![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to gcc.info-10 CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [Apple XNU] / GNUtools / cc|
Return to gcc.info-10 CVS log | Up to [Apple XNU] / GNUtools / cc |
1.1 root 1: This is Info file gcc.info, produced by Makeinfo-1.54 from the input
2: file gcc.texi.
3:
4: This file documents the use and the internals of the GNU compiler.
5:
6: Published by the Free Software Foundation 675 Massachusetts Avenue
7: Cambridge, MA 02139 USA
8:
9: Copyright (C) 1988, 1989, 1992, 1993 Free Software Foundation, Inc.
10:
11: Permission is granted to make and distribute verbatim copies of this
12: manual provided the copyright notice and this permission notice are
13: preserved on all copies.
14:
15: Permission is granted to copy and distribute modified versions of
16: this manual under the conditions for verbatim copying, provided also
17: that the sections entitled "GNU General Public License" and "Protect
18: Your Freedom--Fight `Look And Feel'" are included exactly as in the
19: original, and provided that the entire resulting derived work is
20: distributed under the terms of a permission notice identical to this
21: one.
22:
23: Permission is granted to copy and distribute translations of this
24: manual into another language, under the above conditions for modified
25: versions, except that the sections entitled "GNU General Public
26: License" and "Protect Your Freedom--Fight `Look And Feel'", and this
27: permission notice, may be included in translations approved by the Free
28: Software Foundation instead of in the original English.
29:
30:
31: File: gcc.info, Node: Bug Reporting, Next: Sending Patches, Prev: Bug Lists, Up: Bugs
32:
33: How to Report Bugs
34: ==================
35:
36: The fundamental principle of reporting bugs usefully is this:
37: *report all the facts*. If you are not sure whether to state a fact or
38: leave it out, state it!
39:
40: Often people omit facts because they think they know what causes the
41: problem and they conclude that some details don't matter. Thus, you
42: might assume that the name of the variable you use in an example does
43: not matter. Well, probably it doesn't, but one cannot be sure.
44: Perhaps the bug is a stray memory reference which happens to fetch from
45: the location where that name is stored in memory; perhaps, if the name
46: were different, the contents of that location would fool the compiler
47: into doing the right thing despite the bug. Play it safe and give a
48: specific, complete example. That is the easiest thing for you to do,
49: and the most helpful.
50:
51: Keep in mind that the purpose of a bug report is to enable someone to
52: fix the bug if it is not known. It isn't very important what happens if
53: the bug is already known. Therefore, always write your bug reports on
54: the assumption that the bug is not known.
55:
56: Sometimes people give a few sketchy facts and ask, "Does this ring a
57: bell?" This cannot help us fix a bug, so it is basically useless. We
58: respond by asking for enough details to enable us to investigate. You
59: might as well expedite matters by sending them to begin with.
60:
61: Try to make your bug report self-contained. If we have to ask you
62: for more information, it is best if you include all the previous
63: information in your response, as well as the information that was
64: missing.
65:
66: To enable someone to investigate the bug, you should include all
67: these things:
68:
69: * The version of GNU CC. You can get this by running it with the
70: `-v' option.
71:
72: Without this, we won't know whether there is any point in looking
73: for the bug in the current version of GNU CC.
74:
75: * A complete input file that will reproduce the bug. If the bug is
76: in the C preprocessor, send a source file and any header files
77: that it requires. If the bug is in the compiler proper (`cc1'),
78: run your source file through the C preprocessor by doing `gcc -E
79: SOURCEFILE > OUTFILE', then include the contents of OUTFILE in the
80: bug report. (When you do this, use the same `-I', `-D' or `-U'
81: options that you used in actual compilation.)
82:
83: A single statement is not enough of an example. In order to
84: compile it, it must be embedded in a complete file of compiler
85: input; and the bug might depend on the details of how this is done.
86:
87: Without a real example one can compile, all anyone can do about
88: your bug report is wish you luck. It would be futile to try to
89: guess how to provoke the bug. For example, bugs in register
90: allocation and reloading frequently depend on every little detail
91: of the function they happen in.
92:
93: Even if the input file that fails comes from a GNU program, you
94: should still send the complete test case. Don't ask the GNU CC
95: maintainers to do the extra work of obtaining the program in
96: question--they are all overworked as it is. Also, the problem may
97: depend on what is in the header files on your system; it is
98: unreliable for the GNU CC maintainers to try the problem with the
99: header files available to them. By sending CPP output, you can
100: eliminate this source of uncertainty and save us a certain
101: percentage of wild goose chases.
