UASM 2.57r build problem

[teoberi]

gcc --version

gcc (GCC) 14.2.0
Copyright (C) 2024 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

make -f Makefile-Linux-GCC-64.mak

mkdir GccUnixR
gcc -D UNIX -c -IH -D UNIX -DNDEBUG -O2 -funsigned-char -Wwrite-strings -o GccUnixR/main.o main.c
main.c: In function ‘main’:
main.c:95:17: warning: assignment discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
95 | argv[0] = pEnv;
| ^
gcc -D UNIX -c -IH -D UNIX -DNDEBUG -O2 -funsigned-char -Wwrite-strings -o GccUnixR/apiemu.o apiemu.c
gcc -D UNIX -c -IH -D UNIX -DNDEBUG -O2 -funsigned-char -Wwrite-strings -o GccUnixR/orgfixup.o orgfixup.c
gcc -D UNIX -c -IH -D UNIX -DNDEBUG -O2 -funsigned-char -Wwrite-strings -o GccUnixR/assemble.o assemble.c
assemble.c: In function ‘PassOneChecks’:
assemble.c:1042:32: warning: comparison between pointer and integer
1042 | if ( curr->sym.altname > 1 )
| ^
assemble.c: In function ‘OnePass’:
assemble.c:1181:17: error: implicit declaration of function ‘AddSimdTypes’ [-Wimplicit-function-declaration]
1181 | AddSimdTypes();
| ^~~~~~~~~~~~
In file included from assemble.c:18:
assemble.c: In function ‘get_module_name’:
H/globals.h:51:17: error: implicit declaration of function ‘strupr’; did you mean ‘_strupr’? [-Wimplicit-function-declaration]
51 | #define _strupr strupr
| ^~~~~~
assemble.c:1280:5: note: in expansion of macro ‘_strupr’
1280 | _strupr( ModuleInfo.name );
| ^~~~~~~
assemble.c: In function ‘GetExt’:
assemble.c:1430:23: warning: return discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
1430 | return( EXE_EXT );
| ^
assemble.c:1433:23: warning: return discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
1433 | return( BIN_EXT );
| ^
assemble.c:1435:15: warning: return discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
1435 | return( OBJ_EXT );
| ^
assemble.c:1437:15: warning: return discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
1437 | return( LST_EXT );
| ^
assemble.c:1439:15: warning: return discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
1439 | return( ERR_EXT );
| ^
make: *** [Makefile-Linux-GCC-64.mak:35: GccUnixR/assemble.o] Error 1

[john-terraspace]

Will take a look, I'm on gcc 13.2.0 and it's fine.

[teoberi]

Thank you!
It is a problem for Linux distributions that want to compile UASM from sources.
Until the problem is solved, I will use the binary that for the first time works perfectly with 7-Zip.

[john-terraspace]

Bane of my life ... I had to setup a complete new Linux machine to do the 2.57 build, Ubuntu 24 lts + installed GCC, which brought down 13.x by default - that seems to be fine. GCC 14 however has made more changes. I've already committed a patch to 2.57 branch that fixes the above error for AddSimdTypes - but now it's moaning about strupr.

[grawlinson]

There seems to be an implementation of strupr here:

UASM/assemble.c

Lines 131 to 145 in 540d215

	#if 0
	/* for OW, it would be good to remove the CharUpperA() emulation
	* implemented in apiemu.c. Unfortunately, OW isn't happy with
	* a local, simple version of _strupr() - it still wants to
	* import CharUpperA.
	*/
	char * _strupr( char *src )
	{
	char *dst;
	for ( dst = src; *dst; dst++ )
	if ( *dst >= 'a' && *dst <= 'z' )
	*dst &= ~0x20;
	return( src );
	}
	#endif

However, changing #if 0 to #if 1 results in more compile errors.

compile output

make -f Makefile-Linux-GCC-64.mak
gcc -D __UNIX__ -c -IH -D __UNIX__ -DNDEBUG -O2 -funsigned-char -Wwrite-strings   -o GccUnixR/codegenv2.o codegenv2.c
codegenv2.c: In function ‘hash’:
codegenv2.c:38:29: warning: integer constant is so large that it is unsigned
   38 |         uint_64 fnv_basis = 14695981039346656037;
      |                             ^~~~~~~~~~~~~~~~~~~~
codegenv2.c: In function ‘GenerateInstrHash’:
codegenv2.c:92:21: error: passing argument 1 of ‘hash’ from incompatible pointer type [-Wincompatible-pointer-types]
   92 |         return hash(&hashBuffer, len);
      |                     ^~~~~~~~~~~
      |                     |
      |                     uint_8 (*)[32] {aka unsigned char (*)[32]}
codegenv2.c:35:40: note: expected ‘const uint_8 *’ {aka ‘const unsigned char *’} but argument is of type ‘uint_8 (*)[32]’ {aka ‘unsigned char (*)[32]’}
   35 | static unsigned int hash(const uint_8* data, int size)
      |                          ~~~~~~~~~~~~~~^~~~
codegenv2.c: In function ‘CodeGenV2’:
codegenv2.c:1668:42: error: implicit declaration of function ‘get_curr_srcfile’ [-Wimplicit-function-declaration]
 1668 |                         AddLinnumDataRef(get_curr_srcfile(), GetLineNumber());
      |                                          ^~~~~~~~~~~~~~~~
codegenv2.c:1668:62: error: implicit declaration of function ‘GetLineNumber’ [-Wimplicit-function-declaration]
 1668 |                         AddLinnumDataRef(get_curr_srcfile(), GetLineNumber());
      |                                                              ^~~~~~~~~~~~~
codegenv2.c:1705:44: error: passing argument 7 of ‘BuildMemoryEncoding’ from incompatible pointer type [-Wincompatible-pointer-types]
 1705 |                                 &dispSize, &displacement, matchedInstr, opExpr, &needB, &needX, &needRR, CodeInfo);
      |                                            ^~~~~~~~~~~~~
      |                                            |
      |                                            union <anonymous> *
codegenv2.c:1225:42: note: expected ‘uint_64 *’ {aka ‘long unsigned int *’} but argument is of type ‘union <anonymous> *’
 1225 |         unsigned int* dispSize, uint_64* pDisp, struct Instr_Def* instr, struct expr opExpr[4], bool* needB,
      |                                 ~~~~~~~~~^~~~~
codegenv2.c:1713:54: error: passing argument 3 of ‘BuildVEX’ from incompatible pointer type [-Wincompatible-pointer-types]
 1713 |                         BuildVEX(&needVEX, &vexSize, &vexBytes, matchedInstr, opExpr, needB, needX, opCount);                           /* Create the VEX prefix bytes for both reg and memory operands */
      |                                                      ^~~~~~~~~
      |                                                      |
      |                                                      unsigned char (*)[3]
codegenv2.c:714:69: note: expected ‘unsigned char *’ but argument is of type ‘unsigned char (*)[3]’
  714 | void BuildVEX(bool* needVex, unsigned char* vexSize, unsigned char* vexBytes, struct Instr_Def* instr, struct expr opnd[4], bool needB, bool needX, uint_32 opCount)
      |                                                      ~~~~~~~~~~~~~~~^~~~~~~~
make: *** [Makefile-Linux-GCC-64.mak:35: GccUnixR/codegenv2.o] Error 1

And solving these results in more compile errors. I gave up at this point because I don't know if my fixes were valid.

[grawlinson]

Just noticed that #186 & #216 solve some of these issues. Might be worth pulling them in.

[teoberi]

Fixed by @darealshinji here.