2019-05-31 16:09:56 +08:00
|
|
|
// SPDX-License-Identifier: GPL-2.0-only
|
2006-01-17 00:50:04 +08:00
|
|
|
/*
|
|
|
|
|
* Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
|
2006-05-19 03:09:15 +08:00
|
|
|
* Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
|
2006-01-17 00:50:04 +08:00
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#include <linux/sched.h>
|
|
|
|
|
#include <linux/slab.h>
|
|
|
|
|
#include <linux/spinlock.h>
|
|
|
|
|
#include <linux/completion.h>
|
|
|
|
|
#include <linux/buffer_head.h>
|
2012-03-08 20:10:23 +08:00
|
|
|
#include <linux/mempool.h>
|
2006-02-28 06:23:27 +08:00
|
|
|
#include <linux/gfs2_ondisk.h>
|
2009-04-07 21:13:01 +08:00
|
|
|
#include <linux/bio.h>
|
|
|
|
|
#include <linux/fs.h>
|
2013-05-07 22:58:49 +08:00
|
|
|
#include <linux/list_sort.h>
|
2019-05-03 03:17:40 +08:00
|
|
|
#include <linux/blkdev.h>
|
2006-01-17 00:50:04 +08:00
|
|
|
|
2019-05-03 03:17:40 +08:00
|
|
|
#include "bmap.h"
|
2018-01-17 07:01:33 +08:00
|
|
|
#include "dir.h"
|
2006-01-17 00:50:04 +08:00
|
|
|
#include "gfs2.h"
|
2006-02-28 06:23:27 +08:00
|
|
|
#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>
2007-06-19 03:50:20 +08:00
|
|
|
#include "inode.h"
|
2006-01-17 00:50:04 +08:00
|
|
|
#include "glock.h"
|
|
|
|
|
#include "log.h"
|
|
|
|
|
#include "lops.h"
|
|
|
|
|
#include "meta_io.h"
|
|
|
|
|
#include "recovery.h"
|
|
|
|
|
#include "rgrp.h"
|
|
|
|
|
#include "trans.h"
|
2006-02-28 06:23:27 +08:00
|
|
|
#include "util.h"
|
2009-06-12 15:49:20 +08:00
|
|
|
#include "trace_gfs2.h"
|
2006-01-17 00:50:04 +08:00
|
|
|
|
2007-08-27 20:54:05 +08:00
|
|
|
/**
|
|
|
|
|
* 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
|
|
|
|
|
*/
|
2012-12-14 20:52:14 +08:00
|
|
|
void gfs2_pin(struct gfs2_sbd *sdp, struct buffer_head *bh)
|
2007-08-27 20:54:05 +08:00
|
|
|
{
|
|
|
|
|
struct gfs2_bufdata *bd;
|
|
|
|
|
|
2011-03-30 23:33:25 +08:00
|
|
|
BUG_ON(!current->journal_info);
|
2007-08-27 20:54:05 +08:00
|
|
|
|
|
|
|
|
clear_buffer_dirty(bh);
|
|
|
|
|
if (test_set_buffer_pinned(bh))
|
|
|
|
|
gfs2_assert_withdraw(sdp, 0);
|
|
|
|
|
if (!buffer_uptodate(bh))
|
2018-06-07 18:56:46 +08:00
|
|
|
gfs2_io_error_bh_wd(sdp, bh);
|
2007-08-27 20:54:05 +08:00
|
|
|
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 20:40:29 +08:00
|
|
|
spin_lock(&sdp->sd_ail_lock);
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
if (bd->bd_tr)
|
|
|
|
|
list_move(&bd->bd_ail_st_list, &bd->bd_tr->tr_ail2_list);
|
2011-03-14 20:40:29 +08:00
|
|
|
spin_unlock(&sdp->sd_ail_lock);
|
2007-08-27 20:54:05 +08:00
|
|
|
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>
2010-05-05 03:29:16 +08:00
|
|
|
atomic_inc(&sdp->sd_log_pinned);
|
2009-06-12 15:49:20 +08:00
|
|
|
trace_gfs2_pin(bd, 1);
|
2007-08-27 20:54:05 +08:00
|
|
|
}
|
|
|
|
|
|
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 16:53:19 +08:00
|
|
|
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;
|
2015-03-17 00:52:05 +08:00
|
|
|
struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
|
2017-06-30 20:55:08 +08:00
|
|
|
struct gfs2_rgrpd *rgd = gfs2_glock2rgrp(gl);
|
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 16:53:19 +08:00
|
|
|
unsigned int index = bd->bd_bh->b_blocknr - gl->gl_name.ln_number;
|
|
|
|
|
struct gfs2_bitmap *bi = rgd->rd_bits + index;
|
|
|
|
|
|
2014-04-27 02:57:23 +08:00
|
|
|
if (bi->bi_clone == 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 16:53:19 +08:00
|
|
|
return;
|
|
|
|
|
if (sdp->sd_args.ar_discard)
|
2012-02-08 20:58:32 +08:00
|
|
|
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 16:53:19 +08:00
|
|
|
memcpy(bi->bi_clone + bi->bi_offset,
|
2018-09-27 06:32:46 +08:00
|
|
|
bd->bd_bh->b_data + bi->bi_offset, bi->bi_bytes);
|
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 16:53:19 +08:00
|
|
|
clear_bit(GBF_FULL, &bi->bi_flags);
|
|
|
|
|
rgd->rd_free_clone = rgd->rd_free;
|
2013-11-25 19:16:25 +08:00
|
|
|
rgd->rd_extfail_pt = rgd->rd_free;
|
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 16:53:19 +08:00
|
|
|
}
|
|
|
|
|
|
2007-08-27 20:54:05 +08:00
|
|
|
/**
|
|
|
|
|
* gfs2_unpin - Unpin a buffer
|
|
|
|
|
* @sdp: the filesystem the buffer belongs to
|
|
|
|
|
* @bh: The buffer to unpin
|
|
|
|
|
* @ai:
|
2011-03-30 23:33:25 +08:00
|
|
|
* @flags: The inode dirty flags
|
2007-08-27 20:54:05 +08:00
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static void gfs2_unpin(struct gfs2_sbd *sdp, struct buffer_head *bh,
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
struct gfs2_trans *tr)
|
2007-08-27 20:54:05 +08:00
|
|
|
{
|
|
|
|
|
struct gfs2_bufdata *bd = bh->b_private;
|
|
|
|
|
|
2011-03-30 23:33:25 +08:00
|
|
|
BUG_ON(!buffer_uptodate(bh));
|
|
|
|
|
BUG_ON(!buffer_pinned(bh));
|
2007-08-27 20:54:05 +08:00
|
|
|
|
|
|
|
|
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 16:53:19 +08:00
|
|
|
if (buffer_is_rgrp(bd))
|
|
|
|
|
maybe_release_space(bd);
|
|
|
|
|
|
2011-03-11 19:52:25 +08:00
|
|
|
spin_lock(&sdp->sd_ail_lock);
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
if (bd->bd_tr) {
|
2007-08-27 20:54:05 +08:00
|
|
|
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);
|
|
|
|
|
}
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
bd->bd_tr = tr;
|
|
|
|
|
list_add(&bd->bd_ail_st_list, &tr->tr_ail1_list);
|
2011-03-11 19:52:25 +08:00
|
|
|
spin_unlock(&sdp->sd_ail_lock);
|
|
|
|
|
|
2011-03-30 23:33:25 +08:00
|
|
|
clear_bit(GLF_LFLUSH, &bd->bd_gl->gl_flags);
|
2009-06-12 15:49:20 +08:00
|
|
|
trace_gfs2_pin(bd, 0);
|
2007-08-27 20:54:05 +08:00
|
|
|
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>
2010-05-05 03:29:16 +08:00
|
|
|
atomic_dec(&sdp->sd_log_pinned);
|
2007-08-27 20:54:05 +08:00
|
|
|
}
|
|
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
static void gfs2_log_incr_head(struct gfs2_sbd *sdp)
|
2007-09-17 17:59:52 +08:00
|
|
|
{
|
2012-04-16 16:28:31 +08:00
|
|
|
BUG_ON((sdp->sd_log_flush_head == sdp->sd_log_tail) &&
|
|
|
|
|
(sdp->sd_log_flush_head != sdp->sd_log_head));
|
|
|
|
|
|
2017-06-20 22:52:57 +08:00
|
|
|
if (++sdp->sd_log_flush_head == sdp->sd_jdesc->jd_blocks)
|
2012-04-16 16:28:31 +08:00
|
|
|
sdp->sd_log_flush_head = 0;
|
2007-09-17 17:59:52 +08:00
|
|
|
}
|
|
|
|
|
|
2018-01-17 07:01:33 +08:00
|
|
|
u64 gfs2_log_bmap(struct gfs2_sbd *sdp)
|
2007-09-17 17:59:52 +08:00
|
|
|
{
|
2012-04-16 16:28:31 +08:00
|
|
|
unsigned int lbn = sdp->sd_log_flush_head;
|
|
|
|
|
struct gfs2_journal_extent *je;
|
|
|
|
|
u64 block;
|
|
|
|
|
|
2014-03-03 21:35:57 +08:00
|
|
|
list_for_each_entry(je, &sdp->sd_jdesc->extent_list, list) {
|
|
|
|
|
if ((lbn >= je->lblock) && (lbn < (je->lblock + je->blocks))) {
|
2012-04-16 16:28:31 +08:00
|
|
|
block = je->dblock + lbn - je->lblock;
|
|
|
|
|
gfs2_log_incr_head(sdp);
|
|
|
|
|
return block;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return -1;
|
2007-09-17 17:59:52 +08:00
|
|
|
}
|
|
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
/**
|
|
|
|
|
* 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
|
|
|
|
|
*
|
2018-01-26 01:12:31 +08:00
|
|
|
* This finds the relevant buffers and unlocks them and sets the
|
2012-04-16 16:28:31 +08:00
|
|
|
* 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.
|
|
|
|
|
*/
|
|
|
|
|
|
2019-02-15 19:13:19 +08:00
|
|
|
static void gfs2_end_log_write_bh(struct gfs2_sbd *sdp,
|
|
|
|
|
struct bio_vec *bvec,
|
2017-06-03 15:38:06 +08:00
|
|
|
blk_status_t error)
|
2007-09-17 17:59:52 +08:00
|
|
|
{
|
2012-04-16 16:28:31 +08:00
|
|
|
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)
|
2017-07-06 19:02:21 +08:00
|
|
|
mark_buffer_write_io_error(bh);
|
2012-04-16 16:28:31 +08:00
|
|
|
unlock_buffer(bh);
|
|
|
|
|
next = bh->b_this_page;
|
|
|
|
|
size -= bh->b_size;
|
|
|
|
|
brelse(bh);
|
|
|
|
|
bh = next;
|
|
|
|
|
} while(bh && size);
|
2007-09-17 17:59:52 +08:00
|
|
|
}
|
|
|
|
|
|
2012-02-03 23:21:59 +08:00
|
|
|
/**
|
2012-04-16 16:28:31 +08:00
|
|
|
* gfs2_end_log_write - end of i/o to the log
|
|
|
|
|
* @bio: The bio
|
|
|
|
|
*
|
|
|
|
|
* 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 23:21:59 +08:00
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
|
2015-07-20 21:29:37 +08:00
|
|
|
static void gfs2_end_log_write(struct bio *bio)
|
2012-02-03 23:21:59 +08:00
|
|
|
{
|
2012-04-16 16:28:31 +08:00
|
|
|
struct gfs2_sbd *sdp = bio->bi_private;
|
|
|
|
|
struct bio_vec *bvec;
|
|
|
|
|
struct page *page;
|
2019-02-15 19:13:19 +08:00
|
|
|
struct bvec_iter_all iter_all;
|
2012-04-16 16:28:31 +08:00
|
|
|
|
GFS2: Withdraw for IO errors writing to the journal or statfs
Before this patch, if GFS2 encountered IO errors while writing to
the journal, it would not report the problem, so they would go
unnoticed, sometimes for many hours. Sometimes this would only be
noticed later, when recovery tried to do journal replay and failed
due to invalid metadata at the blocks that resulted in IO errors.
This patch makes GFS2's log daemon check for IO errors. If it
encounters one, it withdraws from the file system and reports
why in dmesg. A similar action is taken when IO errors occur when
writing to the system statfs file.
These errors are also reported back to any callers of fsync, since
that requires the journal to be flushed. Therefore, any IO errors
that would previously go unnoticed are now noticed and the file
system is withdrawn as early as possible, thus preventing further
file system damage.
Also note that this reintroduces superblock variable sd_log_error,
which Christoph removed with commit f729b66fca.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
2017-08-17 00:30:06 +08:00
|
|
|
if (bio->bi_status) {
|
|
|
|
|
fs_err(sdp, "Error %d writing to journal, jid=%u\n",
|
|
|
|
|
bio->bi_status, sdp->sd_jdesc->jd_jid);
|
|
|
|
|
wake_up(&sdp->sd_logd_waitq);
|
|
|
|
|
}
|
2012-04-16 16:28:31 +08:00
|
|
|
|
2019-04-25 15:03:00 +08:00
|
|
|
bio_for_each_segment_all(bvec, bio, iter_all) {
|
2012-04-16 16:28:31 +08:00
|
|
|
page = bvec->bv_page;
|
|
|
|
|
if (page_has_buffers(page))
|
2017-06-03 15:38:06 +08:00
|
|
|
gfs2_end_log_write_bh(sdp, bvec, bio->bi_status);
|
2012-04-16 16:28:31 +08:00
|
|
|
else
|
|
|
|
|
mempool_free(page, gfs2_page_pool);
|
|
|
|
|
}
|
2012-02-03 23:21:59 +08:00
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
bio_put(bio);
|
2012-02-03 23:21:59 +08:00
|
|
|
if (atomic_dec_and_test(&sdp->sd_log_in_flight))
|
|
|
|
|
wake_up(&sdp->sd_log_flush_wait);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
2018-11-09 23:49:01 +08:00
|
|
|
* gfs2_log_submit_bio - Submit any pending log bio
|
|
|
|
|
* @biop: Address of the bio pointer
|
2019-05-03 03:17:40 +08:00
|
|
|
* @opf: REQ_OP | op_flags
|
2012-02-03 23:21:59 +08:00
|
|
|
*
|
2012-04-16 16:28:31 +08:00
|
|
|
* 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 23:21:59 +08:00
|
|
|
*/
|
|
|
|
|
|
2019-05-03 03:17:40 +08:00
|
|
|
void gfs2_log_submit_bio(struct bio **biop, int opf)
|
2012-02-03 23:21:59 +08:00
|
|
|
{
|
2018-11-09 23:49:01 +08:00
|
|
|
struct bio *bio = *biop;
|
|
|
|
|
if (bio) {
|
|
|
|
|
struct gfs2_sbd *sdp = bio->bi_private;
|
2012-04-16 16:28:31 +08:00
|
|
|
atomic_inc(&sdp->sd_log_in_flight);
|
2019-05-03 03:17:40 +08:00
|
|
|
bio->bi_opf = opf;
|
2018-11-09 23:49:01 +08:00
|
|
|
submit_bio(bio);
|
|
|
|
|
*biop = NULL;
|
2012-04-16 16:28:31 +08:00
|
|
|
}
|
|
|
|
|
}
|
2012-02-03 23:21:59 +08:00
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
/**
|
2018-11-09 23:49:01 +08:00
|
|
|
* gfs2_log_alloc_bio - Allocate a bio
|
|
|
|
|
* @sdp: The super block
|
|
|
|
|
* @blkno: The device block number we want to write to
|
|
|
|
|
* @end_io: The bi_end_io callback
|
2012-04-16 16:28:31 +08:00
|
|
|
*
|
2018-11-09 23:49:01 +08:00
|
|
|
* Allocate a new bio, initialize it with the given parameters and return it.
|
2012-04-16 16:28:31 +08:00
|
|
|
*
|
2018-11-09 23:49:01 +08:00
|
|
|
* Returns: The newly allocated bio
|
2012-04-16 16:28:31 +08:00
|
|
|
*/
|
|
|
|
|
|
2018-11-09 23:49:01 +08:00
|
|
|
static struct bio *gfs2_log_alloc_bio(struct gfs2_sbd *sdp, u64 blkno,
|
|
|
|
|
bio_end_io_t *end_io)
|
2012-04-16 16:28:31 +08:00
|
|
|
{
|
|
|
|
|
struct super_block *sb = sdp->sd_vfs;
|
2018-11-09 23:49:01 +08:00
|
|
|
struct bio *bio = bio_alloc(GFP_NOIO, BIO_MAX_PAGES);
|
2012-04-16 16:28:31 +08:00
|
|
|
|
2013-10-12 06:44:27 +08:00
|
|
|
bio->bi_iter.bi_sector = blkno * (sb->s_blocksize >> 9);
|
2017-08-24 01:10:32 +08:00
|
|
|
bio_set_dev(bio, sb->s_bdev);
|
2018-11-09 23:49:01 +08:00
|
|
|
bio->bi_end_io = end_io;
|
2012-04-16 16:28:31 +08:00
|
|
|
bio->bi_private = sdp;
|
|
|
|
|
|
|
|
|
|
return bio;
|
2012-02-03 23:21:59 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
2012-04-16 16:28:31 +08:00
|
|
|
* gfs2_log_get_bio - Get cached log bio, or allocate a new one
|
2018-11-09 23:49:01 +08:00
|
|
|
* @sdp: The super block
|
2012-04-16 16:28:31 +08:00
|
|
|
* @blkno: The device block number we want to write to
|
2018-11-09 23:49:01 +08:00
|
|
|
* @bio: The bio to get or allocate
|
|
|
|
|
* @op: REQ_OP
|
|
|
|
|
* @end_io: The bi_end_io callback
|
|
|
|
|
* @flush: Always flush the current bio and allocate a new one?
