linux-kernel - Re: [PATCH] i2c: iproc: Change driver to use 'BIT' macro

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]

Date:   Thu, 18 Apr 2019 06:21:37 +0000
From:   Peter Rosin <peda@...ntia.se>
To:     Ray Jui <ray.jui@...adcom.com>, Wolfram Sang <wsa@...-dreams.de>
CC:     "linux-i2c@...r.kernel.org" <linux-i2c@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "bcm-kernel-feedback-list@...adcom.com" 
        <bcm-kernel-feedback-list@...adcom.com>,
        Rayagonda Kokatanur <rayagonda.kokatanur@...adcom.com>
Subject: Re: [PATCH] i2c: iproc: Change driver to use 'BIT' macro

On 2019-04-18 01:48, Ray Jui wrote:
> 
> 
> On 4/14/2019 11:56 PM, Peter Rosin wrote:
>> On 2019-04-13 00:59, Peter Rosin wrote:
>>> On 2019-04-03 23:05, Ray Jui wrote:
>>>> Change the iProc I2C driver to use the 'BIT' macro from all '1 << XXX'
>>>> bit operations to get rid of compiler warning and improve readability of
>>>> the code
>>>
>>> All? I see lots more '1 << XXX_SHIFT' matches. I might be behind though?
>>
>> I verified that, and yes indeed, I was behind. That said, see below...
>>
> 
> Right. Previous change that this change depends on is already queued in
> i2c/for-next.
> 
>>> Anyway, if you are cleaning up, I'm just flagging that BIT(XXX_SHIFT) looks
>>> a bit clunky to me. You might consider renaming all those single-bit
>>> XXX_SHIFT macros to simple be
>>>
>>> #define XXX BIT(<xxx>)
>>>
>>> instead of
>>>
>>> #define XXX_SHIFT <xxx>
>>>
>>> but that triggers more churn, so is obviously more error prone. You might
>>> not dare it?
>>>
> 
> With the current code, I don't see how that is cleaner. With XXX_SHIFT
> specified, it makes it very clear to the user that the define a for a
> bit location within a register. You can argue and say it makes the
> define longer, but not less clear.
> 
>>> Cheers,
>>> Peter
>>>
>>>> Signed-off-by: Ray Jui <ray.jui@...adcom.com>
>>>> ---
>>>>  drivers/i2c/busses/i2c-bcm-iproc.c | 6 +++---
>>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/drivers/i2c/busses/i2c-bcm-iproc.c b/drivers/i2c/busses/i2c-bcm-iproc.c
>>>> index 562942d0c05c..a845b8decac8 100644
>>>> --- a/drivers/i2c/busses/i2c-bcm-iproc.c
>>>> +++ b/drivers/i2c/busses/i2c-bcm-iproc.c
>>>> @@ -717,7 +717,7 @@ static int bcm_iproc_i2c_xfer_single_msg(struct bcm_iproc_i2c_dev *iproc_i2c,
>>>>  
>>>>  			/* mark the last byte */
>>>>  			if (i == msg->len - 1)
>>>> -				val |= 1 << M_TX_WR_STATUS_SHIFT;
>>>> +				val |= BIT(M_TX_WR_STATUS_SHIFT);
>>>>  
>>>>  			iproc_i2c_wr_reg(iproc_i2c, M_TX_OFFSET, val);
>>>>  		}
>>>> @@ -844,7 +844,7 @@ static int bcm_iproc_i2c_cfg_speed(struct bcm_iproc_i2c_dev *iproc_i2c)
>>>>  
>>>>  	iproc_i2c->bus_speed = bus_speed;
>>>>  	val = iproc_i2c_rd_reg(iproc_i2c, TIM_CFG_OFFSET);
>>>> -	val &= ~(1 << TIM_CFG_MODE_400_SHIFT);
>>>> +	val &= ~BIT(TIM_CFG_MODE_400_SHIFT);
>>>>  	val |= (bus_speed == 400000) << TIM_CFG_MODE_400_SHIFT;
>>
>> These two statements now no longer "match". One uses BIT and the other open
>> codes the shift. I think that's bad. Losing the _SHIFT suffix and including
>> BIT in the macro expansion, as suggested above, yields:
>>
>> 	val &= ~TIM_CFG_MODE_400;
>> 	if (bus_speed == 400000)
>> 		val |= TIM_CFG_MODE_400;
>>
>> which is perhaps one more line, but also more readable IMO.
>>
> 
> A single line with evaluation embedded is nice and clean to me. I guess
> this is subjective.

The "problem" I had when I looked at the driver was not any one specific
instance. It was just that, for my taste, the code had too many shifts
etc inline with the real code. Replacing 1 << xyz_SHIFT with BIT(xyz_SHIFT)
is not a real improvement, they are just about equal to me, it's just that
there are still too many mechanical things happening that could easily be
tucked away with the suggested approach.

> I'll leave the decision to Wolfram. If he also prefers the above change
> to be made, sure. Otherwise, I'll leave it as it is.

But if you see no value in my suggestion and/or don't what to take the
cleanup one step further, then just leave it as-is.

>> But all this is of course in deep nit-pick-territory...
>>
>> Cheers,
>> Peter
>>
> 
> Thanks,
> 
> Ray
>