m68k: fix access_ok for coldfire
authorArnd Bergmann <arnd@arndb.de>
Tue, 15 Feb 2022 16:59:41 +0000 (17:59 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Mon, 28 Mar 2022 07:58:46 +0000 (09:58 +0200)
commit 26509034bef198525d5936c116cbd0c3fa491c0b upstream.

While most m68k platforms use separate address spaces for user
and kernel space, at least coldfire does not, and the other
ones have a TASK_SIZE that is less than the entire 4GB address
range.

Using the default implementation of __access_ok() stops coldfire
user space from trivially accessing kernel memory.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: stable@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/m68k/include/asm/uaccess.h

index ba67052..60b786e 100644 (file)
 #include <asm/extable.h>
 
 /* We let the MMU do all checking */
-static inline int access_ok(const void __user *addr,
+static inline int access_ok(const void __user *ptr,
                            unsigned long size)
 {
-       /*
-        * XXX: for !CONFIG_CPU_HAS_ADDRESS_SPACES this really needs to check
-        * for TASK_SIZE!
-        */
-       return 1;
+       unsigned long limit = TASK_SIZE;
+       unsigned long addr = (unsigned long)ptr;
+
+       if (IS_ENABLED(CONFIG_CPU_HAS_ADDRESS_SPACES) ||
+           !IS_ENABLED(CONFIG_MMU))
+               return 1;
+
+       return (size <= limit) && (addr <= (limit - size));
 }
 
 /*