Re: net: socket: NULL ptr deref in sendmsg

[Hannes Frederic Sowa <hannes@stressinduktion.org>]

On Fr, 2014-07-25 at 19:23 +0400, Andrey Ryabinin wrote:
> On 07/14/14 01:50, Sasha Levin wrote:
> 
> > 
> > I've tried debugging it, but I don't see a code path that could lead to that.
> > 
> 
> I finally found some time to take look at this and I've found where the problem is.
> 
> Sasha, I suppose there was no usual "Unable to handle NULL pointer deference" after KASAN's report, right?
> 
> This gave me a clue that address 0 is actually mapped and contains valid socket address structure in it.
> I've managed to write a simple code (in attachment), which could easily reproduce this bug.
> 
> I've fixed it with the following patch, please take a look.
> 
> 
> From: Andrey Ryabinin <a.ryabinin@samsung.com>
> Subject: [PATCH] net: sendmsg: fix NULL pointer dereference
> 
> Sasha's report:
> 	> While fuzzing with trinity inside a KVM tools guest running the latest -next
> 	> kernel with the KASAN patchset, I've stumbled on the following spew:
> 	>
> 	> [ 4448.949424] ==================================================================
> 	> [ 4448.951737] AddressSanitizer: user-memory-access on address 0
> 	> [ 4448.952988] Read of size 2 by thread T19638:
> 	> [ 4448.954510] CPU: 28 PID: 19638 Comm: trinity-c76 Not tainted 3.16.0-rc4-next-20140711-sasha-00046-g07d3099-dirty #813
> 	> [ 4448.956823]  ffff88046d86ca40 0000000000000000 ffff880082f37e78 ffff880082f37a40
> 	> [ 4448.958233]  ffffffffb6e47068 ffff880082f37a68 ffff880082f37a58 ffffffffb242708d
> 	> [ 4448.959552]  0000000000000000 ffff880082f37a88 ffffffffb24255b1 0000000000000000
> 	> [ 4448.961266] Call Trace:
> 	> [ 4448.963158] dump_stack (lib/dump_stack.c:52)
> 	> [ 4448.964244] kasan_report_user_access (mm/kasan/report.c:184)
> 	> [ 4448.965507] __asan_load2 (mm/kasan/kasan.c:352)
> 	> [ 4448.966482] ? netlink_sendmsg (net/netlink/af_netlink.c:2339)
> 	> [ 4448.967541] netlink_sendmsg (net/netlink/af_netlink.c:2339)
> 	> [ 4448.968537] ? get_parent_ip (kernel/sched/core.c:2555)
> 	> [ 4448.970103] sock_sendmsg (net/socket.c:654)
> 	> [ 4448.971584] ? might_fault (mm/memory.c:3741)
> 	> [ 4448.972526] ? might_fault (./arch/x86/include/asm/current.h:14 mm/memory.c:3740)
> 	> [ 4448.973596] ? verify_iovec (net/core/iovec.c:64)
> 	> [ 4448.974522] ___sys_sendmsg (net/socket.c:2096)
> 	> [ 4448.975797] ? put_lock_stats.isra.13 (./arch/x86/include/asm/preempt.h:98 kernel/locking/lockdep.c:254)
> 	> [ 4448.977030] ? lock_release_holdtime (kernel/locking/lockdep.c:273)
> 	> [ 4448.978197] ? lock_release_non_nested (kernel/locking/lockdep.c:3434 (discriminator 1))
> 	> [ 4448.979346] ? check_chain_key (kernel/locking/lockdep.c:2188)
> 	> [ 4448.980535] __sys_sendmmsg (net/socket.c:2181)
> 	> [ 4448.981592] ? trace_hardirqs_on_caller (kernel/locking/lockdep.c:2600)
> 	> [ 4448.982773] ? trace_hardirqs_on (kernel/locking/lockdep.c:2607)
> 	> [ 4448.984458] ? syscall_trace_enter (arch/x86/kernel/ptrace.c:1500 (discriminator 2))
> 	> [ 4448.985621] ? trace_hardirqs_on_caller (kernel/locking/lockdep.c:2600)
> 	> [ 4448.986754] SyS_sendmmsg (net/socket.c:2201)
> 	> [ 4448.987708] tracesys (arch/x86/kernel/entry_64.S:542)
> 	> [ 4448.988929] ==================================================================
> 
> This reports means that we've come to netlink_sendmsg() with msg->msg_name == NULL and msg->msg_namelen > 0.
> 
> After this report there was no usual "Unable to handle kernel NULL pointer dereference"
> and this gave me a clue that address 0 is mapped and contains valid socket address structure in it.
> 
> This bug was introduced in f3d3342602f8bcbf37d7c46641cb9bca7618eb1c
> (net: rework recvmsg handler msg_name and msg_namelen logic).
> Commit message states that:
> 	"Set msg->msg_name = NULL if user specified a NULL in msg_name but had a
> 	 non-null msg_namelen in verify_iovec/verify_compat_iovec. This doesn't
> 	 affect sendto as it would bail out earlier while trying to copy-in the
> 	 address."
> But in fact this affects sendto when address 0 is mapped and contains
> socket address structure in it. In such case copy-in address will succeed,
> verify_iovec() function will successfully exit with msg->msg_namelen > 0
> and msg->msg_name == NULL.
> 
> This patch fixes it by assigning m->msg_name to address if move_addr_to_kernel()
> was successful.
> 
> Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
> Cc: Eric Dumazet <edumazet@google.com>
> Cc: <stable@vger.kernel.org>
> Reported-by: Sasha Levin <sasha.levin@oracle.com>
> Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com>
> ---
>  net/compat.c     | 6 ++++--
>  net/core/iovec.c | 6 ++++--
>  2 files changed, 8 insertions(+), 4 deletions(-)
> 
> diff --git a/net/compat.c b/net/compat.c
> index 9a76eaf..eefd989 100644
> --- a/net/compat.c
> +++ b/net/compat.c
> @@ -92,9 +92,11 @@ int verify_compat_iovec(struct msghdr *kern_msg, struct iovec *kern_iov,
>  						      kern_address);
>  			if (err < 0)
>  				return err;
> -		}
> -		if (kern_msg->msg_name)
> +
>  			kern_msg->msg_name = kern_address;
> +		} else
> +			if (kern_msg->msg_name)
> +				kern_msg->msg_name = kern_address;
>  	} else
>  		kern_msg->msg_name = NULL;
> 

Thanks for looking at this! I certainly have overlooked this case.

I wonder, if we allow sendto with valid NULL pointer and positive
msg_namelen to work, why don't we do the same for recvmsg, as in
replacing the VERIFY_WRITE case non-null check with an
access_ok(VERIFY_WRITE, ...) check? We can even get rid of the check
because recvmsg will do it nonetheless when copying the sockaddrs to
user space. So kern_msg->msg_name can always be set to kern_address.

Otherwise I would just set msg_namelen = 0, too, and just not handle
passed in NULL pointers to sockaddrs.

Thanks,
Hannes