On Sun, 17 Dec 2006, Linus Torvalds wrote: > > So we should probably do a "wait_for_page()" in do_no_page()? > > Or maybe only do it for write accesses (since we don't really care about > getting mapped readably)? If so, we need to do it in the write case of > do_no_page() _and_ in the do_wp_page() case. Hmm?
I think we discussed doing exactly this at some earlier time, actually, just to try to throttle people who do lots of page dirtying. Maybe we even do it somewhere, but I tried to see it, and in the normal "nopage()" routine we very much try to _avoid_ locking the page (ie if it's marked PageUptodate() we'll return it whether locked or not). Which is fine - especially for readers, there really isn't any reason to ever delay them getting access to a page just because it's locked for write-out or something (once it's mapped, they'll have access to it regardless of any locked state in the kernel anyway). So I don't actually see any serialization at all that would keep a random page from being paged back in. Linus - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/