2006-01-16 16:50:04 +00:00
|
|
|
/*
|
|
|
|
|
* Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
|
2006-05-18 19:09:15 +00:00
|
|
|
* Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
|
2006-01-16 16:50:04 +00:00
|
|
|
*
|
|
|
|
|
* This copyrighted material is made available to anyone wishing to use,
|
|
|
|
|
* modify, copy, or redistribute it subject to the terms and conditions
|
2006-09-01 15:05:15 +00:00
|
|
|
* of the GNU General Public License version 2.
|
2006-01-16 16:50:04 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#include <linux/sched.h>
|
|
|
|
|
#include <linux/slab.h>
|
|
|
|
|
#include <linux/spinlock.h>
|
|
|
|
|
#include <linux/completion.h>
|
|
|
|
|
#include <linux/buffer_head.h>
|
2012-03-08 12:10:23 +00:00
|
|
|
#include <linux/mempool.h>
|
2006-02-27 22:23:27 +00:00
|
|
|
#include <linux/gfs2_ondisk.h>
|
2009-04-07 13:13:01 +00:00
|
|
|
#include <linux/bio.h>
|
|
|
|
|
#include <linux/fs.h>
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
#include "gfs2.h"
|
2006-02-27 22:23:27 +00:00
|
|
|
#include "incore.h"
|
[GFS2] assertion failure after writing to journaled file, umount
This patch passes all my nasty tests that were causing the code to
fail under one circumstance or another. Here is a complete summary
of all changes from today's git tree, in order of appearance:
1. There are now separate variables for metadata buffer accounting.
2. Variable sd_log_num_hdrs is no longer needed, since the header
accounting is taken care of by the reserve/refund sequence.
3. Fixed a tiny grammatical problem in a comment.
4. Added a new function "calc_reserved" to calculate the reserved
log space. This isn't entirely necessary, but it has two benefits:
First, it simplifies the gfs2_log_refund function greatly.
Second, it allows for easier debugging because I could sprinkle the
code with calls to this function to make sure the accounting is
proper (by adding asserts and printks) at strategic point of the code.
5. In log_pull_tail there apparently was a kludge to fix up the
accounting based on a "pull" parameter. The buffer accounting is
now done properly, so the kludge was removed.
6. File sync operations were making a call to gfs2_log_flush that
writes another journal header. Since that header was unplanned
for (reserved) by the reserve/refund sequence, the free space had
to be decremented so that when log_pull_tail gets called, the free
space is be adjusted properly. (Did I hear you call that a kludge?
well, maybe, but a lot more justifiable than the one I removed).
7. In the gfs2_log_shutdown code, it optionally syncs the log by
specifying the PULL parameter to log_write_header. I'm not sure
this is necessary anymore. It just seems to me there could be
cases where shutdown is called while there are outstanding log
buffers.
8. In the (data)buf_lo_before_commit functions, I changed some offset
values from being calculated on the fly to being constants. That
simplified some code and we might as well let the compiler do the
calculation once rather than redoing those cycles at run time.
9. This version has my rewritten databuf_lo_add function.
This version is much more like its predecessor, buf_lo_add, which
makes it easier to understand. Again, this might not be necessary,
but it seems as if this one works as well as the previous one,
maybe even better, so I decided to leave it in.
10. In databuf_lo_before_commit, a previous data corruption problem
was caused by going off the end of the buffer. The proper solution
is to have the proper limit in place, rather than stopping earlier.
(Thus my previous attempt to fix it is wrong).
If you don't wrap the buffer, you're stopping too early and that
causes more log buffer accounting problems.
11. In lops.h there are two new (previously mentioned) constants for
figuring out the data offset for the journal buffers.
12. There are also two new functions, buf_limit and databuf_limit to
calculate how many entries will fit in the buffer.
13. In function gfs2_meta_wipe, it needs to distinguish between pinned
metadata buffers and journaled data buffers for proper journal buffer
accounting. It can't use the JDATA gfs2_inode flag because it's
sometimes passed the "real" inode and sometimes the "metadata
inode" and the inode flags will be random bits in a metadata
gfs2_inode. It needs to base its decision on which was passed in.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2007-06-18 19:50:20 +00:00
|
|
|
#include "inode.h"
|
2006-01-16 16:50:04 +00:00
|
|
|
#include "glock.h"
|
|
|
|
|
#include "log.h"
|
|
|
|
|
#include "lops.h"
|
|
|
|
|
#include "meta_io.h"
|
|
|
|
|
#include "recovery.h"
|
|
|
|
|
#include "rgrp.h"
|
|
|
|
|
#include "trans.h"
|
2006-02-27 22:23:27 +00:00
|
|
|
#include "util.h"
|
2009-06-12 07:49:20 +00:00
|
|
|
#include "trace_gfs2.h"
|
2006-01-16 16:50:04 +00:00
|
|
|
|
2007-08-27 12:54:05 +00:00
|
|
|
/**
|
|
|
|
|
* gfs2_pin - Pin a buffer in memory
|
|
|
|
|
* @sdp: The superblock
|
|
|
|
|
* @bh: The buffer to be pinned
|
|
|
|
|
*
|
|
|
|
|
* The log lock must be held when calling this function
|
|
|
|
|
*/
|
|
|
|
|
static void gfs2_pin(struct gfs2_sbd *sdp, struct buffer_head *bh)
|
|
|
|
|
{
|
|
|
|
|
struct gfs2_bufdata *bd;
|
|
|
|
|
|
2011-03-30 15:33:25 +00:00
|
|
|
BUG_ON(!current->journal_info);
|
2007-08-27 12:54:05 +00:00
|
|
|
|
|
|
|
|
clear_buffer_dirty(bh);
|
|
|
|
|
if (test_set_buffer_pinned(bh))
|
|
|
|
|
gfs2_assert_withdraw(sdp, 0);
|
|
|
|
|
if (!buffer_uptodate(bh))
|
|
|
|
|
gfs2_io_error_bh(sdp, bh);
|
|
|
|
|
bd = bh->b_private;
|
|
|
|
|
/* If this buffer is in the AIL and it has already been written
|
|
|
|
|
* to in-place disk block, remove it from the AIL.
|
|
|
|
|
*/
|
2011-03-14 12:40:29 +00:00
|
|
|
spin_lock(&sdp->sd_ail_lock);
|
2007-08-27 12:54:05 +00:00
|
|
|
if (bd->bd_ail)
|
|
|
|
|
list_move(&bd->bd_ail_st_list, &bd->bd_ail->ai_ail2_list);
|
2011-03-14 12:40:29 +00:00
|
|
|
spin_unlock(&sdp->sd_ail_lock);
|
2007-08-27 12:54:05 +00:00
|
|
|
get_bh(bh);
|
GFS2: Various gfs2_logd improvements
This patch contains various tweaks to how log flushes and active item writeback
work. gfs2_logd is now managed by a waitqueue, and gfs2_log_reseve now waits
for gfs2_logd to do the log flushing. Multiple functions were rewritten to
remove the need to call gfs2_log_lock(). Instead of using one test to see if
gfs2_logd had work to do, there are now seperate tests to check if there
are two many buffers in the incore log or if there are two many items on the
active items list.
This patch is a port of a patch Steve Whitehouse wrote about a year ago, with
some minor changes. Since gfs2_ail1_start always submits all the active items,
it no longer needs to keep track of the first ai submitted, so this has been
removed. In gfs2_log_reserve(), the order of the calls to
prepare_to_wait_exclusive() and wake_up() when firing off the logd thread has
been switched. If it called wake_up first there was a small window for a race,
where logd could run and return before gfs2_log_reserve was ready to get woken
up. If gfs2_logd ran, but did not free up enough blocks, gfs2_log_reserve()
would be left waiting for gfs2_logd to eventualy run because it timed out.
Finally, gt_logd_secs, which controls how long to wait before gfs2_logd times
out, and flushes the log, can now be set on mount with ar_commit.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2010-05-04 19:29:16 +00:00
|
|
|
atomic_inc(&sdp->sd_log_pinned);
|
2009-06-12 07:49:20 +00:00
|
|
|
trace_gfs2_pin(bd, 1);
|
2007-08-27 12:54:05 +00:00
|
|
|
}
|
|
|
|
|
|
GFS2: Use rbtree for resource groups and clean up bitmap buffer ref count scheme
Here is an update of Bob's original rbtree patch which, in addition, also
resolves the rather strange ref counting that was being done relating to
the bitmap blocks.
Originally we had a dual system for journaling resource groups. The metadata
blocks were journaled and also the rgrp itself was added to a list. The reason
for adding the rgrp to the list in the journal was so that the "repolish
clones" code could be run to update the free space, and potentially send any
discard requests when the log was flushed. This was done by comparing the
"cloned" bitmap with what had been written back on disk during the transaction
commit.
Due to this, there was a requirement to hang on to the rgrps' bitmap buffers
until the journal had been flushed. For that reason, there was a rather
complicated set up in the ->go_lock ->go_unlock functions for rgrps involving
both a mutex and a spinlock (the ->sd_rindex_spin) to maintain a reference
count on the buffers.
However, the journal maintains a reference count on the buffers anyway, since
they are being journaled as metadata buffers. So by moving the code which deals
with the post-journal accounting for bitmap blocks to the metadata journaling
code, we can entirely dispense with the rather strange buffer ref counting
scheme and also the requirement to journal the rgrps.
The net result of all this is that the ->sd_rindex_spin is left to do exactly
one job, and that is to look after the rbtree or rgrps.
This patch is designed to be a stepping stone towards using RCU for the rbtree
of resource groups, however the reduction in the number of uses of the
->sd_rindex_spin is likely to have benefits for multi-threaded workloads,
anyway.
The patch retains ->go_lock and ->go_unlock for rgrps, however these maybe also
be removed in future in favour of calling the functions directly where required
in the code. That will allow locking of resource groups without needing to
actually read them in - something that could be useful in speeding up statfs.
In the mean time though it is valid to dereference ->bi_bh only when the rgrp
is locked. This is basically the same rule as before, modulo the references not
being valid until the following journal flush.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Cc: Benjamin Marzinski <bmarzins@redhat.com>
2011-08-31 08:53:19 +00:00
|
|
|
static bool buffer_is_rgrp(const struct gfs2_bufdata *bd)
|
|
|
|
|
{
|
|
|
|
|
return bd->bd_gl->gl_name.ln_type == LM_TYPE_RGRP;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void maybe_release_space(struct gfs2_bufdata *bd)
|
|
|
|
|
{
|
|
|
|
|
struct gfs2_glock *gl = bd->bd_gl;
|
|
|
|
|
struct gfs2_sbd *sdp = gl->gl_sbd;
|
|
|
|
|
struct gfs2_rgrpd *rgd = gl->gl_object;
|
|
|
|
|
unsigned int index = bd->bd_bh->b_blocknr - gl->gl_name.ln_number;
|
|
|
|
|
struct gfs2_bitmap *bi = rgd->rd_bits + index;
|
|
|
|
|
|
|
|
|
|
if (bi->bi_clone == 0)
|
|
|
|
|
return;
|
|
|
|
|
if (sdp->sd_args.ar_discard)
|
2012-02-08 12:58:32 +00:00
|
|
|
gfs2_rgrp_send_discards(sdp, rgd->rd_data0, bd->bd_bh, bi, 1, NULL);
|
GFS2: Use rbtree for resource groups and clean up bitmap buffer ref count scheme
Here is an update of Bob's original rbtree patch which, in addition, also
resolves the rather strange ref counting that was being done relating to
the bitmap blocks.
Originally we had a dual system for journaling resource groups. The metadata
blocks were journaled and also the rgrp itself was added to a list. The reason
for adding the rgrp to the list in the journal was so that the "repolish
clones" code could be run to update the free space, and potentially send any
discard requests when the log was flushed. This was done by comparing the
"cloned" bitmap with what had been written back on disk during the transaction
commit.
Due to this, there was a requirement to hang on to the rgrps' bitmap buffers
until the journal had been flushed. For that reason, there was a rather
complicated set up in the ->go_lock ->go_unlock functions for rgrps involving
both a mutex and a spinlock (the ->sd_rindex_spin) to maintain a reference
count on the buffers.
However, the journal maintains a reference count on the buffers anyway, since
they are being journaled as metadata buffers. So by moving the code which deals
with the post-journal accounting for bitmap blocks to the metadata journaling
code, we can entirely dispense with the rather strange buffer ref counting
scheme and also the requirement to journal the rgrps.
The net result of all this is that the ->sd_rindex_spin is left to do exactly
one job, and that is to look after the rbtree or rgrps.
This patch is designed to be a stepping stone towards using RCU for the rbtree
of resource groups, however the reduction in the number of uses of the
->sd_rindex_spin is likely to have benefits for multi-threaded workloads,
anyway.
The patch retains ->go_lock and ->go_unlock for rgrps, however these maybe also
be removed in future in favour of calling the functions directly where required
in the code. That will allow locking of resource groups without needing to
actually read them in - something that could be useful in speeding up statfs.
In the mean time though it is valid to dereference ->bi_bh only when the rgrp
is locked. This is basically the same rule as before, modulo the references not
being valid until the following journal flush.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Cc: Benjamin Marzinski <bmarzins@redhat.com>
2011-08-31 08:53:19 +00:00
|
|
|
memcpy(bi->bi_clone + bi->bi_offset,
|
|
|
|
|
bd->bd_bh->b_data + bi->bi_offset, bi->bi_len);
|
|
|
|
|
clear_bit(GBF_FULL, &bi->bi_flags);
|
|
|
|
|
rgd->rd_free_clone = rgd->rd_free;
|
|
|
|
|
}
|
|
|
|
|
|
2007-08-27 12:54:05 +00:00
|
|
|
/**
|
|
|
|
|
* gfs2_unpin - Unpin a buffer
|
|
|
|
|
* @sdp: the filesystem the buffer belongs to
|
|
|
|
|
* @bh: The buffer to unpin
|
|
|
|
|
* @ai:
|
2011-03-30 15:33:25 +00:00
|
|
|
* @flags: The inode dirty flags
|
2007-08-27 12:54:05 +00:00
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static void gfs2_unpin(struct gfs2_sbd *sdp, struct buffer_head *bh,
|
|
|
|
|
struct gfs2_ail *ai)
|
|
|
|
|
{
|
|
|
|
|
struct gfs2_bufdata *bd = bh->b_private;
|
|
|
|
|
|
2011-03-30 15:33:25 +00:00
|
|
|
BUG_ON(!buffer_uptodate(bh));
|
|
|
|
|
BUG_ON(!buffer_pinned(bh));
|
2007-08-27 12:54:05 +00:00
|
|
|
|
|
|
|
|
lock_buffer(bh);
|
|
|
|
|
mark_buffer_dirty(bh);
|
|
|
|
|
clear_buffer_pinned(bh);
|
|
|
|
|
|
GFS2: Use rbtree for resource groups and clean up bitmap buffer ref count scheme
Here is an update of Bob's original rbtree patch which, in addition, also
resolves the rather strange ref counting that was being done relating to
the bitmap blocks.
Originally we had a dual system for journaling resource groups. The metadata
blocks were journaled and also the rgrp itself was added to a list. The reason
for adding the rgrp to the list in the journal was so that the "repolish
clones" code could be run to update the free space, and potentially send any
discard requests when the log was flushed. This was done by comparing the
"cloned" bitmap with what had been written back on disk during the transaction
commit.
Due to this, there was a requirement to hang on to the rgrps' bitmap buffers
until the journal had been flushed. For that reason, there was a rather
complicated set up in the ->go_lock ->go_unlock functions for rgrps involving
both a mutex and a spinlock (the ->sd_rindex_spin) to maintain a reference
count on the buffers.
However, the journal maintains a reference count on the buffers anyway, since
they are being journaled as metadata buffers. So by moving the code which deals
with the post-journal accounting for bitmap blocks to the metadata journaling
code, we can entirely dispense with the rather strange buffer ref counting
scheme and also the requirement to journal the rgrps.
The net result of all this is that the ->sd_rindex_spin is left to do exactly
one job, and that is to look after the rbtree or rgrps.
This patch is designed to be a stepping stone towards using RCU for the rbtree
of resource groups, however the reduction in the number of uses of the
->sd_rindex_spin is likely to have benefits for multi-threaded workloads,
anyway.
The patch retains ->go_lock and ->go_unlock for rgrps, however these maybe also
be removed in future in favour of calling the functions directly where required
in the code. That will allow locking of resource groups without needing to
actually read them in - something that could be useful in speeding up statfs.
In the mean time though it is valid to dereference ->bi_bh only when the rgrp
is locked. This is basically the same rule as before, modulo the references not
being valid until the following journal flush.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Cc: Benjamin Marzinski <bmarzins@redhat.com>
2011-08-31 08:53:19 +00:00
|
|
|
if (buffer_is_rgrp(bd))
|
|
|
|
|
maybe_release_space(bd);
|
|
|
|
|
|
2011-03-11 11:52:25 +00:00
|
|
|
spin_lock(&sdp->sd_ail_lock);
|
2007-08-27 12:54:05 +00:00
|
|
|
if (bd->bd_ail) {
|
|
|
|
|
list_del(&bd->bd_ail_st_list);
|
|
|
|
|
brelse(bh);
|
|
|
|
|
} else {
|
|
|
|
|
struct gfs2_glock *gl = bd->bd_gl;
|
|
|
|
|
list_add(&bd->bd_ail_gl_list, &gl->gl_ail_list);
|
|
|
|
|
atomic_inc(&gl->gl_ail_count);
|
|
|
|
|
}
|
|
|
|
|
bd->bd_ail = ai;
|
|
|
|
|
list_add(&bd->bd_ail_st_list, &ai->ai_ail1_list);
|
2011-03-11 11:52:25 +00:00
|
|
|
spin_unlock(&sdp->sd_ail_lock);
|
|
|
|
|
|
2011-03-30 15:33:25 +00:00
|
|
|
clear_bit(GLF_LFLUSH, &bd->bd_gl->gl_flags);
|
2009-06-12 07:49:20 +00:00
|
|
|
trace_gfs2_pin(bd, 0);
|
2007-08-27 12:54:05 +00:00
|
|
|
unlock_buffer(bh);
|
GFS2: Various gfs2_logd improvements
This patch contains various tweaks to how log flushes and active item writeback
work. gfs2_logd is now managed by a waitqueue, and gfs2_log_reseve now waits
for gfs2_logd to do the log flushing. Multiple functions were rewritten to
remove the need to call gfs2_log_lock(). Instead of using one test to see if
gfs2_logd had work to do, there are now seperate tests to check if there
are two many buffers in the incore log or if there are two many items on the
active items list.
This patch is a port of a patch Steve Whitehouse wrote about a year ago, with
some minor changes. Since gfs2_ail1_start always submits all the active items,
it no longer needs to keep track of the first ai submitted, so this has been
removed. In gfs2_log_reserve(), the order of the calls to
prepare_to_wait_exclusive() and wake_up() when firing off the logd thread has
been switched. If it called wake_up first there was a small window for a race,
where logd could run and return before gfs2_log_reserve was ready to get woken
up. If gfs2_logd ran, but did not free up enough blocks, gfs2_log_reserve()
would be left waiting for gfs2_logd to eventualy run because it timed out.
Finally, gt_logd_secs, which controls how long to wait before gfs2_logd times
out, and flushes the log, can now be set on mount with ar_commit.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2010-05-04 19:29:16 +00:00
|
|
|
atomic_dec(&sdp->sd_log_pinned);
|
2007-08-27 12:54:05 +00:00
|
|
|
}
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
static void gfs2_log_incr_head(struct gfs2_sbd *sdp)
|
2007-09-17 09:59:52 +00:00
|
|
|
{
|
2012-04-16 08:28:31 +00:00
|
|
|
BUG_ON((sdp->sd_log_flush_head == sdp->sd_log_tail) &&
|
|
|
|
|
(sdp->sd_log_flush_head != sdp->sd_log_head));
|
|
|
|
|
|
|
|
|
|
if (++sdp->sd_log_flush_head == sdp->sd_jdesc->jd_blocks) {
|
|
|
|
|
sdp->sd_log_flush_head = 0;
|
|
|
|
|
sdp->sd_log_flush_wrapped = 1;
|
|
|
|
|
}
|
2007-09-17 09:59:52 +00:00
|
|
|
}
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
static u64 gfs2_log_bmap(struct gfs2_sbd *sdp)
|
2007-09-17 09:59:52 +00:00
|
|
|
{
|
2012-04-16 08:28:31 +00:00
|
|
|
unsigned int lbn = sdp->sd_log_flush_head;
|
|
|
|
|
struct gfs2_journal_extent *je;
|
|
|
|
|
u64 block;
|
|
|
|
|
|
|
|
|
|
list_for_each_entry(je, &sdp->sd_jdesc->extent_list, extent_list) {
|
|
|
|
|
if (lbn >= je->lblock && lbn < je->lblock + je->blocks) {
|
|
|
|
|
block = je->dblock + lbn - je->lblock;
|
|
|
|
|
gfs2_log_incr_head(sdp);
|
|
|
|
|
return block;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return -1;
|
2007-09-17 09:59:52 +00:00
|
|
|
}
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
/**
|
|
|
|
|
* gfs2_end_log_write_bh - end log write of pagecache data with buffers
|
|
|
|
|
* @sdp: The superblock
|
|
|
|
|
* @bvec: The bio_vec
|
|
|
|
|
* @error: The i/o status
|
|
|
|
|
*
|
|
|
|
|
* This finds the relavent buffers and unlocks then and sets the
|
|
|
|
|
* error flag according to the status of the i/o request. This is
|
|
|
|
|
* used when the log is writing data which has an in-place version
|
|
|
|
|
* that is pinned in the pagecache.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static void gfs2_end_log_write_bh(struct gfs2_sbd *sdp, struct bio_vec *bvec,
|
|
|
|
|
int error)
|
2007-09-17 09:59:52 +00:00
|
|
|
{
|
2012-04-16 08:28:31 +00:00
|
|
|
struct buffer_head *bh, *next;
|
|
|
|
|
struct page *page = bvec->bv_page;
|
|
|
|
|
unsigned size;
|
|
|
|
|
|
|
|
|
|
bh = page_buffers(page);
|
|
|
|
|
size = bvec->bv_len;
|
|
|
|
|
while (bh_offset(bh) < bvec->bv_offset)
|
|
|
|
|
bh = bh->b_this_page;
|
|
|
|
|
do {
|
|
|
|
|
if (error)
|
|
|
|
|
set_buffer_write_io_error(bh);
|
|
|
|
|
unlock_buffer(bh);
|
|
|
|
|
next = bh->b_this_page;
|
|
|
|
|
size -= bh->b_size;
|
|
|
|
|
brelse(bh);
|
|
|
|
|
bh = next;
|
|
|
|
|
} while(bh && size);
|
2007-09-17 09:59:52 +00:00
|
|
|
}
|
|
|
|
|
|
2012-02-03 15:21:59 +00:00
|
|
|
/**
|
2012-04-16 08:28:31 +00:00
|
|
|
* gfs2_end_log_write - end of i/o to the log
|
|
|
|
|
* @bio: The bio
|
|
|
|
|
* @error: Status of i/o request
|
|
|
|
|
*
|
|
|
|
|
* Each bio_vec contains either data from the pagecache or data
|
|
|
|
|
* relating to the log itself. Here we iterate over the bio_vec
|
|
|
|
|
* array, processing both kinds of data.
|
2012-02-03 15:21:59 +00:00
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
static void gfs2_end_log_write(struct bio *bio, int error)
|
2012-02-03 15:21:59 +00:00
|
|
|
{
|
2012-04-16 08:28:31 +00:00
|
|
|
struct gfs2_sbd *sdp = bio->bi_private;
|
|
|
|
|
struct bio_vec *bvec;
|
|
|
|
|
struct page *page;
|
|
|
|
|
int i;
|
|
|
|
|
|
|
|
|
|
if (error) {
|
|
|
|
|
sdp->sd_log_error = error;
|
|
|
|
|
fs_err(sdp, "Error %d writing to log\n", error);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bio_for_each_segment(bvec, bio, i) {
|
|
|
|
|
page = bvec->bv_page;
|
|
|
|
|
if (page_has_buffers(page))
|
|
|
|
|
gfs2_end_log_write_bh(sdp, bvec, error);
|
|
|
|
|
else
|
|
|
|
|
mempool_free(page, gfs2_page_pool);
|
|
|
|
|
}
|
2012-02-03 15:21:59 +00:00
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
bio_put(bio);
|
2012-02-03 15:21:59 +00:00
|
|
|
if (atomic_dec_and_test(&sdp->sd_log_in_flight))
|
|
|
|
|
wake_up(&sdp->sd_log_flush_wait);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
2012-04-16 08:28:31 +00:00
|
|
|
* gfs2_log_flush_bio - Submit any pending log bio
|
|
|
|
|
* @sdp: The superblock
|
|
|
|
|
* @rw: The rw flags
|
2012-02-03 15:21:59 +00:00
|
|
|
*
|
2012-04-16 08:28:31 +00:00
|
|
|
* Submit any pending part-built or full bio to the block device. If
|
|
|
|
|
* there is no pending bio, then this is a no-op.
|
2012-02-03 15:21:59 +00:00
|
|
|
*/
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
void gfs2_log_flush_bio(struct gfs2_sbd *sdp, int rw)
|
2012-02-03 15:21:59 +00:00
|
|
|
{
|
2012-04-16 08:28:31 +00:00
|
|
|
if (sdp->sd_log_bio) {
|
|
|
|
|
atomic_inc(&sdp->sd_log_in_flight);
|
|
|
|
|
submit_bio(rw, sdp->sd_log_bio);
|
|
|
|
|
sdp->sd_log_bio = NULL;
|
|
|
|
|
}
|
|
|
|
|
}
|
2012-02-03 15:21:59 +00:00
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
/**
|
|
|
|
|
* gfs2_log_alloc_bio - Allocate a new bio for log writing
|
|
|
|
|
* @sdp: The superblock
|
|
|
|
|
* @blkno: The next device block number we want to write to
|
|
|
|
|
*
|
|
|
|
|
* This should never be called when there is a cached bio in the
|
|
|
|
|
* super block. When it returns, there will be a cached bio in the
|
|
|
|
|
* super block which will have as many bio_vecs as the device is
|
|
|
|
|
* happy to handle.
|
|
|
|
|
*
|
|
|
|
|
* Returns: Newly allocated bio
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static struct bio *gfs2_log_alloc_bio(struct gfs2_sbd *sdp, u64 blkno)
|
|
|
|
|
{
|
|
|
|
|
struct super_block *sb = sdp->sd_vfs;
|
|
|
|
|
unsigned nrvecs = bio_get_nr_vecs(sb->s_bdev);
|
|
|
|
|
struct bio *bio;
|
|
|
|
|
|
|
|
|
|
BUG_ON(sdp->sd_log_bio);
|
|
|
|
|
|
|
|
|
|
while (1) {
|
|
|
|
|
bio = bio_alloc(GFP_NOIO, nrvecs);
|
|
|
|
|
if (likely(bio))
|
|
|
|
|
break;
|
|
|
|
|
nrvecs = max(nrvecs/2, 1U);
|
|
|
|
|
}
|
2012-02-03 15:21:59 +00:00
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
bio->bi_sector = blkno * (sb->s_blocksize >> 9);
|
|
|
|
|
bio->bi_bdev = sb->s_bdev;
|
|
|
|
|
bio->bi_end_io = gfs2_end_log_write;
|
|
|
|
|
bio->bi_private = sdp;
|
|
|
|
|
|
|
|
|
|
sdp->sd_log_bio = bio;
|
|
|
|
|
|
|
|
|
|
return bio;
|
2012-02-03 15:21:59 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
2012-04-16 08:28:31 +00:00
|
|
|
* gfs2_log_get_bio - Get cached log bio, or allocate a new one
|
|
|
|
|
* @sdp: The superblock
|
|
|
|
|
* @blkno: The device block number we want to write to
|
|
|
|
|
*
|
|
|
|
|
* If there is a cached bio, then if the next block number is sequential
|
|
|
|
|
* with the previous one, return it, otherwise flush the bio to the
|
|
|
|
|
* device. If there is not a cached bio, or we just flushed it, then
|
|
|
|
|
* allocate a new one.
|
2012-02-03 15:21:59 +00:00
|
|
|
*
|
2012-04-16 08:28:31 +00:00
|
|
|
* Returns: The bio to use for log writes
|
2012-02-03 15:21:59 +00:00
|
|
|
*/
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
static struct bio *gfs2_log_get_bio(struct gfs2_sbd *sdp, u64 blkno)
|
2012-02-03 15:21:59 +00:00
|
|
|
{
|
2012-04-16 08:28:31 +00:00
|
|
|
struct bio *bio = sdp->sd_log_bio;
|
|
|
|
|
u64 nblk;
|
|
|
|
|
|
|
|
|
|
if (bio) {
|
|
|
|
|
nblk = bio->bi_sector + bio_sectors(bio);
|
|
|
|
|
nblk >>= sdp->sd_fsb2bb_shift;
|
|
|
|
|
if (blkno == nblk)
|
|
|
|
|
return bio;
|
|
|
|
|
gfs2_log_flush_bio(sdp, WRITE);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return gfs2_log_alloc_bio(sdp, blkno);
|
2012-02-03 15:21:59 +00:00
|
|
|
}
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
|
2012-02-03 15:21:59 +00:00
|
|
|
/**
|
2012-04-16 08:28:31 +00:00
|
|
|
* gfs2_log_write - write to log
|
2012-02-03 15:21:59 +00:00
|
|
|
* @sdp: the filesystem
|
2012-04-16 08:28:31 +00:00
|
|
|
* @page: the page to write
|
|
|
|
|
* @size: the size of the data to write
|
|
|
|
|
* @offset: the offset within the page
|
2012-02-03 15:21:59 +00:00
|
|
|
*
|
2012-04-16 08:28:31 +00:00
|
|
|
* Try and add the page segment to the current bio. If that fails,
|
|
|
|
|
* submit the current bio to the device and create a new one, and
|
|
|
|
|
* then add the page segment to that.
|
2012-02-03 15:21:59 +00:00
|
|
|
*/
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
static void gfs2_log_write(struct gfs2_sbd *sdp, struct page *page,
|
|
|
|
|
unsigned size, unsigned offset)
|
2012-02-03 15:21:59 +00:00
|
|
|
{
|
2012-04-16 08:28:31 +00:00
|
|
|
u64 blkno = gfs2_log_bmap(sdp);
|
|
|
|
|
struct bio *bio;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
bio = gfs2_log_get_bio(sdp, blkno);
|
|
|
|
|
ret = bio_add_page(bio, page, size, offset);
|
|
|
|
|
if (ret == 0) {
|
|
|
|
|
gfs2_log_flush_bio(sdp, WRITE);
|
|
|
|
|
bio = gfs2_log_alloc_bio(sdp, blkno);
|
|
|
|
|
ret = bio_add_page(bio, page, size, offset);
|
|
|
|
|
WARN_ON(ret == 0);
|
|
|
|
|
}
|
|
|
|
|
}
|
2012-02-03 15:21:59 +00:00
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
/**
|
|
|
|
|
* gfs2_log_write_bh - write a buffer's content to the log
|
|
|
|
|
* @sdp: The super block
|
|
|
|
|
* @bh: The buffer pointing to the in-place location
|
|
|
|
|
*
|
|
|
|
|
* This writes the content of the buffer to the next available location
|
|
|
|
|
* in the log. The buffer will be unlocked once the i/o to the log has
|
|
|
|
|
* completed.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static void gfs2_log_write_bh(struct gfs2_sbd *sdp, struct buffer_head *bh)
|
|
|
|
|
{
|
|
|
|
|
gfs2_log_write(sdp, bh->b_page, bh->b_size, bh_offset(bh));
|
|
|
|
|
}
|
2012-02-03 15:21:59 +00:00
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
/**
|
|
|
|
|
* gfs2_log_write_page - write one block stored in a page, into the log
|
|
|
|
|
* @sdp: The superblock
|
|
|
|
|
* @page: The struct page
|
|
|
|
|
*
|
|
|
|
|
* This writes the first block-sized part of the page into the log. Note
|
|
|
|
|
* that the page must have been allocated from the gfs2_page_pool mempool
|
|
|
|
|
* and that after this has been called, ownership has been transferred and
|
|
|
|
|
* the page may be freed at any time.
|
|
|
|
|
*/
|
2012-02-03 15:21:59 +00:00
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
void gfs2_log_write_page(struct gfs2_sbd *sdp, struct page *page)
|
|
|
|
|
{
|
|
|
|
|
struct super_block *sb = sdp->sd_vfs;
|
|
|
|
|
gfs2_log_write(sdp, page, sb->s_blocksize, 0);
|
2012-02-03 15:21:59 +00:00
|
|
|
}
|
2007-09-17 09:59:52 +00:00
|
|
|
|
2012-04-16 08:40:00 +00:00
|
|
|
static struct page *gfs2_get_log_desc(struct gfs2_sbd *sdp, u32 ld_type,
|
|
|
|
|
u32 ld_length, u32 ld_data1)
|
2007-09-17 09:59:52 +00:00
|
|
|
{
|
2012-04-16 08:28:31 +00:00
|
|
|
void *page = mempool_alloc(gfs2_page_pool, GFP_NOIO);
|
|
|
|
|
struct gfs2_log_descriptor *ld = page_address(page);
|
|
|
|
|
clear_page(ld);
|
2007-09-17 09:59:52 +00:00
|
|
|
ld->ld_header.mh_magic = cpu_to_be32(GFS2_MAGIC);
|
|
|
|
|
ld->ld_header.mh_type = cpu_to_be32(GFS2_METATYPE_LD);
|
|
|
|
|
ld->ld_header.mh_format = cpu_to_be32(GFS2_FORMAT_LD);
|
|
|
|
|
ld->ld_type = cpu_to_be32(ld_type);
|
2012-04-16 08:40:00 +00:00
|
|
|
ld->ld_length = cpu_to_be32(ld_length);
|
|
|
|
|
ld->ld_data1 = cpu_to_be32(ld_data1);
|
2007-09-17 09:59:52 +00:00
|
|
|
ld->ld_data2 = 0;
|
2012-04-16 08:28:31 +00:00
|
|
|
return page;
|
2007-09-17 09:59:52 +00:00
|
|
|
}
|
|
|
|
|
|
2006-01-16 16:50:04 +00:00
|
|
|
static void buf_lo_add(struct gfs2_sbd *sdp, struct gfs2_log_element *le)
|
|
|
|
|
{
|
|
|
|
|
struct gfs2_bufdata *bd = container_of(le, struct gfs2_bufdata, bd_le);
|
GFS2: Tag all metadata with jid
There are two spare field in the header common to all GFS2
metadata. One is just the right size to fit a journal id
in it, and this patch updates the journal code so that each
time a metadata block is modified, we tag it with the journal
id of the node which is performing the modification.
The reason for this is that it should make it much easier to
debug issues which arise if we can tell which node was the
last to modify a particular metadata block.
Since the field is updated before the block is written into
the journal, each journal should only contain metadata which
is tagged with its own journal id. The one exception to this
is the journal header block, which might have a different node's
id in it, if that journal was recovered by another node in the
cluster.
Thus each journal will contain a record of which nodes recovered
it, via the journal header.
The other field in the metadata header could potentially be
used to hold information about what kind of operation was
performed, but for the time being we just zero it on each
transaction so that if we use it for that in future, we'll
know that the information (where it exists) is reliable.
I did consider using the other field to hold the journal
sequence number, however since in GFS2's journaling we write
the modified data into the journal and not the original
data, this gives no information as to what action caused the
modification, so I think we can probably come up with a better
use for those 64 bits in the future.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2009-11-06 16:20:51 +00:00
|
|
|
struct gfs2_meta_header *mh;
|
2006-01-16 16:50:04 +00:00
|
|
|
struct gfs2_trans *tr;
|
|
|
|
|
|
2007-08-27 12:54:05 +00:00
|
|
|
lock_buffer(bd->bd_bh);
|
2007-01-25 10:04:20 +00:00
|
|
|
gfs2_log_lock(sdp);
|
2007-08-27 12:54:05 +00:00
|
|
|
if (!list_empty(&bd->bd_list_tr))
|
|
|
|
|
goto out;
|
2006-02-27 22:23:27 +00:00
|
|
|
tr = current->journal_info;
|
2006-01-16 16:50:04 +00:00
|
|
|
tr->tr_touched = 1;
|
|
|
|
|
tr->tr_num_buf++;
|
|
|
|
|
list_add(&bd->bd_list_tr, &tr->tr_list_buf);
|
|
|
|
|
if (!list_empty(&le->le_list))
|
2007-08-27 12:54:05 +00:00
|
|
|
goto out;
|
2007-11-08 14:25:12 +00:00
|
|
|
set_bit(GLF_LFLUSH, &bd->bd_gl->gl_flags);
|
|
|
|
|
set_bit(GLF_DIRTY, &bd->bd_gl->gl_flags);
|
2006-01-16 16:50:04 +00:00
|
|
|
gfs2_meta_check(sdp, bd->bd_bh);
|
2006-01-18 13:38:44 +00:00
|
|
|
gfs2_pin(sdp, bd->bd_bh);
|
GFS2: Tag all metadata with jid
There are two spare field in the header common to all GFS2
metadata. One is just the right size to fit a journal id
in it, and this patch updates the journal code so that each
time a metadata block is modified, we tag it with the journal
id of the node which is performing the modification.
The reason for this is that it should make it much easier to
debug issues which arise if we can tell which node was the
last to modify a particular metadata block.
Since the field is updated before the block is written into
the journal, each journal should only contain metadata which
is tagged with its own journal id. The one exception to this
is the journal header block, which might have a different node's
id in it, if that journal was recovered by another node in the
cluster.
Thus each journal will contain a record of which nodes recovered
it, via the journal header.
The other field in the metadata header could potentially be
used to hold information about what kind of operation was
performed, but for the time being we just zero it on each
transaction so that if we use it for that in future, we'll
know that the information (where it exists) is reliable.
