[Bug rtl-optimization/49977] New: [4.7 Regression] CFI notes are missed for delayed slot

[Bug rtl-optimization/49977] New: [4.7 Regression] CFI notes are missed for delayed slot

[kkojima at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

           Summary: [4.7 Regression] CFI notes are missed for delayed slot
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: kkojima@gcc.gnu.org
                CC: rth@gcc.gnu.org, hp@gcc.gnu.org
            Target: sh4-unknown-linux-gnu, cris-elf


Many EH tests fail on SH and CRIS.  These failures went away with
-fno-delayed-branch on SH.  The symptoms are quite similar to those
of PR49686.

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[danglin at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

John David Anglin <danglin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|sh4-unknown-linux-gnu,      |sh4-unknown-linux-gnu,
                   |cris-elf                    |cris-elf,
                   |                            |hppa64-hp-hpux11.11
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.08.04 18:20:05
                 CC|                            |danglin at gcc dot gnu.org
     Ever Confirmed|0                           |1

--- Comment #1 from John David Anglin <danglin at gcc dot gnu.org> 2011-08-04 18:20:05 UTC ---
On hppa64-hp-hpux11.11, I'm seeing the the following g++ eh fails with revision
177282:

FAIL: g++.dg/compat/eh/unexpected1 cp_compat_x_tst.o-cp_compat_y_tst.o execute 
FAIL: g++.dg/eh/crossjump1.C execution test
FAIL: g++.dg/eh/forced1.C execution test
FAIL: g++.dg/eh/uncaught1.C execution test
FAIL: g++.dg/eh/unexpected1.C execution test
FAIL: g++.old-deja/g++.abi/cxa_vec.C execution test
FAIL: g++.old-deja/g++.eh/badalloc1.C execution test
FAIL: g++.old-deja/g++.eh/fntry1.C execution test
FAIL: g++.old-deja/g++.eh/rethrow1.C execution test
FAIL: g++.old-deja/g++.eh/rethrow2.C execution test
FAIL: g++.old-deja/g++.eh/rethrow3.C execution test
FAIL: g++.old-deja/g++.eh/rethrow4.C execution test
FAIL: g++.old-deja/g++.eh/rethrow5.C execution test
FAIL: g++.old-deja/g++.eh/rethrow6.C execution test
FAIL: g++.old-deja/g++.mike/eh23.C execution test
FAIL: g++.old-deja/g++.mike/eh39.C execution test
FAIL: g++.old-deja/g++.mike/eh40.C execution test
FAIL: g++.old-deja/g++.robertl/eb31.C execution test

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[rth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

Richard Henderson <rth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
         AssignedTo|unassigned at gcc dot       |rth at gcc dot gnu.org
                   |gnu.org                     |
   Target Milestone|---                         |4.7.0

--- Comment #2 from Richard Henderson <rth at gcc dot gnu.org> 2011-08-04 19:05:11 UTC ---
Mine.

Kaz, can you enumerate some specific tests that are now failing?
It'll save me time trying to track down the problem.  E.g. a cross
to CRIS is hanging in the simulator for *every* execution test,
so I assume there's something wrong with my cris-sim dejagnu setup.
I havn't tried sh-sim yet, but will do so while looking at hppa...

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[rth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

--- Comment #3 from Richard Henderson <rth at gcc dot gnu.org> 2011-08-04 19:50:03 UTC ---
David, most of those c++ tests require system headers to compile,
but g++.old-deja/g++.robertl/eb31.C does not.

Except that I don't see any changes to unwind info in any delay
slots for that test case, so I'm particularly confused about what
could have gone wrong with that test.  Does it work without delayed
branches?

Or is the problem in fact in libgcc somewhere?  I.e. you have to
compile libgcc with -fno-delayed-branch in order to fix things?

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[dave.anglin at bell dot net]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

--- Comment #4 from dave.anglin at bell dot net 2011-08-04 20:28:54 UTC ---
On 4-Aug-11, at 3:50 PM, rth at gcc dot gnu.org wrote:

> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977
>
> Or is the problem in fact in libgcc somewhere?  I.e. you have to
> compile libgcc with -fno-delayed-branch in order to fix things?


