Sure. I will make a patch to clean up all the occurrences of this usages later. Thanks. On Fri, Dec 6, 2019 at 5:27 PM Christoph Hellwig <hch@xxxxxxxxxxxxx> wrote: > > On Fri, Dec 06, 2019 at 04:55:42PM +0800, Liang Chen wrote: > > Same as cache device, the buffer page needs to be put while > > freeing cached_dev. Otherwise a page would be leaked every > > time a cached_dev is stopped. > > > > Signed-off-by: Liang Chen <liangchen.linux@xxxxxxxxx> > > --- > > drivers/md/bcache/super.c | 3 +++ > > 1 file changed, 3 insertions(+) > > > > diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c > > index 77e9869345e7..a573ce1d85aa 100644 > > --- a/drivers/md/bcache/super.c > > +++ b/drivers/md/bcache/super.c > > @@ -1275,6 +1275,9 @@ static void cached_dev_free(struct closure *cl) > > > > mutex_unlock(&bch_register_lock); > > > > + if (dc->sb_bio.bi_inline_vecs[0].bv_page) > > + put_page(bio_first_page_all(&dc->sb_bio)); > > Using bio_first_page_all in the put_page call, but open coding it > for the check looks rather strange. > > The cleanest thing here would be to just add a page pointer to the > cached device structure and use that instead of all the indirections.
