Non-Anchor Discriminator

Overview

The non-Anchor discriminator detector identifies raw Solana programs that lack instruction discrimination. While Anchor programs use 8-byte SHA256-based discriminators, raw programs should still implement instruction routing to prevent instruction confusion attacks. Programs that accept instruction data without any discrimination are vulnerable to executing unexpected instruction handlers.

For remediation guidance, see Non-Anchor Discriminator Remediation.

Why This Is an Issue

Without instruction discrimination, all incoming instruction data flows to a single handler or an ambiguous routing path. An attacker can submit instruction data crafted to trigger unintended code paths, potentially executing administrative functions or bypassing security checks. This is highly prevalent in early Solana programs and custom protocols.

How to Resolve

Before (Vulnerable)

// Vulnerable: processes instruction data without discriminator routing
pub fn process_instruction(
    program_id: &Pubkey,
    accounts: &[AccountInfo],
    instruction_data: &[u8],
) -> ProgramResult {
    // No routing -- all data goes to one handler
    process_data(accounts, instruction_data)
}

After (Fixed)

// Fixed: explicit instruction routing via discriminator byte
pub fn process_instruction(
    program_id: &Pubkey,
    accounts: &[AccountInfo],
    instruction_data: &[u8],
) -> ProgramResult {
    let (tag, rest) = instruction_data.split_first()
        .ok_or(ProgramError::InvalidInstructionData)?;
    match tag {
        0 => process_deposit(accounts, rest),
        1 => process_withdraw(accounts, rest),
        2 => process_admin(accounts, rest),
        _ => Err(ProgramError::InvalidInstructionData),
    }
}

Example JSON Finding

{
  "detector": "non-anchor-discriminator",
  "severity": "medium",
  "confidence": 0.6,
  "message": "Instruction data processed without discriminator routing",
  "location": { "function": "process_instruction", "block": 0, "statement": 1 }
}

Detection Methodology

Entry point analysis: Identifies the program entry point and instruction data handling.
Discriminator pattern search: Looks for branch/match operations on the first bytes of instruction data.
Anchor detection exclusion: Skips programs that use Anchor’s 8-byte discriminator pattern.
Routing coverage: Verifies that all instruction paths include a discriminator check.

Limitations

False positives: Programs with a single instruction that intentionally process all data uniformly. False negatives: Custom routing mechanisms that use instruction data offsets other than the first byte.

Anchor Discriminator Validation – Anchor discriminator issues
Native Discriminator Validation – account discriminators
Instruction Data Parsing – data parsing issues