oasm: Add initial support for labels

A label is like a pin on a map with a number attached to it. OASM keeps track
of a number referred to as the psuedo instruction pointer (PIP). This value is
initially zero by default and is incremented every instruction. The purpose of
this value is to pin a location in the code and mark it to be at a specific address
so that further references to that label would be translated to the previous PIP value
at the time of encountering the label.

Signed-off-by: Ian Moffett <ian@osmora.org>

5 files changed, 53 insertions, 16 deletions

diff --git a/etc/oemu/test-00.s b/etc/oemu/test-00.s
index b34c099..1fa7ff1 100644
--- a/etc/oemu/test-00.s
+++ b/etc/oemu/test-00.s
@@ -1,13 +1,14 @@
-mov x0, #1   ! ~ 0x00000000
-mov x1, #2   ! ~ 0x00000004
-mov x2, #3   ! ~ 0x00000008
-             ! X
-mov x2, #20  ! ~ 0x0000000C
-br x2        ! ~ 0x00000010  -----+
-!!!!!!!!!!!!!! X                  |
-!!!!!!!!!!!!!! X                  |
-             ! X                  |
-dec x0       ! ~ 0x00000014  <----+
-inc x1       ! ~ 0x00000018
-add x2, #3   ! ~ 0x0000001C
-hlt          ! ~ 0x00000020
+mov x0, #1       ! ~ 0x00000000
+mov x1, #2       ! ~ 0x00000004
+mov x2, #3       ! ~ 0x00000008
+                 ! X
+some_label:      ! X
+    mov x2, #20  ! ~ 0x0000000C
+    br x2        ! ~ 0x00000010  -----+
+    !!!!!!!!!!!!!! X                  |
+    !!!!!!!!!!!!!! X                  |
+                 ! X                  |
+    dec x0       ! ~ 0x00000014  <----+
+    inc x1       ! ~ 0x00000018
+    add x2, #3   ! ~ 0x0000001C
+    hlt          ! ~ 0x00000020
diff --git a/usr.bin/oasm/include/oasm/lex.h b/usr.bin/oasm/include/oasm/lex.h
index 62183e0..69e50a1 100644
--- a/usr.bin/oasm/include/oasm/lex.h
+++ b/usr.bin/oasm/include/oasm/lex.h
@@ -103,6 +103,7 @@ typedef enum {
     TT_INC,         /* 'inc' */
     TT_DEC,         /* 'dec' */
     TT_IMM,         /* #<n> */
+    TT_LABEL,       /* 'label: ...' */
 
     /* Register sets */
     __XN_REGS,      /* x0-x15 */
diff --git a/usr.bin/oasm/include/oasm/state.h b/usr.bin/oasm/include/oasm/state.h
index 5f58144..6dd2435 100644
--- a/usr.bin/oasm/include/oasm/state.h
+++ b/usr.bin/oasm/include/oasm/state.h
@@ -39,6 +39,8 @@
  * OASM state:
  *
  * @filename: Filname of unit we are parsing
+ * @pip: Pseudo instruction pointer
+ * @label_ip: IP at current label start
  * @in_fd: Input file descriptor
  * @out_fd: Resulting binary output file descriptor
  * @line: Current line number
@@ -46,6 +48,8 @@
  */
 struct oasm_state {
     char *filename;
+    off_t pip;
+    off_t label_ip;
     int in_fd;
     int out_fd;
     off_t line;
diff --git a/usr.bin/oasm/lex.c b/usr.bin/oasm/lex.c
index a33a570..87ac464 100644
--- a/usr.bin/oasm/lex.c
+++ b/usr.bin/oasm/lex.c
@@ -50,6 +50,20 @@ static char putback = '\0';
 #define S_IMN_HLT  "hlt"
 #define S_IMN_BR   "br"
 
+/* Instruction length */
+#define OSMX64_INST_LEN 4
+
+/*
+ * Update the state when the caller encounters
+ * a newline.
+ */
+static inline void
+lex_newline(struct oasm_state *state)
+{
+    ++state->line;
+    state->pip += OSMX64_INST_LEN;
+}
+
 /*
  * Returns 0 if a char is counted as a
  * skippable token. Otherwise, -1
@@ -63,7 +77,7 @@ lex_skippable(struct oasm_state *state, char c)
     case '\t': return 0;
     case '\r': return 0;
     case '\n':
-        ++state->line;
+        lex_newline(state);
         return 0;
     }
 
@@ -154,6 +168,10 @@ lex_nomstr(struct oasm_state *state, char **res)
             retval = tmp;
             break;
         }
+        if (tmp == ':') {
+            retval = tmp;
+            break;
+        }
         if (tmp == '\n') {
             ++state->line;
             retval = tmp;
@@ -295,7 +313,7 @@ lex_tok(struct oasm_state *state, struct oasm_token *ttp)
 
     switch (c) {
     case '\n':
-        ++state->line;
+        lex_newline(state);
         return 0;
     case '\0':
         return -1;
@@ -306,7 +324,14 @@ lex_tok(struct oasm_state *state, struct oasm_token *ttp)
         ttp->raw = NULL;
 
         lex_putback(c);
-        lex_nomstr(state, &p);
+        c = lex_nomstr(state, &p);
+
+        while (c == ':') {
+            ttp->type = TT_LABEL;
+            ttp->raw = p;
+            state->label_ip = state->pip;
+            return 0;
+        }
 
         /* Arithmetic operation? */
         if ((tok = token_arith(p)) != TT_UNKNOWN) {
@@ -340,6 +365,7 @@ lex_tok(struct oasm_state *state, struct oasm_token *ttp)
             ttp->raw = p;
             return 0;
         }
+
         oasm_err("bad token \"%s\"\n", p);
         lex_try_free(p);
         return -1;
diff --git a/usr.bin/oasm/parse.c b/usr.bin/oasm/parse.c
index 673fbcd..113671f 100644
--- a/usr.bin/oasm/parse.c
+++ b/usr.bin/oasm/parse.c
@@ -49,6 +49,8 @@ static const char *tokstr[] = {
     [ TT_DEC ] = "dec",
     [ TT_MOV ] = "mov",
     [ TT_IMM ]  = "<imm>",
+    [ TT_LABEL] = "<label>",
+
 
     /* X<n> registers */
     [ TT_X0 ]  = "x0",
@@ -150,6 +152,9 @@ parse_tok(struct oasm_state *state, struct oasm_token *tok)
         state->last = tok->type;
         emit_osmx64(&emit_state, tok);
         break;
+    case TT_LABEL:
+        state->last = tok->type;
+        break;
     case TT_HLT:
         state->last = tok->type;
         emit_osmx64(&emit_state, tok);