Re: [PATCH]: Interprocedural detection of readonly andnon-addressable static variables

[Daniel Berlin <dberlin at dberlin dot org>]

On Wed, 2005-06-08 at 18:04 +0200, Gabriel Dos Reis wrote:
> Daniel Berlin <dberlin@dberlin.org> writes:
> 
> | This is part 1 (of 7) of the interprocedural work Kenny and I have been
> | doing for static variables, type-escape analysis, and pure/const
> | detection.
> | 
> | This is the scanning code that finds information about which static
> | variables are written/referenced/address taken in a given procedure.
> 
> Awesome!
> 
> | 2005-06-06  Kenneth Zadeck  <zadeck@naturalbridge.com>
> |        Danny Berlin <dberlin@dberlin.org>
> | 
> |    Interprocedual detection of readonly and non-addressable static
> |    variables.
> | 
> |    * ipa-reference.c, ipa-reference.h, ipa-utils.c, ipa-utils.h: new
> |    files.
> |    * common.opt: Added "fipa-reference".
> |    * opts.c: Added "flag_ipa_reference".
> |    * timevar.def: Added TV_IPA_REFERENCE.
> |    * tree-dfa.c (find_reference_vars) Addition of call to reset
> |    per function information.
> |    * tree-flow.h: Addition of reference_vars_info field for
> |    FUNCTION_DECLS.
> |    * tree-optimize.c: Added pass.
> |    * tree-pass.h: Added pass.
> | 
> | 
> | 
> | Bootstrapped and regtested on i686-pc-linux-gnu and powerpc-linux-gnu.
> | 
> | Okay for mainline?
> 
> I think it makes sense to put the core files in a seperate directory,
> say gcc/ipa.  That helps strenghting the GCC structure. We have too
> much of otherwise reasonably partitioned component files lumped
> together in gcc/. 

I'm not sure how to pull that off, however.

Is it as simple as prefixing the files with ipa/ in the Makefile?
Or did you want me to add a new makefile to that dir?

> 
> Thanks,
> 
> -- Gaby