CVE Index中文
L3 · ShartedReth · Ethereum · integrityreth-state-root

Reth trie bug can omit new state entries; upstream and downstream fixes are available

Reth's trie iterator could compute an incorrect incremental state root. Upstream 2.10.0, Base 1.5.3, op-reth 2.8.1, Morph 1.4.1 and BNB reth-bsc 0.1.4-hotfix contain fixes. Celo, Mantle and Gravity have scoped source backports; other release coverage and full ranges need verification. No exploitation or fund loss is established.

Disclosed
Nov 3, 2026
Updated Nov 6, 2026
Severity
Sharted
No CVE: a breach, not a bug
Splash zone
Nodes and applications using an affected Reth trie implementation; exposure depends on the actual fork and pinned revision.
Reth · Ethereum · integrity
In the wild
Unknown
No exploitation or fund loss is established in the reviewed original fix records.

What happened

The iterator combines stored trie branches with newly hashed state entries. Its previous shortcut could skip a seek when the stored branch was deeper than the seek prefix, leaving a newly inserted leaf out of the incremental root. The patch permits that shortcut only when the keys are equal. The upstream regression test and Celo’s backport explanation describe this concrete correctness failure.

Reth 2.10.0 lists PR 28264, and Base 1.5.3 explicitly says it prevents a state-root mismatch through that same patch. Celo and Mantle published narrower backports tied to their own dependency pins. A merged backport does not establish that every deployed node or every branch already runs it. The op-reth 2.8.1 tag also contains the backport: its release commit merges PR 23924 and rust/Cargo.lock pins the corrected fork revision. Its release notes only mention an unrelated RPC correction, so those notes alone would miss this fix. Full vulnerable-version ranges and other downstream packaged-release thresholds remain incomplete.

L3 reflects an integrity bug in a widely used execution-client family and an actionable operator update. These original records do not establish an attacker stealing assets, remote code execution, a chain-wide compromise or confirmed user losses. Claims about paused bridges or incorrect proof responses require their own original evidence and are not used to expand the verified scope here.

November 6 BNB backports: BNB Reth merged the upstream-equivalent correction on main and separately on its v2.8 integration branch. Verify the actual fork/commit in the deployed artifact; a merged source patch is not evidence of a released build or fleet-wide deployment. These technical records do not establish a chain-wide incident, confirmed user loss or the extent of BNB Chain exposure.

November 6 Morph and Gravity follow-up: Morph’s released morph-reth 1.4.1 hotfix uses upstream Reth 2.5.2 plus the trie fix through morph-l2/reth v2.5.2-morph.1, avoiding an immediate full upstream upgrade. Its release requires upgrading every node, including followers, without a database migration or resync. Gravity separately merged its equivalent correction on main; no fixed Gravity release number was established. These records expand patch coverage, not evidence of theft or a chain-wide compromise.

BNB released hotfix: The node release reth-bsc v0.1.4-hotfix now provides a released node remediation, beyond the earlier source merges. Its dependency tag bnb-chain/reth v0.1.4 is marked prerelease; those are separate repositories and version contexts. Follow the hotfix instructions for the actual reth-bsc deployment. This does not establish that every BNB client uses the affected code, that the separate v2.8 branch has a released fix, or that exploitation occurred.

Fixed versions

ProductSafe fromNote
Upstream Reth2.10.0Contains the trie fix; use a build compatible with the operated chain
OP Stack op-reth2.8.1Tagged lockfile contains op-v2.6.0-2 backport; release notes mention only a separate RPC fix
Base Mainnet and Sepolia node release1.5.3Base recommends upgrading; release notes explicitly link the state-root fix
Morph morph-reth1.4.1Released hotfix; pins morph-l2/reth v2.5.2-morph.1 with the upstream-equivalent fix
BNB reth-bsc0.1.4-hotfixReleased hotfix pins bnb-chain/reth v0.1.4; upgrade and restart, no database migration or hardfork for this fix

Crime-scene timeline

  1. Nov 3, 21:00 UTC

    Upstream merges the iterator-boundary fix

    PR 28264 merges as 8572c88 and includes an incremental-root regression test.

  2. Nov 5, 2026

    Celo and Mantle merge scoped backports

    Celo updates its celo/reth-v1.0 fork pin; Mantle backports onto its production mantle-v1.6.2 line. Other branches require separate verification.

  3. Nov 5, 2026

    Reth 2.10.0, op-reth 2.8.1 and Base 1.5.3 contain the fix

    Upstream lists the trie correction; the op-reth release tag pins its backport despite RPC-only release notes. Base recommends its new build for Mainnet and Sepolia and links the same fix.

  4. Nov 6, 02:19 UTC

    BNB Reth merges its main-line backport

    PR 254 merges as 3b2bd020; it backports the same upstream iterator correction and regression test.

  5. Nov 6, 02:20 UTC

    BNB publishes the patched Reth dependency tag

    bnb-chain/reth v0.1.4 includes PR 254; this dependency tag is marked prerelease.

  6. Nov 6, 02:40 UTC

    BNB Reth merges a separate v2.8 integration backport

    PR 255 merges as 141cc278 into fix/bsc-static-file-td-611. This establishes a branch-specific patch, not a released safe-version threshold.

  7. Nov 6, 02:53 UTC

    BNB releases reth-bsc 0.1.4-hotfix

    The released node hotfix pins reth v0.1.4 and recommends an immediate upgrade for all reth-bsc mainnet/testnet node operators. Install and restart; this fix needs no database migration or hardfork.

  8. Nov 6, 2026

    Morph releases morph-reth 1.4.1

    The hotfix pins the corrected fork and adds a regression test. All Morph nodes, including followers, should upgrade; no database migration or resync is required.

  9. Nov 6, 2026

    Gravity merges a source backport

    PR 464 merges as c407bd9 into main. This is a verified source patch, not a released safe-version threshold.

Who said what

We list each claim and who made it. Primary sources (the vendor, the researcher, the registry) are marked. Where reports disagree, we grade by the worst credible case.