diff --git a/codegen/src/asm.rs b/codegen/src/asm.rs
index 73b713ca1..8ab558e43 100644
--- a/codegen/src/asm.rs
+++ b/codegen/src/asm.rs
@@ -5,6 +5,8 @@
 use crate::{Buffer, Error, Result};
 use opcodes::{for_each_cancun_operator, Cancun as OpCode, OpCode as _};
 
+const MAX_STACK_SIZE: u16 = 1024;
+
 /// Low level assembler implementation for EVM.
 #[derive(Default, Clone, Debug)]
 pub struct Assembler {
@@ -18,8 +20,8 @@ pub struct Assembler {
     gas: u128,
     /// Memory pointer for byte offset.
     pub mp: usize,
-    /// Stack pointer, maximum 1024 items.
-    pub sp: u8,
+    /// Stack pointer, maximum `MAX_STACK_SIZE` items.
+    pub sp: u16,
 }
 
 impl Assembler {
@@ -41,7 +43,7 @@ impl Assembler {
     }
 
     /// Increment stack pointer
-    pub fn increment_sp(&mut self, items: u8) -> Result<()> {
+    pub fn increment_sp(&mut self, items: u16) -> Result<()> {
         if items == 0 {
             return Ok(());
         }
@@ -51,18 +53,20 @@ impl Assembler {
             self.sp,
             self.sp + items
         );
-        self.sp += items;
+        self.sp = self
+            .sp
+            .checked_add(items)
+            .ok_or(Error::StackOverflow(self.sp, items))?;
 
-        // TODO: fix this limitation: should be 1024. (#127)
-        if self.sp > 254 {
-            return Err(Error::StackOverflow(self.sp));
+        if self.sp > MAX_STACK_SIZE {
+            return Err(Error::StackOverflow(self.sp, items));
         }
 
         Ok(())
     }
 
     /// Decrement stack pointer
-    pub fn decrement_sp(&mut self, items: u8) -> Result<()> {
+    pub fn decrement_sp(&mut self, items: u16) -> Result<()> {
         if items == 0 {
             return Ok(());
         }
@@ -118,10 +122,10 @@ impl Assembler {
     /// the stack usages.
     pub fn emit_op(&mut self, opcode: OpCode) -> Result<()> {
         tracing::trace!("emit opcode: {:?}", opcode);
-        self.decrement_sp(opcode.stack_in() as u8)?;
+        self.decrement_sp(opcode.stack_in())?;
         self.emit(opcode.into());
         self.increment_gas(opcode.gas().into());
-        self.increment_sp(opcode.stack_out() as u8)?;
+        self.increment_sp(opcode.stack_out())?;
 
         Ok(())
     }
diff --git a/codegen/src/codegen/function.rs b/codegen/src/codegen/function.rs
index 380983fab..3b5596adb 100644
--- a/codegen/src/codegen/function.rs
+++ b/codegen/src/codegen/function.rs
@@ -137,7 +137,7 @@ impl Function {
     /// Finish code generation.
     pub fn finish(self, jump_table: &mut JumpTable, pc: u16) -> Result<Buffer> {
         let sp = self.masm.sp();
-        if !self.is_main && self.abi.is_none() && self.masm.sp() != self.ty.results().len() as u8 {
+        if !self.is_main && self.abi.is_none() && self.masm.sp() != self.ty.results().len() as u16 {
             return Err(Error::StackNotBalanced(sp));
         }
 
diff --git a/codegen/src/control.rs b/codegen/src/control.rs
index 418c829f4..61d342431 100644
--- a/codegen/src/control.rs
+++ b/codegen/src/control.rs
@@ -38,7 +38,7 @@ pub struct ControlStackFrame {
     result: BlockType,
 
