Implement instructions for known-size bytes

See: #49

evm-instructions.ss

@@ -0,0 +1,109 @@
+(export #t)
+(import
+  :gerbil/gambit/bits :gerbil/gambit/bytes :gerbil/gambit/exact
+  :std/misc/number :std/sugar :std/misc/list :std/format
+  :clan/base :clan/number :clan/with-id :std/srfi/1
+  :clan/poo/object (only-in :clan/poo/mop Type)
+  :clan/crypto/secp256k1
+  ./assembly ./ethereum ./types ./evm-runtime)
+
+;; --------------------------------
+;; General purpose EVM instructions
+;; --------------------------------
+;; Extends instruction set.
+;; Examples include:
+;; - Storing bytes larger than EVM Word
+;; - Storing to memory with reference
+;; - Load from memory with reference
+;; NOTE: Reference is start-offset, length on stack.
+;;
+;; -------------
+;; Byte encoding
+;; -------------
+;; Instructions encode bytes as big-endian.
+;;
+;; For example, given the value: 0x00ff,
+;; After PUSH:  Stack will have "0x00" on top, followed by "0xff".
+;; After STORE: Memory location will have "0x00" at offset, "0xff" at offset+1.
+;;
+;; For larger bytes (larger than a EVM word),
+;; we partition them into smaller chunks,
+;; less than or equal to a EVM word.
+;; We use the notation: "part[n]", to denote the partition index.
+;; E.g. 65-bytes -> | 32-bytes | 32-bytes | 1-byte  |
+;;                  | part0  | part1  | part2 |
+
+
+;; ----------------------
+;; Instruction defintions
+;; ----------------------
+
+;; stack input: -
+;; stack output: part0 part1 ... partn
+(def (&push/any-size bytes (start 0) (len (u8vector-length bytes)))
+  (assert! (<= 1 len) (format "size: ~d not in range: [1,∞]")) ; TODO: provide a concrete upper bound
+  (def end (min len (+ start 32)))
+  (when (< start end)
+    (let ()
+      (def bytes<=32 (subu8vector bytes start end))
+      (def n-bytes (u8vector-length bytes<=32))
+      (&begin (&push/any-size bytes end) [&push-bytes bytes<=32])
+    )))
+
+;; Helper function - Make list of word-sizes
+;; E.g. (sizes/word-size<-size 65 32) -> [32 32 1]
+(def (sizes/word-size<-size size (word-size 32))
+  (def n-words (truncate (/ size word-size)))
+  (def words (make-list n-words word-size))
+  (def rem (modulo size word-size))
+  (if (eq? rem 0) words [words ... rem])
+  )
+
+;; NOTE: Uses brk@ for offset via &brk-cons, dependent on EVM memory layout.
+;; stack input:  part0 part1 ... partn
+;; stack output: -
+(def (&mstore/free/any-size size)
+  (assert! (<= 1 size) (format "size: ~d not in range: [1,∞]")) ; TODO: provide a concrete upper bound
+  (def sizes/base-32 (sizes/word-size<-size size))
+  (&begin (map &brk-cons sizes/base-32) ...)
+  )
+
+;; Helper function - Make list of relative offsets and sizes for partitions.
+;; E.g. (offsets-and-sizes<-size 65) -> [[0 32] [32 32] [64 1]]
+;; List [RelativeOffset Size] <- Nat
+(def (offsets-and-sizes<-size size)
+  (def sizes (sizes/word-size<-size size))
+  (def relative-offsets (iota (length sizes) 0 32))
+  (zip relative-offsets sizes))
+
+;; Given relative-offset and size,
+;; generates EVM code to:
+;; - load specified bytes
+;; - maintain start-offset for loading next segment
+;;
+;; stack input: start-offset
+;; stack output: start-offset bytes[offset:end]
+;;   where offset = offset+relative-offset
+;;         end    = offset+size
+(def &mload-1/any-size
+  (match <>
+    ([relative-offset size]
+     (&begin                   ; start-offset
+      DUP1 relative-offset ADD ; offset       start-offset
+      (&mload size) SWAP1))))  ; start-offset bytes[offset:end]
+
+;; stack input:  start-offset
+;; stack output: part0 part1 ... partn
+(def (&mload/any-size size)
+  (assert! (<= 0 size) (format "size: ~d not in range: [1,∞]")) ; TODO: provide a concrete upper bound
+  (match (offsets-and-sizes<-size size)
+    ;; = 0 bytes
+    ([] (&begin POP 0)
+     )
+    ;; > 0 bytes
+    ([[_ start-size] . rest]
+     (&begin
+      (map &mload-1/any-size (reverse rest)) ...
+      (&mload start-size)))
+    )
+  )

t/100-evm-instructions-integrationtest.ss

@@ -0,0 +1,91 @@
+(export #t)
+
+(import
+  :std/test :clan/number :clan/poo/object
+  ../types ../assembly ../evm-runtime
+  ../testing ../evm-instructions.ss
+  ./10-json-rpc-integrationtest)
+
+;; Stores free memory pointer at free memory location for returning
+(def &store-brk
+  (&begin (&mloadat brk@) DUP1 MSTORE))
+
+(def 100-evm-instructions-integrationtest
+  (test-suite "integration tests for evm instructions"
+    (test-case "EVM-type: &mstore/free/any-size, size = 1"
+      (evm-test [] (&begin
+                    #x20 (&mstoreat brk@ 32)
+                    (&push/any-size #u8(1))    ; stack: 1
+                    (&mstore/free/any-size 1)
+                    &store-brk
+                    )
+                [[Bytes1 . #u8(1)]
+                 [UInt256 . 33]]
+                result-in-memory?: #t
+                result-start: #x20))
+
+    (test-case "EVM-type: &mstore/free/any-size, size = 65"
+      (def 65-bytes (list->u8vector (make-list 65 65)))
+      (u8vector-set! 65-bytes 0 66)
+      (u8vector-set! 65-bytes 32 67)
+      (u8vector-set! 65-bytes 64 68)
+
+      (def 65-bytes/0-32  (subu8vector 65-bytes 0  32))
+      (def 65-bytes/32-64 (subu8vector 65-bytes 32 64))
+      (def 65-bytes/64-65 (subu8vector 65-bytes 64 65))
+
+      (evm-test [] (&begin
+                    #x20 (&mstoreat brk@ 32)
+                    (&push/any-size 65-bytes)
+                    (&mstore/free/any-size 65)
+                    &store-brk
+                    )
+                [[Bytes32 . 65-bytes/0-32]
+                 [Bytes32 . 65-bytes/32-64]
+                 [Bytes1  . 65-bytes/64-65]
+                 [UInt256 . 97] ; 32 (free mem ptr) + 65 (str65) = 97
+                 ]
+                result-in-memory?: #t
+                result-start: #x20)
+      )
+
+    (test-case "EVM-type &mload/free/any-size, size = 65"
+      (def 65-bytes (list->u8vector (make-list 65 65)))
+      (u8vector-set! 65-bytes 0 66)
+      (u8vector-set! 65-bytes 32 67)
+      (u8vector-set! 65-bytes 64 68)
+
+      (def 65-bytes/0-32  (subu8vector 65-bytes 0  32))
+      (def 65-bytes/32-64 (subu8vector 65-bytes 32 64))
+      (def 65-bytes/64-65 (subu8vector 65-bytes 64 65))
+
+      (evm-test [] (&begin
+                    ;; init freememptr
+                    #x20 (&mstoreat brk@ 32)     ; -
+
+                    ;; Store str65 in mem
+                    (&push/any-size 65-bytes)    ; bytes[0-32] bytes[32-64] bytes[64-65]
+                    (&mstore/free/any-size 65)   ; -
+                    (&mloadat/any-size #x20 65)  ; bytes[0-32] bytes[32-64] bytes[64-65]
+                    (&mstore/free/any-size 65)   ; -
+
+                    &store-brk
+                    )
+
+                 ;; Original 65-bytes
+                [[Bytes32 . 65-bytes/0-32]
+                 [Bytes32 . 65-bytes/32-64]
+                 [Bytes1  . 65-bytes/64-65]
+
+                 ;; Duplicate 65-bytes
+                 [Bytes32 . 65-bytes/0-32]
+                 [Bytes32 . 65-bytes/32-64]
+                 [Bytes1  . 65-bytes/64-65]
+
+                 ;; freememptr: 32 + 65 + 65 = 162
+                 [UInt256 . 162        ]
+                 ]
+                result-in-memory?: #t
+                result-start: #x20)
+      )
+    ))

t/80-evm-eval-integrationtest.ss

@@ -263,4 +263,16 @@
 
     (test-case "&marshal UInt8"
       (evm-test [] (&begin brk DUP1 DUP1 (&marshal UInt16 7))
-                [[UInt16 . 2]]))))
+                [[UInt16 . 2]]))
+
+    (test-case "&mstore 1 byte"
+      (evm-test [] (&begin 42 0 (&mstore 1))
+                [[UInt8 . 42]]
+                result-in-memory?: #t))
+
+    (test-case "&mstore 32 bytes"
+      (def maxUInt256 (- (expt 2 256) 1))
+      (evm-test [] (&begin maxUInt256 0 (&mstore 32))
+                [[UInt256 . maxUInt256]]
+                result-in-memory?: #t))
+    ))

t/evm-instructions-test.ss

@@ -0,0 +1,39 @@
+(export #t)
+
+(import
+  :std/test :clan/poo/object
+  ../assembly ../evm-instructions ../types)
+
+;; Verify assembled bytecode of instruction(s)
+(def (check-inst? i/actual i/expected)
+  (check-equal? (assemble/bytes i/actual) (assemble/bytes i/expected)))
+
+;; NOTE: Boxed/Unboxed stack<-mem (load) methods are dependent
+;; on EVM network (eip145).
+;; Hence they are tested in integration tests.
+(def evm-instructions-test
+  (test-suite "test suite for evm-instructions"
+    (test-case "&push/any-size <= 32"
+      (def 1-byte #u8(1))
+      (check-inst? (&push/any-size 1-byte) [1-byte])
+
+      (def 5-bytes #u8(104 101 108 108 111))
+      (check-inst? (&push/any-size 5-bytes) [5-bytes])
+
+      (def 32-bytes (list->u8vector (make-list 32 65)))
+      (check-inst? (&push/any-size 32-bytes) [32-bytes])
+      )
+
+    (test-case "&push/any-size > 32"
+      (def 65-bytes (list->u8vector (make-list 65 65)))
+      (u8vector-set! 65-bytes 0 66)  ; bytes/0-32:  66 65 65 ... 65
+      (u8vector-set! 65-bytes 32 67) ; bytes/32-64: 67 65 65 ... 65
+      (u8vector-set! 65-bytes 64 68) ; bytes/64-65: 68
+
+      (check-inst? (&push/any-size 65-bytes)
+                   [(subu8vector 65-bytes 64 65) ; bytes/64-65: 68
+                    (subu8vector 65-bytes 32 64) ; bytes/32-64: 67 65 65 ... 65
+                    (subu8vector 65-bytes 0  32) ; bytes/0-32:  66 65 65 ... 65
+                    ])
+      )
+    ))