diff --git a/evm/src/cpu/kernel/aggregator.rs b/evm/src/cpu/kernel/aggregator.rs
index 67a0051b..5fb6ad08 100644
--- a/evm/src/cpu/kernel/aggregator.rs
+++ b/evm/src/cpu/kernel/aggregator.rs
@@ -45,7 +45,6 @@ pub(crate) fn combined_kernel() -> Kernel {
         include_str!("asm/sha2/ops.asm"),
         include_str!("asm/sha2/store_pad.asm"),
         include_str!("asm/sha2/temp_words.asm"),
-        include_str!("asm/sha2/util.asm"),
         include_str!("asm/sha2/write_length.asm"),
         include_str!("asm/rlp/encode.asm"),
         include_str!("asm/rlp/decode.asm"),
diff --git a/evm/src/cpu/kernel/asm/sha2/compression.asm b/evm/src/cpu/kernel/asm/sha2/compression.asm
index 5325e7d0..eb9b73b8 100644
--- a/evm/src/cpu/kernel/asm/sha2/compression.asm
+++ b/evm/src/cpu/kernel/asm/sha2/compression.asm
@@ -1,3 +1,11 @@
+// We use memory starting at 320 * num_blocks + 2 (after the message schedule
+// space) as scratch space to store stack values.
+%macro scratch_space_addr_from_num_blocks
+    // stack: num_blocks
+    %mul_const(320)
+    %add_const(2)
+%endmacro
+
 global sha2_compression:
     // stack: message_schedule_addr, retdest
     PUSH 0
diff --git a/evm/src/cpu/kernel/asm/sha2/message_schedule.asm b/evm/src/cpu/kernel/asm/sha2/message_schedule.asm
index d3b5c6c1..78d98634 100644
--- a/evm/src/cpu/kernel/asm/sha2/message_schedule.asm
+++ b/evm/src/cpu/kernel/asm/sha2/message_schedule.asm
@@ -1,3 +1,10 @@
+// We put the message schedule in memory starting at 64 * num_blocks + 2.
+%macro message_schedule_addr_from_num_blocks
+    // stack: num_blocks
+    %mul_const(64)
+    %add_const(2)
+%endmacro
+
 // Precodition: stack contains address of one message block, followed by output address
 // Postcondition: 256 bytes starting at given output address contain the 64 32-bit chunks
 //                of message schedule (in four-byte increments)
@@ -32,7 +39,7 @@ gen_message_schedule_from_block_0_loop:
     // stack: block[0] >> 32, block[0], output_addr, counter, block[1], retdest
     SWAP1
     // stack: block[0], block[0] >> 32, output_addr, counter, block[1], retdest
-    %truncate_to_u32
+    %as_u32
     // stack: block[0] % (1 << 32), block[0] >> 32, output_addr, counter, block[1], retdest
     DUP3
     // stack: output_addr, block[0] % (1 << 32), block[0] >> 32, output_addr, counter, block[1], retdest
@@ -73,7 +80,7 @@ gen_message_schedule_from_block_1_loop:
     // stack: block[1] >> 32, block[1], output_addr, counter, block[0], retdest
     SWAP1
     // stack: block[1], block[1] >> 32, output_addr, counter, block[0], retdest
-    %truncate_to_u32
+    %as_u32
     // stack: block[1] % (1 << 32), block[1] >> 32, output_addr, counter, block[0], retdest
     DUP3
     // stack: output_addr, block[1] % (1 << 32), block[1] >> 32, output_addr, counter, block[0], retdest
diff --git a/evm/src/cpu/kernel/asm/sha2/ops.asm b/evm/src/cpu/kernel/asm/sha2/ops.asm
index f24455ef..7d8054ca 100644
--- a/evm/src/cpu/kernel/asm/sha2/ops.asm
+++ b/evm/src/cpu/kernel/asm/sha2/ops.asm
@@ -1,13 +1,3 @@
-// u32 addition (discarding 2^32 bit)
-%macro add_u32
-    // stack: x, y
-    ADD
-    // stack: x + y
-    %truncate_to_u32
-    // stack: (x + y) & u32::MAX
-%endmacro
-
-
 // 32-bit right rotation
 %macro rotr(rot)
     // stack: value
@@ -25,7 +15,7 @@
     // stack: 32 - rot, value, value >> rot
     SHL
     // stack: value << (32 - rot), value >> rot
-    %truncate_to_u32
+    %as_u32
     // stack: (value << (32 - rot)) % (1 << 32), value >> rot
     ADD
 %endmacro
diff --git a/evm/src/cpu/kernel/asm/sha2/store_pad.asm b/evm/src/cpu/kernel/asm/sha2/store_pad.asm
index ce935ec3..7594eb81 100644
--- a/evm/src/cpu/kernel/asm/sha2/store_pad.asm
+++ b/evm/src/cpu/kernel/asm/sha2/store_pad.asm
@@ -1,3 +1,6 @@
+global sha2:
+    %jump(sha2_store)
+
 global sha2_store:
     // stack: num_bytes, x[0], x[1], ..., x[num_bytes - 1], retdest
     DUP1
@@ -84,6 +87,3 @@ global sha2_pad:
     // stack: num_blocks, retdest
     %message_schedule_addr_from_num_blocks
     %jump(sha2_gen_all_message_schedules)
-
-global sha2:
-    %jump(sha2_store)
diff --git a/evm/src/cpu/kernel/asm/sha2/util.asm b/evm/src/cpu/kernel/asm/sha2/util.asm
deleted file mode 100644
index e090df0a..00000000
--- a/evm/src/cpu/kernel/asm/sha2/util.asm
+++ /dev/null
@@ -1,18 +0,0 @@
-// We put the message schedule in memory starting at 64 * num_blocks + 2.
-%macro message_schedule_addr_from_num_blocks
-    // stack: num_blocks
-    %mul_const(64)
-    %add_const(2)
-%endmacro
-
-// We use memory starting at 320 * num_blocks + 2 (after the message schedule
-// space) as scratch space to store stack values.
-%macro scratch_space_addr_from_num_blocks
-    // stack: num_blocks
-    %mul_const(320)
-    %add_const(2)
-%endmacro
-
-%macro truncate_to_u32
-    %and_const(0xFFFFFFFF)
-%endmacro
diff --git a/evm/src/cpu/kernel/asm/util/basic_macros.asm b/evm/src/cpu/kernel/asm/util/basic_macros.asm
index fb9e6d01..8ac92258 100644
--- a/evm/src/cpu/kernel/asm/util/basic_macros.asm
+++ b/evm/src/cpu/kernel/asm/util/basic_macros.asm
@@ -246,3 +246,16 @@
     %mod_const(2)
     ISZERO
 %endmacro
+
+%macro as_u32
+    %and_const(0xFFFFFFFF)
+%endmacro
+
+// u32 addition (discarding 2^32 bit)
+%macro add_u32
+    // stack: x, y
+    ADD
+    // stack: x + y
+    %as_u32
+    // stack: (x + y) & u32::MAX
+%endmacro
diff --git a/evm/src/cpu/kernel/tests/sha2.rs b/evm/src/cpu/kernel/tests/hash.rs
similarity index 63%
rename from evm/src/cpu/kernel/tests/sha2.rs
rename to evm/src/cpu/kernel/tests/hash.rs
index cb6d580a..de707e71 100644
--- a/evm/src/cpu/kernel/tests/sha2.rs
+++ b/evm/src/cpu/kernel/tests/hash.rs
@@ -8,21 +8,23 @@ use sha2::{Digest, Sha256};
 use crate::cpu::kernel::aggregator::combined_kernel;
 use crate::cpu::kernel::interpreter::run_with_kernel;
 
-#[test]
-fn test_sha2() -> Result<()> {
-    let kernel = combined_kernel();
-    let sha2 = kernel.global_labels["sha2"];
+/// Standard Sha2 implementation.
+fn sha2(input: Vec<u8>) -> U256 {
+    let mut hasher = Sha256::new();
+    hasher.update(input.clone());
+    U256::from(&hasher.finalize()[..])
+}
 
+fn test_hash(hash_fn_label: &str, standard_implementation: &dyn Fn(Vec<u8>) -> U256) -> Result<()> {
+    let kernel = combined_kernel();
     let mut rng = thread_rng();
 
     // Generate a random message, between 0 and 9999 bytes.
     let num_bytes = rng.gen_range(0..10000);
     let message: Vec<u8> = (0..num_bytes).map(|_| rng.gen()).collect();
 
-    // Hash the message using a standard Sha256 implementation.
-    let mut hasher = Sha256::new();
-    hasher.update(message.clone());
-    let expected = U256::from(&hasher.finalize()[..]);
+    // Hash the message using a standard implementation.
+    let expected = standard_implementation(message.clone());
 
     // Load the message onto the stack.
     let mut initial_stack = vec![U256::from(num_bytes)];
@@ -31,8 +33,9 @@ fn test_sha2() -> Result<()> {
     initial_stack.push(U256::from_str("0xdeadbeef").unwrap());
     initial_stack.reverse();
 
-    // Run the sha2 kernel code.
-    let result = run_with_kernel(&kernel, sha2, initial_stack)?;
+    // Run the kernel code.
+    let kernel_function = kernel.global_labels[hash_fn_label];
+    let result = run_with_kernel(&kernel, kernel_function, initial_stack)?;
     let actual = result.stack()[0];
 
     // Check that the result is correct.
@@ -40,3 +43,8 @@ fn test_sha2() -> Result<()> {
 
     Ok(())
 }
+
+#[test]
+fn test_sha2() -> Result<()> {
+    test_hash("sha2", &sha2)
+}
diff --git a/evm/src/cpu/kernel/tests/mod.rs b/evm/src/cpu/kernel/tests/mod.rs
index 36471dc4..45feb238 100644
--- a/evm/src/cpu/kernel/tests/mod.rs
+++ b/evm/src/cpu/kernel/tests/mod.rs
@@ -2,10 +2,10 @@ mod core;
 mod curve_ops;
 mod ecrecover;
 mod exp;
+mod hash;
 mod mpt;
 mod packing;
 mod rlp;
-mod sha2;
 mod transaction_parsing;
 
 use std::str::FromStr;