android/SukiSU-Ultra
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity
2,938 Commits 11 Branches 15 Tags
c4d8c49e5c88532407d80e0be99ca282b7222af3
Commit Graph

710 Commits

Author SHA1 Message Date
ShirkNeko
c4d8c49e5c kernel/manager/ksud: Add switch functionality to sulog 
Signed-off-by: ShirkNeko <109797057+ShirkNeko@users.noreply.github.com>
 2025-11-16 16:53:55 +08:00
ShirkNeko
8442ebcb7a kernel: Optimise multi manager scanning logic 2025-11-16 15:35:26 +08:00
Wang Han
25fbc22f66 kernel: Replace kmalloc() usages with kzalloc() (#2939) 
This ensures we won't use uninitialized pointers for task work.
 2025-11-16 15:24:49 +08:00
ShirkNeko
a4a9df3a25 kernel: fmt 2025-11-16 14:29:10 +08:00
5ec1cff
db6a59ec4e kernel: file_wrapper: copy mode of original inode 
Bionic uses fstat to determine whether an fd is a tty and set proper
buffering flags, so we also need to set the wrapper file's inode mode to
the original inode mode.

see:
https://cs.android.com/android/platform/superproject/main/+/main:bionic/libc/upstream-openbsd/lib/libc/stdio/makebuf.c;l=61-95;drc=9a4b68e20d617b2cb3355071521f16e8c3d538df
 2025-11-15 19:54:16 +08:00
ShirkNeko
46846dce91 kernel: Utilise the existing try_umount to avoid repackaging 2025-11-15 19:29:12 +08:00
5ec1cff
04ca981e4d kernel: no need to remark process on post-fs-data and boot-completed 
- Remark on post-fs-data may unmark zygote unexpectedly, and there is no
necessity to remark on these stages, so simply remove them.
 2025-11-15 17:27:48 +08:00
Wang Han
01ac06c3fd kernel: Prune allowlist only after boot completed 
For unknown reason, packages.list is not reliable during boot for oplus
devices, so we have to disable pruning and re-run pruning after boot.
 2025-11-15 10:54:19 +08:00
ShirkNeko
7e7713ee4a kernel/sulog: Optimise the sulog log format 2025-11-14 12:55:57 +08:00
ShirkNeko
7afcdb3059 kernel: Optimise sulog to prevent deadlocks caused by global lock contexts. 2025-11-14 12:29:55 +08:00
ShirkNeko
5df821ed41 kernel: Resolved potential deadlock issues arising from operations not being performed within locks. 2025-11-14 11:55:06 +08:00
5ec1cff
9f869090d2 kernel: refine syscall_hook_manager 
- Don't unmark process when setuid if syscall tracepoint is in use
- Remark process when app profile updated
- Ensure zygote is marked on first boot
 2025-11-13 20:09:32 +08:00
ShirkNeko
bb8b991110 kernel: restart enable sulog for LKM 2025-11-13 17:46:08 +08:00
ShirkNeko
10548f9243 kernel: fix build 2025-11-13 17:06:27 +08:00
weishu
96d33e62bb kernel: fix zygote mark on first boot (#2924) 2025-11-13 16:31:26 +08:00
ShirkNeko
7be8c15b85 kernel: Remove redundant ksu_handle_inode_permission hooks, calling only for manual_su 2025-11-13 15:02:20 +08:00
5ec1cff
095385f814 kernel: fix wrong show_fdinfo impl 
Signed-off-by: Wang Han <416810799@qq.com>
 2025-11-12 13:25:20 +08:00
ShirkNeko
7e595e1730 kernel: Public ksu_handle_sys_reboot 
- clean code
 2025-11-11 16:49:09 +08:00
Wang Han
ded31c2043 kernel: Fix task flag marking for root and shell UID 
Signed-off-by: Wang Han <416810799@qq.com>
Co-authored-by: 5ec1cff <56485584+5ec1cff@users.noreply.github.com>
 2025-11-11 16:26:37 +08:00
ShirkNeko
8f49898155 kernel: Use task work to install fd 
There are many fd related functions that can sleep, so we have no choice
but move operations to task work. Also close fd when copy_to_user fails.

Co-authored-by: Wang Han <416810799@qq.com>
 2025-11-11 01:10:36 +08:00
ShirkNeko
fd3a22360a kernel: Clean code build 2025-11-09 23:18:48 +08:00
Wang Han
ef36a36e9a kernel: Don't save allowlist on module exit 
This is not needed and may trigger UAF as work is async.
 2025-11-09 23:14:13 +08:00
ShirkNeko
cda7e4c6c0 Clean up kernel code (#2898) 
1) Fix memory leak of callback head in allowlist.c
2) Remove duplicated logic and incorrect log in kernel_umount.c
3) Prevent sleep in kprobe context in ksud.c
4) Remove useless is_unsupported_uid, use euid for security enhance,
   add FIXME in setuid_hook.c
5) Remove useless fd argument for execve hook, fix incorrent pointer
   usage in syscall_hook_manager.c and sucompat.c
6) Use correct errno in supercalls.c

---------

Co-authored-by: Ylarod <me@ylarod.cn>
 2025-11-09 19:20:30 +08:00
ShirkNeko
0b63cc445c kernel: Expose the ksu_handle_sys_read hook 2025-11-09 18:11:30 +08:00
ShirkNeko
2433ced81a kernel: Rollback disable_seccomp 
Co-authored-by: ShirkNeko <109797057+ShirkNeko@users.noreply.github.com>
Co-authored-by: Faris <rissu.ntk@gmail.com>
 2025-11-09 17:00:43 +08:00
ShirkNeko
1f04f13e44 kernel: Remove redundant checks 2025-11-09 16:41:54 +08:00
ShirkNeko
184467c691 kernel: Undo some changes 2025-11-09 16:09:59 +08:00
ShirkNeko
05ed1a3714 fix build 2025-11-09 14:17:27 +08:00
ShirkNeko
163531fcd2 kernel: Simplify state management logic 2025-11-09 14:07:40 +08:00
ShirkNeko
049956aaa9 Fixes only hook in LKM mode __NR_newfstatat ,__NR_faccessat ,__NR_execve system calls 2025-11-09 12:16:23 +08:00
ShirkNeko
6530d06710 kernel: clean build 2025-11-09 04:12:11 +08:00
ShirkNeko
88135d8363 Attempt to refactor and migrate inode_permission, bprm_check_security, and task_alloc entirely to syscall_hook_manager 2025-11-09 02:52:46 +08:00
ShirkNeko
548258f922 kernel: Rewrite the kernel source code (#554) 
* clean unused header

* on_module_mounted in ksud.c

* refact: use app_profile

* unified hook manager

* add zygote to hook target

* move reboot hook to supercall.c

* refactor: kernel_umount setuid_hook

* update mark rules, add init mark tracker

* remove reboot from check_syscall_fastpath

* update setuid_hook, remove uneeded sucompat enable

* log freely

* kernel: Migrate kprobe hook configuration items

* kernel: fix build

* cli: add ksud debug mark

* Fix rustfmt warning

---------

Co-authored-by: Ylarod <me@ylarod.cn>
Co-authored-by: Wang Han <416810799@qq.com>
Co-authored-by: ShirkNeko <109797057+ShirkNeko@users.noreply.github.com>
 2025-11-09 01:14:26 +08:00
weishu
46b9f5fb4b kernel: fix put_task if alloc failed. 2025-11-08 21:44:04 +08:00
ShirkNeko
413e9ab8a9 kernel: Resolved compatibility issues with su when using manual hooks 2025-11-08 20:52:29 +08:00
ShirkNeko
7479c0b81b kernel: Fix build 2025-11-08 19:54:20 +08:00
ShirkNeko
0381d12be2 kernel: Resolve implicit declaration conflicts 2025-11-08 19:47:03 +08:00
Wang Han
4425c88d5a Fix missing unlock on error path 2025-11-08 19:41:11 +08:00
weishu
7828c5c107 kernel: fix save allowlist 2025-11-08 19:40:16 +08:00
ShirkNeko
76046c84cd kernel: remove unused kernel_compat 
Co-authored-by: weishu <twsxtd@gmail.com>
 2025-11-08 19:39:41 +08:00
weishu
623dd15cbf kernel: Use real_parent to avoid interference from ptrace. 2025-11-08 19:30:55 +08:00
ShirkNeko
ab13ed5c16 kernel: remove unused wrapper for 
Co-authored-by: weishu <twsxtd@gmail.com>
 2025-11-08 19:30:44 +08:00
weishu
e171ca15cb kernel: remove ksu_compat_{open,read,write} because we're in the right context now 
Co-authored-by: weishu <twsxtd@gmail.com>
 2025-11-08 19:28:25 +08:00
ShirkNeko
4fc369a059 kernel: remove workqueue for allowlist 
Co-authored-by: weishu <twsxtd@gmail.com>
 2025-11-08 19:18:20 +08:00
ShirkNeko
18ad2afadb Reworking fdwrapper 
Co-authored-by: 5ec1cff <ewtqyqyewtqyqy@gmail.com>
Co-authored-by: 5ec1cff <56485584+5ec1cff@users.noreply.github.com>
Co-authored-by: Ylarod <me@ylarod.cn>
 2025-11-08 19:03:14 +08:00
technotic
1b5b235bd9 add include for crc32.h in sulog.h (for dedup_calc_hash) (#548) 
Co-authored-by: technotic <pixie@technotic.us>
Co-authored-by: ShirkNeko <109797057+ShirkNeko@users.noreply.github.com>
 2025-11-08 18:12:38 +08:00
AlexLiuDev233
fdf5e7104e refactor: kpm: memory management migrate to sukisu side (#539) 
* refactor: kpm: memory management migrate to sukisu side

* fix: build warning in some gki2 device

fix stack frame size warning (maybe) in gki2 device, specialy in ShirkNeko's device

* chore: use pr_info instead of printk

* feat: check the validity of pointers sent from user space
Sometimes, ksud or other root processes might request a kpm ioctl,
but data incorrectly, such as invalid pointer,
which cause the kernel to crash.

If the request is made by ksud at boot time, the situation is even worse,
as it can cause the system to enter the boot loop.

Therefore, I believe we need to check pointer integrity in kernel space to fix this problem.

---------

Co-authored-by: Saksham <typeflu@gmail.com>
Co-authored-by: ShirkNeko <109797057+ShirkNeko@users.noreply.github.com>
 2025-11-08 18:11:29 +08:00
AlexLiuDev233
704f7cba32 kernel: core_hook: disable seccomp in 5.10.2- for allowed uids (#545) 
* kernel: core_hook: disable seccomp in 5.10.2- for allowed uids

we dont have those new fancy things upstream has
lets just do original thing where we disable seccomp

* Update kernel/core_hook.c

* fmt

---------

Co-authored-by: backslashxx <118538522+backslashxx@users.noreply.github.com>
Co-authored-by: Saksham <saksham.mac@icloud.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
 2025-11-08 12:46:25 +08:00
ShirkNeko
d37a78ea2d fix build & cleanup 2025-11-08 11:48:35 +08:00
Ylarod
6c9bf69718 fix build 2025-11-08 11:37:07 +08:00