High Sierra native support for 10Gb ethernet

[Kris Kelvin]

Kris, any luck with this?

Too much going on to put time into this. I've returned my adapter.
However, I believe it is insufficient to disable the boot ROM, as the card doesn't work in Hackintoshes either, as documented here:
http://www.insanelymac.com/forum/topic/330614-10-gbe-native-support-in-high-sierra-10132/

It's also entirely possible that the iMac Pro ships with a different version of the Aquantia kernel extension. Looking forward to 10.13.3...

 

[d5aqoëp]

Any progress with kext patch?

Offtopic:

For those who want to try V2 drivers for Aquantia based cards for Windows, I have extracted these drivers from iMacPro1,1 bootcamp installer package.

Windows 10 X64 only
http://www12.zippyshare.com/v/lAckTVG1/file.html

 

[Squuiid]

Any progress with kext patch?

Offtopic:

For those who want to try V2 drivers for Aquantia based cards for Windows, I have extracted these drivers from iMacPro1,1 bootcamp installer package.

Windows 10 X64 only
http://www12.zippyshare.com/v/lAckTVG1/file.html

LOL. Nice! Thanks for sharing.

 

[mikeboss]

at least two got the card to work. it seems that the Boot Camp driver from iMac Pro changes the device ID ->

http://www.insanelymac.com/forum/to...7-10-gbe-native-support-in-high-sierra-10133/

 

[Squuiid]

it seems that the Boot Camp driver from iMac Pro changes the device ID

What do you mean by this? The driver runs a firmware update to change the device ID? That would be odd/unlikely!

 

[mikeboss]

What do you mean by this? The driver runs a firmware update to change the device ID? That would be odd/unlikely!

that's exactly what to OP in the thread of the insanelymac forum wrote. he also stated that the ASUS driver didn't work anymore in MS-Windows because of this. he had to use the Boot Camp driver.

 

[Kris Kelvin]

If I understood correctly, it was actually the update to macOS 10.13.3 Beta that supposedly updated the Aquantia chip's firmware. And yes, that IS strange.

Besides, there's still the problem of the BootROM that prevents it from working in a cMP.

 

[Squuiid]

Apparently the 10.13.3 update does indeed make this card work however, as Kris has stated, the BootROM is still very much an issue in a cMP. It must be disabled for the card to post in a cMP.

Has anyone made any progress in disabling the BootROM? I contacted Aquantia and they were less than helpful. Rude in fact. Definitely a dead end.

 

[theitsage]

Kris, you could try to short pins 4+5 of the EEPROM while powering on the Mac Pro. this will prevent the system from reading out the ROM of the network card. back in the time when I was flashing EFI ROMs to PC graphics cards, this allowed me to boot the Mac Pro even if it played dead because I really f***ed up the ROM of the card...

https://forums.macrumors.com/threads/how-to-save-your-ati-5770-gpu-after-a-bad-bios-flash.1894948/

You're a legend dude. I located the EEPROM [MX25L3233F] and confirmed pin 4 & 5 orientation prior to soldering. A small piece of thin wire did the job. As far as software, I mounted the NIC in a Thunderbolt enclosure first to install 10.13.3 combo update to make sure the firmware gets flashed. Here's what I'm seeing in a Mac Pro tower [ASUS XG-C100C (AQC107)].

 

[Kris Kelvin]

I located the EEPROM [MX25L3233F] and confirmed pin 4 & 5 orientation prior to soldering. A small piece of thin wire did the job.

Great job! Did you leave the wire on after booting? Does that mean one could just remove the EEPROM chip to make the card work in a Mac Pro tower? Were you successful in establishing an Ethernet link?

 

[mikeboss]

yesss, finally! wonderful news! great to see that you got it working. congrats!

@Kris:

yes, removing the EEPROM chip altogether should give you the same result. shorting these pins completely blocks the ROM from being read out.

 

[theitsage]

Great job! Did you leave the wire on after booting? Does that mean one could just remove the EEPROM chip to make the card work in a Mac Pro tower? Were you successful in establishing an Ethernet link?

The soldered wire shorting out pin 4 & 5 stays put during operation. I wanted to test first prior to doing more evasive operation on the NIC. Next step is to remove the EEPROM completely per @mikeboss' confirmation.

I took a couple of screen captures to show the difference in macOS' detection of this card (10.13 on the left, 10.13.3 on the right):

 

[Kris Kelvin]

In macOS 10.13.3, there's a new kernel extension called AppleEthernetAquantiaAqtionFirmware.kext in /System/Library/Extensions/IONetworkingFamily.kext/Contents/PlugIns/.

That might mean that AQC107 cards installed into a Mac running 10.13.3 will be upgraded automatically (not just during the 10.13.3 upgrade).

That being said, AQC107 cards seem to be getting rare already...

 

[Squuiid]

In macOS 10.13.3, there's a new kernel extension called AppleEthernetAquantiaAqtionFirmware.kext in /System/Library/Extensions/IONetworkingFamily.kext/Contents/PlugIns/.

That might mean that AQC107 cards installed into a Mac running 10.13.3 will be upgraded automatically (not just during the 10.13.3 upgrade).

This is indeed what is happening it would seem.
http://www.insanelymac.com/forum/to...be-native-support-in-high-sierra-10133/page-3
[doublepost=1517315424][/doublepost]

Next step is to remove the EEPROM completely per @mikeboss' confirmation.

theitsage, were you able to give this a shot yet? Any chance of a photo of the chip before and after? Does the red heatsink on the ASUS card need to be removed first or is it located on the back?
Thanks.

 

[pastrychef]

I think that the firmware is "updated" on each reboot of macOS. It seems like it's a soft update and nothing is being flashed. I believe Bluetooth firmware is handled the same way.

My Gigabyte GC-AQC107 has been working perfectly since the official macOS 10.13.3 update on my hackintosh.

 

[Squuiid]

after having a look at the linux driver for the ASUS XG-C100C I'd say it's almost 100% certain that this card will work with macOS 10.13.2

http://dlcdnet.asus.com/pub/ASUS/wireless/XG-C100C/DR_XG_C100C_5005_Linux.zip

Has been confirmed to work. Prob best to edit your 1st post? Many will look at just that and see that it isn't.
[doublepost=1517319916][/doublepost]

I think that the firmware is "updated" on each reboot of macOS. It seems like it's a soft update and nothing is being flashed. I believe Bluetooth firmware is handled the same way.

Best way to check is to put it in a PC or Linux box and see what the device ID is after the supposed 'update'.
This approach would make much more sense however. I can't see Apple applying firmware like this on the fly. I could be wrong though!

 

[mikeboss]

Prob best to edit your 1st post?

done!

 

[Squuiid]

Here we go. Close-up shots of the XG-C100C including EEPROM.
http://www.52asus.com/thread-3845-1-1.html

 

[ydeng]

Has been confirmed to work. Prob best to edit your 1st post? Many will look at just that and see that it isn't.
[doublepost=1517319916][/doublepost]
Best way to check is to put it in a PC or Linux box and see what the device ID is after the supposed 'update'.
This approach would make much more sense however. I can't see Apple applying firmware like this on the fly. I could be wrong though!

Looked at the assembly yesterday, definitely NVRAM update. There is a bunch of spi functions not present in Linux driver.
Only done the first time if the version provided by firmware kext is newer. People at insanelymac has confirmed that after the update, stock ASUS windows driver does NOT work anymore.

 

[pastrychef]

The Gigabyte drivers works for me in Windows 10.

 

[pastrychef]

This is what happens on every reboot for me:

 

[ydeng]

The line right above compares firmware version. Because your existing firmware is newer, there is no update. You can remove AppleEthernetAquantiaAqtionFirmwareDev.kext the message will be gone and card still work. The turned on the debug print in that kext. Everytime the drive loads, it compares the firmware version and try to upload. AppleEthernetAquantiaAqtionFirmware.kext does the same thing without verbose log.

 

[Kris Kelvin]

Does anyone actually have a working card in a Mac Pro? So far, I've only seen success reports from Hackintosh users.

I've (unsuccessfully) been trying to upgrade an ASUS ROG Areion. Behavior in a MacPro5,1 as follows:

Vendor ID: 0x1d6a (Aquantia Corp.)
Device ID: 0xd107 (AQC107 NBase-T/ IEEE 802.3bz Ethernet Controller [AQtion])
Subsystem Vendor: 0x1043 (ASUSTeK Computer Inc.)
Subsystem ID: 0x872e (different from the XG-C100C, which has 0x8741)
Version: 0x0002

• EFI Boot Manager: Doesn't work (just a white screen); LEDs on card light up
• Regular boot: Works fine with card installed; LEDs on card don't light up
• Driver is loaded (AppleEthernetAquantiaAqtion.kext)
• Shows bogus MAC address (02:03:93:5b:9c:01)
• Establishing an Ethernet link is not possible

 

[Squuiid]

Does anyone actually have a working card in a Mac Pro? So far, I've only seen success reports from Hackintosh users.

I've (unsuccessfully) been trying to upgrade an ASUS ROG Areion. Behavior in a MacPro5,1 as follows:

Vendor ID: 0x1d6a (Aquantia Corp.)
Device ID: 0xd107 (AQC107 NBase-T/ IEEE 802.3bz Ethernet Controller [AQtion])
Subsystem Vendor: 0x1043 (ASUSTeK Computer Inc.)
Subsystem ID: 0x872e (different from the XG-C100C, which has 0x8741)
Version: 0x0002

• EFI Boot Manager: Doesn't work (just a white screen); LEDs on card light up
• Regular boot: Works fine with card installed; LEDs on card don't light up
• Driver is loaded (AppleEthernetAquantiaAqtion.kext)
• Shows bogus MAC address (02:03:93:5b:9c:01)
• Establishing an Ethernet link is not possible

I’m assuming you shorted out the bootROM.

On insanelymac forum Hackintosh users have stated that they could not get a link when another NIC was enabled in their system.
As soon as they disabled their other NICs the Aquantia card could then get a link and IP.
Rather troublesome for a cMP I would think.

 

[Kris Kelvin]

I’m assuming you shorted out the bootROM.

Actually no. The Mac Pro booted without issue with the Areion installed (without any modifications).

On insanelymac forum Hackintosh users have stated that they could not get a link when another NIC was enabled in their system.

I saw; I've tried disabling the respective Intel kext, but that didn't help. It's not entirely clear under which circumstances the firmware upgrade takes place.