Its nesting hundreds of spinlocks (255+) make the preempt debug code
unhappy, it also causes fun latencies when you do start KVM/GRU
Although arguably that's the least convincing reason to do all this its
the one that got me to actually compose this series -- I really should
write a new leader..
Least Convincing is a good description...
Tuning operations which only happen once is not very interesting,
especially if that affects
fast paths.
Making all this preemptible also allows making the whole mmu_gather
thing preemptible, which is something we've wanted to do for a long
while, it also allows XPMEM or whatever that thing was called (Andrea
knows) and of course, it moves a part of -rt upstream.
I thought the reason for the preempt off inside the mmu gather region was
to stay on the same CPU for local/remote flushes. How would it change that?
If we decide to keep patch 24, it also simplifies the truncate path
quite a bit.
That sounds like a good thing. Making truncate simpler is always good.
-Andi
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxxx For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>