linux/drivers/block/nbd.c

1074 lines
24 KiB
C
Raw Normal View History

/*
* Network block device - make block devices work over TCP
*
* Note that you can not swap over this thing, yet. Seems to work but
* deadlocks sometimes - you can not swap over TCP in general.
*
* Copyright 1997-2000, 2008 Pavel Machek <pavel@ucw.cz>
* Parts copyright 2001 Steven Whitehouse <steve@chygwyn.com>
*
* This file is released under GPLv2 or later.
*
* (part of code stolen from loop.c)
*/
#include <linux/major.h>
#include <linux/blkdev.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/sched.h>
#include <linux/fs.h>
#include <linux/bio.h>
#include <linux/stat.h>
#include <linux/errno.h>
#include <linux/file.h>
#include <linux/ioctl.h>
#include <linux/mutex.h>
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
#include <linux/compiler.h>
#include <linux/err.h>
#include <linux/kernel.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 11:04:11 +03:00
#include <linux/slab.h>
#include <net/sock.h>
#include <linux/net.h>
#include <linux/kthread.h>
#include <linux/types.h>
#include <linux/debugfs.h>
#include <asm/uaccess.h>
#include <asm/types.h>
#include <linux/nbd.h>
struct nbd_device {
u32 flags;
struct socket * sock; /* If == NULL, device is not ready, yet */
int magic;
spinlock_t queue_lock;
struct list_head queue_head; /* Requests waiting result */
struct request *active_req;
wait_queue_head_t active_wq;
struct list_head waiting_queue; /* Requests to be sent */
wait_queue_head_t waiting_wq;
struct mutex tx_lock;
struct gendisk *disk;
int blksize;
loff_t bytesize;
int xmit_timeout;
bool disconnect; /* a disconnect has been requested by user */
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
struct timer_list timeout_timer;
/* protects initialization and shutdown of the socket */
spinlock_t sock_lock;
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
struct task_struct *task_recv;
struct task_struct *task_send;
#if IS_ENABLED(CONFIG_DEBUG_FS)
struct dentry *dbg_dir;
#endif
};
#if IS_ENABLED(CONFIG_DEBUG_FS)
static struct dentry *nbd_dbg_dir;
#endif
#define nbd_name(nbd) ((nbd)->disk->disk_name)
#define NBD_MAGIC 0x68797548
static unsigned int nbds_max = 16;
static struct nbd_device *nbd_dev;
NBD: add partition support Permit the use of partitions with network block devices (NBD). A new parameter is introduced to define how many partition we want to be able to manage per network block device. This parameter is "max_part". For instance, to manage 63 partitions / loop device, we will do: [on the server side] # nbd-server 1234 /dev/sdb [on the client side] # modprobe nbd max_part=63 # ls -l /dev/nbd* brw-rw---- 1 root disk 43, 0 2008-03-25 11:14 /dev/nbd0 brw-rw---- 1 root disk 43, 64 2008-03-25 11:11 /dev/nbd1 brw-rw---- 1 root disk 43, 640 2008-03-25 11:11 /dev/nbd10 brw-rw---- 1 root disk 43, 704 2008-03-25 11:11 /dev/nbd11 brw-rw---- 1 root disk 43, 768 2008-03-25 11:11 /dev/nbd12 brw-rw---- 1 root disk 43, 832 2008-03-25 11:11 /dev/nbd13 brw-rw---- 1 root disk 43, 896 2008-03-25 11:11 /dev/nbd14 brw-rw---- 1 root disk 43, 960 2008-03-25 11:11 /dev/nbd15 brw-rw---- 1 root disk 43, 128 2008-03-25 11:11 /dev/nbd2 brw-rw---- 1 root disk 43, 192 2008-03-25 11:11 /dev/nbd3 brw-rw---- 1 root disk 43, 256 2008-03-25 11:11 /dev/nbd4 brw-rw---- 1 root disk 43, 320 2008-03-25 11:11 /dev/nbd5 brw-rw---- 1 root disk 43, 384 2008-03-25 11:11 /dev/nbd6 brw-rw---- 1 root disk 43, 448 2008-03-25 11:11 /dev/nbd7 brw-rw---- 1 root disk 43, 512 2008-03-25 11:11 /dev/nbd8 brw-rw---- 1 root disk 43, 576 2008-03-25 11:11 /dev/nbd9 # nbd-client localhost 1234 /dev/nbd0 Negotiation: ..size = 80418240KB bs=1024, sz=80418240 -------NOTE, RFC: partition table is not automatically read. The driver sets bdev->bd_invalidated to 1 to force the read of the partition table of the device, but this is done only on an open of the device. So we have to do a "touch /dev/nbdX" or something like that. It can't be done from the nbd-client or nbd driver because at this level we can't ask to read the partition table and to serve the request at the same time (-> deadlock) If someone has a better idea, I'm open to any suggestion. -------NOTE, RFC # fdisk -l /dev/nbd0 Disk /dev/nbd0: 82.3 GB, 82348277760 bytes 255 heads, 63 sectors/track, 10011 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/nbd0p1 * 1 9965 80043831 83 Linux /dev/nbd0p2 9966 10011 369495 5 Extended /dev/nbd0p5 9966 10011 369463+ 82 Linux swap / Solaris # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 brw-rw---- 1 root disk 43, 1 2008-03-25 11:16 /dev/nbd0p1 brw-rw---- 1 root disk 43, 2 2008-03-25 11:16 /dev/nbd0p2 brw-rw---- 1 root disk 43, 5 2008-03-25 11:16 /dev/nbd0p5 # mount /dev/nbd0p1 /mnt # ls /mnt bin dev initrd lost+found opt sbin sys var boot etc initrd.img media proc selinux tmp vmlinuz cdrom home lib mnt root srv usr # umount /mnt # nbd-client -d /dev/nbd0 # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 -------NOTE On "nbd-client -d", we can do an iocl(BLKRRPART) to update partition table: as the size of the device is 0, we don't have to serve the partition manager request (-> no deadlock). -------NOTE Signed-off-by: Paul Clements <paul.clements@steeleye.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-29 12:02:51 +04:00
static int max_part;
/*
* Use just one lock (or at most 1 per NIC). Two arguments for this:
* 1. Each NIC is essentially a synchronization point for all servers
* accessed through that NIC so there's no need to have more locks
* than NICs anyway.
* 2. More locks lead to more "Dirty cache line bouncing" which will slow
* down each lock to the point where they're actually slower than just
* a single lock.
* Thanks go to Jens Axboe and Al Viro for their LKML emails explaining this!
*/
static DEFINE_SPINLOCK(nbd_lock);
static inline struct device *nbd_to_dev(struct nbd_device *nbd)
{
return disk_to_dev(nbd->disk);
}
static const char *nbdcmd_to_ascii(int cmd)
{
switch (cmd) {
case NBD_CMD_READ: return "read";
case NBD_CMD_WRITE: return "write";
case NBD_CMD_DISC: return "disconnect";
nbd: support FLUSH requests Currently, the NBD device does not accept flush requests from the Linux block layer. If the NBD server opened the target with neither O_SYNC nor O_DSYNC, however, the device will be effectively backed by a writeback cache. Without issuing flushes properly, operation of the NBD device will not be safe against power losses. The NBD protocol has support for both a cache flush command and a FUA command flag; the server will also pass a flag to note its support for these features. This patch adds support for the cache flush command and flag. In the kernel, we receive the flags via the NBD_SET_FLAGS ioctl, and map NBD_FLAG_SEND_FLUSH to the argument of blk_queue_flush. When the flag is active the block layer will send REQ_FLUSH requests, which we translate to NBD_CMD_FLUSH commands. FUA support is not included in this patch because all free software servers implement it with a full fdatasync; thus it has no advantage over supporting flush only. Because I [Paolo] cannot really benchmark it in a realistic scenario, I cannot tell if it is a good idea or not. It is also not clear if it is valid for an NBD server to support FUA but not flush. The Linux block layer gives a warning for this combination, the NBD protocol documentation says nothing about it. The patch also fixes a small problem in the handling of flags: nbd->flags must be cleared at the end of NBD_DO_IT, but the driver was not doing that. The bug manifests itself as follows. Suppose you two different client/server pairs to start the NBD device. Suppose also that the first client supports NBD_SET_FLAGS, and the first server sends NBD_FLAG_SEND_FLUSH; the second pair instead does neither of these two things. Before this patch, the second invocation of NBD_DO_IT will use a stale value of nbd->flags, and the second server will issue an error every time it receives an NBD_CMD_FLUSH command. This bug is pre-existing, but it becomes much more important after this patch; flush failures make the device pretty much unusable, unlike Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Alex Bligh <alex@alex.org.uk> Acked-by: Paul Clements <Paul.Clements@steeleye.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 05:05:23 +04:00
case NBD_CMD_FLUSH: return "flush";
case NBD_CMD_TRIM: return "trim/discard";
}
return "invalid";
}
static void nbd_end_request(struct nbd_device *nbd, struct request *req)
{
int error = req->errors ? -EIO : 0;
struct request_queue *q = req->q;
unsigned long flags;
dev_dbg(nbd_to_dev(nbd), "request %p: %s\n", req,
error ? "failed" : "done");
spin_lock_irqsave(q->queue_lock, flags);
__blk_end_request_all(req, error);
spin_unlock_irqrestore(q->queue_lock, flags);
}
/*
* Forcibly shutdown the socket causing all listeners to error
*/
static void sock_shutdown(struct nbd_device *nbd)
{
spin_lock_irq(&nbd->sock_lock);
if (!nbd->sock) {
spin_unlock_irq(&nbd->sock_lock);
return;
}
dev_warn(disk_to_dev(nbd->disk), "shutting down socket\n");
kernel_sock_shutdown(nbd->sock, SHUT_RDWR);
sockfd_put(nbd->sock);
nbd->sock = NULL;
spin_unlock_irq(&nbd->sock_lock);
del_timer(&nbd->timeout_timer);
}
static void nbd_xmit_timeout(unsigned long arg)
{
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
struct nbd_device *nbd = (struct nbd_device *)arg;
unsigned long flags;
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
if (list_empty(&nbd->queue_head))
return;
nbd->disconnect = true;
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
spin_lock_irqsave(&nbd->sock_lock, flags);
if (nbd->sock)
kernel_sock_shutdown(nbd->sock, SHUT_RDWR);
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
spin_unlock_irqrestore(&nbd->sock_lock, flags);
dev_err(nbd_to_dev(nbd), "Connection timed out, shutting down connection\n");
}
/*
* Send or receive packet.
*/
static int sock_xmit(struct nbd_device *nbd, int send, void *buf, int size,
int msg_flags)
{
struct socket *sock = nbd->sock;
int result;
struct msghdr msg;
struct kvec iov;
unsigned long pflags = current->flags;
if (unlikely(!sock)) {
dev_err(disk_to_dev(nbd->disk),
"Attempted %s on closed socket in sock_xmit\n",
(send ? "send" : "recv"));
return -EINVAL;
}
current->flags |= PF_MEMALLOC;
do {
sock->sk->sk_allocation = GFP_NOIO | __GFP_MEMALLOC;
iov.iov_base = buf;
iov.iov_len = size;
msg.msg_name = NULL;
msg.msg_namelen = 0;
msg.msg_control = NULL;
msg.msg_controllen = 0;
msg.msg_flags = msg_flags | MSG_NOSIGNAL;
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
if (send)
result = kernel_sendmsg(sock, &msg, &iov, 1, size);
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
else
result = kernel_recvmsg(sock, &msg, &iov, 1, size,
msg.msg_flags);
if (result <= 0) {
if (result == 0)
result = -EPIPE; /* short read */
break;
}
size -= result;
buf += result;
} while (size > 0);
tsk_restore_flags(current, pflags, PF_MEMALLOC);
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
if (!send && nbd->xmit_timeout)
mod_timer(&nbd->timeout_timer, jiffies + nbd->xmit_timeout);
return result;
}
static inline int sock_send_bvec(struct nbd_device *nbd, struct bio_vec *bvec,
int flags)
{
int result;
void *kaddr = kmap(bvec->bv_page);
result = sock_xmit(nbd, 1, kaddr + bvec->bv_offset,
bvec->bv_len, flags);
kunmap(bvec->bv_page);
return result;
}
/* always call with the tx_lock held */
static int nbd_send_req(struct nbd_device *nbd, struct request *req)
{
int result, flags;
struct nbd_request request;
unsigned long size = blk_rq_bytes(req);
u32 type;
if (req->cmd_type == REQ_TYPE_DRV_PRIV)
type = NBD_CMD_DISC;
else if (req->cmd_flags & REQ_DISCARD)
type = NBD_CMD_TRIM;
else if (req->cmd_flags & REQ_FLUSH)
type = NBD_CMD_FLUSH;
else if (rq_data_dir(req) == WRITE)
type = NBD_CMD_WRITE;
else
type = NBD_CMD_READ;
memset(&request, 0, sizeof(request));
request.magic = htonl(NBD_REQUEST_MAGIC);
request.type = htonl(type);
if (type != NBD_CMD_FLUSH && type != NBD_CMD_DISC) {
nbd: support FLUSH requests Currently, the NBD device does not accept flush requests from the Linux block layer. If the NBD server opened the target with neither O_SYNC nor O_DSYNC, however, the device will be effectively backed by a writeback cache. Without issuing flushes properly, operation of the NBD device will not be safe against power losses. The NBD protocol has support for both a cache flush command and a FUA command flag; the server will also pass a flag to note its support for these features. This patch adds support for the cache flush command and flag. In the kernel, we receive the flags via the NBD_SET_FLAGS ioctl, and map NBD_FLAG_SEND_FLUSH to the argument of blk_queue_flush. When the flag is active the block layer will send REQ_FLUSH requests, which we translate to NBD_CMD_FLUSH commands. FUA support is not included in this patch because all free software servers implement it with a full fdatasync; thus it has no advantage over supporting flush only. Because I [Paolo] cannot really benchmark it in a realistic scenario, I cannot tell if it is a good idea or not. It is also not clear if it is valid for an NBD server to support FUA but not flush. The Linux block layer gives a warning for this combination, the NBD protocol documentation says nothing about it. The patch also fixes a small problem in the handling of flags: nbd->flags must be cleared at the end of NBD_DO_IT, but the driver was not doing that. The bug manifests itself as follows. Suppose you two different client/server pairs to start the NBD device. Suppose also that the first client supports NBD_SET_FLAGS, and the first server sends NBD_FLAG_SEND_FLUSH; the second pair instead does neither of these two things. Before this patch, the second invocation of NBD_DO_IT will use a stale value of nbd->flags, and the second server will issue an error every time it receives an NBD_CMD_FLUSH command. This bug is pre-existing, but it becomes much more important after this patch; flush failures make the device pretty much unusable, unlike Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Alex Bligh <alex@alex.org.uk> Acked-by: Paul Clements <Paul.Clements@steeleye.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 05:05:23 +04:00
request.from = cpu_to_be64((u64)blk_rq_pos(req) << 9);
request.len = htonl(size);
}
memcpy(request.handle, &req, sizeof(req));
dev_dbg(nbd_to_dev(nbd), "request %p: sending control (%s@%llu,%uB)\n",
req, nbdcmd_to_ascii(type),
(unsigned long long)blk_rq_pos(req) << 9, blk_rq_bytes(req));
result = sock_xmit(nbd, 1, &request, sizeof(request),
(type == NBD_CMD_WRITE) ? MSG_MORE : 0);
if (result <= 0) {
dev_err(disk_to_dev(nbd->disk),
"Send control failed (result %d)\n", result);
return -EIO;
}
if (type == NBD_CMD_WRITE) {
struct req_iterator iter;
block: Convert bio_for_each_segment() to bvec_iter More prep work for immutable biovecs - with immutable bvecs drivers won't be able to use the biovec directly, they'll need to use helpers that take into account bio->bi_iter.bi_bvec_done. This updates callers for the new usage without changing the implementation yet. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Paul Clements <Paul.Clements@steeleye.com> Cc: Jim Paris <jim@jtan.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Nagalakshmi Nandigama <Nagalakshmi.Nandigama@lsi.com> Cc: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> Cc: support@lsi.com Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Tejun Heo <tj@kernel.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Matthew Wilcox <matthew.r.wilcox@intel.com> Cc: Keith Busch <keith.busch@intel.com> Cc: Stephen Hemminger <shemminger@vyatta.com> Cc: Quoc-Son Anh <quoc-sonx.anh@intel.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Jerome Marchand <jmarchan@redhat.com> Cc: Seth Jennings <sjenning@linux.vnet.ibm.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: "Darrick J. Wong" <darrick.wong@oracle.com> Cc: Chris Metcalf <cmetcalf@tilera.com> Cc: Jan Kara <jack@suse.cz> Cc: linux-m68k@lists.linux-m68k.org Cc: linuxppc-dev@lists.ozlabs.org Cc: drbd-user@lists.linbit.com Cc: nbd-general@lists.sourceforge.net Cc: cbe-oss-dev@lists.ozlabs.org Cc: xen-devel@lists.xensource.com Cc: virtualization@lists.linux-foundation.org Cc: linux-raid@vger.kernel.org Cc: linux-s390@vger.kernel.org Cc: DL-MPTFusionLinux@lsi.com Cc: linux-scsi@vger.kernel.org Cc: devel@driverdev.osuosl.org Cc: linux-fsdevel@vger.kernel.org Cc: cluster-devel@redhat.com Cc: linux-mm@kvack.org Acked-by: Geoff Levand <geoff@infradead.org>
2013-11-24 05:19:00 +04:00
struct bio_vec bvec;
/*
* we are really probing at internals to determine
* whether to set MSG_MORE or not...
*/
rq_for_each_segment(bvec, req, iter) {
flags = 0;
if (!rq_iter_last(bvec, iter))
flags = MSG_MORE;
dev_dbg(nbd_to_dev(nbd), "request %p: sending %d bytes data\n",
req, bvec.bv_len);
block: Convert bio_for_each_segment() to bvec_iter More prep work for immutable biovecs - with immutable bvecs drivers won't be able to use the biovec directly, they'll need to use helpers that take into account bio->bi_iter.bi_bvec_done. This updates callers for the new usage without changing the implementation yet. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Paul Clements <Paul.Clements@steeleye.com> Cc: Jim Paris <jim@jtan.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Nagalakshmi Nandigama <Nagalakshmi.Nandigama@lsi.com> Cc: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> Cc: support@lsi.com Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Tejun Heo <tj@kernel.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Matthew Wilcox <matthew.r.wilcox@intel.com> Cc: Keith Busch <keith.busch@intel.com> Cc: Stephen Hemminger <shemminger@vyatta.com> Cc: Quoc-Son Anh <quoc-sonx.anh@intel.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Jerome Marchand <jmarchan@redhat.com> Cc: Seth Jennings <sjenning@linux.vnet.ibm.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: "Darrick J. Wong" <darrick.wong@oracle.com> Cc: Chris Metcalf <cmetcalf@tilera.com> Cc: Jan Kara <jack@suse.cz> Cc: linux-m68k@lists.linux-m68k.org Cc: linuxppc-dev@lists.ozlabs.org Cc: drbd-user@lists.linbit.com Cc: nbd-general@lists.sourceforge.net Cc: cbe-oss-dev@lists.ozlabs.org Cc: xen-devel@lists.xensource.com Cc: virtualization@lists.linux-foundation.org Cc: linux-raid@vger.kernel.org Cc: linux-s390@vger.kernel.org Cc: DL-MPTFusionLinux@lsi.com Cc: linux-scsi@vger.kernel.org Cc: devel@driverdev.osuosl.org Cc: linux-fsdevel@vger.kernel.org Cc: cluster-devel@redhat.com Cc: linux-mm@kvack.org Acked-by: Geoff Levand <geoff@infradead.org>
2013-11-24 05:19:00 +04:00
result = sock_send_bvec(nbd, &bvec, flags);
if (result <= 0) {
dev_err(disk_to_dev(nbd->disk),
"Send data failed (result %d)\n",
result);
return -EIO;
}
}
}
return 0;
}
static struct request *nbd_find_request(struct nbd_device *nbd,
struct request *xreq)
{
struct request *req, *tmp;
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
int err;
err = wait_event_interruptible(nbd->active_wq, nbd->active_req != xreq);
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
if (unlikely(err))
return ERR_PTR(err);
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
spin_lock(&nbd->queue_lock);
list_for_each_entry_safe(req, tmp, &nbd->queue_head, queuelist) {
if (req != xreq)
continue;
list_del_init(&req->queuelist);
spin_unlock(&nbd->queue_lock);
return req;
}
spin_unlock(&nbd->queue_lock);
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
return ERR_PTR(-ENOENT);
}
static inline int sock_recv_bvec(struct nbd_device *nbd, struct bio_vec *bvec)
{
int result;
void *kaddr = kmap(bvec->bv_page);
result = sock_xmit(nbd, 0, kaddr + bvec->bv_offset, bvec->bv_len,
MSG_WAITALL);
kunmap(bvec->bv_page);
return result;
}
/* NULL returned = something went wrong, inform userspace */
static struct request *nbd_read_stat(struct nbd_device *nbd)
{
int result;
struct nbd_reply reply;
struct request *req;
reply.magic = 0;
result = sock_xmit(nbd, 0, &reply, sizeof(reply), MSG_WAITALL);
if (result <= 0) {
dev_err(disk_to_dev(nbd->disk),
"Receive control failed (result %d)\n", result);
return ERR_PTR(result);
}
if (ntohl(reply.magic) != NBD_REPLY_MAGIC) {
dev_err(disk_to_dev(nbd->disk), "Wrong magic (0x%lx)\n",
(unsigned long)ntohl(reply.magic));
return ERR_PTR(-EPROTO);
}
req = nbd_find_request(nbd, *(struct request **)reply.handle);
if (IS_ERR(req)) {
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
result = PTR_ERR(req);
if (result != -ENOENT)
return ERR_PTR(result);
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
dev_err(disk_to_dev(nbd->disk), "Unexpected reply (%p)\n",
reply.handle);
return ERR_PTR(-EBADR);
}
if (ntohl(reply.error)) {
dev_err(disk_to_dev(nbd->disk), "Other side returned error (%d)\n",
ntohl(reply.error));
req->errors++;
return req;
}
dev_dbg(nbd_to_dev(nbd), "request %p: got reply\n", req);
if (rq_data_dir(req) != WRITE) {
struct req_iterator iter;
block: Convert bio_for_each_segment() to bvec_iter More prep work for immutable biovecs - with immutable bvecs drivers won't be able to use the biovec directly, they'll need to use helpers that take into account bio->bi_iter.bi_bvec_done. This updates callers for the new usage without changing the implementation yet. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Paul Clements <Paul.Clements@steeleye.com> Cc: Jim Paris <jim@jtan.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Nagalakshmi Nandigama <Nagalakshmi.Nandigama@lsi.com> Cc: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> Cc: support@lsi.com Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Tejun Heo <tj@kernel.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Matthew Wilcox <matthew.r.wilcox@intel.com> Cc: Keith Busch <keith.busch@intel.com> Cc: Stephen Hemminger <shemminger@vyatta.com> Cc: Quoc-Son Anh <quoc-sonx.anh@intel.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Jerome Marchand <jmarchan@redhat.com> Cc: Seth Jennings <sjenning@linux.vnet.ibm.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: "Darrick J. Wong" <darrick.wong@oracle.com> Cc: Chris Metcalf <cmetcalf@tilera.com> Cc: Jan Kara <jack@suse.cz> Cc: linux-m68k@lists.linux-m68k.org Cc: linuxppc-dev@lists.ozlabs.org Cc: drbd-user@lists.linbit.com Cc: nbd-general@lists.sourceforge.net Cc: cbe-oss-dev@lists.ozlabs.org Cc: xen-devel@lists.xensource.com Cc: virtualization@lists.linux-foundation.org Cc: linux-raid@vger.kernel.org Cc: linux-s390@vger.kernel.org Cc: DL-MPTFusionLinux@lsi.com Cc: linux-scsi@vger.kernel.org Cc: devel@driverdev.osuosl.org Cc: linux-fsdevel@vger.kernel.org Cc: cluster-devel@redhat.com Cc: linux-mm@kvack.org Acked-by: Geoff Levand <geoff@infradead.org>
2013-11-24 05:19:00 +04:00
struct bio_vec bvec;
rq_for_each_segment(bvec, req, iter) {
block: Convert bio_for_each_segment() to bvec_iter More prep work for immutable biovecs - with immutable bvecs drivers won't be able to use the biovec directly, they'll need to use helpers that take into account bio->bi_iter.bi_bvec_done. This updates callers for the new usage without changing the implementation yet. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Paul Clements <Paul.Clements@steeleye.com> Cc: Jim Paris <jim@jtan.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Nagalakshmi Nandigama <Nagalakshmi.Nandigama@lsi.com> Cc: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> Cc: support@lsi.com Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Tejun Heo <tj@kernel.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Matthew Wilcox <matthew.r.wilcox@intel.com> Cc: Keith Busch <keith.busch@intel.com> Cc: Stephen Hemminger <shemminger@vyatta.com> Cc: Quoc-Son Anh <quoc-sonx.anh@intel.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Jerome Marchand <jmarchan@redhat.com> Cc: Seth Jennings <sjenning@linux.vnet.ibm.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: "Darrick J. Wong" <darrick.wong@oracle.com> Cc: Chris Metcalf <cmetcalf@tilera.com> Cc: Jan Kara <jack@suse.cz> Cc: linux-m68k@lists.linux-m68k.org Cc: linuxppc-dev@lists.ozlabs.org Cc: drbd-user@lists.linbit.com Cc: nbd-general@lists.sourceforge.net Cc: cbe-oss-dev@lists.ozlabs.org Cc: xen-devel@lists.xensource.com Cc: virtualization@lists.linux-foundation.org Cc: linux-raid@vger.kernel.org Cc: linux-s390@vger.kernel.org Cc: DL-MPTFusionLinux@lsi.com Cc: linux-scsi@vger.kernel.org Cc: devel@driverdev.osuosl.org Cc: linux-fsdevel@vger.kernel.org Cc: cluster-devel@redhat.com Cc: linux-mm@kvack.org Acked-by: Geoff Levand <geoff@infradead.org>
2013-11-24 05:19:00 +04:00
result = sock_recv_bvec(nbd, &bvec);
if (result <= 0) {
dev_err(disk_to_dev(nbd->disk), "Receive data failed (result %d)\n",
result);
req->errors++;
return req;
}
dev_dbg(nbd_to_dev(nbd), "request %p: got %d bytes data\n",
req, bvec.bv_len);
}
}
return req;
}
static ssize_t pid_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct gendisk *disk = dev_to_disk(dev);
struct nbd_device *nbd = (struct nbd_device *)disk->private_data;
return sprintf(buf, "%d\n", task_pid_nr(nbd->task_recv));
}
static struct device_attribute pid_attr = {
.attr = { .name = "pid", .mode = S_IRUGO},
.show = pid_show,
};
static int nbd_thread_recv(struct nbd_device *nbd)
{
struct request *req;
int ret;
BUG_ON(nbd->magic != NBD_MAGIC);
sk_set_memalloc(nbd->sock->sk);
nbd->task_recv = current;
ret = device_create_file(disk_to_dev(nbd->disk), &pid_attr);
if (ret) {
dev_err(disk_to_dev(nbd->disk), "device_create_file failed!\n");
nbd->task_recv = NULL;
return ret;
}
while (1) {
req = nbd_read_stat(nbd);
if (IS_ERR(req)) {
ret = PTR_ERR(req);
break;
}
nbd_end_request(nbd, req);
}
device_remove_file(disk_to_dev(nbd->disk), &pid_attr);
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
nbd->task_recv = NULL;
return ret;
}
static void nbd_clear_que(struct nbd_device *nbd)
{
struct request *req;
BUG_ON(nbd->magic != NBD_MAGIC);
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
/*
* Because we have set nbd->sock to NULL under the tx_lock, all
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
* 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(nbd->sock);
BUG_ON(nbd->active_req);
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
while (!list_empty(&nbd->queue_head)) {
req = list_entry(nbd->queue_head.next, struct request,
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
queuelist);
list_del_init(&req->queuelist);
req->errors++;
nbd_end_request(nbd, req);
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
}
while (!list_empty(&nbd->waiting_queue)) {
req = list_entry(nbd->waiting_queue.next, struct request,
queuelist);
list_del_init(&req->queuelist);
req->errors++;
nbd_end_request(nbd, req);
}
dev_dbg(disk_to_dev(nbd->disk), "queue cleared\n");
}
static void nbd_handle_req(struct nbd_device *nbd, struct request *req)
{
if (req->cmd_type != REQ_TYPE_FS)
goto error_out;
if (rq_data_dir(req) == WRITE &&
(nbd->flags & NBD_FLAG_READ_ONLY)) {
dev_err(disk_to_dev(nbd->disk),
"Write on read-only\n");
goto error_out;
nbd: support FLUSH requests Currently, the NBD device does not accept flush requests from the Linux block layer. If the NBD server opened the target with neither O_SYNC nor O_DSYNC, however, the device will be effectively backed by a writeback cache. Without issuing flushes properly, operation of the NBD device will not be safe against power losses. The NBD protocol has support for both a cache flush command and a FUA command flag; the server will also pass a flag to note its support for these features. This patch adds support for the cache flush command and flag. In the kernel, we receive the flags via the NBD_SET_FLAGS ioctl, and map NBD_FLAG_SEND_FLUSH to the argument of blk_queue_flush. When the flag is active the block layer will send REQ_FLUSH requests, which we translate to NBD_CMD_FLUSH commands. FUA support is not included in this patch because all free software servers implement it with a full fdatasync; thus it has no advantage over supporting flush only. Because I [Paolo] cannot really benchmark it in a realistic scenario, I cannot tell if it is a good idea or not. It is also not clear if it is valid for an NBD server to support FUA but not flush. The Linux block layer gives a warning for this combination, the NBD protocol documentation says nothing about it. The patch also fixes a small problem in the handling of flags: nbd->flags must be cleared at the end of NBD_DO_IT, but the driver was not doing that. The bug manifests itself as follows. Suppose you two different client/server pairs to start the NBD device. Suppose also that the first client supports NBD_SET_FLAGS, and the first server sends NBD_FLAG_SEND_FLUSH; the second pair instead does neither of these two things. Before this patch, the second invocation of NBD_DO_IT will use a stale value of nbd->flags, and the second server will issue an error every time it receives an NBD_CMD_FLUSH command. This bug is pre-existing, but it becomes much more important after this patch; flush failures make the device pretty much unusable, unlike Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Alex Bligh <alex@alex.org.uk> Acked-by: Paul Clements <Paul.Clements@steeleye.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 05:05:23 +04:00
}
req->errors = 0;
mutex_lock(&nbd->tx_lock);
if (unlikely(!nbd->sock)) {
mutex_unlock(&nbd->tx_lock);
dev_err(disk_to_dev(nbd->disk),
"Attempted send on closed socket\n");
goto error_out;
}
nbd->active_req = req;
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
if (nbd->xmit_timeout && list_empty_careful(&nbd->queue_head))
mod_timer(&nbd->timeout_timer, jiffies + nbd->xmit_timeout);
if (nbd_send_req(nbd, req) != 0) {
dev_err(disk_to_dev(nbd->disk), "Request send failed\n");
req->errors++;
nbd_end_request(nbd, req);
} else {
spin_lock(&nbd->queue_lock);
list_add_tail(&req->queuelist, &nbd->queue_head);
spin_unlock(&nbd->queue_lock);
}
nbd->active_req = NULL;
mutex_unlock(&nbd->tx_lock);
wake_up_all(&nbd->active_wq);
return;
error_out:
req->errors++;
nbd_end_request(nbd, req);
}
static int nbd_thread_send(void *data)
{
struct nbd_device *nbd = data;
struct request *req;
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
nbd->task_send = current;
set_user_nice(current, MIN_NICE);
while (!kthread_should_stop() || !list_empty(&nbd->waiting_queue)) {
/* wait for something to do */
wait_event_interruptible(nbd->waiting_wq,
kthread_should_stop() ||
!list_empty(&nbd->waiting_queue));
/* extract request */
if (list_empty(&nbd->waiting_queue))
continue;
spin_lock_irq(&nbd->queue_lock);
req = list_entry(nbd->waiting_queue.next, struct request,
queuelist);
list_del_init(&req->queuelist);
spin_unlock_irq(&nbd->queue_lock);
/* handle request */
nbd_handle_req(nbd, req);
}
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
nbd->task_send = NULL;
return 0;
}
/*
* We always wait for result of write, for now. It would be nice to make it optional
* in future
* if ((rq_data_dir(req) == WRITE) && (nbd->flags & NBD_WRITE_NOCHK))
* { printk( "Warning: Ignoring result!\n"); nbd_end_request( req ); }
*/
static void nbd_request_handler(struct request_queue *q)
__releases(q->queue_lock) __acquires(q->queue_lock)
{
struct request *req;
block: implement and enforce request peek/start/fetch Till now block layer allowed two separate modes of request execution. A request is always acquired from the request queue via elv_next_request(). After that, drivers are free to either dequeue it or process it without dequeueing. Dequeue allows elv_next_request() to return the next request so that multiple requests can be in flight. Executing requests without dequeueing has its merits mostly in allowing drivers for simpler devices which can't do sg to deal with segments only without considering request boundary. However, the benefit this brings is dubious and declining while the cost of the API ambiguity is increasing. Segment based drivers are usually for very old or limited devices and as converting to dequeueing model isn't difficult, it doesn't justify the API overhead it puts on block layer and its more modern users. Previous patches converted all block low level drivers to dequeueing model. This patch completes the API transition by... * renaming elv_next_request() to blk_peek_request() * renaming blkdev_dequeue_request() to blk_start_request() * adding blk_fetch_request() which is combination of peek and start * disallowing completion of queued (not started) requests * applying new API to all LLDs Renamings are for consistency and to break out of tree code so that it's apparent that out of tree drivers need updating. [ Impact: block request issue API cleanup, no functional change ] Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: James Bottomley <James.Bottomley@HansenPartnership.com> Cc: Mike Miller <mike.miller@hp.com> Cc: unsik Kim <donari75@gmail.com> Cc: Paul Clements <paul.clements@steeleye.com> Cc: Tim Waugh <tim@cyberelk.net> Cc: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com> Cc: David S. Miller <davem@davemloft.net> Cc: Laurent Vivier <Laurent@lvivier.info> Cc: Jeff Garzik <jgarzik@pobox.com> Cc: Jeremy Fitzhardinge <jeremy@xensource.com> Cc: Grant Likely <grant.likely@secretlab.ca> Cc: Adrian McMenamin <adrian@mcmen.demon.co.uk> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> Cc: Borislav Petkov <petkovbb@googlemail.com> Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com> Cc: Alex Dubov <oakad@yahoo.com> Cc: Pierre Ossman <drzeus@drzeus.cx> Cc: David Woodhouse <dwmw2@infradead.org> Cc: Markus Lidel <Markus.Lidel@shadowconnect.com> Cc: Stefan Weinhuber <wein@de.ibm.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Pete Zaitcev <zaitcev@redhat.com> Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-05-08 06:54:16 +04:00
while ((req = blk_fetch_request(q)) != NULL) {
struct nbd_device *nbd;
spin_unlock_irq(q->queue_lock);
nbd = req->rq_disk->private_data;
BUG_ON(nbd->magic != NBD_MAGIC);
dev_dbg(nbd_to_dev(nbd), "request %p: dequeued (flags=%x)\n",
req, req->cmd_type);
if (unlikely(!nbd->sock)) {
dev_err(disk_to_dev(nbd->disk),
"Attempted send on closed socket\n");
req->errors++;
nbd_end_request(nbd, req);
spin_lock_irq(q->queue_lock);
continue;
}
spin_lock_irq(&nbd->queue_lock);
list_add_tail(&req->queuelist, &nbd->waiting_queue);
spin_unlock_irq(&nbd->queue_lock);
wake_up(&nbd->waiting_wq);
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
spin_lock_irq(q->queue_lock);
}
}
static int nbd_set_socket(struct nbd_device *nbd, struct socket *sock)
{
int ret = 0;
spin_lock_irq(&nbd->sock_lock);
if (nbd->sock) {
ret = -EBUSY;
goto out;
}
nbd->sock = sock;
out:
spin_unlock_irq(&nbd->sock_lock);
return ret;
}
static int nbd_dev_dbg_init(struct nbd_device *nbd);
static void nbd_dev_dbg_close(struct nbd_device *nbd);
/* Must be called with tx_lock held */
static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *nbd,
unsigned int cmd, unsigned long arg)
{
switch (cmd) {
case NBD_DISCONNECT: {
struct request sreq;
dev_info(disk_to_dev(nbd->disk), "NBD_DISCONNECT\n");
if (!nbd->sock)
return -EINVAL;
mutex_unlock(&nbd->tx_lock);
fsync_bdev(bdev);
mutex_lock(&nbd->tx_lock);
blk_rq_init(NULL, &sreq);
sreq.cmd_type = REQ_TYPE_DRV_PRIV;
/* Check again after getting mutex back. */
if (!nbd->sock)
return -EINVAL;
nbd->disconnect = true;
nbd_send_req(nbd, &sreq);
return 0;
}
case NBD_CLEAR_SOCK:
sock_shutdown(nbd);
nbd_clear_que(nbd);
BUG_ON(!list_empty(&nbd->queue_head));
BUG_ON(!list_empty(&nbd->waiting_queue));
kill_bdev(bdev);
return 0;
case NBD_SET_SOCK: {
int err;
struct socket *sock = sockfd_lookup(arg, &err);
if (!sock)
return err;
err = nbd_set_socket(nbd, sock);
if (!err && max_part)
bdev->bd_invalidated = 1;
return err;
}
case NBD_SET_BLKSIZE:
nbd->blksize = arg;
nbd->bytesize &= ~(nbd->blksize-1);
bdev->bd_inode->i_size = nbd->bytesize;
set_blocksize(bdev, nbd->blksize);
set_capacity(nbd->disk, nbd->bytesize >> 9);
return 0;
case NBD_SET_SIZE:
nbd->bytesize = arg & ~(nbd->blksize-1);
bdev->bd_inode->i_size = nbd->bytesize;
set_blocksize(bdev, nbd->blksize);
set_capacity(nbd->disk, nbd->bytesize >> 9);
return 0;
case NBD_SET_TIMEOUT:
nbd->xmit_timeout = arg * HZ;
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
if (arg)
mod_timer(&nbd->timeout_timer,
jiffies + nbd->xmit_timeout);
else
del_timer_sync(&nbd->timeout_timer);
return 0;
case NBD_SET_FLAGS:
nbd->flags = arg;
return 0;
case NBD_SET_SIZE_BLOCKS:
nbd->bytesize = ((u64) arg) * nbd->blksize;
bdev->bd_inode->i_size = nbd->bytesize;
set_blocksize(bdev, nbd->blksize);
set_capacity(nbd->disk, nbd->bytesize >> 9);
return 0;
case NBD_DO_IT: {
struct task_struct *thread;
int error;
if (nbd->task_recv)
return -EBUSY;
if (!nbd->sock)
return -EINVAL;
mutex_unlock(&nbd->tx_lock);
if (nbd->flags & NBD_FLAG_READ_ONLY)
set_device_ro(bdev, true);
if (nbd->flags & NBD_FLAG_SEND_TRIM)
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD,
nbd->disk->queue);
nbd: support FLUSH requests Currently, the NBD device does not accept flush requests from the Linux block layer. If the NBD server opened the target with neither O_SYNC nor O_DSYNC, however, the device will be effectively backed by a writeback cache. Without issuing flushes properly, operation of the NBD device will not be safe against power losses. The NBD protocol has support for both a cache flush command and a FUA command flag; the server will also pass a flag to note its support for these features. This patch adds support for the cache flush command and flag. In the kernel, we receive the flags via the NBD_SET_FLAGS ioctl, and map NBD_FLAG_SEND_FLUSH to the argument of blk_queue_flush. When the flag is active the block layer will send REQ_FLUSH requests, which we translate to NBD_CMD_FLUSH commands. FUA support is not included in this patch because all free software servers implement it with a full fdatasync; thus it has no advantage over supporting flush only. Because I [Paolo] cannot really benchmark it in a realistic scenario, I cannot tell if it is a good idea or not. It is also not clear if it is valid for an NBD server to support FUA but not flush. The Linux block layer gives a warning for this combination, the NBD protocol documentation says nothing about it. The patch also fixes a small problem in the handling of flags: nbd->flags must be cleared at the end of NBD_DO_IT, but the driver was not doing that. The bug manifests itself as follows. Suppose you two different client/server pairs to start the NBD device. Suppose also that the first client supports NBD_SET_FLAGS, and the first server sends NBD_FLAG_SEND_FLUSH; the second pair instead does neither of these two things. Before this patch, the second invocation of NBD_DO_IT will use a stale value of nbd->flags, and the second server will issue an error every time it receives an NBD_CMD_FLUSH command. This bug is pre-existing, but it becomes much more important after this patch; flush failures make the device pretty much unusable, unlike Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Alex Bligh <alex@alex.org.uk> Acked-by: Paul Clements <Paul.Clements@steeleye.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 05:05:23 +04:00
if (nbd->flags & NBD_FLAG_SEND_FLUSH)
blk_queue_flush(nbd->disk->queue, REQ_FLUSH);
else
blk_queue_flush(nbd->disk->queue, 0);
thread = kthread_run(nbd_thread_send, nbd, "%s",
nbd_name(nbd));
if (IS_ERR(thread)) {
mutex_lock(&nbd->tx_lock);
return PTR_ERR(thread);
}
nbd_dev_dbg_init(nbd);
error = nbd_thread_recv(nbd);
nbd_dev_dbg_close(nbd);
kthread_stop(thread);
mutex_lock(&nbd->tx_lock);
sock_shutdown(nbd);
nbd_clear_que(nbd);
kill_bdev(bdev);
queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, nbd->disk->queue);
set_device_ro(bdev, false);
nbd: support FLUSH requests Currently, the NBD device does not accept flush requests from the Linux block layer. If the NBD server opened the target with neither O_SYNC nor O_DSYNC, however, the device will be effectively backed by a writeback cache. Without issuing flushes properly, operation of the NBD device will not be safe against power losses. The NBD protocol has support for both a cache flush command and a FUA command flag; the server will also pass a flag to note its support for these features. This patch adds support for the cache flush command and flag. In the kernel, we receive the flags via the NBD_SET_FLAGS ioctl, and map NBD_FLAG_SEND_FLUSH to the argument of blk_queue_flush. When the flag is active the block layer will send REQ_FLUSH requests, which we translate to NBD_CMD_FLUSH commands. FUA support is not included in this patch because all free software servers implement it with a full fdatasync; thus it has no advantage over supporting flush only. Because I [Paolo] cannot really benchmark it in a realistic scenario, I cannot tell if it is a good idea or not. It is also not clear if it is valid for an NBD server to support FUA but not flush. The Linux block layer gives a warning for this combination, the NBD protocol documentation says nothing about it. The patch also fixes a small problem in the handling of flags: nbd->flags must be cleared at the end of NBD_DO_IT, but the driver was not doing that. The bug manifests itself as follows. Suppose you two different client/server pairs to start the NBD device. Suppose also that the first client supports NBD_SET_FLAGS, and the first server sends NBD_FLAG_SEND_FLUSH; the second pair instead does neither of these two things. Before this patch, the second invocation of NBD_DO_IT will use a stale value of nbd->flags, and the second server will issue an error every time it receives an NBD_CMD_FLUSH command. This bug is pre-existing, but it becomes much more important after this patch; flush failures make the device pretty much unusable, unlike Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Alex Bligh <alex@alex.org.uk> Acked-by: Paul Clements <Paul.Clements@steeleye.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 05:05:23 +04:00
nbd->flags = 0;
nbd->bytesize = 0;
bdev->bd_inode->i_size = 0;
set_capacity(nbd->disk, 0);
NBD: add partition support Permit the use of partitions with network block devices (NBD). A new parameter is introduced to define how many partition we want to be able to manage per network block device. This parameter is "max_part". For instance, to manage 63 partitions / loop device, we will do: [on the server side] # nbd-server 1234 /dev/sdb [on the client side] # modprobe nbd max_part=63 # ls -l /dev/nbd* brw-rw---- 1 root disk 43, 0 2008-03-25 11:14 /dev/nbd0 brw-rw---- 1 root disk 43, 64 2008-03-25 11:11 /dev/nbd1 brw-rw---- 1 root disk 43, 640 2008-03-25 11:11 /dev/nbd10 brw-rw---- 1 root disk 43, 704 2008-03-25 11:11 /dev/nbd11 brw-rw---- 1 root disk 43, 768 2008-03-25 11:11 /dev/nbd12 brw-rw---- 1 root disk 43, 832 2008-03-25 11:11 /dev/nbd13 brw-rw---- 1 root disk 43, 896 2008-03-25 11:11 /dev/nbd14 brw-rw---- 1 root disk 43, 960 2008-03-25 11:11 /dev/nbd15 brw-rw---- 1 root disk 43, 128 2008-03-25 11:11 /dev/nbd2 brw-rw---- 1 root disk 43, 192 2008-03-25 11:11 /dev/nbd3 brw-rw---- 1 root disk 43, 256 2008-03-25 11:11 /dev/nbd4 brw-rw---- 1 root disk 43, 320 2008-03-25 11:11 /dev/nbd5 brw-rw---- 1 root disk 43, 384 2008-03-25 11:11 /dev/nbd6 brw-rw---- 1 root disk 43, 448 2008-03-25 11:11 /dev/nbd7 brw-rw---- 1 root disk 43, 512 2008-03-25 11:11 /dev/nbd8 brw-rw---- 1 root disk 43, 576 2008-03-25 11:11 /dev/nbd9 # nbd-client localhost 1234 /dev/nbd0 Negotiation: ..size = 80418240KB bs=1024, sz=80418240 -------NOTE, RFC: partition table is not automatically read. The driver sets bdev->bd_invalidated to 1 to force the read of the partition table of the device, but this is done only on an open of the device. So we have to do a "touch /dev/nbdX" or something like that. It can't be done from the nbd-client or nbd driver because at this level we can't ask to read the partition table and to serve the request at the same time (-> deadlock) If someone has a better idea, I'm open to any suggestion. -------NOTE, RFC # fdisk -l /dev/nbd0 Disk /dev/nbd0: 82.3 GB, 82348277760 bytes 255 heads, 63 sectors/track, 10011 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/nbd0p1 * 1 9965 80043831 83 Linux /dev/nbd0p2 9966 10011 369495 5 Extended /dev/nbd0p5 9966 10011 369463+ 82 Linux swap / Solaris # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 brw-rw---- 1 root disk 43, 1 2008-03-25 11:16 /dev/nbd0p1 brw-rw---- 1 root disk 43, 2 2008-03-25 11:16 /dev/nbd0p2 brw-rw---- 1 root disk 43, 5 2008-03-25 11:16 /dev/nbd0p5 # mount /dev/nbd0p1 /mnt # ls /mnt bin dev initrd lost+found opt sbin sys var boot etc initrd.img media proc selinux tmp vmlinuz cdrom home lib mnt root srv usr # umount /mnt # nbd-client -d /dev/nbd0 # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 -------NOTE On "nbd-client -d", we can do an iocl(BLKRRPART) to update partition table: as the size of the device is 0, we don't have to serve the partition manager request (-> no deadlock). -------NOTE Signed-off-by: Paul Clements <paul.clements@steeleye.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-29 12:02:51 +04:00
if (max_part > 0)
blkdev_reread_part(bdev);
if (nbd->disconnect) /* user requested, ignore socket errors */
return 0;
return error;
}
case NBD_CLEAR_QUE:
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
/*
* This is for compatibility only. The queue is always cleared
* by NBD_DO_IT or NBD_CLEAR_SOCK.
*/
return 0;
case NBD_PRINT_DEBUG:
dev_info(disk_to_dev(nbd->disk),
"next = %p, prev = %p, head = %p\n",
nbd->queue_head.next, nbd->queue_head.prev,
&nbd->queue_head);
return 0;
}
return -ENOTTY;
}
static int nbd_ioctl(struct block_device *bdev, fmode_t mode,
unsigned int cmd, unsigned long arg)
{
struct nbd_device *nbd = bdev->bd_disk->private_data;
int error;
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
BUG_ON(nbd->magic != NBD_MAGIC);
mutex_lock(&nbd->tx_lock);
error = __nbd_ioctl(bdev, nbd, cmd, arg);
mutex_unlock(&nbd->tx_lock);
return error;
}
static const struct block_device_operations nbd_fops =
{
.owner = THIS_MODULE,
.ioctl = nbd_ioctl,
.compat_ioctl = nbd_ioctl,
};
#if IS_ENABLED(CONFIG_DEBUG_FS)
static int nbd_dbg_tasks_show(struct seq_file *s, void *unused)
{
struct nbd_device *nbd = s->private;
if (nbd->task_recv)
seq_printf(s, "recv: %d\n", task_pid_nr(nbd->task_recv));
if (nbd->task_send)
seq_printf(s, "send: %d\n", task_pid_nr(nbd->task_send));
return 0;
}
static int nbd_dbg_tasks_open(struct inode *inode, struct file *file)
{
return single_open(file, nbd_dbg_tasks_show, inode->i_private);
}
static const struct file_operations nbd_dbg_tasks_ops = {
.open = nbd_dbg_tasks_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static int nbd_dbg_flags_show(struct seq_file *s, void *unused)
{
struct nbd_device *nbd = s->private;
u32 flags = nbd->flags;
seq_printf(s, "Hex: 0x%08x\n\n", flags);
seq_puts(s, "Known flags:\n");
if (flags & NBD_FLAG_HAS_FLAGS)
seq_puts(s, "NBD_FLAG_HAS_FLAGS\n");
if (flags & NBD_FLAG_READ_ONLY)
seq_puts(s, "NBD_FLAG_READ_ONLY\n");
if (flags & NBD_FLAG_SEND_FLUSH)
seq_puts(s, "NBD_FLAG_SEND_FLUSH\n");
if (flags & NBD_FLAG_SEND_TRIM)
seq_puts(s, "NBD_FLAG_SEND_TRIM\n");
return 0;
}
static int nbd_dbg_flags_open(struct inode *inode, struct file *file)
{
return single_open(file, nbd_dbg_flags_show, inode->i_private);
}
static const struct file_operations nbd_dbg_flags_ops = {
.open = nbd_dbg_flags_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static int nbd_dev_dbg_init(struct nbd_device *nbd)
{
struct dentry *dir;
if (!nbd_dbg_dir)
return -EIO;
dir = debugfs_create_dir(nbd_name(nbd), nbd_dbg_dir);
if (!dir) {
dev_err(nbd_to_dev(nbd), "Failed to create debugfs dir for '%s'\n",
nbd_name(nbd));
return -EIO;
}
nbd->dbg_dir = dir;
debugfs_create_file("tasks", 0444, dir, nbd, &nbd_dbg_tasks_ops);
debugfs_create_u64("size_bytes", 0444, dir, &nbd->bytesize);
debugfs_create_u32("timeout", 0444, dir, &nbd->xmit_timeout);
debugfs_create_u32("blocksize", 0444, dir, &nbd->blksize);
debugfs_create_file("flags", 0444, dir, &nbd, &nbd_dbg_flags_ops);
return 0;
}
static void nbd_dev_dbg_close(struct nbd_device *nbd)
{
debugfs_remove_recursive(nbd->dbg_dir);
}
static int nbd_dbg_init(void)
{
struct dentry *dbg_dir;
dbg_dir = debugfs_create_dir("nbd", NULL);
if (!dbg_dir)
return -EIO;
nbd_dbg_dir = dbg_dir;
return 0;
}
static void nbd_dbg_close(void)
{
debugfs_remove_recursive(nbd_dbg_dir);
}
#else /* IS_ENABLED(CONFIG_DEBUG_FS) */
static int nbd_dev_dbg_init(struct nbd_device *nbd)
{
return 0;
}
static void nbd_dev_dbg_close(struct nbd_device *nbd)
{
}
static int nbd_dbg_init(void)
{
return 0;
}
static void nbd_dbg_close(void)
{
}
#endif
/*
* And here should be modules and kernel interface
* (Just smiley confuses emacs :-)
*/
static int __init nbd_init(void)
{
int err = -ENOMEM;
int i;
NBD: add partition support Permit the use of partitions with network block devices (NBD). A new parameter is introduced to define how many partition we want to be able to manage per network block device. This parameter is "max_part". For instance, to manage 63 partitions / loop device, we will do: [on the server side] # nbd-server 1234 /dev/sdb [on the client side] # modprobe nbd max_part=63 # ls -l /dev/nbd* brw-rw---- 1 root disk 43, 0 2008-03-25 11:14 /dev/nbd0 brw-rw---- 1 root disk 43, 64 2008-03-25 11:11 /dev/nbd1 brw-rw---- 1 root disk 43, 640 2008-03-25 11:11 /dev/nbd10 brw-rw---- 1 root disk 43, 704 2008-03-25 11:11 /dev/nbd11 brw-rw---- 1 root disk 43, 768 2008-03-25 11:11 /dev/nbd12 brw-rw---- 1 root disk 43, 832 2008-03-25 11:11 /dev/nbd13 brw-rw---- 1 root disk 43, 896 2008-03-25 11:11 /dev/nbd14 brw-rw---- 1 root disk 43, 960 2008-03-25 11:11 /dev/nbd15 brw-rw---- 1 root disk 43, 128 2008-03-25 11:11 /dev/nbd2 brw-rw---- 1 root disk 43, 192 2008-03-25 11:11 /dev/nbd3 brw-rw---- 1 root disk 43, 256 2008-03-25 11:11 /dev/nbd4 brw-rw---- 1 root disk 43, 320 2008-03-25 11:11 /dev/nbd5 brw-rw---- 1 root disk 43, 384 2008-03-25 11:11 /dev/nbd6 brw-rw---- 1 root disk 43, 448 2008-03-25 11:11 /dev/nbd7 brw-rw---- 1 root disk 43, 512 2008-03-25 11:11 /dev/nbd8 brw-rw---- 1 root disk 43, 576 2008-03-25 11:11 /dev/nbd9 # nbd-client localhost 1234 /dev/nbd0 Negotiation: ..size = 80418240KB bs=1024, sz=80418240 -------NOTE, RFC: partition table is not automatically read. The driver sets bdev->bd_invalidated to 1 to force the read of the partition table of the device, but this is done only on an open of the device. So we have to do a "touch /dev/nbdX" or something like that. It can't be done from the nbd-client or nbd driver because at this level we can't ask to read the partition table and to serve the request at the same time (-> deadlock) If someone has a better idea, I'm open to any suggestion. -------NOTE, RFC # fdisk -l /dev/nbd0 Disk /dev/nbd0: 82.3 GB, 82348277760 bytes 255 heads, 63 sectors/track, 10011 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/nbd0p1 * 1 9965 80043831 83 Linux /dev/nbd0p2 9966 10011 369495 5 Extended /dev/nbd0p5 9966 10011 369463+ 82 Linux swap / Solaris # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 brw-rw---- 1 root disk 43, 1 2008-03-25 11:16 /dev/nbd0p1 brw-rw---- 1 root disk 43, 2 2008-03-25 11:16 /dev/nbd0p2 brw-rw---- 1 root disk 43, 5 2008-03-25 11:16 /dev/nbd0p5 # mount /dev/nbd0p1 /mnt # ls /mnt bin dev initrd lost+found opt sbin sys var boot etc initrd.img media proc selinux tmp vmlinuz cdrom home lib mnt root srv usr # umount /mnt # nbd-client -d /dev/nbd0 # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 -------NOTE On "nbd-client -d", we can do an iocl(BLKRRPART) to update partition table: as the size of the device is 0, we don't have to serve the partition manager request (-> no deadlock). -------NOTE Signed-off-by: Paul Clements <paul.clements@steeleye.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-29 12:02:51 +04:00
int part_shift;
BUILD_BUG_ON(sizeof(struct nbd_request) != 28);
NBD: add partition support Permit the use of partitions with network block devices (NBD). A new parameter is introduced to define how many partition we want to be able to manage per network block device. This parameter is "max_part". For instance, to manage 63 partitions / loop device, we will do: [on the server side] # nbd-server 1234 /dev/sdb [on the client side] # modprobe nbd max_part=63 # ls -l /dev/nbd* brw-rw---- 1 root disk 43, 0 2008-03-25 11:14 /dev/nbd0 brw-rw---- 1 root disk 43, 64 2008-03-25 11:11 /dev/nbd1 brw-rw---- 1 root disk 43, 640 2008-03-25 11:11 /dev/nbd10 brw-rw---- 1 root disk 43, 704 2008-03-25 11:11 /dev/nbd11 brw-rw---- 1 root disk 43, 768 2008-03-25 11:11 /dev/nbd12 brw-rw---- 1 root disk 43, 832 2008-03-25 11:11 /dev/nbd13 brw-rw---- 1 root disk 43, 896 2008-03-25 11:11 /dev/nbd14 brw-rw---- 1 root disk 43, 960 2008-03-25 11:11 /dev/nbd15 brw-rw---- 1 root disk 43, 128 2008-03-25 11:11 /dev/nbd2 brw-rw---- 1 root disk 43, 192 2008-03-25 11:11 /dev/nbd3 brw-rw---- 1 root disk 43, 256 2008-03-25 11:11 /dev/nbd4 brw-rw---- 1 root disk 43, 320 2008-03-25 11:11 /dev/nbd5 brw-rw---- 1 root disk 43, 384 2008-03-25 11:11 /dev/nbd6 brw-rw---- 1 root disk 43, 448 2008-03-25 11:11 /dev/nbd7 brw-rw---- 1 root disk 43, 512 2008-03-25 11:11 /dev/nbd8 brw-rw---- 1 root disk 43, 576 2008-03-25 11:11 /dev/nbd9 # nbd-client localhost 1234 /dev/nbd0 Negotiation: ..size = 80418240KB bs=1024, sz=80418240 -------NOTE, RFC: partition table is not automatically read. The driver sets bdev->bd_invalidated to 1 to force the read of the partition table of the device, but this is done only on an open of the device. So we have to do a "touch /dev/nbdX" or something like that. It can't be done from the nbd-client or nbd driver because at this level we can't ask to read the partition table and to serve the request at the same time (-> deadlock) If someone has a better idea, I'm open to any suggestion. -------NOTE, RFC # fdisk -l /dev/nbd0 Disk /dev/nbd0: 82.3 GB, 82348277760 bytes 255 heads, 63 sectors/track, 10011 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/nbd0p1 * 1 9965 80043831 83 Linux /dev/nbd0p2 9966 10011 369495 5 Extended /dev/nbd0p5 9966 10011 369463+ 82 Linux swap / Solaris # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 brw-rw---- 1 root disk 43, 1 2008-03-25 11:16 /dev/nbd0p1 brw-rw---- 1 root disk 43, 2 2008-03-25 11:16 /dev/nbd0p2 brw-rw---- 1 root disk 43, 5 2008-03-25 11:16 /dev/nbd0p5 # mount /dev/nbd0p1 /mnt # ls /mnt bin dev initrd lost+found opt sbin sys var boot etc initrd.img media proc selinux tmp vmlinuz cdrom home lib mnt root srv usr # umount /mnt # nbd-client -d /dev/nbd0 # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 -------NOTE On "nbd-client -d", we can do an iocl(BLKRRPART) to update partition table: as the size of the device is 0, we don't have to serve the partition manager request (-> no deadlock). -------NOTE Signed-off-by: Paul Clements <paul.clements@steeleye.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-29 12:02:51 +04:00
if (max_part < 0) {
printk(KERN_ERR "nbd: max_part must be >= 0\n");
NBD: add partition support Permit the use of partitions with network block devices (NBD). A new parameter is introduced to define how many partition we want to be able to manage per network block device. This parameter is "max_part". For instance, to manage 63 partitions / loop device, we will do: [on the server side] # nbd-server 1234 /dev/sdb [on the client side] # modprobe nbd max_part=63 # ls -l /dev/nbd* brw-rw---- 1 root disk 43, 0 2008-03-25 11:14 /dev/nbd0 brw-rw---- 1 root disk 43, 64 2008-03-25 11:11 /dev/nbd1 brw-rw---- 1 root disk 43, 640 2008-03-25 11:11 /dev/nbd10 brw-rw---- 1 root disk 43, 704 2008-03-25 11:11 /dev/nbd11 brw-rw---- 1 root disk 43, 768 2008-03-25 11:11 /dev/nbd12 brw-rw---- 1 root disk 43, 832 2008-03-25 11:11 /dev/nbd13 brw-rw---- 1 root disk 43, 896 2008-03-25 11:11 /dev/nbd14 brw-rw---- 1 root disk 43, 960 2008-03-25 11:11 /dev/nbd15 brw-rw---- 1 root disk 43, 128 2008-03-25 11:11 /dev/nbd2 brw-rw---- 1 root disk 43, 192 2008-03-25 11:11 /dev/nbd3 brw-rw---- 1 root disk 43, 256 2008-03-25 11:11 /dev/nbd4 brw-rw---- 1 root disk 43, 320 2008-03-25 11:11 /dev/nbd5 brw-rw---- 1 root disk 43, 384 2008-03-25 11:11 /dev/nbd6 brw-rw---- 1 root disk 43, 448 2008-03-25 11:11 /dev/nbd7 brw-rw---- 1 root disk 43, 512 2008-03-25 11:11 /dev/nbd8 brw-rw---- 1 root disk 43, 576 2008-03-25 11:11 /dev/nbd9 # nbd-client localhost 1234 /dev/nbd0 Negotiation: ..size = 80418240KB bs=1024, sz=80418240 -------NOTE, RFC: partition table is not automatically read. The driver sets bdev->bd_invalidated to 1 to force the read of the partition table of the device, but this is done only on an open of the device. So we have to do a "touch /dev/nbdX" or something like that. It can't be done from the nbd-client or nbd driver because at this level we can't ask to read the partition table and to serve the request at the same time (-> deadlock) If someone has a better idea, I'm open to any suggestion. -------NOTE, RFC # fdisk -l /dev/nbd0 Disk /dev/nbd0: 82.3 GB, 82348277760 bytes 255 heads, 63 sectors/track, 10011 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/nbd0p1 * 1 9965 80043831 83 Linux /dev/nbd0p2 9966 10011 369495 5 Extended /dev/nbd0p5 9966 10011 369463+ 82 Linux swap / Solaris # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 brw-rw---- 1 root disk 43, 1 2008-03-25 11:16 /dev/nbd0p1 brw-rw---- 1 root disk 43, 2 2008-03-25 11:16 /dev/nbd0p2 brw-rw---- 1 root disk 43, 5 2008-03-25 11:16 /dev/nbd0p5 # mount /dev/nbd0p1 /mnt # ls /mnt bin dev initrd lost+found opt sbin sys var boot etc initrd.img media proc selinux tmp vmlinuz cdrom home lib mnt root srv usr # umount /mnt # nbd-client -d /dev/nbd0 # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 -------NOTE On "nbd-client -d", we can do an iocl(BLKRRPART) to update partition table: as the size of the device is 0, we don't have to serve the partition manager request (-> no deadlock). -------NOTE Signed-off-by: Paul Clements <paul.clements@steeleye.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-29 12:02:51 +04:00
return -EINVAL;
}
part_shift = 0;
if (max_part > 0) {
NBD: add partition support Permit the use of partitions with network block devices (NBD). A new parameter is introduced to define how many partition we want to be able to manage per network block device. This parameter is "max_part". For instance, to manage 63 partitions / loop device, we will do: [on the server side] # nbd-server 1234 /dev/sdb [on the client side] # modprobe nbd max_part=63 # ls -l /dev/nbd* brw-rw---- 1 root disk 43, 0 2008-03-25 11:14 /dev/nbd0 brw-rw---- 1 root disk 43, 64 2008-03-25 11:11 /dev/nbd1 brw-rw---- 1 root disk 43, 640 2008-03-25 11:11 /dev/nbd10 brw-rw---- 1 root disk 43, 704 2008-03-25 11:11 /dev/nbd11 brw-rw---- 1 root disk 43, 768 2008-03-25 11:11 /dev/nbd12 brw-rw---- 1 root disk 43, 832 2008-03-25 11:11 /dev/nbd13 brw-rw---- 1 root disk 43, 896 2008-03-25 11:11 /dev/nbd14 brw-rw---- 1 root disk 43, 960 2008-03-25 11:11 /dev/nbd15 brw-rw---- 1 root disk 43, 128 2008-03-25 11:11 /dev/nbd2 brw-rw---- 1 root disk 43, 192 2008-03-25 11:11 /dev/nbd3 brw-rw---- 1 root disk 43, 256 2008-03-25 11:11 /dev/nbd4 brw-rw---- 1 root disk 43, 320 2008-03-25 11:11 /dev/nbd5 brw-rw---- 1 root disk 43, 384 2008-03-25 11:11 /dev/nbd6 brw-rw---- 1 root disk 43, 448 2008-03-25 11:11 /dev/nbd7 brw-rw---- 1 root disk 43, 512 2008-03-25 11:11 /dev/nbd8 brw-rw---- 1 root disk 43, 576 2008-03-25 11:11 /dev/nbd9 # nbd-client localhost 1234 /dev/nbd0 Negotiation: ..size = 80418240KB bs=1024, sz=80418240 -------NOTE, RFC: partition table is not automatically read. The driver sets bdev->bd_invalidated to 1 to force the read of the partition table of the device, but this is done only on an open of the device. So we have to do a "touch /dev/nbdX" or something like that. It can't be done from the nbd-client or nbd driver because at this level we can't ask to read the partition table and to serve the request at the same time (-> deadlock) If someone has a better idea, I'm open to any suggestion. -------NOTE, RFC # fdisk -l /dev/nbd0 Disk /dev/nbd0: 82.3 GB, 82348277760 bytes 255 heads, 63 sectors/track, 10011 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/nbd0p1 * 1 9965 80043831 83 Linux /dev/nbd0p2 9966 10011 369495 5 Extended /dev/nbd0p5 9966 10011 369463+ 82 Linux swap / Solaris # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 brw-rw---- 1 root disk 43, 1 2008-03-25 11:16 /dev/nbd0p1 brw-rw---- 1 root disk 43, 2 2008-03-25 11:16 /dev/nbd0p2 brw-rw---- 1 root disk 43, 5 2008-03-25 11:16 /dev/nbd0p5 # mount /dev/nbd0p1 /mnt # ls /mnt bin dev initrd lost+found opt sbin sys var boot etc initrd.img media proc selinux tmp vmlinuz cdrom home lib mnt root srv usr # umount /mnt # nbd-client -d /dev/nbd0 # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 -------NOTE On "nbd-client -d", we can do an iocl(BLKRRPART) to update partition table: as the size of the device is 0, we don't have to serve the partition manager request (-> no deadlock). -------NOTE Signed-off-by: Paul Clements <paul.clements@steeleye.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-29 12:02:51 +04:00
part_shift = fls(max_part);
/*
* Adjust max_part according to part_shift as it is exported
* to user space so that user can know the max number of
* partition kernel should be able to manage.
*
* Note that -1 is required because partition 0 is reserved
* for the whole disk.
*/
max_part = (1UL << part_shift) - 1;
}
nbd: limit module parameters to a sane value The 'max_part' parameter controls the number of maximum partition a nbd device can have. However if a user specifies very large value it would exceed the limitation of device minor number and can cause a kernel oops (or, at least, produce invalid device nodes in some cases). In addition, specifying large 'nbds_max' value causes same problem for the same reason. On my desktop, following command results to the kernel bug: $ sudo modprobe nbd max_part=100000 kernel BUG at /media/Linux_Data/project/linux/fs/sysfs/group.c:65! invalid opcode: 0000 [#1] SMP last sysfs file: /sys/devices/virtual/block/nbd4/range CPU 1 Modules linked in: nbd(+) bridge stp llc kvm_intel kvm asus_atk0110 sg sr_mod cdrom Pid: 2522, comm: modprobe Tainted: G W 2.6.39-leonard+ #159 System manufacturer System Product Name/P5G41TD-M PRO RIP: 0010:[<ffffffff8115aa08>] [<ffffffff8115aa08>] internal_create_group+0x2f/0x166 RSP: 0018:ffff8801009f1de8 EFLAGS: 00010246 RAX: 00000000ffffffef RBX: ffff880103920478 RCX: 00000000000a7bd3 RDX: ffffffff81a2dbe0 RSI: 0000000000000000 RDI: ffff880103920478 RBP: ffff8801009f1e38 R08: ffff880103920468 R09: ffff880103920478 R10: ffff8801009f1de8 R11: ffff88011eccbb68 R12: ffffffff81a2dbe0 R13: ffff880103920468 R14: 0000000000000000 R15: ffff880103920400 FS: 00007f3c49de9700(0000) GS:ffff88011f800000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 00007f3b7fe7c000 CR3: 00000000cd58d000 CR4: 00000000000406e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process modprobe (pid: 2522, threadinfo ffff8801009f0000, task ffff8801009a93a0) Stack: ffff8801009f1e58 ffffffff812e8f6e ffff8801009f1e58 ffffffff812e7a80 ffff880000000010 ffff880103920400 ffff8801002fd0c0 ffff880103920468 0000000000000011 ffff880103920400 ffff8801009f1e48 ffffffff8115ab6a Call Trace: [<ffffffff812e8f6e>] ? device_add+0x4f1/0x5e4 [<ffffffff812e7a80>] ? dev_set_name+0x41/0x43 [<ffffffff8115ab6a>] sysfs_create_group+0x13/0x15 [<ffffffff810b857e>] blk_trace_init_sysfs+0x14/0x16 [<ffffffff811ee58b>] blk_register_queue+0x4c/0xfd [<ffffffff811f3bdf>] add_disk+0xe4/0x29c [<ffffffffa007e2ab>] nbd_init+0x2ab/0x30d [nbd] [<ffffffffa007e000>] ? 0xffffffffa007dfff [<ffffffff8100020f>] do_one_initcall+0x7f/0x13e [<ffffffff8107ab0a>] sys_init_module+0xa1/0x1e3 [<ffffffff814f3542>] system_call_fastpath+0x16/0x1b Code: 41 57 41 56 41 55 41 54 53 48 83 ec 28 0f 1f 44 00 00 48 89 fb 41 89 f6 49 89 d4 48 85 ff 74 0b 85 f6 75 0b 48 83 7f 30 00 75 14 <0f> 0b eb fe b9 ea ff ff ff 48 83 7f 30 00 0f 84 09 01 00 00 49 RIP [<ffffffff8115aa08>] internal_create_group+0x2f/0x166 RSP <ffff8801009f1de8> ---[ end trace 753285ffbf72c57c ]--- Signed-off-by: Namhyung Kim <namhyung@gmail.com> Cc: Laurent Vivier <Laurent.Vivier@bull.net> Cc: Paul Clements <Paul.Clements@steeleye.com> Cc: stable@kernel.org Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-05-28 16:44:46 +04:00
if ((1UL << part_shift) > DISK_MAX_PARTS)
return -EINVAL;
if (nbds_max > 1UL << (MINORBITS - part_shift))
return -EINVAL;
nbd_dev = kcalloc(nbds_max, sizeof(*nbd_dev), GFP_KERNEL);
if (!nbd_dev)
return -ENOMEM;
for (i = 0; i < nbds_max; i++) {
NBD: add partition support Permit the use of partitions with network block devices (NBD). A new parameter is introduced to define how many partition we want to be able to manage per network block device. This parameter is "max_part". For instance, to manage 63 partitions / loop device, we will do: [on the server side] # nbd-server 1234 /dev/sdb [on the client side] # modprobe nbd max_part=63 # ls -l /dev/nbd* brw-rw---- 1 root disk 43, 0 2008-03-25 11:14 /dev/nbd0 brw-rw---- 1 root disk 43, 64 2008-03-25 11:11 /dev/nbd1 brw-rw---- 1 root disk 43, 640 2008-03-25 11:11 /dev/nbd10 brw-rw---- 1 root disk 43, 704 2008-03-25 11:11 /dev/nbd11 brw-rw---- 1 root disk 43, 768 2008-03-25 11:11 /dev/nbd12 brw-rw---- 1 root disk 43, 832 2008-03-25 11:11 /dev/nbd13 brw-rw---- 1 root disk 43, 896 2008-03-25 11:11 /dev/nbd14 brw-rw---- 1 root disk 43, 960 2008-03-25 11:11 /dev/nbd15 brw-rw---- 1 root disk 43, 128 2008-03-25 11:11 /dev/nbd2 brw-rw---- 1 root disk 43, 192 2008-03-25 11:11 /dev/nbd3 brw-rw---- 1 root disk 43, 256 2008-03-25 11:11 /dev/nbd4 brw-rw---- 1 root disk 43, 320 2008-03-25 11:11 /dev/nbd5 brw-rw---- 1 root disk 43, 384 2008-03-25 11:11 /dev/nbd6 brw-rw---- 1 root disk 43, 448 2008-03-25 11:11 /dev/nbd7 brw-rw---- 1 root disk 43, 512 2008-03-25 11:11 /dev/nbd8 brw-rw---- 1 root disk 43, 576 2008-03-25 11:11 /dev/nbd9 # nbd-client localhost 1234 /dev/nbd0 Negotiation: ..size = 80418240KB bs=1024, sz=80418240 -------NOTE, RFC: partition table is not automatically read. The driver sets bdev->bd_invalidated to 1 to force the read of the partition table of the device, but this is done only on an open of the device. So we have to do a "touch /dev/nbdX" or something like that. It can't be done from the nbd-client or nbd driver because at this level we can't ask to read the partition table and to serve the request at the same time (-> deadlock) If someone has a better idea, I'm open to any suggestion. -------NOTE, RFC # fdisk -l /dev/nbd0 Disk /dev/nbd0: 82.3 GB, 82348277760 bytes 255 heads, 63 sectors/track, 10011 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/nbd0p1 * 1 9965 80043831 83 Linux /dev/nbd0p2 9966 10011 369495 5 Extended /dev/nbd0p5 9966 10011 369463+ 82 Linux swap / Solaris # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 brw-rw---- 1 root disk 43, 1 2008-03-25 11:16 /dev/nbd0p1 brw-rw---- 1 root disk 43, 2 2008-03-25 11:16 /dev/nbd0p2 brw-rw---- 1 root disk 43, 5 2008-03-25 11:16 /dev/nbd0p5 # mount /dev/nbd0p1 /mnt # ls /mnt bin dev initrd lost+found opt sbin sys var boot etc initrd.img media proc selinux tmp vmlinuz cdrom home lib mnt root srv usr # umount /mnt # nbd-client -d /dev/nbd0 # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 -------NOTE On "nbd-client -d", we can do an iocl(BLKRRPART) to update partition table: as the size of the device is 0, we don't have to serve the partition manager request (-> no deadlock). -------NOTE Signed-off-by: Paul Clements <paul.clements@steeleye.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-29 12:02:51 +04:00
struct gendisk *disk = alloc_disk(1 << part_shift);
if (!disk)
goto out;
nbd_dev[i].disk = disk;
/*
* The new linux 2.5 block layer implementation requires
* every gendisk to have its very own request_queue struct.
* These structs are big so we dynamically allocate them.
*/
disk->queue = blk_init_queue(nbd_request_handler, &nbd_lock);
if (!disk->queue) {
put_disk(disk);
goto out;
}
/*
* Tell the block layer that we are not a rotational device
*/
queue_flag_set_unlocked(QUEUE_FLAG_NONROT, disk->queue);
queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, disk->queue);
disk->queue->limits.discard_granularity = 512;
blk_queue_max_discard_sectors(disk->queue, UINT_MAX);
disk->queue->limits.discard_zeroes_data = 0;
blk_queue_max_hw_sectors(disk->queue, 65536);
disk->queue->limits.max_sectors = 256;
}
if (register_blkdev(NBD_MAJOR, "nbd")) {
err = -EIO;
goto out;
}
printk(KERN_INFO "nbd: registered device at major %d\n", NBD_MAJOR);
nbd_dbg_init();
for (i = 0; i < nbds_max; i++) {
struct gendisk *disk = nbd_dev[i].disk;
nbd_dev[i].magic = NBD_MAGIC;
INIT_LIST_HEAD(&nbd_dev[i].waiting_queue);
spin_lock_init(&nbd_dev[i].queue_lock);
spin_lock_init(&nbd_dev[i].sock_lock);
INIT_LIST_HEAD(&nbd_dev[i].queue_head);
mutex_init(&nbd_dev[i].tx_lock);
nbd: Fix timeout detection At the moment the nbd timeout just detects hanging tcp operations. This is not enough to detect a hanging or bad connection as expected of a timeout. This patch redesigns the timeout detection to include some more cases. The timeout is now in relation to replies from the server. If the server does not send replies within the timeout the connection will be shut down. The patch adds a continous timer 'timeout_timer' that is setup in one of two cases: - The request list is empty and we are sending the first request out to the server. We want to have a reply within the given timeout, otherwise we consider the connection to be dead. - A server response was received. This means the server is still communicating with us. The timer is reset to the timeout value. The timer is not stopped if the list becomes empty. It will just trigger a timeout which will directly leave the handling routine again as the request list is empty. The whole patch does not use any additional explicit locking. The list_empty() calls are safe to be used concurrently. The timer is locked internally as we just use mod_timer and del_timer_sync(). The patch is based on the idea of Michal Belczyk with a previous different implementation. Cc: Michal Belczyk <belczyk@bsd.krakow.pl> Cc: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Markus Pargmann <mpa@pengutronix.de> Tested-by: Hermann Lauer <Hermann.Lauer@iwr.uni-heidelberg.de> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-17 09:20:00 +03:00
init_timer(&nbd_dev[i].timeout_timer);
nbd_dev[i].timeout_timer.function = nbd_xmit_timeout;
nbd_dev[i].timeout_timer.data = (unsigned long)&nbd_dev[i];
[PATCH] nbd: fix TX/RX race condition Janos Haar of First NetCenter Bt. reported numerous crashes involving the NBD driver. With his help, this was tracked down to bogus bio vectors which in turn was the result of a race condition between the receive/transmit routines in the NBD driver. The bug manifests itself like this: CPU0 CPU1 do_nbd_request add req to queuelist nbd_send_request send req head for each bio kmap send nbd_read_stat nbd_find_request nbd_end_request kunmap When CPU1 finishes nbd_end_request, the request and all its associated bio's are freed. So when CPU0 calls kunmap whose argument is derived from the last bio, it may crash. Under normal circumstances, the race occurs only on the last bio. However, if an error is encountered on the remote NBD server (such as an incorrect magic number in the request), or if there were a bug in the server, it is possible for the nbd_end_request to occur any time after the request's addition to the queuelist. The following patch fixes this problem by making sure that requests are not added to the queuelist until after they have been completed transmission. In order for the receiving side to be ready for responses involving requests still being transmitted, the patch introduces the concept of the active request. When a response matches the current active request, its processing is delayed until after the tranmission has come to a stop. This has been tested by Janos and it has been successful in curing this race condition. From: Herbert Xu <herbert@gondor.apana.org.au> Here is an updated patch which removes the active_req wait in nbd_clear_queue and the associated memory barrier. I've also clarified this in the comment. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Cc: <djani22@dynamicweb.hu> Cc: Paul Clements <Paul.Clements@SteelEye.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-06 11:09:47 +03:00
init_waitqueue_head(&nbd_dev[i].active_wq);
init_waitqueue_head(&nbd_dev[i].waiting_wq);
nbd_dev[i].blksize = 1024;
nbd_dev[i].bytesize = 0;
disk->major = NBD_MAJOR;
NBD: add partition support Permit the use of partitions with network block devices (NBD). A new parameter is introduced to define how many partition we want to be able to manage per network block device. This parameter is "max_part". For instance, to manage 63 partitions / loop device, we will do: [on the server side] # nbd-server 1234 /dev/sdb [on the client side] # modprobe nbd max_part=63 # ls -l /dev/nbd* brw-rw---- 1 root disk 43, 0 2008-03-25 11:14 /dev/nbd0 brw-rw---- 1 root disk 43, 64 2008-03-25 11:11 /dev/nbd1 brw-rw---- 1 root disk 43, 640 2008-03-25 11:11 /dev/nbd10 brw-rw---- 1 root disk 43, 704 2008-03-25 11:11 /dev/nbd11 brw-rw---- 1 root disk 43, 768 2008-03-25 11:11 /dev/nbd12 brw-rw---- 1 root disk 43, 832 2008-03-25 11:11 /dev/nbd13 brw-rw---- 1 root disk 43, 896 2008-03-25 11:11 /dev/nbd14 brw-rw---- 1 root disk 43, 960 2008-03-25 11:11 /dev/nbd15 brw-rw---- 1 root disk 43, 128 2008-03-25 11:11 /dev/nbd2 brw-rw---- 1 root disk 43, 192 2008-03-25 11:11 /dev/nbd3 brw-rw---- 1 root disk 43, 256 2008-03-25 11:11 /dev/nbd4 brw-rw---- 1 root disk 43, 320 2008-03-25 11:11 /dev/nbd5 brw-rw---- 1 root disk 43, 384 2008-03-25 11:11 /dev/nbd6 brw-rw---- 1 root disk 43, 448 2008-03-25 11:11 /dev/nbd7 brw-rw---- 1 root disk 43, 512 2008-03-25 11:11 /dev/nbd8 brw-rw---- 1 root disk 43, 576 2008-03-25 11:11 /dev/nbd9 # nbd-client localhost 1234 /dev/nbd0 Negotiation: ..size = 80418240KB bs=1024, sz=80418240 -------NOTE, RFC: partition table is not automatically read. The driver sets bdev->bd_invalidated to 1 to force the read of the partition table of the device, but this is done only on an open of the device. So we have to do a "touch /dev/nbdX" or something like that. It can't be done from the nbd-client or nbd driver because at this level we can't ask to read the partition table and to serve the request at the same time (-> deadlock) If someone has a better idea, I'm open to any suggestion. -------NOTE, RFC # fdisk -l /dev/nbd0 Disk /dev/nbd0: 82.3 GB, 82348277760 bytes 255 heads, 63 sectors/track, 10011 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/nbd0p1 * 1 9965 80043831 83 Linux /dev/nbd0p2 9966 10011 369495 5 Extended /dev/nbd0p5 9966 10011 369463+ 82 Linux swap / Solaris # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 brw-rw---- 1 root disk 43, 1 2008-03-25 11:16 /dev/nbd0p1 brw-rw---- 1 root disk 43, 2 2008-03-25 11:16 /dev/nbd0p2 brw-rw---- 1 root disk 43, 5 2008-03-25 11:16 /dev/nbd0p5 # mount /dev/nbd0p1 /mnt # ls /mnt bin dev initrd lost+found opt sbin sys var boot etc initrd.img media proc selinux tmp vmlinuz cdrom home lib mnt root srv usr # umount /mnt # nbd-client -d /dev/nbd0 # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 -------NOTE On "nbd-client -d", we can do an iocl(BLKRRPART) to update partition table: as the size of the device is 0, we don't have to serve the partition manager request (-> no deadlock). -------NOTE Signed-off-by: Paul Clements <paul.clements@steeleye.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-29 12:02:51 +04:00
disk->first_minor = i << part_shift;
disk->fops = &nbd_fops;
disk->private_data = &nbd_dev[i];
sprintf(disk->disk_name, "nbd%d", i);
set_capacity(disk, 0);
add_disk(disk);
}
return 0;
out:
while (i--) {
blk_cleanup_queue(nbd_dev[i].disk->queue);
put_disk(nbd_dev[i].disk);
}
kfree(nbd_dev);
return err;
}
static void __exit nbd_cleanup(void)
{
int i;
nbd_dbg_close();
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);
}
}
unregister_blkdev(NBD_MAJOR, "nbd");
kfree(nbd_dev);
printk(KERN_INFO "nbd: unregistered device at major %d\n", NBD_MAJOR);
}
module_init(nbd_init);
module_exit(nbd_cleanup);
MODULE_DESCRIPTION("Network Block Device");
MODULE_LICENSE("GPL");
module_param(nbds_max, int, 0444);
NBD: add partition support Permit the use of partitions with network block devices (NBD). A new parameter is introduced to define how many partition we want to be able to manage per network block device. This parameter is "max_part". For instance, to manage 63 partitions / loop device, we will do: [on the server side] # nbd-server 1234 /dev/sdb [on the client side] # modprobe nbd max_part=63 # ls -l /dev/nbd* brw-rw---- 1 root disk 43, 0 2008-03-25 11:14 /dev/nbd0 brw-rw---- 1 root disk 43, 64 2008-03-25 11:11 /dev/nbd1 brw-rw---- 1 root disk 43, 640 2008-03-25 11:11 /dev/nbd10 brw-rw---- 1 root disk 43, 704 2008-03-25 11:11 /dev/nbd11 brw-rw---- 1 root disk 43, 768 2008-03-25 11:11 /dev/nbd12 brw-rw---- 1 root disk 43, 832 2008-03-25 11:11 /dev/nbd13 brw-rw---- 1 root disk 43, 896 2008-03-25 11:11 /dev/nbd14 brw-rw---- 1 root disk 43, 960 2008-03-25 11:11 /dev/nbd15 brw-rw---- 1 root disk 43, 128 2008-03-25 11:11 /dev/nbd2 brw-rw---- 1 root disk 43, 192 2008-03-25 11:11 /dev/nbd3 brw-rw---- 1 root disk 43, 256 2008-03-25 11:11 /dev/nbd4 brw-rw---- 1 root disk 43, 320 2008-03-25 11:11 /dev/nbd5 brw-rw---- 1 root disk 43, 384 2008-03-25 11:11 /dev/nbd6 brw-rw---- 1 root disk 43, 448 2008-03-25 11:11 /dev/nbd7 brw-rw---- 1 root disk 43, 512 2008-03-25 11:11 /dev/nbd8 brw-rw---- 1 root disk 43, 576 2008-03-25 11:11 /dev/nbd9 # nbd-client localhost 1234 /dev/nbd0 Negotiation: ..size = 80418240KB bs=1024, sz=80418240 -------NOTE, RFC: partition table is not automatically read. The driver sets bdev->bd_invalidated to 1 to force the read of the partition table of the device, but this is done only on an open of the device. So we have to do a "touch /dev/nbdX" or something like that. It can't be done from the nbd-client or nbd driver because at this level we can't ask to read the partition table and to serve the request at the same time (-> deadlock) If someone has a better idea, I'm open to any suggestion. -------NOTE, RFC # fdisk -l /dev/nbd0 Disk /dev/nbd0: 82.3 GB, 82348277760 bytes 255 heads, 63 sectors/track, 10011 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/nbd0p1 * 1 9965 80043831 83 Linux /dev/nbd0p2 9966 10011 369495 5 Extended /dev/nbd0p5 9966 10011 369463+ 82 Linux swap / Solaris # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 brw-rw---- 1 root disk 43, 1 2008-03-25 11:16 /dev/nbd0p1 brw-rw---- 1 root disk 43, 2 2008-03-25 11:16 /dev/nbd0p2 brw-rw---- 1 root disk 43, 5 2008-03-25 11:16 /dev/nbd0p5 # mount /dev/nbd0p1 /mnt # ls /mnt bin dev initrd lost+found opt sbin sys var boot etc initrd.img media proc selinux tmp vmlinuz cdrom home lib mnt root srv usr # umount /mnt # nbd-client -d /dev/nbd0 # ls -l /dev/nbd0* brw-rw---- 1 root disk 43, 0 2008-03-25 11:16 /dev/nbd0 -------NOTE On "nbd-client -d", we can do an iocl(BLKRRPART) to update partition table: as the size of the device is 0, we don't have to serve the partition manager request (-> no deadlock). -------NOTE Signed-off-by: Paul Clements <paul.clements@steeleye.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-29 12:02:51 +04:00
MODULE_PARM_DESC(nbds_max, "number of network block devices to initialize (default: 16)");
module_param(max_part, int, 0444);
MODULE_PARM_DESC(max_part, "number of partitions per device (default: 0)");