This is the mail archive of the gcc-patches@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: [tree-ssa] Removing redundant loads


In message <wvlisrdh9tn.fsf@prospero.boston.redhat.com>, Jason Merrill writes:
 >Well, don't keep me in suspense.  :)
We didn't call it after simplifying a rather hairy BIT_FIELD_REF which had
side effects (via a statement expression if I recall) in one of its operands.

The easiest way to find these is to change avail_expr_p to use
operand_equal_p and wait for the segfaults when we are unable to remove
expressions from the hash table when bootstrapping because expressions
with side effects don't compare equal to themselves.

And, yes, this points to a bigger problem -- namely that expressions with
side effects are ending up in the avail_exprs hash table.  I've got a
fix for that (as well as the BIT_FIELD_REF issue noted above) bootstrapping
right now.

jeff


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