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

[mlazy]

Jalavoui,

still having ucode issues, but the logs show some new scan routines. My logs are probably no good to you, but worth a shot to review. I'll kep playing around with the hacks anyway.

Btw; is this the error thrown because the ucode load fails? (just want to know what I'm looking at)

Apr 26 08:54:50 Sony kernel[0]: error iwl3945: Microcode SW error detected. Restarting 0x82000008.
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: Start IWL Error Log Dump:
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: Status: 0x00010132, Config: 00000000 count: 1
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: Desc Time asrtPC blink2 ilink1 nmiPC Line
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: SYSASSERT (#5) 0000019322 0x0089A 0x00FE0 0x00300 0x00000 57
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: Start IWL Event Log Dump: display count 8, wraps 0
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: 0000000000 0x000000d5 0117
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: 0000000000 0x000000be 0118
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: 0000002107 0x00000000 0651
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: 0000002128 0xffffff22 0454
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: 0000002129 0x00000000 0472
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: 0000002150 0x000000c5 0104
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: 0000019321 0x00000000 0401
Apr 26 08:54:50 Sony kernel[0]: error iwl3945: 0000019323 0x00000100 0125


 846.txt ( 269.85K ) Number of downloads: 3

[pere]

IWI 3945 REV846

With high IOSleep values, i can get all to work, scept the famous SCAN_COMPLETE_NOTIFICATION.
Reboot log attached.

Values for IOSleep on IWI3945-base.c

ret = HOST_COMPLETE_TIMEOUT;
while(test_bit(STATUS_HCMD_ACTIVE, &priv->status)) {
IOSleep(500);
ret--;
if(ret==0)
break;
}

IOSleep(1000);//hack
ret = UCODE_READY_TIMEOUT;
while(!(test_bit(STATUS_READY, &priv->status))) {
IOSleep(500);
ret--;
if(ret==0)
break;

Attached File(s)

 reboot.txt ( 135.6K ) Number of downloads: 4
 iwi3945.kext.zip ( 237.87K ) Number of downloads: 14

 

[jalavoui]

good to have a hack team

Yesterday night tried the Iosleep with several values, 500, 1000, 1500, 2000, the last value makes the microcode works 100% of the times, and i got this:1st scannot ready for 1st scan

this means you get a error that prevents the scan from start - iosleep hack?

mlazy,
you get initial boot errrors,
but after that the scan cmds work ok
better leave it this way?

ok - this is a perfect log:
Apr 26 14:04:44 Macintosh kernel[0]: SCAN_COMPLETE_NOTIFICATION, 0x84
Apr 26 14:04:44 Macintosh kernel[0]: iwl3945: U iwl3945_rx_scan_complete_notif Scan complete: 23 channels (TSF 0x002E9984:00000000) - 1
Apr 26 14:04:44 Macintosh kernel[0]: iwl3945: U iwl3945_rx_scan_complete_notif Scan pass on 5.2GHz took 2190ms
Apr 26 14:04:44 Macintosh kernel[0]: iwl3945: U iwl3945_rx_scan_complete_notif Setting scan to off
Apr 26 14:04:44 Macintosh kernel[0]: iwl3945: U iwl3945_rx_scan_complete_notif Scan took 2560ms
Apr 26 14:04:44 Macintosh kernel[0]: queue_te @ /Users/droberts/Desktop/iwi3945_new-osx10.5/compatibility.cpp:6535
Apr 26 14:04:44 Macintosh kernel[0]: test_function @ /Users/droberts/Desktop/iwi3945_new-osx10.5/compatibility.cpp:6524
Apr 26 14:04:44 Macintosh kernel[0]: iwl3945: U iwl3945_bg_scan_completed SCAN complete scan
Apr 26 14:04:44 Macintosh kernel[0]: ieee80211_scan_completed

the iosleep timeout is very good:
Apr 26 14:04:38 Macintosh kernel[0]: darwin_iwi3945: Ethernet address 00:18:de:a4:ab:39
Apr 26 14:04:38 Macintosh kernel[0]: iwl3945: U iwl3945_send_cmd_sync Wait REPLY_TX_PWR_TABLE_CMD Condition: 49ms left
Apr 26 14:04:38 Macintosh kernel[0]: queue_te @ /Users/droberts/Desktop/iwi3945_new-osx10.5/compatibility.cpp:6535
Apr 26 14:04:38 Macintosh kernel[0]: timei 60000 timei2 -556682884
Apr 26 14:04:38 Macintosh kernel[0]: iwl3945: U iwl3945_alive_start ALIVE processing complete.
Apr 26 14:04:38 Macintosh kernel[0]: __ OUT iwl3945_alive_start __
the line __OUT means that ucode is ok and this was made at the right time - before call enable()

if you all agree this is the best solution

try BoboBear hack
if you get good results i'll put his code in the svn tree
the scan completes ok but doesn't restart (loops)
how should i make this work?
i have some ideas i'll post here and let you add them in xcode:
search for
IOLog("1st scan\n");
you have
ieee80211_sta_req_scan(dev,(u8*)"<hidden>", sizeof("<hidden>"));
this call the scan ok
try
ieee80211_sta_req_scan(dev,NULL,0);

i'm looking at iwl3945_rx_scan_complete_notif() wait...
ok here's the reason for the scan don't restart:
/* No need to wake the master device. */
if (sdata->dev == local->mdev)
continue;

if (sdata->type == IEEE80211_IF_TYPE_STA) {
if (sdata->u.sta.associated)
ieee80211_send_nullfunc(local, sdata, 0);
ieee80211_sta_timer((unsigned long)sdata);
}
i think the driver allways goes to if (sdata->dev == local->mdev)
it should call ieee80211_sta_timer() -> this will make the scan loop

ok - i figure it out - this is because i hacked the local->mdev
i'll comment this code, apply the timer hacks and upload to svn for testing


iwi3945 rel 847
add previous timer,dev hacks

iwi4965 rel 848
this driver don't need timer hacks - i guess...

both driver can call scan by two way search for IOLog("1st scan\n"); and try both ways
i'm expecting the driver to starts scanning in a loop, i.e, it will keeep scanning
if all of this works it's possible it tryes to associate to a network
if this happens you might get a kernel panic
test this and report

i'm very interested in making all this work in mlazy's card
ok - i stop writting - go run some tests

pere - you can ignore the bac crc info

[pere]

The scan 1 not ready for scan was present with IOsleep settled at 2000, with 1000 and 500, the message doesnt appers, but i always get this too:

test_function @ /Users/PERE/iwi3945/compatibility.cpp:6524
Apr 26 05:16:37 Pere kernel[0]: iwl3945: U iwl3945_rx_handle r = 141, i = 140, REPLY_3945_RX, 0x1b
Apr 26 05:16:37 Pere kernel[0]: iwl3945: U iwl3945_rx_reply_rx Bad CRC or FIFO: 0x00000702.
Apr 26 05:16:37 Pere kernel[0]: queue_te @ /Users/PERE/iwi3945/compatibility.cpp:6535

Is this ok?.

Wich hacks must i do now?, please point me where and what values...

[Roal2488]

Am testing the latest iwi3945 release. i posted my log earlier. ive booted with -s and got into a loop like u can see on the picture. did not get KP. hope it helps

Attached File(s)

 Archive.zip ( 1.89MB ) Number of downloads: 11

 

[pere]

Cant use the Iosleep from bobobear, the card refuse to be a F1...
Need to raise to 500 to make it work, then get this:

Apr 26 15:52:29 Pere kernel[0]: iwl3945: U iwl3945_enqueue_hcmd Sending command REPLY_SCAN_CMD (#80), seq: 0x4405, 366 bytes at 5[32]:4
Apr 26 15:52:39 Pere kernel[0]: iwl3945: U iwl3945_send_cmd_sync Wait REPLY_SCAN_CMD Condition: 0ms left
Apr 26 15:52:39 Pere kernel[0]: error iwl3945: Error sending REPLY_SCAN_CMD: time out after 500ms.

