| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <12282462542094-git-send-email-zbr@ioremap.net>
Date: Tue, 2 Dec 2008 22:30:52 +0300
From: Evgeniy Polyakov <zbr@...emap.net>
To: linux-kernel@...r.kernel.org
Cc: netdev@...r.kernel.org, linux-fsdevel@...r.kernel.org,
Evgeniy Polyakov <zbr@...emap.net>
Subject: [3/5] POHMELFS: transactions and network.
POHMELFS uses transactions to maintain coherent IO on the server.
Each transaction can include multiple different commands, so they all
will be completed (possibly with error) as a single entity. When server
goes down, transaction tree is scanned and transactions are resent according
to its last-sent time (scanning interval can be specified via mount option,
more details in Documentation/filesystems/pohmelfs), after predefined number
of failed resends (specified via mount option) transaction will be completed
with error. Each transaction user may specify completion callback which will
be invoked with transaction data and error status.
Network socket is handled by dedicated thread, which receives data and invokes
appropriate callbacks, or submits it for crypto processing into pool of crypto
threads.
Signed-off-by: Evgeniy Polyakov <zbr@...emap.net>
diff --git a/fs/pohmelfs/net.c b/fs/pohmelfs/net.c
new file mode 100644
index 0000000..87ca7da
--- /dev/null
+++ b/fs/pohmelfs/net.c
@@ -0,0 +1,1180 @@
+/*
+ * 2007+ Copyright (c) Evgeniy Polyakov <johnpol@....mipt.ru>
+ * All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#include <linux/fsnotify.h>
+#include <linux/jhash.h>
+#include <linux/in.h>
+#include <linux/in6.h>
+#include <linux/kthread.h>
+#include <linux/pagemap.h>
+#include <linux/poll.h>
+#include <linux/swap.h>
+#include <linux/syscalls.h>
+
+#include "netfs.h"
+
+/*
+ * Async machinery lives here.
+ * All commands being sent to server do _not_ require sync reply,
+ * instead, if it is really needed, like readdir or readpage, caller
+ * sleeps waiting for data, which will be placed into provided buffer
+ * and caller will be awakened.
+ *
+ * Every command response can come without some listener. For example
+ * readdir response will add new objects into cache without appropriate
+ * request from userspace. This is used in cache coherency.
+ *
+ * If object is not found for given data, it is discarded.
+ *
+ * All requests are received by dedicated kernel thread.
+ */
+
+/*
+ * Basic network sending/receiving functions.
+ * Blocked mode is used.
+ */
+static int netfs_data_recv(struct netfs_state *st, void *buf, u64 size)
+{
+ struct msghdr msg;
+ struct kvec iov;
+ int err;
+
+ BUG_ON(!size);
+
+ iov.iov_base = buf;
+ iov.iov_len = size;
+
+ msg.msg_iov = (struct iovec *)&iov;
+ msg.msg_iovlen = 1;
+ msg.msg_name = NULL;
+ msg.msg_namelen = 0;
+ msg.msg_control = NULL;
+ msg.msg_controllen = 0;
+ msg.msg_flags = MSG_DONTWAIT;
+
+ err = kernel_recvmsg(st->socket, &msg, &iov, 1, iov.iov_len,
+ msg.msg_flags);
+ if (err <= 0) {
+ printk("%s: failed to recv data: size: %llu, err: %d.\n", __func__, size, err);
+ if (err == 0)
+ err = -ECONNRESET;
+
+ netfs_state_exit(st);
+ }
+
+ return err;
+}
+
+static int pohmelfs_data_recv(struct netfs_state *st, void *data, unsigned int size)
+{
+ unsigned int revents = 0;
+ unsigned int err_mask = POLLERR | POLLHUP | POLLRDHUP;
+ unsigned int mask = err_mask | POLLIN;
+ int err = 0;
+
+ while (size && !err) {
+ revents = netfs_state_poll(st);
+
+ if (!(revents & mask)) {
+ DEFINE_WAIT(wait);
+
+ for (;;) {
+ prepare_to_wait(&st->thread_wait, &wait, TASK_INTERRUPTIBLE);
+ if (kthread_should_stop())
+ break;
+
+ revents = netfs_state_poll(st);
+
+ if (revents & mask)
+ break;
+
+ if (signal_pending(current))
+ break;
+
+ schedule();
+ continue;
+ }
+ finish_wait(&st->thread_wait, &wait);
+ }
+
+ err = -ECONNRESET;
+ netfs_state_lock(st);
+
+ if (st->socket && (st->read_socket == st->socket) && (revents & POLLIN)) {
+ err = netfs_data_recv(st, data, size);
+ if (err > 0) {
+ data += err;
+ size -= err;
+ err = 0;
+ }
+ }
+
+ if (revents & err_mask) {
+ printk("%s: revents: %x, socket: %p, size: %u, err: %d.\n",
+ __func__, revents, st->socket, size, err);
+ netfs_state_exit(st);
+ err = -ECONNRESET;
+ }
+
+ if (!st->socket) {
+ err = netfs_state_init(st);
+ if (!err)
+ err = -EAGAIN;
+ }
+
+ netfs_state_unlock(st);
+
+ if (kthread_should_stop())
+ err = -ENODEV;
+
+ if (err)
+ printk("%s: socket: %p, read_socket: %p, revents: %x, rev_error: %d, "
+ "should_stop: %d, size: %u, err: %d.\n",
+ __func__, st->socket, st->read_socket,
+ revents, revents & err_mask, kthread_should_stop(), size, err);
+ }
+
+ return err;
+}
+
+int pohmelfs_data_recv_and_check(struct netfs_state *st, void *data, unsigned int size)
+{
+ struct netfs_cmd *cmd = &st->cmd;
+ int err;
+
+ err = pohmelfs_data_recv(st, data, size);
+ if (err)
+ return err;
+
+ return pohmelfs_crypto_process_input_data(&st->eng, cmd->iv, data, NULL, size);
+}
+
+/*
+ * Polling machinery.
+ */
+
+struct netfs_poll_helper
+{
+ poll_table pt;
+ struct netfs_state *st;
+};
+
+static int netfs_queue_wake(wait_queue_t *wait, unsigned mode, int sync, void *key)
+{
+ struct netfs_state *st = container_of(wait, struct netfs_state, wait);
+
+ wake_up(&st->thread_wait);
+ return 1;
+}
+
+static void netfs_queue_func(struct file *file, wait_queue_head_t *whead,
+ poll_table *pt)
+{
+ struct netfs_state *st = container_of(pt, struct netfs_poll_helper, pt)->st;
+
+ st->whead = whead;
+ init_waitqueue_func_entry(&st->wait, netfs_queue_wake);
+ add_wait_queue(whead, &st->wait);
+}
+
+static void netfs_poll_exit(struct netfs_state *st)
+{
+ if (st->whead) {
+ remove_wait_queue(st->whead, &st->wait);
+ st->whead = NULL;
+ }
+}
+
+static int netfs_poll_init(struct netfs_state *st)
+{
+ struct netfs_poll_helper ph;
+
+ ph.st = st;
+ init_poll_funcptr(&ph.pt, &netfs_queue_func);
+
+ st->socket->ops->poll(NULL, st->socket, &ph.pt);
+ return 0;
+}
+
+/*
+ * Get response for readpage command. We search inode and page in its mapping
+ * and copy data into. If it was async request, then we queue page into shared
+ * data and wakeup listener, who will copy it to userspace.
+ *
+ * There is a work in progress of allowing to call copy_to_user() directly from
+ * async receiving kernel thread.
+ */
+static int pohmelfs_read_page_response(struct netfs_state *st)
+{
+ struct pohmelfs_sb *psb = st->psb;
+ struct netfs_cmd *cmd = &st->cmd;
+ struct inode *inode;
+ struct page *page;
+ int err = 0;
+
+ if (cmd->size > PAGE_CACHE_SIZE) {
+ err = -EINVAL;
+ goto err_out_exit;
+ }
+
+ inode = ilookup(st->psb->sb, cmd->id);
+ if (!inode) {
+ printk("%s: failed to find inode: id: %llu.\n", __func__, cmd->id);
+ err = -ENOENT;
+ goto err_out_exit;
+ }
+
+ page = find_get_page(inode->i_mapping, cmd->start >> PAGE_CACHE_SHIFT);
+ if (!page || !PageLocked(page)) {
+ printk("%s: failed to find/lock page: page: %p, id: %llu, start: %llu, index: %llu.\n",
+ __func__, page, cmd->id, cmd->start, cmd->start >> PAGE_CACHE_SHIFT);
+
+ while (cmd->size) {
+ unsigned int sz = min(cmd->size, st->size);
+
+ err = pohmelfs_data_recv(st, st->data, sz);
+ if (err)
+ break;
+
+ cmd->size -= sz;
+ }
+
+ err = -ENODEV;
+ if (page)
+ goto err_out_page_put;
+ goto err_out_put;
+ }
+
+ if (cmd->size) {
+ void *addr;
+
+ addr = kmap(page);
+ err = pohmelfs_data_recv(st, addr, cmd->size);
+ kunmap(page);
+
+ if (err)
+ goto err_out_page_unlock;
+ }
+
+ dprintk("%s: page: %p, start: %llu, size: %u, locked: %d.\n",
+ __func__, page, cmd->start, cmd->size, PageLocked(page));
+
+ SetPageChecked(page);
+ if ((psb->hash_string || psb->cipher_string) && psb->perform_crypto && cmd->size) {
+ err = pohmelfs_crypto_process_input_page(&st->eng, page, cmd->size, cmd->iv);
+ if (err < 0)
+ goto err_out_page_unlock;
+ } else {
+ SetPageUptodate(page);
+ unlock_page(page);
+ page_cache_release(page);
+ }
+
+ pohmelfs_put_inode(POHMELFS_I(inode));
+ wake_up(&st->psb->wait);
+
+ return 0;
+
+err_out_page_unlock:
+ SetPageError(page);
+ unlock_page(page);
+err_out_page_put:
+ page_cache_release(page);
+err_out_put:
+ pohmelfs_put_inode(POHMELFS_I(inode));
+err_out_exit:
+ wake_up(&st->psb->wait);
+ return err;
+}
+
+static int pohmelfs_check_name(struct pohmelfs_inode *parent, struct qstr *str,
+ struct netfs_inode_info *info)
+{
+ struct inode *inode;
+ struct pohmelfs_name *n;
+ int err = 0;
+ u64 ino = 0;
+
+ mutex_lock(&parent->offset_lock);
+ n = pohmelfs_search_hash(parent, str->hash);
+ if (n)
+ ino = n->ino;
+ mutex_unlock(&parent->offset_lock);
+
+ if (!ino)
+ goto out;
+
+ inode = ilookup(parent->vfs_inode.i_sb, ino);
+ if (!inode)
+ goto out;
+
+ dprintk("%s: parent: %llu, inode: %llu.\n", __func__, parent->ino, ino);
+
+ pohmelfs_fill_inode(inode, info);
+ pohmelfs_put_inode(POHMELFS_I(inode));
+ err = -EEXIST;
+out:
+ return err;
+}
+
+/*
+ * Readdir response from server. If special field is set, we wakeup
+ * listener (readdir() call), which will copy data to userspace.
+ */
+static int pohmelfs_readdir_response(struct netfs_state *st)
+{
+ struct inode *inode;
+ struct netfs_cmd *cmd = &st->cmd;
+ struct netfs_inode_info *info;
+ struct pohmelfs_inode *parent = NULL, *npi;
+ int err = 0, last = cmd->ext;
+ struct qstr str;
+
+ if (cmd->size > st->size)
+ return -EINVAL;
+
+ inode = ilookup(st->psb->sb, cmd->id);
+ if (!inode)
+ return -ENOENT;
+ parent = POHMELFS_I(inode);
+
+ if (!cmd->size && cmd->start) {
+ err = -cmd->start;
+ goto out;
+ }
+
+ if (cmd->size) {
+ char *name;
+
+ err = pohmelfs_data_recv_and_check(st, st->data, cmd->size);
+ if (err)
+ goto err_out_put;
+
+ info = (struct netfs_inode_info *)(st->data);
+
+ name = (char *)(info + 1);
+ str.len = cmd->size - sizeof(struct netfs_inode_info) - 1 - cmd->cpad;
+ name[str.len] = 0;
+ str.name = name;
+ str.hash = jhash(str.name, str.len, 0);
+
+ netfs_convert_inode_info(info);
+
+ if (parent) {
+ err = pohmelfs_check_name(parent, &str, info);
+ if (err) {
+ if (err == -EEXIST)
+ err = 0;
+ goto out;
+ }
+ }
+
+ info->ino = cmd->start;
+ if (!info->ino)
+ info->ino = pohmelfs_new_ino(st->psb);
+
+ dprintk("%s: parent: %llu, ino: %llu, name: '%s', hash: %x, len: %u, mode: %o.\n",
+ __func__, parent->ino, info->ino, str.name, str.hash, str.len,
+ info->mode);
+
+ npi = pohmelfs_new_inode(st->psb, parent, &str, info, 0);
+ if (IS_ERR(npi)) {
+ err = PTR_ERR(npi);
+
+ if (err != -EEXIST)
+ goto err_out_put;
+ } else {
+ set_bit(NETFS_INODE_CREATED, &npi->state);
+ }
+ }
+out:
+ if (last) {
+ set_bit(NETFS_INODE_REMOTE_SYNCED, &parent->state);
+ wake_up(&st->psb->wait);
+ }
+ pohmelfs_put_inode(parent);
+
+ return err;
+
+err_out_put:
+ clear_bit(NETFS_INODE_REMOTE_SYNCED, &parent->state);
+ printk("%s: parent: %llu, ino: %llu, cmd_id: %llu.\n", __func__, parent->ino, cmd->start, cmd->id);
+ pohmelfs_put_inode(parent);
+ wake_up(&st->psb->wait);
+ return err;
+}
+
+/*
+ * Lookup command response.
+ * It searches for inode to be looked at (if it exists) and substitutes
+ * its inode information (size, permission, mode and so on), if inode does
+ * not exist, new one will be created and inserted into caches.
+ */
+static int pohmelfs_lookup_response(struct netfs_state *st)
+{
+ struct inode *inode = NULL;
+ struct netfs_cmd *cmd = &st->cmd;
+ struct netfs_inode_info *info;
+ struct pohmelfs_inode *parent = NULL, *npi;
+ int err = -EINVAL;
+ char *name;
+
+ inode = ilookup(st->psb->sb, cmd->id);
+ if (!inode) {
+ printk("%s: lookup response: id: %llu, start: %llu, size: %u.\n",
+ __func__, cmd->id, cmd->start, cmd->size);
+ err = -ENOENT;
+ goto err_out_exit;
+ }
+ parent = POHMELFS_I(inode);
+
+ if (!cmd->size) {
+ err = -cmd->start;
+ goto err_out_put;
+ }
+
+ if (cmd->size < sizeof(struct netfs_inode_info)) {
+ printk("%s: broken lookup response: id: %llu, start: %llu, size: %u.\n",
+ __func__, cmd->id, cmd->start, cmd->size);
+ err = -EINVAL;
+ goto err_out_put;
+ }
+
+ err = pohmelfs_data_recv_and_check(st, st->data, cmd->size);
+ if (err)
+ goto err_out_put;
+
+ info = (struct netfs_inode_info *)(st->data);
+ name = (char *)(info + 1);
+
+ netfs_convert_inode_info(info);
+
+ info->ino = cmd->start;
+ if (!info->ino)
+ info->ino = pohmelfs_new_ino(st->psb);
+
+ dprintk("%s: parent: %llu, ino: %llu, name: '%s', start: %llu.\n",
+ __func__, parent->ino, info->ino, name, cmd->start);
+
+ if (cmd->start)
+ npi = pohmelfs_new_inode(st->psb, parent, NULL, info, 0);
+ else {
+ struct qstr str;
+
+ str.name = name;
+ str.len = cmd->size - sizeof(struct netfs_inode_info) - 1 - cmd->cpad;
+ str.hash = jhash(name, str.len, 0);
+
+ npi = pohmelfs_new_inode(st->psb, parent, &str, info, 0);
+ }
+ if (IS_ERR(npi)) {
+ err = PTR_ERR(npi);
+
+ if (err != -EEXIST)
+ goto err_out_put;
+ } else {
+ set_bit(NETFS_INODE_CREATED, &npi->state);
+ }
+
+ clear_bit(NETFS_COMMAND_PENDING, &parent->state);
+ pohmelfs_put_inode(parent);
+
+ wake_up(&st->psb->wait);
+
+ return 0;
+
+err_out_put:
+ pohmelfs_put_inode(parent);
+err_out_exit:
+ clear_bit(NETFS_COMMAND_PENDING, &parent->state);
+ wake_up(&st->psb->wait);
+ printk("%s: inode: %p, id: %llu, start: %llu, size: %u, err: %d.\n",
+ __func__, inode, cmd->id, cmd->start, cmd->size, err);
+ return err;
+}
+
+/*
+ * Create response, just marks local inode as 'created', so that writeback
+ * for any of its children (or own) would not try to sync it again.
+ */
+static int pohmelfs_create_response(struct netfs_state *st)
+{
+ struct inode *inode;
+ struct netfs_cmd *cmd = &st->cmd;
+
+ inode = ilookup(st->psb->sb, cmd->id);
+ if (!inode) {
+ printk("%s: failed to find inode: id: %llu, start: %llu.\n",
+ __func__, cmd->id, cmd->start);
+ goto err_out_exit;
+ }
+
+ /*
+ * To lock or not to lock?
+ * We actually do not care if it races...
+ */
+ if (cmd->start)
+ make_bad_inode(inode);
+
+ set_bit(NETFS_INODE_CREATED, &POHMELFS_I(inode)->state);
+
+ pohmelfs_put_inode(POHMELFS_I(inode));
+
+ wake_up(&st->psb->wait);
+ return 0;
+
+err_out_exit:
+ wake_up(&st->psb->wait);
+ return -ENOENT;
+}
+
+/*
+ * Object remove response. Just says that remove request has been received.
+ * Used in cache coherency protocol.
+ */
+static int pohmelfs_remove_response(struct netfs_state *st)
+{
+ struct netfs_cmd *cmd = &st->cmd;
+ int err;
+
+ err = pohmelfs_data_recv_and_check(st, st->data, cmd->size);
+ if (err)
+ return err;
+
+ dprintk("%s: parent: %llu, path: '%s'.\n", __func__, cmd->id, (char *)st->data);
+
+ return 0;
+}
+
+/*
+ * Transaction reply processing.
+ *
+ * Find transaction based on its generation number, bump its reference counter,
+ * so that none could free it under us, drop from the trees and lists and
+ * drop reference counter. When it hits zero (when all destinations replied
+ * and all timeout handled by async scanning code), completion will be called
+ * and transaction will be freed.
+ */
+static int pohmelfs_transaction_response(struct netfs_state *st)
+{
+ struct netfs_trans_dst *dst;
+ struct netfs_trans *t = NULL;
+ struct netfs_cmd *cmd = &st->cmd;
+ short err = (signed)cmd->ext;
+
+ mutex_lock(&st->trans_lock);
+ dst = netfs_trans_search(st, cmd->start);
+ if (dst) {
+ netfs_trans_remove_nolock(dst, st);
+ t = dst->trans;
+ }
+ mutex_unlock(&st->trans_lock);
+
+ if (!t) {
+ printk("%s: failed to find transaction: start: %llu: id: %llu, size: %u, ext: %u.\n",
+ __func__, cmd->start, cmd->id, cmd->size, cmd->ext);
+ err = -EINVAL;
+ goto out;
+ }
+
+ t->result = err;
+ netfs_trans_drop_dst_nostate(dst);
+
+out:
+ wake_up(&st->psb->wait);
+ return err;
+}
+
+/*
+ * Inode metadata cache coherency message.
+ */
+static int pohmelfs_page_cache_response(struct netfs_state *st)
+{
+ struct netfs_cmd *cmd = &st->cmd;
+ struct inode *inode;
+
+ dprintk("%s: st: %p, id: %llu, start: %llu, size: %u.\n", __func__, st, cmd->id, cmd->start, cmd->size);
+
+ inode = ilookup(st->psb->sb, cmd->id);
+ if (!inode) {
+ printk("%s: failed to find inode: id: %llu.\n", __func__, cmd->id);
+ return -ENOENT;
+ }
+
+ set_bit(NETFS_INODE_NEED_FLUSH, &POHMELFS_I(inode)->state);
+ pohmelfs_put_inode(POHMELFS_I(inode));
+
+ return 0;
+}
+
+/*
+ * Root capabilities response: export statistics
+ * like used and available size, number of files and dirs,
+ * permissions.
+ */
+static int pohmelfs_root_cap_response(struct netfs_state *st)
+{
+ struct netfs_cmd *cmd = &st->cmd;
+ struct netfs_root_capabilities *cap;
+ struct pohmelfs_sb *psb = st->psb;
+
+ if (cmd->size != sizeof(struct netfs_root_capabilities)) {
+ psb->flags = EPROTO;
+ wake_up(&psb->wait);
+ return -EPROTO;
+ }
+
+ cap = st->data;
+
+ netfs_convert_root_capabilities(cap);
+
+ if (psb->total_size < cap->used + cap->avail)
+ psb->total_size = cap->used + cap->avail;
+ if (cap->avail)
+ psb->avail_size = cap->avail;
+ psb->state_flags = cap->flags;
+
+ if (psb->state_flags & POHMELFS_FLAGS_RO) {
+ psb->sb->s_flags |= MS_RDONLY;
+ printk(KERN_INFO "Mounting POHMELFS (%d) read-only.\n", psb->idx);
+ }
+
+ if (psb->state_flags & POHMELFS_FLAGS_XATTR)
+ printk(KERN_INFO "Mounting POHMELFS (%d) "
+ "with extended attributes support.\n", psb->idx);
+
+ if (atomic_read(&psb->total_inodes) <= 1)
+ atomic_long_set(&psb->total_inodes, cap->nr_files);
+
+ dprintk("%s: total: %llu, avail: %llu, flags: %llx, inodes: %llu.\n",
+ __func__, psb->total_size, psb->avail_size, psb->state_flags, cap->nr_files);
+
+ psb->flags = 0;
+ wake_up(&psb->wait);
+ return 0;
+}
+
+/*
+ * Crypto capabilities of the server, where it says that
+ * it supports or does not requested hash/cipher algorithms.
