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-*-linuxaout Linux-2.x on ARM using a.out binaries
38 Ports to other Linux platforms are in development, and may in fact work
39 already, but no one has sent us success reports for them. Currently no
40 ports to other operating systems are underway, although a few people have
43 If you have a system not listed above (or in the `README' file) and you are
44 really interested in porting it, contact
48 ?? What compiler do I need to build GNU libc?
50 {UD} You must use GNU CC to compile GNU libc. A lot of extensions of GNU CC
51 are used to increase portability and speed.
53 GNU CC is found, like all other GNU packages, on
55 ftp://ftp.gnu.org/pub/gnu
57 and the many mirror sites. ftp.gnu.org is always overloaded, so try to find
60 You always should try to use the latest official release. Older versions
61 may not have all the features GNU libc requires. The current releases of
62 egcs (1.0.2) and GNU CC (2.8.1) should work with the GNU C library (for
63 powerpc see question ?powerpc).
65 ?? When I try to compile glibc I get only error messages.
68 {UD} You definitely need GNU make to translate GNU libc. No other make
69 program has the needed functionality.
71 We recommend version GNU make version 3.75. Versions 3.76 and 3.76.1 have
72 bugs which appear when building big projects like GNU libc. Versions before
73 3.74 have bugs and/or are missing features.
75 ?? Do I need a special linker or archiver?
77 {UD} You may be able to use your system linker, but GNU libc works best with
80 On systems where the native linker does not support weak symbols you will
81 not get a fully ISO C compliant C library. Generally speaking you should
82 use the GNU binutils if they provide at least the same functionality as your
85 Always get the newest release of GNU binutils available. Older releases are
86 known to have bugs that prevent a successful compilation.
88 {ZW} As of release 2.1 a linker supporting symbol versions is required. For
89 Linux, get binutils-2.8.1.0.23 or later. Other systems may have native
90 linker support, but it's moot right now, because glibc has not been ported
93 ??powerpc Which compiler should I use for powerpc?
95 {GK} You want to use egcs 1.0.1 or later (together with the right versions
96 of all the other tools, of course).
98 In fact, egcs 1.0.1 has a serious bug that prevents a clean make, relating
99 to switch statement folding. It also causes the resulting shared libraries
100 to use more memory than they should. There is a patch at:
102 <http://discus.anu.edu.au/~geoffk/egcs-1.0.1-geoffk.diff>
104 Later versions of egcs may fix these problems.
107 ?? Do I need some more things to compile GNU C Library?
109 {UD} Yes, there are some more :-).
111 * GNU gettext. This package contains the tools needed to construct
112 `message catalog' files containing translated versions of system
113 messages. See ftp://ftp.gnu.org/pub/gnu or better any mirror
114 site. (We distribute compiled message catalogs, but they may not be
117 * Some files depend on special tools. E.g., files ending in .gperf
118 need a `gperf' program. The GNU version (part of libg++) is known
119 to work while some vendor versions do not.
121 You should not need these tools unless you change the source files.
123 * Some scripts need perl5 - but at the moment those scripts are not
124 vital for building and installing GNU libc (some data files will not
127 * When compiling for Linux, the header files of the Linux kernel must
128 be available to the compiler as <linux/*.h> and <asm/*.h>.
130 * lots of disk space (~170MB for i?86-linux; more for RISC platforms,
133 * plenty of time. Compiling just the shared and static libraries for
134 i?86-linux takes approximately 1h on an i586@133, or 2.5h on
135 i486@66, or 4.5h on i486@33. Multiply this by 1.5 or 2.0 if you
136 build profiling and/or the highly optimized version as well. For
137 Hurd systems times are much higher.
139 You should avoid compiling in a NFS mounted filesystem. This is
142 James Troup <J.J.Troup@comp.brad.ac.uk> reports a compile time of
143 45h34m for a full build (shared, static, and profiled) on Atari
144 Falcon (Motorola 68030 @ 16 Mhz, 14 Mb memory) and Jan Barte
145 <yann@plato.uni-paderborn.de> reports 22h48m on Atari TT030
146 (Motorola 68030 @ 32 Mhz, 34 Mb memory)
148 If you have some more measurements let me know.
150 ?? What version of the Linux kernel headers should be used?
152 {AJ,UD} The headers from the most recent Linux kernel should be used. The
153 headers used while compiling the GNU C library and the kernel binary used
154 when using the library do not need to match. The GNU C library runs without
155 problems on kernels that are older than the kernel headers used. The other
156 way round (compiling the GNU C library with old kernel headers and running
157 on a recent kernel) does not necessarily work. For example you can't use
158 new kernel features when using old kernel headers for compiling the GNU C
161 ?? The compiler hangs while building iconvdata modules. What's
164 {ZW} This is a problem with all current releases of GCC. Initialization of
165 large static arrays is very slow. The compiler will eventually finish; give
168 The problem will be fixed in egcs 1.1 but probably not before then.
170 ?? When I run `nm -u libc.so' on the produced library I still
171 find unresolved symbols. Can this be ok?
173 {UD} Yes, this is ok. There can be several kinds of unresolved symbols:
175 * magic symbols automatically generated by the linker. These have names
176 like __start_* and __stop_*
178 * symbols starting with _dl_* come from the dynamic linker
180 * weak symbols, which need not be resolved at all (fabs for example)
182 Generally, you should make sure you find a real program which produces
183 errors while linking before deciding there is a problem.
185 ??addon What are these `add-ons'?
187 {UD} To avoid complications with export rules or external source code some
188 optional parts of the libc are distributed as separate packages (e.g., the
189 crypt package, see ?crypt).
191 To use these packages as part of GNU libc, just unpack the tarfiles in the
192 libc source directory and tell the configuration script about them using the
193 --enable-add-ons option. If you give just --enable-add-ons configure tries
194 to find all the add-on packages in your source tree. This may not work. If
195 it doesn't, or if you want to select only a subset of the add-ons, give a
196 comma-separated list of the add-ons to enable:
198 configure --enable-add-ons=crypt,linuxthreads
202 Add-ons can add features (including entirely new shared libraries), override
203 files, provide support for additional architectures, and just about anything
204 else. The existing makefiles do most of the work; only some few stub rules
205 must be written to get everything running.
207 ?? My XXX kernel emulates a floating-point coprocessor for me.
208 Should I enable --with-fp?
210 {ZW} An emulated FPU is just as good as a real one, as far as the C library
211 is concerned. You only need to say --without-fp if your machine has no way
212 to execute floating-point instructions.
214 People who are interested in squeezing the last drop of performance
215 out of their machine may wish to avoid the trap overhead, but this is
216 far more trouble than it's worth: you then have to compile
217 *everything* this way, including the compiler's internal libraries
218 (libgcc.a for GNU C), because the calling conventions change.
220 ?? When compiling GNU libc I get lots of errors saying functions
221 in glibc are duplicated in libgcc.
223 {EY} This is *exactly* the same problem that I was having. The problem was
224 due to the fact that configure didn't correctly detect that the linker flag
225 --no-whole-archive was supported in my linker. In my case it was because I
226 had run ./configure with bogus CFLAGS, and the test failed.
228 One thing that is particularly annoying about this problem is that once this
229 is misdetected, running configure again won't fix it unless you first delete
232 {UD} Starting with glibc-2.0.3 there should be a better test to avoid some
233 problems of this kind. The setting of CFLAGS is checked at the very
234 beginning and if it is not usable `configure' will bark.
236 ?? Why do I get messages about missing thread functions when I use
237 librt? I don't even use threads.
239 {UD} In this case you probably mixed up your installation. librt uses
240 threads internally and has implicit references to the thread library.
241 Normally these references are satisfied automatically but if the thread
242 library is not in the expected place you must tell the linker where it is.
243 When using GNU ld it works like this:
245 gcc -o foo foo.c -Wl,-rpath-link=/some/other/dir -lrt
247 The `/some/other/dir' should contain the thread library. `ld' will use the
248 given path to find the implicitly referenced library while not disturbing
251 ?? What's the problem with configure --enable-omitfp?
253 {AJ} When --enable-omitfp is set the libraries are built without frame
254 pointers. Some compilers produce buggy code for this model and therefore we
255 don't advise using it at the moment.
257 If you use --enable-omitfp, you're on your own. If you encounter problems
258 with a library that was build this way, we advise you to rebuild the library
259 without --enable-omitfp. If the problem vanishes consider tracking the
260 problem down and report it as compiler failure.
262 Since a library build with --enable-omitfp is undebuggable on most systems,
263 debuggable libraries are also built - you can use it by appending "_g" to
266 The compilation of these extra libraries and the compiler optimizations slow
267 down the build process and need more disk space.
269 ? Installation and configuration issues
271 ?? Can I replace the libc on my Linux system with GNU libc?
273 {UD} You cannot replace any existing libc for Linux with GNU libc. It is
274 binary incompatible and therefore has a different major version. You can,
275 however, install it alongside your existing libc.
277 For Linux there are three major libc versions:
279 libc-5 original ELF libc
282 You can have any combination of these three installed. For more information
283 consult documentation for shared library handling. The Makefiles of GNU
284 libc will automatically generate the needed symbolic links which the linker
287 ?? How do I configure GNU libc so that the essential libraries
288 like libc.so go into /lib and the other into /usr/lib?
