Some basic/beginner questions

Some basic/beginner questions

[nico]

Hi,

I want to get a dump of the whole data structure of my sourcecode  
after (all) target independent optimization.
I think the best point to do this is after 'pass_del_ssa'. But I'm  
not sure.
At this point the internal representation of the data structure is  
'generic' - is this right?
I think one dump switch ('fdump-tree-optimized') is very close that  
what I want.
Which of the options is the best way to have a look at the data  
structure? I don't want have a look at debug-informations etc. I only  
want to see control flow, data flow, types and such things.
Another problem is to find in the gcc-sourcecode where this dump  
happens. A hint from this list was to have a look at the tree- 
dump.c . But I think the dumps which are implemented there, based on  
the 'tree-dump-index' in 'tree.h'. And there I can only found a few  
of the possible dump-flags. And no one of them is 'optimized'.

To much questions - I know.

Perhaps somebody recommend me some further readings.

Ciao and thanks,
Nico

Re: Some basic/beginner questions

[Diego Novillo]

On Tue, Jun 28, 2005 at 04:11:24PM +0200, nico wrote:

> I want to get a dump of the whole data structure of my sourcecode  
> after (all) target independent optimization.
> 
Dump flags and switches are dynamically enabled by the pass
manager.  Take a look at tree-optimize.c:init_tree_optimization_passes

Dumps are controlled by TODO_dump_func.  See tree-optimize.c:execute_todo
to follow what happens when the flag is enabled.  The global
variables dump_file and dump_flag tell the passes whether dumps
are enabled.  If you browse a few passes, you'll see that they
all predicate dump actions with 'if (dump_file ...)'.


Diego.