NAK/cmnt: [Trusty SRU v2] Windows guest got 0x5c BSOD when rebooting

On 01.02.2017 14:01, Tim Gardner wrote:
> Patch 1 is actually a back port with relatively significant context
> differences. Always note that in the commit log along with the names
> of the files in conflict so that reviewers can focus their attention
> on specific code differences.
> 
> [PATCH 1/2] KVM: x86: reset RVI upon system reset
> [PATCH 2/2] KVM: x86: Emulator should set DR6 upon GD like real CPU
> 
> rtg
> 

Patch #1 probably could be near (or real) cherry-pick if picking

commit 963fee1656603ce2e91ebb988cd5a92f2af41369
Author: Wanpeng Li <wanpeng.li@linux.intel.com>
Date:   Thu Jul 17 19:03:00 2014 +0800

    KVM: nVMX: Fix virtual interrupt delivery injection

But reason for NAK is patch #2 v2 as it looks to be missing the definition of
DR6_RTM which is in the v1 of that.

iff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 70427d1..4d2cfd8 100644
-Stefan

On 14.02.2017 11:25, Stefan Bader wrote:
> On 01.02.2017 14:01, Tim Gardner wrote:
>> Patch 1 is actually a back port with relatively significant context
>> differences. Always note that in the commit log along with the names
>> of the files in conflict so that reviewers can focus their attention
>> on specific code differences.
>>
>> [PATCH 1/2] KVM: x86: reset RVI upon system reset
>> [PATCH 2/2] KVM: x86: Emulator should set DR6 upon GD like real CPU
>>
>> rtg
>>
> 
> Patch #1 probably could be near (or real) cherry-pick if picking
> 
> commit 963fee1656603ce2e91ebb988cd5a92f2af41369
> Author: Wanpeng Li <wanpeng.li@linux.intel.com>
> Date:   Thu Jul 17 19:03:00 2014 +0800
> 
>     KVM: nVMX: Fix virtual interrupt delivery injection
> 
> But reason for NAK is patch #2 v2 as it looks to be missing the definition of
> DR6_RTM which is in the v1 of that.

... which probably was due to incorrectly stating it to be a cherry-pick instead
of a backport...
> 
> iff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index 70427d1..4d2cfd8 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -157,6 +157,7 @@ enum {
> 
>  #define DR6_BD		(1 << 13)
>  #define DR6_BS		(1 << 14)
> +#define DR6_RTM         (1 << 16)
>  #define DR6_FIXED_1	0xffff0ff0
>  #define DR6_VOLATILE	0x0000e00f
> 
> -Stefan
> 
> 
> 
>

I dropped the use of DR6_RTM in the second patch. Since it was only ever used
in that specific patch and never defined elsewhere, its absense should
be benign. I actually test compiled this time.

[PATCH 1/2 V3] KVM: x86: reset RVI upon system reset
[PATCH 2/2 V3] KVM: x86: Emulator should set DR6 upon GD like real CPU

rtg

Applied to trusty master-next branch.

Thanks.
Cascardo.