Possible New Zaurus Kernel For Oe

[weasel123]

I have noticed that the sl c3000 uses kernel 2.4.20 and the sources are available on sharps site.
I would like to ask wether it is possible to use the new kernel  (2.4.20) with future OE releases, than the old 2.4.18 rmk one.

 IMHO the only reason we are using the 2.4.18 is because of the binary only sd driver from sharp. We should ditch the SD thing (its a propriotory closed format) which has been hindering us from moving to a newer kernel for a long time. We should instead write/use a existing MMC driver and forget about the whole SD thing.

There is a lot of hype about OZ moving to the 2.6 kernels, well that will anyway break the SD compatibility as there is no SD driver for the 2.6 yet, so in the meantime till a 2.6 zaurus kernel comes out I suggest that we move ahead with  newer 2.4 releases than that 2.4.18 release.

[Mickeyl]

2.4.20-embedix is a piece of crap. It doesn't compile for anything else than the C3000 target right now. There's no way we spend anymore seconds with this retarded code than absolutely necessary. Kernel 2.6 for C7x0 is 95% finished. We will ship it in OpenZaurus 3.5.3.

For the 5000 and the 5500, it doesn't look that good because of the SD thing. Then again, 2.4.20 doesn't contain collie code, so we wouldn't gain anything.

[allanjard]

Ah pants - and I just went and bought a new SD card. I could decide on SD or MMC since they were the same price... doh

This slap on head...

Looking forward to seeing the 2.6.x kernels though!

[mussi]

And what about SD support for the 6000 and the clamshell Zaurii in that case? I'd hate not to be able to use that 'slower' (but perfectly usable) storage.

[Joshp]

2.4.20-embedix is a piece of crap. It doesn't compile for anything else than the C3000 target right now. There's no way we spend anymore seconds with this retarded code than absolutely necessary. Kernel 2.6 for C7x0 is 95% finished. We will ship it in OpenZaurus 3.5.3.

For the 5000 and the 5500, it doesn't look that good because of the SD thing. Then again, 2.4.20 doesn't contain collie code, so we wouldn't gain anything.
[div align=\"right\"][a href=\"index.php?act=findpost&pid=72623\"][{POST_SNAPBACK}][/a][/div]

Where there be a 2.6 kernel for the sl-6000?

JP

[Bedman]

Hey,

i don't know if i'm the right person to say. But there is development for the 2.6 kernel of the SL-6000 device.
For more information look at the openzaurus-mailinglist:
http://sourceforge.net/mailarchive/forum.php?forum_id=11924

Ciao Bedman

[sashz]

2.4.20-embedix is a piece of crap. It doesn't compile for anything else than the C3000 target right now. There's no way we spend anymore seconds with this retarded code than absolutely necessary. Kernel 2.6 for C7x0 is 95% finished. We will ship it in OpenZaurus 3.5.3.

For the 5000 and the 5500, it doesn't look that good because of the SD thing. Then again, 2.4.20 doesn't contain collie code, so we wouldn't gain anything.
[div align=\"right\"][a href=\"index.php?act=findpost&pid=72623\"][{POST_SNAPBACK}][/a][/div]

well, this kernel works fine with pdaXrom on SL-C7x0/860  just need fix problem with MMC driver

[Mickeyl]

So you patched it to behave? Did you notice any differences?

[sashz]

So you patched it to behave? Did you notice any differences?
[div align=\"right\"][a href=\"index.php?act=findpost&pid=72781\"][{POST_SNAPBACK}][/a][/div]

There was little problem with PXA SSP initialization and kernel linking, i fixed it and added pdaXrom specific and part of japan patches - devfs support, optimized w100 modes, etc

unzip attachment, there bzip2ed archive with patches, because forum unlike attach bzip2 file

[weasel123]

Kernel 2.6 for C7x0 is 95% finished. We will ship it in OpenZaurus 3.5.3.

For the 5000 and the 5500, it doesn't look that good because of the SD thing. Then again, 2.4.20 doesn't contain collie code, so we wouldn't gain anything.
[div align=\"right\"][a href=\"index.php?act=findpost&pid=72623\"][{POST_SNAPBACK}][/a][/div]

IMHO why dont you forget about the SD support for collie with regard to the 2.6 kernels altogether and just use a genereic mmc driver for collie. I think that the main reason kernel developement fir collie has been retarded is because of the absurd binary only sd module.  

Its because of the binary sd module that we are still stuck with the ancient kernel 2.4.18 rmk-pxa7 etc. compiled using the 2.95 compiler too.

I think its high time we (OZ) like the guys at familiar just forget about supporting SD( propriotory format) and support mmc instead and move ahead. Its s small sacrifice to pay for the added benefits of a newer kernel.

[lpotter]

Kernel 2.6 for C7x0 is 95% finished. We will ship it in OpenZaurus 3.5.3.

I think its high time we (OZ) like the guys at familiar just forget about supporting SD( propriotory format) and support mmc instead and move ahead. Its s small sacrifice to pay for the added benefits of a newer kernel.
[div align=\"right\"][a href=\"index.php?act=findpost&pid=73382\"][{POST_SNAPBACK}][/a][/div]

I don't think the benefits of any new kernel outweighs having a properly working sd slot.
Users want a device that works, they don't really care about which kernel it uses.

[adf]

I'm not sure I understand the issue clearly.

That the new 2.6 won't support Sd is clear enough. ...

It will, however support mmc cards?

is there a serious performance difference? (sd is already slow enough)

My sd casn easily enough be replaced with an mmc, if needed.
(mmc cards are as low as us $68 at price watch)

I agree with potter that a working system should be a pretty high priority--- right now, fior example, I use sharprom on my 6k because it works.

If, though, mmc does work and oz also works well on my 6k, then i would be willing to buy an mmc and replace my sd--and call it good.

of course my last jaunt to the 2.6kernel stuff at sourceforge only showed work on the 5500, the 5600 and the 7&8 series clamshells....
  i hope this doesn't bode ill for 6k support...
2 cents from an end user...

[pframpton]

The SD slot on an SL5x00 doesn't support any of the SD features anyway (the Secure part), would an MMC driver read an SD card - I would have thought it would. As for speed, surely that's hardware based. Is the actual problem that you can't use a generic MMC driver with the Sharp SD reader chip?

Also, according to the website with info on 5x00 kernels, there are other things not working too. http://www.cs.wisc.edu/~lenz/zaurus/

Maybe someone should try building a kernel with MMC support and do some tests to see what speed/perforamnc you actual get, if it does work.

[Mickeyl]

Is the actual problem that you can't use a generic MMC driver with the Sharp SD reader chip?

Bingo. The chipset that drives the SD/MMC slot has zero publically available documentation. That means no SD/MMC for 5500 and 5000d until Sharp releases documentation and/or the driver source code - both which is very unlikely to happen.

Note that this is _no_ problem on all other models since they use standard chipsets.

[adf]

then would this mean the continuation of the 2.4 kernel (and separate crosscompiler) for the 5500 while the other models move to a 2.6 based better integrated OE build system?