[PATCH] Re: [RFC] SFTs on complex type

[Andrew Pinski]

On Dec 29, 2005, at 1:49 PM, Andrew Pinski wrote:

>
> On Dec 29, 2005, at 1:30 PM, Daniel Berlin wrote:
>
>> Uh, so, if that's a bug, why not fix it?
>>
>> 1. the MTAG_P check could go first, since no memory tag ever has
>> subvars.
>
> That seems like a good idea, maybe even better is don't ask for the 
> memory
> tags at all.  (I can provide a patch for that).

Here is the full patch (after Richard committed the fix for the second 
part
already).

OK? Bootstrap and tested on x86_64-linux-gnu and powerpc-darwin with no 
regressions.

Thanks,
Andrew Pinski

ChangeLog:
	* tree-ssa-alias.c (find_used_portions): Handle REALPART_EXPR
	and IMAGPART_EXPR.
	* tree-flow-inline.h (var_can_have_subvars): Handle complex types
	on non gimple variables.  Also add checks at the top for decls and
	mtags.
	* tree-ssa-structalias.c (push_fields_onto_fieldstack): Handle
	complex types.
	* tree-ssa-operands.c (parse_ssa_operands): Handle REALPART_EXPR
	and IMAGPART_EXPR for creating MUST_DEFs.
	(get_expr_operands): Handle SSA_NAME, STRUCT_FIELD_TAG, 
TYPE_MEMORY_TAG,
	and NAME_MEMORY_TAG separately from the DECLs.


-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: complex.diff.txt
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20060106/ce87967d/attachment.txt>