All of lore.kernel.org
 help / color / mirror / Atom feed
From: Robin Murphy <robin.murphy@arm.com>
To: Tomasz Figa <tfiga@chromium.org>
Cc: Yong Zhi <yong.zhi@intel.com>,
	linux-media@vger.kernel.org,
	Sakari Ailus <sakari.ailus@linux.intel.com>,
	"Zheng, Jian Xu" <jian.xu.zheng@intel.com>,
	"Mani, Rajmohan" <rajmohan.mani@intel.com>,
	"Toivonen, Tuukka" <tuukka.toivonen@intel.com>,
	Joerg Roedel <joro@8bytes.org>,
	"open list:IOMMU DRIVERS" <iommu@lists.linux-foundation.org>
Subject: Re: [PATCH 03/12] intel-ipu3: Add DMA API implementation
Date: Thu, 8 Jun 2017 19:07:09 +0100	[thread overview]
Message-ID: <949a467a-e711-d746-859d-fc006bf59773@arm.com> (raw)
In-Reply-To: <CAAFQd5AWGN_qSXGG32D-eWKZRoRme+XCD9v1r8qJ5bthtS9z9w@mail.gmail.com>

On 08/06/17 15:35, Tomasz Figa wrote:
> On Thu, Jun 8, 2017 at 10:22 PM, Robin Murphy <robin.murphy@arm.com> wrote:
>> On 07/06/17 10:47, Tomasz Figa wrote:
>>> Hi Yong,
>>>
>>> +Robin, Joerg, IOMMU ML
>>>
>>> Please see my comments inline.
>>>
>>> On Tue, Jun 6, 2017 at 5:39 AM, Yong Zhi <yong.zhi@intel.com> wrote:
> [snip]
>>>> +
>>>> +/* End of things adapted from arch/arm/mm/dma-mapping.c */
>>>> +static void ipu3_dmamap_sync_single_for_cpu(struct device *dev,
>>>> +                                           dma_addr_t dma_handle, size_t size,
>>>> +                                           enum dma_data_direction dir)
>>>> +{
>>>> +       struct ipu3_mmu *mmu = to_ipu3_mmu(dev);
>>>> +       dma_addr_t daddr = iommu_iova_to_phys(mmu->domain, dma_handle);
>>>> +
>>>> +       clflush_cache_range(phys_to_virt(daddr), size);
>>>
>>> You might need to consider another IOMMU on the way here. Generally,
>>> given that daddr is your MMU DMA address (not necessarily CPU physical
>>> address), you should be able to call
>>>
>>> dma_sync_single_for_cpu(<your pci device>, daddr, size, dir)
>>
>> I'd hope that this IPU complex is some kind of embedded endpoint thing
>> that bypasses the VT-d IOMMU or is always using its own local RAM,
>> because it would be pretty much unworkable otherwise.
> 
> It uses system RAM and, as far as my understanding goes, by default it
> operates without the VT-d IOMMU and that's how it's used right now.

OK, if it *is* behind a DMAR unit then booting with "iommu=force" (or
whatever the exact incantation for intel-iommu is) should be fun...

> I'm suggesting VT-d IOMMU as a way to further strengthen the security
> and error resilience in future (due to the IPU complex being
> non-coherent and also running a closed source firmware).

TBH, doing DMA remapping through *two* IOMMUS will add horrible hardware
overhead, increase the scope for kernel-side bugs, and not much more. If
we don't trust this IOMMU to behave, why are we trying to drive it in
the first place? If we do, then a second IOMMU behind it won't protect
anything that the first one doesn't already.

>> The whole
>> infrastructure isn't really capable of dealing with nested IOMMUs, and
>> nested DMA ops would be an equally horrible idea.
> 
> Could you elaborate a bit more on this? I think we should be able to
> deal with this in a way I suggested before:
> 
> a) the PCI device would use the system DMA ops,
> b) the PCI device would implement a secondary bus for which it would
> provide its own DMA and IOMMU ops.
> c) a secondary device would be registered on the secondary bus,
> d) all memory for the IPU would be managed on behalf of the secondary device.
> 
> In fact, the driver already is designed in a way that all the points
> above are true. If I'm not missing something, the only significant
> missing point is calling into system DMA ops from IPU DMA ops.

I don't believe x86 has any non-coherent DMA ops, therefore the IPU DMA
ops would still probably have to do all their own cache maintenance.
Allocation/mapping, though, would have to be done with the parent DMA
ops first (in case DMA address != physical address), *then* mapped at
the IPU MMU, which is the real killer - if the PCI DMA ops are from
intel-iommu, then there's little need for the IPU MMU mapping to be
anything other than 1:1, so you may as well not bother. If the PCI DMA
ops are from SWIOTLB, then the constraints of having to go through that
first eliminate all the scatter-gather benefit of the IPU MMU.

The IOMMU API ops would have to be handled similarly, by checking for
ops on the parent bus, calling those first if present, then running the
intermediate results through the IPU MMU's own functions. Sure, it's not
impossible, but it's really really grim. Not to mention that all the IPU
MMU's page tables/control structures/etc. would also have to be
DMA-allocated/mapped because it may or may not be operating in physical
address space.

The reasonable option - assuming the topology really is this way - would
seem to be special-casing the IPU in intel-iommu in a similar manner to
integrated graphics, to make sure it gets a passthrough domain for DMA
ops, but still allowing the whole PCI device to be passed through to a
guest VM via VFIO if desired (which is really the only case where nested
translation does start to make sense).

Robin.

> 
> Best regards,
> Tomasz
> 

  reply	other threads:[~2017-06-08 18:07 UTC|newest]

Thread overview: 72+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-06-05 20:39 [PATCH 00/12] Intel IPU3 ImgU patchset Yong Zhi
2017-06-05 20:39 ` [PATCH 01/12] videodev2.h, v4l2-ioctl: add IPU3 meta buffer format Yong Zhi
2017-06-05 20:43   ` Alan Cox
2017-06-09 10:55     ` Sakari Ailus
2017-06-06  4:30   ` Tomasz Figa
2017-06-06  4:30     ` Tomasz Figa
2017-06-06  7:25       ` Sakari Ailus
2017-06-06  8:04         ` Hans Verkuil
2017-06-06 10:09           ` Tomasz Figa
2017-06-16  5:52             ` Tomasz Figa
2017-06-16  8:25               ` Sakari Ailus
2017-06-16  8:35                 ` Tomasz Figa
2017-06-16  8:49                   ` Sakari Ailus
2017-06-16  9:03                     ` Tomasz Figa
2017-06-16  9:19                       ` Sakari Ailus
2017-06-16  9:29                         ` Tomasz Figa
2017-06-19  9:17                   ` Laurent Pinchart
2017-06-19 10:41                     ` Tomasz Figa
2017-06-05 20:39 ` [PATCH 02/12] intel-ipu3: mmu: implement driver Yong Zhi
2017-06-06  8:07   ` Hans Verkuil
2017-06-16  9:21     ` Sakari Ailus
2017-06-06 10:13   ` Tomasz Figa
2017-06-07  8:35     ` Tomasz Figa
2017-06-07  8:35       ` Tomasz Figa
2017-06-08 16:43       ` Sakari Ailus
2017-06-09  5:59         ` Tomasz Figa
2017-06-09 11:16           ` Sakari Ailus
2017-06-09 12:09             ` Tomasz Figa
2017-06-09  8:26       ` Tuukka Toivonen
2017-06-09 12:10         ` Tomasz Figa
2017-06-07 21:59     ` Sakari Ailus
2017-06-07 21:59       ` Sakari Ailus
2017-06-08  7:36       ` Tomasz Figa
2017-06-05 20:39 ` [PATCH 03/12] intel-ipu3: Add DMA API implementation Yong Zhi
2017-06-07  9:47   ` Tomasz Figa
2017-06-07  9:47     ` Tomasz Figa
2017-06-07 17:45     ` Alan Cox
2017-06-08  2:55       ` Tomasz Figa
2017-06-08  2:55         ` Tomasz Figa
2017-06-08 16:47         ` Sakari Ailus
2017-06-08 13:22     ` Robin Murphy
2017-06-08 14:35       ` Tomasz Figa
2017-06-08 14:35         ` Tomasz Figa
2017-06-08 18:07         ` Robin Murphy [this message]
2017-06-09  6:20           ` Tomasz Figa
2017-06-09  6:20             ` Tomasz Figa
2017-06-09 13:05             ` Robin Murphy
2017-06-05 20:39 ` [PATCH 04/12] intel-ipu3: Add user space ABI definitions Yong Zhi
2017-06-06  8:28   ` Hans Verkuil
2017-06-07 22:22     ` Sakari Ailus
2017-09-05 17:31       ` Zhi, Yong
2018-04-27 12:27       ` Sakari Ailus
2017-06-05 20:39 ` [PATCH 06/12] intel-ipu3: css: imgu dma buff pool Yong Zhi
2017-06-05 20:39 ` [PATCH 07/12] intel-ipu3: css: firmware management Yong Zhi
2017-06-06  8:38   ` Hans Verkuil
2017-06-14 21:46     ` Zhi, Yong
2017-06-16 10:15   ` Tomasz Figa
2017-06-05 20:39 ` [PATCH 08/12] intel-ipu3: params: compute and program ccs Yong Zhi
2017-06-05 20:39 ` [PATCH 09/12] intel-ipu3: css hardware setup Yong Zhi
2017-06-05 20:39 ` [PATCH 10/12] intel-ipu3: css pipeline Yong Zhi
2017-06-05 20:39 ` [PATCH 11/12] intel-ipu3: Add imgu v4l2 driver Yong Zhi
2017-06-06  9:08   ` Hans Verkuil
2017-06-09  9:20     ` Sakari Ailus
2017-06-14 23:40       ` Zhi, Yong
2017-06-05 20:39 ` [PATCH 12/12] intel-ipu3: imgu top level pci device Yong Zhi
2017-06-05 20:46 ` [PATCH 00/12] Intel IPU3 ImgU patchset Alan Cox
2017-06-14 22:26   ` Sakari Ailus
2017-06-15  8:26     ` Andy Shevchenko
2017-06-15  8:37       ` Sakari Ailus
2017-06-06  9:14 ` Hans Verkuil
     [not found] ` <1496695157-19926-6-git-send-email-yong.zhi@intel.com>
2017-06-08  8:29   ` [PATCH 05/12] intel-ipu3: css: tables Tomasz Figa
2017-06-09  9:43     ` Sakari Ailus

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=949a467a-e711-d746-859d-fc006bf59773@arm.com \
    --to=robin.murphy@arm.com \
    --cc=iommu@lists.linux-foundation.org \
    --cc=jian.xu.zheng@intel.com \
    --cc=joro@8bytes.org \
    --cc=linux-media@vger.kernel.org \
    --cc=rajmohan.mani@intel.com \
    --cc=sakari.ailus@linux.intel.com \
    --cc=tfiga@chromium.org \
    --cc=tuukka.toivonen@intel.com \
    --cc=yong.zhi@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.