[OpenWrt-Devel] [PATCH 1/2] ath79: add support for Netgear WNDR4300
michal.cieslakiewicz at wp.pl
Fri Oct 11 13:14:48 EDT 2019
On Fri, 11 Oct 2019 14:58:03 +0200
"Adrian Schmutzler" <mail at adrianschmutzler.de> wrote:
> > > Have you checked that there are no usable addresses in 0x1002 and
> > > 0x5006?
> > >
> > Yes, there are wireless addresses inside calibration data at
> > offsets you provided, and they are both equal to LAN MAC.
> Okay, so for &wmac that would mean it should correctly read the
> address from calibration data even if you do not specify it. Thus,
> you should be able to remove the mtd-mac-address for wmac and still
> have the correct MAC address.
> For 0x5006 this is more interesting, as it implies that the vendor
> puts a "wrong" MAC address into the calibration data. However, since
> you have verified it, it seems appropriate to overwrite the address
> in calibration with mtd-mac-address in this case.
> Note that if you remove the mtd-mac-address for wmac, you will need
> to change the label-mac-device alias to ð0 then.
IMHO the valid place for Netgear routers to get MACs from is 0x0-0xc
(beginning of art partition). This is the first Netgear model I came
across that has a real hardware address inside calibration area,
usually there are 0xff's there. I always look into vendor's u-boot
sources to find hints about configuration and all WN[D]Rs I had access
to point at 'the start of the art' :-).
Even though vendor u-boot does not put Wifi into use, it defines where
wmac address is for u-boot commands like 'wmacset' etc.
We should consider leaving it as it is now, as long as we decide to
honour u-boot as the valid option/tool for changing device
It looks like Netgear has its own way of handling these things...
Anyway, I'll gladly read what you as a developer think about all this.
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
More information about the openwrt-devel