You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "ASF GitHub Bot (Jira)" <ji...@apache.org> on 2022/07/18 22:12:00 UTC
[jira] [Work logged] (HIVE-26408) Vectorization: Fix deallocation of scratch columns, don't reuse a child ConstantVectorExpression as an output
[ https://issues.apache.org/jira/browse/HIVE-26408?focusedWorklogId=792373&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-792373 ]
ASF GitHub Bot logged work on HIVE-26408:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 18/Jul/22 22:11
Start Date: 18/Jul/22 22:11
Worklog Time Spent: 10m
Work Description: abstractdog opened a new pull request, #3452:
URL: https://github.com/apache/hive/pull/3452
<!--
Thanks for sending a pull request! Here are some tips for you:
1. If this is your first time, please read our contributor guidelines: https://cwiki.apache.org/confluence/display/Hive/HowToContribute
2. Ensure that you have created an issue on the Hive project JIRA: https://issues.apache.org/jira/projects/HIVE/summary
3. Ensure you have added or run the appropriate tests for your PR:
4. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP]HIVE-XXXXX: Your PR title ...'.
5. Be sure to keep the PR description updated to reflect all changes.
6. Please write your PR title to summarize what this PR proposes.
7. If possible, provide a concise example to reproduce the issue for a faster review.
-->
### What changes were proposed in this pull request?
<!--
Please clarify what changes you are proposing. The purpose of this section is to outline the changes and how this PR fixes the issue.
If possible, please consider writing useful notes for better and faster reviews in your PR. See the examples below.
1. If you refactor some codes with changing classes, showing the class hierarchy will help reviewers.
2. If you fix some SQL features, you can provide some references of other DBMSes.
3. If there is design documentation, please add the link.
4. If there is a discussion in the mailing list, please add the link.
-->
### Why are the changes needed?
<!--
Please clarify why the changes are needed. For instance,
1. If you propose a new API, clarify the use case for a new API.
2. If you fix a bug, you can clarify why it is a bug.
-->
### Does this PR introduce _any_ user-facing change?
<!--
Note that it means *any* user-facing change including all aspects such as the documentation fix.
If yes, please clarify the previous behavior and the change this PR proposes - provide the console output, description, screenshot and/or a reproducable example to show the behavior difference if possible.
If possible, please also clarify if this is a user-facing change compared to the released Hive versions or within the unreleased branches such as master.
If no, write 'No'.
-->
### How was this patch tested?
<!--
If tests were added, say they were added here. Please make sure to add some test cases that check the changes thoroughly including negative and positive cases if possible.
If it was tested in a way different from regular unit tests, please clarify how you tested step by step, ideally copy and paste-able, so that other reviewers can test and check, and descendants can verify in the future.
If tests were not added, please describe why they were not added and/or why it was difficult to add.
-->
Issue Time Tracking
-------------------
Worklog Id: (was: 792373)
Remaining Estimate: 0h
Time Spent: 10m
> Vectorization: Fix deallocation of scratch columns, don't reuse a child ConstantVectorExpression as an output
> -------------------------------------------------------------------------------------------------------------
>
> Key: HIVE-26408
> URL: https://issues.apache.org/jira/browse/HIVE-26408
> Project: Hive
> Issue Type: Bug
> Reporter: László Bodor
> Assignee: László Bodor
> Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> This is similar to HIVE-15588. With a customer query, I reproduced a vectorized expression tree like the below one (I'll attach a simple repro query when it's possible):
> {code}
> selectExpressions: IfExprCondExprColumn(col 67:boolean, col 63:string, col 61:string)(children: StringColumnInList(col 13, values TermDeposit, RecurringDeposit, CertificateOfDeposit) -> 67:boolean, VectorCoalesce(columns [61, 62])(children: VectorUDFAdaptor(from_unixtime(to_unix_timestamp(CAST( _col1 AS DATE)), 'MM-dd-yyyy'))(children: VectorUDFUnixTimeStampDate(col 68)(children: CastStringToDate(col 33:string) -> 68:date) -> 69:bigint) -> 61:string, ConstantVectorExpression(val ) -> 62:string) -> 63:string, ConstantVectorExpression(val ) -> 61:string) -> 62:string
> {code}
> query part was:
> {code}
> CASE WHEN DLY_BAL.PDELP_VALUE in (
> 'TermDeposit', 'RecurringDeposit',
> 'CertificateOfDeposit'
> ) THEN NVL(
> (
> from_unixtime(
> unix_timestamp(
> cast(DLY_BAL.APATD_MTRTY_DATE as date)
> ),
> 'MM-dd-yyyy'
> )
> ),
> ' '
> ) ELSE '' END AS MAT_DTE
> {code}
> Here is the problem described:
> 1. IfExprCondExprColumn has 62:string as its [outputColumn|https://github.com/apache/hive/blob/d3309c0ea9da907af4d27427805084b7331a6c24/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/expressions/IfExprCondExprColumn.java#L64], which is a reused scratch column (see 5) )
> 2. in evaluation time, [isRepeating is reset|https://github.com/apache/hive/blob/d3309c0ea9da907af4d27427805084b7331a6c24/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/expressions/IfExprCondExprColumn.java#L68]
> 3. in order to evaluate IfExprCondExprColumn, the conditional evaluation of children is required, so we go to [conditionalEvaluate|https://github.com/apache/hive/blob/d3309c0ea9da907af4d27427805084b7331a6c24/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/expressions/IfExprCondExprColumn.java#L95]
> 4. one of the children is ConstantVectorExpression(val ) -> 62:string, which belongs to the second branch of VectorCoalesce, so to the '' empty string in NVL's second argument
> 5. in 4) 62: string column is set to an isRepeating column (and it's released by [freeNonColumns|https://github.com/apache/hive/blob/d3309c0ea9da907af4d27427805084b7331a6c24/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorizationContext.java#L2459]), so it's marked as a reusable scratch column
> 6. after the conditional evaluation in 3), the final output of IfExprCondExprColumn set [here|https://github.com/apache/hive/blob/d3309c0ea9da907af4d27427805084b7331a6c24/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/expressions/IfExprCondExprColumn.java#L99], but here we get an exception [here|https://github.com/apache/hive/blob/d3309c0ea9da907af4d27427805084b7331a6c24/storage-api/src/java/org/apache/hadoop/hive/ql/exec/vector/BytesColumnVector.java#L484]:
> {code}
> 2022-07-01T04:26:24,567 ERROR [TezTR-745267_1_35_6_0_0] tez.MapRecordSource: java.lang.AssertionError: Output column number expected to be 0 when isRepeating
> at org.apache.hadoop.hive.ql.exec.vector.BytesColumnVector.setElement(BytesColumnVector.java:494)
> at org.apache.hadoop.hive.ql.exec.vector.expressions.IfExprCondExprColumn.evaluate(IfExprCondExprColumn.java:108)
> at org.apache.hadoop.hive.ql.exec.vector.VectorSelectOperator.process(VectorSelectOperator.java:146)
> at org.apache.hadoop.hive.ql.exec.Operator.vectorForward(Operator.java:969)
> at org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinGenerateResultOperator.forwardBigTableBatch(VectorMapJoinGenerateResultOperator.java:694)
> at org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinInnerBigOnlyStringOperator.processBatch(VectorMapJoinInnerBigOnlyStringOperator.java:371)
> at org.apache.hadoop.hive.ql.exec.vector.mapjoin.VectorMapJoinCommonOperator.process(VectorMapJoinCommonOperator.java:839)
> at org.apache.hadoop.hive.ql.exec.Operator.vectorForward(Operator.java:969)
> {code}
> this is clearly an incorrect scratch column reuse, where we reused the output of some children, and got that vector in an inconsistent state
> this must not be fixed by resetting vectors in more places in IfExprCondExprColumn, as it would just hide the original issue
> I realized that the problem can be easily fixed by simply preventing releasing ConstantVectorExpressions, that's what I'm trying to test now
--
This message was sent by Atlassian Jira
(v8.20.10#820010)