Help - Search - Members - Calendar
Full Version: Status Of 2.6 For Akita (sl-c1000) In Oz/gpe
OESF Forums > Distros, Development, and Model Specific Forums > Distro Support and Discussion > Angstrom & OpenZaurus
obergix
Hi.

I've tried to generate a kernel 2.6 for akita by rebuilding in openembedded (default "branch"/HEAD/trunk... sorry, not so much experienced with monotone).

Kernel is 2.6.14-git3.

It boots, then X/GPE starts, but then I have problems with the touchscreen being unresponsive...

Anyone with a suggestion on how to correct that ?

My gues is that it may lack some modules...

Any help welcome.
obergix
Hmmm... looks like it may be the same as : http://bugs.openembedded.org/show_bug.cgi?id=517
obergix
QUOTE(obergix @ Dec 11 2005, 12:50 PM)
It boots, then X/GPE starts, but then I have problems with the touchscreen being unresponsive...

*


looks like I had lost the touch screen calibration while rebooting from 2.4 kernel image to 2.6.

I used the keybord to launch a terminal, then su and launched xtscal.

After that, the pointer works again in X...

Hope this helps.

Anyway, if others have other issues/solutions with a 2.6 kernel on akita, I'd be glad to know...
Hrw
Touchscreen is different handled under 2.6 so it had to be recalibrated.

About other issues: upgrade to 2.6.15-rc5 (today added) so you will get finished ALSA sound.
obergix
QUOTE(Hrw @ Dec 11 2005, 05:12 PM)
About other issues: upgrade to 2.6.15-rc5 (today added) so you will get finished ALSA sound.
*


I tried monotone pull + monotone update, but nothing seems to have changed...

Where/how would I find it ?

Something else : looks like I have kernel 2.6.14-git3 and modules in /lib/modules/2.6.14-rc1 only ...

So depmod.26 won't work, and no modprobe possible sad.gif

I'm suspecting something incoherent somewhere...

Any clues ?
koen
QUOTE(obergix @ Dec 11 2005, 03:40 PM)
QUOTE(Hrw @ Dec 11 2005, 05:12 PM)
About other issues: upgrade to 2.6.15-rc5 (today added) so you will get finished ALSA sound.
*


I tried monotone pull + monotone update, but nothing seems to have changed...

*



http://handhelds.org/hypermail/oe/47/4772.html
obergix
QUOTE(koen @ Dec 11 2005, 05:55 PM)
QUOTE(obergix @ Dec 11 2005, 03:40 PM)
QUOTE(Hrw @ Dec 11 2005, 05:12 PM)
About other issues: upgrade to 2.6.15-rc5 (today added) so you will get finished ALSA sound.
*


I tried monotone pull + monotone update, but nothing seems to have changed...

*



http://handhelds.org/hypermail/oe/47/4772.html
*


Looks like it helps alot... my repo wasn't synchronised...

Had to start over rebuilding... will let you know of my progress when succeeds.

Many thanks.
obergix
QUOTE(obergix @ Dec 11 2005, 07:47 PM)
QUOTE(koen @ Dec 11 2005, 05:55 PM)

Looks like it helps alot... my repo wasn't synchronised...

Had to start over rebuilding... will let you know of my progress when succeeds.

*



The rebuild has succeeded. Thanks.

Now, I'll keep on testing if kernel 2.6 works...
obergix
I've generated in parallel a gpe-image in .dev and in .oz354 branches.

The latter (so the 3.5.4-RC) only generates a 2.4.20 kernel although I specified
CODE
KERNEL_VERSION = "2.6"
in local.conf...

Dunno if it's a bug or if no kernel 2.6 is supposed to be generated in that branch for akitas ?
koen
QUOTE(obergix @ Dec 13 2005, 07:52 AM)
I've generated in parallel a gpe-image in .dev and in .oz354 branches.

The latter (so the 3.5.4-RC) only generates a 2.4.20 kernel although I specified
CODE
KERNEL_VERSION = "2.6"
in local.conf...

Dunno if it's a bug or if no kernel 2.6 is supposed to be generated in that branch for akitas ?
*


It is, but in the branch it's ZKERNEL_VERSION: http://ewi546.ewi.utwente.nl/tmp/viewmtn/r...2611680e524a2e8
obergix
QUOTE(koen @ Dec 13 2005, 05:30 PM)
QUOTE(obergix @ Dec 13 2005, 07:52 AM)
I've generated in parallel a gpe-image in .dev and in .oz354 branches.

