ExecutionEngine: collect parameters affecting opcode price by Turalchik · Pull Request #571 · neo-project/neo-vm

Turalchik · 2026-04-17T10:55:22Z

This PR adds collection of parameters in the OpcodePriceArgs struct (length, reference count changes, etc.) that affect opcode costs for dynamic pricing. The pricing is calculated in the related PR neo-project/neo#4536.

A partial port of nspcc-dev/neo-go#4087. Ref. nspcc-dev/neo-go#4043.

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

codecov · 2026-04-17T10:56:45Z

Codecov Report

❌ Patch coverage is 88.44444% with 26 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.49%. Comparing base (e5fc3da) to head (68a9eaa).

Files with missing lines	Patch %	Lines
src/Neo.VM/JumpTable/JumpTable.Compound.cs	77.14%	23 Missing and 1 partial ⚠️
src/Neo.VM/JumpTable/JumpTable.Control.cs	88.23%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #571      +/-   ##
==========================================
- Coverage   85.55%   85.49%   -0.06%     
==========================================
  Files          41       42       +1     
  Lines        2720     2896     +176     
  Branches      345      349       +4     
==========================================
+ Hits         2327     2476     +149     
- Misses        303      330      +27     
  Partials       90       90

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

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

AnnaShaleva

Will submit detailed handlers review later.

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

AnnaShaleva

Will check it one more time together with neo-project/neo#4536.

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

AnnaShaleva

We still need to update parameters construction for Map-related opcodes (ref. nspcc-dev/neo-go#4222); other than that the PR looks legit, let's wait for the update.

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

shargon · 2026-04-29T09:10:46Z

+    /// Typ specifies type of <see cref="StackItemType"/> which in most of the cases serves
+	/// as an operand of the given opcode.
+    /// </summary>
+    public StackItemType Type { get; internal set; }


Why this is not initialized every time? Is this required?

It's required for some of opcodes. For other opcodes price doesn't depend on the operand type.

Could you give me an example please?

Sure. CONVERT requires Type since it works differently for Array/Struct and ByteStrings:
https://github.com/Turalchik/neo/blob/f0699631792accf4cdd717073091199f0bd650a3/src/Neo/SmartContract/Fee.cs#L357

PICKITEM requires Type argument since its price depends on whether the operand is a compound type or not:
https://github.com/Turalchik/neo/blob/f0699631792accf4cdd717073091199f0bd650a3/src/Neo/SmartContract/Fee.cs#L368

And APPEND doesn't actually depend on the operand type. RefsDelta and NClonedItems arguments cover possible price fluctuations for different operand types:
https://github.com/Turalchik/neo/blob/f0699631792accf4cdd717073091199f0bd650a3/src/Neo/SmartContract/Fee.cs#L353

In general, different combinations of Type, RefsDelta, Length and NClonedItems are used for every opcode. The set of required parameters is defined separately for every opcode via opcode handler investigation and via benchmarks of edge cases in nspcc-dev/neo-go#4218.

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

shargon · 2026-07-23T06:51:00Z

                    {
                        engine.ReferenceCounter.RemoveStackReference(value);
-                        throw new CatchableException($"The index of {nameof(VMArray)} is out of range, {index}/[0, {array.Count}).");
+                        var r4 = engine.ReferenceCounter.Count;


Should be good if we add some uts for this catcheable exceptions (json)
This can be done in a different PR

shargon · 2026-07-23T06:53:34Z

+        var r2 = engine.ReferenceCounter.Count;
+        engine.Push(x.ConvertTo(toType));
+        var (type, length) = (StackItemType.Any, 0);
+        if (fromType == StackItemType.Array && toType == StackItemType.Struct || fromType == StackItemType.Struct && toType == StackItemType.Array)


From array to array or from struct to struct?

When converting Array to Struct

neo-vm/src/Neo.VM/Types/Array.cs

Lines 81 to 86 in 99095c2

public override StackItem ConvertTo(StackItemType type)

{

if (Type == StackItemType.Array && type == StackItemType.Struct)

return new Struct(new List<StackItem>(InnerList));

return base.ConvertTo(type);

}

or Struct to Array

neo-vm/src/Neo.VM/Types/Struct.cs

Lines 69 to 74 in 99095c2

public override StackItem ConvertTo(StackItemType type)

{

if (type == StackItemType.Array)

return new Array(new List<StackItem>(InnerList));

return base.ConvertTo(type);

}

copying occurs, while converting Array to Array (or Struct to Struct) is a no-op.

but type will be Any in RunStats?

No, Type will be equal to StackItemType.Array

neo-vm/src/Neo.VM/JumpTable/JumpTable.Types.cs

Line 77 in a17f49c

type = StackItemType.Array;

neo-vm/src/Neo.VM/JumpTable/JumpTable.Types.cs

Line 86 in a17f49c

runStats.CollectType(type);

Please check my UT (693ed61)

Yes, Type is equal to StackItemType.Any in cases where the length of the array, structure, byte array, or other parameters do not affect the opcode operation time. When an Array is converted to an Array (or something like that), the memory is not copied, respectively, the instruction executes in constant time relative to the size of the array. The price of CONVERT depends on the size of the data structure only when the Type is equal to StackItemType.Array or StackItemType.ByteString https://github.com/neo-project/neo/blob/7c1e64512984ebdd39c0d26df25abb5670af7760/src/Neo/SmartContract/ApplicationEngine.OpCodePricesV1.cs#L151-L157

I think they should pay as if was converted, these optimizations should be done in compiler side. We can be faster, but they call convert

Price should reflect real resources spent and any imbalance is a problem. We're not spending any real resources on same-type conversions, these operations are fast, price should just reflect that instead of trying to punish anyone. Why anyone is doing that is a separate question, it's the same as NOP price --- we don't care why people use NOP (theoretically they shouldn't), we just charge some tiny amount of GAS for it and move the instruction pointer to the next (more interesting or not) instruction.

cschuchardt88 · 2026-07-23T07:23:13Z

Review summary

This PR is a solid foundation for dynamic opcode pricing: it introduces RunStats, threads it through the jump table, and reworks catchable faults to go through ExecuteThrow with ref-delta accounting (neo-go-aligned). Pricing itself correctly lives in neo-project/neo#4536.

I think this is mergeable after a few verification items, not a redesign.

Must verify before merge

CatchableException migration is complete
ExecuteNext no longer catches CatchableException. All former throw new CatchableException(...) paths in handlers must use ExecuteThrow (the compound/map/buffer cases in the diff look correct). Please confirm with a full-branch grep that nothing still throws CatchableException expecting the old outer catch. A regression test for a former catchable path would help.
Breaking API surface
This breaks:
- JumpTable.DelAction / all handlers
- PostExecuteInstruction
- ExecuteThrow
- Map.Remove
- Struct.Clone
  Please call this out explicitly for Neo.VM consumers and land in lockstep with neo#4536.

Suggestions (non-blocking if correct)

More RunStats unit tests
ASSERTMSG UTF-8 length tests are excellent. Please add similar captures for at least: map remove (index + refs), struct clone (NClonedItems), append/newarray length, and one fault path that uses ExecuteThrow + RefsDelta.
Struct.Clone clone count
Prefer an explicit counter in the clone loop over nClonedItems = MaxStackSize - count for readability and future-proofing.
Map.Remove ordered index
Add a small test: insert A,B,C → remove B → out index == 1 and remaining order A,C (so pricing uses the ordered index neo-go expects).

Nits

Document that each RunStats field is set once per instruction (later Collect* overwrites).
Document that PostExecuteInstruction may receive a null instruction on the synthetic RET path.

Existing discussion

Agree with @AnnaShaleva: no need to special-case RunStats population for uncatchable FAULT.

shargon

As talked with @roman-khimov , my ut must be updated, but the code looks good to me

vncoelho

lgtm
There is no much test for now

vncoelho · 2026-07-23T14:23:31Z

+        RefsDelta = refsDelta;
+    }
+
+    /// <summary>


Maybe some virtual method here for additional info to this stats, that we can in the future trigger for specific case and serve as a flag for price calculation.
Not sure yet.

vncoelho · 2026-07-23T14:23:33Z

+/// <summary>
+/// Contains opcode-specific parameters used to calculate dynamic price.
+/// </summary>
+public struct RunStats


I suggest another name such as OpCodeRunStats, OpcodeStackStats, OpcodePriceStats

Wi1l-B0t · 2026-07-23T15:39:33Z

                ExecutionContext context = CurrentContext!;
                Instruction? currentInstruction = context.CurrentInstruction;
                Instruction instruction = currentInstruction ?? Instruction.RET;
+                RunStats runStats = default;


RunStats can be a field of ExecutionEngine?

It was discussed here #571 (comment).

Wi1l-B0t

LGTM.

ajara87 · 2026-07-23T21:47:51Z

UT fails

AnnaShaleva

LGTM, will approve after unit-test fix.

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

collect parameters that affect price of opcode

47c3397

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

AnnaShaleva reviewed Apr 17, 2026

View reviewed changes

Comment thread src/Neo.VM/OpCodePriceArgs.cs Outdated

Comment thread src/Neo.VM/ExecutionEngine.cs Outdated

Comment thread src/Neo.VM/ExecutionEngine.cs Outdated

Comment thread src/Neo.VM/Types/Struct.cs Outdated

Comment thread src/Neo.VM/OpCodePriceArgs.cs Outdated

AnnaShaleva changed the title ~~collect parameters that affect price of opcode~~ ExecutionEngine: collect parameters affecting opcode price Apr 17, 2026

AnnaShaleva mentioned this pull request Apr 17, 2026

Add dynamic opcode pricing neo-project/neo#4536

Open

18 tasks

Tural Devrishev added 4 commits April 20, 2026 11:49

make handlers return OpcodePriceArgs?

ef35a92

restrict setter access modifiers

3d5240a

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

make out parameter in clone struct

6a36389

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

rename OpcodePriceArgs to OpcodePriceParams

35d29c1

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

Turalchik requested a review from AnnaShaleva April 20, 2026 09:59

Merge remote-tracking branch 'origin/master' into dynamic-opcode-pricing

b3796e7

AnnaShaleva reviewed Apr 20, 2026

View reviewed changes

AnnaShaleva added 4 commits April 20, 2026 20:01

Update src/Neo.VM/JumpTable/JumpTable.Compound.cs

6f26c80

Update src/Neo.VM/JumpTable/JumpTable.Compound.cs

34233c4

Update src/Neo.VM/JumpTable/JumpTable.Push.cs

7a6e6f5

Update src/Neo.VM/JumpTable/JumpTable.Types.cs

b35c0a0

shargon reviewed Apr 21, 2026

View reviewed changes

Comment thread src/Neo.VM/JumpTable/JumpTable.Splice.cs

Tural Devrishev added 3 commits April 21, 2026 11:41

fix setting refsdelta in popitem

d8c2021

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

add comments for OpcodePriceParams fields

2ba50ee

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

rename OpcodePriceArgs file to OpcodePriceParams

56fe960

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

AnnaShaleva reviewed Apr 22, 2026

View reviewed changes

Comment thread src/Neo.VM/OpCodePriceParams.cs Outdated

AnnaShaleva reviewed Apr 22, 2026

View reviewed changes

shargon reviewed Apr 22, 2026

View reviewed changes

Comment thread src/Neo.VM/Types/Struct.cs Outdated

Comment thread src/Neo.VM/JumpTable/JumpTable.Slot.cs Outdated

Tural Devrishev added 3 commits April 28, 2026 21:47

replace OpcodePriceParams to OpCodePriceParams

3587311

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

apply shargon review

38c00c8

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

fix collecting params

01c566a

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

shargon reviewed Apr 29, 2026

View reviewed changes

AnnaShaleva self-requested a review April 29, 2026 09:14

AnnaShaleva reviewed Apr 29, 2026

View reviewed changes

take into account number of elements being shifted

7e91bb9

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

shargon reviewed Jul 22, 2026

View reviewed changes

Comment thread src/Neo.VM/JumpTable/JumpTable.Compound.cs

Turalchik requested a review from shargon July 22, 2026 08:42

roman-khimov mentioned this pull request Jul 22, 2026

[2026-07-23] 🗓️ Core Dev Regular Meeting Discussion neo-project/neo#4582

Closed

shargon reviewed Jul 23, 2026

View reviewed changes

Comment thread src/Neo.VM/JumpTable/JumpTable.Control.cs

Turalchik requested a review from shargon July 23, 2026 10:28

UT to prove Convert it's wrong

693ed61

shargon previously approved these changes Jul 23, 2026

View reviewed changes

vncoelho reviewed Jul 23, 2026

View reviewed changes

Merge branch 'master' into dynamic-opcode-pricing

d1dc530

Wi1l-B0t reviewed Jul 23, 2026

View reviewed changes

Turalchik requested review from Jim8y, Wi1l-B0t and ajara87 July 23, 2026 16:14

Wi1l-B0t previously approved these changes Jul 23, 2026

View reviewed changes

AnnaShaleva reviewed Jul 28, 2026

View reviewed changes

fix CONVERT UT expectations for no-op conversions

68a9eaa

Signed-off-by: Tural Devrishev <tural@nspcc.ru>

Turalchik dismissed stale reviews from Wi1l-B0t and shargon via 68a9eaa July 28, 2026 10:42

Turalchik requested review from AnnaShaleva, Wi1l-B0t, shargon and vncoelho July 28, 2026 10:43

AnnaShaleva approved these changes Jul 28, 2026

View reviewed changes

ajara87 approved these changes Jul 29, 2026

View reviewed changes

Wi1l-B0t approved these changes Jul 29, 2026

View reviewed changes

shargon approved these changes Aug 2, 2026

View reviewed changes

shargon merged commit 3665b50 into neo-project:master Aug 2, 2026
5 checks passed

	public override StackItem ConvertTo(StackItemType type)
	{
	if (Type == StackItemType.Array && type == StackItemType.Struct)
	return new Struct(new List<StackItem>(InnerList));
	return base.ConvertTo(type);
	}

	public override StackItem ConvertTo(StackItemType type)
	{
	if (type == StackItemType.Array)
	return new Array(new List<StackItem>(InnerList));
	return base.ConvertTo(type);
	}

Uh oh!

Conversation

Turalchik commented Apr 17, 2026 • edited by AnnaShaleva Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov Bot commented Apr 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

AnnaShaleva left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

AnnaShaleva left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

AnnaShaleva left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

AnnaShaleva Apr 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

AnnaShaleva Apr 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

shargon Jul 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

cschuchardt88 commented Jul 23, 2026

Review summary

Must verify before merge

Suggestions (non-blocking if correct)

Nits

Existing discussion

Turalchik commented Apr 17, 2026 •

edited by AnnaShaleva

Loading

codecov Bot commented Apr 17, 2026 •

edited

Loading

AnnaShaleva Apr 29, 2026 •

edited

Loading

AnnaShaleva Apr 29, 2026 •

edited

Loading

shargon Jul 23, 2026 •

edited

Loading