2008-07-09 13:30:04

by Stefan Monnier

[permalink] [raw]
Subject: Slow b43 with 2.6.25.9 on an asus wl-700ge home-router

I finally started to use the b43 driver on my home-router and am happy
to have it working at all (with WPA), but I notice that the speed is
pretty terrible: I cannot seem to get more than 100KB/s out of it (I
use it as a personal web proxy, so it ends up slowing down my
web-surfing :-( ).

Originally "iwconfig" told me that the bit rate was 2Mb/s (and even
that should allow me to get a bit more than 100KB/s).
After "iwconfig wlan0 rate 54M", the result is the same (except that
"iwconfig" tells me the bit rate is 54Mb/s).

More specifically, iwconfig tells me things like:

wlan0 IEEE 802.11 ESSID:"test"
Mode:Managed Frequency:2.457 GHz Access Point: 00:10:30:C0:50:50
Bit Rate=54 Mb/s Tx-Power=27 dBm
Retry min limit:7 RTS thr:off Fragment thr=2352 B
Link Quality=95/100 Signal level:-54 dBm Noise level=-63 dBm
Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
Tx excessive retries:0 Invalid misc:0 Missed beacon:0

which seems to indicate that the reception is indeed good.
Any idea what I might want to check?

I did notice that my dmesg says things like
"received packet with own address as source address". This seems
related to my use of bridging, but:
1 - I have no bridging loop (actually I unplugged all ethernet cables,
so the only active network interface in the bridge is wlan0).
2 - taking wlan0 out of the bridge eliminates those messages, but
doesn't improve the bandwidth.

Any idea what might be going on?


Stefan



2008-07-09 15:27:26

by Larry Finger

[permalink] [raw]
Subject: Re: Slow b43 with 2.6.25.9 on an asus wl-700ge home-router

Stefan Monnier wrote:
> I finally started to use the b43 driver on my home-router and am happy
> to have it working at all (with WPA), but I notice that the speed is
> pretty terrible: I cannot seem to get more than 100KB/s out of it (I
> use it as a personal web proxy, so it ends up slowing down my
> web-surfing :-( ).
>
> Originally "iwconfig" told me that the bit rate was 2Mb/s (and even
> that should allow me to get a bit more than 100KB/s).
> After "iwconfig wlan0 rate 54M", the result is the same (except that
> "iwconfig" tells me the bit rate is 54Mb/s).
>
> More specifically, iwconfig tells me things like:
>
> wlan0 IEEE 802.11 ESSID:"test"
> Mode:Managed Frequency:2.457 GHz Access Point: 00:10:30:C0:50:50
> Bit Rate=54 Mb/s Tx-Power=27 dBm
> Retry min limit:7 RTS thr:off Fragment thr=2352 B
> Link Quality=95/100 Signal level:-54 dBm Noise level=-63 dBm
> Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
> Tx excessive retries:0 Invalid misc:0 Missed beacon:0
>
> which seems to indicate that the reception is indeed good.
> Any idea what I might want to check?
>
> I did notice that my dmesg says things like
> "received packet with own address as source address". This seems
> related to my use of bridging, but:
> 1 - I have no bridging loop (actually I unplugged all ethernet cables,
> so the only active network interface in the bridge is wlan0).
> 2 - taking wlan0 out of the bridge eliminates those messages, but
> doesn't improve the bandwidth.
>
> Any idea what might be going on?

What model is your card? Please post the output of 'dmesg | grep b43'.

With the current versions of mac80211, the rate-setting mechanism is very good.
If you force a high rate, i.e. with 'iwconfig ... 54M', you run the risk of
increasing your error rate causing many more retransmits, and might reduce the
throughput. That may not be happening here, but I wonder why the rate capped at
2M. Incidentally, the quality numbers for my BCM4312 located about 2m from the
AP are "Link Quality=93/100 Signal level:-37 dBm Noise level=-71 dBm". My Link
Quality is lower than yours, but S/N is much better - not that any of those
numbers have much validity. My transmit throughput is consistently over 2 Mbs,
but I'm not bridging. Roughly 6 months ago, I did set up routing, but that was
using iptables, not a formal bridge device. That configuration had full speed.

If you shut down the bridge, what speed do you get?

Larry


Larry

2008-07-09 15:59:24

by Michael Büsch

[permalink] [raw]
Subject: Re: Slow b43 with 2.6.25.9 on an asus wl-700ge home-router

On Wednesday 09 July 2008 17:43:49 Stefan Monnier wrote:
> >> Any idea what might be going on?
> > Yeah well. It's a bug somewhere.
>
> The scary part is that there's only Free Software involved, AFAICT.
> That *should* be bug-free, shouldn't it?

<inhuman mode>
Either my sense of irony currently is out of order, or this is the
most stupid question I did _ever_ see.
</inhuman mode>

> > I didn't have time to track it down, yet.
>
> Does that mean that you know about the problem and can reproduce it?

I can reproduce it, but I have no idea why it happens.

--
Greetings Michael.

2008-07-09 13:51:39

by Gábor Stefanik

[permalink] [raw]
Subject: Re: Slow b43 with 2.6.25.9 on an asus wl-700ge home-router

On Wed, Jul 9, 2008 at 7:50 AM, Stefan Monnier <[email protected]> wrote:
> I finally started to use the b43 driver on my home-router and am happy
> to have it working at all (with WPA), but I notice that the speed is
> pretty terrible: I cannot seem to get more than 100KB/s out of it (I
> use it as a personal web proxy, so it ends up slowing down my
> web-surfing :-( ).
>
> Originally "iwconfig" told me that the bit rate was 2Mb/s (and even
> that should allow me to get a bit more than 100KB/s).
> After "iwconfig wlan0 rate 54M", the result is the same (except that
> "iwconfig" tells me the bit rate is 54Mb/s).
>
> More specifically, iwconfig tells me things like:
>
> wlan0 IEEE 802.11 ESSID:"test"
> Mode:Managed Frequency:2.457 GHz Access Point: 00:10:30:C0:50:50
> Bit Rate=54 Mb/s Tx-Power=27 dBm
> Retry min limit:7 RTS thr:off Fragment thr=2352 B
> Link Quality=95/100 Signal level:-54 dBm Noise level=-63 dBm
> Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:0
> Tx excessive retries:0 Invalid misc:0 Missed beacon:0
>
> which seems to indicate that the reception is indeed good.
> Any idea what I might want to check?
>
> I did notice that my dmesg says things like
> "received packet with own address as source address". This seems
> related to my use of bridging, but:
> 1 - I have no bridging loop (actually I unplugged all ethernet cables,
> so the only active network interface in the bridge is wlan0).
> 2 - taking wlan0 out of the bridge eliminates those messages, but
> doesn't improve the bandwidth.
>
> Any idea what might be going on?
>
>
> Stefan
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>

Can you see "PHY transmission error" messages in dmesg? If you get
those, then that's a known bug.

--
Vista: [V]iruses, [I]ntruders, [S]pyware, [T]rojans and [A]dware. :-)

2008-07-09 13:42:35

by Michael Büsch

[permalink] [raw]
Subject: Re: Slow b43 with 2.6.25.9 on an asus wl-700ge home-router

On Wednesday 09 July 2008 07:50:11 Stefan Monnier wrote:
> Any idea what might be going on?

Yeah well. It's a bug somewhere. I didn't have time to track it down, yet.
I have no idea where it is and what kind of bug it is. Pretty time
consuming job. So if you can do a patch that fixes it, I will take it. ;)

--
Greetings Michael.

2008-07-09 15:44:03

by Stefan Monnier

[permalink] [raw]
Subject: Re: Slow b43 with 2.6.25.9 on an asus wl-700ge home-router

>> Any idea what might be going on?
> Yeah well. It's a bug somewhere.

The scary part is that there's only Free Software involved, AFAICT.
That *should* be bug-free, shouldn't it?

> I didn't have time to track it down, yet.

Does that mean that you know about the problem and can reproduce it?

> I have no idea where it is and what kind of bug it is. Pretty time
> consuming job. So if you can do a patch that fixes it, I will take it. ;)

Having too much time on my hand, and being naturally brilliant, I just
came up with the obvious fix, see attached patch.

More seriously: I am not familiar with any part of the kernel code, so
it's very unlikely that I'll find the time and motivation to delve into
it and come up with a patch. But if you could use some other kind of
help (e.g. some things to test, some printks to add to the code to
collect various kinds of data, ...), please holler,


Stefan