1
linux/fs/ocfs2/export.c

285 lines
6.2 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-or-later
/*
* export.c
*
* Functions to facilitate NFS exporting
*
* Copyright (C) 2002, 2005 Oracle. All rights reserved.
*/
#include <linux/fs.h>
#include <linux/types.h>
#include <cluster/masklog.h>
#include "ocfs2.h"
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
#include "alloc.h"
#include "dir.h"
#include "dlmglue.h"
#include "dcache.h"
#include "export.h"
#include "inode.h"
#include "buffer_head_io.h"
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
#include "suballoc.h"
#include "ocfs2_trace.h"
struct ocfs2_inode_handle
{
u64 ih_blkno;
u32 ih_generation;
};
static struct dentry *ocfs2_get_dentry(struct super_block *sb,
struct ocfs2_inode_handle *handle)
{
struct inode *inode;
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
struct ocfs2_super *osb = OCFS2_SB(sb);
u64 blkno = handle->ih_blkno;
int status, set;
struct dentry *result;
trace_ocfs2_get_dentry_begin(sb, handle, (unsigned long long)blkno);
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
if (blkno == 0) {
result = ERR_PTR(-ESTALE);
goto bail;
}
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
inode = ocfs2_ilookup(sb, blkno);
/*
* If the inode exists in memory, we only need to check it's
* generation number
*/
if (inode)
goto check_gen;
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
/*
* This will synchronize us against ocfs2_delete_inode() on
* all nodes
*/
status = ocfs2_nfs_sync_lock(osb, 1);
if (status < 0) {
mlog(ML_ERROR, "getting nfs sync lock(EX) failed %d\n", status);
goto check_err;
}
status = ocfs2_test_inode_bit(osb, blkno, &set);
if (status < 0) {
if (status == -EINVAL) {
/*
* The blkno NFS gave us doesn't even show up
* as an inode, we return -ESTALE to be
* nice
*/
status = -ESTALE;
} else
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
mlog(ML_ERROR, "test inode bit failed %d\n", status);
goto unlock_nfs_sync;
}
trace_ocfs2_get_dentry_test_bit(status, set);
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
/* If the inode allocator bit is clear, this inode must be stale */
if (!set) {
status = -ESTALE;
goto unlock_nfs_sync;
}
inode = ocfs2_iget(osb, blkno, 0, 0);
unlock_nfs_sync:
ocfs2_nfs_sync_unlock(osb, 1);
check_err:
if (status < 0) {
if (status == -ESTALE) {
trace_ocfs2_get_dentry_stale((unsigned long long)blkno,
handle->ih_generation);
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
}
result = ERR_PTR(status);
goto bail;
}
if (IS_ERR(inode)) {
mlog_errno(PTR_ERR(inode));
result = ERR_CAST(inode);
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
goto bail;
}
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
check_gen:
if (handle->ih_generation != inode->i_generation) {
trace_ocfs2_get_dentry_generation((unsigned long long)blkno,
handle->ih_generation,
inode->i_generation);
iput(inode);
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
result = ERR_PTR(-ESTALE);
goto bail;
}
result = d_obtain_alias(inode);
if (IS_ERR(result))
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
mlog_errno(PTR_ERR(result));
ocfs2: fix rare stale inode errors when exporting via nfs For nfs exporting, ocfs2_get_dentry() returns the dentry for fh. ocfs2_get_dentry() may read from disk when the inode is not in memory, without any cross cluster lock. this leads to the file system loading a stale inode. This patch fixes above problem. Solution is that in case of inode is not in memory, we get the cluster lock(PR) of alloc inode where the inode in question is allocated from (this causes node on which deletion is done sync the alloc inode) before reading out the inode itsself. then we check the bitmap in the group (the inode in question allcated from) to see if the bit is clear. if it's clear then it's stale. if the bit is set, we then check generation as the existing code does. We have to read out the inode in question from disk first to know its alloc slot and allot bit. And if its not stale we read it out using ocfs2_iget(). The second read should then be from cache. And also we have to add a per superblock nfs_sync_lock to cover the lock for alloc inode and that for inode in question. this is because ocfs2_get_dentry() and ocfs2_delete_inode() lock on them in reverse order. nfs_sync_lock is locked in EX mode in ocfs2_get_dentry() and in PR mode in ocfs2_delete_inode(). so that mutliple ocfs2_delete_inode() can run concurrently in normal case. [mfasheh@suse.com: build warning fixes and comment cleanups] Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com> Acked-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2009-03-06 06:29:10 -07:00
bail:
trace_ocfs2_get_dentry_end(result);
return result;
}
static struct dentry *ocfs2_get_parent(struct dentry *child)
{
int status;
u64 blkno;
struct dentry *parent;
struct inode *dir = d_inode(child);
ocfs2: fix ocfs2 read inode data panic in ocfs2_iget In some cases, ocfs2_iget() reads the data of inode, which has been deleted for some reason. That will make the system panic. So We should judge whether this inode has been deleted, and tell the caller that the inode is a bad inode. For example, the ocfs2 is used as the backed of nfs, and the client is nfsv3. This issue can be reproduced by the following steps. on the nfs server side, ..../patha/pathb Step 1: The process A was scheduled before calling the function fh_verify. Step 2: The process B is removing the 'pathb', and just completed the call to function dput. Then the dentry of 'pathb' has been deleted from the dcache, and all ancestors have been deleted also. The relationship of dentry and inode was deleted through the function hlist_del_init. The following is the call stack. dentry_iput->hlist_del_init(&dentry->d_u.d_alias) At this time, the inode is still in the dcache. Step 3: The process A call the function ocfs2_get_dentry, which get the inode from dcache. Then the refcount of inode is 1. The following is the call stack. nfsd3_proc_getacl->fh_verify->exportfs_decode_fh->fh_to_dentry(ocfs2_get_dentry) Step 4: Dirty pages are flushed by bdi threads. So the inode of 'patha' is evicted, and this directory was deleted. But the inode of 'pathb' can't be evicted, because the refcount of the inode was 1. Step 5: The process A keep running, and call the function reconnect_path(in exportfs_decode_fh), which call function ocfs2_get_parent of ocfs2. Get the block number of parent directory(patha) by the name of ... Then read the data from disk by the block number. But this inode has been deleted, so the system panic. Process A Process B 1. in nfsd3_proc_getacl | 2. | dput 3. fh_to_dentry(ocfs2_get_dentry) | 4. bdi flush dirty cache | 5. ocfs2_iget | [283465.542049] OCFS2: ERROR (device sdp): ocfs2_validate_inode_block: Invalid dinode #580640: OCFS2_VALID_FL not set [283465.545490] Kernel panic - not syncing: OCFS2: (device sdp): panic forced after error [283465.546889] CPU: 5 PID: 12416 Comm: nfsd Tainted: G W 4.1.12-124.18.6.el6uek.bug28762940v3.x86_64 #2 [283465.548382] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 09/21/2015 [283465.549657] 0000000000000000 ffff8800a56fb7b8 ffffffff816e839c ffffffffa0514758 [283465.550392] 000000000008dc20 ffff8800a56fb838 ffffffff816e62d3 0000000000000008 [283465.551056] ffff880000000010 ffff8800a56fb848 ffff8800a56fb7e8 ffff88005df9f000 [283465.551710] Call Trace: [283465.552516] [<ffffffff816e839c>] dump_stack+0x63/0x81 [283465.553291] [<ffffffff816e62d3>] panic+0xcb/0x21b [283465.554037] [<ffffffffa04e66b0>] ocfs2_handle_error+0xf0/0xf0 [ocfs2] [283465.554882] [<ffffffffa04e7737>] __ocfs2_error+0x67/0x70 [ocfs2] [283465.555768] [<ffffffffa049c0f9>] ocfs2_validate_inode_block+0x229/0x230 [ocfs2] [283465.556683] [<ffffffffa047bcbc>] ocfs2_read_blocks+0x46c/0x7b0 [ocfs2] [283465.557408] [<ffffffffa049bed0>] ? ocfs2_inode_cache_io_unlock+0x20/0x20 [ocfs2] [283465.557973] [<ffffffffa049f0eb>] ocfs2_read_inode_block_full+0x3b/0x60 [ocfs2] [283465.558525] [<ffffffffa049f5ba>] ocfs2_iget+0x4aa/0x880 [ocfs2] [283465.559082] [<ffffffffa049146e>] ocfs2_get_parent+0x9e/0x220 [ocfs2] [283465.559622] [<ffffffff81297c05>] reconnect_path+0xb5/0x300 [283465.560156] [<ffffffff81297f46>] exportfs_decode_fh+0xf6/0x2b0 [283465.560708] [<ffffffffa062faf0>] ? nfsd_proc_getattr+0xa0/0xa0 [nfsd] [283465.561262] [<ffffffff810a8196>] ? prepare_creds+0x26/0x110 [283465.561932] [<ffffffffa0630860>] fh_verify+0x350/0x660 [nfsd] [283465.562862] [<ffffffffa0637804>] ? nfsd_cache_lookup+0x44/0x630 [nfsd] [283465.563697] [<ffffffffa063a8b9>] nfsd3_proc_getattr+0x69/0xf0 [nfsd] [283465.564510] [<ffffffffa062cf60>] nfsd_dispatch+0xe0/0x290 [nfsd] [283465.565358] [<ffffffffa05eb892>] ? svc_tcp_adjust_wspace+0x12/0x30 [sunrpc] [283465.566272] [<ffffffffa05ea652>] svc_process_common+0x412/0x6a0 [sunrpc] [283465.567155] [<ffffffffa05eaa03>] svc_process+0x123/0x210 [sunrpc] [283465.568020] [<ffffffffa062c90f>] nfsd+0xff/0x170 [nfsd] [283465.568962] [<ffffffffa062c810>] ? nfsd_destroy+0x80/0x80 [nfsd] [283465.570112] [<ffffffff810a622b>] kthread+0xcb/0xf0 [283465.571099] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 [283465.572114] [<ffffffff816f11b8>] ret_from_fork+0x58/0x90 [283465.573156] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 Link: http://lkml.kernel.org/r/1554185919-3010-1-git-send-email-sunny.s.zhang@oracle.com Signed-off-by: Shuning Zhang <sunny.s.zhang@oracle.com> Reviewed-by: Joseph Qi <jiangqi903@gmail.com> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: piaojun <piaojun@huawei.com> Cc: "Gang He" <ghe@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-05-13 17:15:56 -07:00
int set;
trace_ocfs2_get_parent(child, child->d_name.len, child->d_name.name,
(unsigned long long)OCFS2_I(dir)->ip_blkno);
ocfs2: fix ocfs2 read inode data panic in ocfs2_iget In some cases, ocfs2_iget() reads the data of inode, which has been deleted for some reason. That will make the system panic. So We should judge whether this inode has been deleted, and tell the caller that the inode is a bad inode. For example, the ocfs2 is used as the backed of nfs, and the client is nfsv3. This issue can be reproduced by the following steps. on the nfs server side, ..../patha/pathb Step 1: The process A was scheduled before calling the function fh_verify. Step 2: The process B is removing the 'pathb', and just completed the call to function dput. Then the dentry of 'pathb' has been deleted from the dcache, and all ancestors have been deleted also. The relationship of dentry and inode was deleted through the function hlist_del_init. The following is the call stack. dentry_iput->hlist_del_init(&dentry->d_u.d_alias) At this time, the inode is still in the dcache. Step 3: The process A call the function ocfs2_get_dentry, which get the inode from dcache. Then the refcount of inode is 1. The following is the call stack. nfsd3_proc_getacl->fh_verify->exportfs_decode_fh->fh_to_dentry(ocfs2_get_dentry) Step 4: Dirty pages are flushed by bdi threads. So the inode of 'patha' is evicted, and this directory was deleted. But the inode of 'pathb' can't be evicted, because the refcount of the inode was 1. Step 5: The process A keep running, and call the function reconnect_path(in exportfs_decode_fh), which call function ocfs2_get_parent of ocfs2. Get the block number of parent directory(patha) by the name of ... Then read the data from disk by the block number. But this inode has been deleted, so the system panic. Process A Process B 1. in nfsd3_proc_getacl | 2. | dput 3. fh_to_dentry(ocfs2_get_dentry) | 4. bdi flush dirty cache | 5. ocfs2_iget | [283465.542049] OCFS2: ERROR (device sdp): ocfs2_validate_inode_block: Invalid dinode #580640: OCFS2_VALID_FL not set [283465.545490] Kernel panic - not syncing: OCFS2: (device sdp): panic forced after error [283465.546889] CPU: 5 PID: 12416 Comm: nfsd Tainted: G W 4.1.12-124.18.6.el6uek.bug28762940v3.x86_64 #2 [283465.548382] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 09/21/2015 [283465.549657] 0000000000000000 ffff8800a56fb7b8 ffffffff816e839c ffffffffa0514758 [283465.550392] 000000000008dc20 ffff8800a56fb838 ffffffff816e62d3 0000000000000008 [283465.551056] ffff880000000010 ffff8800a56fb848 ffff8800a56fb7e8 ffff88005df9f000 [283465.551710] Call Trace: [283465.552516] [<ffffffff816e839c>] dump_stack+0x63/0x81 [283465.553291] [<ffffffff816e62d3>] panic+0xcb/0x21b [283465.554037] [<ffffffffa04e66b0>] ocfs2_handle_error+0xf0/0xf0 [ocfs2] [283465.554882] [<ffffffffa04e7737>] __ocfs2_error+0x67/0x70 [ocfs2] [283465.555768] [<ffffffffa049c0f9>] ocfs2_validate_inode_block+0x229/0x230 [ocfs2] [283465.556683] [<ffffffffa047bcbc>] ocfs2_read_blocks+0x46c/0x7b0 [ocfs2] [283465.557408] [<ffffffffa049bed0>] ? ocfs2_inode_cache_io_unlock+0x20/0x20 [ocfs2] [283465.557973] [<ffffffffa049f0eb>] ocfs2_read_inode_block_full+0x3b/0x60 [ocfs2] [283465.558525] [<ffffffffa049f5ba>] ocfs2_iget+0x4aa/0x880 [ocfs2] [283465.559082] [<ffffffffa049146e>] ocfs2_get_parent+0x9e/0x220 [ocfs2] [283465.559622] [<ffffffff81297c05>] reconnect_path+0xb5/0x300 [283465.560156] [<ffffffff81297f46>] exportfs_decode_fh+0xf6/0x2b0 [283465.560708] [<ffffffffa062faf0>] ? nfsd_proc_getattr+0xa0/0xa0 [nfsd] [283465.561262] [<ffffffff810a8196>] ? prepare_creds+0x26/0x110 [283465.561932] [<ffffffffa0630860>] fh_verify+0x350/0x660 [nfsd] [283465.562862] [<ffffffffa0637804>] ? nfsd_cache_lookup+0x44/0x630 [nfsd] [283465.563697] [<ffffffffa063a8b9>] nfsd3_proc_getattr+0x69/0xf0 [nfsd] [283465.564510] [<ffffffffa062cf60>] nfsd_dispatch+0xe0/0x290 [nfsd] [283465.565358] [<ffffffffa05eb892>] ? svc_tcp_adjust_wspace+0x12/0x30 [sunrpc] [283465.566272] [<ffffffffa05ea652>] svc_process_common+0x412/0x6a0 [sunrpc] [283465.567155] [<ffffffffa05eaa03>] svc_process+0x123/0x210 [sunrpc] [283465.568020] [<ffffffffa062c90f>] nfsd+0xff/0x170 [nfsd] [283465.568962] [<ffffffffa062c810>] ? nfsd_destroy+0x80/0x80 [nfsd] [283465.570112] [<ffffffff810a622b>] kthread+0xcb/0xf0 [283465.571099] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 [283465.572114] [<ffffffff816f11b8>] ret_from_fork+0x58/0x90 [283465.573156] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 Link: http://lkml.kernel.org/r/1554185919-3010-1-git-send-email-sunny.s.zhang@oracle.com Signed-off-by: Shuning Zhang <sunny.s.zhang@oracle.com> Reviewed-by: Joseph Qi <jiangqi903@gmail.com> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: piaojun <piaojun@huawei.com> Cc: "Gang He" <ghe@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-05-13 17:15:56 -07:00
status = ocfs2_nfs_sync_lock(OCFS2_SB(dir->i_sb), 1);
if (status < 0) {
mlog(ML_ERROR, "getting nfs sync lock(EX) failed %d\n", status);
parent = ERR_PTR(status);
goto bail;
}
status = ocfs2_inode_lock(dir, NULL, 0);
if (status < 0) {
if (status != -ENOENT)
mlog_errno(status);
parent = ERR_PTR(status);
ocfs2: fix ocfs2 read inode data panic in ocfs2_iget In some cases, ocfs2_iget() reads the data of inode, which has been deleted for some reason. That will make the system panic. So We should judge whether this inode has been deleted, and tell the caller that the inode is a bad inode. For example, the ocfs2 is used as the backed of nfs, and the client is nfsv3. This issue can be reproduced by the following steps. on the nfs server side, ..../patha/pathb Step 1: The process A was scheduled before calling the function fh_verify. Step 2: The process B is removing the 'pathb', and just completed the call to function dput. Then the dentry of 'pathb' has been deleted from the dcache, and all ancestors have been deleted also. The relationship of dentry and inode was deleted through the function hlist_del_init. The following is the call stack. dentry_iput->hlist_del_init(&dentry->d_u.d_alias) At this time, the inode is still in the dcache. Step 3: The process A call the function ocfs2_get_dentry, which get the inode from dcache. Then the refcount of inode is 1. The following is the call stack. nfsd3_proc_getacl->fh_verify->exportfs_decode_fh->fh_to_dentry(ocfs2_get_dentry) Step 4: Dirty pages are flushed by bdi threads. So the inode of 'patha' is evicted, and this directory was deleted. But the inode of 'pathb' can't be evicted, because the refcount of the inode was 1. Step 5: The process A keep running, and call the function reconnect_path(in exportfs_decode_fh), which call function ocfs2_get_parent of ocfs2. Get the block number of parent directory(patha) by the name of ... Then read the data from disk by the block number. But this inode has been deleted, so the system panic. Process A Process B 1. in nfsd3_proc_getacl | 2. | dput 3. fh_to_dentry(ocfs2_get_dentry) | 4. bdi flush dirty cache | 5. ocfs2_iget | [283465.542049] OCFS2: ERROR (device sdp): ocfs2_validate_inode_block: Invalid dinode #580640: OCFS2_VALID_FL not set [283465.545490] Kernel panic - not syncing: OCFS2: (device sdp): panic forced after error [283465.546889] CPU: 5 PID: 12416 Comm: nfsd Tainted: G W 4.1.12-124.18.6.el6uek.bug28762940v3.x86_64 #2 [283465.548382] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 09/21/2015 [283465.549657] 0000000000000000 ffff8800a56fb7b8 ffffffff816e839c ffffffffa0514758 [283465.550392] 000000000008dc20 ffff8800a56fb838 ffffffff816e62d3 0000000000000008 [283465.551056] ffff880000000010 ffff8800a56fb848 ffff8800a56fb7e8 ffff88005df9f000 [283465.551710] Call Trace: [283465.552516] [<ffffffff816e839c>] dump_stack+0x63/0x81 [283465.553291] [<ffffffff816e62d3>] panic+0xcb/0x21b [283465.554037] [<ffffffffa04e66b0>] ocfs2_handle_error+0xf0/0xf0 [ocfs2] [283465.554882] [<ffffffffa04e7737>] __ocfs2_error+0x67/0x70 [ocfs2] [283465.555768] [<ffffffffa049c0f9>] ocfs2_validate_inode_block+0x229/0x230 [ocfs2] [283465.556683] [<ffffffffa047bcbc>] ocfs2_read_blocks+0x46c/0x7b0 [ocfs2] [283465.557408] [<ffffffffa049bed0>] ? ocfs2_inode_cache_io_unlock+0x20/0x20 [ocfs2] [283465.557973] [<ffffffffa049f0eb>] ocfs2_read_inode_block_full+0x3b/0x60 [ocfs2] [283465.558525] [<ffffffffa049f5ba>] ocfs2_iget+0x4aa/0x880 [ocfs2] [283465.559082] [<ffffffffa049146e>] ocfs2_get_parent+0x9e/0x220 [ocfs2] [283465.559622] [<ffffffff81297c05>] reconnect_path+0xb5/0x300 [283465.560156] [<ffffffff81297f46>] exportfs_decode_fh+0xf6/0x2b0 [283465.560708] [<ffffffffa062faf0>] ? nfsd_proc_getattr+0xa0/0xa0 [nfsd] [283465.561262] [<ffffffff810a8196>] ? prepare_creds+0x26/0x110 [283465.561932] [<ffffffffa0630860>] fh_verify+0x350/0x660 [nfsd] [283465.562862] [<ffffffffa0637804>] ? nfsd_cache_lookup+0x44/0x630 [nfsd] [283465.563697] [<ffffffffa063a8b9>] nfsd3_proc_getattr+0x69/0xf0 [nfsd] [283465.564510] [<ffffffffa062cf60>] nfsd_dispatch+0xe0/0x290 [nfsd] [283465.565358] [<ffffffffa05eb892>] ? svc_tcp_adjust_wspace+0x12/0x30 [sunrpc] [283465.566272] [<ffffffffa05ea652>] svc_process_common+0x412/0x6a0 [sunrpc] [283465.567155] [<ffffffffa05eaa03>] svc_process+0x123/0x210 [sunrpc] [283465.568020] [<ffffffffa062c90f>] nfsd+0xff/0x170 [nfsd] [283465.568962] [<ffffffffa062c810>] ? nfsd_destroy+0x80/0x80 [nfsd] [283465.570112] [<ffffffff810a622b>] kthread+0xcb/0xf0 [283465.571099] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 [283465.572114] [<ffffffff816f11b8>] ret_from_fork+0x58/0x90 [283465.573156] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 Link: http://lkml.kernel.org/r/1554185919-3010-1-git-send-email-sunny.s.zhang@oracle.com Signed-off-by: Shuning Zhang <sunny.s.zhang@oracle.com> Reviewed-by: Joseph Qi <jiangqi903@gmail.com> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: piaojun <piaojun@huawei.com> Cc: "Gang He" <ghe@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-05-13 17:15:56 -07:00
goto unlock_nfs_sync;
}
status = ocfs2_lookup_ino_from_name(dir, "..", 2, &blkno);
if (status < 0) {
parent = ERR_PTR(-ENOENT);
goto bail_unlock;
}
ocfs2: fix ocfs2 read inode data panic in ocfs2_iget In some cases, ocfs2_iget() reads the data of inode, which has been deleted for some reason. That will make the system panic. So We should judge whether this inode has been deleted, and tell the caller that the inode is a bad inode. For example, the ocfs2 is used as the backed of nfs, and the client is nfsv3. This issue can be reproduced by the following steps. on the nfs server side, ..../patha/pathb Step 1: The process A was scheduled before calling the function fh_verify. Step 2: The process B is removing the 'pathb', and just completed the call to function dput. Then the dentry of 'pathb' has been deleted from the dcache, and all ancestors have been deleted also. The relationship of dentry and inode was deleted through the function hlist_del_init. The following is the call stack. dentry_iput->hlist_del_init(&dentry->d_u.d_alias) At this time, the inode is still in the dcache. Step 3: The process A call the function ocfs2_get_dentry, which get the inode from dcache. Then the refcount of inode is 1. The following is the call stack. nfsd3_proc_getacl->fh_verify->exportfs_decode_fh->fh_to_dentry(ocfs2_get_dentry) Step 4: Dirty pages are flushed by bdi threads. So the inode of 'patha' is evicted, and this directory was deleted. But the inode of 'pathb' can't be evicted, because the refcount of the inode was 1. Step 5: The process A keep running, and call the function reconnect_path(in exportfs_decode_fh), which call function ocfs2_get_parent of ocfs2. Get the block number of parent directory(patha) by the name of ... Then read the data from disk by the block number. But this inode has been deleted, so the system panic. Process A Process B 1. in nfsd3_proc_getacl | 2. | dput 3. fh_to_dentry(ocfs2_get_dentry) | 4. bdi flush dirty cache | 5. ocfs2_iget | [283465.542049] OCFS2: ERROR (device sdp): ocfs2_validate_inode_block: Invalid dinode #580640: OCFS2_VALID_FL not set [283465.545490] Kernel panic - not syncing: OCFS2: (device sdp): panic forced after error [283465.546889] CPU: 5 PID: 12416 Comm: nfsd Tainted: G W 4.1.12-124.18.6.el6uek.bug28762940v3.x86_64 #2 [283465.548382] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 09/21/2015 [283465.549657] 0000000000000000 ffff8800a56fb7b8 ffffffff816e839c ffffffffa0514758 [283465.550392] 000000000008dc20 ffff8800a56fb838 ffffffff816e62d3 0000000000000008 [283465.551056] ffff880000000010 ffff8800a56fb848 ffff8800a56fb7e8 ffff88005df9f000 [283465.551710] Call Trace: [283465.552516] [<ffffffff816e839c>] dump_stack+0x63/0x81 [283465.553291] [<ffffffff816e62d3>] panic+0xcb/0x21b [283465.554037] [<ffffffffa04e66b0>] ocfs2_handle_error+0xf0/0xf0 [ocfs2] [283465.554882] [<ffffffffa04e7737>] __ocfs2_error+0x67/0x70 [ocfs2] [283465.555768] [<ffffffffa049c0f9>] ocfs2_validate_inode_block+0x229/0x230 [ocfs2] [283465.556683] [<ffffffffa047bcbc>] ocfs2_read_blocks+0x46c/0x7b0 [ocfs2] [283465.557408] [<ffffffffa049bed0>] ? ocfs2_inode_cache_io_unlock+0x20/0x20 [ocfs2] [283465.557973] [<ffffffffa049f0eb>] ocfs2_read_inode_block_full+0x3b/0x60 [ocfs2] [283465.558525] [<ffffffffa049f5ba>] ocfs2_iget+0x4aa/0x880 [ocfs2] [283465.559082] [<ffffffffa049146e>] ocfs2_get_parent+0x9e/0x220 [ocfs2] [283465.559622] [<ffffffff81297c05>] reconnect_path+0xb5/0x300 [283465.560156] [<ffffffff81297f46>] exportfs_decode_fh+0xf6/0x2b0 [283465.560708] [<ffffffffa062faf0>] ? nfsd_proc_getattr+0xa0/0xa0 [nfsd] [283465.561262] [<ffffffff810a8196>] ? prepare_creds+0x26/0x110 [283465.561932] [<ffffffffa0630860>] fh_verify+0x350/0x660 [nfsd] [283465.562862] [<ffffffffa0637804>] ? nfsd_cache_lookup+0x44/0x630 [nfsd] [283465.563697] [<ffffffffa063a8b9>] nfsd3_proc_getattr+0x69/0xf0 [nfsd] [283465.564510] [<ffffffffa062cf60>] nfsd_dispatch+0xe0/0x290 [nfsd] [283465.565358] [<ffffffffa05eb892>] ? svc_tcp_adjust_wspace+0x12/0x30 [sunrpc] [283465.566272] [<ffffffffa05ea652>] svc_process_common+0x412/0x6a0 [sunrpc] [283465.567155] [<ffffffffa05eaa03>] svc_process+0x123/0x210 [sunrpc] [283465.568020] [<ffffffffa062c90f>] nfsd+0xff/0x170 [nfsd] [283465.568962] [<ffffffffa062c810>] ? nfsd_destroy+0x80/0x80 [nfsd] [283465.570112] [<ffffffff810a622b>] kthread+0xcb/0xf0 [283465.571099] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 [283465.572114] [<ffffffff816f11b8>] ret_from_fork+0x58/0x90 [283465.573156] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 Link: http://lkml.kernel.org/r/1554185919-3010-1-git-send-email-sunny.s.zhang@oracle.com Signed-off-by: Shuning Zhang <sunny.s.zhang@oracle.com> Reviewed-by: Joseph Qi <jiangqi903@gmail.com> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: piaojun <piaojun@huawei.com> Cc: "Gang He" <ghe@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-05-13 17:15:56 -07:00
status = ocfs2_test_inode_bit(OCFS2_SB(dir->i_sb), blkno, &set);
if (status < 0) {
if (status == -EINVAL) {
status = -ESTALE;
} else
mlog(ML_ERROR, "test inode bit failed %d\n", status);
parent = ERR_PTR(status);
goto bail_unlock;
}
trace_ocfs2_get_dentry_test_bit(status, set);
if (!set) {
status = -ESTALE;
parent = ERR_PTR(status);
goto bail_unlock;
}
parent = d_obtain_alias(ocfs2_iget(OCFS2_SB(dir->i_sb), blkno, 0, 0));
bail_unlock:
ocfs2_inode_unlock(dir, 0);
ocfs2: fix ocfs2 read inode data panic in ocfs2_iget In some cases, ocfs2_iget() reads the data of inode, which has been deleted for some reason. That will make the system panic. So We should judge whether this inode has been deleted, and tell the caller that the inode is a bad inode. For example, the ocfs2 is used as the backed of nfs, and the client is nfsv3. This issue can be reproduced by the following steps. on the nfs server side, ..../patha/pathb Step 1: The process A was scheduled before calling the function fh_verify. Step 2: The process B is removing the 'pathb', and just completed the call to function dput. Then the dentry of 'pathb' has been deleted from the dcache, and all ancestors have been deleted also. The relationship of dentry and inode was deleted through the function hlist_del_init. The following is the call stack. dentry_iput->hlist_del_init(&dentry->d_u.d_alias) At this time, the inode is still in the dcache. Step 3: The process A call the function ocfs2_get_dentry, which get the inode from dcache. Then the refcount of inode is 1. The following is the call stack. nfsd3_proc_getacl->fh_verify->exportfs_decode_fh->fh_to_dentry(ocfs2_get_dentry) Step 4: Dirty pages are flushed by bdi threads. So the inode of 'patha' is evicted, and this directory was deleted. But the inode of 'pathb' can't be evicted, because the refcount of the inode was 1. Step 5: The process A keep running, and call the function reconnect_path(in exportfs_decode_fh), which call function ocfs2_get_parent of ocfs2. Get the block number of parent directory(patha) by the name of ... Then read the data from disk by the block number. But this inode has been deleted, so the system panic. Process A Process B 1. in nfsd3_proc_getacl | 2. | dput 3. fh_to_dentry(ocfs2_get_dentry) | 4. bdi flush dirty cache | 5. ocfs2_iget | [283465.542049] OCFS2: ERROR (device sdp): ocfs2_validate_inode_block: Invalid dinode #580640: OCFS2_VALID_FL not set [283465.545490] Kernel panic - not syncing: OCFS2: (device sdp): panic forced after error [283465.546889] CPU: 5 PID: 12416 Comm: nfsd Tainted: G W 4.1.12-124.18.6.el6uek.bug28762940v3.x86_64 #2 [283465.548382] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 09/21/2015 [283465.549657] 0000000000000000 ffff8800a56fb7b8 ffffffff816e839c ffffffffa0514758 [283465.550392] 000000000008dc20 ffff8800a56fb838 ffffffff816e62d3 0000000000000008 [283465.551056] ffff880000000010 ffff8800a56fb848 ffff8800a56fb7e8 ffff88005df9f000 [283465.551710] Call Trace: [283465.552516] [<ffffffff816e839c>] dump_stack+0x63/0x81 [283465.553291] [<ffffffff816e62d3>] panic+0xcb/0x21b [283465.554037] [<ffffffffa04e66b0>] ocfs2_handle_error+0xf0/0xf0 [ocfs2] [283465.554882] [<ffffffffa04e7737>] __ocfs2_error+0x67/0x70 [ocfs2] [283465.555768] [<ffffffffa049c0f9>] ocfs2_validate_inode_block+0x229/0x230 [ocfs2] [283465.556683] [<ffffffffa047bcbc>] ocfs2_read_blocks+0x46c/0x7b0 [ocfs2] [283465.557408] [<ffffffffa049bed0>] ? ocfs2_inode_cache_io_unlock+0x20/0x20 [ocfs2] [283465.557973] [<ffffffffa049f0eb>] ocfs2_read_inode_block_full+0x3b/0x60 [ocfs2] [283465.558525] [<ffffffffa049f5ba>] ocfs2_iget+0x4aa/0x880 [ocfs2] [283465.559082] [<ffffffffa049146e>] ocfs2_get_parent+0x9e/0x220 [ocfs2] [283465.559622] [<ffffffff81297c05>] reconnect_path+0xb5/0x300 [283465.560156] [<ffffffff81297f46>] exportfs_decode_fh+0xf6/0x2b0 [283465.560708] [<ffffffffa062faf0>] ? nfsd_proc_getattr+0xa0/0xa0 [nfsd] [283465.561262] [<ffffffff810a8196>] ? prepare_creds+0x26/0x110 [283465.561932] [<ffffffffa0630860>] fh_verify+0x350/0x660 [nfsd] [283465.562862] [<ffffffffa0637804>] ? nfsd_cache_lookup+0x44/0x630 [nfsd] [283465.563697] [<ffffffffa063a8b9>] nfsd3_proc_getattr+0x69/0xf0 [nfsd] [283465.564510] [<ffffffffa062cf60>] nfsd_dispatch+0xe0/0x290 [nfsd] [283465.565358] [<ffffffffa05eb892>] ? svc_tcp_adjust_wspace+0x12/0x30 [sunrpc] [283465.566272] [<ffffffffa05ea652>] svc_process_common+0x412/0x6a0 [sunrpc] [283465.567155] [<ffffffffa05eaa03>] svc_process+0x123/0x210 [sunrpc] [283465.568020] [<ffffffffa062c90f>] nfsd+0xff/0x170 [nfsd] [283465.568962] [<ffffffffa062c810>] ? nfsd_destroy+0x80/0x80 [nfsd] [283465.570112] [<ffffffff810a622b>] kthread+0xcb/0xf0 [283465.571099] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 [283465.572114] [<ffffffff816f11b8>] ret_from_fork+0x58/0x90 [283465.573156] [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180 Link: http://lkml.kernel.org/r/1554185919-3010-1-git-send-email-sunny.s.zhang@oracle.com Signed-off-by: Shuning Zhang <sunny.s.zhang@oracle.com> Reviewed-by: Joseph Qi <jiangqi903@gmail.com> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: piaojun <piaojun@huawei.com> Cc: "Gang He" <ghe@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-05-13 17:15:56 -07:00
unlock_nfs_sync:
ocfs2_nfs_sync_unlock(OCFS2_SB(dir->i_sb), 1);
bail:
trace_ocfs2_get_parent_end(parent);
return parent;
}
static int ocfs2_encode_fh(struct inode *inode, u32 *fh_in, int *max_len,
struct inode *parent)
{
int len = *max_len;
int type = 1;
u64 blkno;
u32 generation;
__le32 *fh = (__force __le32 *) fh_in;
#ifdef TRACE_HOOKS_ARE_NOT_BRAINDEAD_IN_YOUR_OPINION
#error "You go ahead and fix that mess, then. Somehow"
trace_ocfs2_encode_fh_begin(dentry, dentry->d_name.len,
dentry->d_name.name,
fh, len, connectable);
#endif
if (parent && (len < 6)) {
*max_len = 6;
type = FILEID_INVALID;
goto bail;
} else if (len < 3) {
*max_len = 3;
type = FILEID_INVALID;
goto bail;
}
blkno = OCFS2_I(inode)->ip_blkno;
generation = inode->i_generation;
trace_ocfs2_encode_fh_self((unsigned long long)blkno, generation);
len = 3;
fh[0] = cpu_to_le32((u32)(blkno >> 32));
fh[1] = cpu_to_le32((u32)(blkno & 0xffffffff));
fh[2] = cpu_to_le32(generation);
if (parent) {
blkno = OCFS2_I(parent)->ip_blkno;
generation = parent->i_generation;
fh[3] = cpu_to_le32((u32)(blkno >> 32));
fh[4] = cpu_to_le32((u32)(blkno & 0xffffffff));
fh[5] = cpu_to_le32(generation);
len = 6;
type = 2;
trace_ocfs2_encode_fh_parent((unsigned long long)blkno,
generation);
}
*max_len = len;
bail:
trace_ocfs2_encode_fh_type(type);
return type;
}
static struct dentry *ocfs2_fh_to_dentry(struct super_block *sb,
struct fid *fid, int fh_len, int fh_type)
{
struct ocfs2_inode_handle handle;
if (fh_len < 3 || fh_type > 2)
return NULL;
ocfs2: fix sparse warnings 1. fs/ocfs2/localalloc.c:1224:41: warning: incorrect type in argument 1 (different base types) fs/ocfs2/localalloc.c:1224:41: expected unsigned long long val1 fs/ocfs2/localalloc.c:1224:41: got restricted __le32 [usertype] la_bm_off 2. fs/ocfs2/export.c:258:32: warning: cast to restricted __le32 fs/ocfs2/export.c:259:33: warning: cast to restricted __le32 fs/ocfs2/export.c:260:32: warning: cast to restricted __le32 fs/ocfs2/export.c:272:32: warning: cast to restricted __le32 fs/ocfs2/export.c:273:33: warning: cast to restricted __le32 fs/ocfs2/export.c:274:32: warning: cast to restricted __le32 3. fs/ocfs2/inode.c:1623:13: warning: context imbalance in 'ocfs2_inode_cache_lock' - wrong count at exit fs/ocfs2/inode.c:1630:13: warning: context imbalance in 'ocfs2_inode_cache_unlock' - unexpected unlock 4. fs/ocfs2/refcounttree.c:633:27: warning: incorrect type in assignment (different base types) fs/ocfs2/refcounttree.c:633:27: expected restricted __le32 [usertype] rf_generation fs/ocfs2/refcounttree.c:633:27: got unsigned int 5. fs/ocfs2/dlm/dlmdomain.c:1316:20: warning: context imbalance in 'dlm_query_nodeinfo_handler' - different lock contexts for basic block Link: https://lkml.kernel.org/r/20240328125203.20892-5-heming.zhao@suse.com Signed-off-by: Heming Zhao <heming.zhao@suse.com> Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: Gang He <ghe@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Jun Piao <piaojun@huawei.com> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Mark Fasheh <mark@fasheh.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-03-28 05:52:03 -07:00
handle.ih_blkno = (u64)le32_to_cpu((__force __le32)fid->raw[0]) << 32;
handle.ih_blkno |= (u64)le32_to_cpu((__force __le32)fid->raw[1]);
handle.ih_generation = le32_to_cpu((__force __le32)fid->raw[2]);
return ocfs2_get_dentry(sb, &handle);
}
static struct dentry *ocfs2_fh_to_parent(struct super_block *sb,
struct fid *fid, int fh_len, int fh_type)
{
struct ocfs2_inode_handle parent;
if (fh_type != 2 || fh_len < 6)
return NULL;
ocfs2: fix sparse warnings 1. fs/ocfs2/localalloc.c:1224:41: warning: incorrect type in argument 1 (different base types) fs/ocfs2/localalloc.c:1224:41: expected unsigned long long val1 fs/ocfs2/localalloc.c:1224:41: got restricted __le32 [usertype] la_bm_off 2. fs/ocfs2/export.c:258:32: warning: cast to restricted __le32 fs/ocfs2/export.c:259:33: warning: cast to restricted __le32 fs/ocfs2/export.c:260:32: warning: cast to restricted __le32 fs/ocfs2/export.c:272:32: warning: cast to restricted __le32 fs/ocfs2/export.c:273:33: warning: cast to restricted __le32 fs/ocfs2/export.c:274:32: warning: cast to restricted __le32 3. fs/ocfs2/inode.c:1623:13: warning: context imbalance in 'ocfs2_inode_cache_lock' - wrong count at exit fs/ocfs2/inode.c:1630:13: warning: context imbalance in 'ocfs2_inode_cache_unlock' - unexpected unlock 4. fs/ocfs2/refcounttree.c:633:27: warning: incorrect type in assignment (different base types) fs/ocfs2/refcounttree.c:633:27: expected restricted __le32 [usertype] rf_generation fs/ocfs2/refcounttree.c:633:27: got unsigned int 5. fs/ocfs2/dlm/dlmdomain.c:1316:20: warning: context imbalance in 'dlm_query_nodeinfo_handler' - different lock contexts for basic block Link: https://lkml.kernel.org/r/20240328125203.20892-5-heming.zhao@suse.com Signed-off-by: Heming Zhao <heming.zhao@suse.com> Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: Gang He <ghe@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Jun Piao <piaojun@huawei.com> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Mark Fasheh <mark@fasheh.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-03-28 05:52:03 -07:00
parent.ih_blkno = (u64)le32_to_cpu((__force __le32)fid->raw[3]) << 32;
parent.ih_blkno |= (u64)le32_to_cpu((__force __le32)fid->raw[4]);
parent.ih_generation = le32_to_cpu((__force __le32)fid->raw[5]);
return ocfs2_get_dentry(sb, &parent);
}
const struct export_operations ocfs2_export_ops = {
.encode_fh = ocfs2_encode_fh,
.fh_to_dentry = ocfs2_fh_to_dentry,
.fh_to_parent = ocfs2_fh_to_parent,
.get_parent = ocfs2_get_parent,
.flags = EXPORT_OP_ASYNC_LOCK,
};