iommu/amd: Don't copy GCR3 table root pointer

When iommu is pre_enabled in kdump kernel, if a device is set up with
guest translations (DTE.GV=1), then don't copy GCR3 table root pointer
but move the device over to an empty guest-cr3 table and handle the
faults in the PPR log (which answer them with INVALID). After all these
PPR faults are recoverable for the device and we should not allow the
device to change old-kernels data when we don't have to.

Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>

drivers/iommu/amd_iommu.c

@@ -102,30 +102,6 @@ int amd_iommu_max_glx_val = -1;
 
 static const struct dma_map_ops amd_iommu_dma_ops;
 
-/*
- * This struct contains device specific data for the IOMMU
- */
-struct iommu_dev_data {
-	struct list_head list;		  /* For domain->dev_list */
-	struct list_head dev_data_list;	  /* For global dev_data_list */
-	struct protection_domain *domain; /* Domain the device is bound to */
-	u16 devid;			  /* PCI Device ID */
-	u16 alias;			  /* Alias Device ID */
-	bool iommu_v2;			  /* Device can make use of IOMMUv2 */
-	bool passthrough;		  /* Device is identity mapped */
-	struct {
-		bool enabled;
-		int qdep;
-	} ats;				  /* ATS state */
-	bool pri_tlp;			  /* PASID TLB required for
-					     PPR completions */
-	u32 errata;			  /* Bitmap for errata to apply */
-	bool use_vapic;			  /* Enable device to use vapic mode */
-	bool defer_attach;
-
-	struct ratelimit_state rs;	  /* Ratelimit IOPF messages */
-};
-
 /*
  * general struct to manage commands send to an IOMMU
  */
@@ -386,10 +362,11 @@ static struct iommu_dev_data *find_dev_data(u16 devid)
 	return dev_data;
 }
 
-static struct iommu_dev_data *get_dev_data(struct device *dev)
+struct iommu_dev_data *get_dev_data(struct device *dev)
 {
 	return dev->archdata.iommu;
 }
+EXPORT_SYMBOL(get_dev_data);
 
 /*
 * Find or create an IOMMU group for a acpihid device.
@@ -2540,6 +2517,7 @@ static int dir2prot(enum dma_data_direction direction)
 	else
 		return 0;
 }
+
 /*
  * This function contains common code for mapping of a physically
  * contiguous memory region into DMA address space. It is used by all

drivers/iommu/amd_iommu_init.c

@@ -214,6 +214,7 @@ u16 *amd_iommu_alias_table;
  * for a specific device. It is also indexed by the PCI device id.
  */
 struct amd_iommu **amd_iommu_rlookup_table;
+EXPORT_SYMBOL(amd_iommu_rlookup_table);
 
 /*
  * This table is used to find the irq remapping table for a given device id
@@ -269,6 +270,7 @@ bool translation_pre_enabled(struct amd_iommu *iommu)
 {
 	return (iommu->flags & AMD_IOMMU_FLAG_TRANS_PRE_ENABLED);
 }
+EXPORT_SYMBOL(translation_pre_enabled);
 
 static void clear_translation_pre_enabled(struct amd_iommu *iommu)
 {
@@ -859,6 +861,7 @@ static bool copy_device_table(void)
 	struct amd_iommu *iommu;
 	u16 dom_id, dte_v, irq_v;
 	gfp_t gfp_flag;
+	u64 tmp;
 
 	if (!amd_iommu_pre_enabled)
 		return false;
@@ -910,6 +913,15 @@ static bool copy_device_table(void)
 			old_dev_tbl_cpy[devid].data[0] = old_devtb[devid].data[0];
 			old_dev_tbl_cpy[devid].data[1] = old_devtb[devid].data[1];
 			__set_bit(dom_id, amd_iommu_pd_alloc_bitmap);
+			/* If gcr3 table existed, mask it out */
+			if (old_devtb[devid].data[0] & DTE_FLAG_GV) {
+				tmp = DTE_GCR3_VAL_B(~0ULL) << DTE_GCR3_SHIFT_B;
+				tmp |= DTE_GCR3_VAL_C(~0ULL) << DTE_GCR3_SHIFT_C;
+				old_dev_tbl_cpy[devid].data[1] &= ~tmp;
+				tmp = DTE_GCR3_VAL_A(~0ULL) << DTE_GCR3_SHIFT_A;
+				tmp |= DTE_FLAG_GV;
+				old_dev_tbl_cpy[devid].data[0] &= ~tmp;
+			}
 		}
 
 		irq_v = old_devtb[devid].data[2] & DTE_IRQ_REMAP_ENABLE;

drivers/iommu/amd_iommu_proto.h

@@ -88,4 +88,5 @@ static inline bool iommu_feature(struct amd_iommu *iommu, u64 f)
 }
 
 extern bool translation_pre_enabled(struct amd_iommu *iommu);
+extern struct iommu_dev_data *get_dev_data(struct device *dev);
 #endif /* _ASM_X86_AMD_IOMMU_PROTO_H  */

drivers/iommu/amd_iommu_types.h

@@ -618,6 +618,30 @@ struct devid_map {
 	bool cmd_line;
 };
 
+/*
+ * This struct contains device specific data for the IOMMU
+ */
+struct iommu_dev_data {
+	struct list_head list;		  /* For domain->dev_list */
+	struct list_head dev_data_list;	  /* For global dev_data_list */
+	struct protection_domain *domain; /* Domain the device is bound to */
+	u16 devid;			  /* PCI Device ID */
+	u16 alias;			  /* Alias Device ID */
+	bool iommu_v2;			  /* Device can make use of IOMMUv2 */
+	bool passthrough;		  /* Device is identity mapped */
+	struct {
+		bool enabled;
+		int qdep;
+	} ats;				  /* ATS state */
+	bool pri_tlp;			  /* PASID TLB required for
+					     PPR completions */
+	u32 errata;			  /* Bitmap for errata to apply */
+	bool use_vapic;			  /* Enable device to use vapic mode */
+	bool defer_attach;
+
+	struct ratelimit_state rs;        /* Ratelimit IOPF messages */
+};
+
 /* Map HPET and IOAPIC ids to the devid used by the IOMMU */
 extern struct list_head ioapic_map;
 extern struct list_head hpet_map;

drivers/iommu/amd_iommu_v2.c

@@ -562,14 +562,30 @@ static int ppr_notifier(struct notifier_block *nb, unsigned long e, void *data)
 	unsigned long flags;
 	struct fault *fault;
 	bool finish;
-	u16 tag;
+	u16 tag, devid;
 	int ret;
+	struct iommu_dev_data *dev_data;
+	struct pci_dev *pdev = NULL;
 
 	iommu_fault = data;
 	tag         = iommu_fault->tag & 0x1ff;
 	finish      = (iommu_fault->tag >> 9) & 1;
 
+	devid = iommu_fault->device_id;
+	pdev = pci_get_bus_and_slot(PCI_BUS_NUM(devid), devid & 0xff);
+	if (!pdev)
+		return -ENODEV;
+	dev_data = get_dev_data(&pdev->dev);
+
+	/* In kdump kernel pci dev is not initialized yet -> send INVALID */
 	ret = NOTIFY_DONE;
+	if (translation_pre_enabled(amd_iommu_rlookup_table[devid])
+		&& dev_data->defer_attach) {
+		amd_iommu_complete_ppr(pdev, iommu_fault->pasid,
+				       PPR_INVALID, tag);
+		goto out;
+	}
+
 	dev_state = get_device_state(iommu_fault->device_id);
 	if (dev_state == NULL)
 		goto out;