Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Michael Wu]

[-- Attachment #1.1: Type: text/plain, Size: 525 bytes --]

Alright, I've replaced all + lines with spaces with tabs.

I also fixed one long line. The rest of them are nearly impossible to shorten 
well. The "(fc & IEEE80211_FCTL_FTYPE) == IEEE80211_FTYPE_DATA" style is 
really killing us (unless you want to break after ==, which is rather bad). I 
think we should switch back to a macro for that.

I would prefer if we could get this merged soon and put in that line 
shortening macro later (or whatever solution that's best), but it's your 
call.

Thanks,
-Michael Wu

[-- Attachment #1.2: switch-d80211-to-new-names.bz2 --]
[-- Type: application/x-bzip2, Size: 13938 bytes --]

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[John W. Linville]

On Thu, Jul 27, 2006 at 12:37:14AM -0700, Michael Wu wrote:
> Alright, I've replaced all + lines with spaces with tabs.
> 
> I also fixed one long line. The rest of them are nearly impossible to shorten 
> well. The "(fc & IEEE80211_FCTL_FTYPE) == IEEE80211_FTYPE_DATA" style is 
> really killing us (unless you want to break after ==, which is rather bad). I 
> think we should switch back to a macro for that.
> 
> I would prefer if we could get this merged soon and put in that line 
> shortening macro later (or whatever solution that's best), but it's your 
> call.

Michael,

As usual I'll depend on Jiri to merge d80211 stack patches, then
send me a pull request.  If I apply your "Switch drivers to d80211"
series now, that will undoutedly cause a breakage when Jiri asks me
to pull this later.

I presume that at least parts of those patches will still be necessary
or desirable after the d80211 symbol rename gets merged.  Are you
preparing a new patch series for when that happens?

Thanks,

John
-- 
John W. Linville
linville@tuxdriver.com

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Michael Wu]

[-- Attachment #1: Type: text/plain, Size: 966 bytes --]

On Monday 31 July 2006 13:31, John W. Linville wrote:
> As usual I'll depend on Jiri to merge d80211 stack patches, then
> send me a pull request.  If I apply your "Switch drivers to d80211"
> series now, that will undoutedly cause a breakage when Jiri asks me
> to pull this later.
>
Yeah, there needs to be a new (and smaller) set of patches to switch drivers 
to the d80211.h header.

> I presume that at least parts of those patches will still be necessary
> or desirable after the d80211 symbol rename gets merged.  Are you
> preparing a new patch series for when that happens?
>
I'm not quite sure whether switching d80211_mgmt.h will be worthwhile. If it 
isn't, then I can get to preparing a new patch series. At any rate, I think 
the most important thing right now is fixing the conflicts w/ wireless-dev in 
the patch that updates the d80211 drivers to use the new names so we can get 
the rename stuff into wireless-dev.

-Michael Wu

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Christoph Hellwig]

On Mon, Jul 31, 2006 at 01:51:31PM -0700, Michael Wu wrote:
> On Monday 31 July 2006 13:31, John W. Linville wrote:
> > As usual I'll depend on Jiri to merge d80211 stack patches, then
> > send me a pull request.  If I apply your "Switch drivers to d80211"
> > series now, that will undoutedly cause a breakage when Jiri asks me
> > to pull this later.
> >
> Yeah, there needs to be a new (and smaller) set of patches to switch drivers 
> to the d80211.h header.

NACK again.  Driver should continue to use the ieee80211.h header forever.

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[John W. Linville]

On Tue, Aug 01, 2006 at 10:56:50AM +0100, Christoph Hellwig wrote:
> On Mon, Jul 31, 2006 at 01:51:31PM -0700, Michael Wu wrote:
> > On Monday 31 July 2006 13:31, John W. Linville wrote:
> > > As usual I'll depend on Jiri to merge d80211 stack patches, then
> > > send me a pull request.  If I apply your "Switch drivers to d80211"
> > > series now, that will undoutedly cause a breakage when Jiri asks me
> > > to pull this later.
> > >
> > Yeah, there needs to be a new (and smaller) set of patches to switch drivers 
> > to the d80211.h header.
> 
> NACK again.  Driver should continue to use the ieee80211.h header forever.

