[Bug lto/63894] New: [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[Bug lto/63894] New: [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63894

            Bug ID: 63894
           Summary: [5 Regression] ICE in inline_call, at
                    ipa-inline-transform.c:339 building Firefox
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: lto
          Assignee: unassigned at gcc dot gnu.org
          Reporter: trippels at gcc dot gnu.org
                CC: hubicka at gcc dot gnu.org

Executing: ../../../../gcc_test/usr/local/bin/c++ -o js -Wall
-Wignored-qualifiers -Wsign-compare -Wtype-limits -Werror=char-subscripts
-Werror=comment -Werror=empty-body -Werror=endif-labels
-Werror=int-to-pointer-cast -Werror=missing-braces -Werror=overloaded-virtual
-Werror=pointer-arith -Werror=reorder -Werror=return-type
-Werror=sequence-point -Werror=switch -Werror=trigraphs -Werror=unknown-pragmas
-Werror=unused-label -Werror=unused-value -Werror=write-strings
-Werror=conversion-null -Wno-invalid-offsetof -Wcast-align -flto=64 --param
lto-partitions=64 -mcpu=power7 -fno-semantic-interposition -ffunction-sections
-fdata-sections -fno-rtti -fno-exceptions -fno-math-errno -std=gnu++0x -pthread
-pipe -DNDEBUG -DTRIMMED -O3 -fomit-frame-pointer
/home/trippels/moz-build-dir/js/src/shell/tmpKUuVPB.list -lpthread
-Wl,--hash-style=gnu,--as-needed,--gc-sections,--icf=all -Wl,-z,noexecstack
-Wl,-z,text -Wl,--build-id -Wl,-rpath-link,../../../dist/bin
-Wl,-rpath-link,/home/trippels/moz-build-dir/dist/lib ../libjs_static.a
-rdynamic ../../../nsprpub/lib/ds/libplds4.so
../../../nsprpub/lib/libc/src/libplc4.so ../../../nsprpub/pr/src/libnspr4.so
-lm -ldl -lffi -lz -lm -ldl
/home/trippels/moz-build-dir/js/src/shell/tmpKUuVPB.list:
    INPUT("OSObject.o")
    INPUT("js.o")
    INPUT("jsheaptools.o")
    INPUT("jsoptparse.o")
    INPUT("../editline/editline.o")
    INPUT("../editline/sysunix.o")
    INPUT("../../../mozglue/build/dummy.o")
    INPUT("../../../mfbt/Compression.o")
    INPUT("../../../mfbt/Decimal.o")
    INPUT("../../../mfbt/FloatingPoint.o")
    INPUT("../../../mfbt/HashFunctions.o")
    INPUT("../../../mfbt/JSONWriter.o")
    INPUT("../../../mfbt/Poison.o")
    INPUT("../../../mfbt/SHA1.o")
    INPUT("../../../mfbt/TaggedAnonymousMemory.o")
    INPUT("../../../mfbt/bignum-dtoa.o")
    INPUT("../../../mfbt/bignum.o")
    INPUT("../../../mfbt/cached-powers.o")
    INPUT("../../../mfbt/diy-fp.o")
    INPUT("../../../mfbt/double-conversion.o")
    INPUT("../../../mfbt/fast-dtoa.o")
    INPUT("../../../mfbt/fixed-dtoa.o")
    INPUT("../../../mfbt/strtod.o")
    INPUT("../../../mfbt/unused.o")

lto1: internal compiler error: in inline_call, at ipa-inline-transform.c:339
0x10d422df inline_call(cgraph_edge*, bool, vec<cgraph_edge*, va_heap, vl_ptr>*,
int*, bool, bool*)
        ../../gcc/gcc/ipa-inline-transform.c:334
0x10d3b9d7 inline_small_functions
        ../../gcc/gcc/ipa-inline.c:1827
0x10d3b9d7 ipa_inline
        ../../gcc/gcc/ipa-inline.c:2198
0x10d3b9d7 execute
        ../../gcc/gcc/ipa-inline.c:2568
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.
lto-wrapper: fatal error: ../../../../gcc_test/usr/local/bin/c++ returned 1
exit status
compilation terminated.
/home/trippels/bin/ld: fatal error: lto-wrapper failed
collect2: error: ld returned 1 exit status
make[1]: *** [js] Error 1

adding -r -nostdlib to the invocation triggers another ICE:

lto1: internal compiler error: in update_visibility_by_resolution_info, at
ipa-visibility.c:427

(gdb) l
422              next != node; next = next->same_comdat_group)
423           gcc_assert (!next->externally_visible
424                       || define == (next->resolution ==
LDPR_PREVAILING_DEF_IRONLY
425                                     || next->resolution ==
LDPR_PREVAILING_DEF
426                                     || next->resolution == LDPR_UNDEF
427                                     || next->resolution ==
LDPR_PREVAILING_DEF_IRONLY_EXP));
428
429       if (node->same_comdat_group)
430         for (symtab_node *next = node->same_comdat_group;
431              next != node; next = next->same_comdat_group)
(gdb) p next->externally_visible
$1 = 1
(gdb) p define
$2 = true
(gdb) p next->resolution
$3 = LDPR_PREVAILING_DEF

I will try to reduce this issue.

[Bug lto/63894] [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[hubicka at ucw dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63894

--- Comment #1 from Jan Hubicka <hubicka at ucw dot cz> ---
> (gdb) p next->externally_visible
> $1 = 1
> (gdb) p define
> $2 = true
> (gdb) p next->resolution
> $3 = LDPR_PREVAILING_DEF
> 
> I will try to reduce this issue.

Thanks, the inline-transform ICE is result of Martin's ipa-cp devirtualization
changes.  There are quite few details that needs to be solved correctly, I have
some cleanups on queue to test and commit today.  Hopefully these will care
this (it also reproduces for me, the visibility ICE doesn't, but my tree is
old).
I also get ICE preprocessing one .S file.

Honza

[Bug lto/63894] [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63894

--- Comment #2 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
Created attachment 33993
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=33993&action=edit
testcase

Here's a reduced testcase for ppc64:

trippels@gcc2-power8 % c++ -r -nostdlib -w -flto -std=gnu++0x -O3 js.ii Ion.ii
LoopUnroller.ii Lowering.ii jsapi.ii HelperThreads.ii
lto1: internal compiler error: in inline_call, at ipa-inline-transform.c:339
0x10d4911f inline_call(cgraph_edge*, bool, vec<cgraph_edge*, va_heap, vl_ptr>*,
int*, bool, bool*)
        ../../gcc/gcc/ipa-inline-transform.c:334
0x10d42817 inline_small_functions
        ../../gcc/gcc/ipa-inline.c:1827
0x10d42817 ipa_inline
        ../../gcc/gcc/ipa-inline.c:2198
0x10d42817 execute
        ../../gcc/gcc/ipa-inline.c:2568
Please submit a full bug report,
with preprocessed source if appropriate.

[Bug lto/63894] [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63894

Markus Trippelsdorf <trippels at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-11-16
                 CC|                            |jamborm at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #3 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
The issue is target independent and also happens without -flto:

trippels@gcc2-power8 % cat Lowering.ii
struct A
{
  void *operator new(unsigned long, int);
};
class C
{
public:
  C (int);
};
class D
{
public:
  enum Type
  {
    BOX
  };
  D (int, Type, C);
};
class F
{
public:
  virtual void m_fn1 (int, D);
};
class G : public F, public A
{
};
class K : public G
{
public:
  K (C, D);
};
class J
{
  D m_fn2 (int);
  bool m_fn3 (G *);
};
bool
J::m_fn3 (G *p1)
{
  p1->m_fn1 (0, D (0, D::BOX, 0));
  K *d = new (0) K (0, m_fn2 (0));
  m_fn3 (d);
}

trippels@gcc2-power8 % c++ -c -O3 Lowering.ii
Lowering.ii:43:1: internal compiler error: in inline_call, at
ipa-inline-transform.c:339

[Bug lto/63894] [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63894

Markus Trippelsdorf <trippels at gcc dot gnu.org> changed:

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

[Bug lto/63894] [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63894

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
           Priority|P3                          |P1

[Bug lto/63894] [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63894

--- Comment #4 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
Fixed by r217634. I will add the testcase and then close this bug.

[Bug lto/63894] [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63894

--- Comment #5 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
Author: trippels
Date: Mon Nov 17 09:21:34 2014
New Revision: 217648

URL: https://gcc.gnu.org/viewcvs?rev=217648&root=gcc&view=rev
Log:
Add testcase for PR 63894

2014-11-17  Markus Trippelsdorf  <markus@trippelsdorf.de>

  PR ipa/63894
  * g++.dg/ipa/pr63894.C: New test.

Added:
    trunk/gcc/testsuite/g++.dg/ipa/pr63894.C
Modified:
    trunk/gcc/testsuite/ChangeLog

[Bug ipa/63894] [5 Regression] ICE in inline_call, at ipa-inline-transform.c:339 building Firefox

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63894

Markus Trippelsdorf <trippels at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dcb314 at hotmail dot com

--- Comment #7 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
*** Bug 63912 has been marked as a duplicate of this bug. ***