[Bug objc++/61759] New: internal compiler error: in objc_eh_runtime_type, at objc/objc-next-runtime-abi-01.c:2792

[Bug objc++/61759] New: internal compiler error: in objc_eh_runtime_type, at objc/objc-next-runtime-abi-01.c:2792

[dougmencken at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

            Bug ID: 61759
           Summary: internal compiler error: in objc_eh_runtime_type, at
                    objc/objc-next-runtime-abi-01.c:2792
           Product: gcc
           Version: 4.8.3
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: objc++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dougmencken at gmail dot com

I cannot compile some LibreOffice .mm files using gcc-4.8.3 on OS X 10.5. Many
of them fails with ICE when -O2 is used, but with -O(1) they compiles.

I cannot also compile .cxx files when -x objective-c++ is used on gcc line.

All of them ICE at the same spot: "in objc_eh_runtime_type, at
objc/objc-next-runtime-abi-01.c:2792"

This one (vcl/osx/a11yselectionwrapper.mm) ICEs even with -O:

S=/lo-build && I=$S/instdir && W=$S/workdir &&  mkdir -p
$W/ObjCxxObject/vcl/osx/ $W/Dep/ObjCxxObject/vcl/osx/ && cd /lo-build &&  
/usr/bin/ccache g++ -save-temps -mmacosx-version-min=10.5 -DCPPU_ENV=gcc3
-DLIBO_INTERNAL_ONLY -DMACOSX -DMACOSX_SDK_VERSION=1050
-DMAC_OS_X_VERSION_MAX_ALLOWED=1050 -DMAC_OS_X_VERSION_MIN_REQUIRED=1050
-DNDEBUG -DNO_PTHREAD_PRIORITY -DOPTIMIZE -DOSL_DEBUG_LEVEL=0 -DPOWERPC -DPPC
-DSUPD=440 -DUNIX -DUNX -D_PTHREADS -D_REENTRANT  -DRTL_USING  
-DVCL_DLLIMPLEMENTATION -DCUI_DLL_NAME=\"libcuilo.dylib\"
-DDESKTOP_DETECTOR_DLL_NAME=\"\" -DTK_DLL_NAME=\"libtklo.dylib\"
-DVCLPLUG_SVP_DLL_NAME=\"\" 
-DMACOSX_BUNDLE_IDENTIFIER=\"org.libreoffice.script\"  
-DHAVE_GCC_VISIBILITY_FEATURE -fvisibility=hidden   -Wall -Wendif-labels
-Wextra -Wundef -Wunused-macros -fmessage-length=0 -fno-common -pipe 
-fno-threadsafe-statics -fPIC -Wno-ctor-dtor-privacy -Wno-non-virtual-dtor
-fno-strict-aliasing -fsigned-char -std=gnu++11  -O   -c
$S/vcl/osx/a11yselectionwrapper.mm -o
$W/ObjCxxObject/vcl/osx/a11yselectionwrapper.o -MMD -MT
$W/ObjCxxObject/vcl/osx/a11yselectionwrapper.o -MP -MF
$W/Dep/ObjCxxObject/vcl/osx/a11yselectionwrapper.d_ -I$S/vcl/osx/
-I$W/UnpackedTarball/lcms2/include -I$W/UnpackedTarball/icu/source
-I$W/UnpackedTarball/icu/source/i18n -I$W/UnpackedTarball/icu/source/common
-I$W/UnpackedTarball/boost -I$W/UnpackedTarball/jpeg  -I$S/include  -isystem
/Developer/SDKs/MacOSX10.5.sdk/System/Library/Frameworks/JavaVM.framework/Versions/Current/Headers
-I$S/config_host  -I$W/CustomTarget/officecfg/registry
-I$W/CustomTarget/vcl/generic/fontmanager -I$S/vcl/inc  
-I$W/UnoApiHeadersTarget/udkapi/normal -I$W/UnoApiHeadersTarget/offapi/normal
-I$W/CustomTarget/officecfg/registry  -I$W/UnpackedTarball/nss/dist/public/nss
-I$W/UnpackedTarball/nss/dist/out/include      && mv
$W/Dep/ObjCxxObject/vcl/osx/a11yselectionwrapper.d_
$W/Dep/ObjCxxObject/vcl/osx/a11yselectionwrapper.d

$ gcc -v
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/libexec/gcc/powerpc-unknown-darwin/4.8.3/lto-wrapper
Target: powerpc-unknown-darwin
Configured with: ../gcc-4.8.3/configure --build=powerpc-unknown-darwin
--host=powerpc-unknown-darwin --target=powerpc-unknown-darwin --prefix=/usr
--sysconfdir=/etc --mandir=/usr/share/man --with-slibdir=/usr/lib
--program-prefix= --enable-languages=c,c++,objc,obj-c++
--enable-checking=release --enable-stage1-checking --enable-shared
--enable-static --enable-threads=posix --with-__thread --with-system-zlib
--disable-werror
Thread model: posix
gcc version 4.8.3 (GCC)

[Bug objc++/61759] internal compiler error: in objc_eh_runtime_type, at objc/objc-next-runtime-abi-01.c:2792

[dougmencken at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #1 from Douglas Mencken <dougmencken at gmail dot com> ---
Created attachment 33094
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=33094&action=edit
preprocessed a11yselectionwrapper.mm

Attaching preprocessed a11yselectionwrapper.mm (a11yselectionwrapper.mii)

[Bug objc++/61759] internal compiler error: in objc_eh_runtime_type, at objc/objc-next-runtime-abi-01.c:2792

[dougmencken at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #2 from Douglas Mencken <dougmencken at gmail dot com> ---
Also, there's builder.cxx (one of files which compiles after removing -x
objective-c++ from line, but not with it):

S=/lo-build && I=$S/instdir && W=$S/workdir &&  mkdir -p
$W/CxxObject/vcl/source/window/ $W/Dep/CxxObject/vcl/source/window/ && cd
/lo-build &&   /usr/bin/ccache g++ -mmacosx-version-min=10.5 -DCPPU_ENV=gcc3
-DLIBO_INTERNAL_ONLY -DMACOSX -DMACOSX_SDK_VERSION=1050
-DMAC_OS_X_VERSION_MAX_ALLOWED=1050 -DMAC_OS_X_VERSION_MIN_REQUIRED=1050
-DNDEBUG -DNO_PTHREAD_PRIORITY -DOPTIMIZE -DOSL_DEBUG_LEVEL=0 -DPOWERPC -DPPC
-DSUPD=440 -DUNIX -DUNX -D_PTHREADS -D_REENTRANT  -DRTL_USING  
-DVCL_DLLIMPLEMENTATION -DCUI_DLL_NAME=\"libcuilo.dylib\"
-DDESKTOP_DETECTOR_DLL_NAME=\"\" -DTK_DLL_NAME=\"libtklo.dylib\"
-DVCLPLUG_SVP_DLL_NAME=\"\" 
-DMACOSX_BUNDLE_IDENTIFIER=\"org.libreoffice.script\"  
-DHAVE_GCC_VISIBILITY_FEATURE -fvisibility=hidden   -Wall -Wendif-labels
-Wextra -Wundef -Wunused-macros -fmessage-length=0 -fno-common -pipe 
-fno-threadsafe-statics -fPIC -Wno-ctor-dtor-privacy -Wno-non-virtual-dtor
-fno-strict-aliasing -fsigned-char -std=gnu++11  -Wno-long-double 
-DEXCEPTIONS_ON -fexceptions -fno-enforce-eh-specs -O2  -x objective-c++
-fobjc-exceptions  -c $S/vcl/source/window/builder.cxx -o
$W/CxxObject/vcl/source/window/builder.o -MMD -MT
$W/CxxObject/vcl/source/window/builder.o -MP -MF
$W/Dep/CxxObject/vcl/source/window/builder.d_ -I$S/vcl/source/window/
-I$W/UnpackedTarball/lcms2/include -I$W/UnpackedTarball/icu/source
-I$W/UnpackedTarball/icu/source/i18n -I$W/UnpackedTarball/icu/source/common
-I$W/UnpackedTarball/boost -I$W/UnpackedTarball/jpeg  -I$S/include  -isystem
/Developer/SDKs/MacOSX10.5.sdk/System/Library/Frameworks/JavaVM.framework/Versions/Current/Headers
-I$S/config_host  -I$W/CustomTarget/officecfg/registry
-I$W/CustomTarget/vcl/generic/fontmanager -I$S/vcl/inc  
-I$W/UnoApiHeadersTarget/udkapi/normal -I$W/UnoApiHeadersTarget/offapi/normal
-I$W/CustomTarget/officecfg/registry  -I$W/UnpackedTarball/nss/dist/public/nss
-I$W/UnpackedTarball/nss/dist/out/include      && mv
$W/Dep/CxxObject/vcl/source/window/builder.d_
$W/Dep/CxxObject/vcl/source/window/builder.d

** Error message:
vcl/source/window/builder.cxx: In member function 'void
VclBuilder::loadTranslations(const LanguageTag&, const rtl::OUString&)':
vcl/source/window/builder.cxx:91:6: internal compiler error: in
objc_eh_runtime_type, at objc/objc-next-runtime-abi-01.c:2792
 void VclBuilder::loadTranslations(const LanguageTag &rLanguageTag, const
OUString& rUri)
      ^
libbacktrace could not find executable to open

** Error message for vcl/osx/a11yselectionwrapper.mm was:
./vcl/osx/a11yselectionwrapper.mm:31:61: internal compiler error: in
objc_eh_runtime_type, at objc/objc-next-runtime-abi-01.c:2792
 +(id)selectedChildrenAttributeForElement:(AquaA11yWrapper *)wrapper

[Bug objc++/61759] internal compiler error: in objc_eh_runtime_type, at objc/objc-next-runtime-abi-01.c:2792

[dougmencken at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #3 from Douglas Mencken <dougmencken at gmail dot com> ---
Created attachment 33095
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=33095&action=edit
preprocessed builder.cxx

Attaching preprocessed builder.cxx (builder.mii)

[Bug objc++/61759] internal compiler error: in objc_eh_runtime_type, at objc/objc-next-runtime-abi-01.c:2792

[dougmencken at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #4 from Douglas Mencken <dougmencken at gmail dot com> ---
In 4.9.1, it's not :2792 but :2793 ---

vcl/osx/a11yselectionwrapper.cxx:31:61: internal compiler error: in
objc_eh_runtime_type, at objc/objc-next-runtime-abi-01.c:2793

[Bug objc++/61759] [ICE] [objc] reaching gcc_unreachable in objc_eh_runtime_type at objc/objc-next-runtime-abi-01.c

[dougmencken at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #5 from Douglas Mencken <dougmencken at gmail dot com> ---
Now I know it's question about non-well-supported NeXT Objective C ABIs.

Will try -fgnu-runtime

[Bug objc++/61759] [ICE] [objc] reaching gcc_unreachable in objc_eh_runtime_type at objc/objc-next-runtime-abi-01.c

[dougmencken at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #6 from Douglas Mencken <dougmencken at gmail dot com> ---
Looks like I found the root of the issue ~
GCC ICEs when it meets C++ exception handling (try+catch)

$ cat main.mm
#include <Foundation/Foundation.h>
#include <iostream>

int main (void)
{
        try {
                throw 0;
        } catch (int & e) {
                std::cout << "caught" << std::endl;
        }

        NSString *name = @"GNUstep !";
        NSAutoreleasePool *pool;
        pool = [NSAutoreleasePool new];
        [pool drain];
}

$ g++ main.mm -o test.out -framework Foundation -lobjc
main.mm: In function 'int main()':
main.mm:4:5: internal compiler error: in objc_eh_runtime_type, at
objc/objc-next-runtime-abi-01.c:2804
 int main (void)
     ^
libbacktrace could not find executable to open
Please submit a full bug report,

[Bug objc++/61759] [ICE] [objc++] reaching gcc_unreachable in objc_eh_runtime_type at objc/objc-next-runtime-abi-01.c

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |vital.had at gmail dot com

--- Comment #11 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 107530 has been marked as a duplicate of this bug. ***

[Bug objc++/61759] [ICE] [objc++] reaching gcc_unreachable in objc_eh_runtime_type at objc/objc-next-runtime-abi-01.c

[vital.had at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #12 from Sergey Fedorov <vital.had at gmail dot com> ---
(In reply to Andrew Pinski from comment #11)
> *** Bug 107530 has been marked as a duplicate of this bug. ***

I.e. gcc12 has the same bug too.

[Bug objc++/61759] [ICE] [objc++] reaching gcc_unreachable in objc_eh_runtime_type at objc/objc-next-runtime-abi-01.c

[vital.had at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #13 from Sergey Fedorov <vital.had at gmail dot com> ---
Any update on this?

[Bug objc++/61759] [ICE] [objc++] reaching gcc_unreachable in objc_eh_runtime_type at objc/objc-next-runtime-abi-01.c

[vital.had at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #14 from Sergey Fedorov <vital.had at gmail dot com> ---
(In reply to Eric Gallager from comment #8)

> I'd need to rebuild gcc with debug info to get a better backtrace.

Have you been able to?

[Bug objc++/61759] [ICE] [objc++] reaching gcc_unreachable in objc_eh_runtime_type at objc/objc-next-runtime-abi-01.c

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

Iain Sandoe <iains at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|*-*-darwin                  |powerpc-*-darwin,
                   |                            |i686-*-darwin

--- Comment #15 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Sergey Fedorov from comment #14)
> (In reply to Eric Gallager from comment #8)

With current trunk on a cross to powerpc-apple-darwin9 this reproduces for the
runtime ABI 01 - which is used on 32b Darwin. (i.e the code compiles fine for
'-m64')

A guess would be that we have a conflict between C++ exception handling (DWARF
unwinding) and runtime-abi-01 exception handling (set jump / long jump). 
Probably, there needs to be some way to flag which kind of exception is
present.

For runtime ABI 02 (64bit) this should not happen - so the only really affected
targets are 32b powerpc and i686 Darwin.

===== backtrace for curent trunk:

  * frame #0: 0x00000001021b78a5 cc1objplus`internal_error(gmsgid="in %s, at
%s:%d") at diagnostic.cc:2166:25
    frame #1: 0x00000001021b7d8b
cc1objplus`fancy_abort(file="/src-local/gcc-master/gcc/objc/objc-next-runtime-abi-01.cc",
line=2795, function="objc_eh_runtime_type") at diagnostic.cc:2296:18
    frame #2: 0x000000010003dab7
cc1objplus`::objc_eh_runtime_type(type=0x000000011401c5e8) at
objc-next-runtime-abi-01.cc:2795:3
    frame #3: 0x00000001011501ad
cc1objplus`add_type_for_runtime(type=0x000000011401c5e8) at except.cc:683:40
    frame #4: 0x000000010114f8c4
cc1objplus`gen_eh_region_catch(t=0x0000000117c3b318,
type_or_list=0x000000011401c5e8) at except.cc:424:23
    frame #5: 0x0000000101a400bb
cc1objplus`::lower_catch(state=0x0000000309267190, tp=0x000000014813f690) at
tree-eh.cc:1815:31
    frame #6: 0x0000000101a40cfe
cc1objplus`::lower_eh_constructs_2(state=0x0000000309267190,
gsi=0x0000000309267140) at tree-eh.cc:2107:27
    frame #7: 0x0000000101a40e17
cc1objplus`::lower_eh_constructs_1(state=0x0000000309267190,
pseq=0x0000000309267188) at tree-eh.cc:2153:27
    frame #8: 0x0000000101a40f46
cc1objplus`pass_lower_eh::execute(this=0x000060000260c540,
fun=0x00000001480e8678) const at tree-eh.cc:2198:25
    frame #9: 0x00000001017b5131
cc1objplus`execute_one_pass(pass=0x000060000260c540) at passes.cc:2641:30
    frame #10: 0x00000001017b5570
cc1objplus`::execute_pass_list_1(pass=0x000060000260c540) at passes.cc:2750:28
    frame #11: 0x00000001017b5607
cc1objplus`execute_pass_list(fn=0x00000001480e8678, pass=0x000060000260c240) at
passes.cc:2761:23
    frame #12: 0x0000000100fe00c4
cc1objplus`cgraph_node::analyze(this=0x0000000148128990) at
cgraphunit.cc:694:22
    frame #13: 0x0000000100fe23d1
cc1objplus`::analyze_functions(first_time=true) at cgraphunit.cc:1247:18
    frame #14: 0x0000000100fe5ed4
cc1objplus`symbol_table::finalize_compilation_unit(this=0x0000000114006000) at
cgraphunit.cc:2554:21
    frame #15: 0x0000000101987ad5 cc1objplus`::compile_file() at
toplev.cc:471:41
    frame #16: 0x000000010198b4f0 cc1objplus`::do_compile(no_backend=false) at
toplev.cc:2126:24
    frame #17: 0x000000010198ba34
cc1objplus`toplev::main(this=0x000000030926751a, argc=16,
argv=0x0000000309267668) at toplev.cc:2282:18
    frame #18: 0x0000000102187a64 cc1objplus`main(argc=16,
argv=0x0000000309267668) at main.cc:39:23

[Bug objc++/61759] [ICE] [objc++] reaching gcc_unreachable in objc_eh_runtime_type at objc/objc-next-runtime-abi-01.c

[vital.had at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #16 from Sergey Fedorov <vital.had at gmail dot com> ---
(In reply to Iain Sandoe from comment #15)

Would be nice to have it fixed, otherwise `jack` is broken:
https://github.com/jackaudio/jack2/issues/950
Which, in turn, leaves broken dependencies.

[Bug objc++/61759] [ICE] [objc++] reaching gcc_unreachable in objc_eh_runtime_type at objc/objc-next-runtime-abi-01.c

[vital.had at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61759

--- Comment #17 from Sergey Fedorov <vital.had at gmail dot com> ---
(In reply to Iain Sandoe from comment #15)

Iain, any chance of addressing this one?

If would have really helped with quite a number of usable software for PowerPC
(majority of which are 32-bit, and even on G5 not all software support ppc64
anyway).