ftp://ftp.wolfmountaingroup.org/pub/ndiswrapper/ndiswrapper-1.54-2.6.28.tar.gz
The breakage during unload and subsequent kernel corruption has been
tracked down to link_watch.c and the packet scheduler code and is due to
list corruption in dev_deactivate() being called with an invalid netdev
context.
The linkwatch code is not called by ndiswrapper and the problems are due
to poor error checking and cleanup in Linux proper with network drivers.
As such, ndiswrapper for 2.6.28 is posted as released based upon input
from other folks. Thanks to those folks who downloaded and tested the
code and provided feedback for this release.
The Native Linux wireless drivers and the hacked up pseudo-pirated
microcode they rely on still do not work properly on most of my laptop
wireless drivers or are very flaky for the BCM based drivers and Intel
3945 series drivers in 2.6.28. ndiswrapper almost a required component in
order to use wireless with Linux on a large number of laptops. I see no
way around the current vendors offerings which are heavily dependent on
internal and proprietary microcode without ndiswrapper since the Linux
drivers are essentially impossible to keep working.
Since most people (me included) are not interested in engaging in piracy
by using tools which lift microcode and firmware images from ndis drivers,
ndiswrapper seems a reasonable compromise since any customer who purchases
a card with ndis based drivers has a legal right to use them as ndis
drivers.
Cutting and pasting firmware seems too much like piracy, not to mention
being a great big hassle.
Jeff
On Sat, 27 Dec 2008 12:05:57 MST, [email protected] said:
> The Native Linux wireless drivers and the hacked up pseudo-pirated
> microcode
"pseudo-pirated"?
> On Sat, 27 Dec 2008 12:05:57 MST, [email protected] said:
>
>> The Native Linux wireless drivers and the hacked up pseudo-pirated
>> microcode
>
> "pseudo-pirated"?
>
I don't know what else to call it. Some of the USB based wireless
adapters have their firmware poached, clipped, and out on the internet.
I realize some of it is legit and provided by vendors, but by no means all
of it.
Jeff
>> On Sat, 27 Dec 2008 12:05:57 MST, [email protected] said:
>>
>>> The Native Linux wireless drivers and the hacked up pseudo-pirated
>>> microcode
>>
>> "pseudo-pirated"?
>>
>
> I don't know what else to call it. Some of the USB based wireless
> adapters have their firmware poached, clipped, and out on the internet.
> I realize some of it is legit and provided by vendors, but by no means all
> of it.
>
> Jeff
>
>
>
And most, if not all, of the reverse engineered and hacked firmware is
**SHIT** and does not work properly. My bcml15.sys drivers work great --
the Linux drivers with the crap firmware which locks up all the time and
stops receiving packets does not even make the LED wireless display work.
Jeff
>> On Sat, 27 Dec 2008 12:05:57 MST, jmerkey@xxxxxxxxxxxxxxxxxxxxx said:
> And most, if not all, of the reverse engineered and hacked firmware is
> **SHIT** and does not work properly. My bcml15.sys drivers work great --
> the Linux drivers with the crap firmware which locks up all the time and
> stops receiving packets does not even make the LED wireless display work.
The firmware used by the b43 driver for BCM43XX is _EXACTLY_ the same as used by
the Windows drivers. We are working on reverse-engineering the firmware, but
that is not workiong yet. In fact, the firmware is extracted from the Broadcom
drivers. If the firmware is faulty, it isn't our fault. Of course, our RE is not
perfect, but then we get absolutely no support from the vendor.
Your device may work better with ndiswrapper and bcmwl5.sys than with b43, but I
find that ndiswrapper dies after about 4 hours and cannot be resurrected even by
unloading and reloading the driver. A reboot is needed. On the other hand, b43
just keeps on chugging.
Larry