We cannot carry pointers into the gossmap across localmod addition
or removal.
We didn't notice because the map->chan_arr is not normally resized,
but if we change gossmap.c line 689 to only allocate 1 to start, we see this:
```
VALGRIND=1 valgrind -q --error-exitcode=7 --track-origins=yes --leak-check=full --show-reachable=yes --errors-for-leak-kinds=all plugins/renepay/test/run-mcf > /dev/null
==2349744== Invalid read of size 4
==2349744== at 0x1788C2: gossmap_chan_scid (gossmap.c:558)
==2349744== by 0x1872A2: get_chan_extra_half_by_chan (flow.c:346)
==2349744== by 0x187797: remove_completed_flow (flow.c:488)
==2349744== by 0x187927: remove_completed_flow_set (flow.c:518)
==2349744== by 0x18DF4D: main (run-mcf.c:393)
==2349744== Address 0x4b80f38 is 88 bytes inside a block of size 136 free'd
==2349744== at 0x4848C63: realloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==2349744== by 0x173D71: tal_resize_ (tal.c:744)
==2349744== by 0x177E36: next_free_chan (gossmap.c:336)
==2349744== by 0x177ED3: new_channel (gossmap.c:351)
==2349744== by 0x178441: add_channel (gossmap.c:458)
==2349744== by 0x1798D4: gossmap_apply_localmods (gossmap.c:904)
==2349744== by 0x18DEDB: main (run-mcf.c:388)
==2349744== Block was alloc'd at
==2349744== at 0x4848C63: realloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==2349744== by 0x173D71: tal_resize_ (tal.c:744)
==2349744== by 0x177E36: next_free_chan (gossmap.c:336)
==2349744== by 0x177ED3: new_channel (gossmap.c:351)
==2349744== by 0x178441: add_channel (gossmap.c:458)
==2349744== by 0x178B6D: map_catchup (gossmap.c:635)
==2349744== by 0x178F45: load_gossip_store (gossmap.c:697)
==2349744== by 0x179D71: gossmap_load (gossmap.c:978)
==2349744== by 0x18D22F: main (run-mcf.c:295)
```
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The main function here is payment_reconsider:
* Each payment has a list of pay_flow.
* This is populated in try_paying(), calling add_payflows & sendpay_new_flows.
* When we get a notification, we resolve a pay_flow using one of the pay_flow_failedxxx
or pay_flow_succeeded functions.
* They call payment_reconsider() which cleans up finished flows decides what to do:
often calling try_paying again.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
- remove internal gheap checks
- add check for arc_t.chanidx overflow
- remove outdated comments
- check the delta flow bounds before augmenting along a path
- get_flow_paths uses a dynamic tal array instead of a list.
- fix a unit test that depended on the order of returned flows
- fix bug: lightnind doesn't like if I reuse the partid of a failed
flow, therefore use a higher partid than any of the previous attempts.
- plugin_err instead of LOG_BROKEN if sendpay fails and we cannot get a
an error code.
- fix wrong comments.
- remove the background timer.
- This is a bugfix. Previous to this the MCF network was built using the
knowledge of the min and max liquidity but it didn't take into account
pending HTLCs.
- Also remove the min_prob_success option but hardcode a 90% value.
Removing some options that are not relevant to the user, they're kept
for developer mode only:
- base_fee_penalty
- min_prob_success
- prob_cost_factor
- remove heap.h, not used
Signed-off-by: Lagrang3 <eduardo.quintana@pm.me>
Rusty Russell
Lagrang3