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 assembler?
26 1.5. Which compiler should I use for powerpc?
27 1.6. Which tools should I use for ARM?
28 1.7. Do I need some more things to compile the GNU C Library?
29 1.8. What version of the Linux kernel headers should be used?
30 1.9. The compiler hangs while building iconvdata modules. What's
32 1.10. When I run `nm -u libc.so' on the produced library I still
33 find unresolved symbols. Can this be ok?
34 1.11. What are these `add-ons'?
35 1.12. My XXX kernel emulates a floating-point coprocessor for me.
36 Should I enable --with-fp?
37 1.13. When compiling GNU libc I get lots of errors saying functions
38 in glibc are duplicated in libgcc.
39 1.14. Why do I get messages about missing thread functions when I use
40 librt? I don't even use threads.
41 1.15. What's the problem with configure --enable-omitfp?
42 1.16. I get failures during `make check'. What should I do?
43 1.17. What is symbol versioning good for? Do I need it?
45 2. Installation and configuration issues
47 2.1. Can I replace the libc on my Linux system with GNU libc?
48 2.2. How do I configure GNU libc so that the essential libraries
49 like libc.so go into /lib and the other into /usr/lib?
50 2.3. How should I avoid damaging my system when I install GNU libc?
51 2.4. Do I need to use GNU CC to compile programs that will use the
53 2.5. When linking with the new libc I get unresolved symbols
54 `crypt' and `setkey'. Why aren't these functions in the
56 2.6. When I use GNU libc on my Linux system by linking against
57 the libc.so which comes with glibc all I get is a core dump.
58 2.7. Looking through the shared libc file I haven't found the
59 functions `stat', `lstat', `fstat', and `mknod' and while
60 linking on my Linux system I get error messages. How is
61 this supposed to work?
62 2.8. When I run an executable on one system which I compiled on
63 another, I get dynamic linker errors. Both systems have the same
64 version of glibc installed. What's wrong?
65 2.9. How can I compile gcc 2.7.2.1 from the gcc source code using
67 2.10. The `gencat' utility cannot process the catalog sources which
68 were used on my Linux libc5 based system. Why?
69 2.11. Programs using libc have their messages translated, but other
70 behavior is not localized (e.g. collating order); why?
71 2.12. I have set up /etc/nis.conf, and the Linux libc 5 with NYS
72 works great. But the glibc NIS+ doesn't seem to work.
73 2.13. I have killed ypbind to stop using NIS, but glibc
75 2.14. Under Linux/Alpha, I always get "do_ypcall: clnt_call:
76 RPC: Unable to receive; errno = Connection refused" when using NIS.
77 2.15. After installing glibc name resolving doesn't work properly.
78 2.16. How do I create the databases for NSS?
79 2.17. I have /usr/include/net and /usr/include/scsi as symlinks
80 into my Linux source tree. Is that wrong?
81 2.18. Programs like `logname', `top', `uptime' `users', `w' and
82 `who', show incorrect information about the (number of)
83 users on my system. Why?
84 2.19. After upgrading to glibc 2.1 with symbol versioning I get
85 errors about undefined symbols. What went wrong?
86 2.20. When I start the program XXX after upgrading the library
88 XXX: Symbol `_sys_errlist' has different size in shared
89 object, consider re-linking
90 Why? What should I do?
91 2.21. What do I need for C++ development?
92 2.22. Even statically linked programs need some shared libraries
93 which is not acceptable for me. What can I do?
94 2.23. I just upgraded my Linux system to glibc and now I get
95 errors whenever I try to link any program.
96 2.24. When I use nscd the machine freezes.
97 2.25. I need lots of open files. What do I have to do?
98 2.26. How do I get the same behavior on parsing /etc/passwd and
99 /etc/group as I have with libc5 ?
100 2.27. What needs to be recompiled when upgrading from glibc 2.0 to glibc
102 2.28. Why is extracting files via tar so slow?
103 2.29. Compiling programs I get parse errors in libio.h (e.g. "parse error
104 before `_IO_seekoff'"). How should I fix this?
106 3. Source and binary incompatibilities, and what to do about them
108 3.1. I expect GNU libc to be 100% source code compatible with
109 the old Linux based GNU libc. Why isn't it like this?
110 3.2. Why does getlogin() always return NULL on my Linux box?
111 3.3. Where are the DST_* constants found in <sys/time.h> on many
113 3.4. The prototypes for `connect', `accept', `getsockopt',
114 `setsockopt', `getsockname', `getpeername', `send',
115 `sendto', and `recvfrom' are different in GNU libc from
116 any other system I saw. This is a bug, isn't it?
117 3.5. On Linux I've got problems with the declarations in Linux
119 3.6. I don't include any kernel headers myself but the compiler
120 still complains about redeclarations of types in the kernel
122 3.7. Why don't signals interrupt system calls anymore?
123 3.8. I've got errors compiling code that uses certain string
125 3.9. I get compiler messages "Initializer element not constant" with
126 stdin/stdout/stderr. Why?
127 3.10. I can't compile with gcc -traditional (or
128 -traditional-cpp). Why?
129 3.11. I get some errors with `gcc -ansi'. Isn't glibc ANSI compatible?
130 3.12. I can't access some functions anymore. nm shows that they do
131 exist but linking fails nevertheless.
132 3.13. When using the db-2 library which comes with glibc is used in
133 the Perl db modules the testsuite is not passed. This did not
134 happen with db-1, gdbm, or ndbm.
135 3.14. The pow() inline function I get when including <math.h> is broken.
136 I get segmentation faults when I run the program.
137 3.15. The sys/sem.h file lacks the definition of `union semun'.
138 3.16. Why has <netinet/ip_fw.h> disappeared?
139 3.17. I get floods of warnings when I use -Wconversion and include
140 <string.h> or <math.h>.
141 3.18. After upgrading to glibc 2.1, I receive errors about
142 unresolved symbols, like `_dl_initial_searchlist' and can not
143 execute any binaries. What went wrong?
144 3.19. bonnie reports that char i/o with glibc 2 is much slower than with
145 libc5. What can be done?
149 4.1. After I changed configure.in I get `Autoconf version X.Y.
150 or higher is required for this script'. What can I do?
151 4.2. When I try to compile code which uses IPv6 headers and
152 definitions on my Linux 2.x.y system I am in trouble.
153 Nothing seems to work.
154 4.3. When I set the timezone by setting the TZ environment variable
155 to EST5EDT things go wrong since glibc computes the wrong time
156 from this information.
157 4.4. What other sources of documentation about glibc are available?
158 4.5. The timezone string for Sydney/Australia is wrong since even when
159 daylight saving time is in effect the timezone string is EST.
160 4.6. I've build make 3.77 against glibc 2.1 and now make gets
164 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
168 1.1. What systems does the GNU C Library run on?
170 {UD} This is difficult to answer. The file `README' lists the architectures
171 GNU libc was known to run on *at some time*. This does not mean that it
172 still can be compiled and run on them now.
174 The systems glibc is known to work on as of this release, and most probably
178 i[3456]86-*-linux-gnu Linux-2.x on Intel
179 m68k-*-linux-gnu Linux-2.x on Motorola 680x0
180 alpha-*-linux-gnu Linux-2.x on DEC Alpha
181 powerpc-*-linux-gnu Linux and MkLinux on PowerPC systems
182 sparc-*-linux-gnu Linux-2.x on SPARC
183 sparc64-*-linux-gnu Linux-2.x on UltraSPARC
184 arm-*-none ARM standalone systems
185 arm-*-linux Linux-2.x on ARM
186 arm-*-linuxaout Linux-2.x on ARM using a.out binaries
188 Ports to other Linux platforms are in development, and may in fact work
189 already, but no one has sent us success reports for them. Currently no
190 ports to other operating systems are underway, although a few people have
193 If you have a system not listed above (or in the `README' file) and you are
194 really interested in porting it, contact
199 1.2. What compiler do I need to build GNU libc?
