X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=drivers%2Fblock%2Fnbd.c;h=090796bef78f16374f8c94f2bf70e23e602deabd;hb=97bf2856c6014879bd04983a3e9dfcdac1e7fe85;hp=2f5cc3e251c605f34d72d95824caece8294c8037;hpb=6a77f38946aaee1cd85eeec6cf4229b204c15071;p=linux-2.6.git diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c index 2f5cc3e25..090796bef 100644 --- a/drivers/block/nbd.c +++ b/drivers/block/nbd.c @@ -7,39 +7,9 @@ * Copyright 1997-2000 Pavel Machek * Parts copyright 2001 Steven Whitehouse * - * (part of code stolen from loop.c) + * This file is released under GPLv2 or later. * - * 97-3-25 compiled 0-th version, not yet tested it - * (it did not work, BTW) (later that day) HEY! it works! - * (bit later) hmm, not that much... 2:00am next day: - * yes, it works, but it gives something like 50kB/sec - * 97-4-01 complete rewrite to make it possible for many requests at - * once to be processed - * 97-4-11 Making protocol independent of endianity etc. - * 97-9-13 Cosmetic changes - * 98-5-13 Attempt to make 64-bit-clean on 64-bit machines - * 99-1-11 Attempt to make 64-bit-clean on 32-bit machines - * 01-2-27 Fix to store proper blockcount for kernel (calculated using - * BLOCK_SIZE_BITS, not device blocksize) - * 01-3-11 Make nbd work with new Linux block layer code. It now supports - * plugging like all the other block devices. Also added in MSG_MORE to - * reduce number of partial TCP segments sent. - * 01-12-6 Fix deadlock condition by making queue locks independent of - * the transmit lock. - * 02-10-11 Allow hung xmit to be aborted via SIGKILL & various fixes. - * - * 03-06-22 Make nbd work with new linux 2.5 block layer design. This fixes - * memory corruption from module removal and possible memory corruption - * from sending/receiving disk data. - * 03-06-23 Cosmetic changes. - * 03-06-23 Enhance diagnostics support. - * 03-06-24 Remove unneeded blksize_bits field from nbd_device struct. - * - * 03-06-24 Cleanup PARANOIA usage & code. - * 04-02-19 Remove PARANOIA, plus various cleanups (Paul Clements) - * possible FIXME: make set_sock / set_blksize / set_size / do_it one syscall - * why not: would need verify_area and friends, would share yet another - * structure with userland + * (part of code stolen from loop.c) */ #include @@ -54,11 +24,13 @@ #include #include #include +#include +#include +#include #include -#include - #include +#include #include #include @@ -80,6 +52,7 @@ static unsigned int debugflags; #endif /* NDEBUG */ +static unsigned int nbds_max = 16; static struct nbd_device nbd_dev[MAX_NBD]; /* @@ -135,7 +108,7 @@ static void nbd_end_request(struct request *req) spin_lock_irqsave(q->queue_lock, flags); if (!end_that_request_first(req, uptodate, req->nr_sectors)) { - end_that_request_last(req); + end_that_request_last(req, uptodate); } spin_unlock_irqrestore(q->queue_lock, flags); } @@ -169,7 +142,6 @@ static int sock_xmit(struct socket *sock, int send, void *buf, int size, msg.msg_namelen = 0; msg.msg_control = NULL; msg.msg_controllen = 0; - msg.msg_namelen = 0; msg.msg_flags = msg_flags | MSG_NOSIGNAL; if (send) @@ -229,14 +201,6 @@ static int nbd_send_req(struct nbd_device *lo, struct request *req) request.len = htonl(size); memcpy(request.handle, &req, sizeof(req)); - down(&lo->tx_lock); - - if (!sock || !lo->sock) { - printk(KERN_ERR "%s: Attempted send on closed socket\n", - lo->disk->disk_name); - goto error_out; - } - dprintk(DBG_TX, "%s: request %p: sending control (%s@%llu,%luB)\n", lo->disk->disk_name, req, nbdcmd_to_ascii(nbd_cmd(req)), @@ -275,11 +239,9 @@ static int nbd_send_req(struct nbd_device *lo, struct request *req) } } } - up(&lo->tx_lock); return 0; error_out: - up(&lo->tx_lock); return 1; } @@ -288,9 +250,14 @@ static struct request *nbd_find_request(struct nbd_device *lo, char *handle) struct request *req; struct list_head *tmp; struct request *xreq; + int err; memcpy(&xreq, handle, sizeof(xreq)); + err = wait_event_interruptible(lo->active_wq, lo->active_req != xreq); + if (unlikely(err)) + goto out; + spin_lock(&lo->queue_lock); list_for_each(tmp, &lo->queue_head) { req = list_entry(tmp, struct request, queuelist); @@ -301,7 +268,11 @@ static struct request *nbd_find_request(struct nbd_device *lo, char *handle) return req; } spin_unlock(&lo->queue_lock); - return NULL; + + err = -ENOENT; + +out: + return ERR_PTR(err); } static inline int sock_recv_bvec(struct socket *sock, struct bio_vec *bvec) @@ -315,7 +286,7 @@ static inline int sock_recv_bvec(struct socket *sock, struct bio_vec *bvec) } /* NULL returned = something went wrong, inform userspace */ -struct request *nbd_read_stat(struct nbd_device *lo) +static struct request *nbd_read_stat(struct nbd_device *lo) { int result; struct nbd_reply reply; @@ -329,13 +300,6 @@ struct request *nbd_read_stat(struct nbd_device *lo) lo->disk->disk_name, result); goto harderror; } - req = nbd_find_request(lo, reply.handle); - if (req == NULL) { - printk(KERN_ERR "%s: Unexpected reply (%p)\n", - lo->disk->disk_name, reply.handle); - result = -EBADR; - goto harderror; - } if (ntohl(reply.magic) != NBD_REPLY_MAGIC) { printk(KERN_ERR "%s: Wrong magic (0x%lx)\n", @@ -344,6 +308,19 @@ struct request *nbd_read_stat(struct nbd_device *lo) result = -EPROTO; goto harderror; } + + req = nbd_find_request(lo, reply.handle); + if (unlikely(IS_ERR(req))) { + result = PTR_ERR(req); + if (result != -ENOENT) + goto harderror; + + printk(KERN_ERR "%s: Unexpected reply (%p)\n", + lo->disk->disk_name, reply.handle); + result = -EBADR; + goto harderror; + } + if (ntohl(reply.error)) { printk(KERN_ERR "%s: Other side returned error (%d)\n", lo->disk->disk_name, ntohl(reply.error)); @@ -364,7 +341,8 @@ struct request *nbd_read_stat(struct nbd_device *lo) printk(KERN_ERR "%s: Receive data failed (result %d)\n", lo->disk->disk_name, result); - goto harderror; + req->errors++; + return req; } dprintk(DBG_RX, "%s: request %p: got %d bytes data\n", lo->disk->disk_name, req, bvec->bv_len); @@ -377,36 +355,57 @@ harderror: return NULL; } -void nbd_do_it(struct nbd_device *lo) +static ssize_t pid_show(struct gendisk *disk, char *page) +{ + return sprintf(page, "%ld\n", + (long) ((struct nbd_device *)disk->private_data)->pid); +} + +static struct disk_attribute pid_attr = { + .attr = { .name = "pid", .mode = S_IRUGO }, + .show = pid_show, +}; + +static void nbd_do_it(struct nbd_device *lo) { struct request *req; BUG_ON(lo->magic != LO_MAGIC); + lo->pid = current->pid; + sysfs_create_file(&lo->disk->kobj, &pid_attr.attr); + while ((req = nbd_read_stat(lo)) != NULL) nbd_end_request(req); + + sysfs_remove_file(&lo->disk->kobj, &pid_attr.attr); return; } -void nbd_clear_que(struct nbd_device *lo) +static void nbd_clear_que(struct nbd_device *lo) { struct request *req; BUG_ON(lo->magic != LO_MAGIC); - do { - req = NULL; - spin_lock(&lo->queue_lock); - if (!list_empty(&lo->queue_head)) { - req = list_entry(lo->queue_head.next, struct request, queuelist); - list_del_init(&req->queuelist); - } - spin_unlock(&lo->queue_lock); - if (req) { - req->errors++; - nbd_end_request(req); - } - } while (req); + /* + * Because we have set lo->sock to NULL under the tx_lock, all + * modifications to the list must have completed by now. For + * the same reason, the active_req must be NULL. + * + * As a consequence, we don't need to take the spin lock while + * purging the list here. + */ + BUG_ON(lo->sock); + BUG_ON(lo->active_req); + + while (!list_empty(&lo->queue_head)) { + req = list_entry(lo->queue_head.next, struct request, + queuelist); + list_del_init(&req->queuelist); + req->errors++; + nbd_end_request(req); + } } /* @@ -424,21 +423,16 @@ static void do_nbd_request(request_queue_t * q) struct nbd_device *lo; blkdev_dequeue_request(req); - dprintk(DBG_BLKDEV, "%s: request %p: dequeued (flags=%lx)\n", - req->rq_disk->disk_name, req, req->flags); + dprintk(DBG_BLKDEV, "%s: request %p: dequeued (flags=%x)\n", + req->rq_disk->disk_name, req, req->cmd_type); - if (!(req->flags & REQ_CMD)) + if (!blk_fs_request(req)) goto error_out; lo = req->rq_disk->private_data; BUG_ON(lo->magic != LO_MAGIC); - if (!lo->file) { - printk(KERN_ERR "%s: Request when not-ready\n", - lo->disk->disk_name); - goto error_out; - } nbd_cmd(req) = NBD_CMD_READ; if (rq_data_dir(req) == WRITE) { nbd_cmd(req) = NBD_CMD_WRITE; @@ -452,32 +446,34 @@ static void do_nbd_request(request_queue_t * q) req->errors = 0; spin_unlock_irq(q->queue_lock); - spin_lock(&lo->queue_lock); - - if (!lo->file) { - spin_unlock(&lo->queue_lock); - printk(KERN_ERR "%s: failed between accept and semaphore, file lost\n", - lo->disk->disk_name); + mutex_lock(&lo->tx_lock); + if (unlikely(!lo->sock)) { + mutex_unlock(&lo->tx_lock); + printk(KERN_ERR "%s: Attempted send on closed socket\n", + lo->disk->disk_name); req->errors++; nbd_end_request(req); spin_lock_irq(q->queue_lock); continue; } - list_add(&req->queuelist, &lo->queue_head); - spin_unlock(&lo->queue_lock); + lo->active_req = req; if (nbd_send_req(lo, req) != 0) { printk(KERN_ERR "%s: Request send failed\n", lo->disk->disk_name); - if (nbd_find_request(lo, (char *)&req) != NULL) { - /* we still own req */ - req->errors++; - nbd_end_request(req); - } else /* we're racing with nbd_clear_que */ - printk(KERN_DEBUG "nbd: can't find req\n"); + req->errors++; + nbd_end_request(req); + } else { + spin_lock(&lo->queue_lock); + list_add(&req->queuelist, &lo->queue_head); + spin_unlock(&lo->queue_lock); } + lo->active_req = NULL; + mutex_unlock(&lo->tx_lock); + wake_up_all(&lo->active_wq); + spin_lock_irq(q->queue_lock); continue; @@ -509,7 +505,7 @@ static int nbd_ioctl(struct inode *inode, struct file *file, switch (cmd) { case NBD_DISCONNECT: printk(KERN_INFO "%s: NBD_DISCONNECT\n", lo->disk->disk_name); - sreq.flags = REQ_SPECIAL; + sreq.cmd_type = REQ_TYPE_SPECIAL; nbd_cmd(&sreq) = NBD_CMD_DISC; /* * Set these to sane values in case server implementation @@ -525,20 +521,13 @@ static int nbd_ioctl(struct inode *inode, struct file *file, case NBD_CLEAR_SOCK: error = 0; - down(&lo->tx_lock); + mutex_lock(&lo->tx_lock); lo->sock = NULL; - up(&lo->tx_lock); - spin_lock(&lo->queue_lock); + mutex_unlock(&lo->tx_lock); file = lo->file; lo->file = NULL; - spin_unlock(&lo->queue_lock); nbd_clear_que(lo); - spin_lock(&lo->queue_lock); - if (!list_empty(&lo->queue_head)) { - printk(KERN_ERR "nbd: disconnect: some requests are in progress -> please try again.\n"); - error = -EBUSY; - } - spin_unlock(&lo->queue_lock); + BUG_ON(!list_empty(&lo->queue_head)); if (file) fput(file); return error; @@ -548,8 +537,8 @@ static int nbd_ioctl(struct inode *inode, struct file *file, error = -EINVAL; file = fget(arg); if (file) { - inode = file->f_dentry->d_inode; - if (inode->i_sock) { + inode = file->f_path.dentry->d_inode; + if (S_ISSOCK(inode->i_mode)) { lo->file = file; lo->sock = SOCKET_I(inode); error = 0; @@ -588,7 +577,7 @@ static int nbd_ioctl(struct inode *inode, struct file *file, * FIXME: This code is duplicated from sys_shutdown, but * there should be a more generic interface rather than * calling socket ops directly here */ - down(&lo->tx_lock); + mutex_lock(&lo->tx_lock); if (lo->sock) { printk(KERN_WARNING "%s: shutting down socket\n", lo->disk->disk_name); @@ -596,25 +585,20 @@ static int nbd_ioctl(struct inode *inode, struct file *file, SEND_SHUTDOWN|RCV_SHUTDOWN); lo->sock = NULL; } - up(&lo->tx_lock); - spin_lock(&lo->queue_lock); + mutex_unlock(&lo->tx_lock); file = lo->file; lo->file = NULL; - spin_unlock(&lo->queue_lock); nbd_clear_que(lo); printk(KERN_WARNING "%s: queue cleared\n", lo->disk->disk_name); if (file) fput(file); return lo->harderror; case NBD_CLEAR_QUE: - down(&lo->tx_lock); - if (lo->sock) { - up(&lo->tx_lock); - return 0; /* probably should be error, but that would - * break "nbd-client -d", so just return 0 */ - } - up(&lo->tx_lock); - nbd_clear_que(lo); + /* + * This is for compatibility only. The queue is always cleared + * by NBD_DO_IT or NBD_CLEAR_SOCK. + */ + BUG_ON(!lo->sock && !list_empty(&lo->queue_head)); return 0; case NBD_PRINT_DEBUG: printk(KERN_INFO "%s: next = %p, prev = %p, head = %p\n", @@ -642,12 +626,15 @@ static int __init nbd_init(void) int err = -ENOMEM; int i; - if (sizeof(struct nbd_request) != 28) { - printk(KERN_CRIT "nbd: sizeof nbd_request needs to be 28 in order to work!\n" ); - return -EIO; + BUILD_BUG_ON(sizeof(struct nbd_request) != 28); + + if (nbds_max > MAX_NBD) { + printk(KERN_CRIT "nbd: cannot allocate more than %u nbds; %u requested.\n", MAX_NBD, + nbds_max); + return -EINVAL; } - for (i = 0; i < MAX_NBD; i++) { + for (i = 0; i < nbds_max; i++) { struct gendisk *disk = alloc_disk(1); if (!disk) goto out; @@ -672,15 +659,15 @@ static int __init nbd_init(void) printk(KERN_INFO "nbd: registered device at major %d\n", NBD_MAJOR); dprintk(DBG_INIT, "nbd: debugflags=0x%x\n", debugflags); - devfs_mk_dir("nbd"); - for (i = 0; i < MAX_NBD; i++) { + for (i = 0; i < nbds_max; i++) { struct gendisk *disk = nbd_dev[i].disk; nbd_dev[i].file = NULL; nbd_dev[i].magic = LO_MAGIC; nbd_dev[i].flags = 0; spin_lock_init(&nbd_dev[i].queue_lock); INIT_LIST_HEAD(&nbd_dev[i].queue_head); - init_MUTEX(&nbd_dev[i].tx_lock); + mutex_init(&nbd_dev[i].tx_lock); + init_waitqueue_head(&nbd_dev[i].active_wq); nbd_dev[i].blksize = 1024; nbd_dev[i].bytesize = 0x7ffffc00ULL << 10; /* 2TB */ disk->major = NBD_MAJOR; @@ -689,7 +676,6 @@ static int __init nbd_init(void) disk->private_data = &nbd_dev[i]; disk->flags |= GENHD_FL_SUPPRESS_PARTITION_INFO; sprintf(disk->disk_name, "nbd%d", i); - sprintf(disk->devfs_name, "nbd/%d", i); set_capacity(disk, 0x7ffffc00ULL << 1); /* 2 TB */ add_disk(disk); } @@ -706,15 +692,15 @@ out: static void __exit nbd_cleanup(void) { int i; - for (i = 0; i < MAX_NBD; i++) { + for (i = 0; i < nbds_max; i++) { struct gendisk *disk = nbd_dev[i].disk; + nbd_dev[i].magic = 0; if (disk) { del_gendisk(disk); blk_cleanup_queue(disk->queue); put_disk(disk); } } - devfs_remove("nbd"); unregister_blkdev(NBD_MAJOR, "nbd"); printk(KERN_INFO "nbd: unregistered device at major %d\n", NBD_MAJOR); } @@ -725,6 +711,8 @@ module_exit(nbd_cleanup); MODULE_DESCRIPTION("Network Block Device"); MODULE_LICENSE("GPL"); +module_param(nbds_max, int, 0444); +MODULE_PARM_DESC(nbds_max, "How many network block devices to initialize."); #ifndef NDEBUG module_param(debugflags, int, 0644); MODULE_PARM_DESC(debugflags, "flags for controlling debug output");