5 files changed, 294 insertions, 107 deletions

diff --git a/drivers/vfio/pci/vfio_pci.c b/drivers/vfio/pci/vfio_pci.c
index 09d2e3ffd6f..ac3725440d6 100644
--- a/drivers/vfio/pci/vfio_pci.c
+++ b/drivers/vfio/pci/vfio_pci.c
@@ -201,7 +201,9 @@ static int vfio_pci_get_irq_count(struct vfio_pci_device *vdev, int irq_type)
 
 			return (flags & PCI_MSIX_FLAGS_QSIZE) + 1;
 		}
-	}
+	} else if (irq_type == VFIO_PCI_ERR_IRQ_INDEX)
+		if (pci_is_pcie(vdev->pdev))
+			return 1;
 
 	return 0;
 }
@@ -317,6 +319,17 @@ static long vfio_pci_ioctl(void *device_data,
 		if (info.argsz < minsz || info.index >= VFIO_PCI_NUM_IRQS)
 			return -EINVAL;
 
+		switch (info.index) {
+		case VFIO_PCI_INTX_IRQ_INDEX ... VFIO_PCI_MSIX_IRQ_INDEX:
+			break;
+		case VFIO_PCI_ERR_IRQ_INDEX:
+			if (pci_is_pcie(vdev->pdev))
+				break;
+		/* pass thru to return error */
+		default:
+			return -EINVAL;
+		}
+
 		info.flags = VFIO_IRQ_INFO_EVENTFD;
 
 		info.count = vfio_pci_get_irq_count(vdev, info.index);
@@ -552,11 +565,40 @@ static void vfio_pci_remove(struct pci_dev *pdev)
 	kfree(vdev);
 }
 
+static pci_ers_result_t vfio_pci_aer_err_detected(struct pci_dev *pdev,
+						  pci_channel_state_t state)
+{
+	struct vfio_pci_device *vdev;
+	struct vfio_device *device;
+
+	device = vfio_device_get_from_dev(&pdev->dev);
+	if (device == NULL)
+		return PCI_ERS_RESULT_DISCONNECT;
+
+	vdev = vfio_device_data(device);
+	if (vdev == NULL) {
+		vfio_device_put(device);
+		return PCI_ERS_RESULT_DISCONNECT;
+	}
+
+	if (vdev->err_trigger)
+		eventfd_signal(vdev->err_trigger, 1);
+
+	vfio_device_put(device);
+
+	return PCI_ERS_RESULT_CAN_RECOVER;
+}
+
+static struct pci_error_handlers vfio_err_handlers = {
+	.error_detected = vfio_pci_aer_err_detected,
+};
+
 static struct pci_driver vfio_pci_driver = {
 	.name		= "vfio-pci",
 	.id_table	= NULL, /* only dynamic ids */
 	.probe		= vfio_pci_probe,
 	.remove		= vfio_pci_remove,
+	.err_handler	= &vfio_err_handlers,
 };
 
 static void __exit vfio_pci_cleanup(void)
diff --git a/drivers/vfio/pci/vfio_pci_config.c b/drivers/vfio/pci/vfio_pci_config.c
index aeb00fc2d3b..affa34745be 100644
--- a/drivers/vfio/pci/vfio_pci_config.c
+++ b/drivers/vfio/pci/vfio_pci_config.c
@@ -274,9 +274,10 @@ static int vfio_direct_config_read(struct vfio_pci_device *vdev, int pos,
 	return count;
 }
 
-static int vfio_direct_config_write(struct vfio_pci_device *vdev, int pos,
-				    int count, struct perm_bits *perm,
-				    int offset, __le32 val)
+/* Raw access skips any kind of virtualization */
+static int vfio_raw_config_write(struct vfio_pci_device *vdev, int pos,
+				 int count, struct perm_bits *perm,
+				 int offset, __le32 val)
 {
 	int ret;
 
@@ -287,13 +288,36 @@ static int vfio_direct_config_write(struct vfio_pci_device *vdev, int pos,
 	return count;
 }
 
-/* Default all regions to read-only, no-virtualization */
+static int vfio_raw_config_read(struct vfio_pci_device *vdev, int pos,
+				int count, struct perm_bits *perm,
+				int offset, __le32 *val)
+{
+	int ret;
+
+	ret = vfio_user_config_read(vdev->pdev, pos, val, count);
+	if (ret)
+		return pcibios_err_to_errno(ret);
+
+	return count;
+}
+
+/* Default capability regions to read-only, no-virtualization */
 static struct perm_bits cap_perms[PCI_CAP_ID_MAX + 1] = {
 	[0 ... PCI_CAP_ID_MAX] = { .readfn = vfio_direct_config_read }
 };
 static struct perm_bits ecap_perms[PCI_EXT_CAP_ID_MAX + 1] = {
 	[0 ... PCI_EXT_CAP_ID_MAX] = { .readfn = vfio_direct_config_read }
 };
+/*
+ * Default unassigned regions to raw read-write access.  Some devices
+ * require this to function as they hide registers between the gaps in
+ * config space (be2net).  Like MMIO and I/O port registers, we have
+ * to trust the hardware isolation.
+ */
+static struct perm_bits unassigned_perms = {
+	.readfn = vfio_raw_config_read,
+	.writefn = vfio_raw_config_write
+};
 
 static void free_perm_bits(struct perm_bits *perm)
 {
@@ -779,16 +803,16 @@ int __init vfio_pci_init_perm_bits(void)
 
 	/* Capabilities */
 	ret |= init_pci_cap_pm_perm(&cap_perms[PCI_CAP_ID_PM]);
-	cap_perms[PCI_CAP_ID_VPD].writefn = vfio_direct_config_write;
+	cap_perms[PCI_CAP_ID_VPD].writefn = vfio_raw_config_write;
 	ret |= init_pci_cap_pcix_perm(&cap_perms[PCI_CAP_ID_PCIX]);
-	cap_perms[PCI_CAP_ID_VNDR].writefn = vfio_direct_config_write;
+	cap_perms[PCI_CAP_ID_VNDR].writefn = vfio_raw_config_write;
 	ret |= init_pci_cap_exp_perm(&cap_perms[PCI_CAP_ID_EXP]);
 	ret |= init_pci_cap_af_perm(&cap_perms[PCI_CAP_ID_AF]);
 
 	/* Extended capabilities */
 	ret |= init_pci_ext_cap_err_perm(&ecap_perms[PCI_EXT_CAP_ID_ERR]);
 	ret |= init_pci_ext_cap_pwr_perm(&ecap_perms[PCI_EXT_CAP_ID_PWR]);
-	ecap_perms[PCI_EXT_CAP_ID_VNDR].writefn = vfio_direct_config_write;
+	ecap_perms[PCI_EXT_CAP_ID_VNDR].writefn = vfio_raw_config_write;
 
 	if (ret)
 		vfio_pci_uninit_perm_bits();
@@ -801,9 +825,6 @@ static int vfio_find_cap_start(struct vfio_pci_device *vdev, int pos)
 	u8 cap;
 	int base = (pos >= PCI_CFG_SPACE_SIZE) ? PCI_CFG_SPACE_SIZE :
 						 PCI_STD_HEADER_SIZEOF;
-	base /= 4;
-	pos /= 4;
-
 	cap = vdev->pci_config_map[pos];
 
 	if (cap == PCI_CAP_ID_BASIC)
@@ -813,7 +834,7 @@ static int vfio_find_cap_start(struct vfio_pci_device *vdev, int pos)
 	while (pos - 1 >= base && vdev->pci_config_map[pos - 1] == cap)
 		pos--;
 
-	return pos * 4;
+	return pos;
 }
 
 static int vfio_msi_config_read(struct vfio_pci_device *vdev, int pos,
@@ -1017,13 +1038,9 @@ static int vfio_cap_len(struct vfio_pci_device *vdev, u8 cap, u8 pos)
 		return byte;
 	case PCI_CAP_ID_EXP:
 		/* length based on version */
-		ret = pci_read_config_word(pdev, pos + PCI_EXP_FLAGS, &word);
-		if (ret)
-			return pcibios_err_to_errno(ret);
-
 		vdev->extended_caps = true;
 
-		if ((word & PCI_EXP_FLAGS_VERS) == 1)
+		if ((pcie_caps_reg(pdev) & PCI_EXP_FLAGS_VERS) == 1)
 			return PCI_CAP_EXP_ENDPOINT_SIZEOF_V1;
 		else
 			return PCI_CAP_EXP_ENDPOINT_SIZEOF_V2;
@@ -1230,8 +1247,8 @@ static int vfio_cap_init(struct vfio_pci_device *vdev)
 		}
 
 		/* Sanity check, do we overlap other capabilities? */
-		for (i = 0; i < len; i += 4) {
-			if (likely(map[(pos + i) / 4] == PCI_CAP_ID_INVALID))
+		for (i = 0; i < len; i++) {
+			if (likely(map[pos + i] == PCI_CAP_ID_INVALID))
 				continue;
 
 			pr_warn("%s: %s pci config conflict @0x%x, was cap 0x%x now cap 0x%x\n",
@@ -1239,7 +1256,7 @@ static int vfio_cap_init(struct vfio_pci_device *vdev)
 				pos + i, map[pos + i], cap);
 		}
 
-		memset(map + (pos / 4), cap, len / 4);
+		memset(map + pos, cap, len);
 		ret = vfio_fill_vconfig_bytes(vdev, pos, len);
 		if (ret)
 			return ret;
@@ -1314,8 +1331,8 @@ static int vfio_ecap_init(struct vfio_pci_device *vdev)
 			hidden = true;
 		}
 
-		for (i = 0; i < len; i += 4) {
-			if (likely(map[(epos + i) / 4] == PCI_CAP_ID_INVALID))
+		for (i = 0; i < len; i++) {
+			if (likely(map[epos + i] == PCI_CAP_ID_INVALID))
 				continue;
 
 			pr_warn("%s: %s pci config conflict @0x%x, was ecap 0x%x now ecap 0x%x\n",
@@ -1330,7 +1347,7 @@ static int vfio_ecap_init(struct vfio_pci_device *vdev)
 		 */
 		BUILD_BUG_ON(PCI_EXT_CAP_ID_MAX >= PCI_CAP_ID_INVALID);
 
-		memset(map + (epos / 4), ecap, len / 4);
+		memset(map + epos, ecap, len);
 		ret = vfio_fill_vconfig_bytes(vdev, epos, len);
 		if (ret)
 			return ret;
@@ -1377,10 +1394,12 @@ int vfio_config_init(struct vfio_pci_device *vdev)
 	int ret;
 
 	/*
-	 * Config space, caps and ecaps are all dword aligned, so we can
-	 * use one byte per dword to record the type.
+	 * Config space, caps and ecaps are all dword aligned, so we could
+	 * use one byte per dword to record the type.  However, there are
+	 * no requiremenst on the length of a capability, so the gap between
+	 * capabilities needs byte granularity.
 	 */
-	map = kmalloc(pdev->cfg_size / 4, GFP_KERNEL);
+	map = kmalloc(pdev->cfg_size, GFP_KERNEL);
 	if (!map)
 		return -ENOMEM;
 
@@ -1393,9 +1412,9 @@ int vfio_config_init(struct vfio_pci_device *vdev)
 	vdev->pci_config_map = map;
 	vdev->vconfig = vconfig;
 
-	memset(map, PCI_CAP_ID_BASIC, PCI_STD_HEADER_SIZEOF / 4);
-	memset(map + (PCI_STD_HEADER_SIZEOF / 4), PCI_CAP_ID_INVALID,
-	       (pdev->cfg_size - PCI_STD_HEADER_SIZEOF) / 4);
+	memset(map, PCI_CAP_ID_BASIC, PCI_STD_HEADER_SIZEOF);
+	memset(map + PCI_STD_HEADER_SIZEOF, PCI_CAP_ID_INVALID,
+	       pdev->cfg_size - PCI_STD_HEADER_SIZEOF);
 
 	ret = vfio_fill_vconfig_bytes(vdev, 0, PCI_STD_HEADER_SIZEOF);
 	if (ret)
@@ -1450,6 +1469,22 @@ void vfio_config_free(struct vfio_pci_device *vdev)
 	vdev->msi_perm = NULL;
 }
 
+/*
+ * Find the remaining number of bytes in a dword that match the given
+ * position.  Stop at either the end of the capability or the dword boundary.
+ */
+static size_t vfio_pci_cap_remaining_dword(struct vfio_pci_device *vdev,
+					   loff_t pos)
+{
+	u8 cap = vdev->pci_config_map[pos];
+	size_t i;
+
+	for (i = 1; (pos + i) % 4 && vdev->pci_config_map[pos + i] == cap; i++)
+		/* nop */;
+
+	return i;
+}
+
 static ssize_t vfio_config_do_rw(struct vfio_pci_device *vdev, char __user *buf,
 				 size_t count, loff_t *ppos, bool iswrite)
 {
@@ -1458,55 +1493,48 @@ static ssize_t vfio_config_do_rw(struct vfio_pci_device *vdev, char __user *buf,
 	__le32 val = 0;
 	int cap_start = 0, offset;
 	u8 cap_id;
-	ssize_t ret = count;
+	ssize_t ret;
 
-	if (*ppos < 0 || *ppos + count > pdev->cfg_size)
+	if (*ppos < 0 || *ppos >= pdev->cfg_size ||
+	    *ppos + count > pdev->cfg_size)
 		return -EFAULT;
 
 	/*
-	 * gcc can't seem to figure out we're a static function, only called
-	 * with count of 1/2/4 and hits copy_from_user_overflow without this.
+	 * Chop accesses into aligned chunks containing no more than a
+	 * single capability.  Caller increments to the next chunk.
 	 */
-	if (count > sizeof(val))
-		return -EINVAL;
-
-	cap_id = vdev->pci_config_map[*ppos / 4];
-
-	if (cap_id == PCI_CAP_ID_INVALID) {
-		if (iswrite)
-			return ret; /* drop */
-
-		/*
-		 * Per PCI spec 3.0, section 6.1, reads from reserved and
-		 * unimplemented registers return 0
-		 */
-		if (copy_to_user(buf, &val, count))
-			return -EFAULT;
-
-		return ret;
-	}
+	count = min(count, vfio_pci_cap_remaining_dword(vdev, *ppos));
+	if (count >= 4 && !(*ppos % 4))
+		count = 4;
+	else if (count >= 2 && !(*ppos % 2))
+		count = 2;
+	else
+		count = 1;
 
-	/*
-	 * All capabilities are minimum 4 bytes and aligned on dword
-	 * boundaries.  Since we don't support unaligned accesses, we're
-	 * only ever accessing a single capability.
-	 */
-	if (*ppos >= PCI_CFG_SPACE_SIZE) {
-		WARN_ON(cap_id > PCI_EXT_CAP_ID_MAX);
+	ret = count;
 
-		perm = &ecap_perms[cap_id];
-		cap_start = vfio_find_cap_start(vdev, *ppos);
+	cap_id = vdev->pci_config_map[*ppos];
 
+	if (cap_id == PCI_CAP_ID_INVALID) {
+		perm = &unassigned_perms;
+		cap_start = *ppos;
 	} else {
-		WARN_ON(cap_id > PCI_CAP_ID_MAX);
+		if (*ppos >= PCI_CFG_SPACE_SIZE) {
+			WARN_ON(cap_id > PCI_EXT_CAP_ID_MAX);
 
-		perm = &cap_perms[cap_id];
+			perm = &ecap_perms[cap_id];
+			cap_start = vfio_find_cap_start(vdev, *ppos);
+		} else {
+			WARN_ON(cap_id > PCI_CAP_ID_MAX);
 
-		if (cap_id == PCI_CAP_ID_MSI)
-			perm = vdev->msi_perm;
+			perm = &cap_perms[cap_id];
 
