Palacios Public Git Repository

To checkout Palacios execute

  git clone http://v3vee.org/palacios/palacios.web/palacios.git
This will give you the master branch. You probably want the devel branch or one of the release branches. To switch to the devel branch, simply execute
  cd palacios
  git checkout --track -b devel origin/devel
The other branches are similar.


Remove non-multithreaded OS option. Change VM creation to be fully asynchronous....
[palacios.git] / linux_module / main.c
index a6787ce..9317635 100644 (file)
@@ -67,74 +67,54 @@ static long v3_dev_ioctl(struct file * filp,
            struct v3_guest * guest = kmalloc(sizeof(struct v3_guest), GFP_KERNEL);
 
            if (IS_ERR(guest)) {
-               printk("Error allocating Kernel guest_image\n");
+               printk("Palacios: Error allocating Kernel guest_image\n");
                return -EFAULT;
            }
 
            memset(guest, 0, sizeof(struct v3_guest));
 
-           printk("Starting V3 Guest... (%p)\n", guest);
+           printk("Palacios: Starting V3 Guest...\n");
 
            vm_minor = register_vm(guest);
 
            if (vm_minor == -1) {
-               printk("Too many VMs are currently running\n");
+               printk("Palacios Error: Too many VMs are currently running\n");
                return -EFAULT;
            }
 
            guest->vm_dev = MKDEV(v3_major_num, vm_minor);
 
            if (copy_from_user(&user_image, argp, sizeof(struct v3_guest_img))) {
-               printk("copy from user error getting guest image...\n");
+               printk("Palacios Error: copy from user error getting guest image...\n");
                return -EFAULT;
            }
 
            guest->img_size = user_image.size;
 
-           printk("Allocating kernel memory for guest image (%llu bytes)\n", user_image.size);
+           printk("Palacios: Allocating kernel memory for guest image (%llu bytes)\n", user_image.size);
            guest->img = vmalloc(guest->img_size);
 
            if (IS_ERR(guest->img)) {
-               printk("Error: Could not allocate space for guest image\n");
+               printk("Palacios Error: Could not allocate space for guest image\n");
                return -EFAULT;
            }
 
            if (copy_from_user(guest->img, user_image.guest_data, guest->img_size)) {
-               printk("Error loading guest data\n");
+               printk("Palacios: Error loading guest data\n");
                return -EFAULT;
            }      
 
            strncpy(guest->name, user_image.name, 127);
 
-           printk("Launching VM\n");
+           printk("Palacios: Launching VM\n");
 
            INIT_LIST_HEAD(&(guest->exts));
 
-           init_completion(&(guest->start_done));
-           init_completion(&(guest->thread_done));
-
-           { 
-               struct task_struct * launch_thread = NULL;
-               // At some point we're going to want to allow the user to specify a CPU mask
-               // But for now, well just launch from the local core, and rely on the global cpu mask
-
-               preempt_disable();
-               launch_thread = kthread_create(start_palacios_vm, guest, guest->name);
-               
-               if (IS_ERR(launch_thread)) {
-                   preempt_enable();
-                   printk("Palacios error creating launch thread for vm (%s)\n", guest->name);
-                   return -EFAULT;
-               }
-
-               kthread_bind(launch_thread, smp_processor_id());
-               preempt_enable();
-
-               wake_up_process(launch_thread);
+           if (start_palacios_vm(guest) == -1) {
+               printk("Palacios: Error starting guest\n");
+               return -EFAULT;
            }
 
-           wait_for_completion(&(guest->start_done));
-
            return guest->vm_dev;
            break;
        }