290 {UD,AJ} Like all other GNU packages GNU libc is designed to use a base
291 directory and install all files relative to this. The default is
292 /usr/local, because this is safe (it will not damage the system if installed
293 there). If you wish to install GNU libc as the primary C library on your
294 system, set the base directory to /usr (i.e. run configure --prefix=/usr
295 <other_options>). Note that this can damage your system; see ?safety for
298 Some systems like Linux have a filesystem standard which makes a difference
299 between essential libraries and others. Essential libraries are placed in
300 /lib because this directory is required to be located on the same disk
301 partition as /. The /usr subtree might be found on another
302 partition/disk. If you configure for Linux with --prefix=/usr, then this
303 will be done automatically.
305 To install the essential libraries which come with GNU libc in /lib on
306 systems other than Linux one must explicitly request it. Autoconf has no
307 option for this so you have to use a `configparms' file (see the `INSTALL'
308 file for details). It should contain:
313 The first line specifies the directory for the essential libraries, the
314 second line the directory for system configuration files.
316 ??safety How should I avoid damaging my system when I install GNU libc?
318 {ZW} If you wish to be cautious, do not configure with --prefix=/usr. If
319 you don't specify a prefix, glibc will be installed in /usr/local, where it
320 will probably not break anything. (If you wish to be certain, set the
321 prefix to something like /usr/local/glibc2 which is not used for anything.)
323 The dangers when installing glibc in /usr are twofold:
325 * glibc will overwrite the headers in /usr/include. Other C libraries
326 install a different but overlapping set of headers there, so the
327 effect will probably be that you can't compile anything. You need to
328 rename /usr/include out of the way first. (Do not throw it away; you
329 will then lose the ability to compile programs against your old libc.)
331 * None of your old libraries, static or shared, can be used with a
332 different C library major version. For shared libraries this is not a
333 problem, because the filenames are different and the dynamic linker
334 will enforce the restriction. But static libraries have no version
335 information. You have to evacuate all the static libraries in
336 /usr/lib to a safe location.
338 The situation is rather similar to the move from a.out to ELF which
339 long-time Linux users will remember.
341 ?? Do I need to use GNU CC to compile programs that will use the
344 {ZW} In theory, no; the linker does not care, and the headers are supposed
345 to check for GNU CC before using its extensions to the C language.
347 However, there are currently no ports of glibc to systems where another
348 compiler is the default, so no one has tested the headers extensively
349 against another compiler. You may therefore encounter difficulties. If you
350 do, please report them as bugs.
352 Also, in several places GNU extensions provide large benefits in code
353 quality. For example, the library has hand-optimized, inline assembly
354 versions of some string functions. These can only be used with GCC. See
357 ??crypt When linking with the new libc I get unresolved symbols
358 `crypt' and `setkey'. Why aren't these functions in the
361 {UD} The US places restrictions on exporting cryptographic programs and
362 source code. Until this law gets abolished we cannot ship the cryptographic
363 functions together with glibc.
365 The functions are available, as an add-on (see ?addon). People in the US
366 may get it from the same place they got GNU libc from. People outside the
367 US should get the code from ftp://ftp.ifi.uio.no/pub/gnu, or another archive
368 site outside the USA. The README explains how to install the sources.
370 If you already have the crypt code on your system the reason for the failure
371 is probably that you did not link with -lcrypt. The crypto functions are in
372 a separate library to make it possible to export GNU libc binaries from the
375 ?? When I use GNU libc on my Linux system by linking against
376 the libc.so which comes with glibc all I get is a core dump.
378 {UD} On Linux, gcc sets the dynamic linker to /lib/ld-linux.so.1 unless the
379 user specifies a -dynamic-linker argument. This is the name of the libc5
380 dynamic linker, which does not work with glibc.
382 For casual use of GNU libc you can just specify
383 -dynamic-linker=/lib/ld-linux.so.2
385 which is the glibc dynamic linker, on Linux systems. On other systems the
386 name is /lib/ld.so.1.
388 To change your environment to use GNU libc for compiling you need to change
389 the `specs' file of your gcc. This file is normally found at
391 /usr/lib/gcc-lib/<arch>/<version>/specs
393 In this file you have to change a few things:
395 - change `ld-linux.so.1' to `ld-linux.so.2'
397 - remove all expression `%{...:-lgmon}'; there is no libgmon in glibc
399 - fix a minor bug by changing %{pipe:-} to %|
401 Here is what the gcc-2.7.2 specs file should look like when GNU libc is
404 -----------------------------------------------------------------------
406 %{V} %{v:%{!V:-V}} %{Qy:} %{!Qn:-Qy} %{n} %{T} %{Ym,*} %{Yd,*} %{Wa,*:%*}
412 %{fPIC:-D__PIC__ -D__pic__} %{fpic:-D__PIC__ -D__pic__} %{!m386:-D__i486__} %{posix:-D_POSIX_SOURCE} %{pthread:-D_REENTRANT}
421 %{!shared:crtend.o%s} %{shared:crtendS.o%s} crtn.o%s
424 -m elf_i386 %{shared:-shared} %{!shared: %{!ibcs: %{!static: %{rdynamic:-export-dynamic} %{!dynamic-linker:-dynamic-linker /lib/ld-linux.so.2}} %{static:-static}}}
427 %{!shared: %{pthread:-lpthread} %{profile:-lc_p} %{!profile: -lc}}
433 %{!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}
435 *switches_need_spaces:
439 %{funsigned-char:-D__CHAR_UNSIGNED__}
442 -D__ELF__ -Dunix -Di386 -Dlinux -Asystem(unix) -Asystem(posix) -Acpu(i386) -Amachine(i386)
450 -----------------------------------------------------------------------
452 Things get a bit more complicated if you have GNU libc installed in some
453 other place than /usr, i.e., if you do not want to use it instead of the old
454 libc. In this case the needed startup files and libraries are not found in
455 the regular places. So the specs file must tell the compiler and linker
458 Version 2.7.2.3 does and future versions of GCC will automatically
459 provide the correct specs.
461 ?? Looking through the shared libc file I haven't found the
462 functions `stat', `lstat', `fstat', and `mknod' and while
463 linking on my Linux system I get error messages. How is
464 this supposed to work?
466 {RM} Believe it or not, stat and lstat (and fstat, and mknod) are supposed
467 to be undefined references in libc.so.6! Your problem is probably a missing
468 or incorrect /usr/lib/libc.so file; note that this is a small text file now,
469 not a symlink to libc.so.6. It should look something like this:
471 GROUP ( libc.so.6 libc_nonshared.a )
473 ?? How can I compile gcc 2.7.2.1 from the gcc source code using
476 {AJ} There's only correct support for glibc 2.0.x in gcc 2.7.2.3 or later.
477 But you should get at least gcc 2.8.1 or egcs 1.0.2 (or later versions)
480 ?? The `gencat' utility cannot process the catalog sources which
481 were used on my Linux libc5 based system. Why?
483 {UD} The `gencat' utility provided with glibc complies to the XPG standard.
484 The older Linux version did not obey the standard, so they are not
487 To ease the transition from the Linux version some of the non-standard
488 features are also present in the `gencat' program of GNU libc. This mainly
489 includes the use of symbols for the message number and the automatic
490 generation of header files which contain the needed #defines to map the
493 Here is a simple SED script to convert at least some Linux specific catalog
494 files to the XPG4 form:
496 -----------------------------------------------------------------------
497 # Change catalog source in Linux specific format to standard XPG format.
498 # Ulrich Drepper <drepper@cygnus.com>, 1996.
502 s/\$ #\([^ ]*\).*/\1/
504 s/\$ #[^ ]* *\(.*\)/\$ \1/
510 s/\(.*\)\n\(.*\)/\2 \1/
512 -----------------------------------------------------------------------
514 ?? Programs using libc have their messages translated, but other
515 behavior is not localized (e.g. collating order); why?
517 {ZW} Translated messages are automatically installed, but the locale
518 database that controls other behaviors is not. You need to run localedef to
519 install this database, after you have run `make install'. For example, to
520 set up the French Canadian locale, simply issue the command
522 localedef -i fr_CA -f ISO-8859-1 fr_CA
524 Please see localedata/README in the source tree for further details.
526 ?? I have set up /etc/nis.conf, and the Linux libc 5 with NYS
527 works great. But the glibc NIS+ doesn't seem to work.
529 {TK} The glibc NIS+ implementation uses a /var/nis/NIS_COLD_START file for
530 storing information about the NIS+ server and their public keys, because the
531 nis.conf file does not contain all the necessary information. You have to
532 copy a NIS_COLD_START file from a Solaris client (the NIS_COLD_START file is
533 byte order independent) or generate it with nisinit from the nis-tools
534 package; available at
536 http://www-vt.uni-paderborn.de/~kukuk/linux/nisplus.html
538 ?? I have killed ypbind to stop using NIS, but glibc
541 {TK} For faster NIS lookups, glibc uses the /var/yp/binding/ files from
542 ypbind. ypbind 3.3 and older versions don't always remove these files, so
543 glibc will continue to use them. Other BSD versions seem to work correctly.
544 Until ypbind 3.4 is released, you can find a patch at
546 ftp://ftp.kernel.org/pub/linux/utils/net/NIS/ypbind-3.3-glibc3.diff.
548 ?? Under Linux/Alpha, I always get "do_ypcall: clnt_call:
549 RPC: Unable to receive; errno = Connection refused" when using NIS.
551 {TK} You need a ypbind version which is 64bit clean. Some versions are not
552 64bit clean. A 64bit clean implementation is ypbind-mt. For ypbind 3.3,
553 you need the patch from ftp.kernel.org (See the previous question). I don't
554 know about other versions.
557 ?? After installing glibc name resolving doesn't work properly.
559 {AJ} You probably should read the manual section describing nsswitch.conf
560 (just type `info libc "NSS Configuration File"'). The NSS configuration
561 file is usually the culprit.
564 ?? How do I create the databases for NSS?
566 {AJ} If you have an entry "db" in /etc/nsswitch.conf you should also create
567 the database files. The glibc sources contain a Makefile which does the
568 neccessary conversion and calls to create those files. The file is
569 `db-Makefile' in the subdirectory `nss' and you can call it with `make -f
570 db-Makefile'. Please note that not all services are capable of using a
571 database. Currently passwd, group, ethers, protocol, rpc, services shadow
572 and netgroup are implemented.
575 ?? I have /usr/include/net and /usr/include/scsi as symlinks
576 into my Linux source tree. Is that wrong?
578 {PB} This was necessary for libc5, but is not correct when using glibc.
579 Including the kernel header files directly in user programs usually does not
580 work (see ?kerhdr). glibc provides its own <net/*> and <scsi/*> header
581 files to replace them, and you may have to remove any symlink that you have
582 in place before you install glibc. However, /usr/include/asm and
583 /usr/include/linux should remain as they were.
585 ?? Programs like `logname', `top', `uptime' `users', `w' and
586 `who', show incorrect information about the (number of)
587 users on my system. Why?
591 ?? After upgrading to glibc 2.1 with symbol versioning I get
592 errors about undefined symbols. What went wrong?
594 {AJ} The problem is caused either by wrong program code or tools. In the
595 versioned libc a lot of symbols are now local that were global symbols in
596 previous versions. It seems that programs linked against older versions
597 often accidentally used libc global variables -- something that should not
600 The only way to fix this is to recompile your program. Sorry, that's the
601 price you might have to pay once for quite a number of advantages with
604 ?? When I start the program XXX after upgrading the library
606 XXX: Symbol `_sys_errlist' has different size in shared
607 object, consider re-linking
608 Why? What should I do?
610 {UD} As the message says, relink the binary. The problem is that a few
611 symbols from the library can change in size and there is no way to avoid
612 this. _sys_errlist is a good example. Occasionally there are new error
613 numbers added to the kernel and this must be reflected at user level,
614 breaking programs that refer to them directly.
616 Such symbols should normally not be used at all. There are mechanisms to
617 avoid using them. In the case of _sys_errlist, there is the strerror()
618 function which should _always_ be used instead. So the correct fix is to
619 rewrite that part of the application.
621 In some situations (especially when testing a new library release) it might
622 be possible that a symbol changed size when that should not have happened.
623 So in case of doubt report such a warning message as a problem.
625 ?? What do I need for C++ development?
627 {HJ,AJ} You need either egcs 1.0.2 or gcc-2.8.1 with libstdc++ 2.8.1 (or
628 more recent versions). libg++ 2.7.2 (and the Linux Versions 2.7.2.x) doesn't
629 work very well with the GNU C library due to vtable thunks. If you're
630 upgrading from glibc 2.0.x to 2.1 you have to recompile libstdc++ since the
631 library compiled for 2.0 is not compatible due to the new Large File Support
632 (LFS) in version 2.1.
634 ?? Even statically linked programs need some shared libraries
635 which is not acceptable for me. What can I do?
637 {AJ} NSS (for details just type `info libc "Name Service Switch"') won't
638 work properly without shared libraries. NSS allows using different services
639 (e.g. NIS, files, db, hesiod) by just changing one configuration file
640 (/etc/nsswitch.conf) without relinking any programs. The only disadvantage
641 is that now static libraries need to access shared libraries. This is
642 handled transparently by the GNU C library.
644 A solution is to configure glibc with --enable-static-nss. In this case you
645 can create a static binary that will use only the services dns and files
646 (change /etc/nsswitch.conf for this). You need to link explicitly against
647 all these services. For example:
649 gcc -static test-netdb.c -o test-netdb.c \
650 -lc -lnss_files -lnss_dns -lresolv
652 The problem with this approach is that you've got to link every static
653 program that uses NSS routines with all those libraries.
655 {UD} In fact, one cannot say anymore that a libc compiled with this
656 option is using NSS. There is no switch anymore. Therefore it is
657 *highly* recommended *not* to use --enable-static-nss since this makes
658 the behaviour of the programs on the system inconsistent.
660 ? Source and binary incompatibilities, and what to do about them
662 ?? I expect GNU libc to be 100% source code compatible with
663 the old Linux based GNU libc. Why isn't it like this?
665 {DMT,UD} Not every extension in Linux libc's history was well thought-out.
666 In fact it had a lot of problems with standards compliance and with
667 cleanliness. With the introduction of a new version number these errors can
668 now be corrected. Here is a list of the known source code
671 * _GNU_SOURCE: glibc does not make the GNU extensions available
672 automatically. If a program depends on GNU extensions or some
673 other non-standard functionality, it is necessary to compile it
674 with the C compiler option -D_GNU_SOURCE, or better, to put
675 `#define _GNU_SOURCE' at the beginning of your source files, before
676 any C library header files are included. This difference normally
677 manifests itself in the form of missing prototypes and/or data type
678 definitions. Thus, if you get such errors, the first thing you
679 should do is try defining _GNU_SOURCE and see if that makes the
682 For more information consult the file `NOTES' in the GNU C library
685 * reboot(): GNU libc sanitizes the interface of reboot() to be more
686 compatible with the interface used on other OSes. reboot() as
687 implemented in glibc takes just one argument. This argument
688 corresponds to the third argument of the Linux reboot system call.
689 That is, a call of the form reboot(a, b, c) needs to be changed into
690 reboot(c). Beside this the header <sys/reboot.h> defines the needed
691 constants for the argument. These RB_* constants should be used
692 instead of the cryptic magic numbers.
694 * swapon(): the interface of this function didn't change, but the
695 prototype is in a separate header file <sys/swap.h>. This header
696 file also provides the SWAP_* constants defined by <linux/swap.h>;
697 you should use them for the second argument to swapon().
699 * errno: If a program uses the variable "errno", then it _must_
700 include <errno.h>. The old libc often (erroneously) declared this
701 variable implicitly as a side-effect of including other libc header
702 files. glibc is careful to avoid such namespace pollution, which,
703 in turn, means that you really need to include the header files that
704 you depend on. This difference normally manifests itself in the
705 form of the compiler complaining about references to an undeclared
708 * Linux-specific syscalls: All Linux system calls now have appropriate
709 library wrappers and corresponding declarations in various header files.
710 This is because the syscall() macro that was traditionally used to
711 work around missing syscall wrappers are inherently non-portable and
712 error-prone. The following table lists all the new syscall stubs,
713 the header-file declaring their interface and the system call name.
715 syscall name: wrapper name: declaring header file:
716 ------------- ------------- ----------------------
717 bdflush bdflush <sys/kdaemon.h>
718 syslog ksyslog_ctl <sys/klog.h>
720 * lpd: Older versions of lpd depend on a routine called _validuser().
721 The library does not provide this function, but instead provides
722 __ivaliduser() which has a slightly different interface. Simply
723 upgrading to a newer lpd should fix this problem (e.g., the 4.4BSD
724 lpd is known to be working).
726 * resolver functions/BIND: like on many other systems the functions of
727 the resolver library are not included in libc itself. There is a
728 separate library libresolv. If you get undefined symbol errors for
729 symbols starting with `res_*' simply add -lresolv to your linker
732 * the `signal' function's behavior corresponds to the BSD semantic and
733 not the SysV semantic as it was in libc-5. The interface on all GNU
734 systems shall be the same and BSD is the semantic of choice. To use
735 the SysV behavior simply use `sysv_signal', or define _XOPEN_SOURCE.
736 See ?signal for details.
738 ??getlog Why does getlogin() always return NULL on my Linux box?
740 {UD} The GNU C library has a format for the UTMP and WTMP file which differs
741 from what your system currently has. It was extended to fulfill the needs
742 of the next years when IPv6 is introduced. The record size is different and
743 some fields have different positions. The files written by functions from
744 the one library cannot be read by functions from the other library. Sorry,
745 but this is what a major release is for. It's better to have a cut now than
746 having no means to support the new techniques later.
748 {MK} There is however a (partial) solution for this problem. Please take a
749 look at the file `login/README.utmpd'.
751 ?? Where are the DST_* constants found in <sys/time.h> on many
754 {UD} These constants come from the old BSD days and are not used anymore
755 (libc5 does not actually implement the handling although the constants are
758 Instead GNU libc contains zone database support and compatibility code for
759 POSIX TZ environment variable handling.
761 ?? The prototypes for `connect', `accept', `getsockopt',
762 `setsockopt', `getsockname', `getpeername', `send',
763 `sendto', and `recvfrom' are different in GNU libc from
764 any other system I saw. This is a bug, isn't it?
766 {UD} No, this is no bug. This version of GNU libc already follows the new
767 Single Unix specifications (and I think the POSIX.1g draft which adopted the
768 solution). The type for a parameter describing a size is now `socklen_t', a
771 ??kerhdr On Linux I've got problems with the declarations in Linux
774 {UD,AJ} On Linux, the use of kernel headers is reduced to the minimum. This
775 gives Linus the ability to change the headers more freely. Also, user
776 programs are now insulated from changes in the size of kernel data
779 For example, the sigset_t type is 32 or 64 bits wide in the kernel. In
780 glibc it is 1024 bits wide. This guarantees that when the kernel gets a
781 bigger sigset_t (for POSIX.1e realtime support, say) user programs will not
782 have to be recompiled. Consult the header files for more information about
785 Therefore you shouldn't include Linux kernel header files directly if glibc
786 has defined a replacement. Otherwise you might get undefined results because
789 ?? I don't include any kernel headers myself but the compiler
790 still complains about redeclarations of types in the kernel
793 {UD} The kernel headers before Linux 2.1.61 and 2.0.32 don't work correctly
794 with glibc. Compiling C programs is possible in most cases but C++ programs
795 have (due to the change of the name lookups for `struct's) problems. One
796 prominent example is `struct fd_set'.
798 There might be some problems left but 2.1.61/2.0.32 fix most of the known
799 ones. See the BUGS file for other known problems.
801 ??signal Why don't signals interrupt system calls anymore?
803 {ZW} By default GNU libc uses the BSD semantics for signal(), unlike Linux
804 libc 5 which used System V semantics. This is partially for compatibility
805 with other systems and partially because the BSD semantics tend to make
806 programming with signals easier.
808 There are three differences:
810 * BSD-style signals that occur in the middle of a system call do not
811 affect the system call; System V signals cause the system call to
812 fail and set errno to EINTR.
814 * BSD signal handlers remain installed once triggered. System V signal
815 handlers work only once, so one must reinstall them each time.
817 * A BSD signal is blocked during the execution of its handler. In other
818 words, a handler for SIGCHLD (for example) does not need to worry about
819 being interrupted by another SIGCHLD. It may, however, be interrupted
822 There is general consensus that for `casual' programming with signals, the
823 BSD semantics are preferable. You don't need to worry about system calls
824 returning EINTR, and you don't need to worry about the race conditions
825 associated with one-shot signal handlers.
827 If you are porting an old program that relies on the old semantics, you can
828 quickly fix the problem by changing signal() to sysv_signal() throughout.
829 Alternatively, define _XOPEN_SOURCE before including <signal.h>.
831 For new programs, the sigaction() function allows you to specify precisely
832 how you want your signals to behave. All three differences listed above are
833 individually switchable on a per-signal basis with this function.
835 If all you want is for one specific signal to cause system calls to fail and
836 return EINTR (for example, to implement a timeout) you can do this with
840 ??string I've got errors compiling code that uses certain string
843 {AJ} glibc 2.1 has special string functions that are faster than the normal
844 library functions. Some of the functions are additionally implemented as
845 inline functions and others as macros.
847 The optimized string functions are only used when compiling with
848 optimizations (-O1 or higher). The behavior can be changed with two feature
851 * __NO_STRING_INLINES: Don't do any string optimizations.
852 * __USE_STRING_INLINES: Use assembly language inline functions (might
853 increase code size dramatically).
855 Since some of these string functions are now additionally defined as macros,
856 code like "char *strncpy();" doesn't work anymore (and is unnecessary, since
857 <string.h> has the necessary declarations). Either change your code or
858 define __NO_STRING_INLINES.
860 {UD} Another problem in this area is that gcc still has problems on machines
861 with very few registers (e.g., ix86). The inline assembler code can require
862 almost all the registers and the register allocator cannot always handle
865 One can disable the string optimizations selectively. Instead of writing
867 cp = strcpy (foo, "lkj");
871 cp = (strcpy) (foo, "lkj");
873 This disables the optimization for that specific call.
875 ?? I get compiler messages "Initializer element not constant" with
876 stdin/stdout/stderr. Why?
878 {RM,AJ} Constructs like:
879 static FILE *InPtr = stdin;
881 lead to this message. This is correct behaviour with glibc since stdin is
882 not a constant expression. Please note that a strict reading of ISO C does
883 not allow above constructs.
885 One of the advantages of this is that you can assign to stdin, stdout, and
886 stderr just like any other global variable (e.g. `stdout = my_stream;'),
887 which can be very useful with custom streams that you can write with libio
888 (but beware this is not necessarily portable). The reason to implement it
889 this way were versioning problems with the size of the FILE structure.
892 ?? I can't compile with gcc -traditional (or
893 -traditional-cpp). Why?
895 {AJ} glibc2 does break -traditional and -traditonal-cpp - and will continue
896 to do so. For example constructs of the form:
902 are useful for debugging purposes (you can use foo with your debugger that's
903 why we need the enum) and for compatibility (other systems use defines and
906 ?? I get some errors with `gcc -ansi'. Isn't glibc ANSI compatible?
908 {AJ} The GNU C library is compatible with the ANSI/ISO C standard. If
909 you're using `gcc -ansi', the glibc includes which are specified in the
910 standard follow the standard. The ANSI/ISO C standard defines what has to be
911 in the include files - and also states that nothing else should be in the
912 include files (btw. you can still enable additional standards with feature
915 The GNU C library is conforming to ANSI/ISO C - if and only if you're only
916 using the headers and library functions defined in the standard.
918 ?? I can't access some functions anymore. nm shows that they do
919 exist but linking fails nevertheless.
921 {AJ} With the introduction of versioning in glibc 2.1 it is possible to
922 export only those identifiers (functions, variables) that are really needed
923 by application programs and by other parts of glibc. This way a lot of
924 internal interfaces are now hidden. nm will still show those identifiers
925 but marking them as internal. ISO C states that identifiers beginning with
926 an underscore are internal to the libc. An application program normally
927 shouldn't use those internal interfaces (there are exceptions,
928 e.g. __ivaliduser). If a program uses these interfaces, it's broken. These
929 internal interfaces might change between glibc releases or dropped
935 ?? After I changed configure.in I get `Autoconf version X.Y.
936 or higher is required for this script'. What can I do?
938 {UD} You have to get the specified autoconf version (or a later one)
939 from your favorite mirror of ftp.gnu.org.
941 ?? When I try to compile code which uses IPv6 headers and
942 definitions on my Linux 2.x.y system I am in trouble.
943 Nothing seems to work.
945 {UD} The problem is that IPv6 development still has not reached a point
946 where the headers are stable. There are still lots of incompatible changes
947 made and the libc headers have to follow.
949 Also, make sure you have a suitably recent kernel. As of the 970401
950 snapshot, according to Philip Blundell <Philip.Blundell@pobox.com>, the
951 required kernel version is at least 2.1.30.
953 ?? When I set the timezone by setting the TZ environment variable
954 to EST5EDT things go wrong since glibc computes the wrong time
955 from this information.
957 {UD} The problem is that people still use the braindamaged POSIX method to
958 select the timezone using the TZ environment variable with a format EST5EDT
959 or whatever. People, read the POSIX standard, the implemented behaviour is
960 correct! What you see is in fact the result of the decisions made while
961 POSIX.1 was created. We've only implemented the handling of TZ this way to
962 be POSIX compliant. It is not really meant to be used.
964 The alternative approach to handle timezones which is implemented is the
965 correct one to use: use the timezone database. This avoids all the problems
966 the POSIX method has plus it is much easier to use. Simply run the tzselect
967 shell script, answer the question and use the name printed in the end by
968 making a symlink to /usr/share/zoneinfo/NAME (NAME is the returned value
969 from tzselect) from the file /etc/localtime. That's all. You never again
972 So, please avoid sending bug reports about time related problems if you use
973 the POSIX method and you have not verified something is really broken by
974 reading the POSIX standards.
977 Answers were given by:
978 {UD} Ulrich Drepper, <drepper@cygnus.com>
979 {DMT} David Mosberger-Tang, <davidm@AZStarNet.com>
980 {RM} Roland McGrath, <roland@gnu.org>
981 {AJ} Andreas Jaeger, <aj@arthur.rhein-neckar.de>
982 {EY} Eric Youngdale, <eric@andante.jic.com>
983 {PB} Phil Blundell, <Philip.Blundell@pobox.com>
984 {MK} Mark Kettenis, <kettenis@phys.uva.nl>
985 {ZW} Zack Weinberg, <zack@rabi.phys.columbia.edu>
986 {TK} Thorsten Kukuk, <kukuk@vt.uni-paderborn.de>
987 {GK} Geoffrey Keating, <geoffk@ozemail.com.au>
988 {HJ} H.J. Lu, <hjl@gnu.org>
projects / platform / upstream / glibc.git / blob