Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Free the SSL context after unloading the provider libs #133

Merged
merged 1 commit into from
Jan 30, 2024

Conversation

dprandle
Copy link
Contributor

Otherwise crashes on linux in dtor

  • [ x] Accept to license the library and tools code under the terms
    of the LGPL 2.0 or later
  • [ x] Accept to license the library and tools code under the terms
    of the MPL 2.0
  • [ x] Accept to license the test code under the terms
    of the MIT-0
  • [ x] Read contributions guidelines
  • [ x] Checked coding style
  • [ x] The commits sequence is clean without work in progress/bugged revisions and merge commits. In doubt, squash the commits and/or rebase master

This is the fix for SSL context/provider destruction order

Otherwise crashes on linux in dtor
@ceztko ceztko merged commit 189192e into podofo:master Jan 30, 2024
2 of 3 checks passed
@ceztko
Copy link
Contributor

ceztko commented Jan 30, 2024

It makes sense to me. Just a question: have you ever had a crash related to this?

@dprandle
Copy link
Contributor Author

It makes sense to me. Just a question: have you ever had a crash related to this?

Yes - its how I found it.

It does not crash on Mac, but it does crash on Linux (Ubuntu 22.04.3 LTS) every time. My guess is the difference between the way the two OSs handle memory - it likely happens to work on Mac (and perhaps windows or other OSs).

@ceztko
Copy link
Contributor

ceztko commented Jan 30, 2024

Ok, I asked because we have CI with tests on linux and it doesn't crash. Still, it makes sense to me to have resource disposal in the opposite order of creation, hence I don't need to see a reproduction with my eyes.

@dprandle
Copy link
Contributor Author

dprandle commented Feb 8, 2024

This doesn't surprise me - those types of tests in my experience can only catch so much. We have several different SSL contexts being created and used by several different libraries, each of which utilize several threads. Its unlikely CL tests would catch the issues that may arise in this situation.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants