[PATCH 1/2] KVM: PPC: Pass through program interrupts

[PATCH 1/2] KVM: PPC: Pass through program interrupts

[Alexander Graf]

When we get a program interrupt in guest kernel mode, we try to emulate the
instruction.

If that doesn't fail, we report to the user and try again - at the exact same
instruction pointer. So if the guest kernel really does trigger an invalid
instruction, we loop forever.

So let's better go and forward program exceptions to the guest when we don't
know the instruction we're supposed to emulate.

Signed-off-by: Alexander Graf <agraf@suse.de>
---
 arch/powerpc/kvm/emulate.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/arch/powerpc/kvm/emulate.c b/arch/powerpc/kvm/emulate.c
index 8b0ba0b..b905623 100644
--- a/arch/powerpc/kvm/emulate.c
+++ b/arch/powerpc/kvm/emulate.c
@@ -476,6 +476,7 @@ int kvmppc_emulate_instruction(struct kvm_run *run, struct kvm_vcpu *vcpu)
 			advance = 0;
 			printk(KERN_ERR "Couldn't emulate instruction 0x%08x "
 			       "(op %d xop %d)\n", inst, get_op(inst), get_xop(inst));
+			kvmppc_core_queue_program(vcpu, 0);
 		}
 	}
 
-- 
1.6.0.2

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Avi Kivity]

On 01/10/2010 04:27 AM, Alexander Graf wrote:
> When we get a program interrupt in guest kernel mode, we try to emulate the
> instruction.
>
> If that doesn't fail, we report to the user and try again - at the exact same
> instruction pointer. So if the guest kernel really does trigger an invalid
> instruction, we loop forever.
>
> So let's better go and forward program exceptions to the guest when we don't
> know the instruction we're supposed to emulate.
>
>    

Applied both, and queued for .33.  I was able to guess this from the 
description, but please mention it explicitly in the future.

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

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Alexander Graf]

On 10.01.2010, at 13:22, Avi Kivity wrote:

> On 01/10/2010 04:27 AM, Alexander Graf wrote:
>> When we get a program interrupt in guest kernel mode, we try to emulate the
>> instruction.
>> 
>> If that doesn't fail, we report to the user and try again - at the exact same
>> instruction pointer. So if the guest kernel really does trigger an invalid
>> instruction, we loop forever.
>> 
>> So let's better go and forward program exceptions to the guest when we don't
>> know the instruction we're supposed to emulate.
>> 
>>   
> 
> Applied both, and queued for .33.  I was able to guess this from the description, but please mention it explicitly in the future.

I'm not sure what to mention. From a stability point of view, all the Book3S patches should go in .33 because they all fix severe bugs.

But then again the question is what our goal for .33 is. If we only want to give out the message "KVM for Book3S is coming! Stay tuned!" then we shouldn't do any updates for .33. If we want to have something working to the best of our knowledge, all patches should go into .33.


Alex

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Avi Kivity]

On 01/10/2010 02:32 PM, Alexander Graf wrote:
>
>    
>> Applied both, and queued for .33.  I was able to guess this from the description, but please mention it explicitly in the future.
>>      
> I'm not sure what to mention. From a stability point of view, all the Book3S patches should go in .33 because they all fix severe bugs.
>
> But then again the question is what our goal for .33 is. If we only want to give out the message "KVM for Book3S is coming! Stay tuned!" then we shouldn't do any updates for .33. If we want to have something working to the best of our knowledge, all patches should go into .33.
>    

Given the other 9-patch series, I'll regard .33 as a sneak preview (look 
but don't touch) and .34 will be the first really serious kvm ppc64 release.

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

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Alexander Graf]

On 10.01.2010, at 13:37, Avi Kivity wrote:

> On 01/10/2010 02:32 PM, Alexander Graf wrote:
>> 
>>   
>>> Applied both, and queued for .33.  I was able to guess this from the description, but please mention it explicitly in the future.
>>>     
>> I'm not sure what to mention. From a stability point of view, all the Book3S patches should go in .33 because they all fix severe bugs.
>> 
>> But then again the question is what our goal for .33 is. If we only want to give out the message "KVM for Book3S is coming! Stay tuned!" then we shouldn't do any updates for .33. If we want to have something working to the best of our knowledge, all patches should go into .33.
>>   
> 
> Given the other 9-patch series, I'll regard .33 as a sneak preview (look but don't touch) and .34 will be the first really serious kvm ppc64 release.

That's what I figured too, so I didn't queue anything explicitly for .33 :-).

Alex

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Avi Kivity]

On 01/10/2010 02:22 PM, Avi Kivity wrote:
> On 01/10/2010 04:27 AM, Alexander Graf wrote:
>> When we get a program interrupt in guest kernel mode, we try to 
>> emulate the
>> instruction.
>>
>> If that doesn't fail, we report to the user and try again - at the 
>> exact same
>> instruction pointer. So if the guest kernel really does trigger an 
>> invalid
>> instruction, we loop forever.
>>
>> So let's better go and forward program exceptions to the guest when 
>> we don't
>> know the instruction we're supposed to emulate.
>>
>
> Applied both, and queued for .33.  I was able to guess this from the 
> description, but please mention it explicitly in the future.
>

I get this building ppc32:

scripts/kconfig/conf -s arch/powerpc/Kconfig
   CHK     include/linux/version.h
   CHK     include/generated/utsrelease.h
   CALL    scripts/checksyscalls.sh
<stdin>:1523:2: warning: #warning syscall recvmmsg not implemented
   CC      arch/powerpc/kvm/emulate.o
arch/powerpc/kvm/emulate.c: In function 'kvmppc_emulate_instruction':
arch/powerpc/kvm/emulate.c:467: error: too many arguments to function 
'kvmppc_core_queue_program'
make[1]: *** [arch/powerpc/kvm/emulate.o] Error 1



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

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Alexander Graf]

On 10.01.2010, at 15:26, Avi Kivity wrote:

> On 01/10/2010 02:22 PM, Avi Kivity wrote:
>> On 01/10/2010 04:27 AM, Alexander Graf wrote:
>>> When we get a program interrupt in guest kernel mode, we try to emulate the
>>> instruction.
>>> 
>>> If that doesn't fail, we report to the user and try again - at the exact same
>>> instruction pointer. So if the guest kernel really does trigger an invalid
>>> instruction, we loop forever.
>>> 
>>> So let's better go and forward program exceptions to the guest when we don't
>>> know the instruction we're supposed to emulate.
>>> 
>> 
>> Applied both, and queued for .33.  I was able to guess this from the description, but please mention it explicitly in the future.
>> 
> 
> I get this building ppc32:

PPC32? That shouldn't even allow enabling KVM O_o.

Alex

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Avi Kivity]

On 01/10/2010 05:06 PM, Alexander Graf wrote:
>>
>> I get this building ppc32:
>>      
> PPC32? That shouldn't even allow enabling KVM O_o.
>
>    

You know, the embedded stuff.

Enabled config options:

VIRTUALIZATION=y
KVM=y
KVM_440=y
KVM_E500=n
KVM_EXIT_TIMING=n

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

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Alexander Graf]

On 10.01.2010, at 16:09, Avi Kivity wrote:

> On 01/10/2010 05:06 PM, Alexander Graf wrote:
>>> 
>>> I get this building ppc32:
>>>     
>> PPC32? That shouldn't even allow enabling KVM O_o.
>> 
>>   
> 
> You know, the embedded stuff.

Hm, that patch is based on the big patchset I sent earlier. Are you sure you have applied them in the correct order?

Alex

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Avi Kivity]

On 01/10/2010 05:28 PM, Alexander Graf wrote:
>
> Hm, that patch is based on the big patchset I sent earlier. Are you sure you have applied them in the correct order?
>    

Probably not.  Will reorder and retry.

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

Re: [PATCH 1/2] KVM: PPC: Pass through program interrupts

[Avi Kivity]

On 01/10/2010 05:37 PM, Avi Kivity wrote:
> On 01/10/2010 05:28 PM, Alexander Graf wrote:
>>
>> Hm, that patch is based on the big patchset I sent earlier. Are you 
>> sure you have applied them in the correct order?
>
> Probably not.  Will reorder and retry.
>

No, still fails.  Please take a look at kvm-updates/2.6.34 commit 
7f31fad9aa, subject to git.kernel.org rsync delay (was doing 
bisectability testing).

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