bug(super circuit): evm circuit / rw table / mpt table / block table assigned twice

[lispc]

maybe this line should be removed https://github.com/privacy-scaling-explorations/zkevm-circuits/blob/c78c86dfb7bc9198c0e7b0fcc5e7ecfd21277893/zkevm-circuits/src/super_circuit.rs#L304

Assigning to same columns in two region, can result the assigned "matrix" repeat twice vertically. At least it costs twice rows without any furthur problems if any

[han0110]

Yes I think this is duplicated and should be removed, but I wonder how current SuperCircuit pass the integration test, I guess it's because q_usable is not enabled in the final row (the row right after q_last_step), so the transition from EndBlock to BeginTx is not causing failure.

[lispc]

oh i find rw_table / mpt_table also assigned twice in super circuit, once in state circuit, once standalone.. Will fix them next PR..

[SuccinctPaul]

Hi, all.
I've seen the some table(like exp_table) can be assigned in two ways. Is this a bug?

• The fisrt is the table is assigned when its associated circuit is assigned.
  https://github.com/privacy-scaling-explorations/zkevm-circuits/blob/0778f85fe9ad326d8284e8c57f1d5f9f46b5db33/zkevm-circuits/src/exp_circuit.rs#L338

• The second is the table is assigned by invoking its load method.
  https://github.com/privacy-scaling-explorations/zkevm-circuits/blob/0778f85fe9ad326d8284e8c57f1d5f9f46b5db33/zkevm-circuits/src/evm_circuit.rs#L365

#984 (comment)

[kunxian-xia]

It's not a bug, as the EVM test circuit does not include the synthesize_sub of exp circuit.

[SuccinctPaul]

It's not a bug, as the EVM test circuit does not include the synthesize_sub of exp circuit.

The above seems like will be assigned twice in non-test situation.

[kunxian-xia]

After a quick search, the only place that exp_table.load function is called is in the EVM test circuit.