From cb593385548aaf7b23760bf271d78e20d4f9b530 Mon Sep 17 00:00:00 2001
From: Nero <41307858+nero@users.noreply.github.com>
Date: Thu, 5 Jan 2023 13:08:58 +0000
Subject: [PATCH] awkasm: add bit instructions and refine

---
 host/asm.awk | 53 ++++++++++++++++++++++++++++++++++++++--------------
 1 file changed, 39 insertions(+), 14 deletions(-)

diff --git a/host/asm.awk b/host/asm.awk
index 105a78b..f57efdc 100644
--- a/host/asm.awk
+++ b/host/asm.awk
@@ -53,6 +53,15 @@ BEGIN {
   alu["sub"]=5
   alu["xor"]=6
   alu["cmp"]=7
+  # bit instructions
+  grp2["rol"]=0
+  grp2["ror"]=1
+  grp2["rcl"]=2
+  grp2["rcr"]=3
+  grp2["shl"]=4
+  grp2["sal"]=4
+  grp2["shr"]=5
+  grp2["sar"]=7
   # near conditional jumps
   ops_rel8["jo"]=112+0
   ops_rel8["jno"]=112+1
@@ -161,9 +170,11 @@ function expr(str) {
       }
     } else if (match(ep[k], /^[0-9]/)) {
       if (match(ep[k], /h$/)) {
-        ep[i]="0x" ep[k]
+        ep[k]="0x" ep[k]
       }
       val = val + sign*int(ep[k])
+    } else if (ep[k] == "$") {
+      val = val + sign*pos
     } else if (ep[k] in sym) {
       val = val + sign*sym[ep[k]]
     } else if (ep[k] in prevsym) {
@@ -253,16 +264,19 @@ function push_op_modrm(opcode) {
   } else if (rm2 && wordop && op1 in r16) {
     push_byte(opcode+2+1)
     push_modrm(op2, r16[op1], r16)
+  } else {
+    return 0
   }
+  return 1
 }
 # common encoding: one operand encoded as modrm with fixed spare field
 # operand can be byte or word, encoded in last bit of opcode
 function push_op_fixed_spare(opcode, spare) {
-  if (byteop) {
+  if (op1 in r8 || (substr(op1,1,1)=="[" && byteop) ) {
     push_byte(opcode)
     push_modrm(op1, spare, r8)
     return 1
-  } else if (wordop) {
+  } else if (op1 in r16 || (substr(op1,1,1)=="[" && wordop) ) {
     push_byte(opcode+1)
     push_modrm(op1, spare, r16)
     return 2
@@ -306,9 +320,9 @@ $2=="sym" {
   }
   # take note if we got a instruction size specifier
   op=$(opn)
-  if (op in prefix) {
+  while (op in prefix) {
     push_byte(prefix[op])
-    opn=opn+1
+    opn++
     op=$(opn)
   }
 
@@ -378,6 +392,15 @@ op=="equ" {
   val=crit(a[1])
   submit(val);next
 }
+op==".space" {
+  len=crit(a[1])
+  val=0
+  if (a[2]) val=crit(a[2])
+  for(i=1;i<=len;i++) {
+    push_byte(val)
+  }
+  submit(pos);next
+}
 op=="db" {
   for(i=1;i<=c;i++) {
     if (substr(a[i],1,1)=="\"") {
@@ -392,8 +415,7 @@ op=="dw" {
 }
 # arithmetics: ADD, SUB, XOR etc
 op in alu {
-  push_op_modrm(alu[op]*8)
-  if (!hex) {
+  if (!push_op_modrm(alu[op]*8)) {
     size=push_op_fixed_spare(0x80, alu[op])
     if (size==1) {
       push_byte(imm(op2))
@@ -402,10 +424,16 @@ op in alu {
     }
   }
 }
+# Group 2
+op in grp2 && op2=="1" {
+  push_op_fixed_spare(208, grp2[op])
+}
+op in grp2 && op2=="cl" {
+  push_op_fixed_spare(210, grp2[op])
+}
 # no idea why this made this extra, this is a AND without storing the result
 op=="test" {
-  push_op_modrm(132) # 84
-  if (!hex) {
+  if (!push_op_modrm(132)) { # 84
     size=push_op_fixed_spare(246, 0)
     if (size==1) {
       push_byte(imm(op2))
@@ -414,9 +442,6 @@ op=="test" {
     }
   }
 }
-op=="sar" && op2=="1" {
-  push_op_fixed_spare(208, 0) # D0 /0
-}
 op=="inc" && rm1 {
   push_op_fixed_spare(254, 0) # FE /0
 }
@@ -473,10 +498,10 @@ op=="int" { # CD
 }
 op=="jmp" {
   val=imm(op1)-(pos+2)
-  if (val>-127 && val<128 && ecrit) {
+  if (!wordop || (val>-127 && val<128 && ecrit)) {
     push_byte(235)
     push_signed_byte(val)
-  } else {
+  } else if(wordop) {
     push_byte(233)
     push_word(val-1)
   }