Intel Wireless driver

[Mr. Green]

Here is a picture of the panic with my 2100. The system panics as soon as I load the kext, so i dont have a chance to link the network selector thing.

[ryuu123]

(snip)

 

 

3945abg's radio power is controled by register 0x05c.

 

from NetBSD's code..

#define WPI_UCODE_CLR		0x05c

/* clear "radio off" and "disable command" bits (reversed logic) */
WPI_WRITE(sc, WPI_UCODE_CLR, WPI_RADIO_OFF);
WPI_WRITE(sc, WPI_UCODE_CLR, WPI_DISABLE_CMD);

/* possible flags for register WPI_UCODE_CLR */
#define WPI_RADIO_OFF		(1 << 1)
#define WPI_DISABLE_CMD		(1 << 2)

 

but this stuff is already in iwi3945.c: ipw_up()

 

#define CSR_UCODE_DRV_GP1_CLR   (CSR_BASE+0x05c)

#define CSR_UCODE_SW_BIT_RFKILL					 (0x00000002)
#define CSR_UCODE_DRV_GP1_BIT_CMD_BLOCKED		   (0x00000004)

ipw_write32( CSR_UCODE_DRV_GP1_CLR, CSR_UCODE_SW_BIT_RFKILL);
ipw_write32( CSR_UCODE_DRV_GP1_CLR,
CSR_UCODE_DRV_GP1_BIT_CMD_BLOCKED);

 

abot not calling bg_alive_start:

 

"no ALIVE INTR" means not occur(receive) interrpt

CSR_INT register == BIT_INT_ALIVE

 

priv->status is 0x28. interrupt seems disabled ...

 

--

I wonder where iwi3945 is ported from?

 

firm version is 2.13 ipwraw.ucode.

so i think ported from iwlwifi.

(but seems very earlier version, iwlwifi-0.0.3 or so)

 

correct?

 

ipw3945 needs userspace daemon,

so choosing iwlwifi is better way.

[moseschrist]

well...

 

last iwi2200 version as the mbuf cluster bug

 

sorry for that

 

it's fixed in the

 

iwi2200 rev 416

i hope this version is bug free

i'm unable to upload the .dmg but source code is up to date in svn tree

 

moses,

i think you should add some IODelay in nsgui for power off. it can cause a crash or a firmware reload error if associated to a network

 

i'll get on it ASAP, also will include latest nsgui in the driver installation app.

 

i'm sorry for low updates recently... i'm a bit tied up on porting ieee80211 lib.

[rjtd]

Is WEP working?

[moseschrist]

Is WEP working?

 

not yet, no encryption yet...

[osxonmylaptop]

sorry, not working 4 me

 

 

ups. whats this in dmesg:

-iwi3945: getHardwareAddress 00:18:de:54:70:01 ?

 

don`t know how to get the system.rtf, i`m a noob

thx for your work, guys!

 

 

Hi,

Are you using the scripts for getting the logs?

http://iwidarwin.googlecode.com/files/getlogs3.command.zip

I copied it to my home folder (The default path for Terminal), so I open Terminal > go into superuser "su" > enter root password (You have to enable it from NetInfo Manager) > and then ./getlogs3.command > find your logs in your Hard Disk folder.

Bye

[JustJoe]

this seem to only happen with Acer Aspire laptops!!

Ah ha! I never thought to ask that question.

i get the map is zero error as well. from what i can tell this is a problem with the card not being found on the pci bus. can it be that the card is switched off at boot and needs to be switched back on before the card can be accesed by the driver.

Looks like AppleACPIPCI has no clue it's even there. Could be crazy Acer IRQ sharing gone awry?

i think we need someone to port acerhk (acer hotkey driver) or acer_acpi (kernel module for acer hotkeys through acpi) from linux to get the wireless cards initialized before the driver is loaded. i dunno, the switch works for me in sabayon without installing anything. acer_acpi code is here http://aceracpi.googlecode.com/svn/trunk/a...cpi/acer_acpi.c can someone take a look at it?

