Date:   Wed, 9 Mar 2022 21:19:18 -0800
From:   Randy Dunlap <rdunlap@...radead.org>
To:     Yusuf Khan <yusisamerican@...il.com>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Jason Wang <jasowang@...hat.com>,
        Michael Kelley <mikelley@...rosoft.com>,
        "Michael S. Tsirkin" <mst@...hat.com>,
        gregkh <gregkh@...uxfoundation.org>, javier@...igon.com,
        Arnd Bergmann <arnd@...db.de>, Will Deacon <will@...nel.org>,
        Jens Axboe <axboe@...nel.dk>
Subject: Re: [PATCH v3] drivers: ddcci: upstream DDCCI driver



On 3/9/22 21:01, Yusuf Khan wrote:
> Thanks for the feedback!
> 
> (2) Where would I place that documentation then,
> documentation/devicetree/bindings/?

I like the way that you have all the info in one place, but the
current way Documentation is organized has /sysfs interfaces in
Documentation/ABI/. Then I would put the rest of the doc info in
Documentation/driver-api/ddcci.rst.

> (4) A quick look at bootlin shows a symbol called
> "modalias referenced 176 times.
> 

Well, bootlin is not a kernel driver.
I see places in generic ACPI or generic (other) firmware code
or SPI bus working with modalias strings. Maybe it's OK --
just seems odd to me.

> 
> On Wed, Mar 9, 2022 at 6:49 PM Randy Dunlap <rdunlap@...radead.org> wrote:
>>
>> Hi--
>>
>> On 3/9/22 16:44, Yusuf Khan wrote:
>>> diff --git a/drivers/char/Kconfig b/drivers/char/Kconfig
>>> index 740811893c57..c7aa439d23e7 100644
>>> --- a/drivers/char/Kconfig
>>> +++ b/drivers/char/Kconfig
>>> @@ -451,4 +451,13 @@ config RANDOM_TRUST_BOOTLOADER
>>>       pool. Otherwise, say N here so it will be regarded as device input that
>>>       only mixes the entropy pool.
>>>
>>> +config DDCCI
>>> +     tristate "DDCCI display protocol support"
>>> +     help
>>> +       Display Data Channel Command Interface is a
>>
>>                                                  is an
>>
>> Also, the line above ends with a space. Please check the entire patch
>> for lines that end with SPACE and remove the trailing spaces.
>>
>>> +       interface that allows the kernel to "talk"
>>> +       to most displays made after 2005. Check your
>>> +       display's specification to see if it has
>>> +       support for this.
>>> +
>>>  endmenu
>>
>> (2) ddcci appears to be a char driver, not a misc driver,
>> so its documentation probably should not be in Documentation/misc-devices/.
>>
>> (3) The documentation file ends with:
>>
>> +options ddcci dyndbg
>> +options ddcci-backlight dyndbg
>> \ No newline at end of file
>>
>> so please add a newline at the end of the last line.
>>
>> (4) What is the modalias stuff about?
>> Does the kernel have other drivers that play modalias games
>> like this?
>>
>>
>> (5) This standalone comment might need some more text.
>> Doesn't make much sense by itself.
>>
>> +       /* Special case: sender to 0x6E is always 0x51 */
>>
>> Does that mean something like "reply to 0x6E is always 0x51" ?
>>
>>
>> (6) Be a bit more generous with spaces around operators (kernel style).
>> And be consistent.  This first line is OK, second line not so much.
>>
>> +       xor ^= ((p_flag << 7) | len);
>> +       *(ptr++) = (p_flag << 7)|len;
>>
>> This line could also be improved with some spaces:
>>
>> +       ret = payload_len+3+((quirks & DDCCI_QUIRK_SKIP_FIRST_BYTE)?1:0);
>>
>> (e.g. -- I expect that there are others.)
>>
>>
>> (7) Too much use of likely() and unlikely().
>>
>> (8) I don't see anything in the Kconfig entries such as
>>         depends on I2C
>>
>> so do these 2 drivers work without I2C being enabled in a kernel?
>>
>> Oh, if I build ddcci without CONFIG_I2C:
>>
>> ../drivers/char/ddcci.c: In function ‘__ddcci_write_bytewise’:
>> ../drivers/char/ddcci.c:104:8: error: implicit declaration of function ‘i2c_smbus_write_byte’ [-Werror=implicit-function-declaration]
>>   ret = i2c_smbus_write_byte(client, addr);
>>         ^~~~~~~~~~~~~~~~~~~~
>>   CC      kernel/delayacct.o
>> ../drivers/char/ddcci.c: In function ‘__ddcci_write_block’:
>> ../drivers/char/ddcci.c:165:9: error: implicit declaration of function ‘i2c_master_send’; did you mean ‘i2c_match_id’? [-Werror=implicit-function-declaration]
>>   return i2c_master_send(client, sendbuf, ptr - sendbuf + 1);
>>          ^~~~~~~~~~~~~~~
>>          i2c_match_id
>> ../drivers/char/ddcci.c: In function ‘__ddcci_read’:
>> ../drivers/char/ddcci.c:216:8: error: implicit declaration of function ‘i2c_master_recv’; did you mean ‘i2c_match_id’? [-Werror=implicit-function-declaration]
>>   ret = i2c_master_recv(client, buf, len);
>>         ^~~~~~~~~~~~~~~
>>         i2c_match_id
>> ../drivers/char/ddcci.c: In function ‘ddcci_identify_device’:
>> ../drivers/char/ddcci.c:413:10: error: implicit declaration of function ‘i2c_check_functionality’; did you mean ‘in_lock_functions’? [-Werror=implicit-function-declaration]
>>        && i2c_check_functionality(client->adapter,
>>           ^~~~~~~~~~~~~~~~~~~~~~~
>>           in_lock_functions
>> ../drivers/char/ddcci.c: In function ‘ddcci_module_init’:
>> ../drivers/char/ddcci.c:1835:8: error: implicit declaration of function ‘i2c_add_driver’; did you mean ‘ddcci_add_driver’? [-Werror=implicit-function-declaration]
>>   ret = i2c_add_driver(&ddcci_driver);
>>         ^~~~~~~~~~~~~~
>>         ddcci_add_driver
>> ../drivers/char/ddcci.c: In function ‘ddcci_module_exit’:
>> ../drivers/char/ddcci.c:1868:2: error: implicit declaration of function ‘i2c_del_driver’; did you mean ‘ddcci_del_driver’? [-Werror=implicit-function-declaration]
>>   i2c_del_driver(&ddcci_driver);
>>   ^~~~~~~~~~~~~~
>>   ddcci_del_driver
>>
>>
>>
>> --
>> ~Randy

-- 
~Randy

Powered by blists - more mailing lists