tfp0 patch

In the XNU kernel, task_for_pid is a function that allows a (privileged) process to get the task port of another process on the same host, except the kernel task (process ID 0). A tfp0 patch (or task_for_pid(0) patch) removes this restriction, allowing any executable to call task_for_pid for pid 0 (hence the name) and then use vm_read and vm_write to modify the kernel VM region. The entitlements get-task-allow and task_for_pid-allow are required to make AMFI happy.

Example code

The following C program calls task_for_pid and returns the error code:

#include <mach/mach.h>

// Compile and fakesign with entitlements (on-device; LLVM+Clang and ldid must be installed):
// cc -o tfp0 tfp0.c && ldid -Stfp0.plist tfp0

int main(void) {
    mach_port_t kernel_task = 0;
    return task_for_pid(mach_task_self(), 0, &kernel_task);
}

The returned error code, which can be checked using echo $? in bash after running the test program, will be 0 if the call succeeded. If it did not, a positive number, e.g. 5 (KERN_FAILURE), is returned instead (see kern_return.h for possible values). The entitlements plist (named tfp0.plist in this example) for ldid can look like this:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>get-task-allow</key>
    <true/>
    <key>run-unsigned-code</key>
    <true/>
    <key>task_for_pid-allow</key>
    <true/>
</dict>
</plist>