Re: [PATCH] rcar-vin: fix null pointer dereference in rvin_group_get()

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, Apr 26, 2018 at 05:20:05PM +0200, Niklas Söderlund wrote:
> Hi Simon,
> 
> Thanks for your feedback.
> 
> On 2018-04-25 09:18:51 +0200, Simon Horman wrote:
> > On Wed, Apr 25, 2018 at 01:45:06AM +0200, Niklas Söderlund wrote:
> > > Store the group pointer before disassociating the VIN from the group.
> > > 
> > > Fixes: 3bb4c3bc85bf77a7 ("media: rcar-vin: add group allocator functions")
> > > Reported-by: Colin Ian King <colin.king@xxxxxxxxxxxxx>
> > > Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@xxxxxxxxxxxx>
> > > ---
> > >  drivers/media/platform/rcar-vin/rcar-core.c | 12 +++++++-----
> > >  1 file changed, 7 insertions(+), 5 deletions(-)
> > > 
> > > diff --git a/drivers/media/platform/rcar-vin/rcar-core.c b/drivers/media/platform/rcar-vin/rcar-core.c
> > > index 7bc2774a11232362..d3072e166a1ca24f 100644
> > > --- a/drivers/media/platform/rcar-vin/rcar-core.c
> > > +++ b/drivers/media/platform/rcar-vin/rcar-core.c
> > > @@ -338,19 +338,21 @@ static int rvin_group_get(struct rvin_dev *vin)
> > >  
> > >  static void rvin_group_put(struct rvin_dev *vin)
> > >  {
> > > -	mutex_lock(&vin->group->lock);
> > > +	struct rvin_group *group = vin->group;
> > > +
> > > +	mutex_lock(&group->lock);
> > 
> > Hi Niklas, its not clear to me why moving the lock is safe.
> > Could you explain the locking scheme a little?
> 
> The lock here protects the members of the group struct and not any of 
> the members of the vin struct. The intent of the rvin_group_put() 
> function is:
> 
> 1. Disassociate the vin struct from the group struct. This is done by 
>    removing the pointer to the vin from the group->vin array and 
>    removing the pointer from vin->group to the group struct. Here the 
>    lock is needed to protect access to the group->vin array.
> 
> 2. Decrease the refcount of the struct group and if we are the last one 
>    out release the group.
> 
> The problem with the original code is that I first disassociate group 
> from the vin 'vin->group = NULL' but still use the pointer stored in the 
> vin struct when I try to disassociate the vin from the group 
> 'vin->group->vin[vin->id]'.
> 
> AFIK can tell the locking here is fine, the problem was that I pulled 
> the rug from under my own feet in how I access the lock in order to not 
> having to declare a variable to store the pointer in ;-)
> 
> Do this explanation help put you at ease?

Thanks, I am completely relaxed now :)

Reviewed-by: Simon Horman <horms+renesas@xxxxxxxxxxxx>

> > >  	vin->group = NULL;
> > >  	vin->v4l2_dev.mdev = NULL;
> > >  
> > > -	if (WARN_ON(vin->group->vin[vin->id] != vin))
> > > +	if (WARN_ON(group->vin[vin->id] != vin))
> > >  		goto out;
> > >  
> > > -	vin->group->vin[vin->id] = NULL;
> > > +	group->vin[vin->id] = NULL;
> > >  out:
> > > -	mutex_unlock(&vin->group->lock);
> > > +	mutex_unlock(&group->lock);
> > >  
> > > -	kref_put(&vin->group->refcount, rvin_group_release);
> > > +	kref_put(&group->refcount, rvin_group_release);
> > >  }
> > >  
> > >  /* -----------------------------------------------------------------------------
> > > -- 
> > > 2.17.0
> > > 
> 
> -- 
> Regards,
> Niklas Söderlund
> 



[Index of Archives]     [Linux Samsung SOC]     [Linux Wireless]     [Linux Kernel]     [ATH6KL]     [Linux Bluetooth]     [Linux Netdev]     [Kernel Newbies]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Samba]     [Device Mapper]

  Powered by Linux