Lecture 10, Locks

* demonstrate problem (Ch. 26, x86.py)
* simple lock examples (Ch. 28, x86.py)
* scalability of concurrent structures (Ch. 29)
  - global shared counter
  - sloppy counter (weak/eventual consistency)
  - concurrent linked list

Ch. 28 & 29 of OSTEP (after a quick look back at Ch. 26):

Get the .tgz files for Threads Intro) and Threads (Locks) from
http://pages.cs.wisc.edu/~remzi/OSTEP/Homework/homework.html

(n.b.: They will unpack into the current directory, so create a subdir
and cd into it first.)

First, demonstrate the problem (examples from Ch. 26):

wifi-139-170:ThreadsIntro rdv$ pwd
/Users/rdv/Dropbox/old/rdv/keio/sfc/teaching/rdvteachingweb/undergrad-OS/undergrad-OS-2018/ThreadsIntro
./x86.py -p simple-race.s -t 1 
./x86.py -p simple-race.s -t 1 -M 2000 -R ax,bx -c
./x86.py -p looping-race-nolock.s -t 2 -a bx=1 -M 2000 -c
./x86.py -p looping-race-nolock.s -t 2 -a bx=1 -M 2000 -i 2
./x86.py -p looping-race-nolock.s -t 2 -a bx=1 -M 2000 -i 2 -c
./x86.py -p looping-race-nolock.s -t 2 -a bx=1 -M 2000 -c -R ax,bx
./x86.py -p looping-race-nolock.s -t 2 -a bx=1 -M 2000 -c -R ax,bx -i 2

Now see some possible solutions (from Ch. 28):

First, runs okay:
./x86.py -p flag.s -t 2 -a bx=1 -M count -c -R ax,bx
But this still doesn't work under all circumstances:
./x86.py -p flag.s -t 2 -a bx=1 -M count -c -R ax,bx -i 2
./x86.py -p flag.s -t 2 -a bx=2 -M count -c -R ax,bx
./x86.py -p flag.s -t 2 -a bx=3 -M count -c -R ax,bx
./x86.py -p flag.s -t 2 -a bx=3 -M count -c -R ax,bx -i 2
./x86.py -p flag.s -t 2 -a bx=3 -M count -c -R ax,bx -i 3

Test-and-set example:
./x86.py -p test-and-set.s -t 2 -a bx=2 -M count -c -R ax,bx
./x86.py -p test-and-set.s -t 2 -a bx=2 -M count -c -R ax,bx -i 2

And with three threads:
./x86.py -p test-and-set.s -t 3 -a bx=2 -M count -c -R ax,bx -i 2
./x86.py -p test-and-set.s -t 3 -a bx=3 -M count -c -R ax,bx -i 2