ix4-300d infos & hack ... (bare debian & 3.16 kern) -> DONE

[benoitm974]

Well you were too fast at finding your solution

If you feel there are some missing step to accomplish this please complete the wiki or post pull request!

Thanks you'll see rhisnlitt'e nas is still running at home with uptime over 4 months until I reboot it for upgrades! Very stable.

I followed the instructions from the wiki about "Flashing for booting custom kernel from nand". Flashing went well, but as you can see below I couldn't mount rootfs.

Ok I done it:
stop autoboot and then:

Code: Select all

setenv bootargs_root "root=/dev/sda2 rw"
saveenv
reset

No I feel myself (I think) as Benoitm974 at the beginning of this topic as I continuously replying to my own posts



Well that's it, everything is working now! Thanks Benoitm974 for your work, and thanks Doufas for your encouragement. This was my first kernel compilation ;D

ps.: if anyone knows why jessie is not working, then please share with me.

[Gyuri]

ps.: if anyone knows why jessie is not working, then please share with me.

I had to recompile kernel and enabling:
CONFIG_DEVTMPFS
and
CONFIG_CGROUPS

Now jessie is okay too.

[Gyuri]

Is it possible to use cpufreq on this machine? cpufreq-info says:

Code: Select all

analyzing CPU 0:
  no or unknown cpufreq driver is active on this CPU
  maximum transition latency: 0.00 ms.
analyzing CPU 1:
  no or unknown cpufreq driver is active on this CPU
  maximum transition latency: 0.00 ms.

and I find this:
https://github.com/torvalds/linux/blob/ ... -cpufreq.c

How can I step ahead?

[doufas]

Good to see you got it done. Ive been using wheezy fine and its very stable.

Any reason for wanting jessie?

[Gyuri]

Any reason for wanting jessie?

Wheezy reached EOL (April 26th 2016 (full) / May 2018 (LTS)). But my main reason is that I used some DVB stuff on bananapi, and it required newer packages from something.

The only thing what I'd like to achieve now is to save more power. It's good that I was able to boot from pendrive only. In this case I'll be able to send all HDDs to sleep when there is no activity. It would be nice to do something with the CPU too. Most of the time there is no activity, so in a yearly bases the 11W power consumption (vs bpi 2.5W) leads me more electricity bill. In Hungary 1W is about 1.1€/year, so "using sata port(s) on nearly doubled speed" costs me +9€/year. Okay-okay I know it's not much, but if it's possible to save then it would be nice to do.

So long story short: if anyone knows how to make cpufreq working on this thing, please share

[benoitm974]

Any reason for wanting jessie?

Wheezy reached EOL (April 26th 2016 (full) / May 2018 (LTS)). But my main reason is that I used some DVB stuff on bananapi, and it required newer packages from something.

The only thing what I'd like to achieve now is to save more power. It's good that I was able to boot from pendrive only. In this case I'll be able to send all HDDs to sleep when there is no activity. It would be nice to do something with the CPU too. Most of the time there is no activity, so in a yearly bases the 11W power consumption (vs bpi 2.5W) leads me more electricity bill. In Hungary 1W is about 1.1€/year, so "using sata port(s) on nearly doubled speed" costs me +9€/year. Okay-okay I know it's not much, but if it's possible to save then it would be nice to do.

So long story short: if anyone knows how to make cpufreq working on this thing, please share

Why don't you look for power sleep and lan wake on lan then ?

[Gyuri]

Why don't you look for power sleep and lan wake on lan then ?

It's less comfortable

[doufas]

Ultimately wouldn't that save more power in the end?

Also, you could boot from network too..

[Gyuri]

So long story short: if anyone knows how to make cpufreq working on this thing, please share

I didn't installed lib/modules into the rootfs... After doing it (& copied the kernel source too) cpufreq works.

Code: Select all

root@ix4-300d:~# cpufreq-info
cpufrequtils 008: cpufreq-info (C) Dominik Brodowski 2004-2009
Report errors and bugs to cpufreq@vger.kernel.org, please.
analyzing CPU 0:
  driver: cpufreq-dt
  CPUs which run at the same hardware frequency: 0
  CPUs which need to have their frequency coordinated by software: 0
  maximum transition latency: 1000 us.
  hardware limits: 667 MHz - 1.33 GHz
  available frequency steps: 667 MHz, 1.33 GHz
  available cpufreq governors: conservative, userspace, powersave, ondemand, performance
  current policy: frequency should be within 667 MHz and 1.33 GHz.
                  The governor "ondemand" may decide which speed to use
                  within this range.
  current CPU frequency is 667 MHz (asserted by call to hardware).
  cpufreq stats: 667 MHz:99.04%, 1.33 GHz:0.96%  (261)
analyzing CPU 1:
  driver: cpufreq-dt
  CPUs which run at the same hardware frequency: 1
  CPUs which need to have their frequency coordinated by software: 1
  maximum transition latency: 1000 us.
  hardware limits: 667 MHz - 1.33 GHz
  available frequency steps: 667 MHz, 1.33 GHz
  available cpufreq governors: conservative, userspace, powersave, ondemand, performance
  current policy: frequency should be within 667 MHz and 1.33 GHz.
                  The governor "ondemand" may decide which speed to use
                  within this range.
  current CPU frequency is 667 MHz (asserted by call to hardware).
  cpufreq stats: 667 MHz:99.32%, 1.33 GHz:0.68%  (175)

As you can see it's already running on the minimum freq. (page84: http://www.marvell.com/embedded-process ... 230_OS.PDF)

I tried using the "tuning" function of powertop, but it couldn't save me more power.

About WOL: I still don't want to do the job with WakeOnLan, and I'am afraid I couldn't do it anyway. It seems for me that there is no hardcoded mac address in the nic (is this even possible?), so I won't be able to address the magic packets.

Code: Select all

dmesg
[    2.705049] mvneta d0070000.ethernet eth0: Using random mac address ae:c0:2c:0d:xx:xx
[    2.712703] mvneta d0074000.ethernet eth1: Using random mac address 36:65:ec:ef:xx:xx

Code: Select all

ethtool -e eth0
Cannot get EEPROM data: Operation not supported

I must accept, that this board is eating 11W without load & hdds. And that's it

[Mijzelf]

It seems for me that there is no hardcoded mac address in the nic (is this even possible?), so I won't be able to address the magic packets.

Yes, that is possible. Why bother to add a prom to the nic, if the intended platform already needs some flash to boot anyway?
Many embedded systems have the mac address(es) stored in the u-boot environment, and it's up to u-boot, or even the OS to read it and write it to the nic.

However, in case of a Marvel SoC, you might succeed using 00:00:00:00:00:30

[Gyuri]

I've got a strange boot behavior and I cannot solve the problem. Please help!

I copied the uImage into the nand, and my rootfs is on the 2nd partition of an ssd in sata slot 1.

I can boot only if uart is connected. If I'd like to boot "standalone", then the boot stops somewhere. If I turn on the nas, wait some time and connect uart to see what's up (connect uart cable and press enter), then I get:

Code: Select all

Unknown command '▒▒' - try 'help'
Marvell>>

Do you have any idea how to debug at this stage? Or what would be the right bootcmd to boot without uart connected?

Thanks,
Gyuri

My bootcmd now is:

Code: Select all

bootcmd=nand read.e $loadaddr 0x120000 0x400000; setenv bootargs $console $mtdparts root=/dev/sda2 rw rootdelay=5 ip=$bootargs_end $mvNetConfig; bootm $loadaddr;

Where the variables are:

Code: Select all

loadaddr=0x02000000
console=console=ttyS0,115200
mtdparts=mtdparts=armada-nand:0xe0000@0x0(uboot),0x20000@0xe0000(env),0x20000@0x100000(env2),0x400000@0x120000(zImage),0x400000@0x520000(initrd),0x3F200000@0xE00000(boot),0x40000000@0x0(flash)
bootargs_end=:::DB88FXX81:eth0:none
mvNetConfig=mv_net_config=2,(00:50:43:11:11:11,0:1),mtu=1500

[Mijzelf]

I think you have an electrical problem. Somehow the RX pin gets some noise, which is interpreted as a keystroke, stopping the boot. The serial cable has it's own pull down resistors to prevent that.

If you soldered any cables to the serial port, it might help to insert a pulldown resistor when your serial cable is not connected.

[Gyuri]

I think you have an electrical problem. Somehow the RX pin gets some noise, which is interpreted as a keystroke, stopping the boot.

It was that! I spent a few hours on this Thanks for your help!

[avieros]

so i built a kernel, which boots ok!
however, my rootfs (built with the instructions linked to) gets stuck on the sshd - it times out
then gives me an error about ID0

any chance someone, Gyuri, can link to a small rootfs image to test?

[Gyuri]

any chance someone, Gyuri, can link to a small rootfs image to test?

Sorry, but I couldn't share it. I followed the steps on the site what Doufas suggested, and it's worked for me.
https://blog.night-shade.org.uk/2013/12 ... hf-rootfs/

The only thing what I did to make jessie work is enabling CONFIG_DEVTMPFS and CONFIG_CGROUPS in the kernel config, and recompile (4.9.13).