[PATCH] bootconfig: Fix off-by-one in xbc_node_compose_key_after()

[PATCH] bootconfig: Fix off-by-one in xbc_node_compose_key_after()

[Steven Rostedt]

From: Steven Rostedt (VMware) <rostedt@goodmis.org>

While reviewing some patches for bootconfig, I noticed the following
code in xbc_node_compose_key_after():

	ret = snprintf(buf, size, "%s%s", xbc_node_get_data(node),
		       depth ? "." : "");
	if (ret < 0)
		return ret;
	if (ret > size) {
		size = 0;
	} else {
		size -= ret;
		buf += ret;
	}

But snprintf() returns the number of bytes that would be written, not
the number of bytes that are written (ignoring the nul terminator).
This means that if the number of non null bytes written were to equal
size, then the nul byte, which snprintf() always adds, will overwrite
that last byte.

	ret = snprintf(buf, 5, "hello");
	printf("buf = '%s'\n", buf);
	printf("ret = %d\n", ret);

produces:

	buf = 'hell'
	ret = 5

The string was truncated without ret being greater than 5.
Test (ret >= size) for overwrite.

Cc: stable@vger.kernel.org
Fixes: 76db5a27a827c ("bootconfig: Add Extra Boot Config support")
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
---
diff --git a/lib/bootconfig.c b/lib/bootconfig.c
index a5f701161f6b..42735f2b8860 100644
--- a/lib/bootconfig.c
+++ b/lib/bootconfig.c
@@ -248,7 +248,7 @@ int __init xbc_node_compose_key_after(struct xbc_node *root,
 			       depth ? "." : "");
 		if (ret < 0)
 			return ret;
-		if (ret > size) {
+		if (ret >= size) {
 			size = 0;
 		} else {
 			size -= ret;

Re: [PATCH] bootconfig: Fix off-by-one in xbc_node_compose_key_after()

[Andrew Morton]

On Thu, 13 Aug 2020 18:30:50 -0400 Steven Rostedt <rostedt@goodmis.org> wrote:

> From: Steven Rostedt (VMware) <rostedt@goodmis.org>
> 
> While reviewing some patches for bootconfig, I noticed the following
> code in xbc_node_compose_key_after():
> 
> 	ret = snprintf(buf, size, "%s%s", xbc_node_get_data(node),
> 		       depth ? "." : "");
> 	if (ret < 0)
> 		return ret;
> 	if (ret > size) {
> 		size = 0;
> 	} else {
> 		size -= ret;
> 		buf += ret;
> 	}
> 
> But snprintf() returns the number of bytes that would be written, not
> the number of bytes that are written (ignoring the nul terminator).
> This means that if the number of non null bytes written were to equal
> size, then the nul byte, which snprintf() always adds, will overwrite
> that last byte.
> 
> 	ret = snprintf(buf, 5, "hello");
> 	printf("buf = '%s'\n", buf);
> 	printf("ret = %d\n", ret);
> 
> produces:
> 
> 	buf = 'hell'
> 	ret = 5
> 
> The string was truncated without ret being greater than 5.
> Test (ret >= size) for overwrite.

What are the end-user visible effects of the bug?  IOW, why cc:stable?

Re: [PATCH] bootconfig: Fix off-by-one in xbc_node_compose_key_after()

[Steven Rostedt]

On Thu, 13 Aug 2020 19:38:18 -0700
Andrew Morton <akpm@linux-foundation.org> wrote:

