On 2024/12/10 16:24, Barry Song wrote:
> On Tue, Dec 10, 2024 at 2:41 PM chenridong <chenridong@xxxxxxxxxx> wrote:
>>
>>
>>
>> On 2024/12/10 12:54, Barry Song wrote:
>>> On Mon, Dec 9, 2024 at 4:46 PM Chen Ridong <chenridong@xxxxxxxxxxxxxxx> wrote:
>>>>
>>>> From: Chen Ridong <chenridong@xxxxxxxxxx>
>>>>
>>>> The commit 359a5e1416ca ("mm: multi-gen LRU: retry folios written back
>>>> while isolated") only fixed the issue for mglru. However, this issue
>>>> also exists in the traditional active/inactive LRU. This issue will be
>>>> worse if THP is split, which makes the list longer and needs longer time
>>>> to finish a batch of folios reclaim.
>>>>
>>>> This issue should be fixed in the same way for the traditional LRU.
>>>> Therefore, the common logic was extracted to the 'find_folios_written_back'
>>>> function firstly, which is then reused in the 'shrink_inactive_list'
>>>> function. Finally, retry reclaiming those folios that may have missed the
>>>> rotation for traditional LRU.
>>>
>>> let's drop the cover-letter and refine the changelog.
>>>
>> Will update.
>>
>>>>
>>>> Signed-off-by: Chen Ridong <chenridong@xxxxxxxxxx>
>>>> ---
>>>> include/linux/mmzone.h | 3 +-
>>>> mm/vmscan.c | 108 +++++++++++++++++++++++++++++------------
>>>> 2 files changed, 77 insertions(+), 34 deletions(-)
>>>>
>>>> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
>>>> index b36124145a16..47c6e8c43dcd 100644
>>>> --- a/include/linux/mmzone.h
>>>> +++ b/include/linux/mmzone.h
>>>> @@ -391,6 +391,7 @@ struct page_vma_mapped_walk;
>>>>
>>>> #define LRU_GEN_MASK ((BIT(LRU_GEN_WIDTH) - 1) << LRU_GEN_PGOFF)
>>>> #define LRU_REFS_MASK ((BIT(LRU_REFS_WIDTH) - 1) << LRU_REFS_PGOFF)
>>>> +#define LRU_REFS_FLAGS (BIT(PG_referenced) | BIT(PG_workingset))
>>>>
>>>> #ifdef CONFIG_LRU_GEN
>>>>
>>>> @@ -406,8 +407,6 @@ enum {
>>>> NR_LRU_GEN_CAPS
>>>> };
>>>>
>>>> -#define LRU_REFS_FLAGS (BIT(PG_referenced) | BIT(PG_workingset))
>>>> -
>>>> #define MIN_LRU_BATCH BITS_PER_LONG
>>>> #define MAX_LRU_BATCH (MIN_LRU_BATCH * 64)
>>>>
>>>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>>>> index 76378bc257e3..1f0d194f8b2f 100644
>>>> --- a/mm/vmscan.c
>>>> +++ b/mm/vmscan.c
>>>> @@ -283,6 +283,48 @@ static void set_task_reclaim_state(struct task_struct *task,
>>>> task->reclaim_state = rs;
>>>> }
>>>>
>>>> +/**
>>>> + * find_folios_written_back - Find and move the written back folios to a new list.
>>>> + * @list: filios list
>>>> + * @clean: the written back folios list
>>>> + * @skip: whether skip to move the written back folios to clean list.
