Adding support for Machine Integer version of Bytes, List, and MInt functions #1209

Robertorosmaninho · 2025-06-28T17:17:03Z

This PR introduces the concrete implementation of the following new Machine Integer hooks:

1.  syntax {Width} MInt{Width} ::=MInt{Width} "^MInt" MInt{Width}
2.  syntax {Width} Bytes ::= MInt2Bytes(MInt{Width})
3.  syntax {Width} MInt{Width} ::= Bytes2MInt(Bytes)
4.  syntax {Width} MInt{Width} ::= lengthBytes(Bytes)
5.  syntax {Width} Bytes ::= padRightBytes(Bytes, length: MInt{Width}, value: MInt{Width})
6.  syntax {Width} Bytes ::= padLeftBytes(Bytes, length: MInt{Width}, value: MInt{Width})
7.  syntax {Width} Bytes ::= replaceAtBytes(dest: Bytes, index: MInt{Width}, src: Bytes)
8.  syntax {Width} Bytes ::= substrBytes(Bytes, startIndex: MInt{Width}, endIndex: MInt{Width})
9.  syntax {Width} MInt{Width} ::= Bytes "[" MInt{Width} "]"
10. syntax {Width} Bytes ::= Bytes "[" index: MInt{Width} "<-" value: MInt{Width} "]"
11. syntax {Width} MInt{Width} ::= size(List)
12. syntax {Width} List ::= List "[" index: MInt{Width} "<-" value: KItem "]"
13. syntax {Width} KItem ::= List "[" MInt{Width} "]"

Most of these hooks have an initial implementation for 64 and 256-bit Integers, being the former defined in C++ in its respective collection file while the latter is defined in the llvm_header.inc file, as C++ doesn't support 256-bit integer natively.

The only exceptions for now are 12 and 13, which are only implemented for 64-bit, as we have the immer limitation that the index should be a in64_t.

…d MInt{64} Implementing `hook_MINT_pow256` in LLVM IR Implementing `hook_MINT_pow64` in int.cpp Adding `mint-pow` unit test

Implementing `hook_MINT_MInt2Bytes` in LLVM IR Adding `mint2bytes` unit test

Implementing `hook_MINT_Bytes2MInt` in LLVM IR Adding `bytes2mint` unit test

…} and MInt{64} Implementing `hook_BYTES_length256` in LLVM IR Implementing `hook_BYTES_length64` in bytes.cpp Adding `lengthbytes` unit test

…56} and MInt{64} Implementing `hook_BYTES_padRight256` in LLVM IR Implementing `hook_BYTES_padRight64` in bytes.cpp Adding `padrightbytes` unit test

…6} and MInt{64} Implementing `hook_BYTES_padLeft256` in LLVM IR Implementing `hook_BYTES_padLeft64` in bytes.cpp Adding `padleftbytes` unit test

…256} and MInt{64} Implementing `hook_BYTES_replaceAt256` in LLVM IR Implementing `hook_BYTES_replaceAt64` in bytes.cpp Adding `replaceatbytes` unit test Declaring functions needed by other functions in the `llvm_hearder.inc` Defining `interger_overflow` used in used in `hook_BYTES_pad*` functions in `llvm_hearder.inc`

…} and MInt{64} Implementing `hook_BYTES_substr256` in LLVM IR Implementing `hook_BYTES_substr64` in bytes.cpp Adding `substrbytes` unit test

…nd MInt{64} Implementing `hook_BYTES_get256` in LLVM IR Implementing `hook_BYTES_get64` in bytes.cpp Adding `bytes-get` unit test

…} and MInt{64} Implementing `hook_BYTES_update256` in LLVM IR Implementing `hook_BYTES_update64` in bytes.cpp Adding `bytes-update` unit test

…nd MInt{64} Implementing `hook_LIST_size256` in LLVM IR Implementing `hook_LIST_size64` in bytes.cpp Adding `list-size` unit test

Implementing `hook_LIST_get64` in bytes.cpp Adding `list-get` unit test

…64} only Adding `list-update` unit test

dwightguth

I'd suggest considering the following two factors regarding the hooks added:

Is there a particularly compelling reason why both arguments of hooks with more than one mint arg need to have the same size?
If a mint arg or return value carresponds to a size in bytes or in elements of something in memory, you should be able to use a non parametric mint 64 for it.

None of this is absolutely necessary but if we're adding hooks to the frontend we ought to think about this.

dwightguth · 2025-07-08T15:23:22Z

runtime/collections/lists.cpp

@@ -111,6 +120,10 @@ size_t hook_LIST_size_long(SortList list) {
  return list->size();
 }

+uint64_t hook_LIST_size64(SortList list) {


hook_list_size_long already does this. Check for other similar duplicates that may exist.

But it can return a 32-bit in 32-bit machines, right? As in CreateTerm.cpp, we set this return type fixed in i64; wouldn't it be better to be sure we're always returning the correct type?

Robertorosmaninho · 2025-07-08T16:19:30Z

Is there a particularly compelling reason why both arguments of hooks with more than one mint arg need to have the same size?

Not necessarily, but that's how we defined it in the frontend. We also haven't reached the need to use 2 different bit sizes as well.

If a mint arg or return value corresponds to a size in bytes or in elements of something in memory, you should be able to use a non parametric mint 64 for it.

I agree, and it makes sense, but it will enforce a roundMInt to users of this function if they're using a type different from i64, right? In our case, we're mainly using i256 in EVM, so if we enforce an i64 here, we'll have to delegate the responsibility to truncate the integer to the user instead of the function. My initial idea was to delegate this responsibility to the function to make the interface nicer. What do you think?

Also, @dwightguth, do you have any comments on how I can have:

A List return type in CreateTerm.cpp; and
How is the correct way to handle the list data type in LLVM IR, so we can have i256 versions of the List hooks there as well?

config/llvm_header.inc

mariaKt

Thank you @Robertorosmaninho. I added only two minor comments.

config/llvm_header.inc

runtime/strings/bytes.cpp

rv-jenkins changed the base branch from master to develop June 28, 2025 17:17

Robertorosmaninho force-pushed the mint64-hooks branch from 56451cf to 8482578 Compare July 3, 2025 00:45

Robertorosmaninho added 13 commits July 4, 2025 15:33

Implementing hook_MINT_powMint codegen with support to MInt{256} an…

df2492b

…d MInt{64} Implementing `hook_MINT_pow256` in LLVM IR Implementing `hook_MINT_pow64` in int.cpp Adding `mint-pow` unit test

Implementing hook_MINT_MInt2Bytes codegen with support to MInt{256}

64d46c5

Implementing `hook_MINT_MInt2Bytes` in LLVM IR Adding `mint2bytes` unit test

Implementing hook_MINT_Bytes2MInt codegen with support to MInt{256}

813f91d

Implementing `hook_MINT_Bytes2MInt` in LLVM IR Adding `bytes2mint` unit test

Implementing hook_BYTES_lengthMInt codegen with support to MInt{256…

8229b1a

…} and MInt{64} Implementing `hook_BYTES_length256` in LLVM IR Implementing `hook_BYTES_length64` in bytes.cpp Adding `lengthbytes` unit test

Implementing hook_BYTES_padRightMInt codegen with support to MInt{2…

7797ded

…56} and MInt{64} Implementing `hook_BYTES_padRight256` in LLVM IR Implementing `hook_BYTES_padRight64` in bytes.cpp Adding `padrightbytes` unit test

Implementing hook_BYTES_padLeftMInt codegen with support to MInt{25…

0e0b832

…6} and MInt{64} Implementing `hook_BYTES_padLeft256` in LLVM IR Implementing `hook_BYTES_padLeft64` in bytes.cpp Adding `padleftbytes` unit test

Implementing hook_BYTES_substrMInt codegen with support to MInt{256…

3e5b1e5

…} and MInt{64} Implementing `hook_BYTES_substr256` in LLVM IR Implementing `hook_BYTES_substr64` in bytes.cpp Adding `substrbytes` unit test

Implementing hook_BYTES_getMInt codegen with support to MInt{256} a…

6535b5c

…nd MInt{64} Implementing `hook_BYTES_get256` in LLVM IR Implementing `hook_BYTES_get64` in bytes.cpp Adding `bytes-get` unit test

Implementing hook_BYTES_updateMInt codegen with support to MInt{256…

5178cce

…} and MInt{64} Implementing `hook_BYTES_update256` in LLVM IR Implementing `hook_BYTES_update64` in bytes.cpp Adding `bytes-update` unit test

Implementing hook_LIST_sizeMInt codegen with support to MInt{256} a…

3669211

…nd MInt{64} Implementing `hook_LIST_size256` in LLVM IR Implementing `hook_LIST_size64` in bytes.cpp Adding `list-size` unit test

Implementing hook_LIST_getMInt codegen with support to MInt{64}

5d7dd50

Implementing `hook_LIST_get64` in bytes.cpp Adding `list-get` unit test

Implementing hook_LIST_updateMInt in bytes.cpp for supporting MInt{…

d7efaf7

…64} only Adding `list-update` unit test

Robertorosmaninho force-pushed the mint64-hooks branch from d1d5580 to d7efaf7 Compare July 4, 2025 19:36

Robertorosmaninho requested review from mariaKt and dwightguth July 4, 2025 19:37

Robertorosmaninho self-assigned this Jul 4, 2025

Robertorosmaninho marked this pull request as ready for review July 4, 2025 19:38

dwightguth reviewed Jul 8, 2025

View reviewed changes

Robertorosmaninho requested a review from theo25 July 10, 2025 15:19

theo25 reviewed Jul 10, 2025

View reviewed changes

config/llvm_header.inc Outdated Show resolved Hide resolved

mariaKt requested changes Jul 10, 2025

View reviewed changes

config/llvm_header.inc Outdated Show resolved Hide resolved

runtime/strings/bytes.cpp Outdated Show resolved Hide resolved

Fixing typo

d8749dc

Robertorosmaninho requested review from mariaKt, dwightguth and theo25 July 14, 2025 12:29

Using hook_BYTES_length64 instead of hardcoded LENGTH_MASK

8bc0a9b

theo25 approved these changes Jul 14, 2025

View reviewed changes

Reusing error functions to emit messages on Int hooks as well

60c6c99

mariaKt approved these changes Jul 14, 2025

View reviewed changes

Robertorosmaninho added the automerge label Jul 14, 2025

automergerpr-permission-manager bot merged commit bb5318b into develop Jul 14, 2025
10 checks passed

automergerpr-permission-manager bot deleted the mint64-hooks branch July 14, 2025 19:26

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Adding support for Machine Integer version of Bytes, List, and MInt functions #1209

Adding support for Machine Integer version of Bytes, List, and MInt functions #1209

Uh oh!

Robertorosmaninho commented Jun 28, 2025 •

edited

Loading

Uh oh!

dwightguth left a comment

Uh oh!

dwightguth Jul 8, 2025

Uh oh!

Robertorosmaninho Jul 8, 2025

Uh oh!

Robertorosmaninho commented Jul 8, 2025 •

edited

Loading

Uh oh!

Uh oh!

mariaKt left a comment •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Adding support for Machine Integer version of Bytes, List, and MInt functions #1209

Adding support for Machine Integer version of Bytes, List, and MInt functions #1209

Uh oh!

Conversation

Robertorosmaninho commented Jun 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dwightguth left a comment

Choose a reason for hiding this comment

Uh oh!

dwightguth Jul 8, 2025

Choose a reason for hiding this comment

Uh oh!

Robertorosmaninho Jul 8, 2025

Choose a reason for hiding this comment

Uh oh!

Robertorosmaninho commented Jul 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

mariaKt left a comment • 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!

Robertorosmaninho commented Jun 28, 2025 •

edited

Loading

Robertorosmaninho commented Jul 8, 2025 •

edited

Loading

mariaKt left a comment •

edited

Loading