WEBVTT

1
00:00:06.060 --> 00:00:06.660
William Cheng: Testing.

2
00:01:00.600 --> 00:01:31.980
William Cheng: Good morning. Welcome to lecture 14

3
00:01:33.330 --> 00:01:37.290
William Cheng: Any question about the recorded lecture video

4
00:01:38.970 --> 00:01:45.420
Hardik Mahipal Surana: Well, good manga for so no questions actually. I'm just here to see if anyone else had an equal sign it an opportunity to go through the lecture yet.

5
00:01:46.470 --> 00:01:50.250
William Cheng: Yeah, I think I don't know people studying for that matter more people are

6
00:01:51.870 --> 00:01:56.340
William Cheng: You know, doing Colonel to or something like that. Any question about the Colonel, Simon.

7
00:01:57.600 --> 00:02:09.570
Hardik Mahipal Surana: I'm not yet. I went ahead and looked at the to make note the portable for some part. So we were able to create the devices. Now we're just trying to figure out how to communicate.

8
00:02:10.620 --> 00:02:22.230
Hardik Mahipal Surana: With the occasion like once their devices are created Keisha create will work. And then we were just trying to understand what the flow should be for actually communicating by typing on the keyboard.

9
00:02:23.610 --> 00:02:45.270
William Cheng: Right, so, so, so in Colonel to the Keisha is using the virtual file system. Right. So, you know, once you created the directly, I guess you need to create dev slash TTY zero RY slash dev slash knowledge that says what the budget device, you have to create and and then

10
00:02:46.740 --> 00:02:51.270
William Cheng: I think if you step through the Keisha code. I think Keisha I should call open right

11
00:02:52.020 --> 00:02:53.160
Hardik Mahipal Surana: Yeah, it goes to

12
00:02:53.970 --> 00:03:00.120
William Cheng: Right, so, so, so do open is pretty complicated, right, because open, it's going to do path same resolution.

13
00:03:00.600 --> 00:03:07.290
William Cheng: Right, so you need all the function name VI de see all those things has to work. So once it opens, then

14
00:03:08.910 --> 00:03:23.670
William Cheng: Then, then the Keisha can start calling you know guess they tried to print a prompt, so he has called right so you got to make sure that the do right will eventually get to the special the special right

15
00:03:24.720 --> 00:03:29.280
William Cheng: And then it will call do read to try to read a line from the keyboard. Right.

16
00:03:31.530 --> 00:03:32.280
Hardik Mahipal Surana: Okay, okay.

17
00:03:32.940 --> 00:03:39.570
William Cheng: So as a writer needs to work so you will see a prompt and then whatever you type really needs to work so that you will get a line from the, from the user.

18
00:03:40.170 --> 00:03:49.380
William Cheng: And then then then then the rest is will be similar to Colonel why right will parse the command line and try to sort of figure out what you want the THE CASE OUT TO DO, etc.

19
00:03:51.480 --> 00:04:03.540
Hardik Mahipal Surana: Yeah, actually I had a question regards to partner resolution why those implementing the local function. So I went to the floor. And I found that a ram offense is actually calling we get

20
00:04:04.140 --> 00:04:12.870
Hardik Mahipal Surana: Which caused the ref so when we're looking at for apart. Why do we need to implement the reference count of each component to that we are looking at for

21
00:04:14.280 --> 00:04:17.220
William Cheng: A big goes through the entire part

22
00:04:17.550 --> 00:04:20.280
Hardik Mahipal Surana: And for each component we call look upgrade.

23
00:04:20.760 --> 00:04:29.070
William Cheng: Right, so, so, so the, you know, the point is that if you want to do a correctly so that when you switch the colonel three everything will work.

24
00:04:29.490 --> 00:04:35.910
William Cheng: Then you should increment the reference count, you know, you can look at the REM FS Co. Everything is implementing RAM. There's no disconnect

25
00:04:36.750 --> 00:04:40.620
William Cheng: So, therefore, you don't have to worry about anything. But if you imagine that if you have a real disk.

26
00:04:40.950 --> 00:04:47.580
William Cheng: Then when you are performing that look up, you need to read one directory entry at a time. So you might have to go to the this multiple times.

27
00:04:47.910 --> 00:05:02.880
William Cheng: So in that case of Colonel throw fall asleep, and if you don't increment the reference cow. Maybe one time you're unlucky, you know, the you know the the the the the reference count go to zero for you because of some other thread or something like that, then you got to be in trouble.

28
00:05:05.520 --> 00:05:06.360
Hardik Mahipal Surana: Okay, so

29
00:05:06.840 --> 00:05:07.590
Hardik Mahipal Surana: We have done within

30
00:05:08.040 --> 00:05:13.140
William Cheng: Again, so you know where that's a very important point, right, because in the virtual process them if you

31
00:05:13.620 --> 00:05:23.190
William Cheng: If you have a pointer to one of these objects implement the reference to say I'm still using it. I'm not done with it. So, therefore, you better not delete it. Right. So that's why I increment the reference count.

32
00:05:24.930 --> 00:05:37.320
William Cheng: Okay, and later on when the function returns. I mean, it's a local variable, the stack will get POP So so so so you know the the reference is automatically removed, but if you don't document the reference call you're going to end up with over referencing

33
00:05:38.130 --> 00:05:40.650
Hardik Mahipal Surana: Exactly. So once you're done with the Luca

34
00:05:42.450 --> 00:05:52.110
Hardik Mahipal Surana: And the free from once we're done with the lookup, we should be. And again calling the put to discriminate the reference country because you're done with the look up and we no longer are using that we know

35
00:05:52.920 --> 00:06:00.210
William Cheng: Right, so, so if within the function, right. So, so let's say over here, right. Let's look at this right so you have a function right here.

36
00:06:00.720 --> 00:06:08.760
William Cheng: Okay, I'll be here. You say, you know, like, say, the node or something is there to object has reference can read the file has a reference call and the vino has a reference com

37
00:06:09.360 --> 00:06:15.810
William Cheng: So let's say you have a you know vino pointer over here, you point to something. Right. So as soon as you point to something you should increment the reference count.

38
00:06:16.320 --> 00:06:25.650
William Cheng: Increment reference count. Okay, so when the function is about to return. And this is a local variable, right. So the node star or something like that. Right.

39
00:06:25.980 --> 00:06:28.950
William Cheng: When this function return this pointer is going to disappear.

40
00:06:29.490 --> 00:06:33.210
William Cheng: Right. So, therefore, you have to document the reference com so that they sort of match out

41
00:06:35.010 --> 00:06:35.610
Hardik Mahipal Surana: Okay, okay.

42
00:06:35.910 --> 00:06:43.530
William Cheng: So this way. When you return, then the reference count will be correct. You got to be very, very careful because some of the functions over here will automatically increment the reference now.

43
00:06:44.520 --> 00:06:46.200
Hardik Mahipal Surana: Look up to automate caters increment.

44
00:06:46.230 --> 00:06:54.600
William Cheng: Right, so, so if it automatically included a reference count then then if you detriment. I mean, so, so again if you if you implemented again there will be an extra reference

45
00:06:54.720 --> 00:06:58.290
William Cheng: So therefore, we recommend you with the will remove the actual reference

46
00:06:58.830 --> 00:07:06.600
William Cheng: Right. But if you don't incremental. Are you document it because you say, oh, inside this function they automatically include the the wrestling calm and as soon as you've documented it will

47
00:07:07.020 --> 00:07:08.850
William Cheng: Not go to zero, and then it will get freed up

48
00:07:09.810 --> 00:07:16.110
William Cheng: So again, the matching over here. You need to do it very carefully explicitly increment and why you should definitely explicitly detriment it

49
00:07:16.440 --> 00:07:26.430
William Cheng: But if you if it's incremental implicitly, then you need to decide whether it's a good idea to document or not. Okay. I think most cases if it's, you know, if a function

50
00:07:27.390 --> 00:07:32.850
William Cheng: Sort of increase the reference how you should read the comment very, very carefully and

51
00:07:33.240 --> 00:07:43.770
William Cheng: Or look at the rim of our system CO and to see if its reasonable. I mean, you know, if you only create a reference temporarily or you know what this function return that reference should continue to go up.

52
00:07:44.790 --> 00:07:47.460
William Cheng: Right. So if the reference continue to go on what, then you shouldn't document it.

53
00:07:49.530 --> 00:07:50.760
Hardik Mahipal Surana: I see. Okay.

54
00:07:51.450 --> 00:07:59.820
William Cheng: Okay, so do I. I don't know if there's a fixed rule, you know, I mean, if it's a local variable. And you do this temporarily. Well, then, that in this case you should document it.

55
00:08:00.090 --> 00:08:05.490
William Cheng: But all the other cases, you need to think about it very, very carefully look at a comment and the decide what to do.

56
00:08:06.510 --> 00:08:12.990
Hardik Mahipal Surana: Okay, so if we take the example of the Navy, which takes apart as a string and does

57
00:08:13.290 --> 00:08:21.420
Hardik Mahipal Surana: A tentative look up on each component of the path. So until we reach the very end of the path. And we are done with looking up the last part, we should still store the

58
00:08:21.900 --> 00:08:29.910
Hardik Mahipal Surana: reference count as is right and as an where once we're done with their we need to be treated again and then sort of detriment the reference point for each component

59
00:08:31.650 --> 00:08:45.630
William Cheng: Um, I mean I you know I don't, I, I didn't write the code. So I don't know exactly whether I should do that or not. So I think, you know, again, I mean, I shouldn't really tell you a code right in this case.

60
00:08:45.990 --> 00:08:56.310
William Cheng: Yeah, so it's more appropriate to talk about that in the class Google group, right, you said that was inside the envy, you know, what do you guys think the, you know, there's a, there's a, I guess there's iteration going on.

61
00:08:57.360 --> 00:09:02.550
William Cheng: So again, you should sort of analyze the situation and find out whether you should you know where you should increment where you should detriment. I

62
00:09:03.180 --> 00:09:08.220
William Cheng: Go edit the point over here is that is that, you know, in Colonel to you're dealing with the RAM file system.

63
00:09:08.430 --> 00:09:14.250
William Cheng: So the ranch houses and never go to sleep. So, therefore, you know, you don't have to worry about it. But when you go to Colonel Thierry

64
00:09:15.240 --> 00:09:27.000
William Cheng: You know what, whatever function that that that you call it, there's a possibility that you'll Colonel, so we'll go to sleep. Well, in that case, if you have a pointer to an object that has a reference count, you should increment it as soon as function returns your document.

65
00:09:29.250 --> 00:09:30.600
William Cheng: So that would be a general rule.

66
00:09:31.800 --> 00:09:32.220
Hardik Mahipal Surana: Good.

67
00:09:33.060 --> 00:09:39.420
William Cheng: Whatever function that you call if you're Colonel second fall asleep and you have a pointer to an object that has a reference count.

68
00:09:39.660 --> 00:09:40.740
Hardik Mahipal Surana: Any question you have in

69
00:09:41.100 --> 00:09:43.200
William Cheng: Common and Decker minute right just to be safe.

70
00:09:43.590 --> 00:09:44.250
Hardik Mahipal Surana: Okay, okay.

71
00:09:45.840 --> 00:09:55.980
William Cheng: Now, what happened is that people use local variable to point to, you know, many different things instead of function, right. So, so, yes. So the question is that when you point to one thing you equipment, the reference car.

72
00:09:56.160 --> 00:10:05.850
William Cheng: And then we'll use the point of the point to some other thing that you have a document that references the reference is that a function will go up and down like crazy. Is that correct, well, if you do a very, very carefully. Then we'll be correct.

73
00:10:07.020 --> 00:10:10.890
William Cheng: So you want to make sure that you don't end up with extra references or you don't have, you know,

74
00:10:12.270 --> 00:10:13.020
William Cheng: Under referencing

75
00:10:15.480 --> 00:10:32.400
Hardik Mahipal Surana: God or more general question, not with respect to remix, but in general about Linux. So in a multi processing environment. If we were to manage reference counting to make that also thread safe or should we be using on the text, whenever we are you come into detrimental reference don't

76
00:10:34.020 --> 00:10:42.660
William Cheng: Well, so right now we're talking about reference caught in the kernel. Right. Well, you're talking about multi threading. Are you talking about that's already in the user space program.

77
00:10:43.110 --> 00:10:44.220
Hardik Mahipal Surana: Yeah, and the user space.

78
00:10:45.030 --> 00:10:50.460
William Cheng: Yeah, I mean, all these things happening is that a colonel, the user space program is not going to see it. So, you know,

79
00:10:52.350 --> 00:10:56.910
William Cheng: I mean, they use a space for like one month one and two a month you you don't even use reference counting

80
00:10:58.680 --> 00:11:06.810
Hardik Mahipal Surana: Yeah, we don't. But yeah, makes sense. So what would happen in the kernel. Then in a multi processing. Remember

81
00:11:07.440 --> 00:11:11.760
William Cheng: What's a minute understanding is that Linux is one step process.

82
00:11:12.840 --> 00:11:14.430
Hardik Mahipal Surana: Oh yeah, it's a one by one more.

83
00:11:15.210 --> 00:11:19.680
William Cheng: Right. But, but is that a colonel all the Colonel, everything works together, things are the colonel right

84
00:11:20.010 --> 00:11:23.010
William Cheng: Yeah so. So even if you think about the sort of the kernel.

85
00:11:23.250 --> 00:11:35.130
William Cheng: For the corresponding user through that once a user. So I come inside a colonel it become a colonel threat and all the Colonel's where they need to work with each other. So they all follow the rules, right. So, so, yeah. So therefore, they're all being planted correctly.

86
00:11:36.840 --> 00:11:39.780
William Cheng: I'm the application poem, it doesn't get to put a thread inside the carnal.

87
00:11:40.470 --> 00:11:45.930
Hardik Mahipal Surana: Yeah, it is. It really actually become the colonel Thierry runs it executes the trap and the controls.

88
00:11:45.930 --> 00:11:53.220
William Cheng: Right, so you don't have to worry about the user. The user. The user. So I will not do the right thing because as soon as it become comes out the Colonel.

89
00:11:53.520 --> 00:12:01.620
William Cheng: Then, then it you know it, then all the code is done. They're done by the colonel programmer, so therefore inside Colonel, everything will be done correctly.

90
00:12:03.240 --> 00:12:03.870
Hardik Mahipal Surana: Okay, yeah.

91
00:12:04.350 --> 00:12:06.510
William Cheng: I mean for Linux for football meanings, where

92
00:12:06.510 --> 00:12:09.960
William Cheng: We are the one that writing the code for the Colonel. So we need to make sure that they're done correctly.

93
00:12:17.250 --> 00:12:18.210
Hardik Mahipal Surana: Okay, thank you so much.

94
00:12:18.420 --> 00:12:18.720
Okay.

95
00:12:32.070 --> 00:12:40.710
William Cheng: Anyone else has a question about the lecture 14 or lectures in general or, you know, we seven discussion section.

96
00:12:41.340 --> 00:12:54.120
William Cheng: So I think what I'm trying to do what I'm going to do is that, you know, today I will record the week eight discussion section. So, so, just in case people want to work on Colonel to they can you know that they can look at all the materials.

97
00:12:58.860 --> 00:13:05.010
Lalit Gupta: I have a question regarding elect to functions and proud dot c in the Phoenix source code.

98
00:13:05.400 --> 00:13:05.790
William Cheng: Okay, so

99
00:13:06.480 --> 00:13:09.930
Lalit Gupta: We have to function which is talking for an analyst info.

100
00:13:11.520 --> 00:13:12.120
I'm wanting

101
00:13:15.780 --> 00:13:19.020
Lalit Gupta: Actually provided to us proc info and populist info.

102
00:13:19.470 --> 00:13:20.070
Right.

103
00:13:21.630 --> 00:13:25.140
William Cheng: Well, maybe. Maybe I should switch to take a look there.

104
00:13:37.200 --> 00:13:41.160
William Cheng: Practice see right here. So one is called proc

105
00:13:42.270 --> 00:13:43.740
William Cheng: List of Prague info.

106
00:13:46.920 --> 00:13:47.640
William Cheng: Okay.

107
00:13:50.970 --> 00:13:51.630
Lalit Gupta: What is this

108
00:13:53.250 --> 00:13:53.880
William Cheng: What is what

109
00:13:55.050 --> 00:13:56.730
Lalit Gupta: Is the

110
00:13:58.590 --> 00:13:59.010
Lalit Gupta: Path.

111
00:13:59.700 --> 00:14:04.170
William Cheng: Yeah, so I think this kind of code is used by Python.

112
00:14:05.460 --> 00:14:05.910
Lalit Gupta: Okay.

113
00:14:06.690 --> 00:14:12.630
Lalit Gupta: Yeah, so some of the colonel commands like, you know, Colonel info Colonel Prague, or something like that.

114
00:14:12.840 --> 00:14:22.320
William Cheng: Those are the GDP commands the GD commands are implemented in Python, and the Python is going to call all these you know info function. So there's also the proc list info.

115
00:14:24.870 --> 00:14:33.540
William Cheng: Okay, so, yeah. So I think when you type the command Colonel proc in GDP this code will get executed.

116
00:14:35.430 --> 00:14:37.560
Lalit Gupta: Okay, so if you type Colonel proc

117
00:14:37.860 --> 00:14:53.400
William Cheng: Right. I think this one or the other one is going to get executed. I don't know exactly which one it is. I tried to, you know, try to debug this ones that I couldn't figure out exactly how this work that hard to go from the Python code to call the C code and get all this to work.

118
00:14:54.540 --> 00:15:00.360
Lalit Gupta: Okay, so. But the thing is, in general in GDP, you can actually cause a function like temporarily to see what

119
00:15:01.950 --> 00:15:11.040
Lalit Gupta: Is it possible to call. I was trying to collect call this function of broccoli broccoli listed for directly from God, but I was not able to know

120
00:15:11.640 --> 00:15:24.780
William Cheng: It's not directly. So in we're in GDP, because they Colonel help anyway show you all the colonel GDP commands and then you will guess. One of them is Colonel Prague and then the otherwise Colonel

121
00:15:25.560 --> 00:15:38.340
William Cheng: I can't, I can't remember the you know the example I guess there's a list of command that that you can that you can use. And then what it will do is it will show you all this information if you if you give the right command.

122
00:15:40.320 --> 00:15:45.150
William Cheng: Okay, so let's see if we there's a Python directory CD to Python.

123
00:15:46.650 --> 00:15:52.170
William Cheng: Python. We next over here and then we can say see proctor Python right

124
00:15:53.280 --> 00:16:00.030
William Cheng: Okay. So over here, it will show you what are the possible commands. I guess it doesn't really show you what the commands are

125
00:16:05.040 --> 00:16:05.880
William Cheng: Yeah. Anyways,

126
00:16:07.140 --> 00:16:09.300
William Cheng: I'm not that good at Python. So

127
00:16:10.290 --> 00:16:18.540
William Cheng: Guys over here, you can see that there's a proc lists of something. Right. It says, you know, go to the winnings and listen to load the proc list function or whatever.

128
00:16:18.780 --> 00:16:28.530
William Cheng: And then it has these data structure that it will use and you will follow the pointer and then they will call proc right. Where's progress progress right I guess proc as a constructor for this class.

129
00:16:29.910 --> 00:16:34.740
William Cheng: Anyways, so if you are, you know, the Python programmer, maybe we can figure it out.

130
00:16:36.150 --> 00:16:41.370
Lalit Gupta: So the talking for is not a secret because I was in because some other secret.

131
00:16:43.110 --> 00:16:51.360
Lalit Gupta: In GDP, we can actually call the function and it can give us a certain value is it, is it not possible to do it in profile profitable.

132
00:16:52.350 --> 00:17:01.320
William Cheng: I have no idea. Okay. Okay. So, so the only thing I grab proc info, Colonel. Oops.

133
00:17:07.170 --> 00:17:11.010
William Cheng: Grab properly info Colonel star star see

134
00:17:12.180 --> 00:17:13.530
William Cheng: Guess it's not being used anywhere.

135
00:17:15.270 --> 00:17:16.470
William Cheng: Including

136
00:17:24.630 --> 00:17:28.260
William Cheng: provide detailed debugging information about a given process.

137
00:17:30.930 --> 00:17:37.920
William Cheng: Yes, I know. So in GDP. You say, Colonel help. And then there was sort of show you a bunch of options that you can try all these commands over here.

138
00:17:39.120 --> 00:17:40.800
William Cheng: So it's it's it's

139
00:17:41.910 --> 00:17:46.590
William Cheng: Called a word debug right so in this header file. There are these two things that are here says debug.

140
00:17:47.730 --> 00:17:56.040
William Cheng: I think they means debugging under GDP not you know debug under seat. So, of course, you know, you can also call these functions yourself.

141
00:17:57.210 --> 00:18:04.680
William Cheng: If you know you know what the passes the first argument over here at the first argument we hear his voice star. So, if you look at a co

142
00:18:05.460 --> 00:18:09.780
Lalit Gupta: Yeah services blogger underscore te the type of this pointer to that.

143
00:18:11.370 --> 00:18:12.000
William Cheng: Okay.

144
00:18:13.560 --> 00:18:24.750
William Cheng: Yeah. So over here, he was, he can see that they typecast this one to the prop data structure. So, so if you, you can actually call this function yourself if you want to print this out right well you're running your approval and when you run your code.

145
00:18:25.830 --> 00:18:31.140
William Cheng: So this one. What is argh over here. So proc lists info. Whoops.

146
00:18:37.470 --> 00:18:42.060
William Cheng: Oh, this one on the apprentice. If I should ignore the first argument. It just, you know,

147
00:18:43.800 --> 00:18:50.340
William Cheng: It the okay sir. Make sure the first argument is no so the progress the info less information about all the processes.

148
00:18:51.510 --> 00:18:57.630
William Cheng: Say I saw you and GDP. If you type a type Colonel proc so this function will get calls

149
00:18:59.010 --> 00:19:08.010
William Cheng: Coordinate. Okay. Yes. What is that, oh, OK. So I think I know what I was doing over here right i printer print the information into this buffer.

150
00:19:08.850 --> 00:19:17.730
William Cheng: So it doesn't get printed to standard out it will print into this to this buffer and this buffer is passed back to Python. So Python can actually print it onto the screen.

151
00:19:19.800 --> 00:19:22.740
Lalit Gupta: Are even allocating the buffer, because we're is this before.

152
00:19:23.910 --> 00:19:24.330
Lalit Gupta: Because

153
00:19:25.410 --> 00:19:28.740
Lalit Gupta: Python is a Python is basically allocating memory for the

154
00:19:29.160 --> 00:19:33.330
William Cheng: Enterprise. So I think if you want to call this function yourself, you need to allocate a buffer.

155
00:19:34.470 --> 00:19:35.010
Lalit Gupta: Okay.

156
00:19:35.490 --> 00:19:47.430
William Cheng: Okay, so, so, so you will do something like this, right. So what I would do is that, you know, so let's say I have a function here, right, like this, right, if I have a function over here. I say, Joe char, you know, 4096 or something like that. Right.

157
00:19:47.910 --> 00:19:58.830
William Cheng: Or you know above 4096 and then I will call proc lists info and then all right buff size of buff.

158
00:19:59.910 --> 00:20:10.350
William Cheng: And then over here. And when this function return this one should return a size t. Right. So you can say, you know, size to length equal to this. Right.

159
00:20:10.650 --> 00:20:18.090
William Cheng: And then you can call you know right you know the basically you can take the the buffer and just write it out to write, write it out to standard out

160
00:20:19.260 --> 00:20:23.490
William Cheng: Okay, so you can actually, I can do GDP does. Sorry.

161
00:20:26.250 --> 00:20:41.310
William Cheng: The big, the big 10% s and then maybe I had access and over here buff. Okay, so, so if I write it like this, then it only works if Bob is terminated by zero.

162
00:20:41.970 --> 00:20:49.440
William Cheng: Okay, then what it will do is it will you know if God debugging has tend to include include a then this way, it will print out this buffer.

163
00:20:50.970 --> 00:20:51.270
Oh,

164
00:20:52.920 --> 00:20:56.880
William Cheng: Okay, or you can say, you know, do right I do right

165
00:20:59.430 --> 00:21:08.910
William Cheng: I guess do right over here, require file descriptor so standard in is one and then you can say buff buff and length or something like that.

166
00:21:10.410 --> 00:21:15.150
William Cheng: Yeah, I'm just making this up. I assume that this is going to work out but you should try to see if this actually work.

167
00:21:15.690 --> 00:21:23.220
William Cheng: The stuff with the Python stuff is that if you make one little mistake Python is just going to crash. And then, you know, then, then you have to get out of that debugger and start all over again.

168
00:21:26.370 --> 00:21:29.070
William Cheng: So the yeah the Python stuff is pretty sensitive

169
00:21:34.590 --> 00:21:35.310
Lalit Gupta: So I have

170
00:21:37.080 --> 00:21:41.370
Lalit Gupta: Maybe a slight question on lecture 11 slide 33

171
00:21:42.030 --> 00:21:42.810
Okay.

172
00:21:52.320 --> 00:22:04.710
William Cheng: Let me go there lecture 1153 8383

173
00:22:12.090 --> 00:22:12.510
William Cheng: Okay.

174
00:22:14.130 --> 00:22:15.660
Lalit Gupta: So in this thing, this

175
00:22:17.100 --> 00:22:22.410
Lalit Gupta: What is on the right side of the file descriptor is is the part of the FS right

176
00:22:23.250 --> 00:22:25.650
William Cheng: Yeah, so, so this is a Fs. Yeah.

177
00:22:26.760 --> 00:22:28.170
Lalit Gupta: And what is this middle thing.

178
00:22:30.000 --> 00:22:30.180
William Cheng: The

179
00:22:30.540 --> 00:22:35.100
William Cheng: Table now that this whole thing is in the in the bfs.

180
00:22:36.630 --> 00:22:37.740
Lalit Gupta: Okay, so okay so

181
00:22:38.190 --> 00:22:42.150
William Cheng: Yeah, so the picture that I drove me here. It's an array of fall objects.

182
00:22:44.820 --> 00:22:53.430
William Cheng: Okay, so every one of those everyone roll over here is a file object right every file object has these four fields reference count access cursor position and the I know pointer.

183
00:22:56.730 --> 00:23:02.340
Lalit Gupta: Okay, this is dis dis data structure which is in the middle part of the HDFS

184
00:23:03.240 --> 00:23:13.380
William Cheng: Right. So this was, it's a system. Why, and I it's a system file table or the file object table. So it's a, you know, it says there's a data structure that holds all the file objects.

185
00:23:17.070 --> 00:23:19.050
Lalit Gupta: Or additional one table big table.

186
00:23:19.620 --> 00:23:30.060
William Cheng: Yeah so. So on the left hand side. This is the PR process file descriptor table. So every process is about discovery table in the middle data structure is shared by all processes.

187
00:23:33.030 --> 00:23:33.330
Lalit Gupta: If

188
00:23:35.040 --> 00:23:39.600
Lalit Gupta: You can you point like in Phoenix, what is a system wide table intervening

189
00:23:40.140 --> 00:23:41.160
William Cheng: Oh, there's no such thing.

190
00:23:41.940 --> 00:23:42.660
Lalit Gupta: There's no such thing.

191
00:23:42.780 --> 00:23:50.280
William Cheng: Well, it's just an abstraction that there's a there's a file object table, there's some kind of a data structure for you to find a file object.

192
00:23:50.940 --> 00:23:59.460
William Cheng: Okay, because when you will when you create a file object, you need to keep track of them. So there's going to be some sort of a data structure that keeps a collection of objects.

193
00:24:01.440 --> 00:24:13.500
William Cheng: Okay so i i don't know what kind of data structures. Maybe it's a link list. Maybe it's an array I doesn't really matter what kind of data structures. Right. So it's just a, you know, so it's just a way for you to keep track of a bunch of objects.

194
00:24:17.460 --> 00:24:29.520
Lalit Gupta: And do it. Does the middle thing. Do we have to. Is this the responsibility for our for our or is this is the middle table no responsibility for the as

195
00:24:31.170 --> 00:24:39.930
William Cheng: Well, I mean, you know, I mean, the whole thing is inside of Colonel right so you could call the right hand side. A FS right then the left hand side over here is going to be BFS. Right.

196
00:24:42.030 --> 00:24:44.640
William Cheng: I mean fall object is implemented inside BFS.

197
00:24:48.270 --> 00:24:51.900
William Cheng: Where every one of these things that fall object or fall object is a VM FS object.

198
00:24:53.580 --> 00:24:54.120
Cookie.

199
00:24:58.080 --> 00:25:01.830
William Cheng: Right. So in chapter one. We talked about the file object. Right. So the follow up jag is

200
00:25:02.820 --> 00:25:14.040
William Cheng: You know, which is independent of the, the actual file system. So what's inside apologetic. He has a reference count it has how you open the file. There's a cursor position. All these things are BFS concepts.

201
00:25:17.130 --> 00:25:21.300
Lalit Gupta: Okay, and where to go. Different would be a video or an iOS.

202
00:25:22.380 --> 00:25:28.380
William Cheng: So vino is a weenie x, y. So, so, so, so I know it's what's inside the actual file system. Right.

203
00:25:29.940 --> 00:25:37.920
William Cheng: Okay, so our way you create an object in the virtual file system to refer to the I note when x. That's called a be note.

204
00:25:40.920 --> 00:25:44.490
William Cheng: So, so in a way it's the VM FS representation of IOT

205
00:25:50.760 --> 00:25:51.660
William Cheng: So make sense.

206
00:25:53.460 --> 00:25:54.030
Lalit Gupta: Trying to

207
00:25:54.540 --> 00:25:57.510
William Cheng: Wipe it is. Yeah. But now, Jerry, you have an object inside

208
00:25:58.530 --> 00:26:08.070
William Cheng: Is that actual process. And this one is called. I know, right, then this I know his implementation dependent right it's a it's a FST pendant

209
00:26:09.510 --> 00:26:18.030
William Cheng: Okay, so how do we actually have a, you know, sort of keep track of this particular. I know we need an abstraction inside the virtual file system to keep track of this. I know I

210
00:26:19.950 --> 00:26:29.760
William Cheng: Mean in the example that I also give us in the Windows file system. There's no such thing as I know, right. So, so therefore I haha what do you point to inside the virtual classes ME THERE'S NO SUCH THING AS I KNOW.

211
00:26:31.980 --> 00:26:36.390
William Cheng: So you need to have another abstraction at the bfs level. So in weenies. This is called a vino

212
00:26:43.980 --> 00:26:45.090
Lalit Gupta: Oh key.

213
00:26:45.180 --> 00:26:50.640
William Cheng: Why because, because otherwise, all you have is a voice star pointer and you pointed this one. What can you do with a voice. There's nothing you can do.

214
00:26:54.870 --> 00:27:00.360
Lalit Gupta: Okay, so okay so this the node and okay so this vineyard in the VMware side and there is a note in the

215
00:27:01.590 --> 00:27:02.340
Lalit Gupta: FM side.

216
00:27:02.700 --> 00:27:04.050
William Cheng: Yep. Yeah.

217
00:27:04.080 --> 00:27:14.790
Lalit Gupta: Even though, even though they are on the different side they did. What is the boy, what is inside them has to be similar, because we have to like a bind them.

218
00:27:15.270 --> 00:27:15.540
William Cheng: So,

219
00:27:15.990 --> 00:27:22.590
William Cheng: Yeah, so you have to look at the source code, right. So let's take a look at an example over here. So let's go back to

220
00:27:27.840 --> 00:27:29.580
William Cheng: Okay, so, so let's look at the

221
00:27:30.960 --> 00:27:35.970
William Cheng: Colonel include FS vino hai. So let's see what a B. Notice

222
00:27:38.670 --> 00:27:43.440
William Cheng: Okay, so here's the Beano right you know struck me know right here, I can. You can see that right

223
00:27:44.160 --> 00:27:45.060
Lalit Gupta: Yes, yes, yes.

224
00:27:45.330 --> 00:28:02.610
William Cheng: Okay, so as a lot of stuff to keep track off right over here. Remember the I know number. Okay. So right here I n ot it's a Wii and Wii. No. So this is the vino numbers. I mean, we need you know number I know number. I think they're the same thing.

225
00:28:03.870 --> 00:28:17.220
William Cheng: There is the length of the file. It's called the node length and we and I over here. So here is a voicemail pointer, right, the voice point over here in the weakness implementation. They use his voice, not pointed pointed. I know.

226
00:28:19.770 --> 00:28:20.100
William Cheng: Okay.

227
00:28:21.810 --> 00:28:24.390
William Cheng: So let's take a look at the I know, Colonel.

228
00:28:26.190 --> 00:28:29.550
William Cheng: FS ram FS ram FS see

229
00:28:41.070 --> 00:28:44.790
William Cheng: Okay, so here is the I know data structure.

230
00:28:46.770 --> 00:28:52.800
Lalit Gupta: Yeah, so, so this seems very similar to what is in the diagram of the slide. Well,

231
00:28:53.190 --> 00:29:01.380
William Cheng: So. So the basic idea is here is that the the vino is pretty generic right because he needs to support any kind of I know so i. So, therefore, it's a much bigger data structure.

232
00:29:01.980 --> 00:29:07.320
William Cheng: So over here, this is the I know data structure. So in the referral system that I know data structure is very, very simple. Right.

233
00:29:09.420 --> 00:29:21.270
William Cheng: So this one is like a lecture slide where it has the file size, it doesn't really have ownership. It has, you know, device number and all that kind of stuff, and has a self reference. What I know number it belongs to

234
00:29:28.710 --> 00:29:34.470
William Cheng: Yeah. So anyway, so this this. I know it's very, very simple. So if we look at the system file, file system.

235
00:29:40.170 --> 00:29:44.730
William Cheng: Is the I know here and maybe it's in the include file.

236
00:29:54.300 --> 00:29:57.630
William Cheng: Okay, so here's the system, our system I node.

237
00:30:02.550 --> 00:30:05.700
William Cheng: Okay, anyways, so I guess. Yeah, let's not go through it.

238
00:30:09.090 --> 00:30:11.250
William Cheng: Let's take a look at the RAM file system again.

239
00:30:17.670 --> 00:30:22.770
William Cheng: I'm trying to find a way to get back to the V node.

240
00:30:24.090 --> 00:30:31.020
William Cheng: Okay, so all the referral system over here, you can look at all the functions over here coming with a be node. Okay, so you can actually see that

241
00:30:31.560 --> 00:30:40.080
William Cheng: What they will do is that you know there will use the V nodes V and I don't even remember that when we look at the vino at that age to be an AI is the voicemail pointer.

242
00:30:40.470 --> 00:30:51.810
William Cheng: They will take the voice or partner over here typecast to typecast into the RAM PA system. I know the I know data structure. So, therefore, you can convert a vino to I know right by following the pointer.

243
00:30:55.590 --> 00:31:02.160
Lalit Gupta: Okay, so we can convert so we know it is a big data structure in in that there is a field name V V and underscore I

244
00:31:02.640 --> 00:31:04.410
William Cheng: And that's the most affluent area.

245
00:31:05.370 --> 00:31:07.740
Lalit Gupta: And that we can type costs to the RAM, I note.

246
00:31:08.190 --> 00:31:15.240
William Cheng: And only the RAM file system can do that, right, because as a virtual file system. They have no idea what kind of point to it is because it's only a voice.

247
00:31:15.840 --> 00:31:22.560
William Cheng: BUT WHEN YOU GET INSIDE around faster than the referral systems as well. Of course I know, I know what it is, because I'm the one that set up this pointer.

248
00:31:25.770 --> 00:31:28.920
Lalit Gupta: Okay, but if it does not know what is we notice you notice

249
00:31:30.540 --> 00:31:34.710
William Cheng: Of course, all the FS they know exactly how to support the bfs.

250
00:31:35.970 --> 00:31:43.140
William Cheng: Okay, so, so, so when you write the actual file system for individual actual file system. You all have to know exactly what BFS look like

251
00:31:44.790 --> 00:31:47.730
Lalit Gupta: Also, if it is aware of the data structure of renewed

252
00:31:48.030 --> 00:31:48.810
William Cheng: Yes, yeah.

253
00:31:51.960 --> 00:31:54.120
William Cheng: Okay, so you can see all this implementation over here, right.

254
00:31:54.630 --> 00:32:07.050
William Cheng: Okay. Now over here, you know, this is the code for the RAM file system implementation of one of the via the the function. So, it will take the vino data structure. Of course, you know, you have to know exactly what's inside of you know data structure and then you can work on it. Right.

255
00:32:08.220 --> 00:32:15.660
William Cheng: Okay, right. So, this particular function. What it does is that it will actually initialize the vino based on what's inside the I know

256
00:32:16.950 --> 00:32:25.140
William Cheng: I mean, the name of this function is really weird. It's called ram FS read Dino. So I want you to call it initialized vino but somehow they call it the renal function.

257
00:32:25.620 --> 00:32:38.040
William Cheng: So in this function there was set up all the pointers inside of vino and we just saw that the vino data structure is a really big data structure. Okay, so some of the some of the information over some of the information is that'd be no they are initialized inside this function.

258
00:32:39.120 --> 00:32:48.030
William Cheng: So for example, you can see that, you know, vn i is equal to I know, right. So what does. I know, I know it over here is going to be one of the iOS inside the red ball system.

259
00:32:50.220 --> 00:32:50.610
Cool.

260
00:32:53.520 --> 00:32:59.130
William Cheng: Right. So inside is inside this function you you basically connect BFS to the FS

261
00:33:03.660 --> 00:33:06.720
Lalit Gupta: Okay, so this is some function that we have to somehow

262
00:33:10.290 --> 00:33:15.540
Lalit Gupta: That will actually that we have to buy it to like some function in the air in the

263
00:33:16.980 --> 00:33:17.940
Lalit Gupta: Study BFS.

264
00:33:18.210 --> 00:33:18.600
William Cheng: Yes.

265
00:33:18.630 --> 00:33:24.870
Lalit Gupta: In via telephone functions that we have to bind, which is which. I think one of this function, maybe them.

266
00:33:26.100 --> 00:33:28.920
Lalit Gupta: Or maybe I think maybe some career probably

267
00:33:29.550 --> 00:33:35.040
William Cheng: Well, it's some kind of a initialization function, right, because all the pointer needs to be set up when you initialize the data structure.

268
00:33:35.970 --> 00:33:40.860
William Cheng: If your data center 10 pointers law, then we initialize data we initialize that data structure.

269
00:33:41.040 --> 00:33:43.650
William Cheng: You got to set up all the temperatures, right, if you're going to use them.

270
00:33:44.730 --> 00:33:48.840
Lalit Gupta: Also, this is an answer reviewed be noted in is the initialization function.

271
00:33:50.100 --> 00:33:59.010
William Cheng: So, so, this function is used to initialize every vino so so ram FS read V node. So whereas that function defined

272
00:34:00.030 --> 00:34:12.360
William Cheng: Grab read V node Colonel includes start at H. It's in BFS Colonel include FS BFS that he

273
00:34:15.480 --> 00:34:26.130
William Cheng: OK. So again, this is a rail function pointer I the poly multifunction pointer. It's called the file systems operations. So every file system has, you know, this list of operations.

274
00:34:26.880 --> 00:34:37.560
William Cheng: surprisingly small is only a four, four functions over here. One is called Rubino the other ones could to leave, you know, the other one is called Caribbean. Oh, and then the other one is called amount the fastest them.

275
00:34:38.490 --> 00:34:48.540
William Cheng: So over here, it says Rubino will be passive vino blah blah blah and review them must initialize the family members of the provided penal data structure.

276
00:34:50.550 --> 00:34:53.430
William Cheng: Well guys at this function initialize is, you know, part of the Beano

277
00:34:57.750 --> 00:35:04.320
Lalit Gupta: Previously you were talking about ramps read been over to the difference between GDP node and you'd be good.

278
00:35:04.650 --> 00:35:16.020
William Cheng: Well, so, so, so this function over here it's appalling morphic pointer. So this one is called read vino so when you're using the RAM file system, you know, the functionality will point to will be the emphasis then Rubino

279
00:35:16.290 --> 00:35:21.150
William Cheng: When you're using the system boss is them, then this point I will point to the system file, file system Rubino

280
00:35:23.160 --> 00:35:23.700
Lalit Gupta: Okay.

281
00:35:24.420 --> 00:35:37.890
William Cheng: So, you know, anything that says ops. They are typically I guess in Phoenix, they try to name all these function in a consistent way. So whenever you see ops. That means that they are a real function pointers.

282
00:35:41.940 --> 00:35:47.040
William Cheng: Okay, so therefore they're probably more fake, you know, for, you know, for one. So in this case, this is the fastest them right

283
00:35:47.280 --> 00:35:53.790
William Cheng: file system operations. So for one type of analysis, then all these for function will point to, you know, for function is that the red boxes, then

284
00:35:54.060 --> 00:36:01.050
William Cheng: When you're using the system versus them all these four functions, they will point to the forefront, the corresponding for function in this isn't a PA system.

285
00:36:09.840 --> 00:36:10.410
William Cheng: So,

286
00:36:11.250 --> 00:36:12.360
Lalit Gupta: We got the system by

287
00:36:12.480 --> 00:36:17.490
Lalit Gupta: The file system like in Phoenix, like is there something we have to do first.

288
00:36:18.150 --> 00:36:23.700
William Cheng: System for our system is what you need to work, we need to use in chapter in Colonel three

289
00:36:26.040 --> 00:36:26.340
Lalit Gupta: Okay.

290
00:36:26.460 --> 00:36:27.720
Lalit Gupta: Just don't find file system is

291
00:36:27.900 --> 00:36:30.390
Lalit Gupta: Like you fall, Colonel three. Okay.

292
00:36:30.660 --> 00:36:37.080
William Cheng: Yeah, so your kernel to you're going to work with a very simple process and known as the ramp our system, the ram file system is a fake classes them.

293
00:36:37.560 --> 00:36:53.550
William Cheng: Because it's done in RAM right the RAM is a memory. So therefore, there's no disk. Okay, so if your implementation or BFS is perfect. When you switch to Colonel three, you should be able to work with the system pop them directly. And this isn't bosses and has a disconnect

294
00:36:54.690 --> 00:37:00.690
William Cheng: OK. OK, so, so, so if you implement your polymorphous and perfectly, you should, you know, the switch should be very easy.

295
00:37:04.680 --> 00:37:13.410
William Cheng: Never. It's almost impossible to do perfectly. So, so even if we have a perfect corner to where you go to Colonel three at the beginning you gonna have to, you know, fix some of your bugs.

296
00:37:14.040 --> 00:37:20.190
William Cheng: Because now there's a desk. If you have reference counting problem, you know, then then then then things get a little messy.

297
00:37:25.200 --> 00:37:28.260
Lalit Gupta: Also in ramps adored see if you go to

298
00:37:29.790 --> 00:37:32.010
Lalit Gupta: The ramps grammar for start the fire file.

299
00:37:32.580 --> 00:37:33.090
Okay.

300
00:37:35.820 --> 00:37:36.090
Lalit Gupta: So,

301
00:37:37.500 --> 00:37:43.080
Lalit Gupta: There are 2222 struck

302
00:37:46.860 --> 00:37:50.100
Lalit Gupta: Struck get there. To add a function pointer.

303
00:37:51.360 --> 00:37:56.520
William Cheng: Right. So again, this is you know LPS but anything that's ops. That means that the repetition pointers.

304
00:37:57.930 --> 00:38:03.810
Lalit Gupta: Yeah. So what is the difference between this or direct directory VR ops and the file via ups.

305
00:38:04.650 --> 00:38:15.360
William Cheng: Yeah so. So again, they named the variable pretty carefully. Right. So this is the real function pointer for directly file. This is a real function point of for a non directory file or regular file.

306
00:38:17.100 --> 00:38:28.020
Lalit Gupta: And why why why in the directory. We have a lot of we have like create which is not now an empty node which is normal and whereas in the file ops, we have these know is

307
00:38:28.140 --> 00:38:31.320
Lalit Gupta: There are two things about if you think about a directory file.

308
00:38:31.380 --> 00:38:33.810
William Cheng: What kind of operations to the director of our support.

309
00:38:35.820 --> 00:38:42.270
William Cheng: Right. So here is a directory. The, the operation that's supported by directory file right so directly does directly support read

310
00:38:42.720 --> 00:38:48.270
William Cheng: Well over here says no, right, does it directory support. Right. No, you're not allowed to write to a directory file.

311
00:38:49.230 --> 00:38:58.620
William Cheng: You are not allowed to memory map a direct file into your address space. Are you allowed to create a file inside a directory what the answer is yes. Right. And here is the function that would do it.

312
00:39:00.750 --> 00:39:01.620
Lalit Gupta: Oh, yeah.

313
00:39:02.190 --> 00:39:07.080
William Cheng: Does it directly allow you to create a device. The answer is yes, because here's a function that allow you to do a

314
00:39:07.380 --> 00:39:17.400
William Cheng: Does it directly an artist performer look up on June directory allow you to perform a link function that directory is support on link that directory support making a directory subdirectory

315
00:39:18.180 --> 00:39:23.790
William Cheng: The directory support remove a directory and then over here, you can read a directory entry or as a reader over here.

316
00:39:23.940 --> 00:39:32.220
William Cheng: Is how you read it directly entry. So you can actually call reader over and over and over again. And every time you call it, you will you will, you know, it will return the next directory entry.

317
00:39:33.000 --> 00:39:39.420
William Cheng: The directory also support the function call stat. Yeah. And we also don't know what the other ones are doing

318
00:39:39.900 --> 00:39:42.780
William Cheng: If you look at a file right so here's a, here's a regular file.

319
00:39:43.140 --> 00:39:55.140
William Cheng: A regular fall. What can you do with it what you can read from it. Right. You can read from it. You can write to it. You can also get the status of the file, but you cannot create something inside the file which kind of makes sense, right,

320
00:39:57.180 --> 00:39:59.010
William Cheng: And you can look you can link economically.

321
00:40:00.960 --> 00:40:01.410
William Cheng: Okay, so

322
00:40:03.360 --> 00:40:11.610
William Cheng: One more point here, right. These array of function pointers. They have exactly the same entry in the first one. The second one, because everything is a file.

323
00:40:11.850 --> 00:40:18.450
William Cheng: Or directory is a file a regular file the file. So therefore, they have to, you know, the we have an array of a function ponder that look like this.

324
00:40:19.800 --> 00:40:23.460
Lalit Gupta: Oh, so that we can we can type constant is like in the same manner.

325
00:40:23.550 --> 00:40:31.800
William Cheng: We use the same, same, same, a real function pointer to to to to to access function in the file or access function or directory far

326
00:40:32.460 --> 00:40:33.000
Lalit Gupta: Okay.

327
00:40:33.270 --> 00:40:42.570
William Cheng: So, so that's why before we use that function, we have to check whether it's equal to know are not visible to Know when we shouldn't call it because otherwise you're going to get a segmentation fault. Yeah.

328
00:40:46.800 --> 00:40:52.290
Lalit Gupta: And violated got before the orders, all these things. The reason

329
00:40:53.640 --> 00:40:54.450
William Cheng: I'm

330
00:40:55.920 --> 00:40:56.400
William Cheng: I

331
00:40:57.630 --> 00:40:58.080
William Cheng: Don't know.

332
00:40:59.250 --> 00:41:02.040
William Cheng: I think this is a event C programming.

333
00:41:03.060 --> 00:41:05.550
William Cheng: So I don't know what, what was the meaning of that.

334
00:41:09.150 --> 00:41:11.880
Lalit Gupta: Do we have to worry about this and map for control to

335
00:41:13.200 --> 00:41:13.740
William Cheng: Have to worry about.

336
00:41:14.820 --> 00:41:15.090
Lalit Gupta: And

337
00:41:20.220 --> 00:41:21.840
William Cheng: I think the only um

338
00:41:23.220 --> 00:41:28.320
William Cheng: Let's see a graph mm AP Colonel include started. Ah.

339
00:41:32.100 --> 00:41:36.270
William Cheng: Ok so MF is a system called we don't have to worry about system call until Colonel three

340
00:41:37.950 --> 00:41:42.720
William Cheng: So in Colonel three, you have to implement the system call

341
00:41:44.640 --> 00:41:45.810
William Cheng: I yeah

342
00:41:47.250 --> 00:41:52.110
William Cheng: And I guess the example that we just saw over here right for the directory file.

343
00:41:53.190 --> 00:41:56.640
William Cheng: A map is empty and also for regular father and map is also empty.

344
00:41:56.670 --> 00:42:01.110
William Cheng: As yeah maybe we don't so clearly for Colonel to we don't have to worry about a map.

345
00:42:02.070 --> 00:42:16.980
William Cheng: Okay, so, so, so, so for this fall system and map is not implemented. Right. And also we can also see that there's a lot of other other function over here, acquire release. I've no idea what they are. And then, you know, all these five functions over here. Again, nobody's implementing that

346
00:42:20.040 --> 00:42:26.490
William Cheng: Hey, if you look at the rest of the for the rest of the file is going to be the implementation of all these functions, right, create make know the, you know, look up

347
00:42:27.900 --> 00:42:33.090
William Cheng: So this is a, you know, detailed implementation inside the RAM file system for all these functions.

348
00:42:35.520 --> 00:42:44.910
Lalit Gupta: Also, is it would you recommend or is it is it even possible to go through line by line for ramp Stuart, Stuart. See like

349
00:42:45.030 --> 00:42:48.630
William Cheng: That's what you're that's what you're supposed to do. That's part of the call to assignment.

350
00:42:48.810 --> 00:42:51.300
William Cheng: Is to reach to the call and try to figure out what's going on.

351
00:42:53.430 --> 00:42:53.910
Lalit Gupta: Okay.

352
00:42:55.290 --> 00:43:04.110
Lalit Gupta: So we so we can under so if you try. We can under like it so it so we can understand what exactly every function is doing and

353
00:43:04.440 --> 00:43:09.210
William Cheng: What you're supposed to single step into the RAM file system code to devalue or Colonel to co

354
00:43:10.230 --> 00:43:10.770
Lalit Gupta: Op

355
00:43:10.980 --> 00:43:21.210
William Cheng: We can learn to code is not working. He said, How come it's not working well let's go into the referral system to see how it's the what's going on. And then, and then you should realize that you did something wrong in your kernel to code.

356
00:43:22.320 --> 00:43:22.830
Okay.

357
00:43:25.440 --> 00:43:32.910
William Cheng: So, so my week a discussion section I'm gonna you know I'm basically, I'm going to go through some of the stuff we just talked about. Okay.

358
00:43:38.250 --> 00:43:47.850
Lalit Gupta: So yeah, so right now I like yeah I'm trying to make a picture of it and looks like the whole thing. So yeah, I've currently I have this much question.

359
00:43:48.360 --> 00:43:56.520
William Cheng: Yeah so. So the way I sort of think about writing these kernel to code is that you get enough to get going, right, and then you try to sort of figure out what's wrong with your code.

360
00:43:57.690 --> 00:44:05.760
William Cheng: Okay. I mean, some people I know their programming style is that they need to understand everything perfectly before they start writing code, so that will be a little difficult.

361
00:44:09.990 --> 00:44:14.940
William Cheng: Right, because that means that you have to understand every line of code and the figure out which function that they should call

362
00:44:15.600 --> 00:44:22.740
William Cheng: But I think at this point, you have a pretty good guess right if you're in the if you try to implement it make notes make notes function in the VM Fs.

363
00:44:23.040 --> 00:44:29.820
William Cheng: What is your reach the emphasis that make know so so you don't really have to understand the entire emphasis them to just to see how that will work.

364
00:44:31.710 --> 00:44:32.310
Okay.

365
00:44:36.630 --> 00:44:37.170
Lalit Gupta: So,

366
00:44:38.940 --> 00:44:43.530
Lalit Gupta: So we can start him start writing the function even before reading for 30

367
00:44:44.850 --> 00:44:52.740
William Cheng: Yeah, you just trust that it works perfectly right. So, so read the comment below in the comment box says find the pointer. Why is that, well, we'll, we'll find what pointer.

368
00:44:53.280 --> 00:45:02.580
William Cheng: You look at the header file, you look at the source code and then you try to sort of make an educated guess to see what you know what point do you use and then make sure the pointer is not equal to know and then you call it.

369
00:45:04.050 --> 00:45:11.640
William Cheng: Okay. And if it doesn't work. You say, Why doesn't it work. Did I pass the wrong arguments to, you know, if you pass no pointer. Of course it doesn't work is going to crash.

370
00:45:12.270 --> 00:45:20.100
William Cheng: The round classes. And there's also a bunch of K assert right maybe should take it a little bit closer. Because if the assertion fail. Well, that means that you're passing the wrong argument.

371
00:45:21.570 --> 00:45:21.960
Okay.

372
00:45:31.500 --> 00:45:32.760
Lalit Gupta: Okay, I'll work on

373
00:45:33.540 --> 00:45:39.630
William Cheng: Yeah, so, so I mean this is you know where this is kind of pretty realistic when you go work for a company and they an internship or something like that.

374
00:45:39.810 --> 00:45:49.170
William Cheng: So well, they will do this a while. I want you to implement this new thing, but they all these existing co read existing code to find out what kind of code that you have to write. And then, you know, if you have questions, you can go bug these engineers.

375
00:45:49.560 --> 00:45:52.440
William Cheng: Okay but but don't pass the engineer to tell you every line of code, what you

376
00:45:52.470 --> 00:45:54.480
William Cheng: Need to write right so it's kind of like our assignment.

377
00:45:55.650 --> 00:45:55.950
William Cheng: Yes.

378
00:46:01.170 --> 00:46:02.610
Lalit Gupta: Yeah, the assignment. Interesting.

379
00:46:03.990 --> 00:46:07.140
William Cheng: Yeah. So I think these Brown University assignment. They're very clever.

380
00:46:08.430 --> 00:46:09.360
William Cheng: So there's a pretty good.

381
00:46:11.400 --> 00:46:15.780
Lalit Gupta: Like is there, like, at some point, I think you told like there are like

382
00:46:16.920 --> 00:46:20.070
Lalit Gupta: Two assignment which are provided by the ground in your city.

383
00:46:20.640 --> 00:46:25.740
William Cheng: Yeah drivers and this is the boss is them, they are written there. Yeah, they're written by the Brown University people

384
00:46:27.180 --> 00:46:35.250
Lalit Gupta: Are they able to do it in one semester because it seems like five like five, the fire assignment is like too much for one semester.

385
00:46:36.780 --> 00:46:46.830
William Cheng: Well, the, you know, if you look at the system file system. You know, it's actually pretty simple. Right. And also, if you look at chapter six. You know, we finished this Infosys them.

386
00:46:47.880 --> 00:46:51.960
William Cheng: In very short period of time since the proposal is very, very straightforward.

387
00:46:53.490 --> 00:46:55.950
William Cheng: So it is possible. So I think, you know,

388
00:46:57.390 --> 00:47:02.940
William Cheng: If I say not yet implemented, Colonel.

389
00:47:05.310 --> 00:47:10.170
William Cheng: Start see and then grab two says five Fs.

390
00:47:11.730 --> 00:47:16.260
William Cheng: There's nothing there. So what about one more level deep

391
00:47:17.880 --> 00:47:21.480
William Cheng: Okay, so to implement the system purposes. And there's only two files yet to implement

392
00:47:23.040 --> 00:47:34.980
William Cheng: Right, there's only two files one product here one part here. There's a bunch of functions over here, right. These are the function. I mean, these are the function that we just mentioned, right, instead of seeing grandpa system, you have to implement the corresponding system PA system.

393
00:47:36.300 --> 00:47:43.260
William Cheng: But in this case, there's a disk. So therefore you need to know how to interact with the disk and you need to put your kernel threat to sleep, you know, other kind of stuff.

394
00:47:47.880 --> 00:47:50.550
William Cheng: So I think in Brown University.

395
00:47:53.220 --> 00:48:06.570
William Cheng: I guess there's there's one way to take the operating system class is to actually take it as two classes. One is just the lecture and the other one is the last class. I think it's only half the credit of the regular class. And that's where they implement all the five programming assignments.

396
00:48:08.490 --> 00:48:09.090
Lalit Gupta: Oh,

397
00:48:09.960 --> 00:48:12.870
William Cheng: I think they actually divided them into, you know, two different classes.

398
00:48:19.380 --> 00:48:21.930
William Cheng: But anyways, a USC, you know, since

399
00:48:23.100 --> 00:48:32.400
William Cheng: You know, and also you Brian diversity. These are the costs are for undergrad students a USC CSV to can only be taken by grad students. So again, it's a little different.

400
00:48:39.660 --> 00:48:41.430
William Cheng: Classes more intense because

401
00:48:41.580 --> 00:48:45.000
William Cheng: Yes, and people in Glasgow, they should, they should put more into it.

402
00:48:48.060 --> 00:48:58.170
Lalit Gupta: I like I'm just curious like a little harder to figure the all the assignment because the thing is like without your health, like, pretty much. It is impossible to like figure out what to do.

403
00:48:58.620 --> 00:49:05.340
William Cheng: Yeah, so like, like I said at the beginning, you know, in the, you know, in 2012 when I got when I started this class.

404
00:49:06.060 --> 00:49:19.170
William Cheng: You know, at that time, the book was newly published the assignment was the idea that the although the wiki style was brand new to the world. So, at that time, the only documentation people here was the winnings documentation.

405
00:49:21.030 --> 00:49:33.990
William Cheng: Okay, so at that time. The problem is we're almost impossible to do, right. So over the years. You know, I am more stuff to the programming FAQ, you know, so, so now I think. I mean, you can see how long the problem FAQ is I

406
00:49:35.580 --> 00:49:36.120
William Cheng: Get this done.

407
00:49:38.820 --> 00:49:44.580
Lalit Gupta: Yeah. And the thing is, Dr. Yeah, I think it was like, I think, has a lot of interesting information like

408
00:49:46.200 --> 00:49:46.410
Lalit Gupta: Your

409
00:49:47.520 --> 00:49:50.820
William Cheng: Yeah, so without those information the kernels Sam is pretty much impossible, right.

410
00:49:51.360 --> 00:49:51.870
Yeah.

411
00:49:56.430 --> 00:50:00.270
William Cheng: So, so, by, by the way, you know, you could compare our class with a Brown University class right

412
00:50:01.050 --> 00:50:16.620
William Cheng: Since the class is taken by undergrads. The undergrads actually taking a lot of classes, you know, per semester I say a USC typically for gratitude and they're taking, you know, two classes per semester. So, therefore, you know, for classes to be more intense, which is which is also expected

413
00:50:19.020 --> 00:50:19.290
Of

414
00:50:21.090 --> 00:50:25.290
Lalit Gupta: What I think is goddess Java classes are also like pretty like this.

415
00:50:26.880 --> 00:50:29.730
Lalit Gupta: Slide is if you go from the undergrad classes.

416
00:50:29.790 --> 00:50:32.340
William Cheng: Yeah, supposedly yeah more work.

417
00:50:33.240 --> 00:50:37.020
Lalit Gupta: Yes, or her work and assignment or also make more difficult.

418
00:50:37.500 --> 00:50:37.830
Yeah.

419
00:50:39.570 --> 00:50:45.420
William Cheng: Yeah. So I think, you know, in the end for CSP to that people are willing to spend a lot of time on it in the end. Everything is doable.

420
00:50:45.990 --> 00:51:01.050
William Cheng: Right, given the colonel FAQ and stuff I and given that I'm here to help. Right. People should ask me question. You can see that in the class Google group. There's not enough discussion, which is kind of surprising, but you know I do expect people to have more discussion, the classical but

421
00:51:04.920 --> 00:51:07.170
William Cheng: Anyways. Is that happening. I can. There's nothing I can do.

422
00:51:11.100 --> 00:51:13.950
Lalit Gupta: Focus. I will try to look into, like,

423
00:51:15.120 --> 00:51:17.550
William Cheng: A semi question if you, you know, anything's not clear.

424
00:51:18.390 --> 00:51:19.260
Lalit Gupta: Yes, thank you.

425
00:51:19.920 --> 00:51:23.400
William Cheng: Okay, so I'm gonna I'm it says you're the only person in class, right.

426
00:51:23.640 --> 00:51:24.780
William Cheng: I'm just gonna end the

427
00:51:25.050 --> 00:51:28.650
William Cheng: End the classes. Alright, see ya. Bye bye.