Fix POS round state logic - #210
Conversation
The POS state machine previously detected staleness by comparing
heights only. An equal-height fork switch (pop block N, adopt the competing
block N) leaves the chain height unchanged, so the state machine kept using
`wait_for_next_block.top_hash`, which by then referred to a block deleted
from both the main and the alt DB (switch_to_alternative_blockchain() is
called with keep_disconnected_chain=false on the POS weight/checkpoint
reorg paths). The next entropy lookup then failed with:
"Failed to find block <hash>"
"Failed to get quorum entropy for POS, next block parent <hash>"
and the node sat out POS participation until the next height. Comparing the
top hash as well restarts the POS stages against the new tip immediately.
📝 WalkthroughSummary by CodeRabbit
WalkthroughThe POS loop now detects same-height tip replacements, handles unavailable quorum entropy, and updates test fault-injection height handling. Mainnet seed hostnames changed for three entries. The CMake project version changed from 7.0.2 to 7.0.3. ChangesPOS chain handling
Release and network configuration
Estimated code review effort: 3 (Moderate) | ~20 minutes Sequence Diagram(s)sequenceDiagram
participant POS as POS round
participant ChainContext as chain context
participant QuorumEntropy as quorum entropy
POS->>ChainContext: compare height and cached top-block hash
ChainContext-->>POS: return chain_top_changed
POS->>QuorumEntropy: prepare quorum entropy
QuorumEntropy-->>POS: return entropy state
POS->>POS: restart round when the chain tip changed
Suggested reviewers: Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 3
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/cryptonote_core/pos.cpp`:
- Around line 1054-1055: Update the conditional branches in the affected PoS
logic, including the check on context.wait_for_next_block.height and the
corresponding branch near the second location, to wrap each controlled return
statement in braces. Preserve the existing conditions and return behavior.
- Around line 1047-1057: The POS wait flow currently reads the chain tip
multiple times, allowing height and hash-dependent state to come from different
snapshots. Update wait_for_next_block() and chain_top_changed() to obtain one
locked chain-tip snapshot via the existing chain API, then reuse its height and
hash for the changed check, prev_hash, timing setup, and POS round context
updates.
In `@src/p2p/net_node.inl`:
- Around line 596-600: Verify that seed1.rpcnode.stream, seed3.rpcnode.stream,
and seed5.rpcnode.stream resolve to the intended MAINNET P2P services on port
19090, and confirm these endpoints use independent DNS/provider failure domains.
Apply any required corrections to the default seed entries and the fallback seed
list so bootstrap redundancy is preserved.
🪄 Autofix
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 8ca5b06d-8fd8-40a0-8eba-ee824435fb1f
📒 Files selected for processing (3)
CMakeLists.txtsrc/cryptonote_core/pos.cppsrc/p2p/net_node.inl
| // Returns true if the blockchain's top block differs from the tip cached in | ||
| // the round context — either because a new block arrived (height changed) or | ||
| // because a same-height reorg replaced the top block (hash changed at the | ||
| // same height). | ||
| static bool chain_top_changed(round_context const &context, cryptonote::Blockchain const &blockchain) | ||
| { | ||
| uint64_t const chain_height = blockchain.get_current_blockchain_height(true /*lock*/); | ||
| if (context.wait_for_next_block.height != chain_height) | ||
| return true; | ||
| return context.wait_for_next_block.top_hash != blockchain.get_block_id_by_height(chain_height - 1); | ||
| } |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
# Inspect the lock lifetime and all chain-tip reads.
rg -n -C 12 \
'get_current_blockchain_height|get_block_id_by_height|get_tail_id|chain_top_changed|wait_for_next_block' \
src/cryptonote_core --glob '*.cpp' --glob '*.h'Repository: Beldex-Coin/beldex
Length of output: 50375
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
# Inspect Blockchain tip read implementations and lock names.
fd -a 'blockchain\.(cpp|h)$' src | sed 's#^\./##' | while read -r f; do
echo "===== $f ====="
rg -n -C 8 'get_current_blockchain_height|get_tail_id|get_block_id_by_height' "$f" || true
done
# Focused slice around Blockchain implementation declarations/definitions.
rg -n -C 5 'Blockchain::(get_current_blockchain_height|get_tail_id(?:\b|\(.*\))|get_block_id_by_height)' src/cryptonote_core/blockchain.cpp src/cryptonote_core --glob '*.cpp' --glob '*.h'Repository: Beldex-Coin/beldex
Length of output: 33330
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
# Focused slice around Blockchain implementation declarations/definitions.
rg -n -C 5 'Blockchain::(get_current_blockchain_height|get_tail_id(?:\b|\(.*\))|get_block_id_by_height)' src/cryptonote_core/blockchain.cpp src/cryptonote_core/blockchain.hRepository: Beldex-Coin/beldex
Length of output: 3608
Use one locked chain-tip snapshot for POS state updates.
