Larry is the expert here, so I think it is worthwhile waiting for his advice here, but I do note that the b43 driver has a QOS option that can be disabled. I’ve seen the advice to disable via a config file located in /etc/modprobe.d/ several times across various threads (various distros). For example:
On 01/02/2013 01:26 AM, pilotmm wrote:
>
> Hi lwfinger
>
> Thanks for your help and answering all my questions. I
> appreciate your efforts to improve wireless drivers for the Linux
> kernel.
>
> I have adjusted the channel for the AP and it did make a
> difference. However I am still loosing a connection every second or
> third web-page load. I can still browse web pages without too much
> annoyance. I just do not get that snappy response I am use to. I guess
> that is the best I can expect out of my adapter.
>
> I am curious what software controls the handshaking
> routine. Is that all in the driver or is there a wireless module that is
> driving the module? It seems strange that the device has to be
> reconnected when it misses a hand-shaking? Could it not be re-programmed
> to retry the hand shaking routine? Just some curiosities.
Do you still have that obscenely large signal strength after your reconfiguration?
Are you still getting the “No probe response from AP” messages?
You are certainly welcome to offer fixes to any part of the system in the way of
patches. To answer your question, the responsibilities for communication are
split. First of all, the firmware handles the actual transmission or reception,
the acknowledgment of packets, and anything else that is too time critical to be
handled by the general CPU. The firmware code is not open-source, and is
essentially unchangeable. The MAC layer, aka mac80211, handles the rate control,
the preparation of packets, and other tasks that are specific to the 802.11
protocols. It is responsible for the handshaking. The driver sits in the middle
where it reformats the data provided by either side, and does the things that
are peculiar to this piece of hardware.
If the only problem is while web browsing, you could also have a DNS problem. To
test that, edit /etc/resolv.conf (as root) and add a line that says “nameserver
8.8.8.8” before any other nameserver lines. That server is a public one run by
Google that does IPv6 correctly, and always responds quickly. My ISP’s name
servers are really bad, and I always use Google.
Hello Deano and Larry
Thanks a lot for all your help and explain more about how wireless adapters work. Your tips have made it better. I tried the QOS fix it seems to help, not a complete fix though. It is not perfect but it is good enough for what I need. Thanks a lot for all the help.
Take Care
Mike
Hello,
I’m experiencing similar problems to those of Pilot, with regard to my wireless connection.
I am using an Acer eMachines E625 Laptop with a Broadcom B4312 [14e4:4315] chipset.
Currently, I’m running OpenSuse 12.2
My wireless connection drops after a few minutes of inactivity. It also drops the moment that I open firefox, or move to another page in Firefox. It is impossible to surf with Firefox, although konqueror is a lot more stable and may allow a few minutes of surfing.
My Laptop is dual boot, and when I boot with Windoze Vista, my wireless network connection in completely stable and reliable. I also had OpenSuse 11.3 installed previously on this laptop and the wireless connection worked perfectly then (using SSB).
I have tried the wl, b43, b43legacy and b43 with LP/PHY drivers, as well as ndiswrapper. All yeild exactly the same problem.
Let me know if I can send any info that may help with this issue
Regards,
Michaho
On 01/03/2013 01:36 PM, michaho wrote:
>
> Hello,
>
> I’m experiencing similar problems to those of Pilot, with regard to my
> wireless connection.
> I am using an Acer eMachines E625 Laptop with a Broadcom B4312
> [14e4:4315] chipset.
> Currently, I’m running OpenSuse 12.2
>
> My wireless connection drops after a few minutes of inactivity. It also
> drops the moment that I open firefox, or move to another page in
> Firefox. It is impossible to surf with Firefox, although konqueror is a
> lot more stable and may allow a few minutes of surfing.
>
> My Laptop is dual boot, and when I boot with Windoze Vista, my wireless
> network connection in completely stable and reliable. I also had
> OpenSuse 11.3 installed previously on this laptop and the wireless
> connection worked perfectly then (using SSB).
>
> I have tried the wl, b43, b43legacy and b43 with LP/PHY drivers, as
> well as ndiswrapper. All yeild exactly the same problem.
>
>
> Let me know if I can send any info that may help with this issue
This report is very puzzling. Some of the newer 14e4:4315 units can be trouble;
however, if you used it with 11.3, then yours is an older model and must be the
same as mine. Under extremely high throughput, my device reports PHY
TRANSMISSION ERRORS and will drop out, but it never drops out under those
conditions.
A couple of corrections to what you reported - driver b43 uses a companion
driver to connect to the PCI bus. For older Broadcom units, that driver is ssb.
That has not changed for as long as b43 has been in the kernel, thus you were
using b43 with 11.3. Secondly, b43legacy is for much older devices than yours.
When you tried it, absolutely nothing happened other than loading the driver.
As you see the same problem with wl, b43, and the Windows driver (through
ndiswrapper), it would seem that none of those drivers are at fault. In
addition, your tests involve completely different implementations of the MAC layer.
I am just grasping at straws with the next questions.
Are you using the same AP as you did with 11.3?
Does b43 report any errors when it disconnects?
Please post the output from a ‘dmesg | egrep “ssb|b43”’ command.
Hello All
Just looking further into this driver issue I came across the following article:
Interesting to read that under features they list interference mitigation as not working. I do not know exactly what that means. Yet it might be the reason I see these problems. Maybe firefox is trying to establish multiple connections simultaneously and causing this interference? Just speculating... thought it was a relevant article.
Mike
On 01/03/2013 08:56 PM, pilotmm wrote:
>
> Hello All
>
> Just looking further into this driver issue I came across the
> following article:
>
> ‘b43 - Linux Wireless’
> (http://wireless.kernel.org/en/users/Drivers/b43)
>
> Interesting to read that under features they list interference
> mitigation as not working. I do not know exactly what that means. Yet it
> might be the reason I see these problems. Maybe firefox is trying to
> establish multiple connections simultaneously and causing this
> interference? Just speculating… thought it was a relevant article.
That is interference from other APs than yours, not multiple threads running our
your connection. When you do the command ‘sudo /usr/sbin/iwlist scan’, how many
stations do you see with a channel +/- 2 from the one you use, and how strong
are they? These chips are not very good in a noisy environment.
Hello,
I’m still using the same AP with the same settings that I was using previously with 11.3.
(This AP works fine with other laptops running Win7, and with the problem laptop when booted with Win Vista)
Please find below the output from dmesg when the connection dropped:
dmesg | egrep “ssb|b43”
15.495100] ssb: Found chip with id 0x4312, rev 0x01 and package 0x00
15.495112] ssb: Core 0 found: ChipCommon (cc 0x800, rev 0x16, vendor 0x4243)
15.495120] ssb: Core 1 found: IEEE 802.11 (cc 0x812, rev 0x0F, vendor 0x4243)
15.495127] ssb: Core 2 found: PCMCIA (cc 0x80D, rev 0x0A, vendor 0x4243)
15.495135] ssb: Core 3 found: PCI-E (cc 0x820, rev 0x09, vendor 0x4243)
15.535131] ssb: Sonics Silicon Backplane found on PCI device 0000:02:00.0
17.279140] b43-phy0: Broadcom 4312 WLAN found (core revision 15)
17.391751] Registered led device: b43-phy0::tx
17.391778] Registered led device: b43-phy0::rx
17.391804] Registered led device: b43-phy0::radio
30.462115] b43-phy0: Loading firmware version 478.104 (2008-07-01 00:50:23)
31.837178] b43-phy0 warning: Forced PIO by use_pio module parameter. This should not be needed and will result in lower performance.
88.549084] b43-phy0: Loading firmware version 478.104 (2008-07-01 00:50:23)
89.919179] b43-phy0 warning: Forced PIO by use_pio module parameter. This should not be needed and will result in lower performance.
434.884095] b43-phy0: Loading firmware version 478.104 (2008-07-01 00:50:23)
436.257187] b43-phy0 warning: Forced PIO by use_pio module parameter. This should not be needed and will result in lower performance.
Many thanks,
Michaho
On 01/04/2013 08:56 AM, michaho wrote:
>
> Hello,
>
> I’m still using the same AP with the same settings that I was using
> previously with 11.3.
> (This AP works fine with other laptops running Win7, and with the
> problem laptop when booted with Win Vista)
>
> Please find below the output from dmesg when the connection dropped:
>
> dmesg | egrep “ssb|b43”
> 15.495100] ssb: Found chip with id 0x4312, rev 0x01 and package
> 0x00
> 15.495112] ssb: Core 0 found: ChipCommon (cc 0x800, rev 0x16,
> vendor 0x4243)
> 15.495120] ssb: Core 1 found: IEEE 802.11 (cc 0x812, rev 0x0F,
> vendor 0x4243)
> 15.495127] ssb: Core 2 found: PCMCIA (cc 0x80D, rev 0x0A, vendor
> 0x4243)
> 15.495135] ssb: Core 3 found: PCI-E (cc 0x820, rev 0x09, vendor
> 0x4243)
> 15.535131] ssb: Sonics Silicon Backplane found on PCI device
> 0000:02:00.0
> 17.279140] b43-phy0: Broadcom 4312 WLAN found (core revision 15)
> 17.391751] Registered led device: b43-phy0::tx
> 17.391778] Registered led device: b43-phy0::rx
> 17.391804] Registered led device: b43-phy0::radio
> 30.462115] b43-phy0: Loading firmware version 478.104 (2008-07-01
> 00:50:23)
> 31.837178] b43-phy0 warning: Forced PIO by use_pio module
> parameter. This should not be needed and will result in lower
> performance.
> 88.549084] b43-phy0: Loading firmware version 478.104 (2008-07-01
> 00:50:23)
> 89.919179] b43-phy0 warning: Forced PIO by use_pio module
> parameter. This should not be needed and will result in lower
> performance.
> 434.884095] b43-phy0: Loading firmware version 478.104 (2008-07-01
> 00:50:23)
> 436.257187] b43-phy0 warning: Forced PIO by use_pio module
> parameter. This should not be needed and will result in lower
> performance.
You should not need that pio parameter. In fact, that may be causing your
trouble as that I/O method cannot handle high throughput.
I get the same messages… not sure how to disable PIO however I have a feeling that the following configuration may have something to do with it? These are the commands that I ran… regardless I will play around with this configuration and report back when I know more
sudo touch /etc/modprobe.d/b43.conf echo "options b43 pio=1 qos=0" | sudo tee -a /etc/modprobe.d/b43.conf
Mike
On 01/04/2013 12:46 PM, pilotmm wrote:
>
> I get the same messages… not sure how to disable PIO however I have a
> feeling that the following configuration may have something to do with
> it? These are the commands that I ran… regardless I will play around
> with this configuration and report back when I know more
>
>
> Code:
> --------------------
>
> sudo touch /etc/modprobe.d/b43.conf echo “options b43 pio=1 qos=0” | sudo tee -a /etc/modprobe.d/b43.conf
> --------------------
Get rid of the “pio=1” part. That is forbidding the use of DMA in the transfers.
I don’t know if the qos option is needed. I run without it, but as long as you
are not doing VOIP, etc., then it probably does not make a difference.
I switched to the wl driver from brcmsmac and have had no more dropped connection issues. Thanks for the info!
Eric
Hi,
I’ve removed the pio and qos settings from the b43.conf.
I can now ping continuously without a problem, until I open a browser - Konquorer allows me to surf a few pages, but firefox cuases my connection to drop immediately.
I can check my e-mail using K-mail without a problem
(interestingly, the wireless LED on the laptop remains on even whhen the connection is dropped!)
dmesg | egrep “b43|ssb”
15.343110] ssb: Found chip with id 0x4312, rev 0x01 and package 0x00
15.343121] ssb: Core 0 found: ChipCommon (cc 0x800, rev 0x16, vendor 0x4243)
15.343129] ssb: Core 1 found: IEEE 802.11 (cc 0x812, rev 0x0F, vendor 0x4243)
15.343136] ssb: Core 2 found: PCMCIA (cc 0x80D, rev 0x0A, vendor 0x4243)
15.343144] ssb: Core 3 found: PCI-E (cc 0x820, rev 0x09, vendor 0x4243)
15.383141] ssb: Sonics Silicon Backplane found on PCI device 0000:02:00.0
16.279726] b43-phy0: Broadcom 4312 WLAN found (core revision 15)
16.994480] Registered led device: b43-phy0::tx
16.994535] Registered led device: b43-phy0::rx
16.994591] Registered led device: b43-phy0::radio
30.214086] b43-phy0: Loading firmware version 478.104 (2008-07-01 00:50:23)
103.075090] b43-phy0: Loading firmware version 478.104 (2008-07-01 00:50:23)
What do the pio and qos settings do?
Thanks,
Michael
On 01/05/2013 02:46 PM, michaho wrote:
>
> Hi,
>
> I’ve removed the pio and qos settings from the b43.conf.
> I can now ping continuously without a problem, until I open a browser -
> Konquorer allows me to surf a few pages, but firefox cuases my
> connection to drop immediately.
> I can check my e-mail using K-mail without a problem
> (interestingly, the wireless LED on the laptop remains on even whhen
> the connection is dropped!)
>
>
> dmesg | egrep “b43|ssb”
> 15.343110] ssb: Found chip with id 0x4312, rev 0x01 and package
> 0x00
> 15.343121] ssb: Core 0 found: ChipCommon (cc 0x800, rev 0x16,
> vendor 0x4243)
> 15.343129] ssb: Core 1 found: IEEE 802.11 (cc 0x812, rev 0x0F,
> vendor 0x4243)
> 15.343136] ssb: Core 2 found: PCMCIA (cc 0x80D, rev 0x0A, vendor
> 0x4243)
> 15.343144] ssb: Core 3 found: PCI-E (cc 0x820, rev 0x09, vendor
> 0x4243)
> 15.383141] ssb: Sonics Silicon Backplane found on PCI device
> 0000:02:00.0
> 16.279726] b43-phy0: Broadcom 4312 WLAN found (core revision 15)
> 16.994480] Registered led device: b43-phy0::tx
> 16.994535] Registered led device: b43-phy0::rx
> 16.994591] Registered led device: b43-phy0::radio
> 30.214086] b43-phy0: Loading firmware version 478.104 (2008-07-01
> 00:50:23)
> 103.075090] b43-phy0: Loading firmware version 478.104 (2008-07-01
> 00:50:23)
The pio setting turns off DMA (direct memory access) and turns on PIO
(programmed Input/Output). With DMA, the device transfers data to/from memory
without involving the CPU. With PIO, the CPU reads a byte from memory, and
writes it to the device, or vice versa. PIO takes a lot more CPU.
QoS refers to “quality of service”. The device has 4 different hardware transmit
queues. When the hardware is ready to transmit, it selects a packet from the
highest-priority queue that contains information. With QoS active, high priority
actions such as VOIP (Voice Over Internet Protocol) have an advantage over
normal traffic.
Many thanks for the info.
The PIO is off, but if I look at Systems Monitor, I often see my CPU usage going upto 100% when on the internet.
When the laptop is idling, Systems Monitor indicates approximately 20% CPU usage. When I open a webpage using Konqueror that CPU usage shoots up to over 50% while the page loads. If I open pages in 2 or 3 more tabs, then the CPU usage jumps to 100%. If this 100% level is maintained for a short while, then then network disconnects.
If I open firefox, my CPU usage jumps to 50% before I open a webpage.
My CPU is an AMD Athlon TF20 (1,6GHz) processor, with 2GB of RAM.
Am I maybe looking in the wrong place for my network issue?
Could my laptop just be running out of resources?
Is it normal for a browser to put such load on the CPU?
Many Thanks
Michaho
On 01/06/2013 09:36 AM, michaho wrote:
>
> Many thanks for the info.
>
> The PIO is off, but if I look at Systems Monitor, I often see my CPU
> usage going upto 100% when on the internet.
> When the laptop is idling, Systems Monitor indicates approximately 20%
> CPU usage. When I open a webpage using Konqueror that CPU usage shoots
> up to over 50% while the page loads. If I open pages in 2 or 3 more
> tabs, then the CPU usage jumps to 100%. If this 100% level is maintained
> for a short while, then then network disconnects.
>
> If I open firefox, my CPU usage jumps to 50% before I open a webpage.
>
> My CPU is an AMD Athlon TF20 (1,6GHz) processor, with 2GB of RAM.
>
> Am I maybe looking in the wrong place for my network issue?
> Could my laptop just be running out of resources?
> Is it normal for a browser to put such load on the CPU?
Yes, the browser will place heavy loads on both CPU and memory; however, after
you open Firefox, the CPU load should drop back to zero if you do not open any
web pages other than the default. If it does not, use “top” to find the process
that is running. Does “free” show the use of swap?
I have an HP Mini 110 netbook with a single 1.6 GHz processor and 1 GB of RAM.
On it, both KDE and Gnome are very sluggish; however, it is quite usable with
LXDE. Try your machine with any of the lighter desktops.
On 2013-01-06 16:36, michaho wrote:
> Am I maybe looking in the wrong place for my network issue?
> Could my laptop just be running out of resources?
> Is it normal for a browser to put such load on the CPU?
If the network card is still using PIO, then the CPU usage could go up a
lot, and network would go down if the CPU can’t cope.
You can find out if it is that or the browser by downloading a file from
the command line with wget.
–
Cheers / Saludos,
Carlos E. R.
(from 11.4, with Evergreen, x86_64 “Celadon” (Minas Tirith))
Hi,PIO is definitely off.I find that if I don’t load the network too much, then the connection stays up.I can check my e-mail and surf with Konqueror (as long as I don’t open more than 2 pages at a time!). I have checked my Facebook etc. As long as I only do one thing at a time then there is no problem. If I open too many pages, then my CPU usage jumps to 100% and my connection drops after a few moments. However, Firefox still causes the network connection to drop as soon as it is opened.For now, I’ll just have to be content with Konqueror even though it is a little rough compared to Firefox. I’ll keep digging over the next few days and see if anything obvious appears.Many thanks for all the helpMichaho
On 2013-01-08 19:46, michaho wrote:
>
> Hi,PIO is definitely off.I find that if I don’t load the network too
> much, then the connection stays up.I can check my e-mail and surf with
> Konqueror (as long as I don’t open more than 2 pages at a time!).
Do the download test with wget, as I said.
–
Cheers / Saludos,
Carlos E. R.
(from 11.4, with Evergreen, x86_64 “Celadon” (Minas Tirith))
Hi Carlos,
Been hectically busy here.
Have done a wget. I’m not sure what I’m looking for, so have attached the wget logfile below:
–2013-01-10 18:57:34-- ftp://ftp.regalcctv.dyndns.org/3teamviewer3_setup.zip
=> `3teamviewer3_setup.zip’
Resolving ftp.regalcctv.dyndns.org (ftp.regalcctv.dyndns.org)… 41.185.121.27
Connecting to ftp.regalcctv.dyndns.org (ftp.regalcctv.dyndns.org)|41.185.121.27|:21… connected.
Logging in as anonymous … Logged in!
==> SYST … done. ==> PWD … done.
==> TYPE I … done. ==> CWD not needed.
==> SIZE 3teamviewer3_setup.zip … 1037411
==> PASV … done. ==> RETR 3teamviewer3_setup.zip … done.
Length: 1037411 (1013K) (unauthoritative)
0K .......... .......... .......... .......... .......... 4% 22.3K 43s
50K .......... .......... .......... .......... .......... 9% 8.74K 73s
100K … … … … … 14% 61.0M 46s
150K … … … … … 19% 29.3K 39s
200K … … … … … 24% 33.1K 34s
250K … … … … … 29% 34.0K 30s
300K … … … … … 34% 21.6K 28s
350K … … … … … 39% 37.3K 25s
400K … … … … … 44% 27.7K 23s
450K … … … … … 49% 33.6K 20s
500K … … … … … 54% 39.1K 18s
550K … … … … … 59% 21.1K 16s
600K … … … … … 64% 21.3K 14s
650K … … … … … 69% 26.6K 12s
700K … … … … … 74% 40.5K 10s
750K … … … … … 78% 34.2K 8s
800K … … … … … 83% 33.1K 6s
850K … … … … … 88% 34.5K 4s
900K … … … … … 93% 20.0K 2s
950K … … … … … 98% 42.0K 0s
1000K … … 100% 9.16K=38s
2013-01-10 18:58:18 (26.5 KB/s) - `3teamviewer3_setup.zip’ saved [1037411]
what should I be looking for in this log?
Some other questions:
- is there a way to force an network connection to run at a certain speed (rather than just selecting auto)?
- Is there a way to add a memory buffer, or just allocate more, to the network card?
Many thanks for your help
Michaho