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]
Other format: [Raw text]

RE: Redundant Load Elimination in GCSE


>What you have described is a different problem.
>I agree that the functionality of rtx_equal_for_memref_p should be
>improved,
>However, we can still use its current ability to find (even in only part of
>the cases) two pseudos that point to the same address in GCSE.
>And tell (only in part of the cases) that two memory references are
>equivalent.


True, for this rtx_equal_for_memref_p certainly seems better than
expr_equiv_p 
as it uses the base objects info (reg_known_value) of the two pseudos.

>BTW, does CSE uses rtx_equal_for_memref_p?

I am not aware of any usage of rtx_equal_for_memref_p in CSE or GCSE.

--Sanjiv
 


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