Topic: Mistake when Creating and using multisig address (Read 1775 times)

Thank you for the link with Gavin's example, I will use that as a base case to test from.

The two outputs are going to the same set of keys, I wanted to try using both the p2sh and the "native multisig".  Labeling one of them as change was probably a poor choice.

I still need to work through the signing step, but hopefully working from the link you provided with Gavin's example I wont run into much trouble.

You should put a an address in your signature, I would love to tip you some coin for all the help you've given me!

i think you still have some mistakes.  I don't have the time to take a detailed look however it is important to understand there are two different multisigs in Bitcoin.  There is the original multsig where the script is encoded in the output itself.  I generally call this "native multisig" to be explicit.  Then there is P2SH (pay to script hash) where the payment is sent to a hash of the script (redeem script).

It appears in your tx above that out[0] is a payment using p2sh and out[1] is a payment using "native multisig".  Also it looks like you are using the same keys both outputs.  So you have 7.09 BTC going to the same set of keys, 0.01 BTC being sent using P2SH address (out[0]) and another 7.08 BTC being sent using native multisig (P2SH).

There is no requirement for change.  The protocol just sees inputs and outputs.  We only use a "change" address to balance the equation sum(input) = sum(outputs) + fee when the intended transaction amount is smaller than the input being "spent".

So I am going to assume the input (txid: 3349d7e3d8b25bbe7556d04c2ab37f7d071b7b2679c7ab364f4e12c15761538b index: 1) is worth 7.09 BTC (I don't see it on testnet blockchain but that may be my issue) and you are trying to make a tx with no fee?

If you just want to send the full amount to the P2SH address then you only need a single output worth 7.09 (or the value of the input - intended miner fee)

If you wanted to send 0.01 BTC to the P2SH address then the second output should just be another address in the senders wallet (if you are using a single wallet and sending to yourself I would just grab a "normal" address to use as the second output and the value of that should be value of input - 0.01 - intended fee.

If you haven't already I would do this demo from start to finish (just substitute your keys and your funding transaction) using bitcoind on testnet:
https://gist.github.com/gavinandresen/3966071

Gavin's demo is good because it walks through all of the following:
1) Create a P2SH address (2 of 3)
2) Send funds to the P2SH address
3) Create a tx "spending" funds from the P2SH address
4) Sign the P2SH address in multiple steps (simulating keys on different machines)
5) Broadcast the fully signed transaction.

Make sure you can do all this using nothing but bitcoind before moving on to custom java.  Otherwise it makes it difficult to see if the problem is in your java or your understanding of P2SH or your understanding on bitcoin tx in general.  Once you are able to go through it once, you can do it again, save the output of each step but don't broadcast.  You now have a step by step template and your java should produce the same outputs at each step.

Thank you DeathAndTaxes!

You are right, I was encoding the OP_MULTISIG incorrect.

I took your advice to generate the multisig address using bitcoin for comparison.  The address created and redemption script match the values you listed and  I noticed that I was not including the opcode prefix on my public keys when creating the redemption script.

My updated example, including a transaction to try and spend my source address is below.  Both the change and the destination are (I think) going to multisig addresses, just one is using the redemption script, and one is using a pay 2 script hash.

source address




destination address

change address

Transaction

Why don't you try creating the address in bitcoind and see where it differs.

Yes.


The bitcoin protocol never uses addresses.  Addresses are just for us humans (to avoid mistakes).  The client/app needs to convert the address to the PubKeyHash (Pay2PubKeyHash) or ScriptHash (Pay2ScriptHash) so it should be:
OP_HASH160 || ScriptHash || OP_EQUAL


It also looks like you are using decimal instead of hex value for OP_CHECKMULTISIG.  It should be 0xae not 174.  Everything will always be in binary.  Opcodes are always one byte (2 hex digits).

RedeemScript
5221037783b9aee0cd926306a30f37e2e819d69aa66707a1f96a15cc581368270c5bcf2102b024b 76570311b75fa353a0296c50d41629da101ecfee172449935c1a029af812103685e7c360d84e799 9055e8639776aee4a743c529b4d2c655c03905a754cbbdce53ae

Address:
37S7ghArXes3tU7k6aTUcjrD4aDVFbuQDi

The transactions are unknown? Can you paste the transaction ID?

You could look at the source code from https://coinb.in/multisig/

I’m having some trouble creating a valid testnet multisignature address in java, I’ve included the steps I’ve taken below.  Most of the variables I’m using are hex encoded and I’m using || to represent concatenation.  I hope that someone can help identify my mistake.

Here are the public keys I’m using (these are compressed keys, is that ok?)

pubKey1: 037783b9aee0cd926306a30f37e2e819d69aa66707a1f96a15cc581368270c5bcf
pubKey2: 02b024b76570311b75fa353a0296c50d41629da101ecfee172449935c1a029af81
pubKey3: 03685e7c360d84e7999055e8639776aee4a743c529b4d2c655c03905a754cbbdce


Algorithm for creating redemption script:
n || publicKey0 || publicKey1 || publicKey2 || m || OP_CHECKMULTISIG

My redemption script values:
52 || 037783b9aee0cd926306a30f37e2e819d69aa66707a1f96a15cc581368270c5bcf || 02b024b76570311b75fa353a0296c50d41629da101ecfee172449935c1a029af81 || 03685e7c360d84e7999055e8639776aee4a743c529b4d2c655c03905a754cbbdce || 53 || 174

Redemption script:
52037783b9aee0cd926306a30f37e2e819d69aa66707a1f96a15cc581368270c5bcf02b024b7657 0311b75fa353a0296c50d41629da101ecfee172449935c1a029af8103685e7c360d84e7999055e8 639776aee4a743c529b4d2c655c03905a754cbbdce53ae

Algorithm for creating address from redemption script:
version || hash160(sha256(redemption script)) || checksum

My address values:
c4 || c9d4d1c375a6964fe2cc08100c9178ce9c610999 || 34151294

Address:
c4c9d4d1c375a6964fe2cc08100c9178ce9c61099934151294

Finally my base58 address:
2NBeQknuJrQKY8AHCTV5AVpKzyHryxXXQ9R




I think can spend coins from the address using the redemption script:
52037783b9aee0cd926306a30f37e2e819d69aa66707a1f96a15cc581368270c5bcf02b024b7657 0311b75fa353a0296c50d41629da101ecfee172449935c1a029af8103685e7c360d84e7999055e8 639776aee4a743c529b4d2c655c03905a754cbbdce53ae

I think can send coins to the address using:
OP_HASH160 || 2NBeQknuJrQKY8AHCTV5AVpKzyHryxXXQ9R || OP_EQUAL



When I send coin to shows in blockexplore as unknown, and I can’t seem to spend it.  I’m betting there are some gross mistakes in the steps I’m taking, and I’m hoping the community can help me identify them.

A couple of questions I have are:
1) Can compressed public keys be used to create a multisig address
2) When sending coin using OP_HASH160 || Address || OP_EUQUAL what format should the address be in?
3) Am I using the correct algorithms to create the redemption script and the address?