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]

dead_or_set_p now valid after reload


I was reviewing some old patches I submitted, and found some leftovers
that I think can be removed.

dead_or_set_p use to not be callable after reload completed due to the
REG_DEAD notes not being accurate.  Meisner checked in work to disable
dead_or_set_p (0 && dead_or_set_p) to the m32r.md port.  Law checked
in work to make them accurate.  Here are the two entries:

Fri May  8 18:23:08 1998  Michael Meissner  <meissner@cygnus.com>

    (peepholes): Disable peepholes that call dead_or_set_p.

Thu Oct 15 23:53:29 1998 Bernd Schmidt <crux@pool.informatik.rwth-aachen.de>
			 Jeffrey A Law  (law@cygnus.com)

Now, if I don't miss my guess, we can reenable the use of
dead_or_set_p in m32r.md.


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