提交 1135ca9c 编写于 作者: D Dan Magenheimer 提交者: Greg Kroah-Hartman

staging: ramster: cluster/messaging foundation

Copy cluster subdirectory from ocfs2.  These files implement
the basic cluster discovery, mapping, heartbeat / keepalive, and
messaging ("o2net") that ramster requires for internode communication.
Note: there are NO ramster-specific changes yet; this commit
does NOT pass checkpatch since the copied source files do not.

(Why copy?  This particular part of ocfs2 has never been broken out
for non-ocfs2 use before, some (small) changes are required for ramster
to use that code, and ramster is currently incompatible with real
ocfs2 anyway (requires !CONFIG_OCFS2_FS).  Before ramster can be promoted
out of staging, we will need to work with the ocfs2 maintainers to
see if the code interdependencies can be merged, but for now, for
staging, this seemed to be an expedient way to make use of the ocfs2
core cluster code while still incorporating necessary changes for ramster.)
Signed-off-by: NDan Magenheimer <dan.magenheimer@oracle.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>
上级 32de21f7
obj-$(CONFIG_OCFS2_FS) += ocfs2_nodemanager.o
ocfs2_nodemanager-objs := heartbeat.o masklog.o sys.o nodemanager.o \
quorum.o tcp.o netdebug.o ver.o
此差异已折叠。
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* heartbeat.h
*
* Function prototypes
*
* Copyright (C) 2004 Oracle. 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.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*
*/
#ifndef O2CLUSTER_HEARTBEAT_H
#define O2CLUSTER_HEARTBEAT_H
#include "ocfs2_heartbeat.h"
#define O2HB_REGION_TIMEOUT_MS 2000
#define O2HB_MAX_REGION_NAME_LEN 32
/* number of changes to be seen as live */
#define O2HB_LIVE_THRESHOLD 2
/* number of equal samples to be seen as dead */
extern unsigned int o2hb_dead_threshold;
#define O2HB_DEFAULT_DEAD_THRESHOLD 31
/* Otherwise MAX_WRITE_TIMEOUT will be zero... */
#define O2HB_MIN_DEAD_THRESHOLD 2
#define O2HB_MAX_WRITE_TIMEOUT_MS (O2HB_REGION_TIMEOUT_MS * (o2hb_dead_threshold - 1))
#define O2HB_CB_MAGIC 0x51d1e4ec
/* callback stuff */
enum o2hb_callback_type {
O2HB_NODE_DOWN_CB = 0,
O2HB_NODE_UP_CB,
O2HB_NUM_CB
};
struct o2nm_node;
typedef void (o2hb_cb_func)(struct o2nm_node *, int, void *);
struct o2hb_callback_func {
u32 hc_magic;
struct list_head hc_item;
o2hb_cb_func *hc_func;
void *hc_data;
int hc_priority;
enum o2hb_callback_type hc_type;
};
struct config_group *o2hb_alloc_hb_set(void);
void o2hb_free_hb_set(struct config_group *group);
void o2hb_setup_callback(struct o2hb_callback_func *hc,
enum o2hb_callback_type type,
o2hb_cb_func *func,
void *data,
int priority);
int o2hb_register_callback(const char *region_uuid,
struct o2hb_callback_func *hc);
void o2hb_unregister_callback(const char *region_uuid,
struct o2hb_callback_func *hc);
void o2hb_fill_node_map(unsigned long *map,
unsigned bytes);
void o2hb_exit(void);
int o2hb_init(void);
int o2hb_check_node_heartbeating(u8 node_num);
int o2hb_check_node_heartbeating_from_callback(u8 node_num);
int o2hb_check_local_node_heartbeating(void);
void o2hb_stop_all_regions(void);
int o2hb_get_all_regions(char *region_uuids, u8 numregions);
int o2hb_global_heartbeat_active(void);
#endif /* O2CLUSTER_HEARTBEAT_H */
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* Copyright (C) 2004, 2005 Oracle. 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.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*/
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/string.h>
#include <asm/uaccess.h>
#include "masklog.h"
struct mlog_bits mlog_and_bits = MLOG_BITS_RHS(MLOG_INITIAL_AND_MASK);
EXPORT_SYMBOL_GPL(mlog_and_bits);
struct mlog_bits mlog_not_bits = MLOG_BITS_RHS(0);
EXPORT_SYMBOL_GPL(mlog_not_bits);
static ssize_t mlog_mask_show(u64 mask, char *buf)
{
char *state;
if (__mlog_test_u64(mask, mlog_and_bits))
state = "allow";
else if (__mlog_test_u64(mask, mlog_not_bits))
state = "deny";
else
state = "off";
return snprintf(buf, PAGE_SIZE, "%s\n", state);
}
static ssize_t mlog_mask_store(u64 mask, const char *buf, size_t count)
{
if (!strnicmp(buf, "allow", 5)) {
__mlog_set_u64(mask, mlog_and_bits);
__mlog_clear_u64(mask, mlog_not_bits);
} else if (!strnicmp(buf, "deny", 4)) {
__mlog_set_u64(mask, mlog_not_bits);
__mlog_clear_u64(mask, mlog_and_bits);
} else if (!strnicmp(buf, "off", 3)) {
__mlog_clear_u64(mask, mlog_not_bits);
__mlog_clear_u64(mask, mlog_and_bits);
} else
return -EINVAL;
return count;
}
struct mlog_attribute {
struct attribute attr;
u64 mask;
};
#define to_mlog_attr(_attr) container_of(_attr, struct mlog_attribute, attr)
#define define_mask(_name) { \
.attr = { \
.name = #_name, \
.mode = S_IRUGO | S_IWUSR, \
}, \
.mask = ML_##_name, \
}
static struct mlog_attribute mlog_attrs[MLOG_MAX_BITS] = {
define_mask(TCP),
define_mask(MSG),
define_mask(SOCKET),
define_mask(HEARTBEAT),
define_mask(HB_BIO),
define_mask(DLMFS),
define_mask(DLM),
define_mask(DLM_DOMAIN),
define_mask(DLM_THREAD),
define_mask(DLM_MASTER),
define_mask(DLM_RECOVERY),
define_mask(DLM_GLUE),
define_mask(VOTE),
define_mask(CONN),
define_mask(QUORUM),
define_mask(BASTS),
define_mask(CLUSTER),
define_mask(ERROR),
define_mask(NOTICE),
define_mask(KTHREAD),
};
static struct attribute *mlog_attr_ptrs[MLOG_MAX_BITS] = {NULL, };
static ssize_t mlog_show(struct kobject *obj, struct attribute *attr,
char *buf)
{
struct mlog_attribute *mlog_attr = to_mlog_attr(attr);
return mlog_mask_show(mlog_attr->mask, buf);
}
static ssize_t mlog_store(struct kobject *obj, struct attribute *attr,
const char *buf, size_t count)
{
struct mlog_attribute *mlog_attr = to_mlog_attr(attr);
return mlog_mask_store(mlog_attr->mask, buf, count);
}
static const struct sysfs_ops mlog_attr_ops = {
.show = mlog_show,
.store = mlog_store,
};
static struct kobj_type mlog_ktype = {
.default_attrs = mlog_attr_ptrs,
.sysfs_ops = &mlog_attr_ops,
};
static struct kset mlog_kset = {
.kobj = {.ktype = &mlog_ktype},
};
int mlog_sys_init(struct kset *o2cb_kset)
{
int i = 0;
while (mlog_attrs[i].attr.mode) {
mlog_attr_ptrs[i] = &mlog_attrs[i].attr;
i++;
}
mlog_attr_ptrs[i] = NULL;
kobject_set_name(&mlog_kset.kobj, "logmask");
mlog_kset.kobj.kset = o2cb_kset;
return kset_register(&mlog_kset);
}
void mlog_sys_shutdown(void)
{
kset_unregister(&mlog_kset);
}
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* Copyright (C) 2005 Oracle. 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.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*/
#ifndef O2CLUSTER_MASKLOG_H
#define O2CLUSTER_MASKLOG_H
/*
* For now this is a trivial wrapper around printk() that gives the critical
* ability to enable sets of debugging output at run-time. In the future this
* will almost certainly be redirected to relayfs so that it can pay a
* substantially lower heisenberg tax.
*
* Callers associate the message with a bitmask and a global bitmask is
* maintained with help from /proc. If any of the bits match the message is
* output.
*
* We must have efficient bit tests on i386 and it seems gcc still emits crazy
* code for the 64bit compare. It emits very good code for the dual unsigned
* long tests, though, completely avoiding tests that can never pass if the
* caller gives a constant bitmask that fills one of the longs with all 0s. So
* the desire is to have almost all of the calls decided on by comparing just
* one of the longs. This leads to having infrequently given bits that are
* frequently matched in the high bits.
*
* _ERROR and _NOTICE are used for messages that always go to the console and
* have appropriate KERN_ prefixes. We wrap these in our function instead of
* just calling printk() so that this can eventually make its way through
* relayfs along with the debugging messages. Everything else gets KERN_DEBUG.
* The inline tests and macro dance give GCC the opportunity to quite cleverly
* only emit the appropriage printk() when the caller passes in a constant
* mask, as is almost always the case.
*
* All this bitmask nonsense is managed from the files under
* /sys/fs/o2cb/logmask/. Reading the files gives a straightforward
* indication of which bits are allowed (allow) or denied (off/deny).
* ENTRY deny
* EXIT deny
* TCP off
* MSG off
* SOCKET off
* ERROR allow
* NOTICE allow
*
* Writing changes the state of a given bit and requires a strictly formatted
* single write() call:
*
* write(fd, "allow", 5);
*
* Echoing allow/deny/off string into the logmask files can flip the bits
* on or off as expected; here is the bash script for example:
*
* log_mask="/sys/fs/o2cb/log_mask"
* for node in ENTRY EXIT TCP MSG SOCKET ERROR NOTICE; do
* echo allow >"$log_mask"/"$node"
* done
*
* The debugfs.ocfs2 tool can also flip the bits with the -l option:
*
* debugfs.ocfs2 -l TCP allow
*/
/* for task_struct */
#include <linux/sched.h>
/* bits that are frequently given and infrequently matched in the low word */
/* NOTE: If you add a flag, you need to also update masklog.c! */
#define ML_TCP 0x0000000000000001ULL /* net cluster/tcp.c */
#define ML_MSG 0x0000000000000002ULL /* net network messages */
#define ML_SOCKET 0x0000000000000004ULL /* net socket lifetime */
#define ML_HEARTBEAT 0x0000000000000008ULL /* hb all heartbeat tracking */
#define ML_HB_BIO 0x0000000000000010ULL /* hb io tracing */
#define ML_DLMFS 0x0000000000000020ULL /* dlm user dlmfs */
#define ML_DLM 0x0000000000000040ULL /* dlm general debugging */
#define ML_DLM_DOMAIN 0x0000000000000080ULL /* dlm domain debugging */
#define ML_DLM_THREAD 0x0000000000000100ULL /* dlm domain thread */
#define ML_DLM_MASTER 0x0000000000000200ULL /* dlm master functions */
#define ML_DLM_RECOVERY 0x0000000000000400ULL /* dlm master functions */
#define ML_DLM_GLUE 0x0000000000000800ULL /* ocfs2 dlm glue layer */
#define ML_VOTE 0x0000000000001000ULL /* ocfs2 node messaging */
#define ML_CONN 0x0000000000002000ULL /* net connection management */
#define ML_QUORUM 0x0000000000004000ULL /* net connection quorum */
#define ML_BASTS 0x0000000000008000ULL /* dlmglue asts and basts */
#define ML_CLUSTER 0x0000000000010000ULL /* cluster stack */
/* bits that are infrequently given and frequently matched in the high word */
#define ML_ERROR 0x1000000000000000ULL /* sent to KERN_ERR */
#define ML_NOTICE 0x2000000000000000ULL /* setn to KERN_NOTICE */
#define ML_KTHREAD 0x4000000000000000ULL /* kernel thread activity */
#define MLOG_INITIAL_AND_MASK (ML_ERROR|ML_NOTICE)
#ifndef MLOG_MASK_PREFIX
#define MLOG_MASK_PREFIX 0
#endif
/*
* When logging is disabled, force the bit test to 0 for anything other
* than errors and notices, allowing gcc to remove the code completely.
* When enabled, allow all masks.
*/
#if defined(CONFIG_OCFS2_DEBUG_MASKLOG)
#define ML_ALLOWED_BITS ~0
#else
#define ML_ALLOWED_BITS (ML_ERROR|ML_NOTICE)
#endif
#define MLOG_MAX_BITS 64
struct mlog_bits {
unsigned long words[MLOG_MAX_BITS / BITS_PER_LONG];
};
extern struct mlog_bits mlog_and_bits, mlog_not_bits;
#if BITS_PER_LONG == 32
#define __mlog_test_u64(mask, bits) \
( (u32)(mask & 0xffffffff) & bits.words[0] || \
((u64)(mask) >> 32) & bits.words[1] )
#define __mlog_set_u64(mask, bits) do { \
bits.words[0] |= (u32)(mask & 0xffffffff); \
bits.words[1] |= (u64)(mask) >> 32; \
} while (0)
#define __mlog_clear_u64(mask, bits) do { \
bits.words[0] &= ~((u32)(mask & 0xffffffff)); \
bits.words[1] &= ~((u64)(mask) >> 32); \
} while (0)
#define MLOG_BITS_RHS(mask) { \
{ \
[0] = (u32)(mask & 0xffffffff), \
[1] = (u64)(mask) >> 32, \
} \
}
#else /* 32bit long above, 64bit long below */
#define __mlog_test_u64(mask, bits) ((mask) & bits.words[0])
#define __mlog_set_u64(mask, bits) do { \
bits.words[0] |= (mask); \
} while (0)
#define __mlog_clear_u64(mask, bits) do { \
bits.words[0] &= ~(mask); \
} while (0)
#define MLOG_BITS_RHS(mask) { { (mask) } }
#endif
/*
* smp_processor_id() "helpfully" screams when called outside preemptible
* regions in current kernels. sles doesn't have the variants that don't
* scream. just do this instead of trying to guess which we're building
* against.. *sigh*.
*/
#define __mlog_cpu_guess ({ \
unsigned long _cpu = get_cpu(); \
put_cpu(); \
_cpu; \
})
/* In the following two macros, the whitespace after the ',' just
* before ##args is intentional. Otherwise, gcc 2.95 will eat the
* previous token if args expands to nothing.
