[PATCH] iommu/rockchip: prevent iommus dead loop when two masters share one IOMMU

[PATCH] iommu/rockchip: prevent iommus dead loop when two masters share one IOMMU

[Simon Xue]

When two masters share an IOMMU, calling ops->of_xlate during
the second master's driver init may overwrite iommu->domain set
by the first. This causes the check if (iommu->domain == domain)
in rk_iommu_attach_device() to fail, resulting in the same
iommu->node being added twice to &rk_domain->iommus, which can
lead to an infinite loop in subsequent &rk_domain->iommus operations.

Signed-off-by: Simon Xue <xxm@rock-chips.com>
---
 drivers/iommu/rockchip-iommu.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/iommu/rockchip-iommu.c b/drivers/iommu/rockchip-iommu.c
index 22f74ba33a0e..e6bb3c784017 100644
--- a/drivers/iommu/rockchip-iommu.c
+++ b/drivers/iommu/rockchip-iommu.c
@@ -1157,7 +1157,6 @@ static int rk_iommu_of_xlate(struct device *dev,
 		return -ENOMEM;
 
 	data->iommu = platform_get_drvdata(iommu_dev);
-	data->iommu->domain = &rk_identity_domain;
 	dev_iommu_priv_set(dev, data);
 
 	platform_device_put(iommu_dev);
@@ -1195,6 +1194,8 @@ static int rk_iommu_probe(struct platform_device *pdev)
 	if (!iommu)
 		return -ENOMEM;
 
+	iommu->domain = &rk_identity_domain;
+
 	platform_set_drvdata(pdev, iommu);
 	iommu->dev = dev;
 	iommu->num_mmu = 0;
-- 
2.34.1

Re: [PATCH] iommu/rockchip: prevent iommus dead loop when two masters share one IOMMU

[Robin Murphy]

On 2025-06-20 8:39 am, Simon Xue wrote:
> When two masters share an IOMMU, calling ops->of_xlate during
> the second master's driver init may overwrite iommu->domain set
> by the first. This causes the check if (iommu->domain == domain)
> in rk_iommu_attach_device() to fail, resulting in the same
> iommu->node being added twice to &rk_domain->iommus, which can
> lead to an infinite loop in subsequent &rk_domain->iommus operations.

Indeed this is a property of the IOMMU instance itself so it really 
should be initialised before registration, irrespective of client 
devices. FWIW, if it's possible to take an unexpected 
RK_MMU_IRQ_PAGE_FAULT immediately after requesting the IRQ (e.g. in a 
kdump kernel after a crash with the hardware still running) then I think 
the current code could probably end up dereferencing NULL in 
report_iommu_fault() as well.

Reviewed-by: Robin Murphy <robin.murphy@arm.com>

And probably also:

Cc: stable@vger.kernel.org
Fixes: 25c2325575cc ("iommu/rockchip: Add missing set_platform_dma_ops 
callback")

Thanks,
Robin.

> Signed-off-by: Simon Xue <xxm@rock-chips.com>
> ---
>   drivers/iommu/rockchip-iommu.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/iommu/rockchip-iommu.c b/drivers/iommu/rockchip-iommu.c
> index 22f74ba33a0e..e6bb3c784017 100644
> --- a/drivers/iommu/rockchip-iommu.c
> +++ b/drivers/iommu/rockchip-iommu.c
> @@ -1157,7 +1157,6 @@ static int rk_iommu_of_xlate(struct device *dev,
>   		return -ENOMEM;
>   
>   	data->iommu = platform_get_drvdata(iommu_dev);
> -	data->iommu->domain = &rk_identity_domain;
>   	dev_iommu_priv_set(dev, data);
>   
>   	platform_device_put(iommu_dev);
> @@ -1195,6 +1194,8 @@ static int rk_iommu_probe(struct platform_device *pdev)
>   	if (!iommu)
>   		return -ENOMEM;
>   
> +	iommu->domain = &rk_identity_domain;
> +
>   	platform_set_drvdata(pdev, iommu);
>   	iommu->dev = dev;
>   	iommu->num_mmu = 0;

[PATCH v2] iommu/rockchip: prevent iommus dead loop when two masters share one IOMMU

[Simon Xue]

When two masters share an IOMMU, calling ops->of_xlate during
the second master's driver init may overwrite iommu->domain set
by the first. This causes the check if (iommu->domain == domain)
in rk_iommu_attach_device() to fail, resulting in the same
iommu->node being added twice to &rk_domain->iommus, which can
lead to an infinite loop in subsequent &rk_domain->iommus operations.