Hey, good catch. I have seen in the ioreg's that the cards are not recognized, but I assumed it was an iwi software problem.

Do you think there are other people on this board already working on Acer ACPI? That would probably be the best place to start - pretty sure it's out of the realm of this project.

 

It will be intereseting to know what you find out

 

[osxonmylaptop]

 

Hi,

I never did that before, I did and installed latest (415?). Same {censored}

Here are logs and:

Until yesterday I had Jas 10.4.8 with semthex 8.8.1, now I have Jas 10.4.9 with netkas 8.9.1.

Next week I'll install it on an Acer Travelmate 4002 (mine is 4220) I guess It also has 3945...

 

8-( there is no crying smiley!!!

dmesg.txt

ioreg.txt

NI.txt

system.txt

[ttorok]

iwi2100 latest version:

 

boot with -s

type sh etc/rc

and long scanning

and panic: pic attached.

 

so I do not have any chance to call networkselecor.

[ryuu123]

"no ALIVE INTR" means not occur(receive) interrpt

CSR_INT register == BIT_INT_ALIVE

 

when hot reboot from windows, no interrupt.

but cold boot ->

 

iwi3945: interrupt recieved 0x82000009 masked 0xc531f832 card mask 0xaa000003

 

-> 0x82000009

--> bit 31 25 3 0

 

#define BIT_INT_RX (1<<31)

#define BIT_INT_SWERROR (1<<25)

#define BIT_INT_ALIVE (1<<0)

 

hmm..

 

I'll check init and firmload.

 

>> for 3945 testers here,

plz try cold boot and hot boot,I want to see system.log.

[moscat]

Hi,

Are you using the scripts for getting the logs?

http://iwidarwin.googlecode.com/files/getlogs3.command.zip

I copied it to my home folder (The default path for Terminal), so I open Terminal > go into superuser "su" > enter root password (You have to enable it from NetInfo Manager) > and then ./getlogs3.command > find your logs in your Hard Disk folder.

Bye

 

no, i did not use the scripts, thanks for the advice.

maybee they should be added to the first post,

so newbees like me find them :pirate2:

 

thx, moscat

[cmanbrown]

when hot reboot from windows, no interrupt.

but cold boot ->

 

iwi3945: interrupt recieved 0x82000009 masked 0xc531f832 card mask 0xaa000003

 

-> 0x82000009

--> bit 31 25 3 0

 

#define BIT_INT_RX (1<<31)

#define BIT_INT_SWERROR (1<<25)

#define BIT_INT_ALIVE (1<<0)

 

hmm..

 

I'll check init and firmload.

 

>> for 3945 testers here,

plz try cold boot and hot boot,I want to see system.log.

 

Do you know any way to listen to commands sent from driver to device and vice versa?

 

Where are you getting these interrupts from? Darwin boot?

[asstastic]

Ah ha! I never thought to ask that question.Looks like AppleACPIPCI has no clue it's even there. Could be crazy Acer IRQ sharing gone awry?

Hey, good catch. I have seen in the ioreg's that the cards are not recognized, but I assumed it was an iwi software problem.

Do you think there are other people on this board already working on Acer ACPI? That would probably be the best place to start - pretty sure it's out of the realm of this project.

 

It will be intereseting to know what you find out

 

i doubt it is an irq issue. windows will notify you if there is a resource conflict like that and i didn't see anything in the device manager to sugest that it could be an issue. someone posted earlier that he had indeed ported acer acpi to OSX but when i sent him a pm he never responded and i haven't heard from him since then. chances are he's just a lying {censored}.

 

easiest way to test acer ACPI would be to do a cold boot into OSX from another os while the card is enabled. i haven't been able to try this yet as there is no way to do a cold reboot from XP or 2000. i tried from linux but couldn't get it to do a cold reboot either. if someone has a broken card and is willing to test this under linux i'd appreciate it. or the much easier solution might be to install win98 and try it from there(i think you down shift as you restart the computer to do a cold reboot).

 

unfortunately, after looking deeper into the ioreg outputs i don't think that the io registry can be hacked using a natit style kext to force card detection by the iwi drivers. all the additional ioreg info on the cards looks to be provided by the iwi driver itself as the cards show up under ioreg but are not listed as networking devices.

 

for refrence my card shows up under ioreg like this(extraneous info removed):

   | +-o PCI0@0  <class IOACPIPlatformDevice, registered, matched, active, busy 0, retain count 60>
   | | +-o AppleACPIPCI  <class AppleACPIPCI, !registered, !matched, active, busy 0, retain count 74>
   | |   +-o pci-bridge@1C,1  <class IOPCIDevice, registered, matched, active, busy 0, retain count 11>
   | |   | +-o IOPCI2PCIBridge  <class IOPCI2PCIBridge, !registered, !matched, active, busy 0, retain count 8>
   | |   |   +-o pci8086,4222@0  <class IOPCIDevice, registered, matched, active, busy 0, retain count 11>

 

a working card on a intel 915 chipset looks like this

   | +-o PCI0@0  <class IOACPIPlatformDevice, registered, matched, active, busy 0, retain count 41>
   | | +-o AppleACPIPCI  <class AppleACPIPCI, registered, matched, active, busy 0, retain count 41>
   | |   +-o RP03@1C,2  <class IOPCIDevice, registered, matched, active, busy 0, retain count 10>
   | |   | +-o IOPCI2PCIBridge  <class IOPCI2PCIBridge, registered, matched, active, busy 0, retain count 8>
   | |   |   +-o PXS3@0  <class IOPCIDevice, registered, matched, active, busy 0, retain count 10>
   | |   |     +-o darwin_iwi3945  <class darwin_iwi3945, registered, matched, active, busy 0, retain count 7>
   | |   |       +-o IOEthernetInterface  <class IOEthernetInterface, registered, matched, active, busy 0, retain count 10>
   | |   |         +-o IONetworkStack  <class IONetworkStack, registered, matched, active, busy 0, retain count 11>
   | |   |           +-o IONetworkStackUserClient  <class IONetworkStackUserClient, !registered, !matched, active, busy 0, retain count 5>

another example

   | |   +-o RP01@1C  <class IOPCIDevice, registered, matched, active, busy 0, retain count 10>
   | |   | +-o IOPCI2PCIBridge  <class IOPCI2PCIBridge, registered, matched, active, busy 0, retain count 8>
   | |   |   +-o PXS1@0  <class IOPCIDevice, registered, matched, active, busy 0, retain count 10>
   | |   |     +-o darwin_iwi3945  <class darwin_iwi3945, registered, matched, active, busy 0, retain count 7>
   | |   |       +-o IOEthernetInterface  <class IOEthernetInterface, registered, matched, active, busy 0, retain count 10>
   | |   |         +-o IONetworkStack  <class IONetworkStack, registered, matched, active, busy 0, retain count 9>
   | |   |           +-o IONetworkStackUserClient  <class IONetworkStackUserClient, !registered, !matched, active, busy 0, retain count 5>

 

i would bet that your card is listed right here in the ioreg JustJoe:

   | |   +-o RP01@1C  <class IOPCIDevice, registered, matched, active, busy 0, retain count 10>
   | |   | +-o IOPCI2PCIBridge  <class IOPCI2PCIBridge, registered, matched, active, busy 0, retain count 8>
   | |   |   +-o PXS1@0  <class IOPCIDevice, registered, matched, active, busy 0, retain count 9>

from all the other logs i have seen the entry here shouldn't end at the third line unless theres an unrecognized device on the pci bridge, ie your wireless lan.

 

as i just noticed, my laptop uses a mobile 945 chipset and the logs don't look anything like the ones from people using the mobile 915 chipset.

[ryuu123]

Do you know any way to listen to commands sent from driver to device and vice versa?

 

Where are you getting these interrupts from? Darwin boot?

 

perhaps "RXON" command.

 

I attach my system.log(boot time).

 

# some changes made from svn codes.

system.log.txt

[jalavoui]

finally manage to upload the iwi2200.dmg to svn

 

you can download from

http://code.google.com/p/iwidarwin/

[osxonmylaptop]

i doubt it is an irq issue. windows will notify you if there is a resource conflict like that and i didn't see anything in the device manager to sugest that it could be an issue. someone posted earlier that he had indeed ported acer acpi to OSX but when i sent him a pm he never responded and i haven't heard from him since then. chances are he's just a lying {censored}.

 

easiest way to test acer ACPI would be to do a cold boot into OSX from another os while the card is enabled. i haven't been able to try this yet as there is no way to do a cold reboot from XP or 2000. i tried from linux but couldn't get it to do a cold reboot either. if someone has a broken card and is willing to test this under linux i'd appreciate it. or the much easier solution might be to install win98 and try it from there(i think you down shift as you restart the computer to do a cold reboot).

 

unfortunately, after looking deeper into the ioreg outputs i don't think that the io registry can be hacked using a natit style kext to force card detection by the iwi drivers. all the additional ioreg info on the cards looks to be provided by the iwi driver itself as the cards show up under ioreg but are not listed as networking devices.

 

for refrence my card shows up under ioreg like this(extraneous info removed):

 

from all the other logs i have seen the entry here shouldn't end at the third line unless theres an unrecognized device on the pci bridge, ie your wireless lan.

 

as i just noticed, my laptop uses a mobile 945 chipset and the logs don't look anything like the ones from people using the mobile 915 chipset.

 

Hi all,

I'd like to do some test to help out, I guess that by cold start you mean to boot osx with the card not deactivated by XP. I can load some w98 or ReactOS on a Pendrive or External USB Disk, so then how can I reboot the system? That's because the laptops doesn't have reset button? Suspending or Hibernating doesn't help?

I'm trying to figure out how to reboot my computer without shutting down XP (it's ironic because XP frequently crashes I want to do it and I can't )

Thanks

[JustJoe]

i would bet that your card is listed right here in the ioreg JustJoe:

   | |   +-o RP01@1C  <class IOPCIDevice, registered, matched, active, busy 0, retain count 10>
   | |   | +-o IOPCI2PCIBridge  <class IOPCI2PCIBridge, registered, matched, active, busy 0, retain count 8>
   | |   |   +-o PXS1@0  <class IOPCIDevice, registered, matched, active, busy 0, retain count 9>

You got it... Intel 945GM mobile chipset

   | |   +-o RP01@1C  <class IOPCIDevice, registered, matched, active, busy 0, retain count 11>
   | |   | +-o IOPCI2PCIBridge  <class IOPCI2PCIBridge, !registered, !matched, active, busy 0, retain count 8>
   | |   |   +-o PXS1@0  <class IOPCIDevice, registered, matched, active, busy 0, retain count 11>
   | |   |     +-o darwin_iwi3945  <class darwin_iwi3945, registered, matched, active, busy 0, retain count 7>
   | |   |       +-o IOEthernetInterface  <class IOEthernetInterface, registered, matched, active, busy 0, retain count 10>
   | |   |         +-o IONetworkStack  <class IONetworkStack, registered, matched, active, busy 0, retain count 11>
   | |   |           +-o IONetworkStackUserClient  <class IONetworkStackUserClient, !registered, !matched, active, busy 0, r$

Yeah I don't have that problem, just tryin to help out those who do. You are a good resource, too.

 

PS - I noticed that devices on the 915 chipset are all matched and registered - what gives???

 

[asstastic]

osxonmylaptop, if you can get win98 up and running with the wireless drivers installed then just run this from the command line or a .bat file

C:\WINDOWS\RUNDLL.EXE C:\WINDOWS\SYSTEM\USER.EXE,ExitWindowsExec

your computer should do a 'warm' reboot meaning that the power doesn't cycle or anything. although i've never heard of a version of win98 you could boot from a flash drive.

[moseschrist]

Hi all,

I'd like to do some test to help out, I guess that by cold start you mean to boot osx with the card not deactivated by XP. I can load some w98 or ReactOS on a Pendrive or External USB Disk, so then how can I reboot the system? That's because the laptops doesn't have reset button? Suspending or Hibernating doesn't help?

I'm trying to figure out how to reboot my computer without shutting down XP (it's ironic because XP frequently crashes I want to do it and I can't )

Thanks

 

just use the reaset button, or press the shutdown button for like 4 seconds (more or less, depends on your motherboard) and it will just shutdown without actually shutting down XP.

[asstastic]

just use the reaset button, or press the shutdown button for like 4 seconds (more or less, depends on your motherboard) and it will just shutdown without actually shutting down XP.

 

that would put the computer into hybernation or suspend it depending on the settings. you can set all of those options from the power settings menu hidden in the screensaver tab(yet another reason to use OSX, bill why couldn't you have taken a hint from mac syspref and stuck everything in one window) what we need to do here to test the card by resetting the software only. any sort of power cycle where the laptop shuts off even if for a brief moment won't work.

[moseschrist]

what about pulling the battary out ?

[Airfly]

iwi3945...

 

I think I got interrupt before using networkSelector, and after (1) option calling in networkSelector and received some command like : (#3af539fc), seq:0x00A1, flags=0x8001, len = 0.

log lines start with "..." are I marked in iwi3945.cpp. dont know it usefull or not, Jalavoui.

 

that is too bad , but how to catch all log's lines when using dmesg or log all lines to system.log ? I saw a lots of log's lines on screen but only few of them was caught in dmesg.txt or system.txt.

dmesg1.txt

system1.txt

dmesg2.txt

system2.txt

[moseschrist]

jalavoui:

i'm having some trouble with the ieee80211 lib...

seems like that the ieee80211_crypt*.c files are havily depended on linux kernel's crypto lib...

and i have no idea how to make it work, i have tried some of my best copy&paste capabilities with lots of code changing to make it fit our needs and managed to port the ieee80211_crypt.c file but for the others (those who implement different crypto stuff like wep,wpa and others...) are a no go (at least for me) as i'm stuck here on a dead end...

[jalavoui]

iwi2100 rev 418

read previous posts on how to test

http://code.google.com/p/iwidarwin/

 

moseschrist,

post your changes to iwi2200. i'll try to do something but it realy looks hard to make the lib work

 

iwi3945

i believe the problem (as i see it in logs) is the radio on/off

can someone try to check this in iwi3945.cpp?

we need to find a way to turn the radio on (as i did for iwi2200) or the driver will never work

 

iwi3945 rev 419

i'll try a hack with 0x05c - maybe with the right value we can make it work - please help

test it with option (1) of networkselector.

 

 

can someone check this linux code? http://sourceforge.net/project/showfiles.php?group_id=108766

 

 

the code needs to be changed so that the driver does nothing until option (1) is pressed - i'm having this problem in iwi2100.

 

Airfly,

you have the right logs - maybe you can find a hack for this

 

about the iwi2200 0x30 hack

i've mailed the code to the guy who write the original linux code - he didn't answered back, maybe because the driver isn't expected to work this way - but it works!

so i believe we'll find a way to make it work for iwi3945

don't give up guys

 

i find out that some cards - this should be valid for most of them - need some time to switch the radio state. maybe this is the final solution for the problem.

[moseschrist]

i've mailed the code to the guy who write the original linux code - he didn't answered back, maybe because the driver isn't expected to work this way - but it works!

so i believe we'll find a way to make it work for iwi3945

don't give up guys

 

lol, lets just hope he won't think it's a bug and remove that "feature"

i will commit the changes to the new version and post...

 

i have a question, i need some help creating an installer for nsGUI. i need to make it so it will install the software and than make it so it will run at login time, anyone has any idea on how to do it ?

 

iwi2200 rev 420:

ieee80211_crypt.h functions added and implemented.