[PATCH] Fix PR target/7784 (Sparc)

[Richard Henderson]

On Tue, Mar 18, 2003 at 04:38:13PM +0100, Eric Botcazou wrote:
> This is a regression from gcc 2.95.3 present on all active branches. GCC 
> aborts when compiling the testcase (that I wasn't able to further shrink) on 
> Ultrasparc at -O2 because it can't reload the following address:

Hmm.  I don't see this failure on 3.3 cross compiling from x86?
Should I be able to see it there on native?

I'd really prefer that we fix reload.  It's *supposed* to handle
this.  LEGITIMIZE_RELOAD_ADDRESS should only be needed for 
optmization, not correctness.


r~