On 21 April 2014 11:45, John Crispin <j...@phrozen.org> wrote: > > > On 21/04/2014 10:36, Roman Yeryomin wrote: >> On 21 April 2014 11:05, John Crispin <j...@phrozen.org> wrote: >>> >>> >>> On 21/04/2014 05:18, Claudio Leite wrote: >>>> The Airlink101 AR725W is a RT2880 router with 32MB RAM, 4MB >>>> flash, 100mbit IP175E switch and dual-band 802.11n wireless. >>>> >>>> Notes: - I couldn't get the switch to be recognized without >>>> building it into the kernel, rather than as a module. Since >>>> the driver has it attached to the ethernet port, it needs(?) to >>>> be present when the eth device is probed. >>> >>> adding the phy driver to the kernel config is the correct >>> approach >> >> Which leads to larger kernel for all devices even without that >> switch and those which have other switches. I think this was >> discussed before. The boards with rtl8366xx don't have them >> explicitly enabled in target kernel config, all is done via kernel >> package. Correct me if I'm wrong? >> > > i have never done so for any target i think. which target does this ? > > rtl8366xx is not a phy driver but uses smi which makes it loadable > independent of the netdevcie bring up. > > not having ip175e in the kernel will lead to us having to ship > ralink_soc_eth as a ko aswell i think. > > correct me if i am a wrong ...
The target was ramips, rt2880 and rt305x subtargets at least. I'm not sure how rtl8366xx differs from ip175e but both realtek switch and smi parts are under drivers/net/phy/. I remember when I submitted a patch to exclude realtek switch from target kernel konfig and just use kernel package when it's needed by a board, somebody said it will not work like this. But it did and works now also. Unfortunately I don't have any hw with ip175e to test... Regards, Roman _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
