[OpenWrt-Devel] [PATCH 00/17] atheros: checkpatch fixes

[Florian Fainelli]

2014-06-24 15:30 GMT-07:00 Sergey Ryazanov <ryazanov.s.a at gmail.com>:
> 2014-06-25 0:30 GMT+04:00 Daniel Gimpelevich
> <daniel at gimpelevich.san-francisco.ca.us>:
>> plus the AR2317 watchdog operates completely differently from the
>> AR2315 one.
>
> Hm, I am missed that. Thank you Daniel for your note.
>
>>  What solution do you propose?
>
> As Florian mention, as a solution we can register different platform
> devices: ar2315-wdt for AR2315/16 SoCs and ar2317-wdt for AR2317/18
> SoCs. This should be easy since the code already contains the model
> detection of chip.
>
> As for the boards differences, I've thought about it and came to the
> same conclusion that Florian: we should build separate images for
> different boards each of which have own board=XXX kernel argument.

I think, if the differences we have to abstract for are (hopefully)
limited to GPIOs and the watchdog driver differences, we can have a
small data-base in the kernel, nobdoy should hopefully yell at you for
doing that. If we are talking about a few dozen regulators, SPI, i2c
peripherals, that's another story, and DT could be one way to get that
done in a generic way.

>
> There are another one solution: RedBoot has a builtin string with
> board's name, which is located just after "Board: %s" format string
> and we could find and use this string. But this solution seems ugly.

Right, I would not count on that since:

- there might be platforms that do not use RedBoot
- people might be recompiling their own RedBoot which could place the
string at a different location in the binary
-- 
Florian
_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel