I'm trying to locate where __builtin_va_start is defined in GCC's source code, and see how it is implemented. (I was looking for where va_start is defined and then found that this macro is defined as __builtin_va_start.) I used cscope -r in GCC 9.1's source code directory to search the definition but haven't found it. Can anyone point where this function is defined?
That __builtin_va_start is not defined anywhere. It is a GCC compiler builtin (a bit like sizeof is a compile-time operator). It is an implementation detail related to the <stdarg.h> standard header (provided by the compiler, not the C standard library implementation libc). What really matters are the calling conventions and ABI followed by the generated assembler.
GCC has special code to deal with compiler builtins. And that code is not defining the builtin, but implementing its ad-hoc behavior inside the compiler. And __builtin_va_start is expanded into some compiler-specific internal representation of your compiled C/C++ code, specific to GCC (some GIMPLE perhaps)
From a comment of yours, I would infer that you are interested in implementation details. But that should be in your question
If you study GCC 9.1 source code, look inside some of gcc-9.1.0/gcc/builtins.c (the expand_builtin_va_start function there), and for other builtins inside gcc-9.1.0/gcc/c-family/c-cppbuiltin.c, gcc-9.1.0/gcc/cppbuiltin.c, gcc-9.1.0/gcc/jit/jit-builtins.c
You could write your own GCC plugin (in 2Q2019, for GCC 9, and the C++ code of your plugin might have to change for the future GCC 10) to add your own GCC builtins. BTW, you might even overload the behavior of the existing __builtin_va_start by your own specific code, and/or you might have -at least for research purposes- your own stdarg.h header with #define va_start(v,l) __my_builtin_va_start(v,l) and have your GCC plugin understand your __my_builtin_va_start plugin-specific builtin. Be however aware of the GCC runtime library exception and read its rationale: I am not a lawyer, but I tend to believe that you should (and that legal document requires you to) publish your GCC plugin with some open source license.
You first need to read a textbook on compilers, such as the Dragon book, to understand that an optimizing compiler is mostly transforming internal representations of your compiled code.
You further need to spend months in studying the many internal representations of GCC. Remember, GCC is a very complex program (of about ten millions lines of code). Don't expect to understand it with only a few days of work. Look inside the GCC resource center website.
My dead GCC MELT project had references and slides explaining more of GCC (the design philosophy and architecture of GCC changes slowly; so the concepts are still relevant, even if individual details changed). It took me almost ten years full time to partly understand some of the middle-end layers of GCC. I cannot transmit that knowledge in a StackOverflow answer.
My draft Bismon report (work in progress, funded by H2020, so lot of bureaucracy) has a dozen of pages (in its sections §1.3 and 1.4) introducing the internal representations of GCC.
Related
I would like to create a build of my embedded C code which specifically checks that floating point operations aren't introduced into it by accident. I've tried adding +nofp to my [cortex-m3] processor architecture but GCC for ARM doesn't like that (probably because the cortex-m3 doesn't have a floating point unit). I've tried specifying -mfpu=none but that isn't a permitted option. I've tried leaving -lm off the linker command-line but the linker seems too clever to be fooled by that and is compiling code with double in it and resolving pow() anyway.
This post: https://gcc.gnu.org/legacy-ml/gcc-help/2011-07/msg00093.html from 2011 hints that GCC has no such option, since no-one is interested in it, which surprises me as it seems like a common thing to want, at least from an embedded standpoint, to avoid accidental C-library bloat.
Does anyone know of a way to do this with GCC/newlib without me having to go through and manually hack stuff out of the C library file it chooses?
It is not just a library issue. Your target will use soft-fp, and the compiler will supply floating point code to implement arithmetic operators regardless of the library.
The solution I generally apply is to scan the map file for instances of the compiler supplied floating-point routines. If your code is "fp clean" there will be no such references. The math library and any other code that perform floating-point arithmetic operations will use these operator implementations, so you only need look for these operator calls and can ignore the Newlib math library functions.
The internal soft-fp routines are listed at https://gcc.gnu.org/onlinedocs/gccint/Soft-float-library-routines.html. It is probably feasible to manually check the mapfile for fp symbols but you might write yourself a script or tool to scan the map file for these names to check your. The cross-reference section of the map file will list all modules these symbols are used in so you can use that to identify where the floating point code is used.
The Newlib stdio functions support floating-point by default. If your formatted I/O is limited to printf() you can use iprintf() instead or you can rebuild Newlib with FLOATING_POINT undefined to remove floating point support from all but scanf() (no idea why). You can then use the map file technique again to find "banned" formatted I/O functions (although these are likely to also use the floating point operator functions in any case, so you will already have spotted them indirectly).
An alternative is to use an alternative stdio library to override the Newlib versions. There are any number of "tiny printf" implementations available you could use. If you link such a library as object code or list its library ahead of Newlib in the link command, it will override the Newlib versions.
If I have such a structure：
struct test{
float c,
f,
ops;
};
How can I modify the GCC compiler source code to make it as follows：
struct test{
double c,
f,
ops;
};
I now have such a requirement, I need to modify the gcc source code so that when he compiles a structure of a certain mode, he changes its type to a specified type.
Thank you！
your goal is very ambitious!
A possible approach could be to develop your own GCC plugin doing that job.
My recommendations:
budget several months of your time for that work (and perhaps several years) - at least 6 months full time to get a "proof of concept" thing which would fail on most code bases (in C). For C++, add another full year.
read carefully the C11 standard n1570 (if you target C), and the C++11 standard n3337 (if you target C++). That effort alone may take you a full month.
make your plugin open source software, and put its code quickly (e.g. under LGPL license) on some repository such as github or gitlab.
target the latest available version of GCC. In September 2020, that means GCC 10. Plugins and GCC APIs are changing incompatibly from one version to the next. If you need to stick to GCC 8 specifically, be prepared to spend a big amount of money for companies like AdaCore.
read carefully the documentation on GCC internals. You need to understand the GENERIC representation.
study very carefully the gcc/tree.def and gcc/treestruct.def and gcc/gimple.def files of GCC source code. You need to basically understand every line in them.
study very carefully the gcc/passes.def file of GCC source code. Again, you need to understand every line in that file.
learn to compile GCC from its source code. You certainly want to build it with g++ -Wall -Wextra -g -O1
read this draft report.
ask help in written English on the gcc#gcc.gnu.org mailing list, but have some working plugin before.
consider making a PhD out of this work. It is worth one. Alexandre Lissy in France got a PhD on a very similar topic.
If your code base is in C, consider using Frama-C (or Clang) and design your tool as a C to C transpiler.
Perhaps clever preprocessor tricks like #define float double followed by #undef float might be enough.
Closed. This question needs debugging details. It is not currently accepting answers.
Edit the question to include desired behavior, a specific problem or error, and the shortest code necessary to reproduce the problem. This will help others answer the question.
Closed 6 years ago.
Improve this question
I am working on a custom CFD Solver written in Fortran 90 and MPI.
The code contain 15+ Modules and was initially designed to work with the Intel Fortran compiler. Now since i do not have access to the Intel compiler I need to make it work using the GNU Fortran Compiler.
I made changes in the Makefile that initially had flags suitable for the ifort.
I am using it on Ubuntu with GNU Fortran and Openmpi
I am sorry I am unable to put in anything from the code structure or terminal output due to IP restrictions of my university. Nevertheless,I will try to best describe the issues
So now when I compile the code I am having some strange issues.
The GNU Fortran is not able to read lines that are too long and I get errors during compilation. As a result I have to break it into multiple lines using the '&' symbol
A module D.f90 contains all the Global variables declared. However, now I during compilation i get error is in module B.F90.
The error I get is 'Unclassified Statement Error', I was able to fix it in some subroutines and functions by locally declaring the variables again.
I am not the most experienced person in Fortran, but I thought that the change in compiler should not be a reason for new found syntax errors.
The errors described above so far could be remedied but considering the expanse of the code it is impractical.
I was hoping if anyone could share views on this matter and provide guidance on how to tackle it.
You should start reading three pieces of documentation:
The Fortran 90 standard (alternatively, other versions), which tells you what is legal, standard Fortran and what is not. Whenever you find some error, look at your code and check if what you are doing is legal, standard Fortran. Likely, the code in question will either be completely nonstandard (e.g. REAL*8, although that extension is fairly well understood) or rely on unspecified behaviour that Intel Fortran and GFortran are interpreting in different ways.
The GFortran manual for your version, which tells you how GFortran decides such unspecified cases, what intrinsic functions are available, how to change some options/flags, etc. This would tell you that your problem with the line lengths would be solved by adding -ffree-line-length-none.
The Intel Fortran manual for your version, which in cases of non-standard or unspecified behaviour, will allow you to know what the code you are reading was written to do, e.g. the behaviour that you would expect. In particular, it will allow you to decipher what the compiler flags that are currently being used mean. They may or may not need translation to GFortran, e.g. /Qsave will need to become -f-no-automatic.
A concrete example of interpretative differences within the range allowed be the standard: until Fortran 2003, the units for the "record length" in random access record files were left unspecified. Intel Fortran used "one machine word" (4 bytes in x86) while GFortran used 1 byte. Both were compliant with the standard letter, but incompatible.
Furthermore, even when coding "to standard", you may hit a wall if the compiler does not implement part of the Fnn standard, or it is buggy. Case in point: Intel Fortran 12.0 (old, but it's what I work with) does not the implement the ALLOCATE(y, SOURCE=x) construct for polymorphic x (the "clone allocation"). On the other hand, GFortran has not completely implemented FINAL type-bound procedures (destructors).
In both cases, you will need to find workarounds. For example, for the first issue you can use a special form of the INQUIRE statement (kudos to #haraldkl). In other cases, the workaround might even involve using some kind of feature detection (see autoconf, CMake, etc.) and storing the results as PARAMETER variables in a config.f90 file that is included by your code. Your code would then take decisions based on it, as in:
! config.f90.in (things in #x# would get subtituted by automake, for example)
INTEGER, PARAMETER :: RECORD_LEN_BYTES = #RECORD_LEN_BYTES#
! Some other file which opens a file
INCLUDE "config.f90"
!...
OPEN(u, FILE='DE430.BIN', ACCESS='direct', FORM='unformatted', RECL=56 / RECORD_LEN_BYTES)
People have been having complaints about following the standard since at least the 60s. But those cDEC$ features were put in a for good reasons...
It is valuable to cross compile though and you usually have things caught in one compiler or the other.
For you question #1 "The GNU Fortran is not able to read lines that are too long and I get errors during compilation. As a result I have to break it into multiple lines using the '&' symbol"
In the days of old there was:
options/extended_source
SUBROUTINE...
In fort it is -132, but I have not found a gfortran equivalent to -132 . It may be -ffixed-line-length-n -ffixed-line-length-none -ffree-line-length-n -ffree-line-length-none per the link: http://www.math.uni-leipzig.de/~hellmund/Vorlesung/gfortran.html#SEC8
Also the ifort standard for .f90 and .f95 is the the compiler switch '-free' '-fixed' is the standard <.f90... However one can use -fixed with .f90 and use column 6 and 'D' in column #1... Which is handy with '-D_lines' or '-DD'.
Per the link: https://software.intel.com/sites/default/files/m/f/8/5/8/0/6366-ifort.txt
For you question #2: "A module D.f90 contains all the Global variables declared. However, now I during compilation i get error is in module B.F90. The error I get is 'Unclassified Statement Error', I was able to fix it in some subroutines and functions by locally declaring the variables again."
You probably need to put in the offending line, if you can get an IP waiver.
Making variables local if they are expected to be shared in a /common/ or shared in a module will not work.
If there were in /common/ or PUBLIC then they are shared.
If they are local then they are PRIVATE.
it would be easy to get that error if a PRIVATE statement was in the wrong place, or a USE statement was omitted.
I don't know if this is the right place for things like this, but I am curious about a few aspects of the GCC front-end/back-end architecture:
I know I can compile .o files from C code and link them to C++ code, and I think I can do it the other way round, too. Does this work because the two languages are similar, or because the GCC back-end is really language-independent? Would this work with ADA code too? (I don't even know if that makes sense, since I don't know ADA or if it even has "functions", but the question is understood. If it makes no sense, think "Pascal" or even "my own custom language front-end")
Where would garbage-collection be implemented? For example, a Java front-end. The way I understand, if compiling to a JVM back-end, the "platform" will take care of the GC, and so the front-end needs not do anything about it, but if compiling to native code, would the front-end send garbage-collecting GENERIC code to the back-end, or does it turn on some flag telling the back-end to produce garbage-collecting code? The first makes more sense to me, but that would mean the front-end produces different output based on the target, which seems to miss the point of the GCC's front-end/back-end architecture.
Where would language-specific libraries go? For instance, the standard Java classes or standard C headers. If they are linked in at the end, then could a C program theoretically call functions from the Java library or something like that, since it is just another linked library?
Yes, the backend is at least reasonably language independent. Yes, it works with Ada.
GCJ generates native code which uses a runtime library. The garbage collector is part of the runtime library.
GCJ implements the CNI, which allows you to write code in C++ that can be used as native methods by Java code -- but being able to do this is a consequence of them having designed it in, not just an accidental byproduct of using the same back-end.
It is possible because calling convention is compatible, but name mangling is different (no mangling in C). To call C function from C++ you should declare it with extern "C". And to call C++ function from C you should declare it with mangled name (and may be with additional or different type args). The calling Fortran code is possible in some cases too, but argument passing convention is different (pass by ref in Fortran).
There were actually a converters from C++ to C (cfront) and from fortran to c (f2c) and some solutions from them are still used.
garbage-collection is implemented in run-time library, e.g. boehm. Backend should generate objects compatible with selected GC library.
Compiler driver (g++, gfortran, ..) will add language-specific libraries to linking step.
Is there any way to tell the compiler (gcc/mingw32) when building an object file (lib*.o) to only expose certain functions from the .c file?
The reason I want to do this is that I am statically linking to a 100,000+ line library (SQLite), but am only using a select few of the functions it offers. I am hoping that if I can tell the compiler to only expose those functions, it will optimize out all the code of the functions that are never needed for those few I selected, thus dratically decreasing the size of the library.
I found several possible solutions:
This is what I asked about. It is the gcc equivalent of Windows' dllexpoort:
http://gcc.gnu.org/onlinedocs/gcc-4.6.1/gcc/Code-Gen-Options.html (-fvisibility)
http://gcc.gnu.org/wiki/Visibility
I also discovered link-time code-generation. This allows the linker to see what parts of the code are actually used and get rid of the rest. Using this together with strip and -fwhole-program has given me drastically better results.
http://gcc.gnu.org/onlinedocs/gcc-4.6.1/gcc/Optimize-Options.html (see -flto and -fwhole-program)
Note: This flag only makes sense if you are not compiling the whole program in one call to gcc, which is what I was doing (making a sqlite.o file and then statically linking it in).
The third option which I found but have not yet looked into is mentioned here:
How to remove unused C/C++ symbols with GCC and ld?
That's probably the linkers job, not the compilers. When linking that as a program (.exe), the linker will take care of only importing the relevant symbols, and when linking a DLL, the __dllexport mechanism is probably what you are looking for, or some flags of ld can help you (man ld).