| blob | b0d5d0c14d09a865cf4ab22e1140194ff5e29049 |
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,
13 please let me know.
15 --drepper@cygnus.com
16 \f
17 ? Compiling glibc
19 ?? What systems does the GNU C Library run on?
21 {UD} This is difficult to answer. The file `README' lists the architectures
22 GNU libc was known to run on *at some time*. This does not mean that it
23 still can be compiled and run on them now.
25 The systems glibc is known to work on as of this release, and most probably
26 in the future, are:
28 *-*-gnu GNU Hurd
29 i[3456]86-*-linux-gnu Linux-2.x on Intel
30 m68k-*-linux-gnu Linux-2.x on Motorola 680x0
31 alpha-*-linux-gnu Linux-2.x on DEC Alpha
32 powerpc-*-linux-gnu Linux and MkLinux on PowerPC systems
33 sparc-*-linux-gnu Linux-2.x on SPARC
34 sparc64-*-linux-gnu Linux-2.x on UltraSPARC
35 arm-*-none ARM standalone systems
36 arm-*-linux Linux-2.x on ARM
37 arm-*-linuxaout Linux-2.x on ARM using a.out binaries
39 Ports to other Linux platforms are in development, and may in fact work
40 already, but no one has sent us success reports for them. Currently no
41 ports to other operating systems are underway, although a few people have
42 expressed interest.
44 If you have a system not listed above (or in the `README' file) and you are
45 really interested in porting it, contact
47 <bug-glibc@gnu.org>
49 ??binsize What compiler do I need to build GNU libc?
51 {UD} You must use GNU CC to compile GNU libc. A lot of extensions of GNU CC
52 are used to increase portability and speed.
54 GNU CC is found, like all other GNU packages, on
56 ftp://ftp.gnu.org/pub/gnu
58 and the many mirror sites. ftp.gnu.org is always overloaded, so try to find
59 a local mirror first.
61 You should always try to use the latest official release. Older versions
62 may not have all the features GNU libc requires. The current releases of
63 egcs (1.0.3 and 1.1.1) should work with the GNU C library (for powerpc see
64 ?powerpc; for ARM see ?arm).
66 While the GNU CC should be able to compile glibc it is nevertheless adviced
67 to use EGCS. Comparing the sizes of glibc on Intel compiled with a recent
68 EGCS and gcc 2.8.1 shows this:
70 text data bss dec hex filename
71 egcs-2.93.10 862897 15944 12824 891665 d9b11 libc.so
72 gcc-2.8.1 959965 16468 12152 988585 f15a9 libc.so
74 Make up your own decision.
76 ?? When I try to compile glibc I get only error messages.
77 What's wrong?
79 {UD} You definitely need GNU make to build GNU libc. No other make
80 program has the needed functionality.
82 We recommend version GNU make version 3.75 or 3.77. Versions before 3.75
83 have bugs and/or are missing features. Version 3.76 has bugs which
84 appear when building big projects like GNU libc. 3.76.1 appears to work but
85 some people have reported problems. If you build GNU make 3.77 from source,
86 please read ?make first.
88 ?? Do I need a special linker or assembler?
90 {ZW} If you want a shared library, you need a linker and assembler that
91 understand all the features of ELF, including weak and versioned symbols.
92 The static library can be compiled with less featureful tools, but lacks key
93 features such as NSS.
95 For Linux or Hurd, you want binutils 2.8.1.0.23, 2.9.1, or 2.9.1.0.15 or
96 higher. These are the only versions we've tested and found reliable. Other
97 versions after 2.8.1.0.23 may work but we don't recommend them, especially
98 not when C++ is involved. Earlier versions do not work at all.
100 Other operating systems may come with system tools that have all the
101 necessary features, but this is moot because glibc hasn't been ported to
102 them.
104 ??powerpc Which compiler should I use for powerpc?
106 {GK} You want to use egcs 1.1 or later (together with the right versions
107 of all the other tools, of course).
109 In fact, egcs 1.1 has a bug that causes linuxthreads to be
110 miscompiled, resulting in segmentation faults when using condition
111 variables. There is a temporary patch at:
113 <http://discus.anu.edu.au/~geoffk/egcs-3.diff>
115 Later versions of egcs may fix this problem.
117 ??arm Which tools should I use for ARM?
119 {PB} You should use egcs 1.1 or a later version. For ELF systems some
120 changes are needed to the compiler; a patch against egcs-1.1.x can be found
121 at:
123 <ftp://ftp.netwinder.org/users/p/philb/egcs-1.1.1pre2-diff-981126>
125 Binutils 2.9.1.0.16 or later is also required.
127 ?? Do I need some more things to compile the GNU C Library?
129 {UD} Yes, there are some more :-).
131 * GNU gettext. This package contains the tools needed to construct
132 `message catalog' files containing translated versions of system
133 messages. See ftp://ftp.gnu.org/pub/gnu or better any mirror
134 site. (We distribute compiled message catalogs, but they may not be
135 updated in patches.)
137 * Some files are built with special tools. E.g., files ending in .gperf
138 need a `gperf' program. The GNU version (now available in a separate
139 package, formerly only as part of libg++) is known to work while some
140 vendor versions do not.
142 You should not need these tools unless you change the source files.
144 * Perl 5 is needed if you wish to test an installation of GNU libc
145 as the primary C library.
147 * When compiling for Linux, the header files of the Linux kernel must
148 be available to the compiler as <linux/*.h> and <asm/*.h>.
150 * lots of disk space (~170MB for i?86-linux; more for RISC platforms,
151 as much as 400MB).
153 * plenty of time. Compiling just the shared and static libraries for
154 i?86-linux takes approximately 1h on an i586@133, or 2.5h on
155 i486@66, or 4.5h on i486@33. Multiply this by 1.5 or 2.0 if you
156 build profiling and/or the highly optimized version as well. For
157 Hurd systems times are much higher.
159 You should avoid compiling in a NFS mounted filesystem. This is
160 very slow.
162 James Troup <J.J.Troup@comp.brad.ac.uk> reports a compile time of
163 45h34m for a full build (shared, static, and profiled) on Atari
164 Falcon (Motorola 68030 @ 16 Mhz, 14 Mb memory) and Jan Barte
165 <yann@plato.uni-paderborn.de> reports 22h48m on Atari TT030
166 (Motorola 68030 @ 32 Mhz, 34 Mb memory)
168 If you have some more measurements let me know.
170 ?? What version of the Linux kernel headers should be used?
172 {AJ,UD} The headers from the most recent Linux kernel should be used. The
173 headers used while compiling the GNU C library and the kernel binary used
174 when using the library do not need to match. The GNU C library runs without
175 problems on kernels that are older than the kernel headers used. The other
176 way round (compiling the GNU C library with old kernel headers and running
177 on a recent kernel) does not necessarily work. For example you can't use
178 new kernel features if you used old kernel headers to compile the GNU C
179 library.
181 {ZW} Even if you are using a 2.0 kernel on your machine, we recommend you
182 compile GNU libc with 2.2 kernel headers. That way you won't have to
183 recompile libc if you ever upgrade to kernel 2.2. To tell libc which
184 headers to use, give configure the --with-headers switch
185 (e.g. --with-headers=/usr/src/linux-2.2.0/include).
187 Note that you must configure the 2.2 kernel if you do this, otherwise libc
188 will be unable to find <linux/version.h>. Just change the current directory
189 to the root of the 2.2 tree and do `make include/linux/version.h'.
191 ?? The compiler hangs while building iconvdata modules. What's
192 wrong?
194 {ZW} This is a problem with old versions of GCC. Initialization of large
195 static arrays is very slow. The compiler will eventually finish; give it
196 time.
198 The problem is fixed in egcs 1.1.
200 ?? When I run `nm -u libc.so' on the produced library I still
201 find unresolved symbols. Can this be ok?
203 {UD} Yes, this is ok. There can be several kinds of unresolved symbols:
205 * magic symbols automatically generated by the linker. These have names
206 like __start_* and __stop_*
208 * symbols starting with _dl_* come from the dynamic linker
210 * weak symbols, which need not be resolved at all (fabs for example)
212 Generally, you should make sure you find a real program which produces
213 errors while linking before deciding there is a problem.
215 ??addon What are these `add-ons'?
217 {UD} To avoid complications with export rules or external source code some
218 optional parts of the libc are distributed as separate packages (e.g., the
219 crypt package, see ?crypt).
221 To use these packages as part of GNU libc, just unpack the tarfiles in the
222 libc source directory and tell the configuration script about them using the
223 --enable-add-ons option. If you give just --enable-add-ons configure tries
224 to find all the add-on packages in your source tree. This may not work. If
225 it doesn't, or if you want to select only a subset of the add-ons, give a
226 comma-separated list of the add-ons to enable:
228 configure --enable-add-ons=crypt,linuxthreads
230 for example.
232 Add-ons can add features (including entirely new shared libraries), override
233 files, provide support for additional architectures, and just about anything
234 else. The existing makefiles do most of the work; only some few stub rules
235 must be written to get everything running.
237 ?? My XXX kernel emulates a floating-point coprocessor for me.
238 Should I enable --with-fp?
240 {ZW} An emulated FPU is just as good as a real one, as far as the C library
241 is concerned. You only need to say --without-fp if your machine has no way
242 to execute floating-point instructions.
244 People who are interested in squeezing the last drop of performance
245 out of their machine may wish to avoid the trap overhead, but this is
246 far more trouble than it's worth: you then have to compile
247 *everything* this way, including the compiler's internal libraries
248 (libgcc.a for GNU C), because the calling conventions change.
250 ?? When compiling GNU libc I get lots of errors saying functions
251 in glibc are duplicated in libgcc.
253 {EY} This is *exactly* the same problem that I was having. The problem was
254 due to the fact that configure didn't correctly detect that the linker flag
255 --no-whole-archive was supported in my linker. In my case it was because I
256 had run ./configure with bogus CFLAGS, and the test failed.
258 One thing that is particularly annoying about this problem is that once this
259 is misdetected, running configure again won't fix it unless you first delete
260 config.cache.
262 {UD} Starting with glibc-2.0.3 there should be a better test to avoid some
263 problems of this kind. The setting of CFLAGS is checked at the very
264 beginning and if it is not usable `configure' will bark.
266 ?? Why do I get messages about missing thread functions when I use
267 librt? I don't even use threads.
269 {UD} In this case you probably mixed up your installation. librt uses
270 threads internally and has implicit references to the thread library.
271 Normally these references are satisfied automatically but if the thread
272 library is not in the expected place you must tell the linker where it is.
273 When using GNU ld it works like this:
275 gcc -o foo foo.c -Wl,-rpath-link=/some/other/dir -lrt
277 The `/some/other/dir' should contain the thread library. `ld' will use the
278 given path to find the implicitly referenced library while not disturbing
279 any other link path.
281 ?? What's the problem with configure --enable-omitfp?
283 {AJ} When --enable-omitfp is set the libraries are built without frame
284 pointers. Some compilers produce buggy code for this model and therefore we
285 don't advise using it at the moment.
287 If you use --enable-omitfp, you're on your own. If you encounter problems
288 with a library that was build this way, we advise you to rebuild the library
289 without --enable-omitfp. If the problem vanishes consider tracking the
290 problem down and report it as compiler failure.
292 Since a library built with --enable-omitfp is undebuggable on most systems,
293 debuggable libraries are also built - you can use them by appending "_g" to
294 the library names.
296 The compilation of these extra libraries and the compiler optimizations slow
297 down the build process and need more disk space.
299 ?? I get failures during `make check'. What should I do?
301 {AJ} The testsuite should compile and run cleanly on your system; every
302 failure should be looked into. Depending on the failures, you probably
303 should not install the library at all.
305 You should consider using the `glibcbug' script to report the failure,
306 providing as much detail as possible. If you run a test directly, please
307 remember to set up the environment correctly. You want to test the compiled
308 library - and not your installed one. The best way is to copy the exact
309 command line which failed and run the test from the subdirectory for this
310 test in the sources.
312 There are some failures which are not directly related to the GNU libc:
313 - Some compilers produce buggy code. No compiler gets single precision
314 complex numbers correct on Alpha. Otherwise, the egcs 1.1 release should be
315 ok; gcc 2.8.1 might cause some failures; gcc 2.7.2.x is so buggy that
316 explicit checks have been used so that you can't build with it.
317 - The kernel might have bugs. For example on Linux/Alpha 2.0.34 the
318 floating point handling has quite a number of bugs and therefore most of
319 the test cases in the math subdirectory will fail. Linux 2.2 has
320 fixes for the floating point support on Alpha. The Linux/SPARC kernel has
321 also some bugs in the FPU emulation code (as of Linux 2.2.0).
323 ?? What is symbol versioning good for? Do I need it?
325 {AJ} Symbol versioning solves problems that are related to interface
326 changes. One version of an interface might have been introduced in a
327 previous version of the GNU C library but the interface or the semantics of
328 the function has been changed in the meantime. For binary compatibility
329 with the old library, a newer library needs to still have the old interface
330 for old programs. On the other hand, new programs should use the new
331 interface. Symbol versioning is the solution for this problem. The GNU
332 libc version 2.1 uses symbol versioning by default if the installed binutils
333 supports it.
335 We don't advise building without symbol versioning, since you lose binary
336 compatibility - forever! The binary compatibility you lose is not only
337 against the previous version of the GNU libc (version 2.0) but also against
338 all future versions.
341 ? Installation and configuration issues
343 ?? Can I replace the libc on my Linux system with GNU libc?
345 {UD} You cannot replace any existing libc for Linux with GNU libc. It is
346 binary incompatible and therefore has a different major version. You can,
347 however, install it alongside your existing libc.
349 For Linux there are three major libc versions:
350 libc-4 a.out libc
351 libc-5 original ELF libc
352 libc-6 GNU libc
354 You can have any combination of these three installed. For more information
355 consult documentation for shared library handling. The Makefiles of GNU
356 libc will automatically generate the needed symbolic links which the linker
357 will use.
359 ?? How do I configure GNU libc so that the essential libraries
360 like libc.so go into /lib and the other into /usr/lib?
362 {UD,AJ} Like all other GNU packages GNU libc is designed to use a base
363 directory and install all files relative to this. The default is
364 /usr/local, because this is safe (it will not damage the system if installed
365 there). If you wish to install GNU libc as the primary C library on your
366 system, set the base directory to /usr (i.e. run configure --prefix=/usr
367 <other_options>). Note that this can damage your system; see ?safety for
368 details.
370 Some systems like Linux have a filesystem standard which makes a difference
371 between essential libraries and others. Essential libraries are placed in
372 /lib because this directory is required to be located on the same disk
373 partition as /. The /usr subtree might be found on another
374 partition/disk. If you configure for Linux with --prefix=/usr, then this
375 will be done automatically.
377 To install the essential libraries which come with GNU libc in /lib on
378 systems other than Linux one must explicitly request it. Autoconf has no
379 option for this so you have to use a `configparms' file (see the `INSTALL'
380 file for details). It should contain:
382 slibdir=/lib
383 sysconfdir=/etc
385 The first line specifies the directory for the essential libraries, the
386 second line the directory for system configuration files.
388 ??safety How should I avoid damaging my system when I install GNU libc?
390 {ZW} If you wish to be cautious, do not configure with --prefix=/usr. If
391 you don't specify a prefix, glibc will be installed in /usr/local, where it
392 will probably not break anything. (If you wish to be certain, set the
393 prefix to something like /usr/local/glibc2 which is not used for anything.)
395 The dangers when installing glibc in /usr are twofold:
397 * glibc will overwrite the headers in /usr/include. Other C libraries
398 install a different but overlapping set of headers there, so the effect
399 will probably be that you can't compile anything. You need to rename
400 /usr/include out of the way before running `make install'. (Do not throw
401 it away; you will then lose the ability to compile programs against your
402 old libc.)
404 * None of your old libraries, static or shared, can be used with a
405 different C library major version. For shared libraries this is not a
406 problem, because the filenames are different and the dynamic linker
407 will enforce the restriction. But static libraries have no version
408 information. You have to evacuate all the static libraries in
409 /usr/lib to a safe location.
411 The situation is rather similar to the move from a.out to ELF which
412 long-time Linux users will remember.
414 ?? Do I need to use GNU CC to compile programs that will use the
415 GNU C Library?
417 {ZW} In theory, no; the linker does not care, and the headers are supposed
418 to check for GNU CC before using its extensions to the C language.
420 However, there are currently no ports of glibc to systems where another
421 compiler is the default, so no one has tested the headers extensively
422 against another compiler. You may therefore encounter difficulties. If you
423 do, please report them as bugs.
425 Also, in several places GNU extensions provide large benefits in code
426 quality. For example, the library has hand-optimized, inline assembly
427 versions of some string functions. These can only be used with GCC. See
428 ?string for details.
430 ??crypt When linking with the new libc I get unresolved symbols
431 `crypt' and `setkey'. Why aren't these functions in the
432 libc anymore?
434 {UD} The US places restrictions on exporting cryptographic programs and
435 source code. Until this law gets abolished we cannot ship the cryptographic
436 functions together with glibc.
438 The functions are available, as an add-on (see ?addon). People in the US
439 may get it from the same place they got GNU libc from. People outside the
440 US should get the code from ftp.funet.fi [128.214.248.6] in the directory
441 pub/gnu/funet, or another archive site outside the USA. The README explains
442 how to install the sources.
444 If you already have the crypt code on your system the reason for the failure
445 is probably that you did not link with -lcrypt. The crypto functions are in
446 a separate library to make it possible to export GNU libc binaries from the
447 US.
449 ?? When I use GNU libc on my Linux system by linking against
450 the libc.so which comes with glibc all I get is a core dump.
452 {UD} On Linux, gcc sets the dynamic linker to /lib/ld-linux.so.1 unless the
453 user specifies a -dynamic-linker argument. This is the name of the libc5
454 dynamic linker, which does not work with glibc.
456 For casual use of GNU libc you can just specify to the linker
457 --dynamic-linker=/lib/ld-linux.so.2
459 which is the glibc dynamic linker, on Linux systems. On other systems the
460 name is /lib/ld.so.1. When linking via gcc, you've got to add
461 -Wl,--dynamic-linker=/lib/ld-linux.so.2
463 to the gcc command line.
465 To change your environment to use GNU libc for compiling you need to change
466 the `specs' file of your gcc. This file is normally found at
468 /usr/lib/gcc-lib/<arch>/<version>/specs
470 In this file you have to change a few things:
472 - change `ld-linux.so.1' to `ld-linux.so.2'
474 - remove all expression `%{...:-lgmon}'; there is no libgmon in glibc
476 - fix a minor bug by changing %{pipe:-} to %|
478 Here is what the gcc-2.7.2 specs file should look like when GNU libc is
479 installed at /usr:
481 -----------------------------------------------------------------------
482 *asm:
483 %{V} %{v:%{!V:-V}} %{Qy:} %{!Qn:-Qy} %{n} %{T} %{Ym,*} %{Yd,*} %{Wa,*:%*}
485 *asm_final:
486 %|
488 *cpp:
489 %{fPIC:-D__PIC__ -D__pic__} %{fpic:-D__PIC__ -D__pic__} %{!m386:-D__i486__} %{posix:-D_POSIX_SOURCE} %{pthread:-D_REENTRANT}
491 *cc1:
492 %{profile:-p}
494 *cc1plus:
497 *endfile:
498 %{!shared:crtend.o%s} %{shared:crtendS.o%s} crtn.o%s
500 *link:
501 -m elf_i386 %{shared:-shared} %{!shared: %{!ibcs: %{!static: %{rdynamic:-export-dynamic} %{!dynamic-linker:-dynamic-linker /lib/ld-linux.so.2}} %{static:-static}}}
503 *lib:
504 %{!shared: %{pthread:-lpthread} %{profile:-lc_p} %{!profile: -lc}}
506 *libgcc:
507 -lgcc
509 *startfile:
510 %{!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}
512 *switches_need_spaces:
515 *signed_char:
516 %{funsigned-char:-D__CHAR_UNSIGNED__}
518 *predefines:
519 -D__ELF__ -Dunix -Di386 -Dlinux -Asystem(unix) -Asystem(posix) -Acpu(i386) -Amachine(i386)
521 *cross_compile:
522 0
524 *multilib:
525 . ;
527 -----------------------------------------------------------------------
529 Things get a bit more complicated if you have GNU libc installed in some
530 other place than /usr, i.e., if you do not want to use it instead of the old
531 libc. In this case the needed startup files and libraries are not found in
532 the regular places. So the specs file must tell the compiler and linker
533 exactly what to use.
535 Version 2.7.2.3 does and future versions of GCC will automatically
536 provide the correct specs.
538 ?? Looking through the shared libc file I haven't found the
539 functions `stat', `lstat', `fstat', and `mknod' and while
540 linking on my Linux system I get error messages. How is
541 this supposed to work?
543 {RM} Believe it or not, stat and lstat (and fstat, and mknod) are supposed
544 to be undefined references in libc.so.6! Your problem is probably a missing
545 or incorrect /usr/lib/libc.so file; note that this is a small text file now,
546 not a symlink to libc.so.6. It should look something like this:
548 GROUP ( libc.so.6 libc_nonshared.a )
550 ?? When I run an executable on one system which I compiled on
551 another, I get dynamic linker errors. Both systems have the same
552 version of glibc installed. What's wrong?
554 {ZW} Glibc on one of these systems was compiled with gcc 2.7 or 2.8, the
555 other with egcs (any version). Egcs has functions in its internal
556 `libgcc.a' to support exception handling with C++. They are linked into
557 any program or dynamic library compiled with egcs, whether it needs them or
558 not. Dynamic libraries then turn around and export those functions again
559 unless special steps are taken to prevent them.
561 When you link your program, it resolves its references to the exception
562 functions to the ones exported accidentally by libc.so. That works fine as
563 long as libc has those functions. On the other system, libc doesn't have
564 those functions because it was compiled by gcc 2.8, and you get undefined
565 symbol errors. The symbols in question are named things like
566 `__register_frame_info'.
568 For glibc 2.0, the workaround is to not compile libc with egcs. We've also
569 incorporated a patch which should prevent the EH functions sneaking into
570 libc. It doesn't matter what compiler you use to compile your program.
572 For glibc 2.1, we've chosen to do it the other way around: libc.so
573 explicitly provides the EH functions. This is to prevent other shared
574 libraries from doing it.
