(no subject)

From: pah@promiscua.org
Date: Mon Jun 24 2002 - 00:49:59 EST


        I've just found a bug (an unsignificant bug) in the panic() function!
        There's a possible buffer overflow if the formated string exceeds
1024 characters (I think that the problem is in all kernel releases).
        The problem is in the use of vsprintf() insted of vsnprintf()!

        I know that this doesn't compromise any exploitation by an uid
different than zero, but should be fixed in the case of panic()'s arguments
exceeds the buffer limit (probably by an lkm or something like that) and
cause (probably) a system crash.

        Here is the exploitation by an lkm:

/************** LKM ***********/

 * panic()'s buffer overflow test
 * By: Pedro Hortas
 * E-Mail: pah@promiscua.org

#define __KERNEL__
#define MODULE


#include <linux/module.h>
#include <linux/kernel.h>

int init_module(void) {
        char foo[2048];
        int i;

        for (i = 0; i < sizeof(foo); i++)
                foo[i] = '\x90';
        foo[i - 1] = 0;
        panic("Overflowing panic()'s buffer: %s\n", foo);

        return 0;

int cleanup_module(void) {
        return 0;

/************* END OF LKM ************/

        And here is the patch to fix the problem:

/************* PATCH **************/

diff -urP linux/kernel/panic.c linux-patched/kernel/panic.c
--- linux/kernel/panic.c Sun Sep 30 20:26:08 2001
+++ linux-patched/kernel/panic.c Mon Jun 24 06:18:12 2002
@@ -51,7 +51,7 @@

        va_start(args, fmt);
- vsprintf(buf, fmt, args);
+ vsnprintf(buf, sizeof(buf), fmt, args);
        printk(KERN_EMERG "Kernel panic: %s\n",buf);
        if (in_interrupt())

/************** END OF PATCH *************/

        Pedro Hortas
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Sun Jun 30 2002 - 22:00:06 EST