Fixes: 25c2325575cc ("iommu/rockchip: Add missing set_platform_dma_ops callback")

Signed-off-by: Simon Xue <xxm@rock-chips.com>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>

v2:
   No functional changes.
---
 drivers/iommu/rockchip-iommu.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/iommu/rockchip-iommu.c b/drivers/iommu/rockchip-iommu.c
index 22f74ba33a0e..e6bb3c784017 100644
--- a/drivers/iommu/rockchip-iommu.c
+++ b/drivers/iommu/rockchip-iommu.c
@@ -1157,7 +1157,6 @@ static int rk_iommu_of_xlate(struct device *dev,
 		return -ENOMEM;
 
 	data->iommu = platform_get_drvdata(iommu_dev);
-	data->iommu->domain = &rk_identity_domain;
 	dev_iommu_priv_set(dev, data);
 
 	platform_device_put(iommu_dev);
@@ -1195,6 +1194,8 @@ static int rk_iommu_probe(struct platform_device *pdev)
 	if (!iommu)
 		return -ENOMEM;
 
+	iommu->domain = &rk_identity_domain;
+
 	platform_set_drvdata(pdev, iommu);
 	iommu->dev = dev;
 	iommu->num_mmu = 0;
-- 
2.34.1

[PATCH v3] iommu/rockchip: prevent iommus dead loop when two masters share one IOMMU

[Simon Xue]

When two masters share an IOMMU, calling ops->of_xlate during
the second master's driver init may overwrite iommu->domain set
by the first. This causes the check if (iommu->domain == domain)
in rk_iommu_attach_device() to fail, resulting in the same
iommu->node being added twice to &rk_domain->iommus, which can
lead to an infinite loop in subsequent &rk_domain->iommus operations.

Cc: <stable@vger.kernel.org>
Fixes: 25c2325575cc ("iommu/rockchip: Add missing set_platform_dma_ops callback")
Signed-off-by: Simon Xue <xxm@rock-chips.com>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>

v3:
   Add missing `Cc: stable@vger.kernel.org` in commit message.
   No functional changes.
v2:
   No functional changes.
---
 drivers/iommu/rockchip-iommu.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/iommu/rockchip-iommu.c b/drivers/iommu/rockchip-iommu.c
index 22f74ba33a0e..e6bb3c784017 100644
--- a/drivers/iommu/rockchip-iommu.c
+++ b/drivers/iommu/rockchip-iommu.c
@@ -1157,7 +1157,6 @@ static int rk_iommu_of_xlate(struct device *dev,
 		return -ENOMEM;
 
 	data->iommu = platform_get_drvdata(iommu_dev);
-	data->iommu->domain = &rk_identity_domain;
 	dev_iommu_priv_set(dev, data);
 
 	platform_device_put(iommu_dev);
@@ -1195,6 +1194,8 @@ static int rk_iommu_probe(struct platform_device *pdev)
 	if (!iommu)
 		return -ENOMEM;
 
+	iommu->domain = &rk_identity_domain;
+
 	platform_set_drvdata(pdev, iommu);
 	iommu->dev = dev;
 	iommu->num_mmu = 0;
-- 
2.34.1

Re: [PATCH v3] iommu/rockchip: prevent iommus dead loop when two masters share one IOMMU

[Joerg Roedel]

On Mon, Jun 23, 2025 at 10:00:18AM +0800, Simon Xue wrote:
> When two masters share an IOMMU, calling ops->of_xlate during
> the second master's driver init may overwrite iommu->domain set
> by the first. This causes the check if (iommu->domain == domain)
> in rk_iommu_attach_device() to fail, resulting in the same
> iommu->node being added twice to &rk_domain->iommus, which can
> lead to an infinite loop in subsequent &rk_domain->iommus operations.
> 
> Cc: <stable@vger.kernel.org>
> Fixes: 25c2325575cc ("iommu/rockchip: Add missing set_platform_dma_ops callback")
> Signed-off-by: Simon Xue <xxm@rock-chips.com>
> Reviewed-by: Robin Murphy <robin.murphy@arm.com>
> 
> v3:
>    Add missing `Cc: stable@vger.kernel.org` in commit message.
>    No functional changes.
> v2:
>    No functional changes.
> ---
>  drivers/iommu/rockchip-iommu.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)

Applied for -rc, thanks.