Menu
▾
▴
[coLinux-devel] [ colinux-Bugs-1745597 ] BSOD on Windows Server 2003 with Adaptec AIC-9580W RAID Ctrl
|
From: SourceForge.net <no...@so...> - 2007-08-21 16:12:14
|
Bugs item #1745597, was opened at 2007-06-29 16:23 Message generated for change (Comment added) made by nobody You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=622063&aid=1745597&group_id=98788 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: None Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Nobody/Anonymous (nobody) Assigned to: Nobody/Anonymous (nobody) Summary: BSOD on Windows Server 2003 with Adaptec AIC-9580W RAID Ctrl Initial Comment: I am trying to install coLinux-0.6.4.exe on a system that is running Windows server 2003 and has a Adaptec AIC-9580W RAID controller (aka ServeRAID 8k-l or 8k) installed on it. The installation gets about 98% complete and then it gives the BSOD (blus screen of death) with the STOP: 0x1111777A (0x0, 0x0, 0x0, 0x0) code at the bottom. I have been able to install colinux-0.6.4 on systems that are running Windows Server 2003, but do not have the RAID Controller HW. I was also able to install colinux-0.6.4 on a system that is running Windows 2000 and has a Adaptec AIC-9580W RAID controller (aka ServeRAID 8k-l or 8k) installed on it. I have tired various configuratoins in the boot.ini file (i.e. /NOEXECUTE, /APE, /NOAPE etc), but the BSOD still shows up. I have also tried coLinux development version 0.8.0. This version installs the linux.sys driver (using the colinux-daemon --install-driver commmand) without giving the BSOD, but when coLinux-daemon.exe is executed the BSOD shows up with the following message: IRQL_NOT_LESS_OR_EQUAL Here is some system information on the machine that has Windows Server 2003 and the Adaptec AIC-9580W RAID controller (aka ServeRAID 8k-l or 8k) installed on it: a two socket server utilizing Intel Xeon 5000 series processors Intel 5000X chipset Adaptec AIC-9580W RAID controller (aka ServeRAID 8k-l or 8k) Max memory supported is 32GB. Anyone have any suggestions as to what the cause of the BSOD is? Thank you ---------------------------------------------------------------------- Comment By: Nobody/Anonymous (nobody) Date: 2007-08-21 09:12 Message: Logged In: NO Solved! one uppon a time it happens that there was a linux.sys which did not run with mem=128, running with mem=126 was OK. That was 0.6.1 or 0.6.2! And that linux.sys was installed on the box which BSODs. After deinstalling: No problems any more! Thanks a lot! But i am pretty sure i started the colinux 0.7.1 daemon from the pen drive. Why didn't it complain about that old driver? Was the BSOD faster? Some infos over the old driver: hardware wizzard shows as details only: ROOT\LEGACY_COLINUXDRIVER\0000 dir: 29.05.2004 16:59 237.649 coLinux-0.6.1.tar.gz 05.02.2005 10:57 422.441 colinux-0.6.2.tar.gz 29.05.2004 15:20 274.784 colinux-bridged-net-daemon.exe 29.05.2004 15:20 713.682 colinux-console-fltk.exe 29.05.2004 15:20 402.411 colinux-console-nt.exe 29.05.2004 15:20 338.449 colinux-daemon.exe 29.05.2004 15:20 274.032 colinux-net-daemon.exe 14.07.2004 00:54 62.146.666 Debian-20040605-mit.ext3.1610mb.bz2 09.06.2005 15:18 18.329.320 Debian-3.0r0.ext3.1gb.bz2 10.11.2004 17:05 1.073.741.824 Debian-3.0r2.ext3-mit-backports.1gb 27.06.2004 00:36 22.449.829 Debian-3.0r2.ext3-mit-backports.1gb.bz2 25.04.2004 16:35 1.112 default.colinux.xml 10.11.2004 17:03 1.247 default.colinux.xml.$$$ 10.11.2004 17:04 1.256 default.colinux.xml.old 29.05.2004 15:20 94.327 linux.sys 09.06.2005 15:10 <DIR> netdriver 20.04.2004 06:43 5.014 README 09.06.2005 15:10 60.313 Uninstall.exe 29.05.2004 14:32 1.822.230 vmlinux 29.05.2004 15:20 641.763 vmlinux-modules.tar.gz After deinstalling the old colinux and install 0.8.0 i get, when started from pendrive: Cooperative Linux Daemon, 0.7.1 Compiled on Sat Jul 14 12:15:18 2007 using 'vmlinux' as kernel image configuring 128 MB of virtual RAM mapping cobd4 to \Device\Cdrom0 mapping cobd1 to \??\E:\colinux.disks\swap_file mapping cobd0 to \??\E:\colinux.disks\Debian-3.0r2.ext3-mit-backports.1gb mapping hdc to cobd4 mapping hdb to cobd1 mapping hda to cobd0 configured Slirp as eth0 redirections tcp:15022:22/tcp:15080:80/tcp:5901:5901 mapping cofs0 to \??\E:\colinux.disks\exchange using 'initrd.gz' as initrd image mapping cofs31 to \??\E:\coLinux kernel boot parameters: 'root=/dev/cobd0' colinux: manager open: last error: Das System kann die angegebene Datei nicht finden. daemon: exit code 8320a80e daemon: can't access CoLinuxDriver, please check status driver! back to "E:\coLinux\colinuxstart.cmd"... So i it is still a dream to run a "linux-in-a-box" without an installed driver or admin rights. Mark Russinovich (wininterals etc.) one made that trick with his "filemon" IIRC. ---------------------------------------------------------------------- Comment By: Henry N. (henryn) Date: 2007-08-21 01:56 Message: Logged In: YES user_id=579204 Originator: NO There exist an variable for total ram used by coLinux (your "mem=xx") This variable exist in a struct on daemon and will transfer to the driver linux.sys, this calls than the malloc. (has to do with co_os_physical_memory_pages() and the automatic down rounder for variable "hostmem_used") Very old coLinux has no version check (think before 0.6.4, but not remember). Current Daemon first checks the version of linux.sys and don't start. You can start coLinux from a USB stick. But you need to load ones the driver. After you have the driver updated, you can run without admin rights. If you say, this machine has never been seen coLinux before, then the daemon would not start. Check the hardware device manager and enable "hidden devices", then you can see the colinux driver, under "non PnP drivers". This scren shows the way howto find the driver by mouse clicks: http://www.henrynestler.com/colinux/screenshoots/driver-details-grayed.png Remove this driver and restart windows. Or run "colinux-daemon.exe --remove-driver" (do the same without reboot). ---------------------------------------------------------------------- Comment By: Nobody/Anonymous (nobody) Date: 2007-08-20 15:42 Message: Logged In: NO >>Comment By: Henry N. (henryn) >Date: 2007-08-20 11:58 >Message: >Logged In: YES >user_id=579204 >Originator: NO >Hello, >can't resolve the stack to source, because I can't find a matching >binary linux.sys from there you got the BSOD. I used the normal linux.sys from the windows install. I added file sizes and dates. >Your address "linux+52e6" is inside the >function co_os_malloc, >and the condition for this bug is a call "co_os_malloc(0)". Jepp. i already analyzed the sources but could not see any place where co_os_malloc etc. could(!) be called with a "0". At all locations (i found) that param was a compile time constant value excpt one. So it currently sounds like an ugly pointer error or bad case/table jump. >Are you shure, that you have called >"colinux-daemon.exe --remove-driver" >and "colinux-daemon.exe --install-driver" >between the change of coLinux versions? Hm. No. because i assumed that the machine never saw colinux before. And i was not aware that this could be a problem at all because I spawned colinux from a 4GB USB-Drive without any install because in wanted to test it without admin rights... I was very surprised to BSOD a box (without admin rights, i thought), but if an old installed driver was used, a BSOD is possible. Would colinux-daemon.exe use any installed linux.sys without any version check? That would explain the problem. I'll have a look if a (very old) linux.sys was installed long time ago tomorrow. >Replacing the file linux.sys and the exe daemons in the directory >are not enough. I thought it would, because i did not install it. Maybe some else did. My intention was do have "my own linux on a stick" but if it uses the installed stone aged linux.sys it could always crash. ---------------------------------------------------------------------- Comment By: Henry N. (henryn) Date: 2007-08-20 02:58 Message: Logged In: YES user_id=579204 Originator: NO Hello, can't resolve the stack to source, because I can't find a matching binary linux.sys from there you got the BSOD. Your address "linux+52e6" is inside the function co_os_malloc, and the condition for this bug is a call "co_os_malloc(0)". See line 63: http://colinux.svn.sourceforge.net/viewvc/colinux/branches/devel/src/colinux/os/winnt/kernel/lowlevel/alloc.c?view=markup The 1111777A exist only ones in the source. So your address "linux+52e6" should match an address ones behind the BugCheck-call. > ef6d92db 687a771111 push 1111777Ah <-- BugCheck code > ef6d92e0 ff1518e16def call dword ptr [linux+0xa118 (ef6de118)] <-- calls KeBugCheck > ef6d92e6 <--- this is the sddress you see in stack trace In your disassemble I can see. But, I have no this file, you used. We must use exactly the same linux.sys for analysis from where you got the bug. I see different build date and file size for linux.sys. Nothing of my known files matched with your disassembler output. Are you shure, that you have called "colinux-daemon.exe --remove-driver" and "colinux-daemon.exe --install-driver" between the change of coLinux versions? Replacing the file linux.sys and the exe daemons in the directory are not enough. I have added more informations to the bugcheck 1111777A to see the caller. Please do a clean install from http://www.henrynestler.com/colinux/testing/devel-0.8.0/20070819-Snapshot/ I see, you have WinDbg installed. This is very good. You can send me the output from "kd> !analyze -v" via Mail. In "debugging.txt" have added the way howto resolve the symbols from linux.sys by hands. Hope, this helps. Henry ---------------------------------------------------------------------- Comment By: Henry N. (henryn) Date: 2007-08-08 01:26 Message: Logged In: YES user_id=579204 Originator: NO BSOD 0x1111777A can have more as one initial errors (more as the known and fixed call). Many thanks for the 'STACK_TEXT:', can see the linux calls. Will check it. ---------------------------------------------------------------------- Comment By: Nobody/Anonymous (nobody) Date: 2007-08-07 08:00 Message: Logged In: NO Sorry, I forgot to emphesis: I already tried the "developer driver" as the file date "2007-07-14" indicates. So: BSOD 0x111177A is not fixed in 0.71 Have not tried 0.8 because there is no zip where i can take the driver from. But as 0.8 and 0.71 has the same patches... Have tried Debug version of 0.71, but the symbol information is not acepted by kernel debugger and thatverion ist a bit older. Debug version 0.7.1 (build on 26 March 2007) still show the BSOD. ---------------------------------------------------------------------- Comment By: Nobody/Anonymous (nobody) Date: 2007-08-07 07:51 Message: Logged In: NO linux.sys BSDOD 0x111177A coLinux-0.7.1 Internal Version: 1.1.0.0 linux.sys 69120 byte 2007-07-14 14:15 Debug: linux.sys 139.730 139.730 2007-03-09 02:42 (But symbols are not usable) Starting gives BSOD on an normal Office PC running "Windows XP 2002 SP2 Celeron 2,80GHz, 504MB RAM" 512MiB RAM, 2GiB Swap Trendmicro Virus scanner "Slirp" network So that's no problem of "too much memory", i assume. Invoked by from USB flash drive by: set ETH=eth0=slirp,,tcp:5022:22 set WINSHARE=cofs0=\colinux.disks\exchange set MEM=128 set CDROM0=cobd4=\Device\Cdrom0 hdc=:cobd4 set BOOTDEV=cobd0 set ROOTFS=%BOOTDEV%=\colinux.disks\Debian-3.0r2.ext3-mit-backports.imi set ROOTFS=%BOOTDEV%=\colinux.disks\Debian-3.0r2.ext3-mit-backports.1gb set ROOTFS=%ROOTFS% root=/dev/%BOOTDEV% hda=:%BOOTDEV% set SWAPDEV=cobd1 set SWAPFS=%SWAPDEV%=\colinux.disks\swap_file hdb=:%SWAPDEV% set OPTIONS=-k set OPTIONS=%OPTIONS% -t NT \coLinux\colinux-daemon.exe %OPTIONS% kernel=vmlinux initrd=initrd.gz mem=%MEM% %CDROM0% %SWAPFS% %ROOTFS% %WINSHARE% %ETH% Reducing paramters to minimum does not change behavior. --- Analysis of core dump: kd -z c:\Dump\MEMORY.DMP -i c:\coLinux\ -y c:\Symbols\ *** ERROR: Symbol file could not be found. Defaulted to export symbols for NDIS.sys - Probably caused by : linux.sys ( linux+52e6 ) Windows XP Kernel Version 2600 (Service Pack 2) UP Free x86 compatibl Product: WinNt, suite: TerminalServer SingleUserTS Built by: 2600.xpsp_sp2_gdr Kernel base = 0x804d7000 PsLoadedModuleList = 0x8055a620 Debug session time: Tue Aug 7 11:27:53.602 2007 (GMT+2) System Uptime: 0 days 0:25:19.109 PEB is paged out (Peb.Ldr = 7ffdc00c). Type ".hh dbgerr001" for details BugCheck 1111777A, {0, 0, 0, 0} Probably caused by : linux.sys ( linux+52e6 ) MODULE_NAME: linux FAULTING_MODULE: 804d7000 nt DEBUG_FLR_IMAGE_TIMESTAMP: 40b88e1b DEFAULT_BUCKET_ID: WRONG_SYMBOLS BUGCHECK_STR: 0x111777A LAST_CONTROL_TRANSFER: from ef6d92e6 to 80533523 STACK_TEXT: WARNING: Stack unwind information not available. Following frames may be wrong. ee441b14 ef6d92e6 1111777a 818fc65c ee441b74 nt!KeBugCheck+0x14 ee441b34 ef6d7742 00000000 00000000 ee441bc4 linux+0x52e6 ee441b74 ef6d7aa6 81beb3a8 00000000 818fc638 linux+0x3742 ee441b94 ef6d7c5b 81beb3a8 81705000 80561b40 linux+0x3aa6 ee441bc4 ef6d86c4 81beb3a8 00000011 81705000 linux+0x3c5b ee441c14 ef6d86f3 81beb2f0 81a85418 81705000 linux+0x46c4 ee441c34 804e37f7 81beb2f0 81a85418 806ee2d0 linux+0x46f3 ee441c58 8057ad03 81beb2f0 81a85418 818fc650 nt!IofCallDriver+0x32 ee441d00 8057d281 000007c8 00000000 00000000 nt!SeQueryAuthenticationIdToken+0x6c6 ee441d34 804de7ec 000007c8 00000000 00000000 nt!NtDeviceIoControlFile+0x2a ee441ddc 804f827a f8332b85 82187f48 00000000 nt!ZwYieldExecution+0xb78 ee441de0 f8332b85 82187f48 00000000 0000037f nt!KeInitializeTimer+0x107 ee441de4 82187f48 00000000 0000037f 00000000 NDIS!NdisFreeToBlockPool+0x15e1 ee441de8 00000000 0000037f 00000000 00000000 0x82187f48 STACK_COMMAND: kb FOLLOWUP_IP: linux+52e6 ef6d92e6 83c40c add esp,0Ch SYMBOL_STACK_INDEX: 1 FOLLOWUP_NAME: MachineOwner IMAGE_NAME: linux.sys SYMBOL_NAME: linux+52e6 BUCKET_ID: WRONG_SYMBOLS kd> r eax=ffdff13c ebx=00000000 ecx=00000000 edx=00000032 esi=81beb3a8 edi=818fc65c eip=80533523 esp=ee441afc ebp=ee441b14 iopl=0 nv up ei ng nz na pe nc cs=0008 ss=0010 ds=0023 es=0023 fs=0030 gs=0000 efl=00000286 nt!KeBugCheck+0x14: 80533523 5d pop ebp kd> u nt!KeBugCheck+0x14: 80533523 5d pop ebp 80533524 c20400 ret 4 80533527 90 nop kd> u ef6d92d8 linux+0x52d8: ef6d92d8 83ec0c sub esp,0Ch ef6d92db 687a771111 push 1111777Ah ef6d92e0 ff1518e16def call dword ptr [linux+0xa118 (ef6de118)] ef6d92e6 83c40c add esp,0Ch ef6d92e9 ebd5 jmp linux+0x52c0 (ef6d92c0) ef6d92eb 90 nop ef6d92ec 8d742600 lea esi,[esi] ef6d92f0 55 push ebp linux+0x52c0: ef6d92c0 83ec04 sub esp,4 ef6d92c3 68636f6c78 push 786C6F63h ef6d92c8 53 push ebx ef6d92c9 6a00 push 0 ef6d92cb ff15fce06def call dword ptr [linux+0xa0fc (ef6de0fc)] ef6d92d1 8b5dfc mov ebx,dword ptr [ebp-4] ef6d92d4 89ec mov esp,ebp ef6d92d6 5d pop ebp C:\Programme\Debugging Tools for Windows>dir d:\coLinux\linux.sys 19.03.2007 02:42 139.730 linux.sys C:\Programme\Debugging Tools for Windows>symchk.exe d:\coLinux\linux.sys SYMCHK: linux.sys FAILED - Built without debugging information. SYMCHK: FAILED files = 1 SYMCHK: PASSED + IGNORED files = 0 Source: System Error Cat: (102) ID: 1003 Error code 1111777a, 1. Parameter 00000000, 2. Parameter 00000000, 3. Parameter 00000000, 4. Parameter 00000000. 0000: 53 79 73 74 65 6d 20 45 System E 0008: 72 72 6f 72 20 20 45 72 rror Er 0010: 72 6f 72 20 63 6f 64 65 ror code 0018: 20 31 31 31 31 37 37 37 1111777 0020: 61 20 20 50 61 72 61 6d a Param 0028: 65 74 65 72 73 20 30 30 eters 00 0030: 30 30 30 30 30 30 2c 20 000000, 0038: 30 30 30 30 30 30 30 30 00000000 0040: 2c 20 30 30 30 30 30 30 , 000000 0048: 30 30 2c 20 30 30 30 30 00, 0000 0050: 30 30 30 30 0000 ---------------------------------------------------------------------- Comment By: Henry N. (henryn) Date: 2007-07-02 11:18 Message: Logged In: YES user_id=579204 Originator: NO Yes, the BSOD (0x1111777A) is fixed at 2007-03-24 in Version 0.8.0 http://colinux.svn.sourceforge.net/viewvc/colinux/branches/devel/src/colinux/os/winnt/kernel/lowlevel/misc.c?view=log BSOD also fixed in pre version 0.7.1 (build on 26 March 2007). Version 0.6.4 would no fix, is to old. The changes on linux.sys installer prevents the install of this driver, if you have more than 4GB RAM, but should no crash with IRQL_NOT_LESS_OR_EQUAL, if the daemon tries to load it. How many memory you have? 32GB? What says the screen by pressing keys Win+Pause? The main problem is: You have to many memory. CoLinux would run, if you limit your host memory to 4GB totaly RAM /MAXMEM=4096 in your C:\BOOT.INI Sorry more is not usable by current design ;-( CoLinux can not run on machines with more as 4GB RAM. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=622063&aid=1745597&group_id=98788 |
