feat: merge-train/fairies-v5 - #24769
Merged
Merged
Conversation
Prevents multi-tab concurrency issues in SQLite store creation by guarding it with weblocks. Closes F-829
This is a defensive measure in case more than one SQLite page file is found pointing to the same logical name, which could cause undefined behavior. This could happen with old, pre-weblock controlled versions of wallet/pxe (see #24740), so it's unlikely to be found in the wild, but it gives us graceful coverage if that happens. We detect if a pool contains two valid .opaque files mapped to the same logical SQLite path, then: 1. Acquire the pool Web Lock. 2. Copy the entire pool byte-for-byte into: .aztec-sqlite-quarantine/<timestamp-random>/ 3. Verify the copied directory and file contents. 4. Write a quarantine.json describing the duplicate mappings. 5. Delete the original active pool. 6. Open a new, empty database under the original pool name. 7. Emit a warning log containing the quarantine location. The caller receives a successfully opened but empty store, so wallet/PXE state would need to be recreated or resynchronized. The quarantined bytes remain available for forensic or manual recovery, although there is currently no public API or UI for that. If copying or verification fails, opening fails and the original pool is not intentionally removed. If the pool merely comes from an old version but has no duplicate logical mappings, nothing special happens, it opens normally. --------- Co-authored-by: Gregorio Juliana <gregojquiros@gmail.com>
#24689) ## Motivation The `#[note]` macro generated each `PropertySelector` from the field's position in the struct declaration, but selectors are applied to the note's packed representation. The two only agree when every field packs to one `Field`. For a note with a multi-slot field (a `Point`, an array, a nested struct), every filter on a later field silently constrained an unrelated packed slot — on the constrained read path, where these checks are what bind oracle-returned notes to the contract's criteria. A malicious PXE could satisfy a filter with a note that does not match it. No shipped contract is affected (all first-party filters target single-slot fields with single-slot predecessors); the bug was latent in the library. ## The change - Selector indices are the field's packed offset: the accumulated sum of preceding fields' `Packable::N`, mirroring `derive(Packable)`'s layout. - `PropertySelector<T>` carries the selected field's type. `select`/`sort` reject fields that pack to more than one `Field` at compile time (a one-slot criterion cannot express them), and `select`'s value is typed as the field's type. - Criterion values are compared as `value.pack()[0]` instead of `value.to_field()`, matching what the packed note slot actually contains. - `properties()` statically asserts the note's packed length equals the sum of its field packed lengths, so custom `Packable` layouts get a compile error directing to hand-written selectors instead of wrong ones. A same-length field reorder is not detectable; that residual gap is documented. - Breaking: hand-written `PropertySelector` literals need a type annotation, and every note field type must implement `Packable`. Migration notes included. All existing contracts compile to identical circuits; the PXE already applied indices to the packed layout, so no TS changes. ## Future changes The root cause is that two independent authorities describe the same layout: the note's `Packable` impl owns it, and the macro assumes it. Every assertion in this PR is a consistency check between the two, and one divergence (a custom pack that reorders fields at the same total length) cannot be checked at all and still fails silently. A potential fix (maybe there is a better one) is for the note macro to own the layout: `#[note]` derives `Packable` itself and generates the selectors from the same field list, making divergence impossible by construction, with an explicit custom-packing opt-out that generates neither and leaves both to the author. That is a larger breaking change to the macro surface and is left for a follow-up. Fixes F-800
Collaborator
Author
|
🤖 Auto-merge enabled after 4 hours of inactivity. This PR will be merged automatically once all checks pass. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
BEGIN_COMMIT_OVERRIDE
feat: weblock controlled opfs pool (#24740)
chore: handle legacy duplicate opaque handles (#24743)
fix(aztec-nr)!: compute note property selectors from the packed layout (#24689)
END_COMMIT_OVERRIDE