Opensimpad-0.9.0rc1 On T-sinuspad!

[ric]

Hi all,

yesterday, I managed to run opensimpad-0.9.0rc1 on my stock T-Sinuspad!
Unfortunately, my solution is not permanent so far.

I take no responsibility for any damage that could happen when you try this out!

I'm running the hh.org bootldr that Chris had provided in combination with the RC1. This one seems to have a hardcoded partition table which needs 32MB of flash which the Sinuspad obviously doesn't have.
What I did is this:
After booting into the bootldr, I deleted the partitions home and root.
After that, I created a new root partition with the command

Code: [Select]

partition create root 0x40000 0xfc0000 16This is a partition that uses all the flash memory between the bootloader partition and the end of flash.
I erased this partition with

Code: [Select]

eflash rootNow, I loaded the jffs2-file with

Code: [Select]

load rootand uploaded the file http://opensimpad.org/bootldr-simpad/opens...0rc1_opie.jffs2 to the pad.
After the command

Code: [Select]

boot opie 1.2 started on my machine. I had not much time to try but most applications seemed to work. Some apps made the pad hang which was not so nice... but they were images for testing purposes, anyway.
Next time, I'm going to try one of the images of devernay.

Now to the problem: After suspend or reboot, the bootldr reconstructs the original partition table with the big root partition lasting to address 17c0000 and the home partition which I wouldn't need anyway. This leads to a kernel panic when rebooting into opie. After changing the partition table again and booting, it works again.

Is there a way to change the default partitions other than compiling a new bootldr? Maybe the params are the key. I need to dig deeper into this.
Maybe, I'm going to get the toolchain and build my own image, maybe someone could build an image after some patching of the bootldr.

Any help and hints are appreciated.

Greetings
Richard

[Berend]

Very nice!

I had the same problem and since the partions got reset I didn't bother to actually try to load an image (edit: did try to load but not with edited partition table). Just tried it, flashed the small 0.9.0-rc1 GPE image and booted, but things go very wrong when calibrating and setting up (freezes after each setup step, reboot a few times and finally no response after setup anymore). Fun to see it booting though.

one nitpick:

Code: [Select]

partition create root 0x40000 0xfc0000 16becomes

Code: [Select]

partition define root 0x40000 0xfc0000 16
I'll try some other images this weekend, it would be very cool to get this to work. Thanks.

[ric]

Hi Berend,

in the meantime, I gave the gpe-image a try, too with the same results.
I'm not really sure if there is a major problem with the bootldr itself leading to these problems or the images which maybe rely on an existing home partition or whatever.
Any idea if the issue about the reset partitions can be solved without touching the bootldr sourcecode? I'm not even close to compiling anything since I have no linux machine running (except my server) at this time - shame on me...
But I think there is a certain need to support 16MB Simpads after the recent flooding of sinuspads and the swisscom(?) pads.
Sooner or later, I will dig into the oe toolchain. But I must admit, I have a couple of priorities, first of all getting my new house ready :-)
Maybe it is worth waiting for Angström which I haven't very much info about right now.
I will keep trying the images I can find for now.
Anyone out there who can provide images suitable for 16MB simpads?

Greetings
Richard

[jochen]

Hi Berend,

in the meantime, I gave the gpe-image a try, too with the same results.
I'm not really sure if there is a major problem with the bootldr itself leading to these problems or the images which maybe rely on an existing home partition or whatever.
[div align=\"right\"][a href=\"index.php?act=findpost&pid=104937\"][{POST_SNAPBACK}][/a][/div]

One problem could be the different memory size. While bootloader detects the smaller device with 32MB main memory, the kernel assumes 64MB and behaves strange when trying to use nonexisting memory. I sort of fixed this by adding mem=32M to the kernel parameters on my "pimped" CL4 (added second flash chip to the main board).

[ric]

Just a quick update:
gpe seems to run ok on the sinuspad with the mem=32M parameter. But still no idea how to have the bootloader start with usable partitions.
I'v contacted zecke to get the latest bootldr sources. Maybe this is the way to go.

Greetings
Richard

[albertr]

I might be wrong, but I think that you need to recompile the kernel with static mtd partition definitions, since bootloader passes wrong partition definitions to the kernel.
Try that first before touching bootloader's code.
-albertr

[ric]

Currently, I'm digging into bitbake/openembedded. Maybe eventually, I find some code to change partitions via the kernel which I indeed would prefer over manipulating the bootldr.
At the moment, I have problems building an opie-image from openembedded which I must understand at first.
But thanks for the hints.

Greetings
Richard

[sash]

Hi!

Look at params from guylhem.
He soldered a mmc-card and uses all the 32mb flash for the root partition.

So I think you also can change your /boot/params to permanently take your 16mb flash for root.