From b566dbd73555a4b36272d3b624785b8f831456b0 Mon Sep 17 00:00:00 2001
From: Nicholas Ward <npward@berkeley.edu>
Date: Mon, 8 May 2023 15:30:01 -0700
Subject: [PATCH 1/7] refactor memory/core.asm to make code more reusable

---
 evm/src/cpu/kernel/asm/memory/core.asm | 413 +++++++++++++++----------
 1 file changed, 253 insertions(+), 160 deletions(-)

diff --git a/evm/src/cpu/kernel/asm/memory/core.asm b/evm/src/cpu/kernel/asm/memory/core.asm
index 24f35ada..beb6b420 100644
--- a/evm/src/cpu/kernel/asm/memory/core.asm
+++ b/evm/src/cpu/kernel/asm/memory/core.asm
@@ -1,3 +1,223 @@
+// Load a big-endian u32, consisting of 4 bytes (c_3, c_2, c_1, c_0).
+%macro mload_u32
+    // stack: context, segment, offset
+    DUP3
+    DUP3
+    DUP3
+    MLOAD_GENERAL
+    // stack: c_3, context, segment, offset
+    %shl_const(8)
+    // stack: c_3 << 8, context, segment, offset
+    DUP4
+    %increment
+    DUP4
+    DUP4
+    MLOAD_GENERAL
+    ADD // OR
+    // stack: (c_3 << 8) | c_2, context, segment, offset
+    %shl_const(8)
+    // stack: ((c_3 << 8) | c_2) << 8, context, segment, offset
+    DUP4
+    %add_const(2)
+    DUP4
+    DUP4
+    MLOAD_GENERAL
+    ADD // OR
+    // stack: (((c_3 << 8) | c_2) << 8) | c_1, context, segment, offset
+    %shl_const(8)
+    // stack: ((((c_3 << 8) | c_2) << 8) | c_1) << 8, context, segment, offset
+    SWAP3
+    %add_const(3)
+    SWAP2
+    SWAP1
+    MLOAD_GENERAL
+    ADD // OR
+    // stack: (((((c_3 << 8) | c_2) << 8) | c_1) << 8) | c_0
+%endmacro
+
+// Load a little-endian u32, consisting of 4 bytes (c_0, c_1, c_2, c_3).
+%macro mload_u32_LE
+    // stack: context, segment, offset
+    DUP3
+    DUP3
+    DUP3
+    %mload_u32
+    // stack: c0, context, segment, offset
+    DUP4
+    %increment
+    DUP4
+    DUP4
+    %mload_u32
+    %shl_const(8)
+    ADD
+    // stack: c0 | (c1 << 8), context, segment, offset
+    DUP4
+    %add_const(2)
+    DUP4
+    DUP4
+    %mload_u32
+    %shl_const(16)
+    ADD
+    // stack: c0 | (c1 << 8) | (c2 << 16), context, segment, offset
+    SWAP3
+    %add_const(3)
+    SWAP2
+    SWAP1
+    %mload_u32
+    %shl_const(24)
+    ADD // OR
+    // stack: c0 | (c1 << 8) | (c2 << 16) | (c3 << 24)
+%endmacro
+
+// Load a little-endian u64, consisting of 8 bytes (c_0, ..., c_7).
+%macro mload_u64_LE
+    // stack: context, segment, offset
+    DUP3
+    DUP3
+    DUP3
+    %mload_u32_LE
+    // stack: lo, context, segment, offset
+    SWAP3
+    %add_const(4)
+    SWAP2
+    SWAP1
+    %mload_u32_LE
+    // stack: hi, lo
+    %shl_const(32)
+    // stack: hi << 32, lo
+    ADD // OR
+    // stack: (hi << 32) | lo
+%endmacro
+
+// Load a u256 (big-endian).
+%macro mload_u256
+    // stack: context, segment, offset
+    DUP3
+    DUP3
+    DUP3
+    %mload_u32
+    // stack: c_7, context, segment, offset
+    %shl_const(32)
+    // stack: c7 << 32, context, segment, offset
+    DUP4
+    %add_const(4)
+    DUP4
+    DUP4
+    %mload_u32
+    ADD // OR
+    // stack: (c_7 << 32) | c_6, context, segment, offset
+    %shl_const(32)
+    // stack: ((c_7 << 32) | c_6) << 32, context, segment, offset
+    DUP4
+    %add_const(8)
+    DUP4
+    DUP4
+    %mload_u32
+    ADD // OR
+    // stack: (c_7 << 64) | (c_6 << 32) | c_5, context, segment, offset
+    %shl_const(32)
+    // stack: ((c_7 << 64) | (c_6 << 32) | c_5) << 32, context, segment, offset
+    DUP4
+    %add_const(12)
+    DUP4
+    DUP4
+    %mload_u32
+    ADD // OR
+    // stack: (c_7 << 96) | (c_6 << 64) | (c_5 << 32) | c_4, context, segment, offset
+    %shl_const(32)
+    // stack: ((c_7 << 96) | (c_6 << 64) | (c_5 << 32) | c_4) << 32, context, segment, offset
+    DUP4
+    %add_const(16)
+    DUP4
+    DUP4
+    %mload_u32
+    ADD // OR
+    // stack: (c_7 << 128) | (c_6 << 96) | (c_5 << 64) | (c_4 << 32) | c_3, context, segment, offset
+    %shl_const(32)
+    // stack: ((c_7 << 128) | (c_6 << 96) | (c_5 << 64) | (c_4 << 32) | c_3) << 32, context, segment, offset
+    DUP4
+    %add_const(20)
+    DUP4
+    DUP4
+    %mload_u32
+    ADD // OR
+    // stack: (c_7 << 160) | (c_6 << 128) | (c_5 << 96) | (c_4 << 64) | (c_3 << 32) | c_2, context, segment, offset
+    %shl_const(32)
+    // stack: ((c_7 << 160) | (c_6 << 128) | (c_5 << 96) | (c_4 << 64) | (c_3 << 32) | c_2) << 32, context, segment, offset
+    DUP4
+    %add_const(24)
+    DUP4
+    DUP4
+    %mload_u32
+    ADD // OR
+    // stack: (c_7 << 192) | (c_6 << 160) | (c_5 << 128) | (c_4 << 96) | (c_3 << 64) | (c_2 << 32) | c_1, context, segment, offset
+    %shl_const(32)
+    // stack: ((c_7 << 192) | (c_6 << 160) | (c_5 << 128) | (c_4 << 96) | (c_3 << 64) | (c_2 << 32) | c_1) << 32, context, segment, offset
+    SWAP3
+    %add_const(28)
+    SWAP2
+    SWAP1
+    %mload_u32
+    ADD // OR
+    // stack: (c_7 << 224) | (c_6 << 192) | (c_5 << 160) | (c_4 << 128) | (c_3 << 96) | (c_2 << 64) | (c_1 << 32) | c_0
+%endmacro
+
+// Store a big-endian u32, consisting of 4 bytes (c_3, c_2, c_1, c_0).
+%macro mstore_u32
+    // stack: context, segment, offset, value
+    SWAP3
+    // stack: value, segment, offset, context
+    DUP1
+    // stack: value, value, segment, offset, context
+    %and_const(0xff)
+    // stack: c_0 = value % (1 << 8), value, segment, offset, context
+    SWAP1
+    // stack: value, c_0, segment, offset, context
+    %shr_const(8)
+    // stack: value >> 8, c_0, segment, offset, context
+    DUP1
+    // stack: value >> 8, value >> 8, c_0, segment, offset, context
+    %and_const(0xff)
+    // stack: c_1 = (value >> 8) % (1 << 8), value >> 8, c_0, segment, offset, context
+    SWAP1
+    // stack: value >> 8, c_1, c_0, segment, offset, context
+    %shr_const(8)
+    // stack: value >> 16, c_1, c_0, segment, offset, context
+    DUP1
+    // stack: value >> 16, value >> 16, c_1, c_0, segment, offset, context
+    %and_const(0xff)
+    // stack: c_2 = (value >> 16) % (1 << 8), value >> 16, c_1, c_0, segment, offset, context
+    SWAP1
+    // stack: value >> 16, c_2, c_1, c_0, segment, offset, context
+    %shr_const(8)
+    // stack: value >> 24, c_2, c_1, c_0, segment, offset, context
+    %and_const(0xff)
+    // stack: c_3 = (value >> 24) % (1 << 8), c_2, c_1, c_0, segment, offset, context
+    DUP6
+    DUP6
+    DUP8
+    // stack: context, segment, offset, c_3, c_2, c_1, c_0, segment, offset, context
+    MSTORE_GENERAL
+    // stack: c_2, c_1, c_0, segment, offset, context
+    DUP5
+    %increment
+    DUP5
+    DUP7
+    MSTORE_GENERAL
+    // stack: c_1, c_0, segment, offset, context
+    DUP4
+    %add_const(2)
+    DUP4
+    DUP6
+    MSTORE_GENERAL
+    // stack: c_0, segment, offset, context
+    SWAP2
+    %add_const(3)
+    SWAP2
+    SWAP3
+    MSTORE_GENERAL
+%endmacro
+
 // Load a value from the given segment of the current context's memory space.
 // Note that main memory values are one byte each, but in general memory values
 // can be 256 bits. This macro deals with a single address (unlike MLOAD), so
@@ -33,6 +253,13 @@
     // stack: value
 %endmacro
 
+// Load a single byte from user general memory.
+%macro mload_current_memory
+    // stack: offset
+    %mload_current(@SEGMENT_MEMORY)
+    // stack: value
+%endmacro
+
 // Load a single value from the given segment of kernel (context 0) memory.
 %macro mload_kernel(segment)
     // stack: offset
@@ -71,186 +298,52 @@
 // Load from the kernel a big-endian u32, consisting of 4 bytes (c_3, c_2, c_1, c_0)
 %macro mload_kernel_u32(segment)
     // stack: offset
-    DUP1
-    %mload_kernel($segment)
-    // stack: c_3, offset
-    %shl_const(8)
-    // stack: c_3 << 8, offset
-    DUP2
-    %increment
-    %mload_kernel($segment)
-    ADD // OR
-    // stack: (c_3 << 8) | c_2, offset
-    %shl_const(8)
-    // stack: ((c_3 << 8) | c_2) << 8, offset
-    DUP2
-    %add_const(2)
-    %mload_kernel($segment)
-    ADD // OR
-    // stack: (((c_3 << 8) | c_2) << 8) | c_1, offset
-    %shl_const(8)
-    // stack: ((((c_3 << 8) | c_2) << 8) | c_1) << 8, offset
-    SWAP1
-    %add_const(3)
-    %mload_kernel($segment)
-    ADD // OR
-    // stack: (((((c_3 << 8) | c_2) << 8) | c_1) << 8) | c_0
+    PUSH $segment
+    // stack: segment, offset
+    PUSH 0 // kernel has context 0
+    // stack: context, segment, offset
+    %mload_u32
 %endmacro
 
 // Load from the kernel a little-endian u32, consisting of 4 bytes (c_0, c_1, c_2, c_3).
 %macro mload_kernel_u32_LE(segment)
     // stack: offset
-    DUP1
-    %mload_kernel($segment)
-    // stack: c0                         , offset
-    DUP2
-    %increment
-    %mload_kernel($segment)
-    %shl_const(8)
-    ADD
-    // stack: c0 | (c1 << 8)             , offset
-    DUP2
-    %add_const(2)
-    %mload_kernel($segment)
-    %shl_const(16)
-    ADD
-    // stack: c0 | (c1 << 8) | (c2 << 16), offset
-    SWAP1
-    %add_const(3)
-    %mload_kernel($segment)
-    %shl_const(24)
-    ADD // OR
-    // stack: c0 | (c1 << 8) | (c2 << 16) | (c3 << 24)
+    PUSH $segment
+    // stack: segment, offset
+    PUSH 0 // kernel has context 0
+    // stack: context, segment, offset
+    %mload_u32_LE
 %endmacro
 
-// Load from the kernel a little-endian u64, consisting of 8 bytes
-// (c_0, c_1, c_2, c_3, c_4, c_5, c_6, c_7).
+// Load from the kernel a little-endian u64, consisting of 8 bytes (c_0, ..., c_7).
 %macro mload_kernel_u64_LE(segment)
     // stack: offset
-    DUP1
-    %mload_kernel_u32_LE($segment)
-    // stack: lo, offset
-    SWAP1
-    // stack: offset, lo
-    %add_const(4)
-    %mload_kernel_u32_LE($segment)
-    // stack: hi, lo
-    %shl_const(32)
-    // stack: hi << 32, lo
-    ADD // OR
-    // stack: (hi << 32) | lo
+    PUSH $segment
+    // stack: segment, offset
+    PUSH 0 // kernel has context 0
+    // stack: context, segment, offset
+    %mload_u64_LE
 %endmacro
 
 // Load a u256 (big-endian) from the kernel.
 %macro mload_kernel_u256(segment)
     // stack: offset
-    DUP1
-    %mload_kernel_u32($segment)
-    // stack: c_7, offset
-    %shl_const(32)
-    // stack: c7 << 32, offset
-    DUP2
-    %add_const(4)
-    %mload_kernel_u32($segment)
-    ADD // OR
-    // stack: (c_7 << 32) | c_6, offset
-    %shl_const(32)
-    // stack: ((c_7 << 32) | c_6) << 32, offset
-    DUP2
-    %add_const(8)
-    %mload_kernel_u32($segment)
-    ADD // OR
-    // stack: (c_7 << 64) | (c_6 << 32) | c_5, offset
-    %shl_const(32)
-    // stack: ((c_7 << 64) | (c_6 << 32) | c_5) << 32, offset
-    DUP2
-    %add_const(12)
-    %mload_kernel_u32($segment)
-    ADD // OR
-    // stack: (c_7 << 96) | (c_6 << 64) | (c_5 << 32) | c_4, offset
-    %shl_const(32)
-    // stack: ((c_7 << 96) | (c_6 << 64) | (c_5 << 32) | c_4) << 32, offset
-    DUP2
-    %add_const(16)
-    %mload_kernel_u32($segment)
-    ADD // OR
-    // stack: (c_7 << 128) | (c_6 << 96) | (c_5 << 64) | (c_4 << 32) | c_3, offset
-    %shl_const(32)
-    // stack: ((c_7 << 128) | (c_6 << 96) | (c_5 << 64) | (c_4 << 32) | c_3) << 32, offset
-    DUP2
-    %add_const(20)
-    %mload_kernel_u32($segment)
-    ADD // OR
-    // stack: (c_7 << 160) | (c_6 << 128) | (c_5 << 96) | (c_4 << 64) | (c_3 << 32) | c_2, offset
-    %shl_const(32)
-    // stack: ((c_7 << 160) | (c_6 << 128) | (c_5 << 96) | (c_4 << 64) | (c_3 << 32) | c_2) << 32, offset
-    DUP2
-    %add_const(24)
-    %mload_kernel_u32($segment)
-    ADD // OR
-    // stack: (c_7 << 192) | (c_6 << 160) | (c_5 << 128) | (c_4 << 96) | (c_3 << 64) | (c_2 << 32) | c_1, offset
-    %shl_const(32)
-    // stack: ((c_7 << 192) | (c_6 << 160) | (c_5 << 128) | (c_4 << 96) | (c_3 << 64) | (c_2 << 32) | c_1) << 32, offset
-    DUP2
-    %add_const(28)
-    %mload_kernel_u32($segment)
-    ADD // OR
-    // stack: (c_7 << 224) | (c_6 << 192) | (c_5 << 160) | (c_4 << 128) | (c_3 << 96) | (c_2 << 64) | (c_1 << 32) | c_0, offset
-    SWAP1
-    POP
-    // stack: (c_7 << 224) | (c_6 << 192) | (c_5 << 160) | (c_4 << 128) | (c_3 << 96) | (c_2 << 64) | (c_1 << 32) | c_0
+    PUSH $segment
+    // stack: segment, offset
+    PUSH 0 // kernel has context 0
+    // stack: context, segment, offset
+    %mload_u256
 %endmacro
 
 // Store a big-endian u32, consisting of 4 bytes (c_3, c_2, c_1, c_0),
 // to the kernel.
 %macro mstore_kernel_u32(segment)
     // stack: offset, value
-    SWAP1
-    // stack: value, offset
-    DUP1
-    // stack: value, value, offset
-    %and_const(0xff)
-    // stack: c_0 = value % (1 << 8), value, offset
-    SWAP1
-    // stack: value, c_0, offset
-    %shr_const(8)
-    // stack: value >> 8, c_0, offset
-    DUP1
-    // stack: value >> 8, value >> 8, c_0, offset
-    %and_const(0xff)
-    // stack: c_1 = (value >> 8) % (1 << 8), value >> 8, c_0, offset
-    SWAP1
-    // stack: value >> 8, c_1, c_0, offset
-    %shr_const(8)
-    // stack: value >> 16, c_1, c_0, offset
-    DUP1
-    // stack: value >> 16, value >> 16, c_1, c_0, offset
-    %and_const(0xff)
-    // stack: c_2 = (value >> 16) % (1 << 8), value >> 16, c_1, c_0, offset
-    SWAP1
-    // stack: value >> 16, c_2, c_1, c_0, offset
-    %shr_const(8)
-    // stack: value >> 24, c_2, c_1, c_0, offset
-    %and_const(0xff)
-    // stack: c_3 = (value >> 24) % (1 << 8), c_2, c_1, c_0, offset
-    DUP5
-    // stack: offset, c_3, c_2, c_1, c_0, offset
-    %mstore_kernel($segment)
-    // stack: c_2, c_1, c_0, offset
-    DUP4
-    // stack: offset, c_2, c_1, c_0, offset
-    %increment
-    %mstore_kernel($segment)
-    // stack: c_1, c_0, offset
-    DUP3
-    // stack: offset, c_1, c_0, offset
-    %add_const(2)
-    %mstore_kernel($segment)
-    // stack: c_0, offset
-    SWAP1
-    // stack: offset, c_0
-    %add_const(3)
-    %mstore_kernel($segment)
+    PUSH $segment
+    // stack: segment, offset, value
+    PUSH 0 // kernel has context 0
+    // stack: context, segment, offset, value
+    %mstore_u32
 %endmacro
 
 // Load a single byte from kernel code.

From 446c3b7140159cd662d2cc1d697ed769d6a294d6 Mon Sep 17 00:00:00 2001
From: Nicholas Ward <npward@berkeley.edu>
Date: Mon, 8 May 2023 16:02:43 -0700
Subject: [PATCH 2/7] fix

---
 evm/src/cpu/kernel/asm/memory/core.asm | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/evm/src/cpu/kernel/asm/memory/core.asm b/evm/src/cpu/kernel/asm/memory/core.asm
index beb6b420..38a7adaf 100644
--- a/evm/src/cpu/kernel/asm/memory/core.asm
+++ b/evm/src/cpu/kernel/asm/memory/core.asm
@@ -41,13 +41,13 @@
     DUP3
     DUP3
     DUP3
-    %mload_u32
+    MLOAD_GENERAL
     // stack: c0, context, segment, offset
     DUP4
     %increment
     DUP4
     DUP4
-    %mload_u32
+    MLOAD_GENERAL
     %shl_const(8)
     ADD
     // stack: c0 | (c1 << 8), context, segment, offset
@@ -55,7 +55,7 @@
     %add_const(2)
     DUP4
     DUP4
-    %mload_u32
+    MLOAD_GENERAL
     %shl_const(16)
     ADD
     // stack: c0 | (c1 << 8) | (c2 << 16), context, segment, offset
@@ -63,7 +63,7 @@
     %add_const(3)
     SWAP2
     SWAP1
-    %mload_u32
+    MLOAD_GENERAL
     %shl_const(24)
     ADD // OR
     // stack: c0 | (c1 << 8) | (c2 << 16) | (c3 << 24)

From d9694d95a6dbd3f92bb505a39b65659aae41a127 Mon Sep 17 00:00:00 2001
From: Nicholas Ward <npward@berkeley.edu>
Date: Mon, 8 May 2023 16:22:12 -0700
Subject: [PATCH 3/7] fix

---
 evm/src/cpu/kernel/asm/memory/core.asm | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/evm/src/cpu/kernel/asm/memory/core.asm b/evm/src/cpu/kernel/asm/memory/core.asm
index 38a7adaf..c1c6fe10 100644
--- a/evm/src/cpu/kernel/asm/memory/core.asm
+++ b/evm/src/cpu/kernel/asm/memory/core.asm
@@ -195,20 +195,20 @@
     // stack: c_3 = (value >> 24) % (1 << 8), c_2, c_1, c_0, segment, offset, context
     DUP6
     DUP6
-    DUP8
+    DUP9
     // stack: context, segment, offset, c_3, c_2, c_1, c_0, segment, offset, context
     MSTORE_GENERAL
     // stack: c_2, c_1, c_0, segment, offset, context
     DUP5
     %increment
     DUP5
-    DUP7
+    DUP8
     MSTORE_GENERAL
     // stack: c_1, c_0, segment, offset, context
     DUP4
     %add_const(2)
     DUP4
-    DUP6
+    DUP7
     MSTORE_GENERAL
     // stack: c_0, segment, offset, context
     SWAP2

From 98a75774d4756a39ae89d98ade1d1ace42e458f2 Mon Sep 17 00:00:00 2001
From: Nicholas Ward <npward@berkeley.edu>
Date: Wed, 10 May 2023 08:39:35 -0700
Subject: [PATCH 4/7] cleanup

---
 evm/src/cpu/kernel/asm/memory/core.asm | 7 -------
 1 file changed, 7 deletions(-)

diff --git a/evm/src/cpu/kernel/asm/memory/core.asm b/evm/src/cpu/kernel/asm/memory/core.asm
index c1c6fe10..22b6fadc 100644
--- a/evm/src/cpu/kernel/asm/memory/core.asm
+++ b/evm/src/cpu/kernel/asm/memory/core.asm
@@ -253,13 +253,6 @@
     // stack: value
 %endmacro
 
-// Load a single byte from user general memory.
-%macro mload_current_memory
-    // stack: offset
-    %mload_current(@SEGMENT_MEMORY)
-    // stack: value
-%endmacro
-
 // Load a single value from the given segment of kernel (context 0) memory.
 %macro mload_kernel(segment)
     // stack: offset

From 57bcb45104d9265067bd7d74bd19ac62d61fda10 Mon Sep 17 00:00:00 2001
From: Nicholas Ward <npward@berkeley.edu>
Date: Wed, 10 May 2023 09:12:30 -0700
Subject: [PATCH 5/7] use mstore_unpacking and mload_packing

---
 evm/src/cpu/kernel/asm/memory/core.asm | 162 ++-----------------------
 1 file changed, 10 insertions(+), 152 deletions(-)

diff --git a/evm/src/cpu/kernel/asm/memory/core.asm b/evm/src/cpu/kernel/asm/memory/core.asm
index 22b6fadc..723e9be7 100644
--- a/evm/src/cpu/kernel/asm/memory/core.asm
+++ b/evm/src/cpu/kernel/asm/memory/core.asm
@@ -1,38 +1,9 @@
 // Load a big-endian u32, consisting of 4 bytes (c_3, c_2, c_1, c_0).
 %macro mload_u32
     // stack: context, segment, offset
-    DUP3
-    DUP3
-    DUP3
-    MLOAD_GENERAL
-    // stack: c_3, context, segment, offset
-    %shl_const(8)
-    // stack: c_3 << 8, context, segment, offset
-    DUP4
-    %increment
-    DUP4
-    DUP4
-    MLOAD_GENERAL
-    ADD // OR
-    // stack: (c_3 << 8) | c_2, context, segment, offset
-    %shl_const(8)
-    // stack: ((c_3 << 8) | c_2) << 8, context, segment, offset
-    DUP4
-    %add_const(2)
-    DUP4
-    DUP4
-    MLOAD_GENERAL
-    ADD // OR
-    // stack: (((c_3 << 8) | c_2) << 8) | c_1, context, segment, offset
-    %shl_const(8)
-    // stack: ((((c_3 << 8) | c_2) << 8) | c_1) << 8, context, segment, offset
-    SWAP3
-    %add_const(3)
-    SWAP2
-    SWAP1
-    MLOAD_GENERAL
-    ADD // OR
-    // stack: (((((c_3 << 8) | c_2) << 8) | c_1) << 8) | c_0
+    %stack (addr: 3) -> (addr, 4, %%after)
+    %jump(mload_packing)
+%%after:
 %endmacro
 
 // Load a little-endian u32, consisting of 4 bytes (c_0, c_1, c_2, c_3).
@@ -89,133 +60,20 @@
     // stack: (hi << 32) | lo
 %endmacro
 
-// Load a u256 (big-endian).
+// Load a big-endian u256.
 %macro mload_u256
     // stack: context, segment, offset
-    DUP3
-    DUP3
-    DUP3
-    %mload_u32
-    // stack: c_7, context, segment, offset
-    %shl_const(32)
-    // stack: c7 << 32, context, segment, offset
-    DUP4
-    %add_const(4)
-    DUP4
-    DUP4
-    %mload_u32
-    ADD // OR
-    // stack: (c_7 << 32) | c_6, context, segment, offset
-    %shl_const(32)
-    // stack: ((c_7 << 32) | c_6) << 32, context, segment, offset
-    DUP4
-    %add_const(8)
-    DUP4
-    DUP4
-    %mload_u32
-    ADD // OR
-    // stack: (c_7 << 64) | (c_6 << 32) | c_5, context, segment, offset
-    %shl_const(32)
-    // stack: ((c_7 << 64) | (c_6 << 32) | c_5) << 32, context, segment, offset
-    DUP4
-    %add_const(12)
-    DUP4
-    DUP4
-    %mload_u32
-    ADD // OR
-    // stack: (c_7 << 96) | (c_6 << 64) | (c_5 << 32) | c_4, context, segment, offset
-    %shl_const(32)
-    // stack: ((c_7 << 96) | (c_6 << 64) | (c_5 << 32) | c_4) << 32, context, segment, offset
-    DUP4
-    %add_const(16)
-    DUP4
-    DUP4
-    %mload_u32
-    ADD // OR
-    // stack: (c_7 << 128) | (c_6 << 96) | (c_5 << 64) | (c_4 << 32) | c_3, context, segment, offset
-    %shl_const(32)
-    // stack: ((c_7 << 128) | (c_6 << 96) | (c_5 << 64) | (c_4 << 32) | c_3) << 32, context, segment, offset
-    DUP4
-    %add_const(20)
-    DUP4
-    DUP4
-    %mload_u32
-    ADD // OR
-    // stack: (c_7 << 160) | (c_6 << 128) | (c_5 << 96) | (c_4 << 64) | (c_3 << 32) | c_2, context, segment, offset
-    %shl_const(32)
-    // stack: ((c_7 << 160) | (c_6 << 128) | (c_5 << 96) | (c_4 << 64) | (c_3 << 32) | c_2) << 32, context, segment, offset
-    DUP4
-    %add_const(24)
-    DUP4
-    DUP4
-    %mload_u32
-    ADD // OR
-    // stack: (c_7 << 192) | (c_6 << 160) | (c_5 << 128) | (c_4 << 96) | (c_3 << 64) | (c_2 << 32) | c_1, context, segment, offset
-    %shl_const(32)
-    // stack: ((c_7 << 192) | (c_6 << 160) | (c_5 << 128) | (c_4 << 96) | (c_3 << 64) | (c_2 << 32) | c_1) << 32, context, segment, offset
-    SWAP3
-    %add_const(28)
-    SWAP2
-    SWAP1
-    %mload_u32
-    ADD // OR
-    // stack: (c_7 << 224) | (c_6 << 192) | (c_5 << 160) | (c_4 << 128) | (c_3 << 96) | (c_2 << 64) | (c_1 << 32) | c_0
+    %stack (addr: 3) -> (addr, 32, %%after)
+    %jump(mload_packing)
+%%after:
 %endmacro
 
 // Store a big-endian u32, consisting of 4 bytes (c_3, c_2, c_1, c_0).
 %macro mstore_u32
     // stack: context, segment, offset, value
-    SWAP3
-    // stack: value, segment, offset, context
-    DUP1
-    // stack: value, value, segment, offset, context
-    %and_const(0xff)
-    // stack: c_0 = value % (1 << 8), value, segment, offset, context
-    SWAP1
-    // stack: value, c_0, segment, offset, context
-    %shr_const(8)
-    // stack: value >> 8, c_0, segment, offset, context
-    DUP1
-    // stack: value >> 8, value >> 8, c_0, segment, offset, context
-    %and_const(0xff)
-    // stack: c_1 = (value >> 8) % (1 << 8), value >> 8, c_0, segment, offset, context
-    SWAP1
-    // stack: value >> 8, c_1, c_0, segment, offset, context
-    %shr_const(8)
-    // stack: value >> 16, c_1, c_0, segment, offset, context
-    DUP1
-    // stack: value >> 16, value >> 16, c_1, c_0, segment, offset, context
-    %and_const(0xff)
-    // stack: c_2 = (value >> 16) % (1 << 8), value >> 16, c_1, c_0, segment, offset, context
-    SWAP1
-    // stack: value >> 16, c_2, c_1, c_0, segment, offset, context
-    %shr_const(8)
-    // stack: value >> 24, c_2, c_1, c_0, segment, offset, context
-    %and_const(0xff)
-    // stack: c_3 = (value >> 24) % (1 << 8), c_2, c_1, c_0, segment, offset, context
-    DUP6
-    DUP6
-    DUP9
-    // stack: context, segment, offset, c_3, c_2, c_1, c_0, segment, offset, context
-    MSTORE_GENERAL
-    // stack: c_2, c_1, c_0, segment, offset, context
-    DUP5
-    %increment
-    DUP5
-    DUP8
-    MSTORE_GENERAL
-    // stack: c_1, c_0, segment, offset, context
-    DUP4
-    %add_const(2)
-    DUP4
-    DUP7
-    MSTORE_GENERAL
-    // stack: c_0, segment, offset, context
-    SWAP2
-    %add_const(3)
-    SWAP2
-    SWAP3
-    MSTORE_GENERAL
+    %stack (addr: 3, value) -> (addr, value, 32, %%after)
+    %jump(mstore_unpacking)
+%%after:
 %endmacro
 
 // Load a value from the given segment of the current context's memory space.

From 6e7fa6da027d822bcc71e311764dcfe175c2e654 Mon Sep 17 00:00:00 2001
From: Nicholas Ward <npward@berkeley.edu>
Date: Wed, 10 May 2023 09:21:52 -0700
Subject: [PATCH 6/7] fix

---
 evm/src/cpu/kernel/asm/memory/core.asm | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/evm/src/cpu/kernel/asm/memory/core.asm b/evm/src/cpu/kernel/asm/memory/core.asm
index 723e9be7..8f822b33 100644
--- a/evm/src/cpu/kernel/asm/memory/core.asm
+++ b/evm/src/cpu/kernel/asm/memory/core.asm
@@ -71,7 +71,7 @@
 // Store a big-endian u32, consisting of 4 bytes (c_3, c_2, c_1, c_0).
 %macro mstore_u32
     // stack: context, segment, offset, value
-    %stack (addr: 3, value) -> (addr, value, 32, %%after)
+    %stack (addr: 3, value) -> (addr, value, 4, %%after)
     %jump(mstore_unpacking)
 %%after:
 %endmacro

From 0d9e321647869eb24f7ec8fc59c63758140ee84b Mon Sep 17 00:00:00 2001
From: Nicholas Ward <npward@berkeley.edu>
Date: Wed, 10 May 2023 09:55:35 -0700
Subject: [PATCH 7/7] fix (mstore_unpacking returns offset)

---
 evm/src/cpu/kernel/asm/memory/core.asm | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/evm/src/cpu/kernel/asm/memory/core.asm b/evm/src/cpu/kernel/asm/memory/core.asm
index 8f822b33..0bcb53dc 100644
--- a/evm/src/cpu/kernel/asm/memory/core.asm
+++ b/evm/src/cpu/kernel/asm/memory/core.asm
@@ -74,6 +74,8 @@
     %stack (addr: 3, value) -> (addr, value, 4, %%after)
     %jump(mstore_unpacking)
 %%after:
+    // stack: offset
+    POP
 %endmacro
 
 // Load a value from the given segment of the current context's memory space.