Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1691193 > unrolled thread

[PATCH 0/5] FlexRM support in VFIO platform

Back to article view | Back to linux.kernel

Contents

  [PATCH 0/5] FlexRM support in VFIO platform Anup Patel <anup.patel@broadcom.com> - 2017-07-19 11:40 +0200
    [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS Anup Patel <anup.patel@broadcom.com> - 2017-07-19 11:40 +0200
      Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS Robin Murphy <robin.murphy@arm.com> - 2017-07-19 13:00 +0200
        Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS Anup Patel <anup.patel@broadcom.com> - 2017-07-19 13:20 +0200
          Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS Robin Murphy <robin.murphy@arm.com> - 2017-07-19 13:30 +0200
          Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS Will Deacon <will.deacon@arm.com> - 2017-07-19 13:30 +0200
            Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS Anup Patel <anup.patel@broadcom.com> - 2017-07-19 13:30 +0200
              Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS Anup Patel <anup.patel@broadcom.com> - 2017-07-19 13:40 +0200
              Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS Will Deacon <will.deacon@arm.com> - 2017-07-19 13:40 +0200
    [PATCH 2/5] iommu/arm-smmu: add IOMMU_CAP_BYPASS to the ARM SMMU driver Anup Patel <anup.patel@broadcom.com> - 2017-07-19 11:40 +0200
      Re: [PATCH 2/5] iommu/arm-smmu: add IOMMU_CAP_BYPASS to the ARM SMMU  driver Robin Murphy <robin.murphy@arm.com> - 2017-07-19 13:00 +0200
    Re: [PATCH 0/5] FlexRM support in VFIO platform Robin Murphy <robin.murphy@arm.com> - 2017-07-19 13:00 +0200
      Re: [PATCH 0/5] FlexRM support in VFIO platform Anup Patel <anup.patel@broadcom.com> - 2017-07-19 13:20 +0200
        Re: [PATCH 0/5] FlexRM support in VFIO platform Robin Murphy <robin.murphy@arm.com> - 2017-07-19 13:30 +0200
          Re: [PATCH 0/5] FlexRM support in VFIO platform Anup Patel <anup.patel@broadcom.com> - 2017-07-19 13:30 +0200

#1691193 — [PATCH 0/5] FlexRM support in VFIO platform

This patchset primarily adds Broadcom FlexRM reset module for
VFIO platform driver. We also have minor improvments in IOMMU
and VFIO driver to allow VFIO no-IOMMU mode access to FlexRM.

The patches are based on Linux-4.13-rc1 and can also be
found at flexrm-vfio-v1 branch of
https://github.com/Broadcom/arm64-linux.git

Anup Patel (5):
  iommu: Add capability IOMMU_CAP_BYPASS
  iommu/arm-smmu: add IOMMU_CAP_BYPASS to the ARM SMMU driver
  iommu/arm-smmu-v3: add IOMMU_CAP_BYPASS to the ARM SMMUv3 driver
  vfio: Allow No-IOMMU mode for IOMMUs with bypass capability
  vfio: platform: reset: Add Broadcom FlexRM reset module

 drivers/iommu/arm-smmu-v3.c                        |  2 +
 drivers/iommu/arm-smmu.c                           |  2 +
 drivers/vfio/platform/reset/Kconfig                |  9 +++
 drivers/vfio/platform/reset/Makefile               |  1 +
 .../vfio/platform/reset/vfio_platform_bcmflexrm.c  | 91 ++++++++++++++++++++++
 drivers/vfio/vfio.c                                | 13 +++-
 include/linux/iommu.h                              |  4 +
 7 files changed, 119 insertions(+), 3 deletions(-)
 create mode 100644 drivers/vfio/platform/reset/vfio_platform_bcmflexrm.c

-- 
2.7.4

[toc] | [next] | [standalone]

#1691197 — [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS

Some of the IOMMUs (such as ARM SMMU) are capable of bypassing
transactions for which no IOMMU domain is configured.

This patch adds IOMMU_CAP_BYPASS which can be used by IOMMU
drivers to advertise transation bypass capability of an IOMMU.

Signed-off-by: Anup Patel <anup.patel@broadcom.com>
---
 include/linux/iommu.h | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/include/linux/iommu.h b/include/linux/iommu.h
index 2cb54ad..6bbb4cc 100644
--- a/include/linux/iommu.h
+++ b/include/linux/iommu.h
@@ -101,6 +101,10 @@ enum iommu_cap {
 					   transactions */
 	IOMMU_CAP_INTR_REMAP,		/* IOMMU supports interrupt isolation */
 	IOMMU_CAP_NOEXEC,		/* IOMMU_NOEXEC flag */
+	IOMMU_CAP_BYPASS,		/*
+					 * IOMMU can bypass transactions for
+					 * which domain is not configured
+					 */
 };
 
 /*
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1691470 — Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS

On 19/07/17 10:33, Anup Patel wrote:
> Some of the IOMMUs (such as ARM SMMU) are capable of bypassing
> transactions for which no IOMMU domain is configured.
> 
> This patch adds IOMMU_CAP_BYPASS which can be used by IOMMU
> drivers to advertise transation bypass capability of an IOMMU.

Whatever the intended semantics of this are, I can't help thinking it
would be better served by allowing callers to explicitly allocate their
own IOMMU_DOMAIN_IDENTITY domains. That would also be useful for the
problem we have with legacy virtio devices behind real IOMMUs.

Robin.

> Signed-off-by: Anup Patel <anup.patel@broadcom.com>
> ---
>  include/linux/iommu.h | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/include/linux/iommu.h b/include/linux/iommu.h
> index 2cb54ad..6bbb4cc 100644
> --- a/include/linux/iommu.h
> +++ b/include/linux/iommu.h
> @@ -101,6 +101,10 @@ enum iommu_cap {
>  					   transactions */
>  	IOMMU_CAP_INTR_REMAP,		/* IOMMU supports interrupt isolation */
>  	IOMMU_CAP_NOEXEC,		/* IOMMU_NOEXEC flag */
> +	IOMMU_CAP_BYPASS,		/*
> +					 * IOMMU can bypass transactions for
> +					 * which domain is not configured
> +					 */
>  };
>  
>  /*
> 

[toc] | [prev] | [next] | [standalone]

#1691489 — Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS

On Wed, Jul 19, 2017 at 4:28 PM, Robin Murphy <robin.murphy@arm.com> wrote:
> On 19/07/17 10:33, Anup Patel wrote:
>> Some of the IOMMUs (such as ARM SMMU) are capable of bypassing
>> transactions for which no IOMMU domain is configured.
>>
>> This patch adds IOMMU_CAP_BYPASS which can be used by IOMMU
>> drivers to advertise transation bypass capability of an IOMMU.
>
> Whatever the intended semantics of this are, I can't help thinking it
> would be better served by allowing callers to explicitly allocate their
> own IOMMU_DOMAIN_IDENTITY domains. That would also be useful for the
> problem we have with legacy virtio devices behind real IOMMUs.

We want to use VFIO no-IOMMU mode for FlexRM device but
currently it does not allow on our SOC because IOMMU ops are
registered for platform bus.

Regards,
Anup

[toc] | [prev] | [next] | [standalone]

#1691524 — Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS

On 19/07/17 12:26, Anup Patel wrote:
> On Wed, Jul 19, 2017 at 4:53 PM, Will Deacon <will.deacon@arm.com> wrote:
>> On Wed, Jul 19, 2017 at 04:49:00PM +0530, Anup Patel wrote:
>>> On Wed, Jul 19, 2017 at 4:28 PM, Robin Murphy <robin.murphy@arm.com> wrote:
>>>> On 19/07/17 10:33, Anup Patel wrote:
>>>>> Some of the IOMMUs (such as ARM SMMU) are capable of bypassing
>>>>> transactions for which no IOMMU domain is configured.
>>>>>
>>>>> This patch adds IOMMU_CAP_BYPASS which can be used by IOMMU
>>>>> drivers to advertise transation bypass capability of an IOMMU.
>>>>
>>>> Whatever the intended semantics of this are, I can't help thinking it
>>>> would be better served by allowing callers to explicitly allocate their
>>>> own IOMMU_DOMAIN_IDENTITY domains. That would also be useful for the
>>>> problem we have with legacy virtio devices behind real IOMMUs.
>>>
>>> We want to use VFIO no-IOMMU mode for FlexRM device but
>>> currently it does not allow on our SOC because IOMMU ops are
>>> registered for platform bus.
>>
>> Why do you want to use no-IOMMU mode if you have an IOMMU, and why you do
>> think the individual IOMMU drivers are the place to implement this?
>>
>> NAK to the SMMU patches, for the reasons outlined by Robin.
> 
> We have limited number of SMRs on our SOC.
> 
> There are lot of devices for which we can potentially
> configure SMMU but then due to limited number of
> SMRs so we use SMMU only for certain devices.

Is the stream ID allocation so whacked out that you can't use masking?

Robin.
> For FlexRM device on our SOC, we don't intend to
> use SMMU hence we need VFIO no-IOMMU mode
> working for FlexRM device on our SOC.
> 
> Please re-consider your NAK.
> 
> Regards,
> Anup
> 

[toc] | [prev] | [next] | [standalone]

#1691545 — Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS

On Wed, Jul 19, 2017 at 04:49:00PM +0530, Anup Patel wrote:
> On Wed, Jul 19, 2017 at 4:28 PM, Robin Murphy <robin.murphy@arm.com> wrote:
> > On 19/07/17 10:33, Anup Patel wrote:
> >> Some of the IOMMUs (such as ARM SMMU) are capable of bypassing
> >> transactions for which no IOMMU domain is configured.
> >>
> >> This patch adds IOMMU_CAP_BYPASS which can be used by IOMMU
> >> drivers to advertise transation bypass capability of an IOMMU.
> >
> > Whatever the intended semantics of this are, I can't help thinking it
> > would be better served by allowing callers to explicitly allocate their
> > own IOMMU_DOMAIN_IDENTITY domains. That would also be useful for the
> > problem we have with legacy virtio devices behind real IOMMUs.
> 
> We want to use VFIO no-IOMMU mode for FlexRM device but
> currently it does not allow on our SOC because IOMMU ops are
> registered for platform bus.

Why do you want to use no-IOMMU mode if you have an IOMMU, and why you do
think the individual IOMMU drivers are the place to implement this?

NAK to the SMMU patches, for the reasons outlined by Robin.

Will

[toc] | [prev] | [next] | [standalone]

#1691549 — Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS

On Wed, Jul 19, 2017 at 4:53 PM, Will Deacon <will.deacon@arm.com> wrote:
> On Wed, Jul 19, 2017 at 04:49:00PM +0530, Anup Patel wrote:
>> On Wed, Jul 19, 2017 at 4:28 PM, Robin Murphy <robin.murphy@arm.com> wrote:
>> > On 19/07/17 10:33, Anup Patel wrote:
>> >> Some of the IOMMUs (such as ARM SMMU) are capable of bypassing
>> >> transactions for which no IOMMU domain is configured.
>> >>
>> >> This patch adds IOMMU_CAP_BYPASS which can be used by IOMMU
>> >> drivers to advertise transation bypass capability of an IOMMU.
>> >
>> > Whatever the intended semantics of this are, I can't help thinking it
>> > would be better served by allowing callers to explicitly allocate their
>> > own IOMMU_DOMAIN_IDENTITY domains. That would also be useful for the
>> > problem we have with legacy virtio devices behind real IOMMUs.
>>
>> We want to use VFIO no-IOMMU mode for FlexRM device but
>> currently it does not allow on our SOC because IOMMU ops are
>> registered for platform bus.
>
> Why do you want to use no-IOMMU mode if you have an IOMMU, and why you do
> think the individual IOMMU drivers are the place to implement this?
>
> NAK to the SMMU patches, for the reasons outlined by Robin.

We have limited number of SMRs on our SOC.

There are lot of devices for which we can potentially
configure SMMU but then due to limited number of
SMRs so we use SMMU only for certain devices.

For FlexRM device on our SOC, we don't intend to
use SMMU hence we need VFIO no-IOMMU mode
working for FlexRM device on our SOC.

Please re-consider your NAK.

Regards,
Anup

[toc] | [prev] | [next] | [standalone]

#1691567 — Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS

On Wed, Jul 19, 2017 at 5:00 PM, Will Deacon <will.deacon@arm.com> wrote:
> On Wed, Jul 19, 2017 at 04:56:38PM +0530, Anup Patel wrote:
>> On Wed, Jul 19, 2017 at 4:53 PM, Will Deacon <will.deacon@arm.com> wrote:
>> > On Wed, Jul 19, 2017 at 04:49:00PM +0530, Anup Patel wrote:
>> >> On Wed, Jul 19, 2017 at 4:28 PM, Robin Murphy <robin.murphy@arm.com> wrote:
>> >> > On 19/07/17 10:33, Anup Patel wrote:
>> >> >> Some of the IOMMUs (such as ARM SMMU) are capable of bypassing
>> >> >> transactions for which no IOMMU domain is configured.
>> >> >>
>> >> >> This patch adds IOMMU_CAP_BYPASS which can be used by IOMMU
>> >> >> drivers to advertise transation bypass capability of an IOMMU.
>> >> >
>> >> > Whatever the intended semantics of this are, I can't help thinking it
>> >> > would be better served by allowing callers to explicitly allocate their
>> >> > own IOMMU_DOMAIN_IDENTITY domains. That would also be useful for the
>> >> > problem we have with legacy virtio devices behind real IOMMUs.
>> >>
>> >> We want to use VFIO no-IOMMU mode for FlexRM device but
>> >> currently it does not allow on our SOC because IOMMU ops are
>> >> registered for platform bus.
>> >
>> > Why do you want to use no-IOMMU mode if you have an IOMMU, and why you do
>> > think the individual IOMMU drivers are the place to implement this?
>> >
>> > NAK to the SMMU patches, for the reasons outlined by Robin.
>>
>> We have limited number of SMRs on our SOC.
>>
>> There are lot of devices for which we can potentially
>> configure SMMU but then due to limited number of
>> SMRs so we use SMMU only for certain devices.
>>
>> For FlexRM device on our SOC, we don't intend to
>> use SMMU hence we need VFIO no-IOMMU mode
>> working for FlexRM device on our SOC.
>>
>> Please re-consider your NAK.
>
> I'm afraid it still stands for the current implementation. If you can't
> solve the SMR restriction by grouping things appropriately (which would be
> my strong preference), then I think you'll have to follow-up on Robin's
> suggestion of implementing support for IDENTITY domains in VFIO for no-IOMMU
> mode to be used even when an IOMMU is present.
>

Yes, we have considered making stream-id space for
devices to be continuguous but this cannot be fixed
in our current SOC. We will have to live with this
limitation for our current SOC and it will be only
fixed in future.

Regards,
Anup

[toc] | [prev] | [next] | [standalone]

#1691578 — Re: [PATCH 1/5] iommu: Add capability IOMMU_CAP_BYPASS

On Wed, Jul 19, 2017 at 04:56:38PM +0530, Anup Patel wrote:
> On Wed, Jul 19, 2017 at 4:53 PM, Will Deacon <will.deacon@arm.com> wrote:
> > On Wed, Jul 19, 2017 at 04:49:00PM +0530, Anup Patel wrote:
> >> On Wed, Jul 19, 2017 at 4:28 PM, Robin Murphy <robin.murphy@arm.com> wrote:
> >> > On 19/07/17 10:33, Anup Patel wrote:
> >> >> Some of the IOMMUs (such as ARM SMMU) are capable of bypassing
> >> >> transactions for which no IOMMU domain is configured.
> >> >>
> >> >> This patch adds IOMMU_CAP_BYPASS which can be used by IOMMU
> >> >> drivers to advertise transation bypass capability of an IOMMU.
> >> >
> >> > Whatever the intended semantics of this are, I can't help thinking it
> >> > would be better served by allowing callers to explicitly allocate their
> >> > own IOMMU_DOMAIN_IDENTITY domains. That would also be useful for the
> >> > problem we have with legacy virtio devices behind real IOMMUs.
> >>
> >> We want to use VFIO no-IOMMU mode for FlexRM device but
> >> currently it does not allow on our SOC because IOMMU ops are
> >> registered for platform bus.
> >
> > Why do you want to use no-IOMMU mode if you have an IOMMU, and why you do
> > think the individual IOMMU drivers are the place to implement this?
> >
> > NAK to the SMMU patches, for the reasons outlined by Robin.
> 
> We have limited number of SMRs on our SOC.
> 
> There are lot of devices for which we can potentially
> configure SMMU but then due to limited number of
> SMRs so we use SMMU only for certain devices.
> 
> For FlexRM device on our SOC, we don't intend to
> use SMMU hence we need VFIO no-IOMMU mode
> working for FlexRM device on our SOC.
> 
> Please re-consider your NAK.

I'm afraid it still stands for the current implementation. If you can't
solve the SMR restriction by grouping things appropriately (which would be
my strong preference), then I think you'll have to follow-up on Robin's
suggestion of implementing support for IDENTITY domains in VFIO for no-IOMMU
mode to be used even when an IOMMU is present.

Will

[toc] | [prev] | [next] | [standalone]

#1691205 — [PATCH 2/5] iommu/arm-smmu: add IOMMU_CAP_BYPASS to the ARM SMMU driver

The ARM SMMUv1 and SMMUv2 support bypassing transactions for
which domain is not configured. The patch adds corresponding
IOMMU capability to advertise this fact.

Signed-off-by: Anup Patel <anup.patel@broadcom.com>
---
 drivers/iommu/arm-smmu.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/iommu/arm-smmu.c b/drivers/iommu/arm-smmu.c
index bc89b4d..08a9020 100644
--- a/drivers/iommu/arm-smmu.c
+++ b/drivers/iommu/arm-smmu.c
@@ -1483,6 +1483,8 @@ static bool arm_smmu_capable(enum iommu_cap cap)
 		return true;
 	case IOMMU_CAP_NOEXEC:
 		return true;
+	case IOMMU_CAP_BYPASS:
+		return true;
 	default:
 		return false;
 	}
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1691469 — Re: [PATCH 2/5] iommu/arm-smmu: add IOMMU_CAP_BYPASS to the ARM SMMU driver

On 19/07/17 10:33, Anup Patel wrote:
> The ARM SMMUv1 and SMMUv2 support bypassing transactions for
> which domain is not configured. The patch adds corresponding
> IOMMU capability to advertise this fact.
> 
> Signed-off-by: Anup Patel <anup.patel@broadcom.com>
> ---
>  drivers/iommu/arm-smmu.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/iommu/arm-smmu.c b/drivers/iommu/arm-smmu.c
> index bc89b4d..08a9020 100644
> --- a/drivers/iommu/arm-smmu.c
> +++ b/drivers/iommu/arm-smmu.c
> @@ -1483,6 +1483,8 @@ static bool arm_smmu_capable(enum iommu_cap cap)
>  		return true;
>  	case IOMMU_CAP_NOEXEC:
>  		return true;
> +	case IOMMU_CAP_BYPASS:
> +		return true;

Except when it isn't, of course.

Beware that what we'd like to do in the long term is flip the polarity
of disable_bypass, because there isn't generally a good reason for
Non-Secure DMA to be happening behind Linux's back.

Robin.

>  	default:
>  		return false;
>  	}
> 

[toc] | [prev] | [next] | [standalone]

#1691467

On 19/07/17 10:33, Anup Patel wrote:
> This patchset primarily adds Broadcom FlexRM reset module for
> VFIO platform driver. We also have minor improvments in IOMMU
> and VFIO driver to allow VFIO no-IOMMU mode access to FlexRM.

I'm struggling to understand the IOMMU changes here - what's the
FlexRM's hardware relationship with the IOMMU, and how is it different
from any other device? Furthermore, if there *is* a relevant IOMMU
present, why would no-IOMMU mode need to be involved at all?

Robin.

> The patches are based on Linux-4.13-rc1 and can also be
> found at flexrm-vfio-v1 branch of
> https://github.com/Broadcom/arm64-linux.git
> 
> Anup Patel (5):
>   iommu: Add capability IOMMU_CAP_BYPASS
>   iommu/arm-smmu: add IOMMU_CAP_BYPASS to the ARM SMMU driver
>   iommu/arm-smmu-v3: add IOMMU_CAP_BYPASS to the ARM SMMUv3 driver
>   vfio: Allow No-IOMMU mode for IOMMUs with bypass capability
>   vfio: platform: reset: Add Broadcom FlexRM reset module
> 
>  drivers/iommu/arm-smmu-v3.c                        |  2 +
>  drivers/iommu/arm-smmu.c                           |  2 +
>  drivers/vfio/platform/reset/Kconfig                |  9 +++
>  drivers/vfio/platform/reset/Makefile               |  1 +
>  .../vfio/platform/reset/vfio_platform_bcmflexrm.c  | 91 ++++++++++++++++++++++
>  drivers/vfio/vfio.c                                | 13 +++-
>  include/linux/iommu.h                              |  4 +
>  7 files changed, 119 insertions(+), 3 deletions(-)
>  create mode 100644 drivers/vfio/platform/reset/vfio_platform_bcmflexrm.c
> 

[toc] | [prev] | [next] | [standalone]

#1691481

On Wed, Jul 19, 2017 at 4:27 PM, Robin Murphy <robin.murphy@arm.com> wrote:
> On 19/07/17 10:33, Anup Patel wrote:
>> This patchset primarily adds Broadcom FlexRM reset module for
>> VFIO platform driver. We also have minor improvments in IOMMU
>> and VFIO driver to allow VFIO no-IOMMU mode access to FlexRM.
>
> I'm struggling to understand the IOMMU changes here - what's the
> FlexRM's hardware relationship with the IOMMU, and how is it different
> from any other device? Furthermore, if there *is* a relevant IOMMU
> present, why would no-IOMMU mode need to be involved at all?

We want to have FlexRM device accessible from user-space
using VFIO platform with and without IOMMU.

Currently, if IOMMU ops are available for platform bus then
I cannot access FlexRM device using VFIO no-IOMMU mode.

Since, SMMU can bypass transactions which do not match
any SMRs, we should allow no-IOMMU mode for devices
when IOMMU group is not available.

Regards,
Anup

[toc] | [prev] | [next] | [standalone]

#1691526

On 19/07/17 12:17, Anup Patel wrote:
> On Wed, Jul 19, 2017 at 4:27 PM, Robin Murphy <robin.murphy@arm.com> wrote:
>> On 19/07/17 10:33, Anup Patel wrote:
>>> This patchset primarily adds Broadcom FlexRM reset module for
>>> VFIO platform driver. We also have minor improvments in IOMMU
>>> and VFIO driver to allow VFIO no-IOMMU mode access to FlexRM.
>>
>> I'm struggling to understand the IOMMU changes here - what's the
>> FlexRM's hardware relationship with the IOMMU, and how is it different
>> from any other device? Furthermore, if there *is* a relevant IOMMU
>> present, why would no-IOMMU mode need to be involved at all?
> 
> We want to have FlexRM device accessible from user-space
> using VFIO platform with and without IOMMU.
> 
> Currently, if IOMMU ops are available for platform bus then
> I cannot access FlexRM device using VFIO no-IOMMU mode.

So does the FlexRM hardware master through the SMMU or not? If it does,
why do you need no-IOMMU mode? If it doesn't, then that's yet another
reason to fix the real problem, which is the utterly broken notion of
there being 'an IOMMU' on the platform 'bus', rather than papering over
it in VFIO.

Robin.

> Since, SMMU can bypass transactions which do not match
> any SMRs, we should allow no-IOMMU mode for devices
> when IOMMU group is not available.
> 
> Regards,
> Anup
> 

[toc] | [prev] | [next] | [standalone]

#1691541

On Wed, Jul 19, 2017 at 4:55 PM, Robin Murphy <robin.murphy@arm.com> wrote:
> On 19/07/17 12:17, Anup Patel wrote:
>> On Wed, Jul 19, 2017 at 4:27 PM, Robin Murphy <robin.murphy@arm.com> wrote:
>>> On 19/07/17 10:33, Anup Patel wrote:
>>>> This patchset primarily adds Broadcom FlexRM reset module for
>>>> VFIO platform driver. We also have minor improvments in IOMMU
>>>> and VFIO driver to allow VFIO no-IOMMU mode access to FlexRM.
>>>
>>> I'm struggling to understand the IOMMU changes here - what's the
>>> FlexRM's hardware relationship with the IOMMU, and how is it different
>>> from any other device? Furthermore, if there *is* a relevant IOMMU
>>> present, why would no-IOMMU mode need to be involved at all?
>>
>> We want to have FlexRM device accessible from user-space
>> using VFIO platform with and without IOMMU.
>>
>> Currently, if IOMMU ops are available for platform bus then
>> I cannot access FlexRM device using VFIO no-IOMMU mode.
>
> So does the FlexRM hardware master through the SMMU or not? If it does,
> why do you need no-IOMMU mode? If it doesn't, then that's yet another
> reason to fix the real problem, which is the utterly broken notion of
> there being 'an IOMMU' on the platform 'bus', rather than papering over
> it in VFIO.
>

We are not trying to paper-over the issue. The ARM SMMU will
have limited number of SMRs so on a big SOC with large number
of DMA-capable devices we can run-out of SMRs if we try to
configure SMMU for all DMA-capable devices.

Regards,
Anup

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web