On 2018/9/20 17:41, Andy Duan wrote:
> From: Leonard Crestez Sent: 2018年9月20日 16:51
>> On Thu, 2018-09-20 at 08:45 +0200, Jiri Slaby wrote:
>>> On 09/20/2018, 03:58 AM, YueHaibing wrote:
>>>> Fixes gcc '-Wunused-but-set-variable' warning:
>>>>
>>>> drivers/tty/serial/imx.c: In function 'imx_uart_probe':
>>>> drivers/tty/serial/imx.c:2198:20: warning:
>>>> variable 'rtsirq' set but not used [-Wunused-but-set-variable]
>>>>
>>>> diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c @@
>>>> -2220,7 +2220,6 @@ static int imx_uart_probe(struct
>> platform_device
>>>> *pdev)
>>>>
>>>> rxirq = platform_get_irq(pdev, 0);
>>>> txirq = platform_get_irq(pdev, 1);
>>>> - rtsirq = platform_get_irq(pdev, 2);
>>>
>>> I am not sure this is correct. platform_get_irq has side effects (like
>>> enabling the IRQ). Are you sure this won't change the behaviour (this
>>> is question mostly to IMX fellows)?
>>
>> As far as I can tell there was a request_irq call for rtsirq which was
>> removed by mistake in commit afe9cbb1a6ad ("serial: imx: drop support
>> for IRDA"):
>>
>> - /* do not use RTS IRQ on IrDA */
>> - if (!USE_IRDA(sport)) {
>> - ret = devm_request_irq(&pdev->dev, rtsirq,
>> - imx_rtsint, 0,
>> -
>> dev_name(&pdev->dev), sport);
>> - if (ret)
>> - return ret;
>> - }
>>
>> This should have just removed the IRDA check and request rtsirq
>> unconditionally. Nobody noticed this by testing RTS on imx1, this is an old
>> chip and later variants have a single combined irq.
>>
>> The correct fix for the warning would be to restore that request_irq.
>>
>> --
>> Regards,
>> Leonard
>
> Yes, your explain is very correct! Thanks for your comment.
> We should restore rtsirq request that for i.MX1.
Ok, I will post a new fix patch as Leonard and suggested.
>
> Regards,
> Andy Duan
>
