tx.origin for factory

Question

I want to use a factory so that people can create a standardized contract. People should be able to transfer money directly by this creation which led me to the problem of the msg.sender during the exceution. I read that tx.origin has some security flaws but I wonder if this is also true just for the creation of contracts by a factory. Following example contract:

 pragma solidity ^0.4.0;

    contract Factory {

     address[] public contracts;

       function createContract () 
          payable 
          public 
       {
          Con newCon = (new Con).value(msg.value)();
          contracts.push(newCon);
       }

    }


    contract Con { 

     address owner;
     uint256 valueOwner
;

     function Con() 
        payable 
        public 
     { 
        owner = tx.origin;
        valueOwner = msg.value;
     } 

       function withdraw () 
          public
       {
          if(msg.sender == owner)
             msg.sender.transfer(valueOwner);
       }

    }

I do not see why tx.origin should do any harm here or am I missing some point?

Answer 1

Using tx.origin will create issues when the factory is called from within another contract.

Wallet (A) calls > multisig contract (B) > which calls your factory (C) > sub contract (D).

In this case the tx.origin is wallet A. In most cases users will expect the owner to actually be the multisig B contract, and not A.

To prevent this, just use msg.sender in the factory C, and pass it along to the new contract instance D, preferably in a second method call so people can easily use the etherscan code validator with no constructor parameters.