Message-ID: <20180207000731.32764.95992.stgit@gimli.home>
Date:   Tue, 06 Feb 2018 17:08:14 -0700
From:   Alex Williamson <alex.williamson@...hat.com>
To:     kvm@...r.kernel.org
Cc:     linux-kernel@...r.kernel.org, qemu-devel@...gnu.org,
        alex.williamson@...hat.com
Subject: [RFC PATCH] vfio/pci: Add ioeventfd support

The ioeventfd here is actually irqfd handling of an ioeventfd such as
supported in KVM.  A user is able to pre-program a device write to
occur when the eventfd triggers.  This is yet another instance of
eventfd-irqfd triggering between KVM and vfio.  The impetus for this
is high frequency writes to pages which are virtualized in QEMU.
Enabling this near-direct write path for selected registers within
the virtualized page can improve performance and reduce overhead.
Specifically this is initially targeted at NVIDIA graphics cards where
the driver issues a write to an MMIO register within a virtualized
region in order to allow the MSI interrupt to re-trigger.

Signed-off-by: Alex Williamson <alex.williamson@...hat.com>
---
 drivers/vfio/pci/vfio_pci.c         |   33 +++++++
 drivers/vfio/pci/vfio_pci_private.h |   14 +++
 drivers/vfio/pci/vfio_pci_rdwr.c    |  165 ++++++++++++++++++++++++++++++++---
 include/uapi/linux/vfio.h           |   24 +++++
 4 files changed, 224 insertions(+), 12 deletions(-)

diff --git a/drivers/vfio/pci/vfio_pci.c b/drivers/vfio/pci/vfio_pci.c
index f041b1a6cf66..c8e7297a61a3 100644
--- a/drivers/vfio/pci/vfio_pci.c
+++ b/drivers/vfio/pci/vfio_pci.c
@@ -302,6 +302,7 @@ static void vfio_pci_disable(struct vfio_pci_device *vdev)
 {
 	struct pci_dev *pdev = vdev->pdev;
 	struct vfio_pci_dummy_resource *dummy_res, *tmp;
+	struct vfio_pci_ioeventfd *ioeventfd, *ioeventfd_tmp;
 	int i, bar;
 
 	/* Stop the device from further DMA */
@@ -311,6 +312,14 @@ static void vfio_pci_disable(struct vfio_pci_device *vdev)
 				VFIO_IRQ_SET_ACTION_TRIGGER,
 				vdev->irq_type, 0, 0, NULL);
 
+	/* Device closed, don't need mutex here */
+	list_for_each_entry_safe(ioeventfd, ioeventfd_tmp,
+				 &vdev->ioeventfds_list, next) {
+		vfio_virqfd_disable(&ioeventfd->virqfd);
+		list_del(&ioeventfd->next);
+		kfree(ioeventfd);
+	}
+
 	vdev->virq_disabled = false;
 
 	for (i = 0; i < vdev->num_regions; i++)
@@ -1039,6 +1048,28 @@ static long vfio_pci_ioctl(void *device_data,
 
 		kfree(groups);
 		return ret;
+	} else if (cmd == VFIO_DEVICE_IOEVENTFD) {
+		struct vfio_device_ioeventfd ioeventfd;
+		int count;
+
+		minsz = offsetofend(struct vfio_device_ioeventfd, fd);
+
+		if (copy_from_user(&ioeventfd, (void __user*)arg, minsz))
+			return -EFAULT;
+
+		if (ioeventfd.argsz < minsz)
+			return -EINVAL;
+
+		if (ioeventfd.flags & ~VFIO_DEVICE_IOEVENTFD_SIZE_MASK)
+			return -EINVAL;
+
+		count = ioeventfd.flags & VFIO_DEVICE_IOEVENTFD_SIZE_MASK;
+
+		if (hweight8(count) != 1 || ioeventfd.fd < -1)
+			return -EINVAL;
+
+		return vfio_pci_ioeventfd(vdev, ioeventfd.offset,
+					  ioeventfd.data, count, ioeventfd.fd);
 	}
 
 	return -ENOTTY;
@@ -1217,6 +1248,8 @@ static int vfio_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	vdev->irq_type = VFIO_PCI_NUM_IRQS;
 	mutex_init(&vdev->igate);
 	spin_lock_init(&vdev->irqlock);
+	mutex_init(&vdev->ioeventfds_lock);
+	INIT_LIST_HEAD(&vdev->ioeventfds_list);
 
 	ret = vfio_add_group_dev(&pdev->dev, &vfio_pci_ops, vdev);
 	if (ret) {
diff --git a/drivers/vfio/pci/vfio_pci_private.h b/drivers/vfio/pci/vfio_pci_private.h
index f561ac1c78a0..23797622396e 100644
--- a/drivers/vfio/pci/vfio_pci_private.h
+++ b/drivers/vfio/pci/vfio_pci_private.h
@@ -29,6 +29,15 @@
 #define PCI_CAP_ID_INVALID		0xFF	/* default raw access */
 #define PCI_CAP_ID_INVALID_VIRT		0xFE	/* default virt access */
 
+struct vfio_pci_ioeventfd {
+	struct list_head	next;
+	struct virqfd		*virqfd;
+	loff_t			pos;
+	uint64_t		data;
+	int			bar;
+	int			count;
+};
+
 struct vfio_pci_irq_ctx {
 	struct eventfd_ctx	*trigger;
 	struct virqfd		*unmask;
@@ -95,6 +104,8 @@ struct vfio_pci_device {
 	struct eventfd_ctx	*err_trigger;
 	struct eventfd_ctx	*req_trigger;
 	struct list_head	dummy_resources_list;
+	struct mutex		ioeventfds_lock;
+	struct list_head	ioeventfds_list;
 };
 
 #define is_intx(vdev) (vdev->irq_type == VFIO_PCI_INTX_IRQ_INDEX)
@@ -120,6 +131,9 @@ extern ssize_t vfio_pci_bar_rw(struct vfio_pci_device *vdev, char __user *buf,
 extern ssize_t vfio_pci_vga_rw(struct vfio_pci_device *vdev, char __user *buf,
 			       size_t count, loff_t *ppos, bool iswrite);
 
+extern long vfio_pci_ioeventfd(struct vfio_pci_device *vdev, loff_t offset,
+			       uint64_t data, int count, int fd);
+
 extern int vfio_pci_init_perm_bits(void);
 extern void vfio_pci_uninit_perm_bits(void);
 
diff --git a/drivers/vfio/pci/vfio_pci_rdwr.c b/drivers/vfio/pci/vfio_pci_rdwr.c
index 357243d76f10..55bb4517d4ba 100644
--- a/drivers/vfio/pci/vfio_pci_rdwr.c
+++ b/drivers/vfio/pci/vfio_pci_rdwr.c
@@ -17,6 +17,7 @@
 #include <linux/pci.h>
 #include <linux/uaccess.h>
 #include <linux/io.h>
+#include <linux/vfio.h>
 #include <linux/vgaarb.h>
 
 #include "vfio_pci_private.h"
@@ -113,6 +114,30 @@ static ssize_t do_io_rw(void __iomem *io, char __user *buf,
 	return done;
 }
 
+static int vfio_pci_setup_barmap(struct vfio_pci_device *vdev, int bar)
+{
+	struct pci_dev *pdev = vdev->pdev;
+	int ret;
+	void __iomem *io;
+
+	if (vdev->barmap[bar])
+		return 0;
+
+	ret = pci_request_selected_regions(pdev, 1 << bar, "vfio");
+	if (ret)
+		return ret;
+
+	io = pci_iomap(pdev, bar, 0);
+	if (!io) {
+		pci_release_selected_regions(pdev, 1 << bar);
+		return -ENOMEM;
+	}
+
+	vdev->barmap[bar] = io;
+
+	return 0;
+}
+
 ssize_t vfio_pci_bar_rw(struct vfio_pci_device *vdev, char __user *buf,
 			size_t count, loff_t *ppos, bool iswrite)
 {
@@ -147,22 +172,13 @@ ssize_t vfio_pci_bar_rw(struct vfio_pci_device *vdev, char __user *buf,
 		if (!io)
 			return -ENOMEM;
 		x_end = end;
-	} else if (!vdev->barmap[bar]) {
-		int ret;
-
-		ret = pci_request_selected_regions(pdev, 1 << bar, "vfio");
+	} else {
+		int ret = vfio_pci_setup_barmap(vdev, bar);
 		if (ret)
 			return ret;
 
-		io = pci_iomap(pdev, bar, 0);
-		if (!io) {
-			pci_release_selected_regions(pdev, 1 << bar);
-			return -ENOMEM;
-		}
-
-		vdev->barmap[bar] = io;
-	} else
 		io = vdev->barmap[bar];
+	}
 
 	if (bar == vdev->msix_bar) {
 		x_start = vdev->msix_offset;
@@ -242,3 +258,128 @@ ssize_t vfio_pci_vga_rw(struct vfio_pci_device *vdev, char __user *buf,
 
 	return done;
 }
+
+static int vfio_pci_ioeventfd_handler8(void *opaque, void *data)
+{
+	iowrite8((unsigned long)data, opaque);
+	return 0;
+}
+
+static int vfio_pci_ioeventfd_handler16(void *opaque, void *data)
+{
+	iowrite16((unsigned long)data, opaque);
+	return 0;
+}
+
+static int vfio_pci_ioeventfd_handler32(void *opaque, void *data)
+{
+	iowrite32((unsigned long)data, opaque);
+	return 0;
+}
+
+#ifdef iowrite64
+static int vfio_pci_ioeventfd_handler64(void *opaque, void *data)
+{
+	iowrite64((unsigned long)data, opaque);
+	return 0;
+}
+#endif
+
+long vfio_pci_ioeventfd(struct vfio_pci_device *vdev, loff_t offset,
+			uint64_t data, int count, int fd)
+{
+	struct pci_dev *pdev = vdev->pdev;
+	loff_t pos = offset & VFIO_PCI_OFFSET_MASK;
+	int ret, bar = VFIO_PCI_OFFSET_TO_INDEX(offset);
+	struct vfio_pci_ioeventfd *ioeventfd;
+	int (*handler)(void *, void *);
+	unsigned long val;
+
+	/* Only support ioeventfds into BARs */
+	if (bar > VFIO_PCI_BAR5_REGION_INDEX)
+		return -EINVAL;
+
+	if (pos + count > pci_resource_len(pdev, bar))
+		return -EINVAL;
+
+	/* Disallow ioeventfds working around MSI-X table writes */
+	if (bar == vdev->msix_bar &&
+	    !(pos + count <= vdev->msix_offset ||
+	      pos >= vdev->msix_offset + vdev->msix_size))
+		return -EINVAL;
+
+	switch (count) {
+	case 1:
+		handler = &vfio_pci_ioeventfd_handler8;
+		val = data;
+		break;
+	case 2:
+		handler = &vfio_pci_ioeventfd_handler16;
+		val = le16_to_cpu(data);
+		break;
+	case 4:
+		handler = &vfio_pci_ioeventfd_handler32;
+		val = le32_to_cpu(data);
+		break;
+#ifdef iowrite64
+	case 8:
+		handler = &vfio_pci_ioeventfd_handler64;
+		val = le64_to_cpu(data);
+		break;
+#endif
+	default:
+		return -EINVAL;
+	}
+
+	ret = vfio_pci_setup_barmap(vdev, bar);
+	if (ret)
+		return ret;
+
+	mutex_lock(&vdev->ioeventfds_lock);
+
+	list_for_each_entry(ioeventfd, &vdev->ioeventfds_list, next) {
+		if (ioeventfd->pos == pos && ioeventfd->bar == bar &&
+		    ioeventfd->data == data && ioeventfd->count == count) {
+			if (fd == -1) {
+				vfio_virqfd_disable(&ioeventfd->virqfd);
+				list_del(&ioeventfd->next);
+				kfree(ioeventfd);
+				ret = 0;
+			} else
+				ret = -EEXIST;
+
+			goto out_unlock;
+		}
+	}
+
+	if (fd < 0) {
+		ret = -ENODEV;
+		goto out_unlock;
+	}
+
+	ioeventfd = kzalloc(sizeof(*ioeventfd), GFP_KERNEL);
+	if (!ioeventfd) {
+		ret = -ENOMEM;
+		goto out_unlock;
+	}
+
+	ioeventfd->pos = pos;
+	ioeventfd->bar = bar;
+	ioeventfd->data = data;
+	ioeventfd->count = count;
+
+	ret = vfio_virqfd_enable(vdev->barmap[ioeventfd->bar] + ioeventfd->pos,
+				 handler, NULL, (void *)val,
+				 &ioeventfd->virqfd, fd);
+	if (ret) {
+		kfree(ioeventfd);
+		goto out_unlock;
+	}
+
+	list_add(&ioeventfd->next, &vdev->ioeventfds_list);
+
+out_unlock:
+	mutex_unlock(&vdev->ioeventfds_lock);
+
+	return ret;
+}
diff --git a/include/uapi/linux/vfio.h b/include/uapi/linux/vfio.h
index e3301dbd27d4..07966a5f0832 100644
--- a/include/uapi/linux/vfio.h
+++ b/include/uapi/linux/vfio.h
@@ -503,6 +503,30 @@ struct vfio_pci_hot_reset {
 
 #define VFIO_DEVICE_PCI_HOT_RESET	_IO(VFIO_TYPE, VFIO_BASE + 13)
 
+/**
+ * VFIO_DEVICE_IOEVENTFD - _IOW(VFIO_TYPE, VFIO_BASE + 14,
+ *                              struct vfio_device_ioeventfd)
+ *
+ * Perform a write to the device at the specified device fd offset, with
+ * the specified data and width when the provided eventfd is triggered.
+ *
+ * Return: 0 on success, -errno on failure.
+ */
+struct vfio_device_ioeventfd {
+	__u32	argsz;
+	__u32	flags;
+#define VFIO_DEVICE_IOEVENTFD_8		(1 << 0) /* 1-byte write */
+#define VFIO_DEVICE_IOEVENTFD_16	(1 << 1) /* 2-byte write */
+#define VFIO_DEVICE_IOEVENTFD_32	(1 << 2) /* 4-byte write */
+#define VFIO_DEVICE_IOEVENTFD_64	(1 << 3) /* 8-byte write */
+#define VFIO_DEVICE_IOEVENTFD_SIZE_MASK	(0xf)
+	__u64	offset;			/* device fd offset of write */
+	__u64	data;			/* data to be written */
+	__s32	fd;			/* -1 for de-assignment */
+};
+
+#define VFIO_DEVICE_IOEVENTFD		_IO(VFIO_TYPE, VFIO_BASE + 14)
+
 /* -------- API for Type1 VFIO IOMMU -------- */
 
 /**

Powered by blists - more mailing lists