Messages in this thread | | | Date | Sat, 28 Sep 2013 21:13:39 +0200 | From | Andi Kleen <> | Subject | Re: [PATCH] rwsem: reduce spinlock contention in wakeup code path |
| |
> Of course, since then, we may well have screwed things up and now we > sleep under it, but I still really think it was a mistake to do it in > the first place. > > So if the primary reason for this is really just that f*cking anon_vma > lock, then I would seriously suggest: > > - turn it back into a spinlock (or rwlock_t, since we subsequently > separated the read and write paths)
Yes please. spinlocks/rwlocks have so much nicer performance behavior than rwsems/mutexes (which noone seems to fully understand)
We had also significant performance regressions from every such spinning->sleeping change in the VM (this was just the latest)
And afaik anon_vma is usually hold short.
-Andi
-- ak@linux.intel.com -- Speaking for myself only.
| |