AI-generated analysisPublished automatically and not human-verified. Validated context appears in community notes below.
← Watch feed
Moderate 67 Bitcoin

htlc_wire: fix crash when adding an HTLC

Public commit record

What the developer wrote

Authored by Lagrang3

78/100 · Adequate
htlc_wire: fix crash when adding an HTLC

In line channeld/channeld_wiregen.c:832 `*added+i` is not a tal object hence
the instruction in common/htlc_wire.c:200 `tal_arr(ctx, struct tlv_field, 0);` crashes CLN.
This is fixed by stating that added_htlc is a a varsize_type.

Logs:

2025-08-16T02:25:28.640Z **BROKEN** lightningd: FATAL SIGNAL 6 (version v25.05-200-g79b959b)V
...
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/tal/tal.c:95 (call_error) 0x54f6bc
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/tal/tal.c:169 (check_bounds) 0x54f75a
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/tal/tal.c:178 (to_tal_hdr) 0x54f782
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/tal/tal.c:193 (to_tal_hdr_or_null) 0x54f7c7
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/tal/tal.c:471 (tal_alloc_) 0x54ffe4
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/tal/tal.c:517 (tal_alloc_arr_) 0x5500c4
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: common/htlc_wire.c:200 (fromwire_len_and_tlvstream) 0x48d63d
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: common/htlc_wire.c:234 (fromwire_added_htlc) 0x48dd23
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: channeld/channeld_wiregen.c:832 (fromwire_channeld_got_commitsig) 0x4c61fa
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: lightningd/peer_htlcs.c:2377 (peer_got_commitsig) 0x4549cb
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: lightningd/channel_control.c:1552 (channel_msg) 0x4140fe
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: lightningd/subd.c:560 (sd_msg_read) 0x461513
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/io/io.c:60 (next_plan) 0x544885
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/io/io.c:422 (do_plan) 0x544cea
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/io/io.c:439 (io_ready) 0x544d9d
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: ccan/ccan/io/poll.c:455 (io_loop) 0x54665d
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: lightningd/io_loop_with_timers.c:22 (io_loop_with_timers) 0x42d220
2025-08-16T02:25:28.640Z **BROKEN** lightningd: backtrace: lightningd/lightningd.c:1487 (main) 0x43280f

gdb inspection:
830 *added = num_added ? tal_arr(ctx, struct added_htlc, num_added) : NULL;
831 for (size_t i = 0; i < num_added; i++)
832 fromwire_added_htlc(&cursor, &plen, *added + i);
(gdb) p i
$3 = 1

Changelog-None: crash introduced this release.
Signed-off-by: Lagrang3 <lagrang3@protonmail.com>
[ Added test, removed Changelog --RR ]
✓ Descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Mentions testing or verification
The short version

What changed, and why it matters

This commit fixes a crash bug in Core Lightning (CLN) that occurs when a peer adds more than one HTLC (a payment channel transaction) in a single commitment round. The root cause was a memory-management mistake: the generated wire-deserialization code treated an array of 'added_htlc' records as if each element were an independently allocated memory object, but only the array's head was. When the second element was processed, the code tried to allocate child objects against a non-TAL pointer, triggering an assertion/crash. The fix changes 'added_htlc' to a varsize pointer type so each record is allocated separately. The commit message and a test name indicate this could be hit in normal operation and caused a FATAL SIGNAL 6 (abort).

Recommended action

Upgrade to a CLN release containing this commit. Nodes running the affected release (v25.05-200-g79b959b and nearby commits) are at risk of remote-triggered crash when receiving a commitsig with multiple added HTLCs. Until patched, operators should monitor for unexpected daemon restarts and avoid untrusted high-volume channels if possible.

Security signals we found

01

Denial-of-service vector: remote peer can crash the local lightningd/channeld by adding multiple HTLCs in one commitment signature message

02

Memory allocator invariant violation: passing interior array pointer as tal context

03

FATAL SIGNAL 6 (SIGABRT) backtrace provided in commit message

04

Test test_htlc_tlv_crash was marked xfail and is now enabled, confirming reproducible crash

05

Fix touches generated wire code and declares added_htlc as varsize_type

Risk score

Why this scored 67/100

Our methodology →
Potential impact 20/30
Exploitability 15/25
Stealth signal 8/15
Affected reach 12/15
Confidence 8/10
Evidence quality 4/5
Human-validated context

Community notes

Notes can correct, qualify, or add evidence to the AI analysis. Every note shown here has been validated by a human moderator.

No validated notes yet.

The AI analysis stands alone for now. Submit a note if you can add evidence or important context.