[SPARK-53394][CORE] UninterruptibleLock.isInterruptible should avoid duplicated interrupt #52139
Conversation
This file contains hidden or 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
Member
Author
cloud-fan
reviewed
Aug 27, 2025
core/src/main/scala/org/apache/spark/util/UninterruptibleThread.scala
Outdated
Show resolved
Hide resolved
Contributor
|
Can the existing tests verify this fix? |
Member
Author
@cloud-fan Yes. |
cloud-fan
approved these changes
Aug 27, 2025
Member
|
@Ngone51 What is the assumption? If thread A or thread B interrupts |
vrozov
reviewed
Aug 27, 2025
core/src/main/scala/org/apache/spark/util/UninterruptibleThread.scala
Outdated
Show resolved
Hide resolved
vrozov
reviewed
Aug 27, 2025
core/src/main/scala/org/apache/spark/util/UninterruptibleThread.scala
Outdated
Show resolved
Hide resolved
core/src/main/scala/org/apache/spark/util/UninterruptibleThread.scala
Outdated
Show resolved
Hide resolved
Member
Author
Contributor
|
@Ngone51 since this needs manual test, have you verified the new fix locally? |
Member
Author
|
@cloud-fan Yes, it still works. |
Contributor
|
thanks, merging to master/4.0! |
cloud-fan
pushed a commit
that referenced
this pull request
Aug 29, 2025
…duplicated interrupt ### What changes were proposed in this pull request? This PR fixes `UninterruptibleLock.isInterruptible` to avoid duplicated interruption when the thread is already interrupted. ### Why are the changes needed? The "uninterruptible" semantic of `UninterruptibleThread`is broken (i.e., `UninterruptibleThread` is interruptible even if it's under `runUninterruptibly`) after the fix #50594. The probelm is that the state of `shouldInterruptThread` becomes unsafe when there are multiple interrupts concurrently. For example, thread A could interrupt UninterruptibleThread ut first before UninterruptibleThread enters `runUninterruptibly`. Right after that, another thread B starts to invoke ut.interrupt() and pass through `uninterruptibleLock.isInterruptible` (becasue at this point, `shouldInterruptThread = uninterruptible = false`). Before thread B invokes `super.interrupt()`, UninterruptibleThread ut enters `runUninterruptibly` and pass through `uninterruptibleLock.getAndSetUninterruptible` and set `uninterruptible = true`. Then, thread ut continues the check `uninterruptibleLock.isInterruptPending`. However, `uninterruptibleLock.isInterruptPending` return false at this point (due to `shouldInterruptThread = Thread.interrupted = true`) even though thread B is actully interrupting. *As a result, the state of `shouldInterruptThread` becomes inconsistent between thread B and thread ut.* Then, as `uninterruptibleLock.isInterruptPending` returns false, ut to continute to execute `f`. At the same time, thread B invokes `super.interrupt()`, and `f` could be interrupted ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Manually tested. The issue can be easily reproduced if we run `UninterruptibleThreadSuite.stress test` for 100 times in a row: ``` [info] true did not equal false (UninterruptibleThreadSuite.scala:208) [info] org.scalatest.exceptions.TestFailedException: [info] at org.scalatest.Assertions.newAssertionFailedException(Assertions.scala:472) [info] at org.scalatest.Assertions.newAssertionFailedException$(Assertions.scala:471) [info] at org.scalatest.Assertions$.newAssertionFailedException(Assertions.scala:1231) [info] at org.scalatest.Assertions$AssertionsHelper.macroAssert(Assertions.scala:1295) [info] at org.apache.spark.util.UninterruptibleThreadSuite.$anonfun$new$7(UninterruptibleThreadSuite.scala:208) ... ``` And the issue is gone after the fix. ### Was this patch authored or co-authored using generative AI tooling? No Closes #52139 from Ngone51/fix-uninterruptiable. Authored-by: Yi Wu <yi.wu@databricks.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com> (cherry picked from commit 78871d7) Signed-off-by: Wenchen Fan <wenchen@databricks.com>
Member
Author
|
Thanks! @cloud-fan @vrozov |
zifeif2
pushed a commit
to zifeif2/spark
that referenced
this pull request
Nov 14, 2025
…duplicated interrupt ### What changes were proposed in this pull request? This PR fixes `UninterruptibleLock.isInterruptible` to avoid duplicated interruption when the thread is already interrupted. ### Why are the changes needed? The "uninterruptible" semantic of `UninterruptibleThread`is broken (i.e., `UninterruptibleThread` is interruptible even if it's under `runUninterruptibly`) after the fix apache#50594. The probelm is that the state of `shouldInterruptThread` becomes unsafe when there are multiple interrupts concurrently. For example, thread A could interrupt UninterruptibleThread ut first before UninterruptibleThread enters `runUninterruptibly`. Right after that, another thread B starts to invoke ut.interrupt() and pass through `uninterruptibleLock.isInterruptible` (becasue at this point, `shouldInterruptThread = uninterruptible = false`). Before thread B invokes `super.interrupt()`, UninterruptibleThread ut enters `runUninterruptibly` and pass through `uninterruptibleLock.getAndSetUninterruptible` and set `uninterruptible = true`. Then, thread ut continues the check `uninterruptibleLock.isInterruptPending`. However, `uninterruptibleLock.isInterruptPending` return false at this point (due to `shouldInterruptThread = Thread.interrupted = true`) even though thread B is actully interrupting. *As a result, the state of `shouldInterruptThread` becomes inconsistent between thread B and thread ut.* Then, as `uninterruptibleLock.isInterruptPending` returns false, ut to continute to execute `f`. At the same time, thread B invokes `super.interrupt()`, and `f` could be interrupted ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Manually tested. The issue can be easily reproduced if we run `UninterruptibleThreadSuite.stress test` for 100 times in a row: ``` [info] true did not equal false (UninterruptibleThreadSuite.scala:208) [info] org.scalatest.exceptions.TestFailedException: [info] at org.scalatest.Assertions.newAssertionFailedException(Assertions.scala:472) [info] at org.scalatest.Assertions.newAssertionFailedException$(Assertions.scala:471) [info] at org.scalatest.Assertions$.newAssertionFailedException(Assertions.scala:1231) [info] at org.scalatest.Assertions$AssertionsHelper.macroAssert(Assertions.scala:1295) [info] at org.apache.spark.util.UninterruptibleThreadSuite.$anonfun$new$7(UninterruptibleThreadSuite.scala:208) ... ``` And the issue is gone after the fix. ### Was this patch authored or co-authored using generative AI tooling? No Closes apache#52139 from Ngone51/fix-uninterruptiable. Authored-by: Yi Wu <yi.wu@databricks.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com> (cherry picked from commit 271392a) Signed-off-by: Wenchen Fan <wenchen@databricks.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
What changes were proposed in this pull request?
This PR fixes
UninterruptibleLock.isInterruptibleto avoid duplicated interruption when the thread is already interrupted.Why are the changes needed?
The "uninterruptible" semantic of
UninterruptibleThreadis broken (i.e.,UninterruptibleThreadis interruptible even if it's underrunUninterruptibly) after the fix #50594. The probelm is that the state ofshouldInterruptThreadbecomes unsafe when there are multiple interrupts concurrently.For example, thread A could interrupt UninterruptibleThread ut first before UninterruptibleThread enters
runUninterruptibly. Right after that, another thread B starts to invoke ut.interrupt() and pass throughuninterruptibleLock.isInterruptible(becasue at this point,shouldInterruptThread = uninterruptible = false). Before thread B invokessuper.interrupt(), UninterruptibleThread ut entersrunUninterruptiblyand pass throughuninterruptibleLock.getAndSetUninterruptibleand setuninterruptible = true. Then, thread ut continues the checkuninterruptibleLock.isInterruptPending. However,uninterruptibleLock.isInterruptPendingreturn false at this point (due toshouldInterruptThread = Thread.interrupted = true) even though thread B is actully interrupting. As a result, the state ofshouldInterruptThreadbecomes inconsistent between thread B and thread ut. Then, asuninterruptibleLock.isInterruptPendingreturns false, ut to continute to executef. At the same time, thread B invokessuper.interrupt(), andfcould be interruptedDoes this PR introduce any user-facing change?
No.
How was this patch tested?
Manually tested. The issue can be easily reproduced if we run
UninterruptibleThreadSuite.stress testfor 100 times in a row:And the issue is gone after the fix.
Was this patch authored or co-authored using generative AI tooling?
No