[cfe-dev] computed goto/labels as values in interpreters

[Carsten Mattner]

On Sat, Oct 13, 2012 at 6:43 PM, Carsten Mattner
<carstenmattner at gmail.com> wrote:
> On Sat, Oct 13, 2012 at 2:55 AM, Eli Friedman <eli.friedman at gmail.com> wrote:
>> On Thu, Oct 11, 2012 at 11:36 AM, Carsten Mattner
>> <carstenmattner at gmail.com> wrote:
>>> On Wed, Oct 3, 2012 at 12:37 PM, Carsten Mattner
>>> <carstenmattner at gmail.com> wrote:
>>>> On Thu, Sep 27, 2012 at 4:10 PM, Carsten Mattner
>>>> <carstenmattner at gmail.com> wrote:
>>>>> On Sat, Sep 22, 2012 at 5:28 PM, Carsten Mattner
>>>>> <carstenmattner at gmail.com> wrote:
>>>>>> On Sat, Sep 22, 2012 at 4:51 PM, James Molloy <james at jamesmolloy.co.uk> wrote:
>>>>>>> Hi Carsten,
>>>>>>>
>>>>>>> He's asking for preprocessed source. If you run clang -cc1 in '-E'
>>>>>>> mode instead of the normal '-c', it will run only the preprocessor and
>>>>>>> emit source that is easy to compile elsewhere without dependencies.
>>>>>>
>>>>>> Mailed it privately to Eli, Rafael and you as it's a large file
>>>>>> and may contain info not meant for the public.
>>>>>>
>>>>>> Thanks for the help.
>>>>>
>>>>> Sending the files again, now publicly. Can you test and try to fix
>>>>> the infinite loop now?
>>>>>
>>>>> Any news on the slow and memory intensive compile of beam_emu.c
>>>>> mentioned earlier?
>>>>
>>>> Was somebody able to inspect the issues?
>>>
>>> Anybody?
>>
>> I took a quick look at wxe_funcs.cpp; it doesn't hang for me with
>> trunk clang.  -ftime-report does show us spending an insane amount of
>> time in the loop-unswitch pass, and it's using a lot of memory (around
>> 500 megabytes?), but it does complete successfully.
>
> Thanks Eli, I did another test today and noticed two things.
>
> When I was just building erlang via 'make all' wxe_funcs.cpp
> didn't seem to take long to compile but this step in the erlang/lib/wx
> build process didn't seem to end:
> clang -shared -fPIC  i686-pc-linux-gnu/gl_funcs.o
>  i686-pc-linux-gnu/egl_impl.o -lGL -lGLU -o ../priv/erl_gl.so
> Then I interrupted the build process and touched wxe_funcs.cpp and
> all other C and C++ files lib/wx and ran '/usr/bin/time make' in the top
> directory. This time around wxe_funcs.cpp didn't finish quickly as
> before. I can't reproduce the quick compile of wxe_funcs.cpp in such
> a tree but maybe it's reproducible when building lib/wx for the first
> time. This sounds random and non-conclusive but I'm just writing up
> what I saw.
>
> So the timing for letting lib/wx finish is:
> 2543.73user 3.62system 42:30.43elapsed 99%CPU (0avgtext+0avgdata
> 347264maxresident)k
> 45904inputs+39360outputs (4major+1975612minor)pagefaults 0swaps
>
> Eli, yes it looks like it does at some point finish to build.
> I can also confirm that the resulting build does work with the
> wxWidgets binding including OpenGL functionality.
>
> So I'd say once clang does finish building the "hard to build"
> parts of Erlan/OTP the result looks basically functional.
>
> What can we do to fix the build time and space use?
>
> Do you have an idea what could have made the first build of
> lib/wx process wxe_funcs.cpp quickly but stall at building erl_gl.so?
> That looks similar to the long 'ar ...' step in Erlang's erts/
> directory.
>
> llvm/clang version info:
> clang version 3.2
>  (http://llvm.org/git/clang.git e5a934d3c840872d58724383a83443ed38f1d831)
>  (http://llvm.org/git/llvm.git  c2fcf1a67126a079e34f04bd09a50ae85fdd9469)

Any news on these issues? If nobody is working or planning to work
on it, I'll unsubscribe from the list. Have been hanging around to
track the Erlang build compatibility.