[Bug middle-end/64498] New: [5 Regression] Cannot build Firefox with LTO: ICE in substitute_and_fold_dom_walker::before_dom_children

[Bug middle-end/64498] New: [5 Regression] Cannot build Firefox with LTO: ICE in substitute_and_fold_dom_walker::before_dom_children

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64498

            Bug ID: 64498
           Summary: [5 Regression] Cannot build Firefox with LTO: ICE in
                    substitute_and_fold_dom_walker::before_dom_children
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
          Assignee: unassigned at gcc dot gnu.org
          Reporter: marxin at gcc dot gnu.org
                CC: markus at trippelsdorf dot net

Hello.

With gcc version 5.0.0 20150105 (experimental) (GCC) and latest master branch
of Firefox, compiled with -O2 and -flto, I see following problem in a ltrans:


/home/marxin/Programming/gecko-dev/js/src/builtin/Intl.cpp: In function
‘NewUDateFormat’:
/home/marxin/Programming/gecko-dev/js/src/builtin/Intl.cpp:1840:0: internal
compiler error: Segmentation fault
 NewUDateFormat(JSContext *cx, HandleObject dateTimeFormat)
 ^
0x8de73f crash_signal
    ../../gcc/toplev.c:359
0x98feab gimple_code
    ../../gcc/gimple.h:1545
0x98feab gimple_nop_p
    ../../gcc/gimple.h:5589
0x98feab get_default_value
    ../../gcc/tree-ssa-ccp.c:291
0x98feab get_value
    ../../gcc/tree-ssa-ccp.c:365
0x98feab get_constant_value
    ../../gcc/tree-ssa-ccp.c:384
0xa0556d replace_uses_in
    ../../gcc/tree-ssa-propagate.c:948
0xa0556d substitute_and_fold_dom_walker::before_dom_children(basic_block_def*)
    ../../gcc/tree-ssa-propagate.c:1151
0xdf4022 dom_walker::walk(basic_block_def*)
    ../../gcc/domwalk.c:188
0xa05030 substitute_and_fold(tree_node* (*)(tree_node*), bool
(*)(gimple_stmt_iterator*), bool)
    ../../gcc/tree-ssa-propagate.c:1230
0x990cb0 ccp_finalize
    ../../gcc/tree-ssa-ccp.c:934
0x990cb0 do_ssa_ccp
    ../../gcc/tree-ssa-ccp.c:2374
0x990cb0 execute
    ../../gcc/tree-ssa-ccp.c:2406

Where:

#0  0x0000000000bb4950 in gimple_code (g=0x0) at ../../gcc/gimple.h:1545
#1  0x0000000000bb52cc in gimple_nop_p (g=0x0) at ../../gcc/gimple.h:5589
#2  0x0000000000bb6350 in get_default_value (var=0x7fffe5d24870) at
../../gcc/tree-ssa-ccp.c:291
#3  0x0000000000bb6620 in get_value (var=0x7fffe5d24870) at
../../gcc/tree-ssa-ccp.c:365
#4  0x0000000000bb66b7 in get_constant_value (var=0x7fffe5d24870) at
../../gcc/tree-ssa-ccp.c:384
#5  0x0000000000c48d5e in replace_uses_in (stmt=0x7fffe5d116c0,
get_value=0xbb666e <get_constant_value(tree)>) at
../../gcc/tree-ssa-propagate.c:948
#6  0x0000000000c49484 in substitute_and_fold_dom_walker::before_dom_children
(this=0x7fffffffd430, bb=0x7fffe5d02750) at ../../gcc/tree-ssa-propagate.c:1151
#7  0x000000000108d113 in dom_walker::walk (this=0x7fffffffd430,
bb=0x7fffe5d02750) at ../../gcc/domwalk.c:188
#8  0x0000000000c4975a in substitute_and_fold (get_value_fn=0xbb666e
<get_constant_value(tree)>, fold_fn=0xbbc57c
<ccp_fold_stmt(gimple_stmt_iterator*)>, do_dce=true) at
../../gcc/tree-ssa-propagate.c:1230
#9  0x0000000000bb7c20 in ccp_finalize () at ../../gcc/tree-ssa-ccp.c:934
#10 0x0000000000bbd0a2 in do_ssa_ccp () at ../../gcc/tree-ssa-ccp.c:2374
#11 0x0000000000bbd155 in (anonymous namespace)::pass_ccp::execute
(this=0x1b68c40) at ../../gcc/tree-ssa-ccp.c:2406
#12 0x0000000000a1bbb2 in execute_one_pass (pass=0x1b68c40) at
../../gcc/passes.c:2311
#13 0x0000000000a1bdec in execute_pass_list_1 (pass=0x1b68c40) at
../../gcc/passes.c:2363
#14 0x0000000000a1be1d in execute_pass_list_1 (pass=0x1b67ec0) at
../../gcc/passes.c:2364
#15 0x0000000000a1be5a in execute_pass_list (fn=0x7fffe6a37c78, pass=0x1b67e00)
at ../../gcc/passes.c:2374
#16 0x0000000000745a2c in cgraph_node::expand (this=0x7ffff6a64930) at
../../gcc/cgraphunit.c:1798
#17 0x0000000000745ecb in expand_all_functions () at
../../gcc/cgraphunit.c:1934
#18 0x0000000000746938 in symbol_table::compile (this=0x7ffff6c3a000) at
../../gcc/cgraphunit.c:2284
#19 0x00000000006b154c in lto_main () at ../../gcc/lto/lto.c:3446
#20 0x0000000000ac3b8c in compile_file () at ../../gcc/toplev.c:570
#21 0x0000000000ac5f48 in do_compile () at ../../gcc/toplev.c:2018
#22 0x0000000000ac6152 in toplev::main (this=0x7fffffffd7cf, argc=36,
argv=0x7fffffffd8c8) at ../../gcc/toplev.c:2115
#23 0x000000000067afc9 in main (argc=36, argv=0x7fffffffd8c8) at
../../gcc/main.c:38

in #2:
(gdb) call debug_tree(var)
 <ssa_name 0x7fffe5d24870 nothrowdef_stmt 

    version 578 in-free-list>

in #5:
(gdb) call debug_gimple_stmt(stmt)
# DEBUG code => _578

Unfortunately, problem is in a LTRANS and it's hard to reduce a testcase.
Was there any recent change in CCP? If needed, I can provide dumps.

Thanks,
Martin
>From gcc-bugs-return-472226-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Jan 05 11:08:08 2015
Return-Path: <gcc-bugs-return-472226-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 16934 invoked by alias); 5 Jan 2015 11:08:07 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 16855 invoked by uid 48); 5 Jan 2015 11:07:59 -0000
From: "jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug middle-end/64465] [5 Regression] internal compiler error: verify_flow_info failed
Date: Mon, 05 Jan 2015 11:08:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: middle-end
X-Bugzilla-Version: 5.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jakub at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P1
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 5.0
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-64465-4-ugD84VOFnu@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-64465-4@http.gcc.gnu.org/bugzilla/>
References: <bug-64465-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2015-01/txt/msg00220.txt.bz2
Content-length: 448

https://gcc.gnu.org/bugzilla/show_bug.cgi?idd465

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I get the same ICE as H.J., and the bug is that redirect_all_calls
redirects a call to __builtin_unreachable , before it was
  _10 = __open_alias (__path_6(D), __oflag_3(D), __builtin_va_arg_pack ());
and afterwards
  __builtin_unreachable ("/dev/null", 3);
(note, again there is the bug that the extra arguments aren't dropped).

[Bug middle-end/64498] [5 Regression] Cannot build Firefox with LTO: ICE in substitute_and_fold_dom_walker::before_dom_children

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64498

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |5.0

[Bug middle-end/64498] [5 Regression] Cannot build Firefox with LTO: ICE in substitute_and_fold_dom_walker::before_dom_children

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64498

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hubicka at gcc dot gnu.org

--- Comment #1 from Martin Liška <marxin at gcc dot gnu.org> ---
Problem is hidden in flag_var_tracking_assignments, which is set to false.
Thus, MAY_HAVE_DEBUG_INSNS is set to false. Unfortunately, Firefox contains
couple of third party libraries that have -g as a default option. So that DCE
after a statements removal does not remove debug statements that use result of
a statement and verifier fails.

Martin
>From gcc-bugs-return-472745-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Jan 12 09:01:41 2015
Return-Path: <gcc-bugs-return-472745-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 9229 invoked by alias); 12 Jan 2015 09:01:40 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 9203 invoked by uid 89); 12 Jan 2015 09:01:39 -0000
Authentication-Results: sourceware.org; auth=none
X-Virus-Found: No
X-Spam-SWARE-Status: No, score=-2.1 required=5.0 tests=AWL,BAYES_00,T_RP_MATCHES_RCVD autolearn=ham version=3.3.2
X-Spam-User: qpsmtpd, 2 recipients
X-HELO: atrey.karlin.mff.cuni.cz
Received: from atrey.karlin.mff.cuni.cz (HELO atrey.karlin.mff.cuni.cz) (195.113.26.193) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Mon, 12 Jan 2015 09:01:38 +0000
Received: by atrey.karlin.mff.cuni.cz (Postfix, from userid 4018)	id E600D81E78; Mon, 12 Jan 2015 10:01:35 +0100 (CET)
Date: Mon, 12 Jan 2015 09:01:00 -0000
From: Jan Hubicka <hubicka@ucw.cz>
To: "marxin at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
Cc: gcc-bugs@gcc.gnu.org, jakub@redhat.com
Subject: Re: [Bug middle-end/64498] [5 Regression] Cannot build Firefox with LTO: ICE in substitute_and_fold_dom_walker::before_dom_children
Message-ID: <20150112090135.GB16891@atrey.karlin.mff.cuni.cz>
References: <bug-64498-4@http.gcc.gnu.org/bugzilla/> <bug-64498-4-Q8wx2Th2mT@http.gcc.gnu.org/bugzilla/>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <bug-64498-4-Q8wx2Th2mT@http.gcc.gnu.org/bugzilla/>
User-Agent: Mutt/1.5.21 (2010-09-15)
X-IsSubscribed: yes
X-SW-Source: 2015-01/txt/msg00739.txt.bz2
Content-length: 895

> Problem is hidden in flag_var_tracking_assignments, which is set to false.
> Thus, MAY_HAVE_DEBUG_INSNS is set to false. Unfortunately, Firefox contains
> couple of third party libraries that have -g as a default option. So that DCE
> after a statements removal does not remove debug statements that use result of
> a statement and verifier fails.

flag_var_tracking_assignments is marked as Optimization in common.opt while rest
of debug machinery is not Optimization.  This means that we try to preserve assignments
in functions that was compiled with -g at compile time, but then we never consume
it....

Shall we drop flag_var_tracking_assignments to 0 and remove the debug
statements after streaming in in these cases? Also inliner probably needs to
remove the statements when it inlines from function with
flag_var_tracking_assignments to one with !flag_var_tracking_assignments

Honza

[Bug middle-end/64498] [5 Regression] Cannot build Firefox with LTO: ICE in substitute_and_fold_dom_walker::before_dom_children

[hubicka at ucw dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64498

--- Comment #2 from Jan Hubicka <hubicka at ucw dot cz> ---
> Problem is hidden in flag_var_tracking_assignments, which is set to false.
> Thus, MAY_HAVE_DEBUG_INSNS is set to false. Unfortunately, Firefox contains
> couple of third party libraries that have -g as a default option. So that DCE
> after a statements removal does not remove debug statements that use result of
> a statement and verifier fails.

flag_var_tracking_assignments is marked as Optimization in common.opt while
rest
of debug machinery is not Optimization.  This means that we try to preserve
assignments
in functions that was compiled with -g at compile time, but then we never
consume
it....

Shall we drop flag_var_tracking_assignments to 0 and remove the debug
statements after streaming in in these cases? Also inliner probably needs to
remove the statements when it inlines from function with
flag_var_tracking_assignments to one with !flag_var_tracking_assignments

Honza

[Bug middle-end/64498] [5 Regression] Cannot build Firefox with LTO: ICE in substitute_and_fold_dom_walker::before_dom_children

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64498

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #3)
> If the final result is -g0, then clearing flag_var_tracking_assignments is
> desirable, if the final result is -g and any of the inputs have debug stmts,
> then either you need to turn flag_var_tracking_assignments whenever any
> inputs
> has debug stmts, or remove the debug stmts if you want to keep it cleared.

I think this will properly happen when removing flag_var_tracking_assignments
from Optimization.  The streamer drops all debug stmts at stream-in time
according to global flags.

[Bug middle-end/64498] [5 Regression] Cannot build Firefox with LTO: ICE in substitute_and_fold_dom_walker::before_dom_children

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64498

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |DUPLICATE

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Should be fixed now.

*** This bug has been marked as a duplicate of bug 64415 ***

[Bug middle-end/64498] [5 Regression] Cannot build Firefox with LTO: ICE in substitute_and_fold_dom_walker::before_dom_children

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64498

--- Comment #6 from Martin Liška <marxin at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #5)
> Should be fixed now.
> 
> *** This bug has been marked as a duplicate of bug 64415 ***

I can confirm that it works for me.

Thanks,
Martin
>From gcc-bugs-return-473218-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Wed Jan 14 16:18:17 2015
Return-Path: <gcc-bugs-return-473218-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 2545 invoked by alias); 14 Jan 2015 16:18:16 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 2508 invoked by uid 48); 14 Jan 2015 16:18:12 -0000
From: "joel at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug target/64460] ARM ICE on valid code
Date: Wed, 14 Jan 2015 16:18:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: target
X-Bugzilla-Version: 5.0
X-Bugzilla-Keywords: ice-on-valid-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: joel at gcc dot gnu.org
X-Bugzilla-Status: RESOLVED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: ktkachov at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: bug_status resolution
Message-ID: <bug-64460-4-qdUHDX6kji@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-64460-4@http.gcc.gnu.org/bugzilla/>
References: <bug-64460-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2015-01/txt/msg01212.txt.bz2
Content-length: 1164

https://gcc.gnu.org/bugzilla/show_bug.cgi?idd460

Joel Sherrill <joel at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|---                         |FIXED

--- Comment #12 from Joel Sherrill <joel at gcc dot gnu.org> ---
(In reply to ktkachov from comment #11)
> Joel, can you try trunk after this commit to confirm that it fixes the ICE
> on your end and close this off if ok?

Based on today's results, I am closing this one. The following shows my
compiler version and compiling the original test case.

Thanks.

$ arm-rtems4.11-gcc --version
arm-rtems4.11-gcc (GCC) 5.0.0 20150114 (experimental)
Copyright (C) 2015 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

$  ~/test-gcc/install-head/bin/arm-rtems4.11-gcc -mcpu=xscale -Os -c
/tmp/gumstix.c
$  ~/test-gcc/install-head/bin/arm-rtems4.11-gcc -mcpu=xscale -O2 -c
/tmp/gumstix.c