Signature with negative integer?
2 messages, 19 Jul 2014. Started by Richard Moore.
Richard Moore, Greg Maxwell
Hey all,
I'm wondering if anyone can help explain to me tx 70f7c15c6f62139cc41afa858894650344eda9975b46656d893ee59df8914a3d...
(https://blockchain.info/tx/70f7c15c6f62139cc41afa858894650344eda9975b46656d893ee59df8914a3d)
The input signature script is:
304402206b5c3b1c86748dcf328b9f3a65e10085afcf5d1af5b40970d8ce3a9355e06b5b0220cdbdc23e6d3618e47056fccc60c5f73d1a542186705197e5791e97f0e6582a3201
Which decodes to:
r= 48560432700441876832361368709121298776045893858160378595187765610521057848155 s= -22732680560694206332190468058638664750027418114195068375538144640549433890254
The ECC library I'm using is failing to verify this, which I think makes sense, since I the point needs to be positive, no? But it is obviously valid, as it has been verified and spent. I have tried simply modulo curve.order to positive-ify it, but that didn't seem to work either. Given a point P (with Py < 0) is there some fancy way to bring it into the elliptic curve space, such that Px >= 0 and Py >= 0?
Thanks!
RicMoo
.·´¯`·.¸¸.·´¯`·.¸¸.·´¯`·.¸¸.·´¯`·.¸¸.·´¯`·.¸><(((º>
Richard Moore ~ Founder Genetic Mistakes Software inc. phone: (778) 882-6125 email: ricmoo@geneticmistakes•com www: http://GeneticMistakes.com
On Fri, Jul 18, 2014 at 9:33 PM, Richard Moore <me@ricmoo•com> wrote:
Quoted text (2 lines)
> Hey all, > I'm wondering if anyone can help explain to me tx 70f7c15c6f62139cc41afa858894650344eda9975b46656d893ee59df8914a3d...
A rather timely post. See the other thread on BIP0062. What you're looking at is an example of a well-known-to-implementers-here where invisible and undocumented "over permissiveness" in interpreting invalid encoding in a cryptographic library (OpenSSL in our case) which would have been probably-not-unwelcome in many other protocol uses results in an unexpected consensus critical normative rule in Bitcoin.
Modern releases of Bitcoin core will no longer relay or mine them but they're still valid in blocks should they show up.
BIP62 proposes, among other things, soft-forking (backwards compatible) changes that will strictly limit the DER encoding to avoid ambiguity. If adopted by the network implementations would still need to grandfather in existing weird transactions but could do so on a txid by txid basis since there would be no more broken encoding permitted in blocks, and use different DER decoding code without risk of consensus inconsistency (so long as it uses der decoding which is functionally identical to what BIP62 requires— of course).