Anchor spends crash node #6973
Comments
endothermicdev
added a commit
to endothermicdev/lightning
that referenced
this issue
Jan 3, 2024
Temporary patch to avoid issue ElementsProject#6973 Changelog-None
|
My hunch here is that this is an old(-ish) node, or a node that had a channel open without The solutions here are twofold: prevent using tweaked pubkey outputs to bump anchors, or teach the signer about those tweaked key. I was unable to decode the PSBT, so I cannot verify this. But @michael1011 could probably take a look at the inputs and tell us if this idea is correct. |
|
This is a v2 PSBT. using the following to convert it to v0: we can decode the result: analyzepsbt: |
|
The release |
|
The latest version of liberally has been pulled din and it should be fixed, as far as I know. |
Alright. I am gonna close this for now and will reopen in case we observe the same behaviour again |
Still happening on |
|
linking a patch that at least prevents the crash endothermicdev@5aebe15 |
|
This is the bitcoind decode (after using lightningd to change to PSBT v0) for this one:
While there's no proper backtrace on the second report, the first shows |
|
OK. Looking at the testnet one. We have two inputs: second is the anchor, which is finalized fine, and first is a to-remote output from a previous unilateral close, which for some reason we do not finalize. Interestingly, that output did get spent according to the testnet explorer (which is why I know its script!), so we obviously can spend it somehow, but we're failing to do so in this case. Here's the tx which spent it (as third output): https://blockstream.info/testnet/tx/d1d87af8946333427a76d76f9524a8590e15eb1ef3bf691f32907dc4627781a8 |
|
For the anchor input there is only the |
|
Good point, but we're OK with that (and indeed, you can see the final_scriptwitness for that one). I've added a test, but it works. I think we'll need @endothermicdev 's "don't crash" fix, and more logging to get to the bottom of this if it happens again... |
rustyrussell
pushed a commit
to rustyrussell/lightning
that referenced
this issue
May 6, 2024
Temporary patch to avoid issue ElementsProject#6973 [ Add much more logging! --RR ] Changelog-Fixed: lightningd: avoid crash on signing failure when trying to spend anchor outputs.
rustyrussell
pushed a commit
to rustyrussell/lightning
that referenced
this issue
May 7, 2024
Temporary patch to avoid issue ElementsProject#6973 [ Add much more logging! --RR ] Changelog-Fixed: lightningd: avoid crash on signing failure when trying to spend anchor outputs.
endothermicdev
added a commit
that referenced
this issue
May 7, 2024
Temporary patch to avoid issue #6973 [ Add much more logging! --RR ] Changelog-Fixed: lightningd: avoid crash on signing failure when trying to spend anchor outputs.
rustyrussell
pushed a commit
to rustyrussell/lightning
that referenced
this issue
May 14, 2024
Temporary patch to avoid issue ElementsProject#6973 [ Add much more logging! --RR ] Changelog-Fixed: lightningd: avoid crash on signing failure when trying to spend anchor outputs.
|
I'm closing this for now, will reopen if it happens again. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Issue and Steps to Reproduce
When trying to create what I think is an anchor spend, CLN crashes:
We have had this happen twice:
getinfooutputThe only experimental feature we have enabled are the anchors
The text was updated successfully, but these errors were encountered: