[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference hydra::axp-developer

Title:Alpha Developer Support
Notice:[email protected], 800-332-4786
Moderator:HYDRA::SYSTEM
Created:Mon Jun 06 1994
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:3722
Total number of notes:11359

3524.0. "Symark Software" by HYDRA::VANORDEN () Tue Apr 22 1997 15:48

    Company Name :  Symark Software
    Contact Name :  Marty Kuhrt
    Phone        :  1-800-234-9072 x104
    Fax          :  
    Email        :  [email protected]
    Date/Time in :  22-APR-1997 14:45:39
    Entered by   :  Donna Van Orden
    SPE center   :  MRO

    Category     :  VMS
    OS Version   :  6.2-1h3
    System H/W   :  8200


    Brief Description of Problem:
    -----------------------------


Upgrading his customer to 6.2-1h3 to support 8200.  Failing on reboot:
initodb-%sys$dkdriver.exe load iosb status 00002398
(which means image version mismatch).


He wonders what is new in 6.2-1h3 that could possibility cause the image
mismatch.  He also has a call in to the CSC, but was a little concerned 
because the person who took his call did not seem to know who to assign
the call to.


Donna VO


T.RTitleUserPersonal
Name
DateLines
3524.1HYDRA::VANORDENTue Apr 22 1997 16:5845
    
    
    
    As a courtesy I looked for additional information on 6.2-1h3 and
    DKdriver changes.  The following note from vmsnotes indicated they
    might have some problems considering they have a caching product:
    
    
    
    note 33.8
    
    
    In VMS 6.2 (and continuing thru 7.1) dkdriver has looped back and
        grabbed IRPs off its queue without calling any standard entry.
        
        This makes it basically impossible for 3rd party cache products to
        work on 6.2 thru 7.1 on alpha (problem does not occur on Vax).
        
        I have a driver workaround that's totally unsupported but seems
        to work ok though. Customers nevertheless need to be sure they
        have a cacher that works right.
        
        VIOC is somewhat privileged in that there's special purpose code
        to support it, so I *THINK* it is not affected by this. The others
        have to trap I/O at some standard entry, in general.
        
        Also, the dkdriver listings were censored off 6.2 (that's undone
        now, as I understand) at least on alpha, so 3rd parties would have
        a hard time finding this.
        
        I have been in touch with Raxco and with Rick Cadruvi who wrote a
        couple of these to let them know of the issues & fixups.
        
        However an element of caution is needed in suggesting these
    products
        in view of this. (I only realized the scope of the problem quite
        recently.)
        
        The fixup is pretty rude 'n' crude; instead of looping back in
        dkdriver after remque'ing an IRP, fork first and let the fork
        routine do it and call the start_io entry. I don't believe any
        other disk driver is affected at the moment. This fix does howver
        prevent kernel stack overflows.
        
        Glenn Everhart
3524.2HYDRA::VANORDENTue Apr 22 1997 17:409
    
    I sent a copy of the release notes to Marty and asked for confirmation
    that his product worked on 6.2 (the note I placed in the previous reply 
    means there is potential for problems on this version...)
    
    I've fulfilled my commitment of sending the release notes.  He is
    working with the CSC now...
    
    Donna
3524.3HYDRA::VANORDENThu Apr 24 1997 11:4334
From:	HYDRA::AXPDEVELOPER "[email protected]" 24-APR-1997 03:44:00.24
To:	vanorden
CC:	AXPDEVELOPER
Subj:	FWD: RE: Release Notes for v6.2-1h3

From:	US6RMC::"[email protected]" "Marty Kuhrt"   23-APR-1997 21:52:15.00
To:	hydra::axpdeveloper
CC:	
Subj:	RE: Release Notes for v6.2-1h3

Hello,

I found someone in the VMS internals group to give me a hand 
with the V6.2-1H3 problem.  Looks likes the problem is resolved.

Thanks for your help.
Marty

% ====== Internet headers and postmarks (see DECWRL::GATEWAY.DOC) ======
% Received: from mail13.digital.com by us6rmc.mro.dec.com (5.65/rmc-17Jan97) id 
AA22236; Wed, 23 Apr 97 21:45:42 -0400
% Received: from netcomsv.netcom.com by mail13.digital.com (8.7.5/UNX 
1.5/1.0/WV) id VAA31838; Wed, 23 Apr 1997 21:42:30 -0400 (EDT)
% Received: from symark.UUCP by netcomsv.netcom.com with UUCP (8.6.12/SMI-4.1) 
id SAA03244; Wed, 23 Apr 1997 18:38:15 -0700
% Received: by symark.com (DECUS UUCP /2.0/2.0/2.0/); Wed, 23 Apr 97 17:18:06 
PDT
% Date: Wed, 23 Apr 97 17:18:06 PDT
% Message-Id: <[email protected]>
% From: "Marty Kuhrt" <[email protected]>
% Subject: RE: Release Notes for v6.2-1h3
% To: hydra::axpdeveloper
% X-Vms-Mail-To: UUCP%"[email protected]"

3524.4HYDRA::VANORDENThu Apr 24 1997 11:5726
    From:   HYDRA::AXPDEVELOPER "[email protected]"
    To:     US6RMC::"[email protected]"
    CC:     AXPDEVELOPER
    Subj:   RE: Release Notes for v6.2-1h3
    
    
    Marty,
    
    Thank you for taking the time to send me confirmation that this problem
    has been resolved.
    
    FYI:  I did find mention that in VMS 6.2 through 7.1 the dkdriver
    looped back
    and grabbed IRPs off its queue without calling any standard entry,
    which may
    affect cache products on Alpha.
    
    Remembering you mentioned your product was a cache product but not
    knowing
    anything else about it, I wanted to bring this to your attention.  Let
    me
    know if we need to pursue this further.  
    
    Donna
    Alpha Developer Support
    
3524.5another reply from partnerHYDRA::DONSBACHJeff Donsbach, Software Partner Engineering, DTN 297-6862Thu Apr 24 1997 16:2137
    Date:	24-APR-1997 15:13:23.69
    From:	US6RMC::"[email protected]" "Marty Kuhrt"
    Subj:	RE: Release Notes for v6.2-1h3
    To:	hydra::axpdeveloper
    CC:	
    --------
    >Thank you for taking the time to send me confirmation that this
    problem
    >has been resolved.
    >
    >FYI:  I did find mention that in VMS 6.2 through 7.1 the dkdriver
    looped back
    >and grabbed IRPs off its queue without calling any standard entry,
    which may
    >affect cache products on Alpha.
    
    Yes, we were made aware of this.  We been in contact with the DKDRIVER
    team
    and have received new drivers for those three versions.  The new
    drivers
    are installed with the kitinstal of our caching product.
    
    The problem we were running into was that a driver that worked on a 
    V6.2 Alpha wasn't working on a V6.2-1H3 machine.  We received a new
    driver yesterday and I integrated into the product.
    
    >Remembering you mentioned your product was a cache product but not
    knowing
    >anything else about it, I wanted to bring this to your attention.  Let
    me
    >know if we need to pursue this further.  
    
    I think we have it under control now.
    
    Thanks,
    Marty