Re: Trouble with inline asm on m68k

["William A. Gatliff" <bgat at saturn dot billgatliff dot com>]

Guys:


I've come up with a workaround:

  int _len asm("%d1");

This puts _len into %d1.  By assigning whatever it is I'm trying to
get into %d1 to _len, I get the result I'm looking for.

This accomplishes the goal, but forces me to allocate locals to get
there.  For my immediate problem this is ok, but long-term I would
like to understand what I'm doing wrong in the code below.


Suggestions welcome,

b.g.
  

On Sat, Jan 05, 2002 at 09:37:57PM -0600, William A. Gatliff wrote:
> Guys:
> 
> 
> Maybe it's just too late in the evening, maybe I'm just not reading
> the info pages closely enough...
> 
> I'm trying to do this on a CPU32, m68k-elf-gcc (2.95.3):
> 
> ***
> int len;
> ...
> __asm__ __volatile__ ("move.l %0, %d1" : : "r" (len) : "%d1");
> 
> [bgat@mars] m68k-elf-gcc -c -mcpu32 test.c
> test.c:41: invalid 'asm': operand number out of range
> ***
> 
> Seems like gcc is getting tripped up on the %d1 in the opcode,
> thinking it is an operand number instead of a register name.  Turning
> the "%d1" into just "d1" makes gcc happy, but causes an unresolved
> reference to "d1" at link time.  The assembler doesn't know it is a
> register (which makes sense).
> 
> I'm binding to an asm API, so I have to call out the register name.
> 
> I can't be the first person to have this problem.  What's the solution?
> 
> Or better yet, what section of the infopages do I need to re-read?
> 
> 
> Thanks!
> 
> 
> b.g.
> -- 
> Bill Gatliff
> bgat@billgatliff.com
> 
> ------
> Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
> Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com
> 

-- 
Bill Gatliff
bgat@billgatliff.com

------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com