All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Tian, Kevin" <kevin.tian@intel.com>
To: Jason Gunthorpe <jgg@nvidia.com>, Lu Baolu <baolu.lu@linux.intel.com>
Cc: "Raj, Ashok" <ashok.raj@intel.com>, Will Deacon <will@kernel.org>,
	Jean-Philippe Brucker <jean-philippe@linaro.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Christoph Hellwig <hch@infradead.org>,
	"iommu@lists.linux-foundation.org"
	<iommu@lists.linux-foundation.org>,
	"Pan, Jacob jun" <jacob.jun.pan@intel.com>,
	Robin Murphy <robin.murphy@arm.com>
Subject: RE: [PATCH RFC v2 03/11] iommu/sva: Add iommu_domain type for SVA
Date: Sat, 2 Apr 2022 08:43:16 +0000	[thread overview]
Message-ID: <BN9PR11MB5276A39F481E2E31B0FCFEAE8CE39@BN9PR11MB5276.namprd11.prod.outlook.com> (raw)
In-Reply-To: <20220330190201.GB2120790@nvidia.com>

> From: Jason Gunthorpe <jgg@nvidia.com>
> Sent: Thursday, March 31, 2022 3:02 AM
> 
> On Tue, Mar 29, 2022 at 01:37:52PM +0800, Lu Baolu wrote:
> > @@ -95,6 +101,7 @@ struct iommu_domain {
> >  	void *handler_token;
> >  	struct iommu_domain_geometry geometry;
> >  	struct iommu_dma_cookie *iova_cookie;
> > +	struct iommu_sva_cookie *sva_cookie;
> 
> Cookie is still the wrong word to use here
> 
> > +struct iommu_sva_cookie {
> > +	struct mm_struct *mm;
> > +	ioasid_t pasid;
> > +	refcount_t users;
> 
> Really surprised to see a refcount buried inside the iommu_domain..
> 
> This design seems inside out, the SVA struct should 'enclose' the domain, not
> be a pointer inside it.
> 
> struct iommu_sva_domain {
>        struct kref_t kref;
>        struct mm_struct *mm;
>        ioasid_t pasid;
> 
>        /* All the domains that are linked to this */
>        struct xarray domain_list;
> };
> 
> And then you could have a pointer to that inside the mm_struct instead
> of just the naked pasid.
> 
> > +static __maybe_unused struct iommu_domain *
> 
> Why maybe unused?
> 
> > +iommu_sva_get_domain(struct device *dev, struct mm_struct *mm)
> > +{
> > +	struct iommu_domain *domain;
> > +	ioasid_t pasid = mm->pasid;
> > +
> > +	if (pasid == INVALID_IOASID)
> > +		return NULL;
> > +
> > +	domain = xa_load(&sva_domain_array, pasid);
> > +	if (!domain)
> > +		return iommu_sva_alloc_domain(dev, mm);
> > +	iommu_sva_domain_get_user(domain);
> 
> This assumes any domain is interchangeable with any device, which is
> not the iommu model. We need a domain op to check if a device is
> compatiable with the domain for vfio an iommufd, this should do the
> same.

This suggests that mm_struct needs to include the format information
of the CPU page table so the format can be checked by the domain op?

> 
> It means each mm can have a list of domains associated with it and a
> new domain is auto-created if the device doesn't work with any of the
> existing domains.
> 

mm has only one page table and one format. If a device is incompatible
with an existing domain wrapping that page table, how come creating
another domain could make it compatible?

Thanks
Kevin
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu

WARNING: multiple messages have this Message-ID (diff)
From: "Tian, Kevin" <kevin.tian@intel.com>
To: Jason Gunthorpe <jgg@nvidia.com>, Lu Baolu <baolu.lu@linux.intel.com>
Cc: Joerg Roedel <joro@8bytes.org>,
	Christoph Hellwig <hch@infradead.org>,
	"Raj, Ashok" <ashok.raj@intel.com>, Will Deacon <will@kernel.org>,
	"Robin Murphy" <robin.murphy@arm.com>,
	Jean-Philippe Brucker <jean-philippe@linaro.com>,
	Eric Auger <eric.auger@redhat.com>,
	"Liu, Yi L" <yi.l.liu@intel.com>,
	"Pan, Jacob jun" <jacob.jun.pan@intel.com>,
	"iommu@lists.linux-foundation.org"
	<iommu@lists.linux-foundation.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: RE: [PATCH RFC v2 03/11] iommu/sva: Add iommu_domain type for SVA
Date: Sat, 2 Apr 2022 08:43:16 +0000	[thread overview]
Message-ID: <BN9PR11MB5276A39F481E2E31B0FCFEAE8CE39@BN9PR11MB5276.namprd11.prod.outlook.com> (raw)
In-Reply-To: <20220330190201.GB2120790@nvidia.com>

> From: Jason Gunthorpe <jgg@nvidia.com>
> Sent: Thursday, March 31, 2022 3:02 AM
> 
> On Tue, Mar 29, 2022 at 01:37:52PM +0800, Lu Baolu wrote:
> > @@ -95,6 +101,7 @@ struct iommu_domain {
> >  	void *handler_token;
> >  	struct iommu_domain_geometry geometry;
> >  	struct iommu_dma_cookie *iova_cookie;
> > +	struct iommu_sva_cookie *sva_cookie;
> 
> Cookie is still the wrong word to use here
> 
> > +struct iommu_sva_cookie {
> > +	struct mm_struct *mm;
> > +	ioasid_t pasid;
> > +	refcount_t users;
> 
> Really surprised to see a refcount buried inside the iommu_domain..
> 
> This design seems inside out, the SVA struct should 'enclose' the domain, not
> be a pointer inside it.
> 
> struct iommu_sva_domain {
>        struct kref_t kref;
>        struct mm_struct *mm;
>        ioasid_t pasid;
> 
>        /* All the domains that are linked to this */
>        struct xarray domain_list;
> };
> 
> And then you could have a pointer to that inside the mm_struct instead
> of just the naked pasid.
> 
> > +static __maybe_unused struct iommu_domain *
> 
> Why maybe unused?
> 
> > +iommu_sva_get_domain(struct device *dev, struct mm_struct *mm)
> > +{
> > +	struct iommu_domain *domain;
> > +	ioasid_t pasid = mm->pasid;
> > +
> > +	if (pasid == INVALID_IOASID)
> > +		return NULL;
> > +
> > +	domain = xa_load(&sva_domain_array, pasid);
> > +	if (!domain)
> > +		return iommu_sva_alloc_domain(dev, mm);
> > +	iommu_sva_domain_get_user(domain);
> 
> This assumes any domain is interchangeable with any device, which is
> not the iommu model. We need a domain op to check if a device is
> compatiable with the domain for vfio an iommufd, this should do the
> same.

This suggests that mm_struct needs to include the format information
of the CPU page table so the format can be checked by the domain op?

> 
> It means each mm can have a list of domains associated with it and a
> new domain is auto-created if the device doesn't work with any of the
> existing domains.
> 

mm has only one page table and one format. If a device is incompatible
with an existing domain wrapping that page table, how come creating
another domain could make it compatible?

Thanks
Kevin

  reply	other threads:[~2022-04-02  8:43 UTC|newest]

Thread overview: 124+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-03-29  5:37 [PATCH RFC v2 00/11] iommu: SVA and IOPF refactoring Lu Baolu
2022-03-29  5:37 ` Lu Baolu
2022-03-29  5:37 ` [PATCH RFC v2 01/11] iommu: Add pasid_bits field in struct dev_iommu Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-29 21:00   ` Jacob Pan
2022-03-29 21:00     ` Jacob Pan
2022-03-30  4:30     ` Lu Baolu
2022-03-30  4:30       ` Lu Baolu
2022-03-30  7:05   ` Tian, Kevin
2022-03-30  7:05     ` Tian, Kevin
2022-03-30 11:58     ` Lu Baolu
2022-03-30 11:58       ` Lu Baolu
2022-03-29  5:37 ` [PATCH RFC v2 02/11] iommu: Add iommu_group_singleton_lockdown() Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-29  8:42   ` Tian, Kevin
2022-03-29  8:42     ` Tian, Kevin
2022-03-29 11:42     ` Jason Gunthorpe
2022-03-29 11:42       ` Jason Gunthorpe via iommu
2022-03-30  6:50       ` Tian, Kevin
2022-03-30  6:50         ` Tian, Kevin
2022-03-30 11:57         ` Lu Baolu
2022-03-30 11:57           ` Lu Baolu
2022-03-30 11:58         ` Jason Gunthorpe via iommu
2022-03-30 11:58           ` Jason Gunthorpe
2022-03-30 14:12           ` Tian, Kevin
2022-03-30 14:12             ` Tian, Kevin
2022-03-30 14:30             ` Jason Gunthorpe
2022-03-30 14:30               ` Jason Gunthorpe via iommu
2022-04-02  7:12               ` Tian, Kevin
2022-04-02  7:12                 ` Tian, Kevin
2022-04-02 23:29                 ` Jason Gunthorpe
2022-04-02 23:29                   ` Jason Gunthorpe via iommu
2022-04-06 10:02                 ` Lu Baolu
2022-04-06 10:02                   ` Lu Baolu
2022-04-06 10:44                   ` Tian, Kevin
2022-04-06 10:44                     ` Tian, Kevin
2022-04-06 11:03                     ` Lu Baolu
2022-04-06 11:03                       ` Lu Baolu
2022-04-06 23:56                       ` Tian, Kevin
2022-04-06 23:56                         ` Tian, Kevin
2022-03-30 14:18           ` Tian, Kevin
2022-03-30 14:18             ` Tian, Kevin
2022-03-30 15:04             ` Alex Williamson
2022-03-30 15:04               ` Alex Williamson
2022-04-04  5:43           ` Lu Baolu
2022-04-04  5:43             ` Lu Baolu
2022-04-04 17:24             ` Jason Gunthorpe via iommu
2022-04-04 17:24               ` Jason Gunthorpe
2022-04-05  6:12               ` Lu Baolu
2022-04-05  6:12                 ` Lu Baolu
2022-04-05 14:10                 ` Jason Gunthorpe via iommu
2022-04-05 14:10                   ` Jason Gunthorpe
2022-04-06  9:51                   ` Lu Baolu
2022-04-06  9:51                     ` Lu Baolu
2022-04-01  6:20       ` Yi Liu
2022-04-01  6:20         ` Yi Liu
2022-04-01 11:52         ` Jason Gunthorpe
2022-04-01 11:52           ` Jason Gunthorpe via iommu
2022-03-30  4:59     ` Lu Baolu
2022-03-30  4:59       ` Lu Baolu
2022-03-30  6:55       ` Tian, Kevin
2022-03-30  6:55         ` Tian, Kevin
2022-04-01  5:49     ` Yi Liu
2022-04-01  5:49       ` Yi Liu
2022-03-29  5:37 ` [PATCH RFC v2 03/11] iommu/sva: Add iommu_domain type for SVA Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-29 21:38   ` Jacob Pan
2022-03-29 21:38     ` Jacob Pan
2022-03-30  4:35     ` Lu Baolu
2022-03-30  4:35       ` Lu Baolu
2022-03-30 19:02   ` Jason Gunthorpe
2022-03-30 19:02     ` Jason Gunthorpe via iommu
2022-04-02  8:43     ` Tian, Kevin [this message]
2022-04-02  8:43       ` Tian, Kevin
2022-04-02 23:32       ` Jason Gunthorpe
2022-04-02 23:32         ` Jason Gunthorpe via iommu
2022-04-04  6:09         ` Lu Baolu
2022-04-04  6:09           ` Lu Baolu
2022-04-06  1:00         ` Tian, Kevin
2022-04-06  1:00           ` Tian, Kevin
2022-04-06  1:23           ` Jason Gunthorpe via iommu
2022-04-06  1:23             ` Jason Gunthorpe
2022-04-06  5:58             ` Tian, Kevin
2022-04-06  5:58               ` Tian, Kevin
2022-04-06 12:32               ` Robin Murphy
2022-04-06 12:32                 ` Robin Murphy
2022-04-06 13:06                 ` Jason Gunthorpe via iommu
2022-04-06 13:06                   ` Jason Gunthorpe
2022-04-06 13:37                   ` Robin Murphy
2022-04-06 13:37                     ` Robin Murphy
2022-04-06 14:01                     ` Jason Gunthorpe via iommu
2022-04-06 14:01                       ` Jason Gunthorpe
2022-04-07  0:11                 ` Tian, Kevin
2022-04-07  0:11                   ` Tian, Kevin
2022-03-29  5:37 ` [PATCH RFC v2 04/11] iommu: Add attach/detach_dev_pasid domain ops Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-30 19:08   ` Jason Gunthorpe
2022-03-30 19:08     ` Jason Gunthorpe via iommu
2022-04-04  6:47     ` Lu Baolu
2022-04-04  6:47       ` Lu Baolu
2022-03-29  5:37 ` [PATCH RFC v2 05/11] iommu/vt-d: Remove SVM_FLAG_SUPERVISOR_MODE suport Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-29  5:37 ` [PATCH RFC v2 06/11] iommu/vt-d: Add SVA domain support Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-30 19:09   ` Jason Gunthorpe
2022-03-30 19:09     ` Jason Gunthorpe via iommu
2022-04-04  6:52     ` Lu Baolu
2022-04-04  6:52       ` Lu Baolu
2022-03-29  5:37 ` [PATCH RFC v2 07/11] arm-smmu-v3/sva: " Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-29  5:37 ` [PATCH RFC v2 08/11] iommu/sva: Use attach/detach_pasid_dev in SVA interfaces Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-31 20:59   ` Jacob Pan
2022-03-31 20:59     ` Jacob Pan
2022-03-31 22:26     ` Jason Gunthorpe
2022-03-31 22:26       ` Jason Gunthorpe via iommu
2022-04-04  5:55       ` Lu Baolu
2022-04-04  5:55         ` Lu Baolu
2022-03-29  5:37 ` [PATCH RFC v2 09/11] iommu: Remove SVA related callbacks from iommu ops Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-29  5:37 ` [PATCH RFC v2 10/11] iommu: Per-domain I/O page fault handling Lu Baolu
2022-03-29  5:37   ` Lu Baolu
2022-03-29  5:38 ` [PATCH RFC v2 11/11] iommu: Rename iommu-sva-lib.{c,h} Lu Baolu
2022-03-29  5:38   ` Lu Baolu

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=BN9PR11MB5276A39F481E2E31B0FCFEAE8CE39@BN9PR11MB5276.namprd11.prod.outlook.com \
    --to=kevin.tian@intel.com \
    --cc=ashok.raj@intel.com \
    --cc=baolu.lu@linux.intel.com \
    --cc=hch@infradead.org \
    --cc=iommu@lists.linux-foundation.org \
    --cc=jacob.jun.pan@intel.com \
    --cc=jean-philippe@linaro.com \
    --cc=jgg@nvidia.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=robin.murphy@arm.com \
    --cc=will@kernel.org \
    /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.