wrong stats of TestIssues24349

[winoros]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

Looking into column b's stats:

• partition p0: TopN: (3, 3). Hist: one bucket [2, 2] with count = 1
• partition p1: TopN: (2, 3), Hist: empty
• partition p2: TopN: (1, 2), Hist: two buckets [1, 3] with count = 2 and repeat = 1, [4, 4] with count = 1 and repeat = 1

After fixing the global TopN, it is now 2 with 4 as its occurrence (local TopN of p1 and one single bucket from p0).

Remained buckets(TopN also treat as bucket)

• [3, 3] count = 4, repeat = 4 (local TopN from p0 plus 1 from p2's bucket [1, 3]'s repeat)
• [1, 1] count = 2, repeat = 2
• [1, 3] count = 1, repeat = 0
• [4, 4] count = 1, repeat = 1

After merging them to get the global histogram, we should get one bucket [1, 4] with count = 8 and repeat = 1.
But in the current test result, we have

The count is 10.

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

3. What did you see instead (Required)

4. What is your TiDB version? (Required)

master.

[winoros]

It affects many branches. But I decide to only pick it to v8.5.

It's caused by that Column's histogram uses decoded column value while TopNs of it use encoded value.