ConcurrentWeakMapOperationStressTest.testOperation failed on CI

[qwwdfsad]

https://teamcity.jetbrains.com/viewLog.html?buildId=3878037&tab=buildResultsDiv&buildTypeId=KotlinTools_KotlinxCoroutines_NightlyStress

java.lang.AssertionError: expected:<0> but was:<1>
	at org.junit.Assert.fail(Assert.java:89)
	at org.junit.Assert.failNotEquals(Assert.java:835)
	at org.junit.Assert.assertEquals(Assert.java:120)
	at kotlin.test.junit.JUnitAsserter.assertEquals(JUnitSupport.kt:32)
	at kotlin.test.AssertionsKt__AssertionsKt.assertEquals(Assertions.kt:63)
	at kotlin.test.AssertionsKt.assertEquals(Unknown Source)
	at kotlin.test.AssertionsKt__AssertionsKt.assertEquals$default(Assertions.kt:62)
	at kotlin.test.AssertionsKt.assertEquals$default(Unknown Source)
	at kotlinx.coroutines.internal.ConcurrentWeakMapOperationStressTest.testOperations(ConcurrentWeakMapOperationStressTest.kt:71)

[qwwdfsad]

It doesn't reproduce locally and also has nothing in common with #2187

[qwwdfsad]

Another observation is that assertEquals(v, m[k]) failed, so it may appear that value was prematurely deleted.
But it's not the case, because the very last assert assertEquals(0, m.size) failed with 0 != 1, which likely means that it is possible to observe map in such intermediates state that the key is here, but get returns null

[dkhalanskyjb]

It doesn't reproduce locally

Just to double-check, I tried running this test in a loop on Linux, and 20 minutes later, with nSeconds = 75, the bug did not surface.

[qwwdfsad]

Sorry to bother, I've actually found the bug and working on a more concise reproducer

[qwwdfsad]

The longer I stare at it, the less I believe it's a bug in our implementation. I've studied the implementation once again, and it doesn't seem possible it can reasonably make this very test fail, the test itself also looks reasonable.

I've added an additional diagnostics, let's see if it fails ever again

[dkhalanskyjb]

I left the test to run in a loop for 2 days and 20 hours, but it still didn't crash.

[qwwdfsad]

Closing as "cannot reproduce", let's reopen it if it indeed was a real failure that will appear again and not a random fluke