lightning on mingw

[Reini Urban]

Elliott Slaughter schrieb:

> On Fri, May 16, 2008 at 9:08 AM, Reini Urban <rurban@...> wrote:
>> 2008/5/16 Yaroslav Kavenchuk <kavenchuk@...>:
>>> Reini Urban wrote:
>>>> You'd need latest cvs, because we fixed several bugs in lightning.
>>>> My version is called 1.2c,
>>>> but in the NEWS file they are already calling it 1.3
>>>>
>>>> cvs -d :pserver:anonymous@...:/sources/lightning co
>>>> lightning
>>> Thanks!
>>>
>>> Any source tarball?
>> Ok.
>> http://rurban.xarch.at/software/lightning-cvs-20080516.tar.bz2
>
> Is there something I need to do before compiling? I assumed this would work
>
> ./configure --prefix=${prefix}
> make
> make check
> make install
>
> But I keep getting all tests failing on 'make check' (both with the
> old 1.2 tarball and the new cvs tarball).

latest lightning cvs + mingw:
In file included from C:/gnu/local/include/lightning.h:50,
                   from lightning.c:55,
                   from ../src/eval.d:8269:
C:/gnu/local/include/lightning/funcs-common.h:38: error: syntax error
before '(' token
C:/gnu/local/include/lightning/funcs-common.h:42: warning: no previous
declaration for 'jit_fail'
In file included from ../src/eval.d:8269:
lightning.c:1162:1: warning: "S_operand_ignore" redefined
../src/eval.d:6129:1: warning: this is the location of the previous
definition
In file included from ../src/eval.d:8269:
lightning.c: In function `jit_compile_':
lightning.c:1296: warning: implicit declaration of function `jit_allocai'
lightning.c:3069: error: `setjmp' undeclared (first use in this function)
lightning.c:3069: error: (Each undeclared identifier is reported only once
lightning.c:3069: error: for each function it appears in.)
make: *** [eval.o] Error 1

So you have to fix the mingw header errors (jit_fail conflict, ...) by
yourself. I only fixed it for cygwin.
Please report the error or send patches to lightning@...
--
Reini Urban
http://phpwiki.org/  http://murbreak.at/


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Paolo Bonzini-2]

> latest lightning cvs + mingw:
> In file included from C:/gnu/local/include/lightning.h:50,
>                   from lightning.c:55,
>                   from ../src/eval.d:8269:
> C:/gnu/local/include/lightning/funcs-common.h:38: error: syntax error
> before '(' token

Can you send me a preprocessed output of ../src/eval.d (just add
--save-temps to the CFLAGS and send the resulting eval.i file, bzipped
and to private mail because of the size)?

Paolo


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Elliott Slaughter]

On Sun, May 18, 2008 at 9:57 AM, Paolo Bonzini <bonzini@...> wrote:

>> latest lightning cvs + mingw:
>> In file included from C:/gnu/local/include/lightning.h:50,
>>                  from lightning.c:55,
>>                  from ../src/eval.d:8269:
>> C:/gnu/local/include/lightning/funcs-common.h:38: error: syntax error
>> before '(' token
>
> Can you send me a preprocessed output of ../src/eval.d (just add
> --save-temps to the CFLAGS and send the resulting eval.i file, bzipped and
> to private mail because of the size)?

I don't think I get that far in the build. It warns about automake not
being present, but says I shouldn't need it if I haven't modified
certain files, then goes on to fail anyways (unless I am misreading
the output).

$ CC='gcc -mno-cygwin'; export CC
$ CFLAGS='--save-temps'; export CFLAGS
$ ./configure --prefix=${prefix}
[... config output seems to be fine ...]
$ make
cd . && /bin/sh /clisp-2.45/tools/lightning-cvs/build-aux/missing
--run aclocal-1.9 -I build-aux
/clisp-2.45/tools/lightning-cvs/build-aux/missing: /c/Program
Files/gnuwin32/bin/aclocal-1.9: No such file or directory
WARNING: `aclocal-1.9' is missing on your system.  You should only need it if
         you modified `acinclude.m4' or `configure.ac'.  You might want
         to install the `Automake' and `Perl' packages.  Grab them from
         any GNU archive site.
 cd . && /bin/sh /clisp-2.45/tools/lightning-cvs/build-aux/missing
--run automake-1.9 --gnu
/clisp-2.45/tools/lightning-cvs/build-aux/missing: /c/Program
Files/gnuwin32/bin/automake-1.9: No such file or directory
WARNING: `automake-1.9' is missing on your system.  You should only need it if
         you modified `Makefile.am', `acinclude.m4' or `configure.ac'.
         You might want to install the `Automake' and `Perl' packages.
         Grab them from any GNU archive site.
cd . && /bin/sh /clisp-2.45/tools/lightning-cvs/build-aux/missing --run autoconf
/c/Program Files/gnuwin32/bin/autoconf:
/clisp-2.45/tools/lightning-cvs/c:/progra~1/AutoConf/bin/autom4te: No
such file or directory
/c/Program Files/gnuwin32/bin/autoconf: exec:
/clisp-2.45/tools/lightning-cvs/c:/progra~1/AutoConf/bin/autom4te:
cannot execute: No such file or directory
make: *** [configure] Error 1



