ltc: budget Phase-1 think() verify + compute-thread watchdog (.157 wedge fix) - #246
Merged
Conversation
Phase-1 head-verification was unbudgeted: a backward reorg that makes every head tip fail attempt_verify walks the FULL rooted chain x PPLNS under the exclusive tracker lock -> multi-minute wedge (.157 19:53Z 2026-06-19). The pre-existing io_context watchdog is blind to this (IO thread only try_to_locks, so io_context stays live while the compute thread spins under the exclusive lock). - share_tracker.hpp: THINK_P1_VERIFY_BUDGET=100 bounds the per-cycle Phase-1 verify storm; remainder defers to the next run_think() continuation (lock released+reacquired between chunks). Same set of shares verified as p2pool -> GENTX parity preserved; only chunked. - node.cpp/.hpp: compute-thread deadline watchdog (atomic ns-deadline + generation, IO-thread steady_timer that never touches the tracker lock). On expiry: logs + backtrace, resets m_think_running so a fresh cycle can schedule. Mirrors btc/node. Bounds lock hold; does not change verification semantics.
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| { | ||
| // Phase-1 budget exhausted — defer the remaining heads to the | ||
| // next run_think() continuation (lock released between chunks). | ||
| if (p1_budget_remaining <= 0) { m_think_needs_continue = true; break; } |
ltc/node.cpp included <execinfo.h> and called ::backtrace unconditionally for the Phase-1-budget watchdog stack dump. execinfo.h/backtrace are glibc-only; MSVC has no such header -> Windows x86_64 build failed with C1083 (Cannot open include file: execinfo.h) compiling src/impl/ltc/node.cpp (ltc.vcxproj). Linux/clang/ASan were unaffected (glibc present); btc/node.cpp has the same include but is not built on Windows, so master stayed green. Guard both the include and the backtrace block with #ifndef _WIN32, matching the existing c2pool_refactored.cpp idiom. On MSVC the watchdog still fires, logs the timeout + pending_adds, and runs the identical recovery (reset m_think_running); only the native stack dump is skipped. No change to the Linux runtime path.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fix-bearing SHA 974c2e2 (GPG-signed)
Root cause (.157 wedge 2026-06-19 19:53Z)
Phase-1 head-verification in think() was unbudgeted. A backward reorg that makes every head tip fail attempt_verify walks the FULL rooted chain x PPLNS under the exclusive tracker lock -> multi-minute wedge. The pre-existing io_context-liveness watchdog is blind to this: the IO thread only ever try_to_locks, so io_context stays responsive while the compute thread spins under the exclusive lock (observed live: stratum pseudoshares still accepted, ZERO ASYNC-THINK ticks).
Fix (two pieces, same lock-discipline family as #71 drop_tails)
Verification
git grep THINK_P1_VERIFY_BUDGET 974c2e25.LTC-side only (share_tracker/node); additive, no overlap with btc-heap-opt drop_tails. Consensus/lock-path -> operator merge-tap when green.