http://bugzilla.gdcproject.org/show_bug.cgi?id=196

--- Comment #1 from John Colvin <john.loughran.colvin@gmail.com> ---
Created attachment 86
  --> http://bugzilla.gdcproject.org/attachment.cgi?id=86&action=edit
build log

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=196

Johannes Pfau <johannespfau@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |johannespfau@gmail.com

--- Comment #2 from Johannes Pfau <johannespfau@gmail.com> ---
Can you try to get the ASM output? You need to cd to the correct directory, then run

/Users/johncolvin/Git/GDC/objdir/./gcc/gdc -B/Users/johncolvin/Git/GDC/objdir/./gcc/ -B/opt/gdc/x86_64-apple-darwin15.0.0/bin/ -B/opt/gdc/x86_64-apple-darwin15.0.0/lib/ -isystem /opt/gdc/x86_64-apple-darwin15.0.0/include -isystem /opt/gdc/x86_64-apple-darwin15.0.0/sys-include -Wall -Werror -g -frelease -O2 -nostdinc -pipe -Wno-deprecated -I ../../../../gcc-6-20150719/libphobos/libdruntime -I ./x86_64-apple-darwin15.0.0 -I .  -S ../../../../gcc-6-20150719/libphobos/libdruntime/core/thread.d

(Sounds like something messed up the ASM output. Maybe some incorrect inline ASM somewhere. IIRC there's also problem with emutls which in some cases generates invalid names in ASM)

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=196

--- Comment #3 from John Colvin <john.loughran.colvin@gmail.com> ---
What is the correct directory? Trying to work it out from the various directories in that compilation command:

$ pwd
/Users/johncolvin/Git/GDC/objdir
$ find . -name x86_64-apple-darwin15.0.0 -type d
./prev-x86_64-apple-darwin15.0.0/i386/libstdc++-v3/include/x86_64-apple-darwin15.0.0
./prev-x86_64-apple-darwin15.0.0/libstdc++-v3/include/x86_64-apple-darwin15.0.0
./stage1-x86_64-apple-darwin15.0.0/i386/libstdc++-v3/include/x86_64-apple-darwin15.0.0
./stage1-x86_64-apple-darwin15.0.0/libstdc++-v3/include/x86_64-apple-darwin15.0.0
./x86_64-apple-darwin15.0.0
./x86_64-apple-darwin15.0.0/i386/libstdc++-v3/include/x86_64-apple-darwin15.0.0
./x86_64-apple-darwin15.0.0/libstdc++-v3/include/x86_64-apple-darwin15.0.0

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=196

--- Comment #4 from John Colvin <john.loughran.colvin@gmail.com> ---
Created attachment 87
  --> http://bugzilla.gdcproject.org/attachment.cgi?id=87&action=edit
assembly from thread.d

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=196

--- Comment #5 from John Colvin <john.loughran.colvin@gmail.com> ---
Well I tried the last one of those directories and got something, please find attached.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=196

--- Comment #6 from Iain Buclaw <ibuclaw@gdcproject.org> ---
Something wacky going on in the backend:

```
    .globl
__D4core6thread13onThreadErrorFNbAyaC6object9ThrowableZ5errorC4core6thread11ThreadError
    .data
    .align 3
__D4core6thread13onThreadErrorFNbAyaC6object9ThrowableZ5errorC4core6thread11ThreadError:
    .quad    .4648
    .const_data
    .align 5
.4648:   /* Line 23283 */
    .quad    __D4core6thread11ThreadError6__vtblZ
    .quad    0
    .space 16
    .quad    62
    .quad    LC13
    .quad    2765
    .space 8
    .space 8
    .space 8
```

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=196

--- Comment #7 from John Colvin <john.loughran.colvin@gmail.com> ---
Just tried this again with a few different combinations of gdc and gcc versions, all with the same result.

Is there anything I can do to help with this? remote login details for an OS X machine perhaps?

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=196

--- Comment #8 from Iain Buclaw <ibuclaw@gdcproject.org> ---
First, dustmite it.  This should be possible to reproduce in a cross compiler too.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=196

--- Comment #9 from John Colvin <john.loughran.colvin@gmail.com> ---
(In reply to Iain Buclaw from comment #8)
> First, dustmite it.  This should be possible to reproduce in a cross compiler too.

Ok, after dustmite and a little manual cleaning:

//thread.d
class ThreadError : Error
{
    this(string )
    {
        super(msg, file, line);
    }
}

void onThreadError() {
    __gshared ThreadError = new ThreadError(null);
}

$ /Users/johncolvin/Documents/GDC/objdir/./gcc/gdc -B/Users/johncolvin/Documents/GDC/objdir/./gcc/ -B/opt/gdc/x86_64-apple-darwin15.0.0/bin/ -B/opt/gdc/x86_64-apple-darwin15.0.0/lib/ -isystem /opt/gdc/x86_64-apple-darwin15.0.0/include -isystem /opt/gdc/x86_64-apple-darwin15.0.0/sys-include    -o thread.o -Wall -Werror -g -frelease -O2 -nostdinc -pipe -Wno-deprecated -I ../../../../../gcc-5.2.0/libphobos/libdruntime -I ../x86_64-apple-darwin15.0.0 -I ../  -c thread.d

<stdin>:109:1: error: unexpected token at start of statement
.3614:
^


If I comment out the super line, I get

thread.d:3:5: error: constructor thread.ThreadError.this no match for implicit
super() call in constructor
     this(string )
     ^
Assertion failed: (!fd->semantic3Errors), function ctfeCompile, file
../../gcc-5.2.0/gcc/d/dfrontend/interpret.c, line 699.
cc1d: internal compiler error: Abort trap: 6

instead. The compiler hangs there and Ctrl-c was required.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=196

--- Comment #10 from Iain Buclaw <ibuclaw@gdcproject.org> ---
(In reply to John Colvin from comment #9)
> (In reply to Iain Buclaw from comment #8)
> > First, dustmite it.  This should be possible to reproduce in a cross compiler too.
> 
> Ok, after dustmite and a little manual cleaning:
> 
> //thread.d
> class ThreadError : Error
> {
>     this(string )
>     {
>         super(msg, file, line);
>     }
> }
> 
> void onThreadError() {
>     __gshared ThreadError = new ThreadError(null);
> }
> 
> $ /Users/johncolvin/Documents/GDC/objdir/./gcc/gdc -B/Users/johncolvin/Documents/GDC/objdir/./gcc/ -B/opt/gdc/x86_64-apple-darwin15.0.0/bin/ -B/opt/gdc/x86_64-apple-darwin15.0.0/lib/ -isystem /opt/gdc/x86_64-apple-darwin15.0.0/include -isystem /opt/gdc/x86_64-apple-darwin15.0.0/sys-include    -o thread.o -Wall -Werror -g -frelease -O2 -nostdinc -pipe -Wno-deprecated -I ../../../../../gcc-5.2.0/libphobos/libdruntime -I ../x86_64-apple-darwin15.0.0 -I ../  -c thread.d
> 
> <stdin>:109:1: error: unexpected token at start of statement
> .3614:
> ^
> 

OK, that's a good start.


> 
> If I comment out the super line, I get
> 
> thread.d:3:5: error: constructor thread.ThreadError.this no match for
> implicit super() call in constructor
>      this(string )
>      ^
> Assertion failed: (!fd->semantic3Errors), function ctfeCompile, file
> ../../gcc-5.2.0/gcc/d/dfrontend/interpret.c, line 699.
> cc1d: internal compiler error: Abort trap: 6
> 
> instead. The compiler hangs there and Ctrl-c was required.


That's reproducible with DMD.  Raised upstream: https://issues.dlang.org/show_bug.cgi?id=15172

-- 
You are receiving this mail because:
You are watching all bug changes.