erts: Fix crash in ETS compressed tables #8683
Conversation
CT Test Results 4 files 227 suites 1h 20m 41s ⏱️ Results for commit 36dd893. ♻️ This comment has been updated with latest results. To speed up review, make sure that you have read Contributing to Erlang/OTP and that all checks pass. See the TESTING and DEVELOPMENT HowTo guides for details about how to run test locally. Artifacts// Erlang/OTP Github Action Bot |
The compression was a bit too good, converting ErlSubBits to
ErlHeapBits whenever that resulted in a smaller result, breaking
an assumption that ETS compressed tables made:
size_object(original) == size_object(decompressed)
This commit fixes it by not being as clever, naively encoding
ErlSubBits and the likes as-is regardless of size.
jhogberg
force-pushed
the
john/erts/fix-compressed-ets-crash/OTP-19176
branch
from
9639403
to
36dd893
Compare
|
I'm running into this issue in production and am looking forward to the fix. Any updates on the status? |
|
The fix will be included in the next patch for the 27 track. In the meantime, you can include this branch in your builds. :-) |
Fair enough, was just wondering how solid this fix is as you mentioned adding a test for the bug. |
|
I added a test the morning after, it appears solid so far. :-) |
The compression was a bit too good, converting ErlSubBits to ErlHeapBits whenever that resulted in a smaller result, breaking an assumption that ETS compressed tables made (
size_object(original) == size_object(decompressed)).This commit fixes it by not being as clever, naively encoding ErlSubBits and the likes as-is regardless of size.
(Edit: the PR currently lacks a test for this bug, but it's late and I want CI to run overnight, I'll add one tomorrow)