Upgrade VM, crypto, and rand dependencies by huitseeker · Pull Request #3146 · 0xMiden/protocol

huitseeker · 2026-06-25T18:34:12Z

This updates Miden VM to v0.24, Miden Crypto to v0.27, and rand to v0.10.

The VM upgrade moved package debug info out of MastForest. This PR keeps that debug info attached through scripts, account code, and MAST forest stores by carrying PackageDebugInfoin LoadedMastForest.

This also updates MASM imports and build helpers for the new assembly package API.

Closes #3064

PhilippGackstatter · 2026-06-26T08:12:50Z

@@ -1 +1,4 @@
+#![cfg_attr(not(feature = "concurrent"), no_std)]


Do we need to run benchmarks for no_std or why is this needed? I assume this has something to do with all dependencies being made optional as well?

The issue is Cargo feature unification. The transaction benchmark uses the concurrent proving path. Since 0xMiden/crypto#1073, miden-lifted-stark/concurrent is part of the default crypto stack. In crypto 0.27, miden-crypto/concurrent enables std. On the protocol side, miden-tx/concurrent and miden-processor/concurrent also require std.

That works for the benchmark, but it must not leak into cargo build --no-default-features --target wasm32-unknown-unknown --workspace --lib. So the feature gate keeps normal benchmark builds as the default through default = ["concurrent"], while the no-default library build does not enable the std only benchmark stack.

/cc @WiktorStarczewski @Nashtare

PhilippGackstatter · 2026-06-26T08:31:19Z

It seems like a large part of the changes are related to packages and the way we assemble MASM. Does this not result in many conflicts with #2896? That workstream currently has 4 open PRs. So it seems it would make sense to do this migration after that issue was resolved, wdyt?

Yes, this PR is already doing part of issue #2896. The VM 0.24 upgrade removes the old shape of several Library and Program APIs, so this PR has to move some code to Package and MastForest. That is visible in account component code, note scripts, transaction scripts, and the build scripts.

The part this PR does not do is the project file migration. It adds no miden-project.toml files, and it does not move protocol or standards assembly to the declarative project layout from #3094 and #3107.

I think this leaves us with an ordering choice. If we want the dependency upgrade now, this PR can land first and the project file PRs can rebase over it. If we want less churn in the package migration work, this PR can wait and I can rebase it after #3094 and #3107.

/cc @TomasArrachea @bitwalker

I would merge this after #3094 and #3107, and then do another PR after this one that cleans up the parts of #3094/#3107 that are sort of hacky workarounds for the v0.23 APIs

Looking through this PR, I think I agree with @bitwalker that it would be better to mere #3094 and #3107 first.

This could work well with timing as I think @huitseeker you are out tomorrow (Wednesday).

partylikeits1983

Looks good! Really like the new import syntax. There are still many places where the imports can be consolidated. I opened an issue in the VM for adding unused import checking. I think the new import syntax makes it easier to have unused imports.

For the most part looks good so far, I didn't get through everything just yet, will re-review in 24-48h, or feel free to re-request my review if you address things sooner.

partylikeits1983 · 2026-06-29T12:38:03Z

@@ -1,4 +1,4 @@
-use agglayer::bridge::bridge_in -> bridge
+use agglayer::bridge::bridge_in as bridge


partylikeits1983 · 2026-06-29T12:42:41Z

+pub use {
+    FUNGIBLE_ASSET_MAX_AMOUNT,
+    ASSET_SIZE,
+    ASSET_VALUE_MEMORY_OFFSET,


Very clean!

partylikeits1983 · 2026-06-29T14:18:33Z

+use {
+    ACCOUNT_GET_ID_OFFSET,
+    ACCOUNT_GET_NONCE_OFFSET,
+    ACCOUNT_GET_INITIAL_COMMITMENT_OFFSET,
+    ACCOUNT_COMPUTE_COMMITMENT_OFFSET,
+    ACCOUNT_GET_CODE_COMMITMENT_OFFSET,
+    ACCOUNT_GET_INITIAL_STORAGE_COMMITMENT_OFFSET,
+    ACCOUNT_COMPUTE_STORAGE_COMMITMENT_OFFSET,
+    ACCOUNT_GET_INITIAL_VAULT_ROOT_OFFSET,
+    ACCOUNT_GET_VAULT_ROOT_OFFSET,
+    ACCOUNT_GET_ITEM_OFFSET,
+    ACCOUNT_HAS_STORAGE_SLOT_OFFSET,
+    ACCOUNT_GET_INITIAL_ITEM_OFFSET,
+    ACCOUNT_GET_MAP_ITEM_OFFSET,
+    ACCOUNT_GET_INITIAL_MAP_ITEM_OFFSET,
+    ACCOUNT_GET_ASSET_OFFSET,
+    ACCOUNT_GET_INITIAL_ASSET_OFFSET,
+    ACCOUNT_GET_NUM_PROCEDURES_OFFSET,
+    ACCOUNT_GET_PROCEDURE_ROOT_OFFSET,
+    ACCOUNT_HAS_PROCEDURE_OFFSET,
+} from miden::protocol::kernel_proc_offsets


When experimenting with this branch locally, I noticed its easy to add imports with this new import syntax, and not actually use the imported procedure or const. I opened an issue in the VM to discuss.

Yeah that is a bug if those aren't being used anywhere - we do have unused import checking implemented, but it may be buggy still.

For the record the imports being commented on in this comment are used.

bitwalker

I only looked over the assembly-related bits, but looks good to me, aside from a couple notes.

bitwalker · 2026-06-30T21:32:45Z

I think this whole file gets way simpler once #3094 gets merged - all we have to do is instantiate the project assembler with a package registry that is seeded with miden-core-lib, and request the various target artifacts we want. I'd suggest we rebase this on top of #3094 once that lands

mmagician · 2026-07-01T19:33:14Z

are the changes here correct? It seems like a bunch of constants are re-defined here now.

I think this will likely need to be rebased after #3094 shifted around a lot of these imports, but the re-definition of constants seems a separate issue so just wanted to flag that

mmagician · 2026-07-01T19:35:31Z

+# The number of field elements in an asset.
+const ASSET_SIZE=8
+
+# The size of the memory segment allocated to each note.
+const NOTE_MEM_SIZE=1024
+
+# The pointer at which the output vault root starts.
+const OUTPUT_VAULT_ROOT_PTR=8
+
+# The offset of the output note section.
+const OUTPUT_NOTE_SECTION_OFFSET=16777216
+


seems that constants have been re-defined across more than just one file

mmagician · 2026-07-01T19:58:52Z

        self.mast_store.insert(TransactionKernel::library().mast_forest().clone());
        self.mast_store.insert(program.mast_forest().clone());


IIUC, this used to load the debug info in the previous VM versions, and will now fall back to LoadedMastForest::new(...) which no longer contains debug info

and similarly in other users of TransactionMastStore::insert

mmagician · 2026-07-01T20:00:48Z

+    /// Registers all procedures of the provided [Package] with this store.
+    fn insert_package(&self, package: &Package) {
+        self.insert_loaded(loaded_mast_forest_from_package(package));
+    }


The key function here seems to loaded_mast_forest_from_package which adds the debug info.
It might be worth marking this explicitly in this fn's docs

mmagician · 2026-07-01T20:02:06Z

    /// Registers all procedures of the provided [MastForest] with this store.
    pub fn insert(&self, mast_forest: Arc<MastForest>) {
+        self.insert_loaded(LoadedMastForest::new(mast_forest));
+    }


similarly here (and IIUC) this fn will NOT add debug info, whereas previously it would have by default, because the debug info lived on the MastForest object

mmagician · 2026-07-01T20:05:27Z

    #[must_use]
-    pub fn with_debug_mode(mut self) -> Self {
-        self.exec_options = self.exec_options.with_debugging(true);
+    pub fn with_debug_mode(self) -> Self {
        self
    }


is this dead code now?

huitseeker requested review from PhilippGackstatter, mmagician and partylikeits1983 June 25, 2026 18:34

huitseeker marked this pull request as ready for review June 25, 2026 19:57

huitseeker marked this pull request as draft June 25, 2026 19:58

huitseeker marked this pull request as ready for review June 25, 2026 21:01

PhilippGackstatter reviewed Jun 26, 2026

View reviewed changes

huitseeker force-pushed the issue-3064-dependency-upgrade branch from ce78962 to f09c095 Compare June 28, 2026 13:50

bobbinth mentioned this pull request Jun 28, 2026

feat: migrate protocol crate to miden-project #3094

Merged

partylikeits1983 reviewed Jun 29, 2026

View reviewed changes

huitseeker force-pushed the issue-3064-dependency-upgrade branch 2 times, most recently from a2502cf to b2a6275 Compare June 30, 2026 14:18

bitwalker reviewed Jun 30, 2026

View reviewed changes

huitseeker added 14 commits June 30, 2026 18:10

Upgrade Miden VM and crypto dependencies

7760aac

Address dependency upgrade review findings

82b84c7

Validate kernel API offset exports

d33fbb4

Preserve package debug info after MAST upgrade

f66f146

Tighten dependency upgrade cleanup

3617763

chore: Changelog

d844621

Fix dependency upgrade CI

ad085b9

Fix package rustdoc links

6e29616

Exclude transaction benchmark from no-std build

248808e

Restore transaction benchmark no-std build

84cf6ce

chore: explain transaction bench feature gate

1938df0

chore: tighten MASM imports

d077782

Fix rebase fallout

4a0ee4d

Fix standards package asset extension

1c2a973

huitseeker force-pushed the issue-3064-dependency-upgrade branch from b2a6275 to 1c2a973 Compare June 30, 2026 22:13

PhilippGackstatter mentioned this pull request Jul 1, 2026

Refactor shared build utilites into crate #3179

Open

mmagician reviewed Jul 1, 2026

View reviewed changes

bobbinth mentioned this pull request Jul 1, 2026

feat: migrate transaction scripts to Library #3173

Open

		@@ -1 +1,4 @@
		#![cfg_attr(not(feature = "concurrent"), no_std)]

		@@ -1,4 +1,4 @@
		use agglayer::bridge::bridge_in -> bridge
		use agglayer::bridge::bridge_in as bridge

		self.mast_store.insert(TransactionKernel::library().mast_forest().clone());
		self.mast_store.insert(program.mast_forest().clone());

Uh oh!

Conversation

huitseeker commented Jun 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

huitseeker Jun 26, 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!

partylikeits1983 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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

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!

Uh oh!

Uh oh!

bitwalker left a comment

Choose a reason for hiding this comment

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!

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!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

huitseeker commented Jun 25, 2026 •

edited

Loading

huitseeker Jun 26, 2026 •

edited

Loading

partylikeits1983 left a comment •

edited

Loading