]>
Commit | Line | Data |
---|---|---|
61952351 | 1 | Frequently Asked Questions about the GNU C Library |
f8cac037 | 2 | |
f12944ec UD |
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. | |
f8cac037 | 6 | |
f12944ec | 7 | The GNU C library is very complex. The installation process has not been |
fdacb17d | 8 | completely automated; there are too many variables. You can do substantial |
f12944ec UD |
9 | damage to your system by installing the library incorrectly. Make sure you |
10 | understand what you are undertaking before you begin. | |
f8cac037 | 11 | |
41f27456 RM |
12 | If you have any questions you think should be answered in this document, |
13 | please let me know. | |
f8cac037 RM |
14 | |
15 | --drepper@cygnus.com | |
16 | \f | |
61952351 UD |
17 | ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ |
18 | ||
19 | 1. Compiling glibc | |
20 | ||
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. | |
24 | What's wrong? | |
5edb9387 | 25 | 1.4. Do I need a special linker or assembler? |
8619129f | 26 | 1.5. Which compiler should I use for powerpc? |
348ed515 UD |
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 | |
f12944ec | 31 | wrong? |
348ed515 | 32 | 1.10. When I run `nm -u libc.so' on the produced library I still |
61952351 | 33 | find unresolved symbols. Can this be ok? |
348ed515 UD |
34 | 1.11. What are these `add-ons'? |
35 | 1.12. My XXX kernel emulates a floating-point coprocessor for me. | |
61952351 | 36 | Should I enable --with-fp? |
348ed515 | 37 | 1.13. When compiling GNU libc I get lots of errors saying functions |
61952351 | 38 | in glibc are duplicated in libgcc. |
348ed515 | 39 | 1.14. Why do I get messages about missing thread functions when I use |
a35cb74d | 40 | librt? I don't even use threads. |
348ed515 | 41 | 1.15. What's the problem with configure --enable-omitfp? |
b1418d8f | 42 | 1.16. I get failures during `make check'. What should I do? |
348ed515 | 43 | 1.17. What is symbol versioning good for? Do I need it? |
c7f7281e UD |
44 | 1.18. How can I compile on my fast ix86 machine a working libc for my slow |
45 | i386? After installing libc, programs abort with "Illegal | |
46 | Instruction". | |
150ae521 UD |
47 | 1.19. `make' complains about a missing dlfcn/libdl.so when building |
48 | malloc/libmemprof.so. How can I fix this? | |
61952351 UD |
49 | |
50 | 2. Installation and configuration issues | |
51 | ||
52 | 2.1. Can I replace the libc on my Linux system with GNU libc? | |
53 | 2.2. How do I configure GNU libc so that the essential libraries | |
54 | like libc.so go into /lib and the other into /usr/lib? | |
55 | 2.3. How should I avoid damaging my system when I install GNU libc? | |
56 | 2.4. Do I need to use GNU CC to compile programs that will use the | |
57 | GNU C Library? | |
58 | 2.5. When linking with the new libc I get unresolved symbols | |
59 | `crypt' and `setkey'. Why aren't these functions in the | |
60 | libc anymore? | |
b669ab02 UD |
61 | |
62 | ||
63 | Removed. Does not apply anymore. | |
61952351 UD |
64 | 2.6. When I use GNU libc on my Linux system by linking against |
65 | the libc.so which comes with glibc all I get is a core dump. | |
66 | 2.7. Looking through the shared libc file I haven't found the | |
67 | functions `stat', `lstat', `fstat', and `mknod' and while | |
68 | linking on my Linux system I get error messages. How is | |
69 | this supposed to work? | |
5edb9387 UD |
70 | 2.8. When I run an executable on one system which I compiled on |
71 | another, I get dynamic linker errors. Both systems have the same | |
72 | version of glibc installed. What's wrong? | |
73 | 2.9. How can I compile gcc 2.7.2.1 from the gcc source code using | |
61952351 | 74 | glibc 2.x? |
5edb9387 | 75 | 2.10. The `gencat' utility cannot process the catalog sources which |
61952351 | 76 | were used on my Linux libc5 based system. Why? |
5edb9387 | 77 | 2.11. Programs using libc have their messages translated, but other |
a35cb74d | 78 | behavior is not localized (e.g. collating order); why? |
5edb9387 | 79 | 2.12. I have set up /etc/nis.conf, and the Linux libc 5 with NYS |
61952351 | 80 | works great. But the glibc NIS+ doesn't seem to work. |
5edb9387 | 81 | 2.13. I have killed ypbind to stop using NIS, but glibc |
3dcf8ea6 | 82 | continues using NIS. |
5edb9387 | 83 | 2.14. Under Linux/Alpha, I always get "do_ypcall: clnt_call: |
3dcf8ea6 | 84 | RPC: Unable to receive; errno = Connection refused" when using NIS. |
5edb9387 UD |
85 | 2.15. After installing glibc name resolving doesn't work properly. |
86 | 2.16. How do I create the databases for NSS? | |
87 | 2.17. I have /usr/include/net and /usr/include/scsi as symlinks | |
61952351 | 88 | into my Linux source tree. Is that wrong? |
5edb9387 | 89 | 2.18. Programs like `logname', `top', `uptime' `users', `w' and |
61952351 UD |
90 | `who', show incorrect information about the (number of) |
91 | users on my system. Why? | |
5edb9387 | 92 | 2.19. After upgrading to glibc 2.1 with symbol versioning I get |
61952351 | 93 | errors about undefined symbols. What went wrong? |
5edb9387 | 94 | 2.20. When I start the program XXX after upgrading the library |
61952351 UD |
95 | I get |
96 | XXX: Symbol `_sys_errlist' has different size in shared | |
97 | object, consider re-linking | |
98 | Why? What should I do? | |
5edb9387 UD |
99 | 2.21. What do I need for C++ development? |
100 | 2.22. Even statically linked programs need some shared libraries | |
ff44f2a5 | 101 | which is not acceptable for me. What can I do? |
5edb9387 | 102 | 2.23. I just upgraded my Linux system to glibc and now I get |
fdacb17d | 103 | errors whenever I try to link any program. |
5edb9387 | 104 | 2.24. When I use nscd the machine freezes. |
0155a773 | 105 | 2.25. I need lots of open files. What do I have to do? |
7db169c9 UD |
106 | 2.26. How do I get the same behavior on parsing /etc/passwd and |
107 | /etc/group as I have with libc5 ? | |
b710a6e2 UD |
108 | 2.27. What needs to be recompiled when upgrading from glibc 2.0 to glibc |
109 | 2.1? | |
b7398be5 | 110 | 2.28. Why is extracting files via tar so slow? |
2ee511d9 UD |
111 | 2.29. Compiling programs I get parse errors in libio.h (e.g. "parse error |
112 | before `_IO_seekoff'"). How should I fix this? | |
4f7ea427 UD |
113 | 2.30. After upgrading to glibc 2.1, libraries that were compiled against |
114 | glibc 2.0.x don't work anymore. | |
9de4e203 UD |
115 | 2.31. What happened to the Berkeley DB libraries? Can I still use db |
116 | in /etc/nsswitch.conf? | |
61952351 UD |
117 | |
118 | 3. Source and binary incompatibilities, and what to do about them | |
119 | ||
120 | 3.1. I expect GNU libc to be 100% source code compatible with | |
121 | the old Linux based GNU libc. Why isn't it like this? | |
122 | 3.2. Why does getlogin() always return NULL on my Linux box? | |
123 | 3.3. Where are the DST_* constants found in <sys/time.h> on many | |
124 | systems? | |
125 | 3.4. The prototypes for `connect', `accept', `getsockopt', | |
126 | `setsockopt', `getsockname', `getpeername', `send', | |
127 | `sendto', and `recvfrom' are different in GNU libc from | |
128 | any other system I saw. This is a bug, isn't it? | |
129 | 3.5. On Linux I've got problems with the declarations in Linux | |
130 | kernel headers. | |
131 | 3.6. I don't include any kernel headers myself but the compiler | |
132 | still complains about redeclarations of types in the kernel | |
133 | headers. | |
134 | 3.7. Why don't signals interrupt system calls anymore? | |
135 | 3.8. I've got errors compiling code that uses certain string | |
136 | functions. Why? | |
4775243a UD |
137 | 3.9. I get compiler messages "Initializer element not constant" with |
138 | stdin/stdout/stderr. Why? | |
139 | 3.10. I can't compile with gcc -traditional (or | |
140 | -traditional-cpp). Why? | |
141 | 3.11. I get some errors with `gcc -ansi'. Isn't glibc ANSI compatible? | |
a35cb74d UD |
142 | 3.12. I can't access some functions anymore. nm shows that they do |
143 | exist but linking fails nevertheless. | |
a5f4e34a UD |
144 | 3.13. When using the db-2 library which comes with glibc is used in |
145 | the Perl db modules the testsuite is not passed. This did not | |
146 | happen with db-1, gdbm, or ndbm. | |
9de4e203 UD |
147 | |
148 | Removed. Does not apply anymore. | |
5148d49f UD |
149 | 3.14. The pow() inline function I get when including <math.h> is broken. |
150 | I get segmentation faults when I run the program. | |
05f732b3 | 151 | 3.15. The sys/sem.h file lacks the definition of `union semun'. |
33127459 | 152 | 3.16. Why has <netinet/ip_fw.h> disappeared? |
28ab8526 UD |
153 | 3.17. I get floods of warnings when I use -Wconversion and include |
154 | <string.h> or <math.h>. | |
5ff1a70a UD |
155 | 3.18. After upgrading to glibc 2.1, I receive errors about |
156 | unresolved symbols, like `_dl_initial_searchlist' and can not | |
157 | execute any binaries. What went wrong? | |
7d1de115 UD |
158 | 3.19. bonnie reports that char i/o with glibc 2 is much slower than with |
159 | libc5. What can be done? | |
b93492aa UD |
160 | 3.20. Programs compiled with glibc 2.1 can't read db files made with glibc |
161 | 2.0. What has changed that programs like rpm break? | |
9de4e203 UD |
162 | |
163 | Removed. Does not apply anymore. | |
b5a9efcd UD |
164 | 3.21. Autoconf's AC_CHECK_FUNC macro reports that a function exists, but |
165 | when I try to use it, it always returns -1 and sets errno to ENOSYS. | |
166 | 3.22. My program segfaults when I call fclose() on the FILE* returned | |
167 | from setmntent(). Is this a glibc bug? | |
61952351 | 168 | |
7d1de115 | 169 | 4. Miscellaneous |
61952351 | 170 | |
7d1de115 | 171 | 4.1. After I changed configure.in I get `Autoconf version X.Y. |
61952351 | 172 | or higher is required for this script'. What can I do? |
7d1de115 | 173 | 4.2. When I try to compile code which uses IPv6 headers and |
61952351 UD |
174 | definitions on my Linux 2.x.y system I am in trouble. |
175 | Nothing seems to work. | |
7d1de115 | 176 | 4.3. When I set the timezone by setting the TZ environment variable |
ff44f2a5 UD |
177 | to EST5EDT things go wrong since glibc computes the wrong time |
178 | from this information. | |
7d1de115 UD |
179 | 4.4. What other sources of documentation about glibc are available? |
180 | 4.5. The timezone string for Sydney/Australia is wrong since even when | |
348ed515 | 181 | daylight saving time is in effect the timezone string is EST. |
7d1de115 | 182 | 4.6. I've build make 3.77 against glibc 2.1 and now make gets |
eeabe877 | 183 | segmentation faults. |
2bbc70d5 AJ |
184 | |
185 | Removed. Does not apply anymore, use make 3.79 or newer. | |
c63598bf | 186 | 4.7. Why do so many programs using math functions fail on my AlphaStation? |
8892c471 UD |
187 | 4.8. The conversion table for character set XX does not match with |
188 | what I expect. | |
f8cac037 | 189 | |
61952351 UD |
190 | \f |
191 | ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ | |
f4017d20 | 192 | |
61952351 | 193 | 1. Compiling glibc |
04be94a8 | 194 | |
61952351 | 195 | 1.1. What systems does the GNU C Library run on? |
613a76ff | 196 | |
f12944ec UD |
197 | {UD} This is difficult to answer. The file `README' lists the architectures |
198 | GNU libc was known to run on *at some time*. This does not mean that it | |
199 | still can be compiled and run on them now. | |
f8cac037 | 200 | |
f12944ec UD |
201 | The systems glibc is known to work on as of this release, and most probably |
202 | in the future, are: | |
f8cac037 RM |
203 | |
204 | *-*-gnu GNU Hurd | |
4775243a UD |
205 | i[3456]86-*-linux-gnu Linux-2.x on Intel |
206 | m68k-*-linux-gnu Linux-2.x on Motorola 680x0 | |
2bbc70d5 | 207 | alpha*-*-linux-gnu Linux-2.x on DEC Alpha |
9a0a462c | 208 | powerpc-*-linux-gnu Linux and MkLinux on PowerPC systems |
4775243a UD |
209 | sparc-*-linux-gnu Linux-2.x on SPARC |
210 | sparc64-*-linux-gnu Linux-2.x on UltraSPARC | |
ff44f2a5 | 211 | arm-*-none ARM standalone systems |
348ed515 | 212 | arm-*-linux Linux-2.x on ARM |
ff44f2a5 | 213 | arm-*-linuxaout Linux-2.x on ARM using a.out binaries |
2bbc70d5 AJ |
214 | mips*-*-linux-gnu Linux-2.x on MIPS |
215 | ia64-*-linux-gnu Linux-2.x on ia64 | |
f8cac037 | 216 | |
f12944ec UD |
217 | Ports to other Linux platforms are in development, and may in fact work |
218 | already, but no one has sent us success reports for them. Currently no | |
219 | ports to other operating systems are underway, although a few people have | |
220 | expressed interest. | |
f8cac037 | 221 | |
f12944ec UD |
222 | If you have a system not listed above (or in the `README' file) and you are |
223 | really interested in porting it, contact | |
f8cac037 | 224 | |
4775243a | 225 | <bug-glibc@gnu.org> |
f8cac037 RM |
226 | |
227 | ||
61952351 | 228 | 1.2. What compiler do I need to build GNU libc? |
f8cac037 | 229 | |
f12944ec UD |
230 | {UD} You must use GNU CC to compile GNU libc. A lot of extensions of GNU CC |
231 | are used to increase portability and speed. | |
f8cac037 | 232 | |
61952351 | 233 | GNU CC is found, like all other GNU packages, on |
f12944ec | 234 | |
a35cb74d | 235 | ftp://ftp.gnu.org/pub/gnu |
f12944ec | 236 | |
a35cb74d | 237 | and the many mirror sites. ftp.gnu.org is always overloaded, so try to find |
61952351 | 238 | a local mirror first. |
f8cac037 | 239 | |
b0610668 | 240 | You should always try to use the latest official release. Older versions |
f12944ec | 241 | may not have all the features GNU libc requires. The current releases of |
b8f558b7 UD |
242 | egcs (1.0.3 and 1.1.1) should work with the GNU C library (for powerpc see |
243 | question 1.5; for ARM see question 1.6). | |
f8cac037 | 244 | |
95f7cecb UD |
245 | While the GNU CC should be able to compile glibc it is nevertheless adviced |
246 | to use EGCS. Comparing the sizes of glibc on Intel compiled with a recent | |
247 | EGCS and gcc 2.8.1 shows this: | |
248 | ||
249 | text data bss dec hex filename | |
66f6a52b UD |
250 | egcs-2.93.10 862897 15944 12824 891665 d9b11 libc.so |
251 | gcc-2.8.1 959965 16468 12152 988585 f15a9 libc.so | |
95f7cecb UD |
252 | |
253 | Make up your own decision. | |
5edb9387 | 254 | |
83f6a990 UD |
255 | GNU CC versions 2.95 and above are derived from egcs, and they may do even |
256 | better. | |
257 | ||
6e8afc1c | 258 | Please note that gcc 2.95 and 2.95.x cannot compile glibc on Alpha due to |
bd952512 UD |
259 | problems in the complex float support. |
260 | ||
f8cac037 | 261 | |
61952351 UD |
262 | 1.3. When I try to compile glibc I get only error messages. |
263 | What's wrong? | |
f8cac037 | 264 | |
b1418d8f | 265 | {UD} You definitely need GNU make to build GNU libc. No other make |
f12944ec | 266 | program has the needed functionality. |
f8cac037 | 267 | |
2bbc70d5 AJ |
268 | We recommend version GNU make version 3.79 or newer. Older versions have |
269 | bugs and/or are missing features. | |
f8cac037 | 270 | |
f8cac037 | 271 | |
5edb9387 | 272 | 1.4. Do I need a special linker or assembler? |
f8cac037 | 273 | |
5edb9387 UD |
274 | {ZW} If you want a shared library, you need a linker and assembler that |
275 | understand all the features of ELF, including weak and versioned symbols. | |
276 | The static library can be compiled with less featureful tools, but lacks key | |
277 | features such as NSS. | |
41f27456 | 278 | |
5edb9387 UD |
279 | For Linux or Hurd, you want binutils 2.8.1.0.23, 2.9.1, or 2.9.1.0.15 or |
280 | higher. These are the only versions we've tested and found reliable. Other | |
281 | versions after 2.8.1.0.23 may work but we don't recommend them, especially | |
282 | not when C++ is involved. Earlier versions do not work at all. | |
a379e56a | 283 | |
5edb9387 UD |
284 | Other operating systems may come with system tools that have all the |
285 | necessary features, but this is moot because glibc hasn't been ported to | |
286 | them. | |
f8cac037 | 287 | |
f8cac037 | 288 | |
8619129f | 289 | 1.5. Which compiler should I use for powerpc? |
4775243a | 290 | |
83f6a990 UD |
291 | {GK} You want to use at least gcc 2.95 (together with the right versions |
292 | of all the other tools, of course). See also question question 2.8. | |
4775243a UD |
293 | |
294 | ||
348ed515 UD |
295 | 1.6. Which tools should I use for ARM? |
296 | ||
297 | {PB} You should use egcs 1.1 or a later version. For ELF systems some | |
298 | changes are needed to the compiler; a patch against egcs-1.1.x can be found | |
299 | at: | |
300 | ||
301 | <ftp://ftp.netwinder.org/users/p/philb/egcs-1.1.1pre2-diff-981126> | |
302 | ||
303 | Binutils 2.9.1.0.16 or later is also required. | |
304 | ||
305 | ||
306 | 1.7. Do I need some more things to compile the GNU C Library? | |
f8cac037 | 307 | |
61952351 | 308 | {UD} Yes, there are some more :-). |
78b5ba3e | 309 | |
61952351 UD |
310 | * GNU gettext. This package contains the tools needed to construct |
311 | `message catalog' files containing translated versions of system | |
a35cb74d | 312 | messages. See ftp://ftp.gnu.org/pub/gnu or better any mirror |
61952351 | 313 | site. (We distribute compiled message catalogs, but they may not be |
2f2f52f7 UD |
314 | updated in patches.) Please note that the required minimal version |
315 | (0.10.35) of gettext is alpha software and available from | |
316 | ftp://alpha.gnu.org/gnu . | |
f8cac037 | 317 | |
5edb9387 UD |
318 | * Some files are built with special tools. E.g., files ending in .gperf |
319 | need a `gperf' program. The GNU version (now available in a separate | |
320 | package, formerly only as part of libg++) is known to work while some | |
321 | vendor versions do not. | |
f8cac037 | 322 | |
61952351 | 323 | You should not need these tools unless you change the source files. |
1f205a47 | 324 | |
5edb9387 UD |
325 | * Perl 5 is needed if you wish to test an installation of GNU libc |
326 | as the primary C library. | |
4775243a | 327 | |
61952351 UD |
328 | * When compiling for Linux, the header files of the Linux kernel must |
329 | be available to the compiler as <linux/*.h> and <asm/*.h>. | |
f8cac037 | 330 | |
02228370 | 331 | * lots of disk space (~400MB for i?86-linux; more for RISC platforms). |
af6f3906 | 332 | |
61952351 | 333 | * plenty of time. Compiling just the shared and static libraries for |
b5a9efcd UD |
334 | i?86-linux takes approximately 1h on an AMD-K6@225MHz w/ 96MB of RAM, |
335 | 45mins on a Celeron@400MHz w/ 128MB, and 55mins on a Alpha@533MHz w/ 256MB. | |
336 | Multiply this by 1.5 or 2.0 if you build profiling and/or the highly | |
337 | optimized version as well. For Hurd systems times are much higher. | |
f8cac037 | 338 | |
61952351 UD |
339 | You should avoid compiling in a NFS mounted filesystem. This is |
340 | very slow. | |
0200214b | 341 | |
61952351 | 342 | James Troup <J.J.Troup@comp.brad.ac.uk> reports a compile time of |
4775243a UD |
343 | 45h34m for a full build (shared, static, and profiled) on Atari |
344 | Falcon (Motorola 68030 @ 16 Mhz, 14 Mb memory) and Jan Barte | |
345 | <yann@plato.uni-paderborn.de> reports 22h48m on Atari TT030 | |
346 | (Motorola 68030 @ 32 Mhz, 34 Mb memory) | |
0200214b | 347 | |
83f6a990 | 348 | A full build of the PowerPC library took 1h on a PowerPC 750@400Mhz w/ |
6e8afc1c | 349 | 64MB of RAM, and about 9h on a 601@60Mhz w/ 72Mb. |
83f6a990 | 350 | |
61952351 | 351 | If you have some more measurements let me know. |
0200214b | 352 | |
ba1ffaa1 | 353 | |
348ed515 | 354 | 1.8. What version of the Linux kernel headers should be used? |
a35cb74d | 355 | |
f12944ec UD |
356 | {AJ,UD} The headers from the most recent Linux kernel should be used. The |
357 | headers used while compiling the GNU C library and the kernel binary used | |
358 | when using the library do not need to match. The GNU C library runs without | |
359 | problems on kernels that are older than the kernel headers used. The other | |
360 | way round (compiling the GNU C library with old kernel headers and running | |
361 | on a recent kernel) does not necessarily work. For example you can't use | |
b1418d8f | 362 | new kernel features if you used old kernel headers to compile the GNU C |
f12944ec UD |
363 | library. |
364 | ||
b0610668 | 365 | {ZW} Even if you are using a 2.0 kernel on your machine, we recommend you |
b710a6e2 UD |
366 | compile GNU libc with 2.2 kernel headers. That way you won't have to |
367 | recompile libc if you ever upgrade to kernel 2.2. To tell libc which | |
b0610668 | 368 | headers to use, give configure the --with-headers switch |
b710a6e2 | 369 | (e.g. --with-headers=/usr/src/linux-2.2.0/include). |
b0610668 | 370 | |
b710a6e2 | 371 | Note that you must configure the 2.2 kernel if you do this, otherwise libc |
62595351 | 372 | will be unable to find <linux/version.h>. Just change the current directory |
b710a6e2 | 373 | to the root of the 2.2 tree and do `make include/linux/version.h'. |
b0610668 | 374 | |
f12944ec | 375 | |
348ed515 | 376 | 1.9. The compiler hangs while building iconvdata modules. What's |
f12944ec UD |
377 | wrong? |
378 | ||
5edb9387 UD |
379 | {ZW} This is a problem with old versions of GCC. Initialization of large |
380 | static arrays is very slow. The compiler will eventually finish; give it | |
381 | time. | |
a35cb74d | 382 | |
b8f558b7 | 383 | The problem is fixed in egcs 1.1. |
a35cb74d | 384 | |
f12944ec | 385 | |
348ed515 | 386 | 1.10. When I run `nm -u libc.so' on the produced library I still |
61952351 | 387 | find unresolved symbols. Can this be ok? |
f8cac037 | 388 | |
f12944ec | 389 | {UD} Yes, this is ok. There can be several kinds of unresolved symbols: |
f8cac037 | 390 | |
61952351 UD |
391 | * magic symbols automatically generated by the linker. These have names |
392 | like __start_* and __stop_* | |
f8cac037 | 393 | |
78b5ba3e RM |
394 | * symbols starting with _dl_* come from the dynamic linker |
395 | ||
61952351 | 396 | * weak symbols, which need not be resolved at all (fabs for example) |
f8cac037 RM |
397 | |
398 | Generally, you should make sure you find a real program which produces | |
41f27456 | 399 | errors while linking before deciding there is a problem. |
f8cac037 RM |
400 | |
401 | ||
348ed515 | 402 | 1.11. What are these `add-ons'? |
999493cb | 403 | |
f12944ec | 404 | {UD} To avoid complications with export rules or external source code some |
b669ab02 UD |
405 | optional parts of the libc are distributed as separate packages, e.g., the |
406 | linuxthreads package. | |
999493cb | 407 | |
f12944ec UD |
408 | To use these packages as part of GNU libc, just unpack the tarfiles in the |
409 | libc source directory and tell the configuration script about them using the | |
410 | --enable-add-ons option. If you give just --enable-add-ons configure tries | |
411 | to find all the add-on packages in your source tree. This may not work. If | |
412 | it doesn't, or if you want to select only a subset of the add-ons, give a | |
413 | comma-separated list of the add-ons to enable: | |
613a76ff | 414 | |
b669ab02 | 415 | configure --enable-add-ons=linuxthreads |
41f27456 | 416 | |
61952351 | 417 | for example. |
0200214b | 418 | |
f12944ec UD |
419 | Add-ons can add features (including entirely new shared libraries), override |
420 | files, provide support for additional architectures, and just about anything | |
421 | else. The existing makefiles do most of the work; only some few stub rules | |
422 | must be written to get everything running. | |
613a76ff | 423 | |
bd952512 | 424 | Most add-ons are tightly coupled to a specific GNU libc version. Please |
b669ab02 UD |
425 | check that the add-ons work with the GNU libc. For example the linuxthreads |
426 | add-on has the same numbering scheme as the libc and will in general only | |
427 | work with the corresponding libc. | |
428 | ||
429 | {AJ} With glibc 2.2 the crypt add-on and with glibc 2.1 the localedata | |
430 | add-on have been integrated into the normal glibc distribution, crypt and | |
431 | localedata are therefore not anymore add-ons. | |
bd952512 | 432 | |
613a76ff | 433 | |
348ed515 | 434 | 1.12. My XXX kernel emulates a floating-point coprocessor for me. |
61952351 | 435 | Should I enable --with-fp? |
613a76ff | 436 | |
f12944ec UD |
437 | {ZW} An emulated FPU is just as good as a real one, as far as the C library |
438 | is concerned. You only need to say --without-fp if your machine has no way | |
439 | to execute floating-point instructions. | |
f8cac037 | 440 | |
61952351 UD |
441 | People who are interested in squeezing the last drop of performance |
442 | out of their machine may wish to avoid the trap overhead, but this is | |
443 | far more trouble than it's worth: you then have to compile | |
444 | *everything* this way, including the compiler's internal libraries | |
445 | (libgcc.a for GNU C), because the calling conventions change. | |
a1470b6f | 446 | |
999493cb | 447 | |
