Re: PowerBook G4 wireless problems on YDL 6.0

I've interspersed my comments below for clarification.

On Wed, 9 Apr 2008 10:54:11 +0200
furio ercolessi <furiospin.it> wrote:

> Hi Derick, 
> 
> thanks for looking into this problem.
> 
> On Tue, Apr 08, 2008 at 09:04:08PM -0400, Derick
Centeno wrote:
> > 
> > There are two methods of extracting the firmware
from MacOS X
> > [...]
> 
> Absolutely no problem with that.  Beware that the file
AppleAirPort2
> does not exist any more in Leopard.  Perhaps the good
one is named now
>
/Library/Extensions/I080211Family.kext/Contents/PlugIns/Appl
eAirPortBrcm4311.kext/Contents/MacOS/AppleAirPortBrcm4311
> but I extracted the drivers from a copy of OSX10.3 I
still had around.
> 

I'm using Tiger on my PB.  Thanks for the heads up on
Leopard, I was
considering getting it but I had already thought that it had
put "bells and
whistles" and way too much 3D this and that, and no
read added truly useful
functions into the interface. I'm not moving beyond Tiger
especially as they are
removing controls to PowerPC hardware which I use.  The also
explains why YDL
has become my primary OS, and YDL6 even moreso.

Just to return the favor of providing a likewise potentially
useful point
regarding Leopard's faults.  Remember that the Powerbook has
a fax interface
built in where you could have seen a preview of what you fax
out -- you can't
do that in Windows (not for free) and in Panther (the
release prior to Tiger)
you could.  I've observed that in Tiger this useful feature
(at least for me)
was removed and Apple published nothing regarding this
change on their website.
For me this made Tiger a "downgrade" and most
likely from what you shared,
Leopard may be a more profoundly pretty
"downgrade" as well.
> > [...]
> > So never accept the computer's view on face value
in any operating system.  
> > [...]
> 
> Yes, I have grasped this 30 years ago, no problem here
either 

Glad to hear it, although I've 20 plus your 30 years
experience in Unix! 

Sometimes we all get taken by the "new" this or
that feature as it is packaged
so incredibly well. I fear Apple is doing more
"pretty" than function.  Thank
God for TSS and their focus on function in producing YDL6! 


If I compare what happened at Apple to cars, it's like
someone confused the
value of a Lamborghini or Citroen with a Ford and decided
producing and
supporting exclusively Fords were the way to go.  For some
going on the cheap
makes "business" sense as far as cutting costs in
a certain narrow minded
logic, but for the perceptive consumer seeking value and
utility such cuts or
choices become obvious degradations of value to that
product.

> > The first three instructions under the section:
Configure your... under
> > Linux. Didn't work.  Being the nut that I am I
tried a bit of cheating by
> > merely copying ifcfg-eth0 and naming it
ifcfg-eth1; that didn't work
> > either.  The reason why it didn't work is because
the HWADDR variable in
> > ifcfg-eth1 != the HWADDR variable in ifcfg-eth0. 
Anaconda creates both
> > files with different definitions for the hardware
location accessed for
> > Ethernet (eth0) and Wireless (eth1). 
> 
> Yes, that is the MAC address and every interface has
its own.
> So I start to create a new ifcfg-eth1 using
system-config-network
> (which I believe should be what Anaconda does on a
fresh install).
> I follow Network Configuration -> new Device Type
-> Wireless Connection.
> Here I only see "Other Wireless Card" as a
possible choice.  If I choose
> that, the BCM4306 does not show up in the device list. 
I tried
> "Airport" but it does not work (I guess that
is the Airport of former
> powerbooks).  If I follow
> Network Configuration -> new Device Type ->
Ethernet Connection
> then I do see the BCM4306, but the resulting
configuration does not
> work either.  IN both cases, by "does not
work" I mean that the
> interface refuses to start with
> 
> > #ifconfig eth1 up
> 
> Can you tell me what you have in /etc/modprobe.conf ?

I'm not using the interface at all.  My method, a variant of
what Chris Murtagh
used and demonstrated by-passes the interface entirely.  The
challenge in this
particular situation is that you are still thinking like an
Apple user where
the interface is the function or application.  

We all need to switch to thinking in terms of Linux/Unix
where it is ok, and
more efficient, to by-pass the interface and get to
executing the function which
interests us.  There's time to dress up or correct the
interface later if
that's what you care about.  Just to emphasize the point, if
you are
choosing to not use the interface, why would you or anyone,
care whether
it works or not.  

However, just so I'm not a "stick in the mud" or
non-cooperative, here's my
modprobe.conf:

alias eth0 sungem
alias eth1 ssb
alias snd-card-0 snd-powermac
options snd-card-0 index=0
options snd-powermac index=0

All the best... and I encourage your improvement at poker.


=====
"If I were not a physicist, I would probably be a
musician. I often
think in music. I live my daydreams in music. I see my life
in terms of
music. ... I get most joy in life out of music."  

"What Life Means to Einstein: An Interview by George
Sylvester
Viereck," for the October 26, 1929 issue of The
Saturday Evening Post.

_______________________________________________
yellowdog-general mailing list
yellowdog-generallists.terrasoftsolutions.com
http://lists.terrasoftsolutions.com/mailma
n/listinfo/yellowdog-general
HINT: to Google archives, try  '<keywords>
site:terrasoftsolutions.com'

On Wed, Apr 09, 2008 at 08:20:18AM -0400, Derick Centeno
wrote:
> 
> [considerations on OSX evolution]
> [...]
> For me this made Tiger a "downgrade" and most
likely from what you shared,
> Leopard may be a more profoundly pretty
"downgrade" as well.

I essentially agree, the OSX GUI is also getting slower due
to
new stupid bells and whistles.  But I have no choice, I have
to follow
the world. I teach two simulation courses and my laptop has
to be 
reasonably aligned with the students' machines.  I also rely
heavily
on MacPorts stuff for the OSX side, for Panther
MacPorts/DarwinPorts is
now in a completely abandoned state, and the maintainers'
activity is 
definitely focused on Leopard.

> Thank God for TSS and their focus on function in
producing YDL6!  

Yes, it's very nice to have a company committed to Linux on
PowerPC.

> > [...]  IN both cases, by "does not work"
I mean that the
> > interface refuses to start with
> > 
> > > #ifconfig eth1 up
> > 
> > Can you tell me what you have in
/etc/modprobe.conf ?
> 
> I'm not using the interface at all.  My method, a
variant of what Chris Murtagh
> used and demonstrated by-passes the interface entirely.
 The challenge in this
> particular situation is that you are still thinking
like an Apple user where
> the interface is the function or application.  

No no no, I was thinking UNlike an Apple user and with
"interface"
I was referring to a "network interface" , the
"if" in ifconfig!
You can by-pass those and still be connected only if you go
back to
Kermit on serial ports 

> However, just so I'm not a "stick in the mud"
or non-cooperative, here's my
> modprobe.conf:
> 
> alias eth0 sungem
> alias eth1 ssb    <==============

This was useful.

I solved the problem.  I _believe_ that the key issue was
that the new
Redhat/Fedora derivatives (and YDL 6) assume the presence of
recent 
versions of the Broadcom drivers.  The thread that inspired
me was:
http://forums.fedoraforum.org/printthread.php?t=171176

(not specific to Apple hardware, but specific to Broadcom
devices).

So I did the following (some of these things may not be
relevant but
I report them for sake of completeness):

- Throw away the old drivers I previously stored in
/lib/firmware

- Get the latest 'firmware cutter' program,
    http://download.berlios.de/bcm43xx/b43-fwcutter-009.t
ar.bz2
  unpack it and compile it.  It does not compile under OSX
10.5 with Xcode
  ("error: byteswap.h: No such file or
directory"), but it compiles
  fine on Linux (either Intel or PowerPC).

- Get this firmware file:
  http://downloads.openwrt.org/sources/broadcom-
wl-4.80.53.0.tar.bz2
  and unpack it.

- store the firmware files in /lib/firmware using
  (path)/b43-fwcutter -w /lib/firmware
(path)/broadcom-wl-4.80.53.0/kmod/wl_apsta.o
  For the record, I did this on a Intel Linux and then
copied the files
  but it should not matter how you do it.

- 22 firmware files will be stored in a subdirectory
/lib/firmware/b43/ ;
  they should not stay directly in /lib/firmware as with the
previous
  drivers.

- Make sure that /etc/modprobe.conf contains the line:
	alias eth1 ssb

- Manually edit /etc/sysconfig/network-script/ifcfg-eth1. In
my case:

-------------------------------------------
# Please read /usr/share/doc/initscripts-*/sysconfig.txt
# for the documentation of these parameters.
GATEWAY=192.168.2.1
TYPE=Wireless
DEVICE=eth1
BOOTPROTO=none
HWADDR=00:11:24:27:5e:66
NETMASK=255.255.255.0
IPADDR=192.168.2.129
ONBOOT=yes
USERCTL=yes
IPV6INIT=no
PEERDNS=yes
ESSID=(name of my wireless network)
CHANNEL=9
MODE=Auto
RATE=Auto
DOMAIN=
DHCP_HOSTNAME=
-------------------------------------------

(I use static IPs.  I copied the HWADDR of my card from the
output of 
ifconfig on OSX).

As a final note, my wireless network was originally on
channel 13, which 
is valid in Europe but not in the US.  The driver assumed to
be in the US
and choked on channel 13, insisting that it has to be
<=11.   So, not
being in an antagonist mood, I moved the whole network to
channel 9 which
nobody in the neighborhood seems to use.

These are the relevant section of the boot log:

------------------------------------------------------------
--
Using PowerMac machine description
Total memory = 1536MB; using 4096kB for hash table (at
cfc00000)
Linux version 2.6.23-9.ydl6.1 (rootbuild2.terraplex.com) (gcc
version 4.1.1 20070105 (Red Hat 4.1.1-52.ydl.1)) #1 SMP Sat
Jan 26 19:47:37 EST 2008
[...]
bcm43xx driver
[...]
b43-phy0: Broadcom 4306 WLAN found
b43-phy0 debug: Found PHY: Analog 2, Type 2, Revision 2
b43-phy0 debug: Found Radio: Manuf 0x17F, Version 0x2050,
Revision 2
b43-phy0 debug: DebugFS (CONFIG_DEBUG_FS) not enabled in
kernel config
phy0: Selected rate control algorithm 'simple'
net eth1: device_rename: sysfs_create_symlink failed (-17)
[...]
b43-phy0 debug: Loading firmware version 351.126 (2006-07-29
05:54:02)
b43-phy0 debug: Chip initialized
b43-phy0 debug: 30-bit DMA initialized
b43-phy0 debug: Wireless interface started
b43-phy0 debug: Adding Interface type 2
b43-phy0 debug: Using hardware based encryption for keyidx:
0, mac: ff:ff:ff:ff:ff:ff
ADDRCONF(NETDEV_UP): eth1: link is not ready
[...]
(later proceeding with authentication and bringing the link
up automatically)
[...]
------------------------------------------------------------
--

I have no idea about the origin of the device_rename error.
That error
does not seem to have a practical impact (but time will
tell!).

Hope it helps other people stumbling into this.  I hope that
this stuff
will get fixed in YDL 6.1.

furio

_______________________________________________
yellowdog-general mailing list
yellowdog-generallists.terrasoftsolutions.com
http://lists.terrasoftsolutions.com/mailma
n/listinfo/yellowdog-general
HINT: to Google archives, try  '<keywords>
site:terrasoftsolutions.com'