iommu: Handle freelists when using deferred flushing in iommu drivers

Allow the iommu_unmap_fast to return newly freed page table pages and
pass the freelist to queue_iova in the dma-iommu ops path.

This is useful for iommu drivers (in this case the intel iommu driver)
which need to wait for the ioTLB to be flushed before newly
free/unmapped page table pages can be freed. This way we can still batch
ioTLB free operations and handle the freelists.

Signed-off-by: Tom Murphy <murphyt7@tcd.ie>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Tested-by: Logan Gunthorpe <logang@deltatee.com>
Link: https://lore.kernel.org/r/20201124082057.2614359-2-baolu.lu@linux.intel.com
Signed-off-by: Will Deacon <will@kernel.org>

1 files changed, 36 insertions, 19 deletions

diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 7b32703c0b47..fed0b3f0e25b 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -1243,17 +1243,17 @@ next:
    pages can only be freed after the IOTLB flush has been done. */
 static struct page *domain_unmap(struct dmar_domain *domain,
 				 unsigned long start_pfn,
-				 unsigned long last_pfn)
+				 unsigned long last_pfn,
+				 struct page *freelist)
 {
-	struct page *freelist;
-
 	BUG_ON(!domain_pfn_supported(domain, start_pfn));
 	BUG_ON(!domain_pfn_supported(domain, last_pfn));
 	BUG_ON(start_pfn > last_pfn);
 
 	/* we don't need lock here; nobody else touches the iova range */
 	freelist = dma_pte_clear_level(domain, agaw_to_level(domain->agaw),
-				       domain->pgd, 0, start_pfn, last_pfn, NULL);
+				       domain->pgd, 0, start_pfn, last_pfn,
+				       freelist);
 
 	/* free pgd */
 	if (start_pfn == 0 && last_pfn == DOMAIN_MAX_PFN(domain->gaw)) {
@@ -2011,7 +2011,8 @@ static void domain_exit(struct dmar_domain *domain)
 	if (domain->pgd) {
 		struct page *freelist;
 
-		freelist = domain_unmap(domain, 0, DOMAIN_MAX_PFN(domain->gaw));
+		freelist = domain_unmap(domain, 0,
+					DOMAIN_MAX_PFN(domain->gaw), NULL);
 		dma_free_pagelist(freelist);
 	}
 
@@ -3570,7 +3571,7 @@ static void intel_unmap(struct device *dev, dma_addr_t dev_addr, size_t size)
 	if (dev_is_pci(dev))
 		pdev = to_pci_dev(dev);
 
-	freelist = domain_unmap(domain, start_pfn, last_pfn);
+	freelist = domain_unmap(domain, start_pfn, last_pfn, NULL);
 	if (intel_iommu_strict || (pdev && pdev->untrusted) ||
 			!has_iova_flush_queue(&domain->iovad)) {
 		iommu_flush_iotlb_psi(iommu, domain, start_pfn,
@@ -4636,7 +4637,8 @@ static int intel_iommu_memory_notifier(struct notifier_block *nb,
 			struct page *freelist;
 
 			freelist = domain_unmap(si_domain,
-						start_vpfn, last_vpfn);
+						start_vpfn, last_vpfn,
+						NULL);
 
 			rcu_read_lock();
 			for_each_active_iommu(iommu, drhd)
@@ -5608,10 +5610,8 @@ static size_t intel_iommu_unmap(struct iommu_domain *domain,
 				struct iommu_iotlb_gather *gather)
 {
 	struct dmar_domain *dmar_domain = to_dmar_domain(domain);
-	struct page *freelist = NULL;
 	unsigned long start_pfn, last_pfn;
-	unsigned int npages;
-	int iommu_id, level = 0;
+	int level = 0;
 
 	/* Cope with horrid API which requires us to unmap more than the
 	   size argument if it happens to be a large-page mapping. */
@@ -5623,22 +5623,38 @@ static size_t intel_iommu_unmap(struct iommu_domain *domain,
 	start_pfn = iova >> VTD_PAGE_SHIFT;
 	last_pfn = (iova + size - 1) >> VTD_PAGE_SHIFT;
 
-	freelist = domain_unmap(dmar_domain, start_pfn, last_pfn);
-
-	npages = last_pfn - start_pfn + 1;
-
-	for_each_domain_iommu(iommu_id, dmar_domain)
-		iommu_flush_iotlb_psi(g_iommus[iommu_id], dmar_domain,
-				      start_pfn, npages, !freelist, 0);
-
-	dma_free_pagelist(freelist);
+	gather->freelist = domain_unmap(dmar_domain, start_pfn,
+					last_pfn, gather->freelist);
 
 	if (dmar_domain->max_addr == iova + size)
 		dmar_domain->max_addr = iova;
 
+	iommu_iotlb_gather_add_page(domain, gather, iova, size);
+
 	return size;
 }
 
+static void intel_iommu_tlb_sync(struct iommu_domain *domain,
+				 struct iommu_iotlb_gather *gather)
+{
+	struct dmar_domain *dmar_domain = to_dmar_domain(domain);
+	unsigned long iova_pfn = IOVA_PFN(gather->start);
+	size_t size = gather->end - gather->start;
+	unsigned long start_pfn, last_pfn;
+	unsigned long nrpages;
+	int iommu_id;
+
+	nrpages = aligned_nrpages(gather->start, size);
+	start_pfn = mm_to_dma_pfn(iova_pfn);
+	last_pfn = start_pfn + nrpages - 1;
+
+	for_each_domain_iommu(iommu_id, dmar_domain)
+		iommu_flush_iotlb_psi(g_iommus[iommu_id], dmar_domain,
+				      start_pfn, nrpages, !gather->freelist, 0);
+
+	dma_free_pagelist(gather->freelist);
+}
+
 static phys_addr_t intel_iommu_iova_to_phys(struct iommu_domain *domain,
 					    dma_addr_t iova)
 {
@@ -6098,6 +6114,7 @@ const struct iommu_ops intel_iommu_ops = {
 	.aux_get_pasid		= intel_iommu_aux_get_pasid,
 	.map			= intel_iommu_map,
 	.unmap			= intel_iommu_unmap,
+	.iotlb_sync		= intel_iommu_tlb_sync,
 	.iova_to_phys		= intel_iommu_iova_to_phys,
 	.probe_device		= intel_iommu_probe_device,
 	.probe_finalize		= intel_iommu_probe_finalize,