| Title: | SABLE SYSTEM PUBLIC DISCUSSION |
| Moderator: | COSMIC::PETERSON |
| Created: | Mon Jan 11 1993 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2614 |
| Total number of notes: | 10244 |
Hi Kurt!
Can you have a look at the following errorlog entryof a 2100 4/275?
UNIX 3.2G
SRM 4.5
PAL 1.45
There must be "Quadword-shift" in the errorlogframe because this is
the errorlogframe of an Correctable Memory Error on Memory Modul 2.
(BIU_STAT = 240, MEM ERR REG for MODUL2 is 40001)
But up to now the BIU_STAT was located at Address 1A8 in this errorlog output
it is at address 1B0 !
Also the Memory-Registers: for Modul 2 the Error-Register have been at
address 3F0 in this output it is on 3F8!
--> All registers are schifted backwards - Is this new in 3.2G?
Were does it come from?
thanks for any hint
Helmut
entwama1:root> uerf -R -Z | more
uerf version 4.2-011 (122)
********************************* ENTRY 1. *********************************
----- EVENT INFORMATION -----
EVENT CLASS ERROR EVENT
OS EVENT TYPE 100. CPU EXCEPTION
SEQUENCE NUMBER 44.
OPERATING SYSTEM DEC OSF/1
OCCURRED/LOGGED ON Thu Apr 17 09:06:31 1997
OCCURRED ON SYSTEM entwama1
SYSTEM ID x00060009 CPU TYPE: DEC 2100
SYSTYPE x00000000
PROCESSOR COUNT 4.
PROCESSOR WHO LOGGED x00000002
----- UNIT INFORMATION -----
UNIT CLASS CPU
RECORD ENTRY DUMP:
RECORD HEADER
0000: 002C04B8 00060009 00060101 3355CBF7 *..,...........U3*
0010: 77746E65 31616D61 00000000 00000000 *entwama1........*
0020: 00000004 00000002 10010064 00000000 *........d.......*
0030: 00000000 00000000 *........ *
RECORD BODY
0038: 00000002 00000228 00000220 80000000 *....(... .......*
0048: 00000110 000001A0 0000008A 00000001 *................*
0058: 0000008A 00000001 07B39800 00000000 *................*
0068: 00000004 001382F8 0063FD9C 00000000 *..........c.....*
0078: 00000000 00000000 00000009 00000000 *................*
0088: 00000240 00000000 00004E00 00000000 *@........N......*
0098: 00000600 00000000 00000000 00000000 *................*
00A8: 0046F5E0 FFFFFC00 00000000 00000000 *..F.............*
00B8: 0046F980 FFFFFC00 0046F9B0 FFFFFC00 *..F.......F.....*
00C8: 0046FA10 FFFFFC00 0046F780 FFFFFC00 *..F.......F.....*
00D8: 0046F490 FFFFFC00 00019410 00000000 *..F.............*
00E8: 1FFED340 00000001 A929BA58 FFFFFFFF *@.......X.).....*
00F8: 005E2F80 FFFFFC00 EDCBA987 EDCBA987 *./^.............*
0108: 72727272 40424272 FFFFFFFD CF77BF3F *rrrrrBB@....?.w.*
0118: 00000000 00000000 00010000 00000000 *................*
0128: 0063FD9B 00000000 00000002 00000000 *..c.............*
0138: 0BAB0000 00000000 00000000 FFFFFFFC *................*
0148: 00000001 00000000 3EA65A58 00000000 *........XZ.>....*
0158: 00014125 00000000 00000000 00000000 *%A..............*
0168: 00000000 00000000 00000004 001382F8 *................*
0178: 00014000 00000000 000004F0 00000000 *.@..............*
0188: 00000000 00000000 00003F01 00000000 *.........?......*
0198: 00000003 00000000 FFFFFFFF 00000007 *................*
01A8: 0000142E 00000000 00000240 00000000 *........@.......*
01B8: 00019410 00000000 5000E567 0000000E *........g..P....*
01C8: 00000000 00000000 00019710 00000000 *................*
01D8: 00006970 00000000 0004F833 00002600 *pi......3....&..*
01E8: C00001C5 C00001C5 00000000 00000000 *................*
01F8: 00000000 00000000 00000A00 00000A00 *................*
0208: 20C5CD20 20C5CD20 0D1B4C84 0D1B4C84 * .. .. .L...L..*
0218: 00000868 00000868 C4000000 C8002000 *h...h........ ..*
0228: E0800043 E0800043 2F600B43 0F600B43 *C...C...C.`/C.`.*
0238: 00000000 00000000 00000000 00000000 *................*
0248: 00000000 00000000 0FD7CCE8 0FD7CCE8 *................*
0258: 0FF9EE89 0FF9EE19 0002001A 0002001A *................*
0268: 00000011 000000B8 27060190 FE000082 *...........'....*
0278: 00000000 00000000 E3800010 E3800010 *................*
0288: 20600B43 00600B43 00000000 00000000 *C.` C.`.........*
0298: 5A79A860 00000007 00000010 00000000 *`.yZ............*
02A8: 00000000 00000000 0010603F 00000000 *........?`......*
02B8: 400807FF 00000000 3FF00000 00000000 *...@.......?....*
02C8: 00000000 00000000 000C01FF 00000000 *................*
02D8: 1FF00000 00000000 00440000 00000000 *..........D.....*
02E8: 00110803 00000663 0011080D 00012618 *....c........&..*
02F8: 0011080E 0000821A 0011080F 0003221C *............."..*
0308: 00110810 0003F61E 00110811 0002DA20 *............ ...*
0318: 00110800 0000065D 00110801 0000065F *....]......._...*
0328: 00000008 00000058 00000000 00000000 *....X...........*
0338: 00000000 00000000 00000000 00000000 *................*
0348: 00000000 00000000 00000000 00000000 *................*
0358: 00000000 00000000 00000000 00000000 *................*
0368: 00000000 00000000 00000000 00000000 *................*
0378: 00000000 00000000 00000000 00000000 *................*
0388: 00000008 00000058 00000001 00000000 *....X...........*
0398: 00000000 00000000 00000000 00000000 *................*
03A8: 00000000 00000000 00000000 00000000 *................*
03B8: 00000000 00000000 00000000 00000000 *................*
03C8: 00000000 00000000 00000000 00000000 *................*
03D8: 00000000 00000000 00000000 00000000 *................*
03E8: 00000008 00000058 00000002 00000000 *....X...........*
03F8: 00040001 00000000 069E6800 E2800008 *.........h......*
0408: F083FFFF 00601E83 8005506A 8005506A *......`.jP..jP..*
0418: 0C020C02 0E560E56 00000B45 00000017 *....V.V.E.......*
0428: 20000000 20000000 00000800 00000800 *... ... ........*
0438: 000001D8 000001D8 00000000 00000000 *................*
0448: 00000008 00000058 00000003 00000000 *....X...........*
0458: 00000000 00000000 E2C00008 E2C00008 *................*
0468: 20601E83 00601E83 8015506B 8015506B *..` ..`.kP..kP..*
0478: 071506A1 04230555 0000000D 00000017 *....U.#.........*
0488: 20000000 20000000 00000800 00000800 *... ... ........*
0498: 000001D8 000001D8 00000000 00000000 *................*
04A8: 00000000 00000000 00000000 5E3C7E25 *............%~<^*
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 2580.1 | Who knows ? | ATZIS2::PUTZENLECHNE | wherever is fun, there's always ALPHA | Wed Apr 23 1997 07:28 | 5 |
Hi!
Anybody else who is able to answer the question in .0 ?
Helmut
| |||||
| 2580.2 | Seen on AS4100 | DANGER::HAYES | Wed Apr 23 1997 13:19 | 94 | |
This is a blitz released against the AS4100, certainly similar,
perhaps the console folks can comment.
Dennis
<<< MVBLAB::SYS$SYSDEVICE:[NOTES$LIBRARY]ALPHASERVER_4100.NOTE;1 >>>
-< AlphaServer 4100 >-
================================================================================
Note 93.39 Field Blitzes 39 of 45
MOVMON::DAVIS 79 lines 26-FEB-1997 10:03
-< Errorlog Decode Byte Offset >-
--------------------------------------------------------------------------------
Request log number: 886
+---------------------------+TM
| | | | | | | |
| d | i | g | i | t | a | l | TIME DEPENDENT BLITZ
| | | | | | | |
+---------------------------+
BLITZ TITLE: Alpha Server 4100 - Errorlog Decode Byte Offset
DATE: February 26, 1997
AUTHOR:Ted Gent TD #:
DTN:223-6530
ENET:POBOXA::GENT CROSS REFERENCE #'s:
DEPARTMENT:SBU Engineering (PRISM/TIME/CLD#'s)
INTENDED AUDIENCE: U.S/EUROPE/GIA PRIORITY LEVEL: 2
(1=TIME CRITICAL,
2=NON-TIME CRITICAL)
=====================================================================
Subject : Alpha Server 4100 - Errorlog Decode Byte Offset
Problem
-------
A bug has been found in the Digital Unix and OpenVMS PAL code which
causes, under certain circumstances, the errorlog to be offset by a byte
in Correctable Read Data (MChk 620/630) handling. The potential for the
problem to occur is present in all AlphaServer 4100 SRM Console Code prior
to release 4.8-5.
If you encounter entries in the errorlog indicating a Machine Check code
of 400, an unassigned code, that record may be offset by a byte. Please
note that this only affects individual records, other entries in the same
log file will be correctly recorded and reported.
Description
-----------
1. An I/O-detected 620 CRD error is delivered as a device IPL (21)
interrupt by the IOD.
2. PAL is supposed to raise IPL to MCHK IPL at this point. In all
versions of SRM console prior to V4.8-X, IPL is not raised..
3. During the course of handling the 620 error, PALcode will scrub
the location if the 620 was a memory CRD error. When PAL reads
the bad location in order to write it back (scrub it), if the
error was not a transient, the CPU takes a 630 CRD error since
it sees the same bad ECC.
4. If IPL is at MCHK IPL, which it should be at, the 630 will pend
until the 620 can be fully serviced. The bug in PAL is such that
the 630 can potentially come in on top of the 620. [Software CRD
handlers typically raise IPL to MCHK IPL (31), but there's still
a window between PAL dispatching to the interrupt service routine
(ISR) and when IPL is actually raised by the handler.]
5. The net effect of the 630 on top of the 620 is that PAL passes a
minus one (FFFFFFFF) pointer to the ISR and when this pointer is
added to various other hard-coded offsets, it results in the CRD
frame passed by PAL to the upper level software being off by one
byte. This causes the CRD entry to the screen (if at console) or
the system error log entry (if at UNIX/VMS level) to be shifted
by one byte.
Resolution
----------
Upgrade to the V4.8-5 SRM console if you are experiencing this problem.
Occurrences have been rare and the only documented and analysed case was
from a site being monitored by DPP that was experiencing greater than 15
CRD's per hour.
bc/tg 02/26/97
| |||||