------- Additional Comments From rakdver at atrey dot karlin dot mff dot cuni dot cz 2004-11-12 14:46 ------- Subject: Re: Code for arrays and pointers are not the same
> > Patch here which should fix not pulling the load of q out of the loop: > > http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00957.html > > > > Then the only thing left is for IV-OPTS to be fixed. > > no -- once this is done, ivopts work just fine, producing the same code > as for array access. btw. the patch above might be wrong -- I think ssa names also pass the is_gimple_addressable test: is_gimple_addressable -> is_gimple_id -> is_gimple_variable -> SSA_NAME. So the correct patch would be Index: tree-ssa-loop-im.c =================================================================== RCS file: /cvs/gcc/gcc/gcc/tree-ssa-loop-im.c,v retrieving revision 2.21 diff -c -3 -p -r2.21 tree-ssa-loop-im.c *** tree-ssa-loop-im.c 8 Nov 2004 13:54:39 -0000 2.21 --- tree-ssa-loop-im.c 12 Nov 2004 11:29:40 -0000 *************** stmt_cost (tree stmt) *** 367,373 **** /* Hoisting memory references out should almost surely be a win. */ if (!is_gimple_variable (lhs)) cost += 20; ! if (is_gimple_addressable (rhs) && !is_gimple_variable (rhs)) cost += 20; switch (TREE_CODE (rhs)) --- 367,373 ---- /* Hoisting memory references out should almost surely be a win. */ if (!is_gimple_variable (lhs)) cost += 20; ! if (is_gimple_addressable (rhs) && TREE_CODE (rhs) != SSA_NAME) cost += 20; switch (TREE_CODE (rhs)) -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18431
