state lexicographic ordering can overlap for tag

[DreamWuGit]

code line in https://github.com/appliedzkp/zkevm-circuits/blob/main/zkevm-circuits/src/state_circuit/lexicographic_ordering.rs#L322
be_bytes[0] = row.field_tag().unwrap_or_default() as u8 + ((row.tag() as u8) << 4);.

For RwTableTag::CallContext , its corresponding field tag CallContextFieldTag consists of more than 20 items. when current row is CallContext tag, next row is tag behind RwTableTag::CallContext , like TxLog, it is maybe not distinguishable to use be_bytes[0].
i.e.

• current row is CallContext with CallContextFieldTag::CodeSource ,
  be_bytes[0] = 20 + 10 <<4 = 180.
• next row is TxLog with TxLogFieldTag::Address ,
  be_bytes[0] = 1 + 11 <<4 = 177.

after RW sort , it breaks next row value should be bigger than current row.

[ed255]

I noticed there's already a comment here about this:

zkevm-circuits/zkevm-circuits/src/state_circuit/lexicographic_ordering.rs

Line 63 in db95abe

// + 4 bits for field_tag // TODO: this actually needs 5 bits. Either reduce id

So I would think that @z2trillion (he implemented the current state circuit) is already aware of this and maybe is planning an update?

As long as 5 bits (or more) are reserved for field_tag, this issue should be resolved right?

[z2trillion]

#523 will fixes the TODO. 4 bits for the field tag was previously enough because none of the tests used larger field tags, but #476 does, so we had to fix this issue now.

[DreamWuGit]

closing as #523 merged!