The latter (so the 3.5.4-RC) only generates a 2.4.20 kernel although I specified
CODE
KERNEL_VERSION = "2.6"
in local.conf...

Dunno if it's a bug or if no kernel 2.6 is supposed to be generated in that branch for akitas ?
*


It is, but in the branch it's ZKERNEL_VERSION: http://ewi546.ewi.utwente.nl/tmp/viewmtn/r...2611680e524a2e8
*


OK... btw, I've just noticed that : http://bugs.openembedded.org/show_bug.cgi?id=513

Will try applying this change to my local.conf
obergix
QUOTE(obergix @ Dec 13 2005, 09:46 AM)
Now, I'll keep on testing if kernel 2.6 works...
*


But first, it had to build, of course wink.gif

After I've changed ZKERNEL_VERSION, I've noticed a problem with rebuilding of linux-openzaurus-2.6.14-rc1 :

Fetch failed: http://www.rpsys.net/openzaurus/patches/co...3.patch;patch=1

Looks like the patch was archived from http://www.rpsys.net//openzaurus/patches/ to its archive/ subdir... easy to fix in the .bb ...

Stay tuned for more details wink.gif
obergix
Another patch which was moved as well : hx2750_base-r20.patch sad.gif

Solved again... but now, a show stopper :

No .../build/tmp/work/linux-openzaurus-2.6.14-rc1-r2/defconfig-akita sad.gif

OK, I guess this is it, and 2.6 is not ready yet for being built in that 3.5.4RC branch for akitas...

Will keep on playing in .dev as far as 2.6 is concerned, then.
lardman
Try copying the defconfig across from .dev, it's probably just a simple omission as most people with the cxxxx machines are working from .dev as it is adding more support for those machines.


Si
obergix
QUOTE(lardman @ Dec 14 2005, 01:51 PM)
Try copying the defconfig across from .dev, it's probably just a simple omission as most people with the cxxxx machines are working from .dev as it is adding more support for those machines.

*


Hmmm... looks like there's none in .dev 2.6 kernel packages' files sad.gif
lardman
Try something like this:

org.openembedded.dev/packages/linux/linux-openzaurus-2.6.14-git3/defconfig-cxx00

Si
rpurdie
I'd recommend anyone wanting to build 2.6 kernels for cxx00 devices to use the .dev branch. Recent 2.6 kernels have much better support for these devices but need too many changes to allow them to be imported into the release branch.

3.5.4 will not be released with a 2.6 kernel for the cxx00 but should be closely followed by a 3.5.4.1 release from the dev branch with images/2.6 kernels for these devices. A testing release of 3.5.4.1 is available at

http://ewi546.ewi.utwente.nl/mirror/sam.rp...urus/autobuild/

I'm aware the fstabs need fixing for borzoi/spitz (/home and /mnt/hdd should be automount) and that there are probably two copies of the hostap kernel module installed. Both have been fixed in OE since these images were generated.
obergix
QUOTE(lardman @ Dec 15 2005, 02:09 PM)
Try something like this:

org.openembedded.dev/packages/linux/linux-openzaurus-2.6.14-git3/defconfig-cxx00

Si
*


Thx.

I've integrated it in the .bb and changed accordingly linux-openzaurus.conf for it to be used.

The kernel compilation starts, but then stops on compilation error :
CODE
| sound/arm/pxa2xx-ac97.c: In function `pxa2xx_ac97_read':
| sound/arm/pxa2xx-ac97.c:52: warning: assignment from incompatible pointer type
| sound/arm/pxa2xx-ac97.c: In function `pxa2xx_ac97_write':
| sound/arm/pxa2xx-ac97.c:90: warning: assignment from incompatible pointer type
| sound/arm/pxa2xx-ac97.c: At top level:
| sound/arm/pxa2xx-ac97.c:184: warning: initialization from incompatible pointer type
| sound/arm/pxa2xx-ac97.c:192: warning: initialization from incompatible pointer type
| sound/arm/pxa2xx-ac97.c: In function `pxa2xx_ac97_suspend':
| sound/arm/pxa2xx-ac97.c:284: error: incompatible type for argument 2 of `pxa2xx_ac97_do_suspend'
| sound/arm/pxa2xx-ac97.c: In function `pxa2xx_ac97_probe':
| sound/arm/pxa2xx-ac97.c:353: warning: passing arg 2 of `snd_card_set_pm_callback' from incompatible pointer type
| make[2]: *** [sound/arm/pxa2xx-ac97.o] Erreur 1
| make[1]: *** [sound/arm] Erreur 2
| make: *** [sound] Erreur 2
| FATAL: oe_runmake failed


No much progress then in .oz354 ...
obergix
QUOTE(rpurdie @ Dec 15 2005, 10:33 PM)
A testing release of 3.5.4.1 is available at

http://ewi546.ewi.utwente.nl/mirror/sam.rp...urus/autobuild/
*

I've tested these (gpe / akita) and I've got problems resuming after a suspend.

When pressing the on button to resume, the mail led flashes, and that's all : no resuming sad.gif
lardman
Have you reported it on the bugtracker?

Si
obergix
QUOTE(lardman @ Dec 20 2005, 02:16 PM)
Have you reported it on the bugtracker?

Si
*


Not yet.. but I should.^H^H^H : done : #547 (http://bugs.openembedded.org/show_bug.cgi?id=547)

However, one previous problem with apm that I've reported doesn't seem to attract much interest : http://bugs.openembedded.org/show_bug.cgi?id=496 (nor the workaround suggested : http://bugs.openembedded.org/show_bug.cgi?id=497 ) ... not very encouraging for users sad.gif
lardman
There are only so many people with commit rights, and it's only those with a C1000 who will be willing to commit something (after they've tested it).

I also note that you make no explicit mention of the version of OZ you're running, nor the kernel version (I see the field which says OpenZaurus 3.5.3, but iirc there were no 3.5.3 images for the Akita, so this is a bit confusing).

The suspend script is useful, but ideally the apm behaviour would be fixed. Iirc, someone was talking about this bug on irc and noting that it may be a kernel issue rather than something to do with su.

If you have a specific issue which you think ought to be addressed, you'd be advised to get on irc on either #oe or #openzaurus and talk to the appropriate person (RP might be suitable, but you've not said what kernel version so I can't be sure).


Si
obergix
QUOTE
I also note that you make no explicit mention of the version of OZ you're running, nor the kernel version (I see the field which says OpenZaurus 3.5.3, but iirc there were no 3.5.3 images for the Akita, so this is a bit confusing).

Yeah, it happened on 3.5.4RC ... and btw, a corresponding value in version in bugtracker would be great wink.gif

And the 2.4 standard kernel (only one available for akita onthat one wink.gif

Sorry for not giving this detail in my report in the first place.

QUOTE
The suspend script is useful, but ideally the apm behaviour would be fixed. Iirc, someone was talking about this bug on irc and noting that it may be a kernel issue rather than something to do with su.

Indeed ... but in the meantime.... a usable workaround may help wink.gif
QUOTE
If you have a specific issue which you think ought to be addressed, you'd be advised to get on irc on either #oe or #openzaurus and talk to the appropriate person (RP might be suitable, but you've not said what kernel version so I can't be sure).

I do that from time to time... but expect developers to read the bug tracker sometimes too wink.gif

Thx.
lardman
They will read it, but as apm is closely related to the kernel they won't tend to respond unless they know this info, etc.

Anyway, I fear you may have troubles (with that kernel) as no-one really wants to fiddle with the rubbish Sharp kernel code. The 2.6.x kernel is the main thrust of work now and it's coming along nicely for the Akita afaik.

That said, as a temporary workround, your su thing might be acceptable if you must run with an old kernel - I believe the c1000 image will be released with a 2.4.20 kernel. Perhaps ask about this on irc and see whether any of the devs with a c1000 has had the same issue.


Si
obergix
FYI, interesting discussion about suspend/resume problems on akita with 2.6 in :
http://www.oesf.org/forums/index.php?showt...ndpost&p=108123
obergix
Just a note to report I've just tested gpe-image (gpe-image-akita-20060102193451.rootfs.img + zImage-2.6.14+2.6.15-rc7-akita-20060102193451.bin) generated in .dev on my akita.

The kernel stops the boot with the following messages when booted without the compact flash inserted :
CODE
VFS: Cannot open root device "hda1" or unknown-block(0,0)
Please append a correct "root=" boot option
Kernel panic - not syncing: VFS: Unable ro mount root fs on unknown-block(0,0)


And with the CF inserted, more or less the same :
CODE
Probing IDE interface ide0...
hda: SAMSUNG CF/ATA, CFA DISK drive
...
ide0: reset: success
hda1
ide-cs: hda: Vcc = 3.3, Vpp = 0.0
Kernel panic - not syncing: VFS: Unable ro mount root fs on unknown-block(3,1)


As far as I understand that, it looks like the kernel is told to boot off an IDE drive... maybe the config for a SL-C3x00 ? ... but on the akita, no IDE boot device : only JFFS flash, I think...

Hope this helps,
lardman
Yes, looks like that is the case.

Can you add this to the bugtracker please: http://bugs.openembedded.org/


Si
obergix
QUOTE(lardman @ Jan 4 2006, 07:09 PM)
Can you add this to the bugtracker please ?
*


Done : http://bugs.openembedded.org/show_bug.cgi?id=573
obergix
QUOTE(obergix @ Jan 3 2006, 10:48 PM)
Just a note to report I've just tested gpe-image (gpe-image-akita-20060102193451.rootfs.img + zImage-2.6.14+2.6.15-rc7-akita-20060102193451.bin) generated in .dev on my akita.

*


Just to tell you that the bug has been fixed, and at first sight, it looks like the 2.6 is working pretty well now for akita in .dev...

Will need additional testing, though wink.gif

at least looks like it boots, suspends and resumes allright !
obergix
QUOTE(obergix @ Jan 6 2006, 10:49 AM)
Will need additional testing, though wink.gif

*


My DLink DCF-660W wireless card doesn't seem to work out of the box... whereas it worked allright in 2.4 (3.5.4RC for instance).

However, after modprobing hostap-cs, I made it work. Dunno why it didn't recognize it automatically.

Now, it's dropbear which seems not to respond pretty well...

Actually no... it's just a bit slower than before to respond.

Cool, looks like it works !
JohnX
Alright! biggrin.gif I got 2.6.14+2.6.15-rc7 20060109 installed with a matching GPE image that I built. Like obergix I had to modprobe hostap-cs manually. I'll look into the cause of that later. Right now I'm taking the time to build every program I want out of the OE tree so that I don't contaminate this install with random versions from different feeds.

Random question: Does anyone know off the top of their head how to generate the "Packages" file for a feed. I'd like to just have a feed mounted through samba or nfs, but the "Packages" in my tmp/deploy/ipk only contains info about the packages in gpe-image.
EDIT: The answer is
CODE
bitbake package-index


Alright, off to start digging through this stuff, but so far this release feels really "snappy" (for lack of a better word) compared to the old 2.4 stuff. Great going OE/OZ/GPE devs!

EDIT: OK. I ran into trouble with the alsa drivers. They throw an oops when loaded. The system keeps going fine after that, unless I try to play sound. When I try to play a sound file with madplay or aplay that program locks, but the system stays running. Once I find the right bug tracker I'll post the details to it.

-John
lardman
bugs.openembedded.org
JohnX
Thanks. It's bug 584.

Oh, and unlike previous releases, SD continues to work after a suspend. So that at least is fixed. Is there anything specific I should be testing for, besides just giving it a general thrashing?
lardman
Have a talk with RP (rpurdie) on irc (#oe, #openzaurus) he'll point you in the right direction.

Si
JohnX
Do you have a general idea of when he (and other people) are on IRC? I've been on #oe, #openzaurus, and #gpe since I got back from classes today, 5:30PM PST but it seems like everyone is idle...

EDIT:Nevermind, I got a hold of him. Only had to stay up til 1AM... ph34r.gif
danboid
Great news about 2.6 starting to shape up on the CXX00 models! How complete is 2.6 on these devices now? Can you mount the Z microdrive as a removeable USB disk under 2.6/GPE/Opie on the C3000 and C3100? Does sound in/out work fine under ALSA? Can I run mplayer with bvdd and iwmmx support? Are the touch screen, keyboard and power management OK yet?

Thanks! Can't wait to get 2.6 running on my Z!

dan
JohnX
Ok, here's the current status from a user's point of view. The versions I'm running are a) a week or more old and cool.gif not the 3.5.4 release but the result of building an image from the "unstable" OE data.

-ALSA sound works fine in the kernel I'm running, at least for playing sound. I have no idea if "mic in" works. I had to apply a patch by hand (only concerns C1000) but I'm pretty sure it got into OE by now.

-Suspend works great. My C1000 suspends and resumes very quickly, and once it's resumed there isn't any pause before the UI becomes responsive (GPE at least, haven't tried OPIE recently). Sound continues to play through a resume. blink.gif Timeout before suspending works fine.

-I have no idea at all about using the Z as a USB device connected to a USB host. I don't even know where my cable is. rolleyes.gif

-The bvdd kernel driver is not in 2.6. I have no idea if it will be there eventually, or if it might be replaced with a real accelerated framebuffer. AFAICT the Xserver that OZ/GPE uses does not support 320x240 mode so mplayer is pretty much a lost cause for the time being.

-touch screen works fine.

-keyboard works fine and has a nice keymapping by default. Delay before auto-repeat is a little quick to start with but can be changed by "kbd_mode -d 500" by root.

-Some fixes have showed up in GPE recently that deal with making it act right at 200dpi, which makes fonts show up at humanly readable sizes by defaults. Huzzah!

-I'm just a user and not always up with what the devs are doing so people should feel free to correct me if I've said something wrong...
danboid
Thanks very much for that JohnX- sounds like much progress has been made in recent months with Linux 2.6 Zaurus support.

It would be good to hear what the status is on a fully bvdd accelerated X-server or framebuffer, if such things are being worked on.
lardman
QUOTE
It would be good to hear what the status is on a fully bvdd accelerated X-server or framebuffer


My understanding is that RP is going to work on the accelerated framebuffer (or even has done some work on it already). I don't see why people can't do likewise for the Xserver in the meantime - there are already some accelerated Xservers in OE, but not for the C7x0 or Cxxxx machines. Afaict this is simply due to a lack of time (though I'd ask on #oe & #gpe). I was planning to port the acceleration stuff from the pdaXrom Xserver to the OE version for my c7x0 - it doesn't look like a great deal of patching - but am a bit busy at the moment.

QUOTE
-The bvdd kernel driver is not in 2.6. I have no idea if it will be there eventually, or if it might be replaced with a real accelerated framebuffer. AFAICT the Xserver that OZ/GPE uses does not support 320x240 mode so mplayer is pretty much a lost cause for the time being.


I don't really know what I'm talking about here, but the w100 driver for the c7x0 machines works with mplayer in zoomed mode (using SDL iirc), so I can't see why the bvdd won't also - just needs someone to add it I'd guess.

That said, I'm not an expert on graphics drivers so I may be wrong.


Si
JohnX
QUOTE
I don't really know what I'm talking about here, but the w100 driver for the c7x0 machines works with mplayer in zoomed mode (using SDL iirc), so I can't see why the bvdd won't also - just needs someone to add it I'd guess.

That said, I'm not an expert on graphics drivers so I may be wrong.


Well, that makes two graphics non-experts then. smile.gif I played with mplayer a very small amount on 2.6.14 (or .15? don't remember) and it worked, but played in a little 320x240 box in the center of the screen with -vo x11. In this mode it seemed to be playing fast enough for everyday use. Therefore, by having either 1) the bvdd driver setup like in 2.4 where mplayer uses /dev/bvdd to write directly to the framebuffer -or- 2) an xserver that could switch video modes on the fly (ie "zoom"), we would have workable video. Unfortunately we have neither. But really just as long as I have LTris and LBreakout2, that's all the entertainment I need. smile.gif
obergix
QUOTE(JohnX @ Jan 29 2006, 04:11 AM)
-I have no idea at all about using the Z as a USB device connected to a USB host.  I don't even know where my cable is.  rolleyes.gif

*


Just wanted to complement on this, from my experience on 3.5.4.1-alpha1, which has been released a couple days ago.

This seems to be working with the ohci_hcd driver.

Hope this helps.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2016 Invision Power Services, Inc.