[GRLUG] wifi connection drops under Ubuntu

[Michael Mol]

On Thu, Apr 8, 2010 at 6:58 PM, John-Thomas Richards <jtr at jrichards.org> wrote:
> On Thu, Apr 08, 2010 at 06:41:38PM -0400, Michael Mol wrote:
>> On Thu, Apr 8, 2010 at 6:19 PM, John-Thomas Richards <jtr at jrichards.org> wrote:
>> > I bought an Acer Aspire One for my wife.  It came with Win7 Starter
>> > Edition.  W7SE is a dog.  I installed Ubuntu (full, not Netbook Remix)
>> > on it.  It is much snappier than Win7.  However, wifi does not work as
>> > well under Ubuntu as it does under Win7.  Win7 shows signal strength
>> > around 85% and Ubuntu (NetworkManager) shows it around 42%.  Googling
>> > has revealed that these figures may not be accurate.  I am inclined to
>> > trust them since Win7 never experiences loss of network connectivity
>> > whereas Ubuntu experiences frequent drops.
>>
>> The actual "quality" measurement isn't so simple as a percentage; the
>> driver itself will be reporting a signal strength in dB and a noise
>> strength in dB. Your actual connection quality depends on your
>> signal-to-noise ratio. That "percentage" is going to be some
>> calculation based on those two numbers. If you drop into an xterm and
>> run 'iwconfig (interface)', what does it say?
>
> root at djr-laptop:~# iwconfig wlan0
>          Link Quality=26/70  Signal level=-84 dBm

Can't interpret this right now without looking at my laptop.

>
>> > I disabled power management
>> > on the wifi card (thinking that perhaps Ubuntu is mistakenly reducing
>> > the power to it in an effort to save power).  This does not eliminate
>> > the loss of connectivity.  I am not sure what else to do, but no matter
>> > what, I cannot tell my wife to just boot into Windows.  Any ideas?
>>
>> Watch /var/messages and /var/syslog. There's likely something interesting there.
>
> It just lost connection again.  dmesg shows multiple entries of this:
>
> [ 1927.216729] ath9k: DMA failed to stop in 10 ms AR_CR=0x00000024 AR_DIAG_SW=0x40000020

Sounds like the ath9k driver is having problems?

>
>> It's also possible that the Aspire One generates more EMI when running
>> Ubuntu's drivers and policies than when running Windows' drivers and
>> policies, based on configuration of the device. It may actually *help*
>> to be more aggressive on power policies for things like your hard disk
>> and CPU speed.
>
> How could Ubuntu's drivers and policies result in more EMI?

As far as drivers, it depends on the device and any quirks of the hardware.

As far as policies, any logic that causes more system activity will
result in greater EMI, and if the physical system isn't properly
shielded, that EMI will have an impact outside the actual machine. I
noted power policies in particular because putting a hard drive to
sleep, or underclocking the CPU, necessarily results in less
electrical activity.

It's not *supposed* to be an observable thing, but that's because your
computer is supposed to act as RF shielding for its internal
components.

-- 
:wq