     /// Original stack pointer.
-    pub original_sp: u8,
+    pub original_sp: u16,
 }
 
 impl ControlStackFrame {
@@ -46,7 +46,7 @@ impl ControlStackFrame {
     pub fn new(
         ty: ControlStackFrameType,
         original_pc_offset: u16,
-        original_sp: u8,
+        original_sp: u16,
         result: BlockType,
     ) -> Self {
         Self {
diff --git a/codegen/src/masm/mod.rs b/codegen/src/masm/mod.rs
index b78b57a55..0978a3ef1 100644
--- a/codegen/src/masm/mod.rs
+++ b/codegen/src/masm/mod.rs
@@ -139,7 +139,7 @@ impl MacroAssembler {
             30 => self.asm._push30(),
             31 => self.asm._push31(),
             32 => self.asm._push32(),
-            _ => return Err(Error::StackIndexOutOfRange(len as u8)),
+            _ => return Err(Error::StackIndexOutOfRange(len as u16)),
         }?;
 
         self.asm.emitn(bytes);
@@ -155,12 +155,12 @@ impl MacroAssembler {
     }
 
     /// Get the stack pointer.
-    pub fn sp(&self) -> u8 {
+    pub fn sp(&self) -> u16 {
         self.asm.sp
     }
 
     /// Swap memory by target index.
-    pub fn swap(&mut self, index: u8) -> Result<()> {
+    pub fn swap(&mut self, index: u16) -> Result<()> {
         tracing::trace!("swap index: {}", index);
         match index {
             0 => Ok(()),
@@ -185,7 +185,7 @@ impl MacroAssembler {
     }
 
     /// Duplicate stack item by target index.
-    pub fn dup(&mut self, index: u8) -> Result<()> {
+    pub fn dup(&mut self, index: u16) -> Result<()> {
         tracing::trace!("dup index: {}", index);
         match index {
             0 => Ok(()),
@@ -212,7 +212,7 @@ impl MacroAssembler {
     /// Shift the program counter to the bottom or the top of the
     /// parameters. This is used by the callee function for jumping
     /// back to the caller function.
-    pub fn shift_stack(&mut self, count: u8, from_top: bool) -> Result<()> {
+    pub fn shift_stack(&mut self, count: u16, from_top: bool) -> Result<()> {
         let mut swaps = 0;
 
         if from_top {
diff --git a/codegen/src/masm/ret.rs b/codegen/src/masm/ret.rs
index 84ef44ab2..ec5b65a04 100644
--- a/codegen/src/masm/ret.rs
+++ b/codegen/src/masm/ret.rs
@@ -31,7 +31,7 @@ impl MacroAssembler {
 
     /// Handle the return of a call.
     pub fn call_return(&mut self, results: &[ValType]) -> Result<()> {
-        let len = results.len() as u8;
+        let len = results.len() as u16;
 
         tracing::trace!("cleaning frame stack, target: {}", len + 1);
         // TODO: clean stacks via the count of nested control stacks.
diff --git a/codegen/src/result.rs b/codegen/src/result.rs
index 1889e1bbd..12816f3b4 100644
--- a/codegen/src/result.rs
+++ b/codegen/src/result.rs
@@ -86,16 +86,16 @@ pub enum Error {
     SelectorNotFound,
     /// Failed to index data on stack.
     #[error("Stack index is out of range {0}, max is 255 (0x400)")]
-    StackIndexOutOfRange(u8),
+    StackIndexOutOfRange(u16),
     /// Failed to increment stack pointer.
-    #[error("Stack overflow, max is 1024 stack items, got {0}")]
-    StackOverflow(u8),
+    #[error("Stack overflow, max is 1024 stack items, but add {1} to {0}")]
+    StackOverflow(u16, u16),
     /// Failed to decrement stack pointer.
     #[error("Stack underflow, current stack items {0}, expect at least {1}")]
-    StackUnderflow(u8, u8),
+    StackUnderflow(u16, u16),
     /// Failed to pop stack.
     #[error("Stack not balanced, current stack items {0}")]
-    StackNotBalanced(u8),
+    StackNotBalanced(u16),
     /// Failed to queue host functions.
     #[error("Unsupported host function {0:?}")]
     UnsupportedHostFunc(crate::wasm::HostFunc),
diff --git a/codegen/src/visitor/call.rs b/codegen/src/visitor/call.rs
index c61c93df0..cdf9066af 100644
--- a/codegen/src/visitor/call.rs
+++ b/codegen/src/visitor/call.rs
@@ -89,7 +89,7 @@ impl Function {
 
         // Adjust the stack pointer for the results.
         self.masm._jumpdest()?;
-        self.masm.increment_sp(*results as u8)?;
+        self.masm.increment_sp(*results as u16)?;
         Ok(())
     }