[ltp] Re: spotty wireless with t520 w/Intel wifi and xubuntu 13.04
Sean Leonard
linux-thinkpad@linux-thinkpad.org
Sat, 26 Oct 2013 18:20:52 -0700
--047d7b62535a61e52904e9aecbf9
Content-Type: text/plain; charset=ISO-8859-1
On Sat, 26 Oct 2013 08:06:47 +0300, Michael Livshin <
linux-thinkpad@cmm.kakpryg.net> wrote:
> Even though
> Intel wireless cards are supposed to be well-supported in Linux. Go
> figure.
I, too, have been disappointed in Linux support for Intel devices (mainly
my video card, I haven't had issues with my T410s' wifi), while my
impression has always been that they're pro-Linux.
> Where was I? Ah, yes, NetworkManager. See, that fine piece of software
> is periodically (as in every fscking two minutes) issuing AP scans.
> _While_ you are connected. The idea, as I understand it, is to support
> WiFi roaming transparently -- I mean, who cares for those sedentary
> idiots like your truly who just want to sit at the table, or on the
> couch, and work on their laptop? This cool portable device, which has
> WiFi? While _seated_? Really, who the hell would want to do that,
> right?
I agree that it's odd. It's not my phone, after all, and moving while
using is minimal.
> The NetworkManager maintainers are refusing to introduce a configuration
> knob to turn those moronic periodic scans off, because they figure that
> any problem caused by scanning is always the driver's fault. So
> everyone should just fix their drivers (sound familiar? Yes, all the
> NetworkManager maintainers work at RedHat).
Ah, yes. In that Perfect World, I say the same. In This World, Linux
doesn't have the use base to make them care. Yet ;)
> A passable workaround
> (which has been making using my T420 on the home network bearable for a
> month now) is to patch the NetworkManager source and build your own deb
> for it (but make sure to tweak the version in debian/changelog, or
> apt-get will bring back the official package at the first opportinity).
>
> See the attached patch.
Thanks for the patch! Where is it from, that it may be tracked and
discussions or changes followed?
Also, I wouldn't tweak the version, but tell apt to "hold" it:
'echo "network-manager hold" | dpkg --set-selections'
Or whichever appropriate NM spin you have.
And 'echo "network-manager install" | dpkg --set-selections' to remove that
hold and allow for upgrading.
> (You'd expect some nice individual to maintain a PPA and The Community
> (whatever that is)
That, sir, is everyone here and all the people who work very hard to create
the amazing ecosystem that has come to be so highly functional and Free
Libretto and Open Source.
> Or perhaps (*gasp*) you'd expect sane distributors to just
> fix the stupid problem in their packaged versions, right? Right.)
More deviations from Upstream makes more work. Ideal changes are handled
automagically and all tests pass and bugs don't multiply, but that is for
the good cases.
Regards,
sean
--047d7b62535a61e52904e9aecbf9
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
<p dir=3D"ltr">On=A0 Sat, 26 Oct 2013 08:06:47 +0300, Michael Livshin <<=
a href=3D"mailto:linux-thinkpad@cmm.kakpryg.net">linux-thinkpad@cmm.kakpryg=
.net</a>> wrote:<br>
> Even though<br>
> Intel wireless cards are supposed to be well-supported in Linux. =A0Go=
<br>
> figure.</p>
<p dir=3D"ltr">I, too, have been disappointed in Linux support for Intel de=
vices (mainly my video card, I haven't had issues with my T410s' wi=
fi), while my impression has always been that they're pro-Linux.</p>
<p dir=3D"ltr">> Where was I? =A0Ah, yes, NetworkManager. =A0See, that f=
ine piece of software<br>
> is periodically (as in every fscking two minutes) issuing AP scans.<br=
>
> _While_ you are connected. =A0The idea, as I understand it, is to supp=
ort<br>
> WiFi roaming transparently -- I mean, who cares for those sedentary<br=
>
> idiots like your truly who just want to sit at the table, or on the<br=
>
> couch, and work on their laptop? =A0This cool portable device, which h=
as<br>
> WiFi? =A0While _seated_? =A0Really, who the hell would want to do that=
,<br>
> right?</p>
<p dir=3D"ltr">I agree that it's odd.=A0 It's not my phone, after a=
ll, and moving while using is minimal.</p>
<p dir=3D"ltr">> The NetworkManager maintainers are refusing to introduc=
e a configuration<br>
> knob to turn those moronic periodic scans off, because they figure tha=
t<br>
> any problem caused by scanning is always the driver's fault. =A0So=
<br>
> everyone should just fix their drivers (sound familiar? =A0Yes, all th=
e<br>
> NetworkManager maintainers work at RedHat).</p>
<p dir=3D"ltr">Ah, yes.=A0 In that Perfect World, I say the same.=A0 In Thi=
s World, Linux doesn't have the use base to make them care.=A0 Yet ;)</=
p>
<p dir=3D"ltr">> A passable workaround<br>
> (which has been making using my T420 on the home network bearable for =
a<br>
> month now) is to patch the NetworkManager source and build your own de=
b<br>
> for it (but make sure to tweak the version in debian/changelog, or<br>
> apt-get will bring back the official package at the first opportinity)=
.<br>
><br>
> See the attached patch.</p>
<p dir=3D"ltr">Thanks for the patch!=A0 Where is it from, that it may be tr=
acked and discussions or changes followed?</p>
<p dir=3D"ltr">Also, I wouldn't tweak the version, but tell apt to &quo=
t;hold" it:<br>
'echo "network-manager hold" | dpkg --set-selections'<br>
Or whichever appropriate NM spin you have.<br>
And 'echo "network-manager install" | dpkg --set-selections&#=
39; to remove that hold and allow for upgrading.</p>
<p dir=3D"ltr">> (You'd expect some nice individual to maintain a PP=
A and The Community<br>
> (whatever that is)</p>
<p dir=3D"ltr">That, sir, is everyone here and all the people who work very=
hard to create the amazing ecosystem that has come to be so highly functio=
nal and Free Libretto and Open Source.</p>
<p dir=3D"ltr">> Or perhaps (*gasp*) you'd expect sane distributors =
to just<br>
> fix the stupid problem in their packaged versions, right? =A0Right.)</=
p>
<p dir=3D"ltr">More deviations from Upstream makes more work.=A0 Ideal chan=
ges are handled automagically and all tests pass and bugs don't multipl=
y, but that is for the good cases.<br></p>
<p dir=3D"ltr">Regards,<br>
sean</p>
--047d7b62535a61e52904e9aecbf9--