From 427ebeab6726799bb93f7720f48aea27a7a2e964 Mon Sep 17 00:00:00 2001
From: Lucas Steuernagel <lucas.tnagel@gmail.com>
Date: Mon, 19 Jun 2023 16:27:31 -0300
Subject: [PATCH] Avoid repeated global strings

Signed-off-by: Lucas Steuernagel <lucas.tnagel@gmail.com>
---
 src/emit/binary.rs                            |  16 +-
 .../solidity/global_strings.sol               | 153 ++++++++++++++++++
 2 files changed, 167 insertions(+), 2 deletions(-)
 create mode 100644 tests/codegen_testcases/solidity/global_strings.sol

diff --git a/src/emit/binary.rs b/src/emit/binary.rs
index a6fc429313..991d10e39d 100644
--- a/src/emit/binary.rs
+++ b/src/emit/binary.rs
@@ -59,6 +59,7 @@ pub struct Binary<'a> {
     pub(crate) return_values: HashMap<ReturnCode, IntValue<'a>>,
     /// No initializer for vector_new
     pub(crate) vector_init_empty: PointerValue<'a>,
+    global_strings: RefCell<HashMap<Vec<u8>, PointerValue<'a>>>,
 }
 
 impl<'a> Binary<'a> {
@@ -322,6 +323,7 @@ impl<'a> Binary<'a> {
                 .i8_type()
                 .ptr_type(AddressSpace::default())
                 .const_null(),
+            global_strings: RefCell::new(HashMap::new()),
         }
     }
 
@@ -359,6 +361,11 @@ impl<'a> Binary<'a> {
         data: &[u8],
         constant: bool,
     ) -> PointerValue<'a> {
+        if let Some(emitted_string) = self.global_strings.borrow().get(data) {
+            if constant {
+                return *emitted_string;
+            }
+        }
         let ty = self.context.i8_type().array_type(data.len() as u32);
 
         let gv = self
@@ -372,9 +379,14 @@ impl<'a> Binary<'a> {
         if constant {
             gv.set_constant(true);
             gv.set_unnamed_addr(true);
+            let ptr_val = gv.as_pointer_value();
+            self.global_strings
+                .borrow_mut()
+                .insert(data.to_vec(), ptr_val);
+            ptr_val
+        } else {
+            gv.as_pointer_value()
         }
-
-        gv.as_pointer_value()
     }
 
     /// Wrapper for alloca. Ensures that the alloca is done on the first basic block.
diff --git a/tests/codegen_testcases/solidity/global_strings.sol b/tests/codegen_testcases/solidity/global_strings.sol
new file mode 100644
index 0000000000..7a4d4c248f
--- /dev/null
+++ b/tests/codegen_testcases/solidity/global_strings.sol
@@ -0,0 +1,153 @@
+// RUN: --target solana --emit llvm-ir
+// READ: c.ll
+
+
+// BEGIN-CHECK: "math overflow,\0A"
+// NOT-CHECK: "math overflow,\0A"
+
+import "solana";
+contract ServiceRegistrySolana {
+    struct Service {
+        address serviceOwner;
+        uint64 securityDeposit;
+        address multisig;
+        bytes32 configHash;
+        uint32 threshold;
+        uint32 maxNumAgentInstances;
+        uint32 numAgentInstances;
+        uint8 state;
+        uint32[] agentIds;
+        uint32[] slots;
+        uint64[] bonds;
+        address[] operators;
+        address[] agentInstances;
+        uint32[] agentIdForAgentInstances;
+    }
+
+    address public owner;
+    address public escrow;
+    string public baseURI;
+    uint32 public totalSupply;
+    string public constant CID_PREFIX = "f01701220";
+    uint64 public slashedFunds;
+    address public drainer;
+    string public constant VERSION = "1.0.0";
+    mapping(bytes32 => uint64) public mapOperatorAndServiceIdOperatorBalances;
+    mapping (address => address) public mapAgentInstanceOperators;
+    mapping (address => bool) public mapMultisigs;
+    Service public services;
+
+
+
+    constructor(address a, string memory _baseURI)
+    {
+        
+    }
+
+    function changeOwner(address newOwner) external {
+        
+    }
+
+    function changeDrainer(address newDrainer) external {
+        
+    }
+
+    function transfer(uint32 serviceId, address newServiceOwner) public {
+
+    }
+
+    function create(
+        address serviceOwner,
+        bytes32 configHash,
+        uint32[] memory agentIds,
+        uint32[] memory slots,
+        uint64[] memory bonds,
+        uint32 threshold
+    ) external returns (uint32 serviceId)
+    {
+        
+    }
+
+    function update(
+        bytes32 configHash,
+        uint32[] memory agentIds,
+        uint32[] memory slots,
+        uint64[] memory bonds,
+        uint32 threshold,
+        uint32 serviceId
+    ) external returns (bool success)
+    {
+    }
+
+    function activateRegistration(uint32 serviceId) external payable returns (bool success)
+    {
+
+    }
+
+    function registerAgents(
+        address operator,
+        uint32 serviceId,
+        address[] memory agentInstances,
+        uint32[] agentIds
+    ) external payable returns (bool success)
+    {
+        
+    }
+
+    function deploy(
+        uint32 serviceId,
+        address multisigImplementation,
+        bytes memory data
+    ) external returns (address multisig)
+    {
+        
+    }
+
+    function slash(address[] memory agentInstances, uint64[] memory amounts, uint32 serviceId) external
+        returns (bool success)
+    {
+        
+    }
+
+    function terminate(uint32 serviceId) external returns (bool success, uint64 refund)
+    {
+    }
+
+    function unbond(address operator, uint32 serviceId) external returns (bool success, uint64 refund) {
+
+    }
+
+    function getService(uint32 serviceId) external view returns (Service memory service) {
+        
+    }
+
+    function getAgentParams(uint32 serviceId) external view
+        returns (uint32 numAgentIds, uint32[] memory slots, uint64[] memory bonds)
+    {
+        
+    }
+
+    function getAgentInstances(uint32 serviceId) external view
+        returns (uint32 numAgentInstances, address[] memory agentInstances)
+    {
+
+    }
+
+    function getOperatorBalance(address operator, uint32 serviceId) external view returns (uint64 balance)
+    {
+    }
+
+    function changeMultisigPermission(address multisig, bool permission) external returns (bool success) {
+       
+    }
+
+    function drain() external returns (uint64 amount) {
+       
+    }
+
+    function ownerOf(uint32 id) public view returns (address addr) {
+        
+    }
+
+    function exists() public view {}
+}