Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - rpurdie

Pages: [1] 2
Zaurus - pdaXrom / Pdaxrom Development
« on: July 21, 2006, 11:06:13 am »
I think that post is one of the most rational and balanced I've read here for a while. It correctly sums up the issues and position and puts things very well - nice one!

I'm pleased you got OE working and saw a glimpse of its potential. I suspect OE would value feeback to make the getting started process easier and help in improving the documentation. A new user perspective on that is often useful. Also, please be sure to submit the .bb file for inclusion if you haven't already!


Angstrom & OpenZaurus / 128mb Ram On Cxxx
« on: July 10, 2006, 06:32:02 am »
I have a C860 upgraded with 128Mb RAM total. What can i do for take it working in OZ? For Cacko and PdaxRom I have patch on 2.4.18(2.4.20) kernel and I can build custon kernel with support 128Mb RAM enabled. What's there?

PS. Sorry for my ugly english.
[div align=\"right\"][a href=\"index.php?act=findpost&pid=134521\"][{POST_SNAPBACK}][/a][/div]

Under 2.6, in arch/arm/mach-pxa/corgi.c, the memory value is hardcoded in the fixup routine. You'll need to adjust that. Passing mem=128MB on the commandline might also work but I can't remember which option overrides which.

Angstrom & OpenZaurus / Pxa270 Overlay Dev Funding
« on: June 16, 2006, 11:08:27 am »
"The thing that worries me is kernel development. It seems that within OZ it is only RP who really does anything. I BADLY want to see proper pxa270 overlay support which is as good as or surpasses 2.4's bvdd but it seems that this isn't a priority for RP. RP, if you're reading, may I may make a suggestion? If we raise some funding would you be able to concentrate on this (and pxa270 freq scaling) so we can get this major obstacle, for cxxxx/OZ users, out of the way? Or is there anybody else who thinks they might be up to this task?"

I don't read the forums but someone mentioned this was here. I do intend to look at these issues and its a question of time, not money. There are a lot of demands on my time and I'm juggling them as best I can.

There are a lot of things that go on behind the scenes that people don't realise and its been pointed out to me that perhaps I should illustrate some of them. Yesterday I probably spent about 6 hours working out why the 2.6.17-rc kernels didn't have working PCMCIA when reading CF cards on the Zaurus. This turned out to be a trivial problem with one of the Zaurus patches and is now resolved (with the patch being pulled from the upcomming OZ release as well although it didn't seem to cause a problem under 2.6.16).

The advantage of this behind the scenes work is when 2.6.17 ships in the next  few days, I can instantly provide a kernel which should live up to the standards set by 2.6.16. Having a working cutting edge mainline kernel is important as I will want to develop the overlay code against mainline so we can get it merged and reduce the maintenance load.

The night before I debugged the console rotation code to find out why logos didn't display properly on rotated framebuffers. The fix will be made in mainline and we have a copy of that fix now in use in the forthcomming OZ release along with working logos.

Recently I also spent time applying the IRQ patches that testing was requested for on linux-arm-kernel. I found problems, worked with the author to fix them and am now happy that when there are merged after 2.6.17 comes out, the Zaurus will be unaffected.

This is not a complaint and I don't mind doing this but perhaps you can begin to see how finding a large free chunk of development time for overlay support is difficult. Progress on the Zaurus kernels has always been slow and steady and overlay support will happen! When we do implement it, it will be done properly, get into mainline and hopefully continue to work in many kernel versions thereafter.

Angstrom & OpenZaurus / 2gig Sd On C700 In Oz 3.5.4 ?
« on: May 02, 2006, 03:02:33 pm »
Hi ! I have a c700 for which I have just bought a 2gig Kingston SD card...

The card did not work (getting IO errors) so I brought it back to the store, thinking it was defective..

Now I have another one, and I have the same problem..

Buffer I/O error on device mmcblk0, logical block 0..

Is it a limitation of the hardware ? Of the SD drivers? Or did I just hit a model of SD cards that was not compatible with my device?

[div align=\"right\"][a href=\"index.php?act=findpost&pid=125436\"][{POST_SNAPBACK}][/a][/div]

You're not the first person I've seen mention that although I've yet to work out what the problem is. The changes in OZ might help as it has a much more recent kernel with several MMC/SD changes.


Zaurus - pdaXrom / Psx Running On The Zaurus
« on: April 27, 2006, 07:12:37 am »
OZ forum cannot help you because they rejected my patches for overcloking and overlay as "hacks".
[div align=\"right\"][a href=\"index.php?act=findpost&pid=124776\"][{POST_SNAPBACK}][/a][/div]

You admitted they were hacks yourself ;-)

The OZ kernel has slowly but surely been making sure everything it does is done properly and in a way we can merge with mainline kernels to ensure support for all kernel versions to come. Admittedly this takes longer than quick hacks but the 2.6 OZ kernels are long term projects and this approach has paid dividends already as switching to new kernel version is easy.

Taking the Zaurus reboot patch you showed me recently, it only supported akita. I was able to create one which supported all pxa 2.6 models in a way I think I can push into mainline.

Patches are gratefully received  and will get looked at with a view to being applied. If I feel they need rewriting, they will have to wait until someone does that.  We will continue to move carefully and in a way which looks at the long term outcome rather than the short term. I've applied a better version of the zaurus reboot patches. CPU frequency scaling (using cpufreq rather than intel tools like your version), pxafb overlay support and QVGA on cxx00 models will happen as soon as I find time to sort make nice patches.

Note when I say "we", there is a total of *one* person working on the 2.6 kernel in OZ/OE support for poodle, corgi, shepherd, husky, akita, spitz, borzoi and terrier support (me). Thankfully someone is helping with Tosa (thanks Dirk!).

OE/OZ 2.6 Zaurus Kernel Maintainer/Developer

Angstrom & OpenZaurus / C3200 - Suspend Problem
« on: April 10, 2006, 10:00:52 am »
Suspend works fine on mine but reboot doesn't (system just hangs).
[div align=\"right\"][a href=\"index.php?act=findpost&pid=122487\"][{POST_SNAPBACK}][/a][/div]

Does this happen for every reboot or just occasionally?


Angstrom & OpenZaurus / C3200 - Suspend Problem
« on: April 10, 2006, 09:59:28 am »
I have installed OZ Alpha1 on the 3200, but experience a problem with suspend. After reboot suspend works just one time, after the first resume OS
refuses to get suspended again, here is dmesg:

Stopping tasks: =====================================|
hub 1-0:1.0: error -16
Could not suspend device 1-0:1.0: error -16
Some devices failed to suspend
Restarting tasks... done

And the system continues to work. Here is /proc/version if it matters:
Linux version 2.6.15- (richard@sam) (gcc version 3.4.4) PREEMPT Mon Mar 20 07:55:10 GMT 2006

It seems to me the kernel is unable to suspend HDD ? I don't have any card inserted in slots.
Any ideas ?
[div align=\"right\"][a href=\"index.php?act=findpost&pid=122420\"][{POST_SNAPBACK}][/a][/div]

Could you try removing all the usb modules before suspending/resuming? (rmmod ohci_hcd, pxa27x_udc and any usb related ones). This doesn't look like the HDD but more like some USB module issue.


Angstrom & OpenZaurus / Openzaurus Testing
« on: March 23, 2006, 03:55:17 pm »
In the installation instructions for C3100 it is stated that OZ will be installed to the hard-drive instead of to flash. Is this still the case? Isn't there any way to install to flash instead, seems like a bit of a waste not to use the built-in 128 MB flash rom...
[div align=\"right\"][a href=\"index.php?act=findpost&pid=119816\"][{POST_SNAPBACK}][/a][/div]

that would be nice, can you? Actually working on all the bugs(if any) would be the priority probably...
[div align=\"right\"][a href=\"index.php?act=findpost&pid=119841\"][{POST_SNAPBACK}][/a][/div]

Please keep in mind that most folks really do not want their rootfs on flash when they have a large HDD available as this would bring back all the problems caused with ipkg-link once you try to install apps to HDD.

Having a large rootfs is a very nice thing and simplifies things for most users.
[div align=\"right\"][a href=\"index.php?act=findpost&pid=119868\"][{POST_SNAPBACK}][/a][/div]

Also keep in mind the flash is much slower than the microdrive. Using the microdrive gives quite a performance gain. Sharp couldn't do this in 2.4 due to kernel limitations that no longer exist. I doubt OZ will release images that use the flash for these reasons although you could easily build them using OE after changing the fstab/kernel commandline.

Zaurus - pdaXrom / Porting Pdaxrom To Phycore Board
« on: January 03, 2006, 07:34:41 am »
You can also try OpenEmbedded which will give you more packages then pdaX and contain more lowlevel stuff usefull during prototype development.
[div align=\"right\"][{POST_SNAPBACK}][/a][/div]

I'd like to second the vote for using Openembedded. OE is built specifically to generate images customised to machines and has a much more diverse range of packages than pdaX [a href=\"][/url].

Angstrom & OpenZaurus / Status Of 2.6 For Akita (sl-c1000) In Oz/gpe
« on: December 15, 2005, 03:33:57 pm »
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 release from the dev branch with images/2.6 kernels for these devices. A testing release of is available at

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.

Cxx0 General discussions / C750 Won't Flash - Any Ideas?
« on: September 09, 2005, 07:06:47 am »
Right, I've been doing this for a while so I hopefully haven't made any school-boy errors.

I've flashed a fair few times this week (perhaps 10 times) testing kernels, so my CF card was working as was the transfer method (sftp -> WinXP -> card reader -> CF card), likewise the should be okay.

I press and hold the Okay button, then press the On button - I get the Japanese menu from which I choose option 4, then choose 2 to get the Cf card, then Y to confirm. Then rather than rebooting using the Sharp code, the screen just goes black. I considered the possibility that the backlight wasn't being turned on, so I've left it for 10min or so (which is enough time to flash a kernel) however it doesn't do anything.

When I get bored and then press the on button to see what's happening, there's a sort of flash across the bottom third of the screen (but not bright, it's black - it looks like the difference between the screen being powered and black and switched off).

I press the on button again and get back to the Japanese flashing menu, selecting option 1 will allow me to reboot.

I've no idea what's wrong, if anyone has any suggestions I'd be glad to hear them.

To sum up what I've done to check that I've not made a mistake:

* I've used the Cf card before, have tried another one too and also an SD card. I've also re-partitioned and formatted the CF card.

* I've replaced the

* I've checked the md5sum of the zImage.bin file, and also tried one which flashed without troubles earlier in the week (just in case the zImage.bin was the problem)

* I've checked that there are no extra file extensions (using my Z to look at the contents of the card).

* Power is attached and battery is good.

* I've left the machine unpowered and without the battery overnight (on the off chance)

I'm at my wits' end, any ideas gladly accepted,


[div align=\"right\"][{POST_SNAPBACK}][/a][/div]

I think I know what's wrong. You've flashed a kernel that was bigger than the space available in NAND and it overwrote the ajoining data area. Your NAND is now corrupt.

Thankfully, this is easy to fix. See [a href=\"][/url] and follow the instuctions with the files for your model.

C1000/3x00 General discussions / C3100 2.6 Kernel Tester Urgently Needed
« on: September 09, 2005, 05:57:47 am »
My pleasure.   As we speak, the NAND flash backup of my OZ 3.5.3 image is going back onto the little beast.   If you would be so kind, how long would it take you to create a bootable OZ 3.5.3 image + 2.6? 

[div align=\"right\"][a href=\"index.php?act=findpost&pid=95161\"][{POST_SNAPBACK}][/a][/div]

I have bootable OZ-3.5.4-pre + 2.6 here now although there is at least one serious bug. I wouldn't bother trying OZ-3.5.3 as none of the model specifics are in the OE metadata from that era.

I'm in the process of rewriting almost all the kernel patches. Once this is done, I might have time to make some kernels+images available.


C1000/3x00 General discussions / C3100 2.6 Kernel Tester Urgently Needed
« on: September 09, 2005, 05:55:13 am »

Didn't Ken L. get similar results back on Aug 30th?   

[div align=\"right\"][a href=\"index.php?act=findpost&pid=95162\"][{POST_SNAPBACK}][/a][/div]

He tested akita (C1000), you've tested the C3100 and I have a C3000. I now know detection of each model works. You'll notice you were using the same kernel as the C1000 yet they did different things  

C1000/3x00 General discussions / C3100 2.6 Kernel Tester Urgently Needed
« on: September 09, 2005, 05:12:40 am »
Well, the kernel does boot.   The console messages fly by in PORTRAIT mode though, and I don't have a way to scroll back through the messages.  What I really should have had is a serial console running ... -sigh-
[div align=\"right\"][a href=\"index.php?act=findpost&pid=95154\"][{POST_SNAPBACK}][/a][/div]

Portrait is to be expected on these devices as the framebuffer doesn't support hardware rotation like the c7x0s. Someone needs to add software rotate to fbcon. Its a low priority for me at the moment but will happen. opie/gpe already have software rotation so its not a problem.

In any case, the kernel panics with an expected "Kernel panic - not syncing: no init found.  Try passing init= option to kernel."
[div align=\"right\"][a href=\"index.php?act=findpost&pid=95154\"][{POST_SNAPBACK}][/a][/div]

Yes, that's expected as your microdrive won't have a /sbin/init. If it had, it would probably happily boot from it!

There are a lot of "Bad eraseblock xxx at 0x00000000" where xxx are numbers like 938, 985 and the memory ranges look like 0x7540000 and 0x7b20000 etc.. etc..
[div align=\"right\"][a href=\"index.php?act=findpost&pid=95154\"][{POST_SNAPBACK}][/a][/div]

I need to add a fix for that on Borzoi as it suffers the mtd virtual block problem. This has reminded me, thanks!

I did see messages like "spitz_pci_setpower: on"
[div align=\"right\"][a href=\"index.php?act=findpost&pid=95154\"][{POST_SNAPBACK}][/a][/div]

That's it trying to power the MMC/SD slot which is good.

Pressing the power button results in :

power.c: device-level power management is not supported uet
power: requesting system suspend
[div align=\"right\"][a href=\"index.php?act=findpost&pid=95154\"][{POST_SNAPBACK}][/a][/div]

That's normal for 2.6 kernels. I'll remove the first message at some point as its annoying/confusing.

In any case, here is my attempt to type what's on the screen---->

Creating 3 MTD partitions on "sharpsl-nand":
0x00000000-0x00700000 : "System Area"
Charge LED On
0x00700000-0x2700000 : "Root Filesystem"
0x02700000-0x8000000 : "Home Filesystem"
NET: Registered protocol family 2
spitz_mci_setpower: on
spitz_mci_setpower: on
IP route cache hash table entries: 1024 (order: 0, 4096 bytes)
TCP established hash table entries: 4096 (order: 3, 32768 bytes)
TCP bind hash table entries: 4096 ( order: 2, 16384 bytes)
TCP: Hash tables configured (established 4096 bind 4096)
TCP reno registered
TCP bic registered
NET: Registered protocol family 1
NET: Registered protocol family 17
Probing IDE interface ide0...
mmcblk0: mmc0:a95c SD128 123008KiB
 mmcblk0: p1
hda: HMS360604D5CF00, CFA DISK drive
ide0 at 0xc4820000-0xc4820007,0xc482000e on irq 138
hda: max request size: 128KiB
hda: 7999488 sectors (4095 MB) w/128KiB Cache, CHS=7936/16/63
hda: cache flushes supported
 hda: hda1 hda2 hda3
ide-cs: hda: Vcc = 3.3, Vpp = 0.0
 hda: hda1 hda2 hda3
EXT3-fs: mounted filesystem with ordered data mode.
VFS: mounted root (ext3 filesystem) readonly.
Freeing init memory: 76K
kjournald starting.  Commit interval 5 seconds
Warning: unable to open an initial console
Kernel panic - not syncing: No init found.  Try passing init = option to kernel.

And with that ... my typing hand wants to fall off.  T-t-t-t-t-t-hhats all folks.

[div align=\"right\"][a href=\"index.php?act=findpost&pid=95154\"][{POST_SNAPBACK}][/a][/div]

And this tells me everything I need to know. It can only have printed that if it correctly identified the machine as a Borzoi. Thanks for testing and you can look forward to 2.6 kernels on this device. We might even have 2.6 kernels+images for akita, spitz and borzoi for OZ 3.5.4 (perhaps unofficially).

I can't tell you how much I needed that test report and the results are near perfect! Thanks

C1000/3x00 General discussions / C3100 2.6 Kernel Tester Urgently Needed
« on: September 07, 2005, 04:14:17 am »
i was under the impersion that usb client was a pxa270 thing and not c3k specfic?, i know they have it working on a x50 (usbnet)

some other things that have shown up in the 2.6.13 kernel that my be of use, alsa ac97 support for pxa2xx's and pxa27x power support (throgh sysfs) (note i see the sound driver is i2c, i belive they added suport for that in 2.6.12 for the pxa's)

hope it helps, as for testers where do i sign up, i woudnt mind sound but that wont hold me back from doing testing
[div align=\"right\"][a href=\"index.php?act=findpost&pid=94764\"][{POST_SNAPBACK}][/a][/div]

Yes, a usb client for the pxa27x exists in and we can probably use it. It has several bugs which is why I've put off importing it into the 2.6 kernel - I don't like importing known faulty code as I'll just get complaints.

As for sound, we don't use ac97 but use I2S instead - no mainline support for that. There is a known good I2C driver for the PXA around but it is also not in mainline. We have the kernel patched to add that as the c7x0 already uses it. For cxx00, someone just needs to import the 2.4 Sharp code, clean it up and we'd have an OSS driver. I'd prefer we went one step further and converted both it and the c7x0 driver to alsa though.

Both issues are easily solved, they just need someone to spend time on them. Any help is welcome - kernels/images can be generated from OE.

At the moment, I'm busy pushing some of the cxx00 changes into mainline Linux so we don't have a massive patchset to maintain. I'm also trying to fix the 2.6 images OE is generating as there are a few issues with them. Once this is done, I'll get back to enhacning the kernel.

Pages: [1] 2