All of lore.kernel.org
 help / color / mirror / Atom feed
From: Lu Baolu <baolu.lu@linux.intel.com>
To: Robin Murphy <robin.murphy@arm.com>, Tom Murphy <tmurphy@arista.com>
Cc: baolu.lu@linux.intel.com, "Tian, Kevin" <kevin.tian@intel.com>,
	Ashok Raj <ashok.raj@intel.com>, Dmitry Safonov <dima@arista.com>,
	linux-kernel@vger.kernel.org, iommu@lists.linux-foundation.org,
	jacob.jun.pan@intel.com, David Woodhouse <dwmw2@infradead.org>
Subject: Re: [PATCH v3 1/8] iommu: Add ops entry for supported default domain type
Date: Thu, 9 May 2019 10:30:16 +0800	[thread overview]
Message-ID: <3e0da076-4916-1a02-615c-927c1b3528b8@linux.intel.com> (raw)
In-Reply-To: <bba1f327-21b7-ed3c-8fd4-217ad97a6a7c@arm.com>

Hi Robin,

On 5/7/19 6:28 PM, Robin Murphy wrote:
> On 06/05/2019 16:32, Tom Murphy via iommu wrote:
>> The AMD driver already solves this problem and uses the generic
>> iommu_request_dm_for_dev function. It seems like both drivers have the
>> same problem and could use the same solution. Is there any reason we
>> can't have use the same solution for the intel and amd driver?
>>
>> Could we just  copy the implementation of the AMD driver? It would be
>> nice to have the same behavior across both drivers especially as we
>> move to make both drivers use more generic code.
> 
> TBH I don't think the API really needs to be involved at all here. 
> Drivers can already not provide the requested default domain type if 
> they don't support it, so as long as the driver can ensure that the 
> device ends up with IOMMU or direct DMA ops as appropriate, I don't see 
> any great problem with drivers just returning a passthrough domain when 
> a DMA domain was requested, or vice versa (and logging a message that 
> the requested type was overridden). The only type that we really do have 
> to honour strictly is non-default (i.e. unmanaged) domains.

I agree with you that we only have to honor strictly the non-default
domains. But domain type saved in iommu_domain is consumed in iommu.c
and exposed to user through sysfs. It's not clean if the iommu driver
silently replace the default domain.

Best regards,
Lu Baolu

> 
> Robin.
> 
>> On Mon, Apr 29, 2019 at 3:16 AM Lu Baolu <baolu.lu@linux.intel.com> 
>> wrote:
>>>
>>> This adds an optional ops entry to query the default domain
>>> types supported by the iommu driver for  a specific device.
>>> This is necessary in cases where the iommu driver can only
>>> support a specific type of default domain for a device. In
>>> normal cases, this ops will return IOMMU_DOMAIN_ANY which
>>> indicates that the iommu driver supports both IOMMU_DOMAIN_DMA
>>> and IOMMU_DOMAIN_IDENTITY, hence the static default domain
>>> type will be used.
>>>
>>> Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
>>> ---
>>>   drivers/iommu/iommu.c | 13 ++++++++++---
>>>   include/linux/iommu.h | 11 +++++++++++
>>>   2 files changed, 21 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c
>>> index acd6830e6e9b..1ad9a1f2e078 100644
>>> --- a/drivers/iommu/iommu.c
>>> +++ b/drivers/iommu/iommu.c
>>> @@ -1097,15 +1097,22 @@ struct iommu_group 
>>> *iommu_group_get_for_dev(struct device *dev)
>>>           * IOMMU driver.
>>>           */
>>>          if (!group->default_domain) {
>>> +               unsigned int domain_type = IOMMU_DOMAIN_ANY;
>>>                  struct iommu_domain *dom;
>>>
>>> -               dom = __iommu_domain_alloc(dev->bus, 
>>> iommu_def_domain_type);
>>> -               if (!dom && iommu_def_domain_type != IOMMU_DOMAIN_DMA) {
>>> +               if (ops->def_domain_type)
>>> +                       domain_type = ops->def_domain_type(dev);
>>> +
>>> +               if (domain_type == IOMMU_DOMAIN_ANY)
>>> +                       domain_type = iommu_def_domain_type;
>>> +
>>> +               dom = __iommu_domain_alloc(dev->bus, domain_type);
>>> +               if (!dom && domain_type != IOMMU_DOMAIN_DMA) {
>>>                          dom = __iommu_domain_alloc(dev->bus, 
>>> IOMMU_DOMAIN_DMA);
>>>                          if (dom) {
>>>                                  dev_warn(dev,
>>>                                           "failed to allocate default 
>>> IOMMU domain of type %u; falling back to IOMMU_DOMAIN_DMA",
>>> -                                        iommu_def_domain_type);
>>> +                                        domain_type);
>>>                          }
>>>                  }
>>>
>>> diff --git a/include/linux/iommu.h b/include/linux/iommu.h
>>> index 8239ece9fdfc..ba9a5b996a63 100644
>>> --- a/include/linux/iommu.h
>>> +++ b/include/linux/iommu.h
>>> @@ -79,12 +79,16 @@ struct iommu_domain_geometry {
>>>    *     IOMMU_DOMAIN_DMA        - Internally used for DMA-API 
>>> implementations.
>>>    *                               This flag allows IOMMU drivers to 
>>> implement
>>>    *                               certain optimizations for these 
>>> domains
>>> + *     IOMMU_DOMAIN_ANY        - All domain types defined here
>>>    */
>>>   #define IOMMU_DOMAIN_BLOCKED   (0U)
>>>   #define IOMMU_DOMAIN_IDENTITY  (__IOMMU_DOMAIN_PT)
>>>   #define IOMMU_DOMAIN_UNMANAGED (__IOMMU_DOMAIN_PAGING)
>>>   #define IOMMU_DOMAIN_DMA       (__IOMMU_DOMAIN_PAGING |        \
>>>                                   __IOMMU_DOMAIN_DMA_API)
>>> +#define IOMMU_DOMAIN_ANY       (IOMMU_DOMAIN_IDENTITY |        \
>>> +                                IOMMU_DOMAIN_UNMANAGED |       \
>>> +                                IOMMU_DOMAIN_DMA)
>>>
>>>   struct iommu_domain {
>>>          unsigned type;
>>> @@ -196,6 +200,11 @@ enum iommu_dev_features {
>>>    * @dev_feat_enabled: check enabled feature
>>>    * @aux_attach/detach_dev: aux-domain specific attach/detach entries.
>>>    * @aux_get_pasid: get the pasid given an aux-domain
>>> + * @def_domain_type: get per-device default domain type that the IOMMU
>>> + *             driver is able to support. Valid returns values:
>>> + *             - IOMMU_DOMAIN_DMA: only suports non-identity domain
>>> + *             - IOMMU_DOMAIN_IDENTITY: only supports identity domain
>>> + *             - IOMMU_DOMAIN_ANY: supports all
>>>    * @pgsize_bitmap: bitmap of all possible supported page sizes
>>>    */
>>>   struct iommu_ops {
>>> @@ -251,6 +260,8 @@ struct iommu_ops {
>>>          void (*aux_detach_dev)(struct iommu_domain *domain, struct 
>>> device *dev);
>>>          int (*aux_get_pasid)(struct iommu_domain *domain, struct 
>>> device *dev);
>>>
>>> +       int (*def_domain_type)(struct device *dev);
>>> +
>>>          unsigned long pgsize_bitmap;
>>>   };
>>>
>>> -- 
>>> 2.17.1
>>>
>> _______________________________________________
>> iommu mailing list
>> iommu@lists.linux-foundation.org
>> https://lists.linuxfoundation.org/mailman/listinfo/iommu
>>
> 

WARNING: multiple messages have this Message-ID (diff)
From: Lu Baolu <baolu.lu@linux.intel.com>
To: Robin Murphy <robin.murphy@arm.com>, Tom Murphy <tmurphy@arista.com>
Cc: "Tian, Kevin" <kevin.tian@intel.com>,
	Ashok Raj <ashok.raj@intel.com>, Dmitry Safonov <dima@arista.com>,
	linux-kernel@vger.kernel.org, iommu@lists.linux-foundation.org,
	jacob.jun.pan@intel.com, David Woodhouse <dwmw2@infradead.org>
Subject: Re: [PATCH v3 1/8] iommu: Add ops entry for supported default domain type
Date: Thu, 9 May 2019 10:30:16 +0800	[thread overview]
Message-ID: <3e0da076-4916-1a02-615c-927c1b3528b8@linux.intel.com> (raw)
In-Reply-To: <bba1f327-21b7-ed3c-8fd4-217ad97a6a7c@arm.com>

Hi Robin,

On 5/7/19 6:28 PM, Robin Murphy wrote:
> On 06/05/2019 16:32, Tom Murphy via iommu wrote:
>> The AMD driver already solves this problem and uses the generic
>> iommu_request_dm_for_dev function. It seems like both drivers have the
>> same problem and could use the same solution. Is there any reason we
>> can't have use the same solution for the intel and amd driver?
>>
>> Could we just  copy the implementation of the AMD driver? It would be
>> nice to have the same behavior across both drivers especially as we
>> move to make both drivers use more generic code.
> 
> TBH I don't think the API really needs to be involved at all here. 
> Drivers can already not provide the requested default domain type if 
> they don't support it, so as long as the driver can ensure that the 
> device ends up with IOMMU or direct DMA ops as appropriate, I don't see 
> any great problem with drivers just returning a passthrough domain when 
> a DMA domain was requested, or vice versa (and logging a message that 
> the requested type was overridden). The only type that we really do have 
> to honour strictly is non-default (i.e. unmanaged) domains.

I agree with you that we only have to honor strictly the non-default
domains. But domain type saved in iommu_domain is consumed in iommu.c
and exposed to user through sysfs. It's not clean if the iommu driver
silently replace the default domain.

Best regards,
Lu Baolu

> 
> Robin.
> 
>> On Mon, Apr 29, 2019 at 3:16 AM Lu Baolu <baolu.lu@linux.intel.com> 
>> wrote:
>>>
>>> This adds an optional ops entry to query the default domain
>>> types supported by the iommu driver for  a specific device.
>>> This is necessary in cases where the iommu driver can only
>>> support a specific type of default domain for a device. In
>>> normal cases, this ops will return IOMMU_DOMAIN_ANY which
>>> indicates that the iommu driver supports both IOMMU_DOMAIN_DMA
>>> and IOMMU_DOMAIN_IDENTITY, hence the static default domain
>>> type will be used.
>>>
>>> Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
>>> ---
>>>   drivers/iommu/iommu.c | 13 ++++++++++---
>>>   include/linux/iommu.h | 11 +++++++++++
>>>   2 files changed, 21 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c
>>> index acd6830e6e9b..1ad9a1f2e078 100644
>>> --- a/drivers/iommu/iommu.c
>>> +++ b/drivers/iommu/iommu.c
>>> @@ -1097,15 +1097,22 @@ struct iommu_group 
>>> *iommu_group_get_for_dev(struct device *dev)
>>>           * IOMMU driver.
>>>           */
>>>          if (!group->default_domain) {
>>> +               unsigned int domain_type = IOMMU_DOMAIN_ANY;
>>>                  struct iommu_domain *dom;
>>>
>>> -               dom = __iommu_domain_alloc(dev->bus, 
>>> iommu_def_domain_type);
>>> -               if (!dom && iommu_def_domain_type != IOMMU_DOMAIN_DMA) {
>>> +               if (ops->def_domain_type)
>>> +                       domain_type = ops->def_domain_type(dev);
>>> +
>>> +               if (domain_type == IOMMU_DOMAIN_ANY)
>>> +                       domain_type = iommu_def_domain_type;
>>> +
>>> +               dom = __iommu_domain_alloc(dev->bus, domain_type);
>>> +               if (!dom && domain_type != IOMMU_DOMAIN_DMA) {
>>>                          dom = __iommu_domain_alloc(dev->bus, 
>>> IOMMU_DOMAIN_DMA);
>>>                          if (dom) {
>>>                                  dev_warn(dev,
>>>                                           "failed to allocate default 
>>> IOMMU domain of type %u; falling back to IOMMU_DOMAIN_DMA",
>>> -                                        iommu_def_domain_type);
>>> +                                        domain_type);
>>>                          }
>>>                  }
>>>
>>> diff --git a/include/linux/iommu.h b/include/linux/iommu.h
>>> index 8239ece9fdfc..ba9a5b996a63 100644
>>> --- a/include/linux/iommu.h
>>> +++ b/include/linux/iommu.h
>>> @@ -79,12 +79,16 @@ struct iommu_domain_geometry {
>>>    *     IOMMU_DOMAIN_DMA        - Internally used for DMA-API 
>>> implementations.
>>>    *                               This flag allows IOMMU drivers to 
>>> implement
>>>    *                               certain optimizations for these 
>>> domains
>>> + *     IOMMU_DOMAIN_ANY        - All domain types defined here
>>>    */
>>>   #define IOMMU_DOMAIN_BLOCKED   (0U)
>>>   #define IOMMU_DOMAIN_IDENTITY  (__IOMMU_DOMAIN_PT)
>>>   #define IOMMU_DOMAIN_UNMANAGED (__IOMMU_DOMAIN_PAGING)
>>>   #define IOMMU_DOMAIN_DMA       (__IOMMU_DOMAIN_PAGING |        \
>>>                                   __IOMMU_DOMAIN_DMA_API)
>>> +#define IOMMU_DOMAIN_ANY       (IOMMU_DOMAIN_IDENTITY |        \
>>> +                                IOMMU_DOMAIN_UNMANAGED |       \
>>> +                                IOMMU_DOMAIN_DMA)
>>>
>>>   struct iommu_domain {
>>>          unsigned type;
>>> @@ -196,6 +200,11 @@ enum iommu_dev_features {
>>>    * @dev_feat_enabled: check enabled feature
>>>    * @aux_attach/detach_dev: aux-domain specific attach/detach entries.
>>>    * @aux_get_pasid: get the pasid given an aux-domain
>>> + * @def_domain_type: get per-device default domain type that the IOMMU
>>> + *             driver is able to support. Valid returns values:
>>> + *             - IOMMU_DOMAIN_DMA: only suports non-identity domain
>>> + *             - IOMMU_DOMAIN_IDENTITY: only supports identity domain
>>> + *             - IOMMU_DOMAIN_ANY: supports all
>>>    * @pgsize_bitmap: bitmap of all possible supported page sizes
>>>    */
>>>   struct iommu_ops {
>>> @@ -251,6 +260,8 @@ struct iommu_ops {
>>>          void (*aux_detach_dev)(struct iommu_domain *domain, struct 
>>> device *dev);
>>>          int (*aux_get_pasid)(struct iommu_domain *domain, struct 
>>> device *dev);
>>>
>>> +       int (*def_domain_type)(struct device *dev);
>>> +
>>>          unsigned long pgsize_bitmap;
>>>   };
>>>
>>> -- 
>>> 2.17.1
>>>
>> _______________________________________________
>> iommu mailing list
>> iommu@lists.linux-foundation.org
>> https://lists.linuxfoundation.org/mailman/listinfo/iommu
>>
> 
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu

  reply	other threads:[~2019-05-09  2:36 UTC|newest]

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-04-29  2:09 [PATCH v3 0/8] iommu/vt-d: Delegate DMA domain to generic iommu Lu Baolu
2019-04-29  2:09 ` Lu Baolu
2019-04-29  2:09 ` [PATCH v3 1/8] iommu: Add ops entry for supported default domain type Lu Baolu
2019-04-29  2:09   ` Lu Baolu
2019-05-06 15:32   ` Tom Murphy
2019-05-06 15:32     ` Tom Murphy via iommu
2019-05-07 10:28     ` Robin Murphy
2019-05-07 10:28       ` Robin Murphy
2019-05-09  2:30       ` Lu Baolu [this message]
2019-05-09  2:30         ` Lu Baolu
2019-05-09 16:11         ` Robin Murphy
2019-05-09 16:11           ` Robin Murphy
2019-05-10  5:29           ` Lu Baolu
2019-05-10  5:29             ` Lu Baolu
2019-05-09  2:22     ` Lu Baolu
2019-05-09  2:22       ` Lu Baolu
2019-04-29  2:09 ` [PATCH v3 2/8] iommu/vt-d: Implement apply_resv_region iommu ops entry Lu Baolu
2019-04-29  2:09   ` Lu Baolu
2019-04-29  2:09 ` [PATCH v3 3/8] iommu/vt-d: Expose ISA direct mapping region via iommu_get_resv_regions Lu Baolu
2019-04-29  2:09   ` Lu Baolu
2019-04-29  2:09 ` [PATCH v3 4/8] iommu/vt-d: Enable DMA remapping after rmrr mapped Lu Baolu
2019-04-29  2:09   ` Lu Baolu
2019-04-29  2:09 ` [PATCH v3 5/8] iommu/vt-d: Implement def_domain_type iommu ops entry Lu Baolu
2019-04-29  2:09   ` Lu Baolu
2019-04-29 20:03   ` Christoph Hellwig
2019-04-29 20:03     ` Christoph Hellwig
2019-04-30  2:11     ` Lu Baolu
2019-04-30  2:11       ` Lu Baolu
2019-05-06 15:25       ` Tom Murphy
2019-05-06 15:25         ` Tom Murphy via iommu
2019-05-09  4:31         ` Lu Baolu
2019-05-09  4:31           ` Lu Baolu
2019-04-29  2:09 ` [PATCH v3 6/8] iommu/vt-d: Allow DMA domains to be allocated by iommu ops Lu Baolu
2019-04-29  2:09   ` Lu Baolu
2019-04-29  2:09 ` [PATCH v3 7/8] iommu/vt-d: Remove lazy allocation of domains Lu Baolu
2019-04-29  2:09   ` Lu Baolu
2019-04-29  2:09 ` [PATCH v3 8/8] iommu/vt-d: Implement is_attach_deferred iommu ops entry Lu Baolu
2019-04-29  2:09   ` 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=3e0da076-4916-1a02-615c-927c1b3528b8@linux.intel.com \
    --to=baolu.lu@linux.intel.com \
    --cc=ashok.raj@intel.com \
    --cc=dima@arista.com \
    --cc=dwmw2@infradead.org \
    --cc=iommu@lists.linux-foundation.org \
    --cc=jacob.jun.pan@intel.com \
    --cc=kevin.tian@intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=robin.murphy@arm.com \
    --cc=tmurphy@arista.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.