[RFC][PATCH] Remove limitations with regards to kernel commandline parsing

[Daniel Mierswa <impulze@impulze.org>]

[-- Attachment #1: Type: text/plain, Size: 1295 bytes --]

Currently there are limitations and imho some weird things with regards to parsing kernel command line parameters. Say if you an initramfs and you want to pass whitespace values to it you may end up with something you did not expect. Passing param="foo" to the kernel makes it (using [] notation) [param[foo]], so far so good. Passing real_root=LABEL="abc def" makes it [real_root[LABEL="abc def"]] which is not quite what you want, you'd expect [real_rooŧ[LABEL=abc def]]. Hence my patch (git f-p against master of linus.git is attached) tries to remove those limitations.

> There was a limitation for kernel parameters with regards to quoting. It
> wasn't possible to escape quotes or use quotes to form space-filled
> values _inside_ parameters. This patch attempts to make that possible,
> kernel parameters are now parsed as follows:
> '"param= value"' [param= value][]
> 'param=" value "" combination "' [param][ value  combination ]
> 'param=" \" test"' [param][ " test]
> '"param"=another' [param][another]

CC'ed rusty@rustcorp.com.au since he shows up quite frequent in git log of kernel/params.c

-- 
Mierswa, Daniel

If you still don't like it, that's ok: that's why I'm boss. I simply know better than you do.
               --- Linus Torvalds, comp.os.linux.advocacy, 1996/07/22

[-- Attachment #2: kernel.patch --]
[-- Type: text/plain, Size: 4318 bytes --]

From d8b4272241a4eec1d426fdfcce7fc4fcfe65c284 Mon Sep 17 00:00:00 2001
From: Daniel Mierswa <impulze@impulze.org>
Date: Tue, 7 Jul 2009 00:54:38 +0200
Subject: [PATCH] Attempt to handle quotes in kernel parameters

There was a limitation for kernel parameters with regards to quoting. It
wasn't possible to escape quotes or use quotes to form space-filled
values _inside_ parameters. This patch attempts to make that possible,
kernel parameters are now parsed as follows:
'"param= value"' [param= value][]
'param=" value "" combination "' [param][ value  combination ]
'param=" \" test"' [param][ " test]
'"param"=another' [param][another]

Signed-off-by: Daniel Mierswa <impulze@impulze.org>
---
 kernel/params.c |  144 +++++++++++++++++++++++++++++++++++++++----------------
 1 files changed, 103 insertions(+), 41 deletions(-)

diff --git a/kernel/params.c b/kernel/params.c
index 7f6912c..e6041c0 100644
--- a/kernel/params.c
+++ b/kernel/params.c
@@ -72,58 +72,120 @@ static int parse_one(char *param,
 	return -ENOENT;
 }
 
-/* You can use " around spaces, but can't escape ". */
-/* Hyphens and underscores equivalent in parameter names. */
-static char *next_arg(char *args, char **param, char **val)
+/* handle quotes in tokens (parameter and values)
+ * '" foo bar "' => ' foo bar '
+ * '" foo \" "'  => ' foo " '
+ */
+static void add_token(char ** token, char * args)
 {
-	unsigned int i, equals = 0;
-	int in_quote = 0, quoted = 0;
-	char *next;
+	char * iterator, * last_quote;
+	int in_quotes;
 
-	if (*args == '"') {
-		args++;
-		in_quote = 1;
-		quoted = 1;
+	in_quotes = 0;
+	last_quote = NULL;
+
+	for (iterator = args; *iterator; iterator++) {
+		if (*iterator == '\\' && *(iterator + 1) == '"') {
+			char * mover;
+
+			/* move all characters back */
+			for (mover = iterator; *mover; mover++) {
+				*mover = *(mover + 1);
+			}
+
+			continue;
+		}
+
+		if (*iterator != '"') {
+			continue;
+		}
+
+		if (in_quotes) {
+			char * mover;
+
+			/* move whole string back until current " is reached */
+			for (mover = last_quote; mover != iterator - 1; mover++) {
+				*mover = *(mover + 1);
+			}
+
+			/* ignore the current " and move the rest of the string back */
+			while (*mover) {
+				*mover = *(mover + 2);
+				mover++;
+			}
+
+			/* ignored 2 quotes, decrease the iterator */
+			iterator -= 2;
+			last_quote = NULL;
+		} else {
+			last_quote = iterator;
+		}
+
+		in_quotes = !in_quotes;
 	}
 
-	for (i = 0; args[i]; i++) {
-		if (args[i] == ' ' && !in_quote)
+	*token = args;
+}
+
+static char * next_arg(char * args, char ** param, char ** val)
+{
+	char * token, * next;
+	int in_quotes, is_escaped;
+
+	*param = *val = next = NULL;
+	in_quotes = is_escaped = 0;
+
+	/* tokenizer */
+	for (token = args; *token; token++) {
+		/* parameter or value is finished */
+		if (!in_quotes && *token == ' ') {
+			*token = '\0';
+
+			/* there're still characters in the parameter line */
+			next = token + 1;
+
+			/* remove trailing whitespace */
+			while (*next == ' ') {
+				next++;
+			}
+
 			break;
-		if (equals == 0) {
-			if (args[i] == '=')
-				equals = i;
 		}
-		if (args[i] == '"')
-			in_quote = !in_quote;
-	}
 
-	*param = args;
-	if (!equals)
-		*val = NULL;
-	else {
-		args[equals] = '\0';
-		*val = args + equals + 1;
-
-		/* Don't include quotes in value. */
-		if (**val == '"') {
-			(*val)++;
-			if (args[i-1] == '"')
-				args[i-1] = '\0';
+		/* parameter/value split */
+		if (!in_quotes && *token == '=' && !*param) {
+			*token = '\0';
+			add_token(param, args);
+			args = token + 1;
+			continue;
 		}
-		if (quoted && args[i-1] == '"')
-			args[i-1] = '\0';
+
+		if (!is_escaped && *token == '\\') {
+			is_escaped = 1;
+			continue;
+		}
+
+		if (*token == '"' && !is_escaped) {
+			in_quotes = !in_quotes;
+		}
+
+		/* always reset escape value, only " needs it */
+		is_escaped = 0;
 	}
 
-	if (args[i]) {
-		args[i] = '\0';
-		next = args + i + 1;
-	} else
-		next = args + i;
+	if (!*param) {
+		add_token(param, args);
+	} else {
+		add_token(val, args);
+	}
+
+	/* there're parameters left in the command line */
+	if (next) {
+		return next;
+	}
 
-	/* Chew up trailing spaces. */
-	while (*next == ' ')
-		next++;
-	return next;
+	/* end of the line */
+	return token;
 }
 
 /* Args looks like "foo=bar,bar2 baz=fuz wiz". */
-- 
1.6.3.3