On Wed, Mar 11, 2015 at 12:25 PM, Konstantin Khlebnikov
<khlebnikov@xxxxxxxxxxxxxx> wrote:
> This patch conflicts with my cleanup which is already in mm tree:
> ("page_writeback: clean up mess around cancel_dirty_page()")
> Nothing nontrivial but I've killed cancel_dirty_page() and replaced
> it which account_page_cleaned() symmetrical to account_page_dirtied().
Fair enough. I'll rebase.
> I think this accounting can be done without mem_cgroup_begin_page_stat()
> All page cleaning happens under page is lock.
> Some dirtying is called without page-lock when kernel moves
> dirty status from pte to page, but in this case acconting happens
> under mapping->tree_lock.
>
> Memcg already locks pages when moves them between cgroups,
> maybe it could also lock mapping->tree_lock?
Good suggestion. I'll try it out and report back.
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
