gcc 3.4.0 m68k-elf generates illegal code for ColdFire (maybe a fix)

Bernardo Innocenti bernie@develer.com
Tue Jun 1 21:20:00 GMT 2004


Peter Barada wrote:
>>I believe all your patches are already in CVS, with the
>>exception of the addsi3_5200 fix.
>>
>>Is there anything else I've omitted?
> 
> I'mve updated my uberbaum tree, and notice that I never sent in my
> patch to:
> 
> 1) Use mvz/mvs instructions for ColdFire V4 to zero/sign extend
>    byte/word from memory into a data register

The version handling an immediate value is already in.


> 2) Enable use of output_andsi3/output_iorsi3/output_xorsi3 for
>    ColdFire to allow for using the bclr/bset/bchg instructions for
>    single-bit modifications.

Didn't the ColdFire always support bit operations the
same way the m68k did?  Why should this be special-cased
for ColdFire only?

> 3) Update MULL_COST/MULW_COST for ColdFire v3 and higher cores.

Yummy ;-)


> I'm building up the uberbaum tree now, and will send in a patch as
> soon as I verify it.

Thank you!

-- 
  // Bernardo Innocenti - Develer S.r.l., R&D dept.
\X/  http://www.develer.com/



More information about the Gcc mailing list