[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-24 Thread rguenth at gcc dot gnu.org 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632



--- Comment #8 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-24 
09:26:25 UTC ---

If it's triggered by GC I suggest to try reducing with --param ggc-min-expand=0

--param ggc-min-heapsize=0 (not that this will make it any faster ;))

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-24 Thread zsojka at seznam dot cz 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632



--- Comment #9 from Zdenek Sojka zsojka at seznam dot cz 2012-09-24 09:38:40 
UTC ---

If I remember correctly, with --param ggc-min-expand=0

--param ggc-min-heapsize=0 the program didn't crash - maybe the space was

reallocated of another tree, so the pointer became valid again.



I tried to compile gcc with --disable-bootstrap --enable-checking=valgrind, but

the compilation fails when linking (saying I should use -fPIC). That would give

me the chance to use valgrind to find the invalid memory access (memory

allocations aren't hooked by valgrind here; valgrind checking adds explicit

calls to valgrind routines).



I am not reducing the file any longer, if it's caused by invalid memory

references, the crash could appear and disappear 'randomly' with changes in the

code.

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-24 Thread dimhen at gmail dot com 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632



--- Comment #10 from Dmitry G. Dyachenko dimhen at gmail dot com 2012-09-24 
09:46:12 UTC ---

for me testcase FAIl with -O2 -flto --param ggc-min-heapsize=0

and OK with -O3 -flto --param ggc-min-expand=0

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-24 Thread markus at trippelsdorf dot de 

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632

--- Comment #11 from Markus Trippelsdorf markus at trippelsdorf dot de 
2012-09-24 12:27:28 UTC ---
I get:

==24033== Memcheck, a memory error detector
==24033== Copyright (C) 2002-2012, and GNU GPL'd, by Julian Seward et al.
==24033== Using Valgrind-3.8.1 and LibVEX; rerun with -h for copyright info
==24033== Command:
/var/tmp/foo/usr/local/bin/../libexec/gcc/x86_64-unknown-linux-gnu/4.8.0/cc1plus
-fpreprocessed test.ii -quiet -dumpbase test.ii -mtune=generic -march=x86-64
-auxbase test -O2 -Wfatal-errors -w -std=c++11 -flto -flto-partition=none
--param ggc-min-heapsize=0 -o /tmp/ccj0VbuA.s
==24033== 
==24033== Invalid read of size 2
==24033==at 0x947E70: lto_output_tree(output_block*, tree_node*, bool,
bool) (lto-streamer-out.c:127)
==24033==by 0xB8D64F: streamer_write_tree_body(output_block*, tree_node*,
bool) (tree-streamer-out.c:672)
==24033==by 0x948569: lto_output_tree(output_block*, tree_node*, bool,
bool) (lto-streamer-out.c:339)
==24033==by 0xB8CDE3: streamer_write_tree_body(output_block*, tree_node*,
bool) (tree-streamer-out.c:507)
==24033==by 0x948569: lto_output_tree(output_block*, tree_node*, bool,
bool) (lto-streamer-out.c:339)
==24033==by 0x948CAE: lto_output() (lto-streamer-out.c:751)
==24033==by 0x97DBE0: ipa_write_summaries_2(opt_pass*, lto_out_decl_state*)
(passes.c:2284)
==24033==by 0x97EA2D: ipa_write_summaries() (passes.c:2314)
==24033==by 0x782919: compile() (cgraphunit.c:1866)
==24033==by 0x782B54: finalize_compilation_unit() (cgraphunit.c:2080)
==24033==by 0x5AE5FA: cp_write_global_declarations() (decl2.c:4024)
==24033==by 0xA1D784: compile_file() (toplev.c:560)
==24033==  Address 0x8c46af0 is not stack'd, malloc'd or (recently) free'd
==24033== 
==24033== Invalid read of size 2
==24033==at 0x947E99: lto_output_tree(output_block*, tree_node*, bool,
bool) (tree-streamer.h:63)
==24033==by 0xB8D64F: streamer_write_tree_body(output_block*, tree_node*,
bool) (tree-streamer-out.c:672)
==24033==by 0x948569: lto_output_tree(output_block*, tree_node*, bool,
bool) (lto-streamer-out.c:339)
==24033==by 0xB8CDE3: streamer_write_tree_body(output_block*, tree_node*,
bool) (tree-streamer-out.c:507)
==24033==by 0x948569: lto_output_tree(output_block*, tree_node*, bool,
bool) (lto-streamer-out.c:339)
==24033==by 0x948CAE: lto_output() (lto-streamer-out.c:751)
==24033==by 0x97DBE0: ipa_write_summaries_2(opt_pass*, lto_out_decl_state*)
(passes.c:2284)
==24033==by 0x97EA2D: ipa_write_summaries() (passes.c:2314)
==24033==by 0x782919: compile() (cgraphunit.c:1866)
==24033==by 0x782B54: finalize_compilation_unit() (cgraphunit.c:2080)
==24033==by 0x5AE5FA: cp_write_global_declarations() (decl2.c:4024)
==24033==by 0xA1D784: compile_file() (toplev.c:560)
==24033==  Address 0x8c46af0 is not stack'd, malloc'd or (recently) free'd
==24033== 
test.ii: In member function ‘AirportTileTableIterator::operator++()’:
test.ii:10619:6: internal compiler error: tree code ‘�F���’ is not supported in
LTO streams
  }
  ^
Please submit a full bug report,
with preprocessed source if appropriate.
See http://gcc.gnu.org/bugs.html for instructions.

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-24 Thread rguenth at gcc dot gnu.org 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632



Richard Guenther rguenth at gcc dot gnu.org changed:



   What|Removed |Added



 Status|UNCONFIRMED |ASSIGNED

   Last reconfirmed||2012-09-24

 AssignedTo|unassigned at gcc dot   |rguenth at gcc dot gnu.org

   |gnu.org |

 Ever Confirmed|0   |1



--- Comment #12 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-24 
12:42:27 UTC ---

The backtrace hints at trees containing DECL_VALUE/DEBUG_EXPRs.  We

are indeed writing TREE_BLOCK which has been garbage-collected.



(gdb) up

#3  0x00dcf803 in write_ts_exp_tree_pointers (ob=0x1c809d0, 

expr=0x74de, ref_p=true)

at /space/rguenther/src/svn/trunk/gcc/tree-streamer-out.c:674

674   stream_write_tree (ob, TREE_BLOCK (expr), ref_p);

...

#7  0x00dcea3c in write_ts_decl_common_tree_pointers (ob=0x1c809d0, 

expr=0x74ddc2f8, ref_p=true)

at /space/rguenther/src/svn/trunk/gcc/tree-streamer-out.c:509

509 stream_write_tree (ob, DECL_DEBUG_EXPR (expr), ref_p);

...

#11 0x00ad8e10 in output_struct_function_base (ob=0x1c809d0, 

fn=0x75440c80)

at /space/rguenther/src/svn/trunk/gcc/lto-streamer-out.c:751

748   /* Output all the local variables in the function.  */

749   streamer_write_hwi (ob, VEC_length (tree, fn-local_decls));

750   FOR_EACH_VEC_ELT (tree, fn-local_decls, i, t)

751 stream_write_tree (ob, t, true);



so somehow GC does not mark this block as used (or it is not marked

as used in the block tree and thus removed in remove_unused_scope_block_p).



We indeed forget to walk all locals in clear_unused_block_pointer ().



