[qa] Fix rare failure in p2p-segwit.py #12133

pull sdaftuar wants to merge 1 commits into bitcoin:master from sdaftuar:2018-01-fix-p2p-segwit changing 1 files +8 −6

sdaftuar commented at 6:39 PM on January 9, 2018: member

Avoid creating very small utxos that would violate an assumption in test_non_standard_witness.

Fixes #11953

in test/functional/p2p-segwit.py:1412 in 7fe5b20097 outdated

1411 |              # the signatures as we go.
1412 |              tx.vin.append(CTxIn(COutPoint(i.sha256, i.n), b""))
1413 | -            tx.vout.append(CTxOut(i.nValue, CScript([OP_TRUE])))
1414 |              tx.wit.vtxinwit.append(CTxInWitness())
1415 | -            sign_P2PK_witness_input(witness_program, tx, index, SIGHASH_SINGLE|SIGHASH_ANYONECANPAY, i.nValue, key)
1416 | +            sign_P2PK_witness_input(witness_program, tx, index, SIGHASH_ALL|SIGHASH_ANYONECANPAY, i.nValue, key)

instagibbs commented at 7:05 PM on January 9, 2018:

why this change?

sdaftuar commented at 7:06 PM on January 9, 2018:

It felt wrong to use SIGHASH_SINGLE with an out-of-bounds index.

instagibbs commented at 7:08 PM on January 9, 2018:

ah, ok then comment could use an update a few lines up.

sdaftuar commented at 7:10 PM on January 9, 2018:

Oops! Thanks, will fix.

sdaftuar force-pushed on Jan 9, 2018

Fix rare failure in p2p-segwit.py

Avoid creating very small utxos that would violate an assumption in
test_non_standard_witness.

35c2b1fe3b

sdaftuar force-pushed on Jan 9, 2018
jnewbery commented at 9:39 PM on January 9, 2018: member

Tested ACk 35c2b1fe3b1a046bbd1bfaf1c178e838d047277a
MarcoFalke commented at 5:02 PM on January 10, 2018: member

Thanks!

utACK 35c2b1fe3b1a046bbd1bfaf1c178e838d047277a. Created a similar fix last year, but somehow forgot to submit it.
ryanofsky commented at 9:28 PM on January 10, 2018: member

utACK 35c2b1fe3b1a046bbd1bfaf1c178e838d047277a. Change makes sense, but I couldn't figure out what problem the previous outputs caused in test_non_standard_witness.
MarcoFalke commented at 9:46 PM on January 10, 2018: member

In test_non_standard_witness, you subtract the fee at some point. Having really small outputs, this could lead to negative outputs, provoking the "bad-txns-vout-negative". The smallest observed value depends on the test run and the exception is hit very rarely non-deterministically.
instagibbs commented at 9:49 PM on January 10, 2018: member

Yes the PR comment makes it sound like it's a non-standard witness, rather than non-standard output.
fanquake added the label Tests on Jan 11, 2018
laanwj commented at 1:47 PM on January 11, 2018: member

utACK 35c2b1f
laanwj merged this on Jan 11, 2018
laanwj closed this on Jan 11, 2018
laanwj referenced this in commit 1d2eaba300 on Jan 11, 2018
DrahtBot locked this on Sep 8, 2021

Contributors

Labels

Linked (view graph)

#11953 travis failure in p2p-segwit.py not(b'bad-txns-vout-negative' == b'bad-witness-nonstandard')