Re: ordered memory access

Manfred Spraul (manfreds@colorfullife.com)
Thu, 30 Sep 1999 12:11:36 +0000

Jes Sorensen wrote:
> atomic_foo() should be ordered, at least that the idea behind it as
> far as I understand.

That might be the idea, but that's not the implementation:
Pentium III documentation: (24319201.pdf)
> 7.2.2. Memory Ordering in the P6 Family Processors
> The P6 family processors also use a processor-ordered memory
> ordering model [...]
> 1. Reads can be carried out speculatively and in any order.

I think there is currently no implementation for "ordered read" except
rmb():
- multiple atomic_read()'s don't enforce read ordering.
- there are set_?mb() function, but there are no get_?mb() functions.

I've read that rmb() is expensive on RISC architectures, what about
adding "get_?mb()" functions?

--
	Manfred




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