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: CFG transparent inlining


> > On Fri, 2005-04-29 at 14:57 +0100, Andrew Haley wrote:
> > > Daniel Berlin writes:
> > >  > On Fri, 2005-04-29 at 14:48 +0100, Andrew Haley wrote:
> > >  > > Jan Hubicka writes:
> > >  > >  > 
> > >  > >  > 
> > >  > >  > Bootstrapped/regtested i686-pc-gnu-linux,x86-64-linux,ppc-linux and ia64-linux
> > >  > >  > (last two only in earlier incarnation of patch before resolving some further
> > >  > >  > conflicts on moving mainline)
> > >  > > 
> > >  > > Tested with large code bases? 
> > >  > 
> > >  > SPEC, all of int and FP.
> > > 
> > > I can't remember what is in SPEC.  Has libc been tested, or the kernel?
> > 
> > Not by me :)
> > (My testing has mainly been a side-effect of tree-profiling work :P)
> 
> I also built some C++ packages (mozilla, lyx) and tested SPEC in several
> different setups on tree-profiling (such as --combine with inlining that
> enables quite a lot of transformations).  The mainline patch is not 100%
> identical, but I am going to try out glibc for now.
> Overall I believe that basic inlining works fluently now, side cases such
> as warnings or exception handling pops up, but hopefully our testsuite
> covers these pretty well.

So testing of glibc/kernel wasn't particularly interesting.  glibc fails
with syntax error in the same place as it does on vanilla mainline,
kernel builds and crash booting same place as mainline one does.  I am
investigating what is going wrong...

Honza
> 
> Honza
> > 


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