The problem looks to be in libgcc:

Program received signal SIGSEGV, Segmentation fault.
0xc000000000091924 in _Unwind_DeleteException (exc=0x800000010000a090)
     at ../../../gcc/libgcc/../gcc/unwind.inc:271
271        (*exc->exception_cleanup) (_URC_FOREIGN_EXCEPTION_CAUGHT, exc);
(gdb) p exc
$1 = (struct _Unwind_Exception *) 0x800000010000a090
(gdb) p exc->exception_cleanup
$2 = (_Unwind_Exception_Cleanup_Fn) 0x100000001
(gdb) p/x *exc
$4 = {exception_class = 0x474e5543432b2b00, exception_cleanup =  
0x100000001,
   private_1 = 0x0, private_2 = 0x800003fffdff0940}

Dave
--
John David Anglin    dave.anglin@bell.net

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[kkojima at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

--- Comment #5 from Kazumoto Kojima <kkojima at gcc dot gnu.org> 2011-08-04 21:18:55 UTC ---
(In reply to comment #2)
> Kaz, can you enumerate some specific tests that are now failing?

I've got

FAIL: gcc.dg/cleanup-10.c execution test
FAIL: gcc.dg/cleanup-11.c execution test

FAIL: g++.dg/eh/crossjump1.C execution test
FAIL: g++.dg/eh/unexpected1.C execution test
FAIL: g++.dg/ext/cleanup-10.C execution test
FAIL: g++.dg/ext/cleanup-11.C execution test
FAIL: g++.dg/torture/pr49115.C  -O1  execution test
...

A tiny testcase in #1 of PR49686

int foo (int a)
{
  if (a)
    bar ();
  return 1;
}

is again compiled to

foo:
.LFB0:
    tst    r4,r4
    bt/s    .L2
    sts.l    pr,@-r15
    mov.l    .L3,r0
    jsr    @r0
    nop

with -O1 -fexceptions -fnon-call-exceptions.

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

--- Comment #6 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2011-08-04 21:51:07 UTC ---
(In reply to comment #2)
> E.g. a cross
> to CRIS is hanging in the simulator for *every* execution test,
> so I assume there's something wrong with my cris-sim dejagnu setup.

Likely.  Sounds like the usual ungraceful failure of dejagnu to find the
baseboard file: it just tries to execute the file natively and then oddly times
out instead of just failing; note lack of "cris-axis-elf-run" calls in the
<whatever>.log files.  That is, if you're running dejagnu-1.4.4 or earlier;
dejagnu-1.5 should already have the baseboard.  For future reference,
<http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20204#c7>.

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[rth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

Richard Henderson <rth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |WAITING

--- Comment #7 from Richard Henderson <rth at gcc dot gnu.org> 2011-08-05 02:29:40 UTC ---
Proposed patch:

http://gcc.gnu.org/ml/gcc-patches/2011-08/msg00551.html

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[rth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

--- Comment #8 from Richard Henderson <rth at gcc dot gnu.org> 2011-08-05 16:17:20 UTC ---
Author: rth
Date: Fri Aug  5 16:17:13 2011
New Revision: 177465

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=177465
Log:
PR rtl-opt/49977
        * dwarf2cfi.c (scan_insn_after): Split out of ...
        (scan_trace): ... here.  Correctly place notes wrt sequences.

Modified:
    trunk/gcc/ChangeLog

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[rth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

--- Comment #9 from Richard Henderson <rth at gcc dot gnu.org> 2011-08-05 16:17:54 UTC ---
Author: rth
Date: Fri Aug  5 16:17:46 2011
New Revision: 177466

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=177466
Log:
PR rtl-opt/49977
        * dwarf2cfi.c (scan_insn_after): Split out of ...
        (scan_trace): ... here.  Correctly place notes wrt sequences.

Modified:
    trunk/gcc/dwarf2cfi.c

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[rth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

--- Comment #10 from Richard Henderson <rth at gcc dot gnu.org> 2011-08-05 16:18:57 UTC ---
Fixed?  Dave, if this works for you on pa, please close.

[Bug rtl-optimization/49977] [4.7 Regression] CFI notes are missed for delayed slot

[kkojima at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49977

Kazumoto Kojima <kkojima at gcc dot gnu.org> changed:

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

--- Comment #11 from Kazumoto Kojima <kkojima at gcc dot gnu.org> 2011-08-10 02:41:57 UTC ---
Now the testresult for hppa64-hp-hpux11.11 looks good
http://gcc.gnu.org/ml/gcc-testresults/2011-08/msg00952.html

I'd like to close this PR.