A new driver is going to be released

A new driver is going to be released

[Dmitry Tarnyagin]

Hi,

I'm about to release cw1200 driver to the community (more or less
latest code is available here:
http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=shortlog;h=refs/heads/linux-3.0-ux500
as well).

Technical question: how is it usually done in terms of commit history?
Should I preserve the history and release all the changes in the
driver separately or should I squash everything to a single commit
or...?

Thank you and with best regards,
Dmitry

Re: A new driver is going to be released

[Larry Finger]

On 11/03/2011 05:40 PM, Dmitry Tarnyagin wrote:
> Hi,
>
> I'm about to release cw1200 driver to the community (more or less
> latest code is available here:
> http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=shortlog;h=refs/heads/linux-3.0-ux500
> as well).
>
> Technical question: how is it usually done in terms of commit history?
> Should I preserve the history and release all the changes in the
> driver separately or should I squash everything to a single commit
> or...?

The details of what it took to get your driver to this point are not going to be 
of much interest in the future. The "official" history starts with acceptance 
into the kernel. Two things to keep in mind: (1) the kernel source must compile 
at each step of the way to allow for bisection, thus any changes to Kconfig and 
Makefile must be in the last patch, and (2) each commit should be small enough 
that it is relatively easy to review.

As I do not know the complexity of the driver, it is a little difficult to make 
suggestions; however, one way to accomplish the above is to add each source file 
in a separate patch. The maintainer may submit the final version as a single 
commit, but that is their choice.

Your sending this mail to the wireless ML indicates that you plan to submit the 
driver to the drivers/net/wireless/ part of the source tree. As long as the 
driver employs mac80211, that is appropriate. The other option is 
drivers/staging/. Drivers for that section of the source tree usually are not 
held to standards as high as the wireless section.

Please be certain that all code submitted passes all tests cleanly including 
checkpatch, sparse including endianess, and smatch.

Good luck,

Larry

Re: A new driver is going to be released

[Dan Williams]

On Thu, 2011-11-03 at 20:09 -0500, Larry Finger wrote:
> On 11/03/2011 05:40 PM, Dmitry Tarnyagin wrote:
> > Hi,
> >
> > I'm about to release cw1200 driver to the community (more or less
> > latest code is available here:
> > http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=shortlog;h=refs/heads/linux-3.0-ux500
> > as well).
> >
> > Technical question: how is it usually done in terms of commit history?
> > Should I preserve the history and release all the changes in the
> > driver separately or should I squash everything to a single commit
> > or...?
> 
> The details of what it took to get your driver to this point are not going to be 
> of much interest in the future. The "official" history starts with acceptance 
> into the kernel. Two things to keep in mind: (1) the kernel source must compile 
> at each step of the way to allow for bisection, thus any changes to Kconfig and 
> Makefile must be in the last patch, and (2) each commit should be small enough 
> that it is relatively easy to review.
> 
> As I do not know the complexity of the driver, it is a little difficult to make 
> suggestions; however, one way to accomplish the above is to add each source file 
> in a separate patch. The maintainer may submit the final version as a single 
> commit, but that is their choice.
> 
> Your sending this mail to the wireless ML indicates that you plan to submit the 
> driver to the drivers/net/wireless/ part of the source tree. As long as the 
> driver employs mac80211, that is appropriate. The other option is 
> drivers/staging/. Drivers for that section of the source tree usually are not 
> held to standards as high as the wireless section.

It looks like it's mac80211-based and already in drivers/staging/
according to the git tree Dmitry pointed to.

Dan

Re: A new driver is going to be released

[Dmitry Tarnyagin]

On Fri, Nov 4, 2011 at 3:01 AM, Dan Williams <dcbw@redhat.com> wrote:
> It looks like it's mac80211-based and already in drivers/staging/
> according to the git tree Dmitry pointed to.
>
Correct. But I would move it to drivers/net/wireless during submission.

Re: A new driver is going to be released

[Kalle Valo]

Dmitry Tarnyagin <abi.dmitryt@gmail.com> writes:

> I'm about to release cw1200 driver to the community

Nice!

> (more or less latest code is available here:
> http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=shortlog;h=refs/heads/linux-3.0-ux500
> as well).

I can't access it. I get "404 - Reading tree failed" when I try to
access the tree through gitweb:

http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=tree;h=%5Crefs/heads/linux-3.0-ux500;hb=%5Crefs/heads/linux-3.0-ux500

-- 
Kalle Valo

Re: A new driver is going to be released

[Johannes Berg]

On Fri, 2011-11-04 at 06:32 +0100, Dmitry Tarnyagin wrote:
> On Fri, Nov 4, 2011 at 3:01 AM, Dan Williams <dcbw@redhat.com> wrote:
> > It looks like it's mac80211-based and already in drivers/staging/
> > according to the git tree Dmitry pointed to.
> >
> Correct. But I would move it to drivers/net/wireless during submission.

A word of caution: don't attempt that in its current state.

johannes

Re: A new driver is going to be released

[Dmitry Tarnyagin]

Hi Johannes,

>> Correct. But I would move it to drivers/net/wireless during submission.
>
> A word of caution: don't attempt that in its current state.
>
Thank you for caution. CONFIG_CW1200_USE_STE_EXTENSIONS,
CONFIG_WAKELOCK and HAS_PUT_TASK_STRUCT will be removed
of course, API will be aligned with 3.1. But I'm a bit curious what else could
prevent from submission to drivers/net/wireless? I do not see obvious problems
with code. Could you please give a short advise what should be fixed then if you
see problems?

Thank you and with best regards,
Dmitry

Re: A new driver is going to be released

[Vitaly Wool]

Hi Kalle,

On Fri, Nov 4, 2011 at 9:06 AM, Kalle Valo <kvalo@adurom.com> wrote:
>> (more or less latest code is available here:
>> http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=shortlog;h=refs/heads/linux-3.0-ux500
>> as well).
>
> I can't access it. I get "404 - Reading tree failed" when I try to
> access the tree through gitweb:
>
> http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=tree;h=%5Crefs/heads/linux-3.0-ux500;hb=%5Crefs/heads/linux-3.0-ux500

I have a feeling that your mailer screws up the links because I can
access the tree using the one above but not the one that you get.

Thanks,
   Vitaly

Re: A new driver is going to be released

[Kalle Valo]

Vitaly Wool <vitalywool@gmail.com> writes:

> On Fri, Nov 4, 2011 at 9:06 AM, Kalle Valo <kvalo@adurom.com> wrote:
>>> (more or less latest code is available here:
>>> http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=shortlog;h=refs/heads/linux-3.0-ux500
>>> as well).
>>
>> I can't access it. I get "404 - Reading tree failed" when I try to
>> access the tree through gitweb:
>>
>> http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=tree;h=%5Crefs/heads/linux-3.0-ux500;hb=%5Crefs/heads/linux-3.0-ux500
>
> I have a feeling that your mailer screws up the links because I can
> access the tree using the one above but not the one that you get.

That was it, there was one extra '\'. Should have figured out that
myself, thanks.

-- 
Kalle Valo

Re: A new driver is going to be released

[Johannes Berg]

On Fri, 2011-11-04 at 10:10 +0100, Dmitry Tarnyagin wrote:
> Hi Johannes,
> 
> >> Correct. But I would move it to drivers/net/wireless during submission.
> >
> > A word of caution: don't attempt that in its current state.
> >
> Thank you for caution. CONFIG_CW1200_USE_STE_EXTENSIONS,
> CONFIG_WAKELOCK and HAS_PUT_TASK_STRUCT will be removed
> of course, API will be aligned with 3.1. But I'm a bit curious what else could
> prevent from submission to drivers/net/wireless? I do not see obvious problems
> with code. Could you please give a short advise what should be fixed then if you
> see problems?

Well there is a ton of stuff like compat code too. Also, for example the
worker thread is pretty strange -- kthread_create() is essentially
deprecated, and having a thread to handle interrupts is weird to start
with.

johannes

Re: A new driver is going to be released

[Larry Finger]

On 11/03/2011 09:01 PM, Dan Williams wrote:

> It looks like it's mac80211-based and already in drivers/staging/
> according to the git tree Dmitry pointed to.

A staging driver in a private tree is not the same as one in GregKH's staging 
tree. As this cw1200 driver has been seen by a limited number of developers, it 
needs to be submitted to staging in mainline before an attempt is made to 
convert it to the regular tree.

Larry

Re: A new driver is going to be released

[Anand Jain]

Dmitry Tarnyagin <abi.dmitryt@...> writes:

> 
> Hi,
> 
> I'm about to release cw1200 driver to the community (more or less
> latest code is available here:
>
http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=shortlog;h=refs/heads/linux-3.0-ux500
> as well).
> 
> Technical question: how is it usually done in terms of commit history?
> Should I preserve the history and release all the changes in the
> driver separately or should I squash everything to a single commit
> or...?
> 
> Thank you and with best regards,
> Dmitry
> --
> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
> the body of a message to majordomo@...
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> 

Hi,
 we are exploring the cw1200 driver and found in TODO that IBSS is not
implemented. So can you guide us if we can implement/enhance the same for
Snowball board in ICS code.

thanks
Anand