Istanbul holiman #627
Istanbul holiman #627
Conversation
|
@holiman's tests these were derived from: https://github.com/holiman/IstanbulTests/tree/3c34f2c1a24175f5f4424b64b5e7ed6d822dc6af/GeneralStateTests |
|
Good job, hats off!
|
shemnon
force-pushed
the
istanbul-holiman
branch
from
f374fc7
to
ae4c9c4
Compare
shemnon
force-pushed
the
istanbul-holiman
branch
from
502a10c
to
ae4c9c4
Compare
| "1000000000000000000000000000000000000000": { | ||
| "balance": "", | ||
| "//code": "record the gas, run the operation, then record the gas again. Drop output, subtract 2 for GAS, what is left is the cost of the operation, store in storage 0x01.", | ||
| "//codeLLL": "(asm GAS CHAINID GAS SWAP1 POP SWAP1 SUB 5 SWAP1 SUB 0x01 SSTORE)", |
There was a problem hiding this comment.
substract 2 or 5 ?
looks a good asm code template for gasCost checkup for each opcode
| @@ -0,0 +1,84 @@ | |||
| { | |||
| "sloadGasCost": { | |||
There was a problem hiding this comment.
this test has a very good post condition!
| "pre": { | ||
| "1000000000000000000000000000000000000000": { | ||
| "balance": "0x01f4", | ||
| "//code": "{ [[ 1 ]] (EQ (BALANCE (ADDRESS)) (SELFBALANCE)) }", |
There was a problem hiding this comment.
really need to add lllc support for new opcodes
| "pre": { | ||
| "1000000000000000000000000000000000000000": { | ||
| "balance": "0x01f4", | ||
| "//code": "{ [[ 1 ]] (SELFBALANCE) }", |
There was a problem hiding this comment.
have some more ideas:
Test selfbalance inside codecall delegatecall staticcall ordinary calls
Test selfbalance in create opcode in inicode
Test selfbalance in tr init code
Suicide then test selfbalance return value
Selfbalance after revert operations. And after oog operations in a subcall that doesn't break the whole transaction execution
Selfbalance on different calldepth levels
* this are also to be added to a template for any kind of opcode
|
This are very good tests, @shemnon. thanks! |
|
latest solidity works again. try out this PR it support CHAINID and SELFBALANCE |
update to new CHAINID opcode for LLL Also, the calculation for GAS and target value of CHAINID needed an update (offsetting off by 1 errors)
Update to use SELFBALANCE ASM from new LLLC Also, there was another offsetting off by 1 error pair in the data and the raw bytecode
|
Updated. Old raw bytecode removed from fillers. |
|
geth passes this tests |
Adaptations of @holiman's istanbul tests as state test fillers.
TODO:
refill with latest solidity