> I however doubt this will bring us back to the same performance of the > current spinlock version, as the real overhead should come out of > overscheduling in down_write ai anon_vma_link. Here an initially > spinning lock would help but that's gray area, it greatly depends on > timings, and on very large systems where a cacheline wait with many > cpus forking at the same time takes more than scheduling a semaphore > may not slowdown performance that much. So I think the only way is a > configuration option to switch the locking at compile time, then XPMEM > will depend on that option to be on, I don't see a big deal and this > guarantees embedded isn't screwed up by totally unnecessary locks on UP.
You have said this continually about a CONFIG option. I am unsure how that could be achieved. Could you provide a patch? Thanks, Robin _______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
