This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: entry point of gimplification


On Tue, Jan 5, 2010 at 8:44 PM, Diego Novillo <dnovillo@google.com> wrote:
> On 1/4/10 14:57 , sandeep soni wrote:
>
>> I want to know what is the entry point to the gimplification pass? and
>> given a function body which are the functions in the gcc source that
>> convert the body into equivalent gimple statements?
>
> This is controlled from the callgraph manager. ?You need to start
> looking at cgraphunit.c:cgraph_analyze_functions. ?It traverses the
> cgraph queue converting each node into gimple (the call to
> cgraph_analyze_function).
>
>> Also is there a way in which i can selectively step through the
>> execution of the source related to this?
>
> Yes. ?You need to debug cc1/cc1plus and set a breakpoint in
> cgraph_analyze_functions. ?The function debug_cgraph() will show you the
> callgraph at that point.
>
>
> Diego.
>

Hi,

I recently read in the internal documentation of gcc that the main
entry point to the gimplification pass is the function defined in
gimplify.c: gimplify_function_tree function.Is there something which I
m missing?

plus can someone point to me to of any way in which I can achieve my
aim of speculatively parallelizing the code at runtime?
chiefly I wanna know if gcc supports it ?
(If not) would it be possible to do?

-- 
cheers
sandy


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]