ALLWINNER/h3 malfunction

[Daniel Braniss]

> On 17 Sep 2016, at 20:55, Emmanuel Vadot <manu at bidouilliste.com> wrote:
> 
> On Sat, 17 Sep 2016 20:41:09 +0300
> Daniel Braniss <danny at cs.huji.ac.il <mailto:danny at cs.huji.ac.il>> wrote:
> 
>> 
>>> On 17 Sep 2016, at 6:19 PM, Emmanuel Vadot <manu at bidouilliste.com> wrote:
>>> 
>>> On Wed, 7 Sep 2016 16:34:48 +0300
>>> Daniel Braniss <danny at cs.huji.ac.il> wrote:
>>> 
>>>> Hi All!
>>>> 
>>>> I am playing with an orangepi-one, and I now compiled a kernel without INVARIANTS & WITNESS
>>>> and it panics, when compiled without WITNESS, it boots, but for example the awg comes up
>>>> with a semi random ethernet address, so clearly there is some timing issues. I have placed the panic in
>>>>       http://www.cs.huji.ac.il/~danny/orangepi/op1.panic.1
>>>> 
>>>> and so if someone can point me to where some delay can be added ?
>>>> 
>>>> thanks,
>>>> 	danny
>>>> 
>>> 
>>> Hello Daniel,
>>> 
>>> I'll admit that I've never tested ALLWINNER kernel without INVARIANTS
>>> or WITNESS recently, I'll try that.
>>> I assume that you've patched your DTS for ethernet support, can you
>>> try with the unpatched one ?
>> 
>> i searched high and low for such one, can you send me one?
> 
> It's in the tree in sys/gnu/dts/arm/
> Yes it doesn't have the awg node, I know, h3 emac bindings aren't
> stable for now.

I couldn’t find any orange-one.dts, so I’m using the one I created,
adding the a83t-sid and a83t-ts got both recognized:

sysctl -a | grep sid
kern.proc_vmmap_skip_resident_count: 0
kern.geom.label.ufsid.enable: 1
net.inet.sctp.min_residual: 1452
dev.aw_sid.0.rootkey: 02004620043548045028028e0400005b
dev.aw_sid.0.%parent: simplebus0
dev.aw_sid.0.%pnpinfo: name=eeprom at 01c14000 compat=allwinner,sun8i-a83t-sid
dev.aw_sid.0.%location: 
dev.aw_sid.0.%driver: aw_sid
dev.aw_sid.0.%desc: Allwinner Secure ID Controller
dev.aw_sid.%parent: 

sysctl -a | grep -i therma
dev.aw_thermal.0.gpu: 191C
dev.aw_thermal.0.cluster1: 191C
dev.aw_thermal.0.cluster0: 191C
dev.aw_thermal.0.%parent: simplebus0
dev.aw_thermal.0.%pnpinfo: name=rtp at 01f04000 compat=allwinner,sun8i-a83t-ts
dev.aw_thermal.0.%location: 
dev.aw_thermal.0.%driver: aw_thermal
dev.aw_thermal.0.%desc: Allwinner Thermal Sensor Controller
dev.aw_thermal.%parent:

my guess there is something fishy here since 191C is a bit high…

cheers,
	danny

> 
>>> It seems that it panic because the mod0 clk doesn't have the correct
>>> parent.
>>> 
>>> For the semi random MAC address this is expected. The awg driver
>>> generate a random one if it can't use the SID root key as
>>> the "source", which we don't support for now on H3 (it shouldn't be 
>>> hard to add support for the H3).
>>> 
>>> Cheers,
>>> 
>>> -- 
>>> Emmanuel Vadot <manu at bidouilliste.com> <manu at freebsd.org>
> 
> 
> -- 
> Emmanuel Vadot <manu at bidouilliste.com <mailto:manu at bidouilliste.com>> <manu at freebsd.org <mailto:manu at freebsd.org>>