Intel Wireless driver

[jalavoui]

iwi3945

to check the linux logs,

get the log from previous post

open in console

use reload to load the complete file size

tipe iwl3945 in filter to see driver code only

the log is equal to iwi3945 except for the send/receive commands

for each command send the driver is expect to receive a iwl_rx_handle with reclaim=1

 

it's better to try it first in 10.4

10.5 will have to wait

 

ok

here's the linux log with grep iwl3945

 

maybe the driver don't call the interrupts while in iodelay loop of iwl_send_cmd()

try to add this before the iodelay:

fInterruptSrc->enable()

klog2.log.gz

[uk_rules_ok]

iwi3945

build and source version

 

the sources from linux are in intelwireless site

i've post a linux log - check previous posts

 

the driver is stuck at snd_cmd - should be easy to fix

the timeout delay is ok - the bug is somewhere else

 

 

no kp, here logs.

[ice96]

I find in previos bug:

linux log:

Aug 30 22:17:19 friedi iwl3945: U iwl_enqueue_hcmd Sending command REPLY_BT_CONFIG (#9b), seq: 0x0401, 16 bytes at 1[1]:4

in my:

iwi3945: iwl_enqueue_hcmd Sending command REPLY_BT_CONFIG (#9b), seq: 0x0400, 16 bytes at 0[0]:4

seq is not same

[jalavoui]

ice96,

check iwl_alive_start()

the sequence is different because i've dissabled the send_power cmd -> gives firmware errors. i don't use this in iwi2200

the power setup is done in iokit at the start() function

 

i've build the source with other firmware versions - maybe this help. in the past iv'e got some erros related with firmware versions

 

current source uses firmware 2.14.4 - i'm including the header of other firmware versions. backup the current firmware before replacing it in the xcode project. i use ./hex2string command to create the firmware headers

 

you can find all intel source files in http://intellinuxwireless.org/

iwi3945 is using version 1.0.0-1 -> found in intel main site "stable" sources - linux session

i only use iwlwifi-1.1.20 as a reference to see what they fix in code

 

just remebered that 1.0.0-1 uses firmware 2.14.1 in linux code -> maybe this is the stable version

 

before posting logs of this .dmg try networkselector option 1 a few times

ipw3945.ucode2.14.1.5.h.zip

ipw3945.ucode2.14.3.h.zip

iwi3945_1.dmg

iwi3945_2.dmg

[uk_rules_ok]

heres my logs for the above two files.

[ice96]

Ok, logs:

dmesg.txt

ioreg.txt

system.txt

dmesg.txt

ioreg.txt

system.txt

[koenlek]

Hi people,

 

Nice to see that there is still work going on for 3945! Keep on the good work. I guess many people will be happy with a driver for that one. The problem is that I own the 4965agn. I tested the driver on http://code.google.com/p/iwidarwin/ but this one doesn't work.

Has anybody yet got this card to work? As far as i know it's impossible.

And is anyone still developing a 4965agn driver? Is there a chance that it will become working? Or will it be practically impossible due to the n-draft? And is there a (bigger) chance that it will run on Leopard?

 

@ ice96, is it maybe relatively easy to port 3945 to a 4965 version when it's done?

 

@ jalavoui. It might be an idea to mention the SVN revision number for the SVN-compiled binaries on code.google.com . Because now it is very unclear to see if the binaries are updated or not. A view at the SVN revision number would make people sure whether it's updated or not.

 

I really hope you will succeed in this nice peace op programming!

Greetings Koen

[jalavoui]

iwi3945

looks like the firmware versions make no diferrence

i see in logs

 

: Grabbing access while already held at line 882

 

this can be the cause of firmware errors - i'll try to fix this

nop this can stay as it is

 

i remove functions from class darwin_iwi3945 - this make the code more simple

i suspect that might be a bug by calling functions this way

also make a call to interrupts in snd_cmd

 

ice96,

try this

move the declaration from iwi3945.h

struct iwl_priv *priv;

to base.cpp -> this will make it global and probably will make the driver work better

i've some problems with (darwin_iwi3945 *clone;) and doing this fix the problem

 

bt_config is for bluetooth suport. see in logs: sometimes the driver sends this command susesufully other times it fails

 

i put back to firmware 2.4.4 - think is best for now

 

since you have the card - try this

in snd_cmd -> change the iodelay values... you might find one that works

iwi3945.dmg

iwi3945_src.tar.gz

[ice96]

As I see error is in iwl_alive_start in :

 

/* Configure BT coexistence */

iwl_send_bt_config(priv);

 

what do this command?

 

I think we need to debug priv in this place. maybe I'm wrong.

[jalavoui]

can someone test the .dmg from last post? it's fresh

[auqs10]

can someone test the .dmg from last post? it's fresh

 

Yes, immediately.

[b34r]

Last build logs. Keep them coming!

dmesg.txt

ioreg.txt

system.txt

[ice96]

logs:

ioreg.txt

system.txt

ioreg.txt

[jalavoui]

iwi3945

removed rx_handle from queue_te

iwi3945.dmg

[auqs10]

can someone test the .dmg from last post? it's fresh

 

Here are the logs of the previous version.

I will test also the newest one.

[ice96]

logs:hmm

ioreg.txt

system.txt

[jalavoui]

ice96,

from your logs looks like the driver need more time to complete the cmd

i've raised the timer in snd_cmd

 

all i did was remove all queue_te for iwl_irq_tasklet

iwl_irq_tasklet(priv);

//clone->queue_te(13,OSMemberFunctionCast(thread_call_func_t,clone,&darwin_iwi3945::iwl_irq_tasklet),

 

set the timer in snd_cmd

while (priv->status & STATUS_HCMD_ACTIVE)

{

rc++;

//clone->fInterruptSrc->enable();

mdelay(HOST_COMPLETE_TIMEOUT);

if (rc==10) break;

}

when rc=10 timeout is 500ms= 0,5 second

iwi3945.dmg

[ice96]

Hmmm...

ioreg.txt

system.txt

[auqs10]

Again I am posting logs from a previous version. Comments included.

I will test now your latest jalavoui.

[jalavoui]

iwi3945

add rx_handle into snd_cmd timer

iwi3945.dmg

[Sergey Basmanov]

This is what I see on 10.4.10

iwi3945.dmg from last post.

Hope this will be useful.

dmesg.txt

ioreg.txt

system.txt

[ice96]

I don't no why, but I got kernel panic. Error 14 page foult. Sorry I don't have photo. Maybe I did some thing wrong...

[auqs10]

Logs commented from dmg posted on #3342. Again delaied logs of the previous version.

[ice96]

auqs10,plz test 3345.i don't know kernel panic for me only or not.Can anyone test 3345?

[jalavoui]

iwi3945

this could work - no kernel panic here

iwi3945.dmg