OESF Portables Forum
Everything Else => Zaurus Distro Support and Discussion => Distros, Development, and Model Specific Forums => Archived Forums => Angstrom & OpenZaurus => Topic started by: Hrw on September 07, 2005, 05:13:32 am
-
On my website (http://www.hrw.one.pl/tmp/poodle/) I put r43 kernel for poodle with APM fix which I ported from collie fix made by Hector Martin.
Place zImage-poodle-20050907085254.bin on CF or SD card, rename it to zImage.bin, add proper updater.sh (version from OZ 3.5.3 should work) and flash kernel (initrd does not have to be flashed).
If you got "unresolved symbols" when loading modules then look into "modules" subdirectory where you will find all kernel modules (without out-of-kernel modules like hostap or orinoco). Packages are named a bit different then it was in OZ 3.5.3.
Report does that kernel works better for your poodle (also give info if it does not).
Keep in mind that this kernel does not fix problems like "shutdown on right arrow hold" as this patch is not ported - I'm not familiar with that crap which is called "2.4.18 sharp kernel sources" and dont have device to test.
I dont know does it is for PXA250 or PXA255 device - it's plain OE kernel + one patch.
-
I dont know does it is for PXA250 or PXA255 device - it's plain OE kernel + one patch.
poodle-defconfig builds for pxa250.
-
Report does that kernel works better for your poodle (also give info if it does not).
I can't get the r43 kernel to boot past the OpenZaurus screen with a blinking curser at the top left corner.
I'm using the Hentges T3 image (the latest build) and I have a PXA250. I have removed the checkversion script but it doesn't get that far. It appeared to flash normally but I don't have an md5sum to check the kernel... can you post one?
If you think it makes a difference I can flash the official 3.5.3 oz image?
Has anyone tested this with the Official 3.5.3 oz image?
Thanks,
Greg
-
257c9fc103faec337996b719eb102a9f zImage-poodle-20050907085254.bin
Which you test it should be not important - it should boot.
-
I checked the md5sum on the CF-card and it is ok.
Any ideas?
-
Report does that kernel works better for your poodle (also give info if it does not).
I can't get the r43 kernel to boot past the OpenZaurus screen with a blinking curser at the top left corner.
I'm using the Hentges T3 image (the latest build) and I have a PXA250. I have removed the checkversion script but it doesn't get that far. It appeared to flash normally but I don't have an md5sum to check the kernel... can you post one?
If you think it makes a difference I can flash the official 3.5.3 oz image?
Has anyone tested this with the Official 3.5.3 oz image?
Thanks,
Greg[div align=\"right\"][a href=\"index.php?act=findpost&pid=94841\"][{POST_SNAPBACK}][/a][/div]
I heve the same behaviour with both kernels (standart and new) and any GUI. bootstrap image runs with any kernel.
Hentges T3 image an its kernel works ok too.
-
If you think it makes a difference I can flash the official 3.5.3 oz image?
It boots up without any problems with the official 3.5.3 image.
Maybe it had something to do with the altboot stuff in the Hentges image and me removing the checkversion script to flash this kernel?
Anyhow it's working and I will check different things out later today.
Thanks,
Greg
-
Ok finally tested Hrw's image. Not quite perfect yet (but a bit better). Had some issues though.
I think something's going on with the updater. I tried flashing the new kernel by itself with OZ's updater.sh and I get that stuck blinking cursor problem as well. This is with an OZ353 build already on my poodle. I tried just reflashing the original kernel and it still gets stuck.
I then tried flashing both initrd and the r43 kernel next, it starts up fine minus that 1 minute version mismatch warning. It boots up fine afterwards.
So much for not saving the initrd... once again start over from scratch *sigh*
Anyway, off to APM testing:
First time the button is pushed and held it freezes the machine for a few seconds and the machine becomes responsive again. Then if I press and hold it again, it will suspend. However if I push the button again to turn it on, it will turn on but then turn off suddenly! then when I push it again, it will turn on and stay on. A bit annoying but it's better than stuck off.
This is just first pass testing... more later...
-
First time the button is pushed and held it freezes the machine for a few seconds and the machine becomes responsive again. Then if I press and hold it again, it will suspend. However if I push the button again to turn it on, it will turn on but then turn off suddenly! then when I push it again, it will turn on and stay on.
This is the same pattern I have.
I will add that if you accidentally suspend with the menu/backlight button or the right arrow button... it will turn on and stay on with 'one' push of the on/off button.
Greg
-
Someone with poodle has to port second fix from collie to get that problem with hardware buttons resolved. It won't be me as I don't have 5600 and (more important) I'm one of that guys which touch 2.4.sharp-lineo-crap kernels just from time to time if at all.
-
I don't write very eloquently, so I hope this is not misunderstood.
I'm very appreciative of the effort to patch this kernel however I don't see that this patch has fixed anything but rather made things worse when compared to the last kernels that CoreDump built for his poodle cardfs images. The Hentges T3 poodle image and kernels on his site simply work better then this kernel. I just thought that I should say something before this is put into mainstream.
Do any poodle users here feel differently? Have any other poodle users tested this yet?
Thanks for all your work Hrw,
Greg
-
I haven't tried the hentges kernel, where is the source code available? I'd like to look at that too and compare.
What I'm really curious about is why it was working and now broken. I'm also not in sync with development because I'm a newcomer to all this (I haven't even tried my own hack yet, if everything else is equal it should be the exact same behavior...) Question is, what changed, why'd it break... Anyone familiar enough with monotone with how to check out old versions without making a whole new build environment?
I need a new disk drive for all this, it's terrible (not to mention my build machine is only a celeron...)
-
Keep in mind that this kernel does not fix problems like "shutdown on right arrow hold" as this patch is not ported - I'm not familiar with that crap which is called "2.4.18 sharp kernel sources" and dont have device to test.
Can someone explain how this kernel is better? I thought the hentges kernel already had the APM fix (I'm running it now and haven't had any power off problems in about a week). The ONLY problems I'm still having are the "shutdown on right arrow hold" problems and some backlight problems. I'm grateful someone at least cares though.
-
Heh... more kernel hacking.
Right arrow key problem is fixed in this patch *I THINK... "it works for me" famous last words ;-)*, BUT I do _NOT_ like how apm is working here (same as before, the useless suspend hanging push, then push suspend, then upon resume it suspends right away again, then resume... see other posts).
If you want to try this kernel, here it is: right arrow plus broken apm fix zImage binary (http://q.dyndns.org/~blc/zaurus/zImage-apm-poodle-rightarrow)
The patch, again against -r41 because monotone doesn't seem to want me to checkout anything newer, is in the attachment and also in my webpage directory. I imagine it will patch cleanly against a newer version.
NOTE: My version stuff is really screwed up on this build, so don't use for production use, just to play with. You can workaround the issues if you want though, that's your choice Also it seems to break compatibility with hostap, I might have mucked with some stuff I shouldnt have when customizing other stuff.
I guess I'll have to study how apm works some more or find working code to mess with...
-
eccerr0r: can you open bug in OE bugtracker and put that patch there?
Forums are not bugtrackers.
-
Looked at the hentges oe stuff. did a diff against his 3.5.3-T3 (-R39 ???!) .bb and .inc files and it looks mostly the same as the -R41 that I picked up from monotone. Were there any user mode hacks? all that appears to be changed is mostly the cache bugfix where hentges disabled the workaround.
Seems there are so many bug trackers, not sure which one is still being used, a lot of them are just idle... is poodle kernel bugfixing unmaintained now? (posted on the one on sourceforge since I already had an existing account...)
What was changed in -R42, is -R43 just the experimental APM patch?
-
there was no r42 officially
-
Ah, another one that has the problem, that you cannot upgrade the kernel on poodle without reflashing initrd.
I have this problem, too and it keeps me away from playing around with kernels.
Can this problem be fixed?
-
jinidog: really? There are users which made it.
First upgrade modules, then upgrade kernel.
-
I have this problem for every kernel I flash.
the device stops booting long before the modules comes into the process