DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 652492084D
From:   Jia-Ju Bai <baijiaju1990@gmail.com>
Subject: [Report] net: wireless: iwlegacy: Several possible data races
To:     sgruszka@redhat.com, kvalo@codeaurora.org, davem@davemloft.net
Cc:     linux-wireless@vger.kernel.org, netdev <netdev@vger.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Message-ID: <be47676b-baf3-7655-ff83-8b3ed370c624@gmail.com>
Date:   Sat, 6 Oct 2018 10:25:13 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101
 Thunderbird/52.2.0
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Content-Language: en-US
Sender: linux-wireless-owner@vger.kernel.org
Precedence: bulk

****** Possible race 0 ******
CPU0:
il4965_mac_tx
     il4965_tx_skb
         line 1687: spin_lock_irqsave(&il->lock, flags)
         line 1894: q->write_ptr = il_queue_inc_wrap(...) [WRITE]

CPU1:
il4965_irq_tasklet
     il_txq_update_write_ptr
         line 2749: il_wr(..., txq->q.write_ptr); [READ]
         line 2757: _il_wr(..., txq->q.write_ptr); [READ]

CPU2:
il_bg_watchdog
     il_check_stuck_queue
         line 4817: if (q->read_ptr == q->write_ptr) [READ]

CPU3:
il4965_irq_tasklet
     il4965_rx_handle
         line 4827, function pointer: il->handlers[pkt->hdr.cmd] (il, rxb);
         il4965_hdl_tx
             il_queue_used
                 line 868: return q->write_ptr >= ... q->write_ptr ... 
[READ]

As for q->write_ptr, the WRITE operation in CPU0 is performed with holding
a spinlock il->lock, but the READ operations in CPU1, CPU2 and CPU3
are performed without holding this spinlock, so there may exist data races.

****** Possible race 1 ******
CPU0:
il4965_irq_tasklet
     il4965_rx_handle
         line 4827, function pointer: il->handlers[pkt->hdr.cmd] (il, rxb);
         il4965_hdl_tx
             line 2834: spin_lock_irqsave(&il->sta_lock, flags)
             il4965_tx_queue_reclaim
                 line 2490: q->read_ptr = ... [WRITE]

CPU1:
il_bg_watchdog
     il_check_stuck_queue
         line 4817: q->read_ptr == q->write_ptr [READ]

As for q->read_ptr, the WRITE operation in CPU0 is performed with
holding a spinlock il->sta_lock, but the READ operation in CPU1 is
performed without holding this spinlock, so there may exist a data race.

****** Possible race 2 ******
CPU0:
il4965_mac_tx
     il4965_tx_skb
         line 1687: spin_lock_irqsave(&il->lock, flags)
         line 1863: txq->need_update = 1 [WRITE]

CPU1:
il4965_irq_tasklet
     il_txq_update_write_ptr
         line 2731: if (txq->need_update == 0) [READ]

As for txq->need_update, the WRITE operation in CPU0 is performed with
holding a spinlock il->lock, but the READ operation in CPU1 is performed
without holding this spinlock, so there may exist a data race.

I do not find good ways to fix these possible data races, so I only 
report them.


Best wishes,
Jia-Ju Bai