| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1921.1 | | STAR::KLEINSORGE | Fred Kleinsorge, OpenVMS Engineering | Fri Apr 11 1997 09:37 | 6 |
|
You have a crash dump, so analyze it. How can we guess what the
problem is? Odds are since it was in SYSMAN that some driver barfed in
it's initialization.
|
| 1921.2 | Facts, not guesswork | CSC32::D_BROWN | Dave Brown CSC-VSG/INTDRV | Fri Apr 11 1997 11:34 | 8 |
|
The problem is that I don't have the crash dump. I was hoping that
the foregoing might have looked familiar to someone. Having the crash
dump would be an added benefit; we're working on it. I take it
therefore that this is not a known problem.
I was not intending anyone to "guess" the problem. I was hoping for
more than that.
|
| 1921.3 | Question of Scale... | XDELTA::HOFFMAN | Steve, OpenVMS Engineering | Fri Apr 11 1997 11:47 | 6 |
| :I take it therefore that this is not a known problem.
We have probably thirty-five thousand "known" crashdump footprints
on VAX and Alpha, and there are certainly a large number more that
haven't been uploaded...
|
| 1921.4 | | STAR::KLEINSORGE | Fred Kleinsorge, OpenVMS Engineering | Fri Apr 11 1997 11:52 | 12 |
|
> **** OpenVMS (TM) Alpha Operating System V7.1 - BUGCHECK ****
> ** Code=000001CC: INVEXCEPTN,Exception while above ASTDEL
> ** Crash CPU: 00 Primary CPU: 00 Active CPUS: 00000001
> ** Current Process = STARTUP
> ** Image name = $3$DKA300:[sys0.syscommon.][SYSEXE]SYSMAN.EXE;1
> **** Starting selective memory dump at (date time)...
>
I guess this is what mislead me ;-)
|
| 1921.5 | SYS$GQDDRIVER's doing it | CSC32::D_BROWN | Dave Brown CSC-VSG/INTDRV | Fri Apr 11 1997 13:31 | 13 |
|
Still don't have the dump but we've determined that its
SYS$GQDDRIVER which crashes the system. Finally got the system booted
far enough to have the customer look at the dump. Renaming
SYS$GQDDRIVER permits the system to boot and the installation to
complete.
He's got two GQA devices, an ATI Mach64-CX and a ZLXp-E (TGA). The
crash occurs at SYS$GQDDRIVER_NPRO+06CA4 for which we have no listings.
Is there a way I can get V7.1 SYS$GQDDRIVER listings?
|
| 1921.6 | | STAR::KLEINSORGE | Fred Kleinsorge, OpenVMS Engineering | Fri Apr 11 1997 13:59 | 13 |
|
Could you include the sho crash (or clue crash) output. It looks like
a crash in a common routine that allocates pages for special structures
- like the device information block.
This is so unlikely that either I am not looking at the right code, or
you have stale images on the disk. Please check your
SYS$LOADABLE_IMAGES area to make sure there are no drivers that someone
copied by hand into a sys$specific area and then did an upgrade to
V7.1. Also check SYS$LIBRARY while you are at it.
_Fred
|
| 1921.7 | Right-shift problem? | CSC32::D_BROWN | Dave Brown CSC-VSG/INTDRV | Fri Apr 11 1997 14:41 | 115 |
|
We have no spurious executibles in either SYS$LOADABLE_IMAGES or
SYS$LIBRARY. As the following information shows, we have some kind of a 4-bit
right shift problem afoot:
Crash Time: 11-APR-1997 10:10:47.25
Bugcheck Type: INVEXCEPTN, Exception while above ASTDEL
Node: SUSAN (Standalone)
CPU Type: AlphaStation 200 4/100
VMS Version: V7.1
Current Process: SYSTEM_3
Current Image: $3$DKA300:[SYS0.SYSCOMMON.][SYSEXE]SYSMAN.EXE;1
Failing PC: FFFFFFFF.802E6CA4 SYS$GQDDRIVER_NPRO+06CA4
Failing PS: 10000000.00000604
Module: SYS$GQDDRIVER
Offset: 0000ACA4
Boot Time: 11-APR-1997 10:08:09.00
System Uptime: 0 00:02:38.25
Crash/Primary CPU: 00/00
System/CPU Type: 0D02
Saved Processes: 7
Pagesize: 8 KByte (8192 bytes)
Physical Memory: 32 MByte (4096 PFNs, contiguous memory)
Dumpfile Pagelets: 57129 blocks
Dump Flags: olddump,writecomp,errlogcomp,dump_style
Dump Type: raw,selective
EXE$GL_FLAGS: poolpging,init,bugdump
Paging Files: 1 Pagefile and 1 Swapfile installed
Stack Pointers:
KSP = 00000000.7FFA1B28 ESP = 00000000.7FFA5AC0 SSP = 00000000.7FFAC100
USP = 00000000.7AFF5390
General Registers:
R0 = 00000000.00000000 R1 = 00000000.0000000C R2 = 00000000.7FFA1D10
R3 = FFFFFFFF.8103E5A0 R4 = 00000000.7FFA1B80 R5 = 00000000.7FFA1CF8
R6 = 00000000.7FFA1D40 R7 = 10000000.00000604 R8 = 00000000.00002011
R9 = FFFFFFFF.82000000 R10 = 00000000.00002001 R11 = 00000000.006A4610
R12 = 00000000.00000001 R13 = FFFFFFFF.810928C0 R14 = 00000000.00000000
R15 = 00000000.00000000 R16 = 00000000.000001CC R17 = 00000000.7FFA1B80
R18 = 00000000.7FFA1D40 R19 = 00000000.00000003 R20 = FFFFFFFD.FFE0AAF8
R21 = 00000000.001C11FF R22 = FFFFFFFF.826C0000 R23 = 00000000.0000000D
R24 = FFFFFFFF.81006C18 AI = 00000000.00000003 RA = FFFFFFFF.802E5110
PV = FFFFFFFF.810944C8 R28 = FFFFFFFF.802E50E4 FP = 00000000.7FFA1D90
PC = FFFFFFFF.80069158 PS = 28000000.00000604
Exception Frame:
R2 = FFFFFFFF.81094010 R3 = FFFFFFFF.8118BB80 R4 = FFFFFFFF.81147200
R5 = 00000000.00002031 R6 = FFFFFFFF.82000000 R7 = 00000000.003FFC00
PC = FFFFFFFF.802E6CA4 PS = 10000000.00000604
Signal Array: 64-bit Signal Array:
Arg Count = 00000005 Arg Count = 00000005
Condition = 0000000C Condition = 00000000.0000000C
Argument #2 = 00010000 Argument #2 = 00000000.00010000
Argument #3 = 07FF9418 Argument #3 = 00000000.07FF9418
Argument #4 = 802E6CA4 Argument #4 = FFFFFFFF.802E6CA4
Argument #5 = 00000604 Argument #5 = 10000000.00000604
Mechanism Array:
Arguments = 0000002C Establisher FP = 00000000.7FFA1D90
Flags = 00000000 Exception FP = 00000000.7FFA1D40
Depth = FFFFFFFD Signal Array = 00000000.7FFA1CF8
Handler Data = FFFFFFFF.82000000 Signal64 Array = 00000000.7FFA1D10
R0 = 00000000.00000001 R1 = 00000000.07FF9418 R16 = 00000000.00000000
R17 = 00000000.07FF9400 R18 = 00000000.003FFCA0 R19 = 00000000.00000003
R20 = FFFFFFFD.FFE0AAF8 R21 = 00000000.001C11FF R22 = FFFFFFFF.826C0000
R23 = 00000000.0000000D R24 = FFFFFFFF.81006C18 R25 = 00000000.00000003
R26 = FFFFFFFF.802E5110 R27 = FFFFFFFF.810944C8 R28 = FFFFFFFF.802E50E4
System Registers:
Page Table Base Register (PTBR) 00000000.00000DB6
Processor Base Register (PRBR) FFFFFFFF.8110E000
Privileged Context Block Base (PCBB) 00000000.01B6A080
System Control Block Base (SCBB) 00000000.000001B8
Software Interrupt Summary Register (SISR) 00000000.00000000
Address Space Number (ASN) 00000000.0000003A
AST Summary / AST Enable (ASTSR_ASTEN) 00000000.0000000F
Floating-Point Enable (FEN) 00000000.00000000
Interrupt Priority Level (IPL) 00000000.00000006
Machine Check Error Summary (MCES) 00000000.00000008
Virtual Page Table Base Register (VPTB) FFFFFFFC.00000000
Failing Instruction:
SYS$GQDDRIVER_NPRO+06CA4: LDL R0,(R1)
Instruction Stream (last 20 instructions):
SYS$GQDDRIVER_NPRO+06C54: MB
SYS$GQDDRIVER_NPRO+06C58: BIS R31,#X01,R19
SYS$GQDDRIVER_NPRO+06C5C: SLL R19,R1,R1
SYS$GQDDRIVER_NPRO+06C60: SLL R18,R17,R17
SYS$GQDDRIVER_NPRO+06C64: AND R18,#X03,R18
SYS$GQDDRIVER_NPRO+06C68: BIS R17,R1,R1
SYS$GQDDRIVER_NPRO+06C6C: ADDL R16,R1,R1
SYS$GQDDRIVER_NPRO+06C70: S8ADDL R18,R31,R18
SYS$GQDDRIVER_NPRO+06C74: LDL R1,(R1)
SYS$GQDDRIVER_NPRO+06C78: ZAPNOT R1,#X0F,R28
SYS$GQDDRIVER_NPRO+06C7C: SRL R28,R18,R1
SYS$GQDDRIVER_NPRO+06C80: ZAPNOT R1,#X03,R0
SYS$GQDDRIVER_NPRO+06C84: RET R31,(R26)
SYS$GQDDRIVER_NPRO+06C88: SUBL R17,#X02,R1
SYS$GQDDRIVER_NPRO+06C8C: MB
SYS$GQDDRIVER_NPRO+06C90: BIS R31,#X03,R19
SYS$GQDDRIVER_NPRO+06C94: SLL R19,R1,R1
SYS$GQDDRIVER_NPRO+06C98: SLL R18,R17,R17
SYS$GQDDRIVER_NPRO+06C9C: BIS R17,R1,R1
SYS$GQDDRIVER_NPRO+06CA0: ADDL R16,R1,R1
SYS$GQDDRIVER_NPRO+06CA4: LDL R0,(R1)
SYS$GQDDRIVER_NPRO+06CA8: RET R31,(R26)
SYS$GQDDRIVER_NPRO+06CAC: BIS R31,R31,R31
SYS$GQDDRIVER_NPRO+06CB0: LDQ R1,#X0010(R27)
SYS$GQDDRIVER_NPRO+06CB4: LDL R1,(R1)
|
| 1921.8 | | STAR::KLEINSORGE | Fred Kleinsorge, OpenVMS Engineering | Fri Apr 11 1997 14:56 | 8 |
|
Found it. It is trying to do a 32-bit read from ISA space... and
something is very broken. What rev of the firmware are you running?
Can you give me the call frames so I can see what is attempting to be
read?
|
| 1921.9 | On it's way | CSC32::D_BROWN | Dave Brown CSC-VSG/INTDRV | Fri Apr 11 1997 15:21 | 15 |
|
Fred,
The firmware is as follows:
SRM 6.3-4
PAL 5.56-2
I don't have the call frames but I have mailed to you everything I
have received from the customer so far including the stack. I can have
him E-mail me some more if needed. He's sending in the dump on a tape.
Thanks,
Dave
|
| 1921.10 | Try a released version V7.1! | WRKSYS::BROOKS | | Fri Apr 11 1997 15:33 | 6 |
| V7.1 is this an official released version that you used. There was an
inernal version V7.1 that would hang. It is a problem we have seen
from several Customers. Some prereleased versions were sent to some
Customers. Please try an official version. It should have a date of
March or April--before this there were problems--Good Luck!
Art Brooks
|
| 1921.11 | | STAR::KLEINSORGE | Fred Kleinsorge, OpenVMS Engineering | Fri Apr 11 1997 15:36 | 19 |
|
Just to save me time, do you know which firmware rev (i.e. 3.7, 3.8,
3.9)...
In any case. It looks like an attempt to read a 32-bit register in the
memory mapped space at the end of the frame buffer. That is, the port
address is 3FFCA0, and the ATI has it's registers memory mapped at the
4mb address - 1024... or offset A0 into it... which is register 28,
which is BUS_CNTL, which is the first address read during init, after
the frame buffer is mapped... BUT the base address being passed in is
zero, which indicates that the driver *failed* to map the PCI address
space in sparse mode. Which means typically means that the init of the
card is garbage.
Of course, we don't support the CX card *and* the TGA or TGA2 in the
same box at the same time... so maybe that is the problem?
|
| 1921.12 | | STAR::KLEINSORGE | Fred Kleinsorge, OpenVMS Engineering | Fri Apr 11 1997 15:45 | 4 |
|
We'll take this offline and work the problem. Thanks.
|
| 1921.13 | 3.8 Firmware CD | CSC32::D_BROWN | Dave Brown CSC-VSG/INTDRV | Fri Apr 11 1997 15:49 | 18 |
|
Re: .10 - V7.1 SSB right off CD
Re: .11 - 3.8 CD
>>Of course, we don't support the CX card *and* the TGA or TGA2 in the
>>same box at the same time... so maybe that is the problem?
Customer's reaction to this was "*&*^#%, that's the way you guys
shipped me the system!!"
Is indeed the combination of a ATI Mach64-CX and a ZLXp-E (TGA)
on the same PCI (or in the same box) illegal under V7.1?
Thanks,
Dave
|
| 1921.14 | | STAR::KLEINSORGE | Fred Kleinsorge, OpenVMS Engineering | Fri Apr 11 1997 15:55 | 14 |
| It's an illegal configuration as far as I know... I know I don't
remember it ever being qual'ed or tested.
I will certainly take a look at it, and see if I can figure out what's
going wrong (I'll try plugging in a TGA and a CX card to my mustang).
If you get the dump up someplace I can set host to, I'll give you more
detail on what is out of whack.
VMS has always said that the commodity graphics cards would only work
multi-head in specific configurations. Usually the TGA combination is
benign because it functions purely in PCI space and doesn't muck with
VGA space. But it's certainly possible that something is getting
stepped on by one of the card inits.
|
| 1921.15 | | STAR::KLEINSORGE | Fred Kleinsorge, OpenVMS Engineering | Fri Apr 11 1997 16:46 | 6 |
|
Here's a question for the AlphaStation folks... what would a customer
have to do to actually have such a system manufactured? Was the
ZLXp-E1 substituted for a Mach64-CX when the CX went to EOL?
|
| 1921.16 | It shipped that way | CSC32::D_BROWN | Dave Brown CSC-VSG/INTDRV | Fri Apr 11 1997 17:47 | 8 |
|
The system shipped from Digital as DEC#96018492S; shipping papers
dated 3-October-1995. The customer said he got it with some type of
promotional deal. The system option was an 82-PB411-EB (containing I
beleive a Mach64-CX) with an optional ZLXp-E1; factory configured.
I've instructed him to remove the Mach64-CX
|
| 1921.17 | product management has been notified | WRKSYS::HOUSE | Kenny House, Workstations Engineering | Sun Apr 13 1997 09:06 | 19 |
| re .15 - "what would a customer have to do to actually have such a
system manufactured?" ...
I've extracted this note string (through reply .16) and sent it along
to Product Management with the following preface.
A customer got an AlphaStation 200 4/100 with both a ATI
Mach64-CX and a ZLXp-E (TGA). This combination was apparently
never tested, and indeed seems to have caused his system to
crash.
The question has been raised (by the customer and by the folks
who had to deal with the problem): How did this system get out
the door? Shouldn't something in the order process or in
manufacturing have caught it?
-- Kenny House
|