2009-10-08 04:48:02

by Stephen Rothwell

[permalink] [raw]
Subject: linux-next: build warnings (buffer size is not provably correct)

Hi all,

Today's linux-next build (i386 defconfig) produced these warnings:

In file included from arch/x86/include/asm/uaccess.h:572,
from kernel/capability.c:18:
arch/x86/include/asm/uaccess_32.h: In function 'sys_capset':
arch/x86/include/asm/uaccess_32.h:211: warning: call to 'copy_from_user_overflow' declared with attribute warning: copy_from_user() buffer size is not provably correct
In file included from arch/x86/include/asm/uaccess.h:572,
from include/linux/poll.h:14,
from include/linux/rtc.h:115,
from include/linux/mc146818rtc.h:16,
from drivers/char/nvram.c:108:
arch/x86/include/asm/uaccess_32.h: In function 'nvram_write':
arch/x86/include/asm/uaccess_32.h:211: warning: call to 'copy_from_user_overflow' declared with attribute warning: copy_from_user() buffer size is not provably correct

Brought to light by commit 4a3127693001c61a21d1ce680db6340623f52e93
("x86: Turn the copy_from_user check into an (optional) compile time
warning") from the tip tree.

Clearly these are not problems with the tip tree, but someone may want to
look at them.

--
Cheers,
Stephen Rothwell [email protected]
http://www.canb.auug.org.au/~sfr/


Attachments:
(No filename) (1.22 kB)
(No filename) (198.00 B)
Download all attachments

2009-10-08 13:42:48

by Arjan van de Ven

[permalink] [raw]
Subject: Re: linux-next: build warnings (buffer size is not provably correct)

On Thu, 8 Oct 2009 15:47:10 +1100
Stephen Rothwell <[email protected]> wrote:

> Hi all,
>
> Today's linux-next build (i386 defconfig) produced these warnings:
>
> In file included from arch/x86/include/asm/uaccess.h:572,
> from kernel/capability.c:18:
> arch/x86/include/asm/uaccess_32.h: In function 'sys_capset':

the following needs to go somewhere... Andrew ?

From: Arjan van de Ven <[email protected]>
Subject: [PATCH 7/9] Simplify bound checks in capabilities for copy_from_user
CC: James Morris <[email protected]>

The capabilities syscall has a copy_from_user() call where gcc currently
cannot prove to itself that the copy is always within bounds.

This patch adds a very explicity bound check to prove to gcc that
this copy_from_user cannot overflow its destination buffer.

Signed-off-by: Arjan van de Ven <[email protected]>
Acked-by: James Morris <[email protected]>

diff --git a/kernel/capability.c b/kernel/capability.c
index 4e17041..204f11f 100644
--- a/kernel/capability.c
+++ b/kernel/capability.c
@@ -238,7 +241,7 @@ SYSCALL_DEFINE2(capget, cap_user_header_t, header, cap_user_data_t, dataptr)
SYSCALL_DEFINE2(capset, cap_user_header_t, header, const cap_user_data_t, data)
{
struct __user_cap_data_struct kdata[_KERNEL_CAPABILITY_U32S];
- unsigned i, tocopy;
+ unsigned i, tocopy, copybytes;
kernel_cap_t inheritable, permitted, effective;
struct cred *new;
int ret;
@@ -255,8 +258,11 @@ SYSCALL_DEFINE2(capset, cap_user_header_t, header, const cap_user_data_t, data)
if (pid != 0 && pid != task_pid_vnr(current))
return -EPERM;

- if (copy_from_user(&kdata, data,
- tocopy * sizeof(struct __user_cap_data_struct)))
+ copybytes = tocopy * sizeof(struct __user_cap_data_struct);
+ if (copybytes > sizeof(kdata))
+ return -EFAULT;
+
+ if (copy_from_user(&kdata, data, copybytes))
return -EFAULT;

for (i = 0; i < tocopy; i++) {


--
Arjan van de Ven Intel Open Source Technology Centre
For development, discussion and tips for power savings,
visit http://www.lesswatts.org