348ed515 | 448 | 1.13. When compiling GNU libc I get lots of errors saying functions |
61952351 | 449 | in glibc are duplicated in libgcc. |
5290baf0 | 450 | |
f12944ec UD |
451 | {EY} This is *exactly* the same problem that I was having. The problem was |
452 | due to the fact that configure didn't correctly detect that the linker flag | |
453 | --no-whole-archive was supported in my linker. In my case it was because I | |
454 | had run ./configure with bogus CFLAGS, and the test failed. | |
78b5ba3e | 455 | |
f12944ec UD |
456 | One thing that is particularly annoying about this problem is that once this |
457 | is misdetected, running configure again won't fix it unless you first delete | |
458 | config.cache. | |
78b5ba3e | 459 | |
f12944ec UD |
460 | {UD} Starting with glibc-2.0.3 there should be a better test to avoid some |
461 | problems of this kind. The setting of CFLAGS is checked at the very | |
462 | beginning and if it is not usable `configure' will bark. | |
78b5ba3e | 463 | |
af6f3906 | 464 | |
348ed515 | 465 | 1.14. Why do I get messages about missing thread functions when I use |
a35cb74d | 466 | librt? I don't even use threads. |
4775243a | 467 | |
a35cb74d UD |
468 | {UD} In this case you probably mixed up your installation. librt uses |
469 | threads internally and has implicit references to the thread library. | |
f12944ec UD |
470 | Normally these references are satisfied automatically but if the thread |
471 | library is not in the expected place you must tell the linker where it is. | |
472 | When using GNU ld it works like this: | |
4775243a UD |
473 | |
474 | gcc -o foo foo.c -Wl,-rpath-link=/some/other/dir -lrt | |
475 | ||
f12944ec UD |
476 | The `/some/other/dir' should contain the thread library. `ld' will use the |
477 | given path to find the implicitly referenced library while not disturbing | |
478 | any other link path. | |
4775243a UD |
479 | |
480 | ||
348ed515 | 481 | 1.15. What's the problem with configure --enable-omitfp? |
78b5ba3e | 482 | |
61952351 | 483 | {AJ} When --enable-omitfp is set the libraries are built without frame |
fdacb17d | 484 | pointers. Some compilers produce buggy code for this model and therefore we |
f12944ec | 485 | don't advise using it at the moment. |
66219c07 | 486 | |
fdacb17d | 487 | If you use --enable-omitfp, you're on your own. If you encounter problems |
f12944ec UD |
488 | with a library that was build this way, we advise you to rebuild the library |
489 | without --enable-omitfp. If the problem vanishes consider tracking the | |
490 | problem down and report it as compiler failure. | |
66219c07 | 491 | |
b1418d8f UD |
492 | Since a library built with --enable-omitfp is undebuggable on most systems, |
493 | debuggable libraries are also built - you can use them by appending "_g" to | |
f12944ec | 494 | the library names. |
66219c07 | 495 | |
f12944ec UD |
496 | The compilation of these extra libraries and the compiler optimizations slow |
497 | down the build process and need more disk space. | |
66219c07 | 498 | |
b0610668 | 499 | |
b1418d8f | 500 | 1.16. I get failures during `make check'. What should I do? |
b0610668 | 501 | |
b1418d8f UD |
502 | {AJ} The testsuite should compile and run cleanly on your system; every |
503 | failure should be looked into. Depending on the failures, you probably | |
504 | should not install the library at all. | |
b0610668 UD |
505 | |
506 | You should consider using the `glibcbug' script to report the failure, | |
507 | providing as much detail as possible. If you run a test directly, please | |
508 | remember to set up the environment correctly. You want to test the compiled | |
509 | library - and not your installed one. The best way is to copy the exact | |
510 | command line which failed and run the test from the subdirectory for this | |
511 | test in the sources. | |
512 | ||
513 | There are some failures which are not directly related to the GNU libc: | |
b1418d8f UD |
514 | - Some compilers produce buggy code. No compiler gets single precision |
515 | complex numbers correct on Alpha. Otherwise, the egcs 1.1 release should be | |
516 | ok; gcc 2.8.1 might cause some failures; gcc 2.7.2.x is so buggy that | |
517 | explicit checks have been used so that you can't build with it. | |
b0610668 UD |
518 | - The kernel might have bugs. For example on Linux/Alpha 2.0.34 the |
519 | floating point handling has quite a number of bugs and therefore most of | |
b710a6e2 | 520 | the test cases in the math subdirectory will fail. Linux 2.2 has |
b1418d8f UD |
521 | fixes for the floating point support on Alpha. The Linux/SPARC kernel has |
522 | also some bugs in the FPU emulation code (as of Linux 2.2.0). | |
d32a4020 UD |
523 | - Other tools might have problems. For example bash 2.03 gives a |
524 | segmentation fault running the tst-rpmatch.sh test script. | |
b0610668 | 525 | |
a379e56a | 526 | |
348ed515 | 527 | 1.17. What is symbol versioning good for? Do I need it? |
a379e56a UD |
528 | |
529 | {AJ} Symbol versioning solves problems that are related to interface | |
530 | changes. One version of an interface might have been introduced in a | |
531 | previous version of the GNU C library but the interface or the semantics of | |
532 | the function has been changed in the meantime. For binary compatibility | |
533 | with the old library, a newer library needs to still have the old interface | |
b1418d8f | 534 | for old programs. On the other hand, new programs should use the new |
a379e56a | 535 | interface. Symbol versioning is the solution for this problem. The GNU |
b1418d8f UD |
536 | libc version 2.1 uses symbol versioning by default if the installed binutils |
537 | supports it. | |
a379e56a | 538 | |
b1418d8f UD |
539 | We don't advise building without symbol versioning, since you lose binary |
540 | compatibility - forever! The binary compatibility you lose is not only | |
541 | against the previous version of the GNU libc (version 2.0) but also against | |
542 | all future versions. | |
a379e56a | 543 | |
c7f7281e UD |
544 | |
545 | 1.18. How can I compile on my fast ix86 machine a working libc for my slow | |
546 | i386? After installing libc, programs abort with "Illegal | |
547 | Instruction". | |
548 | ||
549 | {AJ} glibc and gcc might generate some instructions on your machine that | |
550 | aren't available on i386. You've got to tell glibc that you're configuring | |
551 | for i386 with adding i386 as your machine, for example: | |
552 | ||
553 | ../configure --prefix=/usr i386-pc-linux-gnu | |
554 | ||
555 | And you need to tell gcc to only generate i386 code, just add `-mcpu=i386' | |
556 | (just -m386 doesn't work) to your CFLAGS. | |
557 | ||
558 | {UD} This applies not only to the i386. Compiling on a i686 for any older | |
559 | model will also fail if the above methods are not used. | |
560 | ||
150ae521 UD |
561 | |
562 | 1.19. `make' complains about a missing dlfcn/libdl.so when building | |
563 | malloc/libmemprof.so. How can I fix this? | |
564 | ||
565 | {AJ} Older make version (<= 3.78.90) have a bug which was hidden by a bug in | |
2bbc70d5 | 566 | glibc (<= 2.1.2). You need to upgrade make to a newer or fixed version. |
150ae521 UD |
567 | |
568 | After upgrading make, you should remove the file sysd-sorted in your build | |
569 | directory. The problem is that the broken make creates a wrong order for | |
570 | one list in that file. The list has to be recreated with the new make - | |
571 | which happens if you remove the file. | |
572 | ||
573 | You might encounter this bug also in other situations where make scans | |
2bbc70d5 AJ |
574 | directories. I strongly advise to upgrade your make version to 3.79 or |
575 | newer. | |
150ae521 | 576 | |
61952351 UD |
577 | \f |
578 | . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . | |
e6c9a67a | 579 | |
61952351 | 580 | 2. Installation and configuration issues |
e6c9a67a | 581 | |
61952351 | 582 | 2.1. Can I replace the libc on my Linux system with GNU libc? |
e6c9a67a | 583 | |
f12944ec UD |
584 | {UD} You cannot replace any existing libc for Linux with GNU libc. It is |
585 | binary incompatible and therefore has a different major version. You can, | |
586 | however, install it alongside your existing libc. | |
e6c9a67a | 587 | |
61952351 UD |
588 | For Linux there are three major libc versions: |
589 | libc-4 a.out libc | |
590 | libc-5 original ELF libc | |
591 | libc-6 GNU libc | |
e6c9a67a | 592 | |
f12944ec UD |
593 | You can have any combination of these three installed. For more information |
594 | consult documentation for shared library handling. The Makefiles of GNU | |
595 | libc will automatically generate the needed symbolic links which the linker | |
596 | will use. | |
e6c9a67a RM |
597 | |
598 | ||
61952351 UD |
599 | 2.2. How do I configure GNU libc so that the essential libraries |
600 | like libc.so go into /lib and the other into /usr/lib? | |
ec42724d | 601 | |
61952351 UD |
602 | {UD,AJ} Like all other GNU packages GNU libc is designed to use a base |
603 | directory and install all files relative to this. The default is | |
f12944ec UD |
604 | /usr/local, because this is safe (it will not damage the system if installed |
605 | there). If you wish to install GNU libc as the primary C library on your | |
606 | system, set the base directory to /usr (i.e. run configure --prefix=/usr | |
607 | <other_options>). Note that this can damage your system; see question 2.3 for | |
608 | details. | |
609 | ||
610 | Some systems like Linux have a filesystem standard which makes a difference | |
611 | between essential libraries and others. Essential libraries are placed in | |
612 | /lib because this directory is required to be located on the same disk | |
613 | partition as /. The /usr subtree might be found on another | |
614 | partition/disk. If you configure for Linux with --prefix=/usr, then this | |
615 | will be done automatically. | |
ec42724d | 616 | |
61952351 | 617 | To install the essential libraries which come with GNU libc in /lib on |
f12944ec UD |
618 | systems other than Linux one must explicitly request it. Autoconf has no |
619 | option for this so you have to use a `configparms' file (see the `INSTALL' | |
620 | file for details). It should contain: | |
ec42724d RM |
621 | |
622 | slibdir=/lib | |
623 | sysconfdir=/etc | |
624 | ||
f12944ec UD |
625 | The first line specifies the directory for the essential libraries, the |
626 | second line the directory for system configuration files. | |
ec42724d | 627 | |
5290baf0 | 628 | |
61952351 | 629 | 2.3. How should I avoid damaging my system when I install GNU libc? |
ec42724d | 630 | |
f12944ec UD |
631 | {ZW} If you wish to be cautious, do not configure with --prefix=/usr. If |
632 | you don't specify a prefix, glibc will be installed in /usr/local, where it | |
633 | will probably not break anything. (If you wish to be certain, set the | |
634 | prefix to something like /usr/local/glibc2 which is not used for anything.) | |
845dcb57 | 635 | |
61952351 | 636 | The dangers when installing glibc in /usr are twofold: |
845dcb57 | 637 | |
61952351 | 638 | * glibc will overwrite the headers in /usr/include. Other C libraries |
27e309c1 UD |
639 | install a different but overlapping set of headers there, so the effect |
640 | will probably be that you can't compile anything. You need to rename | |
641 | /usr/include out of the way before running `make install'. (Do not throw | |
642 | it away; you will then lose the ability to compile programs against your | |
643 | old libc.) | |
845dcb57 | 644 | |
61952351 UD |
645 | * None of your old libraries, static or shared, can be used with a |
646 | different C library major version. For shared libraries this is not a | |
647 | problem, because the filenames are different and the dynamic linker | |
648 | will enforce the restriction. But static libraries have no version | |
649 | information. You have to evacuate all the static libraries in | |
650 | /usr/lib to a safe location. | |
845dcb57 | 651 | |
61952351 UD |
652 | The situation is rather similar to the move from a.out to ELF which |
653 | long-time Linux users will remember. | |
845dcb57 | 654 | |
845dcb57 | 655 | |
61952351 UD |
656 | 2.4. Do I need to use GNU CC to compile programs that will use the |
657 | GNU C Library? | |
845dcb57 | 658 | |
f12944ec UD |
659 | {ZW} In theory, no; the linker does not care, and the headers are supposed |
660 | to check for GNU CC before using its extensions to the C language. | |
845dcb57 | 661 | |
f12944ec UD |
662 | However, there are currently no ports of glibc to systems where another |
663 | compiler is the default, so no one has tested the headers extensively | |
664 | against another compiler. You may therefore encounter difficulties. If you | |
665 | do, please report them as bugs. | |
845dcb57 | 666 | |
61952351 UD |
667 | Also, in several places GNU extensions provide large benefits in code |
668 | quality. For example, the library has hand-optimized, inline assembly | |
f12944ec UD |
669 | versions of some string functions. These can only be used with GCC. See |
670 | question 3.8 for details. | |
845dcb57 | 671 | |
845dcb57 | 672 | |
61952351 UD |
673 | 2.5. When linking with the new libc I get unresolved symbols |
674 | `crypt' and `setkey'. Why aren't these functions in the | |
675 | libc anymore? | |
845dcb57 UD |
676 | |
677 | ||
b669ab02 UD |
678 | Removed. Does not apply anymore. |
679 | ||
c4029823 | 680 | |
c4029823 | 681 | |
c4029823 | 682 | |
61952351 UD |
683 | 2.6. When I use GNU libc on my Linux system by linking against |
684 | the libc.so which comes with glibc all I get is a core dump. | |
c4029823 | 685 | |
f12944ec | 686 | {UD} On Linux, gcc sets the dynamic linker to /lib/ld-linux.so.1 unless the |
a582750d | 687 | user specifies a --dynamic-linker argument. This is the name of the libc5 |
f12944ec | 688 | dynamic linker, which does not work with glibc. |
61952351 | 689 | |
a379e56a UD |
690 | For casual use of GNU libc you can just specify to the linker |
691 | --dynamic-linker=/lib/ld-linux.so.2 | |
61952351 | 692 | |
f12944ec | 693 | which is the glibc dynamic linker, on Linux systems. On other systems the |
a379e56a UD |
694 | name is /lib/ld.so.1. When linking via gcc, you've got to add |
695 | -Wl,--dynamic-linker=/lib/ld-linux.so.2 | |
696 | ||
697 | to the gcc command line. | |
c4029823 | 698 | |
f12944ec UD |
699 | To change your environment to use GNU libc for compiling you need to change |
700 | the `specs' file of your gcc. This file is normally found at | |
c4029823 UD |
701 | |
702 | /usr/lib/gcc-lib/<arch>/<version>/specs | |
703 | ||
704 | In this file you have to change a few things: | |
705 | ||
61952351 | 706 | - change `ld-linux.so.1' to `ld-linux.so.2' |
c4029823 UD |
707 | |
708 | - remove all expression `%{...:-lgmon}'; there is no libgmon in glibc | |
709 | ||
f4017d20 UD |
710 | - fix a minor bug by changing %{pipe:-} to %| |
711 | ||
f12944ec UD |
712 | Here is what the gcc-2.7.2 specs file should look like when GNU libc is |
713 | installed at /usr: | |
c4029823 UD |
714 | |
715 | ----------------------------------------------------------------------- | |
716 | *asm: | |
717 | %{V} %{v:%{!V:-V}} %{Qy:} %{!Qn:-Qy} %{n} %{T} %{Ym,*} %{Yd,*} %{Wa,*:%*} | |
718 | ||
719 | *asm_final: | |
f4017d20 | 720 | %| |
c4029823 UD |
721 | |
722 | *cpp: | |
68dbb3a6 | 723 | %{fPIC:-D__PIC__ -D__pic__} %{fpic:-D__PIC__ -D__pic__} %{!m386:-D__i486__} %{posix:-D_POSIX_SOURCE} %{pthread:-D_REENTRANT} |
c4029823 UD |
724 | |
725 | *cc1: | |
68dbb3a6 | 726 | %{profile:-p} |
c4029823 UD |
727 | |
728 | *cc1plus: | |
729 | ||
730 | ||
731 | *endfile: | |
68dbb3a6 | 732 | %{!shared:crtend.o%s} %{shared:crtendS.o%s} crtn.o%s |
c4029823 UD |
733 | |
734 | *link: | |
68dbb3a6 | 735 | -m elf_i386 %{shared:-shared} %{!shared: %{!ibcs: %{!static: %{rdynamic:-export-dynamic} %{!dynamic-linker:-dynamic-linker /lib/ld-linux.so.2}} %{static:-static}}} |
c4029823 UD |
736 | |
737 | *lib: | |
68dbb3a6 | 738 | %{!shared: %{pthread:-lpthread} %{profile:-lc_p} %{!profile: -lc}} |
c4029823 UD |
739 | |
740 | *libgcc: | |
68dbb3a6 | 741 | -lgcc |
c4029823 UD |
742 | |
743 | *startfile: | |
61952351 | 744 | %{!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} |
c4029823 UD |
745 | |
746 | *switches_need_spaces: | |
747 | ||
748 | ||
749 | *signed_char: | |
750 | %{funsigned-char:-D__CHAR_UNSIGNED__} | |
751 | ||
752 | *predefines: | |
753 | -D__ELF__ -Dunix -Di386 -Dlinux -Asystem(unix) -Asystem(posix) -Acpu(i386) -Amachine(i386) | |
754 | ||
755 | *cross_compile: | |
756 | 0 | |
757 | ||
758 | *multilib: | |
759 | . ; | |
760 | ||
761 | ----------------------------------------------------------------------- | |
762 | ||
f12944ec UD |
763 | Things get a bit more complicated if you have GNU libc installed in some |
764 | other place than /usr, i.e., if you do not want to use it instead of the old | |
765 | libc. In this case the needed startup files and libraries are not found in | |
766 | the regular places. So the specs file must tell the compiler and linker | |
767 | exactly what to use. | |
0d204b0a | 768 | |
f41c8091 | 769 | Version 2.7.2.3 does and future versions of GCC will automatically |
0d8733c4 | 770 | provide the correct specs. |
c4029823 UD |
771 | |
772 | ||
61952351 UD |
773 | 2.7. Looking through the shared libc file I haven't found the |
774 | functions `stat', `lstat', `fstat', and `mknod' and while | |
775 | linking on my Linux system I get error messages. How is | |
776 | this supposed to work? | |
c4029823 | 777 | |
f12944ec UD |
778 | {RM} Believe it or not, stat and lstat (and fstat, and mknod) are supposed |
779 | to be undefined references in libc.so.6! Your problem is probably a missing | |
780 | or incorrect /usr/lib/libc.so file; note that this is a small text file now, | |
781 | not a symlink to libc.so.6. It should look something like this: | |
c4029823 | 782 | |
ff44f2a5 | 783 | GROUP ( libc.so.6 libc_nonshared.a ) |
1f205a47 | 784 | |
c4029823 | 785 | |
5edb9387 UD |
786 | 2.8. When I run an executable on one system which I compiled on |
787 | another, I get dynamic linker errors. Both systems have the same | |
788 | version of glibc installed. What's wrong? | |
789 | ||
790 | {ZW} Glibc on one of these systems was compiled with gcc 2.7 or 2.8, the | |
791 | other with egcs (any version). Egcs has functions in its internal | |
792 | `libgcc.a' to support exception handling with C++. They are linked into | |
793 | any program or dynamic library compiled with egcs, whether it needs them or | |
794 | not. Dynamic libraries then turn around and export those functions again | |
795 | unless special steps are taken to prevent them. | |
796 | ||
797 | When you link your program, it resolves its references to the exception | |
798 | functions to the ones exported accidentally by libc.so. That works fine as | |
799 | long as libc has those functions. On the other system, libc doesn't have | |
800 | those functions because it was compiled by gcc 2.8, and you get undefined | |
801 | symbol errors. The symbols in question are named things like | |
802 | `__register_frame_info'. | |
803 | ||
804 | For glibc 2.0, the workaround is to not compile libc with egcs. We've also | |
805 | incorporated a patch which should prevent the EH functions sneaking into | |
806 | libc. It doesn't matter what compiler you use to compile your program. | |
807 | ||
808 | For glibc 2.1, we've chosen to do it the other way around: libc.so | |
809 | explicitly provides the EH functions. This is to prevent other shared | |
95f7cecb UD |
810 | libraries from doing it. |
811 | ||
812 | {UD} Starting with glibc 2.1.1 you can compile glibc with gcc 2.8.1 or | |
813 | newer since we have explicitly add references to the functions causing the | |
814 | problem. But you nevertheless should use EGCS for other reasons | |
815 | (see question 1.2). | |
5edb9387 | 816 | |
83f6a990 UD |
817 | {GK} On some Linux distributions for PowerPC, you can see this when you have |
818 | built gcc or egcs from the Web sources (gcc versions 2.95 or earlier), then | |
819 | re-built glibc. This happens because in these versions of gcc, exception | |
820 | handling is implemented using an older method; the people making the | |
821 | distributions are a little ahead of their time. | |
822 | ||
823 | A quick solution to this is to find the libgcc.a file that came with the | |
6e8afc1c | 824 | distribution (it would have been installed under /usr/lib/gcc-lib), do |
83f6a990 UD |
825 | `ar x libgcc.a frame.o' to get the frame.o file out, and add a line saying |
826 | `LDLIBS-c.so += frame.o' to the file `configparms' in the directory you're | |
827 | building in. You can check you've got the right `frame.o' file by running | |
828 | `nm frame.o' and checking that it has the symbols defined that you're | |
829 | missing. | |
830 | ||
831 | This will let you build glibc with the C compiler. The C++ compiler | |
832 | will still be binary incompatible with any C++ shared libraries that | |
833 | you got with your distribution. | |
834 | ||
5edb9387 UD |
835 | |
836 | 2.9. How can I compile gcc 2.7.2.1 from the gcc source code using | |
61952351 | 837 | glibc 2.x? |
ba1ffaa1 | 838 | |
f12944ec | 839 | {AJ} There's only correct support for glibc 2.0.x in gcc 2.7.2.3 or later. |
95f7cecb | 840 | But you should get at least gcc 2.8.1 or egcs 1.1 (or later versions) |
f12944ec | 841 | instead. |
ba1ffaa1 UD |
842 | |
843 | ||
5edb9387 | 844 | 2.10. The `gencat' utility cannot process the catalog sources which |
61952351 | 845 | were used on my Linux libc5 based system. Why? |
47707456 | 846 | |
f12944ec UD |
847 | {UD} The `gencat' utility provided with glibc complies to the XPG standard. |
848 | The older Linux version did not obey the standard, so they are not | |
849 | compatible. | |
47707456 | 850 | |
61952351 | 851 | To ease the transition from the Linux version some of the non-standard |
f12944ec UD |
852 | features are also present in the `gencat' program of GNU libc. This mainly |
853 | includes the use of symbols for the message number and the automatic | |
61952351 UD |
854 | generation of header files which contain the needed #defines to map the |
855 | symbols to integers. | |
47707456 | 856 | |
f12944ec UD |
857 | Here is a simple SED script to convert at least some Linux specific catalog |
858 | files to the XPG4 form: | |
68dbb3a6 | 859 | |
61952351 UD |
860 | ----------------------------------------------------------------------- |
861 | # Change catalog source in Linux specific format to standard XPG format. | |
862 | # Ulrich Drepper <drepper@cygnus.com>, 1996. | |
863 | # | |
864 | /^\$ #/ { | |
865 | h | |
866 | s/\$ #\([^ ]*\).*/\1/ | |
867 | x | |
868 | s/\$ #[^ ]* *\(.*\)/\$ \1/ | |
869 | } | |
68dbb3a6 | 870 | |
61952351 UD |
871 | /^# / { |
872 | s/^# \(.*\)/\1/ | |
873 | G | |
874 | s/\(.*\)\n\(.*\)/\2 \1/ | |
875 | } | |
876 | ----------------------------------------------------------------------- | |
19361cb7 | 877 | |
19361cb7 | 878 | |
5edb9387 | 879 | 2.11. Programs using libc have their messages translated, but other |
a35cb74d UD |
880 | behavior is not localized (e.g. collating order); why? |
881 | ||
882 | {ZW} Translated messages are automatically installed, but the locale | |
f12944ec UD |
883 | database that controls other behaviors is not. You need to run localedef to |
884 | install this database, after you have run `make install'. For example, to | |
885 | set up the French Canadian locale, simply issue the command | |
a35cb74d UD |
886 | |
887 | localedef -i fr_CA -f ISO-8859-1 fr_CA | |
888 | ||
889 | Please see localedata/README in the source tree for further details. | |
890 | ||
891 | ||
5edb9387 | 892 | 2.12. I have set up /etc/nis.conf, and the Linux libc 5 with NYS |
61952351 | 893 | works great. But the glibc NIS+ doesn't seem to work. |
19361cb7 | 894 | |
f12944ec UD |
895 | {TK} The glibc NIS+ implementation uses a /var/nis/NIS_COLD_START file for |
896 | storing information about the NIS+ server and their public keys, because the | |
897 | nis.conf file does not contain all the necessary information. You have to | |
898 | copy a NIS_COLD_START file from a Solaris client (the NIS_COLD_START file is | |
899 | byte order independent) or generate it with nisinit from the nis-tools | |
900 | package; available at | |
901 | ||
612fdf25 | 902 | http://www.suse.de/~kukuk/linux/nisplus.html |
19361cb7 | 903 | |
68dbb3a6 | 904 | |
5edb9387 | 905 | 2.13. I have killed ypbind to stop using NIS, but glibc |
3dcf8ea6 | 906 | continues using NIS. |
4d06461a | 907 | |
f12944ec UD |
908 | {TK} For faster NIS lookups, glibc uses the /var/yp/binding/ files from |
909 | ypbind. ypbind 3.3 and older versions don't always remove these files, so | |
910 | glibc will continue to use them. Other BSD versions seem to work correctly. | |
911 | Until ypbind 3.4 is released, you can find a patch at | |
912 | ||
66f6a52b | 913 | <ftp://ftp.kernel.org/pub/linux/utils/net/NIS/ypbind-3.3-glibc4.diff.gz> |
a35cb74d | 914 | |
4d06461a | 915 | |
5edb9387 | 916 | 2.14. Under Linux/Alpha, I always get "do_ypcall: clnt_call: |
3dcf8ea6 | 917 | RPC: Unable to receive; errno = Connection refused" when using NIS. |
4d06461a | 918 | |
f12944ec UD |
919 | {TK} You need a ypbind version which is 64bit clean. Some versions are not |
920 | 64bit clean. A 64bit clean implementation is ypbind-mt. For ypbind 3.3, | |
921 | you need the patch from ftp.kernel.org (See the previous question). I don't | |
922 | know about other versions. | |
a35cb74d UD |
923 | |
924 | ||
5edb9387 | 925 | 2.15. After installing glibc name resolving doesn't work properly. |
68dbb3a6 | 926 | |
f12944ec UD |
927 | {AJ} You probably should read the manual section describing nsswitch.conf |
928 | (just type `info libc "NSS Configuration File"'). The NSS configuration | |
929 | file is usually the culprit. | |
22d57dd3 | 930 | |
22d57dd3 | 931 | |
5edb9387 | 932 | 2.16. How do I create the databases for NSS? |
3dcf8ea6 UD |
933 | |
934 | {AJ} If you have an entry "db" in /etc/nsswitch.conf you should also create | |
935 | the database files. The glibc sources contain a Makefile which does the | |
a379e56a | 936 | necessary conversion and calls to create those files. The file is |
3dcf8ea6 UD |
937 | `db-Makefile' in the subdirectory `nss' and you can call it with `make -f |
938 | db-Makefile'. Please note that not all services are capable of using a | |
939 | database. Currently passwd, group, ethers, protocol, rpc, services shadow | |
9de4e203 | 940 | and netgroup are implemented. See also question question 2.31. |
3dcf8ea6 UD |
941 | |
942 | ||
5edb9387 | 943 | 2.17. I have /usr/include/net and /usr/include/scsi as symlinks |
61952351 | 944 | into my Linux source tree. Is that wrong? |
22d57dd3 | 945 | |
f12944ec UD |
946 | {PB} This was necessary for libc5, but is not correct when using glibc. |
947 | Including the kernel header files directly in user programs usually does not | |
948 | work (see question 3.5). glibc provides its own <net/*> and <scsi/*> header | |
949 | files to replace them, and you may have to remove any symlink that you have | |
950 | in place before you install glibc. However, /usr/include/asm and | |
951 | /usr/include/linux should remain as they were. | |
22d57dd3 | 952 | |
22d57dd3 | 953 | |
5edb9387 | 954 | 2.18. Programs like `logname', `top', `uptime' `users', `w' and |
61952351 UD |
955 | `who', show incorrect information about the (number of) |
956 | users on my system. Why? | |
22d57dd3 | 957 | |
61952351 | 958 | {MK} See question 3.2. |
22d57dd3 | 959 | |
22d57dd3 | 960 | |
5edb9387 | 961 | 2.19. After upgrading to glibc 2.1 with symbol versioning I get |
61952351 | 962 | errors about undefined symbols. What went wrong? |
26dee9c4 | 963 | |
f12944ec UD |
964 | {AJ} The problem is caused either by wrong program code or tools. In the |
965 | versioned libc a lot of symbols are now local that were global symbols in | |
966 | previous versions. It seems that programs linked against older versions | |
967 | often accidentally used libc global variables -- something that should not | |
968 | happen. | |
26dee9c4 | 969 | |
f12944ec UD |
970 | The only way to fix this is to recompile your program. Sorry, that's the |
971 | price you might have to pay once for quite a number of advantages with | |
972 | symbol versioning. | |
26dee9c4 | 973 | |
26dee9c4 | 974 | |
5edb9387 | 975 | 2.20. When I start the program XXX after upgrading the library |
61952351 UD |
976 | I get |
977 | XXX: Symbol `_sys_errlist' has different size in shared | |
978 | object, consider re-linking | |
979 | Why? What should I do? | |
26dee9c4 | 980 | |
f12944ec UD |
981 | {UD} As the message says, relink the binary. The problem is that a few |
982 | symbols from the library can change in size and there is no way to avoid | |
983 | this. _sys_errlist is a good example. Occasionally there are new error | |
984 | numbers added to the kernel and this must be reflected at user level, | |
985 | breaking programs that refer to them directly. | |
a2b08ee5 | 986 | |
f12944ec UD |
987 | Such symbols should normally not be used at all. There are mechanisms to |
988 | avoid using them. In the case of _sys_errlist, there is the strerror() | |
989 | function which should _always_ be used instead. So the correct fix is to | |
990 | rewrite that part of the application. | |
a2b08ee5 | 991 | |
f12944ec UD |
992 | In some situations (especially when testing a new library release) it might |
993 | be possible that a symbol changed size when that should not have happened. | |
994 | So in case of doubt report such a warning message as a problem. | |
a2b08ee5 | 995 | |
a35cb74d | 996 | |
5edb9387 UD |
997 | 2.21. What do I need for C++ development? |
998 | ||
999 | {HJ,AJ} You need either egcs 1.1 which comes directly with libstdc++ or | |
1000 | gcc-2.8.1 together with libstdc++ 2.8.1.1. egcs 1.1 has the better C++ | |
1001 | support and works directly with glibc 2.1. If you use gcc-2.8.1 with | |
1002 | libstdc++ 2.8.1.1, you need to modify libstdc++ a bit. A patch is available | |
1003 | as: | |
66f6a52b | 1004 | <ftp://alpha.gnu.org/gnu/libstdc++-2.8.1.1-glibc2.1-diff.gz> |
a35cb74d | 1005 | |
5edb9387 UD |
1006 | Please note that libg++ 2.7.2 (and the Linux Versions 2.7.2.x) doesn't work |
1007 | very well with the GNU C library due to vtable thunks. If you're upgrading | |
1008 | from glibc 2.0.x to 2.1 you have to recompile libstdc++ since the library | |
1009 | compiled for 2.0 is not compatible due to the new Large File Support (LFS) | |
1010 | in version 2.1. | |
fb98e2bf UD |
1011 | |
1012 | {UD} But since in the case of a shared libstdc++ the version numbers should | |
1013 | be different existing programs will continue to work. | |
a35cb74d | 1014 | |
ff44f2a5 | 1015 | |
5edb9387 | 1016 | 2.22. Even statically linked programs need some shared libraries |
ff44f2a5 UD |
1017 | which is not acceptable for me. What can I do? |
1018 | ||
f12944ec UD |
1019 | {AJ} NSS (for details just type `info libc "Name Service Switch"') won't |
1020 | work properly without shared libraries. NSS allows using different services | |
1021 | (e.g. NIS, files, db, hesiod) by just changing one configuration file | |
1022 | (/etc/nsswitch.conf) without relinking any programs. The only disadvantage | |
1023 | is that now static libraries need to access shared libraries. This is | |
1024 | handled transparently by the GNU C library. | |
ff44f2a5 | 1025 | |
f12944ec UD |
1026 | A solution is to configure glibc with --enable-static-nss. In this case you |
1027 | can create a static binary that will use only the services dns and files | |
1028 | (change /etc/nsswitch.conf for this). You need to link explicitly against | |
1029 | all these services. For example: | |
ff44f2a5 UD |
1030 | |
1031 | gcc -static test-netdb.c -o test-netdb.c \ | |
1032 | -lc -lnss_files -lnss_dns -lresolv | |
1033 | ||
1034 | The problem with this approach is that you've got to link every static | |
1035 | program that uses NSS routines with all those libraries. | |
1036 | ||
1037 | {UD} In fact, one cannot say anymore that a libc compiled with this | |
1038 | option is using NSS. There is no switch anymore. Therefore it is | |
1039 | *highly* recommended *not* to use --enable-static-nss since this makes | |
1040 | the behaviour of the programs on the system inconsistent. | |
1041 | ||
fdacb17d | 1042 | |
5edb9387 | 1043 | 2.23. I just upgraded my Linux system to glibc and now I get |
fdacb17d UD |
1044 | errors whenever I try to link any program. |
1045 | ||
1046 | {ZW} This happens when you have installed glibc as the primary C library but | |
1047 | have stray symbolic links pointing at your old C library. If the first | |
1048 | `libc.so' the linker finds is libc 5, it will use that. Your program | |
1049 | expects to be linked with glibc, so the link fails. | |
1050 | ||
1051 | The most common case is that glibc put its `libc.so' in /usr/lib, but there | |
1052 | was a `libc.so' from libc 5 in /lib, which gets searched first. To fix the | |
1053 | problem, just delete /lib/libc.so. You may also need to delete other | |
1054 | symbolic links in /lib, such as /lib/libm.so if it points to libm.so.5. | |
1055 | ||
1056 | {AJ} The perl script test-installation.pl which is run as last step during | |
1057 | an installation of glibc that is configured with --prefix=/usr should help | |
1058 | detect these situations. If the script reports problems, something is | |
1059 | really screwed up. | |
1060 | ||
48244d09 | 1061 | |
5edb9387 | 1062 | 2.24. When I use nscd the machine freezes. |
48244d09 | 1063 | |
5edb9387 UD |
1064 | {UD} You cannot use nscd with Linux 2.0.*. There is functionality missing |
1065 | in the kernel and work-arounds are not suitable. Besides, some parts of the | |
1066 | kernel are too buggy when it comes to using threads. | |
48244d09 | 1067 | |
b710a6e2 | 1068 | If you need nscd, you have to use at least a 2.1 kernel. |
48244d09 UD |
1069 | |
1070 | Note that I have at this point no information about any other platform. | |
1071 | ||
0155a773 UD |
1072 | |
1073 | 2.25. I need lots of open files. What do I have to do? | |
1074 | ||
1075 | {AJ} This is at first a kernel issue. The kernel defines limits with | |
1076 | OPEN_MAX the number of simultaneous open files and with FD_SETSIZE the | |
1077 | number of used file descriptors. You need to change these values in your | |
c0389ee4 | 1078 | kernel and recompile the kernel so that the kernel allows more open |
0155a773 UD |
1079 | files. You don't necessarily need to recompile the GNU C library since the |
1080 | only place where OPEN_MAX and FD_SETSIZE is really needed in the library | |
1081 | itself is the size of fd_set which is used by select. | |
1082 | ||
249fd241 UD |
1083 | The GNU C library is now select free. This means it internally has no |
1084 | limits imposed by the `fd_set' type. Instead all places where the | |
0155a773 UD |
1085 | functionality is needed the `poll' function is used. |
1086 | ||
1087 | If you increase the number of file descriptors in the kernel you don't need | |
6e8afc1c | 1088 | to recompile the C library. |
0155a773 UD |
1089 | |
1090 | {UD} You can always get the maximum number of file descriptors a process is | |
1091 | allowed to have open at any time using | |
1092 | ||
1093 | number = sysconf (_SC_OPEN_MAX); | |
1094 | ||
1095 | This will work even if the kernel limits change. | |
1096 | ||
7db169c9 UD |
1097 | |
1098 | 2.26. How do I get the same behavior on parsing /etc/passwd and | |
1099 | /etc/group as I have with libc5 ? | |
1100 | ||
1101 | {TK} The name switch setup in /etc/nsswitch.conf selected by most Linux | |
1102 | distributions does not support +/- and netgroup entries in the files like | |
1103 | /etc/passwd. Though this is the preferred setup some people might have | |
1104 | setups coming over from the libc5 days where it was the default to recognize | |
1105 | lines like this. To get back to the old behaviour one simply has to change | |
1106 | the rules for passwd, group, and shadow in the nsswitch.conf file as | |
1107 | follows: | |
1108 | ||
1109 | passwd: compat | |
1110 | group: compat | |
1111 | shadow: compat | |
1112 | ||
1113 | passwd_compat: nis | |
1114 | group_compat: nis | |
1115 | shadow_compat: nis | |
1116 | ||
b710a6e2 UD |
1117 | |
1118 | 2.27. What needs to be recompiled when upgrading from glibc 2.0 to glibc | |
1119 | 2.1? | |
1120 | ||
1121 | {AJ,CG} If you just upgrade the glibc from 2.0.x (x <= 7) to 2.1, binaries | |
1122 | that have been linked against glibc 2.0 will continue to work. | |
1123 | ||
1124 | If you compile your own binaries against glibc 2.1, you also need to | |
c19559b0 UD |
1125 | recompile some other libraries. The problem is that libio had to be changed |
1126 | and therefore libraries that are based or depend on the libio of glibc, | |
1127 | e.g. ncurses, slang and most C++ libraries, need to be recompiled. If you | |
1128 | experience strange segmentation faults in your programs linked against glibc | |
1129 | 2.1, you might need to recompile your libraries. | |
b710a6e2 UD |
1130 | |
1131 | Another problem is that older binaries that were linked statically against | |
