[Bug regression/48163] New: [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[Bug regression/48163] New: [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48163

           Summary: [4.7 Regression]: ICEs for cris-elf, like
                    gcc.c-torture/compile/calls.c
                    gcc.c-torture/execute/complex-1.c
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: regression
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hp@gcc.gnu.org
                CC: jakub@gcc.gnu.org
              Host: x86_64-unknown-linux-gnu
            Target: cris-axis-elf


With revision 171027 these tests passed.
>From revision 171042 and on, these tests have failed as follows:

Running
/tmp/hpautotest-gcc1/gcc/gcc/testsuite/gcc.c-torture/compile/compile.exp ...
FAIL: gcc.c-torture/compile/calls.c  -O3 -g  (internal compiler error)
FAIL: gcc.c-torture/compile/calls.c  -O3 -g  (test for excess errors)
FAIL: gcc.c-torture/compile/pr21840.c  -O3 -g  (internal compiler error)
FAIL: gcc.c-torture/compile/pr21840.c  -O3 -g  (test for excess errors)
...
FAIL: gcc.c-torture/execute/complex-1.c compilation,  -O3 -g  (internal
compiler error)
...
Running /tmp/hpautotest-gcc1/gcc/gcc/testsuite/gcc.dg/torture/dg-torture.exp
...
FAIL: gcc.dg/torture/builtin-cproj-1.c  -O3 -g  (internal compiler error)
FAIL: gcc.dg/torture/builtin-cproj-1.c  -O3 -g  (test for excess errors)
FAIL: gcc.dg/torture/pr24626-1.c  -O3 -g  (internal compiler error)
FAIL: gcc.dg/torture/pr24626-1.c  -O3 -g  (test for excess errors)

With the messages in the logfile being two different:

Executing on host: /tmp/hpautotest-gcc1/cris-elf/gccobj/gcc/xgcc
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/gcc/   -O3 -g  -w -c   -isystem
/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./newlib/targ-include -isystem
/tmp/hpautotest-gcc1/gcc/newlib/libc/include  -o calls.o
/tmp/hpautotest-gcc1/gcc/gcc/testsuite/gcc.c-torture/compile/calls.c   
(timeout = 300)
/tmp/hpautotest-gcc1/gcc/gcc/testsuite/gcc.c-torture/compile/calls.c: In
function 'f3':

/tmp/hpautotest-gcc1/gcc/gcc/testsuite/gcc.c-torture/compile/calls.c:13:1:
internal compiler error: in wrap_constant, at cselib.c:898


and

Executing on host: /tmp/hpautotest-gcc1/cris-elf/gccobj/gcc/xgcc
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/gcc/
/tmp/hpautotest-gcc1/gcc/gcc/testsuite/gcc.c-torture/execute/complex-1.c  -w 
-O3 -g    -isystem
/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./newlib/targ-include -isystem
/tmp/hpautotest-gcc1/gcc/newlib/libc/include
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./libgloss/cris/
-L/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./libgloss/cris
-L/tmp/hpautotest-gcc1/gcc/libgloss/cris 
-B/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./newlib/
-L/tmp/hpautotest-gcc1/cris-elf/gccobj/cris-elf/./newlib -sim3  -lm   -o
/tmp/hpautotest-gcc1/cris-elf/gccobj/gcc/testsuite/gcc/complex-1.x4    (timeout
= 300)
/tmp/hpautotest-gcc1/gcc/gcc/testsuite/gcc.c-torture/execute/complex-1.c: In
function 'xcexp':

/tmp/hpautotest-gcc1/gcc/gcc/testsuite/gcc.c-torture/execute/complex-1.c:20:1:
internal compiler error: in expand_debug_expr, at cfgexpand.c:2552


Author of one or more suspect patches in revision range CC:ed.