-		if (cap_id > PCI_CAP_ID_BASIC)
-			cap_start = vfio_find_cap_start(vdev, *ppos);
+			if (cap_id == PCI_CAP_ID_MSI)
+				perm = vdev->msi_perm;
+
+			if (cap_id > PCI_CAP_ID_BASIC)
+				cap_start = vfio_find_cap_start(vdev, *ppos);
+		}
 	}
 
 	WARN_ON(!cap_start && cap_id != PCI_CAP_ID_BASIC);
@@ -1546,20 +1574,8 @@ ssize_t vfio_pci_config_rw(struct vfio_pci_device *vdev, char __user *buf,
 
 	pos &= VFIO_PCI_OFFSET_MASK;
 
-	/*
-	 * We want to both keep the access size the caller users as well as
-	 * support reading large chunks of config space in a single call.
-	 * PCI doesn't support unaligned accesses, so we can safely break
-	 * those apart.
-	 */
 	while (count) {
-		if (count >= 4 && !(pos % 4))
-			ret = vfio_config_do_rw(vdev, buf, 4, &pos, iswrite);
-		else if (count >= 2 && !(pos % 2))
-			ret = vfio_config_do_rw(vdev, buf, 2, &pos, iswrite);
-		else
-			ret = vfio_config_do_rw(vdev, buf, 1, &pos, iswrite);
-
+		ret = vfio_config_do_rw(vdev, buf, count, &pos, iswrite);
 		if (ret < 0)
 			return ret;
 
diff --git a/drivers/vfio/pci/vfio_pci_intrs.c b/drivers/vfio/pci/vfio_pci_intrs.c
index a96509187de..4bc704e1b7c 100644
--- a/drivers/vfio/pci/vfio_pci_intrs.c
+++ b/drivers/vfio/pci/vfio_pci_intrs.c
@@ -287,7 +287,8 @@ void vfio_pci_intx_mask(struct vfio_pci_device *vdev)
  * a signal is necessary, which can then be handled via a work queue
  * or directly depending on the caller.
  */
-int vfio_pci_intx_unmask_handler(struct vfio_pci_device *vdev, void *unused)
+static int vfio_pci_intx_unmask_handler(struct vfio_pci_device *vdev,
+					void *unused)
 {
 	struct pci_dev *pdev = vdev->pdev;
 	unsigned long flags;
@@ -746,6 +747,63 @@ static int vfio_pci_set_msi_trigger(struct vfio_pci_device *vdev,
 	return 0;
 }
 
+static int vfio_pci_set_err_trigger(struct vfio_pci_device *vdev,
+				    unsigned index, unsigned start,
+				    unsigned count, uint32_t flags, void *data)
+{
+	int32_t fd = *(int32_t *)data;
+	struct pci_dev *pdev = vdev->pdev;
+
+	if ((index != VFIO_PCI_ERR_IRQ_INDEX) ||
+	    !(flags & VFIO_IRQ_SET_DATA_TYPE_MASK))
+		return -EINVAL;
+
+	/*
+	 * device_lock synchronizes setting and checking of
+	 * err_trigger. The vfio_pci_aer_err_detected() is also
+	 * called with device_lock held.
+	 */
+
+	/* DATA_NONE/DATA_BOOL enables loopback testing */
+
+	if (flags & VFIO_IRQ_SET_DATA_NONE) {
+		device_lock(&pdev->dev);
+		if (vdev->err_trigger)
+			eventfd_signal(vdev->err_trigger, 1);
+		device_unlock(&pdev->dev);
+		return 0;
+	} else if (flags & VFIO_IRQ_SET_DATA_BOOL) {
+		uint8_t trigger = *(uint8_t *)data;
+		device_lock(&pdev->dev);
+		if (trigger && vdev->err_trigger)
+			eventfd_signal(vdev->err_trigger, 1);
+		device_unlock(&pdev->dev);
+		return 0;
+	}
+
+	/* Handle SET_DATA_EVENTFD */
+
+	if (fd == -1) {
+		device_lock(&pdev->dev);
+		if (vdev->err_trigger)
+			eventfd_ctx_put(vdev->err_trigger);
+		vdev->err_trigger = NULL;
+		device_unlock(&pdev->dev);
+		return 0;
+	} else if (fd >= 0) {
+		struct eventfd_ctx *efdctx;
+		efdctx = eventfd_ctx_fdget(fd);
+		if (IS_ERR(efdctx))
+			return PTR_ERR(efdctx);
+		device_lock(&pdev->dev);
+		if (vdev->err_trigger)
+			eventfd_ctx_put(vdev->err_trigger);
+		vdev->err_trigger = efdctx;
+		device_unlock(&pdev->dev);
+		return 0;
+	} else
+		return -EINVAL;
+}
 int vfio_pci_set_irqs_ioctl(struct vfio_pci_device *vdev, uint32_t flags,
 			    unsigned index, unsigned start, unsigned count,
 			    void *data)
@@ -780,6 +838,13 @@ int vfio_pci_set_irqs_ioctl(struct vfio_pci_device *vdev, uint32_t flags,
 			break;
 		}
 		break;
+	case VFIO_PCI_ERR_IRQ_INDEX:
+		switch (flags & VFIO_IRQ_SET_ACTION_TYPE_MASK) {
+		case VFIO_IRQ_SET_ACTION_TRIGGER:
+			if (pci_is_pcie(vdev->pdev))
+				func = vfio_pci_set_err_trigger;
+			break;
+		}
 	}
 
 	if (!func)
diff --git a/drivers/vfio/pci/vfio_pci_private.h b/drivers/vfio/pci/vfio_pci_private.h
index d7e55d03f49..9c6d5d0f3b0 100644
--- a/drivers/vfio/pci/vfio_pci_private.h
+++ b/drivers/vfio/pci/vfio_pci_private.h
@@ -56,6 +56,7 @@ struct vfio_pci_device {
 	bool			has_vga;
 	struct pci_saved_state	*pci_saved_state;
 	atomic_t		refcnt;
+	struct eventfd_ctx	*err_trigger;
 };
 
 #define is_intx(vdev) (vdev->irq_type == VFIO_PCI_INTX_IRQ_INDEX)
diff --git a/drivers/vfio/vfio.c b/drivers/vfio/vfio.c
index fcc12f3e60a..acb7121a931 100644
--- a/drivers/vfio/vfio.c
+++ b/drivers/vfio/vfio.c
@@ -24,8 +24,10 @@
 #include <linux/list.h>
 #include <linux/module.h>
 #include <linux/mutex.h>
+#include <linux/rwsem.h>
 #include <linux/sched.h>
 #include <linux/slab.h>
+#include <linux/stat.h>
 #include <linux/string.h>
 #include <linux/uaccess.h>
 #include <linux/vfio.h>
@@ -57,7 +59,7 @@ struct vfio_iommu_driver {
 struct vfio_container {
 	struct kref			kref;
 	struct list_head		group_list;
-	struct mutex			group_lock;
+	struct rw_semaphore		group_lock;
 	struct vfio_iommu_driver	*iommu_driver;
 	void				*iommu_data;
 };
@@ -392,12 +394,13 @@ static void vfio_device_release(struct kref *kref)
 }
 
 /* Device reference always implies a group reference */
-static void vfio_device_put(struct vfio_device *device)
+void vfio_device_put(struct vfio_device *device)
 {
 	struct vfio_group *group = device->group;
 	kref_put_mutex(&device->kref, vfio_device_release, &group->device_lock);
 	vfio_group_put(group);
 }
+EXPORT_SYMBOL_GPL(vfio_device_put);
 
 static void vfio_device_get(struct vfio_device *device)
 {
@@ -627,6 +630,33 @@ int vfio_add_group_dev(struct device *dev,
 }
 EXPORT_SYMBOL_GPL(vfio_add_group_dev);
 
+/**
+ * Get a reference to the vfio_device for a device that is known to
+ * be bound to a vfio driver.  The driver implicitly holds a
+ * vfio_device reference between vfio_add_group_dev and
+ * vfio_del_group_dev.  We can therefore use drvdata to increment
+ * that reference from the struct device.  This additional
+ * reference must be released by calling vfio_device_put.
+ */
+struct vfio_device *vfio_device_get_from_dev(struct device *dev)
+{
+	struct vfio_device *device = dev_get_drvdata(dev);
+
+	vfio_device_get(device);
+
+	return device;
+}
+EXPORT_SYMBOL_GPL(vfio_device_get_from_dev);
+
+/*
+ * Caller must hold a reference to the vfio_device
+ */
+void *vfio_device_data(struct vfio_device *device)
+{
+	return device->device_data;
+}
+EXPORT_SYMBOL_GPL(vfio_device_data);
+
 /* Given a referenced group, check if it contains the device */
 static bool vfio_dev_present(struct vfio_group *group, struct device *dev)
 {
@@ -675,9 +705,13 @@ EXPORT_SYMBOL_GPL(vfio_del_group_dev);
 static long vfio_ioctl_check_extension(struct vfio_container *container,
 				       unsigned long arg)
 {
-	struct vfio_iommu_driver *driver = container->iommu_driver;
+	struct vfio_iommu_driver *driver;
 	long ret = 0;
 
+	down_read(&container->group_lock);
+
+	driver = container->iommu_driver;
+
 	switch (arg) {
 		/* No base extensions yet */
 	default:
@@ -707,10 +741,12 @@ static long vfio_ioctl_check_extension(struct vfio_container *container,
 						 VFIO_CHECK_EXTENSION, arg);
 	}
 
+	up_read(&container->group_lock);
+
 	return ret;
 }
 
-/* hold container->group_lock */
+/* hold write lock on container->group_lock */
 static int __vfio_container_attach_groups(struct vfio_container *container,
 					  struct vfio_iommu_driver *driver,
 					  void *data)
@@ -741,7 +777,7 @@ static long vfio_ioctl_set_iommu(struct vfio_container *container,
 	struct vfio_iommu_driver *driver;
 	long ret = -ENODEV;
 
-	mutex_lock(&container->group_lock);
+	down_write(&container->group_lock);
 
 	/*
 	 * The container is designed to be an unprivileged interface while
@@ -752,7 +788,7 @@ static long vfio_ioctl_set_iommu(struct vfio_container *container,
 	 * the container is deprivileged and returns to an unset state.
 	 */
 	if (list_empty(&container->group_list) || container->iommu_driver) {
-		mutex_unlock(&container->group_lock);
+		up_write(&container->group_lock);
 		return -EINVAL;
 	}
 
@@ -799,7 +835,7 @@ static long vfio_ioctl_set_iommu(struct vfio_container *container,
 
 	mutex_unlock(&vfio.iommu_drivers_lock);
 skip_drivers_unlock:
-	mutex_unlock(&container->group_lock);
+	up_write(&container->group_lock);
 
 	return ret;
 }
@@ -815,9 +851,6 @@ static long vfio_fops_unl_ioctl(struct file *filep,
 	if (!container)
 		return ret;
 
-	driver = container->iommu_driver;
-	data = container->iommu_data;
-
 	switch (cmd) {
 	case VFIO_GET_API_VERSION:
 		ret = VFIO_API_VERSION;
@@ -829,8 +862,15 @@ static long vfio_fops_unl_ioctl(struct file *filep,
 		ret = vfio_ioctl_set_iommu(container, arg);
 		break;
 	default:
+		down_read(&container->group_lock);
+
+		driver = container->iommu_driver;
+		data = container->iommu_data;
+
 		if (driver) /* passthrough all unrecognized ioctls */
 			ret = driver->ops->ioctl(data, cmd, arg);
+
+		up_read(&container->group_lock);
 	}
 
 	return ret;
@@ -854,7 +894,7 @@ static int vfio_fops_open(struct inode *inode, struct file *filep)
 		return -ENOMEM;
 
 	INIT_LIST_HEAD(&container->group_list);
-	mutex_init(&container->group_lock);
+	init_rwsem(&container->group_lock);
 	kref_init(&container->kref);
 
 	filep->private_data = container;
@@ -881,35 +921,55 @@ static ssize_t vfio_fops_read(struct file *filep, char __user *buf,
 			      size_t count, loff_t *ppos)
 {
 	struct vfio_container *container = filep->private_data;
-	struct vfio_iommu_driver *driver = container->iommu_driver;
+	struct vfio_iommu_driver *driver;
+	ssize_t ret = -EINVAL;
 
-	if (unlikely(!driver || !driver->ops->read))
-		return -EINVAL;
+	down_read(&container->group_lock);
 
-	return driver->ops->read(container->iommu_data, buf, count, ppos);
+	driver = container->iommu_driver;
+	if (likely(driver && driver->ops->read))
+		ret = driver->ops->read(container->iommu_data,
+					buf, count, ppos);
+
+	up_read(&container->group_lock);
+
+	return ret;
 }
 
 static ssize_t vfio_fops_write(struct file *filep, const char __user *buf,
 			       size_t count, loff_t *ppos)
 {
 	struct vfio_container *container = filep->private_data;
-	struct vfio_iommu_driver *driver = container->iommu_driver;
+	struct vfio_iommu_driver *driver;
+	ssize_t ret = -EINVAL;
 
-	if (unlikely(!driver || !driver->ops->write))
-		return -EINVAL;
+	down_read(&container->group_lock);
 
-	return driver->ops->write(container->iommu_data, buf, count, ppos);
+	driver = container->iommu_driver;
+	if (likely(driver && driver->ops->write))
+		ret = driver->ops->write(container->iommu_data,
+					 buf, count, ppos);
+
+	up_read(&container->group_lock);
+
+	return ret;
 }
 
 static int vfio_fops_mmap(struct file *filep, struct vm_area_struct *vma)
 {
 	struct vfio_container *container = filep->private_data;
-	struct vfio_iommu_driver *driver = container->iommu_driver;
+	struct vfio_iommu_driver *driver;
+	int ret = -EINVAL;
 
-	if (unlikely(!driver || !driver->ops->mmap))
-		return -EINVAL;
+	down_read(&container->group_lock);
 
-	return driver->ops->mmap(container->iommu_data, vma);
+	driver = container->iommu_driver;
+	if (likely(driver && driver->ops->mmap))
+		ret = driver->ops->mmap(container->iommu_data, vma);
+
+	up_read(&container->group_lock);
+
+	return ret;
 }
 
 static const struct file_operations vfio_fops = {
@@ -933,7 +993,7 @@ static void __vfio_group_unset_container(struct vfio_group *group)
 	struct vfio_container *container = group->container;
 	struct vfio_iommu_driver *driver;
 
-	mutex_lock(&container->group_lock);
+	down_write(&container->group_lock);
 
 	driver = container->iommu_driver;
 	if (driver)
@@ -951,7 +1011,7 @@ static void __vfio_group_unset_container(struct vfio_group *group)
 		container->iommu_data = NULL;
 	}
 
-	mutex_unlock(&container->group_lock);
+	up_write(&container->group_lock);
 
 	vfio_container_put(container);
 }
@@ -1011,7 +1071,7 @@ static int vfio_group_set_container(struct vfio_group *group, int container_fd)
 	container = f.file->private_data;
 	WARN_ON(!container); /* fget ensures we don't race vfio_release */
 
-	mutex_lock(&container->group_lock);
+	down_write(&container->group_lock);
 
 	driver = container->iommu_driver;
 	if (driver) {
@@ -1029,7 +1089,7 @@ static int vfio_group_set_container(struct vfio_group *group, int container_fd)
 	atomic_inc(&group->container_users);
 
 unlock_out:
-	mutex_unlock(&container->group_lock);
+	up_write(&container->group_lock);
 	fdput(f);
 	return ret;
 }
@@ -1300,6 +1360,9 @@ static const struct file_operations vfio_device_fops = {
  */
 static char *vfio_devnode(struct device *dev, umode_t *mode)
 {
+	if (MINOR(dev->devt) == 0)
+		*mode = S_IRUGO | S_IWUGO;
+
 	return kasprintf(GFP_KERNEL, "vfio/%s", dev_name(dev));
 }