|
2012-04-16 16:28:31 +08:00
|
|
|
*
|
|
|
|
|
* 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
|
2018-11-09 23:49:01 +08:00
|
|
|
* device. If there is no cached bio, or we just flushed it, then
|
2012-04-16 16:28:31 +08:00
|
|
|
* allocate a new one.
|
2012-02-03 23:21:59 +08:00
|
|
|
*
|
2012-04-16 16:28:31 +08:00
|
|
|
* Returns: The bio to use for log writes
|
2012-02-03 23:21:59 +08:00
|
|
|
*/
|
|
|
|
|
|
2018-11-09 23:49:01 +08:00
|
|
|
static struct bio *gfs2_log_get_bio(struct gfs2_sbd *sdp, u64 blkno,
|
|
|
|
|
struct bio **biop, int op,
|
|
|
|
|
bio_end_io_t *end_io, bool flush)
|
2012-02-03 23:21:59 +08:00
|
|
|
{
|
2018-11-09 23:49:01 +08:00
|
|
|
struct bio *bio = *biop;
|
2012-04-16 16:28:31 +08:00
|
|
|
|
|
|
|
|
if (bio) {
|
2018-11-09 23:49:01 +08:00
|
|
|
u64 nblk;
|
|
|
|
|
|
2012-09-26 06:05:12 +08:00
|
|
|
nblk = bio_end_sector(bio);
|
2012-04-16 16:28:31 +08:00
|
|
|
nblk >>= sdp->sd_fsb2bb_shift;
|
2018-11-09 23:49:01 +08:00
|
|
|
if (blkno == nblk && !flush)
|
2012-04-16 16:28:31 +08:00
|
|
|
return bio;
|
2019-05-03 03:17:40 +08:00
|
|
|
gfs2_log_submit_bio(biop, op);
|
2012-04-16 16:28:31 +08:00
|
|
|
}
|
|
|
|
|
|
2018-11-09 23:49:01 +08:00
|
|
|
*biop = gfs2_log_alloc_bio(sdp, blkno, end_io);
|
|
|
|
|
return *biop;
|
2012-02-03 23:21:59 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
2012-04-16 16:28:31 +08:00
|
|
|
* gfs2_log_write - write to log
|
2012-02-03 23:21:59 +08:00
|
|
|
* @sdp: the filesystem
|
2012-04-16 16:28:31 +08:00
|
|
|
* @page: the page to write
|
|
|
|
|
* @size: the size of the data to write
|
|
|
|
|
* @offset: the offset within the page
|
2018-01-17 07:01:33 +08:00
|
|
|
* @blkno: block number of the log entry
|
2012-02-03 23:21:59 +08:00
|
|
|
*
|
2012-04-16 16:28:31 +08:00
|
|
|
* 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 23:21:59 +08:00
|
|
|
*/
|
|
|
|
|
|
2018-01-17 07:01:33 +08:00
|
|
|
void gfs2_log_write(struct gfs2_sbd *sdp, struct page *page,
|
|
|
|
|
unsigned size, unsigned offset, u64 blkno)
|
2012-02-03 23:21:59 +08:00
|
|
|
{
|
2012-04-16 16:28:31 +08:00
|
|
|
struct bio *bio;
|
|
|
|
|
int ret;
|
|
|
|
|
|
2018-11-09 23:49:01 +08:00
|
|
|
bio = gfs2_log_get_bio(sdp, blkno, &sdp->sd_log_bio, REQ_OP_WRITE,
|
|
|
|
|
gfs2_end_log_write, false);
|
2012-04-16 16:28:31 +08:00
|
|
|
ret = bio_add_page(bio, page, size, offset);
|
|
|
|
|
if (ret == 0) {
|
2018-11-09 23:49:01 +08:00
|
|
|
bio = gfs2_log_get_bio(sdp, blkno, &sdp->sd_log_bio,
|
|
|
|
|
REQ_OP_WRITE, gfs2_end_log_write, true);
|
2012-04-16 16:28:31 +08:00
|
|
|
ret = bio_add_page(bio, page, size, offset);
|
|
|
|
|
WARN_ON(ret == 0);
|
|
|
|
|
}
|
|
|
|
|
}
|
2012-02-03 23:21:59 +08:00
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
/**
|
|
|
|
|
* 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)
|
|
|
|
|
{
|
2018-01-17 07:01:33 +08:00
|
|
|
gfs2_log_write(sdp, bh->b_page, bh->b_size, bh_offset(bh),
|
|
|
|
|
gfs2_log_bmap(sdp));
|
2012-04-16 16:28:31 +08:00
|
|
|
}
|
2012-02-03 23:21:59 +08:00
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
/**
|
|
|
|
|
* 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 23:21:59 +08:00
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
void gfs2_log_write_page(struct gfs2_sbd *sdp, struct page *page)
|
|
|
|
|
{
|
|
|
|
|
struct super_block *sb = sdp->sd_vfs;
|
2018-01-17 07:01:33 +08:00
|
|
|
gfs2_log_write(sdp, page, sb->s_blocksize, 0,
|
|
|
|
|
gfs2_log_bmap(sdp));
|
2012-02-03 23:21:59 +08:00
|
|
|
}
|
2007-09-17 17:59:52 +08:00
|
|
|
|
2019-05-03 03:17:40 +08:00
|
|
|
/**
|
|
|
|
|
* gfs2_end_log_read - end I/O callback for reads from the log
|
|
|
|
|
* @bio: The bio
|
|
|
|
|
*
|
|
|
|
|
* Simply unlock the pages in the bio. The main thread will wait on them and
|
|
|
|
|
* process them in order as necessary.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static void gfs2_end_log_read(struct bio *bio)
|
|
|
|
|
{
|
|
|
|
|
struct page *page;
|
|
|
|
|
struct bio_vec *bvec;
|
|
|
|
|
struct bvec_iter_all iter_all;
|
|
|
|
|
|
2019-05-09 04:16:07 +08:00
|
|
|
bio_for_each_segment_all(bvec, bio, iter_all) {
|
2019-05-03 03:17:40 +08:00
|
|
|
page = bvec->bv_page;
|
|
|
|
|
if (bio->bi_status) {
|
|
|
|
|
int err = blk_status_to_errno(bio->bi_status);
|
|
|
|
|
|
|
|
|
|
SetPageError(page);
|
|
|
|
|
mapping_set_error(page->mapping, err);
|
|
|
|
|
}
|
|
|
|
|
unlock_page(page);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bio_put(bio);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* gfs2_jhead_pg_srch - Look for the journal head in a given page.
|
|
|
|
|
* @jd: The journal descriptor
|
|
|
|
|
* @page: The page to look in
|
|
|
|
|
*
|
|
|
|
|
* Returns: 1 if found, 0 otherwise.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static bool gfs2_jhead_pg_srch(struct gfs2_jdesc *jd,
|
|
|
|
|
struct gfs2_log_header_host *head,
|
|
|
|
|
struct page *page)
|
|
|
|
|
{
|
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
|
|
|
|
struct gfs2_log_header_host uninitialized_var(lh);
|
|
|
|
|
void *kaddr = kmap_atomic(page);
|
|
|
|
|
unsigned int offset;
|
|
|
|
|
bool ret = false;
|
|
|
|
|
|
|
|
|
|
for (offset = 0; offset < PAGE_SIZE; offset += sdp->sd_sb.sb_bsize) {
|
|
|
|
|
if (!__get_log_header(sdp, kaddr + offset, 0, &lh)) {
|
|
|
|
|
if (lh.lh_sequence > head->lh_sequence)
|
|
|
|
|
*head = lh;
|
|
|
|
|
else {
|
|
|
|
|
ret = true;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
kunmap_atomic(kaddr);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* gfs2_jhead_process_page - Search/cleanup a page
|
|
|
|
|
* @jd: The journal descriptor
|
|
|
|
|
* @index: Index of the page to look into
|
|
|
|
|
* @done: If set, perform only cleanup, else search and set if found.
|
|
|
|
|
*
|
|
|
|
|
* Find the page with 'index' in the journal's mapping. Search the page for
|
|
|
|
|
* the journal head if requested (cleanup == false). Release refs on the
|
|
|
|
|
* page so the page cache can reclaim it (put_page() twice). We grabbed a
|
|
|
|
|
* reference on this page two times, first when we did a find_or_create_page()
|
|
|
|
|
* to obtain the page to add it to the bio and second when we do a
|
|
|
|
|
* find_get_page() here to get the page to wait on while I/O on it is being
|
|
|
|
|
* completed.
|
|
|
|
|
* This function is also used to free up a page we might've grabbed but not
|
|
|
|
|
* used. Maybe we added it to a bio, but not submitted it for I/O. Or we
|
|
|
|
|
* submitted the I/O, but we already found the jhead so we only need to drop
|
|
|
|
|
* our references to the page.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static void gfs2_jhead_process_page(struct gfs2_jdesc *jd, unsigned long index,
|
|
|
|
|
struct gfs2_log_header_host *head,
|
|
|
|
|
bool *done)
|
|
|
|
|
{
|
|
|
|
|
struct page *page;
|
|
|
|
|
|
|
|
|
|
page = find_get_page(jd->jd_inode->i_mapping, index);
|
|
|
|
|
wait_on_page_locked(page);
|
|
|
|
|
|
|
|
|
|
if (PageError(page))
|
|
|
|
|
*done = true;
|
|
|
|
|
|
|
|
|
|
if (!*done)
|
|
|
|
|
*done = gfs2_jhead_pg_srch(jd, head, page);
|
|
|
|
|
|
|
|
|
|
put_page(page); /* Once for find_get_page */
|
|
|
|
|
put_page(page); /* Once more for find_or_create_page */
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* gfs2_find_jhead - find the head of a log
|
|
|
|
|
* @jd: The journal descriptor
|
|
|
|
|
* @head: The log descriptor for the head of the log is returned here
|
|
|
|
|
*
|
|
|
|
|
* Do a search of a journal by reading it in large chunks using bios and find
|
|
|
|
|
* the valid log entry with the highest sequence number. (i.e. the log head)
|
|
|
|
|
*
|
|
|
|
|
* Returns: 0 on success, errno otherwise
|
|
|
|
|
*/
|
|
|
|
|
int gfs2_find_jhead(struct gfs2_jdesc *jd, struct gfs2_log_header_host *head,
|
|
|
|
|
bool keep_cache)
|
|
|
|
|
{
|
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
|
|
|
|
struct address_space *mapping = jd->jd_inode->i_mapping;
|
|
|
|
|
unsigned int block = 0, blocks_submitted = 0, blocks_read = 0;
|
|
|
|
|
unsigned int bsize = sdp->sd_sb.sb_bsize;
|
|
|
|
|
unsigned int bsize_shift = sdp->sd_sb.sb_bsize_shift;
|
|
|
|
|
unsigned int shift = PAGE_SHIFT - bsize_shift;
|
|
|
|
|
unsigned int readhead_blocks = BIO_MAX_PAGES << shift;
|
|
|
|
|
struct gfs2_journal_extent *je;
|
|
|
|
|
int sz, ret = 0;
|
|
|
|
|
struct bio *bio = NULL;
|
|
|
|
|
struct page *page = NULL;
|
|
|
|
|
bool done = false;
|
|
|
|
|
errseq_t since;
|
|
|
|
|
|
|
|
|
|
memset(head, 0, sizeof(*head));
|
|
|
|
|
if (list_empty(&jd->extent_list))
|
|
|
|
|
gfs2_map_journal_extents(sdp, jd);
|
|
|
|
|
|
|
|
|
|
since = filemap_sample_wb_err(mapping);
|
|
|
|
|
list_for_each_entry(je, &jd->extent_list, list) {
|
|
|
|
|
for (; block < je->lblock + je->blocks; block++) {
|
|
|
|
|
u64 dblock;
|
|
|
|
|
|
|
|
|
|
if (!page) {
|
|
|
|
|
page = find_or_create_page(mapping,
|
|
|
|
|
block >> shift, GFP_NOFS);
|
|
|
|
|
if (!page) {
|
|
|
|
|
ret = -ENOMEM;
|
|
|
|
|
done = true;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (bio) {
|
|
|
|
|
unsigned int off;
|
|
|
|
|
|
|
|
|
|
off = (block << bsize_shift) & ~PAGE_MASK;
|
|
|
|
|
sz = bio_add_page(bio, page, bsize, off);
|
|
|
|
|
if (sz == bsize) { /* block added */
|
|
|
|
|
if (off + bsize == PAGE_SIZE) {
|
|
|
|
|
page = NULL;
|
|
|
|
|
goto page_added;
|
|
|
|
|
}
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
blocks_submitted = block + 1;
|
|
|
|
|
submit_bio(bio);
|
|
|
|
|
bio = NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
dblock = je->dblock + (block - je->lblock);
|
|
|
|
|
bio = gfs2_log_alloc_bio(sdp, dblock, gfs2_end_log_read);
|
|
|
|
|
bio->bi_opf = REQ_OP_READ;
|
|
|
|
|
sz = bio_add_page(bio, page, bsize, 0);
|
|
|
|
|
gfs2_assert_warn(sdp, sz == bsize);
|
|
|
|
|
if (bsize == PAGE_SIZE)
|
|
|
|
|
page = NULL;
|
|
|
|
|
|
|
|
|
|
page_added:
|
|
|
|
|
if (blocks_submitted < blocks_read + readhead_blocks) {
|
|
|
|
|
/* Keep at least one bio in flight */
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
gfs2_jhead_process_page(jd, blocks_read >> shift, head, &done);
|
|
|
|
|
blocks_read += PAGE_SIZE >> bsize_shift;
|
|
|
|
|
if (done)
|
|
|
|
|
goto out; /* found */
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
if (bio)
|
|
|
|
|
submit_bio(bio);
|
|
|
|
|
while (blocks_read < block) {
|
|
|
|
|
gfs2_jhead_process_page(jd, blocks_read >> shift, head, &done);
|
|
|
|
|
blocks_read += PAGE_SIZE >> bsize_shift;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (!ret)
|
|
|
|
|
ret = filemap_check_wb_err(mapping, since);
|
|
|
|
|
|
|
|
|
|
if (!keep_cache)
|
|
|
|
|
truncate_inode_pages(mapping, 0);
|
|
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2012-04-16 16:40:00 +08:00
|
|
|
static struct page *gfs2_get_log_desc(struct gfs2_sbd *sdp, u32 ld_type,
|
|
|
|
|
u32 ld_length, u32 ld_data1)
|
2007-09-17 17:59:52 +08:00
|
|
|
{
|
2012-04-19 17:38:50 +08:00
|
|
|
struct page *page = mempool_alloc(gfs2_page_pool, GFP_NOIO);
|
2012-04-16 16:28:31 +08:00
|
|
|
struct gfs2_log_descriptor *ld = page_address(page);
|
|
|
|
|
clear_page(ld);
|
2007-09-17 17:59:52 +08:00
|
|
|
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 16:40:00 +08:00
|
|
|
ld->ld_length = cpu_to_be32(ld_length);
|
|
|
|
|
ld->ld_data1 = cpu_to_be32(ld_data1);
|
2007-09-17 17:59:52 +08:00
|
|
|
ld->ld_data2 = 0;
|
2012-04-16 16:28:31 +08:00
|
|
|
return page;
|
2007-09-17 17:59:52 +08:00
|
|
|
}
|
|
|
|
|
|
2012-04-16 16:40:00 +08:00
|
|
|
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);
|
|
|
|
|
}
|
|
|
|
|
|
2013-05-07 22:58:49 +08:00
|
|
|
static int blocknr_cmp(void *priv, struct list_head *a, struct list_head *b)
|
|
|
|
|
{
|
|
|
|
|
struct gfs2_bufdata *bda, *bdb;
|
|
|
|
|
|
|
|
|
|
bda = list_entry(a, struct gfs2_bufdata, bd_list);
|
|
|
|
|
bdb = list_entry(b, struct gfs2_bufdata, bd_list);
|
|
|
|
|
|
|
|
|
|
if (bda->bd_bh->b_blocknr < bdb->bd_bh->b_blocknr)
|
|
|
|
|
return -1;
|
|
|
|
|
if (bda->bd_bh->b_blocknr > bdb->bd_bh->b_blocknr)
|
|
|
|
|
return 1;
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2012-04-16 16:40:00 +08:00
|
|
|
static void gfs2_before_commit(struct gfs2_sbd *sdp, unsigned int limit,
|
|
|
|
|
unsigned int total, struct list_head *blist,
|
|
|
|
|
bool is_databuf)
|
2006-01-17 00:50:04 +08:00
|
|
|
{
|
|
|
|
|
struct gfs2_log_descriptor *ld;
|
|
|
|
|
struct gfs2_bufdata *bd1 = NULL, *bd2;
|
2012-04-16 16:28:31 +08:00
|
|
|
struct page *page;
|
2006-01-17 00:50:04 +08:00
|
|
|
unsigned int num;
|
|
|
|
|
unsigned n;
|
|
|
|
|
__be64 *ptr;
|
|
|
|
|
|
2007-07-25 03:05:31 +08:00
|
|
|
gfs2_log_lock(sdp);
|
2013-05-07 22:58:49 +08:00
|
|
|
list_sort(NULL, blist, blocknr_cmp);
|
2012-05-02 00:00:34 +08:00
|
|
|
bd1 = bd2 = list_prepare_entry(bd1, blist, bd_list);
|
2006-01-17 00:50:04 +08:00
|
|
|
while(total) {
|
|
|
|
|
num = total;
|
|
|
|
|
if (total > limit)
|
|
|
|
|
num = limit;
|
2007-07-25 03:05:31 +08:00
|
|
|
gfs2_log_unlock(sdp);
|
2013-05-25 03:02:49 +08:00
|
|
|
page = gfs2_get_log_desc(sdp,
|
|
|
|
|
is_databuf ? GFS2_LOG_DESC_JDATA :
|
|
|
|
|
GFS2_LOG_DESC_METADATA, num + 1, num);
|
2012-04-16 16:28:31 +08:00
|
|
|
ld = page_address(page);
|
2007-07-25 03:05:31 +08:00
|
|
|
gfs2_log_lock(sdp);
|
2012-04-16 16:28:31 +08:00
|
|
|
ptr = (__be64 *)(ld + 1);
|
2006-01-17 00:50:04 +08:00
|
|
|
|
|
|
|
|
n = 0;
|
2012-05-02 00:00:34 +08:00
|
|
|
list_for_each_entry_continue(bd1, blist, bd_list) {
|
2006-01-17 00:50:04 +08:00
|
|
|
*ptr++ = cpu_to_be64(bd1->bd_bh->b_blocknr);
|
2012-04-16 16:40:00 +08:00
|
|
|
if (is_databuf) {
|
|
|
|
|
gfs2_check_magic(bd1->bd_bh);
|
|
|
|
|
*ptr++ = cpu_to_be64(buffer_escaped(bd1->bd_bh) ? 1 : 0);
|
|
|
|
|
}
|
2006-01-17 00:50:04 +08:00
|
|
|
if (++n >= num)
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
2007-07-25 03:05:31 +08:00
|
|
|
gfs2_log_unlock(sdp);
|
2012-04-16 16:28:31 +08:00
|
|
|
gfs2_log_write_page(sdp, page);
|
2007-07-25 03:05:31 +08:00
|
|
|
gfs2_log_lock(sdp);
|
2006-01-17 00:50:04 +08:00
|
|
|
|
|
|
|
|
n = 0;
|
2012-05-02 00:00:34 +08:00
|
|
|
list_for_each_entry_continue(bd2, blist, bd_list) {
|
2007-09-17 17:59:52 +08:00
|
|
|
get_bh(bd2->bd_bh);
|
2007-07-25 03:05:31 +08:00
|
|
|
gfs2_log_unlock(sdp);
|
2007-09-17 17:59:52 +08:00
|
|
|
lock_buffer(bd2->bd_bh);
|
2012-04-16 16:40:00 +08:00
|
|
|
|
|
|
|
|
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);
|
|
|
|
|
}
|
2007-07-25 03:05:31 +08:00
|
|
|
gfs2_log_lock(sdp);
|
2006-01-17 00:50:04 +08:00
|
|
|
if (++n >= num)
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
2007-07-25 03:05:31 +08:00
|
|
|
BUG_ON(total < num);
|
2006-01-17 00:50:04 +08:00
|
|
|
total -= num;
|
|
|
|
|
}
|
2007-07-25 03:05:31 +08:00
|
|
|
gfs2_log_unlock(sdp);
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
|
2014-02-21 23:22:35 +08:00
|
|
|
static void buf_lo_before_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
|
2012-04-16 16:40:00 +08:00
|
|
|
{
|
|
|
|
|
unsigned int limit = buf_limit(sdp); /* 503 for 4k blocks */
|
2014-02-22 05:55:33 +08:00
|
|
|
unsigned int nbuf;
|
2014-02-21 23:22:35 +08:00
|
|
|
if (tr == NULL)
|
|
|
|
|
return;
|
2014-02-22 05:55:33 +08:00
|
|
|
nbuf = tr->tr_num_buf_new - tr->tr_num_buf_rm;
|
|
|
|
|
gfs2_before_commit(sdp, limit, nbuf, &tr->tr_buf, 0);
|
2012-04-16 16:40:00 +08:00
|
|
|
}
|
|
|
|
|
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
static void buf_lo_after_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
|
2006-01-17 00:50:04 +08:00
|
|
|
{
|
2014-02-21 23:22:35 +08:00
|
|
|
struct list_head *head;
|
2006-01-17 00:50:04 +08:00
|
|
|
struct gfs2_bufdata *bd;
|
|
|
|
|
|
2014-02-21 23:22:35 +08:00
|
|
|
if (tr == NULL)
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
return;
|
|
|
|
|
|
2014-02-21 23:22:35 +08:00
|
|
|
head = &tr->tr_buf;
|
2006-01-17 00:50:04 +08:00
|
|
|
while (!list_empty(head)) {
|
2012-05-02 00:00:34 +08:00
|
|
|
bd = list_entry(head->next, struct gfs2_bufdata, bd_list);
|
|
|
|
|
list_del_init(&bd->bd_list);
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
gfs2_unpin(sdp, bd->bd_bh, tr);
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void buf_lo_before_scan(struct gfs2_jdesc *jd,
|
2006-10-14 09:47:13 +08:00
|
|
|
struct gfs2_log_header_host *head, int pass)
|
2006-01-17 00:50:04 +08:00
|
|
|
{
|
|
|
|
|
if (pass != 0)
|
|
|
|
|
return;
|
|
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_found_blocks = 0;
|
|
|
|
|
jd->jd_replayed_blocks = 0;
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
|
2019-03-25 23:34:19 +08:00
|
|
|
static int buf_lo_scan_elements(struct gfs2_jdesc *jd, u32 start,
|
2006-01-17 00:50:04 +08:00
|
|
|
struct gfs2_log_descriptor *ld, __be64 *ptr,
|
|
|
|
|
int pass)
|
|
|
|
|
{
|
2006-06-15 03:32:57 +08:00
|
|
|
struct gfs2_inode *ip = GFS2_I(jd->jd_inode);
|
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-02-28 06:23:27 +08:00
|
|
|
struct gfs2_glock *gl = ip->i_gl;
|
2006-01-17 00:50:04 +08:00
|
|
|
unsigned int blks = be32_to_cpu(ld->ld_data1);
|
|
|
|
|
struct buffer_head *bh_log, *bh_ip;
|
2006-09-05 00:49:07 +08:00
|
|
|
u64 blkno;
|
2006-01-17 00:50:04 +08:00
|
|
|
int error = 0;
|
|
|
|
|
|
|
|
|
|
if (pass != 1 || be32_to_cpu(ld->ld_type) != GFS2_LOG_DESC_METADATA)
|
|
|
|
|
return 0;
|
|
|
|
|
|
2016-07-22 02:02:44 +08:00
|
|
|
gfs2_replay_incr_blk(jd, &start);
|
2006-01-17 00:50:04 +08:00
|
|
|
|
2016-07-22 02:02:44 +08:00
|
|
|
for (; blks; gfs2_replay_incr_blk(jd, &start), blks--) {
|
2006-01-17 00:50:04 +08:00
|
|
|
blkno = be64_to_cpu(*ptr++);
|
|
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_found_blocks++;
|
2006-01-17 00:50:04 +08:00
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
if (gfs2_revoke_check(jd, blkno, start))
|
2006-01-17 00:50:04 +08:00
|
|
|
continue;
|
|
|
|
|
|
|
|
|
|
error = gfs2_replay_read_block(jd, start, &bh_log);
|
2006-09-05 02:47:06 +08:00
|
|
|
if (error)
|
|
|
|
|
return error;
|
2006-01-17 00:50:04 +08:00
|
|
|
|
|
|
|
|
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;
|
gfs2: Warn when a journal replay overwrites a rgrp with buffers
This patch adds some instrumentation in gfs2's journal replay that
indicates when we're about to overwrite a rgrp for which we already
have a valid buffer_head.
When this problem occurs, it's a situation in which this node has
been granted a rgrp glock and subsequently read in buffer_heads for
it, and possibly even made changes to the rgrp bits and/or
allocation values. But now another node has failed and forced us to
replay its journal, but its journal contains a copy of the same
rgrp, without a revoke, which means we're about to overwrite a
rgrp that we now rightfully own, with an obsolete copy. That is
always a problem. It means the other node (which failed and left
its journal to be replayed) failed to flush out its rgrp buffers,
write out the revoke, and invalidate its copy before it released
the glock to our possession.
