feat(vm): Improve tracer trait

[Deniallugo]

What ❔

Changing the way, how tracers are working in VM. Try to make less calls and add more static dispatching

Tracer Api changes:

1. Method before decoding now always empty (zk_evm don't call it)
2. Method before cycle doesn't exist anymore
3. Method after_cycle renamed to finish_cycle and unified with should stop execution
4. Method save_results was removed
5. All methods from ExecutionProcessing trait moved to VmTracer

Vm Changes:
Refund tracer now dispatching statically and not dynamically

Why ❔

Checklist

• PR title corresponds to the body of PR (we generate changelog entries from PRs).
• Tests for the changes have been added / updated.
• Documentation comments have been added / updated.
• Code has been formatted via zk fmt and zk lint.

[codecov]

Codecov Report

Attention: 36 lines in your changes are missing coverage. Please review.

Comparison is base (7e23188) 32.73% compared to head (b266161) 32.75%.
Report is 10 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #121      +/-   ##
==========================================
+ Coverage   32.73%   32.75%   +0.02%     
==========================================
  Files         536      536              
  Lines       28122    28126       +4     
==========================================
+ Hits         9207     9214       +7     
+ Misses      18915    18912       -3     

Files	Coverage Δ
core/bin/system-constants-generator/src/utils.rs	0.00% <ø> (ø)
core/lib/vm/src/tracers/call.rs	50.00% <100.00%> (ø)
core/lib/vm/src/tracers/result_tracer.rs	50.00% <ø> (ø)
core/lib/vm/src/tracers/traits.rs	73.33% <85.71%> (-8.49%)	⬇️
core/lib/vm/src/tracers/validation/mod.rs	0.00% <0.00%> (ø)
core/lib/vm/src/tracers/refunds.rs	55.93% <50.00%> (+2.15%)	⬆️
core/lib/vm/src/tracers/storage_invocations.rs	0.00% <0.00%> (ø)
core/lib/vm/src/implementation/execution.rs	48.27% <46.15%> (-3.58%)	⬇️
core/lib/vm/src/tracers/default_tracers.rs	50.96% <45.45%> (+4.42%)	⬆️

... and 8 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mm-zk]

@Deniallugo - can you add more info in the PR description?
which methods you are removing, what are you changing in the API

[RomanBrodetski]

this is a little hard to follow - we do match, don't return only in the Batch variabt, and then do an if. I think it would be cleaner with

match self.execution_mode {
            VmExecutionMode::OneTx if self.tx_has_been_processed() => 
                      TracerExecutionStatus::Stop(TracerExecutionStopReason::Finish)
            }
            VmExecutionMode::Bootloader if self.ret_from_the_bootloader == Some(RetOpcode::Ok) => 
                    TracerExecutionStatus::Stop(TracerExecutionStopReason::Finish);
            }
            VmExecutionMode::Batch if self.validation_run_out_of_gas() => 
TracerExecutionStatus::Stop(TracerExecutionStopReason::Abort(
                Halt::ValidationOutOfGas,
        _ => 
        TracerExecutionStatus::Continue
            
        };

[montekki]

The two variants can be squashed into one i think with |

[Deniallugo]

We have to always check self.validation_run_out_of_gas. Independent from execution_mode.
Thanks for the other suggestions :)

The two variants can be squashed into one i think with |

Only without if..

[RomanBrodetski]

Why? We dobn't check it now since we return

[RomanBrodetski]

@Deniallugo can you please explain why the RefundTracer needs to be treated specially?

[Deniallugo]

@Deniallugo can you please explain why the RefundTracer needs to be treated specially?

It's an internal tracer for VM and it used almost with all calls and the results of the execution are presented in VmExecutionResultAndLogs

From performance perspective it's better to make it statically dispatched

[montekki]

The two variants can be squashed into one i think with |

[perekopskiy]

Looks really nice! Left a couple of nits

[github-actions]

Detected VM performance changes

Benchmark name	Difference in runtime
decode_shl_sub	-3.3%
event_spam	-2.5%
slot_hash_collision	-3.2%
access_memory	-3.0%
write_and_decode	-3.1%