fix(batch): fix hash join process some remaining chunk builder

[st1page]

I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.

What's changed and what's your intention?

fix #15899
The process_xxx function is only need for the chunk_builder but not need for the remaining_chunk_builder

Checklist

• I have written necessary rustdoc comments
• I have added necessary unit tests and integration tests
• I have added test labels as necessary. See details.
• I have added fuzzing tests or opened an issue to track them. (Optional, recommended for new SQL features Sqlsmith: Sql feature generation #7934).
• My PR contains breaking changes. (If it deprecates some features, please create a tracking issue to remove them in the future).
• All checks passed in ./risedev check (or alias, ./risedev c)
• My PR changes performance-critical code. (Please run macro/micro-benchmarks and show the results.)

• My PR contains critical fixes that are necessary to be merged into the latest release. (Please check out the details)

Documentation

• My PR needs documentation updates. (Please use the Release note section below to summarize the impact on users)

Release note

If this PR includes changes that directly affect users or other significant modifications relevant to the community, kindly draft a release note to provide a concise summary of these changes. Please prioritize highlighting the impact these changes will have on users.

[kwannoel]

I suppose I need to cherry pick this into kwannoel/release-1.7-xxx?

Because this is a follow-up from previous fix here: #15853

[fuyufjh]

LGTM.

Can you please a test case somewhere? The problamtic SQL in fuzz test is:

WITH with_0 AS (SELECT 'g60H9TgFkY' AS col_0 FROM (hop(auction, auction.date_time, INTERVAL '30', INTERVAL '540') AS hop_1 FULL JOIN hop(alltypes2, alltypes2.c11, INTERVAL '93', INTERVAL '1302') AS hop_2 ON hop_1.expires = hop_2.c11) FULL JOIN tumble(bid, bid.date_time, INTERVAL '12') AS tumble_3 ON hop_1.seller = tumble_3.price AND hop_2.c1 WHERE EXISTS (SELECT tumble_4.name AS col_0, hop_5.id AS col_1, (lower(max(DISTINCT 'wZCO4fLGwT' ORDER BY 'wZCO4fLGwT' NULLS FIRST))) AS col_2 FROM tumble(person, person.date_time, INTERVAL '18') AS tumble_4 LEFT JOIN hop(person, person.date_time, INTERVAL '33', INTERVAL '198') AS hop_5 ON tumble_4.credit_card = hop_5.name GROUP BY GROUPING SETS ((hop_5.date_time, hop_5.credit_card, tumble_4.city, tumble_4.state, tumble_4.date_time, hop_5.id), (hop_5.date_time, tumble_4.name, tumble_4.state, hop_5.extra), (tumble_4.name, hop_5.id, tumble_4.credit_card, tumble_4.email_address, tumble_4.state), (tumble_4.id, tumble_4.date_time, hop_5.state, tumble_4.email_address, hop_5.name), (hop_5.email_address, tumble_4.date_time, tumble_4.city, hop_5.id)) HAVING true) GROUP BY hop_2.c5, tumble_3.extra, tumble_3.bidder, hop_2.c13, hop_2.c15, hop_1.id, tumble_3.date_time, hop_2.c1, hop_2.c4, hop_1.expires, hop_1.item_name, hop_1.extra HAVING ((FLOAT '0') IS NOT NULL) LIMIT 7) SELECT tumble_6.c8 AS col_0, tumble_6.c11 AS col_1, (INTERVAL '-56') AS col_2 FROM with_0, tumble(alltypes1, alltypes1.c11, INTERVAL '13') AS tumble_6 WHERE (tumble_6.c11 = tumble_6.c11) GROUP BY tumble_6.c13, tumble_6.c8, tumble_6.c1, tumble_6.c11, tumble_6.c3 HAVING tumble_6.c1 ORDER BY tumble_6.c13 ASC NULLS FIRST

See https://buildkite.com/risingwavelabs/main-cron/builds/2154#018e7577-bd40-4e4f-ac86-a5ff0c4d0010 / fuzz test / logs / risedev-logs.tgz for full logs.

[st1page]

LGTM.

Can you please a test case somewhere? The problamtic SQL in fuzz test is:

WITH with_0 AS (SELECT 'g60H9TgFkY' AS col_0 FROM (hop(auction, auction.date_time, INTERVAL '30', INTERVAL '540') AS hop_1 FULL JOIN hop(alltypes2, alltypes2.c11, INTERVAL '93', INTERVAL '1302') AS hop_2 ON hop_1.expires = hop_2.c11) FULL JOIN tumble(bid, bid.date_time, INTERVAL '12') AS tumble_3 ON hop_1.seller = tumble_3.price AND hop_2.c1 WHERE EXISTS (SELECT tumble_4.name AS col_0, hop_5.id AS col_1, (lower(max(DISTINCT 'wZCO4fLGwT' ORDER BY 'wZCO4fLGwT' NULLS FIRST))) AS col_2 FROM tumble(person, person.date_time, INTERVAL '18') AS tumble_4 LEFT JOIN hop(person, person.date_time, INTERVAL '33', INTERVAL '198') AS hop_5 ON tumble_4.credit_card = hop_5.name GROUP BY GROUPING SETS ((hop_5.date_time, hop_5.credit_card, tumble_4.city, tumble_4.state, tumble_4.date_time, hop_5.id), (hop_5.date_time, tumble_4.name, tumble_4.state, hop_5.extra), (tumble_4.name, hop_5.id, tumble_4.credit_card, tumble_4.email_address, tumble_4.state), (tumble_4.id, tumble_4.date_time, hop_5.state, tumble_4.email_address, hop_5.name), (hop_5.email_address, tumble_4.date_time, tumble_4.city, hop_5.id)) HAVING true) GROUP BY hop_2.c5, tumble_3.extra, tumble_3.bidder, hop_2.c13, hop_2.c15, hop_1.id, tumble_3.date_time, hop_2.c1, hop_2.c4, hop_1.expires, hop_1.item_name, hop_1.extra HAVING ((FLOAT '0') IS NOT NULL) LIMIT 7) SELECT tumble_6.c8 AS col_0, tumble_6.c11 AS col_1, (INTERVAL '-56') AS col_2 FROM with_0, tumble(alltypes1, alltypes1.c11, INTERVAL '13') AS tumble_6 WHERE (tumble_6.c11 = tumble_6.c11) GROUP BY tumble_6.c13, tumble_6.c8, tumble_6.c1, tumble_6.c11, tumble_6.c3 HAVING tumble_6.c1 ORDER BY tumble_6.c13 ASC NULLS FIRST

See https://buildkite.com/risingwavelabs/main-cron/builds/2154#018e7577-bd40-4e4f-ac86-a5ff0c4d0010 / fuzz test / logs / risedev-logs.tgz for full logs.

Yes, I have reproduced this bug in my local environment and ensured that is fixed. But I think the original test case is a little big as a e2e test.

[chenzl25]

Do we need to modify the left semi-join and full outer join as well? because I remember https://github.com/risingwavelabs/risingwave/pull/15853/files this PR modified 3 places.

[fuyufjh]

Yes, I have reproduced this bug in my local environment and ensured that is fixed. But I think the original test case is a little big as a e2e test.

Sounds like you have a more minimal reproducing case? That would be much better for a test case

[st1page]

Yes, I have reproduced this bug in my local environment and ensured that is fixed. But I think the original test case is a little big as a e2e test.

Sounds like you have a more minimal reproducing case? That would be much better for a test case

No, I just use the original case in the fuzz test 🥵

[st1page]

Do we need to modify the left semi-join and full outer join as well? because I remember https://github.com/risingwavelabs/risingwave/pull/15853/files this PR modified 3 places.

The three place in the previous PR is leftOuter, leftAnti and fullOuter. I wrongly changes the leftAnti and fullOuter by mistake.
And no changes needed for leftSemi because it has no code on the joinkey's mismatch path

[st1page]

the sql fuzz test fails because some other issues that the frontend stack overflow.
https://buildkite.com/risingwavelabs/pull-request/builds/45532#018e79dc-9596-4f36-9aed-c57fabef65ce

thread 'rw-main' has overflowed its stack
fatal runtime error: stack overflow

#15916