Xref: news2.mv.net comp.os.msdos.djgpp:3511
From: richard DOT young AT crc DOT doc DOT ca (Richard Young)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: libm.a/linker *BUG* with test program
Date: 6 May 1996 18:51:35 GMT
Organization: Communications Research Centre
Lines: 51
Message-ID: <4mlhnn$inf@crc-news.doc.ca>
References: <25900 DOT 9605051801 AT bylands DOT dur DOT ac DOT uk> <318DE92E DOT 41C6 AT LSTM DOT Ruhr-UNI-Bochum DOT De> <318E1002 DOT 2781 AT LSTM DOT Ruhr-UNI-Bochum DOT De>
NNTP-Posting-Host: yaker.vpcs.doc.ca
Mime-Version: 1.0
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

In article <318E1002 DOT 2781 AT LSTM DOT Ruhr-UNI-Bochum DOT De>, demmer AT LSTM DOT Ruhr-UNI-Bochum DOT De says...

>Thomas Demmer wrote:
>> The solution is to change src/libm/src/s_scalbn.S to:
>> 
>> #include <machine/asm.h>
>> 
>> ENTRY(scalbn)
>>   ... some incorrect code
>>         ret
>> 
>> gcc -c s_scalbn.S
>> 
>> and replace s_scalbn.o in libm.a
>> 
>> with
>> ar -rv libm.a s_scalbn.o
>
>Sometimes I'm pretty stupid...
>
>The correction _MUST_ be
> ENTRY(scalbn)
>         fidl 12(%esp)
>         fldl  4(%esp)
>         fscale          /* Now we have result in ST(0) and */
>                         /* 2nd operand in ST(1) */
>         ffree  %st(1)   
>         ret
 

The compiler does not like the line 

         fidl 12(%esp)

in the above correct scalbn routine.

It should instead be 

         fildl 12(%esp)

I've put the fix into my libm.a and it has eliminated the 
problems that I was having.  

Thanks to Thomas Demmer for his timely solution.

------------------------------------------------

Richard Young
Communications Research Centre
Ottawa, Canada