Remaining generic bits from sparc64-linux merge

[Jeffrey A Law]

  In message <199710021950.PAA04406@jenolan.rutgers.edu>you write:
  > 
  > The following have still not been applied.  This is fine, but I would
  > like some feedback on whether the maintainers have some reservations
  > about applying these changes, and if so what the specific problems
  > are.
I'm making another pass over this stuff....


  > Thu Oct  2 15:28:24 1997  David S. Miller  <davem@tanya.rutgers.edu>
  > 
  > 	* varasm.c (immed_double_const): Add casts to HOST_WIDE_INT where
  > 	necessary.
  > 	(const_hash): Hash val is unsigned long.
  > 	(SYMHASH): Likewise.
I installed this.


  > 	* tree.c (TYPE_HASH): Type of hash val is unsigned long.
This one too.


  > 	* rtl.c: Remove atol extern, include stdlib.h instead.
  > 	(read_rtx): Use atol/atoq based upon disposition of
  > 	HOST_WIDE_INT.
Not yet, we need to take another look at this one.

  > 	* genattrtab.c (write_test_expr): Use proper format string based
  > 	upon disposition of HOST_BITS_PER_WIDE_INT.
  > 	* genemit.c (gen_exp): Likewise.
  > 	* genpeep.c (match_rtx): Likewise.
  > 	* genrecog.c (write_tree_1): Likewise.
Similarly.


  > 	* print-tree.c (print_node_brief): HOST_PTR_PRINTF format wants a
  > 	char pointer, not HOST_WIDE_INT.
  > 	(print_node): Likewise.  Also hash is unsigned long not
  > 	HOST_WIDE_INT.
I installed this.


  > 	* gbl-ctors.h: Do not declare extern for atexit if both WINNT and
  > 	NEED_ATEXIT are not defined.
Not yet.

  > 	* cse.c (canon_hash): Hash is unsigned long not HOST_WIDE_INT.
I installed this.

  > 	* c-lex.c (yyprint): Use proper format string based upon
  > 	disposition of HOST_BITS_PER_WIDE_INT.
  > 	(yylex): Put casts in right place for args to build_int_2.
Not yet.