From b70cdd13c45c61ec836f8054f09b5c82bee6e071 Mon Sep 17 00:00:00 2001
From: Paul Mathieu <paulmathieu@google.com>
Date: Thu, 18 Mar 2021 08:35:23 -0700
Subject: [PATCH] Add partial support for 8-bit loads

We're big-endian.
when we have this in memory:

00 01 02 03

And we need the first byte, we load the first 16-bit word:

0x0001

The first byte is then in the upper part of the word, and requires
a right shift by 8.
So any load into an 8-bit typed container needs to shift stuff.

So far, stores from/to the stack are exempted, they always load/store
full 16-bit words.

And a few othe rminor things. Like string null terminators.
And escaped characters in character literals.

Can you believe it's spelled 'literal', with a single t?
Me neither.
---
 cpu/cpu.vhdl  |  6 +++++-
 dsp/main.c    |  6 ++++--
 tools/cc.ebnf |  2 +-
 tools/cc.py   | 23 +++++++++++++++++++----
 4 files changed, 29 insertions(+), 8 deletions(-)

diff --git a/cpu/cpu.vhdl b/cpu/cpu.vhdl
index b5c96ed..4afd788 100644
--- a/cpu/cpu.vhdl
+++ b/cpu/cpu.vhdl
@@ -121,7 +121,11 @@ begin
         end if;
 
         regn_0 := to_integer(unsigned(load_reg(3 downto 0)));
-        reg_d(regn_0) <= mem_in;
+        if load_addr(0) = '1' then  -- 8-bit load
+          reg_d(regn_0) <= mem_in(7 downto 0) & x"00";
+        else
+          reg_d(regn_0) <= mem_in;
+        end if;
       when BRANCH =>
         inst := x"0000";  -- NOP
         reg_d(14) <= std_logic_vector(unsigned(reg_q(14)) + 2);
diff --git a/dsp/main.c b/dsp/main.c
index c09b058..8c75a08 100644
--- a/dsp/main.c
+++ b/dsp/main.c
@@ -9,11 +9,11 @@ void init() {
 }
 
 int main() {
-  led0->output = 37;
   init();
 
+  led0->output = 42;
+
   while(1) {
-    led0->output = 42;
     uint8_t c = uart_read(uart0);
     uint16_t period;
     if (c == 'c') {
@@ -44,6 +44,8 @@ int main() {
       period = 791;  //  B4, 493.88 Hz, maybe
       uart_writen(uart0, "B4\r\n", 4);
       led0->output = 64;
+    } else {
+      led0->output = 37;
     }
 
     square0->period = period;
diff --git a/tools/cc.ebnf b/tools/cc.ebnf
index 997eef9..b3ae672 100644
--- a/tools/cc.ebnf
+++ b/tools/cc.ebnf
@@ -145,7 +145,7 @@ struct_field: type IDENTIFIER sized_array* ";"
 IDENTIFIER: /[_a-zA-Z]\w*/
 COMMENT: /\/\/.*/
 HEX_LITTERAL: /0x[a-fA-F0-9]+/
-CHARACTER: /'[^']'/
+CHARACTER: /'([^']|\\n|\\r|\\0|\\t)'/
 
 
 %import common.WS
diff --git a/tools/cc.py b/tools/cc.py
index 3b76be7..df5ca23 100644
--- a/tools/cc.py
+++ b/tools/cc.py
@@ -366,7 +366,7 @@ class CcTransform(lark.visitors.Transformer):
     mul = _binary_op(lambda a, b: a*b)
     shl = _binary_op(lambda a, b: a<<b)
 
-    CHARACTER = lambda _, x: ord(x[1])
+    CHARACTER = lambda _, x: ord(ast.literal_eval(x)[0])
     IDENTIFIER = str
     SIGNED_NUMBER = int
     HEX_LITTERAL = lambda _, x: int(x[2:], 16)
@@ -543,6 +543,20 @@ class Load(AsmOp):
     def __init__(self, fun, ops):
         self.fun = fun
         self.dest, self.var = ops
+        fun.log(f'8bit load? {self.var.type} {self.is8bit}')
+
+    @property
+    def is8bit(self):
+        if not isinstance(self.var.type, PodType):
+            return False
+        if self.var.type.pod in ['uint8_t', 'int8_t', 'char']:
+            return True
+        return False
+
+    def _maybecast8bit(self, reg):
+        if self.is8bit:
+            return [f'shr {reg}, {reg}, 8']
+        return []
 
     @property
     def out(self):
@@ -552,13 +566,13 @@ class Load(AsmOp):
         reg = self.dest
         if self.var.name in self.fun.locals:
             src = self.var.stackaddr
-            return [f'load {reg}, [sp, {src}]']
+            return [f'load {reg}, [sp, {src}]']  # no 8bit cast to/from stack
         elif self.var.addr_reg is not None:
-            return [f'load {reg}, [{self.var.addr_reg}]']
+            return [f'load {reg}, [{self.var.addr_reg}]'] + self._maybecast8bit(reg)
         else:
             return [f'set {reg}, {self.var.name}',
                     f'nop  // in case we load a far global',
-                    f'load {reg}, [{reg}]']
+                    f'load {reg}, [{reg}]']  # unsure if we need a cast here
 
 class Store(AsmOp):
     scratch_need = 1
@@ -1254,6 +1268,7 @@ def parse_tree(tree, debug=False):
     for sym, dat in inte.strings.items():
         out += [f'.global {sym}',
                 f'{sym}:']
+        dat += '\0'
         if len(dat) % 2 != 0:
             dat += '\0'
         dat = dat.encode()