gcc 2.95.2 compilation problems
Chris Bourke
I'm trying to compile crypto++ 4.2 in a unix environment using gcc 2.95.2,
but I get the following errors:
gmake -f GNUmakefile
g++ -g -pipe -c 3way.cpp
g++: -pipe is not supported.
gmake: *** [3way.o] Error 1
does 4.2 NOT support gcc 2.95.2 as it says?
Wei Dai
just makes the compile go faster and is not crucial. Edit GNUmakefile in
your Crypto++ directory and delete "-pipe".
Chris Bourke
> Maybe it's something with your OS that makes -pipe not supported. But it
> just makes the compile go faster and is not crucial. Edit GNUmakefile in
> your Crypto++ directory and delete "-pipe".
I got rid of the reference to the -pipe flag, but now I get a whole lot of
error messages:
3way.cpp:20: parse error before `{'
3way.cpp:20: stray '\' in program
3way.cpp:21: stray '\' in program
3way.cpp:22: `a0' was not declared in this scope
3way.cpp:22: stray '\' in program
3way.cpp:23: ANSI C++ forbids declaration `a0' with no type
3way.cpp:23: `a2' was not declared in this scope
3way.cpp:23: stray '\' in program
3way.cpp:24: ANSI C++ forbids declaration `a2' with no type
3way.cpp:24: stray '\' in program
3way.cpp:28: parse error before `{'
3way.cpp:28: stray '\' in program
3way.cpp:29: stray '\' in program
3way.cpp:30: ANSI C++ forbids declaration `b2' with no type
3way.cpp:30: `a2' was not declared in this scope
3way.cpp:30: implicit declaration of function `int rotlFixed(...)'
3way.cpp:30: stray '\' in program
3way.cpp:31: ANSI C++ forbids declaration `b0' with no type
3way.cpp:31: `a0' was not declared in this scope
3way.cpp:31: stray '\' in program
3way.cpp:32: ANSI C++ forbids declaration `a0' with no type
3way.cpp:32: `a1' was not declared in this scope
3way.cpp:32: stray '\' in program
3way.cpp:33: ANSI C++ forbids declaration `a2' with no type
3way.cpp:33: `a1' was not declared in this scope
3way.cpp:33: stray '\' in program
3way.cpp:34: syntax error before `^='
3way.cpp:34: stray '\' in program
3way.cpp:39: stray '\' in program
3way.cpp:40: stray '\' in program
3way.cpp:41: ANSI C++ forbids declaration `c' with no type
3way.cpp:41: `a1' was not declared in this scope
3way.cpp:41: stray '\' in program
3way.cpp:42: ANSI C++ forbids declaration `c' with no type
3way.cpp:42: redefinition of `int c'
3way.cpp:41: `int c' previously defined here
3way.cpp:42: stray '\' in program
3way.cpp:43: ANSI C++ forbids declaration `b0' with no type
3way.cpp:43: redefinition of `int b0'
3way.cpp:31: `int b0' previously defined here
3way.cpp:43: `a1' was not declared in this scope
3way.cpp:43: stray '\' in program
3way.cpp:44: ANSI C++ forbids declaration `b1' with no type
3way.cpp:44: `a1' was not declared in this scope
3way.cpp:44: `a1' was not declared in this scope
3way.cpp:44: stray '\' in program
3way.cpp:45: syntax error before `^='
3way.cpp:45: stray '\' in program
3way.cpp:46: syntax error before `^='
3way.cpp:46: stray '\' in program
3way.cpp:47: syntax error before `^='
3way.cpp:47: stray '\' in program
3way.cpp:51: stray '\' in program
3way.cpp:52: stray '\' in program
3way.cpp:52: stray '\' in program
3way.cpp:53: stray '\' in program
3way.cpp:53: stray '\' in program
3way.cpp:56: `word32' was not declared in this scope
3way.cpp:56: parse error before `,'
3way.cpp: In function `void GenerateRoundConstants(...)':
3way.cpp:58: `rounds' undeclared (first use this function)
3way.cpp:58: (Each undeclared identifier is reported only once
3way.cpp:58: for each function it appears in.)
3way.cpp:60: `rtab' undeclared (first use this function)
3way.cpp:60: `strt' undeclared (first use this function)
3way.cpp: At top level:
3way.cpp:66: syntax error before `::'
3way.cpp:70: `START_E' was not declared in this scope
3way.cpp:70: `rc' was not declared in this scope
3way.cpp:70: ANSI C++ forbids declaration `GenerateRoundConstants' with no type
3way.cpp:70: `int GenerateRoundConstants' redeclared as different kind of symbol
3way.cpp:57: previous declaration of `void GenerateRoundConstants(...)'
3way.cpp:70: initializer list being treated as compound expression
3way.cpp:71: parse error before `for'
3way.cpp:71: parse error before `;'
3way.cpp:71: syntax error before `++'
3way.cpp:75: syntax error before `::'
3way.cpp:80: syntax error before `::'
3way.cpp:89: ANSI C++ forbids declaration `a0' with no type
3way.cpp:89: redefinition of `int a0'
3way.cpp:32: `int a0' previously defined here
3way.cpp:89: `word32' was not declared in this scope
3way.cpp:89: parse error before `)'
3way.cpp:90: ANSI C++ forbids declaration `a1' with no type
3way.cpp:90: `word32' was not declared in this scope
3way.cpp:90: parse error before `)'
3way.cpp:90: `in' was not declared in this scope
3way.cpp:91: ANSI C++ forbids declaration `a2' with no type
3way.cpp:91: redefinition of `int a2'
3way.cpp:33: `int a2' previously defined here
3way.cpp:91: `word32' was not declared in this scope
3way.cpp:91: parse error before `)'
3way.cpp:91: `in' was not declared in this scope
3way.cpp:94: parse error before `for'
3way.cpp:94: parse error before `;'
3way.cpp:94: syntax error before `++'
3way.cpp:97: syntax error before `^='
3way.cpp:98: syntax error before `^='
3way.cpp:99: stray '\' in program
3way.cpp:101: syntax error before `^='
3way.cpp:102: syntax error before `^='
3way.cpp:103: syntax error before `^='
3way.cpp:104: stray '\' in program
3way.cpp:111: parse error before `*'
3way.cpp:112: parse error before `*'
3way.cpp:113: parse error before `*'
3way.cpp:117: syntax error before `::'
3way.cpp:121: `START_D' was not declared in this scope
3way.cpp:121: `rc' was not declared in this scope
3way.cpp:121: ANSI C++ forbids declaration `GenerateRoundConstants' with no type
3way.cpp:121: redefinition of `int GenerateRoundConstants'
3way.cpp:70: `int GenerateRoundConstants' previously defined here
3way.cpp:121: initializer list being treated as compound expression
3way.cpp:122: parse error before `for'
3way.cpp:122: parse error before `;'
3way.cpp:122: syntax error before `++'
3way.cpp:124: stray '\' in program
3way.cpp:125: stray '\' in program
3way.cpp:126: ANSI C++ forbids declaration `k' with no type
3way.cpp:126: implicit declaration of function `int byteReverse(...)'
3way.cpp:126: invalid initializer
3way.cpp:127: ANSI C++ forbids declaration `k' with no type
3way.cpp:127: conflicting types for `int k[1]'
3way.cpp:126: previous declaration as `int k[0]'
3way.cpp:127: invalid initializer
3way.cpp:128: ANSI C++ forbids declaration `k' with no type
3way.cpp:128: conflicting types for `int k[2]'
3way.cpp:127: previous declaration as `int k[1]'
3way.cpp:128: invalid initializer
3way.cpp:129: parse error before `}'
3way.cpp:131: syntax error before `::'
3way.cpp:136: syntax error before `::'
3way.cpp:141: ANSI C++ forbids declaration `a0' with no type
3way.cpp:141: redefinition of `int a0'
3way.cpp:89: `int a0' previously defined here
3way.cpp:141: `word32' was not declared in this scope
3way.cpp:141: parse error before `)'
3way.cpp:142: ANSI C++ forbids declaration `a1' with no type
3way.cpp:142: redefinition of `int a1'
3way.cpp:90: `int a1' previously defined here
3way.cpp:142: `word32' was not declared in this scope
3way.cpp:142: parse error before `)'
3way.cpp:142: `in' was not declared in this scope
3way.cpp:143: ANSI C++ forbids declaration `a2' with no type
3way.cpp:143: redefinition of `int a2'
3way.cpp:91: `int a2' previously defined here
3way.cpp:143: `word32' was not declared in this scope
3way.cpp:143: parse error before `)'
3way.cpp:143: `in' was not declared in this scope
3way.cpp:150: stray '\' in program
3way.cpp:151: parse error before `for'
3way.cpp:151: parse error before `;'
3way.cpp:151: syntax error before `++'
3way.cpp:154: syntax error before `^='
3way.cpp:155: syntax error before `^='
3way.cpp:156: stray '\' in program
3way.cpp:158: syntax error before `^='
3way.cpp:159: syntax error before `^='
3way.cpp:160: syntax error before `^='
3way.cpp:161: stray '\' in program
3way.cpp:162: stray '\' in program
3way.cpp:165: parse error before `*'
3way.cpp:166: parse error before `*'
3way.cpp:167: parse error before `*'
Wei Dai
Chris Bourke
> You forgot to use the -a option when unzipping.
>
Bingo, that was indeed the problem. However, the entire library is still
not compiling properly. It goes along until it tries to compile
cryptlib.cpp, then we get the following:
cryptlib.cpp: In method `class string
CryptoPP::NullRNG()::NullRNG::AlgorithmNam
e() const':
cryptlib.cpp:125: Internal compiler error in `scope_die_for', at
dwarf2out.c:779
4
Please submit a full bug report.
See <URL:http://www.gnu.org/software/gcc/faq.html#bugreport> for
instructions.
gmake: *** [cryptlib.o] Error 1
So, is it really a bug in the gcc compiler or is there something else
wrong?
Wei Dai
Crypto++ 5.0 with GCC 2.95.2 on one of the SourceForge compile farm
machines, so it might be something specific to the GCC for your platform.
Try upgrading it.
Zooko
Wei Dai wrote:
>
> Looks like a compiler bug to me, although I have been able to compile
> Crypto++ 5.0 with GCC 2.95.2 on one of the SourceForge compile farm
> machines, so it might be something specific to the GCC for your platform.
> Try upgrading it.
Did you compile Crypto++ 5.0 on MacOSX? That would be exciting. The Mnet
project [1] has not yet succeeded at compiling and linking any version of
Crypto++ on MacOSX, and that is why Mnet doesn't run on MacOSX yet.
Regards,
Zooko
Wei Dai
> Did you compile Crypto++ 5.0 on MacOSX? That would be exciting. The Mnet
> project [1] has not yet succeeded at compiling and linking any version of
> Crypto++ on MacOSX, and that is why Mnet doesn't run on MacOSX yet.
No I didn't. I think I compiled Crypto++ 5.0 with GCC 2.95.2 on FreeBSD or
Linux.
You can't compile any version of Crypto++ on MacOS X, even with the latest
GCC and binutils? Have you tried to work with GCC developers to debug the
problem?
Brian Quirt
>You can't compile any version of Crypto++ on MacOS X, even with the
>latest GCC and binutils? Have you tried to work with GCC developers to
>debug the problem?
One part of the problem may be that Apple's moving towards gcc 3.x
for new development, but they're only releasing it as part of OSX 10.2.x-
only packages (and since 10.2.x a) is full-price even for those who own
10.1.x, and b) isn't even compatible with all 10.1.x applications, there
are quite a few (myself included) who haven't switched yet (although if I
weren't living on a student budget I might well have switched by now).
-Brian
--
Brian Quirt (bri...@ncf.ca)
Proud Member of the Society for the Conservation of Angular Momentum
Visit the society web site at <http://www.ncf.ca/~cy856/bio/scam.html>
Or visit my web site at <http://www.ncf.ca/~cy856/>
Wei Dai
> project [1] has not yet succeeded at compiling and linking any version of
> Crypto++ on MacOSX, and that is why Mnet doesn't run on MacOSX yet.
I'd like to get this problem fixed. According to this thread:
http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&threadm=adf08a26.0208111546.659b0f78%40posting.google.com&rnum=1&prev=/groups%3Fq%3Dconstructors_used%26hl%3Den%26lr%3D%26ie%3DUTF-8%26sa%3DN%26scoring%3Dd
the ".constructors_used" problem can be fixed by running 'sudo gcc_select
3'.
Jan Gacnik reported compile errors with GCC 3.1 on MacOS. Can either of
you please post the error messages?
Wei Dai
> One part of the problem may be that Apple's moving towards gcc 3.x
> for new development, but they're only releasing it as part of OSX 10.2.x-
> only packages (and since 10.2.x a) is full-price even for those who own
> 10.1.x, and b) isn't even compatible with all 10.1.x applications, there
> are quite a few (myself included) who haven't switched yet (although if I
> weren't living on a student budget I might well have switched by now).
You can download GCC 3.x directly from gcc.gnu.org. Build and install it,
and then run 'sudo gcc_select 3'. Please try that and let me know if it
works.
Jan Gacnik
Hello.
This is message (I sent you the whole make dump to your private
address).
c++ -g -pipe -fno-pic -c rsa.cpp
{standard input}:6807:FATAL:Symbol
__ZN8CryptoPP16PK_FinalTemplateINS_16TF_EncryptorImplINS_16TF_SchemeOpti
onsINS_5TF_ESINS_8PKCS1v15ENS_3RSAEiEES5_NS_28PKCS_EncryptionPaddingSche
meEEEEEEC4INS2_INS_6TF_SSAIS4_NS_3SHAES5_iEES5_NS_27PKCS_SignaturePaddin
gSchemeEEEEEz already defined.
JG
Michael Monsen
Gacnik <gac...@stud.uni-hannover.de> was overheard saying to the guard:
>Hello,
>
>Hello.
>
>This is message (I sent you the whole make dump to your private
>address).
>
>c++ -g -pipe -fno-pic -c rsa.cpp
>{standard input}:6807:FATAL:Symbol
>__ZN8CryptoPP16PK_FinalTemplateINS_16TF_EncryptorImplINS_16TF_SchemeOpti
>onsINS_5TF_ESINS_8PKCS1v15ENS_3RSAEiEES5_NS_28PKCS_EncryptionPaddingSche
>meEEEEEEC4INS2_INS_6TF_SSAIS4_NS_3SHAES5_iEES5_NS_27PKCS_SignaturePaddin
>gSchemeEEEEEz already defined.
Was this ever resolved? I'm seeing the same behavior on my MacOS X
Jaguar (10.2.1) box. I tried using "sudo gcc_select 3" but it did not
make a difference, I still get the error (the result of the command said
that I was already using gcc 3.1)
I'd be happy to provide any other debugging output or other
information needed.
--
**Disclaimer: Opinions cited by me are not necessarily my opinions.
Facts cited by me are not necessarily facts.
Wei Dai
let us know the result.
Michael Monsen
Dai <wei...@weidai.com> was overheard saying to the guard:
>It's probably a bug in GCC 3.1. Please try upgrading to 3.2 or later, and
>let us know the result.
OK, to follow up on this problem:
I used the command "sudo gcc_select 2" to switch the default compiler
to gcc 2.95. Compiling Crypto++ 5.0 worked great, and all of the
tests/benchmarks passed.
If I switch back to gcc 3.1 which is the default compiler for MacOS
10.2.1, then of course I get the error that was talked about in this
thread. Also, it doesn't seem to like the -fno-coalesce-templates option
(perhaps this doesn't exist for gcc 3.1?) After removing that compiler
option it would begin to compile but then error out on rsa.cpp, like
before.
I would happily upgrade to gcc 3.2, except that it is my understanding
that the version of gcc that Apple includes with MacOS X includes some
Apple extensions for MacOS X. I think I will have to either use gcc 2.95
or wait for Apple to update their Developer Tools with gcc 3.2.
When Apple releases gcc 3.2 for MacOS X I'll get back to the mailing
list with the results.
Wei Dai
> I would happily upgrade to gcc 3.2, except that it is my understanding
> that the version of gcc that Apple includes with MacOS X includes some
> Apple extensions for MacOS X. I think I will have to either use gcc 2.95
> or wait for Apple to update their Developer Tools with gcc 3.2.
I have the impression (may have read somewhere) that gcc 3.2 already
works fine on MacOS X without extensions. You might want to double check
that or just try it out.