This provides a quick, high-level overview of the Ethers ASM Dialect for EVM, which is defined by the Ethers ASM Dialect Grammar
Once a program is compiled by a higher level language into ASM (assembly), or hand-coded directly in ASM, it needs to be assembled into bytecode.
The assembly process performs a very small set of operations and is intentionally simple and closely related to the underlying EVM bytecode.
Operations include embedding programs within programs (for example the deployment bootstrap has the runtime embedded in it) and computing the necessary offsets for jump operations.
The Command-Line Assembler can be used to assemble an Ethers ASM Dialect file or to disassemble bytecode into its human-readable (ish) opcodes and literals.
An Opcode may be provided in either a functional or instructional syntax. For Opcodes that require parameters, the functional syntax is recommended and the instructional syntax will raise a warning.
A Label is a position in the program which can be jumped to. A
JUMPDEST is automatically added to this point in the assembled output.
A Literal puts data on the stack when executed using a
A Literal can be provided using a DataHexString or a decimal byte value.
To enter a comment in the Ethers ASM Dialect, any text following a semi-colon (i.e.
;) is ignored by the assembler.
A common case in Ethereum is to have one program embedded in another.
The most common use of this is embedding a Contract runtime bytecode within a deployment bytecode, which can be used as init code.
When deploying a program to Ethereum, an init transaction is used. An init transaction has a null
to address and contains bytecode in the
data bytecode is a program, that when executed returns some other bytecode as a result, this result is the bytecode to be installed.
Therefore it is important that embedded code uses jumps relative to itself, not the entire program it is embedded in, which also means that a jump can only target its own scope, no parent or child scopes. This is enforced by the assembler.
Every program in the Ethers ASM Dialect has a top-leve scope named
A Data Segment allows arbitrary data to be embedded into a program, which can be useful for lookup tables or deploy-time constants.
An empty Data Segment can also be used when a labelled location is required, but without the
JUMPDEST which a Labels adds.
To access the byte offset of a labelled item, use
For a Data Segment or a Labels, there is an additional type of Link, which provides the length of the data or bytecode respectively. A Length Link is accessed by
#foobar and is pushed on the stack as a literal.