Prism54 in 2.6.4-bk2

Prism54 in 2.6.4-bk2

[Luis R. Rodriguez]

[-- Attachment #1: Type: text/plain, Size: 2219 bytes --]

On Sat, Mar 13, 2004 at 03:30:58PM -0500, Luis R. Rodriguez wrote:
> On Sat, Mar 13, 2004 at 05:39:32PM +0000, Andr? Ventura Lemos wrote:
> > ChangeSet@1.1608.81.1, 2004-03-12 12:55:33-05:00, jgarzik@redhat.com
> >   [wireless] Add new Prism54 wireless driver.
> > 
> > :-)
> > 
> > On Sat, 2004-03-13 at 17:08, Margit Schubert-While wrote:
> > > In case nobody noticed, the driver is in 2.6.4-bk2 !
> > > 
> > > Margit
> 
> Hmm. Now what? Should we create a new set of automated patchsets against the latest kernel
> snapshot? Or just not worry about that and keep on happilly with our stable kernel
> patches?
> 
> The prism54 driver snapshot that went into bk2 could be marked as our
> first testing snapshot... If so, we could also start a new ChangeLog.
> 
> Ideas?
> 
> 	Luis

I just checked out the bk2 snapshot and its diffs with our latest
prism54 tree. Here is the diff:

http://prism54.org/~mcgrof/bk2-up.diff

It seems jgarzik took a patch before 2004-03-09 changes and just nuked
WDS code. It's great that it's already on some official kernel snapshot 
patch but prism54 driver project is a very active project and I think
someone jumped the gun. We were preparing the driver for proper
integration per the nevdev list discussion. Anyway, now onto dealing
with it.

Regarding WDS on prism54: on the netdev list we discussed this
but no one got back to me as to whether we should really just nuke this
code. Prism54 driver source *does* include WDS support because hey, the
firmware does. Why wouldn't it go in the driver? We haven't given WDS
much though anyway since it's also been low priority on our TODO list.

Is it already agreed among kdevelopers that even if certain chipsets
support WDS at the hardware/firmware level that another layer is going
to be used for it's support?

Now that prism54 driver is in bk2 if we (prism54 team) want to submit
patches for it, should we always patch against the latest bk snapshot?

Oh and if you get messages telling you your post is pending approval on
the prism54 list because you're not registered, don't worry, we're quick.

-- 
GnuPG Key fingerprint = 113F B290 C6D2 0251 4D84  A34A 6ADD 4937 E20A 525E

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

Re: Prism54 in 2.6.4-bk2

[Jeff Garzik]

Luis R. Rodriguez wrote:
> Regarding WDS on prism54: on the netdev list we discussed this
> but no one got back to me as to whether we should really just nuke this
> code. Prism54 driver source *does* include WDS support because hey, the
> firmware does. Why wouldn't it go in the driver? We haven't given WDS
> much though anyway since it's also been low priority on our TODO list.

The WDS code was dead code as merged.

If you actually use it, I don't mind adding it :)

	Jeff

Re: [Prism54-devel] Re: Prism54 in 2.6.4-bk2

[Luis R. Rodriguez]

On Tue, Mar 16, 2004 at 01:14:35AM -0500, Jeff Garzik wrote:
> Luis R. Rodriguez wrote:
> >Regarding WDS on prism54: on the netdev list we discussed this
> >but no one got back to me as to whether we should really just nuke this
> >code. Prism54 driver source *does* include WDS support because hey, the
> >firmware does. Why wouldn't it go in the driver? We haven't given WDS
> >much though anyway since it's also been low priority on our TODO list.
> 
> The WDS code was dead code as merged.
> 
> If you actually use it, I don't mind adding it :)

I don't know of anybody who uses it. We did consider to drop it but we
just never got around to deciding what we were going to do about it. I
know it's there and it's *supposed* to work. 

Can we get back to you on that?  :)  It is just code that *is*
driver/hardware specific.

Actually can I just send you a patch for 2.6 for the latest 2.6 tree to
match ours? That is, rm -rf prism54/ as is and add our latest patch ?
It'd save a lot of work on our end.

We'll still do the 2.6 / 2.4 split and clean the 2.6 code of 2.4 code,
but can you just give us the opportunity to do it from within our tree?

It'd cure many headaches (on our end at least).

	Luis

-- 
GnuPG Key fingerprint = 113F B290 C6D2 0251 4D84  A34A 6ADD 4937 E20A 525E

Re: [Prism54-devel] Re: Prism54 in 2.6.4-bk2

[Jeff Garzik]

Luis R. Rodriguez wrote:
> On Tue, Mar 16, 2004 at 01:14:35AM -0500, Jeff Garzik wrote:
> 
>>Luis R. Rodriguez wrote:
>>
>>>Regarding WDS on prism54: on the netdev list we discussed this
>>>but no one got back to me as to whether we should really just nuke this
>>>code. Prism54 driver source *does* include WDS support because hey, the
>>>firmware does. Why wouldn't it go in the driver? We haven't given WDS
>>>much though anyway since it's also been low priority on our TODO list.
>>
>>The WDS code was dead code as merged.
>>
>>If you actually use it, I don't mind adding it :)
> 
> 
> I don't know of anybody who uses it. We did consider to drop it but we
> just never got around to deciding what we were going to do about it. I
> know it's there and it's *supposed* to work. 
> 
> Can we get back to you on that?  :)  It is just code that *is*
> driver/hardware specific.

For code is that (a) experimental, (b) for pre-production hardware, or 
(c) rarely if ever used, we would prefer to not merge it at all.

When I see stuff like "TODO: actually give this some thought" and "I 
don't know anybody who uses it", that means it doesn't need to be merged 
in the upstream tree :)


> Actually can I just send you a patch for 2.6 for the latest 2.6 tree to
> match ours? That is, rm -rf prism54/ as is and add our latest patch ?
> It'd save a lot of work on our end.

It depends on how big the patch is, and whether or not it adds code that 
nobody but the dev team uses, etc...  I don't want to add the WDS code, 
since nobody uses it...  and adding the #ifdefs I removed would not be 
desired either.  Those #ifdefs aren't need in the upstream tree.  I plan 
to remove them from other upstream drivers, too.

WRT submitting patches...   send away.  drivers/net patches should go -> 
Jean T -> jgarzik+netdev or simply -> jgarzik+netdev, your choice.  In 
general "50 small patches are better than 1 big patch".  Large updates 
are not reviewable or easily testable.  Large patches tend to fix 20 
bugs, and add 5 new ones.

	Jeff