*: prevent cursor read from being cancelled by GC (#39950)

[ti-chi-bot]

This is an automated cherry-pick of #39950

Signed-off-by: zyguan zhongyangguan@gmail.com

What problem does this PR solve?

Issue Number: close #39447

Problem Summary:

Cursor read might be cancelled by GC since min-start-ts is calculated by min([s.ProcessInfo.CurTxnStartTS for s in all sessions]) however the CurTxnStartTS will be reset once fetch returned.

What is changed and how it works?

1. Let each session hold a list of pretected timestamps.
2. Let session manager calculate the min-start-ts (considering protected timestamps).
3. When start to do cursor read, hold the start-ts until the result set gets close.

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
• No code

Side effects

• Performance regression: Consumes more CPU
• Performance regression: Consumes more Memory
• Breaking backward compatibility

Documentation

• Affects user behaviors
• Contains syntax changes
• Contains variable changes
• Contains experimental features
• Changes MySQL compatibility

Release note

Please refer to Release Notes Language Style Guide to write a quality release note.

None

[ti-chi-bot]

[REVIEW NOTIFICATION]

This pull request has been approved by:

• cfzjywxk
• ekexium

To complete the pull request process, please ask the reviewers in the list to review by filling /cc @reviewer in the comment.
After your PR has acquired the required number of LGTMs, you can assign this pull request to the committer in the list by filling /assign @committer in the comment to help you merge this pull request.

The full list of commands accepted by this bot can be found here.

Reviewer can indicate their review by submitting an approval review.
Reviewer can cancel approval by submitting a request changes review.

[zyguan]

/retest

[cfzjywxk]

@zyguan
Please help to resolve the conflicts.

[cfzjywxk]

/merge

[ti-chi-bot]

This pull request has been accepted and is ready to merge.

Commit hash: 5bed421