feat: add remaining mainnet opcodes and L2 diffs

[leovct]

Edits

edit 1: just a few more opcodes... 🫠 edit 2: all mainnet opcodes have been added! it's time to add L2 opcodes now 🔥 edit 3: all mainnet opcodes and diffs with arbitrum/optimism have been added 🎉

Description

Resolves #6 Resolves #16

This PR adds the remaining mainnet opcodes as well as L2 diffs for Optimism and Arbitrum.

It also makes a few changes:

• The Opcode type has a new field supportedHardforks: string[] and two new functions getHardforksFrom and getHardforksFromTo to make sure we document in which hard fork each opcode is supported (related to #16 ).
• Links pointing to ethereum/execution-specs are now referenced to a commit ID (related to this comment).
• Instead of hardcoding the url of websites such as http://evm.codes in every opcode configuration file, I created a constant and a function to get the URL for a specific opcode number. Same thing for https://evm.codes/playground and for the opcode reference to the ethereum/execution-specs source code. If any URL changes or the Github repository name changes, we just need to update it in one place.

Mainnet Opcode Tracker

Note: If no hard fork is specified next to an opcode, it means the opcode was added in the Frontier hard fork.

Arithmetic

• [x] add
• [x] addmod
• [x] div
• [x] exp
• [x] mod
• [x] mul
• [x] mulmod
• [x] sdiv
• [x] signextend
• [x] smod
• [x] sub

Bitwise

• [x] bitwise_and
• [x] bitwise_or
• [x] bitwise_xor
• [x] bitwise_not
• [x] get_byte
• [x] bitwise_shl (added in Constantinople hard fork)
• [x] bitwise_shr (added in Constantinople hard fork)
• [x] bitwise_sar (added in Constantinople hard fork)

Block

• [x] blockhash
• [x] chainid (added in Istanbul hard fork)
• [x] coinbase
• [x] difficulty (removed in Paris hard fork)
• [x] galimit
• [x] number
• [x] prevrandao (added in Paris hard fork)
• [x] timestamp

Comparison

• [x] less_than
• [x] signed_less_than
• [x] greater_than
• [x] signed_greater_than
• [x] equal
• [x] is_zero

Control Flow

• [x] stop
• [x] jump
• [x] jumpi
• [x] pc
• [x] gas_left
• [x] jumpdest

Environment

• [x] address
• [x] balance
• [x] origin
• [x] caller
• [x] callvalue
• [x] calldataload
• [x] calldatasize
• [x] calldatacopy
• [x] codesize
• [x] codecopy
• [x] gasprice
• [x] extcodesize
• [x] extcodecopy
• [x] returndatasize (added in Byzantium hard fork)
• [x] returndatacopy (added in Byzantium hard fork)
• [x] extcodehash (added in Constantinople hard fork)

Keccak

• [x] keccak

Log

• [x] log0
• [x] log1
• [x] log2
• [x] log3
• [x] log4

Memory

• [x] mstore
• [x] mstore8
• [x] mload
• [x] msize

Stack

• [x] pop
• [x] push[1..32]
• [x] push0 (added in Shanghai hard fork)
• [x] dup[1..16]
• [x] swap[1..16]

Storage

• [x] sload
• [x] sstore

System

• [x] create
• [x] create2 (added in Constantinople hard fork)
• [x] return_
• [x] generic_call
• [x] call
• [x] callcode
• [x] selfdestruct
• [x] delegatecall (added in Homestead hard fork)
• [x] staticcall (added in Byzantium hard fork)
• [x] revert (added in Byzantium hard fork)
• [x] invalid

L2 Opcode Tracker

• [x] Arbitrum opcode diffs
• [x] Optimism opcode diffs

[vercel[bot]]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
evm-diff	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Jul 19, 2023 0:11am

[leovct]

The PR isn't ready yet but just to confirm I'm going in the right direction with these additional changes, wdyt @mds1?

[leovct]

Should we reference the opcode INVALID (0xFE) mentioned here https://www.evm.codes/#fe?fork=shanghai? It's not mentioned in https://github.com/ethereum/execution-specs.

[mds1]

@leovct Yea let's include it. It looks that that opcode is here: https://github.com/ethereum/execution-specs/blob/03f0204e39ec98e456086407189f0ca3af277f68/src/ethereum/shanghai/vm/exceptions.py#L61-L66

But you're right I can't find the 0xfe reference for it, though I did find it in geth here: https://github.com/ethereum/go-ethereum/blob/c7b099b2ea58b3c2b594ae77b607142451ae9129/core/vm/opcodes.go#L221

0xfe is equivalent in functionality to any other undefined opcode, but the difference is 0xfe is guaranteed to remain undefined. I think this is mainly an off-chain consensus guarantee, which is probably why the execution-specs reason doesn't have it specified

[leovct]

@mds1 I've added the latest opcodes missing and the diffs between Arbitrum and Optimism so you can start reviewing if you want :)

I'm also doing a review on my side to make sure I didn't forget anything and that all the fields are properly formatted.

• [ ] Make sure objects are always expanded (prettier doesn't handle it well unfortunately)
• [ ] Make sure name fields are specified in lower case.
• [ ] Make sure all the other fields (e.g. description) begin with a capital letter. If there are several sentences, punctuation must be added.
• [ ] Make sure that examples' input and output field use a string value when there's only one element in the array (instead of an array made of one element)
• [ ] Double-check that notes have not been missed 👀
• [ ] Check all evm.codes playground link to make sure no query parameters have been missed.