Hi.
I have an Intel 7260 that fails to associate correctly in 802.11n mode.
Unfortunately I don't have any access to the AP in question, but I
believe it is an ASUS based device. This is on the 5GHz band; there is
some odd configuration with the router that results in the driver
refusing to associate on the 2.4GHz setup with the following error:
[ 507.495986] iwlwifi 0000:02:00.0: AP 30:00:da:b1:da:d0 beacon interval is 7, refusing due to firmware bug!
On the 5GHz band the driver attempts to associate, but never seems to
actually fully associate to the point of getting an IP. If I pass the
11n_disable=1 parameter to the driver then things work fine.
I've put a debug log of the failure to associate (kernel 3.14 but this
has happened with all the version I've tried) up at:
http://the.earth.li/~noodles/iwlwifi-fail.log
If there is more information I should be providing please let me know.
I'm sure it's related to the AP in question as I've had successful 11n
associations with other setups, but it's annoying to have to explicitly
disable 11n when in this location and then not in others.
J.
--
"evilwm - we sold our souls to the window manager" --
http://www.6809.org.uk/evilwm/
>
> On Mon, Apr 07, 2014 at 07:00:49AM +0000, Grumbach, Emmanuel wrote:
> > > On Mon, Apr 07, 2014 at 08:00:19AM +0300, Emmanuel Grumbach wrote:
> > > > On Mon, Apr 7, 2014 at 3:36 AM, Jonathan McDowell
> > > > <[email protected]>
> > > wrote:
> > > > >
> > > > > I have an Intel 7260 that fails to associate correctly in 802.11n mode.
> > > > > Unfortunately I don't have any access to the AP in question, but
> > > > > I believe it is an ASUS based device. This is on the 5GHz band;
> > > > > there is some odd configuration with the router that results in
> > > > > the driver refusing to associate on the 2.4GHz setup with the
> following error:
> > > > >
> > > > > [ 507.495986] iwlwifi 0000:02:00.0: AP 30:00:da:b1:da:d0 beacon
> > > > > interval is
> > > 7, refusing due to firmware bug!
> > > >
> > > > Ok - this is an interesting settings.
> > > > I'll check with our firmware team what can be done here, but
> > > > beacon
> > > > interval=7 is really insane. I can't find what the spec defines as
> > > > legal values right now.
> > >
> > > I wouldn't rule out the person who configured the AP having made a
> > > mistake; the MAC address is a custom one so they may have tweaked
> > > other settings too. However my phone (running Android) can connect
> > > to the 2.4GHz AP just fine and I think the other users of the AP are
> > > running Macs without problems.
> > >
> > > > > On the 5GHz band the driver attempts to associate, but never
> > > > > seems to actually fully associate to the point of getting an IP.
> > > > > If I pass the
> > > > > 11n_disable=1 parameter to the driver then things work fine.
> > > > >
> > > > This is surprising - can you share your logs?
> > >
> > > The log below is without the 11n_disable flag set - if it helps I
> > > can provide the log of when it works too.
> > >
> > > > > I've put a debug log of the failure to associate (kernel 3.14
> > > > > but this has happened with all the version I've tried) up at:
> > > > >
> > > > > http://the.earth.li/~noodles/iwlwifi-fail.log
> > > > >
> > > >
> > > > I can't access: Access forbidden.
> > > >
> > > Sorry, fixed.
> > >
> > Your log seems healthy - you associate at Apr 6 17:17:20 and
> > disassociate at Apr 6 17:18:05 - was that intentional?
>
> All I did was load the module with the debug parameter set, let Network
> Manager do its thing and wait for the connection to fail, then unload the
> module. NM is reporting that the DHCP times out - log at:
>
> http://the.earth.li/~noodles/iwlfail-nm.log
>
> I'd be blaming Network Manager or some other piece of configuration if it
> wasn't for the fact that disabling 11n makes things work.
>
I really can't see anything bad in this log.
Can you run tracing?
sudo trace-cmd record -e iwlwifi -e iwlwifi_msg -e mac80211 -e cfg80211
this will create a trace.dat file which I'd like to look at.
Thanks.
On Wed, Apr 16, 2014 at 07:58:01PM +0000, Grumbach, Emmanuel wrote:
> > > I really can't see anything bad in this log.
> > > Can you run tracing?
> > >
> > > sudo trace-cmd record -e iwlwifi -e iwlwifi_msg -e mac80211 -e
> > > cfg80211
> > >
> > > this will create a trace.dat file which I'd like to look at.
> >
> > http://the.earth.li/~noodles/iwlwifi/broken-trace.dat.xz
> >
> > and for completeness, with disable_11n=1:
> >
> > http://the.earth.li/~noodles/iwlwifi/working-trace.dat.xz
> >
> > In both cases I started the trace, flipped the hardware wifi disable switch, let
> > Network Manager do its thing.
> >
> Thanks for this. I am on vacation right now and won't be able to look
> at this straight away.
>
> Please don't remove the files from your web server. Also - if you can
> open a bug on bugzilla.kernel.org - this would be best. Cc
> [email protected] to your bug. Thanks!
No problem; bug 74311 raised. I'll leave the files around.
J.
--
I wondered why I lost three stones last week.
> On Sun, Apr 13, 2014 at 11:07:10AM +0000, Grumbach, Emmanuel wrote:
> > > > Your log seems healthy - you associate at Apr 6 17:17:20 and
> > > > disassociate at Apr 6 17:18:05 - was that intentional?
> > >
> > > All I did was load the module with the debug parameter set, let
> > > Network Manager do its thing and wait for the connection to fail,
> > > then unload the module. NM is reporting that the DHCP times out - log at:
> > >
> > > http://the.earth.li/~noodles/iwlfail-nm.log
> > >
> > > I'd be blaming Network Manager or some other piece of configuration
> > > if it wasn't for the fact that disabling 11n makes things work.
> > >
> > I really can't see anything bad in this log.
> > Can you run tracing?
> >
> > sudo trace-cmd record -e iwlwifi -e iwlwifi_msg -e mac80211 -e
> > cfg80211
> >
> > this will create a trace.dat file which I'd like to look at.
>
> http://the.earth.li/~noodles/iwlwifi/broken-trace.dat.xz
>
> and for completeness, with disable_11n=1:
>
> http://the.earth.li/~noodles/iwlwifi/working-trace.dat.xz
>
> In both cases I started the trace, flipped the hardware wifi disable switch, let
> Network Manager do its thing.
>
Thanks for this. I am on vacation right now and won't be able to look at this straight away.
Please don't remove the files from your web server.
Also - if you can open a bug on bugzilla.kernel.org - this would be best.
Cc [email protected] to your bug. Thanks!
On Mon, Apr 07, 2014 at 07:00:49AM +0000, Grumbach, Emmanuel wrote:
> > On Mon, Apr 07, 2014 at 08:00:19AM +0300, Emmanuel Grumbach wrote:
> > > On Mon, Apr 7, 2014 at 3:36 AM, Jonathan McDowell <[email protected]>
> > wrote:
> > > >
> > > > I have an Intel 7260 that fails to associate correctly in 802.11n mode.
> > > > Unfortunately I don't have any access to the AP in question, but I
> > > > believe it is an ASUS based device. This is on the 5GHz band; there is
> > > > some odd configuration with the router that results in the driver
> > > > refusing to associate on the 2.4GHz setup with the following error:
> > > >
> > > > [ 507.495986] iwlwifi 0000:02:00.0: AP 30:00:da:b1:da:d0 beacon interval is
> > 7, refusing due to firmware bug!
> > >
> > > Ok - this is an interesting settings.
> > > I'll check with our firmware team what can be done here, but beacon
> > > interval=7 is really insane. I can't find what the spec defines as
> > > legal values right now.
> >
> > I wouldn't rule out the person who configured the AP having made a
> > mistake; the MAC address is a custom one so they may have tweaked other
> > settings too. However my phone (running Android) can connect to the
> > 2.4GHz AP just fine and I think the other users of the AP are running
> > Macs without problems.
> >
> > > > On the 5GHz band the driver attempts to associate, but never seems to
> > > > actually fully associate to the point of getting an IP. If I pass the
> > > > 11n_disable=1 parameter to the driver then things work fine.
> > > >
> > > This is surprising - can you share your logs?
> >
> > The log below is without the 11n_disable flag set - if it helps I can
> > provide the log of when it works too.
> >
> > > > I've put a debug log of the failure to associate (kernel 3.14 but this
> > > > has happened with all the version I've tried) up at:
> > > >
> > > > http://the.earth.li/~noodles/iwlwifi-fail.log
> > > >
> > >
> > > I can't access: Access forbidden.
> > >
> > Sorry, fixed.
> >
> Your log seems healthy - you associate at Apr 6 17:17:20 and
> disassociate at Apr 6 17:18:05 - was that intentional?
All I did was load the module with the debug parameter set, let Network
Manager do its thing and wait for the connection to fail, then unload
the module. NM is reporting that the DHCP times out - log at:
http://the.earth.li/~noodles/iwlfail-nm.log
I'd be blaming Network Manager or some other piece of configuration if
it wasn't for the fact that disabling 11n makes things work.
> Also - your AP seems to behave badly with uAPSD - can you try to set
> power_scheme=1 as a module parameter to iwlmvm module?
Sure, I'll try that this evening.
J.
--
/-\ | "How the f**k did you work that
|@/ Debian GNU/Linux Developer | out?" -- Pythagoras
\- |
On Sun, Apr 13, 2014 at 11:07:10AM +0000, Grumbach, Emmanuel wrote:
> > > Your log seems healthy - you associate at Apr 6 17:17:20 and
> > > disassociate at Apr 6 17:18:05 - was that intentional?
> >
> > All I did was load the module with the debug parameter set, let Network
> > Manager do its thing and wait for the connection to fail, then unload the
> > module. NM is reporting that the DHCP times out - log at:
> >
> > http://the.earth.li/~noodles/iwlfail-nm.log
> >
> > I'd be blaming Network Manager or some other piece of configuration if it
> > wasn't for the fact that disabling 11n makes things work.
> >
> I really can't see anything bad in this log.
> Can you run tracing?
>
> sudo trace-cmd record -e iwlwifi -e iwlwifi_msg -e mac80211 -e cfg80211
>
> this will create a trace.dat file which I'd like to look at.
http://the.earth.li/~noodles/iwlwifi/broken-trace.dat.xz
and for completeness, with disable_11n=1:
http://the.earth.li/~noodles/iwlwifi/working-trace.dat.xz
In both cases I started the trace, flipped the hardware wifi disable
switch, let Network Manager do its thing.
J.
--
I just Fedexed my soul to hell. I'm *real* clever.
Hi,
On Mon, Apr 7, 2014 at 3:36 AM, Jonathan McDowell <[email protected]> wrote:
> Hi.
>
> I have an Intel 7260 that fails to associate correctly in 802.11n mode.
> Unfortunately I don't have any access to the AP in question, but I
> believe it is an ASUS based device. This is on the 5GHz band; there is
> some odd configuration with the router that results in the driver
> refusing to associate on the 2.4GHz setup with the following error:
>
> [ 507.495986] iwlwifi 0000:02:00.0: AP 30:00:da:b1:da:d0 beacon interval is 7, refusing due to firmware bug!
Ok - this is an interesting settings.
I'll check with our firmware team what can be done here, but beacon
interval=7 is really insane. I can't find what the spec defines as
legal values right now.
>
> On the 5GHz band the driver attempts to associate, but never seems to
> actually fully associate to the point of getting an IP. If I pass the
> 11n_disable=1 parameter to the driver then things work fine.
>
This is surprising - can you share your logs?
> I've put a debug log of the failure to associate (kernel 3.14 but this
> has happened with all the version I've tried) up at:
>
> http://the.earth.li/~noodles/iwlwifi-fail.log
>
I can't access: Access forbidden.
> If there is more information I should be providing please let me know.
> I'm sure it's related to the AP in question as I've had successful 11n
> associations with other setups, but it's annoying to have to explicitly
> disable 11n when in this location and then not in others.
>
> On Mon, Apr 07, 2014 at 08:00:19AM +0300, Emmanuel Grumbach wrote:
> > On Mon, Apr 7, 2014 at 3:36 AM, Jonathan McDowell <[email protected]>
> wrote:
> > >
> > > I have an Intel 7260 that fails to associate correctly in 802.11n mode.
> > > Unfortunately I don't have any access to the AP in question, but I
> > > believe it is an ASUS based device. This is on the 5GHz band; there is
> > > some odd configuration with the router that results in the driver
> > > refusing to associate on the 2.4GHz setup with the following error:
> > >
> > > [ 507.495986] iwlwifi 0000:02:00.0: AP 30:00:da:b1:da:d0 beacon interval is
> 7, refusing due to firmware bug!
> >
> > Ok - this is an interesting settings.
> > I'll check with our firmware team what can be done here, but beacon
> > interval=7 is really insane. I can't find what the spec defines as
> > legal values right now.
>
> I wouldn't rule out the person who configured the AP having made a
> mistake; the MAC address is a custom one so they may have tweaked other
> settings too. However my phone (running Android) can connect to the
> 2.4GHz AP just fine and I think the other users of the AP are running
> Macs without problems.
>
> > > On the 5GHz band the driver attempts to associate, but never seems to
> > > actually fully associate to the point of getting an IP. If I pass the
> > > 11n_disable=1 parameter to the driver then things work fine.
> > >
> > This is surprising - can you share your logs?
>
> The log below is without the 11n_disable flag set - if it helps I can
> provide the log of when it works too.
>
> > > I've put a debug log of the failure to associate (kernel 3.14 but this
> > > has happened with all the version I've tried) up at:
> > >
> > > http://the.earth.li/~noodles/iwlwifi-fail.log
> > >
> >
> > I can't access: Access forbidden.
>
> Sorry, fixed.
>
Your log seems healthy - you associate at Apr 6 17:17:20 and disassociate at Apr 6 17:18:05 - was that intentional?
Also - your AP seems to behave badly with uAPSD - can you try to set power_scheme=1 as a module parameter to iwlmvm module?
On Mon, Apr 07, 2014 at 08:00:19AM +0300, Emmanuel Grumbach wrote:
> On Mon, Apr 7, 2014 at 3:36 AM, Jonathan McDowell <[email protected]> wrote:
> >
> > I have an Intel 7260 that fails to associate correctly in 802.11n mode.
> > Unfortunately I don't have any access to the AP in question, but I
> > believe it is an ASUS based device. This is on the 5GHz band; there is
> > some odd configuration with the router that results in the driver
> > refusing to associate on the 2.4GHz setup with the following error:
> >
> > [ 507.495986] iwlwifi 0000:02:00.0: AP 30:00:da:b1:da:d0 beacon interval is 7, refusing due to firmware bug!
>
> Ok - this is an interesting settings.
> I'll check with our firmware team what can be done here, but beacon
> interval=7 is really insane. I can't find what the spec defines as
> legal values right now.
I wouldn't rule out the person who configured the AP having made a
mistake; the MAC address is a custom one so they may have tweaked other
settings too. However my phone (running Android) can connect to the
2.4GHz AP just fine and I think the other users of the AP are running
Macs without problems.
> > On the 5GHz band the driver attempts to associate, but never seems to
> > actually fully associate to the point of getting an IP. If I pass the
> > 11n_disable=1 parameter to the driver then things work fine.
> >
> This is surprising - can you share your logs?
The log below is without the 11n_disable flag set - if it helps I can
provide the log of when it works too.
> > I've put a debug log of the failure to associate (kernel 3.14 but this
> > has happened with all the version I've tried) up at:
> >
> > http://the.earth.li/~noodles/iwlwifi-fail.log
> >
>
> I can't access: Access forbidden.
Sorry, fixed.
> > If there is more information I should be providing please let me know.
> > I'm sure it's related to the AP in question as I've had successful 11n
> > associations with other setups, but it's annoying to have to explicitly
> > disable 11n when in this location and then not in others.
J.
--
/-\ | Where's your fishing rod, lawn
|@/ Debian GNU/Linux Developer | ornament?
\- |
> On Wed, Apr 16, 2014 at 07:58:01PM +0000, Grumbach, Emmanuel wrote:
> > > > I really can't see anything bad in this log.
> > > > Can you run tracing?
> > > >
> > > > sudo trace-cmd record -e iwlwifi -e iwlwifi_msg -e mac80211 -e
> > > > cfg80211
> > > >
> > > > this will create a trace.dat file which I'd like to look at.
> > >
> > > http://the.earth.li/~noodles/iwlwifi/broken-trace.dat.xz
> > >
> > > and for completeness, with disable_11n=1:
> > >
> > > http://the.earth.li/~noodles/iwlwifi/working-trace.dat.xz
> > >
> > > In both cases I started the trace, flipped the hardware wifi disable
> > > switch, let Network Manager do its thing.
> > >
> > Thanks for this. I am on vacation right now and won't be able to look
> > at this straight away.
> >
> > Please don't remove the files from your web server. Also - if you can
> > open a bug on bugzilla.kernel.org - this would be best. Cc
> > [email protected] to your bug. Thanks!
>
> No problem; bug 74311 raised. I'll leave the files around.
>
Many thanks - see you there.