VM: first integration #53

semuxgo · 2018-08-18T12:36:52Z

This is a standalone EVM implementation derived from EthereumJ v1.8.1. The main changes are:

Added a facade interface between VM and kernel
Made it independent and embedable into any Java-based blockchain project, the only dependencies are
- Commons Lang
- Bouncycastle
- SLF4J
- JUnit and Mockito (test)
Made DataWord immutable and refactored a few classes

@orogvany could you help review the code? also we need to implement all the interfaces in the org.ethereum.vm.client package, in the semux project.

witoldsz · 2018-08-18T20:45:58Z

src/main/java/org/ethereum/vm/DataWord.java

+ }
+
+ public byte getByte(int index) {
+ if (index < 0 || index >= SIZE) {


Please correct me if I am wrong, but isn't the bounds check embedded within Java arrays? I thought it's the primary reason why the arrays are slow – because one cannot disable the bounds check.

orogvany · 2018-08-20T03:06:17Z

It's unclear which are modifications and which are verbatim copies.

I think we do want to fork the project into a module, so we can continue to get bugfixes from mainline.

This combined with license restrictions makes for compelling reason to modularize.

I'm fine with LGPL and using this if you like it better than etheriumJ 1.5.0. I'd rather have bugfixes and modern version with ability to get fixes (even if we need to merge heavily as they diverge)

orogvany · 2018-08-20T04:43:00Z

I looked through the client interfaces, and it looks like we should be able to support them fairly easily. just a few new fields. looks good to me to start work on integration

semuxgo · 2018-08-20T23:58:50Z

This derived EVM is already highly modularized, and it's not using any classes from org.semux.

There isn't much development around EthereumJ, especially after the leave of their team leads, Roman and Nash. They are not even actively implementing the upcoming Constantinople fork, ethereum/pm#53. We have to put resources to maintain the VM and develop new features by ourselves, to stay up-to-date.

In case we want to merge bugfixes from upstream, we can simply follow https:/ethereum/ethereumj/commits/develop/ethereumj-core/src/main/java/org/ethereum/vm. There're only less than 50 commits in the past year.

orogvany · 2018-08-21T00:46:17Z

Ah, I did not know that. I'd still prefer to keep it separated for the licensing issue, even though it does make the build chain a little more difficult.

Alternatively, we could just split the project into 2 sub-modules (or more) if we wanted to keep them in the same github repo. Or we could ask nicely to get it dual licensed as MIT again, so we can continue development without worry of license taint. Long shot, as I'm not sure history of moving towards lgpl, but as its sunsetting development for now, they may be more open to it.

It's also possible we can keep them in the same module but use some mvn tools to get 2 jars output, but I'm not sure if that complies with the lgpl lib requirements. Having 2 libs enforces better separation IMO.

orogvany · 2018-08-21T00:49:37Z

src/main/java/org/ethereum/vm/util/BytecodeCompiler.java

+
+public class BytecodeCompiler {
+
+ public static void main(String[] args) {


should remove this.

It could be useful as an independent tool for de-compile opcode string.

Fair enough, javadoc here might be appropriate to avoid confusion

orogvany · 2018-08-21T01:41:18Z

src/main/java/org/ethereum/vm/client/BlockStore.java

+
+public interface BlockStore {
+
+ byte[] getBlockHashByNumber(int index);


This should be a long, right?
In the above Block class getNumber is a long, and we use long internally.

orogvany · 2018-08-21T01:46:34Z

src/main/java/org/ethereum/vm/client/Block.java

+
+public interface Block {
+
+ BigInteger getGasLimit();


Should gasLimit be set per-block or via configuration?
It seems like potentially each validator can define its own gas limit?
In eth, this is voted on by miners apparently, but I'm not sure how that works with dPOS

Correct. The block gasLimit is determined by miners through consensus rule. This VM module is designed for generic blockchain project, not tailed for Semux BFT.

Ok, we'll just handle in our client impl

orogvany · 2018-08-21T01:48:20Z

src/main/java/org/ethereum/vm/client/Repository.java

+ * the account address
+ * @return true if account exist, false otherwise
+ */
+ boolean isExist(byte[] address);


this reads weird. Can we rename to accountExists(address) or exists(address) ?

orogvany · 2018-08-21T01:52:58Z

src/main/java/org/ethereum/vm/client/Repository.java

+ * the account address
+ * @ImplNote trigger account creation
+ */
+ void createAccount(byte[] address);


Is this createAccount/deleteAccount compatible with our database?
It's not immediately clear if accounts in VM != addresses ?

orogvany · 2018-08-21T01:56:53Z

src/main/java/org/ethereum/vm/client/Repository.java

+ * the account address
+ * @return balance of the account as a <code>BigInteger</code> value
+ */
+ BigInteger getBalance(byte[] address);


There's probably some discrepancy here. We use long as our nano sem in Amount class. There seems to be little point in using BigInt, as we do not support over longs

This is because the base unit in Ethereum is 10^-18 ETH, while the base unit is 10^-9 SEM in Semux. Using Biginteger to handle any decimals.

Ok, that could get tricky. We'll probably want to add a toBigInt to amount class to make sure there's no errors.

Will need to figure out if this means we should move to 10^-18 in our db to match, or what.

semux added 30 commits August 11, 2018 09:24

VM: add the instruction set

b2c8772

Docs: update third party license

1ad8643

VM: add fee schedule

5f337db

VM: add dataword

5571ac6

VM: add message call and log info

25b3d8f

VM: add program stack, memory, and result

3e829cb

VM: add program storage

c2b473d

VM: add execution exceptions

e56dabb

VM: add vm configurations

593bb7d

VM: add precompiled contract

85f2dc5

VM: add program and vm

272bc27

Trivial: rename exceptions and fee schedule

5da0fa0

VM: remove dependency on semux utils

7898559

VM: refine the Transaction facet interface

4e5c0dc

Trivial: refactor code

9f7e344

VM: add transaction executor

8188091

VM: refactor the utility classes

223e5c9

VM: unify data type of nonce, value, gasLimit, and gasPrice

b1d57e9

VM: use byte array wrapper

82a19df

VM: add program invoke and implementations

5a3b1e1

VM: make dataword immutable

527b38c

Trivial: clean up code

f74ce10

VM: rename ProgramPrecompile to Preprocess

aab5532

VM: add unit tests

6fa8053

VM: add mock testing environment

b17d6c2

VM: fix unit tests

aee6541

VM: add bytecode decompiler

f698a21

Tests: add dataword test

c03c2f1

Tests: add memory test

040f0cb

Tests: add more complex and custom vm tests

e15b435

semuxgo force-pushed the vm branch from 450af72 to 3587a8c Compare August 18, 2018 20:44

witoldsz approved these changes Aug 18, 2018

View reviewed changes

semux added 5 commits August 18, 2018 20:51

VM: add range check for BigInteger to DataWord conversion

be2d437

Trivial: correct format

9bd730c

VM: clean up code

55fe11a

Tests: clean up VMComplexTest

00bcc72

VM: clean up transaction executor

b8c15ad

semuxgo force-pushed the vm branch from 3407d83 to b8c15ad Compare August 19, 2018 03:42

semux added 7 commits August 18, 2018 23:47

Tests: remove ingored test cases

fb9c802

VM: apply IntelliJ code inspection suggestion

5bd7ded

Trivial: rename methods

c1feee7

Trivial: correct typo

25ac77b

VM: add pre-compiled contracts

0ec7eac

VM: add precompiled contract tests

090c4a5

Tests: update ECRECOVER test

871bc24

orogvany reviewed Aug 21, 2018

View reviewed changes

semux added 4 commits August 20, 2018 23:38

Trivial: suppress IDE warnings

c704cb2

Trivial: remove unnecessary string constants

60daee0

VM: method signature refactoring

0ddc45b

VM: relocate to a git submodule

bdf3591

semuxgo merged commit c97d0af into semuxproject:develop Aug 21, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

VM: first integration #53

VM: first integration #53

semuxgo commented Aug 18, 2018 •

edited

Loading

witoldsz Aug 18, 2018 •

edited

Loading

orogvany commented Aug 20, 2018 •

edited

Loading

orogvany commented Aug 20, 2018

semuxgo commented Aug 20, 2018 •

edited

Loading

orogvany commented Aug 21, 2018 •

edited

Loading

orogvany Aug 21, 2018

semuxgo Aug 21, 2018

orogvany Aug 21, 2018 •

edited

Loading

orogvany Aug 21, 2018

orogvany Aug 21, 2018

semuxgo Aug 21, 2018

orogvany Aug 21, 2018

orogvany Aug 21, 2018

orogvany Aug 21, 2018

orogvany Aug 21, 2018

semuxgo Aug 21, 2018

orogvany Aug 21, 2018


		public class BytecodeCompiler {

		public static void main(String[] args) {


		public interface BlockStore {

		byte[] getBlockHashByNumber(int index);

VM: first integration #53

VM: first integration #53

Conversation

semuxgo commented Aug 18, 2018 • edited Loading

witoldsz Aug 18, 2018 • edited Loading

Choose a reason for hiding this comment

orogvany commented Aug 20, 2018 • edited Loading

orogvany commented Aug 20, 2018

semuxgo commented Aug 20, 2018 • edited Loading

orogvany commented Aug 21, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

orogvany Aug 21, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

semuxgo commented Aug 18, 2018 •

edited

Loading

witoldsz Aug 18, 2018 •

edited

Loading

orogvany commented Aug 20, 2018 •

edited

Loading

semuxgo commented Aug 20, 2018 •

edited

Loading

orogvany commented Aug 21, 2018 •

edited

Loading

orogvany Aug 21, 2018 •

edited

Loading