Will look for it on code...

EDIT, trying last hacks as javaui commented, still getting timeouts on RXON

[jalavoui]

nop - the scan won't loop like this
the call to ieee80211_sta_work() will fail with previous hack
not very shore about this - need some logs

pere,
if you get cmd timeout it will never work
i think you can change the iosleep in iwl3945_send_cmd_sync to 1
and change the one in iwl3945_mac_open to 100
the problem in your card is the delay in iwl3945_mac_open not the delay in iwl3945_send_cmd_sync
this is probably the best solution to all cards

[Wayfarer247]

Drat, 4965 still doesn't build. Can someone post the latest kext?

[Ruud]

Oke, im back from work and now reinstalling leopard.. will be testing within a hour or two.


Will post the newest 4965 kext then.

[jalavoui]

pere,
maybr your card don't need hacked timeouts
hope you can make it work

iwi3945 rel 849
maybe this work for all cards?


iwi3945/4965
i have a hack for the scan loop - need logs before posting new release

pere, do you change code?
looks like your card as a double call to __iwl3945_up()
this is why it fails to scan

[pere]

pere,
maybr your card don't need hacked timeouts
hope you can make it work

Used 300 on timeouts and stopped the ucode errors.
Reboot log attached, and as you said, its not scanning..
Tried to find where to hack your last comments, find the lines.. but dont know where to add that... sorry, im not coder, but with a little help can hack so much

Attached File(s)

 reboot.txt ( 81.71K ) Number of downloads: 3

 

[Ruud]

pere,
maybr your card don't need hacked timeouts
hope you can make it work

iwi3945 rel 849
maybe this work for all cards?
iwi3945/4965
i have a hack for the scan loop - need logs before posting new release

i will post a log for 4965 within 30 minutes m:)

[pere]

My reboot log, no timeouts. Rel 849

Attached File(s)

 reboot.txt ( 72.97K ) Number of downloads: 7

 

[Ruud]

oke, now on my new install im using kextload iwi4965.kext and now im getting....

cannot resolve dependencies for kernel extension iwi4965.kext/ error loading extension

does anyone knows a solution to this?

oke kextload -t gives me com.apple.iokit.IO80211Family = A valid compatible version of this dependency cannot be found...

strange that kext is in my extensions dir..

any suggestions?

[pere]

oke, now on my new install im using kextload iwi4965.kext and now im getting....

cannot resolve dependencies for kernel extension iwi4965.kext/ error loading extension

does anyone knows a solution to this?

oke kextload -t gives me com.apple.iokit.IO80211Family = A valid compatible version of this dependency cannot be found...

strange that kext is in my extensions dir..

any suggestions?

Have you install any kext related to wireless when you installed? Maybe thats the problem... seems you have an oldie release...

[Ruud]

it was replaced with an atheros compatible one i see. replaced the origional and now it works ..

Logs for lates 4965..

Attached File(s)

 system.rtf ( 22.72K ) Number of downloads: 9

 

[tiger2wander]

Jalavouis, Could you post the flowchart of iwi3945 driver here? so, me or anyone can trace or test it better than ever, please ...

[pere]

@Javalui: pere, do you change code?
looks like your card as a double call to __iwl3945_up()
this is why it fails to scan

Nop, this time i just downloaded the svn and try it.

[jalavoui]

scan call - this is the right way to start a scan

iwi3945 rel 850
might need to change timers for your card

iwi4965 rel 851
i'll check the scan loop for this card

[Ruud]

oke here is my log for 851 4965

next log wil be .txt again

Attached File(s)

 system.rtf ( 24.77K ) Number of downloads: 8