Marvell (Aquantia) 10 Gb Ethernet support thread

cosmos654 · August 12, 2018

A brand new RJ45 10Gb NIC card based on same Aquantia chip by Synology with 5 YEARS warranty and (it semms) better heatsink:

https://www.synology.com/en-uk/products/E10G18-T1

August 12, 2018

1 hour ago, meaganmargaret said:

Well on another thread on a different forum, I mentioned that my new Sonnet card had failed and I returned it. I haven't ordered another one, so I can't help you in this regard.

Maybe you get just a bad one... mine works perfect, no issues at all

d5aqoep · August 23, 2018

On 8/12/2018 at 8:32 PM, DSM2 said:

Maybe you get just a bad one... mine works perfect, no issues at all

Hi dude! Can you let me know if WOL (Wake on LAN) works on your Sonnet Card in Windows?

You can use any WOL app found on Google's Play Store or Apple's App Store depending on your mobile to wake your system up.

Allubz · September 5, 2018

Dear peeps,

I'm at a loss with my ASRock X299 XE board that has the AQC107 chip. I see user jya claims to have it running on this board, and I believe (s)he does, because I managed to get it to work in 10.13.3 and Windows (forcefully installed the iMac Pro bootcamp drivers, before installing any other Aquantia driver). I just can't get it to work in 10.13.6 (latest). I tried both Mieze's patches (original and V3), but to no avail.

In verbose loading 10.13.3 I clearly see there is a firmware check, and it works. In verbose 10.13.6 this doesn't happen, and there is no Aquantia related information to be found anywhere (network settings, IOReg). I tried resetting the network settings a few times, but no luck there either.

Am I missing something? Can I help by uploading some files so people can check? Like I said, IOReg is empty, and the patches are copy/pasted from this thread (doublechecked 'm).

Does the firmware flashing only work when updating from 10.13.2 to 10.13.3, or with any 10.13.3 installation?

I did a direct install of 10.13.3 on one of the two identical boards I have here. Due to some mishap the AQC107 was flashed badly by macOS, leading to Windows not being able to start or install at all. A BIOS flash didn't recover this, as the Aquantia chip itself is flashed. I contacted ASRock about this.

Since I had a second board I went ahead and tried the process again, but this time without first installing the Windows drivers, and making sure everything happened in UEFI mode. As far as I can tell the second X299 board isn't flashed, or it went right this time. The current verbose output I get is as follows: https://i.imgur.com/X10h6sy.jpg

10.13.3 works, Windows works (with forcefully installed iMac Pro drivers), but there's no show at all of anything in macOS, IOReg, verbose boot, etc. of the Aquantia chip in 10.13.6 or 10.14.beta

My EFI's are differing, because I use an iMac18,3 definition to get into 10.13.3 (also 10.13.6 possible) and iMac Pro for 10.13.6 usually. The rest of the EFI folder is identical, of which probably only Mieze's patches matter. I can post it tomorrow if needed.

I also testes the lastes Mojave beta, no luck with either two patches.

Edited September 5, 2018 by Allubz

d5aqoep · September 7, 2018

Just installing 10.13.3 is enough for it to get flashed. No need to upgrade from 10.13.2. But the flashed firmware gets active only after you power off (shutdown) your PC completely and turn off it's power from wall's power socket. Then turn it back on.

Regarding the faulty 1st motherboard, you need to RMA it to ASROCK saying "it is not working".

Apple has recently updated Aquantia drivers for iMac Pro and they don't work with our firmware flashed cards anymore in Windows. Yo have to rely on Old drivers. So at some point these flashed cards might not work in Windows as support for older drivers gets dropped.

Everyone's best bet is to buy a Sonnet card.

Allubz · September 11, 2018

In regard to my problem with the ASRock X299 XE motherboard: It turns out I probably mistyped one of the patches anyway, because it works fine now in 10.13.6 on both motherboards I have here.

I can advise setting the motherboard to non-CSM (UEFI-mode), and not installing any Windows Aquantia drivers, before flashing with 10.13.3. It bricked Windows functionality on the board I took the other route, whereas the second motherboard works fine in Windows (iMac Pro drivers) and any modern macOS. The Windows problems are severe, as the UEFI firmware data of the AQC107 gets corrupted, not allowing Windows to boot, be it either preinstalled, or as installer, or as post-final installer. Ididn't try Live USB.

So, typical lesson: Triple-check your patches if everyone says this is how it works

d5aqoep · September 11, 2018

I managed to successfully flash back the stock firmware on one of my Aquantia card. Firmware files are available on station-drivers website. I plan to keep it that way because it is going to be used in a Windows 10 Only based system. Just make sure to install the Diagnostic driver for your Aquantia card first.

The other baptized card I will continue to use till it stops working. It is still working in latest DP10 of Mojave.

d5aqoep · September 12, 2018

What I did:

1. Extract the Firmware Directory to your Windows 10 desktop. Download AQC_Config.zip from this post and extract the 2 config files into the Firmware Dir on your Desktop.

2. Open AQC107-Config.aqc with notepad and make the necessary changes to the following: The below are the default values for my card. You have to download official drivers from Aquantia card and open the inf file to find your own values pertaining to your manufacturer.

FW_FILE: 'C:\Users\Your Username\Desktop\aquantia_firmware\firmware\asr_d107_1.5.58.clx'
PCI Device ID: 0xd107
PCI Subsystem ID: 0x1 (Mine is 0001 but you need to drop the zeroes if they are first. 
PCI Subsystem Vendor ID: 0x1d6a

(Eg. If your ASRock PCI Device ID is 07B1 you write PCI Device ID: 0x7b1)
(Adjust the firmware path to the actual clx file which contains the firmware. If you are using 10 Gigabit card, choose d107 file. If you are using 5 Gigabit card, choose d108 file.

3. Change the MAC Address by editing these 2 lines

MAC OUI: 'XX-XX-XX' (First 6 digits of your MAC address)
MAC Begin: 'XX-XX-XX' (Last 6 digits of your MAC address)

Find your MAC address printed on your card's sticker or on the box. Dashes are required.

4. MAC End should be

MAC End: '00-FF-FF'
Count : 0x0

5. Uninstall your Aquantia ethernet device and tick "Delete the driver software for this device"

6. Install Diag driver for Aquantia card found in the firmware folder.

7. Open Command Prompt in Administrator

C:\Users\Your Username\Desktop\aquantia_firmware\AQUANTIA\AMD64

(Adjust your path to diag64 folder and replace Your Username with Windows 10 account name and Press Enter)

Then type:

diag --password !h:ahT8uW6 --aqc C:\Users\Your Username\Desktop\aquantia_firmware\AQC107-Config.aqc

(Again adjust your path as per required and replace Your Username with Windows 10 account name and Press Enter)

If your paths are correct, you should see a success message. Reboot windows for firmware change to take effect. Your card should be stock and start accepting Official Aquantia drivers. It will not work in macOS anymore. (Even with the kext patches)

If your card has AQC107S chipset, it will work in macOS. It is really necessary to find a card which has AQC107S chipset for out of the box compatibility with macOS.

So far, cards by Sonnet and Caldigit are the right ones to buy.

AQC_Config.zip

Edited September 12, 2018 by d5aqoep

jb007 · September 25, 2018

On 5/28/2018 at 8:49 AM, Mieze said:

Well, it has always been clear that this patch will most likely fail after the next update of the driver. Try to use the "built-in" patch which should still work as it doesn't rely on any specific code pattern.

I've just updated to Mojave 10.14 (18A391) and my ASUS XG-C100C stopped working.

I was using the Find: 0F84C003 0000 Replace: 90909090 9090 patch, but It's now working fine with this patch.

Thanks for all your hard work @Mieze, much appreciated. :thumbsup_anim:

d5aqoep · January 3, 2019

BTW there is a card Syba 10 gbe based on AQC-107s chipset and it is native in macOS Mojave 10.14. Someone has commented on it on another forum.

My card is arriving today so will confirm.

d5aqoep · January 3, 2019

2 hours ago, meaganmargaret said:

Basically, I expect it will work, but what I'm particularly interested in is longevity. Because my Sonnet worked for about a week or so and then it didn't. Based on reviews on Amazon, others experienced the same thing - maybe a bad batch? But who knows.....

This card is supposed to support WOL, too.....so let us know how it goes.

Got the card. Installed it and works out of the box in Mojave. Wake on Lan also works in Windows. I have finally found a true 10gbe solution which works flawlessly in macOS, Windows and Linux without hackery of kext patches and dodgy firmware updates.

dolgarrenan · January 11, 2019

Has any of you tried or know a way to activate a AQC111? I just moved to an ASUS X299D 2 and it comes bundled with the said model 5Gbs ethernet, I already have a ACPI table for it, but I think the Mieze patch dosn't work with this model.

I'm not even near sure how to make patches so I'm a bit lost right now

Mieze · January 11, 2019

6 hours ago, dolgarrenan said:

Has any of you tried or know a way to activate a AQC111? I just moved to an ASUS X299D 2 and it comes bundled with the said model 5Gbs ethernet, I already have a ACPI table for it, but I think the Mieze patch dosn't work with this model.

In order to use the device with the Apple driver, it's firmware must be updated with macOS 10.13.3 but as the AQC111's device id 0x11b1 isn't supported natively by Apple's driver, I doubt that it will update the firmware and even in case it does, there is no guarantee that Apple's firmware will work on this chip, it might as well render the chip unusable.

Mieze

afrojewelz · January 14, 2019

it is seems aqc-108 on my asrock z270 supercarrier did not work after inject kext patch in clover config.plist,even i open info.plist in AppleEthernetAquantiaAqtion.kext and found out the aqc 108 did in this list.

<key>IONameMatch</key>

<array>

<string>pci1d6a,80b1</string>

</array>

       <key>ForceKextsToLoad</key>
       <array>
           <string>/System/Library/Extensions/tn40xx.kext</string>
           <string>/System/Library/Extensions/IONetworkingFamily.kext/Contents/PlugIns/AppleEthernetAquantiaAqtion.kext</string>

<key>KextsToPatch</key>

</array>

           <dict>
               <key>Disabled</key>
               <false/>
               <key>Find</key>
               <data>
               YnVpbHQtaW4=
               </data>
               <key>InfoPlistPatch</key>
               <false/>
               <key>Name</key>
               <string>AppleEthernetAquantiaAqtion</string>
               <key>Replace</key>
               <data>
               YmJiYmJiYmI=
               </data>
           </dict>

i have no clue but asking for help ,if any other necessory info anyone need,plz ask me ,i will provide them as i could to help.

系统版本： macOS 10.13.6 (17G4015)

config.plist

Edited January 14, 2019 by afrojewelz

d5aqoep · January 28, 2019

On 1/3/2019 at 8:50 PM, meaganmargaret said:

Basically, I expect it will work, but what I'm particularly interested in is longevity. Because my Sonnet worked for about a week or so and then it didn't. Based on reviews on Amazon, others experienced the same thing - maybe a bad batch? But who knows.....

This card is supposed to support WOL, too.....so let us know how it goes.

This Syba card has stopped doing WOL suddenly from yesterday. I installed my previous card it is waking up my PC correctly. So this Syba card is borked in just 20 days. Tried everything from reinstalling windows to resetting my BIOS settings to resetting my ASUS router. It is also taking fairly long to establish a link.

So there it is.... another Chinese trash with non-existent Q&A. If I cant wake up my PC remotely, this card is junk for me.

Mieze · January 28, 2019

1 hour ago, d5aqoep said:

This Syba card has stopped doing WOL suddenly from yesterday. I installed my previous card it is waking up my PC correctly. So this Syba card is borked in just 20 days. Tried everything from reinstalling windows to resetting my BIOS settings to resetting my ASUS router. It is also taking fairly long to establish a link.

So there it is.... another Chinese trash with non-existent Q&A. If I cant wake up my PC remotely, this card is junk for me.

I see no reason why anybody would need WoL as Apple drivers support mDNS offload so that the machine stays visible while sleeping and can be woken up with a connection request using protocols like SSH for example?

Mieze

d5aqoep · January 28, 2019

20 hours ago, Mieze said:

I see no reason why anybody would need WoL as Apple drivers support mDNS offload so that the machine stays visible while sleeping and can be woken up with a connection request using protocols like SSH for example?

Mieze

I use Windows to primarily wake my PC.

An update: I kept they Syba card out of my PC for 2 days and today I plugged it back in. Now my PC wakes up correctly. Must be some capacitor problem stored with too much charge.

But still it is a reliability issue. I will keep an eye on it if it happens again or so.

Edited January 29, 2019 by d5aqoep

Thireus · February 12, 2019

Thank you for the little guide d5aqoep, I managed to flash the firmware back. However, I couldn't manage to set the MAC address. So now I have a card without MAC, which still works fine, but this is far from ideal.

This was my aqc file:

FW_FILE: 'C:\diag\asr_d107_1.5.58.clx'
PCI Device ID: 0xd107
PCI Subsystem ID: 0x8757
PCI Subsystem Vendor ID: 0x1043
PCI Max Lane Width: 4
MAC OUI: 'AA-AA-AA'
MAC Begin: 'BB-BB-BB'
MAC End: '00-FF-FF'
Count : 0x0

[...]

This is the message I would get when attempting to set a MAC address:

C:\Users\Thireus\Desktop\aqua\AQUANTIA\AMD64>diag --password !h:ahT8uW6 --aqc C:\Users\Thireus\Desktop\aqua\AQC107-Config.aqc

FLASH type = Winbond W25Q32
Erase duration 6657205 us

WARNING! Max number of MAC Addresses Programmed!

Override result = FAIL



Flash override with aqc file failed!

Exiting program...

So, I ended up removing all the MAC and Count lines from the AQC107-Config.aqc file. Which led to an empty 00-00-00-00-00-00 MAC.

I have also tried to use the "--flash_fields" parameter to set "mac_addr", but it doesn't appear to be taken into account:

C:\diag\aqua\AQUANTIA\AMD64>diag -s --password !h:ahT8uW6 --aqc C:\diag\AQC107-Config.aqc --flash_fields mac_addr=AA-AA-AA-BB-BB-BB

FLASH type = Winbond W25Q32
Erase duration 6319922 us
Starting Read Whole Flash . Pass
Starting Flash Update . Pass
Starting Flash Update . . Pass
Starting Flash Update . .  Bytes: 0x20000
  Bytes: 0x40000
 . . Pass
Starting Flash Update . . Pass

At the end of the process and after installing the latest drivers, the Ethernet controller is correctly recognised, but the MAC remains empty 00-00-00-00-00-00.

I have tried to change the Count value and MAC End value. But I don't know what they are for. Are you sure MAC Begin and MAC End don't correspond to AA-AA-AA-BB-BB-BB, where MAC Begin is A and MAC End is B? In your example you state that MAC OUI is A and MAC Begin is B. But what is MAC End and Count?

MAC OUI: 'AA-AA-AA'
MAC Begin: 'BB-BB-BB'
MAC End: '00-FF-FF' <-- What is this?
Count : 0x0 <-- What is this?

The diag utility also provides a FLASH menu where we can select an image to be flashed. I have tried and it works well. But I am unable to set the MAC and PCI info settings I want the flash process to use. Do you know where those settings are stored or how I can set them?

Also, is there any chance someone can dump of the clx firmware that the MacOS upgrade patches? So we can easily flash the ethernet card with Apple firmware. The diag utility allows to do this (FLASH MENU --> SAVE FIRMWARE, something like that).

Thank you for your help!

d5aqoep · February 13, 2019

MAC OUI: 'AA-AA-AA'
MAC Begin: 'BB-BB-BB'
MAC End: '00-FF-FF' <-- What is this?
Count : 0x0 <-- What is this?

MAC OUI = First 6 characters of your MAC address.

MAC Begin = Last 6 characters of your MAC address

MAC End and Count leave at those default values.

MAC address should be printed on the sticker at the back of your card.

Edited February 13, 2019 by d5aqoep

Thireus · February 13, 2019

4 hours ago, d5aqoep said:
MAC OUI: 'AA-AA-AA'
MAC Begin: 'BB-BB-BB'
MAC End: '00-FF-FF' <-- What is this?
Count : 0x0 <-- What is this?
MAC OUI = First 6 characters of your MAC address.

MAC Begin = Last 6 characters of your MAC address

MAC End and Count leave at those default values.

MAC address should be printed on the sticker at the back of your card.

This is what I did. But still getting the following message:

WARNING! Max number of MAC Addresses Programmed!

Override result = FAIL

Any idea?

d5aqoep · February 14, 2019

6 hours ago, Thireus said:
This is what I did. But still getting the following message:
WARNING! Max number of MAC Addresses Programmed!

Override result = FAIL
Any idea?

Try this

MAC OUI = Last 6 characters of your MAC address.

MAC Begin = Firstt 6 characters of your MAC address

MAC End and Count leave at those default values.

richtig · March 2, 2019

A Station-Drivers forum user posted a method for restoring operability to a bricked AQC111c controller:

https://www.station-drivers.com/index.php?option=com_kunena&view=topic&defaultmenu=860&Itemid=858&catid=19&id=156&lang=en&limitstart=36#1257

d5aqoep · March 5, 2019

@Thireus

Did you manage to fix your MAC address? If not I may have a fix for you.

Quote

Explanation of fields:
1) FW_FILE: Path to clx file that will be flashed.
2) PCI Device ID/PCI Subsystem ID/PCI Subsystem Vendor ID: The configuration space ids. Should be a hex number with no more
than 4 digits (eg, 0xd108).
3) PCI Max Lane Width: The maximum number of lanes that our device will report to the system. Usually 1, 2 or 4.
4) MAC OUI/MAC Begin/MAC End/Count: Fields to program the MAC Address. The OUI is self explanatory, and covers the 1st 3 bytes
of the MAC Address. For the last 3 bytes, "MAC Begin" tells the diagnostic which value to use as the starting value and "Count"
provides the offset relative to "MAC Begin" (so to get the last 3 bytes, "Count" is added to "MAC Begin"). "MAC End" is the
maximum value for the last 3 bytes: if "MAC Begin" + "Count" is equal to "MAC End", the diag will raise an error. To facilitate easy
burning of multiple chips, the "Count" is automatically incremented when one chip's flash is burnt. An example is
probably better:

   a. The user opens "AQC108-Config.aqc" and sees that "MAC OUI" is "00-17-b6", "MAC Begin" is "00-00-00", "MAC End" is "ff-ff-ff"
       and "Count" is 0x5. They run the diag with "--aqc ./temp/AQC108-Config.aqc --password [password here]". The chip's flash override
       section will be burned with a MAC Address of "00-17-b6-00-00-05" (MAC OUI for 1st 3 bytes, and MAC Begin + Count for last 3).
   b. Afterwards, the user opens "AQC108-Config.aqc" and everything remains the same, EXCEPT "Count" which has been changed to
       0x6 (incremented automatically by the diag). If the user swaps in a new chip, and runs the --aqc option again, the new chip's flash
       override section will be burned with "00-17-b6-00-00-06" (MAC OUI for 1st 3 bytes, and MAC Begin + Count for last 3).
   c. If the user does this for 20 chips, the "Count" will then read 0x19.
   d. For the next batch of 20 chips, if the user wants these chips to start with "00-17-b6-00-f0-00", then the user would change
       "MAC Begin" to "00-f0-00", and change Count back to 0x0 (otherwise, it will start at "00-17-b6-00-f0-19", because the Count was
       at 0x19 in step c). If the user keeps burning multiple chips with --aqc, they will have MAC Addreses of "00-17-b6-00-f0-00",
       "00-17-b6-00-f0-01", "00-17-b6-00-f0-02", etc (and Count will be incremented to 0x1, 0x2, etc automatically by the diag).

5) VPD: A nested structure which corresponds to the VPD structure detailed in the PCI-SIG specification. The "key_order" field
   should not be changed, but all other fields can be changed (most users leave the defaults).

You can read the enclosed firmware readme file enclosed and flash just the MAC address.

Readme-Aquantia-Firmware.txt

Edited March 5, 2019 by d5aqoep

dolgarrenan · March 6, 2019

On 1/11/2019 at 7:07 PM, Mieze said:

In order to use the device with the Apple driver, it's firmware must be updated with macOS 10.13.3 but as the AQC111's device id 0x11b1 isn't supported natively by Apple's driver, I doubt that it will update the firmware and even in case it does, there is no guarantee that Apple's firmware will work on this chip, it might as well render the chip unusable.

Mieze

How about changing device id in linux, like with the intel X550..? could that be done? A dump prior to changing device id could be made just in case..

d5aqoep · March 7, 2019

I have had 100% success in restoring the Apple's wrongly flashed firmware on AQC107, AQC108, AQN107 and AQN108 cards from Aquantia. You can flash back the correct MAC address printed on your card's sticker. The process has been outlined on

Link

here

Marvell (Aquantia) 10 Gb Ethernet support thread

Recommended Posts

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites