This is the mail archive of the gcc-patches@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: [Patch] Fix for PR middle-end/38059, Compile time regression


On Tue, 2008-12-02 at 19:53 +0100, Richard Guenther wrote:

> >
> > If I move the todo_flags_start check above the gate check then things
> > seem to work fine.  I don't know if this change is reasonable though.
> > Maybe we just need to always have a non-optional first pass.
> 
> Yeah.  Cleanest would be an empty pass that just has PROP_gimple_any in its
> properties it provides as the first pass.

So I tried this and it fixed the problem in verify_properties but now I
am running into segmentation violations in the lower_eh pass.  I think
this is because it has never seen empty catch clauses before as they
were getting removed by the 'useless' pass.

Given the various issues I have run into with completely removing the
useless pass I would like to re-propose my original patch which simply
removes the goto optimization from the useless pass and leaves the rest
of it alone.  Maybe the idea of completely removing the pass can be
addressed after the 4.4 branch is made.

Steve Ellcey
sje@cup.hp.com


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