From: Nicolin Chen <nicolinc@nvidia.com>
To: <jgg@nvidia.com>, <kevin.tian@intel.com>, <corbet@lwn.net>,
<will@kernel.org>
Cc: <bagasdotme@gmail.com>, <robin.murphy@arm.com>, <joro@8bytes.org>,
<thierry.reding@gmail.com>, <vdumpa@nvidia.com>,
<jonathanh@nvidia.com>, <shuah@kernel.org>, <jsnitsel@redhat.com>,
<nathan@kernel.org>, <peterz@infradead.org>, <yi.l.liu@intel.com>,
<mshavit@google.com>, <praan@google.com>,
<zhangzekun11@huawei.com>, <iommu@lists.linux.dev>,
<linux-doc@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
<linux-arm-kernel@lists.infradead.org>,
<linux-tegra@vger.kernel.org>, <linux-kselftest@vger.kernel.org>,
<patches@lists.linux.dev>, <mochs@nvidia.com>,
<alok.a.tiwari@oracle.com>, <vasant.hegde@amd.com>,
<dwmw2@infradead.org>, <baolu.lu@linux.intel.com>
Subject: [PATCH v8 11/29] iommufd/access: Bypass access->ops->unmap for internal use
Date: Fri, 4 Jul 2025 18:13:27 -0700 [thread overview]
Message-ID: <820ae86e5455be2cac480d941e9ab7e29b206402.1751677708.git.nicolinc@nvidia.com> (raw)
In-Reply-To: <cover.1751677708.git.nicolinc@nvidia.com>
The access object has been used externally by VFIO mdev devices, allowing
them to pin/unpin physical pages (via needs_pin_pages). Meanwhile, a racy
unmap can occur in this case, so these devices usually implement an unmap
handler, invoked by iommufd_access_notify_unmap().
The new HW queue object will need the same pin/unpin feature, although it
(unlike the mdev case) wants to reject any unmap attempt, during its life
cycle. Instead, it would not implement an unmap handler. Thus, bypass any
access->ops->unmap access call when the access is marked as internal.
Also, an area being pinned by an internal access should reject any unmap
request. This cannot be done inside iommufd_access_notify_unmap() as it's
a per-iopt action. Add a "num_locks" counter in the struct iopt_area, set
that in iopt_area_add_access() when the caller is an internal access.
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Pranjal Shrivastava <praan@google.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
---
drivers/iommu/iommufd/io_pagetable.h | 5 +++--
drivers/iommu/iommufd/device.c | 18 ++++++++++++------
drivers/iommu/iommufd/io_pagetable.c | 6 ++++++
drivers/iommu/iommufd/pages.c | 12 ++++++++++--
4 files changed, 31 insertions(+), 10 deletions(-)
diff --git a/drivers/iommu/iommufd/io_pagetable.h b/drivers/iommu/iommufd/io_pagetable.h
index c115a51d9384..b6064f4ce4af 100644
--- a/drivers/iommu/iommufd/io_pagetable.h
+++ b/drivers/iommu/iommufd/io_pagetable.h
@@ -48,6 +48,7 @@ struct iopt_area {
int iommu_prot;
bool prevent_access : 1;
unsigned int num_accesses;
+ unsigned int num_locks;
};
struct iopt_allowed {
@@ -238,9 +239,9 @@ void iopt_pages_unfill_xarray(struct iopt_pages *pages, unsigned long start,
int iopt_area_add_access(struct iopt_area *area, unsigned long start,
unsigned long last, struct page **out_pages,
- unsigned int flags);
+ unsigned int flags, bool lock_area);
void iopt_area_remove_access(struct iopt_area *area, unsigned long start,
- unsigned long last);
+ unsigned long last, bool unlock_area);
int iopt_pages_rw_access(struct iopt_pages *pages, unsigned long start_byte,
void *data, unsigned long length, unsigned int flags);
diff --git a/drivers/iommu/iommufd/device.c b/drivers/iommu/iommufd/device.c
index 07a4ff753c12..0567faff5680 100644
--- a/drivers/iommu/iommufd/device.c
+++ b/drivers/iommu/iommufd/device.c
@@ -1048,7 +1048,7 @@ static int iommufd_access_change_ioas(struct iommufd_access *access,
}
if (cur_ioas) {
- if (access->ops->unmap) {
+ if (!iommufd_access_is_internal(access) && access->ops->unmap) {
mutex_unlock(&access->ioas_lock);
access->ops->unmap(access->data, 0, ULONG_MAX);
mutex_lock(&access->ioas_lock);
@@ -1255,7 +1255,8 @@ void iommufd_access_notify_unmap(struct io_pagetable *iopt, unsigned long iova,
xa_lock(&ioas->iopt.access_list);
xa_for_each(&ioas->iopt.access_list, index, access) {
- if (!iommufd_lock_obj(&access->obj))
+ if (!iommufd_lock_obj(&access->obj) ||
+ iommufd_access_is_internal(access))
continue;
xa_unlock(&ioas->iopt.access_list);
@@ -1279,6 +1280,7 @@ void iommufd_access_notify_unmap(struct io_pagetable *iopt, unsigned long iova,
void iommufd_access_unpin_pages(struct iommufd_access *access,
unsigned long iova, unsigned long length)
{
+ bool internal = iommufd_access_is_internal(access);
struct iopt_area_contig_iter iter;
struct io_pagetable *iopt;
unsigned long last_iova;
@@ -1305,7 +1307,8 @@ void iommufd_access_unpin_pages(struct iommufd_access *access,
area, iopt_area_iova_to_index(area, iter.cur_iova),
iopt_area_iova_to_index(
area,
- min(last_iova, iopt_area_last_iova(area))));
+ min(last_iova, iopt_area_last_iova(area))),
+ internal);
WARN_ON(!iopt_area_contig_done(&iter));
up_read(&iopt->iova_rwsem);
mutex_unlock(&access->ioas_lock);
@@ -1354,6 +1357,7 @@ int iommufd_access_pin_pages(struct iommufd_access *access, unsigned long iova,
unsigned long length, struct page **out_pages,
unsigned int flags)
{
+ bool internal = iommufd_access_is_internal(access);
struct iopt_area_contig_iter iter;
struct io_pagetable *iopt;
unsigned long last_iova;
@@ -1362,7 +1366,8 @@ int iommufd_access_pin_pages(struct iommufd_access *access, unsigned long iova,
/* Driver's ops don't support pin_pages */
if (IS_ENABLED(CONFIG_IOMMUFD_TEST) &&
- WARN_ON(access->iova_alignment != PAGE_SIZE || !access->ops->unmap))
+ WARN_ON(access->iova_alignment != PAGE_SIZE ||
+ (!internal && !access->ops->unmap)))
return -EINVAL;
if (!length)
@@ -1396,7 +1401,7 @@ int iommufd_access_pin_pages(struct iommufd_access *access, unsigned long iova,
}
rc = iopt_area_add_access(area, index, last_index, out_pages,
- flags);
+ flags, internal);
if (rc)
goto err_remove;
out_pages += last_index - index + 1;
@@ -1419,7 +1424,8 @@ int iommufd_access_pin_pages(struct iommufd_access *access, unsigned long iova,
iopt_area_iova_to_index(area, iter.cur_iova),
iopt_area_iova_to_index(
area, min(last_iova,
- iopt_area_last_iova(area))));
+ iopt_area_last_iova(area))),
+ internal);
}
up_read(&iopt->iova_rwsem);
mutex_unlock(&access->ioas_lock);
diff --git a/drivers/iommu/iommufd/io_pagetable.c b/drivers/iommu/iommufd/io_pagetable.c
index 22fc3a12109f..abf4aadca96c 100644
--- a/drivers/iommu/iommufd/io_pagetable.c
+++ b/drivers/iommu/iommufd/io_pagetable.c
@@ -719,6 +719,12 @@ static int iopt_unmap_iova_range(struct io_pagetable *iopt, unsigned long start,
goto out_unlock_iova;
}
+ /* The area is locked by an object that has not been destroyed */
+ if (area->num_locks) {
+ rc = -EBUSY;
+ goto out_unlock_iova;
+ }
+
if (area_first < start || area_last > last) {
rc = -ENOENT;
goto out_unlock_iova;
diff --git a/drivers/iommu/iommufd/pages.c b/drivers/iommu/iommufd/pages.c
