Hello all, This are the latest version of the patches. Lots of things have changed since the last submission. A few of which I remember: - VNC copy / paste works* (* conditions apply) - client vnc copies get propagated to guest port 3 (/dev/vmch3) - guest writes to port 3 (/dev/vmch3) go straight to client's clipboard - sysfs hooks to autodiscover ports - support for 64 ports in this version (MAX_VIRTIO_SERIAL_PORTS). More ports can be added by introducing a new feature flag to maintain backward compat. However, till this code gets accepted upstream, the value of that #define can change. I think 64 ports are enough for everyone. - remove support for control queue (though this queue could make a comeback for just one use-case that I can currently think of: to prevent rogue (host) userspace putting lots of data into a guest that goes unconsumed for a while, increasing the memory pressure. To prevent this a threshold level can be decided upon and a control message can be sent to host userspace to prevent any more writes to the port. - numerous fixes There still exist a few kmalloc/kfree-related debug logs that spew up in the guest but I haven't been able to track them down. As for the merge with virtio-console, Christian has voiced some oppostion to that idea. For the merge to happen, the kernel folks have to agree in merging the driver as well and I can proceed once we have a resolution on this. Other than that, a few more rebases have to be done to the qemu code to make it apply to qemu-upstream. Please give this a good review. Thanks, Amit.
Amit Shah
2009-Jul-27 18:04 UTC
[PATCH 1/1] virtio_serial: A char device for simple guest <-> host communication
We expose multiple char devices ("ports") for simple communication between the host userspace and guest. Sample offline usages can be: poking around in a guest to find out the file systems used, applications installed, etc. Online usages can be sharing of clipboard data between the guest and the host, sending information about logged-in users to the host, locking the screen or session when a vnc session is closed, and so on. The interface presented to guest userspace is of a simple char device, so it can be used like this: fd = open("/dev/vmch0", O_RDWR); ret = read(fd, buf, 100); ret = write(fd, string, strlen(string)); Each port is to be assigned a unique function, for example, the first 4 ports may be reserved for libvirt usage, the next 4 for generic streaming data and so on. This port-function mapping isn't finalised yet. For requirements, use-cases and some history see http://www.linux-kvm.org/page/VMchannel_Requirements Signed-off-by: Amit Shah <amit.shah at redhat.com> --- drivers/char/Kconfig | 6 + drivers/char/Makefile | 1 + drivers/char/virtio_serial.c | 690 +++++++++++++++++++++++++++++++++++++++++ include/linux/virtio_serial.h | 27 ++ 4 files changed, 724 insertions(+), 0 deletions(-) create mode 100644 drivers/char/virtio_serial.c create mode 100644 include/linux/virtio_serial.h diff --git a/drivers/char/Kconfig b/drivers/char/Kconfig index 6a06913..51adb48 100644 --- a/drivers/char/Kconfig +++ b/drivers/char/Kconfig @@ -679,6 +679,12 @@ config VIRTIO_CONSOLE help Virtio console for use with lguest and other hypervisors. +config VIRTIO_SERIAL + tristate "Virtio serial" + select VIRTIO + select VIRTIO_RING + help + Virtio serial device driver for simple guest and host communication config HVCS tristate "IBM Hypervisor Virtual Console Server support" diff --git a/drivers/char/Makefile b/drivers/char/Makefile index 66f779a..5e1915b 100644 --- a/drivers/char/Makefile +++ b/drivers/char/Makefile @@ -54,6 +54,7 @@ obj-$(CONFIG_HVC_XEN) += hvc_xen.o obj-$(CONFIG_HVC_IUCV) += hvc_iucv.o obj-$(CONFIG_HVC_UDBG) += hvc_udbg.o obj-$(CONFIG_VIRTIO_CONSOLE) += virtio_console.o +obj-$(CONFIG_VIRTIO_SERIAL) += virtio_serial.o obj-$(CONFIG_RAW_DRIVER) += raw.o obj-$(CONFIG_SGI_SNSC) += snsc.o snsc_event.o obj-$(CONFIG_MSPEC) += mspec.o diff --git a/drivers/char/virtio_serial.c b/drivers/char/virtio_serial.c new file mode 100644 index 0000000..f42709d --- /dev/null +++ b/drivers/char/virtio_serial.c @@ -0,0 +1,690 @@ +/* + * VirtIO Serial driver + * + * This is paravirtualised serial guest<->host communication channel + * for relaying short messages and events in either direction. + * + * One PCI device can have multiple serial ports so that different + * applications can communicate without polluting the PCI device space + * in the guest. + * + * Copyright (C) 2009, Red Hat, Inc. + * + * Author(s): Amit Shah <amit.shah at redhat.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * 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. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <linux/cdev.h> +#include <linux/completion.h> +#include <linux/device.h> +#include <linux/err.h> +#include <linux/fs.h> +#include <linux/init.h> +#include <linux/poll.h> +#include <linux/virtio.h> +#include <linux/virtio_serial.h> +#include <linux/workqueue.h> + +struct virtio_serial_struct { + struct work_struct rx_work; + struct work_struct tx_work; + struct work_struct queue_work; + struct work_struct config_work; + + struct list_head port_head; + + struct virtio_device *vdev; + struct class *class; + struct virtqueue *in_vq, *out_vq; + + struct virtio_serial_config *config; +}; + +/* This struct holds individual buffers received for each port */ +struct virtio_serial_port_buffer { + struct list_head list; + + unsigned int len; /* length of the buffer */ + unsigned int offset; /* offset in the buf from which to consume data */ + + char *buf; +}; + +/* This struct is put in each buffer that gets passed to userspace and + * vice-versa + */ +struct virtio_serial_id { + u32 id; /* Port number */ +}; + +struct virtio_serial_port { + /* Next port in the list */ + struct list_head next; + + /* Buffer management */ + struct virtio_serial_port_buffer read_buf; + struct list_head readbuf_head; + struct completion have_data; + + /* Each port associates with a separate char device */ + struct cdev cdev; + struct device *dev; +}; + +static struct virtio_serial_struct virtserial; + +static int major = 60; /* from the experimental range */ + +static struct virtio_serial_port *get_port_from_id(u32 id) +{ + struct virtio_serial_port *port; + struct list_head *ptr; + + list_for_each(ptr, &virtserial.port_head) { + port = list_entry(ptr, struct virtio_serial_port, next); + + if (MINOR(port->dev->devt) == id) + return port; + } + return NULL; +} + +static int get_id_from_port(struct virtio_serial_port *port) +{ + struct virtio_serial_port *match; + struct list_head *ptr; + + list_for_each(ptr, &virtserial.port_head) { + match = list_entry(ptr, struct virtio_serial_port, next); + + if (match == port) + return MINOR(port->dev->devt); + } + return VIRTIO_SERIAL_BAD_ID; +} + +static struct virtio_serial_port *get_port_from_buf(char *buf) +{ + u32 id; + + memcpy(&id, buf, sizeof(id)); + + return get_port_from_id(id); +} + + +static ssize_t virtserial_read(struct file *filp, char __user *ubuf, + size_t count, loff_t *offp) +{ + struct list_head *ptr, *ptr2; + struct virtio_serial_port *port; + struct virtio_serial_port_buffer *buf; + ssize_t ret; + + port = filp->private_data; + + ret = -EINTR; + if (list_empty(&port->readbuf_head)) { + if (filp->f_flags & O_NONBLOCK) + return -EAGAIN; + + if (wait_for_completion_interruptible(&port->have_data) < 0) + return ret; + } + list_for_each_safe(ptr, ptr2, &port->readbuf_head) { + buf = list_entry(ptr, struct virtio_serial_port_buffer, list); + + /* FIXME: other buffers further in this list might + * have data too + */ + if (count > buf->len - buf->offset) + count = buf->len - buf->offset; + + ret = copy_to_user(ubuf, buf->buf + buf->offset, count); + + /* Return the number of bytes actually copied */ + ret = count - ret; + + buf->offset += ret; + + if (buf->len - buf->offset == 0) { + list_del(&buf->list); + kfree(buf->buf); + kfree(buf); + } + /* FIXME: if there's more data requested and more data + * available, return it. + */ + break; + } + return ret; +} + +/* For data that exceeds PAGE_SIZE in size we should send it all in + * one sg to not unnecessarily split up the data. Also some (all?) + * vnc clients don't consume split data. + * + * If we are to keep PAGE_SIZE sized buffers, we then have to stack + * multiple of those in one virtio request. virtio-ring returns to us + * just one pointer for all the buffers. So use this struct to + * allocate the bufs in so that freeing this up later is easier. + */ +struct vbuf { + char **bufs; + struct scatterlist *sg; + unsigned int nent; +}; + +static ssize_t virtserial_write(struct file *filp, const char __user *ubuf, + size_t count, loff_t *offp) +{ + struct virtqueue *out_vq; + struct virtio_serial_port *port; + struct virtio_serial_id id; + struct vbuf *vbuf; + size_t offset, size; + ssize_t ret; + int i, id_len; + + port = filp->private_data; + id.id = get_id_from_port(port); + out_vq = virtserial.out_vq; + + id_len = sizeof(id); + + ret = -EFBIG; + vbuf = kzalloc(sizeof(struct vbuf), GFP_KERNEL); + if (!vbuf) + return ret; + + /* Max. number of buffers clubbed together in one message */ + vbuf->nent = (count + id_len + PAGE_SIZE - 1) / PAGE_SIZE; + + vbuf->bufs = kzalloc(vbuf->nent, GFP_KERNEL); + if (!vbuf->bufs) + goto free_vbuf; + + vbuf->sg = kzalloc(vbuf->nent, GFP_KERNEL); + if (!vbuf->sg) + goto free_bufs; + sg_init_table(vbuf->sg, vbuf->nent); + + i = 0; /* vbuf->bufs[i] */ + offset = 0; /* offset in the user buffer */ + while (count - offset) { + size = min(count - offset + id_len, PAGE_SIZE); + vbuf->bufs[i] = kzalloc(size, GFP_KERNEL); + if (!vbuf->bufs[i]) { + ret = -EFBIG; + goto free_buffers; + } + if (id_len) { + memcpy(vbuf->bufs[i], &id, id_len); + size -= id_len; + } + ret = copy_from_user(vbuf->bufs[i] + id_len, ubuf + offset, size); + offset += size - ret; + + sg_set_buf(&vbuf->sg[i], vbuf->bufs[i], size - ret + id_len); + id_len = 0; /* Pass the port id only in the first buffer */ + i++; + } + if (out_vq->vq_ops->add_buf(out_vq, vbuf->sg, i, 0, vbuf)) { + /* XXX: We can't send the buffer. Report failure */ + ret = 0; + } + /* Tell Host to go! */ + out_vq->vq_ops->kick(out_vq); + + /* We're expected to return the amount of data we wrote */ + return offset; +free_buffers: + while (i--) + kfree(vbuf->bufs[i]); + kfree(vbuf->sg); +free_bufs: + kfree(vbuf->bufs); +free_vbuf: + kfree(vbuf); + return ret; +} + +static long virtserial_ioctl(struct file *filp, unsigned int ioctl, + unsigned long arg) +{ + struct virtio_serial_port *port; + long ret; + + port = filp->private_data; + + ret = -EINVAL; + switch (ioctl) { + default: + break; + } + return ret; +} + +static int virtserial_release(struct inode *inode, struct file *filp) +{ + filp->private_data = NULL; + return 0; +} + +static int virtserial_open(struct inode *inode, struct file *filp) +{ + struct cdev *cdev = inode->i_cdev; + struct virtio_serial_port *port; + + port = container_of(cdev, struct virtio_serial_port, cdev); + + filp->private_data = port; + return 0; +} + +static unsigned int virtserial_poll(struct file *filp, poll_table *wait) +{ + pr_notice("%s\n", __func__); + return 0; +} + +static const struct file_operations virtserial_fops = { + .owner = THIS_MODULE, + .open = virtserial_open, + .read = virtserial_read, + .write = virtserial_write, + .compat_ioctl = virtserial_ioctl, + .unlocked_ioctl = virtserial_ioctl, + .poll = virtserial_poll, + .release = virtserial_release, +}; + +static void virtio_serial_queue_work_handler(struct work_struct *work) +{ + struct scatterlist sg[1]; + struct virtqueue *vq; + char *buf; + + vq = virtserial.in_vq; + while (1) { + buf = kzalloc(PAGE_SIZE, GFP_KERNEL); + if (!buf) + break; + + sg_init_one(sg, buf, PAGE_SIZE); + + if (vq->vq_ops->add_buf(vq, sg, 0, 1, buf) < 0) { + kfree(buf); + break; + } + } + vq->vq_ops->kick(vq); +} + +static void virtio_serial_rx_work_handler(struct work_struct *work) +{ + struct virtio_serial_port *port = NULL; + struct virtio_serial_port_buffer *buf; + struct virtqueue *vq; + char *tmpbuf; + unsigned int tmplen; + + vq = virtserial.in_vq; + while ((tmpbuf = vq->vq_ops->get_buf(vq, &tmplen))) { + port = get_port_from_buf(tmpbuf); + if (!port) { + /* No valid index at start of + * buffer. Drop it. + */ + pr_debug("%s: invalid index in buffer, %c %d\n", + __func__, tmpbuf[0], tmpbuf[0]); + break; + } + buf = kzalloc(sizeof(struct virtio_serial_port_buffer), + GFP_KERNEL); + if (!buf) + break; + + buf->buf = tmpbuf; + buf->len = tmplen; + buf->offset = sizeof(struct virtio_serial_id); + list_add_tail(&buf->list, &port->readbuf_head); + + complete(&port->have_data); + } + /* Allocate buffers for all the ones that got used up */ + schedule_work(&virtserial.queue_work); +} + +static void virtio_serial_tx_work_handler(struct work_struct *work) +{ + struct virtqueue *vq; + struct vbuf *vbuf; + unsigned int tmplen; + int i; + + vq = virtserial.out_vq; + while ((vbuf = vq->vq_ops->get_buf(vq, &tmplen))) { + for (i = 0; i < vbuf->nent; i++) { + kfree(vbuf->bufs[i]); + } + kfree(vbuf->bufs); + kfree(vbuf->sg); + kfree(vbuf); + } +} + +static void rx_intr(struct virtqueue *vq) +{ + schedule_work(&virtserial.rx_work); +} + +static void tx_intr(struct virtqueue *vq) +{ + schedule_work(&virtserial.tx_work); +} + +static void config_intr(struct virtio_device *vdev) +{ + schedule_work(&virtserial.config_work); +} + +static u32 virtserial_get_hot_add_port(struct virtio_serial_config *config) +{ + u32 i; + u32 port_nr; + + for (i = 0; i < virtserial.config->max_nr_ports / 32; i++) { + port_nr = ffs(config->ports_map[i] ^ virtserial.config->ports_map[i]); + if (port_nr) + break; + } + if (unlikely(!port_nr)) + return VIRTIO_SERIAL_BAD_ID; + + /* We used ffs above */ + port_nr--; + + /* FIXME: Do this only when add_port is successful */ + virtserial.config->ports_map[i] |= 1U << port_nr; + + port_nr += i * 32; + return port_nr; +} + +static u32 virtserial_find_next_port(u32 *map, int *map_i) +{ + u32 port_nr; + + while (1) { + port_nr = ffs(*map); + if (port_nr) + break; + + if (unlikely(*map_i >= virtserial.config->max_nr_ports / 32)) + return VIRTIO_SERIAL_BAD_ID; + ++*map_i; + *map = virtserial.config->ports_map[*map_i]; + } + /* We used ffs above */ + port_nr--; + + /* FIXME: Do this only when add_port is successful / reset bit + * in config space if add_port was unsuccessful + */ + *map &= ~(1U << port_nr); + + port_nr += *map_i * 32; + return port_nr; +} + +static int virtserial_add_port(u32 port_nr) +{ + struct virtio_serial_port *port; + dev_t devt; + int ret; + + port = kzalloc(sizeof(struct virtio_serial_port), GFP_KERNEL); + if (!port) + return -ENOMEM; + + devt = MKDEV(major, port_nr); + cdev_init(&port->cdev, &virtserial_fops); + + ret = register_chrdev_region(devt, 1, "virtio-serial"); + if (ret < 0) { + pr_err("%s: error registering chrdev region, ret = %d\n", + __func__, ret); + goto free_cdev; + } + ret = cdev_add(&port->cdev, devt, 1); + if (ret < 0) { + pr_err("%s: error adding cdev, ret = %d\n", __func__, ret); + goto free_cdev; + } + port->dev = device_create(virtserial.class, NULL, devt, NULL, + "vmch%u", port_nr); + if (IS_ERR(port->dev)) { + ret = PTR_ERR(port->dev); + pr_err("%s: Error creating device, ret = %d\n", __func__, ret); + goto free_cdev; + } + INIT_LIST_HEAD(&port->readbuf_head); + init_completion(&port->have_data); + + list_add_tail(&port->next, &virtserial.port_head); + + pr_info("virtio-serial port found at id %u\n", port_nr); + + return 0; +free_cdev: + unregister_chrdev(major, "virtio-serial"); + return ret; +} + +static __u32 get_ports_map_size(__u32 max_ports) +{ + return sizeof(__u32) * ((max_ports + 31) / 32); +} + +static void virtio_serial_config_work_handler(struct work_struct *work) +{ + struct virtio_serial_config *virtserconf; + struct virtio_device *vdev = virtserial.vdev; + u32 i, port_nr; + int ret; + + virtserconf = kmalloc(sizeof(struct virtio_serial_config) + + get_ports_map_size(virtserial.config->max_nr_ports), + GFP_KERNEL); + vdev->config->get(vdev, + offsetof(struct virtio_serial_config, nr_active_ports), + &virtserconf->nr_active_ports, + sizeof(virtserconf->nr_active_ports)); + vdev->config->get(vdev, + offsetof(struct virtio_serial_config, ports_map), + virtserconf->ports_map, + get_ports_map_size(virtserial.config->max_nr_ports)); + + /* Hot-add ports */ + for (i = virtserial.config->nr_active_ports; + i < virtserconf->nr_active_ports; i++) { + port_nr = virtserial_get_hot_add_port(virtserconf); + if (port_nr == VIRTIO_SERIAL_BAD_ID) + continue; + ret = virtserial_add_port(port_nr); + if (!ret) + virtserial.config->nr_active_ports++; + } + kfree(virtserconf); +} + +static int virtserial_probe(struct virtio_device *vdev) +{ + struct virtqueue *vqs[3]; + const char *vq_names[] = { "input", "output" }; + vq_callback_t *vq_callbacks[] = { rx_intr, tx_intr }; + u32 i, map; + int ret, map_i; + u32 max_nr_ports; + + vdev->config->get(vdev, offsetof(struct virtio_serial_config, + max_nr_ports), + &max_nr_ports, + sizeof(max_nr_ports)); + virtserial.config = kmalloc(sizeof(struct virtio_serial_config) + + get_ports_map_size(max_nr_ports), + GFP_KERNEL); + virtserial.config->max_nr_ports = max_nr_ports; + + vdev->config->get(vdev, offsetof(struct virtio_serial_config, + nr_active_ports), + &virtserial.config->nr_active_ports, + sizeof(virtserial.config->nr_active_ports)); + vdev->config->get(vdev, + offsetof(struct virtio_serial_config, ports_map), + virtserial.config->ports_map, + get_ports_map_size(max_nr_ports)); + + virtserial.vdev = vdev; + + ret = vdev->config->find_vqs(vdev, 2, vqs, vq_callbacks, vq_names); + if (ret) + goto fail; + + virtserial.in_vq = vqs[0]; + virtserial.out_vq = vqs[1]; + + INIT_LIST_HEAD(&virtserial.port_head); + + map_i = 0; + map = virtserial.config->ports_map[map_i]; + for (i = 0; i < virtserial.config->nr_active_ports; i++) { + __u32 port_nr; + + port_nr = virtserial_find_next_port(&map, &map_i); + if (unlikely(port_nr == VIRTIO_SERIAL_BAD_ID)) + continue; + + virtserial_add_port(port_nr); + } + INIT_WORK(&virtserial.rx_work, &virtio_serial_rx_work_handler); + INIT_WORK(&virtserial.tx_work, &virtio_serial_tx_work_handler); + INIT_WORK(&virtserial.queue_work, &virtio_serial_queue_work_handler); + INIT_WORK(&virtserial.config_work, &virtio_serial_config_work_handler); + + /* Allocate pages to fill the receive queue */ + schedule_work(&virtserial.queue_work); + + return 0; +fail: + return ret; +} + + +static void virtserial_remove_port_data(struct virtio_serial_port *port) +{ + struct list_head *ptr, *ptr2; + + device_destroy(virtserial.class, port->dev->devt); + unregister_chrdev_region(port->dev->devt, 1); + cdev_del(&port->cdev); + + /* Remove the buffers in which we have unconsumed data */ + list_for_each_safe(ptr, ptr2, &port->readbuf_head) { + struct virtio_serial_port_buffer *buf; + + buf = list_entry(ptr, struct virtio_serial_port_buffer, list); + + list_del(&buf->list); + kfree(buf->buf); + kfree(buf); + } +} + +static void virtserial_remove(struct virtio_device *vdev) +{ + struct list_head *ptr, *ptr2; + char *buf; + int len; + + unregister_chrdev(major, "virtio-serial"); + class_destroy(virtserial.class); + + cancel_work_sync(&virtserial.rx_work); + + /* Free up the unused buffers in the receive queue */ + while ((buf = virtserial.in_vq->vq_ops->get_buf(virtserial.in_vq, &len))) + kfree(buf); + + vdev->config->del_vqs(vdev); + + list_for_each_safe(ptr, ptr2, &virtserial.port_head) { + struct virtio_serial_port *port; + + port = list_entry(ptr, struct virtio_serial_port, next); + + list_del(&port->next); + virtserial_remove_port_data(port); + kfree(port); + } + kfree(virtserial.config); +} + +static struct virtio_device_id id_table[] = { + { VIRTIO_ID_SERIAL, VIRTIO_DEV_ANY_ID }, + { 0 }, +}; + +static struct virtio_driver virtio_serial = { + // .feature_table = features, + // .feature_table_size = ARRAY_SIZE(features), + .driver.name = KBUILD_MODNAME, + .driver.owner = THIS_MODULE, + .id_table = id_table, + .probe = virtserial_probe, + .remove = virtserial_remove, + .config_changed = config_intr, +}; + +static int __init init(void) +{ + int ret; + + virtserial.class = class_create(THIS_MODULE, "virtio-serial"); + if (IS_ERR(virtserial.class)) { + pr_err("Error creating virtio-serial class\n"); + ret = PTR_ERR(virtserial.class); + return ret; + } + ret = register_virtio_driver(&virtio_serial); + if (ret) { + class_destroy(virtserial.class); + return ret; + } + return 0; +} + +static void __exit fini(void) +{ + unregister_virtio_driver(&virtio_serial); +} +module_init(init); +module_exit(fini); + +MODULE_DEVICE_TABLE(virtio, id_table); +MODULE_DESCRIPTION("Virtio serial driver"); +MODULE_LICENSE("GPL"); diff --git a/include/linux/virtio_serial.h b/include/linux/virtio_serial.h new file mode 100644 index 0000000..025dcf1 --- /dev/null +++ b/include/linux/virtio_serial.h @@ -0,0 +1,27 @@ +#ifndef _LINUX_VIRTIO_SERIAL_H +#define _LINUX_VIRTIO_SERIAL_H +#include <linux/types.h> +#include <linux/virtio_config.h> + +/* Guest kernel - Host interface */ + +/* The ID for virtio serial */ +#define VIRTIO_ID_SERIAL 7 + +#define VIRTIO_SERIAL_BAD_ID (~(u32)0) + +struct virtio_serial_config { + __u32 max_nr_ports; + __u32 nr_active_ports; + __u32 ports_map[0 /* (max_nr_ports + 31) / 32 */]; +}; + +#ifdef __KERNEL__ + +/* Guest kernel - Guest userspace interface */ + +/* IOCTL-related */ +#define VIRTIO_SERIAL_IO 0xAF + +#endif /* __KERNEL__ */ +#endif /* _LINUX_VIRTIO_SERIAL_H */ -- 1.6.2.5
Amit Shah
2009-Jul-27 18:04 UTC
[PATCH 3/3] virtio-serial: vnc: support for sending / receiving guest clipboard
- Send ServerCutText message over to the vnc client from qemu on every write to the clipboard port via virtio-serial - On receiving ClientCutText message send over the data to the guest via virtio-serial. Signed-off-by: Amit Shah <amit.shah at redhat.com> --- hw/virtio-serial.c | 15 +++++++++++++++ hw/virtio-serial.h | 4 ++++ vnc.c | 24 ++++++++++++++++++++++++ vnc.h | 2 ++ 4 files changed, 45 insertions(+), 0 deletions(-) diff --git a/hw/virtio-serial.c b/hw/virtio-serial.c index d77af9b..95c715f 100644 --- a/hw/virtio-serial.c +++ b/hw/virtio-serial.c @@ -20,6 +20,7 @@ #include "pci.h" #include "monitor.h" #include "qemu-char.h" +#include "vnc.h" #include "virtio.h" #include "virtio-serial.h" @@ -118,6 +119,10 @@ static size_t flush_buf(uint32_t id, const uint8_t *buf, size_t len) VirtIOSerialPort *port; size_t write_len = 0; + if (id == VIRTIO_SERIAL_CLIPBOARD_PORT && is_vnc_active()) { + vnc_clipboard_data_from_guest(buf, len); + return len; + } port = get_port_from_id(id); if (port && port->hd) { write_len = qemu_chr_write(port->hd, buf, len); @@ -211,6 +216,16 @@ static void write_to_port(VirtIOSerialPort *port, virtio_notify(port->virtserial->vdev, vq); } +void virtio_serial_send_clipboard_to_guest(const uint8_t *buf, size_t len) +{ + VirtIOSerialPort *port = get_port_from_id(VIRTIO_SERIAL_CLIPBOARD_PORT); + + if (!port) { + return; + } + write_to_port(port, buf, len); +} + /* Readiness of the guest to accept data on a port */ static int cons_can_read(void *opaque) { diff --git a/hw/virtio-serial.h b/hw/virtio-serial.h index 08c4b51..44238f6 100644 --- a/hw/virtio-serial.h +++ b/hw/virtio-serial.h @@ -18,6 +18,9 @@ #define VIRTIO_ID_SERIAL 7 #define VIRTIO_SERIAL_BAD_ID (~(uint32_t)0) +/* Port number to function mapping */ +#define VIRTIO_SERIAL_CLIPBOARD_PORT 3 + struct virtio_serial_config { uint32_t max_nr_ports; @@ -32,5 +35,6 @@ struct virtio_serial_config void *virtio_serial_new_port(PCIDevice *dev, int idx); void virtio_serial_monitor_command(Monitor *mon, const char *command, const char *param); +void virtio_serial_send_clipboard_to_guest(const uint8_t *buf, size_t len); #endif diff --git a/vnc.c b/vnc.c index e4e78dc..271b64e 100644 --- a/vnc.c +++ b/vnc.c @@ -29,6 +29,7 @@ #include "qemu_socket.h" #include "qemu-timer.h" #include "acl.h" +#include "hw/virtio-serial.h" #define VNC_REFRESH_INTERVAL (1000 / 30) @@ -671,6 +672,28 @@ static void vnc_copy(VncState *vs, int src_x, int src_y, int dst_x, int dst_y, i vnc_flush(vs); } +void vnc_clipboard_data_from_guest(const uint8_t *buf, size_t len) +{ + VncState *vs; + VncDisplay *vd; + DisplayState *ds; + + ds = vnc_display->ds; + vd = ds->opaque; + + for (vs = vd->clients; vs; vs = vs->next) { + vnc_write_u8(vs, 3); /* ServerCutText */ + vnc_write_u8(vs, 0); /* Padding */ + vnc_write_u8(vs, 0); /* Padding */ + vnc_write_u8(vs, 0); /* Padding */ + vnc_write_u32(vs, len); + while (len--) { + vnc_write_u8(vs, *(buf++)); + } + vnc_flush(vs); + } +} + static void vnc_dpy_copy(DisplayState *ds, int src_x, int src_y, int dst_x, int dst_y, int w, int h) { VncDisplay *vd = ds->opaque; @@ -1239,6 +1262,7 @@ uint32_t read_u32(uint8_t *data, size_t offset) static void client_cut_text(VncState *vs, size_t len, uint8_t *text) { + virtio_serial_send_clipboard_to_guest(text, len); } static void check_pointer_type_change(VncState *vs, int absolute) diff --git a/vnc.h b/vnc.h index 9739c35..00afdbb 100644 --- a/vnc.h +++ b/vnc.h @@ -303,6 +303,8 @@ int vnc_client_io_error(VncState *vs, int ret, int last_errno); void start_client_init(VncState *vs); void start_auth_vnc(VncState *vs); +void vnc_clipboard_data_from_guest(const uint8_t *buf, size_t len); + /* Buffer management */ void buffer_reserve(Buffer *buffer, size_t len); int buffer_empty(Buffer *buffer); -- 1.6.2.5
Anthony Liguori
2009-Jul-27 20:22 UTC
virtio-serial: An interface for host-guest communication
Amit Shah wrote:> Hello all, > > This are the latest version of the patches. > > Lots of things have changed since the last submission. A few of > which I remember: > - VNC copy / paste works* (* conditions apply) > - client vnc copies get propagated to guest port 3 (/dev/vmch3) > - guest writes to port 3 (/dev/vmch3) go straight to client's clipboard >Why 3? Where's the guest application that drives the copy/paste? I expect the first problem you'll run into is that copy/paste daemon has to run as an unprivileged user but /dev/vmch3 is going to be owned by root. You could set udev rules for /dev/vmch3 but that's pretty terrible IMHO. I think you'll find that you need a root daemon that talks to vmchannel and then allows unprivileged connections over a unix socket device. In that case, why even bother having multiple channels since your daemon can multiplex.. Regards, Anthony Liguori
Reasonably Related Threads
- virtio-serial: An interface for host-guest communication
- [RFC PATCH v3] virito-serial: A guest <-> host interface
- [RFC PATCH v3] virito-serial: A guest <-> host interface
- virtio-serial: A guest <-> host interface for simple communication
- virtio-serial: A guest <-> host interface for simple communication