182 replies to this topic

[Meknes]

I don't have to deal with your arguments as I already wrote now 3 times that this is a Plugin support thread and we become OT here, if you don't contribute to the plugin and want to discuss your arguments you are welcome to open your own thread.
 
@luis67
 
Regarding the boot proplems - we start to run out of options, the only remaining thing to try to make ubifs work on your box would be to flash an orginal Image and convert a virgin PLi instead of a backup - and you didn't answer my question how big your backup image is.

42 mb is virgin and 59 mb is backup open pli image,i tried with vigrin and backup and the result is the same.

I have the same problem dm800se

[Lost in Space]

Look, if you only post 'doesn't work' then this 'doesn't help'.

 

Bootlogs and conversion script output (jtr.log on /tmp) would help, producing them to contribute on identifiying and maybe fixing the problem would be more helpfull.

Edited by gutemine, 18 April 2013 - 19:26.

[Persian Prince]

UBIFS for dreambox : https://sourceforge....26be/log/?path=

 

[ims]

I do not know, but build with this do not boots on dm800se more...

[Lost in Space]

Why revert such 'great PLi inventions' like compresion=none for boot?

 

And if you don't apply the recent broadcom nand patches which avoid double bad block handling the fun will not last long.

 

And if you don't have a working block2mtd you will suffer from nandsim forever.

 

It is not only copy & paste ...

Edited by gutemine, 18 April 2013 - 19:38.

[gweny]

thank you very much Persian Prince 

Edited by gweny, 19 April 2013 - 05:35.

[Dimitrij]

offtopic...

Question gutemine.

 

I am often asked about the multiboot for ET.

Now it does not work because of one reason, not mounted device.

root@et9x00:~# df -h
Filesystem                Size      Used Available Use% Mounted on
ubi0:rootfs             105.1M     94.3M     10.8M  90% /
devtmpfs                142.0M      4.0K    142.0M   0% /dev
/dev/sda1                64.0K         0     64.0K   0% /media/hdd
df: /media/sdb1: No such file or directory
df: /media/xtrenboot: No such file or directory
df: /media/usb: No such file or directory
tmpfs                    64.0K         0     64.0K   0% /media
tmpfs  

As I understand the reason block2mtd.

If so, what should be done, for what would the driver was built into the software PLi.

A lot of people came out in support of this patch.
But first, he needs a patch ...

[Lost in Space]

I already wrote now multiply that PLi doesn't want any Multibooting and hence I don't support any Multibooting here.

 

If you don't understand how chroot booting and mounting in such an early stage of the boot process works it is simply ...difficult.

 

Even the great BA patcher(s) for clones are not able to fix simple things like PLi support (2 lines of code in the shellscript) - so maybe ask them :-)

 

And this has nothing to do with block2mtd driver. block2mtd simply makes ubifs extraction much easier then doing it with nandsim. And all the patches for block2mtd are publically available - you only need to apply them correctly and then simply USE the resulting driver

Edited by gutemine, 19 April 2013 - 17:53.

[Dutchdude]

@Gutemine,

 

I was planning to test the upgrade with kernel hold, but as Openpli now has UBIFS in their

build there is no point anymore with currently over 30 kernel updates to put them all on hold.

 

Thx for all your help.

 

DD

[Lost in Space]

You are welcome, as JTR was a workaround to have an ubifs look and see aproach. It was never intended to be a permanent solution of any kind.

 

gutemine

Edited by gutemine, 20 April 2013 - 09:22.

[Dimitrij]

I already wrote now multiply that PLi doesn't want any Multibooting and hence I don't support any Multibooting here.

 

If you don't understand how chroot booting and mounting in such an early stage of the boot process works it is simply ...difficult.

 

Even the great BA patcher(s) for clones are not able to fix simple things like PLi support (2 lines of code in the shellscript) - so maybe ask them :-)

1)I do not care who and what he thinks about multibut on openPli.
I do this for the users.

2)Yes, it's right up my street, and I do not hide it.
As should look these two lines shellscript?

3)I do this one, and so I ask you for help.
I know that you are a pro in this matter.

Edited by Dima73, 21 April 2013 - 07:15.

[Lost in Space]

Well, the problem in this case is that this is also a kind of test if and how long it takes 'them' to fix it - like this thread is a test for the ubifs changes.

 

And tests should not be prematurely terminated, just because the observer knows the answers to the questions

 

Mabye this is not fair, but there is a reason behing everything - and in case of Multibooting it is pretty obvious.

[Lost in Space]

Update:

 

The 7020hd and 8000 now have latest drivers and kernels and ubifs already in the current OE 3.0 build.

 

As the upgrade now works too instead of flashing a current image if you want to keep you existing image setup you can simply do an ubifs conversion after the upgrade with dFlash by changing backup tool from mkfs.jffs2 to mkfs.ubifs and flash the resulting backup.

 

As now also 800se and 500hd have at least latest drivers and kernel with ubifs support if you want to convert these images to ubifs (as they stay to be shipped with jffs2) dFlash will also do the job with backup/restore as explained above.

 

So there is no need for using JTR any more and we could close this thread to prevent further confusion.

 

gutemine

Edited by gutemine, 21 April 2013 - 16:07.

[isson]

it's working

 

root@dm500hd:~# df -h
Filesystem                Size      Used Available Use% Mounted on
ubi0:rootfs              41.3M     29.5M     11.8M  71% /
devtmpfs                 66.1M      4.0K     66.1M   0% /dev
tmpfs                    64.0K         0     64.0K   0% /media
/dev/sda1                14.5G    494.9M     13.3G   4% /media/usb
tmpfs                    66.3M    136.0K     66.1M   0% /var/volatile

 

 

[Lost in Space]

Yes, why not, as with OE 2.0 Images this is working with dFlash now since 2 weeks already and with Marsu even longer.

 

PS: But don't tell the others that you still have that many Freespace, as it should be < 5MB - at least I was told this is the reason why PLI doesn't want ubifs on 500hd and 800se

Edited by gutemine, 21 April 2013 - 17:40.

[mika-nl]

it's working

 

root@dm500hd:~# df -h
Filesystem                Size      Used Available Use% Mounted on
ubi0:rootfs              41.3M     29.5M     11.8M  71% /
devtmpfs                 66.1M      4.0K     66.1M   0% /dev
tmpfs                    64.0K         0     64.0K   0% /media
/dev/sda1                14.5G    494.9M     13.3G   4% /media/usb
tmpfs                    66.3M    136.0K     66.1M   0% /var/volatile

 

What did you remove to get free space ?

 

MiKa

[Lost in Space]

Probably nothing, with zlib compression you easily get 9MB Freespace and with xz compression you get 1-2MB more of free space but you need some patience on the backup.

 

zlib compressed ubifs is still faster then jffs2 with zlib and compareable to jffs2 with lzo as PLi has and it still boots faster, only xz will cost you half of the boot time gains.

 

And don't forget to remove dFlash after the Migration for a fair compare.

 

But the checked in recepies only copied the favor_lzo compression from DMM which is pretty useless on the boxes with small Flash

Edited by gutemine, 21 April 2013 - 22:17.

[pieterg]

But the checked in recepies only copied the favor_lzo compression from DMM which is pretty useless on the boxes with small Flash

Not entirely true, you should tweak it to your needs of course, you can simply gain another ~8MB, as we have. But that's not the point, for the average user the few seconds boot time improvement do not outweigh the loss in rootfs space, because of the ubi and ubifs overhead. UBI(FS) was designed for big nand devices, on a 61MB partition the overhead is just too much.

So we'll keep the 500hd/800se at jffs2, for the benefit of the needs of the average user.

[Lost in Space]

I don't have a problem with this decision as DMM did the same for their OE 2.0 on these boxes.

 

I only disliked the argument that only 5MB remain - which is simply not true.

 

And that what you call UBIFS overhead simply improves the performance. And it is NOT only boottime. That what you save Flashspace with jffs2 are structures that then need to be rebuilt in Memory. Therefore ubifs also saves some Memory when then box is running already, which improves enigma2 agility, especially for users running memory eating things like CrossEPG,...

 

And with dFlash the users still can covert the images on these boxes to UBIFS even when it isn't delivered out of the box.

 

gutemine

Edited by gutemine, 22 April 2013 - 08:43.

[robi62]

Does this work as well on the old dm800??

It would  be nice for this old box to bootup faster...........as it takes a few miutes on OE2

Edited by robi62, 22 April 2013 - 19:13.