Port lexer to leg.

8 files changed, 146 insertions, 118 deletions

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..ae79679
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,5 @@
+.redo
+minias
+*.s
+*.o
+*.inc
\ No newline at end of file
diff --git a/README.md b/README.md
index 4eb7c2b..2512215 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
-# dumbas
+# minias
 
-A dumb assembler for x86_64, written for fun and learning.
+A mini assembler for x86_64, written for fun and learning.
 
 Goals:
 
@@ -14,6 +14,19 @@ Non Goals:
 - Assemble other architectures.
 - Work as a library.
 
+# Building
+
+```
+redo minias
+```
+
+or 
+
+```
+leg asm.peg > asm.peg.inc
+cc -O2 *.c -o minias
+```
+
 # Resources
 
 - [elf spec](https://refspecs.linuxfoundation.org/elf/elf.pdf)
diff --git a/asm.peg b/asm.peg
new file mode 100644
index 0000000..946f6e6
--- /dev/null
+++ b/asm.peg
@@ -0,0 +1,80 @@
+
+line  = s:stmt (eol | !. ) { yy->v = s; }
+      | eol { yy->v.kind = ASM_BLANK; }
+      | . { yy->v.kind = ASM_SYNTAX_ERROR; }
+
+stmt = d:directive {$$ = d;}
+      | i:instr { $$ = i; }
+      | l:label {  $$ = l; } 
+
+directive = ".glob" "o"? "l" ws i:ident
+             { $$.globl = (Globl){.kind = ASM_DIR_GLOBL, .name = i.ident.name }; }
+           | ".data" { $$.kind = ASM_DIR_DATA; }
+           | ".text" { $$.kind = ASM_DIR_TEXT; }
+           | ".balign" ws n:number 
+             { $$.balign = (Balign){.kind = ASM_DIR_BALIGN, .align = n.number.v }; }
+           | ".byte" ws n:number
+             { $$.byte = (Byte){.kind = ASM_DIR_BYTE, .b = (uint8_t)n.number.v }; }
+
+label = i:ident ':'
+         { $$.label = (Label){.kind = ASM_LABEL, .name = i.ident.name}; }
+
+instr = "nop"{ $$.kind = ASM_NOP; }
+       | "leave" { $$.kind = ASM_LEAVE; }
+       | "ret" { $$.kind = ASM_RET; }
+       | i:jmp { $$ = i; }
+       | i:add { $$ = i; }
+
+jmp = "jmp" ws i:ident
+       { $$.jmp = (Jmp){.kind = ASM_JMP, .target = i.ident.name}; }
+
+add = "addq" ws s:r-m64 ws? ',' ws? d:r64
+      { $$.add = (Add){ .kind = ASM_ADD, .type = 'q', .src = dupv(&s), .dst = dupv(&d) } }
+    | "addq" ws s:r64 ws? ',' ws? d:r-m64
+      { $$.add = (Add){ .kind = ASM_ADD, .type = 'q', .src = dupv(&s), .dst = dupv(&d) } }
+    | "addq" ws s:imm ws? ',' ws? d:r-m64
+      { $$.add = (Add){ .kind = ASM_ADD, .type = 'q', .src = dupv(&s), .dst = dupv(&d) } }
+
+r-m64 = r:r64 { $$ = r; }
+      | m:m { $$ = m; }
+
+m = '(' ws? r:r64 ws? ')'
+  {  $$.memarg = (Memarg){ .kind = ASM_MEMARG, .c = 0, .l = NULL, .reg = r.kind  }  }
+  | <'-'?[0-9]+> ws? '(' ws? r:r64 ws? ')'
+  { $$.memarg = (Memarg){ .kind = ASM_MEMARG, .c = strtoll(yytext, NULL, 10), .l = NULL, .reg = r.kind  }  }
+  | i:ident  ws? '(' ws? r:r64 ws? ')'
+  { $$.memarg = (Memarg){ .kind = ASM_MEMARG, .c = 0, .l = i.ident.name, .reg = r.kind  }  }
+
+r64 = "%rax" { $$.kind = ASM_RAX }
+    | "%rcx" { $$.kind = ASM_RCX }
+    | "%rdx" { $$.kind = ASM_RDX }
+    | "%rbx" { $$.kind = ASM_RBX }
+    | "%rsp" { $$.kind = ASM_RSP }
+    | "%rbp" { $$.kind = ASM_RBP }
+    | "%rsi" { $$.kind = ASM_RSI }
+    | "%rdi" { $$.kind = ASM_RDI }
+    | "%r8"  { $$.kind = ASM_R8 }
+    | "%r9"  { $$.kind = ASM_R9 }
+    | "%r10" { $$.kind = ASM_R10 }
+    | "%r11" { $$.kind = ASM_R11 }
+    | "%r12" { $$.kind = ASM_R12 }
+    | "%r13" { $$.kind = ASM_R13 }
+    | "%r14" { $$.kind = ASM_R14 }
+    | "%r15" { $$.kind = ASM_R15 }
+
+
+imm = '$' i:ident
+       { $$.imm = (Imm){.kind = ASM_IMM, .l = i.ident.name, .c = 0 };  }
+     | '$' <'-'?[0-9]+>
+       { $$.imm = (Imm){.kind = ASM_IMM, .l = NULL, .c = strtoll(yytext, NULL, 10) }; }
+
+ident = <[_a-zA-Z][_a-zA-Z0-9]*>
+         { $$.ident = (Ident){ .kind = ASM_IDENT, .name = xstrdup(yytext) }; } 
+
+number = <'-'?[0-9]+>
+          { $$.number = (Number){ .kind = ASM_NUMBER, .v = strtoll(yytext, NULL, 10) }; }
+
+ws = [ \t]+
+
+eol = ws? "\n"
+
diff --git a/asmparser.peg b/asmparser.peg
deleted file mode 100644
index 5a625ae..0000000
--- a/asmparser.peg
+++ /dev/null
@@ -1,96 +0,0 @@
-%prefix "asmparser"
-
-%value "Parsev"
-