I don't anticipate the d80211 naming conventions to ever make it out
of wireless-dev.  By the time we are ready to push that stuff upstream,
we will have cleaned-up our messes.

This does raise the question: Should we start taking patches to
wireless-dev that migrate the current (i.e. ieee80211/softmac) stack
out of the kernel?  This would include (re-)moving the current stack
code, pointing non-migrated drivers (ipw2[12]00, zd1211rw) at the old
code, moving drivers out of drivers/net/wireless/d80211 up a level,
removing the softmac-based version of the bcm43xx driver, etc.

Are we ready for this?  Who wants to be the wireless janitor?

Whether Michael's patches come before or after this clean-up
really doesn't matter.  I'd probably rather have them now.  It only
complicates the migration slightly, while accomplishing something
useful.

John
-- 
John W. Linville
linville@tuxdriver.com

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Jiri Benc]

> On Tue, Aug 01, 2006 at 10:56:50AM +0100, Christoph Hellwig wrote:
> > NACK again.  Driver should continue to use the ieee80211.h header forever.

When a patch that renames constants in d80211 is merged and d80211 stack
is moved to ieee80211/ directory, there will be only slight changes (if
any) needed for most drivers. So yes, fullmac drivers will continue to
use ieee80211.h, constants and structures they use will be still the
same, but the header itself will be completely different.

One notable exception is ipw2100 and ipw2200 drivers, but that's another
story.

On Tue, 1 Aug 2006 08:21:49 -0400, John W. Linville wrote:
> This does raise the question: Should we start taking patches to
> wireless-dev that migrate the current (i.e. ieee80211/softmac) stack
> out of the kernel?

I think we should, at least some of them.

> This would include (re-)moving the current stack
> code,

This can be done easily just before merging, no reason for one more
breakage of everyone's drivers now. Furthermore, if we do this now, it
will be more difficult to track Linus' tree.

> pointing non-migrated drivers (ipw2[12]00, zd1211rw) at the old
> code,

Yes. Rather than moving, zd1211 should be ported to d80211 - this will
also allow using of more advanced features of the hw.

> moving drivers out of drivers/net/wireless/d80211 up a level,

I think they should be moved along with the stack - i.e. just before
merging.

> removing the softmac-based version of the bcm43xx driver, etc.

Ditto.

> Whether Michael's patches come before or after this clean-up
> really doesn't matter.  I'd probably rather have them now.  It only
> complicates the migration slightly, while accomplishing something
> useful.

Do you have a plan when you will merge rt2x00 patches so I can apply
Michael's renaming patch(es) without risk of conflicts?

Thanks,

 Jiri

-- 
Jiri Benc
SUSE Labs

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[John W. Linville]

On Tue, Aug 01, 2006 at 03:58:37PM +0200, Jiri Benc wrote:

> Do you have a plan when you will merge rt2x00 patches so I can apply
> Michael's renaming patch(es) without risk of conflicts?

Working on it ~now.  I hit a snag in that Ivo's patches seem to rely on
his radio button patch, which I had ignored until now.  I'll probably
pull that in this morning (or figure-out how to not need it) and get
something pushed to wireless-dev sometime today.

John
-- 
John W. Linville
linville@tuxdriver.com

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Ivo Van Doorn]

Hi,

> > Do you have a plan when you will merge rt2x00 patches so I can apply
> > Michael's renaming patch(es) without risk of conflicts?
>
> Working on it ~now.  I hit a snag in that Ivo's patches seem to rely on
> his radio button patch, which I had ignored until now.  I'll probably
> pull that in this morning (or figure-out how to not need it) and get
> something pushed to wireless-dev sometime today.

Ehm, that should not have happened. The rt2x00 version I had send as patches
had the radio button integrated as they had been in wireless-dev already.
The patch to convert it to the rfkill driver is one of the patches
I'll send later.

