Simple tx ID malleability fix, opcode proposal: OP_TXHASHVERIFY
4 messages, 17 Sep 2016. Started by Rune K. Svendsen.
Rune K. Svendsen (2), Luke Dashjr, Nick ODell
I would really like to be able to create transactions that are immune to transaction ID malleability now, so I have been thinking of the simplest solution possible, in order to get a BIP through without too much trouble.
An opcode we could call OP_TXHASHVERIFY could be introduced. It would be defined to work only if added to a scriptSig as the very first operation, and would abort if the hash of the transaction **with all OP_TXHASHVERIFY operations (including stack push) removed** does not match what has been pushed on the stack.
So, in order to produce a transaction with one or more inputs protected against tx ID malleability, one would:
1. Calculate tx ID of the tx: TX_HASH 2. For each input you wish to protect, add "0x32 $TX_HASH OP_TXHASHVERIFY" to the beginning of the scriptSig
When evaluating OP_TXHASHVERIFY, we make a copy of the tx in question, and remove the "0x32 <32 bytes> OP_TXHASHVERIFY" sequence from the beginning of all scriptSigs (if present), and abort if the tx copy hash does not match the top stack item.
This is a very simple solution that only adds 34 bytes per input, and when something better becomes available (eg. Segwit), we will stop using this. But in the meantime it's very valuable to be able to not worry about tx ID malleability.
Please let me know what you think.
/Rune
On Saturday, September 17, 2016 8:45:17 PM Rune K. Svendsen via bitcoin-dev wrote:
Quoted text (28 lines)
> I would really like to be able to create transactions that are immune to > transaction ID malleability now, so I have been thinking of the simplest > solution possible, in order to get a BIP through without too much trouble. > > An opcode we could call OP_TXHASHVERIFY could be introduced. It would be > defined to work only if added to a scriptSig as the very first operation, > and would abort if the hash of the transaction **with all OP_TXHASHVERIFY > operations (including stack push) removed** does not match what has been > pushed on the stack. > > So, in order to produce a transaction with one or more inputs protected > against tx ID malleability, one would: > > 1. Calculate tx ID of the tx: TX_HASH > 2. For each input you wish to protect, add "0x32 $TX_HASH OP_TXHASHVERIFY" > to the beginning of the scriptSig > > When evaluating OP_TXHASHVERIFY, we make a copy of the tx in question, and > remove the "0x32 <32 bytes> OP_TXHASHVERIFY" sequence from the beginning of > all scriptSigs (if present), and abort if the tx copy hash does not match > the top stack item. > > This is a very simple solution that only adds 34 bytes per input, and when > something better becomes available (eg. Segwit), we will stop using this. > But in the meantime it's very valuable to be able to not worry about tx ID > malleability. > > Please let me know what you think.
First of all, this is likely to be more trouble than segwit to deploy (mainly just because SegWit is already implemented and tested).
Secondly, it wouldn't fix your problem: anyone malleating the transaction would simply update the hash before this opcode...
Luke
I hadn't thought of that... There is a solution, I think, but it makes the operation less simple.
If a transaction contains at least two OP_TXHASHVERIFY-protected inputs, signed without ANYONECANPAY, their signatures would cover the other input's OP_TXHASHVERIFY hash, right?
/Rune
On Sat, Sep 17, 2016 at 10:56 PM, Matt Corallo <lf-lists@mattcorallo•com> wrote:
Quoted text (50 lines)
> (removing the list) > > Because the tx hash in your construction is not signed, someone wishing to > maleate a transaction may do so by also updating the hash in the scriptSig. > > Matt > > On September 17, 2016 4:45:17 PM EDT, "Rune K. Svendsen via bitcoin-dev" < > bitcoin-dev@lists•linuxfoundation•org> wrote: > >> I would really like to be able to create transactions that are immune to >> transaction ID malleability now, so I have been thinking of the simplest >> solution possible, in order to get a BIP through without too much trouble. >> >> An opcode we could call OP_TXHASHVERIFY could be introduced. It would be >> defined to work only if added to a scriptSig as the very first operation, >> and would abort if the hash of the transaction **with all OP_TXHASHVERIFY >> operations (including stack push) removed** does not match what has been >> pushed on the stack. >> >> So, in order to produce a transaction with one or more inputs protected >> against tx ID malleability, one would: >> >> 1. Calculate tx ID of the tx: TX_HASH >> 2. For each input you wish to protect, add "0x32 $TX_HASH >> OP_TXHASHVERIFY" to the beginning of the scriptSig >> >> When evaluating OP_TXHASHVERIFY, we make a copy of the tx in question, >> and remove the "0x32 <32 bytes> OP_TXHASHVERIFY" sequence from the >> beginning of all scriptSigs (if present), and abort if the tx copy hash >> does not match the top stack item. >> >> This is a very simple solution that only adds 34 bytes per input, and >> when something better becomes available (eg. Segwit), we will stop using >> this. But in the meantime it's very valuable to be able to not worry about >> tx ID malleability. >> >> Please let me know what you think. >> >> >> >> /Rune >> >> ------------------------------ >> >> bitcoin-dev mailing list >> bitcoin-dev@lists•linuxfoundation•org >> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev >> >>
Then you have a new problem. Hash1 must contain Hash2 and the transaction, but Hash2 must contain Hash1 and the transaction. A circular dependency.
--Nick
On Sat, Sep 17, 2016 at 3:14 PM, Rune K. Svendsen via bitcoin-dev <bitcoin-dev@lists•linuxfoundation•org> wrote:
Quoted text (70 lines)
> I hadn't thought of that... There is a solution, I think, but it makes the > operation less simple. > > If a transaction contains at least two OP_TXHASHVERIFY-protected inputs, > signed without ANYONECANPAY, their signatures would cover the other input's > OP_TXHASHVERIFY hash, right? > > > /Rune > > > On Sat, Sep 17, 2016 at 10:56 PM, Matt Corallo <lf-lists@mattcorallo•com> > wrote: >> >> (removing the list) >> >> Because the tx hash in your construction is not signed, someone wishing to >> maleate a transaction may do so by also updating the hash in the scriptSig. >> >> Matt >> >> On September 17, 2016 4:45:17 PM EDT, "Rune K. Svendsen via bitcoin-dev" >> <bitcoin-dev@lists•linuxfoundation•org> wrote: >>> >>> I would really like to be able to create transactions that are immune to >>> transaction ID malleability now, so I have been thinking of the simplest >>> solution possible, in order to get a BIP through without too much trouble. >>> >>> An opcode we could call OP_TXHASHVERIFY could be introduced. It would be >>> defined to work only if added to a scriptSig as the very first operation, >>> and would abort if the hash of the transaction **with all OP_TXHASHVERIFY >>> operations (including stack push) removed** does not match what has been >>> pushed on the stack. >>> >>> So, in order to produce a transaction with one or more inputs protected >>> against tx ID malleability, one would: >>> >>> 1. Calculate tx ID of the tx: TX_HASH >>> 2. For each input you wish to protect, add "0x32 $TX_HASH >>> OP_TXHASHVERIFY" to the beginning of the scriptSig >>> >>> When evaluating OP_TXHASHVERIFY, we make a copy of the tx in question, >>> and remove the "0x32 <32 bytes> OP_TXHASHVERIFY" sequence from the beginning >>> of all scriptSigs (if present), and abort if the tx copy hash does not match >>> the top stack item. >>> >>> This is a very simple solution that only adds 34 bytes per input, and >>> when something better becomes available (eg. Segwit), we will stop using >>> this. But in the meantime it's very valuable to be able to not worry about >>> tx ID malleability. >>> >>> Please let me know what you think. >>> >>> >>> >>> /Rune >>> >>> ________________________________ >>> >>> bitcoin-dev mailing list >>> bitcoin-dev@lists•linuxfoundation•org >>> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev > > > > _______________________________________________ > bitcoin-dev mailing list > bitcoin-dev@lists•linuxfoundation•org > https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev >