Conversation
Tests verify: 1. All ERA files are parseable 2. All blocks have correct tree-hash roots 3. ERA boundary states are available at correct slots 4. Import works and block root index is populated 5. All imported blocks are fetchable and have correct roots 6. Final head matches expected value 7. Genesis state integrity (1024 validators, slot 0) Test vectors: 13 ERA files, 767 blocks, 832 slots, minimal preset
There was a problem hiding this comment.
What's the point of this test? If you load all era files and get to a node that has the same head as the metadata -> Then call read all blocks from DB and check their hash chain you are good.
There was a problem hiding this comment.
Redundant, collapse with test mentioned above that loads all era files and checks the blocks hash chains in the DB post import
There was a problem hiding this comment.
Just do this checks in the same test mentioned above. In 1 test:
- import all era files
- check blocks hash chains
- check states have the expected roots
There was a problem hiding this comment.
Why so much code? The files should have the same name, just use that to compare byte for byte
There was a problem hiding this comment.
Why? Just do basic equality of the bytes
There was a problem hiding this comment.
I don't like this tests, they should cover a block having the wrong block root.
| err.contains("slot mismatch"), | ||
| "expected slot mismatch error: {err}" | ||
| ); | ||
| } |
There was a problem hiding this comment.
For the corrupt ones you need to test for this errors specifically:
era root mismatch for era {era_number}: expected {expected_root:?}, got {actual_root:?}
era state slot mismatch: expected {end_slot}, got {}
block summary root post-capella mismatch for era
block summary root pre-capella mismatch for era {}: {:?} != {:?}
block root mismatch at slot {slot}: expected {expected_block_root:?}, got {block_root:?}
Plus implement verification that the state_root matches from the initial trusted state root supplied by the user. For that mutate a state by changing a single state.balances[i] value
- 3 new corrupt ERA files: era0-wrong-root, era8-corrupt-block-summary, era2-wrong-block-root - 3 new test functions: rejects_wrong_era_root, rejects_corrupt_block_summary, rejects_wrong_block_root - Python generator for reproducible corruption (create_corrupt_files.py) - All 8 corruption tests passing
| era_dir | ||
| .import_era_file(&store, era, &spec, None, None) | ||
| .unwrap_or_else(|e| panic!("import ERA {era}: {e}")); | ||
| } |
There was a problem hiding this comment.
Use the builder code from beacon_node/beacon_chain/src/builder.rs don't re-implement it in the test
There was a problem hiding this comment.
WTF is this verification? You need to check that the chain of blocks is correct, read all blocks in DB and check that they form hash chain. Check that the stored states are the expected states, and that the head matches the test metadata
| /// The trusted state root feature catches it. | ||
| #[test] | ||
| fn era_rejects_mutated_state_with_trusted_root() { | ||
| let tmp = era_dir_with_corrupt("era3-wrong-state-root.era", "-00003-"); |
There was a problem hiding this comment.
How did you generate this file?
1 participant
Issue Addressed
Which issue # does this PR address?
Proposed Changes
Please list or describe the changes introduced by this PR.
Additional Info
Please provide any additional information. For example, future considerations
or information useful for reviewers.