Created attachment 35088 [details] Gimple dump spawn /home/dave/gnu/gcc/objdir/gcc/xgcc -B/home/dave/gnu/gcc/objdir/gcc/ /home/ dave/gnu/gcc/gcc/gcc/testsuite/gcc.dg/pr29215.c -fno-diagnostics-show-caret -fdi agnostics-color=never -O2 -fdump-tree-gimple -S -o pr29215.s PASS: gcc.dg/pr29215.c (test for excess errors) FAIL: gcc.dg/pr29215.c scan-tree-dump-not gimple "memcpy" This is revision 221548. Revision 221505 was okay.
Ok, that's probably caused by 2015-03-20 Richard Biener <rguenther@suse.de> PR middle-end/64715 * tree-chrec.c (chrec_fold_poly_cst): Use useless_type_conversion_p for type comparison and gcc_checking_assert. (chrec_fold_plus_poly_poly): Likewise. (chrec_fold_multiply_poly_poly): Likewise. (chrec_convert_1): Likewise. * gimplify.c (gimplify_expr): Remove premature folding of &X + CST to &MEM[&X, CST]. and should be fixed by adjusting the testcase to scan not the gimple dump but the ccp1 dump. Can you try that?
Created attachment 35111 [details] gcc5-pr65506.patch Indeed, have verified this with the cross-compiler and the attached patch should cure this.
Looks good to me.
On 2015-03-23 10:05 AM, jakub at gcc dot gnu.org wrote: > Indeed, have verified this with the cross-compiler and the attached patch > should cure this. I have applied the patch for testing but hit pr65522 regression. Restarted build without ada. Will take about a day to confirm fix. Dave
Author: jakub Date: Mon Mar 23 20:04:43 2015 New Revision: 221607 URL: https://gcc.gnu.org/viewcvs?rev=221607&root=gcc&view=rev Log: 2015-03-23 Jakub Jelinek <jakub@redhat.com> PR testsuite/65506 * gcc.dg/pr29215.c: Dump and analyze ccp1 dump instead of gimple dump. Modified: trunk/gcc/testsuite/ChangeLog trunk/gcc/testsuite/gcc.dg/pr29215.c
Fixed.
*** Bug 65529 has been marked as a duplicate of this bug. ***
This test was also failing for target arm-none-eabi, also fixed by Jakub's r221607.