wireguard: Fix timeout in unit test #16001
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
gandro
added
area/encryption
|
As this PR only touches unit test code, I will not run the full CI, just the Travis and runtime tests. |
gandro
force-pushed
the
pr/gandro/fix-wireguard-tests
branch
from
7e797e8
to
cc4f15c
Compare
gandro
added
kind/bug/CI
|
test-runtime |
This commit fixes a deadlock in a unit test which ironically tests for deadlocks. The unit test in question ensures that the `wireguard.Agent` `UpdatePeer` method does not create a deadlock if a concurrent IPCache update is performed. The previous version of this test wanted to ensure this by taking a read-lock on the IPCache, which would ensure that only `UpdatePeer` would make progress (as it also just takes an RLock). However, that approach could lead to a timeout when `ipCache.Upsert` was invoked before `wgAgent.UpdatePeer`, as due to the FIFO nature of underlying mutex implementation, `UpdatePeer` will never obtain an `RLock` if there is a waiting writer. This commit addresses this by taking the `wgAgent` lock instead. This means that `UpdatePeer` will lock the IPCache and then wait for the `wgAgent` lock to become available. Any concurrent IPCache updates will also be blocked until `UpdatePeer` has finished as before. This commit also introduces some additional checks to ensure the spawned go routines have actually been scheduled. This is still best effort, as there is easy way to ensure that a certain method is blocked on a particular mutex. Signed-off-by: Sebastian Wicki <sebastian@isovalent.com>
gandro
force-pushed
the
pr/gandro/fix-wireguard-tests
branch
from
cc4f15c
to
fc3a3a0
Compare
|
Travis is green, I'm marking this ready to merge. As mentioned above, this commit only changes unit tests ran by Travis, no impact on Jenkins/GithubActions CI or the Cilium code itself. |
gandro
added
the
ready-to-merge
aanm
moved this from Needs backport from master
to Backport done to v1.10
in 1.10.0-rc2
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.
This commit fixes a deadlock in a unit test which ironically tests for
deadlocks. The unit test in question ensures that the
wireguard.AgentUpdatePeermethod does not create a deadlock if a concurrent IPCacheupdate is performed.
The previous version of this test wanted to ensure this by taking a
read-lock on the IPCache, which would ensure that only
UpdatePeerwould make progress (as it also just takes an RLock). However, that
approach could lead to a timeout when
ipCache.Upsertwas invokedbefore
wgAgent.UpdatePeer, as due to the FIFO nature of underlyingmutex implementation,
UpdatePeerwill never obtain anRLockif thereis a waiting writer.
This commit addresses this by taking the
wgAgentlock instead. Thismeans that
UpdatePeerwill lock the IPCache and then wait for thewgAgentlock to become available. Any concurrent IPCache updates willalso be blocked until
UpdatePeerhas finished as before.This commit also introduces some additional checks to ensure the spawned
go routines have actually been scheduled. This is still best effort, as
there is easy way to ensure that a certain method is blocked on a
particular mutex.
Fixes: #15937