iMX233-OLinuXino-MICRO stability issues (kernel oops)

jap · September 30, 2012, 04:30:50 PM

archlinux was constantly crashing on the micro board. Removing R17 fixed this issue for me. No oops was seen after the fix.

Thank you!

dpwhittaker · October 01, 2012, 12:52:50 AM

15 hours of top -d 0 with the resistor removed. This seems to be a good fix. I'd recommend doing this on any future production runs.

Kean · October 01, 2012, 07:18:25 AM

This sounds like an easy fix, and I will test it tomorrow.

@Lub/OLIMEX - can you explain what R17 is for? It is also used on the Maxi and doesn't cause problems.

Kean

LubOlimex · October 01, 2012, 10:47:36 AM

Hey guys,

At the end of differential clock signals it is usually good to have a terminating resistor to stop signal bouncing and interference (electrical termination).

As far as I can see everybody here agreed that removing R17 improved the stability issues, and will suggest to the board designers to remove it in the next revision of the board.

Thank you guys for the help and the testing!

Cheers,
Lub

fred · October 01, 2012, 06:31:40 PM

Hey guys from olimex,

Is that all ?

- Is the board really stable ?

- Who informs the other users regarding this error ?

- What happens to the boards at the resellers ?

...

Fred

stefan · October 02, 2012, 10:49:08 AM

An Application Note from Micron states that terminating resistor is not required on single DDR IC.

olimex · October 02, 2012, 11:20:10 AM

@Fred we still do tests to be sure that removing the termination resistor solves this problem
once this is identified we will change the user manual, wiki and the schematic in the next run
the boards which are by the distributors will be delivered with R17 of course and customer should do small intervention to remove it if this is the cause of the problem

andersop · October 05, 2012, 02:40:14 AM

After I removed R17 I have not seen the kernel oops. I am running a stress tester on my application and the board does occasionally restart - but it simply reboots with no oops message, so I am not 100% sure if this is a hardware stability problem or some issue with my application... just my experience so far.

Kean · October 05, 2012, 06:29:58 PM

Since removing R17 on one of my MICRO boards, I can report 2.5 days perfect uptime. Fantastic!

cnoviello · October 09, 2012, 12:39:05 PM

Hi,
I'm experiencing stability issues with some Olinuxino-Maxi boards. I exclude that problem is related to software, since it runs correctly on other boards. It seems that kernel freeze and there is no way to restart board. The only solution is powering off it. Do you think that removing R17 resistor can improve stability of Maxi too?

Thanks

andersop · October 09, 2012, 11:48:41 PM

Quote from: cnoviello on October 09, 2012, 12:39:05 PM
Hi,
I'm experiencing stability issues with some Olinuxino-Maxi boards. I exclude that problem is related to software, since it runs correctly on other boards. It seems that kernel freeze and there is no way to restart board. The only solution is powering off it. Do you think that removing R17 resistor can improve stability of Maxi too?

Thanks

It can't hurt to try. I think at this point we (the community / olimex) are not completely sure about this fix so more real-world test data would always be helpful. Removing R17 is very simple with a hot-air station, and moderately easy with a decent soldering iron/wick, so give it a try and post back with your findings!

albertdh · October 10, 2012, 04:08:55 AM

I too have been seeing stability problems with my iMX233-OLinuXino-MICRO. My problems manifested as crashes soon after the USB host driver would shut down the USB system. Sometimes the kernel would manage to dump an OOPS and other times not. The module would not run linux for more than 20 minutes before trouble showed up.

I got the same sort of behavior from the 2.6.35-x kernels and the 3.6 kernel on github [https://github.com/koliqi/imx23-olinuxino]. Thank you to 'kolqui' on github for the port! The same problems happened for the Debian, Arch Linux, and OpenEmbedded images provided by Olimex

On Friday I removed R17 with a pair of soldering irons. I have been running gpsd against a USB GPS reciever with local and remote clients and an ssh client over a USB to ethernet link [3]. Uptime says the combination has been running for 3.9 days[2] and the kernel logs are still clean. This is with currently up to date Arch Linux with gpsd and a few more networking packages than originally provided.

So, I think the removal of the resistor has made the MICRO reliable enough to develop into a kite aerial photography controller and wireless relay. Thanks Olimex for the chance to try!

Albert.

[1] https://github.com/koliqi/imx23-olinuxino

[2] 19:47:49 up 3 days, 22:59, 3 users, load average: 0.98, 0.88, 0.87

[3] a dm9601 USB 1.1 to 100BaseT cheap unit. I had to configure it into the kernel myself, but there are good recipes on the net for that.

Edited to change MINI to MICRO. AdH

petri · December 13, 2012, 07:04:02 PM

I had a probelems also with early MAXI version of the board. I just removed R17 and it *seems* to be fixed instability problems. I'll report if it didn't, but for now with uptime of 22min (record) it really looks good!

ritzBerl · June 10, 2013, 02:23:15 PM

any updates from olimex if removing R17 solves this issue (stable and reliable)?

Chax · June 24, 2013, 07:45:31 PM

I also had the strangest problems, gcc failing to compile a simple program on the Olinuxino Maxi.

I removed R17 from my Maxi board but I *THINK* the main reason for the failures had something to do with u-boot. I used some vanilla 2013.1 version but now the wiki by Robert Nelson (http://www.eewiki.net/display/linuxonarm/iMX233-OLinuXino) also includes some voltage fixes for the iMX23 and u-boot. A couple of days ago I recompiled uboot and this seems to solve the strange problems for me.

I have not run any stress tests yet. I will do this in a couple of weeks with the micro and the maxi board I own.