Operating System - OpenVMS
1748117 Members
3488 Online
108758 Solutions
New Discussion юеВ

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

 
SOLVED
Go to solution
John Travell
Valued Contributor

V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

Folks, I seem to have found a brand new way to crash VMS...
In truth it revolves around what is most probably unsupported hardware, but a bugcheck does seem an ungraceful way to say this thing is not going to work.
The background is that I am trying to find out which, if any, of the CD writers I have will work in my XP1000.
The hardware is an LG DVDRAM drive, model GSA-4163B, connected as DQA0, reported correctly both by the console and $show device/full', but not by CDRECORD.
All it takes to crash a fully patched V8.2 system is :

$ mount/system dqa0: 'label'
any CD will do, I have tried several...

I cannot rule out the possibility that the drive is faulty, but it works in a PC...

See the attachment. I also have three dumps available, but since it appears to be a reproducible problem, I do not expect anyone will need them.

Does anyone have a list of the devices supported either by the CDRECORD in V8.2, or the new 'copy/recordable'? in V8.3. It clearly does not match the list for DVDWRITE, and in any case, I do not have the budget to pay for the commercial product.

I do NOT have a support agreement with HP, so cannot do anything about escalating it myself.
JT:
12 REPLIES 12
Jeff Chisholm
Valued Contributor

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

Hi John,
It's been awhile. Drop me an email, I'll need some details. Jeff
le plus ca change...
Heuser-Hofmann
Frequent Advisor

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

I saw similar crashes on I64/V8.3 due to the (old!!) cdrecord program.

Does the crash occur without the cdrecord call?

Eberhard
John Travell
Valued Contributor

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

Jeff, I haven't got your email address, mine is in my profile.

Eberhard, I am now sure that CDRECORD is not relevant to the crash, the third bugcheck I triggered I just tried to mount the disk without first invoking CDRECORD.
The reason for including the comment about CDRECORD was to illustrate that something could not make up its mind about what the drive really is.

Apart from a couple of irrelevancies, my complete DCL recall buffer from first login after reboot was:
$ sho dev dqa0: /full
$ mount /over=id $1$DKA200:
$ dir $1$DKA200:[dumps] (to verify that I had the correct directory name)
$ ana/cr sys$system: (to copy the previous dump to safety)
$ dism $1$DKA200:
$ mount/sys dqa0: alpha083

I have not taken a detailed look at all three crashes yet, but they do vary slightly in detail. All are from the same PC, SYS$DQDRIVER+06308. The first was a PGFIPLHI, from a partially patched V8.2 disk, with a failure VA different to the later crashes.
I have a newly built V8.2 disk with all of the latest patches installed, so I booted off that, and triggered both the second and third crashes, both INVEXCEPTN with the same failure VA.

For the time being I have suspended my search for a VMS supported CD writer among my limited supply of these things, and plan to upgrade to V8.3 very soon. I will then see if the crash still occurs, and report back the details. However, I believe that all anyone needs to do to reproduce it is to fit a similar drive into an XP1000, or possibly any Alpha using the same IDE chip (Cypress 82C693 IDE), then try to $ mount/sys DQA0:...

JT:
Volker Halle
Honored Contributor

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

John,

crashes with very similar footprints (INVEXCPETN and PGFIPLHI with the same instruction seen in your crash) have been seen during internal OpenVMS testing in the past and have been attributed to 'unsupported drive'...

Volker.
Hoff
Honored Contributor

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

I'd first ensure that the OpenVMS Alpha V8.2 DRIVER ECO kit is loaded here.

The Plextor PX-716A is one of the unsupported drives that has been discussed by and found to work by various folks, both on a varity of AlphaServer and AlphaStation systems in general, and bootable on the older SRM found on this particular AlphaStation XP1000 box. This per various newsgroup discussions. And yes, this drive is not supported by HP, so YMMV.

I've sent John some pointers and related details off-line.

Stephen Hoffman
HoffmanLabs
atul sardana
Frequent Advisor

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

Dear John,

As per your problem and attached crash dumps your system CPU Type:COMPAQ Professional Workstation XP1000 with VMS Version:V8.2 does not supported $1$dqa0: (DVDRAM GSA-4163B)CDROM because driver of this cdrom is not in server firmware.

You please replace it with Hp combo drive for rectification this problem

thanks,
Atul Sardana
I love VMS
Volker Halle
Honored Contributor
Solution

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

John,

the crash happens in a routine, which moves a sector worth of data from the drive into memory on a READ operation. The data is moved word-by-word, R6 is pointing to the buffer in memory, R4 is the no. of words already moved. R5 seems to be the total no. of words to be moved - in your crash, this amounts to 124. blocks.

The failing VA always seems to be on an Alpha page boundary. This looks like the buffer address has not been set up correctly by the caller. As R4 is > 0, the data read/copied so far should be in memory preceeding the failing VA.

Volker.
Jur van der Burg
Respected Contributor

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

>As per your problem and attached crash dumps >your system CPU Type:COMPAQ Professional >Workstation XP1000 with VMS Version:V8.2 does >not supported $1$dqa0: (DVDRAM GSA-4163B)CDROM >because driver of this cdrom is not in server >firmware.

That looks like you have no clue. This is a driver issue, it has nothing to do with firmware.

Having said that, I could crash the system in a similar way by creating an invalid iso cd rom. Is there an acp in the dump? what's the data size on the disk? If it's more than 2Gb it's an invalid dvd.

Jur.

John Travell
Valued Contributor

Re: V8.2 INVEXCEPTN Bugcheck from SYS$DQDRIVER+06308

I concur about the 'have no clue'... Although device firmware has now been proven to be relevant.
I have two of these LG GSA-4163B drives, one with A105 firmware, the other originally with A101 firmware. With the A101 firmware VMS returned ...
$ mount/sys dqa0: vmsdoc082
%MOUNT-F-DRVERR, fatal drive error
However, with the A105 (and A106) firmware the $ mount/sys led to the crash. I confirmed the firmware made the difference by updating the drive with A101 firmware to A105, and then testing with that device. There is no possibility of my having mistaken which is which, one has a black bezel, the other white...

To be honest, I have not yet done any detailed work on this crash yet, as the planned CD write was a pre-cursor to an upgrade to V8.3. The CD I wanted to burn was a copy of a borrowed 8.3 distro CD. It seems no-one noticed the volume label on the mount/sys that crashed the system...

I am now on a clean build of V8.3, fully patched, and have savesets for fully patched clean builds for every version from V7.3 onwards, and a spare disk to restore them to.
I expect to be able to spend some serious time on this issue during the next week.
I will also be awarding points as soon as I have a real picture of how much value each response has given me.
JT: