dual-funding: final final final final?? #6391
Conversation
niftynei
force-pushed
the
nifty/df-final-final
branch
4 times, most recently
from
0d6d5d6 to
92e8df4
Compare
We're adding signed types to the spec! This adds the support mechanisms for them.
the witnesses are maddeningly weird now (you concat everything together) we also changed some things to be s64's (it's a teeny tiny change)
As per lightning/bolts@cd3c99e we should send the next_funding_txid if we've sent our commitment sigs, but we haven't received the peer's tx_signatures. Note that we send here, but don't verify that it's arrived.
niftynei
force-pushed
the
nifty/df-final-final
branch
from
92e8df4 to
c235f60
Compare
| */ | ||
| tlvs = tlv_channel_reestablish_tlvs_new(tmpctx); | ||
| if (!tx_state->remote_funding_sigs_rcvd) | ||
| tlvs->next_funding = &tx_state->funding.txid; |
There was a problem hiding this comment.
Would it make sense to tal_steal or tal a copy onto tlvs? Just so we don't accidentally free the pointee.
| const u8 *data = witness->witness_data; | ||
| size_t size, max = tal_count(data); | ||
| bool ok; | ||
|
|
||
| wally_tx_witness_stack_free(in->final_witness); | ||
|
|
There was a problem hiding this comment.
@niftynei @rustyrussell You need to set this to NULL after freeing it, otherwise if line 72 returns you have a dangling pointer and potential later use-after-free here. Alternately use wally_psbt_set_input_final_witness(NULL) which will free any existing witness and set the value to NULL.
There was a problem hiding this comment.
Thanks for the note! Just added a PR to fix this.
|
@niftynei wally now exposes witness se/serialization using the psbt-spec format as used here: #6882 specifically fedfe6f If I understand your change now uses that same serialization format? If so then your manual serialization can be removed once that PR is rebased and merged (has tal issues I need to look at in the last 2 commits). |
What if we finished off the dual-funding spec??
This PR adds/updates the following. Note that it's missing/still needs the check for the
next_funding_txid, which I'm hoping to get up soon ™️ .funding_contributionfrom an unsigned to a signed int. This is future-proofing for a world where splicing exists.next_funding_txidif we remember the peer on reconnect + haven't received their tx-sigs yet. Needs a little bit of work.