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: TREE_ADDRESSABLE problem


> On Thu, Nov 20, 2003 at 11:43:11PM +0100, Jan Hubicka wrote:
> > I probably can it manage to copy the argument list/return value too and
> > inject the new values into structure once I am finished, better ideas?
> 
> How about we NOT have the front-end manage this flag at all, and
> compute it during optimization always?
This is what I had in mind when I was suggesting that we can recompute
it on start of optimization (remove_useless_* can easilly take care of
this bit too)

However we need to ensure that VAR_DECLs won't be modified in some other
ways by the optimizers?  Is this the case or verify_stmts just randomly
hit one of multiple changes we do into trees?

Perhaps we want do both - duplicate var_decls and compute addressable
flag as part of remove_useless_* or rewrite_into_ssa?

Honza
> 
> 
> r~


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