576 {UD} Starting with glibc 2.1.1 you can compile glibc with gcc 2.8.1 or
577 newer since we have explicitly add references to the functions causing the
578 problem. But you nevertheless should use EGCS for other reasons
579 (see ?binsize).
581 ?? How can I compile gcc 2.7.2.1 from the gcc source code using
582 glibc 2.x?
584 {AJ} There's only correct support for glibc 2.0.x in gcc 2.7.2.3 or later.
585 But you should get at least gcc 2.8.1 or egcs 1.1 (or later versions)
586 instead.
588 ?? The `gencat' utility cannot process the catalog sources which
589 were used on my Linux libc5 based system. Why?
591 {UD} The `gencat' utility provided with glibc complies to the XPG standard.
592 The older Linux version did not obey the standard, so they are not
593 compatible.
595 To ease the transition from the Linux version some of the non-standard
596 features are also present in the `gencat' program of GNU libc. This mainly
597 includes the use of symbols for the message number and the automatic
598 generation of header files which contain the needed #defines to map the
599 symbols to integers.
601 Here is a simple SED script to convert at least some Linux specific catalog
602 files to the XPG4 form:
604 -----------------------------------------------------------------------
605 # Change catalog source in Linux specific format to standard XPG format.
606 # Ulrich Drepper <drepper@cygnus.com>, 1996.
607 #
608 /^\$ #/ {
609 h
610 s/\$ #\([^ ]*\).*/\1/
611 x
612 s/\$ #[^ ]* *\(.*\)/\$ \1/
613 }
615 /^# / {
616 s/^# \(.*\)/\1/
617 G
618 s/\(.*\)\n\(.*\)/\2 \1/
619 }
620 -----------------------------------------------------------------------
622 ?? Programs using libc have their messages translated, but other
623 behavior is not localized (e.g. collating order); why?
625 {ZW} Translated messages are automatically installed, but the locale
626 database that controls other behaviors is not. You need to run localedef to
627 install this database, after you have run `make install'. For example, to
628 set up the French Canadian locale, simply issue the command
630 localedef -i fr_CA -f ISO-8859-1 fr_CA
632 Please see localedata/README in the source tree for further details.
634 ?? I have set up /etc/nis.conf, and the Linux libc 5 with NYS
635 works great. But the glibc NIS+ doesn't seem to work.
637 {TK} The glibc NIS+ implementation uses a /var/nis/NIS_COLD_START file for
638 storing information about the NIS+ server and their public keys, because the
639 nis.conf file does not contain all the necessary information. You have to
640 copy a NIS_COLD_START file from a Solaris client (the NIS_COLD_START file is
641 byte order independent) or generate it with nisinit from the nis-tools
642 package; available at
644 http://www-vt.uni-paderborn.de/~kukuk/linux/nisplus.html
646 ?? I have killed ypbind to stop using NIS, but glibc
647 continues using NIS.
649 {TK} For faster NIS lookups, glibc uses the /var/yp/binding/ files from
650 ypbind. ypbind 3.3 and older versions don't always remove these files, so
651 glibc will continue to use them. Other BSD versions seem to work correctly.
652 Until ypbind 3.4 is released, you can find a patch at
654 ftp://ftp.kernel.org/pub/linux/utils/net/NIS/ypbind-3.3-glibc4.diff.gz
656 ?? Under Linux/Alpha, I always get "do_ypcall: clnt_call:
657 RPC: Unable to receive; errno = Connection refused" when using NIS.
659 {TK} You need a ypbind version which is 64bit clean. Some versions are not
660 64bit clean. A 64bit clean implementation is ypbind-mt. For ypbind 3.3,
661 you need the patch from ftp.kernel.org (See the previous question). I don't
662 know about other versions.
665 ?? After installing glibc name resolving doesn't work properly.
667 {AJ} You probably should read the manual section describing nsswitch.conf
668 (just type `info libc "NSS Configuration File"'). The NSS configuration
669 file is usually the culprit.
672 ?? How do I create the databases for NSS?
674 {AJ} If you have an entry "db" in /etc/nsswitch.conf you should also create
675 the database files. The glibc sources contain a Makefile which does the
676 necessary conversion and calls to create those files. The file is
677 `db-Makefile' in the subdirectory `nss' and you can call it with `make -f
678 db-Makefile'. Please note that not all services are capable of using a
679 database. Currently passwd, group, ethers, protocol, rpc, services shadow
680 and netgroup are implemented.
682 ?? I have /usr/include/net and /usr/include/scsi as symlinks
683 into my Linux source tree. Is that wrong?
685 {PB} This was necessary for libc5, but is not correct when using glibc.
686 Including the kernel header files directly in user programs usually does not
687 work (see ?kerhdr). glibc provides its own <net/*> and <scsi/*> header
688 files to replace them, and you may have to remove any symlink that you have
689 in place before you install glibc. However, /usr/include/asm and
690 /usr/include/linux should remain as they were.
692 ?? Programs like `logname', `top', `uptime' `users', `w' and
693 `who', show incorrect information about the (number of)
694 users on my system. Why?
696 {MK} See ?getlog.
698 ?? After upgrading to glibc 2.1 with symbol versioning I get
699 errors about undefined symbols. What went wrong?
701 {AJ} The problem is caused either by wrong program code or tools. In the
702 versioned libc a lot of symbols are now local that were global symbols in
703 previous versions. It seems that programs linked against older versions
704 often accidentally used libc global variables -- something that should not
705 happen.
707 The only way to fix this is to recompile your program. Sorry, that's the
708 price you might have to pay once for quite a number of advantages with
709 symbol versioning.
711 ?? When I start the program XXX after upgrading the library
712 I get
713 XXX: Symbol `_sys_errlist' has different size in shared
714 object, consider re-linking
715 Why? What should I do?
717 {UD} As the message says, relink the binary. The problem is that a few
718 symbols from the library can change in size and there is no way to avoid
719 this. _sys_errlist is a good example. Occasionally there are new error
720 numbers added to the kernel and this must be reflected at user level,
721 breaking programs that refer to them directly.
723 Such symbols should normally not be used at all. There are mechanisms to
724 avoid using them. In the case of _sys_errlist, there is the strerror()
725 function which should _always_ be used instead. So the correct fix is to
726 rewrite that part of the application.
728 In some situations (especially when testing a new library release) it might
729 be possible that a symbol changed size when that should not have happened.
730 So in case of doubt report such a warning message as a problem.
732 ?? What do I need for C++ development?
734 {HJ,AJ} You need either egcs 1.1 which comes directly with libstdc++ or
735 gcc-2.8.1 together with libstdc++ 2.8.1.1. egcs 1.1 has the better C++
736 support and works directly with glibc 2.1. If you use gcc-2.8.1 with
737 libstdc++ 2.8.1.1, you need to modify libstdc++ a bit. A patch is available
738 as:
739 ftp://alpha.gnu.org/gnu/libstdc++-2.8.1.1-glibc2.1-diff.gz
741 Please note that libg++ 2.7.2 (and the Linux Versions 2.7.2.x) doesn't work
742 very well with the GNU C library due to vtable thunks. If you're upgrading
743 from glibc 2.0.x to 2.1 you have to recompile libstdc++ since the library
744 compiled for 2.0 is not compatible due to the new Large File Support (LFS)
745 in version 2.1.
747 {UD} But since in the case of a shared libstdc++ the version numbers should
748 be different existing programs will continue to work.
750 ?? Even statically linked programs need some shared libraries
751 which is not acceptable for me. What can I do?
753 {AJ} NSS (for details just type `info libc "Name Service Switch"') won't
754 work properly without shared libraries. NSS allows using different services
755 (e.g. NIS, files, db, hesiod) by just changing one configuration file
756 (/etc/nsswitch.conf) without relinking any programs. The only disadvantage
757 is that now static libraries need to access shared libraries. This is
758 handled transparently by the GNU C library.
760 A solution is to configure glibc with --enable-static-nss. In this case you
761 can create a static binary that will use only the services dns and files
762 (change /etc/nsswitch.conf for this). You need to link explicitly against
763 all these services. For example:
765 gcc -static test-netdb.c -o test-netdb.c \
766 -lc -lnss_files -lnss_dns -lresolv
768 The problem with this approach is that you've got to link every static
769 program that uses NSS routines with all those libraries.
771 {UD} In fact, one cannot say anymore that a libc compiled with this
772 option is using NSS. There is no switch anymore. Therefore it is
773 *highly* recommended *not* to use --enable-static-nss since this makes
774 the behaviour of the programs on the system inconsistent.
776 ?? I just upgraded my Linux system to glibc and now I get
777 errors whenever I try to link any program.
779 {ZW} This happens when you have installed glibc as the primary C library but
780 have stray symbolic links pointing at your old C library. If the first
781 `libc.so' the linker finds is libc 5, it will use that. Your program
782 expects to be linked with glibc, so the link fails.
784 The most common case is that glibc put its `libc.so' in /usr/lib, but there
785 was a `libc.so' from libc 5 in /lib, which gets searched first. To fix the
786 problem, just delete /lib/libc.so. You may also need to delete other
787 symbolic links in /lib, such as /lib/libm.so if it points to libm.so.5.
789 {AJ} The perl script test-installation.pl which is run as last step during
790 an installation of glibc that is configured with --prefix=/usr should help
791 detect these situations. If the script reports problems, something is
792 really screwed up.
794 ?? When I use nscd the machine freezes.
796 {UD} You cannot use nscd with Linux 2.0.*. There is functionality missing
797 in the kernel and work-arounds are not suitable. Besides, some parts of the
798 kernel are too buggy when it comes to using threads.
800 If you need nscd, you have to use at least a 2.1 kernel.
802 Note that I have at this point no information about any other platform.
804 ?? I need lots of open files. What do I have to do?
806 {AJ} This is at first a kernel issue. The kernel defines limits with
807 OPEN_MAX the number of simultaneous open files and with FD_SETSIZE the
808 number of used file descriptors. You need to change these values in your
809 kernel and recompile the kernel so that the kernel allows to use more open
810 files. You don't necessarily need to recompile the GNU C library since the
811 only place where OPEN_MAX and FD_SETSIZE is really needed in the library
812 itself is the size of fd_set which is used by select.
814 The GNU C library is now (nearly) select free. This means it internally has
815 no limits imposed by the `fd_set' type. Instead almost all places where the
816 functionality is needed the `poll' function is used.
818 If you increase the number of file descriptors in the kernel you don't need
819 to recompile the C library. The remaining select calls are in the RPC code.
820 If your RPC daemons don't need more than FD_SETSIZE file descriptors, you
821 don't need to change anything at all.
823 {UD} You can always get the maximum number of file descriptors a process is
824 allowed to have open at any time using
826 number = sysconf (_SC_OPEN_MAX);
828 This will work even if the kernel limits change.
830 ?? How do I get the same behavior on parsing /etc/passwd and
831 /etc/group as I have with libc5 ?
833 {TK} The name switch setup in /etc/nsswitch.conf selected by most Linux
834 distributions does not support +/- and netgroup entries in the files like
835 /etc/passwd. Though this is the preferred setup some people might have
836 setups coming over from the libc5 days where it was the default to recognize
837 lines like this. To get back to the old behaviour one simply has to change
838 the rules for passwd, group, and shadow in the nsswitch.conf file as
839 follows:
841 passwd: compat
842 group: compat
843 shadow: compat
845 passwd_compat: nis
846 group_compat: nis
847 shadow_compat: nis
849 ??libs What needs to be recompiled when upgrading from glibc 2.0 to glibc
850 2.1?
852 {AJ,CG} If you just upgrade the glibc from 2.0.x (x <= 7) to 2.1, binaries
853 that have been linked against glibc 2.0 will continue to work.
855 If you compile your own binaries against glibc 2.1, you also need to
856 recompile some other libraries. The problem is that libio had to be
857 changed and therefore libraries that are based or depend on the libio
858 of glibc, e.g. ncurses or slang, need to be recompiled. If you
859 experience strange segmentation faults in your programs linked against
860 glibc 2.1, you might need to recompile your libraries.
862 Another problem is that older binaries that were linked statically against
863 glibc 2.0 will reference the older nss modules (libnss_files.so.1 instead of
864 libnss_files.so.2), so don't remove them. Also, the old glibc-2.0 compiled
865 static libraries (libfoo.a) which happen to depend on the older libio
866 behavior will be broken by the glibc 2.1 upgrade. We plan to produce a
867 compatibility library that people will be able to link in if they want
868 to compile a static library generated against glibc 2.0 into a program
869 on a glibc 2.1 system. You just add -lcompat and you should be fine.
871 The glibc-compat add-on will provide the libcompat.a library, the older
872 nss modules, and a few other files. Together, they should make it
873 possible to do development with old static libraries on a glibc 2.1
874 system. This add-on is still in development. You can get it from
875 ftp://alpha.gnu.org/gnu/glibc-compat-2.1.tar.gz
876 but please keep in mind that it is experimental.
878 ?? Why is extracting files via tar so slow?
880 {AJ} Extracting of tar archives might be quite slow since tar has to look up
881 userid and groupids and doesn't cache negative results. If you have nis or
882 nisplus in your /etc/nsswitch.conf for the passwd and/or group database,
883 each file extractions needs a network connection. There are two possible
884 solutions:
886 - do you really need NIS/NIS+ (some Linux distributions add by default
887 nis/nisplus even if it's not needed)? If not, just remove the entries.
889 - if you need NIS/NIS+, use the Name Service Cache Daemon nscd that comes
890 with glibc 2.1.
892 ?? Compiling programs I get parse errors in libio.h (e.g. "parse error
893 before `_IO_seekoff'"). How should I fix this?
895 {AJ} You might get the following errors when upgrading to glibc 2.1:
897 In file included from /usr/include/stdio.h:57,
898 from ...
899 /usr/include/libio.h:335: parse error before `_IO_seekoff'
900 /usr/include/libio.h:335: parse error before `_G_off64_t'
901 /usr/include/libio.h:336: parse error before `_IO_seekpos'
902 /usr/include/libio.h:336: parse error before `_G_fpos64_t'
904 The problem is a wrong _G_config.h file in your include path. The
905 _G_config.h file that comes with glibc 2.1 should be used and not one from
906 libc5 or from a compiler directory. To check which _G_config.h file the
907 compiler uses, compile your program with `gcc -E ...|grep G_config.h' and
908 remove that file. Your compiler should pick up the file that has been
909 installed by glibc 2.1 in your include directory.
911 ?? After upgrading to glibc 2.1, libraries that were compiled against
912 glibc 2.0.x don't work anymore.
914 {AJ} See ?libs.
917 ? Source and binary incompatibilities, and what to do about them
919 ?? I expect GNU libc to be 100% source code compatible with
920 the old Linux based GNU libc. Why isn't it like this?
922 {DMT,UD} Not every extension in Linux libc's history was well thought-out.
923 In fact it had a lot of problems with standards compliance and with
924 cleanliness. With the introduction of a new version number these errors can
925 now be corrected. Here is a list of the known source code
926 incompatibilities:
928 * _GNU_SOURCE: glibc does not make the GNU extensions available
929 automatically. If a program depends on GNU extensions or some
930 other non-standard functionality, it is necessary to compile it
931 with the C compiler option -D_GNU_SOURCE, or better, to put
932 `#define _GNU_SOURCE' at the beginning of your source files, before
933 any C library header files are included. This difference normally
934 manifests itself in the form of missing prototypes and/or data type
935 definitions. Thus, if you get such errors, the first thing you
936 should do is try defining _GNU_SOURCE and see if that makes the
937 problem go away.
939 For more information consult the file `NOTES' in the GNU C library
940 sources.
942 * reboot(): GNU libc sanitizes the interface of reboot() to be more
943 compatible with the interface used on other OSes. reboot() as
944 implemented in glibc takes just one argument. This argument
945 corresponds to the third argument of the Linux reboot system call.
946 That is, a call of the form reboot(a, b, c) needs to be changed into
947 reboot(c). Beside this the header <sys/reboot.h> defines the needed
948 constants for the argument. These RB_* constants should be used
949 instead of the cryptic magic numbers.
951 * swapon(): the interface of this function didn't change, but the
952 prototype is in a separate header file <sys/swap.h>. This header
953 file also provides the SWAP_* constants defined by <linux/swap.h>;
954 you should use them for the second argument to swapon().
956 * errno: If a program uses the variable "errno", then it _must_
957 include <errno.h>. The old libc often (erroneously) declared this
958 variable implicitly as a side-effect of including other libc header
959 files. glibc is careful to avoid such namespace pollution, which,
960 in turn, means that you really need to include the header files that
961 you depend on. This difference normally manifests itself in the
962 form of the compiler complaining about references to an undeclared
963 symbol "errno".
965 * Linux-specific syscalls: All Linux system calls now have appropriate
966 library wrappers and corresponding declarations in various header files.
967 This is because the syscall() macro that was traditionally used to
968 work around missing syscall wrappers are inherently non-portable and
969 error-prone. The following table lists all the new syscall stubs,
970 the header-file declaring their interface and the system call name.
972 syscall name: wrapper name: declaring header file:
973 ------------- ------------- ----------------------
974 bdflush bdflush <sys/kdaemon.h>
975 syslog ksyslog_ctl <sys/klog.h>
977 * lpd: Older versions of lpd depend on a routine called _validuser().
978 The library does not provide this function, but instead provides
979 __ivaliduser() which has a slightly different interface. Simply
980 upgrading to a newer lpd should fix this problem (e.g., the 4.4BSD
981 lpd is known to be working).
983 * resolver functions/BIND: like on many other systems the functions of
984 the resolver library are not included in libc itself. There is a
985 separate library libresolv. If you get undefined symbol errors for
986 symbols starting with `res_*' simply add -lresolv to your linker
987 command line.
989 * the `signal' function's behavior corresponds to the BSD semantic and
990 not the SysV semantic as it was in libc-5. The interface on all GNU
991 systems shall be the same and BSD is the semantic of choice. To use
992 the SysV behavior simply use `sysv_signal', or define _XOPEN_SOURCE.
993 See ?signal for details.
995 ??getlog Why does getlogin() always return NULL on my Linux box?
997 {UD} The GNU C library has a format for the UTMP and WTMP file which differs
998 from what your system currently has. It was extended to fulfill the needs
999 of the next years when IPv6 is introduced. The record size is different and
1000 some fields have different positions. The files written by functions from
1001 the one library cannot be read by functions from the other library. Sorry,
1002 but this is what a major release is for. It's better to have a cut now than
1003 having no means to support the new techniques later.
1005 {MK} There is however a (partial) solution for this problem. Please take a
1006 look at the file `login/README.utmpd'.
1008 ?? Where are the DST_* constants found in <sys/time.h> on many
1009 systems?
1011 {UD} These constants come from the old BSD days and are not used anymore
1012 (libc5 does not actually implement the handling although the constants are
1013 defined).
1015 Instead GNU libc contains zone database support and compatibility code for
1016 POSIX TZ environment variable handling. For former is very much preferred
1017 (see ?tzdb).
1019 ?? The prototypes for `connect', `accept', `getsockopt',
1020 `setsockopt', `getsockname', `getpeername', `send',
1021 `sendto', and `recvfrom' are different in GNU libc from
1022 any other system I saw. This is a bug, isn't it?
1024 {UD} No, this is no bug. This version of GNU libc already follows the new
1025 Single Unix specifications (and I think the POSIX.1g draft which adopted the
1026 solution). The type for a parameter describing a size is now `socklen_t', a
1027 new type.
1029 ??kerhdr On Linux I've got problems with the declarations in Linux
1030 kernel headers.
1032 {UD,AJ} On Linux, the use of kernel headers is reduced to the minimum. This
1033 gives Linus the ability to change the headers more freely. Also, user
1034 programs are now insulated from changes in the size of kernel data
1035 structures.
1037 For example, the sigset_t type is 32 or 64 bits wide in the kernel. In
1038 glibc it is 1024 bits wide. This guarantees that when the kernel gets a
1039 bigger sigset_t (for POSIX.1e realtime support, say) user programs will not
1040 have to be recompiled. Consult the header files for more information about
1041 the changes.
1043 Therefore you shouldn't include Linux kernel header files directly if glibc
1044 has defined a replacement. Otherwise you might get undefined results because
1045 of type conflicts.
1047 ?? I don't include any kernel headers myself but the compiler
1048 still complains about redeclarations of types in the kernel
1049 headers.
1051 {UD} The kernel headers before Linux 2.1.61 and 2.0.32 don't work correctly
1052 with glibc. Compiling C programs is possible in most cases but C++ programs
1053 have (due to the change of the name lookups for `struct's) problems. One
1054 prominent example is `struct fd_set'.
1056 There might be some problems left but 2.1.61/2.0.32 fix most of the known
1057 ones. See the BUGS file for other known problems.
1059 ??signal Why don't signals interrupt system calls anymore?
1061 {ZW} By default GNU libc uses the BSD semantics for signal(), unlike Linux
1062 libc 5 which used System V semantics. This is partially for compatibility
1063 with other systems and partially because the BSD semantics tend to make
1064 programming with signals easier.
1066 There are three differences:
1068 * BSD-style signals that occur in the middle of a system call do not
1069 affect the system call; System V signals cause the system call to
1070 fail and set errno to EINTR.
1072 * BSD signal handlers remain installed once triggered. System V signal
1073 handlers work only once, so one must reinstall them each time.
1075 * A BSD signal is blocked during the execution of its handler. In other
1076 words, a handler for SIGCHLD (for example) does not need to worry about
1077 being interrupted by another SIGCHLD. It may, however, be interrupted
1078 by other signals.
1080 There is general consensus that for `casual' programming with signals, the
1081 BSD semantics are preferable. You don't need to worry about system calls
1082 returning EINTR, and you don't need to worry about the race conditions
1083 associated with one-shot signal handlers.
1085 If you are porting an old program that relies on the old semantics, you can
1086 quickly fix the problem by changing signal() to sysv_signal() throughout.
1087 Alternatively, define _XOPEN_SOURCE before including <signal.h>.
1089 For new programs, the sigaction() function allows you to specify precisely
1090 how you want your signals to behave. All three differences listed above are
1091 individually switchable on a per-signal basis with this function.
1093 If all you want is for one specific signal to cause system calls to fail and
1094 return EINTR (for example, to implement a timeout) you can do this with
1095 siginterrupt().
1098 ??string I've got errors compiling code that uses certain string
1099 functions. Why?
