From: Ben Peddell <killer DOT lightspeed AT bigpond DOT com>
User-Agent: Mozilla/5.0 (X11; U; Linux i586; en-US; rv:1.1) Gecko/20020826
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Bug 00314 -- div() still broken
References: <3e9c6920$0$21928$afc38c87@> <fA9pa.20115$1s1 DOT 299825 AT newsfeeds DOT bigpond DOT com> <3EA5477F DOT 2020901 AT cyberoptics DOT com> <SzSpa.21592$1s1 DOT 332473 AT newsfeeds DOT bigpond DOT com> <3ea85d95$0$12489$4c41069e AT reader1 DOT ash DOT ops DOT us DOT uu DOT net> <Ah9qa.22112$1s1 DOT 343387 AT newsfeeds DOT bigpond DOT com> <3ea97d4c$0$19415$4c41069e AT reader1 DOT ash DOT ops DOT us DOT uu DOT net>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
Lines: 36
Message-ID: <xvvqa.22796$1s1.351097@newsfeeds.bigpond.com>
Date: Sat, 26 Apr 2003 23:17:04 +1000
NNTP-Posting-Host: 144.134.89.66
X-Trace: newsfeeds.bigpond.com 1051362397 144.134.89.66 (Sat, 26 Apr 2003 23:06:37 EST)
NNTP-Posting-Date: Sat, 26 Apr 2003 23:06:37 EST
Organization: Telstra BigPond Internet Services (http://www.bigpond.com)
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Eric Rudd wrote:
> Ben Peddell wrote:
> 
>> Obviously, some sort of optimization puts the div_t structure in 
>> %edx:%eax instead of 16(%ebp). It would be a good optimization if it 
>> didn't break the libraries. 
> 
> 
> Yes, indeed.  So, it looks as if there is a gcc bug here.  Where are the 
> calling conventions documented?  I couldn't find them in the gcc.ixx 
> docs.  I think I ought to isolate this and submit a bug report, if gcc 
> is indeed at fault.  I suppose a workaround would be to recompile libc, 
> but then one wonders whether all the other libraries are also broken.
> 
> -Eric Rudd
> rudd AT cyberoptics DOT com
> 

I'm pretty sure I saw somewhere that char, short, int and long are 
returned in eax. long long is returned in edx:eax. Anything longer than 
that is returned in memory.
I guess that without the optimization, gcc returns the div_t structure 
in memory, but with the optimization, gcc returns it in edx:eax since it 
sees that it is the length of a long long.

GCC 3.2.1 (the gcc I have under DJGPP) does not optimize the div_t into 
a long long when given -O3 or -fomit-frame-pointer.
In fact, -fomit-frame-pointer does nothing.

I went to www.delorie.com to get gcc 3.2.2, and had a look at your bug 
report whilst I was there.
What command line did you use?

Anyway, I'm getting gcc 3.2.2.
I'll get back tomorrow when I install it.