Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Fri, 21 Oct 2022 10:57:19 +0100
From:   Mark Rutland <mark.rutland@arm.com>
To:     Andrew Morton <akpm@linux-foundation.org>
Cc:     syzbot <syzbot+96e659d35b9d6b541152@syzkaller.appspotmail.com>,
        jiebin.sun@intel.com, linux-kernel@vger.kernel.org,
        syzkaller-bugs@googlegroups.com, tim.c.chen@linux.intel.com
Subject: Re: [syzbot] BUG: unable to handle kernel paging request in
 percpu_counter_add_batch
Message-ID: <Y1Js/wNrjzTyYSwD@FVFF77S0Q05N>
References: <000000000000e528a905eb822440@google.com>
 <20221020194427.68571e92bc33d98a05c852aa@linux-foundation.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20221020194427.68571e92bc33d98a05c852aa@linux-foundation.org>
Precedence: bulk

On Thu, Oct 20, 2022 at 07:44:27PM -0700, Andrew Morton wrote:
> On Thu, 20 Oct 2022 19:24:40 -0700 syzbot <syzbot+96e659d35b9d6b541152@syzkaller.appspotmail.com> wrote:

> Well.  freeque() messes with ns->percpu_msg_hdrs but msg_exit_ns()
> already did percpu_counter_destroy(&ns->percpu_msg_hdrs);
> 
> I'm thinking something like this?
> 
> --- a/ipc/msg.c~a
> +++ a/ipc/msg.c
> @@ -1329,11 +1329,11 @@ fail_msg_bytes:
>  #ifdef CONFIG_IPC_NS
>  void msg_exit_ns(struct ipc_namespace *ns)
>  {
> -	percpu_counter_destroy(&ns->percpu_msg_bytes);
> -	percpu_counter_destroy(&ns->percpu_msg_hdrs);
>  	free_ipcs(ns, &msg_ids(ns), freeque);
>  	idr_destroy(&ns->ids[IPC_MSG_IDS].ipcs_idr);
>  	rhashtable_destroy(&ns->ids[IPC_MSG_IDS].key_ht);
> +	percpu_counter_destroy(&ns->percpu_msg_bytes);
> +	percpu_counter_destroy(&ns->percpu_msg_hdrs);
>  }
>  #endif

That seems to fix the issue for me, testing with the reproducer in my prior
reply. Previously that would trigger within seconds, and with the above applied
I couldn't get it to trigger after 15+ minutes of repeated test runs.

FWIW:

  Tested-by: Mark Rutland <mark.rutland@arm.com>

Thanks,
Mark.