sqlbase: avoid a race in sqlbase.CancelChecker #35548
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
This change is |
asubiotto
approved these changes
Mar 8, 2019
Prior to the distsql-ification of planNode execution, all the execution steps in a local query were interleaved (using coroutine-style concurrency), so that the calls to `(*CancelChecker).Check()` were all sequential. With distsql now it's possible for different planNode `Next()` or `startExec()` methods to be running in concurrent goroutines on multiple cores, i.e. truly parallel. This in turn requires atomic access to the counter in the cancel checker. Without atomic access, there is a race condition and the possibility that the cancel checker does not work well (some increments can be performed two times, which could cause the condition of the check to occasionally fail). Found with SQLSmith. Release note: None
knz
force-pushed
the
20190308-atomic-ancel
branch
from
fbc9b43 to
1ca3629
Compare
Good idea. Done. |
|
TFYR! bors r+ |
craig bot
pushed a commit
that referenced
this pull request
Mar 8, 2019
35521: sql: reduce non-determinism in the show_trace logic tests r=knz a=knz Fixes #33720. First commit from #35519. The test is really about asserting the KV operations sent on behalf of SQL statements. The distsender traffic is really irrelevant. Since that's where most of the test flakes / non-determinism is coming from, simply remove it. Release note: None 35548: sqlbase: avoid a race in sqlbase.CancelChecker r=knz a=knz Fixes #35539. Prior to the distsql-ification of planNode execution, all the execution steps in a local query were interleaved (using coroutine-style concurrency), so that the calls to `(*CancelChecker).Check()` were all sequential. With distsql now it's possible for different planNode `Next()` or `startExec()` methods to be running in concurrent goroutines on multiple cores, i.e. truly parallel. This in turn requires atomic access to the counter in the cancel checker. Without atomic access, there is a race condition and the possibility that the cancel checker does not work well (some increments can be performed two times, which could cause the condition of the check to occasionally fail). Found with SQLSmith. Release note: None Co-authored-by: Raphael 'kena' Poss <[email protected]>
Build succeeded |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #35539.
Prior to the distsql-ification of planNode execution, all the
execution steps in a local query were interleaved (using
coroutine-style concurrency), so that the calls to
(*CancelChecker).Check()were all sequential.With distsql now it's possible for different planNode
Next()orstartExec()methods to be running in concurrent goroutines onmultiple cores, i.e. truly parallel.
This in turn requires atomic access to the counter in the cancel
checker. Without atomic access, there is a race condition and the
possibility that the cancel checker does not work well (some
increments can be performed two times, which could cause the condition
of the check to occasionally fail).
Found with SQLSmith.
Release note: None