On Thu, 2010-06-10 at 11:50 +0200, Peter Zijlstra wrote:
> On Thu, 2010-06-10 at 10:21 +0800, Zhang, Yanmin wrote:
> > . The ABI mostly includes the definition of struct perf_event_attr,
> > guest_perf_counter,
> > and hypercalls.
>
> Note that perf_event_attr isn't guaranteed to be stable betwee
On Thu, 2010-06-10 at 10:21 +0800, Zhang, Yanmin wrote:
> . The ABI mostly includes the definition of struct perf_event_attr,
> guest_perf_counter,
> and hypercalls.
Note that perf_event_attr isn't guaranteed to be stable between kernels,
it can grow.
--
To unsubscribe from this list: send the li
On Thu, 2010-06-10 at 06:06 +0300, Avi Kivity wrote:
> On 06/10/2010 05:21 AM, Zhang, Yanmin wrote:
> >
> >> I see. So put it in a union. Or perhaps not even in a union - what if
> >> a kvm guest is also acting as a kvm host?
> >>
> > My patch has consideration on it. I compiled kernel with
On 06/10/2010 05:21 AM, Zhang, Yanmin wrote:
I see. So put it in a union. Or perhaps not even in a union - what if
a kvm guest is also acting as a kvm host?
My patch has consideration on it. I compiled kernel with host and guest support
at the same time. The accessing to perf_event->sh
On Wed, 2010-06-09 at 12:41 +0300, Avi Kivity wrote:
> On 06/09/2010 12:21 PM, Zhang, Yanmin wrote:
> >
> >> One thing that's missing is documentation of the guest/host ABI. It
> >> will be a requirement for inclusion, but it will also be a great help
> >> for review, so please provide it ASAP.
>
On 06/09/2010 01:08 PM, Peter Zijlstra wrote:
On Wed, 2010-06-09 at 12:41 +0300, Avi Kivity wrote:
Disabling the watchdog is unfortunate. Why is it necessary?
perf always uses NMI, so we disable the nmi_watchdog when a perf_event is
set up in case they might have impact.
* Peter Zijlstra wrote:
> On Wed, 2010-06-09 at 12:41 +0300, Avi Kivity wrote:
> >
> > >> Disabling the watchdog is unfortunate. Why is it necessary?
> > >>
> > > perf always uses NMI, so we disable the nmi_watchdog when a perf_event is
> > > set up in case they might have impact.
> > >
On Wed, 2010-06-09 at 12:41 +0300, Avi Kivity wrote:
>
> >> Disabling the watchdog is unfortunate. Why is it necessary?
> >>
> > perf always uses NMI, so we disable the nmi_watchdog when a perf_event is
> > set up in case they might have impact.
> >
>
> Ok. Is that the case for the ha
On 06/09/2010 12:30 PM, Zhang, Yanmin wrote:
On Wed, 2010-06-09 at 11:59 +0300, Avi Kivity wrote:
On 06/09/2010 06:30 AM, Zhang, Yanmin wrote:
From: Zhang, Yanmin
Based on Ingo's idea, I implement a para virt interface for perf to support
statistics collection in guest os. That means
On 06/09/2010 12:21 PM, Zhang, Yanmin wrote:
One thing that's missing is documentation of the guest/host ABI. It
will be a requirement for inclusion, but it will also be a great help
for review, so please provide it ASAP.
I will add such document. It will includes:
1) Data struct perf_e
On Wed, 2010-06-09 at 11:59 +0300, Avi Kivity wrote:
> On 06/09/2010 06:30 AM, Zhang, Yanmin wrote:
> > From: Zhang, Yanmin
> >
> > Based on Ingo's idea, I implement a para virt interface for perf to support
> > statistics collection in guest os. That means we could run tool perf in
> > guest
> >
On Wed, 2010-06-09 at 11:33 +0300, Avi Kivity wrote:
> On 06/09/2010 06:30 AM, Zhang, Yanmin wrote:
> > From: Zhang, Yanmin
> >
> > Based on Ingo's idea, I implement a para virt interface for perf to support
> > statistics collection in guest os. That means we could run tool perf in
> > guest
> >
On 06/09/2010 06:30 AM, Zhang, Yanmin wrote:
From: Zhang, Yanmin
Based on Ingo's idea, I implement a para virt interface for perf to support
statistics collection in guest os. That means we could run tool perf in guest
os directly.
Great thanks to Peter Zijlstra. He is really the architect and
On 06/09/2010 06:30 AM, Zhang, Yanmin wrote:
From: Zhang, Yanmin
Based on Ingo's idea, I implement a para virt interface for perf to support
statistics collection in guest os. That means we could run tool perf in guest
os directly.
Great thanks to Peter Zijlstra. He is really the architect and
From: Zhang, Yanmin
Based on Ingo's idea, I implement a para virt interface for perf to support
statistics collection in guest os. That means we could run tool perf in guest
os directly.
Great thanks to Peter Zijlstra. He is really the architect and gave me
architecture
design suggestions. I al
15 matches
Mail list logo
