[solved] Need help starting WiFi after standby or suspend

[danielm]

This is a desperate cry for help. Right now I am very upset with my Pandora and I feel like taking a hammer and smashing it to pieces! Since I got this expensive toy, I spent more time fixing things than actually playing with it :-(

Here is my current problem: Some days ago, SuperZaxxon 1.55 was released and I updated the OS with the internal "update OS" function. Since then, my WiFi doesn't switch on when I wake up my Pandora from standby or suspend.

This leaves me with two choices:

1) Reboot and suffer from the nubs malfunctioning for at least 5 minutes (no, "a quick nub-dance" is not enough in my case!)

2) Restart NetworkManager in the terminal

Naturally, I started using choice 2), but that is not a solution for my problem. To make things worse, every now and then, I get a funny (not!) error message telling me that "NetworkManager is already running". Then I am only left with choice 1), reboot. The last OS I had to reboot that often was MS Windows 95 :-(

Apart from DMESG, I have no means of finding out what is going on on my Pandora. And DMESG is not helping at all.

Is there any other way to find out what is wrong with the SuperZaxxon installation? I recently installed 1.54 from scratch and I didn't temper with anything.

Or is there any way to send NetworkManager to hell? Either by killing it and starting it again automatically every time the Pandora wakes up, or by un-installing it and installing something that works. (I know that NetworkManager is buggy and I know it will never get fixed and I know that there is an alternative)

Yes, I am asking for quite a lot, but i am not capable of solving this myself. The whole thing makes me feel terribly stupid and like I got screwed :-(

 

[mosschops]

Re-install the update from a file. IIRC there is a note which says you need to run the upgrade twice due to just this happening. Same happened to me but I just read and followed the instructions

 

[danielm]

Ah, then I have to wait for an update-PND to be released? I tried the OS-updater twice, but the second time it didn't install anything (not a big surprise ...).

I was actually wondering why no update file was released, like with v1.54. The update function can be terribly dangerous ...

 

[TrashyMG]

The update function can be terribly dangerous ...

Actually I've had no issues using it since the 1.52 update which was known to break the wifi temporarily if not ran twice. And I've ran it on all three Pandora models I own.

 

[danielm]

Thanks for the info! "Update Pandora OS" doesn't do anything since i used it to update from 1.54 to 1.55.

I just ran the old 1.54 offline updater (to downgrade, which didn't work) and then "Update Pandora OS" two times ... WiFi is still broken.

Sorry, but I am sooooo tired of my Pandora :-( I didn't mess with anything and I don't want to re-flash again and maybe lose my settings, desktop icons, etc. :-(

 

[danielm]

Found out that the Power Manager caused my problem. The "power button action" was set to suspend, which apparently does something different than the suspend that is activated on flicking the power switch when the "power button action" is set to "nothing"...

 

[levi]

Flicking the power switch right normally causes the system to go into 'low power' mode, which boils down to: underclock, wifi, usb, bluetooth, tv-out off. I don't know what suspend mode does on the Pandora, but on a PC it normally means write RAM to disc and power off. Coming back up from suspend is a bit more fraught, as you need to be able to repopulate all the RAM before any stuff starts to run, and it took a number of years before it worked properly on Linux on x86, so I'm not massively surprised it's broken on Pandora. Given low power mode means we only need to charge every few days if not heavily using our Pandoras, most of us live with that.

 

[_jr_]

Found out that the Power Manager caused my problem. The "power button action" was set to suspend, which apparently does something different than the suspend that is activated on flicking the power switch when the "power button action" is set to "nothing"...

it probably isn't running op_power.sh which takes care of reenabling wifi and bt after wakeup. A comment indicates that this is neccessary because of missing kernel functionality. I'm not sure if it is worth trying to work around this in xfce, too.

 

[Binky]

I think suspend is suspend-to-ram, there being no always-present disk space to suspend to.

 

[levi]

Yes, that would make sense, as I've noticed the XFCE power manager also references the 'hibernate' state, which sounds more along the lines I was thinking.

 

[danielm]

Flicking the power switch right normally causes the system to go into 'low power' mode, which boils down to: underclock, wifi, usb, bluetooth, tv-out off. I don't know what suspend mode does on the Pandora, but on a PC it normally means write RAM to disc and power off. Coming back up from suspend is a bit more fraught, as you need to be able to repopulate all the RAM before any stuff starts to run, and it took a number of years before it worked properly on Linux on x86, so I'm not massively surprised it's broken on Pandora. Given low power mode means we only need to charge every few days if not heavily using our Pandoras, most of us live with that.

Ah, and I thought the power switch would activate suspend when no power is connected, and low power mode with the "charger" connected. Looks like I got this wrong again ...

 

[dimag0g]

Is it possible to teach xfce to run op_power.sh when coming back from suspend?

 

[WizardStan]

Not a good idea. op_power.sh IS running when it comes back from suspend.

What happens is you hit the power switch, which causes op_power.sh to start running: it turns the screen and wifi off, does all it's power saving things, and then issues a "CPU sleep" command, right there in the middle of the script. When you hit the power switch again it wakes the CPU, the script then continues exactly where it left off, re-enabling and restoring everything to the way it was when the script started.

I have no idea what danielm's particular problem was in that case, but right now Levi is wrong (he may not have been wrong at the time, I can't remember when this functionality was added) and suspend does exactly what danielm thought that it did. Unless you are having a specific problem that we can help solve there's no need to go modifying anything.

 

[danielm]

Wizard, are you confusing low power mode with suspend to RAM?

Suspend is still not handled well on the Pandora, and probably never will. And there will probably never be an update of NetworkManager that might fix the suspend problem with WiFi either. I started disabling WiFi before I suspend my Pandora, and so far, WiFi seems to work afterwards when I re-enable it.

 

[WizardStan]

Suspend is still not handled well on the Pandora, and probably never will

I use it all the time. What doesn't work about it? Have you updated the OS recently? We get weeks of downtime thanks to suspend mode.

Wizard, are you confusing low power mode with suspend to RAM?

No, why would you think that? The process I described is exactly what it does in order to suspend. There is one additional step in there I skipped over, and that is that it checks if the charger is plugged in; if so then it switches to low power mode instead of turning off the CPU and then exits the script. This is exactly like what you said earlier, you weren't wrong, that is how it behaves.

 

[danielm]

Suspend is still not handled well on the Pandora, and probably never will

I use it all the time. What doesn't work about it? Have you updated the OS recently? We get weeks of downtime thanks to suspend mode.

That's all written in my first posting here in this thread. My workaround is to deactivate WiFi before I suspend, which seams to work well.

I have re-installed the OS and it is up to date.

 

[dimag0g]

It's hard to tell what your problem is. Pandora suspend script (op_power.sh) has the code which disables wifi before suspending. Because this code works for me, there is no way I could find what's wrong in your case.

You might compare code from op_power.sh and op_wifi.sh (assuming that the latter script works well for you). Normally there is no difference. 

You might also add some sleep after rmmod is called to make sure there is no race condition when wifi driver is removed.