Get rid of with_ext_cost_counter

[matklad]

We have this bit of logic:

nearcore/runtime/near-vm-logic/src/gas_counter.rs

Line 12 in f852488

pub fn with_ext_cost_counter(f: impl FnOnce(&mut HashMap<ExtCosts, u64>)) {

What it does is that it records the cost we charge for in a thread local. We use in the param estimator to make some asserts – basically, we run something, and we sanity check that the expected number of host functions was called. Eg, here:

nearcore/runtime/runtime-params-estimator/src/lib.rs

Line 645 in f852488

fn_cost(ctx, "read_memory_1Mib_10k", ExtCosts::read_memory_byte, 1024 * 1024 * 10_000)

1024 * 1024 * 10_000 is used to assert that we indeed read 1 meg ten thousand times.

The problem here is that we have a second, better infra for that, ProfileData:

nearcore/core/primitives-core/src/profile.rs

Lines 46 to 52 in f852488

	/// Profile of gas consumption.
	/// When add new cost, the new cost should also be append to Cost::ALL
	#[derive(Clone, PartialEq, Eq, BorshSerialize, BorshDeserialize)]
	pub struct ProfileData {
	data: DataArray,
	}

It computes the breakdown of costs for contracts, and is a stable thing which is available via RPC. So, we should port the estimator to profile data

[matklad]

cc @jakmeier

[stale]

This issue has been automatically marked as stale because it has not had recent activity in the last 2 months.
It will be closed in 7 days if no further activity occurs.
Thank you for your contributions.

[stale]

This issue has been automatically marked as stale because it has not had recent activity in the last 2 months.
It will be closed in 7 days if no further activity occurs.
Thank you for your contributions.

[matklad]

cc @akhi3030 you might want to look into this one

[akhi3030]

@matklad: I spent some time looking into this. An initial question is that with_ext_cost_counter appears to be used in production code as well.

https://github.com/near/nearcore/blob/master/runtime/near-vm-logic/src/logic.rs#L2362 storage_write() calls gas_counter.pay_base() which calls inc_ext_costs_counter() which calls with_ext_cost_counter().

Is the goal of this issue to get rid of uses of with_ext_cost_counter or just in the parameter estimator?

[matklad]

@akhi3030 production code uses this function to write this info, but the only reader of this information is the estimator. So, but removing it from the estimator, we'll make it dead, and could remove it from prod code as well.

[akhi3030]

Got it, thanks, makes sense.