Closed JeremyTubongbanua closed 3 months ago
running into a 16 KB leak after establishing SSH connection, converting to draft while I investigate
After establishing an SSH session with 2 fresh atSigns, we get 16KB lost.
Threaded leak summary
==68541== LEAK SUMMARY:
==68541== definitely lost: 16,384 bytes in 2 blocks
==68541== indirectly lost: 0 bytes in 0 blocks
==68541== possibly lost: 272 bytes in 1 blocks
==68541== still reachable: 121,622 bytes in 201 blocks
==68541== suppressed: 0 bytes in 0 blocks
Main process leak summary
==68485== LEAK SUMMARY:
==68485== definitely lost: 16,384 bytes in 2 blocks
==68485== indirectly lost: 0 bytes in 0 blocks
==68485== possibly lost: 0 bytes in 0 blocks
==68485== still reachable: 0 bytes in 0 blocks
==68485== suppressed: 0 bytes in 0 blocks
Full valgrind logs: https://gist.github.com/JeremyTubongbanua/a49a783ea400506d871239873224d1a8
After establishing an SSH session with 2 fresh atSigns, we get 16KB lost.
Threaded leak summary
==68541== LEAK SUMMARY: ==68541== definitely lost: 16,384 bytes in 2 blocks ==68541== indirectly lost: 0 bytes in 0 blocks ==68541== possibly lost: 272 bytes in 1 blocks ==68541== still reachable: 121,622 bytes in 201 blocks ==68541== suppressed: 0 bytes in 0 blocks
Main process leak summary
==68485== LEAK SUMMARY: ==68485== definitely lost: 16,384 bytes in 2 blocks ==68485== indirectly lost: 0 bytes in 0 blocks ==68485== possibly lost: 0 bytes in 0 blocks ==68485== still reachable: 0 bytes in 0 blocks ==68485== suppressed: 0 bytes in 0 blocks
Full valgrind logs: https://gist.github.com/JeremyTubongbanua/a49a783ea400506d871239873224d1a8
The reason for this leak is at_c related. See https://github.com/atsign-foundation/at_c/pull/328 for an upcoming fix.
Next commit will most likely be a commit hash change for atsdk so that the C Daemon will use the latest trunk hash for the memory leak fix.
- What I did
- How to verify it
new leak summary (after closing thread)
notice that the 585 bytes that were usually suppressed are no longer lost.
- Description for the changelog fix: no longer lose 585B every time SSH session is established