On Wed, 2008-01-23 at 09:05 -0800, Linus Torvalds wrote:

quoted text
> So here's even a patch to get you started. Do this:
> 
> 	git revert 204ec841fbea3e5138168edbc3a76d46747cc987
> 
> and then use this appended patch on top of that as a starting point for 
> something that compiles and *possibly* works.
> 
> And no, I do *not* guarantee that this is right either! I have not tested 
> it or thought about it a lot, and S390 tends to be odd about some of these 
> things. In particular, I actually suspect that we should possibly do this 
> the same way we do
> 
> 	ptep_clear_flush_young()
> 
> except we would do "ptep_clear_flush_wrprotect()". So even though this is 
> a revert plus a simple patch to make it compile again (we've changed how 
> we do dirty bits), I think a patch like this needs testing and other 
> people like Nick and Peter to ack it.
> 
> Nick? Peter? Testing? Other comments?

It would need some addition piece to not call msync_interval() for
MS_SYNC, and remove the balance_dirty_pages_ratelimited_nr() stuff.

But yeah, this pte walker is much better. 

As for s390, I think they only differ on the dirty bit, and we should
not be touching that.

--
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/