No node should ever release a glock until its metadata has been
written to the journal and revoked and invalidated..
We also kludge around the problem and refuse to replace our good
copy with the journals bad copy by not marking the buffer dirty,
but never do it silently. That's wallpapering over a larger problem
that still exists. IOW, if this situation can happen to this node,
it can also happen to a different node and we wouldn't even know it
or be able to circumvent it: Suppose we have a 3-node cluster:
Node 1 fails, leaving an obsolete rgrp block in its journal without
a revoke. Node 2 grabs the rgrp as soon as the rgrp glock is
released and starts making changes, allocating and freeing blocks
from the rgrp, etc. Node 3 replays the journal from node 1,
oblivious and unaware that it's about to overwrite node 2's changes.
So we still need to be vocal and log the error to make it apparent
that a corruption path still exists in gfs2.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2019-02-28 04:26:59 +08:00
|
|
|
else {
|
|
|
|
|
struct gfs2_meta_header *mh =
|
|
|
|
|
(struct gfs2_meta_header *)bh_ip->b_data;
|
|
|
|
|
|
|
|
|
|
if (mh->mh_type == cpu_to_be32(GFS2_METATYPE_RG)) {
|
|
|
|
|
struct gfs2_rgrpd *rgd;
|
|
|
|
|
|
|
|
|
|
rgd = gfs2_blk2rgrpd(sdp, blkno, false);
|
|
|
|
|
if (rgd && rgd->rd_addr == blkno &&
|
|
|
|
|
rgd->rd_bits && rgd->rd_bits->bi_bh) {
|
|
|
|
|
fs_info(sdp, "Replaying 0x%llx but we "
|
|
|
|
|
"already have a bh!\n",
|
|
|
|
|
(unsigned long long)blkno);
|
|
|
|
|
fs_info(sdp, "busy:%d, pinned:%d\n",
|
|
|
|
|
buffer_busy(rgd->rd_bits->bi_bh) ? 1 : 0,
|
|
|
|
|
buffer_pinned(rgd->rd_bits->bi_bh));
|
2019-05-09 22:21:48 +08:00
|
|
|
gfs2_dump_glock(NULL, rgd->rd_gl, true);
|
gfs2: Warn when a journal replay overwrites a rgrp with buffers
This patch adds some instrumentation in gfs2's journal replay that
indicates when we're about to overwrite a rgrp for which we already
have a valid buffer_head.
When this problem occurs, it's a situation in which this node has
been granted a rgrp glock and subsequently read in buffer_heads for
it, and possibly even made changes to the rgrp bits and/or
allocation values. But now another node has failed and forced us to
replay its journal, but its journal contains a copy of the same
rgrp, without a revoke, which means we're about to overwrite a
rgrp that we now rightfully own, with an obsolete copy. That is
always a problem. It means the other node (which failed and left
its journal to be replayed) failed to flush out its rgrp buffers,
write out the revoke, and invalidate its copy before it released
the glock to our possession.
No node should ever release a glock until its metadata has been
written to the journal and revoked and invalidated..
We also kludge around the problem and refuse to replace our good
copy with the journals bad copy by not marking the buffer dirty,
but never do it silently. That's wallpapering over a larger problem
that still exists. IOW, if this situation can happen to this node,
it can also happen to a different node and we wouldn't even know it
or be able to circumvent it: Suppose we have a 3-node cluster:
Node 1 fails, leaving an obsolete rgrp block in its journal without
a revoke. Node 2 grabs the rgrp as soon as the rgrp glock is
released and starts making changes, allocating and freeing blocks
from the rgrp, etc. Node 3 replays the journal from node 1,
oblivious and unaware that it's about to overwrite node 2's changes.
So we still need to be vocal and log the error to make it apparent
that a corruption path still exists in gfs2.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2019-02-28 04:26:59 +08:00
|
|
|
}
|
|
|
|
|
}
|
2006-01-17 00:50:04 +08:00
|
|
|
mark_buffer_dirty(bh_ip);
|
gfs2: Warn when a journal replay overwrites a rgrp with buffers
This patch adds some instrumentation in gfs2's journal replay that
indicates when we're about to overwrite a rgrp for which we already
have a valid buffer_head.
When this problem occurs, it's a situation in which this node has
been granted a rgrp glock and subsequently read in buffer_heads for
it, and possibly even made changes to the rgrp bits and/or
allocation values. But now another node has failed and forced us to
replay its journal, but its journal contains a copy of the same
rgrp, without a revoke, which means we're about to overwrite a
rgrp that we now rightfully own, with an obsolete copy. That is
always a problem. It means the other node (which failed and left
its journal to be replayed) failed to flush out its rgrp buffers,
write out the revoke, and invalidate its copy before it released
the glock to our possession.
No node should ever release a glock until its metadata has been
written to the journal and revoked and invalidated..
We also kludge around the problem and refuse to replace our good
copy with the journals bad copy by not marking the buffer dirty,
but never do it silently. That's wallpapering over a larger problem
that still exists. IOW, if this situation can happen to this node,
it can also happen to a different node and we wouldn't even know it
or be able to circumvent it: Suppose we have a 3-node cluster:
Node 1 fails, leaving an obsolete rgrp block in its journal without
a revoke. Node 2 grabs the rgrp as soon as the rgrp glock is
released and starts making changes, allocating and freeing blocks
from the rgrp, etc. Node 3 replays the journal from node 1,
oblivious and unaware that it's about to overwrite node 2's changes.
So we still need to be vocal and log the error to make it apparent
that a corruption path still exists in gfs2.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2019-02-28 04:26:59 +08:00
|
|
|
}
|
2006-01-17 00:50:04 +08:00
|
|
|
brelse(bh_log);
|
|
|
|
|
brelse(bh_ip);
|
|
|
|
|
|
|
|
|
|
if (error)
|
|
|
|
|
break;
|
|
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_replayed_blocks++;
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return error;
|
|
|
|
|
}
|
|
|
|
|
|
2013-07-03 22:13:32 +08:00
|
|
|
/**
|
|
|
|
|
* gfs2_meta_sync - Sync all buffers associated with a glock
|
|
|
|
|
* @gl: The glock
|
|
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
static void gfs2_meta_sync(struct gfs2_glock *gl)
|
|
|
|
|
{
|
|
|
|
|
struct address_space *mapping = gfs2_glock2aspace(gl);
|
2015-03-17 00:52:05 +08:00
|
|
|
struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
|
2013-07-03 22:13:32 +08:00
|
|
|
int error;
|
|
|
|
|
|
2013-12-07 00:19:54 +08:00
|
|
|
if (mapping == NULL)
|
|
|
|
|
mapping = &sdp->sd_aspace;
|
|
|
|
|
|
2013-07-03 22:13:32 +08:00
|
|
|
filemap_fdatawrite(mapping);
|
|
|
|
|
error = filemap_fdatawait(mapping);
|
|
|
|
|
|
|
|
|
|
if (error)
|
2015-03-17 00:52:05 +08:00
|
|
|
gfs2_io_error(gl->gl_name.ln_sbd);
|
2013-07-03 22:13:32 +08:00
|
|
|
}
|
|
|
|
|
|
2006-01-17 00:50:04 +08:00
|
|
|
static void buf_lo_after_scan(struct gfs2_jdesc *jd, int error, int pass)
|
|
|
|
|
{
|
2006-06-15 03:32:57 +08:00
|
|
|
struct gfs2_inode *ip = GFS2_I(jd->jd_inode);
|
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-01-17 00:50:04 +08:00
|
|
|
|
|
|
|
|
if (error) {
|
2006-09-22 05:05:23 +08:00
|
|
|
gfs2_meta_sync(ip->i_gl);
|
2006-01-17 00:50:04 +08:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (pass != 1)
|
|
|
|
|
return;
|
|
|
|
|
|
2006-09-22 05:05:23 +08:00
|
|
|
gfs2_meta_sync(ip->i_gl);
|
2006-01-17 00:50:04 +08:00
|
|
|
|
|
|
|
|
fs_info(sdp, "jid=%u: Replayed %u of %u blocks\n",
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_jid, jd->jd_replayed_blocks, jd->jd_found_blocks);
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
|
2014-02-21 23:22:35 +08:00
|
|
|
static void revoke_lo_before_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
|
2006-01-17 00:50:04 +08:00
|
|
|
{
|
|
|
|
|
struct gfs2_meta_header *mh;
|
|
|
|
|
unsigned int offset;
|
2019-04-05 19:16:14 +08:00
|
|
|
struct list_head *head = &sdp->sd_log_revokes;
|
2007-09-02 22:39:43 +08:00
|
|
|
struct gfs2_bufdata *bd;
|
2012-04-16 16:28:31 +08:00
|
|
|
struct page *page;
|
2012-04-16 16:40:00 +08:00
|
|
|
unsigned int length;
|
2006-01-17 00:50:04 +08:00
|
|
|
|
2013-06-15 00:38:29 +08:00
|
|
|
gfs2_write_revokes(sdp);
|
2006-01-17 00:50:04 +08:00
|
|
|
if (!sdp->sd_log_num_revoke)
|
|
|
|
|
return;
|
|
|
|
|
|
2012-04-16 16:40:00 +08:00
|
|
|
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);
|
2006-01-17 00:50:04 +08:00
|
|
|
offset = sizeof(struct gfs2_log_descriptor);
|
|
|
|
|
|
2012-05-02 00:00:34 +08:00
|
|
|
list_for_each_entry(bd, head, bd_list) {
|
2006-01-17 00:50:04 +08:00
|
|
|
sdp->sd_log_num_revoke--;
|
|
|
|
|
|
2006-09-05 00:49:07 +08:00
|
|
|
if (offset + sizeof(u64) > sdp->sd_sb.sb_bsize) {
|
2006-01-17 00:50:04 +08:00
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
gfs2_log_write_page(sdp, page);
|
|
|
|
|
page = mempool_alloc(gfs2_page_pool, GFP_NOIO);
|
|
|
|
|
mh = page_address(page);
|
|
|
|
|
clear_page(mh);
|
2006-01-17 00:50:04 +08:00
|
|
|
mh->mh_magic = cpu_to_be32(GFS2_MAGIC);
|
2006-03-31 04:46:23 +08:00
|
|
|
mh->mh_type = cpu_to_be32(GFS2_METATYPE_LB);
|
|
|
|
|
mh->mh_format = cpu_to_be32(GFS2_FORMAT_LB);
|
2006-01-17 00:50:04 +08:00
|
|
|
offset = sizeof(struct gfs2_meta_header);
|
|
|
|
|
}
|
|
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
*(__be64 *)(page_address(page) + offset) = cpu_to_be64(bd->bd_blkno);
|
2006-09-05 00:49:07 +08:00
|
|
|
offset += sizeof(u64);
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
gfs2_assert_withdraw(sdp, !sdp->sd_log_num_revoke);
|
|
|
|
|
|
2012-04-16 16:28:31 +08:00
|
|
|
gfs2_log_write_page(sdp, page);
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
static void revoke_lo_after_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
|
2011-04-14 23:50:31 +08:00
|
|
|
{
|
2019-04-05 19:16:14 +08:00
|
|
|
struct list_head *head = &sdp->sd_log_revokes;
|
2019-06-06 20:33:38 +08:00
|
|
|
struct gfs2_bufdata *bd;
|
|
|
|
|
struct gfs2_glock *gl;
|
2019-04-05 11:41:38 +08:00
|
|
|
|
2019-06-06 20:33:38 +08:00
|
|
|
while (!list_empty(head)) {
|
|
|
|
|
bd = list_entry(head->next, struct gfs2_bufdata, bd_list);
|
|
|
|
|
list_del_init(&bd->bd_list);
|
|
|
|
|
gl = bd->bd_gl;
|
|
|
|
|
if (atomic_dec_return(&gl->gl_revokes) == 0) {
|
|
|
|
|
clear_bit(GLF_LFLUSH, &gl->gl_flags);
|
|
|
|
|
gfs2_glock_queue_put(gl);
|
gfs2: Fix occasional glock use-after-free
This patch has to do with the life cycle of glocks and buffers. When
gfs2 metadata or journaled data is queued to be written, a gfs2_bufdata
object is assigned to track the buffer, and that is queued to various
lists, including the glock's gl_ail_list to indicate it's on the active
items list. Once the page associated with the buffer has been written,
it is removed from the ail list, but its life isn't over until a revoke
has been successfully written.
So after the block is written, its bufdata object is moved from the
glock's gl_ail_list to a file-system-wide list of pending revokes,
sd_log_le_revoke. At that point the glock still needs to track how many
revokes it contributed to that list (in gl_revokes) so that things like
glock go_sync can ensure all the metadata has been not only written, but
also revoked before the glock is granted to a different node. This is
to guarantee journal replay doesn't replay the block once the glock has
been granted to another node.
Ross Lagerwall recently discovered a race in which an inode could be
evicted, and its glock freed after its ail list had been synced, but
while it still had unwritten revokes on the sd_log_le_revoke list. The
evict decremented the glock reference count to zero, which allowed the
glock to be freed. After the revoke was written, function
revoke_lo_after_commit tried to adjust the glock's gl_revokes counter
and clear its GLF_LFLUSH flag, at which time it referenced the freed
glock.
This patch fixes the problem by incrementing the glock reference count
in gfs2_add_revoke when the glock's first bufdata object is moved from
the glock to the global revokes list. Later, when the glock's last such
bufdata object is freed, the reference count is decremented. This
guarantees that whichever process finishes last (the revoke writing or
the evict) will properly free the glock, and neither will reference the
glock after it has been freed.
Reported-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
2019-04-05 04:11:11 +08:00
|
|
|
}
|
2011-04-14 23:50:31 +08:00
|
|
|
kmem_cache_free(gfs2_bufdata_cachep, bd);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2006-01-17 00:50:04 +08:00
|
|
|
static void revoke_lo_before_scan(struct gfs2_jdesc *jd,
|
2006-10-14 09:47:13 +08:00
|
|
|
struct gfs2_log_header_host *head, int pass)
|
2006-01-17 00:50:04 +08:00
|
|
|
{
|
|
|
|
|
if (pass != 0)
|
|
|
|
|
return;
|
|
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_found_revokes = 0;
|
|
|
|
|
jd->jd_replay_tail = head->lh_tail;
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
|
2019-03-25 23:34:19 +08:00
|
|
|
static int revoke_lo_scan_elements(struct gfs2_jdesc *jd, u32 start,
|
2006-01-17 00:50:04 +08:00
|
|
|
struct gfs2_log_descriptor *ld, __be64 *ptr,
|
|
|
|
|
int pass)
|
|
|
|
|
{
|
2006-06-15 03:32:57 +08:00
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-01-17 00:50:04 +08:00
|
|
|
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;
|
2006-09-05 00:49:07 +08:00
|
|
|
u64 blkno;
|
2006-01-17 00:50:04 +08:00
|
|
|
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);
|
|
|
|
|
|
2016-07-22 02:02:44 +08:00
|
|
|
for (; blks; gfs2_replay_incr_blk(jd, &start), blks--) {
|
2006-01-17 00:50:04 +08:00
|
|
|
error = gfs2_replay_read_block(jd, start, &bh);
|
|
|
|
|
if (error)
|
|
|
|
|
return error;
|
|
|
|
|
|
|
|
|
|
if (!first)
|
|
|
|
|
gfs2_metatype_check(sdp, bh, GFS2_METATYPE_LB);
|
|
|
|
|
|
2006-09-05 00:49:07 +08:00
|
|
|
while (offset + sizeof(u64) <= sdp->sd_sb.sb_bsize) {
|
2006-01-17 00:50:04 +08:00
|
|
|
blkno = be64_to_cpu(*(__be64 *)(bh->b_data + offset));
|
|
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
error = gfs2_revoke_add(jd, blkno, start);
|
2008-01-29 06:35:13 +08:00
|
|
|
if (error < 0) {
|
|
|
|
|
brelse(bh);
|
2006-01-17 00:50:04 +08:00
|
|
|
return error;
|
2008-01-29 06:35:13 +08:00
|
|
|
}
|
2006-01-17 00:50:04 +08:00
|
|
|
else if (error)
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_found_revokes++;
|
2006-01-17 00:50:04 +08:00
|
|
|
|
|
|
|
|
if (!--revokes)
|
|
|
|
|
break;
|
2006-09-05 00:49:07 +08:00
|
|
|
offset += sizeof(u64);
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
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)
|
|
|
|
|
{
|
2006-06-15 03:32:57 +08:00
