Re: Break point

["Steve Richter" <srichter@xxxxxxxxxxxxx>]

so when a pgm is in debug mode, non debug jobs that call the pgm will run slower than when the pgm is not being debugged by any job on the system.

 

do ile breakpoints insert traps in the pgm object also?

 

does the power pc chip have the capability to do it another way?

 

any idea how the pentium implements breakpoints?

 

thanks,

 

-Steve

-----Original Message-----
From: Leif Svalgaard <leif@leif.org>
To: MI400@midrange.com <MI400@midrange.com>
Date: Monday, June 04, 2001 6:03 PM
Subject: Re: Break point

From: Steve Richter

To: MI400@midrange.com

Sent: Monday, June 04, 2001 4:37 PM

Subject: Re: Break point

From: Steve Richter

A debugger feature plea:  When in debug mode, provide an "AddAddrBkp" cmd. When an address range is referenced in the job being debugged, call a debug handler pgm.  Similar to the action of the breakpoint handing pgm that is called when an instruction is executed.

 

===> that may be hard. The way the orignal breakpoint (BRK "here") is by compiling into a dummy instruction.

Is you ADDBKP "here", the dummy instruction is actually replaced in the program object by a trap instruction.

 

 

Leif,

 

When you AddBkp, a trap instruction is inserted in the pgm object?

 

===> yes, absolutely

 

What happens when the pgm is called by a non debug mode job?

 

===> the trap that is taken detects that the program is not under debug

in this job and does nothing. I didn't design this  :-)