On Thu, Oct 22, 2020 at 06:01:27PM +0800, Zhuoliang Zhang wrote:
> From: zhuoliang zhang <zhuoliang.zh...@mediatek.com>
>
> we found that the following race condition exists in
> xfrm_alloc_userspi flow:
>
> user thread state_hash_work thread
> ---- ----
> xfrm_alloc_userspi()
> __find_acq_core()
> /*alloc new xfrm_state:x*/
> xfrm_state_alloc()
> /*schedule state_hash_work thread*/
> xfrm_hash_grow_check() xfrm_hash_resize()
> xfrm_alloc_spi /*hold lock*/
> x->id.spi = htonl(spi)
> spin_lock_bh(&net->xfrm.xfrm_state_lock)
> /*waiting lock release*/ xfrm_hash_transfer()
> spin_lock_bh(&net->xfrm.xfrm_state_lock) /*add x into
> hlist:net->xfrm.state_byspi*/
>
> hlist_add_head_rcu(&x->byspi)
>
> spin_unlock_bh(&net->xfrm.xfrm_state_lock)
>
> /*add x into hlist:net->xfrm.state_byspi 2 times*/
> hlist_add_head_rcu(&x->byspi)
>
> 1. a new state x is alloced in xfrm_state_alloc() and added into the bydst
> hlist
> in __find_acq_core() on the LHS;
> 2. on the RHS, state_hash_work thread travels the old bydst and tranfers
> every xfrm_state
> (include x) into the new bydst hlist and new byspi hlist;
> 3. user thread on the LHS gets the lock and adds x into the new byspi hlist
> again.
>
> So the same xfrm_state (x) is added into the same list_hash
> (net->xfrm.state_byspi) 2 times that makes the list_hash become
> an inifite loop.
>
> To fix the race, x->id.spi = htonl(spi) in the xfrm_alloc_spi() is moved
> to the back of spin_lock_bh, sothat state_hash_work thread no longer add x
> which id.spi is zero into the hash_list.
>
> Fixes: f034b5d4efdf ("[XFRM]: Dynamic xfrm_state hash table sizing.")
> Signed-off-by: zhuoliang zhang <zhuoliang.zh...@mediatek.com>
Applied, thanks a lot!
One remark, please don't use base64 encoding when you send patches.
I had to hand edit your patch to get it applied.