201 {UD} You must use GNU CC to compile GNU libc. A lot of extensions of GNU CC
202 are used to increase portability and speed.
204 GNU CC is found, like all other GNU packages, on
206 ftp://ftp.gnu.org/pub/gnu
208 and the many mirror sites. ftp.gnu.org is always overloaded, so try to find
209 a local mirror first.
211 You should always try to use the latest official release. Older versions
212 may not have all the features GNU libc requires. The current releases of
213 egcs (1.0.3 and 1.1.1) should work with the GNU C library (for powerpc see
214 question 1.5; for ARM see question 1.6).
216 {ZW} Due to problems with C++ exception handling, you must use EGCS (any
217 version) to compile version 2.1 of GNU libc. See question 2.8 for details.
220 1.3. When I try to compile glibc I get only error messages.
223 {UD} You definitely need GNU make to build GNU libc. No other make
224 program has the needed functionality.
226 We recommend version GNU make version 3.75 or 3.77. Versions before 3.75
227 have bugs and/or are missing features. Version 3.76 has bugs which
228 appear when building big projects like GNU libc. 3.76.1 appears to work but
229 some people have reported problems. If you build GNU make 3.77 from source,
230 please read question 4.6 first.
233 1.4. Do I need a special linker or assembler?
235 {ZW} If you want a shared library, you need a linker and assembler that
236 understand all the features of ELF, including weak and versioned symbols.
237 The static library can be compiled with less featureful tools, but lacks key
238 features such as NSS.
240 For Linux or Hurd, you want binutils 2.8.1.0.23, 2.9.1, or 2.9.1.0.15 or
241 higher. These are the only versions we've tested and found reliable. Other
242 versions after 2.8.1.0.23 may work but we don't recommend them, especially
243 not when C++ is involved. Earlier versions do not work at all.
245 Other operating systems may come with system tools that have all the
246 necessary features, but this is moot because glibc hasn't been ported to
250 1.5. Which compiler should I use for powerpc?
252 {GK} You want to use egcs 1.1 or later (together with the right versions
253 of all the other tools, of course).
255 In fact, egcs 1.1 has a bug that causes linuxthreads to be
256 miscompiled, resulting in segmentation faults when using condition
257 variables. There is a temporary patch at:
259 <http://discus.anu.edu.au/~geoffk/egcs-3.diff>
261 Later versions of egcs may fix this problem.
264 1.6. Which tools should I use for ARM?
266 {PB} You should use egcs 1.1 or a later version. For ELF systems some
267 changes are needed to the compiler; a patch against egcs-1.1.x can be found
270 <ftp://ftp.netwinder.org/users/p/philb/egcs-1.1.1pre2-diff-981126>
272 Binutils 2.9.1.0.16 or later is also required.
275 1.7. Do I need some more things to compile the GNU C Library?
277 {UD} Yes, there are some more :-).
279 * GNU gettext. This package contains the tools needed to construct
280 `message catalog' files containing translated versions of system
281 messages. See ftp://ftp.gnu.org/pub/gnu or better any mirror
282 site. (We distribute compiled message catalogs, but they may not be
285 * Some files are built with special tools. E.g., files ending in .gperf
286 need a `gperf' program. The GNU version (now available in a separate
287 package, formerly only as part of libg++) is known to work while some
288 vendor versions do not.
290 You should not need these tools unless you change the source files.
292 * Perl 5 is needed if you wish to test an installation of GNU libc
293 as the primary C library.
295 * When compiling for Linux, the header files of the Linux kernel must
296 be available to the compiler as <linux/*.h> and <asm/*.h>.
298 * lots of disk space (~170MB for i?86-linux; more for RISC platforms,
301 * plenty of time. Compiling just the shared and static libraries for
302 i?86-linux takes approximately 1h on an i586@133, or 2.5h on
303 i486@66, or 4.5h on i486@33. Multiply this by 1.5 or 2.0 if you
304 build profiling and/or the highly optimized version as well. For
305 Hurd systems times are much higher.
307 You should avoid compiling in a NFS mounted filesystem. This is
310 James Troup <J.J.Troup@comp.brad.ac.uk> reports a compile time of
311 45h34m for a full build (shared, static, and profiled) on Atari
312 Falcon (Motorola 68030 @ 16 Mhz, 14 Mb memory) and Jan Barte
313 <yann@plato.uni-paderborn.de> reports 22h48m on Atari TT030
314 (Motorola 68030 @ 32 Mhz, 34 Mb memory)
316 If you have some more measurements let me know.
319 1.8. What version of the Linux kernel headers should be used?
321 {AJ,UD} The headers from the most recent Linux kernel should be used. The
322 headers used while compiling the GNU C library and the kernel binary used
323 when using the library do not need to match. The GNU C library runs without
324 problems on kernels that are older than the kernel headers used. The other
325 way round (compiling the GNU C library with old kernel headers and running
326 on a recent kernel) does not necessarily work. For example you can't use
327 new kernel features if you used old kernel headers to compile the GNU C
330 {ZW} Even if you are using a 2.0 kernel on your machine, we recommend you
331 compile GNU libc with 2.2 kernel headers. That way you won't have to
332 recompile libc if you ever upgrade to kernel 2.2. To tell libc which
333 headers to use, give configure the --with-headers switch
334 (e.g. --with-headers=/usr/src/linux-2.2.0/include).
336 Note that you must configure the 2.2 kernel if you do this, otherwise libc
337 will be unable to find <linux/version.h>. Just change the current directory
338 to the root of the 2.2 tree and do `make include/linux/version.h'.
341 1.9. The compiler hangs while building iconvdata modules. What's
344 {ZW} This is a problem with old versions of GCC. Initialization of large
345 static arrays is very slow. The compiler will eventually finish; give it
348 The problem is fixed in egcs 1.1.
351 1.10. When I run `nm -u libc.so' on the produced library I still
352 find unresolved symbols. Can this be ok?
354 {UD} Yes, this is ok. There can be several kinds of unresolved symbols:
356 * magic symbols automatically generated by the linker. These have names
357 like __start_* and __stop_*
359 * symbols starting with _dl_* come from the dynamic linker
361 * weak symbols, which need not be resolved at all (fabs for example)
363 Generally, you should make sure you find a real program which produces
364 errors while linking before deciding there is a problem.
367 1.11. What are these `add-ons'?
369 {UD} To avoid complications with export rules or external source code some
370 optional parts of the libc are distributed as separate packages (e.g., the
371 crypt package, see question 2.5).
373 To use these packages as part of GNU libc, just unpack the tarfiles in the
374 libc source directory and tell the configuration script about them using the
375 --enable-add-ons option. If you give just --enable-add-ons configure tries
376 to find all the add-on packages in your source tree. This may not work. If
377 it doesn't, or if you want to select only a subset of the add-ons, give a
378 comma-separated list of the add-ons to enable:
380 configure --enable-add-ons=crypt,linuxthreads
384 Add-ons can add features (including entirely new shared libraries), override
385 files, provide support for additional architectures, and just about anything
386 else. The existing makefiles do most of the work; only some few stub rules
387 must be written to get everything running.
390 1.12. My XXX kernel emulates a floating-point coprocessor for me.
391 Should I enable --with-fp?
393 {ZW} An emulated FPU is just as good as a real one, as far as the C library
394 is concerned. You only need to say --without-fp if your machine has no way
395 to execute floating-point instructions.
397 People who are interested in squeezing the last drop of performance
398 out of their machine may wish to avoid the trap overhead, but this is
399 far more trouble than it's worth: you then have to compile
400 *everything* this way, including the compiler's internal libraries
401 (libgcc.a for GNU C), because the calling conventions change.
404 1.13. When compiling GNU libc I get lots of errors saying functions
405 in glibc are duplicated in libgcc.
407 {EY} This is *exactly* the same problem that I was having. The problem was
408 due to the fact that configure didn't correctly detect that the linker flag
409 --no-whole-archive was supported in my linker. In my case it was because I
410 had run ./configure with bogus CFLAGS, and the test failed.
