[Hardy][Intrepid] Low bandwidth with rt2400 / rt2500 drivers

I confirm this behaviour with a MSI CB54G2 adapter (rt2500pci/cardbus)

This behaviour is still in 2.6.24-7

My ralink rt2400 PCI card also works with wrong bandwidth (~52 kb/s instead of 132kb/s).

Related upstream discussion: http://rt2x00.serialmonkey.com/phpBB2/viewtopic.php?t=4579

I have the same problem on Hardy Alpha 3 and Alpha 4 (can't remember earlier). This doesn't happen in Feisty and Gutsy.

However, I installed 64-bit Debian Sid and there is no slowdown at all. I even get 54Mbps instead of typical 11Mbps (!) over WPA-protected access point.

I have built rt2500-source twice:
* 1:1.1.0-b4-3 for 2.6.21-2-amd64 kernel
* 1:1.1.0-b4+cvs20070924-3 for 2.6.24-1-amd64 kernel

Details follow:

# uname -smvro
Linux 2.6.24-1-amd64 #1 SMP Mon Feb 11 16:01:54 UTC 2008 x86_64 GNU/Linux

# COLUMNS=160 dpkg -l rt2500*
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Installed/Config-files/Unpacked/Failed-config/Half-installed
|/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err: uppercase=bad)
||/ Name Version Description
+++-==================================-==================================-====================================================================================
un rt2500 <none> (no description available)
un rt2500-2.6.18-4-amd64 <none> (no description available)
un rt2500-2.6.21-2-amd64 <none> (no description available)
un rt2500-2.6.24-1-amd64 <none> (no description available)
pn rt2500-modules-2.6.18-4-amd64 <none> (no description available)
ii rt2500-modules-2.6.21-2-amd64 1:1.1.0-b4-3+2.6.21-5 RT2500 wireless network drivers
ii rt2500-modules-2.6.24-1-amd64 1:1.1.0-b4+cvs20070924-3+2.6.24-4 rt2500 wireless network driver
ii rt2500-source 1:1.1.0-b4+cvs20070924-3 source for rt2500 wireless network driver

# iwconfig eth2
eth2 RT2500 Wireless ESSID:"my-essid"
          Mode:Managed Frequency=2.437 GHz Access Point: 00:11:95:7E:00:86
          Bit Rate=54 Mb/s Tx-Power:-3 dBm
          RTS thr:off Fragment thr:off
          Encryption key:3B5D-3C7D-207E-37DC-EEED-D301-E4 Security mode:open
          Link Quality=84/100 Signal level:-51 dBm Noise level:-80 dBm
          Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
          Tx excessive retries:0 Invalid misc:0 Missed beacon:0

You've got 54 Mbs directly because you are using a different driver.

rt2500-source = legacy driver (driver name: rt2500).

The driver shipped with hardy is part of the rt2x00 set, included in the kernel tree (driver name: rt2500pci)

See http://rt2x00.serialmonkey.com/wiki/index.php/Downloads

Revelant discussion here:
http://rt2x00.serialmonkey.com/phpBB2/viewtopic.php?p=28726#28726

Upstream fixed this in 2.6.25 but will not backport it.

So shall we close this bugreport as won'tfix?

It think Ubuntu developers should help with backporting drivers. Hardy is LTS and shouldn't be buggy...

Hi Guys,

Do you by chance have a pointer to the upstream patch that went into 2.6.25 to resolve this? I didn't see anything in the upstream thread and it would be helpful to the kernel team. Thanks.

When I asked him, he said he was not doing any backporting unless it
was a segfault.
(http://rt2x00.serialmonkey.com/phpBB2/viewtopic.php?t=4579&start=14).

On Mon, Mar 10, 2008 at 10:23 PM, Leann Ogasawara <email address hidden> wrote:
> Hi Guys,
>
> Do you by chance have a pointer to the upstream patch that went into
> 2.6.25 to resolve this? I didn't see anything in the upstream thread
> and it would be helpful to the kernel team. Thanks.
>
> ** Changed in: linux-ubuntu-modules-2.6.24 (Ubuntu)
> Status: Confirmed => Incomplete
>
> --
> [Hardy] Backport rt2400/rt2500 drivers
> https://bugs.launchpad.net/bugs/190515
> You received this bug notification because you are a direct subscriber
> of the bug.
>

@Nicolas Deschildre
> You've got 54 Mbs directly because you are using a different driver.
> rt2500-source = legacy driver (driver name: rt2500).

Since the Debianers used rt2500, I have tried the same - but without luck. The version from Ubuntu repository didn't want to compile with module-assistant (it died with error).

Your post gave me another idea, to use the rt2500 straight from CVS. And it worked!
I've used the "rt2500 1.1.0 CVS 2008030802" version, which compiled and worked fine on 2.6.24-11-generic kernel. It gave me full 54Mb/s :) Yes, I blacklisted manually rt2x00lib, rt2x00pci and rt2500pci but I think it was worth it :D

Today I've installed the upgrades and along with them came the 2.6.24-12-generic kernel. As expected, the wi-fi card worked no loger, since the relevant modules were left in the previous kernel. So, using a routine I used to in my Gentoo days, I reinstalled the driver after the kernel upgrade. It was only a matter of:

"make clean ; make ; make install ; modprobe rt2500"

And there it is again:
# iwconfig wlan0
wlan0 RT2500 Wireless ESSID:"przestrzen"
          Mode:Managed Frequency=2.437 GHz Access Point: 00:AA:BB:CC:DD:EE
          Bit Rate=54 Mb/s Tx-Power:-3 dBm
          RTS thr:off Fragment thr:off
          Encryption key:XXXX-XXXX-XXXX-XXXX-XXXX-XXXX-XX Security mode:open
          Link Quality=85/100 Signal level:-49 dBm Noise level:-79 dBm
          Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
          Tx excessive retries:0 Invalid misc:0 Missed beacon:0

I appreciate the serialmonkey work very much, but just as a quick remedy I'm reverting for now to the legacy rt2500 driver (that works :)

Friendly,
Wiktor

My rt2400 card works well with rt2400-source built with module-assistant (after blacklisting rt2400pci)

Could a switch to the legacy drivers be considered at this point for the rt2400/2500 hardware, knowing that the legacy drivers have worked great on previous releases, and that the 2.6.24 rt2x00 won't be fixed (cf previous posts)?

I'll redo some tests, but IIRC, the range of the wifi link was also diminished...

Hi everyone,
I have an Asus WL-107g wlan card (Ralink RT2500 chipset) that works great under Edubuntu 7.10 but with Hardy Heron (with the latest updates until now) the traffic is very slow (about 40Kb/s i.s.o 400Kb/s).
Thanks for the support, bye
Jef

I confirm this bug with my PC54G2. Tthe bit rate is set to 1Mb/s.

Confirmed...after upgrading from 7.10 to 8.04 it defaults to about half the signal level and sets the bit rate 1M.

Running 'sudo iwconfig wlan0 rate 54M' increases the tranfer rate but the signal level is still a lot lower then in 7.10.

$ iwconfig wlan0
wlan0 IEEE 802.11g ESSID:"freebsdap"
          Mode:Managed Frequency:2.447 GHz Access Point: 00:14:6C:72:86:52
          Bit Rate=54 Mb/s Tx-Power=27 dBm
          Retry min limit:7 RTS thr:off Fragment thr=2346 B
          Link Quality=45/100 Signal level=-60 dBm
          Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
          Tx excessive retries:0 Invalid misc:0 Missed beacon:0

$ sudo lspci -vvv
03:00.0 Network controller: RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)
 Subsystem: ASUSTeK Computer Inc. Unknown device 107f
 Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop- ParErr- Stepping- SERR- FastB2B-
 Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=slow >TAbort- <TAbort- <MAbort- >SERR- <PERR-
 Latency: 64, Cache Line Size: 32 bytes
 Interrupt: pin A routed to IRQ 11
 Region 0: Memory at 38000000 (32-bit, non-prefetchable) [size=8K]
 Capabilities: [40] Power Management version 2
  Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
  Status: D0 PME-Enable- DSel=0 DScale=0 PME-

Hi Nicolas,

Just wanted to let you know that one of the Ubuntu kernel team devs has a patchset to bring the rt2x00 code in Hardy uptodate with what upstream has. It is currently being discussed since we are in Beta Freeze for Hardy and almost at the kernel freeze. If it does get pulled in, hopefully it may help resolve this issue. Thanks.

Nice!
Let's hope the patchset will be considered stable enough for inclusion...

On Thu, Mar 27, 2008 at 12:46 AM, Leann Ogasawara <email address hidden> wrote:
> Hi Nicolas,
>
> Just wanted to let you know that one of the Ubuntu kernel team devs has
> a patchset to bring the rt2x00 code in Hardy uptodate with what upstream
> has. It is currently being discussed since we are in Beta Freeze for
> Hardy and almost at the kernel freeze. If it does get pulled in,
> hopefully it may help resolve this issue. Thanks.
>
> ** Changed in: linux-ubuntu-modules-2.6.24 (Ubuntu)
> Importance: Undecided => Medium
> Status: Incomplete => Triaged
>
>
>
> --
> [Hardy] Low bandwidth with rt2400 / rt2500 drivers
> https://bugs.launchpad.net/bugs/190515
> You received this bug notification because you are a direct subscriber
> of the bug.
>

I'm not sure why this is "Medium". Most computers these days are doorstops if they don't have reliable networking.

You have a LTS release due next month and you are persisting with releasing broken drivers for Ralink chipsets. There was a point about a month ago where someone needed to make a call and either revert to the old drivers or commit to getting the new ones working in time for Hardy.

I for one am using the old drivers.

Thought I'd just say that I ran into this problem after upgrading Gutsy to Hardy, i.e. slow connection, 50% signal strength, but that running the command 'sudo iwconfig wlan0 rate 54M' has fixed both aspects of the problem. So to me it seems the driver works but is just being configured wrongly. Maybe for the Hardy release all that needs to happen is for the rate to be set to 54M by default? My thinking might be too simplistic though.

P.S. I'm using kernel 2.6.24-14-generic with a rt2500pci card.

I've the same problem in Hardy Beta, Kernel 2.6.24-16 and rt2500pci card.
I've 10/10 Mbps Internet Optical Fiber connection, but speed indication on Ralink card is 1 Mb/s and signal is 50/70 %, very slow speed (50/100 kbps).

I'm having the same problem after upgrading from Gutsy to Hardy. Considering that this chipset is used in many "low end" wireless cards I believe this is a problem that should be addressed before Hardy's final release. In the current state many folks are going to be left with nearly useless wireless cards.

As Matt S. above suggests maybe this is as simple as changing a default setting?

2.6.24-16-generic
RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)

I've installed linux-backports-modules-2.6.24-16-generic and my card (rt2500pci based) now initialize at 54Mbps at boot time.
WEP 128 is working.

Great work !

Installed latest backports packages, but my card still initializes at 1Mbps. I can set the rate to 54Mbps manually and get decent speed/signal.

$ uname -a; lspci | grep RaL
Linux paragon 2.6.24-16-generic #1 SMP Thu Apr 10 13:23:42 UTC 2008 i686 GNU/Linux
00:0b.0 Network controller: RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)

Same problem here with rt2500pci. I've got a more reliable connection using Compat Wireless driver available here:
http://linuxwireless.sipsolutions.net/en/users/Download

Using the latest package do not solve the problem, downloading the old package (25 March 2008) the wireless card seem to work better. The older packages of Compat Wireless are available here:
http://www.orbit-lab.org/kernel/compat-wireless-2.6/

Hope this helps

I have the same problem here with a Hercules Wireless PCI G card.
I downloaded the drivers from rt2x00.serialmonkey.com and now everything works well.
The priority of this should be high as this is a serious problem and very easy to fix.

'sudo iwconfig wlan0 rate 54M' and all the problems (slow connection and 50% of signal strength) with rt2500pci are gone. Thanks!

uname -a; lspci | grep RaL
Linux Aragorn 2.6.24-16-386 #1 Thu Apr 10 12:50:06 UTC 2008 i686 GNU/Linux
03:00.0 Network controller: RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)

'sudo iwconfig wlan0 rate 54M' works for me too...how can i automate it!?!?
thanks

but the led of my PCMCIA card does blink no more!!!
WHY!!!

My network card worked fine on both 7.x-releases, but now in Hardy it does not.

This is how I solved it (and shouldn't have had to... FAR too much hassle):
0) Install build essentials...
1) Download the rt2500 serialmonkey drivers, compile and install. (wget, make, sudo make install)
2) Blacklist the old drivers (add rt2500pci, rt2x00pci and rt2x00lib to /etc/modprobe.d/blacklist or something)
3) Add 'alias ra0 rt2500' to /etc/modprobe.d/aliases.
4) Add 'ifconfig ra0 up' to /etc/rc.local
5) Add 'auto wlan0' and 'iface ra0 inet dhcp' to /etc/networking/interfaces
Optional: 6) Remove knetworkmanager, it doesn't work anymore anyway...

I mean... really? This all 'just worked' in the previous versions, and now it was like one and a half day of IRC/web search for help fixing it!
All this on an LTS!?

Hi, I'm another guy trying substitute his Windows XP by Ubuntu 8.04 (...and I'm about to give up).
The last problem I found so far is exactly defined in this bug:
MSI PC54G2 with rt2500 drivers and Internet extremely slow.

Question: Is adding the lines...
    ifconfig wlan0 up
    iwconfig wlan0 rate 54M
to /etc/rc.local a temporal solution (but 100% functional) till this bug is fixed?
I understood that the only consequence of this bug is that the adapter is initialized with a baudrate 1Mb. Are there any other additional consequeces?

I've tried this solution, after deactivating IPv6 (which is another source of speed problems), but internet is still quite slow (compared with the same system running Widnows). So I'm not sure if I should look for other problems (hardware? router? Firefox 3?) or try to compile the last drivers as described in the last post.

By the way, despite my complainments: good work.

'sudo iwconfig wlan0 rate 54M' helped me out with RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01) / Subsystem: Linksys WMP54G 2.0 PCI Adapter
Thanks a lot for the workaround.

Hi, I succeeded in installing the serialmonkey drivers and now I have full speed internet again, at least till the next update.
Because the process is not easy at all (at least for newbies like me), I have documented it in ubuntu forums:
http://ubuntuforums.org/showthread.php?p=4894088

Greetings

I agree with previous posters. This issue is absolutely not "medium". LTS = RaLink 2500 fixed, period. I don't claim to be an expert, but I went through this ugly wireless setup process three releases ago to get WPA2 working, and seeing it borked again just sucks. Fix this. Please.

Please try and see if installing linux-backports-modules-hardy solves your problems.

Have the same problem the only solution for me was to switch back to an older kernel. I'm waiting till this is resolved in the newer kernel.

Btw, it's still a problem in 2.6.24-17-generic. I'm using 2.6.24-17-generic for now

Question on Andreas Gnaus 2008-05-07 21:38:03 UTC contribution: in Synaptic, I see a couple of similar units. Do you mean the complete backport package for all modules that have had a backport, or only linux-backports-modules-hardy-rt ? Do you have success with it ?

It took me a day, many reboots, and only worse performance:
I installed linux-backports-modules-hardy (NOT the -386, as I inadvertently did, or the -rt, but the -generic, as automatically selected).
Speed still set to 1 Mb instead of 54Mb.
Dramatic worse performance after a connection was broken due to bad signal: I could not het the PC to work as a router, except after rebooting.
Installing linux-backports-modules-hardy STRONGLY DISCOURAGED.

I recently had to re-build my system (now: ASUS M2N-VM HDMI motherboard, EVGA NVidia e-GeForce 8800GT 512, AMD AM2 Athlon 64 4800+, Corsair 512MBx2 DDR2 PC5300 RAM, Sony DWG120A DVD+/-RW and - RaLink (Belkin) RT2500 PCI 802.11g (F5D7000) wireless network card) and installed Hardy 32bit. The new system is amazing - Hardy is wonderful - the best yet. I would love to use AMD64 but couldn't run Second Life.

However, like the others above i needed to use a speed fix. I have added a start up script that upgrades the speed to 54M. But, although this works, the driver is unstable and often crashes the wireless link. I find I have to switch off and re-boot to re-set. This is unacceptable as I am re-booting several times an hour. I cannot believe this new driver was added to an enterprise edition when the old one has been working faultlessly since Dapper came out.

Some short comments:

- This is definitely a regression. In Gutsy (even Feisty!) there was a reliable version of the driver. In Hardy there is a crappier version related to the code that was merged into the main linux tree. I quote from the front page rtx00 upstream web site:

rt2x00 enters mainline kernel - January 2008
Hi All,
two days ago, January 24th, Linux Kernel 2.6.24 was released. This is the first mainline kernel that includes (sadly a somewhat buggy) rt2x00 release.

- It is more than just the rate issue. The code seems to have other problems. I have very variable success with associating under the same setup that worked perfectly under the hardy rt2500pci module.

- I have linux-backports-modules-2.6.24-16-generic installed. doesn't help. Sometimes I get good performance, other times I can't even associate.

What can be done to fix this? It kills/blocks/subverts networking on this popular network hardware (rt2500) in Hardy.

my second dot point I should have said gutsy instead of hardy.

My point is gutsy had a better rt2500pci than hardy, all because of the chronological accident of hardy locking in on a kernel version that had a temporarily bad merge of the rt2x00 tree.

Wayne, what do you recommend to do? What exact bug has to be tracked? This very one states confirmed (this is good) and triaged (does not sound as good).
Do you recommend to install the backports package and wait TLS to honour the 'S' in its acronym? How in earth has this been possible to happen (a regression to a full year old version of the tree)?
Thank you.

I confirm this bug is *not* fixed with linux-backports-modules-2.6.24-16-generic (version 2.6.24-16.14). The rate is still 1M by default.

According to this relevant upstream discussion: http://rt2x00.serialmonkey.com/phpBB/viewtopic.php?f=5&t=4579&st=0&sk=t&sd=a&start=30
it seems the bug is still not fixed upstream.

On Sunday, the instability with my wireless card had increased to the level where it was requiring a re-boot every 30 secs or so. It was so unstable as to make the distribution unusable and I had decided to try and return to Dapper till the bug was fixed in Hardy.

However, there was a rare window of stability on Sunday night and I decided to update the machine. Twice it aborted because the wireless card got lost. On the 3rd attempt it updated. Since then, the wireless card has not crashed once. I do not know if the speed bug is still present because my system runs a startup script with the line: iwconfig wlan0 rate 54M

For now the system is working brilliantly so as far as I am concerned the problem is fixed. It would be nice to know however, that if I ever had to re-install the OS then the wireless card would work perfectly at the start. I still can not believe Hardy was rteleased with this bug: a rare black mark. But, I bet owners of Vista wish their bugs could be cleared so quickly - he he.

i've been playing around with hardy proposed and hardy backports modules.

i've noticed that there are two different versions of the module (at least).

using "modinfo rt2500pci" there seems to be a 2.0.10 version and a 2.1.5 version

the 2.0.10 version seems to work better for me.

I think 2.1.5 is the backports?

My advice: remove all backports, add hardy proposed as a repository in Synaptic, and get the latest kernel.

Compiling and installing an older module mitigate the problem in my situation.
Compiling the latest vanilla kernel doesn't solve the problem, the bug is still there in upstream.
The older module is available as Compat Wireless package version 2.0.14.
This package is downloadable in this page: http://alexit.wordpress.com/2008/01/10/rt2x00-2014-nuova-vita-ai-driver-wireless-ralink/
At the end of the post, before the comments, there is a link saying "compat-wireless-2.6-rt2×00-2.0.14.tar.bz2" that point to that old compat wireless package.
Download the package, extract it, type "make" and then "make unload" and "make load". In this way the older module will be loaded. For further instructions see here: http://linuxwireless.org/en/users/Download#Buildingandinstalling

This worked for me, hope this helps someone. Like i said before, the upstream bug is still present on vanilla kernel 2.6.26-rc4 and affect other distributions like fedora.

I just put the following file in /etc/network/if-up.d/ and made it executable to fix this problem.

I've tried just about everything suggested to fix this issue, short of using ndiswrapper with the Windows drivers - nothing helps to keep the network connected consistently at a reasonable speed.

Comment 49: Sebastian Urban
>I just put the following file in /etc/network/if-up.d/ and made it executable to fix this problem.

I tried this and it seemingly fixed a majority of the speed issue. The quality issue remains but I guess simply setting the rate works well.

This should be a recommended step to fixing the issue.

Using
 -> linux-backports-modules-2.6.24-19-generic
does not fix the 1M problem. Still low signal quality and 1M speed after reboot.

With Ubuntu kernel 2.6.24.18-generic, I was able to get my RT2500 to work full speed by using the rt2500-cvs-2008060319 driver. On a different machine using Ubuntu kernel 2.6.24.19-generic and the same RT2500 card, I had no choice, but to resort to using the winxp driver under ndiswrapper.

So far(about 1hr), ndiswrapper seems to be working great for the second machine. I'll give that machine at least a week, before I actually declare it as WORKING. My past experience with ndiswrapper has produced many failures on some machines after 1 week. I know, weird, but welcome to my life. The first machine with the cvs driver has been working fullspeed for a good 3 weeks now. But I'll be updating the kernel to latest in the repos tonight, so I expect I'll probably have some issues with it of course.

Man what a bummer, As far back as breezy, I never had to do any more than punch in the ssid for my RT2500. But then hardy came along. I'll try to keep you updated on my progress.

I can also confirm that linux-backports-modules-hardy does not work for me. (RaLink RT2500, UbuntuStudio 8.04)

I get about 3 mins of wireless at 800kbs, then it drops to 0.2kbs.

I was going to spend the last 3 days solid doing some tutorial screencasts for Inkscape, but instead I have spent 3 days testing various scripts and commands suggested here, reading help forums and (mostly) rebooting.

I'm another person who can't understand why Hardy has an update that takes us backwards. Just like sixgun, I've enjoyed problem-free RT2500 use since Breezy - until now.

I'll stick with Feisty ....and live without "compiz detail zooms" in my screencasts. :(

Can we at least agree that people having no network access on their laptops for approaching *8 MONTHS* now is of High importance, and mark this accordingly?

My laptop is a brick with Hardy on it.

I agree this issue is of high importance.

On my desktop I've changed to an atheros based card to work around it

It is not just ubuntus fault, the rt2x00 project is light on developers right now and it's unfortunate that as they have been merged into the mainline tree they have had so many problems.

I can confirm the problem. Please repair this quickly: it looks so bad in a LTS version. And many computers are affected, because this is a very common wireless chipset.

How is it possible that such a small problem takes so long to solve?
A lot of people are using wireless with this chipset.
Anyway thanks in advance.

Using driver rt2500usb for an internal wireless chip on my laptop.
lsusb output:
Bus 005 Device 002: ID 148f:2573 RaLink Technology, Corp.

Tested with kernel 2.6.24-19-generic (linux-image-generic, version 2.6.24-19.36) installed as an update to clean Hardy 8.04.1 install. Result: WiFi was detected but speed was set at 1MB/s as described in the bug.

Installed linux-backport-modules and rebooted. Result: WiFi connected at full speed 54MB/s

If you are experiencing problems with this chipset, try installing the linux-backport-modules-hardy package and reboot.

sudo apt-get install linux-backport-modules-hardy

Do the recent updates only fix the problem for USB chipsets?

I just tried backports-modules again with my minipci rt2500; it definitely doesn't fix the problem. Connection is at 1MB/s and file transfer speeds match this.

I confirm the same failure with the backports-modules as Shawn.

So far, the machine that I installed the CVS driver on has worked fine for the last month. The machine with the NDiswrapper driver, I decided to try installing the CVS driver again, and it did work, but not without a minor complication. I was receiving two errors like "no rule to make target" and "make: *** /lib/modules/2.6.24-19-generic/build: No such file or directory. Stop". Fortunately it just took an easy fix with a symlink to my header files. Here is a neutral copy and paste friendly version of the command I used.

ln -s /usr/src/linux-headers-`uname -r` /lib/modules/`uname -r`/build

I do plan to test the ndiswrapper driver again, because I didn't let it go a full week. I just wanted to make sure that I could duplicate my results with the CVS driver. So, I'll be getting back with you on that.

A simple solution that worked for me is here: http://ubuntuforums.org/showthread.php?t=836237

The "iwconfig wlan0 rate 54m" solution, with the minipci-based cards at least, appears to make iwconfig report the "proper" 54mb rate, but the actual data rate is still sub-1mb. stephspynx, you can confirm this by transferring a very large file between two computers on your local LAN.

Yes, on my internal network it's definitely increased from 30 Kb/s to an average 300 Kb/s between local machines.

This is an output from my card info:

02:00.0 Network controller: RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)
 Subsystem: Belkin F5D7010 Wireless G Notebook Network Card
 Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop- ParErr- Stepping- SERR- FastB2B-
 Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=slow >TAbort- <TAbort- <MAbort- >SERR- <PERR-
 Latency: 64, Cache Line Size: 64 bytes
 Interrupt: pin A routed to IRQ 5
 Region 0: Memory at 34000000 (32-bit, non-prefetchable) [size=8K]
 Capabilities: [40] Power Management version 2
  Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
  Status: D0 PME-Enable- DSel=0 DScale=0 PME-

The Ubuntu Kernel Team is planning to move to the 2.6.27 kernel for the upcoming Intrepid Ibex 8.10 release. As a result, the kernel team would appreciate it if you could please test this newer 2.6.27 Ubuntu kernel. There are one of two ways you should be able to test:

1) If you are comfortable installing packages on your own, the linux-image-2.6.27-* package is currently available for you to install and test.

--or--

2) The upcoming Alpha5 for Intrepid Ibex 8.10 will contain this newer 2.6.27 Ubuntu kernel. Alpha5 is set to be released Thursday Sept 4. Please watch http://www.ubuntu.com/testing for Alpha5 to be announced. You should then be able to test via a LiveCD.

Please let us know immediately if this newer 2.6.27 kernel resolves the bug reported here or if the issue remains. More importantly, please open a new bug report for each new bug/regression introduced by the 2.6.27 kernel and tag the bug report with 'linux-2.6.27'. Also, please specifically note if the issue does or does not appear in the 2.6.26 kernel. Thanks again, we really appreicate your help and feedback.

The issue is not fixed in 2.6.27 for my rt2500 : still the same symptoms. (I have not tested the 2.6.26)

It seems to depend on the wireless access point you are using, too.

pre-up iwconfig wlan0 rate 54M

worked for me when I used a T-Sinus 1054 DSL AP (shipped by T-Online in Germany), but with the "Arcor-DSL Wlan-Modem 200" built by Zyxel, the bandwidth was as low as ever.
However, thanks to Cedric Schieli's post I installed linux-backports-modules-hardy, and it works like a charm now. Thanks for backporting the module!

I still have the same problems with kernel 2.6.27.1 for my Asus WL-107g wlan card (Ralink RT2500 chipset).
I was not able to test it with 2.6.26 .

With kernel 2.6.26 the Asus WL-107g wlan card (Ralink RT2500 chipset) was also at low speed.
Sorry for my wrong answer in the previous post.

I have an Edimax EW7128g, using the ralink RT61 driver, which apparently should "work out of the box" on Ubuntu Hardy (http://www.linuxemporium.co.uk/products/wireless/). Anyway, it does, and iwconfig shows that it is correctly recognised as connecting via 54M 11g . However, download speed is approximately 1Mbps (compared to 10Mbps on Windows), and upload around 200Kbps (compared to 4Mbps). I also experience random freezes whilst downloading large files (see separate bug for other reports on similar freezes https://bugs.launchpad.net/ubuntu/+bug/228633). When I place the computer next to the wireless router, the signal strength is around 50%. Strangely, after moving to another room with 3 intervening walls, the signal strength increases to 60%, although the download speed decreases.

Downloading the backport modules as suggested above does not help. Interestingly, if I change the speed of the connection to 11M by typing "iwconfig wlan0 rate 11M", the download speed increases to around 4Mbps, and the upload speed to around 2Mbps when I am next to the router. Now, moving into the other room again, where the signal strength is around 60%, the download speed decreases to about 700Kbps, but the upload speed remains at 2Mbps. Windows achieves the same speeds (10Mbps and 4Mbps for download and upload respectively) wherever I place the computer.

So far, it seems that the freezing only occurs when the rate is set to 54M, and not when it is set to 11M.

I am running Hardy, with kernel 2.6.24-19.

I have not yet tried ndiswrapper on the windows driver, but thought my observations may possibly be of interest/use to someone.

To correct and clarify:

It is actually the Edimax EW 7318Ug that I am using, which is a USB wifi dongle, and I am running Hardy in 64-bit format.

Hi Niskitonf

I can confirm that with my Asus / AMD Athlon and Belkin card system -
changing from 54M to 11M seems to have cleared the bug - I haven't
crashed once since the mod.

Thanks for that Niskitonf - I have been without a stable system since I
upgraded to Hardy Heron back about Easter. What a relief and luxury not
to have to re-boot 7 times a night. Just wish Ubuntu spent more time
getting their priorities right. I very nearly gave up and installed a
different distro. Your solution came just as I was about to change for
good. We must be able to connect to the internet or we will always look
elsewhere - when will programmers learn this priority. Keep the customer
at all costs and worry about luxuries later is basic business sense. I
would have been so sad as Hardy seems to be easily my best distro yet.
It is so much better than Vista; but then, so is everything else
including XP - he he.

Ta Vici

niskitonf wrote:
> I have an Edimax EW7128g, using the ralink RT61 driver, which apparently
> should "work out of the box" on Ubuntu Hardy
> (http://www.linuxemporium.co.uk/products/wireless/). Anyway, it does,
> and iwconfig shows that it is correctly recognised as connecting via 54M
> 11g . However, download speed is approximately 1Mbps (compared to 10Mbps
> on Windows), and upload around 200Kbps (compared to 4Mbps). I also
> experience random freezes whilst downloading large files (see separate
> bug for other reports on similar freezes
> https://bugs.launchpad.net/ubuntu/+bug/228633). When I place the
> computer next to the wireless router, the signal strength is around 50%.
> Strangely, after moving to another room with 3 intervening walls, the
> signal strength increases to 60%, although the download speed decreases.
>
> Downloading the backport modules as suggested above does not help.
> Interestingly, if I change the speed of the connection to 11M by typing
> "iwconfig wlan0 rate 11M", the download speed increases to around 4Mbps,
> and the upload speed to around 2Mbps when I am next to the router. Now,
> moving into the other room again, where the signal strength is around
> 60%, the download speed decreases to about 700Kbps, but the upload speed
> remains at 2Mbps. Windows achieves the same speeds (10Mbps and 4Mbps for
> download and upload respectively) wherever I place the computer.
>
> So far, it seems that the freezing only occurs when the rate is set to
> 54M, and not when it is set to 11M.
>
> I am running Hardy, with kernel 2.6.24-19.
>
> I have not yet tried ndiswrapper on the windows driver, but thought my
> observations may possibly be of interest/use to someone.
>
>

Ha - I typed too soon

I just got lucky and had a rare moment of calm after slowing down to
11M. The driver now crashes my machine all the time. It is almost
unusable. What on earth are the developers doing allowing this to
continue when the old driver was perfect for years. Why should I or
anybody have to learn how to re-install the old one. I will certainly
have to remove ubuntu from my machine soon as I need a stable web link.
I don't feel confident in doing a back port. I am very un-geek. Such a
shame. A brilliant distro ruined by a silly driver problem.

Take note Ubuntu - people have old cards and use Linux because they
expect their old hardware to work.

Vici

P.S. My PC crashed and had to be re-booted twice whilst writing and
sending this message!!!!!!!!!!!!!

Vici wrote:
> Hi Niskitonf
>
> I can confirm that with my Asus / AMD Athlon and Belkin card system -
> changing from 54M to 11M seems to have cleared the bug - I haven't
> crashed once since the mod.
>
> Thanks for that Niskitonf - I have been without a stable system since I
> upgraded to Hardy Heron back about Easter. What a relief and luxury not
> to have to re-boot 7 times a night. Just wish Ubuntu spent more time
> getting their priorities right. I very nearly gave up and installed a
> different distro. Your solution came just as I was about to change for
> good. We must be able to connect to the internet or we will always look
> elsewhere - when will programmers learn this priority. Keep the customer
> at all costs and worry about luxuries later is basic business sense. I
> would have been so sad as Hardy seems to be easily my best distro yet.
> It is so much better than Vista; but then, so is everything else
> including XP - he he.
>
> Ta Vici
>
>
> niskitonf wrote:
>
>> I have an Edimax EW7128g, using the ralink RT61 driver, which apparently
>> should "work out of the box" on Ubuntu Hardy
>> (http://www.linuxemporium.co.uk/products/wireless/). Anyway, it does,
>> and iwconfig shows that it is correctly recognised as connecting via 54M
>> 11g . However, download speed is approximately 1Mbps (compared to 10Mbps
>> on Windows), and upload around 200Kbps (compared to 4Mbps). I also
>> experience random freezes whilst downloading large files (see separate
>> bug for other reports on similar freezes
>> https://bugs.launchpad.net/ubuntu/+bug/228633). When I place the
>> computer next to the wireless router, the signal strength is around 50%.
>> Strangely, after moving to another room with 3 intervening walls, the
>> signal strength increases to 60%, although the download speed decreases.
>>
>> Downloading the backport modules as suggested above does not help.
>> Interestingly, if I change the speed of the connection to 11M by typing
>> "iwconfig wlan0 rate 11M", the download speed increases to around 4Mbps,
>> and the upload speed to around 2Mbps when I am next to the router. Now,
>> moving into the other room again, where the signal strength is around
>> 60%, the download speed decreases to about 700Kbps, but the upload speed
>> remains at 2Mbps. Windows achieves the same speeds (10Mbps and 4Mbps for
>> download and upload respectively) wherever I place the com...

Are people still working on this bug? i noticed immediately when upgrading from Gutsy to Hardy that my connection signal dropped from ~97% to ~37%, and it's way too slow.
And yes, I'm using the rt2500 driver

Using the Intrepid LiveCD Alpha6, I can still reproduce this bug.

In my case, typing in a command line
sudo iwconfig wlan0 rate 54M
brought the network up to normal speed.

I tried setting the rate to 54M manually for my rt2500 in Hardy, but it continues to run slow. Why is it that the success of these solutions seems to be so hit and miss? How could they mess something up that seems to effect about every ralink and has no one solution to fix it?

Oh yeah, and as to update on my previous comment back on 2008-08-18, it seems that using the ndiswrapper driver is stable. I used ndisgtk to install it.

sixgun, where did you get the driver to use with ndiswrapper?

The driver I used was pulled from an installation cd that came with the wifi card. If you don't have a disk for your card, try downloading the driver at the following website. http://www.yournewdriver.com/Ralink_RT2500_Wireless_LAN_Card_7584.htm
It appears to be a generic type of driver that supports several different RT2500's.

I also get 1Mbit/s out of my rt2500 on an up to date Intrepid, and I can force it to 54M with iwconfig.

Binary package hint: linux-image-2.6.27-4-generic
02:09.0 Network controller: RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)
         Subsystem: Micro-Star International Co., Ltd. Device 6833
         Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
         Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=slow >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
         Latency: 64, Cache Line Size: 64 bytes
         Interrupt: pin A routed to IRQ 22
         Region 0: Memory at fbffc000 (32-bit, non-prefetchable) [size=8K]
         Capabilities: [40] Power Management version 2
                 Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
                 Status: D0 PME-Enable- DSel=0 DScale=0 PME-
         Kernel driver in use: rt2500pci
         Kernel modules: rt2500pci
This card, when used with:
% iwconfig wlan0 rate 54M auto
will never stay at 54M, but instead will immediately drop to 1M and stay there.
 Setting the rate using:
% iwconfig wlan0 rate 54M
locks the card at 54M, which is fine.
linux-image-2.6.27-4-generic 2.6.27-4.6

Just to update, with:

pre-up iwconfig wlan0 rate 11M

in my /etc/network/interfaces file, I still seem to get reasonable speeds, decent signal quality, and no crashing during downloads.

Hi All - Time is up

We have been searching for a cure for, what was, a perfectly good driver
for almost 6months with me on board. The solution discussed now is at
best a 'get us through' solution. The fix works fine when broadband is
behaving itself but, when it gets erratic at peak load times, the
wireless driver crashes and requires a full re-boot to clear (is there a
way of resetting the driver from the command line - the rest of the
machine seems OK). The other night, after I had re-booted 10 times in
half an hour, I decided that enough is enough. I will get rid of this
wireless board - great board but, useless without a driver.

Of course, I could install a dedicated driver (which I have) but am not
sure how to remove the other one first and then install with ndiswrapper
(any simple instructions anyone). Given that this is such a show
stopper for so many people, I am astounded that Ubuntu have not solved
this problem; especially as the old driver works. I believe it is a hang
over from the previous distribution so we approach at least a year and a
half still with the problem!!! A disgrace. Who cares about Ubuntu if we
cant use it.

Can anybody recommend a card that does work out of the box by the way,
that would be a great help. My Partner's Sony Viao laptop works
faultlessly from a straight install so I know my problem is my driver.

Regards Vici

P.S. I hope you all get your cards working soon and don't have to wait
another few years.

I believe about all you have to do to ensure the other drivers don't interfere is:
sudo gedit /etc/modprobe.d/blacklist

Then add the following blacklist lines to the end of the file and save:
blacklist rt2x00lib
blacklist rt2x00pci
blacklist rt2500pci
blacklist rt2500usb

Also, if you have installed the CVS, go to
System > Administration > Hardware Drivers
and make sure that the CVS driver isn't enable

To install ndiswrapper, open synaptic, and look up and install ndisgtk
After ndisgtk installed, go to System > Administration > Windows Wireless Drivers

Click the "Install New Driver" button.

Click the "location" box to open a dialog, and locate the proper .inf file.

The file you need, and the location may of course vary from mine, but in my case the file I needed was
"Rt2500.inf"
and was located in
"/media/cdrom/Software/WinXP/"

Select the Rt2500.inf file, click "Open"

Click "Install"

If all is well, it should say "Hardware present: Yes", and then you are free to configure the interface.

Be aware, if you have a pure ralink card, and your interface name was "ra0" or "ra1", it will now be "wlan0" or "wlan1".

And if you plan to use WPA, good luck with that. I've been struggling with getting my rt2500 to work and keep working with encryption on any driver. My only luck has been a combination of using the latest CVS driver, and Rutilt. Just make sure that you can access a network that is completely unsecured, unencrypted, and has a visible ssid, before attempting WPA.

Oh thanks for that Sixgun

That is a great help. I have downloaded the driver so will try this when
I get an hour free. Didn't crash last night at all but did many times
each night all week. Seems stupid that we have to go through this
rigmarole when there is a perfectly good driver available. I never ever
crashed before my upgrade to Hardy. Dapper worked faultlessly for years.

Thanks, with relief, Vici

sixgun wrote:
> I believe about all you have to do to ensure the other drivers don't interfere is:
> sudo gedit /etc/modprobe.d/blacklist
>
> Then add the following blacklist lines to the end of the file and save:
> blacklist rt2x00lib
> blacklist rt2x00pci
> blacklist rt2500pci
> blacklist rt2500usb
>
> Also, if you have installed the CVS, go to
> System > Administration > Hardware Drivers
> and make sure that the CVS driver isn't enable
>
> To install ndiswrapper, open synaptic, and look up and install ndisgtk
> After ndisgtk installed, go to System > Administration > Windows Wireless Drivers
>
> Click the "Install New Driver" button.
>
> Click the "location" box to open a dialog, and locate the proper .inf
> file.
>
> The file you need, and the location may of course vary from mine, but in my case the file I needed was
> "Rt2500.inf"
> and was located in
> "/media/cdrom/Software/WinXP/"
>
> Select the Rt2500.inf file, click "Open"
>
> Click "Install"
>
> If all is well, it should say "Hardware present: Yes", and then you are
> free to configure the interface.
>
> Be aware, if you have a pure ralink card, and your interface name was
> "ra0" or "ra1", it will now be "wlan0" or "wlan1".
>
> And if you plan to use WPA, good luck with that. I've been struggling
> with getting my rt2500 to work and keep working with encryption on any
> driver. My only luck has been a combination of using the latest CVS
> driver, and Rutilt. Just make sure that you can access a network that is
> completely unsecured, unencrypted, and has a visible ssid, before
> attempting WPA.
>
>

@Canonical:

How much money would we have to pay to get this bug fixed?
We all like Ubuntu, know that it is an open source project and development resources are expensive. A lot of people seem to be affected. So if everybody pays, say $ 20, will this be enough to hire a developer to fix this problem? Perhaps we can open a "shared" support request.

Hey Sebastian

I see where you are coming from but, the fact remains that there is 'NO
BUG TO FIX'. All that is required is that the old driver is re-instated
in the basic build - something that works does not need fixing - a heap
of junk that stops people using the complete system needs throwing away
not fixing. What pray was the point of a new driver anyway for an old
chip set - I am lost for an explanation. How much is spent on
developing Ubuntu - the real costs must be astronomical - yet all that
money is wasted if you can't use the system. Upgrades to Gnome, Open
Office, Firefox - these are all irrelevant if you cant log on. A basic
need is to link to the internet (hence web books) without that the whole
computer is a useless pile of junk. I have to say I am torn between
paying for a new wireless card and dumping Ubuntu all together.

I will try six packs recipe for loading the dedicated driver but, if
that fails it will be goodbye Ubuntu. What a pity, it is such a
brilliant OS. But my main requirement is the web and I can't do it
without constantly crashing.

I hope this will make it abundantly clear to Canonical why they misspend
their resources - it is the basic things that need to be right - not the
bells and whistles - hence, the enduring popularity of Debian. I know
how geeks are, they only want to work on the bells and whistles - but it
is basics that pay the bills. I would not pay $20 to fix this bug - I
would put the cash towards a new wireless card which would cost about
$20. If I wasn't retired, I would have paid for a new one already. In
fact, I think I will make the jump into Geekdom and install Debian
instead. How hard, will it be to suss out the driver installation and
have a usable system? One that always makes operability rather than
frills its basic requirement.

Regards Vici

Sebastian Urban wrote:
> @Canonical:
>
> How much money would we have to pay to get this bug fixed?
> We all like Ubuntu, know that it is an open source project and development resources are expensive. A lot of people seem to be affected. So if everybody pays, say $ 20, will this be enough to hire a developer to fix this problem? Perhaps we can open a "shared" support request.
>
>

I think that Canonical does an excellent job of balancing bug fixes and new features. Just look how buggy openSUSE and Fedora are. Ubuntu is in my opinion a very stable OS. We all have to keep in mind that we got Ubuntu for free and that fixing bugs does cost money. (even if the fix is as simple as using the old driver by default)

I for my part am willing to pay and I encourage other people to do the same. Open source software lives from people either fixing bugs thenselves or paying other people to do that for them. As I have no time for kernel hacking at the moment I am willing to pay.

@Vici
> I see where you are coming from but, the fact remains that there is 'NO
> BUG TO FIX'.

I just updated a hardy with rt2400pci to intrepid, and the network performance is even worse than before. if this is not a bug, what is it?

btw, now (8.10) only `sudo iwconfig wlan0 rate 11M` is accepted here, any higher rates gives:

Error for wireless request "Set Bit Rate" (8B20) :
    SET failed on device wlan0 ; Invalid argument.

in 7.10 it worked like a charm, since then... from bad to worse.

Let me see - where is the no bug - ah yes - the no bug is in the
original driver - so the fix for the bug is to remove the broken driver
and use the original - it has worked faultlessly for years.

My last contribution - I am installing Debian Lenny and will hand
install the correct driver - no bug fix required thanks. So long and
thanks for all the fish. I do not want to crash till Xmas. I have
enjoyed Ubuntu and promoted it everywhere but, I need to have control again.

Byeeeee Vici

mnemoc wrote:
> @Vici
>
>> I see where you are coming from but, the fact remains that there is 'NO
>> BUG TO FIX'.
>>
>
> I just updated a hardy with rt2400pci to intrepid, and the network
> performance is even worse than before. if this is not a bug, what is it?
>
>

For the record I'd like to add a "me too"

I have an rt2500, worked well in the past (old driver)

It was unusably slow in hardy but could be forced to full speed with:

sudo iwconfig wlan0 rate 54M

I have the same situation on intrepid.

Is there a bug filed for intrepit?

Confirm - things get even worser with Ubuntu 8.10. Using the rt2500pci driver and NetworkManager my WLAN-Router started to hang after some ammount of traffic (just using aptitude update was enougt). Had to reboot the router to be able to connect again.

Just after compiling the rt2500 from serialmonkey and loading this driver all problems disappeared. Router is stable for few days now (downloaded few Gigabytes for testing). This represents just a workaround (NOT a solution) as rt2500 driver is not compatible with Network-Manager.

I guess not every PCI card behaves like mine in conjunction with the rt2500pci driver, so developers can't reproduce this bug or do not have the time. This bug now exists for a year... I lost any hope to get this fixed. Currently I'm searching for a Intel-based wlan pci card as I had no problems so far with (any) IBM laptop that use intel hardware.

Anyway - for sake of completeness here are my results :

PC boots and rt2500pci driver is loaded. iwconfig shows a speed of just 1MBit for wlan0: NetworkManager connects normally the access point. (The following appeared with Intrepid: After some traffic my router hangs - no DNS queries possible. Current connections continue as IP is already in clients cache. Have to restart my router.)
Using iwconfig wlan0 rate xyM I can boost speed up noticably. iwconfig still shows 1M for speed. I have to reconnect using the NetworkManager to get iwconfig show the new speed I setup previosly.

Sourceforge bug added as a cross-reference:
https://sourceforge.net/tracker/index.php?func=detail&aid=2216782&group_id=107832&atid=648844

I got a working work-around, just using standard Ubuntu stuff. Basically, this is reinstalling the legacy drivers, with workaround for bug 183818, and on my system with a recent kernel: 2.6.24-21. As not everyboby might know all commands, I'll lay out what I did:

1) install rt2500-source, using "Synaptic Package Manager"
2) open a Terminal (under accessories), and type:
cd /usr/src/modules/rt2500
vi rtmp_main.c
3) in vi (or any other standard editor), change on line 254, the line
    SET_MODULE_OWNER(net_dev);
into:
#ifdef __BUGFIX__
 https://bugs.launchpad.net/ubuntu/+source/rt2500/+bug/183818/
    SET_MODULE_OWNER(net_dev);
#endif
4) type in the terminal
sudo make
  sudo mv /lib/modules/2.6.24-21-generic/kernel/drivers/net/wireless/rt2x00/ rt2x00_orig
  sudo mkdir /lib/modules/2.6.24-21-generic/kernel/drivers/net/wireless/rt2x00/
  sudo cp -p rt2500.ko /lib/modules/2.6.24-21-generic/kernel/drivers/net/wireless/rt2x00/
sudo depmod

5) at this point in time, I rebooted my system, and got normal rt2500 functionality

Note that I did NOT use the debian style calls as described in the file INSTALL.Debian
    sudo module-assistant prepare
    sudo module-assistant get rt2500
    sudo module-assistant build rt2500
Also note that installing a new kernel version will require that you do these actions again.

I guess this bug report should be renamed to "Low bandwidth with rt2400pci / rt2500pci drivers". I hear too often: "workaround, just compile rt2500 drivers from serialmonkey". But this is a completely different driver - not working with network-manager. This hint doesn't help to solve my problem. It just tells me to use different software with different functionality.

Again: rt2500 works, speed ok, router ok BUT no network-manager support.

rt2500pci doesn't work, low speed, (at least my) router unstable (beginning with intrepid) BUT with network-manager support.

Is there a better place to post this bug? I guess the rt2x00pci drivers are heavily related to linux kernel development.

Just to say I am NOT suffering this bug on 8.10 64bits, either with NetWorkManager or with my manual configuration and I use an rt2500pci card:
----------------------------------------------------
  *-network
       description: Wireless interface
       product: RT2500 802.11g Cardbus/mini-PCI
       vendor: RaLink
       physical id: 0
       bus info: pci@0000:05:00.0
       logical name: wmaster0
       version: 01
       serial: 00:08:d3:05:8c:a8
       width: 32 bits
       clock: 33MHz
       capabilities: pm bus_master cap_list logical ethernet physical wireless
       configuration: broadcast=yes driver=rt2500pci ip=192.168.0.10 latency=64 module=rt2500pci multicast=yes wireless=IEEE 802.11bg
----------------------------------------------------
auto wlan0
iface wlan0 inet dhcp
wpa-driver wext
wpa-conf /etc/wpa_supplicant.conf
-------------------------------------------------------
ap_scan=1
ctrl_interface=/var/run/wpa_supplicant
network={
        ssid="XXXXXX"
        scan_ssid=0
        proto=RSN WPA
        key_mgmt=WPA-PSK
        psk="xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
        pairwise=CCMP TKIP
        group=CCMP TKIP
}
-------------------------------------------
and that's all, no rate command :

wlan0 IEEE 802.11bg ESSID:"XXXXXX"
          Mode:Managed Frequency:2.437 GHz Access Point: 1E:C2:88:FB:2F:08
          Bit Rate=54 Mb/s Tx-Power=26 dBm
          Retry min limit:7 RTS thr:off Fragment thr=2352 B
          Power Management:off
          Link Quality=93/100 Signal level:-34 dBm
          Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
          Tx excessive retries:0 Invalid misc:0 Missed beacon:0

I have a RaLink RT2500 802.11g Cardbus/mini-PCI.

My solution for intrepid is to put in
/etc/rc.local this line: iwconfig wlan0 rate 54M

Now I have nothing to do it work perfectly.

Intrepid amd64

2.6.27-9-generic

RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)

Doesn't even associate until I run:

iwconfig wlan0 rate 54M

I just tried to install Ubuntu at my friend's computer. He has this wlan chip and of course this caused big problem. I found a howto on the internet to compile a driver which didn't work with network manager, but with rutilt (or similar). This worked well until a restart. Now I found the workaround here, but dunno how to remove the "new" driver again, and set the old driver from the blacklist again away. Could anyone please fix this for Intrepid? Thank you so much.

By the way:
According to https://wiki.ubuntu.com/Bugs/Importance
this bug should has importance "high", because of three reasons:
1. Has a *severe impact* on a small portion of Ubuntu users (estimated)
2. Makes a default Ubuntu installation *generally unusable* for some users

    * For example, if the system fails to boot, or X fails to start, on a certain make and model of computer

3. A problem with an essential hardware component (disk controller, *laptop built-in wireless*, video card, keyboard, mouse)

A system with this bug is generally unusable, a system update takes 6 hours and often the connection crashed, so even that cannot help. You cannot surf in the internet, the connection is slower that with a modem 10 years ago...

Just for information: We got it working again after 2 hours of work. The workaround described here works. Thanks.

This is turning into a genuine fiasco. I'm very aware that most work on Linux is on a volontary basis but upstream's arrogance and attitude, quite frankly, sucks. Look, if the driver is a rt2xxx "generic", "catch all", then the damn thing should work with rt2400 and rt2500 chipsets. If he doesn't want to back port (wait, isn't your crap supposed to work with 45 & 25? What's "back" about it?) then the driver name should be changed to rt 26* or rt28* or whatever since it IS NOT GENERIC TO ALL RT2xxx chipsets! Furthermore, as was stated above, sine it doesn't work and isn't generic to the family any more, then the legacy drivers should be reinstated.

So, exactly where is the issue and why have we gone 3 versions without support for a very common and popular wireless chipset? Three versions and hundreds of broken laptops later we're still discussing this? The driver is NOT generic to the whole chipset series. The driver IS broken. The author refuses to make his generic driver generic any more. Rename the damn thing, reinstate the legacy drivers and move on.

Sorry, but all this pussy footing around and breakage for hundreds of users over one asshole's arrogance and everybody's refusal to call him the elitist prick that he is or just ignore him is beyond frustrating.

As far as I know the legacy driver is useless because it does not work with NetworkManager.

Sebastian, driver compatibility with NetworkManager is kind of pointless if it means my wireless card doesn't work.

I'd rather have a working wireless card using either static configuration or something more elegant like Wicd.

Marty wrote:
> Sebastian, driver compatibility with NetworkManager is kind of pointless
> if it means my wireless card doesn't work.
>
> I'd rather have a working wireless card using either static
> configuration or something more elegant like Wicd.
>
>
For me the easy solution was buying another pcmcia card.

In my opinion the current state of this kernel module is SHAMEFUL.

Yes the caps were intentional.

I test from a nearly 6 year old laptop. This kernel module is useless in
every distro I've tried.

For those that have built in wireless that is rt25* use ndiswrapper,
windows or another card.

D-Link WNA-2330, $30 at my local officedespot works great.

Oh yeah use Linux it is great! everything just works, do not worry about
virii because you cannot connect :\

-mark

P.S. No, I'm not a smart a**

Per a decision made by the Ubuntu Kernel Team, bugs will longer be assigned to the ubuntu-kernel-team in Launchpad as part of the bug triage process. The ubuntu-kernel-team is being unassigned from this bug report. Refer to https://wiki.ubuntu.com/KernelTeamBugPolicies for more information. Thanks.

I am a bit confused, could someone confirm my thoughts, please?

- There are two GPL drivers for this wlan card:

- One driver is the in the linux kernel and is called rt2x00 [1]. This driver works, but is not compatible to network manager.
(The last tarball seems to be from 'Apr 30 2007' [2] , so this seems a bit dead, but there are still updates from time to time in the linux kernel (e.g. [3]), so it is not dead. Why does it not work with network manager? Is there a problem in the driver or in network manager? Is there already a bug report for this issue?

- The other driver is called rt2500pci. This driver is shipped with Ubuntu and is used by default for the rt2500 chips. This driver works with network manager, but is extremely slow. Is there a bug report against exactly this package?

I don't understand completely what's going on here, could someone please help me? Thanks!

[1] http://rt2x00.serialmonkey.com/wiki/index.php/Main_Page
[2] http://sourceforge.net/projects/rt2400/
[3] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=2bb057d07a0bc17475a7bf897fc41667ab08b73f

No, there is two GPL drivers for rt2500 chips. The two come from serialmonkey. One is the old "legacy" driver that is not compatible with Network Manager, the other one (rt2x00) is included in Ubuntu and works with NetworkManager, as explained in tour [1] link. It uses the new "Linux wireless extensions (generic)".

Ah, I see, thank you for the answer. Sometimes things are complicated, sorry.

In a vanilla linux kernel is there the new driver that works with network manager or the old legacy driver that only works with rutilt?

So the bug is actually that in the new driver the possible speed is not recognised correctly and so the driver is so very slow. Is this so hard to fix? For me it sounds as if this could be easyly hard coded, if it's not possible to fix the bug to find out the correct possible speed.
Thanks!

Don't know exactly as I do not suffer this bug with the default driver. After connection, I am at 1Mb/s rate, but each time I check thereafter, the rate slowly raise each time, to reach 54Mb/s after 2 minutes.

Its still a bug and a very bad one. Even setting manually to 54Mb will not always help. At this point it makes Linux almost unusable! I am almost at the point to look into the source myself :-)

Just do it! :)

*Your* help is needed and everyone will be very thankful.
The bug is known and confirmed by the devs, it is in the vanilla kernel but nobody seems to be working on it atm. Just look at the kernel bug tracker.

This is the link to the forum thread:
http://rt2x00.serialmonkey.com/phpBB/viewtopic.php?f=5&t=4579&p=31664#p31664

Thanks!

I have this problem on all three of my boxes. Rate drops to 11 and then to 1
1. An Asus EEEpc 701 (not sure of the chipset at the moment)
2. My compaq desktop with RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)
3. My HP DV9201ca (dv9000 series) Broadcom BCM4311 (rev 01)

I'm starting to think this problem isn't limited to a specific driver / hardware set.
Are there other reports that confirm this?

I tried the fixes to add the rate change at start-up and I believe it only changed the reporting, not the rate.

This is with ibex on all machines 2.6.27-7-generic, and the amd64 kernel on the HP.

Wifi is a must for market share. That the out-of-the-box everyone wants.
Hope this is fixed with the next update/release

The initial bit rate of 1M is still present in Jaunty Alpha4. I used the script at http://launchpadlibrarian.net/14908987/ralink-fix to have the bit rate set to 54M on startup. That should go in /etc/network/if-up.d. With this in place I get 73/100 link quality while two feet from my wireless router.

uname -r = 2.6.28-7-generic

mark@stingray:~$ modinfo rt2500pci
filename: /lib/modules/2.6.28-7-generic/kernel/drivers/net/wireless/rt2x00/rt2500pci.ko
license: GPL
description: Ralink RT2500 PCI & PCMCIA Wireless LAN driver.
version: 2.2.1
author: http://rt2x00.serialmonkey.com
srcversion: C0463573D1E4C8EE73ABD0E
alias: pci:v00001814d00000201sv*sd*bc*sc*i*
depends: rt2x00lib,rt2x00pci,eeprom_93cx6
vermagic: 2.6.28-7-generic SMP mod_unload modversions 586

mark@stingray:~$ lspci | grep Ra
05:00.0 Network controller: RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)

What a nasty bug! Can I pay someone to fix this? Or can't a Ubuntu guy look at this? Upstream knows about this bug, but does not do anything in order to get this fixed, it seems at least :(

On Monday 16 February 2009, goto wrote:
> What a nasty bug! Can I pay someone to fix this? Or can't a Ubuntu guy
> look at this? Upstream knows about this bug, but does not do anything in
> order to get this fixed, it seems at least :(

Well I have been shouting for ages that more help is needed on the rt2x00 drivers.

Since there is only 1 developer for all rt2x00 drivers, and that person can't spend
too much time on those drivers because of work etc., the total development hours/week
for the drivers is low.

so each bug needs to be prioritized before time is spend time on it. And bugs like
 "I can't associate to an AP using rt2500pci" and "Where are the rt2800pci/usb" drivers",
simply have higher priority then "I have to issue a 'iwconfig wlan0 rate 54M'"

Ivo

IvD: Sorry if I sounded rude, it wasn't meant so. I can't say how thankful I am, that you offer free drivers. But it's hard to be quiet, if one can't change anything.

This driver appears in the base kernel. Closing off the tasks for linux-restricted-modules and linux-update-modules.

Could all of those affected by this issue confirm whether they have linux-backport-modules-2.6.24 installed. If not could they test with that installed.

Specificially could they:

1) confirm the base speed selected after association
2) if the speed is 1M confirm whether the command below allows higher speeds:
            iwconfig wlan0 rate 54M

(You may need change the name of your ethernet device to match that on your system.)

2009/3/2 Andy Whitcroft <email address hidden>:
> Could all of those affected by this issue confirm whether they have
> linux-backport-modules-2.6.24 installed.  If not could they test with
> that installed.

I'm sorry to say, don't expect too many responses. All affected people
will have changed to another wireless card ages ago.

--
Francisco Vila. Badajoz (Spain)

The incredible carnival of Badajoz
http://www.flickr.com/photos/pacovila/tags/carnaval/show/

> If not could they test with that installed.
>
>Specificially could they:
>
>1) confirm the base speed selected after association
>2) if the speed is 1M confirm whether the command below allows higher speeds:
> iwconfig wlan0 rate 54M

1) base speed always 1M after association
2) Yes, but in my case within 2 minutes speed raise to 54 M (by steps, each time I look for speed)

Intrepid (2.6.27):
lsmod :
rt2500pci 26368 0
rt2x00pci 16256 1 rt2500pci
rt2x00lib 40576 2 rt2500pci,rt2x00pci

lspci :
05:00.0 Network controller: RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)

In Intrepid, the problem still there. However linux-backport-modules-2.6.24 cannot be selected there (only 2.6.27 ones)

> 1) confirm the base speed selected after association

I tested with all versions of Ubuntu including backports. ALWAYS speed was 1M after association and stayed at that level. Even worse, the link quality is very bad with this driver since this bug appeared.

> 2) if the speed is 1M confirm whether the command below allows higher speeds:
> iwconfig wlan0 rate 54M

This raises speed to the desired rate but link quality is very bad, so effective speed is not good compared to legacy rt2500 driver at same rate. I think this two things (1M after assoc and link quality) are related.

Andy Whitcroft wrote:
> Could all of those affected by this issue confirm whether they have
> linux-backport-modules-2.6.24 installed. If not could they test with
> that installed.
>

I do not have linux-backport-modules-2.6.24 installed. I will gladly
test if you send directions on how to regress from 2.26.27. Using
standard 2.26.27 module, not backport 2.26.27, the device is usable at
short range but not reliable at 30 feet from AP.

As Francisco noted it was easier for me to just get another wifi card.

Using 2.26.28-8 (current jaunty) the situation is similar: device
initiates at 1M, responds to iwconfig wlanx rate 54M, but is useless at
30 feet from an AP (access point).

I do not understand the object/goal of regression. Isn't the goal to
move forward not backwards?! Problem: the new driver/module doesn't work
well; solution: use the old driver!

When will the solution be making the new driver/module work well?

-mark

> Specificially could they:
>
> 1) confirm the base speed selected after association
> 2) if the speed is 1M confirm whether the command below allows higher speeds:
> iwconfig wlan0 rate 54M
>
> (You may need change the name of your ethernet device to match that on
> your system.)
>

The problem is still there. I've also noticed something else. The link quality is rather low at 1M (41/100) and increases to 57/100 when I set the rate to 54M. My guess is that the drivers sees a low link quality and therefore selects 1M. I'v looked for some technical documentation of the driver but there seems to be nothing there! I do not have the time to reverse engineer the driver, but I suspect it should not be hard to fix.

For those on Jaunty, there is an updated linux-backport-modules as of this morning which includes updated drivers for the rt wireless drivers. It would be helpful if those of you on Jaunty could test that and report back here.

Do you think it is fixed there? I am not so sure, because it's afaik not even fixed upstream. Thanks though :)

Andy Whitcroft wrote:
> For those on Jaunty, there is an updated linux-backport-modules as of
> this morning which includes updated drivers for the rt wireless drivers.
> It would be helpful if those of you on Jaunty could test that and report
> back here.
>
Very, even VERY bad; network manager broke, neither a rt2500 or an
artheros card would work, perhaps they would have with a command line. I
purged them.

-mark

@markthecarp -- appologies there there was a bad backports module uploaded which did not work at all. It should be resolved now if you are able to retest. Thanks in advance.

Andy Whitcroft wrote:
> @markthecarp -- appologies there there was a bad backports module
> uploaded which did not work at all. It should be resolved now if you
> are able to retest. Thanks in advance.
>

Andy these are *outstanding!

I first tried them around 11:30 UTC; initial rate 36-54M at close range
(4 feet), at 30' the rates seem to top out at 48M.

I used speakeasy.net/speedtest and regardless of the bit rate am getting
speeds 6700 kbps which is less than 100kbps lower than I get directly
connected to the modem.

However with one intervening brick wall performance drops dramatically
(1M bit rate/700 kbps).

uname -r
2.6.28-9-generic

modinfo rt2500pci
filename: /lib/modules/2.6.28-9-generic/updates/rt2500pci.ko
license: GPL
description: Ralink RT2500 PCI & PCMCIA Wireless LAN driver.
version: 2.3.0
author: http://rt2x00.serialmonkey.com
srcversion: 070F51D7EDEA76F54A51843
alias: pci:v00001814d00000201sv*sd*bc*sc*i*
depends: rt2x00lib,rt2x00pci,eeprom_93cx6
vermagic: 2.6.28-9-generic SMP mod_unload modversions 586

-mark

@mark -- excellent that sounds like a major improvement.

@all -- could those of you on Intrepid and Jaunty please try out the linux-backport-modules and report back here.

Andy: Is it enough to test a Jaunty live CD? Without installation and without adding any modules?

@goto -- you would need to install l-b-m and then reboot which would not work on a CD. However if you install the image onto a USB stick using usb-creator then that has persistance and you can reboot that to test.

To summarise the status/package updates. This problem is related to the quality of the rt driver in the base kernel. This is not something we propose to fix in the kernel version as the changes to the latest are too great, therefore closing the kernel task Won't Fix. However, the driver is already backported for Intrepid and Jaunty in the linux-backports-modules-2.6.27/28 packages for those releases. Therefore closing FIx Commited for Intrepid and Jaunty. For Hardy the kernel changes are too great for a backport to be feasable therefore closing that task Won't Fix.

So this bug is fixed in Ubuntu but you don't push the fix back upstream?

On Tue, Mar 24, 2009 at 02:16:26PM -0000, goto wrote:
> So this bug is fixed in Ubuntu but you don't push the fix back upstream?

This is fixed much later upstream and has been backported to this older
Ubuntu release.

Oh, nice. Could you point me to the commit in the upstream project? So one could close this [1] now, as it should be fixed in linux 2.6.29?

Thanks!

[1] http://bugzilla.kernel.org/show_bug.cgi?id=9273

Hello,

tested the "fixed" driver using ubuntu-proposed repository and this packages:

linux-image-2.6.27-14-generic
linux-backports-modules-2.6.27-14-generic

I can confirm that the driver now initializes fine at a much higher rate (in my case at about 48Mbit/s), while link quality ist displayed at a very high level (by network-manager).

BUT, running a web-based dsl-speed-test only the first run (after initialization of my wireless connection my network-manager) gives me a good result (for me about 6MBit/s which maches exaclty what should be) and any further run of the test shows different speeds, but all of them are much lower, simetimes as low as 1Mbit/s.

I repeated this test switching multiple times between this driver and the legacy rt2500 driver and everytime get the same results. Somehow the speed is very unstble for me.

BTW. The legacy driver passed the dsl-speed-test constantly at about 5.8Mbit/s at each run.

I can confirm that using linux-backports-modules-2.6.28-11-generic on Jaunty fixes the problem for me.

Is there any reason why the fixed driver is in the backports package and not installed by default?

Not for me, I still have the problem eventhough I installed linux-backports-modules-jaunty.

This has not been fixed yet. I just performed a clean install of jaunty and suffered a godforsaken slow connection (like I did back in the Intrepid days). I had to do sudo iwconfig wlan0 rate 54M and sudo iwconfig wlan0 rate fixed to get a decent speed again.

Reopening.

Reverting. Still present in jaunty

Sorry, I'm a moron who needs to read the comments

this issue is still present in Karmic
~$ uname -a
Linux neclaptop 2.6.31-13-generic #42-Ubuntu SMP Thu Oct 8 20:04:50 UTC 2009 i686 GNU/Linux

~$ modinfo rt2500pci
filename: /lib/modules/2.6.31-13-generic/kernel/drivers/net/wireless/rt2x00/rt2500pci.ko
license: GPL
description: Ralink RT2500 PCI & PCMCIA Wireless LAN driver.
version: 2.3.0
author: http://rt2x00.serialmonkey.com
srcversion: 017B67DDC35630260307788
alias: pci:v00001814d00000201sv*sd*bc*sc*i*
depends: rt2x00lib,rt2x00pci,eeprom_93cx6
vermagic: 2.6.31-13-generic SMP mod_unload modversions 586

I confirm this bug with Karmic. To set the bit rate manually doesn't work anymore.
The bit rate seems to have been changed but the connection is still very slow and the connectivity is poor.

~$ iwconfig
lo no wireless extensions.

wmaster0 no wireless extensions.

wlan0 IEEE 802.11bg ESSID:"Reseau_maison"
          Mode:Managed Frequency:2.457 GHz Access Point: 00:07:CB:53:DD:50
          Bit Rate=54 Mb/s Tx-Power=20 dBm
          Retry long limit:7 RTS thr:off Fragment thr:off
          Power Management:on
          Link Quality=33/70 Signal level=-77 dBm
          Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
          Tx excessive retries:0 Invalid misc:0 Missed beacon:0

~$ uname -a
Linux bellerophon 2.6.31-14-generic #48-Ubuntu SMP Fri Oct 16 14:04:26 UTC 2009 i686 GNU/Linux

~$ modinfo rt2500pci
filename: /lib/modules/2.6.31-14-generic/kernel/drivers/net/wireless/rt2x00/rt2500pci.ko
license: GPL
description: Ralink RT2500 PCI & PCMCIA Wireless LAN driver.
version: 2.3.0
author: http://rt2x00.serialmonkey.com
srcversion: 017B67DDC35630260307788
alias: pci:v00001814d00000201sv*sd*bc*sc*i*
depends: rt2x00lib,rt2x00pci,eeprom_93cx6
vermagic: 2.6.31-14-generic SMP mod_unload modversions 586

@ Stéphane B.:
You could try if a different bitrate helps: 11M instead of 54M. Does this help?

No, it doesn't do anything.

I can confirm this bug in 9.10 also. Is this bug active and being worked on? Should be a high priority because no one is maintaining the legacy serialmonkey rt2500 drivers any longer.

Hopefully this will help:
http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6-stable.git;a=commit;h=4789666e13fb0b2d45feb1b4a5119a1b997ec84c

On Sun, Nov 22, 2009 at 12:38, Cybjit <email address hidden> wrote:
> Hopefully this will help:
> http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6-stable.git;a=commit;h=4789666e13fb0b2d45feb1b4a5119a1b997ec84c

lovely magic numbers

It's been committed a few months ago but still has not made into 9.10

One last plea for someone to fix this and get it into a ubuntu 9.10 repo before I bite the bullet and buy a replacement wifi card. I think it is really poor how support for this card has effectively been dropped as soon as it makes it's way into the linux kernel tree.

the only walkaround I currently know is to use ndiswrapper with 2.6.32 kernel (there is another bug in Ubuntu, which prevents to use ndiswrapper with the current kernel). You can find instructions for where to get the new kernel here: http://ubuntuforums.org/showpost.php?p=8694409&postcount=35 To use ndiswrapper you will have to blacklist the native rt2x00 driver.

@foxy123: Are you using Karmic? My rt2500 device, Asus WL-167G (usb), runs perfectly out of the box with the default driver (included in the kernel) in Karmic and it's configured with network-manager.

I had been using 2.3.31-17 so tried upgrading to 2.6.32 but still can't get more than 1Mb/s through my 10Mb/s Internet connection.

@papukaija - I can believe the rt2500pci kernel module is providing adequate support for some devices - but not certainly not for my RaLink PCI card.

Details of my configuration below. Also more than happy to provide more information if required.

paul@locke:~$ uname -r
2.6.32-020632-generic

paul@locke:~$ iwconfig wlan0
wlan0 IEEE 802.11bg ESSID:"174A"
          Mode:Managed Frequency:2.442 GHz Access Point: 00:14:7C:B6:14:C8
          Bit Rate=54 Mb/s Tx-Power=20 dBm
          Retry long limit:7 RTS thr:off Fragment thr:off
          Power Management:off *
          Link Quality=62/70 Signal level=-48 dBm
          Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
          Tx excessive retries:0 Invalid misc:0 Missed beacon:0

* Note: I have had to disable power management due to this bug: http://www.gossamer-threads.com/lists/linux/kernel/1129333
Saying that, regardless of whether I enable or disable it I still get the poor throughput.

lspci -v

03:02.0 Network controller: RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)
 Subsystem: Belkin Device 700a
 Flags: bus master, slow devsel, latency 64, IRQ 18
 Memory at ecefa000 (32-bit, non-prefetchable) [size=8K]
 Capabilities: [40] Power Management version 2
 Kernel driver in use: rt2500pci
 Kernel modules: rt2500pci

Just upgraded Ubuntu from 8.04 ( I used rt2500 driver in it, compiled it myself) to 10.04 and see very slow (less then 1 Mbit) wifi connection with rt2500pci.
The same problem with 2.6.34 from ppa.
Unfortunately, I can't compile rt2500 with 2.6.32 and newer kernels, so I need a solution.
I tried to change speed with iwconfig, it helps, but only for several seconds.
And see many frame errors on wi-fi interface of wi-fi router, which also run linux :-)
Is there any way to get back my wi-fi connection? :-)

@ sles:
Try if this helps:

sudo iwconfig wlan0 rate 11M

@ sles:

Whenever I notice my rt2500pci's (linksys wmp54g) connection is slow, I right-click the NetworkManager applet on the Ubuntu desktop, deselect "Enable Wireless", wait 2 seconds, then enable it again. This usually works for a couple of hours. Sometimes it's only a few minutes, sometimes more than a day. Maybe I'll write a script for this one day.

Good luck.

OK, this appears to be an effective workaround: disable power management for the wireless chipset. As follows:

Applications - Accessories - Terminal
type (use copy/paste):
gksudo gedit /etc/rc.local

Press Enter.

Add this line, just above exit 0 (use copy/paste):
iwconfig wlan0 power off

So it should become something like this (example):

#!/bin/sh -e
#
# rc.local
#
# This script is executed at the end of each multiuser runlevel.
# Make sure that the script will "exit 0" on success or any other
# value on error.
#
# In order to enable or disable this script just change the execution
# bits.
#
# By default this script does nothing.
iwconfig wlan0 power off
exit 0

Save, close and reboot. The problem should be fixed permanently.

Please report your findings.

The power management issue is fixed in maverick as it has the 2.6.34 kernel.

Hm, seems my RT2560 based card is now also affected by this issue, although I actually didn't notice until now.. (well, recently, I exchanged my modem router, but that should not be an issue - a netbook with an Intel wireless device runs at full 54M bandwidth all the time, while my RT2x00 card slows down at the same time).

The card's power management is off by default on my system (by rc.local), yet it seems the card slows down from time to time to rates around 1400mbit/s. At first I thought this had something to do with pptp being enabled over the wireless network, but the problem persists when I disconnect from VPN. After shutting down the wireless connection via NetworkManager, and bringing it up again, bandwidth returns back to normal.

iwconfig permanently reports 54M, signal strength is 80% most of the time. Is there a safe way to undoubtedly tell the bandwidth reduction I am experiencing here is actually the same issue as reported here?

like with the stock kernel in lucid, the bug is also persistent with the modules in linux-backports-modules-wireless-2.6.32-24-generic

Hi, every body, I find a workaround for his untimely disconnection to operate with rt2500 lucid in ad-hoc mode. Because the downside is to stop the update has some major security on file.
go here:

http://forum.ubuntu-fr.org/viewtopic.php?id=419044

Assistance will be welcome.
PS: sorry for my English Littel

see you soon!

I am still experiencing this bug in karmic.

Today I tried maverick (running from a USB stick), unfortunately the problem is still present. First the connection is fine (1.0 MB/s in my case), later the speed drops to 64 kB/s - 100 kB/s. (I'm testing the speed with 'scp'.) Then I tried:
# iwconfig wlan0 power off
# iwconfig wlan0 rate 54M fixed

Then iwconfig will claim the speed is 54M, but actually it's much slower. As I wrote earlier, my workaround is right-click the NetworkManager applet on the Ubuntu desktop, deselect "Enable Wireless", wait 2 seconds, then enable it again. Then the speed is good again for 1 minute to 1 day. (varies)

ubuntu@ubuntu:~$ uname -a
Linux ubuntu 2.6.35-22-generic #33-Ubuntu SMP Sun Sep 19 20:34:50 UTC 2010 i686 GNU/Linux

ubuntu@ubuntu:~$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 10.10
Release: 10.10
Codename: maverick

ubuntu@ubuntu:~$ modinfo rt2500pci
filename: /lib/modules/2.6.35-22-generic/kernel/drivers/net/wireless/rt2x00/rt2500pci.ko
license: GPL
description: Ralink RT2500 PCI & PCMCIA Wireless LAN driver.
version: 2.3.0
author: http://rt2x00.serialmonkey.com
srcversion: A92C9A44EDA76B8F1D30911
alias: pci:v00001814d00000201sv*sd*bc*sc*i*
depends: rt2x00pci,rt2x00lib,eeprom_93cx6
vermagic: 2.6.35-22-generic SMP mod_unload modversions 686

Hello,

The rt2500 chipset doesn't have any publicly datasheet available, so its development/bugs fixing is hard.

I have asked more information on this really well known problem on the rt2x00's forum, and also where the datasheets are (to try to fix the bug by myself), here is the answer from Ivo van Doorn, administrator of the rt2x00 project:

"The specsheets are not publicly available. Only those for rt61 and rt73 have been released publicly by Ralink on their website, all older chipsets have not had this privilege and thus remain closed (although they are available to rt2x00 team members)."

So only the t2x00 team members are able to fix this bug, because they are the only ones to have access to the datasheet.

Since this is an older chipset, I can understand that they prefer working on more recent chipset.

Forgot the link to the answer:
http://rt2x00.serialmonkey.com/phpBB/viewtopic.php?p=34505#p34505

There seems to be an infinite amount of trust in the so-called
specsheets from Ralink.
If you would take a look at the publicly available sheets for rt61 and
rt73, you would
notice that there isn't that much useful in there to resolve
throughput issues as
experienced on rt2500pci/usb rt61pci/rt73usb (yes the problem is not
only rt2500pci).

But it is correct that the main focus currently is on the rt2800pci/usb drivers.
We still want to maintain the older drivers, but we simply lack the
time and developers
for working on all drivers at the same time.

On Tue, Nov 2, 2010 at 8:40 PM, Loic Pefferkorn
<email address hidden> wrote:
> Forgot the link to the answer:
> http://rt2x00.serialmonkey.com/phpBB/viewtopic.php?p=34505#p34505
>
> --
> [Hardy][Intrepid] Low bandwidth with rt2400 / rt2500 drivers
> https://bugs.launchpad.net/bugs/190515
> You received this bug notification because you are a direct subscriber
> of the bug.
>

Status changed to 'Confirmed' because the bug affects multiple users.

This bug is still present in Precise.

Using an A-Link PCI card
("Ralink corp. RT2500 Wireless 802.11bg (rev 01)" according to lspci),
I get about 1.4 Mbit/s regardless of what iwconfig shows, and
so many dropped packets it's practically unusable.

Disabling power management as per comment #161 helped somewhat:
I get higher speeds, up to 10 Mbit/s immediately after boot, but soon
it drops back to 1.4 Mbit/s range. It doesn't, however, drop packets
anymore, so it's usable although sluggish.

(Speeds measured with netperf.)

This bug is indeed no longer present in Quantal
(in my machine anyway). :-)

A followup to my comment #172 above: things are not perfect with quantal either.
The card behaves differently, starting up with good speed consostently, but
it's not stable, eventually losing connection totally sooner or later,
and too often sooner rather than later.

I have updated my old laptop Easy Note some days ago .... to Ubuntu 14.04
and the ralink card (rt2500pci) cease to work
it connenct and disconnet continuosly, and internet wasn't working(no ping, no navigation)
 I try all the solutions without ANY modification .....

at last, I try the workaround of comments #161
and....

ALL goes fine!!!!!!

MANY MANY MANY TAHNKS

(sorry 4 my bad english)

outdated flavor, report about a newer active version if needed