transactions with more than MAX_SIGOPS sigops

[gavinandresen]

Transactions with more than MAX_SIGOPS sigops should not be added to the memory pool, because they can (obviously) never make it into a block.

Probably a good idea to DoS-disconnect clients who send you such transactions, too....

Thanks to makomk for finding this on testnet. This would only be an issue on main-net if you were relaying/mining non-standard transactions or running an updated git HEAD or OP_EVAL backport which defines CHECKMULTISIG as a standard transaction type.

[ashleyholman]

I thought I would take a look at some old bugs - this one being 2 years old now.

To test this, I need to make a valid transaction with >20,000 sigops (MAX_BLOCK_SIGOPS). Anyone got any pointers on how I could do that?

[sipa]

OP_CHECKMULTISIG can count as up to 20 sigops.

[sipa]

See #4150 by the way.

[ashleyholman]

The code in that thread does appear to fix this exact issue. The only difference is that it introduces a tighter limit of MAX_BLOCK_SIGOPS / 5 per tx. If that one gets merged, I guess this should be closed off.

[petertodd]

@ashleyholman You can test it with this: https://github.com/petertodd/tx-flood-attack The default settings make >20,000 sigops transactions.

Be warned that my test program will likely take all your Bitcoins, cause your wife to leave you, and shoot your dog. (though your resulting country music career may be successful)

[laanwj]

#4150 was merged, closing