Add get_checkpoint_block to the fork choice spec #3308

saltiniroberto · 2023-03-28T04:59:24Z

This PR introduces the function get_checkpoint_block and replaces any computation of checkpoint blocks in the fork choice spec with a call to the newly introduced function.

The objective of this PR is to increase the readability of the spec by better clarifying when we are interested in calculating an checkpoint block.

ralexstokes

looks good just from a refactor perspective and I like the increased semantic content

hwwhww

It looks like a nice-to-have to me. No strong opinion here.

/cc @adiasg

Some CircleCI didn't start with the latest commit. It may need a kick with a new commit. 😬

hwwhww · 2023-03-28T07:53:09Z

specs/phase0/fork-choice.md

@@ -506,7 +516,7 @@ def on_block(store: Store, signed_block: SignedBeaconBlock) -> None:
    finalized_slot = compute_start_slot_at_epoch(store.finalized_checkpoint.epoch)
    assert block.slot > finalized_slot
    # Check block is a descendant of the finalized block at the checkpoint finalized slot
-    assert get_ancestor(store, block.parent_root, finalized_slot) == store.finalized_checkpoint.root


it seems no need to modify this one

@hwwhww Why do you think that there may be no need to modify this one?

because we need to define the variable finalized_slot to check assert block.slot > finalized_slot anyway, the change of this change doesn't simplify much here.

The objective of this PR is not so much about simplifying the code, but rather making explicit when we are computing an epoch boundary block.

For example, in this line, we are not only checking that store.finalized_checkpoint.root is an ancestor of block.parent_root, but also that store.finalized_checkpoint.root is an epoch boundary block in the chain of block.parent_root, which I don't think that it is immediate in the original code.

But the difference between checking whether a block is just an ancestor or it is also an epoch boundary block is significant when performing security analysis on the code.

specs/phase0/fork-choice.md

potuz

You need to change the corresponding files in different forks that changed on_block, eg. Bellatrix

saltiniroberto · 2023-03-30T23:57:22Z

You need to change the corresponding files in different forks that changed on_block, eg. Bellatrix

👍
Done.

ppopth · 2023-04-05T09:14:14Z

Please change the p2p network specs as well.

specs/phase0/p2p-interface.md:  `get_ancestor(store, block.parent_root, compute_start_slot_at_epoch(store.finalized_checkpoint.epoch))
specs/phase0/p2p-interface.md:  `get_ancestor(store, aggregate.data.beacon_block_root, compute_start_slot_at_epoch(store.finalized_checkpoint.epoch))
specs/phase0/p2p-interface.md:  `get_ancestor(store, attestation.data.beacon_block_root, compute_start_slot_at_epoch(attestation.data.target.epoch)) == attestation.data.target.root`
specs/phase0/p2p-interface.md:  `get_ancestor(store, attestation.data.beacon_block_root, compute_start_slot_at_epoch(store.finalized_checkpoint.epoch))

ppopth · 2023-04-05T09:19:12Z

We should also update the pyspec correspondingly, if we want to apply this patch.

For example, in tests/core/pyspec/eth2spec/test/phase0/fork_choice/test_on_block.py, we should change the following.

finalized_slot = spec.compute_start_slot_at_epoch(store.finalized_checkpoint.epoch)
ancestor_at_finalized_slot = spec.get_ancestor(store, pre_store_justified_checkpoint_root, finalized_slot)

Please run git grep get_ancestor tests/ to see all the places that need change.

djrtwo · 2023-04-06T13:32:01Z

specs/deneb/fork-choice.md

+    assert store.finalized_checkpoint.root == get_ancestor_at_epoch_boundary(
+        store,
+        block.parent_root,
+        store.finalized_checkpoint.epoch,
+    )


Doing an assignment before the assert is preferable to reduce line count and increase readability. ditto at a few other places in this patch

Suggested change

assert store.finalized_checkpoint.root == get_ancestor_at_epoch_boundary(

store,

block.parent_root,

store.finalized_checkpoint.epoch,

)

epoch_boundary_root = get_ancestor_at_epoch_boundary(store, block.parent_root, store.finalized_checkpoint.epoch)

assert store.finalized_checkpoint.root == epoch_boundary_root

Done here in all other places marked with "ditto"

djrtwo · 2023-04-06T13:32:19Z

specs/bellatrix/fork-choice.md

@@ -170,7 +170,11 @@ def on_block(store: Store, signed_block: SignedBeaconBlock) -> None:
    finalized_slot = compute_start_slot_at_epoch(store.finalized_checkpoint.epoch)
    assert block.slot > finalized_slot
    # Check block is a descendant of the finalized block at the checkpoint finalized slot
-    assert get_ancestor(store, block.parent_root, finalized_slot) == store.finalized_checkpoint.root
+    assert store.finalized_checkpoint.root == get_ancestor_at_epoch_boundary(


djrtwo · 2023-04-06T13:32:25Z

specs/phase0/fork-choice.md

    correct_finalized = (
        store.finalized_checkpoint.epoch == GENESIS_EPOCH
-        or store.finalized_checkpoint.root == get_ancestor(store, block_root, finalized_slot)
+        or store.finalized_checkpoint.root == get_ancestor_at_epoch_boundary(


djrtwo · 2023-04-06T13:32:35Z

specs/phase0/fork-choice.md

@@ -506,7 +520,11 @@ def on_block(store: Store, signed_block: SignedBeaconBlock) -> None:
    finalized_slot = compute_start_slot_at_epoch(store.finalized_checkpoint.epoch)
    assert block.slot > finalized_slot
    # Check block is a descendant of the finalized block at the checkpoint finalized slot
-    assert get_ancestor(store, block.parent_root, finalized_slot) == store.finalized_checkpoint.root
+    assert store.finalized_checkpoint.root == get_ancestor_at_epoch_boundary(


adiasg

The fork choice & p2p spec do not refer to epoch boundary blocks before this PR. More generally, the spec refers to epoch boundary blocks as part of Checkpoints.

I'd prefer to avoid introducing the "epoch boundary block" nomenclature here, and instead rename get_ancestor_at_epoch_boundary to get_checkpoint_block.

saltiniroberto · 2023-04-18T05:51:32Z

make test report was

================================================================================================================================================= short test summary info ==================================================================================================================================================
FAILED eth2spec/test/deneb/unittests/polynomial_commitments/test_polynomial_commitments.py::test_validate_kzg_g1_not_in_g1 - AssertionError: expected an assertion error, but got none.
FAILED eth2spec/test/deneb/unittests/polynomial_commitments/test_polynomial_commitments.py::test_validate_kzg_g1_not_on_curve - AssertionError: expected an assertion error, but got none.
================================================================================================================================= 2 failed, 806 passed, 146 skipped in 2537.18s (0:42:17) ==================================================================================================================================

saltiniroberto · 2023-04-18T06:00:55Z

We should also update the pyspec correspondingly, if we want to apply this patch.

For example, in tests/core/pyspec/eth2spec/test/phase0/fork_choice/test_on_block.py, we should change the following.
finalized_slot = spec.compute_start_slot_at_epoch(store.finalized_checkpoint.epoch)
ancestor_at_finalized_slot = spec.get_ancestor(store, pre_store_justified_checkpoint_root, finalized_slot)
Please run git grep get_ancestor tests/ to see all the places that need change.

Done

saltiniroberto · 2023-04-18T06:01:08Z

The fork choice & p2p spec do not refer to epoch boundary blocks before this PR. More generally, the spec refers to epoch boundary blocks as part of Checkpoints.

I'd prefer to avoid introducing the "epoch boundary block" nomenclature here, and instead rename get_ancestor_at_epoch_boundary to get_checkpoint_block.

Done

saltiniroberto added 2 commits March 28, 2023 15:43

Introduce get_epoch_boundary_block

201f113

Add toc

ddbd82e

ralexstokes reviewed Mar 28, 2023

View reviewed changes

hwwhww reviewed Mar 28, 2023

View reviewed changes

hwwhww added the scope:fork-choice label Mar 28, 2023

Rename get_epoch_boundary_block to get_ancestor_at_epoch_boundary

c7029ce

mkalinin approved these changes Mar 29, 2023

View reviewed changes

potuz suggested changes Mar 29, 2023

View reviewed changes

Apply changes to Bellatrix and Daneb

e255d09

hwwhww added 2 commits April 5, 2023 11:19

Merge branch 'dev' into pr3308

cdcf458

make linter happy

4cac761

djrtwo reviewed Apr 6, 2023

View reviewed changes

saltiniroberto added 2 commits April 8, 2023 14:00

Apply changes to p2p-interface.md

4138609

Merge remote-tracking branch 'upstream/dev' into add_get_boundary_block

6d7b9e3

adiasg suggested changes Apr 12, 2023

View reviewed changes

saltiniroberto added 6 commits April 18, 2023 13:11

Merge remote-tracking branch 'upstream/dev' into add_get_boundary_block

fdf2dc7

Rename get_ancestor_at_epoch_boundary to get_checkpoint_block

334114d

Break long statement into two statements

36fcb81

Fix copy and past error

c985605

Applied changes to tests

b5bd90d

Fix lint erorrs

313439a

saltiniroberto requested a review from adiasg April 18, 2023 05:48

Fixed doc in get_checkpoint_block

ffb8459

saltiniroberto changed the title ~~Add get_epoch_boundary_block to the fork choice spec~~ Add get_checkpoint_block to the fork choice spec Apr 18, 2023

adiasg approved these changes Apr 26, 2023

View reviewed changes

Merge branch 'dev' into pr3308

a044c0c

hwwhww merged commit acccbd7 into ethereum:dev May 1, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add get_checkpoint_block to the fork choice spec #3308

Add get_checkpoint_block to the fork choice spec #3308

saltiniroberto commented Mar 28, 2023 •

edited

Loading

ralexstokes left a comment

hwwhww left a comment

hwwhww Mar 28, 2023

saltiniroberto Mar 28, 2023

hwwhww Mar 28, 2023

saltiniroberto Mar 29, 2023 •

edited

Loading

saltiniroberto Mar 29, 2023

potuz left a comment

saltiniroberto commented Mar 30, 2023 •

edited

Loading

ppopth commented Apr 5, 2023

ppopth commented Apr 5, 2023 •

edited

Loading

djrtwo Apr 6, 2023

saltiniroberto Apr 18, 2023

djrtwo Apr 6, 2023

djrtwo Apr 6, 2023

djrtwo Apr 6, 2023

adiasg left a comment

saltiniroberto commented Apr 18, 2023

saltiniroberto commented Apr 18, 2023

saltiniroberto commented Apr 18, 2023

Add get_checkpoint_block to the fork choice spec #3308

Add get_checkpoint_block to the fork choice spec #3308

Conversation

saltiniroberto commented Mar 28, 2023 • edited Loading

ralexstokes left a comment

Choose a reason for hiding this comment

hwwhww left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

saltiniroberto Mar 29, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

potuz left a comment

Choose a reason for hiding this comment

saltiniroberto commented Mar 30, 2023 • edited Loading

ppopth commented Apr 5, 2023

ppopth commented Apr 5, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

adiasg left a comment

Choose a reason for hiding this comment

saltiniroberto commented Apr 18, 2023

saltiniroberto commented Apr 18, 2023

saltiniroberto commented Apr 18, 2023

saltiniroberto commented Mar 28, 2023 •

edited

Loading

saltiniroberto Mar 29, 2023 •

edited

Loading

saltiniroberto commented Mar 30, 2023 •

edited

Loading

ppopth commented Apr 5, 2023 •

edited

Loading