Services can name their process name arbitrarily, for instance the service in
com.google.android.gms that is responsible for SafetyNet is named
com.google.android.gms.unstable. There are many apps out in the wild use
dedicated services with special names to detect root, and previously the user
is expected to add all of them to the hide list.
In this commit, we change from targeting process names to component names.
On Android, component names are composed of <pkg>/<cls>. When targeting
component names, we can always know what application spawned the new process.
This means that if the user adds a package name to the hidelist, MagiskHide can
now target ALL possible processes of that specific application.
To abide with this change, the default SafetyNet target is now changed from
com.google.android.gms.unstable (process name) to
com.google.android.gms/.droidguard.DroidGuardService (component name)
The database should only be accessed by a single process, which is magiskd.
This means 'magisk --sqlite [SQL]' has to be updated to pass the SQL command to the daemon.
In addition, open the database connection with SQLITE_OPEN_FULLMUTEX to support multithread in magiskd.
The utils function may be called in any situation, such as in daemon. We
should guarantee that all the resource got from this function released
normally.
Signed-off-by: yuchenlin <npes87184@gmail.com>
Introduce a new communication method between Magisk and Magisk Manager.
Magisk used to hardcode classnames and send broadcast/start activities to
specific components. This new method makes no assumption of any class names,
so Magisk Manager can easily be fully obfuscated.
In addition, the new method connects Magisk and Magisk Manager with random
abstract Linux sockets instead of socket files in filesystems, bypassing
file system complexities (selinux, permissions and such)
Boot services tend to fail in the middle when the kernel loads a sepolicy live.
It seems that moving full patch (allow magisk * * *) to late_start is still not enough to fix service startup failures.
So screw it, apply all patched in magiskinit, which makes sure that all rules are only loaded in a single step.
The only down side is that some OEM with a HUGE set of secontexts (e.g. Samsung) might suffer a slightly longer boot time, which IS the reason why the rules are split to 2 parts in the first place.
In previous versions, magiskinit will not early mount if /sepolicy is detected. However on OP5/5T latest betas, the devices are fully trebelized,
but for some reason the file /sepolicy still exists, making magiskinit think it is NOT a treble device and doesn't work properly.
So to properly fix this issue, I will have to use the "official" way - check fstab in device trees. Any block mentioned in the fstab in device trees
are supposed to be early mounted. Currently magiskinit will only mount system and vendor even if other partitions exists in the dtb fstab, since other
partitions are not used to construct sepolicy (currently).
These changes can also fix#373, since we dynamically detect PARTNAME from device trees.
BusyBox is unable to run properly on non-root applications due to seccomp introduced in Android 8.0.
The SDK-21 libc.a has system call wrappers that uses the system calls on the whitelist, so binaries compiled with the updated libc can work properly.
1. Introduce new applet: imgtool for better separation from the main program
2. Actually mount the image and check statvfs for free space in the image
This shall eliminate any possible module installation failure from image resizing issues.
The &cmd will return a pointer which point to a pointer of cmdline.
It is a memory address which is usually 8 bytes in 64 bits machine.
However, the struct cmdline is 4 bytes. This will cause setting zero
beyond the bound.
Below is a simple example to show the differentiation:
struct cmdline {
char skip_initramfs;
char slot[3];
};
static void parse_cmdline(struct cmdline *cmd)
{
printf("%lu\n", sizeof(*cmd)); /* 4 */
printf("%lu\n", sizeof(&cmd)); /* 8 */
}
int main()
{
struct cmdline cmd;
parse_cmdline(&cmd);
return 0;
}
This patch prevents this.
Signed-off-by: npes87184 <npes87184@gmail.com>
In previous implementations, proc_monitor checks whether the mount namespace of an app is actually separated from zygote using a list generated at startup.
However, for some unknown reason, some devices (e.g. Samsung) has multiple zygote servers running in the background.
This means that app processes spawned from the unlisted zygotes are not checked whether the separation is done or not, causing MagiskHide unmount stuffs in the namespace of zygote, and since zygote is the "mother" of all apps, all apps will no longer have root access.
Since I'm not sure of the reason why multiple zygotes exists, so instead of checking the namespace against a list, compare the current namespace against the parent process's namespace.
This will make sure the namespace is NOT the same as the parent process, which is supposed to be the zygote server.
It's not important to check the return value of unlink(2) or even verify
that the file exists. If this code is running, it means the system has
rebooted, and thus the update file, if any, should be removed so that
MagiskManager doesn't keep displaying the same message. We also handle
this before we handle "disable" so that disabled modules don't keep
requesting a reboot to update.
Unlike other common OEMs, Samsung use uppper case partition name.
e.g: /dev/block/platform/11120000.ufs/by-name/SYSTEM
This will cause setup_block() fails to find a match partition.
Thus, we should use strcasecmp instead of strcmp.
Signed-off-by: msdx321 <msdx321@gmail.com>
This fixes an issue where if /system or /vendor was already mounted
prior to magiskinit running, then they would get unmounted.
Signed-off-by: Andrew Gunnerson <andrewgunnerson@gmail.com>
