public inbox for b43-dev@lists.infradead.org
 help / color / mirror / Atom feed
* Packet loss on bcm4331
@ 2012-08-14 19:07 George Wright
  2012-08-14 19:29 ` Larry Finger
  0 siblings, 1 reply; 3+ messages in thread
From: George Wright @ 2012-08-14 19:07 UTC (permalink / raw)
  To: b43-dev

I have a MacBook with a bcm4331 chip which mostly works in Linux (Fedora
17 w/kernel 3.5.1), but I'm noticing horrible packet loss and throughput
a lot of the time.

The setup is basically a small apartment with an 802.11bgn router
sitting in the living room; I've had the laptop sit directly on top of
the router and run ping to another computer wired directly into the
access point and experienced 75% packet loss. File transfers are
experiencing throughput in the region of 200kB/s to 800kB/s. Oddly
enough, the amount of packet loss is significantly reduced if I'm
connected in my bedroom (~20 feet from the router and with a couple of
walls in between). I've also experienced the same problems with my
backup wireless router (a WRT54GL running dd-wrt).

By comparison, on OS X I get 0% packet loss and throughput in the region
of 2MB-3MB/s.

System information:

[george at aluminium ~]$ uname -a
Linux aluminium 3.5.1-1.fc17.x86_64 #1 SMP Thu Aug 9 17:50:43 UTC 2012
x86_64 x86_64 x86_64 GNU/Linux

[george at aluminium ~]$ lspci -vnn | grep 43 -A7
04:00.0 Network controller [0280]: Broadcom Corporation BCM4331
802.11a/b/g/n [14e4:4331] (rev 02)
	Subsystem: Apple Computer Inc. Device [106b:00ef]
	Flags: bus master, fast devsel, latency 0, IRQ 17
	Memory at c1900000 (64-bit, non-prefetchable) [size=16K]
	Capabilities: <access denied>
	Kernel driver in use: bcma-pci-bridge

In dmesg the relevant messages seem to be:

[ 3062.855352] wlan0: authenticate with d8:c7:c8:eb:08:20
[ 3062.865459] wlan0: direct probe to d8:c7:c8:eb:08:20 (try 1/3)
[ 3063.065716] wlan0: direct probe to d8:c7:c8:eb:08:20 (try 2/3)
[ 3063.266339] wlan0: direct probe to d8:c7:c8:eb:08:20 (try 3/3)
[ 3063.466953] wlan0: authentication with d8:c7:c8:eb:08:20 timed out
[ 3063.707190] wlan0: authenticate with d8:c7:c8:eb:0a:40
[ 3063.716868] wlan0: direct probe to d8:c7:c8:eb:0a:40 (try 1/3)
[ 3063.917113] wlan0: direct probe to d8:c7:c8:eb:0a:40 (try 2/3)
[ 3064.117756] wlan0: direct probe to d8:c7:c8:eb:0a:40 (try 3/3)
[ 3064.318329] wlan0: authentication with d8:c7:c8:eb:0a:40 timed out
[ 3064.563442] wlan0: authenticate with d8:c7:c8:eb:06:e0
[ 3064.563696] wlan0: send auth to d8:c7:c8:eb:06:e0 (try 1/3)
[ 3064.763538] wlan0: send auth to d8:c7:c8:eb:06:e0 (try 2/3)
[ 3064.771511] wlan0: authenticated
[ 3064.772525] wlan0: associate with d8:c7:c8:eb:06:e0 (try 1/3)
[ 3064.782600] wlan0: RX AssocResp from d8:c7:c8:eb:06:e0 (capab=0x431
status=0 aid=7)
[ 3064.782775] wlan0: associated
[ 3107.530689] ieee80211 phy0: wlan0: No probe response from AP
d8:c7:c8:eb:06:e0 after 500ms, disconnecting.
[ 3107.552816] cfg80211: Calling CRDA to update world regulatory domain
[ 3107.555845] cfg80211: World regulatory domain updated:
[ 3107.555850] cfg80211:   (start_freq - end_freq @ bandwidth),
(max_antenna_gain, max_eirp)
[ 3107.555854] cfg80211:   (2402000 KHz - 2472000 KHz @ 40000 KHz), (300
mBi, 2000 mBm)
[ 3107.555856] cfg80211:   (2457000 KHz - 2482000 KHz @ 20000 KHz), (300
mBi, 2000 mBm)
[ 3107.555858] cfg80211:   (2474000 KHz - 2494000 KHz @ 20000 KHz), (300
mBi, 2000 mBm)
[ 3107.555860] cfg80211:   (5170000 KHz - 5250000 KHz @ 40000 KHz), (300
mBi, 2000 mBm)
[ 3107.555863] cfg80211:   (5735000 KHz - 5835000 KHz @ 40000 KHz), (300
mBi, 2000 mBm)
[ 3107.556124] cfg80211: Calling CRDA for country: CA
[ 3107.559776] cfg80211: Regulatory domain changed to country: CA
[ 3107.559781] cfg80211:   (start_freq - end_freq @ bandwidth),
(max_antenna_gain, max_eirp)
[ 3107.559786] cfg80211:   (2402000 KHz - 2472000 KHz @ 40000 KHz), (300
mBi, 2700 mBm)
[ 3107.559790] cfg80211:   (5170000 KHz - 5250000 KHz @ 40000 KHz), (300
mBi, 1700 mBm)
[ 3107.559794] cfg80211:   (5250000 KHz - 5330000 KHz @ 40000 KHz), (300
mBi, 2000 mBm)
[ 3107.559798] cfg80211:   (5490000 KHz - 5710000 KHz @ 40000 KHz), (300
mBi, 2000 mBm)
[ 3107.559801] cfg80211:   (5735000 KHz - 5835000 KHz @ 40000 KHz), (300
mBi, 3000 mBm)

