Slow connection with rt2500

oponek · **Joined:** Thu Feb 21, 2008 2:08 pm **Posts:** 1

I have the same problem with rt2400 card. Hope it will be fixed in Ubuntu Hardy...

IvD · **Posted:** Mon Feb 25, 2008 4:23 pm

The fix is in linux 2.6.25.
And no, I am not backporting any fixes to older kernels (unless it fixes segmentation faults / null pointers faults or other crashes).

nand · **Joined:** Thu Feb 14, 2008 5:18 pm **Posts:** 10

Ok thanks for the info!

ngc2997 · **Joined:** Fri Jan 11, 2008 2:31 pm **Posts:** 5

Hm does that mean that unless I am using a legacy driver (not working with Gnome's Network Manager), or compile a newer kernel, this won't be fixed in Hardy? If so, that is a slightly unpleasant perspective for the next six months, meaning I will have to go 'back' to a wired connection.. *sigh*

Best wishes,
Karsten

thnker · **Joined:** Sat Apr 05, 2008 3:03 pm **Posts:** 6

I have this problem too - 1 Mb/s bit rate. I'm using kernel version 2.6.24.4-69 on a mac G4 (ppc) with a linksys wireless g pci card. Being a newbie to linux, I'm not sure I understand the current situation with the rt2x00 driver. From this thread, I see that there is a thing called rt2x00.git that contains a fixed version of the rt2x00 driver but also, the rt2x00.git contains an entire kernel.

a) Is the kernel in the rt2x00.git a fedora core 8 kernel or is it a beta version of fedora core 9?

b) Is there any way to get the fixed version of rt2x00 driver without installing a beta version of a kernel?

c) Would the old rt2500 driver work better for me than the best version of rt2x00 that I'm able to get with fedora core 8 and if so, how do I switch from using the rt2x00 driver to the rt2500 driver?

d) What is a git?

Sorry for so many questions. I've spent days getting to where I am now on this problem and still feel lost. Thanks in advance for any help you can offer.

IvD · **Posted:** Wed Apr 09, 2008 9:53 pm

thnker wrote:

a) Is the kernel in the rt2x00.git a fedora core 8 kernel or is it a beta version of fedora core 9?

rt2x00.git is not fedora.
rt2x00.git is a branch of the Linux Kernel, the Fedora kernel is a branch of the Linux Kernel. The Fedora kernel contains many patches from rt2x00.git which are not yet in the main Linux kernel.

thnker wrote:

b) Is there any way to get the fixed version of rt2x00 driver without installing a beta version of a kernel?

Either use the latest Fedora kernel, or go to linuxwireless.org and download the wireless-compat package. But that version contains the same rt2x00 version that is currently found in the latest Fedora kernel...
Note that with latest Fedora kernel I mean the development kernel that is not yet available through Yum(?) but is available for download on the Fedora website.

thnker wrote:

c) Would the old rt2500 driver work better for me than the best version of rt2x00 that I'm able to get with fedora core 8 and if so, how do I switch from using the rt2x00 driver to the rt2500 driver?

If you don't intend to use NetworkManager and want to write all networking configuration manually, then yes you can use the rt2500 package. Simply read the INSTALL and README files provided in that package. And Google is your friend when you want complete howtos for Fedora and rt2500.

thnker wrote:

d) What is a git?

http://en.wikipedia.org/wiki/Git_(software)

thnker · **Joined:** Sat Apr 05, 2008 3:03 pm **Posts:** 6

I just installed 2.6.25 and all of my problems relating to the rt2x00 wireless driver are fixed except that I'm getting a low throughput rate. When using a wired connection on this machine, I can download the latest fedora 8 core at about 600 kB/sec (almost full ISP connection bitrate). However, with the wireless connection (rt2x00) the same download occurs at about 30 kB/sec.

Is there a setting somewhere that is misconfigured so as to cause the low bitrate?

ifconfig reports:

wlan0 Link encap:Ethernet HWaddr 00:0F:66:ED:70:A7
inet addr:192.168.2.100 Bcast:192.168.2.255 Mask:255.255.255.0
inet6 addr: fe80::20f:66ff:feed:70a7/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:2399 errors:0 dropped:0 overruns:0 frame:0
TX packets:2518 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:2786385 (2.6 MiB) TX bytes:278028 (271.5 KiB)

iwconfig reports:

wlan0 IEEE 802.11 ESSID:"linksys"
Mode:Managed Frequency:2.437 GHz Access Point: 00:12:17:09:55:4C
Bit Rate=54 Mb/s Tx-Power=23 dBm
Retry min limit:7 RTS thr:off Fragment thr=2352 B
Encryption key:<hidden>
Link Quality=62/100 Signal level=-21 dBm
Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
Tx excessive retries:0 Invalid misc:0 Missed beacon:0

Rikz · **Posted:** Fri Apr 18, 2008 7:57 am

I have the same problem with rt2500 driver from 2.6.25 kernel in case you are collecting some statistics.

Hazzl · **Joined:** Thu May 17, 2007 9:58 am **Posts:** 41

I have the same problems (as already described in this thread. When the link is first established, iwconfig shows a rate of 1M, however, the speed doesn't improve even after manually setting the rate to 54M

With the latest rt2x00 git kernel (2.6.25-rc9-wl-30069-g026c62a) I get some interesting looking warnings in the kernel log

Quote:

Apr 20 22:08:15 eressea kernel: [ 3174.692609] phy0 -> rt2x00lib_rxdone: Warning - Frame received with unrecognized signal,signal=0x09, plcp=0.
Apr 20 22:20:31 eressea kernel: [ 3891.556084] phy0 -> rt2x00lib_rxdone: Warning - Frame received with unrecognized signal,signal=0x0c, plcp=0.

Can this help to trace down the causes for the missing performance?

IvD · **Posted:** Sun Apr 20, 2008 9:19 pm

Unfortunately no, that bug is a hardware bug which causes a frame to be marked with an incorrect RX rate. rt2x00 compensates this by setting the RX rate to something mac80211 will accept (if rt2x00 wouldn't do that, the frame would be dropped by mac80211).

osuwop · **Joined:** Mon Apr 21, 2008 9:25 pm **Posts:** 1

IvD wrote:

Unfortunately no, that bug is a hardware bug which causes a frame to be marked with an incorrect RX rate. rt2x00 compensates this by setting the RX rate to something mac80211 will accept (if rt2x00 wouldn't do that, the frame would be dropped by mac80211).

so are you saying that this speed issue is a bug that cannot be corrected due to it being a hardware issue? also are people confirming that this speed issue is NOT corrected in .25?

IvD · **Posted:** Mon Apr 21, 2008 9:37 pm

No I am saying that the error message reported by Hazzl is caused by a hardware error AND that this error message has _nothing_ to do with the rate issue.

I have heard mixed reports about the 2.6.25 kernel, in some rc releases people reported high speeds, but with the follow-up rc release the rate was slow again. Changes in rt2x00 between those rc releases were minimal and didn't affect the TX code.
Based on what I heard from other drivers, is that _perhaps_ the issue is caused by the rate selection module provided by mac80211 in combination with bad link quality reporting by the driver.
In addition another possible cause had to do with short preamble mode.
Both causes are under investigation.

mikkoc · **Joined:** Thu Jan 31, 2008 5:01 pm **Posts:** 3

I can confirm that going from 2.6.24 to .25 did NOT fix any speed/rate issue for me.
I still get the 1M rate, even after manually setting it to 54M.
The download speed is around 30-40 KB/s, but not very stable, it tends to drop very often, web pages always timeout.

Result: back to old rt2500, which works like a charm

lspci --v:

Code:

00:10.0 Network controller: RaLink RT2500 802.11g Cardbus/mini-PCI (rev 01)
        Subsystem: Micro-Star International Co., Ltd. Device 6834          
        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: 32, Cache Line Size: 32 bytes                                                               
        Interrupt: pin A routed to IRQ 12                                                                    
        Region 0: Memory at da100000 (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

Toon · **Joined:** Sat Apr 26, 2008 9:28 am **Posts:** 1

IvD wrote:

I have heard mixed reports about the 2.6.25 kernel, in some rc releases people reported high speeds, but with the follow-up rc release the rate was slow again. Changes in rt2x00 between those rc releases were minimal and didn't affect the TX code.
Based on what I heard from other drivers, is that _perhaps_ the issue is caused by the rate selection module provided by mac80211 in combination with bad link quality reporting by the driver.
In addition another possible cause had to do with short preamble mode.
Both causes are under investigation.

Yeah, sounds plausible.
I'm currently running Gentoo Linux with kernel 2.6.25-r1.
I'm using the in-kernel rt2500pci driver.
The link speed is (automatically) set to 1Mb.
Interesting observation: in the NetworkManager icon the signal strength indication is fluctuating wildly.
I didn't have problems with an earlier CVS-version of the rt2500pci driver on 2.6.23.

Regards,
Toon.

madwolfa · **Joined:** Sun May 18, 2008 5:23 am **Posts:** 3

Confirmed, the same problem with 2.6.25-4 under Ubuntu Hardy 8.04 (rt2500pci 2.0.14).. I'm now building the 2.6.26-rc2 to check, if it's still there..

This problem is dreading me for a few months already (since the release of 2.6.24)... I can't imagine why it wasn't fixed yet with so many complaints.

rt2x00.serialmonkey.com

Slow connection with rt2500

Who is online