[U-Boot] [PATCH v3 0/9] From Simon:

[Simon Glass]

Hi Heiko,

On 2 December 2016 at 04:30, Heiko Stübner <heiko at sntech.de> wrote:
> Am Mittwoch, 30. November 2016, 19:20:30 schrieb Simon Glass:
>> Hi Heiko,
>>
>> On 30 November 2016 at 04:52, Heiko Stübner <heiko at sntech.de> wrote:
>> > Am Dienstag, 29. November 2016, 19:47:10 schrieb Simon Glass:
>> >> Hi Heiko,
>> >>
>> >> On 29 November 2016 at 16:45, Heiko Stübner <heiko at sntech.de> wrote:
>> >> > Hi Simon,
>> >> >
>> >> > Am Sonntag, 27. November 2016, 10:01:40 schrieb Simon Glass:
>> >> >> This v3 patch is an update on Sjoerd's original v2 series from
>> >> >> Feburary.
>> >> >> I have dealt with the changes requested at the time, and adjusted the
>> >> >> way
>> >> >> that the speed change is handled.
>> >> >>
>> >> >> Tested on firefly-rk3288, rock2.
>> >> >
>> >> > Tested on a firefly as well. Though I get mixed results in my netboot
>> >> > environment. At 100MBit (manually limited) everything seems fine:
>> >> >
>> >> > ----------------------
>> >> > Speed: 100, full duplex
>> >> > BOOTP broadcast 1
>> >> > BOOTP broadcast 2
>> >> > DHCP client bound to address 192.168.140.58 (269 ms)
>> >> > Using ethernet at ff290000 device
>> >> > TFTP from server 192.168.140.1; our IP address is 192.168.140.58
>> >> > Filename 'hstuebner/firefly.vmlinuz'.
>> >> > Load address: 0x4000000
>> >> > Loading:
>> >> > #################################################################
>> >> >
>> >> >          ##############################################################
>> >> >          ###
>> >> >          ##############################################################
>> >> >          ###
>> >> >          ##############################################################
>> >> >          ###
>> >> >          ##############################################################
>> >> >          ###
>> >> >          ##############################################################
>> >> >          ###
>> >> >          ##############################################################
>> >> >          ###
>> >> >          #########################
>> >> >          2.9 MiB/s
>> >> >
>> >> > done
>> >> > Bytes transferred = 7033483 (6b528b hex)
>> >> > ----------------------
>> >> >
>> >> > 5 out of 5 boots worked fine.
>> >> >
>> >> >
>> >> > but at 1000MHz I only get:
>> >> >
>> >> > ----------------------
>> >> > Speed: 1000, full duplex
>> >> > BOOTP broadcast 1
>> >> > BOOTP broadcast 2
>> >> > DHCP client bound to address 192.168.140.57 (270 ms)
>> >> > Using ethernet at ff290000 device
>> >> > TFTP from server 192.168.140.1; our IP address is 192.168.140.57
>> >> > Filename 'hstuebner/firefly.vmlinuz'.
>> >> > Load address: 0x4000000
>> >> > Loading: #####T #T #T ##T T #T #T T ####T #T ####
>> >> > Retry count exceeded; starting again
>> >> > ----------------------
>> >> >
>> >> > on 5 boots. 1 lonely boot also worked at 1000MBit for some unknown
>> >> > reason.
>> >> > I'm not sure if just my switch is some special snowflake (TL-SG1024
>> >> > from
>> >> > TP- Link) or there is some other voodoo at work here.
>> >> >
>> >> > The rootfs over nfs seems to work fine on 1000MBit though.
>> >>
>> >> Yes I see some timeouts, although I don't think it is anything to do
>> >> with the driver conversion. Or does this not happen unless my patches
>> >> are applied?
>> >
>> > nope, really looks like some other parts are just not fast enough? And the
>> > setup of GRF and so on, look the same as in the linux kernel, so should be
>> > fine as well.
>>
>> So to be clear, are you saying that the same problem happens on a
>> downstream tree that you tried, or not?
>
> nope, didn't say that. The firefly actually is my first board using the internal
> ethernet. My other boards normally use usb-ethernet adapters as the arc-emac
> used in other socs isn't supported at all yet.
>
>
>> In terms of 'not fast enough' you could try turning off HDMI if that
>> is on. Also see veyron_init() and see if you can adapt that to firefly
>> to get the clock speed up?
>
> Sadly not until after the 11th of december. But I guess we could just merge it
> now and hopefully fixup the special case later on. It seems to work always at
> 100MBit and it seems also at 1000MBit on your side.

I decided to hold off until we get a bit more testing. Let me know
what you fine.

Regards,
Simon