On Fri, Jul 11, 2008 at 11:25:03PM +0000, Andrew Doran wrote: > On Fri, Jul 11, 2008 at 02:04:16PM +0100, David Brownlee wrote: > > On Fri, 11 Jul 2008, Andrew Doran wrote: > > > > >In its current form SA threading is a regressive proposition. Even if all > > >the remaining issues are addressed, what benefits would it bring over and > > >above 1:1 threading? > > > > As I understand it there are two factors in favour of keeping SA: > > > > - There was a report of a significant performance gain on single > > core embedded arm32 > > I have heard similar stories put about but I have not seen any test cases > posted. Without evidence I find it difficult to classify them as anything > other than myth based on myth. > > > - If we keep SA for 5.0 it allows people to run a 4.0 userland > > on a 5.0 kernel while upgrading (and makes COMPAT_40 useful). > > That's not something that we support. Admittedly there is no public > statement to that effect, as far as I am aware. I'm confused. Isn't something like this _exactly_ why we have COMPAT options? I'm also confused as to why it is so important that SA not come back, especially as an option (that can be turned off). Take care, Bill
Attachment:
pgpMO5RfIqJOR.pgp
Description: PGP signature