Welcome! Log In Create A New Profile

Advanced

Linux Kernel 4.4.89 Oxnas (OX820) Pogo V3 package and Debian rootfs

Posted by bodhi 
It is sort of a combination. You have to copy some files and then apply the main patch.

You git clone the repo and find the patch itself under "/target/linux/oxnas/patches-3.14".

Under "/target/linux/oxnas/files" you find the missing files or rather the directory structure that you need to copy directly into your kernel source tree.

And if I remember correctly that should be it.
Copy the files and then apply the patch.

@Daniel:
Anything else that needs to be done?
I am of course talking solely about compiling the kernel. Ignoring any other openwrt related stuff, as me personally, I am only interested in running Debian on my Pogos.
Of course I still HIGHLY value the work that kref and now you as a member of the openwrt community are doing and/or have been doing. ;-)
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 14, 2014 03:07PM
Hello Bodhi,
I would like to test the new uboot and kernel with booting from SATA. This means I don't want to change any NAND content. Can you please tell me what to do to get this working.

- shv
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 22, 2014 02:11PM
bodhi,
I didn't try that yet. ALARM strategy was to put stage1 and uboot on specific sector location on the hard drive. Maybe I can just replace the uboot with the new one while using the old stage1 primary loader. I want to prevent a Fat32 partition just for boot files.

Another strategy could be to make a uboot_spl.bin primary loader for SATA which loads the uboot and uboot env from flash. Do you think that I can use your uboot.spl.bin for that strategy? What clock frequency set with your primary loader?

- shv



Edited 2 time(s). Last edit at 07/24/2014 10:49AM by shv.
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 25, 2014 12:01AM
Shv,

I dont think my spl image works in SATA mode. But you could try. NAND and SATA versions are built differently. I'd doubt that it would work.

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
bodhi,
if I try to compile your git sources with ext4 or fat configuration it stops due to compilation problems. NAND and SATA configuration seams to be OK.

-shv
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 25, 2014 02:31PM
shv Wrote:
-------------------------------------------------------
> bodhi,
> if I try to compile your git sources with ext4 or
> fat configuration it stops due to compilation
> problems. NAND and SATA configuration seams to be
> OK.
>
> -shv

I've just found the same thing. There's compilation problem with FAT. I'm not sure yet but I think #define's has some problem when I use ox820_fat_config.

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 25, 2014 03:39PM
@shv,

I found an inconsistency between NAND and FAT compilation options. That was why the LED flashing code that I've added compiled for NAND but not for FAT. U-Boot has built OK now. Would you like to test it?

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
@bodhi,
yes I would like to test it but it seams not to be in your git repository yet.

-shv
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 26, 2014 02:02AM
@shv,

I did not want to commit to GitHub before testing. I'm having problem booting with the SATA spl. It did not get loaded by ROM during powered up. I don't know if my eSata enclosure is the culprit or not (I don't have a 22 pin-eSATA+power cable to try a direct connect right now).

Here attached is the updated board/ox820/ox820.c file. You could try to rebuild with it. If you have any problem rebuilding, I could upload binaries to Dropbox.

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
Attachments:
open | download - ox820.c (9.9 KB)
@bodhi,
we should continue uboot related topics in the right thread: http://forum.doozan.com/read.php?3,16017,16831#msg-16831

-shv
IDE performance compared to 2.6.31.14 kernel:

write: dd if=/dev/zero of=/media/ide0/test.tmp bs=500K count=1024
read: dd if=/media/ide0/test.tmp of=/dev/null bs=500K count=1024

2.6.31.14 (850MHz) - old test results
read: 82 MB/s
write: 70 MB/s

3.15.3 (850MHz):
read: 59 MB/s
write: 50 MB/s



Edited 7 time(s). Last edit at 07/26/2014 12:55PM by shv.
I will try to get my mini pcie sata and/or USB3.0 controllers working with my Pogo V3 Pro the next few days. I am really interested in seeing how these will perform with the new kernel. Or first if they will work at all.

@shv:
Still very interesting to see that the old kernel performs best, when it comes to SATA..
@ingmar_k,
at least with ext4. Navym wrote somewhere that the performance of ext2 is much better.
you can find my performance tests on navyms kernel within some messages in the following thread: http://archlinuxarm.org/forum/viewtopic.php?f=55&t=6193&start=20#p35077

@bodhi,
the write perfomance compared with the older kernel form navym seams to be the same but the read performance seams to be a bit worse.

for information: http://archlinuxarm.org/forum/viewtopic.php?f=55&t=6193&start=30#p35164



Edited 7 time(s). Last edit at 07/26/2014 12:50PM by shv.
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 26, 2014 01:52PM
@shv,

Nice works! interesting to see the hard numbers. Have you tried Ext2 to see if read performance is back to normal?

@ingmar_k,

Quote

Still very interesting to see that the old kernel performs best, when it comes to SATA..

Indeed! I'm wondering could the OXNAS SATA driver be improved somehow.

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
I woud guess that the implemented DMA mechanism simply was better suited to the CPU, in the stock kernel patches.
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 26, 2014 02:35PM
ingmar_k,

Quote

> I woud guess that the implemented DMA mechanism
> simply was better suited to the CPU, in the stock
> kernel patches.

Agree. That sounds very probable to me too. Too bad if we have to go back to Ext2 to get the performance back.

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
I have a problem with one Pro plug that the network startup is not working.
Setting up networking....
Configuring network interfaces...SIOCSIFHWADDR: No such device
Failed to bring up eth0.
Any idea?

-shv
If I remember correctly, the old implementation also used some sort of caching mechanism.
Something maybe like CMA. That could account for the Performance difference.

@shv:
MAC-Address correctly set in /etc/network/Interfaces?
The error message reads a bit like the MAC address could not be determined and/or set. But I am just guessing.



Edited 2 time(s). Last edit at 07/26/2014 03:31PM by ingmar_k.
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 26, 2014 03:32PM
shv,

Quote

network startup is not working.

Does the /etc/network/interfaces contain the MAC address?

Quote

hwaddress ether XX:XX:XX:XX:XX:XX

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
@bodhi,

Yes, the MAC address is set in /etc/network/interfaces. I have one Squeeze rootfs where the folloing where the problem occurs if the MAC address in the boot environment and /etc/network/interfaces are the same. If I use certain values values it works.
boot.env: ethaddr=00:25:31:01:66:5F
/etc/network/interfaces: hwaddress ether 00:25:31:01:C9:74
If I adapt ethaddr to hwaddress ether the problem occurs. Can you please try to reproduce that problem?

With another Squeeze rootfs this "workaround" does not work.

There is no problem if I use my old kernel within both rootfs if "ethaddr" = "hwaddress ether".

-shv



Edited 2 time(s). Last edit at 07/27/2014 04:31AM by shv.
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 27, 2014 01:34PM
@shv,

I do have both ethaddr and /etc/network/interfaces set to the same value. No problem on kernel 3.15.3. And it is the same rootfs that I have since squeeze.

I think if you look at serial console output closely, it might give out some hints.

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
I decided to play around with my other Pogoplug - this one is a non-PCI Pro model. I noticed that when I run fw_printenv, I get an ominous error message:
Warning: Bad CRC, using default environment
bootcmd=bootp; setenv bootargs root=/dev/nfs nfsroot=${serverip}:${rootpath} ip=${ipaddr}:${serverip}:${gatewayip}:${netmask}:${hostname}::off; bootm
bootdelay=5
baudrate=115200

I don't have a serial cable, so I'm nervous about running fw_setenv and rendering the unit unbootable. Any tips?
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 27, 2014 09:12PM
agiorgio,

> I don't have a serial cable, so I'm nervous about
> running fw_setenv and rendering the unit
> unbootable. Any tips?

Pls describe what are you trying to do, preferably with the full log. The CRC error usually indicates that the envs area in NAND has not been saved, therefore has no checksum.

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
I decided to upgrade this pogoplug to Debian Wheezy, and I wanted to setup the netconsole while I'm working on it. However, I didn't know if I should run fw_setenv with that ominous message.
Which uboot and which kernel are currently on the device? With the old uboot the netconsole doesn't work. With the old kernel you must use blparam instead of fw_setenv! If you have the old ALARM uboot and kernel image in the flash you could use my Wheezy rootfs: http://pogoplug.cwsurf.de/forum/viewtopic.php?f=5&t=114

-shv
Re: Linux Kernel 3.15.3 Pogo V3 (Oxnas OX820) package and rootfs (Flattened Device Tree)
July 28, 2014 01:13PM
agiorgio,

Quote

> I decided to upgrade this pogoplug to Debian Wheezy

Did you follow a procedure, instruction …? which u-boot is currently on your Pogo? if not, which system were you running before, and what exactly did you do before seeing CRC error with fw_printenv?

There is not enough information, so a wrong advice could mess up your u-boot envs and brick your box!

-bodhi
===========================
Wiki
latest Kirkwood kernel builds and rootfs
latest u-boot-kirkwood builds
latest Oxnas kernel builds and rootfs
latest u-boot-oxnas builds
latest MVEBU Armada kernel builds and rootfs
U-Boot & Kernel Booting process
bodhi's u-boot GitHub
bodhi's corner
The SATA performance issues come down to this:
The manufacturers kernel had hacked SATA transfers through the SRAM instead of DMA because the DMA is not directly hooked to the oxsemi SATA controller on the SoC. It was an ugly hack, but it made it work at a much better performance. Only problem, was it hit SO many inner workings of the kernel that it was a total PITA to port forward. I mean it hit A LOT.
The plug was running Debian Squeeze, and I replaced the root filesystem on the USB hard disk with shv's Wheezy one. That seemed to work perfectly. It's been a while since I hacked that plug to run Debian, and I don't remember exactly what I did. I'll probably end up just leaving it alone, since I'm just using it as a remote backup device.
Author:

Your Email:


Subject:


Spam prevention:
Please, enter the code that you see below in the input field. This is for blocking bots that try to post this form automatically. If the code is hard to read, then just try to guess it right. If you enter the wrong code, a new image is created and you get another chance to enter it right.
Message: