|
| 1 | +--- |
| 2 | +simd: '0290' |
| 3 | +title: Relax Fee Payer Constraint |
| 4 | +authors: |
| 5 | + - Andrew Fitzgerald (Anza) |
| 6 | +category: Standard |
| 7 | +type: Core |
| 8 | +status: Review |
| 9 | +created: 2025-05-29 |
| 10 | +feature: |
| 11 | +--- |
| 12 | + |
| 13 | +## Summary |
| 14 | + |
| 15 | +This proposal aims to relax the handling of invalid fee payers in Solana. |
| 16 | +Currently, if a transaction with an invalid fee payer is included in a block, |
| 17 | +the entire block is rejected. |
| 18 | +This proposal suggests that instead of rejecting the entire block, the |
| 19 | +transaction with the invalid fee payer should simply skip execution. |
| 20 | + |
| 21 | +## Motivation |
| 22 | + |
| 23 | +The current constraint forces block-validation to be synchronous since in order |
| 24 | +to determine if a block is valid or not, some subset of transactions must be |
| 25 | +executed in order to determine if the fee payer has sufficient funds to pay the |
| 26 | +transaction fees. |
| 27 | +By relaxing this constraint, we move one step closer towards asynchronous |
| 28 | +validation/execution. |
| 29 | + |
| 30 | +## New Terminology |
| 31 | + |
| 32 | +None. |
| 33 | + |
| 34 | +## Detailed Design |
| 35 | + |
| 36 | +A transaction has a statically determined fee `fee` lamports. |
| 37 | +A transaction can successfully pay fees if: |
| 38 | + |
| 39 | +- The fee payer account has exactly `fee` lamports. |
| 40 | +- The fee payer account has at least X + `rent_exempt_reserve` lamports. |
| 41 | + |
| 42 | +If the fee payer account does not meet either of these conditions, the transaction |
| 43 | +may be included in a block, but it must not be executed. |
| 44 | +The transaction will have no effect on the state. |
| 45 | + |
| 46 | +Invalid fee payer transactions will count their requested, |
| 47 | +or default, compute units (CUs) towards block limits. |
| 48 | +This is intended to make it strictly cheaper to process invalid fee payer |
| 49 | +transactions compared to valid fee payer transactions of the same construction. |
| 50 | + |
| 51 | +## Alternatives Considered |
| 52 | + |
| 53 | +- Requiring some sort of fee-payer lock up/bonding mechanism to ensure that fee |
| 54 | + payers have sufficient funds to pay for the transaction fees. |
| 55 | + - This is more complex compared to this proposal. |
| 56 | + |
| 57 | +## Impact |
| 58 | + |
| 59 | +Transactions that are unable to pay fees may be included in blocks. |
| 60 | + |
| 61 | +## Security Considerations |
| 62 | + |
| 63 | +- Possible attack vector where a malicious leader can spam transactions with |
| 64 | + invalid fee payers. Mitigation for this is charging full CUs for these. |
| 65 | + |
| 66 | +## Drawbacks |
| 67 | + |
| 68 | +- If there is no interest in simplifying block-validation to allow for |
| 69 | + asynchronous, this proposal is not necesary. |
| 70 | +- Concern about data propagation for without paying fees to the network (burn): |
| 71 | + - This concern has been raised in the past when this has been discussed. |
| 72 | + - However, the concern is largely invalid since even without this proposal, |
| 73 | + a malicious leader could still propagate data through the network for free |
| 74 | + by simply using an invalid fee payer. |
| 75 | + |
| 76 | +## Backwards Compatibility |
| 77 | + |
| 78 | +- All previously valid transactions and blocks are still valid. |
| 79 | +- Blocks produced after the change may be rejected by previous versions of the |
| 80 | + validator client(s). |
0 commit comments