 buffers used for those queries could result in the response of a query
2553 writing beyond the alloca allocated buffer created by
2554 _nss_dns_gethostbyname4_r. Buffer management is simplified to remove
2555 the overflow. Thanks to the Google Security Team and Red Hat for
2556 reporting the security impact of this issue, and Robert Holiday of
2557 Ciena for reporting the related bug 18665. (CVE-2015-7547)
2559 The following bugs are resolved with this release:
2561 [89] localedata: Locales nb_NO and nn_NO should transliterate æøå
2562 [887] math: Math library function "logb" and "nextafter" inconsistent
2563 [2542] math: Incorrect return from float gamma (-0X1.FA471547C2FE5P+1)
2564 [2543] math: Incorrect return from float gamma (-0X1.9260DCP+1)
2565 [2558] math: Incorrect return from double gamma (-0X1.FA471547C2FE5P+1)
2566 [2898] libc: [improve] warning: the use of `mktemp' is dangerous, better
2568 [4404] localedata: German translation of "Alarm clock" is misleading
2569 [6799] math: nextafter() and nexttoward() doen't set errno on
2570 overflow/underflow errors
2571 [6803] math: scalb(), scalbln(), scalbn() do not set errno on
2573 [10432] nis: _nss_nis_setnetgrent assertion failure
2574 [11460] libc: fts has no LFS support
2575 [12926] network: getaddrinfo()/make_request() may spin forever
2576 [13065] nptl: Race condition in pthread barriers
2577 [13690] nptl: pthread_mutex_unlock potentially cause invalid access
2578 [14341] dynamic-link: Dynamic linker crash when DT_JMPREL and DT_REL{,A}
2580 [14551] math: [ldbl-128ibm] strtold overflow handling for IBM long double
2581 [14912] libc: Rename non-installed bits/*.h headers
2582 [15002] libc: Avoid undefined behavior in posix_fallocate overflow check
2583 [15367] math: Let gcc use __builtin_isinf
2584 [15384] math: One constant fewer in ieee754/dbl-64/wordsize-64/s_finite.c
2585 [15421] math: lgamma wrongly sets signgam for ISO C
2586 [15470] math: [arm] On ARM llrintl() and llroundl() do not raise
2587 FE_INVALID with argument out of range
2588 [15491] math: [i386/x86_64] x86 nearbyint implementations wrongly clear
2590 [15786] dynamic-link: ifunc resolver functions can smash function
2592 [15918] math: Unnecessary check for equality in hypotf()
2593 [16061] localedata: Review / update transliteration data
2594 [16068] math: [i386/x86_64] x86 and x86_64 fesetenv exclude state they
2596 [16141] time: strptime %z offset restriction
2597 [16171] math: drem should be alias of remainder
2598 [16296] math: fegetround is pure?
2599 [16347] math: [ldbl-128ibm] ldbl-128/e_lgammal_r.c may not be suitable.
2600 [16364] libc: sleep may leave SIGCHLD blocked on sync cancellation on
2602 [16399] math: [mips] lrint / llrint / lround / llround missing exceptions
2603 [16415] math: Clean up ldbl-128 / ldbl-128ibm expm1l for large positive
2605 [16422] math: [powerpc] math-float, math-double failing llrint tests with
2606 "Exception "Inexact" set" on ppc32
2607 [16495] localedata: nl_NL: date_fmt: shuffle year/month around
2608 [16517] math: Missing underflow exception from tanf/tan/tanl
2609 [16519] math: Missing underflow exception from sinhf
2610 [16520] math: Missing underflow exception from tanhf
2611 [16521] math: Missing underflow exception from exp2
2612 [16620] math: [ldbl-128ibm] exp10l spurious overflows / bad directed
2614 [16734] stdio: fopen calls mmap to allocate its buffer
2615 [16961] math: nan function incorrect handling of bad sequences
2616 [16962] math: nan function unbounded stack allocation (CVE-2014-9761)
2617 [16973] localedata: Fix lang_lib/lang_term as per ISO 639-2
2618 [16985] locale: localedef: confusing error message when opening output
2620 [17118] math: ctanh(INFINITY + 2 * I) returns incorrect value
2621 [17197] locale: Redundant shift character in iconv conversion output at
2623 [17243] libc: trunk/posix/execl.c:53: va_args problem ?
2624 [17244] libc: trunk/sysdeps/unix/sysv/linux/semctl.c:116: va_args muxup ?
2625 [17250] dynamic-link: static linking breaks nss loading
2626 (getaddrinfo/getpwnam/etc...)
2627 [17404] libc: atomic_exchange_rel lacking a barrier on MIPS16, GCC before
2629 [17441] math: isnan() should use __builtin_isnan() in GCC
2630 [17514] nptl: Assert failure unlocking ERRORCHECK mutex after timedlock
2631 (related to lock elision)
2632 [17787] manual: Exponent on page 324 of the PDF ends prematurely
2633 [17886] time: strptime should be able to parse "Z" as a timezone with %z
2634 [17887] time: strptime should be able to parse "+01:00" style timezones
2635 [17905] libc: catopen() Multiple unbounded stack allocations
2637 [18084] libc: backtrace (..., 0) dumps core on x86
2638 [18086] libc: nice() sets errno to 0 on success
2639 [18240] libc: hcreate, hcreate_r should fail with ENOMEM if element count
2640 is too large (CVE-2015-8778)
2641 [18251] dynamic-link: SONAME missing when audit modules provides path
2642 [18265] libc: add attributes for wchar string and memory functions
2643 [18370] math: csqrt missing underflows
2644 [18421] libc: [hppa] read-only segment has dynamic relocations
2645 [18472] libc: Obsolete syscall wrappers should be compat symbols
2646 [18480] libc: hppa glibc miscompilation in sched_setaffinity()
2647 [18491] localedata: Update tr_TR LC_CTYPE as part of Unicode updates
2648 [18525] localedata: Remove locale timezone information
2649 [18560] libc: [powerpc] spurious bits/ipc.h definitions
2650 [18568] localedata: Update locale data to Unicode 8.0
2651 [18589] locale: sort-test.sh fails at random
2652 [18595] math: ctan, ctanh missing underflows
2653 [18604] libc: assert macro-expands its argument
2654 [18610] math: S390: fetestexcept() reports any exception if DXC-code
2655 contains a vector instruction exception.
2656 [18611] math: j1, jn missing errno setting on underflow
2657 [18618] localedata: sync Chechen locale definitions with other *_RU
2659 [18647] math: powf(-0x1.000002p0, 0x1p30) returns 0 instead of +inf
2660 [18661] libc: Some x86-64 assembly codes don't align stack to 16 bytes
2661 [18665] network: In send_dg, the recvfrom function is NOT always using the
2662 buffer size of a newly created buffer (CVE-2015-7547)
2663 [18674] libc: [i386] trunk/sysdeps/i386/tst-auditmod3b.c:84: possible
2665 [18675] libc: fpathconf(_PC_NAME_MAX) fails against large filesystems for
2667 [18681] libc: regexp.h is obsolete and buggy, and should be desupported
2668 [18699] math: tilegx cproj() for various complex infinities does not yield
2670 [18724] libc: Harden put*ent functions against data injection
2671 [18743] nptl: PowerPC: findutils testcase fails with --enable-lock-elision
2672 [18755] build: build errors with -DNDEBUG
2673 [18757] stdio: fmemopen fails to set errno on failure
2674 [18778] dynamic-link: ld.so crashes if failed dlopen causes libpthread to
2676 [18781] libc: openat64 lacks O_LARGEFILE
2677 [18787] libc: [hppa] sysdeps/unix/sysv/linux/hppa/bits/atomic.h:71:6:
2678 error: can’t find a register in class ‘R1_REGS’ while reloading ‘asm’
2679 [18789] math: [ldbl-128ibm] sinhl inaccurate near 0
2680 [18790] math: [ldbl-128ibm] tanhl inaccurate
2681 [18795] libc: stpncpy fortification misses buffer lengths that are
2682 statically too large
2683 [18796] build: build fails for --disable-mathvec
2684 [18803] math: hypot missing underflows
2685 [18820] stdio: fmemopen may leak memory on failure
2686 [18823] math: csqrt spurious underflows
2687 [18824] math: fma spurious underflows
2688 [18825] math: pow missing underflows
2689 [18857] math: [ldbl-128ibm] nearbyintl wrongly uses signaling comparisons
2690 [18868] nptl: pthread_barrier_init typo has in-theory-undefined behavior
2691 [18870] build: sem_open.c fails to compile with missing symbol
2693 [18872] stdio: Fix memory leak in printf_positional
2694 [18873] libc: posix_fallocate overflow check ineffective
2695 [18875] math: Excess precision leads incorrect libm
2696 [18877] libc: arm: mmap offset regression
2697 [18887] libc: memory corruption when using getmntent on blank lines
2698 [18918] localedata: hu_HU: change time to HH:MM:SS format
2699 [18921] libc: Regression: extraneous stat() and fstat() performed by
2701 [18928] dynamic-link: LD_POINTER_GUARD is not ignored for privileged
2702 binaries (CVE-2015-8777)
2703 [18951] math: tgamma missing underflows
2704 [18952] math: [ldbl-128/ldbl-128ibm] lgammal spurious "invalid", incorrect
2706 [18953] localedata: lt_LT: change currency symbol to the euro
2707 [18956] math: powf inaccuracy
2708 [18961] math: [i386] exp missing underflows
2709 [18966] math: [i386] exp10 missing underflows
2710 [18967] math: math.h XSI POSIX namespace (gamma, isnan, scalb)
2711 [18969] build: multiple string test failures due to missing locale
2713 [18970] libc: Reference of pthread_setcancelstate in libc.a
2714 [18977] math: float / long double Bessel functions not in XSI POSIX
2715 [18980] math: i386 libm functions return with excess range and precision
2716 [18981] math: i386 scalb*, ldexp return with excess range and precision
2717 [18982] stdio: va_list and vprintf
2718 [18985] time: Passing out of range data to strftime() causes a segfault
2720 [19003] math: [x86_64] fma4 version of pow inappropriate contraction
2721 [19007] libc: FAIL: elf/check-localplt with -z now and binutils 2.26
2722 [19012] locale: iconv_open leaks memory on error path
2723 [19016] math: clog, clog10 inaccuracy
2724 [19018] nptl: Mangle function pointers in tls_dtor_list
2725 [19032] math: [i386] acosh (-qNaN) spurious "invalid" exception
2726 [19046] math: ldbl-128 / ldbl-128ibm lgamma bad overflow handling
2727 [19048] malloc: malloc: arena free list can become cyclic, increasing
2729 [19049] math: [powerpc] erfc incorrect zero sign
2730 [19050] math: [powerpc] log* incorrect zero sign
2731 [19058] math: [x86_64] Link fail with -fopenmp and -flto
2732 [19059] math: nexttoward overflow incorrect in non-default rounding modes
2733 [19071] math: ldbl-96 lroundl incorrect just below powers of 2
2734 [19074] network: Data race in _res_hconf_reorder_addrs
2735 [19076] math: [ldbl-128ibm] log1pl (-1) wrong sign of infinity
2736 [19077] math: [ldbl-128ibm] logl (1) incorrect sign of zero result
2737 [19078] math: [ldbl-128ibm] expl overflow incorrect in non-default
2739 [19079] math: dbl-64/wordsize-64 lround based on llround incorrect for
2741 [19085] math: ldbl-128 lrintl, lroundl missing exceptions for 32-bit long
2742 [19086] manual: posix_fallocate64 documented argument order is wrong.
2743 [19088] math: lround, llround missing exceptions close to overflow
2745 [19094] math: lrint, llrint missing exceptions close to overflow threshold
2746 [19095] math: dbl-64 lrint incorrect for 64-bit long
2747 [19122] dynamic-link: Unnecessary PLT relocations in librtld.os
2748 [19124] dynamic-link: ld.so failed to build with older assmebler
2749 [19125] math: [powerpc32] llroundf, llround incorrect exceptions
2750 [19129] dynamic-link: [arm] Concurrent lazy TLSDESC resolution can crash
2751 [19134] math: [powerpc32] lround, lroundf spurious exceptions
2752 [19137] libc: i386/epoll_pwait.S doesn't support cancellation
2753 [19143] nptl: Remove CPU set size checking from sched_setaffinity,
2754 pthread_setaffinity_np
2755 [19156] math: [ldbl-128] j0l spurious underflows
2756 [19164] nptl: tst-getcpu fails with many possible CPUs
2757 [19168] math: math/test-ildoubl and math/test-ldouble failure
2758 [19174] nptl: PowerPC: TLE enabled pthread mutex performs poorly.
2759 [19178] dynamic-link: ELF_RTYPE_CLASS_EXTERN_PROTECTED_DATA confuses
2761 [19181] math: [i386/x86_64] fesetenv (FE_DFL_ENV), fesetenv
2762 (FE_NOMASK_ENV) do not clear SSE exceptions
2763 [19182] malloc: malloc deadlock between ptmalloc_lock_all and
2764 _int_new_arena/reused_arena
2765 [19189] math: [ldbl-128] log1pl (-qNaN) spurious "invalid" exception
2766 [19201] math: dbl-64 remainder incorrect sign of zero result
2767 [19205] math: bits/math-finite.h conditions do not match math.h and
2769 [19209] math: bits/math-finite.h wrongly maps ldexp to scalbn
2770 [19211] math: lgamma functions do not set signgam for -ffinite-math-only
2771 for C99-based standards
2772 [19212] libc: features.h not -Wundef clean
2773 [19213] math: [i386/x86_64] log* (1) incorrect zero sign for -ffinite-
2775 [19214] libc: Family and model identification for AMD CPU's are incorrect.
2776 [19219] libc: GLIBC build fails for ia64 with missing __nearbyintl
2777 [19228] math: [powerpc] nearbyint wrongly clears "inexact", leaves traps
2779 [19235] math: [powerpc64] lround, lroundf, llround, llroundf spurious
2780 "inexact" exceptions
2781 [19238] math: [powerpc] round, roundf spurious "inexact" for integer
2783 [19242] libc: strtol incorrect in Turkish locales
2784 [19243] malloc: reused_arena can pick an arena on the free list, leading
2785 to an assertion failure and reference count corruption
2786 [19253] time: tzset() ineffective when temporary TZ did not include DST
2788 [19266] math: strtod ("NAN(I)") incorrect in Turkish locales
2789 [19270] math: [hppa] Shared libm missing __isnanl
2790 [19285] libc: [hppa] sysdeps/unix/sysv/linux/hppa/bits/mman.h: missing
2791 MAP_HUGETLB and MAP_STACK defines
2792 [19313] nptl: Wrong __cpu_mask for x32
2793 [19347] libc: grantpt: try to force a specific gid even without pt_chown
2794 [19349] math: [ldbl-128ibm] tanhl inaccurate for small arguments
2795 [19350] math: [ldbl-128ibm] sinhl spurious overflows
2796 [19351] math: [ldbl-128ibm] logl inaccurate near 1
2797 [19363] time: x32: times() return value wrongly truncates/sign extends
2799 [19367] dynamic-link: Improve branch prediction on Silvermont
2800 [19369] network: Default domain name not reset by res_ninit when "search"
2801 / "domain" entry is removed from resolv.conf
2802 [19375] math: powerpc: incorrect results for POWER7 logb with negative
2804 [19385] localedata: bg_BG: time separator should be colon, not comma
2805 [19408] libc: linux personality syscall wrapper may erroneously return an
2806 error on 32-bit architectures
2807 [19415] libc: dladdr returns wrong names on hppa
2808 [19432] libc: iconv rejects redundant escape sequences in IBM900, IBM903,
2809 IBM905, IBM907, and IBM909
2810 [19439] math: Unix98 isinf and isnan functions conflict with C++11
2811 [19443] build: build failures with -DDEBUG
2812 [19451] build: Make check fails on test-double-vlen2
2813 [19462] libc: Glibc failed to build with -Os
2814 [19465] math: Wrong code with -Os
2815 [19466] time: time/tst-mktime2.c is compiled into an infinite loop with
2817 [19467] string: Fast_Unaligned_Load needs to be enabled for Excavator core
2819 [19475] libc: Glibc 2.22 doesn't build on sparc [PATCH]
2820 [19486] math: S390: Math tests fail with "Exception Inexact set".
2821 [19529] libc: [ARM]: FAIL: stdlib/tst-makecontext
2822 [19550] libc: [mips] mmap negative offset handling inconsistent with other
2824 [19590] math: Fail to build shared objects that use libmvec.so functions.
2828 * The following bugs are resolved with this release:
2830 438, 2981, 4719, 6544, 6792, 11216, 12836, 13028, 13064, 13151, 13152,
2831 14094, 14113, 14292, 14841, 14906, 14958, 15319, 15467, 15790, 15969,
2832 16159, 16339, 16350, 16351, 16352, 16353, 16361, 16512, 16526, 16538,
2833 16559, 16560, 16704, 16783, 16850, 17053, 17090, 17195, 17269, 17293,
2834 17322, 17403, 17475, 17523, 17542, 17569, 17581, 17588, 17596, 17620,
2835 17621, 17628, 17631, 17692, 17711, 17715, 17776, 17779, 17792, 17833,
2836 17836, 17841, 17912, 17916, 17930, 17932, 17944, 17949, 17964, 17965,
2837 17967, 17969, 17977, 17978, 17987, 17991, 17996, 17998, 17999, 18007,
2838 18019, 18020, 18029, 18030, 18032, 18034, 18036, 18038, 18039, 18042,
2839 18043, 18046, 18047, 18049, 18068, 18078, 18080, 18093, 18100, 18104,
2840 18110, 18111, 18116, 18125, 18128, 18134, 18138, 18185, 18196, 18197,
2841 18206, 18210, 18211, 18217, 18219, 18220, 18221, 18234, 18244, 18245,
2842 18247, 18287, 18319, 18324, 18333, 18346, 18371, 18383, 18397, 18400,
2843 18409, 18410, 18412, 18418, 18422, 18434, 18444, 18457, 18468, 18469,
2844 18470, 18479, 18483, 18495, 18496, 18497, 18498, 18502, 18507, 18508,
2845 18512, 18513, 18519, 18520, 18522, 18527, 18528, 18529, 18530, 18532,
2846 18533, 18534, 18536, 18539, 18540, 18542, 18544, 18545, 18546, 18547,
2847 18549, 18553, 18557, 18558, 18569, 18583, 18585, 18586, 18592, 18593,
2848 18594, 18602, 18612, 18613, 18619, 18633, 18635, 18641, 18643, 18648,
2849 18657, 18676, 18694, 18696, 18887.
2851 * Cache information can be queried via sysconf() function on s390 e.g. with
2852 _SC_LEVEL1_ICACHE_SIZE as argument.
2854 * A buffer overflow in gethostbyname_r and related functions performing DNS
2855 requests has been fixed. If the NSS functions were called with a
2856 misaligned buffer, the buffer length change due to pointer alignment was
2857 not taken into account. This could result in application crashes or,
2858 potentially arbitrary code execution, using crafted, but syntactically
2859 valid DNS responses. (CVE-2015-1781)
2861 * The time zone file parser has been made more robust against crafted time
2862 zone files, avoiding heap buffer overflows related to the processing of
2863 the tzh_ttisstdcnt and tzh_ttisgmtcnt fields, and a stack overflow due to
2864 large time zone data files. Overly long time zone specifiers in the TZ
2865 variable no longer result in stack overflows and crashes.
2867 * A powerpc and powerpc64 optimization for TLS, similar to TLS descriptors
2868 for LD and GD on x86 and x86-64, has been implemented. You will need
2869 binutils-2.24 or later to enable this optimization.
2871 * Character encoding and ctype tables were updated to Unicode 7.0.0, using
2872 new generator scripts contributed by Pravin Satpute and Mike FABIAN (Red
2873 Hat). These updates cause user visible changes, such as the fix for bug
2876 * CVE-2014-8121 The NSS backends shared internal state between the getXXent
2877 and getXXbyYY NSS calls for the same database, causing a denial-of-service
2878 condition in some applications.
2880 * Added vector math library named libmvec with the following vectorized x86_64
2881 implementations: cos, cosf, sin, sinf, sincos, sincosf, log, logf, exp, expf,
2883 The library can be disabled with --disable-mathvec. Use of the functions is
2884 enabled with -fopenmp -ffast-math starting from -O1 for GCC version >= 4.9.0.
2885 Shared library libmvec.so is linked in as needed when using -lm (no need to
2886 specify -lmvec explicitly for not static builds).
2887 Visit <https://sourceware.org/glibc/wiki/libmvec> for detailed information.
2889 * A new fmemopen implementation has been added with the goal of POSIX
2890 compliance. The new implementation fixes the following long-standing
2891 issues: BZ#6544, BZ#11216, BZ#12836, BZ#13151, BZ#13152, and BZ#14292. The
2892 old implementation is still present for use be by existing binaries.
2894 * The 32-bit sparc sigaction ABI was inadvertently broken in the 2.20 and 2.21
2895 releases. It has been fixed to match 2.19 and older, but binaries built
2896 against 2.20 and 2.21 might need to be recompiled. See BZ#18694.
2898 * Port to Native Client running on ARMv7-A (--host=arm-nacl).
2899 Contributed by Roland McGrath (Google).
2901 * The header <regexp.h> is deprecated, and will be removed in a future
2902 release. Use of this header will trigger a deprecation warning.
2903 Application developers should update their code to use <regex.h> instead.
2905 This header was formerly part of SUS, but was deprecated in 1994 and
2906 removed from the standard in 2001. Also, the glibc implementation
2907 leaks memory. See BZ#18681 for more details.
2911 * The following bugs are resolved with this release:
2913 6652, 10672, 12674, 12847, 12926, 13862, 14132, 14138, 14171, 14498,
2914 15215, 15378, 15884, 16009, 16418, 16191, 16469, 16576, 16617, 16618,
2915 16619, 16657, 16740, 16857, 17192, 17266, 17273, 17344, 17363, 17370,
2916 17371, 17411, 17460, 17475, 17485, 17501, 17506, 17508, 17522, 17555,
2917 17570, 17571, 17572, 17573, 17574, 17582, 17583, 17584, 17585, 17589,
2918 17594, 17601, 17608, 17616, 17625, 17630, 17633, 17634, 17635, 17647,
2919 17653, 17657, 17658, 17664, 17665, 17668, 17682, 17702, 17717, 17719,
2920 17722, 17723, 17724, 17725, 17732, 17733, 17744, 17745, 17746, 17747,
2921 17748, 17775, 17777, 17780, 17781, 17782, 17791, 17793, 17796, 17797,
2922 17801, 17803, 17806, 17834, 17844, 17848, 17868, 17869, 17870, 17885,
2925 * CVE-2015-1472 CVE-2015-1473 Under certain conditions wscanf can allocate
2926 too little memory for the to-be-scanned arguments and overflow the
2927 allocated buffer. The implementation now correctly computes the required
2928 buffer size when using malloc, and switches to malloc from alloca as
2931 * A new semaphore algorithm has been implemented in generic C code for all
2932 machines. Previous custom assembly implementations of semaphore were
2933 difficult to reason about or ensure that they were safe. The new version
2934 of semaphore supports machines with 64-bit or 32-bit atomic operations.
2935 The new semaphore algorithm is used by sem_init, sem_open, sem_post,
2936 sem_wait, sem_timedwait, sem_trywait, and sem_getvalue.
2938 * Port to Altera Nios II has been contributed by Mentor Graphics.
2940 * Optimized strcpy, stpcpy, strncpy, stpncpy, strcmp, and strncmp
2941 implementations for powerpc64/powerpc64le.
2942 Implemented by Adhemerval Zanella (IBM).
2944 * Added support for TSX lock elision of pthread mutexes on powerpc32, powerpc64
2945 and powerpc64le. This may improve lock scaling of existing programs on
2946 HTM capable systems. The lock elision code is only enabled with
2947 --enable-lock-elision=yes. Also, the TSX lock elision implementation for
2948 powerpc will issue a transaction abort on every syscall to avoid side
2949 effects being visible outside transactions.
2951 * Optimized strcpy, stpcpy, strchrnul and strrchr implementations for
2952 AArch64. Contributed by ARM Ltd.
2954 * i386 memcpy functions optimized with SSE2 unaligned load/store.
2956 * CVE-2104-7817 The wordexp function could ignore the WRDE_NOCMD flag
2957 under certain input conditions resulting in the execution of a shell for
2958 command substitution when the application did not request it. The
2959 implementation now checks WRDE_NOCMD immediately before executing the
2960 shell and returns the error WRDE_CMDSUB as expected.
2962 * CVE-2012-3406 printf-style functions could run into a stack overflow when
2963 processing format strings with a large number of format specifiers.
2965 * CVE-2014-9402 The nss_dns implementation of getnetbyname could run into an
2966 infinite loop if the DNS response contained a PTR record of an unexpected
2969 * The minimum GCC version that can be used to build this version of the GNU
2970 C Library is GCC 4.6. Older GCC versions, and non-GNU compilers, can
2971 still be used to compile programs using the GNU C Library.
2973 * The GNU C Library is now built with -Werror by default. This can be
2974 disabled by configuring with --disable-werror.
2976 * New locales: tu_IN, bh_IN, raj_IN, ce_RU.
2978 * The obsolete sigvec function has been removed. This was the original
2979 4.2BSD interface that inspired the POSIX.1 sigaction interface, which
2980 programs have been using instead for about 25 years. Of course, ABI
2981 compatibility for old binaries using sigvec remains intact.
2983 * Merged gettext 0.19.3 into the intl subdirectory. This fixes building
2984 with newer versions of bison.
2986 * Support for MIPS o32 FPXX, FP64A and FP64 ABI Extensions.
2987 The original MIPS o32 hard-float ABI requires an FPU where double-precision
2988 registers overlay two consecutive single-precision registers. MIPS32R2
2989 introduced a new FPU mode (FR=1) where double-precision registers extend the
2990 corresponding single-precision registers which is incompatible with the
2991 o32 hard-float ABI. The MIPS SIMD ASE and the MIPSR6 architecture both
2992 require the use of FR=1 making a transition necessary. New o32 ABI
2993 extensions enable users to migrate over time from the original o32 ABI
2994 through to the updated o32 FP64 ABI. To achieve this the dynamic linker now
2995 tracks the ABI of any loaded object and verifies that new objects are
2996 compatible. Mode transitions will also be requested as required and
2997 unsupportable objects will be rejected. The ABI checks include both soft and
2998 hard float ABIs for o32, n32 and n64.
3000 GCC 5 with GNU binutils 2.25 onwards:
3001 It is strongly recommended that all o32 system libraries are built using the
3002 new o32 FPXX ABI (-mfpxx) to facilitate the transition as this is compatible
3003 with the original and all new o32 ABI extensions. Configure a MIPS GCC
3004 compiler using --with-fp-32=xx to set this by default.
3008 * The following bugs are resolved with this release:
3010 6804, 9894, 11505, 12994, 13347, 13651, 14308, 14770, 15119, 15132, 15347,
3011 15514, 15698, 15804, 15894, 15946, 16002, 16064, 16095, 16194, 16198,
3012 16275, 16284, 16287, 16315, 16348, 16349, 16354, 16357, 16362, 16447,
3013 16516, 16532, 16539, 16545, 16561, 16562, 16564, 16574, 16599, 16600,
3014 16609, 16610, 16611, 16613, 16619, 16623, 16629, 16632, 16634, 16639,
3015 16642, 16648, 16649, 16670, 16674, 16677, 16680, 16681, 16683, 16689,
3016 16695, 16701, 16706, 16707, 16712, 16713, 16714, 16724, 16731, 16739,
3017 16740, 16743, 16754, 16758, 16759, 16760, 16770, 16786, 16789, 16791,
3018 16796, 16799, 16800, 16815, 16823, 16824, 16831, 16838, 16839, 16849,
3019 16854, 16876, 16877, 16878, 16882, 16885, 16888, 16890, 16892, 16912,
3020 16915, 16916, 16917, 16918, 16922, 16927, 16928, 16932, 16943, 16958,
3021 16965, 16966, 16967, 16977, 16978, 16984, 16990, 16996, 17009, 17022,
3022 17031, 17042, 17048, 17050, 17058, 17061, 17062, 17069, 17075, 17078,
3023 17079, 17084, 17086, 17088, 17092, 17097, 17125, 17135, 17137, 17150,
3024 17153, 17187, 17213, 17259, 17261, 17262, 17263, 17319, 17325, 17354.
3026 * Reverted change of ABI data structures for s390 and s390x:
3027 On s390 and s390x the size of struct ucontext and jmp_buf was increased in
3028 2.19. This change is reverted in 2.20. The introduced 2.19 symbol versions
3029 of getcontext, setjmp, _setjmp, __sigsetjmp, longjmp, _longjmp, siglongjmp
3030 are preserved pointing straight to the same implementation as the old ones.
3031 Given that, new callers will simply provide a too-big buffer to these
3032 functions. Any applications/libraries out there that embed jmp_buf or
3033 ucontext_t in an ABI-relevant data structure that have already been rebuilt
3034 against 2.19 headers will have to rebuilt again. This is necessary in any
3035 case to revert the breakage in their ABI caused by the glibc change.
3037 * Support for file description locks is added to systems running the
3038 Linux kernel. The standard file locking interfaces are extended to
3039 operate on file descriptions, not file descriptors, via the use of
3040 F_OFD_GETLK, F_OFD_SETLK, and F_OFD_SETLKW. File description locks
3041 are associated with an open file instead of a process.
3043 * Optimized strchr implementation for AArch64. Contributed by ARM Ltd.
3045 * The minimum Linux kernel version that this version of the GNU C Library
3046 can be used with is 2.6.32.
3048 * Running the testsuite no longer terminates as soon as a test fails.
3049 Instead, a file tests.sum (xtests.sum from "make xcheck") is generated,
3050 with PASS or FAIL lines for individual tests. A summary of the results is
3051 printed, including a list of failing lists, and "make check" exits with
3052 error status if there were any unexpected failures. "make check
3053 stop-on-test-failure=y" may be used to keep the old behavior.
3055 * The am33 port, which had not worked for several years, has been removed
3058 * The _BSD_SOURCE and _SVID_SOURCE feature test macros are no longer
3059 supported; they now act the same as _DEFAULT_SOURCE (but generate a
3060 warning). Except for cases where _BSD_SOURCE enabled BSD interfaces that
3061 conflicted with POSIX (support for which was removed in 2.19), the
3062 interfaces those macros enabled remain available when compiling with
3063 _GNU_SOURCE defined, with _DEFAULT_SOURCE defined, or without any feature
3064 test macros defined.
3066 * Optimized strcmp implementation for ARMv7. Contributed by ARM Ltd.
3068 * Added support for TX lock elision of pthread mutexes on s390 and s390x.
3069 This may improve lock scaling of existing programs on TX capable systems.
3070 The lock elision code is only built with --enable-lock-elision=yes and
3071 then requires a GCC version supporting the TX builtins. With lock elision
3072 default mutexes are elided via __builtin_tbegin, if the cpu supports
3073 transactions. By default lock elision is not enabled and the elision code
3076 * CVE-2014-4043 The posix_spawn_file_actions_addopen implementation did not
3077 copy the path argument. This allowed programs to cause posix_spawn to
3078 deference a dangling pointer, or use an unexpected pathname argument if
3079 the string was modified after the posix_spawn_file_actions_addopen
3082 * All supported architectures now use the main glibc sysdeps directory
3083 instead of some being in a separate "ports" directory (which was
3084 distributed separately before glibc 2.17).
3086 * The NPTL implementation of POSIX pthreads is no longer an "add-on".
3087 On configurations that support it (all Linux configurations), it's now
3088 used regardless of the --enable-add-ons switch to configure. It is no
3089 longer possible to build such configurations without pthreads support.
3091 * Locale names, including those obtained from environment variables (LANG
3092 and the LC_* variables), are more tightly checked for proper syntax.
3093 setlocale will now fail (with EINVAL) for locale names that are overly
3094 long, contain slashes without starting with a slash, or contain ".." path
3095 components. (CVE-2014-0475) Previously, some valid locale names were
3096 silently replaced with the "C" locale when running in AT_SECURE mode
3097 (e.g., in a SUID program). This is no longer necessary because of the
3100 * On x86-64, the dynamic linker's lazy-binding support is now compatible
3101 with application code using Intel MPX instructions. (With all previous
3102 versions, the MPX register state could be clobbered when making calls
3103 into or out of a shared library.) Note that while the new dynamic
3104 linker is compatible with all known x86 hardware whether or not it
3105 supports Intel MPX, some x86 instruction-set emulators might fail to
3106 handle the new instruction encodings. This is known to affect Valgrind
3107 versions up through 3.9 (but will be fixed in the forthcoming 3.10
3108 release), and might affect other tools that do instruction emulation.
3110 * Support for loadable gconv transliteration modules has been removed.
3111 The support for transliteration modules has been non-functional for
3112 over a decade, and the removal is prompted by security defects. The
3113 normal gconv conversion modules are still supported. Transliteration
3114 with //TRANSLIT is still possible, and the //IGNORE specifier
3115 continues to be supported. (CVE-2014-5119)
3117 * Decoding a crafted input sequence in the character sets IBM933, IBM935,
3118 IBM937, IBM939, IBM1364 could result in an out-of-bounds array read,
3119 resulting a denial-of-service security vulnerability in applications which
3120 use functions related to iconv. (CVE-2014-6040)
3124 * The following bugs are resolved with this release:
3126 156, 387, 431, 762, 832, 926, 2801, 4772, 6786, 6787, 6807, 6810, 6981,
3127 7003, 9721, 9954, 10253, 10278, 11087, 11157, 11214, 12100, 12486, 12751,
3128 12986, 13028, 13982, 13985, 14029, 14032, 14120, 14143, 14155, 14286,
3129 14547, 14699, 14752, 14782, 14876, 14910, 15004, 15048, 15073, 15089,
3130 15128, 15218, 15268, 15277, 15308, 15362, 15374, 15400, 15425, 15427,
3131 15483, 15522, 15531, 15532, 15593, 15601, 15608, 15609, 15610, 15632,
3132 15640, 15670, 15672, 15680, 15681, 15723, 15734, 15735, 15736, 15748,
3133 15749, 15754, 15760, 15763, 15764, 15797, 15799, 15825, 15843, 15844,
3134 15846, 15847, 15849, 15850, 15855, 15856, 15857, 15859, 15867, 15886,
3135 15887, 15890, 15892, 15893, 15895, 15897, 15901, 15905, 15909, 15915,
3136 15917, 15919, 15921, 15923, 15939, 15941, 15948, 15963, 15966, 15968,
3137 15985, 15988, 15997, 16032, 16034, 16036, 16037, 16038, 16041, 16046,
3138 16055, 16071, 16072, 16074, 16077, 16078, 16103, 16112, 16143, 16144,
3139 16146, 16150, 16151, 16153, 16167, 16169, 16172, 16195, 16214, 16245,
3140 16271, 16274, 16283, 16289, 16293, 16314, 16316, 16330, 16337, 16338,
3141 16356, 16365, 16366, 16369, 16372, 16375, 16379, 16381, 16384, 16385,
3142 16386, 16387, 16390, 16394, 16398, 16400, 16407, 16408, 16414, 16430,
3143 16431, 16453, 16474, 16506, 16510, 16529
3145 * Slovenian translations for glibc messages have been contributed by the
3146 Translation Project's Slovenian team of translators.
3148 * The public headers no longer use __unused nor __block. This change is to
3149 support compiling programs that are derived from BSD sources and use
3150 __unused internally, and to support compiling with Clang's -fblock
3151 extension which uses __block.
3153 * CVE-2012-4412 The strcoll implementation caches indices and rules for
3154 large collation sequences to optimize multiple passes. This cache
3155 computation may overflow for large collation sequences and may cause a
3156 stack or buffer overflow. This is now fixed to use a slower algorithm
3157 which does not use a cache if there is an integer overflow.
3159 * CVE-2012-4424 The strcoll implementation uses malloc to cache indices and
3160 rules for large collation sequences to optimize multiple passes and falls
3161 back to alloca if malloc fails, resulting in a possible stack overflow.
3162 The implementation now falls back to an uncached collation sequence lookup
3165 * CVE-2013-4788 The pointer guard used for pointer mangling was not
3166 initialized for static applications resulting in the security feature
3167 being disabled. The pointer guard is now correctly initialized to a
3168 random value for static applications. Existing static applications need
3169 to be recompiled to take advantage of the fix (bug 15754).
3171 * CVE-2013-4237 The readdir_r function could write more than NAME_MAX bytes
3172 to the d_name member of struct dirent, or omit the terminating NUL
3173 character. (Bugzilla #14699).
3175 * CVE-2013-4332 The pvalloc, valloc, memalign, posix_memalign and
3176 aligned_alloc functions could allocate too few bytes or corrupt the
3177 heap when passed very large allocation size values (Bugzilla #15855,
3180 * CVE-2013-4458 Stack overflow in getaddrinfo with large number of results
3181 for AF_INET6 has been fixed (Bugzilla #16072).
3183 * New locales: ak_GH, anp_IN, ar_SS, cmn_TW, hak_TW, lzh_TW, nan_TW, pap_AW,
3184 pap_CW, quz_PE, the_NP.
3186 * Substantially revised locales: gd_GB, ht_HT
3188 * The LC_ADDRESS field was updated to support country_car for almost all
3191 * ISO 1427 definitions were updated.
3193 * ISO 3166 definitions were updated.
3195 * The localedef utility now supports --big-endian and --little-endian
3196 command-line options to generate locales for a different system from that
3197 for which the C library was built.
3199 * Binary locale files now only depend on the endianness of the system for
3200 which they are generated and not on other properties of that system. As a
3201 consequence, binary files generated with new localedef may be incompatible
3202 with old versions of the GNU C Library, and binary files generated with
3203 old localedef may be incompatible with this version of the GNU C Library,
3204 in the following circumstances:
3206 + Locale files may be incompatible on m68k systems.
3208 + Locale archive files (but not separate files for individual locales) may
3209 be incompatible on systems where plain "char" is signed.
3211 * The configure option --disable-versioning has been removed. Builds with
3212 --disable-versioning had not worked for several years.
3214 * ISO 639 definitions were updated for Chiga (cgg) and Chinese (gan, hak, czh,
3215 cjy, lzh, cmn, mnp, cdo, czo, cpx, wuu, hsn, yue).
3217 * SystemTap probes for malloc have been introduced.
3219 * SystemTap probes for slow multiple precision fallback paths of
3220 transcendental functions have been introduced.
3222 * Support for powerpc64le has been added.
3224 * The soft-float powerpc port now supports e500 processors.
3226 * Support for STT_GNU_IFUNC symbols added for ppc32/power4+ and ppc64.
3228 * A new feature test macro _DEFAULT_SOURCE is available to enable the same
3229 set of header declarations that are enabled by default, even when other
3230 feature test macros or compiler options such as -std=c99 would otherwise
3231 disable some of those declarations.
3233 * The _BSD_SOURCE feature test macro no longer enables BSD interfaces that
3234 conflict with POSIX. The libbsd-compat library (which was a dummy library
3235 that did nothing) has also been removed.
3237 * Preliminary documentation about Multi-Thread, Async-Signal and
3238 Async-Cancel Safety has been added.
3240 * Change of ABI data structures for s390 and s390x:
3241 On s390 and s390x the size of struct ucontext and jmp_buf was increased to
3242 allow for future hardware extensions. All library functions that accept or
3243 return these structures were versioned in order to provide backwards
3244 compatibility with existing code. However, not all cases can be handled
3245 with interface versioning. If an application embeds either structure into
3246 another structure and that structure is passed to another compilation unit
3247 which expects the newer larger-sized object then any access to the new
3248 fields will result in undefined behaviour. Similarly any access to fields
3249 that were moved by the enlarging of the structures will cause undefined
3250 behaviour. To fix the undefined behaviour all objects that are part of the
3251 final in-memory image must agree on the size of structures, and this may
3252 require recompilation.
3256 * The following bugs are resolved with this release:
3258 2546, 2560, 5159, 6809, 7006, 10060, 10062, 10283, 10357, 10686, 11120,
3259 11561, 12310, 12387, 12492, 12515, 12723, 13550, 13889, 13951, 13988,
3260 14142, 14176, 14200, 14256, 14280, 14293, 14317, 14327, 14478, 14496,
3261 14582, 14686, 14812, 14888, 14894, 14907, 14908, 14909, 14920, 14941,
3262 14952, 14964, 14981, 14982, 14985, 14991, 14994, 14996, 15000, 15003,
3263 15006, 15007, 15014, 15020, 15022, 15023, 15036, 15054, 15055, 15062,
3264 15078, 15084, 15085, 15086, 15100, 15160, 15214, 15221, 15232, 15234,
3265 15283, 15285, 15287, 15304, 15305, 15307, 15309, 15327, 15330, 15335,
3266 15336, 15337, 15339, 15342, 15346, 15359, 15361, 15366, 15380, 15381,
3267 15394, 15395, 15405, 15406, 15409, 15416, 15418, 15419, 15423, 15424,
3268 15426, 15429, 15431, 15432, 15441, 15442, 15448, 15465, 15480, 15485,
3269 15488, 15490, 15492, 15493, 15497, 15506, 15529, 15536, 15553, 15577,
3270 15583, 15618, 15627, 15631, 15654, 15655, 15666, 15667, 15674, 15711,
3273 * CVE-2013-2207 Incorrectly granting access to another user's pseudo-terminal
3274 has been fixed by disabling the use of pt_chown (Bugzilla #15755).
3275 Distributions can re-enable building and using pt_chown via the new configure
3276 option `--enable-pt_chown'. Enabling the use of pt_chown carries with it
3277 considerable security risks and should only be used if the distribution
3278 understands and accepts the risks.
3280 * CVE-2013-0242 Buffer overrun in regexp matcher has been fixed (Bugzilla
3283 * CVE-2013-1914 Stack overflow in getaddrinfo with many results has been
3284 fixed (Bugzilla #15330).
3286 * Add support for calling C++11 thread_local object destructors on thread
3287 and program exit. This needs compiler support for offloading C++11
3288 destructor calls to glibc.
3290 * Improved worst case performance of libm functions with double inputs and
3293 * Support for priority inherited mutexes in pthread condition variables on
3294 non-x86 architectures.
3296 * Port to Xilinx MicroBlaze contributed by David Holsgrove.
3298 * Optimized string functions for AArch64. Implemented by Marcus Shawcroft.
3300 * Optimized string functions for ARM. Implemented by Will Newton and
3303 * Optimized libm functions for SPARC. Implemented by David S. Miller.
3305 * Support for building more of ARM glibc as Thumb-2 code. Implemented by
3308 * Support for building most of MIPS glibc as MIPS16 code. Implemented by
3309 Chung-Lin Tang, Maciej W. Rozycki and Maxim Kuvyrkov.
3311 * Added a benchmark framework to track performance of functions in glibc.
3313 * New <math.h> macro named issignaling to check for a signaling NaN (sNaN).
3314 It is based on draft TS 18661 and currently enabled as a GNU extension.
3316 * On Linux, the clock function now uses the clock_gettime system call
3317 for improved precision, rather than old times system call.
3319 * Added support for version-3 tz format files. This is needed when using
3320 the tz database release 2013e or later, and affects a few unusual cases --
3321 currently only TZ='America/Godthab' for time stamps after 2037.
3323 * Added new API functions pthread_getattr_default_np and
3324 pthread_setattr_default_np to get and set the default pthread
3325 attributes of a process.
3327 * Added support for TSX lock elision for pthread mutexes on i386 and x86-64.
3328 This may improve lock scaling of existing programs on TSX capable systems.
3329 When the --enable-lock-elision=yes parameter is specified at configure
3330 time lock elision will be enabled by default for all PTHREAD_MUTEX_DEFAULT
3333 * Support added for AT_HWCAP2 (to coincide with Linux kernel 3.10
3334 availability). Implemented by Ryan S. Arnold.
3336 * Support added for POWER8 platform. Implemented by Ryan S. Arnold.
3340 * The following bugs are resolved with this release:
3342 1349, 3439, 3479, 3665, 5044, 5246, 5298, 5400, 6530, 6677, 6778, 6808,
3343 9685, 9914, 10014, 10038, 10114, 10191, 10631, 10873, 11438, 11607, 11638,
3344 11741, 12140, 13013, 13412, 13542, 13601, 13603, 13604, 13629, 13679,
3345 13696, 13698, 13717, 13741, 13759, 13761, 13763, 13881, 13939, 13950,
3346 13952, 13966, 14042, 14047, 14090, 14150, 14151, 14152, 14154, 14157,
3347 14166, 14173, 14195, 14197, 14237, 14246, 14251, 14252, 14283, 14298,
3348 14303, 14307, 14328, 14331, 14336, 14337, 14347, 14349, 14368, 14376,
3349 14417, 14447, 14459, 14476, 14477, 14501, 14505, 14510, 14516, 14518,
3350 14519, 14530, 14532, 14538, 14543, 14544, 14545, 14557, 14562, 14568,
3351 14576, 14579, 14583, 14587, 14595, 14602, 14610, 14621, 14638, 14645,
3352 14648, 14652, 14660, 14661, 14669, 14672, 14683, 14694, 14716, 14719,
3353 14743, 14767, 14783, 14784, 14785, 14793, 14796, 14797, 14801, 14803,
3354 14805, 14807, 14811, 14815, 14821, 14822, 14824, 14828, 14831, 14833,
3355 14835, 14838, 14856, 14863, 14865, 14866, 14868, 14869, 14871, 14872,
3356 14879, 14889, 14893, 14898, 14914.
3358 * Optimization of memcpy for MIPS.
3360 * CVE-2011-4609 svc_run() produces high cpu usage when accept fails with
3361 EMFILE has been fixed (Bugzilla #14889).
3363 * The add-on ports collection is now distributed in the "ports" subdirectory
3364 of the main GNU C Library distribution, rather than separately.
3366 * Port to ARM AArch64 contributed by Linaro.
3368 * Support for STT_GNU_IFUNC symbols added for s390 and s390x.
3369 Optimized versions of memcpy, memset, and memcmp added for System z10 and
3371 Implemented by Andreas Krebbel.
3373 * The new function secure_getenv allows secure access to the environment,
3374 returning NULL if running in a SUID/SGID process. This function replaces
3375 the internal function __secure_getenv.
3377 * SystemTap static probes have been added into the dynamic linker.
3378 Implemented by Gary Benson.
3380 * Optimizations of string functions strstr, strcasestr and memmem.
3381 Implemented by Maxim Kuvyrkov.
3383 * The minimum Linux kernel version that this version of the GNU C Library
3384 can be used with is 2.6.16.
3386 * Optimizations of string functions memchr, wcschr, wcscpy, and wcsrchr for
3387 powerpc POWER7. Implemented by Will Schmidt.
3389 * New configure option --disable-nscd builds the C library such that it
3390 never attempts to contact the Name Service Caching Daemon (nscd).
3391 New configure option --disable-build-nscd avoids building nscd itself;
3392 this is the default if --disable-nscd is used.
3394 * Improved support for cross-compilation, including cross-testing and
3395 bootstrap builds without a previously built glibc.
3397 * Several testsuite tests are now able to test multiple IFUNC variants of an
3398 interface, rather than just testing the one that would be chooen by
3401 * New configure options --with-bugurl and --with-pkgversion, for
3402 distributors to use to embed their bug-reporting and package version
3403 information in --help and --version output.
3405 * The ttyname and ttyname_r functions on Linux now fall back to searching for
3406 the tty file descriptor in /dev/pts or /dev if /proc is not available. This
3407 allows creation of chroots without the procfs mounted on /proc.
3409 * The `crypt' function now fails if passed salt bytes that violate the
3410 specification for those values. On Linux, the `crypt' function will
3411 consult /proc/sys/crypto/fips_enabled to determine if "FIPS mode" is
3412 enabled, and fail on encrypted strings using the MD5 or DES algorithm
3413 when the mode is enabled.
3415 * The `clock_*' suite of functions (declared in <time.h>) is now available
3416 directly in the main C library. Previously it was necessary to link with
3417 -lrt to use these functions. This change has the effect that a
3418 single-threaded program that uses a function such as `clock_gettime' (and
3419 is not linked with -lrt) will no longer implicitly load the pthreads
3420 library at runtime and so will not suffer the overheads associated with
3421 multi-thread support in other code such as the C++ runtime library.
3423 * New locales: ayc_PE, doi_IN, ia_FR, mni_IN, nhn_MX, niu_NU, niu_NZ,
3428 * The following bugs are resolved with this release:
3430 174, 208, 350, 369, 411, 706, 766, 2074, 2541, 2547, 2548, 2550, 2551,
3431 2552, 2553, 2554, 2562, 2563, 2565, 2566, 2570, 2576, 2636, 2678, 3335,
3432 3440, 3748, 3768, 3866, 3868, 3906, 3976, 3992, 4026, 4108, 4596, 4822,
3433 5077, 5461, 5805, 5993, 6471, 6486, 6578, 6649, 6730, 6770, 6794, 6884,
3434 6890, 6894, 6895, 6907, 6911, 6959, 7064, 9739, 9902, 10110, 10135, 10140,
3435 10153, 10210, 10254, 10346, 10375, 10545, 10716, 10846, 10882, 11174,
3436 11261, 11322, 11365, 11451, 11494, 11521, 11677, 11837, 11959, 12047,
3437 12097, 12193, 12194, 12297, 12298, 12301, 12340, 12354, 12416, 12495,
3438 13058, 13223, 13361, 13525, 13526, 13527, 13528, 13529, 13530, 13531,
3439 13532, 13533, 13547, 13551, 13552, 13553, 13555, 13556, 13559, 13563,
3440 13566, 13576, 13579, 13583, 13592, 13594, 13613, 13618, 13637, 13656,
3441 13658, 13673, 13691, 13695, 13704, 13705, 13706, 13718, 13726, 13738,
3442 13739, 13743, 13750, 13758, 13760, 13761, 13775, 13786, 13787, 13792,
3443 13806, 13824, 13840, 13841, 13844, 13846, 13848, 13851, 13852, 13854,
3444 13871, 13872, 13873, 13879, 13882, 13883, 13884, 13885, 13886, 13892,
3445 13895, 13908, 13910, 13911, 13912, 13913, 13914, 13915, 13916, 13917,
3446 13918, 13919, 13920, 13921, 13922, 13923, 13924, 13926, 13927, 13928,
3447 13938, 13941, 13942, 13954, 13955, 13956, 13963, 13967, 13968, 13970,
3448 13973, 13979, 13983, 13986, 13996, 14012, 14027, 14033, 14034, 14036,
3449 14040, 14043, 14044, 14048, 14049, 14050, 14053, 14055, 14059, 14064,
3450 14075, 14080, 14083, 14103, 14104, 14109, 14112, 14117, 14122, 14123,
3451 14134, 14153, 14183, 14188, 14199, 14210, 14218, 14229, 14241, 14273,
3454 * Support for the x32 ABI on x86-64 added. The x32 target is selected by
3455 configuring glibc with:
3456 BUILD_CC='gcc' CC='gcc -mx32' CXX='g++ -mx32'
3457 Visit <http://sites.google.com/site/x32abi/> for more x32 ABI info.
3458 Implemented by H.J. Lu.
3462 + define static_assert
3464 + do not declare gets
3466 + declare at_quick_exit and quick_exit also for ISO C11
3468 + aligned_alloc. NB: The code is deliberately allows the size parameter
3469 to not be a multiple of the alignment. This is a moronic requirement
3470 in the standard but it is only a requirement on the caller, not the
3473 + timespec_get added
3475 + uchar.h support added
3477 + CMPLX, CMPLXF, CMPLXL added
3479 Implemented by Ulrich Drepper.
3481 * Support for the IA-64 has been moved to ports.
3483 * Remove support for anything but ELF binary format
3485 * Checking versions of poll, ppoll added.
3486 Implemented by Ulrich Drepper.
3488 * More generic and 64-bit performance optimizations to math functions.
3489 Implemented by Ulrich Drepper.
3491 * New configure option --enable-obsolete-rpc makes the deprecated RPC
3492 headers and functions available at compile time as they were before
3493 version 2.14. This option will be removed at some time in the future
3494 after the TI-RPC library becomes fully sufficient for the needs of
3495 existing applications.
3497 * Compatibility code for Linux kernel versions before 2.4 has been removed.
3498 Note that glibc is not expected to work with any Linux kernel version
3501 * New header <sys/auxv.h> and function getauxval allowing easy access to
3502 the AT_* key-value pairs passed from the Linux kernel. The header also
3503 defines the HWCAP_* bits associated with the AT_HWCAP key.
3505 * New locales: mag_IN
3507 * New configure option --enable-systemtap builds SystemTap static probes
3508 into libc for setjmp and longjmp and into libpthread for various operations.
3509 So far the setjmp/longjmp probes and some of the libpthread probes are
3510 provided only for i*86 and x86_64.
3511 Implemented by Roland McGrath and Rayson Ho.
3513 * Optimized expf for x86-32 and x86-64. Implemented by Liubov Dmitrieva.
3515 * More optimized functions for PowerPC. Implemented by Adhemerval Zanella
3518 * More optimized functions for SPARC. Implemented by David S. Miller.
3520 * Improved support for cross-compilation, especially bootstrap builds
3521 without a previously built glibc.
3523 * Ports for the TILE-Gx and TILEPro families of processors. Contributed by
3524 Chris Metcalf from Tilera.
3526 * Support for the old ARM ABI has been removed from ports. Only the EABI is
3527 now supported for ARM processors.
3529 * The hard-float variant of the ARM EABI now uses /lib/ld-linux-armhf.so.3
3530 as the name of the dynamic linker, to distinguish it from the
3531 /lib/ld-linux.so.3 dynamic linker used for the base EABI.
3533 * Support for CRIS has been removed from ports.
3535 * A new class of installed header has been documented for low-level
3536 platform-specific functionality. PowerPC added the first instance with a
3537 function to provide time base register access. Contributed by Tulio
3538 Magno Quites Machado Filho.
3540 * ix86 configurations now install header files that are consistent with
3541 what x86-64 configurations install. These same header files can be used
3542 for -m32, -m64, or -mx32 builds.
3543 Contributed by H.J. Lu.
3545 * Math library bug fixes. A thorough audit of all open math library bugs was
3546 conducted by Joseph Myers. Significant progress was made on many math
3547 library bugs resulting in more accurate exceptions and function results.
3548 Many thanks to all those that contributed including Andreas Jaeger for his
3549 patch review and work on the x87 trigonometric instruction issues.
3551 * Timezone data is no longer installed. Timezone-related binaries and scripts
3552 will continue to be installed. Users should obtain their timezone data from
3553 their distribution provider or from the tzdata package at
3554 <ftp://munnari.oz.au/pub/>.
3558 * The following bugs are resolved with this release:
3560 6779, 6783, 9696, 10103, 10709, 11589, 11929, 12403, 12786, 12840, 12847,
3561 12868, 12852, 12874, 12885, 12892, 12906, 12907, 12922, 12935, 12962,
3562 13007, 13021, 13061, 13062, 13067, 13068, 13085, 13088, 13090, 13092,
3563 13096, 13114, 13118, 13123, 13134, 13138, 13147, 13150, 13166, 13179,
3564 13185, 13189, 13192, 13268, 13276, 13282, 13291, 13305, 13328, 13335,
3565 13337, 13344, 13358, 13367, 13413, 13416, 13423, 13439, 13446, 13472,
3566 13484, 13506, 13515, 13523, 13524, 13538, 13540
3568 * New program pldd to list loaded object of a process
3569 Implemented by Ulrich Drepper.
3571 * Add nss_db support back to glibc. No more dependency on Berkeley db
3572 and support for initgroups lookups.
3573 Implemented by Ulrich Drepper.
3575 * Optimized strcpy, strncpy, stpcpy, stpncpy for SSE2 and SSSE3 on x86-32.
3576 Contributed by HJ Lu.
3578 * Improved strcpy, strncpy, stpcpy, stpncpy for SSE2 and SSSE3 on x86-64.
3579 Contributed by HJ Lu.
3581 * Optimized strcat, strncat on x86-64 and optimized wcscmp, wcslen, strnlen
3582 on x86-32 and x86-64.
3583 Contributed by Liubov Dmitrieva.
3585 * Optimized strchr and strrchr for SSE on x86-32.
3586 Contributed by Liubov Dmitrieva.
3588 * Optimized memchr, memrchr, rawmemchr, memcmp, wmemcmp, wcschr, wcscpy
3589 for x86-64 and x86-32.
3590 Contributed by Liubov Dmitrieva.
3592 * New interfaces: scandirat, scandirat64
3593 Implemented by Ulrich Drepper.
3595 * Checking versions of FD_SET, FD_CLR, and FD_ISSET added.
3596 Implemented by Ulrich Drepper.
3598 * nscd now also caches the netgroup database.
3599 Implemented by Ulrich Drepper.
3601 * Integrate libm with gcc's -ffinite-math-only option.
3602 Implemented by Ulrich Drepper.
3604 * Lots of generic, 64-bit, and x86-64-specific performance optimizations
3605 to math functions. Implemented by Ulrich Drepper.
3607 * Optimized strcasecmp and strncasecmp for AVX on x86-64.
3608 Implemented by Ulrich Drepper.
3610 * New Linux interfaces: process_vm_readv, process_vm_writev
3612 * Optimized strcasecmp and strncasecmp for SSSE3 and SSE4.2 on x86-32.
3613 Implemented by Ulrich Drepper.
3615 * Optimized nearbyint and strcasecmp for PPC.
3616 Implemented by Adhemerval Zanella.
3618 * New locales: bho_IN, unm_US, es_CU, ta_LK
3622 * The following bugs are resolved with this release:
3624 386, 6420, 7101, 9730, 9732, 9809, 10138, 10149, 10157, 11099, 11257,
3625 11258, 11487, 11532, 11558, 11578, 11634, 11653, 11668, 11697, 11724,
3626 11781, 11799, 11820, 11837, 11857, 11884, 11892, 11895, 11901, 11945,
3627 11947, 11952, 11987, 12052, 12083, 12158, 12178, 12200, 12346, 12350,
3628 12393, 12420, 12432, 12445, 12449, 12453, 12454, 12460, 12469, 12489,
3629 12509, 12510, 12511, 12518, 12527, 12541, 12545, 12551, 12582, 12583,
3630 12587, 12597, 12601, 12611, 12625, 12626, 12631, 12650, 12653, 12655,
3631 12660, 12671, 12681, 12684, 12685, 12711, 12713, 12714, 12717, 12723,
3632 12724, 12734, 12738, 12746, 12766, 12775, 12777, 12782, 12788, 12792,
3633 12795, 12811, 12813, 12814, 12841
3635 * The RPC implementation in libc is obsoleted. Old programs keep working
3636 but new programs cannot be linked with the routines in libc anymore.
3637 Programs in need of RPC functionality must be linked against TI-RPC.
3638 The TI-RPC implementation is IPv6 enabled and there are other benefits.
3640 Visible changes of this change include (obviously) the inability to link
3641 programs using RPC functions without referencing the TI-RPC library and the
3642 removal of the RPC headers from the glibc headers.
3643 Implemented by Ulrich Drepper.
3645 * New Linux interfaces: clock_adjtime, name_to_handle_at, open_by_handle_at,
3646 syncfs, setns, sendmmsg
3648 * New locales: os_RU, bem_ZA, en_ZA, ff_SN, sw_KE, sw_TZ, lb_LU, wae_CH,
3649 yue_HK, lij_IT, mhr_RU
3651 * New encodings: CP770, CP771, CP772, CP773, CP774
3653 * New program sotruss to trace calls through PLTs
3654 Implemented by Ulrich Drepper.
3656 * The malloc hook implementation is marked deprecated and will be removed
3657 from the default implementation in the next version. The design never
3658 worked ever since the introduction of threads. Even programs which do
3659 not create threads themselves can use multiple threads created internally.
3663 * The following bugs are resolved with this release:
3665 3268, 6812, 7066, 10085, 10484, 10563, 10851, 11149, 11155, 11577, 11611,
3666 11640, 11655, 11701, 11840, 11856, 11883, 11903, 11904, 11968, 11979,
3667 12005, 12037, 12067, 12077, 12078, 12092, 12093, 12107, 12108, 12113,
3668 12140, 12159, 12167, 12191, 12194, 12201, 12204, 12205, 12207, 12348,
3671 * New Linux interfaces: prlimit, prlimit64, fanotify_init, fanotify_mark
3673 * POWER7 optimizations: memset, memcmp, strncmp
3675 * New optimized string functions for x86-64: strnlen (SSE2),
3676 strcasecmp (SSE2, SSSE3, SSE4.2), strncasecmp (SSE2, SSSE3, SSE4.2)
3677 Implemented by Ulrich Drepper.
3681 * The following bugs are resolved with this release:
3683 3662, 4457, 5553, 10162, 10401, 10414, 10554, 10824, 10864, 10915,
3684 10918, 10936, 10939, 10958, 10968, 10969, 10972, 10992, 11000, 11001,
3685 11007, 11010, 11027, 11039, 11040, 11041, 11043, 11046, 11056, 11070,
3686 11093, 11115, 11125, 11126, 11127, 11134, 11141, 11149, 11183, 11184,
3687 11185, 11186, 11187, 11188, 11189, 11190, 11191, 11192, 11193, 11194,
3688 11200, 11230, 11235, 11242, 11254, 11258, 11271, 11272, 11276, 11279,
3689 11287, 11292, 11319, 11332, 11333, 11387, 11389, 11390, 11394, 11397,
3690 11410, 11438, 11449, 11470, 11471, 11520, 11537, 11538, 11571
3692 * New interfaces: pthread_getname_np, pthread_setname_np
3694 * New Linux interface: recvmmsg
3696 * STT_GNU_IFUNC implemented for Sparc by David Miller.
3698 * The dynamic linker now recognizes supported ABI versions from the
3699 EI_ABIVERSION field in the ELF header.
3700 Implemented by Ulrich Drepper.
3702 * New NIS mode selector ADJUNCT_AS_SHADOW. The passwd.adjunct.byname table
3703 will not be used to fill in password fields in the passwd.byname replies.
3704 Instead it is used to synthesize the shadow.byname table, should it be
3705 missing. This is a useful mode in some installations involving Solaris.
3706 Implemented by Ulrich Drepper.
3708 * New locales: kok_IN, sq_MK, cv_RU
3712 * The following bugs are resolved with this release:
3714 3270, 7094, 9924, 9986, 10011, 10085, 10107, 10159, 10162, 10166, 10183,
3715 10195, 10196, 10200, 10203, 10207, 10211, 10213, 10217, 10221, 10229,
3716 10262, 10286, 10312, 10315, 10319, 10349, 10360, 10391, 10402, 10416,
3717 10418, 10422, 10425, 10446, 10448, 10452, 10540, 10553, 10560, 10564,
3718 10609, 10643, 10692, 10713, 10717, 10730, 10731, 10742, 10780, 10784,
3719 10789, 10817, 10823, 10840, 10847
3721 * New interfaces: execvpe, pthread_sigqueue, mkstemps, mkstemps64, mkostemps,
3723 Implemented by Ulrich Drepper.
3725 * Checking version of longjmp added that fails if an uninitialized stack
3726 frame would be created. Implemented by Ulrich Drepper.
3728 * STT_GNU_IFUNC is now supported in static executables.
3729 Implemented by H.J. Lu.
3731 * STT_GNU_IFUNC implemented for PPC by Alan Modra.
3733 * New optimized string functions for x86-64: strstr, strcasestr, memcmp,
3734 strcspn, strpbrk, strspn, strcpy, stpcpy, strncpy, strcmp (SSE2, SSE4.2),
3735 strncmp (SSE2, SSE4.2), strchr (SSE4.2), strrchr (SSE4.2).
3736 Contributed by H.J. Lu.
3738 strlen, rawmemchr, strcmp (SSSE3), strncmp (SSSE3).
3739 Implemented by Ulrich Drepper.
3741 * New optimized string functions for x86: strlen, strcspn, strspn, strpbrk,
3743 Contributed by H.J. Lu.
3745 * Support for fma instruction in AVX on x86-64.
3746 Implemented by H.J. Lu and Ulrich Drepper.
3748 * AVX support in x86-64 auditing support in ld.so.
3749 Implemented by H.J. Lu.
3751 * STB_GNU_UNIQUE support added.
3752 Implemented by Ulrich Drepper.
3754 * Implement second fallback mode for optimized DNS lookup for even more
3755 broken environments. If two requests from the same port are not
3756 handled correctly close the socket and open a new one before sending
3757 the second request. The 'single-request-reopen' option in /etc/resolv.conf
3758 can be used to select this mode right away, instead of rediscovering the
3759 necessity is every process again.
3760 Implemented by Ulrich Drepper.
3762 * New resolver flag RES_USE_DNSSEC to enable use of verified lookup.
3763 Implemented by Adam Tkac.
3765 * Optimized iconv conversions for S390x.
3766 Implemented by Andreas Krebbel.
3768 * Using condvars with PI mutexes is now more efficient due to kernel
3769 support for requeueing to PI futexes. NPTL support added for x86-64.
3770 Implemented by Ulrich Drepper.
3772 * New locales: ps_AF, my_MM
3776 * The following bugs are resolved with this release:
3778 697, 5381, 5807, 6411, 6545, 7040, 7058, 7067, 7080, 7083, 7095, 7098,
3779 9677, 9694, 9697, 9701, 9704, 9705, 9706, 9720, 9726, 9730, 9733, 9736,
3780 9741, 9750, 9753, 9759, 9781, 9791, 9793, 9823, 9833, 9844, 9880, 9881,
3781 9891, 9893, 9895, 9913, 9916, 9920, 9948, 9955, 9957, 9963, 9985, 10052,
3782 10069, 10086, 10087, 10090, 10092, 10093, 10100, 10118, 10128
3784 * New interfaces: preadv, preadv64, pwritev, pwritev64, malloc_info
3785 Implemented by Ulrich Drepper.
3787 * New Linux interfaces: accept4, fallocate, fallocate64.
3788 Implemented by Ulrich Drepper.
3790 * Correct declarations of string function when used in C++ code. This
3791 could lead to compile errors for invalid C++ code.
3793 * XPG7/POSIX 2008 compilation environment. Many GNU-specific functions are
3796 * New POSIX 2008 interface: psiginfo
3797 Implemented by Ulrich Drepper.
3799 * New ISO C++1x interfaces: quick_exit, at_quick_exit
3800 Implemented by Ulrich Drepper.
3802 * Support for selecting between multiple function definitions at runtime
3803 using STT_GNU_IFUNC symbols. Implemented by Ulrich Drepper.
3805 * The libcrypt library can now use the hash function implementations in
3806 NSS. Implemented by Ulrich Drepper.
3808 * The malloc implementation can be compiled to be less memory efficient
3809 but higher performing in multi-threaded programs.
3810 Implemented by Ulrich Drepper.
3812 * New locales: nan_TW@latin, ks_IN
3814 * Faster strlen, strchr, strchrnul, strrchr, memchr, and rawmemchr for x86-64.
3815 Implemented by Ulrich Drepper.
3817 * Extended printf hook support. It is possible to use user-defined types
3818 and extend existing format specifiers.
3819 Implemented by Ulrich Drepper.
3821 * Handling for group shadow files has been added.
3822 Implemented by Ulrich Drepper.
3824 * DNS IPv4-IPv6 parallel lookup now deals better with broken DNS
3825 servers (the case, e.g., for some people using the built-in DNS
3826 server in ADSL modems/routers). There is a once-per-process timeout
3827 in case of a broken server. To avoid it, users can run nscd or put
3828 'options single-request' in /etc/resolv.conf.
3829 Implemented by Ulrich Drepper.
3833 * The following bugs are resolved with this release:
3835 3406, 5209, 5210, 5381, 5794, 5814, 5911, 6428, 6438, 6442, 6447, 6461,
3836 6472, 6505, 6544, 6589, 6612, 6634, 6653, 6654, 6657, 6698, 6712, 6713,
3837 6719, 6723, 6724, 6740, 6763, 6771, 6790, 6791, 6817, 6824, 6839, 6843,
3838 6845, 6856, 6857, 6860, 6867, 6875, 6919, 6920, 6942, 6947, 6955, 6966,
3839 6968, 6974, 6980, 6995, 7008, 7009, 7029
3841 * Unified lookup for getaddrinfo: IPv4 and IPv6 addresses are now looked
3842 up at the same time. Implemented by Ulrich Drepper.
3844 * TLS descriptors for LD and GD on x86 and x86-64.
3845 Implemented by Alexandre Oliva.
3847 * getaddrinfo now handles DCCP and UDPlite.
3848 Implemented by Ulrich Drepper.
3850 * New fixed-size conversion macros: htobe16, htole16, be16toh, le16toh,
3851 htobe32, htole32, be32toh, le32toh, htobe64, htole64, be64toh, le64toh.
3852 Implemented by Ulrich Drepper.
3854 * New implementation of memmem, strstr, and strcasestr which is O(n).
3855 Implemented by Eric Blake.
3857 * New Linux interfaces: inotify_init1, dup3, epoll_create1, pipe2
3859 * Implement "e" option for popen to open file descriptor with the
3860 close-on-exec flag set. Implemented by Ulrich Drepper.
3862 * Implement "b" mode for fmemopen. In this mode writes writes don't
3863 implicitly add a NUL byte and seeks from the end of the buffer really
3864 use the buffer end, not the string length as the basis.
3865 Implemented by Ulrich Drepper.
3867 * Many functions, exported and internal, now atomically set the close-on-exec
3868 flag when run on a sufficiently new kernel. Implemented by Ulrich Drepper.
3870 * Sorting rules for some Indian languages (Kannada, Gurumukhi, Telugu,
3872 Implemented by Pravin Satpute.
3874 * New locales: sd_IN, sd_IN@devanagari, ks_IN@devanagari, ht_HT, en_AG, nl_AW.
3878 * The following bugs are resolved with this release:
3880 2549, 3112, 4314, 4407, 4997, 5012, 5040, 5112, 5204, 5208, 5209, 5220,
3881 5222, 5224, 5225, 5237, 5238, 5240, 5245, 5277, 5346, 5375, 5378, 5382,
3882 5424, 5427, 5428, 5435, 5436, 5439, 5441, 5442, 5443, 5451, 5452, 5454,
3883 5455, 5463, 5464, 5465, 5475, 5477, 5478, 5520, 5541, 5545, 5553, 5600,
3884 5602, 5607, 5614, 5627, 5628, 5736, 5737, 5741, 5753, 5760, 5762, 5768,
3885 5774, 5777, 5778, 5779, 5786, 5790, 5818, 5854, 5857, 5903, 5939, 5979,
3886 5983, 5995, 5998, 6004, 6007, 6020, 6021, 6024, 6040, 6042
3888 * New locales: bo_CN, bo_IN, shs_CA.
3890 * New encoding: HP-ROMAN9, HP-GREEK8, HP-THAI8, HP-TURKISH8.
3892 * Sorting rules for some Indian languages (Devanagari and Gujarati).
3893 Implemented by Pravin Satpute.
3895 * IPV6 addresses in /etc/resolv.conf can now have a scope ID
3897 * nscd caches now all timeouts for DNS entries
3898 Implemented by Ulrich Drepper.
3900 * nscd is more efficient and wakes up less often.
3901 Implemented by Ulrich Drepper.
3903 * More checking functions: asprintf, dprintf, obstack_printf, vasprintf,
3904 vdprintf, and obstack_vprintf.
3905 Implemented by Jakub Jelinek.
3907 * Faster memset for x86-64.
3908 Implemented by Harsha Jagasia and H.J. Lu.
3910 * Faster memcpy on x86.
3911 Implemented by Ulrich Drepper.
3913 * ARG_MAX is not anymore constant on Linux. Use sysconf(_SC_ARG_MAX).
3914 Implemented by Ulrich Drepper.
3916 * Faster sqrt and sqrtf implemention for some PPC variants.
3917 Implemented by Steven Munroe.
3921 * The following bugs are resolved with this release:
3923 4125, 4126, 4178, 4359, 4407, 4512, 4514, 4525, 4554, 4556, 4557, 4566,
3924 4582, 4586, 4588, 4599, 4610, 4647, 4702, 4705, 4726, 4745, 4772, 4773,
3925 4775, 4776, 4792, 4813, 4814, 4816, 4833, 4858, 4860, 4896, 4905, 4925,
3926 4936, 4937, 4938, 4941, 4946, 4963, 4972, 5010, 5028, 5043, 5058, 5063,
3927 5071, 5103, 5104, 5112, 5113, 5184, 5186
3929 * More checking functions: fread, fread_unlocked, open*, mq_open.
3930 Implemented by Jakub Jelinek and Ulrich Drepper.
3932 * Extend fortification to C++. Implemented by Jakub Jelinek.
3934 * Implement 'm' modifier for scanf. Add stricter C99/SUS compliance
3935 by not recognizing 'a' as a modifier when those specs are requested.
3936 Implemented by Jakub Jelinek.
3938 * PPC optimizations to math and string functions.
3939 Implemented by Steven Munroe.
3941 * New interfaces: mkostemp, mkostemp64. Like mkstemp* but allow additional
3942 options to be passed. Implemented by Ulrich Drepper.
3944 * More CPU set manipulation functions. Implemented by Ulrich Drepper.
3946 * New Linux interfaces: signalfd, eventfd, eventfd_read, and eventfd_write.
3947 Implemented by Ulrich Drepper.
3949 * Handle private futexes in the NPTL implementation.
3950 Implemented by Jakub Jelinek and Ulrich Drepper.
3952 * Add support for O_CLOEXEC. Implement in Hurd. Use throughout libc.
3953 Implemented by Roland McGrath and Ulrich Drepper.
3955 * Linux/x86-64 vDSO support. Implemented by Ulrich Drepper.
3957 * SHA-256 and SHA-512 based password encryption.
3958 Implemented by Ulrich Drepper.
3960 * New locales: ber_DZ, ber_MA, en_NG, fil_PH, fur_IT, fy_DE, ha_NG, ig_NG,
3961 ik_CA, iu_CA, li_BE, li_NL, nds_DE, nds_NL, pap_AN, sc_IT, tk_TM, ug_CN,
3964 + New iconv modules: MAC-CENTRALEUROPE, ISO-8859-9E, KOI8-RU.
3965 Implemented by Ulrich Drepper.
3969 * The following bugs are resolved with this release:
3971 3156, 3213, 3285, 3291, 3306, 3313, 3320, 3322, 3325, 3326, 3334, 3348,
3972 3352, 3363, 3369, 3426, 3427, 3429, 3451, 3458, 3465, 3480, 3483, 3493,
3973 3514, 3515, 3559, 3632, 3664, 3673, 3674, 3745, 3747, 3818, 3840, 3842,
3974 3851, 3855, 3884, 3885, 3902, 3905, 3919, 3944, 3954, 3955, 3957, 3991,
3975 3995, 3996, 4024, 4040, 4069, 4070, 4074, 4076, 4096, 4101, 4102, 4114,
3976 4130, 4131, 4181, 4276, 4306, 4342, 4344, 4349, 4364, 4368, 4381, 4392,
3977 4403, 4405, 4406, 4411, 4438, 4439, 4455, 4456, 4465, 4512, 4514, 4586,
3980 * New Linux interfaces: epoll_pwait, sched_getcpu.
3982 * New generic interfaces: strerror_l.
3984 * nscd can now cache the services database. Implemented by Ulrich Drepper.
3988 * The following bugs are resolved with this release:
3990 39, 68, 192, 352, 388, 531, 935, 1006, 1201, 1203, 1386, 1782, 1783, 1784,
3991 1785, 1786, 1787, 1951, 1974, 1996, 2072, 2098, 2125, 2134, 2158, 2182,
3992 2349, 2376, 2386, 2389, 2415, 2418, 2420, 2423, 2450, 2451, 2466, 2472,
3993 2473, 2477, 2498, 2499, 2501, 2502, 2505, 2507, 2508, 2509, 2517, 2526,
3994 2569, 2571, 2592, 2611, 2625, 2632, 2644, 2662, 2680, 2683, 2684, 2693,
3995 2695, 2703, 2734, 2751, 2764, 2766, 2775, 2792, 2821, 2823, 2832, 2841,
3996 2843, 2883, 2892, 2908, 2914, 2926, 2961, 2978, 2980, 2987, 2997, 2998,
3997 3013, 3018, 3034, 3040, 3044, 3123, 3124, 3137, 3138, 3143, 3155, 3189,
3998 3225, 3251, 3252, 3253, 3273
4000 * For Linux, the sorting of addresses returned by getaddrinfo now also
4001 handles rules 3, 4, and 7 from RFC 3484. I.e., all rules are handled.
4002 Implemented by Ulrich Drepper.
4004 * Allow system admin to configure getaddrinfo with the /etc/gai.conf file.
4005 Implemented by Ulrich Drepper.
4007 * New Linux interfaces: splice, tee, sync_file_range, vmsplice.
4009 * New iconv module for MIK. Contributed by Alexander Shopov.
4011 * For sites with broken group and/or passwd database, the auto-propagate
4012 option of nscd can prevent creating ID lookup entries from the results
4013 of a name lookup and vice versa. This usually is no problem but some
4014 site might have problems with the default behavior.
4015 Implemented by Ulrich Drepper.
4017 * Iterating over entire database in NIS can be slow. With the
4018 SETENT_BATCH_READ option in /etc/default/nss a system admin can decide
4019 to trade time for memory. The entire database will be read at once.
4020 Implemented by Ulrich Drepper.
4022 * The interfaces introduced in RFC 3542 have been implemented by
4025 * Support for the new ELF hash table format was added by Ulrich Drepper.
4027 * Support for priority inheritance mutexes added by Jakub Jelinek and
4030 * Support for priority protected mutexes added by Jakub Jelinek.
4032 * New locales: nr_ZA, pa_PK, ca_AD, ca_FR, ca_IT, el_CY, tr_CY, as_IN,
4033 or_IN, csb_PL, fy_NL, sr_ME.
4037 * More overflow detection functions.
4039 * New iconv converters for IBM1025, IBM1122, IBM1137, IBM1153,
4040 IBM1154, IBM1155, IBM1156, IBM1157, and IBM1158 contributed by Jiro Sekiba.
4042 More converters for IBM803, IBM901, IBM902, IBM921, IBM1008, IBM1097,
4043 IBM1112, IBM1123, IBM1130, IBM1140, IBM1141, IBM1142, IBM1143, IBM1144,
4044 IBM1145, IBM1146, IBM1147, IBM1148, IBM1149, IBM1166, IBM1167, IBM4517,
4045 IBM4899, IBM4909, IBM4971, IBM5347, IBM9030, IBM9066, IBM9448, IBM12712,
4046 IBM16804, IBM1364, IBM1371, IBM1388, IBM1390, and IBM1399 contributed
4047 by Masahide Washizawa.
4049 * It is now possible to install audit libraries for the dynamic linker using
4050 LD_AUDIT. Implemented by Ulrich Drepper.
4052 * The LinuxThreads add-on, providing pthreads on Linux 2.4 kernels, is no
4053 longer supported. The new NPTL implementation requires Linux 2.6 kernels.
4054 For a libc and libpthread that works well on Linux 2.4 kernels, we
4055 recommend using the stable 2.3 branch.
4057 * The new function open_wmemstream acts like open_memstream,
4058 but uses a wchar_t wide character buffer.
4060 * The new function ppoll is an improved version of poll, similar to pselect.
4062 * New interfaces: fdopendir, openat, fstatat, fchmodat, fchownat,
4063 futimesat, faccessat, mkdirat, mkfifoat, mknodat,
4064 renameat, unlinkat, linkat, symlinkat, readlinkat.
4066 * New Linux kernel system calls: unshare,
4067 inotify_init, inotify_add_watch, inotify_rm_watch.
4069 * The euidaccess function is now also known by the alias eaccess,
4070 for compatibility with some other systems.
4072 * Timezone data updated to 2006b version.
4076 * The following bugs are resolved with this release:
4078 38, 253, 549, 622, 653, 721, 758, 851, 877, 915, 934, 955, 961,
4079 1016, 1037, 1076, 1079, 1080, 1081, 1082, 1083, 1084, 1085, 1086,
4080 1087, 1088, 1090, 1091, 1092, 1093, 1094, 1095, 1096, 1097, 1098,
4081 1099, 1100, 1101, 1102, 1103, 1104, 1105, 1106, 1107, 1108, 1109,
4082 1110, 1111, 1112, 1113, 1125, 1137, 1138, 1249, 1250, 1251, 1252,
4083 1253, 1254, 1350, 1358, 1394, 1438, 1498, 1534
4085 Visit <http://sources.redhat.com/bugzilla/> for the details of each bug.
4087 * As of this release, GCC 4 can be used to compile the C Library.
4089 * Timezone data updated to 2005m version.
4093 * The following bugs are resolved with this release:
4095 284, 592, 602, 626, 633, 640, 650, 661, 671, 681, 693, 700, 710, 719,
4096 722, 723, 725, 726, 727, 728, 729, 730, 731, 732, 733, 734, 735, 736,
4097 737, 738, 739, 740, 741, 742, 743, 744, 745, 765, 767, 768, 769, 776,
4098 777, 787, 821, 822, 823, 825
4100 Visit <http://sources.redhat.com/bugzilla/> for the details of each bug.
4104 * Support for RFC 3678. Real implementations exist only for Linux so far.
4105 Implemented by Ulrich Drepper.
4107 * nscd can now cache entries persistently. Expiring entries are reloaded.
4108 For speedups the cache can be shared in memory with client processes.
4109 Implemented by Ulrich Drepper.
4111 * nscd can now perform SELinux checks.
4112 Implemented by Matthew Rickard <mjricka@epoch.ncsc.mil>.
4114 * getaddrinfo queries are now cached. Canonical name lookup is performed
4116 Implemented by Ulrich Drepper.
4118 * The nothrow function attribute is used when headers are used by gcc when
4119 compiling C code. This can avoid significant amounts of exception
4122 * The malloc functions perform more error checking and are stricter when
4123 it comes to reacting on errors. The default action is to terminate
4124 the process after showing an error message. Implemented by Ulrich Drepper.
4126 * Reverse lookups of IPv6 addresses does not use bit string or .ip6.int
4127 lookups anymore unless explicitly requested. Implemented by Ulrich Drepper.
4129 * Namespaces in ld.so are implemented. DSOs can be loaded in separate
4130 namespaces using the new function dlmopen(). This feature is of course,
4131 like most other dynamic loading functionality, not available in statically
4132 linked applications. Implemented by Ulrich Drepper.
4134 * Low-overhead boundary checking variants of string and some stdio functions
4135 were added. These are to be used in conjunction with a gcc patch by
4136 Jakub Jelinek which adds calls to these functions if possible.
4137 Implemented by Jakub Jelinek and Ulrich Drepper.
4139 * Old code for several operating systems and machine architectures that
4140 have not been in working condition in a long time have been removed from
4141 the main source tree maintained by the GNU C Library's maintainers.
4142 These files are now reside in the separate `ports' source module
4143 that is usable as an add-on when building the library.
4147 * New functions `dladdr1' and `dlinfo' in <dlfcn.h> provide more ways to
4148 interrogate the dynamic linker, compatible with the Solaris interface.
4150 * ELF thread-local storage support (TLS) now works on PowerPC and PowerPC64;
4151 implemented by Paul Mackerras, Steven Munroe, and Roland McGrath.
4153 * getifaddrs now uses the netlink interface on Linux to get its information.
4154 Implemented by Thorsten Kukuk.
4156 * getaddrinfo now implements AI_V4MAPPED, AI_ALL, and AI_ADDRCONF.
4157 Implemented by Ulrich Drepper.
4159 * support for non-executable stacks on x86 has been added. Changes mostly
4162 * regex is now much faster for multibyte locales. Changes by Jakub Jelinek
4165 * getaddrinfo now performs destination address selection according to
4170 * Thread-safe interfaces for many functions that access locale data
4171 were added in version 2.3, but these features were omitted from NEWS.
