channeld: avoid overflow in reloading of channel from db.

We used to just throw htlcs into the channel with a flag to tell it to
ignore overflow.  Instead, we can insert them in order (which is the same as
id order) which always must be valid.

This helps when we turn the balance into a struct amount_msat which will get
upset with overflows.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This commit is contained in:
Rusty Russell 2019-02-21 14:15:55 +10:30
parent b8e484b508
commit 93dcd5fed7
1 changed files with 37 additions and 34 deletions

View File

@ -290,7 +290,7 @@ static enum channel_add_err add_htlc(struct channel *channel,
bool enforce_aggregate_limits)
{
struct htlc *htlc, *old;
s64 msat_in_htlcs, fee_msat, balance_msat;
s64 msat_in_htlcs, fee_msat;
enum side sender = htlc_state_owner(state), recipient = !sender;
const struct htlc **committed, **adding, **removing;
const struct channel_view *view;
@ -371,8 +371,7 @@ static enum channel_add_err add_htlc(struct channel *channel,
* its local commitment transaction...
* - SHOULD fail the channel.
*/
if (enforce_aggregate_limits
&& tal_count(committed) - tal_count(removing) + tal_count(adding)
if (tal_count(committed) - tal_count(removing) + tal_count(adding)
> channel->config[recipient].max_accepted_htlcs) {
return CHANNEL_ERR_TOO_MANY_HTLCS;
}
@ -388,6 +387,9 @@ static enum channel_add_err add_htlc(struct channel *channel,
* local commitment transaction:
* - SHOULD fail the channel.
*/
/* We don't enforce this for channel_force_htlcs: some might already
* be fulfilled/failed */
if (enforce_aggregate_limits
&& msat_in_htlcs > channel->config[recipient].max_htlc_value_in_flight.millisatoshis) {
return CHANNEL_ERR_MAX_HTLC_VALUE_EXCEEDED;
@ -420,32 +422,33 @@ static enum channel_add_err add_htlc(struct channel *channel,
assert(fee_msat >= 0);
/* Figure out what balance sender would have after applying all
* pending changes. */
balance_msat = view->owed_msat[sender];
if (enforce_aggregate_limits) {
/* Figure out what balance sender would have after applying all
* pending changes. */
s64 balance_msat = view->owed_msat[sender];
assert(balance_msat >= 0);
for (i = 0; i < tal_count(removing); i++)
balance_msat += balance_removing_htlc(removing[i], sender);
assert(balance_msat >= 0);
for (i = 0; i < tal_count(adding); i++)
balance_msat += balance_adding_htlc(adding[i], sender);
assert(balance_msat >= 0);
for (i = 0; i < tal_count(removing); i++)
balance_msat += balance_removing_htlc(removing[i], sender);
assert(balance_msat >= 0);
for (i = 0; i < tal_count(adding); i++)
balance_msat += balance_adding_htlc(adding[i], sender);
/* This is a little subtle:
*
* The change is being applied to the receiver but it will
* come back to the sender after revoke_and_ack. So the check
* here is that the balance to the sender doesn't go below the
* sender's reserve. */
if (enforce_aggregate_limits
&& balance_msat - fee_msat < (s64)channel->config[!sender].channel_reserve.satoshis * 1000) {
status_trace("balance = %"PRIu64
", fee is %"PRIu64
", reserve is %s",
balance_msat, fee_msat,
type_to_string(tmpctx, struct amount_sat,
&channel->config[!sender].channel_reserve));
return CHANNEL_ERR_CHANNEL_CAPACITY_EXCEEDED;
/* This is a little subtle:
*
* The change is being applied to the receiver but it will
* come back to the sender after revoke_and_ack. So the check
* here is that the balance to the sender doesn't go below the
* sender's reserve. */
if (balance_msat - fee_msat < (s64)channel->config[!sender].channel_reserve.satoshis * 1000) {
status_trace("balance = %"PRIu64
", fee is %"PRIu64
", reserve is %s",
balance_msat, fee_msat,
type_to_string(tmpctx, struct amount_sat,
&channel->config[!sender].channel_reserve));
return CHANNEL_ERR_CHANNEL_CAPACITY_EXCEEDED;
}
}
dump_htlc(htlc, "NEW:");
@ -479,7 +482,6 @@ enum channel_add_err channel_add_htlc(struct channel *channel,
else
state = RCVD_ADD_HTLC;
/* FIXME: check expiry etc. against config. */
return add_htlc(channel, state, id, msatoshi, cltv_expiry,
payment_hash, routing, htlcp, true);
}
@ -918,6 +920,8 @@ bool channel_force_htlcs(struct channel *channel,
const enum side *failed_sides)
{
size_t i;
struct htlc *htlc;
struct htlc_map_iter it;
if (tal_count(hstates) != tal_count(htlcs)) {
status_trace("#hstates %zu != #htlcs %zu",
@ -1052,12 +1056,11 @@ bool channel_force_htlcs(struct channel *channel,
htlc->failed_scid = NULL;
}
for (i = 0; i < tal_count(htlcs); i++) {
struct htlc *htlc;
htlc = channel_get_htlc(channel,
htlc_state_owner(hstates[i]),
htlcs[i].id);
/* Now adjust balances. The balance never goes negative, because
* we do them in id order. */
for (htlc = htlc_map_first(channel->htlcs, &it);
htlc;
htlc = htlc_map_next(channel->htlcs, &it)) {
if (!adjust_balance(channel, htlc))
return false;
}