Host Crash reproducible

[PAETH CLAUDIUSRAPHAEL]

Hi, i finally add it as a BUG here AFTER following all advices ( IRC, testing, reproducing failure, compaing results, check forums, etc. );

[ Sorry for sounding washy, im a bit tired now of all that crap, hopefully you are not and can take a deep look in it ]

Definition of the Bug as far as i can describe:

[HOST-SYSTEM]

[Soft-Ware:OS]

The Host Operating Systems i can reproduce these bugs on ( so far )are:

a) Windows NT 6.3.1 x86 U1 [ Windows 8.1 ( Embedded / Enterprise )Update 1 32-Bit ] NVHD

b) Windows NT 6.1.2 x86 [ Windows 7 ( Embedded Standard / Enterprise ) ] Service Pack 2 32-Bit ] NVHD

[Soft-Ware:VBOX]

VirtualBox 4.3.28 r100309 with ext-pack same version

[same effect/other symptoms in]

VirtualBox 4.3.29 r100811 with ext-pack from released version r100309 AND extpack for testing version [r100811]

HINT: Any version before is running ( not fine, but running )!

[Hard-Ware]

[A:AMD]

• Dell Optiplex 740
• CPU: AMD-Athlon X2 2.7GHz ( AMD-VX )
• NIC:
• onboard lan Marvell yukon GBit
• Wifi 54g D-link (PCI) originally; was exchanged with a firewire-pci
• 100mbit lan(pci with pxe-bios)

SND:

• onboard realtek ac 97

VGA:

• onboard nvidia ? ( unused)
• Ati v5700 Fire-GL (pci-express 16x)

HDD:

• 4x Seagate Barracuda ES 750GB 15k (non-raid use on onboard SATA AHCI )

ODD:

• external LG DVD+-RW whatever USB2

USB:

• 4 RootHubs ( 3 in use ) USB 2

[B:Intel]

• FS Esprimo

CPU:

• Pentium D 945 Presler ( VT-X)

HDD:

• 1x Samsung 160GB ( internal SATA AHCI )
• 1x IDE changeable rack
• 1x ide smart card/Compactflash-boot/SDHC/MS

VGA:

• Intel 945G (EE)

SND:

• onboard intel HD

ODD:

• none

HINT: These machines are in amounts up to 50 available, were in seldom use and bought from local energy provider as sorted-out and should be used in future for kindergarten-network as experimental stations; they are already checked and stress-tested and i interchanged systems with the given setup just to ensure it is not a single-machine that is faulty. Btw.: all these machines ran in energy efficient lowest modes all their lifetime just used for some bureaucracy, so even if cpu-ram-gpu-stresstest weren't reliable, they function normally in any other case ( before and after reproducing the given failure of vbox )

What is the problem ?

In my webchat IRC with MichalN in the #vbox channel i had kind brainstorm of possibilities, because i wouldn't report a bug without further investigations. Whatever, the problem to report a bug is, i can get NO MINIDUMP at all, independent of settings, because due to bugs in the given version, from a given OS in native VHD-Boot there are issues which in most cases, do not lead to a dump. believe me i tried all options. That is why i asked at vbox-channel for possible workarounds, like special logging-modes for vbox or possibilities to sandbox vbox, so the host-system can't crush and can be investigated. Nonen option so far given.

The problem in its symptomatic:

When running 64-Bit guests the host system crushes almost immediately! Linux live systems run. ( at least in the release version ) in the testing-version they crush also immediately and with a bsod clock_watchdog_timeout for some millisecs of the host - followed by immediate restart and no dump at all ); at least in testing version vbox comes so far to log the usual vm-init inside. in the release version it never got so far.

