runtime: Implement rent-adjusted stake delegations for amended SIMD-0392 (formerly SIMD-0488) by joncinque · Pull Request #11332 · anza-xyz/agave

joncinque · 2026-03-16T16:38:12Z

Problem

As described in SIMD-0488, we need a mechanism for adjusting stake delegations during epoch rewards payout.

Summary of Changes

Implement the new calculation logic. The commits can be read in order:

add the feature gate
adjust the calculation
tweak existing tests, add calculation test
add distribution test

Note: leaving this in draft until the SIMD goes through, there's still one more open question about respecting the minimum SOL delegation amount

codecov-commenter · 2026-03-16T17:07:58Z

Codecov Report

❌ Patch coverage is 98.57143% with 8 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.2%. Comparing base (0ef8fb3) to head (d8a8278).
⚠️ Report is 10 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff            @@
##           master   #11332    +/-   ##
========================================
  Coverage    83.1%    83.2%            
========================================
  Files         847      847            
  Lines      320546   321076   +530     
========================================
+ Hits       266680   267261   +581     
+ Misses      53866    53815    -51

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

joncinque · 2026-03-17T11:41:07Z

Although the SIMD isn't in yet, for the sake of moving things forward quickly, please take a look when you have a moment. The actual code changes are pretty small, so this should be an easy backport if we make that decision.

t-nelson · 2026-03-18T23:24:34Z

runtime/src/bank/partitioned_epoch_rewards/calculation.rs


 #[derive(Default)]
 struct RewardsAccumulator {
+    /// Accounts receiving voting commissions at the epoch boundary


please avoid unrelated changes. this should be a separate pr

The other comments "/// Number of stake rewards, used to figure out size of rewards vector" and "/// Accumulated reward lamports, only used for sanity checks" are very important, since it helps show what's actually important in all of these structs.

Adding comments to two out of three fields in a struct felt silly, so I added this one too. The other ones are important to this change though.

t-nelson · 2026-03-18T23:30:01Z

runtime/src/inflation_rewards/mod.rs

+            stake.delegation.stake = new_delegation;
+            // Deactivate stake if needed
+            if new_delegation == 0 {
+                stake.delegation.deactivation_epoch = rewarded_epoch;


since phone guys don't know what encapsulation is, can we have a comment here about this deactivation being immediate, unlike an authority-requested deactivation which is cast to the next epoch boundary

t-nelson · 2026-03-18T23:53:08Z

runtime/src/inflation_rewards/mod.rs

    }
 }

+#[allow(clippy::too_many_arguments)]


i think it'd be worthwhile to do a cleanup preliminary pr where we just pass the whole stake account down

I don't think it would really help to be honest -- it's either we pass the whole account and rent sysvar, or we pass the current and minimum lamport amount. I found the lamport amounts to be much simpler

ok but you literally added the "fuck you guys i'm lazy" attribute

I was answering the suggestion -- if you really want to refactor this function, I can do it

#11681 if you want it

t-nelson · 2026-03-19T00:06:37Z

runtime/src/inflation_rewards/mod.rs

-    })
+        ))
+    } else {
+        // No rewards to pay out, but still might need to modify delegation


seems like it'd be more maintainable to combine these blocks. for the sake of this operation, we don't care whether the account has earned rewards

I implemented your suggestion, but it makes the code much harder to understand (and thus, maintain) in my opinion. I've pushed it up, let me know what you think

joncinque · 2026-03-19T12:02:33Z

Noting that this will be rebased on #9380 to use the feature gate from there

2501babe · 2026-03-19T12:04:33Z

runtime/src/inflation_rewards/mod.rs

+        let new_delegation = std::cmp::min(
+            stake.delegation.stake.saturating_add(staker_rewards),
+            current_lamports
+                .saturating_add(staker_rewards)
+                .saturating_sub(minimum_lamports),
+        );


this looks correct to me

#### Problem As mentioned during this [SIMD-0392 amendment](solana-foundation/solana-improvement-documents#488 (comment)), block metadata should carry the information of when a stake account is deactivting. #### Summary of changes Add a new `DeactivatingStake` variant of `RewardType`, to be used in anza-xyz/agave#11332.

* reward-info: Add variant for deactivating stake #### Problem As mentioned during this [SIMD-0392 amendment](solana-foundation/solana-improvement-documents#488 (comment)), block metadata should carry the information of when a stake account is deactivting. #### Summary of changes Add a new `DeactivatingStake` variant of `RewardType`, to be used in anza-xyz/agave#11332. * Change wording

mergify · 2026-03-24T23:44:41Z

If this PR represents a change to the public RPC API:

Make sure it includes a complementary update to rpc-client/ (example)
Open a follow-up PR to update the JavaScript client @solana/kit (example)

Thank you for keeping the RPC clients in sync with the server API @joncinque.

joncinque · 2026-03-24T23:56:23Z

This now contains the new reward type of DeactivatedStake, to be used when a stake becomes inactive, so that all deactivations (include forced ones due to rent) can be tracked. It will still require #9380

t-nelson · 2026-03-31T16:08:17Z

accounts-db/src/stake_rewards.rs

        }
    }

+    pub fn new_with_pre_stake_account(


Suggested change

pub fn new_with_pre_stake_account(

#[cfg(feature = "dev-context-only-utils")]

pub fn new_with_pre_stake_account(

'cause it calls Pubkey::unique()

This is already gated with dev-context-only-utils on line 97

t-nelson · 2026-03-31T16:20:34Z

accounts-db/src/stake_rewards.rs

    let vote_state =
        vote_state::VoteStateV4::deserialize(vote_account.data(), voter_pubkey).unwrap();
    let credits_observed = vote_state.credits();
+    let last_epoch = vote_state.epoch_credits.last().map(|c| c.0).unwrap_or(0);


is the _or(0) the right thing to do here? seems like current_epoch - 1 is more correct, but this doesn't not work? just unclear whether it will encourage us to write more misleading tests

Sure why not, it can't hurt to pass in the epoch too. I was going with the simplest way to just create a deactivated stake account

t-nelson · 2026-03-31T16:21:50Z

rpc/src/rpc.rs

+                            && (reward_type == RewardType::Staking
+                                || reward_type == RewardType::DeactivatedStake))


in retrospect, we should have added a helper on RewardType

The thought did cross my mind 🙃

t-nelson · 2026-03-31T16:27:40Z

runtime/src/bank/partitioned_epoch_rewards/distribution.rs

            ))
            .map_err(|_| DistributionError::UnableToSetState)?;
+
+        let reward_type = if partitioned_stake_reward.stake.delegation.stake(


thanks for reminding me how worthlessly ambiguous the stake interface is 🫠

joncinque requested review from 2501babe, brooksprumo, igor56D and t-nelson March 17, 2026 11:39

t-nelson reviewed Mar 19, 2026

View reviewed changes

joncinque changed the title ~~runtime: Implement rent-adjusted stake delegations for SIMD-0488~~ runtime: Implement rent-adjusted stake delegations for amended SIMD-0392 (formerly SIMD-0488) Mar 19, 2026

joncinque mentioned this pull request Mar 19, 2026

SIMD-0392: Relax post-execution balance check #9380

Open

2501babe reviewed Mar 19, 2026

View reviewed changes

joncinque mentioned this pull request Mar 23, 2026

reward-info: Add variant for deactivated stake anza-xyz/solana-sdk#634

Merged

joncinque added 6 commits March 24, 2026 22:45

Add feature gate and assert during distribution

2175b89

Add lamport balances, adjust calculation

cd71a27

Add tests, adjust existing tests

4858056

Add distribution test

59aff48

Combine branches for maintainability?

c4ca302

Add new reward-info type

d4a007c

joncinque force-pushed the rad branch from 9b02385 to d4a007c Compare March 24, 2026 23:44

joncinque added 3 commits March 25, 2026 00:45

Remove comments to get this in quicker

66826a6

Add new reward type everywhere, pedantically

a84e8bb

Update all lockfiles

d8a8278

joncinque requested a review from t-nelson March 24, 2026 23:57

This was referenced Mar 25, 2026

chore(deps): bump solana-reward-info from 5.0.0 to 6.0.0 #11539

Closed

feat(storage-proto): bump reward-info to 6.1.0 #11647

Merged

t-nelson reviewed Mar 31, 2026

View reviewed changes

joncinque mentioned this pull request Mar 31, 2026

runtime: Bundle parameters in reward calculation functions #11681

Open

	pub fn new_with_pre_stake_account(
	#[cfg(feature = "dev-context-only-utils")]
	pub fn new_with_pre_stake_account(

		&& (reward_type == RewardType::Staking
		\|\| reward_type == RewardType::DeactivatedStake))

Conversation

joncinque commented Mar 16, 2026

Problem

Summary of Changes

Uh oh!

codecov-commenter commented Mar 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

joncinque commented Mar 17, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

joncinque commented Mar 19, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mergify bot commented Mar 24, 2026

Uh oh!

joncinque commented Mar 24, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

codecov-commenter commented Mar 16, 2026 •

edited

Loading