Re: [PATCH 1/2] Staging: comedi: introduce outl_1564_* and inl_1564_* helper functions in hwdrv_apci1564.c

[Chase Southwood <chase.southwood@yahoo.com>]

Hi Dan,


>On Saturday, March 1, 2014 6:46 AM, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>>On Sat, Mar 01, 2014 at 04:28:27AM -0600, Chase Southwood wrote:
>> This patch introduces a handful of outl and inl helper functions with the
>> ultimate goal of improving code readability and allowing several lines
>> which violate the character limit to be shortened in a sane way.
>> 
>> Cc: Dan Carpenter <dan.carpenter@oracle.com>
>> Signed-off-by: Chase Southwood <chase.southwood@yahoo.com>
>> ---
>> This patchset serves as a replacement to my previous cleanup patchset for
>> hwdrv_apci1564.c
>> 
>> Dan,
>> After spending a little bit more time with this and trying out different
>> ways of cleaning this up, I decided that making helper functions for all
>> of the most common register sets would look the best, but I haven't made
>> a helper for a few of the least common inl/outl calls because if I did,
>> the sheer number of helper functions would get quite ridiculous.
>> Let me know if you think my selections of what to make into helper
>> functions seems appropriate.
>> 
>
>Yeah.  You're right...  It's kind of a lot of helper functions.
>
>I wonder if we could just do something like:
>
>static void outl_amcc(struct addi_private *devpriv, unsigned int cmd,
>              unsigned int reg)
>{
>    outl(cmd, devpriv->i_IobaseAmcc + APCI1564_DIGITAL_IP + reg);
>}
>
>And then change APCI1564_DIGITAL_IP_INTERRUPT_MODE1 to be:
>
>#define APCI1564_DIGITAL_IP_INTERRUPT_MODE1 (0x4 + 0x4)
>

I like this idea.  Just to clarify though, basically all of the macros would
change to something like 

#define APCI1564_DIGITAL_IP 0x4 #define APCI1564_DIGITAL_IP_INTERRUPT_MODE1 (0x4 + 0x4) #define APCI1564_DIGITAL_IP_INTERRUPT_MODE2		 (0x4 + 0x8) #define APCI1564_DIGITAL_IP_IRQ (0x4 + 0x10) #define APCI1564_DIGITAL_OP 0x18 #define APCI1564_DIGITAL_OP_RW 0x18 #define APCI1564_DIGITAL_OP_INTERRUPT (0x18 + 0x4) #define APCI1564_DIGITAL_OP_IRQ (0x18 + 0xc)
etc... and then we just have the single helper function
(the corrected version from your follow up email) and then
the calls would be something to the effect of:

outl_amcc(devpriv, cmd, APCI1564_DIGITAL_IP_INTERRUPT_MODE1);
or whichever macro is appropriate?  It definitely trims down
the length of the function calls by removing the dereference of
devpriv and the addition to get the proper register...I like that.

>
>The only problem with that would be i_APCI1564_Reset().  Is
>i_APCI1564_Reset() buggy?  Ian or Hartley might know.  Take a look at
>other comedi drivers as well to see what they do.
>

I agree.  I'll look into the other addi-data drivers (the layout of each
appears pretty similar) or see if Ian or Hartley can shed more light on the
reset function, because I have a sneaking suspicion that a good few of the lines
in it already are buggy, and it seems like there's a chance that it's not clearing
all of the registers that it should be, either.  I could be wrong about that though.

At any rate, I'll see what I can do.

Thanks,
Chase

>
>regards,
>
>dan carpenter