Tested systems:

• 64-bit guest on 32-bit host is the problem;
• i counterchecked 32-bit ones also

• Fedora
• 17
• 19
• 21

( Workstation, Server, Live, DVD , CD, Netinstall )

• Ubuntu
• all LTS versions plus vivid (15.04) and winy (15.10)

• Windows
• any you can think of

• Debian
• Jessie ( 8 )
• Wheezy ( 7.1 and 7.6 )
• did i forget some, maybe ...

okay to get some more specific things:

side-problem NIC

as MichalN mentioned, due to change from wifi to onboard lan i checked that ones too, it makes a difference in time it takes until crash, but nevertheless changes nothing at all; so forget about it; right now in this moment i am running the stable-version 4.2.30 r100415 with winy werewolf (15.10) besides this ticket-creation without a problem with full install and stress-test inside 64 bit vm on the A-Host, the AMD-one, with all options given and any hardware acceleration and all hardware on the host enabled, soi can say -- It is NOT a HW-problem and not even a host-problem i would say.

I know it is problematic without a dump at all; however let me go on with that:

i tested any old version i could find in my libraries and which i regularly used on all my machines, especially on the given ones. Btw. i just checked if it is dependent of the generation/featuresets by checking on m HP Elitebooks, too and on a panasonic toughbook; It is not; That ones were Core-duo Core2-Duo Turion X2 i5; Where was I, ah yes: Old versions

Old versions of VBOX tested so far:

• 3.2.8
• 4.0.8-71778
• 4.1.12
• 4.1.14-77440
• 4.1.22-80657
• 4.2.0-80737
• 4.2 ( 4.2.16 + 4.2.4[r81684] )
• 4.3 ( 4.3.6[r91406] + 4.3.8[92456] )

... these ones are counterchecked on linux hosts as well, which have not at all any problem with the 64guests on 32hosts (whereas the mentioned systems above that were used as guests i have also available as hosts ) plus the actual

release:

• 4.3.28-100309

testing:

• 4.3.29-100811

and the actual stable:

• 4.2.30-100415

Here is what i came to as a definitive reproducable point:

When using the actual release or testing:

-ONLY Linux-live systems based on debian run at all;

• UNTIL fully installed, the CRASH follows right after logging in the created user-account first-time.

• Windows-64-Guests ( Setup OR OVA OR prepared vhd's dism/firadisk/grub4dos) crash immediately.

• Linux-64-Guests crash right in the moment a user account is logged in in a graphical system

• Fedora-based linuxes are critically slow

• debian-based run acceptable up to perfect

• ovas with virtualization features in it are crushing in the same manner, comparable to Fedora 21 Workstation, which is the fastest crusher of all tested systems(IMMEDIATE-RIGHT-NOW-TOTAL-RESTART-OF-HOST)

Okay that was long enough - i am tired and have no concentration anymore, please see this as a possibility to make something of it; i am available for further test-scenarios, BUT DO NOT FORGET: the given HOST-OSs can not produce a dump.

I saved the chat-log of my chat with MichalN if it helps and am willed to help out. added also 2 screenshots just for a prove of stable is running ubuntu winy and a short device overview.

Give me reproducable test-case and i follow to my best; If necessary i can make a video of the crash and the none existance of the dump if you wish to. In case you do not believe. I have visual studio available if it helps and am able to install a set of progs to help narrow the problem, because these ones are not really production systems at the moment. Only thing is you have to explain what it is for if i dont know and i might double check for resulting contents, because i run some more networks chained where these machines are appended to.

To be precise: i am willed to spend my time to check on the HOST-A ( the AMD ) as the Intel-Machines go out for setup and deliverance for a while for other projects.

The other systems are my personal ones and i am not willed to make crucial tests on them anymore as long as crazy shutdowns are not handled.

So whatever: I hope you have an idea of an idea of an idea of what might could eventually maybe happen here and feel able to give some straight questions.

Thanks for your interest.

PAETH CLAUDIUSRAPHAEL S466531257BOSS

P.s.:

By the way, if you think that it is not necessary to give compatibility to these OLD machines then i need you to tell that clearly and unmisleading and ALSO clarify the tech-specs of your products for the given case. If instead you accept that these kind of machines are right now even in industry-leading nations like germany still in use by MANY private AND business users and organizations, especially those who teach and take care of our children then PLEASE do me and maybe others the chance to make it usable. Your decision. I just ask to be precise about this issue.

[PAETH CLAUDIUSRAPHAEL]

Short-Device-Overview

[PAETH CLAUDIUSRAPHAEL]

VBox-Stable-Runs-Ubuntu-Winy-On-AMD

[PAETH CLAUDIUSRAPHAEL]

VBox-IRC-Chat-Log--Discussion-About-Possibilities-Of-Total-crash

[Frank Mehnert]

Sorry but it is very difficult to get the required information from your above report. Perhaps you could take some time and read your comments and remove any prose before hitting the submit button. Actually I don't worry if you are tired of if Germany is a leading industry nation, such prose does not belong to a bugtracker ticket.

Regarding your problem: You listed several old versions of VBox but did not clearly state which of these versions work for you (don't crash) or not. Or is it 4.2.30 and every version after it (all version 4.3.x crash the host)? And did you try the latest 5.0 release candidate 1?

Also could you please attach a VBox.log file from a VM session when you run your VM with 4.3.28? If I understood you right then you are able to start the VM and eventually it crashes.

[PAETH CLAUDIUSRAPHAEL]

Replying to frank: good to see someone has an eye on it so far ;) thanks

[PAETH CLAUDIUSRAPHAEL]

Yes as i stated all versions given run fine BUT not the actual release and testing

[PAETH CLAUDIUSRAPHAEL]

i found the only log it made, in that one where i reduced the proc speed to 40% and braked the host os manually with some operations, ill add it now

[PAETH CLAUDIUSRAPHAEL]

Only-Run-Where-Log-Was-Created-Due-To-Heavy-BrakingOut-Of-Host

[PAETH CLAUDIUSRAPHAEL]

In regard to my comment belonging to germany as an example for a nation which uses these kind of machines nowadays, even if they are that old: It was a question if you look exactly. The question was: Is there any known incompatibility with machines of these specs? If so i haven't found and would like you to precise this information, as for example many organizations like the described ones are in need of this old stuff and virtualization solutions like VirtualBox running on them, because there is no money at all and it is therefore a very need to be able to run those machines in future too. Just to be clear about that and in hope i got to an understandable point now. If not, forget about it.

[PAETH CLAUDIUSRAPHAEL]

As you are really working heavily on it ( prove me wrong ) here is a message for you:

By the way and maybe lastly:

A) I can not see where i propose to make a prose or any kind of offensive verbal tackle to you in person or to your organization or one of that members in general. If you ( in person ) are able to read and think and read your comments before prose than we have an option to accept that:

B) I did the whole day nothing but testing aggressively in a structured manner on a wholeset of given possibilities to find a solution to a definitive problem, because the machine-setup where this is happening does not give a dump at all. So if you be so kind to accept that, too, you might accept also, that:

C) A comment, placed at the end of my bug report in straight words articulating the need of specific platform information for compatibility to even older systems, that are technically able to spend the features needed and are up to the actual version or at least to the last one before the actual one of those that are listed and tested and used actively, is in no case a prose; especially if a real scenario is given and an indirect description why i do this work and how necessary it is, that your product is delivering these functions. Because your product is the only product that is in any case a choice to let these by example defined institutions let survive. Especially if you accept, that

D) Oracle, or to be detailed www.virtualbox.org, if this makes a difference for you (in person) is giving just one final possibility for problems like the one i showed off; There is no interactive form to press available informations into a scheme that may make it workable and in your ( you, in person ) eyes workable and in no way understandable as a prose, but as a kind relief to help out for the hope of being helped. So, if you can accept that actually the number of machines used or will be used in this kind of scenario is approximately ( to my least knowledge ) in a range of about 20k ( though not under my range of giveable help ) then you might realize, why you (in person) should be glad that you can help with that, especially as i am willed to go on further with you, IF YOU CAN PROVIDE A SCHEME TO DO SO; So, if you can accept that, too, then:

E) Take your head out of your scheme kid and do some for mankind, because ( though i know, due to the relations to partner-projects in nations like Kenya, Ghanda, Brazil and Israel and some more nations like these one given FOR EXAMPLE ) are not even in a reach of these techniques or techs; So you might realize, that there is also a potential for Oracle or VirtualBox, to be ongoing the ONLY GLOBAL PLAYER In THIS NICHE ); so hopefully you accepted that, too, then

F ) yourself, read your comment and deliver this document inhouse to any interested person and realize that you have made a big fault, because you obviously are not able to think, or to express it in detail and schematic: You ( in person ) are in need of a schematic way to handle problems, but You ( in person ) have obviously no idea of what living and helping is, besides the fact that before someone can fillout a scheme, someone has to fit in a scheme and set up a scheme. So do your work, homeboy and never again get in touch with any other living machine ( they are called humans and animals if you need an example ), until you accept that, too.

Just to be sure you understand me IN DETAIL and IN PERSON : If you wish to be handled carefully and kindly, then act yourself carefully and kindly.

Thanks for your interest.

P.s.: I am 37 years old. Fought me up from non-existance out of a non-home and drug-fooled parental advisory to a self-employed, beneficial and socially active worker helping others to help until my PolyArthritis which runs me actually to total physical knockout actively since 1996, followed by ParkinsonSympthomatics due to heavy mismedication with opioides (its what you in person call painkillers) and i am not giving up on anything, because i am also a father of a 3yrs old daughter which is in need of you know what? Come on guess schematic toyboy... a f'n old pc at kindergarten and later at school running virtualbox to have access to linux-systems which are delivering excellent software driven by the opensource community and people like me ( not by you, im sure, at least not in person, but you may give me advice to know it better ), but the problem is, there is no money, and that crappy pitch of cash, what you ( in person may swallow to none at one evening ) that is needed to have a not-so-old machine that runs these needed tools directly, without the effort of adminastrating and orchestrating them are, guess what my dear fellow, not in any case available or even affordable if they exist. So if you accept, that too:

G) you get to work . NOW.

I offer you a restart from ground-zero for the mental health of yourself, but if you think you can play on this kind of stuff like a nervous wannabe-gee-too-dumb-too-be-a-member-of-allday-chronical-poverty, go on; i won't hinder you in any way: i just watch and see you dyin of your acts. So hopefully you at least are able to READ, before you PROSE and you are also able to accept:

That ohne has to do some to get others getting done. Wisdom stops now. Show what you are. Or let it be, let it be, let it be. I think i find my own solution, for those that need it. As always.

Hope you have a nice stay on planet earth.

Thanks for your interest.

PAETH CLAUDIUSRAPHAEL S466531257 BOSS

[Frank Mehnert]

I suggest you read the bugtracker guidelines which you can find here. Most of the stuff which you wrote is useless for this bug report. If you want to talk, go to the forums. If you want to have bugs resolved, provide the required information in a dense form without superfluous explanations which are not really relevant to help finding the problem.

Sorry, if you don't follow the rules nobody will help you.

[Michael Thayer]

Since I have read the description and comments anyway, this is my understanding of the problem:

VirtualBox 4.3.28 and testing cause two 64-bit host systems running Windows 7 and 8.1 32-bit versions to blue-screen when running various 64-bit Debian-based guests. VirtualBox 4.3.8 and various earlier versions have been found not to show this problem.

I'm afraid that I am not the person to help with Windows host problems and don't know if anyone who can will have enough free development time to look at this. If you are interested in investigating this further yourself, a good start would be to find out in exactly which version of VirtualBox this behaviour first appeared (i.e. by trying versions between 4.3.8 and 4.3.28 and finding the oldest one which is affected).