-ffixed-di vs. strlen

Richard Henderson rth@redhat.com
Fri Feb 13 22:46:00 GMT 2004


On Fri, Feb 13, 2004 at 03:04:05PM -0500, Ian Lance Taylor wrote:
> If we do track local registers, then the i386 backend could choose to
> not do string expansions when there di is dedicated to a local
> register.

Eh.  Maybe.

> It's not
> definitive--there could be a real compiler bug and the local dedicated
> register could be a red herring--so the error message would have to

Actually, if you can identify the actual REG rtx that's live,
then you can look at REG_USERVAR_P and find out whose bug it is.


r~



More information about the Gcc mailing list