[Bug 1443] PowerPC in_cksum() requires optimization

bugzilla-daemon at rtems.org bugzilla-daemon at rtems.org
Wed Aug 31 14:18:16 UTC 2011


https://www.rtems.org/bugzilla/show_bug.cgi?id=1443

--- Comment #5 from dufault at hda.com 2011-08-31 09:18:14 CDT ---
(In reply to comment #4)
> (In reply to comment #3)
> > (In reply to comment #2)
> > > This version would introduce the first
> > (Wasabi Systems BSD advertising clause)
> > > > 
> > > in the cpukit/libnetworking.  I don't think it is good to have another
> > > advertising clause in the system.  Maybe we can fix the current version or use
> > > the current FreeBSD code?  We should also do some performance tests since this
> > > code is critical.
> > 
> > When I reported the bug I tried FreeBSD and it didn't work either.
> How tracking the bug down and fixing it? One thing I know, is much of the
> powerpc stuff being fairly careless about types.

It's the inline assembler.  I'm sure the constraints are wrong.  I don't
understand it, I don't have time to learn it.
> 
(Not continuing the advertising discussion, I think everyone agrees we're
better without it.)
> 
> > 2. Anyone with time to look at this should look at the DragonflyBSD one. 
> As all of the RTEMS networking code is based on FreeBSD. I am not keen on
> seeing other niche OSes code being adopted.

It's a better design for in_cksum(), a small piece of code with a well-defined
interface requiring optimization frequently replaced with different
implementations on different architectures.  The DragonflyBSD one is a better
basis for someone needing to add a faster machine specific implementation where
one isn't already available with the appropriate license.

However, I should not have brought this up in the context of this bug, so let's
save any argument for another time.

Maybe Til or someone else who already groks GCC assembler constraints will have
time to fix the in-line assembler, apparently no one else is bothered by this. 
It only kicks in when compiled without optimization.

-- 
Configure bugmail: https://www.rtems.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching all bug changes.



More information about the bugs mailing list