Fix ed25519 builtin program handling

[jstarry]

Problem

Master and v1.9 branches are incompatible with v1.8.

The ed25519 program was originally added as a "builtin" program which means that ed25519 instructions would be loaded by the native loader. However, on master and v1.9, ed25519 was changed to be a "precompile" program to prevent it from being invoked by on-chain programs (it always returns success so this would be confusing).

On devnet/testnet/mainnet-beta, the ed25519 program has already been activated as a builtin, and so both precompile programs, secp256k1 and ed25519, are still invokable because the prevent_calling_precompiles_as_programs feature wasn't back ported to v1.8 and so hasn't been activated yet. In order to transition those clusters to v1.9, the precompiles need to be handled as builtins until the prevent_calling_precompiles_as_programs feature is activated.

Summary of Changes

At a high level, this PR just ensures that ed25519 is treated the same way as secp256k1. When the prevent_calling_precompiles_as_programs feature is activated, both programs will be transitioned from builtins to precompiles.

• Re-add ed25519 to the list of builtin programs. Since the program is already activated on all clusters, it can be added directly to the genesis builtin list.
• Ensure that ed25519 is removed as a builtin when the prevent_calling_precompiles_as_programs feature is activated
• Ensure that ed25519 is added as a precompile when prevent_calling_precompiles_as_programs feature is activated
• Add integration tests and the ability to deactivate features in program tests

Fixes #23180

this regression is introduced at #19918 and #19930

[rmekdma]

Is this typo?

[t-nelson]

Are you a bot? Elaborate

[ryoqun]

hehe, i'm not familiar with those tokio based integration tests. but it seems these are nice. :)

also, it's pretty fast compared to the dreadful local-cluster thing.. ;)

2842-running 3 tests
2843-test test_failure ... ok
2844-test test_success ... ok
2845:test test_success_call_builtin_program ... ok
2846-
2847-test result: ok. 3 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.30s

so, this test is checking ed25519 instruction still success as long as it's directly embedded into tx (i.e. not via CPI) regardless prevent_calling_precompiles_as_programs is activated or not.

[ryoqun]

lgtm

[behzadnouri]

Apparently this causing panic when starting a validator off master on testnet:

thread 'main' panicked at 'assertion failed: !self.freeze_started()', runtime/src/bank.rs:5383:9
stack backtrace:
   0: rust_begin_unwind
             at /rustc/6db0a0e9a4a2f55b1a85954e114ada0b45c32e45/library/std/src/panicking.rs:498:5
   1: core::panicking::panic_fmt
             at /rustc/6db0a0e9a4a2f55b1a85954e114ada0b45c32e45/library/core/src/panicking.rs:107:14
   2: core::panicking::panic
             at /rustc/6db0a0e9a4a2f55b1a85954e114ada0b45c32e45/library/core/src/panicking.rs:48:5
   3: solana_runtime::bank::Bank::add_builtin_account
   4: solana_runtime::bank::Bank::add_builtin
   5: solana_runtime::bank::Bank::finish_init
   6: solana_runtime::bank::Bank::new_from_fields
   7: solana_runtime::serde_snapshot::bank_from_streams
   8: solana_runtime::snapshot_utils::rebuild_bank_from_snapshots
   9: solana_runtime::snapshot_utils::bank_from_latest_snapshot_archives
  10: solana_ledger::bank_forks_utils::load
  11: solana_core::validator::new_banks_from_ledger
  12: solana_core::validator::Validator::new
  13: solana_validator::main

[t-nelson]

Apparently this causing panic when starting a validator off master on testnet:

thread 'main' panicked at 'assertion failed: !self.freeze_started()', runtime/src/bank.rs:5383:9
stack backtrace:
   0: rust_begin_unwind
             at /rustc/6db0a0e9a4a2f55b1a85954e114ada0b45c32e45/library/std/src/panicking.rs:498:5
   1: core::panicking::panic_fmt
             at /rustc/6db0a0e9a4a2f55b1a85954e114ada0b45c32e45/library/core/src/panicking.rs:107:14
   2: core::panicking::panic
             at /rustc/6db0a0e9a4a2f55b1a85954e114ada0b45c32e45/library/core/src/panicking.rs:48:5
   3: solana_runtime::bank::Bank::add_builtin_account
   4: solana_runtime::bank::Bank::add_builtin
   5: solana_runtime::bank::Bank::finish_init
   6: solana_runtime::bank::Bank::new_from_fields
   7: solana_runtime::serde_snapshot::bank_from_streams
   8: solana_runtime::snapshot_utils::rebuild_bank_from_snapshots
   9: solana_runtime::snapshot_utils::bank_from_latest_snapshot_archives
  10: solana_ledger::bank_forks_utils::load
  11: solana_core::validator::new_banks_from_ledger
  12: solana_core::validator::Validator::new
  13: solana_validator::main

Yep! Fix is here #23233

[github-actions]

This PR has been automatically locked since there has not been any activity in past 14 days after it was merged.