Testing



Index: gcc/tree-ssa-live.c

===

--- gcc/tree-ssa-live.c (revision 191664)

+++ gcc/tree-ssa-live.c (working copy)

@@ -620,11 +620,6 @@ clear_unused_block_pointer_1 (tree *tp,

   if (EXPR_P (*tp)  TREE_BLOCK (*tp)

!TREE_USED (TREE_BLOCK (*tp)))

 TREE_SET_BLOCK (*tp, NULL);

-  if (TREE_CODE (*tp) == VAR_DECL  DECL_DEBUG_EXPR_IS_FROM (*tp))

-{

-  tree debug_expr = DECL_DEBUG_EXPR (*tp);

-  walk_tree (debug_expr, clear_unused_block_pointer_1, NULL, NULL);

-}

   return NULL_TREE;

 }



@@ -636,6 +631,16 @@ clear_unused_block_pointer ()

 {

   basic_block bb;

   gimple_stmt_iterator gsi;

+  tree t;

+  unsigned i;

+

+  FOR_EACH_VEC_ELT (tree, cfun-local_decls, i, t)

+if (TREE_CODE (t) == VAR_DECL  DECL_DEBUG_EXPR_IS_FROM (t))

+  {

+   tree debug_expr = DECL_DEBUG_EXPR (t);

+   walk_tree (debug_expr, clear_unused_block_pointer_1, NULL, NULL);

+  }

+

   FOR_EACH_BB (bb)

 for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi); gsi_next (gsi))

   {

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-24 Thread rguenth at gcc dot gnu.org 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632



--- Comment #13 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-24 
15:03:01 UTC ---

Author: rguenth

Date: Mon Sep 24 15:02:53 2012

New Revision: 191669



URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=191669

Log:

2012-09-24  Richard Guenther  rguent...@suse.de



PR middle-end/54632

* tree-ssa-live.c (clear_unused_block_pointer_1): Do not

handle DECL_DEBUG_EXPR_IS_FROM here...

(clear_unused_block_pointer): ... but here when walking all

local decls.



Modified:

trunk/gcc/ChangeLog

trunk/gcc/tree-ssa-live.c

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-24 Thread rguenth at gcc dot gnu.org 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632



Richard Guenther rguenth at gcc dot gnu.org changed:



   What|Removed |Added



 Status|ASSIGNED|RESOLVED

 Resolution||FIXED



--- Comment #14 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-24 
15:03:28 UTC ---

Fixed.

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-23 Thread dimhen at gmail dot com 

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632

--- Comment #3 from Dmitry G. Dyachenko dimhen at gmail dot com 2012-09-23 
09:22:20 UTC ---
191649 FAIL

and, sorry, i cant reduce testcase :(

if i run 
'g++ -DHAVE_CONFIG_H -I. -I/home/dimhen/src/CSPbuild/shared/cplib -I../..
-I/usr/local/include -DUNIX -DLINUX -I/home/dimhen/src/CSPbuild
-I/home/dimhen/src/CSPbuild/reader/include
-I/home/dimhen/src/CSPbuild/CSP/src/include
-I/home/dimhen/src/CSPbuild/shared/include
-I/home/dimhen/src/CSPbuild/shared/include/atl
-I/home/dimhen/src/CSPbuild/shared/include/cplib -DNO_ASN1_RT_REDUNDANT
-D_COMPACT -fno-common -Ofast -flto -flto-partition=none -pipe -m64
-Woverloaded-virtual -fvisibility-inlines-hidden -MT DateTime.lo -MD -MP -MF
.deps/DateTime.Tpo -c /home/dimhen/src/CSPbuild/shared/cplib/DateTime.cpp 
-fPIC -DPIC -o .libs/DateTime.o'

i receive 
/home/dimhen/src/CSPbuild/shared/cplib/DateTime.cpp: In function
'CryptoPro::GeneralizedTimeToTm(std::basic_stringchar, std::char_traitschar,
std::allocatorchar  const, unsigned int, bool)':
/home/dimhen/src/CSPbuild/shared/cplib/DateTime.cpp:934:1: internal compiler
error: tree code '�F��D��`Hc���D�Hc�f;�`�bFD���(D����
   
   ��
   

HD���H!��=�' is not supported in LTO streams
 } // namespace CryptoPro
 ^
Please submit a full bug report,
with preprocessed source if appropriate.
See http://gcc.gnu.org/bugs.html for instructions.

if i add '-save-temps' and recompile ii-file, then i not receive any err
message


$ LANG=C g++ -DHAVE_CONFIG_H -I. -I/home/dimhen/src/CSPbuild/shared/cplib
-I../.. -I/usr/local/include -DUNIX -DLINUX -I/home/dimhen/src/CSPbuild
-I/home/dimhen/src/CSPbuild/reader/include
-I/home/dimhen/src/CSPbuild/CSP/src/include
-I/home/dimhen/src/CSPbuild/shared/include
-I/home/dimhen/src/CSPbuild/shared/include/atl
-I/home/dimhen/src/CSPbuild/shared/include/cplib -DNO_ASN1_RT_REDUNDANT
-D_COMPACT -fno-common -Ofast -flto -flto-partition=none -pipe -m64
-Woverloaded-virtual -fvisibility-inlines-hidden -MT DateTime.lo -MD -MP -MF
.deps/DateTime.Tpo -c /home/dimhen/src/CSPbuild/shared/cplib/DateTime.cpp 
-fPIC -DPIC -o .libs/DateTime.o -save-temps
g++: warning: -pipe ignored because -save-temps specified
$ ll -tr
итого 1712
-rw-rw-r--. 1 dimhen dimhen  22385 сент. 23 13:04 Makefile
-rw-rw-r--. 1 dimhen dimhen260 сент. 23 13:06 Blob.lo
-rw-rw-r--. 1 dimhen dimhen 854468 сент. 23 13:18 DateTime.ii
-rw-rw-r--. 1 dimhen dimhen 864446 сент. 23 13:18 DateTime.s

$ LANG=C g++ -DHAVE_CONFIG_H -I. -I/home/dimhen/src/CSPbuild/shared/cplib
-I../.. -I/usr/local/include -DUNIX -DLINUX -I/home/dimhen/src/CSPbuild
-I/home/dimhen/src/CSPbuild/reader/include
-I/home/dimhen/src/CSPbuild/CSP/src/include
-I/home/dimhen/src/CSPbuild/shared/include
-I/home/dimhen/src/CSPbuild/shared/include/atl
-I/home/dimhen/src/CSPbuild/shared/include/cplib -DNO_ASN1_RT_REDUNDANT
-D_COMPACT -fno-common -Ofast -flto -flto-partition=none -pipe -m64
-Woverloaded-virtual -fvisibility-inlines-hidden -MT DateTime.lo -MD -MP -MF
.deps/DateTime.Tpo -c ./DateTime.ii  -fPIC -DPIC -o .libs/DateTime.o

so, how i can help?

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-23 Thread dimhen at gmail dot com 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632



--- Comment #4 from Dmitry G. Dyachenko dimhen at gmail dot com 2012-09-23 
09:33:32 UTC ---

new in r191649 is eer.location -- DateTime.cpp:934:1

line 934 is outermost '}' in src file

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-23 Thread zsojka at seznam dot cz 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632



--- Comment #5 from Zdenek Sojka zsojka at seznam dot cz 2012-09-23 10:50:29 
UTC ---

Created attachment 28253

  -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=28253

partially reduced testcase



Reducing goes very slowly, this has been reducing for over a day

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-23 Thread dimhen at gmail dot com 

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632

--- Comment #6 from Dmitry G. Dyachenko dimhen at gmail dot com 2012-09-23 
10:55:08 UTC ---
valgrind show no any appropriate with debug-compiled gcc, only over-optimized
access in parser.

