From 61d3688581cd38376f1ed0e9440d930af22053f6 Mon Sep 17 00:00:00 2001
From: Nero <41307858+nero@users.noreply.github.com>
Date: Mon, 2 Jan 2023 01:37:05 +0000
Subject: [PATCH] Implement addition expressions, critical expressions, proper
 modrm

---
 host/asm.awk | 135 ++++++++++++++++++++++++++++++++++-----------------
 1 file changed, 90 insertions(+), 45 deletions(-)

diff --git a/host/asm.awk b/host/asm.awk
index 7880d99..105a78b 100644
--- a/host/asm.awk
+++ b/host/asm.awk
@@ -42,6 +42,7 @@ BEGIN {
   modstr["dibp"]=3
   modstr["si"]=4
   modstr["di"]=5
+  modstr["bp"]=6
   modstr["bx"]=7
   # ALU operations
   alu["add"]=0
@@ -119,6 +120,7 @@ function err(str) {
 }
 # submit a assembling result to output
 # set label of current line to off
+# this outputs a listing line
 function submit(off) {
   if (of=="lst") printf("%04X %-18s %s\n", off, hex, $0)
   if (of=="hex" && hex) printf("%s", hex)
@@ -132,23 +134,61 @@ function submit(off) {
   pos=pos+length(hex)/2
   hex=""
 }
-# gets a symbol's values
-function getsym(name) {
-  if (substr(name,1,1)=="0" || int(name)) {
-    return int(name)
+# evaluate an expression
+# globals set:
+#   ecrit: value known after first pass, 1=yes, 0=no
+#   eregs: concatenated list of registers that add to this expr (for modrm)
+function expr(str) {
+  val=0
+  sign=1
+  ecrit=1
+  eregs=""
+  gsub(/-/, "+-", str)
+  split(str, ep,"+")
+  for (k in ep) {
+    if (substr(ep[k],1,1)=="-") {
+      gsub(/^-/, "", ep[k])
+      sign = -1
+    } else {
+      sign = 1
+    }
+    if (substr(ep[k],1,1)==".") ep[k] = plabel ep[k]
+    if (ep[k] in r8 || ep[k] in r16 || ep[k] in sreg) {
+      if (sign > 0) {
+        eregs=eregs ep[k]
+      } else {
+        err("Registers cannot be subtractive in expressions")
+      }
+    } else if (match(ep[k], /^[0-9]/)) {
+      if (match(ep[k], /h$/)) {
+        ep[i]="0x" ep[k]
+      }
+      val = val + sign*int(ep[k])
+    } else if (ep[k] in sym) {
+      val = val + sign*sym[ep[k]]
+    } else if (ep[k] in prevsym) {
+      val = val + sign*prevsym[ep[k]]
+      ecrit=0
+    } else {
+      err("Undefined label " ep[k])
+      ecrit=0
+    }
   }
-  if (name in r8 || name in r16 || name in sreg) {
-    err("Register name " name " is not a valid immediate")
-    return 0
-  }
-  if (substr(name,1,1)==".") name = plabel name
-  if (name in prevsym) {
-    return prevsym[name]
-  } else if (name in sym) {
-    return sym[name]
-  }
-  return 0
+  return val
 }
+
+function imm(str) {
+  val = expr(str)
+  if (eregs) err("Registers not allowed here")
+  return val
+}
+
+function crit(str) {
+  val = imm(str)
+  if (!ecrit) err("Labels from below not allowed here")
+  return val
+}
+
 function push_byte(val) {
   #print("; pushb " val)
   if (val<0 || val>=256) err("Value " val " does not fit in byte")
@@ -165,29 +205,27 @@ function push_word(val) {
 }
 # rs is the register set (r8, r16) that can show up in str
 function push_modrm(str, spare, rs) {
-  mod=3
+  mod=0
   rm=0
   if (str in rs) {
     mod=3
     rm=rs[str]
   } else if (substr(str,1,1)=="[") {
-    rmap=""
     gsub(/^\[|\]$/, "", str)
-    split(str,rmp,"+")
-    disp=0
-    for (i in rmp) {
-      if (rmp[i] in r16) {
-        rmap=rmap rmp[i]
-      } else {
-        disp = disp + getsym(rmp[i])
-        mod=2
-      }
+    disp=expr(str)
+    if (!ecrit || disp) {
+      mod=2
     }
-    if (!rmap) {
+    if (!eregs) {
       mod=0
       rm=6
-    } else if (rmap in modstr) {
-      rm=modstr[rmap]
+    } else if (eregs in modstr) {
+      rm=modstr[eregs]
+      # [BP] is unencodable, this combination is read as [0000]
+      # so we upgrade [BP] to [BP+00]
+      if (mod==0 && rm==6) {
+        mod=1
+      }
     } else {
       err("Bad modR/M")
     }
@@ -333,11 +371,11 @@ op=="cpu" {
   next
 }
 op=="org" {
-  pos=getsym(a[1])
+  pos=crit(a[1])
   submit(pos);next
 }
 op=="equ" {
-  val=getsym(a[1])
+  val=crit(a[1])
   submit(val);next
 }
 op=="db" {
@@ -346,11 +384,11 @@ op=="db" {
       for(j=2;j<length(a[i]);j++) {
         push_byte(_ord_[substr(a[i],j,1)])
       }
-    } else push_byte(getsym(a[i]))
+    } else push_byte(imm(a[i]))
   }
 }
 op=="dw" {
-  for(i=1;i<=c;i++) push_word(getsym(a[i]))
+  for(i=1;i<=c;i++) push_word(imm(a[i]))
 }
 # arithmetics: ADD, SUB, XOR etc
 op in alu {
@@ -358,9 +396,9 @@ op in alu {
   if (!hex) {
     size=push_op_fixed_spare(0x80, alu[op])
     if (size==1) {
-      push_byte(getsym(op2))
+      push_byte(imm(op2))
     } else if (size==2) {
-      push_word(getsym(op2))
+      push_word(imm(op2))
     }
   }
 }
@@ -370,9 +408,9 @@ op=="test" {
   if (!hex) {
     size=push_op_fixed_spare(246, 0)
     if (size==1) {
-      push_byte(getsym(op2))
+      push_byte(imm(op2))
     } else if (size==2) {
-      push_word(getsym(op2))
+      push_word(imm(op2))
     }
   }
 }
@@ -407,10 +445,10 @@ op=="mov" {
   # reg <- imm
   } else if (op1 in r8) {
     push_byte(176+r8[op1])
-    push_byte(getsym(op2))
+    push_byte(imm(op2))
   } else if (op1 in r16) {
     push_byte(184+r16[op1])
-    push_word(getsym(op2))
+    push_word(imm(op2))
   # modrm <- imm
   } else if (byteop && rm1) {
     push_byte(198)
@@ -431,15 +469,22 @@ op=="int" && op1=="3" { # CC breakpoint
 }
 op=="int" { # CD
   push_byte(205)
-  push_byte(getsym(op1))
+  push_byte(imm(op1))
 }
-op=="jmp" && wordop {
-  push_byte(233)
-  push_word(getsym(op1)-(pos+3))
+op=="jmp" {
+  val=imm(op1)-(pos+2)
+  if (val>-127 && val<128 && ecrit) {
+    push_byte(235)
+    push_signed_byte(val)
+  } else {
+    push_byte(233)
+    push_word(val-1)
+  }
+  submit(pos);next
 }
 op=="call" && wordop {
   push_byte(232)
-  push_word(getsym(op1)-(pos+3))
+  push_word(imm(op1)-(pos+3))
 }
 op=="neg" {
   push_op_fixed_spare(246, 3)
@@ -447,7 +492,7 @@ op=="neg" {
 # opcodes with rel8 encoding
 op in ops_rel8 && byteop && c==1 {
   push_byte(ops_rel8[op])
-  push_signed_byte(getsym(op1)-(pos+2))
+  push_signed_byte(imm(op1)-(pos+2))
 }
 # opcodes without arguments
 op in ops_sb { push_byte(ops_sb[op]) }