index cbdde642d2af..301c232462bd 100644
--- a/drivers/iommu/iommufd/pages.c
+++ b/drivers/iommu/iommufd/pages.c
@@ -2111,7 +2111,7 @@ iopt_pages_get_exact_access(struct iopt_pages *pages, unsigned long index,
*/
int iopt_area_add_access(struct iopt_area *area, unsigned long start_index,
unsigned long last_index, struct page **out_pages,
- unsigned int flags)
+ unsigned int flags, bool lock_area)
{
struct iopt_pages *pages = area->pages;
struct iopt_pages_access *access;
@@ -2124,6 +2124,8 @@ int iopt_area_add_access(struct iopt_area *area, unsigned long start_index,
access = iopt_pages_get_exact_access(pages, start_index, last_index);
if (access) {
area->num_accesses++;
+ if (lock_area)
+ area->num_locks++;
access->users++;
iopt_pages_fill_from_xarray(pages, start_index, last_index,
out_pages);
@@ -2145,6 +2147,8 @@ int iopt_area_add_access(struct iopt_area *area, unsigned long start_index,
access->node.last = last_index;
access->users = 1;
area->num_accesses++;
+ if (lock_area)
+ area->num_locks++;
interval_tree_insert(&access->node, &pages->access_itree);
mutex_unlock(&pages->mutex);
return 0;
@@ -2166,7 +2170,7 @@ int iopt_area_add_access(struct iopt_area *area, unsigned long start_index,
* must stop using the PFNs before calling this.
*/
void iopt_area_remove_access(struct iopt_area *area, unsigned long start_index,
- unsigned long last_index)
+ unsigned long last_index, bool unlock_area)
{
struct iopt_pages *pages = area->pages;
struct iopt_pages_access *access;
@@ -2177,6 +2181,10 @@ void iopt_area_remove_access(struct iopt_area *area, unsigned long start_index,
goto out_unlock;
WARN_ON(area->num_accesses == 0 || access->users == 0);
+ if (unlock_area) {
+ WARN_ON(area->num_locks == 0);
+ area->num_locks--;
+ }
area->num_accesses--;
access->users--;
if (access->users)
--
2.43.0
next prev parent reply other threads:[~2025-07-05 1:14 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-07-05 1:13 [PATCH v8 00/29] iommufd: Add vIOMMU infrastructure (Part-4 HW QUEUE) Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 01/29] iommufd: Report unmapped bytes in the error path of iopt_unmap_iova_range Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 02/29] iommufd: Correct virt_id kdoc at struct iommu_vdevice_alloc Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 03/29] iommufd/viommu: Explicitly define vdev->virt_id Nicolin Chen
2025-07-05 15:03 ` Vasant Hegde
2025-07-05 1:13 ` [PATCH v8 04/29] iommu: Use enum iommu_hw_info_type for type in hw_info op Nicolin Chen
2025-07-09 16:47 ` Jason Gunthorpe
2025-07-10 2:57 ` Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 05/29] iommu: Add iommu_copy_struct_to_user helper Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 06/29] iommu: Pass in a driver-level user data structure to viommu_init op Nicolin Chen
2025-07-07 8:05 ` Vasant Hegde
2025-07-05 1:13 ` [PATCH v8 07/29] iommufd/viommu: Allow driver-specific user data for a vIOMMU object Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 08/29] iommufd/selftest: Support user_data in mock_viommu_alloc Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 09/29] iommufd/selftest: Add coverage for viommu data Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 10/29] iommufd/access: Add internal APIs for HW queue to use Nicolin Chen
2025-07-05 1:13 ` Nicolin Chen [this message]
2025-07-05 1:13 ` [PATCH v8 12/29] iommufd/viommu: Add driver-defined vDEVICE support Nicolin Chen
2025-07-07 7:41 ` Vasant Hegde
2025-07-05 1:13 ` [PATCH v8 13/29] iommufd/viommu: Introduce IOMMUFD_OBJ_HW_QUEUE and its related struct Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 14/29] iommufd/viommu: Add IOMMUFD_CMD_HW_QUEUE_ALLOC ioctl Nicolin Chen
2025-07-07 7:41 ` Vasant Hegde
2025-07-09 19:02 ` Nicolin Chen
2025-07-09 17:09 ` Jason Gunthorpe
2025-07-09 18:55 ` Nicolin Chen
2025-07-10 5:34 ` Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 15/29] iommufd/driver: Add iommufd_hw_queue_depend/undepend() helpers Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 16/29] iommufd/selftest: Add coverage for IOMMUFD_CMD_HW_QUEUE_ALLOC Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 17/29] iommufd: Add mmap interface Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 18/29] iommufd/selftest: Add coverage for the new " Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 19/29] Documentation: userspace-api: iommufd: Update HW QUEUE Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 20/29] iommu: Allow an input type in hw_info op Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 21/29] iommufd: Allow an input data_type via iommu_hw_info Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 22/29] iommufd/selftest: Update hw_info coverage for an input data_type Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 23/29] iommu/arm-smmu-v3-iommufd: Add vsmmu_size/type and vsmmu_init impl ops Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 24/29] iommu/arm-smmu-v3-iommufd: Add hw_info to impl_ops Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 25/29] iommu/tegra241-cmdqv: Use request_threaded_irq Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 26/29] iommu/tegra241-cmdqv: Simplify deinit flow in tegra241_cmdqv_remove_vintf() Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 27/29] iommu/tegra241-cmdqv: Do not statically map LVCMDQs Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 28/29] iommu/tegra241-cmdqv: Add user-space use support Nicolin Chen
2025-07-05 1:13 ` [PATCH v8 29/29] iommu/tegra241-cmdqv: Add IOMMU_VEVENTQ_TYPE_TEGRA241_CMDQV support Nicolin Chen
2025-07-09 17:33 ` [PATCH v8 00/29] iommufd: Add vIOMMU infrastructure (Part-4 HW QUEUE) Jason Gunthorpe
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=820ae86e5455be2cac480d941e9ab7e29b206402.1751677708.git.nicolinc@nvidia.com \
--to=nicolinc@nvidia.com \
--cc=alok.a.tiwari@oracle.com \
--cc=bagasdotme@gmail.com \
--cc=baolu.lu@linux.intel.com \
--cc=corbet@lwn.net \
--cc=dwmw2@infradead.org \
--cc=iommu@lists.linux.dev \
--cc=jgg@nvidia.com \
--cc=jonathanh@nvidia.com \
--cc=joro@8bytes.org \
--cc=jsnitsel@redhat.com \
--cc=kevin.tian@intel.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-doc@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=linux-tegra@vger.kernel.org \
--cc=mochs@nvidia.com \
--cc=mshavit@google.com \
--cc=nathan@kernel.org \
--cc=patches@lists.linux.dev \
--cc=peterz@infradead.org \
--cc=praan@google.com \
--cc=robin.murphy@arm.com \
--cc=shuah@kernel.org \
--cc=thierry.reding@gmail.com \
--cc=vasant.hegde@amd.com \
--cc=vdumpa@nvidia.com \
--cc=will@kernel.org \
--cc=yi.l.liu@intel.com \
--cc=zhangzekun11@huawei.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®