1 Frequently Asked Questions about the GNU C Library
3 This document tries to answer questions a user might have when installing
4 and using glibc. Please make sure you read this before sending questions or
5 bug reports to the maintainers.
7 The GNU C library is very complex. The installation process has not been
8 completely automated; there are too many variables. You can do substantial
9 damage to your system by installing the library incorrectly. Make sure you
10 understand what you are undertaking before you begin.
12 If you have any questions you think should be answered in this document,
19 ?? What systems does the GNU C Library run on?
21 {UD} This is difficult to answer. The file `README' lists the architectures
22 GNU libc was known to run on *at some time*. This does not mean that it
23 still can be compiled and run on them now.
25 The systems glibc is known to work on as of this release, and most probably
29 i[3456]86-*-linux-gnu Linux-2.x on Intel
30 m68k-*-linux-gnu Linux-2.x on Motorola 680x0
31 alpha*-*-linux-gnu Linux-2.x on DEC Alpha
32 powerpc-*-linux-gnu Linux and MkLinux on PowerPC systems
33 sparc-*-linux-gnu Linux-2.x on SPARC
34 sparc64-*-linux-gnu Linux-2.x on UltraSPARC
35 arm-*-none ARM standalone systems
36 arm-*-linux Linux-2.x on ARM
37 arm-*-linuxaout Linux-2.x on ARM using a.out binaries
38 mips*-*-linux-gnu Linux-2.x on MIPS
39 ia64-*-linux-gnu Linux-2.x on ia64
41 Ports to other Linux platforms are in development, and may in fact work
42 already, but no one has sent us success reports for them. Currently no
43 ports to other operating systems are underway, although a few people have
46 If you have a system not listed above (or in the `README' file) and you are
47 really interested in porting it, contact
51 ??binsize What compiler do I need to build GNU libc?
53 {UD} You must use GNU CC to compile GNU libc. A lot of extensions of GNU CC
54 are used to increase portability and speed.
56 GNU CC is found, like all other GNU packages, on
58 ftp://ftp.gnu.org/pub/gnu
60 and the many mirror sites. ftp.gnu.org is always overloaded, so try to find
63 You should always try to use the latest official release. Older versions
64 may not have all the features GNU libc requires. The current releases of
65 egcs (1.0.3 and 1.1.1) should work with the GNU C library (for powerpc see
66 ?powerpc; for ARM see ?arm).
68 While the GNU CC should be able to compile glibc it is nevertheless adviced
69 to use EGCS. Comparing the sizes of glibc on Intel compiled with a recent
70 EGCS and gcc 2.8.1 shows this:
72 text data bss dec hex filename
73 egcs-2.93.10 862897 15944 12824 891665 d9b11 libc.so
74 gcc-2.8.1 959965 16468 12152 988585 f15a9 libc.so
76 Make up your own decision.
78 GNU CC versions 2.95 and above are derived from egcs, and they may do even
81 Please note that gcc 2.95 and 2.95.x cannot compile glibc on Alpha due to
82 problems in the complex float support.
84 ?? When I try to compile glibc I get only error messages.
87 {UD} You definitely need GNU make to build GNU libc. No other make
88 program has the needed functionality.
90 We recommend version GNU make version 3.79 or newer. Older versions have
91 bugs and/or are missing features.
93 ?? Do I need a special linker or assembler?
95 {ZW} If you want a shared library, you need a linker and assembler that
96 understand all the features of ELF, including weak and versioned symbols.
97 The static library can be compiled with less featureful tools, but lacks key
100 For Linux or Hurd, you want binutils 2.8.1.0.23, 2.9.1, or 2.9.1.0.15 or
101 higher. These are the only versions we've tested and found reliable. Other
102 versions after 2.8.1.0.23 may work but we don't recommend them, especially
103 not when C++ is involved. Earlier versions do not work at all.
105 Other operating systems may come with system tools that have all the
106 necessary features, but this is moot because glibc hasn't been ported to
109 ??powerpc Which compiler should I use for powerpc?
111 {GK} You want to use at least gcc 2.95 (together with the right versions
112 of all the other tools, of course). See also question ?excpt.
114 ??arm Which tools should I use for ARM?
116 {PB} You should use egcs 1.1 or a later version. For ELF systems some
117 changes are needed to the compiler; a patch against egcs-1.1.x can be found
120 <ftp://ftp.netwinder.org/users/p/philb/egcs-1.1.1pre2-diff-981126>
122 Binutils 2.9.1.0.16 or later is also required.
124 ?? Do I need some more things to compile the GNU C Library?
126 {UD} Yes, there are some more :-).
128 * GNU gettext. This package contains the tools needed to construct
129 `message catalog' files containing translated versions of system
130 messages. See ftp://ftp.gnu.org/pub/gnu or better any mirror
131 site. (We distribute compiled message catalogs, but they may not be
132 updated in patches.) Please note that the required minimal version
133 (0.10.35) of gettext is alpha software and available from
134 ftp://alpha.gnu.org/gnu .
136 * Some files are built with special tools. E.g., files ending in .gperf
137 need a `gperf' program. The GNU version (now available in a separate
138 package, formerly only as part of libg++) is known to work while some
139 vendor versions do not.
141 You should not need these tools unless you change the source files.
143 * Perl 5 is needed if you wish to test an installation of GNU libc
144 as the primary C library.
146 * When compiling for Linux, the header files of the Linux kernel must
147 be available to the compiler as <linux/*.h> and <asm/*.h>.
149 * lots of disk space (~400MB for i?86-linux; more for RISC platforms).
151 * plenty of time. Compiling just the shared and static libraries for
152 i?86-linux takes approximately 1h on an AMD-K6@225MHz w/ 96MB of RAM,
153 45mins on a Celeron@400MHz w/ 128MB, and 55mins on a Alpha@533MHz w/ 256MB.
154 Multiply this by 1.5 or 2.0 if you build profiling and/or the highly
155 optimized version as well. For Hurd systems times are much higher.
157 You should avoid compiling in a NFS mounted filesystem. This is
160 James Troup <J.J.Troup@comp.brad.ac.uk> reports a compile time of
161 45h34m for a full build (shared, static, and profiled) on Atari
162 Falcon (Motorola 68030 @ 16 Mhz, 14 Mb memory) and Jan Barte
163 <yann@plato.uni-paderborn.de> reports 22h48m on Atari TT030
164 (Motorola 68030 @ 32 Mhz, 34 Mb memory)
166 A full build of the PowerPC library took 1h on a PowerPC 750@400Mhz w/
167 64MB of RAM, and about 9h on a 601@60Mhz w/ 72Mb.
169 If you have some more measurements let me know.
171 ?? What version of the Linux kernel headers should be used?
173 {AJ,UD} The headers from the most recent Linux kernel should be used. The
174 headers used while compiling the GNU C library and the kernel binary used
175 when using the library do not need to match. The GNU C library runs without
176 problems on kernels that are older than the kernel headers used. The other
177 way round (compiling the GNU C library with old kernel headers and running
178 on a recent kernel) does not necessarily work. For example you can't use
179 new kernel features if you used old kernel headers to compile the GNU C
182 {ZW} Even if you are using a 2.0 kernel on your machine, we recommend you
183 compile GNU libc with 2.2 kernel headers. That way you won't have to
184 recompile libc if you ever upgrade to kernel 2.2. To tell libc which
185 headers to use, give configure the --with-headers switch
186 (e.g. --with-headers=/usr/src/linux-2.2.0/include).
188 Note that you must configure the 2.2 kernel if you do this, otherwise libc
189 will be unable to find <linux/version.h>. Just change the current directory
190 to the root of the 2.2 tree and do `make include/linux/version.h'.
192 ?? The compiler hangs while building iconvdata modules. What's
195 {ZW} This is a problem with old versions of GCC. Initialization of large
196 static arrays is very slow. The compiler will eventually finish; give it
199 The problem is fixed in egcs 1.1.
201 ?? When I run `nm -u libc.so' on the produced library I still
202 find unresolved symbols. Can this be ok?
204 {UD} Yes, this is ok. There can be several kinds of unresolved symbols:
206 * magic symbols automatically generated by the linker. These have names
207 like __start_* and __stop_*
209 * symbols starting with _dl_* come from the dynamic linker
211 * weak symbols, which need not be resolved at all (fabs for example)
213 Generally, you should make sure you find a real program which produces
214 errors while linking before deciding there is a problem.
216 ??addon What are these `add-ons'?
218 {UD} To avoid complications with export rules or external source code some
219 optional parts of the libc are distributed as separate packages, e.g., the
220 linuxthreads package.
222 To use these packages as part of GNU libc, just unpack the tarfiles in the
223 libc source directory and tell the configuration script about them using the
224 --enable-add-ons option. If you give just --enable-add-ons configure tries
225 to find all the add-on packages in your source tree. This may not work. If
226 it doesn't, or if you want to select only a subset of the add-ons, give a
227 comma-separated list of the add-ons to enable:
229 configure --enable-add-ons=linuxthreads
233 Add-ons can add features (including entirely new shared libraries), override
234 files, provide support for additional architectures, and just about anything
235 else. The existing makefiles do most of the work; only some few stub rules
236 must be written to get everything running.
238 Most add-ons are tightly coupled to a specific GNU libc version. Please
239 check that the add-ons work with the GNU libc. For example the linuxthreads
240 add-on has the same numbering scheme as the libc and will in general only
241 work with the corresponding libc.
243 {AJ} With glibc 2.2 the crypt add-on and with glibc 2.1 the localedata
244 add-on have been integrated into the normal glibc distribution, crypt and
245 localedata are therefore not anymore add-ons.
247 ?? My XXX kernel emulates a floating-point coprocessor for me.
248 Should I enable --with-fp?
250 {ZW} An emulated FPU is just as good as a real one, as far as the C library
251 is concerned. You only need to say --without-fp if your machine has no way
252 to execute floating-point instructions.
254 People who are interested in squeezing the last drop of performance
255 out of their machine may wish to avoid the trap overhead, but this is
256 far more trouble than it's worth: you then have to compile
257 *everything* this way, including the compiler's internal libraries
258 (libgcc.a for GNU C), because the calling conventions change.
260 ?? When compiling GNU libc I get lots of errors saying functions
261 in glibc are duplicated in libgcc.
263 {EY} This is *exactly* the same problem that I was having. The problem was
264 due to the fact that configure didn't correctly detect that the linker flag
265 --no-whole-archive was supported in my linker. In my case it was because I
266 had run ./configure with bogus CFLAGS, and the test failed.
268 One thing that is particularly annoying about this problem is that once this
269 is misdetected, running configure again won't fix it unless you first delete
272 {UD} Starting with glibc-2.0.3 there should be a better test to avoid some
273 problems of this kind. The setting of CFLAGS is checked at the very
274 beginning and if it is not usable `configure' will bark.
276 ?? Why do I get messages about missing thread functions when I use
277 librt? I don't even use threads.
279 {UD} In this case you probably mixed up your installation. librt uses
280 threads internally and has implicit references to the thread library.
281 Normally these references are satisfied automatically but if the thread
282 library is not in the expected place you must tell the linker where it is.
283 When using GNU ld it works like this:
285 gcc -o foo foo.c -Wl,-rpath-link=/some/other/dir -lrt
287 The `/some/other/dir' should contain the thread library. `ld' will use the
288 given path to find the implicitly referenced library while not disturbing
291 ?? What's the problem with configure --enable-omitfp?
293 {AJ} When --enable-omitfp is set the libraries are built without frame
294 pointers. Some compilers produce buggy code for this model and therefore we
295 don't advise using it at the moment.
297 If you use --enable-omitfp, you're on your own. If you encounter problems
298 with a library that was build this way, we advise you to rebuild the library
299 without --enable-omitfp. If the problem vanishes consider tracking the
300 problem down and report it as compiler failure.
302 Since a library built with --enable-omitfp is undebuggable on most systems,
303 debuggable libraries are also built - you can use them by appending "_g" to
306 The compilation of these extra libraries and the compiler optimizations slow
307 down the build process and need more disk space.
309 ?? I get failures during `make check'. What should I do?
311 {AJ} The testsuite should compile and run cleanly on your system; every
312 failure should be looked into. Depending on the failures, you probably
313 should not install the library at all.
315 You should consider using the `glibcbug' script to report the failure,
316 providing as much detail as possible. If you run a test directly, please
317 remember to set up the environment correctly. You want to test the compiled
318 library - and not your installed one. The best way is to copy the exact
319 command line which failed and run the test from the subdirectory for this
322 There are some failures which are not directly related to the GNU libc:
323 - Some compilers produce buggy code. No compiler gets single precision
324 complex numbers correct on Alpha. Otherwise, the egcs 1.1 release should be
325 ok; gcc 2.8.1 might cause some failures; gcc 2.7.2.x is so buggy that
326 explicit checks have been used so that you can't build with it.
327 - The kernel might have bugs. For example on Linux/Alpha 2.0.34 the
328 floating point handling has quite a number of bugs and therefore most of
329 the test cases in the math subdirectory will fail. Linux 2.2 has
330 fixes for the floating point support on Alpha. The Linux/SPARC kernel has
331 also some bugs in the FPU emulation code (as of Linux 2.2.0).
332 - Other tools might have problems. For example bash 2.03 gives a
333 segmentation fault running the tst-rpmatch.sh test script.
335 ?? What is symbol versioning good for? Do I need it?
337 {AJ} Symbol versioning solves problems that are related to interface
338 changes. One version of an interface might have been introduced in a
339 previous version of the GNU C library but the interface or the semantics of
340 the function has been changed in the meantime. For binary compatibility
341 with the old library, a newer library needs to still have the old interface
342 for old programs. On the other hand, new programs should use the new
343 interface. Symbol versioning is the solution for this problem. The GNU
344 libc version 2.1 uses symbol versioning by default if the installed binutils
347 We don't advise building without symbol versioning, since you lose binary
348 compatibility - forever! The binary compatibility you lose is not only
349 against the previous version of the GNU libc (version 2.0) but also against
352 ?? How can I compile on my fast ix86 machine a working libc for my slow
353 i386? After installing libc, programs abort with "Illegal
356 {AJ} glibc and gcc might generate some instructions on your machine that
357 aren't available on i386. You've got to tell glibc that you're configuring
358 for i386 with adding i386 as your machine, for example:
360 ../configure --prefix=/usr i386-pc-linux-gnu
362 And you need to tell gcc to only generate i386 code, just add `-mcpu=i386'
363 (just -m386 doesn't work) to your CFLAGS.
365 {UD} This applies not only to the i386. Compiling on a i686 for any older
366 model will also fail if the above methods are not used.
368 ?? `make' complains about a missing dlfcn/libdl.so when building
369 malloc/libmemprof.so. How can I fix this?
371 {AJ} Older make version (<= 3.78.90) have a bug which was hidden by a bug in
372 glibc (<= 2.1.2). You need to upgrade make to a newer or fixed version.
374 After upgrading make, you should remove the file sysd-sorted in your build
375 directory. The problem is that the broken make creates a wrong order for
376 one list in that file. The list has to be recreated with the new make -
377 which happens if you remove the file.
379 You might encounter this bug also in other situations where make scans
380 directories. I strongly advise to upgrade your make version to 3.79 or
384 ? Installation and configuration issues
386 ?? Can I replace the libc on my Linux system with GNU libc?
388 {UD} You cannot replace any existing libc for Linux with GNU libc. It is
389 binary incompatible and therefore has a different major version. You can,
390 however, install it alongside your existing libc.
392 For Linux there are three major libc versions:
394 libc-5 original ELF libc
397 You can have any combination of these three installed. For more information
398 consult documentation for shared library handling. The Makefiles of GNU
399 libc will automatically generate the needed symbolic links which the linker
402 ?? How do I configure GNU libc so that the essential libraries
403 like libc.so go into /lib and the other into /usr/lib?
405 {UD,AJ} Like all other GNU packages GNU libc is designed to use a base
406 directory and install all files relative to this. The default is
407 /usr/local, because this is safe (it will not damage the system if installed
408 there). If you wish to install GNU libc as the primary C library on your
409 system, set the base directory to /usr (i.e. run configure --prefix=/usr
410 <other_options>). Note that this can damage your system; see ?safety for
413 Some systems like Linux have a filesystem standard which makes a difference
414 between essential libraries and others. Essential libraries are placed in
415 /lib because this directory is required to be located on the same disk
416 partition as /. The /usr subtree might be found on another
417 partition/disk. If you configure for Linux with --prefix=/usr, then this
418 will be done automatically.
420 To install the essential libraries which come with GNU libc in /lib on
421 systems other than Linux one must explicitly request it. Autoconf has no
422 option for this so you have to use a `configparms' file (see the `INSTALL'
423 file for details). It should contain:
428 The first line specifies the directory for the essential libraries, the
429 second line the directory for system configuration files.
431 ??safety How should I avoid damaging my system when I install GNU libc?
433 {ZW} If you wish to be cautious, do not configure with --prefix=/usr. If
434 you don't specify a prefix, glibc will be installed in /usr/local, where it
435 will probably not break anything. (If you wish to be certain, set the
436 prefix to something like /usr/local/glibc2 which is not used for anything.)
438 The dangers when installing glibc in /usr are twofold:
440 * glibc will overwrite the headers in /usr/include. Other C libraries
441 install a different but overlapping set of headers there, so the effect
442 will probably be that you can't compile anything. You need to rename
443 /usr/include out of the way before running `make install'. (Do not throw
444 it away; you will then lose the ability to compile programs against your
447 * None of your old libraries, static or shared, can be used with a
448 different C library major version. For shared libraries this is not a
449 problem, because the filenames are different and the dynamic linker
450 will enforce the restriction. But static libraries have no version
451 information. You have to evacuate all the static libraries in
452 /usr/lib to a safe location.
454 The situation is rather similar to the move from a.out to ELF which
455 long-time Linux users will remember.
457 ?? Do I need to use GNU CC to compile programs that will use the
460 {ZW} In theory, no; the linker does not care, and the headers are supposed
461 to check for GNU CC before using its extensions to the C language.
463 However, there are currently no ports of glibc to systems where another
464 compiler is the default, so no one has tested the headers extensively
465 against another compiler. You may therefore encounter difficulties. If you
466 do, please report them as bugs.
468 Also, in several places GNU extensions provide large benefits in code
469 quality. For example, the library has hand-optimized, inline assembly
470 versions of some string functions. These can only be used with GCC. See
473 ??crypt When linking with the new libc I get unresolved symbols
474 `crypt' and `setkey'. Why aren't these functions in the
478 Removed. Does not apply anymore.
480 ?? When I use GNU libc on my Linux system by linking against
481 the libc.so which comes with glibc all I get is a core dump.
483 {UD} On Linux, gcc sets the dynamic linker to /lib/ld-linux.so.1 unless the
484 user specifies a --dynamic-linker argument. This is the name of the libc5
485 dynamic linker, which does not work with glibc.
487 For casual use of GNU libc you can just specify to the linker
488 --dynamic-linker=/lib/ld-linux.so.2
490 which is the glibc dynamic linker, on Linux systems. On other systems the
491 name is /lib/ld.so.1. When linking via gcc, you've got to add
492 -Wl,--dynamic-linker=/lib/ld-linux.so.2
494 to the gcc command line.
496 To change your environment to use GNU libc for compiling you need to change
497 the `specs' file of your gcc. This file is normally found at
499 /usr/lib/gcc-lib/<arch>/<version>/specs
501 In this file you have to change a few things:
503 - change `ld-linux.so.1' to `ld-linux.so.2'
505 - remove all expression `%{...:-lgmon}'; there is no libgmon in glibc
507 - fix a minor bug by changing %{pipe:-} to %|
509 Here is what the gcc-2.7.2 specs file should look like when GNU libc is
512 -----------------------------------------------------------------------
514 %{V} %{v:%{!V:-V}} %{Qy:} %{!Qn:-Qy} %{n} %{T} %{Ym,*} %{Yd,*} %{Wa,*:%*}
520 %{fPIC:-D__PIC__ -D__pic__} %{fpic:-D__PIC__ -D__pic__} %{!m386:-D__i486__} %{posix:-D_POSIX_SOURCE} %{pthread:-D_REENTRANT}
529 %{!shared:crtend.o%s} %{shared:crtendS.o%s} crtn.o%s
532 -m elf_i386 %{shared:-shared} %{!shared: %{!ibcs: %{!static: %{rdynamic:-export-dynamic} %{!dynamic-linker:-dynamic-linker /lib/ld-linux.so.2}} %{static:-static}}}
535 %{!shared: %{pthread:-lpthread} %{profile:-lc_p} %{!profile: -lc}}
541 %{!shared: %{pg:gcrt1.o%s} %{!pg:%{p:gcrt1.o%s} %{!p:%{profile:gcrt1.o%s} %{!profile:crt1.o%s}}}} crti.o%s %{!shared:crtbegin.o%s} %{shared:crtbeginS.o%s}
543 *switches_need_spaces:
547 %{funsigned-char:-D__CHAR_UNSIGNED__}
550 -D__ELF__ -Dunix -Di386 -Dlinux -Asystem(unix) -Asystem(posix) -Acpu(i386) -Amachine(i386)
558 -----------------------------------------------------------------------
560 Things get a bit more complicated if you have GNU libc installed in some
561 other place than /usr, i.e., if you do not want to use it instead of the old
562 libc. In this case the needed startup files and libraries are not found in
563 the regular places. So the specs file must tell the compiler and linker
566 Version 2.7.2.3 does and future versions of GCC will automatically
567 provide the correct specs.
569 ?? Looking through the shared libc file I haven't found the
570 functions `stat', `lstat', `fstat', and `mknod' and while
571 linking on my Linux system I get error messages. How is
572 this supposed to work?
574 {RM} Believe it or not, stat and lstat (and fstat, and mknod) are supposed
575 to be undefined references in libc.so.6! Your problem is probably a missing
576 or incorrect /usr/lib/libc.so file; note that this is a small text file now,
577 not a symlink to libc.so.6. It should look something like this:
579 GROUP ( libc.so.6 libc_nonshared.a )
581 ??excpt When I run an executable on one system which I compiled on
582 another, I get dynamic linker errors. Both systems have the same
583 version of glibc installed. What's wrong?
585 {ZW} Glibc on one of these systems was compiled with gcc 2.7 or 2.8, the
586 other with egcs (any version). Egcs has functions in its internal
587 `libgcc.a' to support exception handling with C++. They are linked into
588 any program or dynamic library compiled with egcs, whether it needs them or
589 not. Dynamic libraries then turn around and export those functions again
590 unless special steps are taken to prevent them.
592 When you link your program, it resolves its references to the exception
593 functions to the ones exported accidentally by libc.so. That works fine as
594 long as libc has those functions. On the other system, libc doesn't have
595 those functions because it was compiled by gcc 2.8, and you get undefined
596 symbol errors. The symbols in question are named things like
597 `__register_frame_info'.
599 For glibc 2.0, the workaround is to not compile libc with egcs. We've also
600 incorporated a patch which should prevent the EH functions sneaking into
601 libc. It doesn't matter what compiler you use to compile your program.
603 For glibc 2.1, we've chosen to do it the other way around: libc.so
604 explicitly provides the EH functions. This is to prevent other shared
605 libraries from doing it.
607 {UD} Starting with glibc 2.1.1 you can compile glibc with gcc 2.8.1 or
608 newer since we have explicitly add references to the functions causing the
609 problem. But you nevertheless should use EGCS for other reasons
612 {GK} On some Linux distributions for PowerPC, you can see this when you have
613 built gcc or egcs from the Web sources (gcc versions 2.95 or earlier), then
614 re-built glibc. This happens because in these versions of gcc, exception
615 handling is implemented using an older method; the people making the
616 distributions are a little ahead of their time.
618 A quick solution to this is to find the libgcc.a file that came with the
619 distribution (it would have been installed under /usr/lib/gcc-lib), do
620 `ar x libgcc.a frame.o' to get the frame.o file out, and add a line saying
621 `LDLIBS-c.so += frame.o' to the file `configparms' in the directory you're
622 building in. You can check you've got the right `frame.o' file by running
623 `nm frame.o' and checking that it has the symbols defined that you're
626 This will let you build glibc with the C compiler. The C++ compiler
627 will still be binary incompatible with any C++ shared libraries that
628 you got with your distribution.
630 ?? How can I compile gcc 2.7.2.1 from the gcc source code using
633 {AJ} There's only correct support for glibc 2.0.x in gcc 2.7.2.3 or later.
634 But you should get at least gcc 2.8.1 or egcs 1.1 (or later versions)
637 ?? The `gencat' utility cannot process the catalog sources which
638 were used on my Linux libc5 based system. Why?
640 {UD} The `gencat' utility provided with glibc complies to the XPG standard.
641 The older Linux version did not obey the standard, so they are not
644 To ease the transition from the Linux version some of the non-standard
645 features are also present in the `gencat' program of GNU libc. This mainly
646 includes the use of symbols for the message number and the automatic
647 generation of header files which contain the needed #defines to map the
650 Here is a simple SED script to convert at least some Linux specific catalog
651 files to the XPG4 form:
653 -----------------------------------------------------------------------
654 # Change catalog source in Linux specific format to standard XPG format.
655 # Ulrich Drepper <drepper@cygnus.com>, 1996.
659 s/\$ #\([^ ]*\).*/\1/
661 s/\$ #[^ ]* *\(.*\)/\$ \1/
667 s/\(.*\)\n\(.*\)/\2 \1/
669 -----------------------------------------------------------------------
671 ?? Programs using libc have their messages translated, but other
672 behavior is not localized (e.g. collating order); why?
674 {ZW} Translated messages are automatically installed, but the locale
675 database that controls other behaviors is not. You need to run localedef to
676 install this database, after you have run `make install'. For example, to
677 set up the French Canadian locale, simply issue the command
679 localedef -i fr_CA -f ISO-8859-1 fr_CA
681 Please see localedata/README in the source tree for further details.
683 ?? I have set up /etc/nis.conf, and the Linux libc 5 with NYS
684 works great. But the glibc NIS+ doesn't seem to work.
686 {TK} The glibc NIS+ implementation uses a /var/nis/NIS_COLD_START file for
687 storing information about the NIS+ server and their public keys, because the
688 nis.conf file does not contain all the necessary information. You have to
689 copy a NIS_COLD_START file from a Solaris client (the NIS_COLD_START file is
690 byte order independent) or generate it with nisinit from the nis-tools
691 package; available at
693 http://www.suse.de/~kukuk/linux/nisplus.html
695 ?? I have killed ypbind to stop using NIS, but glibc
698 {TK} For faster NIS lookups, glibc uses the /var/yp/binding/ files from
699 ypbind. ypbind 3.3 and older versions don't always remove these files, so
700 glibc will continue to use them. Other BSD versions seem to work correctly.
701 Until ypbind 3.4 is released, you can find a patch at
703 <ftp://ftp.kernel.org/pub/linux/utils/net/NIS/ypbind-3.3-glibc4.diff.gz>
705 ?? Under Linux/Alpha, I always get "do_ypcall: clnt_call:
706 RPC: Unable to receive; errno = Connection refused" when using NIS.
708 {TK} You need a ypbind version which is 64bit clean. Some versions are not
709 64bit clean. A 64bit clean implementation is ypbind-mt. For ypbind 3.3,
710 you need the patch from ftp.kernel.org (See the previous question). I don't
711 know about other versions.
714 ?? After installing glibc name resolving doesn't work properly.
716 {AJ} You probably should read the manual section describing nsswitch.conf
717 (just type `info libc "NSS Configuration File"'). The NSS configuration
718 file is usually the culprit.
721 ?? How do I create the databases for NSS?
723 {AJ} If you have an entry "db" in /etc/nsswitch.conf you should also create
724 the database files. The glibc sources contain a Makefile which does the
725 necessary conversion and calls to create those files. The file is
726 `db-Makefile' in the subdirectory `nss' and you can call it with `make -f
727 db-Makefile'. Please note that not all services are capable of using a
728 database. Currently passwd, group, ethers, protocol, rpc, services shadow
729 and netgroup are implemented. See also question ?nssdb.
731 ?? I have /usr/include/net and /usr/include/scsi as symlinks
732 into my Linux source tree. Is that wrong?
734 {PB} This was necessary for libc5, but is not correct when using glibc.
735 Including the kernel header files directly in user programs usually does not
736 work (see ?kerhdr). glibc provides its own <net/*> and <scsi/*> header
737 files to replace them, and you may have to remove any symlink that you have
738 in place before you install glibc. However, /usr/include/asm and
739 /usr/include/linux should remain as they were.
741 ?? Programs like `logname', `top', `uptime' `users', `w' and
742 `who', show incorrect information about the (number of)
743 users on my system. Why?
747 ?? After upgrading to glibc 2.1 with symbol versioning I get
748 errors about undefined symbols. What went wrong?
750 {AJ} The problem is caused either by wrong program code or tools. In the
751 versioned libc a lot of symbols are now local that were global symbols in
752 previous versions. It seems that programs linked against older versions
753 often accidentally used libc global variables -- something that should not
756 The only way to fix this is to recompile your program. Sorry, that's the
757 price you might have to pay once for quite a number of advantages with
760 ?? When I start the program XXX after upgrading the library
762 XXX: Symbol `_sys_errlist' has different size in shared
763 object, consider re-linking
764 Why? What should I do?
766 {UD} As the message says, relink the binary. The problem is that a few
767 symbols from the library can change in size and there is no way to avoid
768 this. _sys_errlist is a good example. Occasionally there are new error
769 numbers added to the kernel and this must be reflected at user level,
770 breaking programs that refer to them directly.
772 Such symbols should normally not be used at all. There are mechanisms to
773 avoid using them. In the case of _sys_errlist, there is the strerror()
774 function which should _always_ be used instead. So the correct fix is to
775 rewrite that part of the application.
777 In some situations (especially when testing a new library release) it might
778 be possible that a symbol changed size when that should not have happened.
779 So in case of doubt report such a warning message as a problem.
781 ?? What do I need for C++ development?
783 {HJ,AJ} You need either egcs 1.1 which comes directly with libstdc++ or
784 gcc-2.8.1 together with libstdc++ 2.8.1.1. egcs 1.1 has the better C++
785 support and works directly with glibc 2.1. If you use gcc-2.8.1 with
786 libstdc++ 2.8.1.1, you need to modify libstdc++ a bit. A patch is available
788 <ftp://alpha.gnu.org/gnu/libstdc++-2.8.1.1-glibc2.1-diff.gz>
790 Please note that libg++ 2.7.2 (and the Linux Versions 2.7.2.x) doesn't work
791 very well with the GNU C library due to vtable thunks. If you're upgrading
792 from glibc 2.0.x to 2.1 you have to recompile libstdc++ since the library
793 compiled for 2.0 is not compatible due to the new Large File Support (LFS)
796 {UD} But since in the case of a shared libstdc++ the version numbers should
797 be different existing programs will continue to work.
799 ?? Even statically linked programs need some shared libraries
800 which is not acceptable for me. What can I do?
802 {AJ} NSS (for details just type `info libc "Name Service Switch"') won't
803 work properly without shared libraries. NSS allows using different services
804 (e.g. NIS, files, db, hesiod) by just changing one configuration file
805 (/etc/nsswitch.conf) without relinking any programs. The only disadvantage
806 is that now static libraries need to access shared libraries. This is
807 handled transparently by the GNU C library.
809 A solution is to configure glibc with --enable-static-nss. In this case you
810 can create a static binary that will use only the services dns and files
811 (change /etc/nsswitch.conf for this). You need to link explicitly against
812 all these services. For example:
814 gcc -static test-netdb.c -o test-netdb.c \
815 -lc -lnss_files -lnss_dns -lresolv
817 The problem with this approach is that you've got to link every static
818 program that uses NSS routines with all those libraries.
820 {UD} In fact, one cannot say anymore that a libc compiled with this
821 option is using NSS. There is no switch anymore. Therefore it is
822 *highly* recommended *not* to use --enable-static-nss since this makes
823 the behaviour of the programs on the system inconsistent.
825 ?? I just upgraded my Linux system to glibc and now I get
826 errors whenever I try to link any program.
828 {ZW} This happens when you have installed glibc as the primary C library but
829 have stray symbolic links pointing at your old C library. If the first
830 `libc.so' the linker finds is libc 5, it will use that. Your program
831 expects to be linked with glibc, so the link fails.
833 The most common case is that glibc put its `libc.so' in /usr/lib, but there
834 was a `libc.so' from libc 5 in /lib, which gets searched first. To fix the
835 problem, just delete /lib/libc.so. You may also need to delete other
836 symbolic links in /lib, such as /lib/libm.so if it points to libm.so.5.
838 {AJ} The perl script test-installation.pl which is run as last step during
839 an installation of glibc that is configured with --prefix=/usr should help
840 detect these situations. If the script reports problems, something is
843 ?? When I use nscd the machine freezes.
845 {UD} You cannot use nscd with Linux 2.0.*. There is functionality missing
846 in the kernel and work-arounds are not suitable. Besides, some parts of the
847 kernel are too buggy when it comes to using threads.
849 If you need nscd, you have to use at least a 2.1 kernel.
851 Note that I have at this point no information about any other platform.
853 ?? I need lots of open files. What do I have to do?
855 {AJ} This is at first a kernel issue. The kernel defines limits with
856 OPEN_MAX the number of simultaneous open files and with FD_SETSIZE the
857 number of used file descriptors. You need to change these values in your
858 kernel and recompile the kernel so that the kernel allows more open
859 files. You don't necessarily need to recompile the GNU C library since the
860 only place where OPEN_MAX and FD_SETSIZE is really needed in the library
861 itself is the size of fd_set which is used by select.
863 The GNU C library is now select free. This means it internally has no
864 limits imposed by the `fd_set' type. Instead all places where the
865 functionality is needed the `poll' function is used.
867 If you increase the number of file descriptors in the kernel you don't need
868 to recompile the C library.
870 {UD} You can always get the maximum number of file descriptors a process is
871 allowed to have open at any time using
873 number = sysconf (_SC_OPEN_MAX);
875 This will work even if the kernel limits change.
877 ?? How do I get the same behavior on parsing /etc/passwd and
878 /etc/group as I have with libc5 ?
880 {TK} The name switch setup in /etc/nsswitch.conf selected by most Linux
881 distributions does not support +/- and netgroup entries in the files like
882 /etc/passwd. Though this is the preferred setup some people might have
883 setups coming over from the libc5 days where it was the default to recognize
884 lines like this. To get back to the old behaviour one simply has to change
885 the rules for passwd, group, and shadow in the nsswitch.conf file as
896 ??libs What needs to be recompiled when upgrading from glibc 2.0 to glibc
899 {AJ,CG} If you just upgrade the glibc from 2.0.x (x <= 7) to 2.1, binaries
900 that have been linked against glibc 2.0 will continue to work.
902 If you compile your own binaries against glibc 2.1, you also need to
903 recompile some other libraries. The problem is that libio had to be changed
904 and therefore libraries that are based or depend on the libio of glibc,
905 e.g. ncurses, slang and most C++ libraries, need to be recompiled. If you
906 experience strange segmentation faults in your programs linked against glibc
907 2.1, you might need to recompile your libraries.
909 Another problem is that older binaries that were linked statically against
910 glibc 2.0 will reference the older nss modules (libnss_files.so.1 instead of
911 libnss_files.so.2), so don't remove them. Also, the old glibc-2.0 compiled
912 static libraries (libfoo.a) which happen to depend on the older libio
913 behavior will be broken by the glibc 2.1 upgrade. We plan to produce a
914 compatibility library that people will be able to link in if they want
915 to compile a static library generated against glibc 2.0 into a program
916 on a glibc 2.1 system. You just add -lcompat and you should be fine.
918 The glibc-compat add-on will provide the libcompat.a library, the older
919 nss modules, and a few other files. Together, they should make it
920 possible to do development with old static libraries on a glibc 2.1
921 system. This add-on is still in development. You can get it from
922 <ftp://alpha.gnu.org/gnu/glibc/glibc-compat-2.1.tar.gz>
923 but please keep in mind that it is experimental.
925 ?? Why is extracting files via tar so slow?
927 {AJ} Extracting of tar archives might be quite slow since tar has to look up
928 userid and groupids and doesn't cache negative results. If you have nis or
929 nisplus in your /etc/nsswitch.conf for the passwd and/or group database,
930 each file extractions needs a network connection. There are two possible
933 - do you really need NIS/NIS+ (some Linux distributions add by default
934 nis/nisplus even if it's not needed)? If not, just remove the entries.
936 - if you need NIS/NIS+, use the Name Service Cache Daemon nscd that comes
939 ?? Compiling programs I get parse errors in libio.h (e.g. "parse error
940 before `_IO_seekoff'"). How should I fix this?
942 {AJ} You might get the following errors when upgrading to glibc 2.1:
944 In file included from /usr/include/stdio.h:57,
946 /usr/include/libio.h:335: parse error before `_IO_seekoff'
947 /usr/include/libio.h:335: parse error before `_G_off64_t'
948 /usr/include/libio.h:336: parse error before `_IO_seekpos'
949 /usr/include/libio.h:336: parse error before `_G_fpos64_t'
951 The problem is a wrong _G_config.h file in your include path. The
952 _G_config.h file that comes with glibc 2.1 should be used and not one from
953 libc5 or from a compiler directory. To check which _G_config.h file the
954 compiler uses, compile your program with `gcc -E ...|grep G_config.h' and
955 remove that file. Your compiler should pick up the file that has been
956 installed by glibc 2.1 in your include directory.
958 ?? After upgrading to glibc 2.1, libraries that were compiled against
959 glibc 2.0.x don't work anymore.
963 ??nssdb What happened to the Berkeley DB libraries? Can I still use db
964 in /etc/nsswitch.conf?
966 {AJ} Due to too many incompatible changes in disk layout and API of Berkeley
967 DB and a too tight coupling of libc and libdb, the db library has been
968 removed completely from glibc 2.2. The only place that really used the
969 Berkeley DB was the NSS db module.
971 The NSS db module has been rewritten to support a number of different
972 versions of Berkeley DB for the NSS db module. Currently the releases 2.x
973 and 3.x of Berkeley DB are supported. The older db 1.85 library is not
974 supported. You can use the version from glibc 2.1.x or download a version
975 from Sleepycat Software (http://www.sleepycat.com). The library has to be
976 compiled as shared library and installed in the system lib directory
977 (normally /lib). The library needs to have a special soname to be found by
980 If public structures change in a new Berkeley db release, this needs to be
983 Currently the code searches for libraries with a soname of "libdb.so.3"
984 (that's the name from db 2.4.14 which comes with glibc 2.1.x) and
985 "libdb-3.0.so" (the name used by db 3.0.55 as default).
988 ? Source and binary incompatibilities, and what to do about them
990 ?? I expect GNU libc to be 100% source code compatible with
991 the old Linux based GNU libc. Why isn't it like this?
993 {DMT,UD} Not every extension in Linux libc's history was well thought-out.
994 In fact it had a lot of problems with standards compliance and with
995 cleanliness. With the introduction of a new version number these errors can
996 now be corrected. Here is a list of the known source code
999 * _GNU_SOURCE: glibc does not make the GNU extensions available
1000 automatically. If a program depends on GNU extensions or some
1001 other non-standard functionality, it is necessary to compile it
1002 with the C compiler option -D_GNU_SOURCE, or better, to put
1003 `#define _GNU_SOURCE' at the beginning of your source files, before
1004 any C library header files are included. This difference normally
1005 manifests itself in the form of missing prototypes and/or data type
1006 definitions. Thus, if you get such errors, the first thing you
1007 should do is try defining _GNU_SOURCE and see if that makes the
1010 For more information consult the file `NOTES' in the GNU C library
1013 * reboot(): GNU libc sanitizes the interface of reboot() to be more
1014 compatible with the interface used on other OSes. reboot() as
1015 implemented in glibc takes just one argument. This argument
1016 corresponds to the third argument of the Linux reboot system call.
1017 That is, a call of the form reboot(a, b, c) needs to be changed into
1018 reboot(c). Beside this the header <sys/reboot.h> defines the needed
1019 constants for the argument. These RB_* constants should be used
1020 instead of the cryptic magic numbers.
1022 * swapon(): the interface of this function didn't change, but the
1023 prototype is in a separate header file <sys/swap.h>. This header
1024 file also provides the SWAP_* constants defined by <linux/swap.h>;
1025 you should use them for the second argument to swapon().
1027 * errno: If a program uses the variable "errno", then it _must_
1028 include <errno.h>. The old libc often (erroneously) declared this
1029 variable implicitly as a side-effect of including other libc header
1030 files. glibc is careful to avoid such namespace pollution, which,
1031 in turn, means that you really need to include the header files that
1032 you depend on. This difference normally manifests itself in the
1033 form of the compiler complaining about references to an undeclared
1036 * Linux-specific syscalls: All Linux system calls now have appropriate
1037 library wrappers and corresponding declarations in various header files.
1038 This is because the syscall() macro that was traditionally used to
1039 work around missing syscall wrappers are inherently non-portable and
1040 error-prone. The following table lists all the new syscall stubs,
1041 the header-file declaring their interface and the system call name.
1043 syscall name: wrapper name: declaring header file:
1044 ------------- ------------- ----------------------
1045 bdflush bdflush <sys/kdaemon.h>
1046 syslog ksyslog_ctl <sys/klog.h>
1048 * lpd: Older versions of lpd depend on a routine called _validuser().
1049 The library does not provide this function, but instead provides
1050 __ivaliduser() which has a slightly different interface. Simply
1051 upgrading to a newer lpd should fix this problem (e.g., the 4.4BSD
1052 lpd is known to be working).
1054 * resolver functions/BIND: like on many other systems the functions of
1055 the resolver library are not included in libc itself. There is a
1056 separate library libresolv. If you get undefined symbol errors for
1057 symbols starting with `res_*' simply add -lresolv to your linker
1060 * the `signal' function's behavior corresponds to the BSD semantic and
1061 not the SysV semantic as it was in libc-5. The interface on all GNU
1062 systems shall be the same and BSD is the semantic of choice. To use
1063 the SysV behavior simply use `sysv_signal', or define _XOPEN_SOURCE.
1064 See ?signal for details.
1066 ??getlog Why does getlogin() always return NULL on my Linux box?
1068 {UD} The GNU C library has a format for the UTMP and WTMP file which differs
1069 from what your system currently has. It was extended to fulfill the needs
1070 of the next years when IPv6 is introduced. The record size is different and
1071 some fields have different positions. The files written by functions from
1072 the one library cannot be read by functions from the other library. Sorry,
1073 but this is what a major release is for. It's better to have a cut now than
1074 having no means to support the new techniques later.
1076 ?? Where are the DST_* constants found in <sys/time.h> on many
1079 {UD} These constants come from the old BSD days and are not used anymore
1080 (libc5 does not actually implement the handling although the constants are
1083 Instead GNU libc contains zone database support and compatibility code for
1084 POSIX TZ environment variable handling. For former is very much preferred
1087 ?? The prototypes for `connect', `accept', `getsockopt',
1088 `setsockopt', `getsockname', `getpeername', `send',
1089 `sendto', and `recvfrom' are different in GNU libc from
1090 any other system I saw. This is a bug, isn't it?
1092 {UD} No, this is no bug. This version of GNU libc already follows the new
1093 Single Unix specifications (and I think the POSIX.1g draft which adopted the
1094 solution). The type for a parameter describing a size is now `socklen_t', a
1097 ??kerhdr On Linux I've got problems with the declarations in Linux
1100 {UD,AJ} On Linux, the use of kernel headers is reduced to the minimum. This
1101 gives Linus the ability to change the headers more freely. Also, user
1102 programs are now insulated from changes in the size of kernel data
1105 For example, the sigset_t type is 32 or 64 bits wide in the kernel. In
1106 glibc it is 1024 bits wide. This guarantees that when the kernel gets a
1107 bigger sigset_t (for POSIX.1e realtime support, say) user programs will not
1108 have to be recompiled. Consult the header files for more information about
1111 Therefore you shouldn't include Linux kernel header files directly if glibc
1112 has defined a replacement. Otherwise you might get undefined results because
1115 ?? I don't include any kernel headers myself but the compiler
1116 still complains about redeclarations of types in the kernel
1119 {UD} The kernel headers before Linux 2.1.61 and 2.0.32 don't work correctly
1120 with glibc. Compiling C programs is possible in most cases but C++ programs
1121 have (due to the change of the name lookups for `struct's) problems. One
1122 prominent example is `struct fd_set'.
1124 There might be some problems left but 2.1.61/2.0.32 fix most of the known
1125 ones. See the BUGS file for other known problems.
1127 ??signal Why don't signals interrupt system calls anymore?
1129 {ZW} By default GNU libc uses the BSD semantics for signal(), unlike Linux
1130 libc 5 which used System V semantics. This is partially for compatibility
1131 with other systems and partially because the BSD semantics tend to make
1132 programming with signals easier.
1134 There are three differences:
1136 * BSD-style signals that occur in the middle of a system call do not
1137 affect the system call; System V signals cause the system call to
1138 fail and set errno to EINTR.
1140 * BSD signal handlers remain installed once triggered. System V signal
1141 handlers work only once, so one must reinstall them each time.
1143 * A BSD signal is blocked during the execution of its handler. In other
1144 words, a handler for SIGCHLD (for example) does not need to worry about
1145 being interrupted by another SIGCHLD. It may, however, be interrupted
1148 There is general consensus that for `casual' programming with signals, the
1149 BSD semantics are preferable. You don't need to worry about system calls
1150 returning EINTR, and you don't need to worry about the race conditions
1151 associated with one-shot signal handlers.
1153 If you are porting an old program that relies on the old semantics, you can
1154 quickly fix the problem by changing signal() to sysv_signal() throughout.
1155 Alternatively, define _XOPEN_SOURCE before including <signal.h>.
1157 For new programs, the sigaction() function allows you to specify precisely
1158 how you want your signals to behave. All three differences listed above are
1159 individually switchable on a per-signal basis with this function.
1161 If all you want is for one specific signal to cause system calls to fail and
1162 return EINTR (for example, to implement a timeout) you can do this with
1166 ??string I've got errors compiling code that uses certain string
1169 {AJ} glibc 2.1 has special string functions that are faster than the normal
1170 library functions. Some of the functions are additionally implemented as
1171 inline functions and others as macros. This might lead to problems with
1172 existing codes but it is explicitly allowed by ISO C.
