[v6,1/1] rtl8821au: new package

Adding a kernel module based USB wifi driver. There is no official
version of this driver that works properly on ARM and with newer USB
WiFi cards. This driver version builds module 8821au and is based on a
kernel module release from ASUS with various fixes integrated in.

Signed-off-by: Christian Stewart <christian@paral.in>
---

I've switched to use ulli-kroll's branch of rtl8821au. While it's based
on an older vendor version, it has binary firmware and around ~1800
commits of work to improve the driver. I think this work is probably
more valuable than the vendor's updates.

 package/Config.in                                  |  1 +
 package/rtl8821au/0001-disable-default-debug.patch | 24 ++++++++++++++++++++++
 package/rtl8821au/Config.in                        | 10 +++++++++
 package/rtl8821au/rtl8821au.hash                   |  1 +
 package/rtl8821au/rtl8821au.mk                     | 21 +++++++++++++++++++
 5 files changed, 57 insertions(+)
 create mode 100644 package/rtl8821au/0001-disable-default-debug.patch
 create mode 100644 package/rtl8821au/Config.in
 create mode 100644 package/rtl8821au/rtl8821au.hash
 create mode 100644 package/rtl8821au/rtl8821au.mk

Hi all,  

Note : I'm new on the list, so I had insert this tread by hand !

I'm the one working on the RTL8821AU driver.

On Thu, 23 Jul 2015, Christian Stewart wrote:

> Adding a kernel module based USB wifi driver. There is no official
> version of this driver that works properly on ARM and with newer USB
> WiFi cards. This driver version builds module 8821au and is based on a
> kernel module release from ASUS with various fixes integrated in.

This message will be commited please change.

The I think the difference between the version from ASUS and mine based
the one Edimax are in the handling of the LED.
Maybe there are more ...

> I've switched to use ulli-kroll's branch of rtl8821au. While it's based
> on an older vendor version, it has binary firmware and around ~1800
> commits of work to improve the driver. I think this work is probably
> more valuable than the vendor's updates.
> 

-> git send-email --compose
Will add some message an intreductory message

> --- /dev/null
> +++ b/package/rtl8821au/0001-disable-default-debug.patch

I do add some kind of "disable debug message"
This will take some time, I'm currently working in the code.

> +config BR2_PACKAGE_RTL8821AU
> +     bool "rtl8821au"
> +     depends on BR2_LINUX_KERNEL
> +     help
> +       rtl8821au wifi driver
> +
> +       https://github.com/paralin/rtl8821au

Please fix this URL !

Maybe add some line "out of tree driver"
And the USB3 issue I found inthe HW as you can read in
Readme.md

Greetings from Cologne
Hans Ulli Kroll

Hey Hans,

It's not unusual in Buildroot to have long commit messages when a package
is first added.

I've addressed your concerns and respun the patch.

Christian

On Fri, Jul 24, 2015 at 6:50 AM Hans Ulli Kroll <ulli.kroll@googlemail.com>
wrote:

> Hi all,
>
> Note : I'm new on the list, so I had insert this tread by hand !
>
> I'm the one working on the RTL8821AU driver.
>
> On Thu, 23 Jul 2015, Christian Stewart wrote:
>
> > Adding a kernel module based USB wifi driver. There is no official
> > version of this driver that works properly on ARM and with newer USB
> > WiFi cards. This driver version builds module 8821au and is based on a
> > kernel module release from ASUS with various fixes integrated in.
>
> This message will be commited please change.
>
> The I think the difference between the version from ASUS and mine based
> the one Edimax are in the handling of the LED.
> Maybe there are more ...
>
> > I've switched to use ulli-kroll's branch of rtl8821au. While it's based
> > on an older vendor version, it has binary firmware and around ~1800
> > commits of work to improve the driver. I think this work is probably
> > more valuable than the vendor's updates.
> >
>
> -> git send-email --compose
> Will add some message an intreductory message
>
> > --- /dev/null
> > +++ b/package/rtl8821au/0001-disable-default-debug.patch
>
> I do add some kind of "disable debug message"
> This will take some time, I'm currently working in the code.
>
> > +config BR2_PACKAGE_RTL8821AU
> > +     bool "rtl8821au"
> > +     depends on BR2_LINUX_KERNEL
> > +     help
> > +       rtl8821au wifi driver
> > +
> > +       https://github.com/paralin/rtl8821au
>
> Please fix this URL !
>
> Maybe add some line "out of tree driver"
> And the USB3 issue I found inthe HW as you can read in
> Readme.md
>
> Greetings from Cologne
> Hans Ulli Kroll
>

Ulli, all,

On 2015-07-24 15:50 +0200, Hans Ulli Kroll spake thusly:
> Note : I'm new on the list, so I had insert this tread by hand !

Welcome! :-)

> I'm the one working on the RTL8821AU driver.