Ivo

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[John W. Linville]

On Tue, Aug 01, 2006 at 04:25:05PM +0200, Ivo Van Doorn wrote:
> Hi,
> 
> >> Do you have a plan when you will merge rt2x00 patches so I can apply
> >> Michael's renaming patch(es) without risk of conflicts?
> >
> >Working on it ~now.  I hit a snag in that Ivo's patches seem to rely on
> >his radio button patch, which I had ignored until now.  I'll probably
> >pull that in this morning (or figure-out how to not need it) and get
> >something pushed to wireless-dev sometime today.
> 
> Ehm, that should not have happened. The rt2x00 version I had send as patches
> had the radio button integrated as they had been in wireless-dev already.
> The patch to convert it to the rfkill driver is one of the patches
> I'll send later.

The rfkill stuff seems to be behind a CONFIG_RT2X00_BUTTON define,
which keys off some driver-specific Kconfig stuff.  They were
getting turned-on by default w/ 'allmodconfig' and 'allyesconfig'.
Turning them off in .config lets them compile.

In general, I'd prefer not to have Kconfig-settable options that
don't compile.  But in this case I guess we'll just live with it.
After all, it is a development tree... :-)

John
-- 
John W. Linville
linville@tuxdriver.com

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Ivo van Doorn]

On Tuesday 01 August 2006 19:11, John W. Linville wrote:
> On Tue, Aug 01, 2006 at 04:25:05PM +0200, Ivo Van Doorn wrote:
> > Hi,
> > 
> > >> Do you have a plan when you will merge rt2x00 patches so I can apply
> > >> Michael's renaming patch(es) without risk of conflicts?
> > >
> > >Working on it ~now.  I hit a snag in that Ivo's patches seem to rely on
> > >his radio button patch, which I had ignored until now.  I'll probably
> > >pull that in this morning (or figure-out how to not need it) and get
> > >something pushed to wireless-dev sometime today.
> > 
> > Ehm, that should not have happened. The rt2x00 version I had send as patches
> > had the radio button integrated as they had been in wireless-dev already.
> > The patch to convert it to the rfkill driver is one of the patches
> > I'll send later.
> 
> The rfkill stuff seems to be behind a CONFIG_RT2X00_BUTTON define,
> which keys off some driver-specific Kconfig stuff.  They were
> getting turned-on by default w/ 'allmodconfig' and 'allyesconfig'.
> Turning them off in .config lets them compile.

Ah, I have spotted the problem. That was indeed accidently added during the last
patch series. It was part of the patch I was keeping behind.
I'll send the rfkill patch as soon as possible,  that should fix that problem,
in case the rfkill is considered not good enough for inclusion yet, I'll create a
patch to fix this issue correctly.

> In general, I'd prefer not to have Kconfig-settable options that
> don't compile.  But in this case I guess we'll just live with it.
> After all, it is a development tree... :-)

Hehe, true. But I'll do my best to fix this issue soon. :)

Ivo

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Ulrich Kunitz]

On 06-08-01 15:58 Jiri Benc wrote:

> > pointing non-migrated drivers (ipw2[12]00, zd1211rw) at the old
> > code,
> 
> Yes. Rather than moving, zd1211 should be ported to d80211 - this will
> also allow using of more advanced features of the hw.

I have currently no idea, when this will happen. Currently we are
still working on the basic plumbing of the driver.

However I would support the descape-preparing clean-up, if
pointless renaming is minimized. In the ideal case only header
includes should be changed. I would support a split between
protocol related headers and stack-related stuff.

Cheers,

Uli

-- 
Uli Kunitz

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Christoph Hellwig]

On Tue, Aug 01, 2006 at 08:21:49AM -0400, John W. Linville wrote:
> I don't anticipate the d80211 naming conventions to ever make it out
> of wireless-dev.  By the time we are ready to push that stuff upstream,
> we will have cleaned-up our messes.

Yes, absolutely.

> This does raise the question: Should we start taking patches to
> wireless-dev that migrate the current (i.e. ieee80211/softmac) stack
> out of the kernel?  This would include (re-)moving the current stack
> code, pointing non-migrated drivers (ipw2[12]00, zd1211rw) at the old
> code, moving drivers out of drivers/net/wireless/d80211 up a level,
> removing the softmac-based version of the bcm43xx driver, etc.

I think you first need to come up with a way to deal with ipw2[12]00,
dealing with that kind of devices is priority number one before the
devicescape stack can go anywhere.  Once that is handled all software
MAC drivers in wireless-dev should be migrated to the device-scape based
stack.

After that we're in the right position to deal with last big step of
the merge:

 a) deal with the userspace interface issue.  Either rever to the single
    network device and eth%d name interface (at least for the non-AP case)
    so existing userspace won't know about the name underlying stack at
    all or alternatively go through the painfull design process for a new
    userspace interface.  I think the first variant will be a lot easier..
 b) do all the small compare and review work.  See how different the interface
    for software mac devices are for softmac vs dscape.  See if any of the
    old code should be reintegrated into dcape, port forward dscape API
    changes to mainline to make merging more gradually.  In the best case
    we'd migrate the current mainline code fully to the finally dscape
    revisited API so the new code really is a drop in, but I don't expect
    that happen, the concepts are just too different.  At least we should
    try to make the interface diff as small a possible.

Once we're there I'm happy to help out on that step b) as much as I can.

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Michael Buesch]

On Saturday 05 August 2006 13:46, Christoph Hellwig wrote:
> On Tue, Aug 01, 2006 at 08:21:49AM -0400, John W. Linville wrote:
> > I don't anticipate the d80211 naming conventions to ever make it out
> > of wireless-dev.  By the time we are ready to push that stuff upstream,
> > we will have cleaned-up our messes.
> 
> Yes, absolutely.
> 
> > This does raise the question: Should we start taking patches to
> > wireless-dev that migrate the current (i.e. ieee80211/softmac) stack
> > out of the kernel?  This would include (re-)moving the current stack
> > code, pointing non-migrated drivers (ipw2[12]00, zd1211rw) at the old
> > code, moving drivers out of drivers/net/wireless/d80211 up a level,
> > removing the softmac-based version of the bcm43xx driver, etc.

Yes, I personally would like to see that happen now.

> I think you first need to come up with a way to deal with ipw2[12]00,
> dealing with that kind of devices is priority number one before the
> devicescape stack can go anywhere.  Once that is handled all software
> MAC drivers in wireless-dev should be migrated to the device-scape based
> stack.

Well, I think we all agreed on what to do with ipw at the wireless summit.
As far as I remember all agreed to move the softmac-only drivers to their
own directory. Simply the reverse of what is currently in wireless-dev.
Currently we have d80211 drivers in their own dir.
After that move was done, we port softmac drivers over to d80211
and delete the softmac version once it's stabilized. If no softmac
driver is left, we delete the old ieee80211 subsystem.

> After that we're in the right position to deal with last big step of
> the merge:
> 
>  a) deal with the userspace interface issue.  Either rever to the single
>     network device and eth%d name interface (at least for the non-AP case)
>     so existing userspace won't know about the name underlying stack at
>     all or alternatively go through the painfull design process for a new
>     userspace interface.  I think the first variant will be a lot easier..

Not in the long term, IMHO.

-- 
Greetings Michael.

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Michael Buesch]

On Saturday 05 August 2006 19:30, Michael Buesch wrote:
> On Saturday 05 August 2006 13:46, Christoph Hellwig wrote:
> > On Tue, Aug 01, 2006 at 08:21:49AM -0400, John W. Linville wrote:
> > > I don't anticipate the d80211 naming conventions to ever make it out
> > > of wireless-dev.  By the time we are ready to push that stuff upstream,
> > > we will have cleaned-up our messes.
> > 
> > Yes, absolutely.
> > 
> > > This does raise the question: Should we start taking patches to
> > > wireless-dev that migrate the current (i.e. ieee80211/softmac) stack
> > > out of the kernel?  This would include (re-)moving the current stack
> > > code, pointing non-migrated drivers (ipw2[12]00, zd1211rw) at the old
> > > code, moving drivers out of drivers/net/wireless/d80211 up a level,
> > > removing the softmac-based version of the bcm43xx driver, etc.
> 
> Yes, I personally would like to see that happen now.

I forgot to say why ;)

It's getting a major pain to support two up-to-date bcm43xx drivers.
It's getting nontrivial to port drivers over from bcm43xx-softmac
to bcm43xx-d80211; even with tools like quilt and wiggle.

What I'd like to see is: After 2.6.18 got out, we freeze bcm43xx-softmac
and only allow small bugfixes to it. All development work would go
into the d80211 branch.
That saves me a _lot_ of time and pain.

-- 
Greetings Michael.

[PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Michael Wu]

[-- Attachment #1.1: Type: text/plain, Size: 234 bytes --]

Big patches are fun.

These two attached patches also roll in a switch to MAC_FMT and MAC_ARG, and 
use IEEE80211_DATA_LEN when possible. Otherwise, it's the same as before 
except split into two patches.

Thanks,
-Michael Wu

[-- Attachment #1.2: switch-d80211-to-new-names.bz2 --]
[-- Type: application/x-bzip2, Size: 14981 bytes --]

[-- Attachment #1.3: switch-drivers-to-new-names.bz2 --]
[-- Type: application/x-bzip2, Size: 2688 bytes --]

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Jiri Benc]

On Mon, 24 Jul 2006 23:02:50 -0700, Michael Wu wrote:
> These two attached patches also roll in a switch to MAC_FMT and MAC_ARG, and 
> use IEEE80211_DATA_LEN when possible. Otherwise, it's the same as before 
> except split into two patches.

Nice, thanks a lot.

One more thing (hopefully the last one), could you try to keep lines
under 80 characters? I understand that with such long identifiers it is
not always possible (it dramatically reduces readability of the code if
you try to force the limit in every case) but there are some places
where it is feasible.

Also, when changing a line that is indented with spaces instead of tab
characters, it's nice to convert that spaces to tabs. That's not a
requirement, though; I will modify the patch if you won't do it.

Thanks,

 Jiri

-- 
Jiri Benc
SUSE Labs

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Michael Wu]

[-- Attachment #1: Type: text/plain, Size: 841 bytes --]

On Wednesday 26 July 2006 09:32, Jiri Benc wrote:
> One more thing (hopefully the last one), could you try to keep lines
> under 80 characters? I understand that with such long identifiers it is
> not always possible (it dramatically reduces readability of the code if
> you try to force the limit in every case) but there are some places
> where it is feasible.
>
Ah, yeah, that was a problem..

> Also, when changing a line that is indented with spaces instead of tab
> characters, it's nice to convert that spaces to tabs. That's not a
> requirement, though; I will modify the patch if you won't do it.
>
I still want to convert d80211_mgmt.h. Once that's done, I'll look through for 
those tabs/spaces & long lines formatting issues. I won't mind if you start 
fixing some of those issues now, however. :)

-Michael Wu

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Jiri Benc]

On Wed, 26 Jul 2006 09:43:55 -0700, Michael Wu wrote:
> I still want to convert d80211_mgmt.h. Once that's done, I'll look through for 
> those tabs/spaces & long lines formatting issues. I won't mind if you start 
> fixing some of those issues now, however. :)

I just meant running some s/ {8}/\t/ on your patch :-)

Also, I don't want to fix all of tab issues in d80211 by one big patch.
I always have several patches pending (and other people probably too)
and it leads to unnecessary rejects when applying them. Instead, I
gradually correct them when changing corresponding lines.

Thanks,

 Jiri

-- 
Jiri Benc
SUSE Labs

Re: [PATCH dscape] d80211: Switch d80211.h to IEEE80211_ style names

[Michael Wu]

[-- Attachment #1: Type: text/plain, Size: 534 bytes --]

On Wednesday 26 July 2006 09:53, Jiri Benc wrote:
> I just meant running some s/ {8}/\t/ on your patch :-)
>
> Also, I don't want to fix all of tab issues in d80211 by one big patch.
> I always have several patches pending (and other people probably too)
> and it leads to unnecessary rejects when applying them. Instead, I
> gradually correct them when changing corresponding lines.
>
Ah. Well, if you don't mind doing the space->tabs fix, please do. I don't know 
if I can get to that soon enough.

Thanks,
-Michael Wu

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]