I purchased my hw modules (2.4, 5, 6 GHz) from compex. They were not recognized on my std Ubuntu 24.04 installation. According to compex the hw bit needs to either be hard coded into the module or a software patch applied. Because I am using all three modules in one machine (Dell R730xd) and using PCIE to M.2 riser cables they had to hardcode the identification bit and I had to send the modules back. I would ask your hardware manufacturer. I was told that the encoding will vary depending on the use of the module. You are right in that Wifi 7 has proved to be exceedingly difficult compared to the wifi 5 router I built. I have almost given up myself but continue to fight the good fight. Cheers Prashant gupta On September 20, 2024 6:17:01 PM Dieter Knueppel <dknueppel@xxxxxxxxx> wrote: I'm wondering, is this the correct mailing list for my questions? I know you are all working on best effort basis, but at least my questions related to supported HW shouldn't be too complicated to answer. The information I found so far are almost not existent, stone old, misleading or even proofed to be wrong (e.g. https://wikidevi.wi-cat.ru/Ath12k, 17cb:1107 equals HERALD-BE as well, not sure who maintains this page). I'm almost at the point moving back to my legacy HW. Anyway, I would still be very happy to receive any feedback :-) Thanks a lot, Dieter P.S.: Due to my background in mobile networks radio interface, i would have a good starting point to contribute myself. Unfortunately I'm involved in plenty of other things and don't have the time for another job :-(. Am 19.09.2024 um 11:41 schrieb Dieter Knueppel: Minor update: I just figured out, that even the physical machine with std. Ubuntu 24.04 server (before the VM gets started and PCI device vanishes on PM) reports: "Unknown hardware version found for WCN7850: 0xf" Hence the earlier question, whether "MSI HERALD-BE" or "Gigabyte GC-WIFI7" is supported, seems still to be valid. Am 18.09.2024 um 19:54 schrieb Dieter Knueppel: Hi Jeff, thanks a lot for your update on ath11/ath12 related vfio support. That's actually bad news! Do you know whether vfio support vanished per accident or intention? I.e. I'm wondering on why the ath11 patch haven't made it into mainline? Assume there are no other 802.11be Chipsets supported by Linux, which can be used as AP within a VM? Kind regards, Dieter Am 18.09.2024 um 18:13 schrieb Jeff Johnson: Resend since I had a typo in the ath12k e-mail list On 9/18/2024 9:05 AM, Jeff Johnson wrote: On 9/18/2024 7:40 AM, Dieter Knueppel wrote: Dear developer team, I have to admit, the earlier post "HTT timeout error with NCM86" is my fault, sorry for causing confusion. Due to earlier tests with legacy cards, the PCIe slot was still forced to Gen2 :-( Putting it back to <auto>, with "MSI HERALD-BE", similar to "Gigabyte WC-WIFI7" i consistently get: "Unknown hardware version found for WCN7850: 0xf" Which nails down to the question whether any of these adapter is already supported? The M.2 NCM865 card as such seems to be supported. I'm wondering about the difference, as there seems to be no additional logic compiled on one of the two PCIe boards, just bit of glue logic. For better support on ath12k driver issues please include the ath12k driver list (I've added it to this reply). In your original e-mail you said: I'm trying to get "MSI HERALD-BE NCM865" up and running within a kvm The ath12k driver does not support running in a VM. Even prior ath11k-based hardware doesn't support it. For ath11k there are some manual steps that have made it work, so if you are adventurous you can try doing something similar with ath12k. Refer to: https://lore.kernel.org/all/adcb785e-4dc7-4c4a-b341-d53b72e13467@xxxxxxxxx/ /jeff
