[PATCH] kvm: arm: vgic: Let one looping code instead of two looping code

[PATCH] kvm: arm: vgic: Let one looping code instead of two looping code

[Chen Gang]

Use one looping instead of two, so can let code more simpler and get a
little better performance.

Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
---
 virt/kvm/arm/vgic.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/virt/kvm/arm/vgic.c b/virt/kvm/arm/vgic.c
index cf23c13..eb1a51b 100644
--- a/virt/kvm/arm/vgic.c
+++ b/virt/kvm/arm/vgic.c
@@ -1959,9 +1959,6 @@ int kvm_vgic_create(struct kvm *kvm)
 		if (!mutex_trylock(&vcpu->mutex))
 			goto out_unlock;
 		vcpu_lock_idx = i;
-	}
-
-	kvm_for_each_vcpu(i, vcpu, kvm) {
 		if (vcpu->arch.has_run_once)
 			goto out_unlock;
 	}
-- 
1.9.3

[PATCH] kvm: arm: vgic: Let one looping code instead of two looping code

[Christoffer Dall]

On Sun, Nov 30, 2014 at 12:18:04PM +0800, Chen Gang wrote:
> Use one looping instead of two, so can let code more simpler and get a
> little better performance.
> 
> Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>

Performance?  Please, this is vgic_create....

This does nothing else than introduce churn.

-Christoffer

[PATCH] kvm: arm: vgic: Let one looping code instead of two looping code

[Chen Gang]

On 11/30/2014 06:07 PM, Christoffer Dall wrote:
> On Sun, Nov 30, 2014 at 12:18:04PM +0800, Chen Gang wrote:
>> Use one looping instead of two, so can let code more simpler and get a
>> little better performance.
>>
>> Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
> 
> Performance?  Please, this is vgic_create....
> 

I guess your meaning is "this code is not performance sensitive", do not
mean "1 looping can not get a little better performance than 2 looping".
If what I guess is incorrect, please let me know.

> This does nothing else than introduce churn.
> 

I don't think it does nothing, it can let code simpler than before:
remove the redundant looping -- the 2 looping do the same thing which
is commented just above the firstly looping.

And I don't know which churn this patch will introduce. At least, no one
want to waste him/her time resources to intend churn, please provide
more details about it.

If kvm does not like me, or does not like members to find issues based
on reading source code, please let me know (no reply means kvm does not
like me or this way), I will quit, so can save time resources both.


Thanks.
-- 
Chen Gang

Open share and attitude like air water and life which God blessed

[PATCH] kvm: arm: vgic: Let one looping code instead of two looping code

[Christoffer Dall]

On Mon, Dec 01, 2014 at 10:04:32PM +0800, Chen Gang wrote:
> On 11/30/2014 06:07 PM, Christoffer Dall wrote:
> > On Sun, Nov 30, 2014 at 12:18:04PM +0800, Chen Gang wrote:
> >> Use one looping instead of two, so can let code more simpler and get a
> >> little better performance.
> >>
> >> Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
> > 
> > Performance?  Please, this is vgic_create....
> > 
> 
> I guess your meaning is "this code is not performance sensitive", do not
> mean "1 looping can not get a little better performance than 2 looping".
> If what I guess is incorrect, please let me know.
> 

I doubt if this is even measureable.  Have you even looked at how this
compiles?

> > This does nothing else than introduce churn.
> > 
> 
> I don't think it does nothing, it can let code simpler than before:
> remove the redundant looping -- the 2 looping do the same thing which
> is commented just above the firstly looping.

I don't think there's any big benefit here.

> 
> And I don't know which churn this patch will introduce. At least, no one
> want to waste him/her time resources to intend churn, please provide
> more details about it.
> 
> If kvm does not like me, or does not like members to find issues based
> on reading source code, please let me know (no reply means kvm does not
> like me or this way), I will quit, so can save time resources both.
> 
I appreciate your efforts and welcome anybody to contribute to the code
base.  But you seem to be insisting on changing some insignificant
details, and we just have bigger fish to fry right now.  Sorry.

This is not about anyone not liking someone, it is about the code and
whether there's a need to take this patch right now or not.

-Christoffer

[PATCH] kvm: arm: vgic: Let one looping code instead of two looping code

[Chen Gang]

On 11/30/14 19:08, Christoffer Dall wrote:
> On Mon, Dec 01, 2014 at 10:04:32PM +0800, Chen Gang wrote:
>> On 11/30/2014 06:07 PM, Christoffer Dall wrote:
>>> On Sun, Nov 30, 2014 at 12:18:04PM +0800, Chen Gang wrote:
>>>> Use one looping instead of two, so can let code more simpler and get a
>>>> little better performance.
>>>>
>>>> Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
>>>
>>> Performance?  Please, this is vgic_create....
>>>
>>
>> I guess your meaning is "this code is not performance sensitive", do not
>> mean "1 looping can not get a little better performance than 2 looping".
>> If what I guess is incorrect, please let me know.
>>
> 
> I doubt if this is even measureable.  Have you even looked at how this
> compiles?
> 

I did not look at the related assembly code for O2.

But for me, it must be a little better than 2 loops: as far as I know,
the compiler should/can not merge the 2 loops into 1 loop (it will
change logical working flow, at least, need notice about lock/unlock).


>>> This does nothing else than introduce churn.
>>>
>>
>> I don't think it does nothing, it can let code simpler than before:
>> remove the redundant looping -- the 2 looping do the same thing which
>> is commented just above the firstly looping.
> 
> I don't think there's any big benefit here.
> 

Really, no big benefit here.

>>
>> And I don't know which churn this patch will introduce. At least, no one
>> want to waste him/her time resources to intend churn, please provide
>> more details about it.
>>
>> If kvm does not like me, or does not like members to find issues based
>> on reading source code, please let me know (no reply means kvm does not
>> like me or this way), I will quit, so can save time resources both.
>>
> I appreciate your efforts and welcome anybody to contribute to the code
> base.  But you seem to be insisting on changing some insignificant
> details, and we just have bigger fish to fry right now.  Sorry.
> 
> This is not about anyone not liking someone, it is about the code and
> whether there's a need to take this patch right now or not.
> 

OK, thanks. Next, I shall try to find some bugs for it. And they are not
urgent, since they are only found by reading source code.


And could you provide any information or suggestion for kvm common test?
(e.g. "make check" in gcc or qemu). At present, I only let my kvm
related patches pass compiling, not give additional test.

Thanks.
-- 
Chen Gang

Open, share, and attitude like air, water, and life which God blessed