[Bug regression/48163] [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48163

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.7.0

[Bug debug/48163] [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48163

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2011.03.17 12:04:00
         AssignedTo|unassigned at gcc dot       |jakub at gcc dot gnu.org
                   |gnu.org                     |
     Ever Confirmed|0                           |1

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-03-17 12:04:00 UTC ---
Created attachment 23689
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23689
gcc47-pr48163-1.patch

Fix for the expand_debug_expr issue.  I was very much surprised that for
data.passed_pointer parameters (those aren't DECL_BY_REFERENCE)
DECL_INCOMING_RTL doesn't reflect where it actually is passed and there is no
way to find the fact that it is pass_by_reference (except for repeating all the
cumulative args/etc. stuff done by function.c).

Either we could use some flag on the PARM_DECL to say this, but I think it is
much better to just adjust DECL_INCOMING_RTL in that case.  With the additional
dwarf2out.c change, e.g. on powerpc-linux (32-bit) we are now able to emit
locations for:
struct S { char buf[16]; };

void foo (struct S a, int b, int c, int d, int e, int f, int g, int h, struct S
i)
{
}
even for a and i parameters, which is something we weren't able to do before.

[Bug debug/48163] [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48163

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-03-17 12:07:03 UTC ---
Created attachment 23690
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23690
gcc47-pr48163-2.patch

Patch to fix the wrap_constant ICE.  In the (rare) case where a constant is
casted to a function pointer and called the constant is VOIDmode, which doesn't
work too well with cselib_lookup.  Furthermore, there is no point to cselib
looking the constant up, we can just add the constant as is.

[Bug debug/48163] [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48163

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-03-17 20:34:00 UTC ---
Author: jakub
Date: Thu Mar 17 20:33:54 2011
New Revision: 171113

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=171113
Log:
    PR debug/48163
    * function.c (assign_parms): For data.passed_pointer parms
    use MEM of data.entry_parm instead of data.entry_parm itself
    as DECL_INCOMING_RTL.
    * dwarf2out.c (rtl_for_decl_location): Use DECL_INCOMING_RTL
    also when passed and declared mode is the same, DECL_RTL
    is a MEM with pseudo as address and DECL_INCOMING_RTL is
    a MEM too.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/dwarf2out.c
    trunk/gcc/function.c

[Bug debug/48163] [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48163

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-03-17 20:35:02 UTC ---
Author: jakub
Date: Thu Mar 17 20:34:59 2011
New Revision: 171114

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=171114
Log:
    PR debug/48163
    * var-tracking.c (prepare_call_arguments): If CALL target
    is a non-SYMBOL_REF CONSTANT_P, just add that into the list as
    pc instead of looking it up using cselib_lookup and use
    Pmode for it if x has VOIDmode.
    * dwarf2out.c (gen_subprogram_die): If also both first and
    second CONCAT arguments are VOIDmode, use mode of CONCAT itself.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/dwarf2out.c
    trunk/gcc/var-tracking.c

[Bug debug/48163] [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48163

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-03-17 20:41:02 UTC ---
I hope both issues should be fixed now, can you please retest on cris-elf? 
Thanks.

[Bug debug/48163] [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48163

Hans-Peter Nilsson <hp at gcc dot gnu.org> changed:

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

--- Comment #6 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2011-03-18 02:44:04 UTC ---
(In reply to comment #5)
> I hope both issues should be fixed now, can you please retest on cris-elf? 

Back to no regressions, last tested at r171118.  Thanks!
Closing.

[Bug debug/48163] [4.7 Regression]: ICEs for cris-elf, like gcc.c-torture/compile/calls.c gcc.c-torture/execute/complex-1.c

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48163

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

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

--- Comment #7 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
vt_add_function_parameter needs to be adjusted after the assign_parms change:

  if (!vt_get_decl_and_offset (incoming, &decl, &offset))
    {
      if (REG_P (incoming) || MEM_P (incoming))
    {
      /* This means argument is passed by invisible reference.  */
      offset = 0;
      decl = parm;
      incoming = gen_rtx_MEM (GET_MODE (decl_rtl), incoming);
    }
      else
    {
      if (!vt_get_decl_and_offset (decl_rtl, &decl, &offset))
        return;
      offset += byte_lowpart_offset (GET_MODE (incoming),
                     GET_MODE (decl_rtl));
    }
    }

This generates MEM of MEM incoming locations.