chronologic / chronos-protocol Goto Github PK

We define the transaction encoding to be

    byte    TemporalUnit           [1  byte ]
    address recipient              [32 bytes]
    uint256 value                  [32 bytes]
    uint256 callGas                [32 bytes]
    uint256 gasPrice               [32 bytes]
    uint256 executionWindowStart   [32 bytes]
    uint256 executionWindowLength  [32 bytes]
    uint256 bounty                 [32 bytes]
    uint256 fee                    [32 bytes]
    bytes   callData               [32+bytes]

where TemporalUnit is either 0x01 for block number based scheduling or 0x02 for unix timestamp based scheduling. Transaction parameters recipient, value, callGas, gasPrice, executionWindowStart, executionWindowLength, bounty, and fee are ABI encoded. The end of the string of bytes contains the callData.

Serialized this will look like this:

0x010000000000000000000000007ed1e469fcb3ee19c0366d829e291451be638e59000000000000000000000000000000000000000000000000000000000000000a00000000000000000000000000000000000000000000000
00000000000000014000000000000000000000000000000000000000000000000000000000000001e000000000000000000000000000000000000000000000000000000000000002800000000000000000000000000000000
00000000000000000000000000000032000000000000000000000000000000000000000000000000000000000000003c0000000000000000000000000000000000000000000000000000000000000046

As I'm working on building up this repo using the bytes encoding I will use this GH issue to document possible security concerns that occur to me.

1. Pre-image attack. Since we now use the hash of the encoded transaction bytes, it is possible for many ScheduledTransactions to be deployed pointing to the same data represented by IPFS hash. We may then get an attacker who attempts to break the system by scheduling the same transaction n times. We must build timenodes which are resilient to this kind of attack.

As a means of generalizing the conditional execution object to something nicely usable by other projects in the community such as state channels or Plasma contracts, or perhaps within future extensions of this project, it would be nice if callWithData returned bytes using returndatasize.