1101 {AJ} glibc 2.1 has special string functions that are faster than the normal
1102 library functions. Some of the functions are additionally implemented as
1103 inline functions and others as macros. This might lead to problems with
1104 existing codes but it is explicitly allowed by ISO C.
1106 The optimized string functions are only used when compiling with
1107 optimizations (-O1 or higher). The behavior can be changed with two feature
1108 macros:
1110 * __NO_STRING_INLINES: Don't do any string optimizations.
1111 * __USE_STRING_INLINES: Use assembly language inline functions (might
1112 increase code size dramatically).
1114 Since some of these string functions are now additionally defined as macros,
1115 code like "char *strncpy();" doesn't work anymore (and is unnecessary, since
1116 <string.h> has the necessary declarations). Either change your code or
1117 define __NO_STRING_INLINES.
1119 {UD} Another problem in this area is that gcc still has problems on machines
1120 with very few registers (e.g., ix86). The inline assembler code can require
1121 almost all the registers and the register allocator cannot always handle
1122 this situation.
1124 One can disable the string optimizations selectively. Instead of writing
1126 cp = strcpy (foo, "lkj");
1128 one can write
1130 cp = (strcpy) (foo, "lkj");
1132 This disables the optimization for that specific call.
1134 ?? I get compiler messages "Initializer element not constant" with
1135 stdin/stdout/stderr. Why?
1137 {RM,AJ} Constructs like:
1138 static FILE *InPtr = stdin;
1140 lead to this message. This is correct behaviour with glibc since stdin is
1141 not a constant expression. Please note that a strict reading of ISO C does
1142 not allow above constructs.
1144 One of the advantages of this is that you can assign to stdin, stdout, and
1145 stderr just like any other global variable (e.g. `stdout = my_stream;'),
1146 which can be very useful with custom streams that you can write with libio
1147 (but beware this is not necessarily portable). The reason to implement it
1148 this way were versioning problems with the size of the FILE structure.
1150 To fix those programs you've got to initialize the variable at run time.
1151 This can be done, e.g. in main, like:
1153 static FILE *InPtr;
1154 int main(void)
1155 {
1156 InPtr = stdin;
1157 }
1159 or by constructors (beware this is gcc specific):
1161 static FILE *InPtr;
1162 static void inPtr_construct (void) __attribute__((constructor));
1163 static void inPtr_construct (void) { InPtr = stdin; }
1166 ?? I can't compile with gcc -traditional (or
1167 -traditional-cpp). Why?
1169 {AJ} glibc2 does break -traditional and -traditonal-cpp - and will continue
1170 to do so. For example constructs of the form:
1172 enum {foo
1173 #define foo foo
1174 }
1176 are useful for debugging purposes (you can use foo with your debugger that's
1177 why we need the enum) and for compatibility (other systems use defines and
1178 check with #ifdef).
1180 ?? I get some errors with `gcc -ansi'. Isn't glibc ANSI compatible?
1182 {AJ} The GNU C library is compatible with the ANSI/ISO C standard. If
1183 you're using `gcc -ansi', the glibc includes which are specified in the
1184 standard follow the standard. The ANSI/ISO C standard defines what has to be
1185 in the include files - and also states that nothing else should be in the
1186 include files (btw. you can still enable additional standards with feature
1187 flags).
1189 The GNU C library is conforming to ANSI/ISO C - if and only if you're only
1190 using the headers and library functions defined in the standard.
1192 ?? I can't access some functions anymore. nm shows that they do
1193 exist but linking fails nevertheless.
1195 {AJ} With the introduction of versioning in glibc 2.1 it is possible to
1196 export only those identifiers (functions, variables) that are really needed
1197 by application programs and by other parts of glibc. This way a lot of
1198 internal interfaces are now hidden. nm will still show those identifiers
1199 but marking them as internal. ISO C states that identifiers beginning with
1200 an underscore are internal to the libc. An application program normally
1201 shouldn't use those internal interfaces (there are exceptions,
1202 e.g. __ivaliduser). If a program uses these interfaces, it's broken. These
1203 internal interfaces might change between glibc releases or dropped
1204 completely.
1206 ?? When using the db-2 library which comes with glibc is used in
1207 the Perl db modules the testsuite is not passed. This did not
1208 happen with db-1, gdbm, or ndbm.
1210 {UD} You are using an outdated copy of the DB_File Perl module. In fact db-2
1211 finally removed the handling of zero-sized keys which was one of the features
1212 tested by the old Perl testsuite and therefore you see an error. But this
1213 never was documented and guaranteed, only broken programs used this feature.
1215 Consequently db-2 does not need to support this feature and instead signals
1216 an error which leads to easier debugging. The DB_File module maintainer
1217 Paul Marquess <pmarquess@bfsec.bt.co.uk> acknowledged this change and fixed
1218 the testsuite so that if you use DB_File v1.60 or later you should not have
1219 any more problems with db-2.
1221 ?? The pow() inline function I get when including <math.h> is broken.
1222 I get segmentation faults when I run the program.
1224 {UD} Nope, the implementation is correct. The problem is with egcs version
1225 prior to 1.1. I.e., egcs 1.0 to 1.0.3 are all broken (at least on Intel).
1226 If you have to use this compiler you must define __NO_MATH_INLINES before
1227 including <math.h> to prevent the inline functions from being used. egcs 1.1
1228 fixes the problem. I don't know about gcc 2.8 and 2.8.1.
1230 ?? The sys/sem.h file lacks the definition of `union semun'.
1232 {UD} Nope. This union has to be provided by the user program. Former glibc
1233 versions defined this but it was an error since it does not make much sense
1234 when thinking about it. The standards describing the System V IPC functions
1235 define it this way and therefore programs must be adopted.
1237 ?? Why has <netinet/ip_fw.h> disappeared?
1239 {AJ} The corresponding Linux kernel data structures and constants are
1240 totally different in Linux 2.0 and Linux 2.2. This situation has to be
1241 taken care in user programs using the firewall structures and therefore
1242 those programs (ipfw is AFAIK the only one) should deal with this problem
1243 themselves.
1245 ?? I get floods of warnings when I use -Wconversion and include
1246 <string.h> or <math.h>.
1248 {ZW} <string.h> and <math.h> intentionally use prototypes to override
1249 argument promotion. -Wconversion warns about all these. You can safely
1250 ignore the warnings.
1252 -Wconversion isn't really intended for production use, only for shakedown
1253 compiles after converting an old program to standard C.
1256 ?? After upgrading to glibc 2.1, I receive errors about
1257 unresolved symbols, like `_dl_initial_searchlist' and can not
1258 execute any binaries. What went wrong?
1260 {AJ} This normally happens if your libc and ld (dynamic linker) are from
1261 different releases of glibc. For example, the dynamic linker
1262 /lib/ld-linux.so.2 comes from glibc 2.0.x, but the version of libc.so.6 is
1263 from glibc 2.1.
1265 The path /lib/ld-linux.so.2 is hardcoded in every glibc2 binary but
1266 libc.so.6 is searched via /etc/ld.so.cache and in some special directories
1267 like /lib and /usr/lib. If you run configure with another prefix than /usr
1268 and put this prefix before /lib in /etc/ld.so.conf, your system will break.
1270 So what can you do? Either of the following should work:
1272 * Run `configure' with the same prefix argument you've used for glibc 2.0.x
1273 so that the same paths are used.
1274 * Replace /lib/ld-linux.so.2 with a link to the dynamic linker from glibc
1275 2.1.
1277 You can even call the dynamic linker by hand if everything fails. You've
1278 got to set LD_LIBRARY_PATH so that the corresponding libc is found and also
1279 need to provide an absolute path to your binary:
1281 LD_LIBRARY_PATH=<path-where-libc.so.6-lives> \
1282 <path-where-corresponding-dynamic-linker-lives>/ld-linux.so.2 \
1283 <path-to-binary>/binary
1285 For example `LD_LIBRARY_PATH=/libold /libold/ld-linux.so.2 /bin/mv ...'
1286 might be useful in fixing a broken system (if /libold contains dynamic
1287 linker and corresponding libc).
1289 With that command line no path is used. To further debug problems with the
1290 dynamic linker, use the LD_DEBUG environment variable, e.g.
1291 `LD_DEBUG=help echo' for the help text.
1293 If you just want to test this release, don't put the lib directory in
1294 /etc/ld.so.conf. You can call programs directly with full paths (as above).
1295 When compiling new programs against glibc 2.1, you've got to specify the
1296 correct paths to the compiler (option -I with gcc) and linker (options
1297 --dynamic-linker, -L and --rpath).
1299 ?? bonnie reports that char i/o with glibc 2 is much slower than with
1300 libc5. What can be done?
1302 {AJ} The GNU C library uses thread safe functions by default and libc5 used
1303 non thread safe versions. The non thread safe functions have in glibc the
1304 suffix `_unlocked', for details check <stdio.h>. Using `putc_unlocked' etc.
1305 instead of `putc' should give nearly the same speed with bonnie (bonnie is a
1306 benchmark program for measuring disk access).
1308 ?? Programs compiled with glibc 2.1 can't read db files made with glibc
1309 2.0. What has changed that programs like rpm break?
1311 {AJ} The GNU C library 2.1 uses db2 instead of db1 which was used in version
1312 2.0. The internal formats of the actual db files are different. To convert
1313 the db files from db1 format to db2 format, you can use the programs
1314 `db_dump185' and `db_load'. Alternativly programs can be linked with db1
1315 using `-ldb1' instead of linking with db2 which uses `-ldb'. Linking with
1316 db1 might be preferable if older programs need to access the db file.
1318 db2 supports the old db1 programming interface and also a new programming
1319 interface. For compilation with the old API, <db_185.h> has to be included
1320 (and not <db.h>) and you can link with either `-ldb1' or `-ldb' for either
1321 of the db formats.
1324 ? Miscellaneous
1326 ?? After I changed configure.in I get `Autoconf version X.Y.
1327 or higher is required for this script'. What can I do?
1329 {UD} You have to get the specified autoconf version (or a later one)
1330 from your favorite mirror of ftp.gnu.org.
1332 ?? When I try to compile code which uses IPv6 headers and
1333 definitions on my Linux 2.x.y system I am in trouble.
1334 Nothing seems to work.
1336 {UD} The problem is that IPv6 development still has not reached a point
1337 where the headers are stable. There are still lots of incompatible changes
1338 made and the libc headers have to follow.
1340 {PB} The 2.1 release of GNU libc aims to comply with the current versions of
1341 all the relevant standards. The IPv6 support libraries for older Linux
1342 systems used a different naming convention and so code written to work with
1343 them may need to be modified. If the standards make incompatible changes in
1344 the future then the libc may need to change again.
1346 IPv6 will not work with a 2.0.x kernel. When kernel 2.2 is released it
1347 should contain all the necessary support; until then you should use the
1348 latest 2.1.x release you can find. As of 98/11/26 the currently recommended
1349 kernel for IPv6 is 2.1.129.
1351 Also, as of the 2.1 release the IPv6 API provided by GNU libc is not
1352 100% complete. In particular the getipnodebyname and getipnodebyaddr
1353 functions are not implemented.
1355 ??tzdb When I set the timezone by setting the TZ environment variable
1356 to EST5EDT things go wrong since glibc computes the wrong time
1357 from this information.
1359 {UD} The problem is that people still use the braindamaged POSIX method to
1360 select the timezone using the TZ environment variable with a format EST5EDT
1361 or whatever. People, if you insist on using TZ instead of the timezone
1362 database (see below), read the POSIX standard, the implemented behaviour is
1363 correct! What you see is in fact the result of the decisions made while
1364 POSIX.1 was created. We've only implemented the handling of TZ this way to
1365 be POSIX compliant. It is not really meant to be used.
1367 The alternative approach to handle timezones which is implemented is the
1368 correct one to use: use the timezone database. This avoids all the problems
1369 the POSIX method has plus it is much easier to use. Simply run the tzselect
1370 shell script, answer the question and use the name printed in the end by
1371 making a symlink /etc/localtime pointing to /usr/share/zoneinfo/NAME (NAME
1372 is the returned value from tzselect). That's all. You never again have to
1373 worry.
1375 So, please avoid sending bug reports about time related problems if you use
1376 the POSIX method and you have not verified something is really broken by
1377 reading the POSIX standards.
1379 ?? What other sources of documentation about glibc are available?
1381 {AJ} The FSF has a page about the GNU C library at
1382 <http://www.gnu.org/software/libc/>. The problem data base of open and
1383 solved bugs in GNU libc is available at
1384 <http://www-gnats.gnu.org:8080/cgi-bin/wwwgnats.pl>. Eric Green has written
1385 a HowTo for converting from Linux libc5 to glibc2. The HowTo is accessable
1386 via the FSF page and at <http://www.imaxx.net/~thrytis/glibc>. Frodo
1387 Looijaard describes a different way installing glibc2 as secondary libc at
1388 <http://huizen.dds.nl/~frodol/glibc>.
1390 Please note that this is not a complete list.
1392 ?? The timezone string for Sydney/Australia is wrong since even when
1393 daylight saving time is in effect the timezone string is EST.
1395 {UD} The problem for some timezones is that the local authorities decided
1396 to use the term "summer time" instead of "daylight saving time". In this
1397 case the abbreviation character `S' is the same as the standard one. So,
1398 for Sydney we have
1400 Eastern Standard Time = EST
1401 Eastern Summer Time = EST
1403 Great! To get this bug fixed convince the authorities to change the laws
1404 and regulations of the country this effects. glibc behaves correctly.
1406 ??make I've build make 3.77 against glibc 2.1 and now make gets
1407 segmentation faults.
1409 {AJ} GNU make 3.77 has support for 64 bit filesystems which is slightly
1410 broken (and one of the new features in the GNU C library 2.1 is 64 bit
1411 filesystem support :-( ). To get a working make you can use either make
1412 3.75 or patch 3.77. A working patch is available via RedHat's Rawhide server
1413 (ftp://rawhide.redhat.com/SRPMS/SRPMS/make-3.77-*src.rpm).
1415 \f
1416 Answers were given by:
1417 {UD} Ulrich Drepper, <drepper@cygnus.com>
1418 {DMT} David Mosberger-Tang, <davidm@AZStarNet.com>
1419 {RM} Roland McGrath, <roland@gnu.org>
1420 {AJ} Andreas Jaeger, <aj@arthur.rhein-neckar.de>
1421 {EY} Eric Youngdale, <eric@andante.jic.com>
1422 {PB} Phil Blundell, <Philip.Blundell@pobox.com>
1423 {MK} Mark Kettenis, <kettenis@phys.uva.nl>
1424 {ZW} Zack Weinberg, <zack@rabi.phys.columbia.edu>
1425 {TK} Thorsten Kukuk, <kukuk@vt.uni-paderborn.de>
1426 {GK} Geoffrey Keating, <geoffk@ozemail.com.au>
1427 {HJ} H.J. Lu, <hjl@gnu.org>
1428 {CG} Cristian Gafton, <gafton@redhat.com>
1429 \f
1430 Local Variables:
1431 mode:outline
1432 outline-regexp:"\\?"
1433 fill-column:76
1434 End: