OESF Portables Forum
Everything Else => Zaurus Distro Support and Discussion => Distros, Development, and Model Specific Forums => Archived Forums => Debian => Topic started by: tux on October 23, 2007, 06:45:54 am
-
The recent comments about the use of the angstrom kernels and the Titchylinux rootfs raise a few questions. I mean the fact that the kernel is EABI and the system is OABI. One suggestion has been to use the armel rootfs that is mentioned in the sound thread.
However, another direction has shown up in the C1000 and Debian thread.
Here is the link that scottlfa gave: Pokylinux 2.6.21 Kernels (http://pokylinux.org/releases/blinky-3.0/). By the way his post gives very clear steps to setting up debian on the C1000, presumably adaptable to the 860 and 700 series?
I've had the Pokylinux site bookmarked for a while but not had the time to look more closely. (I think there is a Pokylinux thread somewhere. ) The site is easy to navigate and all the files are easily accessible. The documentation seems fairly clear as well. It should be easy to download and set up their version of the development environment. Those who would like to roll their own kernels have clear instructions on obtaining the files ansd setting up the environment. (A bit of a contrast to another site we know! )
It seems quite a promising site in lots of ways. For example: if I flash the 2.621 kernel for my 3200 and follow the steps for getting the modules in the right place, I should have a working Titchy without the angstrom kernel. Presumably the kernel and rootfs architecture should now match. Would this solve some of the problems with multimedia? (I must try it!) There are also rootfs files on that site. What would happen with those? What about the PokyLinux distribution? That is for some other brave soul to try!
Hope to see some suggestions/comments when I get back from the real world!
But I think it is well worth a good look round!
-
Poky kernels are also EABI and built with OpenEmbedded, so you might as well face the fact that you all need OE to get your zaurusses to do something usefull with kernel 2.6
-
yeah i tried poky already some months ago, and it looked very similar to angstrom
the main problem are that packages were incompatible between angstrom and poky (as said by poky developpers - "not supported"), although i tried some and they were working... (for example i had bluetooth working).
and there was little packages available too when i tried it.
but with poky installed, your zaurus looks quite like an actual pda, with dates, contacts and so on. so it looks promising for people looking for a PDA
-
Poky kernels are also EABI and built with OpenEmbedded, so you might as well face the fact that you all need OE to get your zaurusses to do something usefull with kernel 2.6
What a disappointment! I thought there might be a shortcut there.
Please correct me if I am wrong here but I thought that it didn't matter what build system was used to produce a kernel. What matters is what choices are set for the build process to follow. Assuming that is true, then the bitbake system would produce an OABI kernel and packages if the options were set for that as opposed to the setup you use, would it not.
For that matter any build system could surely have the options/switches set to produce either EABI code or OABI code as required?
Anyway, to get back to the point of my first post, it looks like I am going to have to wait for a fully built mainstream Debian armel system or find/build an OABI version of the kernel. What build system to use and which kernel source to start with is a matter for further consideration.
Why do I want that? TitchyLinux gives me a system which uses the apt-get install root just like my desktops. All, well almost all, the packages that run on my desktop machines are available through apt-get. Those packages that aren't available can be compiled, by me, from the tars from the developers site, just like my desktops. I can use the same development environment that my desktop uses.
What is missing for me? At the moment the multimedia is problematic. I can live without that, especially since there appear to be ways to sort it out. Why is there a multimedia problem: apparently because the kernels we have available are EABI and the normal Debian arm is OABI. Well that will be cured eventually.
As I said above there are two routes: wait for the Debian armel branch to be fully developed or get/build a kernel for the presently fully developed Debian arm system. Whichever comes first doesn't matter.
I suspect that the OABI kernel will be available first: I'll happily use it and equally happily transfer to the armel version when it is completed.
-
Poky kernels are also EABI and built with OpenEmbedded, so you might as well face the fact that you all need OE to get your zaurusses to do something usefull with kernel 2.6
What a disappointment! I thought there might be a shortcut there.
Please correct me if I am wrong here but I thought that it didn't matter what build system was used to produce a kernel. What matters is what choices are set for the build process to follow. Assuming that is true, then the bitbake system would produce an OABI kernel and packages if the options were set for that as opposed to the setup you use, would it not.
For that matter any build system could surely have the options/switches set to produce either EABI code or OABI code as required?
Anyway, to get back to the point of my first post, it looks like I am going to have to wait for a fully built mainstream Debian armel system or find/build an OABI version of the kernel. What build system to use and which kernel source to start with is a matter for further consideration.
Why do I want that? TitchyLinux gives me a system which uses the apt-get install root just like my desktops. All, well almost all, the packages that run on my desktop machines are available through apt-get. Those packages that aren't available can be compiled, by me, from the tars from the developers site, just like my desktops. I can use the same development environment that my desktop uses.
What is missing for me? At the moment the multimedia is problematic. I can live without that, especially since there appear to be ways to sort it out. Why is there a multimedia problem: apparently because the kernels we have available are EABI and the normal Debian arm is OABI. Well that will be cured eventually.
As I said above there are two routes: wait for the Debian armel branch to be fully developed or get/build a kernel for the presently fully developed Debian arm system. Whichever comes first doesn't matter.
I suspect that the OABI kernel will be available first: I'll happily use it and equally happily transfer to the armel version when it is completed.
the problem you have is not whether it is a EABI or OABI kernel. the kernel 2.6 just does not 100% support the zaurus hardware yet. some features are just not working yet.
as far as the packages are concerned, both armel debian and angstrom are build using OE, so you might as well bite the bullet and do what Koen said and use OE if you want to use debian armel. OE is just a build system which is supposed to make configuration and patching easier for you since it automates most of that. you can use your own crosscompiler or use OE to build one for you. it is just a matter of what you prefer. the end result will be a compiled kernel either way as long as you got all the required patches. OE can be configured to build other things too. It is just badly/inadequately documented and requires a steeper learning curve to get started with than rolling your own. however, it is conceivable that you will save some time down the road when doing further updating and patching... The OE stuff can help in streamlining development and building, but it is not perfect and has areas that aren't ideal. It is a tool like any others that have their positives and negatives...
-
you all need OE to get your zaurusses to do something usefull with kernel 2.6
So you are trying to convince me that my Zaurus, which I use for about 80% of my computing, and which is running a 2.6 kernel, is useless?
EDIT: It's not running Angstrom/OZ, either, and last time I tried to run Angstrom on it, it wouldn't boot. I guess having a paperweight must be much more useful than having a device that I can code, browse the web, check e-mail, and compile on in my pocket...
-
you all need OE to get your zaurusses to do something usefull with kernel 2.6
So you are trying to convince me that my Zaurus, which I use for about 80% of my computing, and which is running a 2.6 kernel, is useless?
EDIT: It's not running Angstrom/OZ, either, and last time I tried to run Angstrom on it, it wouldn't boot. I guess having a paperweight must be much more useful than having a device that I can code, browse the web, check e-mail, and compile on in my pocket...
DELETED (In fact I hate "mine is bigger than yours" discussions. It's a road to nowhere)
-
I think it's worthwhile to check if the problem is related to OABI vs EABI or not. I removed the dust of my Poky build system, and it's building a cross-toolchain as we speak, and hopefully a OABI kernel after that. We'll see...
-
I think it's worthwhile to check if the problem is related to OABI vs EABI or not. I removed the dust of my Poky build system, and it's building a cross-toolchain as we speak, and hopefully a OABI kernel after that. We'll see...
Thanks cortez, hope that it all goes smoothly. Please post your impressions of the build system: I presume it is the bitbake thing? I'd like to repeat what you are doing. I'll set myself up to try this weekend. More sleep deprivation!
I really need to find time to test these cross-compiler systems out, if I'm going to get into this. I expect that I'm going to have to try and build some of the multimedia stuff myself. Meanie's comment about hardware support in 2.6 kernels seems on target! Meanie's comments usually are.
It will be interesting to see if an OABI kernel helps with the problems though.
So it looks like my guess about OABI kernels coming through before a fully developed armel system was correct.
Cheers
-
The Poky build system is OE/bitbake indeed. The Poky README (http://svn.o-hand.com/repos/poky/trunk/README) describes how to get started. The complete build system can be checked out by simply running svn co http://svn.o-hand.com/repos/poky/trunk (http://svn.o-hand.com/repos/poky/trunk) poky
Default setup and building is simple, adjusting the default configuration is somewhat more complicated than I was used to in the pdaXrom-builder.
-
The Poky build system is OE/bitbake indeed. The Poky README (http://svn.o-hand.com/repos/poky/trunk/README) describes how to get started. The complete build system can be checked out by simply running svn co http://svn.o-hand.com/repos/poky/trunk (http://svn.o-hand.com/repos/poky/trunk) poky
Default setup and building is simple, adjusting the default configuration is somewhat more complicated than I was used to in the pdaXrom-builder.
Thank you, I'll get on with that now.
Quote:'adjusting the default configuration is somewhat more complicated than I was used to in the pdaXrom-builder'. Why is this not a surprise to me?
-
The Poky build system is OE/bitbake indeed. The Poky README (http://svn.o-hand.com/repos/poky/trunk/README) describes how to get started. The complete build system can be checked out by simply running svn co http://svn.o-hand.com/repos/poky/trunk (http://svn.o-hand.com/repos/poky/trunk) poky
Default setup and building is simple, adjusting the default configuration is somewhat more complicated than I was used to in the pdaXrom-builder.
Thank you, I'll get on with that now.
Quote:'adjusting the default configuration is somewhat more complicated than I was used to in the pdaXrom-builder'. Why is this not a surprise to me?
I've set up the poky/bitbake buildsystem. Now for more self-education. While doing this I had a little look around. OpenHand have done a lot of work in this sort of area, e.g. Nokia 770. You can google for them and find lots of interesting stuff. I think this link might be of great interest:Open hand Projects (http://projects.o-hand.com/).
There are some highly relevant files to this thread and some useful pointers.
Lots of things are turning up.
-
See attached the 2.6.21 OABI Poky kernel. I'm currently not able to test, so if maybe someone else could test this it would be great.
It's not actually a .tar.gz, so just rename the file to zImage-2.6.21-spitz-20071024075428.bin
-
See attached the 2.6.21 OABI Poky kernel. I'm currently not able to test, so if maybe someone else could test this it would be great.
It's not actually a .tar.gz, so just rename the file to zImage-2.6.21-spitz-20071024075428.bin
I've downloaded it. I'll need to get the matching modules ( using 2.6.20 at the moment) but then I'll test things out. I'll try and do it by midnight. Got to sort the laundry out!! If only real life didn't get in the way!
-
See attached the matching modules. The forum only allows to upload tar.gz files, so just remove the tar.gz from the filename.
-
See attached the matching modules. The forum only allows to upload tar.gz files, so just remove the tar.gz from the filename.
If I had any medals to hand out? What a star!
-
the problem you have is not whether it is a EABI or OABI kernel. the kernel 2.6 just does not 100% support the zaurus hardware yet.
Last I checked the kernel in OE it only lacked migrating the fb to SRAM for pxa270 zaurusses, everything else works on c7x0 and cxxxx.
-
See attached the matching modules. The forum only allows to upload tar.gz files, so just remove the tar.gz from the filename.
If I had any medals to hand out? What a star!
Well cortez, I have mixed news! I took your files and renamed them as required. I didn't know which updater.sh to use so I put the one from the poky site on. I untarred the modules to the right spot. I rebooted and pressed OK at the appropriate moment and the flashing menu came up. It failed with a message about wrong machine. I do have a 3200. I suppose I might need another updater.sh? Or maybe the zImage should have been renamed to zImage.bin. I'll try that idea first.
I assume of course that the machine will reboot to my original setup when I put the battery back in! It did!
Cheers - I'll persevere!
-
You need updater.sh from here (http://pokylinux.org/releases/blinky-3.0/updater.sh.spitz) and rename the zImage file to zImage.bin
Then flash again, keep your fingers crossed and hope this will fix some of our problems
-
You need updater.sh from here (http://pokylinux.org/releases/blinky-3.0/updater.sh.spitz) and rename the zImage file to zImage.bin
Then flash again, keep your fingers crossed and hope this will fix some of our problems
flashed and booted :
- no modules.dep (and some other files in /lib/modules/2.6.21/)
- no X after fixing the previous error through depmod : illegal instruction
Wifi is working ...
Chero.
-
You need updater.sh from here (http://pokylinux.org/releases/blinky-3.0/updater.sh.spitz) and rename the zImage file to zImage.bin
Then flash again, keep your fingers crossed and hope this will fix some of our problems
Thanks again cortez, I'd just tried flashing again with an updater.sh from that japanese site whose name has BOX in it. The kernel flashed. I got the Opened hand logo on booting and I'm looking at a text login. I got many messages about not being able to find the modules. I copied the modules onto the hard disk using tar zxvf with the tgz file being on the SD card. I changed into the microdrive / directory to do it and the files showed uo when I listed the lib/modules directory. They are there.
That file you've pointed me to looks like it is the one I downloaded from the poky site and that I tried to use. I'll try again with the one I've just downloaded from the link. But first I'll try a text login. It works and I can see the modules. I have a 2.6.20 directory and a 2.6.21 directory. The console doesn't have the keytable set to the correct one.
I got some nice error messages about the soundhack that comes with Titchy though. I feel rather comforted by that! I have much to play with!!
Later this evening... I tried depmod and the keys are working. I took the wireless card out and reinserted it. I appear to have a connection. YES!!!!
So what have I missed about the modules?
Really must post this message mustn't I?
-
Here is the output from dmesg after I'd done depmod, reinserted the wifi card and ssh'd into root.
debian:/# dmesg
Linux version 2.6.21 (egoe@egoe-laptop) (gcc version 3.4.4) #1 PREEMPT Wed Oct 24 13:52:39 CEST 2007
CPU: XScale-PXA270 [69054117] revision 7 (ARMv5TE), cr=0000397f
Machine: SHARP Borzoi
Ignoring unrecognised tag 0x00000000
Memory policy: ECC disabled, Data cache writeback
On node 0 totalpages: 16384
DMA zone: 128 pages used for memmap
DMA zone: 0 pages reserved
DMA zone: 16256 pages, LIFO batch:3
Normal zone: 0 pages used for memmap
Run Mode clock: 208.00MHz (*16)
Turbo Mode clock: 416.00MHz (*2.0, active)
Memory clock: 208.00MHz (/2)
System bus clock: 208.00MHz
CPU0: D VIVT undefined 5 cache
CPU0: I cache: 32768 bytes, associativity 32, 32 byte lines, 32 sets
CPU0: D cache: 32768 bytes, associativity 32, 32 byte lines, 32 sets
Built 1 zonelists. Total pages: 16256
Kernel command line: console=ttyS0,115200n8 console=tty1 noinitrd root=/dev/hda1 rootfstype=ext3 rootdelay=1 rw fbcon=rotate:1 dyntick=enable quiet
PID hash table entries: 256 (order: 8, 1024 bytes)
Console: colour dummy device 80x30
Dentry cache hash table entries: 8192 (order: 3, 32768 bytes)
Inode-cache hash table entries: 4096 (order: 2, 16384 bytes)
Memory: 64MB = 64MB total
Memory: 62244KB available (2276K code, 237K data, 108K init)
Calibrating delay loop... 415.33 BogoMIPS (lpj=2076672)
Mount-cache hash table entries: 512
CPU: Testing write buffer coherency: ok
NET: Registered protocol family 16
Sharp Scoop Device found at 0x10800000 -> 0xc4800000
Sharp Scoop Device found at 0x08800040 -> 0xc4802040
Time: pxa_timer clocksource has been installed.
NET: Registered protocol family 2
IP route cache hash table entries: 1024 (order: 0, 4096 bytes)
TCP established hash table entries: 2048 (order: 2, 16384 bytes)
TCP bind hash table entries: 2048 (order: 1, 8192 bytes)
TCP: Hash tables configured (established 2048 bind 2048)
TCP reno registered
NetWinder Floating Point Emulator V0.97 (double precision)
io scheduler noop registered
io scheduler anticipatory registered (default)
Corgi Backlight Driver Initialized.
Console: switching to colour frame buffer device 80x30
Load PXA Overlay driver successfully!
pxa2xx-uart.0: ttyS0 at MMIO 0x40100000 (irq = 22) is a FFUART
pccard: PCMCIA card inserted into slot 0
pcmcia: registering new device pcmcia0.0
pxa2xx-uart.1: ttyS1 at MMIO 0x40200000 (irq = 21) is a BTUART
pxa2xx-uart.2: ttyS2 at MMIO 0x40700000 (irq = 20) is a STUART
loop: loaded (max 8 devices)
Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 50MHz system bus speed for PIO modes; override with idebus=xx
Sharp SL series flash device: 800000 at 0
Using static partision definition
Creating 1 MTD partitions on "sharpsl-flash":
0x00140000-0x007f0000 : "Boot PROM Filesystem"
NAND device: Manufacturer ID: 0xec, Chip ID: 0xf1 (Samsung NAND 128MiB 3,3V 8-bit)
Scanning device for bad blocks
Bad eraseblock 367 at 0x02de0000
Creating 3 MTD partitions on "sharpsl-nand":
0x00000000-0x00700000 : "System Area"
0x00700000-0x02700000 : "Root Filesystem"
0x02700000-0x08000000 : "Home Filesystem"
input: Spitz Keyboard as /class/input/input0
power.c: Adding power management to input layer
input: Corgi Touchscreen as /class/input/input1
sa1100-rtc sa1100-rtc: rtc core: registered sa1100-rtc as rtc0
I2C: i2c-0: PXA I2C adapter
I2C: i2c-1: PXA I2C adapter
Registered led device: spitz:amber
Registered led device: spitz:green
TCP cubic registered
NET: Registered protocol family 1
NET: Registered protocol family 17
XScale iWMMXt coprocessor detected.
Probing IDE interface ide0...
pccard: PCMCIA card inserted into slot 1
pcmcia: registering new device pcmcia1.0
mmcblk0: mmc0:dae3 S128B 125440KiB
mmcblk0: p1
hda: SanDisk SDCFX3-16384, CFA DISK drive
ide0 at 0xc4820000-0xc4820007,0xc482000e on irq 138
hda: max request size: 128KiB
hda: 32014080 sectors (16391 MB) w/1KiB Cache, CHS=31760/16/63
hda: hda1 hda2 < hda5 >
ide-cs: hda: Vpp = 0.0
sa1100-rtc sa1100-rtc: setting the system clock to 2007-10-24 19:43:13 (1193254993)
Waiting 1sec before mounting root device...
kjournald starting. Commit interval 5 seconds
EXT3 FS on hda1, internal journal
EXT3-fs: mounted filesystem with ordered data mode.
VFS: Mounted root (ext3 filesystem).
Freeing init memory: 108K
Adding 176672k swap on /dev/hda5. Priority:-1 extents:1 across:176672k
EXT3 FS on hda1, internal journal
pccard: card ejected from slot 1
pccard: PCMCIA card inserted into slot 1
pcmcia: registering new device pcmcia1.0
orinoco 0.15 (David Gibson <hermes@gibson.dropbear.id.au>, Pavel Roskin <proski@gnu.org>, et al)
orinoco_cs 0.15 (David Gibson <hermes@gibson.dropbear.id.au>, Pavel Roskin <proski@gnu.org>, et al)
pcmcia: request for exclusive IRQ could not be fulfilled.
pcmcia: the driver needs updating to supported shared IRQ lines.
eth0: Hardware identity 801b:0000:0001:0000
eth0: Station identity 001f:0000:0001:0008
eth0: Firmware determined as Intersil 1.8.0
eth0: Ad-hoc demo mode supported
eth0: IEEE standard IBSS ad-hoc mode supported
eth0: WEP supported, 104-bit key
eth0: MAC address 00:02:6F:34:4D:6B
eth0: Station name "Prism I"
eth0: ready
eth0: orinoco_cs at 1.0, irq 137, io 0xc4840000-0xc484003f
ieee80211_crypt: registered algorithm 'NULL'
hostap_cs: 0.4.4-kernel (Jouni Malinen <jkmaline@cc.hut.fi>)
eth0: New link status: Connected (0001)
NET: Registered protocol family 10
eth0: no IPv6 routers present
debian:/#
The 'bolded' bits are some messages that I remember as different to the 'angstrom' kernel or that new to me. I don't know if it has any significane that the init memory is reported as 108, I think the previous version said 84.
No sign of life with the X-windows system as yet.
I'll try apt-get update and see what ensues. I'll also try un-installing the soundhack. Then a reboot to see what happens.
This is interesting. I don't know whether to say it is promising though!
-
Wow. Now we have five favors to choose from!
