[ANNOUNCE] kvm-75 release

[ANNOUNCE] kvm-75 release

[Avi Kivity]

Ballooning is finally enabled in kvm-75.  You will need a 2.6.27+ host 
for ballooning to work properly.

Changes from kvm-74:
- add 1280x768 std-vga resolution
- ballooning (Anthony Liguori)
   - requires 2.6.27+ host to work correctly
- fix bad memory sizing on ia64 (Xiantao Zhang)
- restore option roms to their original content on reset
   - fixes extboot failures after reset
- enable ia64 power manangement (Xiantao Zhang)
- add cld/std/cli/sti/clc instructions to testsuite (Mohammed Gamal)
- fix compile errors on hosts that lack MADV_DONTFORK
- avoid accessing CMOS NVRAM from ACPI AML (Gleb Natapov)
   - fixes rare reboot failures with Windows
   - fixes ACPI error messages in Windows event log
- add mp state for ia64 (Xiantao Zhang)
- add missing prototypes (Jes Sorensen)
- fix compiler warnings (Jes Sorensen)
- add in/out instructions to emulator (Mohammed Gamal)
- use Windows pcr instead of guest's tr for tpr patching cpu id
   - fixes hibernate/resume with guest smp
- account for realmode/npt/ept page faults in statistics
- fix locking around dirty log tracking
- flush tlbs after fetching the dirty log
- fix random segfaults with npt (Joerg Roedel)
- fix setting accessed bit on non-speculative sptes
- fix tlb flush during page zap
- slightly optimize event injection with npt
- add MC5_MISC msr support (Joerg Roedel)
   - fixes Linux bootup warnings
- make s390 facility bits future proof (Christian Borntraeger)
- improve s390 guest Kconfig help (Christian Borntraeger)
- add cld/std instructions to emulator (Mohammed Gamal)
- fix address overflow in ept (Sheng Yang)
- add statistic for guest irq injections
- remove unnecessary remote tlb flush (Joerg Roedel)


Notes:
     If you use the modules bundled with kvm-75, you can use any version
of Linux from 2.6.16 upwards.  You may also use kvm-75 userspace with
the kvm modules provided by Linux 2.6.22 or above.  Some features may
only be available in newer releases.

http://kvm.qumranet.com


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

Re: [ANNOUNCE] kvm-75 release

[Jim Paris]

Avi Kivity wrote:
> - restore option roms to their original content on reset
>   - fixes extboot failures after reset

It doesn't seem to work here:

$ dmesg | grep kvm | tail -1
[1649282.904413] loaded kvm module (kvm-75)
$ sudo kvm | head -1
QEMU PC emulator version 0.9.1 (kvm-75), Copyright (c) 2003-2008 Fabrice Bellard
$ sudo /usr/bin/kvm -M pc -m 512 -smp 1 -name svn -boot c -drive file=/dev/mapper/raid1-svn,if=virtio,index=0,boot=on

At GRUB, I hit "c" for a command prompt, then type "reboot".
At the reboot, the BIOS says:

  QEMU BIOS - build: 09/03/08
  $Revision: 1.182 $ $Date: 2007/08/01 17:09:51 $
  Options: apmbios pcibios eltorito rombios32
  
  ata1 master: QEMU DVD-ROM ATAPI-4 CD-Rom/DVD-Rom
  
  Booting from Hard Disk...
  Boot from Hard Disk failed: could not read the boot disk
  FATAL: No bootable device.

With if=ide, rebooting in this manner works fine.

-jim

Re: [ANNOUNCE] kvm-75 release

[Avi Kivity]

Jim Paris wrote:
> Avi Kivity wrote:
>   
>> - restore option roms to their original content on reset
>>   - fixes extboot failures after reset
>>     
>
> It doesn't seem to work here:
>
> $ dmesg | grep kvm | tail -1
> [1649282.904413] loaded kvm module (kvm-75)
> $ sudo kvm | head -1
> QEMU PC emulator version 0.9.1 (kvm-75), Copyright (c) 2003-2008 Fabrice Bellard
> $ sudo /usr/bin/kvm -M pc -m 512 -smp 1 -name svn -boot c -drive file=/dev/mapper/raid1-svn,if=virtio,index=0,boot=on
>
> At GRUB, I hit "c" for a command prompt, then type "reboot".
>   

grub is probably booting via int 19 or jumping to the reset vector, so 
kvm doesn't see the reboot.  If that's the case, we can either implement 
the chipset's memory control registers properly and teach the bios to 
program them, or add a hack in the bios to trigger the copying without 
the full reset.

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

Re: [ANNOUNCE] kvm-75 release

[Jim Paris]

Avi Kivity wrote:
> Jim Paris wrote:
>> Avi Kivity wrote:
>>   
>>> - restore option roms to their original content on reset
>>>   - fixes extboot failures after reset
>>
>> It doesn't seem to work here:
>
> grub is probably booting via int 19 or jumping to the reset vector, so  
> kvm doesn't see the reboot.

You're right:

ENTRY(grub_reboot)
        call    EXT_C(prot_to_real)
        .code16
        /* cold boot */
        movw    $0x0472, %di
        movw    %ax, (%di)
        ljmp    $0xFFFF, $0x0000

> If that's the case, we can either implement  
> the chipset's memory control registers properly and teach the bios to  
> program them, or add a hack in the bios to trigger the copying without  
> the full reset.

My testing was a bit flawed.  With kvm-74, I noticed that reboot from
Linux had the problem, and reboot from grub failed the same way.  With
kvm-75, I only tested reboot from grub.  I can verify that reboot
from Linux works fine with virtio and kvm-75, and so the current
behavior is actually fine for us.

Thanks,
-jim