Sanity-check system deposit tx gas marginal wrt EIP-7623

[emhane]

we need to sanity-check something: with the increased calldata cost, what effect does this have on system deposit transactions gas usage? Specifically the L1BlockInfo? I think we have some margin in the gas-limit of these, but it may be good to sanity-check anyway.

Ref: ethereum-optimism/pm#25 (comment)

[elihaims]

discussed with @ajsutton in slack and don't think that this is an issue, but would catch it quickly in kurtosis/devnet if it was

[sebastianst]

From Slack Definitely a good callout. I see the max usage is 64k in @elihaims's statistics (can you post it here too?). The worst-case increased gas usage with EIP-7623 is by a factor of 10/4 (that's how non-zero calldata tokens are priced before/after, multiplied by 4). So even if those txs were calldata-only-txs, they would increase usage to 160k, still well below the limit. But in actuality, they aren't calldata-only txs, so their cost probably just stays the same. I think much more gas goes into storage writes etc.

But an end2end test/Kurtosis test will deliver the definitive proof.

[elihaims]

[refcell]

cc @meyer9 in case you are working on an E2E test for this

[meyer9]

Not currently - my e2e test for network upgrade txs might cover this. It checks to make sure they were successfully executed, although not for older upgrade txs.