Chinaunix

标题: 怎么会这样 Fast Data Access MMU Miss [打印本页]

作者: zgccpy    时间: 2003-09-25 09:56
标题: 怎么会这样 Fast Data Access MMU Miss
ok boot cdrom            
Boot device: /pci@1f,0/pci@1/pci@1/ide@e/cdrom@2:f  File and args:              
SunOS Release 5.9 Version Generic 64-bit
Copyright 1983-2002 Sun Microsystems, Inc.  All rights reserved.
Use is subject to license terms.
Fast Data Access MMU Miss
ok
ok
ok boot cdrom
Resetting ...


screen not found.
keyboard not found.
Keyboard not present.  Using ttya for input and output.

Netra t1 (UltraSPARC-IIi 440MHz), No Keyboard
OpenBoot 3.10.24 ME, 512 MB memory installed, Serial #12753110.
Ethernet address 8:0:20:c2:98:d6, Host ID: 80c298d6.



Executing last command: boot cdrom
Boot device: /pci@1f,0/pci@1/pci@1/ide@e/cdrom@2:f  File and args:
Fast Data Access MMU Miss
ok
作者: baijianyi    时间: 2003-09-25 10:12
标题: 怎么会这样 Fast Data Access MMU Miss
应该是主板坏了
baijianyi@aytk.com.cn
作者: metor78    时间: 2003-09-25 10:17
标题: 怎么会这样 Fast Data Access MMU Miss
内存也有可能
作者: zgccpy    时间: 2003-09-25 10:30
标题: 怎么会这样 Fast Data Access MMU Miss
真的假的,我是重装系统时出现的这个问题
作者: zgccpy    时间: 2003-09-25 10:36
标题: 怎么会这样 Fast Data Access MMU Miss
There are two common Solaris[TM] 2.5.1 or 2.6 boot problems that are similar, but
have different solutions. There is some confusion as to what solution to use for
what problem. Both problems are that systems fail to boot to a Solaris 2.5.1 or
2.6 CD-ROM or JumpStart[TM] image. SRDB documents have been written for each
problem and this is a description of each of these and how to tell the difference.



PROBLEM #1

(see also SRDB 20576)

Any system that has a 450MHz UltraSPARC(R) II CPU or the 440MHz UltraSPARC IIi CPU
will not be able to boot Solaris 2.5.1 HW 11/97, Solaris 2.6 HW 3/98 or Solaris
2.6 HW 5/98 CD-ROM or JumpStart images of these Operating Systems. It will fail
with "hme0:link down" or keyboard input multiple repeats.

These systems must be booted with a cdrom called the Operating Environment
Installation CD February 2000 (OECD), which is Part No. 704-7076-10. Booting
this OECD will then prompt for the 2.5.1 or 2.6 Solaris CD for the install
of the OS. This OECD can also be used to modify a JumpStart image to boot/load
2.5.1 or 2.6 on these systems. See the manual "Installing Solaris Software for
Selected Hardware" Part No. 806-4005-10 for procedure for modifying a jumpstart
image.

Once a jumpstart image is modified with the OECD, this image can be used to load
Solaris 2.5.1/2.6 on these systems. If the method of installation is via a cdrom,
then the OECD must always be used. The manual (included with the OECD) also
describes how to use the OECD for CD-ROM-based installs. Here is a list of the
systems affected:

        Ultra 5         Ultra 10
        Netra T105      Ultra 60
        Netra T1120     Netra T1125
        Ultra 80        Enterprise 420R
        Netra T1400     Netra T1405
        Ultra 450       Enterprise 220R

Only the above systems with 450MHz or 440MHz CPUs are effected by this problem.
You can check the speed of the cpu(s) by running the command ".speed" from the
OK prompt. The problem is permanently fixed by loading the most current kernel
patch. Patching the kernel to this level is one of the functions of the OECD. The
OECD actually has a Solaris 7 kernel that it boots to, then it kicks off an install
that prompts for the 2.5.1 or 2.6 cdrom for the OS install. It will load the kernel
patch during this install, so the kernel will be patched upon the first boot to
the new OS on the disk.

This CD-ROM is not included with Solaris and is usually included with the system
in the "Binary Code License" package. The OECD can be ordered by Sun personnel
from this website: http://acac.central/SAG/Templates/CD_Zero.html

NOTE: The OECD can also be used to fix the "NOTICE: Can't find driver for console
frame buffer" install/boot problem mentioned in SRDB 19271.

NOTE: The Ultra Enterprise[TM] 450 with a 480MHz CPU requires a special procedure.
See SRDB 24408.



PROBLEM #2

(see also SRDB 20149)

Enterprise Server systems that use the 400MHz UltraSPARC II with 8mb ecache CPU
module(s) will fail to boot to Solaris 2.5.1 or Solaris 2.6 cdrom (or JumpStart
image) with "Fast Data Access MMU Miss" and/or "mutex_enter: bad mutex". SRDB
20149 describes this problem and has the exact procedure to boot/load these
systems with 2.5.1/2.6.

The problem is that the kernel on the 2.5.1/2.6 CD-ROM cannot load with the
400MHz/8MB ecache CPU until the kernel patch is loaded. The workaround (seen
in the SRDB) is to run the command "limit-ecache-size" from the OK prompt before
booting to the 2.5.1/2.6 cdrom or JumpStart image. It will be able to boot the
cdrom or image immediately after this command is run. After the OS is loaded,
you will need to run this command again from the OK prompt to boot the system
until you load the latest kernel patch. Here are the systems (that potentially
have the 400MHz 8mb cache cpu installed) affected by this problem:

        E3000   E3500   E4000
        E4500   E5000   E5500
        E6000   E6500   

Only systems that have the 400MHz/8MB cache cpu(s) are affected.



Final Note:

Do not confuse these two different problems. Do not use the OECD on an Enterprise
server in the model range E3000 - E6500 (with 400MHz/8MB cache). These servers
are fixed with the limit-ecache-size workaround to boot/load 2.5.1/2.6. The
Ultra[TM] 5 - Ultra 450 group of systems (with 440/450 MHz CPU) are the ones that
need the OE cdrom to boot/load 2.5.1/2.6.
作者: 我是海风    时间: 2003-09-25 10:38
标题: 怎么会这样 Fast Data Access MMU Miss
这个我遇到过,当时用reset-all就搞定了。
作者: zgccpy    时间: 2003-09-25 10:54
标题: 怎么会这样 Fast Data Access MMU Miss
[quote]原帖由 "我是海风"]这个我遇到过,当时用reset-all就搞定了。[/quote 发表:
   

您的机器型号跟偶的一样吗
作者: 我是海风    时间: 2003-09-25 10:57
标题: 怎么会这样 Fast Data Access MMU Miss
我的服务器是古董:E3000
作者: zgccpy    时间: 2003-09-25 10:59
标题: 怎么会这样 Fast Data Access MMU Miss
但愿偶也可以搞定 第一次装就碰到这种问题,真是不幸,偶的一个朋友跟偶的配置一摸一样,他的安装就没有问题,郁闷ing
作者: zgccpy    时间: 2003-09-25 16:22
标题: 怎么会这样 Fast Data Access MMU Miss
没有遇到同样的问题了吗,顶一下
作者: mkwf    时间: 2003-09-26 00:09
标题: 怎么会这样 Fast Data Access MMU Miss
在OK下面 setdefaults之后重启试试看。
作者: mmmmn    时间: 2003-09-26 09:13
标题: 怎么会这样 Fast Data Access MMU Miss
检查一下solaris9要求的最低OBP版本,我觉的你的OBP版本过低了
作者: zgccpy    时间: 2003-09-26 10:17
标题: 怎么会这样 Fast Data Access MMU Miss
昨日回家使用reset-all,reset等方法,依然如故,继续顶
作者: race    时间: 2003-09-26 10:48
标题: 怎么会这样 Fast Data Access MMU Miss
我在一个老的SPARC20上装8也是这个错误提示,后来换别的硬盘好了。
作者: baijianyi    时间: 2003-09-26 11:04
标题: 怎么会这样 Fast Data Access MMU Miss
还是换主板吧!一定没错。如方便找我吧。010-82572879 白
作者: 东方蜘蛛    时间: 2003-09-26 11:16
标题: 怎么会这样 Fast Data Access MMU Miss
PROM设置问题,是板卡的版本同步问题,不影响使用
init 0 (or STOP+A)
set-defaults
reset-all
boot (or boot cdrom)
作者: zgccpy    时间: 2003-09-26 12:47
标题: 怎么会这样 Fast Data Access MMU Miss
[quote]原帖由 "mmmmn"]检查一下solaris9要求的最低OBP版本,我觉的你的OBP版本过低了[/quote 发表:
     


偶的一个朋友跟偶的配置一摸一样,他的安装就没有问题
作者: race    时间: 2003-09-26 13:27
标题: 怎么会这样 Fast Data Access MMU Miss
你把他的硬盘拿过来装上,用光盘启动一下看看
作者: zgccpy    时间: 2003-09-26 13:54
标题: 怎么会这样 Fast Data Access MMU Miss
有道理,大家请看这个,服务器安装9肯定是没有问题的,Netra t1 Model 105 Servers 这是偶的型号
http://docs.sun.com/db/doc/816-1664-10/6m82lv9lf?a=view
作者: johnyou    时间: 2003-09-26 15:05
标题: 怎么会这样 Fast Data Access MMU Miss
换CD或CDROM,或boot net -s (有环境的情况)试试。
作者: zgccpy    时间: 2003-09-26 15:18
标题: 怎么会这样 Fast Data Access MMU Miss
非常感谢大家的帮忙,即使搞不好也高兴
作者: johnyou    时间: 2003-09-26 17:57
标题: 怎么会这样 Fast Data Access MMU Miss
我见过有一台CDROM太老,CD太新就出现过这样的错误。尽量试试boot net -s看看。
作者: sunmarmot    时间: 2003-09-26 22:42
提示: 作者被禁止或删除 内容自动屏蔽
作者: zgccpy    时间: 2003-09-27 09:33
标题: 怎么会这样 Fast Data Access MMU Miss
使用 set-defaults,依然如故,今天换光驱试试
作者: 五行二    时间: 2003-09-27 11:29
提示: 作者被禁止或删除 内容自动屏蔽
作者: zgccpy    时间: 2003-09-27 14:14
标题: 怎么会这样 Fast Data Access MMU Miss
非常感谢大家的热情帮助,更换光驱后问题解决
作者: sevenhundred    时间: 2003-09-27 17:28
标题: 怎么会这样 Fast Data Access MMU Miss
我也遇过这种提示。当时有CPU CACHE方面的出错提示。后来换CPU后就好了。
作者: suncn    时间: 2003-09-27 18:00
标题: 怎么会这样 Fast Data Access MMU Miss
我在安装一台4800的时候也碰到这个问题,用2222A接D240的时候有BUG,BUG ID是4727426,以后大家注意一下!

This problem is on the Cauldron (X2222A) pci card connected to a D240
on a Sun Fire (Serengeti) machine.

Problem: When a probe-scsi-all command is issued at the OBP level of a
         domain,
         a) the command does not display the entire path to the
            devices connected to it.
         b) issuing a boot command immediately fails.
            A reset or reset-all is required.
            

The probe-scsi-all output does not show the entire path to the
devices connected to a cauldron pci card.

  
{10} ok probe-scsi-all
/ssm@0,0/pci@1d,600000/SUNW,qlc@1
/ssm@0,0/pci@1d,700000/scsi@2,1
/ssm@0,0/pci@1d,700000/scsi@2
/ssm@0,0/pci@1c,600000/pci@1/scsi@2,1
/ssm@0,0/pci@1c,600000/pci@1/scsi@2    <<<<this is incomplete path
Target 0
  Unit 0   Disk     FUJITSU MAJ3182M SUN18G 0804
Target 3
  Unit 0   Processor             2   
Target 6
  Unit 0   Removable Read Only device    TOSHIBA DVD-ROM SD-M14011007

The following scenarios has happen in the Field resulting
in not able to boot

1. using incomplete boot path
  ok>; boot  /ssm@0,0/pci@1c,600000/pci@1/scsi@2
  ERROR: Can't open boot-device
  
2. using complete boot path to DVDrom

  ok>;boot /ssm@0,0/pci@1c,600000/pci@1/scsi@2,0/disk@6,0:f
  ERROR: Fast Data Access MMU Miss


It appears that a boot command is issued right after the probe-scsi-all command,

the boot comamnd will fail with Can't open boot-device,  Fast Data Access MMU
Miss. This has been tested on system controller firmware 5.12.6 and 5.13.2

When a reset or reset-all command is issued, booting off the Cauldron (X2222A)
card is successful.

{10} ok reset-all
Resetting ...
Copying IO prom to Cpu dram
.{/N0/SB4/P0} @(#) lpost        5.12.6  2001/11/29 12:41
{/N0/SB4/P0} Copyright 2001 Sun Microsystems, Inc. All rights reserved.
.............................{/N0/SB4/P0} Running PCI IO Controller Basic Tests
{/N0/SB4/P0} Jumping to memory 00000000.00000020 [00000010]
{/N0/SB4/P0} System PCI IO post code running from memory
{/N0/SB4/P0} @(#) lpost         5.12.6  2001/11/29 12:41
{/N0/SB4/P0} Copyright 2001 Sun Microsystems, Inc. All rights reserved.
{/N0/SB4/P0} Subtest: PCI IO Controller Register Initialization for aid 0x1c
{/N0/SB4/P0} Running PCI IO Controller Functional Tests
{/N0/SB4/P0} Running SBBC Basic Tests
{/N0/SB4/P0} Subtest: SBBC PCI Reg Initialization for aid 0x1c
{/N0/SB4/P0} Running PCI IO Controller Basic Tests
{/N0/SB4/P0} Subtest: PCI IO Controller Register Initialization for aid 0x1d
{/N0/SB4/P0} Running PCI IO Controller Functional Tests
{/N0/SB4/P0} @(#) lpost         5.12.6  2001/11/29 12:41
{/N0/SB4/P0} Copyright 2001 Sun Microsystems, Inc. All rights reserved.
{/N0/SB4/P1} @(#) lpost         5.12.6  2001/11/29 12:41
{/N0/SB4/P1} Copyright 2001 Sun Microsystems, Inc. All rights reserved.
{/N0/IB8/P0} Passed
{/N0/IB8/P1} Passed
{/N0/SB4/P0} Running Domain Level Tests
{/N0/SB4/P2} @(#) lpost         5.12.6  2001/11/29 12:41
{/N0/SB4/P3} @(#) lpost         5.12.6  2001/11/29 12:41
{/N0/SB4/P0} GLOBAL IO SRAM on board 8
{/N0/SB4/P2} Copyright 2001 Sun Microsystems, Inc. All rights reserved.
{/N0/SB4/P3} Copyright 2001 Sun Microsystems, Inc. All rights reserved.
{/N0/SB4/P0} Running Domain Basic Tests
{/N0/SB4/P0} Running Domain Advanced Tests
{/N0/SB4/P0} Running Domain Stick Sync Tests
{/N0/SB4/P0} Running Domain Verify Stick Sync Tests
{/N0/SB4/P0} DCB_DECOMP_OBP command succeeded
{/N0/SB4/P0} GLOBAL IO SRAM on board 8
{/N0/SB4/P0} Committing retained memory 00000000.00b74000-00000000.00b75fff
{/N0/SB4/P0} Retaining 00000000.00b74000-00000000.00b75fff
{/N0/SB4/P0} Committing retained memory 00000000.00b92000-00000000.00b93fff
{/N0/SB4/P0} Retaining 00000000.00b92000-00000000.00b93fff
{/N0/SB4/P1} GLOBAL IO SRAM on board 8
{/N0/SB4/P0}  CPU 17 clearing 00000000.00000000 to 00000000.002dd000
{/N0/SB4/P2} GLOBAL IO SRAM on board 8
{/N0/SB4/P0}  CPU 18 clearing 00000000.002dd000 to 00000000.005ba000
{/N0/SB4/P3} GLOBAL IO SRAM on board 8
{/N0/SB4/P0}  CPU 19 clearing 00000000.005ba000 to 00000000.00897000
{/N0/SB4/P0}  CPU 16 clearing 00000000.00897000 to 00000000.00b74000
{/N0/SB4/P0}  CPU 17 clearing 00000000.00b76000 to 00000000.00b7d000
{/N0/SB4/P0}  CPU 18 clearing 00000000.00b7d000 to 00000000.00b84000
{/N0/SB4/P0}  CPU 19 clearing 00000000.00b84000 to 00000000.00b8b000
{/N0/SB4/P0}  CPU 16 clearing 00000000.00b8b000 to 00000000.00b92000
{/N0/SB4/P0}  CPU 17 clearing 00000000.00b94000 to 00000000.808af000
{/N0/SB4/P0}  CPU 18 clearing 00000000.808af000 to 00000001.005ca000
{/N0/SB4/P0}  CPU 19 clearing 00000001.005ca000 to 00000001.802e5000
{/N0/SB4/P0}  CPU 16 clearing 00000001.802e5000 to 00000002.00000000
{/N0/SB4/P0} GLOBAL IO SRAM on board 8
{/N0/SB4/P0} Decompress OBP done
{/N0/SB4/P0} DCB_ENTER_OBP command succeeded
{/N0/SB4/P1} DCB_ENTER_OBP command succeeded
{/N0/SB4/P2} DCB_ENTER_OBP command succeeded
{/N0/SB4/P3} DCB_ENTER_OBP command succeeded


Sun Fire 4800
OpenFirmware version 5.12.6 (11/29/01 12:41)
Copyright 2001 Sun Microsystems, Inc.  All rights reserved.
SmartFirmware, Copyright (C) 1996-2001.  All rights reserved.
8192 MB memory installed, Serial #14199680.
Ethernet address 8:0:20:d8:ab:80, Host ID: 80d8ab80.


{10} ok boot /ssm@0,0/pci@1c,600000/pci@1/scsi@2,0/disk@6,0:f   <---ok
SunOS Release 5.8 Version Generic_108528-13 64-bit
Copyright 1983-2001 Sun Microsystems, Inc.  All rights reserved.
|

{10} ok boot /ssm@0,0/pci@1c,600000/pci@1/scsi@2/disk@sd0,0   <---ok
-
作者: ljhappy    时间: 2008-12-11 14:52
Ultra 5出现此问题,更换光驱问题解决




欢迎光临 Chinaunix (http://bbs.chinaunix.net/) Powered by Discuz! X3.2