Re: [PATCH v2 1/2] amd/iommu: Preserve domain ids inside the kdump kernel

From: Jason Gunthorpe
Date: Mon Nov 24 2025 - 09:45:35 EST


On Fri, Nov 21, 2025 at 02:41:15PM +0530, Sairaj Kodilkar wrote:
> Currently AMD IOMMU driver does not reserve domain ids programmed in the
> DTE while reusing the device table inside kdump kernel. This can cause
> reallocation of these domain ids for newer domains that are created by
> the kdump kernel, which can lead to potential IO_PAGE_FAULTs
>
> Hence reserve these ids inside pdom_ids.
>
> Fixes: 38e5f33ee359 ("iommu/amd: Reuse device table for kdump")
> Signed-off-by: Sairaj Kodilkar <sarunkod@xxxxxxx>
> Reported-by: Jason Gunthorpe <jgg@xxxxxxxxxx>
> Reviewed-by: Vasant Hegde <vasant.hegde@xxxxxxx>
> ---
> drivers/iommu/amd/init.c | 23 ++++++++++++++++++++++-
> 1 file changed, 22 insertions(+), 1 deletion(-)

This seems OK

Reviewed-by: Jason Gunthorpe <jgg@xxxxxxxxxx>

But the a point of this work was to remove this code:

diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c
index 48bca4dc8eb61f..1cd799913cbcd6 100644
--- a/drivers/iommu/amd/iommu.c
+++ b/drivers/iommu/amd/iommu.c
@@ -2024,7 +2024,6 @@ static void set_dte_entry(struct amd_iommu *iommu,
phys_addr_t top_paddr, unsigned int top_level)
{
u16 domid;
- u32 old_domid;
struct dev_table_entry *initial_dte;
struct dev_table_entry new = {};
struct protection_domain *domain = dev_data->domain;
@@ -2080,7 +2079,6 @@ static void set_dte_entry(struct amd_iommu *iommu,
if (dev_data->ats_enabled)
new.data[1] |= DTE_FLAG_IOTLB;

- old_domid = READ_ONCE(dte->data[1]) & DEV_DOMID_MASK;
new.data[1] |= domid;

/*
@@ -2096,15 +2094,6 @@ static void set_dte_entry(struct amd_iommu *iommu,
set_dte_gcr3_table(iommu, dev_data, &new);

update_dte256(iommu, dev_data, &new);
-
- /*
- * A kdump kernel might be replacing a domain ID that was copied from
- * the previous kernel--if so, it needs to flush the translation cache
- * entries for the old domain ID that is being overwritten
- */
- if (old_domid) {
- amd_iommu_flush_tlb_domid(iommu, old_domid);
- }
}

/*

Under the reasoning that:
- domids in use by the prior kernel are reserved in the IDA and are
never used by this kernel
- domids in the IDA must be clean
- There is no reason to flush a domid until it is returned to the IDA
- detach_device() calls amd_iommu_domain_flush_all() before the
domain can be freed and the domid returned the IDA which clears the
IOTLB

Please add a patch?

Jason