On Thu, Sep 11, 2014 at 04:53:54PM -0400, Dan Streetman wrote: > Update ordering of a changed zspage in its fullness group LRU list, > even if it has not moved to a different fullness group. > > This is needed by zsmalloc shrinking, which partially relies on each > class fullness group list to be kept in LRU order, so the oldest can > be reclaimed first. Currently, LRU ordering is only updated when > a zspage changes fullness groups.
Just something I saw. fix_fullness_group() is called from zs_free(), which means that removing an object from a zspage moves it to the front of the LRU. Not sure if that is what we want. If anything that makes it a _better_ candidate for reclaim as the zspage is now contains fewer objects that we'll have to decompress and writeback. Seth > > Signed-off-by: Dan Streetman <ddstr...@ieee.org> > Cc: Minchan Kim <minc...@kernel.org> > --- > mm/zsmalloc.c | 10 ++++------ > 1 file changed, 4 insertions(+), 6 deletions(-) > > diff --git a/mm/zsmalloc.c b/mm/zsmalloc.c > index fedb70f..51db622 100644 > --- a/mm/zsmalloc.c > +++ b/mm/zsmalloc.c > @@ -467,16 +467,14 @@ static enum fullness_group fix_fullness_group(struct > zs_pool *pool, > BUG_ON(!is_first_page(page)); > > get_zspage_mapping(page, &class_idx, &currfg); > - newfg = get_fullness_group(page); > - if (newfg == currfg) > - goto out; > - > class = &pool->size_class[class_idx]; > + newfg = get_fullness_group(page); > + /* Need to do this even if currfg == newfg, to update lru */ > remove_zspage(page, class, currfg); > insert_zspage(page, class, newfg); > - set_zspage_mapping(page, class_idx, newfg); > + if (currfg != newfg) > + set_zspage_mapping(page, class_idx, newfg); > > -out: > return newfg; > } > > -- > 1.8.3.1 > -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