Hmmm, i am under impression that BUILD_CONFIG=bootstrap-O1 has no effect, and
BOOT_CFLAGS='-O0 -g' has no effect too. -- gcc is compiled with '-O2 -g'

But, at least, i made it :)

according http://gcc.gnu.org/install/build.html

~/src/gcc-current/configure --prefix=/usr/local/gcc_current
--with-multilib-list=m64 --enable-__cxa_atexit --enable-shared
--enable-checking=df,fold,rtl,tree,yes --enable-gnu-unique-object
--enable-linker-build-id --enable-languages=c,c++,lto --enable-plugin
--enable-version-specific-runtime-libs --enable-bootstrap4

BUILD_CONFIG=bootstrap-O1 make -j7

make install -j4

cp /usr/local/gcc_current/lib/gcc/x86_64-unknown-linux-gnu/lib64/libgcc_s.so
/usr/local/gcc_current/lib/gcc/x86_64-unknown-linux-gnu/4.8.0/

$g++ -v
Using built-in specs.
COLLECT_GCC=g++
COLLECT_LTO_WRAPPER=/usr/local/gcc_current/libexec/gcc/x86_64-unknown-linux-gnu/4.8.0/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: /home/dimhen/src/gcc-current/configure
--prefix=/usr/local/gcc_current --with-multilib-list=m64 --enable-__cxa_atexit
--enable-shared --enable-checking=df,fold,rtl,tree,yes
--enable-gnu-unique-object --enable-linker-build-id
--enable-languages=c,c++,lto --enable-plugin
--enable-version-specific-runtime-libs --enable-bootstrap4
Thread model: posix
gcc version 4.8.0 20120923 (experimental) [trunk revision 191649] (GCC)

$ /usr/local/gcc_current/libexec/gcc/x86_64-unknown-linux-gnu/4.8.0/cc1plus
-quiet -v -I . -I /home/dimhen/src/CSPbuild/shared/cplib -I ../.. -I
/usr/local/include -I /home/dimhen/src/CSPbuild -I
/home/dimhen/src/CSPbuild/reader/include -I
/home/dimhen/src/CSPbuild/CSP/src/include -I
/home/dimhen/src/CSPbuild/shared/include -I
/home/dimhen/src/CSPbuild/shared/include/atl -I
/home/dimhen/src/CSPbuild/shared/include/cplib -MD .libs/DateTime.d -MF
.deps/DateTime.Tpo -MP -MT DateTime.lo -D_GNU_SOURCE -D HAVE_CONFIG_H -D UNIX
-D LINUX -D NO_ASN1_RT_REDUNDANT -D _COMPACT -D PIC
/home/dimhen/src/CSPbuild/shared/cplib/DateTime.cpp -quiet -dumpbase
DateTime.cpp -m64 -mtune=generic -march=x86-64 -auxbase-strip .libs/DateTime.o
-Ofast -Woverloaded-virtual -version -fno-common -flto -flto-partition=none
-fvisibility-inlines-hidden -fPIC -o .libs/DateTime.o



/home/dimhen/src/CSPbuild/shared/cplib/DateTime.cpp: In function
'CryptoPro::GeneralizedTimeToTm(std::basic_stringchar, std::char_traitschar,
std::allocatorchar  const, unsigned int, bool)':
/home/dimhen/src/CSPbuild/shared/cplib/DateTime.cpp:934:1: internal compiler
error: tree code '�F��D���Hc��' is not supported in LTO streams
 } // namespace CryptoPro
 ^
Please submit a full bug report,
with preprocessed source if appropriate.
See http://gcc.gnu.org/bugs.html for instructions.

