On Wed, 27 Jun 2007, Nick Piggin wrote:
quoted text
> 
> I don't know why my unlock sequence should be that much slower? Unlocked
> mov vs unlocked add? Definitely in dumb micro-benchmark testing it wasn't
> twice as slow (IIRC).

Oh, that releasing "add" can be unlocked, and only the holder of the lock 
ever touches that field?

I must not have looked closely enough. In that case, I withdraw that 
objection, and the sequence-number-based spinlock sounds like a perfectly 
fine one.

Yes, the add will be slightly slower than the plain byte move, and the 
locked xadd will be slightly slower than a regular locked add, but 
compared to the serialization cost, that should be small. For some reason 
I thought you needed a locked instruction for the unlock too.

So try it with just a byte counter, and test some stupid micro-benchmark 
on both a P4 and a Core 2 Duo, and if it's in the noise, maybe we can make 
it the normal spinlock sequence just because it isn't noticeably slower.

In fact, I think a "incb <mem>" instruction is even a byte shorter than 
"movb $1,mem", and with "unlock" being inlined, that could actually be a 
slight _win_.

			Linus
-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/