Intel MIC Host Driver Changes for Virtio Devices.

This patch introduces the host "Virtio over PCIe" interface for
Intel MIC. It allows creating user space backends on the host and instantiating
virtio devices for them on the Intel MIC card. It uses the existing VRINGH
infrastructure in the kernel to access virtio rings from the host. A character
device per MIC is exposed with IOCTL, mmap and poll callbacks. This allows the
user space backend to:
(a) add/remove a virtio device via a device page.
(b) map (R/O) virtio rings and device page to user space.
(c) poll for availability of data.
(d) copy a descriptor or entire descriptor chain to/from the card.
(e) modify virtio configuration.
(f) handle virtio device reset.
The buffers are copied over using CPU copies for this initial patch
and host initiated MIC DMA support is planned for future patches.
The avail and desc virtio rings are in host memory and the used ring
is in card memory to maximize writes across PCIe for performance.

Co-author: Sudeep Dutt <sudeep.dutt@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Caz Yokoyama <Caz.Yokoyama@intel.com>
Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com>
Signed-off-by: Nikhil Rao <nikhil.rao@intel.com>
Signed-off-by: Harshavardhan R Kharche <harshavardhan.r.kharche@intel.com>
Signed-off-by: Sudeep Dutt <sudeep.dutt@intel.com>
Acked-by: Yaozu (Eddie) Dong <eddie.dong@intel.com>
Reviewed-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

3 files changed, 240 insertions, 1 deletions

diff --git a/include/uapi/linux/Kbuild b/include/uapi/linux/Kbuild
index d37263fbdca7..33d2b8fe166d 100644
--- a/include/uapi/linux/Kbuild
+++ b/include/uapi/linux/Kbuild
@@ -242,6 +242,7 @@ header-y += mei.h
 header-y += mempolicy.h
 header-y += meye.h
 header-y += mic_common.h
+header-y += mic_ioctl.h
 header-y += mii.h
 header-y += minix_fs.h
 header-y += mman.h
diff --git a/include/uapi/linux/mic_common.h b/include/uapi/linux/mic_common.h
index a9091e529183..364ac751bd04 100644
--- a/include/uapi/linux/mic_common.h
+++ b/include/uapi/linux/mic_common.h
@@ -21,7 +21,60 @@
 #ifndef __MIC_COMMON_H_
 #define __MIC_COMMON_H_
 
-#include <linux/types.h>
+#include <linux/virtio_ring.h>
+
+#ifndef __KERNEL__
+#define ALIGN(a, x)	(((a) + (x) - 1) & ~((x) - 1))
+#define __aligned(x)	__attribute__ ((aligned(x)))
+#endif
+
+#define mic_aligned_size(x) ALIGN(sizeof(x), 8)
+
+/**
+ * struct mic_device_desc: Virtio device information shared between the
+ * virtio driver and userspace backend
+ *
+ * @type: Device type: console/network/disk etc.  Type 0/-1 terminates.
+ * @num_vq: Number of virtqueues.
+ * @feature_len: Number of bytes of feature bits.  Multiply by 2: one for
+   host features and one for guest acknowledgements.
+ * @config_len: Number of bytes of the config array after virtqueues.
+ * @status: A status byte, written by the Guest.
+ * @config: Start of the following variable length config.
+ */
+struct mic_device_desc {
+	__s8 type;
+	__u8 num_vq;
+	__u8 feature_len;
+	__u8 config_len;
+	__u8 status;
+	__u64 config[0];
+} __aligned(8);
+
+/**
+ * struct mic_device_ctrl: Per virtio device information in the device page
+ * used internally by the host and card side drivers.
+ *
+ * @vdev: Used for storing MIC vdev information by the guest.
+ * @config_change: Set to 1 by host when a config change is requested.
+ * @vdev_reset: Set to 1 by guest to indicate virtio device has been reset.
+ * @guest_ack: Set to 1 by guest to ack a command.
+ * @host_ack: Set to 1 by host to ack a command.
+ * @used_address_updated: Set to 1 by guest when the used address should be
+ * updated.
+ * @c2h_vdev_db: The doorbell number to be used by guest. Set by host.
+ * @h2c_vdev_db: The doorbell number to be used by host. Set by guest.
+ */
+struct mic_device_ctrl {
+	__u64 vdev;
+	__u8 config_change;
+	__u8 vdev_reset;
+	__u8 guest_ack;
+	__u8 host_ack;
+	__u8 used_address_updated;
+	__s8 c2h_vdev_db;
+	__s8 h2c_vdev_db;
+} __aligned(8);
 
 /**
  * struct mic_bootparam: Virtio device independent information in device page
@@ -42,6 +95,117 @@ struct mic_bootparam {
 	__u8 shutdown_card;
 } __aligned(8);
 
+/**
+ * struct mic_device_page: High level representation of the device page
+ *
+ * @bootparam: The bootparam structure is used for sharing information and
+ * status updates between MIC host and card drivers.
+ * @desc: Array of MIC virtio device descriptors.
+ */
+struct mic_device_page {
+	struct mic_bootparam bootparam;
+	struct mic_device_desc desc[0];
+};
+/**
+ * struct mic_vqconfig: This is how we expect the device configuration field
+ * for a virtqueue to be laid out in config space.
+ *
+ * @address: Guest/MIC physical address of the virtio ring
+ * (avail and desc rings)
+ * @used_address: Guest/MIC physical address of the used ring
+ * @num: The number of entries in the virtio_ring
+ */
+struct mic_vqconfig {
+	__u64 address;
+	__u64 used_address;
+	__u16 num;
+} __aligned(8);
+
+/*
+ * The alignment to use between consumer and producer parts of vring.
+ * This is pagesize for historical reasons.
+ */
+#define MIC_VIRTIO_RING_ALIGN		4096
+
+#define MIC_MAX_VRINGS			4
+#define MIC_VRING_ENTRIES		128
+
+/*
+ * Max vring entries (power of 2) to ensure desc and avail rings
+ * fit in a single page
+ */
+#define MIC_MAX_VRING_ENTRIES		128
+
+/**
+ * Max size of the desc block in bytes: includes:
+ *	- struct mic_device_desc
+ *	- struct mic_vqconfig (num_vq of these)
+ *	- host and guest features
+ *	- virtio device config space
+ */
+#define MIC_MAX_DESC_BLK_SIZE		256
+
+/**
+ * struct _mic_vring_info - Host vring info exposed to userspace backend
+ * for the avail index and magic for the card.
+ *
+ * @avail_idx: host avail idx
+ * @magic: A magic debug cookie.
+ */
+struct _mic_vring_info {
+	__u16 avail_idx;
+	int magic;
+};
+
+/**
+ * struct mic_vring - Vring information.
+ *
+ * @vr: The virtio ring.
+ * @info: Host vring information exposed to the userspace backend for the
+ * avail index and magic for the card.
+ * @va: The va for the buffer allocated for vr and info.
+ * @len: The length of the buffer required for allocating vr and info.
+ */
+struct mic_vring {
+	struct vring vr;
+	struct _mic_vring_info *info;
+	void *va;
+	int len;
+};
+
+#define mic_aligned_desc_size(d) ALIGN(mic_desc_size(d), 8)
+
+#ifndef INTEL_MIC_CARD
+static inline unsigned mic_desc_size(const struct mic_device_desc *desc)
+{
+	return mic_aligned_size(*desc)
+		+ desc->num_vq * mic_aligned_size(struct mic_vqconfig)
+		+ desc->feature_len * 2
+		+ desc->config_len;
+}
+
+static inline struct mic_vqconfig *
+mic_vq_config(const struct mic_device_desc *desc)
+{
+	return (struct mic_vqconfig *)(desc + 1);
+}
+
+static inline __u8 *mic_vq_features(const struct mic_device_desc *desc)
+{
+	return (__u8 *)(mic_vq_config(desc) + desc->num_vq);
+}
+
+static inline __u8 *mic_vq_configspace(const struct mic_device_desc *desc)
+{
+	return mic_vq_features(desc) + desc->feature_len * 2;
+}
+static inline unsigned mic_total_desc_size(struct mic_device_desc *desc)
+{
+	return mic_aligned_desc_size(desc) +
+		mic_aligned_size(struct mic_device_ctrl);
+}
+#endif
+
 /* Device page size */
 #define MIC_DP_SIZE 4096
 
diff --git a/include/uapi/linux/mic_ioctl.h b/include/uapi/linux/mic_ioctl.h
new file mode 100644
index 000000000000..0e6cbf3e5292
--- /dev/null
+++ b/include/uapi/linux/mic_ioctl.h
@@ -0,0 +1,74 @@
+/*
+ * Intel MIC Platform Software Stack (MPSS)
+ *
+ * Copyright(c) 2013 Intel Corporation.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * The full GNU General Public License is included in this distribution in
+ * the file called "COPYING".
+ *
+ * Intel MIC Host driver.
+ *
+ */
+#ifndef _MIC_IOCTL_H_
+#define _MIC_IOCTL_H_
+
+/*
+ * mic_copy - MIC virtio descriptor copy.
+ *
+ * @iov: An array of IOVEC structures containing user space buffers.
+ * @iovcnt: Number of IOVEC structures in iov.
+ * @vr_idx: The vring index.
+ * @update_used: A non zero value results in used index being updated.
+ * @out_len: The aggregate of the total length written to or read from
+ *	the virtio device.
+ */
+struct mic_copy_desc {
+#ifdef __KERNEL__
+	struct iovec __user *iov;
+#else
+	struct iovec *iov;
+#endif
+	int iovcnt;
+	__u8 vr_idx;
+	__u8 update_used;
+	__u32 out_len;
+};
+
+/*
+ * Add a new virtio device
+ * The (struct mic_device_desc *) pointer points to a device page entry
+ *	for the virtio device consisting of:
+ *	- struct mic_device_desc
+ *	- struct mic_vqconfig (num_vq of these)
+ *	- host and guest features
+ *	- virtio device config space
+ * The total size referenced by the pointer should equal the size returned
+ * by desc_size() in mic_common.h
+ */
+#define MIC_VIRTIO_ADD_DEVICE _IOWR('s', 1, struct mic_device_desc *)
+
+/*
+ * Copy the number of entries in the iovec and update the used index
+ * if requested by the user.
+ */
+#define MIC_VIRTIO_COPY_DESC	_IOWR('s', 2, struct mic_copy_desc *)
+
+/*
+ * Notify virtio device of a config change
+ * The (__u8 *) pointer points to config space values for the device
+ * as they should be written into the device page. The total size
+ * referenced by the pointer should equal the config_len field of struct
+ * mic_device_desc.
+ */
+#define MIC_VIRTIO_CONFIG_CHANGE _IOWR('s', 5, __u8 *)
+
+#endif