102:
103: * The command arguments you gave GNU CC or GNU C++ to compile that
104: example and observe the bug. For example, did you use `-O'? To
105: guarantee you won't omit something important, list all the options.
106:
107: If we were to try to guess the arguments, we would probably guess
108: wrong and then we would not encounter the bug.
109:
110: * The type of machine you are using, and the operating system name
111: and version number.
112:
113: * The operands you gave to the `configure' command when you installed
114: the compiler.
115:
116: * A complete list of any modifications you have made to the compiler
117: source. (We don't promise to investigate the bug unless it
118: happens in an unmodified compiler. But if you've made
119: modifications and don't tell us, then you are sending us on a wild
120: goose chase.)
121:
122: Be precise about these changes. A description in English is not
123: enough--send a context diff for them.
124:
125: Adding files of your own (such as a machine description for a
126: machine we don't support) is a modification of the compiler source.
127:
128: * Details of any other deviations from the standard procedure for
129: installing GNU CC.
130:
131: * A description of what behavior you observe that you believe is
132: incorrect. For example, "The compiler gets a fatal signal," or,
133: "The assembler instruction at line 208 in the output is incorrect."
134:
135: Of course, if the bug is that the compiler gets a fatal signal,
136: then one can't miss it. But if the bug is incorrect output, the
137: maintainer might not notice unless it is glaringly wrong. None of
138: us has time to study all the assembler code from a 50-line C
139: program just on the chance that one instruction might be wrong.
140: We need *you* to do this part!
141:
142: Even if the problem you experience is a fatal signal, you should
143: still say so explicitly. Suppose something strange is going on,
144: such as, your copy of the compiler is out of synch, or you have
145: encountered a bug in the C library on your system. (This has
146: happened!) Your copy might crash and the copy here would not. If
147: you said to expect a crash, then when the compiler here fails to
148: crash, we would know that the bug was not happening. If you don't
149: say to expect a crash, then we would not know whether the bug was
150: happening. We would not be able to draw any conclusion from our
151: observations.
152:
153: If the problem is a diagnostic when compiling GNU CC with some
154: other compiler, say whether it is a warning or an error.
155:
156: Often the observed symptom is incorrect output when your program
157: is run. Sad to say, this is not enough information unless the
158: program is short and simple. None of us has time to study a large
159: program to figure out how it would work if compiled correctly,
160: much less which line of it was compiled wrong. So you will have
161: to do that. Tell us which source line it is, and what incorrect
162: result happens when that line is executed. A person who
163: understands the program can find this as easily as finding a bug
164: in the program itself.
165:
166: * If you send examples of assembler code output from GNU CC or GNU
167: C++, please use `-g' when you make them. The debugging information
168: includes source line numbers which are essential for correlating
169: the output with the input.
170:
171: * If you wish to mention something in the GNU CC source, refer to it
172: by context, not by line number.
173:
174: The line numbers in the development sources don't match those in
175: your sources. Your line numbers would convey no useful
176: information to the maintainers.
177:
178: * Additional information from a debugger might enable someone to
179: find a problem on a machine which he does not have available.
180: However, you need to think when you collect this information if
181: you want it to have any chance of being useful.
182:
183: For example, many people send just a backtrace, but that is never
184: useful by itself. A simple backtrace with arguments conveys little
185: about GNU CC because the compiler is largely data-driven; the same
186: functions are called over and over for different RTL insns, doing
187: different things depending on the details of the insn.
188:
189: Most of the arguments listed in the backtrace are useless because
190: they are pointers to RTL list structure. The numeric values of the
191: pointers, which the debugger prints in the backtrace, have no
192: significance whatever; all that matters is the contents of the
193: objects they point to (and most of the contents are other such
194: pointers).
195:
196: In addition, most compiler passes consist of one or more loops that
197: scan the RTL insn sequence. The most vital piece of information
198: about such a loop--which insn it has reached--is usually in a
199: local variable, not in an argument.
