Skip to contentSkip to conversation

iommu/vt-d: Avoid iova flush queue in strict mode

PATCH In review Lu Baolu 1 participant 4 messages Dec 19, 2019 – Dec 28, 2019 IOMMU

lore
Share & tools
lore

Lu Baolu

author
#0 lore raw

PATCH 1/2 iommu/vt-d: Avoid iova flush queue in strict mode

If Intel IOMMU strict mode is enabled by users, it's unnecessary to create the iova flush queue.

---
 drivers/iommu/intel-iommu.c | 24 +++++++++++++++---------
 1 file changed, 15 insertions(+), 9 deletions(-)
drivers/iommu/intel-iommu.c +15 -9
index 0726705d2d89..93e1d7c3ac7c 100644
@@ -1858,10 +1858,12 @@ static int domain_init(struct dmar_domain *domain, struct intel_iommu *iommu,
 
 	init_iova_domain(&domain->iovad, VTD_PAGE_SIZE, IOVA_START_PFN);
 
-	err = init_iova_flush_queue(&domain->iovad,
-				    iommu_flush_iova, iova_entry_free);
-	if (err)
-		return err;
+	if (!intel_iommu_strict) {
+		err = init_iova_flush_queue(&domain->iovad,
+					    iommu_flush_iova, iova_entry_free);
+		if (err)
+			return err;
+	}
 
 	domain_reserve_special_ranges(domain);
 
@@ -5199,6 +5201,7 @@ static struct iommu_domain *intel_iommu_domain_alloc(unsigned type)
 {
 	struct dmar_domain *dmar_domain;
 	struct iommu_domain *domain;
+	int ret;
 
 	switch (type) {
 	case IOMMU_DOMAIN_DMA:
@@ -5215,11 +5218,14 @@ static struct iommu_domain *intel_iommu_domain_alloc(unsigned type)
 			return NULL;
 		}
 
-		if (type == IOMMU_DOMAIN_DMA &&
-		    init_iova_flush_queue(&dmar_domain->iovad,
-					  iommu_flush_iova, iova_entry_free)) {
-			pr_warn("iova flush queue initialization failed\n");
-			intel_iommu_strict = 1;
+		if (!intel_iommu_strict && type == IOMMU_DOMAIN_DMA) {
+			ret = init_iova_flush_queue(&dmar_domain->iovad,
+						    iommu_flush_iova,
+						    iova_entry_free);
+			if (ret) {
+				pr_warn("iova flush queue initialization failed\n");
+				intel_iommu_strict = 1;
+			}
 		}
 
 		domain_update_iommu_cap(dmar_domain);
Signature · 1 line
2.17.1
author
#1 lore raw

PATCH 2/2 iommu/vt-d: Loose requirement for flush queue initializaton

Currently if flush queue initialization fails, we return error or enforce the system-wide strict mode. These are unnecessary because we always check the existence of a flush queue before queuing any iova's for lazy flushing. Printing a informational message is enough.

---
 drivers/iommu/intel-iommu.c | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)
drivers/iommu/intel-iommu.c +6 -8
index 93e1d7c3ac7c..65eee8cf6d59 100644
@@ -1854,15 +1854,15 @@ static int domain_init(struct dmar_domain *domain, struct intel_iommu *iommu,
 {
 	int adjust_width, agaw;
 	unsigned long sagaw;
-	int err;
+	int ret;
 
 	init_iova_domain(&domain->iovad, VTD_PAGE_SIZE, IOVA_START_PFN);
 
 	if (!intel_iommu_strict) {
-		err = init_iova_flush_queue(&domain->iovad,
+		ret = init_iova_flush_queue(&domain->iovad,
 					    iommu_flush_iova, iova_entry_free);
-		if (err)
-			return err;
+		if (ret)
+			pr_info("iova flush queue initialization failed\n");
 	}
 
 	domain_reserve_special_ranges(domain);
@@ -5222,10 +5222,8 @@ static struct iommu_domain *intel_iommu_domain_alloc(unsigned type)
 			ret = init_iova_flush_queue(&dmar_domain->iovad,
 						    iommu_flush_iova,
 						    iova_entry_free);
-			if (ret) {
-				pr_warn("iova flush queue initialization failed\n");
-				intel_iommu_strict = 1;
-			}
+			if (ret)
+				pr_info("iova flush queue initialization failed\n");
 		}
 
 		domain_update_iommu_cap(dmar_domain);
Signature · 1 line
2.17.1
author
#2 lore raw

Re: [PATCH 2/2] iommu/vt-d: Loose requirement for flush queue initializaton

In reply to Lu Baolu
Quoted 7 lines · Lu Baolu

Currently if flush queue initialization fails, we return error or enforce the system-wide strict mode. These are unnecessary because we always check the existence of a flush queue before queuing any iova's for lazy flushing. Printing a informational message is enough.

Queued for v5.6.

Thanks,
-baolu
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu