2023-11-30 20:09:18

by Pasha Tatashin

[permalink] [raw]
Subject: [PATCH] vfio: account iommu allocations

iommu allocations should be accounted in order to allow admins to
monitor and limit the amount of iommu memory.

Signed-off-by: Pasha Tatashin <[email protected]>
---
drivers/vfio/vfio_iommu_type1.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)

This patch is spinned of from the series:
https://lore.kernel.org/all/[email protected]

diff --git a/drivers/vfio/vfio_iommu_type1.c b/drivers/vfio/vfio_iommu_type1.c
index eacd6ec04de5..b2854d7939ce 100644
--- a/drivers/vfio/vfio_iommu_type1.c
+++ b/drivers/vfio/vfio_iommu_type1.c
@@ -1436,7 +1436,7 @@ static int vfio_iommu_map(struct vfio_iommu *iommu, dma_addr_t iova,
list_for_each_entry(d, &iommu->domain_list, next) {
ret = iommu_map(d->domain, iova, (phys_addr_t)pfn << PAGE_SHIFT,
npage << PAGE_SHIFT, prot | IOMMU_CACHE,
- GFP_KERNEL);
+ GFP_KERNEL_ACCOUNT);
if (ret)
goto unwind;

@@ -1750,7 +1750,8 @@ static int vfio_iommu_replay(struct vfio_iommu *iommu,
}

ret = iommu_map(domain->domain, iova, phys, size,
- dma->prot | IOMMU_CACHE, GFP_KERNEL);
+ dma->prot | IOMMU_CACHE,
+ GFP_KERNEL_ACCOUNT);
if (ret) {
if (!dma->iommu_mapped) {
vfio_unpin_pages_remote(dma, iova,
@@ -1845,7 +1846,8 @@ static void vfio_test_domain_fgsp(struct vfio_domain *domain, struct list_head *
continue;

ret = iommu_map(domain->domain, start, page_to_phys(pages), PAGE_SIZE * 2,
- IOMMU_READ | IOMMU_WRITE | IOMMU_CACHE, GFP_KERNEL);
+ IOMMU_READ | IOMMU_WRITE | IOMMU_CACHE,
+ GFP_KERNEL_ACCOUNT);
if (!ret) {
size_t unmapped = iommu_unmap(domain->domain, start, PAGE_SIZE);

--
2.43.0.rc2.451.g8631bc7472-goog


2023-12-04 15:46:30

by Jason Gunthorpe

[permalink] [raw]
Subject: Re: [PATCH] vfio: account iommu allocations

On Thu, Nov 30, 2023 at 08:09:00PM +0000, Pasha Tatashin wrote:
> iommu allocations should be accounted in order to allow admins to
> monitor and limit the amount of iommu memory.
>
> Signed-off-by: Pasha Tatashin <[email protected]>
> ---
> drivers/vfio/vfio_iommu_type1.c | 8 +++++---
> 1 file changed, 5 insertions(+), 3 deletions(-)
>
> This patch is spinned of from the series:
>
>https://lore.kernel.org/all/[email protected]

Reviewed-by: Jason Gunthorpe <[email protected]>

Jason

2023-12-04 20:32:13

by Pasha Tatashin

[permalink] [raw]
Subject: Re: [PATCH] vfio: account iommu allocations

On Mon, Dec 4, 2023 at 10:46 AM Jason Gunthorpe <[email protected]> wrote:
>
> On Thu, Nov 30, 2023 at 08:09:00PM +0000, Pasha Tatashin wrote:
> > iommu allocations should be accounted in order to allow admins to
> > monitor and limit the amount of iommu memory.
> >
> > Signed-off-by: Pasha Tatashin <[email protected]>
> > ---
> > drivers/vfio/vfio_iommu_type1.c | 8 +++++---
> > 1 file changed, 5 insertions(+), 3 deletions(-)
> >
> > This patch is spinned of from the series:
> >
> >https://lore.kernel.org/all/[email protected]
>
> Reviewed-by: Jason Gunthorpe <[email protected]>

Thank you,
Pasha

>
> Jason

2023-12-05 00:01:44

by Alex Williamson

[permalink] [raw]
Subject: Re: [PATCH] vfio: account iommu allocations

On Thu, 30 Nov 2023 20:09:00 +0000
Pasha Tatashin <[email protected]> wrote:

> iommu allocations should be accounted in order to allow admins to
> monitor and limit the amount of iommu memory.
>
> Signed-off-by: Pasha Tatashin <[email protected]>
> ---
> drivers/vfio/vfio_iommu_type1.c | 8 +++++---
> 1 file changed, 5 insertions(+), 3 deletions(-)
>
> This patch is spinned of from the series:
> https://lore.kernel.org/all/[email protected]
>
> diff --git a/drivers/vfio/vfio_iommu_type1.c b/drivers/vfio/vfio_iommu_type1.c
> index eacd6ec04de5..b2854d7939ce 100644
> --- a/drivers/vfio/vfio_iommu_type1.c
> +++ b/drivers/vfio/vfio_iommu_type1.c
> @@ -1436,7 +1436,7 @@ static int vfio_iommu_map(struct vfio_iommu *iommu, dma_addr_t iova,
> list_for_each_entry(d, &iommu->domain_list, next) {
> ret = iommu_map(d->domain, iova, (phys_addr_t)pfn << PAGE_SHIFT,
> npage << PAGE_SHIFT, prot | IOMMU_CACHE,
> - GFP_KERNEL);
> + GFP_KERNEL_ACCOUNT);
> if (ret)
> goto unwind;
>
> @@ -1750,7 +1750,8 @@ static int vfio_iommu_replay(struct vfio_iommu *iommu,
> }
>
> ret = iommu_map(domain->domain, iova, phys, size,
> - dma->prot | IOMMU_CACHE, GFP_KERNEL);
> + dma->prot | IOMMU_CACHE,
> + GFP_KERNEL_ACCOUNT);
> if (ret) {
> if (!dma->iommu_mapped) {
> vfio_unpin_pages_remote(dma, iova,
> @@ -1845,7 +1846,8 @@ static void vfio_test_domain_fgsp(struct vfio_domain *domain, struct list_head *
> continue;
>
> ret = iommu_map(domain->domain, start, page_to_phys(pages), PAGE_SIZE * 2,
> - IOMMU_READ | IOMMU_WRITE | IOMMU_CACHE, GFP_KERNEL);
> + IOMMU_READ | IOMMU_WRITE | IOMMU_CACHE,
> + GFP_KERNEL_ACCOUNT);
> if (!ret) {
> size_t unmapped = iommu_unmap(domain->domain, start, PAGE_SIZE);
>

Applied to vfio next branch for v6.8. Thanks,

Alex