Please help - Free SCO OpenDesktop installation fails miserably!!

Please help - Free SCO OpenDesktop installation fails miserably!!

Post by JOH » Mon, 14 Oct 1996 04:00:00



Hello SCO users/professionals:

My computer invariably crashes at around 78% through the installation progress.
This is the error message:

Unexpected trap in kernel mode:

cr0  0x800100113  cr2  0x701A2004  cr3  0x000020000  tlb  0x00000000
ss   0x00000000   uesp 0x0002D55E  efl  0x00010246   ipl  0x00000000
cs   0x00000158   eip  0xF0052388  err  0x00000000   trap 0x0000000E
eax  0x0005AABC   ecx  0x00000002  edx  0x00000000   ebx  0x701A1F28
esp  0xE0000B9C   ebp  0xE0000BD8  esi  0xC3Ea40CC   edi  0x00000000
ds   0x00000160   es   0x00000160  fs   0x00000000   gs   0x00000000
cpu  0x00000001   Status: Executing ccs script ../SCO/nfs/2.0.0Dp/cntl/ccs

PANIC:  k_trap - Kernel mode trap type 0x0000000E

Dump not completed
   **  Safe to Power Off  **
           -or-
**  Press Any Key to Reboot  **

The following is my computer configuration:

Dell XPS Pentium 133
64 MB EDO 60ns
Adaptec 2940 UltraFast PCI adapter
Hitachi 4x SCSI CDROM at id 2
Quantum Firefall 1080S at id 0
3COM Etherlink III 3c509
PS/2 keyboard
Number Nine Imagine 128 Video Card
Mouse Systems PC Mouse II (optical mouse) on com1

I specify the following options in install:
use SCSI CDROM at id 2
install SCO gateway for Netware software
install "" documentation
manually specify the 3COM card (did not detect it)
use whole disk for SCO
specified IP address, netmask, etc.
watched it work until the above failure

Any help or suggestions would be appreciated.  Thanks!
John

 
 
 

Please help - Free SCO OpenDesktop installation fails miserably!!

Post by Bela Lubki » Wed, 16 Oct 1996 04:00:00



> My computer invariably crashes at around 78% through the installation progress.
> This is the error message:

> Unexpected trap in kernel mode:

> cr0  0x800100113  cr2  0x701A2004  cr3  0x000020000  tlb  0x00000000
> ss   0x00000000   uesp 0x0002D55E  efl  0x00010246   ipl  0x00000000
> cs   0x00000158   eip  0xF0052388  err  0x00000000   trap 0x0000000E
> eax  0x0005AABC   ecx  0x00000002  edx  0x00000000   ebx  0x701A1F28
> esp  0xE0000B9C   ebp  0xE0000BD8  esi  0xC3Ea40CC   edi  0x00000000
> ds   0x00000160   es   0x00000160  fs   0x00000000   gs   0x00000000
> cpu  0x00000001   Status: Executing ccs script ../SCO/nfs/2.0.0Dp/cntl/ccs

> PANIC:  k_trap - Kernel mode trap type 0x0000000E

> Dump not completed
>    **  Safe to Power Off  **
>            -or-
> **  Press Any Key to Reboot  **

> The following is my computer configuration:

> Dell XPS Pentium 133
> 64 MB EDO 60ns
> Adaptec 2940 UltraFast PCI adapter
> Hitachi 4x SCSI CDROM at id 2
> Quantum Firefall 1080S at id 0
> 3COM Etherlink III 3c509
> PS/2 keyboard
> Number Nine Imagine 128 Video Card
> Mouse Systems PC Mouse II (optical mouse) on com1

> I specify the following options in install:
> use SCSI CDROM at id 2
> install SCO gateway for Netware software
> install "" documentation
> manually specify the 3COM card (did not detect it)
> use whole disk for SCO
> specified IP address, netmask, etc.
> watched it work until the above failure

Thanks for the reasonably complete description.  (Note: all of the hex
values in the register dump should be 8 digits.  You've got 9 digits in
a couple of them, which makes the rest also suspect -- but in this case
it looks like the important registers are correct.)

This panic is occurring at an impossible place.  By that I mean that
I've looked at the code in the install kernel at the %eip address you
show, and I do not see how it could be panicing.  The instruction which
is causing the panic is

  Sdskstrategy+0xec:              div     [ebx+0xdc]

You can see in the register dump that %ebx is 0x701a1f28, so %ebx+0xdc is
0x701a2004, which is exactly the address seen in %cr2.  %cr2 is the
register that shows the address of the illegal memory reference (and
0x701a2004 is certainly an illegal address in the context of
installation).  So everything checks out.  The problem is that one of
the immediate antecedent instructions is

  Sdskstrategy+0x68:              mov     eax,[ebx+0xdc]

%ebx does not change between these two instructions.  So if the
antecedent instruction succeeded, the following must also succeed.

This makes me think that the problem is caused by a hardware failure.
It's extremely unlikely that the value of a register would get corrupted
inside the CPU.  So let's instead imagine that an interrupt occurred
between those two instructions.  The interrupt routine saved all the
registers on the stack, did its business, restored the registers, and
continued.  Unfortunately, there was a cache or memory failure -- a
false value was read -- when the %ebx register was restored.

Conclusion: either you have a bad SIMM, or your motherboard cache is
bad.

Next step: try installing with the external cache disabled (leave the
internal cache enabled).  This will make it somewhat, but not
disastrously, slower.  If it still fails, you have a bad SIMM.  If it
succeeds, it's very likely that the problem was in the cache, because
without the external cache, we're working the memory even harder.

If you have a bad SIMM (i.e. install still fails with the external cache
disabled), see if you can remove half the memory.  (This depends on your
memory configuration: if two 32MB SIMMs, you may not be able to remove
one; if 4 16MB SIMMs, you can probably remove 2.)  Remove half, install,
if it fails, swap in the other half, try again.  Then swap only one, try
to isolate down to the single bad SIMM, then get it replaced.  [Yeah,
this is incredibly tedious :-(]

Quote:>Bela<

PS: until this problem is solved, write down and save every register
    dump -- being careful to get the numbers exactly right.  If you have
    other examples from earlier, forward them to me.  If all of the
    failures are identical (which seems highly unlikely from this
    example), tell me how many have occurred.  Let's take this part to
    direct email...

 
 
 

1. Redhat 5.2 Hard Drive installation failing miserably

I went through the process of downloading RH 5.2 and would like to install
it on my old hard drive. I followed the installation manual which said to
simply copy the base and RPMS directories and then boot with the boot disk.
When I do this the installation goes fine until it asks my where the files
are. I receive a message saying the drive doesn't appear to contain an
installation tree................Please HELP

Thank you
Quint

2. strtok

3. ConfigXF86 fails miserably - help!

4. Question: Separate compilation of Pascal in UNIX?

5. HELP: I want to use the network in DOS under SCO OpenDesktop?

6. A few questions??

7. HELP! FREE SCO Openserver installation problem

8. Shared filesystem for MacOS (Classic), Linux and NetBSD

9. Xconfigurator fails miserably

10. PROBLEM: Kernel compiles fail miserably...

11. upgrade to COL2.3 fails miserably

12. SuSE 7.3 installation hosed, reinstall failed; please help