mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
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


  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®