Mysterious BUGCHECK while booting 7.3-1 on DEC2000 - DEC

This is a discussion on Mysterious BUGCHECK while booting 7.3-1 on DEC2000 - DEC ; Hiya, Just when I thought things were all right, I now having BUGCHECKs while trying to boot VMS on my lovely Jensen. Backstory: Yesterday bought an EISA S3 928 board, and while it even worked, it displayed in preposterous resolution ...

+ Reply to Thread
Results 1 to 14 of 14

Thread: Mysterious BUGCHECK while booting 7.3-1 on DEC2000

  1. Mysterious BUGCHECK while booting 7.3-1 on DEC2000

    Hiya,

    Just when I thought things were all right, I now having BUGCHECKs
    while trying to boot VMS on my lovely Jensen.

    Backstory: Yesterday bought an EISA S3 928 board, and while it even
    worked, it displayed in preposterous resolution none of my monitors
    couldn't handle. So I but the ISA board back, ECUed the machine... and
    all hell broke loose.

    Bugcheck Type: INVEXCEPTN, Exception while above ASTDEL
    No process, active image nor module. Offset: 00000001(!)
    With booting with 0,1000 it was revealed that it happens right after
    "entering SWAPPER main loop".

    So I put back EISA board and the system mysteriously recovered and was
    booting... right until today morning- when it started crashing again.

    In desperation I started plugging out boards, to get system working
    only after I pulled out VGA and started with serial console. So that's
    the solution, but does anyone have a clue what does the machine
    actually not like? Did anyone encounter such behaviour before?

    Rambo


  2. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000

    Rambo writes:

    >Bugcheck Type: INVEXCEPTN, Exception while above ASTDEL
    >No process, active image nor module. Offset: 00000001(!)
    >With booting with 0,1000 it was revealed that it happens right after
    >"entering SWAPPER main loop".


    It's likely a crash in a driver or other system code which runs at high
    IPL (well, we already know its at an elevated IPL). The concept of the
    "active process" or "active image" really isn't valid at such IPLs since
    things such as timer or hardware interrupts can happen at any time, while
    any process (or none) are active. You'd need someone who knows how to read
    VMS crash dumps to really find what's going on, there's not enough info
    there. My guess is hardware, something broke when removing or inserting a
    card, or one is not seated properly.




  3. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000


    "Rambo" wrote in message
    news:1176826788.840054.194790@n59g2000hsh.googlegr oups.com...
    > Hiya,
    >
    > Just when I thought things were all right, I now having BUGCHECKs
    > while trying to boot VMS on my lovely Jensen.
    >
    > Backstory: Yesterday bought an EISA S3 928 board, and while it even
    > worked, it displayed in preposterous resolution none of my monitors
    > couldn't handle. So I but the ISA board back, ECUed the machine... and
    > all hell broke loose.
    >


    There is no graphics server for the S3 928

    > Bugcheck Type: INVEXCEPTN, Exception while above ASTDEL
    > No process, active image nor module. Offset: 00000001(!)
    > With booting with 0,1000 it was revealed that it happens right after
    > "entering SWAPPER main loop".
    >
    > So I put back EISA board and the system mysteriously recovered and was
    > booting... right until today morning- when it started crashing again.
    >
    > In desperation I started plugging out boards, to get system working
    > only after I pulled out VGA and started with serial console. So that's
    > the solution, but does anyone have a clue what does the machine
    > actually not like? Did anyone encounter such behaviour before?
    >
    > Rambo
    >


    Well, first you need to look at the crash dump. Try SHOW CRASH for
    starters. SHO STACK. SHO CALL/SUMM.




  4. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000

    > Well, first you need to look at the crash dump. Try SHOW CRASH for
    > starters. SHO STACK. SHO CALL/SUMM.


    I copied and wrote all to file, if anyone cares to decipher it ...
    http://rambo.id.uw.edu.pl/dump


  5. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000


    "Rambo" wrote in message
    news:1177154086.908849.218370@y5g2000hsa.googlegro ups.com...
    >> Well, first you need to look at the crash dump. Try SHOW CRASH for
    >> starters. SHO STACK. SHO CALL/SUMM.

    >
    > I copied and wrote all to file, if anyone cares to decipher it ...
    > http://rambo.id.uw.edu.pl/dump
    >


    Well, I can tell you what happened, but can only speculate on why.

    The crash occured in the interrupt dispatch code trying to service a console
    interrupt. A critical data structure - the sysgen configuration datablock
    (SYSG) pointer in OPA$AR_CRB is not filled in. So the system crashed when
    it fetched a zero pointer and then tried to access the structure.

    The cell is filled in by OPDRIVER during initialization, and I'm going to
    take a wild guess as to what happened...

    On the Jensen, the firmware wasn't written by the guys in Marlboro, it was
    written by guys in England. Depending on how the console is set, it creates
    a structure in the HWRPB called the CTB (console terminal block). A system
    with workstation capabilities uses a WORKSTATION CTB type, a system with
    just a serial line uses a TERMINAL type.

    Originally, the thought was that depending on the way the console was setup,
    you would get different types - and that is how it works on the Laser and
    Jensen systems. But later on, the FW folks decided that they would always
    use the WORKSTATION type and simply set the "alternate console" bit to tell
    the system to use the serial port. But in some of the early systems you
    would get different types. The downside to this was that on the Jensen in
    particular - is you set the console to serial - graphics would not work at
    all - because the mouse & keyboard would not be configured.

    Now, there is another type of CTB - "GRAPHIC" (well, there are a couple
    others - but we would have halted if we saw them) - which I don't know the
    circumstances of why it might be used at all - perhaps if there was no mouse
    (or no keyboard) or some other funky decision by the FW... I think that the
    CTB was set to something other than workstation or terminal (i.e. graphic)
    and the cell was not filled in - and when the first interrupt came in....
    poof

    I would replace the unsupported S3 928 with a "supported" card. A Compaq
    QVISION for example - which is what the system originally shipped with.

    Note: V7.3-1 is no longer supported. I'm not even sure that the Jensen is
    still supported.




  6. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000

    On Sat, 21 Apr 2007 10:09:17 -0400, "FredK"
    wrote:

    >On the Jensen, the firmware wasn't written by the guys in Marlboro, it was
    >written by guys in England.


    Not correct, and not complete...

    The SRM console was written by the engineering team in Ayr, Scotland.
    Calling them "guys in England" is fighting words. :-)

    The ARC firmware (a.k.a. NT firmware or NT console) originated in the
    source code for the ARC console on a MIPS workstation written by
    Microsoft engineers, and then ported to the Jensen by the engineering
    team in DECwest, Bellevue, WA, USA.

    John

  7. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000

    > Well, I can tell you what happened, but can only speculate on why.
    >
    > The crash occured in the interrupt dispatch code trying to service a console
    > interrupt. A critical data structure - the sysgen configuration datablock
    > (SYSG) pointer in OPA$AR_CRB is not filled in. So the system crashed when
    > it fetched a zero pointer and then tried to access the structure.
    > The cell is filled in by OPDRIVER during initialization, and I'm going to
    > take a wild guess as to what happened...


    Makes sense: i did run AUTOGEN when the unsupported S3 card was inside
    (see my Mozilla memory post) and maybe that messed up OPADRIVER state.
    The catch is, that Jensen works only when I throw out the graphics,
    and crashes
    when I put one back (ironically, it was working perfectly well with
    that one before
    I tried S3).

    > I would replace the unsupported S3 928 with a "supported" card. A Compaq
    > QVISION for example - which is what the system originally shipped with.


    Yep, a 1024/E is coming my way, will try to run and fix it.

    > Note: V7.3-1 is no longer supported. I'm not even sure that the Jensen is
    > still supported.


    It's not, 7.3-1 was the last to support it officially- although few
    folks encourage
    me to go and try running 8.3 and see what happens... :-)


  8. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000


    "John DeRosa" wrote in message
    news:q37k23di6pskmlbdhuhu5a3cpt9pi8h9bo@4ax.com...
    > On Sat, 21 Apr 2007 10:09:17 -0400, "FredK"
    > wrote:
    >
    >>On the Jensen, the firmware wasn't written by the guys in Marlboro, it was
    >>written by guys in England.

    >
    > Not correct, and not complete...
    >
    > The SRM console was written by the engineering team in Ayr, Scotland.
    > Calling them "guys in England" is fighting words. :-)
    >


    Brain fart on my part :-) How are ya John? Long time...

    > The ARC firmware (a.k.a. NT firmware or NT console) originated in the
    > source code for the ARC console on a MIPS workstation written by
    > Microsoft engineers, and then ported to the Jensen by the engineering
    > team in DECwest, Bellevue, WA, USA.
    >


    Yup. I still have the "specs" and the bruises from working on Bob Supniks
    "Console Unification" team.




  9. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000


    "Rambo" wrote in message
    news:1177175236.046518.189840@p77g2000hsh.googlegr oups.com...
    >> Well, I can tell you what happened, but can only speculate on why.
    >>
    >> The crash occured in the interrupt dispatch code trying to service a
    >> console
    >> interrupt. A critical data structure - the sysgen configuration
    >> datablock
    >> (SYSG) pointer in OPA$AR_CRB is not filled in. So the system crashed
    >> when
    >> it fetched a zero pointer and then tried to access the structure.
    >> The cell is filled in by OPDRIVER during initialization, and I'm going to
    >> take a wild guess as to what happened...

    >
    > Makes sense: i did run AUTOGEN when the unsupported S3 card was inside
    > (see my Mozilla memory post) and maybe that messed up OPADRIVER state.
    > The catch is, that Jensen works only when I throw out the graphics,
    > and crashes
    > when I put one back (ironically, it was working perfectly well with
    > that one before
    > I tried S3).
    >
    >> I would replace the unsupported S3 928 with a "supported" card. A Compaq
    >> QVISION for example - which is what the system originally shipped with.

    >
    > Yep, a 1024/E is coming my way, will try to run and fix it.
    >


    One other thing to try is to run the ECU. The QVision is something that the
    FW will recognize even if it is fully reset with no ECU run. Anything else
    needs a valid ECU entry.

    Autogen probably isn't the culprit here... it is something in the
    configuration state of the console.

    As to your comment about running V8.3 - for the most part we never remove
    the code that supported the old systems - we just don't update it or support
    it. Providing you have enough memory, it might work fine. I've got V8.3 on
    my Tadpole.



  10. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000

    On Mon, 23 Apr 2007 08:06:17 -0400, "FredK"
    wrote:

    >
    >"John DeRosa" wrote in message
    >news:q37k23di6pskmlbdhuhu5a3cpt9pi8h9bo@4ax.com...
    >> On Sat, 21 Apr 2007 10:09:17 -0400, "FredK"
    >> wrote:
    >>
    >>>On the Jensen, the firmware wasn't written by the guys in Marlboro, it was
    >>>written by guys in England.

    >>
    >> Not correct, and not complete...
    >>
    >> The SRM console was written by the engineering team in Ayr, Scotland.
    >> Calling them "guys in England" is fighting words. :-)
    >>

    >
    >Brain fart on my part :-) How are ya John? Long time...


    Ya know, I saw your username but didn't know if you were _that_ Fred.
    :-) I'm good. What's with the dec.com email address? Is that for
    real?

    >
    >> The ARC firmware (a.k.a. NT firmware or NT console) originated in the
    >> source code for the ARC console on a MIPS workstation written by
    >> Microsoft engineers, and then ported to the Jensen by the engineering
    >> team in DECwest, Bellevue, WA, USA.
    >>

    >
    >Yup. I still have the "specs" and the bruises from working on Bob Supniks
    >"Console Unification" team.


    Ah, memories.

    I remember a phone call from a shall-be-nameless individual, asking me
    if the ARC console was going to obey SRM Chapter 6. I said, "No."
    "Why not?" "Because I don't want to. And Dave Cutler doesn't want
    to." Or words to that effect.

    The other person found the remainder of the conversation to be equally
    unsatisfying.

    Boy, was that fun. I still get a smile on my face.

  11. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000


    "John DeRosa" wrote in message
    news:aiip23hg58606n55afv9oan2ivp6cao2l1@4ax.com...
    > On Mon, 23 Apr 2007 08:06:17 -0400, "FredK"
    > wrote:
    >>>

    >>
    >>Brain fart on my part :-) How are ya John? Long time...

    >
    > Ya know, I saw your username but didn't know if you were _that_ Fred.
    > :-) I'm good. What's with the dec.com email address? Is that for
    > real?
    >


    It's me. But the mail address is bogus to keep the spammers at bay.
    Getting anyone at hp is easy if you can spell their name... just first name
    dot lastname at hp dot com.


    >
    > I remember a phone call from a shall-be-nameless individual, asking me
    > if the ARC console was going to obey SRM Chapter 6. I said, "No."
    > "Why not?" "Because I don't want to. And Dave Cutler doesn't want
    > to." Or words to that effect.
    >
    > The other person found the remainder of the conversation to be equally
    > unsatisfying.
    >
    > Boy, was that fun. I still get a smile on my face.


    Let me take a guess... Stephen?




  12. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000

    > >> QVISION for example - which is what the system originally shipped with.
    >
    > > Yep, a 1024/E is coming my way, will try to run and fix it.

    >
    > One other thing to try is to run the ECU. The QVision is something that the
    > FW will recognize even if it is fully reset with no ECU run. Anything else
    > needs a valid ECU entry.


    It arrived, I put it in and behold! System switched into nice
    "framebuffer"
    OpenVMS Operator Console, and X started.

    http://rambo.id.uw.edu.pl/pic.php,DE...le.jpg,big,0,0
    http://rambo.id.uw.edu.pl/pic.php,DE...e2.jpg,big,0,0

    Rambo


  13. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000


    "Rambo" wrote in message
    news:1177366783.012046.92110@p77g2000hsh.googlegro ups.com...
    >> >> QVISION for example - which is what the system originally shipped
    >> >> with.

    >>
    >> > Yep, a 1024/E is coming my way, will try to run and fix it.

    >>
    >> One other thing to try is to run the ECU. The QVision is something that
    >> the
    >> FW will recognize even if it is fully reset with no ECU run. Anything
    >> else
    >> needs a valid ECU entry.

    >
    > It arrived, I put it in and behold! System switched into nice
    > "framebuffer"
    > OpenVMS Operator Console, and X started.
    >
    > http://rambo.id.uw.edu.pl/pic.php,DE...le.jpg,big,0,0
    > http://rambo.id.uw.edu.pl/pic.php,DE...e2.jpg,big,0,0
    >
    > Rambo
    >


    My compliments on the creative use of the SUN systems as a monitor stand!
    :-)




  14. Re: Mysterious BUGCHECK while booting 7.3-1 on DEC2000

    > My compliments on the creative use of the SUN systems as a monitor stand!
    > :-)


    Well, in defense of Suns , I like them too- basically I'm not really
    fond of
    "superiority of Easter over X-mas" discussions :-D :-D, though
    addmittably
    they score rather low in my "like it" list .

    Rambo



+ Reply to Thread