mirror of
https://github.com/topjohnwu/Magisk.git
synced 2024-11-30 21:45:27 +00:00
288 lines
16 KiB
Markdown
288 lines
16 KiB
Markdown
|
## Available Tools
|
||
|
Magisk comes with a lot of tools for installation, programs running as a daemon, and utilities for developers. This documentation covers 3 binaries, and many more tools are available as applets. The relation between tools are shown below:
|
||
|
|
||
|
```
|
||
|
magiskboot /* binary */
|
||
|
magiskinit /* binary */
|
||
|
magiskpolicy -> magiskinit
|
||
|
supolicy -> magiskinit /* alias of magiskpolicy */
|
||
|
magisk /* binary */
|
||
|
magiskhide -> magisk
|
||
|
resetprop -> magisk
|
||
|
su -> magisk
|
||
|
```
|
||
|
|
||
|
### magiskboot
|
||
|
A tool to unpack / repack boot images, parse and patch cpio and dtbs, hex patch binaries, compress / decompress with multiple algorithms. It is used to install Magisk into boot images.
|
||
|
|
||
|
`magiskboot` natively supports (which means it does not call external tools) all popular compression methods including `gzip` (used everywhere for compressing kernel and ramdisk), `lz4` (used to compress kernel in modern devices like Pixel), `lz4_legacy` (legacy LZ4 block format with special metadata used [only on LG](https://events.static.linuxfound.org/sites/events/files/lcjpcojp13_klee.pdf) to compress kernel), `lzma` (LZMA1 algorithm natively supported in Linux kernel, used in some custom kernel to compress ramdisk), `xz` (LZMA2 algorithm, very high compression rate, used in Magisk for high compression mode and storing binaries), and `bzip2` (used in desktop Linux boot images to create bzImage, haven't seen on Android yet).
|
||
|
|
||
|
The concept of `magiskboot` is to keep the images as intact as possible. For unpacking, it extracts the large chunks of data (kernel, ramdisk, second, dtb, extra etc.) and decompress them if possible. When repacking a boot image, the original boot image has to be provided so it can use the original headers (including MTK specific headers) with only changing the necessary entries such as the data chunk sizes, and re-compress all data with the original compression method. The same concept also applies to CPIO patching: it does not extract all files, modify in file system, archive all files back to cpio as usually done to create Linux `initramfs`, instead we do modifications directly in the cpio level in memory without involving any data extraction.
|
||
|
|
||
|
Command help message:
|
||
|
|
||
|
```
|
||
|
Usage: magiskboot <action> [args...]
|
||
|
|
||
|
Supported actions:
|
||
|
--parse <bootimg>
|
||
|
Parse <bootimg> only, do not unpack. Return values:
|
||
|
0:OK 1:error 2:insufficient boot partition size
|
||
|
3:chromeos 4:ELF32 5:ELF64
|
||
|
|
||
|
--unpack <bootimg>
|
||
|
Unpack <bootimg> to kernel, ramdisk.cpio, (second), (dtb), (extra) into
|
||
|
the current directory. Return value is the same as --parse
|
||
|
|
||
|
--repack <origbootimg> [outbootimg]
|
||
|
Repack kernel, ramdisk.cpio[.ext], second, dtb... from current directory
|
||
|
to [outbootimg], or new-boot.img if not specified.
|
||
|
It will compress ramdisk.cpio with the same method used in <origbootimg>,
|
||
|
or attempt to find ramdisk.cpio.[ext], and repack directly with the
|
||
|
compressed ramdisk file
|
||
|
|
||
|
--hexpatch <file> <hexpattern1> <hexpattern2>
|
||
|
Search <hexpattern1> in <file>, and replace with <hexpattern2>
|
||
|
|
||
|
--cpio <incpio> [commands...]
|
||
|
Do cpio commands to <incpio> (modifications are done directly)
|
||
|
Each command is a single argument, use quotes if necessary
|
||
|
Supported commands:
|
||
|
rm [-r] ENTRY
|
||
|
Remove ENTRY, specify [-r] to remove recursively
|
||
|
mkdir MODE ENTRY
|
||
|
Create directory ENTRY in permissions MODE
|
||
|
ln TARGET ENTRY
|
||
|
Create a symlink to TARGET with the name ENTRY
|
||
|
mv SOURCE DEST
|
||
|
Move SOURCE to DEST
|
||
|
add MODE ENTRY INFILE
|
||
|
Add INFILE as ENTRY in permissions MODE; replaces ENTRY if exists
|
||
|
extract [ENTRY OUT]
|
||
|
Extract ENTRY to OUT, or extract all entries to current directory
|
||
|
test
|
||
|
Test the current cpio's patch status. Return value:
|
||
|
0:stock 1:Magisk 2:other (phh, SuperSU, Xposed)
|
||
|
patch KEEPVERITY KEEPFORCEENCRYPT
|
||
|
Ramdisk patches. KEEP**** are boolean values
|
||
|
backup ORIG [SHA1]
|
||
|
Create ramdisk backups from ORIG
|
||
|
SHA1 of stock boot image is optional
|
||
|
restore
|
||
|
Restore ramdisk from ramdisk backup stored within incpio
|
||
|
magisk ORIG HIGHCOMP KEEPVERITY KEEPFORCEENCRYPT [SHA1]
|
||
|
Do Magisk patches and backups all in one step
|
||
|
Create ramdisk backups from ORIG
|
||
|
HIGHCOMP, KEEP**** are boolean values
|
||
|
SHA1 of stock boot image is optional
|
||
|
sha1
|
||
|
Print stock boot SHA1 if previously stored
|
||
|
|
||
|
--dtb-<cmd> <dtb>
|
||
|
Do dtb related cmds to <dtb> (modifications are done directly)
|
||
|
Supported commands:
|
||
|
dump
|
||
|
Dump all contents from dtb for debugging
|
||
|
test
|
||
|
Check if fstab has verity/avb flags. Return value:
|
||
|
0:no flags 1:flag exists
|
||
|
patch
|
||
|
Search for fstab and remove verity/avb
|
||
|
|
||
|
--compress[=method] <infile> [outfile]
|
||
|
Compress <infile> with [method] (default: gzip), optionally to [outfile]
|
||
|
<infile>/[outfile] can be '-' to be STDIN/STDOUT
|
||
|
Supported methods: gzip xz lzma bzip2 lz4 lz4_legacy
|
||
|
|
||
|
--decompress <infile> [outfile]
|
||
|
Detect method and decompress <infile>, optionally to [outfile]
|
||
|
<infile>/[outfile] can be '-' to be STDIN/STDOUT
|
||
|
Supported methods: gzip xz lzma bzip2 lz4 lz4_legacy
|
||
|
|
||
|
--sha1 <file>
|
||
|
Print the SHA1 checksum for <file>
|
||
|
|
||
|
--cleanup
|
||
|
Cleanup the current working directory
|
||
|
```
|
||
|
|
||
|
### magiskinit
|
||
|
This tool is created to unify Magisk support for both legacy "normal" devices and new `skip_initramfs` devices. The compiled binary will replace `init` in the ramdisk, so things could be done even before `init` is started.
|
||
|
|
||
|
`magiskinit` is responsible for constructing a proper rootfs on devices which the actual rootfs is placed in the system partition instead of ramdisk in `boot.img`, such as the Pixel familiy and most Treble enabled devices, or I like to call it `skip_initramfs` devices: it will parse kernel cmdline, mount sysfs, parse through uevent files to make the system (or vendor if available) block device node, then copy rootfs files from system. For normal "traditional" devices, it will simply swap `init` back to the original one and continue on to the next stage.
|
||
|
|
||
|
With a proper rootfs, `magiskinit` goes on and does all pre-init operations to setup a Magisk environment. It patches rootfs on the fly, providing fundamental support such as patching `init`, `init.rc`, run preliminary `sepolicy` patches, and extracts `magisk` and `init.magisk.rc` (these two files are embedded into `magiskinit`). Once all is done, it will spawn another process (`magiskinit_daemon`) to asynchronously run a full `sepolicy` patch, then starts monitoring the main Magisk daemon to make sure it is always running (a.k.a invincible mode); at the same time, it will execute the original `init` to hand the boot process back.
|
||
|
|
||
|
### magiskpolicy
|
||
|
(This tool is aliased to `supolicy` for compatibility with SuperSU's sepolicy tool)
|
||
|
|
||
|
This tool is an applet of `magiskinit`: once `magiskinit` had finished its mission in the pre-init stage, it will preserve an entry point for `magiskpolicy`. This tool could be used for advanced developers messing with `sepolicy`, a compiled binary containing SELinux rules. Normally Linux server admins directly modifies the SELinux policy sources (`*.te`) and recompile the `sepolicy` binary, but here we directly patch the binary file since we don't have access to the sources.
|
||
|
|
||
|
All processes spawned from the Magisk daemon, including root shells and all its forks, are running in the context `u:r:su:s0`. Magisk splits the built in patches into 2 parts: preliminary and full
|
||
|
|
||
|
- The preliminary patch should allow all Magisk internal procedures to run properly (can be done manually by the `--magisk` option). It also contains quite a few additional patches so most scripts can run in the daemon before the full patch is done
|
||
|
- The full patch adds the rule `allow su * * *` on top of the preliminary rules. This is done because stock Samsung ROMs do not support permissive; adding this rule makes the domain effectively permissive. Due to the concern of greatly increasing the boot time, the Magisk daemon will not wait for this patch to finish until the boot stage `late_start` triggers. What this means is that **only `late_start` service mode is guaranteed to run in a fully patched environment**. For non-Samsung devices it doesn't matter because `u:r:su:s0` is permissive anyways, but for full compatibility, it is **highly recommended to run boot scripts in `late_start` service mode**.
|
||
|
|
||
|
Command help message:
|
||
|
|
||
|
```
|
||
|
Usage: magiskpolicy [--options...] [policystatements...]
|
||
|
|
||
|
Options:
|
||
|
--live directly apply patched policy live
|
||
|
--magisk built-in rules for a Magisk selinux environment
|
||
|
--load FILE load policies from <infile>
|
||
|
--save FILE save policies to <outfile>
|
||
|
|
||
|
If no input file is specified, it will load from current policies
|
||
|
If neither --live nor --save is specified, nothing will happen
|
||
|
|
||
|
One policy statement should be treated as one parameter;
|
||
|
this means a full policy statement should be enclosed in quotes;
|
||
|
multiple policy statements can be provided in a single command
|
||
|
|
||
|
The statements has a format of "<action> [args...]"
|
||
|
Use '*' in args to represent every possible match.
|
||
|
Collections wrapped in curly brackets can also be used as args.
|
||
|
|
||
|
Supported policy statements:
|
||
|
|
||
|
Type 1:
|
||
|
"<action> source-class target-class permission-class permission"
|
||
|
Action: allow, deny, auditallow, auditdeny
|
||
|
|
||
|
Type 2:
|
||
|
"<action> source-class target-class permission-class ioctl range"
|
||
|
Action: allowxperm, auditallowxperm, dontauditxperm
|
||
|
|
||
|
Type 3:
|
||
|
"<action> class"
|
||
|
Action: create, permissive, enforcing
|
||
|
|
||
|
Type 4:
|
||
|
"attradd class attribute"
|
||
|
|
||
|
Type 5:
|
||
|
"typetrans source-class target-class permission-class default-class (optional: object-name)"
|
||
|
|
||
|
Notes:
|
||
|
- typetrans does not support the all match '*' syntax
|
||
|
- permission-class cannot be collections
|
||
|
- source-class and target-class can also be attributes
|
||
|
|
||
|
Example: allow { source1 source2 } { target1 target2 } permission-class *
|
||
|
Will be expanded to:
|
||
|
|
||
|
allow source1 target1 permission-class { all-permissions }
|
||
|
allow source1 target2 permission-class { all-permissions }
|
||
|
allow source2 target1 permission-class { all-permissions }
|
||
|
allow source2 target2 permission-class { all-permissions }
|
||
|
```
|
||
|
|
||
|
|
||
|
### magisk
|
||
|
The magisk binary contains all the magic of Magisk, providing all the features Magisk has to offer. When called with the name `magisk`, it works as an utility tool with many helper functions, and also the entry point for `init` to start Magisk services. These helper functions are extensively used by the [Magisk Module Template](https://github.com/topjohnwu/magisk-module-template) and Magisk Manager.
|
||
|
|
||
|
Command help message:
|
||
|
|
||
|
```
|
||
|
Usage: magisk [applet [arguments]...]
|
||
|
or: magisk [options]...
|
||
|
|
||
|
Options:
|
||
|
-c print current binary version
|
||
|
-v print running daemon version
|
||
|
-V print running daemon version code
|
||
|
--list list all available applets
|
||
|
--install [SOURCE] DIR symlink all applets to DIR. SOURCE is optional
|
||
|
--createimg IMG SIZE create ext4 image. SIZE is interpreted in MB
|
||
|
--imgsize IMG report ext4 image used/total size
|
||
|
--resizeimg IMG SIZE resize ext4 image. SIZE is interpreted in MB
|
||
|
--mountimg IMG PATH mount IMG to PATH and prints the loop device
|
||
|
--umountimg PATH LOOP unmount PATH and delete LOOP device
|
||
|
--[init service] start init service
|
||
|
--unlock-blocks set BLKROSET flag to OFF for all block devices
|
||
|
--restorecon fix selinux context on Magisk files and folders
|
||
|
--clone-attr SRC DEST clone permission, owner, and selinux context
|
||
|
|
||
|
Supported init services:
|
||
|
daemon, post-fs, post-fs-data, service
|
||
|
|
||
|
Supported applets:
|
||
|
su, resetprop, magiskhide
|
||
|
```
|
||
|
|
||
|
### su
|
||
|
An applet of `magisk`, the MagiskSU entry point, the good old `su` command.
|
||
|
|
||
|
Command help message:
|
||
|
|
||
|
```
|
||
|
Usage: su [options] [-] [user [argument...]]
|
||
|
|
||
|
Options:
|
||
|
-c, --command COMMAND pass COMMAND to the invoked shell
|
||
|
-h, --help display this help message and exit
|
||
|
-, -l, --login pretend the shell to be a login shell
|
||
|
-m, -p,
|
||
|
--preserve-environment preserve the entire environment
|
||
|
-s, --shell SHELL use SHELL instead of the default /system/bin/sh
|
||
|
-v, --version display version number and exit
|
||
|
-V display version code and exit,
|
||
|
this is used almost exclusively by Superuser.apk
|
||
|
-mm, -M,
|
||
|
--mount-master run in the global mount namespace,
|
||
|
use if you need to publicly apply mounts
|
||
|
```
|
||
|
|
||
|
Note: even though the `-Z, --context` option is not listed above, it actually still exists for compatibility with apps using SuperSU. However MagiskSU will silently ignore the option since it's no more relevant.
|
||
|
|
||
|
### resetprop
|
||
|
An applet of `magisk`, an advanced system property manipulation utility. Here's some background knowledge:
|
||
|
|
||
|
System properties are stored in a hybrid trie/binary tree data structure in memory. These properties are allowed to be read by many processes (natively via `libcutils`, in shells via the `getprop` command); however, only the `init` process have direct write access to the memory of property data. `init` provides a `property_service` to accept property update requests and acts as a gatekeeper, doing things such as preventing **read-only** props to be overridden and storing **persist** props to non-volatile storages. In addition, property triggers registered in `*.rc` scripts are also handled here.
|
||
|
|
||
|
`resetprop` is created by pulling out the portion of source code managing properties from AOSP and try to mimic what `init` is doing. With some hackery the result is that we have direct access to the data structure, bypassing the need to go through `property_service` to gain arbitrary control. Here is a small implementation detail: the data structure and the stack-like memory allocation method does not support removing props (they are **designed NOT** to be removed); prop deletion is accomplished by detaching the target node from the tree structure, making it effectively invisible. As we cannot reclaim the memory allocated to store the property, this wastes a few bytes of memory but it shouldn't be a big deal unless you are adding and deleting hundreds of thousands of props over and over again.
|
||
|
|
||
|
Due to the fact that we bypassed `property_service`, there are a few things developer should to be aware of:
|
||
|
|
||
|
- `on property:foo=bar` triggers registered in `*.rc` scripts will not be triggered when props are changed. This could be a good thing or a bad thing, depending on what behavior you expect. The default behavior of `resetprop` matches the original `setprop`, which **WILL** trigger events (implemented by deleting the prop and set the props via `property_service`), but there is a flag (`-n`) to disable it if you need this special behavior.
|
||
|
- persist props are stored both in memory and in `/data/property`. By default, deleting props will **NOT** remove it from persistent storage, meaning the prop will be restored after the next reboot; reading props will **NOT** read from persistent storage, as this is the behavior of normal `getprop`. With the flag `-p` enabled, deleting props will remove the prop **BOTH** in memory and `/data/property`; props will be read from **BOTH** in memory and persistent storage.
|
||
|
|
||
|
Command help message:
|
||
|
|
||
|
```
|
||
|
Usage: resetprop [flags] [options...]
|
||
|
|
||
|
Options:
|
||
|
-h, --help show this message
|
||
|
(no arguments) print all properties
|
||
|
NAME get property
|
||
|
NAME VALUE set property entry NAME with VALUE
|
||
|
--file FILE load props from FILE
|
||
|
--delete NAME delete property
|
||
|
|
||
|
Flags:
|
||
|
-v print verbose output to stderr
|
||
|
-n set properties without init triggers
|
||
|
only affects setprop
|
||
|
-p access actual persist storage
|
||
|
only affects getprop and deleteprop
|
||
|
```
|
||
|
|
||
|
### magiskhide
|
||
|
An applet of `magisk`, the CLI to control MagiskHide. Use this tool to communicate with the daemon to change MagiskHide settings.
|
||
|
|
||
|
Command help message:
|
||
|
|
||
|
```
|
||
|
Usage: magiskhide [--options [arguments...] ]
|
||
|
|
||
|
Options:
|
||
|
--enable Start magiskhide
|
||
|
--disable Stop magiskhide
|
||
|
--add PROCESS Add PROCESS to the hide list
|
||
|
--rm PROCESS Remove PROCESS from the hide list
|
||
|
--ls Print out the current hide list
|
||
|
```
|