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: [C++ patch] final? unit-at-a-time patch


> 
> On Saturday, Jun 21, 2003, at 18:31 US/Eastern, Jan Hubicka wrote:
> 
> >Hi,
> >this patch finally implements unit-at-a-time mode in C++.  I
> >regtested/bootstrapped it in both unit-at-a-time and 
> >non-unit-at-a-time mode.
> >In unit-at-a-time mode it still fails with testcase 
> >pretty1.c/pretty2.c, but I
> >can trigger same failure by deferring the functions via inline 
> >keyword, so it
> >is latent bug I will try to fix shortly too.
> >
> >The libstdc++ is about 11K smaller than without unit-at-a-time and 
> >compiles in
> >the same time.  I got similar results on eon benchmark (2% code size 
> >decrease)
> >and few other C++ programs I compiled so I think it can be enabled by 
> >default
> >at -O3 level unless we hit some inexpected problems. (I still expect
> >some but I will try to take care to fix these)
> >
> >OK for mainline?
> 
> 
> I would try it on a platform where there is no comdat support.
That sounds like good idea.   In theory it should just work as frontend
then should emit just static functions and unit-at-a-time should be just
happy about it but...
What platofrms do miss it?  Is -fno-weak supposed to acomplish this?

Honza
> 
> 
> Thanks,
> Andrew Pinski


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