WEBVTT 00:08.000 --> 00:11.660 Now, if I ask you a question that what is schedulable? 00:11.660 --> 00:13.670 Is it threads or processes? 00:13.820 --> 00:17.630 So your kernel or operating system do not schedule the processes. 00:17.630 --> 00:25.160 It schedule the threads by schedule I mean that how operating system decides that which thread to allocate 00:25.190 --> 00:27.380 which CPU for execution. 00:28.350 --> 00:31.600 So it is the thread which is a schedulable entity. 00:31.620 --> 00:33.520 It is not a process. 00:33.540 --> 00:40.320 So in the diagram you can see on the right hand side, I have drawn a block diagram of a computer system 00:40.320 --> 00:45.060 which is mainly divided into two parts the user space and kernel space. 00:45.090 --> 00:51.060 Kernel space is a space where your kernel or operating system is running, whereas in user space all 00:51.060 --> 00:53.670 your applications are running right. 00:54.670 --> 00:58.120 Now, let us suppose that in user space there is a user process. 00:58.150 --> 00:58.810 P1. 00:58.840 --> 01:04.270 That user process P1 has two threads that is the thread T1 and threat T2. 01:04.300 --> 01:05.140 Right. 01:05.920 --> 01:12.550 Similarly, there is another process P2, which has two more threads, let us say T3 and T4. 01:12.700 --> 01:13.510 Right. 01:14.820 --> 01:19.620 So now your kernel or operating system will going to schedule the threads. 01:19.650 --> 01:22.380 T1, T2, T3 and T4. 01:22.410 --> 01:30.360 These four threads will compete for these three CPUs which are attached to the system as hardware resources. 01:31.100 --> 01:36.990 So remember, the operating system sees threads as schedulable entity and not a process. 01:37.010 --> 01:42.980 Here the competition for CPU is between four threads that is T1, T2, T3 and T4. 01:43.850 --> 01:49.160 As I said in the beginning section of this course, that thread is nothing, but it is a basic unit 01:49.190 --> 01:51.350 of execution flow, right? 01:51.350 --> 01:57.980 And whatever is execution flow is supposed to be allocated CPU by the operating system for execution. 01:58.550 --> 02:05.390 However, the fact that operating system sees thread as a schedulable entity and not the process is 02:05.390 --> 02:08.150 violated in certain error conditions. 02:08.390 --> 02:15.590 The first error condition is that that for some reason if a thread commits some crime and it is segmentation 02:15.590 --> 02:20.750 fault, then it is the entire process which is terminated by the operating system. 02:20.750 --> 02:21.470 Right? 02:21.470 --> 02:24.540 That is including all the threads of a process. 02:24.560 --> 02:32.600 So in this example, suppose the thread t1 do some illegal thing and and it is given a segmentation 02:32.600 --> 02:35.120 fault signal by the operating system. 02:35.120 --> 02:40.520 Then the entire process P1 will going to be terminated by the operating system. 02:40.520 --> 02:41.140 Right. 02:41.150 --> 02:46.730 It simply means that the thread T2 pays the penalty for the crime which is committed by the thread. 02:46.760 --> 02:48.260 T1 Right. 02:49.450 --> 02:54.490 Similarly, a signal is delivered per process and not per thread. 02:54.520 --> 02:59.920 Now, if you don't have an idea about signals, then it is better to understand this point. 02:59.950 --> 03:07.510 When you study signals at a high level, let me explain that when the kernel space or operating system 03:07.510 --> 03:11.770 delivers the signal as then it is delivered. 03:11.800 --> 03:17.260 Then the signal s is delivered at a process level and not at the thread level. 03:17.290 --> 03:21.160 One example of the signal s is segmentation fault signal. 03:21.520 --> 03:25.630 That is sigsegv signal, right. 03:25.660 --> 03:32.020 Whenever the signal is delivered by the operating system to the process, the entire process is terminated 03:32.020 --> 03:34.030 with segmentation fault error. 03:34.360 --> 03:40.990 Similarly, when you have a multithreaded program which is running and when you press control C, it 03:40.990 --> 03:45.880 is not one particular thread which is terminated, but the entire application is terminated. 03:45.880 --> 03:46.480 Right? 03:46.480 --> 03:52.310 So when you press control C the, the operating system delivers another type of signal to the process, 03:52.310 --> 03:55.520 which is called sigterm signal, right? 03:55.610 --> 04:02.060 So here what I am trying to say is that signal is something which works at the process level and not 04:02.060 --> 04:03.320 at the thread level. 04:03.350 --> 04:09.420 Now remember that we discussed the race condition which is created when a new thread is created, right? 04:09.440 --> 04:16.580 So the race condition on thread creation is due to the fact that which thread the kernel chooses to 04:16.580 --> 04:21.290 allocate CPU, whether it's a parent thread or it's a child thread. 04:21.290 --> 04:21.920 Right? 04:21.920 --> 04:25.790 So because of this indeterminism, the race condition is there. 04:26.940 --> 04:33.090 And the last point is that that the kernel schedules threads on multiple CPUs as per the scheduling 04:33.090 --> 04:33.840 policy. 04:33.870 --> 04:38.100 You must have studied in your academics that there are various types of scheduling policy. 04:38.100 --> 04:45.050 For example, first come, first serve policy or shortest job, first scheduling policy, etcetera. 04:45.060 --> 04:45.840 Right. 04:47.630 --> 04:53.870 So as part of this policy scheme, the kernel operating system chooses one thread among many which are 04:53.870 --> 04:57.410 competing for the CPU and allocates the CPU to that thread. 04:58.280 --> 05:02.630 Also note that in this diagram I have drawn three CPUs. 05:02.660 --> 05:03.420 Right? 05:03.440 --> 05:10.820 So it means that your operating system has three CPUs on which it has to schedule all the live threads 05:10.820 --> 05:12.020 running in the system. 05:13.420 --> 05:21.700 It can choose thread T1 to run on CPU one and it can choose thread t2 to run on CPU two and it can choose 05:21.700 --> 05:24.990 thread T3 to run on CPU three and so on. 05:25.000 --> 05:25.810 Right. 05:27.240 --> 05:33.150 So all this theoretical concept will help you to understand the advanced features of Multithreading, 05:33.150 --> 05:37.200 which we will going to discuss in this course in the subsequent sections.