[PATCH 0/6] [PULL REQUEST] Intel IOMMU updates for Linux v6.1

[PATCH 0/6] [PULL REQUEST] Intel IOMMU updates for Linux v6.1

[Lu Baolu]

Hi Joerg,

This includes patches queued for v6.1. It includes:

 - Decouple PASID and PRI from SVA
 - Add ESRTPS & ESIRTPS capability check
 - Cleanups

The whole series is also available at:
https://github.com/LuBaolu/intel-iommu/commits/vtd-update-for-v6.1

Please consider them for next.

Best regards,
Baolu

Lu Baolu (5):
  iommu/vt-d: Remove unnecessary SVA data accesses in page fault path
  iommu/vt-d: Decouple PASID & PRI enabling from SVA
  iommu/vt-d: Remove pasid_set_eafe()
  iommu/vt-d: Avoid unnecessary global IRTE cache invalidation
  iommu/vt-d: Avoid unnecessary global DMA cache invalidation

Yi Liu (1):
  iommu/vt-d: Rename cap_5lp_support to cap_fl5lp_support

 drivers/iommu/intel/iommu.h         |  7 ++-
 drivers/iommu/intel/cap_audit.c     |  4 +-
 drivers/iommu/intel/iommu.c         | 87 ++++++++---------------------
 drivers/iommu/intel/irq_remapping.c |  6 +-
 drivers/iommu/intel/pasid.c         | 12 +---
 drivers/iommu/intel/svm.c           | 62 +++-----------------
 drivers/iommu/intel/Kconfig         |  5 +-
 7 files changed, 45 insertions(+), 138 deletions(-)

-- 
2.34.1

[PATCH 1/6] iommu/vt-d: Remove unnecessary SVA data accesses in page fault path

[Lu Baolu]

The existing I/O page fault handling code accesses the per-PASID SVA data
structures. This is unnecessary and makes the fault handling code only
suitable for SVA scenarios. This removes the SVA data accesses from the
I/O page fault reporting and responding code, so that the fault handling
code could be generic.

Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20220914011821.400986-1-baolu.lu@linux.intel.com
---
 drivers/iommu/intel/iommu.h |  2 +-
 drivers/iommu/intel/svm.c   | 60 +++++--------------------------------
 2 files changed, 8 insertions(+), 54 deletions(-)

diff --git a/drivers/iommu/intel/iommu.h b/drivers/iommu/intel/iommu.h
index 74b0e19e23ee..b5fb7706e97c 100644
--- a/drivers/iommu/intel/iommu.h
+++ b/drivers/iommu/intel/iommu.h
@@ -586,6 +586,7 @@ struct intel_iommu {
 #ifdef CONFIG_INTEL_IOMMU_SVM
 	struct page_req_dsc *prq;
 	unsigned char prq_name[16];    /* Name for PRQ interrupt */
+	unsigned long prq_seq_number;
 	struct completion prq_complete;
 	struct ioasid_allocator_ops pasid_allocator; /* Custom allocator for PASIDs */
 #endif
@@ -761,7 +762,6 @@ struct intel_svm_dev {
 	struct device *dev;
 	struct intel_iommu *iommu;
 	struct iommu_sva sva;
-	unsigned long prq_seq_number;
 	u32 pasid;
 	int users;
 	u16 did;
diff --git a/drivers/iommu/intel/svm.c b/drivers/iommu/intel/svm.c
index 8bcfb93dda56..d1cab931dcb0 100644
--- a/drivers/iommu/intel/svm.c
+++ b/drivers/iommu/intel/svm.c
@@ -48,23 +48,6 @@ static void *pasid_private_find(ioasid_t pasid)
 	return xa_load(&pasid_private_array, pasid);
 }
 
-static struct intel_svm_dev *
-svm_lookup_device_by_sid(struct intel_svm *svm, u16 sid)
-{
-	struct intel_svm_dev *sdev = NULL, *t;
-
-	rcu_read_lock();
-	list_for_each_entry_rcu(t, &svm->devs, list) {
-		if (t->sid == sid) {
-			sdev = t;
-			break;
-		}
-	}
-	rcu_read_unlock();
-
-	return sdev;
-}
-
 static struct intel_svm_dev *
 svm_lookup_device_by_dev(struct intel_svm *svm, struct device *dev)
 {
@@ -706,11 +689,10 @@ static void handle_bad_prq_event(struct intel_iommu *iommu,
 
 static irqreturn_t prq_event_thread(int irq, void *d)
 {
-	struct intel_svm_dev *sdev = NULL;
 	struct intel_iommu *iommu = d;
-	struct intel_svm *svm = NULL;
 	struct page_req_dsc *req;
 	int head, tail, handled;
+	struct pci_dev *pdev;
 	u64 address;
 
 	/*
@@ -730,8 +712,6 @@ static irqreturn_t prq_event_thread(int irq, void *d)
 			pr_err("IOMMU: %s: Page request without PASID\n",
 			       iommu->name);
 bad_req:
-			svm = NULL;
-			sdev = NULL;
 			handle_bad_prq_event(iommu, req, QI_RESP_INVALID);
 			goto prq_advance;
 		}
@@ -758,34 +738,19 @@ static irqreturn_t prq_event_thread(int irq, void *d)
 		if (unlikely(req->lpig && !req->rd_req && !req->wr_req))
 			goto prq_advance;
 
-		if (!svm || svm->pasid != req->pasid) {
-			/*
-			 * It can't go away, because the driver is not permitted
-			 * to unbind the mm while any page faults are outstanding.
-			 */
-			svm = pasid_private_find(req->pasid);
-			if (IS_ERR_OR_NULL(svm) || (svm->flags & SVM_FLAG_SUPERVISOR_MODE))
-				goto bad_req;
-		}
-
-		if (!sdev || sdev->sid != req->rid) {
-			sdev = svm_lookup_device_by_sid(svm, req->rid);
-			if (!sdev)
-				goto bad_req;
-		}
-
-		sdev->prq_seq_number++;
-
+		pdev = pci_get_domain_bus_and_slot(iommu->segment,
+						   PCI_BUS_NUM(req->rid),
+						   req->rid & 0xff);
 		/*
 		 * If prq is to be handled outside iommu driver via receiver of
 		 * the fault notifiers, we skip the page response here.
 		 */
-		if (intel_svm_prq_report(iommu, sdev->dev, req))
+		if (!pdev || intel_svm_prq_report(iommu, &pdev->dev, req))
 			handle_bad_prq_event(iommu, req, QI_RESP_INVALID);
 
-		trace_prq_report(iommu, sdev->dev, req->qw_0, req->qw_1,
+		trace_prq_report(iommu, &pdev->dev, req->qw_0, req->qw_1,
 				 req->priv_data[0], req->priv_data[1],
-				 sdev->prq_seq_number);
+				 iommu->prq_seq_number++);
 prq_advance:
 		head = (head + sizeof(*req)) & PRQ_RING_MASK;
 	}
@@ -881,8 +846,6 @@ int intel_svm_page_response(struct device *dev,
 			    struct iommu_page_response *msg)
 {
 	struct iommu_fault_page_request *prm;
-	struct intel_svm_dev *sdev = NULL;
-	struct intel_svm *svm = NULL;
 	struct intel_iommu *iommu;
 	bool private_present;
 	bool pasid_present;
@@ -901,8 +864,6 @@ int intel_svm_page_response(struct device *dev,
 	if (!msg || !evt)
 		return -EINVAL;
 
-	mutex_lock(&pasid_mutex);
-
 	prm = &evt->fault.prm;
 	sid = PCI_DEVID(bus, devfn);
 	pasid_present = prm->flags & IOMMU_FAULT_PAGE_REQUEST_PASID_VALID;
@@ -919,12 +880,6 @@ int intel_svm_page_response(struct device *dev,
 		goto out;
 	}
 
-	ret = pasid_to_svm_sdev(dev, prm->pasid, &svm, &sdev);
-	if (ret || !sdev) {
-		ret = -ENODEV;
-		goto out;
-	}
-
 	/*
 	 * Per VT-d spec. v3.0 ch7.7, system software must respond
 	 * with page group response if private data is present (PDP)
@@ -954,6 +909,5 @@ int intel_svm_page_response(struct device *dev,
 		qi_submit_sync(iommu, &desc, 1, 0);
 	}
 out:
-	mutex_unlock(&pasid_mutex);
 	return ret;
 }
-- 
2.34.1

[PATCH 2/6] iommu/vt-d: Decouple PASID & PRI enabling from SVA

[Lu Baolu]

Previously the PCI PASID and PRI capabilities are enabled in the path of
iommu device probe only if INTEL_IOMMU_SVM is configured and the device
supports ATS. As we've already decoupled the I/O page fault handler from
SVA, we could also decouple PASID and PRI enabling from it to make room
for growth of new features like kernel DMA with PASID, SIOV and nested
translation.

At the same time, the iommu_enable_dev_iotlb() helper is also called in
iommu_dev_enable_feature(dev, IOMMU_DEV_FEAT_SVA) path. It's unnecessary
and duplicate. This cleanups this helper to make the code neat.

Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20220915085814.2261409-1-baolu.lu@linux.intel.com
---
 drivers/iommu/intel/iommu.h |  1 -
 drivers/iommu/intel/iommu.c | 78 ++++++++-----------------------------
 drivers/iommu/intel/Kconfig |  5 +--
 3 files changed, 18 insertions(+), 66 deletions(-)

diff --git a/drivers/iommu/intel/iommu.h b/drivers/iommu/intel/iommu.h
index b5fb7706e97c..8f29a183467d 100644
--- a/drivers/iommu/intel/iommu.h
+++ b/drivers/iommu/intel/iommu.h
@@ -742,7 +742,6 @@ extern int dmar_ir_support(void);
 void *alloc_pgtable_page(int node);
 void free_pgtable_page(void *vaddr);
 void iommu_flush_write_buffer(struct intel_iommu *iommu);
-int intel_iommu_enable_pasid(struct intel_iommu *iommu, struct device *dev);
 struct intel_iommu *device_to_iommu(struct device *dev, u8 *bus, u8 *devfn);
 
 #ifdef CONFIG_INTEL_IOMMU_SVM
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index e84039720e2b..6d04ee5ea3ab 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -199,6 +199,11 @@ static inline void context_set_domain_id(struct context_entry *context,
 	context->hi |= (value & ((1 << 16) - 1)) << 8;
 }
 
+static inline void context_set_pasid(struct context_entry *context)
+{
+	context->lo |= CONTEXT_PASIDE;
+}
+
 static inline int context_domain_id(struct context_entry *c)
 {
 	return((c->hi >> 8) & 0xffff);
@@ -1350,21 +1355,18 @@ static void __iommu_flush_iotlb(struct intel_iommu *iommu, u16 did,
 }
 
 static struct device_domain_info *
-iommu_support_dev_iotlb(struct dmar_domain *domain, struct intel_iommu *iommu,
-			u8 bus, u8 devfn)
+domain_lookup_dev_info(struct dmar_domain *domain,
+		       struct intel_iommu *iommu, u8 bus, u8 devfn)
 {
 	struct device_domain_info *info;
 	unsigned long flags;
 
-	if (!iommu->qi)
-		return NULL;
-
 	spin_lock_irqsave(&domain->lock, flags);
 	list_for_each_entry(info, &domain->devices, link) {
 		if (info->iommu == iommu && info->bus == bus &&
 		    info->devfn == devfn) {
 			spin_unlock_irqrestore(&domain->lock, flags);
-			return info->ats_supported ? info : NULL;
+			return info;
 		}
 	}
 	spin_unlock_irqrestore(&domain->lock, flags);
@@ -1389,7 +1391,7 @@ static void domain_update_iotlb(struct dmar_domain *domain)
 	spin_unlock_irqrestore(&domain->lock, flags);
 }
 
-static void iommu_enable_dev_iotlb(struct device_domain_info *info)
+static void iommu_enable_pci_caps(struct device_domain_info *info)
 {
 	struct pci_dev *pdev;
 
@@ -1412,7 +1414,6 @@ static void iommu_enable_dev_iotlb(struct device_domain_info *info)
 		info->pfsid = pci_dev_id(pf_pdev);
 	}
 
-#ifdef CONFIG_INTEL_IOMMU_SVM
 	/* The PCIe spec, in its wisdom, declares that the behaviour of
 	   the device if you enable PASID support after ATS support is
 	   undefined. So always enable PASID support on devices which
@@ -1425,7 +1426,7 @@ static void iommu_enable_dev_iotlb(struct device_domain_info *info)
 	    (info->pasid_enabled ? pci_prg_resp_pasid_required(pdev) : 1)  &&
 	    !pci_reset_pri(pdev) && !pci_enable_pri(pdev, PRQ_DEPTH))
 		info->pri_enabled = 1;
-#endif
+
 	if (info->ats_supported && pci_ats_page_aligned(pdev) &&
 	    !pci_enable_ats(pdev, VTD_PAGE_SHIFT)) {
 		info->ats_enabled = 1;
@@ -1448,16 +1449,16 @@ static void iommu_disable_dev_iotlb(struct device_domain_info *info)
 		info->ats_enabled = 0;
 		domain_update_iotlb(info->domain);
 	}
-#ifdef CONFIG_INTEL_IOMMU_SVM
+
 	if (info->pri_enabled) {
 		pci_disable_pri(pdev);
 		info->pri_enabled = 0;
 	}
+
 	if (info->pasid_enabled) {
 		pci_disable_pasid(pdev);
 		info->pasid_enabled = 0;
 	}
-#endif
 }
 
 static void __iommu_flush_dev_iotlb(struct device_domain_info *info,
@@ -1907,7 +1908,7 @@ static int domain_context_mapping_one(struct dmar_domain *domain,
 				      u8 bus, u8 devfn)
 {
 	struct device_domain_info *info =
-			iommu_support_dev_iotlb(domain, iommu, bus, devfn);
+			domain_lookup_dev_info(domain, iommu, bus, devfn);
 	u16 did = domain_id_iommu(domain, iommu);
 	int translation = CONTEXT_TT_MULTI_LEVEL;
 	struct context_entry *context;
@@ -1980,6 +1981,8 @@ static int domain_context_mapping_one(struct dmar_domain *domain,
 			context_set_sm_dte(context);
 		if (info && info->pri_supported)
 			context_set_sm_pre(context);
+		if (info && info->pasid_supported)
+			context_set_pasid(context);
 	} else {
 		struct dma_pte *pgd = domain->pgd;
 		int agaw;
@@ -2037,7 +2040,7 @@ static int domain_context_mapping_one(struct dmar_domain *domain,
 	} else {
 		iommu_flush_write_buffer(iommu);
 	}
-	iommu_enable_dev_iotlb(info);
+	iommu_enable_pci_caps(info);
 
 	ret = 0;
 
@@ -4572,52 +4575,6 @@ static void intel_iommu_get_resv_regions(struct device *device,
 	list_add_tail(&reg->list, head);
 }
 
-int intel_iommu_enable_pasid(struct intel_iommu *iommu, struct device *dev)
-{
-	struct device_domain_info *info = dev_iommu_priv_get(dev);
-	struct context_entry *context;
-	struct dmar_domain *domain;
-	u64 ctx_lo;
-	int ret;
-
-	domain = info->domain;
-	if (!domain)
-		return -EINVAL;
-
-	spin_lock(&iommu->lock);
-	ret = -EINVAL;
-	if (!info->pasid_supported)
-		goto out;
-
-	context = iommu_context_addr(iommu, info->bus, info->devfn, 0);
-	if (WARN_ON(!context))
-		goto out;
-
-	ctx_lo = context[0].lo;
-
-	if (!(ctx_lo & CONTEXT_PASIDE)) {
-		ctx_lo |= CONTEXT_PASIDE;
-		context[0].lo = ctx_lo;
-		wmb();
-		iommu->flush.flush_context(iommu,
-					   domain_id_iommu(domain, iommu),
-					   PCI_DEVID(info->bus, info->devfn),
-					   DMA_CCMD_MASK_NOBIT,
-					   DMA_CCMD_DEVICE_INVL);
-	}
-
-	/* Enable PASID support in the device, if it wasn't already */
-	if (!info->pasid_enabled)
-		iommu_enable_dev_iotlb(info);
-
-	ret = 0;
-
- out:
-	spin_unlock(&iommu->lock);
-
-	return ret;
-}
-
 static struct iommu_group *intel_iommu_device_group(struct device *dev)
 {
 	if (dev_is_pci(dev))
@@ -4641,9 +4598,6 @@ static int intel_iommu_enable_sva(struct device *dev)
 	if (!(iommu->flags & VTD_FLAG_SVM_CAPABLE))
 		return -ENODEV;
 
-	if (intel_iommu_enable_pasid(iommu, dev))
-		return -ENODEV;
-
 	if (!info->pasid_enabled || !info->pri_enabled || !info->ats_enabled)
 		return -EINVAL;
 
diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
index c48005147ac5..b7dff5092fd2 100644
--- a/drivers/iommu/intel/Kconfig
+++ b/drivers/iommu/intel/Kconfig
@@ -20,6 +20,8 @@ config INTEL_IOMMU
 	select SWIOTLB
 	select IOASID
 	select PCI_ATS
+	select PCI_PRI
+	select PCI_PASID
 	help
 	  DMA remapping (DMAR) devices support enables independent address
 	  translations for Direct Memory Access (DMA) from devices.
@@ -47,10 +49,7 @@ config INTEL_IOMMU_DEBUGFS
 config INTEL_IOMMU_SVM
 	bool "Support for Shared Virtual Memory with Intel IOMMU"
 	depends on X86_64
-	select PCI_PASID
-	select PCI_PRI
 	select MMU_NOTIFIER
-	select IOASID
 	select IOMMU_SVA
 	help
 	  Shared Virtual Memory (SVM) provides a facility for devices
-- 
2.34.1

[PATCH 3/6] iommu/vt-d: Remove pasid_set_eafe()

[Lu Baolu]

It is not used anywhere in the tree. Remove it to avoid dead code.
No functional change intended.

Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Link: https://lore.kernel.org/r/20220915081645.1834555-1-baolu.lu@linux.intel.com
---
 drivers/iommu/intel/pasid.c | 10 ----------
 1 file changed, 10 deletions(-)

diff --git a/drivers/iommu/intel/pasid.c b/drivers/iommu/intel/pasid.c
index c5e7e8b020a5..ccaf32949254 100644
--- a/drivers/iommu/intel/pasid.c
+++ b/drivers/iommu/intel/pasid.c
@@ -392,16 +392,6 @@ pasid_set_flpm(struct pasid_entry *pe, u64 value)
 	pasid_set_bits(&pe->val[2], GENMASK_ULL(3, 2), value << 2);
 }
 
-/*
- * Setup the Extended Access Flag Enable (EAFE) field (Bit 135)
- * of a scalable mode PASID entry.
- */
-static inline void
-pasid_set_eafe(struct pasid_entry *pe)
-{
-	pasid_set_bits(&pe->val[2], 1 << 7, 1 << 7);
-}
-
 static void
 pasid_cache_invalidation_with_pasid(struct intel_iommu *iommu,
 				    u16 did, u32 pasid)
-- 
2.34.1

[PATCH 4/6] iommu/vt-d: Rename cap_5lp_support to cap_fl5lp_support

[Lu Baolu]

From: Yi Liu <yi.l.liu@intel.com>

This renaming better describes it is for first level page table (a.k.a
first stage page table since VT-d spec 3.4).

Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20220916071326.2223901-1-yi.l.liu@intel.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
---
 drivers/iommu/intel/iommu.h     | 2 +-
 drivers/iommu/intel/cap_audit.c | 4 ++--
 drivers/iommu/intel/iommu.c     | 2 +-
 drivers/iommu/intel/pasid.c     | 2 +-
 drivers/iommu/intel/svm.c       | 2 +-
 5 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/iommu/intel/iommu.h b/drivers/iommu/intel/iommu.h
index 8f29a183467d..99cc75ecac63 100644
--- a/drivers/iommu/intel/iommu.h
+++ b/drivers/iommu/intel/iommu.h
@@ -146,7 +146,7 @@
 /*
  * Decoding Capability Register
  */
-#define cap_5lp_support(c)	(((c) >> 60) & 1)
+#define cap_fl5lp_support(c)	(((c) >> 60) & 1)
 #define cap_pi_support(c)	(((c) >> 59) & 1)
 #define cap_fl1gp_support(c)	(((c) >> 56) & 1)
 #define cap_read_drain(c)	(((c) >> 55) & 1)
diff --git a/drivers/iommu/intel/cap_audit.c b/drivers/iommu/intel/cap_audit.c
index 3ee68393122f..806986696841 100644
--- a/drivers/iommu/intel/cap_audit.c
+++ b/drivers/iommu/intel/cap_audit.c
@@ -37,7 +37,7 @@ static inline void check_dmar_capabilities(struct intel_iommu *a,
 	MINIMAL_FEATURE_IOMMU(b, ecap, ECAP_MHMV_MASK);
 	MINIMAL_FEATURE_IOMMU(b, ecap, ECAP_IRO_MASK);
 
-	CHECK_FEATURE_MISMATCH(a, b, cap, 5lp_support, CAP_FL5LP_MASK);
+	CHECK_FEATURE_MISMATCH(a, b, cap, fl5lp_support, CAP_FL5LP_MASK);
 	CHECK_FEATURE_MISMATCH(a, b, cap, fl1gp_support, CAP_FL1GP_MASK);
 	CHECK_FEATURE_MISMATCH(a, b, cap, read_drain, CAP_RD_MASK);
 	CHECK_FEATURE_MISMATCH(a, b, cap, write_drain, CAP_WD_MASK);
@@ -84,7 +84,7 @@ static int cap_audit_hotplug(struct intel_iommu *iommu, enum cap_audit_type type
 		goto out;
 	}
 
-	CHECK_FEATURE_MISMATCH_HOTPLUG(iommu, cap, 5lp_support, CAP_FL5LP_MASK);
+	CHECK_FEATURE_MISMATCH_HOTPLUG(iommu, cap, fl5lp_support, CAP_FL5LP_MASK);
 	CHECK_FEATURE_MISMATCH_HOTPLUG(iommu, cap, fl1gp_support, CAP_FL1GP_MASK);
 	CHECK_FEATURE_MISMATCH_HOTPLUG(iommu, cap, read_drain, CAP_RD_MASK);
 	CHECK_FEATURE_MISMATCH_HOTPLUG(iommu, cap, write_drain, CAP_WD_MASK);
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 6d04ee5ea3ab..81a449b5cb91 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -404,7 +404,7 @@ static unsigned long __iommu_calculate_sagaw(struct intel_iommu *iommu)
 {
 	unsigned long fl_sagaw, sl_sagaw;
 
-	fl_sagaw = BIT(2) | (cap_5lp_support(iommu->cap) ? BIT(3) : 0);
+	fl_sagaw = BIT(2) | (cap_fl5lp_support(iommu->cap) ? BIT(3) : 0);
 	sl_sagaw = cap_sagaw(iommu->cap);
 
 	/* Second level only. */
diff --git a/drivers/iommu/intel/pasid.c b/drivers/iommu/intel/pasid.c
index ccaf32949254..c30ddac40ee5 100644
--- a/drivers/iommu/intel/pasid.c
+++ b/drivers/iommu/intel/pasid.c
@@ -519,7 +519,7 @@ int intel_pasid_setup_first_level(struct intel_iommu *iommu,
 		}
 	}
 
-	if ((flags & PASID_FLAG_FL5LP) && !cap_5lp_support(iommu->cap)) {
+	if ((flags & PASID_FLAG_FL5LP) && !cap_fl5lp_support(iommu->cap)) {
 		pr_err("No 5-level paging support for first-level on %s\n",
 		       iommu->name);
 		return -EINVAL;
diff --git a/drivers/iommu/intel/svm.c b/drivers/iommu/intel/svm.c
index d1cab931dcb0..7d08eb034f2d 100644
--- a/drivers/iommu/intel/svm.c
+++ b/drivers/iommu/intel/svm.c
@@ -164,7 +164,7 @@ void intel_svm_check(struct intel_iommu *iommu)
 	}
 
 	if (cpu_feature_enabled(X86_FEATURE_LA57) &&
-	    !cap_5lp_support(iommu->cap)) {
+	    !cap_fl5lp_support(iommu->cap)) {
 		pr_err("%s SVM disabled, incompatible paging mode\n",
 		       iommu->name);
 		return;
-- 
2.34.1

[PATCH 5/6] iommu/vt-d: Avoid unnecessary global IRTE cache invalidation

[Lu Baolu]

Some VT-d hardware implementations invalidate all interrupt remapping
hardware translation caches as part of SIRTP flow. The VT-d spec adds
a ESIRTPS (Enhanced Set Interrupt Remap Table Pointer Support, section
11.4.2 in VT-d spec) capability bit to indicate this.

The spec also states in 11.4.4 that hardware also performs global
invalidation on all interrupt remapping caches as part of Interrupt
Remapping Disable operation if ESIRTPS capability bit is set.

This checks the ESIRTPS capability bit and skip software global cache
invalidation if it's set.

Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jerry Snitselaar <jsnitsel@redhat.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20220921065741.3572495-1-baolu.lu@linux.intel.com
---
 drivers/iommu/intel/iommu.h         | 1 +
 drivers/iommu/intel/irq_remapping.c | 6 ++++--
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/iommu/intel/iommu.h b/drivers/iommu/intel/iommu.h
index 99cc75ecac63..bddf6c69587d 100644
--- a/drivers/iommu/intel/iommu.h
+++ b/drivers/iommu/intel/iommu.h
@@ -146,6 +146,7 @@
 /*
  * Decoding Capability Register
  */
+#define cap_esirtps(c)		(((c) >> 62) & 1)
 #define cap_fl5lp_support(c)	(((c) >> 60) & 1)
 #define cap_pi_support(c)	(((c) >> 59) & 1)
 #define cap_fl1gp_support(c)	(((c) >> 56) & 1)
diff --git a/drivers/iommu/intel/irq_remapping.c b/drivers/iommu/intel/irq_remapping.c
index 2e9683e970f8..5962bb5027d0 100644
--- a/drivers/iommu/intel/irq_remapping.c
+++ b/drivers/iommu/intel/irq_remapping.c
@@ -494,7 +494,8 @@ static void iommu_set_irq_remapping(struct intel_iommu *iommu, int mode)
 	 * Global invalidation of interrupt entry cache to make sure the
 	 * hardware uses the new irq remapping table.
 	 */
-	qi_global_iec(iommu);
+	if (!cap_esirtps(iommu->cap))
+		qi_global_iec(iommu);
 }
 
 static void iommu_enable_irq_remapping(struct intel_iommu *iommu)
@@ -680,7 +681,8 @@ static void iommu_disable_irq_remapping(struct intel_iommu *iommu)
 	 * global invalidation of interrupt entry cache before disabling
 	 * interrupt-remapping.
 	 */
-	qi_global_iec(iommu);
+	if (!cap_esirtps(iommu->cap))
+		qi_global_iec(iommu);
 
 	raw_spin_lock_irqsave(&iommu->register_lock, flags);
 
-- 
2.34.1

[PATCH 6/6] iommu/vt-d: Avoid unnecessary global DMA cache invalidation

[Lu Baolu]

Some VT-d hardware implementations invalidate all DMA remapping hardware
translation caches as part of SRTP flow. The VT-d spec adds a ESRTPS
(Enhanced Set Root Table Pointer Support, section 11.4.2 in VT-d spec)
capability bit to indicate this. With this bit set, software has no need
to issue the global invalidation request.

Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Jerry Snitselaar <jsnitsel@redhat.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20220919062523.3438951-3-baolu.lu@linux.intel.com
---
 drivers/iommu/intel/iommu.h | 1 +
 drivers/iommu/intel/iommu.c | 7 +++++++
 2 files changed, 8 insertions(+)

diff --git a/drivers/iommu/intel/iommu.h b/drivers/iommu/intel/iommu.h
index bddf6c69587d..92023dff9513 100644
--- a/drivers/iommu/intel/iommu.h
+++ b/drivers/iommu/intel/iommu.h
@@ -146,6 +146,7 @@
 /*
  * Decoding Capability Register
  */
+#define cap_esrtps(c)		(((c) >> 63) & 1)
 #define cap_esirtps(c)		(((c) >> 62) & 1)
 #define cap_fl5lp_support(c)	(((c) >> 60) & 1)
 #define cap_pi_support(c)	(((c) >> 59) & 1)
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 81a449b5cb91..a8b36c3fddf1 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -1239,6 +1239,13 @@ static void iommu_set_root_entry(struct intel_iommu *iommu)
 
 	raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
 
+	/*
+	 * Hardware invalidates all DMA remapping hardware translation
+	 * caches as part of SRTP flow.
+	 */
+	if (cap_esrtps(iommu->cap))
+		return;
+
 	iommu->flush.flush_context(iommu, 0, 0, 0, DMA_CCMD_GLOBAL_INVL);
 	if (sm_supported(iommu))
 		qi_flush_pasid_cache(iommu, 0, QI_PC_GLOBAL, 0);
-- 
2.34.1

Re: [PATCH 0/6] [PULL REQUEST] Intel IOMMU updates for Linux v6.1

[Joerg Roedel]

Hi Baolu,

On Fri, Sep 23, 2022 at 08:42:00AM +0800, Lu Baolu wrote:
> Lu Baolu (5):
>   iommu/vt-d: Remove unnecessary SVA data accesses in page fault path
>   iommu/vt-d: Decouple PASID & PRI enabling from SVA
>   iommu/vt-d: Remove pasid_set_eafe()
>   iommu/vt-d: Avoid unnecessary global IRTE cache invalidation
>   iommu/vt-d: Avoid unnecessary global DMA cache invalidation
> 
> Yi Liu (1):
>   iommu/vt-d: Rename cap_5lp_support to cap_fl5lp_support

These don't apply cleanly on v6.0-rc4 or -rc7. Please rebase these
patches to my x86/vt-d branch and re-send.

Thanks,

	Joerg

Re: [PATCH 0/6] [PULL REQUEST] Intel IOMMU updates for Linux v6.1

[Baolu Lu]

Hi Joerg,

On 2022/9/26 20:15, Joerg Roedel wrote:
> Hi Baolu,
> 
> On Fri, Sep 23, 2022 at 08:42:00AM +0800, Lu Baolu wrote:
>> Lu Baolu (5):
>>    iommu/vt-d: Remove unnecessary SVA data accesses in page fault path
>>    iommu/vt-d: Decouple PASID & PRI enabling from SVA
>>    iommu/vt-d: Remove pasid_set_eafe()
>>    iommu/vt-d: Avoid unnecessary global IRTE cache invalidation
>>    iommu/vt-d: Avoid unnecessary global DMA cache invalidation
>>
>> Yi Liu (1):
>>    iommu/vt-d: Rename cap_5lp_support to cap_fl5lp_support
> 
> These don't apply cleanly on v6.0-rc4 or -rc7. Please rebase these
> patches to my x86/vt-d branch and re-send.

This pull request is based on the next branch. There's a conflict
between

de9f8a91eb32 iommu/dma: Clean up Kconfig (core branch)

and

iommu/vt-d: Decouple PASID & PRI enabling from SVA (this series).

Do you prefer applying de9f8a91eb32 to vt-d branch, or let me send a new
v6.0-rc7 based pull request?

Best regards,
baolu

Re: [PATCH 0/6] [PULL REQUEST] Intel IOMMU updates for Linux v6.1

[Joerg Roedel]

On Mon, Sep 26, 2022 at 08:38:09PM +0800, Baolu Lu wrote:
> Do you prefer applying de9f8a91eb32 to vt-d branch, or let me send a new
> v6.0-rc7 based pull request?

Yes, please send a v6.0-rc7 pull-request. I'd like to carry the conflict
resolution in the merge commit for next.

Regards,

	Joerg

Re: [PATCH 0/6] [PULL REQUEST] Intel IOMMU updates for Linux v6.1

[Baolu Lu]

On 2022/9/26 21:01, Joerg Roedel wrote:
> On Mon, Sep 26, 2022 at 08:38:09PM +0800, Baolu Lu wrote:
>> Do you prefer applying de9f8a91eb32 to vt-d branch, or let me send a new
>> v6.0-rc7 based pull request?
> Yes, please send a v6.0-rc7 pull-request. I'd like to carry the conflict
> resolution in the merge commit for next.

Sure. A new pull request has been posted:

https://lore.kernel.org/linux-iommu/20220926131529.4045281-1-baolu.lu@linux.intel.com/

Best regards,
baolu

[RESEND PATCH 0/6] [PULL REQUEST] Intel IOMMU updates for Linux v6.1

[Lu Baolu]

Hi Joerg,

This is a resent pull request. It's based on v6.0-rc7. The original pull
request was posted here:

https://lore.kernel.org/linux-iommu/20220923004206.3630441-1-baolu.lu@linux.intel.com/

Please pull.

Best regards,
Baolu


Lu Baolu (5):
  iommu/vt-d: Remove unnecessary SVA data accesses in page fault path
  iommu/vt-d: Decouple PASID & PRI enabling from SVA
  iommu/vt-d: Remove pasid_set_eafe()
  iommu/vt-d: Avoid unnecessary global IRTE cache invalidation
  iommu/vt-d: Avoid unnecessary global DMA cache invalidation

Yi Liu (1):
  iommu/vt-d: Rename cap_5lp_support to cap_fl5lp_support

 drivers/iommu/intel/iommu.h         |  7 ++-
 drivers/iommu/intel/cap_audit.c     |  4 +-
 drivers/iommu/intel/iommu.c         | 87 ++++++++---------------------
 drivers/iommu/intel/irq_remapping.c |  6 +-
 drivers/iommu/intel/pasid.c         | 12 +---
 drivers/iommu/intel/svm.c           | 62 +++-----------------
 drivers/iommu/intel/Kconfig         |  5 +-
 7 files changed, 45 insertions(+), 138 deletions(-)

-- 
2.34.1

[PATCH 2/6] iommu/vt-d: Decouple PASID & PRI enabling from SVA

[Lu Baolu]

Previously the PCI PASID and PRI capabilities are enabled in the path of
iommu device probe only if INTEL_IOMMU_SVM is configured and the device
supports ATS. As we've already decoupled the I/O page fault handler from
SVA, we could also decouple PASID and PRI enabling from it to make room
for growth of new features like kernel DMA with PASID, SIOV and nested
translation.

At the same time, the iommu_enable_dev_iotlb() helper is also called in
iommu_dev_enable_feature(dev, IOMMU_DEV_FEAT_SVA) path. It's unnecessary
and duplicate. This cleanups this helper to make the code neat.

Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20220915085814.2261409-1-baolu.lu@linux.intel.com
---
 drivers/iommu/intel/iommu.h |  1 -
 drivers/iommu/intel/iommu.c | 78 ++++++++-----------------------------
 drivers/iommu/intel/Kconfig |  5 +--
 3 files changed, 18 insertions(+), 66 deletions(-)

diff --git a/drivers/iommu/intel/iommu.h b/drivers/iommu/intel/iommu.h
index b5fb7706e97c..8f29a183467d 100644
--- a/drivers/iommu/intel/iommu.h
+++ b/drivers/iommu/intel/iommu.h
@@ -742,7 +742,6 @@ extern int dmar_ir_support(void);
 void *alloc_pgtable_page(int node);
 void free_pgtable_page(void *vaddr);
 void iommu_flush_write_buffer(struct intel_iommu *iommu);
-int intel_iommu_enable_pasid(struct intel_iommu *iommu, struct device *dev);
 struct intel_iommu *device_to_iommu(struct device *dev, u8 *bus, u8 *devfn);
 
 #ifdef CONFIG_INTEL_IOMMU_SVM
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 31bc50e538a3..af17177b6d76 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -199,6 +199,11 @@ static inline void context_set_domain_id(struct context_entry *context,
 	context->hi |= (value & ((1 << 16) - 1)) << 8;
 }
 
+static inline void context_set_pasid(struct context_entry *context)
+{
+	context->lo |= CONTEXT_PASIDE;
+}
+
 static inline int context_domain_id(struct context_entry *c)
 {
 	return((c->hi >> 8) & 0xffff);
@@ -1350,21 +1355,18 @@ static void __iommu_flush_iotlb(struct intel_iommu *iommu, u16 did,
 }
 
 static struct device_domain_info *
-iommu_support_dev_iotlb(struct dmar_domain *domain, struct intel_iommu *iommu,
-			u8 bus, u8 devfn)
+domain_lookup_dev_info(struct dmar_domain *domain,
+		       struct intel_iommu *iommu, u8 bus, u8 devfn)
 {
 	struct device_domain_info *info;
 	unsigned long flags;
 
-	if (!iommu->qi)
-		return NULL;
-
 	spin_lock_irqsave(&domain->lock, flags);
 	list_for_each_entry(info, &domain->devices, link) {
 		if (info->iommu == iommu && info->bus == bus &&
 		    info->devfn == devfn) {
 			spin_unlock_irqrestore(&domain->lock, flags);
-			return info->ats_supported ? info : NULL;
+			return info;
 		}
 	}
 	spin_unlock_irqrestore(&domain->lock, flags);
@@ -1389,7 +1391,7 @@ static void domain_update_iotlb(struct dmar_domain *domain)
 	spin_unlock_irqrestore(&domain->lock, flags);
 }
 
-static void iommu_enable_dev_iotlb(struct device_domain_info *info)
+static void iommu_enable_pci_caps(struct device_domain_info *info)
 {
 	struct pci_dev *pdev;
 
@@ -1412,7 +1414,6 @@ static void iommu_enable_dev_iotlb(struct device_domain_info *info)
 		info->pfsid = pci_dev_id(pf_pdev);
 	}
 
-#ifdef CONFIG_INTEL_IOMMU_SVM
 	/* The PCIe spec, in its wisdom, declares that the behaviour of
 	   the device if you enable PASID support after ATS support is
 	   undefined. So always enable PASID support on devices which
@@ -1425,7 +1426,7 @@ static void iommu_enable_dev_iotlb(struct device_domain_info *info)
 	    (info->pasid_enabled ? pci_prg_resp_pasid_required(pdev) : 1)  &&
 	    !pci_reset_pri(pdev) && !pci_enable_pri(pdev, PRQ_DEPTH))
 		info->pri_enabled = 1;
-#endif
+
 	if (info->ats_supported && pci_ats_page_aligned(pdev) &&
 	    !pci_enable_ats(pdev, VTD_PAGE_SHIFT)) {
 		info->ats_enabled = 1;
@@ -1448,16 +1449,16 @@ static void iommu_disable_dev_iotlb(struct device_domain_info *info)
 		info->ats_enabled = 0;
 		domain_update_iotlb(info->domain);
 	}
-#ifdef CONFIG_INTEL_IOMMU_SVM
+
 	if (info->pri_enabled) {
 		pci_disable_pri(pdev);
 		info->pri_enabled = 0;
 	}
+
 	if (info->pasid_enabled) {
 		pci_disable_pasid(pdev);
 		info->pasid_enabled = 0;
 	}
-#endif
 }
 
 static void __iommu_flush_dev_iotlb(struct device_domain_info *info,
@@ -1907,7 +1908,7 @@ static int domain_context_mapping_one(struct dmar_domain *domain,
 				      u8 bus, u8 devfn)
 {
 	struct device_domain_info *info =
-			iommu_support_dev_iotlb(domain, iommu, bus, devfn);
+			domain_lookup_dev_info(domain, iommu, bus, devfn);
 	u16 did = domain_id_iommu(domain, iommu);
 	int translation = CONTEXT_TT_MULTI_LEVEL;
 	struct context_entry *context;
@@ -1980,6 +1981,8 @@ static int domain_context_mapping_one(struct dmar_domain *domain,
 			context_set_sm_dte(context);
 		if (info && info->pri_supported)
 			context_set_sm_pre(context);
+		if (info && info->pasid_supported)
+			context_set_pasid(context);
 	} else {
 		struct dma_pte *pgd = domain->pgd;
 		int agaw;
@@ -2037,7 +2040,7 @@ static int domain_context_mapping_one(struct dmar_domain *domain,
 	} else {
 		iommu_flush_write_buffer(iommu);
 	}
-	iommu_enable_dev_iotlb(info);
+	iommu_enable_pci_caps(info);
 
 	ret = 0;
 
@@ -4574,52 +4577,6 @@ static void intel_iommu_get_resv_regions(struct device *device,
 	list_add_tail(&reg->list, head);
 }
 
-int intel_iommu_enable_pasid(struct intel_iommu *iommu, struct device *dev)
-{
-	struct device_domain_info *info = dev_iommu_priv_get(dev);
-	struct context_entry *context;
-	struct dmar_domain *domain;
-	u64 ctx_lo;
-	int ret;
-
-	domain = info->domain;
-	if (!domain)
-		return -EINVAL;
-
-	spin_lock(&iommu->lock);
-	ret = -EINVAL;
-	if (!info->pasid_supported)
-		goto out;
-
-	context = iommu_context_addr(iommu, info->bus, info->devfn, 0);
-	if (WARN_ON(!context))
-		goto out;
-
-	ctx_lo = context[0].lo;
-
-	if (!(ctx_lo & CONTEXT_PASIDE)) {
-		ctx_lo |= CONTEXT_PASIDE;
-		context[0].lo = ctx_lo;
-		wmb();
-		iommu->flush.flush_context(iommu,
-					   domain_id_iommu(domain, iommu),
-					   PCI_DEVID(info->bus, info->devfn),
-					   DMA_CCMD_MASK_NOBIT,
-					   DMA_CCMD_DEVICE_INVL);
-	}
-
-	/* Enable PASID support in the device, if it wasn't already */
-	if (!info->pasid_enabled)
-		iommu_enable_dev_iotlb(info);
-
-	ret = 0;
-
- out:
-	spin_unlock(&iommu->lock);
-
-	return ret;
-}
-
 static struct iommu_group *intel_iommu_device_group(struct device *dev)
 {
 	if (dev_is_pci(dev))
@@ -4643,9 +4600,6 @@ static int intel_iommu_enable_sva(struct device *dev)
 	if (!(iommu->flags & VTD_FLAG_SVM_CAPABLE))
 		return -ENODEV;
 
-	if (intel_iommu_enable_pasid(iommu, dev))
-		return -ENODEV;
-
 	if (!info->pasid_enabled || !info->pri_enabled || !info->ats_enabled)
 		return -EINVAL;
 
diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
index 39a06d245f12..cd0ec7ed48b6 100644
--- a/drivers/iommu/intel/Kconfig
+++ b/drivers/iommu/intel/Kconfig
@@ -21,6 +21,8 @@ config INTEL_IOMMU
 	select IOASID
 	select IOMMU_DMA
 	select PCI_ATS
+	select PCI_PRI
+	select PCI_PASID
 	help
 	  DMA remapping (DMAR) devices support enables independent address
 	  translations for Direct Memory Access (DMA) from devices.
@@ -48,10 +50,7 @@ config INTEL_IOMMU_DEBUGFS
 config INTEL_IOMMU_SVM
 	bool "Support for Shared Virtual Memory with Intel IOMMU"
 	depends on X86_64
-	select PCI_PASID
-	select PCI_PRI
 	select MMU_NOTIFIER
-	select IOASID
 	select IOMMU_SVA
 	help
 	  Shared Virtual Memory (SVM) provides a facility for devices
-- 
2.34.1