*/
#define __mlog_printk(level, fmt, args...) \
printk(level "(%s,%u,%lu):%s:%d " fmt, current->comm, \
task_pid_nr(current), __mlog_cpu_guess, \
__PRETTY_FUNCTION__, __LINE__ , ##args)
#define mlog(mask, fmt, args...) do { \
u64 __m = MLOG_MASK_PREFIX | (mask); \
if ((__m & ML_ALLOWED_BITS) && \
__mlog_test_u64(__m, mlog_and_bits) && \
!__mlog_test_u64(__m, mlog_not_bits)) { \
if (__m & ML_ERROR) \
__mlog_printk(KERN_ERR, "ERROR: "fmt , ##args); \
else if (__m & ML_NOTICE) \
__mlog_printk(KERN_NOTICE, fmt , ##args); \
else __mlog_printk(KERN_INFO, fmt , ##args); \
} \
} while (0)
#define mlog_errno(st) do { \
int _st = (st); \
if (_st != -ERESTARTSYS && _st != -EINTR && \
_st != AOP_TRUNCATED_PAGE && _st != -ENOSPC) \
mlog(ML_ERROR, "status = %lld\n", (long long)_st); \
} while (0)
#define mlog_bug_on_msg(cond, fmt, args...) do { \
if (cond) { \
mlog(ML_ERROR, "bug expression: " #cond "\n"); \
mlog(ML_ERROR, fmt, ##args); \
BUG(); \
} \
} while (0)
#include <linux/kobject.h>
#include <linux/sysfs.h>
int mlog_sys_init(struct kset *o2cb_subsys);
void mlog_sys_shutdown(void);
#endif /* O2CLUSTER_MASKLOG_H */
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* netdebug.c
*
* debug functionality for o2net
*
* Copyright (C) 2005, 2008 Oracle. 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.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*
*/
#ifdef CONFIG_DEBUG_FS
#include <linux/module.h>
#include <linux/types.h>
#include <linux/slab.h>
#include <linux/idr.h>
#include <linux/kref.h>
#include <linux/seq_file.h>
#include <linux/debugfs.h>
#include <linux/uaccess.h>
#include "tcp.h"
#include "nodemanager.h"
#define MLOG_MASK_PREFIX ML_TCP
#include "masklog.h"
#include "tcp_internal.h"
#define O2NET_DEBUG_DIR "o2net"
#define SC_DEBUG_NAME "sock_containers"
#define NST_DEBUG_NAME "send_tracking"
#define STATS_DEBUG_NAME "stats"
#define NODES_DEBUG_NAME "connected_nodes"
#define SHOW_SOCK_CONTAINERS 0
#define SHOW_SOCK_STATS 1
static struct dentry *o2net_dentry;
static struct dentry *sc_dentry;
static struct dentry *nst_dentry;
static struct dentry *stats_dentry;
static struct dentry *nodes_dentry;
static DEFINE_SPINLOCK(o2net_debug_lock);
static LIST_HEAD(sock_containers);
static LIST_HEAD(send_tracking);
void o2net_debug_add_nst(struct o2net_send_tracking *nst)
{
spin_lock(&o2net_debug_lock);
list_add(&nst->st_net_debug_item, &send_tracking);
spin_unlock(&o2net_debug_lock);
}
void o2net_debug_del_nst(struct o2net_send_tracking *nst)
{
spin_lock(&o2net_debug_lock);
if (!list_empty(&nst->st_net_debug_item))
list_del_init(&nst->st_net_debug_item);
spin_unlock(&o2net_debug_lock);
}
static struct o2net_send_tracking
*next_nst(struct o2net_send_tracking *nst_start)
{
struct o2net_send_tracking *nst, *ret = NULL;
assert_spin_locked(&o2net_debug_lock);
list_for_each_entry(nst, &nst_start->st_net_debug_item,
st_net_debug_item) {
/* discover the head of the list */
if (&nst->st_net_debug_item == &send_tracking)
break;
/* use st_task to detect real nsts in the list */
if (nst->st_task != NULL) {
ret = nst;
break;
}
}
return ret;
}
static void *nst_seq_start(struct seq_file *seq, loff_t *pos)
{
struct o2net_send_tracking *nst, *dummy_nst = seq->private;
spin_lock(&o2net_debug_lock);
nst = next_nst(dummy_nst);
spin_unlock(&o2net_debug_lock);
return nst;
}
static void *nst_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
struct o2net_send_tracking *nst, *dummy_nst = seq->private;
spin_lock(&o2net_debug_lock);
nst = next_nst(dummy_nst);
list_del_init(&dummy_nst->st_net_debug_item);
if (nst)
list_add(&dummy_nst->st_net_debug_item,
&nst->st_net_debug_item);
spin_unlock(&o2net_debug_lock);
return nst; /* unused, just needs to be null when done */
}
static int nst_seq_show(struct seq_file *seq, void *v)
{