--
Elliott Slaughter

"Any road followed precisely to its end leads precisely nowhere." -
Frank Herbert


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Reini Urban]

2008/5/19 Elliott Slaughter <elliottslaughter@...>:

> On Sun, May 18, 2008 at 9:57 AM, Paolo Bonzini <bonzini@...> wrote:
>>> latest lightning cvs + mingw:
>>> In file included from C:/gnu/local/include/lightning.h:50,
>>>                  from lightning.c:55,
>>>                  from ../src/eval.d:8269:
>>> C:/gnu/local/include/lightning/funcs-common.h:38: error: syntax error
>>> before '(' token
>>
>> Can you send me a preprocessed output of ../src/eval.d (just add
>> --save-temps to the CFLAGS and send the resulting eval.i file, bzipped and
>> to private mail because of the size)?
>
> I don't think I get that far in the build. It warns about automake not
> being present, but says I shouldn't need it if I haven't modified
> certain files, then goes on to fail anyways (unless I am misreading
> the output).
>
> $ CC='gcc -mno-cygwin'; export CC
> $ CFLAGS='--save-temps'; export CFLAGS
> $ ./configure --prefix=${prefix}
> [... config output seems to be fine ...]
> $ make
> cd . && /bin/sh /clisp-2.45/tools/lightning-cvs/build-aux/missing
> --run aclocal-1.9 -I build-aux
> /clisp-2.45/tools/lightning-cvs/build-aux/missing: /c/Program
> Files/gnuwin32/bin/aclocal-1.9: No such file or directory
> WARNING: `aclocal-1.9' is missing on your system.  You should only need it if
>         you modified `acinclude.m4' or `configure.ac'.  You might want
>         to install the `Automake' and `Perl' packages.  Grab them from
>         any GNU archive site.
>  cd . && /bin/sh /clisp-2.45/tools/lightning-cvs/build-aux/missing
> --run automake-1.9 --gnu
> /clisp-2.45/tools/lightning-cvs/build-aux/missing: /c/Program
> Files/gnuwin32/bin/automake-1.9: No such file or directory
> WARNING: `automake-1.9' is missing on your system.  You should only need it if
>         you modified `Makefile.am', `acinclude.m4' or `configure.ac'.
>         You might want to install the `Automake' and `Perl' packages.
>         Grab them from any GNU archive site.
> cd . && /bin/sh /clisp-2.45/tools/lightning-cvs/build-aux/missing --run autoconf
> /c/Program Files/gnuwin32/bin/autoconf:
> /clisp-2.45/tools/lightning-cvs/c:/progra~1/AutoConf/bin/autom4te: No
> such file or directory
> /c/Program Files/gnuwin32/bin/autoconf: exec:
> /clisp-2.45/tools/lightning-cvs/c:/progra~1/AutoConf/bin/autom4te:
> cannot execute: No such file or directory
> make: *** [configure] Error 1

I have a cygwin mingw32 cross-compiler setup - plain cygwin with a
sane shell and sane tools, so maybe I'll get farther. But I have no
plain MSYS setup unfortunately.

make distclean # optional
mkdir .build
cd .build
CFLAGS="-mno-cygwin" ../configure --target=i686-pc-mingw32
make
make check
=> 19 of 19 tests failed

But manual inspection of the test results (tests/*.log) show that all
succeeded,
so it looks like a \r\n issue to me.
I'll try to reproduce the jit_fail macro issue with the MSYS headers.
w32api with -mno-cygwin is doing fine.
--
Reini Urban
http://phpwiki.org/ http://murbreak.at/


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Paolo Bonzini-2]

> But manual inspection of the test results (tests/*.log) show that all
> succeeded,
> so it looks like a \r\n issue to me.

Ok, this has been fixed.

Paolo


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Reini Urban]

2008/5/19 Paolo Bonzini <bonzini@...>:
>
>> But manual inspection of the test results (tests/*.log) show that all
>> succeeded,
>> so it looks like a \r\n issue to me.
>
> Ok, this has been fixed.

My fix, which passes now all tests under mingw32 is:

diff -u tests/run-test~ tests/run-test
--- tests/run-test~     2007-01-29 10:57:46.000000000 +0100
+++ tests/run-test      2008-05-19 14:20:49.062500000 +0200
@@ -1,6 +1,6 @@
 #! /bin/sh

-./$1 > $1.log
+./$1 | sed -e 's,0x0d,,' > $1.log
 if test $? = 77; then
   exit 77
 fi



--
Reini Urban
http://phpwiki.org/ http://murbreak.at/


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Yaroslav Kavenchuk]

Paolo Bonzini wrote:
> Can you send me a preprocessed output of ../src/eval.d (just add
> --save-temps to the CFLAGS and send the resulting eval.i file, bzipped
> and to private mail because of the size)?

You got a file?
If not - get here:
  http://kavenchuk.googlepages.com/eval.i.tar.bz2

--
WBR, Yaroslav Kavenchuk.


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Paolo Bonzini-2]

>> Can you send me a preprocessed output of ../src/eval.d (just add
>> --save-temps to the CFLAGS and send the resulting eval.i file, bzipped
>> and to private mail because of the size)?
>
> You got a file?
> If not - get here:
>  http://kavenchuk.googlepages.com/eval.i.tar.bz2

Thanks.

Something is defining __unused__ to "(void)". :-(

You can test this file: if it doesn't work with an error message on thee
last line, the culprit is probably MinGW.  If it works, the culprit is
CLISP.  If it works with another error message, then we know nothing
more than we did. :-)

---start---
#include <sys/types.h>
#include <errno.h>
#include <io.h>
#include <locale.h>
#include <malloc.h>
#include <setjmp.h>
#include <stdarg.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/timeb.h>
#include <time.h>

#include <_mingw.h>
#include <iconv.h>
#include <libintl.h>
#include <sigsegv.h>

#include <windows.h>
#include <basetsd.h>
#include <basetyps.h>
#include <cguid.h>
#include <commctrl.h>
#include <oaidl.h>
#include <objbase.h>
#include <objfwd.h>
#include <objidl.h>
#include <ole2.h>
#include <oleauto.h>
#include <olectlid.h>
#include <oleidl.h>
#include <poppack.h>
#include <prsht.h>
#include <pshpack2.h>
#include <pshpack4.h>
#include <rpc.h>
#include <rpcdce.h>
#include <rpcdcep.h>
#include <rpcndr.h>
#include <rpcnsi.h>
#include <rpcnsip.h>
#include <rpcnterr.h>
#include <shellapi.h>
#include <shlguid.h>
#include <shlobj.h>
#include <unknwn.h>
#include <winbase.h>
#include <wincon.h>
#include <windef.h>
#include <winerror.h>
#include <wingdi.h>
#include <winnetwk.h>
#include <winnls.h>
#include <winnt.h>
#include <winreg.h>
#include <winsock2.h>
#include <winsvc.h>
#include <winuser.h>
#include <winver.h>
#include <ws2tcpip.h>
#include <wtypes.h>
static int jit_fail(const char *, const char*, int, const char *)
    __attribute__((__unused__));
---end---

Paolo


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Yaroslav Kavenchuk-2]

Paolo Bonzini wrote:
> You can test this file: if it doesn't work with an error message on thee
> last line, the culprit is probably MinGW.  If it works, the culprit is
> CLISP.  If it works with another error message, then we know nothing
> more than we did. :-)
>

$ cat unused_test.c
#include <sys/types.h>
#include <errno.h>
#include <io.h>
#include <locale.h>
#include <malloc.h>
#include <setjmp.h>
#include <stdarg.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/timeb.h>
#include <time.h>

#include <_mingw.h>
#include <iconv.h>
#include <libintl.h>
#include <sigsegv.h>

#include <windows.h>
#include <basetsd.h>
#include <basetyps.h>
#include <cguid.h>
#include <commctrl.h>
#include <oaidl.h>
#include <objbase.h>
#include <objfwd.h>
#include <objidl.h>
#include <ole2.h>
#include <oleauto.h>
#include <olectlid.h>
#include <oleidl.h>
#include <poppack.h>
#include <prsht.h>
#include <pshpack2.h>
#include <pshpack4.h>
#include <rpc.h>
#include <rpcdce.h>
#include <rpcdcep.h>
#include <rpcndr.h>
#include <rpcnsi.h>
#include <rpcnsip.h>
#include <rpcnterr.h>
#include <shellapi.h>
#include <shlguid.h>
#include <shlobj.h>
#include <unknwn.h>
#include <winbase.h>
#include <wincon.h>
#include <windef.h>
#include <winerror.h>
#include <wingdi.h>
#include <winnetwk.h>
#include <winnls.h>
#include <winnt.h>
#include <winreg.h>
#include <winsock2.h>
#include <winsvc.h>
#include <winuser.h>
#include <winver.h>
#include <ws2tcpip.h>
#include <wtypes.h>
static int jit_fail(const char *, const char*, int, const char *)
     __attribute__((__unused__));

int
main ()
{
   return 0;
};


$ gcc -mno-cygwin -I/usr/local/include unused_test.c

$


All works. What next?

--
WBR, Yaroslav Kavenchuk.


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Paolo Bonzini-2]

> All works. What next?

I'll leave this to the CLISP guys.

Paolo


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Sam Steingold]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Paolo Bonzini wrote:
|>> Can you send me a preprocessed output of ../src/eval.d (just add
|>> --save-temps to the CFLAGS and send the resulting eval.i file,
|>> bzipped and to private mail because of the size)?
|>
|> You got a file?
|> If not - get here:
|>  http://kavenchuk.googlepages.com/eval.i.tar.bz2
|
| Thanks.
|
| Something is defining __unused__ to "(void)". :-(

clisp does not do that.
clisp defines "unused" (without underscores) to "(void)" in lispbibl.d

does this patch help?



- --- eval.d.~1.251.~ 2008-03-25 18:05:52.000000000 -0400
+++ eval.d 2008-05-21 09:58:28.000019000 -0400
@@ -8266,6 +8266,7 @@ global maygc void init_cclosures (void)

~ #if defined(USE_JITC)
~  #if defined(lightning)
+  #undef unused
~   #include "lightning.c"
~  #else
~   #error USE_JITC: what is your JITC flavor?

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFINCqlPp1Qsf2qnMcRAq0KAJ46GfyFaCsEmPYXLRtx4CRXd0M2RwCfVxgo
6/h0FXOF+9qlzoVkKmGm8xw=
=9r8u
-----END PGP SIGNATURE-----



_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Yaroslav Kavenchuk]

Sam Steingold wrote:
> does this patch help?
>

Partially:

gcc -mno-cygwin -I/usr/local/include  -Igllib --save-temps -W -Wswitch
-Wcomment -Wpointer-arith -Wimplicit -Wreturn-type
-Wmissing-declarations -Wno-sign-compare -falign-functions=4 -D_WIN32 -g
-DDEBUG_OS_ERROR -DDEBUG_SPVW -DDEBUG_BYTECODE -DSAFETY=3 -DUNICODE
-DHEAPCODES -DUSE_JITC -Dlightning -DDYNAMIC_FFI -I. -Wno-unused -c eval.c
In file included from C:/msys/local/include/lightning.h:44,
                  from lightning.c:55,
                  from ../src/eval.d:8270:
C:/msys/local/include/lightning/asm.h:48:1: warning: "_r1" redefined
In file included from C:/msys/local/include/lightning/asm.h:46,
                  from C:/msys/local/include/lightning.h:44,
                  from lightning.c:55,
                  from ../src/eval.d:8270:
C:/msys/local/include/lightning/asm-i386.h:125:1: warning: this is the
location of the previous definition
In file included from ../src/eval.d:8270:
lightning.c:1162:1: warning: "S_operand_ignore" redefined
../src/eval.d:6129:1: warning: this is the location of the previous
definition
In file included from ../src/eval.d:7:
../src/lispbibl.d:9042: warning: volatile register variables don't work
as you might wish
../src/eval.d: In function `interpret_bytecode_':
../src/eval.d:5768: warning: variable 'byteptr' might be clobbered by
`longjmp' or `vfork'
In file included from ../src/eval.d:8270:
lightning.c: In function `jit_compile_':
lightning.c:3069: error: `setjmp' undeclared (first use in this function)
lightning.c:3069: error: (Each undeclared identifier is reported only once
lightning.c:3069: error: for each function it appears in.)
make: *** [eval.o] Error 1


--
WBR, Yaroslav Kavenchuk.


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Yaroslav Kavenchuk]

Yaroslav Kavenchuk wrote:
> In file included from ../src/eval.d:8270:
> lightning.c: In function `jit_compile_':
> lightning.c:3069: error: `setjmp' undeclared (first use in this
> function)
> lightning.c:3069: error: (Each undeclared identifier is reported only
> once
> lightning.c:3069: error: for each function it appears in.)
> make: *** [eval.o] Error 1

What is `setjmp'? Function from <setjmp.h> or other?

When first - from eval.i:

= (((_ul )((int)(((unsigned long) ((setjmp)))))) - ((_ul )(_jit.x.pc)))

It is function call?

When not function - where it defined?

Any help?

--
WBR, Yaroslav Kavenchuk.


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Paolo Bonzini-2]

Yaroslav Kavenchuk wrote:

> Yaroslav Kavenchuk wrote:
>> In file included from ../src/eval.d:8270:
>> lightning.c: In function `jit_compile_':
>> lightning.c:3069: error: `setjmp' undeclared (first use in this
>> function)
>> lightning.c:3069: error: (Each undeclared identifier is reported only
>> once
>> lightning.c:3069: error: for each function it appears in.)
>> make: *** [eval.o] Error 1
>
> What is `setjmp'? Function from <setjmp.h> or other?
>
> When first - from eval.i:
>
> = (((_ul )((int)(((unsigned long) ((setjmp)))))) - ((_ul )(_jit.x.pc)))
>
> It is function call?
>
> When not function - where it defined?

I think the problem is that setjmp is a macro in mingw, or something
like that.

Paolo


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Yaroslav Kavenchuk]

Paolo Bonzini wrote:

>> What is `setjmp'? Function from <setjmp.h> or other?
>>
>> When first - from eval.i:
>>
>> = (((_ul )((int)(((unsigned long) ((setjmp)))))) - ((_ul
> )(_jit.x.pc)))
>> It is function call?
>>
>> When not function - where it defined?
>
> I think the problem is that setjmp is a macro in mingw, or something
> like that.
>

Yes.

#define setjmp(x) _setjmp(x)

What should be?

--
WBR, Yaroslav Kavenchuk.


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Paolo Bonzini-2]

> #define    setjmp(x)    _setjmp(x)
>
> What should be?

#define setjmp _setjmp

would be better, because then you can take the address of it.  You can
put at some strategic place (don't know where)

#undef setjmp
#define setjmp _setjmp

Paolo


_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning

[Sam Steingold]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Paolo Bonzini wrote:
|> #define    setjmp(x)    _setjmp(x)
|>
|> What should be?
|
| #define setjmp _setjmp
|
| would be better, because then you can take the address of it.  You can
| put at some strategic place (don't know where)
|
| #undef setjmp
| #define setjmp _setjmp

we already have this in lispbibl.d:

/* non-local exits */
#include <setjmp.h>
#if defined(UNIX) && defined(HAVE__JMP) && !defined(UNIX_LINUX) &&
!defined(UNIX_GNU) && !defined(UNIX_BEOS) && !defined(UNIX_CYGWIN32)
~  /* The following routines are more efficient (don't use with
signal-masks): */
~  #undef setjmp
~  #undef longjmp
~  #define setjmp  _setjmp
~  #define longjmp  _longjmp
~  #ifdef LONGJMP_RETURNS
~    /* _longjmp(jmpbuf,value) can return if jmpbuf is invalid. */
~    #undef longjmp
~    #define longjmp(x,y)  (_longjmp(x,y), NOTREACHED)
~  #endif
#endif
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFINXfZPp1Qsf2qnMcRAgzgAJ9P56YFqmWHTv/lohgfXK5nP1jkHQCgsJ1z
LqAcy0yEhjpuEkv/RxfcrGI=
=1S4F
-----END PGP SIGNATURE-----



_______________________________________________
Lightning mailing list
Lightning@...
http://lists.gnu.org/mailman/listinfo/lightning