[128] in linux-scsi channel archive
Re: NCR53c810 SCSI problem in Slackware 2.2
daemon@ATHENA.MIT.EDU (Camm Maguire)
Thu Mar 30 19:02:49 1995
Date: Thu, 30 Mar 95 16:20:54 -0500
From: camm@enhanced.com (Camm Maguire)
To: linux-scsi@vger.rutgers.edu
Greetings all, and thank you for your replies!
Thanks to your pointers, the problem does seem to lie somewhere in the
motherboard/BIOS/PCI bus system. Please tell me if the following
occurrences mean anything to you. I need to either pinpoint this
error and fix it promptly, or return my machine, which I hesitate to
do as my vendor is far away.
In addition, I would greatly appreciate knowing if anyone has Linux
1.2 successfully working on a Pentium PCI system with SiS Chipsset and
NCRc810 SCSI controller, so that I can at least tell my vendor that
the problem is definitely not software related.
1) One respondent enquired as to whether this was the only error I was
getting. The answer is no, many different kernel panics and system
hangs have occurred, but are usually somewhat regular given any
particular BIOS (Phoenix) setting. This respondent suggested using
the following command as a foolproof test:
dd if=/dev/sda of=/dev/null bs=1048576 count=500
This indeed gives a kernel panic with the default settings, but
proceeds normally if the external cache is disabled. However, the
rest of the installation fails due to a segmentation fault or system
hang either in the mkswap step or in the process of
uncompressing/installing the software disk sets. Even the slow format
of the native Linux partition seems to proceed normally if the
external cache is disabled. Other errors seen include a kernel panic
in the allocate_device routine (197e14) and a "Kernel panic: VFS: LRU
block list corrupted" error message.
2) If I run mkswap before running the above dd command, I get a long
series of "free_one_pmd: bad directory entry ###" errors, and the
mkswap is killed. I can then run mkswap again with no apparent
problems.
3) For the purpose of talking to my vendor's technician, I've looked
for a DOS program symptomatic of the problems. The clearest one I
found was a little shareware program called rt302 on the simtel
archive, which exhaustively tests system memory. This program works
normally on two 486 PC's at work, but fails with a runtime error of
"integer divide by 0" on my system. My understanding was that pentium
systems were supposed to by binary compatible with 486 PCs, but I'm
not sure. In addition, the following strange DOS events occur:
1) Infrequent but noticeable system hangs in large programs
such as spreadsheets or word-processors
2) One program misreading the motherboard speed to report
the CPU as a 66.7 MHz pentium, instead of a 75 MHz.
3) Norton System Info reporting the system BIOS as "Thursday,
March 30", etc.
4) My Phoenix BIOS options for the SQ588 Motherboard are listed below,
together with their default settings and an optional '*' to indicate
my experimentation with different settings. While the SCSI controller
appears to be in Slot #1, only the IRQ setting for Slot #4 is detected
by Linux.
System info:
Time: <time>
Date: <date>
Video System: EGA/VGA
Diskette A: 1.44 MB, 3 1/2 "
Diskette B: Not installed
Fixed Disk Info:
IDE Adapter 0 Master: (None)
IDE Adapter 0 Slave: (None)
IDE Adapter 1 Master: (None)
IDE Adapter 1 Slave: (None)
Large Disk Access Mode: DOS *
Integrated Peripherals:
COM Port: Auto *
COM Port: Auto *
LPT Port: 378, IRQ 7 *
Diskette Controller: Enabled
Parallel Port Mode: Standard
Local Bus IDE: Enabled *
Memory Cache:
External Cache: Enabled *
Cache System BIOS: Enabled *
Cache Video BIOS: Enabled *
Cache Speed: Slower
Cache burst r/w Cycle: 2T (15 ns SRAM default)
Non-cacheable area #1: Disabled
Allocation: Local DRAM
Start Address: 0
Size: 64 KB
Non-cacheable area #2: Disabled
Allocation: Local DRAM
Start Address: 0
Size: 64 KB
Memory Shadow:
System BIOS: <Always Enabled>
Video BIOS: Enabled *
Shadow Option ROMs:
C800-CFFF: Disabled
D000-D7FF: Disabled
D800-DFFF: Disabled
Advanced Chip Control:
Fast Reset Latency: 2 us
16-bit I/O recovery time: 3 BusClk
8-bit I/O recovery time: 5 BusClk
16-bit mem, I/O wait state: 1 Ws
CPU to PCI burst memory write: Enabled *
PCI Device:
Base I/O 3000
Base Memory Address: 0080000000
Multimedia Mode: Disabled
Slot #1:
Interrupt: IRQ 9 *
Edge/Level: Edge *
Slot #2:
Interrupt: IRQ 9 *
Edge/Level: Edge *
Slot #3:
Interrupt: IRQ 9 *
Edge/Level: Edge *
Slot #4:
Interrupt: IRQ 9 *
Edge/Level: Edge *
Boot Options:
Sequence: A: then C:
Setup prompt: Enabled
POST errors: Enabled
Floppy Check: Enabled
Summary Screen: Enabled
Security and Anti-Virus:
Supervisor Password: <none>
Password on boot: Disabled
Diskette Access: Supervisor
Fixed Disk Boot Sector: Normal
Green PC Features:
Power Save Mode: Disabled
System Suspend Timer: 3 min.
Hard Disk Standby Timer: Disabled
VGA with Power Down Feature: None
Suspend Switch Select: Enabled *
APM SMI Function Support: Enabled *
VGA Access Detection: Disabled
IRQ1-15/NMI Activity: Enabled *
IRQ1 Monitor: On *
IRQ2 Monitor: On *
IRQ3 Monitor: On *
IRQ4 Monitor: On *
IRQ5 Monitor: On *
IRQ6 Monitor: On *
IRQ7 Monitor: On *
IRQ8 Monitor: On *
IRQ9 Monitor: On *
IRQ10 Monitor: On *
IRQ11 Monitor: On *
IRQ12 Monitor: On *
IRQ13 Monitor: On *
IRQ14 Monitor: On *
IRQ15 Monitor: On *
Thank you again for any and all help/insights!
--
Camm Maguire camm@enhanced.com
==========================================================================
"The earth is but one country, and mankind its citizens." -- Baha'u'llah
