Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1402736 > unrolled thread

Re: [PATCH v2 2/9] powerpc/kvm: make hypervisor state restore a function

Started byGautham R Shenoy <ego@linux.vnet.ibm.com>
First post2016-05-18 08:30 +0200
Last post2016-05-19 16:30 +0200
Articles 3 — 2 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.


Contents

  Re: [PATCH v2 2/9] powerpc/kvm: make hypervisor state restore a  function Gautham R Shenoy <ego@linux.vnet.ibm.com> - 2016-05-18 08:30 +0200
    Re: [PATCH v2 2/9] powerpc/kvm: make hypervisor state restore a function Shreyas B Prabhu <shreyas@linux.vnet.ibm.com> - 2016-05-18 09:10 +0200
      Re: [PATCH v2 2/9] powerpc/kvm: make hypervisor state restore a  function Gautham R Shenoy <ego@linux.vnet.ibm.com> - 2016-05-19 16:30 +0200

#1402736 — Re: [PATCH v2 2/9] powerpc/kvm: make hypervisor state restore a function

FromGautham R Shenoy <ego@linux.vnet.ibm.com>
Date2016-05-18 08:30 +0200
SubjectRe: [PATCH v2 2/9] powerpc/kvm: make hypervisor state restore a function
Message-ID<rA39v-6Ng-5@gated-at.bofh.it>
Hi Shreyas,

On Tue, May 03, 2016 at 01:54:31PM +0530, Shreyas B. Prabhu wrote:
> In the current code, when the thread wakes up in reset vector, some
> of the state restore code and check for whether a thread needs to
> branch to kvm is duplicated. Reorder the code such that this
> duplication is avoided.
> 
> At a higher level this is what the change looks like-

I have manually verified that the code flow in the new patch is has
the same effect as whatever we were doing earlier. There a couple of
comments inline.

> diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S
> index 7716ceb..7ebfbb0 100644
> --- a/arch/powerpc/kernel/exceptions-64s.S
> +++ b/arch/powerpc/kernel/exceptions-64s.S
> @@ -107,25 +107,8 @@ BEGIN_FTR_SECTION
>  	beq	9f
> 
>  	cmpwi	cr3,r13,2
> +	bl	power7_restore_hyp_resource
> 
> -	/*
> -	 * Check if last bit of HSPGR0 is set. This indicates whether we are
> -	 * waking up from winkle.
> -	 */
> -	GET_PACA(r13)
> -	clrldi	r5,r13,63
> -	clrrdi	r13,r13,1
> -	cmpwi	cr4,r5,1
> -	mtspr	SPRN_HSPRG0,r13
> -
> -	lbz	r0,PACA_THREAD_IDLE_STATE(r13)
> -	cmpwi   cr2,r0,PNV_THREAD_NAP
> -	bgt     cr2,8f				/* Either sleep or Winkle */
> -
> -	/* Waking up from nap should not cause hypervisor state loss */
> -	bgt	cr3,.
> -
> -	/* Waking up from nap */
>  	li	r0,PNV_THREAD_RUNNING
>  	stb	r0,PACA_THREAD_IDLE_STATE(r13)	/* Clear thread state */
> 
> @@ -143,13 +126,9 @@ BEGIN_FTR_SECTION
> 
>  	/* Return SRR1 from power7_nap() */
>  	mfspr	r3,SPRN_SRR1
> -	beq	cr3,2f
> -	b	power7_wakeup_noloss
> -2:	b	power7_wakeup_loss
> -
> -	/* Fast Sleep wakeup on PowerNV */
> -8:	GET_PACA(r13)

In the old code, we do a GET_PACA(r13) before invoking the
power7_wakeup_tb_loss. In the new code we don't. Can you explain
this omission ?


[..snip..]

> @@ -420,33 +451,9 @@ common_exit:
> 
>  hypervisor_state_restored:
> 
> -	li	r5,PNV_THREAD_RUNNING
> -	stb     r5,PACA_THREAD_IDLE_STATE(r13)
> -
>  	mtspr	SPRN_SRR1,r16
> -#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
> -	li      r0,KVM_HWTHREAD_IN_KERNEL
> -	stb     r0,HSTATE_HWTHREAD_STATE(r13)
> -	/* Order setting hwthread_state vs. testing hwthread_req */
> -	sync
> -	lbz     r0,HSTATE_HWTHREAD_REQ(r13)
> -	cmpwi   r0,0
> -	beq     6f
> -	b       kvm_start_guest
> -6:
> -#endif
> -
> -	REST_NVGPRS(r1)
> -	REST_GPR(2, r1)
> -	ld	r3,_CCR(r1)
> -	ld	r4,_MSR(r1)
> -	ld	r5,_NIP(r1)
> -	addi	r1,r1,INT_FRAME_SIZE
> -	mtcr	r3
> -	mfspr	r3,SPRN_SRR1		/* Return SRR1 */
> -	mtspr	SPRN_SRR1,r4
> -	mtspr	SPRN_SRR0,r5
> -	rfid
> +	mtlr	r17
> +	blr


Perhaps you could add a comment against this blr to indicate that we
go back to the reset vector right after the call to
power7_restore_hyp_resource. 

> 
>  fastsleep_workaround_at_exit:
>  	li	r3,1
> -- 
> 2.4.11
> 

[toc] | [next] | [standalone]


#1402753 — Re: [PATCH v2 2/9] powerpc/kvm: make hypervisor state restore a function

FromShreyas B Prabhu <shreyas@linux.vnet.ibm.com>
Date2016-05-18 09:10 +0200
SubjectRe: [PATCH v2 2/9] powerpc/kvm: make hypervisor state restore a function
Message-ID<rA3Md-7f3-5@gated-at.bofh.it>
In reply to#1402736
Hi Gautham,

On 05/18/2016 11:55 AM, Gautham R Shenoy wrote:
> Hi Shreyas,
> 
> On Tue, May 03, 2016 at 01:54:31PM +0530, Shreyas B. Prabhu wrote:
>> In the current code, when the thread wakes up in reset vector, some
>> of the state restore code and check for whether a thread needs to
>> branch to kvm is duplicated. Reorder the code such that this
>> duplication is avoided.
>>
>> At a higher level this is what the change looks like-
> 
> I have manually verified that the code flow in the new patch is has
> the same effect as whatever we were doing earlier. There a couple of
> comments inline.
> 
Thanks for the review!

>> diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S
>> index 7716ceb..7ebfbb0 100644
>> --- a/arch/powerpc/kernel/exceptions-64s.S
>> +++ b/arch/powerpc/kernel/exceptions-64s.S
>> @@ -107,25 +107,8 @@ BEGIN_FTR_SECTION
>>  	beq	9f
>>
>>  	cmpwi	cr3,r13,2
>> +	bl	power7_restore_hyp_resource
>>
>> -	/*
>> -	 * Check if last bit of HSPGR0 is set. This indicates whether we are
>> -	 * waking up from winkle.
>> -	 */
>> -	GET_PACA(r13)
>> -	clrldi	r5,r13,63
>> -	clrrdi	r13,r13,1
>> -	cmpwi	cr4,r5,1
>> -	mtspr	SPRN_HSPRG0,r13
>> -
>> -	lbz	r0,PACA_THREAD_IDLE_STATE(r13)
>> -	cmpwi   cr2,r0,PNV_THREAD_NAP
>> -	bgt     cr2,8f				/* Either sleep or Winkle */
>> -
>> -	/* Waking up from nap should not cause hypervisor state loss */
>> -	bgt	cr3,.
>> -
>> -	/* Waking up from nap */
>>  	li	r0,PNV_THREAD_RUNNING
>>  	stb	r0,PACA_THREAD_IDLE_STATE(r13)	/* Clear thread state */
>>
>> @@ -143,13 +126,9 @@ BEGIN_FTR_SECTION
>>
>>  	/* Return SRR1 from power7_nap() */
>>  	mfspr	r3,SPRN_SRR1
>> -	beq	cr3,2f
>> -	b	power7_wakeup_noloss
>> -2:	b	power7_wakeup_loss
>> -
>> -	/* Fast Sleep wakeup on PowerNV */
>> -8:	GET_PACA(r13)
> 
> In the old code, we do a GET_PACA(r13) before invoking the
> power7_wakeup_tb_loss. In the new code we don't. Can you explain
> this omission ?

GET_PACA(13) is the called in the beginning of
power7_restore_hyp_resource. So r13 contains pointer to PACA when
power7_wakeup_tb_loss invoked later in the same function.
> 
> 
> [..snip..]
> 
>> @@ -420,33 +451,9 @@ common_exit:
>>
>>  hypervisor_state_restored:
>>
>> -	li	r5,PNV_THREAD_RUNNING
>> -	stb     r5,PACA_THREAD_IDLE_STATE(r13)
>> -
>>  	mtspr	SPRN_SRR1,r16
>> -#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
>> -	li      r0,KVM_HWTHREAD_IN_KERNEL
>> -	stb     r0,HSTATE_HWTHREAD_STATE(r13)
>> -	/* Order setting hwthread_state vs. testing hwthread_req */
>> -	sync
>> -	lbz     r0,HSTATE_HWTHREAD_REQ(r13)
>> -	cmpwi   r0,0
>> -	beq     6f
>> -	b       kvm_start_guest
>> -6:
>> -#endif
>> -
>> -	REST_NVGPRS(r1)
>> -	REST_GPR(2, r1)
>> -	ld	r3,_CCR(r1)
>> -	ld	r4,_MSR(r1)
>> -	ld	r5,_NIP(r1)
>> -	addi	r1,r1,INT_FRAME_SIZE
>> -	mtcr	r3
>> -	mfspr	r3,SPRN_SRR1		/* Return SRR1 */
>> -	mtspr	SPRN_SRR1,r4
>> -	mtspr	SPRN_SRR0,r5
>> -	rfid
>> +	mtlr	r17
>> +	blr
> 
> 
> Perhaps you could add a comment against this blr to indicate that we
> go back to the reset vector right after the call to
> power7_restore_hyp_resource. 

Ok. I'll do that.
> 
>>
>>  fastsleep_workaround_at_exit:
>>  	li	r3,1
>> -- 
>> 2.4.11
>>

[toc] | [prev] | [next] | [standalone]


#1403777

FromGautham R Shenoy <ego@linux.vnet.ibm.com>
Date2016-05-19 16:30 +0200
Message-ID<rAx7E-Wa-3@gated-at.bofh.it>
In reply to#1402753
Hi Shreyas,

On Wed, May 18, 2016 at 12:37:56PM +0530, Shreyas B Prabhu wrote:

[..snip..]
> >> diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S
> >> index 7716ceb..7ebfbb0 100644
> >> --- a/arch/powerpc/kernel/exceptions-64s.S
> >> +++ b/arch/powerpc/kernel/exceptions-64s.S
> >> @@ -107,25 +107,8 @@ BEGIN_FTR_SECTION
> >>  	beq	9f
> >>
> >>  	cmpwi	cr3,r13,2
> >> +	bl	power7_restore_hyp_resource
> >>
> >> -	/*
> >> -	 * Check if last bit of HSPGR0 is set. This indicates whether we are
> >> -	 * waking up from winkle.
> >> -	 */
> >> -	GET_PACA(r13)
> >> -	clrldi	r5,r13,63
> >> -	clrrdi	r13,r13,1
> >> -	cmpwi	cr4,r5,1
> >> -	mtspr	SPRN_HSPRG0,r13
> >> -
> >> -	lbz	r0,PACA_THREAD_IDLE_STATE(r13)
> >> -	cmpwi   cr2,r0,PNV_THREAD_NAP
> >> -	bgt     cr2,8f				/* Either sleep or Winkle */
> >> -
> >> -	/* Waking up from nap should not cause hypervisor state loss */
> >> -	bgt	cr3,.
> >> -
> >> -	/* Waking up from nap */
> >>  	li	r0,PNV_THREAD_RUNNING
> >>  	stb	r0,PACA_THREAD_IDLE_STATE(r13)	/* Clear thread state */
> >>
> >> @@ -143,13 +126,9 @@ BEGIN_FTR_SECTION
> >>
> >>  	/* Return SRR1 from power7_nap() */
> >>  	mfspr	r3,SPRN_SRR1
> >> -	beq	cr3,2f
> >> -	b	power7_wakeup_noloss
> >> -2:	b	power7_wakeup_loss
> >> -
> >> -	/* Fast Sleep wakeup on PowerNV */
> >> -8:	GET_PACA(r13)
> > 
> > In the old code, we do a GET_PACA(r13) before invoking the
> > power7_wakeup_tb_loss. In the new code we don't. Can you explain
> > this omission ?
> 
> GET_PACA(13) is the called in the beginning of
> power7_restore_hyp_resource. So r13 contains pointer to PACA when
> power7_wakeup_tb_loss invoked later in the same function.

Ah, I see it now. So the GET_PACA(r13) at 8: was anyway redundant in
the older code.

You can add my Reviewed-by: to this patch.

--
Thanks and Regards
gautham.

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web