On Tue, Jun 15, 2004 at 09:17:48PM +0200, Margit Schubert-While wrote:
> As an experiment, I have just tried merging all source code for a
> GUPD (Grand Unified Prism54 Driver) :-)
> prism54.h - 1063 lines
> prism54.c - 5583 lines
> With everything defined static (data and functions), it works and
> saves a chunk of generated code.
>
> Is this something we should be looking at ?
>
> Some points :
> With this setup, we can move the driver up into net/wireless.
> We get rid of the horrible prism include dependencies.
> (All structures/defs are available to all routines)
> Everything gets defined static allowing the compiler to produce
> compact code. This, though, means that kernel stack traces are
> less informative. (But, of course, we don't get oops'es in prism54
> do we ? :-) )
>
> OK Discuss.
I like the idea, I think we should just time the change appropriately to
alllow current work out there to get included / branched in.
---
What's left / being worked on (near future)?
* WPA - PSK - I'm working on this, hope to have a patch this week for cvs
testing.
* WDS - Umm someone on the list had mentioned they had used WDS
successfully and had some patches (?) Are you still alive? Are you
still working on this?
* ethtool support - Basic ethtool support is easy althought this can
wait
* Move verbose/debug printing to use netif_msg instead of our nasty ifdefs.
No one is working on this yet though so it can wait.
---
Anyone else working on anything for prism54 and would like us to wait
for you before merging all into the proposed GUPD?
Anyone disagree with this approach? Comments?
PS: please CC netdev
Luis
--
GnuPG Key fingerprint = 113F B290 C6D2 0251 4D84 A34A 6ADD 4937 E20A 525E
pgp8VWHpqMxAo.pgp
Description: PGP signature
|