a great many fixes

evm/src/cpu/kernel/aggregator.rs

@@ -48,10 +48,11 @@ pub(crate) fn combined_kernel() -> Kernel {
         include_str!("asm/secp256k1/moddiv.asm"),
         include_str!("asm/secp256k1/lift_x.asm"),
         include_str!("asm/secp256k1/inverse_scalar.asm"),
-        include_str!("asm/sha2.asm"),
+        include_str!("asm/sha2/constants.asm"),
-        include_str!("asm/sha2_constants.asm"),
+        include_str!("asm/sha2/memory.asm"),
-        include_str!("asm/sha2_memory.asm"),
+        include_str!("asm/sha2/ops.asm"),
-        include_str!("asm/sha2_ops.asm"),
+        include_str!("asm/sha2/sha2.asm"),
+        include_str!("asm/sha2/write_length.asm"),
         include_str!("asm/ecrecover.asm"),
         include_str!("asm/rlp/encode.asm"),
         include_str!("asm/rlp/decode.asm"),

evm/src/cpu/kernel/asm/sha2/memory.asm

@@ -94,6 +94,9 @@
     %mload_kernel_general_u32
     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
 %endmacro
 
 // Store a single byte to kernel general memory.

evm/src/cpu/kernel/asm/sha2/sha2.asm

@@ -40,36 +40,36 @@ sha2_store_end:
     //JUMP
     %jump(sha2_pad)
 
-global test_sha2_read:
+//global test_sha2_read:
-    JUMPDEST
+//    JUMPDEST
-    // stack: retdest
+//    // stack: retdest
-    push 0
+//    push 0
-    // stack: 0, retdest
+//    // stack: 0, retdest
-    %mload_kernel_general
+//    %mload_kernel_general
-    // stack: counter=num_bytes, retdest
+//    // stack: counter=num_bytes, retdest
-test_sha2_read_loop:
+//test_sha2_read_loop:
-    JUMPDEST
+//    JUMPDEST
-    // stack: counter, retdest, [stack]
+//    // stack: counter, retdest, [stack]
-    dup1
+//    dup1
-    // stack: addr=counter, counter, retdest, [stack]
+//    // stack: addr=counter, counter, retdest, [stack]
-    %mload_kernel_general
+//    %mload_kernel_general
-    // stack: value, counter, retdest, [stack]
+//    // stack: value, counter, retdest, [stack]
-    swap2
+//    swap2
-    // stack: retdest, counter, value, [stack]
+//    // stack: retdest, counter, value, [stack]
-    swap1
+//    swap1
-    // stack: counter, retdest, value, [stack]
+//    // stack: counter, retdest, value, [stack]
-    %decrement
+//    %decrement
-    // stack: counter-1, retdest, value, [stack]
+//    // stack: counter-1, retdest, value, [stack]
-    dup1
+//    dup1
-    iszero
+//    iszero
-    %jumpi(test_sha2_read_end)
+//    %jumpi(test_sha2_read_end)
-    %jump(test_sha2_read_loop)
+//    %jump(test_sha2_read_loop)
-test_sha2_read_end:
+//test_sha2_read_end:
-    // stack: counter=0, retdest, [stack]
+//    // stack: counter=0, retdest, [stack]
-    JUMPDEST
+//    JUMPDEST
-    pop
+//    pop
-    // stack: retdest, [stack]
+//    // stack: retdest, [stack]
-    JUMP
+//    JUMP
 
 // Precodition: input is in memory, starting at 0 of kernel general segment, of the form
 //              num_bytes, x[0], x[1], ..., x[num_bytes - 1]
@@ -120,7 +120,10 @@ global sha2_pad:
     push 0
     %mstore_kernel_general
     // stack: retdest
-    JUMP
+    //JUMP
+    push 100
+    push 1
+    %jump(sha2_gen_message_schedule_from_block)
 
 // Precodition: stack contains address of one message block, followed by output address
 // Postcondition: 256 addresses starting at given output address, contain 32-bit chunks
@@ -321,11 +324,16 @@ sha2_gen_message_schedule_remaining_loop:
     // stack: counter, output_addr + 4, block[0], block[1], retdest
     %decrement
     // stack: counter - 1, output_addr + 4, block[0], block[1], retdest
+    dup1
     iszero
     %jumpi(sha2_gen_message_schedule_remaining_end)
     %jump(sha2_gen_message_schedule_remaining_loop)
 sha2_gen_message_schedule_remaining_end:
     JUMPDEST
+    // stack: counter=0, output_addr, block[0], block[1], retdest
+    %pop4
+    STOP
+    JUMP
 
-global sha2_gen_all_message_schedules:
+//global sha2_gen_all_message_schedules:
-    JUMPDEST
+//  JUMPDEST

evm/src/cpu/kernel/asm/sha2/write_length.asm

@@ -1,16 +1,20 @@
 %macro sha2_write_length
+    // stack: last_addr, length
+    swap1
     // stack: length, last_addr
     push 1
     push 8
     shl
 
     // stack: 1 << 8, length, last_addr
-    dup2
+    dup1
-    // stack: length, 1 << 8, length, last_addr
+    // stack: 1 << 8, 1 << 8, length, last_addr
-    mod
-    // stack: length % (1 << 8), length, last_addr
     dup3
-    // stack: last_addr, length % (1 << 8), length, last_addr
+    // stack: length, 1 << 8, 1 << 8, length, last_addr
+    mod
+    // stack: length % (1 << 8), 1 << 8, length, last_addr
+    dup3
+    // stack: last_addr, length % (1 << 8), 1 << 8, length, last_addr
     %mstore_kernel_general
     
     // stack: 1 << 8, length, last_addr

evm/src/cpu/kernel/tests/sha2.rs

@@ -40,7 +40,6 @@ fn test_sha2_store() -> Result<()> {
     let memory_after_storing = after_storing.memory;
     dbg!(memory_after_storing);
 
-
     // let load_initial_stack = vec![U256::from_str("0xdeadbeef").unwrap()];
     // let stack_after_loading = run(&kernel.code, test_sha2_read, load_initial_stack)?.stack;
     // dbg!(stack_after_loading);