kernel_optimize_test/fs/lockd/svcsubs.c
Jeff Layton 1c8c601a8c locks: protect most of the file_lock handling with i_lock
Having a global lock that protects all of this code is a clear
scalability problem. Instead of doing that, move most of the code to be
protected by the i_lock instead. The exceptions are the global lists
that the ->fl_link sits on, and the ->fl_block list.

->fl_link is what connects these structures to the
global lists, so we must ensure that we hold those locks when iterating
over or updating these lists.

Furthermore, sound deadlock detection requires that we hold the
blocked_list state steady while checking for loops. We also must ensure
that the search and update to the list are atomic.

For the checking and insertion side of the blocked_list, push the
acquisition of the global lock into __posix_lock_file and ensure that
checking and update of the  blocked_list is done without dropping the
lock in between.

On the removal side, when waking up blocked lock waiters, take the
global lock before walking the blocked list and dequeue the waiters from
the global list prior to removal from the fl_block list.

With this, deadlock detection should be race free while we minimize
excessive file_lock_lock thrashing.

Finally, in order to avoid a lock inversion problem when handling
/proc/locks output we must ensure that manipulations of the fl_block
list are also protected by the file_lock_lock.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2013-06-29 12:57:42 +04:00

453 lines
10 KiB
C

/*
* linux/fs/lockd/svcsubs.c
*
* Various support routines for the NLM server.
*
* Copyright (C) 1996, Olaf Kirch <okir@monad.swb.de>
*/
#include <linux/types.h>
#include <linux/string.h>
#include <linux/time.h>
#include <linux/in.h>
#include <linux/slab.h>
#include <linux/mutex.h>
#include <linux/sunrpc/svc.h>
#include <linux/sunrpc/addr.h>
#include <linux/nfsd/nfsfh.h>
#include <linux/nfsd/export.h>
#include <linux/lockd/lockd.h>
#include <linux/lockd/share.h>
#include <linux/module.h>
#include <linux/mount.h>
#define NLMDBG_FACILITY NLMDBG_SVCSUBS
/*
* Global file hash table
*/
#define FILE_HASH_BITS 7
#define FILE_NRHASH (1<<FILE_HASH_BITS)
static struct hlist_head nlm_files[FILE_NRHASH];
static DEFINE_MUTEX(nlm_file_mutex);
#ifdef NFSD_DEBUG
static inline void nlm_debug_print_fh(char *msg, struct nfs_fh *f)
{
u32 *fhp = (u32*)f->data;
/* print the first 32 bytes of the fh */
dprintk("lockd: %s (%08x %08x %08x %08x %08x %08x %08x %08x)\n",
msg, fhp[0], fhp[1], fhp[2], fhp[3],
fhp[4], fhp[5], fhp[6], fhp[7]);
}
static inline void nlm_debug_print_file(char *msg, struct nlm_file *file)
{
struct inode *inode = file_inode(file->f_file);
dprintk("lockd: %s %s/%ld\n",
msg, inode->i_sb->s_id, inode->i_ino);
}
#else
static inline void nlm_debug_print_fh(char *msg, struct nfs_fh *f)
{
return;
}
static inline void nlm_debug_print_file(char *msg, struct nlm_file *file)
{
return;
}
#endif
static inline unsigned int file_hash(struct nfs_fh *f)
{
unsigned int tmp=0;
int i;
for (i=0; i<NFS2_FHSIZE;i++)
tmp += f->data[i];
return tmp & (FILE_NRHASH - 1);
}
/*
* Lookup file info. If it doesn't exist, create a file info struct
* and open a (VFS) file for the given inode.
*
* FIXME:
* Note that we open the file O_RDONLY even when creating write locks.
* This is not quite right, but for now, we assume the client performs
* the proper R/W checking.
*/
__be32
nlm_lookup_file(struct svc_rqst *rqstp, struct nlm_file **result,
struct nfs_fh *f)
{
struct nlm_file *file;
unsigned int hash;
__be32 nfserr;
nlm_debug_print_fh("nlm_lookup_file", f);
hash = file_hash(f);
/* Lock file table */
mutex_lock(&nlm_file_mutex);
hlist_for_each_entry(file, &nlm_files[hash], f_list)
if (!nfs_compare_fh(&file->f_handle, f))
goto found;
nlm_debug_print_fh("creating file for", f);
nfserr = nlm_lck_denied_nolocks;
file = kzalloc(sizeof(*file), GFP_KERNEL);
if (!file)
goto out_unlock;
memcpy(&file->f_handle, f, sizeof(struct nfs_fh));
mutex_init(&file->f_mutex);
INIT_HLIST_NODE(&file->f_list);
INIT_LIST_HEAD(&file->f_blocks);
/* Open the file. Note that this must not sleep for too long, else
* we would lock up lockd:-) So no NFS re-exports, folks.
*
* We have to make sure we have the right credential to open
* the file.
*/
if ((nfserr = nlmsvc_ops->fopen(rqstp, f, &file->f_file)) != 0) {
dprintk("lockd: open failed (error %d)\n", nfserr);
goto out_free;
}
hlist_add_head(&file->f_list, &nlm_files[hash]);
found:
dprintk("lockd: found file %p (count %d)\n", file, file->f_count);
*result = file;
file->f_count++;
nfserr = 0;
out_unlock:
mutex_unlock(&nlm_file_mutex);
return nfserr;
out_free:
kfree(file);
goto out_unlock;
}
/*
* Delete a file after having released all locks, blocks and shares
*/
static inline void
nlm_delete_file(struct nlm_file *file)
{
nlm_debug_print_file("closing file", file);
if (!hlist_unhashed(&file->f_list)) {
hlist_del(&file->f_list);
nlmsvc_ops->fclose(file->f_file);
kfree(file);
} else {
printk(KERN_WARNING "lockd: attempt to release unknown file!\n");
}
}
/*
* Loop over all locks on the given file and perform the specified
* action.
*/
static int
nlm_traverse_locks(struct nlm_host *host, struct nlm_file *file,
nlm_host_match_fn_t match)
{
struct inode *inode = nlmsvc_file_inode(file);
struct file_lock *fl;
struct nlm_host *lockhost;
again:
file->f_locks = 0;
spin_lock(&inode->i_lock);
for (fl = inode->i_flock; fl; fl = fl->fl_next) {
if (fl->fl_lmops != &nlmsvc_lock_operations)
continue;
/* update current lock count */
file->f_locks++;
lockhost = (struct nlm_host *) fl->fl_owner;
if (match(lockhost, host)) {
struct file_lock lock = *fl;
spin_unlock(&inode->i_lock);
lock.fl_type = F_UNLCK;
lock.fl_start = 0;
lock.fl_end = OFFSET_MAX;
if (vfs_lock_file(file->f_file, F_SETLK, &lock, NULL) < 0) {
printk("lockd: unlock failure in %s:%d\n",
__FILE__, __LINE__);
return 1;
}
goto again;
}
}
spin_unlock(&inode->i_lock);
return 0;
}
static int
nlmsvc_always_match(void *dummy1, struct nlm_host *dummy2)
{
return 1;
}
/*
* Inspect a single file
*/
static inline int
nlm_inspect_file(struct nlm_host *host, struct nlm_file *file, nlm_host_match_fn_t match)
{
nlmsvc_traverse_blocks(host, file, match);
nlmsvc_traverse_shares(host, file, match);
return nlm_traverse_locks(host, file, match);
}
/*
* Quick check whether there are still any locks, blocks or
* shares on a given file.
*/
static inline int
nlm_file_inuse(struct nlm_file *file)
{
struct inode *inode = nlmsvc_file_inode(file);
struct file_lock *fl;
if (file->f_count || !list_empty(&file->f_blocks) || file->f_shares)
return 1;
spin_lock(&inode->i_lock);
for (fl = inode->i_flock; fl; fl = fl->fl_next) {
if (fl->fl_lmops == &nlmsvc_lock_operations) {
spin_unlock(&inode->i_lock);
return 1;
}
}
spin_unlock(&inode->i_lock);
file->f_locks = 0;
return 0;
}
/*
* Loop over all files in the file table.
*/
static int
nlm_traverse_files(void *data, nlm_host_match_fn_t match,
int (*is_failover_file)(void *data, struct nlm_file *file))
{
struct hlist_node *next;
struct nlm_file *file;
int i, ret = 0;
mutex_lock(&nlm_file_mutex);
for (i = 0; i < FILE_NRHASH; i++) {
hlist_for_each_entry_safe(file, next, &nlm_files[i], f_list) {
if (is_failover_file && !is_failover_file(data, file))
continue;
file->f_count++;
mutex_unlock(&nlm_file_mutex);
/* Traverse locks, blocks and shares of this file
* and update file->f_locks count */
if (nlm_inspect_file(data, file, match))
ret = 1;
mutex_lock(&nlm_file_mutex);
file->f_count--;
/* No more references to this file. Let go of it. */
if (list_empty(&file->f_blocks) && !file->f_locks
&& !file->f_shares && !file->f_count) {
hlist_del(&file->f_list);
nlmsvc_ops->fclose(file->f_file);
kfree(file);
}
}
}
mutex_unlock(&nlm_file_mutex);
return ret;
}
/*
* Release file. If there are no more remote locks on this file,
* close it and free the handle.
*
* Note that we can't do proper reference counting without major
* contortions because the code in fs/locks.c creates, deletes and
* splits locks without notification. Our only way is to walk the
* entire lock list each time we remove a lock.
*/
void
nlm_release_file(struct nlm_file *file)
{
dprintk("lockd: nlm_release_file(%p, ct = %d)\n",
file, file->f_count);
/* Lock file table */
mutex_lock(&nlm_file_mutex);
/* If there are no more locks etc, delete the file */
if (--file->f_count == 0 && !nlm_file_inuse(file))
nlm_delete_file(file);
mutex_unlock(&nlm_file_mutex);
}
/*
* Helpers function for resource traversal
*
* nlmsvc_mark_host:
* used by the garbage collector; simply sets h_inuse only for those
* hosts, which passed network check.
* Always returns 0.
*
* nlmsvc_same_host:
* returns 1 iff the two hosts match. Used to release
* all resources bound to a specific host.
*
* nlmsvc_is_client:
* returns 1 iff the host is a client.
* Used by nlmsvc_invalidate_all
*/
static int
nlmsvc_mark_host(void *data, struct nlm_host *hint)
{
struct nlm_host *host = data;
if ((hint->net == NULL) ||
(host->net == hint->net))
host->h_inuse = 1;
return 0;
}
static int
nlmsvc_same_host(void *data, struct nlm_host *other)
{
struct nlm_host *host = data;
return host == other;
}
static int
nlmsvc_is_client(void *data, struct nlm_host *dummy)
{
struct nlm_host *host = data;
if (host->h_server) {
/* we are destroying locks even though the client
* hasn't asked us too, so don't unmonitor the
* client
*/
if (host->h_nsmhandle)
host->h_nsmhandle->sm_sticky = 1;
return 1;
} else
return 0;
}
/*
* Mark all hosts that still hold resources
*/
void
nlmsvc_mark_resources(struct net *net)
{
struct nlm_host hint;
dprintk("lockd: nlmsvc_mark_resources for net %p\n", net);
hint.net = net;
nlm_traverse_files(&hint, nlmsvc_mark_host, NULL);
}
/*
* Release all resources held by the given client
*/
void
nlmsvc_free_host_resources(struct nlm_host *host)
{
dprintk("lockd: nlmsvc_free_host_resources\n");
if (nlm_traverse_files(host, nlmsvc_same_host, NULL)) {
printk(KERN_WARNING
"lockd: couldn't remove all locks held by %s\n",
host->h_name);
BUG();
}
}
/**
* nlmsvc_invalidate_all - remove all locks held for clients
*
* Release all locks held by NFS clients.
*
*/
void
nlmsvc_invalidate_all(void)
{
/*
* Previously, the code would call
* nlmsvc_free_host_resources for each client in
* turn, which is about as inefficient as it gets.
* Now we just do it once in nlm_traverse_files.
*/
nlm_traverse_files(NULL, nlmsvc_is_client, NULL);
}
static int
nlmsvc_match_sb(void *datap, struct nlm_file *file)
{
struct super_block *sb = datap;
return sb == file->f_file->f_path.dentry->d_sb;
}
/**
* nlmsvc_unlock_all_by_sb - release locks held on this file system
* @sb: super block
*
* Release all locks held by clients accessing this file system.
*/
int
nlmsvc_unlock_all_by_sb(struct super_block *sb)
{
int ret;
ret = nlm_traverse_files(sb, nlmsvc_always_match, nlmsvc_match_sb);
return ret ? -EIO : 0;
}
EXPORT_SYMBOL_GPL(nlmsvc_unlock_all_by_sb);
static int
nlmsvc_match_ip(void *datap, struct nlm_host *host)
{
return rpc_cmp_addr(nlm_srcaddr(host), datap);
}
/**
* nlmsvc_unlock_all_by_ip - release local locks by IP address
* @server_addr: server's IP address as seen by clients
*
* Release all locks held by clients accessing this host
* via the passed in IP address.
*/
int
nlmsvc_unlock_all_by_ip(struct sockaddr *server_addr)
{
int ret;
ret = nlm_traverse_files(server_addr, nlmsvc_match_ip, NULL);
return ret ? -EIO : 0;
}
EXPORT_SYMBOL_GPL(nlmsvc_unlock_all_by_ip);