system_powerdown not working for qemu-kvm 0.12.4?

system_powerdown not working for qemu-kvm 0.12.4?

[Teck Choon Giam]

Hi,

Anyone encountered the same issue as me about system_powerdown no
longer working since upgraded to qemu-kvm 0.12.4?

Thanks.

Kindest regards,
Giam Teck Choon

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Michael Tokarev]

15.05.2010 05:19, Teck Choon Giam пишет:
> Hi,
>
> Anyone encountered the same issue as me about system_powerdown no
> longer working since upgraded to qemu-kvm 0.12.4?

Works For Me (tm)

Tried with winXP and win7 and linux guests.

/mjt

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

On 05/15/2010 04:19 AM, Teck Choon Giam wrote:
> Hi,
>
> Anyone encountered the same issue as me about system_powerdown no
> longer working since upgraded to qemu-kvm 0.12.4?
>    

Compared with what version?

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

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Teck Choon Giam]

On Sun, May 16, 2010 at 7:52 PM, Avi Kivity <avi@redhat.com> wrote:
> On 05/15/2010 04:19 AM, Teck Choon Giam wrote:
>>
>> Hi,
>>
>> Anyone encountered the same issue as me about system_powerdown no
>> longer working since upgraded to qemu-kvm 0.12.4?
>>
>
> Compared with what version?
>
> --
> error compiling committee.c: too many arguments to function
>
>

Hi Avi,

Sorry for this late reply.

Here are my testing:

freebsd 8: qemu-kvm-0.12.3 system_powerdown not working and
qemu-kvm-0.12.4 system_powerdown not working
centos 5: qemu-kvm-0.12.3 system_powerdown working and qemu-kvm-0.12.4
system_powerdown not working
windows 2008 R2: qemu-kvm-0.12.3 system_powerdown working and
qemu-kvm-0.12.4 system_powerdown not working

If you want me to test on other qemu-kvm version, please state which version.

Thanks.

Kindest regards,
Giam Teck Choon

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

On 05/19/2010 12:23 PM, Teck Choon Giam wrote:
> On Sun, May 16, 2010 at 7:52 PM, Avi Kivity<avi@redhat.com>  wrote:
>    
>> On 05/15/2010 04:19 AM, Teck Choon Giam wrote:
>>      
>>> Hi,
>>>
>>> Anyone encountered the same issue as me about system_powerdown no
>>> longer working since upgraded to qemu-kvm 0.12.4?
>>>
>>>        
>> Compared with what version?
>>
>> --
>> error compiling committee.c: too many arguments to function
>>
>>
>>      
> Hi Avi,
>
> Sorry for this late reply.
>
> Here are my testing:
>
> freebsd 8: qemu-kvm-0.12.3 system_powerdown not working and
> qemu-kvm-0.12.4 system_powerdown not working
> centos 5: qemu-kvm-0.12.3 system_powerdown working and qemu-kvm-0.12.4
> system_powerdown not working
> windows 2008 R2: qemu-kvm-0.12.3 system_powerdown working and
> qemu-kvm-0.12.4 system_powerdown not working
>
> If you want me to test on other qemu-kvm version, please state which version.
>    

Can you try to bisect between qemu-kvm-0.12.3 and 0.12.4 to see which 
commit introduced the regression?

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Teck Choon Giam]

Hi Avi,

Sorry for such super late reply as I have been caught up with personal
matters thus unable to do testing etc.

The following are my report about system_powerdown on various versions:

system_powerdown Status Report:
|=========================================================|
| Version            |       Guest OS       |   Status    |
|---------------------------------------------------------|
| qemu-kvm-0.12.1.2  | FreeBSD 8.1 (i386)   | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.1.2  | CentOS 5.5 (x86_64)  | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.1.2  | Windows 2008 R2      | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.2    | FreeBSD 8.1 (i386)   | Not Working |
|---------------------------------------------------------|
| qemu-kvm-0.12.2    | CentOS 5.5 (x86_64)  | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.2    | Windows 2008 R2      | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.3    | FreeBSD 8.1 (i386)   | Not Working |
|---------------------------------------------------------|
| qemu-kvm-0.12.3    | CentOS 5.5 (x86_64)  | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.3    | Windows 2008 R2      | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.4    | FreeBSD 8.1 (i386)   | Not Working |
|---------------------------------------------------------|
| qemu-kvm-0.12.4    | CentOS 5.5 (x86_64)  | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.4    | Windows 2008 R2      | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.5    | FreeBSD 8.1 (i386)   | Not Working |
|---------------------------------------------------------|
| qemu-kvm-0.12.5    | CentOS 5.5 (x86_64)  | Working     |
|---------------------------------------------------------|
| qemu-kvm-0.12.5    | Windows 2008 R2      | Working     |
|=========================================================|

>
> Can you try to bisect between qemu-kvm-0.12.3 and 0.12.4 to see which commit
> introduced the regression?
>

Actually I am not so sure about how to do the bisecting as the below
steps always produce different configure for me.  Any pointers?

# cd /usr/src
# git clone git://git.kernel.org/pub/scm/virt/kvm/qemu-kvm.git
# cd qemu-kvm
# ./configure --help|grep cpu-emulation
  --disable-cpu-emulation  disables use of qemu cpu emulation code
# git bisect reset master
We are not bisecting.
# git bisect good qemu-kvm-0.12.1.2
You need to start by "git bisect start"
Do you want me to do it for you [Y/n]? y
# git bisect bad qemu-kvm-0.12.2
Bisecting: 14 revisions left to test after this (roughly 4 steps)
[66dbb62824845e91808171a675998706ce359c71] Handle TFTP ERROR from client
# ./configure --help|grep cpu-emulation
show nothing when bisecting... ... configure script is different :(

Thanks.

Kindest regards,
Giam Teck Choon

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

  On 08/15/2010 02:32 AM, Teck Choon Giam wrote:
>> Can you try to bisect between qemu-kvm-0.12.3 and 0.12.4 to see which commit
>> introduced the regression?
>>
> Actually I am not so sure about how to do the bisecting as the below
> steps always produce different configure for me.  Any pointers?
>
> # cd /usr/src
> # git clone git://git.kernel.org/pub/scm/virt/kvm/qemu-kvm.git
> # cd qemu-kvm
> # ./configure --help|grep cpu-emulation
>    --disable-cpu-emulation  disables use of qemu cpu emulation code
> # git bisect reset master
> We are not bisecting.
> # git bisect good qemu-kvm-0.12.1.2
> You need to start by "git bisect start"
> Do you want me to do it for you [Y/n]? y
> # git bisect bad qemu-kvm-0.12.2
> Bisecting: 14 revisions left to test after this (roughly 4 steps)
> [66dbb62824845e91808171a675998706ce359c71] Handle TFTP ERROR from client
> # ./configure --help|grep cpu-emulation
> show nothing when bisecting... ... configure script is different :(
>

That's fine - you'll be running upstream qemu instead of qemu-kvm.  Just 
remember to add -enable-kvm to the command line.  Use ./configure 
--target-list=x86_64-softtmmu to cut down on compile time.

I'm betting 73b48d914f9 is the cause, but let's see the full bisect.

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

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Teck Choon Giam]

>
> That's fine - you'll be running upstream qemu instead of qemu-kvm.  Just
> remember to add -enable-kvm to the command line.  Use ./configure
> --target-list=x86_64-softtmmu to cut down on compile time.

Yes, I am doing so about the --target-list but missed out the
-enable-kvm command option prior to start each Guest OS.  Here is my
script:

-----------------------------------8<-----------------------------------
#!/bin/sh

kerver=`uname -r`
KERNELDIR="--kerneldir=/usr/src/linux-`uname -r`"
make clean
./configure --prefix=/usr/local/kvm ${KERNELDIR} \
--target-list=x86_64-softmmu

make
make INSTALL_MOD_STRIP=1 install
mkinitrd -v -f --builtin=ehci-hcd --builtin=uhci-hcd
--builtin=ohci-hcd --builtin=usb-storage /boot/initrd-${kerver}.img
${kerver}
depmod -v `uname -r`
-----------------------------------8<-----------------------------------

Instead of rebooting... I do the following then start the FreeBSD
Guest OS to test:

1. Stop all Guest OSes.
2. rmmod -v kvm-intel
3. rmmod -v kvm
4. modprobe -v kvm
5. modprobe -v kvm-intel

If I add -enable-kvm to the command line to start the guest OS I got:

No SMP KVM support, use '-smp 1'
failed to initialize KVM

Well... ... changed to use '-smp 1' as shown and the FreeBSD 8.1 Guest
OS started without issues.  However for CentOS 5.5 Guest OS, since I
am using virtio... ... running in qemu instead of qemu-kvm will be an
issue?  I have problem to start the Guest OS and it always goes to the
CentOS installer since I have the following:

---SNAP---
-boot cd \
-cdrom /home/kvm/images/CentOS-5.5-x86_64-bin-1of8.iso \
-drive file=/dev/VM/centos5,if=virtio,boot=on \
---SNAP---

So I will not test on other Guest OS but just focus on FreeBSD instead.

>
> I'm betting 73b48d914f9 is the cause, but let's see the full bisect.
>

I don't know whether I have done it right or not about bisecting but
each good ones I just use git bisect good until the one that is
failing then I use git bisect bad.  The following are for git bisect
good qemu-kvm-0.12.1.2 and git bisect bad qemu-kvm-0.12.2:

# git bisect log
git bisect start
# good: [a6ec654a7a863afff41b491a02ffd696c862cb41] Merge branch
'stable-0.12-upstream' into stable-0.12
git bisect good a6ec654a7a863afff41b491a02ffd696c862cb41
# bad: [c01cfac552861ca4d82e359791a2d79da7f80cb5] device assignment:
default requires IOMMU
git bisect bad c01cfac552861ca4d82e359791a2d79da7f80cb5
# good: [66dbb62824845e91808171a675998706ce359c71] Handle TFTP ERROR from client
git bisect good 66dbb62824845e91808171a675998706ce359c71
# good: [04babf6c6f8ccf69f1219db5fea233d679702e90] roms: rework rom
loading via fw
git bisect good 04babf6c6f8ccf69f1219db5fea233d679702e90
# good: [3999bf32440c1ea2ceb85eef008cc56a069af13f] Qemu's internal
TFTP server breaks lock-step-iness of TFTP
git bisect good 3999bf32440c1ea2ceb85eef008cc56a069af13f
# good: [e389e937a7b94186449e0590bdc8f04ecbb1ab0b] Update version and
changelog for release
git bisect good e389e937a7b94186449e0590bdc8f04ecbb1ab0b
# bad: [b874ce1db7d8654850c8a6606b95ffb1c7d22ce2] Merge remote branch
'upstream/stable-0.12' into stable-0.12
git bisect bad b874ce1db7d8654850c8a6606b95ffb1c7d22ce2


The following are git bisect log between git bisect good
qemu-kvm-0.12.1.2 and git bisect bad qemu-kvm-0.12.3 instead:

git bisect start
# good: [a6ec654a7a863afff41b491a02ffd696c862cb41] Merge branch
'stable-0.12-upstream' into stable-0.12
git bisect good a6ec654a7a863afff41b491a02ffd696c862cb41
# bad: [69a5ecafa27daeb943dc2ee65b1470844f23f934] Merge branch
'stable-0.12-merge' into stable-0.12
git bisect bad 69a5ecafa27daeb943dc2ee65b1470844f23f934
# good: [d0d888bc6d1a106609b9af42ecb552c6c34a85c5] qcow2: Return
0/-errno in qcow2_alloc_cluster_offset
git bisect good d0d888bc6d1a106609b9af42ecb552c6c34a85c5
# good: [7ebc79037c5f426bfb08cc506670bf7dd3912430] virtio-net: fix
network stall under load
git bisect good 7ebc79037c5f426bfb08cc506670bf7dd3912430
# good: [6173d56bdcb53389c54e803873e6bf8f87836a4f] Merge remote branch
'qemu-kvm/uq/stable-0.12' into stable-0.12
git bisect good 6173d56bdcb53389c54e803873e6bf8f87836a4f
# bad: [59691c0cb129c9aa955be22573b43b26534f9db4] KVM: Request setting
of nmi_pending and sipi_vector
git bisect bad 59691c0cb129c9aa955be22573b43b26534f9db4
# bad: [dec2eb9d724b21581500aea911dd13f7bfbea59e] Fix kvm_load_mpstate
for vcpu hot add
git bisect bad dec2eb9d724b21581500aea911dd13f7bfbea59e


Let me know if I have done anything wrong about the bisecting... ...

Thanks.

Kindest regards,
Giam Teck Choon

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

  On 08/15/2010 07:15 PM, Teck Choon Giam wrote:
>
> Let me know if I have done anything wrong about the bisecting... ...
>

All looks fine, but what are the results?  git should say something like 
'xxxxxxxxx is first bad commit' which is the interesting part.

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

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Teck Choon Giam]

On Mon, Aug 16, 2010 at 12:19 AM, Avi Kivity <avi@redhat.com> wrote:
>  On 08/15/2010 07:15 PM, Teck Choon Giam wrote:
>>
>> Let me know if I have done anything wrong about the bisecting... ...
>>
>
> All looks fine, but what are the results?  git should say something like
> 'xxxxxxxxx is first bad commit' which is the interesting part.


# git bisect reset master
We are not bisecting.
# git bisect start
# git bisect good
# git bisect good qemu-kvm-0.12.1.2
# git bisect bad qemu-kvm-0.12.2
Bisecting: 14 revisions left to test after this (roughly 4 steps)
[66dbb62824845e91808171a675998706ce359c71] Handle TFTP ERROR from client
# git bisect good
Bisecting: 7 revisions left to test after this (roughly 3 steps)
[04babf6c6f8ccf69f1219db5fea233d679702e90] roms: rework rom loading via fw
# git bisect good
Bisecting: 3 revisions left to test after this (roughly 2 steps)
[3999bf32440c1ea2ceb85eef008cc56a069af13f] Qemu's internal TFTP server
breaks lock-step-iness of TFTP
# git bisect good
Bisecting: 1 revision left to test after this (roughly 1 step)
[e389e937a7b94186449e0590bdc8f04ecbb1ab0b] Update version and
changelog for release
# git bisect good
Bisecting: 0 revisions left to test after this (roughly 0 steps)
[b874ce1db7d8654850c8a6606b95ffb1c7d22ce2] Merge remote branch
'upstream/stable-0.12' into stable-0.12
# git bisect bad
b874ce1db7d8654850c8a6606b95ffb1c7d22ce2 is the first bad commit


Thanks.

Kindest regards,
Giam Teck Choon

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Teck Choon Giam]

On Mon, Aug 16, 2010 at 1:53 AM, Teck Choon Giam
<giamteckchoon@gmail.com> wrote:
> On Mon, Aug 16, 2010 at 12:19 AM, Avi Kivity <avi@redhat.com> wrote:
>>  On 08/15/2010 07:15 PM, Teck Choon Giam wrote:
>>>
>>> Let me know if I have done anything wrong about the bisecting... ...
>>>
>>
>> All looks fine, but what are the results?  git should say something like
>> 'xxxxxxxxx is first bad commit' which is the interesting part.
>
>
> # git bisect reset master
> We are not bisecting.
> # git bisect start
> # git bisect good
> # git bisect good qemu-kvm-0.12.1.2
> # git bisect bad qemu-kvm-0.12.2
> Bisecting: 14 revisions left to test after this (roughly 4 steps)
> [66dbb62824845e91808171a675998706ce359c71] Handle TFTP ERROR from client
> # git bisect good
> Bisecting: 7 revisions left to test after this (roughly 3 steps)
> [04babf6c6f8ccf69f1219db5fea233d679702e90] roms: rework rom loading via fw
> # git bisect good
> Bisecting: 3 revisions left to test after this (roughly 2 steps)
> [3999bf32440c1ea2ceb85eef008cc56a069af13f] Qemu's internal TFTP server
> breaks lock-step-iness of TFTP
> # git bisect good
> Bisecting: 1 revision left to test after this (roughly 1 step)
> [e389e937a7b94186449e0590bdc8f04ecbb1ab0b] Update version and
> changelog for release
> # git bisect good
> Bisecting: 0 revisions left to test after this (roughly 0 steps)
> [b874ce1db7d8654850c8a6606b95ffb1c7d22ce2] Merge remote branch
> 'upstream/stable-0.12' into stable-0.12
> # git bisect bad
> b874ce1db7d8654850c8a6606b95ffb1c7d22ce2 is the first bad commit
>

Hi Avi,

Is there anything else you need from me about this system_powerdown
testing?  I really hope qemu/qemu-kvm will support system_powerdown
for FreeBSD guest... ...

Thanks.

Kindest regards,
Giam Teck Choon

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Ruben Kerkhof]

Hi Avi,

On Sun, Aug 15, 2010 at 13:00, Avi Kivity <avi@redhat.com> wrote:

> I'm betting 73b48d914f9 is the cause, but let's see the full bisect.

system_powerdown with FreeBSD 8.1 works on 73b48d914f9.

I've bisected the issue to commit 76e617c.

Let me know if I should do some other tests.

Regards, Ruben

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

  On 10/11/2010 06:29 PM, Ruben Kerkhof wrote:
> Hi Avi,
>
> On Sun, Aug 15, 2010 at 13:00, Avi Kivity<avi@redhat.com>  wrote:
>
> >  I'm betting 73b48d914f9 is the cause, but let's see the full bisect.
>
> system_powerdown with FreeBSD 8.1 works on 73b48d914f9.
>
> I've bisected the issue to commit 76e617c.
>
> Let me know if I should do some other tests.

It works on 73b48d914f9 but fails on 76e617c due to the seabios change.

Can you try 76e617c but with SeaBIOS 0.5.0?  If that works, keep 
qemu-kvm on 76e617c but bisect SeaBIOS from 0.5.0 to 0.5.1 to see which 
commit broke freebsd.

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

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Ruben Kerkhof]

On Mon, Oct 11, 2010 at 18:57, Avi Kivity <avi@redhat.com> wrote:
>  On 10/11/2010 06:29 PM, Ruben Kerkhof wrote:
>>
>> Hi Avi,
>>
>> On Sun, Aug 15, 2010 at 13:00, Avi Kivity<avi@redhat.com>  wrote:
>>
>> >  I'm betting 73b48d914f9 is the cause, but let's see the full bisect.
>>
>> system_powerdown with FreeBSD 8.1 works on 73b48d914f9.
>>
>> I've bisected the issue to commit 76e617c.
>>
>> Let me know if I should do some other tests.
>
> It works on 73b48d914f9 but fails on 76e617c due to the seabios change.
>
> Can you try 76e617c but with SeaBIOS 0.5.0?  If that works, keep qemu-kvm on
> 76e617c but bisect SeaBIOS from 0.5.0 to 0.5.1 to see which commit broke
> freebsd.
>

Sure,

5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6 is the first bad commit
commit 5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6
Author: Kevin O'Connor <kevin@koconnor.net>
Date:   Wed Dec 30 12:36:22 2009 -0500

    Commit compiled dsdt file; misc comment updates.

    Commit new dsdt with recent changes.
    Add some misc comments.
    Also, fix uninitialized warning in mptable code.

Regards, Ruben

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

  On 10/11/2010 07:53 PM, Ruben Kerkhof wrote:
> On Mon, Oct 11, 2010 at 18:57, Avi Kivity<avi@redhat.com>  wrote:
> >    On 10/11/2010 06:29 PM, Ruben Kerkhof wrote:
> >>
> >>  Hi Avi,
> >>
> >>  On Sun, Aug 15, 2010 at 13:00, Avi Kivity<avi@redhat.com>    wrote:
> >>
> >>  >    I'm betting 73b48d914f9 is the cause, but let's see the full bisect.
> >>
> >>  system_powerdown with FreeBSD 8.1 works on 73b48d914f9.
> >>
> >>  I've bisected the issue to commit 76e617c.
> >>
> >>  Let me know if I should do some other tests.
> >
> >  It works on 73b48d914f9 but fails on 76e617c due to the seabios change.
> >
> >  Can you try 76e617c but with SeaBIOS 0.5.0?  If that works, keep qemu-kvm on
> >  76e617c but bisect SeaBIOS from 0.5.0 to 0.5.1 to see which commit broke
> >  freebsd.
> >
>
> Sure,
>
> 5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6 is the first bad commit
> commit 5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6
> Author: Kevin O'Connor<kevin@koconnor.net>
> Date:   Wed Dec 30 12:36:22 2009 -0500
>
>      Commit compiled dsdt file; misc comment updates.
>
>      Commit new dsdt with recent changes.
>      Add some misc comments.
>      Also, fix uninitialized warning in mptable code.
>
> Regards, Ruben

Gleb, Kevin, any ideas?

(summary: qemu-kvm doesn't acpi shutdown freebsd 8.1 with this commit; 
qemu.git does.  May be due to interrupt polarity which kvm implements 
but qemu does not)

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

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Kevin O'Connor]

On Tue, Oct 12, 2010 at 08:49:58AM +0200, Avi Kivity wrote:
>  On 10/11/2010 07:53 PM, Ruben Kerkhof wrote:
> >5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6 is the first bad commit
> >commit 5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6
> >Author: Kevin O'Connor<kevin@koconnor.net>
> Gleb, Kevin, any ideas?
> 
> (summary: qemu-kvm doesn't acpi shutdown freebsd 8.1 with this
> commit; qemu.git does.  May be due to interrupt polarity which kvm
> implements but qemu does not)

The only thing in commit 5c99b6c9 that could cause an issue is that it
has the compiled acpi changes actually made in commit 29f4b912, but I
don't see how that would be a problem to reboots:

diff --git a/src/acpi-dsdt.dsl b/src/acpi-dsdt.dsl
index cee038a..2bede25 100644
--- a/src/acpi-dsdt.dsl
+++ b/src/acpi-dsdt.dsl
@@ -58,7 +58,10 @@ DefinitionBlock (
 #define prt_slot2(nr) prt_slot(nr, LNKB, LNKC, LNKD, LNKA)
 #define prt_slot3(nr) prt_slot(nr, LNKC, LNKD, LNKA, LNKB)
                prt_slot0(0x0000),
-               prt_slot1(0x0001),
+               Package() { 0x0001ffff, 0, 0, 9 },
+               Package() { 0x0001ffff, 1, LNKB, 0 },
+               Package() { 0x0001ffff, 2, LNKC, 0 },
+               Package() { 0x0001ffff, 3, LNKD, 0 },
                prt_slot2(0x0002),
                prt_slot3(0x0003),
                prt_slot0(0x0004),

Can you confirm that commit 4c94b7ea works reliably while commit
5c99b6c9 does not?

-Kevin

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Gleb Natapov]

On Tue, Oct 12, 2010 at 03:11:24AM -0400, Kevin O'Connor wrote:
> On Tue, Oct 12, 2010 at 08:49:58AM +0200, Avi Kivity wrote:
> >  On 10/11/2010 07:53 PM, Ruben Kerkhof wrote:
> > >5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6 is the first bad commit
> > >commit 5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6
> > >Author: Kevin O'Connor<kevin@koconnor.net>
> > Gleb, Kevin, any ideas?
> > 
> > (summary: qemu-kvm doesn't acpi shutdown freebsd 8.1 with this
> > commit; qemu.git does.  May be due to interrupt polarity which kvm
> > implements but qemu does not)
> 
> The only thing in commit 5c99b6c9 that could cause an issue is that it
> has the compiled acpi changes actually made in commit 29f4b912, but I
> don't see how that would be a problem to reboots:
> 
It makes line 0 of device 1 to be active low instead of active high.

> diff --git a/src/acpi-dsdt.dsl b/src/acpi-dsdt.dsl
> index cee038a..2bede25 100644
> --- a/src/acpi-dsdt.dsl
> +++ b/src/acpi-dsdt.dsl
> @@ -58,7 +58,10 @@ DefinitionBlock (
>  #define prt_slot2(nr) prt_slot(nr, LNKB, LNKC, LNKD, LNKA)
>  #define prt_slot3(nr) prt_slot(nr, LNKC, LNKD, LNKA, LNKB)
>                 prt_slot0(0x0000),
> -               prt_slot1(0x0001),
> +               Package() { 0x0001ffff, 0, 0, 9 },
> +               Package() { 0x0001ffff, 1, LNKB, 0 },
> +               Package() { 0x0001ffff, 2, LNKC, 0 },
> +               Package() { 0x0001ffff, 3, LNKD, 0 },
>                 prt_slot2(0x0002),
>                 prt_slot3(0x0003),
>                 prt_slot0(0x0004),
> 
> Can you confirm that commit 4c94b7ea works reliably while commit
> 5c99b6c9 does not?
> 
> -Kevin

--
			Gleb.

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

  On 10/12/2010 09:14 AM, Gleb Natapov wrote:
> On Tue, Oct 12, 2010 at 03:11:24AM -0400, Kevin O'Connor wrote:
> >  On Tue, Oct 12, 2010 at 08:49:58AM +0200, Avi Kivity wrote:
> >  >   On 10/11/2010 07:53 PM, Ruben Kerkhof wrote:
> >  >  >5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6 is the first bad commit
> >  >  >commit 5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6
> >  >  >Author: Kevin O'Connor<kevin@koconnor.net>
> >  >  Gleb, Kevin, any ideas?
> >  >
> >  >  (summary: qemu-kvm doesn't acpi shutdown freebsd 8.1 with this
> >  >  commit; qemu.git does.  May be due to interrupt polarity which kvm
> >  >  implements but qemu does not)
> >
> >  The only thing in commit 5c99b6c9 that could cause an issue is that it
> >  has the compiled acpi changes actually made in commit 29f4b912, but I
> >  don't see how that would be a problem to reboots:
> >
> It makes line 0 of device 1 to be active low instead of active high.

Correctly, according to the documentation (the line is marked IRQ9OUT#, 
which indicates it's active low, though that isn't set down elsewhere).

Qemu however emulates it as active high.  Other guests work, which is wierd.

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

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

  On 10/12/2010 09:25 AM, Avi Kivity wrote:
> On 10/12/2010 09:14 AM, Gleb Natapov wrote:
>> On Tue, Oct 12, 2010 at 03:11:24AM -0400, Kevin O'Connor wrote:
>> > On Tue, Oct 12, 2010 at 08:49:58AM +0200, Avi Kivity wrote:
>> > > On 10/11/2010 07:53 PM, Ruben Kerkhof wrote:
>> > > >5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6 is the first bad commit
>> > > >commit 5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6
>> > > >Author: Kevin O'Connor<kevin@koconnor.net>
>> > > Gleb, Kevin, any ideas?
>> > >
>> > > (summary: qemu-kvm doesn't acpi shutdown freebsd 8.1 with this
>> > > commit; qemu.git does. May be due to interrupt polarity which kvm
>> > > implements but qemu does not)
>> >
>> > The only thing in commit 5c99b6c9 that could cause an issue is that it
>> > has the compiled acpi changes actually made in commit 29f4b912, but I
>> > don't see how that would be a problem to reboots:
>> >
>> It makes line 0 of device 1 to be active low instead of active high.
>
> Correctly, according to the documentation (the line is marked 
> IRQ9OUT#, which indicates it's active low, though that isn't set down 
> elsewhere).
>
> Qemu however emulates it as active high. Other guests work, which is 
> wierd.
>

Aha! The piix4 specification updates says:

>
> • IRQ9OUT#/GPO29 should be labeled as IRQ9OUT/GPO29

So it looks like it should be active high, and seabios is wrong in 
listing it as active low. Perhaps other OSes have quirks to force it to 
active high.

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

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

  On 10/12/2010 09:33 AM, Avi Kivity wrote:
>
> Aha! The piix4 specification updates says:
>
>>
>> • IRQ9OUT#/GPO29 should be labeled as IRQ9OUT/GPO29
>

Even more explicitly:

> 14. IRQ9OUT# Is Active HI
> Problem: The signal identified as IRQ9OUT#/GPO29, pin F3, is not 
> active level LO, it is active level HI, when APIC Chip Select 
> (XBCS[8]) is set.
> Implication: This signal is typically used in Dual Processor capable 
> systems and is connected to an IOAPIC. If the IOAPIC input is 
> programmed for level LO, and SCIs or SM Bus events in the 
> PIIX4/PIIX4E/PIIX4M are programmed to be reported on IRQ9OUT, devices 
> using these will not be recognized by the IOAPIC and will not work 
> correctly.
> Workaround: Program the appropriate input of the IOAPIC to active 
> level HI.
> Status: This will not be fixed in the PIIX4/PIIX4E/PIIX4M. This is 
> planned to be incorporated into the PIIX4 datasheet as a specification 
> change.



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

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Gleb Natapov]

On Tue, Oct 12, 2010 at 09:33:42AM +0200, Avi Kivity wrote:
>  On 10/12/2010 09:25 AM, Avi Kivity wrote:
> >On 10/12/2010 09:14 AM, Gleb Natapov wrote:
> >>On Tue, Oct 12, 2010 at 03:11:24AM -0400, Kevin O'Connor wrote:
> >>> On Tue, Oct 12, 2010 at 08:49:58AM +0200, Avi Kivity wrote:
> >>> > On 10/11/2010 07:53 PM, Ruben Kerkhof wrote:
> >>> > >5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6 is the first bad commit
> >>> > >commit 5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6
> >>> > >Author: Kevin O'Connor<kevin@koconnor.net>
> >>> > Gleb, Kevin, any ideas?
> >>> >
> >>> > (summary: qemu-kvm doesn't acpi shutdown freebsd 8.1 with this
> >>> > commit; qemu.git does. May be due to interrupt polarity which kvm
> >>> > implements but qemu does not)
> >>>
> >>> The only thing in commit 5c99b6c9 that could cause an issue is that it
> >>> has the compiled acpi changes actually made in commit 29f4b912, but I
> >>> don't see how that would be a problem to reboots:
> >>>
> >>It makes line 0 of device 1 to be active low instead of active high.
> >
> >Correctly, according to the documentation (the line is marked
> >IRQ9OUT#, which indicates it's active low, though that isn't set
> >down elsewhere).
> >
> >Qemu however emulates it as active high. Other guests work, which
> >is wierd.
> >
> 
> Aha! The piix4 specification updates says:
> 
> >
> >• IRQ9OUT#/GPO29 should be labeled as IRQ9OUT/GPO29
> 
> So it looks like it should be active high, and seabios is wrong in
> listing it as active low. Perhaps other OSes have quirks to force it
> to active high.
> 
It is even stranger then that. Seabios creates interrupt override entry
for irq9 to be active high, level triggered, but it should be edge
triggered.

--
			Gleb.

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Avi Kivity]

  On 10/12/2010 09:40 AM, Gleb Natapov wrote:
> >  >
> >  >• IRQ9OUT#/GPO29 should be labeled as IRQ9OUT/GPO29
> >
> >  So it looks like it should be active high, and seabios is wrong in
> >  listing it as active low. Perhaps other OSes have quirks to force it
> >  to active high.
> >
> It is even stranger then that. Seabios creates interrupt override entry
> for irq9 to be active high, level triggered, but it should be edge
> triggered.

So two bugs, not one...

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

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Ruben Kerkhof]

Hi Kevin,

On Tue, Oct 12, 2010 at 09:11, Kevin O'Connor <kevin@koconnor.net> wrote:
> On Tue, Oct 12, 2010 at 08:49:58AM +0200, Avi Kivity wrote:
>>  On 10/11/2010 07:53 PM, Ruben Kerkhof wrote:
>> >5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6 is the first bad commit
>> >commit 5c99b6c984682ddb1d4543a7e27a1f4ca633e6a6
>> >Author: Kevin O'Connor<kevin@koconnor.net>
>> Gleb, Kevin, any ideas?
>>
>> (summary: qemu-kvm doesn't acpi shutdown freebsd 8.1 with this
>> commit; qemu.git does.  May be due to interrupt polarity which kvm
>> implements but qemu does not)
>
> The only thing in commit 5c99b6c9 that could cause an issue is that it
> has the compiled acpi changes actually made in commit 29f4b912, but I
> don't see how that would be a problem to reboots:
>
> diff --git a/src/acpi-dsdt.dsl b/src/acpi-dsdt.dsl
> index cee038a..2bede25 100644
> --- a/src/acpi-dsdt.dsl
> +++ b/src/acpi-dsdt.dsl
> @@ -58,7 +58,10 @@ DefinitionBlock (
>  #define prt_slot2(nr) prt_slot(nr, LNKB, LNKC, LNKD, LNKA)
>  #define prt_slot3(nr) prt_slot(nr, LNKC, LNKD, LNKA, LNKB)
>                prt_slot0(0x0000),
> -               prt_slot1(0x0001),
> +               Package() { 0x0001ffff, 0, 0, 9 },
> +               Package() { 0x0001ffff, 1, LNKB, 0 },
> +               Package() { 0x0001ffff, 2, LNKC, 0 },
> +               Package() { 0x0001ffff, 3, LNKD, 0 },
>                prt_slot2(0x0002),
>                prt_slot3(0x0003),
>                prt_slot0(0x0004),
>
> Can you confirm that commit 4c94b7ea works reliably while commit
> 5c99b6c9 does not?
>
> -Kevin
>

Yes, that's correct.

Is there anything I can do to help resolve this?
Would rolling back this commit have any impact on other operating systems?

Regards,

Ruben

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Gleb Natapov]

On Fri, Oct 15, 2010 at 03:45:06AM +0200, Ruben Kerkhof wrote:
> Is there anything I can do to help resolve this?
> Would rolling back this commit have any impact on other operating systems?
> 
Patch below should fix the problem.

diff --git a/src/acpi-dsdt.dsl b/src/acpi-dsdt.dsl
index bb0a176..dafcf45 100644
--- a/src/acpi-dsdt.dsl
+++ b/src/acpi-dsdt.dsl
@@ -73,7 +73,7 @@ DefinitionBlock (
 #define prt_slot3(nr) prt_slot(nr, LNKC, LNKD, LNKA, LNKB)
                prt_slot0(0x0000),
                /* Device 1 is power mgmt device, and can only use irq 9 */
-               Package() { 0x0001ffff, 0, 0, 9 },
+               Package() { 0x0001ffff, 0, LNKS, 0 },
                Package() { 0x0001ffff, 1, LNKB, 0 },
                Package() { 0x0001ffff, 2, LNKC, 0 },
                Package() { 0x0001ffff, 3, LNKD, 0 },
@@ -634,6 +634,46 @@ DefinitionBlock (
                     Store (TMP, PRQ3)
                 }
         }
+        Device(LNKS){
+                Name(_HID, EISAID("PNP0C0F"))     // PCI interrupt link
+                Name(_UID, 5)
+                Name(_PRS, ResourceTemplate(){
+                    Interrupt (, Level, ActiveHigh, Shared)
+                        { 9 }
+                })
+                Method (_STA, 0, NotSerialized)
+                {
+                    Store (0x0B, Local0)
+                    If (And (0x80, PRQ0, Local1))
+                    {
+                         Store (0x09, Local0)
+                    }
+                    Return (Local0)
+                }
+                Method (_DIS, 0, NotSerialized)
+                {
+                    Or (PRQ0, 0x80, PRQ0)
+                }
+                Method (_CRS, 0, NotSerialized)
+                {
+                    Name (PRR0, ResourceTemplate ()
+                    {
+                        Interrupt (, Level, ActiveHigh, Shared)
+                            {9}
+                    })
+                    CreateDWordField (PRR0, 0x05, TMP)
+                    Store (PRQ0, Local0)
+                    If (LLess (Local0, 0x80))
+                    {
+                        Store (Local0, TMP)
+                    }
+                    Else
+                    {
+                        Store (Zero, TMP)
+                    }
+                    Return (PRR0)
+                }
+        }
     }
 
     /*
--
			Gleb.

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Ruben Kerkhof]

On Fri, Oct 15, 2010 at 09:01, Gleb Natapov <gleb@redhat.com> wrote:
> On Fri, Oct 15, 2010 at 03:45:06AM +0200, Ruben Kerkhof wrote:
>> Is there anything I can do to help resolve this?
>> Would rolling back this commit have any impact on other operating systems?
>>
> Patch below should fix the problem.
>
> diff --git a/src/acpi-dsdt.dsl b/src/acpi-dsdt.dsl
> index bb0a176..dafcf45 100644
> --- a/src/acpi-dsdt.dsl
> +++ b/src/acpi-dsdt.dsl
> @@ -73,7 +73,7 @@ DefinitionBlock (
>  #define prt_slot3(nr) prt_slot(nr, LNKC, LNKD, LNKA, LNKB)
>                prt_slot0(0x0000),
>                /* Device 1 is power mgmt device, and can only use irq 9 */
> -               Package() { 0x0001ffff, 0, 0, 9 },
> +               Package() { 0x0001ffff, 0, LNKS, 0 },
>                Package() { 0x0001ffff, 1, LNKB, 0 },
>                Package() { 0x0001ffff, 2, LNKC, 0 },
>                Package() { 0x0001ffff, 3, LNKD, 0 },
> @@ -634,6 +634,46 @@ DefinitionBlock (
>                     Store (TMP, PRQ3)
>                 }
>         }
> +        Device(LNKS){
> +                Name(_HID, EISAID("PNP0C0F"))     // PCI interrupt link
> +                Name(_UID, 5)
> +                Name(_PRS, ResourceTemplate(){
> +                    Interrupt (, Level, ActiveHigh, Shared)
> +                        { 9 }
> +                })
> +                Method (_STA, 0, NotSerialized)
> +                {
> +                    Store (0x0B, Local0)
> +                    If (And (0x80, PRQ0, Local1))
> +                    {
> +                         Store (0x09, Local0)
> +                    }
> +                    Return (Local0)
> +                }
> +                Method (_DIS, 0, NotSerialized)
> +                {
> +                    Or (PRQ0, 0x80, PRQ0)
> +                }
> +                Method (_CRS, 0, NotSerialized)
> +                {
> +                    Name (PRR0, ResourceTemplate ()
> +                    {
> +                        Interrupt (, Level, ActiveHigh, Shared)
> +                            {9}
> +                    })
> +                    CreateDWordField (PRR0, 0x05, TMP)
> +                    Store (PRQ0, Local0)
> +                    If (LLess (Local0, 0x80))
> +                    {
> +                        Store (Local0, TMP)
> +                    }
> +                    Else
> +                    {
> +                        Store (Zero, TMP)
> +                    }
> +                    Return (PRR0)
> +                }
> +        }
>     }
>
>     /*
> --
>                        Gleb.
>

It does indeed.

Thanks a lot!

Ruben

Re: system_powerdown not working for qemu-kvm 0.12.4?

[Teck Choon Giam]

Thanks a lot for all :)

Kindest regards,
Giam Teck Choon