From: Dave Jiang <dave.jiang@intel.com>
To: Jason Gunthorpe <jgg@nvidia.com>
Cc: alex.williamson@redhat.com, kwankhede@nvidia.com,
tglx@linutronix.de, vkoul@kernel.org, megha.dey@intel.com,
jacob.jun.pan@intel.com, ashok.raj@intel.com, yi.l.liu@intel.com,
baolu.lu@intel.com, kevin.tian@intel.com,
sanjay.k.kumar@intel.com, tony.luck@intel.com,
dan.j.williams@intel.com, eric.auger@redhat.com,
pbonzini@redhat.com, dmaengine@vger.kernel.org,
linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Subject: Re: [PATCH v6 15/20] vfio/mdev: idxd: ims domain setup for the vdcm
Date: Wed, 26 May 2021 18:15:17 -0700 [thread overview]
Message-ID: <f5606802-2ff3-2f54-0ff1-c1f1dd59f52c@intel.com> (raw)
In-Reply-To: <20210527005444.GV1002214@nvidia.com>
On 5/26/2021 5:54 PM, Jason Gunthorpe wrote:
> On Wed, May 26, 2021 at 05:22:22PM -0700, Dave Jiang wrote:
>> On 5/23/2021 4:50 PM, Jason Gunthorpe wrote:
>>> On Fri, May 21, 2021 at 05:20:37PM -0700, Dave Jiang wrote:
>>>> @@ -77,8 +80,18 @@ int idxd_mdev_host_init(struct idxd_device *idxd, struct mdev_driver *drv)
>>>> return rc;
>>>> }
>>>> + ims_info.max_slots = idxd->ims_size;
>>>> + ims_info.slots = idxd->reg_base + idxd->ims_offset;
>>>> + idxd->ims_domain = pci_ims_array_create_msi_irq_domain(idxd->pdev, &ims_info);
>>>> + if (!idxd->ims_domain) {
>>>> + dev_warn(dev, "Fail to acquire IMS domain\n");
>>>> + iommu_dev_disable_feature(dev, IOMMU_DEV_FEAT_AUX);
>>>> + return -ENODEV;
>>>> + }
>>> I'm quite surprised that every mdev doesn't create its own ims_domain
>>> in its probe function.
>>>
>>> This places a global total limit on the # of vectors which makes me
>>> ask what was the point of using IMS in the first place ?
>>>
>>> The entire idea for IMS was to make the whole allocation system fully
>>> dynamic based on demand.
>> Hi Jason, thank you for the review of the series.
>>
>> My understanding is that the driver creates a single IMS domain for the
>> device and provides the address base and IMS numbers for the domain based on
>> device IMS resources. So the IMS region needs to be contiguous. Each mdev
>> can call msi_domain_alloc_irqs() and acquire the number of IMS vectors it
>> desires and the DEV MSI core code will keep track of which vectors are being
>> used. This allows the mdev devices to dynamically allocate based on demand.
>> If the driver allocates a domain per mdev, it'll needs to do internal
>> accounting of the base and vector numbers for each of those domains that the
>> MSI core already provides. Isn't that what we are trying to avoid? As mdevs
>> come and go, that partitioning will become fragmented.
> I suppose it depends entirely on how the HW works.
>
> If the HW has a fixed number of interrupt vectors organized in a
> single table then by all means allocate a single domain that spans the
> entire fixed HW vector space. But then why do we have a ims_size
> variable here??
>
> However, that really begs the question of why the HW is using IMS at
> all? I'd expect needing 2x-10x the max MSI-X vector size before
> reaching for IMS.
>
> So does IDXD really have like a 4k - 40k entry linear IMS vector table
> to wrap a shared domain around?
>
> Basically, that isn't really "scalable" it is just "bigger".
>
> Fully scalable would be for every mdev to point to its own 2k entry
> IMS table that is allocated on the fly. Every mdev gets a domain and
> every domain is fully utilized by the mdev in emulating
> MSI-X. Basically for a device like idxd every PASID would have to map
> to a IMS vector table array.
>
> I suppose that was not what was done?
At least not for first gen of hardware. DSA 1.0 has 2k of IMS entries
total. ims_size is what is read from the device cap register. For MSIX,
the device only has 1 misc vector and 8 I/O vectors. That's why IMS is
being used for mdevs. We will discuss with our hardware people your
suggestion.
>
> Jason
next prev parent reply other threads:[~2021-05-27 1:15 UTC|newest]
Thread overview: 64+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-22 0:19 [PATCH v6 00/20] Add VFIO mediated device support and DEV-MSI support for the idxd driver Dave Jiang
2021-05-22 0:19 ` [PATCH v6 01/20] vfio/mdev: idxd: add theory of operation documentation for idxd mdev Dave Jiang
2021-05-22 0:19 ` [PATCH v6 02/20] dmaengine: idxd: add external module driver support for dsa_bus_type Dave Jiang
2021-05-22 0:19 ` [PATCH v6 03/20] dmaengine: idxd: add IMS offset and size retrieval code Dave Jiang
2021-05-22 0:19 ` [PATCH v6 04/20] dmaengine: idxd: add portal offset for IMS portals Dave Jiang
2021-05-22 0:19 ` [PATCH v6 05/20] vfio: mdev: common lib code for setting up Interrupt Message Store Dave Jiang
2021-05-24 0:02 ` Jason Gunthorpe
2021-05-28 1:49 ` Dave Jiang
2021-05-28 12:21 ` Jason Gunthorpe
2021-05-28 16:37 ` Dave Jiang
2021-05-28 16:39 ` Jason Gunthorpe
2021-05-31 10:41 ` Thomas Gleixner
2021-05-31 13:48 ` Thomas Gleixner
2021-05-31 15:24 ` Jason Gunthorpe
2021-06-08 15:57 ` Dave Jiang
2021-06-08 17:22 ` Jason Gunthorpe
2021-06-10 13:00 ` Thomas Gleixner
2021-05-22 0:19 ` [PATCH v6 06/20] vfio/mdev: idxd: add PCI config for read/write for mdev Dave Jiang
2021-05-22 0:19 ` [PATCH v6 07/20] vfio/mdev: idxd: Add administrative commands emulation " Dave Jiang
2021-05-22 0:19 ` [PATCH v6 08/20] vfio/mdev: idxd: Add mdev device context initialization Dave Jiang
2021-05-22 0:20 ` [PATCH v6 09/20] vfio/mdev: Add mmio read/write support for mdev Dave Jiang
2021-05-22 0:20 ` [PATCH v6 10/20] vfio/mdev: idxd: add mdev type as a new wq type Dave Jiang
2021-05-22 0:20 ` [PATCH v6 11/20] vfio/mdev: idxd: Add basic driver setup for idxd mdev Dave Jiang
2021-05-23 23:27 ` Jason Gunthorpe
2021-05-23 23:52 ` Jason Gunthorpe
2021-05-22 0:20 ` [PATCH v6 12/20] vfio: move VFIO PCI macros to common header Dave Jiang
2021-06-04 3:47 ` Alex Williamson
2021-05-22 0:20 ` [PATCH v6 13/20] vfio/mdev: idxd: add mdev driver registration and helper functions Dave Jiang
2021-05-23 23:32 ` Jason Gunthorpe
2021-05-23 23:46 ` Jason Gunthorpe
2021-05-22 0:20 ` [PATCH v6 14/20] vfio/mdev: idxd: add 1dwq-v1 mdev type Dave Jiang
2021-05-23 23:47 ` Jason Gunthorpe
2021-05-22 0:20 ` [PATCH v6 15/20] vfio/mdev: idxd: ims domain setup for the vdcm Dave Jiang
2021-05-23 23:50 ` Jason Gunthorpe
2021-05-27 0:22 ` Dave Jiang
2021-05-27 0:54 ` Jason Gunthorpe
2021-05-27 1:15 ` Dave Jiang [this message]
2021-05-27 1:41 ` Raj, Ashok
2021-05-27 13:36 ` Jason Gunthorpe
2021-05-31 14:02 ` Thomas Gleixner
2021-05-31 16:57 ` Jason Gunthorpe
2021-05-31 23:55 ` Thomas Gleixner
2021-06-01 12:16 ` Jason Gunthorpe
2021-05-22 0:20 ` [PATCH v6 16/20] vfio/mdev: idxd: add new wq state for mdev Dave Jiang
2021-05-22 0:20 ` [PATCH v6 17/20] vfio/mdev: idxd: add error notification from host driver to mediated device Dave Jiang
2021-05-22 0:20 ` [PATCH v6 18/20] vfio: move vfio_pci_set_ctx_trigger_single to common code Dave Jiang
2021-05-22 0:21 ` [PATCH v6 19/20] vfio: mdev: Add device request interface Dave Jiang
2021-05-23 22:38 ` Jason Gunthorpe
2021-05-22 0:21 ` [PATCH v6 20/20] vfio: mdev: idxd: setup request interrupt Dave Jiang
2021-05-23 23:22 ` [PATCH v6 00/20] Add VFIO mediated device support and DEV-MSI support for the idxd driver Jason Gunthorpe
2021-06-02 15:40 ` Dave Jiang
2021-06-02 23:17 ` Jason Gunthorpe
2021-06-03 1:11 ` Tian, Kevin
2021-06-03 1:49 ` Jason Gunthorpe
2021-06-03 5:52 ` Tian, Kevin
2021-06-03 11:23 ` Jason Gunthorpe
2021-06-04 3:40 ` Alex Williamson
2021-06-07 6:22 ` Tian, Kevin
2021-06-07 18:13 ` Dave Jiang
2021-06-07 19:11 ` Jason Gunthorpe
2021-06-08 16:02 ` Dave Jiang
2021-06-11 18:21 ` Dave Jiang
2021-06-11 18:29 ` Jason Gunthorpe
2021-06-07 18:28 ` Jason Gunthorpe
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=f5606802-2ff3-2f54-0ff1-c1f1dd59f52c@intel.com \
--to=dave.jiang@intel.com \
--cc=alex.williamson@redhat.com \
--cc=ashok.raj@intel.com \
--cc=baolu.lu@intel.com \
--cc=dan.j.williams@intel.com \
--cc=dmaengine@vger.kernel.org \
--cc=eric.auger@redhat.com \
--cc=jacob.jun.pan@intel.com \
--cc=jgg@nvidia.com \
--cc=kevin.tian@intel.com \
--cc=kvm@vger.kernel.org \
--cc=kwankhede@nvidia.com \
--cc=linux-kernel@vger.kernel.org \
--cc=megha.dey@intel.com \
--cc=pbonzini@redhat.com \
--cc=sanjay.k.kumar@intel.com \
--cc=tglx@linutronix.de \
--cc=tony.luck@intel.com \
--cc=vkoul@kernel.org \
--cc=yi.l.liu@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).