Messages in this thread | | | Date | Sun, 8 Oct 2000 08:27:24 -0700 (PDT) | From | Linus Torvalds <> | Subject | Re: VM in v2.4.0test9 |
| |
On Wed, 4 Oct 2000, Rik van Riel wrote: > > The potential for this bug has been around since 2.3.51, when > different balance_ratios for different zones became possible.
No, the bug has been around since your VM.
You must NOT depend on some global "freepages" thing.
You MUST do your freepages comparisons on a per-zone basis.
Why? Think NUMA. The global freepages number is NOT USABLE. Never will be. Because it's fundamentally a non-valid number to use - it has nothing to do with any reality, and never will.
This is exactly my argument and beef with Andreas zone patches. They simply cannot work in the end because they use global information that should not be used.
Rik, why do you think the old code used the per-zone information?
Don't do this patch. Fix it for real instead. You can use the global amount of free memory for heuristics, but you MUST NOT use it for deciding to not page stuff out etc - because you will _always_ end up with cases where you decide that you globally have enough memory even though locally there is a memory shortage (or the other way around - you end up thinking that there is a global memory shortage even if all local memories are plenty fine, so you end up trying to swap stuff out when it shouldn't be done).
Linus
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/
| |