GTI Mirabox and Netgear RN120/RN104 (Armada 370) Installation & Kernel Upgrade (Linux-6.1.8)

Hi Trond,

OK. Let me take a look.

-bodhi
===========================
Forum Wiki
bodhi's corner

bodhi,

Those options would also apply to the Kirkwood kernel for NETGEAR ReadyNAS NV+ v2 as well.

Ray

Hi Ray,

> Those options would also apply to the Kirkwood
> kernel for NETGEAR ReadyNAS NV+ v2 as well.

It's a bit late for Kirkwood kernel. Please make sure you boot up your ReadyNAS NV+ v2, pos tthe boot log, and make the request before next kernel release.

-bodhi
===========================
Forum Wiki
bodhi's corner

bodhi,

Not a problem for me a I usually build my own kernels using your patches. Just thought it would be helpful to have the config options there for any current or future users of your kernel packages who might be using with that device.

Ray

Hi Trond,

> I checked. Contrary to what I believed,
> 'ide-disk1' and 'ide-disk2' work perfectly on the
> RN102. Sorry for my confusion!

So for the RN104. Does disk-activity trigger cause all 4 LEDs to flash when only one HDD is accessed?

-bodhi
===========================
Forum Wiki
bodhi's corner

Hi Trond,

Here is the next kernel build (release version). Please try the LCD.

UPDATED: Please see 1st post for download link.

-bodhi
===========================
Forum Wiki
bodhi's corner



Edited 3 time(s). Last edit at 01/28/2023 06:49PM by bodhi.

Hi bodhi,

Today I upgraded both my RN102s to Linux 6.1.8. No issues. 'dmesg' outputs attached. Thanks!

I currently do not have access to my RN104 box, but I'll upgrade it in a few weeks time.

Regards,
Trond Melen

Kernel linux-6.1.8-mvebu-370xp-tld-1 package has been uploaded. Please see 1st post for download link.

-bodhi
===========================
Forum Wiki
bodhi's corner

Hi bodhi,

My Netgear RN104 has been upgraded to your latest Linux kernel 6.1.8 for Armada 370XP. No issues. Boot log attached. Thanks!

I can't find anything about it in the boot log, but the 16 by 2 character LCD display is now supported. It says "Linux-6.1.8-mveb" for about 3 seconds early in the boot process. Then it turns dark until it says "Power off." for about 1 second just before the box is shutting down. I've not yet figured out how to make it show anything else on the display, but I've posted my findings.

The LED indicators for the disk bays can be turned on and off through the states 'default-on' and 'none'. For the RN102 they can indicate disk activity in the corresponding bay through the states 'ide-disk1' and 'ide-disk2', but this fails for the RN104. 'ide-disk1' and 'ide-disk2' do not work, and 'ide-disk3' and 'ide-disk4' are not defined. Selecting 'disk-activity' make all 4 LEDs indicate activity on the SSD connected to the eSata port, so you're close!

To avoid the system clock lagging, it still must be synchronized with the RTC every minute using 'crontab'. 'sleep 600' finishes in 10:55 minutes independently of the system clock:

tme@rn104:~$ sudo crontab -l | tail -1
#* * * * * /sbin/hwclock -s

tme@rn104:~$ ssh root@router date; sleep 600; ssh root@router date;
Sat Feb 11 08:46:58 CET 2023
Sat Feb 11 08:57:53 CET 2023

tme@rn104:~$ sudo crontab -l | tail -1
* * * * * /sbin/hwclock -s

tme@rn104:~$ ssh root@router date; sleep 600; ssh root@router date;
Sat Feb 11 08:59:02 CET 2023
Sat Feb 11 09:09:57 CET 2023

I'm uncertain if this indicates that the RN102/RN104 is under-clocked so that the CPU performance is degraded. When requested to generate 2 GiB of random numbers and dump them, the box overestimates it's own performance:

tme@rn104:~$ ssh root@router date -uR; echo -ne '\a'; dd if=/dev/urandom of=/dev/null bs=32k count=65536; echo -ne '\a'; ssh root@router date -uR
Sat, 11 Feb 2023 10:07:45 +0000
65536+0 records in
65536+0 records out
2147483648 bytes (2.1 GB, 2.0 GiB) copied, 59.5075 s, 36.1 MB/s
Sat, 11 Feb 2023 10:08:51 +0000

Since it finishes in about 66 s on a wall clock (in my case the router's clock) rather than 59.5 s on it's own clock, the actual data rate is about 32.5 MB/s rather than 36.1. How many seconds does you Mirabox spend on the same task?

Regards,
Trond Melen

EDIT: Typos only. :TIDE



Edited 2 time(s). Last edit at 02/11/2023 10:39AM by tme.

Hi Trond,

> I can't find anything about it in the boot log,
> but the 16 by 2 character LCD display is now
> supported. It says "Linux-6.1.8-mveb" for about 3
> seconds early in the boot process. Then it turns
> dark until it says "Power off." for about 1 second
> just before the box is shutting down. I've not yet
> figured out how to make it show anything else on
> the display, but I've posted my
> findings.

Please see my comments in that thread.

> The LED indicators for the disk bays can be turned
> on and off through the states 'default-on' and
> 'none'. For the RN102 they can indicate disk
> activity in the corresponding bay through the
> states 'ide-disk1' and 'ide-disk2', but this fails
> for the RN104. 'ide-disk1' and 'ide-disk2' do not
> work, and 'ide-disk3' and 'ide-disk4' are not
> defined. Selecting 'disk-activity' make all 4 LEDs
> indicate activity on the SSD connected to the
> eSata port, so you're close!

OK I think we can try to control the RN104 2 LEDs the same way with the RN102 to see if works. I'll compile another DTS to set it in the defautl trigger.

Now I think it's the time for me to patch that part of the kernel again to support 4 HDD LEDs. I did not bother to do that since nobody has requested 4 triggers :) And I don't have any 4-bay NAS (no need to).

> RN102/RN104 is under-clocked so that the CPU
> performance is degraded.

I also briefly thought about this. But if it was not up to the spec, then Marvell would have downgraded it to the next processor speed, as in we have the same Kirkwood SoC but CPU is 1Ghz, 1.2Ghz, ...

This one is spec at 1.2Ghz. I wonder if there is a way to underclock it to 1.Ghz.

In the Wiki thread, there was some example with the Pogo V4 where people over clock the chip

Quote

Overclock

Overclock the Pogo V4/Mobile to 1Ghz
Pogo V4/Mobile overlocked with USB2TTL adaptor
Pogo V4/Mobile overlocked with no resistor needed

-bodhi
===========================
Forum Wiki
bodhi's corner



Edited 1 time(s). Last edit at 02/11/2023 04:38PM by bodhi.

> Now I think it's the time for me to patch that
> part of the kernel again to support 4 HDD LEDs. I
> did not bother to do that since nobody has
> requested 4 triggers :) And I don't have any 4-bay
> NAS (no need to).

I think if I will work on this, it will be a different patch altogether (do it in a proper way, not a hack as I did from way back).

-bodhi
===========================
Forum Wiki
bodhi's corner

Hi Trond,

> I'm uncertain if this indicates that the
> RN102/RN104 is under-clocked so that the CPU
> performance is degraded. When requested to
> generate 2 GiB of random numbers and dump them,
> the box overestimates it's own performance:
>

> tme@rn104:~$ ssh root@router date -uR; echo -ne
> '\a'; dd if=/dev/urandom of=/dev/null bs=32k
> count=65536; echo -ne '\a'; ssh root@router date
> -uR
> Sat, 11 Feb 2023 10:07:45 +0000
> 65536+0 records in
> 65536+0 records out
> 2147483648 bytes (2.1 GB, 2.0 GiB) copied, 59.5075
> s, 36.1 MB/s
> Sat, 11 Feb 2023 10:08:51 +0000
>

>
> Since it finishes in about 66 s on a wall clock
> (in my case the router's clock) rather than 59.5 s
> on it's own clock, the actual data rate is about
> 32.5 MB/s rather than 36.1. How many seconds does
> you Mirabox spend on the same task?

root@Mirabox:~# time dd if=/dev/urandom of=/dev/null bs=32k count=65536
65536+0 records in
65536+0 records out
2147483648 bytes (2.1 GB, 2.0 GiB) copied, 98.914 s, 21.7 MB/s

real	1m38.923s
user	0m0.090s
sys	1m38.722s

And the 1m38.923s is about the same as on my phone stop watch (1m39s).

-bodhi
===========================
Forum Wiki
bodhi's corner