[PATCH v2 0/5] iommufd: Iterate the cache invalidation array in the core

[PATCH v2 0/5] iommufd: Iterate the cache invalidation array in the core

[Nicolin Chen]

The vIOMMU cache_invalidate() and the nested-HWPT cache_invalidate_user()
ops are each handed the full user invalidation array and must report, via
array->entry_num, how many of its entries they handled. That makes every
driver open-code the same array walk, with real downsides:

 - each driver carries its own loop and sub-array bookkeeping;
 - the ARM SMMUv3 driver allocates a buffer sized to the whole array just
   to iterate over it;
 - hand-rolling the loop left the ARM SMMUv3 driver with two long-standing
   bugs:
    1) on a conversion failure it counts commands that it converted but
       never issued, so user space skips invalidations that never reached
       the cmdq;
    2) it rejects a zero-length array, which the uAPI documents as a valid
       request that only probes the data type.

The walk is identical for every driver, so move it into the iommufd core.

The core now drives the iteration:

 - it invokes the op on a sub-array starting at the first not-yet-handled
   entry;
 - the op handles one chunk from the front of that sub-array and reports
   the count via array->entry_num;
 - the core advances and re-invokes until the whole array is consumed or
   the op returns an error.

A driver then only has to handle one bounded chunk per call, e.g. the ARM
SMMUv3 op copies a single cmdq batch into a fixed on-stack buffer and drops
its whole-array allocation. An op still handling the entire array in one
call keeps working, so each driver converts independently.

These are long-standing corner cases, so this targets for-next, not for-rc.

This is on Github:
https://github.com/nicolinc/iommufd/commits/iommufd_invalidation_loop-v2

[Note to Jason and Will]
This has some conflicts with Ashish's ARM_SMMU_OPT_REPEAT_TLBI_CFGI series:
https://lore.kernel.org/all/20260609073204.1760077-1-amhetre@nvidia.com/

Changelog
v2
 * Add "Reviewed-by" from Kevin to patches 2-5
 * Patch-1: Allow the ATC_INV Global bit gated on ssid_bits, correcting
   the wrong every-device claim: per the spec it only broadens a single
   device's invalidation across its PASIDs
 * Patch-1: Move the FEAT_ATS check into the allowlist switch
 * Patch-1: Gate the TTL range field on FEAT_RANGE_INV too
 * Patch-1: Accept only asid_bits of the ASID field
 * Patch-1: Reject Reserved range field value combinations
 * Patch-1: Reject an ATC_INV Size above 52
 * Patch-1: Add local smmu and data variables to simplify the long lines
 * Patch-1: Document the valid-command contract in the uAPI header
 * Patch-1: Note that unchecked out-of-range values are UNPREDICTABLE
 * Patch-1: Note that SSID/Global are IGNORED, not RES0, when SSV == 0
 * Patch-2: Consolidate the two invalidation loops into one
 * Patch-2: Multiply by the size_t entry_len to avoid a u32 overflow
 * Patch-3/5: Return 0 directly on a zero-length array
 * Patch-4: Use a processed counter and an out label like the mock driver
v1
 https://lore.kernel.org/all/cover.1782767398.git.nicolinc@nvidia.com/

Nicolin Chen (5):
  iommu/arm-smmu-v3-iommufd: Reject unsupported bits in invalidation
    commands
  iommufd: Iterate the cache invalidation array in the core
  iommufd/selftest: Convert cache invalidation mocks to the core array
    loop
  iommu/arm-smmu-v3-iommufd: Convert cache invalidation to the core
    array loop
  iommu/vt-d: Convert nested cache invalidation to the core array loop

 include/linux/iommu.h                         |   6 +-
 include/linux/iommufd.h                       |   2 +
 include/uapi/linux/iommufd.h                  |   4 +-
 .../arm/arm-smmu-v3/arm-smmu-v3-iommufd.c     | 179 ++++++++++++++----
 drivers/iommu/intel/nested.c                  |  54 +++---
 drivers/iommu/iommufd/hw_pagetable.c          |  25 ++-
 drivers/iommu/iommufd/selftest.c              | 147 +++++++-------
 7 files changed, 264 insertions(+), 153 deletions(-)

-- 
2.43.0

[PATCH v2 1/5] iommu/arm-smmu-v3-iommufd: Reject unsupported bits in invalidation commands

[Nicolin Chen]

The arm_vsmmu_cache_invalidate() op hands a guest's invalidation commands
to the trusted main command queue after enforcing only the VMID or the SID,
and passes the rest of the command through to the queue unchanged.

That lets a guest set bits the host never meant to forward: a reserved or
undefined bit makes a command malformed; per the Arm SMMUv3 specification,
in its section 4.1.3 "Command errors", a CERROR_ILL is raised, among other
cases, when:

  A valid command opcode is used and a Reserved or undefined field is
  optionally detected as non-zero, which results in the command being
  treated as malformed.

Restrict each opcode to the fields that the driver supports and reject the
command with -EIO if it sets any other bit, before the command reaches the
queue. This stops the host from forwarding any bit whose meaning it does
not control. Document this contract in the uAPI header, so user space must
take the responsibility to forward valid commands only.

Some fields and whole opcodes are legal only on an SMMU that implements
the matching feature, so accept them conditionally:

 - NUM, SCALE, TG and TTL need FEAT_RANGE_INV.
 - ASID is limited to asid_bits, since its upper 8 bits are RES0 on an
   SMMU that only supports 8-bit ASIDs.
 - ATC_INV needs FEAT_ATS. Per the specification's section 4.5 "ATS and
   PRI", CMD_ATC_INV is ILLEGAL when:

     SMMU_IDR0.ATS == 0 and this command is issued on a Non-secure or
     Secure Command queue.

 - SSV, SSID and Global need a non-zero ssid_bits. Without it, setting
   them is not illegal but CONSTRAINED UNPREDICTABLE, which a guest
   should not be able to provoke. Global also takes effect only when
   SSV == 1, broadening the invalidation from the one SSID to all the
   PASIDs of the single device that the SID field addresses.

Some values inside the accepted fields are Reserved too:

 - NUM == 0, SCALE == 0 and TTL == 0 together are a Reserved combination
   and cause a CERROR_ILL.
 - TTL == 0b01 with a 16KB TG is Reserved without SMMU_IDR5.DS, which
   the driver does not support.
 - NUM, SCALE and TTL turn RES0 when TG == 0.
 - An ATC_INV Size above 52, the invalidate-all span, is permitted to
   raise a CERROR_ILL.

Reject these Reserved values the same way. In contrast, an out-of-range
address or ID value is defined as CONSTRAINED UNPREDICTABLE that is
scoped to the guest itself, so it does not deserve a check.

Fixes: d68beb276ba2 ("iommu/arm-smmu-v3: Support IOMMU_HWPT_INVALIDATE using a VIOMMU object")
Cc: stable@vger.kernel.org
Assisted-by: Claude:claude-fable-5
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
---
 include/uapi/linux/iommufd.h                  |   4 +-
 .../arm/arm-smmu-v3/arm-smmu-v3-iommufd.c     | 111 ++++++++++++++++--
 2 files changed, 103 insertions(+), 12 deletions(-)

diff --git a/include/uapi/linux/iommufd.h b/include/uapi/linux/iommufd.h
index 0425d452d41ed..691b810ca574f 100644
--- a/include/uapi/linux/iommufd.h
+++ b/include/uapi/linux/iommufd.h
@@ -909,7 +909,9 @@ struct iommu_hwpt_vtd_s1_invalidate {
  *     CMDQ_OP_CFGI_CD
  *     CMDQ_OP_CFGI_CD_ALL
  *
- * -EIO will be returned if the command is not supported.
+ * User space must forward only valid commands: the kernel rejects, with
+ * -EIO, any command carrying an unsupported opcode, an unsupported field,
+ * or a field value that the underlying SMMU hardware does not implement.
  */
 struct iommu_viommu_arm_smmuv3_invalidate {
 	__aligned_le64 cmd[2];
diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-iommufd.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-iommufd.c
index 1e9f7d2de3441..ee68a13025cab 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-iommufd.c
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-iommufd.c
@@ -305,6 +305,25 @@ struct arm_vsmmu_invalidation_cmd {
 	};
 };
 
