From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 4616 invoked by alias); 20 Jul 2009 00:01:35 -0000 Received: (qmail 30535 invoked by uid 48); 20 Jul 2009 00:01:19 -0000 Date: Mon, 20 Jul 2009 00:01:00 -0000 Message-ID: <20090720000119.30534.qmail@sourceware.org> X-Bugzilla-Reason: CC References: Subject: [Bug rtl-optimization/40797] [4.5 Regression] ICE in df_refs_verify, at df-scan.c:4361 In-Reply-To: Reply-To: gcc-bugzilla@gcc.gnu.org To: gcc-bugs@gcc.gnu.org From: "kkojima at gcc dot gnu dot org" Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org X-SW-Source: 2009-07/txt/msg01583.txt.bz2 ------- Comment #4 from kkojima at gcc dot gnu dot org 2009-07-20 00:01 ------- It started to fail after revision 142809: r142809 | zadeck | 2008-12-18 22:38:39 +0900 (Thu, 18 Dec 2008) | 47 lines 2008-12-18 Kenneth Zadeck PR rtl-optimization/37922 * dse.c (bb_info): Added regs_live field. (look_for_hardregs): New function. (replace_read): Added regs_live parameter and code to check that shift sequence does not clobber live hardregs. (check_mem_read_rtx): Added parameter to replace_read. (dse_step1): Added regs_live bitmap and initialize it. (rest_of_handle_dse): Added DF_NOTES problem and earlier call to df_analyze. ... which adds df_analyze to dse.c. Since 142809 is 4.4.0 and the current 4.4.1 doesn't fail, I'll run another binary search on 4.4 branch to see where the ICE went away. BTW, I've found that with --enable-checking=df, all 4.3/4.4/4.5 SH compilers I tried fail for the test case with the similar ICE in df_refs_verify, though they fail in different passes. Oops. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40797