wait_for_next_block() reads the chain height, then chain_top_changed() reads it a second time. get_current_blockchain_height(true) releases the lock, and get_block_id_by_height() also acquires its own lock, so the comparison and the state update can observe different chain tips. Use one atomic snapshot, such as get_tail_id(chain_height_out), and reuse that height/hash for prev_hash, timing setup, and the POS context.
🧰 Tools
🪛 Clang (14.0.6)
[warning] 1051-1051: use a trailing return type for this function
(modernize-use-trailing-return-type)
[warning] 1054-1054: statement should be inside braces
(readability-braces-around-statements)
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/cryptonote_core/pos.cpp` around lines 1047 - 1057, The POS wait flow
currently reads the chain tip multiple times, allowing height and hash-dependent
state to come from different snapshots. Update wait_for_next_block() and
chain_top_changed() to obtain one locked chain-tip snapshot via the existing
chain API, then reuse its height and hash for the changed check, prev_hash,
timing setup, and POS round context updates.
| if (context.wait_for_next_block.height != chain_height) | ||
| return true; |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win
Add braces around the new conditional branches.
Clang reports readability-braces-around-statements at both locations. Add braces before later edits change the controlled statement unintentionally.
Proposed fix
- if (context.wait_for_next_block.height != chain_height)
- return true;
+ if (context.wait_for_next_block.height != chain_height)
+ {
+ return true;
+ }
- if (chain_top_changed(context, blockchain))
- return goto_wait_for_next_block_and_clear_round_data(context);
+ if (chain_top_changed(context, blockchain))
+ {
+ return goto_wait_for_next_block_and_clear_round_data(context);
+ }Also applies to: 1138-1139
🧰 Tools
🪛 Clang (14.0.6)
[warning] 1054-1054: statement should be inside braces
(readability-braces-around-statements)
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/cryptonote_core/pos.cpp` around lines 1054 - 1055, Update the conditional
branches in the affected PoS logic, including the check on
context.wait_for_next_block.height and the corresponding branch near the second
location, to wrap each controlled return statement in braces. Preserve the
existing conditions and return behavior.
Source: Linters/SAST tools
| full_addrs.insert("seed1.rpcnode.stream:19090"); | ||
| full_addrs.insert("seed2.rpcnode.stream:19090"); | ||
| full_addrs.insert("seed3.beldex.io:19090"); | ||
| full_addrs.insert("seed3.rpcnode.stream:19090"); | ||
| full_addrs.insert("seed4.rpcnode.stream:19090"); | ||
| full_addrs.insert("seed5.beldex.io:19090"); | ||
| full_addrs.insert("seed5.rpcnode.stream:19090"); |
There was a problem hiding this comment.
🩺 Stability & Availability | 🔵 Trivial
Verify bootstrap redundancy before release.
All five MAINNET seed entries now use rpcnode.stream. A DNS or provider outage for this domain can remove every default bootstrap endpoint. The fallback path also reuses the same seed list at Line 1463-1466. Confirm that seed1.rpcnode.stream, seed3.rpcnode.stream, and seed5.rpcnode.stream resolve to the intended MAINNET P2P services on port 19090. Confirm that the seed infrastructure has independent failure domains.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/p2p/net_node.inl` around lines 596 - 600, Verify that
seed1.rpcnode.stream, seed3.rpcnode.stream, and seed5.rpcnode.stream resolve to
the intended MAINNET P2P services on port 19090, and confirm these endpoints use
independent DNS/provider failure domains. Apply any required corrections to the
default seed entries and the fallback seed list so bootstrap redundancy is
preserved.
2 participants
The POS state machine previously detected staleness by comparing
heights only. An equal-height fork switch (pop block N, adopt the competing
block N) leaves the chain height unchanged, so the state machine kept using
wait_for_next_block.top_hash, which by then referred to a block deletedfrom both the main and the alt DB (switch_to_alternative_blockchain() is
called with keep_disconnected_chain=false on the POS weight/checkpoint
reorg paths). The next entropy lookup then failed with:
"Failed to find block "
"Failed to get quorum entropy for POS, next block parent "
and the node sat out POS participation until the next height. Comparing the
top hash as well restarts the POS stages against the new tip immediately.