[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#847947: firmware-libertas: Wifi connection drastically drops after a few seconds on some hotspots



Package: firmware-libertas
Version: 20161130-1
Severity: important

Hello,

I am new to the bug reporting on Debian and thus I apologize in advance for the
likely poor quality of this bug report. I will try to explain throughly the 
issue although it is a little bit difficult to reproduce.

- Case where it happens : *One* particular wifi hoptspot which happens to be
the one I use at home.
My phone tethering, work wifi hotspot and university wifi hotspot all work
flawlessly though.

- Expected behavior : Connect to the wifi, get redirected to the login portal,
browse the internet with optimal speed.

- Actual behavior : Connect to the wifi, get redirected to the login portal,
browse the internet at speeds up to 64Kbps (~7-8Ko/s observed).

- Details :
-- The internet does work decently, as in it's not 7-8Ko/s of basic networking
handshakes pretending to be data. It is actual data and any website can be
loaded privided they have high timeout numbers and you are patient enough.

-- This issue doesn't exist in the testing (20160824-1) version of the package,
only on the sid one (20161130-1).

-- I am running a Microsoft Surface Book right now, on a custom kernel image which
forces the wifi card to never go on power save mode as this results in the
described behavior for any version of the firmware-libertas package.

-- I tried other network configurations, either disabling powersave through iw,
or through networkmanager (which is the manager I use). Also no analysing of
either kernel module, package, networkmanager, wpa_supplicant logs showed any
different behavior with the working wifi access points described earlier.

-- Wifi IP packets show no sign of refusal to serve data from my access point
either.

-- Downgrading to the testing version (20160824-1) solved my issue.


- Possible fix :
The might have been some powersave-related checks for it being enabled added
in the source code that prevent the kernel patches from disabling it on
hardware where it is necessary because Microsoft apparently tried some
creativity in their hardware design this time around...

- Addendum :
I'm sorry I don't really have much more to this case to say. I hope someone
who actually edited that code upstream sees this and can bring in some
insight.
I've been fighting with this bugs for a whole week now and nothing was ever
concluant from what I looked at, which is why I don't even have logs to show
for it.

Thanks for reading.

-- System Information:
Debian Release: stretch/sid
  APT prefers unstable
  APT policy: (500, 'unstable'), (500, 'testing'), (1, 'experimental')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 4.9.0-rc8-mssp4+ (SMP w/4 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
Init: systemd (via /run/systemd/system)

firmware-libertas depends on no packages.

firmware-libertas recommends no packages.

Versions of packages firmware-libertas suggests:
ii  initramfs-tools  0.125

-- no debconf information


Reply to: