*
* Copyright (C) 2006 by Russ Cox <rsc@swtch.com>
* Copyright (C) 2004-2005 by Latchesar Ionkov <lucho@ionkov.net>
- * Copyright (C) 2004-2005 by Eric Van Hensbergen <ericvh@gmail.com>
+ * Copyright (C) 2004-2008 by Eric Van Hensbergen <ericvh@gmail.com>
* Copyright (C) 1997-2002 by Ron Minnich <rminnich@sarnoff.com>
*
* This program is free software; you can redistribute it and/or modify
#include <linux/module.h>
#include <linux/net.h>
#include <linux/ipv6.h>
+#include <linux/kthread.h>
#include <linux/errno.h>
#include <linux/kernel.h>
#include <linux/un.h>
#include <linux/inet.h>
#include <linux/idr.h>
#include <linux/file.h>
+#include <linux/parser.h>
#include <net/9p/9p.h>
+#include <net/9p/client.h>
#include <net/9p/transport.h>
#define P9_PORT 564
+#define MAX_SOCK_BUF (64*1024)
+#define MAXPOLLWADDR 2
+
+/**
+ * struct p9_fd_opts - per-transport options
+ * @rfd: file descriptor for reading (trans=fd)
+ * @wfd: file descriptor for writing (trans=fd)
+ * @port: port to connect to (trans=tcp)
+ *
+ */
+
+struct p9_fd_opts {
+ int rfd;
+ int wfd;
+ u16 port;
+};
+
+/**
+ * struct p9_trans_fd - transport state
+ * @rd: reference to file to read from
+ * @wr: reference of file to write to
+ * @conn: connection state reference
+ *
+ */
struct p9_trans_fd {
struct file *rd;
struct file *wr;
+ struct p9_conn *conn;
};
-static int p9_socket_open(struct p9_transport *trans, struct socket *csocket);
-static int p9_fd_open(struct p9_transport *trans, int rfd, int wfd);
-static int p9_fd_read(struct p9_transport *trans, void *v, int len);
-static int p9_fd_write(struct p9_transport *trans, void *v, int len);
-static unsigned int p9_fd_poll(struct p9_transport *trans,
- struct poll_table_struct *pt);
-static void p9_fd_close(struct p9_transport *trans);
+/*
+ * Option Parsing (code inspired by NFS code)
+ * - a little lazy - parse all fd-transport options
+ */
-struct p9_transport *p9_trans_create_tcp(const char *addr, int port)
-{
+enum {
+ /* Options that take integer arguments */
+ Opt_port, Opt_rfdno, Opt_wfdno, Opt_err,
+};
+
+static const match_table_t tokens = {
+ {Opt_port, "port=%u"},
+ {Opt_rfdno, "rfdno=%u"},
+ {Opt_wfdno, "wfdno=%u"},
+ {Opt_err, NULL},
+};
+
+enum {
+ Rworksched = 1, /* read work scheduled or running */
+ Rpending = 2, /* can read */
+ Wworksched = 4, /* write work scheduled or running */
+ Wpending = 8, /* can write */
+};
+
+struct p9_poll_wait {
+ struct p9_conn *conn;
+ wait_queue_t wait;
+ wait_queue_head_t *wait_addr;
+};
+
+/**
+ * struct p9_conn - fd mux connection state information
+ * @mux_list: list link for mux to manage multiple connections (?)
+ * @client: reference to client instance for this connection
+ * @err: error state
+ * @req_list: accounting for requests which have been sent
+ * @unsent_req_list: accounting for requests that haven't been sent
+ * @req: current request being processed (if any)
+ * @tmp_buf: temporary buffer to read in header
+ * @rsize: amount to read for current frame
+ * @rpos: read position in current frame
+ * @rbuf: current read buffer
+ * @wpos: write position for current frame
+ * @wsize: amount of data to write for current frame
+ * @wbuf: current write buffer
+ * @poll_wait: array of wait_q's for various worker threads
+ * @poll_waddr: ????
+ * @pt: poll state
+ * @rq: current read work
+ * @wq: current write work
+ * @wsched: ????
+ *
+ */
+
+struct p9_conn {
+ struct list_head mux_list;
+ struct p9_client *client;
int err;
- struct p9_transport *trans;
- struct socket *csocket;
- struct sockaddr_in sin_server;
+ struct list_head req_list;
+ struct list_head unsent_req_list;
+ struct p9_req_t *req;
+ char tmp_buf[7];
+ int rsize;
+ int rpos;
+ char *rbuf;
+ int wpos;
+ int wsize;
+ char *wbuf;
+ struct list_head poll_pending_link;
+ struct p9_poll_wait poll_wait[MAXPOLLWADDR];
+ poll_table pt;
+ struct work_struct rq;
+ struct work_struct wq;
+ unsigned long wsched;
+};
- csocket = NULL;
- trans = kmalloc(sizeof(struct p9_transport), GFP_KERNEL);
- if (!trans)
- return ERR_PTR(-ENOMEM);
+static DEFINE_SPINLOCK(p9_poll_lock);
+static LIST_HEAD(p9_poll_pending_list);
+static struct workqueue_struct *p9_mux_wq;
+static struct task_struct *p9_poll_task;
- trans->write = p9_fd_write;
- trans->read = p9_fd_read;
- trans->close = p9_fd_close;
- trans->poll = p9_fd_poll;
+static void p9_mux_poll_stop(struct p9_conn *m)
+{
+ unsigned long flags;
+ int i;
- sin_server.sin_family = AF_INET;
- sin_server.sin_addr.s_addr = in_aton(addr);
- sin_server.sin_port = htons(port);
- sock_create_kern(PF_INET, SOCK_STREAM, IPPROTO_TCP, &csocket);
+ for (i = 0; i < ARRAY_SIZE(m->poll_wait); i++) {
+ struct p9_poll_wait *pwait = &m->poll_wait[i];
- if (!csocket) {
- P9_EPRINTK(KERN_ERR, "p9_trans_tcp: problem creating socket\n");
- err = -EIO;
- goto error;
+ if (pwait->wait_addr) {
+ remove_wait_queue(pwait->wait_addr, &pwait->wait);
+ pwait->wait_addr = NULL;
+ }
}
- err = csocket->ops->connect(csocket,
- (struct sockaddr *)&sin_server,
- sizeof(struct sockaddr_in), 0);
- if (err < 0) {
- P9_EPRINTK(KERN_ERR,
- "p9_trans_tcp: problem connecting socket to %s\n",
- addr);
- goto error;
+ spin_lock_irqsave(&p9_poll_lock, flags);
+ list_del_init(&m->poll_pending_link);
+ spin_unlock_irqrestore(&p9_poll_lock, flags);
+}
+
+/**
+ * p9_conn_cancel - cancel all pending requests with error
+ * @m: mux data
+ * @err: error code
+ *
+ */
+
+static void p9_conn_cancel(struct p9_conn *m, int err)
+{
+ struct p9_req_t *req, *rtmp;
+ unsigned long flags;
+ LIST_HEAD(cancel_list);
+
+ P9_DPRINTK(P9_DEBUG_ERROR, "mux %p err %d\n", m, err);
+
+ spin_lock_irqsave(&m->client->lock, flags);
+
+ if (m->err) {
+ spin_unlock_irqrestore(&m->client->lock, flags);
+ return;
}
- err = p9_socket_open(trans, csocket);
- if (err < 0)
+ m->err = err;
+
+ list_for_each_entry_safe(req, rtmp, &m->req_list, req_list) {
+ req->status = REQ_STATUS_ERROR;
+ if (!req->t_err)
+ req->t_err = err;
+ list_move(&req->req_list, &cancel_list);
+ }
+ list_for_each_entry_safe(req, rtmp, &m->unsent_req_list, req_list) {
+ req->status = REQ_STATUS_ERROR;
+ if (!req->t_err)
+ req->t_err = err;
+ list_move(&req->req_list, &cancel_list);
+ }
+ spin_unlock_irqrestore(&m->client->lock, flags);
+
+ list_for_each_entry_safe(req, rtmp, &cancel_list, req_list) {
+ list_del(&req->req_list);
+ P9_DPRINTK(P9_DEBUG_ERROR, "call back req %p\n", req);
+ p9_client_cb(m->client, req);
+ }
+}
+
+static unsigned int
+p9_fd_poll(struct p9_client *client, struct poll_table_struct *pt)
+{
+ int ret, n;
+ struct p9_trans_fd *ts = NULL;
+
+ if (client && client->status == Connected)
+ ts = client->trans;
+
+ if (!ts)
+ return -EREMOTEIO;
+
+ if (!ts->rd->f_op || !ts->rd->f_op->poll)
+ return -EIO;
+
+ if (!ts->wr->f_op || !ts->wr->f_op->poll)
+ return -EIO;
+
+ ret = ts->rd->f_op->poll(ts->rd, pt);
+ if (ret < 0)
+ return ret;
+
+ if (ts->rd != ts->wr) {
+ n = ts->wr->f_op->poll(ts->wr, pt);
+ if (n < 0)
+ return n;
+ ret = (ret & ~POLLOUT) | (n & ~POLLIN);
+ }
+
+ return ret;
+}
+
+/**
+ * p9_fd_read- read from a fd
+ * @client: client instance
+ * @v: buffer to receive data into
+ * @len: size of receive buffer
+ *
+ */
+
+static int p9_fd_read(struct p9_client *client, void *v, int len)
+{
+ int ret;
+ struct p9_trans_fd *ts = NULL;
+
+ if (client && client->status != Disconnected)
+ ts = client->trans;
+
+ if (!ts)
+ return -EREMOTEIO;
+
+ if (!(ts->rd->f_flags & O_NONBLOCK))
+ P9_DPRINTK(P9_DEBUG_ERROR, "blocking read ...\n");
+
+ ret = kernel_read(ts->rd, ts->rd->f_pos, v, len);
