linux-next: manual merge of the kvm tree with Linus' and the tip trees

linux-next: manual merge of the kvm tree with Linus' and the tip trees

[Stephen Rothwell]

Hi all,

Today's linux-next merge of the kvm tree got a conflict in:

  arch/x86/include/asm/cpufeatures.h

between commit:

  a89f040fa34e ("x86/cpufeatures: Add X86_BUG_CPU_INSECURE")

from Linus' tree, commit:

  76b043848fd2 ("x86/retpoline: Add initial retpoline support")

from the tip tree and commit:

  18c71ce9c882 ("x86/CPU/AMD: Add the Secure Encrypted Virtualization CPU feature")

from the kvm tree.

I fixed it up (see below) and can carry the fix as necessary. This
is now fixed as far as linux-next is concerned, but any non trivial
conflicts should be mentioned to your upstream maintainer when your tree
is submitted for merging.  You may also want to consider cooperating
with the maintainer of the conflicting tree to minimise any particularly
complex conflicts.

-- 
Cheers,
Stephen Rothwell

diff --cc arch/x86/include/asm/cpufeatures.h
index f275447862f4,13143dd8bdc2..000000000000
--- a/arch/x86/include/asm/cpufeatures.h
+++ b/arch/x86/include/asm/cpufeatures.h
@@@ -202,11 -201,9 +202,11 @@@
  #define X86_FEATURE_HW_PSTATE		( 7*32+ 8) /* AMD HW-PState */
  #define X86_FEATURE_PROC_FEEDBACK	( 7*32+ 9) /* AMD ProcFeedbackInterface */
  #define X86_FEATURE_SME			( 7*32+10) /* AMD Secure Memory Encryption */
 -#define X86_FEATURE_SEV			( 7*32+11) /* AMD Secure Encrypted Virtualization */
 -
 +#define X86_FEATURE_PTI			( 7*32+11) /* Kernel Page Table Isolation enabled */
 +#define X86_FEATURE_RETPOLINE		( 7*32+12) /* Generic Retpoline mitigation for Spectre variant 2 */
 +#define X86_FEATURE_RETPOLINE_AMD	( 7*32+13) /* AMD Retpoline mitigation for Spectre variant 2 */
  #define X86_FEATURE_INTEL_PPIN		( 7*32+14) /* Intel Processor Inventory Number */
- #define X86_FEATURE_INTEL_PT		( 7*32+15) /* Intel Processor Trace */
++#define X86_FEATURE_SEV			( 7*32+15) /* AMD Secure Encrypted Virtualization */
  #define X86_FEATURE_AVX512_4VNNIW	( 7*32+16) /* AVX-512 Neural Network Instructions */
  #define X86_FEATURE_AVX512_4FMAPS	( 7*32+17) /* AVX-512 Multiply Accumulation Single precision */
  

Re: linux-next: manual merge of the kvm tree with Linus' and the tip trees

[Thomas Gleixner]

On Mon, 15 Jan 2018, Stephen Rothwell wrote:
> Hi all,
> --- a/arch/x86/include/asm/cpufeatures.h
> +++ b/arch/x86/include/asm/cpufeatures.h
> @@@ -202,11 -201,9 +202,11 @@@
>   #define X86_FEATURE_HW_PSTATE		( 7*32+ 8) /* AMD HW-PState */
>   #define X86_FEATURE_PROC_FEEDBACK	( 7*32+ 9) /* AMD ProcFeedbackInterface */
>   #define X86_FEATURE_SME			( 7*32+10) /* AMD Secure Memory Encryption */
>  -#define X86_FEATURE_SEV			( 7*32+11) /* AMD Secure Encrypted Virtualization */
>  -
>  +#define X86_FEATURE_PTI			( 7*32+11) /* Kernel Page Table Isolation enabled */
>  +#define X86_FEATURE_RETPOLINE		( 7*32+12) /* Generic Retpoline mitigation for Spectre variant 2 */
>  +#define X86_FEATURE_RETPOLINE_AMD	( 7*32+13) /* AMD Retpoline mitigation for Spectre variant 2 */
>   #define X86_FEATURE_INTEL_PPIN		( 7*32+14) /* Intel Processor Inventory Number */
> - #define X86_FEATURE_INTEL_PT		( 7*32+15) /* Intel Processor Trace */

Where is that patch which moves INTEL_PT ?

commit 632218d8dee296c8b397c13b7b3b4be090a1c043
Author: Paolo Bonzini <pbonzini@redhat.com>
Date:   Mon Dec 11 04:30:50 2017 +0800

    x86: cpufeature: move processor tracing out of scattered features
    
    Processor tracing is already enumerated in word 9 (CPUID[7,0].EBX),
    so do not duplicate it in the scattered features word.
    
    Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Can KVM folks please stop doing random changes to the cpufeatures code
without talking to x86 maintainers and Borislav?

This wants to go through TIP or at least reviewed and acked.

Thanks,

	tglx

Re: linux-next: manual merge of the kvm tree with Linus' and the tip trees

[Thomas Gleixner]

On Mon, 15 Jan 2018, Thomas Gleixner wrote:
> On Mon, 15 Jan 2018, Stephen Rothwell wrote:
> > Hi all,
> > --- a/arch/x86/include/asm/cpufeatures.h
> > +++ b/arch/x86/include/asm/cpufeatures.h
> > @@@ -202,11 -201,9 +202,11 @@@
> >   #define X86_FEATURE_HW_PSTATE		( 7*32+ 8) /* AMD HW-PState */
> >   #define X86_FEATURE_PROC_FEEDBACK	( 7*32+ 9) /* AMD ProcFeedbackInterface */
> >   #define X86_FEATURE_SME			( 7*32+10) /* AMD Secure Memory Encryption */
> >  -#define X86_FEATURE_SEV			( 7*32+11) /* AMD Secure Encrypted Virtualization */
> >  -
> >  +#define X86_FEATURE_PTI			( 7*32+11) /* Kernel Page Table Isolation enabled */
> >  +#define X86_FEATURE_RETPOLINE		( 7*32+12) /* Generic Retpoline mitigation for Spectre variant 2 */
> >  +#define X86_FEATURE_RETPOLINE_AMD	( 7*32+13) /* AMD Retpoline mitigation for Spectre variant 2 */
> >   #define X86_FEATURE_INTEL_PPIN		( 7*32+14) /* Intel Processor Inventory Number */
> > - #define X86_FEATURE_INTEL_PT		( 7*32+15) /* Intel Processor Trace */
> 
> Where is that patch which moves INTEL_PT ?
> 
> commit 632218d8dee296c8b397c13b7b3b4be090a1c043
> Author: Paolo Bonzini <pbonzini@redhat.com>
> Date:   Mon Dec 11 04:30:50 2017 +0800
> 
>     x86: cpufeature: move processor tracing out of scattered features
>     
>     Processor tracing is already enumerated in word 9 (CPUID[7,0].EBX),
>     so do not duplicate it in the scattered features word.
>     
>     Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> 
> Can KVM folks please stop doing random changes to the cpufeatures code
> without talking to x86 maintainers and Borislav?
> 
> This wants to go through TIP or at least reviewed and acked.

In fact it needs to go through TIP. We spent a lot of effort to make the
backporting of all this mess simple and this is just shooting a hole in it.

Please drop that change and we sort something out how it can be done proper.

Dammit, we have a well established process for stuff like that.

Thanks,

	tglx

Re: linux-next: manual merge of the kvm tree with Linus' and the tip trees

[Paolo Bonzini]

On 15/01/2018 19:36, Thomas Gleixner wrote:
>> Can KVM folks please stop doing random changes to the cpufeatures code
>> without talking to x86 maintainers and Borislav?
>>
>> This wants to go through TIP or at least reviewed and acked.
> In fact it needs to go through TIP. We spent a lot of effort to make the
> backporting of all this mess simple and this is just shooting a hole in it.

I do understand why you want this to go through TIP, but I'm not sure
why a change to Processor Tracing is related to PTI or retpolines.  I'm
also not sure why it is a problem for backportability, since we always
try to send pull requests after TIP.  Is it because 7*32+15 will be free
in 4.16 but not earlier?

FWIW, no changes for IBRS or RSB stuffing are going through the KVM tree
before the bare metal parts are there.  I posted those mostly for people
(mostly the cloud providers that had been left in the dark) who wanted
something to apply quickly and didn't care about bare metal protection
because they could assume that any attack path passed through a vmexit.

Paolo

> Please drop that change and we sort something out how it can be done proper.
> 
> Dammit, we have a well established process for stuff like that.

Re: linux-next: manual merge of the kvm tree with Linus' and the tip trees

[Ingo Molnar]

* Paolo Bonzini <pbonzini@redhat.com> wrote:

> On 15/01/2018 19:36, Thomas Gleixner wrote:
> >> Can KVM folks please stop doing random changes to the cpufeatures code
> >> without talking to x86 maintainers and Borislav?
> >>
> >> This wants to go through TIP or at least reviewed and acked.
> > In fact it needs to go through TIP. We spent a lot of effort to make the
> > backporting of all this mess simple and this is just shooting a hole in it.
> 
> I do understand why you want this to go through TIP, but I'm not sure
> why a change to Processor Tracing is related to PTI or retpolines.  I'm
> also not sure why it is a problem for backportability, since we always
> try to send pull requests after TIP.  Is it because 7*32+15 will be free
> in 4.16 but not earlier?

It is because certain central x86 changes (such as changes to processor flags)
are kept on a v4.14 base to keep the PTI backporting efforts manageable.

Please revert (or rebase) this change from the KVM tree, and submit it separately, 
as it should have been done to begin with. Please also follow this process in the 
future: all x86 changes outside arch/x86/kvm/ need an explicit ack from an x86 
maintainer.

Thanks,

	Ingo

Re: linux-next: manual merge of the kvm tree with Linus' and the tip trees

[Paolo Bonzini]

On 16/01/2018 01:55, Ingo Molnar wrote:
> 
> * Paolo Bonzini <pbonzini@redhat.com> wrote:
> 
>> On 15/01/2018 19:36, Thomas Gleixner wrote:
>>>> Can KVM folks please stop doing random changes to the cpufeatures code
>>>> without talking to x86 maintainers and Borislav?
>>>>
>>>> This wants to go through TIP or at least reviewed and acked.
>>> In fact it needs to go through TIP. We spent a lot of effort to make the
>>> backporting of all this mess simple and this is just shooting a hole in it.
>>
>> I do understand why you want this to go through TIP, but I'm not sure
>> why a change to Processor Tracing is related to PTI or retpolines.  I'm
>> also not sure why it is a problem for backportability, since we always
>> try to send pull requests after TIP.  Is it because 7*32+15 will be free
>> in 4.16 but not earlier?
> 
> It is because certain central x86 changes (such as changes to processor flags)
> are kept on a v4.14 base to keep the PTI backporting efforts manageable.
> 
> Please revert (or rebase) this change from the KVM tree, and submit it separately, 
> as it should have been done to begin with. Please also follow this process in the 
> future: all x86 changes outside arch/x86/kvm/ need an explicit ack from an x86 
> maintainer.

I've always done it like that until
https://marc.info/?l=kvm&m=149335647027790 got no response for three
months, then I thought you didn't care.

We will drop Intel PT support and delay it to 4.17.  Luwei, since your
patches have issues with incorrect use of the MSR bitmap, this is
probably a good thing anyway (better bisectability).  Please repost your
patches at the end of the merge window, then we will wait for an ack
from Thomas/Ingo/Peter.

(Radim, just tell me if you prefer to leave the git tree surgery to me.
  Rebasing on top of commit 5cb0944c0c66004c0d9006a7f0fba5782ae38f69
should be okay for other arches).

Paolo

Re: linux-next: manual merge of the kvm tree with Linus' and the tip trees

[Thomas Gleixner]

On Tue, 16 Jan 2018, Paolo Bonzini wrote:

> On 16/01/2018 01:55, Ingo Molnar wrote:
> > 
> > * Paolo Bonzini <pbonzini@redhat.com> wrote:
> > 
> >> On 15/01/2018 19:36, Thomas Gleixner wrote:
> >>>> Can KVM folks please stop doing random changes to the cpufeatures code
> >>>> without talking to x86 maintainers and Borislav?
> >>>>
> >>>> This wants to go through TIP or at least reviewed and acked.
> >>> In fact it needs to go through TIP. We spent a lot of effort to make the
> >>> backporting of all this mess simple and this is just shooting a hole in it.
> >>
> >> I do understand why you want this to go through TIP, but I'm not sure
> >> why a change to Processor Tracing is related to PTI or retpolines.  I'm
> >> also not sure why it is a problem for backportability, since we always
> >> try to send pull requests after TIP.  Is it because 7*32+15 will be free
> >> in 4.16 but not earlier?
> > 
> > It is because certain central x86 changes (such as changes to processor flags)
> > are kept on a v4.14 base to keep the PTI backporting efforts manageable.
> > 
> > Please revert (or rebase) this change from the KVM tree, and submit it separately, 
> > as it should have been done to begin with. Please also follow this process in the 
> > future: all x86 changes outside arch/x86/kvm/ need an explicit ack from an x86 
> > maintainer.
> 
> I've always done it like that until
> https://marc.info/?l=kvm&m=149335647027790 got no response for three
> months, then I thought you didn't care.

Well, I certainly cared, but was kaisered enough to not look.

> We will drop Intel PT support and delay it to 4.17.  Luwei, since your
> patches have issues with incorrect use of the MSR bitmap, this is
> probably a good thing anyway (better bisectability).  Please repost your
> patches at the end of the merge window, then we will wait for an ack
> from Thomas/Ingo/Peter.

Can we get all cpu feature bit specific patches now please so we can move
them through TIP?

Thanks,

	tglx