[Info-vax] Volatile, was: Re: yet another sys$qiow question

[Johnny Billquist]

On 2015-08-19 17:35, "VAXman- "@SendSpamHere.ORG wrote:
> Question:  If "volatile" is necessary, shouldn't this code keep spinning its
> tail???
>
> #pragma environment save
> #ifndef __NEW_STARLET
> #define __NEW_STARLET New_and_Improved
> #endif
>
> #include <descrip.h>		// OpenVMS descriptor structure definitions
> #include <efndef.h>		// OpenVMS event flag number definitions
> #include <iodef.h>		// OpenVMS $QIO I/O function code definitions
> #include <iosbdef.h>		// OpenVMS I/O status block structure definitions
> #include <starlet.h>		// OpenVMS system service prototype & codes
> #include <stsdef.h>		// OpenVMS system message format definitions
>
> int main()
>    {
>      int sts = 0, ctr = 0;
>      IOSB iosb = { 0, 0, 0};
>
>      unsigned short channel;
>      $DESCRIPTOR(sys$output,"SYS$OUTPUT");
>
>      char star[] ={'|','/','-','\\'};
>
>      if (!((sts=SYS$ASSIGN(
> 				&sys$output,		// device name
> 				&channel,		// channel
> 				0,			// access mode
> 				0			// mailbox name
> 			 ))&STS$M_SUCCESS)) return sts;
>
>      while (iosb.iosb$w_status == 0)
>        {
> 	if (!((sts=SYS$QIOW(
> 				EFN$C_ENF,		// event flag number
> 				channel,		// channel
> 				IO$_WRITEVBLK,		// QIO function
> 				&iosb,			// I/O status block
> 				0,			// AST address
> 				0,			// AST parameter
> 				&star[(ctr++)&3],	// P1: VA of buffer
> 				1,			// P2: number of bytes
> 				0,			// P3: n/a
> 				0x8D000000,		// P4: carriage control
> 				0,			// P5: n/a
> 				0			// P6: n/a
> 			   ))&STS$M_SUCCESS)) return sts;
>        }
>    }
>
> #pragma environment restore

First question is (obviousy) how IOSB is defined. (It's a typedef in 
some include file, I would assume...)

But that aside this works for the exact reason I talked about before.
SYS$QIOW() is a function call inside the loop. And thus, the compiler 
cannot know what gets changed by that function call, so the iosb cannot 
be cached between iterations in the loop.

	Johnny