Closed snake-4 closed 5 months ago
This only seems to happen when there's a bind mount module is present. Granted I've only tested with the built-in Systemless Hosts.
An alternative is to use this module instead of Systemless Hosts (It works on Magisk): https://github.com/symbuzzer/systemless-hosts-KernelSU-module/
I use bind mounts to avoid other detector apps - you wouldn't consider this change?
I use bind mounts to avoid other detector apps - you wouldn't consider this change?
Bind mounts are already hidden since 2.0.3 but this specific trace is not hidden. Though I guess most apps don't check this trace because it's kind of unreliable and may also happen on a stock device.
Implemented in f200aa4.
Compared to stock kernel, Magisk adds
errors=remount-ro
to alluserdata
mounts. Should we remount them without the flag? It may be an unreliable detection vector but it's still an abnormality.Magisk patched kernel:
/dev/block/bootdevice/by-name/userdata on /data type ext4 (rw,seclabel,nosuid,nodev,noatime,nobarrier,noauto_da_alloc,errors=remount-ro,data=ordered)
Stock kernel:
/dev/block/bootdevice/by-name/userdata on /data type ext4 (rw,seclabel,nosuid,nodev,noatime,nobarrier,noauto_da_alloc,data=ordered)