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

Re: [Xen-devel] [PATCH v5 06/10] AMD/IOMMU: don't blindly allocate interrupt remapping tables


  • To: Jan Beulich <jbeulich@xxxxxxxx>
  • From: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
  • Date: Wed, 18 Sep 2019 12:36:41 +0100
  • Authentication-results: esa2.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: "xen-devel@xxxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxxx>, Suravee Suthikulpanit <suravee.suthikulpanit@xxxxxxx>
  • Delivery-date: Wed, 18 Sep 2019 11:36:48 +0000
  • Ironport-sdr: /wutAcBcwfVNmCTDVhp8bmCd5LB8eMVHhTeXGuVrJGk/mtj4nm2tcJBlZaO79ObXPdClWmOMH7 pxU/97Np15BHgzMOzpzLRijl/JQTAPG1YENgS95sF6/nA9qz7dW9GHd2Libpiz8j6wOCFP1yJc s0fWX8JjvxyZ35e0knUwgfWP7spFROQOVwm+ZfvohBHZ6TOwTt3Y96xynCzFWOSzZz1fmi1eEx bI/F8tU9MlBR2eia/Z8K5nO+xGdRzsB7Q/nup9tz6hTaZ6kKS0ULXYDOB/RTN8X7THASVex5fc 2cg=
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>
  • Openpgp: preference=signencrypt

On 18/09/2019 09:55, Jan Beulich wrote:
> On 17.09.2019 15:10, Andrew Cooper wrote:
>> On 06/08/2019 14:09, Jan Beulich wrote:
>>> ACPI tables are free to list far more device coordinates than there are
>>> actual devices. By delaying the table allocations for most cases, and
>>> doing them only when an actual device is known to be present at a given
>>> position, overall memory used for the tables goes down from over 500k
>>> pages to just over 1k ones.
>> This is slightly awkward grammar.  While I don't think it is strictly
>> speaking incorrect, it would be more normal to phrase as "just over 1k
>> pages".
> Changed, albeit to me the double "pages" sounds odd as well. Would
> "of them" be any better than "ones"?

You are drawing a comparison between 500k pages, and 1k pages.

Changing the 'pages' qualifier introduced ambiguity - consider the case
where "ones" is substituted for "bytes", which would be a legitimate way
to describe the memory reduction, but is misleading due to the
difference in units.

Alternatively, the trailing qualifier can be dropped, and the sentence
end at "1k.", as the "pages" is implied.

>
>>> ---
>>> TBD: This retains prior (but suspicious) behavior of not calling
>>>      amd_iommu_set_intremap_table() for "invalid" IVRS mapping entries.
>>>      Since DTE.IV=0 means un-remapped interrupts, I wonder if this needs
>>>      changing.
>> How would an invalid entry get DTE.V set in the first place?
> DTE.V gets set by amd_iommu_set_root_page_table(), which in turn gets
> called from amd_iommu_setup_domain_device() alone. It's only the
> latter function's callers which obtain (and possibly check) the
> corresponding IVRS mappings entry. Hence to me there's a sufficient
> disconnect between setting of DTE.V and DTE.IV.
>
> Plus, looking at e.g. amd_iommu_add_device(), there's ample room for
> not even making it to amd_iommu_set_intremap_table(), due to earlier
> errors.
>
>> Whatever the old behaviour may have been, we shouldn't have code which
>> comes with a chance of having non-remapped interrupts by accident.
> We can't make amd_iommu_set_root_page_table() set dte->iv to 1, as
> it gets called only after amd_iommu_set_intremap_table() in
> amd_iommu_add_device(). But we could of course make it do so when
> it finds dte->it_root be zero. Yet I wonder if it wasn't more safe
> to have DTEs start out with the field set this way.
>
> Along these lines I'm also not convinced it is a good idea for
> amd_iommu_set_intremap_table() to have a "valid" parameter in the
> first place: It's okay as long as all callers pass iommu_intremap,
> but it would seem to me that - as already said - we'd want DTEs be
> set this way right when a DT gets allocated. If you agree, I'll
> happily add a patch doing so to the end of this series (there's
> meanwhile a patch re-arranging DT allocation anyway).

I've been looking through the spec, and this is rather complicated.  We
need to set V and TV to inhibit DMA, and IV and IntCtl to inhibit
interrupts.

Why not initialise every entry in the device table when we create it to
a safe, blocked state.  That way, the only way a device starts
functioning appropriately is via a successful call to 
amd_iommu_set_root_page_table() and amd_iommu_set_intremap_table(),
which seems to be far safer behaviour overall.

~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®.