Yeah, great! I saw that Christian and I did some work on the same
module, but based on diffferent code-dump by different vendors. It would
be great if you two could synchronise your work on this, to avoid
duplication and to get the best driver possible, with a ultimate goal of
submitting it for inclusion in the mainline. ;-)

> On Thu, 23 Jul 2015, Christian Stewart wrote:
> > Adding a kernel module based USB wifi driver. There is no official
> > version of this driver that works properly on ARM and with newer USB
> > WiFi cards. This driver version builds module 8821au and is based on a
> > kernel module release from ASUS with various fixes integrated in.
> 
> This message will be commited please change.

No, that's OK to get some details when adding a new package.

> The I think the difference between the version from ASUS and mine based
> the one Edimax are in the handling of the LED.
> Maybe there are more ...

Well, if that's the only difference, we can consider that to be a minor
issue that can be resolvdd by back-/forward-porting it from one code
base to the other.

> > I've switched to use ulli-kroll's branch of rtl8821au. While it's based
> > on an older vendor version, it has binary firmware and around ~1800
> > commits of work to improve the driver. I think this work is probably
> > more valuable than the vendor's updates.
> > 
> 
> -> git send-email --compose
> Will add some message an intreductory message

That's OK the way Christian did: that part isindeed not part of the
commit message, as it lise below the '---' line, so git wil ignore it
when the patch is applied.

> > --- /dev/null
> > +++ b/package/rtl8821au/0001-disable-default-debug.patch
> 
> I do add some kind of "disable debug message"
> This will take some time, I'm currently working in the code.

Great! :-)

> > +config BR2_PACKAGE_RTL8821AU
> > +     bool "rtl8821au"
> > +     depends on BR2_LINUX_KERNEL
> > +     help
> > +       rtl8821au wifi driver
> > +
> > +       https://github.com/paralin/rtl8821au
> 
> Please fix this URL !
> 
> Maybe add some line "out of tree driver"

Well, if it were an in-tree module, we would not have a package for it,
as it would come with the Linux source tree. So, this is obviously an
out-of-tree module. ;-)

> And the USB3 issue I found inthe HW as you can read in
> Readme.md

I see that Christian added that in the new commit log and help text.

Thanks! :-)

Regards,
Yann E. MORIN.

On Sat, 25 Jul 2015, Yann E. MORIN wrote:

> Ulli, all,
> 
> On 2015-07-24 15:50 +0200, Hans Ulli Kroll spake thusly:
> > Note : I'm new on the list, so I had insert this tread by hand !
> 
> Welcome! :-)
> 

Thanks

> > The I think the difference between the version from ASUS and mine based
> > the one Edimax are in the handling of the LED.
> > Maybe there are more ...
> 
> Well, if that's the only difference, we can consider that to be a minor
> issue that can be resolvdd by back-/forward-porting it from one code
> base to the other.
> 

This ways only a quick look at some code I'm working (at this time) on.
As I said there is maybe more.

> > > --- /dev/null
> > > +++ b/package/rtl8821au/0001-disable-default-debug.patch
> > 
> > I do add some kind of "disable debug message"
> > This will take some time, I'm currently working in the code.
> 
> Great! :-)
> 

So, now the driver is *allways* build with debug info, see
CONFIG_RTLWIFI_DEBUG
but you *must* enable this while loading the driver i.e.

insmod rtl8821au debug=5

To be very verbose, default is debug=0 for no debug output.

For the code itself :
I've done *a lot* functions renames and other things.
I use the rtl8821ae driver as some kind of pattern and the rtl8192cu 
driver for the USB part.

Ulli

Ulli,

On Sun, Jul 26, 2015 at 12:59 PM Hans Ulli Kroll <ulli.kroll@googlemail.com>
wrote:

[snip]

So, now the driver is *allways* build with debug info, see
> CONFIG_RTLWIFI_DEBUG
> but you *must* enable this while loading the driver i.e.
>
> insmod rtl8821au debug=5
>
> To be very verbose, default is debug=0 for no debug output.
>

That's a bit odd. When loading the driver as-is from your repository
without any arguments, I get a very verbose amount of logging to the kernel
logs. Are you sure this is the case? With the patch in the merged
repository most of the debugging is disabled through header define comments.


>
> For the code itself :
> I've done *a lot* functions renames and other things.
> I use the rtl8821ae driver as some kind of pattern and the rtl8192cu
> driver for the USB part.


Great!

~ Christian

On Sun, 26 Jul 2015, Christian Stewart wrote:
Hi,

I've done a quick test during my coffee break ...

> That's a bit odd. When loading the driver as-is from your repository
> without any arguments, I get a very verbose amount of logging to the kernel
> logs. Are you sure this is the case? With the patch in the merged
> repository most of the debugging is disabled through header define comments.
> 
> 

I missed some old dev_info() in rtl8821au/.
Now changed.
Pushed.

Ulli