+ if (ret <= 0 && ret != -ERESTARTSYS && ret != -EAGAIN)
+ client->status = Disconnected;
+ return ret;
+}
+
+/**
+ * p9_read_work - called when there is some data to be read from a transport
+ * @work: container of work to be done
+ *
+ */
+
+static void p9_read_work(struct work_struct *work)
+{
+ int n, err;
+ struct p9_conn *m;
+
+ m = container_of(work, struct p9_conn, rq);
+
+ if (m->err < 0)
+ return;
+
+ P9_DPRINTK(P9_DEBUG_TRANS, "start mux %p pos %d\n", m, m->rpos);
+
+ if (!m->rbuf) {
+ m->rbuf = m->tmp_buf;
+ m->rpos = 0;
+ m->rsize = 7; /* start by reading header */
+ }
+
+ clear_bit(Rpending, &m->wsched);
+ P9_DPRINTK(P9_DEBUG_TRANS, "read mux %p pos %d size: %d = %d\n", m,
+ m->rpos, m->rsize, m->rsize-m->rpos);
+ err = p9_fd_read(m->client, m->rbuf + m->rpos,
+ m->rsize - m->rpos);
+ P9_DPRINTK(P9_DEBUG_TRANS, "mux %p got %d bytes\n", m, err);
+ if (err == -EAGAIN) {
+ clear_bit(Rworksched, &m->wsched);
+ return;
+ }
+
+ if (err <= 0)
goto error;
- return trans;
+ m->rpos += err;
+
+ if ((!m->req) && (m->rpos == m->rsize)) { /* header read in */
+ u16 tag;
+ P9_DPRINTK(P9_DEBUG_TRANS, "got new header\n");
+
+ n = le32_to_cpu(*(__le32 *) m->rbuf); /* read packet size */
+ if (n >= m->client->msize) {
+ P9_DPRINTK(P9_DEBUG_ERROR,
+ "requested packet size too big: %d\n", n);
+ err = -EIO;
+ goto error;
+ }
+
+ tag = le16_to_cpu(*(__le16 *) (m->rbuf+5)); /* read tag */
+ P9_DPRINTK(P9_DEBUG_TRANS,
+ "mux %p pkt: size: %d bytes tag: %d\n", m, n, tag);
+ m->req = p9_tag_lookup(m->client, tag);
+ if (!m->req) {
+ P9_DPRINTK(P9_DEBUG_ERROR, "Unexpected packet tag %d\n",
+ tag);
+ err = -EIO;
+ goto error;
+ }
+
+ if (m->req->rc == NULL) {
+ m->req->rc = kmalloc(sizeof(struct p9_fcall) +
+ m->client->msize, GFP_KERNEL);
+ if (!m->req->rc) {
+ m->req = NULL;
+ err = -ENOMEM;
+ goto error;
+ }
+ }
+ m->rbuf = (char *)m->req->rc + sizeof(struct p9_fcall);
+ memcpy(m->rbuf, m->tmp_buf, m->rsize);
+ m->rsize = n;
+ }
+
+ /* not an else because some packets (like clunk) have no payload */
+ if ((m->req) && (m->rpos == m->rsize)) { /* packet is read in */
+ P9_DPRINTK(P9_DEBUG_TRANS, "got new packet\n");
+ spin_lock(&m->client->lock);
+ list_del(&m->req->req_list);
+ spin_unlock(&m->client->lock);
+ p9_client_cb(m->client, m->req);
+
+ m->rbuf = NULL;
+ m->rpos = 0;
+ m->rsize = 0;
+ m->req = NULL;
+ }
+
+ if (!list_empty(&m->req_list)) {
+ if (test_and_clear_bit(Rpending, &m->wsched))
+ n = POLLIN;
+ else
+ n = p9_fd_poll(m->client, NULL);
+
+ if (n & POLLIN) {
+ P9_DPRINTK(P9_DEBUG_TRANS, "sched read work %p\n", m);
+ queue_work(p9_mux_wq, &m->rq);
+ } else
+ clear_bit(Rworksched, &m->wsched);
+ } else
+ clear_bit(Rworksched, &m->wsched);
+
+ return;
error:
- if (csocket)
- sock_release(csocket);
+ p9_conn_cancel(m, err);
+ clear_bit(Rworksched, &m->wsched);
+}
+
+/**
+ * p9_fd_write - write to a socket
+ * @client: client instance
+ * @v: buffer to send data from
+ * @len: size of send buffer
+ *
+ */
+
+static int p9_fd_write(struct p9_client *client, void *v, int len)
+{
+ int ret;
+ mm_segment_t oldfs;
+ struct p9_trans_fd *ts = NULL;
- kfree(trans);
- return ERR_PTR(err);
+ if (client && client->status != Disconnected)
+ ts = client->trans;
+
+ if (!ts)
+ return -EREMOTEIO;
+
+ if (!(ts->wr->f_flags & O_NONBLOCK))
+ P9_DPRINTK(P9_DEBUG_ERROR, "blocking write ...\n");
+
+ oldfs = get_fs();
+ set_fs(get_ds());
+ /* The cast to a user pointer is valid due to the set_fs() */
+ ret = vfs_write(ts->wr, (void __user *)v, len, &ts->wr->f_pos);
+ set_fs(oldfs);
+
+ if (ret <= 0 && ret != -ERESTARTSYS && ret != -EAGAIN)
+ client->status = Disconnected;
+ return ret;
}
-EXPORT_SYMBOL(p9_trans_create_tcp);
-struct p9_transport *p9_trans_create_unix(const char *addr)
+/**
+ * p9_write_work - called when a transport can send some data
+ * @work: container for work to be done
+ *
+ */
+
+static void p9_write_work(struct work_struct *work)
{
- int err;
- struct socket *csocket;
- struct sockaddr_un sun_server;
- struct p9_transport *trans;
+ int n, err;
+ struct p9_conn *m;
+ struct p9_req_t *req;
- csocket = NULL;
- trans = kmalloc(sizeof(struct p9_transport), GFP_KERNEL);
- if (!trans)
- return ERR_PTR(-ENOMEM);
+ m = container_of(work, struct p9_conn, wq);
- trans->write = p9_fd_write;
- trans->read = p9_fd_read;
- trans->close = p9_fd_close;
- trans->poll = p9_fd_poll;
+ if (m->err < 0) {
+ clear_bit(Wworksched, &m->wsched);
+ return;
+ }
- if (strlen(addr) > UNIX_PATH_MAX) {
- P9_EPRINTK(KERN_ERR, "p9_trans_unix: address too long: %s\n",
- addr);
- err = -ENAMETOOLONG;
- goto error;
+ if (!m->wsize) {
+ if (list_empty(&m->unsent_req_list)) {
+ clear_bit(Wworksched, &m->wsched);
+ return;
+ }
+
+ spin_lock(&m->client->lock);
+ req = list_entry(m->unsent_req_list.next, struct p9_req_t,
+ req_list);
+ req->status = REQ_STATUS_SENT;
+ list_move_tail(&req->req_list, &m->req_list);
+
+ m->wbuf = req->tc->sdata;
+ m->wsize = req->tc->size;
+ m->wpos = 0;
+ spin_unlock(&m->client->lock);
}
- sun_server.sun_family = PF_UNIX;
- strcpy(sun_server.sun_path, addr);
- sock_create_kern(PF_UNIX, SOCK_STREAM, 0, &csocket);
- err = csocket->ops->connect(csocket, (struct sockaddr *)&sun_server,
- sizeof(struct sockaddr_un) - 1, 0);
- if (err < 0) {
- P9_EPRINTK(KERN_ERR,
- "p9_trans_unix: problem connecting socket: %s: %d\n",
- addr, err);
- goto error;
+ P9_DPRINTK(P9_DEBUG_TRANS, "mux %p pos %d size %d\n", m, m->wpos,
+ m->wsize);
+ clear_bit(Wpending, &m->wsched);
+ err = p9_fd_write(m->client, m->wbuf + m->wpos, m->wsize - m->wpos);
+ P9_DPRINTK(P9_DEBUG_TRANS, "mux %p sent %d bytes\n", m, err);
+ if (err == -EAGAIN) {
+ clear_bit(Wworksched, &m->wsched);
+ return;
}
- err = p9_socket_open(trans, csocket);
if (err < 0)
goto error;
+ else if (err == 0) {
+ err = -EREMOTEIO;
+ goto error;
+ }
+
+ m->wpos += err;
+ if (m->wpos == m->wsize)
+ m->wpos = m->wsize = 0;
+
+ if (m->wsize == 0 && !list_empty(&m->unsent_req_list)) {
+ if (test_and_clear_bit(Wpending, &m->wsched))
+ n = POLLOUT;
+ else
+ n = p9_fd_poll(m->client, NULL);
- return trans;
+ if (n & POLLOUT) {
+ P9_DPRINTK(P9_DEBUG_TRANS, "sched write work %p\n", m);
+ queue_work(p9_mux_wq, &m->wq);
+ } else
+ clear_bit(Wworksched, &m->wsched);
+ } else
+ clear_bit(Wworksched, &m->wsched);
+
+ return;
error:
- if (csocket)
- sock_release(csocket);
+ p9_conn_cancel(m, err);
+ clear_bit(Wworksched, &m->wsched);
+}
- kfree(trans);
- return ERR_PTR(err);
+static int p9_pollwake(wait_queue_t *wait, unsigned mode, int sync, void *key)
+{
+ struct p9_poll_wait *pwait =
+ container_of(wait, struct p9_poll_wait, wait);
+ struct p9_conn *m = pwait->conn;
+ unsigned long flags;
+ DECLARE_WAITQUEUE(dummy_wait, p9_poll_task);
+
+ spin_lock_irqsave(&p9_poll_lock, flags);
+ if (list_empty(&m->poll_pending_link))
+ list_add_tail(&m->poll_pending_link, &p9_poll_pending_list);
+ spin_unlock_irqrestore(&p9_poll_lock, flags);
+
+ /* perform the default wake up operation */
+ return default_wake_function(&dummy_wait, mode, sync, key);
}
-EXPORT_SYMBOL(p9_trans_create_unix);
-struct p9_transport *p9_trans_create_fd(int rfd, int wfd)
+/**
+ * p9_pollwait - add poll task to the wait queue
+ * @filp: file pointer being polled
+ * @wait_address: wait_q to block on
+ * @p: poll state
+ *
+ * called by files poll operation to add v9fs-poll task to files wait queue
+ */
+
+static void
+p9_pollwait(struct file *filp, wait_queue_head_t *wait_address, poll_table *p)
{
- int err;
- struct p9_transport *trans;
+ struct p9_conn *m = container_of(p, struct p9_conn, pt);
+ struct p9_poll_wait *pwait = NULL;
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(m->poll_wait); i++) {
+ if (m->poll_wait[i].wait_addr == NULL) {
+ pwait = &m->poll_wait[i];
+ break;
+ }
+ }
- if (rfd == ~0 || wfd == ~0) {
- printk(KERN_ERR "v9fs: Insufficient options for proto=fd\n");
- return ERR_PTR(-ENOPROTOOPT);
+ if (!pwait) {
+ P9_DPRINTK(P9_DEBUG_ERROR, "not enough wait_address slots\n");
+ return;
}
- trans = kmalloc(sizeof(struct p9_transport), GFP_KERNEL);
- if (!trans)
+ pwait->conn = m;
+ pwait->wait_addr = wait_address;
+ init_waitqueue_func_entry(&pwait->wait, p9_pollwake);
+ add_wait_queue(wait_address, &pwait->wait);
+}
+
+/**
+ * p9_conn_create - allocate and initialize the per-session mux data
+ * @client: client instance
+ *
+ * Note: Creates the polling task if this is the first session.
+ */
+
+static struct p9_conn *p9_conn_create(struct p9_client *client)
+{
+ int n;
+ struct p9_conn *m;
+
+ P9_DPRINTK(P9_DEBUG_TRANS, "client %p msize %d\n", client,
+ client->msize);
+ m = kzalloc(sizeof(struct p9_conn), GFP_KERNEL);
+ if (!m)
return ERR_PTR(-ENOMEM);
- trans->write = p9_fd_write;
- trans->read = p9_fd_read;
- trans->close = p9_fd_close;
- trans->poll = p9_fd_poll;
+ INIT_LIST_HEAD(&m->mux_list);
+ m->client = client;
- err = p9_fd_open(trans, rfd, wfd);
- if (err < 0)
- goto error;
+ INIT_LIST_HEAD(&m->req_list);
+ INIT_LIST_HEAD(&m->unsent_req_list);
+ INIT_WORK(&m->rq, p9_read_work);
+ INIT_WORK(&m->wq, p9_write_work);
+ INIT_LIST_HEAD(&m->poll_pending_link);
+ init_poll_funcptr(&m->pt, p9_pollwait);
- return trans;
+ n = p9_fd_poll(client, &m->pt);
+ if (n & POLLIN) {
+ P9_DPRINTK(P9_DEBUG_TRANS, "mux %p can read\n", m);
+ set_bit(Rpending, &m->wsched);
+ }
-error:
- kfree(trans);
- return ERR_PTR(err);
+ if (n & POLLOUT) {
+ P9_DPRINTK(P9_DEBUG_TRANS, "mux %p can write\n", m);
+ set_bit(Wpending, &m->wsched);
+ }
+
+ return m;
}
-EXPORT_SYMBOL(p9_trans_create_fd);
-static int p9_socket_open(struct p9_transport *trans, struct socket *csocket)
+/**
+ * p9_poll_mux - polls a mux and schedules read or write works if necessary
+ * @m: connection to poll
+ *
+ */
+
+static void p9_poll_mux(struct p9_conn *m)
{
- int fd, ret;
+ int n;
- csocket->sk->sk_allocation = GFP_NOIO;
- fd = sock_map_fd(csocket);
- if (fd < 0) {
- P9_EPRINTK(KERN_ERR, "p9_socket_open: failed to map fd\n");
- return fd;
+ if (m->err < 0)
+ return;
+
+ n = p9_fd_poll(m->client, NULL);
+ if (n < 0 || n & (POLLERR | POLLHUP | POLLNVAL)) {
+ P9_DPRINTK(P9_DEBUG_TRANS, "error mux %p err %d\n", m, n);
+ if (n >= 0)
+ n = -ECONNRESET;
+ p9_conn_cancel(m, n);
}
- ret = p9_fd_open(trans, fd, fd);
- if (ret < 0) {
- P9_EPRINTK(KERN_ERR, "p9_socket_open: failed to open fd\n");
- sockfd_put(csocket);
- return ret;
+ if (n & POLLIN) {
+ set_bit(Rpending, &m->wsched);
+ P9_DPRINTK(P9_DEBUG_TRANS, "mux %p can read\n", m);
+ if (!test_and_set_bit(Rworksched, &m->wsched)) {
+ P9_DPRINTK(P9_DEBUG_TRANS, "sched read work %p\n", m);
+ queue_work(p9_mux_wq, &m->rq);
+ }
}
- ((struct p9_trans_fd *)trans->priv)->rd->f_flags |= O_NONBLOCK;
+ if (n & POLLOUT) {
+ set_bit(Wpending, &m->wsched);
+ P9_DPRINTK(P9_DEBUG_TRANS, "mux %p can write\n", m);
+ if ((m->wsize || !list_empty(&m->unsent_req_list))
+ && !test_and_set_bit(Wworksched, &m->wsched)) {
+ P9_DPRINTK(P9_DEBUG_TRANS, "sched write work %p\n", m);
+ queue_work(p9_mux_wq, &m->wq);
+ }
+ }
+}
+
+/**
+ * p9_fd_request - send 9P request
+ * The function can sleep until the request is scheduled for sending.
+ * The function can be interrupted. Return from the function is not
+ * a guarantee that the request is sent successfully.
+ *
+ * @client: client instance
+ * @req: request to be sent
+ *
+ */
+
+static int p9_fd_request(struct p9_client *client, struct p9_req_t *req)
+{
+ int n;
+ struct p9_trans_fd *ts = client->trans;
+ struct p9_conn *m = ts->conn;
+
+ P9_DPRINTK(P9_DEBUG_TRANS, "mux %p task %p tcall %p id %d\n", m,
+ current, req->tc, req->tc->id);
+ if (m->err < 0)
+ return m->err;
+
+ spin_lock(&client->lock);
+ req->status = REQ_STATUS_UNSENT;
+ list_add_tail(&req->req_list, &m->unsent_req_list);
+ spin_unlock(&client->lock);
+
+ if (test_and_clear_bit(Wpending, &m->wsched))
+ n = POLLOUT;
+ else
+ n = p9_fd_poll(m->client, NULL);
+
+ if (n & POLLOUT && !test_and_set_bit(Wworksched, &m->wsched))
+ queue_work(p9_mux_wq, &m->wq);
return 0;
}
-static int p9_fd_open(struct p9_transport *trans, int rfd, int wfd)
+static int p9_fd_cancel(struct p9_client *client, struct p9_req_t *req)
+{
+ int ret = 1;
+
+ P9_DPRINTK(P9_DEBUG_TRANS, "client %p req %p\n", client, req);
+
+ spin_lock(&client->lock);
+ list_del(&req->req_list);
+
+ if (req->status == REQ_STATUS_UNSENT) {
+ req->status = REQ_STATUS_FLSHD;
+ ret = 0;
+ }
+
+ spin_unlock(&client->lock);
+
+ return ret;
+}
+
+/**
+ * parse_options - parse mount options into session structure
+ * @options: options string passed from mount
+ * @opts: transport-specific structure to parse options into
+ *
+ * Returns 0 upon success, -ERRNO upon failure
+ */
+
+static int parse_opts(char *params, struct p9_fd_opts *opts)
+{
+ char *p;
+ substring_t args[MAX_OPT_ARGS];
+ int option;
+ char *options;
+ int ret;
+
+ opts->port = P9_PORT;
+ opts->rfd = ~0;
+ opts->wfd = ~0;
+
+ if (!params)
+ return 0;
+
+ options = kstrdup(params, GFP_KERNEL);
+ if (!options) {
+ P9_DPRINTK(P9_DEBUG_ERROR,
+ "failed to allocate copy of option string\n");
+ return -ENOMEM;
+ }
+
+ while ((p = strsep(&options, ",")) != NULL) {
+ int token;
+ int r;
+ if (!*p)
+ continue;
+ token = match_token(p, tokens, args);
+ r = match_int(&args[0], &option);
+ if (r < 0) {
+ P9_DPRINTK(P9_DEBUG_ERROR,
+ "integer field, but no integer?\n");
+ ret = r;
+ continue;
+ }
+ switch (token) {
+ case Opt_port:
+ opts->port = option;
+ break;
+ case Opt_rfdno:
+ opts->rfd = option;
+ break;
+ case Opt_wfdno:
+ opts->wfd = option;
+ break;
+ default:
+ continue;
+ }
+ }
+ kfree(options);
+ return 0;
+}
+
+static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
{
struct p9_trans_fd *ts = kmalloc(sizeof(struct p9_trans_fd),
GFP_KERNEL);
return -EIO;
}
- trans->priv = ts;
- trans->status = Connected;
+ client->trans = ts;
+ client->status = Connected;
+
+ return 0;
+}
+
+static int p9_socket_open(struct p9_client *client, struct socket *csocket)
+{
+ int fd, ret;
+
+ csocket->sk->sk_allocation = GFP_NOIO;
+ fd = sock_map_fd(csocket, 0);
+ if (fd < 0) {
+ P9_EPRINTK(KERN_ERR, "p9_socket_open: failed to map fd\n");
+ return fd;
+ }
+
+ ret = p9_fd_open(client, fd, fd);
+ if (ret < 0) {
+ P9_EPRINTK(KERN_ERR, "p9_socket_open: failed to open fd\n");
+ sockfd_put(csocket);
+ return ret;
+ }
+
+ ((struct p9_trans_fd *)client->trans)->rd->f_flags |= O_NONBLOCK;
return 0;
}
/**
- * p9_fd_read- read from a fd
- * @v9ses: session information
- * @v: buffer to receive data into
- * @len: size of receive buffer
+ * p9_mux_destroy - cancels all pending requests and frees mux resources
+ * @m: mux to destroy
*
*/
-static int p9_fd_read(struct p9_transport *trans, void *v, int len)
-{
- int ret;
- struct p9_trans_fd *ts = NULL;
- if (trans && trans->status != Disconnected)
- ts = trans->priv;
+static void p9_conn_destroy(struct p9_conn *m)
+{
+ P9_DPRINTK(P9_DEBUG_TRANS, "mux %p prev %p next %p\n", m,
+ m->mux_list.prev, m->mux_list.next);
- if (!ts)
- return -EREMOTEIO;
+ p9_mux_poll_stop(m);
+ cancel_work_sync(&m->rq);
+ cancel_work_sync(&m->wq);
- if (!(ts->rd->f_flags & O_NONBLOCK))
- P9_DPRINTK(P9_DEBUG_ERROR, "blocking read ...\n");
+ p9_conn_cancel(m, -ECONNRESET);
- ret = kernel_read(ts->rd, ts->rd->f_pos, v, len);
- if (ret <= 0 && ret != -ERESTARTSYS && ret != -EAGAIN)
- trans->status = Disconnected;
- return ret;
+ m->client = NULL;
+ kfree(m);
}
/**
- * p9_fd_write - write to a socket
- * @v9ses: session information
- * @v: buffer to send data from
- * @len: size of send buffer
+ * p9_fd_close - shutdown file descriptor transport
+ * @client: client instance
*
*/
-static int p9_fd_write(struct p9_transport *trans, void *v, int len)
+
+static void p9_fd_close(struct p9_client *client)
{
- int ret;
- mm_segment_t oldfs;
- struct p9_trans_fd *ts = NULL;
+ struct p9_trans_fd *ts;
- if (trans && trans->status != Disconnected)
- ts = trans->priv;
+ if (!client)
+ return;
+ ts = client->trans;
if (!ts)
- return -EREMOTEIO;
+ return;
- if (!(ts->wr->f_flags & O_NONBLOCK))
- P9_DPRINTK(P9_DEBUG_ERROR, "blocking write ...\n");
+ client->status = Disconnected;
- oldfs = get_fs();
- set_fs(get_ds());
- /* The cast to a user pointer is valid due to the set_fs() */
- ret = vfs_write(ts->wr, (void __user *)v, len, &ts->wr->f_pos);
- set_fs(oldfs);
+ p9_conn_destroy(ts->conn);
- if (ret <= 0 && ret != -ERESTARTSYS && ret != -EAGAIN)
- trans->status = Disconnected;
- return ret;
+ if (ts->rd)
+ fput(ts->rd);
+ if (ts->wr)
+ fput(ts->wr);
+
+ kfree(ts);
}
-static unsigned int
-p9_fd_poll(struct p9_transport *trans, struct poll_table_struct *pt)
+/*
+ * stolen from NFS - maybe should be made a generic function?
+ */
+static inline int valid_ipaddr4(const char *buf)
{
- int ret, n;
- struct p9_trans_fd *ts = NULL;
- mm_segment_t oldfs;
+ int rc, count, in[4];
+
+ rc = sscanf(buf, "%d.%d.%d.%d", &in[0], &in[1], &in[2], &in[3]);
+ if (rc != 4)
+ return -EINVAL;
+ for (count = 0; count < 4; count++) {
+ if (in[count] > 255)
+ return -EINVAL;
+ }
+ return 0;
+}
- if (trans && trans->status == Connected)
- ts = trans->priv;
+static int
+p9_fd_create_tcp(struct p9_client *client, const char *addr, char *args)
+{
+ int err;
+ struct socket *csocket;
+ struct sockaddr_in sin_server;
+ struct p9_fd_opts opts;
+ struct p9_trans_fd *p = NULL; /* this gets allocated in p9_fd_open */
- if (!ts)
- return -EREMOTEIO;
+ err = parse_opts(args, &opts);
+ if (err < 0)
+ return err;
- if (!ts->rd->f_op || !ts->rd->f_op->poll)
- return -EIO;
+ if (valid_ipaddr4(addr) < 0)
+ return -EINVAL;
- if (!ts->wr->f_op || !ts->wr->f_op->poll)
- return -EIO;
+ csocket = NULL;
- oldfs = get_fs();
- set_fs(get_ds());
+ sin_server.sin_family = AF_INET;
+ sin_server.sin_addr.s_addr = in_aton(addr);
+ sin_server.sin_port = htons(opts.port);
+ sock_create_kern(PF_INET, SOCK_STREAM, IPPROTO_TCP, &csocket);
- ret = ts->rd->f_op->poll(ts->rd, pt);
- if (ret < 0)
- goto end;
+ if (!csocket) {
+ P9_EPRINTK(KERN_ERR, "p9_trans_tcp: problem creating socket\n");
+ err = -EIO;
+ goto error;
+ }
- if (ts->rd != ts->wr) {
- n = ts->wr->f_op->poll(ts->wr, pt);
- if (n < 0) {
- ret = n;
- goto end;
- }
- ret = (ret & ~POLLOUT) | (n & ~POLLIN);
+ err = csocket->ops->connect(csocket,
+ (struct sockaddr *)&sin_server,
+ sizeof(struct sockaddr_in), 0);
+ if (err < 0) {
+ P9_EPRINTK(KERN_ERR,
+ "p9_trans_tcp: problem connecting socket to %s\n",
+ addr);
+ goto error;
}
-end:
- set_fs(oldfs);
- return ret;
+ err = p9_socket_open(client, csocket);
+ if (err < 0)
+ goto error;
+
+ p = (struct p9_trans_fd *) client->trans;
+ p->conn = p9_conn_create(client);
+ if (IS_ERR(p->conn)) {
+ err = PTR_ERR(p->conn);
+ p->conn = NULL;
+ goto error;
+ }
+
+ return 0;
+
+error:
+ if (csocket)
+ sock_release(csocket);
+
+ kfree(p);
+
+ return err;
}
+static int
+p9_fd_create_unix(struct p9_client *client, const char *addr, char *args)
+{
+ int err;
+ struct socket *csocket;
+ struct sockaddr_un sun_server;
+ struct p9_trans_fd *p = NULL; /* this gets allocated in p9_fd_open */
+
+ csocket = NULL;
+
+ if (strlen(addr) > UNIX_PATH_MAX) {
+ P9_EPRINTK(KERN_ERR, "p9_trans_unix: address too long: %s\n",
+ addr);
+ err = -ENAMETOOLONG;
+ goto error;
+ }
+
+ sun_server.sun_family = PF_UNIX;
+ strcpy(sun_server.sun_path, addr);
+ sock_create_kern(PF_UNIX, SOCK_STREAM, 0, &csocket);
+ err = csocket->ops->connect(csocket, (struct sockaddr *)&sun_server,
+ sizeof(struct sockaddr_un) - 1, 0);
+ if (err < 0) {
+ P9_EPRINTK(KERN_ERR,
+ "p9_trans_unix: problem connecting socket: %s: %d\n",
+ addr, err);
+ goto error;
+ }
+
+ err = p9_socket_open(client, csocket);
+ if (err < 0)
+ goto error;
+
+ p = (struct p9_trans_fd *) client->trans;
+ p->conn = p9_conn_create(client);
+ if (IS_ERR(p->conn)) {
+ err = PTR_ERR(p->conn);
+ p->conn = NULL;
+ goto error;
+ }
+
+ return 0;
+
+error:
+ if (csocket)
+ sock_release(csocket);
+
+ kfree(p);
+ return err;
+}
+
+static int
+p9_fd_create(struct p9_client *client, const char *addr, char *args)
+{
+ int err;
+ struct p9_fd_opts opts;
+ struct p9_trans_fd *p = NULL; /* this get allocated in p9_fd_open */
+
+ parse_opts(args, &opts);
+
+ if (opts.rfd == ~0 || opts.wfd == ~0) {
+ printk(KERN_ERR "v9fs: Insufficient options for proto=fd\n");
+ return -ENOPROTOOPT;
+ }
+
+ err = p9_fd_open(client, opts.rfd, opts.wfd);
+ if (err < 0)
+ goto error;
+
+ p = (struct p9_trans_fd *) client->trans;
+ p->conn = p9_conn_create(client);
+ if (IS_ERR(p->conn)) {
+ err = PTR_ERR(p->conn);
+ p->conn = NULL;
+ goto error;
+ }
+
+ return 0;
+
+error:
+ kfree(p);
+ return err;
+}
+
+static struct p9_trans_module p9_tcp_trans = {
+ .name = "tcp",
+ .maxsize = MAX_SOCK_BUF,
+ .def = 1,
+ .create = p9_fd_create_tcp,
+ .close = p9_fd_close,
+ .request = p9_fd_request,
+ .cancel = p9_fd_cancel,
+ .owner = THIS_MODULE,
+};
+
+static struct p9_trans_module p9_unix_trans = {
+ .name = "unix",
+ .maxsize = MAX_SOCK_BUF,
+ .def = 0,
+ .create = p9_fd_create_unix,
+ .close = p9_fd_close,
+ .request = p9_fd_request,
+ .cancel = p9_fd_cancel,
+ .owner = THIS_MODULE,
+};
+
+static struct p9_trans_module p9_fd_trans = {
+ .name = "fd",
+ .maxsize = MAX_SOCK_BUF,
+ .def = 0,
+ .create = p9_fd_create,
+ .close = p9_fd_close,
+ .request = p9_fd_request,
+ .cancel = p9_fd_cancel,
+ .owner = THIS_MODULE,
+};
+
/**
- * p9_sock_close - shutdown socket
- * @trans: private socket structure
+ * p9_poll_proc - poll worker thread
+ * @a: thread state and arguments
+ *
+ * polls all v9fs transports for new events and queues the appropriate
+ * work to the work queue
*
*/
-static void p9_fd_close(struct p9_transport *trans)
+
+static int p9_poll_proc(void *a)
{
- struct p9_trans_fd *ts;
+ unsigned long flags;
- if (!trans)
- return;
+ P9_DPRINTK(P9_DEBUG_TRANS, "start %p\n", current);
+ repeat:
+ spin_lock_irqsave(&p9_poll_lock, flags);
+ while (!list_empty(&p9_poll_pending_list)) {
+ struct p9_conn *conn = list_first_entry(&p9_poll_pending_list,
+ struct p9_conn,
+ poll_pending_link);
+ list_del_init(&conn->poll_pending_link);
+ spin_unlock_irqrestore(&p9_poll_lock, flags);
- ts = xchg(&trans->priv, NULL);
+ p9_poll_mux(conn);
- if (!ts)
- return;
+ spin_lock_irqsave(&p9_poll_lock, flags);
+ }
+ spin_unlock_irqrestore(&p9_poll_lock, flags);
- trans->status = Disconnected;
- if (ts->rd)
- fput(ts->rd);
- if (ts->wr)
- fput(ts->wr);
- kfree(ts);
+ set_current_state(TASK_INTERRUPTIBLE);
+ if (list_empty(&p9_poll_pending_list)) {
+ P9_DPRINTK(P9_DEBUG_TRANS, "sleeping...\n");
+ schedule();
+ }
+ __set_current_state(TASK_RUNNING);
+
+ if (!kthread_should_stop())
+ goto repeat;
+
+ P9_DPRINTK(P9_DEBUG_TRANS, "finish\n");
+ return 0;
}
+int p9_trans_fd_init(void)
+{
+ p9_mux_wq = create_workqueue("v9fs");
+ if (!p9_mux_wq) {
+ printk(KERN_WARNING "v9fs: mux: creating workqueue failed\n");
+ return -ENOMEM;
+ }
+
+ p9_poll_task = kthread_run(p9_poll_proc, NULL, "v9fs-poll");
+ if (IS_ERR(p9_poll_task)) {
+ destroy_workqueue(p9_mux_wq);
+ printk(KERN_WARNING "v9fs: mux: creating poll task failed\n");
+ return PTR_ERR(p9_poll_task);
+ }
+
+ v9fs_register_trans(&p9_tcp_trans);
+ v9fs_register_trans(&p9_unix_trans);
+ v9fs_register_trans(&p9_fd_trans);
+
+ return 0;
+}
+
+void p9_trans_fd_exit(void)
+{
+ kthread_stop(p9_poll_task);
+ v9fs_unregister_trans(&p9_tcp_trans);
+ v9fs_unregister_trans(&p9_unix_trans);
+ v9fs_unregister_trans(&p9_fd_trans);
+
+ destroy_workqueue(p9_mux_wq);
+}