[PATCH] net: cpsw: Fix crashes on beaglebone black

[Sascha Hauer]

On Mon, Jan 05, 2015 at 01:35:21PM +0100, Teresa Gámez wrote:
> Hello Philippe,
> 
> > 
> > [<9fe32349>] (unwind_backtrace+0x1/0x58) from [<9fe258e9>] (panic+0x1d/0x34)
> > [<9fe258e9>] (panic+0x1d/0x34) from [<9fe327e1>] (do_exception+0xd/0x10)
> > [<9fe327e1>] (do_exception+0xd/0x10) from [<9fe32849>] 
> > (do_data_abort+0x21/0x2c)
> > [<9fe32849>] (do_data_abort+0x21/0x2c) from [<9fe32508>] 
> > (data_abort+0x48/0x60)
> > 
> > 
> 
> I suspect the issue comes from the dummy cpsw_emac1 node which adds a
> non existing phy in the beaglebone device tree. So the mdiobus_scan
> fails. This patch will workaround the issue but I'm unsure about right
> solution here.

You are right, this is not the ideal solution. Nevertheless I just sent
out a similar patch. It should be good enough for now.

In the longer run the phy shouldn't be probed for in the devices probe
function. There also should be a way to disable a slave in the device
tree, right now both slaves are defined in the dtsi file and a status =
"disabled" is ignored (Does this work in the kernel?).
Also the cpsw driver should accept a proper phy binding. The official
agreed-upon binding is phy-handle = <&ethphy>, not phy_id = <&miibus>,
<0>. The driver should at least optionally support this.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |