| Message ID | 20180529211724.4531-6-josef@toxicpanda.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, May 29, 2018 at 05:17:16PM -0400, Josef Bacik wrote: > From: Tejun Heo <tj@kernel.org> > > For backcharging we need to know who the page belongs to when swapping > it out. > > Signed-off-by: Tejun Heo <tj@kernel.org> > Signed-off-by: Josef Bacik <jbacik@fb.com> > --- > mm/page_io.c | 10 ++++++++++ > 1 file changed, 10 insertions(+) > > diff --git a/mm/page_io.c b/mm/page_io.c > index a552cb37e220..61e1268e5dbc 100644 > --- a/mm/page_io.c > +++ b/mm/page_io.c > @@ -339,6 +339,16 @@ int __swap_writepage(struct page *page, struct writeback_control *wbc, > goto out; > } > bio->bi_opf = REQ_OP_WRITE | REQ_SWAP | wbc_to_write_flags(wbc); > +#if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP) > + if (page->mem_cgroup) { > + struct cgroup_subsys_state *blkcg_css; > + > + blkcg_css = cgroup_get_e_css(page->mem_cgroup->css.cgroup, > + &io_cgrp_subsys); > + bio_associate_blkcg(bio, blkcg_css); > + css_put(blkcg_css); > + } > +#endif This looks reasonable, but it probably warrants a helper function. bio_associate_blkcg_from_page() or something?
On Tue, May 29, 2018 at 05:17:16PM -0400, Josef Bacik wrote: > From: Tejun Heo <tj@kernel.org> > > For backcharging we need to know who the page belongs to when swapping > it out. > > Signed-off-by: Tejun Heo <tj@kernel.org> > Signed-off-by: Josef Bacik <jbacik@fb.com> > --- > mm/page_io.c | 10 ++++++++++ > 1 file changed, 10 insertions(+) > > diff --git a/mm/page_io.c b/mm/page_io.c > index a552cb37e220..61e1268e5dbc 100644 > --- a/mm/page_io.c > +++ b/mm/page_io.c > @@ -339,6 +339,16 @@ int __swap_writepage(struct page *page, struct writeback_control *wbc, > goto out; > } > bio->bi_opf = REQ_OP_WRITE | REQ_SWAP | wbc_to_write_flags(wbc); > +#if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP) > + if (page->mem_cgroup) { > + struct cgroup_subsys_state *blkcg_css; > + > + blkcg_css = cgroup_get_e_css(page->mem_cgroup->css.cgroup, > + &io_cgrp_subsys); > + bio_associate_blkcg(bio, blkcg_css); > + css_put(blkcg_css); > + } > +#endif So, this ignores the cases where bdev_write_page() is the one which does the writes. If my reading is correct, only brd, zram, btt and pmem implement bdev_ops->rw_page() and take bdev_write_page() path, so it shouldn't be a problem in majority of cases. I don't think we need to address ->rw_page() case right now but it might be a good idea to add a comment explaining the ommission. Thanks.
diff --git a/mm/page_io.c b/mm/page_io.c index a552cb37e220..61e1268e5dbc 100644 --- a/mm/page_io.c +++ b/mm/page_io.c @@ -339,6 +339,16 @@ int __swap_writepage(struct page *page, struct writeback_control *wbc, goto out; } bio->bi_opf = REQ_OP_WRITE | REQ_SWAP | wbc_to_write_flags(wbc); +#if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP) + if (page->mem_cgroup) { + struct cgroup_subsys_state *blkcg_css; + + blkcg_css = cgroup_get_e_css(page->mem_cgroup->css.cgroup, + &io_cgrp_subsys); + bio_associate_blkcg(bio, blkcg_css); + css_put(blkcg_css); + } +#endif count_swpout_vm_event(page); set_page_writeback(page); unlock_page(page);