Welcome! Log In Create A New Profile

Advanced

2017.07- 2024.10 U-Boot Kirkwood

Posted by bodhi 
Last post moved here;:
http://forum.doozan.com/read.php?3,33178

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
Thanks for putting this together for us @bodhi, hope you are well ... I just upgraded successfully uboot to latest version on my GFN ... only issue is ethaddr :

root@GoFlexNet1:~/uboot# fw_setenv ethaddr 'XX:XX:XX:XX:XX:XX'
Can't overwrite "ethaddr"

Does setting the MAC address "correctly" matter ? ie can I just leave it with what it's set to ?

(next job is rootfs/kernel), so my uname -a is :

root@GoFlexNet1:~/uboot# uname -a
Linux GoFlexNet1 3.16.0-kirkwood-tld-3 #2 PREEMPT Sat Sep 27 02:01:41 PDT 2014 armv5tel GNU/Linux

(Wheezy)

I don't 100% understand the device tree stuff, but presumably once I try to boot new kernel from dropbox image, I will need to update the uboot to make it device tree aware ?
DonCharisma Wrote:
-------------------------------------------------------
> Thanks for putting this together for us @bodhi, ho
> pe you are well ... I just upgraded successfully u
> boot to latest version on my GFN ... only issue is
> ethaddr :
>
> root@GoFlexNet1:~/uboot# fw_setenv ethaddr 'XX:XX:
> XX:XX:XX:XX'
> Can't overwrite "ethaddr"
>
> Does setting the MAC address "correctly" matter ?
> ie can I just leave it with what it's set to ?
>
> (next job is rootfs/kernel), so my uname -a is :
>
> root@GoFlexNet1:~/uboot# uname -a
> Linux GoFlexNet1 3.16.0-kirkwood-tld-3 #2 PREEMPT
> Sat Sep 27 02:01:41 PDT 2014 armv5tel GNU/Linux
>
> (Wheezy)
>
> I don't 100% understand the device tree stuff, but
> presumably once I try to boot new kernel from drop
> box image, I will need to update the uboot to make
> it device tree aware ?

Hi Don :)

Good to see you posting again!

No need to worry, the latest uboot is FDT aware. The only thing to pay attention to is if you have flashed the default envs image that comes with the uboot image, it will automatically look for the DTB file in /boot/dts. This folder is populated when you install the kernel 4.10.0 (as instructed).

The default MAC address was randomly generated local MAC address, so it can be used w/o problem. But you should be able to change it after upgrading the ketnel to 4.10.0.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
bodhi Wrote:
-------------------------------------------------------
> DonCharisma Wrote:
> --------------------------------------------------
> -----
> > Thanks for putting this together for us @bodhi,
> ho
> > pe you are well ... I just upgraded successfully
> u
> > boot to latest version on my GFN ... only issue
> is
> > ethaddr :
> >
> > root@GoFlexNet1:~/uboot# fw_setenv ethaddr 'XX:X
> X:
> > XX:XX:XX:XX'
> > Can't overwrite "ethaddr"
> >
> > Does setting the MAC address "correctly" matter
> ?
> > ie can I just leave it with what it's set to ?
> >
> > (next job is rootfs/kernel), so my uname -a is :
> >
> > root@GoFlexNet1:~/uboot# uname -a
> > Linux GoFlexNet1 3.16.0-kirkwood-tld-3 #2 PREEMP
> T
> > Sat Sep 27 02:01:41 PDT 2014 armv5tel GNU/Linux
> >
> > (Wheezy)
> >
> > I don't 100% understand the device tree stuff, b
> ut
> > presumably once I try to boot new kernel from dr
> op
> > box image, I will need to update the uboot to ma
> ke
> > it device tree aware ?
>
> Hi Don :)
>
> Good to see you posting again!
>
> No need to worry, the latest uboot is FDT aware.
> The only thing to pay attention to is if you have
> flashed the default envs image that comes with the
> uboot image, it will automatically look for the DT
> B file in /boot/dts. This folder is populated when
> you install the kernel 4.10.0 (as instructed).
>
> The default MAC address was randomly generated loc
> al MAC address, so it can be used w/o problem. But
> you should be able to change it after upgrading th
> e ketnel to 4.10.0.

Hey bodhi ... good to be back and speak again, apologies for absence ;) ...

I flashed the default envs from the file you supplied, which I think caused the MAC to change ... will try changing it after upgrading debian/kernel ...

Only issue left now is that netconsole seems to be masking some of the serial output (immediately after starting netconsole in uboot), will try switching off netconsole as I'm preferring serial connection these days. Also noticed netconsole seems to slow the boot process as it does ping first to see if server up, so another reason for using serial :)

Cheers
Don,

> Only issue left now is that netconsole seems to be
> masking some of the serial output (immediately aft
> er starting netconsole in uboot), will try switchi
> ng off netconsole as I'm preferring serial connect
> ion these days. Also noticed netconsole seems to s
> low the boot process as it does ping first to see
> if server up, so another reason for using serial :
> )

netconsole will mask serial output if using the same terminal for both serial console and netconsole. To see full log, you need 2 terminal sessions. With 1 terminal, when netconsole start, serial output stops, and vice versa.

If you have set the ping count to 5 like I suggested in the instruction, you could reduce the ping count to 2 to speed it up. This is needed only in a noisy network so that we dont get false positive (the 1st ping usually not reliable in a noisy network).
fw_setenv preboot_nc 'setenv nc_ready 0; for pingstat in 1 2 3 4 5; do; sleep 1; if run if_netconsole; then setenv nc_ready 1; fi; done; if test $nc_ready -eq 1; then run start_netconsole; fi'

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
bodhi Wrote:
-------------------------------------------------------
> Don,
>
> > Only issue left now is that netconsole seems to
> be
> > masking some of the serial output (immediately a
> ft
> > er starting netconsole in uboot), will try switc
> hi
> > ng off netconsole as I'm preferring serial conne
> ct
> > ion these days. Also noticed netconsole seems to
> s
> > low the boot process as it does ping first to se
> e
> > if server up, so another reason for using serial
> :
> > )
>
> netconsole will mask serial output if using the sa
> me terminal for both serial console and netconsole
> . To see full log, you need 2 terminal sessions. W
> ith 1 terminal, when netconsole start, serial outp
> ut stops, and vice versa.
>
> If you have set the ping count to 5 like I suggest
> ed in the instruction, you could reduce the ping c
> ount to 2 to speed it up. This is needed only in a
> noisy network so that we dont get false positive (
> the 1st ping usually not reliable in a noisy netwo
> rk).
>
> fw_setenv preboot_nc 'setenv nc_ready 0; for pings
> tat in 1 2 3 4 5; do; sleep 1; if run if_netconsol
> e; then setenv nc_ready 1; fi; done; if test $nc_r
> eady -eq 1; then run start_netconsole; fi'
>

Thanks bodhi, more investigation and I can see what you're saying. Sorry about repeated question in another thread !

I have solved this by :

fw_setenv preboot

ie, preboot doesn't now do anything, and of course messages now all go to serial

Don Charisma ... because anything is possible with Charisma

My blog - http://DonCharisma.org
Our commercial site - http://DonCharisma.com
I was updating all my pogoplugs to the latest U-boot and ran into a USB disk problem.

I have one drive that this uboot will just not work with, it throws timeout errors (see below). Google shows that a lot of people have this same problem, but nobody seems to have a solution yet. All of the other dozen drives I have work fine.
This is the boot drive, but that doesn't matter -- it gets these errors even when just scanning the USB bus if this drive is plugged in.

This drive works fine under Debian, it is only uboot that has trouble with it.

With the uboot.mtd0.pinkpogo.davygravy-2012-02-20.kwb, it works 95% of the time and only fails to boot occasionally. But with uboot.2016.05-tld-1.pogo_e02.mtd0.kwb it fails 99% of the time. I have tried other uboots, uboot-pogo_e02-archlinux-2014.kwb will not even recognize that drive.

Natually, I want to have a uboot that has ext4 support and the faster disk read than the davygravy one -- which is painfully slow.

I have experience writing linux drivers, and the ability to build a uboot, so I could try to find & fix the problem. Question is, is it worthwhile? And would you (bodhi) take the changes if I could fix it? And any hints or tips would be appreciated.

**** Here is a snippet of the console output *****
U-Boot 2016.05-tld-1 (Jun 12 2016 - 13:23:43 -0700)
starting USB...
...
USB0: USB EHCI 1.00
scanning bus 0 for devices... 4 USB Device(s) found
scanning usb for storage devices... 2 Storage Device(s) found
1 bytes read in 330 ms (0 Bytes/s)
Found bootable drive on usb 0:1
EHCI timed out on TD - token=0x80008d80
EHCI timed out on TD - token=0x80008d80
EHCI timed out on TD - token=0x80008d80
EHCI timed out on TD - token=0x1f8c80
EHCI timed out on TD - token=0x80008d80
EHCI timed out on TD - token=0x80008d80
EHCI timed out on TD - token=0x80008d80
EHCI timed out on TD - token=0x80008d80
rayvt Wrote:
-------------------------------------------------------
> I was updating all my pogoplugs to the latest U-bo
> ot and ran into a USB disk problem.
>
> I have one drive that this uboot will just not wor
> k with, it throws timeout errors (see below). Goo
> gle shows that a lot of people have this same prob
> lem, but nobody seems to have a solution yet. All
> of the other dozen drives I have work fine.
> This is the boot drive, but that doesn't matter --
> it gets these errors even when just scanning the U
> SB bus if this drive is plugged in.
>
> This drive works fine under Debian, it is only ubo
> ot that has trouble with it.
>
> With the uboot.mtd0.pinkpogo.davygravy-2012-02-20.
> kwb, it works 95% of the time and only fails to bo
> ot occasionally. But with uboot.2016.05-tld-1.pog
> o_e02.mtd0.kwb it fails 99% of the time. I have t
> ried other uboots, uboot-pogo_e02-archlinux-2014.k
> wb will not even recognize that drive.
>
> Natually, I want to have a uboot that has ext4 sup
> port and the faster disk read than the davygravy o
> ne -- which is painfully slow.
>
> I have experience writing linux drivers, and the a
> bility to build a uboot, so I could try to find &
> fix the problem. Question is, is it worthwhile?
> And would you (bodhi) take the changes if I could
> fix it? And any hints or tips would be appreciate
> d.
>
> **** Here is a snippet of the console output ****
> *
> U-Boot 2016.05-tld-1 (Jun 12 2016 - 13:23:43 -0700
> )
> starting USB...
> ...
> USB0: USB EHCI 1.00
> scanning bus 0 for devices... 4 USB Device(s) foun
> d
> scanning usb for storage devices... 2 Stora
> ge Device(s) found
> 1 bytes read in 330 ms (0 Bytes/s)
> Found bootable drive on usb 0:1
> EHCI timed out on TD - token=0x80008d80
> EHCI timed out on TD - token=0x80008d80
> EHCI timed out on TD - token=0x80008d80
> EHCI timed out on TD - token=0x1f8c80
> EHCI timed out on TD - token=0x80008d80
> EHCI timed out on TD - token=0x80008d80
> EHCI timed out on TD - token=0x80008d80
> EHCI timed out on TD - token=0x80008d80

This problem is very common. There is no need to have new version.

- Did you try a different flash drive?
- Also try setting the env usb_ready_retry to 60 seconds in serial console:

setenv usb_ready_retry 60

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
I can't find any ide commands in this uBoot. I'm trying to boot from an SSD on a sata interface. Can this uBoot do that?
hotqj Wrote:
-------------------------------------------------------
> I can't find any ide commands in this uBoot. I'm t
> rying to boot from an SSD on a sata interface. Can
> this uBoot do that?

Yes, of course, the ide command is built in. But depending which u-boot. Which box do you have? Please post the serial console boot log when you try this command.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
Last posts moved here:
http://forum.doozan.com/read.php?3,34031,34031#msg-34031

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 05/10/2017 03:40PM by bodhi.
Re: my April 27 post

"[bodhi] - This problem is very common. .. Did you try a different flash drive? "

Yes, I know. ;-(
I tried every USB drive I have, which is a lot. Plus, googling the error message gets a potload of hits, all people complaining about the problem.

I found the problem in UBOOT and fixed it. See the thread: http://forum.doozan.com/read.php?3,35295 "[Solved] EHCI timed out on TD - token=XXXX"

Now uboot works reliably on every USB drive I have. I will be submitting the patch to denx as well.

Regards, ray
Thanks for your contribution, ray!

Just in time :) I have rebased uboot kirwood to 2017.05. Will review and take your patch before releasing.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
hi bodhi,

> I have rebased uboot kirwood to 20
> 17.05. Will review and take your patch before rele
> asing.

do you have a timeline for that?
i have to flash an goflex.net or two and are in doubt to wait...

tnx
jay
Hi jay,

Probably within 1, 2 weeks.

But upgrading this u-boot version is very easy, it's a drop-in replacement. You can keep your current envs as is. You will only need to flash u-boot image, no new set up need.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
hi bodhi,
is there a way to get the installed uboot version from the booted (goflexnet) system (rescue or debian)?
means without serial or netcat...
and: can i stay at 2015.10 with the newest rootfs? what are the benefits if using 2016.5 or 2017.5?
tnx
jay



Edited 2 time(s). Last edit at 07/06/2017 08:34PM by jay.
bodhi Wrote:
-------------------------------------------------------
> Thanks for your contribution, ray!
>
> Just in time :) I have rebased uboot kirwood to 2017.05. Will review and take your patch before releasing.
>
Bodhi, will you please add a support for FIT image?
jay,

> is there a way to get the installed uboot version
> from the booted (goflexnet) system (rescue or debi
> an)?
> means without serial or netcat...
> and:

Yes. Let me find the command.

> can i stay at 2015.10 with the newest rootfs?
> what are the benefits if using 2016.5 or 2017.5?

Yes, you can stay with 2015.10. There is no need to update u-boot if you have not run into any limitation. 2015.10 version is relatively new, and it has most the features.

The only benefit in installing new u-boot is to keep up with new features people have upstreamed. I need to do this once in a while so that I don't fall too far behind in architectural changes in mainline. Even doing yearly like I've been doing, it takes a bit of works to rebase to mainline code. Mostly because I always test each u-boot image (14 images) before I release.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
jay,

To check the current installed u-boot version in mtd0:
grep -a --null-data U-Boot /dev/mtd0ro

Afer executing the above comand, you might find a lof of errors in dmesg as a side effect of this. Don't panic! it is normal.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
hi bodhi,
thanks a lot for the command :-)

i have another little problem:
goflexnet with 2016.05 uboot,
linux-4.11.3-kirkwood-tld-2-bodhi
Debian-4.4.0-kirkwood-tld-1-rootfs-bodhi

with the 2010 uboot the device got the 1000Mb/s link without problems.
with the new uboot there ist mostly just a 100Mb/s link.
after dis- and reconnecting the lan cable some times, it switches to 1000.

uboot:

Net: egiga0
88E1116 Initialized on egiga0
No link on egiga0
ping failed; host 192.168.0.242 is not alive
No link on egiga0
ping failed; host 192.168.0.242 is not alive
No link on egiga0
ping failed; host 192.168.0.242 is not alive
No link on egiga0
ping failed; host 192.168.0.242 is not alive
No link on egiga0
ping failed; host 192.168.0.242 is not alive
Hit any key to stop autoboot: 0

or

88E1116 Initialized on egiga0
Using egiga0 device

ARP Retry count exceeded; starting again
ping failed; host 192.168.0.242 is not alive
Using egiga0 device

ARP Retry count exceeded; starting again
ping failed; host 192.168.0.242 is not alive
Using egiga0 device

ARP Retry count exceeded; starting again
ping failed; host 192.168.0.242 is not alive
Using egiga0 device

ARP Retry count exceeded; starting again
ping failed; host 192.168.0.242 is not alive
Using egiga0 device

ARP Retry count exceeded; starting again
ping failed; host 192.168.0.242 is not alive
Hit any key to stop autoboot: 0

dmesg:
[ 6.021878] mv643xx_eth: MV-643xx 10/100/1000 ethernet driver version 1.4
[ 7.132484] mv643xx_eth_port mv643xx_eth_port.0 eth0: port 0 with MAC address 00:10:75:12:34:56
[ 33.702344] mv643xx_eth_port mv643xx_eth_port.0 eth0: link up, 100 Mb/s, full duplex, flow control disabled

any hint?
tnx
jay
jay,

Did you set the MAC address to the GoFlex Net real MAC address? in u-boot default envs image I used a random local MAC address. After you flashed new u-boot and its default envs, you should also set the MAC address u-boot env (ethaddr) to the real one.

I suspect this has something to do with this new default MAC address. Once you have set it to the real one and reboot, the switch or router should have this MAC addess and negotiate the 1Gbits link.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
hi bodhi,

> Did you set the MAC address to the GoFlex Net real
> MAC address?
sure.
ethaddr is the real hardware mac.

jay
jay,

It is a weird problem that I think related to your configuration (eg. the switch/router/cable that you are connecting the GF Net to). You could make sure that the GFNet is running at Gbits, and then shutdown, power up.

I have the exact configuration that you have for one of my GF Net boxes.
Linux version 4.11.3-kirkwood-tld-2 (root@tldDebian) (gcc version 4.9.2 (Debian 4.9.2-10) ) #1 PREEMPT Tue Jun 6 17:01:17 PDT 2017
uboot_version=U-Boot 2016.05-tld-1 (Jun 12 2016 - 13:27:19 -0700)
I did not experience this at all. Mine connects to an unmanaged 1Gbs switch. What type of switch you are connecting to?

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
hi bodhi,

> It is a weird problem
yes, really weird.

> I did not experience this at all. Mine connects to
> an unmanaged 1Gbs switch. What type of switch you
> are connecting to?
i tested with 2 cables and different switch ports.
my switch is an OfficeConnect Managed Gigabit Switch 3CDSG8 (HP 1900 Switch Series)
i use some vlans, but the tested goflex ports are unmanaged.
i think its just a little incompability in the autoneg of the actual kernel nic driver.
doesnt disturb further...

jay
Hi jay,

> i think its just a little incompability in the aut
> oneg of the actual kernel nic driver.

It could be. Although I found unmanaged switches handle the changes in speed better (I think because they dont do anything sophisticate). I recalled when I tested the HP Thin Client T5325 I kept switching the same port for 1Gbs and 100Mbs NICs back and forth but did not see this negotiation confusion,

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
FYI. I will be on travel for a few weeks again! So not sure if I will have time to release the 2017.05 u-boots.

u-boot 2017.07 mainline has been released a few days ago, so by the the time I get back home, it will be a rebase to 2017.07 :))

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
hi bodhi,

> I tested the HP Thin Client T5325 I kept switch
> ing the same port for 1Gbs and 100Mbs NICs back an
> d forth but did not see this negotiation confusion

well, as said, it doesn't disturb further.
its just my test environment...

> not sure if I will have time to release the 2017.05 u-boots.

n.p. 4 me ;)
i now use the above configuration until you release a new main system based on stretch :-)

have fun
jay
bodhi Wrote:
-------------------------------------------------------
> u-boot 2017.07 mainline has been released a few days ago, so by the the time I get back home, it will be a rebase to 2017.07 :))
>
This is great. But, please add a support for FIT.
> This is great. But, please add a support for FIT.

FIT has a very low prriority on my TODO list. For sure it will not be in the next u-boot rebase.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
Last posts moved here:

http://forum.doozan.com/read.php?3,35937

Please post your booting problem in a new thread.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)
Sorry, you can't reply to this topic. It has been closed.