Re: ia64 ltoff22x/ldxmov relaxation

[David Mosberger <davidm at napali dot hpl dot hp dot com>]

>>>>> On Sat, 1 Mar 2003 00:09:20 -0800, Richard Henderson <rth at redhat dot com> said:

  >> It certainly would complicate life, so it's not something I'd just
  >> without good reasons.

  Rich> Would it help if the binaries went into a different region?
  Rich> E.g. place the whole binary at 0x1000...0000?

That might work.  But then again, it might get in the way.

There is no 0x10..0 region, there is only 0x00..0 and 0x20..0.  The
former is currently used for x86 emulation only, though there is
currently nothing that would stop you from using it on ia64.  If you
want to play with this, I'd pick an starting address that is >4GB to
catch pointer-truncation bugs, but not too big to avoid problems with
running out of mappable address space when using 4KB page size).

The reason I'm saying it might get in the way is that so far I have
treated region 0 as sort of an available region so that if we ever
needed another special region, we could use it (e.g., like the 0x80..0
region which is now used for huge pages).

	--david