200:
201: What you need to provide in addition to a backtrace are the values
202: of the local variables for several stack frames up. When a local
203: variable or an argument is an RTX, first print its value and then
204: use the GDB command `pr' to print the RTL expression that it points
205: to. (If GDB doesn't run on your machine, use your debugger to call
206: the function `debug_rtx' with the RTX as an argument.) In
207: general, whenever a variable is a pointer, its value is no use
208: without the data it points to.
209:
210: Here are some things that are not necessary:
211:
212: * A description of the envelope of the bug.
213:
214: Often people who encounter a bug spend a lot of time investigating
215: which changes to the input file will make the bug go away and which
216: changes will not affect it.
217:
218: This is often time consuming and not very useful, because the way
219: we will find the bug is by running a single example under the
220: debugger with breakpoints, not by pure deduction from a series of
221: examples. You might as well save your time for something else.
222:
223: Of course, if you can find a simpler example to report *instead* of
224: the original one, that is a convenience. Errors in the output
225: will be easier to spot, running under the debugger will take less
226: time, etc. Most GNU CC bugs involve just one function, so the
227: most straightforward way to simplify an example is to delete all
228: the function definitions except the one where the bug occurs.
229: Those earlier in the file may be replaced by external declarations
230: if the crucial function depends on them. (Exception: inline
231: functions may affect compilation of functions defined later in the
232: file.)
233:
234: However, simplification is not vital; if you don't want to do this,
235: report the bug anyway and send the entire test case you used.
236:
237: * In particular, some people insert conditionals `#ifdef BUG' around
238: a statement which, if removed, makes the bug not happen. These
239: are just clutter; we won't pay any attention to them anyway.
240: Besides, you should send us cpp output, and that can't have
241: conditionals.
242:
243: * A patch for the bug.
244:
245: A patch for the bug is useful if it is a good one. But don't omit
246: the necessary information, such as the test case, on the
247: assumption that a patch is all we need. We might see problems
248: with your patch and decide to fix the problem another way, or we
249: might not understand it at all.
250:
251: Sometimes with a program as complicated as GNU CC it is very hard
252: to construct an example that will make the program follow a
253: certain path through the code. If you don't send the example, we
254: won't be able to construct one, so we won't be able to verify that
255: the bug is fixed.
256:
257: And if we can't understand what bug you are trying to fix, or why
258: your patch should be an improvement, we won't install it. A test
259: case will help us to understand.
260:
261: *Note Sending Patches::, for guidelines on how to make it easy for
262: us to understand and install your patches.
263:
264: * A guess about what the bug is or what it depends on.
265:
266: Such guesses are usually wrong. Even I can't guess right about
267: such things without first using the debugger to find the facts.
268:
269: * A core dump file.
270:
271: We have no way of examining a core dump for your type of machine
272: unless we have an identical system--and if we do have one, we
273: should be able to reproduce the crash ourselves.
274:
275:
276: File: gcc.info, Node: Sending Patches, Prev: Bug Reporting, Up: Bugs
277:
278: Sending Patches for GNU CC
279: ==========================
280:
281: If you would like to write bug fixes or improvements for the GNU C
282: compiler, that is very helpful. When you send your changes, please
283: follow these guidelines to avoid causing extra work for us in studying
284: the patches.
285:
286: If you don't follow these guidelines, your information might still be
287: useful, but using it will take extra work. Maintaining GNU C is a lot
288: of work in the best of circumstances, and we can't keep up unless you do
289: your best to help.
290:
291: * Send an explanation with your changes of what problem they fix or
292: what improvement they bring about. For a bug fix, just include a
293: copy of the bug report, and explain why the change fixes the bug.
294:
295: (Referring to a bug report is not as good as including it, because
296: then we will have to look it up, and we have probably already
297: deleted it if we've already fixed the bug.)
298:
299: * Always include a proper bug report for the problem you think you
300: have fixed. We need to convince ourselves that the change is
301: right before installing it. Even if it is right, we might have
302: trouble judging it if we don't have a way to reproduce the problem.
303:
304: * Include all the comments that are appropriate to help people
305: reading the source in the future understand why this change was
306: needed.
307:
308: * Don't mix together changes made for different reasons. Send them
309: *individually*.
310:
311: If you make two changes for separate reasons, then we might not
312: want to install them both. We might want to install just one. If
313: you send them all jumbled together in a single set of diffs, we
314: have to do extra work to disentangle them--to figure out which
315: parts of the change serve which purpose. If we don't have time
316: for this, we might have to ignore your changes entirely.
317:
318: If you send each change as soon as you have written it, with its
319: own explanation, then the two changes never get tangled up, and we
320: can consider each one properly without any extra work to
321: disentangle them.
322:
323: Ideally, each change you send should be impossible to subdivide
324: into parts that we might want to consider separately, because each
325: of its parts gets its motivation from the other parts.
326:
327: * Send each change as soon as that change is finished. Sometimes
328: people think they are helping us by accumulating many changes to
329: send them all together. As explained above, this is absolutely
330: the worst thing you could do.
331:
332: Since you should send each change separately, you might as well
333: send it right away. That gives us the option of installing it
334: immediately if it is important.
335:
336: * Use `diff -c' to make your diffs. Diffs without context are hard
337: for us to install reliably. More than that, they make it hard for
338: us to study the diffs to decide whether we want to install them.
339: Unidiff format is better than contextless diffs, but not as easy
340: to read as `-c' format.
341:
342: If you have GNU diff, use `diff -cp', which shows the name of the
343: function that each change occurs in.
344:
345: * Write the change log entries for your changes. We get lots of
346: changes, and we don't have time to do all the change log writing
347: ourselves.
348:
349: Read the `ChangeLog' file to see what sorts of information to put
350: in, and to learn the style that we use. The purpose of the change
351: log is to show people where to find what was changed. So you need
352: to be specific about what functions you changed; in large
353: functions, it's often helpful to indicate where within the
354: function the change was.
355:
356: On the other hand, once you have shown people where to find the
357: change, you need not explain its purpose. Thus, if you add a new
358: function, all you need to say about it is that it is new. If you
359: feel that the purpose needs explaining, it probably does--but the
360: explanation will be much more useful if you put it in comments in
361: the code.
362:
363: If you would like your name to appear in the header line for who
364: made the change, send us the header line.
365:
366: * When you write the fix, keep in mind that we can't install a
367: change that would break other systems.
368:
369: People often suggest fixing a problem by changing
370: machine-independent files such as `toplev.c' to do something
371: special that a particular system needs. Sometimes it is totally
372: obvious that such changes would break GNU CC for almost all users.
373: We can't possibly make a change like that. At best it might tell
374: us how to write another patch that would solve the problem
375: acceptably.
376:
377: Sometimes people send fixes that *might* be an improvement in
378: general--but it is hard to be sure of this. It's hard to install
379: such changes because we have to study them very carefully. Of
380: course, a good explanation of the reasoning by which you concluded
381: the change was correct can help convince us.
382:
383: The safest changes are changes to the configuration files for a
384: particular machine. These are safe because they can't create new
385: bugs on other machines.
386:
387: Please help us keep up with the workload by designing the patch in
388: a form that is good to install.
389:
390:
391: File: gcc.info, Node: Service, Next: VMS, Prev: Bugs, Up: Top
392:
393: How To Get Help with GNU CC
394: ***************************
395:
396: If you need help installing, using or changing GNU CC, there are two
397: ways to find it:
398:
399: * Send a message to a suitable network mailing list. First try
400: `[email protected]', and if that brings no response, try
401: `[email protected]'.
402:
403: * Look in the service directory for someone who might help you for a
404: fee. The service directory is found in the file named `SERVICE'
405: in the GNU CC distribution.
406:
407:
408: File: gcc.info, Node: VMS, Next: Portability, Prev: Service, Up: Top
409:
410: Using GNU CC on VMS
411: *******************
412:
413: * Menu:
414:
415: * Include Files and VMS:: Where the preprocessor looks for the include files.
416: * Global Declarations:: How to do globaldef, globalref and globalvalue with
417: GNU CC.
418: * VMS Misc:: Misc information.
419:
420:
421: File: gcc.info, Node: Include Files and VMS, Next: Global Declarations, Up: VMS
422:
423: Include Files and VMS
424: =====================
425:
426: Due to the differences between the filesystems of Unix and VMS, GNU
427: CC attempts to translate file names in `#include' into names that VMS
428: will understand. The basic strategy is to prepend a prefix to the
429: specification of the include file, convert the whole filename to a VMS
430: filename, and then try to open the file. GNU CC tries various prefixes
431: one by one until one of them succeeds:
432:
433: 1. The first prefix is the `GNU_CC_INCLUDE:' logical name: this is
434: where GNU C header files are traditionally stored. If you wish to
435: store header files in non-standard locations, then you can assign
436: the logical `GNU_CC_INCLUDE' to be a search list, where each
437: element of the list is suitable for use with a rooted logical.
438:
439: 2. The next prefix tried is `SYS$SYSROOT:[SYSLIB.]'. This is where
440: VAX-C header files are traditionally stored.
441:
442: 3. If the include file specification by itself is a valid VMS
443: filename, the preprocessor then uses this name with no prefix in
444: an attempt to open the include file.
445:
446: 4. If the file specification is not a valid VMS filename (i.e. does
447: not contain a device or a directory specifier, and contains a `/'
448: character), the preprocessor tries to convert it from Unix syntax
449: to VMS syntax.
450:
451: Conversion works like this: the first directory name becomes a
452: device, and the rest of the directories are converted into
453: VMS-format directory names. For example, the name `X11/foobar.h'
454: is translated to `X11:[000000]foobar.h' or `X11:foobar.h',
455: whichever one can be opened. This strategy allows you to assign a
456: logical name to point to the actual location of the header files.
457:
458: 5. If none of these strategies succeeds, the `#include' fails.
459:
460: Include directives of the form:
461:
462: #include foobar
463:
464: are a common source of incompatibility between VAX-C and GNU CC. VAX-C
465: treats this much like a standard `#include <foobar.h>' directive. That
466: is incompatible with the ANSI C behavior implemented by GNU CC: to
467: expand the name `foobar' as a macro. Macro expansion should eventually
468: yield one of the two standard formats for `#include':
469:
470: #include "FILE"
471: #include <FILE>
472:
473: If you have this problem, the best solution is to modify the source
474: to convert the `#include' directives to one of the two standard forms.
475: That will work with either compiler. If you want a quick and dirty fix,
476: define the file names as macros with the proper expansion, like this:
477:
478: #define stdio <stdio.h>
479:
480: This will work, as long as the name doesn't conflict with anything else
481: in the program.
482:
483: Another source of incompatibility is that VAX-C assumes that:
484:
485: #include "foobar"
486:
487: is actually asking for the file `foobar.h'. GNU CC does not make this
488: assumption, and instead takes what you ask for literally; it tries to
489: read the file `foobar'. The best way to avoid this problem is to
490: always specify the desired file extension in your include directives.
491:
492: GNU CC for VMS is distributed with a set of include files that is
493: sufficient to compile most general purpose programs. Even though the
494: GNU CC distribution does not contain header files to define constants
495: and structures for some VMS system-specific functions, there is no
496: reason why you cannot use GNU CC with any of these functions. You first
497: may have to generate or create header files, either by using the public
498: domain utility `UNSDL' (which can be found on a DECUS tape), or by
499: extracting the relevant modules from one of the system macro libraries,
500: and using an editor to construct a C header file.
501:
502: A `#include' file name cannot contain a DECNET node name. The
503: preprocessor reports an I/O error if you attempt to use a node name,
504: whether explicitly, or implicitly via a logical name.
505:
506:
507: File: gcc.info, Node: Global Declarations, Next: VMS Misc, Prev: Include Files and VMS, Up: VMS
508:
509: Global Declarations and VMS
510: ===========================
511:
512: GNU CC does not provide the `globalref', `globaldef' and
513: `globalvalue' keywords of VAX-C. You can get the same effect with an
514: obscure feature of GAS, the GNU assembler. (This requires GAS version
515: 1.39 or later.) The following macros allow you to use this feature in
516: a fairly natural way:
517:
518: #ifdef __GNUC__
519: #define GLOBALREF(TYPE,NAME) \
520: TYPE NAME \
521: asm ("_$$PsectAttributes_GLOBALSYMBOL$$" #NAME)
522: #define GLOBALDEF(TYPE,NAME,VALUE) \
523: TYPE NAME \
524: asm ("_$$PsectAttributes_GLOBALSYMBOL$$" #NAME) \
525: = VALUE
526: #define GLOBALVALUEREF(TYPE,NAME) \
527: const TYPE NAME[1] \
528: asm ("_$$PsectAttributes_GLOBALVALUE$$" #NAME)
529: #define GLOBALVALUEDEF(TYPE,NAME,VALUE) \
530: const TYPE NAME[1] \
531: asm ("_$$PsectAttributes_GLOBALVALUE$$" #NAME) \
532: = {VALUE}
533: #else
534: #define GLOBALREF(TYPE,NAME) \
535: globalref TYPE NAME
536: #define GLOBALDEF(TYPE,NAME,VALUE) \
537: globaldef TYPE NAME = VALUE
538: #define GLOBALVALUEDEF(TYPE,NAME,VALUE) \
539: globalvalue TYPE NAME = VALUE
540: #define GLOBALVALUEREF(TYPE,NAME) \
541: globalvalue TYPE NAME
542: #endif
543:
544: (The `_$$PsectAttributes_GLOBALSYMBOL' prefix at the start of the name
545: is removed by the assembler, after it has modified the attributes of
546: the symbol). These macros are provided in the VMS binaries
547: distribution in a header file `GNU_HACKS.H'. An example of the usage
548: is:
549:
550: GLOBALREF (int, ijk);
551: GLOBALDEF (int, jkl, 0);
552:
553: The macros `GLOBALREF' and `GLOBALDEF' cannot be used
554: straightforwardly for arrays, since there is no way to insert the array
555: dimension into the declaration at the right place. However, you can
556: declare an array with these macros if you first define a typedef for the
557: array type, like this:
558:
559: typedef int intvector[10];
560: GLOBALREF (intvector, foo);
561:
562: Array and structure initializers will also break the macros; you can
563: define the initializer to be a macro of its own, or you can expand the
564: `GLOBALDEF' macro by hand. You may find a case where you wish to use
565: the `GLOBALDEF' macro with a large array, but you are not interested in
566: explicitly initializing each element of the array. In such cases you
567: can use an initializer like: `{0,}', which will initialize the entire
568: array to `0'.
569:
570: A shortcoming of this implementation is that a variable declared with
571: `GLOBALVALUEREF' or `GLOBALVALUEDEF' is always an array. For example,
572: the declaration:
573:
574: GLOBALVALUEREF(int, ijk);
575:
576: declares the variable `ijk' as an array of type `int [1]'. This is
577: done because a globalvalue is actually a constant; its "value" is what
578: the linker would normally consider an address. That is not how an
579: integer value works in C, but it is how an array works. So treating
580: the symbol as an array name gives consistent results--with the
581: exception that the value seems to have the wrong type. *Don't try to
582: access an element of the array.* It doesn't have any elements. The
583: array "address" may not be the address of actual storage.
584:
585: The fact that the symbol is an array may lead to warnings where the
586: variable is used. Insert type casts to avoid the warnings. Here is an
587: example; it takes advantage of the ANSI C feature allowing macros that
588: expand to use the same name as the macro itself.
589:
590: GLOBALVALUEREF (int, ss$_normal);
591: GLOBALVALUEDEF (int, xyzzy,123);
592: #ifdef __GNUC__
593: #define ss$_normal ((int) ss$_normal)
594: #define xyzzy ((int) xyzzy)
595: #endif
596:
597: Don't use `globaldef' or `globalref' with a variable whose type is
598: an enumeration type; this is not implemented. Instead, make the
599: variable an integer, and use a `globalvaluedef' for each of the
600: enumeration values. An example of this would be:
601:
602: #ifdef __GNUC__
603: GLOBALDEF (int, color, 0);
604: GLOBALVALUEDEF (int, RED, 0);
605: GLOBALVALUEDEF (int, BLUE, 1);
606: GLOBALVALUEDEF (int, GREEN, 3);
607: #else
608: enum globaldef color {RED, BLUE, GREEN = 3};
609: #endif
610:
611:
612: File: gcc.info, Node: VMS Misc, Prev: Global Declarations, Up: VMS
613:
614: Other VMS Issues
615: ================
616:
617: GNU CC automatically arranges for `main' to return 1 by default if
618: you fail to specify an explicit return value. This will be interpreted
619: by VMS as a status code indicating a normal successful completion.
620: Version 1 of GNU CC did not provide this default.
621:
622: GNU CC on VMS works only with the GNU assembler, GAS. You need
623: version 1.37 or later of GAS in order to produce value debugging
624: information for the VMS debugger. Use the ordinary VMS linker with the
625: object files produced by GAS.
626:
627: Under previous versions of GNU CC, the generated code would
628: occasionally give strange results when linked to the sharable `VAXCRTL'
629: library. Now this should work.
630:
631: A caveat for use of `const' global variables: the `const' modifier
632: must be specified in every external declaration of the variable in all
633: of the source files that use that variable. Otherwise the linker will
634: issue warnings about conflicting attributes for the variable. Your
635: program will still work despite the warnings, but the variable will be
636: placed in writable storage.
637:
638: Although the VMS linker does distinguish between upper and lower case
639: letters in global symbols, most VMS compilers convert all such symbols
640: into upper case and most run-time library routines also have upper case
641: names. To be able to reliably call such routines, GNU CC (by means of
642: the assembler GAS) converts global symbols into upper case like other
643: VMS compilers. However, since the usual practice in C is to distinguish
644: case, GNU CC (via GAS) tries to preserve usual C behavior by augmenting
645: each name that is not all lower case. This means truncating the name
646: to at most 23 characters and then adding more characters at the end
647: which encode the case pattern of those 23. Names which contain at
648: least one dollar sign are an exception; they are converted directly into
649: upper case without augmentation.
650:
651: Name augmentation yields bad results for programs that use
652: precompiled libraries (such as Xlib) which were generated by another
653: compiler. You can use the compiler option `/NOCASE_HACK' to inhibit
654: augmentation; it makes external C functions and variables
655: case-independent as is usual on VMS. Alternatively, you could write
656: all references to the functions and variables in such libraries using
657: lower case; this will work on VMS, but is not portable to other
658: systems. The compiler option `/NAMES' also provides control over
659: global name handling.
660:
661: Function and variable names are handled somewhat differently with GNU
662: C++. The GNU C++ compiler performs "name mangling" on function names,
663: which means that it adds information to the function name to describe
664: the data types of the arguments that the function takes. One result of
665: this is that the name of a function can become very long. Since the
666: VMS linker only recognizes the first 31 characters in a name, special
667: action is taken to ensure that each function and variable has a unique
668: name that can be represented in 31 characters.
669:
670: If the name (plus a name augmentation, if required) is less than 32
671: characters in length, then no special action is performed. If the name
672: is longer than 31 characters, the assembler (GAS) will generate a hash
673: string based upon the function name, truncate the function name to 23
674: characters, and append the hash string to the truncated name. If the
675: `/VERBOSE' compiler option is used, the assembler will print both the
676: full and truncated names of each symbol that is truncated.
677:
678: The `/NOCASE_HACK' compiler option should not be used when you are
679: compiling programs that use libg++. libg++ has several instances of
680: objects (i.e. `Filebuf' and `filebuf') which become indistinguishable
681: in a case-insensitive environment. This leads to cases where you need
682: to inhibit augmentation selectively (if you were using libg++ and Xlib
683: in the same program, for example). There is no special feature for
684: doing this, but you can get the result by defining a macro for each
685: mixed case symbol for which you wish to inhibit augmentation. The
686: macro should expand into the lower case equivalent of itself. For
687: example:
688:
689: #define StuDlyCapS studlycaps
690:
691: These macro definitions can be placed in a header file to minimize
692: the number of changes to your source code.
693:
694:
695: File: gcc.info, Node: Portability, Next: Interface, Prev: VMS, Up: Top
696:
697: GNU CC and Portability
698: **********************
699:
700: The main goal of GNU CC was to make a good, fast compiler for
701: machines in the class that the GNU system aims to run on: 32-bit
702: machines that address 8-bit bytes and have several general registers.
703: Elegance, theoretical power and simplicity are only secondary.
704:
705: GNU CC gets most of the information about the target machine from a
706: machine description which gives an algebraic formula for each of the
707: machine's instructions. This is a very clean way to describe the
708: target. But when the compiler needs information that is difficult to
709: express in this fashion, I have not hesitated to define an ad-hoc
710: parameter to the machine description. The purpose of portability is to
711: reduce the total work needed on the compiler; it was not of interest
712: for its own sake.
713:
714: GNU CC does not contain machine dependent code, but it does contain
715: code that depends on machine parameters such as endianness (whether the
716: most significant byte has the highest or lowest address of the bytes in
717: a word) and the availability of autoincrement addressing. In the
718: RTL-generation pass, it is often necessary to have multiple strategies
719: for generating code for a particular kind of syntax tree, strategies
720: that are usable for different combinations of parameters. Often I have
721: not tried to address all possible cases, but only the common ones or
722: only the ones that I have encountered. As a result, a new target may
723: require additional strategies. You will know if this happens because
724: the compiler will call `abort'. Fortunately, the new strategies can be
725: added in a machine-independent fashion, and will affect only the target
726: machines that need them.
727:
728:
729: File: gcc.info, Node: Interface, Next: Passes, Prev: Portability, Up: Top
730:
731: Interfacing to GNU CC Output
732: ****************************
733:
734: GNU CC is normally configured to use the same function calling
735: convention normally in use on the target system. This is done with the
736: machine-description macros described (*note Target Macros::.).
737:
738: However, returning of structure and union values is done differently
739: on some target machines. As a result, functions compiled with PCC
740: returning such types cannot be called from code compiled with GNU CC,
741: and vice versa. This does not cause trouble often because few Unix
742: library routines return structures or unions.
743:
744: GNU CC code returns structures and unions that are 1, 2, 4 or 8 bytes
745: long in the same registers used for `int' or `double' return values.
746: (GNU CC typically allocates variables of such types in registers also.)
747: Structures and unions of other sizes are returned by storing them into
748: an address passed by the caller (usually in a register). The
749: machine-description macros `STRUCT_VALUE' and `STRUCT_INCOMING_VALUE'
750: tell GNU CC where to pass this address.
751:
752: By contrast, PCC on most target machines returns structures and
753: unions of any size by copying the data into an area of static storage,
754: and then returning the address of that storage as if it were a pointer
755: value. The caller must copy the data from that memory area to the
756: place where the value is wanted. This is slower than the method used
757: by GNU CC, and fails to be reentrant.
758:
759: On some target machines, such as RISC machines and the 80386, the
760: standard system convention is to pass to the subroutine the address of
761: where to return the value. On these machines, GNU CC has been
762: configured to be compatible with the standard compiler, when this method
763: is used. It may not be compatible for structures of 1, 2, 4 or 8 bytes.
764:
765: GNU CC uses the system's standard convention for passing arguments.
766: On some machines, the first few arguments are passed in registers; in
767: others, all are passed on the stack. It would be possible to use
768: registers for argument passing on any machine, and this would probably
769: result in a significant speedup. But the result would be complete
770: incompatibility with code that follows the standard convention. So this
771: change is practical only if you are switching to GNU CC as the sole C
772: compiler for the system. We may implement register argument passing on
773: certain machines once we have a complete GNU system so that we can
774: compile the libraries with GNU CC.
775:
776: On some machines (particularly the Sparc), certain types of arguments
777: are passed "by invisible reference". This means that the value is
778: stored in memory, and the address of the memory location is passed to
779: the subroutine.
780:
781: If you use `longjmp', beware of automatic variables. ANSI C says
782: that automatic variables that are not declared `volatile' have undefined
783: values after a `longjmp'. And this is all GNU CC promises to do,
784: because it is very difficult to restore register variables correctly,
785: and one of GNU CC's features is that it can put variables in registers
786: without your asking it to.
787:
788: If you want a variable to be unaltered by `longjmp', and you don't
789: want to write `volatile' because old C compilers don't accept it, just
790: take the address of the variable. If a variable's address is ever
791: taken, even if just to compute it and ignore it, then the variable
792: cannot go in a register:
793:
794: {
795: int careful;
796: &careful;
797: ...
798: }
799:
800: Code compiled with GNU CC may call certain library routines. Most of
801: them handle arithmetic for which there are no instructions. This
802: includes multiply and divide on some machines, and floating point
803: operations on any machine for which floating point support is disabled
804: with `-msoft-float'. Some standard parts of the C library, such as
805: `bcopy' or `memcpy', are also called automatically. The usual function
806: call interface is used for calling the library routines.
807:
808: These library routines should be defined in the library `libgcc.a',
809: which GNU CC automatically searches whenever it links a program. On
810: machines that have multiply and divide instructions, if hardware
811: floating point is in use, normally `libgcc.a' is not needed, but it is
812: searched just in case.
813:
814: Each arithmetic function is defined in `libgcc1.c' to use the
815: corresponding C arithmetic operator. As long as the file is compiled
816: with another C compiler, which supports all the C arithmetic operators,
817: this file will work portably. However, `libgcc1.c' does not work if
818: compiled with GNU CC, because each arithmetic function would compile
819: into a call to itself!
820:
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.