$ valgrind
/usr/local/gcc_current/libexec/gcc/x86_64-unknown-linux-gnu/4.8.0/cc1plus
-quiet -v -I . -I /home/dimhen/src/CSPbuild/shared/cplib -I ../.. -I
/usr/local/include -I /home/dimhen/src/CSPbuild -I
/home/dimhen/src/CSPbuild/reader/include -I
/home/dimhen/src/CSPbuild/CSP/src/include -I
/home/dimhen/src/CSPbuild/shared/include -I
/home/dimhen/src/CSPbuild/shared/include/atl -I
/home/dimhen/src/CSPbuild/shared/include/cplib -MD .libs/DateTime.d -MF
.deps/DateTime.Tpo -MP -MT DateTime.lo -D_GNU_SOURCE -D HAVE_CONFIG_H -D UNIX
-D LINUX -D NO_ASN1_RT_REDUNDANT -D _COMPACT -D PIC
/home/dimhen/src/CSPbuild/shared/cplib/DateTime.cpp -quiet -dumpbase
DateTime.cpp -m64 -mtune=generic -march=x86-64 -auxbase-strip .libs/DateTime.o
-Ofast -Woverloaded-virtual -version -fno-common -flto -flto-partition=none
-fvisibility-inlines-hidden -fPIC -o .libs/DateTime.o
[dimhen@dim cplib]$ valgrind
/usr/local/gcc_current/libexec/gcc/x86_64-unknown-linux-gnu/4.8.0/cc1plus
-quiet -v -I . -I /home/dimhen/src/CSPbuild/shared/cplib -I ../.. -I
/usr/local/include -I /home/dimhen/src/CSPbuild -I
/home/dimhen/src/CSPbuild/reader/include -I
/home/dimhen/src/CSPbuild/CSP/src/include -I
/home/dimhen/src/CSPbuild/shared/include -I
/home/dimhen/src/CSPbuild/shared/include/atl -I
/home/dimhen/src/CSPbuild/shared/include/cplib -MD .libs/DateTime.d -MF
.deps/DateTime.Tpo -MP -MT DateTime.lo -D_GNU_SOURCE -D HAVE_CONFIG_H -D UNIX
-D LINUX -D NO_ASN1_RT_REDUNDANT -D _COMPACT -D PIC
/home/dimhen/src/CSPbuild/shared/cplib/DateTime.cpp -quiet -dumpbase
DateTime.cpp -m64 -mtune=generic -march=x86-64 -auxbase-strip .libs/DateTime.o
-Ofast -Woverloaded-virtual -version -fno-common -flto -flto-partition=none
-fvisibility-inlines-hidden -fPIC -o .libs/DateTime.o
==2065== Memcheck, a memory error detector
==2065== Copyright (C) 2002-2012, and GNU GPL'd, by Julian Seward et al.
==2065== Using Valgrind-3.9.0.SVN and LibVEX; rerun with -h for copyright info
==2065== Command:
/usr/local/gcc_current/libexec/gcc/x86_64-unknown-linux-gnu/4.8.0/cc1plus

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-23 Thread markus at trippelsdorf dot de 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632



Markus Trippelsdorf markus at trippelsdorf dot de changed:



   What|Removed |Added



 CC||markus at trippelsdorf dot

   ||de



--- Comment #7 from Markus Trippelsdorf markus at trippelsdorf dot de 
2012-09-23 13:05:43 UTC ---

(In reply to comment #5)

 Created attachment 28253 [details]

 partially reduced testcase

 

 Reducing goes very slowly, this has been reducing for over a day



If the failure depends somehow on the size of the input file then delta will

approach it only asymptotically...

[Bug lto/54632] [4.8 Regression] not supported in LTO streams : tree code '�F ��D�� `

2012-09-20 Thread rguenth at gcc dot gnu.org 

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54632

Richard Guenther rguenth at gcc dot gnu.org changed:

   What|Removed |Added

 CC||dehao at gcc dot gnu.org
   Target Milestone|--- |4.8.0
Summary|not supported in LTO|[4.8 Regression] not
   |streams : tree code '�F |supported in LTO streams :
   |��D�� ` |tree code '�F ��D�� `

--- Comment #2 from Richard Guenther rguenth at gcc dot gnu.org 2012-09-20 
11:31:49 UTC ---
It's outputting a garbage collected tree.  Suspect it was a BLOCK.