Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Date:   Fri, 15 Apr 2022 13:55:02 +0200
From:   Maciej Fijalkowski <maciej.fijalkowski@intel.com>
To:     Ivan Vecera <ivecera@redhat.com>
Cc:     netdev@vger.kernel.org, Fei Liu <feliu@redhat.com>,
        "moderated list:INTEL ETHERNET DRIVERS" 
        <intel-wired-lan@lists.osuosl.org>, mschmidt@redhat.com,
        Brett Creeley <brett.creeley@intel.com>,
        open list <linux-kernel@vger.kernel.org>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        "David S. Miller" <davem@davemloft.net>
Subject: Re: [Intel-wired-lan] [PATCH net] ice: Protect vf_state check by
 cfg_lock in ice_vc_process_vf_msg()
Message-ID: <YlldFriBVkKEgbBs@boxer>
References: <20220413072259.3189386-1-ivecera@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220413072259.3189386-1-ivecera@redhat.com>
Precedence: bulk

On Wed, Apr 13, 2022 at 09:22:59AM +0200, Ivan Vecera wrote:
> Previous patch labelled "ice: Fix incorrect locking in
> ice_vc_process_vf_msg()"  fixed an issue with ignored messages

tiny tiny nit: double space after "
Also, has mentioned patch landed onto some tree so that we could provide
SHA-1 of it? If not, then maybe squashing this one with the mentioned one
would make sense?

> sent by VF driver but a small race window still left.
> 
> Recently caught trace during 'ip link set ... vf 0 vlan ...' operation:
> 
> [ 7332.995625] ice 0000:3b:00.0: Clearing port VLAN on VF 0
> [ 7333.001023] iavf 0000:3b:01.0: Reset indication received from the PF
> [ 7333.007391] iavf 0000:3b:01.0: Scheduling reset task
> [ 7333.059575] iavf 0000:3b:01.0: PF returned error -5 (IAVF_ERR_PARAM) to our request 3
> [ 7333.059626] ice 0000:3b:00.0: Invalid message from VF 0, opcode 3, len 4, error -1
> 
> Setting of VLAN for VF causes a reset of the affected VF using
> ice_reset_vf() function that runs with cfg_lock taken:
> 
> 1. ice_notify_vf_reset() informs IAVF driver that reset is needed and
>    IAVF schedules its own reset procedure
> 2. Bit ICE_VF_STATE_DIS is set in vf->vf_state
> 3. Misc initialization steps
> 4. ice_sriov_post_vsi_rebuild() -> ice_vf_set_initialized() and that
>    clears ICE_VF_STATE_DIS in vf->vf_state
> 
> Step 3 is mentioned race window because IAVF reset procedure runs in
> parallel and one of its step is sending of VIRTCHNL_OP_GET_VF_RESOURCES
> message (opcode==3). This message is handled in ice_vc_process_vf_msg()
> and if it is received during the mentioned race window then it's
> marked as invalid and error is returned to VF driver.
> 
> Protect vf_state check in ice_vc_process_vf_msg() by cfg_lock to avoid
> this race condition.
> 
> Fixes: e6ba5273d4ed ("ice: Fix race conditions between virtchnl handling and VF ndo ops")
> Tested-by: Fei Liu <feliu@redhat.com>
> Signed-off-by: Ivan Vecera <ivecera@redhat.com>
> ---
>  drivers/net/ethernet/intel/ice/ice_virtchnl.c | 38 +++++++++----------
>  1 file changed, 17 insertions(+), 21 deletions(-)
> 
> diff --git a/drivers/net/ethernet/intel/ice/ice_virtchnl.c b/drivers/net/ethernet/intel/ice/ice_virtchnl.c
> index 5612c032f15a..553287a75b50 100644
> --- a/drivers/net/ethernet/intel/ice/ice_virtchnl.c
> +++ b/drivers/net/ethernet/intel/ice/ice_virtchnl.c
> @@ -3625,44 +3625,39 @@ void ice_vc_process_vf_msg(struct ice_pf *pf, struct ice_rq_event_info *event)
>  		return;
>  	}
>  
> +	mutex_lock(&vf->cfg_lock);
> +
>  	/* Check if VF is disabled. */
>  	if (test_bit(ICE_VF_STATE_DIS, vf->vf_states)) {
>  		err = -EPERM;
> -		goto error_handler;
> -	}
> -
> -	ops = vf->virtchnl_ops;
> -
> -	/* Perform basic checks on the msg */
> -	err = virtchnl_vc_validate_vf_msg(&vf->vf_ver, v_opcode, msg, msglen);
> -	if (err) {
> -		if (err == VIRTCHNL_STATUS_ERR_PARAM)
> -			err = -EPERM;
> -		else
> -			err = -EINVAL;
> +	} else {
> +		/* Perform basic checks on the msg */
> +		err = virtchnl_vc_validate_vf_msg(&vf->vf_ver, v_opcode, msg,
> +						  msglen);
> +		if (err) {
> +			if (err == VIRTCHNL_STATUS_ERR_PARAM)
> +				err = -EPERM;
> +			else
> +				err = -EINVAL;
> +		}

The chunk above feels a bit like unnecessary churn, no?
Couldn't this patch be simply focused only on extending critical section?

>  	}
> -
> -error_handler:
>  	if (err) {
>  		ice_vc_send_msg_to_vf(vf, v_opcode, VIRTCHNL_STATUS_ERR_PARAM,
>  				      NULL, 0);
>  		dev_err(dev, "Invalid message from VF %d, opcode %d, len %d, error %d\n",
>  			vf_id, v_opcode, msglen, err);
> -		ice_put_vf(vf);
> -		return;
> +		goto finish;
>  	}
>  
> -	mutex_lock(&vf->cfg_lock);
> -
>  	if (!ice_vc_is_opcode_allowed(vf, v_opcode)) {
>  		ice_vc_send_msg_to_vf(vf, v_opcode,
>  				      VIRTCHNL_STATUS_ERR_NOT_SUPPORTED, NULL,
>  				      0);
> -		mutex_unlock(&vf->cfg_lock);
> -		ice_put_vf(vf);
> -		return;
> +		goto finish;
>  	}
>  
> +	ops = vf->virtchnl_ops;
> +
>  	switch (v_opcode) {
>  	case VIRTCHNL_OP_VERSION:
>  		err = ops->get_ver_msg(vf, msg);
> @@ -3773,6 +3768,7 @@ void ice_vc_process_vf_msg(struct ice_pf *pf, struct ice_rq_event_info *event)
>  			 vf_id, v_opcode, err);
>  	}
>  
> +finish:
>  	mutex_unlock(&vf->cfg_lock);
>  	ice_put_vf(vf);
>  }
> -- 
> 2.35.1
> 
> _______________________________________________
> Intel-wired-lan mailing list
> Intel-wired-lan@osuosl.org
> https://lists.osuosl.org/mailman/listinfo/intel-wired-lan