[RFC] kernel: restore non-kprobe support, sucompat toggle for non-kp. #2475
Conversation
…kprobe by default" This reverts commit 500ff9b.
Restoring this will allow a user to build with manual hooks even with kprobes enabled. If a user uses manual hooks, he has to CONFIG_KSU_WITH_KPROBES=n on his kernel config. This is another take of 500ff9b. Signed-off-by: backslashxx <118538522+backslashxx@users.noreply.github.com>
ext4_unregister_sysfs() on de29115
this is not really needed, if a user on 6.1 and further wants to hook vfs_statx all he has to do is point to the proper member. `ksu_handle_stat(&dfd, &filename->name, &flags);` another is to hook vfs_fstatat or newfstatat instead
I have no idea if this is needed or any useful for manual hooks users. While this is not exactly the same thing, this *CAN* achieve the same results. The complete disabling of all KernelSU hooks. While we can use a single variable for this, this is done like how vfs_read_hook, input_hook and execve_hook are disabled, this will allow proper feature extension, such as devpts hook toggling etc. While this is probably not so useful for us, lets still port it for the sake of feature parity. Signed-off-by: backslashxx <118538522+backslashxx@users.noreply.github.com>
|
+1, manual hooks are useful for both GKI and non-GKI users. Non-GKI users benefit from compatibility, while GKI users get an alternative. Supporting both keeps the flexibility, even if it turns into an #ifdef maze, just having the option is huge. I've been using manual hooks, even tho I am a gki user. ChiseWaguri/android_kernel_xiaomi_marble@3322eec |
+1 |
|
Just reverting is fine; we don't want any extra complexity. |
this will revert 500ff9b
"kernel: remove unused CONFIG guard becuase GKI kernel enable kprobe by default"
This pr can allow gki kernels to be built with manual hooks. It also comes with a port of
sucompat toggle for the sake of feature parity.
If this gets accepted, I also plan to get manual hooks guide on par with kprobes hook.
backslashxx#5
ofcourse, this will depend now on KernelSU policy, as this will re-introduce an ifdef hell.
This will make maintenance harder, you now have to keep track a lot of different states, etc etc,
but even if this does NOT get accepted, this can still be a demo on how you can have manual
hooks on gki kernels.