gcl-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Gcl-devel] Re: PLT code and Mach-O


From: Vadim V. Zhytnikov
Subject: [Gcl-devel] Re: PLT code and Mach-O
Date: Thu, 11 Mar 2004 10:07:43 +0300
User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; ru-RU; rv:1.5) Gecko/20031006

Camm Maguire ?????:

Greetings!  OK, could you both (i.e Vadim/Mike and Aurelien) please
try cvs, stable and/or head, now?

Should compile out of the box.  (compiler::link nil "foo") should work
for both of you.  LEADING_UNDERSCORE/GNU_LD should be autodetected.
Aurelien should get an empty si::*plt-table* -- don't know about
windows.

Please confirm/refute.


Sorry but still some problem on windows. Configure and make logs
in attachment.


Take care,

"Vadim V. Zhytnikov" <address@hidden> writes:


Camm Maguire ?????:

Greetings, and thanks!  Unfortunately, I've lost your output of the
original plt.h, but if I recall, the list below can be generated from
that one by 1) dropping one underscore
2) dropping the hyperbolics
3) also dropping main and alloca, which we do not need.
Right?  If so I'll shortly have a commit to test.


Right.  Actually 2) is weaker - only _inverse_
hyperbolic make some trouble.


Take care,
"Vadim V. Zhytnikov" <address@hidden> writes:


Camm Maguire writes:


Greetings.  Basically yes, but for clarity:
=============================================================================
g.c
=============================================================================
typedef struct {
const char *n;
unsigned long ad;
} Plt;
#include <math.h>
static Plt mplt[]={{"cos",(unsigned long)(void *)cos}};
int
main() {
return 0;
}
=============================================================================
gcc -g -Wall -c g.c -o g.o
g.c:7: warning: `mplt' defined but not used
=============================================================================

It works exactly this way.
(


Thanks for your help with this!
Take care,

Here is plt.h which makes plt.o happy
=====================================
MY_PLT(main),
MY_PLT(alloca),
MY_PLT(_setjmp),
MY_PLT(acos),
/*MY_PLT(acosh),*/
MY_PLT(asin),
/*MY_PLT(asinh),*/
MY_PLT(atan),
/*MY_PLT(atanh),*/
MY_PLT(cos),
MY_PLT(cosh),
MY_PLT(exp),
MY_PLT(getc),
MY_PLT(log),
MY_PLT(putc),
MY_PLT(sin),
MY_PLT(sinh),
MY_PLT(tan),
MY_PLT(tanh)
======================================
I commented out three inverse hyperbolic
functions since something weird is hapenning
to them. There is a comment in math.h
======================================
/*
* TODO: asinh, acosh, atanh
*/
======================================




"Vadim V. Zhytnikov" <address@hidden> writes:



Camm Maguire ?????:



Greetings!
"Vadim V. Zhytnikov" <address@hidden> writes:




Camm Maguire:




Greetings!

<snip>

The last possibility is that we put in some configure magic to skip
the plt stuff on windows and macosx.  I don't like this much, as I
fear that eventually the binutils change will catch up there too.

I was about to propose this as a temporary solution for 2.6.2 release.
The problem sounds too serious to make serious improvements in a hurry.
On the other hand I'd to see 2.6.2 released really soon.
Thank to Camm we now have a fix for Linux - at present the only
system where problem with new binutils shows up.
Further investigation and improvement could be moved to 2.7.X
and if more robust scheme will be found it the we can release
2.6.3.

This is possible, but before deciding, could we briefly pursue 2
questions:
1) Do we know what happens on mingw with binutils .8?  Can we find
out
if not?
2) Could you or Mike try to compile that simple little test with
'cos'
only, let me know if it works, and why not if not?

Do you mean plttest.c with just one cos(d) call?






I don't want to spend a lot of time either, but these should be quick
I'm hoping.  I'd like to have the greatest possible chance of never
having to look back at the 2.6 series.
Take care,



--
     Vadim V. Zhytnikov

      <address@hidden>
     <address@hidden>

Attachment: make.log.gz
Description: GNU Zip compressed data

Attachment: configure.log.gz
Description: GNU Zip compressed data

Attachment: config.log.gz
Description: GNU Zip compressed data


reply via email to

[Prev in Thread] Current Thread [Next in Thread]