The wlan and bluetooth driver bcmdhd4398 cannot be loaded because cfg80211.ko has been converted to a vendor module. You can find more information about this conversion at: ee1e4b7c5e.
The reason why cfg80211.ko cannot be loaded is because the rfkill.ko module in system_dlkm has protected symbols. To fix this issue, we simply need to delete the protected symbol.
"There is no ABI stability between GKI and GKI modules. In order for modules to load correctly during runtime, GKI and GKI modules must be built and updated together."
We need to find a better solution to address this problem.
Co-authored-by: aviraxp <wanghan1995315@gmail.com>
- Restore kernels
- 5.15.**94.2**: WSA: **2305**.40000.(4~6).0
- 5.15.**104.1**: WSA: **2306**.40000.4.0
- 5.15.**104.2**: WSA: **2307**.40000.6.0
This is because it is still being distributed by MS official.
- Optimize workflow
Unless there is a destructive update, all settings are the same, so
there are many unnecessary descriptions.
Therefore, the next time the device code is changed or the kernel
version is updated to higher than 5.15, the current configuration will
be restored again.
- WSA latest version: `2307.40000.6.0`
-> Kernel version: `5.15.104.2`
- Next kernel version: `5.15.104.3`
-> We haven't yet confirmed the release of this kernel.
- Bluejay running GrapheneOS is on Kernel version `5.10.187`.
- Tried to flash other versions `5.10.xxx` and all went into bootloop.
Unsure how this version would go
For android13 gki kernels, Google puts lots of unused file to kernel
source tree, while the Github Action only has 14G disk size and the ci
may fail because of "No space left on device".
The `repo` tool will try pull all the refs and tags with depth=INT_MAX if it found the SHA1 hash doesn't exist on remote server even if you force it to fetch with `depth=1` and `--no-tags`. So we use the version `v2.16` which doesn't check it.
And also, using a fixed repo version can avoid randomly failure of building.
Compared with markdown, yaml may be more focused on the required
information, and will not easily destroy the content in the template.
And I think the "feature request" might not require users to upload
their logs.
It looks like this:
Co-authored-by: dabao1955 <195328750@qq.com>
