2023-05-18 02:09:38

by Jon Pan-Doh

Subject: [PATCH RESEND] iommu/amd: Fix domain flush size when syncing iotlb

When running on an AMD vIOMMU, we observed multiple invalidations (of
decreasing power of 2 aligned sizes) when unmapping a single page.

Domain flush takes gather bounds (end-start) as size param. However,
gather->end is defined as the last inclusive address (start + size - 1).
This leads to an off by 1 error.

With this patch, verified that 1 invalidation occurs when unmapping a
single page.

Fixes: a270be1b3fdf ("iommu/amd: Use only natural aligned flushes in a VM")
Cc: <[email protected]> # 5.15.x
Suggested-by: Gary Zibrat <[email protected]>
Tested-by: Sudheer Dantuluri <[email protected]>
Acked-by: Nadav Amit <[email protected]>
Reviewed-by: Vasant Hegde <[email protected]>
Signed-off-by: Jon Pan-Doh <[email protected]>
---
drivers/iommu/amd/iommu.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c
index 5a505ba5467e..da45b1ab042d 100644
--- a/drivers/iommu/amd/iommu.c
+++ b/drivers/iommu/amd/iommu.c
@@ -2378,7 +2378,7 @@ static void amd_iommu_iotlb_sync(struct iommu_domain *domain,
unsigned long flags;

spin_lock_irqsave(&dom->lock, flags);
- domain_flush_pages(dom, gather->start, gather->end - gather->start, 1);
+ domain_flush_pages(dom, gather->start, gather->end - gather->start + 1, 1);
amd_iommu_domain_flush_complete(dom);
spin_unlock_irqrestore(&dom->lock, flags);
}
--
2.40.0.634.g4ca3ef3211-goog

2023-05-19 17:21:30

by Jerry Snitselaar

Subject: Re: [PATCH RESEND] iommu/amd: Fix domain flush size when syncing iotlb

On Wed, May 17, 2023 at 06:50:54PM -0700, Jon Pan-Doh wrote:
> When running on an AMD vIOMMU, we observed multiple invalidations (of
> decreasing power of 2 aligned sizes) when unmapping a single page.
>
> Domain flush takes gather bounds (end-start) as size param. However,
> gather->end is defined as the last inclusive address (start + size - 1).
> This leads to an off by 1 error.
>
> With this patch, verified that 1 invalidation occurs when unmapping a
> single page.
>
> Fixes: a270be1b3fdf ("iommu/amd: Use only natural aligned flushes in a VM")
> Cc: <[email protected]> # 5.15.x
> Suggested-by: Gary Zibrat <[email protected]>
> Tested-by: Sudheer Dantuluri <[email protected]>
> Acked-by: Nadav Amit <[email protected]>
> Reviewed-by: Vasant Hegde <[email protected]>
> Signed-off-by: Jon Pan-Doh <[email protected]>
> ---

Reviewed-by: Jerry Snitselaar <[email protected]

> drivers/iommu/amd/iommu.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c
> index 5a505ba5467e..da45b1ab042d 100644
> --- a/drivers/iommu/amd/iommu.c
> +++ b/drivers/iommu/amd/iommu.c
> @@ -2378,7 +2378,7 @@ static void amd_iommu_iotlb_sync(struct iommu_domain *domain,
> unsigned long flags;
>
> spin_lock_irqsave(&dom->lock, flags);
> - domain_flush_pages(dom, gather->start, gather->end - gather->start, 1);
> + domain_flush_pages(dom, gather->start, gather->end - gather->start + 1, 1);
> amd_iommu_domain_flush_complete(dom);
> spin_unlock_irqrestore(&dom->lock, flags);
> }
> --
> 2.40.0.634.g4ca3ef3211-goog
>