Re: [PATCH 0/7] kvmtool: Cleanup kernel loading

[Will Deacon]

On Wed, Nov 18, 2015 at 10:29:30AM +, Andre Przywara wrote:
> On 02/11/15 14:58, Will Deacon wrote:
> > On Fri, Oct 30, 2015 at 06:26:53PM +, Andre Przywara wrote:
> >> this series cleans up kvmtool's kernel loading functionality a bit.
> >> It has been broken out of a previous series I sent [1] and contains
> >> just the cleanup and bug fix parts, which should be less controversial
> >> and thus easier to merge ;-)
> >> I will resend the pipe loading part later on as a separate series.
> >>
> >> The first patch properly abstracts kernel loading to move
> >> responsibility into each architecture's code. It removes quite some
> >> ugly code from the generic kvm.c file.
> >> The later patches address the naive usage of read(2) to, well, read
> >> data from files. Doing this without coping with the subtleties of
> >> the UNIX read semantics (returning with less or none data read is not
> >> an error) can provoke hard to debug failures.
> >> So these patches make use of the existing and one new wrapper function
> >> to make sure we read everything we actually wanted to.
> >> The last patch moves the ARM kernel loading code into the proper
> >> location to be in line with the other architectures.
> >>
> >> Please have a look and give some comments!
> > 
> > Looks good to me, but I'd like to see some comments from some mips/ppc/x86
> > people on the changes you're making over there.
> 
> Sounds reasonable, but no answers yet.
> 
> Can you take at least patch 1 and 2 meanwhile, preferably 6 and 7 (the
> ARM parts) also if you are OK with it?
> I have other patches that depend on 1/7 and 2/7, so having them upstream
> would help me and reduce further dependency churn.
> I am happy to resend the remaining patches for further discussion later.

We let them sit on the list for a while with no comments, so I just pushed
out your series. If a bug report shows up, then we can always revert the
offending patch if there's no quick fix.

Will
--
To unsubscribe from this list: send the line "unsubscribe kvm-ppc" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] KVM: powerpc: kvmppc_visible_gpa can be boolean

[Paolo Bonzini]

On 16/11/2015 04:10, Yaowei Bai wrote:
> In another patch kvm_is_visible_gfn is maken return bool due to this
> function only returns zero or one as its return value, let's also make
> kvmppc_visible_gpa return bool to keep consistent.
> 
> No functional change.
> 
> Signed-off-by: Yaowei Bai 
> ---
>  arch/powerpc/kvm/book3s_pr.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/powerpc/kvm/book3s_pr.c b/arch/powerpc/kvm/book3s_pr.c
> index 64891b0..70fb08d 100644
> --- a/arch/powerpc/kvm/book3s_pr.c
> +++ b/arch/powerpc/kvm/book3s_pr.c
> @@ -512,7 +512,7 @@ static void kvmppc_patch_dcbz(struct kvm_vcpu *vcpu, 
> struct kvmppc_pte *pte)
>   put_page(hpage);
>  }
>  
> -static int kvmppc_visible_gpa(struct kvm_vcpu *vcpu, gpa_t gpa)
> +static bool kvmppc_visible_gpa(struct kvm_vcpu *vcpu, gpa_t gpa)
>  {
>   ulong mp_pa = vcpu->arch.magic_page_pa;
>  
> @@ -521,7 +521,7 @@ static int kvmppc_visible_gpa(struct kvm_vcpu *vcpu, 
> gpa_t gpa)
>  
>   gpa &= ~0xFFFULL;
>   if (unlikely(mp_pa) && unlikely((mp_pa & KVM_PAM) == (gpa & KVM_PAM))) {
> - return 1;
> + return true;
>   }
>  
>   return kvm_is_visible_gfn(vcpu->kvm, gpa >> PAGE_SHIFT);
> 

Applied all three patches to kvm/queue, thanks.

Paolo
--
To unsubscribe from this list: send the line "unsubscribe kvm-ppc" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 0/7] kvmtool: Cleanup kernel loading

[Andre Przywara]

Hi Will,

On 02/11/15 14:58, Will Deacon wrote:
> On Fri, Oct 30, 2015 at 06:26:53PM +, Andre Przywara wrote:
>> Hi,
> 
> Hello Andre,
> 
>> this series cleans up kvmtool's kernel loading functionality a bit.
>> It has been broken out of a previous series I sent [1] and contains
>> just the cleanup and bug fix parts, which should be less controversial
>> and thus easier to merge ;-)
>> I will resend the pipe loading part later on as a separate series.
>>
>> The first patch properly abstracts kernel loading to move
>> responsibility into each architecture's code. It removes quite some
>> ugly code from the generic kvm.c file.
>> The later patches address the naive usage of read(2) to, well, read
>> data from files. Doing this without coping with the subtleties of
>> the UNIX read semantics (returning with less or none data read is not
>> an error) can provoke hard to debug failures.
>> So these patches make use of the existing and one new wrapper function
>> to make sure we read everything we actually wanted to.
>> The last patch moves the ARM kernel loading code into the proper
>> location to be in line with the other architectures.
>>
>> Please have a look and give some comments!
> 
> Looks good to me, but I'd like to see some comments from some mips/ppc/x86
> people on the changes you're making over there.

Sounds reasonable, but no answers yet.

Can you take at least patch 1 and 2 meanwhile, preferably 6 and 7 (the
ARM parts) also if you are OK with it?
I have other patches that depend on 1/7 and 2/7, so having them upstream
would help me and reduce further dependency churn.
I am happy to resend the remaining patches for further discussion later.

Cheers,
Andre.
--
To unsubscribe from this list: send the line "unsubscribe kvm-ppc" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html