Summary
kora-lib: Unrecognized Instruction Types Create Empty Stubs That Bypass Fee Payer Policy
When inner CPI instructions use instruction types not recognized by Kora's parser (including Token-2022 extension instructions like ConfidentialTransfer, TransferFeeExtension::WithdrawWithheldTokens, etc.), they are reconstructed as stub instructions with empty accounts and empty data. These stubs fail deserialization during fee payer policy validation and are silently skipped, meaning any fee payer usage within those instructions goes completely unchecked.
Severity
Medium
Affected Component
- File:
crates/lib/src/transaction/instruction_util.rs - Functions:
reconstruct_system_instruction(),reconstruct_spl_token_instruction() - Lines: 750–753, 1187–1189
Root Cause
The instruction reconstruction functions have a catch-all _ => arm for unrecognized instruction types that creates a stub CompiledInstruction with the correct program_id_index but empty accounts and empty data. When this stub reaches the fee payer policy parsing (parse_system_instructions / parse_token_instructions), deserialization of empty data fails. The parsing functions also have a catch-all _ => {} that silently skips the failed instruction. The result: the instruction exists in all_instructions (so program allowlist checks pass), but fee payer policy is never enforced on it.
Vulnerable Code
Stub Creation
// crates/lib/src/transaction/instruction_util.rs:750-753
// System program, unrecognized instruction type:
_ => {
log::error!("Unsupported system instruction type: {}", instruction_type);
Ok(Self::build_default_compiled_instruction(program_id_index))
}
// crates/lib/src/transaction/instruction_util.rs:1187-1189
// SPL Token program, unrecognized instruction type:
_ => {
log::error!("Unsupported token instruction type: {}", instruction_type);
Ok(Self::build_default_compiled_instruction(program_id_index))
}
The stub builder:
pub fn build_default_compiled_instruction(program_id_index: u8) -> CompiledInstruction {
CompiledInstruction {
program_id_index,
accounts: vec![], // <-- No accounts
data: vec![], // <-- No data
}
}
Silent Skip During Policy Parsing
// In parse_system_instructions:
if let Ok(system_instruction) = bincode::deserialize::<SystemInstruction>(&instruction.data) {
match system_instruction {
// ... known types handled ...
_ => {} // <-- Unrecognized: silently skipped
}
}
// If deserialize fails (empty data), the entire `if let Ok` block is skipped.
// The instruction is not added to any policy check map.
// In parse_token_instructions:
if let Ok(token_instruction) = TokenInstruction::unpack(&instruction.data) {
match token_instruction {
// ... known types handled ...
_ => {} // <-- Unrecognized: silently skipped
}
}
// Same: empty data causes unpack to fail, instruction completely invisible to policy.
Proof of Concept
Affected Token-2022 Extension Instructions
The following Token-2022 extension instruction types are NOT handled by Kora's parser and would produce empty stubs:
| Extension | Instruction | Risk if Fee Payer is Authority |
|---|---|---|
TransferFeeExtension |
WithdrawWithheldTokensFromMint |
Fee payer as withdraw authority can drain withheld fees |
TransferFeeExtension |
WithdrawWithheldTokensFromAccounts |
Same |
TransferFeeExtension |
HarvestWithheldTokensToMint |
Fee collection manipulation |
ConfidentialTransfer |
Transfer |
Hidden transfer amounts bypass fee tracking |
ConfidentialTransfer |
Withdraw |
Hidden withdrawals |
InterestBearingMint |
UpdateRate |
Fee payer as rate authority can manipulate interest |
TransferHook |
Execute |
Arbitrary hook execution |
GroupMemberPointer |
Update |
Metadata manipulation |
MetadataPointer |
Update |
Metadata manipulation |
PermanentDelegate |
Transfer (via delegate) |
Delegate-based unauthorized transfers |
Code Path Trace
1. Transaction contains an inner CPI instruction:
Program: Token-2022
Type: "withdrawWithheldTokensFromMint" (TransferFeeExtension)
Accounts: [fee_payer (as withdraw_withheld_authority), mint, destination]
2. RPC returns this as a Parsed inner instruction
3. reconstruct_spl_token_instruction() is called:
- instruction_type = "withdrawWithheldTokensFromMint"
- No match in the known types (transfer, transferChecked, burn, etc.)
- Falls through to _ => arm
- Returns: CompiledInstruction { program_id_index, accounts: [], data: [] }
4. Stub is added to all_instructions
→ validate_programs() sees Token-2022 program ID → PASS (allowed)
→ validate_disallowed_accounts() sees no accounts in the stub → PASS
5. parse_token_instructions() processes the stub:
- TokenInstruction::unpack(&[]) → Err (empty data)
- if let Ok(...) block skipped entirely
- Instruction not added to any ParsedSPLInstructionType map
6. validate_fee_payer_usage() iterates parsed SPL instructions:
- No entry for "withdrawWithheldTokensFromMint"
- Fee payer's usage as withdraw_withheld_authority is NEVER checked
7. Transaction is signed by Kora
8. On-chain: fee_payer (as withdraw authority) withdraws withheld
transfer fees from the mint to attacker's account
Verifiable Test
#[test]
fn test_unrecognized_instruction_produces_empty_stub() {
// Simulate what happens for an unrecognized Token-2022 instruction
let program_id_index: u8 = 3; // Token-2022 at index 3
// This is what the catch-all arm produces:
let stub = IxUtils::build_default_compiled_instruction(program_id_index);
assert_eq!(stub.accounts.len(), 0); // No accounts
assert_eq!(stub.data.len(), 0); // No data
// Attempt to parse it:
let result = TokenInstruction::unpack(&stub.data);
assert!(result.is_err()); // Cannot parse empty data
// Therefore: fee payer policy is never applied to this instruction
// The fee payer could be the withdraw_withheld_authority in the
// REAL instruction, but the stub has zero accounts, invisible.
}
References
crates/lib/src/transaction/instruction_util.rs:750-753, system instruction catch-allcrates/lib/src/transaction/instruction_util.rs:1187-1189, SPL token instruction catch-allcrates/lib/src/transaction/instruction_util.rs:316-319,build_default_compiled_instruction- SPL Token-2022 instruction types, full list of extension instructions
Impact
- Fee Payer Policy Bypass: Token-2022 extension instructions that use the fee payer as an authority are invisible to policy enforcement.
- Forward-Looking Risk: As Solana and SPL Token-2022 add new instruction types, they will automatically bypass all fee payer policy checks in Kora.
- Precondition: Requires the fee payer to hold some authority role (e.g.,
withdraw_withheld_authority,permanent_delegate) for Token-2022 accounts. This is unlikely in typical deployments but possible in misconfigured setups.
Affected versions
Security releases
Kodem intelligence
Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.
Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.
Already deployed Kodem?
See it in your environmentNew to Kodem? Get a demo →Remediation advice
Reject transactions containing inner instructions with unrecognized types (fail-secure):
// In reconstruct_system_instruction:
_ => {
return Err(KoraError::InvalidTransaction(format!(
"Unrecognized system instruction type '{}' in CPI, \
cannot validate fee payer policy. Transaction rejected.",
instruction_type
)));
}
// In reconstruct_spl_token_instruction:
_ => {
return Err(KoraError::InvalidTransaction(format!(
"Unrecognized SPL Token instruction type '{}' in CPI, \
cannot validate fee payer policy. Transaction rejected.",
instruction_type
)));
}
Alternatively, maintain a list of known-safe instruction types that don't involve authority checks, and only reject truly unknown types.
Frequently Asked Questions
- What is GHSA-X442-M7CC-HR92? GHSA-X442-M7CC-HR92 is a medium-severity security vulnerability in kora-lib (rust), affecting versions < 2.0.5. It is fixed in 2.0.5.
- Which versions of kora-lib are affected by GHSA-X442-M7CC-HR92? kora-lib (rust) versions < 2.0.5 is affected.
- Is there a fix for GHSA-X442-M7CC-HR92? Yes. GHSA-X442-M7CC-HR92 is fixed in 2.0.5. Upgrade to this version or later.
- Is GHSA-X442-M7CC-HR92 exploitable, and should I be worried? Whether GHSA-X442-M7CC-HR92 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
- What actually determines whether GHSA-X442-M7CC-HR92 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
- How do I fix GHSA-X442-M7CC-HR92? Upgrade
kora-libto 2.0.5 or later.