linux-kernel - Re: [PATCH 2/2 v5] tty/serial/8250: use mctrl

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <6a4f1001-b023-c972-7b36-6d2f8f9a3fa8@denx.de>
Date:   Thu, 13 Jun 2019 07:32:39 +0200
From:   Stefan Roese <sr@...x.de>
To:     Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
Cc:     linux-serial@...r.kernel.org, linux-kernel@...r.kernel.org,
        Yegor Yefremov <yegorslists@...glemail.com>,
        Mika Westerberg <mika.westerberg@...ux.intel.com>,
        Giulio Benetti <giulio.benetti@...ronovasrl.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Subject: Re: [PATCH 2/2 v5] tty/serial/8250: use mctrl_gpio helpers

On 12.06.19 11:16, Andy Shevchenko wrote:
> On Wed, Jun 12, 2019 at 10:13:05AM +0200, Stefan Roese wrote:
>> On 11.06.19 16:48, Andy Shevchenko wrote:
>>> On Tue, Jun 11, 2019 at 04:02:54PM +0200, Stefan Roese wrote:
>>>> On 11.06.19 14:44, Andy Shevchenko wrote:
>>>>> On Tue, Jun 11, 2019 at 12:56:03PM +0200, Stefan Roese wrote:
>>>
>>>>>>     static inline void serial8250_out_MCR(struct uart_8250_port *up, int value)
>>>>>>     {
>>>>>>     	serial_out(up, UART_MCR, value);
>>>>>> +
>>>>>> +	if (up->gpios) {
>>>>>> +		int mctrl_gpio = 0;
>>>>>> +
>>>>>> +		if (value & UART_MCR_RTS)
>>>>>> +			mctrl_gpio |= TIOCM_RTS;
>>>>>> +		if (value & UART_MCR_DTR)
>>>>>> +			mctrl_gpio |= TIOCM_DTR;
>>>>>> +
>>>>>> +		mctrl_gpio_set(up->gpios, mctrl_gpio);
>>>>>> +	}
>>>>>>     }
>>>
>>>>>>     static inline int serial8250_in_MCR(struct uart_8250_port *up)
>>>>>>     {
>>>>>> -	return serial_in(up, UART_MCR);
>>>>>> +	int mctrl;
>>>>>> +
>>>>>> +	mctrl = serial_in(up, UART_MCR);
>>>>>> +
>>>>>> +	if (up->gpios) {
>>>>>> +		int mctrl_gpio = 0;
>>>>>> +
>>>>>> +		/* save current MCR values */
>>>>>> +		if (mctrl & UART_MCR_RTS)
>>>>>> +			mctrl_gpio |= TIOCM_RTS;
>>>>>> +		if (mctrl & UART_MCR_DTR)
>>>>>> +			mctrl_gpio |= TIOCM_DTR;
>>>>>> +
>>>>>> +		mctrl_gpio = mctrl_gpio_get_outputs(up->gpios, &mctrl_gpio);
>>>>>> +		if (mctrl_gpio & TIOCM_RTS)
>>>>>> +			mctrl |= UART_MCR_RTS;
>>>>>> +		else
>>>>>> +			mctrl &= ~UART_MCR_RTS;
>>>>>> +
>>>>>> +		if (mctrl_gpio & TIOCM_DTR)
>>>>>> +			mctrl |= UART_MCR_DTR;
>>>>>> +		else
>>>>>> +			mctrl &= ~UART_MCR_DTR;
>>>>>> +	}
>>>>>> +
>>>>>> +	return mctrl;
>>>>>>     }
>>>>>
>>>>> These are using OR logic with potentially volatile data. Shouldn't we mask
>>>>> unused bits in UART_MCR in case of up->gpios != NULL?
>>>>
>>>> Sorry, I don't see, which bits you are referring to? Could you please be
>>>> a bit more specific with the variable / macro meant (example)?
>>>
>>> I meant that we double write values in the out() which might have some
>>> consequences, though I hope nothing wrong with it happens.
>>
>> Where is the double write to a register? Sorry, I fail to spot it.
> 
> Not to the one register. From the functional point of view the same signal is
> set up twice: once per UART register, once per GPIO pins.
> 
>>> In the in() we read the all bits in the register.
>>>
>>> As now I look at the implementation of mctrl_gpio_get_outputs(),
>>> I think we rather get helpers for conversion between TIOCM and UART_MCR values,
>>> so, they can be used in get_mctrl() / set_mctrl() and above.
>>
>> Do you something like this in mind?
> 
> More likely
> 
> static inline int serial8250_MCR_to_TIOCM(int mcr)

MSR_to_TIOCM (see below) ...

> {
> 	int tiocm = 0;
> 
> 	if (mcr & ...)
> 		tiocm |= ...;
> 	...
> 
> 	return tiocm;
> }
> 
> static inline int serial8250_TIOCM_to_MCR(int tiocm)
> {
> 	... in a similar way ...
> }

While implementing such wrapper functions I noticed, that get_mctrl() /
set_mctrl() need TIOCM->MCR and MSR->TIOCM (notice MSR vs MCR here) but
serial8250_in_MCR() needs MCR->TIOCM. So there is not that much
overlay here. Additionally the wrappers would need to handle all bits
and only some of them are needed in serial8250_in/out_MCR(), so I would
need to add masking here as well.

For my taste its not really worth adding these wrappers as they won't
make things much clearer (if at all).

Thanks,
Stefan