2.6 much worse than 2.4 on poor wlan reception
Hi,
I have big problems with kernel 2.6 and WLAN. Quite often the connection interrupts completely, I can't transfer anything for minutes - making 2.6 unusable for me :-(
I'm only about 5 meters away from my AP, but unfortunately there's a celeiling between me and the AP to the reception is poor.
My hardware is a SMC 2632 PCMCIA card (802.11b) in a IBM Thinkpad A30 and a SMC AP.
lspci:
02:00.0 CardBus bridge: Ricoh Co Ltd RL5c476 II (rev 80) 02:00.1 CardBus bridge: Ricoh Co Ltd RL5c476 II (rev 80)
blade:~# cardctl ident Socket 0: no product info available Socket 1: product info: "SMC", "SMC2632W", "Version 01.02", "" manfid: 0x0156, 0x0002 function: 6 (network)
The wireless-tools have version 26+27pre10-3, pcmcia-cs has 3.2.5, both from Debian Sid. I have tried kernels 2.6.0, .1 and .2. Every kernel 2.4 I've had running until now worked very well.
-----------------------------------------------------------
2.4.20:
blade:~# iwconfig eth1
eth1 IEEE 802.11-DS ESSID:"WLAN" Nickname:"Prism I"
Mode:Managed Frequency:2.412GHz Access Point:00:60:B3:17:F8:8C
Bit Rate:11Mb/s Tx-Power=15 dBm Sensitivity:1/3
Retry min limit:8 RTS thr:off Fragment thr:off
Encryption key:[ secret ] Security mode:open
Power Management:off
Link Quality:1/92 Signal level:-101 dBm Noise level:-149 dBm
Rx invalid nwid:0 Rx invalid crypt:661 Rx invalid frag:0
Tx excessive retries:2751 Invalid misc:0 Missed beacon:0lsmod:
orinoco_cs 4776 1 orinoco 32068 0 [orinoco_cs] hermes 6244 0 [orinoco_cs orinoco] ds 7060 2 [orinoco_cs] yenta_socket 10080 2 pcmcia_core 44928 0 [orinoco_cs ds yenta_socket]
tmm@blade:~$ netio -u 10.0.0.15 NETIO - Network Throughput Benchmark, Version 1.21 (C) 1997-2003 Kai Uwe Rommel
UDP connection established.
Packet size 1k bytes: 557 KByte/s (0%) Tx, 261 KByte/s (73%) Rx. Packet size 2k bytes: 553 KByte/s (0%) Tx, 89 KByte/s (90%) Rx. Packet size 4k bytes: 626 KByte/s (0%) Tx, 61444 Byte/s (92%) Rx. Packet size 8k bytes: 517 KByte/s (2%) Tx, 13393 Byte/s (98%) Rx. Packet size 16k bytes: 565 KByte/s (3%) Tx, 10924 Byte/s (98%) Rx. Packet size 32k bytes: 565 KByte/s (3%) Tx, 13787 Byte/s (98%) Rx.
Done.
-----------------------------------------------------------
2.6.2:
blade:~# iwconfig eth1
eth1 IEEE 802.11-DS ESSID:"WLAN" Nickname:"Prism I"
Mode:Managed Frequency:2.412GHz Access Point:00:60:B3:17:F8:8C
Bit Rate:11Mb/s Tx-Power=15 dBm Sensitivity:1/3
Retry min limit:8 RTS thr:off Fragment thr:off
Encryption key:[ secret ] Security mode:open
Power Management:off
Link Quality:1/92 Signal level:-101 dBm Noise level:-149 dBm
Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
Tx excessive retries:0 Invalid misc:0 Missed beacon:0orinoco_cs 9192 1 orinoco 43980 1 orinoco_cs hermes 8512 2 orinoco_cs,orinoco ds 15940 5 orinoco_cs i82365 20876 1 pcmcia_core 71456 3 orinoco_cs,ds,i82365
tmm@blade:~$ netio -u 10.0.0.15
NETIO - Network Throughput Benchmark, Version 1.21
(C) 1997-2003 Kai Uwe Rommel
UDP connection established.
Packet size 1k bytes: 289 KByte/s (99%) Tx, 223 KByte/s (75%) Rx. Packet size 2k bytes: 0 Byte/s (100%) Tx, 21 KByte/s (91%) Rx. Packet size 4k bytes: 440 KByte/s (99%) Tx, 61333 Byte/s (94%) Rx. Packet size 8k bytes: 382 KByte/s (96%) Tx, 14438 Byte/s (98%) Rx. Packet size 16k bytes: 369 KByte/s (96%) Tx, 2365 Byte/s (98%) Rx. Packet size 32k bytes: 239 KByte/s (98%) Tx, 3005 Byte/s (98%) Rx.
Done.
There was a break when netio transfered the 2k blocks.
My log is full of entries like this one:
Mar 1 17:54:12 blade kernel: eth1: New link status: AP Out of Range (0004) Mar 1 17:54:12 blade kernel: eth1: New link status: AP In Range (0005) Mar 1 17:54:16 blade kernel: eth1: New link status: AP Out of Range (0004) Mar 1 17:54:16 blade kernel: eth1: New link status: AP In Range (0005) Mar 1 17:54:19 blade kernel: eth1: New link status: AP Out of Range (0004) Mar 1 17:54:20 blade kernel: eth1: New link status: AP In Range (0005) Mar 1 17:54:22 blade kernel: eth1: New link status: AP Out of Range
(0004)
Kernel 2.4 works far better in the poor reception situation I have, anyone any idea what I could do without moving the AP or laptop? When I'm near my AP everything works fine with 2.6 too.
BTW: removing the PCMCIA card when it's in use freezes my system completely, that was no problem with 2.4.
Thanks a lot!
Signature · 1 line
Thomas Mueller - http://www.tmueller.com for pgp key (95702B3B)
Quoted 4 lines · Thomas Mueller
I have big problems with kernel 2.6 and WLAN. Quite often the connection interrupts completely, I can't transfer anything for minutes - making 2.6 unusable for me :-(
Quoted 8 lines
blade:~# iwconfig eth1
eth1 IEEE 802.11-DS ESSID:"WLAN" Nickname:"Prism I"
Mode:Managed Frequency:2.412GHz Access Point:00:60:B3:17:F8:8C
Bit Rate:11Mb/s Tx-Power=15 dBm Sensitivity:1/3
Retry min limit:8 RTS thr:off Fragment thr:off
Encryption key:[ secret ] Security mode:open
Power Management:off
Link Quality:1/92 Signal level:-101 dBm Noise level:-149 dBmI have Prism 2.5 cards. I run them with hostap driver. Link quality of 1/92 is very bad. You are on the edge of losing connection. (At least this is the case for my hardware).
Let's see how much errors do you have. Do this:
# cat /proc/net/wireless /proc/net/dev
On my system wlan1 has "Link Quality:66/92" (wlan0 is an AP, has no meaningful Link Quality)
Inter-| sta-| Quality | Discarded packets | Missed | WE face | tus | link level noise | nwid crypt frag retry misc | beacon | 16 wlan0: 0000 0 0 0 0 9178 13 23727 142280 0 wlan1: 0000 66. 195. 156. 0 0 0 3422 309008 0
Inter-| Receive | Transmit face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed
wlan0:1829928225 2539552 0 80689 0 0 0 0 1966341385 2579089 0 42 0 0 0 0
wlan1:143242367 907236 0 0 0 0 0 0 11481487 74357 0 0 0 0 0 0
So, wlan1: 907236 rx packets, 3422 retries
(and 309008 packets wasn't for me, wrong MAC, I suppose ;) ) What's your numbers?
Quoted 13 lines
There was a break when netio transfered the 2k blocks.
My log is full of entries like this one:
Mar 1 17:54:12 blade kernel: eth1: New link status: AP Out of Range (0004) Mar 1 17:54:12 blade kernel: eth1: New link status: AP In Range (0005) Mar 1 17:54:16 blade kernel: eth1: New link status: AP Out of Range (0004) Mar 1 17:54:16 blade kernel: eth1: New link status: AP In Range (0005) Mar 1 17:54:19 blade kernel: eth1: New link status: AP Out of Range (0004) Mar 1 17:54:20 blade kernel: eth1: New link status: AP In Range (0005) Mar 1 17:54:22 blade kernel: eth1: New link status: AP Out of Range
Quoted 5 lines
(0004)
Kernel 2.4 works far better in the poor reception situation I have, anyone any idea what I could do without moving the AP or laptop? When I'm near my AP everything works fine with 2.6 too.
Is your orinoco driver is the same for 2.4 and 2.6? Maybe 2.6 one has a bit lower max retry count or some such?
Quoted 2 lines
BTW: removing the PCMCIA card when it's in use freezes my system completely, that was no problem with 2.4.
No oops? No SysRq?
--
vda
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
"Thomas" == Thomas Mueller <[email protected]> writes:
Thomas> Kernel 2.4 works far better in the poor reception situation I Thomas> have, anyone any idea what I could do without moving the AP or Thomas> laptop? When I'm near my AP everything works fine with 2.6 Thomas> too.
Start out by forcing it to a lower link speed, at that signal quality you really don't want to try and go above 2MBit/sec. If you keep trying to do 11MBit/sec the card will constantly try the higher rate and then lose signal, drop down and try again. Fixing the rate should improve the situation - at least it has always done so for me ;-)
Cheers,
Jes
Hi Jes,
Jes Sorensen meinte am Friday, dem 05. March 2004:
Quoted 10 lines
Thomas> Kernel 2.4 works far better in the poor reception situation I Thomas> have, anyone any idea what I could do without moving the AP or Thomas> laptop? When I'm near my AP everything works fine with 2.6 Thomas> too.
Start out by forcing it to a lower link speed, at that signal quality you really don't want to try and go above 2MBit/sec. If you keep trying to do 11MBit/sec the card will constantly try the higher rate and then lose signal, drop down and try again. Fixing the rate should improve the situation - at least it has always done so for me ;-)
I tried that (rate 2M auto) but that doesn't improve anything. I still loose the connection to the AP, I don't think that happens less frequent than at 11 MBit/s.
Signature · 1 line
MfG Thomas Mueller - http://www.tmueller.com for pgp key (95702B3B)
Hi Denis,
Quoted 9 lines
Quoted 8 lines
blade:~# iwconfig eth1 eth1 IEEE 802.11-DS ESSID:"WLAN" Nickname:"Prism I" Mode:Managed Frequency:2.412GHz Access Point:00:60:B3:17:F8:8C Bit Rate:11Mb/s Tx-Power=15 dBm Sensitivity:1/3 Retry min limit:8 RTS thr:off Fragment thr:off Encryption key:[ secret ] Security mode:open Power Management:off Link Quality:1/92 Signal level:-101 dBm Noise level:-149 dBm
Quoted 4 lines
I have Prism 2.5 cards. I run them with hostap driver. Link quality of 1/92 is very bad. You are on the edge of losing connection. (At least this is the case for my hardware).
Yes I am. When I move some meters in the room I loose connection with kernel 2.4 too.
Quoted 3 lines
Let's see how much errors do you have. Do this:
# cat /proc/net/wireless /proc/net/dev
[..]
tmm@blade:~$ cat /proc/net/wireless /proc/net/dev
Inter-| sta-| Quality | Discarded packets |
Missed | WE
face | tus | link level noise | nwid crypt frag retry misc |
beacon | 16
eth1: 0000 0. 150. 107. 0 8 0 0 0
0
Inter-| Receive |
Transmit
face |bytes packets errs drop fifo frame compressed multicast|bytes
packets errs drop fifo colls carrier compressed
lo: 58298 881 0 0 0 0 0 0
58298 881 0 0 0 0 0 0
eth0: 0 0 0 0 0 0 0 0
3456 14 0 0 0 0 0 0
eth1: 532101 1336 0 0 0 0 0 0
223614 1299 466 0 0 0 0 0
sit0: 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0That's really interesting, thanks for that hint!
Transmit: 1299 packets, 466 errs - argh.
When I can't transmit anything 'errs' increases by one every few seconds.
As comparison: kernel 2.4.20 has 1743 packets and 9 errs at the moment. So the interesting question is: why is the error rate with kernel 2.6 that high?
Quoted 15 lines
Quoted 15 lines
There was a break when netio transfered the 2k blocks.
My log is full of entries like this one:
Mar 1 17:54:12 blade kernel: eth1: New link status: AP Out of Range (0004) Mar 1 17:54:12 blade kernel: eth1: New link status: AP In Range (0005) Mar 1 17:54:16 blade kernel: eth1: New link status: AP Out of Range (0004) Mar 1 17:54:16 blade kernel: eth1: New link status: AP In Range (0005) Mar 1 17:54:19 blade kernel: eth1: New link status: AP Out of Range (0004) Mar 1 17:54:20 blade kernel: eth1: New link status: AP In Range (0005) Mar 1 17:54:22 blade kernel: eth1: New link status: AP Out of Range(0004)
Quoted 6 lines
Quoted 3 lines
Kernel 2.4 works far better in the poor reception situation I have, anyone any idea what I could do without moving the AP or laptop? When I'm near my AP everything works fine with 2.6 too.
Is your orinoco driver is the same for 2.4 and 2.6? Maybe 2.6 one has a bit lower max retry count or some such?
2.6.2 has version 0.13e, 2.4.23 has 0.13d. I diffed the orinoco.* but there are only small changes.
Quoted 4 lines
Quoted 2 lines
BTW: removing the PCMCIA card when it's in use freezes my system completely, that was no problem with 2.4.
No oops? No SysRq?
Nope, it just freezes :-(
Signature · 1 line
MfG Thomas Mueller - http://www.tmueller.com for pgp key (95702B3B)