[Bug middle-end/46510] New: r166812 breaks bootstrap on x86_64-apple-darwin10

[Bug middle-end/46510] New: r166812 breaks bootstrap on x86_64-apple-darwin10

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

           Summary: r166812 breaks bootstrap on x86_64-apple-darwin10
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: howarth@nitro.med.uc.edu


The commit...

Author: hubicka
Date: Tue Nov 16 18:15:17 2010
New Revision: 166812

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=166812
Log:

    * cgraph.h (+varpool_can_remove_if_no_refs): Move here from ...;
    when !flag_toplevel_reorder do not remove unless variable is
    COMDAT or ARTIFICIAL.
    * ipa.c (varpool_can_remove_if_no_refs): ... here.
    (cgraph_remove_unreachable_nodes): Only analyzed nodes needs to stay.
    * cgraphunit.c (cgraph_analyze_functions): Dump varpool, too.
    * varpool.c (decide_is_variable_needed): Do not handle visibility issues.
    (varpool_finalize_decl): Likewise.
    (varpool_remove_unreferenced_decls): Use varpool_mark_needed_node; update
    outdated comment on DECL_RTL_SET_P check.

breaks the bootstrap of x86_64-apple-darwin10. The bootstrap now fails with...

libtool: compile:  /sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/./gcc/xgcc
-B/sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/./gcc/
-B/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/bin/
-B/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/lib/ -isystem
/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/include -isystem
/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/sys-include -DHAVE_CONFIG_H -I.
-I../../../gcc-4.6-20101116/libgomp
-I../../../gcc-4.6-20101116/libgomp/config/bsd
-I../../../gcc-4.6-20101116/libgomp/config/posix
-I../../../gcc-4.6-20101116/libgomp -Wall -pthread -Werror -g -O2 -MT loop.lo
-MD -MP -MF .deps/loop.Tpo -c ../../../gcc-4.6-20101116/libgomp/loop.c 
-fno-common -DPIC -o .libs/loop.o
../../../gcc-4.6-20101116/libgomp/barrier.c:41:1: internal compiler error: in
decide_is_variable_needed, at varpool.c:338

The same error occurs when compiling libgomp/iter.c:333:1,
libgomp/iter_ull.c:340:1, libgomp/env.c:658:1 and libgomp/loop.c:619:1.

../gcc-4.6-20101116/configure --prefix=/sw --prefix=/sw/lib/gcc4.6
--mandir=/sw/share/man --infodir=/sw/lib/gcc4.6/info
--enable-languages=c,c++,fortran,objc,obj-c++,java --with-gmp=/sw
--with-libiconv-prefix=/sw --with-ppl=/sw --with-cloog=/sw --with-mpc=/sw
--with-system-zlib --x-includes=/usr/X11R6/include --x-libraries=/usr/X11R6/lib
--program-suffix=-fsf-4.6 --enable-checking=yes --enable-cloog-backend=isl

[Bug middle-end/46510] r166812 breaks bootstrap on x86_64-apple-darwin10

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|x86_64-apple-darwin10       |*-*-darwin*
               Host|x86_64-apple-darwin10       |
              Build|x86_64-apple-darwin10       |

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> 2010-11-16 23:05:55 UTC ---
powerpc-darwin is broken the same way:
http://gcc.gnu.org/ml/gcc-regression/2010-11/msg00205.html

[Bug middle-end/46510] r166812 breaks bootstrap on x86_64-apple-darwin10

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #2 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-11-16 23:07:53 UTC ---
Created attachment 22425
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22425
preprocessed source for libgomp/barrier.c

Created with...

/sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/./gcc/xgcc
-B/sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/./gcc/
-B/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/bin/
-B/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/lib/ -isystem
/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/include -isystem
/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/sys-include -DHAVE_CONFIG_H -I.
-I../../../gcc-4.6-20101116/libgomp
-I../../../gcc-4.6-20101116/libgomp/config/bsd
-I../../../gcc-4.6-20101116/libgomp/config/posix
-I../../../gcc-4.6-20101116/libgomp -Wall -pthread -Werror -g -O2 -MT
barrier.lo -MD -MP -MF .deps/barrier.Tpo -c
../../../gcc-4.6-20101116/libgomp/barrier.c -fno-common -DPIC --save-temps -v
-o .libs/barrier.o

[Bug middle-end/46510] r166812 breaks bootstrap on x86_64-apple-darwin10

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dominiq at lps dot ens.fr

--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2010-11-16 23:37:49 UTC ---
*** Bug 46511 has been marked as a duplicate of this bug. ***

[Bug middle-end/46510] r166812 breaks bootstrap on x86_64-apple-darwin10

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #4 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2010-11-16 23:42:24 UTC ---
A backtrace can be found in pr46511.

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1
   Target Milestone|---                         |4.6.0
            Summary|r166812 breaks bootstrap on |[4.6 Regression] r166812
                   |x86_64-apple-darwin10       |breaks bootstrap on
                   |                            |x86_64-apple-darwin10

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

Jan Hubicka <hubicka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2010.11.17 16:21:39
     Ever Confirmed|0                           |1

--- Comment #5 from Jan Hubicka <hubicka at gcc dot gnu.org> 2010-11-17 16:21:39 UTC ---
Can you, please, try if the following works?
Index: tree-emutls.c
===================================================================
--- tree-emutls.c       (revision 166864)
+++ tree-emutls.c       (working copy)
@@ -324,7 +324,8 @@ new_emutls_decl (tree decl)
       record_references_in_initializer (to, false);
     }

-  varpool_finalize_decl (to);
+  if (!DECL_EXTERNAL (to))
+    varpool_finalize_decl (to);
   return to;
 }

We should not finalize external vars.

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #6 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-11-17 16:34:33 UTC ---
(In reply to comment #5)
> Can you, please, try if the following works?
> Index: tree-emutls.c
> ===================================================================
> --- tree-emutls.c       (revision 166864)
> +++ tree-emutls.c       (working copy)
> @@ -324,7 +324,8 @@ new_emutls_decl (tree decl)
>        record_references_in_initializer (to, false);
>      }
> 
> -  varpool_finalize_decl (to);
> +  if (!DECL_EXTERNAL (to))
> +    varpool_finalize_decl (to);
>    return to;
>  }
> 
> We should not finalize external vars.

Converts the ICE to a segfault...

libtool: compile:  /sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/./gcc/xgcc
-B/sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/./gcc/
-B/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/bin/
-B/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/lib/ -isystem
/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/include -isystem
/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/sys-include -DHAVE_CONFIG_H -I.
-I../../../gcc-4.6-20101117/libgomp
-I../../../gcc-4.6-20101117/libgomp/config/bsd
-I../../../gcc-4.6-20101117/libgomp/config/posix
-I../../../gcc-4.6-20101117/libgomp -Wall -pthread -Werror -g -O2 -MT loop.lo
-MD -MP -MF .deps/loop.Tpo -c ../../../gcc-4.6-20101117/libgomp/loop.c 
-fno-common -DPIC -o .libs/loop.o
../../../gcc-4.6-20101117/libgomp/barrier.c: In function 'GOMP_barrier':
../../../gcc-4.6-20101117/libgomp/barrier.c:41:1: internal compiler error:
Segmentation fault

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #7 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-11-17 16:38:47 UTC ---
Backtraces as...

gdb /sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/./gcc/cc1
...
(gdb) r -quiet -v -I . -I ../../../gcc-4.6-20101117/libgomp -I
../../../gcc-4.6-20101117/libgomp/config/bsd -I
../../../gcc-4.6-20101117/libgomp/config/posix -I
../../../gcc-4.6-20101117/libgomp -iprefix
/sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/gcc/../lib/gcc/x86_64-apple-darwin10.5.0/4.6.0/
-isystem /sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/./gcc/include
-isystem /sw/src/fink.build/gcc46-4.6.0-1000/darwin_objdir/./gcc/include-fixed
-MD .libs/barrier.d -MF .deps/barrier.Tpo -MP -MT barrier.lo -D__DYNAMIC__
-D_REENTRANT -D HAVE_CONFIG_H -D PIC -isystem
/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/include -isystem
/sw/lib/gcc4.6/x86_64-apple-darwin10.5.0/sys-include
../../../gcc-4.6-20101117/libgomp/barrier.c -fPIC
-feliminate-unused-debug-symbols -quiet -dumpbase barrier.c
-mmacosx-version-min=10.6.5 -mtune=generic -auxbase-strip .libs/barrier.o -g
-O2 -Wall -Werror -version -fno-common -o /var/tmp//cckhMI20.s
...

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0x0000000000000000
0x00000001009e08b2 in gen_emutls_addr (decl=0x14340a8c0, d=0x7fff5fbfe650) at
../../gcc-4.6-20101117/gcc/tree-emutls.c:418
418          cdecl = cvar->decl;
(gdb) bt
#0  0x00000001009e08b2 in gen_emutls_addr (decl=0x14340a8c0, d=0x7fff5fbfe650)
at ../../gcc-4.6-20101117/gcc/tree-emutls.c:418
#1  0x00000001009e1181 in lower_emutls_1 (ptr=0x1434d86a8,
walk_subtrees=0x7fff5fbfdd9c, cb_data=0x7fff5fbfe5a0) at
../../gcc-4.6-20101117/gcc/tree-emutls.c:527
#2  0x0000000100c29101 in walk_tree_1 (tp=0x1434d86a8, func=0x1009e0d5e
<lower_emutls_1>, data=0x7fff5fbfe5a0, pset=0x0, lh=0) at
../../gcc-4.6-20101117/gcc/tree.c:10241
#3  0x0000000100c2b9b5 in walk_tree_1 (tp=0x1434d86e8, func=0x1009e0d5e
<lower_emutls_1>, data=0x7fff5fbfe5a0, pset=0x0, lh=0) at
../../gcc-4.6-20101117/gcc/tree.c:10490
#4  0x0000000100c2b9b5 in walk_tree_1 (tp=0x1434ff688, func=0x1009e0d5e
<lower_emutls_1>, data=0x7fff5fbfe5a0, pset=0x0, lh=0) at
../../gcc-4.6-20101117/gcc/tree.c:10490
#5  0x00000001006960b8 in walk_gimple_op (stmt=0x1434ff640,
callback_op=0x1009e0d5e <lower_emutls_1>, wi=0x7fff5fbfe5a0) at
../../gcc-4.6-20101117/gcc/gimple.c:1361
#6  0x00000001009e1468 in lower_emutls_stmt (stmt=0x1434ff640,
d=0x7fff5fbfe650) at ../../gcc-4.6-20101117/gcc/tree-emutls.c:557
#7  0x00000001009e1a1f in lower_emutls_function_body (node=0x1434d7dc0) at
../../gcc-4.6-20101117/gcc/tree-emutls.c:663
#8  0x00000001009e2318 in ipa_lower_emutls () at
../../gcc-4.6-20101117/gcc/tree-emutls.c:765
#9  0x000000010084eed2 in execute_one_pass (pass=0x1010ab3a0) at
../../gcc-4.6-20101117/gcc/passes.c:1564
#10 0x000000010084fee3 in execute_ipa_pass_list (pass=0x1010ab3a0) at
../../gcc-4.6-20101117/gcc/passes.c:1931
#11 0x0000000100cdf555 in ipa_passes () at
../../gcc-4.6-20101117/gcc/cgraphunit.c:1696
#12 0x0000000100cdf738 in cgraph_optimize () at
../../gcc-4.6-20101117/gcc/cgraphunit.c:1765
#13 0x0000000100cdcad5 in cgraph_finalize_compilation_unit () at
../../gcc-4.6-20101117/gcc/cgraphunit.c:1017
#14 0x0000000100030c41 in c_write_global_declarations () at
../../gcc-4.6-20101117/gcc/c-decl.c:9837
#15 0x0000000100969368 in compile_file () at
../../gcc-4.6-20101117/gcc/toplev.c:882
#16 0x000000010096c470 in do_compile () at
../../gcc-4.6-20101117/gcc/toplev.c:2313
#17 0x000000010096c5b8 in toplev_main (argc=54, argv=0x7fff5fbfe950) at
../../gcc-4.6-20101117/gcc/toplev.c:2374
#18 0x0000000100115548 in main (argc=54, argv=0x7fff5fbfe950) at
../../gcc-4.6-20101117/gcc/main.c:36

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #8 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2010-11-17 21:39:08 UTC ---
As I said in pr46511, removing the assert fixes the bootstrap and does not seem
to introduce any regression (finishing regtesting):

--- /opt/gcc/_clean/gcc/varpool.c    2010-11-16 19:21:22.000000000 +0100
+++ /opt/gcc/work/gcc/varpool.c    2010-11-17 10:59:15.000000000 +0100
@@ -335,7 +335,7 @@ decide_is_variable_needed (struct varpoo
   if (node->force_output)
     return true;

-  gcc_assert (!DECL_EXTERNAL (decl));
+  /* gcc_assert (!DECL_EXTERNAL (decl)); */

   /* Externally visible variables must be output.  The exception is
      COMDAT variables that must be output only when they are needed.  */

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

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

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

--- Comment #9 from Richard Guenther <rguenth at gcc dot gnu.org> 2010-11-18 10:12:07 UTC ---
*** Bug 46530 has been marked as a duplicate of this bug. ***

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[ramana at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

Ramana Radhakrishnan <ramana at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|*-*-darwin*                 |*-*-darwin*, arm*-eabi
                 CC|                            |ramana at gcc dot gnu.org

--- Comment #10 from Ramana Radhakrishnan <ramana at gcc dot gnu.org> 2010-11-18 15:27:07 UTC ---
Also occurs for arm-eabi with the tls tests in the testsuite. 


FAIL: gcc.dg/tls/opt-14.c (internal compiler error)

gcc.dg/tls/opt-14.c:28:1: internal compiler error: in
decide_is_variable_needed, at varpool.c:338


GCC configured with. 

/arm/scratch/pdtltest/work/sandboxes/fsf/trunk/source/gcc-fsf/configure
--target=arm-eabi --with-cpu=cortex-a9 --with-float=softfp --with-fpu=neon

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[ramana at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #11 from Ramana Radhakrishnan <ramana at gcc dot gnu.org> 2010-11-18 15:29:24 UTC ---
(In reply to comment #10)
> Also occurs for arm-eabi with the tls tests in the testsuite. 
> 
> 
> FAIL: gcc.dg/tls/opt-14.c (internal compiler error)
> 
> gcc.dg/tls/opt-14.c:28:1: internal compiler error: in
> decide_is_variable_needed, at varpool.c:338
> 
> 
> GCC configured with. 
> 
> /arm/scratch/pdtltest/work/sandboxes/fsf/trunk/source/gcc-fsf/configure
> --target=arm-eabi --with-cpu=cortex-a9 --with-float=softfp --with-fpu=neon

And so do all these tests. 

arm-sim/-mthumb: gcc.dg/tls/nonpic-1.c (internal compiler error)
arm-sim/-mthumb: gcc.dg/tls/opt-1.c (internal compiler error)
arm-sim/-mthumb: gcc.dg/tls/opt-14.c (internal compiler error)
arm-sim/-mthumb: gcc.dg/tls/opt-3.c (internal compiler error)
arm-sim/-mthumb: gcc.dg/tls/opt-5.c (internal compiler error)
arm-sim/-mthumb: gcc.dg/tls/opt-6.c (internal compiler error)
arm-sim/-mthumb: gcc.dg/tls/opt-9.c (internal compiler error)
arm-sim/-mthumb: gcc.dg/tls/pic-1.c (internal compiler error)
arm-sim: gcc.dg/tls/nonpic-1.c (internal compiler error)
arm-sim: gcc.dg/tls/opt-1.c (internal compiler error)
arm-sim: gcc.dg/tls/opt-14.c (internal compiler error)
arm-sim: gcc.dg/tls/opt-3.c (internal compiler error)
arm-sim: gcc.dg/tls/opt-5.c (internal compiler error)
arm-sim: gcc.dg/tls/opt-6.c (internal compiler error)
arm-sim: gcc.dg/tls/opt-9.c (internal compiler error)
arm-sim: gcc.dg/tls/pic-1.c (internal compiler error)

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #12 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2010-11-18 15:49:53 UTC ---
(In reply to comments #10 and #11)
> Also occurs for arm-eabi with the tls tests in the testsuite. 
> ...

What happens if you remove the gcc_assert at line 338 of varpool.c (see comment
#8)?

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[dave at hiauly1 dot hia.nrc.ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #13 from dave at hiauly1 dot hia.nrc.ca 2010-11-18 23:37:15 UTC ---
On Thu, 18 Nov 2010, dominiq at lps dot ens.fr wrote:

> What happens if you remove the gcc_assert at line 338 of varpool.c (see comment
> #8)?

hppa2.0w-hp-hpux11.11  bootstraps successfully with the assert removed.

Dave

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #14 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2010-11-18 23:44:09 UTC ---
> hppa2.0w-hp-hpux11.11  bootstraps successfully with the assert removed.

Can you also regtest? On x86_64-apple-darwin10 I did not see any regression and
powerpc-apple-darwin9 is currently regtesting without anything obvious so far
(at obj-c++).

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[dave at hiauly1 dot hia.nrc.ca]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #15 from dave at hiauly1 dot hia.nrc.ca 2010-11-18 23:53:19 UTC ---
> powerpc-apple-darwin9 is currently regtesting without anything obvious so far
> (at obj-c++).

My MacPro is darwin9, and I'm only setup to do 32-bit builds at the moment.

Dave

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #16 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-11-19 13:48:45 UTC ---
I can confirm that removing the gcc_assert (!DECL_EXTERNAL (decl)) as suggested
in comment 8 does eliminate the bootstrap failures without regressions on
x86_64-apple-darwin10...

http://gcc.gnu.org/ml/gcc-testresults/2010-11/msg01621.html

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #17 from Jan Hubicka <hubicka at gcc dot gnu.org> 2010-11-21 13:25:03 UTC ---
Created attachment 22476
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22476
Proposed patch

Hi,
can you please try this variant?

Honza

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #18 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2010-11-21 16:12:48 UTC ---
> can you please try this variant?

Applied on tom of revision 167004, I still get on *darwin*

../../../work/libgomp/barrier.c:41:1: internal compiler error: in
decide_is_variable_needed, at varpool.c:341

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[hubicka at ucw dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #19 from Jan Hubicka <hubicka at ucw dot cz> 2010-11-21 16:16:51 UTC ---
> ../../../work/libgomp/barrier.c:41:1: internal compiler error: in
> decide_is_variable_needed, at varpool.c:341

Hmm, interesting, get_emutls_init_templ_addr should no longer call
finalize_decl
on externals.  Can you, please, post the backtrace?  Perhaps there is some
other
place that is wrong about this.

Honza

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #20 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-11-21 17:20:40 UTC ---
Created attachment 22477
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22477
backtrace for patch proposed in comment 17

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #21 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-11-21 17:25:19 UTC ---
(In reply to comment #19)
> > ../../../work/libgomp/barrier.c:41:1: internal compiler error: in
> > decide_is_variable_needed, at varpool.c:341
> 
> Hmm, interesting, get_emutls_init_templ_addr should no longer call
> finalize_decl
> on externals.  Can you, please, post the backtrace?  Perhaps there is some
> other
> place that is wrong about this.
> 
> Honza

Honza,
    I don't know how to do a backtrace in gdb when the compiler exits in gdb
during the run as...

Program exited with code 04.

so I uploaded a single step walk from the 32nd instance of the varpool.c:341
breakpoint.

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[hubicka at ucw dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #22 from Jan Hubicka <hubicka at ucw dot cz> 2010-11-21 17:47:33 UTC ---
> Honza,
>     I don't know how to do a backtrace in gdb when the compiler exits in gdb
> during the run as...
> 
> Program exited with code 04.

breakpoint on internal_error should help. I will take a look.

Honza

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #23 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-11-21 18:02:40 UTC ---
(In reply to comment #22)
> > Honza,
> >     I don't know how to do a backtrace in gdb when the compiler exits in gdb
> > during the run as...
> > 
> > Program exited with code 04.
> 
> breakpoint on internal_error should help. I will take a look.
> 
> Honza

That doesn't work here.