> On Thu, 13 Aug 2020 18:30:50 -0400 Steven Rostedt <rostedt@goodmis.org> wrote:
> 
> > From: Steven Rostedt (VMware) <rostedt@goodmis.org>
> > 
> > While reviewing some patches for bootconfig, I noticed the following
> > code in xbc_node_compose_key_after():
> > 
> > 	ret = snprintf(buf, size, "%s%s", xbc_node_get_data(node),
> > 		       depth ? "." : "");
> > 	if (ret < 0)
> > 		return ret;
> > 	if (ret > size) {
> > 		size = 0;
> > 	} else {
> > 		size -= ret;
> > 		buf += ret;
> > 	}
> > 
> > But snprintf() returns the number of bytes that would be written, not
> > the number of bytes that are written (ignoring the nul terminator).
> > This means that if the number of non null bytes written were to equal
> > size, then the nul byte, which snprintf() always adds, will overwrite
> > that last byte.
> > 
> > 	ret = snprintf(buf, 5, "hello");
> > 	printf("buf = '%s'\n", buf);
> > 	printf("ret = %d\n", ret);
> > 
> > produces:
> > 
> > 	buf = 'hell'
> > 	ret = 5
> > 
> > The string was truncated without ret being greater than 5.
> > Test (ret >= size) for overwrite.  
> 
> What are the end-user visible effects of the bug?  IOW, why cc:stable?
> 

Hmm, looking at it at a wider view, it may not be an issue. The tools
code calls this code, and I looked to see if it was possible to corrupt
the buffer by an incorrect size. But now that I'm looking at the else
part of the section, it may not be a problem as it may act the same.

That is, ret == size will make size = 0 with the size -= ret, and we
get the same result.

OK, you can drop the patch. Thanks for the review!

Although, there's no error message if the buffer is not big enough to
hold the fields.

Masami?

-- Steve

Re: [PATCH] bootconfig: Fix off-by-one in xbc_node_compose_key_after()

[Masami Hiramatsu]

On Thu, 13 Aug 2020 23:04:06 -0400
Steven Rostedt <rostedt@goodmis.org> wrote:

> On Thu, 13 Aug 2020 19:38:18 -0700
> Andrew Morton <akpm@linux-foundation.org> wrote:
> 
> > On Thu, 13 Aug 2020 18:30:50 -0400 Steven Rostedt <rostedt@goodmis.org> wrote:
> > 
> > > From: Steven Rostedt (VMware) <rostedt@goodmis.org>
> > > 
> > > While reviewing some patches for bootconfig, I noticed the following
> > > code in xbc_node_compose_key_after():
> > > 
> > > 	ret = snprintf(buf, size, "%s%s", xbc_node_get_data(node),
> > > 		       depth ? "." : "");
> > > 	if (ret < 0)
> > > 		return ret;
> > > 	if (ret > size) {
> > > 		size = 0;
> > > 	} else {
> > > 		size -= ret;
> > > 		buf += ret;
> > > 	}
> > > 
> > > But snprintf() returns the number of bytes that would be written, not
> > > the number of bytes that are written (ignoring the nul terminator).
> > > This means that if the number of non null bytes written were to equal
> > > size, then the nul byte, which snprintf() always adds, will overwrite
> > > that last byte.
> > > 
> > > 	ret = snprintf(buf, 5, "hello");
> > > 	printf("buf = '%s'\n", buf);
> > > 	printf("ret = %d\n", ret);
> > > 
> > > produces:
> > > 
> > > 	buf = 'hell'
> > > 	ret = 5
> > > 
> > > The string was truncated without ret being greater than 5.
> > > Test (ret >= size) for overwrite.  
> > 
> > What are the end-user visible effects of the bug?  IOW, why cc:stable?
> > 
> 
> Hmm, looking at it at a wider view, it may not be an issue. The tools
> code calls this code, and I looked to see if it was possible to corrupt
> the buffer by an incorrect size. But now that I'm looking at the else
> part of the section, it may not be a problem as it may act the same.
> 
> That is, ret == size will make size = 0 with the size -= ret, and we
> get the same result.
> 
> OK, you can drop the patch. Thanks for the review!

Thanks Andrew, you're right. If size == ret, then size -= ret makes
 size = 0 and after that, the loop doesn't change the size. 

> 
> Although, there's no error message if the buffer is not big enough to
> hold the fields.
> 
> Masami?

Ah, I think we need a patch to fix a comment of the function. It says

 * Returns the total length of the key stored in @buf.

But this must be

  * Returns the total bytes of the key which would be written.

As similar to the snprintf(), so that the user can compare the result
with the given buffer size.

Thank you,

-- 
Masami Hiramatsu <mhiramat@kernel.org>