Kasumi and mount configuration
About 1075 wordsAbout 4 min
YukiSU v1.8.0 integrates Kasumi API 18 and the userspace Kagami controller. The Kasumi button at the top right of Home opens Status, Settings, Rules, and Logs tabs. Module cards also provide per-module mount configuration.
Three separate controls
| Control | Location | Effect and lifecycle |
|---|---|---|
| Kasumi Feature 106 | Settings → Feature controls | Controls the built-in Kasumi enable request; disabling does not immediately tear down existing views, so fully reboot as prompted |
| Built-in mounting | Settings | Selects the built-in Kagami mounting engine; takes effect after reboot |
| Kasumi views and individual options | Kasumi → Settings | Configures kernel views, hiding, and related capabilities; supported runtime settings are applied after saving, while mount-backend settings wait for the next boot |
Built-in mounting does not require installing a separate MetaModule. If an external MetaModule owns mounting, the page displays its ID and disables inapplicable built-in mount controls. Do not have two engines process the same module tree. See MetaModule for mount selection.
Refresh status after saving configuration. “Saved, but kernel apply failed” means configuration is persisted but runtime application is incomplete. Preserve the error, resolve the cause, then retry applying and refresh status.
Status
Check kernel availability, protocol compatibility, and version first. Status also shows storage/mount information, actual module strategies, and active views. After changing a module mode, fully reboot and read its actual strategy on Status.
Installing the v1.8.0 APK does not replace an old LKM or running controller. If interfaces are incompatible, sync the matching ksud, install a compatible LKM, and fully reboot.
API 18 adds a global scope for individual rules; the manager allows it only when the kernel reports rule_bypass. Soft reboot separately checks capabilities such as safe_rule_retire. Do not use the API version alone to decide support; see Soft reboot.
Mount strategies and module rules
Module cards offer auto, kasumi, overlay, magic, and none, plus rules for absolute target paths. Path rules additionally support hide. none selects a mount strategy; it does not disable the entire module, whose scripts still follow its state and lifecycle.
The Modules menu can filter by strategy, show only active Kasumi modules, and check mount conflicts. The conflict list identifies modules competing for the same paths. Inspect the corresponding logs for script execution and mounts created by scripts.
Settings control whether OverlayFS and Magic Mount may participate in planning, the storage filesystem (automatic, ext4, erofs, or tmpfs), mount-source name, temporary directory, and mirror path. After storage or backend changes, fully reboot and inspect actual strategies again.
Kernel views and hiding rules
Available controls depend on capabilities reported by the running kernel, including mount hiding, maps mapping, OverlayFS xattr hiding, statfs views, and SELinux Guard. v1.8.0 removes aggressive mount hiding; do not continue configuring that mode from older instructions.
SELinux Guard is disabled by default. It experimentally applies bounded timing compensation to SELinux context writes. It neither changes Enforcing/Permissive nor replaces App Profile domains and SELinux rules. Enable it when investigating that class of compatibility problem and compare target-app behavior.
User hide rules require absolute paths and cannot hide the reserved /system/bin/su through this entry point. Rules distinguish pending, binding, bound, and error states. A rule can remain pending while its parent directory does not exist; if the path exists but binding does not complete, or an error appears, inspect logs and retry. Maps rules address path, device, and inode information for mappings; they do not change the real origin of the file.
“Clear rules” changes active rules; it is not a complete uninstaller or a reset of every persistent setting. Record the original configuration first and verify mount/view effects inside the target application process.
Global rules: bypassing process policy
“Global rules (bypass policy)” skips ordinary process policy for selected Kasumi rules, affecting all processes, including managers or root processes normally exempted. It does not grant root or change OverlayFS/Magic Mount rules.
- Module switch: makes every Kasumi rule of that module global.
- Path-rule switch: makes only that rule global. A per-rule off setting cannot override an enabled module switch.
- User hide rules: existing rules have editable scope. “Make the new rule global” beside the input affects only the next path saved with Add.
This is off by default. Enable it when root and ordinary apps must share the same file view. Hiding paths required by the manager or tools removes their access too; record the rule before changing it so you can undo it from the same page. Saving module configuration attempts to apply Kasumi rules immediately; mount-backend changes still require reboot.
Conflicting kernel components
Kasumi refuses activation alongside recognized conflicting components: kasumi_lkm, pathmask, procguard, nomount, susfs_guard_lkm, hybridmount, and running-kernel configurations CONFIG_KSU_SUSFS=y, CONFIG_NOMOUNT=y, or CONFIG_HYBRIDMOUNT=y. Loading these conflicting .ko modules while Kasumi is active is also rejected.
The dialog distinguishes an already loaded module blocking Kasumi, a built-in feature conflict, and a newly rejected module load. “I understand” acknowledges the warning only. It does not change settings, unload components, or remove the restriction.
Choose one implementation: disable the conflicting module's load source or disable Kasumi, then fully reboot. For a built-in conflict, keep Kasumi disabled or use a kernel without that feature. This check covers recognized components only; an absent warning does not establish compatibility with arbitrary kernel modifications.
Reboot and diagnosis
v1.8.0 permits Kasumi soft reboot when capabilities and restoration requirements are satisfied. Kasumi activation alone no longer blocks it. Follow the daemon support state shown in Manager and the specific remedies in Soft reboot. After disabling Feature 106, a full reboot is still needed to end its current kernel lifecycle. A full reboot also ends temporary late-load root.
Include the following in an issue report:
- YukiSU,
ksud, and Kasumi versions and support state; - built-in mount setting, external mount owner, and global/per-module configuration;
- requested versus actual strategies and rule states;
- original logs and file/mount observations from the reproducing process.
The Kasumi Logs tab supports refresh and clear. Preserve relevant output before clearing it. Include actual strategies and rule states in screenshots, together with file or mount results from the target process.
Source references
Kasumi UI, settings and capabilities, controller, kernel lifecycle, and soft-reboot restrictions.
kasumi.h, KasumiMountConfigDialog.kt, sucompat_module_guard.c, kasumi_config_guard.c, KasumiConflictDialog.kt.
Copyright
Copyright Ownership:Anatdx
License under:Attribution-NonCommercial-ShareAlike 4.0 International (CC-BY-NC-SA-4.0)