Intel Wireless 2100, 2200bg, 2915bg, 3945abg, 4965agn, IWIDarwin drivers for mac os x 10.4 and 10.5 *under development* Options

[adamisnt]

iwi4965 rel 882

return rc and atomic_xchg renenabled

Got this:

Apr 29 15:09:13 adam-notebook kernel[0]: IOPCCard info: iwl4965: Intel® Wireless WiFi Link 4965AGN driver for Linux, 1.2.23d
Apr 29 15:09:13 adam-notebook kernel[0]: IOPCCard info: iwl4965: Copyright© 2003-2007 Intel Corporation
Apr 29 15:09:13 adam-notebook kernel[0]: ieee80211_alloc_hw @ /Users/Adam/Desktop/root/compatibility.cpp:5610
Apr 29 15:09:13 adam-notebook kernel[0]: sta_info_init @ /Users/Adam/Desktop/root/compatibility.cpp:5442
Apr 29 15:09:13 adam-notebook kernel[0]: ieee80211_if_sdata_init @ /Users/Adam/Desktop/root/compatibility.cpp:2089
Apr 29 15:09:13 adam-notebook kernel[0]: ieee80211_alloc_hw [OK]
Apr 29 15:09:13 adam-notebook kernel[0]: iwl4965: U iwl4965_pci_probe pci_resource_len = 0x00000008
Apr 29 15:09:13 adam-notebook kernel[0]: iwl4965: U iwl4965_pci_probe pci_resource_base = 0x3420c000
Apr 29 15:09:13 adam-notebook kernel[0]: iwl4965: U iwl4965_set_rxon_chain rx chain 280E
Apr 29 15:09:13 adam-notebook kernel[0]: IOPCCard info: iwl4965: Detected Intel Wireless WiFi Link 4965AGN
Apr 29 15:09:13 adam-notebook kernel[0]: iwl4965: U iwl4965_set_rxon_channel Staging channel set to 6 [3]
Apr 29 15:09:13 adam-notebook kernel[0]: init_waitqueue_head stubbed out!
Apr 29 15:09:13 adam-notebook kernel[0]: IOPanelDevice: Machine Is Laptop!
Apr 29 15:09:13 adam-notebook kernel[0]: IOPanelDevice: Registered OK!
Apr 29 15:09:13 adam-notebook kernel[0]: iwl4965: U iwl4965_pci_probe MAC address: 00:1d:e0:54:19:fd

Does this mean anything?

Apr 29 15:09:22 adam-notebook kernel[0]: iwl4965: U iwl4965_rx_scan_results_notif Scan ch.res: 11 [802.11bg] (TSF: 0x00000000:00454456) - 3 elapsed=21117
usec (30ms since last)
Apr 29 15:09:22 adam-notebook kernel[0]: iwl4965: U iwl4965_rx_handle r = 255, i = 253, STATISTICS_NOTIFICATION, 0x9d
Apr 29 15:09:22 adam-notebook kernel[0]: iwl4965: U iwl4965_hw_rx_statistics Statistics notification received (480 vs 484).
Apr 29 15:09:22 adam-notebook kernel[0]: iwl4965: U iwl4965_rx_handle r = 255, i = 254, SCAN_COMPLETE_NOTIFICATION, 0x84
Apr 29 15:09:22 adam-notebook kernel[0]: iwl4965: U iwl4965_rx_scan_complete_notif Scan complete: 11 channels (TSF 0x004545D6:00000000) - 1
Apr 29 15:09:22 adam-notebook kernel[0]: iwl4965: U iwl4965_rx_scan_complete_notif Scan pass on 2.4GHz took 150ms

Hope this helps.

Attached File(s)

 systemlog_882.txt ( 623.82K ) Number of downloads: 6

 

[Doctadre]

@adamsint. How did you get this?

Apr 29 15:09:13 adam-notebook kernel[0]: iwl4965: U iwl4965_pci_probe pci_resource_len = 0x00000008
Apr 29 15:09:13 adam-notebook kernel[0]: iwl4965: U iwl4965_pci_probe pci_resource_base = 0x3420c000
Apr 29 15:09:13 adam-notebook kernel[0]: iwl4965: U iwl4965_set_rxon_chain rx chain 280E
Apr 29 15:09:13 adam-notebook kernel[0]: IOPCCard info: iwl4965: Detected Intel Wireless WiFi Link 4965AGN
Apr 29 15:09:13 adam-notebook kernel[0]: iwl4965: U iwl4965_set_rxon_channel Staging channel set to 6 [3]

These are the lines that make an error for me. What lines did you hack?

[i_Geek]

iwl4965_set_rxon_channel Staging channel set to 6 [3]

isn`t this suppose to be good ?)

[jalavoui]

keep hacking
some tips:
if you enable return rc the driver does the scan as expected in linux but
if you check it the scan must complete 2 bands (a, them bg or bg,a) before get a scan_complete
after this it should repeat the process
i hacked this using a timer - search for
ieee80211_sta_timer() -> the //hack is here but this is not very good because this functions is called later in ieee80211_scan_completed

setup_timer() - last line is a hack (add_timer) - this call IOPCCardAddTimer()
since the timer as no value the code will call
else timei=3000;
wich means this will loop every 3 seconds
maybe 3 seconds is not enought to complete scaning a and bg bands
try to increase this value

if you manage to get a iwl3945_rx_scan_complete_notif()
look at what follows
you can get a call to ieee80211_scan_completed or iwl3945_bg_request_scan (in case there are still bands to scan)

i'd like you do try to follow this code and find a way to make the scan loop after it finish scanning the 2 bands

the iwi4965 works the same way (the functions names start with iwl4965)

but before you do all this make shure you don't get timeouts sending commands or it will never work

[pere]

IWI3945 REL881
.
Did all the hacks,except ieee80211_sta_timer() cant find it on the iwi3945 code.
Used a timei=6000

First time reboot, no errors, did a scan A/BG, i enjoyed watching the log and couldnt pick the first lines cause console reload... XD
Tried to boot again to pick the reboot log, then no longer booted, got frozen.
Booted on safe mode, manual load, no errors on load (i think), log attached A/BG scan notifications.
Should give to 3945 a higher timei?.


EDIT: Very very high timei (24000)
No errors on boot.
GOT SCAN_COMPLETE_NOTIFICATION
But i cant see it completes the bg band....
Log and kext attached.

Apr 29 16:42:21 Pere kernel[0]: iwl3945: U iwl3945_rx_handle r = 59, i = 58, SCAN_COMPLETE_NOTIFICATION, 0x84\
Apr 29 16:42:21 Pere kernel[0]: iwl3945: U iwl3945_rx_scan_complete_notif Scan complete: 13 channels (TSF 0x01525FE2:00000000) - 1\
Apr 29 16:42:21 Pere kernel[0]: iwl3945: U iwl3945_rx_scan_complete_notif Scan pass on 2.4GHz took 320ms\

Attached File(s)

 Succesfully_reboot.txt ( 743.81K ) Number of downloads: 14
 iwi3945.kext.zip ( 238.29K ) Number of downloads: 14

 

[mlazy]

881 logs:

1. Clean build no hacks:

 881_clean.txt ( 22.24K ) Number of downloads: 3


2. Atomic hack=immediate freeze after kextload


3. return_rc hack:

 881return_rc.txt ( 930.21K ) Number of downloads: 2


4. Send power mode hack=immediate freeze:

 881sndpwrmode.jpg ( 71.26K ) Number of downloads: 18


I'm gonna play with the atomic hack now..

BTW, I get BT erors sometimes, but don't have bluetooth at all.....is it something else?

[jalavoui]

EDIT: Very very high timei (24000)
No errors on boot.
GOT SCAN_COMPLETE_NOTIFICATION
But i cant see it completes the bg band....
Log and kext attached.

this mean you have to wait 24 seconds for each scan!
you need to find a better way - or maybe not
the problem sill exists with the scan loop
this is caused by ieee80211_sta_timer hack
this hack make the driver scan again but it also prevents scan to complete
you need to make the driver call this and complete the iee station_work code

[pere]

this mean you have to wait 24 seconds for each scan!
you need to find a better way - or maybe not
the problem sill exists with the scan loop
this is caused by ieee80211_sta_timer hack
this hack make the driver scan again but it also prevents scan to complete
you need to make the driver call this and complete the iee station_work code

Javalui, i have been looking for
ieee80211_sta_timer()

But i cant find it on the code...
Even i look for any "hack" or "//hack", cant find it...

EDIT: Or is one of this...

Attached File(s)

 Capture.jpg ( 25.41K ) Number of downloads: 43

 

[kazzi]

He means the commented hack is there. You need to search for the call of ieee80211_sta_timer()

[adamisnt]

From looking at the linux logs for 4965, it seems the scan doesn't take more than 2 seconds for each frequency.

Apr 27 13:16:27 jpw-vaio kernel: iwl4965: I iwl4965_rx_scan_complete_notif Scan pass on 2.4GHz took 348ms
Apr 27 13:16:29 jpw-vaio kernel: iwl4965: I iwl4965_rx_scan_complete_notif Scan pass on 5.2GHz took 1720ms

[pere]

He means the commented hack is there. You need to search for the call of ieee80211_sta_timer()

Yep, you are right.

Commented this line:
//set_bit(IEEE80211_STA_REQ_SCAN, &ifsta->request);//hack

Had to reboot several times to get one log without errors, here it is.

Attached File(s)

 reboot.txt ( 923.28K ) Number of downloads: 17

 

[mlazy]

Yep, you are right.

Commented this line:
//set_bit(IEEE80211_STA_REQ_SCAN, &ifsta->request);//hack

Had to reboot several times to get one log without errors, here it is.

Post your kext too please. I'd love to test it as well. Do you have all of Jalavoui's hacks enabled at thtis point?

[pere]

Post your kext too please. I'd love to test it as well. Do you have all of Jalavoui's hacks enabled at thtis point?

Yep.
Take it. (will lower the timei from 24000 to 12000 in this kext, attached the log from previous kext).

Check if you can complete the A/BG bands, if dont, i will raise the timei again... If you cant see the A/BG bands, then try my previous kext.
And remember, im not sure my kext is working well with all the hacks, Javalui has to check it.

IWI 3954 Rel 881
EDIT.
Hacks:
uncommented //return rc
changed else timei=3000;>else timei=12000
enabled atomic_xchg
commented //set_bit(IEEE80211_STA_REQ_SCAN, &ifsta->request);//hack

Attached File(s)

 iwi3945.kext.zip ( 238.3K ) Number of downloads: 33
 reboot_all_hacks_timei_24000.txt ( 923.28K ) Number of downloads: 15

 

[jalavoui]

From looking at the linux logs for 4965, it seems the scan doesn't take more than 2 seconds for each frequency.

Apr 27 13:16:27 jpw-vaio kernel: iwl4965: I iwl4965_rx_scan_complete_notif Scan pass on 2.4GHz took 348ms
Apr 27 13:16:29 jpw-vaio kernel: iwl4965: I iwl4965_rx_scan_complete_notif Scan pass on 5.2GHz took 1720ms

yes - this can be important - same thing for 3945
can someone try to make this work within this timer
also need cxxdev to post another linux log for 4965 with more data (scans, association, dissasociate,etc)
and it will be nice someone build the linux version for 3945 and post the log so we can match iwi3945 against the same linux version

pere,
do you get any scan call? - i can't see it in your logs
since you're not using this hack
//set_bit(IEEE80211_STA_REQ_SCAN, &ifsta->request);//hack
you need to comment add_timer(timer); in setup_timer -> put //add_timer(timer);//hack
and in IOPCCardAddTimer
make else timei=0; you can do this because the timer is enabled in code when you use mod_timer(&ifsta->timer this is done after associating,etc

the ieee80211_sta_timer is called when you get a iwl3945_rx_scan_complete_notif -> ieee80211_scan_completed ->add iolog debug to see if this is called
if the driver don't start a scan you can force it to make one scan loop in check_firstup()
in theory the scan loop should be called when you start calling it in check_firstup
someting is wrong i hope you can follow the code and figure it out
this is the same for iwi4965
ok - finish editing

[pere]

yes - this can be important - same thing for 3945
can someone try to make this work within this timer
also need cxxdev to post another linux log for 4965 with more data (scans, association, dissasociate,etc)
and it will be nice someone build the linux version for 3945 and post the log so we can match iwi3945 against the same linux version

pere,
do you get any scan call? - i can't see it in your logs

Nop.
I have recheck console right now...


@ JAVALUI : you need to comment add_timer(timer); in setup_timer -> put //add_timer(timer);//hack
and in IOPCCardAddTimer
make else timei=0; you can do this because the timer is enabled in code when you use mod_timer(&ifsta->timer this is done after associating,etc

the ieee80211_sta_timer is called when you get a iwl3945_rx_scan_complete_notif -> ieee80211_scan_completed ->add iolog debug to see if this is called
if the driver don't start a scan you can force it to make one scan loop in check_firstup()
in theory the scan loop should be called when you start calling it in check_firstup

Done.
EDIT:
Reboot log, no errors, no scan calls.
Please Javalui, im not coder, if you need me to add for example the iolog, write me the code and tell me where to put it, and i will do it happily.

Attached File(s)

 reboot_log.txt ( 367.88K ) Number of downloads: 3
 iwi3945.kext.zip ( 238.23K ) Number of downloads: 5

 

[jalavoui]

maybe doing this can make the driver more stable:
//iwl3945_send_bt_config(priv);
if this work you might remove the iosleep(2000) hack

[adamisnt]

Jalavoui,

I assume that you have seen this already. If not...

Linux 4965 development page.

I am going to dive into this documentation with the hope I can contribute.

Edit: Nevermind - waaay over my head.

[Miguel Madrid]

maybe doing this can make the driver more stable:
//iwl3945_send_bt_config(priv);
if this work you might remove the iosleep(2000) hack

Jalavoui,

I did test yesterday to disable the BT config call (http://forum.insanelymac.com/index.php?showtopic=84072&st=1700#)

The BT error doesn't come up... but as I said, there could be some cards with BT and others without (this is why there are several Device ID).

Different cards could mean different startup secuence... with less or more errors...

In the mean time, i think it's a good idea to keep it hacked.... and see what happend...

On the 4965 I don't see the BT error, altough I have a dual Wifi/BT card...

I'm currently reinstaling OSX as I had a few problems with a KP trying to hack the REQ_Scan....

[jalavoui]

ok -i'll add the hacks for testing for both cards
wait...

[redfez]

no card found (3945)