+/* Reject the range field values that the spec defines as Reserved */
+static int arm_vsmmu_validate_range(struct arm_vsmmu_invalidation_cmd *cmd)
+{
+	bool range = cmd->cmd.data[0] & (CMDQ_TLBI_0_NUM | CMDQ_TLBI_0_SCALE);
+	u8 ttl = FIELD_GET(CMDQ_TLBI_1_TTL, cmd->cmd.data[1]);
+	u8 tg = FIELD_GET(CMDQ_TLBI_1_TG, cmd->cmd.data[1]);
+
+	/* NUM, SCALE and TTL are RES0 when TG == 0 */
+	if (!tg)
+		return (range || ttl) ? -EIO : 0;
+	/* TTL == 0b01 with a 16KB TG requires SMMU_IDR5.DS, unsupported */
+	if (tg == 2 && ttl == 1)
+		return -EIO;
+	/* NUM == 0, SCALE == 0 with TTL == 0 is a reserved combination */
+	if (!range && !ttl)
+		return -EIO;
+	return 0;
+}
+
 /*
  * Convert, in place, the raw invalidation command into an internal format that
  * can be passed to arm_smmu_cmdq_issue_cmdlist(). Internally commands are
@@ -315,33 +334,103 @@ struct arm_vsmmu_invalidation_cmd {
 static int arm_vsmmu_convert_user_cmd(struct arm_vsmmu *vsmmu,
 				      struct arm_vsmmu_invalidation_cmd *cmd)
 {
+	struct arm_smmu_device *smmu = vsmmu->smmu;
+	u64 allowed[2] = { CMDQ_0_OP };
+	u64 *data = cmd->cmd.data;
+
 	/* Commands are le64 stored in u64 */
-	cmd->cmd.data[0] = le64_to_cpu(cmd->ucmd.cmd[0]);
-	cmd->cmd.data[1] = le64_to_cpu(cmd->ucmd.cmd[1]);
+	data[0] = le64_to_cpu(cmd->ucmd.cmd[0]);
+	data[1] = le64_to_cpu(cmd->ucmd.cmd[1]);
+
+	/* Collect the fields userspace is allowed to set for each opcode */
+	switch (data[0] & CMDQ_0_OP) {
+	case CMDQ_OP_TLBI_NH_VA:
+		/* An SMMU with 8-bit ASIDs treats the upper 8 bits as RES0 */
+		allowed[0] |= FIELD_PREP(CMDQ_TLBI_0_ASID,
+					 GENMASK(smmu->asid_bits - 1, 0));
+		fallthrough;
+	case CMDQ_OP_TLBI_NH_VAA:
+		/* NUM/SCALE/TG/TTL are range fields gated on FEAT_RANGE_INV */
+		if (smmu->features & ARM_SMMU_FEAT_RANGE_INV) {
+			if (arm_vsmmu_validate_range(cmd))
+				return -EIO;
+			allowed[0] |= CMDQ_TLBI_0_NUM | CMDQ_TLBI_0_SCALE;
+			allowed[1] |= CMDQ_TLBI_1_TG | CMDQ_TLBI_1_TTL;
+		}
+		allowed[0] |= CMDQ_TLBI_0_VMID;
+		allowed[1] |= CMDQ_TLBI_1_LEAF | CMDQ_TLBI_1_VA_MASK;
+		break;
+	case CMDQ_OP_TLBI_NH_ASID:
+		/* An SMMU with 8-bit ASIDs treats the upper 8 bits as RES0 */
+		allowed[0] |= FIELD_PREP(CMDQ_TLBI_0_ASID,
+					 GENMASK(smmu->asid_bits - 1, 0));
+		fallthrough;
+	case CMDQ_OP_TLBI_NH_ALL:
+		allowed[0] |= CMDQ_TLBI_0_VMID;
+		break;
+	case CMDQ_OP_ATC_INV:
+		/* ATC_INV is illegal unless the SMMU implements ATS */
+		if (!(smmu->features & ARM_SMMU_FEAT_ATS))
+			return -EIO;
+		/* A Size above 52 (invalidate-all) may raise a CERROR_ILL */
+		if (FIELD_GET(CMDQ_ATC_1_SIZE, data[1]) > ATC_INV_SIZE_ALL)
+			return -EIO;
+		/*
+		 * SSV/SSID/Global need substream support. SSID and Global are
+		 * IGNORED (not RES0) when SSV == 0, so they need no SSV check.
+		 */
+		if (smmu->ssid_bits)
+			allowed[0] |= CMDQ_0_SSV | CMDQ_ATC_0_SSID |
+				      CMDQ_ATC_0_GLOBAL;
+		allowed[0] |= CMDQ_ATC_0_SID;
+		allowed[1] |= CMDQ_ATC_1_SIZE | CMDQ_ATC_1_ADDR_MASK;
+		break;
+	case CMDQ_OP_CFGI_CD:
+		/* No SSV for CFGI_CD; SSID requires substream support */
+		if (smmu->ssid_bits)
+			allowed[0] |= CMDQ_CFGI_0_SSID;
+		allowed[1] |= CMDQ_CFGI_1_LEAF;
+		fallthrough;
+	case CMDQ_OP_CFGI_CD_ALL:
+		allowed[0] |= CMDQ_CFGI_0_SID;
+		break;
+	}
+
+	/*
+	 * Reject any other bit, e.g. a RES0 bit or a Secure bit, before the
+	 * command reaches the trusted main cmdq, so a guest cannot wedge the
+	 * shared queue for every device with a CERROR_ILL.
+	 *
+	 * By contrast, an out-of-range address or ID value does not need a
+	 * check: the spec defines it as CONSTRAINED UNPREDICTABLE, which is
+	 * scoped to the guest itself and does not raise a CERROR_ILL.
+	 */
+	if ((data[0] & ~allowed[0]) || (data[1] & ~allowed[1]))
+		return -EIO;
 
-	switch (cmd->cmd.data[0] & CMDQ_0_OP) {
+	switch (data[0] & CMDQ_0_OP) {
 	case CMDQ_OP_TLBI_NSNH_ALL:
 		/* Convert to NH_ALL */
-		cmd->cmd.data[0] = CMDQ_OP_TLBI_NH_ALL |
-			      FIELD_PREP(CMDQ_TLBI_0_VMID, vsmmu->vmid);
-		cmd->cmd.data[1] = 0;
+		data[0] = CMDQ_OP_TLBI_NH_ALL |
+			  FIELD_PREP(CMDQ_TLBI_0_VMID, vsmmu->vmid);
+		data[1] = 0;
 		break;
 	case CMDQ_OP_TLBI_NH_VA:
 	case CMDQ_OP_TLBI_NH_VAA:
 	case CMDQ_OP_TLBI_NH_ALL:
 	case CMDQ_OP_TLBI_NH_ASID:
-		cmd->cmd.data[0] &= ~CMDQ_TLBI_0_VMID;
-		cmd->cmd.data[0] |= FIELD_PREP(CMDQ_TLBI_0_VMID, vsmmu->vmid);
+		data[0] &= ~CMDQ_TLBI_0_VMID;
+		data[0] |= FIELD_PREP(CMDQ_TLBI_0_VMID, vsmmu->vmid);
 		break;
 	case CMDQ_OP_ATC_INV:
 	case CMDQ_OP_CFGI_CD:
 	case CMDQ_OP_CFGI_CD_ALL: {
-		u32 sid, vsid = FIELD_GET(CMDQ_CFGI_0_SID, cmd->cmd.data[0]);
+		u32 sid, vsid = FIELD_GET(CMDQ_CFGI_0_SID, data[0]);
 
 		if (arm_vsmmu_vsid_to_sid(vsmmu, vsid, &sid))
 			return -EIO;
-		cmd->cmd.data[0] &= ~CMDQ_CFGI_0_SID;
-		cmd->cmd.data[0] |= FIELD_PREP(CMDQ_CFGI_0_SID, sid);
+		data[0] &= ~CMDQ_CFGI_0_SID;
+		data[0] |= FIELD_PREP(CMDQ_CFGI_0_SID, sid);
 		break;
 	}
 	default:
-- 
2.43.0

[PATCH v2 2/5] iommufd: Iterate the cache invalidation array in the core

[Nicolin Chen]

The cache invalidation ops, cache_invalidate_user() for a nested HWPT and
the cache_invalidate() for a vIOMMU, are each handed the full user request
array and report how many of the array entries they handled by setting the
array->entry_num. Every driver therefore implements its own loop over the
array, and a driver wanting to process that array in fixed-size chunks
(e.g. to issue commands out of a fixed-size on-stack buffer) has to carry
the loop and its sub-array bookkeeping all on its own.

Move the iteration into the iommufd core instead. Invoke the op with a
sub-array that starts at the first not-yet-handled entry, let it handle a
prefix of that sub-array and report the count via array->entry_num, then
advance the base pointer and re-invoke the op until the entire array has
been consumed or until the op returns an error along the way.

A driver that handles the entire window in one single call, as all of the
current drivers happen to do, finishes the loop in just one pass, so this
does not change any of the existing behavior. It instead lets each of the
drivers convert to bounded chunk processing on its own, done by each of the
subsequent changes.

Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
---
 include/linux/iommu.h                |  6 ++++--
 include/linux/iommufd.h              |  2 ++
 drivers/iommu/iommufd/hw_pagetable.c | 25 +++++++++++++++----------
 3 files changed, 21 insertions(+), 12 deletions(-)

diff --git a/include/linux/iommu.h b/include/linux/iommu.h
index d20aa6f6863ab..969758f87e445 100644
--- a/include/linux/iommu.h
+++ b/include/linux/iommu.h
@@ -773,8 +773,10 @@ struct iommu_ops {
  *                         passes in the cache invalidation requests, in form
  *                         of a driver data structure. The driver must update
  *                         array->entry_num to report the number of handled
- *                         invalidation requests. The driver data structure
- *                         must be defined in include/uapi/linux/iommufd.h
+ *                         invalidation requests. A driver may handle fewer than
+ *                         the requested, in which case the core re-invokes the
+ *                         op for the remainder. The driver data structure must
+ *                         be defined in include/uapi/linux/iommufd.h
  * @iova_to_phys: translate iova to physical address
  * @enforce_cache_coherency: Prevent any kind of DMA from bypassing IOMMU_CACHE,
  *                           including no-snoop TLPs on PCIe or other platform
diff --git a/include/linux/iommufd.h b/include/linux/iommufd.h
index 6e7efe83bc5d8..3087f5b2def84 100644
--- a/include/linux/iommufd.h
+++ b/include/linux/iommufd.h
@@ -154,6 +154,8 @@ struct iommufd_hw_queue {
  *                    The @array passes in the cache invalidation requests, in
  *                    form of a driver data structure. A driver must update the
  *                    array->entry_num to report the number of handled requests.
+ *                    A driver may handle fewer than the requested entry_num, in
+ *                    which case the core re-invokes the op for the remainder.
  *                    The data structure of the array entry must be defined in
  *                    include/uapi/linux/iommufd.h
  * @vdevice_size: Size of the driver-defined vDEVICE structure per this vIOMMU
diff --git a/drivers/iommu/iommufd/hw_pagetable.c b/drivers/iommu/iommufd/hw_pagetable.c
index 623cc608ca0cd..644daf18849f4 100644
--- a/drivers/iommu/iommufd/hw_pagetable.c
+++ b/drivers/iommu/iommufd/hw_pagetable.c
@@ -501,6 +501,8 @@ int iommufd_hwpt_invalidate(struct iommufd_ucmd *ucmd)
 		.entry_len = cmd->entry_len,
 		.entry_num = cmd->entry_num,
 	};
+	struct iommufd_hw_pagetable *hwpt = NULL;
+	struct iommufd_viommu *viommu = NULL;
 	struct iommufd_object *pt_obj;
 	u32 done_num = 0;
 	int rc;
@@ -527,31 +529,34 @@ int iommufd_hwpt_invalidate(struct iommufd_ucmd *ucmd)
 		goto out;
 	}
 	if (pt_obj->type == IOMMUFD_OBJ_HWPT_NESTED) {
-		struct iommufd_hw_pagetable *hwpt =
-			container_of(pt_obj, struct iommufd_hw_pagetable, obj);
-
+		hwpt = container_of(pt_obj, struct iommufd_hw_pagetable, obj);
 		if (!hwpt->domain->ops ||
 		    !hwpt->domain->ops->cache_invalidate_user) {
 			rc = -EOPNOTSUPP;
 			goto out_put_pt;
 		}
-		rc = hwpt->domain->ops->cache_invalidate_user(hwpt->domain,
-							      &data_array);
 	} else if (pt_obj->type == IOMMUFD_OBJ_VIOMMU) {
-		struct iommufd_viommu *viommu =
-			container_of(pt_obj, struct iommufd_viommu, obj);
-
+		viommu = container_of(pt_obj, struct iommufd_viommu, obj);
 		if (!viommu->ops || !viommu->ops->cache_invalidate) {
 			rc = -EOPNOTSUPP;
 			goto out_put_pt;
 		}
-		rc = viommu->ops->cache_invalidate(viommu, &data_array);
 	} else {
 		rc = -EINVAL;
 		goto out_put_pt;
 	}
 
-	done_num = data_array.entry_num;
+	do {
+		if (viommu)
+			rc = viommu->ops->cache_invalidate(viommu, &data_array);
+		else
+			rc = hwpt->domain->ops->cache_invalidate_user(
+				hwpt->domain, &data_array);
+
+		done_num += data_array.entry_num;
+		data_array.uptr += data_array.entry_num * data_array.entry_len;
+		data_array.entry_num = cmd->entry_num - done_num;
+	} while (!rc && done_num != cmd->entry_num);
 
 out_put_pt:
 	iommufd_put_object(ucmd->ictx, pt_obj);
-- 
2.43.0

[PATCH v2 3/5] iommufd/selftest: Convert cache invalidation mocks to the core array loop

[Nicolin Chen]

The vIOMMU and the nested-domain selftest invalidation mocks each used to
walk the whole request array on their own, with the vIOMMU mock even
allocating a buffer sized to the entire array in order to do so first.

The iommufd core now iterates the request array itself and re-invokes the
op with the not-yet-handled sub-array, so handle just a single request per
call out of the front of that sub-array and report one handled entry via
the array->entry_num. Drop both of the loops and the kzalloc_objs() in the
viommu callback function, and keep returning a success for an empty array
as a probe of the selftest data type.

Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
---
 drivers/iommu/iommufd/selftest.c | 147 +++++++++++++++----------------
 1 file changed, 72 insertions(+), 75 deletions(-)

diff --git a/drivers/iommu/iommufd/selftest.c b/drivers/iommu/iommufd/selftest.c
index af07c642a5260..75846cf1f9c44 100644
--- a/drivers/iommu/iommufd/selftest.c
+++ b/drivers/iommu/iommufd/selftest.c
@@ -631,70 +631,63 @@ mock_viommu_alloc_domain_nested(struct iommufd_viommu *viommu, u32 flags,
 static int mock_viommu_cache_invalidate(struct iommufd_viommu *viommu,
 					struct iommu_user_data_array *array)
 {
-	struct iommu_viommu_invalidate_selftest *cmds;
-	struct iommu_viommu_invalidate_selftest *cur;
-	struct iommu_viommu_invalidate_selftest *end;
-	int rc;
+	struct iommu_viommu_invalidate_selftest cmd;
+	struct mock_dev *mdev;
+	struct device *dev;
+	u32 processed = 0;
+	int rc = 0;
+	int i;
 
-	/* A zero-length array is allowed to validate the array type */
-	if (array->entry_num == 0 &&
-	    array->type == IOMMU_VIOMMU_INVALIDATE_DATA_SELFTEST) {
-		array->entry_num = 0;
-		return 0;
+	if (array->type != IOMMU_VIOMMU_INVALIDATE_DATA_SELFTEST) {
+		rc = -EINVAL;
+		goto out;
 	}
 
-	cmds = kzalloc_objs(*cmds, array->entry_num);
-	if (!cmds)
-		return -ENOMEM;
-	cur = cmds;
-	end = cmds + array->entry_num;
+	/*
+	 * The core re-invokes this op for the remaining requests, so handle one
+	 * request per call. A zero-length array only probes the type, validated
+	 * above.
+	 */
+	if (!array->entry_num)
+		return 0;
 
-	static_assert(sizeof(*cmds) == 3 * sizeof(u32));
-	rc = iommu_copy_struct_from_full_user_array(
-		cmds, sizeof(*cmds), array,
-		IOMMU_VIOMMU_INVALIDATE_DATA_SELFTEST);
+	rc = iommu_copy_struct_from_user_array(
+		&cmd, array, IOMMU_VIOMMU_INVALIDATE_DATA_SELFTEST, 0,
+		cache_id);
 	if (rc)
 		goto out;
 
-	while (cur != end) {
-		struct mock_dev *mdev;
-		struct device *dev;
-		int i;
-
-		if (cur->flags & ~IOMMU_TEST_INVALIDATE_FLAG_ALL) {
-			rc = -EOPNOTSUPP;
-			goto out;
-		}
-
-		if (cur->cache_id > MOCK_DEV_CACHE_ID_MAX) {
-			rc = -EINVAL;
-			goto out;
-		}
+	if (cmd.flags & ~IOMMU_TEST_INVALIDATE_FLAG_ALL) {
+		rc = -EOPNOTSUPP;
+		goto out;
+	}
 
-		xa_lock(&viommu->vdevs);
-		dev = iommufd_viommu_find_dev(viommu,
-					      (unsigned long)cur->vdev_id);
-		if (!dev) {
-			xa_unlock(&viommu->vdevs);
-			rc = -EINVAL;
-			goto out;
-		}
-		mdev = container_of(dev, struct mock_dev, dev);
+	if (cmd.cache_id > MOCK_DEV_CACHE_ID_MAX) {
+		rc = -EINVAL;
+		goto out;
+	}
 
-		if (cur->flags & IOMMU_TEST_INVALIDATE_FLAG_ALL) {
-			/* Invalidate all cache entries and ignore cache_id */
-			for (i = 0; i < MOCK_DEV_CACHE_NUM; i++)
-				mdev->cache[i] = 0;
-		} else {
-			mdev->cache[cur->cache_id] = 0;
-		}
+	xa_lock(&viommu->vdevs);
+	dev = iommufd_viommu_find_dev(viommu, (unsigned long)cmd.vdev_id);
+	if (!dev) {
 		xa_unlock(&viommu->vdevs);
-
-		cur++;
+		rc = -EINVAL;
+		goto out;
+	}
+	mdev = container_of(dev, struct mock_dev, dev);
+
+	if (cmd.flags & IOMMU_TEST_INVALIDATE_FLAG_ALL) {
+		/* Invalidate all cache entries and ignore cache_id */
+		for (i = 0; i < MOCK_DEV_CACHE_NUM; i++)
+			mdev->cache[i] = 0;
+	} else {
+		mdev->cache[cmd.cache_id] = 0;
 	}
+	xa_unlock(&viommu->vdevs);
+
+	processed = 1;
 out:
-	array->entry_num = cur - cmds;
-	kfree(cmds);
+	array->entry_num = processed;
 	return rc;
 }
 
@@ -875,42 +868,46 @@ mock_domain_cache_invalidate_user(struct iommu_domain *domain,
 	struct mock_iommu_domain_nested *mock_nested = to_mock_nested(domain);
 	struct iommu_hwpt_invalidate_selftest inv;
 	u32 processed = 0;
-	int i = 0, j;
 	int rc = 0;
+	int i;
 
 	if (array->type != IOMMU_HWPT_INVALIDATE_DATA_SELFTEST) {
 		rc = -EINVAL;
 		goto out;
 	}
 
-	for ( ; i < array->entry_num; i++) {
-		rc = iommu_copy_struct_from_user_array(&inv, array,
-						       IOMMU_HWPT_INVALIDATE_DATA_SELFTEST,
-						       i, iotlb_id);
-		if (rc)
-			break;
+	/*
+	 * The core re-invokes this op for the remaining requests, so handle one
+	 * request per call. A zero-length array only probes the type, validated
+	 * above.
+	 */
+	if (!array->entry_num)
+		return 0;
 
-		if (inv.flags & ~IOMMU_TEST_INVALIDATE_FLAG_ALL) {
-			rc = -EOPNOTSUPP;
-			break;
-		}
+	rc = iommu_copy_struct_from_user_array(
+		&inv, array, IOMMU_HWPT_INVALIDATE_DATA_SELFTEST, 0, iotlb_id);
+	if (rc)
+		goto out;
 
-		if (inv.iotlb_id > MOCK_NESTED_DOMAIN_IOTLB_ID_MAX) {
-			rc = -EINVAL;
-			break;
-		}
+	if (inv.flags & ~IOMMU_TEST_INVALIDATE_FLAG_ALL) {
+		rc = -EOPNOTSUPP;
+		goto out;
+	}
 
-		if (inv.flags & IOMMU_TEST_INVALIDATE_FLAG_ALL) {
-			/* Invalidate all mock iotlb entries and ignore iotlb_id */
-			for (j = 0; j < MOCK_NESTED_DOMAIN_IOTLB_NUM; j++)
-				mock_nested->iotlb[j] = 0;
-		} else {
-			mock_nested->iotlb[inv.iotlb_id] = 0;
-		}
+	if (inv.iotlb_id > MOCK_NESTED_DOMAIN_IOTLB_ID_MAX) {
+		rc = -EINVAL;
+		goto out;
+	}
 
-		processed++;
+	if (inv.flags & IOMMU_TEST_INVALIDATE_FLAG_ALL) {
+		/* Invalidate all mock iotlb entries and ignore iotlb_id */
+		for (i = 0; i < MOCK_NESTED_DOMAIN_IOTLB_NUM; i++)
+			mock_nested->iotlb[i] = 0;
+	} else {
+		mock_nested->iotlb[inv.iotlb_id] = 0;
 	}
 
+	processed = 1;
 out:
 	array->entry_num = processed;
 	return rc;
-- 
2.43.0

[PATCH v2 4/5] iommu/arm-smmu-v3-iommufd: Convert cache invalidation to the core array loop

[Nicolin Chen]

arm_vsmmu_cache_invalidate() allocated a buffer for the entire user request
array, walked the array converting each of the commands, and issued those
converted commands to the cmdq in CMDQ_BATCH_ENTRIES sized chunks, carrying
the sub-array bookkeeping all on its own.

The iommufd core now iterates the invalidation array and re-invokes the op
with the not-yet-handled sub-array, so the driver only has to proceed with
a single chunk per call.

Instead of a per-array allocation, use a fixed on-stack batch to copy from
the userspace array. If the copy fails due to nonzero padding (VMM violates
the ABI), fail the entire batch.

Convert the whole batch before issuing any of it: a malformed command is a
userspace bug, so the first illegal command fails the batch as a unit,
issuing nothing and leaving array->entry_num at zero, the same way the copy
above bails on nonzero padding. A batch that converts cleanly is issued in
full, so the op returns either a handled count with no error or zero with
an error.

A zero-length array now returns success once the data type gets validated,
matching the documented probe behavior, rather than the -EINVAL that the
full-array copy helper would previously return.

This also fixes two long-standing bugs:
 1) On a conversion failure the old code reported commands that it had
    converted but not yet issued, so user space advanced its consumer
    index past invalidations that never reached the cmdq.
 2) A zero-length array was rejected with -EINVAL, although the uAPI
    documents it as a valid request that only probes the data type.

Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
---
 .../arm/arm-smmu-v3/arm-smmu-v3-iommufd.c     | 68 +++++++++++--------
 1 file changed, 38 insertions(+), 30 deletions(-)

diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-iommufd.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-iommufd.c
index ee68a13025cab..19d2849a5294e 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-iommufd.c
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-iommufd.c
@@ -443,49 +443,57 @@ int arm_vsmmu_cache_invalidate(struct iommufd_viommu *viommu,
 			       struct iommu_user_data_array *array)
 {
 	struct arm_vsmmu *vsmmu = container_of(viommu, struct arm_vsmmu, core);
+	struct arm_vsmmu_invalidation_cmd cmds[CMDQ_BATCH_ENTRIES - 1];
 	struct arm_smmu_device *smmu = vsmmu->smmu;
-	struct arm_vsmmu_invalidation_cmd *last;
-	struct arm_vsmmu_invalidation_cmd *cmds;
-	struct arm_vsmmu_invalidation_cmd *cur;
-	struct arm_vsmmu_invalidation_cmd *end;
+	struct iommu_user_data_array batch = {
+		.type = array->type,
+		.uptr = array->uptr,
+		.entry_len = array->entry_len,
+	};
+	u32 processed = 0;
 	int ret;
-
-	cmds = kzalloc_objs(*cmds, array->entry_num);
-	if (!cmds)
-		return -ENOMEM;
-	cur = cmds;
-	end = cmds + array->entry_num;
+	u32 i;
 
 	static_assert(sizeof(*cmds) == 2 * sizeof(u64));
+
+	if (array->type != IOMMU_VIOMMU_INVALIDATE_DATA_ARM_SMMUV3) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	/* A zero-length array only probes the type, validated above */
+	if (!array->entry_num)
+		return 0;
+
+	/*
+	 * The core re-invokes this op for the remaining requests, so copy one
+	 * cmdq batch worth of commands into a fixed on-stack buffer rather than
+	 * allocating for the whole array.
+	 */
+	batch.entry_num = min_t(u32, array->entry_num, ARRAY_SIZE(cmds));
 	ret = iommu_copy_struct_from_full_user_array(
-		cmds, sizeof(*cmds), array,
+		cmds, sizeof(*cmds), &batch,
 		IOMMU_VIOMMU_INVALIDATE_DATA_ARM_SMMUV3);
 	if (ret)
 		goto out;
 
-	last = cmds;
-	while (cur != end) {
-		ret = arm_vsmmu_convert_user_cmd(vsmmu, cur);
+	/*
+	 * Convert the whole batch. Sending an illegal command is a VMM bug, so
+	 * a single one fails the entire batch, issuing nothing.
+	 */
+	for (i = 0; i < batch.entry_num; i++) {
+		ret = arm_vsmmu_convert_user_cmd(vsmmu, &cmds[i]);
 		if (ret)
 			goto out;
-
-		/* FIXME work in blocks of CMDQ_BATCH_ENTRIES and copy each block? */
-		cur++;
-		if (cur != end && (cur - last) != CMDQ_BATCH_ENTRIES - 1)
-			continue;
-
-		/* FIXME always uses the main cmdq rather than trying to group by type */
-		ret = arm_smmu_cmdq_issue_cmdlist(smmu, &smmu->cmdq, &last->cmd,
-						  cur - last, true);
-		if (ret) {
-			cur--;
-			goto out;
-		}
-		last = cur;
 	}
+
+	/* FIXME always uses the main cmdq rather than trying to group by type */
+	ret = arm_smmu_cmdq_issue_cmdlist(smmu, &smmu->cmdq, &cmds->cmd,
+					  batch.entry_num, true);
+	if (!ret)
+		processed = batch.entry_num;
 out:
-	array->entry_num = cur - cmds;
-	kfree(cmds);
+	array->entry_num = processed;
 	return ret;
 }
 
-- 
2.43.0

[PATCH v2 5/5] iommu/vt-d: Convert nested cache invalidation to the core array loop

[Nicolin Chen]

intel_nested_cache_invalidate_user() used to walk the whole request array
on its own, copying and then flushing one single entry at a time.

The iommufd core now iterates the request array itself and re-invokes the
op with the not-yet-handled sub-array, so handle just a single request per
call out of the front of that sub-array and report one handled entry via
the array->entry_num. An empty array keeps returning a success, used as a
probe of IOMMU_HWPT_INVALIDATE_DATA_VTD_S1 support.

Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
---
 drivers/iommu/intel/nested.c | 54 ++++++++++++++++++++----------------
 1 file changed, 30 insertions(+), 24 deletions(-)

diff --git a/drivers/iommu/intel/nested.c b/drivers/iommu/intel/nested.c
index 2b979bec56cef..36e0e8e85c065 100644
--- a/drivers/iommu/intel/nested.c
+++ b/drivers/iommu/intel/nested.c
@@ -93,7 +93,7 @@ static int intel_nested_cache_invalidate_user(struct iommu_domain *domain,
 {
 	struct dmar_domain *dmar_domain = to_dmar_domain(domain);
 	struct iommu_hwpt_vtd_s1_invalidate inv_entry;
-	u32 index, processed = 0;
+	u32 processed = 0;
 	int ret = 0;
 
 	if (array->type != IOMMU_HWPT_INVALIDATE_DATA_VTD_S1) {
@@ -101,31 +101,37 @@ static int intel_nested_cache_invalidate_user(struct iommu_domain *domain,
 		goto out;
 	}
 
-	for (index = 0; index < array->entry_num; index++) {
-		ret = iommu_copy_struct_from_user_array(&inv_entry, array,
-							IOMMU_HWPT_INVALIDATE_DATA_VTD_S1,
-							index, __reserved);
-		if (ret)
-			break;
-
-		if ((inv_entry.flags & ~IOMMU_VTD_INV_FLAGS_LEAF) ||
-		    inv_entry.__reserved) {
-			ret = -EOPNOTSUPP;
-			break;
-		}
-
-		if (!IS_ALIGNED(inv_entry.addr, VTD_PAGE_SIZE) ||
-		    ((inv_entry.npages == U64_MAX) && inv_entry.addr)) {
-			ret = -EINVAL;
-			break;
-		}
-
-		cache_tag_flush_range(dmar_domain, inv_entry.addr,
-				      inv_entry.addr + nrpages_to_size(inv_entry.npages) - 1,
-				      inv_entry.flags & IOMMU_VTD_INV_FLAGS_LEAF);
-		processed++;
+	/*
+	 * The core re-invokes this op for the remaining requests, so handle one
+	 * request per call. A zero-length array only probes the type, validated
+	 * above.
+	 */
+	if (!array->entry_num)
+		return 0;
+
+	ret = iommu_copy_struct_from_user_array(
+		&inv_entry, array, IOMMU_HWPT_INVALIDATE_DATA_VTD_S1, 0,
+		__reserved);
+	if (ret)
+		goto out;
+
+	if ((inv_entry.flags & ~IOMMU_VTD_INV_FLAGS_LEAF) ||
+	    inv_entry.__reserved) {
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+
+	if (!IS_ALIGNED(inv_entry.addr, VTD_PAGE_SIZE) ||
+	    (inv_entry.npages == U64_MAX && inv_entry.addr)) {
+		ret = -EINVAL;
+		goto out;
 	}
 
+	cache_tag_flush_range(dmar_domain, inv_entry.addr,
+			      inv_entry.addr +
+				      nrpages_to_size(inv_entry.npages) - 1,
+			      inv_entry.flags & IOMMU_VTD_INV_FLAGS_LEAF);
+	processed = 1;
 out:
 	array->entry_num = processed;
 	return ret;
-- 
2.43.0