This is Info file f/g77.info, produced by Makeinfo version 1.68 from the input file ../../../src/gcc-2.95.3/gcc/f/g77.texi. INFO-DIR-SECTION Programming START-INFO-DIR-ENTRY * g77: (g77). The GNU Fortran compiler. END-INFO-DIR-ENTRY This file documents the use and the internals of the GNU Fortran (`g77') compiler. It corresponds to the GCC-2.95 version of `g77'. Published by the Free Software Foundation 59 Temple Place - Suite 330 Boston, MA 02111-1307 USA Copyright (C) 1995-1999 Free Software Foundation, Inc. Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies. Permission is granted to copy and distribute modified versions of this manual under the conditions for verbatim copying, provided also that the sections entitled "GNU General Public License," "Funding for Free Software," and "Protect Your Freedom--Fight `Look And Feel'" are included exactly as in the original, and provided that the entire resulting derived work is distributed under the terms of a permission notice identical to this one. Permission is granted to copy and distribute translations of this manual into another language, under the above conditions for modified versions, except that the sections entitled "GNU General Public License," "Funding for Free Software," and "Protect Your Freedom--Fight `Look And Feel'", and this permission notice, may be included in translations approved by the Free Software Foundation instead of in the original English. Contributed by James Craig Burley (). Inspired by a first pass at translating `g77-0.5.16/f/DOC' that was contributed to Craig by David Ronis ().  File: g77.info, Node: Surprising Interpretations of Code, Next: Aliasing Assumed To Work, Prev: Unused Arguments, Up: Working Programs Surprising Interpretations of Code ---------------------------------- The `-Wsurprising' option can help find bugs involving expression evaluation or in the way `DO' loops with non-integral iteration variables are handled. Cases found by this option might indicate a difference of interpretation between the author of the code involved, and a standard-conforming compiler such as `g77'. Such a difference might produce actual bugs. In any case, changing the code to explicitly do what the programmer might have expected it to do, so `g77' and other compilers are more likely to follow the programmer's expectations, might be worthwhile, especially if such changes make the program work better.  File: g77.info, Node: Aliasing Assumed To Work, Next: Output Assumed To Flush, Prev: Surprising Interpretations of Code, Up: Working Programs Aliasing Assumed To Work ------------------------ The `-falias-check', `-fargument-alias', `-fargument-noalias', and `-fno-argument-noalias-global' options, introduced in version 0.5.20 and `g77''s version 2.7.2.2.f.2 of `gcc', were withdrawn as of `g77' version 0.5.23 due to their not being supported by `gcc' version 2.8. These options, which control the assumptions regarding aliasing (overlapping) of writes and reads to main memory (core) made by the `gcc' back end, might well be added back (in some form) in a future version of `gcc'. However, these options *are* supported by `egcs'. The information below still is useful, but applies to only those versions of `g77' that support the alias analysis implied by support for these options. These options are effective only when compiling with `-O' (specifying any level other than `-O0') or with `-falias-check'. The default for Fortran code is `-fargument-noalias-global'. (The default for C code and code written in other C-based languages is `-fargument-alias'. These defaults apply regardless of whether you use `g77' or `gcc' to compile your code.) Note that, on some systems, compiling with `-fforce-addr' in effect can produce more optimal code when the default aliasing options are in effect (and when optimization is enabled). If your program is not working when compiled with optimization, it is possible it is violating the Fortran standards (77 and 90) by relying on the ability to "safely" modify variables and arrays that are aliased, via procedure calls, to other variables and arrays, without using `EQUIVALENCE' to explicitly set up this kind of aliasing. (The FORTRAN 77 standard's prohibition of this sort of overlap, generally referred to therein as "storage assocation", appears in Sections 15.9.3.6. This prohibition allows implementations, such as `g77', to, for example, implement the passing of procedures and even values in `COMMON' via copy operations into local, perhaps more efficiently accessed temporaries at entry to a procedure, and, where appropriate, via copy operations back out to their original locations in memory at exit from that procedure, without having to take into consideration the order in which the local copies are updated by the code, among other things.) To test this hypothesis, try compiling your program with the `-fargument-alias' option, which causes the compiler to revert to assumptions essentially the same as made by versions of `g77' prior to 0.5.20. If the program works using this option, that strongly suggests that the bug is in your program. Finding and fixing the bug(s) should result in a program that is more standard-conforming and that can be compiled by `g77' in a way that results in a faster executable. (You might want to try compiling with `-fargument-noalias', a kind of half-way point, to see if the problem is limited to aliasing between dummy arguments and `COMMON' variables--this option assumes that such aliasing is not done, while still allowing aliasing among dummy arguments.) An example of aliasing that is invalid according to the standards is shown in the following program, which might *not* produce the expected results when executed: I = 1 CALL FOO(I, I) PRINT *, I END SUBROUTINE FOO(J, K) J = J + K K = J * K PRINT *, J, K END The above program attempts to use the temporary aliasing of the `J' and `K' arguments in `FOO' to effect a pathological behavior--the simultaneous changing of the values of *both* `J' and `K' when either one of them is written. The programmer likely expects the program to print these values: 2 4 4 However, since the program is not standard-conforming, an implementation's behavior when running it is undefined, because subroutine `FOO' modifies at least one of the arguments, and they are aliased with each other. (Even if one of the assignment statements was deleted, the program would still violate these rules. This kind of on-the-fly aliasing is permitted by the standard only when none of the aliased items are defined, or written, while the aliasing is in effect.) As a practical example, an optimizing compiler might schedule the `J =' part of the second line of `FOO' *after* the reading of `J' and `K' for the `J * K' expression, resulting in the following output: 2 2 2 Essentially, compilers are promised (by the standard and, therefore, by programmers who write code they claim to be standard-conforming) that if they cannot detect aliasing via static analysis of a single program unit's `EQUIVALENCE' and `COMMON' statements, no such aliasing exists. In such cases, compilers are free to assume that an assignment to one variable will not change the value of another variable, allowing it to avoid generating code to re-read the value of the other variable, to re-schedule reads and writes, and so on, to produce a faster executable. The same promise holds true for arrays (as seen by the called procedure)--an element of one dummy array cannot be aliased with, or overlap, any element of another dummy array or be in a `COMMON' area known to the procedure. (These restrictions apply only when the procedure defines, or writes to, one of the aliased variables or arrays.) Unfortunately, there is no way to find *all* possible cases of violations of the prohibitions against aliasing in Fortran code. Static analysis is certainly imperfect, as is run-time analysis, since neither can catch all violations. (Static analysis can catch all likely violations, and some that might never actually happen, while run-time analysis can catch only those violations that actually happen during a particular run. Neither approach can cope with programs mixing Fortran code with routines written in other languages, however.) Currently, `g77' provides neither static nor run-time facilities to detect any cases of this problem, although other products might. Run-time facilities are more likely to be offered by future versions of `g77', though patches improving `g77' so that it provides either form of detection are welcome.  File: g77.info, Node: Output Assumed To Flush, Next: Large File Unit Numbers, Prev: Aliasing Assumed To Work, Up: Working Programs Output Assumed To Flush ----------------------- For several versions prior to 0.5.20, `g77' configured its version of the `libf2c' run-time library so that one of its configuration macros, `ALWAYS_FLUSH', was defined. This was done as a result of a belief that many programs expected output to be flushed to the operating system (under UNIX, via the `fflush()' library call) with the result that errors, such as disk full, would be immediately flagged via the relevant `ERR=' and `IOSTAT=' mechanism. Because of the adverse effects this approach had on the performance of many programs, `g77' no longer configures `libf2c' (now named `libg2c' in its `g77' incarnation) to always flush output. If your program depends on this behavior, either insert the appropriate `CALL FLUSH' statements, or modify the sources to the `libg2c', rebuild and reinstall `g77', and relink your programs with the modified library. (Ideally, `libg2c' would offer the choice at run-time, so that a compile-time option to `g77' or `f2c' could result in generating the appropriate calls to flushing or non-flushing library routines.) *Note Always Flush Output::, for information on how to modify the `g77' source tree so that a version of `libg2c' can be built and installed with the `ALWAYS_FLUSH' macro defined.  File: g77.info, Node: Large File Unit Numbers, Next: Floating-point precision, Prev: Output Assumed To Flush, Up: Working Programs Large File Unit Numbers ----------------------- If your program crashes at run time with a message including the text `illegal unit number', that probably is a message from the run-time library, `libg2c'. The message means that your program has attempted to use a file unit number that is out of the range accepted by `libg2c'. Normally, this range is 0 through 99, and the high end of the range is controlled by a `libg2c' source-file macro named `MXUNIT'. If you can easily change your program to use unit numbers in the range 0 through 99, you should do so. Otherwise, see *Note Larger File Unit Numbers::, for information on how to change `MXUNIT' in `libg2c' so you can build and install a new version of `libg2c' that supports the larger unit numbers you need. *Note:* While `libg2c' places a limit on the range of Fortran file-unit numbers, the underlying library and operating system might impose different kinds of limits. For example, some systems limit the number of files simultaneously open by a running program. Information on how to increase these limits should be found in your system's documentation.  File: g77.info, Node: Floating-point precision, Next: Inconsistent Calling Sequences, Prev: Large File Unit Numbers, Up: Working Programs Floating-point precision ------------------------ If your program depends on exact IEEE 754 floating-point handling it may help on some systems--specifically x86 or m68k hardware--to use the `-ffloat-store' option or to reset the precision flag on the floating-point unit. *Note Optimize Options::. However, it might be better simply to put the FPU into double precision mode and not take the performance hit of `-ffloat-store'. On x86 and m68k GNU systems you can do this with a technique similar to that for turning on floating-point exceptions (*note Floating-point Exception Handling::.). The control word could be set to double precision by replacing the `__setfpucw' call with one like this: __setfpucw ((_FPU_DEFAULT & ~_FPU_EXTENDED) | _FPU_DOUBLE); (It is not clear whether this has any effect on the operation of the GNU maths library, but we have no evidence of it causing trouble.) Some targets (such as the Alpha) may need special options for full IEEE conformance. *Note Hardware Models and Configurations: (gcc)Submodel Options.  File: g77.info, Node: Inconsistent Calling Sequences, Prev: Floating-point precision, Up: Working Programs Inconsistent Calling Sequences ------------------------------ Code containing inconsistent calling sequences in the same file is normally rejected--see *Note GLOBALS::. (Use, say, `ftnchek' to ensure consistency across source files. *Note Generating Skeletons and Prototypes with `f2c': f2c Skeletons and Prototypes.) Mysterious errors, which may appear to be code generation problems, can appear specifically on the x86 architecture with some such inconsistencies. On x86 hardware, floating-point return values of functions are placed on the floating-point unit's register stack, not the normal stack. Thus calling a `REAL' or `DOUBLE PRECISION' `FUNCTION' as some other sort of procedure, or vice versa, scrambles the floating-point stack. This may break unrelated code executed later. Similarly if, say, external C routines are written incorrectly.  File: g77.info, Node: Overly Convenient Options, Next: Faster Programs, Prev: Working Programs, Up: Collected Fortran Wisdom Overly Convenient Command-line Options ====================================== These options should be used only as a quick-and-dirty way to determine how well your program will run under different compilation models without having to change the source. Some are more problematic than others, depending on how portable and maintainable you want the program to be (and, of course, whether you are allowed to change it at all is crucial). You should not continue to use these command-line options to compile a given program, but rather should make changes to the source code: `-finit-local-zero' (This option specifies that any uninitialized local variables and arrays have default initialization to binary zeros.) Many other compilers do this automatically, which means lots of Fortran code developed with those compilers depends on it. It is safer (and probably would produce a faster program) to find the variables and arrays that need such initialization and provide it explicitly via `DATA', so that `-finit-local-zero' is not needed. Consider using `-Wuninitialized' (which requires `-O') to find likely candidates, but do not specify `-finit-local-zero' or `-fno-automatic', or this technique won't work. `-fno-automatic' (This option specifies that all local variables and arrays are to be treated as if they were named in `SAVE' statements.) Many other compilers do this automatically, which means lots of Fortran code developed with those compilers depends on it. The effect of this is that all non-automatic variables and arrays are made static, that is, not placed on the stack or in heap storage. This might cause a buggy program to appear to work better. If so, rather than relying on this command-line option (and hoping all compilers provide the equivalent one), add `SAVE' statements to some or all program unit sources, as appropriate. Consider using `-Wuninitialized' (which requires `-O') to find likely candidates, but do not specify `-finit-local-zero' or `-fno-automatic', or this technique won't work. The default is `-fautomatic', which tells `g77' to try and put variables and arrays on the stack (or in fast registers) where possible and reasonable. This tends to make programs faster. *Note:* Automatic variables and arrays are not affected by this option. These are variables and arrays that are *necessarily* automatic, either due to explicit statements, or due to the way they are declared. Examples include local variables and arrays not given the `SAVE' attribute in procedures declared `RECURSIVE', and local arrays declared with non-constant bounds (automatic arrays). Currently, `g77' supports only automatic arrays, not `RECURSIVE' procedures or other means of explicitly specifying that variables or arrays are automatic. `-fGROUP-intrinsics-hide' Change the source code to use `EXTERNAL' for any external procedure that might be the name of an intrinsic. It is easy to find these using `-fGROUP-intrinsics-disable'.  File: g77.info, Node: Faster Programs, Prev: Overly Convenient Options, Up: Collected Fortran Wisdom Faster Programs =============== Aside from the usual `gcc' options, such as `-O', `-ffast-math', and so on, consider trying some of the following approaches to speed up your program (once you get it working). * Menu: * Aligned Data:: * Prefer Automatic Uninitialized Variables:: * Avoid f2c Compatibility:: * Use Submodel Options::  File: g77.info, Node: Aligned Data, Next: Prefer Automatic Uninitialized Variables, Up: Faster Programs Aligned Data ------------ On some systems, such as those with Pentium Pro CPUs, programs that make heavy use of `REAL(KIND=2)' (`DOUBLE PRECISION') might run much slower than possible due to the compiler not aligning these 64-bit values to 64-bit boundaries in memory. (The effect also is present, though to a lesser extent, on the 586 (Pentium) architecture.) The Intel x86 architecture generally ensures that these programs will work on all its implementations, but particular implementations (such as Pentium Pro) perform better with more strict alignment. (Such behavior isn't unique to the Intel x86 architecture.) Other architectures might *demand* 64-bit alignment of 64-bit data. There are a variety of approaches to use to address this problem: * Order your `COMMON' and `EQUIVALENCE' areas such that the variables and arrays with the widest alignment guidelines come first. For example, on most systems, this would mean placing `COMPLEX(KIND=2)', `REAL(KIND=2)', and `INTEGER(KIND=2)' entities first, followed by `REAL(KIND=1)', `INTEGER(KIND=1)', and `LOGICAL(KIND=1)' entities, then `INTEGER(KIND=6)' entities, and finally `CHARACTER' and `INTEGER(KIND=3)' entities. The reason to use such placement is it makes it more likely that your data will be aligned properly, without requiring you to do detailed analysis of each aggregate (`COMMON' and `EQUIVALENCE') area. Specifically, on systems where the above guidelines are appropriate, placing `CHARACTER' entities before `REAL(KIND=2)' entities can work just as well, but only if the number of bytes occupied by the `CHARACTER' entities is divisible by the recommended alignment for `REAL(KIND=2)'. By ordering the placement of entities in aggregate areas according to the simple guidelines above, you avoid having to carefully count the number of bytes occupied by each entity to determine whether the actual alignment of each subsequent entity meets the alignment guidelines for the type of that entity. If you don't ensure correct alignment of `COMMON' elements, the compiler may be forced by some systems to violate the Fortran semantics by adding padding to get `DOUBLE PRECISION' data properly aligned. If the unfortunate practice is employed of overlaying different types of data in the `COMMON' block, the different variants of this block may become misaligned with respect to each other. Even if your platform doesn't require strict alignment, `COMMON' should be laid out as above for portability. (Unfortunately the FORTRAN 77 standard didn't anticipate this possible requirement, which is compiler-independent on a given platform.) * Use the (x86-specific) `-malign-double' option when compiling programs for the Pentium and Pentium Pro architectures (called 586 and 686 in the `gcc' configuration subsystem). The warning about this in the `gcc' manual isn't generally relevant to Fortran, but using it will force `COMMON' to be padded if necessary to align `DOUBLE PRECISION' data. When `DOUBLE PRECISION' data is forcibly aligned in `COMMON' by `g77' due to specifying `-malign-double', `g77' issues a warning about the need to insert padding. In this case, each and every program unit that uses the same `COMMON' area must specify the same layout of variables and their types for that area and be compiled with `-malign-double' as well. `g77' will issue warnings in each case, but as long as every program unit using that area is compiled with the same warnings, the resulting object files should work when linked together unless the program makes additional assumptions about `COMMON' area layouts that are outside the scope of the FORTRAN 77 standard, or uses `EQUIVALENCE' or different layouts in ways that assume no padding is ever inserted by the compiler. * Ensure that `crt0.o' or `crt1.o' on your system guarantees a 64-bit aligned stack for `main()'. The recent one from GNU (`glibc2') will do this on x86 systems, but we don't know of any other x86 setups where it will be right. Read your system's documentation to determine if it is appropriate to upgrade to a more recent version to obtain the optimal alignment. Progress is being made on making this work "out of the box" on future versions of `g77', `gcc', and some of the relevant operating systems (such as GNU/Linux). A package that tests the degree to which a Fortran compiler (such as `g77') aligns 64-bit floating-point variables and arrays is available at `ftp://alpha.gnu.org/gnu/g77/align/'.  File: g77.info, Node: Prefer Automatic Uninitialized Variables, Next: Avoid f2c Compatibility, Prev: Aligned Data, Up: Faster Programs Prefer Automatic Uninitialized Variables ---------------------------------------- If you're using `-fno-automatic' already, you probably should change your code to allow compilation with `-fautomatic' (the default), to allow the program to run faster. Similarly, you should be able to use `-fno-init-local-zero' (the default) instead of `-finit-local-zero'. This is because it is rare that every variable affected by these options in a given program actually needs to be so affected. For example, `-fno-automatic', which effectively `SAVE's every local non-automatic variable and array, affects even things like `DO' iteration variables, which rarely need to be `SAVE'd, and this often reduces run-time performances. Similarly, `-fno-init-local-zero' forces such variables to be initialized to zero--when `SAVE'd (such as when `-fno-automatic'), this by itself generally affects only startup time for a program, but when not `SAVE'd, it can slow down the procedure every time it is called. *Note Overly Convenient Command-Line Options: Overly Convenient Options, for information on the `-fno-automatic' and `-finit-local-zero' options and how to convert their use into selective changes in your own code.  File: g77.info, Node: Avoid f2c Compatibility, Next: Use Submodel Options, Prev: Prefer Automatic Uninitialized Variables, Up: Faster Programs Avoid f2c Compatibility ----------------------- If you aren't linking with any code compiled using `f2c', try using the `-fno-f2c' option when compiling *all* the code in your program. (Note that `libf2c' is *not* an example of code that is compiled using `f2c'--it is compiled by a C compiler, typically `gcc'.)  File: g77.info, Node: Use Submodel Options, Prev: Avoid f2c Compatibility, Up: Faster Programs Use Submodel Options -------------------- Using an appropriate `-m' option to generate specific code for your CPU may be worthwhile, though it may mean the executable won't run on other versions of the CPU that don't support the same instruction set. *Note Hardware Models and Configurations: (gcc)Submodel Options. For instance on an x86 system the compiler might have been built--as shown by `g77 -v'--for the target `i386-pc-linux-gnu', i.e. an `i386' CPU. In that case to generate code best optimized for a Pentium you could use the option `-march=pentium'. For recent CPUs that don't have explicit support in the released version of `gcc', it *might* still be possible to get improvements with certain `-m' options. `-fomit-frame-pointer' can help performance on x86 systems and others. It will, however, inhibit debugging on the systems on which it is not turned on anyway by `-O'.  File: g77.info, Node: Trouble, Next: Open Questions, Prev: Collected Fortran Wisdom, Up: Top Known Causes of Trouble with GNU Fortran **************************************** This section describes known problems that affect users of GNU Fortran. Most of these are not GNU Fortran bugs per se--if they were, we would fix them. But the result for a user might be like the result of a bug. Some of these problems are due to bugs in other software, some are missing features that are too much work to add, and some are places where people's opinions differ as to what is best. Information on bugs that show up when configuring, porting, building, or installing `g77' is not provided here. *Note Problems Installing::. To find out about major bugs discovered in the current release and possible workarounds for them, see `ftp://alpha.gnu.org/g77.plan'. (Note that some of this portion of the manual is lifted directly from the `gcc' manual, with minor modifications to tailor it to users of `g77'. Anytime a bug seems to have more to do with the `gcc' portion of `g77', see *Note Known Causes of Trouble with GNU CC: (gcc)Trouble.) * Menu: * But-bugs:: Bugs really in other programs or elsewhere. * Known Bugs:: Bugs known to be in this version of `g77'. * Missing Features:: Features we already know we want to add later. * Disappointments:: Regrettable things we can't change. * Non-bugs:: Things we think are right, but some others disagree. * Warnings and Errors:: Which problems in your code get warnings, and which get errors.  File: g77.info, Node: But-bugs, Next: Known Bugs, Up: Trouble Bugs Not In GNU Fortran ======================= These are bugs to which the maintainers often have to reply, "but that isn't a bug in `g77'...". Some of these already are fixed in new versions of other software; some still need to be fixed; some are problems with how `g77' is installed or is being used; some are the result of bad hardware that causes software to misbehave in sometimes bizarre ways; some just cannot be addressed at this time until more is known about the problem. Please don't re-report these bugs to the `g77' maintainers--if you must remind someone how important it is to you that the problem be fixed, talk to the people responsible for the other products identified below, but preferably only after you've tried the latest versions of those products. The `g77' maintainers have their hands full working on just fixing and improving `g77', without serving as a clearinghouse for all bugs that happen to affect `g77' users. *Note Collected Fortran Wisdom::, for information on behavior of Fortran programs, and the programs that compile them, that might be *thought* to indicate bugs. * Menu: * Signal 11 and Friends:: Strange behavior by any software. * Cannot Link Fortran Programs:: Unresolved references. * Large Common Blocks:: Problems on older GNU/Linux systems. * Debugger Problems:: When the debugger crashes. * NeXTStep Problems:: Misbehaving executables. * Stack Overflow:: More misbehaving executables. * Nothing Happens:: Less behaving executables. * Strange Behavior at Run Time:: Executables misbehaving due to bugs in your program. * Floating-point Errors:: The results look wrong, but....  File: g77.info, Node: Signal 11 and Friends, Next: Cannot Link Fortran Programs, Up: But-bugs Signal 11 and Friends --------------------- A whole variety of strange behaviors can occur when the software, or the way you are using the software, stresses the hardware in a way that triggers hardware bugs. This might seem hard to believe, but it happens frequently enough that there exist documents explaining in detail what the various causes of the problems are, what typical symptoms look like, and so on. Generally these problems are referred to in this document as "signal 11" crashes, because the Linux kernel, running on the most popular hardware (the Intel x86 line), often stresses the hardware more than other popular operating systems. When hardware problems do occur under GNU/Linux on x86 systems, these often manifest themselves as "signal 11" problems, as illustrated by the following diagnostic: sh# g77 myprog.f gcc: Internal compiler error: program f771 got fatal signal 11 sh# It is *very* important to remember that the above message is *not* the only one that indicates a hardware problem, nor does it always indicate a hardware problem. In particular, on systems other than those running the Linux kernel, the message might appear somewhat or very different, as it will if the error manifests itself while running a program other than the `g77' compiler. For example, it will appear somewhat different when running your program, when running Emacs, and so on. How to cope with such problems is well beyond the scope of this manual. However, users of Linux-based systems (such as GNU/Linux) should review `http://www.bitwizard.nl/sig11/', a source of detailed information on diagnosing hardware problems, by recognizing their common symptoms. Users of other operating systems and hardware might find this reference useful as well. If you know of similar material for another hardware/software combination, please let us know so we can consider including a reference to it in future versions of this manual.  File: g77.info, Node: Cannot Link Fortran Programs, Next: Large Common Blocks, Prev: Signal 11 and Friends, Up: But-bugs Cannot Link Fortran Programs ---------------------------- On some systems, perhaps just those with out-of-date (shared?) libraries, unresolved-reference errors happen when linking `g77'-compiled programs (which should be done using `g77'). If this happens to you, try appending `-lc' to the command you use to link the program, e.g. `g77 foo.f -lc'. `g77' already specifies `-lg2c -lm' when it calls the linker, but it cannot also specify `-lc' because not all systems have a file named `libc.a'. It is unclear at this point whether there are legitimately installed systems where `-lg2c -lm' is insufficient to resolve code produced by `g77'. If your program doesn't link due to unresolved references to names like `_main', make sure you're using the `g77' command to do the link, since this command ensures that the necessary libraries are loaded by specifying `-lg2c -lm' when it invokes the `gcc' command to do the actual link. (Use the `-v' option to discover more about what actually happens when you use the `g77' and `gcc' commands.) Also, try specifying `-lc' as the last item on the `g77' command line, in case that helps.  File: g77.info, Node: Large Common Blocks, Next: Debugger Problems, Prev: Cannot Link Fortran Programs, Up: But-bugs Large Common Blocks ------------------- On some older GNU/Linux systems, programs with common blocks larger than 16MB cannot be linked without some kind of error message being produced. This is a bug in older versions of `ld', fixed in more recent versions of `binutils', such as version 2.6.  File: g77.info, Node: Debugger Problems, Next: NeXTStep Problems, Prev: Large Common Blocks, Up: But-bugs Debugger Problems ----------------- There are some known problems when using `gdb' on code compiled by `g77'. Inadequate investigation as of the release of 0.5.16 results in not knowing which products are the culprit, but `gdb-4.14' definitely crashes when, for example, an attempt is made to print the contents of a `COMPLEX(KIND=2)' dummy array, on at least some GNU/Linux machines, plus some others. Attempts to access assumed-size arrays are also known to crash recent versions of `gdb'. (`gdb''s Fortran support was done for a different compiler and isn't properly compatible with `g77'.)  File: g77.info, Node: NeXTStep Problems, Next: Stack Overflow, Prev: Debugger Problems, Up: But-bugs NeXTStep Problems ----------------- Developers of Fortran code on NeXTStep (all architectures) have to watch out for the following problem when writing programs with large, statically allocated (i.e. non-stack based) data structures (common blocks, saved arrays). Due to the way the native loader (`/bin/ld') lays out data structures in virtual memory, it is very easy to create an executable wherein the `__DATA' segment overlaps (has addresses in common) with the `UNIX STACK' segment. This leads to all sorts of trouble, from the executable simply not executing, to bus errors. The NeXTStep command line tool `ebadexec' points to the problem as follows: % /bin/ebadexec a.out /bin/ebadexec: __LINKEDIT segment (truncated address = 0x3de000 rounded size = 0x2a000) of executable file: a.out overlaps with UNIX STACK segment (truncated address = 0x400000 rounded size = 0x3c00000) of executable file: a.out (In the above case, it is the `__LINKEDIT' segment that overlaps the stack segment.) This can be cured by assigning the `__DATA' segment (virtual) addresses beyond the stack segment. A conservative estimate for this is from address 6000000 (hexadecimal) onwards--this has always worked for me [Toon Moene]: % g77 -segaddr __DATA 6000000 test.f % ebadexec a.out ebadexec: file: a.out appears to be executable % Browsing through `egcs/gcc/f/Makefile.in', you will find that the `f771' program itself also has to be linked with these flags--it has large statically allocated data structures. (Version 0.5.18 reduces this somewhat, but probably not enough.) (The above item was contributed by Toon Moene ().)  File: g77.info, Node: Stack Overflow, Next: Nothing Happens, Prev: NeXTStep Problems, Up: But-bugs Stack Overflow -------------- `g77' code might fail at runtime (probably with a "segmentation violation") due to overflowing the stack. This happens most often on systems with an environment that provides substantially more heap space (for use when arbitrarily allocating and freeing memory) than stack space. Often this can be cured by increasing or removing your shell's limit on stack usage, typically using `limit stacksize' (in `csh' and derivatives) or `ulimit -s' (in `sh' and derivatives). Increasing the allowed stack size might, however, require changing some operating system or system configuration parameters. You might be able to work around the problem by compiling with the `-fno-automatic' option to reduce stack usage, probably at the expense of speed. *Note Maximum Stackable Size::, for information on patching `g77' to use different criteria for placing local non-automatic variables and arrays on the stack. However, if your program uses large automatic arrays (for example, has declarations like `REAL A(N)' where `A' is a local array and `N' is a dummy or `COMMON' variable that can have a large value), neither use of `-fno-automatic', nor changing the cut-off point for `g77' for using the stack, will solve the problem by changing the placement of these large arrays, as they are *necessarily* automatic. `g77' currently provides no means to specify that automatic arrays are to be allocated on the heap instead of the stack. So, other than increasing the stack size, your best bet is to change your source code to avoid large automatic arrays. Methods for doing this currently are outside the scope of this document. (*Note:* If your system puts stack and heap space in the same memory area, such that they are effectively combined, then a stack overflow probably indicates a program that is either simply too large for the system, or buggy.)  File: g77.info, Node: Nothing Happens, Next: Strange Behavior at Run Time, Prev: Stack Overflow, Up: But-bugs Nothing Happens --------------- It is occasionally reported that a "simple" program, such as a "Hello, World!" program, does nothing when it is run, even though the compiler reported no errors, despite the program containing nothing other than a simple `PRINT' statement. This most often happens because the program has been compiled and linked on a UNIX system and named `test', though other names can lead to similarly unexpected run-time behavior on various systems. Essentially this problem boils down to giving your program a name that is already known to the shell you are using to identify some other program, which the shell continues to execute instead of your program when you invoke it via, for example: sh# test sh# Under UNIX and many other system, a simple command name invokes a searching mechanism that might well not choose the program located in the current working directory if there is another alternative (such as the `test' command commonly installed on UNIX systems). The reliable way to invoke a program you just linked in the current directory under UNIX is to specify it using an explicit pathname, as in: sh# ./test Hello, World! sh# Users who encounter this problem should take the time to read up on how their shell searches for commands, how to set their search path, and so on. The relevant UNIX commands to learn about include `man', `info' (on GNU systems), `setenv' (or `set' and `env'), `which', and `find'.  File: g77.info, Node: Strange Behavior at Run Time, Next: Floating-point Errors, Prev: Nothing Happens, Up: But-bugs Strange Behavior at Run Time ---------------------------- `g77' code might fail at runtime with "segmentation violation", "bus error", or even something as subtle as a procedure call overwriting a variable or array element that it is not supposed to touch. These can be symptoms of a wide variety of actual bugs that occurred earlier during the program's run, but manifested themselves as *visible* problems some time later. Overflowing the bounds of an array--usually by writing beyond the end of it--is one of two kinds of bug that often occurs in Fortran code. (Compile your code with the `-fbounds-check' option to catch many of these kinds of errors at program run time.) The other kind of bug is a mismatch between the actual arguments passed to a procedure and the dummy arguments as declared by that procedure. Both of these kinds of bugs, and some others as well, can be difficult to track down, because the bug can change its behavior, or even appear to not occur, when using a debugger. That is, these bugs can be quite sensitive to data, including data representing the placement of other data in memory (that is, pointers, such as the placement of stack frames in memory). `g77' now offers the ability to catch and report some of these problems at compile, link, or run time, such as by generating code to detect references to beyond the bounds of most arrays (except assumed-size arrays), and checking for agreement between calling and called procedures. Future improvements are likely to be made in the procedure-mismatch area, at least. In the meantime, finding and fixing the programming bugs that lead to these behaviors is, ultimately, the user's responsibility, as difficult as that task can sometimes be. One runtime problem that has been observed might have a simple solution. If a formatted `WRITE' produces an endless stream of spaces, check that your program is linked against the correct version of the C library. The configuration process takes care to account for your system's normal `libc' not being ANSI-standard, which will otherwise cause this behaviour. If your system's default library is ANSI-standard and you subsequently link against a non-ANSI one, there might be problems such as this one. Specifically, on Solaris2 systems, avoid picking up the `BSD' library from `/usr/ucblib'.  File: g77.info, Node: Floating-point Errors, Prev: Strange Behavior at Run Time, Up: But-bugs Floating-point Errors --------------------- Some programs appear to produce inconsistent floating-point results compiled by `g77' versus by other compilers. Often the reason for this behavior is the fact that floating-point values are represented on almost all Fortran systems by *approximations*, and these approximations are inexact even for apparently simple values like 0.1, 0.2, 0.3, 0.4, 0.6, 0.7, 0.8, 0.9, 1.1, and so on. Most Fortran systems, including all current ports of `g77', use binary arithmetic to represent these approximations. Therefore, the exact value of any floating-point approximation as manipulated by `g77'-compiled code is representable by adding some combination of the values 1.0, 0.5, 0.25, 0.125, and so on (just keep dividing by two) through the precision of the fraction (typically around 23 bits for `REAL(KIND=1)', 52 for `REAL(KIND=2)'), then multiplying the sum by a integral power of two (in Fortran, by `2**N') that typically is between -127 and +128 for `REAL(KIND=1)' and -1023 and +1024 for `REAL(KIND=2)', then multiplying by -1 if the number is negative. So, a value like 0.2 is exactly represented in decimal--since it is a fraction, `2/10', with a denominator that is compatible with the base of the number system (base 10). However, `2/10' cannot be represented by any finite number of sums of any of 1.0, 0.5, 0.25, and so on, so 0.2 cannot be exactly represented in binary notation. (On the other hand, decimal notation can represent any binary number in a finite number of digits. Decimal notation cannot do so with ternary, or base-3, notation, which would represent floating-point numbers as sums of any of `1/1', `1/3', `1/9', and so on. After all, no finite number of decimal digits can exactly represent `1/3'. Fortunately, few systems use ternary notation.) Moreover, differences in the way run-time I/O libraries convert between these approximations and the decimal representation often used by programmers and the programs they write can result in apparent differences between results that do not actually exist, or exist to such a small degree that they usually are not worth worrying about. For example, consider the following program: PRINT *, 0.2 END When compiled by `g77', the above program might output `0.20000003', while another compiler might produce a executable that outputs `0.2'. This particular difference is due to the fact that, currently, conversion of floating-point values by the `libg2c' library, used by `g77', handles only double-precision values. Since `0.2' in the program is a single-precision value, it is converted to double precision (still in binary notation) before being converted back to decimal. The conversion to binary appends *binary* zero digits to the original value--which, again, is an inexact approximation of 0.2--resulting in an approximation that is much less exact than is connoted by the use of double precision. (The appending of binary zero digits has essentially the same effect as taking a particular decimal approximation of `1/3', such as `0.3333333', and appending decimal zeros to it, producing `0.33333330000000000'. Treating the resulting decimal approximation as if it really had 18 or so digits of valid precision would make it seem a very poor approximation of `1/3'.) As a result of converting the single-precision approximation to double precision by appending binary zeros, the conversion of the resulting double-precision value to decimal produces what looks like an incorrect result, when in fact the result is *inexact*, and is probably no less inaccurate or imprecise an approximation of 0.2 than is produced by other compilers that happen to output the converted value as "exactly" `0.2'. (Some compilers behave in a way that can make them appear to retain more accuracy across a conversion of a single-precision constant to double precision. *Note Context-Sensitive Constants::, to see why this practice is illusory and even dangerous.) Note that a more exact approximation of the constant is computed when the program is changed to specify a double-precision constant: PRINT *, 0.2D0 END Future versions of `g77' and/or `libg2c' might convert single-precision values directly to decimal, instead of converting them to double precision first. This would tend to result in output that is more consistent with that produced by some other Fortran implementations. A useful source of information on floating-point computation is David Goldberg, `What Every Computer Scientist Should Know About Floating-Point Arithmetic', Computing Surveys, 23, March 1991, pp. 5-48. An online version is available at `http://docs.sun.com/', and there is a supplemented version, in PostScript form, at `http://www.validgh.com/goldberg/paper.ps'. Information related to the IEEE 754 floating-point standard by a leading light can be found at `http://http.cs.berkeley.edu/%7Ewkahan/ieee754status/'; see also slides from the short course referenced from `http://http.cs.berkeley.edu/%7Efateman/'. `http://www.linuxsupportline.com/%7Ebillm/' has a brief guide to IEEE 754, a somewhat x86-GNU/Linux-specific FAQ, and library code for GNU/Linux x86 systems. The supplement to the PostScript-formatted Goldberg document, referenced above, is available in HTML format. See `Differences Among IEEE 754 Implementations' by Doug Priest, available online at `http://www.validgh.com/goldberg/addendum.html'. This document explores some of the issues surrounding computing of extended (80-bit) results on processors such as the x86, especially when those results are arbitrarily truncated to 32-bit or 64-bit values by the compiler as "spills". (*Note:* `g77' specifically, and `gcc' generally, does arbitrarily truncate 80-bit results during spills as of this writing. It is not yet clear whether a future version of the GNU compiler suite will offer 80-bit spills as an option, or perhaps even as the default behavior.) The GNU C library provides routines for controlling the FPU, and other documentation about this. *Note Floating-point precision::, regarding IEEE 754 conformance.