linux-stable/fs/jbd2/journal.c

3188 lines
89 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0+
/*
* linux/fs/jbd2/journal.c
*
* Written by Stephen C. Tweedie <sct@redhat.com>, 1998
*
* Copyright 1998 Red Hat corp --- All Rights Reserved
*
* Generic filesystem journal-writing code; part of the ext2fs
* journaling system.
*
* This file manages journals: areas of disk reserved for logging
* transactional updates. This includes the kernel journaling thread
* which is responsible for scheduling updates to the log.
*
* We do not actually manage the physical storage of the journal in this
* file: that is left to a per-journal policy function, which allows us
* to store the journal within a filesystem-specified area for ext2
* journaling (ext2 can use a reserved inode for storing the log).
*/
#include <linux/module.h>
#include <linux/time.h>
#include <linux/fs.h>
#include <linux/jbd2.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/freezer.h>
#include <linux/pagemap.h>
#include <linux/kthread.h>
#include <linux/poison.h>
#include <linux/proc_fs.h>
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
#include <linux/seq_file.h>
#include <linux/math64.h>
#include <linux/hash.h>
#include <linux/log2.h>
#include <linux/vmalloc.h>
#include <linux/backing-dev.h>
#include <linux/bitops.h>
#include <linux/ratelimit.h>
#include <linux/sched/mm.h>
#define CREATE_TRACE_POINTS
#include <trace/events/jbd2.h>
#include <linux/uaccess.h>
#include <asm/page.h>
#ifdef CONFIG_JBD2_DEBUG
static ushort jbd2_journal_enable_debug __read_mostly;
module_param_named(jbd2_debug, jbd2_journal_enable_debug, ushort, 0644);
MODULE_PARM_DESC(jbd2_debug, "Debugging level for jbd2");
#endif
EXPORT_SYMBOL(jbd2_journal_extend);
EXPORT_SYMBOL(jbd2_journal_stop);
EXPORT_SYMBOL(jbd2_journal_lock_updates);
EXPORT_SYMBOL(jbd2_journal_unlock_updates);
EXPORT_SYMBOL(jbd2_journal_get_write_access);
EXPORT_SYMBOL(jbd2_journal_get_create_access);
EXPORT_SYMBOL(jbd2_journal_get_undo_access);
jbd2: Add buffer triggers Filesystems often to do compute intensive operation on some metadata. If this operation is repeated many times, it can be very expensive. It would be much nicer if the operation could be performed once before a buffer goes to disk. This adds triggers to jbd2 buffer heads. Just before writing a metadata buffer to the journal, jbd2 will optionally call a commit trigger associated with the buffer. If the journal is aborted, an abort trigger will be called on any dirty buffers as they are dropped from pending transactions. ocfs2 will use this feature. Initially I tried to come up with a more generic trigger that could be used for non-buffer-related events like transaction completion. It doesn't tie nicely, because the information a buffer trigger needs (specific to a journal_head) isn't the same as what a transaction trigger needs (specific to a tranaction_t or perhaps journal_t). So I implemented a buffer set, with the understanding that journal/transaction wide triggers should be implemented separately. There is only one trigger set allowed per buffer. I can't think of any reason to attach more than one set. Contrast this with a journal or transaction in which multiple places may want to watch the entire transaction separately. The trigger sets are considered static allocation from the jbd2 perspective. ocfs2 will just have one trigger set per block type, setting the same set on every bh of the same type. Signed-off-by: Joel Becker <joel.becker@oracle.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: <linux-ext4@vger.kernel.org> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2008-09-11 22:35:47 +00:00
EXPORT_SYMBOL(jbd2_journal_set_triggers);
EXPORT_SYMBOL(jbd2_journal_dirty_metadata);
EXPORT_SYMBOL(jbd2_journal_forget);
EXPORT_SYMBOL(jbd2_journal_flush);
EXPORT_SYMBOL(jbd2_journal_revoke);
EXPORT_SYMBOL(jbd2_journal_init_dev);
EXPORT_SYMBOL(jbd2_journal_init_inode);
EXPORT_SYMBOL(jbd2_journal_check_used_features);
EXPORT_SYMBOL(jbd2_journal_check_available_features);
EXPORT_SYMBOL(jbd2_journal_set_features);
EXPORT_SYMBOL(jbd2_journal_load);
EXPORT_SYMBOL(jbd2_journal_destroy);
EXPORT_SYMBOL(jbd2_journal_abort);
EXPORT_SYMBOL(jbd2_journal_errno);
EXPORT_SYMBOL(jbd2_journal_ack_err);
EXPORT_SYMBOL(jbd2_journal_clear_err);
EXPORT_SYMBOL(jbd2_log_wait_commit);
EXPORT_SYMBOL(jbd2_journal_start_commit);
EXPORT_SYMBOL(jbd2_journal_force_commit_nested);
EXPORT_SYMBOL(jbd2_journal_wipe);
EXPORT_SYMBOL(jbd2_journal_blocks_per_page);
EXPORT_SYMBOL(jbd2_journal_invalidate_folio);
EXPORT_SYMBOL(jbd2_journal_try_to_free_buffers);
EXPORT_SYMBOL(jbd2_journal_force_commit);
EXPORT_SYMBOL(jbd2_journal_inode_ranged_write);
EXPORT_SYMBOL(jbd2_journal_inode_ranged_wait);
EXPORT_SYMBOL(jbd2_journal_finish_inode_data_buffers);
EXPORT_SYMBOL(jbd2_journal_init_jbd_inode);
EXPORT_SYMBOL(jbd2_journal_release_jbd_inode);
EXPORT_SYMBOL(jbd2_journal_begin_ordered_truncate);
EXPORT_SYMBOL(jbd2_inode_cache);
static int jbd2_journal_create_slab(size_t slab_size);
jbd2: use a single printk for jbd_debug() Since the jbd_debug() is implemented with two separate printk() calls, it can lead to corrupted and misleading debug output like the following (see lines marked with "*"): [ 290.339362] (fs/jbd2/journal.c, 203): kjournald2: kjournald2 wakes [ 290.339365] (fs/jbd2/journal.c, 155): kjournald2: commit_sequence=42103, commit_request=42104 [ 290.339369] (fs/jbd2/journal.c, 158): kjournald2: OK, requests differ [* 290.339376] (fs/jbd2/journal.c, 648): jbd2_log_wait_commit: [* 290.339379] (fs/jbd2/commit.c, 370): jbd2_journal_commit_transaction: JBD2: want 42104, j_commit_sequence=42103 [* 290.339382] JBD2: starting commit of transaction 42104 [ 290.339410] (fs/jbd2/revoke.c, 566): jbd2_journal_write_revoke_records: Wrote 0 revoke records [ 290.376555] (fs/jbd2/commit.c, 1088): jbd2_journal_commit_transaction: JBD2: commit 42104 complete, head 42079 i.e. the debug output from log_wait_commit and journal_commit_transaction have become interleaved. The output should have been: (fs/jbd2/journal.c, 648): jbd2_log_wait_commit: JBD2: want 42104, j_commit_sequence=42103 (fs/jbd2/commit.c, 370): jbd2_journal_commit_transaction: JBD2: starting commit of transaction 42104 It is expected that this is not easy to replicate -- I was only able to cause it on preempt-rt kernels, and even then only under heavy I/O load. Reported-by: Paul Gortmaker <paul.gortmaker@windriver.com> Suggested-by: "Theodore Ts'o" <tytso@mit.edu> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2013-06-13 03:04:04 +00:00
#ifdef CONFIG_JBD2_DEBUG
void __jbd2_debug(int level, const char *file, const char *func,
unsigned int line, const char *fmt, ...)
{
struct va_format vaf;
va_list args;
if (level > jbd2_journal_enable_debug)
return;
va_start(args, fmt);
vaf.fmt = fmt;
vaf.va = &args;
printk(KERN_DEBUG "%s: (%s, %u): %pV", file, func, line, &vaf);
jbd2: use a single printk for jbd_debug() Since the jbd_debug() is implemented with two separate printk() calls, it can lead to corrupted and misleading debug output like the following (see lines marked with "*"): [ 290.339362] (fs/jbd2/journal.c, 203): kjournald2: kjournald2 wakes [ 290.339365] (fs/jbd2/journal.c, 155): kjournald2: commit_sequence=42103, commit_request=42104 [ 290.339369] (fs/jbd2/journal.c, 158): kjournald2: OK, requests differ [* 290.339376] (fs/jbd2/journal.c, 648): jbd2_log_wait_commit: [* 290.339379] (fs/jbd2/commit.c, 370): jbd2_journal_commit_transaction: JBD2: want 42104, j_commit_sequence=42103 [* 290.339382] JBD2: starting commit of transaction 42104 [ 290.339410] (fs/jbd2/revoke.c, 566): jbd2_journal_write_revoke_records: Wrote 0 revoke records [ 290.376555] (fs/jbd2/commit.c, 1088): jbd2_journal_commit_transaction: JBD2: commit 42104 complete, head 42079 i.e. the debug output from log_wait_commit and journal_commit_transaction have become interleaved. The output should have been: (fs/jbd2/journal.c, 648): jbd2_log_wait_commit: JBD2: want 42104, j_commit_sequence=42103 (fs/jbd2/commit.c, 370): jbd2_journal_commit_transaction: JBD2: starting commit of transaction 42104 It is expected that this is not easy to replicate -- I was only able to cause it on preempt-rt kernels, and even then only under heavy I/O load. Reported-by: Paul Gortmaker <paul.gortmaker@windriver.com> Suggested-by: "Theodore Ts'o" <tytso@mit.edu> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2013-06-13 03:04:04 +00:00
va_end(args);
}
#endif
/* Checksumming functions */
static __be32 jbd2_superblock_csum(journal_t *j, journal_superblock_t *sb)
{
__u32 csum;
__be32 old_csum;
old_csum = sb->s_checksum;
sb->s_checksum = 0;
csum = jbd2_chksum(j, ~0, (char *)sb, sizeof(journal_superblock_t));
sb->s_checksum = old_csum;
return cpu_to_be32(csum);
}
/*
* Helper function used to manage commit timeouts
*/
static void commit_timeout(struct timer_list *t)
{
journal_t *journal = from_timer(journal, t, j_commit_timer);
wake_up_process(journal->j_task);
}
/*
* kjournald2: The main thread function used to manage a logging device
* journal.
*
* This kernel thread is responsible for two things:
*
* 1) COMMIT: Every so often we need to commit the current state of the
* filesystem to disk. The journal thread is responsible for writing
* all of the metadata buffers to disk. If a fast commit is ongoing
* journal thread waits until it's done and then continues from
* there on.
*
* 2) CHECKPOINT: We cannot reuse a used section of the log file until all
* of the data in that part of the log has been rewritten elsewhere on
* the disk. Flushing these old buffers to reclaim space in the log is
* known as checkpointing, and this thread is responsible for that job.
*/
static int kjournald2(void *arg)
{
journal_t *journal = arg;
transaction_t *transaction;
/*
* Set up an interval timer which can be used to trigger a commit wakeup
* after the commit interval expires
*/
timer_setup(&journal->j_commit_timer, commit_timeout, 0);
set_freezable();
/* Record that the journal thread is running */
journal->j_task = current;
wake_up(&journal->j_wait_done_commit);
/*
* Make sure that no allocations from this kernel thread will ever
* recurse to the fs layer because we are responsible for the
* transaction commit and any fs involvement might get stuck waiting for
* the trasn. commit.
*/
memalloc_nofs_save();
/*
* And now, wait forever for commit wakeup events.
*/
write_lock(&journal->j_state_lock);
loop:
if (journal->j_flags & JBD2_UNMOUNT)
goto end_loop;
jbd2_debug(1, "commit_sequence=%u, commit_request=%u\n",
journal->j_commit_sequence, journal->j_commit_request);
if (journal->j_commit_sequence != journal->j_commit_request) {
jbd2_debug(1, "OK, requests differ\n");
write_unlock(&journal->j_state_lock);
del_timer_sync(&journal->j_commit_timer);
jbd2_journal_commit_transaction(journal);
write_lock(&journal->j_state_lock);
goto loop;
}
wake_up(&journal->j_wait_done_commit);
if (freezing(current)) {
/*
* The simpler the better. Flushing journal isn't a
* good idea, because that depends on threads that may
* be already stopped.
*/
jbd2_debug(1, "Now suspending kjournald2\n");
write_unlock(&journal->j_state_lock);
try_to_freeze();
write_lock(&journal->j_state_lock);
} else {
/*
* We assume on resume that commits are already there,
* so we don't sleep
*/
DEFINE_WAIT(wait);
int should_sleep = 1;
prepare_to_wait(&journal->j_wait_commit, &wait,
TASK_INTERRUPTIBLE);
if (journal->j_commit_sequence != journal->j_commit_request)
should_sleep = 0;
transaction = journal->j_running_transaction;
if (transaction && time_after_eq(jiffies,
transaction->t_expires))
should_sleep = 0;
if (journal->j_flags & JBD2_UNMOUNT)
should_sleep = 0;
if (should_sleep) {
write_unlock(&journal->j_state_lock);
schedule();
write_lock(&journal->j_state_lock);
}
finish_wait(&journal->j_wait_commit, &wait);
}
jbd2_debug(1, "kjournald2 wakes\n");
/*
* Were we woken up by a commit wakeup event?
*/
transaction = journal->j_running_transaction;
if (transaction && time_after_eq(jiffies, transaction->t_expires)) {
journal->j_commit_request = transaction->t_tid;
jbd2_debug(1, "woke because of timeout\n");
}
goto loop;
end_loop:
del_timer_sync(&journal->j_commit_timer);
journal->j_task = NULL;
wake_up(&journal->j_wait_done_commit);
jbd2_debug(1, "Journal thread exiting.\n");
write_unlock(&journal->j_state_lock);
return 0;
}
static int jbd2_journal_start_thread(journal_t *journal)
{
struct task_struct *t;
t = kthread_run(kjournald2, journal, "jbd2/%s",
journal->j_devname);
if (IS_ERR(t))
return PTR_ERR(t);
wait_event(journal->j_wait_done_commit, journal->j_task != NULL);
return 0;
}
static void journal_kill_thread(journal_t *journal)
{
write_lock(&journal->j_state_lock);
journal->j_flags |= JBD2_UNMOUNT;
while (journal->j_task) {
write_unlock(&journal->j_state_lock);
wake_up(&journal->j_wait_commit);
wait_event(journal->j_wait_done_commit, journal->j_task == NULL);
write_lock(&journal->j_state_lock);
}
write_unlock(&journal->j_state_lock);
}
/*
* jbd2_journal_write_metadata_buffer: write a metadata buffer to the journal.
*
* Writes a metadata buffer to a given disk block. The actual IO is not
* performed but a new buffer_head is constructed which labels the data
* to be written with the correct destination disk block.
*
* Any magic-number escaping which needs to be done will cause a
* copy-out here. If the buffer happens to start with the
* JBD2_MAGIC_NUMBER, then we can't write it to the log directly: the
* magic number is only written to the log for descripter blocks. In
* this case, we copy the data and replace the first word with 0, and we
* return a result code which indicates that this buffer needs to be
* marked as an escaped buffer in the corresponding log descriptor
* block. The missing word can then be restored when the block is read
* during recovery.
*
* If the source buffer has already been modified by a new transaction
* since we took the last commit snapshot, we use the frozen copy of
* that data for IO. If we end up using the existing buffer_head's data
* for the write, then we have to make sure nobody modifies it while the
* IO is in progress. do_get_write_access() handles this.
*
* The function returns a pointer to the buffer_head to be used for IO.
*
*
* Return value:
* <0: Error
* >=0: Finished OK
*
* On success:
* Bit 0 set == escape performed on the data
* Bit 1 set == buffer copy-out performed (kfree the data after IO)
*/
int jbd2_journal_write_metadata_buffer(transaction_t *transaction,
struct journal_head *jh_in,
struct buffer_head **bh_out,
sector_t blocknr)
{
int need_copy_out = 0;
int done_copy_out = 0;
int do_escape = 0;
char *mapped_data;
struct buffer_head *new_bh;
struct folio *new_folio;
unsigned int new_offset;
struct buffer_head *bh_in = jh2bh(jh_in);
journal_t *journal = transaction->t_journal;
/*
* The buffer really shouldn't be locked: only the current committing
* transaction is allowed to write it, so nobody else is allowed
* to do any IO.
*
* akpm: except if we're journalling data, and write() output is
* also part of a shared mapping, and another thread has
* decided to launch a writepage() against this buffer.
*/
J_ASSERT_BH(bh_in, buffer_jbddirty(bh_in));
new_bh = alloc_buffer_head(GFP_NOFS|__GFP_NOFAIL);
/* keep subsequent assertions sane */
atomic_set(&new_bh->b_count, 1);
jbd2: Make state lock a spinlock Bit-spinlocks are problematic on PREEMPT_RT if functions which might sleep on RT, e.g. spin_lock(), alloc/free(), are invoked inside the lock held region because bit spinlocks disable preemption even on RT. A first attempt was to replace state lock with a spinlock placed in struct buffer_head and make the locking conditional on PREEMPT_RT and DEBUG_BIT_SPINLOCKS. Jan pointed out that there is a 4 byte hole in struct journal_head where a regular spinlock fits in and he would not object to convert the state lock to a spinlock unconditionally. Aside of solving the RT problem, this also gains lockdep coverage for the journal head state lock (bit-spinlocks are not covered by lockdep as it's hard to fit a lockdep map into a single bit). The trivial change would have been to convert the jbd_*lock_bh_state() inlines, but that comes with the downside that these functions take a buffer head pointer which needs to be converted to a journal head pointer which adds another level of indirection. As almost all functions which use this lock have a journal head pointer readily available, it makes more sense to remove the lock helper inlines and write out spin_*lock() at all call sites. Fixup all locking comments as well. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joseph Qi <joseph.qi@linux.alibaba.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Jan Kara <jack@suse.com> Cc: linux-ext4@vger.kernel.org Link: https://lore.kernel.org/r/20190809124233.13277-7-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:32 +00:00
spin_lock(&jh_in->b_state_lock);
repeat:
/*
* If a new transaction has already done a buffer copy-out, then
* we use that version of the data for the commit.
*/
if (jh_in->b_frozen_data) {
done_copy_out = 1;
new_folio = virt_to_folio(jh_in->b_frozen_data);
new_offset = offset_in_folio(new_folio, jh_in->b_frozen_data);
} else {
new_folio = jh2bh(jh_in)->b_folio;
new_offset = offset_in_folio(new_folio, jh2bh(jh_in)->b_data);
}
mapped_data = kmap_local_folio(new_folio, new_offset);
jbd2: Add buffer triggers Filesystems often to do compute intensive operation on some metadata. If this operation is repeated many times, it can be very expensive. It would be much nicer if the operation could be performed once before a buffer goes to disk. This adds triggers to jbd2 buffer heads. Just before writing a metadata buffer to the journal, jbd2 will optionally call a commit trigger associated with the buffer. If the journal is aborted, an abort trigger will be called on any dirty buffers as they are dropped from pending transactions. ocfs2 will use this feature. Initially I tried to come up with a more generic trigger that could be used for non-buffer-related events like transaction completion. It doesn't tie nicely, because the information a buffer trigger needs (specific to a journal_head) isn't the same as what a transaction trigger needs (specific to a tranaction_t or perhaps journal_t). So I implemented a buffer set, with the understanding that journal/transaction wide triggers should be implemented separately. There is only one trigger set allowed per buffer. I can't think of any reason to attach more than one set. Contrast this with a journal or transaction in which multiple places may want to watch the entire transaction separately. The trigger sets are considered static allocation from the jbd2 perspective. ocfs2 will just have one trigger set per block type, setting the same set on every bh of the same type. Signed-off-by: Joel Becker <joel.becker@oracle.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: <linux-ext4@vger.kernel.org> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2008-09-11 22:35:47 +00:00
/*
jbd2/ocfs2: Fix block checksumming when a buffer is used in several transactions OCFS2 uses t_commit trigger to compute and store checksum of the just committed blocks. When a buffer has b_frozen_data, checksum is computed for it instead of b_data but this can result in an old checksum being written to the filesystem in the following scenario: 1) transaction1 is opened 2) handle1 is opened 3) journal_access(handle1, bh) - This sets jh->b_transaction to transaction1 4) modify(bh) 5) journal_dirty(handle1, bh) 6) handle1 is closed 7) start committing transaction1, opening transaction2 8) handle2 is opened 9) journal_access(handle2, bh) - This copies off b_frozen_data to make it safe for transaction1 to commit. jh->b_next_transaction is set to transaction2. 10) jbd2_journal_write_metadata() checksums b_frozen_data 11) the journal correctly writes b_frozen_data to the disk journal 12) handle2 is closed - There was no dirty call for the bh on handle2, so it is never queued for any more journal operation 13) Checkpointing finally happens, and it just spools the bh via normal buffer writeback. This will write b_data, which was never triggered on and thus contains a wrong (old) checksum. This patch fixes the problem by calling the trigger at the moment data is frozen for journal commit - i.e., either when b_frozen_data is created by do_get_write_access or just before we write a buffer to the log if b_frozen_data does not exist. We also rename the trigger to t_frozen as that better describes when it is called. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Mark Fasheh <mfasheh@suse.com> Signed-off-by: Joel Becker <joel.becker@oracle.com>
2010-07-14 05:56:33 +00:00
* Fire data frozen trigger if data already wasn't frozen. Do this
* before checking for escaping, as the trigger may modify the magic
* offset. If a copy-out happens afterwards, it will have the correct
* data in the buffer.
jbd2: Add buffer triggers Filesystems often to do compute intensive operation on some metadata. If this operation is repeated many times, it can be very expensive. It would be much nicer if the operation could be performed once before a buffer goes to disk. This adds triggers to jbd2 buffer heads. Just before writing a metadata buffer to the journal, jbd2 will optionally call a commit trigger associated with the buffer. If the journal is aborted, an abort trigger will be called on any dirty buffers as they are dropped from pending transactions. ocfs2 will use this feature. Initially I tried to come up with a more generic trigger that could be used for non-buffer-related events like transaction completion. It doesn't tie nicely, because the information a buffer trigger needs (specific to a journal_head) isn't the same as what a transaction trigger needs (specific to a tranaction_t or perhaps journal_t). So I implemented a buffer set, with the understanding that journal/transaction wide triggers should be implemented separately. There is only one trigger set allowed per buffer. I can't think of any reason to attach more than one set. Contrast this with a journal or transaction in which multiple places may want to watch the entire transaction separately. The trigger sets are considered static allocation from the jbd2 perspective. ocfs2 will just have one trigger set per block type, setting the same set on every bh of the same type. Signed-off-by: Joel Becker <joel.becker@oracle.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: <linux-ext4@vger.kernel.org> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2008-09-11 22:35:47 +00:00
*/
jbd2/ocfs2: Fix block checksumming when a buffer is used in several transactions OCFS2 uses t_commit trigger to compute and store checksum of the just committed blocks. When a buffer has b_frozen_data, checksum is computed for it instead of b_data but this can result in an old checksum being written to the filesystem in the following scenario: 1) transaction1 is opened 2) handle1 is opened 3) journal_access(handle1, bh) - This sets jh->b_transaction to transaction1 4) modify(bh) 5) journal_dirty(handle1, bh) 6) handle1 is closed 7) start committing transaction1, opening transaction2 8) handle2 is opened 9) journal_access(handle2, bh) - This copies off b_frozen_data to make it safe for transaction1 to commit. jh->b_next_transaction is set to transaction2. 10) jbd2_journal_write_metadata() checksums b_frozen_data 11) the journal correctly writes b_frozen_data to the disk journal 12) handle2 is closed - There was no dirty call for the bh on handle2, so it is never queued for any more journal operation 13) Checkpointing finally happens, and it just spools the bh via normal buffer writeback. This will write b_data, which was never triggered on and thus contains a wrong (old) checksum. This patch fixes the problem by calling the trigger at the moment data is frozen for journal commit - i.e., either when b_frozen_data is created by do_get_write_access or just before we write a buffer to the log if b_frozen_data does not exist. We also rename the trigger to t_frozen as that better describes when it is called. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Mark Fasheh <mfasheh@suse.com> Signed-off-by: Joel Becker <joel.becker@oracle.com>
2010-07-14 05:56:33 +00:00
if (!done_copy_out)
jbd2_buffer_frozen_trigger(jh_in, mapped_data,
jbd2/ocfs2: Fix block checksumming when a buffer is used in several transactions OCFS2 uses t_commit trigger to compute and store checksum of the just committed blocks. When a buffer has b_frozen_data, checksum is computed for it instead of b_data but this can result in an old checksum being written to the filesystem in the following scenario: 1) transaction1 is opened 2) handle1 is opened 3) journal_access(handle1, bh) - This sets jh->b_transaction to transaction1 4) modify(bh) 5) journal_dirty(handle1, bh) 6) handle1 is closed 7) start committing transaction1, opening transaction2 8) handle2 is opened 9) journal_access(handle2, bh) - This copies off b_frozen_data to make it safe for transaction1 to commit. jh->b_next_transaction is set to transaction2. 10) jbd2_journal_write_metadata() checksums b_frozen_data 11) the journal correctly writes b_frozen_data to the disk journal 12) handle2 is closed - There was no dirty call for the bh on handle2, so it is never queued for any more journal operation 13) Checkpointing finally happens, and it just spools the bh via normal buffer writeback. This will write b_data, which was never triggered on and thus contains a wrong (old) checksum. This patch fixes the problem by calling the trigger at the moment data is frozen for journal commit - i.e., either when b_frozen_data is created by do_get_write_access or just before we write a buffer to the log if b_frozen_data does not exist. We also rename the trigger to t_frozen as that better describes when it is called. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Mark Fasheh <mfasheh@suse.com> Signed-off-by: Joel Becker <joel.becker@oracle.com>
2010-07-14 05:56:33 +00:00
jh_in->b_triggers);
jbd2: Add buffer triggers Filesystems often to do compute intensive operation on some metadata. If this operation is repeated many times, it can be very expensive. It would be much nicer if the operation could be performed once before a buffer goes to disk. This adds triggers to jbd2 buffer heads. Just before writing a metadata buffer to the journal, jbd2 will optionally call a commit trigger associated with the buffer. If the journal is aborted, an abort trigger will be called on any dirty buffers as they are dropped from pending transactions. ocfs2 will use this feature. Initially I tried to come up with a more generic trigger that could be used for non-buffer-related events like transaction completion. It doesn't tie nicely, because the information a buffer trigger needs (specific to a journal_head) isn't the same as what a transaction trigger needs (specific to a tranaction_t or perhaps journal_t). So I implemented a buffer set, with the understanding that journal/transaction wide triggers should be implemented separately. There is only one trigger set allowed per buffer. I can't think of any reason to attach more than one set. Contrast this with a journal or transaction in which multiple places may want to watch the entire transaction separately. The trigger sets are considered static allocation from the jbd2 perspective. ocfs2 will just have one trigger set per block type, setting the same set on every bh of the same type. Signed-off-by: Joel Becker <joel.becker@oracle.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: <linux-ext4@vger.kernel.org> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2008-09-11 22:35:47 +00:00
/*
* Check for escaping
*/
if (*((__be32 *)mapped_data) == cpu_to_be32(JBD2_MAGIC_NUMBER)) {
need_copy_out = 1;
do_escape = 1;
}
kunmap_local(mapped_data);
/*
* Do we need to do a data copy?
*/
if (need_copy_out && !done_copy_out) {
char *tmp;
jbd2: Make state lock a spinlock Bit-spinlocks are problematic on PREEMPT_RT if functions which might sleep on RT, e.g. spin_lock(), alloc/free(), are invoked inside the lock held region because bit spinlocks disable preemption even on RT. A first attempt was to replace state lock with a spinlock placed in struct buffer_head and make the locking conditional on PREEMPT_RT and DEBUG_BIT_SPINLOCKS. Jan pointed out that there is a 4 byte hole in struct journal_head where a regular spinlock fits in and he would not object to convert the state lock to a spinlock unconditionally. Aside of solving the RT problem, this also gains lockdep coverage for the journal head state lock (bit-spinlocks are not covered by lockdep as it's hard to fit a lockdep map into a single bit). The trivial change would have been to convert the jbd_*lock_bh_state() inlines, but that comes with the downside that these functions take a buffer head pointer which needs to be converted to a journal head pointer which adds another level of indirection. As almost all functions which use this lock have a journal head pointer readily available, it makes more sense to remove the lock helper inlines and write out spin_*lock() at all call sites. Fixup all locking comments as well. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joseph Qi <joseph.qi@linux.alibaba.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Jan Kara <jack@suse.com> Cc: linux-ext4@vger.kernel.org Link: https://lore.kernel.org/r/20190809124233.13277-7-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:32 +00:00
spin_unlock(&jh_in->b_state_lock);
tmp = jbd2_alloc(bh_in->b_size, GFP_NOFS);
if (!tmp) {
brelse(new_bh);
return -ENOMEM;
}
jbd2: Make state lock a spinlock Bit-spinlocks are problematic on PREEMPT_RT if functions which might sleep on RT, e.g. spin_lock(), alloc/free(), are invoked inside the lock held region because bit spinlocks disable preemption even on RT. A first attempt was to replace state lock with a spinlock placed in struct buffer_head and make the locking conditional on PREEMPT_RT and DEBUG_BIT_SPINLOCKS. Jan pointed out that there is a 4 byte hole in struct journal_head where a regular spinlock fits in and he would not object to convert the state lock to a spinlock unconditionally. Aside of solving the RT problem, this also gains lockdep coverage for the journal head state lock (bit-spinlocks are not covered by lockdep as it's hard to fit a lockdep map into a single bit). The trivial change would have been to convert the jbd_*lock_bh_state() inlines, but that comes with the downside that these functions take a buffer head pointer which needs to be converted to a journal head pointer which adds another level of indirection. As almost all functions which use this lock have a journal head pointer readily available, it makes more sense to remove the lock helper inlines and write out spin_*lock() at all call sites. Fixup all locking comments as well. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joseph Qi <joseph.qi@linux.alibaba.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Jan Kara <jack@suse.com> Cc: linux-ext4@vger.kernel.org Link: https://lore.kernel.org/r/20190809124233.13277-7-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:32 +00:00
spin_lock(&jh_in->b_state_lock);
if (jh_in->b_frozen_data) {
jbd2_free(tmp, bh_in->b_size);
goto repeat;
}
jh_in->b_frozen_data = tmp;
memcpy_from_folio(tmp, new_folio, new_offset, bh_in->b_size);
new_folio = virt_to_folio(tmp);
new_offset = offset_in_folio(new_folio, tmp);
done_copy_out = 1;
jbd2: Add buffer triggers Filesystems often to do compute intensive operation on some metadata. If this operation is repeated many times, it can be very expensive. It would be much nicer if the operation could be performed once before a buffer goes to disk. This adds triggers to jbd2 buffer heads. Just before writing a metadata buffer to the journal, jbd2 will optionally call a commit trigger associated with the buffer. If the journal is aborted, an abort trigger will be called on any dirty buffers as they are dropped from pending transactions. ocfs2 will use this feature. Initially I tried to come up with a more generic trigger that could be used for non-buffer-related events like transaction completion. It doesn't tie nicely, because the information a buffer trigger needs (specific to a journal_head) isn't the same as what a transaction trigger needs (specific to a tranaction_t or perhaps journal_t). So I implemented a buffer set, with the understanding that journal/transaction wide triggers should be implemented separately. There is only one trigger set allowed per buffer. I can't think of any reason to attach more than one set. Contrast this with a journal or transaction in which multiple places may want to watch the entire transaction separately. The trigger sets are considered static allocation from the jbd2 perspective. ocfs2 will just have one trigger set per block type, setting the same set on every bh of the same type. Signed-off-by: Joel Becker <joel.becker@oracle.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: <linux-ext4@vger.kernel.org> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
2008-09-11 22:35:47 +00:00
/*
* This isn't strictly necessary, as we're using frozen
* data for the escaping, but it keeps consistency with
* b_frozen_data usage.
*/
jh_in->b_frozen_triggers = jh_in->b_triggers;
}
/*
* Did we need to do an escaping? Now we've done all the
* copying, we can finally do so.
*/
if (do_escape) {
mapped_data = kmap_local_folio(new_folio, new_offset);
*((unsigned int *)mapped_data) = 0;
kunmap_local(mapped_data);
}
folio_set_bh(new_bh, new_folio, new_offset);
new_bh->b_size = bh_in->b_size;
new_bh->b_bdev = journal->j_dev;
new_bh->b_blocknr = blocknr;
new_bh->b_private = bh_in;
set_buffer_mapped(new_bh);
set_buffer_dirty(new_bh);
*bh_out = new_bh;
/*
* The to-be-written buffer needs to get moved to the io queue,
* and the original buffer whose contents we are shadowing or
* copying is moved to the transaction's shadow queue.
*/
JBUFFER_TRACE(jh_in, "file as BJ_Shadow");
spin_lock(&journal->j_list_lock);
__jbd2_journal_file_buffer(jh_in, transaction, BJ_Shadow);
spin_unlock(&journal->j_list_lock);
set_buffer_shadow(bh_in);
jbd2: Make state lock a spinlock Bit-spinlocks are problematic on PREEMPT_RT if functions which might sleep on RT, e.g. spin_lock(), alloc/free(), are invoked inside the lock held region because bit spinlocks disable preemption even on RT. A first attempt was to replace state lock with a spinlock placed in struct buffer_head and make the locking conditional on PREEMPT_RT and DEBUG_BIT_SPINLOCKS. Jan pointed out that there is a 4 byte hole in struct journal_head where a regular spinlock fits in and he would not object to convert the state lock to a spinlock unconditionally. Aside of solving the RT problem, this also gains lockdep coverage for the journal head state lock (bit-spinlocks are not covered by lockdep as it's hard to fit a lockdep map into a single bit). The trivial change would have been to convert the jbd_*lock_bh_state() inlines, but that comes with the downside that these functions take a buffer head pointer which needs to be converted to a journal head pointer which adds another level of indirection. As almost all functions which use this lock have a journal head pointer readily available, it makes more sense to remove the lock helper inlines and write out spin_*lock() at all call sites. Fixup all locking comments as well. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joseph Qi <joseph.qi@linux.alibaba.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Jan Kara <jack@suse.com> Cc: linux-ext4@vger.kernel.org Link: https://lore.kernel.org/r/20190809124233.13277-7-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:32 +00:00
spin_unlock(&jh_in->b_state_lock);
return do_escape | (done_copy_out << 1);
}
/*
* Allocation code for the journal file. Manage the space left in the
* journal, so that we can begin checkpointing when appropriate.
*/
/*
* Called with j_state_lock locked for writing.
* Returns true if a transaction commit was started.
*/
static int __jbd2_log_start_commit(journal_t *journal, tid_t target)
{
/* Return if the txn has already requested to be committed */
if (journal->j_commit_request == target)
return 0;
/*
* The only transaction we can possibly wait upon is the
* currently running transaction (if it exists). Otherwise,
* the target tid must be an old one.
*/
if (journal->j_running_transaction &&
journal->j_running_transaction->t_tid == target) {
/*
* We want a new commit: OK, mark the request and wakeup the
* commit thread. We do _not_ do the commit ourselves.
*/
journal->j_commit_request = target;
jbd2_debug(1, "JBD2: requesting commit %u/%u\n",
journal->j_commit_request,
journal->j_commit_sequence);
journal->j_running_transaction->t_requested = jiffies;
wake_up(&journal->j_wait_commit);
return 1;
} else if (!tid_geq(journal->j_commit_request, target))
/* This should never happen, but if it does, preserve
the evidence before kjournald goes into a loop and
increments j_commit_sequence beyond all recognition. */
WARN_ONCE(1, "JBD2: bad log_start_commit: %u %u %u %u\n",
journal->j_commit_request,
journal->j_commit_sequence,
target, journal->j_running_transaction ?
journal->j_running_transaction->t_tid : 0);
return 0;
}
int jbd2_log_start_commit(journal_t *journal, tid_t tid)
{
int ret;
write_lock(&journal->j_state_lock);
ret = __jbd2_log_start_commit(journal, tid);
write_unlock(&journal->j_state_lock);
return ret;
}
/*
* Force and wait any uncommitted transactions. We can only force the running
* transaction if we don't have an active handle, otherwise, we will deadlock.
* Returns: <0 in case of error,
* 0 if nothing to commit,
* 1 if transaction was successfully committed.
*/
static int __jbd2_journal_force_commit(journal_t *journal)
{
transaction_t *transaction = NULL;
tid_t tid;
int need_to_start = 0, ret = 0;
read_lock(&journal->j_state_lock);
if (journal->j_running_transaction && !current->journal_info) {
transaction = journal->j_running_transaction;
if (!tid_geq(journal->j_commit_request, transaction->t_tid))
need_to_start = 1;
} else if (journal->j_committing_transaction)
transaction = journal->j_committing_transaction;
if (!transaction) {
/* Nothing to commit */
read_unlock(&journal->j_state_lock);
return 0;
}
tid = transaction->t_tid;
read_unlock(&journal->j_state_lock);
if (need_to_start)
jbd2_log_start_commit(journal, tid);
ret = jbd2_log_wait_commit(journal, tid);
if (!ret)
ret = 1;
return ret;
}
/**
* jbd2_journal_force_commit_nested - Force and wait upon a commit if the
* calling process is not within transaction.
*
* @journal: journal to force
* Returns true if progress was made.
*
* This is used for forcing out undo-protected data which contains
* bitmaps, when the fs is running out of space.
*/
int jbd2_journal_force_commit_nested(journal_t *journal)
{
int ret;
ret = __jbd2_journal_force_commit(journal);
return ret > 0;
}
/**
* jbd2_journal_force_commit() - force any uncommitted transactions
* @journal: journal to force
*
* Caller want unconditional commit. We can only force the running transaction
* if we don't have an active handle, otherwise, we will deadlock.
*/
int jbd2_journal_force_commit(journal_t *journal)
{
int ret;
J_ASSERT(!current->journal_info);
ret = __jbd2_journal_force_commit(journal);
if (ret > 0)
ret = 0;
return ret;
}
/*
* Start a commit of the current running transaction (if any). Returns true
jbd2: Fix return value of jbd2_journal_start_commit() The function jbd2_journal_start_commit() returns 1 if either a transaction is committing or the function has queued a transaction commit. But it returns 0 if we raced with somebody queueing the transaction commit as well. This resulted in ext4_sync_fs() not functioning correctly (description from Arthur Jones): In the case of a data=ordered umount with pending long symlinks which are delayed due to a long list of other I/O on the backing block device, this causes the buffer associated with the long symlinks to not be moved to the inode dirty list in the second phase of fsync_super. Then, before they can be dirtied again, kjournald exits, seeing the UMOUNT flag and the dirty pages are never written to the backing block device, causing long symlink corruption and exposing new or previously freed block data to userspace. This can be reproduced with a script created by Eric Sandeen <sandeen@redhat.com>: #!/bin/bash umount /mnt/test2 mount /dev/sdb4 /mnt/test2 rm -f /mnt/test2/* dd if=/dev/zero of=/mnt/test2/bigfile bs=1M count=512 touch /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename ln -s /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename /mnt/test2/link umount /mnt/test2 mount /dev/sdb4 /mnt/test2 ls /mnt/test2/ This patch fixes jbd2_journal_start_commit() to always return 1 when there's a transaction committing or queued for commit. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> CC: Eric Sandeen <sandeen@redhat.com> CC: linux-ext4@vger.kernel.org
2009-02-10 16:27:46 +00:00
* if a transaction is going to be committed (or is currently already
* committing), and fills its tid in at *ptid
*/
int jbd2_journal_start_commit(journal_t *journal, tid_t *ptid)
{
int ret = 0;
write_lock(&journal->j_state_lock);
if (journal->j_running_transaction) {
tid_t tid = journal->j_running_transaction->t_tid;
jbd2: Fix return value of jbd2_journal_start_commit() The function jbd2_journal_start_commit() returns 1 if either a transaction is committing or the function has queued a transaction commit. But it returns 0 if we raced with somebody queueing the transaction commit as well. This resulted in ext4_sync_fs() not functioning correctly (description from Arthur Jones): In the case of a data=ordered umount with pending long symlinks which are delayed due to a long list of other I/O on the backing block device, this causes the buffer associated with the long symlinks to not be moved to the inode dirty list in the second phase of fsync_super. Then, before they can be dirtied again, kjournald exits, seeing the UMOUNT flag and the dirty pages are never written to the backing block device, causing long symlink corruption and exposing new or previously freed block data to userspace. This can be reproduced with a script created by Eric Sandeen <sandeen@redhat.com>: #!/bin/bash umount /mnt/test2 mount /dev/sdb4 /mnt/test2 rm -f /mnt/test2/* dd if=/dev/zero of=/mnt/test2/bigfile bs=1M count=512 touch /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename ln -s /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename /mnt/test2/link umount /mnt/test2 mount /dev/sdb4 /mnt/test2 ls /mnt/test2/ This patch fixes jbd2_journal_start_commit() to always return 1 when there's a transaction committing or queued for commit. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> CC: Eric Sandeen <sandeen@redhat.com> CC: linux-ext4@vger.kernel.org
2009-02-10 16:27:46 +00:00
__jbd2_log_start_commit(journal, tid);
/* There's a running transaction and we've just made sure
* it's commit has been scheduled. */
if (ptid)
*ptid = tid;
jbd2: Fix return value of jbd2_journal_start_commit() The function jbd2_journal_start_commit() returns 1 if either a transaction is committing or the function has queued a transaction commit. But it returns 0 if we raced with somebody queueing the transaction commit as well. This resulted in ext4_sync_fs() not functioning correctly (description from Arthur Jones): In the case of a data=ordered umount with pending long symlinks which are delayed due to a long list of other I/O on the backing block device, this causes the buffer associated with the long symlinks to not be moved to the inode dirty list in the second phase of fsync_super. Then, before they can be dirtied again, kjournald exits, seeing the UMOUNT flag and the dirty pages are never written to the backing block device, causing long symlink corruption and exposing new or previously freed block data to userspace. This can be reproduced with a script created by Eric Sandeen <sandeen@redhat.com>: #!/bin/bash umount /mnt/test2 mount /dev/sdb4 /mnt/test2 rm -f /mnt/test2/* dd if=/dev/zero of=/mnt/test2/bigfile bs=1M count=512 touch /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename ln -s /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename /mnt/test2/link umount /mnt/test2 mount /dev/sdb4 /mnt/test2 ls /mnt/test2/ This patch fixes jbd2_journal_start_commit() to always return 1 when there's a transaction committing or queued for commit. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> CC: Eric Sandeen <sandeen@redhat.com> CC: linux-ext4@vger.kernel.org
2009-02-10 16:27:46 +00:00
ret = 1;
} else if (journal->j_committing_transaction) {
/*
* If commit has been started, then we have to wait for
* completion of that transaction.
*/
jbd2: Fix return value of jbd2_journal_start_commit() The function jbd2_journal_start_commit() returns 1 if either a transaction is committing or the function has queued a transaction commit. But it returns 0 if we raced with somebody queueing the transaction commit as well. This resulted in ext4_sync_fs() not functioning correctly (description from Arthur Jones): In the case of a data=ordered umount with pending long symlinks which are delayed due to a long list of other I/O on the backing block device, this causes the buffer associated with the long symlinks to not be moved to the inode dirty list in the second phase of fsync_super. Then, before they can be dirtied again, kjournald exits, seeing the UMOUNT flag and the dirty pages are never written to the backing block device, causing long symlink corruption and exposing new or previously freed block data to userspace. This can be reproduced with a script created by Eric Sandeen <sandeen@redhat.com>: #!/bin/bash umount /mnt/test2 mount /dev/sdb4 /mnt/test2 rm -f /mnt/test2/* dd if=/dev/zero of=/mnt/test2/bigfile bs=1M count=512 touch /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename ln -s /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename /mnt/test2/link umount /mnt/test2 mount /dev/sdb4 /mnt/test2 ls /mnt/test2/ This patch fixes jbd2_journal_start_commit() to always return 1 when there's a transaction committing or queued for commit. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> CC: Eric Sandeen <sandeen@redhat.com> CC: linux-ext4@vger.kernel.org
2009-02-10 16:27:46 +00:00
if (ptid)
*ptid = journal->j_committing_transaction->t_tid;
ret = 1;
}
write_unlock(&journal->j_state_lock);
return ret;
}
/*
* Return 1 if a given transaction has not yet sent barrier request
* connected with a transaction commit. If 0 is returned, transaction
* may or may not have sent the barrier. Used to avoid sending barrier
* twice in common cases.
*/
int jbd2_trans_will_send_data_barrier(journal_t *journal, tid_t tid)
{
int ret = 0;
transaction_t *commit_trans;
if (!(journal->j_flags & JBD2_BARRIER))
return 0;
read_lock(&journal->j_state_lock);
/* Transaction already committed? */
if (tid_geq(journal->j_commit_sequence, tid))
goto out;
commit_trans = journal->j_committing_transaction;
if (!commit_trans || commit_trans->t_tid != tid) {
ret = 1;
goto out;
}
/*
* Transaction is being committed and we already proceeded to
* submitting a flush to fs partition?
*/
if (journal->j_fs_dev != journal->j_dev) {
if (!commit_trans->t_need_data_flush ||
commit_trans->t_state >= T_COMMIT_DFLUSH)
goto out;
} else {
if (commit_trans->t_state >= T_COMMIT_JFLUSH)
goto out;
}
ret = 1;
out:
read_unlock(&journal->j_state_lock);
return ret;
}
EXPORT_SYMBOL(jbd2_trans_will_send_data_barrier);
/*
* Wait for a specified commit to complete.
* The caller may not hold the journal lock.
*/
int jbd2_log_wait_commit(journal_t *journal, tid_t tid)
{
int err = 0;
read_lock(&journal->j_state_lock);
#ifdef CONFIG_PROVE_LOCKING
/*
* Some callers make sure transaction is already committing and in that
* case we cannot block on open handles anymore. So don't warn in that
* case.
*/
if (tid_gt(tid, journal->j_commit_sequence) &&
(!journal->j_committing_transaction ||
journal->j_committing_transaction->t_tid != tid)) {
read_unlock(&journal->j_state_lock);
jbd2_might_wait_for_commit(journal);
read_lock(&journal->j_state_lock);
}
#endif
#ifdef CONFIG_JBD2_DEBUG
if (!tid_geq(journal->j_commit_request, tid)) {
printk(KERN_ERR
"%s: error: j_commit_request=%u, tid=%u\n",
__func__, journal->j_commit_request, tid);
}
#endif
while (tid_gt(tid, journal->j_commit_sequence)) {
jbd2_debug(1, "JBD2: want %u, j_commit_sequence=%u\n",
tid, journal->j_commit_sequence);
read_unlock(&journal->j_state_lock);
wake_up(&journal->j_wait_commit);
wait_event(journal->j_wait_done_commit,
!tid_gt(tid, journal->j_commit_sequence));
read_lock(&journal->j_state_lock);
}
read_unlock(&journal->j_state_lock);
if (unlikely(is_journal_aborted(journal)))
err = -EIO;
return err;
}
/*
* Start a fast commit. If there's an ongoing fast or full commit wait for
* it to complete. Returns 0 if a new fast commit was started. Returns -EALREADY
* if a fast commit is not needed, either because there's an already a commit
* going on or this tid has already been committed. Returns -EINVAL if no jbd2
* commit has yet been performed.
*/
int jbd2_fc_begin_commit(journal_t *journal, tid_t tid)
{
if (unlikely(is_journal_aborted(journal)))
return -EIO;
/*
* Fast commits only allowed if at least one full commit has
* been processed.
*/
if (!journal->j_stats.ts_tid)
return -EINVAL;
write_lock(&journal->j_state_lock);
if (tid <= journal->j_commit_sequence) {
write_unlock(&journal->j_state_lock);
return -EALREADY;
}
if (journal->j_flags & JBD2_FULL_COMMIT_ONGOING ||
(journal->j_flags & JBD2_FAST_COMMIT_ONGOING)) {
DEFINE_WAIT(wait);
prepare_to_wait(&journal->j_fc_wait, &wait,
TASK_UNINTERRUPTIBLE);
write_unlock(&journal->j_state_lock);
schedule();
finish_wait(&journal->j_fc_wait, &wait);
return -EALREADY;
}
journal->j_flags |= JBD2_FAST_COMMIT_ONGOING;
write_unlock(&journal->j_state_lock);
jbd2_journal_lock_updates(journal);
return 0;
}
EXPORT_SYMBOL(jbd2_fc_begin_commit);
/*
* Stop a fast commit. If fallback is set, this function starts commit of
* TID tid before any other fast commit can start.
*/
static int __jbd2_fc_end_commit(journal_t *journal, tid_t tid, bool fallback)
{
jbd2_journal_unlock_updates(journal);
if (journal->j_fc_cleanup_callback)
journal->j_fc_cleanup_callback(journal, 0, tid);
write_lock(&journal->j_state_lock);
journal->j_flags &= ~JBD2_FAST_COMMIT_ONGOING;
if (fallback)
journal->j_flags |= JBD2_FULL_COMMIT_ONGOING;
write_unlock(&journal->j_state_lock);
wake_up(&journal->j_fc_wait);
if (fallback)
return jbd2_complete_transaction(journal, tid);
return 0;
}
int jbd2_fc_end_commit(journal_t *journal)
{
return __jbd2_fc_end_commit(journal, 0, false);
}
EXPORT_SYMBOL(jbd2_fc_end_commit);
int jbd2_fc_end_commit_fallback(journal_t *journal)
{
tid_t tid;
read_lock(&journal->j_state_lock);
tid = journal->j_running_transaction ?
journal->j_running_transaction->t_tid : 0;
read_unlock(&journal->j_state_lock);
return __jbd2_fc_end_commit(journal, tid, true);
}
EXPORT_SYMBOL(jbd2_fc_end_commit_fallback);
/* Return 1 when transaction with given tid has already committed. */
int jbd2_transaction_committed(journal_t *journal, tid_t tid)
{
int ret = 1;
read_lock(&journal->j_state_lock);
if (journal->j_running_transaction &&
journal->j_running_transaction->t_tid == tid)
ret = 0;
if (journal->j_committing_transaction &&
journal->j_committing_transaction->t_tid == tid)
ret = 0;
read_unlock(&journal->j_state_lock);
return ret;
}
EXPORT_SYMBOL(jbd2_transaction_committed);
/*
* When this function returns the transaction corresponding to tid
* will be completed. If the transaction has currently running, start
* committing that transaction before waiting for it to complete. If
* the transaction id is stale, it is by definition already completed,
* so just return SUCCESS.
*/
int jbd2_complete_transaction(journal_t *journal, tid_t tid)
{
int need_to_wait = 1;
read_lock(&journal->j_state_lock);
if (journal->j_running_transaction &&
journal->j_running_transaction->t_tid == tid) {
if (journal->j_commit_request != tid) {
/* transaction not yet started, so request it */
read_unlock(&journal->j_state_lock);
jbd2_log_start_commit(journal, tid);
goto wait_commit;
}
} else if (!(journal->j_committing_transaction &&
journal->j_committing_transaction->t_tid == tid))
need_to_wait = 0;
read_unlock(&journal->j_state_lock);
if (!need_to_wait)
return 0;
wait_commit:
return jbd2_log_wait_commit(journal, tid);
}
EXPORT_SYMBOL(jbd2_complete_transaction);
/*
* Log buffer allocation routines:
*/
int jbd2_journal_next_log_block(journal_t *journal, unsigned long long *retp)
{
unsigned long blocknr;
write_lock(&journal->j_state_lock);
J_ASSERT(journal->j_free > 1);
blocknr = journal->j_head;
journal->j_head++;
journal->j_free--;
if (journal->j_head == journal->j_last)
journal->j_head = journal->j_first;
write_unlock(&journal->j_state_lock);
return jbd2_journal_bmap(journal, blocknr, retp);
}
/* Map one fast commit buffer for use by the file system */
int jbd2_fc_get_buf(journal_t *journal, struct buffer_head **bh_out)
{
unsigned long long pblock;
unsigned long blocknr;
int ret = 0;
struct buffer_head *bh;
int fc_off;
*bh_out = NULL;
if (journal->j_fc_off + journal->j_fc_first < journal->j_fc_last) {
fc_off = journal->j_fc_off;
blocknr = journal->j_fc_first + fc_off;
journal->j_fc_off++;
} else {
ret = -EINVAL;
}
if (ret)
return ret;
ret = jbd2_journal_bmap(journal, blocknr, &pblock);
if (ret)
return ret;
bh = __getblk(journal->j_dev, pblock, journal->j_blocksize);
if (!bh)
return -ENOMEM;
journal->j_fc_wbuf[fc_off] = bh;
*bh_out = bh;
return 0;
}
EXPORT_SYMBOL(jbd2_fc_get_buf);
/*
* Wait on fast commit buffers that were allocated by jbd2_fc_get_buf
* for completion.
*/
int jbd2_fc_wait_bufs(journal_t *journal, int num_blks)
{
struct buffer_head *bh;
int i, j_fc_off;
j_fc_off = journal->j_fc_off;
/*
* Wait in reverse order to minimize chances of us being woken up before
* all IOs have completed
*/
for (i = j_fc_off - 1; i >= j_fc_off - num_blks; i--) {
bh = journal->j_fc_wbuf[i];
wait_on_buffer(bh);
/*
* Update j_fc_off so jbd2_fc_release_bufs can release remain
* buffer head.
*/
if (unlikely(!buffer_uptodate(bh))) {
journal->j_fc_off = i + 1;
return -EIO;
}
put_bh(bh);
journal->j_fc_wbuf[i] = NULL;
}
return 0;
}
EXPORT_SYMBOL(jbd2_fc_wait_bufs);
int jbd2_fc_release_bufs(journal_t *journal)
{
struct buffer_head *bh;
int i, j_fc_off;
j_fc_off = journal->j_fc_off;
for (i = j_fc_off - 1; i >= 0; i--) {
bh = journal->j_fc_wbuf[i];
if (!bh)
break;
put_bh(bh);
journal->j_fc_wbuf[i] = NULL;
}
return 0;
}
EXPORT_SYMBOL(jbd2_fc_release_bufs);
/*
* Conversion of logical to physical block numbers for the journal
*
* On external journals the journal blocks are identity-mapped, so
* this is a no-op. If needed, we can use j_blk_offset - everything is
* ready.
*/
int jbd2_journal_bmap(journal_t *journal, unsigned long blocknr,
unsigned long long *retp)
{
int err = 0;
unsigned long long ret;
sector_t block = blocknr;
if (journal->j_bmap) {
err = journal->j_bmap(journal, &block);
if (err == 0)
*retp = block;
} else if (journal->j_inode) {
ret = bmap(journal->j_inode, &block);
if (ret || !block) {
printk(KERN_ALERT "%s: journal block not found "
"at offset %lu on %s\n",
__func__, blocknr, journal->j_devname);
err = -EIO;
jbd2_journal_abort(journal, err);
} else {
*retp = block;
}
} else {
*retp = blocknr; /* +journal->j_blk_offset */
}
return err;
}
/*
* We play buffer_head aliasing tricks to write data/metadata blocks to
* the journal without copying their contents, but for journal
* descriptor blocks we do need to generate bona fide buffers.
*
* After the caller of jbd2_journal_get_descriptor_buffer() has finished modifying
* the buffer's contents they really should run flush_dcache_page(bh->b_page).
* But we don't bother doing that, so there will be coherency problems with
* mmaps of blockdevs which hold live JBD-controlled filesystems.
*/
struct buffer_head *
jbd2_journal_get_descriptor_buffer(transaction_t *transaction, int type)
{
journal_t *journal = transaction->t_journal;
struct buffer_head *bh;
unsigned long long blocknr;
journal_header_t *header;
int err;
err = jbd2_journal_next_log_block(journal, &blocknr);
if (err)
return NULL;
bh = __getblk(journal->j_dev, blocknr, journal->j_blocksize);
if (!bh)
return NULL;
atomic_dec(&transaction->t_outstanding_credits);
lock_buffer(bh);
memset(bh->b_data, 0, journal->j_blocksize);
header = (journal_header_t *)bh->b_data;
header->h_magic = cpu_to_be32(JBD2_MAGIC_NUMBER);
header->h_blocktype = cpu_to_be32(type);
header->h_sequence = cpu_to_be32(transaction->t_tid);
set_buffer_uptodate(bh);
unlock_buffer(bh);
BUFFER_TRACE(bh, "return this buffer");
return bh;
}
void jbd2_descriptor_block_csum_set(journal_t *j, struct buffer_head *bh)
{
struct jbd2_journal_block_tail *tail;
__u32 csum;
if (!jbd2_journal_has_csum_v2or3(j))
return;
tail = (struct jbd2_journal_block_tail *)(bh->b_data + j->j_blocksize -
sizeof(struct jbd2_journal_block_tail));
tail->t_checksum = 0;
csum = jbd2_chksum(j, j->j_csum_seed, bh->b_data, j->j_blocksize);
tail->t_checksum = cpu_to_be32(csum);
}
/*
* Return tid of the oldest transaction in the journal and block in the journal
* where the transaction starts.
*
* If the journal is now empty, return which will be the next transaction ID
* we will write and where will that transaction start.
*
* The return value is 0 if journal tail cannot be pushed any further, 1 if
* it can.
*/
int jbd2_journal_get_log_tail(journal_t *journal, tid_t *tid,
unsigned long *block)
{
transaction_t *transaction;
int ret;
read_lock(&journal->j_state_lock);
spin_lock(&journal->j_list_lock);
transaction = journal->j_checkpoint_transactions;
if (transaction) {
*tid = transaction->t_tid;
*block = transaction->t_log_start;
} else if ((transaction = journal->j_committing_transaction) != NULL) {
*tid = transaction->t_tid;
*block = transaction->t_log_start;
} else if ((transaction = journal->j_running_transaction) != NULL) {
*tid = transaction->t_tid;
*block = journal->j_head;
} else {
*tid = journal->j_transaction_sequence;
*block = journal->j_head;
}
ret = tid_gt(*tid, journal->j_tail_sequence);
spin_unlock(&journal->j_list_lock);
read_unlock(&journal->j_state_lock);
return ret;
}
/*
* Update information in journal structure and in on disk journal superblock
* about log tail. This function does not check whether information passed in
* really pushes log tail further. It's responsibility of the caller to make
* sure provided log tail information is valid (e.g. by holding
* j_checkpoint_mutex all the time between computing log tail and calling this
* function as is the case with jbd2_cleanup_journal_tail()).
*
* Requires j_checkpoint_mutex
*/
int __jbd2_update_log_tail(journal_t *journal, tid_t tid, unsigned long block)
{
unsigned long freed;
int ret;
BUG_ON(!mutex_is_locked(&journal->j_checkpoint_mutex));
/*
* We cannot afford for write to remain in drive's caches since as
* soon as we update j_tail, next transaction can start reusing journal
* space and if we lose sb update during power failure we'd replay
* old transaction with possibly newly overwritten data.
*/
ret = jbd2_journal_update_sb_log_tail(journal, tid, block, REQ_FUA);
if (ret)
goto out;
write_lock(&journal->j_state_lock);
freed = block - journal->j_tail;
if (block < journal->j_tail)
freed += journal->j_last - journal->j_first;
trace_jbd2_update_log_tail(journal, tid, block, freed);
jbd2_debug(1,
"Cleaning journal tail from %u to %u (offset %lu), "
"freeing %lu\n",
journal->j_tail_sequence, tid, block, freed);
journal->j_free += freed;
journal->j_tail_sequence = tid;
journal->j_tail = block;
write_unlock(&journal->j_state_lock);
out:
return ret;
}
/*
* This is a variation of __jbd2_update_log_tail which checks for validity of
* provided log tail and locks j_checkpoint_mutex. So it is safe against races
* with other threads updating log tail.
*/
void jbd2_update_log_tail(journal_t *journal, tid_t tid, unsigned long block)
{
fs/jbd2, locking/mutex, sched/wait: Use mutex_lock_io() for journal->j_checkpoint_mutex When an ext4 fs is bogged down by a lot of metadata IOs (in the reported case, it was deletion of millions of files, but any massive amount of journal writes would do), after the journal is filled up, tasks which try to access the filesystem and aren't currently performing the journal writes end up waiting in __jbd2_log_wait_for_space() for journal->j_checkpoint_mutex. Because those mutex sleeps aren't marked as iowait, this condition can lead to misleadingly low iowait and /proc/stat:procs_blocked. While iowait propagation is far from strict, this condition can be triggered fairly easily and annotating these sleeps correctly helps initial diagnosis quite a bit. Use the new mutex_lock_io() for journal->j_checkpoint_mutex so that these sleeps are properly marked as iowait. Reported-by: Mingbo Wan <mingbo@fb.com> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Jan Kara <jack@suse.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Theodore Ts'o <tytso@mit.edu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: kernel-team@fb.com Link: http://lkml.kernel.org/r/1477673892-28940-5-git-send-email-tj@kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-10-28 16:58:12 +00:00
mutex_lock_io(&journal->j_checkpoint_mutex);
if (tid_gt(tid, journal->j_tail_sequence))
__jbd2_update_log_tail(journal, tid, block);
mutex_unlock(&journal->j_checkpoint_mutex);
}
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
struct jbd2_stats_proc_session {
journal_t *journal;
struct transaction_stats_s *stats;
int start;
int max;
};
static void *jbd2_seq_info_start(struct seq_file *seq, loff_t *pos)
{
return *pos ? NULL : SEQ_START_TOKEN;
}
static void *jbd2_seq_info_next(struct seq_file *seq, void *v, loff_t *pos)
{
(*pos)++;
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
return NULL;
}
static int jbd2_seq_info_show(struct seq_file *seq, void *v)
{
struct jbd2_stats_proc_session *s = seq->private;
if (v != SEQ_START_TOKEN)
return 0;
seq_printf(seq, "%lu transactions (%lu requested), "
"each up to %u blocks\n",
s->stats->ts_tid, s->stats->ts_requested,
s->journal->j_max_transaction_buffers);
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
if (s->stats->ts_tid == 0)
return 0;
seq_printf(seq, "average: \n %ums waiting for transaction\n",
jiffies_to_msecs(s->stats->run.rs_wait / s->stats->ts_tid));
seq_printf(seq, " %ums request delay\n",
(s->stats->ts_requested == 0) ? 0 :
jiffies_to_msecs(s->stats->run.rs_request_delay /
s->stats->ts_requested));
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
seq_printf(seq, " %ums running transaction\n",
jiffies_to_msecs(s->stats->run.rs_running / s->stats->ts_tid));
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
seq_printf(seq, " %ums transaction was being locked\n",
jiffies_to_msecs(s->stats->run.rs_locked / s->stats->ts_tid));
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
seq_printf(seq, " %ums flushing data (in ordered mode)\n",
jiffies_to_msecs(s->stats->run.rs_flushing / s->stats->ts_tid));
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
seq_printf(seq, " %ums logging transaction\n",
jiffies_to_msecs(s->stats->run.rs_logging / s->stats->ts_tid));
seq_printf(seq, " %lluus average transaction commit time\n",
div_u64(s->journal->j_average_commit_time, 1000));
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
seq_printf(seq, " %lu handles per transaction\n",
s->stats->run.rs_handle_count / s->stats->ts_tid);
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
seq_printf(seq, " %lu blocks per transaction\n",
s->stats->run.rs_blocks / s->stats->ts_tid);
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
seq_printf(seq, " %lu logged blocks per transaction\n",
s->stats->run.rs_blocks_logged / s->stats->ts_tid);
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
return 0;
}
static void jbd2_seq_info_stop(struct seq_file *seq, void *v)
{
}
static const struct seq_operations jbd2_seq_info_ops = {
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
.start = jbd2_seq_info_start,
.next = jbd2_seq_info_next,
.stop = jbd2_seq_info_stop,
.show = jbd2_seq_info_show,
};
static int jbd2_seq_info_open(struct inode *inode, struct file *file)
{
journal_t *journal = pde_data(inode);
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
struct jbd2_stats_proc_session *s;
int rc, size;
s = kmalloc(sizeof(*s), GFP_KERNEL);
if (s == NULL)
return -ENOMEM;
size = sizeof(struct transaction_stats_s);
s->stats = kmalloc(size, GFP_KERNEL);
if (s->stats == NULL) {
kfree(s);
return -ENOMEM;
}
spin_lock(&journal->j_history_lock);
memcpy(s->stats, &journal->j_stats, size);
s->journal = journal;
spin_unlock(&journal->j_history_lock);
rc = seq_open(file, &jbd2_seq_info_ops);
if (rc == 0) {
struct seq_file *m = file->private_data;
m->private = s;
} else {
kfree(s->stats);
kfree(s);
}
return rc;
}
static int jbd2_seq_info_release(struct inode *inode, struct file *file)
{
struct seq_file *seq = file->private_data;
struct jbd2_stats_proc_session *s = seq->private;
kfree(s->stats);
kfree(s);
return seq_release(inode, file);
}
static const struct proc_ops jbd2_info_proc_ops = {
.proc_open = jbd2_seq_info_open,
.proc_read = seq_read,
.proc_lseek = seq_lseek,
.proc_release = jbd2_seq_info_release,
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
};
static struct proc_dir_entry *proc_jbd2_stats;
static void jbd2_stats_proc_init(journal_t *journal)
{
journal->j_proc_entry = proc_mkdir(journal->j_devname, proc_jbd2_stats);
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
if (journal->j_proc_entry) {
proc_create_data("info", S_IRUGO, journal->j_proc_entry,
&jbd2_info_proc_ops, journal);
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
}
}
static void jbd2_stats_proc_exit(journal_t *journal)
{
remove_proc_entry("info", journal->j_proc_entry);
remove_proc_entry(journal->j_devname, proc_jbd2_stats);
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
}
/* Minimum size of descriptor tag */
static int jbd2_min_tag_size(void)
{
/*
* Tag with 32-bit block numbers does not use last four bytes of the
* structure
*/
return sizeof(journal_block_tag_t) - 4;
}
/**
* jbd2_journal_shrink_scan()
* @shrink: shrinker to work on
* @sc: reclaim request to process
*
* Scan the checkpointed buffer on the checkpoint list and release the
* journal_head.
*/
static unsigned long jbd2_journal_shrink_scan(struct shrinker *shrink,
struct shrink_control *sc)
{
jbd2,ext4: dynamically allocate the jbd2-journal shrinker In preparation for implementing lockless slab shrink, use new APIs to dynamically allocate the jbd2-journal shrinker, so that it can be freed asynchronously via RCU. Then it doesn't need to wait for RCU read-side critical section when releasing the struct journal_s. Link: https://lkml.kernel.org/r/20230911094444.68966-32-zhengqi.arch@bytedance.com Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Acked-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Abhinav Kumar <quic_abhinavk@quicinc.com> Cc: Alasdair Kergon <agk@redhat.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andreas Gruenbacher <agruenba@redhat.com> Cc: Anna Schumaker <anna@kernel.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Bob Peterson <rpeterso@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Carlos Llamas <cmllamas@google.com> Cc: Chandan Babu R <chandan.babu@oracle.com> Cc: Chao Yu <chao@kernel.org> Cc: Chris Mason <clm@fb.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Christian Koenig <christian.koenig@amd.com> Cc: Chuck Lever <cel@kernel.org> Cc: Coly Li <colyli@suse.de> Cc: Dai Ngo <Dai.Ngo@oracle.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: "Darrick J. Wong" <djwong@kernel.org> Cc: Dave Chinner <david@fromorbit.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David Airlie <airlied@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Sterba <dsterba@suse.com> Cc: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Cc: Gao Xiang <hsiangkao@linux.alibaba.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Huang Rui <ray.huang@amd.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jaegeuk Kim <jaegeuk@kernel.org> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Jeff Layton <jlayton@kernel.org> Cc: Jeffle Xu <jefflexu@linux.alibaba.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Josef Bacik <josef@toxicpanda.com> Cc: Juergen Gross <jgross@suse.com> Cc: Kent Overstreet <kent.overstreet@gmail.com> Cc: Kirill Tkhai <tkhai@ya.ru> Cc: Marijn Suijten <marijn.suijten@somainline.org> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Mike Snitzer <snitzer@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Nadav Amit <namit@vmware.com> Cc: Neil Brown <neilb@suse.de> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: Olga Kornievskaia <kolga@netapp.com> Cc: Paul E. McKenney <paulmck@kernel.org> Cc: Richard Weinberger <richard@nod.at> Cc: Rob Clark <robdclark@gmail.com> Cc: Rob Herring <robh@kernel.org> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: Sean Paul <sean@poorly.run> Cc: Sergey Senozhatsky <senozhatsky@chromium.org> Cc: Song Liu <song@kernel.org> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: Tom Talpey <tom@talpey.com> Cc: Trond Myklebust <trond.myklebust@hammerspace.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Cc: Yue Hu <huyue2@coolpad.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-11 09:44:30 +00:00
journal_t *journal = shrink->private_data;
unsigned long nr_to_scan = sc->nr_to_scan;
unsigned long nr_shrunk;
unsigned long count;
count = percpu_counter_read_positive(&journal->j_checkpoint_jh_count);
trace_jbd2_shrink_scan_enter(journal, sc->nr_to_scan, count);
nr_shrunk = jbd2_journal_shrink_checkpoint_list(journal, &nr_to_scan);
count = percpu_counter_read_positive(&journal->j_checkpoint_jh_count);
trace_jbd2_shrink_scan_exit(journal, nr_to_scan, nr_shrunk, count);
return nr_shrunk;
}
/**
* jbd2_journal_shrink_count()
* @shrink: shrinker to work on
* @sc: reclaim request to process
*
* Count the number of checkpoint buffers on the checkpoint list.
*/
static unsigned long jbd2_journal_shrink_count(struct shrinker *shrink,
struct shrink_control *sc)
{
jbd2,ext4: dynamically allocate the jbd2-journal shrinker In preparation for implementing lockless slab shrink, use new APIs to dynamically allocate the jbd2-journal shrinker, so that it can be freed asynchronously via RCU. Then it doesn't need to wait for RCU read-side critical section when releasing the struct journal_s. Link: https://lkml.kernel.org/r/20230911094444.68966-32-zhengqi.arch@bytedance.com Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Acked-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Abhinav Kumar <quic_abhinavk@quicinc.com> Cc: Alasdair Kergon <agk@redhat.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andreas Gruenbacher <agruenba@redhat.com> Cc: Anna Schumaker <anna@kernel.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Bob Peterson <rpeterso@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Carlos Llamas <cmllamas@google.com> Cc: Chandan Babu R <chandan.babu@oracle.com> Cc: Chao Yu <chao@kernel.org> Cc: Chris Mason <clm@fb.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Christian Koenig <christian.koenig@amd.com> Cc: Chuck Lever <cel@kernel.org> Cc: Coly Li <colyli@suse.de> Cc: Dai Ngo <Dai.Ngo@oracle.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: "Darrick J. Wong" <djwong@kernel.org> Cc: Dave Chinner <david@fromorbit.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David Airlie <airlied@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Sterba <dsterba@suse.com> Cc: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Cc: Gao Xiang <hsiangkao@linux.alibaba.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Huang Rui <ray.huang@amd.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jaegeuk Kim <jaegeuk@kernel.org> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Jeff Layton <jlayton@kernel.org> Cc: Jeffle Xu <jefflexu@linux.alibaba.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Josef Bacik <josef@toxicpanda.com> Cc: Juergen Gross <jgross@suse.com> Cc: Kent Overstreet <kent.overstreet@gmail.com> Cc: Kirill Tkhai <tkhai@ya.ru> Cc: Marijn Suijten <marijn.suijten@somainline.org> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Mike Snitzer <snitzer@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Nadav Amit <namit@vmware.com> Cc: Neil Brown <neilb@suse.de> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: Olga Kornievskaia <kolga@netapp.com> Cc: Paul E. McKenney <paulmck@kernel.org> Cc: Richard Weinberger <richard@nod.at> Cc: Rob Clark <robdclark@gmail.com> Cc: Rob Herring <robh@kernel.org> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: Sean Paul <sean@poorly.run> Cc: Sergey Senozhatsky <senozhatsky@chromium.org> Cc: Song Liu <song@kernel.org> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: Tom Talpey <tom@talpey.com> Cc: Trond Myklebust <trond.myklebust@hammerspace.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Cc: Yue Hu <huyue2@coolpad.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-11 09:44:30 +00:00
journal_t *journal = shrink->private_data;
unsigned long count;
count = percpu_counter_read_positive(&journal->j_checkpoint_jh_count);
trace_jbd2_shrink_count(journal, sc->nr_to_scan, count);
return count;
}
/*
* If the journal init or create aborts, we need to mark the journal
* superblock as being NULL to prevent the journal destroy from writing
* back a bogus superblock.
*/
static void journal_fail_superblock(journal_t *journal)
{
struct buffer_head *bh = journal->j_sb_buffer;
brelse(bh);
journal->j_sb_buffer = NULL;
}
/*
* Check the superblock for a given journal, performing initial
* validation of the format.
*/
static int journal_check_superblock(journal_t *journal)
{
journal_superblock_t *sb = journal->j_superblock;
int num_fc_blks;
int err = -EINVAL;
if (sb->s_header.h_magic != cpu_to_be32(JBD2_MAGIC_NUMBER) ||
sb->s_blocksize != cpu_to_be32(journal->j_blocksize)) {
printk(KERN_WARNING "JBD2: no valid journal superblock found\n");
return err;
}
if (be32_to_cpu(sb->s_header.h_blocktype) != JBD2_SUPERBLOCK_V1 &&
be32_to_cpu(sb->s_header.h_blocktype) != JBD2_SUPERBLOCK_V2) {
printk(KERN_WARNING "JBD2: unrecognised superblock format ID\n");
return err;
}
if (be32_to_cpu(sb->s_maxlen) > journal->j_total_len) {
printk(KERN_WARNING "JBD2: journal file too short\n");
return err;
}
if (be32_to_cpu(sb->s_first) == 0 ||
be32_to_cpu(sb->s_first) >= journal->j_total_len) {
printk(KERN_WARNING
"JBD2: Invalid start block of journal: %u\n",
be32_to_cpu(sb->s_first));
return err;
}
/*
* If this is a V2 superblock, then we have to check the
* features flags on it.
*/
if (!jbd2_format_support_feature(journal))
return 0;
if ((sb->s_feature_ro_compat &
~cpu_to_be32(JBD2_KNOWN_ROCOMPAT_FEATURES)) ||
(sb->s_feature_incompat &
~cpu_to_be32(JBD2_KNOWN_INCOMPAT_FEATURES))) {
printk(KERN_WARNING "JBD2: Unrecognised features on journal\n");
return err;
}
num_fc_blks = jbd2_has_feature_fast_commit(journal) ?
jbd2_journal_get_num_fc_blks(sb) : 0;
if (be32_to_cpu(sb->s_maxlen) < JBD2_MIN_JOURNAL_BLOCKS ||
be32_to_cpu(sb->s_maxlen) - JBD2_MIN_JOURNAL_BLOCKS < num_fc_blks) {
printk(KERN_ERR "JBD2: journal file too short %u,%d\n",
be32_to_cpu(sb->s_maxlen), num_fc_blks);
return err;
}
if (jbd2_has_feature_csum2(journal) &&
jbd2_has_feature_csum3(journal)) {
/* Can't have checksum v2 and v3 at the same time! */
printk(KERN_ERR "JBD2: Can't enable checksumming v2 and v3 "
"at the same time!\n");
return err;
}
if (jbd2_journal_has_csum_v2or3_feature(journal) &&
jbd2_has_feature_checksum(journal)) {
/* Can't have checksum v1 and v2 on at the same time! */
printk(KERN_ERR "JBD2: Can't enable checksumming v1 and v2/3 "
"at the same time!\n");
return err;
}
/* Load the checksum driver */
if (jbd2_journal_has_csum_v2or3_feature(journal)) {
if (sb->s_checksum_type != JBD2_CRC32C_CHKSUM) {
printk(KERN_ERR "JBD2: Unknown checksum type\n");
return err;
}
journal->j_chksum_driver = crypto_alloc_shash("crc32c", 0, 0);
if (IS_ERR(journal->j_chksum_driver)) {
printk(KERN_ERR "JBD2: Cannot load crc32c driver.\n");
err = PTR_ERR(journal->j_chksum_driver);
journal->j_chksum_driver = NULL;
return err;
}
/* Check superblock checksum */
if (sb->s_checksum != jbd2_superblock_csum(journal, sb)) {
printk(KERN_ERR "JBD2: journal checksum error\n");
err = -EFSBADCRC;
return err;
}
}
return 0;
}
static int journal_revoke_records_per_block(journal_t *journal)
{
int record_size;
int space = journal->j_blocksize - sizeof(jbd2_journal_revoke_header_t);
if (jbd2_has_feature_64bit(journal))
record_size = 8;
else
record_size = 4;
if (jbd2_journal_has_csum_v2or3(journal))
space -= sizeof(struct jbd2_journal_block_tail);
return space / record_size;
}
/*
* Load the on-disk journal superblock and read the key fields into the
* journal_t.
*/
static int journal_load_superblock(journal_t *journal)
{
int err;
struct buffer_head *bh;
journal_superblock_t *sb;
bh = getblk_unmovable(journal->j_dev, journal->j_blk_offset,
journal->j_blocksize);
if (bh)
err = bh_read(bh, 0);
if (!bh || err < 0) {
pr_err("%s: Cannot read journal superblock\n", __func__);
brelse(bh);
return -EIO;
}
journal->j_sb_buffer = bh;
sb = (journal_superblock_t *)bh->b_data;
journal->j_superblock = sb;
err = journal_check_superblock(journal);
if (err) {
journal_fail_superblock(journal);
return err;
}
journal->j_tail_sequence = be32_to_cpu(sb->s_sequence);
journal->j_tail = be32_to_cpu(sb->s_start);
journal->j_first = be32_to_cpu(sb->s_first);
journal->j_errno = be32_to_cpu(sb->s_errno);
journal->j_last = be32_to_cpu(sb->s_maxlen);
if (be32_to_cpu(sb->s_maxlen) < journal->j_total_len)
journal->j_total_len = be32_to_cpu(sb->s_maxlen);
/* Precompute checksum seed for all metadata */
if (jbd2_journal_has_csum_v2or3(journal))
journal->j_csum_seed = jbd2_chksum(journal, ~0, sb->s_uuid,
sizeof(sb->s_uuid));
journal->j_revoke_records_per_block =
journal_revoke_records_per_block(journal);
if (jbd2_has_feature_fast_commit(journal)) {
journal->j_fc_last = be32_to_cpu(sb->s_maxlen);
journal->j_last = journal->j_fc_last -
jbd2_journal_get_num_fc_blks(sb);
journal->j_fc_first = journal->j_last + 1;
journal->j_fc_off = 0;
}
return 0;
}
/*
* Management for journal control blocks: functions to create and
* destroy journal_t structures, and to initialise and read existing
* journal blocks from disk. */
/* First: create and setup a journal_t object in memory. We initialise
* very few fields yet: that has to wait until we have created the
* journal structures from from scratch, or loaded them from disk. */
static journal_t *journal_init_common(struct block_device *bdev,
struct block_device *fs_dev,
unsigned long long start, int len, int blocksize)
{
static struct lock_class_key jbd2_trans_commit_key;
journal_t *journal;
int err;
int n;
journal = kzalloc(sizeof(*journal), GFP_KERNEL);
if (!journal)
return ERR_PTR(-ENOMEM);
journal->j_blocksize = blocksize;
journal->j_dev = bdev;
journal->j_fs_dev = fs_dev;
journal->j_blk_offset = start;
journal->j_total_len = len;
jbd2_init_fs_dev_write_error(journal);
err = journal_load_superblock(journal);
if (err)
goto err_cleanup;
init_waitqueue_head(&journal->j_wait_transaction_locked);
init_waitqueue_head(&journal->j_wait_done_commit);
init_waitqueue_head(&journal->j_wait_commit);
init_waitqueue_head(&journal->j_wait_updates);
init_waitqueue_head(&journal->j_wait_reserved);
init_waitqueue_head(&journal->j_fc_wait);
mutex_init(&journal->j_abort_mutex);
mutex_init(&journal->j_barrier);
mutex_init(&journal->j_checkpoint_mutex);
spin_lock_init(&journal->j_revoke_lock);
spin_lock_init(&journal->j_list_lock);
spin_lock_init(&journal->j_history_lock);
rwlock_init(&journal->j_state_lock);
journal->j_commit_interval = (HZ * JBD2_DEFAULT_MAX_COMMIT_AGE);
journal->j_min_batch_time = 0;
journal->j_max_batch_time = 15000; /* 15ms */
atomic_set(&journal->j_reserved_credits, 0);
lockdep_init_map(&journal->j_trans_commit_map, "jbd2_handle",
&jbd2_trans_commit_key, 0);
/* The journal is marked for error until we succeed with recovery! */
journal->j_flags = JBD2_ABORT;
/* Set up a default-sized revoke table for the new mount. */
err = jbd2_journal_init_revoke(journal, JOURNAL_REVOKE_DEFAULT_HASH);
if (err)
goto err_cleanup;
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
/*
* journal descriptor can store up to n blocks, we need enough
* buffers to write out full descriptor block.
*/
err = -ENOMEM;
n = journal->j_blocksize / jbd2_min_tag_size();
journal->j_wbufsize = n;
journal->j_fc_wbuf = NULL;
journal->j_wbuf = kmalloc_array(n, sizeof(struct buffer_head *),
GFP_KERNEL);
if (!journal->j_wbuf)
goto err_cleanup;
err = percpu_counter_init(&journal->j_checkpoint_jh_count, 0,
GFP_KERNEL);
if (err)
goto err_cleanup;
journal->j_shrink_transaction = NULL;
jbd2,ext4: dynamically allocate the jbd2-journal shrinker In preparation for implementing lockless slab shrink, use new APIs to dynamically allocate the jbd2-journal shrinker, so that it can be freed asynchronously via RCU. Then it doesn't need to wait for RCU read-side critical section when releasing the struct journal_s. Link: https://lkml.kernel.org/r/20230911094444.68966-32-zhengqi.arch@bytedance.com Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Acked-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Abhinav Kumar <quic_abhinavk@quicinc.com> Cc: Alasdair Kergon <agk@redhat.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andreas Gruenbacher <agruenba@redhat.com> Cc: Anna Schumaker <anna@kernel.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Bob Peterson <rpeterso@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Carlos Llamas <cmllamas@google.com> Cc: Chandan Babu R <chandan.babu@oracle.com> Cc: Chao Yu <chao@kernel.org> Cc: Chris Mason <clm@fb.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Christian Koenig <christian.koenig@amd.com> Cc: Chuck Lever <cel@kernel.org> Cc: Coly Li <colyli@suse.de> Cc: Dai Ngo <Dai.Ngo@oracle.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: "Darrick J. Wong" <djwong@kernel.org> Cc: Dave Chinner <david@fromorbit.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David Airlie <airlied@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Sterba <dsterba@suse.com> Cc: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Cc: Gao Xiang <hsiangkao@linux.alibaba.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Huang Rui <ray.huang@amd.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jaegeuk Kim <jaegeuk@kernel.org> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Jeff Layton <jlayton@kernel.org> Cc: Jeffle Xu <jefflexu@linux.alibaba.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Josef Bacik <josef@toxicpanda.com> Cc: Juergen Gross <jgross@suse.com> Cc: Kent Overstreet <kent.overstreet@gmail.com> Cc: Kirill Tkhai <tkhai@ya.ru> Cc: Marijn Suijten <marijn.suijten@somainline.org> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Mike Snitzer <snitzer@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Nadav Amit <namit@vmware.com> Cc: Neil Brown <neilb@suse.de> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: Olga Kornievskaia <kolga@netapp.com> Cc: Paul E. McKenney <paulmck@kernel.org> Cc: Richard Weinberger <richard@nod.at> Cc: Rob Clark <robdclark@gmail.com> Cc: Rob Herring <robh@kernel.org> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: Sean Paul <sean@poorly.run> Cc: Sergey Senozhatsky <senozhatsky@chromium.org> Cc: Song Liu <song@kernel.org> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: Tom Talpey <tom@talpey.com> Cc: Trond Myklebust <trond.myklebust@hammerspace.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Cc: Yue Hu <huyue2@coolpad.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-11 09:44:30 +00:00
journal->j_shrinker = shrinker_alloc(0, "jbd2-journal:(%u:%u)",
MAJOR(bdev->bd_dev),
MINOR(bdev->bd_dev));
if (!journal->j_shrinker) {
err = -ENOMEM;
goto err_cleanup;
jbd2,ext4: dynamically allocate the jbd2-journal shrinker In preparation for implementing lockless slab shrink, use new APIs to dynamically allocate the jbd2-journal shrinker, so that it can be freed asynchronously via RCU. Then it doesn't need to wait for RCU read-side critical section when releasing the struct journal_s. Link: https://lkml.kernel.org/r/20230911094444.68966-32-zhengqi.arch@bytedance.com Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Acked-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Abhinav Kumar <quic_abhinavk@quicinc.com> Cc: Alasdair Kergon <agk@redhat.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andreas Gruenbacher <agruenba@redhat.com> Cc: Anna Schumaker <anna@kernel.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Bob Peterson <rpeterso@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Carlos Llamas <cmllamas@google.com> Cc: Chandan Babu R <chandan.babu@oracle.com> Cc: Chao Yu <chao@kernel.org> Cc: Chris Mason <clm@fb.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Christian Koenig <christian.koenig@amd.com> Cc: Chuck Lever <cel@kernel.org> Cc: Coly Li <colyli@suse.de> Cc: Dai Ngo <Dai.Ngo@oracle.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: "Darrick J. Wong" <djwong@kernel.org> Cc: Dave Chinner <david@fromorbit.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David Airlie <airlied@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Sterba <dsterba@suse.com> Cc: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Cc: Gao Xiang <hsiangkao@linux.alibaba.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Huang Rui <ray.huang@amd.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jaegeuk Kim <jaegeuk@kernel.org> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Jeff Layton <jlayton@kernel.org> Cc: Jeffle Xu <jefflexu@linux.alibaba.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Josef Bacik <josef@toxicpanda.com> Cc: Juergen Gross <jgross@suse.com> Cc: Kent Overstreet <kent.overstreet@gmail.com> Cc: Kirill Tkhai <tkhai@ya.ru> Cc: Marijn Suijten <marijn.suijten@somainline.org> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Mike Snitzer <snitzer@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Nadav Amit <namit@vmware.com> Cc: Neil Brown <neilb@suse.de> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: Olga Kornievskaia <kolga@netapp.com> Cc: Paul E. McKenney <paulmck@kernel.org> Cc: Richard Weinberger <richard@nod.at> Cc: Rob Clark <robdclark@gmail.com> Cc: Rob Herring <robh@kernel.org> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: Sean Paul <sean@poorly.run> Cc: Sergey Senozhatsky <senozhatsky@chromium.org> Cc: Song Liu <song@kernel.org> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: Tom Talpey <tom@talpey.com> Cc: Trond Myklebust <trond.myklebust@hammerspace.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Cc: Yue Hu <huyue2@coolpad.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-11 09:44:30 +00:00
}
journal->j_shrinker->scan_objects = jbd2_journal_shrink_scan;
journal->j_shrinker->count_objects = jbd2_journal_shrink_count;
journal->j_shrinker->batch = journal->j_max_transaction_buffers;
journal->j_shrinker->private_data = journal;
shrinker_register(journal->j_shrinker);
return journal;
err_cleanup:
percpu_counter_destroy(&journal->j_checkpoint_jh_count);
jbd2: Fix memory leak in journal_init_common() There is a memory leak reported by kmemleak: unreferenced object 0xff11000105903b80 (size 64): comm "mount", pid 3382, jiffies 4295032021 (age 27.826s) hex dump (first 32 bytes): 04 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00 ................ ff ff ff ff 00 00 00 00 00 00 00 00 00 00 00 00 ................ backtrace: [<ffffffffae86ac40>] __kmalloc_node+0x50/0x160 [<ffffffffaf2486d8>] crypto_alloc_tfmmem.isra.0+0x38/0x110 [<ffffffffaf2498e5>] crypto_create_tfm_node+0x85/0x2f0 [<ffffffffaf24a92c>] crypto_alloc_tfm_node+0xfc/0x210 [<ffffffffaedde777>] journal_init_common+0x727/0x1ad0 [<ffffffffaede1715>] jbd2_journal_init_inode+0x2b5/0x500 [<ffffffffaed786b5>] ext4_load_and_init_journal+0x255/0x2440 [<ffffffffaed8b423>] ext4_fill_super+0x8823/0xa330 ... The root cause was traced to an error handing path in journal_init_common() when malloc memory failed in register_shrinker(). The checksum driver is used to reference to checksum algorithm via cryptoapi and the user should release the memory when the driver is no longer needed or the journal initialization failed. Fix it by calling crypto_free_shash() on the "err_cleanup" error handing path in journal_init_common(). Fixes: c30713084ba5 ("jbd2: move load_superblock() into journal_init_common()") Signed-off-by: Li Zetao <lizetao1@huawei.com> Reviewed-by: Jan Kara <jack@suse.cz> Reviewed-by: Zhang Yi <yi.zhang@huawei.com> Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com> Link: https://lore.kernel.org/r/20230911025138.983101-1-lizetao1@huawei.com Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2023-09-11 02:51:38 +00:00
if (journal->j_chksum_driver)
crypto_free_shash(journal->j_chksum_driver);
kfree(journal->j_wbuf);
jbd2_journal_destroy_revoke(journal);
journal_fail_superblock(journal);
kfree(journal);
return ERR_PTR(err);
}
/* jbd2_journal_init_dev and jbd2_journal_init_inode:
*
* Create a journal structure assigned some fixed set of disk blocks to
* the journal. We don't actually touch those disk blocks yet, but we
* need to set up all of the mapping information to tell the journaling
* system where the journal blocks are.
*
*/
/**
* journal_t * jbd2_journal_init_dev() - creates and initialises a journal structure
* @bdev: Block device on which to create the journal
* @fs_dev: Device which hold journalled filesystem for this journal.
* @start: Block nr Start of journal.
* @len: Length of the journal in blocks.
* @blocksize: blocksize of journalling device
*
* Returns: a newly created journal_t *
*
* jbd2_journal_init_dev creates a journal which maps a fixed contiguous
* range of blocks on an arbitrary block device.
*
*/
journal_t *jbd2_journal_init_dev(struct block_device *bdev,
struct block_device *fs_dev,
unsigned long long start, int len, int blocksize)
{
journal_t *journal;
journal = journal_init_common(bdev, fs_dev, start, len, blocksize);
if (IS_ERR(journal))
return ERR_CAST(journal);
snprintf(journal->j_devname, sizeof(journal->j_devname),
"%pg", journal->j_dev);
strreplace(journal->j_devname, '/', '!');
jbd2_stats_proc_init(journal);
return journal;
}
/**
* journal_t * jbd2_journal_init_inode () - creates a journal which maps to a inode.
* @inode: An inode to create the journal in
*
* jbd2_journal_init_inode creates a journal which maps an on-disk inode as
* the journal. The inode must exist already, must support bmap() and
* must have all data blocks preallocated.
*/
journal_t *jbd2_journal_init_inode(struct inode *inode)
{
journal_t *journal;
sector_t blocknr;
int err = 0;
blocknr = 0;
err = bmap(inode, &blocknr);
if (err || !blocknr) {
pr_err("%s: Cannot locate journal superblock\n", __func__);
return err ? ERR_PTR(err) : ERR_PTR(-EINVAL);
}
jbd2_debug(1, "JBD2: inode %s/%ld, size %lld, bits %d, blksize %ld\n",
inode->i_sb->s_id, inode->i_ino, (long long) inode->i_size,
inode->i_sb->s_blocksize_bits, inode->i_sb->s_blocksize);
journal = journal_init_common(inode->i_sb->s_bdev, inode->i_sb->s_bdev,
blocknr, inode->i_size >> inode->i_sb->s_blocksize_bits,
inode->i_sb->s_blocksize);
if (IS_ERR(journal))
return ERR_CAST(journal);
journal->j_inode = inode;
snprintf(journal->j_devname, sizeof(journal->j_devname),
"%pg-%lu", journal->j_dev, journal->j_inode->i_ino);
strreplace(journal->j_devname, '/', '!');
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
jbd2_stats_proc_init(journal);
return journal;
}
/*
* Given a journal_t structure, initialise the various fields for
* startup of a new journaling session. We use this both when creating
* a journal, and after recovering an old journal to reset it for
* subsequent use.
*/
static int journal_reset(journal_t *journal)
{
journal_superblock_t *sb = journal->j_superblock;
unsigned long long first, last;
first = be32_to_cpu(sb->s_first);
last = be32_to_cpu(sb->s_maxlen);
if (first + JBD2_MIN_JOURNAL_BLOCKS > last + 1) {
printk(KERN_ERR "JBD2: Journal too short (blocks %llu-%llu).\n",
first, last);
journal_fail_superblock(journal);
return -EINVAL;
}
journal->j_first = first;
journal->j_last = last;
jbd2: continue to record log between each mount For a newly mounted file system, the journal committing thread always record new transactions from the start of the journal area, no matter whether the journal was clean or just has been recovered. So the logdump code in debugfs cannot dump continuous logs between each mount, it is disadvantageous to analysis corrupted file system image and locate the file system inconsistency bugs. If we get a corrupted file system in the running products and want to find out what has happened, besides lookup the system log, one effective way is to backtrack the journal log. But we may not always run e2fsck before each mount and the default fsck -a mode also cannot always checkout all inconsistencies, so it could left over some inconsistencies into the next mount until we detect it. Finally, transactions in the journal may probably discontinuous and some relatively new transactions has been covered, it becomes hard to analyse. If we could record transactions continuously between each mount, we could acquire more useful info from the journal. Like this: |Previous mount checkpointed/recovered logs|Current mount logs | |{------}{---}{--------} ... {------}| ... |{======}{========}...000000| And yes the journal area is limited and cannot record everything, the problematic transaction may also be covered even if we do this, but this is still useful for fuzzy tests and short-running products. This patch save the head blocknr in the superblock after flushing the journal or unmounting the file system, let the next mount could continue to record new transaction behind it. This change is backward compatible because the old kernel does not care about the head blocknr of the journal. It is also fine if we mount a clean old image without valid head blocknr, we fail back to set it to s_first just like before. Finally, for the case of mount an unclean file system, we could also get the journal head easily after scanning/replaying the journal, it will continue to record new transaction after the recovered transactions. Signed-off-by: Zhang Yi <yi.zhang@huawei.com> Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20230322013353.1843306-2-yi.zhang@huaweicloud.com Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2023-03-22 01:33:51 +00:00
if (journal->j_head != 0 && journal->j_flags & JBD2_CYCLE_RECORD) {
/*
* Disable the cycled recording mode if the journal head block
* number is not correct.
*/
if (journal->j_head < first || journal->j_head >= last) {
printk(KERN_WARNING "JBD2: Incorrect Journal head block %lu, "
"disable journal_cycle_record\n",
journal->j_head);
journal->j_head = journal->j_first;
}
} else {
journal->j_head = journal->j_first;
}
journal->j_tail = journal->j_head;
journal->j_free = journal->j_last - journal->j_first;
journal->j_tail_sequence = journal->j_transaction_sequence;
journal->j_commit_sequence = journal->j_transaction_sequence - 1;
journal->j_commit_request = journal->j_commit_sequence;
journal->j_max_transaction_buffers = jbd2_journal_get_max_txn_bufs(journal);
/*
* Now that journal recovery is done, turn fast commits off here. This
* way, if fast commit was enabled before the crash but if now FS has
* disabled it, we don't enable fast commits.
*/
jbd2_clear_feature_fast_commit(journal);
/*
* As a special case, if the on-disk copy is already marked as needing
* no recovery (s_start == 0), then we can safely defer the superblock
* update until the next commit by setting JBD2_FLUSHED. This avoids
* attempting a write to a potential-readonly device.
*/
if (sb->s_start == 0) {
jbd2_debug(1, "JBD2: Skipping superblock update on recovered sb "
"(start %ld, seq %u, errno %d)\n",
journal->j_tail, journal->j_tail_sequence,
journal->j_errno);
journal->j_flags |= JBD2_FLUSHED;
} else {
/* Lock here to make assertions happy... */
fs/jbd2, locking/mutex, sched/wait: Use mutex_lock_io() for journal->j_checkpoint_mutex When an ext4 fs is bogged down by a lot of metadata IOs (in the reported case, it was deletion of millions of files, but any massive amount of journal writes would do), after the journal is filled up, tasks which try to access the filesystem and aren't currently performing the journal writes end up waiting in __jbd2_log_wait_for_space() for journal->j_checkpoint_mutex. Because those mutex sleeps aren't marked as iowait, this condition can lead to misleadingly low iowait and /proc/stat:procs_blocked. While iowait propagation is far from strict, this condition can be triggered fairly easily and annotating these sleeps correctly helps initial diagnosis quite a bit. Use the new mutex_lock_io() for journal->j_checkpoint_mutex so that these sleeps are properly marked as iowait. Reported-by: Mingbo Wan <mingbo@fb.com> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Jan Kara <jack@suse.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Theodore Ts'o <tytso@mit.edu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: kernel-team@fb.com Link: http://lkml.kernel.org/r/1477673892-28940-5-git-send-email-tj@kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-10-28 16:58:12 +00:00
mutex_lock_io(&journal->j_checkpoint_mutex);
/*
* Update log tail information. We use REQ_FUA since new
* transaction will start reusing journal space and so we
* must make sure information about current log tail is on
* disk before that.
*/
jbd2_journal_update_sb_log_tail(journal,
journal->j_tail_sequence,
journal->j_tail, REQ_FUA);
mutex_unlock(&journal->j_checkpoint_mutex);
}
return jbd2_journal_start_thread(journal);
}
/*
* This function expects that the caller will have locked the journal
* buffer head, and will return with it unlocked
*/
static int jbd2_write_superblock(journal_t *journal, blk_opf_t write_flags)
{
struct buffer_head *bh = journal->j_sb_buffer;
journal_superblock_t *sb = journal->j_superblock;
int ret = 0;
/* Buffer got discarded which means block device got invalidated */
if (!buffer_mapped(bh)) {
unlock_buffer(bh);
return -EIO;
}
/*
* Always set high priority flags to exempt from block layer's
* QOS policies, e.g. writeback throttle.
*/
write_flags |= JBD2_JOURNAL_REQ_FLAGS;
if (!(journal->j_flags & JBD2_BARRIER))
write_flags &= ~(REQ_FUA | REQ_PREFLUSH);
trace_jbd2_write_superblock(journal, write_flags);
if (buffer_write_io_error(bh)) {
/*
* Oh, dear. A previous attempt to write the journal
* superblock failed. This could happen because the
* USB device was yanked out. Or it could happen to
* be a transient write error and maybe the block will
* be remapped. Nothing we can do but to retry the
* write and hope for the best.
*/
printk(KERN_ERR "JBD2: previous I/O error detected "
"for journal superblock update for %s.\n",
journal->j_devname);
clear_buffer_write_io_error(bh);
set_buffer_uptodate(bh);
}
if (jbd2_journal_has_csum_v2or3(journal))
sb->s_checksum = jbd2_superblock_csum(journal, sb);
get_bh(bh);
bh->b_end_io = end_buffer_write_sync;
submit_bh(REQ_OP_WRITE | write_flags, bh);
wait_on_buffer(bh);
if (buffer_write_io_error(bh)) {
clear_buffer_write_io_error(bh);
set_buffer_uptodate(bh);
ret = -EIO;
}
if (ret) {
printk(KERN_ERR "JBD2: I/O error when updating journal superblock for %s.\n",
journal->j_devname);
if (!is_journal_aborted(journal))
jbd2_journal_abort(journal, ret);
}
return ret;
}
/**
* jbd2_journal_update_sb_log_tail() - Update log tail in journal sb on disk.
* @journal: The journal to update.
* @tail_tid: TID of the new transaction at the tail of the log
* @tail_block: The first block of the transaction at the tail of the log
* @write_flags: Flags for the journal sb write operation
*
* Update a journal's superblock information about log tail and write it to
* disk, waiting for the IO to complete.
*/
int jbd2_journal_update_sb_log_tail(journal_t *journal, tid_t tail_tid,
unsigned long tail_block,
blk_opf_t write_flags)
{
journal_superblock_t *sb = journal->j_superblock;
int ret;
if (is_journal_aborted(journal))
return -EIO;
if (jbd2_check_fs_dev_write_error(journal)) {
jbd2_journal_abort(journal, -EIO);
return -EIO;
}
BUG_ON(!mutex_is_locked(&journal->j_checkpoint_mutex));
jbd2_debug(1, "JBD2: updating superblock (start %lu, seq %u)\n",
tail_block, tail_tid);
lock_buffer(journal->j_sb_buffer);
sb->s_sequence = cpu_to_be32(tail_tid);
sb->s_start = cpu_to_be32(tail_block);
ret = jbd2_write_superblock(journal, write_flags);
if (ret)
goto out;
/* Log is no longer empty */
write_lock(&journal->j_state_lock);
WARN_ON(!sb->s_sequence);
journal->j_flags &= ~JBD2_FLUSHED;
write_unlock(&journal->j_state_lock);
out:
return ret;
}
/**
* jbd2_mark_journal_empty() - Mark on disk journal as empty.
* @journal: The journal to update.
* @write_flags: Flags for the journal sb write operation
*
* Update a journal's dynamic superblock fields to show that journal is empty.
* Write updated superblock to disk waiting for IO to complete.
*/
static void jbd2_mark_journal_empty(journal_t *journal, blk_opf_t write_flags)
{
journal_superblock_t *sb = journal->j_superblock;
bool had_fast_commit = false;
BUG_ON(!mutex_is_locked(&journal->j_checkpoint_mutex));
lock_buffer(journal->j_sb_buffer);
if (sb->s_start == 0) { /* Is it already empty? */
unlock_buffer(journal->j_sb_buffer);
return;
}
jbd2_debug(1, "JBD2: Marking journal as empty (seq %u)\n",
journal->j_tail_sequence);
sb->s_sequence = cpu_to_be32(journal->j_tail_sequence);
sb->s_start = cpu_to_be32(0);
jbd2: continue to record log between each mount For a newly mounted file system, the journal committing thread always record new transactions from the start of the journal area, no matter whether the journal was clean or just has been recovered. So the logdump code in debugfs cannot dump continuous logs between each mount, it is disadvantageous to analysis corrupted file system image and locate the file system inconsistency bugs. If we get a corrupted file system in the running products and want to find out what has happened, besides lookup the system log, one effective way is to backtrack the journal log. But we may not always run e2fsck before each mount and the default fsck -a mode also cannot always checkout all inconsistencies, so it could left over some inconsistencies into the next mount until we detect it. Finally, transactions in the journal may probably discontinuous and some relatively new transactions has been covered, it becomes hard to analyse. If we could record transactions continuously between each mount, we could acquire more useful info from the journal. Like this: |Previous mount checkpointed/recovered logs|Current mount logs | |{------}{---}{--------} ... {------}| ... |{======}{========}...000000| And yes the journal area is limited and cannot record everything, the problematic transaction may also be covered even if we do this, but this is still useful for fuzzy tests and short-running products. This patch save the head blocknr in the superblock after flushing the journal or unmounting the file system, let the next mount could continue to record new transaction behind it. This change is backward compatible because the old kernel does not care about the head blocknr of the journal. It is also fine if we mount a clean old image without valid head blocknr, we fail back to set it to s_first just like before. Finally, for the case of mount an unclean file system, we could also get the journal head easily after scanning/replaying the journal, it will continue to record new transaction after the recovered transactions. Signed-off-by: Zhang Yi <yi.zhang@huawei.com> Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20230322013353.1843306-2-yi.zhang@huaweicloud.com Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2023-03-22 01:33:51 +00:00
sb->s_head = cpu_to_be32(journal->j_head);
if (jbd2_has_feature_fast_commit(journal)) {
/*
* When journal is clean, no need to commit fast commit flag and
* make file system incompatible with older kernels.
*/
jbd2_clear_feature_fast_commit(journal);
had_fast_commit = true;
}
jbd2_write_superblock(journal, write_flags);
if (had_fast_commit)
jbd2_set_feature_fast_commit(journal);
/* Log is no longer empty */
write_lock(&journal->j_state_lock);
journal->j_flags |= JBD2_FLUSHED;
write_unlock(&journal->j_state_lock);
}
/**
* __jbd2_journal_erase() - Discard or zeroout journal blocks (excluding superblock)
* @journal: The journal to erase.
* @flags: A discard/zeroout request is sent for each physically contigous
* region of the journal. Either JBD2_JOURNAL_FLUSH_DISCARD or
* JBD2_JOURNAL_FLUSH_ZEROOUT must be set to determine which operation
* to perform.
*
* Note: JBD2_JOURNAL_FLUSH_ZEROOUT attempts to use hardware offload. Zeroes
* will be explicitly written if no hardware offload is available, see
* blkdev_issue_zeroout for more details.
*/
static int __jbd2_journal_erase(journal_t *journal, unsigned int flags)
{
int err = 0;
unsigned long block, log_offset; /* logical */
unsigned long long phys_block, block_start, block_stop; /* physical */
loff_t byte_start, byte_stop, byte_count;
/* flags must be set to either discard or zeroout */
if ((flags & ~JBD2_JOURNAL_FLUSH_VALID) || !flags ||
((flags & JBD2_JOURNAL_FLUSH_DISCARD) &&
(flags & JBD2_JOURNAL_FLUSH_ZEROOUT)))
return -EINVAL;
if ((flags & JBD2_JOURNAL_FLUSH_DISCARD) &&
!bdev_max_discard_sectors(journal->j_dev))
return -EOPNOTSUPP;
/*
* lookup block mapping and issue discard/zeroout for each
* contiguous region
*/
log_offset = be32_to_cpu(journal->j_superblock->s_first);
block_start = ~0ULL;
for (block = log_offset; block < journal->j_total_len; block++) {
err = jbd2_journal_bmap(journal, block, &phys_block);
if (err) {
pr_err("JBD2: bad block at offset %lu", block);
return err;
}
if (block_start == ~0ULL) {
block_start = phys_block;
block_stop = block_start - 1;
}
/*
* last block not contiguous with current block,
* process last contiguous region and return to this block on
* next loop
*/
if (phys_block != block_stop + 1) {
block--;
} else {
block_stop++;
/*
* if this isn't the last block of journal,
* no need to process now because next block may also
* be part of this contiguous region
*/
if (block != journal->j_total_len - 1)
continue;
}
/*
* end of contiguous region or this is last block of journal,
* take care of the region
*/
byte_start = block_start * journal->j_blocksize;
byte_stop = block_stop * journal->j_blocksize;
byte_count = (block_stop - block_start + 1) *
journal->j_blocksize;
truncate_inode_pages_range(journal->j_dev->bd_inode->i_mapping,
byte_start, byte_stop);
if (flags & JBD2_JOURNAL_FLUSH_DISCARD) {
err = blkdev_issue_discard(journal->j_dev,
byte_start >> SECTOR_SHIFT,
byte_count >> SECTOR_SHIFT,
GFP_NOFS);
} else if (flags & JBD2_JOURNAL_FLUSH_ZEROOUT) {
err = blkdev_issue_zeroout(journal->j_dev,
byte_start >> SECTOR_SHIFT,
byte_count >> SECTOR_SHIFT,
GFP_NOFS, 0);
}
if (unlikely(err != 0)) {
pr_err("JBD2: (error %d) unable to wipe journal at physical blocks %llu - %llu",
err, block_start, block_stop);
return err;
}
/* reset start and stop after processing a region */
block_start = ~0ULL;
}
return blkdev_issue_flush(journal->j_dev);
}
/**
* jbd2_journal_update_sb_errno() - Update error in the journal.
* @journal: The journal to update.
*
* Update a journal's errno. Write updated superblock to disk waiting for IO
* to complete.
*/
void jbd2_journal_update_sb_errno(journal_t *journal)
{
journal_superblock_t *sb = journal->j_superblock;
int errcode;
lock_buffer(journal->j_sb_buffer);
errcode = journal->j_errno;
if (errcode == -ESHUTDOWN)
errcode = 0;
jbd2_debug(1, "JBD2: updating superblock error (errno %d)\n", errcode);
sb->s_errno = cpu_to_be32(errcode);
jbd2_write_superblock(journal, REQ_FUA);
}
EXPORT_SYMBOL(jbd2_journal_update_sb_errno);
/**
* jbd2_journal_load() - Read journal from disk.
* @journal: Journal to act on.
*
* Given a journal_t structure which tells us which disk blocks contain
* a journal, read the journal from disk to initialise the in-memory
* structures.
*/
int jbd2_journal_load(journal_t *journal)
{
int err;
journal_superblock_t *sb = journal->j_superblock;
/*
* Create a slab for this blocksize
*/
err = jbd2_journal_create_slab(be32_to_cpu(sb->s_blocksize));
if (err)
return err;
/* Let the recovery code check whether it needs to recover any
* data from the journal. */
err = jbd2_journal_recover(journal);
if (err) {
pr_warn("JBD2: journal recovery failed\n");
return err;
}
if (journal->j_failed_commit) {
printk(KERN_ERR "JBD2: journal transaction %u on %s "
"is corrupt.\n", journal->j_failed_commit,
journal->j_devname);
return -EFSCORRUPTED;
}
/*
* clear JBD2_ABORT flag initialized in journal_init_common
* here to update log tail information with the newest seq.
*/
journal->j_flags &= ~JBD2_ABORT;
/* OK, we've finished with the dynamic journal bits:
* reinitialise the dynamic contents of the superblock in memory
* and reset them on disk. */
err = journal_reset(journal);
if (err) {
pr_warn("JBD2: journal reset failed\n");
return err;
}
journal->j_flags |= JBD2_LOADED;
return 0;
}
/**
* jbd2_journal_destroy() - Release a journal_t structure.
* @journal: Journal to act on.
*
* Release a journal_t structure once it is no longer in use by the
* journaled object.
* Return <0 if we couldn't clean up the journal.
*/
int jbd2_journal_destroy(journal_t *journal)
{
int err = 0;
/* Wait for the commit thread to wake up and die. */
journal_kill_thread(journal);
/* Force a final log commit */
if (journal->j_running_transaction)
jbd2_journal_commit_transaction(journal);
/* Force any old transactions to disk */
/* Totally anal locking here... */
spin_lock(&journal->j_list_lock);
while (journal->j_checkpoint_transactions != NULL) {
spin_unlock(&journal->j_list_lock);
fs/jbd2, locking/mutex, sched/wait: Use mutex_lock_io() for journal->j_checkpoint_mutex When an ext4 fs is bogged down by a lot of metadata IOs (in the reported case, it was deletion of millions of files, but any massive amount of journal writes would do), after the journal is filled up, tasks which try to access the filesystem and aren't currently performing the journal writes end up waiting in __jbd2_log_wait_for_space() for journal->j_checkpoint_mutex. Because those mutex sleeps aren't marked as iowait, this condition can lead to misleadingly low iowait and /proc/stat:procs_blocked. While iowait propagation is far from strict, this condition can be triggered fairly easily and annotating these sleeps correctly helps initial diagnosis quite a bit. Use the new mutex_lock_io() for journal->j_checkpoint_mutex so that these sleeps are properly marked as iowait. Reported-by: Mingbo Wan <mingbo@fb.com> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Jan Kara <jack@suse.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Theodore Ts'o <tytso@mit.edu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: kernel-team@fb.com Link: http://lkml.kernel.org/r/1477673892-28940-5-git-send-email-tj@kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-10-28 16:58:12 +00:00
mutex_lock_io(&journal->j_checkpoint_mutex);
err = jbd2_log_do_checkpoint(journal);
mutex_unlock(&journal->j_checkpoint_mutex);
/*
* If checkpointing failed, just free the buffers to avoid
* looping forever
*/
if (err) {
jbd2_journal_destroy_checkpoint(journal);
spin_lock(&journal->j_list_lock);
break;
}
spin_lock(&journal->j_list_lock);
}
J_ASSERT(journal->j_running_transaction == NULL);
J_ASSERT(journal->j_committing_transaction == NULL);
J_ASSERT(journal->j_checkpoint_transactions == NULL);
spin_unlock(&journal->j_list_lock);
/*
* OK, all checkpoint transactions have been checked, now check the
* writeback errseq of fs dev and abort the journal if some buffer
* failed to write back to the original location, otherwise the
* filesystem may become inconsistent.
*/
if (!is_journal_aborted(journal) &&
jbd2_check_fs_dev_write_error(journal))
jbd2_journal_abort(journal, -EIO);
if (journal->j_sb_buffer) {
if (!is_journal_aborted(journal)) {
fs/jbd2, locking/mutex, sched/wait: Use mutex_lock_io() for journal->j_checkpoint_mutex When an ext4 fs is bogged down by a lot of metadata IOs (in the reported case, it was deletion of millions of files, but any massive amount of journal writes would do), after the journal is filled up, tasks which try to access the filesystem and aren't currently performing the journal writes end up waiting in __jbd2_log_wait_for_space() for journal->j_checkpoint_mutex. Because those mutex sleeps aren't marked as iowait, this condition can lead to misleadingly low iowait and /proc/stat:procs_blocked. While iowait propagation is far from strict, this condition can be triggered fairly easily and annotating these sleeps correctly helps initial diagnosis quite a bit. Use the new mutex_lock_io() for journal->j_checkpoint_mutex so that these sleeps are properly marked as iowait. Reported-by: Mingbo Wan <mingbo@fb.com> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Jan Kara <jack@suse.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Theodore Ts'o <tytso@mit.edu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: kernel-team@fb.com Link: http://lkml.kernel.org/r/1477673892-28940-5-git-send-email-tj@kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-10-28 16:58:12 +00:00
mutex_lock_io(&journal->j_checkpoint_mutex);
jbd2: fix FS corruption possibility in jbd2_journal_destroy() on umount path On umount path, jbd2_journal_destroy() writes latest transaction ID (->j_tail_sequence) to be used at next mount. The bug is that ->j_tail_sequence is not holding latest transaction ID in some cases. So, at next mount, there is chance to conflict with remaining (not overwritten yet) transactions. mount (id=10) write transaction (id=11) write transaction (id=12) umount (id=10) <= the bug doesn't write latest ID mount (id=10) write transaction (id=11) crash mount [recovery process] transaction (id=11) transaction (id=12) <= valid transaction ID, but old commit must not replay Like above, this bug become the cause of recovery failure, or FS corruption. So why ->j_tail_sequence doesn't point latest ID? Because if checkpoint transactions was reclaimed by memory pressure (i.e. bdev_try_to_free_page()), then ->j_tail_sequence is not updated. (And another case is, __jbd2_journal_clean_checkpoint_list() is called with empty transaction.) So in above cases, ->j_tail_sequence is not pointing latest transaction ID at umount path. Plus, REQ_FLUSH for checkpoint is not done too. So, to fix this problem with minimum changes, this patch updates ->j_tail_sequence, and issue REQ_FLUSH. (With more complex changes, some optimizations would be possible to avoid unnecessary REQ_FLUSH for example though.) BTW, journal->j_tail_sequence = ++journal->j_transaction_sequence; Increment of ->j_transaction_sequence seems to be unnecessary, but ext3 does this. Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Cc: stable@vger.kernel.org
2016-03-10 04:47:25 +00:00
write_lock(&journal->j_state_lock);
journal->j_tail_sequence =
++journal->j_transaction_sequence;
write_unlock(&journal->j_state_lock);
jbd2_mark_journal_empty(journal, REQ_PREFLUSH | REQ_FUA);
mutex_unlock(&journal->j_checkpoint_mutex);
} else
err = -EIO;
brelse(journal->j_sb_buffer);
}
jbd2,ext4: dynamically allocate the jbd2-journal shrinker In preparation for implementing lockless slab shrink, use new APIs to dynamically allocate the jbd2-journal shrinker, so that it can be freed asynchronously via RCU. Then it doesn't need to wait for RCU read-side critical section when releasing the struct journal_s. Link: https://lkml.kernel.org/r/20230911094444.68966-32-zhengqi.arch@bytedance.com Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Acked-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Abhinav Kumar <quic_abhinavk@quicinc.com> Cc: Alasdair Kergon <agk@redhat.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andreas Gruenbacher <agruenba@redhat.com> Cc: Anna Schumaker <anna@kernel.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Bob Peterson <rpeterso@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Carlos Llamas <cmllamas@google.com> Cc: Chandan Babu R <chandan.babu@oracle.com> Cc: Chao Yu <chao@kernel.org> Cc: Chris Mason <clm@fb.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Christian Koenig <christian.koenig@amd.com> Cc: Chuck Lever <cel@kernel.org> Cc: Coly Li <colyli@suse.de> Cc: Dai Ngo <Dai.Ngo@oracle.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: "Darrick J. Wong" <djwong@kernel.org> Cc: Dave Chinner <david@fromorbit.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David Airlie <airlied@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Sterba <dsterba@suse.com> Cc: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Cc: Gao Xiang <hsiangkao@linux.alibaba.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Huang Rui <ray.huang@amd.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jaegeuk Kim <jaegeuk@kernel.org> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Jeff Layton <jlayton@kernel.org> Cc: Jeffle Xu <jefflexu@linux.alibaba.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Josef Bacik <josef@toxicpanda.com> Cc: Juergen Gross <jgross@suse.com> Cc: Kent Overstreet <kent.overstreet@gmail.com> Cc: Kirill Tkhai <tkhai@ya.ru> Cc: Marijn Suijten <marijn.suijten@somainline.org> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Mike Snitzer <snitzer@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Nadav Amit <namit@vmware.com> Cc: Neil Brown <neilb@suse.de> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: Olga Kornievskaia <kolga@netapp.com> Cc: Paul E. McKenney <paulmck@kernel.org> Cc: Richard Weinberger <richard@nod.at> Cc: Rob Clark <robdclark@gmail.com> Cc: Rob Herring <robh@kernel.org> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: Sean Paul <sean@poorly.run> Cc: Sergey Senozhatsky <senozhatsky@chromium.org> Cc: Song Liu <song@kernel.org> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: Tom Talpey <tom@talpey.com> Cc: Trond Myklebust <trond.myklebust@hammerspace.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Cc: Yue Hu <huyue2@coolpad.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-11 09:44:30 +00:00
if (journal->j_shrinker) {
percpu_counter_destroy(&journal->j_checkpoint_jh_count);
jbd2,ext4: dynamically allocate the jbd2-journal shrinker In preparation for implementing lockless slab shrink, use new APIs to dynamically allocate the jbd2-journal shrinker, so that it can be freed asynchronously via RCU. Then it doesn't need to wait for RCU read-side critical section when releasing the struct journal_s. Link: https://lkml.kernel.org/r/20230911094444.68966-32-zhengqi.arch@bytedance.com Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Acked-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Abhinav Kumar <quic_abhinavk@quicinc.com> Cc: Alasdair Kergon <agk@redhat.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andreas Gruenbacher <agruenba@redhat.com> Cc: Anna Schumaker <anna@kernel.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Bob Peterson <rpeterso@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Carlos Llamas <cmllamas@google.com> Cc: Chandan Babu R <chandan.babu@oracle.com> Cc: Chao Yu <chao@kernel.org> Cc: Chris Mason <clm@fb.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Christian Koenig <christian.koenig@amd.com> Cc: Chuck Lever <cel@kernel.org> Cc: Coly Li <colyli@suse.de> Cc: Dai Ngo <Dai.Ngo@oracle.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: "Darrick J. Wong" <djwong@kernel.org> Cc: Dave Chinner <david@fromorbit.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David Airlie <airlied@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Sterba <dsterba@suse.com> Cc: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Cc: Gao Xiang <hsiangkao@linux.alibaba.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Huang Rui <ray.huang@amd.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jaegeuk Kim <jaegeuk@kernel.org> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Jeff Layton <jlayton@kernel.org> Cc: Jeffle Xu <jefflexu@linux.alibaba.com> Cc: Joel Fernandes (Google) <joel@joelfernandes.org> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Josef Bacik <josef@toxicpanda.com> Cc: Juergen Gross <jgross@suse.com> Cc: Kent Overstreet <kent.overstreet@gmail.com> Cc: Kirill Tkhai <tkhai@ya.ru> Cc: Marijn Suijten <marijn.suijten@somainline.org> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Mike Snitzer <snitzer@kernel.org> Cc: Minchan Kim <minchan@kernel.org> Cc: Muchun Song <muchun.song@linux.dev> Cc: Nadav Amit <namit@vmware.com> Cc: Neil Brown <neilb@suse.de> Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com> Cc: Olga Kornievskaia <kolga@netapp.com> Cc: Paul E. McKenney <paulmck@kernel.org> Cc: Richard Weinberger <richard@nod.at> Cc: Rob Clark <robdclark@gmail.com> Cc: Rob Herring <robh@kernel.org> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: Sean Paul <sean@poorly.run> Cc: Sergey Senozhatsky <senozhatsky@chromium.org> Cc: Song Liu <song@kernel.org> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Steven Price <steven.price@arm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com> Cc: Tom Talpey <tom@talpey.com> Cc: Trond Myklebust <trond.myklebust@hammerspace.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Cc: Yue Hu <huyue2@coolpad.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-09-11 09:44:30 +00:00
shrinker_free(journal->j_shrinker);
}
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
if (journal->j_proc_entry)
jbd2_stats_proc_exit(journal);
iput(journal->j_inode);
if (journal->j_revoke)
jbd2_journal_destroy_revoke(journal);
if (journal->j_chksum_driver)
crypto_free_shash(journal->j_chksum_driver);
kfree(journal->j_fc_wbuf);
kfree(journal->j_wbuf);
kfree(journal);
return err;
}
/**
* jbd2_journal_check_used_features() - Check if features specified are used.
* @journal: Journal to check.
* @compat: bitmask of compatible features
* @ro: bitmask of features that force read-only mount
* @incompat: bitmask of incompatible features
*
* Check whether the journal uses all of a given set of
* features. Return true (non-zero) if it does.
**/
int jbd2_journal_check_used_features(journal_t *journal, unsigned long compat,
unsigned long ro, unsigned long incompat)
{
journal_superblock_t *sb;
if (!compat && !ro && !incompat)
return 1;
if (!jbd2_format_support_feature(journal))
return 0;
sb = journal->j_superblock;
if (((be32_to_cpu(sb->s_feature_compat) & compat) == compat) &&
((be32_to_cpu(sb->s_feature_ro_compat) & ro) == ro) &&
((be32_to_cpu(sb->s_feature_incompat) & incompat) == incompat))
return 1;
return 0;
}
/**
* jbd2_journal_check_available_features() - Check feature set in journalling layer
* @journal: Journal to check.
* @compat: bitmask of compatible features
* @ro: bitmask of features that force read-only mount
* @incompat: bitmask of incompatible features
*
* Check whether the journaling code supports the use of
* all of a given set of features on this journal. Return true
* (non-zero) if it can. */
int jbd2_journal_check_available_features(journal_t *journal, unsigned long compat,
unsigned long ro, unsigned long incompat)
{
if (!compat && !ro && !incompat)
return 1;
if (!jbd2_format_support_feature(journal))
return 0;
if ((compat & JBD2_KNOWN_COMPAT_FEATURES) == compat &&
(ro & JBD2_KNOWN_ROCOMPAT_FEATURES) == ro &&
(incompat & JBD2_KNOWN_INCOMPAT_FEATURES) == incompat)
return 1;
return 0;
}
static int
jbd2_journal_initialize_fast_commit(journal_t *journal)
{
journal_superblock_t *sb = journal->j_superblock;
unsigned long long num_fc_blks;
num_fc_blks = jbd2_journal_get_num_fc_blks(sb);
if (journal->j_last - num_fc_blks < JBD2_MIN_JOURNAL_BLOCKS)
return -ENOSPC;
/* Are we called twice? */
WARN_ON(journal->j_fc_wbuf != NULL);
journal->j_fc_wbuf = kmalloc_array(num_fc_blks,
sizeof(struct buffer_head *), GFP_KERNEL);
if (!journal->j_fc_wbuf)
return -ENOMEM;
journal->j_fc_wbufsize = num_fc_blks;
journal->j_fc_last = journal->j_last;
journal->j_last = journal->j_fc_last - num_fc_blks;
journal->j_fc_first = journal->j_last + 1;
journal->j_fc_off = 0;
journal->j_free = journal->j_last - journal->j_first;
journal->j_max_transaction_buffers =
jbd2_journal_get_max_txn_bufs(journal);
return 0;
}
/**
* jbd2_journal_set_features() - Mark a given journal feature in the superblock
* @journal: Journal to act on.
* @compat: bitmask of compatible features
* @ro: bitmask of features that force read-only mount
* @incompat: bitmask of incompatible features
*
* Mark a given journal feature as present on the
* superblock. Returns true if the requested features could be set.
*
*/
int jbd2_journal_set_features(journal_t *journal, unsigned long compat,
unsigned long ro, unsigned long incompat)
{
#define INCOMPAT_FEATURE_ON(f) \
((incompat & (f)) && !(sb->s_feature_incompat & cpu_to_be32(f)))
#define COMPAT_FEATURE_ON(f) \
((compat & (f)) && !(sb->s_feature_compat & cpu_to_be32(f)))
journal_superblock_t *sb;
if (jbd2_journal_check_used_features(journal, compat, ro, incompat))
return 1;
if (!jbd2_journal_check_available_features(journal, compat, ro, incompat))
return 0;
/* If enabling v2 checksums, turn on v3 instead */
if (incompat & JBD2_FEATURE_INCOMPAT_CSUM_V2) {
incompat &= ~JBD2_FEATURE_INCOMPAT_CSUM_V2;
incompat |= JBD2_FEATURE_INCOMPAT_CSUM_V3;
}
/* Asking for checksumming v3 and v1? Only give them v3. */
if (incompat & JBD2_FEATURE_INCOMPAT_CSUM_V3 &&
compat & JBD2_FEATURE_COMPAT_CHECKSUM)
compat &= ~JBD2_FEATURE_COMPAT_CHECKSUM;
jbd2_debug(1, "Setting new features 0x%lx/0x%lx/0x%lx\n",
compat, ro, incompat);
sb = journal->j_superblock;
if (incompat & JBD2_FEATURE_INCOMPAT_FAST_COMMIT) {
if (jbd2_journal_initialize_fast_commit(journal)) {
pr_err("JBD2: Cannot enable fast commits.\n");
return 0;
}
}
/* Load the checksum driver if necessary */
if ((journal->j_chksum_driver == NULL) &&
INCOMPAT_FEATURE_ON(JBD2_FEATURE_INCOMPAT_CSUM_V3)) {
journal->j_chksum_driver = crypto_alloc_shash("crc32c", 0, 0);
if (IS_ERR(journal->j_chksum_driver)) {
printk(KERN_ERR "JBD2: Cannot load crc32c driver.\n");
journal->j_chksum_driver = NULL;
return 0;
}
/* Precompute checksum seed for all metadata */
journal->j_csum_seed = jbd2_chksum(journal, ~0, sb->s_uuid,
sizeof(sb->s_uuid));
}
lock_buffer(journal->j_sb_buffer);
/* If enabling v3 checksums, update superblock */
if (INCOMPAT_FEATURE_ON(JBD2_FEATURE_INCOMPAT_CSUM_V3)) {
sb->s_checksum_type = JBD2_CRC32C_CHKSUM;
sb->s_feature_compat &=
~cpu_to_be32(JBD2_FEATURE_COMPAT_CHECKSUM);
}
/* If enabling v1 checksums, downgrade superblock */
if (COMPAT_FEATURE_ON(JBD2_FEATURE_COMPAT_CHECKSUM))
sb->s_feature_incompat &=
~cpu_to_be32(JBD2_FEATURE_INCOMPAT_CSUM_V2 |
JBD2_FEATURE_INCOMPAT_CSUM_V3);
sb->s_feature_compat |= cpu_to_be32(compat);
sb->s_feature_ro_compat |= cpu_to_be32(ro);
sb->s_feature_incompat |= cpu_to_be32(incompat);
unlock_buffer(journal->j_sb_buffer);
journal->j_revoke_records_per_block =
journal_revoke_records_per_block(journal);
return 1;
#undef COMPAT_FEATURE_ON
#undef INCOMPAT_FEATURE_ON
}
/*
* jbd2_journal_clear_features() - Clear a given journal feature in the
* superblock
* @journal: Journal to act on.
* @compat: bitmask of compatible features
* @ro: bitmask of features that force read-only mount
* @incompat: bitmask of incompatible features
*
* Clear a given journal feature as present on the
* superblock.
*/
void jbd2_journal_clear_features(journal_t *journal, unsigned long compat,
unsigned long ro, unsigned long incompat)
{
journal_superblock_t *sb;
jbd2_debug(1, "Clear features 0x%lx/0x%lx/0x%lx\n",
compat, ro, incompat);
sb = journal->j_superblock;
sb->s_feature_compat &= ~cpu_to_be32(compat);
sb->s_feature_ro_compat &= ~cpu_to_be32(ro);
sb->s_feature_incompat &= ~cpu_to_be32(incompat);
journal->j_revoke_records_per_block =
journal_revoke_records_per_block(journal);
}
EXPORT_SYMBOL(jbd2_journal_clear_features);
/**
* jbd2_journal_flush() - Flush journal
* @journal: Journal to act on.
* @flags: optional operation on the journal blocks after the flush (see below)
*
* Flush all data for a given journal to disk and empty the journal.
* Filesystems can use this when remounting readonly to ensure that
* recovery does not need to happen on remount. Optionally, a discard or zeroout
* can be issued on the journal blocks after flushing.
*
* flags:
* JBD2_JOURNAL_FLUSH_DISCARD: issues discards for the journal blocks
* JBD2_JOURNAL_FLUSH_ZEROOUT: issues zeroouts for the journal blocks
*/
int jbd2_journal_flush(journal_t *journal, unsigned int flags)
{
int err = 0;
transaction_t *transaction = NULL;
write_lock(&journal->j_state_lock);
/* Force everything buffered to the log... */
if (journal->j_running_transaction) {
transaction = journal->j_running_transaction;
__jbd2_log_start_commit(journal, transaction->t_tid);
} else if (journal->j_committing_transaction)
transaction = journal->j_committing_transaction;
/* Wait for the log commit to complete... */
if (transaction) {
tid_t tid = transaction->t_tid;
write_unlock(&journal->j_state_lock);
jbd2_log_wait_commit(journal, tid);
} else {
write_unlock(&journal->j_state_lock);
}
/* ...and flush everything in the log out to disk. */
spin_lock(&journal->j_list_lock);
while (!err && journal->j_checkpoint_transactions != NULL) {
spin_unlock(&journal->j_list_lock);
fs/jbd2, locking/mutex, sched/wait: Use mutex_lock_io() for journal->j_checkpoint_mutex When an ext4 fs is bogged down by a lot of metadata IOs (in the reported case, it was deletion of millions of files, but any massive amount of journal writes would do), after the journal is filled up, tasks which try to access the filesystem and aren't currently performing the journal writes end up waiting in __jbd2_log_wait_for_space() for journal->j_checkpoint_mutex. Because those mutex sleeps aren't marked as iowait, this condition can lead to misleadingly low iowait and /proc/stat:procs_blocked. While iowait propagation is far from strict, this condition can be triggered fairly easily and annotating these sleeps correctly helps initial diagnosis quite a bit. Use the new mutex_lock_io() for journal->j_checkpoint_mutex so that these sleeps are properly marked as iowait. Reported-by: Mingbo Wan <mingbo@fb.com> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Jan Kara <jack@suse.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Theodore Ts'o <tytso@mit.edu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: kernel-team@fb.com Link: http://lkml.kernel.org/r/1477673892-28940-5-git-send-email-tj@kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-10-28 16:58:12 +00:00
mutex_lock_io(&journal->j_checkpoint_mutex);
err = jbd2_log_do_checkpoint(journal);
mutex_unlock(&journal->j_checkpoint_mutex);
spin_lock(&journal->j_list_lock);
}
spin_unlock(&journal->j_list_lock);
if (is_journal_aborted(journal))
return -EIO;
fs/jbd2, locking/mutex, sched/wait: Use mutex_lock_io() for journal->j_checkpoint_mutex When an ext4 fs is bogged down by a lot of metadata IOs (in the reported case, it was deletion of millions of files, but any massive amount of journal writes would do), after the journal is filled up, tasks which try to access the filesystem and aren't currently performing the journal writes end up waiting in __jbd2_log_wait_for_space() for journal->j_checkpoint_mutex. Because those mutex sleeps aren't marked as iowait, this condition can lead to misleadingly low iowait and /proc/stat:procs_blocked. While iowait propagation is far from strict, this condition can be triggered fairly easily and annotating these sleeps correctly helps initial diagnosis quite a bit. Use the new mutex_lock_io() for journal->j_checkpoint_mutex so that these sleeps are properly marked as iowait. Reported-by: Mingbo Wan <mingbo@fb.com> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Jan Kara <jack@suse.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Theodore Ts'o <tytso@mit.edu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: kernel-team@fb.com Link: http://lkml.kernel.org/r/1477673892-28940-5-git-send-email-tj@kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-10-28 16:58:12 +00:00
mutex_lock_io(&journal->j_checkpoint_mutex);
if (!err) {
err = jbd2_cleanup_journal_tail(journal);
if (err < 0) {
mutex_unlock(&journal->j_checkpoint_mutex);
goto out;
}
err = 0;
}
/* Finally, mark the journal as really needing no recovery.
* This sets s_start==0 in the underlying superblock, which is
* the magic code for a fully-recovered superblock. Any future
* commits of data to the journal will restore the current
* s_start value. */
jbd2_mark_journal_empty(journal, REQ_FUA);
if (flags)
err = __jbd2_journal_erase(journal, flags);
mutex_unlock(&journal->j_checkpoint_mutex);
write_lock(&journal->j_state_lock);
J_ASSERT(!journal->j_running_transaction);
J_ASSERT(!journal->j_committing_transaction);
J_ASSERT(!journal->j_checkpoint_transactions);
J_ASSERT(journal->j_head == journal->j_tail);
J_ASSERT(journal->j_tail_sequence == journal->j_transaction_sequence);
write_unlock(&journal->j_state_lock);
out:
return err;
}
/**
* jbd2_journal_wipe() - Wipe journal contents
* @journal: Journal to act on.
* @write: flag (see below)
*
* Wipe out all of the contents of a journal, safely. This will produce
* a warning if the journal contains any valid recovery information.
* Must be called between journal_init_*() and jbd2_journal_load().
*
* If 'write' is non-zero, then we wipe out the journal on disk; otherwise
* we merely suppress recovery.
*/
int jbd2_journal_wipe(journal_t *journal, int write)
{
int err;
J_ASSERT (!(journal->j_flags & JBD2_LOADED));
if (!journal->j_tail)
return 0;
printk(KERN_WARNING "JBD2: %s recovery information on journal\n",
write ? "Clearing" : "Ignoring");
err = jbd2_journal_skip_recovery(journal);
if (write) {
/* Lock to make assertions happy... */
mutex_lock_io(&journal->j_checkpoint_mutex);
jbd2_mark_journal_empty(journal, REQ_FUA);
mutex_unlock(&journal->j_checkpoint_mutex);
}
return err;
}
/**
* jbd2_journal_abort () - Shutdown the journal immediately.
* @journal: the journal to shutdown.
* @errno: an error number to record in the journal indicating
* the reason for the shutdown.
*
* Perform a complete, immediate shutdown of the ENTIRE
* journal (not of a single transaction). This operation cannot be
* undone without closing and reopening the journal.
*
* The jbd2_journal_abort function is intended to support higher level error
* recovery mechanisms such as the ext2/ext3 remount-readonly error
* mode.
*
* Journal abort has very specific semantics. Any existing dirty,
* unjournaled buffers in the main filesystem will still be written to
* disk by bdflush, but the journaling mechanism will be suspended
* immediately and no further transaction commits will be honoured.
*
* Any dirty, journaled buffers will be written back to disk without
* hitting the journal. Atomicity cannot be guaranteed on an aborted
* filesystem, but we _do_ attempt to leave as much data as possible
* behind for fsck to use for cleanup.
*
* Any attempt to get a new transaction handle on a journal which is in
* ABORT state will just result in an -EROFS error return. A
* jbd2_journal_stop on an existing handle will return -EIO if we have
* entered abort state during the update.
*
* Recursive transactions are not disturbed by journal abort until the
* final jbd2_journal_stop, which will receive the -EIO error.
*
* Finally, the jbd2_journal_abort call allows the caller to supply an errno
* which will be recorded (if possible) in the journal superblock. This
* allows a client to record failure conditions in the middle of a
* transaction without having to complete the transaction to record the
* failure to disk. ext3_error, for example, now uses this
* functionality.
*
*/
void jbd2_journal_abort(journal_t *journal, int errno)
{
transaction_t *transaction;
/*
* Lock the aborting procedure until everything is done, this avoid
* races between filesystem's error handling flow (e.g. ext4_abort()),
* ensure panic after the error info is written into journal's
* superblock.
*/
mutex_lock(&journal->j_abort_mutex);
/*
* ESHUTDOWN always takes precedence because a file system check
* caused by any other journal abort error is not required after
* a shutdown triggered.
*/
write_lock(&journal->j_state_lock);
if (journal->j_flags & JBD2_ABORT) {
int old_errno = journal->j_errno;
write_unlock(&journal->j_state_lock);
if (old_errno != -ESHUTDOWN && errno == -ESHUTDOWN) {
journal->j_errno = errno;
jbd2_journal_update_sb_errno(journal);
}
mutex_unlock(&journal->j_abort_mutex);
return;
}
/*
* Mark the abort as occurred and start current running transaction
* to release all journaled buffer.
*/
pr_err("Aborting journal on device %s.\n", journal->j_devname);
journal->j_flags |= JBD2_ABORT;
journal->j_errno = errno;
transaction = journal->j_running_transaction;
if (transaction)
__jbd2_log_start_commit(journal, transaction->t_tid);
write_unlock(&journal->j_state_lock);
/*
* Record errno to the journal super block, so that fsck and jbd2
* layer could realise that a filesystem check is needed.
*/
jbd2_journal_update_sb_errno(journal);
mutex_unlock(&journal->j_abort_mutex);
}
/**
* jbd2_journal_errno() - returns the journal's error state.
* @journal: journal to examine.
*
* This is the errno number set with jbd2_journal_abort(), the last
* time the journal was mounted - if the journal was stopped
* without calling abort this will be 0.
*
* If the journal has been aborted on this mount time -EROFS will
* be returned.
*/
int jbd2_journal_errno(journal_t *journal)
{
int err;
read_lock(&journal->j_state_lock);
if (journal->j_flags & JBD2_ABORT)
err = -EROFS;
else
err = journal->j_errno;
read_unlock(&journal->j_state_lock);
return err;
}
/**
* jbd2_journal_clear_err() - clears the journal's error state
* @journal: journal to act on.
*
* An error must be cleared or acked to take a FS out of readonly
* mode.
*/
int jbd2_journal_clear_err(journal_t *journal)
{
int err = 0;
write_lock(&journal->j_state_lock);
if (journal->j_flags & JBD2_ABORT)
err = -EROFS;
else
journal->j_errno = 0;
write_unlock(&journal->j_state_lock);
return err;
}
/**
* jbd2_journal_ack_err() - Ack journal err.
* @journal: journal to act on.
*
* An error must be cleared or acked to take a FS out of readonly
* mode.
*/
void jbd2_journal_ack_err(journal_t *journal)
{
write_lock(&journal->j_state_lock);
if (journal->j_errno)
journal->j_flags |= JBD2_ACK_ERR;
write_unlock(&journal->j_state_lock);
}
int jbd2_journal_blocks_per_page(struct inode *inode)
{
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
return 1 << (PAGE_SHIFT - inode->i_sb->s_blocksize_bits);
}
/*
* helper functions to deal with 32 or 64bit block numbers.
*/
size_t journal_tag_bytes(journal_t *journal)
{
size_t sz;
if (jbd2_has_feature_csum3(journal))
return sizeof(journal_block_tag3_t);
sz = sizeof(journal_block_tag_t);
if (jbd2_has_feature_csum2(journal))
sz += sizeof(__u16);
if (jbd2_has_feature_64bit(journal))
return sz;
else
return sz - sizeof(__u32);
}
/*
* JBD memory management
*
* These functions are used to allocate block-sized chunks of memory
* used for making copies of buffer_head data. Very often it will be
* page-sized chunks of data, but sometimes it will be in
* sub-page-size chunks. (For example, 16k pages on Power systems
* with a 4k block file system.) For blocks smaller than a page, we
* use a SLAB allocator. There are slab caches for each block size,
* which are allocated at mount time, if necessary, and we only free
* (all of) the slab caches when/if the jbd2 module is unloaded. For
* this reason we don't need to a mutex to protect access to
* jbd2_slab[] allocating or releasing memory; only in
* jbd2_journal_create_slab().
*/
#define JBD2_MAX_SLABS 8
static struct kmem_cache *jbd2_slab[JBD2_MAX_SLABS];
static const char *jbd2_slab_names[JBD2_MAX_SLABS] = {
"jbd2_1k", "jbd2_2k", "jbd2_4k", "jbd2_8k",
"jbd2_16k", "jbd2_32k", "jbd2_64k", "jbd2_128k"
};
static void jbd2_journal_destroy_slabs(void)
{
int i;
for (i = 0; i < JBD2_MAX_SLABS; i++) {
kmem_cache_destroy(jbd2_slab[i]);
jbd2_slab[i] = NULL;
}
}
static int jbd2_journal_create_slab(size_t size)
{
static DEFINE_MUTEX(jbd2_slab_create_mutex);
int i = order_base_2(size) - 10;
size_t slab_size;
if (size == PAGE_SIZE)
return 0;
if (i >= JBD2_MAX_SLABS)
return -EINVAL;
if (unlikely(i < 0))
i = 0;
mutex_lock(&jbd2_slab_create_mutex);
if (jbd2_slab[i]) {
mutex_unlock(&jbd2_slab_create_mutex);
return 0; /* Already created */
}
slab_size = 1 << (i+10);
jbd2_slab[i] = kmem_cache_create(jbd2_slab_names[i], slab_size,
slab_size, 0, NULL);
mutex_unlock(&jbd2_slab_create_mutex);
if (!jbd2_slab[i]) {
printk(KERN_EMERG "JBD2: no memory for jbd2_slab cache\n");
return -ENOMEM;
}
return 0;
}
static struct kmem_cache *get_slab(size_t size)
{
int i = order_base_2(size) - 10;
BUG_ON(i >= JBD2_MAX_SLABS);
if (unlikely(i < 0))
i = 0;
BUG_ON(jbd2_slab[i] == NULL);
return jbd2_slab[i];
}
void *jbd2_alloc(size_t size, gfp_t flags)
{
void *ptr;
BUG_ON(size & (size-1)); /* Must be a power of 2 */
if (size < PAGE_SIZE)
ptr = kmem_cache_alloc(get_slab(size), flags);
else
ptr = (void *)__get_free_pages(flags, get_order(size));
/* Check alignment; SLUB has gotten this wrong in the past,
* and this can lead to user data corruption! */
BUG_ON(((unsigned long) ptr) & (size-1));
return ptr;
}
void jbd2_free(void *ptr, size_t size)
{
if (size < PAGE_SIZE)
kmem_cache_free(get_slab(size), ptr);
else
free_pages((unsigned long)ptr, get_order(size));
};
/*
* Journal_head storage management
*/
static struct kmem_cache *jbd2_journal_head_cache;
#ifdef CONFIG_JBD2_DEBUG
static atomic_t nr_journal_heads = ATOMIC_INIT(0);
#endif
static int __init jbd2_journal_init_journal_head_cache(void)
{
J_ASSERT(!jbd2_journal_head_cache);
jbd2_journal_head_cache = kmem_cache_create("jbd2_journal_head",
sizeof(struct journal_head),
0, /* offset */
SLAB_TEMPORARY | SLAB_TYPESAFE_BY_RCU,
NULL); /* ctor */
if (!jbd2_journal_head_cache) {
printk(KERN_EMERG "JBD2: no memory for journal_head cache\n");
return -ENOMEM;
}
return 0;
}
static void jbd2_journal_destroy_journal_head_cache(void)
{
kmem_cache_destroy(jbd2_journal_head_cache);
jbd2_journal_head_cache = NULL;
}
/*
* journal_head splicing and dicing
*/
static struct journal_head *journal_alloc_journal_head(void)
{
struct journal_head *ret;
#ifdef CONFIG_JBD2_DEBUG
atomic_inc(&nr_journal_heads);
#endif
ret = kmem_cache_zalloc(jbd2_journal_head_cache, GFP_NOFS);
if (!ret) {
jbd2_debug(1, "out of memory for journal_head\n");
pr_notice_ratelimited("ENOMEM in %s, retrying.\n", __func__);
ret = kmem_cache_zalloc(jbd2_journal_head_cache,
GFP_NOFS | __GFP_NOFAIL);
}
jbd2: Make state lock a spinlock Bit-spinlocks are problematic on PREEMPT_RT if functions which might sleep on RT, e.g. spin_lock(), alloc/free(), are invoked inside the lock held region because bit spinlocks disable preemption even on RT. A first attempt was to replace state lock with a spinlock placed in struct buffer_head and make the locking conditional on PREEMPT_RT and DEBUG_BIT_SPINLOCKS. Jan pointed out that there is a 4 byte hole in struct journal_head where a regular spinlock fits in and he would not object to convert the state lock to a spinlock unconditionally. Aside of solving the RT problem, this also gains lockdep coverage for the journal head state lock (bit-spinlocks are not covered by lockdep as it's hard to fit a lockdep map into a single bit). The trivial change would have been to convert the jbd_*lock_bh_state() inlines, but that comes with the downside that these functions take a buffer head pointer which needs to be converted to a journal head pointer which adds another level of indirection. As almost all functions which use this lock have a journal head pointer readily available, it makes more sense to remove the lock helper inlines and write out spin_*lock() at all call sites. Fixup all locking comments as well. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Jan Kara <jack@suse.cz> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joseph Qi <joseph.qi@linux.alibaba.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Jan Kara <jack@suse.com> Cc: linux-ext4@vger.kernel.org Link: https://lore.kernel.org/r/20190809124233.13277-7-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:32 +00:00
if (ret)
spin_lock_init(&ret->b_state_lock);
return ret;
}
static void journal_free_journal_head(struct journal_head *jh)
{
#ifdef CONFIG_JBD2_DEBUG
atomic_dec(&nr_journal_heads);
memset(jh, JBD2_POISON_FREE, sizeof(*jh));
#endif
kmem_cache_free(jbd2_journal_head_cache, jh);
}
/*
* A journal_head is attached to a buffer_head whenever JBD has an
* interest in the buffer.