1174 The optimized string functions are only used when compiling with
1175 optimizations (-O1 or higher). The behavior can be changed with two feature
1178 * __NO_STRING_INLINES: Don't do any string optimizations.
1179 * __USE_STRING_INLINES: Use assembly language inline functions (might
1180 increase code size dramatically).
1182 Since some of these string functions are now additionally defined as macros,
1183 code like "char *strncpy();" doesn't work anymore (and is unnecessary, since
1184 <string.h> has the necessary declarations). Either change your code or
1185 define __NO_STRING_INLINES.
1187 {UD} Another problem in this area is that gcc still has problems on machines
1188 with very few registers (e.g., ix86). The inline assembler code can require
1189 almost all the registers and the register allocator cannot always handle
1192 One can disable the string optimizations selectively. Instead of writing
1194 cp = strcpy (foo, "lkj");
1198 cp = (strcpy) (foo, "lkj");
1200 This disables the optimization for that specific call.
1202 ?? I get compiler messages "Initializer element not constant" with
1203 stdin/stdout/stderr. Why?
1205 {RM,AJ} Constructs like:
1206 static FILE *InPtr = stdin;
1208 lead to this message. This is correct behaviour with glibc since stdin is
1209 not a constant expression. Please note that a strict reading of ISO C does
1210 not allow above constructs.
1212 One of the advantages of this is that you can assign to stdin, stdout, and
1213 stderr just like any other global variable (e.g. `stdout = my_stream;'),
1214 which can be very useful with custom streams that you can write with libio
1215 (but beware this is not necessarily portable). The reason to implement it
1216 this way were versioning problems with the size of the FILE structure.
1218 To fix those programs you've got to initialize the variable at run time.
1219 This can be done, e.g. in main, like:
1227 or by constructors (beware this is gcc specific):
1230 static void inPtr_construct (void) __attribute__((constructor));
1231 static void inPtr_construct (void) { InPtr = stdin; }
1234 ?? I can't compile with gcc -traditional (or
1235 -traditional-cpp). Why?
1237 {AJ} glibc2 does break -traditional and -traditonal-cpp - and will continue
1238 to do so. For example constructs of the form:
1244 are useful for debugging purposes (you can use foo with your debugger that's
1245 why we need the enum) and for compatibility (other systems use defines and
1248 ?? I get some errors with `gcc -ansi'. Isn't glibc ANSI compatible?
1250 {AJ} The GNU C library is compatible with the ANSI/ISO C standard. If
1251 you're using `gcc -ansi', the glibc includes which are specified in the
1252 standard follow the standard. The ANSI/ISO C standard defines what has to be
1253 in the include files - and also states that nothing else should be in the
1254 include files (btw. you can still enable additional standards with feature
1257 The GNU C library is conforming to ANSI/ISO C - if and only if you're only
1258 using the headers and library functions defined in the standard.
1260 ?? I can't access some functions anymore. nm shows that they do
1261 exist but linking fails nevertheless.
1263 {AJ} With the introduction of versioning in glibc 2.1 it is possible to
1264 export only those identifiers (functions, variables) that are really needed
1265 by application programs and by other parts of glibc. This way a lot of
1266 internal interfaces are now hidden. nm will still show those identifiers
1267 but marking them as internal. ISO C states that identifiers beginning with
1268 an underscore are internal to the libc. An application program normally
1269 shouldn't use those internal interfaces (there are exceptions,
1270 e.g. __ivaliduser). If a program uses these interfaces, it's broken. These
1271 internal interfaces might change between glibc releases or dropped
1274 ?? When using the db-2 library which comes with glibc is used in
1275 the Perl db modules the testsuite is not passed. This did not
1276 happen with db-1, gdbm, or ndbm.
1278 Removed. Does not apply anymore.
1280 ?? The pow() inline function I get when including <math.h> is broken.
1281 I get segmentation faults when I run the program.
1283 {UD} Nope, the implementation is correct. The problem is with egcs version
1284 prior to 1.1. I.e., egcs 1.0 to 1.0.3 are all broken (at least on Intel).
1285 If you have to use this compiler you must define __NO_MATH_INLINES before
1286 including <math.h> to prevent the inline functions from being used. egcs 1.1
1287 fixes the problem. I don't know about gcc 2.8 and 2.8.1.
1289 ?? The sys/sem.h file lacks the definition of `union semun'.
1291 {UD} Nope. This union has to be provided by the user program. Former glibc
1292 versions defined this but it was an error since it does not make much sense
1293 when thinking about it. The standards describing the System V IPC functions
1294 define it this way and therefore programs must be adopted.
1296 ?? Why has <netinet/ip_fw.h> disappeared?
1298 {AJ} The corresponding Linux kernel data structures and constants are
1299 totally different in Linux 2.0 and Linux 2.2. This situation has to be
1300 taken care in user programs using the firewall structures and therefore
1301 those programs (ipfw is AFAIK the only one) should deal with this problem
1304 ?? I get floods of warnings when I use -Wconversion and include
1305 <string.h> or <math.h>.
1307 {ZW} <string.h> and <math.h> intentionally use prototypes to override
1308 argument promotion. -Wconversion warns about all these. You can safely
1309 ignore the warnings.
1311 -Wconversion isn't really intended for production use, only for shakedown
1312 compiles after converting an old program to standard C.
1315 ?? After upgrading to glibc 2.1, I receive errors about
1316 unresolved symbols, like `_dl_initial_searchlist' and can not
1317 execute any binaries. What went wrong?
1319 {AJ} This normally happens if your libc and ld (dynamic linker) are from
1320 different releases of glibc. For example, the dynamic linker
1321 /lib/ld-linux.so.2 comes from glibc 2.0.x, but the version of libc.so.6 is
1324 The path /lib/ld-linux.so.2 is hardcoded in every glibc2 binary but
1325 libc.so.6 is searched via /etc/ld.so.cache and in some special directories
1326 like /lib and /usr/lib. If you run configure with another prefix than /usr
1327 and put this prefix before /lib in /etc/ld.so.conf, your system will break.
1329 So what can you do? Either of the following should work:
1331 * Run `configure' with the same prefix argument you've used for glibc 2.0.x
1332 so that the same paths are used.
1333 * Replace /lib/ld-linux.so.2 with a link to the dynamic linker from glibc
1336 You can even call the dynamic linker by hand if everything fails. You've
1337 got to set LD_LIBRARY_PATH so that the corresponding libc is found and also
1338 need to provide an absolute path to your binary:
1340 LD_LIBRARY_PATH=<path-where-libc.so.6-lives> \
1341 <path-where-corresponding-dynamic-linker-lives>/ld-linux.so.2 \
1342 <path-to-binary>/binary
1344 For example `LD_LIBRARY_PATH=/libold /libold/ld-linux.so.2 /bin/mv ...'
1345 might be useful in fixing a broken system (if /libold contains dynamic
1346 linker and corresponding libc).
1348 With that command line no path is used. To further debug problems with the
1349 dynamic linker, use the LD_DEBUG environment variable, e.g.
1350 `LD_DEBUG=help echo' for the help text.
1352 If you just want to test this release, don't put the lib directory in
1353 /etc/ld.so.conf. You can call programs directly with full paths (as above).
1354 When compiling new programs against glibc 2.1, you've got to specify the
1355 correct paths to the compiler (option -I with gcc) and linker (options
1356 --dynamic-linker, -L and --rpath).
1358 ?? bonnie reports that char i/o with glibc 2 is much slower than with
1359 libc5. What can be done?
1361 {AJ} The GNU C library uses thread safe functions by default and libc5 used
1362 non thread safe versions. The non thread safe functions have in glibc the
1363 suffix `_unlocked', for details check <stdio.h>. Using `putc_unlocked' etc.
1364 instead of `putc' should give nearly the same speed with bonnie (bonnie is a
1365 benchmark program for measuring disk access).
1367 ?? Programs compiled with glibc 2.1 can't read db files made with glibc
1368 2.0. What has changed that programs like rpm break?
1370 Removed. Does not apply anymore.
1372 ?? Autoconf's AC_CHECK_FUNC macro reports that a function exists, but
1373 when I try to use it, it always returns -1 and sets errno to ENOSYS.
1375 {ZW} You are using a 2.0 Linux kernel, and the function you are trying to
1376 use is only implemented in 2.1/2.2. Libc considers this to be a function
1377 which exists, because if you upgrade to a 2.2 kernel, it will work. One
1378 such function is sigaltstack.
1380 Your program should check at runtime whether the function works, and
1381 implement a fallback. Note that Autoconf cannot detect unimplemented
1382 functions in other systems' C libraries, so you need to do this anyway.
1384 ?? My program segfaults when I call fclose() on the FILE* returned
1385 from setmntent(). Is this a glibc bug?
1387 {GK} No. Don't do this. Use endmntent(), that's what it's for.
1389 In general, you should use the correct deallocation routine. For instance,
1390 if you open a file using fopen(), you should deallocate the FILE * using
1391 fclose(), not free(), even though the FILE * is also a pointer.
1393 In the case of setmntent(), it may appear to work in most cases, but it
1394 won't always work. Unfortunately, for compatibility reasons, we can't
1395 change the return type of setmntent() to something other than FILE *.
1400 ?? After I changed configure.in I get `Autoconf version X.Y.
1401 or higher is required for this script'. What can I do?
1403 {UD} You have to get the specified autoconf version (or a later one)
1404 from your favorite mirror of ftp.gnu.org.
1406 ?? When I try to compile code which uses IPv6 headers and
1407 definitions on my Linux 2.x.y system I am in trouble.
1408 Nothing seems to work.
1410 {UD} The problem is that IPv6 development still has not reached a point
1411 where the headers are stable. There are still lots of incompatible changes
1412 made and the libc headers have to follow.
1414 {PB} The 2.1 release of GNU libc aims to comply with the current versions of
1415 all the relevant standards. The IPv6 support libraries for older Linux
1416 systems used a different naming convention and so code written to work with
1417 them may need to be modified. If the standards make incompatible changes in
1418 the future then the libc may need to change again.
1420 IPv6 will not work with a 2.0.x kernel. When kernel 2.2 is released it
1421 should contain all the necessary support; until then you should use the
1422 latest 2.1.x release you can find. As of 98/11/26 the currently recommended
1423 kernel for IPv6 is 2.1.129.
1425 Also, as of the 2.1 release the IPv6 API provided by GNU libc is not
1428 ??tzdb When I set the timezone by setting the TZ environment variable
1429 to EST5EDT things go wrong since glibc computes the wrong time
1430 from this information.
1432 {UD} The problem is that people still use the braindamaged POSIX method to
1433 select the timezone using the TZ environment variable with a format EST5EDT
1434 or whatever. People, if you insist on using TZ instead of the timezone
1435 database (see below), read the POSIX standard, the implemented behaviour is
1436 correct! What you see is in fact the result of the decisions made while
1437 POSIX.1 was created. We've only implemented the handling of TZ this way to
1438 be POSIX compliant. It is not really meant to be used.
1440 The alternative approach to handle timezones which is implemented is the
1441 correct one to use: use the timezone database. This avoids all the problems
1442 the POSIX method has plus it is much easier to use. Simply run the tzselect
1443 shell script, answer the question and use the name printed in the end by
1444 making a symlink /etc/localtime pointing to /usr/share/zoneinfo/NAME (NAME
1445 is the returned value from tzselect). That's all. You never again have to
1448 So, please avoid sending bug reports about time related problems if you use
1449 the POSIX method and you have not verified something is really broken by
1450 reading the POSIX standards.
1452 ?? What other sources of documentation about glibc are available?
1454 {AJ} The FSF has a page about the GNU C library at
1455 <http://www.gnu.org/software/libc/>. The problem data base of open and
1456 solved bugs in GNU libc is available at
1457 <http://www-gnats.gnu.org:8080/cgi-bin/wwwgnats.pl>. Eric Green has written
1458 a HowTo for converting from Linux libc5 to glibc2. The HowTo is accessible
1459 via the FSF page and at <http://www.imaxx.net/~thrytis/glibc>. Frodo
1460 Looijaard describes a different way installing glibc2 as secondary libc at
1461 <http://huizen.dds.nl/~frodol/glibc>.
1463 Please note that this is not a complete list.
1465 ?? The timezone string for Sydney/Australia is wrong since even when
1466 daylight saving time is in effect the timezone string is EST.
1468 {UD} The problem for some timezones is that the local authorities decided
1469 to use the term "summer time" instead of "daylight saving time". In this
1470 case the abbreviation character `S' is the same as the standard one. So,
1473 Eastern Standard Time = EST
1474 Eastern Summer Time = EST
1476 Great! To get this bug fixed convince the authorities to change the laws
1477 and regulations of the country this effects. glibc behaves correctly.
1479 ??make I've build make 3.77 against glibc 2.1 and now make gets
1480 segmentation faults.
1482 Removed. Does not apply anymore, use make 3.79 or newer.
1484 ?? Why do so many programs using math functions fail on my AlphaStation?
1486 {AO} The functions floor() and floorf() use an instruction that is not
1487 implemented in some old PALcodes of AlphaStations. This may cause
1488 `Illegal Instruction' core dumps or endless loops in programs that
1489 catch these signals. Updating the firmware to a 1999 release has
1490 fixed the problem on an AlphaStation 200 4/166.
1492 ?? The conversion table for character set XX does not match with
1495 {UD} I don't doubt for a minute that some of the conversion tables contain
1496 errors. We tried the best we can and relied on automatic generation of the
1497 data to prevent human-introduced errors but this still is no guarantee. If
1498 you think you found a problem please send a bug report describing it and
1499 give an authoritive reference. The latter is important since otherwise
1500 the current behaviour is as good as the proposed one.
1502 Before doing this look through the list of known problem first:
1504 - the GBK (simplified Chinese) encoding is based on Unicode tables. This
1505 is good. These tables, however, differ slightly from the tables used
1506 by the M$ people. The differences are these [+ Unicode, - M$]:
1513 In addition the Unicode tables contain mappings for the GBK characters
1514 0xA8BC, 0xA8BF, 0xA989 to 0xA995, and 0xFE50 to 0xFEA0.
1516 - when mapping from EUC-CN to GBK and vice versa we ignore the fact that
1517 the coded character at position 0xA1A4 maps to different Unicode
1518 characters. Since the iconv() implementation can do whatever it wants
1519 if it cannot directly map a character this is a perfectly good solution
1520 since the semantics and appearance of the character does not change.
1523 Answers were given by:
1524 {UD} Ulrich Drepper, <drepper@cygnus.com>
1525 {DMT} David Mosberger-Tang, <davidm@AZStarNet.com>
1526 {RM} Roland McGrath, <roland@gnu.org>
1527 {AJ} Andreas Jaeger, <aj@suse.de>
1528 {EY} Eric Youngdale, <eric@andante.jic.com>
1529 {PB} Phil Blundell, <Philip.Blundell@pobox.com>
1530 {MK} Mark Kettenis, <kettenis@phys.uva.nl>
1531 {ZW} Zack Weinberg, <zack@rabi.phys.columbia.edu>
1532 {TK} Thorsten Kukuk, <kukuk@suse.de>
1533 {GK} Geoffrey Keating, <geoffk@ozemail.com.au>
1534 {HJ} H.J. Lu, <hjl@gnu.org>
1535 {CG} Cristian Gafton, <gafton@redhat.com>
1536 {AO} Alexandre Oliva, <oliva@lsd.ic.unicamp.br>
1540 outline-regexp:"\\?"
projects / platform / upstream / glibc.git / blob