From - Fri Sep 05 08:07:12 1997
Path: news.mitre.org!blanket.mitre.org!nntprelay.mathworks.com!newsfeed1-hme1!newsfeed.internetmci.com!208.206.176.15!dimensional.com!not-for-mail
From: Eric F. Richards <efricha@flatland.dimensional.com>
Newsgroups: vmsnet.internals,comp.os.vms
Subject: Re: Intercepting RMS calls from the "outside"
Date: 4 Sep 1997 22:42:08 GMT
Organization: Dimensional Communications
Lines: 41
Message-ID: <5undg0$5kg$1@quasar.dimensional.com>
References: <340cb6d1.336133@news> <5ukc22$npo@wnnews1.netlink.net.nz>
NNTP-Posting-Host: flatland.dimensional.com
X-Newsreader: TIN [UNIX 1.3 unoff BETA 970424; sun4m SunOS 4.1.4]
Xref: news.mitre.org vmsnet.internals:5286 comp.os.vms:178040

In vmsnet.internals Don Stokes <don@toyunix.zl2tnm.gen.nz> wrote:
> In article <340cb6d1.336133@news>,
> Eric F. Richards <efricha@alumni.cs.colorado.edu> wrote:
> >What I'm thinking is that I could drop a piece of code in P1 space for
> >the given process which would redirect SYS$PUT through itself, and it
> >would give the appropriate CHME #xx instruction.  

> DCLCOMPLETE does pretty much exactly this, 

Thanks for the pointer!   I have a question, though...  Have some of the 
rules changed in VMS for system services?  You have the following code:

                SETIPL #IPL$_ASTDEL             ; Crank up IPL shields
                                                ; We don't wanna be killed now!
                $CRETVA_S inadr=pagetweak,acmode=#PSL$C_KERNEL
                blbc R0, 99$                    ; Create memory over the page
                movc3 #512, vectmp, @pagetweak  ; Restore vectors from copy
                $SETPRT_S inadr=pagetweak,acmode=#PSL$C_KERNEL,prot=#PRT$C_URKW
                blbc R0, 99$                    ; Set page protection to URKW
                $LKWSET_S inadr=pagetweak,acmode=#PSL$C_KERNEL
                blbc R0, 99$                    ; Lock the buggers down
                SETIPL #0                       ; Shields down

I thought calling a system service at IPL > 0 was a gross violation of
synchronization.  Is it?  What _is_ an acceptible way to handle this?

> >3)  (the real killer)  How would you do something like this on an
> >    Alpha?  (particularly item 1)

> If anyone can do this, I'd love them to make the change to DCLCOMPLETE.


VAXman?  This'd make my life easier, too....  (My goal is to be able to
intercept user-mode calls to $PUT and invoke user-mode code in the process's 
context.)

-- 
Eric F. Richards
efricha@alumni.cs.colorado.edu
"The weird part is that I can feel productive even when I'm doomed."
 - Dilbert