03 05 Processes Part 5 — Transcript

Full transcript

1. 0:08[Music]
2. 0:20next would be synchronization
3. 0:23okay
4. 0:24so synchronization either the sending or
5. 0:26receiving of messages or neither or both
6. 0:30me either blocking or not blocking
7. 0:34okay
8. 0:37next so
9. 0:39the producer consumer message passing
10. 0:42okay
11. 0:43so what do we have here the producer
12. 0:46produce an item
13. 0:48in the next produced
14. 0:50and then send next produced on the
15. 0:52consumer side
16. 0:54okay so while true same thing with that
17. 0:56of the producer receive
18. 0:59next consumed
19. 1:00okay
20. 1:02and then of course they consume the item
21. 1:04in the next consume
22. 1:07okay so that's the implementation of
23. 1:09pesos passing
24. 1:11on
25. 1:13that problem
26. 1:15okay
27. 1:16so next would be buffering
28. 1:18okay so queue up messages attached to
29. 1:20the link
30. 1:21implemented in one of three ways so
31. 1:23you've got the zero
32. 1:24capacity bounded capacity
33. 1:27and unbounded capacity
34. 1:31okay so with a zero capacity messages
35. 1:34cannot be stored in the queue or senders
36. 1:37must block until receivers accept the
37. 1:39message
38. 1:41okay
39. 1:42or the bounded capacity there is a
40. 1:44certain predetermined
41. 1:46infinite capacity in the queue
42. 1:50senders must block if the queue is full
43. 1:52until the space becomes available in the
44. 1:55in the queue
45. 1:57but maybe either blocking or
46. 1:58non-blocking otherwise
47. 2:01okay so unbounded capacity the queue has
48. 2:04a theoretical infinite capacity so
49. 2:06senders can never force to block
50. 2:10okay
51. 2:14next
52. 2:15so
53. 2:16we have here an example of inter process
54. 2:18communications in unix okay
55. 2:21so the posix
56. 2:24so for the implementation of the shared
57. 2:26memory the the process first creates a
58. 2:29shared memory segment
59. 2:30also used to open an existing segment
60. 2:33and then set the size of the object okay
61. 2:37so we use
62. 2:38uh mnap
63. 2:40to memory map okay so a file pointer to
64. 2:43a shared memory object so reading and
65. 2:45writing to shared memory is done
66. 2:48by using the pointer returned memory map
67. 2:50or map
68. 2:52okay
69. 2:53so this is the
70. 2:56complete program implementing the shared
71. 2:58memory and opposing systems
72. 3:02okay
73. 3:06this is the continuation
74. 3:08earlier this is the posix implementation
75. 3:11on the producer
76. 3:12and the posix implementation on the
77. 3:14consumer
78. 3:16okay
79. 3:17now how about on other operating systems
80. 3:19like mac
81. 3:21okay
82. 3:23so mac communication is message based
83. 3:27so even system calls are messages
84. 3:31each task gets two ports at creation so
85. 3:35get kernel and notify
86. 3:37okay
87. 3:38so the kernel sends notification of
88. 3:40events to the notify mail box
89. 3:43three system calls are used for message
90. 3:45transfer you've got message send
91. 3:48sends a message to a mailbox
92. 3:50message received receives a message
93. 3:53and
94. 3:54message rpc
95. 3:56sends a message and waits for exactly
96. 3:58one message and response from the sender
97. 4:01so the port allocate
98. 4:03creates a new mailbox an associated
99. 4:06queue for holding messages so the
100. 4:08default size is eight
101. 4:11all right
102. 4:15next
103. 4:17only one task at a time can own or
104. 4:20receive message from any given mailbox
105. 4:23but this are transferable so messages
106. 4:27from the same sender
107. 4:30to the same receiver are guaranteed to
108. 4:32arrive in the first in first off order
109. 4:35but no guarantees are made regarding
110. 4:38messages from multiple senders
111. 4:40okay
112. 4:41so messages consists of a fixed length
113. 4:45header followed by a variable length
114. 4:47data
115. 4:48the header contains mailbox number
116. 4:51okay
117. 4:52or just the address of the receiver and
118. 4:55the sender
119. 4:56so the section or data section consists
120. 4:58of a list of type data items
121. 5:01it's containing a type
122. 5:04size and value
123. 5:06okay
124. 5:08now here's the implementation of message
125. 5:11passing on the client okay client code
126. 5:14on mac
127. 5:16okay so if the receiver's mailbox is
128. 5:18full the sender has four choices
129. 5:20okay
130. 5:21so wait indefinitely until there is a
131. 5:23room
132. 5:24in the mailbox
133. 5:26wait at most 10 milliseconds
134. 5:29okay so or do not wait at all
135. 5:32okay
136. 5:33so temporarily
137. 5:35cache the message with the kernel for
138. 5:37delivery when the mailbox becomes
139. 5:39available
140. 5:40so only one such message can be pending
141. 5:42at any given time
142. 5:44from any given sender to any given
143. 5:46receiver so normally
144. 5:49only used by certain system tasks such
145. 5:52as the prince pooler
146. 5:53which must notify the client
147. 5:56for the completion of their job
148. 5:58but cannot wait around for the mailbox
149. 6:01to become available
150. 6:03okay so we have a lot of technical terms
151. 6:07and technical discussions here
152. 6:09okay
153. 6:11next
154. 6:13so receive calls must specify the
155. 6:16mailbox
156. 6:17or mailbox set from which they wish to
157. 6:20receive message
158. 6:22okay
159. 6:23so port status
160. 6:25okay so the purge status reports the
161. 6:28number of messages waiting at any given
162. 6:30mailbox so if there are no messages
163. 6:32available in the mailbox so we're using
164. 6:34sept
165. 6:35the receiver can either block for n
166. 6:38milliseconds
167. 6:39or not block at all so in order to avoid
168. 6:41delays caused by capping messages
169. 6:44multiple times
170. 6:45okay mac remaps the memory space
171. 6:49for the message from the sender address
172. 6:51space to the receiver's address space
173. 6:54so using virtual memory techniques
174. 6:56to be covered
175. 6:58on the next chapter that we have
176. 7:00okay so it does not actually move the
177. 7:02message
178. 7:04anywhere at all so when the sending and
179. 7:06receiving tasks are both on the same
180. 7:09computer
181. 7:10okay
182. 7:12so that's the implementation for mac how
183. 7:14about implementation for windows
184. 7:17okay
185. 7:19so for windows
186. 7:20message passing centric via advanced
187. 7:23local procedures call or the lpc
188. 7:25facility
189. 7:26okay it only works between processes on
190. 7:30the same system
191. 7:34okay
192. 7:35so
193. 7:35it uses ports
194. 7:37like mailbox
195. 7:39supports here
196. 7:41pertains to the socket okay socket or
197. 7:45communication socket okay so in other
198. 7:48operating systems that we discussed they
199. 7:50call it mailboxes to establish and
200. 7:52maintain communication channels
201. 7:55all right so communication works as
202. 7:57follows so the client open a handle to
203. 8:00subsystems connection port object
204. 8:02the client sends a connection request
205. 8:04the server creates two private
206. 8:07communication ports and returns the
207. 8:08handle to one of the client
208. 8:12so the client and server use a
209. 8:15corresponding port handle to send
210. 8:17message or callbacks and to listen for
211. 8:20replies
212. 8:22okay
213. 8:23now in here
214. 8:25we have the local procedure calls in
215. 8:27windows
216. 8:28okay
217. 8:29so if you will observe here
218. 8:31it establishes connections from the
219. 8:33client going to the server
220. 8:36okay so and it uses
221. 8:39a different procedure calls to do the
222. 8:41tasks
223. 8:43all right
224. 8:54[Music]
225. 9:01you