linux-kernel - Re: [PATCH] staging: android: vsoc: fix copy_from

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20190415063904.GH31633@kadam>
Date:   Mon, 15 Apr 2019 09:39:04 +0300
From:   Dan Carpenter <dan.carpenter@...cle.com>
To:     Vincent Stehlé <vincent.stehle@...oste.net>
Cc:     devel@...verdev.osuosl.org, linux-kernel@...r.kernel.org,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Subject: Re: [PATCH] staging: android: vsoc: fix copy_from_user overrun

On Mon, Apr 15, 2019 at 09:32:44AM +0300, Dan Carpenter wrote:
> On Sun, Apr 14, 2019 at 05:37:26PM +0200, Vincent Stehlé wrote:
> > The `np->permission' structure is smaller than the `np' structure but
> > sizeof(*np) worth of data is copied in there. Fix the size passed to
> > copy_from_user() to avoid overrun.
> > 
> > Fixes: 3d2ec9dcd5539d42 ("staging: Android: Add 'vsoc' driver for cuttlefish.")
> > Signed-off-by: Vincent Stehlé <vincent.stehle@...oste.net>
> > Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
> > ---
> >  drivers/staging/android/vsoc.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/staging/android/vsoc.c b/drivers/staging/android/vsoc.c
> > index 8a75bd27c4133..00a1ec7b91549 100644
> > --- a/drivers/staging/android/vsoc.c
> > +++ b/drivers/staging/android/vsoc.c
> > @@ -259,7 +259,8 @@ do_create_fd_scoped_permission(struct vsoc_device_region *region_p,
> >  	atomic_t *owner_ptr = NULL;
> >  	struct vsoc_device_region *managed_region_p;
> >  
> > -	if (copy_from_user(&np->permission, &arg->perm, sizeof(*np)) ||
> > +	if (copy_from_user(&np->permission,
> > +			   &arg->perm, sizeof(np->permission)) ||
> 
> The original code was probably correct...  This is a common thing where
> people use "&p->first_struct_member" to represent the whole struct.
> It seems like kind of a horrible thing to do and I can't explain why
> people do it, but they do...

I have reviewed the code in a less totally lazy way and your patch is
correct.

Reviewed-by: Dan Carpenter <dan.carpenter@...cle.com>

The bug is harmless, though.  We copy over the list pointers with user
data and then immediately write the correct data to it.

It should probably be a static checker warning when we copy to non
__user tagged pointers.  I think someone was adding this to Sparse but I
should probably add it to Smatch as well.

regards,
dan carpenter