Question

I have the following code in a kernel module that walks up the process tree and prints the process names and uids up until the init process:

// recursivly walk the task's parent until we reach init

void parent_task_walk(struct task_struct *task) {

    struct task_struct *parent;
    char filename[MAX_FILE_LEN];

    if (task && task->mm) {

        parent = get_task_parent(task);

        printk("%s (uid:%d)", exe_from_mm(task->mm, filename, MAX_FILE_LEN),
            get_task_uid(task));

        if (parent && task->pid != 1) {
            printk(", ");
            parent_task_walk(parent);
        }
    }

}

NOTE: I make use of some macros that reference the real kernel functions, as this is for a kernel module the spans multiple versions. The source code is in this file: https://github.com/cormander/tpe-lkm/blob/319e1e29ea23055cca1c0a3bce3c865def14d3d2/core.c#L61

The output ends up looking something like this:

/bin/bash (uid:500), /usr/sbin/sshd (uid:500), /usr/sbin/sshd (uid:0), /usr/sbin/sshd (uid:0), /sbin/init (uid:0)


It’s a recursive function. As you can imagine, things go bad when you launch 200 bash shells and then trigger the event. I’m not sure exactly what happens, but the machine freezes. Kernel ran out of stack space I’m assuming, went OOM, and shot itself?

I’m wondering what’s the best way to handle that case. I see a few options:

1) stop walking the process tree after N processes

2) stop walking after some character array (that will eventually be printed) is full

3) make use of a goto instead of a recursive function, and yet still obey a new rule from options #1 and #2

4) use some other non-recursion method that you’ll articulate for me

This is happening in kernel space, so not the most hospitable environment. Can anyone give any pointers as to the best approach to take for this?

Answer 1

Indeed you kernel crash probably happens because you are using a recursive function.

You could just use a very simple loop instead of your recursive function… Do you come from a functional programming only world? A while loop is standard in C, but if you really want to use a goto you could… 😉 Here is the code using a loop.

void parent_task_walk(struct task_struct* task) 
{

    struct task_struct *parent = NULL;
    char filename[MAX_FILE_LEN];


    while (task && task->mm) {

        parent = get_task_parent(task);

        printk("%s (uid:%d)",
               exe_from_mm(task->mm, filename, MAX_FILE_LEN),
               get_task_uid(task));

        if (parent && task->pid != 1)
            printk(", ");

        task = parent;
    }

}