[PATCH] Fix PR optimization/10157

Dale Johannesen dalej@apple.com
Wed Apr 2 21:09:00 GMT 2003


On Wednesday, April 2, 2003, at 12:09  PM, Eric Botcazou wrote:
>> My understanding from the doc is that Sparc should be setting
>> AVOID_CCMODE_COPIES if it doesn't support this pattern.
>> Can we get a consensus on this?  If my understanding is wrong
>> I think the doc needs to be changed.
>
> My understanding of the sentence:
>
> "You should only define this macro if support for copying to/from 
> `CCmode' is
> incomplete."
>
> is that AVOID_CCMODE_COPIES should be defined if the port features the
> "movcc" pattern but doesn't want it to be used in every possible 
> circumstance
> for some reason.

Interesting interpretation.  I'd say if it's not there at all it's
as incomplete as you can get.

> For example, quoted from d30v.h:
>
> /* On the D30V, copying to/from CCmode is complete, but since there 
> are only
>    two CC registers usable for conditional tests, this helps gcse not
> compound the reload problem.  */
> #define AVOID_CCMODE_COPIES

This is clearly not doing what the doc says to do, although there seems
to be a good reason for it.



More information about the Gcc-patches mailing list