From: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
To: <jgg@nvidia.com>, <nicolinc@nvidia.com>
Cc: <linux-kernel@vger.kernel.org>, <robin.murphy@arm.com>,
<will@kernel.org>, <joro@8bytes.org>, <kevin.tian@intel.com>,
<jsnitsel@redhat.com>, <vasant.hegde@amd.com>,
<iommu@lists.linux.dev>, <santosh.shukla@amd.com>,
<sairaj.arunkodilkar@amd.com>, <jon.grimm@amd.com>,
<prashanthpra@google.com>, <wvw@google.com>, <wnliu@google.com>,
<gptran@google.com>, <kpsingh@google.com>,
Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Subject: [PATCH 7/8] iommu/amd: Add support for nested domain allocation
Date: Wed, 20 Aug 2025 11:30:08 +0000 [thread overview]
Message-ID: <20250820113009.5233-8-suravee.suthikulpanit@amd.com> (raw)
In-Reply-To: <20250820113009.5233-1-suravee.suthikulpanit@amd.com>
The child domain is allocated with IOMMU_DOMAIN_NESTED type to store
stage-1 translation (i.e. GVA->GPA). This includes the GCR3 root pointer
table along with guest (v2) page tables. The struct iommu_hwpt_amd_v2
contains this information, and is passed from user-space as a parameter
of the struct iommu_ops.domain_alloc_nested().
The parent domain is tracked using the struct protection_domain.parent.
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
---
drivers/iommu/amd/Makefile | 2 +-
drivers/iommu/amd/amd_iommu.h | 5 ++
drivers/iommu/amd/amd_iommu_types.h | 5 ++
drivers/iommu/amd/iommu.c | 22 ++++----
drivers/iommu/amd/nested.c | 79 +++++++++++++++++++++++++++++
5 files changed, 103 insertions(+), 10 deletions(-)
create mode 100644 drivers/iommu/amd/nested.c
diff --git a/drivers/iommu/amd/Makefile b/drivers/iommu/amd/Makefile
index 59c04a67f398..9ccf9d61810c 100644
--- a/drivers/iommu/amd/Makefile
+++ b/drivers/iommu/amd/Makefile
@@ -1,3 +1,3 @@
# SPDX-License-Identifier: GPL-2.0-only
-obj-y += iommu.o init.o quirks.o io_pgtable.o io_pgtable_v2.o ppr.o pasid.o
+obj-y += iommu.o init.o quirks.o io_pgtable.o io_pgtable_v2.o ppr.o pasid.o nested.o
obj-$(CONFIG_AMD_IOMMU_DEBUGFS) += debugfs.o
diff --git a/drivers/iommu/amd/amd_iommu.h b/drivers/iommu/amd/amd_iommu.h
index 3ff380afb9f4..8e86d5b1d915 100644
--- a/drivers/iommu/amd/amd_iommu.h
+++ b/drivers/iommu/amd/amd_iommu.h
@@ -8,6 +8,7 @@
#define AMD_IOMMU_H
#include <linux/iommu.h>
+#include <uapi/linux/iommufd.h>
#include "amd_iommu_types.h"
@@ -190,4 +191,8 @@ void amd_iommu_domain_set_pgtable(struct protection_domain *domain,
struct dev_table_entry *get_dev_table(struct amd_iommu *iommu);
struct iommu_dev_data *search_dev_data(struct amd_iommu *iommu, u16 devid);
+/* NESTED */
+struct iommu_domain *
+amd_iommu_domain_alloc_nested(struct device *dev, struct iommu_domain *parent,
+ u32 flags, const struct iommu_user_data *user_data);
#endif /* AMD_IOMMU_H */
diff --git a/drivers/iommu/amd/amd_iommu_types.h b/drivers/iommu/amd/amd_iommu_types.h
index 5343b99913e4..94f51a09b364 100644
--- a/drivers/iommu/amd/amd_iommu_types.h
+++ b/drivers/iommu/amd/amd_iommu_types.h
@@ -20,6 +20,8 @@
#include <linux/irqreturn.h>
#include <linux/io-pgtable.h>
+#include <uapi/linux/iommufd.h>
+
/*
* Maximum number of IOMMUs supported
*/
@@ -605,6 +607,9 @@ struct protection_domain {
struct mmu_notifier mn; /* mmu notifier for the SVA domain */
struct list_head dev_data_list; /* List of pdom_dev_data */
+
+ struct protection_domain *parent; /* Nested parent domain */
+ struct iommu_hwpt_amd_v2 guest_hwpt;
};
/*
diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c
index 46682c8ba28d..ea790a8997ee 100644
--- a/drivers/iommu/amd/iommu.c
+++ b/drivers/iommu/amd/iommu.c
@@ -2616,6 +2616,7 @@ amd_iommu_domain_alloc_paging_flags(struct device *dev, u32 flags,
const struct iommu_user_data *user_data)
{
+ struct iommu_domain *dom = ERR_PTR(-EOPNOTSUPP);
struct iommu_dev_data *dev_data = dev_iommu_priv_get(dev);
struct amd_iommu *iommu = get_amd_iommu_from_dev(dev);
const u32 supported_flags = IOMMU_HWPT_ALLOC_DIRTY_TRACKING |
@@ -2626,29 +2627,31 @@ amd_iommu_domain_alloc_paging_flags(struct device *dev, u32 flags,
if ((flags & ~supported_flags) || user_data || !is_nest_parent_supported(flags))
return ERR_PTR(-EOPNOTSUPP);
- pr_debug("%s: IOMMU devid=%#x, flags=%#x\n", __func__, dev_data->devid, flags);
+ pr_debug("%s: IOMMU devid=%#x, flags=%#x, supported_flags=%#x\n", __func__, dev_data->devid, flags, supported_flags);
switch (flags & supported_flags) {
case IOMMU_HWPT_ALLOC_DIRTY_TRACKING:
case IOMMU_HWPT_ALLOC_DIRTY_TRACKING | IOMMU_HWPT_ALLOC_NEST_PARENT:
case IOMMU_HWPT_ALLOC_NEST_PARENT:
/* Allocate domain with v1 page table for dirty tracking */
- if (!amd_iommu_hd_support(iommu))
- break;
- return do_iommu_domain_alloc(dev, flags, PD_MODE_V1);
+ if (amd_iommu_hd_support(iommu))
+ dom = do_iommu_domain_alloc(dev, flags, PD_MODE_V1);
+ break;
case IOMMU_HWPT_ALLOC_PASID:
/* Allocate domain with v2 page table if IOMMU supports PASID. */
- if (!amd_iommu_pasid_supported())
- break;
- return do_iommu_domain_alloc(dev, flags, PD_MODE_V2);
+ if (amd_iommu_pasid_supported())
+ dom = do_iommu_domain_alloc(dev, flags, PD_MODE_V2);
+ break;
case 0:
/* If nothing specific is required use the kernel commandline default */
- return do_iommu_domain_alloc(dev, 0, amd_iommu_pgtable);
+ dom = do_iommu_domain_alloc(dev, 0, amd_iommu_pgtable);
+ break;
default:
pr_err("%s: Unhandled flag : 0x%x\n", __func__, flags);
break;
}
- return ERR_PTR(-EOPNOTSUPP);
+
+ return dom;
}
void amd_iommu_domain_free(struct iommu_domain *dom)
@@ -3113,6 +3116,7 @@ const struct iommu_ops amd_iommu_ops = {
.release_domain = &release_domain,
.identity_domain = &identity_domain.domain,
.domain_alloc_paging_flags = amd_iommu_domain_alloc_paging_flags,
+ .domain_alloc_nested = amd_iommu_domain_alloc_nested,
.domain_alloc_sva = amd_iommu_domain_alloc_sva,
.probe_device = amd_iommu_probe_device,
.release_device = amd_iommu_release_device,
diff --git a/drivers/iommu/amd/nested.c b/drivers/iommu/amd/nested.c
new file mode 100644
index 000000000000..09f2a455af33
--- /dev/null
+++ b/drivers/iommu/amd/nested.c
@@ -0,0 +1,79 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2023 Advanced Micro Devices, Inc.
+ */
+
+#define pr_fmt(fmt) "AMD-Vi: " fmt
+#define dev_fmt(fmt) pr_fmt(fmt)
+
+#include <linux/iommu.h>
+#include <uapi/linux/iommufd.h>
+
+#include "amd_iommu.h"
+#include "amd_iommu_types.h"
+
+const struct iommu_domain_ops nested_domain_ops = {
+ .free = amd_iommu_domain_free,
+};
+
+static int udata_to_iommu_hwpt_amd_v2(const struct iommu_user_data *user_data,
+ struct iommu_hwpt_amd_v2 *hwpt)
+{
+ if (!user_data)
+ return -EINVAL;
+
+ if (user_data->type != IOMMU_HWPT_DATA_AMD_V2)
+ return -EOPNOTSUPP;
+
+ return iommu_copy_struct_from_user(hwpt, user_data,
+ IOMMU_HWPT_DATA_AMD_V2,
+ dte);
+}
+
+struct iommu_domain *
+amd_iommu_domain_alloc_nested(struct device *dev, struct iommu_domain *parent,
+ u32 flags, const struct iommu_user_data *user_data)
+{
+ int ret;
+ struct iommu_hwpt_amd_v2 hwpt;
+ struct protection_domain *pdom;
+
+ if (parent->ops != amd_iommu_ops.default_domain_ops)
+ return ERR_PTR(-EINVAL);
+
+ ret = udata_to_iommu_hwpt_amd_v2(user_data, &hwpt);
+ if (ret)
+ return ERR_PTR(ret);
+
+ pdom = kzalloc(sizeof(*pdom), GFP_KERNEL);
+ if (IS_ERR(pdom))
+ return ERR_PTR(-ENOMEM);
+
+ pdom->id = amd_iommu_pdom_id_alloc();
+ if (!pdom->id)
+ goto out_err;
+
+ pr_debug("%s: Allocating nested domain with parent domid=%#x\n",
+ __func__, to_pdomain(parent)->id);
+
+ spin_lock_init(&pdom->lock);
+ INIT_LIST_HEAD(&pdom->dev_list);
+ INIT_LIST_HEAD(&pdom->dev_data_list);
+ xa_init(&pdom->iommu_array);
+
+ pdom->pd_mode = PD_MODE_V2;
+ pdom->iop.pgtbl.cfg.amd.nid = NUMA_NO_NODE;
+ pdom->parent = to_pdomain(parent);
+ pdom->domain.ops = &nested_domain_ops;
+ pdom->domain.type = IOMMU_DOMAIN_NESTED;
+ pdom->domain.geometry.aperture_start = 0;
+ pdom->domain.geometry.aperture_end = ((1ULL << PM_LEVEL_SHIFT(amd_iommu_gpt_level)) - 1);
+ pdom->domain.geometry.force_aperture = true;
+ pdom->domain.pgsize_bitmap = pdom->iop.pgtbl.cfg.pgsize_bitmap;
+ memcpy(&pdom->guest_hwpt, &hwpt, sizeof(struct iommu_hwpt_amd_v2));
+
+ return &pdom->domain;
+out_err:
+ kfree(pdom);
+ return ERR_PTR(-EINVAL);
+}
--
2.34.1
next prev parent reply other threads:[~2025-08-20 11:31 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-08-20 11:30 [PATCH 0/8] iommu/amd: Introduce Nested Translation support Suravee Suthikulpanit
2025-08-20 11:30 ` [PATCH 1/8] iommu/amd: Make amd_iommu_pdom_id_alloc() non-static Suravee Suthikulpanit
2025-09-02 13:03 ` Jason Gunthorpe
2025-08-20 11:30 ` [PATCH 2/8] iommu/amd: Making device attach / detach helpers non-static Suravee Suthikulpanit
2025-09-02 13:10 ` Jason Gunthorpe
2025-08-20 11:30 ` [PATCH 3/8] iommu/amd: Making amd_iommu_pdev_enable_cap_ats non-static Suravee Suthikulpanit
2025-08-20 11:30 ` [PATCH 4/8] iommu/amd: Introduce struct gcr3_tbl_info.giov Suravee Suthikulpanit
2025-09-02 13:07 ` Jason Gunthorpe
2025-08-20 11:30 ` [PATCH 5/8] iommufd: Introduce data struct for AMD nested domain allocation Suravee Suthikulpanit
2025-09-02 13:09 ` Jason Gunthorpe
2025-08-20 11:30 ` [PATCH 6/8] iommu/amd: Add support for nest parent " Suravee Suthikulpanit
2025-09-02 13:12 ` Jason Gunthorpe
2025-08-20 11:30 ` Suravee Suthikulpanit [this message]
2025-08-22 5:39 ` [PATCH 7/8] iommu/amd: Add support for nested " kernel test robot
2025-08-22 19:51 ` Nicolin Chen
2025-08-22 21:16 ` Jason Gunthorpe
2025-08-22 21:45 ` Nicolin Chen
2025-08-25 13:38 ` Jason Gunthorpe
2025-09-01 7:09 ` Sairaj Kodilkar
2025-09-02 11:42 ` Jason Gunthorpe
2025-09-02 13:18 ` Jason Gunthorpe
2025-08-20 11:30 ` [PATCH 8/8] iommu/amd: Add support for nested domain attach/detach Suravee Suthikulpanit
2025-08-22 20:20 ` Nicolin Chen
2025-08-28 0:36 ` Suthikulpanit, Suravee
2025-09-02 13:25 ` Jason Gunthorpe
2025-09-03 8:27 ` Sairaj Kodilkar
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20250820113009.5233-8-suravee.suthikulpanit@amd.com \
--to=suravee.suthikulpanit@amd.com \
--cc=gptran@google.com \
--cc=iommu@lists.linux.dev \
--cc=jgg@nvidia.com \
--cc=jon.grimm@amd.com \
--cc=joro@8bytes.org \
--cc=jsnitsel@redhat.com \
--cc=kevin.tian@intel.com \
--cc=kpsingh@google.com \
--cc=linux-kernel@vger.kernel.org \
--cc=nicolinc@nvidia.com \
--cc=prashanthpra@google.com \
--cc=robin.murphy@arm.com \
--cc=sairaj.arunkodilkar@amd.com \
--cc=santosh.shukla@amd.com \
--cc=vasant.hegde@amd.com \
--cc=will@kernel.org \
--cc=wnliu@google.com \
--cc=wvw@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®