[PATCH 0/3] drm/etnaviv: IOMMU related fixes

[PATCH 0/3] drm/etnaviv: IOMMU related fixes

[Michael Walle]

This patch series fixes usage of the etnaviv driver with GPUs behind a
IOMMU. It was tested on a NXP LS1028A SoC. Together with Lucas' MMU patches
[1] there are not more (GPU internal) MMU nor (system) IOMMU faults on the
LS1028A.

[1] https://lists.freedesktop.org/archives/etnaviv/2021-August/003682.html

Michael Walle (3):
  drm/etnaviv: use PLATFORM_DEVID_NONE
  drm/etnaviv: fix dma configuration of the virtual device
  drm/etnaviv: use a 32 bit mask as coherent DMA mask

 drivers/gpu/drm/etnaviv/etnaviv_drv.c | 41 ++++++++++++++++++++-------
 1 file changed, 31 insertions(+), 10 deletions(-)

-- 
2.30.2

[PATCH 1/3] drm/etnaviv: use PLATFORM_DEVID_NONE

[Michael Walle]

There is already a macro for the magic value. Use it.

Signed-off-by: Michael Walle <michael@walle.cc>
---
 drivers/gpu/drm/etnaviv/etnaviv_drv.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
index 7dcc6392792d..2509b3e85709 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
@@ -653,7 +653,7 @@ static int __init etnaviv_init(void)
 		if (!of_device_is_available(np))
 			continue;
 
-		pdev = platform_device_alloc("etnaviv", -1);
+		pdev = platform_device_alloc("etnaviv", PLATFORM_DEVID_NONE);
 		if (!pdev) {
 			ret = -ENOMEM;
 			of_node_put(np);
-- 
2.30.2

[PATCH 2/3] drm/etnaviv: fix dma configuration of the virtual device

[Michael Walle]

The DMA configuration of the virtual device is inherited from the first
actual etnaviv device. Unfortunately, this doesn't work with an IOMMU:

[    5.191008] Failed to set up IOMMU for device (null); retaining platform DMA ops

This is because there is no associated iommu_group with the device. The
group is set in iommu_group_add_device() which is eventually called by
device_add() via the platform bus:
  device_add()
    blocking_notifier_call_chain()
      iommu_bus_notifier()
        iommu_probe_device()
          __iommu_probe_device()
            iommu_group_get_for_dev()
              iommu_group_add_device()

Move of_dma_configure() into the probe function, which is called after
device_add(). Normally, the platform code will already call it itself
if .of_node is set. Unfortunately, this isn't the case here.

Also move the dma mask assignemnts to probe() to keep all DMA related
settings together.

Signed-off-by: Michael Walle <michael@walle.cc>
---
 drivers/gpu/drm/etnaviv/etnaviv_drv.c | 24 +++++++++++++++---------
 1 file changed, 15 insertions(+), 9 deletions(-)

diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
index 2509b3e85709..ff6425f6ebad 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
@@ -589,6 +589,7 @@ static int compare_str(struct device *dev, void *data)
 static int etnaviv_pdev_probe(struct platform_device *pdev)
 {
 	struct device *dev = &pdev->dev;
+	struct device_node *first_node = NULL;
 	struct component_match *match = NULL;
 
 	if (!dev->platform_data) {
@@ -598,6 +599,9 @@ static int etnaviv_pdev_probe(struct platform_device *pdev)
 			if (!of_device_is_available(core_node))
 				continue;
 
+			if (!first_node)
+				first_node = core_node;
+
 			drm_of_component_match_add(&pdev->dev, &match,
 						   compare_of, core_node);
 		}
@@ -609,6 +613,17 @@ static int etnaviv_pdev_probe(struct platform_device *pdev)
 			component_match_add(dev, &match, compare_str, names[i]);
 	}
 
+	pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
+	pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
+
+	/*
+	 * Apply the same DMA configuration to the virtual etnaviv
+	 * device as the GPU we found. This assumes that all Vivante
+	 * GPUs in the system share the same DMA constraints.
+	 */
+	if (first_node)
+		of_dma_configure(&pdev->dev, first_node, true);
+
 	return component_master_add_with_match(dev, &etnaviv_master_ops, match);
 }
 
@@ -659,15 +674,6 @@ static int __init etnaviv_init(void)
 			of_node_put(np);
 			goto unregister_platform_driver;
 		}
-		pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
-		pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
-
-		/*
-		 * Apply the same DMA configuration to the virtual etnaviv
-		 * device as the GPU we found. This assumes that all Vivante
-		 * GPUs in the system share the same DMA constraints.
-		 */
-		of_dma_configure(&pdev->dev, np, true);
 
 		ret = platform_device_add(pdev);
 		if (ret) {
-- 
2.30.2

[PATCH 3/3] drm/etnaviv: use a 32 bit mask as coherent DMA mask

[Michael Walle]

The STLB and the first command buffer (which is used to set up the TLBs)
has a 32 bit size restriction in hardware. There seems to be no way to
specify addresses larger than 32 bit. Keep it simple and restict the
addresses to the lower 4 GiB range for all coherent DMA memory
allocations.

Signed-off-by: Michael Walle <michael@walle.cc>
---
 drivers/gpu/drm/etnaviv/etnaviv_drv.c | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
index ff6425f6ebad..0b756ecb1bc2 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
@@ -613,8 +613,23 @@ static int etnaviv_pdev_probe(struct platform_device *pdev)
 			component_match_add(dev, &match, compare_str, names[i]);
 	}
 
-	pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
-	pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
+	/*
+	 * PTA and MTLB can have 40 bit base addresses, but
+	 * unfortunately, an entry in the MTLB can only point to a
+	 * 32 bit base address of a STLB. Moreover, to initialize the
+	 * MMU we need a command buffer with a 32 bit address because
+	 * without an MMU there is only an indentity mapping between
+	 * the internal 32 bit addresses and the bus addresses.
+	 *
+	 * To make things easy, we set the dma_coherent_mask to 32
+	 * bit to make sure we are allocating the command buffers and
+	 * TLBs in the lower 4 GiB address space.
+	 */
+	if (dma_set_mask(&pdev->dev, DMA_BIT_MASK(40)) ||
+	    dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(32))) {
+		dev_dbg(&pdev->dev, "No suitable DMA available\n");
+		return -ENODEV;
+	}
 
 	/*
 	 * Apply the same DMA configuration to the virtual etnaviv
-- 
2.30.2

Re: [PATCH 2/3] drm/etnaviv: fix dma configuration of the virtual device

[Russell King (Oracle)]

On Thu, Aug 26, 2021 at 02:10:05PM +0200, Michael Walle wrote:
> +	pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
> +	pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;

Please use dma_coerce_mask_and_coherent() here instead.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!

Re: [PATCH 2/3] drm/etnaviv: fix dma configuration of the virtual device

[Michael Walle]

Am 2021-08-26 14:14, schrieb Russell King (Oracle):
> On Thu, Aug 26, 2021 at 02:10:05PM +0200, Michael Walle wrote:
>> +	pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
>> +	pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
> 
> Please use dma_coerce_mask_and_coherent() here instead.

It will be removed altogether in patch 3/3. I just moved the
current code around here. I could drop the code moving in 2/3. I
wasn't sure whats best.

-michael

Re: [PATCH 3/3] drm/etnaviv: use a 32 bit mask as coherent DMA mask

[Russell King (Oracle)]

On Thu, Aug 26, 2021 at 02:10:06PM +0200, Michael Walle wrote:
> -	pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
> -	pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
> +	/*
> +	 * PTA and MTLB can have 40 bit base addresses, but
> +	 * unfortunately, an entry in the MTLB can only point to a
> +	 * 32 bit base address of a STLB. Moreover, to initialize the
> +	 * MMU we need a command buffer with a 32 bit address because
> +	 * without an MMU there is only an indentity mapping between
> +	 * the internal 32 bit addresses and the bus addresses.
> +	 *
> +	 * To make things easy, we set the dma_coherent_mask to 32
> +	 * bit to make sure we are allocating the command buffers and
> +	 * TLBs in the lower 4 GiB address space.
> +	 */
> +	if (dma_set_mask(&pdev->dev, DMA_BIT_MASK(40)) ||
> +	    dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(32))) {
> +		dev_dbg(&pdev->dev, "No suitable DMA available\n");
> +		return -ENODEV;
> +	}

This makes no sense. In the previous patch, you initialised
pdev->dev.dma_mask ot point at the coherent mask, implying that
it wasn't already set - for which dma_coerce_mask_and_coherent()
should be used. Now you're just calling dma_set_mask(), which will
fail if pdev->dev.dma_mask hasn't already been set to point at
something.

If it's already been initialised to point at something, then you
shouldn't be overwriting it in the driver, and you should've used
dma_set_mask_and_coherent() in your previous patch.

Confused.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!

Re: [PATCH 3/3] drm/etnaviv: use a 32 bit mask as coherent DMA mask

[Michael Walle]

Am 2021-08-26 14:19, schrieb Russell King (Oracle):
> On Thu, Aug 26, 2021 at 02:10:06PM +0200, Michael Walle wrote:
>> -	pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
>> -	pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
>> +	/*
>> +	 * PTA and MTLB can have 40 bit base addresses, but
>> +	 * unfortunately, an entry in the MTLB can only point to a
>> +	 * 32 bit base address of a STLB. Moreover, to initialize the
>> +	 * MMU we need a command buffer with a 32 bit address because
>> +	 * without an MMU there is only an indentity mapping between
>> +	 * the internal 32 bit addresses and the bus addresses.
>> +	 *
>> +	 * To make things easy, we set the dma_coherent_mask to 32
>> +	 * bit to make sure we are allocating the command buffers and
>> +	 * TLBs in the lower 4 GiB address space.
>> +	 */
>> +	if (dma_set_mask(&pdev->dev, DMA_BIT_MASK(40)) ||
>> +	    dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(32))) {
>> +		dev_dbg(&pdev->dev, "No suitable DMA available\n");
>> +		return -ENODEV;
>> +	}
> 
> This makes no sense. In the previous patch, you initialised
> pdev->dev.dma_mask ot point at the coherent mask, implying that
> it wasn't already set - for which dma_coerce_mask_and_coherent()
> should be used. Now you're just calling dma_set_mask(), which will
> fail if pdev->dev.dma_mask hasn't already been set to point at
> something.
> 
> If it's already been initialised to point at something, then you
> shouldn't be overwriting it in the driver, and you should've used
> dma_set_mask_and_coherent() in your previous patch.
> 
> Confused.

Mh, I see that moving these two lines was a bad idea. See commit
message in patch 2/3:
> Also move the dma mask assignemnts to probe() to keep all DMA related
> settings together.

The actual fix in patch 2/3 is the move of the of_dma_configure()
not the dma_mask assignments.

-michael

Re: [PATCH 3/3] drm/etnaviv: use a 32 bit mask as coherent DMA mask

[Michael Walle]

Am 2021-08-26 14:10, schrieb Michael Walle:
> The STLB and the first command buffer (which is used to set up the 
> TLBs)
> has a 32 bit size restriction in hardware. There seems to be no way to
> specify addresses larger than 32 bit. Keep it simple and restict the
> addresses to the lower 4 GiB range for all coherent DMA memory
> allocations.
> 
> Signed-off-by: Michael Walle <michael@walle.cc>

Suggested-by: Lucas Stach <l.stach@pengutronix.de>

is missing here. sorry, will add it in the next version.

-michael

Re: [PATCH 1/3] drm/etnaviv: use PLATFORM_DEVID_NONE

[Christian Gmeiner]

Am Do., 26. Aug. 2021 um 14:10 Uhr schrieb Michael Walle <michael@walle.cc>:
>
> There is already a macro for the magic value. Use it.
>
> Signed-off-by: Michael Walle <michael@walle.cc>

Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>

I will wait for v2 for the rest of the changes to review.

> ---
>  drivers/gpu/drm/etnaviv/etnaviv_drv.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> index 7dcc6392792d..2509b3e85709 100644
> --- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> +++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> @@ -653,7 +653,7 @@ static int __init etnaviv_init(void)
>                 if (!of_device_is_available(np))
>                         continue;
>
> -               pdev = platform_device_alloc("etnaviv", -1);
> +               pdev = platform_device_alloc("etnaviv", PLATFORM_DEVID_NONE);
>                 if (!pdev) {
>                         ret = -ENOMEM;
>                         of_node_put(np);
> --
> 2.30.2
>


-- 
greets
--
Christian Gmeiner, MSc

https://christian-gmeiner.info/privacypolicy

Re: [PATCH 1/3] drm/etnaviv: use PLATFORM_DEVID_NONE

[Christian Gmeiner]

Am Do., 26. Aug. 2021 um 14:10 Uhr schrieb Michael Walle <michael@walle.cc>:
>
> There is already a macro for the magic value. Use it.
>
> Signed-off-by: Michael Walle <michael@walle.cc>

Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>

I will wait for v2 for the rest of the changes to review.

> ---
>  drivers/gpu/drm/etnaviv/etnaviv_drv.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> index 7dcc6392792d..2509b3e85709 100644
> --- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> +++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> @@ -653,7 +653,7 @@ static int __init etnaviv_init(void)
>                 if (!of_device_is_available(np))
>                         continue;
>
> -               pdev = platform_device_alloc("etnaviv", -1);
> +               pdev = platform_device_alloc("etnaviv", PLATFORM_DEVID_NONE);
>                 if (!pdev) {
>                         ret = -ENOMEM;
>                         of_node_put(np);
> --
> 2.30.2
>


-- 
greets
--
Christian Gmeiner, MSc

https://christian-gmeiner.info/privacypolicy

Re: [PATCH 2/3] drm/etnaviv: fix dma configuration of the virtual device

[Robin Murphy]

On 2021-08-26 13:10, Michael Walle wrote:
> The DMA configuration of the virtual device is inherited from the first
> actual etnaviv device. Unfortunately, this doesn't work with an IOMMU:
> 
> [    5.191008] Failed to set up IOMMU for device (null); retaining platform DMA ops
> 
> This is because there is no associated iommu_group with the device. The
> group is set in iommu_group_add_device() which is eventually called by
> device_add() via the platform bus:
>    device_add()
>      blocking_notifier_call_chain()
>        iommu_bus_notifier()
>          iommu_probe_device()
>            __iommu_probe_device()
>              iommu_group_get_for_dev()
>                iommu_group_add_device()
> 
> Move of_dma_configure() into the probe function, which is called after
> device_add(). Normally, the platform code will already call it itself
> if .of_node is set. Unfortunately, this isn't the case here.
> 
> Also move the dma mask assignemnts to probe() to keep all DMA related
> settings together.

I assume the driver must already keep track of the real GPU platform 
device in order to map registers, request interrupts, etc. correctly - 
can't it also correctly use that device for DMA API calls and avoid the 
need for these shenanigans altogether?

FYI, IOMMU configuration is really supposed to *only* run at 
add_device() time as above - the fact that it's currently hooked in to 
be retriggered by of_dma_configure() on DT platforms actually turns out 
to lead to various issues within the IOMMU API, and the plan to change 
that is slowly climbing up my to-do list.

Robin.

> Signed-off-by: Michael Walle <michael@walle.cc>
> ---
>   drivers/gpu/drm/etnaviv/etnaviv_drv.c | 24 +++++++++++++++---------
>   1 file changed, 15 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> index 2509b3e85709..ff6425f6ebad 100644
> --- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> +++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> @@ -589,6 +589,7 @@ static int compare_str(struct device *dev, void *data)
>   static int etnaviv_pdev_probe(struct platform_device *pdev)
>   {
>   	struct device *dev = &pdev->dev;
> +	struct device_node *first_node = NULL;
>   	struct component_match *match = NULL;
>   
>   	if (!dev->platform_data) {
> @@ -598,6 +599,9 @@ static int etnaviv_pdev_probe(struct platform_device *pdev)
>   			if (!of_device_is_available(core_node))
>   				continue;
>   
> +			if (!first_node)
> +				first_node = core_node;
> +
>   			drm_of_component_match_add(&pdev->dev, &match,
>   						   compare_of, core_node);
>   		}
> @@ -609,6 +613,17 @@ static int etnaviv_pdev_probe(struct platform_device *pdev)
>   			component_match_add(dev, &match, compare_str, names[i]);
>   	}
>   
> +	pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
> +	pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
> +
> +	/*
> +	 * Apply the same DMA configuration to the virtual etnaviv
> +	 * device as the GPU we found. This assumes that all Vivante
> +	 * GPUs in the system share the same DMA constraints.
> +	 */
> +	if (first_node)
> +		of_dma_configure(&pdev->dev, first_node, true);
> +
>   	return component_master_add_with_match(dev, &etnaviv_master_ops, match);
>   }
>   
> @@ -659,15 +674,6 @@ static int __init etnaviv_init(void)
>   			of_node_put(np);
>   			goto unregister_platform_driver;
>   		}
> -		pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
> -		pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
> -
> -		/*
> -		 * Apply the same DMA configuration to the virtual etnaviv
> -		 * device as the GPU we found. This assumes that all Vivante
> -		 * GPUs in the system share the same DMA constraints.
> -		 */
> -		of_dma_configure(&pdev->dev, np, true);
>   
>   		ret = platform_device_add(pdev);
>   		if (ret) {
> 

Re: [PATCH 2/3] drm/etnaviv: fix dma configuration of the virtual device

[Lucas Stach]

Am Donnerstag, dem 26.08.2021 um 16:00 +0100 schrieb Robin Murphy:
> On 2021-08-26 13:10, Michael Walle wrote:
> > The DMA configuration of the virtual device is inherited from the first
> > actual etnaviv device. Unfortunately, this doesn't work with an IOMMU:
> > 
> > [    5.191008] Failed to set up IOMMU for device (null); retaining platform DMA ops
> > 
> > This is because there is no associated iommu_group with the device. The
> > group is set in iommu_group_add_device() which is eventually called by
> > device_add() via the platform bus:
> >    device_add()
> >      blocking_notifier_call_chain()
> >        iommu_bus_notifier()
> >          iommu_probe_device()
> >            __iommu_probe_device()
> >              iommu_group_get_for_dev()
> >                iommu_group_add_device()
> > 
> > Move of_dma_configure() into the probe function, which is called after
> > device_add(). Normally, the platform code will already call it itself
> > if .of_node is set. Unfortunately, this isn't the case here.
> > 
> > Also move the dma mask assignemnts to probe() to keep all DMA related
> > settings together.
> 
> I assume the driver must already keep track of the real GPU platform 
> device in order to map registers, request interrupts, etc. correctly - 
> can't it also correctly use that device for DMA API calls and avoid the 
> need for these shenanigans altogether?
> 
Not without a bigger rework. There's still quite a bit of midlayer
issues in DRM, where dma-buf imports are dma-mapped and cached via the
virtual DRM device instead of the real GPU device. Also etnaviv is able
to coalesce multiple Vivante GPUs in a single system under one virtual
DRM device, which is used on i.MX6 where the 2D and 3D GPUs are
separate peripherals, but have the same DMA constraints.

Effectively we would need to handle N devices for the dma-mapping in a
lot of places instead of only dealing with the one virtual DRM device.
It would probably be the right thing to anyways, but it's not something
that can be changed short-term. I'm also not yet sure about the
performance implications, as we might run into some cache maintenance
bottlenecks if we dma synchronize buffers to multiple real device
instead of doing it a single time with the virtual DRM device. I know,
I know, this has a lot of assumptions baked in that could fall apart if
someone builds a SoC with multiple Vivante GPUs that have differing DMA
constraints, but up until now hardware designers have not been *that*
crazy, fortunately.

Regards,
Lucas

> FYI, IOMMU configuration is really supposed to *only* run at 
> add_device() time as above - the fact that it's currently hooked in to 
> be retriggered by of_dma_configure() on DT platforms actually turns out 
> to lead to various issues within the IOMMU API, and the plan to change 
> that is slowly climbing up my to-do list.
> 
> Robin.
> 
> > Signed-off-by: Michael Walle <michael@walle.cc>
> > ---
> >   drivers/gpu/drm/etnaviv/etnaviv_drv.c | 24 +++++++++++++++---------
> >   1 file changed, 15 insertions(+), 9 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> > index 2509b3e85709..ff6425f6ebad 100644
> > --- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> > +++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> > @@ -589,6 +589,7 @@ static int compare_str(struct device *dev, void *data)
> >   static int etnaviv_pdev_probe(struct platform_device *pdev)
> >   {
> >   	struct device *dev = &pdev->dev;
> > +	struct device_node *first_node = NULL;
> >   	struct component_match *match = NULL;
> >   
> >   	if (!dev->platform_data) {
> > @@ -598,6 +599,9 @@ static int etnaviv_pdev_probe(struct platform_device *pdev)
> >   			if (!of_device_is_available(core_node))
> >   				continue;
> >   
> > +			if (!first_node)
> > +				first_node = core_node;
> > +
> >   			drm_of_component_match_add(&pdev->dev, &match,
> >   						   compare_of, core_node);
> >   		}
> > @@ -609,6 +613,17 @@ static int etnaviv_pdev_probe(struct platform_device *pdev)
> >   			component_match_add(dev, &match, compare_str, names[i]);
> >   	}
> >   
> > +	pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
> > +	pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
> > +
> > +	/*
> > +	 * Apply the same DMA configuration to the virtual etnaviv
> > +	 * device as the GPU we found. This assumes that all Vivante
> > +	 * GPUs in the system share the same DMA constraints.
> > +	 */
> > +	if (first_node)
> > +		of_dma_configure(&pdev->dev, first_node, true);
> > +
> >   	return component_master_add_with_match(dev, &etnaviv_master_ops, match);
> >   }
> >   
> > @@ -659,15 +674,6 @@ static int __init etnaviv_init(void)
> >   			of_node_put(np);
> >   			goto unregister_platform_driver;
> >   		}
> > -		pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
> > -		pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
> > -
> > -		/*
> > -		 * Apply the same DMA configuration to the virtual etnaviv
> > -		 * device as the GPU we found. This assumes that all Vivante
> > -		 * GPUs in the system share the same DMA constraints.
> > -		 */
> > -		of_dma_configure(&pdev->dev, np, true);
> >   
> >   		ret = platform_device_add(pdev);
> >   		if (ret) {
> > 

Re: [PATCH 2/3] drm/etnaviv: fix dma configuration of the virtual device

[Lucas Stach]

Am Donnerstag, dem 26.08.2021 um 16:00 +0100 schrieb Robin Murphy:
> On 2021-08-26 13:10, Michael Walle wrote:
> > The DMA configuration of the virtual device is inherited from the first
> > actual etnaviv device. Unfortunately, this doesn't work with an IOMMU:
> > 
> > [    5.191008] Failed to set up IOMMU for device (null); retaining platform DMA ops
> > 
> > This is because there is no associated iommu_group with the device. The
> > group is set in iommu_group_add_device() which is eventually called by
> > device_add() via the platform bus:
> >    device_add()
> >      blocking_notifier_call_chain()
> >        iommu_bus_notifier()
> >          iommu_probe_device()
> >            __iommu_probe_device()
> >              iommu_group_get_for_dev()
> >                iommu_group_add_device()
> > 
> > Move of_dma_configure() into the probe function, which is called after
> > device_add(). Normally, the platform code will already call it itself
> > if .of_node is set. Unfortunately, this isn't the case here.
> > 
> > Also move the dma mask assignemnts to probe() to keep all DMA related
> > settings together.
> 
> I assume the driver must already keep track of the real GPU platform 
> device in order to map registers, request interrupts, etc. correctly - 
> can't it also correctly use that device for DMA API calls and avoid the 
> need for these shenanigans altogether?
> 
Not without a bigger rework. There's still quite a bit of midlayer
issues in DRM, where dma-buf imports are dma-mapped and cached via the
virtual DRM device instead of the real GPU device. Also etnaviv is able
to coalesce multiple Vivante GPUs in a single system under one virtual
DRM device, which is used on i.MX6 where the 2D and 3D GPUs are
separate peripherals, but have the same DMA constraints.

Effectively we would need to handle N devices for the dma-mapping in a
lot of places instead of only dealing with the one virtual DRM device.
It would probably be the right thing to anyways, but it's not something
that can be changed short-term. I'm also not yet sure about the
performance implications, as we might run into some cache maintenance
bottlenecks if we dma synchronize buffers to multiple real device
instead of doing it a single time with the virtual DRM device. I know,
I know, this has a lot of assumptions baked in that could fall apart if
someone builds a SoC with multiple Vivante GPUs that have differing DMA
constraints, but up until now hardware designers have not been *that*
crazy, fortunately.

Regards,
Lucas

> FYI, IOMMU configuration is really supposed to *only* run at 
> add_device() time as above - the fact that it's currently hooked in to 
> be retriggered by of_dma_configure() on DT platforms actually turns out 
> to lead to various issues within the IOMMU API, and the plan to change 
> that is slowly climbing up my to-do list.
> 
> Robin.
> 
> > Signed-off-by: Michael Walle <michael@walle.cc>
> > ---
> >   drivers/gpu/drm/etnaviv/etnaviv_drv.c | 24 +++++++++++++++---------
> >   1 file changed, 15 insertions(+), 9 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/etnaviv/etnaviv_drv.c b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> > index 2509b3e85709..ff6425f6ebad 100644
> > --- a/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> > +++ b/drivers/gpu/drm/etnaviv/etnaviv_drv.c
> > @@ -589,6 +589,7 @@ static int compare_str(struct device *dev, void *data)
> >   static int etnaviv_pdev_probe(struct platform_device *pdev)
> >   {
> >   	struct device *dev = &pdev->dev;
> > +	struct device_node *first_node = NULL;
> >   	struct component_match *match = NULL;
> >   
> >   	if (!dev->platform_data) {
> > @@ -598,6 +599,9 @@ static int etnaviv_pdev_probe(struct platform_device *pdev)
> >   			if (!of_device_is_available(core_node))
> >   				continue;
> >   
> > +			if (!first_node)
> > +				first_node = core_node;
> > +
> >   			drm_of_component_match_add(&pdev->dev, &match,
> >   						   compare_of, core_node);
> >   		}
> > @@ -609,6 +613,17 @@ static int etnaviv_pdev_probe(struct platform_device *pdev)
> >   			component_match_add(dev, &match, compare_str, names[i]);
> >   	}
> >   
> > +	pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
> > +	pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
> > +
> > +	/*
> > +	 * Apply the same DMA configuration to the virtual etnaviv
> > +	 * device as the GPU we found. This assumes that all Vivante
> > +	 * GPUs in the system share the same DMA constraints.
> > +	 */
> > +	if (first_node)
> > +		of_dma_configure(&pdev->dev, first_node, true);
> > +
> >   	return component_master_add_with_match(dev, &etnaviv_master_ops, match);
> >   }
> >   
> > @@ -659,15 +674,6 @@ static int __init etnaviv_init(void)
> >   			of_node_put(np);
> >   			goto unregister_platform_driver;
> >   		}
> > -		pdev->dev.coherent_dma_mask = DMA_BIT_MASK(40);
> > -		pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
> > -
> > -		/*
> > -		 * Apply the same DMA configuration to the virtual etnaviv
> > -		 * device as the GPU we found. This assumes that all Vivante
> > -		 * GPUs in the system share the same DMA constraints.
> > -		 */
> > -		of_dma_configure(&pdev->dev, np, true);
> >   
> >   		ret = platform_device_add(pdev);
> >   		if (ret) {
> > 

Re: [PATCH 2/3] drm/etnaviv: fix dma configuration of the virtual device

[Robin Murphy]

On 2021-08-26 16:17, Lucas Stach wrote:
> Am Donnerstag, dem 26.08.2021 um 16:00 +0100 schrieb Robin Murphy:
>> On 2021-08-26 13:10, Michael Walle wrote:
>>> The DMA configuration of the virtual device is inherited from the first
>>> actual etnaviv device. Unfortunately, this doesn't work with an IOMMU:
>>>
>>> [    5.191008] Failed to set up IOMMU for device (null); retaining platform DMA ops
>>>
>>> This is because there is no associated iommu_group with the device. The
>>> group is set in iommu_group_add_device() which is eventually called by
>>> device_add() via the platform bus:
>>>     device_add()
>>>       blocking_notifier_call_chain()
>>>         iommu_bus_notifier()
>>>           iommu_probe_device()
>>>             __iommu_probe_device()
>>>               iommu_group_get_for_dev()
>>>                 iommu_group_add_device()
>>>
>>> Move of_dma_configure() into the probe function, which is called after
>>> device_add(). Normally, the platform code will already call it itself
>>> if .of_node is set. Unfortunately, this isn't the case here.
>>>
>>> Also move the dma mask assignemnts to probe() to keep all DMA related
>>> settings together.
>>
>> I assume the driver must already keep track of the real GPU platform
>> device in order to map registers, request interrupts, etc. correctly -
>> can't it also correctly use that device for DMA API calls and avoid the
>> need for these shenanigans altogether?
>>
> Not without a bigger rework. There's still quite a bit of midlayer
> issues in DRM, where dma-buf imports are dma-mapped and cached via the
> virtual DRM device instead of the real GPU device. Also etnaviv is able
> to coalesce multiple Vivante GPUs in a single system under one virtual
> DRM device, which is used on i.MX6 where the 2D and 3D GPUs are
> separate peripherals, but have the same DMA constraints.

Sure, I wouldn't expect it to be trivial to fix properly, but I wanted 
to point out that this is essentially a hack, relying on an implicit 
side-effect of of_dma_configure() which is already slated for removal. 
As such, I for one am not going to be too sympathetic if it stops 
working in future.

Furthermore, even today it doesn't work in general - it might be OK for 
LS1028A with a single GPU block behind an SMMU, but as soon as you have 
multiple GPU blocks with distinct SMMU StreamIDs, or behind different 
IOMMU instances, then you're stuffed again.

Although in fact I think it's also broken even for LS1028A, since AFAICS 
there's no guarantee that the relevant SMMU instance will actually be 
probed, or the SMMU driver even loaded, when etnaviv_pdev_probe() runs.

> Effectively we would need to handle N devices for the dma-mapping in a
> lot of places instead of only dealing with the one virtual DRM device.
> It would probably be the right thing to anyways, but it's not something
> that can be changed short-term. I'm also not yet sure about the
> performance implications, as we might run into some cache maintenance
> bottlenecks if we dma synchronize buffers to multiple real device
> instead of doing it a single time with the virtual DRM device. I know,
> I know, this has a lot of assumptions baked in that could fall apart if
> someone builds a SoC with multiple Vivante GPUs that have differing DMA
> constraints, but up until now hardware designers have not been *that*
> crazy, fortunately.

I'm not too familiar with the component stuff, but would it be viable to 
just have etnaviv_gpu_platform_probe() set up the first GPU which comes 
along as the master component and fundamental DRM device, then treat any 
subsequent ones as subcomponents as before? That would at least stand to 
be more robust in terms of obviating the of_dma_configure() hack (only 
actual bus code should ever be calling that), even if it won't do 
anything for the multiple IOMMU mapping or differing DMA constraints 
problems.

Thanks,
Robin.