Locate targets under mutex lock #2976
Merged
+557
−269
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
vinistock
added
bugfix
Graphite App
vinistock
force-pushed
the
12-11-locate_targets_under_mutex_lock
branch
from
006f5b1 to
6980656
Compare
st0012
reviewed
Dec 16, 2024
vinistock
force-pushed
the
12-11-locate_targets_under_mutex_lock
branch
from
6980656 to
b51049e
Compare
|
@st0012 I made a new attempt at this one and split by commit. I realized that documents needed more than one piece of the global state (encoding + mutex), so we can actually pass the state down without making it too messy. The benefit is that now there's no way to locate targets bypassing the lock, as you had suggested. Let me know what you think! |
vinistock
force-pushed
the
12-11-locate_targets_under_mutex_lock
branch
from
b51049e to
a227dd4
Compare
Merge activity
|
This was referenced Jan 7, 2025
Closed
vinistock
added a commit
that referenced
this pull request
Jan 9, 2025
### Motivation Closes #3027 The deadlock bug was introduced in #2976. Essentially, if the editor opens a URI for a non-existing file, we try to acquire the same lock twice. Once [here](https://github.com/Shopify/ruby-lsp/blob/17d5fc733b758bbbedaf36fb17ac3914548086bf/lib/ruby_lsp/base_server.rb#L53) to verify if the request requires a document to be parsed and the second time [here](https://github.com/Shopify/ruby-lsp/blob/17d5fc733b758bbbedaf36fb17ac3914548086bf/lib/ruby_lsp/store.rb#L68) when we read the file from disk and save it in the store. Trying to acquire the same mutex lock twice raises the deadlock error. ### Implementation I stopped synchronizing anything inside basic store operations. It just increases the chance of mistakes like these. I moved the responsibility of synchronizing back to the server. ### Automated Tests To catch this regression, we need a client/server integration test because the first synchronization happens only by reading a request from the STDIN pipe. I added one to ensure we don't hit the same issue again.
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.
Motivation
I'm hoping this is the solution for #2446
My hypothesis is that we're seeing a concurrency issue where threads switch in the middle of locating targets and then the document gets mutated. That would render the location we're currently searching for invalid, which can then lead to infinite loops.
Implementation
My idea is to move our mutex into the global state so that we can use it in more places. Then we lock the mutex while locating targets to avoid having any document edits be applied in the middle.
Note: I experimented with passing the mutex down to the document instances, but it was a bit messier.