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-01-17 00:50:04 +08:00
|
|
|
|
|
|
|
|
if (error) {
|
2014-03-07 06:19:15 +08:00
|
|
|
gfs2_revoke_clean(jd);
|
2006-01-17 00:50:04 +08:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (pass != 1)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
fs_info(sdp, "jid=%u: Found %u revoke tags\n",
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_jid, jd->jd_found_revokes);
|
2006-01-17 00:50:04 +08:00
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
gfs2_revoke_clean(jd);
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
|
2007-09-17 17:59:52 +08:00
|
|
|
/**
|
|
|
|
|
* databuf_lo_before_commit - Scan the data buffers, writing as we go
|
|
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
|
2014-02-21 23:22:35 +08:00
|
|
|
static void databuf_lo_before_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
|
2007-09-17 17:59:52 +08:00
|
|
|
{
|
2014-02-22 05:55:33 +08:00
|
|
|
unsigned int limit = databuf_limit(sdp);
|
|
|
|
|
unsigned int nbuf;
|
2014-02-21 23:22:35 +08:00
|
|
|
if (tr == NULL)
|
|
|
|
|
return;
|
2014-02-22 05:55:33 +08:00
|
|
|
nbuf = tr->tr_num_databuf_new - tr->tr_num_databuf_rm;
|
|
|
|
|
gfs2_before_commit(sdp, limit, nbuf, &tr->tr_databuf, 1);
|
2006-02-08 19:50:51 +08:00
|
|
|
}
|
|
|
|
|
|
2019-03-25 23:34:19 +08:00
|
|
|
static int databuf_lo_scan_elements(struct gfs2_jdesc *jd, u32 start,
|
2006-02-08 19:50:51 +08:00
|
|
|
struct gfs2_log_descriptor *ld,
|
|
|
|
|
__be64 *ptr, int pass)
|
|
|
|
|
{
|
2006-06-15 03:32:57 +08:00
|
|
|
struct gfs2_inode *ip = GFS2_I(jd->jd_inode);
|
2006-02-28 06:23:27 +08:00
|
|
|
struct gfs2_glock *gl = ip->i_gl;
|
2006-02-08 19:50:51 +08:00
|
|
|
unsigned int blks = be32_to_cpu(ld->ld_data1);
|
|
|
|
|
struct buffer_head *bh_log, *bh_ip;
|
2006-09-05 00:49:07 +08:00
|
|
|
u64 blkno;
|
|
|
|
|
u64 esc;
|
2006-02-08 19:50:51 +08:00
|
|
|
int error = 0;
|
|
|
|
|
|
|
|
|
|
if (pass != 1 || be32_to_cpu(ld->ld_type) != GFS2_LOG_DESC_JDATA)
|
|
|
|
|
return 0;
|
|
|
|
|
|
2016-07-22 02:02:44 +08:00
|
|
|
gfs2_replay_incr_blk(jd, &start);
|
|
|
|
|
for (; blks; gfs2_replay_incr_blk(jd, &start), blks--) {
|
2006-02-08 19:50:51 +08:00
|
|
|
blkno = be64_to_cpu(*ptr++);
|
|
|
|
|
esc = be64_to_cpu(*ptr++);
|
|
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_found_blocks++;
|
2006-02-08 19:50:51 +08:00
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
if (gfs2_revoke_check(jd, blkno, start))
|
2006-02-08 19:50:51 +08:00
|
|
|
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);
|
|
|
|
|
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_replayed_blocks++;
|
2006-02-08 19:50:51 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return error;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* FIXME: sort out accounting for log blocks etc. */
|
|
|
|
|
|
|
|
|
|
static void databuf_lo_after_scan(struct gfs2_jdesc *jd, int error, int pass)
|
|
|
|
|
{
|
2006-06-15 03:32:57 +08:00
|
|
|
struct gfs2_inode *ip = GFS2_I(jd->jd_inode);
|
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
|
2006-02-08 19:50:51 +08:00
|
|
|
|
|
|
|
|
if (error) {
|
2006-09-22 05:05:23 +08:00
|
|
|
gfs2_meta_sync(ip->i_gl);
|
2006-02-08 19:50:51 +08:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (pass != 1)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
/* data sync? */
|
2006-09-22 05:05:23 +08:00
|
|
|
gfs2_meta_sync(ip->i_gl);
|
2006-02-08 19:50:51 +08:00
|
|
|
|
|
|
|
|
fs_info(sdp, "jid=%u: Replayed %u of %u data blocks\n",
|
2014-03-07 06:19:15 +08:00
|
|
|
jd->jd_jid, jd->jd_replayed_blocks, jd->jd_found_blocks);
|
2006-02-08 19:50:51 +08:00
|
|
|
}
|
|
|
|
|
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
static void databuf_lo_after_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
|
2006-02-08 19:50:51 +08:00
|
|
|
{
|
2014-02-21 23:22:35 +08:00
|
|
|
struct list_head *head;
|
2006-02-08 19:50:51 +08:00
|
|
|
struct gfs2_bufdata *bd;
|
|
|
|
|
|
2014-02-21 23:22:35 +08:00
|
|
|
if (tr == NULL)
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
return;
|
|
|
|
|
|
2014-02-21 23:22:35 +08:00
|
|
|
head = &tr->tr_databuf;
|
2006-02-08 19:50:51 +08:00
|
|
|
while (!list_empty(head)) {
|
2012-05-02 00:00:34 +08:00
|
|
|
bd = list_entry(head->next, struct gfs2_bufdata, bd_list);
|
|
|
|
|
list_del_init(&bd->bd_list);
|
GFS2: replace gfs2_ail structure with gfs2_trans
In order to allow transactions and log flushes to happen at the same
time, gfs2 needs to move the transaction accounting and active items
list code into the gfs2_trans structure. As a first step toward this,
this patch removes the gfs2_ail structure, and handles the active items
list in the gfs_trans structure. This keeps gfs2 from allocating an ail
structure on log flushes, and gives us a struture that can later be used
to store the transaction accounting outside of the gfs2 superblock
structure.
With this patch, at the end of a transaction, gfs2 will add the
gfs2_trans structure to the superblock if there is not one already.
This structure now has the active items fields that were previously in
gfs2_ail. This is not necessary in the case where the transaction was
simply used to add revokes, since these are never written outside of the
journal, and thus, don't need an active items list.
Also, in order to make sure that the transaction structure is not
removed while it's still in use by gfs2_trans_end, unlocking the
sd_log_flush_lock has to happen slightly later in ending the
transaction.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2013-04-06 09:31:46 +08:00
|
|
|
gfs2_unpin(sdp, bd->bd_bh, tr);
|
2006-02-08 19:50:51 +08:00
|
|
|
}
|
2006-01-17 00:50:04 +08:00
|
|
|
}
|
|
|
|
|
|
2006-02-08 19:50:51 +08:00
|
|
|
|
2019-04-04 22:32:41 +08:00
|
|
|
static const struct gfs2_log_operations gfs2_buf_lops = {
|
2006-01-17 00:50:04 +08:00
|
|
|
.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 22:53:09 +08:00
|
|
|
.lo_name = "buf",
|
2006-01-17 00:50:04 +08:00
|
|
|
};
|
|
|
|
|
|
2019-04-04 22:32:41 +08:00
|
|
|
static const struct gfs2_log_operations gfs2_revoke_lops = {
|
2006-01-17 00:50:04 +08:00
|
|
|
.lo_before_commit = revoke_lo_before_commit,
|
2011-04-14 23:50:31 +08:00
|
|
|
.lo_after_commit = revoke_lo_after_commit,
|
2006-01-17 00:50:04 +08:00
|
|
|
.lo_before_scan = revoke_lo_before_scan,
|
|
|
|
|
.lo_scan_elements = revoke_lo_scan_elements,
|
|
|
|
|
.lo_after_scan = revoke_lo_after_scan,
|
2006-09-05 22:53:09 +08:00
|
|
|
.lo_name = "revoke",
|
2006-01-17 00:50:04 +08:00
|
|
|
};
|
|
|
|
|
|
2019-04-04 22:32:41 +08:00
|
|
|
static const struct gfs2_log_operations gfs2_databuf_lops = {
|
2006-01-17 00:50:04 +08:00
|
|
|
.lo_before_commit = databuf_lo_before_commit,
|
2006-02-08 19:50:51 +08:00
|
|
|
.lo_after_commit = databuf_lo_after_commit,
|
|
|
|
|
.lo_scan_elements = databuf_lo_scan_elements,
|
|
|
|
|
.lo_after_scan = databuf_lo_after_scan,
|
2006-09-05 22:53:09 +08:00
|
|
|
.lo_name = "databuf",
|
2006-01-17 00:50:04 +08:00
|
|
|
};
|
|
|
|
|
|
2006-04-07 23:17:32 +08:00
|
|
|
const struct gfs2_log_operations *gfs2_log_ops[] = {
|
2007-09-17 17:59:52 +08:00
|
|
|
&gfs2_databuf_lops,
|
2006-01-17 00:50:04 +08:00
|
|
|
&gfs2_buf_lops,
|
2007-09-17 17:59:52 +08:00
|
|
|
&gfs2_revoke_lops,
|
2006-09-05 22:53:09 +08:00
|
|
|
NULL,
|
2006-01-17 00:50:04 +08:00
|
|
|
};
|
|
|
|
|
|