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #24 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2010-11-21 18:32:49 UTC ---
> breakpoint on internal_error should help. I will take a look.

with breakpoint at fancy_abort:

#0  fancy_abort (file=0x100e4f182 "../../work/gcc/varpool.c", line=341,
function=0x100e4f28f "decide_is_variable_needed") at
../../work/gcc/diagnostic.c:893
#1  0x0000000100d4d541 in decide_is_variable_needed (node=0x142d0b6e8,
decl=0x142ddfaa0) at ../../work/gcc/varpool.c:341
#2  0x0000000100d4dab0 in varpool_finalize_decl (decl=0x142ddfaa0) at
../../work/gcc/varpool.c:424
#3  0x00000001009e1ed3 in new_emutls_decl (decl=0x142d088c0) at
../../work/gcc/tree-emutls.c:332
#4  0x00000001009e3ae6 in ipa_lower_emutls () at
../../work/gcc/tree-emutls.c:728
#5  0x0000000100850420 in execute_one_pass (pass=0x1010ad480) at
../../work/gcc/passes.c:1564
#6  0x0000000100851431 in execute_ipa_pass_list (pass=0x1010ad480) at
../../work/gcc/passes.c:1931
#7  0x0000000100ce211a in ipa_passes () at ../../work/gcc/cgraphunit.c:1696
#8  0x0000000100ce22fd in cgraph_optimize () at
../../work/gcc/cgraphunit.c:1765
#9  0x0000000100cdf69a in cgraph_finalize_compilation_unit () at
../../work/gcc/cgraphunit.c:1017
#10 0x000000010003100d in c_write_global_declarations () at
../../work/gcc/c-decl.c:9837
#11 0x000000010096aa40 in compile_file () at ../../work/gcc/toplev.c:888
#12 0x000000010096db48 in do_compile () at ../../work/gcc/toplev.c:2319
#13 0x000000010096dc98 in toplev_main (argc=8, argv=0x7fff5fbfd900) at
../../work/gcc/toplev.c:2382
#14 0x0000000100115df7 in main (argc=8, argv=0x7fff5fbfd900) at
../../work/gcc/main.c:36

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #25 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2010-11-21 18:37:09 UTC ---
This seems to work:


--- ../_clean/gcc/tree-emutls.c    2010-11-19 18:13:00.000000000 +0100
+++ gcc/tree-emutls.c    2010-11-21 19:34:11.000000000 +0100
@@ -257,7 +257,12 @@ get_emutls_init_templ_addr (tree decl)
             targetm.emutls.tmpl_section);
     }

-  varpool_finalize_decl (to);
+  /* Create varpool node for the new variable and finalize it if it is
+     not external one.  */
+  if (DECL_EXTERNAL (to))
+    varpool_node (to);
+  else
+    varpool_finalize_decl (to);
   return build_fold_addr_expr (to);
 }

@@ -324,7 +329,12 @@ new_emutls_decl (tree decl)
       record_references_in_initializer (to, false);
     }

-  varpool_finalize_decl (to);
+  /* Create varpool node for the new variable and finalize it if it is
+     not external one.  */
+  if (DECL_EXTERNAL (to))
+    varpool_node (to);
+  else
+    varpool_finalize_decl (to);
   return to;
 }

Re: [Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[Jan Hubicka]

> This seems to work:
This is OK, thanks!

Honza

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[hubicka at ucw dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #26 from Jan Hubicka <hubicka at ucw dot cz> 2010-11-21 18:55:03 UTC ---
> This seems to work:
This is OK, thanks!

Honza

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

--- Comment #27 from Jan Hubicka <hubicka at gcc dot gnu.org> 2010-11-21 23:02:20 UTC ---
Author: hubicka
Date: Sun Nov 21 23:02:15 2010
New Revision: 167014

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=167014
Log:

    PR target/46510
    * tree-emutls.c (get_emutls_init_templ_addr, new_emutls_decl): Do not
    finalize external decls.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/tree-emutls.c

[Bug middle-end/46510] [4.6 Regression] r166812 breaks bootstrap on x86_64-apple-darwin10

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46510

Jack Howarth <howarth at nitro dot med.uc.edu> changed:

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

--- Comment #28 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-11-22 00:03:38 UTC ---
Fixed at r167014.