f771 assertion failure

Jeffrey A Law law@cygnus.com
Sun Sep 28 00:27:00 GMT 1997


  In message < 199709280711.AAA26493@dot.cygnus.com >you write:
  > The problem: ffebad_index_ and ffebad_places_ are both QImode
  > globals and are adjacent.  GCC is not recognizing that the
  > ldq_u/mskbl/stq_u sequence effectively aliases both varibles.
  > 
  > I thought this problem was fixed last year, which is the 
  > confusing part.  Although I just remembered I'm using haifa
  > this time, I thought all that was handled in alias.c.
Are you sure you have the latest alias.c?  I thought Jim checked
in a fix for this recently:

Wed Sep  3 10:39:42 1997  Jim Wilson  <wilson@cygnus.com>

        * alias.c (true_dependence): Address with AND can alias scalars.
        (anti_dependence, output_dependence): Likewise.


Jeff



More information about the Gcc mailing list