>>>> + */
>>>> +static inline void find_folios_written_back(struct list_head *list,
>>>> + struct list_head *clean, bool skip)
>>>> +{
>>>> + struct folio *folio;
>>>> + struct folio *next;
>>>> +
>>>> + list_for_each_entry_safe_reverse(folio, next, list, lru) {
>>>> + if (!folio_evictable(folio)) {
>>>> + list_del(&folio->lru);
>>>> + folio_putback_lru(folio);
>>>> + continue;
>>>> + }
>>>> +
>>>> + if (folio_test_reclaim(folio) &&
>>>> + (folio_test_dirty(folio) || folio_test_writeback(folio))) {
>>>> + /* restore LRU_REFS_FLAGS cleared by isolate_folio() */
>>>> + if (lru_gen_enabled() && folio_test_workingset(folio))
>>>> + folio_set_referenced(folio);
>>>> + continue;
>>>> + }
>>>> +
>>>> + if (skip || folio_test_active(folio) || folio_test_referenced(folio) ||
>>>> + folio_mapped(folio) || folio_test_locked(folio) ||
>>>> + folio_test_dirty(folio) || folio_test_writeback(folio)) {
>>>> + /* don't add rejected folios to the oldest generation */
>>>> + if (lru_gen_enabled())
>>>> + set_mask_bits(&folio->flags, LRU_REFS_MASK | LRU_REFS_FLAGS,
>>>> + BIT(PG_active));
>>>> + continue;
>>>> + }
>>>> +
>>>> + /* retry folios that may have missed folio_rotate_reclaimable() */
>>>> + list_move(&folio->lru, clean);
>>>> + }
>>>> +}
>>>> +
>>>> /*
>>>> * flush_reclaim_state(): add pages reclaimed outside of LRU-based reclaim to
>>>> * scan_control->nr_reclaimed.
>>>> @@ -1907,6 +1949,25 @@ static int current_may_throttle(void)
>>>> return !(current->flags & PF_LOCAL_THROTTLE);
>>>> }
>>>>
>>>> +static inline void acc_reclaimed_stat(struct reclaim_stat *stat,
>>>> + struct reclaim_stat *curr)
>>>> +{
>>>> + int i;
>>>> +
>>>> + stat->nr_dirty += curr->nr_dirty;
>>>> + stat->nr_unqueued_dirty += curr->nr_unqueued_dirty;
>>>> + stat->nr_congested += curr->nr_congested;
>>>> + stat->nr_writeback += curr->nr_writeback;
>>>> + stat->nr_immediate += curr->nr_immediate;
>>>> + stat->nr_pageout += curr->nr_pageout;
>>>> + stat->nr_ref_keep += curr->nr_ref_keep;
>>>> + stat->nr_unmap_fail += curr->nr_unmap_fail;
>>>> + stat->nr_lazyfree_fail += curr->nr_lazyfree_fail;
>>>> + stat->nr_demoted += curr->nr_demoted;
>>>> + for (i = 0; i < ANON_AND_FILE; i++)
>>>> + stat->nr_activate[i] = curr->nr_activate[i];
>>>> +}
>>>
>>> you had no this before, what's the purpose of this？
>>>
>>
>> We may call shrink_folio_list twice, and the 'stat curr' will reset in
>> the shrink_folio_list function. We should accumulate the stats as a
>> whole, which will then be used to calculate the cost and return it to
>> the caller.
>
> Does mglru have the same issue? If so, we may need to send a patch to
> fix mglru's stat accounting as well. By the way, the code is rather
> messy—could it be implemented as shown below instead?
>
I have checked the code (in the evict_folios function) again, and it
appears that 'reclaimed' should correspond to sc->nr_reclaimed, which
accumulates the results twice. Should I address this issue with a
separate patch?
if (!cgroup_reclaim(sc))
__count_vm_events(item, reclaimed);
__count_memcg_events(memcg, item, reclaimed);
__count_vm_events(PGSTEAL_ANON + type, reclaimed);
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 1f0d194f8b2f..40d2ddde21f5 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1094,7 +1094,6 @@ static unsigned int shrink_folio_list(struct
> list_head *folio_list,
> struct swap_iocb *plug = NULL;
>
> folio_batch_init(&free_folios);
> - memset(stat, 0, sizeof(*stat));
> cond_resched();
> do_demote_pass = can_demote(pgdat->node_id, sc);
>
> @@ -1949,25 +1948,6 @@ static int current_may_throttle(void)
> return !(current->flags & PF_LOCAL_THROTTLE);
> }
>
> -static inline void acc_reclaimed_stat(struct reclaim_stat *stat,
> - struct reclaim_stat *curr)
> -{
> - int i;
> -
> - stat->nr_dirty += curr->nr_dirty;
> - stat->nr_unqueued_dirty += curr->nr_unqueued_dirty;
> - stat->nr_congested += curr->nr_congested;
> - stat->nr_writeback += curr->nr_writeback;
> - stat->nr_immediate += curr->nr_immediate;
> - stat->nr_pageout += curr->nr_pageout;
> - stat->nr_ref_keep += curr->nr_ref_keep;
> - stat->nr_unmap_fail += curr->nr_unmap_fail;
> - stat->nr_lazyfree_fail += curr->nr_lazyfree_fail;
> - stat->nr_demoted += curr->nr_demoted;
> - for (i = 0; i < ANON_AND_FILE; i++)
> - stat->nr_activate[i] = curr->nr_activate[i];
> -}
> -
> /*
> * shrink_inactive_list() is a helper for shrink_node(). It returns the number
> * of reclaimed pages
> @@ -1981,7 +1961,7 @@ static unsigned long
> shrink_inactive_list(unsigned long nr_to_scan,
> unsigned long nr_scanned;
> unsigned int nr_reclaimed = 0;
> unsigned long nr_taken;
> - struct reclaim_stat stat, curr;
> + struct reclaim_stat stat;
> bool file = is_file_lru(lru);
> enum vm_event_item item;
> struct pglist_data *pgdat = lruvec_pgdat(lruvec);
> @@ -2022,9 +2002,8 @@ static unsigned long
> shrink_inactive_list(unsigned long nr_to_scan,
>
> memset(&stat, 0, sizeof(stat));
> retry:
> - nr_reclaimed += shrink_folio_list(&folio_list, pgdat, sc, &curr, false);
> + nr_reclaimed += shrink_folio_list(&folio_list, pgdat, sc, &stat, false);
> find_folios_written_back(&folio_list, &clean_list, skip_retry);
> - acc_reclaimed_stat(&stat, &curr);
>
> spin_lock_irq(&lruvec->lru_lock);
> move_folios_to_lru(lruvec, &folio_list);
>
This seems much better. But we have extras works to do:
1. In the shrink_folio_list function:
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1089,12 +1089,12 @@ static unsigned int shrink_folio_list(struct
list_head *folio_list,
LIST_HEAD(ret_folios);
LIST_HEAD(demote_folios);
unsigned int nr_reclaimed = 0;
- unsigned int pgactivate = 0;
+ unsigned int pgactivate = stat->nr_activate[0] +
stat->nr_activate[1];
+ unsigned int nr_demote = 0;
bool do_demote_pass;
struct swap_iocb *plug = NULL;
folio_batch_init(&free_folios);
- memset(stat, 0, sizeof(*stat));
cond_resched();
do_demote_pass = can_demote(pgdat->node_id, sc);
@@ -1558,7 +1558,8 @@ static unsigned int shrink_folio_list(struct
list_head *folio_list,
/* Migrate folios selected for demotion */
stat->nr_demoted = demote_folio_list(&demote_folios, pgdat);
- nr_reclaimed += stat->nr_demoted;
+ stat->nr_demoted += nr_demote;
+ nr_reclaimed += nr_demote;
/* Folios that could not be demoted are still in @demote_folios */
if (!list_empty(&demote_folios)) {
/* Folios which weren't demoted go back on @folio_list */
@@ -1586,7 +1587,7 @@ static unsigned int shrink_folio_list(struct
list_head *folio_list,
}
}
- pgactivate = stat->nr_activate[0] + stat->nr_activate[1];
+ pgactivate = stat->nr_activate[0] + stat->nr_activate[1] -
pgactivate;
mem_cgroup_uncharge_folios(&free_folios);
try_to_unmap_flush();
2. Outsize of the shrink_folio_list function, The callers should memset
the stat.
If you think this will be better, I will update like this.
>>
>> Thanks,
>> Ridong
>>
>>>> +
>>>> /*
>>>> * shrink_inactive_list() is a helper for shrink_node(). It returns the number
>>>> * of reclaimed pages
>>>> @@ -1916,14 +1977,16 @@ static unsigned long shrink_inactive_list(unsigned long nr_to_scan,
>>>> enum lru_list lru)
>>>> {
>>>> LIST_HEAD(folio_list);
>>>> + LIST_HEAD(clean_list);
>>>> unsigned long nr_scanned;
>>>> unsigned int nr_reclaimed = 0;
>>>> unsigned long nr_taken;
>>>> - struct reclaim_stat stat;
>>>> + struct reclaim_stat stat, curr;
>>>> bool file = is_file_lru(lru);
>>>> enum vm_event_item item;
>>>> struct pglist_data *pgdat = lruvec_pgdat(lruvec);
>>>> bool stalled = false;
>>>> + bool skip_retry = false;
>>>>
>>>> while (unlikely(too_many_isolated(pgdat, file, sc))) {
>>>> if (stalled)
>>>> @@ -1957,10 +2020,20 @@ static unsigned long shrink_inactive_list(unsigned long nr_to_scan,
>>>> if (nr_taken == 0)
>>>> return 0;
>>>>
>>>> - nr_reclaimed = shrink_folio_list(&folio_list, pgdat, sc, &stat, false);
>>>> + memset(&stat, 0, sizeof(stat));
>>>> +retry:
>>>> + nr_reclaimed += shrink_folio_list(&folio_list, pgdat, sc, &curr, false);
>>>> + find_folios_written_back(&folio_list, &clean_list, skip_retry);
>>>> + acc_reclaimed_stat(&stat, &curr);
>>>>
>>>> spin_lock_irq(&lruvec->lru_lock);
>>>> move_folios_to_lru(lruvec, &folio_list);
>>>> + if (!list_empty(&clean_list)) {
>>>> + list_splice_init(&clean_list, &folio_list);
>>>> + skip_retry = true;
>>>> + spin_unlock_irq(&lruvec->lru_lock);
>>>> + goto retry;
>
> This is rather confusing. We're still jumping to retry even though
> skip_retry=true is set. Can we find a clearer approach for this?
>
> It was somewhat acceptable before we introduced the extracted
> function find_folios_written_back(). However, it has become
> harder to follow now that skip_retry is passed across functions.
>
> I find renaming skip_retry to is_retry more intuitive. The logic
> is that since we are already retrying, find_folios_written_back()
> shouldn’t move folios to the clean list again. The intended semantics
> are: we have retris, don’t retry again.
>
Reasonable. Will update.
Thanks,
Ridong
>
>>>> + }
>>>>
>>>> __mod_lruvec_state(lruvec, PGDEMOTE_KSWAPD + reclaimer_offset(),
>>>> stat.nr_demoted);
>>>> @@ -4567,8 +4640,6 @@ static int evict_folios(struct lruvec *lruvec, struct scan_control *sc, int swap
>>>> int reclaimed;
>>>> LIST_HEAD(list);
>>>> LIST_HEAD(clean);
>>>> - struct folio *folio;
>>>> - struct folio *next;
>>>> enum vm_event_item item;
>>>> struct reclaim_stat stat;
>>>> struct lru_gen_mm_walk *walk;
>>>> @@ -4597,34 +4668,7 @@ static int evict_folios(struct lruvec *lruvec, struct scan_control *sc, int swap
>>>> scanned, reclaimed, &stat, sc->priority,
>>>> type ? LRU_INACTIVE_FILE : LRU_INACTIVE_ANON);
>>>>
>>>> - list_for_each_entry_safe_reverse(folio, next, &list, lru) {
>>>> - if (!folio_evictable(folio)) {
>>>> - list_del(&folio->lru);
>>>> - folio_putback_lru(folio);
>>>> - continue;
>>>> - }
>>>> -
>>>> - if (folio_test_reclaim(folio) &&
>>>> - (folio_test_dirty(folio) || folio_test_writeback(folio))) {
>>>> - /* restore LRU_REFS_FLAGS cleared by isolate_folio() */
>>>> - if (folio_test_workingset(folio))
>>>> - folio_set_referenced(folio);
>>>> - continue;
>>>> - }
>>>> -
>>>> - if (skip_retry || folio_test_active(folio) || folio_test_referenced(folio) ||
>>>> - folio_mapped(folio) || folio_test_locked(folio) ||
>>>> - folio_test_dirty(folio) || folio_test_writeback(folio)) {
>>>> - /* don't add rejected folios to the oldest generation */
>>>> - set_mask_bits(&folio->flags, LRU_REFS_MASK | LRU_REFS_FLAGS,
>>>> - BIT(PG_active));
>>>> - continue;
>>>> - }
>>>> -
>>>> - /* retry folios that may have missed folio_rotate_reclaimable() */
>>>> - list_move(&folio->lru, &clean);
>>>> - }
>>>> -
>>>> + find_folios_written_back(&list, &clean, skip_retry);
>>>> spin_lock_irq(&lruvec->lru_lock);
>>>>
>>>> move_folios_to_lru(lruvec, &list);
>>>> --
>>>> 2.34.1
>>>>
>>>
>
> Thanks
> Barry