struct o2net_send_tracking *nst, *dummy_nst = seq->private;
ktime_t now;
s64 sock, send, status;
spin_lock(&o2net_debug_lock);
nst = next_nst(dummy_nst);
if (!nst)
goto out;
now = ktime_get();
sock = ktime_to_us(ktime_sub(now, nst->st_sock_time));
send = ktime_to_us(ktime_sub(now, nst->st_send_time));
status = ktime_to_us(ktime_sub(now, nst->st_status_time));
/* get_task_comm isn't exported. oh well. */
seq_printf(seq, "%p:\n"
" pid: %lu\n"
" tgid: %lu\n"
" process name: %s\n"
" node: %u\n"
" sc: %p\n"
" message id: %d\n"
" message type: %u\n"
" message key: 0x%08x\n"
" sock acquiry: %lld usecs ago\n"
" send start: %lld usecs ago\n"
" wait start: %lld usecs ago\n",
nst, (unsigned long)task_pid_nr(nst->st_task),
(unsigned long)nst->st_task->tgid,
nst->st_task->comm, nst->st_node,
nst->st_sc, nst->st_id, nst->st_msg_type,
nst->st_msg_key,
(long long)sock,
(long long)send,
(long long)status);
out:
spin_unlock(&o2net_debug_lock);
return 0;
}
static void nst_seq_stop(struct seq_file *seq, void *v)
{
}
static const struct seq_operations nst_seq_ops = {
.start = nst_seq_start,
.next = nst_seq_next,
.stop = nst_seq_stop,
.show = nst_seq_show,
};
static int nst_fop_open(struct inode *inode, struct file *file)
{
struct o2net_send_tracking *dummy_nst;
struct seq_file *seq;
int ret;
dummy_nst = kmalloc(sizeof(struct o2net_send_tracking), GFP_KERNEL);
if (dummy_nst == NULL) {
ret = -ENOMEM;
goto out;
}
dummy_nst->st_task = NULL;
ret = seq_open(file, &nst_seq_ops);
if (ret)
goto out;
seq = file->private_data;
seq->private = dummy_nst;
o2net_debug_add_nst(dummy_nst);
dummy_nst = NULL;
out:
kfree(dummy_nst);
return ret;
}
static int nst_fop_release(struct inode *inode, struct file *file)
{
struct seq_file *seq = file->private_data;
struct o2net_send_tracking *dummy_nst = seq->private;
o2net_debug_del_nst(dummy_nst);
return seq_release_private(inode, file);
}
static const struct file_operations nst_seq_fops = {
.open = nst_fop_open,
.read = seq_read,
.llseek = seq_lseek,
.release = nst_fop_release,
};
void o2net_debug_add_sc(struct o2net_sock_container *sc)
{
spin_lock(&o2net_debug_lock);
list_add(&sc->sc_net_debug_item, &sock_containers);
spin_unlock(&o2net_debug_lock);
}
void o2net_debug_del_sc(struct o2net_sock_container *sc)
{
spin_lock(&o2net_debug_lock);
list_del_init(&sc->sc_net_debug_item);
spin_unlock(&o2net_debug_lock);
}
struct o2net_sock_debug {
int dbg_ctxt;
struct o2net_sock_container *dbg_sock;
};
static struct o2net_sock_container
*next_sc(struct o2net_sock_container *sc_start)
{
struct o2net_sock_container *sc, *ret = NULL;
assert_spin_locked(&o2net_debug_lock);
list_for_each_entry(sc, &sc_start->sc_net_debug_item,
sc_net_debug_item) {
/* discover the head of the list miscast as a sc */
if (&sc->sc_net_debug_item == &sock_containers)
break;
/* use sc_page to detect real scs in the list */
if (sc->sc_page != NULL) {
ret = sc;
break;
}
}
return ret;
}
static void *sc_seq_start(struct seq_file *seq, loff_t *pos)
{
struct o2net_sock_debug *sd = seq->private;
struct o2net_sock_container *sc, *dummy_sc = sd->dbg_sock;
spin_lock(&o2net_debug_lock);
sc = next_sc(dummy_sc);
spin_unlock(&o2net_debug_lock);
return sc;
}
static void *sc_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
struct o2net_sock_debug *sd = seq->private;
struct o2net_sock_container *sc, *dummy_sc = sd->dbg_sock;
spin_lock(&o2net_debug_lock);
sc = next_sc(dummy_sc);
list_del_init(&dummy_sc->sc_net_debug_item);
if (sc)
list_add(&dummy_sc->sc_net_debug_item, &sc->sc_net_debug_item);
spin_unlock(&o2net_debug_lock);
return sc; /* unused, just needs to be null when done */
}
#ifdef CONFIG_OCFS2_FS_STATS
# define sc_send_count(_s) ((_s)->sc_send_count)
# define sc_recv_count(_s) ((_s)->sc_recv_count)
# define sc_tv_acquiry_total_ns(_s) (ktime_to_ns((_s)->sc_tv_acquiry_total))
# define sc_tv_send_total_ns(_s) (ktime_to_ns((_s)->sc_tv_send_total))
# define sc_tv_status_total_ns(_s) (ktime_to_ns((_s)->sc_tv_status_total))
# define sc_tv_process_total_ns(_s) (ktime_to_ns((_s)->sc_tv_process_total))
#else
# define sc_send_count(_s) (0U)
# define sc_recv_count(_s) (0U)
# define sc_tv_acquiry_total_ns(_s) (0LL)
# define sc_tv_send_total_ns(_s) (0LL)
# define sc_tv_status_total_ns(_s) (0LL)
# define sc_tv_process_total_ns(_s) (0LL)
#endif
/* So that debugfs.ocfs2 can determine which format is being used */
#define O2NET_STATS_STR_VERSION 1
static void sc_show_sock_stats(struct seq_file *seq,
struct o2net_sock_container *sc)
{
if (!sc)
return;
seq_printf(seq, "%d,%u,%lu,%lld,%lld,%lld,%lu,%lld\n", O2NET_STATS_STR_VERSION,
sc->sc_node->nd_num, (unsigned long)sc_send_count(sc),
(long long)sc_tv_acquiry_total_ns(sc),
(long long)sc_tv_send_total_ns(sc),
(long long)sc_tv_status_total_ns(sc),
(unsigned long)sc_recv_count(sc),
(long long)sc_tv_process_total_ns(sc));
}
static void sc_show_sock_container(struct seq_file *seq,
struct o2net_sock_container *sc)
{
struct inet_sock *inet = NULL;
__be32 saddr = 0, daddr = 0;
__be16 sport = 0, dport = 0;
if (!sc)
return;
if (sc->sc_sock) {
inet = inet_sk(sc->sc_sock->sk);
/* the stack's structs aren't sparse endian clean */
saddr = (__force __be32)inet->inet_saddr;
daddr = (__force __be32)inet->inet_daddr;
sport = (__force __be16)inet->inet_sport;
dport = (__force __be16)inet->inet_dport;
}
/* XXX sigh, inet-> doesn't have sparse annotation so any
* use of it here generates a warning with -Wbitwise */
seq_printf(seq, "%p:\n"
" krefs: %d\n"
" sock: %pI4:%u -> "
"%pI4:%u\n"
" remote node: %s\n"
" page off: %zu\n"
" handshake ok: %u\n"
" timer: %lld usecs\n"
" data ready: %lld usecs\n"
" advance start: %lld usecs\n"
" advance stop: %lld usecs\n"
" func start: %lld usecs\n"
" func stop: %lld usecs\n"
" func key: 0x%08x\n"
" func type: %u\n",
sc,
atomic_read(&sc->sc_kref.refcount),
&saddr, inet ? ntohs(sport) : 0,
&daddr, inet ? ntohs(dport) : 0,
sc->sc_node->nd_name,
sc->sc_page_off,
sc->sc_handshake_ok,
(long long)ktime_to_us(sc->sc_tv_timer),
(long long)ktime_to_us(sc->sc_tv_data_ready),
(long long)ktime_to_us(sc->sc_tv_advance_start),
(long long)ktime_to_us(sc->sc_tv_advance_stop),
(long long)ktime_to_us(sc->sc_tv_func_start),
(long long)ktime_to_us(sc->sc_tv_func_stop),
sc->sc_msg_key,
sc->sc_msg_type);
}
static int sc_seq_show(struct seq_file *seq, void *v)
{
struct o2net_sock_debug *sd = seq->private;
struct o2net_sock_container *sc, *dummy_sc = sd->dbg_sock;
spin_lock(&o2net_debug_lock);
sc = next_sc(dummy_sc);
if (sc) {
if (sd->dbg_ctxt == SHOW_SOCK_CONTAINERS)
sc_show_sock_container(seq, sc);
else
sc_show_sock_stats(seq, sc);
}
spin_unlock(&o2net_debug_lock);
return 0;
}
static void sc_seq_stop(struct seq_file *seq, void *v)
{
}
static const struct seq_operations sc_seq_ops = {
.start = sc_seq_start,
.next = sc_seq_next,
.stop = sc_seq_stop,
.show = sc_seq_show,
};
static int sc_common_open(struct file *file, struct o2net_sock_debug *sd)
{
struct o2net_sock_container *dummy_sc;
struct seq_file *seq;
int ret;
dummy_sc = kmalloc(sizeof(struct o2net_sock_container), GFP_KERNEL);
if (dummy_sc == NULL) {
ret = -ENOMEM;
goto out;
}
dummy_sc->sc_page = NULL;
ret = seq_open(file, &sc_seq_ops);
if (ret)
goto out;
seq = file->private_data;
seq->private = sd;
sd->dbg_sock = dummy_sc;
o2net_debug_add_sc(dummy_sc);
dummy_sc = NULL;
out:
kfree(dummy_sc);
return ret;
}
static int sc_fop_release(struct inode *inode, struct file *file)
{
struct seq_file *seq = file->private_data;
struct o2net_sock_debug *sd = seq->private;
struct o2net_sock_container *dummy_sc = sd->dbg_sock;
o2net_debug_del_sc(dummy_sc);
return seq_release_private(inode, file);
}
static int stats_fop_open(struct inode *inode, struct file *file)
{
struct o2net_sock_debug *sd;
sd = kmalloc(sizeof(struct o2net_sock_debug), GFP_KERNEL);
if (sd == NULL)
return -ENOMEM;
sd->dbg_ctxt = SHOW_SOCK_STATS;
sd->dbg_sock = NULL;
return sc_common_open(file, sd);
}
static const struct file_operations stats_seq_fops = {
.open = stats_fop_open,
.read = seq_read,
.llseek = seq_lseek,
.release = sc_fop_release,
};
static int sc_fop_open(struct inode *inode, struct file *file)
{
struct o2net_sock_debug *sd;
sd = kmalloc(sizeof(struct o2net_sock_debug), GFP_KERNEL);
if (sd == NULL)
return -ENOMEM;
sd->dbg_ctxt = SHOW_SOCK_CONTAINERS;
sd->dbg_sock = NULL;
return sc_common_open(file, sd);
}
static const struct file_operations sc_seq_fops = {
.open = sc_fop_open,
.read = seq_read,
.llseek = seq_lseek,
.release = sc_fop_release,
};
static int o2net_fill_bitmap(char *buf, int len)
{
unsigned long map[BITS_TO_LONGS(O2NM_MAX_NODES)];
int i = -1, out = 0;
o2net_fill_node_map(map, sizeof(map));
while ((i = find_next_bit(map, O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES)
out += snprintf(buf + out, PAGE_SIZE - out, "%d ", i);
out += snprintf(buf + out, PAGE_SIZE - out, "\n");
return out;
}
static int nodes_fop_open(struct inode *inode, struct file *file)
{
char *buf;
buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
if (!buf)
return -ENOMEM;
i_size_write(inode, o2net_fill_bitmap(buf, PAGE_SIZE));
file->private_data = buf;
return 0;
}
static int o2net_debug_release(struct inode *inode, struct file *file)
{
kfree(file->private_data);
return 0;
}
static ssize_t o2net_debug_read(struct file *file, char __user *buf,
size_t nbytes, loff_t *ppos)
{
return simple_read_from_buffer(buf, nbytes, ppos, file->private_data,
i_size_read(file->f_mapping->host));
}
static const struct file_operations nodes_fops = {
.open = nodes_fop_open,
.release = o2net_debug_release,
.read = o2net_debug_read,
.llseek = generic_file_llseek,
};
void o2net_debugfs_exit(void)
{
debugfs_remove(nodes_dentry);
debugfs_remove(stats_dentry);
debugfs_remove(sc_dentry);
debugfs_remove(nst_dentry);
debugfs_remove(o2net_dentry);
}
int o2net_debugfs_init(void)
{
mode_t mode = S_IFREG|S_IRUSR;
o2net_dentry = debugfs_create_dir(O2NET_DEBUG_DIR, NULL);
if (o2net_dentry)
nst_dentry = debugfs_create_file(NST_DEBUG_NAME, mode,
o2net_dentry, NULL, &nst_seq_fops);
if (nst_dentry)
sc_dentry = debugfs_create_file(SC_DEBUG_NAME, mode,
o2net_dentry, NULL, &sc_seq_fops);
if (sc_dentry)
stats_dentry = debugfs_create_file(STATS_DEBUG_NAME, mode,
o2net_dentry, NULL, &stats_seq_fops);
if (stats_dentry)
nodes_dentry = debugfs_create_file(NODES_DEBUG_NAME, mode,
o2net_dentry, NULL, &nodes_fops);
if (nodes_dentry)
return 0;
o2net_debugfs_exit();
mlog_errno(-ENOMEM);
return -ENOMEM;
}
#endif /* CONFIG_DEBUG_FS */
此差异已折叠。
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* nodemanager.h
*
* Function prototypes
*
* Copyright (C) 2004 Oracle. 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.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*
*/
#ifndef O2CLUSTER_NODEMANAGER_H
#define O2CLUSTER_NODEMANAGER_H
#include "ocfs2_nodemanager.h"
/* This totally doesn't belong here. */
#include <linux/configfs.h>
#include <linux/rbtree.h>
enum o2nm_fence_method {
O2NM_FENCE_RESET = 0,
O2NM_FENCE_PANIC,
O2NM_FENCE_METHODS, /* Number of fence methods */
};
struct o2nm_node {
spinlock_t nd_lock;
struct config_item nd_item;
char nd_name[O2NM_MAX_NAME_LEN+1]; /* replace? */
__u8 nd_num;
/* only one address per node, as attributes, for now. */
__be32 nd_ipv4_address;
__be16 nd_ipv4_port;
struct rb_node nd_ip_node;
/* there can be only one local node for now */
int nd_local;
unsigned long nd_set_attributes;
};
struct o2nm_cluster {
struct config_group cl_group;
unsigned cl_has_local:1;
u8 cl_local_node;
rwlock_t cl_nodes_lock;
struct o2nm_node *cl_nodes[O2NM_MAX_NODES];
struct rb_root cl_node_ip_tree;
unsigned int cl_idle_timeout_ms;
unsigned int cl_keepalive_delay_ms;
unsigned int cl_reconnect_delay_ms;
enum o2nm_fence_method cl_fence_method;
/* this bitmap is part of a hack for disk bitmap.. will go eventually. - zab */
unsigned long cl_nodes_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)];
};
extern struct o2nm_cluster *o2nm_single_cluster;
u8 o2nm_this_node(void);
int o2nm_configured_node_map(unsigned long *map, unsigned bytes);
struct o2nm_node *o2nm_get_node_by_num(u8 node_num);
struct o2nm_node *o2nm_get_node_by_ip(__be32 addr);
void o2nm_node_get(struct o2nm_node *node);
void o2nm_node_put(struct o2nm_node *node);
int o2nm_depend_item(struct config_item *item);
void o2nm_undepend_item(struct config_item *item);
int o2nm_depend_this_node(void);
void o2nm_undepend_this_node(void);
#endif /* O2CLUSTER_NODEMANAGER_H */
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* ocfs2_heartbeat.h
*
* On-disk structures for ocfs2_heartbeat
*
* Copyright (C) 2002, 2004 Oracle. 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.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*/
#ifndef _OCFS2_HEARTBEAT_H
#define _OCFS2_HEARTBEAT_H
struct o2hb_disk_heartbeat_block {
__le64 hb_seq;
__u8 hb_node;
__u8 hb_pad1[3];
__le32 hb_cksum;
__le64 hb_generation;
__le32 hb_dead_ms;
};
#endif /* _OCFS2_HEARTBEAT_H */
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* ocfs2_nodemanager.h
*
* Header describing the interface between userspace and the kernel
* for the ocfs2_nodemanager module.
*
* Copyright (C) 2002, 2004 Oracle. 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.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*
*/
#ifndef _OCFS2_NODEMANAGER_H
#define _OCFS2_NODEMANAGER_H
#define O2NM_API_VERSION 5
#define O2NM_MAX_NODES 255
#define O2NM_INVALID_NODE_NUM 255
/* host name, group name, cluster name all 64 bytes */
#define O2NM_MAX_NAME_LEN 64 // __NEW_UTS_LEN
/*
* Maximum number of global heartbeat regions allowed.
* **CAUTION** Changing this number will break dlm compatibility.
*/
#define O2NM_MAX_REGIONS 32
#endif /* _OCFS2_NODEMANAGER_H */
此差异已折叠。
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* Copyright (C) 2005 Oracle. 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.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*
*/
#ifndef O2CLUSTER_QUORUM_H
#define O2CLUSTER_QUORUM_H
void o2quo_init(void);
void o2quo_exit(void);
void o2quo_hb_up(u8 node);
void o2quo_hb_down(u8 node);
void o2quo_hb_still_up(u8 node);
void o2quo_conn_up(u8 node);
void o2quo_conn_err(u8 node);
void o2quo_disk_timeout(void);
#endif /* O2CLUSTER_QUORUM_H */
此差异已折叠。
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* sys.h
*
* Function prototypes for o2cb sysfs interface
*
* Copyright (C) 2005 Oracle. 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,
* version 2 of the License.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*
*/
#ifndef O2CLUSTER_SYS_H
#define O2CLUSTER_SYS_H
void o2cb_sys_shutdown(void);
int o2cb_sys_init(void);
#endif /* O2CLUSTER_SYS_H */
此差异已折叠。
此差异已折叠。
此差异已折叠。
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* ver.c
*
* version string
*
* Copyright (C) 2002, 2005 Oracle. 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.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*/
#include <linux/module.h>
#include <linux/kernel.h>
#include "ver.h"
#define CLUSTER_BUILD_VERSION "1.5.0"
#define VERSION_STR "OCFS2 Node Manager " CLUSTER_BUILD_VERSION
void cluster_print_version(void)
{
printk(KERN_INFO "%s\n", VERSION_STR);
}
MODULE_DESCRIPTION(VERSION_STR);
MODULE_VERSION(CLUSTER_BUILD_VERSION);
此差异已折叠。
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册