ddb(4): Enhance lexer functionality for specialized commands

Add a db_read_token_flags() variant of db_read_token() with configurable
parameters.

Allow specifying an explicit radix for tNUMBER lexing.  It overrides the
default inference and db_radix setting.

Also provide the option of yielding any lexed whitespace (tWSPACE) (instead
of ignoring it).  This is useful for whitespace-sensitive CS_OWN commands.

Reviewed by:	markj
Differential Revision:	https://reviews.freebsd.org/D21459

2 files changed, 75 insertions, 10 deletions

diff --git a/sys/ddb/db_lex.c b/sys/ddb/db_lex.c
index 8e25cc898fc1..5b81652bc1f1 100644
--- a/sys/ddb/db_lex.c
+++ b/sys/ddb/db_lex.c
@@ -38,6 +38,7 @@ __FBSDID("$FreeBSD$");
 
 #include <sys/param.h>
 #include <sys/libkern.h>
+#include <sys/lock.h>
 
 #include <ddb/ddb.h>
 #include <ddb/db_lex.h>
@@ -45,7 +46,7 @@ __FBSDID("$FreeBSD$");
 static char	db_line[DB_MAXLINE];
 static char *	db_lp, *db_endlp;
 
-static int	db_lex(void);
+static int	db_lex(int);
 static void 	db_flush_line(void);
 static int 	db_read_char(void);
 static void 	db_unread_char(int);
@@ -124,23 +125,24 @@ db_unread_char(c)
 static int	db_look_token = 0;
 
 void
-db_unread_token(t)
-	int	t;
+db_unread_token(int t)
 {
 	db_look_token = t;
 }
 
 int
-db_read_token()
+db_read_token_flags(int flags)
 {
 	int	t;
 
+	MPASS((flags & ~(DRT_VALID_FLAGS_MASK)) == 0);
+
 	if (db_look_token) {
 	    t = db_look_token;
 	    db_look_token = 0;
 	}
 	else
-	    t = db_lex();
+	    t = db_lex(flags);
 	return (t);
 }
 
@@ -158,22 +160,46 @@ db_flush_lex(void)
 }
 
 static int
-db_lex(void)
+db_lex(int flags)
 {
-	int	c;
+	int	c, n, radix_mode;
+	bool	lex_wspace;
+
+	switch (flags & DRT_RADIX_MASK) {
+	case DRT_DEFAULT_RADIX:
+		radix_mode = -1;
+		break;
+	case DRT_OCTAL:
+		radix_mode = 8;
+		break;
+	case DRT_DECIMAL:
+		radix_mode = 10;
+		break;
+	case DRT_HEXADECIMAL:
+		radix_mode = 16;
+		break;
+	}
+
+	lex_wspace = ((flags & DRT_WSPACE) != 0);
 
 	c = db_read_char();
-	while (c <= ' ' || c > '~') {
+	for (n = 0; c <= ' ' || c > '~'; n++) {
 	    if (c == '\n' || c == -1)
 		return (tEOL);
 	    c = db_read_char();
 	}
+	if (lex_wspace && n != 0) {
+	    db_unread_char(c);
+	    return (tWSPACE);
+	}
 
 	if (c >= '0' && c <= '9') {
 	    /* number */
 	    int	r, digit = 0;
 
-	    if (c > '0')
+	    if (radix_mode != -1)
+		r = radix_mode;
+	    else if (c > '0')
 		r = db_radix;
 	    else {
 		c = db_read_char();
diff --git a/sys/ddb/db_lex.h b/sys/ddb/db_lex.h
index e9ed3a660393..3dbb42575533 100644
--- a/sys/ddb/db_lex.h
+++ b/sys/ddb/db_lex.h
@@ -38,13 +38,51 @@
 /*
  * Lexical analyzer.
  */
+
+/*
+ * Options and flags can configure db_read_token() => db_lex() behavior.
+ *
+ * When a radix other than DRT_DEFAULT_RADIX is used, it overrides
+ * auto-detection, as well as the user-specified db_radix, in db_lex() of
+ * 'tNUMBER' tokens.
+ */
+enum {
+	/* Infer or use db_radix using the old logic. */
+	DRT_DEFAULT_RADIX = 0,
+	/* The following set an explicit base for tNUMBER lex. */
+	DRT_OCTAL,
+	DRT_DECIMAL,
+	DRT_HEXADECIMAL,
+};
+#define	DRT_RADIX_MASK	0x3
+/*
+ * Flag bit powers of two for db_read_token_flags.
+ * The low 2 bits are reserved for radix selection.
+ */
+enum {
+	_DRT_WSPACE = 2,
+};
+#ifndef BIT
+#define	BIT(n)	(1ull << (n))
+#endif
+enum {
+	DRT_WSPACE = BIT(_DRT_WSPACE),
+};
+#define	DRT_VALID_FLAGS_MASK	((int)DRT_RADIX_MASK | DRT_WSPACE)
+
 void	 db_flush_lex(void);
 char	*db_get_line(void);
 void	 db_inject_line(const char *command);
 int	 db_read_line(void);
-int	 db_read_token(void);
+int	 db_read_token_flags(int);
 void	 db_unread_token(int t);
 
+static inline int
+db_read_token(void)
+{
+	return (db_read_token_flags(0));
+}
+
 extern db_expr_t	db_tok_number;
 #define	TOK_STRING_SIZE		120
 extern char	db_tok_string[TOK_STRING_SIZE];
@@ -84,5 +122,6 @@ extern char	db_tok_string[TOK_STRING_SIZE];
 #define	tSTRING		32
 #define	tQUESTION	33
 #define	tBIT_NOT	34
+#define	tWSPACE		35
 
 #endif /* !_DDB_DB_LEX_H_ */