1132 | glibc 2.0 will reference the older nss modules (libnss_files.so.1 instead of | |
1133 | libnss_files.so.2), so don't remove them. Also, the old glibc-2.0 compiled | |
1134 | static libraries (libfoo.a) which happen to depend on the older libio | |
1135 | behavior will be broken by the glibc 2.1 upgrade. We plan to produce a | |
1136 | compatibility library that people will be able to link in if they want | |
1137 | to compile a static library generated against glibc 2.0 into a program | |
1138 | on a glibc 2.1 system. You just add -lcompat and you should be fine. | |
1139 | ||
1140 | The glibc-compat add-on will provide the libcompat.a library, the older | |
1141 | nss modules, and a few other files. Together, they should make it | |
1142 | possible to do development with old static libraries on a glibc 2.1 | |
8d8c6efa | 1143 | system. This add-on is still in development. You can get it from |
df08cc56 | 1144 | <ftp://alpha.gnu.org/gnu/glibc/glibc-compat-2.1.tar.gz> |
b710a6e2 UD |
1145 | but please keep in mind that it is experimental. |
1146 | ||
b7398be5 UD |
1147 | |
1148 | 2.28. Why is extracting files via tar so slow? | |
1149 | ||
1150 | {AJ} Extracting of tar archives might be quite slow since tar has to look up | |
1151 | userid and groupids and doesn't cache negative results. If you have nis or | |
1152 | nisplus in your /etc/nsswitch.conf for the passwd and/or group database, | |
1153 | each file extractions needs a network connection. There are two possible | |
1154 | solutions: | |
1155 | ||
1156 | - do you really need NIS/NIS+ (some Linux distributions add by default | |
1157 | nis/nisplus even if it's not needed)? If not, just remove the entries. | |
1158 | ||
1159 | - if you need NIS/NIS+, use the Name Service Cache Daemon nscd that comes | |
1160 | with glibc 2.1. | |
1161 | ||
2ee511d9 UD |
1162 | |
1163 | 2.29. Compiling programs I get parse errors in libio.h (e.g. "parse error | |
1164 | before `_IO_seekoff'"). How should I fix this? | |
1165 | ||
1166 | {AJ} You might get the following errors when upgrading to glibc 2.1: | |
1167 | ||
1168 | In file included from /usr/include/stdio.h:57, | |
1169 | from ... | |
1170 | /usr/include/libio.h:335: parse error before `_IO_seekoff' | |
1171 | /usr/include/libio.h:335: parse error before `_G_off64_t' | |
1172 | /usr/include/libio.h:336: parse error before `_IO_seekpos' | |
1173 | /usr/include/libio.h:336: parse error before `_G_fpos64_t' | |
1174 | ||
1175 | The problem is a wrong _G_config.h file in your include path. The | |
1176 | _G_config.h file that comes with glibc 2.1 should be used and not one from | |
1177 | libc5 or from a compiler directory. To check which _G_config.h file the | |
1178 | compiler uses, compile your program with `gcc -E ...|grep G_config.h' and | |
1179 | remove that file. Your compiler should pick up the file that has been | |
1180 | installed by glibc 2.1 in your include directory. | |
1181 | ||
4f7ea427 UD |
1182 | |
1183 | 2.30. After upgrading to glibc 2.1, libraries that were compiled against | |
1184 | glibc 2.0.x don't work anymore. | |
1185 | ||
1186 | {AJ} See question 2.27. | |
1187 | ||
9de4e203 UD |
1188 | |
1189 | 2.31. What happened to the Berkeley DB libraries? Can I still use db | |
1190 | in /etc/nsswitch.conf? | |
1191 | ||
1192 | {AJ} Due to too many incompatible changes in disk layout and API of Berkeley | |
1193 | DB and a too tight coupling of libc and libdb, the db library has been | |
1194 | removed completely from glibc 2.2. The only place that really used the | |
1195 | Berkeley DB was the NSS db module. | |
1196 | ||
1197 | The NSS db module has been rewritten to support a number of different | |
1198 | versions of Berkeley DB for the NSS db module. Currently the releases 2.x | |
1199 | and 3.x of Berkeley DB are supported. The older db 1.85 library is not | |
1200 | supported. You can use the version from glibc 2.1.x or download a version | |
1201 | from Sleepycat Software (http://www.sleepycat.com). The library has to be | |
1202 | compiled as shared library and installed in the system lib directory | |
1203 | (normally /lib). The library needs to have a special soname to be found by | |
1204 | the NSS module. | |
1205 | ||
1206 | If public structures change in a new Berkeley db release, this needs to be | |
1207 | reflected in glibc. | |
1208 | ||
1209 | Currently the code searches for libraries with a soname of "libdb.so.3" | |
1210 | (that's the name from db 2.4.14 which comes with glibc 2.1.x) and | |
1211 | "libdb-3.0.so" (the name used by db 3.0.55 as default). | |
1212 | ||
61952351 UD |
1213 | \f |
1214 | . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . | |
a5a0310d | 1215 | |
61952351 | 1216 | 3. Source and binary incompatibilities, and what to do about them |
a5a0310d | 1217 | |
61952351 UD |
1218 | 3.1. I expect GNU libc to be 100% source code compatible with |
1219 | the old Linux based GNU libc. Why isn't it like this? | |
a5a0310d | 1220 | |
f12944ec UD |
1221 | {DMT,UD} Not every extension in Linux libc's history was well thought-out. |
1222 | In fact it had a lot of problems with standards compliance and with | |
1223 | cleanliness. With the introduction of a new version number these errors can | |
1224 | now be corrected. Here is a list of the known source code | |
61952351 | 1225 | incompatibilities: |
af6f3906 | 1226 | |
61952351 UD |
1227 | * _GNU_SOURCE: glibc does not make the GNU extensions available |
1228 | automatically. If a program depends on GNU extensions or some | |
1229 | other non-standard functionality, it is necessary to compile it | |
1230 | with the C compiler option -D_GNU_SOURCE, or better, to put | |
1231 | `#define _GNU_SOURCE' at the beginning of your source files, before | |
1232 | any C library header files are included. This difference normally | |
1233 | manifests itself in the form of missing prototypes and/or data type | |
1234 | definitions. Thus, if you get such errors, the first thing you | |
1235 | should do is try defining _GNU_SOURCE and see if that makes the | |
1236 | problem go away. | |
af6f3906 | 1237 | |
61952351 UD |
1238 | For more information consult the file `NOTES' in the GNU C library |
1239 | sources. | |
af6f3906 | 1240 | |
61952351 UD |
1241 | * reboot(): GNU libc sanitizes the interface of reboot() to be more |
1242 | compatible with the interface used on other OSes. reboot() as | |
1243 | implemented in glibc takes just one argument. This argument | |
1244 | corresponds to the third argument of the Linux reboot system call. | |
1245 | That is, a call of the form reboot(a, b, c) needs to be changed into | |
1246 | reboot(c). Beside this the header <sys/reboot.h> defines the needed | |
1247 | constants for the argument. These RB_* constants should be used | |
1248 | instead of the cryptic magic numbers. | |
1249 | ||
1250 | * swapon(): the interface of this function didn't change, but the | |
1251 | prototype is in a separate header file <sys/swap.h>. This header | |
1252 | file also provides the SWAP_* constants defined by <linux/swap.h>; | |
1253 | you should use them for the second argument to swapon(). | |
1254 | ||
1255 | * errno: If a program uses the variable "errno", then it _must_ | |
1256 | include <errno.h>. The old libc often (erroneously) declared this | |
1257 | variable implicitly as a side-effect of including other libc header | |
1258 | files. glibc is careful to avoid such namespace pollution, which, | |
1259 | in turn, means that you really need to include the header files that | |
1260 | you depend on. This difference normally manifests itself in the | |
1261 | form of the compiler complaining about references to an undeclared | |
1262 | symbol "errno". | |
dd7d45e8 | 1263 | |
61952351 UD |
1264 | * Linux-specific syscalls: All Linux system calls now have appropriate |
1265 | library wrappers and corresponding declarations in various header files. | |
1266 | This is because the syscall() macro that was traditionally used to | |
1267 | work around missing syscall wrappers are inherently non-portable and | |
1268 | error-prone. The following table lists all the new syscall stubs, | |
1269 | the header-file declaring their interface and the system call name. | |
dd7d45e8 | 1270 | |
61952351 UD |
1271 | syscall name: wrapper name: declaring header file: |
1272 | ------------- ------------- ---------------------- | |
1273 | bdflush bdflush <sys/kdaemon.h> | |
1274 | syslog ksyslog_ctl <sys/klog.h> | |
dd7d45e8 | 1275 | |
61952351 UD |
1276 | * lpd: Older versions of lpd depend on a routine called _validuser(). |
1277 | The library does not provide this function, but instead provides | |
1278 | __ivaliduser() which has a slightly different interface. Simply | |
1279 | upgrading to a newer lpd should fix this problem (e.g., the 4.4BSD | |
1280 | lpd is known to be working). | |
dd7d45e8 | 1281 | |
61952351 UD |
1282 | * resolver functions/BIND: like on many other systems the functions of |
1283 | the resolver library are not included in libc itself. There is a | |
1284 | separate library libresolv. If you get undefined symbol errors for | |
1285 | symbols starting with `res_*' simply add -lresolv to your linker | |
1286 | command line. | |
dd7d45e8 | 1287 | |
61952351 UD |
1288 | * the `signal' function's behavior corresponds to the BSD semantic and |
1289 | not the SysV semantic as it was in libc-5. The interface on all GNU | |
1290 | systems shall be the same and BSD is the semantic of choice. To use | |
1291 | the SysV behavior simply use `sysv_signal', or define _XOPEN_SOURCE. | |
1292 | See question 3.7 for details. | |
1cab5444 | 1293 | |
1cab5444 | 1294 | |
61952351 UD |
1295 | 3.2. Why does getlogin() always return NULL on my Linux box? |
1296 | ||
f12944ec UD |
1297 | {UD} The GNU C library has a format for the UTMP and WTMP file which differs |
1298 | from what your system currently has. It was extended to fulfill the needs | |
1299 | of the next years when IPv6 is introduced. The record size is different and | |
1300 | some fields have different positions. The files written by functions from | |
1301 | the one library cannot be read by functions from the other library. Sorry, | |
1302 | but this is what a major release is for. It's better to have a cut now than | |
1303 | having no means to support the new techniques later. | |
1cab5444 | 1304 | |
6973fc01 | 1305 | |
61952351 UD |
1306 | 3.3. Where are the DST_* constants found in <sys/time.h> on many |
1307 | systems? | |
6973fc01 | 1308 | |
f12944ec UD |
1309 | {UD} These constants come from the old BSD days and are not used anymore |
1310 | (libc5 does not actually implement the handling although the constants are | |
1311 | defined). | |
6973fc01 | 1312 | |
f12944ec | 1313 | Instead GNU libc contains zone database support and compatibility code for |
8b4a4715 | 1314 | POSIX TZ environment variable handling. For former is very much preferred |
7d1de115 | 1315 | (see question 4.3). |
6973fc01 UD |
1316 | |
1317 | ||
61952351 UD |
1318 | 3.4. The prototypes for `connect', `accept', `getsockopt', |
1319 | `setsockopt', `getsockname', `getpeername', `send', | |
1320 | `sendto', and `recvfrom' are different in GNU libc from | |
1321 | any other system I saw. This is a bug, isn't it? | |
f4017d20 | 1322 | |
f12944ec UD |
1323 | {UD} No, this is no bug. This version of GNU libc already follows the new |
1324 | Single Unix specifications (and I think the POSIX.1g draft which adopted the | |
1325 | solution). The type for a parameter describing a size is now `socklen_t', a | |
1326 | new type. | |
f4017d20 | 1327 | |
f4017d20 | 1328 | |
61952351 UD |
1329 | 3.5. On Linux I've got problems with the declarations in Linux |
1330 | kernel headers. | |
f4017d20 | 1331 | |
f12944ec UD |
1332 | {UD,AJ} On Linux, the use of kernel headers is reduced to the minimum. This |
1333 | gives Linus the ability to change the headers more freely. Also, user | |
8f1c9b09 | 1334 | programs are now insulated from changes in the size of kernel data |
f12944ec | 1335 | structures. |
f4017d20 | 1336 | |
f12944ec UD |
1337 | For example, the sigset_t type is 32 or 64 bits wide in the kernel. In |
1338 | glibc it is 1024 bits wide. This guarantees that when the kernel gets a | |
1339 | bigger sigset_t (for POSIX.1e realtime support, say) user programs will not | |
1340 | have to be recompiled. Consult the header files for more information about | |
1341 | the changes. | |
61952351 | 1342 | |
f12944ec UD |
1343 | Therefore you shouldn't include Linux kernel header files directly if glibc |
1344 | has defined a replacement. Otherwise you might get undefined results because | |
1345 | of type conflicts. | |
f4017d20 | 1346 | |
f4017d20 | 1347 | |
61952351 UD |
1348 | 3.6. I don't include any kernel headers myself but the compiler |
1349 | still complains about redeclarations of types in the kernel | |
1350 | headers. | |
1351 | ||
f12944ec UD |
1352 | {UD} The kernel headers before Linux 2.1.61 and 2.0.32 don't work correctly |
1353 | with glibc. Compiling C programs is possible in most cases but C++ programs | |
1354 | have (due to the change of the name lookups for `struct's) problems. One | |
1355 | prominent example is `struct fd_set'. | |
61952351 | 1356 | |
f12944ec UD |
1357 | There might be some problems left but 2.1.61/2.0.32 fix most of the known |
1358 | ones. See the BUGS file for other known problems. | |
61952351 UD |
1359 | |
1360 | ||
1361 | 3.7. Why don't signals interrupt system calls anymore? | |
1362 | ||
f12944ec UD |
1363 | {ZW} By default GNU libc uses the BSD semantics for signal(), unlike Linux |
1364 | libc 5 which used System V semantics. This is partially for compatibility | |
1365 | with other systems and partially because the BSD semantics tend to make | |
1366 | programming with signals easier. | |
f4017d20 UD |
1367 | |
1368 | There are three differences: | |
1369 | ||
1370 | * BSD-style signals that occur in the middle of a system call do not | |
1371 | affect the system call; System V signals cause the system call to | |
1372 | fail and set errno to EINTR. | |
1373 | ||
1374 | * BSD signal handlers remain installed once triggered. System V signal | |
1375 | handlers work only once, so one must reinstall them each time. | |
1376 | ||
1377 | * A BSD signal is blocked during the execution of its handler. In other | |
1378 | words, a handler for SIGCHLD (for example) does not need to worry about | |
61952351 | 1379 | being interrupted by another SIGCHLD. It may, however, be interrupted |
f4017d20 UD |
1380 | by other signals. |
1381 | ||
1382 | There is general consensus that for `casual' programming with signals, the | |
1383 | BSD semantics are preferable. You don't need to worry about system calls | |
1384 | returning EINTR, and you don't need to worry about the race conditions | |
1385 | associated with one-shot signal handlers. | |
1386 | ||
1387 | If you are porting an old program that relies on the old semantics, you can | |
1388 | quickly fix the problem by changing signal() to sysv_signal() throughout. | |
1389 | Alternatively, define _XOPEN_SOURCE before including <signal.h>. | |
1390 | ||
1391 | For new programs, the sigaction() function allows you to specify precisely | |
1392 | how you want your signals to behave. All three differences listed above are | |
1393 | individually switchable on a per-signal basis with this function. | |
1394 | ||
f12944ec UD |
1395 | If all you want is for one specific signal to cause system calls to fail and |
1396 | return EINTR (for example, to implement a timeout) you can do this with | |
f4017d20 UD |
1397 | siginterrupt(). |
1398 | ||
1399 | ||
61952351 UD |
1400 | 3.8. I've got errors compiling code that uses certain string |
1401 | functions. Why? | |
1402 | ||
f12944ec | 1403 | {AJ} glibc 2.1 has special string functions that are faster than the normal |
fdacb17d | 1404 | library functions. Some of the functions are additionally implemented as |
a25f2023 UD |
1405 | inline functions and others as macros. This might lead to problems with |
1406 | existing codes but it is explicitly allowed by ISO C. | |
04be94a8 | 1407 | |
04be94a8 | 1408 | The optimized string functions are only used when compiling with |
fdacb17d | 1409 | optimizations (-O1 or higher). The behavior can be changed with two feature |
f12944ec | 1410 | macros: |
61952351 UD |
1411 | |
1412 | * __NO_STRING_INLINES: Don't do any string optimizations. | |
1413 | * __USE_STRING_INLINES: Use assembly language inline functions (might | |
1414 | increase code size dramatically). | |
04be94a8 | 1415 | |
f12944ec UD |
1416 | Since some of these string functions are now additionally defined as macros, |
1417 | code like "char *strncpy();" doesn't work anymore (and is unnecessary, since | |
fdacb17d | 1418 | <string.h> has the necessary declarations). Either change your code or |
f12944ec | 1419 | define __NO_STRING_INLINES. |
04be94a8 | 1420 | |
f12944ec UD |
1421 | {UD} Another problem in this area is that gcc still has problems on machines |
1422 | with very few registers (e.g., ix86). The inline assembler code can require | |
1423 | almost all the registers and the register allocator cannot always handle | |
1424 | this situation. | |
04be94a8 | 1425 | |
61952351 | 1426 | One can disable the string optimizations selectively. Instead of writing |
04be94a8 UD |
1427 | |
1428 | cp = strcpy (foo, "lkj"); | |
1429 | ||
1430 | one can write | |
1431 | ||
1432 | cp = (strcpy) (foo, "lkj"); | |
1433 | ||
61952351 UD |
1434 | This disables the optimization for that specific call. |
1435 | ||
4775243a UD |
1436 | |
1437 | 3.9. I get compiler messages "Initializer element not constant" with | |
1438 | stdin/stdout/stderr. Why? | |
1439 | ||
1440 | {RM,AJ} Constructs like: | |
66f6a52b | 1441 | static FILE *InPtr = stdin; |
4775243a | 1442 | |
fdacb17d UD |
1443 | lead to this message. This is correct behaviour with glibc since stdin is |
1444 | not a constant expression. Please note that a strict reading of ISO C does | |
f12944ec | 1445 | not allow above constructs. |
4775243a | 1446 | |
f12944ec UD |
1447 | One of the advantages of this is that you can assign to stdin, stdout, and |
1448 | stderr just like any other global variable (e.g. `stdout = my_stream;'), | |
1449 | which can be very useful with custom streams that you can write with libio | |
fdacb17d | 1450 | (but beware this is not necessarily portable). The reason to implement it |
f12944ec | 1451 | this way were versioning problems with the size of the FILE structure. |
4775243a | 1452 | |
fdacb17d UD |
1453 | To fix those programs you've got to initialize the variable at run time. |
1454 | This can be done, e.g. in main, like: | |
1455 | ||
66f6a52b UD |
1456 | static FILE *InPtr; |
1457 | int main(void) | |
1458 | { | |
1459 | InPtr = stdin; | |
1460 | } | |
fdacb17d UD |
1461 | |
1462 | or by constructors (beware this is gcc specific): | |
1463 | ||
66f6a52b UD |
1464 | static FILE *InPtr; |
1465 | static void inPtr_construct (void) __attribute__((constructor)); | |
1466 | static void inPtr_construct (void) { InPtr = stdin; } | |
fdacb17d | 1467 | |
4775243a UD |
1468 | |
1469 | 3.10. I can't compile with gcc -traditional (or | |
1470 | -traditional-cpp). Why? | |
1471 | ||
1472 | {AJ} glibc2 does break -traditional and -traditonal-cpp - and will continue | |
fdacb17d | 1473 | to do so. For example constructs of the form: |
f12944ec | 1474 | |
66f6a52b UD |
1475 | enum {foo |
1476 | #define foo foo | |
1477 | } | |
f12944ec UD |
1478 | |
1479 | are useful for debugging purposes (you can use foo with your debugger that's | |
1480 | why we need the enum) and for compatibility (other systems use defines and | |
1481 | check with #ifdef). | |
4775243a UD |
1482 | |
1483 | ||
1484 | 3.11. I get some errors with `gcc -ansi'. Isn't glibc ANSI compatible? | |
1485 | ||
1486 | {AJ} The GNU C library is compatible with the ANSI/ISO C standard. If | |
f12944ec | 1487 | you're using `gcc -ansi', the glibc includes which are specified in the |
fdacb17d | 1488 | standard follow the standard. The ANSI/ISO C standard defines what has to be |
f12944ec UD |
1489 | in the include files - and also states that nothing else should be in the |
1490 | include files (btw. you can still enable additional standards with feature | |
1491 | flags). | |
4775243a | 1492 | |
f12944ec UD |
1493 | The GNU C library is conforming to ANSI/ISO C - if and only if you're only |
1494 | using the headers and library functions defined in the standard. | |
4775243a | 1495 | |
a35cb74d UD |
1496 | |
1497 | 3.12. I can't access some functions anymore. nm shows that they do | |
1498 | exist but linking fails nevertheless. | |
1499 | ||
f12944ec UD |
1500 | {AJ} With the introduction of versioning in glibc 2.1 it is possible to |
1501 | export only those identifiers (functions, variables) that are really needed | |
1502 | by application programs and by other parts of glibc. This way a lot of | |
1503 | internal interfaces are now hidden. nm will still show those identifiers | |
1504 | but marking them as internal. ISO C states that identifiers beginning with | |
1505 | an underscore are internal to the libc. An application program normally | |
1506 | shouldn't use those internal interfaces (there are exceptions, | |
1507 | e.g. __ivaliduser). If a program uses these interfaces, it's broken. These | |
1508 | internal interfaces might change between glibc releases or dropped | |
1509 | completely. | |
a35cb74d | 1510 | |
a5f4e34a UD |
1511 | |
1512 | 3.13. When using the db-2 library which comes with glibc is used in | |
1513 | the Perl db modules the testsuite is not passed. This did not | |
1514 | happen with db-1, gdbm, or ndbm. | |
1515 | ||
9de4e203 UD |
1516 | Removed. Does not apply anymore. |
1517 | ||
1518 | ||
a5f4e34a | 1519 | |
5148d49f UD |
1520 | |
1521 | 3.14. The pow() inline function I get when including <math.h> is broken. | |
1522 | I get segmentation faults when I run the program. | |
1523 | ||
1524 | {UD} Nope, the implementation is correct. The problem is with egcs version | |
1525 | prior to 1.1. I.e., egcs 1.0 to 1.0.3 are all broken (at least on Intel). | |
1526 | If you have to use this compiler you must define __NO_MATH_INLINES before | |
1527 | including <math.h> to prevent the inline functions from being used. egcs 1.1 | |
1528 | fixes the problem. I don't know about gcc 2.8 and 2.8.1. | |
1529 | ||
05f732b3 UD |
1530 | |
1531 | 3.15. The sys/sem.h file lacks the definition of `union semun'. | |
1532 | ||
1533 | {UD} Nope. This union has to be provided by the user program. Former glibc | |
1534 | versions defined this but it was an error since it does not make much sense | |
1535 | when thinking about it. The standards describing the System V IPC functions | |
1536 | define it this way and therefore programs must be adopted. | |
1537 | ||
33127459 UD |
1538 | |
1539 | 3.16. Why has <netinet/ip_fw.h> disappeared? | |
1540 | ||
1541 | {AJ} The corresponding Linux kernel data structures and constants are | |
b710a6e2 | 1542 | totally different in Linux 2.0 and Linux 2.2. This situation has to be |
33127459 UD |
1543 | taken care in user programs using the firewall structures and therefore |
1544 | those programs (ipfw is AFAIK the only one) should deal with this problem | |
1545 | themselves. | |
1546 | ||
28ab8526 UD |
1547 | |
1548 | 3.17. I get floods of warnings when I use -Wconversion and include | |
1549 | <string.h> or <math.h>. | |
1550 | ||
1551 | {ZW} <string.h> and <math.h> intentionally use prototypes to override | |
1552 | argument promotion. -Wconversion warns about all these. You can safely | |
1553 | ignore the warnings. | |
1554 | ||
1555 | -Wconversion isn't really intended for production use, only for shakedown | |
1556 | compiles after converting an old program to standard C. | |
1557 | ||
5ff1a70a UD |
1558 | |
1559 | 3.18. After upgrading to glibc 2.1, I receive errors about | |
1560 | unresolved symbols, like `_dl_initial_searchlist' and can not | |
1561 | execute any binaries. What went wrong? | |
1562 | ||
1563 | {AJ} This normally happens if your libc and ld (dynamic linker) are from | |
1564 | different releases of glibc. For example, the dynamic linker | |
1565 | /lib/ld-linux.so.2 comes from glibc 2.0.x, but the version of libc.so.6 is | |
1566 | from glibc 2.1. | |
1567 | ||
1568 | The path /lib/ld-linux.so.2 is hardcoded in every glibc2 binary but | |
1569 | libc.so.6 is searched via /etc/ld.so.cache and in some special directories | |
1570 | like /lib and /usr/lib. If you run configure with another prefix than /usr | |
1571 | and put this prefix before /lib in /etc/ld.so.conf, your system will break. | |
1572 | ||
1573 | So what can you do? Either of the following should work: | |
1574 | ||
1575 | * Run `configure' with the same prefix argument you've used for glibc 2.0.x | |
1576 | so that the same paths are used. | |
1577 | * Replace /lib/ld-linux.so.2 with a link to the dynamic linker from glibc | |
1578 | 2.1. | |
1579 | ||
1580 | You can even call the dynamic linker by hand if everything fails. You've | |
1581 | got to set LD_LIBRARY_PATH so that the corresponding libc is found and also | |
1582 | need to provide an absolute path to your binary: | |
1583 | ||
1584 | LD_LIBRARY_PATH=<path-where-libc.so.6-lives> \ | |
1585 | <path-where-corresponding-dynamic-linker-lives>/ld-linux.so.2 \ | |
1586 | <path-to-binary>/binary | |
1587 | ||
1588 | For example `LD_LIBRARY_PATH=/libold /libold/ld-linux.so.2 /bin/mv ...' | |
1589 | might be useful in fixing a broken system (if /libold contains dynamic | |
1590 | linker and corresponding libc). | |
1591 | ||
1592 | With that command line no path is used. To further debug problems with the | |
1593 | dynamic linker, use the LD_DEBUG environment variable, e.g. | |
1594 | `LD_DEBUG=help echo' for the help text. | |
1595 | ||
1596 | If you just want to test this release, don't put the lib directory in | |
1597 | /etc/ld.so.conf. You can call programs directly with full paths (as above). | |
1598 | When compiling new programs against glibc 2.1, you've got to specify the | |
1599 | correct paths to the compiler (option -I with gcc) and linker (options | |
1600 | --dynamic-linker, -L and --rpath). | |
1601 | ||
61952351 | 1602 | |
7d1de115 UD |
1603 | 3.19. bonnie reports that char i/o with glibc 2 is much slower than with |
1604 | libc5. What can be done? | |
1605 | ||
1606 | {AJ} The GNU C library uses thread safe functions by default and libc5 used | |
1607 | non thread safe versions. The non thread safe functions have in glibc the | |
1608 | suffix `_unlocked', for details check <stdio.h>. Using `putc_unlocked' etc. | |
1609 | instead of `putc' should give nearly the same speed with bonnie (bonnie is a | |
1610 | benchmark program for measuring disk access). | |
9f6b6d8d | 1611 | |
b93492aa UD |
1612 | |
1613 | 3.20. Programs compiled with glibc 2.1 can't read db files made with glibc | |
1614 | 2.0. What has changed that programs like rpm break? | |
1615 | ||
9de4e203 UD |
1616 | Removed. Does not apply anymore. |
1617 | ||
b93492aa | 1618 | |
b93492aa | 1619 | |
b5a9efcd UD |
1620 | |
1621 | 3.21. Autoconf's AC_CHECK_FUNC macro reports that a function exists, but | |
1622 | when I try to use it, it always returns -1 and sets errno to ENOSYS. | |
1623 | ||
1624 | {ZW} You are using a 2.0 Linux kernel, and the function you are trying to | |
1625 | use is only implemented in 2.1/2.2. Libc considers this to be a function | |
1626 | which exists, because if you upgrade to a 2.2 kernel, it will work. One | |
1627 | such function is sigaltstack. | |
1628 | ||
1629 | Your program should check at runtime whether the function works, and | |
1630 | implement a fallback. Note that Autoconf cannot detect unimplemented | |
1631 | functions in other systems' C libraries, so you need to do this anyway. | |
1632 | ||
1633 | ||
1634 | 3.22. My program segfaults when I call fclose() on the FILE* returned | |
1635 | from setmntent(). Is this a glibc bug? | |
1636 | ||
1637 | {GK} No. Don't do this. Use endmntent(), that's what it's for. | |
1638 | ||
1639 | In general, you should use the correct deallocation routine. For instance, | |
1640 | if you open a file using fopen(), you should deallocate the FILE * using | |
1641 | fclose(), not free(), even though the FILE * is also a pointer. | |
1642 | ||
1643 | In the case of setmntent(), it may appear to work in most cases, but it | |
1644 | won't always work. Unfortunately, for compatibility reasons, we can't | |
1645 | change the return type of setmntent() to something other than FILE *. | |
1646 | ||
9f6b6d8d UD |
1647 | \f |
1648 | . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . | |
1649 | ||
7d1de115 | 1650 | 4. Miscellaneous |
61952351 | 1651 | |
7d1de115 | 1652 | 4.1. After I changed configure.in I get `Autoconf version X.Y. |
61952351 UD |
1653 | or higher is required for this script'. What can I do? |
1654 | ||
1655 | {UD} You have to get the specified autoconf version (or a later one) | |
a35cb74d | 1656 | from your favorite mirror of ftp.gnu.org. |
61952351 | 1657 | |
04be94a8 | 1658 | |
7d1de115 | 1659 | 4.2. When I try to compile code which uses IPv6 headers and |
61952351 UD |
1660 | definitions on my Linux 2.x.y system I am in trouble. |
1661 | Nothing seems to work. | |
1662 | ||
f12944ec UD |
1663 | {UD} The problem is that IPv6 development still has not reached a point |
1664 | where the headers are stable. There are still lots of incompatible changes | |
1665 | made and the libc headers have to follow. | |
61952351 | 1666 | |
348ed515 UD |
1667 | {PB} The 2.1 release of GNU libc aims to comply with the current versions of |
1668 | all the relevant standards. The IPv6 support libraries for older Linux | |
1669 | systems used a different naming convention and so code written to work with | |
1670 | them may need to be modified. If the standards make incompatible changes in | |
1671 | the future then the libc may need to change again. | |
1672 | ||
1673 | IPv6 will not work with a 2.0.x kernel. When kernel 2.2 is released it | |
1674 | should contain all the necessary support; until then you should use the | |
1675 | latest 2.1.x release you can find. As of 98/11/26 the currently recommended | |
1676 | kernel for IPv6 is 2.1.129. | |
1677 | ||
1678 | Also, as of the 2.1 release the IPv6 API provided by GNU libc is not | |
b669ab02 | 1679 | 100% complete. |
04be94a8 | 1680 | |
ff44f2a5 | 1681 | |
7d1de115 | 1682 | 4.3. When I set the timezone by setting the TZ environment variable |
ff44f2a5 UD |
1683 | to EST5EDT things go wrong since glibc computes the wrong time |
1684 | from this information. | |
1685 | ||
f12944ec UD |
1686 | {UD} The problem is that people still use the braindamaged POSIX method to |
1687 | select the timezone using the TZ environment variable with a format EST5EDT | |
8b4a4715 UD |
1688 | or whatever. People, if you insist on using TZ instead of the timezone |
1689 | database (see below), read the POSIX standard, the implemented behaviour is | |
f12944ec UD |
1690 | correct! What you see is in fact the result of the decisions made while |
1691 | POSIX.1 was created. We've only implemented the handling of TZ this way to | |
1692 | be POSIX compliant. It is not really meant to be used. | |
1693 | ||
1694 | The alternative approach to handle timezones which is implemented is the | |
1695 | correct one to use: use the timezone database. This avoids all the problems | |
1696 | the POSIX method has plus it is much easier to use. Simply run the tzselect | |
1697 | shell script, answer the question and use the name printed in the end by | |
8b4a4715 UD |
1698 | making a symlink /etc/localtime pointing to /usr/share/zoneinfo/NAME (NAME |
1699 | is the returned value from tzselect). That's all. You never again have to | |
1700 | worry. | |
f12944ec UD |
1701 | |
1702 | So, please avoid sending bug reports about time related problems if you use | |
1703 | the POSIX method and you have not verified something is really broken by | |
1704 | reading the POSIX standards. | |
ff44f2a5 | 1705 | |
fdacb17d | 1706 | |
7d1de115 | 1707 | 4.4. What other sources of documentation about glibc are available? |
fdacb17d UD |
1708 | |
1709 | {AJ} The FSF has a page about the GNU C library at | |
1710 | <http://www.gnu.org/software/libc/>. The problem data base of open and | |
1711 | solved bugs in GNU libc is available at | |
1712 | <http://www-gnats.gnu.org:8080/cgi-bin/wwwgnats.pl>. Eric Green has written | |
9de4e203 | 1713 | a HowTo for converting from Linux libc5 to glibc2. The HowTo is accessible |
fdacb17d UD |
1714 | via the FSF page and at <http://www.imaxx.net/~thrytis/glibc>. Frodo |
1715 | Looijaard describes a different way installing glibc2 as secondary libc at | |
1716 | <http://huizen.dds.nl/~frodol/glibc>. | |
1717 | ||
1718 | Please note that this is not a complete list. | |
1719 | ||
348ed515 | 1720 | |
7d1de115 | 1721 | 4.5. The timezone string for Sydney/Australia is wrong since even when |
348ed515 UD |
1722 | daylight saving time is in effect the timezone string is EST. |
1723 | ||
1724 | {UD} The problem for some timezones is that the local authorities decided | |
1725 | to use the term "summer time" instead of "daylight saving time". In this | |
1726 | case the abbreviation character `S' is the same as the standard one. So, | |
1727 | for Sydney we have | |
1728 | ||
1729 | Eastern Standard Time = EST | |
1730 | Eastern Summer Time = EST | |
1731 | ||
1732 | Great! To get this bug fixed convince the authorities to change the laws | |
1733 | and regulations of the country this effects. glibc behaves correctly. | |
1734 | ||
eeabe877 | 1735 | |
7d1de115 | 1736 | 4.6. I've build make 3.77 against glibc 2.1 and now make gets |
eeabe877 UD |
1737 | segmentation faults. |
1738 | ||
2bbc70d5 AJ |
1739 | Removed. Does not apply anymore, use make 3.79 or newer. |
1740 | ||
1741 | ||
eeabe877 | 1742 | |
c63598bf UD |
1743 | |
1744 | 4.7. Why do so many programs using math functions fail on my AlphaStation? | |
1745 | ||
1746 | {AO} The functions floor() and floorf() use an instruction that is not | |
1747 | implemented in some old PALcodes of AlphaStations. This may cause | |
1748 | `Illegal Instruction' core dumps or endless loops in programs that | |
1749 | catch these signals. Updating the firmware to a 1999 release has | |
1750 | fixed the problem on an AlphaStation 200 4/166. | |
1751 | ||
8892c471 UD |
1752 | |
1753 | 4.8. The conversion table for character set XX does not match with | |
1754 | what I expect. | |
1755 | ||
1756 | {UD} I don't doubt for a minute that some of the conversion tables contain | |
1757 | errors. We tried the best we can and relied on automatic generation of the | |
1758 | data to prevent human-introduced errors but this still is no guarantee. If | |
1759 | you think you found a problem please send a bug report describing it and | |
1760 | give an authoritive reference. The latter is important since otherwise | |
1761 | the current behaviour is as good as the proposed one. | |
1762 | ||
1763 | Before doing this look through the list of known problem first: | |
1764 | ||
1765 | - the GBK (simplified Chinese) encoding is based on Unicode tables. This | |
1766 | is good. These tables, however, differ slightly from the tables used | |
1767 | by the M$ people. The differences are these [+ Unicode, - M$]: | |
1768 | ||
1769 | +0xA1AA 0x2015 | |
1770 | +0xA844 0x2014 | |
1771 | -0xA1AA 0x2014 | |
1772 | -0xA844 0x2015 | |
1773 | ||
1774 | In addition the Unicode tables contain mappings for the GBK characters | |
1775 | 0xA8BC, 0xA8BF, 0xA989 to 0xA995, and 0xFE50 to 0xFEA0. | |
1776 | ||
ffa156af UD |
1777 | - when mapping from EUC-CN to GBK and vice versa we ignore the fact that |
1778 | the coded character at position 0xA1A4 maps to different Unicode | |
1779 | characters. Since the iconv() implementation can do whatever it wants | |
1780 | if it cannot directly map a character this is a perfectly good solution | |
1781 | since the semantics and appearance of the character does not change. | |
1782 | ||
f8cac037 | 1783 | \f |
61952351 UD |
1784 | ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ |
1785 | ||
f8cac037 RM |
1786 | Answers were given by: |
1787 | {UD} Ulrich Drepper, <drepper@cygnus.com> | |
613a76ff | 1788 | {DMT} David Mosberger-Tang, <davidm@AZStarNet.com> |
dd7d45e8 | 1789 | {RM} Roland McGrath, <roland@gnu.org> |
9de4e203 | 1790 | {AJ} Andreas Jaeger, <aj@suse.de> |
22d57dd3 | 1791 | {EY} Eric Youngdale, <eric@andante.jic.com> |
a5a0310d | 1792 | {PB} Phil Blundell, <Philip.Blundell@pobox.com> |
af6f3906 | 1793 | {MK} Mark Kettenis, <kettenis@phys.uva.nl> |
f4017d20 | 1794 | {ZW} Zack Weinberg, <zack@rabi.phys.columbia.edu> |
612fdf25 | 1795 | {TK} Thorsten Kukuk, <kukuk@suse.de> |
8619129f | 1796 | {GK} Geoffrey Keating, <geoffk@ozemail.com.au> |
a35cb74d | 1797 | {HJ} H.J. Lu, <hjl@gnu.org> |
b710a6e2 | 1798 | {CG} Cristian Gafton, <gafton@redhat.com> |
c63598bf | 1799 | {AO} Alexandre Oliva, <oliva@lsd.ic.unicamp.br> |
f8cac037 RM |
1800 | \f |
1801 | Local Variables: | |
61952351 UD |
1802 | mode:outline |
1803 | outline-regexp:"\\?" | |
f12944ec | 1804 | fill-column:76 |
f8cac037 | 1805 | End: |