I'm loading the b43 module with the following parameters:

options b43 nohwcrypt=1 qos=0 hwpctl=1

I've also tried it with hwpctl=0 to no avail.

Can anyone suggest what may be going on here?

Thanks,

George

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Packet loss on bcm4331
  2012-08-14 19:07 Packet loss on bcm4331 George Wright
@ 2012-08-14 19:29 ` Larry Finger
  2012-08-14 20:05   ` Rafał Miłecki
  0 siblings, 1 reply; 3+ messages in thread
From: Larry Finger @ 2012-08-14 19:29 UTC (permalink / raw)
  To: b43-dev

On 08/14/2012 02:07 PM, George Wright wrote:
> I have a MacBook with a bcm4331 chip which mostly works in Linux (Fedora
> 17 w/kernel 3.5.1), but I'm noticing horrible packet loss and throughput
> a lot of the time.
>
> The setup is basically a small apartment with an 802.11bgn router
> sitting in the living room; I've had the laptop sit directly on top of
> the router and run ping to another computer wired directly into the
> access point and experienced 75% packet loss. File transfers are
> experiencing throughput in the region of 200kB/s to 800kB/s. Oddly
> enough, the amount of packet loss is significantly reduced if I'm
> connected in my bedroom (~20 feet from the router and with a couple of
> walls in between). I've also experienced the same problems with my
> backup wireless router (a WRT54GL running dd-wrt).

In b43, power management of 802.11n devices has never been implemented in the 
fullest form. Thus, it is likely that the gain setting is too high when you are 
close to the AP and the resulting distortion is causing failures.

> By comparison, on OS X I get 0% packet loss and throughput in the region
> of 2MB-3MB/s.

As Broadcom knows the specs for the device, they could get it right. We have to 
rely on reverse engineering.

--snip--

> I'm loading the b43 module with the following parameters:
>
> options b43 nohwcrypt=1 qos=0 hwpctl=1
>
> I've also tried it with hwpctl=0 to no avail.
>
> Can anyone suggest what may be going on here?

It is unlikely that any of the options will make any difference in this case..

Larry

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Packet loss on bcm4331
  2012-08-14 19:29 ` Larry Finger
@ 2012-08-14 20:05   ` Rafał Miłecki
  0 siblings, 0 replies; 3+ messages in thread
From: Rafał Miłecki @ 2012-08-14 20:05 UTC (permalink / raw)
  To: b43-dev

2012/8/14 Larry Finger <Larry.Finger@lwfinger.net>:
> On 08/14/2012 02:07 PM, George Wright wrote:
>>
>> I have a MacBook with a bcm4331 chip which mostly works in Linux (Fedora
>> 17 w/kernel 3.5.1), but I'm noticing horrible packet loss and throughput
>> a lot of the time.
>>
>> The setup is basically a small apartment with an 802.11bgn router
>> sitting in the living room; I've had the laptop sit directly on top of
>> the router and run ping to another computer wired directly into the
>> access point and experienced 75% packet loss. File transfers are
>> experiencing throughput in the region of 200kB/s to 800kB/s. Oddly
>> enough, the amount of packet loss is significantly reduced if I'm
>> connected in my bedroom (~20 feet from the router and with a couple of
>> walls in between). I've also experienced the same problems with my
>> backup wireless router (a WRT54GL running dd-wrt).
>
>
> In b43, power management of 802.11n devices has never been implemented in
> the fullest form. Thus, it is likely that the gain setting is too high when
> you are close to the AP and the resulting distortion is causing failures.

I promise to work on this, but I can provide only N-PHY support (we
have specs for them). I'll try my best to do some power management for
N-PHY for 3.7 merge window.
Right now I'm focused on SoC support for my BCM4706 SoC. But
fortunately I'm getting everything pushed upstream slowly.

Unfortunately BCM4331 is HT-PHY which I can't really improve. We don't
have any specs about that PHY type and support is mostly pure luck.

I may try to improve is a little since I get BCM4706 support (my SoC
has two BCM4331 connected via PCIE), but don't expect much :(

-- 
Rafa?

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2012-08-14 20:05 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-08-14 19:07 Packet loss on bcm4331 George Wright
2012-08-14 19:29 ` Larry Finger
2012-08-14 20:05   ` Rafał Miłecki

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox