[PATCH V2] iommu: Don't reserve 0-length IOVA region

[PATCH V2] iommu: Don't reserve 0-length IOVA region

[Ashish Mhetre]

When the bootloader/firmware doesn't setup the framebuffers, their
address and size are 0 in "iommu-addresses" property. If IOVA region is
reserved with 0 length, then it ends up corrupting the IOVA rbtree with
an entry which has pfn_hi < pfn_lo.
If we intend to use display driver in kernel without framebuffer then
it's causing the display IOMMU mappings to fail as entire valid IOVA
space is reserved when address and length are passed as 0.
An ideal solution would be firmware removing the "iommu-addresses"
property and corresponding "memory-region" if display is not present.
But the kernel should be able to handle this by checking for size of
IOVA region and skipping the IOVA reservation if size is 0. Also, add
a warning if firmware is requesting 0-length IOVA region reservation.

Fixes: a5bf3cfce8cb ("iommu: Implement of_iommu_get_resv_regions()")
Signed-off-by: Ashish Mhetre <amhetre@nvidia.com>
---
 drivers/iommu/of_iommu.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/iommu/of_iommu.c b/drivers/iommu/of_iommu.c
index 47302b637cc0..42cffb0ee5e2 100644
--- a/drivers/iommu/of_iommu.c
+++ b/drivers/iommu/of_iommu.c
@@ -264,6 +264,10 @@ void of_iommu_get_resv_regions(struct device *dev, struct list_head *list)
 					prot |= IOMMU_CACHE;
 
 				maps = of_translate_dma_region(np, maps, &iova, &length);
+				if (length == 0) {
+					dev_warn(dev, "Cannot reserve IOVA region of 0 size\n");
+					continue;
+				}
 				type = iommu_resv_region_get_type(dev, &phys, iova, length);
 
 				region = iommu_alloc_resv_region(iova, length, prot, type,
-- 
2.17.1

Re: [PATCH V2] iommu: Don't reserve 0-length IOVA region

[Ashish Mhetre]

On 12/5/2023 12:26 PM, Ashish Mhetre wrote:
> When the bootloader/firmware doesn't setup the framebuffers, their
> address and size are 0 in "iommu-addresses" property. If IOVA region is
> reserved with 0 length, then it ends up corrupting the IOVA rbtree with
> an entry which has pfn_hi < pfn_lo.
> If we intend to use display driver in kernel without framebuffer then
> it's causing the display IOMMU mappings to fail as entire valid IOVA
> space is reserved when address and length are passed as 0.
> An ideal solution would be firmware removing the "iommu-addresses"
> property and corresponding "memory-region" if display is not present.
> But the kernel should be able to handle this by checking for size of
> IOVA region and skipping the IOVA reservation if size is 0. Also, add
> a warning if firmware is requesting 0-length IOVA region reservation.
>
> Fixes: a5bf3cfce8cb ("iommu: Implement of_iommu_get_resv_regions()")
> Signed-off-by: Ashish Mhetre <amhetre@nvidia.com>
> ---
>   drivers/iommu/of_iommu.c | 4 ++++
>   1 file changed, 4 insertions(+)
>
> diff --git a/drivers/iommu/of_iommu.c b/drivers/iommu/of_iommu.c
> index 47302b637cc0..42cffb0ee5e2 100644
> --- a/drivers/iommu/of_iommu.c
> +++ b/drivers/iommu/of_iommu.c
> @@ -264,6 +264,10 @@ void of_iommu_get_resv_regions(struct device *dev, struct list_head *list)
>   					prot |= IOMMU_CACHE;
>   
>   				maps = of_translate_dma_region(np, maps, &iova, &length);
> +				if (length == 0) {
> +					dev_warn(dev, "Cannot reserve IOVA region of 0 size\n");
> +					continue;
> +				}
>   				type = iommu_resv_region_get_type(dev, &phys, iova, length);
>   
>   				region = iommu_alloc_resv_region(iova, length, prot, type,
Hi all,

Can you please review this patch and provide feedback?

Thank you,
Ashish Mhetre

Re: [PATCH V2] iommu: Don't reserve 0-length IOVA region

[Robin Murphy]

On 2023-12-05 6:56 am, Ashish Mhetre wrote:
> When the bootloader/firmware doesn't setup the framebuffers, their
> address and size are 0 in "iommu-addresses" property. If IOVA region is
> reserved with 0 length, then it ends up corrupting the IOVA rbtree with
> an entry which has pfn_hi < pfn_lo.
> If we intend to use display driver in kernel without framebuffer then
> it's causing the display IOMMU mappings to fail as entire valid IOVA
> space is reserved when address and length are passed as 0.
> An ideal solution would be firmware removing the "iommu-addresses"
> property and corresponding "memory-region" if display is not present.
> But the kernel should be able to handle this by checking for size of
> IOVA region and skipping the IOVA reservation if size is 0. Also, add
> a warning if firmware is requesting 0-length IOVA region reservation.

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

> Fixes: a5bf3cfce8cb ("iommu: Implement of_iommu_get_resv_regions()")
> Signed-off-by: Ashish Mhetre <amhetre@nvidia.com>
> ---
>   drivers/iommu/of_iommu.c | 4 ++++
>   1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/iommu/of_iommu.c b/drivers/iommu/of_iommu.c
> index 47302b637cc0..42cffb0ee5e2 100644
> --- a/drivers/iommu/of_iommu.c
> +++ b/drivers/iommu/of_iommu.c
> @@ -264,6 +264,10 @@ void of_iommu_get_resv_regions(struct device *dev, struct list_head *list)
>   					prot |= IOMMU_CACHE;
>   
>   				maps = of_translate_dma_region(np, maps, &iova, &length);
> +				if (length == 0) {
> +					dev_warn(dev, "Cannot reserve IOVA region of 0 size\n");
> +					continue;
> +				}
>   				type = iommu_resv_region_get_type(dev, &phys, iova, length);
>   
>   				region = iommu_alloc_resv_region(iova, length, prot, type,

Re: [PATCH V2] iommu: Don't reserve 0-length IOVA region

[Joerg Roedel]

On Tue, Dec 05, 2023 at 12:26:56PM +0530, Ashish Mhetre wrote:
>  drivers/iommu/of_iommu.c | 4 ++++
>  1 file changed, 4 insertions(+)

Applied, thanks.