Finish buf refactor #917
Finish buf refactor #917
Conversation
g-r-a-n-t
force-pushed
the
finish-buf-refactor
branch
3 times, most recently
from
0bf96fd
to
4e29dc5
Compare
g-r-a-n-t
force-pushed
the
finish-buf-refactor
branch
from
4e29dc5
to
9896787
Compare
| @@ -112,7 +112,7 @@ pub struct MemoryBuffer { | |||
| } | |||
|
|
|||
| /// The start of the buffer in EVM memory. | |||
| pub fn offset(self) -> u256 { | |||
| pub unsafe fn offset(self) -> u256 { | |||
There was a problem hiding this comment.
Why does this need to be unsafe ?
There was a problem hiding this comment.
Im not sure if it needs to be and we can always change it back. it's kind of like getting a raw pointer tho, so I figure it should be unsafe
crates/library/std/src/evm.fe
Outdated
| } | ||
|
|
||
| pub fn eq(_ x: u256, _ y: u256) -> u256 { | ||
| unsafe { return __eq(x, y) } | ||
| unsafe { return __eq(x, y) } |
| unsafe { return __staticcall(gas, u256(addr), buf.offset(), buf.input_len(), buf.offset(), buf.output_len()) == 1 } | ||
| } | ||
|
|
||
| pub unsafe fn call_2(gas: u256, addr: address, value: u256, input_offset: u256, input_len: u256, output_offset: u256, output_len: u256) -> bool { |
There was a problem hiding this comment.
I guess these functions are made for scenarios where manual memory allocation is required (for example in precompiles.fe)
There was a problem hiding this comment.
yeah, I spent a decent amount of time trying to figure out how to make a low-cost abstraction for static_call usage in precompiles with RawCallBuffer, but just couldnt make it work, so I ended up doing this
| @@ -238,40 +238,40 @@ pub unsafe fn call_data_size() -> u256 { | |||
| return __calldatasize() | |||
| } | |||
|
|
|||
| pub fn call_data_copy(buf: MemoryBuffer, from_offset f: u256) { | |||
| unsafe { __calldatacopy(buf.offset(), f, buf.len()) } | |||
| pub fn call_data_copy(mut buf: MemoryBuffer, from_offset f: u256) { | |||
There was a problem hiding this comment.
There are more functions that accept immutable bufs (e.g call_data_copy, log , return_mem), do we need to change those also ?
There was a problem hiding this comment.
call_data_copy?? typo?
the other ones are not writing to the buf, so they can stay as immutable
g-r-a-n-t
force-pushed
the
finish-buf-refactor
branch
from
9896787
to
bd4bad6
Compare
What was wrong?
There were still some
std::evmfunctions that used memory offset and length parameters instead ofMemoryBuffer.How was it fixed?
To-Do