412 One thing that is particularly annoying about this problem is that once this
413 is misdetected, running configure again won't fix it unless you first delete
416 {UD} Starting with glibc-2.0.3 there should be a better test to avoid some
417 problems of this kind. The setting of CFLAGS is checked at the very
418 beginning and if it is not usable `configure' will bark.
421 1.14. Why do I get messages about missing thread functions when I use
422 librt? I don't even use threads.
424 {UD} In this case you probably mixed up your installation. librt uses
425 threads internally and has implicit references to the thread library.
426 Normally these references are satisfied automatically but if the thread
427 library is not in the expected place you must tell the linker where it is.
428 When using GNU ld it works like this:
430 gcc -o foo foo.c -Wl,-rpath-link=/some/other/dir -lrt
432 The `/some/other/dir' should contain the thread library. `ld' will use the
433 given path to find the implicitly referenced library while not disturbing
437 1.15. What's the problem with configure --enable-omitfp?
439 {AJ} When --enable-omitfp is set the libraries are built without frame
440 pointers. Some compilers produce buggy code for this model and therefore we
441 don't advise using it at the moment.
443 If you use --enable-omitfp, you're on your own. If you encounter problems
444 with a library that was build this way, we advise you to rebuild the library
445 without --enable-omitfp. If the problem vanishes consider tracking the
446 problem down and report it as compiler failure.
448 Since a library built with --enable-omitfp is undebuggable on most systems,
449 debuggable libraries are also built - you can use them by appending "_g" to
452 The compilation of these extra libraries and the compiler optimizations slow
453 down the build process and need more disk space.
456 1.16. I get failures during `make check'. What should I do?
458 {AJ} The testsuite should compile and run cleanly on your system; every
459 failure should be looked into. Depending on the failures, you probably
460 should not install the library at all.
462 You should consider using the `glibcbug' script to report the failure,
463 providing as much detail as possible. If you run a test directly, please
464 remember to set up the environment correctly. You want to test the compiled
465 library - and not your installed one. The best way is to copy the exact
466 command line which failed and run the test from the subdirectory for this
469 There are some failures which are not directly related to the GNU libc:
470 - Some compilers produce buggy code. No compiler gets single precision
471 complex numbers correct on Alpha. Otherwise, the egcs 1.1 release should be
472 ok; gcc 2.8.1 might cause some failures; gcc 2.7.2.x is so buggy that
473 explicit checks have been used so that you can't build with it.
474 - The kernel might have bugs. For example on Linux/Alpha 2.0.34 the
475 floating point handling has quite a number of bugs and therefore most of
476 the test cases in the math subdirectory will fail. Linux 2.2 has
477 fixes for the floating point support on Alpha. The Linux/SPARC kernel has
478 also some bugs in the FPU emulation code (as of Linux 2.2.0).
481 1.17. What is symbol versioning good for? Do I need it?
483 {AJ} Symbol versioning solves problems that are related to interface
484 changes. One version of an interface might have been introduced in a
485 previous version of the GNU C library but the interface or the semantics of
486 the function has been changed in the meantime. For binary compatibility
487 with the old library, a newer library needs to still have the old interface
488 for old programs. On the other hand, new programs should use the new
489 interface. Symbol versioning is the solution for this problem. The GNU
490 libc version 2.1 uses symbol versioning by default if the installed binutils
493 We don't advise building without symbol versioning, since you lose binary
494 compatibility - forever! The binary compatibility you lose is not only
495 against the previous version of the GNU libc (version 2.0) but also against
499 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
501 2. Installation and configuration issues
503 2.1. Can I replace the libc on my Linux system with GNU libc?
505 {UD} You cannot replace any existing libc for Linux with GNU libc. It is
506 binary incompatible and therefore has a different major version. You can,
507 however, install it alongside your existing libc.
509 For Linux there are three major libc versions:
511 libc-5 original ELF libc
514 You can have any combination of these three installed. For more information
515 consult documentation for shared library handling. The Makefiles of GNU
516 libc will automatically generate the needed symbolic links which the linker
520 2.2. How do I configure GNU libc so that the essential libraries
521 like libc.so go into /lib and the other into /usr/lib?
523 {UD,AJ} Like all other GNU packages GNU libc is designed to use a base
524 directory and install all files relative to this. The default is
525 /usr/local, because this is safe (it will not damage the system if installed
526 there). If you wish to install GNU libc as the primary C library on your
527 system, set the base directory to /usr (i.e. run configure --prefix=/usr
528 <other_options>). Note that this can damage your system; see question 2.3 for
531 Some systems like Linux have a filesystem standard which makes a difference
532 between essential libraries and others. Essential libraries are placed in
533 /lib because this directory is required to be located on the same disk
534 partition as /. The /usr subtree might be found on another
535 partition/disk. If you configure for Linux with --prefix=/usr, then this
536 will be done automatically.
538 To install the essential libraries which come with GNU libc in /lib on
539 systems other than Linux one must explicitly request it. Autoconf has no
540 option for this so you have to use a `configparms' file (see the `INSTALL'
541 file for details). It should contain:
546 The first line specifies the directory for the essential libraries, the
547 second line the directory for system configuration files.
550 2.3. How should I avoid damaging my system when I install GNU libc?
552 {ZW} If you wish to be cautious, do not configure with --prefix=/usr. If
553 you don't specify a prefix, glibc will be installed in /usr/local, where it
554 will probably not break anything. (If you wish to be certain, set the
555 prefix to something like /usr/local/glibc2 which is not used for anything.)
557 The dangers when installing glibc in /usr are twofold:
559 * glibc will overwrite the headers in /usr/include. Other C libraries
560 install a different but overlapping set of headers there, so the
561 effect will probably be that you can't compile anything. You need to
562 rename /usr/include out of the way first. (Do not throw it away; you
563 will then lose the ability to compile programs against your old libc.)
565 * None of your old libraries, static or shared, can be used with a
566 different C library major version. For shared libraries this is not a
567 problem, because the filenames are different and the dynamic linker
568 will enforce the restriction. But static libraries have no version
569 information. You have to evacuate all the static libraries in
570 /usr/lib to a safe location.
572 The situation is rather similar to the move from a.out to ELF which
573 long-time Linux users will remember.
576 2.4. Do I need to use GNU CC to compile programs that will use the
579 {ZW} In theory, no; the linker does not care, and the headers are supposed
580 to check for GNU CC before using its extensions to the C language.
582 However, there are currently no ports of glibc to systems where another
583 compiler is the default, so no one has tested the headers extensively
584 against another compiler. You may therefore encounter difficulties. If you
585 do, please report them as bugs.
587 Also, in several places GNU extensions provide large benefits in code
588 quality. For example, the library has hand-optimized, inline assembly
589 versions of some string functions. These can only be used with GCC. See
590 question 3.8 for details.
593 2.5. When linking with the new libc I get unresolved symbols
594 `crypt' and `setkey'. Why aren't these functions in the
597 {UD} The US places restrictions on exporting cryptographic programs and
598 source code. Until this law gets abolished we cannot ship the cryptographic
599 functions together with glibc.
601 The functions are available, as an add-on (see question 1.11). People in the US
602 may get it from the same place they got GNU libc from. People outside the
603 US should get the code from ftp://ftp.ifi.uio.no/pub/gnu, or another archive
604 site outside the USA. The README explains how to install the sources.
606 If you already have the crypt code on your system the reason for the failure
607 is probably that you did not link with -lcrypt. The crypto functions are in
608 a separate library to make it possible to export GNU libc binaries from the
612 2.6. When I use GNU libc on my Linux system by linking against
613 the libc.so which comes with glibc all I get is a core dump.
615 {UD} On Linux, gcc sets the dynamic linker to /lib/ld-linux.so.1 unless the
616 user specifies a -dynamic-linker argument. This is the name of the libc5
617 dynamic linker, which does not work with glibc.
619 For casual use of GNU libc you can just specify to the linker
620 --dynamic-linker=/lib/ld-linux.so.2
622 which is the glibc dynamic linker, on Linux systems. On other systems the
623 name is /lib/ld.so.1. When linking via gcc, you've got to add
624 -Wl,--dynamic-linker=/lib/ld-linux.so.2
626 to the gcc command line.
628 To change your environment to use GNU libc for compiling you need to change
629 the `specs' file of your gcc. This file is normally found at
631 /usr/lib/gcc-lib/<arch>/<version>/specs
633 In this file you have to change a few things:
635 - change `ld-linux.so.1' to `ld-linux.so.2'
637 - remove all expression `%{...:-lgmon}'; there is no libgmon in glibc
639 - fix a minor bug by changing %{pipe:-} to %|
641 Here is what the gcc-2.7.2 specs file should look like when GNU libc is
644 -----------------------------------------------------------------------
646 %{V} %{v:%{!V:-V}} %{Qy:} %{!Qn:-Qy} %{n} %{T} %{Ym,*} %{Yd,*} %{Wa,*:%*}
652 %{fPIC:-D__PIC__ -D__pic__} %{fpic:-D__PIC__ -D__pic__} %{!m386:-D__i486__} %{posix:-D_POSIX_SOURCE} %{pthread:-D_REENTRANT}
661 %{!shared:crtend.o%s} %{shared:crtendS.o%s} crtn.o%s
664 -m elf_i386 %{shared:-shared} %{!shared: %{!ibcs: %{!static: %{rdynamic:-export-dynamic} %{!dynamic-linker:-dynamic-linker /lib/ld-linux.so.2}} %{static:-static}}}
667 %{!shared: %{pthread:-lpthread} %{profile:-lc_p} %{!profile: -lc}}
673 %{!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}
675 *switches_need_spaces:
679 %{funsigned-char:-D__CHAR_UNSIGNED__}
682 -D__ELF__ -Dunix -Di386 -Dlinux -Asystem(unix) -Asystem(posix) -Acpu(i386) -Amachine(i386)
690 -----------------------------------------------------------------------
692 Things get a bit more complicated if you have GNU libc installed in some
693 other place than /usr, i.e., if you do not want to use it instead of the old
694 libc. In this case the needed startup files and libraries are not found in
695 the regular places. So the specs file must tell the compiler and linker
698 Version 2.7.2.3 does and future versions of GCC will automatically
699 provide the correct specs.
702 2.7. Looking through the shared libc file I haven't found the
703 functions `stat', `lstat', `fstat', and `mknod' and while
704 linking on my Linux system I get error messages. How is
705 this supposed to work?
707 {RM} Believe it or not, stat and lstat (and fstat, and mknod) are supposed
708 to be undefined references in libc.so.6! Your problem is probably a missing
709 or incorrect /usr/lib/libc.so file; note that this is a small text file now,
710 not a symlink to libc.so.6. It should look something like this:
712 GROUP ( libc.so.6 libc_nonshared.a )
715 2.8. When I run an executable on one system which I compiled on
716 another, I get dynamic linker errors. Both systems have the same
717 version of glibc installed. What's wrong?
719 {ZW} Glibc on one of these systems was compiled with gcc 2.7 or 2.8, the
720 other with egcs (any version). Egcs has functions in its internal
721 `libgcc.a' to support exception handling with C++. They are linked into
722 any program or dynamic library compiled with egcs, whether it needs them or
723 not. Dynamic libraries then turn around and export those functions again
724 unless special steps are taken to prevent them.
726 When you link your program, it resolves its references to the exception
727 functions to the ones exported accidentally by libc.so. That works fine as
728 long as libc has those functions. On the other system, libc doesn't have
729 those functions because it was compiled by gcc 2.8, and you get undefined
730 symbol errors. The symbols in question are named things like
731 `__register_frame_info'.
733 For glibc 2.0, the workaround is to not compile libc with egcs. We've also
734 incorporated a patch which should prevent the EH functions sneaking into
735 libc. It doesn't matter what compiler you use to compile your program.
737 For glibc 2.1, we've chosen to do it the other way around: libc.so
738 explicitly provides the EH functions. This is to prevent other shared
739 libraries from doing it. You must therefore compile glibc 2.1 with EGCS.
740 Again, it doesn't matter what compiler you use for your programs.
743 2.9. How can I compile gcc 2.7.2.1 from the gcc source code using
746 {AJ} There's only correct support for glibc 2.0.x in gcc 2.7.2.3 or later.
747 But you should get at least gcc 2.8.1 or egcs 1.0.2 (or later versions)
751 2.10. The `gencat' utility cannot process the catalog sources which
752 were used on my Linux libc5 based system. Why?
754 {UD} The `gencat' utility provided with glibc complies to the XPG standard.
755 The older Linux version did not obey the standard, so they are not
758 To ease the transition from the Linux version some of the non-standard
759 features are also present in the `gencat' program of GNU libc. This mainly
760 includes the use of symbols for the message number and the automatic
761 generation of header files which contain the needed #defines to map the
764 Here is a simple SED script to convert at least some Linux specific catalog
765 files to the XPG4 form:
767 -----------------------------------------------------------------------
768 # Change catalog source in Linux specific format to standard XPG format.
769 # Ulrich Drepper <drepper@cygnus.com>, 1996.
773 s/\$ #\([^ ]*\).*/\1/
775 s/\$ #[^ ]* *\(.*\)/\$ \1/
781 s/\(.*\)\n\(.*\)/\2 \1/
783 -----------------------------------------------------------------------
786 2.11. Programs using libc have their messages translated, but other
787 behavior is not localized (e.g. collating order); why?
789 {ZW} Translated messages are automatically installed, but the locale
790 database that controls other behaviors is not. You need to run localedef to
791 install this database, after you have run `make install'. For example, to
792 set up the French Canadian locale, simply issue the command
794 localedef -i fr_CA -f ISO-8859-1 fr_CA
796 Please see localedata/README in the source tree for further details.
799 2.12. I have set up /etc/nis.conf, and the Linux libc 5 with NYS
800 works great. But the glibc NIS+ doesn't seem to work.
802 {TK} The glibc NIS+ implementation uses a /var/nis/NIS_COLD_START file for
803 storing information about the NIS+ server and their public keys, because the
804 nis.conf file does not contain all the necessary information. You have to
805 copy a NIS_COLD_START file from a Solaris client (the NIS_COLD_START file is
806 byte order independent) or generate it with nisinit from the nis-tools
807 package; available at
809 http://www-vt.uni-paderborn.de/~kukuk/linux/nisplus.html
812 2.13. I have killed ypbind to stop using NIS, but glibc
815 {TK} For faster NIS lookups, glibc uses the /var/yp/binding/ files from
816 ypbind. ypbind 3.3 and older versions don't always remove these files, so
817 glibc will continue to use them. Other BSD versions seem to work correctly.
818 Until ypbind 3.4 is released, you can find a patch at
820 ftp://ftp.kernel.org/pub/linux/utils/net/NIS/ypbind-3.3-glibc4.diff.gz
823 2.14. Under Linux/Alpha, I always get "do_ypcall: clnt_call:
824 RPC: Unable to receive; errno = Connection refused" when using NIS.
826 {TK} You need a ypbind version which is 64bit clean. Some versions are not
827 64bit clean. A 64bit clean implementation is ypbind-mt. For ypbind 3.3,
828 you need the patch from ftp.kernel.org (See the previous question). I don't
829 know about other versions.
832 2.15. After installing glibc name resolving doesn't work properly.
834 {AJ} You probably should read the manual section describing nsswitch.conf
835 (just type `info libc "NSS Configuration File"'). The NSS configuration
836 file is usually the culprit.
839 2.16. How do I create the databases for NSS?
841 {AJ} If you have an entry "db" in /etc/nsswitch.conf you should also create
842 the database files. The glibc sources contain a Makefile which does the
843 necessary conversion and calls to create those files. The file is
844 `db-Makefile' in the subdirectory `nss' and you can call it with `make -f
845 db-Makefile'. Please note that not all services are capable of using a
846 database. Currently passwd, group, ethers, protocol, rpc, services shadow
847 and netgroup are implemented.
