unexpected high gas used with bootstrap predeploy

[shunjizhan]

when calling bootstrap multiple times, gasUsed will be extremely high after the first time.

Reproduce

prepare

• checkout this branch
• put KEY=<yout private key> into .env, make sure you have enough jitosol (>1) and ACA (>20) on acala fork testnet for testing
• yarn build

route bootstrap without gasdrop

GAS_DROP=0 yarn hardhat run scripts/test.ts --network acalaFork

this will perform deployDropAndBootstrapStakeRouter for 2 times, and we can observe that in the second run, gasUsed is extremely high, which is abnormal.

-------------------- RUN 0 --------------------
{
  usedGas: '49,773',
  usedStorage: '-64'
}
{
  usedGas: '3,008,857',    // looks normal
  usedStorage: '22,974'    // expected since we need to deploy router for the first time
}

-------------------- RUN 1 --------------------
{
  usedGas: '49,773',
  usedStorage: '-64'
}
{
  usedGas: '24,592,742',  // ❓❓❓
  usedStorage: '0'        // expected since we do not need to deploy router now
}

route bootstrap with gasdrop 3 ACA

GAS_DROP=3 yarn hardhat run scripts/test.ts --network acalaFork

with gas drop, first run is fine, but second run will fail. My guess is that gasUsed will exceed block gas limit

-------------------- RUN 0 --------------------
{
  usedGas: '49,773',
  usedStorage: '-64'
}
{
  usedGas: '3,043,770',
  usedStorage: '22,974'    // expected since we need to deploy router for the first time
}

-------------------- RUN 1 --------------------
Error: cannot estimate gas; transaction may fail or may require manual gas limit
...
execution reverted: TRANSFER_FROM_FAILED

[shunjizhan]

oh the failure for dropping ACA is expected, since we didn't deploy router for the second run and so didn't 'top up ACA for the router

UPDATE: this is still unexpected

[shunjizhan]

for the high gas used, it could be due to calculating router address? will do some validation

[shunjizhan]

ok it's indeed an asset router issue AcalaNetwork/asset-router#63