From 33a4671260b567173c2c931992cd096513f4b9ca Mon Sep 17 00:00:00 2001
From: Andreas Baumann <mail@andreasbaumann.cc>
Date: Thu, 23 Jul 2020 17:28:43 +0200
Subject: parsing $ and $$ in asm-i386 - freestanding libc: added ungetc
 function with one pushback character - asm-i386: for now adding $ and $$ as
 special addresses to the operands in the expression tree

---
 ecomp-c/asm-i386.c             | 49 ++++++++++++++++++++++++++++++++++--------
 ecomp-c/libc-freestanding.c    | 16 +++++++++++++-
 ecomp-c/tests/asm-i386/elf.asm |  3 +++
 3 files changed, 58 insertions(+), 10 deletions(-)

diff --git a/ecomp-c/asm-i386.c b/ecomp-c/asm-i386.c
index 3cb6ff1..a0a65a7 100644
--- a/ecomp-c/asm-i386.c
+++ b/ecomp-c/asm-i386.c
@@ -70,8 +70,8 @@ enum {
 };
 
 static int DEBUG_GETCHAR = 0;
-static int DEBUG_SCANNER = 1;
-static int DEBUG_PARSER = 1;
+static int DEBUG_SCANNER = 0;
+static int DEBUG_PARSER = 0;
 
 /* scanner */
 
@@ -92,6 +92,8 @@ typedef enum {
 	S_minus,
 	S_star,
 	S_slash,
+	S_current_org,
+	S_current_addr,
 	S_eof
 } S_Symbol;
 
@@ -112,6 +114,8 @@ static char *symname[S_eof+1] = {
 	"-",
 	"*",
 	"/",
+	"$$",
+	"$",
 	"eof"
 };
 
@@ -203,6 +207,11 @@ static int getChar( void )
 	return c;
 }
 
+static void ungetChar( int c )
+{
+	ungetc( c, stdin );
+}
+
 static int isWhite( int c )
 {
 	if( c == ' ' || c == '\r' || c == '\t' ) return 1;
@@ -373,8 +382,18 @@ static S_Symbol getSym( void )
 			s = S_newline;
 			break;
 		case '$':
-			hexnumber( );
-			s = S_number;
+			look = getChar( );
+			if( look == '$' ) {
+				s = S_current_org;
+			} else if( ( look >= '0' && look <= '9' ) ||
+				( look >= 'a' && look <= 'f' ) ||
+				( look >= 'A' && look <= 'F' ) ) {
+				ungetChar( look );
+				s = S_number;
+				hexnumber( );
+			} else {
+				s = S_current_addr;
+			}				
 			break;
 		case '\"':
 			string( );
@@ -612,7 +631,9 @@ typedef struct Symbol {
 struct Symbol *symbol = NULL;
 
 enum {
-	ADDRESS_UNDEFINED = 0x7FFFFFFF
+	ADDRESS_UNDEFINED = 0x7FFFFFFF,
+	ADDRESS_CURRENT_ORG = 0x7FFFFFFE,
+	ADDRESS_CURRENT_ADDR = 0x7FFFFFFD
 };
 
 static Symbol *get_symbol( char *name )
@@ -1060,7 +1081,17 @@ static ExpressionNode *parseFactor( void )
 		node = create_expression_node( );
 		node->type = EXPRESSION_NODE_TYPE_VAR;
 		node->symbol = symbol;
-		sym = getSym( );			
+		sym = getSym( );
+	} else if( sym == S_current_org ) {
+		node = create_expression_node( );
+		node->type = EXPRESSION_NODE_TYPE_CONST;
+		node->integer_value = ADDRESS_CURRENT_ORG;
+		sym = getSym( );
+	} else if( sym == S_current_addr ) {
+		node = create_expression_node( );
+		node->type = EXPRESSION_NODE_TYPE_CONST;
+		node->integer_value = ADDRESS_CURRENT_ADDR;
+		sym = getSym( );
 	} else {
 		Abort( "Expected a literal, a variable or a constant." );
 	}
@@ -1105,7 +1136,7 @@ static ExpressionNode *parseExpression( void )
 	return node;
 }
 
-static int evaluate_const_expression_as_integer( ExpressionNode *node )
+static int evaluateExpression( ExpressionNode *node )
 {
 	switch( node->type ) {
 		case EXPRESSION_NODE_TYPE_CONST:
@@ -1138,13 +1169,13 @@ static OperandInfo *parseOperand( OpcodeInfo *opcode_info )
 			assign_label( operand_info, ident );
 		}
 		sym = getSym( );
-	} else if( sym == S_number ) {
+	} else if( sym == S_number || sym == S_current_org || sym == S_current_addr ) {
 		/* absolute operand, like in mov eax, $1,
 		 * we have assemble-time expressions here
 		 */
 		operand_info->type = OPERAND_ABSOLUTE;
 		operand_info->node = parseExpression( );
-		operand_info->num = evaluate_const_expression_as_integer( operand_info->node );
+		operand_info->num = evaluateExpression( operand_info->node );
 		free_expression_node( operand_info->node );
 	} else if( sym == S_string ) {
 		operand_info->type = OPERAND_ABSOLUTE;
diff --git a/ecomp-c/libc-freestanding.c b/ecomp-c/libc-freestanding.c
index ca41ae7..f37aa83 100644
--- a/ecomp-c/libc-freestanding.c
+++ b/ecomp-c/libc-freestanding.c
@@ -399,9 +399,10 @@ typedef struct {
 	int readpos;
 	int size;
 	char buf[INTERNAL_STDIO_BUFSIZE];
+	int pushback;
 } FILE;
 
-static FILE fds[3] = { { STDIN_FILENO, 0, 0, "" }, { STDOUT_FILENO, 0, 0, "" }, { STDERR_FILENO, 0, 0, "" } };
+static FILE fds[3] = { { STDIN_FILENO, 0, 0, "", 0 }, { STDOUT_FILENO, 0, 0, "", 0 }, { STDERR_FILENO, 0, 0, "", 0 } };
 
 FILE *stdin = &fds[0];
 FILE *stdout = &fds[1];
@@ -586,6 +587,12 @@ int fgetc( FILE *stream )
 {	
 	int c;
 	
+	if( stream->pushback != 0 ) {
+		c = stream->pushback;
+		stream->pushback = 0;
+		return c;
+	}
+	
 	if( stream->size == 0 ) {
 		int n = read_string( stream->fileno, stream->buf, INTERNAL_STDIO_BUFSIZE );
 		if( n < 0 ) {
@@ -606,6 +613,13 @@ int fgetc( FILE *stream )
 	return c;
 }
 
+int ungetc( int c, FILE *stream )
+{
+	stream->pushback = c;
+
+	return c;
+}
+
 int getchar( void )
 {
 	return fgetc( stdin );
diff --git a/ecomp-c/tests/asm-i386/elf.asm b/ecomp-c/tests/asm-i386/elf.asm
index e23f61d..3fba61e 100644
--- a/ecomp-c/tests/asm-i386/elf.asm
+++ b/ecomp-c/tests/asm-i386/elf.asm
@@ -2,6 +2,9 @@ format binary
 use32
 org	$08048000
 ehdr:
+mov	eax, $$
+mov	ebx, $
+mov	ecx, $1
 db	$7F, "ELF"	; e_ident: magic
 db	1		; EI_CLASS: ELFCLASS32
 db	1		; EI_BYTE: ELFDATA2LSB (little endian, 2's complement)
-- 
cgit v1.2.3-54-g00ecf