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: [lto][patch] Move the call to execute_all_ipa_transforms to cgraphunit.c


> > How this picture change in LTO? I would expect that RTL compilation
> > queue is pretty much unchanged.
> 
> For each function it is. I am just reordering some passes, and because
> of that I had to add a call fixup_cfg. It introduced that strange
> regression I mentioned. No idea why.

Well, if you reorder the passes, you always need fixup_cfg to be first
in the block of local passes containing the RTL queue.  That way you
should not get any CFG verification failures.
I don't see how resonably you can end up needing 2 fixup_cfg's unless
you are adding new block of local passes somewhere in the queue.

Or simply disable the pass for time being if you want...

Honza
> 
> > Honza
> >
> 
> 
> Cheers,
> -- 
> Rafael Avila de Espindola
> 
> Google | Gordon House | Barrow Street | Dublin 4 | Ireland
> Registered in Dublin, Ireland | Registration Number: 368047


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