[PATCH 0/2]KVM: In-kernel PIT model

[PATCH 0/2]KVM: In-kernel PIT model

[Yang, Sheng]

Hi,

Here is the updated in-kernel PIT model. The main improvements are:

1. The support of SMP PAE host is OK now.

2. The strategy of inject timer interrupt changed. To prevent much more 
interrupt injection in some SMP guest(for example, more than 50k trying for 
1k interrupts per second), the strategy was changed to try to ensure last 
interrupt was injected, then inject the next one. Also add time out to in 
case of interrupts lost.

3. Find a bug of QEMU PIT causing a little more PIT interrupts were 
injected(36.4 interrupts were injected per second, causing 2.6.16 guest timer 
run faster more than 1/10). Disable QEMU PIT now.

The test result is pretty well with both UP and SMP guest, mostly with default 
kernel paramters. But RHEL5/5.1 SMP 
PAE guest need "clocksource=pit" to make it stable, otherwise may meet the 
same problem in currently RHEL5.1 SMP PAE guest. If hang occurs, seems only 
the time out part of PIT interrupts would be injected. For seems it's due to 
other clocksource, I would check that later.

-- 
Thanks
Yang, Sheng

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

Re: [PATCH 0/2]KVM: In-kernel PIT model

[Avi Kivity]

Yang, Sheng wrote:
> Hi,
>
> Here is the updated in-kernel PIT model. The main improvements are:
>   

Looking good; apart from some very minor comments all that is missing is 
save/restore support.

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


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/