z/VM 5.3 SYSTEM FAILURE HTT001
"\"Fish\" \(David B. Trout\)" <fish-6N/[email protected]> Sat, 27 Nov 2010 19:53:05 -0800
| Newsgroups | gmane.comp.emulators.hercules390.vm |
|---|---|
| Message-ID | <000001cb8eaf$c6b62d40$542287c0$@org> |
I'm having a problem but I can't tell at this early stage whether it is wit=
h
Hercules or z/VM 5.3 itself.
Whenever I attempt an IPL of a guest whose storage is defined greater[1]
than 4GB the system (z/VM itself) abends (crashes) with:
HCPDMP908I SYSTEM FAILURE ON CPU 0002, CODE - HTT001
There is no preceding Hercules error message[2], so I'm not sure what went
wrong.
Something else is weird too. After VM crashes and restarts itself, it issue=
s
the following error
HCPZCO6718I Using parm disk 1 on volume 530RES (device 0123).
HCPZCO6718I Parm disk resides on blocks 57200 through 229992.
The directory on volume 530RES at address 0123 has been brought online.
HCPERP6311I DASD 0215 AN OPERATION WAS TERMINATED BECAUSE A
HCPERP6311I CHANNEL PROGRAM CHECK OCCURRED
HCPERP6303I SENSE =3D INVALID
HCPERP6304I IRB =3D 08C0401F 7AC32870 00203600 00800000
HCPERP6305I USERID =3D SYSTEM
HCPERP2216I CHANNEL PATH ID =3D 02
Again, with no associated Hercules error message at all. I'm wondering, is
this is maybe VM's way of reporting "after the fact" what actually occurred
previously? I.e. is VM trying to tell me the reason it originally crashed
(HTT001) was because of a Channel Program Check? (and it's only now telling
me about it after it restarts?) Or maybe it doesn't bother clearing memory
during an abend restart, and some of its control storage is still damaged?
(but that seems like a stupid thing to do)
What is VM trying to tell me? Is it really THAT sick/broken?!
The messages manual has this to say about code HTT001:
HTT001
Explanation: CP fault in primary space caused by an
inappropriate reference to an address in
storage. This abend is usually the result
of a translation exception.
User response: Examine the dump and get the last two trace
entries on the failing processor. From the
second trace entry back, note the type of
exception, location of the failing instruction,
and the old PSW. Use the command:
BLOCK PFXPG 0 FIELD PFXTSKBK
to locate the TSKBK, then issue BLOCK TSKBK
on that address to find the register contents
at the time of the failure. With this information,
refer to z/VM: Diagnosis Guide for information
on gathering the additional documentation you
need to assist IBM in diagnosing the problem;
then contact your IBM support personnel.
I managed to capture the dump and have already done the "DUMPLOAD", so the
dump is now on MAINT's A disk[3]. I have not yet attempted to analyze it
using "VMDUMPTL" however. I thought I'd ask you guys for help first. :)
SOME QUESTIONS:
1. Is this a known problem? Has anyone else
experienced the same/similar behavior?
2. Does APAR VM64031 [4] describe my problem?
I'm thinking it *may*, but I need to analyze
the dump first.
3. If it turns out that I do need the PTF,
can someone download it for me please?
Apparently my IBM customer is not reg-
istered as being for a company that uses
IBM software (that's my guess) because
whenever I try to download it, it asks
me to add my customer number to my profile
and each time I try doing that it rejects
it as being not in their database! [5]
HELP!
Thanks. :)
(p.s. this is with Herc 3.07, but with some *extremely minor* fixes applied=
)
(p.p.s I'll try to do some more detailed Hercules PTT tracing to see if
anything obvious pops up.)
--=20
"Fish"=A0(David B. Trout)=20
[email protected]
[1] When I use 2000M it works fine. If I use 3000M, "mvs" abends very early
in its IPL with disabled wait 108A. If I use 5000M, z/VM always crashes. It
doesn't matter whether z/VM itself has 2000M or 3000M OR 5000M or even 6GB.
z/VM itself works fine no matter how much storage I give it -- *IFF* the VM
guest I'm trying to IPL is defined with 2GB or less of storage. But if the
VM guest has 2GB or more (especially > 4GB), then VM itself always crashes
no matter HOW much storage it itself has!
[2] But that may be because I'm using the Hercules 'OSTAILOR VM' setting. I
may need to try my tests again with a slightly different setting to see if =
I
can prod Herc into issuing a message. I'll let you know.
[3] After DUMPLOAD'ing it to MAINT's "A" disk, I then attached a tape drive
and did a TAPE DUMP of the file, so if anyone wants a copy to help me
analyze it, just ask. It's a Hercules .HET tape file and is only 11.9MB in
size.=20
[4]
http://www-01.ibm.com/support/docview.wss?rs=3D0&context=3DSWG00&dc=3DDB550=
&uid=3Dis
g1VM64031&loc=3Den_US&cs=3Dutf-8&lang=3D
(or, same but shorter: http://tinyurl.com/24v7xxn)
[5] My IBM account information is as follows:
-------------------------------------------------------------
Customer number : #######
Software customer number:=20
Company name : SOFTWARE DEV LABORATORIES
Country code : 000
Country name : United States
Company type : 0 -- Normal Customer
The Customer Service Administrator has also been registered
and can start registering other users for the company.
User name : David Trout
User IBM ID : [email protected]
User IBMLink ID: US######
-------------------------------------------------------------
Is the fact that my account is "Normal Customer" account (without any type
of "Software customer number") might be what the problem is? Are they
telling me I can't order a fix for their software because according to them
I haven't purchased any of their software and thus am not a "software
customer"?! This is so frustrating! Why is IBM's web site so fricking hard
to use?! (sigh!)
------------------------------------
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/H390-VM/
<*> Your email settings:
Individual Email | Traditional
<*> To change settings online go to:
http://groups.yahoo.com/group/H390-VM/join
(Yahoo! ID required)
<*> To change settings via email:
[email protected]=20
[email protected]
<*> To unsubscribe from this group, send an email to:
[email protected]
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/