+ */
+static int pohmelfs_crypto_cap_response(struct netfs_state *st)
+{
+ struct netfs_cmd *cmd = &st->cmd;
+ struct netfs_crypto_capabilities *cap;
+ struct pohmelfs_sb *psb = st->psb;
+ int err = 0;
+
+ if (cmd->size != sizeof(struct netfs_crypto_capabilities)) {
+ psb->flags = EPROTO;
+ wake_up(&psb->wait);
+ return -EPROTO;
+ }
+
+ cap = st->data;
+
+ dprintk("%s: cipher '%s': %s, hash: '%s': %s.\n",
+ __func__,
+ psb->cipher_string, (cap->cipher_strlen)?"SUPPORTED":"NOT SUPPORTED",
+ psb->hash_string, (cap->hash_strlen)?"SUPPORTED":"NOT SUPPORTED");
+
+ if (!cap->hash_strlen) {
+ if (psb->hash_strlen && psb->crypto_fail_unsupported)
+ err = -ENOTSUPP;
+ psb->hash_strlen = 0;
+ kfree(psb->hash_string);
+ psb->hash_string = NULL;
+ }
+
+ if (!cap->cipher_strlen) {
+ if (psb->cipher_strlen && psb->crypto_fail_unsupported)
+ err = -ENOTSUPP;
+ psb->cipher_strlen = 0;
+ kfree(psb->cipher_string);
+ psb->cipher_string = NULL;
+ }
+
+ return err;
+}
+
+/*
+ * Capabilities handshake response.
+ */
+static int pohmelfs_capabilities_response(struct netfs_state *st)
+{
+ struct netfs_cmd *cmd = &st->cmd;
+ int err = 0;
+
+ err = pohmelfs_data_recv(st, st->data, cmd->size);
+ if (err)
+ return err;
+
+ switch (cmd->id) {
+ case POHMELFS_CRYPTO_CAPABILITIES:
+ return pohmelfs_crypto_cap_response(st);
+ case POHMELFS_ROOT_CAPABILITIES:
+ return pohmelfs_root_cap_response(st);
+ default:
+ break;
+ }
+ return -EINVAL;
+}
+
+/*
+ * Receiving extended attribute.
+ * Does not work properly if received size is more than requested one,
+ * it should not happen with current request/reply model though.
+ */
+static int pohmelfs_getxattr_response(struct netfs_state *st)
+{
+ struct pohmelfs_sb *psb = st->psb;
+ struct netfs_cmd *cmd = &st->cmd;
+ struct pohmelfs_mcache *m;
+ short error = (signed short)cmd->ext, err;
+ unsigned int sz, total_size;
+
+ m = pohmelfs_mcache_search(psb, cmd->id);
+
+ dprintk("%s: id: %llu, gen: %llu, err: %d.\n",
+ __func__, cmd->id, (m)?m->gen:0, error);
+
+ if (!m)
+ return -ENOENT;
+
+ if (cmd->size) {
+ sz = min_t(unsigned int, cmd->size, m->size);
+ err = pohmelfs_data_recv_and_check(st, m->data, sz);
+ if (err) {
+ error = err;
+ goto out;
+ }
+
+ m->size = sz;
+ total_size = cmd->size - sz;
+
+ while (total_size) {
+ sz = min(total_size, st->size);
+
+ err = pohmelfs_data_recv_and_check(st, st->data, sz);
+ if (err) {
+ error = err;
+ break;
+ }
+
+ total_size -= sz;
+ }
+ }
+
+out:
+ m->err = error;
+ complete(&m->complete);
+ pohmelfs_mcache_put(psb, m);
+
+ return error;
+}
+
+int pohmelfs_data_lock_response(struct netfs_state *st)
+{
+ struct pohmelfs_sb *psb = st->psb;
+ struct netfs_cmd *cmd = &st->cmd;
+ struct pohmelfs_mcache *m;
+ short err = (signed short)cmd->ext;
+ u64 id = cmd->id;
+
+ m = pohmelfs_mcache_search(psb, id);
+
+ dprintk("%s: id: %llu, gen: %llu, err: %d.\n",
+ __func__, cmd->id, (m)?m->gen:0, err);
+
+ if (!m) {
+ pohmelfs_data_recv(st, st->data, cmd->size);
+ return -ENOENT;
+ }
+
+ if (cmd->size)
+ err = pohmelfs_data_recv_and_check(st, &m->info, cmd->size);
+
+ m->err = err;
+ complete(&m->complete);
+ pohmelfs_mcache_put(psb, m);
+
+ return err;
+}
+
+static void __inline__ netfs_state_reset(struct netfs_state *st)
+{
+ netfs_state_lock(st);
+ netfs_state_exit(st);
+ netfs_state_init(st);
+ netfs_state_unlock(st);
+}
+
+/*
+ * Main receiving function, called from dedicated kernel thread.
+ */
+static int pohmelfs_recv(void *data)
+{
+ int err = -EINTR;
+ struct netfs_state *st = data;
+ struct netfs_cmd *cmd = &st->cmd;
+
+ while (!kthread_should_stop()) {
+ /*
+ * If socket will be reset after this statement, then
+ * pohmelfs_data_recv() will just fail and loop will
+ * start again, so it can be done without any locks.
+ *
+ * st->read_socket is needed to prevents state machine
+ * breaking between this data reading and subsequent one
+ * in protocol specific functions during connection reset.
+ * In case of reset we have to read next command and do
+ * not expect data for old command to magically appear in
+ * new connection.
+ */
+ st->read_socket = st->socket;
+ err = pohmelfs_data_recv(st, cmd, sizeof(struct netfs_cmd));
+ if (err) {
+ msleep(1000);
+ continue;
+ }
+
+ netfs_convert_cmd(cmd);
+
+ dprintk("%s: cmd: %u, id: %llu, start: %llu, size: %u, "
+ "ext: %u, csize: %u, cpad: %u.\n",
+ __func__, cmd->cmd, cmd->id, cmd->start,
+ cmd->size, cmd->ext, cmd->csize, cmd->cpad);
+
+ if (cmd->csize) {
+ struct pohmelfs_crypto_engine *e = &st->eng;
+
+ if (unlikely(cmd->csize > e->size/2)) {
+ netfs_state_reset(st);
+ continue;
+ }
+
+ if (e->hash && unlikely(cmd->csize != st->psb->crypto_attached_size)) {
+ dprintk("%s: cmd: cmd: %u, id: %llu, start: %llu, size: %u, "
+ "csize: %u != digest size %u.\n",
+ __func__, cmd->cmd, cmd->id, cmd->start, cmd->size,
+ cmd->csize, st->psb->crypto_attached_size);
+ netfs_state_reset(st);
+ continue;
+ }
+
+ err = pohmelfs_data_recv(st, e->data, cmd->csize);
+ if (err) {
+ netfs_state_reset(st);
+ continue;
+ }
+
+#ifdef CONFIG_POHMELFS_DEBUG
+ {
+ unsigned int i;
+ unsigned char *hash = e->data;
+
+ dprintk("%s: received hash: ", __func__);
+ for (i=0; i<cmd->csize; ++i) {
+ printk("%02x ", hash[i]);
+ }
+ printk("\n");
+ }
+#endif
+ cmd->size -= cmd->csize;
+ }
+
+ /*
+ * This should catch protocol breakage and random garbage instead of commands.
+ */
+ if (unlikely((cmd->size > st->size) && (cmd->cmd != NETFS_XATTR_GET))) {
+ netfs_state_reset(st);
+ continue;
+ }
+
+ switch (cmd->cmd) {
+ case NETFS_READ_PAGE:
+ err = pohmelfs_read_page_response(st);
+ break;
+ case NETFS_READDIR:
+ err = pohmelfs_readdir_response(st);
+ break;
+ case NETFS_LOOKUP:
+ err = pohmelfs_lookup_response(st);
+ break;
+ case NETFS_CREATE:
+ err = pohmelfs_create_response(st);
+ break;
+ case NETFS_REMOVE:
+ err = pohmelfs_remove_response(st);
+ break;
+ case NETFS_TRANS:
+ err = pohmelfs_transaction_response(st);
+ break;
+ case NETFS_PAGE_CACHE:
+ err = pohmelfs_page_cache_response(st);
+ break;
+ case NETFS_CAPABILITIES:
+ err = pohmelfs_capabilities_response(st);
+ break;
+ case NETFS_LOCK:
+ err = pohmelfs_data_lock_response(st);
+ break;
+ case NETFS_XATTR_GET:
+ err = pohmelfs_getxattr_response(st);
+ break;
+ default:
+ printk("%s: wrong cmd: %u, id: %llu, start: %llu, size: %u, ext: %u.\n",
+ __func__, cmd->cmd, cmd->id, cmd->start, cmd->size, cmd->ext);
+ netfs_state_lock(st);
+ netfs_state_exit(st);
+ netfs_state_init(st);
+ netfs_state_unlock(st);
+ break;
+ }
+ }
+
+ while (!kthread_should_stop())
+ schedule_timeout_uninterruptible(msecs_to_jiffies(10));
+
+ return err;
+}
+
+int netfs_state_init(struct netfs_state *st)
+{
+ int err;
+ struct pohmelfs_ctl *ctl = &st->ctl;
+
+ err = sock_create(ctl->addr.sa_family, ctl->type, ctl->proto, &st->socket);
+ if (err)
+ goto err_out_exit;
+
+ st->socket->sk->sk_allocation = GFP_NOIO;
+ st->socket->sk->sk_sndtimeo = st->socket->sk->sk_rcvtimeo = msecs_to_jiffies(60000);
+
+ err = kernel_connect(st->socket, (struct sockaddr *)&ctl->addr, ctl->addrlen, 0);
+ if (err) {
+ printk("%s: failed to connect to server: idx: %u, err: %d.\n",
+ __func__, st->psb->idx, err);
+ goto err_out_release;
+ }
+ st->socket->sk->sk_sndtimeo = st->socket->sk->sk_rcvtimeo = msecs_to_jiffies(10000);
+
+ err = netfs_poll_init(st);
+ if (err)
+ goto err_out_release;
+
+ if (st->socket->ops->family == AF_INET) {
+ struct sockaddr_in *sin = (struct sockaddr_in *)&ctl->addr;
+ printk(KERN_INFO "%s: (re)connected to peer %u.%u.%u.%u:%d.\n", __func__,
+ NIPQUAD(sin->sin_addr.s_addr), ntohs(sin->sin_port));
+ } else if (st->socket->ops->family == AF_INET6) {
+ struct sockaddr_in6 *sin = (struct sockaddr_in6 *)&ctl->addr;
+ printk(KERN_INFO "%s: (re)connected to peer "
+ "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x:%d",
+ __func__, NIP6(sin->sin6_addr), ntohs(sin->sin6_port));
+ }
+
+ return 0;
+
+err_out_release:
+ sock_release(st->socket);
+err_out_exit:
+ st->socket = NULL;
+ return err;
+}
+
+void netfs_state_exit(struct netfs_state *st)
+{
+ if (st->socket) {
+ netfs_poll_exit(st);
+ st->socket->ops->shutdown(st->socket, 2);
+
+ if (st->socket->ops->family == AF_INET) {
+ struct sockaddr_in *sin = (struct sockaddr_in *)&st->ctl.addr;
+ printk("%s: disconnected from peer %u.%u.%u.%u:%d.\n", __func__,
+ NIPQUAD(sin->sin_addr.s_addr), ntohs(sin->sin_port));
+ } else if (st->socket->ops->family == AF_INET6) {
+ struct sockaddr_in6 *sin = (struct sockaddr_in6 *)&st->ctl.addr;
+ printk("%s: disconnected from peer "
+ "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x:%d",
+ __func__, NIP6(sin->sin6_addr), ntohs(sin->sin6_port));
+ }
+
+ sock_release(st->socket);
+ st->socket = NULL;
+ st->read_socket = NULL;
+ }
+}
+
+int pohmelfs_state_init_one(struct pohmelfs_sb *psb, struct pohmelfs_config *conf)
+{
+ struct netfs_state *st = &conf->state;
+ int err = -ENOMEM;
+
+ mutex_init(&st->__state_lock);
+ init_waitqueue_head(&st->thread_wait);
+
+ st->psb = psb;
+ st->trans_root = RB_ROOT;
+ mutex_init(&st->trans_lock);
+
+ st->size = psb->trans_data_size;
+ st->data = kmalloc(st->size, GFP_KERNEL);
+ if (!st->data)
+ goto err_out_exit;
+
+ if (psb->perform_crypto) {
+ err = pohmelfs_crypto_engine_init(&st->eng, psb);
+ if (err)
+ goto err_out_free_data;
+ }
+
+ err = netfs_state_init(st);
+ if (err)
+ goto err_out_free_engine;
+
+ st->thread = kthread_run(pohmelfs_recv, st, "pohmelfs/%u", psb->idx);
+ if (IS_ERR(st->thread)) {
+ err = PTR_ERR(st->thread);
+ goto err_out_netfs_exit;
+ }
+
+ if (!psb->active_state)
+ psb->active_state = conf;
+
+ dprintk("%s: conf: %p, st: %p, socket: %p.\n",
+ __func__, conf, st, st->socket);
+ return 0;
+
+err_out_netfs_exit:
+ netfs_state_exit(st);
+err_out_free_engine:
+ pohmelfs_crypto_engine_exit(&st->eng);
+err_out_free_data:
+ kfree(st->data);
+err_out_exit:
+ return err;
+
+}
+
+void pohmelfs_state_flush_transactions(struct netfs_state *st)
+{
+ struct rb_node *rb_node;
+ struct netfs_trans_dst *dst;
+
+ for (rb_node = rb_first(&st->trans_root); rb_node; ) {
+ dst = rb_entry(rb_node, struct netfs_trans_dst, state_entry);
+ rb_node = rb_next(rb_node);
+
+ dst->trans->result = -EINVAL;
+ netfs_trans_remove_nolock(dst, st);
+ netfs_trans_finish_send(dst->trans, st->psb);
+
+ netfs_trans_drop_dst_nostate(dst);
+ }
+}
+
+static void pohmelfs_state_exit_one(struct pohmelfs_config *c)
+{
+ struct netfs_state *st = &c->state;
+
+ dprintk("%s: exiting, st: %p.\n", __func__, st);
+ if (st->thread) {
+ kthread_stop(st->thread);
+ st->thread = NULL;
+ }
+
+ netfs_state_lock(st);
+ netfs_state_exit(st);
+ netfs_state_unlock(st);
+
+ pohmelfs_state_flush_transactions(st);
+
+ pohmelfs_crypto_engine_exit(&st->eng);
+ kfree(st->data);
+
+ kfree(c);
+}
+
+/*
+ * Initialize network stack. It searches for given ID in global
+ * configuration table, this contains information of the remote server
+ * (address (any supported by socket interface) and port, protocol and so on).
+ */
+int pohmelfs_state_init(struct pohmelfs_sb *psb)
+{
+ int err = -ENOMEM;
+
+ err = pohmelfs_copy_config(psb);
+ if (err) {
+ pohmelfs_state_exit(psb);
+ return err;
+ }
+
+ return 0;
+}
+
+void pohmelfs_state_exit(struct pohmelfs_sb *psb)
+{
+ struct pohmelfs_config *c, *tmp;
+
+ list_for_each_entry_safe(c, tmp, &psb->state_list, config_entry) {
+ list_del(&c->config_entry);
+ pohmelfs_state_exit_one(c);
+ }
+}
+
+void pohmelfs_switch_active(struct pohmelfs_sb *psb)
+{
+ struct pohmelfs_config *c = psb->active_state;
+
+ if (!list_empty(&psb->state_list)) {
+ if (c->config_entry.next != &psb->state_list) {
+ psb->active_state = list_entry(c->config_entry.next,
+ struct pohmelfs_config, config_entry);
+ } else {
+ psb->active_state = list_entry(psb->state_list.next,
+ struct pohmelfs_config, config_entry);
+ }
+
+ dprintk("%s: empty: %d, active %p -> %p.\n",
+ __func__, list_empty(&psb->state_list), c,
+ psb->active_state);
+ } else
+ psb->active_state = NULL;
+}
+
+void pohmelfs_check_states(struct pohmelfs_sb *psb)
+{
+ struct pohmelfs_config *c, *tmp;
+ LIST_HEAD(delete_list);
+
+ mutex_lock(&psb->state_lock);
+ list_for_each_entry_safe(c, tmp, &psb->state_list, config_entry) {
+ if (pohmelfs_config_check(c, psb->idx)) {
+
+ if (psb->active_state == c)
+ pohmelfs_switch_active(psb);
+ list_move(&c->config_entry, &delete_list);
+ }
+ }
+ pohmelfs_copy_config(psb);
+ mutex_unlock(&psb->state_lock);
+
+ list_for_each_entry_safe(c, tmp, &delete_list, config_entry) {
+ list_del(&c->config_entry);
+ pohmelfs_state_exit_one(c);
+ }
+}
diff --git a/fs/pohmelfs/netfs.h b/fs/pohmelfs/netfs.h
new file mode 100644
index 0000000..9b80c44
--- /dev/null
+++ b/fs/pohmelfs/netfs.h
@@ -0,0 +1,934 @@
+/*
+ * 2007+ Copyright (c) Evgeniy Polyakov <johnpol@....mipt.ru>
+ * All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#ifndef __NETFS_H
+#define __NETFS_H
+
+#include <linux/types.h>
+#include <linux/connector.h>
+
+#define POHMELFS_CN_IDX 5
+#define POHMELFS_CN_VAL 0
+
+#define POHMELFS_CTLINFO_ACK 1
+#define POHMELFS_NOINFO_ACK 2
+
+
+/*
+ * Network command structure.
+ * Will be extended.
+ */
+struct netfs_cmd
+{
+ __u16 cmd; /* Command number */
+ __u16 csize; /* Attached crypto information size */
+ __u16 cpad; /* Attached padding size */
+ __u16 ext; /* External flags */
+ __u32 size; /* Size of the attached data */
+ __u32 trans; /* Transaction id */
+ __u64 id; /* Object ID to operate on. Used for feedback.*/
+ __u64 start; /* Start of the object. */
+ __u64 iv; /* IV sequence */
+ __u8 data[0];
+};
+
+static inline void netfs_convert_cmd(struct netfs_cmd *cmd)
+{
+ cmd->id = __be64_to_cpu(cmd->id);
+ cmd->start = __be64_to_cpu(cmd->start);
+ cmd->iv = __be64_to_cpu(cmd->iv);
+ cmd->cmd = __be16_to_cpu(cmd->cmd);
+ cmd->ext = __be16_to_cpu(cmd->ext);
+ cmd->csize = __be16_to_cpu(cmd->csize);
+ cmd->cpad = __be16_to_cpu(cmd->cpad);
+ cmd->size = __be32_to_cpu(cmd->size);
+}
+
+#define NETFS_TRANS_SINGLE_DST (1<<0)
+
+enum {
+ NETFS_READDIR = 1, /* Read directory for given inode number */
+ NETFS_READ_PAGE, /* Read data page from the server */
+ NETFS_WRITE_PAGE, /* Write data page to the server */
+ NETFS_CREATE, /* Create directory entry */
+ NETFS_REMOVE, /* Remove directory entry */
+
+ NETFS_LOOKUP, /* Lookup single object */
+ NETFS_LINK, /* Create a link */
+ NETFS_TRANS, /* Transaction */
+ NETFS_OPEN, /* Open intent */
+ NETFS_INODE_INFO, /* Metadata cache coherency synchronization message */
+
+ NETFS_PAGE_CACHE, /* Page cache invalidation message */
+ NETFS_READ_PAGES, /* Read multiple contiguous pages in one go */
+ NETFS_RENAME, /* Rename object */
+ NETFS_CAPABILITIES, /* Capabilities of the client, for example supported crypto */
+ NETFS_LOCK, /* Distributed lock message */
+
+ NETFS_XATTR_SET, /* Set extended attribute */
+ NETFS_XATTR_GET, /* Get extended attribute */
+ NETFS_CMD_MAX
+};
+
+enum {
+ POHMELFS_FLAGS_ADD = 0, /* Network state control message for ADD */
+ POHMELFS_FLAGS_DEL, /* Network state control message for DEL */
+ POHMELFS_FLAGS_SHOW, /* Network state control message for SHOW */
+ POHMELFS_FLAGS_CRYPTO, /* Crypto data control message */
+};
+
+/*
+ * Always wanted to copy it from socket headers into public one,
+ * since they are __KERNEL__ protected there.
+ */
+#define _K_SS_MAXSIZE 128
+
+struct saddr
+{
+ unsigned short sa_family;
+ char addr[_K_SS_MAXSIZE];
+};
+
+enum {
+ POHMELFS_CRYPTO_HASH = 0,
+ POHMELFS_CRYPTO_CIPHER,
+};
+
+struct pohmelfs_crypto
+{
+ unsigned int idx; /* Config index */
+ unsigned short strlen; /* Size of the attached crypto string including 0-byte
+ * "cbc(aes)" for example */
+ unsigned short type; /* HMAC, cipher, both */
+ unsigned int keysize; /* Key size */
+ unsigned char data[0]; /* Algorithm string, key and IV */
+};
+
+/*
+ * Configuration command used to create table of different remote servers.
+ */
+struct pohmelfs_ctl
+{
+ unsigned int idx; /* Config index */
+ unsigned int type; /* Socket type */
+ unsigned int proto; /* Socket protocol */
+ unsigned int addrlen; /* Size of the address */
+ unsigned short unused; /* Align structure by 4 bytes */
+ struct saddr addr; /* Remote server address */
+};
+
+/*
+ * Ack for userspace about requested command.
+ */
+struct pohmelfs_cn_ack
+{
+ struct cn_msg msg;
+ int error;
+ int msg_num;
+ int unused[3];
+ struct pohmelfs_ctl ctl;
+};
+
+/*
+ * Inode info structure used to sync with server.
+ * Check what stat() returns.
+ */
+struct netfs_inode_info
+{
+ unsigned int mode;
+ unsigned int nlink;
+ unsigned int uid;
+ unsigned int gid;
+ unsigned int blocksize;
+ unsigned int padding;
+ __u64 ino;
+ __u64 blocks;
+ __u64 rdev;
+ __u64 size;
+ __u64 version;
+};
+
+static inline void netfs_convert_inode_info(struct netfs_inode_info *info)
+{
+ info->mode = __cpu_to_be32(info->mode);
+ info->nlink = __cpu_to_be32(info->nlink);
+ info->uid = __cpu_to_be32(info->uid);
+ info->gid = __cpu_to_be32(info->gid);
+ info->blocksize = __cpu_to_be32(info->blocksize);
+ info->blocks = __cpu_to_be64(info->blocks);
+ info->rdev = __cpu_to_be64(info->rdev);
+ info->size = __cpu_to_be64(info->size);
+ info->version = __cpu_to_be64(info->version);
+ info->ino = __cpu_to_be64(info->ino);
+}
+
+/*
+ * Cache state machine.
+ */
+enum {
+ NETFS_COMMAND_PENDING = 0, /* Command is being executed */
+ NETFS_INODE_CREATED, /* Inode was created locally */
+ NETFS_INODE_REMOTE_SYNCED, /* Inode was synced to server */
+ NETFS_INODE_OWNED, /* Inode is owned by given host */
+ NETFS_INODE_NEED_FLUSH, /* Inode has to be flushed to the server */
+};
+
+/*
+ * Path entry, used to create full path to object by single command.
