>I'm with Jeremy on this one, I don't see the point adding a l.lw, >one look in the arch manual will make it clear that l.lwz and l.lws is >equivalent on 32-bit.
I think we should at least rephrase the manual, because it's not so obvious at the moment. I don't see why the users should have to read and compare the l.lws and l.lwz sections and wonder whether they have missed something, as those instructions actually seem to do the same thing on 32-bit platforms (!). >The only thing that is confusing at the moment is that l.lws isn't >supported in or1200, something that is easily fixed by the patch >proposed by Jeppe. Having a default, like l.lw does sign extension on 64-bit platforms, introduces more room for mistakes. I would favor a strategy that forces you to be explicit: - On 32-bit implementations, only l.lw should be valid. - On 64-bit implementations, only l.lws and l.lwz should be allowed. Regards, rdiez _______________________________________________ OpenRISC mailing list [email protected] http://lists.openrisc.net/listinfo/openrisc
