Jump to content

AppleIntelE1000e.kext for 10.8/10.7/10.6/10.5


  • Please log in to reply
698 replies to this topic

#441
ila225

ila225

    InsanelyMac Protégé

  • Members
  • PipPip
  • 56 posts
  • Gender:Male

 

I did remove the GenericUSB kext

 

So, here's my results, after performing the commands you requested:

 

 

beastie:~ mm2margaret$ ls -l /System/Library/Extensions/IONetworkingFamily.kext/Contents/PlugIns/ | grep E1000
drwxr-xr-x  3 root  wheel  102 Nov  6 16:28 AppleIntelE1000e.kext
 
And:
 
beastie:~ mm2margaret$ sudo touch /System/Library/Extensions
beastie:~ mm2margaret$ sudo kextcache -u /
 
(meaning no errors, I think, because it didn't kick out anything....)
 
And, sadly, still no luck.....still getting a 169.254 address, and even a valid manual ip address assigned, I cannot ping it's default gateway, while the realtek can (if it's plugged in via an ethernet jack).
 
 

 

That is ok, one thing at a time.

 

Although your card was loaded, I wouldn't be surprised of weird behavior by the OS given the problems with permissions or cache. Granted, you could be booting without caches already, but it doesn't hurt to be sure.

 

I'm assuming you can still find your card being loaded properly and everything.

 

Next step you could try is to reset your network interface settings, by moving the config file out of the system configuration folder and then restarting:

mv /Library/Preferences/SystemConfiguration/NetworkInterfaces.plist ~/Desktop

Try that, see if it helps.



#442
meaganmargaret

meaganmargaret

    InsanelyMac Protégé

  • Members
  • Pip
  • 18 posts
  • Gender:Female

Nope.  Still getting a 169.254 address.  That did disable my Realtek, however.  I'm not terribly worried about that, however.  And, also, fyi, I'm not using the Kernal Cache, either...and this is, for sure, driver version 2.4.14.



#443
ila225

ila225

    InsanelyMac Protégé

  • Members
  • PipPip
  • 56 posts
  • Gender:Male

Now I'm running out of ideas... =/



#444
Konner

Konner

    Cat

  • Members
  • PipPipPip
  • 163 posts
  • Gender:Male

I wanna go ahead and post here, I have constant connection drops when downloading and trying to listen to music or browse the web. I've tried different versions and still the same problem.

 

I'm using Mavericks, and it's really getting annoying having to reboot to get internet back. can someone please help me?

 

the device is 

Intel® 82583V Gigabit Ethernet Controller

#445
meaganmargaret

meaganmargaret

    InsanelyMac Protégé

  • Members
  • Pip
  • 18 posts
  • Gender:Female

So, if you're desparate, go out and buy a realtek add-in card, like I did, of the 8111 variety and take the driver from this site and install it in IONetworking.  It worked for me.

 

But please don't ask me for tech support, because I was barely able to get this to work, I'm just not technically competent beyond the very, very basics.   But I managed it, and now, at least I have a gigabit ethernet adapter that works, and the costs was only about $15.   The time and frustration alone troubleshooting the built-in Intel adapter far exceeded a $15........not that I would stop troubleshooting the built-in Intel card, but I think we've reached a dead end here....



#446
Mieze

Mieze

    Giant Cat

  • Coders
  • 548 posts
  • Gender:Female
  • Location:Germany
  • Interests:Cats

It's really fatal, which means that kernel failed in locking physical memory for output buffer. I have never seen that.

Hey hnak,

 

maybe I can help you to fix the problem because I have already seen and successfully fixed that problem in my Realtek driver. getPhysicalSegmentsWithCoalesce() tends to fail when the maximum number of segments is too low so that the system tries to coalesce the buffer by copying the fragments to a large, contiguous buffer. As mbufs for TSO operations may be highly fragmented choosing 32 for the maximum number of segments is not enough. After experimenting I found out that 40 eliminated the problem.

 

Mieze



#447
Gismo

Gismo

    InsanelyMac Protégé

  • Members
  • PipPip
  • 56 posts
  • Gender:Male

 

I wanna go ahead and post here, I have constant connection drops when downloading and trying to listen to music or browse the web. I've tried different versions and still the same problem.

 

I'm using Mavericks, and it's really getting annoying having to reboot to get internet back. can someone please help me?

 

the device is 

Intel® 82583V Gigabit Ethernet Controller

 

Hello hnak,

 

Can you fix this Problem in Mavericks 10.9. I have the same problem.

 

Regards



#448
bcc9

bcc9

    InsanelyMac Legend

  • Coders
  • 1,278 posts
  • Gender:Male

I was running 2.4.14 fine under 10.8, but after upgrading to 10.9, the ethernet breaks after about 1 day of uptime. The system begins flooding the LAN with packets, causing collisions and saturating my gig-e hub. I just upgraded to 2.5.4d, hope that works better.

When the ethernet is malfunctioning, the display is also unresponsive so I can't really debug it further.

Ok, it took a lot longer for the bug to re-occur for me with the current release, but now it has, failing in the same way as the old version. Pretty nasty bug. And, it looks like this driver is no longer being actively maintained?
I have experience fixing ethernet drivers under *nix, but if this driver is going to continue being unmaintained, then the recommendations to switch to a pci-e gig-e card seem like good advice.

However I would disagree that a realtek nic or a usb ethernet solution are good choices - realtek doesn't have a great track record with their open source driver support (look at the bug history of the linux realtek driver if you want more info; basically you have a 3rd party user trying to guess the correct tweaks to support new hardware, with no public hardware specs).

So for users not willing to beta test this driver, I would recommend switching to an intel gig-e nic, one of the models currently supported natively by OSX. Looks like an intel 82574L based NIC, as mentioned by hannibal&karlson12 would be the cheapest solution. Such as the intel EXPI9301CTBLK which can be had for about $30.

... unless someone is working on stabilizing this driver for osx 10.9?

#449
Mieze

Mieze

    Giant Cat

  • Coders
  • 548 posts
  • Gender:Female
  • Location:Germany
  • Interests:Cats

Ok, it took a lot longer for the bug to re-occur for me with the current release, but now it has, failing in the same way as the old version. Pretty nasty bug. And, it looks like this driver is no longer being actively maintained?
I have experience fixing ethernet drivers under *nix, but if this driver is going to continue being unmaintained, then the recommendations to switch to a pci-e gig-e card seem like good advice.

However I would disagree that a realtek nic or a usb ethernet solution are good choices - realtek doesn't have a great track record with their open source driver support (look at the bug history of the linux realtek driver if you want more info; basically you have a 3rd party user trying to guess the correct tweaks to support new hardware, with no public hardware specs).

So for users not willing to beta test this driver, I would recommend switching to an intel gig-e nic, one of the models currently supported natively by OSX. Looks like an intel 82574L based NIC, as mentioned by hannibal&karlson12 would be the cheapest solution. Such as the intel EXPI9301CTBLK which can be had for about $30.

... unless someone is working on stabilizing this driver for osx 10.9?

I checked the source code during the last days and found a number of serious bugs with regard to transmission queue handling. It's not a wonder that the driver fails because it's unable to recover from a stalled queue.

 

Frankly, I'm not sure if it's better to fix the driver or start from scratch. Judging from my experience I think that it's saver and more efficient to reimplement all the code involved with packet transmission and reception than to warp some compatibility code around a linux driver because OS X's driver architecture differs a lot from linux. Only the hardware initialization is the same.

 

Although Realtek's policy not to provide any documentation without NDA is a major disadvantage, their gigabit NIC's aren't bad at all. They might lack a few of the advanced features of Intel or Broadcom NIC's but they work reliably  and fast, provided that the driver knows of their specific bugs. Yes, it's frustrating to have to reverse engineer a Windows driver in order to learn how to write good an OS X driver, but that doesn't matter if you are a user as it only affects the driver developer so that there is no reason not to buy a Realtek add-in card.

 

I'm currently making plans for my next project, a Haswell build and in case I decide to buy a board with an Intel NIC, I will need a reliable driver for it. Maybe I'll take care of the driver after the Christmas holidays?

 

Mieze



#450
ila225

ila225

    InsanelyMac Protégé

  • Members
  • PipPip
  • 56 posts
  • Gender:Male

...

I'm currently making plans for my next project, a Haswell build and in case I decide to buy a board with an Intel NIC, I will need a reliable driver for it. Maybe I'll take care of the driver after the Christmas holidays?

 

Mieze

Oh we all hope you have the time for it. =)



#451
bcc9

bcc9

    InsanelyMac Legend

  • Coders
  • 1,278 posts
  • Gender:Male
Historically, the realtek open source driver has been a bit unstable, with driver bugs including years worth of SMB protocol performance issues, poor jumbo frame support, stuck queues, etc. Meanwhile the intel 10/100/1000 driver was stable.

I think the stark difference in stability has been a direct byproduct of the different way these two companies handle driver support, NDA headaches from realtek being a part of that. While the realtek driver has seemed to finally reach stability under linux a few years ago, apparently the support methodology has remained unchanged.

First party intel driver source is available as soon as new hardware comes out, but for realtek, the driver source normally lags behind. I think it's rather telling about realtek's commitment that their last OSX driver release was in 9/2011.

On top of this, under OSX, the OS provides native, officially supported drivers for some intel nics and no native drivers for realtek ethernet.

This combined with the driver support issues, and the historical track record, are why IMO buying a realtek nic to get a stable ethernet driver is not the wisest choice, especially given how cheap some intel nics are.

That's not to say your realtek osx driver isn't a good one, I don't have reason to believe it isn't.

Should an end user need to be bothered with these details? Certainly not, but the details affect which choice is wisest for them nonetheless.

I agree that for the intel e1000e, osx support should be quite fixable (given that intel has a stable working open source solution available for reference).

I welcome your efforts. It's quite frustrating that the best solution at the moment appears to be to buy a NIC, given that the on-board hardware and intel open source are perfectly capable already.

#452
Mieze

Mieze

    Giant Cat

  • Coders
  • 548 posts
  • Gender:Female
  • Location:Germany
  • Interests:Cats

Historically, the realtek open source driver has been a bit unstable, with driver bugs including years worth of SMB protocol performance issues, poor jumbo frame support, stuck queues, etc. Meanwhile the intel 10/100/1000 driver was stable.

I think the stark difference in stability has been a direct byproduct of the different way these two companies handle driver support, NDA headaches from realtek being a part of that. While the realtek driver has seemed to finally reach stability under linux a few years ago, apparently the support methodology has remained unchanged.

First party intel driver source is available as soon as new hardware comes out, but for realtek, the driver source normally lags behind. I think it's rather telling about realtek's commitment that their last OSX driver release was in 9/2011.

 

 

 

Realtek makes inexpensive semiconductors for the Windows mass market. That's their business and that's why they have a business culture of the Microsoft style while Intel has shown a strong commitment to linux during the last decade. Nevertheless the progress of their linux driver during the last years is remarkable.

 

On top of this, under OSX, the OS provides native, officially supported drivers for some intel nics and no native drivers for realtek ethernet.

 

I don't remember Apple has ever used a Realtek NIC. Although they published the source of a driver for the RTL8139 as sample code years ago, I see no reason why they should write a driver for chips they don't use. Besides, the number of Apple supported Intel chips is limited and taking the price into account it boils down to 82574L based cards.

 

This combined with the driver support issues, and the historical track record, are why IMO buying a realtek nic to get a stable ethernet driver is not the wisest choice, especially given how cheap some intel nics are.

 

Well, Intel NICs have their bugs too, in particular the 82574L (used on cheap add-in cards) which is prone to receiver deadlocks.

 

That's not to say your realtek osx driver isn't a good one, I don't have reason to believe it isn't.

 

But I have: 8 month of stable operation on my OS X home server (24/7 uptime) without a single loss of network connectivity.

 

Should an end user need to be bothered with these details? Certainly not, but the details affect which choice is wisest for them nonetheless.

Yes of course, at least when you are using a hackintosh you should know what you are doing as it's your own responsibility to get a stable working machine. Otherwise I would recommend to go to the next Apple Store and pick a real Mac.

 

Mieze



#453
bcc9

bcc9

    InsanelyMac Legend

  • Coders
  • 1,278 posts
  • Gender:Male
Glad you agree.
Now back to the bug reports :)

#454
meaganmargaret

meaganmargaret

    InsanelyMac Protégé

  • Members
  • Pip
  • 18 posts
  • Gender:Female

So, I want to state that I'm more than willing to beta test any 1000e driver.  I posted about the realtek only because I was able to get it to work.  I needed at least one wired ethernet port to work, and I have that.

 

So, if any of you wants people willing to assist in beta testing drivers for 1000e, I'm still in...



#455
frankiee

frankiee

    InsanelyMac Geek

  • Members
  • PipPipPipPip
  • 217 posts
  • Gender:Male
  • Location:Earth
  • Interests:Everything
kernel[0]: failed to getphysicalsegment in outputPacket.
 
I also got this twice.
 
Can also confirm problems with large network copies, in my case: Finder Error -36, no lockup though.


#456
Neso

Neso

    InsanelyMac Protégé

  • Members
  • Pip
  • 34 posts

it's my first post here, but i had to because i had problem with my intel network adapter from ASUS rampage III gene. After using AppleIntelE1000e.kext for 2.14 up to 2.5, i had network connection, but on wake from sleep finder would froze, only the cursor was moving. Same would happen if i chose network adapter and made it inactive in system preferences.After searching for two month (because i needed sleep function), today i tried with Intel82566MM.kext seeing it had mu device id: 10ce in the .plist. Now sleep is working fine, inactive also works and i have network support. So if you have same dev_id or supported dev_id and have problems with sleep try this kext. I haven't tried yet but it says that this kext doesn't support some function of vmware so read before using.  



#457
questor42

questor42

    InsanelyMac Protégé

  • Members
  • Pip
  • 2 posts
hnak, thanks for your work on the ethernet driver.
 
I have a Z77X-UD5H which has both an Intel GbE (AppleIntelE1000e-v2.5.4d) and Atheros GbE (ALXEthernet-v1.0.2).
 
I'm running 10.9.0.  WOL is working on the Atheros, but not working on the Intel.  The Atheros driver source is available here:  http://code.google.c.../downloads/list
 
Thanks again!


#458
tele

tele

    InsanelyMac Protégé

  • Members
  • Pip
  • 33 posts

here running osx 10.9.0 and i have gigabyte ga-z87x-oc with Intel GbE LAN chipset.

after starting up the DHCP does NOT connect to my Netgear DG834gt, and it appears as IP 169.254.x.x

 

if i disconnect and reconnect the ethernet cable from the router i get correct DHCP IP 192.168.0.x
internet seems to work but IT'S VERY SLOW, and after shut down i have same issue when start up the computer again.
 
my network specification taken from system profiler:

 

Type: Ethernet Controller

  Bus: PCI

  Vendor ID: 0x8086

  Device ID: 0x1503b

  Subsystem Vendor ID: 0x1458

  Subsystem ID: 0xe000

  Revision ID: 0x0004

  BSD name: en0

  Kext name: AppleIntelE1000e.kext

  Location: /System/Library/Extensions/IONetworkingFamily.kext/Contents/PlugIns/AppleIntelE1000e.kext

  Version: 2.4.14



#459
genzai

genzai

    InsanelyMac Geek

  • Members
  • PipPipPip
  • 169 posts
  • Gender:Male
  • Location:San Francisco

 

  Version: 2.1.14

Why don't you start by using a newer version of the driver, especially since your on the newest intel chips. 2.4.14 is what most people find to be the most stable. You may also try 2.5.x and see if it works for you.

g\



#460
tele

tele

    InsanelyMac Protégé

  • Members
  • Pip
  • 33 posts

hey genzai

i reading you must reset kext cache and repair permission to replace a kext file, then restart and install the ethernet driver again… 

so i must delete the 2.4.14 kext from beast 6.0.1 and install the Network hnak's AppleIntelE1000e v2.5.4d from 6.0.0  ?

 

do you suggest to do this manually using app like kext wizard  ?

actually i'm a total noob, so could you explain how to "try" most recent ethernet driver without messing up all the system ?

if is possible, you can help me here by chat or PM ?

 

thanks a lot for you advice







1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users

© 2014 InsanelyMac  |   News  |   Forum  |   Downloads  |   OSx86 Wiki  |   Mac Netbook  |   PHP hosting by CatN  |   Designed by Ed Gain  |   Logo by irfan  |   Privacy Policy