Re: [Patch] make cpsw work on another eval board

[christos%astron.com@localhost (Christos Zoulas)]

In article <20140224162411.GA9330%tazenda.kollasch.net@localhost>,
Jonathan A. Kollasch <jakllsch%kollasch.net@localhost> wrote:
>On Mon, Feb 24, 2014 at 03:35:14PM +0900, Ryota Ozaki wrote:
>> Hi all,
>> 
>> I'm trying to get NetBSD running on CKB-3352(*).
>> 
>> (*) http://www.computex.co.jp/products/ckb/ckb_3352.htm
>> 
>> The board is based on TI AM3352 so that I can boot
>> NetBSD with -m evbarm kernel=BEAGLEBONE configuration.
>> However, cpsw doesn't work because the board has
>> a different PHY (MICREL KSZ9021RNI) from BeagleBone
>> (Black)'s (SMSC LAN8710A); cpsw needs to set RGMII mode.
>> Another issue is that the uboot of the board doesn't
>> support networking features, so cpsw and PHY are not
>> initialized as the kernel expects; cpsw needs to do
>> some additional initializations and PHY needs to
>> adjust clock timing for cpsw.
>> 
>> I wrote two patches to make cpsw work on the board.
>> The patches can be found at
>> https://gist.github.com/ozaki-r/9182826 .

- Typo "initialze", also can you add some symbolic #defines instead of
  explaining what the bits do in comments?
- Why don't you cache the result of cpsw_phy_has_1000t(sc) in sc so that
  you don't have to call it again?
- It might be better to write a function to manipulate ale_entry[] (get
  and set the mac) instead of open-coding it.
- KNF is return 0; instead of return (0);

LGTM otherwise.

christos