daveh@marob.masa.com (Dave Hammond) (12/28/89)
OS: Xenix 386, release 2.3.2, update R ("GT" upgrade kit)
Hardware: Everex 25mhz 386, 6 MB 32 bit RAM
Peripherals: 80387, ESDI drive, 60MB QIC tape, Arnet Smart-8 serial expander.
The system boots and operates normally for a period of time, but will
eventually panic with an error:
Kernel Panic - Non-recoverable Kernel Page Fault.
The most recent panic (this morning) was followed by:
Kernel Double Panic - Stack Fault.
The panics can occur hours or days apart, and (as yet) can not be
attributed to any specific condition. The most obvious hint that the
problem is surfacing is when an exec?() call causes a segmentation
violation, without ever loading the program. If the shell is attempting
the exec?(), this results in a "Memory fault, core dumped" message and a
return to the shell prompt.
System load does not appear to be an obvious factor. Nor does command
argument length appear to be an issue, as a simple one word command
line, like "ps" or "vi", can fail.
The boot screen configuration info follows:
10 bits of I/O address decoding
Wed Dec 27 11:30:03
device address vector dma comment
----------------------------------------------------------------------------
%fpu - 35 - type=80387
%floppy 0x3F2-0x3F7 06 2 unit=0 type=96ds15
%serial 0x3F8-0x3FF 04 - unit=0 type=Standard nports=1
%serial 0x2F8-0x2FF 03 - unit=1 type=Standard nports=1
%parallel 0x378-0x37A 07 - unit=0
%parallel 0x3BC-0x3BE 07 - unit=1
%tape 0x300-0x304 05 1 type=W
Arnet Smartport with 8 ports on COM4
%console - - - unit=mono type=2
Wed Dec 27 11:30:04
%disk 0x1F0-0x1F7 36 - type=W0 unit=0 cyls=1224 hds=7 secs=36
rootdev 1/40, pipedev 1/40, swapdev 1/41
mem: total = 10112k, reserved = 4k, kernel = 1764k, user = 8344k
kernel: drivers = 4k, 12 screens = 68k, 1024 i/o bufs = 1024k, msg bufs = 8k
nswap = 25000, swplo = 0, Hz = 50, maximum user process size = 15844k
Advice as to remedial action, debugging procedures, etc is desparately
sought! E-mail is preferred, however, I read this newsgroup daily.
__
Dave Hammond
EMAIL: daveh@marob.masa.com
VOICE: 516-872-3535
FAX: 516-872-3579karl@ddsw1.MCS.COM (Karl Denninger) (12/29/89)
In article <2599259F.31C0@marob.masa.com> daveh@marob.masa.com (Dave Hammond) writes: >OS: Xenix 386, release 2.3.2, update R ("GT" upgrade kit) >Hardware: Everex 25mhz 386, 6 MB 32 bit RAM >Peripherals: 80387, ESDI drive, 60MB QIC tape, Arnet Smart-8 serial expander. > >The system boots and operates normally for a period of time, but will >eventually panic with an error: > > Kernel Panic - Non-recoverable Kernel Page Fault. > >The most recent panic (this morning) was followed by: > > Kernel Double Panic - Stack Fault. > >The panics can occur hours or days apart, and (as yet) can not be >attributed to any specific condition. Try running 80286 things, such as the 286 compiler. That seems to provoke it in some cases. >The most obvious hint that the >problem is surfacing is when an exec?() call causes a segmentation >violation, without ever loading the program. If the shell is attempting >the exec?(), this results in a "Memory fault, core dumped" message and a >return to the shell prompt. > >Advice as to remedial action, debugging procedures, etc is desparately >sought! E-mail is preferred, however, I read this newsgroup daily. Try removing the Arnet support. We had just such a problem and it disappeared when we got rid of the Anvil board in our system! -- Karl Denninger (karl@ddsw1.MCS.COM, <well-connected>!ddsw1!karl) Public Access Data Line: [+1 708 566-8911], Voice: [+1 708 566-8910] Macro Computer Solutions, Inc. "Quality Solutions at a Fair Price"
asv@gaboon.UUCP (Stan Voket) (12/29/89)
In article <2599259F.31C0@marob.masa.com> daveh@marob.masa.com (Dave Hammond) writes: >OS: Xenix 386, release 2.3.2, update R ("GT" upgrade kit) >device address vector dma comment >---------------------------------------------------------------------------- >%parallel 0x378-0x37A 07 - unit=0 >%parallel 0x3BC-0x3BE 07 - unit=1 >Advice as to remedial action, debugging procedures, etc is desparately >sought! E-mail is preferred, however, I read this newsgroup daily. Dave, You've got poth parallel ports using the same interrupt vector. This is your problem. (I've been guilty of it myself. :-)) -- +----------------------------------------------------------------------+ | - Stan Voket, asv@gaboon - OR - ...uunet!hsi!stpstn!gaboon!asv | | Land Line: (203) 746-4489 TELEX 4996516 - | +----------------------------------------------------------------------+
daveh@marob.masa.com (Dave Hammond) (12/29/89)
Thanks to all those who took the time to offer advice and consolation. After returing to a stock kernel and rebuilding driver by driver, I discovered a conflict between the system's hardware cache and the address which I had assigned to the Arnet board's 64k buffer area. -- Dave Hammond daveh@marob.masa.com uunet!masa.com!marob!daveh
daveh@marob.masa.com (Dave Hammond) (12/29/89)
In article <2599259F.31C0@marob.masa.com> I wrote: >>OS: Xenix 386, release 2.3.2, update R ("GT" upgrade kit) >>device address vector dma comment >>---------------------------------------------------------------------------- >>%parallel 0x378-0x37A 07 - unit=0 >>%parallel 0x3BC-0x3BE 07 - unit=1 In article <4987@gaboon.UUCP> asv@gaboon.UUCP (Stan Voket) writes: >You've got poth parallel ports using the same interrupt vector. This is >your problem. (I've been guilty of it myself. :-)) Both printers were installed on interrupt 7 because the alternate parallel driver interrupt (5) was used by the tape drive. In practise, it is rare that both printers are active simultaneously. In fact, I have witnessed both printers active at the same time, and all appears to go well. Perhaps this is because the printers have large buffers, so the actual data transmission time is quite short? Also, can an interrupt conflict, like this, actually panic a system or will it just bog down printing when both printers are active? -- Dave Hammond daveh@marob.masa.com uunet!masa.com!marob!daveh