Jump to content

nForce LAN driver


eno
 Share

740 posts in this topic

Recommended Posts

I just know, that with Debian and Ubuntu my network card were workied very well, but I didn't try install any driver.

 

Ok, that confirms the forcedeth driver work and thus the problem is either something in your install or something in my driver (your log suggest the driver is working but I am still puzzled at the lack of the generation of timer interrupts).

Link to comment
Share on other sites

I just know, that with Debian and Ubuntu my network card were workied very well, but I didn't try install any driver.

 

My computer works like a EXtreMM@tor, but both lan cards dont work in iDeneb, very strange...ubuntu works very well.

 

eno, the ideneb, is a "OS X distro", the main causes of ethernet problems?

Link to comment
Share on other sites

My computer works like a EXtreMM@tor, but both lan cards dont work in iDeneb, very strange...ubuntu works very well.

 

eno, the ideneb, is a "OS X distro", the main causes of ethernet problems?

 

I don't know if the iDeneb distribution is the problem. I know that with my motherboard, which also has an MCP55 0x037310de, Leo4All, and Voodoo 1.0 there are no problems.

 

There are a couple of things I want to try to see if we can isolate the problem. I will post here once I am ready to start debugging this (probably later today).

Link to comment
Share on other sites

I'm tryed connect to second port and what i had in system.log

 

Last login: Sat Dec 20 12:54:30 on console

evg-rs-macpro31:~ evgr$ tail -f /var/log/system.log

Dec 20 12:54:30 evg-rs-macpro31 loginwindow[35]: USER_PROCESS: 35 console

Dec 20 12:54:30 evg-rs-macpro31 com.apple.launchd[1] (com.apple.UserEventAgent-LoginWindow[106]): Exited: Terminated

Dec 20 12:54:36 evg-rs-macpro31 /System/Library/CoreServices/coreservicesd[55]: SFLSharePointsEntry::CreateDSRecord: dsCreateRecordAndOpen(evg r's Public Folder) returned -14135

Dec 20 12:54:39 evg-rs-macpro31 com.apple.loginwindow[35]: ODUEthernetAddress(): GetEthernetAddress()== 5

Dec 20 12:54:41 evg-rs-macpro31 /System/Library/CoreServices/Finder.app/Contents/MacOS/Finder[129]: installAutoFSMonitor: open failed

Dec 20 12:54:42 evg-rs-macpro31 /System/Library/CoreServices/SystemUIServer.app/Contents/MacOS/SystemUIServer[127]: CPSGetProcessInfo(): This call is deprecated and should not be called anymore.

Dec 20 12:54:42 evg-rs-macpro31 /System/Library/CoreServices/SystemUIServer.app/Contents/MacOS/SystemUIServer[127]: CPSPBGetProcessInfo(): This call is deprecated and should not be called anymore.

Dec 20 12:55:08 evg-rs-macpro31 login[192]: USER_PROCESS: 192 ttys000

Dec 20 12:55:08 evg-rs-macpro31 /System/Library/CoreServices/Finder.app/Contents/MacOS/Finder[129]: CPSGetProcessInfo(): This call is deprecated and should not be called anymore.

Dec 20 12:55:08 evg-rs-macpro31 /System/Library/CoreServices/Finder.app/Contents/MacOS/Finder[129]: CPSPBGetProcessInfo(): This call is deprecated and should not be called anymore.

Dec 20 12:55:28 evg-rs-macpro31 login[203]: USER_PROCESS: 203 ttys001

Dec 20 12:56:08 evg-rs-macpro31 sudo[214]: evgr : TTY=ttys001 ; PWD=/Users/evgr ; USER=root ; COMMAND=/sbin/kextload -v /Users/evgr/Desktop/nForceLAN.kext

Dec 20 12:56:08 evg-rs-macpro31 kernel[0]: nForceLAN: Version 0.62.6n beta_osx initializing

Dec 20 12:56:08 evg-rs-macpro31 kernel[0]: nForceLAN: setting optimization mode: cpu

Dec 20 12:56:08 evg-rs-macpro31 kernel[0]: nForceLAN: found nVidia MCP55 Ethernet Controller with ID: 0x037310de Rev: 0xa3 (subsystem ID: 0x82391043)

Dec 20 12:56:08 evg-rs-macpro31 kernel[0]: nForceLAN: Version 0.62.6n beta_osx initializing

Dec 20 12:56:08 evg-rs-macpro31 kernel[0]: nForceLAN: setting optimization mode: cpu

Dec 20 12:56:08 evg-rs-macpro31 kernel[0]: nForceLAN: found nVidia MCP55 Ethernet Controller with ID: 0x037310de Rev: 0xa3 (subsystem ID: 0x82391043)

Dec 20 12:56:08 evg-rs-macpro31 kernel[0]: nForceLAN: Ethernet address 00:1e:8c:5e:55:2e

Dec 20 12:56:08 evg-rs-macpro31 kernel[0]: nForceLAN: Network link down.

Dec 20 12:56:08 evg-rs-macpro31 kernel[0]: nForceLAN: Ethernet address 00:1e:8c:5e:5e:2a

Dec 20 12:56:09 evg-rs-macpro31 kernel[0]: nForceLAN: Network link down.

Dec 20 12:56:10 evg-rs-macpro31 kernel[0]: nForceLAN: Link speed now 100Mbps(duplex = Full, flow control = Yes), code 0x10064.

Dec 20 12:56:25 evg-rs-macpro31 mDNSResponder[34]: mDNSPlatformSendUDP sendto failed to send packet on InterfaceID 01009600 en0/4 to 224.0.0.251:5353 skt 7 error -1 errno 65 (No route to host) 994280479

Dec 20 12:56:28 evg-rs-macpro31 mDNSResponder[34]: mDNSPlatformSendUDP sendto failed to send packet on InterfaceID 01009600 en0/4 to 224.0.0.251:5353 skt 7 error -1 errno 65 (No route to host) 994282803

Dec 20 12:56:52 evg-rs-macpro31 mDNSResponder[34]: mDNSPlatformSendUDP sendto failed to send packet on InterfaceID 01009600 en0/4 to 224.0.0.251:5353 skt 7 error -1 errno 65 (No route to host) 994307533

Dec 20 12:57:47 evg-rs-macpro31 login[203]: DEAD_PROCESS: 203 ttys001

Dec 20 12:58:14 evg-rs-macpro31 mDNSResponder[34]: mDNSPlatformSendUDP sendto failed to send packet on InterfaceID 01009600 en0/4 to 224.0.0.251:5353 skt 7 error -1 errno 65 (No route to host) 994388695

 

I'm not shure, that 0x037310de is an ID of my Ethernet Controller. If it isn't, could any one tell me where can I see it also?

Link to comment
Share on other sites

Alright, I think I found the hpet problem. It looks like on some motherboards the NIC shares a line interrupt with the hpet. What this means is the nForceLAN kext could possibly be stealing interrupts from the hpet (very bad). The interrupt filter is supposed to prevent this from occurring but it appears not to be working in this case (it works with usb/sata/etc sharing an interrupt with the NIC).

 

I am in the process of updating the code to support MSI/MSI-X (message signalled interrupts) to avoid problems with shared legacy interrupts. The new version will allow the user to override the new default (MSI/MSI-X) and use line interrupts if needed. Look for the MSI version sometime this week.

Link to comment
Share on other sites

Alright, I think I found the hpet problem. It looks like on some motherboards the NIC shares a line interrupt with the hpet. What this means is the nForceLAN kext could possibly be stealing interrupts from the hpet (very bad). The interrupt filter is supposed to prevent this from occurring but it appears not to be working in this case (it works with usb/sata/etc sharing an interrupt with the NIC).

 

I am in the process of updating the code to support MSI/MSI-X (message signalled interrupts) to avoid problems with shared legacy interrupts. The new version will allow the user to override the new default (MSI/MSI-X) and use line interrupts if needed. Look for the MSI version sometime this week.

Hi eno!

Thank you for the good idea. I found that USB 2.0 has the same problem too. At the moment I have two reports with successfully working USBEHCI after deleting AppleHPET and AppleIntelCPUPowerManagement (because it depends on HPET).

If you have any questions or propositions call me.

Regards!

Link to comment
Share on other sites

I've got an eVGA 680i, I'm using nForceLan.kext, and I'm using the ToH 9.2.0 kernel. My network adapters will work when I boot up, but after transferring a little bit of data, they stop being able to communicate with the network. If I do a ping -f to my router, i can see that about 490 packets are able to make it through before the adapter stops working. Interestingly, I can take the network cable out and put it in the second adapter that's integrated into the motherboard, and it will briefly work and then stop, exactly like the other one. I've tried the unplug for 30 seconds trick and it didn't work. If anyone has had this problem, I would be very interested to know how you resolved it. Thanks.

Link to comment
Share on other sites

I've got an eVGA 680i, I'm using nForceLan.kext, and I'm using the ToH 9.2.0 kernel. My network adapters will work when I boot up, but after transferring a little bit of data, they stop being able to communicate with the network. If I do a ping -f to my router, i can see that about 490 packets are able to make it through before the adapter stops working. Interestingly, I can take the network cable out and put it in the second adapter that's integrated into the motherboard, and it will briefly work and then stop, exactly like the other one. I've tried the unplug for 30 seconds trick and it didn't work. If anyone has had this problem, I would be very interested to know how you resolved it. Thanks.

 

I am now noticing the same thing on my MCP55 (this is good as now I can debug it) but not on my MCP79. So far I have determined (and you have confirmed) that something goes wrong when the NIC has gone through the RX ring exactly once (512 packets). I expect to have the problem fixed this weekend (Monday at the latest). Look for the Message Signaled Interrupt enabled release for the fix.

Link to comment
Share on other sites

First off - thanks a bunch for working so hard on this stuff. Most of this is way beyond me!

 

So I've been trying to get my MCP55 ports working but can't find anything good. After enough research - I figured out I kept forgetting to delete forcedeth kext so after that, i installed your nforceLan kext with osx86 tools and restarted. I did the 30 second trick and started with the -v argument. It shows both ports - and it shows one of 'em being up with 100mb full duplex.

 

I log into osx and everything works fine. I attempt to ping yahoo.com first and it doesn't work. I go into settings - I get my internal ip address 192.168 and all hte comcast information (DNS , etc..) but after a few minutes it just goes into a 169.x.x.x address. After a few more minutes the whole computer freezes.

 

I restart and try the other port and the same exact thing happens. I removed the kext reinstalled it again - same thing.

 

When I don't have any kexts for ethernet installed, my computer runs awesome for a long long time (4 days this week) with a bunch of {censored} running so I know it's the ethernet working.

 

I either want to get this running (i hope) or just settle for some other pci nic that I know will work but I see most people getting it to work in this forum so I really want it to work.

 

If you give me instructions please try to be as obvious as possible - I'm still trying to learn most of hte terminology and procedures for this stuff.

 

(btw - both nic's work fine on my windows xp 64bit installation).

 

My specs:

Asus Striker Extreme (680i)

Intel Q6600 2.4GHz

4gb of ram

Nvidia 8800gts 320mb

Soundblaster Live 5.1 sound card

3 sata hard drives (1 osx, 1xp, 1 a fat32 share between the os's)

1 sata dvd drive

 

iDeneb 10.5.5 (v1.3) installation - everything worked flawless from the installation

 

Thanks!

Link to comment
Share on other sites

The problem you describe is well known. It appears that some NICs (including the MCP55) need the rx buffer pointer to change after each use (buffers are reused in 0.62.4 - 0.62.6).