-line <- s:stmt eol { $$ = s; }
-      / eol { $$.kind = ASM_BLANK; }
-      / . { $$.kind = ASM_SYNTAX_ERROR; }
-
-stmt <- d:directive {$$ = d;}
-      / i:instr { $$ = i; }
-      / l:label {  $$ = l; } 
-
-directive <- ".glob" "o"? "l" ws i:ident
-             { $$.globl = (Globl){.kind = ASM_DIR_GLOBL, .name = i.ident.name }; }
-           / ".data" { $$.kind = ASM_DIR_DATA; }
-           / ".text" { $$.kind = ASM_DIR_TEXT; }
-           / ".balign" ws n:number 
-             { $$.balign = (Balign){.kind = ASM_DIR_BALIGN, .align = n.number.v }; }
-           / ".byte" ws n:number
-             { $$.byte = (Byte){.kind = ASM_DIR_BYTE, .b = (uint8_t)n.number.v }; }
-
-label <- i:ident ':'
-         { $$.label = (Label){.kind = ASM_LABEL, .name = i.ident.name}; }
-
-instr <- "nop"{ $$.kind = ASM_NOP; }
-       / "leave" { $$.kind = ASM_LEAVE; }
-       / "ret" { $$.kind = ASM_RET; }
-       / i:jmp { $$ = i; }
-       / i:add { $$ = i; }
-
-jmp <- "jmp" ws i:ident
-       { $$.jmp = (Jmp){.kind = ASM_JMP, .target = i.ident.name}; }
-
-add <- "add" 'q'? ws s:m ws? ',' ws? d:r64
-     { $$.add = mkadd('q', s, d); }
-     / "add" 'q'? ws s:imm ws? ',' ws? d:r64
-     { $$.add = mkadd('q', s, d); }
-     / "add" 'q'? ws s:r64 ws? ',' ws? d:m
-     { $$.add = mkadd('q', s, d); }
-     / "add" 'q'? ws s:r64 ws? ',' ws? d:r64
-     { $$.add = mkadd('q', s, d); }
-     / "addq" ws s:imm ws? ',' ws? d:m
-     { $$.add = mkadd('q', s, d); }
-
-m <- '(' ws? r:r64 ws? ')'
-   {  $$.memarg = (Memarg){ .kind = ASM_MEMARG, .c = 0, .l = NULL, .reg = r.kind  };  }
-   / <'-'?[0-9]+> ws? '(' ws? r:r64 ws? ')'
-   { $$.memarg = (Memarg){ .kind = ASM_MEMARG, .c = strtoll($1, NULL, 10), .l = NULL, .reg = r.kind  };  }
-   / i:ident  ws? '(' ws? r:r64 ws? ')'
-   { $$.memarg = (Memarg){ .kind = ASM_MEMARG, .c = 0, .l = i.ident.name, .reg = r.kind  };  }
-
-r64 <- "%rax" { $$.kind = ASM_RAX; }
-     / "%rcx" { $$.kind = ASM_RCX; }
-     / "%rdx" { $$.kind = ASM_RDX; }
-     / "%rbx" { $$.kind = ASM_RBX; }
-     / "%rsp" { $$.kind = ASM_RSP; }
-     / "%rbp" { $$.kind = ASM_RBP; }
-     / "%rsi" { $$.kind = ASM_RSI; }
-     / "%rdi" { $$.kind = ASM_RDI; }
-     / "%r8"  { $$.kind = ASM_R8; }
-     / "%r9"  { $$.kind = ASM_R9; }
-     / "%r10" { $$.kind = ASM_R10; }
-     / "%r11" { $$.kind = ASM_R11; }
-     / "%r12" { $$.kind = ASM_R12; }
-     / "%r13" { $$.kind = ASM_R13; }
-     / "%r14" { $$.kind = ASM_R14; }
-     / "%r15" { $$.kind = ASM_R15; }
-
-imm <- '$' i:ident
-       { $$.imm = (Imm){.kind = ASM_IMM, .l = i.ident.name, .c = 0 };  }
-     / '$' <'-'?[0-9]+>
-       { $$.imm = (Imm){.kind = ASM_IMM, .l = NULL, .c = strtoll($1, NULL, 10) }; }
-
-ident <- <[_a-zA-Z][_a-zA-Z0-9]*>
-         { $$.ident = (Ident){ .kind = ASM_IDENT, .name = xstrdup($1) }; } 
-
-number <- <'-'?[0-9]+>
-          { $$.number = (Number){ .kind = ASM_NUMBER, .v = strtoll($1, NULL, 10) }; }
-
-ws <- [ \t]+
-
-eol <- ws? ("\n" / (! .))
-
-%source {
-
-static Parsev *dupv(Parsev *p) { 
-  Parsev *r = xmalloc(sizeof(Parsev));
-  *r = *p;
-  return r;
-}
-
-static Add mkadd(char t, Parsev s, Parsev d) {
-  return (Add){ .kind = ASM_ADD, .type = 'q', .src = dupv(&s), .dst = dupv(&d) };
-}
-
-}
\ No newline at end of file
diff --git a/default.do b/default.do
new file mode 100644
index 0000000..6787b8d
--- /dev/null
+++ b/default.do
@@ -0,0 +1,28 @@
+#!/bin/sh
+set -eu
+
+case "$1" in
+  *.o)
+    test "$1" = "main.o" && redo-ifchange asm.peg.inc
+    redo-ifchange "${1%.o}.c" minias.h
+    set -x
+    ${CC:- cc} ${CFLAGS:- -O -Og} -c -o "$3" "${1%.o}.c"
+  ;;
+  asm.peg.inc)
+    redo-ifchange asm.peg
+    set -x
+    leg asm.peg > "$3"
+  ;;
+  minias)
+    obj="main.o util.o "
+    redo-ifchange $obj
+    ${CC:- cc} ${LDFLAGS:-} -o "$3" $obj
+  ;;
+  fmt)
+    clang-format -i main.c util.c >&2
+  ;;
+  *)
+    echo "don't know how to build $1" 2>&1
+    exit 1
+  ;;
+esac
\ No newline at end of file
diff --git a/main.c b/main.c
index 4806722..69de674 100644
--- a/main.c
+++ b/main.c
@@ -1,4 +1,4 @@
-#include "dumbas.h"
+#include "minias.h"
 
 static AsmLine *allasm = NULL;
 
@@ -113,45 +113,43 @@ static void initsections(void) {
   text->hdr.sh_addralign = 4;
 }
 
-static const char *dbg_str[] = {"Evaluating rule", "Matched rule",
-                                "Abandoning rule"};
-#define PCC_DEBUG(event, rule, level, pos, buffer, length)                     \
-  fprintf(stderr, "%*s%s %s @%zu [%.*s]\n", (int)((level)*2), "",              \
-          dbg_str[event], rule, pos, (int)(length), buffer)
+static Parsev *dupv(Parsev *p) { 
+  Parsev *r = xmalloc(sizeof(Parsev));
+  *r = *p;
+  return r;
+}
 
-#include "asmparser.c" // XXX resolve dependency cycle.
+#define YYSTYPE Parsev
+#define YY_CTX_LOCAL
+#define YY_CTX_MEMBERS Parsev v;
+#include "asm.peg.inc"
 
 void parse(void) {
-  int more;
   uint64_t lineno;
-  Parsev v;
   AsmLine *l, *prevl;
-  asmparser_context_t *ctx;
+  yycontext ctx;
 
-  ctx = asmparser_create(NULL);
+  memset(&ctx, 0, sizeof(yycontext));
   prevl = NULL;
   lineno = 0;
 
-  do {
-    more = asmparser_parse(ctx, &v);
+  while (yyparse(&ctx)) {
     lineno += 1;
-    if (v.kind == ASM_SYNTAX_ERROR) {
+    if (ctx.v.kind == ASM_SYNTAX_ERROR) {
       fprintf(stderr, "<stdin>:%lu: syntax error\n", lineno);
       exit(1);
     }
-    if (v.kind == ASM_BLANK)
+    if (ctx.v.kind == ASM_BLANK)
       continue;
     l = zalloc(sizeof(AsmLine));
-    l->v = v;
+    l->v = ctx.v;
     l->lineno = lineno;
     if (prevl)
       prevl->next = l;
     else
       allasm = l;
     prevl = l;
-  } while (more);
-
-  asmparser_destroy(ctx);
+  }
 }
 
 /* Shorthand helpers to write section bytes. */
diff --git a/dumbas.h b/minias.h
index 9889733..9889733 100644
--- a/dumbas.h
+++ b/minias.h
diff --git a/util.c b/util.c
index 6e09683..5ab9c48 100644
--- a/util.c
+++ b/util.c
@@ -1,4 +1,4 @@
-#include "dumbas.h"
+#include "minias.h"
 
 static void vwarn(const char *fmt, va_list ap) {
   vfprintf(stderr, fmt, ap);