guard vfsMap with a mutex (upstream psanford/sqlite3vfs#19) - #1
Merged
Conversation
registering a new vfs writes vfsMap while callbacks read it from other threads, which trips the runtime concurrent map access check
Exercises concurrent RegisterVFS (vfsMap write in newVFS) against in-flight VFS callbacks (vfsMap read in vfsFromC) driven through real sqlite connections. Fails under -race without the vfsMux fix from upstream PR psanford#19.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Merges upstream psanford/sqlite3vfs#19 ("guard vfsMap with a mutex" by @maxencehenneron) into our fork, plus a regression test.
vfsMapwas written bynewVFS(onRegisterVFS) and read byvfsFromCon every VFS callback with no synchronization. A registration racing an in-flight callback is a Go runtime "concurrent map read and map write" — a fatal, unrecoverable whole-process crash. The fix adds async.RWMutex(matching the existingfileMux/fileMappattern in the same file): write-lock innewVFS, read-lock invfsFromC.Evaluation
vfsMapaccess sites exist in the repo (write innewVFS, read invfsFromC); the patch covers both. Initialization is package-init, safe.vfsmap_race_test.go) driving real sqlite connections against a tmp VFS while registering VFSes from another goroutine:main+go test -race: DATA RACE reported betweennewVFSandvfsFromC(viagoVFSFullPathnamecgo callback), matching the production crash trace in the upstream PR descriptiongo test -race -count=2: ok-race;go vetandgofmtclean.Why this matters for the litestream fork
Litestream registers/uses VFSes from concurrent goroutines, so this crash mode is reachable in our deployment path.