 

I am testing the fix right now. My MCP55 is working with full Message Signaled Interrupt support but it is currently taking about 4.5% of the cpu. I am trying to get the cpu utilization down and I will release 0.62.7n beta today even if utilization remains high.

 

First off - thanks a bunch for working so hard on this stuff. Most of this is way beyond me!

 

So I've been trying to get my MCP55 ports working but can't find anything good. After enough research - I figured out I kept forgetting to delete forcedeth kext so after that, i installed your nforceLan kext with osx86 tools and restarted. I did the 30 second trick and started with the -v argument. It shows both ports - and it shows one of 'em being up with 100mb full duplex.

 

I log into osx and everything works fine. I attempt to ping yahoo.com first and it doesn't work. I go into settings - I get my internal ip address 192.168 and all hte comcast information (DNS , etc..) but after a few minutes it just goes into a 169.x.x.x address. After a few more minutes the whole computer freezes.

 

I restart and try the other port and the same exact thing happens. I removed the kext reinstalled it again - same thing.

 

When I don't have any kexts for ethernet installed, my computer runs awesome for a long long time (4 days this week) with a bunch of {censored} running so I know it's the ethernet working.

 

I either want to get this running (i hope) or just settle for some other pci nic that I know will work but I see most people getting it to work in this forum so I really want it to work.

 

If you give me instructions please try to be as obvious as possible - I'm still trying to learn most of hte terminology and procedures for this stuff.

 

(btw - both nic's work fine on my windows xp 64bit installation).

 

My specs:

Asus Striker Extreme (680i)

Intel Q6600 2.4GHz

4gb of ram

Nvidia 8800gts 320mb

Soundblaster Live 5.1 sound card

3 sata hard drives (1 osx, 1xp, 1 a fat32 share between the os's)

1 sata dvd drive

 

iDeneb 10.5.5 (v1.3) installation - everything worked flawless from the installation

 

Thanks!

Link to comment
Share on other sites

Ok. It looks like cpu optimization uses a lot of cpu when MSI is enabled and now throughput optimization uses < 1% cpu. Go figure.

 

Anyway, I have released 0.62.7n beta. The default is to use MSI and throughput optimization. Please give it a try any let me know if there are any problems. I especially want to know if the HPET can now be enabled on those machines where it was causing crashes.

Link to comment
Share on other sites

hi eno...

i used all ur versions...but adding driverid to the plist in every version (its not a big job)...

according to u r driverid list there are 42 driver ids those r working but u have added only 19 ids to the plist...is there a perticular reason for this?or they will work without adding the driverid?or there is a limit to max no of driverids?.........

Link to comment
Share on other sites

hi eno...

i used all ur versions...but adding driverid to the plist in every version (its not a big job)...

according to u r driverid list there are 42 driver ids those r working but u have added only 19 ids to the plist...is there a perticular reason for this?or they will work without adding the driverid?or there is a limit to max no of driverids?.........

 

Some of the device IDs are masked. For example 0x037210de&0xfffeffff matches both 0x037210de and 0x037310de. I don't know why the masked IDs don't always match.

 

Sorry for the newb in me but - in case something is going wrong or the problem didn't get better - should I do anything - change any settings - or just let you know?

 

I have no idea how to access the settings like HPET or MSI or throughput optimization.

 

You shouldn't need to do anything besides add the new nForceLAN.kext to your /System/Library/Extensions , delete forcedeth.kext if it is installed, and reboot.

 

If you are still having problems PM me and I can walk you through another fix I have in mind.

Link to comment
Share on other sites

OMG Eno if I knew where you lived I'd so buy you a six pack. I'm on my osx installation right now and everything seems to work. I'm just going to set terminal to ping yahoo.com for a while and we'll see if it'll keep working. I'll turn on a download too and see if that works too - but for now we are all good.

 

Thank you again!

Link to comment
Share on other sites

  • Allan unpinned this topic
 Share

×
×
  • Create New...