[PATCH] Fix user struct leakage with locked IPC shem segment

[PATCH] Fix user struct leakage with locked IPC shem segment

[Pavel Emelianov]

When user locks an ipc shmem segmant with SHM_LOCK ctl and the
segment is already locked the shmem_lock() function returns 0. 
After this the subsequent code leaks the existing user struct:

== ipc/shm.c: sys_shmctl() ==
     ...
     err = shmem_lock(shp->shm_file, 1, user);
     if (!err) {
          shp->shm_perm.mode |= SHM_LOCKED;
          shp->mlock_user = user;
     }
     ...
==

Other results of this are:
1. the new shp->mlock_user is not get-ed and will point to freed
   memory when the task dies.
2. the RLIMIT_MEMLOCK is screwed on both user structs.

The exploit looks like this:

==
    id = shmget(...);
    setresuid(uid, 0, 0);
    shmctl(id, SHM_LOCK, NULL);
    setresuid(uid + 1, 0, 0);
    shmctl(id, SHM_LOCK, NULL);
==

My solution is to return 0 to the userspace and do not change the
segment's user.

Signed-off-by: Pavel Emelianov <xemul@openvz.org>

---

--- ./ipc/shm.c.shlfix	2007-07-06 10:58:57.000000000 +0400
+++ ./ipc/shm.c	2007-07-16 16:12:34.000000000 +0400
@@ -715,7 +715,7 @@ asmlinkage long sys_shmctl (int shmid, i
 			struct user_struct * user = current->user;
 			if (!is_file_hugepages(shp->shm_file)) {
 				err = shmem_lock(shp->shm_file, 1, user);
-				if (!err) {
+				if (!err && !(shp->shm_perm.mode & SHM_LOCKED)){
 					shp->shm_perm.mode |= SHM_LOCKED;
 					shp->mlock_user = user;
 				}

Re: [PATCH] Fix user struct leakage with locked IPC shem segment

[Andrew Morton]

On Mon, 16 Jul 2007 16:24:12 +0400
Pavel Emelianov <xemul@openvz.org> wrote:

> When user locks an ipc shmem segmant with SHM_LOCK ctl and the
> segment is already locked the shmem_lock() function returns 0. 
> After this the subsequent code leaks the existing user struct:

I'm curious.  For the past few months, people@openvz.org have discovered
(and fixed) an ongoing stream of obscure but serious and quite
long-standing bugs.

How are you discovering these bugs?

> == ipc/shm.c: sys_shmctl() ==
>      ...
>      err = shmem_lock(shp->shm_file, 1, user);
>      if (!err) {
>           shp->shm_perm.mode |= SHM_LOCKED;
>           shp->mlock_user = user;
>      }
>      ...
> ==
> 
> Other results of this are:
> 1. the new shp->mlock_user is not get-ed and will point to freed
>    memory when the task dies.

That sounds fairly serious - can this lead to memory corruption and crashes?

> 2. the RLIMIT_MEMLOCK is screwed on both user structs.
> 
> The exploit looks like this:
> 
> ==
>     id = shmget(...);
>     setresuid(uid, 0, 0);
>     shmctl(id, SHM_LOCK, NULL);
>     setresuid(uid + 1, 0, 0);
>     shmctl(id, SHM_LOCK, NULL);
> ==
> 
> My solution is to return 0 to the userspace and do not change the
> segment's user.
> 
> Signed-off-by: Pavel Emelianov <xemul@openvz.org>
> 
> ---
> 
> --- ./ipc/shm.c.shlfix	2007-07-06 10:58:57.000000000 +0400
> +++ ./ipc/shm.c	2007-07-16 16:12:34.000000000 +0400
> @@ -715,7 +715,7 @@ asmlinkage long sys_shmctl (int shmid, i
>  			struct user_struct * user = current->user;
>  			if (!is_file_hugepages(shp->shm_file)) {
>  				err = shmem_lock(shp->shm_file, 1, user);
> -				if (!err) {
> +				if (!err && !(shp->shm_perm.mode & SHM_LOCKED)){
>  					shp->shm_perm.mode |= SHM_LOCKED;
>  					shp->mlock_user = user;
>  				}

Re: [Devel] Re: [PATCH] Fix user struct leakage with locked IPC shem segment

[Kirill Korotaev]

Andrew Morton wrote:
> On Mon, 16 Jul 2007 16:24:12 +0400
> Pavel Emelianov <xemul@openvz.org> wrote:
> 
> 
>>When user locks an ipc shmem segmant with SHM_LOCK ctl and the
>>segment is already locked the shmem_lock() function returns 0. 
>>After this the subsequent code leaks the existing user struct:
> 
> 
> I'm curious.  For the past few months, people@openvz.org have discovered
> (and fixed) an ongoing stream of obscure but serious and quite
> long-standing bugs.

thanks a lot :@)

> How are you discovering these bugs?

Not sure what to answer :) Just trying to do our best.

This bug was thought over by Pavel for about 3 month after a single
uid leak in container was detected by beancounters' kernel memory accounting...

>>== ipc/shm.c: sys_shmctl() ==
>>     ...
>>     err = shmem_lock(shp->shm_file, 1, user);
>>     if (!err) {
>>          shp->shm_perm.mode |= SHM_LOCKED;
>>          shp->mlock_user = user;
>>     }
>>     ...
>>==
>>
>>Other results of this are:
>>1. the new shp->mlock_user is not get-ed and will point to freed
>>   memory when the task dies.
> 
> 
> That sounds fairly serious - can this lead to memory corruption and crashes?

Yes it can. According to Pavel when the shmem segment is destroyed it
puts the mlock_user pointer, which can already be stalled.

Kirill

Re: [Devel] Re: [PATCH] Fix user struct leakage with locked IPC shem segment

[Andrew Morton]

On Tue, 17 Jul 2007 13:07:55 +0400 Kirill Korotaev <dev@sw.ru> wrote:

> Andrew Morton wrote:
> > On Mon, 16 Jul 2007 16:24:12 +0400
> > Pavel Emelianov <xemul@openvz.org> wrote:
> > 
> > 
> >>When user locks an ipc shmem segmant with SHM_LOCK ctl and the
> >>segment is already locked the shmem_lock() function returns 0. 
> >>After this the subsequent code leaks the existing user struct:
> > 
> > 
> > I'm curious.  For the past few months, people@openvz.org have discovered
> > (and fixed) an ongoing stream of obscure but serious and quite
> > long-standing bugs.
> 
> thanks a lot :@)
> 
> > How are you discovering these bugs?
> 
> Not sure what to answer :) Just trying to do our best.

hm, OK, I was visualising some mysterious Russian bugfinding machine or
something.

Don't stop ;)

> This bug was thought over by Pavel for about 3 month after a single
> uid leak in container was detected by beancounters' kernel memory accounting...
> 
> >>== ipc/shm.c: sys_shmctl() ==
> >>     ...
> >>     err = shmem_lock(shp->shm_file, 1, user);
> >>     if (!err) {
> >>          shp->shm_perm.mode |= SHM_LOCKED;
> >>          shp->mlock_user = user;
> >>     }
> >>     ...
> >>==
> >>
> >>Other results of this are:
> >>1. the new shp->mlock_user is not get-ed and will point to freed
> >>   memory when the task dies.
> > 
> > 
> > That sounds fairly serious - can this lead to memory corruption and crashes?
> 
> Yes it can. According to Pavel when the shmem segment is destroyed it
> puts the mlock_user pointer, which can already be stalled.

OK, thanks, I'll feed a copy in stable@kernel.org's direction.