4172 Many functions have variants with an `_l' suffix that take a `locale_t'
4173 object as a parameter rather than consulting the current locale.
4174 The new functions `newlocale', `duplocale', and `freelocale' in <locale.h>
4175 create and maintain `locale_t' objects. Additionally, the new function
4176 `uselocale' sets "the current locale" (as used by functions not so
4177 parameterized) set for an individual thread. These features were added
4178 in version 2.3, implemented by Ulrich Drepper and Roland McGrath.
4180 * The functions getresuid, getresgid, setresuid, and setresgid, which
4181 have long been available on Linux, are now declared in <unistd.h>
4182 and are now also available on the Hurd.
4184 * ELF thread-local storage support (TLS) now works on x86-64.
4186 * The new dynamic string token $LIB is expanded in shared library names.
4187 This normally expands to lib, but on some 64-bit platforms to lib64 instead.
4189 * Aldy Hernandez contributed complete software floating point support for
4190 PowerPC machines with no FPU.
4192 * fexecve is implemented on Linux.
4194 * The `btowc' function should work at least twice as fast due to
4195 specialized callbacks in the iconv modules. Implemented by Bruno Haible.
4197 * With approriate thread add-ons cancelable functions are now implemented
4198 in libc.so as well. No need to call the function in libpthread. This
4199 change allowed to finally disable the incorrect and expensive handling
4200 of weak definition in ld.so.
4202 * Yet more PLT entries in libc.so have been removed. We finally arrived
4203 at the bare minimum. Startup times improved appropriately.
4205 * Support for the new Linux/x86 system call interface was added. The
4206 AT_SYSINFO auxiliary vector entry is recognized and handled.
4210 * Masahide Washizawa contributed iconv modules for IBM1163 and IBM1164
4213 * iconv (the program and the interface) now accepts empty names (excluding
4214 options like //TRANSLIT) to mean "use charset of current locale".
4216 * localedef can now transliterate characters in strings which are not in
4217 the provided charmap. The information from the input locale is used.
4219 * Prelinking support was added for ELF targets. This requires additional
4220 tools and recent versions of the GNU binutils. Contributed by Jakub Jelinek.
4222 * Read-only stdio streams now use mmap to speed up operation by eliminating
4223 copying and buffer underflows. To use add 'm' to the mode string of
4224 the fopen/fdopen/freopen call. Implemented by Ulrich Drepper.
4226 * The malloc functions were completely rewritten by Wolfram Gloger based
4227 on Doug Lea's malloc-2.7.0.c.
4229 * Isamu Hasegawa contributed a completely new and POSIX-conformant
4230 implementation of regex.
4232 * Bruno Haible upgraded the iconv and locale implementation to support
4235 * Contents of the LC_* and LANG environment variables in the CEN style are
4236 not recognized anymore. It never was used. Change by Ulrich Drepper.
4238 * The runtime (ld.so, libc, libpthread for Linux) now can handle the ELF
4239 thread-local storage (TLS) ABI on some platforms.
4240 Changes by Ulrich Drepper. SH support by Kaz Kojima.
4242 * Bruno Haible contributed iconv converters for ISO-2022-JP-3, SHIFT JIS-X0213,
4243 EUC-JISX0213, and TSCII.
4245 * New header <ifaddrs.h> with functions `getifaddrs' and `freeifaddrs':
4246 BSD-compatible interface for getting all network interface addresses.
4247 Implementation for IPv4 by Roland McGrath.
4249 * Loading of locale data is faster due to the introduction of a locale
4250 archive. Implemented by Roland McGrath and Ulrich Drepper.
4252 * Startup times are significantly reduced by not using exported functions
4253 inside the library itself. Changes by Jakub Jelinek, Roland McGrath,
4256 * Steven Munroe contributed a port to PowerPC64/Linux.
4260 * The Hurd now uses the GNU libio implementation of stdio rather than the
4261 old GNU stdio implementation, and uses a new ABI (libc.so.0.3).
4263 * The Hurd on x86 now has the `ioperm' function and <sys/io.h> header file
4264 with the same behavior as the Linux system call of the same name.
4268 * Stephen Moshier implemented log2, log10, powl and cbrtl for the
4269 128-bit long double format.
4271 * Masahide Washizawa contributed iconv modules for IBM1132, IBM1133, IBM1160,
4272 IBM1161, and IBM1162 charsets.
4274 * Andreas Jaeger contributed a port to x86-64/Linux.
4276 * Peter Bruin contributed a port to PowerPC/Hurd.
4278 * libc functions using I/O streams now can handle wide-oriented streams
4281 * optimizations in the dynamic linker. Binaries created by recent binutils
4282 versions start up quicker due to reduced time spent on relocations.
4284 * Support for use of gcc3 added by Jakub Jelinek and HJ Lu.
4288 * Stephen Moshier implemented cosh, expm1, log1p, acos, sinh, tanh,
4289 asinh, atanh, j0 for the 128-bit long double format.
4291 * Bruno Haible updated all the code handling Unicode in some form to
4292 support Unicode 3.1.
4294 * Speed of regex for single-byte locales is back to previous levels.
4295 Patch by Isamu Hasegawa.
4297 * Alpha, SPARC, and IA-64 now also using floating stacks.
4299 * Startup time of internationalized applications greatly improved through
4300 iconv cache. Use iconvconfig to generate the cache file.
4301 Contributed by Ulrich Drepper.
4303 * The IA-64 specific part of ld.so was rewritten to eliminate some pretty
4304 severe performance problems. Patch by David Mosberger.
4306 * The Hurd port got a lot more functionality like AIO, various stdio
4307 extensions, etc. Mainly done by Roland McGrath.
4309 * mtrace can now lookup symbols in shared libraries.
4313 * Intel's IA-64 math library is largely integrated. It provides fast and
4314 accurate implementatations for most basic and standard math functions
4315 in float, double, and long double format.
4317 * Stephen Moshier implemented j0, j1, jn, y0, y1, yn, lgamma, erf, erfc,
4318 and asin for the 96-bit long double format and asin, log, tan for the
4319 128-bit long double format.
4321 * The beginning of a last-bit accurate math library by IBM Haifa were added.
4322 The basic double functions exist today. Contributed by Abraham Ziv
4323 <ziv@il.ibm.com>, Moshe Olshansky <olshansk@il.ibm.com>, Ealan Henis
4324 <ealan@il.ibm.com>, and Anna Reitman <reitman@il.ibm.com>.
4326 * An asynchronous name lookup library was added. The interface is designed
4327 after POSIX AIO. The proposal was circulated beforehand to get comments.
4328 No negative ones came in. Implemented by Ulrich Drepper.
4330 * Port to S390/64bit contributed by Martin Schwidefsky
4331 <schwidefsky@de.ibm.com>.
4333 * David Mosberger <davidm@hpl.hp.com> implemented the setcontext family
4334 of functions for Linux/IA-64.
4336 * The RPC code is now thread safe. Threads can now use the same service
4337 of different services at the same time. Patch by Eric Norum
4338 <eric.norum@usask.ca> with some help by Ulrich Drepper.
4340 * Martin Schwidefsky <schwidefsky@de.ibm.com> implemented the setcontext
4341 family of functions for Linux/S390.
4343 * Ulrich Drepper <drepper@redhat.com> implemented the setcontext family
4344 of functions for Linux/x86.
4346 * Port to Linux/CRIS contributed by Axis Communications.
4350 * Lots of headers were cleaned up. Using the tool in the conform/ subdir
4351 we can now check for namespace violations and missing declarations. The
4352 result is that almost all headers are now Unix-compliant (as defined in
4353 the upcoming XPG6). The negative side is that some programs might need
4354 corrections, too, if they depend on the incorrect form of the headers in
4355 previous versions which defined too many symbols and included too many
4358 * regex now handles multibyte character sets correctly.
4359 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
4361 * iconv (the program) does now conform to the upcoming XPG6 and handles
4362 charmaps. Instead of the charset names the path of charmaps can be
4363 provided and the conversion happens based on this data.
4364 Contributed by Ulrich Drepper.
4366 * The locale program now provides detailed information about the installed
4371 only lists the names of the supported locales
4375 provides details such as country, language, and codeset name.
4376 Contributed by Ulrich Drepper.
4380 * The gencat program now parses the input file according to the charset
4381 selected by the LC_CTYPE category. This is important for stateful
4382 character sets. To make generating catalogs easier there is a way
4383 to overwrite the charset selected by the locale: before the first
4384 message or $ quote line the catalog can contain a line like
4386 $ codeset=ISO-8859-2
4388 to select the charset (ISO-8859-2 in this case).
4390 Implemented by Shinya Hanataka and Ulrich Drepper.
4392 * New codeset conversion modules: IBM-922 (Estonia PC codeset),
4393 IBM-1124 (Ukraine PC codeset), IBM-1129 (Vietnamese PC codeset).
4394 Contributed by Masahide Washizawa <washi@jp.ibm.com>.
4396 * Optimized string functions for Alpha ev6 and ev67 by Richard Henderson
4397 <rth@redhat.com> and Rick Gorton <rick.gorton@alpha-processor.com>.
4399 * The LANGUAGE environment variable is now ignored unless the locale is
4400 changed from the default "C" locale.
4402 * The usual bug fixes.
4406 * Greg McGary added runtime support for bounds checking using gcc's
4407 new -fbounded-pointers option. ix86 target is complete. PowerPC
4410 * Thorsten Kukuk added secure mode support to nscd.
4412 * The Berkeley db libraries have been removed.
4414 Related, the nss_db module is now in a separate package since it
4415 obviously requires a database library being available.
4417 * Wide character I/O streams implemented by Ulrich Drepper.
4419 * Functions from the extended socket API added by Ulrich Drepper.
4421 * Functions feenableexcept and fedisableexcept to control the
4422 behaviour of individual exceptions have been added by Andreas Jaeger.
4424 * ldconfig program added by Andreas Jaeger and Jakub Jelinek.
4426 * The resolver code has been updated from bind 8.2.3-T5B which supports
4427 threads. The integration was done by Andreas Jaeger, Adam D. Bradley,
4430 This change could in some situations effect backward compatibility. Since
4431 now `_res' is a thread-local instead of a global variable, modifying it
4432 in one thread does not have any effect in other threads.
4434 The resolver library was also extended to allow IPv6 as the transport
4435 protocol for the requests. This work was done by Stig Venaas.
4437 * Compatibility code for K&R C compilers has been removed from the
4438 header files. A ISO C compiler is needed to use the library
4439 (conforming to either C89 or C99 standard).
4441 * Complete rewrite of the localedef program to support multibyte character
4442 sets. Implement handling of ISO 14651 and ISO 14652. Rewrite strcoll,
4443 strxfrm, wcscoll, and wcsxfrm functions. Make isw*() functions work.
4444 Implemented by Ulrich Drepper.
4446 Bruno Haible significantly improved the generation and use of the data
4447 structures for the wide character tables.
4449 * Plural handling in gettext implemented by Ulrich Drepper.
4451 * The utmp daemon has been removed.
4453 * The port to MIPS-Linux has been finished by Andreas Jaeger.
4455 * A port to Hitachi SH3 and SH4 has been contributed by Kazumoto Kojima
4458 * POSIX clocks and timers implemented by Kaz Kylheku and Ulrich Drepper.
4460 * POSIX barriers implemented by Kaz Kylheku.
4462 * POSIX spawn function family implemented by Ulrich Drepper.
4464 * POSIX spinlocks are now available.
4466 * Timed wait functions for mutex, rwlock, and semaphores are implemented.
4468 * the configure option --enable-kernel=X.Y.Z allows to strip out
4469 compatibility for kernel versions before X.Y.Z. This is currently only
4470 implemented for Linux.
4472 * the sockaddr_in6 structure changed. The IPv6 working group added a new
4473 field sin6_scope_id. This means that all programs using IPv6 should be
4474 recompiled. Don't expect binary compatibility with previous glibc
4477 * various conversion modules for IBM character sets contributed by
4480 * IA-64 port by Jes Sorensen and HJ Lu.
4494 * New ISO C 9x function _Exit, imaxabs, and imaxdiv are added.
4496 * New xdr functions are added; some rpc functions are now 64bit clean.
4498 * Fixed a number of bugs and memory leaks (especially in NIS+ code).
4500 * Fixed known incompatibilities with glibc 2.0.
4502 * New functions lock64, strchrnul, rawmemchr, getutmp and getutmpx.
4504 * Optimized a number of functions (especially the ELF dynamic loader).
4506 * Update timezone data files.
4508 * lots of charmaps corrections
4510 * some new locale definitions and charmaps
4515 * Richard Henderson corrected size of struct timeval on Linux/Alpha to
4516 conform to POSIX member type requirements. Symbol versions have been
4517 adjusted as needed within the library, and for direct use by applications,
4518 but there is potential for problems if third-party libraries use
4519 struct timeval as part of their interface. This does not present
4520 a problem for X and other "essential" system libraries.
4522 * An additional locale model to support C++ Standard Library locale
4523 model and probably more was implemented by Ulrich Drepper.
4525 * Eric Youngdale and Ulrich Drepper implemented versioning of objects on
4528 * Miles Bader provided the `argp' function family to support hierachical
4529 command line argument parsing, layered on top of getopt.
4531 * strtod accepts new hexadecimal floating-point format from ISO C 9X.
4533 * printf has two new specifiers %a and %A to print hexadecimal flaoting-point
4536 * scanf recognizes the %a and %A format for scanning floating point numbers.
4538 * the new headers <stdint.h> and <inttypes.h> from ISO C 9X provides
4539 information and interfaces for the available integer types.
4541 * about 130 new math functions were added to implement the ISO C9x math
4544 * the new header <complex.h> contains definitions of the complex math
4545 functions from ISO C 9X.
4547 * the new header <tgmath.h> defines generic macros to use complex or
4548 real valued functions.
4550 * Thorsten Kukuk provided an implementation for NIS+, securelevel 0, 1 and 2.
4552 * Andreas Jaeger provided a test suite for the math library.
4554 * Mark Kettenis implemented the utmpx interface and an utmp daemon.
4556 * Ulrich Drepper added character set conversion functions (iconv).
4558 * Optimized string functions have been added.
4560 * The localedata addon is now part of glibc.
4562 * An implementation of profiling shared libraries was added by Ulrich Drepper.
4564 * Thorsten Kukuk and Ulrich Drepper provided an implementation for a caching
4565 daemon for NSS (nscd).
4567 Missing a better place here are some numbers on improvements. Under
4568 Linux 2.1.125 un-tar-ing the kernel sources takes
4572 using local files 12.19s 6.88s 22.91s
4574 using NIS 13.92s 8.91s 26.34s
4576 using NIS & nscd 10.37s 7.34s 25.30s
4578 using NIS+ 27.57s 30.37s 640.46s
4580 using NIS+ & nscd 10.25s 7.83s 26.51s
4582 using NIS & old nscd [1] 13.83s 8.32s 29.60s
4584 Keep in mind that non-namelookup related operations dominate above times.
4585 It was just a common complain that using NIS+ unpacking the kernel is
4588 [1] The old nscd implementation is not available anymore since it was
4589 distributed with glibc up to version 2.0.98 and thus is now replaced.
4591 * Tim Waugh provided an implementation of the POSIX.2 wordexp function family.
4593 * Mark Kettenis provided a Hesiod NSS module.
4595 * The ELF dynamic loader knows how to process dynamic string tokens ($ORIGIN
4596 and $PLATFORM) in RPATHs and similar strings (Ulrich Drepper).
4598 * rcmd can now handle netgroups (Dick Streefland).
4600 * A port to the ARM was contributed by Phil Blundell, Pat Beirne and Scott
4603 * Support for the IPv6 protocol has been added to the socket API, as per the
4604 latest draft standards.
4606 * Support for Linux 2.2 has been added.
4608 * Interface changes relative to the latest 2.0.x release:
4609 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4610 addseverity NEW: Unix98
4611 alphasort64 NEW: LFS
4612 argp_err_exit_status NEW: argp, GNU ext
4613 argp_error NEW: argp, GNU ext
4614 argp_failure NEW: argp, GNU ext
4615 argp_help NEW: argp, GNU ext
4616 argp_parse NEW: argp, GNU ext
4617 argp_program_bug_address NEW: argp, GNU ext
4618 argp_program_version NEW: argp, GNU ext
4619 argp_program_version_hook NEW: argp, GNU ext
4620 argp_state_help NEW: argp, GNU ext
4621 argp_usage NEW: argp, GNU ext
4622 authdes_create NEW: Secure RPC
4623 authdes_getucred NEW: Secure RPC
4624 authdes_pk_create NEW: Secure RPC
4625 backtrace NEW: GNU ext.
4626 backtrace_symbols NEW: GNU ext.
4627 backtrace_symbols_fd NEW: GNU ext.
4629 cacosf NEW: ISO C 9x
4630 cacosh NEW: ISO C 9x
4631 cacoshf NEW: ISO C 9x
4632 cacoshl NEW: ISO C 9x
4633 cacosl NEW: ISO C 9x
4640 casinf NEW: ISO C 9x
4641 casinh NEW: ISO C 9x
4642 casinhf NEW: ISO C 9x
4643 casinhl NEW: ISO C 9x
4644 casinl NEW: ISO C 9x
4646 catanf NEW: ISO C 9x
4647 catanh NEW: ISO C 9x
4648 catanhf NEW: ISO C 9x
4649 catanhl NEW: ISO C 9x
4650 catanl NEW: ISO C 9x
4651 cbc_crypt NEW: Secure RPC
4655 ccoshf NEW: ISO C 9x
4656 ccoshl NEW: ISO C 9x
4662 cimagf NEW: ISO C 9x
4663 cimagl NEW: ISO C 9x
4664 clearerr_locked REMOVED
4665 clntunix_create NEW: sunrpc ext
4667 clog10 NEW: ISO C 9x
4668 clog10f NEW: ISO C 9x
4669 clog10l NEW: ISO C 9x
4679 cprojf NEW: ISO C 9x
4680 cprojl NEW: ISO C 9x
4682 crealf NEW: ISO C 9x
4683 creall NEW: ISO C 9x
4688 csinhf NEW: ISO C 9x
4689 csinhl NEW: ISO C 9x
4692 csqrtf NEW: ISO C 9x
4693 csqrtl NEW: ISO C 9x
4697 ctanhf NEW: ISO C 9x
4698 ctanhl NEW: ISO C 9x
4700 des_setparity NEW: Secure RPC
4701 ecb_crypt NEW: Secure RPC
4702 endutxent NEW: Unix98
4704 exp10f NEW: ISO C 9x
4705 exp10l NEW: ISO C 9x
4709 fattach NEW: STREAMS
4710 fdetach NEW: STREAMS
4714 feclearexcept NEW: ISO C 9x
4715 fegetenv NEW: ISO C 9x
4716 fegetexceptflag NEW: ISO C 9x
4717 fegetround NEW: ISO C 9x
4718 feholdexcept NEW: ISO C 9x
4720 feraiseexcept NEW: ISO C 9x
4721 ferror_locked REMOVED
4722 fesetenv NEW: ISO C 9x
4723 fesetexceptflag NEW: ISO C 9x
4724 fesetround NEW: ISO C 9x
4725 fetestexcept NEW: ISO C 9x
4726 feupdateenv NEW: ISO C 9x
4727 fflush_locked REMOVED
4731 fgets_unlocked NEW: GNU ext.
4732 fileno_locked REMOVED
4744 fputc_locked REMOVED
4745 fputs_unlocked NEW: GNU ext.
4746 fread_unlocked NEW: GNU ext.
4751 fstatvfs NEW: Unix98
4755 ftruncate64 NEW: LFS
4757 fwrite_unlocked NEW: GNU ext.
4758 gai_strerror NEW: IPv6
4762 getchar_locked REMOVED
4764 getdate_err NEW: Unix98
4765 getdate_r NEW: GNU ext.
4767 getnameinfo NEW: IPv6
4768 getnetname NEW: Secure RPC
4769 getpmsg NEW: STREAMS
4770 getpt NEW: Unix98 PTY
4771 getrlimit64 NEW: LFS
4772 getutxent NEW: Unix98
4773 getutxid NEW: Unix98
4774 getutxline NEW: Unix98
4775 glob64 NEW: GNU ext.
4776 globfree64 NEW: GNU ext.
4777 gnu_get_libc_release NEW: GNU ext.
4778 gnu_get_libc_version NEW: GNU ext.
4779 grantpt NEW: Unix98 PTY
4780 host2netname NEW: Secure RPC
4782 iconv_close NEW: iconv
4783 iconv_open NEW: iconv
4784 if_freenameindex NEW: IPv6
4785 if_indextoname NEW: IPv6
4786 if_nameindex NEW: IPv6
4787 if_nametoindex NEW: IPv6
4788 in6addr_any NEW: IPv6
4789 in6addr_loopback NEW: IPv6
4790 inet6_isipv4mapped NEW: IPv6
4791 isastream NEW: STREAMS
4792 iswblank NEW: GNU ext.
4793 key_decryptsession NEW: Secure RPC
4794 key_decryptsession_pk NEW: Secure RPC
4795 key_encryptsession NEW: Secure RPC
4796 key_encryptsession_pk NEW: Secure RPC
4797 key_gendes NEW: Secure RPC
4798 key_get_conv NEW: Secure RPC
4799 key_secretkey_is_set NEW: Secure RPC
4800 key_setnet NEW: Secure RPC
4801 key_setsecret NEW: Secure RPC
4802 llrint NEW: ISO C 9x
4803 llrintf NEW: ISO C 9x
4804 llrintl NEW: ISO C 9x
4805 llround NEW: ISO C 9x
4806 llroundf NEW: ISO C 9x
4807 llroundl NEW: ISO C 9x
4812 lrintf NEW: ISO C 9x
4813 lrintl NEW: ISO C 9x
4814 lround NEW: ISO C 9x
4815 lroundf NEW: ISO C 9x
4816 lroundl NEW: ISO C 9x
4818 makecontext NEW: Unix98
4819 mempcpy NEW: GNU ext.
4822 modify_ldt NEW: kernel
4826 nearbyint NEW: ISO C 9x
4827 nearbyintf NEW: ISO C 9x
4828 nearbyintl NEW: ISO C 9x
4829 netname2host NEW: Secure RPC
4830 netname2user NEW: Secure RPC
4831 nexttoward NEW: ISO C 9x
4832 nexttowardf NEW: ISO C 9x
4833 nexttowardl NEW: ISO C 9x
4837 passwd2des NEW: Secure RPC
4839 pow10f NEW: GNU ext.
4840 pow10l NEW: GNU ext.
4843 printf_size NEW: GNU ext.
4844 printf_size_info NEW: GNU ext.
4845 profil_counter REMOVED
4846 pthread_mutexattr_getkind_np REPLACED
4847 pthread_mutexattr_setkind_np REPLACED
4848 ptsname NEW: Unix98 PTY
4849 ptsname_r NEW: Unix98 PTY
4851 putchar_locked REMOVED
4852 putgrent NEW: GNU ext.
4854 putpmsg NEW: STREAMS
4855 pututxline NEW: Unix98
4859 readdir64_r NEW: LFS
4860 remquo NEW: ISO C 9x
4861 remquof NEW: ISO C 9x
4862 remquol NEW: ISO C 9x
4864 roundf NEW: ISO C 9x
4865 roundl NEW: ISO C 9x
4867 scalbln NEW: ISO C 9x
4868 scalblnf NEW: ISO C 9x
4869 scalblnl NEW: ISO C 9x
4871 sendfile NEW: kernel
4872 setcontext NEW: Unix98
4873 setrlimit64 NEW: LFS
4874 setutxent NEW: Unix98
4876 sigignore NEW: Unix98
4877 sigqueue NEW: POSIX.1b
4878 sigrelse NEW: Unix98
4879 sigset NEW: POSIX.1b
4880 sigtimedwait NEW: POSIX.1b
4881 sigwaitinfo NEW: POSIX.1b
4882 sincos NEW: GNU ext.
4883 sincosf NEW: GNU ext.
4884 sincosl NEW: GNU ext.
4888 strcasestr NEW: GNU ext.
4889 strtoimax NEW: ISO C 9x
4890 strtoumax NEW: ISO C 9x
4891 strverscmp NEW: GNU ext.
4892 svcauthdes_stats NEW: Secure RPC
4893 svcunix_create NEW: sunrpc etx
4894 svcunixfd_create NEW: sunrpc ext
4895 swapcontext NEW: Unix98
4896 tcgetsid NEW: Unix98 PTY
4897 tdestroy NEW: GNU ext.
4898 tgamma NEW: ISO C 9x
4899 tgammaf NEW: ISO C 9x
4900 tgammal NEW: ISO C 9x
4904 truncf NEW: ISO C 9x
4905 truncl NEW: ISO C 9x
4907 unlockpt NEW: Unix98 PTY
4908 updwtmpx NEW: Unix98
4909 user2netname NEW: Secure RPC
4910 utmpxname NEW: Unix98
4911 versionsort NEW: GNU ext.
4912 versionsort64 NEW: GNU ext.
4914 wcscasecmp NEW: GNU ext.
4915 wcsncasecmp NEW: GNU ext.
4916 wcsnlen NEW: GNU ext.
4917 wcstoimax NEW: ISO C 9x
4918 wcstoll NEW: ISO C 9x
4919 wcstoull NEW: ISO C 9x
4920 wcstoumax NEW: ISO C 9x
4922 wordexp NEW: POSIX.2
4923 wordfree NEW: POSIX.2
4924 write_profiling REMOVED
4925 xdecrypt NEW: Secure RPC
4926 xdr_authdes_cred NEW: Secure RPC
4927 xdr_authdes_verf NEW: Secure RPC
4928 xdr_cryptkeyarg NEW: Secure RPC
4929 xdr_cryptkeyarg2 NEW: Secure RPC
4930 xdr_cryptkeyres NEW: Secure RPC
4931 xdr_getcredres NEW: Secure RPC
4932 xdr_key_netstarg NEW: Secure RPC
4933 xdr_key_netstres NEW: Secure RPC
4934 xdr_keybuf NEW: Secure RPC
4935 xdr_keystatus NEW: Secure RPC
4936 xdr_netnamestr NEW: Secure RPC
4937 xdr_sizeof NEW: Secure RPC
4938 xdr_unixcred NEW: sunrpc ext
4939 xencrypt NEW: Secure RPC
4940 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4951 * inet_ntoa is thread-safe
4953 * updwtmp is moved from libutil to libc
4955 * rewrite of cbrt function
4957 * update of timezone data
4971 * add atoll function
4973 * fix complex problems in Berkeley DB code
4975 * fix math functions
4979 * fixed lots of header problems (especially Linux/GNU specific)
4981 * dynamic loader preserves all registers
4983 * Roland McGrath provided support for handling of auxiliary objects in
4984 the ELF dynamic loader.
4986 * support for parallel builds is improved
4990 * GNU extensions are no longer declared by default. To enable them you
4991 must define the macro `_GNU_SOURCE' in your program or compile with
4994 * The library has changed from using GNU ld symbol aliases to using weak
4995 symbols where available. The ELF object file format supports weak
4996 symbols; GNU ld also supports weak symbols in the a.out format. (There
4997 is also now support for other GNU ld extensions in ELF. Use the
4998 `--with-elf' option to configure to indicate you have ELF, and
4999 `--with-gnu-ld' if using GNU ld.) This change resulted in the deletion
5000 of many files which contained only symbol aliases, reducing the size of
5001 the source and the compiled library; many other files were renamed to
5002 less cryptic names previously occupied by the symbol alias files.
5003 There is a new header file <elf.h> for programs which operate on
5004 files in the ELF format.
5006 * Converted to Autoconf version 2, so `configure' has more options.
5007 Run `configure --help' to see the details.
5009 * The library can now be configured to build profiling, highly-optimized
5010 (but undebuggable), and/or shared libraries (ELF with GNU ld only). The
5011 `--enable-profile', `--enable-omitfp', and `--enable-shared' options to
5012 `configure' enable building these extra libraries. The shared library is
5013 built by default when using both ELF and GNU ld. When shared libraries
5014 are enabled, the new library `-ldl' is available for arbitrary run-time
5015 loading of shared objects; its interface is defined in <dlfcn.h>. The
5016 new header file <link.h> gives access to the internals of the run-time
5017 dynamic linker, `ld.so'. The shell script `ldd' is similar to the
5018 application of same name on other systems and it provides information
5019 about dynamically linked binaries.
5021 * The C library now provides the run-time support code for profiling
5022 executables compiled with `-pg'. Programs can control the profiling code
5023 through the interface in <sys/gmon.h>. The `gmon.out' files written by
5024 the GNU C library can be read only by GNU `gprof' (from GNU binutils);
5025 the support for this file format was contributed by David Mosberger-Tang.
5027 * The math code has been replaced with a math library based on fdlibm from
5028 Sun, and modified by JT Conklin and Ulrich Drepper with i387 support, by
5029 Ian Taylor with `float' functions and by Ulrich Drepper with `long double'
5030 functions. The math functions now reside in a separate library, so
5031 programs using them will need to use `-lm' their linking commands.
5033 * John C. Bowman contributed optimized ix87 assembler inline functions.
5035 * Ulrich Drepper has contributed support for an `/etc/nsswitch.conf'
5036 mechanism similar to that found in Solaris 2. This is now used for the
5037 group, passwd, hosts, networks, services, protocols, rpc, ethers,
5038 shadow, netgroup, publickey, and alias databases. The `nsswitch.conf'
5039 file controls what services are used for each individual database. This
5040 works by loading shared libraries with names specified in `nsswitch.conf',
5041 so service modules can be changed or added at any time without even
5042 relinking any program. Currently there are the file, db, and NIS based
5043 NSS services available.
5045 * The new functions `strtoq' and `strtouq' parse integer values from
5046 strings, like `strtol' and `strtoul', but they return `long long int' and
5047 `unsigned long long int' values, respectively (64-bit quantities).
5049 * The new functions `strtof' and `strtold' parse floating-point values from
5050 strings, like `strtod', but they return `float' and `long double' values,
5051 respectively (on some machines `double' and `long double' are the same).
5053 * Ulrich Drepper has contributed new implementations of the floating-point
5054 printing and reading code used in the `printf' family of functions and
5055 `strtod', `strtof', and `strtold'. These new functions are perfectly
5056 accurate, and much faster than the old ones.
5058 * The implementation of the POSIX locale model was completely rewritten by
5059 Ulrich Drepper. This includes the new programs `localedef' and `locale'
5060 to compile the POSIX locale definition.
5062 * The former dummy implementations of the strcoll and strxfrm function are
5063 now replaced by fully functional code contributed by Ulrich Drepper. The
5064 collation information comes from the POSIX locale definitions.
5066 * The new header <langinfo.h> defines an interface for accessing
5067 various locale-dependent data (using the locale chosen with `setlocale').
5069 * Ulrich Drepper has contributed a new suite of functions for operation on
5070 wide-character and multibyte-character strings, in <wchar.h>;
5071 and classification and case conversion of wide characters, in <wctype.h>.
5072 These new functions are conforming to the ISO C, Amendement 1 specification.
5074 * There is now a second implementation of the standard I/O library available.
5075 It comes from GNU libg++ as was written by Per Bothner, heavily modified
5076 by Hongjiu Lu and made thread safe by Ulrich Drepper.
5078 * You can now use positional parameter specifications in format strings
5079 for the `printf' and `scanf' families of functions. For example,
5080 `printf ("Number %2$d, Mr %1$s\n", "Jones", 6);'' prints
5081 ``Number 6, Mr Jones''. This is mainly useful when providing different
5082 format strings for different languages, whose grammars may dictate
5083 different orderings of the values being printed. To support this
5084 feature, the interface for `register_printf_handler' has changed; see
5085 the header file <printf.h> for details.
5087 * The `printf' and `scanf' families of functions now understand a new
5088 formatting flag for numeric conversions: the ' flag (e.g. %'d or %'f) says
5089 to group numbers as indicated by the locale; for `scanf' and friends, this
5090 says to accept as valid only a number with all the proper grouping
5091 separators in the right places. In the default "C" locale, numbers are
5092 not grouped; but locales for specific countries will define the usual
5093 conventions (i.e. separate thousands with `,' in the US locale).
5095 * The pgrp functions have been regularized, slightly incompatibly but much
5096 less confusingly. The core functions are now `getpgid' and `setpgid',
5097 which take arguments for the PID to operate on; the POSIX.1 `getpgrp' (no
5098 argument) and BSD `setpgrp' (identical to `setpgid') functions are
5099 provided for compatibility. There is no longer an incompatible `getpgrp'
5100 with an argument declared under _BSD_SOURCE; no BSD code uses it.
5102 * The new header file <fts.h> and suite of functions simplify programs that
5103 operate on directory trees. This code comes from 4.4 BSD.
5105 * The resolver code has been updated from the BIND 4.9.5-P1 release.
5106 Parts of the code were heavily modified by Ulrich Drepper to fit in the
5107 NSS scheme used in glibc.
5109 * There is a new malloc debugging hook `__memalign_hook'.
5111 * There are new typedefs `ushort' for `unsigned short int' and `uint' for
5112 `unsigned int' in <sys/types.h>. These are for compatibility only and
5113 their use is discouraged.
5115 * The `-lmcheck' library to enable standard malloc debugging hooks is now
5116 done differently, so that it works even without GNU ld.
5118 * New function `euidaccess' checks allowed access to a file like `access',
5119 but using the effective IDs instead of the real IDs.
5121 * The time zone data files have been updated for the latest and greatest
5122 local time conventions of the countries of the world.
5124 * The new function `dirfd' extracts the file descriptor used by a DIR stream;
5127 * The new functions `ecvt', `fcvt', and `gcvt' provide an obsolete interface
5128 for formatting floating-point numbers. They are provided only for
5129 compatibility; new programs should use `sprintf' instead. There are
5130 also equivalent function for the `long double' floating-point type and
5131 all functions also exist in a reentrant form.
5133 * The new auxiliary library `-lutil' from 4.4 BSD contains various
5134 functions for maintaining the login-record files (primarily of use to
5135 system programs such as `login'), and convenient functions for
5136 allocating and initializing a pseudo-terminal (pty) device.
5138 * Ulrich Drepper has contributed new support for System V style
5139 shared memory and IPC on systems that support it.
5141 * Ulrich Drepper has contributed several miscellaneous new functions found
5142 in System V: The `hsearch' family of functions provide an effective
5143 implementation of hash tables; `a64l' and `l64a' provide a very simple
5144 binary to ASCII mapping; `drand48' and friends provide a 48-bit random
5147 * Ulrich Drepper has contributed new reentrant counterparts for the
5148 `random' and `hsearch' families of functions; `random_r', `hsearch_r', etc.
5150 * Ulrich Drepper has contributed new, highly-optimized versions of several
5151 string functions for the i486/Pentium family of processors.
5153 * Ulrich Drepper has updated the Linux-specific code, based largely
5154 on work done in Hongjiu Lu's version of GNU libc for Linux.
5155 The GNU library now supports Linux versions 2.0.10 and later,
5156 using the ELF object file format (i[3456]86-*-linux).
5158 * Andreas Schwab has ported the C library to Linux/m68k (m68k-*-linux).
5160 * David Mosberger-Tang and Richard Henderson have ported the C library
5161 to Linux/Alpha (alpha-*-linux). Richard Henderson contributed the
5162 dynamic linking support for ELF/Alpha.
5164 * Richard Henderson contributed several Alpha optimized assembler function
5165 for arithmetic and string handling.
5167 * Ulrich Drepper has contributed a new set of message catalog functions to
5168 support multiple languages using the <libintl.h> interface, for use with
5169 his new package GNU gettext. Translation volunteers have contributed
5170 catalogs of the library's messages in Spanish, German, and Korean.
5172 * For compatibility with XPG4, Ulrich Drepper has contributed the `gencat'
5173 program and the `catgets' function for reading the catalog files it
5174 creates. (The <libintl.h> interface is preferred; we include the
5175 <nl_types.h> interface using `catgets' only for source compatibility with
5176 programs already written to use it.)
5178 * New header file <values.h> gives SVID-compatible names for <limits.h>
5181 * Various new macros, declarations, and small header files for compatibility
5184 * New function `group_member' is a convenient way to check if a process has
5185 a given effective group ID.
5187 * When using GCC 2.7 and later, the socket functions are now declared in a
5188 special way so that passing an argument of type `struct sockaddr_in *',
5189 `struct sockaddr_ns *', or `struct sockaddr_un *' instead of the generic
5190 `struct sockaddr *' type, does not generate a type-clash warning.
5192 * New function `error' declared in header file <error.h> is a convenient
5193 function for printing error messages and optionally exiting; this is the
5194 canonical function used in GNU programs. The new functions `err', `warn',
5195 and friends in header file <err.h> are the canonical 4.4 BSD interface for
5196 doing the same thing.
5198 * The <glob.h> interface has several new flags from 4.4 BSD that extend the
5199 POSIX.2 `glob' function to do ~ and {...} expansion.
5201 * New function `unsetenv' complements `setenv' for compatibility with 4.4 BSD.
5202 `clearenv' which is used in POSIX.9 is also available.
5204 * New function `getsid' returns session ID number on systems that support it.
5206 * We have incorporated the 4.4 BSD `db' library (version 1.85). New header
5207 files <db.h> and <mpool.h> provide a rich set of functions for several
5208 types of simple databases stored in memory and in files, and <ndbm.h> is
5209 an old `ndbm'-compatible interface using the `db' functions. Link with
5210 `-ldb' to get these functions.
5212 * New macro `strdupa' copies a string like `strdup', but uses local stack
5213 space from `alloca' instead of dynamic heap space from `malloc'.
5215 * New function `strnlen' is like `strlen' but searches only a given maximum
5216 number of characters for the null terminator. `stpncpy', `strndup' and
5217 `strndupa' are similar variants for the `stpcpy', `strdup' and `strdupa'
5220 * New function `statfs' in header <sys/statfs.h>.
5222 * The new <argz.h> and <envz.h> interfaces contributed by Miles Bader
5223 provide convenient functions for operating on blocks of null-terminated
5226 * A new suite of functions in <utmp.h> handle all the details of reading
5227 and writing the utmp file.
5229 * An implementation of the NIS/YP(tm) based NSS service was contributed by
5232 * Paul Eggert and Ulrich Drepper modified the `strftime' function to be
5233 completely POSIX compliant and also implemented the extended functionality
5234 to handle alternate digit representation and alternate era date formats.
5236 * Ulrich Drepper provided an implementation of the `strptime' function
5237 defined in XPG4.2 which transforms a string into a `struct tm' value.
5239 * Paul Eggert provided the tzselect shell script as part of the timezone
5240 code. The shell script makes it easy to select the correct timezone
5243 * The implementation of the malloc family of functions is completely replaced
5244 by a new implementation by Doug Lea with many improvements by Wolfram Gloger.
5245 The implementation uses the mmap function (if available) and it is
5246 optimized for the use in multi threaded programs.
5248 * Ulrich Drepper contributed a MD5 "encryption" for the crypt family of
5249 functions. This new functionality is usable by specifying a special
5250 salt string and it is compatible with implementation on *BSD systems.
5252 * Lots of functions from the XPG4.2 standard were added by Ulrich Drepper:
5253 `getsubopt' to handle second level command line options, `bsd_signal'
5254 to access BSD style `signal' functionality, the obsolete `regexp' style
5257 * the `lchown' function is available on system which support this
5260 * The implementation of the shadow password handling function was contributed
5263 * David Mosberger-Tang changed the SunRPC implementation to be 64bit safe.
5265 * POSIX.1g support was added. The <sys/select.h> header is available,
5266 `isfdtype' and `pselect' are implemented. Craig Metz contributed an
5267 implementation of `getaddrinfo'.
5271 * For cross-compilation you should now set `BUILD_CC' instead of `HOST_CC'.
5273 * New header file <fstab.h> and new functions `getfsspec', `getfsent' and
5274 friends, for parsing /etc/fstab. This code comes from 4.4 BSD.
5276 * The new function `daemon' from 4.4 BSD is useful for server programs that
5277 want to put themselves in the background.
5279 * Joel Sherrill has contributed support for several standalone boards that
5280 run without an operating system.
5282 * `printf', `scanf' and friends now accept a `q' type modifier for long
5283 long int as well as `ll'. Formats using these might be `%qu' or `%lld'.
5285 * All of the code taken from BSD (notably most of the math and networking
5286 routines) has been updated from the BSD 4.4-Lite release.
5288 * The resolver code has been updated from the BIND-4.9.3-BETA9 release.
5290 * The new functions `getdomainname' and `setdomainname' fetch or change the
5291 YP/NIS domain name. These are system calls which exist on systems which
5294 * The time zone data files have been updated for the latest international
5297 * The SunRPC programs `portmap' and `rpcinfo' are now installed in
5298 $(sbindir) (usually /usr/local/sbin) instead of $(bindir).
5302 * The C library now includes support for Sun RPC, from Sun's free
5303 RPCSRC-4.0 distribution. The `portmap', `rpcinfo', and `rpcgen' programs
5304 are included. (There is still no support for YP.)
5306 * Tom Quinn has contributed a port of the C library to SGI machines running
5307 Irix 4 (mips-sgi-irix4).
5309 * The new `lockf' function is a simplified interface to the locking
5310 facilities of `fcntl', included for compatibility.
5312 * New time functions `timegm', `timelocal', and `dysize' for compatibility.
5314 * New header file <sys/timeb.h> and new function `ftime' for compatibility.
5316 * New header files <poll.h> and <sys/poll.h> and new function `poll' for
5319 * The error message printed by `assert' for a failed assertion now includes
5320 the name of the program (if using GNU ld) and the name of the calling
5321 function (with versions of GCC that support this).
5323 * The `psignal' function is now declared in <signal.h>, not <stdio.h>.
5325 * The library now includes the <sys/mman.h> header file and memory
5326 management functions `mmap', `munmap', `mprotect', `msync', and
5327 `madvise', on systems that support those facilities.
5329 * The interface for `mcheck' has changed slightly: the function called to
5330 abort the program when an allocation inconsistency is detected now takes
5331 an argument that indicates the type of failure. The new function
5332 `mprobe' lets you request a consistency check for a particular block at
5333 any time (checks are normally done only when you call `free' or `realloc'
5336 * It is now possible to easily cross-compile the C library, building on one
5337 system a library to run on another machine and/or operating system. All
5338 you need to do is set the variable `HOST_CC' in `configparms' to the
5339 native compiler for programs to run on the machine you are building on (a
5340 few generator programs are used on Unix systems); set `CC' to the
5343 * The new function `fexecve' (only implemented on the GNU system) executes
5344 a program file given a file descriptor already open on the file.
5348 * Brendan Kehoe has contributed most of a port to the DEC Alpha
5349 running OSF/1 (alpha-dec-osf1). He says it is 75% complete.
5351 * You can set the variable `libprefix' in `configparms' to specify a prefix
5352 to be prepended to installed library files; this makes it easy to install
5353 the GNU C library to be linked as `-lgnuc' or whatever.
5355 * The new `stpncpy' is a cross between `stpcpy' and `strncpy': It
5356 copies a limited number of characters from a string, and returns the
5357 address of the last character written.
5359 * You no longer need to check for whether the installed `stddef.h' is
5360 compatible with the GNU C library. configure now checks for you.
5362 * You can now define a per-stream `fileno' function to convert the
5363 stream's cookie into an integral file descriptor.
5365 * ``malloc (0)'' no longer returns a null pointer. Instead, it
5366 allocates zero bytes of storage, and returns a unique pointer which
5367 you can pass to `realloc' or `free'. The behavior is undefined if
5368 you dereference this pointer.
5370 * The C library now runs on Sony NEWS m68k machines running either
5371 NewsOS 3 or NewsOS 4.
5373 * The new `syscall' function is a system-dependent primitive function
5374 for invoking system calls. It has the canonical behavior on Unix
5375 systems, including unreliable return values for some calls (such as
5376 `pipe', `fork' and `getppid').
5378 * The error code `EWOULDBLOCK' is now obsolete; it is always defined
5379 to `EAGAIN', which is the preferred name. On systems whose kernels
5380 use two distinct codes, the C library now translates EWOULDBLOCK to
5381 EAGAIN in every system call function.
5385 * The GNU C Library Reference Manual is now distributed with the library.
5386 `make dvi' will produce a DVI file of the printed manual.
5387 `make info' will produce Info files that you can read on line using C-h i
5388 in Emacs or the `info' program.
5389 Please send comments on the manual to bug-glibc-manual@gnu.org.
5391 * The library now supports SVR4 on i386s (i386-unknown-sysv4).
5393 * Brendan Kehoe has contributed a port to Sun SPARCs running Solaris 2.
5395 * Jason Merrill has contributed a port to the Sequent Symmetry running
5396 Dynix version 3 (i386-sequent-dynix).
5398 * The library has been ported to i386s running SCO 3.2.4 (also known as SCO
5399 ODT 2.0; i386-unknown-sco3.2.4) or SCO 3.2 (i386-unknown-sco3.2).
5401 * New function `memory_warnings' lets you arrange to get warnings when
5402 malloc is running out of memory to allocate, like Emacs gives you.
5404 * The C library now contains the relocating allocator used in Emacs 19 for
5405 its editing buffers. This allocator (ralloc) minimizes allocation
5406 overhead and fragmentation by moving allocated regions around whenever it
5407 needs to. You always refer to a ralloc'd region with a "handle" (a
5408 pointer to a pointer--an object of type `void **').
5410 * There is a new `printf' format: `%m' gives you the string corresponding
5411 to the error code in `errno'.
5413 * In `scanf' formats, you can now use `%as' or `%a[' to do the normal `%s'
5414 or `%[' conversion, but instead of filling in a fixed-sized buffer you
5415 pass, the `a' modifier says to fill in a `char **' you pass with a
5418 * The `fnmatch' function supports the new flag bits `FNM_LEADING_DIR' and
5419 `FNM_CASEFOLD'. `FNM_LEADING_DIR' lets a pattern like `foo*' match a
5420 name like `foo/bar'. `FNM_CASEFOLD' says to ignore case in matching.
5422 * `mkstemp' is a traditional Unix function to atomically create and open a
5423 uniquely-named temporary file.
5427 * The standard location for the file that says what the local timezone is
5428 has changed again. It is now `/usr/local/etc/localtime' (or more
5429 precisely, `${prefix}/etc/localtime') rather than `/etc/localtime'.
5431 * The distribution no longer contains any files with names longer than 14
5434 * `struct ttyent' has two new flag bits: TTY_TRUSTED and TTY_CONSOLE.
5435 These are set by the new `trusted' and `console' keywords in `/etc/ttys'.
5437 * New functions `ttyslot' and `syslog' from 4.4 BSD.
5441 * The configuration process has changed quite a bit. The `configure'
5442 script is now used just like the configuration scripts for other GNU
5443 packages. The `sysdeps' directory hierarchy is much rearranged.
5444 The file `INSTALL' explains the new scheme in detail.
5446 * The header files no longer need to be processed into ANSI C and
5447 traditional C versions. There is just one set of files to install, and
5448 it will work with ANSI or old C compilers (including `gcc -traditional').
5450 * Brendan Kehoe and Ian Lance Taylor have ported the library to the
5451 MIPS DECStation running Ultrix 4.
5453 * The Sun 4 startup code (crt0) can now properly load SunOS 4 shared libraries.
5454 Tom Quinn contributed the initial code. The GNU C library can NOT yet be
5455 made itself into a shared library.
5457 * Yet further improved support for the i386, running 4.3 BSD-like systems
5458 (such as Mach 3 with the Unix single-server), or System V.
5460 * New function `strncasecmp' to do case-insensitive string comparison
5461 with limited length.
5463 * New function `strsep' is a reentrant alternative to `strtok'.
5465 * New functions `scandir' and `alphasort' for searching directories.
5467 * New function `setenv' is a better interface to `putenv'.
5469 * Ian Lance Taylor has contributed an implementation of the SVID `ftw'
5470 function for traversing a directory tree.
5472 * The GNU obstack package is now also part of the C library.
5473 The new function `open_obstack_stream' creates a stdio stream that
5474 writes onto an obstack; `obstack_printf' and `obstack_vprintf' do
5475 formatted output directly to an obstack.
5477 * Miscellaneous new functions: reboot, nice, sigaltstack (4.4 BSD only),
5478 cfmakeraw, getusershell, getpass, swab, getttyent, seteuid, setegid.
5480 * `FNM_FILE_NAME' is another name for `FNM_PATHNAME', used with `fnmatch'.
5482 * The new functions `strfry' and `memfrob' do mysterious and wonderful
5483 things to your strings.
5485 * There are some new test programs: test-fseek, testmb, and testrand.
5487 * Some work has been done to begin porting the library to 4.4 BSD and Linux.
5488 These ports are not finished, but are a good starting place for really
5489 supporting those systems.
5491 * `/etc/localtime' is now the standard location for the file that says what
5492 the local timezone is, rather than `/usr/local/lib/zoneinfo/localtime'.
5493 This follows the general principle that `/etc' is the place for all local
5494 configuration files.
5496 * The C library header files now use `extern "C"' when used by the C++
5497 compiler, so the C library should now work with C++ code.
5499 * The header file <bstring.h> is gone. <string.h> now declares bcopy,
5500 bcmp, bzero, and ffs. (Update: nowadays these functions are declared
5503 * Mike Haertel (of GNU e?grep and malloc fame) has written a new sorting
5504 function which uses the `merge sort' algorithm, and is said to be
5505 significantly faster than the old GNU `qsort' function. Merge sort is
5506 now the standard `qsort' function. The new algorithm can require a lot
5507 of temporary storage; so, the old sorting function is called when the
5508 required storage is not available.
5510 * The C library now includes Michael Glad's Ultra Fast Crypt, which
5511 provides the Unix `crypt' function, plus some other entry points.
5513 * The code and header files taken from 4.4 BSD have been updated with the
5514 latest files released from Berkeley.
5516 ----------------------------------------------------------------------
5517 Copyright information:
5519 Copyright (C) 1992-2019 Free Software Foundation, Inc.
5521 Permission is granted to anyone to make or distribute verbatim copies
5522 of this document as received, in any medium, provided that the
5523 copyright notice and this permission notice are preserved,
5524 thus giving the recipient permission to redistribute in turn.
5526 Permission is granted to distribute modified versions
5527 of this document, or of portions of it,
5528 under the above conditions, provided also that they
5529 carry prominent notices stating who last changed them.
5532 version-control: never