Re: [PATCH 02/14] ARM : SAMSUNG : Add RS485 support.

[Paul Schilling <paul.s.schilling@gmail.com>]

I am working on resubmitting this stuff right now.  I have 14 patches
that should be
broken into more like 20 patches.  Most of them are ARM S3C and
specifically S3C2416 patches.
I am working on a patch right now to fix a Atomic wait in the DMA
code.  That causes
20 milliseconds of no interrupts whenever a sound is played on the samsung S3C
platform.

The code below wasn't supposed to go out and I have patch that removes
that chunk
of code almost ready that just needs testing before I send it.

On Tue, Nov 1, 2011 at 7:18 AM, Alan Cox <alan@linux.intel.com> wrote:
>> +config SAMSUNG_HAS_RS485
>> +     bool "Enable RS485 support for Samsung"
>> +     depends on SERIAL_SAMSUNG && (MACH_CONDOR2440 ||
>> MACH_CONDOR2416 || MACH_MINI2440)
>> +     default y if (MACH_CONDOR2440 || MACH_CONDOR2416)
>> +     default n if (MACH_MINI2440)
>
> There really needs to be more ARM thinking about doing this sort of
> stuff at runtime. If you can detect the board type at runtime then you
> need to be moving towards kicking board specifics out of the depends,
> and treating it as an "include support for this, if the board has it"
> so you can move towards less build time only configuration.
>

The chunk of code that says FIXME is an optimization that could be
implemented to
decrease interrupts by using DMA until the last byte is ready to be
sent.  I have a new
patch that explains in a comment what the code is for.

>> +/* FIXME */
>> +     #if 0
>> +     if (last_state != en) {
>> +
>
> So this doesn't look ready to submit either...
>

I have a patch that I haven't sent yet that fixes the code formating
issues.  I need to compile and test it before I send it out.

>
>> +             if ((utrstat & S3C2410_UTRSTAT_TXE) ? 1 : 0) {
>> +                     if (cfg->gpio_transmit_en > -1) {
>> +
>> gpio_set_value(cfg->gpio_transmit_en, 0);
>> +                     }
>
> Lots of excess brackets (see CodingStyle) - removing a few of the
> excess ones would make it easier to read later.
>
> The later bits become a real mess of ifdefs - much not your fault, the
> combination of your ifdefs and existing lack of design push it beyond
> ugly and really want addressing.
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-serial" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html