#!/bin/sh # This is a shell archive (produced by GNU sharutils 4.2). # To extract the files from this archive, save it to some FILE, remove # everything before the `!/bin/sh' line above, then type `sh FILE'. # # Made on 1999-02-24 17:02 EST by . # Source directory was `/var/home/jcej/projects/ACE_wrappers/docs/tutorials/018'. # # Existing files will *not* be overwritten unless `-c' is specified. # # This shar contains: # length mode name # ------ ---------- ------------------------------------------ # 416 -rw-rw-r-- hdr # 64 -rw-rw-r-- bodies # 2603 -rw-rw-r-- page01.pre # 430 -rw-rw-r-- page02.pre # 1282 -rw-rw-r-- page03.pre # 689 -rw-rw-r-- page04.pre # 260 -rw-rw-r-- page05.pre # 1523 -rw-rw-r-- page06.pre # 478 -rw-rw-r-- page07.pre # save_IFS="${IFS}" IFS="${IFS}:" gettext_dir=FAILED locale_dir=FAILED first_param="$1" for dir in $PATH do if test "$gettext_dir" = FAILED && test -f $dir/gettext \ && ($dir/gettext --version >/dev/null 2>&1) then set `$dir/gettext --version 2>&1` if test "$3" = GNU then gettext_dir=$dir fi fi if test "$locale_dir" = FAILED && test -f $dir/shar \ && ($dir/shar --print-text-domain-dir >/dev/null 2>&1) then locale_dir=`$dir/shar --print-text-domain-dir` fi done IFS="$save_IFS" if test "$locale_dir" = FAILED || test "$gettext_dir" = FAILED then echo=echo else TEXTDOMAINDIR=$locale_dir export TEXTDOMAINDIR TEXTDOMAIN=sharutils export TEXTDOMAIN echo="$gettext_dir/gettext -s" fi touch -am 1231235999 $$.touch >/dev/null 2>&1 if test ! -f 1231235999 && test -f $$.touch; then shar_touch=touch else shar_touch=: echo $echo 'WARNING: not restoring timestamps. Consider getting and' $echo "installing GNU \`touch', distributed in GNU File Utilities..." echo fi rm -f 1231235999 $$.touch # if mkdir _sh30135; then $echo 'x -' 'creating lock directory' else $echo 'failed to create lock directory' exit 1 fi # ============= hdr ============== if test -f 'hdr' && test "$first_param" != -c; then $echo 'x -' SKIPPING 'hdr' '(file already exists)' else $echo 'x -' extracting 'hdr' '(text)' sed 's/^X//' << 'SHAR_EOF' > 'hdr' && X X X ACE Tutorial 018 X
ACE Tutorial 018
X
The FIFO Nature of ACE_Token
X

SHAR_EOF $shar_touch -am 1114141798 'hdr' && chmod 0664 'hdr' || $echo 'restore of' 'hdr' 'failed' if ( md5sum --help 2>&1 | grep 'sage: md5sum \[' ) >/dev/null 2>&1 \ && ( md5sum --version 2>&1 | grep -v 'textutils 1.12' ) >/dev/null; then md5sum -c << SHAR_EOF >/dev/null 2>&1 \ || $echo 'hdr:' 'MD5 check failed' 66dbcd27e23cdcc9c230089e9c289bcb hdr SHAR_EOF else shar_count="`LC_ALL= LC_CTYPE= LANG= wc -c < 'hdr'`" test 416 -eq "$shar_count" || $echo 'hdr:' 'original size' '416,' 'current size' "$shar_count!" fi fi # ============= bodies ============== if test -f 'bodies' && test "$first_param" != -c; then $echo 'x -' SKIPPING 'bodies' '(file already exists)' else $echo 'x -' extracting 'bodies' '(text)' sed 's/^X//' << 'SHAR_EOF' > 'bodies' && PAGE=2 token.cpp Test_T.h Test_T.cpp Token_i.h Mutex_i.h output SHAR_EOF $shar_touch -am 1114165298 'bodies' && chmod 0664 'bodies' || $echo 'restore of' 'bodies' 'failed' if ( md5sum --help 2>&1 | grep 'sage: md5sum \[' ) >/dev/null 2>&1 \ && ( md5sum --version 2>&1 | grep -v 'textutils 1.12' ) >/dev/null; then md5sum -c << SHAR_EOF >/dev/null 2>&1 \ || $echo 'bodies:' 'MD5 check failed' 22e70b25b6f23655b44d31fcf1a669f8 bodies SHAR_EOF else shar_count="`LC_ALL= LC_CTYPE= LANG= wc -c < 'bodies'`" test 64 -eq "$shar_count" || $echo 'bodies:' 'original size' '64,' 'current size' "$shar_count!" fi fi # ============= page01.pre ============== if test -f 'page01.pre' && test "$first_param" != -c; then $echo 'x -' SKIPPING 'page01.pre' '(file already exists)' else $echo 'x -' extracting 'page01.pre' '(text)' sed 's/^X//' << 'SHAR_EOF' > 'page01.pre' && X Welcome to Tutorial 18!

We've seen various ACE methods for synchronization in this and other tutorial sections. Something we haven't yet seen is the ACE_Token. ACE_Token has a really cool thing: it behaves in a FIFO manner.

Why is that cool?

In the other tutorials, you may have found that one thread will end up with all of the work. Even though other threads are available, the OS scheduling and lock management just causes it to happen. With ACE_Token, the threads are queued up on the token and served in a traditional first-in-first-out manner.

Why is FIFO important?

Well, if your app is running in a bunch of threads and each is doing the same thing on the local host then FIFO may not be important. However, take the case where each thread is connected to a remote system. Let's say you have a dozen threads in your app and each is connected to a different remote system. Each of the threads will be given a block of data which will be passed to the remote for some intense calculation. If you use the FIFO then you'll spread the work more-or-less evenly between the remote peers. If you use the traditional mutex then one peer may get the lion's share of the work.

It gets down to a personal decision based on the application's needs. Consider your application, examine its behavior & decide for yourself if you want to spread the work evenly or if it's OK to let some threads work harder than others.

Kirthika's abstract:

SHAR_EOF $shar_touch -am 0224170199 'page01.pre' && chmod 0664 'page01.pre' || $echo 'restore of' 'page01.pre' 'failed' if ( md5sum --help 2>&1 | grep 'sage: md5sum \[' ) >/dev/null 2>&1 \ && ( md5sum --version 2>&1 | grep -v 'textutils 1.12' ) >/dev/null; then md5sum -c << SHAR_EOF >/dev/null 2>&1 \ || $echo 'page01.pre:' 'MD5 check failed' 89cd84f46240da6d7d4d50e4d303e17a page01.pre SHAR_EOF else shar_count="`LC_ALL= LC_CTYPE= LANG= wc -c < 'page01.pre'`" test 2603 -eq "$shar_count" || $echo 'page01.pre:' 'original size' '2603,' 'current size' "$shar_count!" fi fi # ============= page02.pre ============== if test -f 'page02.pre' && test "$first_param" != -c; then $echo 'x -' SKIPPING 'page02.pre' '(file already exists)' else $echo 'x -' extracting 'page02.pre' '(text)' sed 's/^X//' << 'SHAR_EOF' > 'page02.pre' && Our main() just keeps getting simpler! I guess that's a good thing.

What we've done is create two Task-derived objects that test different locking mechanisms. The Token object uses ACE_Token and the Mutex object uses ACE_Mutex. When you execute the application you should see quite a difference in thread utilization. At the end of the tutorial I've included a link to the output of a typical run of the application.

SHAR_EOF $shar_touch -am 1114153698 'page02.pre' && chmod 0664 'page02.pre' || $echo 'restore of' 'page02.pre' 'failed' if ( md5sum --help 2>&1 | grep 'sage: md5sum \[' ) >/dev/null 2>&1 \ && ( md5sum --version 2>&1 | grep -v 'textutils 1.12' ) >/dev/null; then md5sum -c << SHAR_EOF >/dev/null 2>&1 \ || $echo 'page02.pre:' 'MD5 check failed' a36353959f7874c8e31884d2acd7eb43 page02.pre SHAR_EOF else shar_count="`LC_ALL= LC_CTYPE= LANG= wc -c < 'page02.pre'`" test 430 -eq "$shar_count" || $echo 'page02.pre:' 'original size' '430,' 'current size' "$shar_count!" fi fi # ============= page03.pre ============== if test -f 'page03.pre' && test "$first_param" != -c; then $echo 'x -' SKIPPING 'page03.pre' '(file already exists)' else $echo 'x -' extracting 'page03.pre' '(text)' sed 's/^X//' << 'SHAR_EOF' > 'page03.pre' && Our Test object is a simple derivative of ACE_Task. Although we've certainly used templates in other tutorials, this is the first time we've created one of our own.

In a lot of ways, you can think of templates as the sophisticated cousin of the old C-style pre-processor macros. With templates, however, you get better type-checking and much easier debugging. There are certainly other benefits but these are my favorites.

Our template's MUTEX parameter is used to set the mutex type mutex_t. That'll be used in svc() so that we can protect shared resources needed during the processing of data received on our message queue.

Note at the bottom how we have to include the cpp file associated with us. Most compilers have to see the definition of templated classes along with their declaration. You might be tempted, therefore, to just put the definitions in the header file. Resist that temptation because templates are one of the fastest growing areas of compilers. Including the definition like we do here leads to long compile times and overly-large binaries. With luck, the compilers will get smarter in the future and we won't need definition inclusion. If you've already got them broken out then you'll save yourself a lot of time!

SHAR_EOF $shar_touch -am 1114154698 'page03.pre' && chmod 0664 'page03.pre' || $echo 'restore of' 'page03.pre' 'failed' if ( md5sum --help 2>&1 | grep 'sage: md5sum \[' ) >/dev/null 2>&1 \ && ( md5sum --version 2>&1 | grep -v 'textutils 1.12' ) >/dev/null; then md5sum -c << SHAR_EOF >/dev/null 2>&1 \ || $echo 'page03.pre:' 'MD5 check failed' 4f59a65824406215b1d28b2335b9c691 page03.pre SHAR_EOF else shar_count="`LC_ALL= LC_CTYPE= LANG= wc -c < 'page03.pre'`" test 1282 -eq "$shar_count" || $echo 'page03.pre:' 'original size' '1282,' 'current size' "$shar_count!" fi fi # ============= page04.pre ============== if test -f 'page04.pre' && test "$first_param" != -c; then $echo 'x -' SKIPPING 'page04.pre' '(file already exists)' else $echo 'x -' extracting 'page04.pre' '(text)' sed 's/^X//' << 'SHAR_EOF' > 'page04.pre' && Our Test implementation looks much like the other ACE_Task derivatives we've used in the past. The most obvious change is the addition of the run() method. run() will activate the threads and put a few messages into the queue. We could have done that in main() but it just makes more sense here.

Notice how svc() grabs the guard after getting a message from the queue. Since we constructed our Task baseclass with ACE_MT_SYNCH, we know that the queue is already thread-safe. Our purpose in grabbing the additional lock is to show how ACE_Token and ACE_Mutex behave differently. In a real app, you'd be doing this to protect shared resources that the threads might clobber.

SHAR_EOF $shar_touch -am 1114161298 'page04.pre' && chmod 0664 'page04.pre' || $echo 'restore of' 'page04.pre' 'failed' if ( md5sum --help 2>&1 | grep 'sage: md5sum \[' ) >/dev/null 2>&1 \ && ( md5sum --version 2>&1 | grep -v 'textutils 1.12' ) >/dev/null; then md5sum -c << SHAR_EOF >/dev/null 2>&1 \ || $echo 'page04.pre:' 'MD5 check failed' 428df54cf8483880bfcace604cd9ae3c page04.pre SHAR_EOF else shar_count="`LC_ALL= LC_CTYPE= LANG= wc -c < 'page04.pre'`" test 689 -eq "$shar_count" || $echo 'page04.pre:' 'original size' '689,' 'current size' "$shar_count!" fi fi # ============= page05.pre ============== if test -f 'page05.pre' && test "$first_param" != -c; then $echo 'x -' SKIPPING 'page05.pre' '(file already exists)' else $echo 'x -' extracting 'page05.pre' '(text)' sed 's/^X//' << 'SHAR_EOF' > 'page05.pre' && Here we create simple derivatives of our Test templated class. Each is parameterized with our mutex of choice and "named". Using the Test template we're able to reuse all of the code with practially no retyping and certainly much less chance of error!
SHAR_EOF $shar_touch -am 1114161298 'page05.pre' && chmod 0664 'page05.pre' || $echo 'restore of' 'page05.pre' 'failed' if ( md5sum --help 2>&1 | grep 'sage: md5sum \[' ) >/dev/null 2>&1 \ && ( md5sum --version 2>&1 | grep -v 'textutils 1.12' ) >/dev/null; then md5sum -c << SHAR_EOF >/dev/null 2>&1 \ || $echo 'page05.pre:' 'MD5 check failed' 8f427c59ed060b06d99dfa81e0447454 page05.pre SHAR_EOF else shar_count="`LC_ALL= LC_CTYPE= LANG= wc -c < 'page05.pre'`" test 260 -eq "$shar_count" || $echo 'page05.pre:' 'original size' '260,' 'current size' "$shar_count!" fi fi # ============= page06.pre ============== if test -f 'page06.pre' && test "$first_param" != -c; then $echo 'x -' SKIPPING 'page06.pre' '(file already exists)' else $echo 'x -' extracting 'page06.pre' '(text)' sed 's/^X//' << 'SHAR_EOF' > 'page06.pre' && That's it for the code, now let's take a quick look at some output. The first set of output is from the Token test, the second is Mutex test. Notice how the threads are evenly utilized in the Token test. Each thread gets to see exactly two messages. There's also an interesting side-effect that the messages are processed in order. (You can't rely on that, it just happend in this particular run.)

With the Mutex test, however, we see that the first thread gets no less than 1/2 of all messages. In fact, if we didn't have the governor in svc() it might have gotten them all!

Why does this happen?

Primarily because of time slicing. Even though each thread takes time to do work (1 second in our test), it can still own the timeslice when it gets back to the mutex acquire. Since the other threads are still switched out, the current thread regets the lock and continues. On the other hand, the ACE_Token is very careful about the order in which the acquisition is allowed and more evenly distributes the work.

Play around with the sleep() call in svc(). You'll find that as you decrease it, there is more chance that even the Token test will do most of its work in one thread. You're still at the mercy of the OS time slicing. In reality, though, it will take a moment or two for work to be done. The end goal isn't necessarily to distribute the work evenly over all threads but, rather, to distribute it evenly among available threads. The distinction is subtle but important.

SHAR_EOF $shar_touch -am 1114162298 'page06.pre' && chmod 0664 'page06.pre' || $echo 'restore of' 'page06.pre' 'failed' if ( md5sum --help 2>&1 | grep 'sage: md5sum \[' ) >/dev/null 2>&1 \ && ( md5sum --version 2>&1 | grep -v 'textutils 1.12' ) >/dev/null; then md5sum -c << SHAR_EOF >/dev/null 2>&1 \ || $echo 'page06.pre:' 'MD5 check failed' 9a4cc7f2ddce9c585a86113da4f5eb8d page06.pre SHAR_EOF else shar_count="`LC_ALL= LC_CTYPE= LANG= wc -c < 'page06.pre'`" test 1523 -eq "$shar_count" || $echo 'page06.pre:' 'original size' '1523,' 'current size' "$shar_count!" fi fi # ============= page07.pre ============== if test -f 'page07.pre' && test "$first_param" != -c; then $echo 'x -' SKIPPING 'page07.pre' '(file already exists)' else $echo 'x -' extracting 'page07.pre' '(text)' sed 's/^X//' << 'SHAR_EOF' > 'page07.pre' && And now we're at the end of another Tutorial. As always, feel free to send in questions and comments. There are certainly more implementation possibilites and I'll gladly integrate yours into these Tutorials.

SHAR_EOF $shar_touch -am 1114162598 'page07.pre' && chmod 0664 'page07.pre' || $echo 'restore of' 'page07.pre' 'failed' if ( md5sum --help 2>&1 | grep 'sage: md5sum \[' ) >/dev/null 2>&1 \ && ( md5sum --version 2>&1 | grep -v 'textutils 1.12' ) >/dev/null; then md5sum -c << SHAR_EOF >/dev/null 2>&1 \ || $echo 'page07.pre:' 'MD5 check failed' e1a201541e51dd42654d31440eb4f36c page07.pre SHAR_EOF else shar_count="`LC_ALL= LC_CTYPE= LANG= wc -c < 'page07.pre'`" test 478 -eq "$shar_count" || $echo 'page07.pre:' 'original size' '478,' 'current size' "$shar_count!" fi fi rm -fr _sh30135 exit 0