[Qemu-devel] Problem with QEMU on KVM

[Qemu-devel] Problem with QEMU on KVM

[Gerhard Wiesinger]

Hello,

Booting DOS 6.22 with NT4 Bootloader works fine. With enabled KVM it 
boots, but hangs at counting down the boot loader selection menu (30s, 
29s, hang). Booting Knoppix 6.2 from CD wwith KVM orks well.

Versions (0.12.3):
x86_64-softmmu: OK
i386-softmmu: OK
x86_64-softmmu with -enable-kvm: Boot problem
i386-softmmu with -enable-kvm: Boot problem

Host is: Fedora 11
2.6.30.10-105.2.23.fc11.x86 on AMD Phenom II Quad Core

Any ideas?

Thanx.

Ciao,
Gerhard

--
http://www.wiesinger.com/

Re: [Qemu-devel] Problem with QEMU on KVM

[Mulyadi Santosa]

Hi...

On Sun, Apr 11, 2010 at 03:09, Gerhard Wiesinger <lists@wiesinger.com> wrote:
> Hello,
>
> Booting DOS 6.22 with NT4 Bootloader works fine. With enabled KVM it boots,
> but hangs at counting down the boot loader selection menu (30s, 29s, hang).
> Booting Knoppix 6.2 from CD wwith KVM orks well.
>
> Versions (0.12.3):
> x86_64-softmmu: OK
> i386-softmmu: OK
> x86_64-softmmu with -enable-kvm: Boot problem
> i386-softmmu with -enable-kvm: Boot problem
>
> Host is: Fedora 11
> 2.6.30.10-105.2.23.fc11.x86 on AMD Phenom II Quad Core

IMHO, the problem is, we have to find out first, how the bootloader do
the timing. Is it using tight loop? Counting from TSC? PIT? or else?

I suggest to dump the executed TB in Qemu. Please read about -d option.

NB: I am not an expert, just trying to help....

-- 
regards,

Mulyadi Santosa
Freelance Linux trainer and consultant

blog: the-hydra.blogspot.com
training: mulyaditraining.blogspot.com

Re: [Qemu-devel] Problem with QEMU on KVM

[Gerhard Wiesinger]

On Sun, 11 Apr 2010, Mulyadi Santosa wrote:

> Hi...
>
> On Sun, Apr 11, 2010 at 03:09, Gerhard Wiesinger <lists@wiesinger.com> wrote:
>> Hello,
>>
>> Booting DOS 6.22 with NT4 Bootloader works fine. With enabled KVM it boots,
>> but hangs at counting down the boot loader selection menu (30s, 29s, hang).
>> Booting Knoppix 6.2 from CD wwith KVM orks well.
>>
>> Versions (0.12.3):
>> x86_64-softmmu: OK
>> i386-softmmu: OK
>> x86_64-softmmu with -enable-kvm: Boot problem
>> i386-softmmu with -enable-kvm: Boot problem
>>
>> Host is: Fedora 11
>> 2.6.30.10-105.2.23.fc11.x86 on AMD Phenom II Quad Core
>
> IMHO, the problem is, we have to find out first, how the bootloader do
> the timing. Is it using tight loop? Counting from TSC? PIT? or else?
>
> I suggest to dump the executed TB in Qemu. Please read about -d option.
>
> NB: I am not an expert, just trying to help....

OK, uses the following ports:
Port 0x20: 8259 interrupt controller
Port 0x40: 8253 timer

Interrupt 0x1A:
ah=0x00: fetches system timer counters
ah=0x02: reads the clock
ah=0x04: fetches date

So there must be something wrong with KVM with the above functionality (I 
guess the timers).

Ciao,
Gerhard

--
http://www.wiesinger.com/

Re: [Qemu-devel] Problem with QEMU on KVM

[Mulyadi Santosa]

Hi Gerhard...

On Sun, Apr 11, 2010 at 20:52, Gerhard Wiesinger <lists@wiesinger.com> wrote:
> OK, uses the following ports:
> Port 0x20: 8259 interrupt controller
> Port 0x40: 8253 timer
>
> Interrupt 0x1A:
> ah=0x00: fetches system timer counters
> ah=0x02: reads the clock
> ah=0x04: fetches date
>
> So there must be something wrong with KVM with the above functionality (I
> guess the timers).

Hmm, my silly guess is, maybe the timer is seen as decremented....or
at least "stuck".

I have very little knowledge about git, but maybe you can start doing
git bisect to narrow which git commit that introduce such behaviour.
Meanwhile, let's wait for comments from one of the KVM developers.

-- 
regards,

Mulyadi Santosa
Freelance Linux trainer and consultant

blog: the-hydra.blogspot.com
training: mulyaditraining.blogspot.com

Re: [Qemu-devel] Problem with QEMU on KVM

[Jamie Lokier]

Mulyadi Santosa wrote:
> Hi Gerhard...
> 
> On Sun, Apr 11, 2010 at 20:52, Gerhard Wiesinger <lists@wiesinger.com> wrote:
> > OK, uses the following ports:
> > Port 0x20: 8259 interrupt controller
> > Port 0x40: 8253 timer
> >
> > Interrupt 0x1A:
> > ah=0x00: fetches system timer counters
> > ah=0x02: reads the clock
> > ah=0x04: fetches date
> >
> > So there must be something wrong with KVM with the above functionality (I
> > guess the timers).
> 
> Hmm, my silly guess is, maybe the timer is seen as decremented....or
> at least "stuck".
> 
> I have very little knowledge about git, but maybe you can start doing
> git bisect to narrow which git commit that introduce such behaviour.
> Meanwhile, let's wait for comments from one of the KVM developers.

There are various -no-kvm-XXX options to try:

-no-kvm-irqchip disable KVM kernel mode PIC/IOAPIC/LAPIC
-no-kvm-pit     disable KVM kernel mode PIT
-no-kvm-pit-reinjection disable KVM kernel mode PIT interrupt reinjection

Any of them might be causing this problem.

They disable in-kernel emulation of the interrupt controller and timer
chips, and force the qemu version to be used.  It's possible the
in-kernel emulation is buggier than the qemu version.

-- Jamie

Re: [Qemu-devel] Problem with QEMU on KVM

[Mulyadi Santosa]

Hi Jamie...

On Mon, Apr 12, 2010 at 19:07, Jamie Lokier <jamie@shareable.org> wrote:
> There are various -no-kvm-XXX options to try:
>
> -no-kvm-irqchip disable KVM kernel mode PIC/IOAPIC/LAPIC
> -no-kvm-pit     disable KVM kernel mode PIT
> -no-kvm-pit-reinjection disable KVM kernel mode PIT interrupt reinjection

I try to read the source code without too much luck understanding the
meaning of the above parameters, especially the reinjection. Please
CMIIW, interrupt reinjection is a way to handle lost ticks, right?


-- 
regards,

Mulyadi Santosa
Freelance Linux trainer and consultant

blog: the-hydra.blogspot.com
training: mulyaditraining.blogspot.com

[Qemu-devel] Re: Problem with QEMU on KVM

[Jan Kiszka]

Mulyadi Santosa wrote:
> Hi Jamie...
> 
> On Mon, Apr 12, 2010 at 19:07, Jamie Lokier <jamie@shareable.org> wrote:
>> There are various -no-kvm-XXX options to try:
>>
>> -no-kvm-irqchip disable KVM kernel mode PIC/IOAPIC/LAPIC
>> -no-kvm-pit     disable KVM kernel mode PIT
>> -no-kvm-pit-reinjection disable KVM kernel mode PIT interrupt reinjection
> 
> I try to read the source code without too much luck understanding the
> meaning of the above parameters, especially the reinjection. Please
> CMIIW, interrupt reinjection is a way to handle lost ticks, right?

Those switches only exist in the qemu-kvm branch. The fact that
-enable-kvm has some effect for you makes me think that you tried
upstream KVM support (ie. the one that comes with vanilla QEMU), right?

It would be interesting to check if you can reproduce the problem also
with qemu-kvm(-0.12.3 or git head). As you may also face an issue of the
kvm kernel bits, it would be furthermore helpful to check with the
latest kernel from kvm.git or the modules build via kvm-kmod as well.

Jan

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

Re: [Qemu-devel] Problem with QEMU on KVM

[Gerhard Wiesinger]

On Mon, 12 Apr 2010, Jamie Lokier wrote:

> Mulyadi Santosa wrote:
>> Hi Gerhard...
>>
>> On Sun, Apr 11, 2010 at 20:52, Gerhard Wiesinger <lists@wiesinger.com> wrote:
>>> OK, uses the following ports:
>>> Port 0x20: 8259 interrupt controller
>>> Port 0x40: 8253 timer
>>>
>>> Interrupt 0x1A:
>>> ah=0x00: fetches system timer counters
>>> ah=0x02: reads the clock
>>> ah=0x04: fetches date
>>>
>>> So there must be something wrong with KVM with the above functionality (I
>>> guess the timers).
>>
>> Hmm, my silly guess is, maybe the timer is seen as decremented....or
>> at least "stuck".
>>
>> I have very little knowledge about git, but maybe you can start doing
>> git bisect to narrow which git commit that introduce such behaviour.
>> Meanwhile, let's wait for comments from one of the KVM developers.
>
> There are various -no-kvm-XXX options to try:
>
> -no-kvm-irqchip disable KVM kernel mode PIC/IOAPIC/LAPIC
> -no-kvm-pit     disable KVM kernel mode PIT
> -no-kvm-pit-reinjection disable KVM kernel mode PIT interrupt reinjection
>
> Any of them might be causing this problem.
>
> They disable in-kernel emulation of the interrupt controller and timer
> chips, and force the qemu version to be used.  It's possible the
> in-kernel emulation is buggier than the qemu version.

Hello Jamie, thanx so far.

I'm starting in this way:
/root/download/qemu/qemu-0.12.3/x86_64-softmmu/qemu-system-x86_64 -enable-kvm -a_lot_of_other_options

All options you mentioned don't work:
/root/download/qemu/qemu-0.12.3/x86_64-softmmu/qemu-system-x86_64: invalid option -- '-no-kvm-irqchip'
/root/download/qemu/qemu-0.12.3/x86_64-softmmu/qemu-system-x86_64: invalid option -- '-no-kvm-pit'
/root/download/qemu/qemu-0.12.3/x86_64-softmmu/qemu-system-x86_64: invalid option -- '-no-kvm-no-kvm-irqchip'

Is there another way of startup or is it working only on newer versions?

BTW: For me it is very unclear what's the difference between the 
following commands:
qemu
qemu -enable-kvm
kvm

How can I build the kvm executable? Different source tree? Newer version 
from branch?

Maybe someone can explain the different versions of qemu/kvm and branches 
etc. and possible features in detail.

Thnx.

Ciao,
Gerhard

--
http://www.wiesinger.com/

[Qemu-devel] Re: Problem with QEMU on KVM

[Gerhard Wiesinger]

On Tue, 13 Apr 2010, Jan Kiszka wrote:

> Mulyadi Santosa wrote:
>> Hi Jamie...
>>
>> On Mon, Apr 12, 2010 at 19:07, Jamie Lokier <jamie@shareable.org> wrote:
>>> There are various -no-kvm-XXX options to try:
>>>
>>> -no-kvm-irqchip disable KVM kernel mode PIC/IOAPIC/LAPIC
>>> -no-kvm-pit     disable KVM kernel mode PIT
>>> -no-kvm-pit-reinjection disable KVM kernel mode PIT interrupt reinjection
>>
>> I try to read the source code without too much luck understanding the
>> meaning of the above parameters, especially the reinjection. Please
>> CMIIW, interrupt reinjection is a way to handle lost ticks, right?
>
> Those switches only exist in the qemu-kvm branch. The fact that
> -enable-kvm has some effect for you makes me think that you tried
> upstream KVM support (ie. the one that comes with vanilla QEMU), right?
>

I tried 0.12.3 stable and also qemu git master:
http://git.savannah.gnu.org/cgit/qemu.git/log/

> It would be interesting to check if you can reproduce the problem also
> with qemu-kvm(-0.12.3 or git head). As you may also face an issue of the
> kvm kernel bits, it would be furthermore helpful to check with the
> latest kernel from kvm.git or the modules build via kvm-kmod as well.

Can you please post links to the different git repositories, since a lot 
of different repositories exist here (qemu-kvm and kernel).

Thnx.

Ciao,
Gerhard

--
http://www.wiesinger.com/

[Qemu-devel] Re: Problem with QEMU on KVM

[Jan Kiszka]

Gerhard Wiesinger wrote:
> On Tue, 13 Apr 2010, Jan Kiszka wrote:
> 
>> Mulyadi Santosa wrote:
>>> Hi Jamie...
>>>
>>> On Mon, Apr 12, 2010 at 19:07, Jamie Lokier <jamie@shareable.org> wrote:
>>>> There are various -no-kvm-XXX options to try:
>>>>
>>>> -no-kvm-irqchip disable KVM kernel mode PIC/IOAPIC/LAPIC
>>>> -no-kvm-pit     disable KVM kernel mode PIT
>>>> -no-kvm-pit-reinjection disable KVM kernel mode PIT interrupt reinjection
>>> I try to read the source code without too much luck understanding the
>>> meaning of the above parameters, especially the reinjection. Please
>>> CMIIW, interrupt reinjection is a way to handle lost ticks, right?
>> Those switches only exist in the qemu-kvm branch. The fact that
>> -enable-kvm has some effect for you makes me think that you tried
>> upstream KVM support (ie. the one that comes with vanilla QEMU), right?
>>
> 
> I tried 0.12.3 stable and also qemu git master:
> http://git.savannah.gnu.org/cgit/qemu.git/log/
> 
>> It would be interesting to check if you can reproduce the problem also
>> with qemu-kvm(-0.12.3 or git head). As you may also face an issue of the
>> kvm kernel bits, it would be furthermore helpful to check with the
>> latest kernel from kvm.git or the modules build via kvm-kmod as well.
> 
> Can you please post links to the different git repositories, since a lot 
> of different repositories exist here (qemu-kvm and kernel).

http://www.linux-kvm.org/page/Code

HTH,
Jan

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

Re: [Qemu-devel] Problem with QEMU on KVM

[Gerhard Wiesinger]

On Sat, 10 Apr 2010, Gerhard Wiesinger wrote:

> Hello,
>
> Booting DOS 6.22 with NT4 Bootloader works fine. With enabled KVM it boots, 
> but hangs at counting down the boot loader selection menu (30s, 29s, hang). 
> Booting Knoppix 6.2 from CD wwith KVM orks well.
>
> Versions (0.12.3):
> x86_64-softmmu: OK
> i386-softmmu: OK
> x86_64-softmmu with -enable-kvm: Boot problem
> i386-softmmu with -enable-kvm: Boot problem
>
> Host is: Fedora 11
> 2.6.30.10-105.2.23.fc11.x86 on AMD Phenom II Quad Core

Hello,

OK, works well with version from
git://git.kernel.org/pub/scm/virt/kvm/qemu-kvm.git

The only thing which is broken so far in DOS: PS/2 mouse does not work 
correctly with VNC (buttons are pressed even if no button is pressed, position is 
random). This also happens with the unstable version of QEMU, so this is 
looks like not to be KVM related:
http://git.savannah.gnu.org/cgit/qemu.git/
(Works well with 0.12.3 stable)

Any ideas from which change this might come (PS/2 mouse code seems to be 
stable only keyboard LEDs added by Gerd)?

Ciao,
Gerhard

--
http://www.wiesinger.com/

Re: [Qemu-devel] Problem with QEMU on KVM

[Jamie Lokier]

Gerhard Wiesinger wrote:
> Any ideas from which change this might come (PS/2 mouse code seems to be 
> stable only keyboard LEDs added by Gerd)?

No, but you might find it with "git bisect" since you have a working
and non-working version.

-- Jamie

Re: [Qemu-devel] Problem with QEMU on KVM

[Gerhard Wiesinger]

On Wed, 21 Apr 2010, Jamie Lokier wrote:

> Gerhard Wiesinger wrote:
>> Any ideas from which change this might come (PS/2 mouse code seems to be
>> stable only keyboard LEDs added by Gerd)?
>
> No, but you might find it with "git bisect" since you have a working
> and non-working version.

Tried that, but no useful output (at least for me):
git bisect start
git bisect bad
git bisect good v0.12.3
Bisecting: a merge base must be tested
[749717a0ea2f60d33d01c1e37fa24dfa7250dfc0] Add S390 maintainer information

Had also a look at ps2.c/ps2.h, pckbd.c: There were no changes which could 
explain such a behaviour. Also keyboard is very strange which switches to 
CAPS LOCK (shift state) very unmotivated.

Can someone reproduce this behaviour, maybe with VNC and also with SDL?

Any further ideas?

Thnx.

Ciao,
Gerhard

--
http://www.wiesinger.com/

Re: [Qemu-devel] Problem with QEMU on KVM

[Gerd Hoffmann]

   Hi,

> Had also a look at ps2.c/ps2.h, pckbd.c: There were no changes which
> could explain such a behaviour. Also keyboard is very strange which
> switches to CAPS LOCK (shift state) very unmotivated.

Hmm.  Recent qemu tries hard to make sure vncviewer and guest 
capslock/numlock state are in sync.  It is a heuristic though and can 
fail in certain cases.  Try "-vnc $display,no-lock-key-sync", does that 
help?

cheers,
   Gerd

Re: [Qemu-devel] Problem with QEMU on KVM

[Gerhard Wiesinger]

On Thu, 22 Apr 2010, Gerd Hoffmann wrote:

>  Hi,
>
>> Had also a look at ps2.c/ps2.h, pckbd.c: There were no changes which
>> could explain such a behaviour. Also keyboard is very strange which
>> switches to CAPS LOCK (shift state) very unmotivated.
>
> Hmm.  Recent qemu tries hard to make sure vncviewer and guest 
> capslock/numlock state are in sync.  It is a heuristic though and can fail in 
> certain cases.  Try "-vnc $display,no-lock-key-sync", does that help?

Tracked down the problem of the mouse that te mouse is in absolute 
mouse mode (verified by logging, therefore coordinates are very high and absolute 
and not relative):
static void pointer_event(VncState *vs, int button_mask, int x, int y)
{
...
     if (vs->absolute) {
         kbd_mouse_event(ds_get_width(vs->ds) > 1 ?
                           x * 0x7FFF / (ds_get_width(vs->ds) - 1) : 0x4000,
                         ds_get_height(vs->ds) > 1 ?
                           y * 0x7FFF / (ds_get_height(vs->ds) - 1) : 0x4000,
                         dz, buttons);
     } else if (vnc_has_feature(vs, VNC_FEATURE_POINTER_TYPE_CHANGE)) {
         x -= 0x7FFF;
         y -= 0x7FFF;
/////////////////////////////////////////////////////////////
// We get here in (verified by core dump), but we should get into last
// else
/////////////////////////////////////////////////////////////
         kbd_mouse_event(x, y, dz, buttons);
     } else {
         if (vs->last_x != -1)
             kbd_mouse_event(x - vs->last_x,
                             y - vs->last_y,
                             dz, buttons);
         vs->last_x = x;
         vs->last_y = y;
     }

...
}

Guess problems comes from the following commit (not yet verified):
commit 37c34d9d5d87ea9d51760310c8863b82cb8c055a
Author: Anthony Liguori <aliguori@us.ibm.com>
Date:   Wed Mar 10 09:38:29 2010 -0600

     input: make vnc use mouse mode notifiers

     When we switch to absolute mode, we send out a notification (if the client
     supports it).  Today, we only send this notification when the client sends us
     a mouse event and we're in the wrong mode.

     Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

Ciao,
Gerhard

--
http://www.wiesinger.com/

Re: [Qemu-devel] Problem with QEMU on KVM

[Gerhard Wiesinger]

On Fri, 23 Apr 2010, Gerhard Wiesinger wrote:

> On Thu, 22 Apr 2010, Gerd Hoffmann wrote:
>
>>  Hi,
>> 
>>> Had also a look at ps2.c/ps2.h, pckbd.c: There were no changes which
>>> could explain such a behaviour. Also keyboard is very strange which
>>> switches to CAPS LOCK (shift state) very unmotivated.
>> 
>> Hmm.  Recent qemu tries hard to make sure vncviewer and guest 
>> capslock/numlock state are in sync.  It is a heuristic though and can fail 
>> in certain cases.  Try "-vnc $display,no-lock-key-sync", does that help?
>
> Tracked down the problem of the mouse that te mouse is in absolute mouse mode 
> (verified by logging, therefore coordinates are very high and absolute and 
> not relative):
> static void pointer_event(VncState *vs, int button_mask, int x, int y)
> {
> ...
>    if (vs->absolute) {
>        kbd_mouse_event(ds_get_width(vs->ds) > 1 ?
>                          x * 0x7FFF / (ds_get_width(vs->ds) - 1) : 0x4000,
>                        ds_get_height(vs->ds) > 1 ?
>                          y * 0x7FFF / (ds_get_height(vs->ds) - 1) : 0x4000,
>                        dz, buttons);
>    } else if (vnc_has_feature(vs, VNC_FEATURE_POINTER_TYPE_CHANGE)) {
>        x -= 0x7FFF;
>        y -= 0x7FFF;
> /////////////////////////////////////////////////////////////
> // We get here in (verified by core dump), but we should get into last
> // else
> /////////////////////////////////////////////////////////////
>        kbd_mouse_event(x, y, dz, buttons);
>    } else {
>        if (vs->last_x != -1)
>            kbd_mouse_event(x - vs->last_x,
>                            y - vs->last_y,
>                            dz, buttons);
>        vs->last_x = x;
>        vs->last_y = y;
>    }
>
> ...
> }
>
> Guess problems comes from the following commit (not yet verified):
> commit 37c34d9d5d87ea9d51760310c8863b82cb8c055a
> Author: Anthony Liguori <aliguori@us.ibm.com>
> Date:   Wed Mar 10 09:38:29 2010 -0600
>
>    input: make vnc use mouse mode notifiers
>
>    When we switch to absolute mode, we send out a notification (if the 
> client
>    supports it).  Today, we only send this notification when the client 
> sends us
>    a mouse event and we're in the wrong mode.
>
>    Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>


Ok, verified:
git revert -n 37c34d9d5d87ea9d51760310c8863b82cb8c055a
=> works well.

Ciao,
Gerhard

--
http://www.wiesinger.com/

Re: [Qemu-devel] Problem with QEMU on KVM

[Gerhard Wiesinger]

On Sat, 24 Apr 2010, Gerhard Wiesinger wrote:
>> Guess problems comes from the following commit (not yet verified):
>> commit 37c34d9d5d87ea9d51760310c8863b82cb8c055a
>> Author: Anthony Liguori <aliguori@us.ibm.com>
>> Date:   Wed Mar 10 09:38:29 2010 -0600
>>
>>    input: make vnc use mouse mode notifiers
>>
>>    When we switch to absolute mode, we send out a notification (if the 
>> client
>>    supports it).  Today, we only send this notification when the client 
>> sends us
>>    a mouse event and we're in the wrong mode.
>>
>>    Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
>
>
> Ok, verified:
> git revert -n 37c34d9d5d87ea9d51760310c8863b82cb8c055a
> => works well.

Still got no feedback and saw no further changes to fix this problem.

Thnx.

Ciao,
Gerhard

--
http://www.wiesinger.com/

Re: [Qemu-devel] Problem with QEMU on KVM

[Anthony Liguori]

On 05/04/2010 12:09 AM, Gerhard Wiesinger wrote:
> On Sat, 24 Apr 2010, Gerhard Wiesinger wrote:
>>> Guess problems comes from the following commit (not yet verified):
>>> commit 37c34d9d5d87ea9d51760310c8863b82cb8c055a
>>> Author: Anthony Liguori <aliguori@us.ibm.com>
>>> Date:   Wed Mar 10 09:38:29 2010 -0600
>>>
>>>    input: make vnc use mouse mode notifiers
>>>
>>>    When we switch to absolute mode, we send out a notification (if 
>>> the client
>>>    supports it).  Today, we only send this notification when the 
>>> client sends us
>>>    a mouse event and we're in the wrong mode.
>>>
>>>    Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
>>
>>
>> Ok, verified:
>> git revert -n 37c34d9d5d87ea9d51760310c8863b82cb8c055a
>> => works well.
>
> Still got no feedback and saw no further changes to fix this problem.

Fix is on the list.

Regards,

Anthony Liguori

> Thnx.
>
> Ciao,
> Gerhard
>
> -- 
> http://www.wiesinger.com/
>

Problem with QEMU / KVM

[K D]

[-- Attachment #1: Type: text/plain, Size: 2867 bytes --]

Hi

I built 2.6.27.10 kernel with KVM configured as built-in.

CONFIG_HAVE_KVM=y
CONFIG_VIRTUALIZATION=y
CONFIG_KVM=y
CONFIG_KVM_INTEL=y

I built qemu-kvm 0.12.2 from sources pointing to above kernel. When I spawn a VM, it hangs at boot. Pl see below. I copied qemu-kvm binary and bios, vga etc binaries. I installed grub on a manually created img file. This image has nothing but grub, its stage1,1_5,stage2, my kernel and initrd files.

Am spawning VM as below:

qemu-system-x86_64 -hda vmhd.img -L . -curses -show-cursor

grub> root (hd0,0)
 Filesystem type is ext2fs, partition type 0x83

grub> kernel /boot/vmos root=/dev/ram0 ramdisk_size=32768 rw single
   [Linux-bzImage, setup=0x1800, size=0x1cb9b0]

grub> initrd /boot/initrdfs.gz
   [Linux-initrd @ 0x6751000, 0x189b587 bytes]

grub> boot



I copied all these binaries from my dev host to this linux box.
 
-rw-r--r-- 1 root root 131072 May  2 21:31 bios.bin
-rw-r--r-- 1 root root   1025 May  2 21:31 linuxboot.bin
-rw-r--r-- 1 root root   1024 May  2 21:31 multiboot.bin
-rw-r--r-- 1 root root 524288 May  2 21:31 ppc_rom.bin
-rw-r--r-- 1 root root  72192 May  2 21:31 pxe-e1000.bin
-rw-r--r-- 1 root root  56832 May  2 21:31 pxe-i82559er.bin
-rw-r--r-- 1 root root  56320 May  2 21:31 pxe-ne2k_pci.bin
-rw-r--r-- 1 root root  56832 May  2 21:31 pxe-pcnet.bin
-rw-r--r-- 1 root root  56320 May  2 21:31 pxe-rtl8139.bin
-rw-r--r-- 1 root root  56320 May  2 21:31 pxe-virtio.bin
-rwxr-xr-x 1 root root   8960 May  2 21:31 vapic.bin
-rw-r--r-- 1 root root  35840 May  2 21:31 vgabios-cirrus.bin
-rw-r--r-- 1 root root  39936 May  2 21:31 vgabios.bin



Appreciate your help.



________________________________
From: Anthony Liguori <aliguori@linux.vnet.ibm.com>
To: Gerhard Wiesinger <lists@wiesinger.com>
Cc: qemu-devel@nongnu.org; Gerd Hoffmann <kraxel@redhat.com>
Sent: Tue, May 4, 2010 6:32:50 AM
Subject: Re: [Qemu-devel] Problem with QEMU on KVM

On 05/04/2010 12:09 AM, Gerhard Wiesinger wrote:
> On Sat, 24 Apr 2010, Gerhard Wiesinger wrote:
>>> Guess problems comes from the following commit (not yet verified):
>>> commit 37c34d9d5d87ea9d51760310c8863b82cb8c055a
>>> Author: Anthony Liguori <aliguori@us.ibm.com>
>>> Date:   Wed Mar 10 09:38:29 2010 -0600
>>> 
>>>    input: make vnc use mouse mode notifiers
>>> 
>>>    When we switch to absolute mode, we send out a notification (if the client
>>>    supports it).  Today, we only send this notification when the client sends us
>>>    a mouse event and we're in the wrong mode.
>>> 
>>>    Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
>> 
>> 
>> Ok, verified:
>> git revert -n 37c34d9d5d87ea9d51760310c8863b82cb8c055a
>> => works well.
> 
> Still got no feedback and saw no further changes to fix this problem.

Fix is on the list.

Regards,

Anthony Liguori

> Thnx.
> 
> Ciao,
> Gerhard
> 
> -- http://www.wiesinger.com/
> 


      

[-- Attachment #2: Type: text/html, Size: 4455 bytes --]

[Qemu-devel] Problem with QEMU / KVM

[K D]

[-- Attachment #1: Type: text/plain, Size: 2867 bytes --]

Hi

I built 2.6.27.10 kernel with KVM configured as built-in.

CONFIG_HAVE_KVM=y
CONFIG_VIRTUALIZATION=y
CONFIG_KVM=y
CONFIG_KVM_INTEL=y

I built qemu-kvm 0.12.2 from sources pointing to above kernel. When I spawn a VM, it hangs at boot. Pl see below. I copied qemu-kvm binary and bios, vga etc binaries. I installed grub on a manually created img file. This image has nothing but grub, its stage1,1_5,stage2, my kernel and initrd files.

Am spawning VM as below:

qemu-system-x86_64 -hda vmhd.img -L . -curses -show-cursor

grub> root (hd0,0)
 Filesystem type is ext2fs, partition type 0x83

grub> kernel /boot/vmos root=/dev/ram0 ramdisk_size=32768 rw single
   [Linux-bzImage, setup=0x1800, size=0x1cb9b0]

grub> initrd /boot/initrdfs.gz
   [Linux-initrd @ 0x6751000, 0x189b587 bytes]

grub> boot



I copied all these binaries from my dev host to this linux box.
 
-rw-r--r-- 1 root root 131072 May  2 21:31 bios.bin
-rw-r--r-- 1 root root   1025 May  2 21:31 linuxboot.bin
-rw-r--r-- 1 root root   1024 May  2 21:31 multiboot.bin
-rw-r--r-- 1 root root 524288 May  2 21:31 ppc_rom.bin
-rw-r--r-- 1 root root  72192 May  2 21:31 pxe-e1000.bin
-rw-r--r-- 1 root root  56832 May  2 21:31 pxe-i82559er.bin
-rw-r--r-- 1 root root  56320 May  2 21:31 pxe-ne2k_pci.bin
-rw-r--r-- 1 root root  56832 May  2 21:31 pxe-pcnet.bin
-rw-r--r-- 1 root root  56320 May  2 21:31 pxe-rtl8139.bin
-rw-r--r-- 1 root root  56320 May  2 21:31 pxe-virtio.bin
-rwxr-xr-x 1 root root   8960 May  2 21:31 vapic.bin
-rw-r--r-- 1 root root  35840 May  2 21:31 vgabios-cirrus.bin
-rw-r--r-- 1 root root  39936 May  2 21:31 vgabios.bin



Appreciate your help.



________________________________
From: Anthony Liguori <aliguori@linux.vnet.ibm.com>
To: Gerhard Wiesinger <lists@wiesinger.com>
Cc: qemu-devel@nongnu.org; Gerd Hoffmann <kraxel@redhat.com>
Sent: Tue, May 4, 2010 6:32:50 AM
Subject: Re: [Qemu-devel] Problem with QEMU on KVM

On 05/04/2010 12:09 AM, Gerhard Wiesinger wrote:
> On Sat, 24 Apr 2010, Gerhard Wiesinger wrote:
>>> Guess problems comes from the following commit (not yet verified):
>>> commit 37c34d9d5d87ea9d51760310c8863b82cb8c055a
>>> Author: Anthony Liguori <aliguori@us.ibm.com>
>>> Date:   Wed Mar 10 09:38:29 2010 -0600
>>> 
>>>    input: make vnc use mouse mode notifiers
>>> 
>>>    When we switch to absolute mode, we send out a notification (if the client
>>>    supports it).  Today, we only send this notification when the client sends us
>>>    a mouse event and we're in the wrong mode.
>>> 
>>>    Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
>> 
>> 
>> Ok, verified:
>> git revert -n 37c34d9d5d87ea9d51760310c8863b82cb8c055a
>> => works well.
> 
> Still got no feedback and saw no further changes to fix this problem.

Fix is on the list.

Regards,

Anthony Liguori

> Thnx.
> 
> Ciao,
> Gerhard
> 
> -- http://www.wiesinger.com/
> 


      

[-- Attachment #2: Type: text/html, Size: 4447 bytes --]