[core] Queue::drop is acquiring the snatch guard as well
#8691
+5
−2
There are no files selected for viewing
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -854,7 +854,11 @@ impl Device { | |
| user_closures.mappings, | ||
| user_closures.blas_compact_ready, | ||
| queue_empty, | ||
| ) = queue_result; | ||
| // Queue::drop is acquiring the snatch lock as well | ||
| drop(snatch_guard); | ||
| } else { | ||
| drop(snatch_guard); | ||
| }; | ||
|
|
||
| // Based on the queue empty status, and the current finished submission index, determine the result of the poll. | ||
|
|
@@ -909,7 +913,6 @@ impl Device { | |
|
|
||
| // Don't hold the locks while calling release_gpu_resources. | ||
| drop(fence); | ||
|
Member
There was a problem hiding this comment. question: I think it's fine—but I'm not positive—to no longer be locking the snatch lock when we take the device lost closure. Tagging in @cwfitzgerald to double-check my conclusion here. Does that sound right, Connor?
Member
There was a problem hiding this comment. I don't see any reason why this would be problematic. The snatch lock doesn't guard the device lost closure in any way. |
||
|
|
||
| if should_release_gpu_resource { | ||
| self.release_gpu_resources(); | ||
|
|
||
Oops, something went wrong.
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Outdated suggestions cannot be applied.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.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
question: It seems like the potential for
Queue::dropgetting called here is when:Device::maintainon this thread (let's call it thread 1) while holding a last strong ref. to the queue on another thread (let's call that thread 2).queue, and follow the branch to callqueue.maintain(…).Does that sound right? Are you using
wgpufrom multiple threads?There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Correct - thread 1 is calling
Instance::poll_all(true)in a loop, and other threads 2+ create/drop queuesThere was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Could we add a (like eric's) to say:
This kind of thing is really hard to reason about and the effects are deeply non-local so retaining as much of the knowledge as we can about the decisions that were made is super important.