This repository has been archived by the owner on Nov 17, 2023. It is now read-only.
Fix memory error in threaded engine #6084
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
|
Great catch! |
|
This rare failure problem has bugged me for more than a year and I have to say you make my day! Thanks for the great observation |
|
@GaiYu0 Does this patch fix your problem with threaded engine as well? |
|
@tqchen You are welcome :) |
|
The fix works for me. @sifmelcara Thank you very much! |
Guneet-Dhillon
pushed a commit
to Guneet-Dhillon/mxnet
that referenced
this pull request
Sep 13, 2017
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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 bug make mxnet frequently get segfault on my machine, after some debugging and code tracing I realized it happens in the following condition:
DeleteOperatorDeleteOperatorpushesThreadedOpr::Delete(threaded_opr);into the threaded engine.ThreadedOpr::Delete(threaded_opr);is not satisfied because operationAis using a dependent variable ofThreadedOpr::Delete(threaded_opr);, soThreadedOpr::Delete(threaded_opr)waitsA.Afinishes and callsThreadedEngine::OnComplete(A).Aas complete (these code).threaded_oprpoints to is destroyed and the memory address may be occupied by other objects.threaded_opr->temporaryin this line, and get a non-zero garbage value, then executes ThreadedOpr::Delete(threaded_opr); and destroy an innocent objectthreaded_oprpoints to. (Probably also athreaded_oprbecause object pool being used)This patch fixes the issue.
cc original author @tqchen