*
* Whenever a buffer has an attached journal_head, its ->b_state:BH_JBD bit
* is set. This bit is tested in core kernel code where we need to take
* JBD-specific actions. Testing the zeroness of ->b_private is not reliable
* there.
*
* When a buffer has its BH_JBD bit set, its ->b_count is elevated by one.
*
* When a buffer has its BH_JBD bit set it is immune from being released by
* core kernel code, mainly via ->b_count.
*
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
* A journal_head is detached from its buffer_head when the journal_head's
* b_jcount reaches zero. Running transaction (b_transaction) and checkpoint
* transaction (b_cp_transaction) hold their references to b_jcount.
*
* Various places in the kernel want to attach a journal_head to a buffer_head
* _before_ attaching the journal_head to a transaction. To protect the
* journal_head in this situation, jbd2_journal_add_journal_head elevates the
* journal_head's b_jcount refcount by one. The caller must call
* jbd2_journal_put_journal_head() to undo this.
*
* So the typical usage would be:
*
* (Attach a journal_head if needed. Increments b_jcount)
* struct journal_head *jh = jbd2_journal_add_journal_head(bh);
* ...
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
* (Get another reference for transaction)
* jbd2_journal_grab_journal_head(bh);
* jh->b_transaction = xxx;
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
* (Put original reference)
* jbd2_journal_put_journal_head(jh);
*/
/*
* Give a buffer_head a journal_head.
*
* May sleep.
*/
struct journal_head *jbd2_journal_add_journal_head(struct buffer_head *bh)
{
struct journal_head *jh;
struct journal_head *new_jh = NULL;
repeat:
if (!buffer_jbd(bh))
new_jh = journal_alloc_journal_head();
jbd_lock_bh_journal_head(bh);
if (buffer_jbd(bh)) {
jh = bh2jh(bh);
} else {
J_ASSERT_BH(bh,
(atomic_read(&bh->b_count) > 0) ||
(bh->b_folio && bh->b_folio->mapping));
if (!new_jh) {
jbd_unlock_bh_journal_head(bh);
goto repeat;
}
jh = new_jh;
new_jh = NULL; /* We consumed it */
set_buffer_jbd(bh);
bh->b_private = jh;
jh->b_bh = bh;
get_bh(bh);
BUFFER_TRACE(bh, "added journal_head");
}
jh->b_jcount++;
jbd_unlock_bh_journal_head(bh);
if (new_jh)
journal_free_journal_head(new_jh);
return bh->b_private;
}
/*
* Grab a ref against this buffer_head's journal_head. If it ended up not
* having a journal_head, return NULL
*/
struct journal_head *jbd2_journal_grab_journal_head(struct buffer_head *bh)
{
struct journal_head *jh = NULL;
jbd_lock_bh_journal_head(bh);
if (buffer_jbd(bh)) {
jh = bh2jh(bh);
jh->b_jcount++;
}
jbd_unlock_bh_journal_head(bh);
return jh;
}
EXPORT_SYMBOL(jbd2_journal_grab_journal_head);
static void __journal_remove_journal_head(struct buffer_head *bh)
{
struct journal_head *jh = bh2jh(bh);
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
J_ASSERT_JH(jh, jh->b_transaction == NULL);
J_ASSERT_JH(jh, jh->b_next_transaction == NULL);
J_ASSERT_JH(jh, jh->b_cp_transaction == NULL);
J_ASSERT_JH(jh, jh->b_jlist == BJ_None);
J_ASSERT_BH(bh, buffer_jbd(bh));
J_ASSERT_BH(bh, jh2bh(jh) == bh);
BUFFER_TRACE(bh, "remove journal_head");
jbd2: Free journal head outside of locked region On PREEMPT_RT bit-spinlocks have the same semantics as on PREEMPT_RT=n, i.e. they disable preemption. That means functions which are not safe to be called in preempt disabled context on RT trigger a might_sleep() assert. The journal head bit spinlock is mostly held for short code sequences with trivial RT safe functionality, except for one place: jbd2_journal_put_journal_head() invokes __journal_remove_journal_head() with the journal head bit spinlock held. __journal_remove_journal_head() invokes kmem_cache_free() which must not be called with preemption disabled on RT. Jan suggested to rework the removal function so the actual free happens outside the bit-spinlocked region. Split it into two parts: - Do the sanity checks and the buffer head detach under the lock - Do the actual free after dropping the lock There is error case handling in the free part which needs to dereference the b_size field of the now detached buffer head. Due to paranoia (caused by ignorance) the size is retrieved in the detach function and handed into the free function. Might be over-engineered, but better safe than sorry. This makes the journal head bit-spinlock usage RT compliant and also avoids nested locking which is not covered by lockdep. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: linux-ext4@vger.kernel.org Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan Kara <jack@suse.com> Signed-off-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20190809124233.13277-8-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:33 +00:00
/* Unlink before dropping the lock */
bh->b_private = NULL;
jh->b_bh = NULL; /* debug, really */
clear_buffer_jbd(bh);
}
static void journal_release_journal_head(struct journal_head *jh, size_t b_size)
{
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
if (jh->b_frozen_data) {
printk(KERN_WARNING "%s: freeing b_frozen_data\n", __func__);
jbd2: Free journal head outside of locked region On PREEMPT_RT bit-spinlocks have the same semantics as on PREEMPT_RT=n, i.e. they disable preemption. That means functions which are not safe to be called in preempt disabled context on RT trigger a might_sleep() assert. The journal head bit spinlock is mostly held for short code sequences with trivial RT safe functionality, except for one place: jbd2_journal_put_journal_head() invokes __journal_remove_journal_head() with the journal head bit spinlock held. __journal_remove_journal_head() invokes kmem_cache_free() which must not be called with preemption disabled on RT. Jan suggested to rework the removal function so the actual free happens outside the bit-spinlocked region. Split it into two parts: - Do the sanity checks and the buffer head detach under the lock - Do the actual free after dropping the lock There is error case handling in the free part which needs to dereference the b_size field of the now detached buffer head. Due to paranoia (caused by ignorance) the size is retrieved in the detach function and handed into the free function. Might be over-engineered, but better safe than sorry. This makes the journal head bit-spinlock usage RT compliant and also avoids nested locking which is not covered by lockdep. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: linux-ext4@vger.kernel.org Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan Kara <jack@suse.com> Signed-off-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20190809124233.13277-8-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:33 +00:00
jbd2_free(jh->b_frozen_data, b_size);
}
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
if (jh->b_committed_data) {
printk(KERN_WARNING "%s: freeing b_committed_data\n", __func__);
jbd2: Free journal head outside of locked region On PREEMPT_RT bit-spinlocks have the same semantics as on PREEMPT_RT=n, i.e. they disable preemption. That means functions which are not safe to be called in preempt disabled context on RT trigger a might_sleep() assert. The journal head bit spinlock is mostly held for short code sequences with trivial RT safe functionality, except for one place: jbd2_journal_put_journal_head() invokes __journal_remove_journal_head() with the journal head bit spinlock held. __journal_remove_journal_head() invokes kmem_cache_free() which must not be called with preemption disabled on RT. Jan suggested to rework the removal function so the actual free happens outside the bit-spinlocked region. Split it into two parts: - Do the sanity checks and the buffer head detach under the lock - Do the actual free after dropping the lock There is error case handling in the free part which needs to dereference the b_size field of the now detached buffer head. Due to paranoia (caused by ignorance) the size is retrieved in the detach function and handed into the free function. Might be over-engineered, but better safe than sorry. This makes the journal head bit-spinlock usage RT compliant and also avoids nested locking which is not covered by lockdep. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: linux-ext4@vger.kernel.org Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan Kara <jack@suse.com> Signed-off-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20190809124233.13277-8-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:33 +00:00
jbd2_free(jh->b_committed_data, b_size);
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
}
journal_free_journal_head(jh);
}
/*
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
* Drop a reference on the passed journal_head. If it fell to zero then
* release the journal_head from the buffer_head.
*/
void jbd2_journal_put_journal_head(struct journal_head *jh)
{
struct buffer_head *bh = jh2bh(jh);
jbd_lock_bh_journal_head(bh);
J_ASSERT_JH(jh, jh->b_jcount > 0);
--jh->b_jcount;
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
if (!jh->b_jcount) {
__journal_remove_journal_head(bh);
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
jbd_unlock_bh_journal_head(bh);
jbd2: Free journal head outside of locked region On PREEMPT_RT bit-spinlocks have the same semantics as on PREEMPT_RT=n, i.e. they disable preemption. That means functions which are not safe to be called in preempt disabled context on RT trigger a might_sleep() assert. The journal head bit spinlock is mostly held for short code sequences with trivial RT safe functionality, except for one place: jbd2_journal_put_journal_head() invokes __journal_remove_journal_head() with the journal head bit spinlock held. __journal_remove_journal_head() invokes kmem_cache_free() which must not be called with preemption disabled on RT. Jan suggested to rework the removal function so the actual free happens outside the bit-spinlocked region. Split it into two parts: - Do the sanity checks and the buffer head detach under the lock - Do the actual free after dropping the lock There is error case handling in the free part which needs to dereference the b_size field of the now detached buffer head. Due to paranoia (caused by ignorance) the size is retrieved in the detach function and handed into the free function. Might be over-engineered, but better safe than sorry. This makes the journal head bit-spinlock usage RT compliant and also avoids nested locking which is not covered by lockdep. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: linux-ext4@vger.kernel.org Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan Kara <jack@suse.com> Signed-off-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20190809124233.13277-8-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:33 +00:00
journal_release_journal_head(jh, bh->b_size);
__brelse(bh);
jbd2: Free journal head outside of locked region On PREEMPT_RT bit-spinlocks have the same semantics as on PREEMPT_RT=n, i.e. they disable preemption. That means functions which are not safe to be called in preempt disabled context on RT trigger a might_sleep() assert. The journal head bit spinlock is mostly held for short code sequences with trivial RT safe functionality, except for one place: jbd2_journal_put_journal_head() invokes __journal_remove_journal_head() with the journal head bit spinlock held. __journal_remove_journal_head() invokes kmem_cache_free() which must not be called with preemption disabled on RT. Jan suggested to rework the removal function so the actual free happens outside the bit-spinlocked region. Split it into two parts: - Do the sanity checks and the buffer head detach under the lock - Do the actual free after dropping the lock There is error case handling in the free part which needs to dereference the b_size field of the now detached buffer head. Due to paranoia (caused by ignorance) the size is retrieved in the detach function and handed into the free function. Might be over-engineered, but better safe than sorry. This makes the journal head bit-spinlock usage RT compliant and also avoids nested locking which is not covered by lockdep. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: linux-ext4@vger.kernel.org Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan Kara <jack@suse.com> Signed-off-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20190809124233.13277-8-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:33 +00:00
} else {
jbd2: Fix oops in jbd2_journal_remove_journal_head() jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
2011-06-13 19:38:22 +00:00
jbd_unlock_bh_journal_head(bh);
jbd2: Free journal head outside of locked region On PREEMPT_RT bit-spinlocks have the same semantics as on PREEMPT_RT=n, i.e. they disable preemption. That means functions which are not safe to be called in preempt disabled context on RT trigger a might_sleep() assert. The journal head bit spinlock is mostly held for short code sequences with trivial RT safe functionality, except for one place: jbd2_journal_put_journal_head() invokes __journal_remove_journal_head() with the journal head bit spinlock held. __journal_remove_journal_head() invokes kmem_cache_free() which must not be called with preemption disabled on RT. Jan suggested to rework the removal function so the actual free happens outside the bit-spinlocked region. Split it into two parts: - Do the sanity checks and the buffer head detach under the lock - Do the actual free after dropping the lock There is error case handling in the free part which needs to dereference the b_size field of the now detached buffer head. Due to paranoia (caused by ignorance) the size is retrieved in the detach function and handed into the free function. Might be over-engineered, but better safe than sorry. This makes the journal head bit-spinlock usage RT compliant and also avoids nested locking which is not covered by lockdep. Suggested-by: Jan Kara <jack@suse.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: linux-ext4@vger.kernel.org Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan Kara <jack@suse.com> Signed-off-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20190809124233.13277-8-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2019-08-09 12:42:33 +00:00
}
}
EXPORT_SYMBOL(jbd2_journal_put_journal_head);
/*
* Initialize jbd inode head
*/
void jbd2_journal_init_jbd_inode(struct jbd2_inode *jinode, struct inode *inode)
{
jinode->i_transaction = NULL;
jinode->i_next_transaction = NULL;
jinode->i_vfs_inode = inode;
jinode->i_flags = 0;
jinode->i_dirty_start = 0;
jinode->i_dirty_end = 0;
INIT_LIST_HEAD(&jinode->i_list);
}
/*
* Function to be called before we start removing inode from memory (i.e.,
* clear_inode() is a fine place to be called from). It removes inode from
* transaction's lists.
*/
void jbd2_journal_release_jbd_inode(journal_t *journal,
struct jbd2_inode *jinode)
{
if (!journal)
return;
restart:
spin_lock(&journal->j_list_lock);
/* Is commit writing out inode - we have to wait */
if (jinode->i_flags & JI_COMMIT_RUNNING) {
wait_queue_head_t *wq;
DEFINE_WAIT_BIT(wait, &jinode->i_flags, __JI_COMMIT_RUNNING);
wq = bit_waitqueue(&jinode->i_flags, __JI_COMMIT_RUNNING);
prepare_to_wait(wq, &wait.wq_entry, TASK_UNINTERRUPTIBLE);
spin_unlock(&journal->j_list_lock);
schedule();
finish_wait(wq, &wait.wq_entry);
goto restart;
}
if (jinode->i_transaction) {
list_del(&jinode->i_list);
jinode->i_transaction = NULL;
}
spin_unlock(&journal->j_list_lock);
}
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
#ifdef CONFIG_PROC_FS
#define JBD2_STATS_PROC_NAME "fs/jbd2"
static void __init jbd2_create_jbd_stats_proc_entry(void)
{
proc_jbd2_stats = proc_mkdir(JBD2_STATS_PROC_NAME, NULL);
}
static void __exit jbd2_remove_jbd_stats_proc_entry(void)
{
if (proc_jbd2_stats)
remove_proc_entry(JBD2_STATS_PROC_NAME, NULL);
}
#else
#define jbd2_create_jbd_stats_proc_entry() do {} while (0)
#define jbd2_remove_jbd_stats_proc_entry() do {} while (0)
#endif
struct kmem_cache *jbd2_handle_cache, *jbd2_inode_cache;
static int __init jbd2_journal_init_inode_cache(void)
{
J_ASSERT(!jbd2_inode_cache);
jbd2_inode_cache = KMEM_CACHE(jbd2_inode, 0);
if (!jbd2_inode_cache) {
pr_emerg("JBD2: failed to create inode cache\n");
return -ENOMEM;
}
return 0;
}
static int __init jbd2_journal_init_handle_cache(void)
{
J_ASSERT(!jbd2_handle_cache);
jbd2_handle_cache = KMEM_CACHE(jbd2_journal_handle, SLAB_TEMPORARY);
if (!jbd2_handle_cache) {
printk(KERN_EMERG "JBD2: failed to create handle cache\n");
return -ENOMEM;
}
return 0;
}
static void jbd2_journal_destroy_inode_cache(void)
{
kmem_cache_destroy(jbd2_inode_cache);
jbd2_inode_cache = NULL;
}
static void jbd2_journal_destroy_handle_cache(void)
{
kmem_cache_destroy(jbd2_handle_cache);
jbd2_handle_cache = NULL;
}
/*
* Module startup and shutdown
*/
static int __init journal_init_caches(void)
{
int ret;
ret = jbd2_journal_init_revoke_record_cache();
if (ret == 0)
ret = jbd2_journal_init_revoke_table_cache();
if (ret == 0)
ret = jbd2_journal_init_journal_head_cache();
if (ret == 0)
ret = jbd2_journal_init_handle_cache();
if (ret == 0)
ret = jbd2_journal_init_inode_cache();
if (ret == 0)
ret = jbd2_journal_init_transaction_cache();
return ret;
}
static void jbd2_journal_destroy_caches(void)
{
jbd2_journal_destroy_revoke_record_cache();
jbd2_journal_destroy_revoke_table_cache();
jbd2_journal_destroy_journal_head_cache();
jbd2_journal_destroy_handle_cache();
jbd2_journal_destroy_inode_cache();
jbd2_journal_destroy_transaction_cache();
jbd2_journal_destroy_slabs();
}
static int __init journal_init(void)
{
int ret;
BUILD_BUG_ON(sizeof(struct journal_superblock_s) != 1024);
ret = journal_init_caches();
if (ret == 0) {
jbd2_create_jbd_stats_proc_entry();
} else {
jbd2_journal_destroy_caches();
}
return ret;
}
static void __exit journal_exit(void)
{
#ifdef CONFIG_JBD2_DEBUG
int n = atomic_read(&nr_journal_heads);
if (n)
printk(KERN_ERR "JBD2: leaked %d journal_heads!\n", n);
#endif
jbd2: jbd2 stats through procfs The patch below updates the jbd stats patch to 2.6.20/jbd2. The initial patch was posted by Alex Tomas in December 2005 (http://marc.info/?l=linux-ext4&m=113538565128617&w=2). It provides statistics via procfs such as transaction lifetime and size. Sometimes, investigating performance problems, i find useful to have stats from jbd about transaction's lifetime, size, etc. here is a patch for review and inclusion probably. for example, stats after creation of 3M files in htree directory: [root@bob ~]# cat /proc/fs/jbd/sda/history R/C tid wait run lock flush log hndls block inlog ctime write drop close R 261 8260 2720 0 0 750 9892 8170 8187 C 259 750 0 4885 1 R 262 20 2200 10 0 770 9836 8170 8187 R 263 30 2200 10 0 3070 9812 8170 8187 R 264 0 5000 10 0 1340 0 0 0 C 261 8240 3212 4957 0 R 265 8260 1470 0 0 4640 9854 8170 8187 R 266 0 5000 10 0 1460 0 0 0 C 262 8210 2989 4868 0 R 267 8230 1490 10 0 4440 9875 8171 8188 R 268 0 5000 10 0 1260 0 0 0 C 263 7710 2937 4908 0 R 269 7730 1470 10 0 3330 9841 8170 8187 R 270 0 5000 10 0 830 0 0 0 C 265 8140 3234 4898 0 C 267 720 0 4849 1 R 271 8630 2740 20 0 740 9819 8170 8187 C 269 800 0 4214 1 R 272 40 2170 10 0 830 9716 8170 8187 R 273 40 2280 0 0 3530 9799 8170 8187 R 274 0 5000 10 0 990 0 0 0 where, R - line for transaction's life from T_RUNNING to T_FINISHED C - line for transaction's checkpointing tid - transaction's id wait - for how long we were waiting for new transaction to start (the longest period journal_start() took in this transaction) run - real transaction's lifetime (from T_RUNNING to T_LOCKED lock - how long we were waiting for all handles to close (time the transaction was in T_LOCKED) flush - how long it took to flush all data (data=ordered) log - how long it took to write the transaction to the log hndls - how many handles got to the transaction block - how many blocks got to the transaction inlog - how many blocks are written to the log (block + descriptors) ctime - how long it took to checkpoint the transaction write - how many blocks have been written during checkpointing drop - how many blocks have been dropped during checkpointing close - how many running transactions have been closed to checkpoint this one all times are in msec. [root@bob ~]# cat /proc/fs/jbd/sda/info 280 transaction, each upto 8192 blocks average: 1633ms waiting for transaction 3616ms running transaction 5ms transaction was being locked 1ms flushing data (in ordered mode) 1799ms logging transaction 11781 handles per transaction 5629 blocks per transaction 5641 logged blocks per transaction Signed-off-by: Johann Lombardi <johann.lombardi@bull.net> Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Mingming Cao <cmm@us.ibm.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com>
2008-01-29 04:58:27 +00:00
jbd2_remove_jbd_stats_proc_entry();
jbd2_journal_destroy_caches();
}
MODULE_LICENSE("GPL");
module_init(journal_init);
module_exit(journal_exit);