----- Message d'origine ----
De : Andi Hellmund <mail@xxxxxxxxxxxxxxxx>
à : charfi asma <charfiasma@xxxxxxxx>
Cc : Philip Herron <redbrain@xxxxxxxxxxx>; Ian Lance Taylor <iant@xxxxxxxxxx>
Envoyà le : Sam 2 octobre 2010, 18h 30min 16s
Objet : Re: Tr : Re : Re : Re : add a new gcc fe based on gcalc or sample_fe
On 09/29/2010 11:56 AM, charfi asma wrote:
>
>
>> ----- Message d'origine ----
>>
>
>> De : Ian Lance Taylor<iant@xxxxxxxxxx>
>> Ã : charfi asma<charfiasma@xxxxxxxx>
>> Cc : gcc-help@xxxxxxxxxxx
>> Envoyà le : Lun 27 septembre 2010, 18h 44min 35s
>> Objet : Re: Re : add a new gcc fe based on gcalc or sample_fe
>>
>> charfi asma<charfiasma@xxxxxxxx> writes:
>>
>>
>>
>>> I tried to modify the language by exporting LANG env var to en_US.UTF-8
>>> I even add those lines in my .bash_profile file and reboot my computer:
>>>
>>> export LC_ALL=en_US.UTF-8
>>> export LANG=en_US.UTF-8
>>> export LC_CTYPE=en_US.UTF-8
>>>
>>> [root@is010178 charfi]# echo $LANG
>>> en_US.UTF-8
>>>
>>> but I still have errors in french
>>>
>>>
>> Odd. Is LC_MESSAGES set in the environment?
>>
>>
>>
>>
>>> 1. when I git clone the gcc-dev directory and build the UML front end for
the
>>> first time I get this error:
>>>
>>>
>>> gcc -g -O2 -DIN_GCC -W -Wall -Wwrite-strings -Wcast-qual
>>> -Wstrict-prototypes
>>>
>>> -Wmissing-prototypes -Wmissing-format-attribute -pedantic -Wno-long-long
>>> -Wno-variadic-macros -Wno-overlength-strings -Wold-style-definition
>>> -Wc++-compat
>>>
>>> -fno-common -DHAVE_CONFIG_H -DGENERATOR_FILE -o build/gengtype \
>>> build/gengtype.o build/errors.o build/gengtype-lex.o
>>> build/gengtype-parse.o ../build-i686-pc-linux-gnu/libiberty/libiberty.a
>>> make[2]: *** Pas de rÃgle pour fabriquer la cible  ../../gcc-dev/gcc/uml/u
>>>
> Â,
>
>>>
>>>
>>
>>
>>> nÃcessaire pour  s-gtype Â. ArrÃt.
>>>
>>> --> In english it becomes: no rule to build the target Â
>>> ../../gcc-dev/gcc/uml/u
>>>
>>>  required for  s-gtype Â. Stop
>>>
>>>
>> This seems to be looking for a file gcc/uml/u. That seems odd. Is that
>> exactly what it says? Does uml/u appear in the Makefile?
>>
>> I don't know anything about the UML frontend and it is not part of
>> standard gcc; have you tried asking the author?
>>
>>
>>
> Hey Asma,
>
>> May be I should mention that I do not use any lexer or parser, is it normal
>> that
>> ( in the error message) gcc seems looking for build/gengtype.o build/errors.o
>> build/gengtype-lex.o ?
>>
>>
> I have no idea what's wrong with your code. Maybe, you could send me
> your code, but please only the front-end and not the complete gcc source
> tree :-) I could then try to build your front-end on my system and check
> what's wrong...
>
> Hello,
>
> you find attached my uml front end (even if I am so dummy that I wanted to
send
> you all the gcc tree, I can not because it is too big ;)
>
> I tried also to follow the sample_fe (not the gcalc) cause sample_fe did not
> require any lexer or parser but I get this error.
>
> In file included from ../../gcc-dev/gcc/sample_fe/gsfe.c:3:
> ../../gcc-dev/gcc/gcc.h:45: attention : âstruct cl_decoded_optionâ declared
> inside parameter list
> ../../gcc-dev/gcc/sample_fe/gsfe.c:9: erreur: two or more data types in
> declaration specifiers
> ../../gcc-dev/gcc/sample_fe/gsfe.c:9: erreur: conflicting types for
> âlang_specific_driverâ
> ../../gcc-dev/gcc/gcc.h:44: note: previous declaration of
>âlang_specific_driverâ
>
>
> was here
> make[2]: *** [sample_fe/gsfe.o] Erreur 1
> make[2]: leave the directory  /export/home/charfi/Bureau/build_sfe2/gcc Â
> make[1]: *** [all-gcc] Erreur 2
> make[1]: leave the directory  /export/home/charfi/Bureau/build_sfe2 Â
> make: *** [all] Erreur 2
>
>
> of couse I do
> - change the GTY in the driver to suit the change in GCC4.6 (GTY must be
> declared before identifier)
>
> - change the prototype of lang_specific_driver ( struct cl_decoded_option
> **in_decoded_options ATTRIBUTE_UNUSED, unsigned int *in_decoded_options_count
> ATTRIBUTE_UNUSED, int *in_added_libraries ATTRIBUTE_UNUSED) rather than
>
> lang_specific_driver (int *in_argc, const char *const **in_argv, int
> *in_added_libraries ATTRIBUTE_UNUSED)
>
> to recreate the error, so can try to compile the sample_fe in the gcc-dev
>(4.6).
>
> thanks
>
> Asma
>
>
>
Hey Asma,
please find attached a version of your uml front-end which "works" with
4.6! By saying "works", I just mean that it compiles correctly, but I
didn't look closely at the logic of your code. So, please give it a try
in your Cygwin environment ...
Best regards,
Andi
Hello Andi,
thank you for your help, I build this fe in my environment (but I am not using
Cygwin, I am using the shell of my mandriva 2010) and it works (configure, make
make install without errors ;)
but when I compile a file "test.uml" using the guml driver, I get this error:
I al trying only to transform a main function to generic (In my file.uml I put
only: int main() {return 0;}
[charfi@is010178 test_gcalc]$ guml test.uml -fdump-tree-all
Analyzing compilation unit
Performing interprocedural optimizations
<*free_lang_data> <visibility> <early_local_cleanups> <whole-program>
<inline>Assembling functions:
uml1: internal compiler error: in cgraph_mark_reachable_node, at cgraph.c:1687
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
I think that there is sth missing that said that my node is reachable, the error
is occuring in the function "cgraph_mark_reachable_node"
any idea ?
thanks
Asma
