InsaneKnight

As commented in https://github.com/openwrt/openwrt/issues/8878#issuecomment-1762463046 , if only the ubiblock for the new root squashfs does not fall into the gap, flashing sysupgrade image mostly works for this change.

> I do not think we will be able to get this integrated into the > upstream Linux kernel. > ath10k and ath9k should run into a kernel panic when...

> When you add a compatibility string it should only load for one card > when when it is possible to add multiple PCIe cards into the slot: > See:...

@mpratt14 You seem not get my point. tp-link archer c7 v2 has the following DT definition: > &pcie1 { >         status = "okay"; > >         ***@***.***,0 { >                 compatible =...

Besides, the old approach loading cal data from file system does not have this issue, because ath9k driver will not load a blob named for ath10k.

> After it is replaced with a stand-along ath9k card (e.g. ar9580) > > When you replace the pcie card, update the DTS to reflect the new > card? Otherwise,...

> ok I understand now > > in my opinion the proper fix would be to edit ath9k and ath10k drivers > to look for their own compatible property in...

> it would not hurt to just add another generic compatible > "atheros,ath9k" or similar. even if upstream wouldn't like that, its > small enough to keep in the "hack"...

Personally I am okay to generate image via imagebuilder with tuned device tree for my archerc7v2 with replaced wlan card, as I have to use my wdr4900v2 in this way...

I have pushed another revision, but it seems the back ported mac80211 drivers cannot properly use newly exported symbols from the Linux kernel proper.