linux-wireless.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* frustrated by carl9170
@ 2012-01-23 16:59 Brad Bellomo
  2012-01-23 18:15 ` Christian Lamparter
  0 siblings, 1 reply; 3+ messages in thread
From: Brad Bellomo @ 2012-01-23 16:59 UTC (permalink / raw)
  To: linux-wireless

I have a NetGear wireless n draft dongle that uses the carl9170 driver
on an Arch system.  This module works for anywhere from several
minutes to several hours.  Then I have lose all ability to use TCP/IP
over my wireless until I modprobe -r carl9170, modprobe carl9170, and
reconnect my network.  Although ar9170usb worked fine, I would rather
troubleshoot this than revert to a deprecated driver, or at worst
case, buy a better supported device.  I do not see anything relevant
in my dmesg log.  Does this device log errors somewhere else?  Is
carl9170 actively being developed?  I don't mind digging into code
myself, but if no one is working on carl9170, I would rather buy new
hardware (recommendations appreciated) than spend time fixing
something with no future.

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

* Re: frustrated by carl9170
  2012-01-23 16:59 frustrated by carl9170 Brad Bellomo
@ 2012-01-23 18:15 ` Christian Lamparter
       [not found]   ` <CALY7HCJ_9kh7YYk4YBzVkzK9180vf3H7mun-Gp7eCPNACK8pyA@mail.gmail.com>
  0 siblings, 1 reply; 3+ messages in thread
From: Christian Lamparter @ 2012-01-23 18:15 UTC (permalink / raw)
  To: Brad Bellomo; +Cc: linux-wireless

Hi,

On Monday, January 23, 2012 05:59:18 PM Brad Bellomo wrote:
> I have a NetGear wireless n draft dongle that uses the carl9170 driver
> on an Arch system.  This module works for anywhere from several
> minutes to several hours.  Then I have lose all ability to use TCP/IP
> over my wireless until I modprobe -r carl9170, modprobe carl9170, and
> reconnect my network.  Although ar9170usb worked fine, I would rather
> troubleshoot this than revert to a deprecated driver, or at worst
> case, buy a better supported device.  I do not see anything relevant
> in my dmesg log.  Does this device log errors somewhere else?  Is
> carl9170 actively being developed?  I don't mind digging into code
> myself, but if no one is working on carl9170, I would rather buy new
> hardware (recommendations appreciated) than spend time fixing
> something with no future.
There's not a lot of debug information in your post. e.g.: what's
the exact hardware? From where is the carl9170 driver coming from,
[kernel-version, or compat-wireless?] what's the firmware version?
Furthermore, a bit of background information about your setup wouldn't
hurt either.
[e.g.: What AP (and firmware)? Do you use HT20/HT40? which frequency band
and encryption setting].

When the device "stops". Do you know if it fails to receive
any data or if it has problems with outgoing traffic? Also what
is wpa_supplicant doing at the time when it fails?

The driver itself exports a debugfs interface under
/sys/kernel/debug/ieee80211/phy$X/carl9170

Of course, you are welcome to debug and run experiments on your own.
If you come across something peculiar (which can be reproduced) then
let me know.

Regards,
	Chr

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

* Re: frustrated by carl9170
       [not found]   ` <CALY7HCJ_9kh7YYk4YBzVkzK9180vf3H7mun-Gp7eCPNACK8pyA@mail.gmail.com>
@ 2012-01-26 14:05     ` Christian Lamparter
  0 siblings, 0 replies; 3+ messages in thread
From: Christian Lamparter @ 2012-01-26 14:05 UTC (permalink / raw)
  To: Brad Bellomo; +Cc: linux-wireless

Please keep the CC.

On Wednesday, January 25, 2012 02:39:38 AM Brad Bellomo wrote:
> There isn't much debug information as I was hoping people would help
> tell me how to debug this, rather than trying to debug it for me via
> e-mail.
Ok, but then were do we start? There's a lot more than just a driver
and a firmware that could be buggy. So, if you want to debug this
all by yourself, then be prepared to be frustrated a whole lot more.

> Does this driver log anything outside what I see as dmesg or
> debugfs?
No, the more interesting stuff is logged by the stack above.
You can take a look what the wpa_supplicant is doing in the
syslog [or wpa_cli] and mac80211/cfg80211 has an event viewer
[iw event] and trace points [needs to be enabled in the
kernel config at compile time of course]

> I don't have a debug directory under my kernel, is there
> something I need to do?
There no directory called debug? Well, then I assume that
the "debug Filesystem" (CONFIG_DEBUG_FS) wasn't enabled at
compile time. I'm afraid you have to enable it yourself
and build a kernel. I hope ARCH Linux has a few good wiki-pages
on the subject because you'll definitely need it.

> Is this driver being actively developed or should I just give
> up and buy new hardware?
I think it is :D.  But seriously, if you want to buy something else
go ahead. There's ath9k_htc, but you may end up having the same
problem [so, I would go for cheap products first].

> If it helps:
> 
> I am using a NETGEAR WNDA3100.
Not really, I have three of those and they are working fine in my
setup. Of course, I'm aware that my testing environment is just "mine"
so I won't even think about claiming that the driver/fw will always
work anytime for anyone anywhere.

> The question about outgoing or incoming or both being stopped is a
> good one, but I don't know how to check this.  I won't be able to ping
> to or from the machine either way.
Well, the debugfs interface exports a few statistics like tx and rx counters.
But you could also setup a monitor interface and wait until it dies.
If you still see incoming frames [e.g.: beacons from the AP] you know that
the device is still able to receive. And if you see ACKs and other responses
directed to your device MAC you know that it also sends alright. 
 
> I have replaced both the firmware and driver many times
> troubleshooting this problem, so I don't know where the current driver
> came from, but believe I am using the driver from the 3.0 kernel and
> firmware 1.9.4.  All other versions that I have tried that load (i.e.
> not mismatched driver/firmware) behave the same.
good to know.
 
> I am unsure about HT20/HT40 - I have not messed with this, so it is
> probably the default.
ar9170usb never supported 11n, so you may want to modprobe the module
with a noht=1 parameter to see if it also misbehaves when 11n is
turned off.

Reagards,
	Chr

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

end of thread, other threads:[~2012-01-26 14:05 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-01-23 16:59 frustrated by carl9170 Brad Bellomo
2012-01-23 18:15 ` Christian Lamparter
     [not found]   ` <CALY7HCJ_9kh7YYk4YBzVkzK9180vf3H7mun-Gp7eCPNACK8pyA@mail.gmail.com>
2012-01-26 14:05     ` Christian Lamparter

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).