add vendor_boot patching for some odd devices

[ukriu]

This fixes (or adds) #8835

[ukriu]

The same method to detect this case has been used in tiann/KernelSU#2650 and KernelSU-Next/KernelSU-Next#541 (comment)
And has been working perfectly fine

[topjohnwu]

Thank you for your contribution. I'll need further thinking to make sure if there are better ways to detect this scenario.

[ukriu]

Thank you for your contribution. I'll need further thinking to make sure if there are better ways to detect this scenario.

Just a heads up, there is a motorola device which has a compatible ramdisk in vendor_boot as well as boot, the current check picks the boot partition for this device, but you still could remove one of the magiskboot unpacking steps to save time since both work for the device.

As for the speed, I think its fine since it only takes less than a second from my testing. But it still is a bit messy so I'm looking forward to new ideas if you have them.

Currently the devices which need this are the Pixel 6 family, some Pixel 7 devices (not sure), and about 10 to 15 Xiaomi Mediatek devices which have officially gotten an upgrade to Android 15; so it would be nice if this case is acknowledged

[vvb2060]

No pixel device has this behavior, I think you might have made a mistake

[ukriu]

No pixel device has this behavior, I think you might have made a mistake

Pixel 6/6 Pro/6a, while I only have confirmed it with 6a.
You can grab the latest "bluejay" firmware and test for yourself

[vvb2060]

I checked this https://dl.google.com/dl/android/aosp/bluejay-ota-bp2a.250705.008.a1-14519571.zip
Amazing, it seems that this is because the device uses the new layout but does not want to repartition. Is adding a ramdisk to boot.img invalid? It will overwrite the file with the same name in the vendor_boot ramdisk.

[ukriu]

Is adding a ramdisk to boot.img invalid? It will overwrite the file with the same name in the vendor_boot ramdisk.

That can be done and is being used here (KernelSU-Next/KernelSU-Next@512f845), I have tried an early release of it on my xiaomi device and it does work but I wasn't sure how stable it was.

[ukriu]

@topjohnwu heya, have you decided on it yet..?
If this is too excessive then supporting only patching via the patch script is a viable option too (since for MTK Xiaomi devices, writing to those partitions in runtime is not permitted) and it can have support partially.
Or what vvb2060 suggested could be viable too, then I'll close this PR.

[topjohnwu]

@ukriu my plan is to either
a. have a way to detect this, or
b. add a new option in the app to allow user toggling

I just haven't have time to look into it yet