Amiga Unix Wiki

Because AmigaOS just isn't obscure enough today!

User Tools

Site Tools


tips-tricks

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
tips-tricks [2018/08/16 20:27]
wiki_admin
tips-tricks [2022/04/15 21:35]
wiki_admin missing link added
Line 1: Line 1:
 +This page contains miscellaneous small bits that haven't yet been further written into separate articles. Perhaps one day...
 +
 ====== Daemons ====== ====== Daemons ======
  
-A full install of Amix UNIX includes (and starts) some system daemons that are unlikely to be of much interest to the user. Below are brief descriptions of many of these daemons, how to disable them if they are started at boot, and how to start them if they are not.+A full install of Amix UNIX includes (and starts) some system daemons that are unlikely to be of much interest to the user in the modern day. Below are brief descriptions of many of these daemons, how to disable them if they are started at boot, and how to start them if they are not. There are a few other daemons (flopd, scrmenu, etc) I'm still looking into.
  
-lp+**lp** \\ 
 +Printing services. Unless you want to experiment with this, disable it. The only printers likely to work with AMIX at this point are pure postscript printers. 
  
-Printing services. Unless you want to experiment with this, disable it. The only printers likely to work with AMIX at this point are pure postscript printers. Stop: kill `ps -e | grep lp | cut -f 4 -d ' '` Disable: mv /etc/rc2.d/S80lp /etc/rc2.d/s80lp+Stop: kill `ps -e | grep lp | cut -f 4 -d ' '\\ 
 +Disable: mv /etc/rc2.d/S80lp /etc/rc2.d/s80lp \\
  
-news+**news** \\ 
 +Maybe you use Usenet. I bet you don't want to run an NNTP server though. 
  
-Maybe you use Usenet. I bet you don't want to run an NNTP server though. Stop: /etc/init.d/news stop Disable: mv /etc/rc2.d/S91news /etc/rc2.d/s91news+Stop: /etc/init.d/news stop \\ 
 +Disable: mv /etc/rc2.d/S91news /etc/rc2.d/s91news
  
-smtpd +**smtpd** \\ 
- +Mail services. If you intend to use this, go ahead and leave it running. I've never seen the system send mail for anything it runs on its own, so it should be fine to disable. 
-Mail services. If you intend to use this, go ahead and leave it running. I've never seen the system send mail for anything it runs on its own, so it should be fine to disable. Stop: kill `ps -e | grep smtpd | cut -f 4 -d ' '` Disable: mv /etc/rc2.d/S88smtpd /etc/rc2.d/s88smtpd +
- +
-rpc +
- +
-Various old rpc services that you probably shouldn't use anyway. Some people who have used more modern UNIX might think you need these for NFS to work. Nope. You can turn this off. Stop: /etc/init.d/rpc stop Disable: mv /etc/rc2.d/S75rpc /etc/rc2.d/s75rpc +
- +
-in.named +
- +
-DNS nameserver. If you are using DNS and not just hosts, your system will be much happier if this is running. For details on how to configure in.named, see details on this page. It is not enabled by default. Start: /usr/sbin/in.named Enable: Create an init script in.named, start it at 70. A simple example is attached to this page, place this in /etc/init.d/in.named and set it executable. Enable this at boot with the following: ln /etc/init.d/in.named /etc/rc2.d/S70in.named Now you can start/stop in.named with /etc/init.d/in.named start / stop +
- +
-There are a few other daemons (flopd, scrmenu, etc) I'm still looking into. +
- +
-====== Tips & tricks ====== +
- +
-This will be a mess at first, it's a scribble pad for me at the moment as I work. +
- +
-bash 3.0 on AMIX Compiles correctly after I fixed one source code file. AMIX uses deprecated trap designation in /etc/profile, so for CTRL-C to work you need to add a section like this to your /etc/profile:+
  
-if [ “$BASH” ] then trap 2 3 fi+Stop: kill `ps -e | grep smtpd | cut -f 4 -d ' '` \\ 
 +Disable: mv /etc/rc2.d/S88smtpd /etc/rc2.d/s88smtpd
  
-AMIX 2.1 Don't use tarUse gtar. tar segfaults until you patch to level 2a (2.1c)+**rpc** \\ 
 +Various old rpc services that you probably shouldn't use anywaySome people who have used more modern UNIX might think you need these for NFS to work. NopeYou can turn this off
  
-When configuring software for compilation you will have better luck specifying host/build/target as m68k-cbm-sysv4 than letting configure guess on its own (typically m68k-unknown-sysv4) +Stop: /etc/init.d/rpc stop \\ 
 +Disable: mv /etc/rc2.d/S75rpc /etc/rc2.d/s75rpc
  
-Edit /etc/inet/rc.inet to add default router, not /etc/defaultrouter (ignored by system) +**in.named** \\ 
 +DNS nameserver. If you are using DNS and not just hosts, your system will be much happier if this is running. For details on how to configure in.namedsee details on [[networking|this page]]. It is not enabled by default.
  
-Export CC=/usr/local/bin/gcc as some applications keep trying to use gcc-1.4.in /usr/public/bin during compilation +Start: /usr/sbin/in.named \\ 
 +Enable: Create an init script in.named, start it at 70. Enable this at boot with the following: ln /etc/init.d/in.named /etc/rc2.d/S70in.named Now you can start/stop in.named with /etc/init.d/in.named start stop
  
-Mounting NFS shares from servers without NFS shares will panic the system. Oops! Don't Do That ™. Fixed in 2a patchlevel. +====== Vulnerabilities ======
  
-/usr/include/sys/socket.h, and types.h are too buggy to be used in many situations. Replace with files from 2a patchlevel. +Believe it or not, Amiga Unix is not obscure enough to //not// be mentioned by major security firmsThere is a vulnerability in the included "Finger" command:
  
-Use sysadm to enable XDM, old method from 2.01/2.03 doesn't work anymore. +<code> 
 +Amiga Unix allows access to files using finger 
 +CVE-1999-1218
  
-Don't install 2a patchdiskHD hack install method leaves kernel uncompilable with patchesCopy the include files socket.h and types.h only+Details: amiga-finger (522) reported Feb 11993The finger command in Amiga allows non-privileged users to access files that they would not normally be able to access.
  
-Patches break w and netstat (at least) without kernel upgrade. +Consequences: Gain Access
  
-Adding swapdd if=/dev/zero of=/tmp/swap bs=1024k count=50 && swap -a /tmp/swap 0 102400 +Remedy As root, modify the permission of the existing /usr/bin/finger to prevent misuse.  
 +Issue the following command: # /bin/chmod 0755 /usr/bin/finger
  
-Compilation will fail on large binaries even with 100MB or more of swap. An error like this will result: ld: cc1: libelf error: Memory error: output file space elf_update:. This is due to artificial restrictions imposed by ulimit. Set everything possible to “unlimited”, ie: ulimit -H unlimited, etc. Certain of these users cannot set, if you get this error you must become root and unset all limits. Type ulimit -? to see what you can fiddle with. +— OR —
  
-collect2.c in gcc-2.95.3 has a bug for AMIXcomment out the SYS_SIGLIST defsHope this is right, guessing here. (it compiled afterDamncompiling gcc-2.95.3 is proving to be really educational in “FOR CRYING OUT LOUD JUST COMPILE” kind of way. This link will be helpful diagnosing late stage1 compile problems: http://sources.redhat.com/ml/crossgcc/1997/msg00326.html … appears 2.7.2.3 is buggy, wonderfulIt may be helpful to symlink /usr/X to /usr/X11 since /usr/X11 appears often on newer systems.+As rootinstall the pubsrc package from the distribution tape. 
 +In the /usr/src/pub/cmd/finger/src/finger.c file, add the line "setuid(getuid());" (without the quotes) immediately before the line reading "display_finger(finger_list);" 
 +Optionallysave copy of the existing /usr/bin/finger and modify its permission to prevent misuse: 
 +/bin/mv /usr/bin/finger /usr/bin/finger.orig 
 +# /bin/chmod 0755 /usr/bin/finger.orig 
 +In the /usr/src/pub/cmd/finger directory, issue the commands: 
 +# cd /usr/src/pub/cmd/finger 
 +# make install 
 +</code> 
 +Source: [[https://exchange.xforce.ibmcloud.com/vulnerabilities/522]]
  
 ====== Performance ====== ====== Performance ======
  
-Bonnie_Amix2.1c_A3000D +Overall, performance is…really poor. The NFS tests took a really long time. I won't have a good idea just how bad until I get the other tests completed, but I recall various quips and quotes from Usenet and reviews indicating disk performance being slow. There are patches to improve performance, but I think these are for 2.01/2.03. If it's possible to apply them to 2.1c I'll provide results for that as well. 
-Overall, performance is…really poor. The NFS tests took a really long time. I won't have a good idea just how bad until I get the other tests completed, but I recall various quips and quotes from Usenet and reviews indicating disk performance being slow. There are patches to improve performance, but I think these are for 2.01/2.03. If it's possible to apply them to 2.1c I'll provide results for that as well. On to the results. +
- +
-Bonnie testing parameters +
- +
-Bonnie was run with a 100MB test set. For AMIX it was compiled with gcc-2.7.2.3, using CFLAGS=-O2 -m68020 -fomit-frame-pointer. +
- +
-Key to X axis labels: +
- +
-OChr: Output a character at a time as fast as possible. OBlk: Output intelligently as fast as possible. RW: “Rewrite test” - Read some data, change it, write it, and read it again, intelligently, as fast as possible. IChr: Read a character at a time as fast as possible. IBlk: Read intelligently as fast as possible. +
- +
-Bonnie performance on disk +
- +
-First there's the disk performance using the UFS filesystem. I'm not going to bother testing S5 as it has so many limitations. Plotted are the rates in KB/sec and the average CPU utilization during the test. Image +
- +
-Bonnie performance on NFS +
- +
-This was a test of NFS performance. The NFS server was an i686 Debian GNU/Linux 5.0 system with 100Mb NIC writing to a RAID 5. I assure you the bottleneck was not the Linux machine. Plotted are the rates in KB/sec and the average CPU utilization during the test. Image +
- +
-Performance summary, disk and NFS +
- +
-This compares the data rates for disk and NFS, and disregards the CPU utilization. Plotted are the rates in KB/sec and the average CPU utilization during the test. Image +
- +
-Summary+
  
 I/O on Amix is relatively slow and uses relatively high amounts of CPU power to do the work. NFS is much slower. An advantage of NFS, though, is you can use RAID-backed storage at the other end. There is no such option for disk-based storage on AMIX. I/O on Amix is relatively slow and uses relatively high amounts of CPU power to do the work. NFS is much slower. An advantage of NFS, though, is you can use RAID-backed storage at the other end. There is no such option for disk-based storage on AMIX.
Line 88: Line 71:
 Results for NFS were much tighter across 4 runs than the disk data. The first disk test for IBlk was a bit of an aberration, using far less CPU and having less speed. Probably another process was kicked off by the system during this test and multitasking caused Bonnie to get less CPU time. Results for NFS were much tighter across 4 runs than the disk data. The first disk test for IBlk was a bit of an aberration, using far less CPU and having less speed. Probably another process was kicked off by the system during this test and multitasking caused Bonnie to get less CPU time.
  
-====== Benchmarking an old OS... who cares? ======+====== Tips & tricks ======
  
-This collection of pages is intended to provide some insight as to how Amiga UNIX performs relative to other OSs. Information would not mind seeing here includes:+This will be a mess at first, it's a scribble pad for me at the moment as I work.
  
-Performance of a different OS on the same hardware Performance of newer/better hardware vs “AMIX level” hardware Methodologies to increase Amiga UNIX performance +bash 3.0 on AMIX Compiles correctly after fixed one source code fileAMIX uses deprecated trap designation in /etc/profileso for CTRL-C to work you need to add section like this to your /etc/profile:
- +
-So, yes, fairly dull. The way would like to see this laid out is a seperate page for each set of testsThis waypeople can easily open results up in a new window (or tabs) for comparison. If you would like to submit your own results, first take a look at the list of tests you should run. You do not have to run all of these, but you should run as many as you can. Next, take look at my test results and try to get your report to look similar. +
- +
-The tests +
- +
-The tests are tailored for UNIX systems, but if you can get them to run in AmigaOS (and most of them are sufficiently basic that this should not be an issue) more power to you. +
- +
-Click here to see the list of tests +
- +
-The results +
- +
-Results are split into two categoriesAMIX-capable hardware, which by its nature should keep the machines to roughly the same spec; and other hardware, which must be classic Amiga hardware but does not need to be able to run AMIX.+
  
-AMIX-capable hardware Amiga UNIX 2.1c on A3000D, run by failure+<code> 
 +if [ "$BASH"
 +then 
 +trap - 2 
 +fi 
 +</code>
  
-other hardware+  * Don't use tar. Use gtar. tar segfaults until you patch to level 2a (2.1c).  
 +  * When configuring software for compilation you will have better luck specifying host/build/target as m68k-cbm-sysv4 than letting configure guess on its own (typically m68k-unknown-sysv4)  
 +  * Edit /etc/inet/rc.inet to add default router, not /etc/defaultrouter (ignored by system)  
 +  * Export CC=/usr/local/bin/gcc as some applications keep trying to use gcc-1.4.2 in /usr/public/bin during compilation  
 +  * Mounting NFS shares from servers without NFS shares will panic the system. Oops! Don't Do That ™. Fixed in 2a patchlevel.  
 +  * /usr/include/sys/socket.h, and types.h are too buggy to be used in many situations. Replace with files from 2a patchlevel.  
 +  * Use sysadm to enable XDM, old method from 2.01/2.03 doesn't work anymore.  
 +  * Patches break w and netstat (at least) without kernel upgrade.  
 +  * Adding swap: dd if=/dev/zero of=/tmp/swap bs=1024k count=50 && swap -a /tmp/swap 0 102400  
 +  * Compilation will fail on large binaries even with 100MB or more of swap. An error like this will result: ld: cc1: libelf error: Memory error: output file space elf_update:. This is due to artificial restrictions imposed by ulimit. Set everything possible to “unlimited”, ie: ulimit -H unlimited, etc. Certain of these users cannot set, if you get this error you must become root and unset all limits. Type ulimit -? to see what you can fiddle with.  
 +  * collect2.c in gcc-2.95.3 has a bug for AMIX, comment out the SYS_SIGLIST defs. Hope this is right, guessing here. (it compiled after) Damn, compiling gcc-2.95.3 is proving to be really educational in a “FOR CRYING OUT LOUD JUST COMPILE” kind of way. This link will be helpful diagnosing late stage1 compile problems: http://sources.redhat.com/ml/crossgcc/1997/msg00326.html … appears 2.7.2.3 is buggy, wonderful. 
 +  * It may be helpful to symlink /usr/X to /usr/X11 since /usr/X11 appears often on newer systems.
  
 ====== mkfs ====== ====== mkfs ======
Line 114: Line 101:
 Going to need this info to write this page later... Going to need this info to write this page later...
  
 +<code>
 # mkfs -F ufs /dev/dsk/c6d0s6 1841152 # mkfs -F ufs /dev/dsk/c6d0s6 1841152
 Mkfs: make ufs file system?  Mkfs: make ufs file system? 
Line 318: Line 306:
 /dev/dsk/c6d0s6    823205          823196    0%       /mnt /dev/dsk/c6d0s6    823205          823196    0%       /mnt
 # #
 +</code>
 +
 +
 +====== Internal Timers in Amiga Unix ======
 +
 +By David Miller (Commodore Applications and Technical Support, Commodore Business Machines, Inc.)
 +
 +Copied for safekeeping from [[http://amigadev.elowar.com/read/ADCD_2.1/AmigaMail_Vol2_guide/node0101.html|http://amigadev.elowar.com archives]]
 +
 +The Sleep() function
 +
 +When you want your program to pause for a number of seconds then continue, you will typically use the sleep(3) function (the notation NAME(SECTION#) refers to the manual page NAME in the SECTION# chapter of the UNIX Reference Manuals; section 1 is commands, section 2 is system calls, and section 3 is library functions).  There is also a sleep(1) program which provides the same function to shell scripts.  For example:
 +
 +<code>
 +In C:
 +
 +    main()
 +    {
 +            printf("Hello world!\n");
 +            sleep(5);
 +    }
 +
 +
 +Shell:
 +
 +    #!/usr/bin/sh
 +
 +    echo "Hello world!"
 +    sleep 5
 +</code>
 +
 +The following example is an implementation of the sleep(3) function.
 +
 +<code>
 +Example 1: SLEEP() Using ALARM()
 +--------------------------------
 +
 +      # include       <signal.h>
 +      static void trap ()
 +      {
 +      }
 +      unsigned sleep ( unsigned duration )
 +      {
 +              void            (*oldsig)();
 +              unsigned        oldtime;
 +              oldsig = signal(SIGALRM, trap);
 +    10          oldtime = alarm(0);
 +    11          if (oldtime && oldtime < duration)
 +    12                  alarm(oldtime);
 +    13          else
 +    14                  alarm(duration);
 +    15          pause();
 +    16          signal(SIGALRM, oldsig);
 +    17          if (oldtime > duration)
 +    18                  alarm(oldtime - duration);
 +    19          if (oldtime && oldtime < duration)
 +    20                  return(duration - oldtime);
 +    21          else
 +    22                  return 0;
 +    23  }
 +
 +   Line            Explanation
 +   ----            -----------
 +      1   The file signal.h defines the parameters for the signal(2)
 +          function.
 +
 +    2-4   trap() is the simple signal handler that just sets a flag that
 +          another piece of code will examine.
 +
 +      5   Define the sleep() function.
 +
 +    6-8   oldsig is a variable that will be used to save the previous
 +          state of the signal handler; oldtime will be used to save the
 +          state of the alarm clock.
 +
 +      9   Establish trap() as the current signal handler for the alarm
 +          signal.  The previous handler is saved in oldsig.
 +
 +     10   Clear the alarm clock, saving the current state.
 +
 +  11-14   If the previous setting of alarm was sooner than duration, use
 +          the old value, otherwise use duration to set the alarm.
 +
 +     15   Pause the process until the alarm goes off.
 +
 +     16   Restore the old signal handler.
 +
 +  17-18   If the old alarm setting was later than duration, reset the
 +          alarm with the difference between duration and oldtime (the
 +          time remaining until the previous alarm).
 +
 +  19-22   If an existing alarm is making sleep return early, return the
 +          time remaining on the requested sleep.
 +
 +</code>
 +
 +To schedule its wake up time, sleep(3) uses the alarm(2) system call. The alarm(2) system call asks the OS to deliver a signal (basically a software interrupt) in some number of seconds according to the system clock. However, setting an alarm for 2 seconds does not mean that you will receive an alarm signal in exactly two seconds.
 +
 +The OS processes alarm requests once every second. Each time an alarm request is processed, the number of seconds remaining for that alarm is decremented by one.  When the number of seconds remaining reaches zero,
 +the OS delivers a signal to the process.  Given this, if a process places a 1 second alarm request 1 microsecond before the OS does its alarm processing, the signal will arrive in 1 microsecond, not in 1 second. An alarm set for N seconds actually means:
 +
 +    deliver a signal after N-1 seconds, but before N seconds.
 +
 +If you'd like to see for yourself, try running this shell script:
 +
 +<code>
 +    $ for i in 1 2 3 4 5 6 7 8 9 10
 +    > do
 +    >       time sleep 2
 +    > done
 +</code>
 +
 +How many times did the process actually take more than 2 seconds?  If you try it with both the Korn shell, ksh(1), and the System V shell, sh(1), you'll find that under ksh(1) it takes about 1.2 seconds and under sh(1) it takes about 1.8 seconds (There is a difference because sh(1) counts the time required to start the sleep process whereas ksh(1) only counts the actual running time).  The sleep(1) program rarely, if ever, actually sleeps for 2 seconds.
 +
 +If you are writing a daemon that checks for some event every 5 minutes, or if you want to pause the output to give the user a chance to read it, alarm's 1 second granularity is fine.  But what about that daemon that needs to wake up every second?  Waking up after 1 microsecond could cause the process to run almost continuously.  For any sort of realtime processing, one second is a very long time.  So how do you sleep for less than one second reliably?
 +
 +The answer is do not use an alarm, use an interval timer.
 +
 +**Interval Timers**
 +
 +Each interval timer has a resolution of 1 tick of the system's clock, or 1 microsecond (whichever is larger). Additionally, you can configure an interval timer to automatically restart itself. The system provides each process with three independent interval timers:
 +
 +ITIMER_REAL
 +
 +This timer will count down in real time. That is, this timer will continue to run when your process is waiting for the OS to perform a system call, or when the OS preempts your process. When the timer expires, the OS will deliver a SIGALARM signal.
 +
 +ITIMER_VIRTUAL
 +
 +This timer counts down only when your process is running. If your process makes a system call, or is preempted, this timer will stop counting. The timer will resume when your process resumes execution. When this timer reaches zero, the process will receive a SIGVTALRM signal.
 +
 +Possible uses for this timer include checkpointing (saving data after some period of execution) and multithreading. The virtual timer is more desireable for these applications since it counts only when the process is running; there is no reason to perform a checkpoint or switch threads if the process has been idle.
 +
 +ITIMER_PROF
 +
 +This timer will stop counting any time your process is preempted by the OS, but will not stop when the process is waiting for a system call to return.  When it expires the OS generates a SIGPROF signal.
 +
 +This timer is designed to be used for execution profiling by interpreters. By having a profiling timer send a signal every second, or fraction of a second, and examining the current position in the interpreted code, the process can determine where the most execution time is being spent.
 +
 +All three timers operate on the following structure:
 +
 +<code>
 +    struct itimerval
 +    {
 +            struct timeval it_interval;
 +            struct timeval it_value;
 +    }
 +</code>
 +
 +The timeval structure looks like this:
 +
 +<code>
 +      struct timeval
 +      {
 +            long tv_sec;
 +            long tv_usec;
 +      }
 +</code>
 +
 +Both of these structures are defined in the <sys/time.h> include file.
 +
 +Note that System V Release 4.0 does not guarantee that these are the only members of these structures, nor that they will occur in this order.  You must initialize the members individually.  This can be annoying and tedious, but it allows the structure to be expanded in future releases.
 +
 +You set and examine timers using these two functions:
 +
 +<code>    int getitimer(int which, struct itimerval *myvalue) </code>
 +
 +places the current timer setting into myvalue
 +
 +<code>    int setitimer(int which, struct itimerval *myvalue, struct itimerval *myovalue); </code>
 +
 +sets the timer.  It gets the new time from myvalue and places the previous setting in myovalue.
 +
 +Now, let us take a look at the sleep function again.  The code in example 2 creates a new version of sleep that will sleep an exact number of seconds.
 +
 +<code>
 +Example 2: SLEEP() Using an Interval Timer
 +------------------------------------------
 +
 +      # include       <signal.h>
 +      # include       <sys/time.h>
 +      static void trap ()
 +      {
 +      }
 +      unsigned mysleep ( unsigned duration )
 +      {
 +             void                    (*oldsig)();
 +             struct itimerval        oldtime;
 +    10         struct itimerval        newtime;
 +    11         oldsig = signal(SIGALRM, trap);
 +    12         getitimer(ITIMER_REAL, &oldtime);
 +    13         if (oldtime.it_value.tv_sec == 0
 +                        && oldtime.it_value.tv_usec == 0
 +                        || oldtime.it_value.tv_sec >= duration)
 +    14         {
 +    15                 newtime.it_interval.tv_sec = 0;
 +    16                 newtime.it_interval.tv_usec = 0;
 +    17                 newtime.it_value.tv_sec = duration;
 +    18                 newtime.it_value.tv_usec = 0;
 +    19                 setitimer(ITIMER_REAL, &newtime, NULL);
 +    20         }
 +    21         pause();
 +    22         signal(SIGALRM, oldsig);
 +    23         if (oldtime.it_value.tv_sec > duration)
 +    24         {
 +    25                 oldtime.it_value.tv_sec -= duration;
 +    26                 setitimer(ITIMER_REAL, &newtime, NULL);
 +    27         }
 +    28         if (oldtime && oldtime < n)
 +    29                 return(n - oldtime);
 +    30         else
 +    31                 return 0;
 +    32  }
 +
 +
 +  Line        Explanation
 +  ----        -----------
 +   1-2    The file signal.h defines the parameters of the signal(2)
 +          function.  The file <sys/time.h> defines ITIMER_REAL and the
 +          structures used by getitimer(3) and setitimer(3).
 +
 +   3-5    trap() is the simple signal handler that just sets a flag that
 +          another piece of code will examine.
 +
 +   6-7    Define the mysleep() function.
 +
 +  8-10    oldsig will hold the previous state of the signal handler;
 +          oldtime will hold the state of the timer; and newtime will be
 +          used to set the new timer parameters.
 +
 +     11   Establish trap() as the current signal handler for the alarm
 +          signal.  The previous handler is saved in oldsig.
 +
 +     12   Fetch the current settings of the timer.
 +
 +  13-20   If the timer was idle (both parts of it_value are zero) or it
 +          is set to go off later than duration (it_value.tv_sec is greater
 +          than duration), set the timer to go off in  duration seconds.
 +
 +     21   pause(2) the process until the timer expires.
 +
 +     22   Restore the old signal handler.
 +
 +  23-27   If the old timer setting was later than duration, reset the
 +          timer with the difference between duration and
 +          oldtime.it_value.tv_sec (the time remaining until expiration of
 +          the previous setting).
 +
 +  28-32   If an existing timer is making mysleep() return early, return
 +          the time remaining on the requested mysleep().
 +</code>
 +
 +This example uses what is called a one-shot timer. The timer goes off once, and then stops. By supplying an interval setting, the timer becomes a clock, generating alarm signals on a regular basis.
 +
 +The code fragment in Example 3 shows how to set the timer to produce a 1.5 second clock that will start ticking in 1 minute.
 +
 +<code>
 +Example 3: Using an Interval Timer as a Clock
 +---------------------------------------------
 +
 +      # include       <sys/time.h>
 +      ...
 +              {
 +                      struct itimerval        newtime;
 +                      newtime.it_value.tv_sec = 60;
 +                      newtime.it_value.tv_usec = 0;
 +                      newtime.it_interval.tv_sec = 5;
 +                      newtime.it_interval.tv_usec = 500000;
 +                      setitimer(ITIMER_REAL, &newtime, 0);
 +    10          }
 +    11  ...
 +
 +
 +  Line        Explanation
 +  ----        -----------
 +       The file <sys/time.h> defines ITIMER_REAL and the structures
 +         used by getitimer(3) and setitimer(3).
 +
 +       Other code.
 +
 +       Start of block.
 +
 +       Structure newtime will hold the setting for the timer.
 +
 +   5-6   Set it_value to expire in 1 minute.
 +
 +   7-8   Set it_interval to reload it_value with 5 seconds and 500,000.
 +         microseconds (one half of a second).
 +
 +       Load the new settings into the ITIMER_REAL timer.
 +
 +    10   End of block.
 +
 +    11   Other code.
 +</code>
 +
 +When the time interval specified by it_value expires, the contents of it_interval is copied into it_value and the timer is restarted. If the interval specified by it_interval is zero, the timer stops.  Timers can be stopped at anytime by calling setitimer(3) with the members of it_value set to zero.
 +
 +So there you have the realtime interval timer. The other timers work exactly the same way, varying only in when the timer is running.  For more information on interval timers see UNIX System V Release 4 - Programmer's Reference Manual, published by Prentice Hall. Next time, we'll explore context switching, multithreading, and light weight processes.
tips-tricks.txt · Last modified: 2022/04/15 21:35 by wiki_admin