[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 10/12] docs/migration: Specify X86_{CPUID, MSR}_POLICY records


  • To: Jan Beulich <jbeulich@xxxxxxxx>
  • From: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
  • Date: Thu, 9 Jan 2020 15:30:19 +0000
  • Authentication-results: esa3.hc3370-68.iphmx.com; dkim=none (message not signed) header.i=none; spf=None smtp.pra=andrew.cooper3@xxxxxxxxxx; spf=Pass smtp.mailfrom=Andrew.Cooper3@xxxxxxxxxx; spf=None smtp.helo=postmaster@xxxxxxxxxxxxxxx
  • Autocrypt: addr=andrew.cooper3@xxxxxxxxxx; prefer-encrypt=mutual; keydata= mQINBFLhNn8BEADVhE+Hb8i0GV6mihnnr/uiQQdPF8kUoFzCOPXkf7jQ5sLYeJa0cQi6Penp VtiFYznTairnVsN5J+ujSTIb+OlMSJUWV4opS7WVNnxHbFTPYZVQ3erv7NKc2iVizCRZ2Kxn srM1oPXWRic8BIAdYOKOloF2300SL/bIpeD+x7h3w9B/qez7nOin5NzkxgFoaUeIal12pXSR Q354FKFoy6Vh96gc4VRqte3jw8mPuJQpfws+Pb+swvSf/i1q1+1I4jsRQQh2m6OTADHIqg2E ofTYAEh7R5HfPx0EXoEDMdRjOeKn8+vvkAwhviWXTHlG3R1QkbE5M/oywnZ83udJmi+lxjJ5 YhQ5IzomvJ16H0Bq+TLyVLO/VRksp1VR9HxCzItLNCS8PdpYYz5TC204ViycobYU65WMpzWe LFAGn8jSS25XIpqv0Y9k87dLbctKKA14Ifw2kq5OIVu2FuX+3i446JOa2vpCI9GcjCzi3oHV e00bzYiHMIl0FICrNJU0Kjho8pdo0m2uxkn6SYEpogAy9pnatUlO+erL4LqFUO7GXSdBRbw5 gNt25XTLdSFuZtMxkY3tq8MFss5QnjhehCVPEpE6y9ZjI4XB8ad1G4oBHVGK5LMsvg22PfMJ ISWFSHoF/B5+lHkCKWkFxZ0gZn33ju5n6/FOdEx4B8cMJt+cWwARAQABtClBbmRyZXcgQ29v cGVyIDxhbmRyZXcuY29vcGVyM0BjaXRyaXguY29tPokCOgQTAQgAJAIbAwULCQgHAwUVCgkI CwUWAgMBAAIeAQIXgAUCWKD95wIZAQAKCRBlw/kGpdefoHbdD/9AIoR3k6fKl+RFiFpyAhvO 59ttDFI7nIAnlYngev2XUR3acFElJATHSDO0ju+hqWqAb8kVijXLops0gOfqt3VPZq9cuHlh IMDquatGLzAadfFx2eQYIYT+FYuMoPZy/aTUazmJIDVxP7L383grjIkn+7tAv+qeDfE+txL4 SAm1UHNvmdfgL2/lcmL3xRh7sub3nJilM93RWX1Pe5LBSDXO45uzCGEdst6uSlzYR/MEr+5Z JQQ32JV64zwvf/aKaagSQSQMYNX9JFgfZ3TKWC1KJQbX5ssoX/5hNLqxMcZV3TN7kU8I3kjK mPec9+1nECOjjJSO/h4P0sBZyIUGfguwzhEeGf4sMCuSEM4xjCnwiBwftR17sr0spYcOpqET ZGcAmyYcNjy6CYadNCnfR40vhhWuCfNCBzWnUW0lFoo12wb0YnzoOLjvfD6OL3JjIUJNOmJy RCsJ5IA/Iz33RhSVRmROu+TztwuThClw63g7+hoyewv7BemKyuU6FTVhjjW+XUWmS/FzknSi dAG+insr0746cTPpSkGl3KAXeWDGJzve7/SBBfyznWCMGaf8E2P1oOdIZRxHgWj0zNr1+ooF /PzgLPiCI4OMUttTlEKChgbUTQ+5o0P080JojqfXwbPAyumbaYcQNiH1/xYbJdOFSiBv9rpt TQTBLzDKXok86LkCDQRS4TZ/ARAAkgqudHsp+hd82UVkvgnlqZjzz2vyrYfz7bkPtXaGb9H4 Rfo7mQsEQavEBdWWjbga6eMnDqtu+FC+qeTGYebToxEyp2lKDSoAsvt8w82tIlP/EbmRbDVn 7bhjBlfRcFjVYw8uVDPptT0TV47vpoCVkTwcyb6OltJrvg/QzV9f07DJswuda1JH3/qvYu0p vjPnYvCq4NsqY2XSdAJ02HrdYPFtNyPEntu1n1KK+gJrstjtw7KsZ4ygXYrsm/oCBiVW/OgU g/XIlGErkrxe4vQvJyVwg6YH653YTX5hLLUEL1NS4TCo47RP+wi6y+TnuAL36UtK/uFyEuPy wwrDVcC4cIFhYSfsO0BumEI65yu7a8aHbGfq2lW251UcoU48Z27ZUUZd2Dr6O/n8poQHbaTd 6bJJSjzGGHZVbRP9UQ3lkmkmc0+XCHmj5WhwNNYjgbbmML7y0fsJT5RgvefAIFfHBg7fTY/i kBEimoUsTEQz+N4hbKwo1hULfVxDJStE4sbPhjbsPCrlXf6W9CxSyQ0qmZ2bXsLQYRj2xqd1 bpA+1o1j2N4/au1R/uSiUFjewJdT/LX1EklKDcQwpk06Af/N7VZtSfEJeRV04unbsKVXWZAk uAJyDDKN99ziC0Wz5kcPyVD1HNf8bgaqGDzrv3TfYjwqayRFcMf7xJaL9xXedMcAEQEAAYkC HwQYAQgACQUCUuE2fwIbDAAKCRBlw/kGpdefoG4XEACD1Qf/er8EA7g23HMxYWd3FXHThrVQ HgiGdk5Yh632vjOm9L4sd/GCEACVQKjsu98e8o3ysitFlznEns5EAAXEbITrgKWXDDUWGYxd pnjj2u+GkVdsOAGk0kxczX6s+VRBhpbBI2PWnOsRJgU2n10PZ3mZD4Xu9kU2IXYmuW+e5KCA vTArRUdCrAtIa1k01sPipPPw6dfxx2e5asy21YOytzxuWFfJTGnVxZZSCyLUO83sh6OZhJkk b9rxL9wPmpN/t2IPaEKoAc0FTQZS36wAMOXkBh24PQ9gaLJvfPKpNzGD8XWR5HHF0NLIJhgg 4ZlEXQ2fVp3XrtocHqhu4UZR4koCijgB8sB7Tb0GCpwK+C4UePdFLfhKyRdSXuvY3AHJd4CP 4JzW0Bzq/WXY3XMOzUTYApGQpnUpdOmuQSfpV9MQO+/jo7r6yPbxT7CwRS5dcQPzUiuHLK9i nvjREdh84qycnx0/6dDroYhp0DFv4udxuAvt1h4wGwTPRQZerSm4xaYegEFusyhbZrI0U9tJ B8WrhBLXDiYlyJT6zOV2yZFuW47VrLsjYnHwn27hmxTC/7tvG3euCklmkn9Sl9IAKFu29RSo d5bD8kMSCYsTqtTfT6W4A3qHGvIDta3ptLYpIAOD2sY3GYq2nf3Bbzx81wZK14JdDDHUX2Rs 6+ahAA==
  • Cc: Stefano Stabellini <sstabellini@xxxxxxxxxx>, Wei Liu <wl@xxxxxxx>, Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>, George Dunlap <George.Dunlap@xxxxxxxxxxxxx>, Tim Deegan <tim@xxxxxxx>, Marek Marczykowski-Górecki <marmarek@xxxxxxxxxxxxxxxxxxxxxx>, Julien Grall <julien.grall@xxxxxxx>, Ian Jackson <ian.jackson@xxxxxxxxxx>, Xen-devel <xen-devel@xxxxxxxxxxxxxxxxxxxx>
  • Delivery-date: Thu, 09 Jan 2020 15:30:26 +0000
  • Ironport-sdr: neq3m4V5UNUIx2XU9M5XO9bbatP8bNkVJ5VjSJ0ZimTev7eSSHDRv9c9+H/82huTrxm7CUz65L ZHy7WHFyiHzIXQLDFidYyyBlmNLEvfgzR2fRsBuJDm7d0PVU65rMBEqY1qeW7aIUtmV5bzIe02 hFvCWLxAJ/eIwZLxk8+YWR9jIOkkZrFCKI5+qd/u/esZyCXg6/5avMjczYCDLCu3X2aBnJtl7g 7pbNDTKIIBv/yEpm+1jjACeOKbnOZwFGbkMqazwLvGiKK/QdVWvAKnJ8UXUf2yHEw7z2snBuqq Bh4=
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>
  • Openpgp: preference=signencrypt

On 03/01/2020 15:30, Jan Beulich wrote:
> On 03.01.2020 15:55, Andrew Cooper wrote:
>> On 03/01/2020 14:49, Jan Beulich wrote:
>>> On 24.12.2019 16:19, Andrew Cooper wrote:
>>>> @@ -439,6 +449,34 @@ def verify_record_static_data_end(self, content):
>>>>              raise RecordError("Static data end record found in v2 stream")
>>>>  
>>>>  
>>>> +    def verify_record_x86_cpuid_policy(self, content):
>>>> +        """ x86 CPUID policy record """
>>>> +
>>>> +        if self.version < 3:
>>>> +            raise RecordError("x86 CPUID policy record found in v2 
>>>> stream")
>>>> +
>>>> +        sz = calcsize(X86_CPUID_POLICY_FORMAT)
>>>> +        contentsz = len(content)
>>>> +
>>>> +        if contentsz < sz or (contentsz % sz) != 0:
>>>> +            raise RecordError("Record length %u, expected multiple of %u" 
>>>> %
>>>> +                              (contentsz, sz))
>>>> +
>>>> +
>>>> +    def verify_record_x86_msr_policy(self, content):
>>>> +        """ x86 MSR policy record """
>>>> +
>>>> +        if self.version < 3:
>>>> +            raise RecordError("x86 MSR policy record found in v2 stream")
>>>> +
>>>> +        sz = calcsize(X86_MSR_POLICY_FORMAT)
>>>> +        contentsz = len(content)
>>>> +
>>>> +        if contentsz < sz or (contentsz % sz) != 0:
>>>> +            raise RecordError("Record length %u, expected multiple of %u" 
>>>> %
>>>> +                              (contentsz, sz))
>>> While I can't even see a theoretical case of the CPUID array
>>> having zero elements, is it really entirely implausible to have
>>> an empty MSRs array? I.e. wouldn't the left side of the "or"
>>> better go away?
>> MSRs will never have 0 entries, because unlike CPUID, we can't omit
>> records with 0s as their content.  This becomes ambiguous when the
>> policy default is nonzero.
> Isn't the same true for CPUID, in particular some of the non-boolean
> fields?

I perhaps misspoke.  We can omit CPUID leave(s) based on information
already sent (max_leaf and/or the absence of an enumerating feature). 
In these cases, the destination side can still fill the remainder in
with 0's.

Top level leaves without an encompassing feature do need sending in full.

The best practical example is for not sending all 63 subleaves of the
xsave state leaf, when most of them are outside of the policy XCR0|XSS bits.

>
>> When we do gain more MSRs, I will see about organising elision based on
>> CPUID features, so we don't have to send a 0 for every single MSR in the
>> policy, but MSRs without CPUID enumeration must always be sent.
>>
>> This means that the one MSR we have currently (MSR_INTEL_PLATFORM_INFO
>> for CPUID Faulting, which we also virtualise on AMD hardware) shall
>> unconditionally be present forever more.
> Hmm, yes. Still the special casing of there needing to be at least
> one entry looks a little odd here (and also for CPUID). I would
> find it more logical if there was just the remainder-must-be-zero
> check. But this is libxc code, so I'm not the one to really judge
> anyway.

The migration stream is split into records with no playload (markers
with external control flow meaning), and data records, which have a payload.

It is an error for a data record to have no payload, because it means
there is a source side generation bug.  In the case of Xen returning 0
MSRs, the record would be omitted entirely, rather than be sent with 0
MSRs worth of data.

~Andrew

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.