Dong, Eddie wrote:
>
> These 3 patches are actually independent. Previously you say you want
> to see SMP guest first. Now Xin got it up, and this one works fine.
> (Xin is based on this since very beginning we suspect Linux AP is
> blocked
> by the widely used lock.) So, I think we can check in th
Avi Kivity wrote:
> Dong, Eddie wrote:
>> Avi Kivity wrote:
>>
>>> Dong, Eddie wrote:
>>>
Avi:
Any comments to this patch? Seems not in yet.
The sequence is apic-timer-TMCCT.patch,
>>> apic-timer-last5.patch and apic-timer-pending4.patch.
>>>
thx,eddie
Dong, Eddie wrote:
> Avi Kivity wrote:
>
>> Dong, Eddie wrote:
>>
>>> Avi:
>>> Any comments to this patch? Seems not in yet.
>>> The sequence is apic-timer-TMCCT.patch,
>>>
>> apic-timer-last5.patch and apic-timer-pending4.patch.
>>
>>> thx,eddie
>>>
>>>
>> I'd li
Avi:
Here is the rebased one toward time issue fix and lock removal. The
comming SMP serial patches are based on APIC lock removal.
thanks, eddie
1-tmcct.patch
Description: 1-tmcct.patch
2-last.patch
Description: 2-last.patch
3-pending.patch
Description: 3-pending.patch
-
Avi Kivity wrote:
> Dong, Eddie wrote:
>> Avi:
>> Any comments to this patch? Seems not in yet.
>> The sequence is apic-timer-TMCCT.patch,
> apic-timer-last5.patch and apic-timer-pending4.patch.
>> thx,eddie
>>
>
> I'd like to get guest smp in first...
>
Avi:
We have got SMP L
Dong, Eddie wrote:
> Avi:
> Any comments to this patch? Seems not in yet.
> The sequence is apic-timer-TMCCT.patch, apic-timer-last5.patch and
> apic-timer-pending4.patch.
> thx,eddie
>
I'd like to get guest smp in first...
--
error compiling committee.c: too many arguments to fu
ent: 2007年8月14日 11:25
>To: Avi Kivity
>Cc: kvm-devel
>Subject: Re: [kvm-devel] add back pending timer irqs for
>kernel APIC timer
>
>Here are rebased and fixed patches, please check.
>thanks, eddie
>
>
apic-timer-tmcct.patch
Description: apic-timer-tmcct.patch
apic
Here are rebased and fixed patches, please check.
thanks, eddie
commit 9b1a9ad955d9ac44ee6e60b4a565edf044d3dcc7
Author: root <[EMAIL PROTECTED](none)>
Date: Fri Aug 10 22:38:59 2007 +0800
APIC_TMCCT is a dynamically running count which always
need to be recalculated at the time it is r
Dong, Eddie wrote:
> Avi Kivity wrote:
>
>>> 2: stop hrtimer when the guest is descheduled to increase
>>> scalibility and remove apic->lock.
>>>
>>>
>> Why is this important? An hrtimer is just an entry on a list, no?
>>
>
> Does it need an IRQ to drive? If we have ten VMs, we got
Avi Kivity wrote:
>> 2: stop hrtimer when the guest is descheduled to increase
>> scalibility and remove apic->lock.
>>
>
> Why is this important? An hrtimer is just an entry on a list, no?
Does it need an IRQ to drive? If we have ten VMs, we got extra 10 IRQs.
>
> Do you mean, just if the g
Dong, Eddie wrote:
> [EMAIL PROTECTED] wrote:
>
kvm_pt_update_irq is within intr_assist where host IRQ is disabled.
It can be updated only after we return to guest.
>>> What if it runs on another cpu?
>>>
>> This case will be forbidden in next patch. We need t
Dong, Eddie wrote:
>> Note that my old lapic branch did a similar "tdf" thing as well and it
>> worked quite nicely, so I think Eddie is on the right track with
>> adding this.
>>
>
> I just copied the whole APIC timer stuff from your old patch at that
> time as TODO. maybe I missed something
[EMAIL PROTECTED] wrote:
>>> kvm_pt_update_irq is within intr_assist where host IRQ is disabled.
>>> It can be updated only after we return to guest.
>>>
>>
>> What if it runs on another cpu?
>
> This case will be forbidden in next patch. We need to remove
> apic->lock too.
>
To avoid confuse,
>> kvm_pt_update_irq is within intr_assist where host IRQ is disabled.
>> It can be updated only after we return to guest.
>>
>
> What if it runs on another cpu?
This case will be forbidden in next patch. We need to remove apic->lock
too.
>
>>> How about "kvm_inject_pending_timer_irqs"
>
>
>
> Note that my old lapic branch did a similar "tdf" thing as well and it
> worked quite nicely, so I think Eddie is on the right track with
> adding this.
I just copied the whole APIC timer stuff from your old patch at that
time as TODO. maybe I missed something :-(
Time virtualization has
Dong, Eddie wrote:
>>
>> While this improves throughput, doesn't it decrease responsiveness?
>> Suppose the guest is sleeping and there is no activity except
>> for lapic
>> interrupts. Won't the wakeup get deferred indefinitely?
>>
>
> didn't catch here. do u mean the guest is in HLT ? There
Avi Kivity wrote:
> What does "pt" mean in this context? "pending timer"?
>
> Suggested descriptive names below.
Periodic timer. Comments added.
>> if (apic_lvtt_period(apic)) {
>> -u32 offset;
>> -u32 tmict = apic_get_reg(apic, APIC_TMICT);
>> -
>> -o
On Mon, 2007-08-13 at 11:50 +0300, Avi Kivity wrote:
> > This patch keep track of the pending irqs and inject them back to guest
> > eventually
> > even the guest may be descheduled. This is also what we did in Xen.
> > BTW, This policy will also be applied to future kernel PIT, I just do it
> > s
Dong, Eddie wrote:
> [EMAIL PROTECTED] wrote:
>
>>Add back pending irqs for apic timer to get precise guestAPIC
>> timer interrupt.
>>
>>Signed-off-by: Yaozu (Eddie) Dong <[EMAIL PROTECTED]>
>>
>>
>>
> a typo, please use this one.
>
> With above patches, now guest timer is much
Dong, Eddie wrote:
> Avi Kivity wrote:
>
>> Dong, Eddie wrote:
>>
>>> Add back pending irqs for apic timer to get precise guest
>>> APIC timer interrupt.
>>>
>>>
>> Can you explain the problem and the solution in more detail?
>>
>
> Today guest sleep 10 seconds only get
Avi Kivity wrote:
> Dong, Eddie wrote:
>> Add back pending irqs for apic timer to get precise guest
>> APIC timer interrupt.
>>
>
> Can you explain the problem and the solution in more detail?
Today guest sleep 10 seconds only get about 5-6 seconds in host time.
A design philosiphy in v
Dong, Eddie wrote:
> Add back pending irqs for apic timer to get precise guest
> APIC timer interrupt.
>
Can you explain the problem and the solution in more detail?
-
This SF.net email is sponsored by: Splunk Inc
[EMAIL PROTECTED] wrote:
>Add back pending irqs for apic timer to get precise guestAPIC
> timer interrupt.
>
>Signed-off-by: Yaozu (Eddie) Dong <[EMAIL PROTECTED]>
>
>
a typo, please use this one.
With above patches, now guest timer is much accurate! sleep 60
get exactly 60 second
23 matches
Mail list logo