+ */
+struct netfs_path_entry
+{
+ __u8 len; /* Data length, if less than 5 */
+ __u8 unused[5]; /* then data is embedded here */
+
+ __u16 mode; /* mode of the object (dir, file and so on) */
+
+ char data[];
+};
+
+static inline void netfs_convert_path_entry(struct netfs_path_entry *e)
+{
+ e->mode = __cpu_to_be16(e->mode);
+};
+
+/*
+ * POHMELFS capabilities: information about supported
+ * crypto operations (hash/cipher, modes, key sizes and so on),
+ * root informaion (used/available size, number of objects, permissions)
+ */
+enum pohmelfs_capabilities {
+ POHMELFS_CRYPTO_CAPABILITIES = 0,
+ POHMELFS_ROOT_CAPABILITIES,
+};
+
+/* Read-only mount */
+#define POHMELFS_FLAGS_RO (1<<0)
+/* Extended attributes support on/off */
+#define POHMELFS_FLAGS_XATTR (1<<1)
+
+struct netfs_root_capabilities
+{
+ __u64 nr_files;
+ __u64 used, avail;
+ __u64 flags;
+};
+
+static inline void netfs_convert_root_capabilities(struct netfs_root_capabilities *cap)
+{
+ cap->nr_files = __cpu_to_be64(cap->nr_files);
+ cap->used = __cpu_to_be64(cap->used);
+ cap->avail = __cpu_to_be64(cap->avail);
+ cap->flags = __cpu_to_be64(cap->flags);
+}
+
+struct netfs_crypto_capabilities
+{
+ unsigned short hash_strlen; /* Hash string length, like "hmac(sha1) including 0 byte "*/
+ unsigned short cipher_strlen; /* Cipher string length with the same format */
+ unsigned int cipher_keysize; /* Cipher key size */
+};
+
+static inline void netfs_convert_crypto_capabilities(struct netfs_crypto_capabilities *cap)
+{
+ cap->hash_strlen = __cpu_to_be16(cap->hash_strlen);
+ cap->cipher_strlen = __cpu_to_be16(cap->cipher_strlen);
+ cap->cipher_keysize = __cpu_to_be32(cap->cipher_keysize);
+}
+
+enum pohmelfs_lock_type {
+ POHMELFS_LOCK_GRAB = (1<<15),
+
+ POHMELFS_READ_LOCK = 0,
+ POHMELFS_WRITE_LOCK,
+};
+
+struct netfs_lock
+{
+ __u64 start;
+ __u64 ino;
+ __u32 size;
+ __u32 type;
+};
+
+static inline void netfs_convert_lock(struct netfs_lock *lock)
+{
+ lock->start = __cpu_to_be64(lock->start);
+ lock->ino = __cpu_to_be64(lock->ino);
+ lock->size = __cpu_to_be32(lock->size);
+ lock->type = __cpu_to_be32(lock->type);
+}
+
+#ifdef __KERNEL__
+
+#include <linux/kernel.h>
+#include <linux/completion.h>
+#include <linux/rbtree.h>
+#include <linux/net.h>
+#include <linux/poll.h>
+
+/*
+ * Private POHMELFS cache of objects in directory.
+ */
+struct pohmelfs_name
+{
+ struct rb_node hash_node;
+
+ struct list_head sync_del_entry, sync_create_entry;
+
+ u64 ino;
+
+ u32 hash;
+ u32 mode;
+ u32 len;
+
+ char *data;
+};
+
+/*
+ * POHMELFS inode. Main object.
+ */
+struct pohmelfs_inode
+{
+ struct list_head inode_entry; /* Entry in superblock list.
+ * Objects which are not bound to dentry require to be dropped
+ * in ->put_super()
+ */
+ struct rb_root hash_root; /* The same, but indexed by name hash and len */
+ struct mutex offset_lock; /* Protect both above trees */
+
+ struct list_head sync_del_list, sync_create_list; /* Sync list (create is not used).
+ * It contains children scheduled to be removed
+ */
+
+ unsigned int drop_count;
+
+ int lock_type; /* How this inode is locked: read or write */
+
+ int error; /* Transaction error for given inode */
+
+ long state; /* State machine above */
+
+ u64 ino; /* Inode number */
+ u64 total_len; /* Total length of all children names, used to create offsets */
+
+ struct inode vfs_inode;
+};
+
+struct netfs_trans;
+typedef int (* netfs_trans_complete_t)(struct page **pages, unsigned int page_num,
+ void *private, int err);
+
+struct netfs_state;
+struct pohmelfs_sb;
+
+struct netfs_trans
+{
+ /*
+ * Transaction header and attached contiguous data live here.
+ */
+ struct iovec iovec;
+
+ /*
+ * Pages attached to transaction.
+ */
+ struct page **pages;
+
+ /*
+ * List and protecting lock for transaction destination
+ * network states.
+ */
+ struct mutex dst_lock;
+ struct list_head dst_list;
+
+ /*
+ * Number of users for given transaction.
+ * For example each network state attached to transaction
+ * via dst_list increases it.
+ */
+ atomic_t refcnt;
+
+ /*
+ * Number of pages attached to given transaction.
+ * Some slots in above page array can be NULL, since
+ * for example page can be under writeback already,
+ * so we skip it in this transaction.
+ */
+ unsigned int page_num;
+
+ /*
+ * Transaction flags: single dst or broadcast and so on.
+ */
+ unsigned int flags;
+
+ /*
+ * Size of the data, which can be placed into
+ * iovec.iov_base area.
+ */
+ unsigned int total_size;
+
+ /*
+ * Number of pages to be sent to remote server.
+ * Usually equal to above page_num, but in case of partial
+ * writeback it can accumulate only pages already completed
+ * previous writeback.
+ */
+ unsigned int attached_pages;
+
+ /*
+ * Attached number of bytes in all above pages.
+ */
+ unsigned int attached_size;
+
+ /*
+ * Unique transacton generation number.
+ * Used as identity in the network state tree of transactions.
+ */
+ unsigned int gen;
+
+ /*
+ * Transaction completion status.
+ */
+ int result;
+
+ /*
+ * Superblock this transaction belongs to
+ */
+ struct pohmelfs_sb *psb;
+
+ /*
+ * Crypto engine, which processed this transaction.
+ * Can be not NULL only if crypto engine holds encrypted pages.
+ */
+ struct pohmelfs_crypto_engine *eng;
+
+ /* Private data */
+ void *private;
+
+ /* Completion callback, invoked just before transaction is destroyed */
+ netfs_trans_complete_t complete;
+};
+
+static inline int netfs_trans_cur_len(struct netfs_trans *t)
+{
+ return (signed)(t->total_size - t->iovec.iov_len);
+}
+
+static inline void *netfs_trans_current(struct netfs_trans *t)
+{
+ return t->iovec.iov_base + t->iovec.iov_len;
+}
+
+struct netfs_trans *netfs_trans_alloc(struct pohmelfs_sb *psb, unsigned int size,
+ unsigned int flags, unsigned int nr);
+void netfs_trans_free(struct netfs_trans *t);
+int netfs_trans_finish(struct netfs_trans *t, struct pohmelfs_sb *psb);
+int netfs_trans_finish_send(struct netfs_trans *t, struct pohmelfs_sb *psb);
+
+static inline void netfs_trans_reset(struct netfs_trans *t)
+{
+ t->complete = NULL;
+}
+
+struct netfs_trans_dst
+{
+ struct list_head trans_entry;
+ struct rb_node state_entry;
+
+ unsigned long send_time;
+
+ /*
+ * Times this transaction was resent to its old or new,
+ * depending on flags, destinations. When it reaches maximum
+ * allowed number, specified in superblock->trans_retries,
+ * transaction will be freed with ETIMEDOUT error.
+ */
+ unsigned int retries;
+
+ struct netfs_trans *trans;
+ struct netfs_state *state;
+};
+
+struct netfs_trans_dst *netfs_trans_search(struct netfs_state *st, unsigned int gen);
+void netfs_trans_drop_dst(struct netfs_trans_dst *dst);
+void netfs_trans_drop_dst_nostate(struct netfs_trans_dst *dst);
+void netfs_trans_drop_trans(struct netfs_trans *t, struct netfs_state *st);
+void netfs_trans_drop_last(struct netfs_trans *t, struct netfs_state *st);
+int netfs_trans_resend(struct netfs_trans *t, struct pohmelfs_sb *psb);
+int netfs_trans_remove_nolock(struct netfs_trans_dst *dst, struct netfs_state *st);
+
+int netfs_trans_init(void);
+void netfs_trans_exit(void);
+
+struct pohmelfs_crypto_engine
+{
+ u64 iv; /* Crypto IV for current operation */
+ unsigned long timeout; /* Crypto waiting timeout */
+ unsigned int size; /* Size of crypto scratchpad */
+ void *data; /* Temporal crypto scratchpad */
+ /*
+ * Crypto operations performed on objects.
+ */
+ struct crypto_hash *hash;
+ struct crypto_ablkcipher *cipher;
+
+ struct pohmelfs_crypto_thread *thread; /* Crypto thread which hosts this engine */
+
+ struct page **pages;
+ unsigned int page_num;
+};
+
+struct pohmelfs_crypto_thread
+{
+ struct list_head thread_entry;
+
+ struct task_struct *thread;
+ struct pohmelfs_sb *psb;
+
+ struct pohmelfs_crypto_engine eng;
+
+ struct netfs_trans *trans;
+
+ wait_queue_head_t wait;
+ int error;
+
+ unsigned int size;
+ struct page *page;
+};
+
+void pohmelfs_crypto_thread_make_ready(struct pohmelfs_crypto_thread *th);
+
+/*
+ * Network state, attached to one server.
+ */
+struct netfs_state
+{
+ struct mutex __state_lock; /* Can not allow to use the same socket simultaneously */
+ struct netfs_cmd cmd; /* Cached command */
+ struct netfs_inode_info info; /* Cached inode info */
+
+ void *data; /* Cached some data */
+ unsigned int size; /* Size of that data */
+
+ struct pohmelfs_sb *psb; /* Superblock */
+
+ struct task_struct *thread; /* Async receiving thread */
+
+ /* Waiting/polling machinery */
+ wait_queue_t wait;
+ wait_queue_head_t *whead;
+ wait_queue_head_t thread_wait;
+
+ struct mutex trans_lock;
+ struct rb_root trans_root;
+
+ struct pohmelfs_ctl ctl; /* Remote peer */
+
+ struct socket *socket; /* Socket object */
+ struct socket *read_socket; /* Cached pointer to socket object.
+ * Used to determine if between lock drops socket was changed.
+ * Never used to read data or any kind of access.
+ */
+ /*
+ * Crypto engines to process incoming data.
+ */
+ struct pohmelfs_crypto_engine eng;
+};
+
+int netfs_state_init(struct netfs_state *st);
+void netfs_state_exit(struct netfs_state *st);
+
+static inline void netfs_state_lock(struct netfs_state *st)
+{
+ mutex_lock(&st->__state_lock);
+}
+
+static inline void netfs_state_unlock(struct netfs_state *st)
+{
+ BUG_ON(!mutex_is_locked(&st->__state_lock));
+
+ mutex_unlock(&st->__state_lock);
+}
+
+static inline unsigned int netfs_state_poll(struct netfs_state *st)
+{
+ unsigned int revents = POLLHUP | POLLERR;
+
+ netfs_state_lock(st);
+ if (st->socket)
+ revents = st->socket->ops->poll(NULL, st->socket, NULL);
+ netfs_state_unlock(st);
+
+ return revents;
+}
+
+struct pohmelfs_config;
+
+struct pohmelfs_sb
+{
+ struct rb_root path_root;
+ struct mutex path_lock;
+
+ struct rb_root mcache_root;
+ struct mutex mcache_lock;
+ atomic_long_t mcache_gen;
+ unsigned long mcache_timeout;
+
+ unsigned int idx;
+
+ unsigned int trans_retries;
+
+ atomic_t trans_gen;
+
+ unsigned int crypto_attached_size;
+ unsigned int crypto_align_size;
+
+ unsigned int crypto_fail_unsupported;
+
+ unsigned int crypto_thread_num;
+ struct list_head crypto_active_list, crypto_ready_list;
+ struct mutex crypto_thread_lock;
+
+ unsigned int trans_max_pages;
+ unsigned long trans_data_size;
+ unsigned long trans_timeout;
+
+ unsigned long drop_scan_timeout;
+ unsigned long trans_scan_timeout;
+
+ unsigned long wait_on_page_timeout;
+
+ struct list_head flush_list;
+ struct list_head drop_list;
+ spinlock_t ino_lock;
+ u64 ino;
+
+ /*
+ * Remote nodes POHMELFS connected to.
+ */
+ struct list_head state_list;
+ struct mutex state_lock;
+
+ /*
+ * Currently active state to request data from.
+ */
+ struct pohmelfs_config *active_state;
+
+
+ wait_queue_head_t wait;
+
+ /*
+ * Timed checks: stale transactions, inodes to be freed and so on.
+ */
+ struct delayed_work dwork;
+ struct delayed_work drop_dwork;
+
+ struct super_block *sb;
+
+ /*
+ * Algorithm strings.
+ */
+ char *hash_string;
+ char *cipher_string;
+
+ u8 *hash_key;
+ u8 *cipher_key;
+
+ /*
+ * Algorithm string lengths.
+ */
+ unsigned int hash_strlen;
+ unsigned int cipher_strlen;
+ unsigned int hash_keysize;
+ unsigned int cipher_keysize;
+
+ /*
+ * Controls whether to perfrom crypto processing or not.
+ */
+ int perform_crypto;
+
+ /*
+ * POHMELFS statistics.
+ */
+ u64 total_size;
+ u64 avail_size;
+ atomic_long_t total_inodes;
+
+ /*
+ * Xattr support, read-only and so on.
+ */
+ u64 state_flags;
+
+ /*
+ * Temporary storage to detect changes in the wait queue.
+ */
+ long flags;
+};
+
+static inline void netfs_trans_update(struct netfs_cmd *cmd,
+ struct netfs_trans *t, unsigned int size)
+{
+ unsigned int sz = ALIGN(size, t->psb->crypto_align_size);
+
+ t->iovec.iov_len += sizeof(struct netfs_cmd) + sz;
+ cmd->cpad = __cpu_to_be16(sz - size);
+}
+
+static inline struct pohmelfs_sb *POHMELFS_SB(struct super_block *sb)
+{
+ return sb->s_fs_info;
+}
+
+static inline struct pohmelfs_inode *POHMELFS_I(struct inode *inode)
+{
+ return container_of(inode, struct pohmelfs_inode, vfs_inode);
+}
+
+static inline u64 pohmelfs_new_ino(struct pohmelfs_sb *psb)
+{
+ u64 ino;
+
+ spin_lock(&psb->ino_lock);
+ ino = psb->ino++;
+ spin_unlock(&psb->ino_lock);
+
+ return ino;
+}
+
+static inline void pohmelfs_put_inode(struct pohmelfs_inode *pi)
+{
+ struct pohmelfs_sb *psb = POHMELFS_SB(pi->vfs_inode.i_sb);
+
+ spin_lock(&psb->ino_lock);
+ list_move_tail(&pi->inode_entry, &psb->drop_list);
+ pi->drop_count++;
+ spin_unlock(&psb->ino_lock);
+}
+
+struct pohmelfs_config
+{
+ struct list_head config_entry;
+
+ struct netfs_state state;
+};
+
+struct pohmelfs_config_group
+{
+ /*
+ * Entry in the global config group list.
+ */
+ struct list_head group_entry;
+
+ /*
+ * Index of the current group.
+ */
+ unsigned int idx;
+ /*
+ * Number of config_list entries in this group entry.
+ */
+ unsigned int num_entry;
+ /*
+ * Algorithm strings.
+ */
+ char *hash_string;
+ char *cipher_string;
+
+ /*
+ * Algorithm string lengths.
+ */
+ unsigned int hash_strlen;
+ unsigned int cipher_strlen;
+
+ /*
+ * Key and its size.
+ */
+ unsigned int hash_keysize;
+ unsigned int cipher_keysize;
+ u8 *hash_key;
+ u8 *cipher_key;
+
+ /*
+ * List of config entries (network state info) for given idx.
+ */
+ struct list_head config_list;
+};
+
+int __init pohmelfs_config_init(void);
+void pohmelfs_config_exit(void);
+int pohmelfs_copy_config(struct pohmelfs_sb *psb);
+int pohmelfs_copy_crypto(struct pohmelfs_sb *psb);
+int pohmelfs_config_check(struct pohmelfs_config *config, int idx);
+int pohmelfs_state_init_one(struct pohmelfs_sb *psb, struct pohmelfs_config *conf);
+
+extern const struct file_operations pohmelfs_dir_fops;
+extern const struct inode_operations pohmelfs_dir_inode_ops;
+
+int pohmelfs_state_init(struct pohmelfs_sb *psb);
+void pohmelfs_state_exit(struct pohmelfs_sb *psb);
+void pohmelfs_state_flush_transactions(struct netfs_state *st);
+
+void pohmelfs_fill_inode(struct inode *inode, struct netfs_inode_info *info);
+
+void pohmelfs_name_del(struct pohmelfs_inode *parent, struct pohmelfs_name *n);
+void pohmelfs_free_names(struct pohmelfs_inode *parent);
+struct pohmelfs_name *pohmelfs_search_hash(struct pohmelfs_inode *pi, u32 hash);
+
+void pohmelfs_inode_del_inode(struct pohmelfs_sb *psb, struct pohmelfs_inode *pi);
+
+struct pohmelfs_inode *pohmelfs_create_entry_local(struct pohmelfs_sb *psb,
+ struct pohmelfs_inode *parent, struct qstr *str, u64 start, int mode);
+
+int pohmelfs_write_inode_create(struct inode *inode, struct netfs_trans *trans);
+
+struct pohmelfs_inode *pohmelfs_new_inode(struct pohmelfs_sb *psb,
+ struct pohmelfs_inode *parent, struct qstr *str,
+ struct netfs_inode_info *info, int link);
+
+int pohmelfs_setattr(struct dentry *dentry, struct iattr *attr);
+int pohmelfs_setattr_raw(struct inode *inode, struct iattr *attr);
+
+int pohmelfs_meta_command(struct pohmelfs_inode *pi, unsigned int cmd_op, unsigned int flags,
+ netfs_trans_complete_t complete, void *priv, u64 start);
+int pohmelfs_meta_command_data(struct pohmelfs_inode *pi, u64 id, unsigned int cmd_op, char *addon,
+ unsigned int flags, netfs_trans_complete_t complete, void *priv, u64 start);
+
+void pohmelfs_check_states(struct pohmelfs_sb *psb);
+void pohmelfs_switch_active(struct pohmelfs_sb *psb);
+
+struct pohmelfs_path_entry
+{
+ struct rb_node path_entry;
+ struct list_head entry;
+ u8 len, link;
+ u8 unused[2];
+ atomic_t refcnt;
+ u32 mode;
+ u32 hash;
+ u64 ino;
+ struct pohmelfs_path_entry *parent;
+ char *name;
+};
+
+void pohmelfs_remove_path_entry(struct pohmelfs_sb *psb, struct pohmelfs_path_entry *e);
+void pohmelfs_remove_path_entry_by_ino(struct pohmelfs_sb *psb, u64 ino);
+struct pohmelfs_path_entry * pohmelfs_add_path_entry(struct pohmelfs_sb *psb,
+ u64 parent_ino, u64 ino, struct qstr *str, int link, unsigned int mode);
+int pohmelfs_rename_path_entry(struct pohmelfs_sb *psb, u64 ino, u64 parent_ino, struct qstr *str);
+int pohmelfs_change_path_entry(struct pohmelfs_sb *psb, u64 ino, unsigned int mode);
+int pohmelfs_construct_path(struct pohmelfs_inode *pi, void *data, int len);
+int pohmelfs_construct_path_string(struct pohmelfs_inode *pi, void *data, int len);
+
+int pohmelfs_path_length(struct pohmelfs_inode *pi);
+int pohmelfs_path_length_create(struct pohmelfs_inode *pi);
+
+struct pohmelfs_crypto_completion
+{
+ struct completion complete;
+ int error;
+};
+
+int pohmelfs_trans_crypt(struct netfs_trans *t, struct pohmelfs_sb *psb);
+void pohmelfs_crypto_exit(struct pohmelfs_sb *psb);
+int pohmelfs_crypto_init(struct pohmelfs_sb *psb);
+
+int pohmelfs_crypto_engine_init(struct pohmelfs_crypto_engine *e, struct pohmelfs_sb *psb);
+void pohmelfs_crypto_engine_exit(struct pohmelfs_crypto_engine *e);
+
+int pohmelfs_crypto_process_input_data(struct pohmelfs_crypto_engine *e, u64 iv,
+ void *data, struct page *page, unsigned int size);
+int pohmelfs_crypto_process_input_page(struct pohmelfs_crypto_engine *e,
+ struct page *page, unsigned int size, u64 iv);
+
+static inline u64 pohmelfs_gen_iv(struct netfs_trans *t)
+{
+ u64 iv = t->gen;
+
+ iv <<= 32;
+ iv |= ((unsigned long)t) & 0xffffffff;
+
+ return iv;
+}
+
+int pohmelfs_data_lock(struct pohmelfs_inode *pi, u64 start, u32 size, int type);
+int pohmelfs_data_unlock(struct pohmelfs_inode *pi, u64 start, u32 size, int type);
+int pohmelfs_data_lock_response(struct netfs_state *st);
+
+int __init pohmelfs_mcache_init(void);
+void pohmelfs_mcache_exit(void);
+
+//#define CONFIG_POHMELFS_DEBUG
+
+#ifdef CONFIG_POHMELFS_DEBUG
+#define dprintka(f, a...) printk(f, ##a)
+#define dprintk(f, a...) printk("%d: " f, task_pid_vnr(current), ##a)
+#else
+#define dprintka(f, a...) do {} while (0)
+#define dprintk(f, a...) do {} while (0)
+#endif
+
+static inline void netfs_trans_get(struct netfs_trans *t)
+{
+ atomic_inc(&t->refcnt);
+}
+
+static inline void netfs_trans_put(struct netfs_trans *t)
+{
+ if (atomic_dec_and_test(&t->refcnt)) {
+ dprintk("%s: t: %p, gen: %u, err: %d.\n",
+ __func__, t, t->gen, t->result);
+ if (t->complete)
+ t->complete(t->pages, t->page_num,
+ t->private, t->result);
+ netfs_trans_free(t);
+ }
+}
+
+struct pohmelfs_mcache
+{
+ struct rb_node mcache_entry;
+ struct completion complete;
+
+ atomic_t refcnt;
+
+ u64 gen;
+
+ void *data;
+ u64 start;
+ u32 size;
+ int err;
+
+ struct netfs_inode_info info;
+};
+
+struct pohmelfs_mcache *pohmelfs_mcache_alloc(struct pohmelfs_sb *psb, u64 start,
+ unsigned int size, void *data);
+void pohmelfs_mcache_free(struct pohmelfs_sb *psb, struct pohmelfs_mcache *m);
+struct pohmelfs_mcache *pohmelfs_mcache_search(struct pohmelfs_sb *psb, u64 gen);
+void pohmelfs_mcache_remove_locked(struct pohmelfs_sb *psb, struct pohmelfs_mcache *m);
+
+static inline void pohmelfs_mcache_get(struct pohmelfs_mcache *m)
+{
+ atomic_inc(&m->refcnt);
+}
+
+static inline void pohmelfs_mcache_put(struct pohmelfs_sb *psb,
+ struct pohmelfs_mcache *m)
+{
+ if (atomic_dec_and_test(&m->refcnt))
+ pohmelfs_mcache_free(psb, m);
+}
+
+#endif /* __KERNEL__*/
+
+#endif /* __NETFS_H */
diff --git a/fs/pohmelfs/trans.c b/fs/pohmelfs/trans.c
new file mode 100644
index 0000000..1eda0a9
--- /dev/null
+++ b/fs/pohmelfs/trans.c
@@ -0,0 +1,704 @@
+/*
+ * 2007+ Copyright (c) Evgeniy Polyakov <johnpol@....mipt.ru>
+ * All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#include <linux/module.h>
+#include <linux/crypto.h>
+#include <linux/fs.h>
+#include <linux/jhash.h>
+#include <linux/hash.h>
+#include <linux/ktime.h>
+#include <linux/mempool.h>
+#include <linux/mm.h>
+#include <linux/mount.h>
+#include <linux/pagemap.h>
+#include <linux/parser.h>
+#include <linux/poll.h>
+#include <linux/swap.h>
+#include <linux/slab.h>
+#include <linux/statfs.h>
+#include <linux/writeback.h>
+
+#include "netfs.h"
+
+static struct kmem_cache *netfs_trans_dst;
+static mempool_t *netfs_trans_dst_pool;
+
+static void netfs_trans_init_static(struct netfs_trans *t, int num, int size)
+{
+ t->page_num = num;
+ t->total_size = size;
+ atomic_set(&t->refcnt, 1);
+
+ mutex_init(&t->dst_lock);
+ INIT_LIST_HEAD(&t->dst_list);
+}
+
+static int netfs_trans_send_pages(struct netfs_trans *t, struct netfs_state *st)
+{
+ int err = 0;
+ unsigned int i, attached_pages = t->attached_pages, ci;
+ struct msghdr msg;
+ struct page **pages = (t->eng)?t->eng->pages:t->pages;
+ struct page *p;
+ unsigned int size;
+
+ msg.msg_name = NULL;
+ msg.msg_namelen = 0;
+ msg.msg_control = NULL;
+ msg.msg_controllen = 0;
+ msg.msg_flags = MSG_WAITALL | MSG_MORE;
+
+ ci = 0;
+ for (i=0; i<t->page_num; ++i) {
+ struct page *page = pages[ci];
+ struct netfs_cmd cmd;
+ struct iovec io;
+
+ p = t->pages[i];
+
+ if (!p)
+ continue;
+
+ size = page_private(p);
+
+ io.iov_base = &cmd;
+ io.iov_len = sizeof(struct netfs_cmd);
+
+ cmd.cmd = NETFS_WRITE_PAGE;
+ cmd.ext = 0;
+ cmd.id = 0;
+ cmd.size = size;
+ cmd.start = p->index;
+ cmd.start <<= PAGE_CACHE_SHIFT;
+ cmd.csize = 0;
+ cmd.cpad = 0;
+ cmd.iv = pohmelfs_gen_iv(t);
+
+ netfs_convert_cmd(&cmd);
+
+ msg.msg_iov = &io;
+ msg.msg_iovlen = 1;
+ msg.msg_flags = MSG_WAITALL | MSG_MORE;
+
+ err = kernel_sendmsg(st->socket, &msg, (struct kvec *)msg.msg_iov, 1, sizeof(struct netfs_cmd));
+ if (err <= 0) {
+ printk("%s: %d/%d failed to send transaction header: t: %p, gen: %u, err: %d.\n",
+ __func__, i, t->page_num, t, t->gen, err);
+ if (err == 0)
+ err = -ECONNRESET;
+ goto err_out;
+ }
+
+ msg.msg_flags = MSG_WAITALL|(attached_pages == 1)?0:MSG_MORE;
+
+ err = kernel_sendpage(st->socket, page, 0, size, msg.msg_flags);
+ if (err <= 0) {
+ printk("%s: %d/%d failed to send transaction page: t: %p, gen: %u, size: %u, err: %d.\n",
+ __func__, i, t->page_num, t, t->gen, size, err);
+ if (err == 0)
+ err = -ECONNRESET;
+ goto err_out;
+ }
+
+ dprintk("%s: %d/%d sent t: %p, gen: %u, page: %p/%p, size: %u.\n",
+ __func__, i, t->page_num, t, t->gen, page, p, size);
+
+ err = 0;
+ attached_pages--;
+ if (!attached_pages)
+ break;
+ ci++;
+
+ continue;
+
+err_out:
+ printk("%s: t: %p, gen: %u, err: %d.\n", __func__, t, t->gen, err);
+ netfs_state_exit(st);
+ break;
+ }
+
+ return err;
+}
+
+int netfs_trans_send(struct netfs_trans *t, struct netfs_state *st)
+{
+ int err;
+ struct msghdr msg;
+
+ netfs_state_lock(st);
+ if (!st->socket) {
+ err = netfs_state_init(st);
+ if (err)
+ goto err_out_unlock_return;
+ }
+
+ msg.msg_iov = &t->iovec;
+ msg.msg_iovlen = 1;
+ msg.msg_name = NULL;
+ msg.msg_namelen = 0;
+ msg.msg_control = NULL;
+ msg.msg_controllen = 0;
+ msg.msg_flags = MSG_WAITALL;
+
+ if (t->attached_pages)
+ msg.msg_flags |= MSG_MORE;
+
+ err = kernel_sendmsg(st->socket, &msg, (struct kvec *)msg.msg_iov, 1, t->iovec.iov_len);
+ if (err <= 0) {
+ printk("%s: failed to send contig transaction: t: %p, gen: %u, size: %u, err: %d.\n",
+ __func__, t, t->gen, t->iovec.iov_len, err);
+ if (err == 0)
+ err = -ECONNRESET;
+ goto err_out_unlock_return;
+ }
+
+ dprintk("%s: sent %s transaction: t: %p, gen: %u, size: %u, page_num: %u.\n",
+ __func__, (t->page_num)?"partial":"full",
+ t, t->gen, t->iovec.iov_len, t->page_num);
+
+ err = 0;
+ if (t->attached_pages)
+ err = netfs_trans_send_pages(t, st);
+
+err_out_unlock_return:
+ netfs_state_unlock(st);
+
+ dprintk("%s: t: %p, gen: %u, err: %d.\n",
+ __func__, t, t->gen, err);
+
+ t->result = err;
+ return err;
+}
+
+static inline int netfs_trans_cmp(unsigned int gen, unsigned int new)
+{
+ if (gen < new)
+ return 1;
+ if (gen > new)
+ return -1;
+ return 0;
+}
+
+struct netfs_trans_dst *netfs_trans_search(struct netfs_state *st, unsigned int gen)
+{
+ struct rb_root *root = &st->trans_root;
+ struct rb_node *n = root->rb_node;
+ struct netfs_trans_dst *tmp, *ret = NULL;
+ struct netfs_trans *t;
+ int cmp;
+
+ while (n) {
+ tmp = rb_entry(n, struct netfs_trans_dst, state_entry);
+ t = tmp->trans;
+
+ cmp = netfs_trans_cmp(t->gen, gen);
+ if (cmp < 0)
+ n = n->rb_left;
+ else if (cmp > 0)
+ n = n->rb_right;
+ else {
+ ret = tmp;
+ break;
+ }
+ }
+
+ return ret;
+}
+
+static int netfs_trans_insert(struct netfs_trans_dst *ndst, struct netfs_state *st)
+{
+ struct rb_root *root = &st->trans_root;
+ struct rb_node **n = &root->rb_node, *parent = NULL;
+ struct netfs_trans_dst *ret = NULL, *tmp;
+ struct netfs_trans *t = NULL, *new = ndst->trans;
+ int cmp;
+
+ while (*n) {
+ parent = *n;
+
+ tmp = rb_entry(parent, struct netfs_trans_dst, state_entry);
+ t = tmp->trans;
+
+ cmp = netfs_trans_cmp(t->gen, new->gen);
+ if (cmp < 0)
+ n = &parent->rb_left;
+ else if (cmp > 0)
+ n = &parent->rb_right;
+ else {
+ ret = tmp;
+ break;
+ }
+ }
+
+ if (ret) {
+ printk("%s: exist: old: gen: %u, flags: %x, send_time: %lu, "
+ "new: gen: %u, flags: %x, send_time: %lu.\n",
+ __func__, t->gen, t->flags, ret->send_time,
+ new->gen, new->flags, ndst->send_time);
+ return -EEXIST;
+ }
+
+ rb_link_node(&ndst->state_entry, parent, n);
+ rb_insert_color(&ndst->state_entry, root);
+ ndst->send_time = jiffies;
+
+ return 0;
+}
+
+int netfs_trans_remove_nolock(struct netfs_trans_dst *dst, struct netfs_state *st)
+{
+ if (dst && dst->state_entry.rb_parent_color) {
+ rb_erase(&dst->state_entry, &st->trans_root);
+ dst->state_entry.rb_parent_color = 0;
+ return 1;
+ }
+ return 0;
+}
+
+static int netfs_trans_remove_state(struct netfs_trans_dst *dst)
+{
+ int ret;
+ struct netfs_state *st = dst->state;
+
+ mutex_lock(&st->trans_lock);
+ ret = netfs_trans_remove_nolock(dst, st);
+ mutex_unlock(&st->trans_lock);
+
+ return ret;
+}
+
+/*
+ * Create new destination for given transaction associated with given network state.
+ * Transaction's reference counter is bumped and will be dropped when either
+ * reply is received or when async timeout detection task will fail resending
+ * and drop transaction.
+ */
+static int netfs_trans_push_dst(struct netfs_trans *t, struct netfs_state *st)
+{
+ struct netfs_trans_dst *dst;
+ int err;
+
+ dst = mempool_alloc(netfs_trans_dst_pool, GFP_KERNEL);
+ if (!dst)
+ return -ENOMEM;
+
+ dst->retries = 0;
+ dst->send_time = 0;
+ dst->state = st;
+ dst->trans = t;
+ netfs_trans_get(t);
+
+ mutex_lock(&st->trans_lock);
+ err = netfs_trans_insert(dst, st);
+ mutex_unlock(&st->trans_lock);
+
+ if (err)
+ goto err_out_free;
+
+ mutex_lock(&t->dst_lock);
+ list_add_tail(&dst->trans_entry, &t->dst_list);
+ mutex_unlock(&t->dst_lock);
+
+ return 0;
+
+err_out_free:
+ t->result = err;
+ netfs_trans_put(t);
+ mempool_free(dst, netfs_trans_dst_pool);
+ return err;
+}
+
+static void netfs_trans_free_dst(struct netfs_trans_dst *dst)
+{
+ netfs_trans_put(dst->trans);
+ mempool_free(dst, netfs_trans_dst_pool);
+}
+
+static void netfs_trans_remove_dst(struct netfs_trans_dst *dst)
+{
+ netfs_trans_remove_state(dst);
+ netfs_trans_free_dst(dst);
+}
+
+/*
+ * Drop destination transaction entry when we know it.
+ */
+void netfs_trans_drop_dst(struct netfs_trans_dst *dst)
+{
+ struct netfs_trans *t = dst->trans;
+
+ mutex_lock(&t->dst_lock);
+ list_del_init(&dst->trans_entry);
+ mutex_unlock(&t->dst_lock);
+
+ netfs_trans_remove_dst(dst);
+}
+
+/*
+ * Drop destination transaction entry when we know it and when we
+ * already removed dst from state tree.
+ */
+void netfs_trans_drop_dst_nostate(struct netfs_trans_dst *dst)
+{
+ struct netfs_trans *t = dst->trans;
+
+ mutex_lock(&t->dst_lock);
+ list_del_init(&dst->trans_entry);
+ mutex_unlock(&t->dst_lock);
+
+ netfs_trans_free_dst(dst);
+}
+
+/*
+ * This drops destination transaction entry from appropriate network state
+ * tree and drops related reference counter. It is possible that transaction
+ * will be freed here if its reference counter hits zero.
+ * Destination transaction entry will be freed.
+ */
+void netfs_trans_drop_trans(struct netfs_trans *t, struct netfs_state *st)
+{
+ struct netfs_trans_dst *dst, *tmp, *ret = NULL;
+
+ mutex_lock(&t->dst_lock);
+ list_for_each_entry_safe(dst, tmp, &t->dst_list, trans_entry) {
+ if (dst->state == st) {
+ ret = dst;
+ list_del(&dst->trans_entry);
+ break;
+ }
+ }
+ mutex_unlock(&t->dst_lock);
+
+ if (ret)
+ netfs_trans_remove_dst(ret);
+}
+
+/*
+ * This drops destination transaction entry from appropriate network state
+ * tree and drops related reference counter. It is possible that transaction
+ * will be freed here if its reference counter hits zero.
+ * Destination transaction entry will be freed.
+ */
+void netfs_trans_drop_last(struct netfs_trans *t, struct netfs_state *st)
+{
+ struct netfs_trans_dst *dst, *tmp, *ret;
+
+ mutex_lock(&t->dst_lock);
+ ret = list_entry(t->dst_list.prev, struct netfs_trans_dst, trans_entry);
+ if (ret->state != st) {
+ ret = NULL;
+ list_for_each_entry_safe(dst, tmp, &t->dst_list, trans_entry) {
+ if (dst->state == st) {
+ ret = dst;
+ list_del_init(&dst->trans_entry);
+ break;
+ }
+ }
+ } else {
+ list_del(&ret->trans_entry);
+ }
+ mutex_unlock(&t->dst_lock);
+
+ if (ret)
+ netfs_trans_remove_dst(ret);
+}
+
+static int netfs_trans_push(struct netfs_trans *t, struct netfs_state *st)
+{
+ int err;
+
+ err = netfs_trans_push_dst(t, st);
+ if (err)
+ return err;
+
+ err = netfs_trans_send(t, st);
+ if (err)
+ goto err_out_free;
+
+ if (t->flags & NETFS_TRANS_SINGLE_DST)
+ pohmelfs_switch_active(st->psb);
+
+ return 0;
+
+err_out_free:
+ t->result = err;
+ netfs_trans_drop_last(t, st);
+
+ return err;
+}
+
+int netfs_trans_finish_send(struct netfs_trans *t, struct pohmelfs_sb *psb)
+{
+ struct pohmelfs_config *c;
+ int err = -ENODEV;
+ struct netfs_state *st;
+#if 0
+ dprintk("%s: t: %p, gen: %u, size: %u, page_num: %u, active: %p.\n",
+ __func__, t, t->gen, t->iovec.iov_len, t->page_num, psb->active_state);
+#endif
+ mutex_lock(&psb->state_lock);
+ if ((t->flags & NETFS_TRANS_SINGLE_DST) && psb->active_state) {
+ st = &psb->active_state->state;
+
+ err = -EPIPE;
+ if (netfs_state_poll(st) & POLLOUT) {
+ err = netfs_trans_push_dst(t, st);
+ if (!err) {
+ err = netfs_trans_send(t, st);
+ if (err) {
+ netfs_trans_drop_last(t, st);
+ } else {
+ pohmelfs_switch_active(psb);
+ goto out;
+ }
+ }
+ }
+ pohmelfs_switch_active(psb);
+ }
+
+ list_for_each_entry(c, &psb->state_list, config_entry) {
+ st = &c->state;
+
+ err = netfs_trans_push(t, st);
+ if (!err && (t->flags & NETFS_TRANS_SINGLE_DST))
+ break;
+ }
+out:
+ mutex_unlock(&psb->state_lock);
+#if 0
+ dprintk("%s: fully sent t: %p, gen: %u, size: %u, page_num: %u, err: %d.\n",
+ __func__, t, t->gen, t->iovec.iov_len, t->page_num, err);
+#endif
+ if (err)
+ t->result = err;
+ return err;
+}
+
+int netfs_trans_finish(struct netfs_trans *t, struct pohmelfs_sb *psb)
+{
+ int err;
+ struct netfs_cmd *cmd = t->iovec.iov_base;
+
+ t->gen = atomic_inc_return(&psb->trans_gen);
+
+ cmd->size = t->iovec.iov_len - sizeof(struct netfs_cmd) +
+ t->attached_size + t->attached_pages * sizeof(struct netfs_cmd);
+ cmd->cmd = NETFS_TRANS;
+ cmd->start = t->gen;
+ cmd->id = 0;
+
+ if (psb->perform_crypto) {
+ cmd->ext = psb->crypto_attached_size;
+ cmd->csize = psb->crypto_attached_size;
+ }
+#if 0
+ dprintk("%s: trans: %llu, crypto_attached_size: %u, attached_size: %u, attached_pages: %d, trans_size: %u.\n",
+ __func__, cmd->start, psb->crypto_attached_size, t->attached_size, t->attached_pages, cmd->size);
+#endif
+ err = pohmelfs_trans_crypt(t, psb);
+ if (err)
+ t->result = err;
+ netfs_trans_put(t);
+ return err;
+}
+
+/*
+ * Resend transaction to remote server(s).
+ * If new servers were added into superblock, we can try to send data
+ * to them too.
+ *
+ * It is called under superblock's state_lock, so we can safely
+ * dereference psb->state_list. Also, transaction's reference counter is
+ * bumped, so it can not go away under us, thus we can safely access all
+ * its members. State is locked.
+ *
+ * This function returns 0 if transaction was successfully sent to at
+ * least one destination target.
+ */
+int netfs_trans_resend(struct netfs_trans *t, struct pohmelfs_sb *psb)
+{
+ struct netfs_trans_dst *dst;
+ struct netfs_state *st;
+ struct pohmelfs_config *c;
+ int err, exist, error = -ENODEV;
+
+ list_for_each_entry(c, &psb->state_list, config_entry) {
+ st = &c->state;
+
+ exist = 0;
+ mutex_lock(&t->dst_lock);
+ list_for_each_entry(dst, &t->dst_list, trans_entry) {
+ if (st == dst->state) {
+ exist = 1;
+ break;
+ }
+ }
+ mutex_unlock(&t->dst_lock);
+
+ if (exist) {
+ if (!(t->flags & NETFS_TRANS_SINGLE_DST) ||
+ (c->config_entry.next == &psb->state_list)) {
+ dprintk("%s: resending st: %p, t: %p, gen: %u.\n",
+ __func__, st, t, t->gen);
+ err = netfs_trans_send(t, st);
+ if (!err)
+ error = 0;
+ }
+ continue;
+ }
+
+ dprintk("%s: pushing/resending st: %p, t: %p, gen: %u.\n",
+ __func__, st, t, t->gen);
+ err = netfs_trans_push(t, st);
+ if (err)
+ continue;
+ error = 0;
+ if (t->flags & NETFS_TRANS_SINGLE_DST)
+ break;
+ }
+
+ t->result = error;
+ return error;
+}
+
+void *netfs_trans_add(struct netfs_trans *t, unsigned int size)
+{
+ struct iovec *io = &t->iovec;
+ void *ptr;
+
+ if (size > t->total_size) {
+ ptr = ERR_PTR(-EINVAL);
+ goto out;
+ }
+
+ if (io->iov_len + size > t->total_size) {
+ dprintk("%s: too big size t: %p, gen: %u, iov_len: %u, size: %u, total: %u.\n",
+ __func__, t, t->gen, io->iov_len, size, t->total_size);
+ ptr = ERR_PTR(-E2BIG);
+ goto out;
+ }
+
+ ptr = io->iov_base + io->iov_len;
+ io->iov_len += size;
+
+out:
+ dprintk("%s: t: %p, gen: %u, size: %u, total: %u.\n",
+ __func__, t, t->gen, size, io->iov_len);
+ return ptr;
+}
+
+void netfs_trans_free(struct netfs_trans *t)
+{
+ if (t->eng)
+ pohmelfs_crypto_thread_make_ready(t->eng->thread);
+ kfree(t);
+}
+
+struct netfs_trans *netfs_trans_alloc(struct pohmelfs_sb *psb, unsigned int size,
+ unsigned int flags, unsigned int nr)
+{
+ struct netfs_trans *t;
+ unsigned int num, cont, pad, size_no_trans;
+ unsigned int crypto_added = 0;
+ struct netfs_cmd *cmd;
+
+ if (psb->perform_crypto)
+ crypto_added = psb->crypto_attached_size;
+
+ /*
+ * |sizeof(struct netfs_trans)|
+ * |sizeof(struct netfs_cmd)| - transaction header
+ * |size| - buffer with requested size
+ * |padding| - crypto padding, zero bytes
+ * |nr * sizeof(struct page *)| - array of page pointers
+ *
+ * Overall size should be less than PAGE_SIZE for guaranteed allocation.
+ */
+
+ cont = size;
+ size = ALIGN(size, psb->crypto_align_size);
+ pad = size - cont;
+
+ size_no_trans = size + sizeof(struct netfs_cmd) * 2 + crypto_added;
+
+ cont = sizeof(struct netfs_trans) + size_no_trans;
+
+ num = (PAGE_SIZE - cont)/sizeof(struct page *);
+
+ if (nr > num)
+ nr = num;
+
+ t = kzalloc(cont + nr*sizeof(struct page *), GFP_NOIO);
+ if (!t)
+ goto err_out_exit;
+
+ memset(t, 0, sizeof(struct netfs_trans));
+
+ t->iovec.iov_base = (void *)(t + 1);
+ t->pages = (struct page **)(t->iovec.iov_base + size_no_trans);
+
+ /*
+ * Reserving space for transaction header.
+ */
+ t->iovec.iov_len = sizeof(struct netfs_cmd) + crypto_added;
+
+ netfs_trans_init_static(t, nr, size_no_trans);
+
+ t->flags = flags;
+ t->psb = psb;
+
+ cmd = (struct netfs_cmd *)t->iovec.iov_base;
+
+ cmd->size = size;
+ cmd->cpad = pad;
+ cmd->csize = crypto_added;
+
+ dprintk("%s: t: %p, gen: %u, size: %u, padding: %u, align_size: %u, flags: %x, "
+ "page_num: %u, base: %p, pages: %p.\n",
+ __func__, t, t->gen, size, pad, psb->crypto_align_size, flags, nr,
+ t->iovec.iov_base, t->pages);
+
+ return t;
+
+err_out_exit:
+ return NULL;
+}
+
+int netfs_trans_init(void)
+{
+ int err = -ENOMEM;
+
+ netfs_trans_dst = kmem_cache_create("netfs_trans_dst", sizeof(struct netfs_trans_dst),
+ 0, 0, NULL);
+ if (!netfs_trans_dst)
+ goto err_out_exit;
+
+ netfs_trans_dst_pool = mempool_create_slab_pool(256, netfs_trans_dst);
+ if (!netfs_trans_dst_pool)
+ goto err_out_free;
+
+ return 0;
+
+err_out_free:
+ kmem_cache_destroy(netfs_trans_dst);
+err_out_exit:
+ return err;
+}
+
+void netfs_trans_exit(void)
+{
+ mempool_destroy(netfs_trans_dst_pool);
+ kmem_cache_destroy(netfs_trans_dst);
+}
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists