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: Remove unit-at-a-time checks and flag_inline_trees


> On 7/23/08 7:32 AM, Jan Hubicka wrote:
> 
> >Bootstrapped/regtested i686-linux, will commit it tomorrow if there are no 
> >complains.
> 
> This is great!  I only have a few minor grammar nits.

Thanks for corrections (and moral support ;)!  Some of the overall
cgraph comments will need rewrite, I will look into that after the
DECL_INLINE mess is cleaned up.
> 
> Hmm?  Shouldn't we still check optimize_size?  Or have you moved this 
> elsewhere?

Early inliner always inline for size, because the overall unit growth
limits can be resonably checked before we have idea how the whole unit
is large.  This is aimed to get rid of the C++ wrappers as early as
possible and early optimizers have some real work to do (early
optimizing empty wrappers hardly helps).
The speed inlining was really used only in non-unit-at-a-time
> 
> 
> 
> Thanks for doing this.  I'm glad that we are finally getting rid of 
> non-unit-at-a-time.  I think this will create a few merge conflicts with 
> the tuples branch, but they shouldn't be hard to fix.

I hope it won't be that bad.  It almost exclusively tough the frontends.
Perhaps callgraph building code, but it is very simple.

Honza
> 
> 
> Diego.


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