I did consider using the other field to hold the journal
sequence number, however since in GFS2's journaling we write
the modified data into the journal and not the original
data, this gives no information as to what action caused the
modification, so I think we can probably come up with a better
use for those 64 bits in the future.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2009-11-06 16:20:51 +00:00
|
|
|
mh = (struct gfs2_meta_header *)bd->bd_bh->b_data;
|
|
|
|
|
mh->__pad0 = cpu_to_be64(0);
|
|
|
|
|
mh->mh_jid = cpu_to_be32(sdp->sd_jdesc->jd_jid);
|
2006-01-16 16:50:04 +00:00
|
|
|
sdp->sd_log_num_buf++;
|
|
|
|
|
list_add(&le->le_list, &sdp->sd_log_le_buf);
|
|
|
|
|
tr->tr_num_buf_new++;
|
2007-08-27 12:54:05 +00:00
|
|
|
out:
|
|
|
|
|
gfs2_log_unlock(sdp);
|
|
|
|
|
unlock_buffer(bd->bd_bh);
|
2006-01-16 16:50:04 +00:00
|
|
|
}
|
|
|
|
|
|
2012-04-16 08:40:00 +00:00
|
|
|
static void gfs2_check_magic(struct buffer_head *bh)
|
|
|
|
|
{
|
|
|
|
|
void *kaddr;
|
|
|
|
|
__be32 *ptr;
|
|
|
|
|
|
|
|
|
|
clear_buffer_escaped(bh);
|
|
|
|
|
kaddr = kmap_atomic(bh->b_page);
|
|
|
|
|
ptr = kaddr + bh_offset(bh);
|
|
|
|
|
if (*ptr == cpu_to_be32(GFS2_MAGIC))
|
|
|
|
|
set_buffer_escaped(bh);
|
|
|
|
|
kunmap_atomic(kaddr);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void gfs2_before_commit(struct gfs2_sbd *sdp, unsigned int limit,
|
|
|
|
|
unsigned int total, struct list_head *blist,
|
|
|
|
|
bool is_databuf)
|
2006-01-16 16:50:04 +00:00
|
|
|
{
|
|
|
|
|
struct gfs2_log_descriptor *ld;
|
|
|
|
|
struct gfs2_bufdata *bd1 = NULL, *bd2;
|
2012-04-16 08:28:31 +00:00
|
|
|
struct page *page;
|
2006-01-16 16:50:04 +00:00
|
|
|
unsigned int num;
|
|
|
|
|
unsigned n;
|
|
|
|
|
__be64 *ptr;
|
|
|
|
|
|
2007-07-24 19:05:31 +00:00
|
|
|
gfs2_log_lock(sdp);
|
2012-04-16 08:40:00 +00:00
|
|
|
bd1 = bd2 = list_prepare_entry(bd1, blist, bd_le.le_list);
|
2006-01-16 16:50:04 +00:00
|
|
|
while(total) {
|
|
|
|
|
num = total;
|
|
|
|
|
if (total > limit)
|
|
|
|
|
num = limit;
|
2007-07-24 19:05:31 +00:00
|
|
|
gfs2_log_unlock(sdp);
|
2012-04-16 08:40:00 +00:00
|
|
|
page = gfs2_get_log_desc(sdp, GFS2_LOG_DESC_METADATA, num + 1, num);
|
2012-04-16 08:28:31 +00:00
|
|
|
ld = page_address(page);
|
2007-07-24 19:05:31 +00:00
|
|
|
gfs2_log_lock(sdp);
|
2012-04-16 08:28:31 +00:00
|
|
|
ptr = (__be64 *)(ld + 1);
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
n = 0;
|
2012-04-16 08:40:00 +00:00
|
|
|
list_for_each_entry_continue(bd1, blist, bd_le.le_list) {
|
2006-01-16 16:50:04 +00:00
|
|
|
*ptr++ = cpu_to_be64(bd1->bd_bh->b_blocknr);
|
2012-04-16 08:40:00 +00:00
|
|
|
if (is_databuf) {
|
|
|
|
|
gfs2_check_magic(bd1->bd_bh);
|
|
|
|
|
*ptr++ = cpu_to_be64(buffer_escaped(bd1->bd_bh) ? 1 : 0);
|
|
|
|
|
}
|
2006-01-16 16:50:04 +00:00
|
|
|
if (++n >= num)
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
2007-07-24 19:05:31 +00:00
|
|
|
gfs2_log_unlock(sdp);
|
2012-04-16 08:28:31 +00:00
|
|
|
gfs2_log_write_page(sdp, page);
|
2007-07-24 19:05:31 +00:00
|
|
|
gfs2_log_lock(sdp);
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
n = 0;
|
2012-04-16 08:40:00 +00:00
|
|
|
list_for_each_entry_continue(bd2, blist, bd_le.le_list) {
|
2007-09-17 09:59:52 +00:00
|
|
|
get_bh(bd2->bd_bh);
|
2007-07-24 19:05:31 +00:00
|
|
|
gfs2_log_unlock(sdp);
|
2007-09-17 09:59:52 +00:00
|
|
|
lock_buffer(bd2->bd_bh);
|
2012-04-16 08:40:00 +00:00
|
|
|
|
|
|
|
|
if (buffer_escaped(bd2->bd_bh)) {
|
|
|
|
|
void *kaddr;
|
|
|
|
|
page = mempool_alloc(gfs2_page_pool, GFP_NOIO);
|
|
|
|
|
ptr = page_address(page);
|
|
|
|
|
kaddr = kmap_atomic(bd2->bd_bh->b_page);
|
|
|
|
|
memcpy(ptr, kaddr + bh_offset(bd2->bd_bh),
|
|
|
|
|
bd2->bd_bh->b_size);
|
|
|
|
|
kunmap_atomic(kaddr);
|
|
|
|
|
*(__be32 *)ptr = 0;
|
|
|
|
|
clear_buffer_escaped(bd2->bd_bh);
|
|
|
|
|
unlock_buffer(bd2->bd_bh);
|
|
|
|
|
brelse(bd2->bd_bh);
|
|
|
|
|
gfs2_log_write_page(sdp, page);
|
|
|
|
|
} else {
|
|
|
|
|
gfs2_log_write_bh(sdp, bd2->bd_bh);
|
|
|
|
|
}
|
2007-07-24 19:05:31 +00:00
|
|
|
gfs2_log_lock(sdp);
|
2006-01-16 16:50:04 +00:00
|
|
|
if (++n >= num)
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
2007-07-24 19:05:31 +00:00
|
|
|
BUG_ON(total < num);
|
2006-01-16 16:50:04 +00:00
|
|
|
total -= num;
|
|
|
|
|
}
|
2007-07-24 19:05:31 +00:00
|
|
|
gfs2_log_unlock(sdp);
|
2006-01-16 16:50:04 +00:00
|
|
|
}
|
|
|
|
|
|
2012-04-16 08:40:00 +00:00
|
|
|
static void buf_lo_before_commit(struct gfs2_sbd *sdp)
|
|
|
|
|
{
|
|
|
|
|
unsigned int limit = buf_limit(sdp); /* 503 for 4k blocks */
|
|
|
|
|
|
|
|
|
|
gfs2_before_commit(sdp, limit, sdp->sd_log_num_buf,
|
|
|
|
|
&sdp->sd_log_le_buf, 0);
|
|
|
|
|
}
|
|
|
|
|
|
2006-01-16 16:50:04 +00:00
|
|
|
static void buf_lo_after_commit(struct gfs2_sbd *sdp, struct gfs2_ail *ai)
|
|
|
|
|
{
|
|
|
|
|
struct list_head *head = &sdp->sd_log_le_buf;
|
|
|
|
|
struct gfs2_bufdata *bd;
|
|
|
|
|
|
|
|
|
|
while (!list_empty(head)) {
|
|
|
|
|
bd = list_entry(head->next, struct gfs2_bufdata, bd_le.le_list);
|
|
|
|
|
list_del_init(&bd->bd_le.le_list);
|
|
|
|
|
sdp->sd_log_num_buf--;
|
|
|
|
|
|
2006-01-18 13:38:44 +00:00
|
|
|
gfs2_unpin(sdp, bd->bd_bh, ai);
|
2006-01-16 16:50:04 +00:00
|
|
|
}
|
|
|
|
|
gfs2_assert_warn(sdp, !sdp->sd_log_num_buf);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void buf_lo_before_scan(struct gfs2_jdesc *jd,
|
2006-10-14 01:47:13 +00:00
|
|
|
struct gfs2_log_header_host *head, int pass)
|
2006-01-16 16:50:04 +00:00
|
|
|
{
|
2006-06-14 19:32:57 +00:00
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
if (pass != 0)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
sdp->sd_found_blocks = 0;
|
|
|
|
|
sdp->sd_replayed_blocks = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int buf_lo_scan_elements(struct gfs2_jdesc *jd, unsigned int start,
|
|
|
|
|
struct gfs2_log_descriptor *ld, __be64 *ptr,
|
|
|
|
|
int pass)
|
|
|
|
|
{
|
2006-06-14 19:32:57 +00:00
|
|
|
struct gfs2_inode *ip = GFS2_I(jd->jd_inode);
|
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-02-27 22:23:27 +00:00
|
|
|
struct gfs2_glock *gl = ip->i_gl;
|
2006-01-16 16:50:04 +00:00
|
|
|
unsigned int blks = be32_to_cpu(ld->ld_data1);
|
|
|
|
|
struct buffer_head *bh_log, *bh_ip;
|
2006-09-04 16:49:07 +00:00
|
|
|
u64 blkno;
|
2006-01-16 16:50:04 +00:00
|
|
|
int error = 0;
|
|
|
|
|
|
|
|
|
|
if (pass != 1 || be32_to_cpu(ld->ld_type) != GFS2_LOG_DESC_METADATA)
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
gfs2_replay_incr_blk(sdp, &start);
|
|
|
|
|
|
|
|
|
|
for (; blks; gfs2_replay_incr_blk(sdp, &start), blks--) {
|
|
|
|
|
blkno = be64_to_cpu(*ptr++);
|
|
|
|
|
|
|
|
|
|
sdp->sd_found_blocks++;
|
|
|
|
|
|
|
|
|
|
if (gfs2_revoke_check(sdp, blkno, start))
|
|
|
|
|
continue;
|
|
|
|
|
|
|
|
|
|
error = gfs2_replay_read_block(jd, start, &bh_log);
|
2006-09-04 18:47:06 +00:00
|
|
|
if (error)
|
|
|
|
|
return error;
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
bh_ip = gfs2_meta_new(gl, blkno);
|
|
|
|
|
memcpy(bh_ip->b_data, bh_log->b_data, bh_log->b_size);
|
|
|
|
|
|
|
|
|
|
if (gfs2_meta_check(sdp, bh_ip))
|
|
|
|
|
error = -EIO;
|
|
|
|
|
else
|
|
|
|
|
mark_buffer_dirty(bh_ip);
|
|
|
|
|
|
|
|
|
|
brelse(bh_log);
|
|
|
|
|
brelse(bh_ip);
|
|
|
|
|
|
|
|
|
|
if (error)
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
sdp->sd_replayed_blocks++;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return error;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void buf_lo_after_scan(struct gfs2_jdesc *jd, int error, int pass)
|
|
|
|
|
{
|
2006-06-14 19:32:57 +00:00
|
|
|
struct gfs2_inode *ip = GFS2_I(jd->jd_inode);
|
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
if (error) {
|
2006-09-21 21:05:23 +00:00
|
|
|
gfs2_meta_sync(ip->i_gl);
|
2006-01-16 16:50:04 +00:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (pass != 1)
|
|
|
|
|
return;
|
|
|
|
|
|
2006-09-21 21:05:23 +00:00
|
|
|
gfs2_meta_sync(ip->i_gl);
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
fs_info(sdp, "jid=%u: Replayed %u of %u blocks\n",
|
|
|
|
|
jd->jd_jid, sdp->sd_replayed_blocks, sdp->sd_found_blocks);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void revoke_lo_add(struct gfs2_sbd *sdp, struct gfs2_log_element *le)
|
|
|
|
|
{
|
2011-04-14 15:50:31 +00:00
|
|
|
struct gfs2_bufdata *bd = container_of(le, struct gfs2_bufdata, bd_le);
|
|
|
|
|
struct gfs2_glock *gl = bd->bd_gl;
|
2006-01-16 16:50:04 +00:00
|
|
|
struct gfs2_trans *tr;
|
|
|
|
|
|
2006-02-27 22:23:27 +00:00
|
|
|
tr = current->journal_info;
|
2006-01-16 16:50:04 +00:00
|
|
|
tr->tr_touched = 1;
|
|
|
|
|
tr->tr_num_revoke++;
|
|
|
|
|
sdp->sd_log_num_revoke++;
|
2011-04-14 15:50:31 +00:00
|
|
|
atomic_inc(&gl->gl_revokes);
|
|
|
|
|
set_bit(GLF_LFLUSH, &gl->gl_flags);
|
2006-01-16 16:50:04 +00:00
|
|
|
list_add(&le->le_list, &sdp->sd_log_le_revoke);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void revoke_lo_before_commit(struct gfs2_sbd *sdp)
|
|
|
|
|
{
|
|
|
|
|
struct gfs2_log_descriptor *ld;
|
|
|
|
|
struct gfs2_meta_header *mh;
|
|
|
|
|
unsigned int offset;
|
|
|
|
|
struct list_head *head = &sdp->sd_log_le_revoke;
|
2007-09-02 14:39:43 +00:00
|
|
|
struct gfs2_bufdata *bd;
|
2012-04-16 08:28:31 +00:00
|
|
|
struct page *page;
|
2012-04-16 08:40:00 +00:00
|
|
|
unsigned int length;
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
if (!sdp->sd_log_num_revoke)
|
|
|
|
|
return;
|
|
|
|
|
|
2012-04-16 08:40:00 +00:00
|
|
|
length = gfs2_struct2blk(sdp, sdp->sd_log_num_revoke, sizeof(u64));
|
|
|
|
|
page = gfs2_get_log_desc(sdp, GFS2_LOG_DESC_REVOKE, length, sdp->sd_log_num_revoke);
|
2012-04-16 08:28:31 +00:00
|
|
|
ld = page_address(page);
|
2006-01-16 16:50:04 +00:00
|
|
|
offset = sizeof(struct gfs2_log_descriptor);
|
|
|
|
|
|
2011-04-14 15:50:31 +00:00
|
|
|
list_for_each_entry(bd, head, bd_le.le_list) {
|
2006-01-16 16:50:04 +00:00
|
|
|
sdp->sd_log_num_revoke--;
|
|
|
|
|
|
2006-09-04 16:49:07 +00:00
|
|
|
if (offset + sizeof(u64) > sdp->sd_sb.sb_bsize) {
|
2006-01-16 16:50:04 +00:00
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
gfs2_log_write_page(sdp, page);
|
|
|
|
|
page = mempool_alloc(gfs2_page_pool, GFP_NOIO);
|
|
|
|
|
mh = page_address(page);
|
|
|
|
|
clear_page(mh);
|
2006-01-16 16:50:04 +00:00
|
|
|
mh->mh_magic = cpu_to_be32(GFS2_MAGIC);
|
2006-03-30 20:46:23 +00:00
|
|
|
mh->mh_type = cpu_to_be32(GFS2_METATYPE_LB);
|
|
|
|
|
mh->mh_format = cpu_to_be32(GFS2_FORMAT_LB);
|
2006-01-16 16:50:04 +00:00
|
|
|
offset = sizeof(struct gfs2_meta_header);
|
|
|
|
|
}
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
*(__be64 *)(page_address(page) + offset) = cpu_to_be64(bd->bd_blkno);
|
2006-09-04 16:49:07 +00:00
|
|
|
offset += sizeof(u64);
|
2006-01-16 16:50:04 +00:00
|
|
|
}
|
|
|
|
|
gfs2_assert_withdraw(sdp, !sdp->sd_log_num_revoke);
|
|
|
|
|
|
2012-04-16 08:28:31 +00:00
|
|
|
gfs2_log_write_page(sdp, page);
|
2006-01-16 16:50:04 +00:00
|
|
|
}
|
|
|
|
|
|
2011-04-14 15:50:31 +00:00
|
|
|
static void revoke_lo_after_commit(struct gfs2_sbd *sdp, struct gfs2_ail *ai)
|
|
|
|
|
{
|
|
|
|
|
struct list_head *head = &sdp->sd_log_le_revoke;
|
|
|
|
|
struct gfs2_bufdata *bd;
|
|
|
|
|
struct gfs2_glock *gl;
|
|
|
|
|
|
|
|
|
|
while (!list_empty(head)) {
|
|
|
|
|
bd = list_entry(head->next, struct gfs2_bufdata, bd_le.le_list);
|
|
|
|
|
list_del_init(&bd->bd_le.le_list);
|
|
|
|
|
gl = bd->bd_gl;
|
|
|
|
|
atomic_dec(&gl->gl_revokes);
|
|
|
|
|
clear_bit(GLF_LFLUSH, &gl->gl_flags);
|
|
|
|
|
kmem_cache_free(gfs2_bufdata_cachep, bd);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2006-01-16 16:50:04 +00:00
|
|
|
static void revoke_lo_before_scan(struct gfs2_jdesc *jd,
|
2006-10-14 01:47:13 +00:00
|
|
|
struct gfs2_log_header_host *head, int pass)
|
2006-01-16 16:50:04 +00:00
|
|
|
{
|
2006-06-14 19:32:57 +00:00
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
if (pass != 0)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
sdp->sd_found_revokes = 0;
|
|
|
|
|
sdp->sd_replay_tail = head->lh_tail;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int revoke_lo_scan_elements(struct gfs2_jdesc *jd, unsigned int start,
|
|
|
|
|
struct gfs2_log_descriptor *ld, __be64 *ptr,
|
|
|
|
|
int pass)
|
|
|
|
|
{
|
2006-06-14 19:32:57 +00:00
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-01-16 16:50:04 +00:00
|
|
|
unsigned int blks = be32_to_cpu(ld->ld_length);
|
|
|
|
|
unsigned int revokes = be32_to_cpu(ld->ld_data1);
|
|
|
|
|
struct buffer_head *bh;
|
|
|
|
|
unsigned int offset;
|
2006-09-04 16:49:07 +00:00
|
|
|
u64 blkno;
|
2006-01-16 16:50:04 +00:00
|
|
|
int first = 1;
|
|
|
|
|
int error;
|
|
|
|
|
|
|
|
|
|
if (pass != 0 || be32_to_cpu(ld->ld_type) != GFS2_LOG_DESC_REVOKE)
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
offset = sizeof(struct gfs2_log_descriptor);
|
|
|
|
|
|
|
|
|
|
for (; blks; gfs2_replay_incr_blk(sdp, &start), blks--) {
|
|
|
|
|
error = gfs2_replay_read_block(jd, start, &bh);
|
|
|
|
|
if (error)
|
|
|
|
|
return error;
|
|
|
|
|
|
|
|
|
|
if (!first)
|
|
|
|
|
gfs2_metatype_check(sdp, bh, GFS2_METATYPE_LB);
|
|
|
|
|
|
2006-09-04 16:49:07 +00:00
|
|
|
while (offset + sizeof(u64) <= sdp->sd_sb.sb_bsize) {
|
2006-01-16 16:50:04 +00:00
|
|
|
blkno = be64_to_cpu(*(__be64 *)(bh->b_data + offset));
|
|
|
|
|
|
|
|
|
|
error = gfs2_revoke_add(sdp, blkno, start);
|
2008-01-28 22:35:13 +00:00
|
|
|
if (error < 0) {
|
|
|
|
|
brelse(bh);
|
2006-01-16 16:50:04 +00:00
|
|
|
return error;
|
2008-01-28 22:35:13 +00:00
|
|
|
}
|
2006-01-16 16:50:04 +00:00
|
|
|
else if (error)
|
|
|
|
|
sdp->sd_found_revokes++;
|
|
|
|
|
|
|
|
|
|
if (!--revokes)
|
|
|
|
|
break;
|
2006-09-04 16:49:07 +00:00
|
|
|
offset += sizeof(u64);
|
2006-01-16 16:50:04 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
brelse(bh);
|
|
|
|
|
offset = sizeof(struct gfs2_meta_header);
|
|
|
|
|
first = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void revoke_lo_after_scan(struct gfs2_jdesc *jd, int error, int pass)
|
|
|
|
|
{
|
2006-06-14 19:32:57 +00:00
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-01-16 16:50:04 +00:00
|
|
|
|
|
|
|
|
if (error) {
|
|
|
|
|
gfs2_revoke_clean(sdp);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (pass != 1)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
fs_info(sdp, "jid=%u: Found %u revoke tags\n",
|
|
|
|
|
jd->jd_jid, sdp->sd_found_revokes);
|
|
|
|
|
|
|
|
|
|
gfs2_revoke_clean(sdp);
|
|
|
|
|
}
|
|
|
|
|
|
2006-02-08 11:50:51 +00:00
|
|
|
/**
|
|
|
|
|
* databuf_lo_add - Add a databuf to the transaction.
|
|
|
|
|
*
|
|
|
|
|
* This is used in two distinct cases:
|
|
|
|
|
* i) In ordered write mode
|
|
|
|
|
* We put the data buffer on a list so that we can ensure that its
|
|
|
|
|
* synced to disk at the right time
|
|
|
|
|
* ii) In journaled data mode
|
|
|
|
|
* We need to journal the data block in the same way as metadata in
|
|
|
|
|
* the functions above. The difference is that here we have a tag
|
|
|
|
|
* which is two __be64's being the block number (as per meta data)
|
|
|
|
|
* and a flag which says whether the data block needs escaping or
|
|
|
|
|
* not. This means we need a new log entry for each 251 or so data
|
|
|
|
|
* blocks, which isn't an enormous overhead but twice as much as
|
|
|
|
|
* for normal metadata blocks.
|
|
|
|
|
*/
|
2006-01-16 16:50:04 +00:00
|
|
|
static void databuf_lo_add(struct gfs2_sbd *sdp, struct gfs2_log_element *le)
|
|
|
|
|
{
|
2006-02-08 11:50:51 +00:00
|
|
|
struct gfs2_bufdata *bd = container_of(le, struct gfs2_bufdata, bd_le);
|
2006-02-27 22:23:27 +00:00
|
|
|
struct gfs2_trans *tr = current->journal_info;
|
2006-02-08 11:50:51 +00:00
|
|
|
struct address_space *mapping = bd->bd_bh->b_page->mapping;
|
2006-06-14 19:32:57 +00:00
|
|
|
struct gfs2_inode *ip = GFS2_I(mapping->host);
|
2006-01-16 16:50:04 +00:00
|
|
|
|
2007-08-27 12:54:05 +00:00
|
|
|
lock_buffer(bd->bd_bh);
|
2007-01-25 10:04:20 +00:00
|
|
|
gfs2_log_lock(sdp);
|
2007-09-28 12:49:05 +00:00
|
|
|
if (tr) {
|
|
|
|
|
if (!list_empty(&bd->bd_list_tr))
|
|
|
|
|
goto out;
|
|
|
|
|
tr->tr_touched = 1;
|
|
|
|
|
if (gfs2_is_jdata(ip)) {
|
|
|
|
|
tr->tr_num_buf++;
|
|
|
|
|
list_add(&bd->bd_list_tr, &tr->tr_list_buf);
|
|
|
|
|
}
|
2007-06-20 13:34:06 +00:00
|
|
|
}
|
[GFS2] assertion failure after writing to journaled file, umount
This patch passes all my nasty tests that were causing the code to
fail under one circumstance or another. Here is a complete summary
of all changes from today's git tree, in order of appearance:
1. There are now separate variables for metadata buffer accounting.
2. Variable sd_log_num_hdrs is no longer needed, since the header
accounting is taken care of by the reserve/refund sequence.
3. Fixed a tiny grammatical problem in a comment.
4. Added a new function "calc_reserved" to calculate the reserved
log space. This isn't entirely necessary, but it has two benefits:
First, it simplifies the gfs2_log_refund function greatly.
Second, it allows for easier debugging because I could sprinkle the
code with calls to this function to make sure the accounting is
proper (by adding asserts and printks) at strategic point of the code.
5. In log_pull_tail there apparently was a kludge to fix up the
accounting based on a "pull" parameter. The buffer accounting is
now done properly, so the kludge was removed.
6. File sync operations were making a call to gfs2_log_flush that
writes another journal header. Since that header was unplanned
for (reserved) by the reserve/refund sequence, the free space had
to be decremented so that when log_pull_tail gets called, the free
space is be adjusted properly. (Did I hear you call that a kludge?
well, maybe, but a lot more justifiable than the one I removed).
7. In the gfs2_log_shutdown code, it optionally syncs the log by
specifying the PULL parameter to log_write_header. I'm not sure
this is necessary anymore. It just seems to me there could be
cases where shutdown is called while there are outstanding log
buffers.
8. In the (data)buf_lo_before_commit functions, I changed some offset
values from being calculated on the fly to being constants. That
simplified some code and we might as well let the compiler do the
calculation once rather than redoing those cycles at run time.
9. This version has my rewritten databuf_lo_add function.
This version is much more like its predecessor, buf_lo_add, which
makes it easier to understand. Again, this might not be necessary,
but it seems as if this one works as well as the previous one,
maybe even better, so I decided to leave it in.
10. In databuf_lo_before_commit, a previous data corruption problem
was caused by going off the end of the buffer. The proper solution
is to have the proper limit in place, rather than stopping earlier.
(Thus my previous attempt to fix it is wrong).
If you don't wrap the buffer, you're stopping too early and that
causes more log buffer accounting problems.
11. In lops.h there are two new (previously mentioned) constants for
figuring out the data offset for the journal buffers.
12. There are also two new functions, buf_limit and databuf_limit to
calculate how many entries will fit in the buffer.
13. In function gfs2_meta_wipe, it needs to distinguish between pinned
metadata buffers and journaled data buffers for proper journal buffer
accounting. It can't use the JDATA gfs2_inode flag because it's
sometimes passed the "real" inode and sometimes the "metadata
inode" and the inode flags will be random bits in a metadata
gfs2_inode. It needs to base its decision on which was passed in.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2007-06-18 19:50:20 +00:00
|
|
|
if (!list_empty(&le->le_list))
|
2007-08-27 12:54:05 +00:00
|
|
|
goto out;
|
[GFS2] assertion failure after writing to journaled file, umount
This patch passes all my nasty tests that were causing the code to
fail under one circumstance or another. Here is a complete summary
of all changes from today's git tree, in order of appearance:
1. There are now separate variables for metadata buffer accounting.
2. Variable sd_log_num_hdrs is no longer needed, since the header
accounting is taken care of by the reserve/refund sequence.
3. Fixed a tiny grammatical problem in a comment.
4. Added a new function "calc_reserved" to calculate the reserved
log space. This isn't entirely necessary, but it has two benefits:
First, it simplifies the gfs2_log_refund function greatly.
Second, it allows for easier debugging because I could sprinkle the
code with calls to this function to make sure the accounting is
proper (by adding asserts and printks) at strategic point of the code.
5. In log_pull_tail there apparently was a kludge to fix up the
accounting based on a "pull" parameter. The buffer accounting is
now done properly, so the kludge was removed.
6. File sync operations were making a call to gfs2_log_flush that
writes another journal header. Since that header was unplanned
for (reserved) by the reserve/refund sequence, the free space had
to be decremented so that when log_pull_tail gets called, the free
space is be adjusted properly. (Did I hear you call that a kludge?
well, maybe, but a lot more justifiable than the one I removed).
7. In the gfs2_log_shutdown code, it optionally syncs the log by
specifying the PULL parameter to log_write_header. I'm not sure
this is necessary anymore. It just seems to me there could be
cases where shutdown is called while there are outstanding log
buffers.
8. In the (data)buf_lo_before_commit functions, I changed some offset
values from being calculated on the fly to being constants. That
simplified some code and we might as well let the compiler do the
calculation once rather than redoing those cycles at run time.
9. This version has my rewritten databuf_lo_add function.
This version is much more like its predecessor, buf_lo_add, which
makes it easier to understand. Again, this might not be necessary,
but it seems as if this one works as well as the previous one,
maybe even better, so I decided to leave it in.
10. In databuf_lo_before_commit, a previous data corruption problem
was caused by going off the end of the buffer. The proper solution
is to have the proper limit in place, rather than stopping earlier.
(Thus my previous attempt to fix it is wrong).
If you don't wrap the buffer, you're stopping too early and that
causes more log buffer accounting problems.
11. In lops.h there are two new (previously mentioned) constants for
figuring out the data offset for the journal buffers.
12. There are also two new functions, buf_limit and databuf_limit to
calculate how many entries will fit in the buffer.
13. In function gfs2_meta_wipe, it needs to distinguish between pinned
metadata buffers and journaled data buffers for proper journal buffer
accounting. It can't use the JDATA gfs2_inode flag because it's
sometimes passed the "real" inode and sometimes the "metadata
inode" and the inode flags will be random bits in a metadata
gfs2_inode. It needs to base its decision on which was passed in.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2007-06-18 19:50:20 +00:00
|
|
|
|
2007-11-08 14:25:12 +00:00
|
|
|
set_bit(GLF_LFLUSH, &bd->bd_gl->gl_flags);
|
|
|
|
|
set_bit(GLF_DIRTY, &bd->bd_gl->gl_flags);
|
[GFS2] assertion failure after writing to journaled file, umount
This patch passes all my nasty tests that were causing the code to
fail under one circumstance or another. Here is a complete summary
of all changes from today's git tree, in order of appearance:
1. There are now separate variables for metadata buffer accounting.
2. Variable sd_log_num_hdrs is no longer needed, since the header
accounting is taken care of by the reserve/refund sequence.
3. Fixed a tiny grammatical problem in a comment.
4. Added a new function "calc_reserved" to calculate the reserved
log space. This isn't entirely necessary, but it has two benefits:
First, it simplifies the gfs2_log_refund function greatly.
Second, it allows for easier debugging because I could sprinkle the
code with calls to this function to make sure the accounting is
proper (by adding asserts and printks) at strategic point of the code.
5. In log_pull_tail there apparently was a kludge to fix up the
accounting based on a "pull" parameter. The buffer accounting is
now done properly, so the kludge was removed.
6. File sync operations were making a call to gfs2_log_flush that
writes another journal header. Since that header was unplanned
for (reserved) by the reserve/refund sequence, the free space had
to be decremented so that when log_pull_tail gets called, the free
space is be adjusted properly. (Did I hear you call that a kludge?
well, maybe, but a lot more justifiable than the one I removed).
7. In the gfs2_log_shutdown code, it optionally syncs the log by
specifying the PULL parameter to log_write_header. I'm not sure
this is necessary anymore. It just seems to me there could be
cases where shutdown is called while there are outstanding log
buffers.
8. In the (data)buf_lo_before_commit functions, I changed some offset
values from being calculated on the fly to being constants. That
simplified some code and we might as well let the compiler do the
calculation once rather than redoing those cycles at run time.
9. This version has my rewritten databuf_lo_add function.
This version is much more like its predecessor, buf_lo_add, which
makes it easier to understand. Again, this might not be necessary,
but it seems as if this one works as well as the previous one,
maybe even better, so I decided to leave it in.
10. In databuf_lo_before_commit, a previous data corruption problem
was caused by going off the end of the buffer. The proper solution
is to have the proper limit in place, rather than stopping earlier.
(Thus my previous attempt to fix it is wrong).
If you don't wrap the buffer, you're stopping too early and that
causes more log buffer accounting problems.
11. In lops.h there are two new (previously mentioned) constants for
figuring out the data offset for the journal buffers.
12. There are also two new functions, buf_limit and databuf_limit to
calculate how many entries will fit in the buffer.
13. In function gfs2_meta_wipe, it needs to distinguish between pinned
metadata buffers and journaled data buffers for proper journal buffer
accounting. It can't use the JDATA gfs2_inode flag because it's
sometimes passed the "real" inode and sometimes the "metadata
inode" and the inode flags will be random bits in a metadata
gfs2_inode. It needs to base its decision on which was passed in.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2007-06-18 19:50:20 +00:00
|
|
|
if (gfs2_is_jdata(ip)) {
|
|
|
|
|
gfs2_pin(sdp, bd->bd_bh);
|
|
|
|
|
tr->tr_num_databuf_new++;
|
2007-09-02 09:48:13 +00:00
|
|
|
sdp->sd_log_num_databuf++;
|
2010-02-05 05:45:25 +00:00
|
|
|
list_add_tail(&le->le_list, &sdp->sd_log_le_databuf);
|
2007-09-02 09:48:13 +00:00
|
|
|
} else {
|
2010-02-05 05:45:25 +00:00
|
|
|
list_add_tail(&le->le_list, &sdp->sd_log_le_ordered);
|
2007-08-27 12:54:05 +00:00
|
|
|
}
|
|
|
|
|
out:
|
2006-01-16 16:50:04 +00:00
|
|
|
gfs2_log_unlock(sdp);
|
2007-08-27 12:54:05 +00:00
|
|
|
unlock_buffer(bd->bd_bh);
|
2006-01-16 16:50:04 +00:00
|
|
|
}
|
|
|
|
|
|
2007-09-17 09:59:52 +00:00
|
|
|
/**
|
|
|
|
|
* databuf_lo_before_commit - Scan the data buffers, writing as we go
|
|
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static void databuf_lo_before_commit(struct gfs2_sbd *sdp)
|
|
|
|
|
{
|
2012-04-16 08:40:00 +00:00
|
|
|
unsigned int limit = buf_limit(sdp) / 2;
|
2007-09-17 09:59:52 +00:00
|
|
|
|
2012-04-16 08:40:00 +00:00
|
|
|
gfs2_before_commit(sdp, limit, sdp->sd_log_num_databuf,
|
|
|
|
|
&sdp->sd_log_le_databuf, 1);
|
2006-02-08 11:50:51 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int databuf_lo_scan_elements(struct gfs2_jdesc *jd, unsigned int start,
|
|
|
|
|
struct gfs2_log_descriptor *ld,
|
|
|
|
|
__be64 *ptr, int pass)
|
|
|
|
|
{
|
2006-06-14 19:32:57 +00:00
|
|
|
struct gfs2_inode *ip = GFS2_I(jd->jd_inode);
|
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-02-27 22:23:27 +00:00
|
|
|
struct gfs2_glock *gl = ip->i_gl;
|
2006-02-08 11:50:51 +00:00
|
|
|
unsigned int blks = be32_to_cpu(ld->ld_data1);
|
|
|
|
|
struct buffer_head *bh_log, *bh_ip;
|
2006-09-04 16:49:07 +00:00
|
|
|
u64 blkno;
|
|
|
|
|
u64 esc;
|
2006-02-08 11:50:51 +00:00
|
|
|
int error = 0;
|
|
|
|
|
|
|
|
|
|
if (pass != 1 || be32_to_cpu(ld->ld_type) != GFS2_LOG_DESC_JDATA)
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
gfs2_replay_incr_blk(sdp, &start);
|
|
|
|
|
for (; blks; gfs2_replay_incr_blk(sdp, &start), blks--) {
|
|
|
|
|
blkno = be64_to_cpu(*ptr++);
|
|
|
|
|
esc = be64_to_cpu(*ptr++);
|
|
|
|
|
|
|
|
|
|
sdp->sd_found_blocks++;
|
|
|
|
|
|
|
|
|
|
if (gfs2_revoke_check(sdp, blkno, start))
|
|
|
|
|
continue;
|
|
|
|
|
|
|
|
|
|
error = gfs2_replay_read_block(jd, start, &bh_log);
|
|
|
|
|
if (error)
|
|
|
|
|
return error;
|
|
|
|
|
|
|
|
|
|
bh_ip = gfs2_meta_new(gl, blkno);
|
|
|
|
|
memcpy(bh_ip->b_data, bh_log->b_data, bh_log->b_size);
|
|
|
|
|
|
|
|
|
|
/* Unescape */
|
|
|
|
|
if (esc) {
|
|
|
|
|
__be32 *eptr = (__be32 *)bh_ip->b_data;
|
|
|
|
|
*eptr = cpu_to_be32(GFS2_MAGIC);
|
|
|
|
|
}
|
|
|
|
|
mark_buffer_dirty(bh_ip);
|
|
|
|
|
|
|
|
|
|
brelse(bh_log);
|
|
|
|
|
brelse(bh_ip);
|
|
|
|
|
|
|
|
|
|
sdp->sd_replayed_blocks++;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return error;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* FIXME: sort out accounting for log blocks etc. */
|
|
|
|
|
|
|
|
|
|
static void databuf_lo_after_scan(struct gfs2_jdesc *jd, int error, int pass)
|
|
|
|
|
{
|
2006-06-14 19:32:57 +00:00
|
|
|
struct gfs2_inode *ip = GFS2_I(jd->jd_inode);
|
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-02-08 11:50:51 +00:00
|
|
|
|
|
|
|
|
if (error) {
|
2006-09-21 21:05:23 +00:00
|
|
|
gfs2_meta_sync(ip->i_gl);
|
2006-02-08 11:50:51 +00:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (pass != 1)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
/* data sync? */
|
2006-09-21 21:05:23 +00:00
|
|
|
gfs2_meta_sync(ip->i_gl);
|
2006-02-08 11:50:51 +00:00
|
|
|
|
|
|
|
|
fs_info(sdp, "jid=%u: Replayed %u of %u data blocks\n",
|
|
|
|
|
jd->jd_jid, sdp->sd_replayed_blocks, sdp->sd_found_blocks);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void databuf_lo_after_commit(struct gfs2_sbd *sdp, struct gfs2_ail *ai)
|
|
|
|
|
{
|
|
|
|
|
struct list_head *head = &sdp->sd_log_le_databuf;
|
|
|
|
|
struct gfs2_bufdata *bd;
|
|
|
|
|
|
|
|
|
|
while (!list_empty(head)) {
|
|
|
|
|
bd = list_entry(head->next, struct gfs2_bufdata, bd_le.le_list);
|
2006-08-22 20:25:50 +00:00
|
|
|
list_del_init(&bd->bd_le.le_list);
|
2006-02-08 11:50:51 +00:00
|
|
|
sdp->sd_log_num_databuf--;
|
|
|
|
|
gfs2_unpin(sdp, bd->bd_bh, ai);
|
|
|
|
|
}
|
2006-01-16 16:50:04 +00:00
|
|
|
gfs2_assert_warn(sdp, !sdp->sd_log_num_databuf);
|
|
|
|
|
}
|
|
|
|
|
|
2006-02-08 11:50:51 +00:00
|
|
|
|
2006-04-07 15:17:32 +00:00
|
|
|
const struct gfs2_log_operations gfs2_buf_lops = {
|
2006-01-16 16:50:04 +00:00
|
|
|
.lo_add = buf_lo_add,
|
|
|
|
|
.lo_before_commit = buf_lo_before_commit,
|
|
|
|
|
.lo_after_commit = buf_lo_after_commit,
|
|
|
|
|
.lo_before_scan = buf_lo_before_scan,
|
|
|
|
|
.lo_scan_elements = buf_lo_scan_elements,
|
|
|
|
|
.lo_after_scan = buf_lo_after_scan,
|
2006-09-05 14:53:09 +00:00
|
|
|
.lo_name = "buf",
|
2006-01-16 16:50:04 +00:00
|
|
|
};
|
|
|
|
|
|
2006-04-07 15:17:32 +00:00
|
|
|
const struct gfs2_log_operations gfs2_revoke_lops = {
|
2006-01-16 16:50:04 +00:00
|
|
|
.lo_add = revoke_lo_add,
|
|
|
|
|
.lo_before_commit = revoke_lo_before_commit,
|
2011-04-14 15:50:31 +00:00
|
|
|
.lo_after_commit = revoke_lo_after_commit,
|
2006-01-16 16:50:04 +00:00
|
|
|
.lo_before_scan = revoke_lo_before_scan,
|
|
|
|
|
.lo_scan_elements = revoke_lo_scan_elements,
|
|
|
|
|
.lo_after_scan = revoke_lo_after_scan,
|
2006-09-05 14:53:09 +00:00
|
|
|
.lo_name = "revoke",
|
2006-01-16 16:50:04 +00:00
|
|
|
};
|
|
|
|
|
|
2006-04-07 15:17:32 +00:00
|
|
|
const struct gfs2_log_operations gfs2_rg_lops = {
|
2006-09-05 14:53:09 +00:00
|
|
|
.lo_name = "rg",
|
2006-01-16 16:50:04 +00:00
|
|
|
};
|
|
|
|
|
|
2006-04-07 15:17:32 +00:00
|
|
|
const struct gfs2_log_operations gfs2_databuf_lops = {
|
2006-01-16 16:50:04 +00:00
|
|
|
.lo_add = databuf_lo_add,
|
|
|
|
|
.lo_before_commit = databuf_lo_before_commit,
|
2006-02-08 11:50:51 +00:00
|
|
|
.lo_after_commit = databuf_lo_after_commit,
|
|
|
|
|
.lo_scan_elements = databuf_lo_scan_elements,
|
|
|
|
|
.lo_after_scan = databuf_lo_after_scan,
|
2006-09-05 14:53:09 +00:00
|
|
|
.lo_name = "databuf",
|
2006-01-16 16:50:04 +00:00
|
|
|
};
|
|
|
|
|
|
2006-04-07 15:17:32 +00:00
|
|
|
const struct gfs2_log_operations *gfs2_log_ops[] = {
|
2007-09-17 09:59:52 +00:00
|
|
|
&gfs2_databuf_lops,
|
2006-01-16 16:50:04 +00:00
|
|
|
&gfs2_buf_lops,
|
|
|
|
|
&gfs2_rg_lops,
|
2007-09-17 09:59:52 +00:00
|
|
|
&gfs2_revoke_lops,
|
2006-09-05 14:53:09 +00:00
|
|
|
NULL,
|
2006-01-16 16:50:04 +00:00
|
|
|
};
|
|
|
|
|
|
