Operating Systems 2021F: Tutorial 8 - Revision history

Soma: /* 3000pc-rendezvous-timeout.c */

2022-01-25T19:14:34Z

3000pc-rendezvous-timeout.c

← Older revision		Revision as of 19:14, 25 January 2022
Line 1,098:		Line 1,098:
	return -1;		return -1;
	}		}
	</~~source~~>		</syntaxhighlight>

	===Makefile===		===Makefile===

	<~~source~~ lang="makefile" line>		<syntaxhighlight lang="makefile" line>
	CC = gcc		CC = gcc

Soma at 05:36, 23 November 2021

2021-11-23T05:36:53Z

← Older revision		Revision as of 05:36, 23 November 2021
Line 1:		Line 1:
	~~'''This tutorial is still being developed.'''~~

	In this tutorial you'll be learning about two implementations of the [https://en.wikipedia.org/wiki/Producer%E2%80%93consumer_problem producer-consumer problem], a classic example of a concurrency problem. The [http://pages.cs.wisc.edu/~remzi/OSTEP/ class textbook] covers concurrency in great detail in Chapters 25-34, and the producer-consumer problem is covered in [http://pages.cs.wisc.edu/~remzi/OSTEP/threads-cv.pdf Chapter 30 (Condition Variables)] and [http://pages.cs.wisc.edu/~remzi/OSTEP/threads-sema.pdf Chapter 31 (Semaphores)]. While you should look at this part of the textbook, note that we will not be covering this material in the same level of detail, as should be clear from this tutorial.		In this tutorial you'll be learning about two implementations of the [https://en.wikipedia.org/wiki/Producer%E2%80%93consumer_problem producer-consumer problem], a classic example of a concurrency problem. The [http://pages.cs.wisc.edu/~remzi/OSTEP/ class textbook] covers concurrency in great detail in Chapters 25-34, and the producer-consumer problem is covered in [http://pages.cs.wisc.edu/~remzi/OSTEP/threads-cv.pdf Chapter 30 (Condition Variables)] and [http://pages.cs.wisc.edu/~remzi/OSTEP/threads-sema.pdf Chapter 31 (Semaphores)]. While you should look at this part of the textbook, note that we will not be covering this material in the same level of detail, as should be clear from this tutorial.

Soma: /* A: Getting Started */

2021-11-23T05:35:45Z

A: Getting Started

← Older revision		Revision as of 05:35, 23 November 2021
Line 13:		Line 13:
	# Run both programs with the same arguments of your choice. Do this a few times. Do they behave the same way? Do you notice any differences?		# Run both programs with the same arguments of your choice. Do this a few times. Do they behave the same way? Do you notice any differences?
	# Repeat the above experiment, this time running each program under strace (with the -f flag to trace children too). Do you notice any difference in the system calls each program makes?		# Repeat the above experiment, this time running each program under strace (with the -f flag to trace children too). Do you notice any difference in the system calls each program makes?
	# In the next question, you will be playing around with <tt>trace</tt> from bcc-tools. ~~For now, you can think of~~ <tt>trace</tt> ~~as a useful alternative~~ to ~~programs like~~ <tt>~~gdb~~</tt>. ~~We will be covering exactly how it works in an upcoming tutorial. In order to use <tt>trace</tt>~~, it needs to be run with root privileges. ~~To get trace working with~~ sudo~~, we need to modify the <tt>secure_path</tt> setting in <tt>/etc/sudoers</tt>. To do this, run <tt>sudo visudo</tt> and append <tt>:/usr/share/bcc/tools</tt> to~~ the ~~<tt>secure_path</tt> string on line 12.~~		# In the next question, you will be playing around with <tt>trace</tt> from bcc-tools. <tt>trace</tt> is very similar to <tt>bpftrace</tt>. As any tool based on eBPF, it needs to be run '''with root privileges'''. (sudo should work on the class VM)
	# Here are some example <tt>trace</tt> commands to play with. You can learn more about <tt>trace</tt> from its man page. Note these assume you unpacked the 3000pc.zip in the Documents folder. ~~'''Make sure to run these as root. Follow the instructions in the previous question to get it working with sudo.'''~~		# Here are some example <tt>trace</tt> commands to play with. You can learn more about <tt>trace</tt> from its man page. Note these assume you unpacked the 3000pc.zip in the Documents folder.

	Report on the string passed to pick_word() on its return:		Report on the string passed to pick_word() on its return:

Soma at 05:32, 23 November 2021

2021-11-23T05:32:39Z

← Older revision		Revision as of 05:32, 23 November 2021
Line 86:		Line 86:
	===3000pc-fifo.c===		===3000pc-fifo.c===

	<~~source~~ lang="c" line>		<syntaxhighlight lang="c" line>
	/* 3000pc-fifo.c Simple producer-consumer with a fifo		/* 3000pc-fifo.c Simple producer-consumer with a fifo
	* Original Version Copyright (C) 2017 Anil Somayaji		* Original Version Copyright (C) 2017 Anil Somayaji
Line 300:		Line 300:
	return -1;		return -1;
	}		}
	</~~source~~>		</syntaxhighlight>

	===3000pc-rendezvous.c===		===3000pc-rendezvous.c===

	<~~source~~ lang="c" line>		<syntaxhighlight lang="c" line>
	/* 3000pc-rendezvous.c More complex producer-consumer using mmap shared memory and pthread_cond_wait		/* 3000pc-rendezvous.c More complex producer-consumer using mmap shared memory and pthread_cond_wait
	* Original Version Copyright (C) 2017 Anil Somayaji		* Original Version Copyright (C) 2017 Anil Somayaji
Line 668:		Line 668:
	return -1;		return -1;
	}		}
	</~~source~~>		</syntaxhighlight>

	===3000pc-rendezvous-timeout.c===		===3000pc-rendezvous-timeout.c===

	<~~source~~ lang="c" line>		<syntaxhighlight lang="c" line>
	/* 3000pc-rendezvous-timeout.c More complex producer-consumer using		/* 3000pc-rendezvous-timeout.c More complex producer-consumer using
	mmap shared memory and pthread_cond_wait		mmap shared memory and pthread_cond_wait
Line 1,123:		Line 1,123:
	@rm -rf *.o		@rm -rf *.o
	@rm -rf $(EXEC)		@rm -rf $(EXEC)
	</~~source~~>		</syntaxhighlight>

Soma: /* A: Getting Started */

2021-11-23T05:30:59Z

A: Getting Started

← Older revision		Revision as of 05:30, 23 November 2021
Line 29:		Line 29:

	trace -p 6312 'p:pthread:sem_wait(int sem) "entry: %d", sem' 'r:pthread:sem_wait(int sem) "exit: %d", sem'		trace -p 6312 'p:pthread:sem_wait(int sem) "entry: %d", sem' 'r:pthread:sem_wait(int sem) "exit: %d", sem'





	If you want to automatically trace the producer process, use		If you want to automatically trace the producer process, use

	trace -p `pidof -s 3000pc-fifo` ...		trace -p `pidof -s 3000pc-fifo` ...

	To trace the consumer, use		To trace the consumer, use

Soma: /* A: Getting Started */

2021-11-23T05:30:38Z

A: Getting Started

← Older revision		Revision as of 05:30, 23 November 2021
Line 30:		Line 30:
	trace -p 6312 'p:pthread:sem_wait(int sem) "entry: %d", sem' 'r:pthread:sem_wait(int sem) "exit: %d", sem'		trace -p 6312 'p:pthread:sem_wait(int sem) "entry: %d", sem' 'r:pthread:sem_wait(int sem) "exit: %d", sem'

	If you want to automatically trace the ~~consumer~~ process (the ~~child process)~~, ~~run this after running 3000fifo:~~

			If you want to automatically trace the producer process, use

			trace -p `pidof -s 3000pc-fifo` ...

			To trace the consumer, use

	trace -p `pidof 3000pc-fifo \| awk '{print $2}'` ~~'r:pthread:read "%s", arg2'~~		trace -p `pidof 3000pc-fifo \| awk '{print $2}'` ...

	==B: Producer/Consumer with Pipes==		==B: Producer/Consumer with Pipes==

Soma: /* A: Getting Started */

2021-11-23T05:26:24Z

A: Getting Started

← Older revision		Revision as of 05:26, 23 November 2021
Line 25:		Line 25:

	(Normally read and write are in the main C library and so would be denoted by r:c:read; however we are linking against the pthread library as well and it replaces the read implementation.)		(Normally read and write are in the main C library and so would be denoted by r:c:read; however we are linking against the pthread library as well and it replaces the read implementation.)

	Print calls to sem_wait in process 6312 and the value of the semaphore at function entry '''and''' return.		Print calls to sem_wait in process 6312 and the value of the semaphore at function entry '''and''' return.

	trace -p 6312 'p:pthread:sem_wait(int sem) "entry: %d", sem' 'r:pthread:sem_wait(int sem) "exit: %d", sem'		trace -p 6312 'p:pthread:sem_wait(int sem) "entry: %d", sem' 'r:pthread:sem_wait(int sem) "exit: %d", sem'

	If you want to automatically trace the consumer process (the child process), run this after running 3000fifo:		If you want to automatically trace the consumer process (the child process), run this after running 3000fifo:

			trace -p `pidof 3000pc-fifo \| awk '{print $2}'` 'r:pthread:read "%s", arg2'

	~~trace -p `pidof 3000pc-fifo \| awk '{print $2}'` 'r:pthread:read "%s", arg2'~~

	==B: Producer/Consumer with Pipes==		==B: Producer/Consumer with Pipes==

Soma at 05:25, 23 November 2021

2021-11-23T05:25:51Z

← Older revision		Revision as of 05:25, 23 November 2021
Line 25:		Line 25:

	(Normally read and write are in the main C library and so would be denoted by r:c:read; however we are linking against the pthread library as well and it replaces the read implementation.)		(Normally read and write are in the main C library and so would be denoted by r:c:read; however we are linking against the pthread library as well and it replaces the read implementation.)

	Print calls to sem_wait in process 6312 and the value of the semaphore at function entry '''and''' return.		Print calls to sem_wait in process 6312 and the value of the semaphore at function entry '''and''' return.

	trace 'p:pthread:sem_wait(int sem) "entry: %d", sem' 'r:pthread:sem_wait(int sem) "exit: %d", sem'		trace -p 6312 'p:pthread:sem_wait(int sem) "entry: %d", sem' 'r:pthread:sem_wait(int sem) "exit: %d", sem'

			If you want to automatically trace the consumer process (the child process), run this after running 3000fifo:

			trace -p `pidof 3000pc-fifo \| awk '{print $2}'` 'r:pthread:read "%s", arg2'

	==B: Producer/Consumer with Pipes==		==B: Producer/Consumer with Pipes==

Soma: /* C: Producer/Consumer with Shared Memory */

2021-11-23T05:13:02Z

C: Producer/Consumer with Shared Memory

← Older revision		Revision as of 05:13, 23 November 2021
Line 64:		Line 64:
	<li>Change the calls to pthread_mutexattr_setpshared and pthread_condattr_setpshared to take 0 instead of 1. How does the behavior or 3000pc-rendezvous change? Does anything break? Demonstrate this using your answers to question 2 and question 3.</li>		<li>Change the calls to pthread_mutexattr_setpshared and pthread_condattr_setpshared to take 0 instead of 1. How does the behavior or 3000pc-rendezvous change? Does anything break? Demonstrate this using your answers to question 2 and question 3.</li>
	<li>3000pc-rendezvous can deadlock when run for long enough. 3000pc-rendezvous-timeout should not deadlock, but it may pause sometimes. What is the difference between these two programs? Why would this effect the tendency to deadlock?		<li>3000pc-rendezvous can deadlock when run for long enough. 3000pc-rendezvous-timeout should not deadlock, but it may pause sometimes. What is the difference between these two programs? Why would this effect the tendency to deadlock?
	<li>Benchmark all three programs using the time command. For example, run 3000fifo as follows:<br><br>		<li>Benchmark all three programs using the time command. For example, run 3000fifo as follows:<br /><br />
	<tt>  time (./3000pc-fifo 100000 0 0 > fifo-output.log)</tt><br><br>		<tt>  time (./3000pc-fifo 100000 0 0 > fifo-output.log)</tt><br /><br />
	Which one is fastest? Why?		Which one is fastest? Why?
	<li>(Advanced) Implement multiple producers and consumers in 3000pc-rendezvous. Compare your results to what you did in 3000pc-fifo.</li>		<li>(Advanced) Implement multiple producers and consumers in 3000pc-rendezvous. Compare your results to what you did in 3000pc-fifo.</li>

Soma: /* C: Producer/Consumer with Shared Memory */

2021-11-23T05:12:44Z

C: Producer/Consumer with Shared Memory

← Older revision		Revision as of 05:12, 23 November 2021
Line 64:		Line 64:
	<li>Change the calls to pthread_mutexattr_setpshared and pthread_condattr_setpshared to take 0 instead of 1. How does the behavior or 3000pc-rendezvous change? Does anything break? Demonstrate this using your answers to question 2 and question 3.</li>		<li>Change the calls to pthread_mutexattr_setpshared and pthread_condattr_setpshared to take 0 instead of 1. How does the behavior or 3000pc-rendezvous change? Does anything break? Demonstrate this using your answers to question 2 and question 3.</li>
	<li>3000pc-rendezvous can deadlock when run for long enough. 3000pc-rendezvous-timeout should not deadlock, but it may pause sometimes. What is the difference between these two programs? Why would this effect the tendency to deadlock?		<li>3000pc-rendezvous can deadlock when run for long enough. 3000pc-rendezvous-timeout should not deadlock, but it may pause sometimes. What is the difference between these two programs? Why would this effect the tendency to deadlock?
	<li>Benchmark all three programs using the time command. For example, run 3000fifo as follows:<br>		<li>Benchmark all three programs using the time command. For example, run 3000fifo as follows:<br><br>
	<tt>  time (./3000pc-fifo 100000 0 0 > fifo-output.log)</tt><br>		<tt>  time (./3000pc-fifo 100000 0 0 > fifo-output.log)</tt><br><br>
	Which one is fastest? Why?		Which one is fastest? Why?
	<li>(Advanced) Implement multiple producers and consumers in 3000pc-rendezvous. Compare your results to what you did in 3000pc-fifo.</li>		<li>(Advanced) Implement multiple producers and consumers in 3000pc-rendezvous. Compare your results to what you did in 3000pc-fifo.</li>