29 replies to this topic

[WTE]

Some issues with ubifs existed also in Xtrend boxes. I've experienced corruption to root filesystem in some boxes after power off from the switch or power outage. But this was back when the first versions of ubifs images were used, haven't seen these problems for many months now. Maybe DMM needs some finetuning as well.

 

It's normal behavior of ubifs. Ubifs write the data not immediately to the disk.

To force it execute command sync.

 

Linux must never been shutdown by killing the power The sync was optimized in newer releases so probably that's why you didn't see this ubifs issue anymore

[malakudi]

Corrupting flash (making STB unbootable) because of a power outage is not an option. I don't know what changed, but this should never happen. You don't expect to have a UPS for the STB, do you?

Edited by malakudi, 31 March 2013 - 21:57.

[Lost in Space]

UBIFS can nicely recover power failures als tonns of Smartphones use it due to their big flash where dozends of users take out the batteries on a daily basis when it hangs to get it operational again.

 

You are forgetting that actually NOKIA is the original creator of ubifs to overcome the jffs2 limitations, and these guys know what they do.

 

That DMM forgot a sync in their umountfs script is something which was fixed within 1 day, so where is the problem in an experimental image?

 

Since then the thread you linked is dead ... except my suggestion to drop writenfi due to the bad block limitation and switch to nandwrite completely.

 

But off course there are also boxes with jffs2 suddenly not booting anyore, and ubifs will not do better on this job either - BUT just assuming that it is bad is not right, as CT, VU+, cann't be wrong :-)

Edited by gutemine, 31 March 2013 - 22:10.

[offnow]

An older image with the newer drivers does not boot anymore - it is like the box has no flash anymore...

 

And as I wrote before - it is not done with a simple "y" in kernel-config - some other things have to be done to compile an image with pli3.0

[kocus37]

http://en.wikipedia.org/wiki/UBIFS

 

http://www.linux-mtd.../doc/ubifs.html

 

coted from the site above:

 

"Power-cuts tolerance:

 

Year 2011 note: however, there is an unsolved unstable bits issue which makes UBI/UBIFS fail to recover after a power cut on modern SLC and MLC flashes. This issue has not been observed on older SLC NANDs back at the time UBI/UBIFS was being developed. Note, the below text is quite old and has been written before the unstable bits issue has been first discovered."

 

this information is update 16/03/2013

 

no solution yet.

 

UBIFS is faster than JFFS2 in all cases

 

  Overwhelmingly so during mount

JFFS2 makes more efficient use of space

Conclusion:

Use JFFS2 on small partitions (< 16 MB)

Use UBIFS in other cases

 

coted from http://www.embedded-...tems-slides.pdf 

dated 2009

 

JFFS2 is better for small partitions under 16mb but above ubifs is better in all except power-cuts problem, better in scalability,fast mount,write-back support,tolerance to unclean reboots,fast I/O,on-the-flight compression,recoverability,integrity.

[pieterg]

There are no problems with ubifs itself, just with oe's module-utils and volatiles init scripts which do not sync their initial states.

Those issues are solved in the latest oe-core (and openpli 4)

[Lost in Space]

You are not telling us any real news (and old slides and presentations) - do you want to read also the probably now never solved list of jffs2 problems?

 

But as I already said, if you are really scared, then please send e-mails to CT, VU+, and all the mobile phone vendors that they instantly should stop using UBIFS ... gues what will happen ?

 

And BTW MOST of the other boxes put their kernel and loaders into RAW flash, you know how well this is protected against corruption even when it is read only ?

 

The worst thing that can happen in all these cases is a re-flash, and I think nobody in this board is scared by such a situatuon ...

 

But we are running in circles - let the USERS simply try it out, if enough of them are using it on a daily basis your sample is big enough to do the judging. If you ware just sitting an waiting and repeating arguments - you will miss life :-)

 

And yes, most problems are either already fixed or in an acceptable state, so please stop worrying - or you should not enter any airplanes or cars either.

 

And in the JTR Thread I clearly said that oyu should use only the latest up-to-date images for conversion.

 

Happy Easter!

gutemine

Edited by gutemine, 31 March 2013 - 23:16.

[Lost in Space]

I also spend a couple of hours just to create the JTR Plugin PLi Special Edition, and yes I'm aware that it is not a simple Y in your OE to build complete images with these changes.

 

But all the informations are publically available and you have lots of examples for the other boxes too. And maybe I'm also testing your capabilities to adapt to changes and supporting the Dreambox community, as I do the same for the PLi images within my Plugins.

Edited by gutemine, 1 April 2013 - 18:35.

[offnow]

Feel free to make your own image and let us know the changes...

[Lost in Space]

I'm patient, you know ;-)