This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Your recent reload changes


On Tue, Aug 22, 2000 at 04:57:21PM -0600, Jeffrey A Law wrote:
> I wonder why #0 & #1 aren't RELOAD_FOR_OUTPUT_ADDRESS or something like
> that since they're for the output value.

Because we didn't reload the entire mem for output.  We reloaded the
memory address, which is of course an input to the insn.  Which is
why I was wondering why we wern't just using RELOAD_FOR_INPUT.

Guessing wildly, I'd say we have INPUT and OUTPUT_ADDRESS for some
optional reload nastiness.  But I'm not sure I really want to know.

> I'm not sure.  The reload dependency stuff is really the right way to go
> to describe all this stuff.  But I'm not sure if we've got the time right
> now to tackle that kind of redesign.  I haven't looked at it in the last
> month or so, but my understanding was most of the basic bits where ready,
> but that we were going to need a fair amount of test & benchmark time

Ug.  Ok, so it sounds like option 1 (hack ia64) is the only option.
I'll look at doing that tomorrow.


r~

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]