core: fix data races in unit tests #609
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
tien-nguyen-sm
approved these changes
Oct 21, 2024
We get this data race when running tests
WARNING: DATA RACE
Write at 0x00c0117bbf40 by goroutine 77211:
github.com/ethereum/go-ethereum/core.TestInsertChainWithSidecars()
/home/runner/work/ronin/ronin/core/blockchain_test.go:4215 +0x49fd
testing.tRunner()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1576 +0x216
testing.(*T).Run.func1()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1629 +0x47
Previous write at 0x00c0117bbf40 by goroutine 82133:
github.com/ethereum/go-ethereum/core.TestInsertChainWithSidecars.func6.1()
/home/runner/work/ronin/ronin/core/blockchain_test.go:4179 +0x5a4
testing.tRunner()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1576 +0x216
testing.(*T).Run.func1()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1629 +0x47
The data race happens because 2 goroutines use the same variables. This commit
makes them use their own variables.
We get this data race when running tests
WARNING: DATA RACE
Read at 0x000001faa090 by goroutine 23:
github.com/ethereum/go-ethereum/core.NewBlockChain()
/home/runner/work/ronin/ronin/core/blockchain.go:247 +0xc4
github.com/ethereum/go-ethereum/eth/protocols/eth.newTestBackendWithGenerator()
/home/runner/work/ronin/ronin/eth/protocols/eth/handler_test.go:74 +0x4c4
github.com/ethereum/go-ethereum/eth/protocols/eth.testGetBlockReceipts()
/home/runner/work/ronin/ronin/eth/protocols/eth/handler_test.go:569 +0x1f5
github.com/ethereum/go-ethereum/eth/protocols/eth.TestGetBlockReceipts66()
/home/runner/work/ronin/ronin/eth/protocols/eth/handler_test.go:528 +0x33
testing.tRunner()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1576 +0x216
testing.(*T).Run.func1()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1629 +0x47
Previous write at 0x000001faa090 by goroutine 19:
github.com/ethereum/go-ethereum/core.NewBlockChain()
/home/runner/work/ronin/ronin/core/blockchain.go:248 +0xe4
github.com/ethereum/go-ethereum/eth/protocols/eth.newTestBackendWithGenerator()
/home/runner/work/ronin/ronin/eth/protocols/eth/handler_test.go:74 +0x4c4
github.com/ethereum/go-ethereum/eth/protocols/eth.newTestBackend()
/home/runner/work/ronin/ronin/eth/protocols/eth/handler_test.go:61 +0x64
github.com/ethereum/go-ethereum/eth/protocols/eth.testGetBlockHeaders()
/home/runner/work/ronin/ronin/eth/protocols/eth/handler_test.go:126 +0x53
github.com/ethereum/go-ethereum/eth/protocols/eth.TestGetBlockHeaders66()
/home/runner/work/ronin/ronin/eth/protocols/eth/handler_test.go:121 +0x33
testing.tRunner()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1576 +0x216
testing.(*T).Run.func1()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1629 +0x47
In the tests, the cacheConfig passed to NewBlockChain is nil so the
defaultCacheConfig is used. However, the defaultCacheConfig.TriesInMemory is 0,
so we fall into the path that set the cacheConfig.TriesInMemory to
DefaultTriesInMemory which is 128. This set is made to the global
defaultCacheConfig which causes data race with other accesses. This commit set
the defaultCacheConfig.TriesInMemory to DefaultTriesInMemory.
We get multiples data races while running vote pool tests
WARNING: DATA RACE
Read at 0x00c0002a3ce0 by goroutine 19:
github.com/ethereum/go-ethereum/core/vote.(*VotePool).verifyStructureSizeOfVotePool()
/home/runner/work/ronin/ronin/core/vote/vote_pool_test.go:77 +0xaa
github.com/ethereum/go-ethereum/core/vote.testVotePool()
/home/runner/work/ronin/ronin/core/vote/vote_pool_test.go:160 +0xe84
github.com/ethereum/go-ethereum/core/vote.TestValidVotePool()
/home/runner/work/ronin/ronin/core/vote/vote_pool_test.go:85 +0x33
testing.tRunner()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1576 +0x216
testing.(*T).Run.func1()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1629 +0x47
Previous write at 0x00c0002a3ce0 by goroutine 23:
runtime.mapassign()
/opt/hostedtoolcache/go/1.20.10/x64/src/runtime/map.go:578 +0x0
github.com/ethereum/go-ethereum/core/vote.(*VotePool).putVote()
/home/runner/work/ronin/ronin/core/vote/vote_pool.go:236 +0x346
github.com/ethereum/go-ethereum/core/vote.(*VotePool).putIntoVotePool()
/home/runner/work/ronin/ronin/core/vote/vote_pool.go:212 +0x978
github.com/ethereum/go-ethereum/core/vote.(*VotePool).loop()
/home/runner/work/ronin/ronin/core/vote/vote_pool.go:123 +0x504
github.com/ethereum/go-ethereum/core/vote.NewVotePool.func1()
/home/runner/work/ronin/ronin/core/vote/vote_pool.go:98 +0x39
Currently, when getting vote pool's stats for checking in unit test we access
these fields directly without holding lock. This commit creates some helper
funtion to get these stats safely with read lock hold.
The current timeout is 5s which is not enough when running tests with race detector. Increase that timeout to 15s.
We get this data race when running TestTable_BucketIPLimit
Write at 0x00c00029c000 by goroutine 64:
github.com/ethereum/go-ethereum/p2p/discover.(*Table).addSeenNodeSync()
/home/runner/work/ronin/ronin/p2p/discover/table.go:570 +0x70a
github.com/ethereum/go-ethereum/p2p/discover.(*Table).addSeenNode.func1()
/home/runner/work/ronin/ronin/p2p/discover/table.go:527 +0x47
Previous read at 0x00c00029c000 by goroutine 57:
github.com/ethereum/go-ethereum/p2p/discover.checkIPLimitInvariant()
/home/runner/work/ronin/ronin/p2p/discover/table_test.go:187 +0x105
github.com/ethereum/go-ethereum/p2p/discover.TestTable_BucketIPLimit()
/home/runner/work/ronin/ronin/p2p/discover/table_test.go:177 +0x2e4
testing.tRunner()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1576 +0x216
testing.(*T).Run.func1()
/opt/hostedtoolcache/go/1.20.10/x64/src/testing/testing.go:1629 +0x47
This commit changes TestTable_BucketIPLimit to use synchronous version of
addSeenNode to avoid the data race.
|
Pass all tests with race detector enabled: https://github.com/axieinfinity/ronin/actions/runs/11437046076/job/31815765570 |
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.
We get this data race when running tests
The data race happens because 2 goroutines use the same variables. This commit
makes them use their own variables.
We get this data race when running tests
In the tests, the cacheConfig passed to NewBlockChain is nil so the
defaultCacheConfig is used. However, the defaultCacheConfig.TriesInMemory is 0,
so we fall into the path that set the cacheConfig.TriesInMemory to
DefaultTriesInMemory which is 128. This set is made to the global
defaultCacheConfig which causes data race with other accesses. This commit set
the defaultCacheConfig.TriesInMemory to DefaultTriesInMemory.
We get multiples data races while running vote pool tests
Currently, when getting vote pool's stats for checking in unit test we access
these fields directly without holding lock. This commit creates some helper
funtion to get these stats safely with read lock hold.
internal/cmdtest: increase the cli command test timeout
The current timeout is 5s which is not enough when running tests with race
detector. Increase that timeout to 15s.
p2p/discover: use synchronous version of addSeenNode
We get this data race when running TestTable_BucketIPLimit
This commit changes TestTable_BucketIPLimit to use synchronous version of
addSeenNode to avoid the data race.