AutoIt3Wrapper pagefaulting a lot?

[exolon]

When I run a program from SciTE, AutoIt3Wrapper seems to wait for it to finish, pagefaulting 262 times every second or so. This seems like a slightly excessive overhead, and I'm not sure what it's for. It's nothing major but if there's a way to avoid it, that'd be better.

Does anyone know why this is? If I'm going to run a program for a while, I usually just kill AutoIt3Wrapper in the task manager, with no ill effects. Sounds like it's polling something repeatedly.

[PsaltyDS]

When I run a program from SciTE, AutoIt3Wrapper seems to wait for it to finish, pagefaulting 262 times every second or so. This seems like a slightly excessive overhead, and I'm not sure what it's for. It's nothing major but if there's a way to avoid it, that'd be better.

Does anyone know why this is? If I'm going to run a program for a while, I usually just kill AutoIt3Wrapper in the task manager, with no ill effects. Sounds like it's polling something repeatedly.

It's running your un-compiled script. If you don't want that overhead, compile the script and run the .exe.

[exolon]

It's running your un-compiled script. If you don't want that overhead, compile the script and run the .exe.

No, I don't think so; if this was the case, my program would terminate or hang when I kill AutoIt3Wrapper. But like I said, I can kill it with no ill effects.

I think the wrapper is kind of like a build manager, invoking the desired programs during the build/run process (like the obfuscator, UPX etc), and blocks until the user program has terminated.

As you say, compiling the script and running the .exe also avoids the overhead, just like killing AutoIt3Wrapper after the program has started.

[trancexx]

No, I don't think so; if this was the case, my program would terminate or hang when I kill AutoIt3Wrapper.

Why?

[Jos]

When you hit F5(Run) then AutoIt3Wrapper script is started 2 times:

1. This is the original one that shells AutoIt3 with your script as input and Monitors STDOUT of this shelled session which is displayed in the SciTE Outputpane.

2. This is a Second instance which sole purpose is to monitor if either the Autoit3 program ended (or got killed) or the AutoIt3Wrapper has ended. This is done this way to allow people to used the Tools/Stop Executing (Ctrl+Break) from SciTE to kill the execution of the Script.

Jos

Edited October 7, 2009 by Jos

[exolon]

When you hit F5(Run) then AutoIt3Wrapper script is started 2 times:

1. This is the original one that shells AutoIt3 with your script as input and Monitors STDOUT of this shelled session which is displayed in the SciTE Outputpane.

2. This is a Second instance which sole purpose is to monitor if either the Autoit3 program ended (or got killed) or the AutoIt3Wrapper has ended. This is done this way to allow people to used the Tools/Stop Executing (Ctrl+Break) from SciTE to kill the execution of the Script.

Thanks for this explanation. I modified my program to print to stdout every 500ms and had a look at the two instances of AutoIt3Wrapper in the task manager.

If I kill the one causing lots of pagefaults, there's no visible effect, except hitting ctrl-break will then fail to terminate the program (contrary to the "forcing abrupt termination" message):

>Running:(3.3.0.0):C:\Program Files\AutoIt3\autoit3.exe "C:\Documents and Settings\Ulysses\My Documents\code!\autoit\blinkenwords.au3"    
SleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleepSleep
>Process failed to respond; forcing abrupt termination...
>Exit code: 1    Time: 17.255

If I kill the other one, the running program terminates immediately along with the other wrapper.

So I guess that the wrapper instance that determines whether "the Autoit3 program ended (or got killed) or the AutoIt3Wrapper has ended" is the one that causes the pagefaults. Alright

[Jos]

So I guess that the wrapper instance that determines whether "the Autoit3 program ended (or got killed) or the AutoIt3Wrapper has ended" is the one that causes the pagefaults. Alright

Conclusion sounds correct (Not tested it myself).

The first instance shell itself again with this line in the Run section:

; Run second version as Watcher to kill The running AutoItscript when AutoIt3Wrapper is killed.
    Global $CW = Run(@ScriptFullPath & " /Watcher " & @AutoItPID & " " & $Pid)

This is the code executing in the second "watcher" version:

Case $T_Var = "/Watcher"
            ; when AutoIt3Wrapper is lanched as watcher to see if the original AutoIt3Wrapper is canceled.
            $H_Cmp = $CMDLINE[$x + 1]
            $H_au3 = $CMDLINE[$x + 2]
            While ProcessExists($H_Cmp) And ProcessExists($H_au3)
                Sleep(500)
            WEnd
            Sleep(500)
            If ProcessExists($H_au3) Then
                ProcessClose($H_au3)
                _RefreshSystemTray()
            EndIf
            Exit

That's really all that is happening.

Jos

Edited October 7, 2009 by Jos

[exolon]

This is the code executing in the second "watcher" version:

Case $T_Var = "/Watcher"
            ; when AutoIt3Wrapper is lanched as watcher to see if the original AutoIt3Wrapper is canceled.
            $H_Cmp = $CMDLINE[$x + 1]
            $H_au3 = $CMDLINE[$x + 2]
            While ProcessExists($H_Cmp) And ProcessExists($H_au3)
                Sleep(500)
            WEnd
            Sleep(500)
            If ProcessExists($H_au3) Then
                ProcessClose($H_au3)
                _RefreshSystemTray()
            EndIf
            Exit

That's really all that is happening.

Yes, this makes sense now. I did a quick check with the following snippet:

For $x = 1 To 30
    ProcessExists("scite.exe")
;   ProcessGetStats("scite.exe")
    ConsoleWrite("x")
    Sleep(500)
Next

...and each of those calls costs 108 pagefaults on my machine. Maybe this is unavoidable in the Windows API that gets called - I don't remember using such calls in Windows programming before. Not a big deal really, I was just interested to find out what it was

[Jos]

Yes, this makes sense now. I did a quick check with the following snippet:

For $x = 1 To 30
    ProcessExists("scite.exe")
;   ProcessGetStats("scite.exe")
    ConsoleWrite("x")
    Sleep(500)
Next

...and each of those calls costs 108 pagefaults on my machine. Maybe this is unavoidable in the Windows API that gets called - I don't remember using such calls in Windows programming before. Not a big deal really, I was just interested to find out what it was

Same here, always interested when things like this happen or things can be improved.

Jos