Hash join statistic is incorrect when spill is triggered

[xzhangxian1008]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

Trigger spill in hash join.

2. What did you expect to see? (Required)

correct statistic

3. What did you see instead (Required)

+------------------------------------+--------------+-----------+-----------+----------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------------------------------------------------------------------------------------------------------------------------------------+----------+---------+
| id                                 | estRows      | actRows   | task      | access object  | execution info | operator info | memory   | disk    |
+------------------------------------+--------------+-----------+-----------+----------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------------------------------------------------------------------------------------------------------------------------------------+----------+---------+
| Sort_10                            | 1.00         | 5         | root      |                | time:1m15.8s, open:657.4µs, close:29.5µs, loops:2, RU:1199689.70                                                                                                                                                                                                                                                                         | test.orders.o_orderpriority                                                                                                               | 1.21 KB  | 0 Bytes |
| └─Projection_12                    | 1.00         | 5         | root      |                | time:1m15.8s, open:645.4µs, close:28.6µs, loops:5, Concurrency:OFF | test.orders.o_orderpriority, Column#26                                                                                                    | 34.0 KB  | N/A     |
|   └─HashAgg_13                     | 1.00         | 5         | root      |                | time:1m15.8s, open:641.8µs, close:27µs, loops:5, partial_worker:{wall_time:1m15.813126507s, concurrency:5, task_num:2588, tot_wait:6m18.325465785s, tot_exec:611.285923ms, tot_time:6m19.065554441s, max:1m15.813112828s, p95:1m15.813112828s}, final_worker:{wall_time:1m15.813146976s, concurrency:5, task_num:21, tot_wait:16.761µs, tot_exec:13.302µs, tot_time:6m19.065664816s, max:1m15.813135023s, p95:1m15.813135023s}                                             | group by:test.orders.o_orderpriority, funcs:count(1)->Column#26, funcs:firstrow(test.orders.o_orderpriority)->test.orders.o_orderpriority | 481.9 KB | 0 Bytes |
|     └─HashJoin_26                  | 2281928.20   | 2574425   | root      |                | time:1m15.8s, open:585.6µs, close:24.4µs, loops:2589, build_hash_table:{total:160.9ms, fetch:-7.013241164s, build:7.51s}, probe:{concurrency:5, total:3m25s, max:40.5s, probe:3.09s, fetch and wait:3m21.9s, probe_collision:19336081}                                                                                                                                                                                                                                     | semi join, left side:TableReader_29, equal:[eq(test.orders.o_orderkey, test.lineitem.l_orderkey)]                                         | 1.00 GB  | 9.45 GB |                                                                                                                                                       
|       ├─TableReader_32(Build)      | 240004648.80 | 189670348 | root      |                | time:17.3s, open:425.7µs, close:3.61µs, loops:184597, cop_task: {num: 5628, max: 309.5ms, min: 268.9µs, avg: 58.1ms, p95: 143.3ms, max_proc_keys: 80864, p95_proc_keys: 79840, tot_proc: 5m15.4s, tot_wait: 747.8ms, copr_cache_hit_ratio: 0.00, build_task_duration: 383.2µs, max_distsql_concurrency: 15}, rpc_info:{Cop:{num_rpc:5628, total_time:5m26.7s}}                                                                                                             | data:Selection_31 | 18.5 MB  | N/A     |                                                                                                                                                       
|       │ └─Selection_31             | 240004648.80 | 189670348 | cop[tikv] |                | tikv_task:{proc max:0s, min:0s, avg: 54.1ms, p80:78ms, p95:133ms, iters:315458, tasks:5628}, scan_detail: {total_process_keys: 299982707, total_process_keys_size: 58990940562, total_keys: 299988321, get_snapshot_time: 403.4ms, rocksdb: {key_skipped_count: 299982707, block: {cache_hit_count: 1883561}}}, time_detail: {total_process_time: 5m15.4s, total_suspend_time: 1.11s, total_wait_time: 747.8ms, total_kv_read_wall_time: 4m37.7s, tikv_wall_time: 5m18.8s} | lt(test.lineitem.l_commitdate, test.lineitem.l_receiptdate)                                                                               | N/A      | N/A     |                                                                                                                                                       
|       │   └─TableFullScan_30       | 300005811.00 | 300005811 | cop[tikv] | table:lineitem | tikv_task:{proc max:0s, min:0s, avg: 49.4ms, p80:71.6ms, p95:121.7ms, iters:315458, tasks:5628}                                                                                                                                                                                                                                                                                                                                                                            | keep order:false | N/A      | N/A     |
|       └─TableReader_29(Probe)      | 2852410.25   | 2807525   | root      |                | time:4.4s, open:143.5µs, close:15.2µs, loops:2734, cop_task: {num: 395, max: 1.04s, min: 214.8µs, avg: 150.6ms, p95: 658.5ms, max_proc_keys: 879577, p95_proc_keys: 871392, tot_proc: 59s, tot_wait: 54.9ms, copr_cache_hit_ratio: 0.21, build_task_duration: 81.4µs, max_distsql_concurrency: 15}, rpc_info:{Cop:{num_rpc:395, total_time:59.5s}}                                                                                                                         | data:Selection_28 | 2.08 MB  | N/A     |                                                                                                                                                       
|         └─Selection_28             | 2852410.25   | 2807525   | cop[tikv] |                | tikv_task:{proc max:1.02s, min:1ms, avg: 151.1ms, p80:291ms, p95:652ms, iters:74858, tasks:395}, scan_detail: {total_process_keys: 74173248, total_process_keys_size: 11266342171, total_keys: 74173559, get_snapshot_time: 39.4ms, rocksdb: {key_skipped_count: 74173248, block: {cache_hit_count: 376619}}}, time_detail: {total_process_time: 59s, total_suspend_time: 115.9ms, total_wait_time: 54.9ms, total_kv_read_wall_time: 54.6s, tikv_wall_time: 59.2s}         | ge(test.orders.o_orderdate, 1995-01-01 00:00:00.000000), lt(test.orders.o_orderdate, 1995-04-01 00:00:00.000000)                          | N/A      | N/A     |                                                                                                                                                       
|           └─TableFullScan_27       | 75000000.00  | 75000000  | cop[tikv] | table:orders   | tikv_task:{proc max:970ms, min:1ms, avg: 141.5ms, p80:275ms, p95:611ms, iters:74858, tasks:395} | keep order:false                                                                                                                          | N/A      | N/A     |                                                                                                                                                       
+------------------------------------+--------------+-----------+-----------+----------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------------------------------------------------------------------------------------------------------------------------------------+----------+---------+
10 rows in set (1 min 15.82 sec)

4. What is your TiDB version? (Required)

nightly(9b70321)

[hawkingrei]

explain select /*+ tidb_hj(orders) */
        o_orderpriority,
        count(*) as order_count
from
        orders use index()
where
        o_orderdate >= '1995-01-01'
        and o_orderdate < date_add('1995-01-01', interval '3' month)
        and exists (
                select
                        *
                from
                        lineitem use index()
                where
                        l_orderkey = o_orderkey
                        and l_commitdate < l_receiptdate
        )
group by
        o_orderpriority
order by
        o_orderpriority

the sql looks like this.