we're talking about two different things. I think you're talking about
simultaneous writes, which are serialized by the page lock. that's OK.
the refcount of a page in the pagecache, clean or dirty is 1 (when it's not
being currently read/written and not mapped by any process). shrink_mmap
can recycle it if it's clean. the problem is with the way shrink_mmap
checks if the page is dirty - it checks if any of page->buffers is busy.
if page->buffers is NULL, it just assumes the page is clean and recycles it.
stephen suggested bumping up the refcount to avoid this problem, but I don't
quite see where this can be done.
ganesh
-
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/