[PATCH] Add iwlwifi wireless drivers (resend)

[PATCH] Add iwlwifi wireless drivers (resend)

[Zhu Yi]

[ My previous post doesn't show up on the list for a whole day. So I
resend it now. I guess the an 800k mail is too big for the list... ]

This patch (against 2.6.23-rc3) adds the mac80211 based
wireless drivers for the Intel PRO/Wireless 3945ABG/BG Network
Connection and Intel Wireless WiFi Link AGN (4965) adapters. I hope it
can be merged into 2.6.24 kernel. You comments are also welcome!

Signed-off-by: Zhu Yi <yi.zhu@intel.com>
---
 MAINTAINERS                         |   10 +
 drivers/net/wireless/Kconfig        |  129 +
 drivers/net/wireless/Makefile       |   29 +
 drivers/net/wireless/iwl-3945-hw.h  |  117 +
 drivers/net/wireless/iwl-3945-rs.c  |  985 ++++
 drivers/net/wireless/iwl-3945-rs.h  |  221 +
 drivers/net/wireless/iwl-3945.c     | 2286 +++++++++
 drivers/net/wireless/iwl-3945.h     |   60 +
 drivers/net/wireless/iwl-4965-hw.h  |  954 ++++
 drivers/net/wireless/iwl-4965-rs.c  | 2162 ++++++++
 drivers/net/wireless/iwl-4965-rs.h  |  286 ++
 drivers/net/wireless/iwl-4965.c     | 4764 +++++++++++++++++
 drivers/net/wireless/iwl-4965.h     |  368 ++
 drivers/net/wireless/iwl-base.c     | 9603
+++++++++++++++++++++++++++++++++++
 drivers/net/wireless/iwl-channel.h  |  163 +
 drivers/net/wireless/iwl-commands.h | 1515 ++++++
 drivers/net/wireless/iwl-debug.h    |  149 +
 drivers/net/wireless/iwl-eeprom.h   |  336 ++
 drivers/net/wireless/iwl-helpers.h  |  255 +
 drivers/net/wireless/iwl-hw.h       |  606 +++
 drivers/net/wireless/iwl-io.h       |  486 ++
 drivers/net/wireless/iwl-priv.h     |  305 ++
 drivers/net/wireless/iwl-spectrum.h |   91 +
 drivers/net/wireless/iwlwifi.h      |  715 +++


Please find the patch (and the original mail) from:

http://intellinuxwireless.org/iwlwifi/v4-Add-iwlwifi-wireless-drivers.patch


Thanks,
-yi

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 652 bytes --]

On Fri, 2007-08-17 at 13:28 +0800, Zhu Yi wrote:
> [ My previous post doesn't show up on the list for a whole day. So I
> resend it now. I guess the an 800k mail is too big for the list... ]

Heh, yes.

> This patch (against 2.6.23-rc3) adds the mac80211 based
> wireless drivers for the Intel PRO/Wireless 3945ABG/BG Network
> Connection and Intel Wireless WiFi Link AGN (4965) adapters. I hope it
> can be merged into 2.6.24 kernel. You comments are also welcome!

I was recently reviewing encryption stuff---can you comment on why you
don't allow disabling hardware keys? For proper operation you really
should allow that.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 190 bytes --]

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 271 bytes --]

+	  See <file:Documentation/networking/README.iwlwifi> for
+	  information on the capabilities currently enabled in this
+	  driver and for tips for debugging any issues or problems.

-ENOENT. It's not in wireless-dev either, has it been written at all?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 190 bytes --]

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 493 bytes --]

On Fri, 2007-08-17 at 13:38 +0200, Johannes Berg wrote:
> +	  See <file:Documentation/networking/README.iwlwifi> for
> +	  information on the capabilities currently enabled in this
> +	  driver and for tips for debugging any issues or problems.
> 
> -ENOENT. It's not in wireless-dev either, has it been written at all?

Ah, found it in iwlwifi.git, but it's mostly a list of TODOs and
documentation that only applies to the out-of-tree version, except for
the legalese.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 190 bytes --]

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Zhu Yi]

On Fri, 2007-08-17 at 13:41 +0200, Johannes Berg wrote:
> Ah, found it in iwlwifi.git, but it's mostly a list of TODOs and
> documentation that only applies to the out-of-tree version, except for
> the legalese. 

Yeah, the document is far lagged behind the code so I didn't include it
this time.

Thanks,
-yi

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Zhu Yi]

On Fri, 2007-08-17 at 13:26 +0200, Johannes Berg wrote:
> I was recently reviewing encryption stuff---can you comment on why you
> don't allow disabling hardware keys? For proper operation you really
> should allow that. 

I don't think there is any special reason, just haven't enabled it yet.
We enable the hwcrypto and find it works good, then we switched to
enable something else. For the end users, I think they'd always want to
use hwcrypto if it is supported by the hardware. For developers, I'd
agree it is useful. I'll add it to the TODO list.

Thanks,
-yi

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 840 bytes --]

On Mon, 2007-08-20 at 10:56 +0800, Zhu Yi wrote:
> On Fri, 2007-08-17 at 13:26 +0200, Johannes Berg wrote:
> > I was recently reviewing encryption stuff---can you comment on why you
> > don't allow disabling hardware keys? For proper operation you really
> > should allow that. 
> 
> I don't think there is any special reason, just haven't enabled it yet.
> We enable the hwcrypto and find it works good, then we switched to
> enable something else. For the end users, I think they'd always want to
> use hwcrypto if it is supported by the hardware. For developers, I'd
> agree it is useful. I'll add it to the TODO list.

I guess the relevant thing is that if an association is lost then the
key shouldn't be retained in the hardware because it should no longer be
used, yet you don't allow to disable that key.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 190 bytes --]

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 426 bytes --]

On Mon, 2007-08-20 at 10:39 +0800, Zhu Yi wrote:
> On Fri, 2007-08-17 at 13:41 +0200, Johannes Berg wrote:
> > Ah, found it in iwlwifi.git, but it's mostly a list of TODOs and
> > documentation that only applies to the out-of-tree version, except for
> > the legalese. 
> 
> Yeah, the document is far lagged behind the code so I didn't include it
> this time.

Better remove the references as well then.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 190 bytes --]

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Tomas Winkler]

On 8/20/07, Johannes Berg <johannes@sipsolutions.net> wrote:
> On Mon, 2007-08-20 at 10:56 +0800, Zhu Yi wrote:
> > On Fri, 2007-08-17 at 13:26 +0200, Johannes Berg wrote:
> > > I was recently reviewing encryption stuff---can you comment on why you
> > > don't allow disabling hardware keys? For proper operation you really
> > > should allow that.
> >
> > I don't think there is any special reason, just haven't enabled it yet.
> > We enable the hwcrypto and find it works good, then we switched to
> > enable something else. For the end users, I think they'd always want to
> > use hwcrypto if it is supported by the hardware. For developers, I'd
> > agree it is useful. I'll add it to the TODO list.
>
> I guess the relevant thing is that if an association is lost then the
> key shouldn't be retained in the hardware because it should no longer be
> used, yet you don't allow to disable that key.

iwlwifi has notion of 'station' in driver and firmware. The whole
instance of station including keys has to be removed on association
lost. We need get at least mac address of the station that has lost
association into driver to do that. There is not comfortable interface
right now mac80211 to do that. The same goes for association process
where station has to be added we use rate scale init for that right
now.
There is some table_notification handler but it only updates driver
about number of stations.

There was a whole discussion we led about if rate scale algorithm and
notion of station should separated from the driver or not. I'm not
sure it was closed yet.

Thanks
Tomas

> johannes
>
>

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1399 bytes --]

On Mon, 2007-08-20 at 15:10 +0300, Tomas Winkler wrote:

> iwlwifi has notion of 'station' in driver and firmware. The whole
> instance of station including keys has to be removed on association
> lost. We need get at least mac address of the station that has lost
> association into driver to do that. There is not comfortable interface
> right now mac80211 to do that. The same goes for association process
> where station has to be added we use rate scale init for that right
> now.

Thanks for the explanation.

> There is some table_notification handler but it only updates driver
> about number of stations.

Right. In a recent patch I removed functionality for setting a NULL key
to the hardware for such STAs and remarked something along the lines of
"If hardware needs that, we should not abuse the set_key() function to
notify it of new STAs"

Seems that is the case with your hardware as well as some atheros
hardware the code was initially written for. I'm sure we're willing to
consider patches that extend the sta_table_notification() handle to
include the MAC address of the station that was added/removed.

> There was a whole discussion we led about if rate scale algorithm and
> notion of station should separated from the driver or not. I'm not
> sure it was closed yet.

I only recall the rate scaling vs. driver but nothing about STAs.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 190 bytes --]

Re: [PATCH] Add iwlwifi wireless drivers (resend)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 666 bytes --]

On Fri, 2007-08-17 at 13:28 +0800, Zhu Yi wrote:

> This patch (against 2.6.23-rc3)

+#if IWL == 4965
+	modes[G_11N].mode = MODE_ATHEROS_TURBOG;
+	modes[G_11N].channels = channels;
+	modes[G_11N].num_rates = 13;        /* OFDM & CCK */
+	modes[G_11N].rates = rates;
+	modes[G_11N].num_channels = 0;
+
+	modes[A_11N].mode = MODE_ATHEROS_TURBO;
+	modes[A_11N].channels = &channels[ARRAY_SIZE(iwl_eeprom_band_1)];
+	modes[A_11N].rates = &rates[4];
+	modes[A_11N].num_rates = 9; /* just OFDM */
+	modes[A_11N].num_channels = 0;
+#endif

What's that? Poor man's attempt to support 802.11n with a mac80211 that
has no idea what 802.11n is?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 190 bytes --]