Skip to contentSkip to conversation

Intel IOMMU updates for v6.14

GIT PULL New Lu Baolu 2 participants 7 messages Jan 7, 2025

Subsystems (2)
lore
Share & tools
lore

Lu Baolu

author maintainer
#0 lore raw

Hi Joerg,

The following changes have been queued for v6.14-rc1. They are about new features and code refactoring, including:

- Remove domain_alloc_paging()
- Remove capability audit code
- Draining PRQ in sva unbind path when FPD bit set
- Link cache tags of same iommu unit together

These patches are based on v6.13-rc6. The complete patches are also available at:

https://github.com/LuBaolu/intel-iommu/commits/vtd-update-for-v6.14

Please consider them for iommu next.

Best regards,
baolu

Jason Gunthorpe (1):
iommu/vt-d: Remove domain_alloc_paging()

Kees Bakker (1):
iommu/vt-d: Avoid use of NULL after WARN_ON_ONCE

Lu Baolu (2):
iommu/vt-d: Remove iommu cap audit
iommu/vt-d: Draining PRQ in sva unbind path when FPD bit set

Zhenzhong Duan (1):
iommu/vt-d: Link cache tags of same iommu unit together

 drivers/iommu/intel/Makefile        |   2 +-
 drivers/iommu/intel/cache.c         |  11 +-
 drivers/iommu/intel/cap_audit.c     | 217 ----------------------------
 drivers/iommu/intel/cap_audit.h     | 131 -----------------
 drivers/iommu/intel/iommu.c         |  32 +---
 drivers/iommu/intel/irq_remapping.c |   8 -
 drivers/iommu/intel/pasid.c         |  22 ++-
 drivers/iommu/intel/pasid.h         |   6 +
 8 files changed, 42 insertions(+), 387 deletions(-)
 delete mode 100644 drivers/iommu/intel/cap_audit.c
 delete mode 100644 drivers/iommu/intel/cap_audit.h
Signature · 1 line
2.43.0
author maintainer
#1 lore raw

PATCH 1/5 iommu/vt-d: Avoid use of NULL after WARN_ON_ONCE

From: Kees Bakker <[email protected]>

There is a WARN_ON_ONCE to catch an unlikely situation when domain_remove_dev_pasid can't find the pasid. In case it nevertheless happens we must avoid using a NULL pointer.

---
 drivers/iommu/intel/iommu.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/iommu/intel/iommu.c +4 -3
index 79e0da9eb626..8f75c11a3ec4 100644
@@ -4090,13 +4090,14 @@ void domain_remove_dev_pasid(struct iommu_domain *domain,
 			break;
 		}
 	}
-	WARN_ON_ONCE(!dev_pasid);
 	spin_unlock_irqrestore(&dmar_domain->lock, flags);
 
 	cache_tag_unassign_domain(dmar_domain, dev, pasid);
 	domain_detach_iommu(dmar_domain, iommu);
-	intel_iommu_debugfs_remove_dev_pasid(dev_pasid);
-	kfree(dev_pasid);
+	if (!WARN_ON_ONCE(!dev_pasid)) {
+		intel_iommu_debugfs_remove_dev_pasid(dev_pasid);
+		kfree(dev_pasid);
+	}
 }
 
 static void intel_iommu_remove_dev_pasid(struct device *dev, ioasid_t pasid,
Signature · 1 line
2.43.0
author maintainer
#2 lore raw

PATCH 2/5 iommu/vt-d: Remove domain_alloc_paging()

In reply to Lu Baolu

From: Jason Gunthorpe <[email protected]>

This is duplicated by intel_iommu_domain_alloc_paging_flags(), just remove it.

---
 drivers/iommu/intel/iommu.c | 16 ----------------
 1 file changed, 16 deletions(-)
drivers/iommu/intel/iommu.c +0 -16
index 8f75c11a3ec4..113c66dcc32e 100644
@@ -4446,21 +4446,6 @@ static struct iommu_domain identity_domain = {
 	},
 };
 
-static struct iommu_domain *intel_iommu_domain_alloc_paging(struct device *dev)
-{
-	struct device_domain_info *info = dev_iommu_priv_get(dev);
-	struct intel_iommu *iommu = info->iommu;
-	struct dmar_domain *dmar_domain;
-	bool first_stage;
-
-	first_stage = first_level_by_default(iommu);
-	dmar_domain = paging_domain_alloc(dev, first_stage);
-	if (IS_ERR(dmar_domain))
-		return ERR_CAST(dmar_domain);
-
-	return &dmar_domain->domain;
-}
-
 const struct iommu_ops intel_iommu_ops = {
 	.blocked_domain		= &blocking_domain,
 	.release_domain		= &blocking_domain,
@@ -4469,7 +4454,6 @@ const struct iommu_ops intel_iommu_ops = {
 	.hw_info		= intel_iommu_hw_info,
 	.domain_alloc_paging_flags = intel_iommu_domain_alloc_paging_flags,
 	.domain_alloc_sva	= intel_svm_domain_alloc,
-	.domain_alloc_paging	= intel_iommu_domain_alloc_paging,
 	.domain_alloc_nested	= intel_iommu_domain_alloc_nested,
 	.probe_device		= intel_iommu_probe_device,
 	.release_device		= intel_iommu_release_device,
Signature · 1 line
2.43.0
author maintainer
#3 lore raw

PATCH 3/5 iommu/vt-d: Remove iommu cap audit

In reply to Lu Baolu

The capability audit code was introduced by commit <ad3d19029979> "iommu/vt-d: Audit IOMMU Capabilities and add helper functions", aiming to verify the consistency of capabilities across all IOMMUs for supported features.

Nowadays, all the kAPIs of the iommu subsystem have evolved to be device oriented, in preparation for supporting heterogeneous IOMMU architectures. There is no longer a need to require capability consistence among IOMMUs for any feature.

Remove the iommu cap audit code to make the driver align with the design in the iommu core.

---
 drivers/iommu/intel/Makefile        |   2 +-
 drivers/iommu/intel/cap_audit.c     | 217 ----------------------------
 drivers/iommu/intel/cap_audit.h     | 131 -----------------
 drivers/iommu/intel/iommu.c         |   9 --
 drivers/iommu/intel/irq_remapping.c |   8 -
 5 files changed, 1 insertion(+), 366 deletions(-)
 delete mode 100644 drivers/iommu/intel/cap_audit.c
 delete mode 100644 drivers/iommu/intel/cap_audit.h
drivers/iommu/intel/Makefile +1 -1
index d3bb0798092d..6c7528130cf9 100644
@@ -1,7 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-$(CONFIG_DMAR_TABLE) += dmar.o
 obj-$(CONFIG_INTEL_IOMMU) += iommu.o pasid.o nested.o cache.o prq.o
-obj-$(CONFIG_DMAR_TABLE) += trace.o cap_audit.o
+obj-$(CONFIG_DMAR_TABLE) += trace.o
 obj-$(CONFIG_DMAR_PERF) += perf.o
 obj-$(CONFIG_INTEL_IOMMU_DEBUGFS) += debugfs.o
 obj-$(CONFIG_INTEL_IOMMU_SVM) += svm.o
drivers/iommu/intel/cap_audit.c deleted +0 -217

217 lines not on this page · raw message

drivers/iommu/intel/cap_audit.h deleted +0 -131

131 lines not on this page · raw message

drivers/iommu/intel/iommu.c +0 -9
index 113c66dcc32e..a1e254417307 100644
@@ -29,7 +29,6 @@
 #include "../irq_remapping.h"
 #include "../iommu-pages.h"
 #include "pasid.h"
-#include "cap_audit.h"
 #include "perfmon.h"
 
 #define ROOT_SIZE		VTD_PAGE_SIZE
@@ -2118,10 +2117,6 @@ static int __init init_dmars(void)
 	struct intel_iommu *iommu;
 	int ret;
 
-	ret = intel_cap_audit(CAP_AUDIT_STATIC_DMAR, NULL);
-	if (ret)
-		goto free_iommu;
-
 	for_each_iommu(iommu, drhd) {
 		if (drhd->ignored) {
 			iommu_disable_translation(iommu);
@@ -2617,10 +2612,6 @@ static int intel_iommu_add(struct dmar_drhd_unit *dmaru)
 	struct intel_iommu *iommu = dmaru->iommu;
 	int ret;
 
-	ret = intel_cap_audit(CAP_AUDIT_HOTPLUG_DMAR, iommu);
-	if (ret)
-		goto out;
-
 	/*
 	 * Disable translation if already enabled prior to OS handover.
 	 */
drivers/iommu/intel/irq_remapping.c +0 -8
index 466c1412dd45..4431543bcfe3 100644
@@ -24,7 +24,6 @@
 #include "iommu.h"
 #include "../irq_remapping.h"
 #include "../iommu-pages.h"
-#include "cap_audit.h"
 
 enum irq_mode {
 	IRQ_REMAPPING,
@@ -727,9 +726,6 @@ static int __init intel_prepare_irq_remapping(void)
 	if (dmar_table_init() < 0)
 		return -ENODEV;
 
-	if (intel_cap_audit(CAP_AUDIT_STATIC_IRQR, NULL))
-		return -ENODEV;
-
 	if (!dmar_ir_support())
 		return -ENODEV;
 
@@ -1534,10 +1530,6 @@ static int dmar_ir_add(struct dmar_drhd_unit *dmaru, struct intel_iommu *iommu)
 	int ret;
 	int eim = x2apic_enabled();
 
-	ret = intel_cap_audit(CAP_AUDIT_HOTPLUG_IRQR, iommu);
-	if (ret)
-		return ret;
-
 	if (eim && !ecap_eim_support(iommu->ecap)) {
 		pr_info("DRHD %Lx: EIM not supported by DRHD, ecap %Lx\n",
 			iommu->reg_phys, iommu->ecap);
Signature · 1 line
2.43.0
author maintainer
#4 lore raw

PATCH 4/5 iommu/vt-d: Draining PRQ in sva unbind path when FPD bit set

In reply to Lu Baolu

When a device uses a PASID for SVA (Shared Virtual Address), it's possible that the PASID entry is marked as non-present and FPD bit set before the device flushes all ongoing DMA requests and removes the SVA domain. This can occur when an exception happens and the process terminates before the device driver stops DMA and calls the iommu driver to unbind the PASID.

There's no need to drain the PRQ in the mm release path. Instead, the PRQ will be drained in the SVA unbind path. But in such case, intel_pasid_tear_down_entry() only checks the presence of the pasid entry and returns directly.

Add the code to clear the FPD bit and drain the PRQ.

Fixes c43e1ccdebf2 ("iommu/vt-d: Drain PRQs when domain removed from RID")
---
 drivers/iommu/intel/pasid.c | 22 +++++++++++++++++++++-
 drivers/iommu/intel/pasid.h |  6 ++++++
 2 files changed, 27 insertions(+), 1 deletion(-)
drivers/iommu/intel/pasid.c +21 -1
index 5b7d85f1e143..fb59a7d35958 100644
@@ -244,11 +244,31 @@ void intel_pasid_tear_down_entry(struct intel_iommu *iommu, struct device *dev,
 
 	spin_lock(&iommu->lock);
 	pte = intel_pasid_get_entry(dev, pasid);
-	if (WARN_ON(!pte) || !pasid_pte_is_present(pte)) {
+	if (WARN_ON(!pte)) {
 		spin_unlock(&iommu->lock);
 		return;
 	}
 
+	if (!pasid_pte_is_present(pte)) {
+		if (!pasid_pte_is_fault_disabled(pte)) {
+			WARN_ON(READ_ONCE(pte->val[0]) != 0);
+			spin_unlock(&iommu->lock);
+			return;
+		}
+
+		/*
+		 * When a PASID is used for SVA by a device, it's possible
+		 * that the pasid entry is non-present with the Fault
+		 * Processing Disabled bit set. Clear the pasid entry and
+		 * drain the PRQ for the PASID before return.
+		 */
+		pasid_clear_entry(pte);
+		spin_unlock(&iommu->lock);
+		intel_iommu_drain_pasid_prq(dev, pasid);
+
+		return;
+	}
+
 	did = pasid_get_domain_id(pte);
 	pgtt = pasid_pte_get_pgtt(pte);
 	intel_pasid_clear_entry(dev, pasid, fault_ignore);
drivers/iommu/intel/pasid.h +6 -0
index 082f4fe20216..668d8ece6b14 100644
@@ -73,6 +73,12 @@ static inline bool pasid_pte_is_present(struct pasid_entry *pte)
 	return READ_ONCE(pte->val[0]) & PASID_PTE_PRESENT;
 }
 
+/* Get FPD(Fault Processing Disable) bit of a PASID table entry */
+static inline bool pasid_pte_is_fault_disabled(struct pasid_entry *pte)
+{
+	return READ_ONCE(pte->val[0]) & PASID_PTE_FPD;
+}
+
 /* Get PGTT field of a PASID table entry */
 static inline u16 pasid_pte_get_pgtt(struct pasid_entry *pte)
 {
Signature · 1 line
2.43.0
author maintainer
#5 lore raw

PATCH 5/5 iommu/vt-d: Link cache tags of same iommu unit together

In reply to Lu Baolu

From: Zhenzhong Duan <[email protected]>

Cache tag invalidation requests for a domain are accumulated until a different iommu unit is found when traversing the cache_tags linked list. But cache tags of same iommu unit can be distributed in the linked list, this make batched flush less efficient. E.g., one device backed by iommu0 is attached to a domain in between two devices attaching backed by iommu1.

Group cache tags together for same iommu unit in cache_tag_assign() to maximize the performance of batched flush.

---
 drivers/iommu/intel/cache.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/iommu/intel/cache.c +10 -1
index 09694cca8752..fc35cba59145 100644
@@ -47,6 +47,7 @@ static int cache_tag_assign(struct dmar_domain *domain, u16 did,
 	struct device_domain_info *info = dev_iommu_priv_get(dev);
 	struct intel_iommu *iommu = info->iommu;
 	struct cache_tag *tag, *temp;
+	struct list_head *prev;
 	unsigned long flags;
 
 	tag = kzalloc(sizeof(*tag), GFP_KERNEL);
@@ -65,6 +66,7 @@ static int cache_tag_assign(struct dmar_domain *domain, u16 did,
 		tag->dev = iommu->iommu.dev;
 
 	spin_lock_irqsave(&domain->cache_lock, flags);
+	prev = &domain->cache_tags;
 	list_for_each_entry(temp, &domain->cache_tags, node) {
 		if (cache_tage_match(temp, did, iommu, dev, pasid, type)) {
 			temp->users++;
@@ -73,8 +75,15 @@ static int cache_tag_assign(struct dmar_domain *domain, u16 did,
 			trace_cache_tag_assign(temp);
 			return 0;
 		}
+		if (temp->iommu == iommu)
+			prev = &temp->node;
 	}
-	list_add_tail(&tag->node, &domain->cache_tags);
+	/*
+	 * Link cache tags of same iommu unit together, so corresponding
+	 * flush ops can be batched for iommu unit.
+	 */
+	list_add(&tag->node, prev);
+
 	spin_unlock_irqrestore(&domain->cache_lock, flags);
 	trace_cache_tag_assign(tag);
 
Signature · 1 line
2.43.0
maintainer Applied
#6 lore raw
In reply to Lu Baolu
Quoted 10 lines · Lu Baolu

Hi Joerg,

The following changes have been queued for v6.14-rc1. They are about new features and code refactoring, including:

- Remove domain_alloc_paging()
- Remove capability audit code
- Draining PRQ in sva unbind path when FPD bit set
- Link cache tags of same iommu unit together

Quoted 6 lines

These patches are based on v6.13-rc6. The complete patches are also available at:

https://github.com/LuBaolu/intel-iommu/commits/vtd-update-for-v6.14

Please consider them for iommu next.

Applied, thanks.