[PATCH] AP/master mode success with ZD1211LnxDrv_2_22_0_0 + 2.6.25.1

[Hin-Tak Leung-3]

subscriber-only bounced, 2nd try.

--- On Wed, 7/5/08, Hin-Tak Leung <htl10@...> wrote:

> Here is my ugly patch to make the vendor 2.22 driver work
> with kernel
> 2.6.25.1, with a ZD1211B-based USB stick, and working in
> access-point/master mode against a managed-mode Realtek
> client.
> - the whole point of porting the vendor driver to 2.6.25.1
> is the master
> mode, really. The stick was bought for a old laptop with
> only one USB 1.1
> port, so master mode has only been tested in my case in
> Pure-B (802.11b)
> mode. But the vendor driver seems to be quite alright and
> stable and fast,
> and I ran windows update on the never-been-used Vista
> partition of the
> realtek client and ip-masquaraded 70MB through the other
> side of zd1211.
>
> a little comment:
> - granted, the vendor driver code itself is ugly as hell,
> and have some
> obvious endianness issues and 64-bit issues - the
> realtek-equiped machine
> is 64-bit and it is quite obvious the vendor driver does
> not work there;
> and I don't believe the vendor driver works on
> big-endian arch either - but
> zd1211rw still doesn't work in master/ad-hoc mode,
> despite some words from
>  time to time that it is easy(!) in the new stack for some
> years now...
>
> - the wiki is inaccurate - the vendor-based driver has been
> abandoned for almost a year now, and the last vendor driver
> was releasd a bit before christmas; so the vendor driver is
> actually more up-to-date in tracking
> kernel changes than the vendor-based community driver. And
> the fact that
> zd1211rw still does not do any mode but managed is not
> given enough emphasis. e.g. The realtek driver actually
> stops one from setting illegal/unsupported iwconfig values,
> but the zd1211rw driver doesn't flag
> an error on iwconfig, but one has to issue another iwconfig
> to see that cell is invalid and there is still no
> indication that
> ad-hoc mode doesn't work unless one looks it up that
> cell-invalid is 'ad-hoc mode does not work'. This
> behavior is misleading.
>
> That said, I did use the zd1211rw driver on the 64-bit
> machine briefly and I know the code is clean and
> endian-aware, so thanks for that; and I'll try to tidy
> up the vendor v2.22-kernel-2.6.25.1 patch, and would
> probably
> write in again explaining why I did what I did to make it
> work.
> So on the caveat that the attached patch works for me on an
> old 32-bit intel pentium laptop with a usb 1.1 port running
> in master mode (plus
> using dnsmasq for serving clients), I hope you would put it
> up somewhere for downoad and hope that it would be useful to
> others. It is heavily
> based on a patch by Constantine Shulyupin for v2.15, so if
> he is listening,
> thanks.
>
>
>      
> ___________________________________________________________
>
> Yahoo! For Good. Give and get cool things for free, reduce
> waste and help our planet. Plus find hidden Yahoo! treasure
>
>
> http://green.yahoo.com/uk/earth-day/

      __________________________________________________________
Sent from Yahoo! Mail.
A Smarter Email http://uk.docs.yahoo.com/nowyoucan.html

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Zd1211-devs mailing list - http://zd1211.wiki.sourceforge.net/
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/zd1211-devs

[Hin-Tak Leung-3]

Sorry to reply to my own post - nobody seems to find my post interesting?

I have a somewhat interesting issue/extension to my previous patch: I need 802.11g
compatibility (basically because my client gets a bit confused and the client driver hasn't got the functionality to drop to 11b mode). The vendor driver automatically
decides to go into pure 11b mode if connected to USB 1.1 host. I think this is wrong -
it should stay in mixed mode regardless, but tell client to drop to B mode, if they can,
or limit the transfer rate to B mode from access point side.

Indeed, if I just comment out the USB 1.1 decision, the vendor driver starts in mixed mode and a 11g client can connect (and then the vendor driver limit the rate to b rate),
and I have useable network connectivity.


      __________________________________________________________
Sent from Yahoo! Mail.
A Smarter Email http://uk.docs.yahoo.com/nowyoucan.html

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Zd1211-devs mailing list - http://zd1211.wiki.sourceforge.net/
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/zd1211-devs

[jengstrom ()]

I find your posts very interesting.

Disclaimer: I'm not quite the kernel driver hacker (yet).

I wanted to use my zd1211(A) stick as a access point on my home office router/firewall.

Platform: i686
Kernel version: 2.6.18-r6 (patched, near hardened linux)

I tried the following, in order:
1. Got zd1211 sources from Sourceforge SVN. Compiled cleanly but failed during driver/hw initialization.
2. Downloaded zd1211 2.22.0.0 found on the net. Compiled dirtier and failed in the same manner.
3. Applied your patch and it seems to work fine!


Does anybody here know since which kernel version the 2.22.0.0 drivers were supposed to be broken?
(Would be weird if the driver was supposed to work without your patch for my kernel version.)


I also use USB 1.1, so the master now appears fixed to 802.11b mode.
I agree with your suggestion and rationale on leaving it in mixed mode.
So I'm trying that next.


Note: The stick is a Canyon WF518, older model (newer ones use Ralink RT73).

Thanks a lot!
/John

Hin-Tak Leung-3 wrote:

Sorry to reply to my own post - nobody seems to find my post interesting?

I have a somewhat interesting issue/extension to my previous patch: I need 802.11g
compatibility (basically because my client gets a bit confused and the client driver hasn't got the functionality to drop to 11b mode). The vendor driver automatically
decides to go into pure 11b mode if connected to USB 1.1 host. I think this is wrong -
it should stay in mixed mode regardless, but tell client to drop to B mode, if they can,
or limit the transfer rate to B mode from access point side.

Indeed, if I just comment out the USB 1.1 decision, the vendor driver starts in mixed mode and a 11g client can connect (and then the vendor driver limit the rate to b rate),
and I have useable network connectivity.


      __________________________________________________________
Sent from Yahoo! Mail.
A Smarter Email http://uk.docs.yahoo.com/nowyoucan.html

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Zd1211-devs mailing list - http://zd1211.wiki.sourceforge.net/
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/zd1211-devs