Re: RFC: add inheritance support to nonoverlapping_component_refs_p

[Dan Nicolaescu <dann at godzilla dot ICS dot UCI dot EDU>]

Richard Henderson <rth@redhat.com> writes:

  > On Wed, May 29, 2002 at 04:24:43PM -0700, Dan Nicolaescu wrote:
  > > Another failing test, global pointer vars that point to objects that
  > > have "char" fields are put in alias set 0 (this might be OK though).
  > 
  > I wouldn't expect alias sets to have anything to do with this.
  > In fact, all these tests should be run at -O1 so that we don't
  > use alias sets.
  > 
  > We're trying to test alias resolution based on location within
  > a larger object, not whether types of the accessed objects differ.

OK, I think I know what's going on. 

Here is a new set of testcases, in C this time, no reason to be in C++.

Attachment: alias-global.c
Description: alias-global.c

When compiling fail_c and fail_i 
nonoverlapping_memrefs_p (x, y) returns false for

x = (mem/s/j:QI (plus:SI (reg/v/f:SI 107)
        (const_int 8 [0x8])) [0 <variable>.f1+0 S1 A64])
                                ^^^^^^^^^^^^
                                this is probably a problem
y = (mem/f:SI (lo_sum:SI (reg/f:SI 110)
        (symbol_ref:SI ("fs1"))) [0 fs1+0 S4 A32])

because decl_for_component_ref (exprx) is NULL. 

so true_dependence returns 1. 

If fail_i is compiled with -fstrict-aliasing true_dependence for the
above mems returns false because of the DIFFERENT_ALIAS_SETS_P test. 

If fail_c is compiled with -fstrict-aliasing true_dependence for the
above mems returns true because fs1.f1 is in alias set zero, so 
DIFFERENT_ALIAS_SETS_P can't return true as in fail_i.