Pekka Enberg

The `perf top` shows we're apparently in `step()` function: ``` 16.77% limbo_sim [.] limbo_core::vdbe::Program::step 6.84% [kernel] [k] io_idle 3.35% [kernel] [k] __lock_acquire 2.91% limbo_sim [.] limbo_core::storage::pager::Pager::cacheflush 2.45% limbo_sim [.] ::sync...

`RUST_LOG=trace` claims we're busy looping in `Halt`: ``` [2024-12-25T10:00:19Z TRACE limbo_core::vdbe] 15 Halt 0 0 0 0 [2024-12-25T10:00:19Z TRACE limbo_core::vdbe] Halt auto_commit true [2024-12-25T10:00:19Z TRACE limbo_core::vdbe] 15 Halt 0 0...

Something funny going on: This passes: ``` $ RUST_LOG=trace cargo run -- --seed=13207426936588942312 --maximum-size=5 ``` This panics: ``` $ RUST_LOG=trace cargo run -- --seed=13207426936588942312 --maximum-size=6 ``` ``` [2024-12-25T10:04:36Z ERROR limbo_sim]...

@alpaylan I am not sure. @jussisaurio commented that https://github.com/tursodatabase/limbo/pull/533 fixes _some_ issues that sound similar to what we're seeing here

With 548f66e1cde48b009f79e70de07dce3a2d0e4872 and cf1a3fb3e1b7ef03bfa24f6035fdc6feaa901237, the infinite loop now _always_ reproduces on Linux.

This no longer reproduces so closing.

Mobibench runs fine with Turso as of f9967f809a35c9f92d122abd71a196dad3ddf1e9

We no longer see this as of d5b06b96efabea31c1ba2f77b6418296f715a3c9 -- likely because we're limiting stress to a single thread. Closing for now.

Fixed by @PThorpe92 in 55e79a72c10c93a7612a60bf088d88bd324bc00a

This does not reproduce as of d5b06b96efabea31c1ba2f77b6418296f715a3c9. Closing.