korg/linux
0
0
Fork 0
mirror of https://mirrors.bfsu.edu.cn/git/linux.git synced 2024-12-15 06:55:13 +08:00
Code Issues Packages Projects Releases Wiki Activity
6b54698aec
linux/fs/gfs2/glock.h
Bob Peterson f2e70d8f2f gfs2: fix GL_SKIP node_scope problems 
Before this patch, when a glock was locked, the very first holder on the
queue would unlock the lockref and call the go_instantiate glops function
(if one existed), unless GL_SKIP was specified. When we introduced the new
node-scope concept, we allowed multiple holders to lock glocks in EX mode
and share the lock.

But node-scope introduced a new problem: if the first holder has GL_SKIP
and the next one does NOT, since it is not the first holder on the queue,
the go_instantiate op was not called. Eventually the GL_SKIP holder may
call the instantiate sub-function (e.g. gfs2_rgrp_bh_get) but there was
still a window of time in which another non-GL_SKIP holder assumes the
instantiate function had been called by the first holder. In the case of
rgrp glocks, this led to a NULL pointer dereference on the buffer_heads.

This patch tries to fix the problem by introducing two new glock flags:

GLF_INSTANTIATE_NEEDED, which keeps track of when the instantiate function
needs to be called to "fill in" or "read in" the object before it is
referenced.

GLF_INSTANTIATE_IN_PROG which is used to determine when a process is
in the process of reading in the object. Whenever a function needs to
reference the object, it checks the GLF_INSTANTIATE_NEEDED flag, and if
set, it sets GLF_INSTANTIATE_IN_PROG and calls the glops "go_instantiate"
function.

As before, the gl_lockref spin_lock is unlocked during the IO operation,
which may take a relatively long amount of time to complete. While
unlocked, if another process determines go_instantiate is still needed,
it sees GLF_INSTANTIATE_IN_PROG is set, and waits for the go_instantiate
glop operation to be completed. Once GLF_INSTANTIATE_IN_PROG is cleared,
it needs to check GLF_INSTANTIATE_NEEDED again because the other process's
go_instantiate operation may not have been successful.

Functions that previously called the instantiate sub-functions now call
directly into gfs2_instantiate so the new bits are managed properly.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2021-10-25 08:42:19 +02:00

360 lines
11 KiB
C
Raw Blame History

/* SPDX-License-Identifier: GPL-2.0-only */
/*
* Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
* Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
*/
#ifndef __GLOCK_DOT_H__
#define __GLOCK_DOT_H__
#include <linux/sched.h>
#include <linux/parser.h>
#include "incore.h"
#include "util.h"
/* Options for hostdata parser */
enum {
Opt_jid,
Opt_id,
Opt_first,
Opt_nodir,
Opt_err,
};
/*
* lm_lockname types
*/
#define LM_TYPE_RESERVED 0x00
#define LM_TYPE_NONDISK 0x01
#define LM_TYPE_INODE 0x02
#define LM_TYPE_RGRP 0x03
#define LM_TYPE_META 0x04
#define LM_TYPE_IOPEN 0x05
#define LM_TYPE_FLOCK 0x06
#define LM_TYPE_PLOCK 0x07
#define LM_TYPE_QUOTA 0x08
#define LM_TYPE_JOURNAL 0x09
/*
* lm_lock() states
*
* SHARED is compatible with SHARED, not with DEFERRED or EX.
* DEFERRED is compatible with DEFERRED, not with SHARED or EX.
*/
#define LM_ST_UNLOCKED 0
#define LM_ST_EXCLUSIVE 1
#define LM_ST_DEFERRED 2
#define LM_ST_SHARED 3
/*
* lm_lock() flags
*
* LM_FLAG_TRY
* Don't wait to acquire the lock if it can't be granted immediately.
*
* LM_FLAG_TRY_1CB
* Send one blocking callback if TRY is set and the lock is not granted.
*
* LM_FLAG_NOEXP
* GFS sets this flag on lock requests it makes while doing journal recovery.
* These special requests should not be blocked due to the recovery like
* ordinary locks would be.
*
* LM_FLAG_ANY
* A SHARED request may also be granted in DEFERRED, or a DEFERRED request may
* also be granted in SHARED. The preferred state is whichever is compatible
* with other granted locks, or the specified state if no other locks exist.
*
* LM_FLAG_PRIORITY
* Override fairness considerations. Suppose a lock is held in a shared state
* and there is a pending request for the deferred state. A shared lock
* request with the priority flag would be allowed to bypass the deferred
* request and directly join the other shared lock. A shared lock request
* without the priority flag might be forced to wait until the deferred
* requested had acquired and released the lock.
*
* LM_FLAG_NODE_SCOPE
* This holder agrees to share the lock within this node. In other words,
* the glock is held in EX mode according to DLM, but local holders on the
* same node can share it.
*/
#define LM_FLAG_TRY 0x0001
#define LM_FLAG_TRY_1CB 0x0002
#define LM_FLAG_NOEXP 0x0004
#define LM_FLAG_ANY 0x0008
#define LM_FLAG_PRIORITY 0x0010
#define LM_FLAG_NODE_SCOPE 0x0020
#define GL_ASYNC 0x0040
#define GL_EXACT 0x0080
#define GL_SKIP 0x0100
#define GL_NOCACHE 0x0400
/*
* lm_async_cb return flags
*
* LM_OUT_ST_MASK
* Masks the lower two bits of lock state in the returned value.
*
* LM_OUT_CANCELED
* The lock request was canceled.
*
*/
#define LM_OUT_ST_MASK 0x00000003
#define LM_OUT_CANCELED 0x00000008
#define LM_OUT_ERROR 0x00000004
/*
* lm_recovery_done() messages
*/
#define LM_RD_GAVEUP 308
#define LM_RD_SUCCESS 309
#define GLR_TRYFAILED 13
#define GL_GLOCK_MAX_HOLD (long)(HZ / 5)
#define GL_GLOCK_DFT_HOLD (long)(HZ / 5)
#define GL_GLOCK_MIN_HOLD (long)(10)
#define GL_GLOCK_HOLD_INCR (long)(HZ / 20)
#define GL_GLOCK_HOLD_DECR (long)(HZ / 40)
struct lm_lockops {
const char *lm_proto_name;
int (*lm_mount) (struct gfs2_sbd *sdp, const char *table);
void (*lm_first_done) (struct gfs2_sbd *sdp);
void (*lm_recovery_result) (struct gfs2_sbd *sdp, unsigned int jid,
unsigned int result);
void (*lm_unmount) (struct gfs2_sbd *sdp);
void (*lm_withdraw) (struct gfs2_sbd *sdp);
void (*lm_put_lock) (struct gfs2_glock *gl);
int (*lm_lock) (struct gfs2_glock *gl, unsigned int req_state,
unsigned int flags);