850 2.17. I have /usr/include/net and /usr/include/scsi as symlinks
851 into my Linux source tree. Is that wrong?
853 {PB} This was necessary for libc5, but is not correct when using glibc.
854 Including the kernel header files directly in user programs usually does not
855 work (see question 3.5). glibc provides its own <net/*> and <scsi/*> header
856 files to replace them, and you may have to remove any symlink that you have
857 in place before you install glibc. However, /usr/include/asm and
858 /usr/include/linux should remain as they were.
861 2.18. Programs like `logname', `top', `uptime' `users', `w' and
862 `who', show incorrect information about the (number of)
863 users on my system. Why?
865 {MK} See question 3.2.
868 2.19. After upgrading to glibc 2.1 with symbol versioning I get
869 errors about undefined symbols. What went wrong?
871 {AJ} The problem is caused either by wrong program code or tools. In the
872 versioned libc a lot of symbols are now local that were global symbols in
873 previous versions. It seems that programs linked against older versions
874 often accidentally used libc global variables -- something that should not
877 The only way to fix this is to recompile your program. Sorry, that's the
878 price you might have to pay once for quite a number of advantages with
882 2.20. When I start the program XXX after upgrading the library
884 XXX: Symbol `_sys_errlist' has different size in shared
885 object, consider re-linking
886 Why? What should I do?
888 {UD} As the message says, relink the binary. The problem is that a few
889 symbols from the library can change in size and there is no way to avoid
890 this. _sys_errlist is a good example. Occasionally there are new error
891 numbers added to the kernel and this must be reflected at user level,
892 breaking programs that refer to them directly.
894 Such symbols should normally not be used at all. There are mechanisms to
895 avoid using them. In the case of _sys_errlist, there is the strerror()
896 function which should _always_ be used instead. So the correct fix is to
897 rewrite that part of the application.
899 In some situations (especially when testing a new library release) it might
900 be possible that a symbol changed size when that should not have happened.
901 So in case of doubt report such a warning message as a problem.
904 2.21. What do I need for C++ development?
906 {HJ,AJ} You need either egcs 1.1 which comes directly with libstdc++ or
907 gcc-2.8.1 together with libstdc++ 2.8.1.1. egcs 1.1 has the better C++
908 support and works directly with glibc 2.1. If you use gcc-2.8.1 with
909 libstdc++ 2.8.1.1, you need to modify libstdc++ a bit. A patch is available
911 ftp://alpha.gnu.org/gnu/libstdc++-2.8.1.1-glibc2.1-diff.gz
913 Please note that libg++ 2.7.2 (and the Linux Versions 2.7.2.x) doesn't work
914 very well with the GNU C library due to vtable thunks. If you're upgrading
915 from glibc 2.0.x to 2.1 you have to recompile libstdc++ since the library
916 compiled for 2.0 is not compatible due to the new Large File Support (LFS)
919 {UD} But since in the case of a shared libstdc++ the version numbers should
920 be different existing programs will continue to work.
923 2.22. Even statically linked programs need some shared libraries
924 which is not acceptable for me. What can I do?
926 {AJ} NSS (for details just type `info libc "Name Service Switch"') won't
927 work properly without shared libraries. NSS allows using different services
928 (e.g. NIS, files, db, hesiod) by just changing one configuration file
929 (/etc/nsswitch.conf) without relinking any programs. The only disadvantage
930 is that now static libraries need to access shared libraries. This is
931 handled transparently by the GNU C library.
933 A solution is to configure glibc with --enable-static-nss. In this case you
934 can create a static binary that will use only the services dns and files
935 (change /etc/nsswitch.conf for this). You need to link explicitly against
936 all these services. For example:
938 gcc -static test-netdb.c -o test-netdb.c \
939 -lc -lnss_files -lnss_dns -lresolv
941 The problem with this approach is that you've got to link every static
942 program that uses NSS routines with all those libraries.
944 {UD} In fact, one cannot say anymore that a libc compiled with this
945 option is using NSS. There is no switch anymore. Therefore it is
946 *highly* recommended *not* to use --enable-static-nss since this makes
947 the behaviour of the programs on the system inconsistent.
950 2.23. I just upgraded my Linux system to glibc and now I get
951 errors whenever I try to link any program.
953 {ZW} This happens when you have installed glibc as the primary C library but
954 have stray symbolic links pointing at your old C library. If the first
955 `libc.so' the linker finds is libc 5, it will use that. Your program
956 expects to be linked with glibc, so the link fails.
958 The most common case is that glibc put its `libc.so' in /usr/lib, but there
959 was a `libc.so' from libc 5 in /lib, which gets searched first. To fix the
960 problem, just delete /lib/libc.so. You may also need to delete other
961 symbolic links in /lib, such as /lib/libm.so if it points to libm.so.5.
963 {AJ} The perl script test-installation.pl which is run as last step during
964 an installation of glibc that is configured with --prefix=/usr should help
965 detect these situations. If the script reports problems, something is
969 2.24. When I use nscd the machine freezes.
971 {UD} You cannot use nscd with Linux 2.0.*. There is functionality missing
972 in the kernel and work-arounds are not suitable. Besides, some parts of the
973 kernel are too buggy when it comes to using threads.
975 If you need nscd, you have to use at least a 2.1 kernel.
977 Note that I have at this point no information about any other platform.
980 2.25. I need lots of open files. What do I have to do?
982 {AJ} This is at first a kernel issue. The kernel defines limits with
983 OPEN_MAX the number of simultaneous open files and with FD_SETSIZE the
984 number of used file descriptors. You need to change these values in your
985 kernel and recompile the kernel so that the kernel allows to use more open
986 files. You don't necessarily need to recompile the GNU C library since the
987 only place where OPEN_MAX and FD_SETSIZE is really needed in the library
988 itself is the size of fd_set which is used by select.
990 The GNU C library is now (nearly) select free. This means it internally has
991 no limits imposed by the `fd_set' type. Instead almost all places where the
992 functionality is needed the `poll' function is used.
994 If you increase the number of file descriptors in the kernel you don't need
995 to recompile the C library. The remaining select calls are in the RPC code.
996 If your RPC daemons don't need more than FD_SETSIZE file descriptors, you
997 don't need to change anything at all.
999 {UD} You can always get the maximum number of file descriptors a process is
1000 allowed to have open at any time using
1002 number = sysconf (_SC_OPEN_MAX);
1004 This will work even if the kernel limits change.
1007 2.26. How do I get the same behavior on parsing /etc/passwd and
1008 /etc/group as I have with libc5 ?
1010 {TK} The name switch setup in /etc/nsswitch.conf selected by most Linux
1011 distributions does not support +/- and netgroup entries in the files like
1012 /etc/passwd. Though this is the preferred setup some people might have
1013 setups coming over from the libc5 days where it was the default to recognize
1014 lines like this. To get back to the old behaviour one simply has to change
1015 the rules for passwd, group, and shadow in the nsswitch.conf file as
1027 2.27. What needs to be recompiled when upgrading from glibc 2.0 to glibc
1030 {AJ,CG} If you just upgrade the glibc from 2.0.x (x <= 7) to 2.1, binaries
1031 that have been linked against glibc 2.0 will continue to work.
1033 If you compile your own binaries against glibc 2.1, you also need to
1034 recompile some other libraries. The problem is that libio had to be
1035 changed and therefore libraries that are based or depend on the libio
1036 of glibc, e.g. ncurses or slang, need to be recompiled. If you
1037 experience strange segmentation faults in your programs linked against
1038 glibc 2.1, you might need to recompile your libraries.
1040 Another problem is that older binaries that were linked statically against
1041 glibc 2.0 will reference the older nss modules (libnss_files.so.1 instead of
1042 libnss_files.so.2), so don't remove them. Also, the old glibc-2.0 compiled
1043 static libraries (libfoo.a) which happen to depend on the older libio
1044 behavior will be broken by the glibc 2.1 upgrade. We plan to produce a
1045 compatibility library that people will be able to link in if they want
1046 to compile a static library generated against glibc 2.0 into a program
1047 on a glibc 2.1 system. You just add -lcompat and you should be fine.
1049 The glibc-compat add-on will provide the libcompat.a library, the older
1050 nss modules, and a few other files. Together, they should make it
1051 possible to do development with old static libraries on a glibc 2.1
1052 system. This add-on is still in development. You can get it from
1053 ftp://alpha.gnu.org/gnu/glibc-compat-2.1.tar.gz
1054 but please keep in mind that it is experimental.
1057 2.28. Why is extracting files via tar so slow?
1059 {AJ} Extracting of tar archives might be quite slow since tar has to look up
1060 userid and groupids and doesn't cache negative results. If you have nis or
1061 nisplus in your /etc/nsswitch.conf for the passwd and/or group database,
1062 each file extractions needs a network connection. There are two possible
1065 - do you really need NIS/NIS+ (some Linux distributions add by default
1066 nis/nisplus even if it's not needed)? If not, just remove the entries.
1068 - if you need NIS/NIS+, use the Name Service Cache Daemon nscd that comes
1072 2.29. Compiling programs I get parse errors in libio.h (e.g. "parse error
1073 before `_IO_seekoff'"). How should I fix this?
1075 {AJ} You might get the following errors when upgrading to glibc 2.1:
1077 In file included from /usr/include/stdio.h:57,
1079 /usr/include/libio.h:335: parse error before `_IO_seekoff'
1080 /usr/include/libio.h:335: parse error before `_G_off64_t'
1081 /usr/include/libio.h:336: parse error before `_IO_seekpos'
1082 /usr/include/libio.h:336: parse error before `_G_fpos64_t'
1084 The problem is a wrong _G_config.h file in your include path. The
1085 _G_config.h file that comes with glibc 2.1 should be used and not one from
1086 libc5 or from a compiler directory. To check which _G_config.h file the
1087 compiler uses, compile your program with `gcc -E ...|grep G_config.h' and
1088 remove that file. Your compiler should pick up the file that has been
1089 installed by glibc 2.1 in your include directory.
1092 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1094 3. Source and binary incompatibilities, and what to do about them
1096 3.1. I expect GNU libc to be 100% source code compatible with
1097 the old Linux based GNU libc. Why isn't it like this?
1099 {DMT,UD} Not every extension in Linux libc's history was well thought-out.
1100 In fact it had a lot of problems with standards compliance and with
1101 cleanliness. With the introduction of a new version number these errors can
1102 now be corrected. Here is a list of the known source code
1105 * _GNU_SOURCE: glibc does not make the GNU extensions available
1106 automatically. If a program depends on GNU extensions or some
1107 other non-standard functionality, it is necessary to compile it
1108 with the C compiler option -D_GNU_SOURCE, or better, to put
1109 `#define _GNU_SOURCE' at the beginning of your source files, before
1110 any C library header files are included. This difference normally
1111 manifests itself in the form of missing prototypes and/or data type
1112 definitions. Thus, if you get such errors, the first thing you
1113 should do is try defining _GNU_SOURCE and see if that makes the
1116 For more information consult the file `NOTES' in the GNU C library
1119 * reboot(): GNU libc sanitizes the interface of reboot() to be more
1120 compatible with the interface used on other OSes. reboot() as
1121 implemented in glibc takes just one argument. This argument
1122 corresponds to the third argument of the Linux reboot system call.
1123 That is, a call of the form reboot(a, b, c) needs to be changed into
1124 reboot(c). Beside this the header <sys/reboot.h> defines the needed
1125 constants for the argument. These RB_* constants should be used
1126 instead of the cryptic magic numbers.
1128 * swapon(): the interface of this function didn't change, but the
1129 prototype is in a separate header file <sys/swap.h>. This header
1130 file also provides the SWAP_* constants defined by <linux/swap.h>;
1131 you should use them for the second argument to swapon().
1133 * errno: If a program uses the variable "errno", then it _must_
1134 include <errno.h>. The old libc often (erroneously) declared this
1135 variable implicitly as a side-effect of including other libc header
1136 files. glibc is careful to avoid such namespace pollution, which,
1137 in turn, means that you really need to include the header files that
1138 you depend on. This difference normally manifests itself in the
1139 form of the compiler complaining about references to an undeclared
1142 * Linux-specific syscalls: All Linux system calls now have appropriate
1143 library wrappers and corresponding declarations in various header files.
1144 This is because the syscall() macro that was traditionally used to
1145 work around missing syscall wrappers are inherently non-portable and
1146 error-prone. The following table lists all the new syscall stubs,
1147 the header-file declaring their interface and the system call name.
1149 syscall name: wrapper name: declaring header file:
1150 ------------- ------------- ----------------------
1151 bdflush bdflush <sys/kdaemon.h>
1152 syslog ksyslog_ctl <sys/klog.h>
1154 * lpd: Older versions of lpd depend on a routine called _validuser().
1155 The library does not provide this function, but instead provides
1156 __ivaliduser() which has a slightly different interface. Simply
1157 upgrading to a newer lpd should fix this problem (e.g., the 4.4BSD
1158 lpd is known to be working).
1160 * resolver functions/BIND: like on many other systems the functions of
1161 the resolver library are not included in libc itself. There is a
1162 separate library libresolv. If you get undefined symbol errors for
1163 symbols starting with `res_*' simply add -lresolv to your linker
1166 * the `signal' function's behavior corresponds to the BSD semantic and
1167 not the SysV semantic as it was in libc-5. The interface on all GNU
1168 systems shall be the same and BSD is the semantic of choice. To use
1169 the SysV behavior simply use `sysv_signal', or define _XOPEN_SOURCE.
1170 See question 3.7 for details.
1173 3.2. Why does getlogin() always return NULL on my Linux box?
1175 {UD} The GNU C library has a format for the UTMP and WTMP file which differs
1176 from what your system currently has. It was extended to fulfill the needs
1177 of the next years when IPv6 is introduced. The record size is different and
1178 some fields have different positions. The files written by functions from
1179 the one library cannot be read by functions from the other library. Sorry,
1180 but this is what a major release is for. It's better to have a cut now than
1181 having no means to support the new techniques later.
1183 {MK} There is however a (partial) solution for this problem. Please take a
1184 look at the file `login/README.utmpd'.
1187 3.3. Where are the DST_* constants found in <sys/time.h> on many
1190 {UD} These constants come from the old BSD days and are not used anymore
1191 (libc5 does not actually implement the handling although the constants are
1194 Instead GNU libc contains zone database support and compatibility code for
1195 POSIX TZ environment variable handling. For former is very much preferred
1199 3.4. The prototypes for `connect', `accept', `getsockopt',
1200 `setsockopt', `getsockname', `getpeername', `send',
1201 `sendto', and `recvfrom' are different in GNU libc from
1202 any other system I saw. This is a bug, isn't it?
1204 {UD} No, this is no bug. This version of GNU libc already follows the new
1205 Single Unix specifications (and I think the POSIX.1g draft which adopted the
1206 solution). The type for a parameter describing a size is now `socklen_t', a
1210 3.5. On Linux I've got problems with the declarations in Linux
1213 {UD,AJ} On Linux, the use of kernel headers is reduced to the minimum. This
1214 gives Linus the ability to change the headers more freely. Also, user
1215 programs are now insulated from changes in the size of kernel data
1218 For example, the sigset_t type is 32 or 64 bits wide in the kernel. In
1219 glibc it is 1024 bits wide. This guarantees that when the kernel gets a
1220 bigger sigset_t (for POSIX.1e realtime support, say) user programs will not
1221 have to be recompiled. Consult the header files for more information about
1224 Therefore you shouldn't include Linux kernel header files directly if glibc
1225 has defined a replacement. Otherwise you might get undefined results because
1229 3.6. I don't include any kernel headers myself but the compiler
1230 still complains about redeclarations of types in the kernel
1233 {UD} The kernel headers before Linux 2.1.61 and 2.0.32 don't work correctly
1234 with glibc. Compiling C programs is possible in most cases but C++ programs
1235 have (due to the change of the name lookups for `struct's) problems. One
1236 prominent example is `struct fd_set'.
1238 There might be some problems left but 2.1.61/2.0.32 fix most of the known
1239 ones. See the BUGS file for other known problems.
1242 3.7. Why don't signals interrupt system calls anymore?
1244 {ZW} By default GNU libc uses the BSD semantics for signal(), unlike Linux
1245 libc 5 which used System V semantics. This is partially for compatibility
1246 with other systems and partially because the BSD semantics tend to make
1247 programming with signals easier.
1249 There are three differences:
1251 * BSD-style signals that occur in the middle of a system call do not
1252 affect the system call; System V signals cause the system call to
1253 fail and set errno to EINTR.
1255 * BSD signal handlers remain installed once triggered. System V signal
1256 handlers work only once, so one must reinstall them each time.
1258 * A BSD signal is blocked during the execution of its handler. In other
1259 words, a handler for SIGCHLD (for example) does not need to worry about
1260 being interrupted by another SIGCHLD. It may, however, be interrupted
1263 There is general consensus that for `casual' programming with signals, the
1264 BSD semantics are preferable. You don't need to worry about system calls
1265 returning EINTR, and you don't need to worry about the race conditions
1266 associated with one-shot signal handlers.
1268 If you are porting an old program that relies on the old semantics, you can
1269 quickly fix the problem by changing signal() to sysv_signal() throughout.
1270 Alternatively, define _XOPEN_SOURCE before including <signal.h>.
1272 For new programs, the sigaction() function allows you to specify precisely
1273 how you want your signals to behave. All three differences listed above are
1274 individually switchable on a per-signal basis with this function.
1276 If all you want is for one specific signal to cause system calls to fail and
1277 return EINTR (for example, to implement a timeout) you can do this with
1281 3.8. I've got errors compiling code that uses certain string
1284 {AJ} glibc 2.1 has special string functions that are faster than the normal
1285 library functions. Some of the functions are additionally implemented as
1286 inline functions and others as macros. This might lead to problems with
1287 existing codes but it is explicitly allowed by ISO C.
1289 The optimized string functions are only used when compiling with
1290 optimizations (-O1 or higher). The behavior can be changed with two feature
1293 * __NO_STRING_INLINES: Don't do any string optimizations.
1294 * __USE_STRING_INLINES: Use assembly language inline functions (might
1295 increase code size dramatically).
1297 Since some of these string functions are now additionally defined as macros,
1298 code like "char *strncpy();" doesn't work anymore (and is unnecessary, since
1299 <string.h> has the necessary declarations). Either change your code or
1300 define __NO_STRING_INLINES.
1302 {UD} Another problem in this area is that gcc still has problems on machines
1303 with very few registers (e.g., ix86). The inline assembler code can require
1304 almost all the registers and the register allocator cannot always handle
1307 One can disable the string optimizations selectively. Instead of writing
1309 cp = strcpy (foo, "lkj");
1313 cp = (strcpy) (foo, "lkj");
1315 This disables the optimization for that specific call.
1318 3.9. I get compiler messages "Initializer element not constant" with
1319 stdin/stdout/stderr. Why?
1321 {RM,AJ} Constructs like:
1322 static FILE *InPtr = stdin;
1324 lead to this message. This is correct behaviour with glibc since stdin is
1325 not a constant expression. Please note that a strict reading of ISO C does
1326 not allow above constructs.
1328 One of the advantages of this is that you can assign to stdin, stdout, and
1329 stderr just like any other global variable (e.g. `stdout = my_stream;'),
1330 which can be very useful with custom streams that you can write with libio
1331 (but beware this is not necessarily portable). The reason to implement it
1332 this way were versioning problems with the size of the FILE structure.
1334 To fix those programs you've got to initialize the variable at run time.
1335 This can be done, e.g. in main, like:
1343 or by constructors (beware this is gcc specific):
1346 static void inPtr_construct (void) __attribute__((constructor));
1347 static void inPtr_construct (void) { InPtr = stdin; }
1350 3.10. I can't compile with gcc -traditional (or
1351 -traditional-cpp). Why?
1353 {AJ} glibc2 does break -traditional and -traditonal-cpp - and will continue
1354 to do so. For example constructs of the form:
1360 are useful for debugging purposes (you can use foo with your debugger that's
1361 why we need the enum) and for compatibility (other systems use defines and
1365 3.11. I get some errors with `gcc -ansi'. Isn't glibc ANSI compatible?
1367 {AJ} The GNU C library is compatible with the ANSI/ISO C standard. If
1368 you're using `gcc -ansi', the glibc includes which are specified in the
1369 standard follow the standard. The ANSI/ISO C standard defines what has to be
1370 in the include files - and also states that nothing else should be in the
1371 include files (btw. you can still enable additional standards with feature
1374 The GNU C library is conforming to ANSI/ISO C - if and only if you're only
1375 using the headers and library functions defined in the standard.
1378 3.12. I can't access some functions anymore. nm shows that they do
1379 exist but linking fails nevertheless.
1381 {AJ} With the introduction of versioning in glibc 2.1 it is possible to
1382 export only those identifiers (functions, variables) that are really needed
1383 by application programs and by other parts of glibc. This way a lot of
1384 internal interfaces are now hidden. nm will still show those identifiers
1385 but marking them as internal. ISO C states that identifiers beginning with
1386 an underscore are internal to the libc. An application program normally
1387 shouldn't use those internal interfaces (there are exceptions,
1388 e.g. __ivaliduser). If a program uses these interfaces, it's broken. These
1389 internal interfaces might change between glibc releases or dropped
1393 3.13. When using the db-2 library which comes with glibc is used in
1394 the Perl db modules the testsuite is not passed. This did not
1395 happen with db-1, gdbm, or ndbm.
1397 {UD} You are using an outdated copy of the DB_File Perl module. In fact db-2
1398 finally removed the handling of zero-sized keys which was one of the features
1399 tested by the old Perl testsuite and therefore you see an error. But this
1400 never was documented and guaranteed, only broken programs used this feature.
1402 Consequently db-2 does not need to support this feature and instead signals
1403 an error which leads to easier debugging. The DB_File module maintainer
1404 Paul Marquess <pmarquess@bfsec.bt.co.uk> acknowledged this change and fixed
1405 the testsuite so that if you use DB_File v1.60 or later you should not have
1406 any more problems with db-2.
1409 3.14. The pow() inline function I get when including <math.h> is broken.
1410 I get segmentation faults when I run the program.
1412 {UD} Nope, the implementation is correct. The problem is with egcs version
1413 prior to 1.1. I.e., egcs 1.0 to 1.0.3 are all broken (at least on Intel).
1414 If you have to use this compiler you must define __NO_MATH_INLINES before
1415 including <math.h> to prevent the inline functions from being used. egcs 1.1
1416 fixes the problem. I don't know about gcc 2.8 and 2.8.1.
1419 3.15. The sys/sem.h file lacks the definition of `union semun'.
1421 {UD} Nope. This union has to be provided by the user program. Former glibc
1422 versions defined this but it was an error since it does not make much sense
1423 when thinking about it. The standards describing the System V IPC functions
1424 define it this way and therefore programs must be adopted.
1427 3.16. Why has <netinet/ip_fw.h> disappeared?
1429 {AJ} The corresponding Linux kernel data structures and constants are
1430 totally different in Linux 2.0 and Linux 2.2. This situation has to be
1431 taken care in user programs using the firewall structures and therefore
1432 those programs (ipfw is AFAIK the only one) should deal with this problem
1436 3.17. I get floods of warnings when I use -Wconversion and include
1437 <string.h> or <math.h>.
1439 {ZW} <string.h> and <math.h> intentionally use prototypes to override
1440 argument promotion. -Wconversion warns about all these. You can safely
1441 ignore the warnings.
1443 -Wconversion isn't really intended for production use, only for shakedown
1444 compiles after converting an old program to standard C.
1447 3.18. After upgrading to glibc 2.1, I receive errors about
1448 unresolved symbols, like `_dl_initial_searchlist' and can not
1449 execute any binaries. What went wrong?
1451 {AJ} This normally happens if your libc and ld (dynamic linker) are from
1452 different releases of glibc. For example, the dynamic linker
1453 /lib/ld-linux.so.2 comes from glibc 2.0.x, but the version of libc.so.6 is
1456 The path /lib/ld-linux.so.2 is hardcoded in every glibc2 binary but
1457 libc.so.6 is searched via /etc/ld.so.cache and in some special directories
1458 like /lib and /usr/lib. If you run configure with another prefix than /usr
1459 and put this prefix before /lib in /etc/ld.so.conf, your system will break.
1461 So what can you do? Either of the following should work:
1463 * Run `configure' with the same prefix argument you've used for glibc 2.0.x
1464 so that the same paths are used.
1465 * Replace /lib/ld-linux.so.2 with a link to the dynamic linker from glibc
1468 You can even call the dynamic linker by hand if everything fails. You've
1469 got to set LD_LIBRARY_PATH so that the corresponding libc is found and also
1470 need to provide an absolute path to your binary:
1472 LD_LIBRARY_PATH=<path-where-libc.so.6-lives> \
1473 <path-where-corresponding-dynamic-linker-lives>/ld-linux.so.2 \
1474 <path-to-binary>/binary
1476 For example `LD_LIBRARY_PATH=/libold /libold/ld-linux.so.2 /bin/mv ...'
1477 might be useful in fixing a broken system (if /libold contains dynamic
1478 linker and corresponding libc).
1480 With that command line no path is used. To further debug problems with the
1481 dynamic linker, use the LD_DEBUG environment variable, e.g.
1482 `LD_DEBUG=help echo' for the help text.
1484 If you just want to test this release, don't put the lib directory in
1485 /etc/ld.so.conf. You can call programs directly with full paths (as above).
1486 When compiling new programs against glibc 2.1, you've got to specify the
1487 correct paths to the compiler (option -I with gcc) and linker (options
1488 --dynamic-linker, -L and --rpath).
1491 3.19. bonnie reports that char i/o with glibc 2 is much slower than with
1492 libc5. What can be done?
1494 {AJ} The GNU C library uses thread safe functions by default and libc5 used
1495 non thread safe versions. The non thread safe functions have in glibc the
1496 suffix `_unlocked', for details check <stdio.h>. Using `putc_unlocked' etc.
1497 instead of `putc' should give nearly the same speed with bonnie (bonnie is a
1498 benchmark program for measuring disk access).
1501 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1505 4.1. After I changed configure.in I get `Autoconf version X.Y.
1506 or higher is required for this script'. What can I do?
1508 {UD} You have to get the specified autoconf version (or a later one)
1509 from your favorite mirror of ftp.gnu.org.
1512 4.2. When I try to compile code which uses IPv6 headers and
1513 definitions on my Linux 2.x.y system I am in trouble.
1514 Nothing seems to work.
1516 {UD} The problem is that IPv6 development still has not reached a point
1517 where the headers are stable. There are still lots of incompatible changes
1518 made and the libc headers have to follow.
1520 {PB} The 2.1 release of GNU libc aims to comply with the current versions of
1521 all the relevant standards. The IPv6 support libraries for older Linux
1522 systems used a different naming convention and so code written to work with
1523 them may need to be modified. If the standards make incompatible changes in
1524 the future then the libc may need to change again.
1526 IPv6 will not work with a 2.0.x kernel. When kernel 2.2 is released it
1527 should contain all the necessary support; until then you should use the
1528 latest 2.1.x release you can find. As of 98/11/26 the currently recommended
1529 kernel for IPv6 is 2.1.129.
1531 Also, as of the 2.1 release the IPv6 API provided by GNU libc is not
1532 100% complete. In particular the getipnodebyname and getipnodebyaddr
1533 functions are not implemented.
1536 4.3. When I set the timezone by setting the TZ environment variable
1537 to EST5EDT things go wrong since glibc computes the wrong time
1538 from this information.
1540 {UD} The problem is that people still use the braindamaged POSIX method to
1541 select the timezone using the TZ environment variable with a format EST5EDT
1542 or whatever. People, if you insist on using TZ instead of the timezone
1543 database (see below), read the POSIX standard, the implemented behaviour is
1544 correct! What you see is in fact the result of the decisions made while
1545 POSIX.1 was created. We've only implemented the handling of TZ this way to
1546 be POSIX compliant. It is not really meant to be used.
1548 The alternative approach to handle timezones which is implemented is the
1549 correct one to use: use the timezone database. This avoids all the problems
1550 the POSIX method has plus it is much easier to use. Simply run the tzselect
1551 shell script, answer the question and use the name printed in the end by
1552 making a symlink /etc/localtime pointing to /usr/share/zoneinfo/NAME (NAME
1553 is the returned value from tzselect). That's all. You never again have to
1556 So, please avoid sending bug reports about time related problems if you use
1557 the POSIX method and you have not verified something is really broken by
1558 reading the POSIX standards.
1561 4.4. What other sources of documentation about glibc are available?
1563 {AJ} The FSF has a page about the GNU C library at
1564 <http://www.gnu.org/software/libc/>. The problem data base of open and
1565 solved bugs in GNU libc is available at
1566 <http://www-gnats.gnu.org:8080/cgi-bin/wwwgnats.pl>. Eric Green has written
1567 a HowTo for converting from Linux libc5 to glibc2. The HowTo is accessable
1568 via the FSF page and at <http://www.imaxx.net/~thrytis/glibc>. Frodo
1569 Looijaard describes a different way installing glibc2 as secondary libc at
1570 <http://huizen.dds.nl/~frodol/glibc>.
1572 Please note that this is not a complete list.
1575 4.5. The timezone string for Sydney/Australia is wrong since even when
1576 daylight saving time is in effect the timezone string is EST.
1578 {UD} The problem for some timezones is that the local authorities decided
1579 to use the term "summer time" instead of "daylight saving time". In this
1580 case the abbreviation character `S' is the same as the standard one. So,
1583 Eastern Standard Time = EST
1584 Eastern Summer Time = EST
1586 Great! To get this bug fixed convince the authorities to change the laws
1587 and regulations of the country this effects. glibc behaves correctly.
1590 4.6. I've build make 3.77 against glibc 2.1 and now make gets
1591 segmentation faults.
1593 {AJ} GNU make 3.77 has support for 64 bit filesystems which is slightly
1594 broken (and one of the new features in the GNU C library 2.1 is 64 bit
1595 filesystem support :-( ). To get a working make you can use either make
1596 3.75 or patch 3.77. A working patch is available via RedHat's Rawhide server
1597 (ftp://rawhide.redhat.com/SRPMS/SRPMS/make-3.77-*src.rpm).
1600 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
1602 Answers were given by:
1603 {UD} Ulrich Drepper, <drepper@cygnus.com>
1604 {DMT} David Mosberger-Tang, <davidm@AZStarNet.com>
1605 {RM} Roland McGrath, <roland@gnu.org>
1606 {AJ} Andreas Jaeger, <aj@arthur.rhein-neckar.de>
1607 {EY} Eric Youngdale, <eric@andante.jic.com>
1608 {PB} Phil Blundell, <Philip.Blundell@pobox.com>
1609 {MK} Mark Kettenis, <kettenis@phys.uva.nl>
1610 {ZW} Zack Weinberg, <zack@rabi.phys.columbia.edu>
1611 {TK} Thorsten Kukuk, <kukuk@vt.uni-paderborn.de>
1612 {GK} Geoffrey Keating, <geoffk@ozemail.com.au>
1613 {HJ} H.J. Lu, <hjl@gnu.org>
1614 {CG} Cristian Gafton, <gafton@redhat.com>
1618 outline-regexp:"\\?"