[Bug tree-optimization/51000] [4.7 Regression] ICE: in vect_get_store_cost, at tree-vect-stmts.c:923 on powerpc-apple-darwin9

[irar at il dot ibm.com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51000

Ira Rosen <irar at il dot ibm.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P1                          |P3
             Status|ASSIGNED                    |NEW
   Target Milestone|4.7.0                       |---

--- Comment #3 from Ira Rosen <irar at il dot ibm.com> 2011-11-09 13:46:41 UTC ---
The patch looks ok, but if I change 

Index: gfortran.dg/alloc_comp_assign_2.f90
===================================================================
--- gfortran.dg/alloc_comp_assign_2.f90 (revision 181190)
+++ gfortran.dg/alloc_comp_assign_2.f90 (working copy)
@@ -26,7 +26,7 @@

   y(1) = b ((/x(1),x(2)/))
   y(2) = b ((/x(2),x(1)/))
-  forall (k = 1:2, j=1:2, i = 1:4, y(k)%at(j)%i(i) <= 10)
+  forall (k = 1:2, j=1:2, i = 1:16, y(k)%at(j)%i(i) <= 10)
     y(k)%at(j)%i(i) =  j*4-i+k
   end forall
   if (any ((/(((y(k)%at(i)%i(j), j = 1,4), i = 1,2), k = 1,2)/) .ne. &

making niters sufficient for vectorization, it fails in vect_remove_stores with

alloc_comp_assign_2.f90: In function 'MAIN__':
alloc_comp_assign_2.f90:7:0: internal compiler error: in gsi_for_stmt, at
gimple-iterator.c:560

trying to get gsi for 
VIEW_CONVERT_EXPR<<unnamed-unsigned:8>>(temp.41[mi.42_1293]) = patt.353_2093;


I don't think you can reproduce the exact same behavior on x86, because it
supports misaligned stores.