From: "Paul Crone" <paul AT pkcrone DOT globalnet DOT co DOT uk>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: inline asm problem
Date: Sat, 6 May 2000 23:37:26 +0100
Organization: GXSN
Lines: 53
Message-ID: <8f26ln$htd$1@gxsn.com>
References: <007f01bfb76e$0020d320$2f5293c3 AT pkcrone> <8f1t7r DOT 3vs63bt DOT 0 AT buerssner-17104 DOT user DOT cis DOT dfn DOT de>
NNTP-Posting-Host: 195.147.91.95
Mime-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Trace: 957652471 1NNUCNF1G5B5FC393C gxsn.com
X-Complaints-To: abuse AT gxsn DOT com
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2314.1300
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Thanks, problem solved...I used the stack alternative.
Paul
Dieter Buerssner <buers AT gmx DOT de> wrote in message
news:8f1t7r DOT 3vs63bt DOT 0 AT buerssner-17104 DOT user DOT cis DOT dfn DOT de...
> Paul Crone wrote:
>
> >Please help.  I've used the following inline assembly code function in
> >my program using gpp 2.8.1 and all was well.  I've just made a new
> >installation on a different PC using gpp 2.9.5.2 and get a compiler
> >error.  What have I missed?
>
> There is a change in the handling of inline assembly in recent
> versions of gcc. You are not allowed anymore, to put registers
> in the input-list into the clobber list
>
> >inline void set_vesa_bank_pm(int bank_number)
> >{
> >   asm (
> >   " call *%0 "
> >    :                             /* no outputs */
> >    : "r" (pm_bank_switcher),     /* function pointer in any register */
> >   "b" (0),                    /* set %ebx to zero */
> >   "d" (bank_number)           /* bank number in %edx */
> >
> >    : "%eax",                     /* clobber list (we have no way of */
> >   "%ebx",                     /* knowing which registers the VESA */
> >   "%ecx",                     /* code is going to change, so we */
> >   "%edx",                     /* have to assume the worst and list */
> >   "%esi",                     /* them all here) */
> >   "%edi"
> >   );
> >}
>
> On possible (untested) fix, is to use temporary vars like:
>
> /* Assume, the VESA-function does not change ebp */
>   int unused1, unused2, unused3;
>   asm (" call *%5 "
>        : "=b" (unused1), "=d" (unused2), "=a" (unused3)
>        : "0" (0), "1" (bank_number), "2" (pm_bank_switcher)
>        : "ecx", "esi", "edi" /* all other registers */);
>
> Or you push all the registers yourself onto the stack, call the function,
> and pop the registers again.
>
>   /* Untested */
>   asm ("pushal; call *%2; popal"
>        :
>        : "b" (0), "d" (bank_number), "r" (pm_bank_switcher));
>
> --
> Dieter Buerssner