[Bug tree-optimization/20501] gcc.dg/vect/vect-93.c fails on ia64-hpux

dorit at il dot ibm dot com gcc-bugzilla@gcc.gnu.org
Sun Mar 20 12:26:00 GMT 2005


------- Additional Comments From dorit at il dot ibm dot com  2005-03-20 12:26 -------
Thanks.

This patch should fix the problem (the message "Alignment of access forced 
using peeling" should not be printed when we're not going to vectorize the loop 
due to unsupported unaligned access. this patch moves the printing of this 
message to after the check that the alignment of all accesses is supported):

Index: tree-vect-analyze.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/tree-vect-analyze.c,v
retrieving revision 2.13
diff -c -3 -p -r2.13 tree-vect-analyze.c
*** tree-vect-analyze.c 17 Mar 2005 21:08:06 -0000      2.13
--- tree-vect-analyze.c 20 Mar 2005 12:17:39 -0000
*************** vect_enhance_data_refs_alignment (loop_v
*** 1183,1190 ****
        }

        DR_MISALIGNMENT (dr0) = 0;
-       if (vect_print_dump_info (REPORT_ALIGNMENT, LOOP_LOC (loop_vinfo)))
-       fprintf (vect_dump, "Alignment of access forced using peeling.");
      }
  }

--- 1183,1188 ----
*************** vect_analyze_data_refs_alignment (loop_v
*** 1260,1265 ****
--- 1258,1266 ----
          && (vect_print_dump_info (REPORT_ALIGNMENT, LOOP_LOC (loop_vinfo))))
        fprintf (vect_dump, "Vectorizing an unaligned access.");
      }
+   if (LOOP_VINFO_UNALIGNED_DR (loop_vinfo)
+       && vect_print_dump_info (REPORT_ALIGNMENT, LOOP_LOC (loop_vinfo)))
+     fprintf (vect_dump, "Alignment of access forced using peeling.");

    return true;
  }




-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20501



More information about the Gcc-bugs mailing list