void (*lm_cancel) (struct gfs2_glock *gl);
const match_table_t *lm_tokens;
};
extern struct workqueue_struct *gfs2_delete_workqueue;
static inline struct gfs2_holder *gfs2_glock_is_locked_by_me(struct gfs2_glock *gl)
{
struct gfs2_holder *gh;
struct pid *pid;
/* Look in glock's list of holders for one with current task as owner */
spin_lock(&gl->gl_lockref.lock);
pid = task_pid(current);
list_for_each_entry(gh, &gl->gl_holders, gh_list) {
if (!test_bit(HIF_HOLDER, &gh->gh_iflags))
break;
if (test_bit(HIF_MAY_DEMOTE, &gh->gh_iflags))
continue;
if (gh->gh_owner_pid == pid)
goto out;
}
gh = NULL;
out:
spin_unlock(&gl->gl_lockref.lock);
return gh;
}
static inline int gfs2_glock_is_held_excl(struct gfs2_glock *gl)
{
return gl->gl_state == LM_ST_EXCLUSIVE;
}
static inline int gfs2_glock_is_held_dfrd(struct gfs2_glock *gl)
{
return gl->gl_state == LM_ST_DEFERRED;
}
static inline int gfs2_glock_is_held_shrd(struct gfs2_glock *gl)
{
return gl->gl_state == LM_ST_SHARED;
}
static inline struct address_space *gfs2_glock2aspace(struct gfs2_glock *gl)
{
if (gl->gl_ops->go_flags & GLOF_ASPACE)
return (struct address_space *)(gl + 1);
return NULL;
}
extern int gfs2_glock_get(struct gfs2_sbd *sdp, u64 number,
const struct gfs2_glock_operations *glops,
int create, struct gfs2_glock **glp);
extern void gfs2_glock_hold(struct gfs2_glock *gl);
extern void gfs2_glock_put(struct gfs2_glock *gl);
extern void gfs2_glock_queue_put(struct gfs2_glock *gl);
extern void __gfs2_holder_init(struct gfs2_glock *gl, unsigned int state,
u16 flags, struct gfs2_holder *gh,
unsigned long ip);
static inline void gfs2_holder_init(struct gfs2_glock *gl, unsigned int state,
u16 flags, struct gfs2_holder *gh) {
__gfs2_holder_init(gl, state, flags, gh, _RET_IP_);
}
extern void gfs2_holder_reinit(unsigned int state, u16 flags,
struct gfs2_holder *gh);
extern void gfs2_holder_uninit(struct gfs2_holder *gh);
extern int gfs2_glock_nq(struct gfs2_holder *gh);
extern int gfs2_glock_poll(struct gfs2_holder *gh);
extern int gfs2_instantiate(struct gfs2_holder *gh);
extern int gfs2_glock_wait(struct gfs2_holder *gh);
extern int gfs2_glock_async_wait(unsigned int num_gh, struct gfs2_holder *ghs);
extern void gfs2_glock_dq(struct gfs2_holder *gh);
extern void gfs2_glock_dq_wait(struct gfs2_holder *gh);
extern void gfs2_glock_dq_uninit(struct gfs2_holder *gh);
extern int gfs2_glock_nq_num(struct gfs2_sbd *sdp, u64 number,
const struct gfs2_glock_operations *glops,
unsigned int state, u16 flags,
struct gfs2_holder *gh);
extern int gfs2_glock_nq_m(unsigned int num_gh, struct gfs2_holder *ghs);
extern void gfs2_glock_dq_m(unsigned int num_gh, struct gfs2_holder *ghs);
extern void gfs2_dump_glock(struct seq_file *seq, struct gfs2_glock *gl,
bool fsid);
#define GLOCK_BUG_ON(gl,x) do { if (unlikely(x)) { \
gfs2_dump_glock(NULL, gl, true); \
BUG(); } } while(0)
#define gfs2_glock_assert_warn(gl, x) do { if (unlikely(!(x))) { \
gfs2_dump_glock(NULL, gl, true); \
gfs2_assert_warn((gl)->gl_name.ln_sbd, (x)); } } \
while (0)
#define gfs2_glock_assert_withdraw(gl, x) do { if (unlikely(!(x))) { \
gfs2_dump_glock(NULL, gl, true); \
gfs2_assert_withdraw((gl)->gl_name.ln_sbd, (x)); } } \
while (0)
extern __printf(2, 3)
void gfs2_print_dbg(struct seq_file *seq, const char *fmt, ...);
/**
* gfs2_glock_nq_init - initialize a holder and enqueue it on a glock
* @gl: the glock
* @state: the state we're requesting
* @flags: the modifier flags
* @gh: the holder structure
*
* Returns: 0, GLR_*, or errno
*/
static inline int gfs2_glock_nq_init(struct gfs2_glock *gl,
unsigned int state, u16 flags,
struct gfs2_holder *gh)
{
int error;
__gfs2_holder_init(gl, state, flags, gh, _RET_IP_);
error = gfs2_glock_nq(gh);
if (error)
gfs2_holder_uninit(gh);
return error;
}
extern void gfs2_glock_cb(struct gfs2_glock *gl, unsigned int state);
extern void gfs2_glock_complete(struct gfs2_glock *gl, int ret);
extern bool gfs2_queue_delete_work(struct gfs2_glock *gl, unsigned long delay);
extern void gfs2_cancel_delete_work(struct gfs2_glock *gl);
extern bool gfs2_delete_work_queued(const struct gfs2_glock *gl);
extern void gfs2_flush_delete_work(struct gfs2_sbd *sdp);
extern void gfs2_gl_hash_clear(struct gfs2_sbd *sdp);
extern void gfs2_glock_finish_truncate(struct gfs2_inode *ip);
extern void gfs2_glock_thaw(struct gfs2_sbd *sdp);
extern void gfs2_glock_add_to_lru(struct gfs2_glock *gl);
extern void gfs2_glock_free(struct gfs2_glock *gl);
extern int __init gfs2_glock_init(void);
extern void gfs2_glock_exit(void);
extern void gfs2_create_debugfs_file(struct gfs2_sbd *sdp);
extern void gfs2_delete_debugfs_file(struct gfs2_sbd *sdp);
extern void gfs2_register_debugfs(void);
extern void gfs2_unregister_debugfs(void);
extern const struct lm_lockops gfs2_dlm_ops;
static inline void gfs2_holder_mark_uninitialized(struct gfs2_holder *gh)
{
gh->gh_gl = NULL;
}
static inline bool gfs2_holder_initialized(struct gfs2_holder *gh)
{
return gh->gh_gl;
}
static inline bool gfs2_holder_queued(struct gfs2_holder *gh)
{
return !list_empty(&gh->gh_list);
}
/**
* glock_set_object - set the gl_object field of a glock
* @gl: the glock
* @object: the object
*/
static inline void glock_set_object(struct gfs2_glock *gl, void *object)
{
spin_lock(&gl->gl_lockref.lock);
if (gfs2_assert_warn(gl->gl_name.ln_sbd, gl->gl_object == NULL))
gfs2_dump_glock(NULL, gl, true);
gl->gl_object = object;
spin_unlock(&gl->gl_lockref.lock);
}
/**
* glock_clear_object - clear the gl_object field of a glock
* @gl: the glock
* @object: the object
*
* I'd love to similarly add this:
* else if (gfs2_assert_warn(gl->gl_sbd, gl->gl_object == object))
* gfs2_dump_glock(NULL, gl, true);
* Unfortunately, that's not possible because as soon as gfs2_delete_inode
* frees the block in the rgrp, another process can reassign it for an I_NEW
* inode in gfs2_create_inode because that calls new_inode, not gfs2_iget.
* That means gfs2_delete_inode may subsequently try to call this function
* for a glock that's already pointing to a brand new inode. If we clear the
* new inode's gl_object, we'll introduce metadata corruption. Function
* gfs2_delete_inode calls clear_inode which calls gfs2_clear_inode which also
* tries to clear gl_object, so it's more than just gfs2_delete_inode.
*
*/
static inline void glock_clear_object(struct gfs2_glock *gl, void *object)
{
spin_lock(&gl->gl_lockref.lock);
if (gl->gl_object == object)
gl->gl_object = NULL;
spin_unlock(&gl->gl_lockref.lock);
}
static inline void gfs2_holder_allow_demote(struct gfs2_holder *gh)
{
struct gfs2_glock *gl = gh->gh_gl;
spin_lock(&gl->gl_lockref.lock);
set_bit(HIF_MAY_DEMOTE, &gh->gh_iflags);
spin_unlock(&gl->gl_lockref.lock);
}
static inline void gfs2_holder_disallow_demote(struct gfs2_holder *gh)
{
struct gfs2_glock *gl = gh->gh_gl;
spin_lock(&gl->gl_lockref.lock);
clear_bit(HIF_MAY_DEMOTE, &gh->gh_iflags);
spin_unlock(&gl->gl_lockref.lock);
}
extern void gfs2_inode_remember_delete(struct gfs2_glock *gl, u64 generation);
extern bool gfs2_inode_already_deleted(struct gfs2_glock *gl, u64 generation);
#endif /* __GLOCK_DOT_H__ */
Reference in New Issue View Git Blame Copy Permalink