iommu/s390: add support for IOMMU passthrough

index 474e1f8d1d3c..8fe4c7a72c0b 100644
@@ -144,7 +144,7 @@ struct zpci_dev {
 	u8		util_str_avail	: 1;
 	u8		irqs_registered	: 1;
 	u8		tid_avail	: 1;
-	u8		reserved	: 1;
+	u8		rtr_avail	: 1; /* Relaxed translation allowed */
 	unsigned int	devfn;		/* DEVFN part of the RID*/
 
 	u8 pfip[CLP_PFIP_NR_SEGMENTS];	/* pci function internal path */

index 3fff2f7095c8..7ebff39c84b3 100644
@@ -156,7 +156,9 @@ struct clp_rsp_query_pci_grp {
 	u16			:  4;
 	u16 noi			: 12;	/* number of interrupts */
 	u8 version;
-	u8			:  6;
+	u8			:  2;
+	u8 rtr			:  1;	/* Relaxed translation requirement */
+	u8			:  3;
 	u8 frame		:  1;
 	u8 refresh		:  1;	/* TLB refresh mode */
 	u16			:  3;

index 14bf7e8d06b7..27248686e588 100644
@@ -112,6 +112,7 @@ static void clp_store_query_pci_fngrp(struct zpci_dev *zdev,
 	zdev->version = response->version;
 	zdev->maxstbl = response->maxstbl;
 	zdev->dtsm = response->dtsm;
+	zdev->rtr_avail = response->rtr;
 
 	switch (response->version) {
 	case 1:

index 857afbc4828f..3fa113c4ad8f 100644
@@ -19,6 +19,7 @@
 #include <linux/jump_label.h>
 #include <linux/pci.h>
 #include <linux/printk.h>
+#include <linux/dma-direct.h>
 
 #include <asm/pci_clp.h>
 #include <asm/pci_dma.h>

@@ -283,10 +284,34 @@ static struct zpci_bus *zpci_bus_alloc(int topo, bool topo_is_tid)
 	return zbus;
 }
 
+static void pci_dma_range_setup(struct pci_dev *pdev)
+{
+	struct zpci_dev *zdev = to_zpci(pdev);
+	struct bus_dma_region *map;
+	u64 aligned_end;
+
+	map = kzalloc(sizeof(*map), GFP_KERNEL);
+	if (!map)
+		return;
+
+	map->cpu_start = 0;
+	map->dma_start = PAGE_ALIGN(zdev->start_dma);
+	aligned_end = PAGE_ALIGN_DOWN(zdev->end_dma + 1);
+	if (aligned_end >= map->dma_start)
+		map->size = aligned_end - map->dma_start;
+	else
+		map->size = 0;
+	WARN_ON_ONCE(map->size == 0);
+
+	pdev->dev.dma_range_map = map;
+}
+
 void pcibios_bus_add_device(struct pci_dev *pdev)
 {
 	struct zpci_dev *zdev = to_zpci(pdev);
 
+	pci_dma_range_setup(pdev);
+
 	/*
 	 * With pdev->no_vf_scan the common PCI probing code does not
 	 * perform PF/VF linking.

index 8fe4c7a72c0b..7cc6a7cc66e7 100644
@@ -217,6 +217,7 @@ extern struct airq_iv *zpci_aif_sbv;
 struct zpci_dev *zpci_create_device(u32 fid, u32 fh, enum zpci_state state);
 int zpci_add_device(struct zpci_dev *zdev);
 int zpci_enable_device(struct zpci_dev *);
+int zpci_reenable_device(struct zpci_dev *zdev);
 int zpci_disable_device(struct zpci_dev *);
 int zpci_scan_configured_device(struct zpci_dev *zdev, u32 fh);
 int zpci_deconfigure_device(struct zpci_dev *zdev);

@@ -245,6 +246,7 @@ void update_uid_checking(bool new);
 /* IOMMU Interface */
 int zpci_init_iommu(struct zpci_dev *zdev);
 void zpci_destroy_iommu(struct zpci_dev *zdev);
+int zpci_iommu_register_ioat(struct zpci_dev *zdev, u8 *status);
 
 #ifdef CONFIG_PCI
 static inline bool zpci_use_mio(struct zpci_dev *zdev)

index 9b9e7fdd5380..8c40154ff50f 100644
@@ -433,7 +433,6 @@ static void kvm_s390_pci_dev_release(struct zpci_dev *zdev)
 static int kvm_s390_pci_register_kvm(void *opaque, struct kvm *kvm)
 {
 	struct zpci_dev *zdev = opaque;
-	u8 status;
 	int rc;
 
 	if (!zdev)

@@ -480,13 +479,7 @@ static int kvm_s390_pci_register_kvm(void *opaque, struct kvm *kvm)
 	 */
 	zdev->gisa = (u32)virt_to_phys(&kvm->arch.sie_page2->gisa);
 
-	rc = zpci_enable_device(zdev);
-	if (rc)
-		goto clear_gisa;
-
-	/* Re-register the IOMMU that was already created */
-	rc = zpci_register_ioat(zdev, 0, zdev->start_dma, zdev->end_dma,
-				virt_to_phys(zdev->dma_table), &status);
+	rc = zpci_reenable_device(zdev);
 	if (rc)
 		goto clear_gisa;
 

@@ -516,7 +509,6 @@ static void kvm_s390_pci_unregister_kvm(void *opaque)
 {
 	struct zpci_dev *zdev = opaque;
 	struct kvm *kvm;
-	u8 status;
 
 	if (!zdev)
 		return;

@@ -550,12 +542,7 @@ static void kvm_s390_pci_unregister_kvm(void *opaque)
 			goto out;
 	}
 
-	if (zpci_enable_device(zdev))
-		goto out;
-
-	/* Re-register the IOMMU that was already created */
-	zpci_register_ioat(zdev, 0, zdev->start_dma, zdev->end_dma,
-			   virt_to_phys(zdev->dma_table), &status);
+	zpci_reenable_device(zdev);
 
 out:
 	spin_lock(&kvm->arch.kzdev_list_lock);

index 88f72745fa59..fa879351efb5 100644
@@ -124,14 +124,13 @@ int zpci_register_ioat(struct zpci_dev *zdev, u8 dmaas,
 	struct zpci_fib fib = {0};
 	u8 cc;
 
-	WARN_ON_ONCE(iota & 0x3fff);
 	fib.pba = base;
 	/* Work around off by one in ISM virt device */
 	if (zdev->pft == PCI_FUNC_TYPE_ISM && limit > base)
 		fib.pal = limit + (1 << 12);
 	else
 		fib.pal = limit;
-	fib.iota = iota | ZPCI_IOTA_RTTO_FLAG;
+	fib.iota = iota;
 	fib.gd = zdev->gisa;
 	cc = zpci_mod_fc(req, &fib, status);
 	if (cc)

@@ -690,6 +689,23 @@ int zpci_enable_device(struct zpci_dev *zdev)
 }
 EXPORT_SYMBOL_GPL(zpci_enable_device);
 
+int zpci_reenable_device(struct zpci_dev *zdev)
+{
+	u8 status;
+	int rc;
+
+	rc = zpci_enable_device(zdev);
+	if (rc)
+		return rc;
+
+	rc = zpci_iommu_register_ioat(zdev, &status);
+	if (rc)
+		zpci_disable_device(zdev);
+
+	return rc;
+}
+EXPORT_SYMBOL_GPL(zpci_reenable_device);
+
 int zpci_disable_device(struct zpci_dev *zdev)
 {
 	u32 fh = zdev->fh;

@@ -739,7 +755,6 @@ EXPORT_SYMBOL_GPL(zpci_disable_device);
  */
 int zpci_hot_reset_device(struct zpci_dev *zdev)
 {
-	u8 status;
 	int rc;
 
 	lockdep_assert_held(&zdev->state_lock);

@@ -758,19 +773,9 @@ int zpci_hot_reset_device(struct zpci_dev *zdev)
 			return rc;
 	}
 
-	rc = zpci_enable_device(zdev);
-	if (rc)
-		return rc;
-
-	if (zdev->dma_table)
-		rc = zpci_register_ioat(zdev, 0, zdev->start_dma, zdev->end_dma,
-					virt_to_phys(zdev->dma_table), &status);
-	if (rc) {
-		zpci_disable_device(zdev);
-		return rc;
-	}
+	rc = zpci_reenable_device(zdev);
 
-	return 0;
+	return rc;
 }
 
 /**

index 2de1ea6c3a8c..0ecad08e1b1e 100644
@@ -52,7 +52,6 @@ static DEVICE_ATTR_RO(mio_enabled);
 
 static int _do_recover(struct pci_dev *pdev, struct zpci_dev *zdev)
 {
-	u8 status;
 	int ret;
 
 	pci_stop_and_remove_bus_device(pdev);

@@ -70,16 +69,8 @@ static int _do_recover(struct pci_dev *pdev, struct zpci_dev *zdev)
 			return ret;
 	}
 
-	ret = zpci_enable_device(zdev);
-	if (ret)
-		return ret;
+	ret = zpci_reenable_device(zdev);
 
-	if (zdev->dma_table) {
-		ret = zpci_register_ioat(zdev, 0, zdev->start_dma, zdev->end_dma,
-					 virt_to_phys(zdev->dma_table), &status);
-		if (ret)
-			zpci_disable_device(zdev);
-	}
 	return ret;
 }
 

55 lines not on this page · raw message

index 007ccfdad495..e1c76e0f9c2b 100644
@@ -16,7 +16,7 @@
 
 #include "dma-iommu.h"
 
-static const struct iommu_ops s390_iommu_ops;
+static const struct iommu_ops s390_iommu_ops, s390_iommu_rtr_ops;
 
 static struct kmem_cache *dma_region_table_cache;
 static struct kmem_cache *dma_page_table_cache;

@@ -432,9 +432,11 @@ static int blocking_domain_attach_device(struct iommu_domain *domain,
 		return 0;
 
 	s390_domain = to_s390_domain(zdev->s390_domain);
-	spin_lock_irqsave(&s390_domain->list_lock, flags);
-	list_del_rcu(&zdev->iommu_list);
-	spin_unlock_irqrestore(&s390_domain->list_lock, flags);
+	if (zdev->dma_table) {
+		spin_lock_irqsave(&s390_domain->list_lock, flags);
+		list_del_rcu(&zdev->iommu_list);
+		spin_unlock_irqrestore(&s390_domain->list_lock, flags);
+	}
 
 	zpci_unregister_ioat(zdev, 0);
 	zdev->dma_table = NULL;

@@ -762,7 +764,13 @@ int zpci_init_iommu(struct zpci_dev *zdev)
 	if (rc)
 		goto out_err;
 
-	rc = iommu_device_register(&zdev->iommu_dev, &s390_iommu_ops, NULL);
+	if (zdev->rtr_avail) {
+		rc = iommu_device_register(&zdev->iommu_dev,
+					   &s390_iommu_rtr_ops, NULL);
+	} else {
+		rc = iommu_device_register(&zdev->iommu_dev, &s390_iommu_ops,
+					   NULL);
+	}
 	if (rc)
 		goto out_sysfs;
 

@@ -826,6 +834,39 @@ static int __init s390_iommu_init(void)
 }
 subsys_initcall(s390_iommu_init);
 
+static int s390_attach_dev_identity(struct iommu_domain *domain,
+				    struct device *dev)
+{
+	struct zpci_dev *zdev = to_zpci_dev(dev);
+	u8 status;
+	int cc;
+
+	blocking_domain_attach_device(&blocking_domain, dev);
+
+	/* If we fail now DMA remains blocked via blocking domain */
+	cc = s390_iommu_domain_reg_ioat(zdev, domain, &status);
+
+	/*
+	 * If the device is undergoing error recovery the reset code
+	 * will re-establish the new domain.
+	 */
+	if (cc && status != ZPCI_PCI_ST_FUNC_NOT_AVAIL)
+		return -EIO;
+
+	zdev_s390_domain_update(zdev, domain);
+
+	return 0;
+}
+
+static const struct iommu_domain_ops s390_identity_ops = {
+	.attach_dev = s390_attach_dev_identity,
+};
+
+static struct iommu_domain s390_identity_domain = {
+	.type = IOMMU_DOMAIN_IDENTITY,
+	.ops = &s390_identity_ops,
+};
+
 static struct iommu_domain blocking_domain = {
 	.type = IOMMU_DOMAIN_BLOCKED,
 	.ops = &(const struct iommu_domain_ops) {

@@ -833,23 +874,31 @@ static struct iommu_domain blocking_domain = {
 	}
 };
 
-static const struct iommu_ops s390_iommu_ops = {
-	.blocked_domain		= &blocking_domain,
-	.release_domain		= &blocking_domain,
-	.capable = s390_iommu_capable,
-	.domain_alloc_paging = s390_domain_alloc_paging,
-	.probe_device = s390_iommu_probe_device,
-	.device_group = generic_device_group,
-	.pgsize_bitmap = SZ_4K,
-	.get_resv_regions = s390_iommu_get_resv_regions,
-	.default_domain_ops = &(const struct iommu_domain_ops) {
-		.attach_dev	= s390_iommu_attach_device,
-		.map_pages	= s390_iommu_map_pages,
-		.unmap_pages	= s390_iommu_unmap_pages,
-		.flush_iotlb_all = s390_iommu_flush_iotlb_all,
-		.iotlb_sync      = s390_iommu_iotlb_sync,
-		.iotlb_sync_map  = s390_iommu_iotlb_sync_map,
-		.iova_to_phys	= s390_iommu_iova_to_phys,
-		.free		= s390_domain_free,
+#define S390_IOMMU_COMMON_OPS() \
+	.blocked_domain		= &blocking_domain, \
+	.release_domain		= &blocking_domain, \
+	.capable = s390_iommu_capable, \
+	.domain_alloc_paging = s390_domain_alloc_paging, \
+	.probe_device = s390_iommu_probe_device, \
+	.device_group = generic_device_group, \
+	.pgsize_bitmap = SZ_4K, \
+	.get_resv_regions = s390_iommu_get_resv_regions, \
+	.default_domain_ops = &(const struct iommu_domain_ops) { \
+		.attach_dev	= s390_iommu_attach_device, \
+		.map_pages	= s390_iommu_map_pages, \
+		.unmap_pages	= s390_iommu_unmap_pages, \
+		.flush_iotlb_all = s390_iommu_flush_iotlb_all, \
+		.iotlb_sync      = s390_iommu_iotlb_sync, \
+		.iotlb_sync_map  = s390_iommu_iotlb_sync_map, \
+		.iova_to_phys	= s390_iommu_iova_to_phys, \
+		.free		= s390_domain_free, \
 	}
+
+static const struct iommu_ops s390_iommu_ops = {
+	S390_IOMMU_COMMON_OPS()
+};
+
+static const struct iommu_ops s390_iommu_rtr_ops = {
+	.identity_domain	= &s390_identity_domain,
+	S390_IOMMU_COMMON_OPS()
 };