[00/03][PATCH] kvm-ia64 updates for linux-2.6.28-rc2

[00/03][PATCH] kvm-ia64 updates for linux-2.6.28-rc2

[Zhang, Xiantao]

Hi, Avi
	Please review and apply the three patches!  The last two are key fixes for linux-2.6.28-rc2, and please also push them into upstream. 
Thanks!
Xiantao

Re: [00/03][PATCH] kvm-ia64 updates for linux-2.6.28-rc2

[Avi Kivity]

Zhang, Xiantao wrote:
> Hi, Avi
> 	Please review and apply the three patches!  The last two are key fixes for linux-2.6.28-rc2, and please also push them into upstream. 
>   

Applied all, thanks.  Is the third patch needed for 2.6.28?  What 
scenario does it fix?

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

Re: [00/03][PATCH] kvm-ia64 updates for linux-2.6.28-rc2

[Jes Sorensen]

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

Avi> Zhang, Xiantao wrote:
>> Hi, Avi Please review and apply the three patches!  The last two
>> are key fixes for linux-2.6.28-rc2, and please also push them into
>> upstream.

Avi> Applied all, thanks.  Is the third patch needed for 2.6.28?  What
Avi> scenario does it fix?

Please!

The third patch reorganizes the data structures which saves us a ton
of space so we can go from 8 to 60 vcpus .... Would like it in as I
have stuff that I want to fix that goes on top of this.

Thanks,
Jes

Re: [00/03][PATCH] kvm-ia64 updates for linux-2.6.28-rc2

[Avi Kivity]

Jes Sorensen wrote:
> Avi> Applied all, thanks.  Is the third patch needed for 2.6.28?  What
> Avi> scenario does it fix?
>
> Please!
>
> The third patch reorganizes the data structures which saves us a ton
> of space so we can go from 8 to 60 vcpus .... Would like it in as I
> have stuff that I want to fix that goes on top of this.
>
>   

No, that's the first patch.  And it was applied, but for 2.6.29, not 2.6.28.

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

Re: [00/03][PATCH] kvm-ia64 updates for linux-2.6.28-rc2

[Jes Sorensen]

Avi Kivity wrote:
> Jes Sorensen wrote:
>> The third patch reorganizes the data structures which saves us a ton
>> of space so we can go from 8 to 60 vcpus .... Would like it in as I
>> have stuff that I want to fix that goes on top of this.  
> 
> No, that's the first patch.  And it was applied, but for 2.6.29, not 
> 2.6.28.

DOH! You're right! I just looked at the 3rd patch in my inbox, without
checking the subject line :-)

I'll let Xiantao answer, I don't know the 3rd patch.

cheers
Jes

RE: [00/03][PATCH] kvm-ia64 updates for linux-2.6.28-rc2

[Zhang, Xiantao]

Avi Kivity wrote:
> Zhang, Xiantao wrote:
>> Hi, Avi
>> 	Please review and apply the three patches!  The last two are key
>> fixes for linux-2.6.28-rc2, and please also push them into upstream. 
>> 
> 
> Applied all, thanks.  Is the third patch needed for 2.6.28?  What
> scenario does it fix?
Hi, Avi 
	It is a potential issue. If the vcpu goes to halt with the lock taken, other vcpus wants to take write lock, they have to wait until the all halted vcpus are waken up, so  it may impact performance.  :)
Thanks
Xiantao

Re: [00/03][PATCH] kvm-ia64 updates for linux-2.6.28-rc2

[Avi Kivity]

Zhang, Xiantao wrote:
> 	It is a potential issue. If the vcpu goes to halt with the lock taken, other vcpus wants to take write lock, they have to wait until the all halted vcpus are waken up, so  it may impact performance.  :)

What is the impact?  Taking slots_lock for write is very rare.

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

RE: [00/03][PATCH] kvm-ia64 updates for linux-2.6.28-rc2

[Zhang, Xiantao]

Avi Kivity wrote:
> Zhang, Xiantao wrote:
>> 	It is a potential issue. If the vcpu goes to halt with the lock
>> taken, other vcpus wants to take write lock, they have to wait until
>> the all halted vcpus are waken up, so  it may impact performance. 
>> :)   
> 
> What is the impact?  Taking slots_lock for write is very rare.
Agree, so I just say it is a potential issue in future. If not proper for 2.6.28, it should be also Okay to merge it into 2.6.29:)
Xiantao