1) Debian Titchy: Angstrom EABI kernel + OABI rootfs
2) Debian Kurobox: Angstrom EABI kernel + OABI rootfs
3) Debian egoe: Cortez's OABI kernel + OABI rootfs
4) Debian Angstrom: Angstrom kernel + EABI rootfs
5) Debian yonggun: 2or0's patched Angstrom kernel + EABI rootfs
-
Definitely promising! This boot said it found the modules and loaded them. lsmod lists the usual suspects. How do I get it to tell me which modules 2.6.20 or 2.6.21? Both hostap and orinoco load. I thought I'd blacklisted hostap? I've still got a text login, though the boot messages say that xdm is loaded.The keyboard is working. I think I'll uninstall zaurus-hardware-support and reboot. Then re-install it. Might try the same with titchy package? Removing Zaurus-hardware-support triggers the removal of all the gui stuff.
I think most of the answers are in this forum, I'll go look later! No idea about multimedia yet. Let's see if the system can get back to where it was before with this kernel and modules, compiled fresh by cortez. Must try and duplicate that job in order to start to understand bitbake the poky/openhanded way.
This is definitely on the way to work and with clearer access to the build system and source tree.
-
I'm glad I installed the ssh server before all this! it doesn't half make it easy to work.
I am struggling to get the gui to come alive. I'm wondering if the error messages concerning soundhack are somehow to do with this? I do know that the link to kdrive is still in place. I've just installed gdm, to see if this might help. I got messages about the soundhack package at the end of gdm install. I'm also going to have a look at the scripts. I have a superstitious feeling that the soundhack part of Zaurus-hardware -support is the problem. I've been wrong before. But it seems impossible to have the other hardware support bits without the soundhack thing. I suppose I can unpick the deb file and repack it without the soundhack part, if I'm not just being paranoid.
Any hints or tips will be gratefully received.
I'm un-installing gdm. I'm going to check if the updater.sh that cortez showed me works with this kernel and i think i'll go back to the Titchy version for a while.
-
Hi Cortez, the updater you pointed me to was the one I'd downloaded from poky and it didn't work. The one from the japanese site did. As you've read, a fair amount of success was had. But I'm going back to Titchy to remove some of the things I tried to get x working. When I have it back to where it was I'm going to have another go. I have a feeling that the packages from neil, the hardware support ones, are clashing with the stuff you have given me.
I'm going to have a thorough look at the scripts from the working x system with Titchy before I get back to the poky stuff. I have a feeling that I'd need to set up the X system more by hand as it were. I believe that the soundhack part of titchy is the fly in the ointment. Even if it doesn't cause the x problems I'm sure that it will get in the way of the sound stuff.
I did try gdm, it tried to start but offered me the chance to look at the X server messages to diagnose the problems. There were no visible messages and the system hung! xdm just hung!
Perhaps a cleverer fellow than me will have this sorted by tomorrow!!
Cheers
-
I kept gdm. It took two minutes to reflash the 2.6.20 kernel and as soon as it rebooted X came up with the gdm. I wish I'd used it before. Much more civilised than xdm! So, if nothing else, this excursion into an OABI kernel has led me to a nicer window manager!
Thats all for tonight! I should be sensible and go to bed. The poky/OABI/cortez kernel thing is going to take a bit more work. maybe the hardware is going to have to be configured without the packages from neil's site. I need to search these threads for tips.
There are some very promising things happening though.
I'll check the Wiki and see if those anti-spam bits are still working!
-
So I'm thinking how to proceed. I think I want to go with the OABI kernel(2.6.21) and modules. thanks to cortez + poky, and an OABI rootfs, thanks to Titchy and the Japanese 'mirror' at the bigbridge site. Why OABI? Because, at the moment, the mainstream debian for arm is OABI.
But if I've used Titchy and the support packages that Neil has produced the soundhack part throws up errors and this seems to cut the gui off at the knees. I am not certain about that, this is groping for an explanation on my part.
It is easy to use the kernel provided by cortez from the poky build system if you use the updater.sh from the bigbridge site. It is fairly straightforward to copy the 2.6.21 modules to the right place. By the way, does it matter that the 2.6.20 modules are also in lib?
It's also very simple to back out by flashing the 2.6.20 kernel from the bigbridge site.
One idea: reflash the cortez 2.6.21 kernel. Dig around for the soundhack bits and erase them somehow. Then see what happens. Probably still have to delve into the X scripts and settings. By the way X still points to the kdrive thingy!
This next idea looks more likely to work, to my naive eye.
Second idea: reinstall the TitchyLinux from scratch and not use the zaurus-hardware-support package. Maybe I could use the cortez kernel at this stage? I could make sure that I have a copy of the working keyboard table and put it in manually. In the sources.list I could just put the mainstream stuff, leave out or comment out the repository from Titchy. I might even do a dist-upgrade to lenny before carrying on. I'd make sure I had the cortez/poky kernel and modules in place, by whatever means before going further.
How to get the gui? Install xorg? I know it takes more memory and would probably need the kdrive substituted in for the final system. Where to lay my sticky hands on the kdrive for the Z? How to put it in and set it up. Obviously look at the present set up and copy files.
I'm also hoping that the more experienced tinkerers will jump in and say " Do it this way you clot!"
Obviously look at the present set up and copy files. :- this should be quite easy in my particular case. The microdrive I swapped out had a working system on it. I can shove the Hitachi drive into my reader and browse, copy, print and inspect all the bits and pieces.
Any and all advice welcome! The more the merrier.
-
I think I've found out how to turn off Soundhack. There is a directory /etc/rcS.d : the README in that directory says that renaming any of the scripts to have K at the front disables it. These scripts run at boot. There is one which links to soundhack in order to run it at boot time. I have put K at the start of its name. rebooting into the gui has gone through without a hitch.
By the way the soundhack thing is a script that loads various alsa modules that the Titchy set up puts in /opt/alsa14. So I should see some differences in the lsmod report.
I'm going to try and flash the OABI/cortez/poky kernel again and see what happens depending on the lsmod message.
Wish me luck!
lsmod does not show any sound modules loaded! So I think I should have a go. After all 'Who Dares, Wins!'
-
Well cortez's present is installed. as it booted it told me that the gmpc was being loaded and there were plenty of error messages. This pleased me. It shows that I did manage to turn Soundhack off.
The gdm attempted to load but bottled out. but i got to the command line and can do that reliably. The keymap is correct. I'm actually using ssh and a terminal on my desktop. I can see better there!
I'm taking a telly break. The gui/ X-system will probably take more time. This is going to work without too much hassle.
-
The gui/ X-system will probably take more time. This is going to work without too much hassle.
Well not too much hassle if I had a clue! The /opt/kdrive directory and contents are there. The scripts have not been touched. The only change to the startup from Titchy original is that I've managed to disable soundhack. It happily does the gui with the angstrom kernel and the OABI rootfs. It clangs to a halt with the Oabi kernel and OABI rootfs. Surely I don't need a different x driver?
Should I remove all the titchy packages and start the x adventure from the ordinary repository, still leaving /opt/kdrive in place? What will happen?
Well the worst that can happen is that I'll have to restore the titchy repository and go through that cycle again! So unless some kindly guru tells me what to do by the time I come back from my fitness campaign, I'll go for it!
It should be simpler than this! i'm missing something obvious!!
-
The gui/ X-system will probably take more time. This is going to work without too much hassle.
Well not too much hassle if I had a clue! The /opt/kdrive directory and contents are there. The scripts have not been touched. The only change to the startup from Titchy original is that I've managed to disable soundhack. It happily does the gui with the angstrom kernel and the OABI rootfs. It clangs to a halt with the Oabi kernel and OABI rootfs. Surely I don't need a different x driver?
Should I remove all the titchy packages and start the x adventure from the ordinary repository, still leaving /opt/kdrive in place? What will happen?
Well the worst that can happen is that I'll have to restore the titchy repository and go through that cycle again! So unless some kindly guru tells me what to do by the time I come back from my fitness campaign, I'll go for it!
It should be simpler than this! i'm missing something obvious!!
My guess is that the /opt/kdrive stuff has been borrowed from Angstrom and needs the eabi kernel. When trying to launch the Xfbdev in there I get "illegal instruction" what doesn't look very promising to me.
I installed Xephyr, but don't know how to use it. I had to install xfonts, now it at least tries to launch (blank screen) but then drops out. Maybe we could have a look how they were running xorg on pdaX, maybe it's just some settings ... (but it won't be this evening)
Chero.
-
My guess is that the /opt/kdrive stuff has been borrowed from Angstrom and needs the eabi kernel. When trying to launch the Xfbdev in there I get "illegal instruction" what doesn't look very promising to me.
I installed Xephyr, but don't know how to use it. I had to install xfonts, now it at least tries to launch (blank screen) but then drops out. Maybe we could have a look how they were running xorg on pdaX, maybe it's just some settings ... (but it won't be this evening)
Chero.
that thought occurred to me too. But a dev from angstrom said it was just the sound stuff that couldn't cope with either OABI or EABI. I'm sure they said that.
I'd be tempted to check out the other route, EABI + EABI, but does it involve uboot?
As regards the attempt to launch: gdm starts and then says that the xserver bottled out. It didn't actually say 'bottled out' you understand.
I'll reflash to EABI kernel (takes 5 minutes to get back to a working system) and look again from BigZ's keyboard. I did aptitude install some sound stuff that was suggested in another thread.
-
2or0's EABI works very well. Sound plays clear and loud. Speed is very good too. But you need UBOOT. He said you may do it without wiping your internal drive clean. But I am not sure (because I set up everything from the ground up).
-
2or0's EABI works very well. Sound plays clear and loud. Speed is very good too. But you need UBOOT. He said you may do it without wiping your internal drive clean. But I am not sure (because I set up everything from the ground up).
You don't need uboot for EABI, angstrom uses EABI and does not use uboot on the zaurus.
-
2or0's EABI works very well. Sound plays clear and loud. Speed is very good too. But you need UBOOT. He said you may do it without wiping your internal drive clean. But I am not sure (because I set up everything from the ground up).
You don't need uboot for EABI, angstrom uses EABI and does not use uboot on the zaurus.
Thats what I thought, We are already using the right kernel. You just need to find a way to get the debian-eabi-rootfs on the microdrive (booting into emergency with D+M and make sure you have gnu-tar on the SD or CF for untarring it ?)
-
2or0's EABI works very well. Sound plays clear and loud. Speed is very good too. But you need UBOOT. He said you may do it without wiping your internal drive clean. But I am not sure (because I set up everything from the ground up).
You don't need uboot for EABI, angstrom uses EABI and does not use uboot on the zaurus.
Thats what I thought, We are already using the right kernel. You just need to find a way to get the debian-eabi-rootfs on the microdrive (booting into emergency with D+M and make sure you have gnu-tar on the SD or CF for untarring it ?)
Bangs head on the nearest wall!
This thread Installing Debian Base, Without Needing A Network Connection! (https://www.oesf.org/forum/index.php?showtopic=24827) seems to give the answer. I should have thought of it I started it with a copied post from ZDevil.
So the steps would be summarised as :
1. Get into 'emergency' and produce a formatted, clean drive.
2. Put Eabi kernel + matching updater + EABI rootfs + gnu-tar on SD
3. Flash and cross your fingers.
For 2 the armel/EABI rootfs that everybody appears to be using should do. The gnu-tar and updater.sh from bigbox mentioned in the thread I've linked to should work.
But you could presumably do the untarring manually from the emergency system. Or even do the partitioning/formatting from a working system and untar the armel fs from a mounted sd card. Am I being a bit delusional here?
Looks like an attempt at this is worthwhile. I need to look at the armel rootfs to check modules version.
I suppose I ought to try the 1,2,3 method as listed above if the bits I have look compatible. There's another weekend gone.
Thanks koen, Chero for pointing out the obvious. (Obvious to you, oblivious to me!)
I'd really like to avoid using uboot. Any system that has so many posts about how to remove it or escape from it has to be a concern!
-
But I still have some tinkering to do to see if I can get OABI + OABI working!
If anyone has an idea what I'm missing in getting the OABI + OABI system to use the stuff in kdrive please let me hear about it. If any drive-by system-builders can re-assure me that the stuff from angstrom, which is the kdrive contents, will work with OABI kernels I'd appreciate it. Obviously I'd be pleased if you stopped me going down a blind alley!
-
But I still have some tinkering to do to see if I can get OABI + OABI working!
If anyone has an idea what I'm missing in getting the OABI + OABI system to use the stuff in kdrive please let me hear about it. If any drive-by system-builders can re-assure me that the stuff from angstrom, which is the kdrive contents, will work with OABI kernels I'd appreciate it. Obviously I'd be pleased if you stopped me going down a blind alley!
I'd go for the kdrive of pdaXrom for a first try. No garanties cause pdaX uses some old libc, but you can always try ...
-
Do you remember the film "Butch Cassidy and the Sundance Kid"? The bit were they jump off the cliff into the river: successful outcome. I'm just about to try something I think of as the equivalent.
I hope it doesn't turn out like the end of the picture where they emerge from their cover to meet the entire Bolivian army!
You'll hear about it whatever the outcome. it is such a simple-minded and obvious thing to do. may it work!
-
I put the original TitchyLinux method files onto the SD card. I replaced the zImage.bin, it is angstrom 2.6.20, with the OABI kernel 2.6.21 that cortez kindly compiled and updated.
I'll emphasise: the only change is to put the cortez/poky/OABI kernel on the SD card. The rest of the files are as from the TitchyLinux site.
I did the OK menu trick and took the usual choices. It has updated the kernel, formatted the microdrive and is now extracting the Debian installer.
Those of you that have done this will recognise the steps. So it is looking like the procedure is working. Assuming that I end up with a working command line system I will look closely at the file system, not use the instructions from neil's site about using zaurus-hardware-support, I've got the keyboard file ready to install separately.
Stage 4 has arrived, the kernel booted into the installer. The kernel is the cortez one: the Openedhand logo showed up. Very smart it is too!
This should take a while.
ZDevil you may need to update that howto already! I'll read it while BigZ is busy!
-
Cool! Why don't you write me the bits and I put them in the guide. I am not the author but just a collector of notes. The guide is meant to be the work of everybody here.
-
Cool! Why don't you write me the bits and I put them in the guide. I am not the author but just a collector of notes. The guide is meant to be the work of everybody here.
Well I've not met the Bolivian army... but..
I have to find a way to get the modules into the act. The installer couldn't fid the modules for the wifi card or any other network device. CLANG! Yes I did go out to a shell and look around. If I put the tar file on the SD card will the installer put them in? Almost certainly not. If I put the files on the SD card and mount in the shell will I be able to tar zxvf to get the modules in place. If I do, will that do any good?
By the way, this is looking like you need to have all these extra steps. The TitchyLinux method is no more complicated, initially, than normal install of Debian on a desktop. I'll try those things...
-
Hey, I really don't see the extra steps you mention ... I did exactly what I wrote and things run ...
So tell me more ...
-
Hey, I really don't see the extra steps you mention ... I did exactly what I wrote and things run ...
So tell me more ...
Apologies: I mean this method I'm attempting now! Sorry!
-
No apology. A long shot: Isn't it true that we first have to extract the kernel module tarball together with the rootfs before flashing the kernel?
-
No apology. A long shot: Isn't it true that we first have to extract the kernel module tarball together with the rootfs before flashing the kernel?
I was hoping to avoid all that!
I expect that the rootfs from bigbridge may have to be used for success here, unless I go through the install using neil's files until I get a command line system. Then copy the modules, flash the cortezkernel and hope to get the gui up. I'll keep reading your how to. Eventually sense will break through my foggy brain cells.
-
AFAIK the bigbridge rootfs has the kernel modules only for 2.6.17 and 2.6.20 (and they are EABI). Does the kernel you are testing match this? (Sorry I lost track of which one you're playing with now ...)
-
AFAIK the bigbridge rootfs has the kernel modules only for 2.6.17 and 2.6.20 (and they are EABI). Does the kernel you are testing match this? (Sorry I lost track of which one you're playing with now ...)
I'm not surprised! I'm trying to get OABI + OABI working. So I am using the cortez 2.6.21 kernel So I think I must eventually try setting up the BigBridge way first. Copy the 2.6.21 modules over and flash cortez's kernel. I suppose what i just tried might work if I built a 2.6.20 OABI kernel: sometime before Christmas I think. Cheers.. I'm going to see if the shell will let me cut my way out of this jungle.
-
Still trying to get OABI + OABI. I started again with the Bigbridge method. I did some twiddles to get the wifi card up. I cheated and copied my interfaces file and my sources.list to the SD card and then to BigZ. I then used tar zxvf from the 2.6.21 modules file on the SD card to the root of the filesystem. Apparently successful.
In order to get the wifi up I had to modprobe the orinoco modules. Then ifup eth0.
I then flashed the OABI 2.6.21 kernel. On booting, it gave the error messages concerning Neil's soundhack package. More significantly it tried to load the 2.6.21 module and said it couldn't find them. What now? Any suggestions? Anybody?
This is crazy! It said it loaded the modules this time.... Which ones?
cortez, if you are around, am I missing something simple here? Is there a script to play with?
Any way wifi is up and I'm going to do dist-upgrade and leave it to stew....
-
Still trying to get OABI + OABI. I started again with the Bigbridge method. I did some twiddles to get the wifi card up. I cheated and copied my interfaces file and my sources.list to the SD card and then to BigZ. I then used tar zxvf from the 2.6.21 modules file on the SD card to the root of the filesystem. Apparently successful.
In order to get the wifi up I had to modprobe the orinoco modules. Then ifup eth0.
I then flashed the OABI 2.6.21 kernel. On booting, it gave the error messages concerning Neil's soundhack package. More significantly it tried to load the 2.6.21 module and said it couldn't find them. What now? Any suggestions? Anybody?
This is crazy! It said it loaded the modules this time.... Which ones?
cortez, if you are around, am I missing something simple here? Is there a script to play with?
Any way wifi is up and I'm going to do dist-upgrade and leave it to stew....
perhaps there's an automatic depmod -a command after boot...
anyway unless you absolutely want 2.6.21, you could also flash 2.6.23 from http://www.angstrom-distribution.org/unsta...spitz/20071024/ (http://www.angstrom-distribution.org/unstable/images/spitz/20071024/)
untar the module.tgz and depmod -a
then reboot
-
perhaps there's an automatic depmod -a command after boot...
anyway unless you absolutely want 2.6.21, you could also flash 2.6.23 from http://www.angstrom-distribution.org/unsta...spitz/20071024/ (http://www.angstrom-distribution.org/unstable/images/spitz/20071024/)
untar the module.tgz and depmod -a
then reboot
JP, I'm trying to match the OABI rootfs that comes from neil and the arm Debian side of thins and an OABI kernel. This in the naive hope that the sound/multimedia things would clear up!
-
JP, I'm trying to match the OABI rootfs that comes from neil and the arm Debian side of thins and an OABI kernel. This in the naive hope that the sound/multimedia things would clear up!
I think I'll put BigZ back on lenny with the original way of setting up. I think that if I want OABI kernel + OABI rootfs, without a lot of fiddles I'll need to do a 2.6.20 OABI kernel. I have the bitbake build system. This gives me a reason to use it. One of my experiments allowed me to set up the OABI filesystem and flash the OABI 2.6.21 kernel. The matching modules had to be transplanted after install. Too many fiddles... too little success...
So I'm thinking that if I build OABI 2.6.20, as a first attempt with bitbake, I might be able to do that experiment again with this kernel. I know that the 2.6.20 modules that Neil's installer uses are EABI. So, I don't think it will work straight away. But if I then put the OABI 2.6.20 modules in, yes I'll need to build those I suppose, as replacements for the EABI ones, the system should be able to reboot with a chance of working as well as the original. I'll still have to do manual fixes instead of using zaurus-hardware-support. When I have a working system I can tarball it and use it like the Bigbridge one. So could other people?
That looks like a plan, but for the long term.
For the short term, back to the original and learn it thoroughly! So unless someone comes up with a cunning plan I'm leaving off this thread.
-
cortez, if you are around, am I missing something simple here? Is there a script to play with?
Sorry, but I'm afraid I currently can't help you out. I started from scratch again, using the EABI 2.6.23 Angstrom kernel in combination with the ADS EABI rootfs.
-
cortez, if you are around, am I missing something simple here? Is there a script to play with?
Sorry, but I'm afraid I currently can't help you out. I started from scratch again, using the EABI 2.6.23 Angstrom kernel in combination with the ADS EABI rootfs.
Not a problem. The route you are going looks interesting, I don't blame you in the least. I'm having a few brain waves about the hardware support deb.
That'll kep me busy!