Release 3.2 kernel issue tracking thread

[okiseiji]

Hello, I have a question,


I have received my pandora rebirth, and on my mail, it' written:


It's included on your OS. Press the right shoulder button while


switching on the machine and select "Boot experimental kernel".


Ok so I press the R shoulder button, but what do you mean by "while switching on the machine"?


Thx

 

[Linux-SWAT]

Hi !


While powering on.

 

[okiseiji]

Hi!!


I'm on the boot menu:


default boot


power off


usb serial prompt


serial prompt


Which one is the kernel3.2?


THXXX

 

[Linux-SWAT]

Seems you don't have the latest version of the OS.


Download this :


http://openpandora.o...erZaxxon-B1.zip


Then follow the procedure (don't download the old package, use the one above) explained here, under "complete firmware image" :


http://www.open-pand...emid=32&lang=en

 

[okiseiji]

Oki je vais essayer merci!!

 

[Askarus]

You have to load usb modules :

sudo modprobe ehci-hcd

Do someone have a fix for the middle click ? I tried tweaking xorg.conf without success.

I typed it in the terminal, but nothing happened.

 

[Linux-SWAT]

if you type

lsusb

before and after, you'll see the module is loaded, so your device should work.

 

[okiseiji]

Hi,


so I have installed boot experimental kermel.


Now My applications don't run (like firefox, gba emulator, deadbeef)


I have to erase them and download them again on the SD card?

 

[Link]

^


No it is my understanding that the new experiemental kernel doesn't support all .PND files yet. It won't read the 64GB SDXC cards I'm using at all, but don't go erasing you .PNDs and trying to reinstall them, there are some things that work, but it won't happen overnight regarding the new experimental OS to take hold. I'd do some resarch to see what's working before erasing.

 

[notaz]

Talking about SDXC, I've bought random 64GB SDXC card and it works fine. I guess I'll have to send some debug kernels, I'll PM you someday, hopefully.

 

[jottt]

I'm not quite sure if thats related to the kernel but:


When I start a terminal and enter (with the smplayer2 pnd)


./mplayer2.sh -radio device=/dev/radio0 adevice=hw=1.0:arate=96000 -rawaudio rate=96000 radio://89.3/capture


The system starts playing (even though I don't hear anything....probably wrong audio device but thats not the issue)


When I now press CTRL + C to stop it, I get the usual prompt but the terminal does not show what I enter. It still reacts on WHAT I enter and (for example) a "ping localhost" still prints the output. I have to close an restart the terminal to have my text back.


This also happens when in a SSH-Session. I have to reconnect.


I cant say what happens with the older kernel as mplayer does not even start (because of an issue with the module).


EDIT: changed the error description...it's actually a bit different than I thought.

 

[Darkknight512]

When flicking the power switch to go into low power mode and come back it shows swap device read error. I have a swap partition on my second sd card. The error shows 6 times with different numbers.

 

[Link]

Talking about SDXC, I've bought random 64GB SDXC card and it works fine. I guess I'll have to send some debug kernels, I'll PM you someday, hopefully.

Yes, this would be interesting to test out. I'm hoping to have a few newer SDXC cards from Pretec in the next week or two as well.

 

[notaz]

When flicking the power switch to go into low power mode and come back it shows swap device read error. I have a swap partition on my second sd card. The error shows 6 times with different numbers.

The cards are "removed" on suspend so not much can be done here, except maybe suspend script could unmout all swap partitions and mount them back after suspend.

 

[Galaxis]

Suspending the system when an ext3-formatted SD card is active seems to cause problems too. Just had a ton of ext3 errors scrolling through while shutting down a system after it was in suspend once (that was when booting off NAND, didn't install SuperZaxxon to an SD card yet).


The eject thing will make running off SD cards (which was what I usually did up to now) quite unattractive..

 

[Darkknight512]

When flicking the power switch to go into low power mode and come back it shows swap device read error. I have a swap partition on my second sd card. The error shows 6 times with different numbers.

The cards are "removed" on suspend so not much can be done here, except maybe suspend script could unmout all swap partitions and mount them back after suspend.

The main problem is, I can't get back into the Os when that happens though.

 

[Linux-SWAT]

It's like if you remove a ram bar when the unit is on.

 

[Grench]

When flicking the power switch to go into low power mode and come back it shows swap device read error. I have a swap partition on my second sd card. The error shows 6 times with different numbers.

The cards are "removed" on suspend so not much can be done here, except maybe suspend script could unmout all swap partitions and mount them back after suspend.

Since mount/umount works on partitions - not the physical card...


Would it be possible to force suspend to not umount partitions of the type /swap?


Yeah - that probably gets into then forcing suspend to continue to supply power to the SD slots during suspend... Hmm...


Crazy idea #2.


What would happen if you were to erase/format the NAND as /swap and boot from the SD card? Would the NAND be protected during suspend?

 

[notaz]

The problem is that cards are basically removed and unmount isn't done at all by kernel itself, I've only added a bit of that handling to suspend script. It's like yanking out SATA cable on a running PC. There is a kernel option to not do that "remove" thing on suspend, but if someone swaps a card while suspended, kernel help text says that the whole card is almost guaranteed to be corrupted (the kernel could be smarter here really and check if card was changed or something..).


Meanwhile NAND is always being suspened/resumed fine.

 

[Grench]

The problem is that cards are basically removed and unmount isn't done at all by kernel itself, I've only added a bit of that handling to suspend script. It's like yanking out SATA cable on a running PC. There is a kernel option to not do that "remove" thing on suspend, but if someone swaps a card while suspended, kernel help text says that the whole card is almost guaranteed to be corrupted (the kernel could be smarter here really and check if card was changed or something..).


Meanwhile NAND is always being suspened/resumed fine.

Card corruption concerns sounds like they're leaving partial writes pending during suspend?


You could take the easy way out...


When you get the signal for the device to suspend, have it pop up an, "Are you sure?" box, defaulting to 'No' that details how if the user removes or replaces an SD card during the suspend that it will likely destroy the data on the card, cause an earthquake and do uspeakable acts to the user's pets. Have all 6 LED lights flash a bright SOS signal while this pop-up is up. Also be sure to include a checkbox for, "I get it - don't show this again."


I think that those of us who boot from SD would rather have it not "remove" on suspend - if they screw up they should have known better and they know it.


I can see where it would be better for NAND boot users - they would be the ones who would swap cards on a running device then blame someone else for it.


Hmm... I realize that suspend is supposed to be an ultimate low power mode - but would it help to have a red LED fade in/out to remind the user that the device is in suspend mode so they don't swap cards thinking it's powered off?