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,
17 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
21 1.1. What systems does the GNU C Library run on?
22 1.2. What compiler do I need to build GNU libc?
23 1.3. When I try to compile glibc I get only error messages.
25 1.4. Do I need a special linker or archiver?
26 1.5. Which compiler should I use for powerpc?
27 1.6. Do I need some more things to compile GNU C Library?
28 1.7. What version of the Linux kernel headers should be used?
29 1.8. The compiler hangs while building iconvdata modules. What's
31 1.9. When I run `nm -u libc.so' on the produced library I still
32 find unresolved symbols. Can this be ok?
33 1.10. What are these `add-ons'?
34 1.11. My XXX kernel emulates a floating-point coprocessor for me.
35 Should I enable --with-fp?
36 1.12. When compiling GNU libc I get lots of errors saying functions
37 in glibc are duplicated in libgcc.
38 1.13. Why do I get messages about missing thread functions when I use
39 librt? I don't even use threads.
40 1.14. What's the problem with configure --enable-omitfp?
42 2. Installation and configuration issues
44 2.1. Can I replace the libc on my Linux system with GNU libc?
45 2.2. How do I configure GNU libc so that the essential libraries
46 like libc.so go into /lib and the other into /usr/lib?
47 2.3. How should I avoid damaging my system when I install GNU libc?
48 2.4. Do I need to use GNU CC to compile programs that will use the
50 2.5. When linking with the new libc I get unresolved symbols
51 `crypt' and `setkey'. Why aren't these functions in the
53 2.6. When I use GNU libc on my Linux system by linking against
54 the libc.so which comes with glibc all I get is a core dump.
55 2.7. Looking through the shared libc file I haven't found the
56 functions `stat', `lstat', `fstat', and `mknod' and while
57 linking on my Linux system I get error messages. How is
58 this supposed to work?
59 2.8. How can I compile gcc 2.7.2.1 from the gcc source code using
61 2.9. The `gencat' utility cannot process the catalog sources which
62 were used on my Linux libc5 based system. Why?
63 2.10. Programs using libc have their messages translated, but other
64 behavior is not localized (e.g. collating order); why?
65 2.11. I have set up /etc/nis.conf, and the Linux libc 5 with NYS
66 works great. But the glibc NIS+ doesn't seem to work.
67 2.12. I have killed ypbind to stop using NIS, but glibc
69 2.13. Under Linux/Alpha, I always get "do_ypcall: clnt_call:
70 RPC: Unable to receive; errno = Connection refused" when using NIS.
71 2.14. After installing glibc name resolving doesn't work properly.
72 2.15. I have /usr/include/net and /usr/include/scsi as symlinks
73 into my Linux source tree. Is that wrong?
74 2.16. Programs like `logname', `top', `uptime' `users', `w' and
75 `who', show incorrect information about the (number of)
76 users on my system. Why?
77 2.17. After upgrading to glibc 2.1 with symbol versioning I get
78 errors about undefined symbols. What went wrong?
79 2.18. When I start the program XXX after upgrading the library
81 XXX: Symbol `_sys_errlist' has different size in shared
82 object, consider re-linking
83 Why? What should I do?
84 2.19. What do I need for C++ development?
85 2.20. Even statically linked programs need some shared libraries
86 which is not acceptable for me. What can I do?
88 3. Source and binary incompatibilities, and what to do about them
90 3.1. I expect GNU libc to be 100% source code compatible with
91 the old Linux based GNU libc. Why isn't it like this?
92 3.2. Why does getlogin() always return NULL on my Linux box?
93 3.3. Where are the DST_* constants found in <sys/time.h> on many
95 3.4. The prototypes for `connect', `accept', `getsockopt',
96 `setsockopt', `getsockname', `getpeername', `send',
97 `sendto', and `recvfrom' are different in GNU libc from
98 any other system I saw. This is a bug, isn't it?
99 3.5. On Linux I've got problems with the declarations in Linux
101 3.6. I don't include any kernel headers myself but the compiler
102 still complains about redeclarations of types in the kernel
104 3.7. Why don't signals interrupt system calls anymore?
105 3.8. I've got errors compiling code that uses certain string
107 3.9. I get compiler messages "Initializer element not constant" with
108 stdin/stdout/stderr. Why?
109 3.10. I can't compile with gcc -traditional (or
110 -traditional-cpp). Why?
111 3.11. I get some errors with `gcc -ansi'. Isn't glibc ANSI compatible?
112 3.12. I can't access some functions anymore. nm shows that they do
113 exist but linking fails nevertheless.
117 4.1. After I changed configure.in I get `Autoconf version X.Y.
118 or higher is required for this script'. What can I do?
119 4.2. When I try to compile code which uses IPv6 headers and
120 definitions on my Linux 2.x.y system I am in trouble.
121 Nothing seems to work.
122 4.3. When I set the timezone by setting the TZ environment variable
123 to EST5EDT things go wrong since glibc computes the wrong time
124 from this information.
127 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
131 1.1. What systems does the GNU C Library run on?
133 {UD} This is difficult to answer. The file `README' lists the architectures
134 GNU libc was known to run on *at some time*. This does not mean that it
135 still can be compiled and run on them now.
