Realtek RTL8191SE

[GiladGressel]

HI guys,

I upgraded my asus1201n to natty 11.04 on a fresh install

i downloaded a livecd and ran from USB. (upgrading in 10.10 was broken...)

The wifi signals all appear to be strong and connect very fast (compared to the wrapper i was using in 10.10)

however it's still broken. It worked really fast for about 20 minutes, but I left it downloading a torrent and when I came back it had stopped working. The signal is still there (it doesn't continually ask for password etc) but there is no activity when you try to browse or download

please note -- this problem appears to only exist in WPA password networks. Right now I'm on a cafe wifi with no problems

so... still broken. will probably go to ndswrapper later on.
also of interest is that the asus1201n has an annoying error with the NVidia graphics. This error might be more than annoying if I was to play any serious 3d games.
https://bugs.launchpad.net/ubuntu/+s...ey/+bug/771788

-gilad

[paralia20105]

OMG thank you so much!! my wireless now works !!! thank you!

For me too. Thank you so much !!

[hackwater]

I too am struggling with this issue in Natty (64-bit Kubuntu; I did a fresh install rather than upgrade my Lucid install). I've tried compiling the module from the source available on the Realtek site to no avail. The rmmod/modprobe combination can successfully force the wireless back into operation, but it isn't a permanent fix, as it will go down again.

I note in the release notes from Realtek that the version of the driver publicly available is good for kernels 2.6.27 - 2.6.37; I suppose it's possible that they are not supporting 2.6.38 yet. I also note that this driver seems to be the same driver as for the 8192SE; I'm wondering if that's an issue.

In Windows, my laptop warns about interference from Bluetooth and the wireless card; since I wasn't using Bluetooth, I disabled it in Windows and attempted to disable it in Kubuntu, though I'm not 100% certain I succeeded. (I disabled the Powered option in the Bluetooth settings and stopped the bluetooth service.) This does not seem to affect the wireless's behavior.

I am now trying to contact Realtek to see if there's a driver for Natty/11.04 and/or the 2.6.38 Linux kernel; I will update with their response if any.

[neo1786]

i had similar problems... mine was 8192ce tho

http://ubuntuforums.org/showthread.php?t=1790306

[hackwater]

While waiting for Realtek to get back to me, I installed the 3.0 kernel (aka 2.6.40). I've been running using this kernel for around 9 hours with no problems. So the good news is that this will likely be fixed by the time Oneiric comes out. The not-so-bad news is that right now, I'm running on release candidate 4 of that particular Linux kernel, so they've had 3 candidates to review. I suspect this kernel is close to stable. The not-so-good-news is that these test kernels are not meant for every day use. I won't be getting kernel updates unless I add the kernels PPA to my APT repos, which is a scary idea. If this is the only way to go forward, so be it. But I'm going to test an alternative, which is the latest drivers from Realtek.

Yep, five hours after I e-mailed them, Realtek came back with a zipped package containing the March 29, 2011 driver source. Ignore the date in the filename; I base my date on the Readme and the Release note, which, incidentally, mentions fixing the "crash issue caused by LPS ps-poll and null tx desc wrong," this last being a staple of my syslog when the wireless goes down.

I haven't tried it yet, but that's my next attempt: 2.6.38 with the 0620.2011/March 29, 2011 Realtek drivers. Hopefully that will take care of things until Oneiric releases. I've attached the drivers in case they are helpful to others having the same issue.

[hsoulen]

While waiting for Realtek to get back to me, I installed the 3.0 kernel (aka 2.6.40). I've been running using this kernel for around 9 hours with no problems. So the good news is that this will likely be fixed by the time Oneiric comes out. The not-so-bad news is that right now, I'm running on release candidate 4 of that particular Linux kernel, so they've had 3 candidates to review. I suspect this kernel is close to stable. The not-so-good-news is that these test kernels are not meant for every day use. I won't be getting kernel updates unless I add the kernels PPA to my APT repos, which is a scary idea. If this is the only way to go forward, so be it. But I'm going to test an alternative, which is the latest drivers from Realtek.

Yep, five hours after I e-mailed them, Realtek came back with a zipped package containing the March 29, 2011 driver source. Ignore the date in the filename; I base my date on the Readme and the Release note, which, incidentally, mentions fixing the "crash issue caused by LPS ps-poll and null tx desc wrong," this last being a staple of my syslog when the wireless goes down.

I haven't tried it yet, but that's my next attempt: 2.6.38 with the 0620.2011/March 29, 2011 Realtek drivers. Hopefully that will take care of things until Oneiric releases. I've attached the drivers in case they are helpful to others having the same issue.

I am also having the same issue with an RTL8192SE since moving to Natty. I also wrote to Realtek and they sent me the same driver... And then the real fun began!

The zip contains driver source for several cards so be very careful when using "make install" as it will add driver modules for cards you do not have present.

So... You can edit the makefiles to trim out all but your drivers, or just delete the other sub-dirs and deal with the compile warnings/errors.

But for me, after successfully compiling this driver and getting it installed I get kernel panics on every boot. I have to boot to recovery kernel and get into single user mode, then manually rm the driver to be able to boot again. Sent logs to RT but have not heard back for a solution.

So for me the problem persists. Realtek drivers have been working pretty stable for me through 10.10 but on upgrade to 11.04 neither the Ubuntu kernel driver nor the Realtek driver are in any way reliable.

A beer for anyone who can get this card stable!

Hank

[hackwater]

I am also having the same issue with an RTL8192SE since moving to Natty. I also wrote to Realtek and they sent me the same driver... And then the real fun began!

The zip contains driver source for several cards so be very careful when using "make install" as it will add driver modules for cards you do not have present.

So... You can edit the makefiles to trim out all but your drivers, or just delete the other sub-dirs and deal with the compile warnings/errors.

Thanks for the warning. I'm going to try editing the main makefile under the theory that if it can't call the sub-directory's makefiles from that one, I should be OK for compilation purposes.

But for me, after successfully compiling this driver and getting it installed I get kernel panics on every boot. I have to boot to recovery kernel and get into single user mode, then manually rm the driver to be able to boot again. Sent logs to RT but have not heard back for a solution.

I will hope that the June 20, 2011 date is somehow meaningful (i.e., they haven't updated their documentation in the Readme/Release notes), but if kernel panics ensue, I will continue using the 3.0 kernel and keep myself up-to-date as judiciously as possible, but sadly, manually.

So for me the problem persists. Realtek drivers have been working pretty stable for me through 10.10 but on upgrade to 11.04 neither the Ubuntu kernel driver nor the Realtek driver are in any way reliable.

A beer for anyone who can get this card stable!

Somebody must have; it definitely works for me and one other guy using the 3.0 kernel. See this bug. He installed from the daily builds; I went more conservative and installed from the RC4 mainline build. I also had to update my module init tools with the Oneiric deb. So 4 debs total: the two header debs, the kernel imagte, and the module init tools, essentially following the recipe on the MainlineBuilds wiki page.

I've only had a couple of minor issues with 3.0 so far: it seems slower than 2.6.38, and I couldn't get it to see my printer until I powered the printer down, unplugged the USB cable, and plugged it back in on a different USB port, powering it back on. But the wireless has been going strong, so there's a half-baked solution there (certainly beats kernel panics).

Will report back later (or perhaps sooner, if kernel panic ensues) the results of compiling the driver I posted.

[hsoulen]

Thanks "hackwater".

Looks like your driver was actually newer then the one I got a week or so ago, must have been some changes since I sent in my feedback.


Edited the makefile to only look for RTL8192se

sudo su
make clean (force of habit)
make
make install

Compile went fine (no errors), module installed and after an init 6 it came up and sure enough wireless was enabled.

But... Now the box freezes up after a few seconds to a minute, before I can even connect to a network.

dmesg has no reference to what happens when it freezes, just stops logging, as the only change was this driver I am back to the drawing board. Hacking the driver out now

Appreciate the effort, hopefully you will have better luck. I am on an Asus 1201n BTW.

Cheers,

Hank

[hackwater]

I booted back into the 2.6.38 kernel and put some load on my connection until it went down halfway through a 12 minute YouTube video. I then opened a terminal and

cd Downloads/realtek/kernel-module-source-directory
sudo su
rmmod r8192se_pci
make
make install
modprobe rtl8192se

Note the name change for the module. No kernel panics so far, and I've successfully watched various trailers on YouTube for execrable entertainment (the things I'll do for testing), as this sort of load has proven most successful in the past in bringing down my connection. Connection is still up. Gonna try a reboot and some more tests (aka YouTube vids).

[hsoulen]

Ok so decided not to give up on the new driver!

Hacked it out and went back to the drawing board.

Re-compiled, brought the driver up and did a bit of surfing. Rebooted (actually powered down) and it seems stable so far with no further freezes.

One thing I did do, I killed conky. I know this sounds strange but after watching when the freezing was happening I started to think it might be when conky started probing wlan0. Nothing empirical, simply a guess but so far so good.

Will report back should anything change, but for the moment it seems the driver that hackwater posted is working for me with kernel 2.6.38.

Thanks a bunch for the post, seems that RT took my kernel panics seriously!

Cheers,

Hank