Booting up an old Windows 3.1 harddisk image

Booting up an old Windows 3.1 harddisk image

[John Stoffel]

Hi Guys,

I'm not a subscriber to the kvm mailing list, so please copy me in
your replies.

I've got an old image of an i486 disk running (I think!) Windows 3.1
which I want to bring up and play with, just to make sure I can use
it.  

    > file hda-caroline486.image 
    hda-caroline486.image: x86 boot sector, Microsoft Windows XP mbr;
    partition 1: ID=0x6, active, starthead 1, startsector 17, 254983
    sectors, code offset 0x33

I basically just dd'd the entire disk into a file.  I can mount it
using a loop device and then examine the partitions, so I think it's
fine from that viewpoint.

My server is a Debian 5.0 box, running on AMD Quad Core CPU, 8gb of
RAM.  I've got other KVM guests running just fine.  

So I tried to use the following to boot the image:

  > sudo qemu -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -m \
  128 -vga std -vnc quad:44

And I get the following in a VNC screen:

    Starting SeaBIOS (version 0.5.1-20100616_222654-volta)

    Booting from Hard Disk...

    HIMEM: DOS XMS Driver, Version 3.07 - 02/14/92
    Extended Memory Specification (XMS) Version 3.0
    Copyright 1988-1992 Microsoft Corp.

    Installed A20 handler number 2.
    64K High Memory Area is available.


    MICROSOFT Expanded Memory Manager 386  Version 4.44
    Copyright Microsoft Corporation 1986, 1991

    _


And that's it.  So it looks like I'm missing a driver or something
here.  Do I need to define keyboard and mouse for this sucker?  The
hardware is long gone, but I think it was a Gateway.  Total guess.

Thanks for any hints,
John

Re: Booting up an old Windows 3.1 harddisk image

[Avi Kivity]

On 10/18/2011 02:59 AM, John Stoffel wrote:
> Hi Guys,
>
> I'm not a subscriber to the kvm mailing list, so please copy me in
> your replies.
>
> I've got an old image of an i486 disk running (I think!) Windows 3.1
> which I want to bring up and play with, just to make sure I can use
> it.  
>
>     > file hda-caroline486.image 
>     hda-caroline486.image: x86 boot sector, Microsoft Windows XP mbr;
>     partition 1: ID=0x6, active, starthead 1, startsector 17, 254983
>     sectors, code offset 0x33
>
> I basically just dd'd the entire disk into a file.  I can mount it
> using a loop device and then examine the partitions, so I think it's
> fine from that viewpoint.
>
> My server is a Debian 5.0 box, running on AMD Quad Core CPU, 8gb of
> RAM.  

What's the host kernel version?

> I've got other KVM guests running just fine.  
>
> So I tried to use the following to boot the image:
>
>   > sudo qemu -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -m \
>   128 -vga std -vnc quad:44

No need to sudo.

>
> And I get the following in a VNC screen:
>
>     Starting SeaBIOS (version 0.5.1-20100616_222654-volta)
>
>     Booting from Hard Disk...
>
>     HIMEM: DOS XMS Driver, Version 3.07 - 02/14/92
>     Extended Memory Specification (XMS) Version 3.0
>     Copyright 1988-1992 Microsoft Corp.
>
>     Installed A20 handler number 2.
>     64K High Memory Area is available.
>
>
>     MICROSOFT Expanded Memory Manager 386  Version 4.44
>     Copyright Microsoft Corporation 1986, 1991
>
>     _
>
>
> And that's it.  So it looks like I'm missing a driver or something
> here.  Do I need to define keyboard and mouse for this sucker?  The
> hardware is long gone, but I think it was a Gateway.  Total guess.
>

It's probably a bug in kvm.  Try collecting a complete trace as in
http://www.linux-kvm.org/page/Tracing and posting it here, maybe
something will jump out.

Also try out -no-kvm.

-- 
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.

Re: Booting up an old Windows 3.1 harddisk image

[Jan Kiszka]

On 2011-10-18 12:42, Avi Kivity wrote:
> On 10/18/2011 02:59 AM, John Stoffel wrote:
>> Hi Guys,
>>
>> I'm not a subscriber to the kvm mailing list, so please copy me in
>> your replies.
>>
>> I've got an old image of an i486 disk running (I think!) Windows 3.1
>> which I want to bring up and play with, just to make sure I can use
>> it.  
>>
>>     > file hda-caroline486.image 
>>     hda-caroline486.image: x86 boot sector, Microsoft Windows XP mbr;
>>     partition 1: ID=0x6, active, starthead 1, startsector 17, 254983
>>     sectors, code offset 0x33
>>
>> I basically just dd'd the entire disk into a file.  I can mount it
>> using a loop device and then examine the partitions, so I think it's
>> fine from that viewpoint.
>>
>> My server is a Debian 5.0 box, running on AMD Quad Core CPU, 8gb of
>> RAM.  
> 
> What's the host kernel version?
> 
>> I've got other KVM guests running just fine.  
>>
>> So I tried to use the following to boot the image:
>>
>>   > sudo qemu -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -m \
>>   128 -vga std -vnc quad:44
> 
> No need to sudo.
> 
>>
>> And I get the following in a VNC screen:
>>
>>     Starting SeaBIOS (version 0.5.1-20100616_222654-volta)
>>
>>     Booting from Hard Disk...
>>
>>     HIMEM: DOS XMS Driver, Version 3.07 - 02/14/92
>>     Extended Memory Specification (XMS) Version 3.0
>>     Copyright 1988-1992 Microsoft Corp.
>>
>>     Installed A20 handler number 2.
>>     64K High Memory Area is available.
>>
>>
>>     MICROSOFT Expanded Memory Manager 386  Version 4.44
>>     Copyright Microsoft Corporation 1986, 1991
>>
>>     _
>>
>>
>> And that's it.  So it looks like I'm missing a driver or something
>> here.  Do I need to define keyboard and mouse for this sucker?  The
>> hardware is long gone, but I think it was a Gateway.  Total guess.
>>
> 
> It's probably a bug in kvm.  Try collecting a complete trace as in
> http://www.linux-kvm.org/page/Tracing and posting it here, maybe
> something will jump out.
> 
> Also try out -no-kvm.

A shot from the hips: Jörg and I are just hunting down 16 bit task
switch issues on AMD hosts. If this problem may be related can be
checked by loading kvm-amd with npt=0, ie. disabling nested paging.

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux

Re: Booting up an old Windows 3.1 harddisk image

[John Stoffel]

>>>>> "Avi" == Avi Kivity <avi@redhat.com> writes:

Avi> On 10/18/2011 02:59 AM, John Stoffel wrote:
>> Hi Guys,
>> 
>> I'm not a subscriber to the kvm mailing list, so please copy me in
>> your replies.
>> 
>> I've got an old image of an i486 disk running (I think!) Windows 3.1
>> which I want to bring up and play with, just to make sure I can use
>> it.  
>> 
>> > file hda-caroline486.image 
>> hda-caroline486.image: x86 boot sector, Microsoft Windows XP mbr;
>> partition 1: ID=0x6, active, starthead 1, startsector 17, 254983
>> sectors, code offset 0x33
>> 
>> I basically just dd'd the entire disk into a file.  I can mount it
>> using a loop device and then examine the partitions, so I think it's
>> fine from that viewpoint.
>> 
>> My server is a Debian 5.0 box, running on AMD Quad Core CPU, 8gb of
>> RAM.  

Avi> What's the host kernel version?

I thought I was running the stock Debian version, but I'm actually
running my own 3.1.0-rc4 kernel.  A bit out of date, but I don't
remember seeing any major KVM breakage in the later part of the
3.1-rc# series.

  Linux version 3.1.0-rc4-custom (john@quad) (gcc version 4.4.5 (Debian
  4.4.5-8) ) #1 SMP Wed Aug 31 12:12:32 EDT 2011

>> I've got other KVM guests running just fine.  
>> 
>> So I tried to use the following to boot the image:
>> 
>> > sudo qemu -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -m \
>> 128 -vga std -vnc quad:44

Avi> No need to sudo.

>> And I get the following in a VNC screen:
>> 
>> Starting SeaBIOS (version 0.5.1-20100616_222654-volta)
>> 
>> Booting from Hard Disk...
>> 
>> HIMEM: DOS XMS Driver, Version 3.07 - 02/14/92
>> Extended Memory Specification (XMS) Version 3.0
>> Copyright 1988-1992 Microsoft Corp.
>> 
>> Installed A20 handler number 2.
>> 64K High Memory Area is available.
>> 
>> 
>> MICROSOFT Expanded Memory Manager 386  Version 4.44
>> Copyright Microsoft Corporation 1986, 1991
>> 
>> _
>> 
>> 
>> And that's it.  So it looks like I'm missing a driver or something
>> here.  Do I need to define keyboard and mouse for this sucker?  The
>> hardware is long gone, but I think it was a Gateway.  Total guess.
>> 

Avi> It's probably a bug in kvm.  Try collecting a complete trace as in
Avi> http://www.linux-kvm.org/page/Tracing and posting it here, maybe
Avi> something will jump out.

Avi> Also try out -no-kvm.

Thanks, I'll give this a try tonight when I'm home again.

John

Re: Booting up an old Windows 3.1 harddisk image

[Avi Kivity]

On 10/18/2011 03:08 PM, John Stoffel wrote:
> >>>>> "Avi" == Avi Kivity <avi@redhat.com> writes:
>
> Avi> On 10/18/2011 02:59 AM, John Stoffel wrote:
> >> Hi Guys,
> >> 
> >> I'm not a subscriber to the kvm mailing list, so please copy me in
> >> your replies.
> >> 
> >> I've got an old image of an i486 disk running (I think!) Windows 3.1
> >> which I want to bring up and play with, just to make sure I can use
> >> it.  
> >> 
> >> > file hda-caroline486.image 
> >> hda-caroline486.image: x86 boot sector, Microsoft Windows XP mbr;
> >> partition 1: ID=0x6, active, starthead 1, startsector 17, 254983
> >> sectors, code offset 0x33
> >> 
> >> I basically just dd'd the entire disk into a file.  I can mount it
> >> using a loop device and then examine the partitions, so I think it's
> >> fine from that viewpoint.
> >> 
> >> My server is a Debian 5.0 box, running on AMD Quad Core CPU, 8gb of
> >> RAM.  
>
> Avi> What's the host kernel version?
>
> I thought I was running the stock Debian version, but I'm actually
> running my own 3.1.0-rc4 kernel.  A bit out of date, but I don't
> remember seeing any major KVM breakage in the later part of the
> 3.1-rc# series.
>
>   Linux version 3.1.0-rc4-custom (john@quad) (gcc version 4.4.5 (Debian
>   4.4.5-8) ) #1 SMP Wed Aug 31 12:12:32 EDT 2011

That's plenty recent.  I was worried about having an old emulator.

-- 
error compiling committee.c: too many arguments to function

Re: Booting up an old Windows 3.1 harddisk image

[Avi Kivity]

On 10/18/2011 03:07 PM, Jan Kiszka wrote:
> > 
> > It's probably a bug in kvm.  Try collecting a complete trace as in
> > http://www.linux-kvm.org/page/Tracing and posting it here, maybe
> > something will jump out.
> > 
> > Also try out -no-kvm.
>
> A shot from the hips: Jörg and I are just hunting down 16 bit task
> switch issues on AMD hosts. If this problem may be related can be
> checked by loading kvm-amd with npt=0, ie. disabling nested paging.

Do you mean that you suspect that kvm's task switch emulation is more
accurate than the hardware?  Gleb will have a heart attack.

Another area to look at is A20 emulation, based on where the hang occurs.

-- 
error compiling committee.c: too many arguments to function

Re: Booting up an old Windows 3.1 harddisk image

[Jan Kiszka]

On 2011-10-18 15:13, Avi Kivity wrote:
> On 10/18/2011 03:07 PM, Jan Kiszka wrote:
>>>
>>> It's probably a bug in kvm.  Try collecting a complete trace as in
>>> http://www.linux-kvm.org/page/Tracing and posting it here, maybe
>>> something will jump out.
>>>
>>> Also try out -no-kvm.
>>
>> A shot from the hips: Jörg and I are just hunting down 16 bit task
>> switch issues on AMD hosts. If this problem may be related can be
>> checked by loading kvm-amd with npt=0, ie. disabling nested paging.
> 
> Do you mean that you suspect that kvm's task switch emulation is more
> accurate than the hardware?  Gleb will have a heart attack.

As it looks like (for AMD). But the last proof is yet missing - /me is
instrumenting right now.

> Another area to look at is A20 emulation, based on where the hang occurs.

Yeah, maybe. Traces may tell.

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux

Re: Booting up an old Windows 3.1 harddisk image

[John Stoffel]

>>>>> "John" == John Stoffel <john@stoffel.org> writes:

>>>>> "Avi" == Avi Kivity <avi@redhat.com> writes:
Avi> On 10/18/2011 02:59 AM, John Stoffel wrote:
>>> Hi Guys,
>>> 
>>> I'm not a subscriber to the kvm mailing list, so please copy me in
>>> your replies.
>>> 
>>> I've got an old image of an i486 disk running (I think!) Windows 3.1
>>> which I want to bring up and play with, just to make sure I can use
>>> it.  
>>> 
>>> > file hda-caroline486.image 
>>> hda-caroline486.image: x86 boot sector, Microsoft Windows XP mbr;
>>> partition 1: ID=0x6, active, starthead 1, startsector 17, 254983
>>> sectors, code offset 0x33
>>> 
>>> I basically just dd'd the entire disk into a file.  I can mount it
>>> using a loop device and then examine the partitions, so I think it's
>>> fine from that viewpoint.
>>> 
>>> My server is a Debian 5.0 box, running on AMD Quad Core CPU, 8gb of
>>> RAM.  

Avi> What's the host kernel version?

John> I thought I was running the stock Debian version, but I'm actually
John> running my own 3.1.0-rc4 kernel.  A bit out of date, but I don't
John> remember seeing any major KVM breakage in the later part of the
John> 3.1-rc# series.

John>   Linux version 3.1.0-rc4-custom (john@quad) (gcc version 4.4.5 (Debian
John>   4.4.5-8) ) #1 SMP Wed Aug 31 12:12:32 EDT 2011

>>> I've got other KVM guests running just fine.  
>>> 
>>> So I tried to use the following to boot the image:
>>> 
>>> > sudo qemu -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -m \
>>> 128 -vga std -vnc quad:44

Avi> No need to sudo.

Ok, didn't do it this time.  Still no luck from what I see.

>>> And I get the following in a VNC screen:
>>> 
>>> Starting SeaBIOS (version 0.5.1-20100616_222654-volta)
>>> 
>>> Booting from Hard Disk...
>>> 
>>> HIMEM: DOS XMS Driver, Version 3.07 - 02/14/92
>>> Extended Memory Specification (XMS) Version 3.0
>>> Copyright 1988-1992 Microsoft Corp.
>>> 
>>> Installed A20 handler number 2.
>>> 64K High Memory Area is available.
>>> 
>>> 
>>> MICROSOFT Expanded Memory Manager 386  Version 4.44
>>> Copyright Microsoft Corporation 1986, 1991
>>> 
>>> _
>>> 
>>> 
>>> And that's it.  So it looks like I'm missing a driver or something
>>> here.  Do I need to define keyboard and mouse for this sucker?  The
>>> hardware is long gone, but I think it was a Gateway.  Total guess.
>>> 

Avi> It's probably a bug in kvm.  Try collecting a complete trace as
Avi> in http://www.linux-kvm.org/page/Tracing and posting it here,
Avi> maybe something will jump out.

I tried doing that, but it's huge, mostly because I've got three other
KVM sessions running.  Let me poke at the trace-command to see how I
can limit it to just a single process to make it smaller.

Avi> Also try out -no-kvm.

This doesn't exist in my version of 'qemu' or should I really be doing
something else?  I'm running:

   > qemu --version
   QEMU PC emulator version 0.12.5 (Debian 0.12.5+dfsg-3squeeze1),
   Copyright (c) 2003-2008 Fabrice Bellard

Oh wait, I see, I should be doing:

    > kvm -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -m 128 -vga std -vnc quad:4
    open /dev/kvm: Permission denied
    Could not initialize KVM, will disable KVM support

instead.  And it looks like I don't have KVM support anyway when I run
as myself.

John> Thanks, I'll give this a try tonight when I'm home again.

Ok, so I ssh'd into home from work and now I can see it starting up,
but it's horribly slow.  Heh.  

So I then tried bumping down the memory to just 4 megs (-m 4) to see
what would happen.  No change.  

Thanks for all your help guys.
John


-- 

Re: Booting up an old Windows 3.1 harddisk image

[Avi Kivity]

On 10/18/2011 07:08 PM, John Stoffel wrote:
> Avi> Also try out -no-kvm.
>
> This doesn't exist in my version of 'qemu' or should I really be doing
> something else?  I'm running:
>
>    > qemu --version
>    QEMU PC emulator version 0.12.5 (Debian 0.12.5+dfsg-3squeeze1),
>    Copyright (c) 2003-2008 Fabrice Bellard
>
> Oh wait, I see, I should be doing:
>
>     > kvm -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -m 128 -vga std -vnc quad:4
>     open /dev/kvm: Permission denied
>     Could not initialize KVM, will disable KVM support
>
> instead.  And it looks like I don't have KVM support anyway when I run
> as myself.
>
> John> Thanks, I'll give this a try tonight when I'm home again.
>
> Ok, so I ssh'd into home from work and now I can see it starting up,
> but it's horribly slow.  Heh.  

Ok, so tcg works?  It's a kvm bug then.

Please try with npt=0 (module parameter to kvm-amd), and also post a
trace as per http://www.linux-kvm.org/page/Tracing.

-- 
error compiling committee.c: too many arguments to function

Re: Booting up an old Windows 3.1 harddisk image

[Jun Koi]

On Wed, Oct 19, 2011 at 1:31 AM, Avi Kivity <avi@redhat.com> wrote:
> On 10/18/2011 07:08 PM, John Stoffel wrote:
>> Avi> Also try out -no-kvm.
>>
>> This doesn't exist in my version of 'qemu' or should I really be doing
>> something else?  I'm running:
>>
>>    > qemu --version
>>    QEMU PC emulator version 0.12.5 (Debian 0.12.5+dfsg-3squeeze1),
>>    Copyright (c) 2003-2008 Fabrice Bellard
>>
>> Oh wait, I see, I should be doing:
>>
>>     > kvm -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -m 128 -vga std -vnc quad:4
>>     open /dev/kvm: Permission denied
>>     Could not initialize KVM, will disable KVM support
>>
>> instead.  And it looks like I don't have KVM support anyway when I run
>> as myself.
>>
>> John> Thanks, I'll give this a try tonight when I'm home again.
>>
>> Ok, so I ssh'd into home from work and now I can see it starting up,
>> but it's horribly slow.  Heh.
>
> Ok, so tcg works?  It's a kvm bug then.
>
> Please try with npt=0 (module parameter to kvm-amd), and also post a
> trace as per http://www.linux-kvm.org/page/Tracing.
>

if i understand correctly, this tracing facility only supports Linux
guest, but not Windows guest?

thanks,
Jun

Re: Booting up an old Windows 3.1 harddisk image

[Jun Koi]

On Wed, Oct 19, 2011 at 10:02 AM, Jun Koi <junkoi2004@gmail.com> wrote:
> On Wed, Oct 19, 2011 at 1:31 AM, Avi Kivity <avi@redhat.com> wrote:
>> On 10/18/2011 07:08 PM, John Stoffel wrote:
>>> Avi> Also try out -no-kvm.
>>>
>>> This doesn't exist in my version of 'qemu' or should I really be doing
>>> something else?  I'm running:
>>>
>>>    > qemu --version
>>>    QEMU PC emulator version 0.12.5 (Debian 0.12.5+dfsg-3squeeze1),
>>>    Copyright (c) 2003-2008 Fabrice Bellard
>>>
>>> Oh wait, I see, I should be doing:
>>>
>>>     > kvm -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -m 128 -vga std -vnc quad:4
>>>     open /dev/kvm: Permission denied
>>>     Could not initialize KVM, will disable KVM support
>>>
>>> instead.  And it looks like I don't have KVM support anyway when I run
>>> as myself.
>>>
>>> John> Thanks, I'll give this a try tonight when I'm home again.
>>>
>>> Ok, so I ssh'd into home from work and now I can see it starting up,
>>> but it's horribly slow.  Heh.
>>
>> Ok, so tcg works?  It's a kvm bug then.
>>
>> Please try with npt=0 (module parameter to kvm-amd), and also post a
>> trace as per http://www.linux-kvm.org/page/Tracing.
>>
>
> if i understand correctly, this tracing facility only supports Linux
> guest, but not Windows guest?
>

hmm think again, since it seems there is no requirement on guest, i
guess this supports all kind of guest OS.
does this mean we can trace every instructions executed in both
userspace & kernelspace of the guest?

thanks,
Jun

Re: Booting up an old Windows 3.1 harddisk image

[John Stoffel]

Avi & Guys,

Thanks for your patience, I just got the sucker booting by switching
the -M option to be an isapc system.  And I was wrong about it being a
Gateway, it's a packard Bell system running Windows 3.1, talk about a
blast from the past.  I think I still have the keyboard around
somehwere.

Anyway, here's my commanto run it, with npt=0 in the kvm_amd modules.

  > kvm -no-acpi -no-hpet -cpu 486 -hda hda-caroline486.image -M isapc \
    -m 4 -vga std

and I'll see about changing the defaults back on.  

Thanks,
John

--