137 The systems glibc is known to work on as of this release, and most probably
141 i[3456]86-*-linux-gnu Linux-2.x on Intel
142 m68k-*-linux-gnu Linux-2.x on Motorola 680x0
143 alpha-*-linux-gnu Linux-2.x on DEC Alpha
144 powerpc-*-linux-gnu Linux and MkLinux on PowerPC systems
145 sparc-*-linux-gnu Linux-2.x on SPARC
146 sparc64-*-linux-gnu Linux-2.x on UltraSPARC
147 arm-*-none ARM standalone systems
148 arm-*-linuxaout Linux-2.x on ARM using a.out binaries
150 Ports to other Linux platforms are in development, and may in fact work
151 already, but no one has sent us success reports for them. Currently no
152 ports to other operating systems are underway, although a few people have
155 If you have a system not listed above (or in the `README' file) and you are
156 really interested in porting it, contact
161 1.2. What compiler do I need to build GNU libc?
163 {UD} You must use GNU CC to compile GNU libc. A lot of extensions of GNU CC
164 are used to increase portability and speed.
166 GNU CC is found, like all other GNU packages, on
168 ftp://ftp.gnu.org/pub/gnu
170 and the many mirror sites. ftp.gnu.org is always overloaded, so try to find
171 a local mirror first.
173 You always should try to use the latest official release. Older versions
174 may not have all the features GNU libc requires. The current releases of
175 egcs (1.0.2) and GNU CC (2.8.1) should work with the GNU C library (for
176 powerpc see question question 1.5).
179 1.3. When I try to compile glibc I get only error messages.
182 {UD} You definitely need GNU make to translate GNU libc. No other make
183 program has the needed functionality.
185 We recommend version GNU make version 3.75. Versions 3.76 and 3.76.1 have
186 bugs which appear when building big projects like GNU libc. Versions before
187 3.74 have bugs and/or are missing features.
190 1.4. Do I need a special linker or archiver?
192 {UD} You may be able to use your system linker, but GNU libc works best with
195 On systems where the native linker does not support weak symbols you will
196 not get a fully ISO C compliant C library. Generally speaking you should
197 use the GNU binutils if they provide at least the same functionality as your
200 Always get the newest release of GNU binutils available. Older releases are
201 known to have bugs that prevent a successful compilation.
203 {ZW} As of release 2.1 a linker supporting symbol versions is required. For
204 Linux, get binutils-2.8.1.0.23 or later. Other systems may have native
205 linker support, but it's moot right now, because glibc has not been ported
209 1.5. Which compiler should I use for powerpc?
211 {GK} You want to use egcs 1.0.1 or later (together with the right versions
212 of all the other tools, of course).
214 In fact, egcs 1.0.1 has a serious bug that prevents a clean make, relating
215 to switch statement folding. It also causes the resulting shared libraries
216 to use more memory than they should. There is a patch at:
218 <http://discus.anu.edu.au/~geoffk/egcs-1.0.1-geoffk.diff>
220 Later versions of egcs may fix these problems.
223 1.6. Do I need some more things to compile GNU C Library?
225 {UD} Yes, there are some more :-).
227 * GNU gettext. This package contains the tools needed to construct
228 `message catalog' files containing translated versions of system
229 messages. See ftp://ftp.gnu.org/pub/gnu or better any mirror
230 site. (We distribute compiled message catalogs, but they may not be
233 * Some files depend on special tools. E.g., files ending in .gperf
234 need a `gperf' program. The GNU version (part of libg++) is known
235 to work while some vendor versions do not.
237 You should not need these tools unless you change the source files.
239 * Some scripts need perl5 - but at the moment those scripts are not
240 vital for building and installing GNU libc (some data files will not
243 * When compiling for Linux, the header files of the Linux kernel must
244 be available to the compiler as <linux/*.h> and <asm/*.h>.
246 * lots of disk space (~170MB for i?86-linux; more for RISC platforms,
249 * plenty of time. Compiling just the shared and static libraries for
250 i?86-linux takes approximately 1h on an i586@133, or 2.5h on
251 i486@66, or 4.5h on i486@33. Multiply this by 1.5 or 2.0 if you
252 build profiling and/or the highly optimized version as well. For
253 Hurd systems times are much higher.
255 You should avoid compiling in a NFS mounted filesystem. This is
258 James Troup <J.J.Troup@comp.brad.ac.uk> reports a compile time of
259 45h34m for a full build (shared, static, and profiled) on Atari
260 Falcon (Motorola 68030 @ 16 Mhz, 14 Mb memory) and Jan Barte
261 <yann@plato.uni-paderborn.de> reports 22h48m on Atari TT030
262 (Motorola 68030 @ 32 Mhz, 34 Mb memory)
264 If you have some more measurements let me know.
267 1.7. What version of the Linux kernel headers should be used?
269 {AJ,UD} The headers from the most recent Linux kernel should be used. The
270 headers used while compiling the GNU C library and the kernel binary used
271 when using the library do not need to match. The GNU C library runs without
272 problems on kernels that are older than the kernel headers used. The other
273 way round (compiling the GNU C library with old kernel headers and running
274 on a recent kernel) does not necessarily work. For example you can't use
275 new kernel features when using old kernel headers for compiling the GNU C
279 1.8. The compiler hangs while building iconvdata modules. What's
282 {ZW} This is a problem with all current releases of GCC. Initialization of
283 large static arrays is very slow. The compiler will eventually finish; give
286 The problem will be fixed in egcs 1.1 but probably not before then.
289 1.9. When I run `nm -u libc.so' on the produced library I still
290 find unresolved symbols. Can this be ok?
292 {UD} Yes, this is ok. There can be several kinds of unresolved symbols:
294 * magic symbols automatically generated by the linker. These have names
295 like __start_* and __stop_*
297 * symbols starting with _dl_* come from the dynamic linker
299 * weak symbols, which need not be resolved at all (fabs for example)
301 Generally, you should make sure you find a real program which produces
302 errors while linking before deciding there is a problem.
305 1.10. What are these `add-ons'?
307 {UD} To avoid complications with export rules or external source code some
308 optional parts of the libc are distributed as separate packages (e.g., the
309 crypt package, see question 2.5).
311 To use these packages as part of GNU libc, just unpack the tarfiles in the
312 libc source directory and tell the configuration script about them using the
313 --enable-add-ons option. If you give just --enable-add-ons configure tries
314 to find all the add-on packages in your source tree. This may not work. If
315 it doesn't, or if you want to select only a subset of the add-ons, give a
316 comma-separated list of the add-ons to enable:
318 configure --enable-add-ons=crypt,linuxthreads
322 Add-ons can add features (including entirely new shared libraries), override
323 files, provide support for additional architectures, and just about anything
324 else. The existing makefiles do most of the work; only some few stub rules
325 must be written to get everything running.
328 1.11. My XXX kernel emulates a floating-point coprocessor for me.
329 Should I enable --with-fp?
331 {ZW} An emulated FPU is just as good as a real one, as far as the C library
332 is concerned. You only need to say --without-fp if your machine has no way
333 to execute floating-point instructions.
335 People who are interested in squeezing the last drop of performance
336 out of their machine may wish to avoid the trap overhead, but this is
337 far more trouble than it's worth: you then have to compile
338 *everything* this way, including the compiler's internal libraries
339 (libgcc.a for GNU C), because the calling conventions change.
342 1.12. When compiling GNU libc I get lots of errors saying functions
343 in glibc are duplicated in libgcc.
345 {EY} This is *exactly* the same problem that I was having. The problem was
346 due to the fact that configure didn't correctly detect that the linker flag
347 --no-whole-archive was supported in my linker. In my case it was because I
348 had run ./configure with bogus CFLAGS, and the test failed.
350 One thing that is particularly annoying about this problem is that once this
351 is misdetected, running configure again won't fix it unless you first delete
354 {UD} Starting with glibc-2.0.3 there should be a better test to avoid some
355 problems of this kind. The setting of CFLAGS is checked at the very
356 beginning and if it is not usable `configure' will bark.
359 1.13. Why do I get messages about missing thread functions when I use
360 librt? I don't even use threads.
362 {UD} In this case you probably mixed up your installation. librt uses
363 threads internally and has implicit references to the thread library.
364 Normally these references are satisfied automatically but if the thread
365 library is not in the expected place you must tell the linker where it is.
366 When using GNU ld it works like this:
368 gcc -o foo foo.c -Wl,-rpath-link=/some/other/dir -lrt
370 The `/some/other/dir' should contain the thread library. `ld' will use the
371 given path to find the implicitly referenced library while not disturbing
375 1.14. What's the problem with configure --enable-omitfp?
377 {AJ} When --enable-omitfp is set the libraries are built without frame
378 pointers. Some compilers produce buggy code for this model and therefore we
379 don't advise using it at the moment.
381 If you use --enable-omitfp, you're on your own. If you encounter problems
382 with a library that was build this way, we advise you to rebuild the library
383 without --enable-omitfp. If the problem vanishes consider tracking the
384 problem down and report it as compiler failure.
386 Since a library build with --enable-omitfp is undebuggable on most systems,
387 debuggable libraries are also built - you can use it by appending "_g" to
390 The compilation of these extra libraries and the compiler optimizations slow
391 down the build process and need more disk space.
394 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
396 2. Installation and configuration issues
398 2.1. Can I replace the libc on my Linux system with GNU libc?
400 {UD} You cannot replace any existing libc for Linux with GNU libc. It is
401 binary incompatible and therefore has a different major version. You can,
402 however, install it alongside your existing libc.
404 For Linux there are three major libc versions:
406 libc-5 original ELF libc
409 You can have any combination of these three installed. For more information
410 consult documentation for shared library handling. The Makefiles of GNU
411 libc will automatically generate the needed symbolic links which the linker
415 2.2. How do I configure GNU libc so that the essential libraries
416 like libc.so go into /lib and the other into /usr/lib?
418 {UD,AJ} Like all other GNU packages GNU libc is designed to use a base
419 directory and install all files relative to this. The default is
420 /usr/local, because this is safe (it will not damage the system if installed
421 there). If you wish to install GNU libc as the primary C library on your
422 system, set the base directory to /usr (i.e. run configure --prefix=/usr
423 <other_options>). Note that this can damage your system; see question 2.3 for
426 Some systems like Linux have a filesystem standard which makes a difference
427 between essential libraries and others. Essential libraries are placed in
428 /lib because this directory is required to be located on the same disk
429 partition as /. The /usr subtree might be found on another
430 partition/disk. If you configure for Linux with --prefix=/usr, then this
431 will be done automatically.
433 To install the essential libraries which come with GNU libc in /lib on
434 systems other than Linux one must explicitly request it. Autoconf has no
435 option for this so you have to use a `configparms' file (see the `INSTALL'
436 file for details). It should contain:
441 The first line specifies the directory for the essential libraries, the
442 second line the directory for system configuration files.
445 2.3. How should I avoid damaging my system when I install GNU libc?
447 {ZW} If you wish to be cautious, do not configure with --prefix=/usr. If
448 you don't specify a prefix, glibc will be installed in /usr/local, where it
449 will probably not break anything. (If you wish to be certain, set the
450 prefix to something like /usr/local/glibc2 which is not used for anything.)
452 The dangers when installing glibc in /usr are twofold:
454 * glibc will overwrite the headers in /usr/include. Other C libraries
455 install a different but overlapping set of headers there, so the
456 effect will probably be that you can't compile anything. You need to
457 rename /usr/include out of the way first. (Do not throw it away; you
458 will then lose the ability to compile programs against your old libc.)
460 * None of your old libraries, static or shared, can be used with a
461 different C library major version. For shared libraries this is not a
462 problem, because the filenames are different and the dynamic linker
463 will enforce the restriction. But static libraries have no version
464 information. You have to evacuate all the static libraries in
465 /usr/lib to a safe location.
467 The situation is rather similar to the move from a.out to ELF which
468 long-time Linux users will remember.
471 2.4. Do I need to use GNU CC to compile programs that will use the
474 {ZW} In theory, no; the linker does not care, and the headers are supposed
475 to check for GNU CC before using its extensions to the C language.
477 However, there are currently no ports of glibc to systems where another
478 compiler is the default, so no one has tested the headers extensively
479 against another compiler. You may therefore encounter difficulties. If you
480 do, please report them as bugs.
482 Also, in several places GNU extensions provide large benefits in code
483 quality. For example, the library has hand-optimized, inline assembly
484 versions of some string functions. These can only be used with GCC. See
485 question 3.8 for details.
488 2.5. When linking with the new libc I get unresolved symbols
489 `crypt' and `setkey'. Why aren't these functions in the
492 {UD} The US places restrictions on exporting cryptographic programs and
493 source code. Until this law gets abolished we cannot ship the cryptographic
494 functions together with glibc.
496 The functions are available, as an add-on (see question 1.10). People in the US
497 may get it from the same place they got GNU libc from. People outside the
498 US should get the code from ftp://ftp.ifi.uio.no/pub/gnu, or another archive
499 site outside the USA. The README explains how to install the sources.
501 If you already have the crypt code on your system the reason for the failure
502 is probably that you did not link with -lcrypt. The crypto functions are in
503 a separate library to make it possible to export GNU libc binaries from the
507 2.6. When I use GNU libc on my Linux system by linking against
508 the libc.so which comes with glibc all I get is a core dump.
510 {UD} On Linux, gcc sets the dynamic linker to /lib/ld-linux.so.1 unless the
511 user specifies a -dynamic-linker argument. This is the name of the libc5
512 dynamic linker, which does not work with glibc.
514 For casual use of GNU libc you can just specify
515 -dynamic-linker=/lib/ld-linux.so.2
517 which is the glibc dynamic linker, on Linux systems. On other systems the
518 name is /lib/ld.so.1.
520 To change your environment to use GNU libc for compiling you need to change
521 the `specs' file of your gcc. This file is normally found at
523 /usr/lib/gcc-lib/<arch>/<version>/specs
525 In this file you have to change a few things:
527 - change `ld-linux.so.1' to `ld-linux.so.2'
529 - remove all expression `%{...:-lgmon}'; there is no libgmon in glibc
531 - fix a minor bug by changing %{pipe:-} to %|
533 Here is what the gcc-2.7.2 specs file should look like when GNU libc is
536 -----------------------------------------------------------------------
538 %{V} %{v:%{!V:-V}} %{Qy:} %{!Qn:-Qy} %{n} %{T} %{Ym,*} %{Yd,*} %{Wa,*:%*}
544 %{fPIC:-D__PIC__ -D__pic__} %{fpic:-D__PIC__ -D__pic__} %{!m386:-D__i486__} %{posix:-D_POSIX_SOURCE} %{pthread:-D_REENTRANT}
553 %{!shared:crtend.o%s} %{shared:crtendS.o%s} crtn.o%s
556 -m elf_i386 %{shared:-shared} %{!shared: %{!ibcs: %{!static: %{rdynamic:-export-dynamic} %{!dynamic-linker:-dynamic-linker /lib/ld-linux.so.2}} %{static:-static}}}
559 %{!shared: %{pthread:-lpthread} %{profile:-lc_p} %{!profile: -lc}}
565 %{!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}
567 *switches_need_spaces:
571 %{funsigned-char:-D__CHAR_UNSIGNED__}
574 -D__ELF__ -Dunix -Di386 -Dlinux -Asystem(unix) -Asystem(posix) -Acpu(i386) -Amachine(i386)
582 -----------------------------------------------------------------------
584 Things get a bit more complicated if you have GNU libc installed in some
585 other place than /usr, i.e., if you do not want to use it instead of the old
586 libc. In this case the needed startup files and libraries are not found in
587 the regular places. So the specs file must tell the compiler and linker
590 Version 2.7.2.3 does and future versions of GCC will automatically
591 provide the correct specs.
594 2.7. Looking through the shared libc file I haven't found the
595 functions `stat', `lstat', `fstat', and `mknod' and while
596 linking on my Linux system I get error messages. How is
597 this supposed to work?
599 {RM} Believe it or not, stat and lstat (and fstat, and mknod) are supposed
600 to be undefined references in libc.so.6! Your problem is probably a missing
601 or incorrect /usr/lib/libc.so file; note that this is a small text file now,
602 not a symlink to libc.so.6. It should look something like this:
604 GROUP ( libc.so.6 libc_nonshared.a )
607 2.8. How can I compile gcc 2.7.2.1 from the gcc source code using
610 {AJ} There's only correct support for glibc 2.0.x in gcc 2.7.2.3 or later.
611 But you should get at least gcc 2.8.1 or egcs 1.0.2 (or later versions)
615 2.9. The `gencat' utility cannot process the catalog sources which
616 were used on my Linux libc5 based system. Why?
618 {UD} The `gencat' utility provided with glibc complies to the XPG standard.
619 The older Linux version did not obey the standard, so they are not
622 To ease the transition from the Linux version some of the non-standard
623 features are also present in the `gencat' program of GNU libc. This mainly
624 includes the use of symbols for the message number and the automatic
625 generation of header files which contain the needed #defines to map the
628 Here is a simple SED script to convert at least some Linux specific catalog
629 files to the XPG4 form:
631 -----------------------------------------------------------------------
632 # Change catalog source in Linux specific format to standard XPG format.
633 # Ulrich Drepper <drepper@cygnus.com>, 1996.
637 s/\$ #\([^ ]*\).*/\1/
639 s/\$ #[^ ]* *\(.*\)/\$ \1/
645 s/\(.*\)\n\(.*\)/\2 \1/
647 -----------------------------------------------------------------------
650 2.10. Programs using libc have their messages translated, but other
651 behavior is not localized (e.g. collating order); why?
653 {ZW} Translated messages are automatically installed, but the locale
654 database that controls other behaviors is not. You need to run localedef to
655 install this database, after you have run `make install'. For example, to
656 set up the French Canadian locale, simply issue the command
658 localedef -i fr_CA -f ISO-8859-1 fr_CA
660 Please see localedata/README in the source tree for further details.
663 2.11. I have set up /etc/nis.conf, and the Linux libc 5 with NYS
664 works great. But the glibc NIS+ doesn't seem to work.
666 {TK} The glibc NIS+ implementation uses a /var/nis/NIS_COLD_START file for
667 storing information about the NIS+ server and their public keys, because the
668 nis.conf file does not contain all the necessary information. You have to
669 copy a NIS_COLD_START file from a Solaris client (the NIS_COLD_START file is
670 byte order independent) or generate it with nisinit from the nis-tools
671 package; available at
673 http://www-vt.uni-paderborn.de/~kukuk/linux/nisplus.html
676 2.12. I have killed ypbind to stop using NIS, but glibc
679 {TK} For faster NIS lookups, glibc uses the /var/yp/binding/ files from
680 ypbind. ypbind 3.3 and older versions don't always remove these files, so
681 glibc will continue to use them. Other BSD versions seem to work correctly.
682 Until ypbind 3.4 is released, you can find a patch at
684 ftp://ftp.kernel.org/pub/linux/utils/net/NIS/ypbind-3.3-glibc3.diff.
687 2.13. Under Linux/Alpha, I always get "do_ypcall: clnt_call:
688 RPC: Unable to receive; errno = Connection refused" when using NIS.
690 {TK} You need a ypbind version which is 64bit clean. Some versions are not
691 64bit clean. A 64bit clean implementation is ypbind-mt. For ypbind 3.3,
692 you need the patch from ftp.kernel.org (See the previous question). I don't
693 know about other versions.
696 2.14. After installing glibc name resolving doesn't work properly.
698 {AJ} You probably should read the manual section describing nsswitch.conf
699 (just type `info libc "NSS Configuration File"'). The NSS configuration
700 file is usually the culprit.
703 2.15. I have /usr/include/net and /usr/include/scsi as symlinks
704 into my Linux source tree. Is that wrong?
706 {PB} This was necessary for libc5, but is not correct when using glibc.
707 Including the kernel header files directly in user programs usually does not
708 work (see question 3.5). glibc provides its own <net/*> and <scsi/*> header
709 files to replace them, and you may have to remove any symlink that you have
710 in place before you install glibc. However, /usr/include/asm and
711 /usr/include/linux should remain as they were.
714 2.16. Programs like `logname', `top', `uptime' `users', `w' and
715 `who', show incorrect information about the (number of)
716 users on my system. Why?
718 {MK} See question 3.2.
721 2.17. After upgrading to glibc 2.1 with symbol versioning I get
722 errors about undefined symbols. What went wrong?
724 {AJ} The problem is caused either by wrong program code or tools. In the
725 versioned libc a lot of symbols are now local that were global symbols in
726 previous versions. It seems that programs linked against older versions
727 often accidentally used libc global variables -- something that should not
730 The only way to fix this is to recompile your program. Sorry, that's the
731 price you might have to pay once for quite a number of advantages with
735 2.18. When I start the program XXX after upgrading the library
737 XXX: Symbol `_sys_errlist' has different size in shared
738 object, consider re-linking
739 Why? What should I do?
741 {UD} As the message says, relink the binary. The problem is that a few
742 symbols from the library can change in size and there is no way to avoid
743 this. _sys_errlist is a good example. Occasionally there are new error
744 numbers added to the kernel and this must be reflected at user level,
745 breaking programs that refer to them directly.
747 Such symbols should normally not be used at all. There are mechanisms to
748 avoid using them. In the case of _sys_errlist, there is the strerror()
749 function which should _always_ be used instead. So the correct fix is to
750 rewrite that part of the application.
752 In some situations (especially when testing a new library release) it might
753 be possible that a symbol changed size when that should not have happened.
754 So in case of doubt report such a warning message as a problem.
757 2.19. What do I need for C++ development?
759 {HJ,AJ} You need either egcs 1.0.2 or gcc-2.8.1 with libstdc++ 2.8.1 (or
760 more recent versions). libg++ 2.7.2 (and the Linux Versions 2.7.2.x) doesn't
761 work very well with the GNU C library due to vtable thunks. If you're
762 upgrading from glibc 2.0.x to 2.1 you have to recompile libstdc++ since the
763 library compiled for 2.0 is not compatible due to the new Large File Support
764 (LFS) in version 2.1.
767 2.20. Even statically linked programs need some shared libraries
768 which is not acceptable for me. What can I do?
770 {AJ} NSS (for details just type `info libc "Name Service Switch"') won't
771 work properly without shared libraries. NSS allows using different services
772 (e.g. NIS, files, db, hesiod) by just changing one configuration file
773 (/etc/nsswitch.conf) without relinking any programs. The only disadvantage
774 is that now static libraries need to access shared libraries. This is
775 handled transparently by the GNU C library.
777 A solution is to configure glibc with --enable-static-nss. In this case you
778 can create a static binary that will use only the services dns and files
779 (change /etc/nsswitch.conf for this). You need to link explicitly against
780 all these services. For example:
782 gcc -static test-netdb.c -o test-netdb.c \
783 -lc -lnss_files -lnss_dns -lresolv
785 The problem with this approach is that you've got to link every static
786 program that uses NSS routines with all those libraries.
788 {UD} In fact, one cannot say anymore that a libc compiled with this
789 option is using NSS. There is no switch anymore. Therefore it is
790 *highly* recommended *not* to use --enable-static-nss since this makes
791 the behaviour of the programs on the system inconsistent.
794 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
796 3. Source and binary incompatibilities, and what to do about them
798 3.1. I expect GNU libc to be 100% source code compatible with
799 the old Linux based GNU libc. Why isn't it like this?
801 {DMT,UD} Not every extension in Linux libc's history was well thought-out.
802 In fact it had a lot of problems with standards compliance and with
803 cleanliness. With the introduction of a new version number these errors can
804 now be corrected. Here is a list of the known source code
807 * _GNU_SOURCE: glibc does not make the GNU extensions available
808 automatically. If a program depends on GNU extensions or some
809 other non-standard functionality, it is necessary to compile it
810 with the C compiler option -D_GNU_SOURCE, or better, to put
811 `#define _GNU_SOURCE' at the beginning of your source files, before
812 any C library header files are included. This difference normally
813 manifests itself in the form of missing prototypes and/or data type
814 definitions. Thus, if you get such errors, the first thing you
815 should do is try defining _GNU_SOURCE and see if that makes the
818 For more information consult the file `NOTES' in the GNU C library
821 * reboot(): GNU libc sanitizes the interface of reboot() to be more
822 compatible with the interface used on other OSes. reboot() as
823 implemented in glibc takes just one argument. This argument
824 corresponds to the third argument of the Linux reboot system call.
825 That is, a call of the form reboot(a, b, c) needs to be changed into
826 reboot(c). Beside this the header <sys/reboot.h> defines the needed
827 constants for the argument. These RB_* constants should be used
828 instead of the cryptic magic numbers.
830 * swapon(): the interface of this function didn't change, but the
831 prototype is in a separate header file <sys/swap.h>. This header
832 file also provides the SWAP_* constants defined by <linux/swap.h>;
833 you should use them for the second argument to swapon().
835 * errno: If a program uses the variable "errno", then it _must_
836 include <errno.h>. The old libc often (erroneously) declared this
837 variable implicitly as a side-effect of including other libc header
838 files. glibc is careful to avoid such namespace pollution, which,
839 in turn, means that you really need to include the header files that
840 you depend on. This difference normally manifests itself in the
841 form of the compiler complaining about references to an undeclared
844 * Linux-specific syscalls: All Linux system calls now have appropriate
845 library wrappers and corresponding declarations in various header files.
846 This is because the syscall() macro that was traditionally used to
847 work around missing syscall wrappers are inherently non-portable and
848 error-prone. The following table lists all the new syscall stubs,
849 the header-file declaring their interface and the system call name.
851 syscall name: wrapper name: declaring header file:
852 ------------- ------------- ----------------------
853 bdflush bdflush <sys/kdaemon.h>
854 syslog ksyslog_ctl <sys/klog.h>
856 * lpd: Older versions of lpd depend on a routine called _validuser().
857 The library does not provide this function, but instead provides
858 __ivaliduser() which has a slightly different interface. Simply
859 upgrading to a newer lpd should fix this problem (e.g., the 4.4BSD
860 lpd is known to be working).
862 * resolver functions/BIND: like on many other systems the functions of
863 the resolver library are not included in libc itself. There is a
864 separate library libresolv. If you get undefined symbol errors for
865 symbols starting with `res_*' simply add -lresolv to your linker
868 * the `signal' function's behavior corresponds to the BSD semantic and
869 not the SysV semantic as it was in libc-5. The interface on all GNU
870 systems shall be the same and BSD is the semantic of choice. To use
871 the SysV behavior simply use `sysv_signal', or define _XOPEN_SOURCE.
872 See question 3.7 for details.
875 3.2. Why does getlogin() always return NULL on my Linux box?
877 {UD} The GNU C library has a format for the UTMP and WTMP file which differs
878 from what your system currently has. It was extended to fulfill the needs
879 of the next years when IPv6 is introduced. The record size is different and
880 some fields have different positions. The files written by functions from
881 the one library cannot be read by functions from the other library. Sorry,
882 but this is what a major release is for. It's better to have a cut now than
883 having no means to support the new techniques later.
885 {MK} There is however a (partial) solution for this problem. Please take a
886 look at the file `login/README.utmpd'.
889 3.3. Where are the DST_* constants found in <sys/time.h> on many
892 {UD} These constants come from the old BSD days and are not used anymore
893 (libc5 does not actually implement the handling although the constants are
896 Instead GNU libc contains zone database support and compatibility code for
897 POSIX TZ environment variable handling.
900 3.4. The prototypes for `connect', `accept', `getsockopt',
901 `setsockopt', `getsockname', `getpeername', `send',
902 `sendto', and `recvfrom' are different in GNU libc from
903 any other system I saw. This is a bug, isn't it?
905 {UD} No, this is no bug. This version of GNU libc already follows the new
906 Single Unix specifications (and I think the POSIX.1g draft which adopted the
907 solution). The type for a parameter describing a size is now `socklen_t', a
911 3.5. On Linux I've got problems with the declarations in Linux
914 {UD,AJ} On Linux, the use of kernel headers is reduced to the minimum. This
915 gives Linus the ability to change the headers more freely. Also, user
916 programs are now insulated from changes in the size of kernel data
919 For example, the sigset_t type is 32 or 64 bits wide in the kernel. In
920 glibc it is 1024 bits wide. This guarantees that when the kernel gets a
921 bigger sigset_t (for POSIX.1e realtime support, say) user programs will not
922 have to be recompiled. Consult the header files for more information about
925 Therefore you shouldn't include Linux kernel header files directly if glibc
926 has defined a replacement. Otherwise you might get undefined results because
930 3.6. I don't include any kernel headers myself but the compiler
931 still complains about redeclarations of types in the kernel
934 {UD} The kernel headers before Linux 2.1.61 and 2.0.32 don't work correctly
935 with glibc. Compiling C programs is possible in most cases but C++ programs
936 have (due to the change of the name lookups for `struct's) problems. One
937 prominent example is `struct fd_set'.
939 There might be some problems left but 2.1.61/2.0.32 fix most of the known
940 ones. See the BUGS file for other known problems.
943 3.7. Why don't signals interrupt system calls anymore?
945 {ZW} By default GNU libc uses the BSD semantics for signal(), unlike Linux
946 libc 5 which used System V semantics. This is partially for compatibility
947 with other systems and partially because the BSD semantics tend to make
948 programming with signals easier.
950 There are three differences:
952 * BSD-style signals that occur in the middle of a system call do not
953 affect the system call; System V signals cause the system call to
954 fail and set errno to EINTR.
956 * BSD signal handlers remain installed once triggered. System V signal
957 handlers work only once, so one must reinstall them each time.
959 * A BSD signal is blocked during the execution of its handler. In other
960 words, a handler for SIGCHLD (for example) does not need to worry about
961 being interrupted by another SIGCHLD. It may, however, be interrupted
964 There is general consensus that for `casual' programming with signals, the
965 BSD semantics are preferable. You don't need to worry about system calls
966 returning EINTR, and you don't need to worry about the race conditions
967 associated with one-shot signal handlers.
969 If you are porting an old program that relies on the old semantics, you can
970 quickly fix the problem by changing signal() to sysv_signal() throughout.
971 Alternatively, define _XOPEN_SOURCE before including <signal.h>.
973 For new programs, the sigaction() function allows you to specify precisely
974 how you want your signals to behave. All three differences listed above are
975 individually switchable on a per-signal basis with this function.
977 If all you want is for one specific signal to cause system calls to fail and
978 return EINTR (for example, to implement a timeout) you can do this with
982 3.8. I've got errors compiling code that uses certain string
985 {AJ} glibc 2.1 has special string functions that are faster than the normal
986 library functions. Some of the functions are implemented as inline functions
987 and others as macros.
989 The optimized string functions are only used when compiling with
990 optimizations (-O1 or higher). The behavior can be changed with two feature
993 * __NO_STRING_INLINES: Don't do any string optimizations.
994 * __USE_STRING_INLINES: Use assembly language inline functions (might
995 increase code size dramatically).
997 Since some of these string functions are now additionally defined as macros,
998 code like "char *strncpy();" doesn't work anymore (and is unnecessary, since
999 <string.h> has the necessary declarations). Either change your code or
1000 define __NO_STRING_INLINES.
1002 {UD} Another problem in this area is that gcc still has problems on machines
1003 with very few registers (e.g., ix86). The inline assembler code can require
1004 almost all the registers and the register allocator cannot always handle
1007 One can disable the string optimizations selectively. Instead of writing
1009 cp = strcpy (foo, "lkj");
1013 cp = (strcpy) (foo, "lkj");
1015 This disables the optimization for that specific call.
1018 3.9. I get compiler messages "Initializer element not constant" with
1019 stdin/stdout/stderr. Why?
1021 {RM,AJ} Constructs like:
1022 static FILE *InPtr = stdin;
1024 lead to this message. This is correct behaviour with glibc since stdin is
1025 not a constant expression. Please note that a strict reading of ISO C does
1026 not allow above constructs.
1028 One of the advantages of this is that you can assign to stdin, stdout, and
1029 stderr just like any other global variable (e.g. `stdout = my_stream;'),
1030 which can be very useful with custom streams that you can write with libio
1031 (but beware this is not necessarily portable). The reason to implement it
1032 this way were versioning problems with the size of the FILE structure.
1035 3.10. I can't compile with gcc -traditional (or
1036 -traditional-cpp). Why?
1038 {AJ} glibc2 does break -traditional and -traditonal-cpp - and will continue
1039 to do so. For example constructs of the form:
1045 are useful for debugging purposes (you can use foo with your debugger that's
1046 why we need the enum) and for compatibility (other systems use defines and
1050 3.11. I get some errors with `gcc -ansi'. Isn't glibc ANSI compatible?
1052 {AJ} The GNU C library is compatible with the ANSI/ISO C standard. If
1053 you're using `gcc -ansi', the glibc includes which are specified in the
1054 standard follow the standard. The ANSI/ISO C standard defines what has to be
1055 in the include files - and also states that nothing else should be in the
1056 include files (btw. you can still enable additional standards with feature
1059 The GNU C library is conforming to ANSI/ISO C - if and only if you're only
1060 using the headers and library functions defined in the standard.
1063 3.12. I can't access some functions anymore. nm shows that they do
1064 exist but linking fails nevertheless.
1066 {AJ} With the introduction of versioning in glibc 2.1 it is possible to
1067 export only those identifiers (functions, variables) that are really needed
1068 by application programs and by other parts of glibc. This way a lot of
1069 internal interfaces are now hidden. nm will still show those identifiers
1070 but marking them as internal. ISO C states that identifiers beginning with
1071 an underscore are internal to the libc. An application program normally
1072 shouldn't use those internal interfaces (there are exceptions,
1073 e.g. __ivaliduser). If a program uses these interfaces, it's broken. These
1074 internal interfaces might change between glibc releases or dropped
1078 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1082 4.1. After I changed configure.in I get `Autoconf version X.Y.
1083 or higher is required for this script'. What can I do?
1085 {UD} You have to get the specified autoconf version (or a later one)
1086 from your favorite mirror of ftp.gnu.org.
1089 4.2. When I try to compile code which uses IPv6 headers and
1090 definitions on my Linux 2.x.y system I am in trouble.
1091 Nothing seems to work.
1093 {UD} The problem is that IPv6 development still has not reached a point
1094 where the headers are stable. There are still lots of incompatible changes
1095 made and the libc headers have to follow.
1097 Also, make sure you have a suitably recent kernel. As of the 970401
1098 snapshot, according to Philip Blundell <Philip.Blundell@pobox.com>, the
1099 required kernel version is at least 2.1.30.
1102 4.3. When I set the timezone by setting the TZ environment variable
1103 to EST5EDT things go wrong since glibc computes the wrong time
1104 from this information.
1106 {UD} The problem is that people still use the braindamaged POSIX method to
1107 select the timezone using the TZ environment variable with a format EST5EDT
1108 or whatever. People, read the POSIX standard, the implemented behaviour is
1109 correct! What you see is in fact the result of the decisions made while
1110 POSIX.1 was created. We've only implemented the handling of TZ this way to
1111 be POSIX compliant. It is not really meant to be used.
1113 The alternative approach to handle timezones which is implemented is the
1114 correct one to use: use the timezone database. This avoids all the problems
1115 the POSIX method has plus it is much easier to use. Simply run the tzselect
1116 shell script, answer the question and use the name printed in the end by
1117 making a symlink to /usr/share/zoneinfo/NAME (NAME is the returned value
1118 from tzselect) from the file /etc/localtime. That's all. You never again
1121 So, please avoid sending bug reports about time related problems if you use
1122 the POSIX method and you have not verified something is really broken by
1123 reading the POSIX standards.
1126 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
1128 Answers were given by:
1129 {UD} Ulrich Drepper, <drepper@cygnus.com>
1130 {DMT} David Mosberger-Tang, <davidm@AZStarNet.com>
1131 {RM} Roland McGrath, <roland@gnu.org>
1132 {AJ} Andreas Jaeger, <aj@arthur.rhein-neckar.de>
1133 {EY} Eric Youngdale, <eric@andante.jic.com>
1134 {PB} Phil Blundell, <Philip.Blundell@pobox.com>
1135 {MK} Mark Kettenis, <kettenis@phys.uva.nl>
1136 {ZW} Zack Weinberg, <zack@rabi.phys.columbia.edu>
1137 {TK} Thorsten Kukuk, <kukuk@vt.uni-paderborn.de>
1138 {GK} Geoffrey Keating, <geoffk@ozemail.com.au>
1139 {HJ} H.J. Lu, <hjl@gnu.org>
1143 outline-regexp:"\\?"