Re: [PATCH V5 06/10] Per-memcg background reclaim.

[Ying Han <yinghan@google.com>]

[-- Attachment #1: Type: text/plain, Size: 10432 bytes --]

On Sun, Apr 17, 2011 at 8:51 PM, Minchan Kim <minchan.kim@gmail.com> wrote:

> On Sat, Apr 16, 2011 at 8:23 AM, Ying Han <yinghan@google.com> wrote:
> > This is the main loop of per-memcg background reclaim which is
> implemented in
> > function balance_mem_cgroup_pgdat().
> >
> > The function performs a priority loop similar to global reclaim. During
> each
> > iteration it invokes balance_pgdat_node() for all nodes on the system,
> which
> > is another new function performs background reclaim per node. After
> reclaiming
> > each node, it checks mem_cgroup_watermark_ok() and breaks the priority
> loop if
> > it returns true.
> >
> > changelog v5..v4:
> > 1. remove duplicate check on nodes_empty()
> > 2. add logic to check if the per-memcg lru is empty on the zone.
> > 3. make per-memcg kswapd to reclaim SWAP_CLUSTER_MAX per zone. It make
> senses
> > since it helps to balance the pressure across zones within the memcg.
> >
> > changelog v4..v3:
> > 1. split the select_victim_node and zone_unreclaimable to a seperate
> patches
> > 2. remove the logic tries to do zone balancing.
> >
> > changelog v3..v2:
> > 1. change mz->all_unreclaimable to be boolean.
> > 2. define ZONE_RECLAIMABLE_RATE macro shared by zone and per-memcg
> reclaim.
> > 3. some more clean-up.
> >
> > changelog v2..v1:
> > 1. move the per-memcg per-zone clear_unreclaimable into uncharge stage.
> > 2. shared the kswapd_run/kswapd_stop for per-memcg and global background
> > reclaim.
> > 3. name the per-memcg memcg as "memcg-id" (css->id). And the global
> kswapd
> > keeps the same name.
> > 4. fix a race on kswapd_stop while the per-memcg-per-zone info could be
> accessed
> > after freeing.
> > 5. add the fairness in zonelist where memcg remember the last zone
> reclaimed
> > from.
> >
> > Signed-off-by: Ying Han <yinghan@google.com>
> > ---
> >  mm/vmscan.c |  157
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  1 files changed, 157 insertions(+), 0 deletions(-)
> >
> > diff --git a/mm/vmscan.c b/mm/vmscan.c
> > index 06036d2..39e6300 100644
> > --- a/mm/vmscan.c
> > +++ b/mm/vmscan.c
> > @@ -47,6 +47,8 @@
> >
> >  #include <linux/swapops.h>
> >
> > +#include <linux/res_counter.h>
> > +
> >  #include "internal.h"
> >
> >  #define CREATE_TRACE_POINTS
> > @@ -111,6 +113,8 @@ struct scan_control {
> >         * are scanned.
> >         */
> >        nodemask_t      *nodemask;
> > +
> > +       int priority;
> >  };
> >
> >  #define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))
> > @@ -2631,11 +2635,164 @@ static void kswapd_try_to_sleep(struct kswapd
> *kswapd_p, int order,
> >        finish_wait(wait_h, &wait);
> >  }
> >
> > +#ifdef CONFIG_CGROUP_MEM_RES_CTLR
> > +/*
> > + * The function is used for per-memcg LRU. It scanns all the zones of
> the
> > + * node and returns the nr_scanned and nr_reclaimed.
> > + */
> > +static void balance_pgdat_node(pg_data_t *pgdat, int order,
> > +                                       struct scan_control *sc)
> > +{
> > +       int i;
> > +       unsigned long total_scanned = 0;
> > +       struct mem_cgroup *mem_cont = sc->mem_cgroup;
> > +       int priority = sc->priority;
> > +       enum lru_list l;
> > +
> > +       /*
> > +        * This dma->highmem order is consistant with global reclaim.
> > +        * We do this because the page allocator works in the opposite
> > +        * direction although memcg user pages are mostly allocated at
> > +        * highmem.
> > +        */
> > +       for (i = 0; i < pgdat->nr_zones; i++) {
> > +               struct zone *zone = pgdat->node_zones + i;
> > +               unsigned long scan = 0;
> > +
> > +               for_each_evictable_lru(l)
> > +                       scan += mem_cgroup_zone_nr_pages(mem_cont, zone,
> l);
> > +
> > +               if (!populated_zone(zone) || !scan)
> > +                       continue;
>
> Do we really need this double check?

Isn't only _scan_ check enough?
>

yes. will change on next post.


> And shouldn't we consider non-swap case?
>

good point. we don't need to count the anon lru in non-swap case. A new
function will be added to count the memcg_zone_reclaimable per zone.


>
> > +
> > +               sc->nr_scanned = 0;
> > +               shrink_zone(priority, zone, sc);
> > +               total_scanned += sc->nr_scanned;
> > +
> > +               /*
> > +                * If we've done a decent amount of scanning and
> > +                * the reclaim ratio is low, start doing writepage
> > +                * even in laptop mode
> > +                */
> > +               if (total_scanned > SWAP_CLUSTER_MAX * 2 &&
> > +                   total_scanned > sc->nr_reclaimed + sc->nr_reclaimed /
> 2) {
> > +                       sc->may_writepage = 1;
>
> I don't want to add more random write any more although we don't have
> a trouble of real memory shortage.
>


> Do you have any reason to reclaim memory urgently as writing dirty pages?
> Maybe if we wait a little bit of time, flusher would write out the page.
>

We would like to reduce the writing dirty pages from page reclaim,
especially from direct reclaim. AFAIK, the try_to_free_mem_cgroup_pages()
still need to write dirty pages when there is a need. removing this from the
per-memcg kswap will only add more pressure to the per-memcg direct reclaim,
which seems to be worse. (stack overflow as one example which we would like
to get rid of)


>
> > +               }
> > +       }
> > +
> > +       sc->nr_scanned = total_scanned;
> > +       return;
>
> unnecessary return.
>
> removed.


> > +}
> > +
> > +/*
> > + * Per cgroup background reclaim.
> > + * TODO: Take off the order since memcg always do order 0
> > + */
> > +static unsigned long balance_mem_cgroup_pgdat(struct mem_cgroup
> *mem_cont,
> > +                                             int order)
> > +{
> > +       int i, nid;
> > +       int start_node;
> > +       int priority;
> > +       bool wmark_ok;
> > +       int loop;
> > +       pg_data_t *pgdat;
> > +       nodemask_t do_nodes;
> > +       unsigned long total_scanned;
> > +       struct scan_control sc = {
> > +               .gfp_mask = GFP_KERNEL,
> > +               .may_unmap = 1,
> > +               .may_swap = 1,
> > +               .nr_to_reclaim = SWAP_CLUSTER_MAX,
> > +               .swappiness = vm_swappiness,
> > +               .order = order,
> > +               .mem_cgroup = mem_cont,
> > +       };
> > +
> > +loop_again:
> > +       do_nodes = NODE_MASK_NONE;
> > +       sc.may_writepage = !laptop_mode;
>
> I think it depends on urgency(ie, priority, reclaim
> ratio(#reclaim/#scanning) or something), not laptop_mode in case of
> memcg.
> As I said earlier,it wold be better to avoid random write.
>

I agree that we would like to avoid it. but not sure if we should remove it
here, since it add more pressure to the direct reclaim case.

>
> > +       sc.nr_reclaimed = 0;
> > +       total_scanned = 0;
> > +
> > +       for (priority = DEF_PRIORITY; priority >= 0; priority--) {
> > +               sc.priority = priority;
> > +               wmark_ok = false;
> > +               loop = 0;
> > +
> > +               /* The swap token gets in the way of swapout... */
> > +               if (!priority)
> > +                       disable_swap_token();
> > +
> > +               if (priority == DEF_PRIORITY)
> > +                       do_nodes = node_states[N_ONLINE];
> > +
> > +               while (1) {
> > +                       nid = mem_cgroup_select_victim_node(mem_cont,
> > +                                                       &do_nodes);
> > +
> > +                       /* Indicate we have cycled the nodelist once
>
> Fix comment style.
>

Fixed.

>
> > +                        * TODO: we might add MAX_RECLAIM_LOOP for
> preventing
> > +                        * kswapd burning cpu cycles.
> > +                        */
> > +                       if (loop == 0) {
> > +                               start_node = nid;
> > +                               loop++;
> > +                       } else if (nid == start_node)
> > +                               break;
> > +
> > +                       pgdat = NODE_DATA(nid);
> > +                       balance_pgdat_node(pgdat, order, &sc);
> > +                       total_scanned += sc.nr_scanned;
> > +
> > +                       /* Set the node which has at least
>
> Fix comment style.
>
> Fixed.


> > +                        * one reclaimable zone
> > +                        */
> > +                       for (i = pgdat->nr_zones - 1; i >= 0; i--) {
> > +                               struct zone *zone = pgdat->node_zones +
> i;
> > +
> > +                               if (!populated_zone(zone))
> > +                                       continue;
> > +                       }
>
> I can't understand your comment and logic.
> The comment mentioned reclaimable zone but the logic checks just
> populated_zone. What's meaning?
>

I will move the comment to another patch which adds the zone unreclaimable.

--Ying

>
> > +                       if (i < 0)
> > +                               node_clear(nid, do_nodes);
> > +
> > +                       if (mem_cgroup_watermark_ok(mem_cont,
> > +
> CHARGE_WMARK_HIGH)) {
> > +                               wmark_ok = true;
> > +                               goto out;
> > +                       }
> > +
> > +                       if (nodes_empty(do_nodes)) {
> > +                               wmark_ok = true;
> > +                               goto out;
> > +                       }
> > +               }
> > +
> > +               if (total_scanned && priority < DEF_PRIORITY - 2)
> > +                       congestion_wait(WRITE, HZ/10);
> > +
> > +               if (sc.nr_reclaimed >= SWAP_CLUSTER_MAX)
> > +                       break;
> > +       }
> > +out:
> > +       if (!wmark_ok) {
> > +               cond_resched();
> > +
> > +               try_to_freeze();
> > +
> > +               goto loop_again;
> > +       }
> > +
> > +       return sc.nr_reclaimed;
> > +}
> > +#else
> >  static unsigned long balance_mem_cgroup_pgdat(struct mem_cgroup
> *mem_cont,
> >                                                        int order)
> >  {
> >        return 0;
> >  }
> > +#endif
> >
> >  /*
> >  * The background pageout daemon, started as a kernel thread
> > --
> > 1.7.3.1
> >
> >
>
>
>
> --
> Kind regards,
> Minchan Kim
>

[-- Attachment #2: Type: text/html, Size: 14099 bytes --]