aboutsummaryrefslogtreecommitdiff
path: root/fs
diff options
context:
space:
mode:
authorDavid S. Miller <davem@davemloft.net>2008-10-01 06:12:56 -0700
committerDavid S. Miller <davem@davemloft.net>2008-10-01 06:12:56 -0700
commitb262e60309e1b0eb25d300c7e739427d5316abb1 (patch)
treebf319d78c79bb5cb617ff0c8340c73aa349bba15 /fs
parent93c8b90f01f0dc73891da4e84b26524b61d29d66 (diff)
parent0523820482dcb42784572ffd2296c2f08c275a2b (diff)
Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Conflicts: drivers/net/wireless/ath9k/core.c drivers/net/wireless/ath9k/main.c net/core/dev.c
Diffstat (limited to 'fs')
-rw-r--r--fs/9p/vfs_inode.c3
-rw-r--r--fs/bfs/dir.c2
-rw-r--r--fs/dcache.c10
-rw-r--r--fs/exec.c2
-rw-r--r--fs/ocfs2/aops.c2
-rw-r--r--fs/partitions/check.c4
-rw-r--r--fs/proc/generic.c4
-rw-r--r--fs/proc/proc_misc.c12
-rw-r--r--fs/ubifs/budget.c114
-rw-r--r--fs/ubifs/debug.c2
-rw-r--r--fs/ubifs/dir.c3
-rw-r--r--fs/ubifs/file.c20
-rw-r--r--fs/ubifs/find.c19
-rw-r--r--fs/ubifs/gc.c20
-rw-r--r--fs/ubifs/misc.h49
-rw-r--r--fs/ubifs/super.c25
-rw-r--r--fs/ubifs/tnc.c116
-rw-r--r--fs/ubifs/ubifs-media.h2
-rw-r--r--fs/ubifs/ubifs.h14
-rw-r--r--fs/udf/file.c1
-rw-r--r--fs/udf/ialloc.c44
-rw-r--r--fs/xfs/linux-2.6/xfs_aops.c4
-rw-r--r--fs/xfs/linux-2.6/xfs_super.c20
-rw-r--r--fs/xfs/xfs_buf_item.c44
-rw-r--r--fs/xfs/xfs_dfrag.c9
-rw-r--r--fs/xfs/xfs_inode.c94
-rw-r--r--fs/xfs/xfs_log.c62
-rw-r--r--fs/xfs/xfs_log_priv.h1
-rw-r--r--fs/xfs/xfs_vnodeops.c26
29 files changed, 399 insertions, 329 deletions
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index c95295c6504..e83aa5ebe86 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -626,8 +626,7 @@ static struct dentry *v9fs_vfs_lookup(struct inode *dir, struct dentry *dentry,
return NULL;
error:
- if (fid)
- p9_client_clunk(fid);
+ p9_client_clunk(fid);
return ERR_PTR(result);
}
diff --git a/fs/bfs/dir.c b/fs/bfs/dir.c
index 87ee5ccee34..ed8feb052df 100644
--- a/fs/bfs/dir.c
+++ b/fs/bfs/dir.c
@@ -125,8 +125,8 @@ static int bfs_create(struct inode *dir, struct dentry *dentry, int mode,
inode->i_ino);
if (err) {
inode_dec_link_count(inode);
- iput(inode);
mutex_unlock(&info->bfs_lock);
+ iput(inode);
return err;
}
mutex_unlock(&info->bfs_lock);
diff --git a/fs/dcache.c b/fs/dcache.c
index 80e93956ace..e7a1a99b746 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -1395,6 +1395,10 @@ struct dentry * __d_lookup(struct dentry * parent, struct qstr * name)
if (dentry->d_parent != parent)
goto next;
+ /* non-existing due to RCU? */
+ if (d_unhashed(dentry))
+ goto next;
+
/*
* It is safe to compare names since d_move() cannot
* change the qstr (protected by d_lock).
@@ -1410,10 +1414,8 @@ struct dentry * __d_lookup(struct dentry * parent, struct qstr * name)
goto next;
}
- if (!d_unhashed(dentry)) {
- atomic_inc(&dentry->d_count);
- found = dentry;
- }
+ atomic_inc(&dentry->d_count);
+ found = dentry;
spin_unlock(&dentry->d_lock);
break;
next:
diff --git a/fs/exec.c b/fs/exec.c
index 32993beecbe..cecee501ce7 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -752,11 +752,11 @@ static int exec_mmap(struct mm_struct *mm)
tsk->active_mm = mm;
activate_mm(active_mm, mm);
task_unlock(tsk);
- mm_update_next_owner(old_mm);
arch_pick_mmap_layout(mm);
if (old_mm) {
up_read(&old_mm->mmap_sem);
BUG_ON(active_mm != old_mm);
+ mm_update_next_owner(old_mm);
mmput(old_mm);
return 0;
}
diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 506c24fb507..a53da146627 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -594,7 +594,7 @@ static int ocfs2_direct_IO_get_blocks(struct inode *inode, sector_t iblock,
goto bail;
}
- if (!ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb)) && !p_blkno) {
+ if (!ocfs2_sparse_alloc(OCFS2_SB(inode->i_sb)) && !p_blkno && create) {
ocfs2_error(inode->i_sb,
"Inode %llu has a hole at block %llu\n",
(unsigned long long)OCFS2_I(inode)->ip_blkno,
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 7d6b34e201d..ecc3330972e 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -499,9 +499,9 @@ int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
if (!size)
continue;
if (from + size > get_capacity(disk)) {
- printk(KERN_ERR " %s: p%d exceeds device capacity\n",
+ printk(KERN_WARNING
+ "%s: p%d exceeds device capacity\n",
disk->disk_name, p);
- continue;
}
res = add_partition(disk, p, from, size, state->parts[p].flags);
if (res) {
diff --git a/fs/proc/generic.c b/fs/proc/generic.c
index bca0f81eb68..7821589a17d 100644
--- a/fs/proc/generic.c
+++ b/fs/proc/generic.c
@@ -547,8 +547,8 @@ static int proc_register(struct proc_dir_entry * dir, struct proc_dir_entry * dp
for (tmp = dir->subdir; tmp; tmp = tmp->next)
if (strcmp(tmp->name, dp->name) == 0) {
- printk(KERN_WARNING "proc_dir_entry '%s' already "
- "registered\n", dp->name);
+ printk(KERN_WARNING "proc_dir_entry '%s/%s' already registered\n",
+ dir->name, dp->name);
dump_stack();
break;
}
diff --git a/fs/proc/proc_misc.c b/fs/proc/proc_misc.c
index 00f10a2dcf1..29e20c6b1f7 100644
--- a/fs/proc/proc_misc.c
+++ b/fs/proc/proc_misc.c
@@ -183,6 +183,9 @@ static int meminfo_read_proc(char *page, char **start, off_t off,
"SReclaimable: %8lu kB\n"
"SUnreclaim: %8lu kB\n"
"PageTables: %8lu kB\n"
+#ifdef CONFIG_QUICKLIST
+ "Quicklists: %8lu kB\n"
+#endif
"NFS_Unstable: %8lu kB\n"
"Bounce: %8lu kB\n"
"WritebackTmp: %8lu kB\n"
@@ -190,8 +193,7 @@ static int meminfo_read_proc(char *page, char **start, off_t off,
"Committed_AS: %8lu kB\n"
"VmallocTotal: %8lu kB\n"
"VmallocUsed: %8lu kB\n"
- "VmallocChunk: %8lu kB\n"
- "Quicklists: %8lu kB\n",
+ "VmallocChunk: %8lu kB\n",
K(i.totalram),
K(i.freeram),
K(i.bufferram),
@@ -216,6 +218,9 @@ static int meminfo_read_proc(char *page, char **start, off_t off,
K(global_page_state(NR_SLAB_RECLAIMABLE)),
K(global_page_state(NR_SLAB_UNRECLAIMABLE)),
K(global_page_state(NR_PAGETABLE)),
+#ifdef CONFIG_QUICKLIST
+ K(quicklist_total_size()),
+#endif
K(global_page_state(NR_UNSTABLE_NFS)),
K(global_page_state(NR_BOUNCE)),
K(global_page_state(NR_WRITEBACK_TEMP)),
@@ -223,8 +228,7 @@ static int meminfo_read_proc(char *page, char **start, off_t off,
K(committed),
(unsigned long)VMALLOC_TOTAL >> 10,
vmi.used >> 10,
- vmi.largest_chunk >> 10,
- K(quicklist_total_size())
+ vmi.largest_chunk >> 10
);
len += hugetlb_report_meminfo(page + len);
diff --git a/fs/ubifs/budget.c b/fs/ubifs/budget.c
index 15409815747..73db464cd08 100644
--- a/fs/ubifs/budget.c
+++ b/fs/ubifs/budget.c
@@ -302,18 +302,6 @@ long long ubifs_calc_available(const struct ubifs_info *c, int min_idx_lebs)
int subtract_lebs;
long long available;
- /*
- * Force the amount available to the total size reported if the used
- * space is zero.
- */
- if (c->lst.total_used <= UBIFS_INO_NODE_SZ &&
- c->budg_data_growth + c->budg_dd_growth == 0) {
- /* Do the same calculation as for c->block_cnt */
- available = c->main_lebs - 2;
- available *= c->leb_size - c->dark_wm;
- return available;
- }
-
available = c->main_bytes - c->lst.total_used;
/*
@@ -714,34 +702,106 @@ void ubifs_release_dirty_inode_budget(struct ubifs_info *c,
}
/**
- * ubifs_budg_get_free_space - return amount of free space.
+ * ubifs_reported_space - calculate reported free space.
+ * @c: the UBIFS file-system description object
+ * @free: amount of free space
+ *
+ * This function calculates amount of free space which will be reported to
+ * user-space. User-space application tend to expect that if the file-system
+ * (e.g., via the 'statfs()' call) reports that it has N bytes available, they
+ * are able to write a file of size N. UBIFS attaches node headers to each data
+ * node and it has to write indexind nodes as well. This introduces additional
+ * overhead, and UBIFS it has to report sligtly less free space to meet the
+ * above expectetion.
+ *
+ * This function assumes free space is made up of uncompressed data nodes and
+ * full index nodes (one per data node, tripled because we always allow enough
+ * space to write the index thrice).
+ *
+ * Note, the calculation is pessimistic, which means that most of the time
+ * UBIFS reports less space than it actually has.
+ */
+long long ubifs_reported_space(const struct ubifs_info *c, uint64_t free)
+{
+ int divisor, factor, f;
+
+ /*
+ * Reported space size is @free * X, where X is UBIFS block size
+ * divided by UBIFS block size + all overhead one data block
+ * introduces. The overhead is the node header + indexing overhead.
+ *
+ * Indexing overhead calculations are based on the following formula:
+ * I = N/(f - 1) + 1, where I - number of indexing nodes, N - number
+ * of data nodes, f - fanout. Because effective UBIFS fanout is twice
+ * as less than maximum fanout, we assume that each data node
+ * introduces 3 * @c->max_idx_node_sz / (@c->fanout/2 - 1) bytes.
+ * Note, the multiplier 3 is because UBIFS reseves thrice as more space
+ * for the index.
+ */
+ f = c->fanout > 3 ? c->fanout >> 1 : 2;
+ factor = UBIFS_BLOCK_SIZE;
+ divisor = UBIFS_MAX_DATA_NODE_SZ;
+ divisor += (c->max_idx_node_sz * 3) / (f - 1);
+ free *= factor;
+ do_div(free, divisor);
+ return free;
+}
+
+/**
+ * ubifs_get_free_space - return amount of free space.
* @c: UBIFS file-system description object
*
- * This function returns amount of free space on the file-system.
+ * This function calculates amount of free space to report to user-space.
+ *
+ * Because UBIFS may introduce substantial overhead (the index, node headers,
+ * alighment, wastage at the end of eraseblocks, etc), it cannot report real
+ * amount of free flash space it has (well, because not all dirty space is
+ * reclamable, UBIFS does not actually know the real amount). If UBIFS did so,
+ * it would bread user expectetion about what free space is. Users seem to
+ * accustomed to assume that if the file-system reports N bytes of free space,
+ * they would be able to fit a file of N bytes to the FS. This almost works for
+ * traditional file-systems, because they have way less overhead than UBIFS.
+ * So, to keep users happy, UBIFS tries to take the overhead into account.
*/
-long long ubifs_budg_get_free_space(struct ubifs_info *c)
+long long ubifs_get_free_space(struct ubifs_info *c)
{
- int min_idx_lebs, rsvd_idx_lebs;
+ int min_idx_lebs, rsvd_idx_lebs, lebs;
long long available, outstanding, free;
- /* Do exactly the same calculations as in 'do_budget_space()' */
spin_lock(&c->space_lock);
min_idx_lebs = ubifs_calc_min_idx_lebs(c);
+ outstanding = c->budg_data_growth + c->budg_dd_growth;
- if (min_idx_lebs > c->lst.idx_lebs)
- rsvd_idx_lebs = min_idx_lebs - c->lst.idx_lebs;
- else
- rsvd_idx_lebs = 0;
-
- if (rsvd_idx_lebs > c->lst.empty_lebs + c->freeable_cnt + c->idx_gc_cnt
- - c->lst.taken_empty_lebs) {
+ /*
+ * Force the amount available to the total size reported if the used
+ * space is zero.
+ */
+ if (c->lst.total_used <= UBIFS_INO_NODE_SZ && !outstanding) {
spin_unlock(&c->space_lock);
- return 0;
+ return (long long)c->block_cnt << UBIFS_BLOCK_SHIFT;
}
available = ubifs_calc_available(c, min_idx_lebs);
- outstanding = c->budg_data_growth + c->budg_dd_growth;
- c->min_idx_lebs = min_idx_lebs;
+
+ /*
+ * When reporting free space to user-space, UBIFS guarantees that it is
+ * possible to write a file of free space size. This means that for
+ * empty LEBs we may use more precise calculations than
+ * 'ubifs_calc_available()' is using. Namely, we know that in empty
+ * LEBs we would waste only @c->leb_overhead bytes, not @c->dark_wm.
+ * Thus, amend the available space.
+ *
+ * Note, the calculations below are similar to what we have in
+ * 'do_budget_space()', so refer there for comments.
+ */
+ if (min_idx_lebs > c->lst.idx_lebs)
+ rsvd_idx_lebs = min_idx_lebs - c->lst.idx_lebs;
+ else
+ rsvd_idx_lebs = 0;
+ lebs = c->lst.empty_lebs + c->freeable_cnt + c->idx_gc_cnt -
+ c->lst.taken_empty_lebs;
+ lebs -= rsvd_idx_lebs;
+ available += lebs * (c->dark_wm - c->leb_overhead);
spin_unlock(&c->space_lock);
if (available > outstanding)
diff --git a/fs/ubifs/debug.c b/fs/ubifs/debug.c
index b9cb7747375..d7f7645779f 100644
--- a/fs/ubifs/debug.c
+++ b/fs/ubifs/debug.c
@@ -538,7 +538,7 @@ void dbg_dump_node(const struct ubifs_info *c, const void *node)
printk(KERN_DEBUG "\t%d orphan inode numbers:\n", n);
for (i = 0; i < n; i++)
printk(KERN_DEBUG "\t ino %llu\n",
- le64_to_cpu(orph->inos[i]));
+ (unsigned long long)le64_to_cpu(orph->inos[i]));
break;
}
default:
diff --git a/fs/ubifs/dir.c b/fs/ubifs/dir.c
index 5c96f1fb701..526c01ec800 100644
--- a/fs/ubifs/dir.c
+++ b/fs/ubifs/dir.c
@@ -426,7 +426,7 @@ static int ubifs_readdir(struct file *file, void *dirent, filldir_t filldir)
while (1) {
dbg_gen("feed '%s', ino %llu, new f_pos %#x",
- dent->name, le64_to_cpu(dent->inum),
+ dent->name, (unsigned long long)le64_to_cpu(dent->inum),
key_hash_flash(c, &dent->key));
ubifs_assert(dent->ch.sqnum > ubifs_inode(dir)->creat_sqnum);
@@ -587,7 +587,6 @@ static int ubifs_unlink(struct inode *dir, struct dentry *dentry)
if (err) {
if (err != -ENOSPC)
return err;
- err = 0;
budgeted = 0;
}
diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c
index 4071d1cae29..3d698e2022b 100644
--- a/fs/ubifs/file.c
+++ b/fs/ubifs/file.c
@@ -793,7 +793,7 @@ static int do_truncation(struct ubifs_info *c, struct inode *inode,
int err;
struct ubifs_budget_req req;
loff_t old_size = inode->i_size, new_size = attr->ia_size;
- int offset = new_size & (UBIFS_BLOCK_SIZE - 1);
+ int offset = new_size & (UBIFS_BLOCK_SIZE - 1), budgeted = 1;
struct ubifs_inode *ui = ubifs_inode(inode);
dbg_gen("ino %lu, size %lld -> %lld", inode->i_ino, old_size, new_size);
@@ -811,8 +811,15 @@ static int do_truncation(struct ubifs_info *c, struct inode *inode,
/* A funny way to budget for truncation node */
req.dirtied_ino_d = UBIFS_TRUN_NODE_SZ;
err = ubifs_budget_space(c, &req);
- if (err)
- return err;
+ if (err) {
+ /*
+ * Treat truncations to zero as deletion and always allow them,
+ * just like we do for '->unlink()'.
+ */
+ if (new_size || err != -ENOSPC)
+ return err;
+ budgeted = 0;
+ }
err = vmtruncate(inode, new_size);
if (err)
@@ -869,7 +876,12 @@ static int do_truncation(struct ubifs_info *c, struct inode *inode,
err = ubifs_jnl_truncate(c, inode, old_size, new_size);
mutex_unlock(&ui->ui_mutex);
out_budg:
- ubifs_release_budget(c, &req);
+ if (budgeted)
+ ubifs_release_budget(c, &req);
+ else {
+ c->nospace = c->nospace_rp = 0;
+ smp_wmb();
+ }
return err;
}
diff --git a/fs/ubifs/find.c b/fs/ubifs/find.c
index adee7b5ddea..47814cde240 100644
--- a/fs/ubifs/find.c
+++ b/fs/ubifs/find.c
@@ -211,14 +211,8 @@ static const struct ubifs_lprops *scan_for_dirty(struct ubifs_info *c,
* dirty index heap, and it falls-back to LPT scanning if the heaps are empty
* or do not have an LEB which satisfies the @min_space criteria.
*
- * Note:
- * o LEBs which have less than dead watermark of dirty space are never picked
- * by this function;
- *
- * Returns zero and the LEB properties of
- * found dirty LEB in case of success, %-ENOSPC if no dirty LEB was found and a
- * negative error code in case of other failures. The returned LEB is marked as
- * "taken".
+ * Note, LEBs which have less than dead watermark of free + dirty space are
+ * never picked by this function.
*
* The additional @pick_free argument controls if this function has to return a
* free or freeable LEB if one is present. For example, GC must to set it to %1,
@@ -231,6 +225,10 @@ static const struct ubifs_lprops *scan_for_dirty(struct ubifs_info *c,
*
* In addition @pick_free is set to %2 by the recovery process in order to
* recover gc_lnum in which case an index LEB must not be returned.
+ *
+ * This function returns zero and the LEB properties of found dirty LEB in case
+ * of success, %-ENOSPC if no dirty LEB was found and a negative error code in
+ * case of other failures. The returned LEB is marked as "taken".
*/
int ubifs_find_dirty_leb(struct ubifs_info *c, struct ubifs_lprops *ret_lp,
int min_space, int pick_free)
@@ -245,7 +243,7 @@ int ubifs_find_dirty_leb(struct ubifs_info *c, struct ubifs_lprops *ret_lp,
int lebs, rsvd_idx_lebs = 0;
spin_lock(&c->space_lock);
- lebs = c->lst.empty_lebs;
+ lebs = c->lst.empty_lebs + c->idx_gc_cnt;
lebs += c->freeable_cnt - c->lst.taken_empty_lebs;
/*
@@ -317,7 +315,7 @@ int ubifs_find_dirty_leb(struct ubifs_info *c, struct ubifs_lprops *ret_lp,
lp = idx_lp;
if (lp) {
- ubifs_assert(lp->dirty >= c->dead_wm);
+ ubifs_assert(lp->free + lp->dirty >= c->dead_wm);
goto found;
}
@@ -509,7 +507,6 @@ int ubifs_find_free_space(struct ubifs_info *c, int min_space, int *free,
rsvd_idx_lebs = 0;
lebs = c->lst.empty_lebs + c->freeable_cnt + c->idx_gc_cnt -
c->lst.taken_empty_lebs;
- ubifs_assert(lebs + c->lst.idx_lebs >= c->min_idx_lebs);
if (rsvd_idx_lebs < lebs)
/*
* OK to allocate an empty LEB, but we still don't want to go
diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c
index d0f3dac2908..02aba36fe3d 100644
--- a/fs/ubifs/gc.c
+++ b/fs/ubifs/gc.c
@@ -334,15 +334,21 @@ int ubifs_garbage_collect_leb(struct ubifs_info *c, struct ubifs_lprops *lp)
err = move_nodes(c, sleb);
if (err)
- goto out;
+ goto out_inc_seq;
err = gc_sync_wbufs(c);
if (err)
- goto out;
+ goto out_inc_seq;
err = ubifs_change_one_lp(c, lnum, c->leb_size, 0, 0, 0, 0);
if (err)
- goto out;
+ goto out_inc_seq;
+
+ /* Allow for races with TNC */
+ c->gced_lnum = lnum;
+ smp_wmb();
+ c->gc_seq += 1;
+ smp_wmb();
if (c->gc_lnum == -1) {
c->gc_lnum = lnum;
@@ -363,6 +369,14 @@ int ubifs_garbage_collect_leb(struct ubifs_info *c, struct ubifs_lprops *lp)
out:
ubifs_scan_destroy(sleb);
return err;
+
+out_inc_seq:
+ /* We may have moved at least some nodes so allow for races with TNC */
+ c->gced_lnum = lnum;
+ smp_wmb();
+ c->gc_seq += 1;
+ smp_wmb();
+ goto out;
}
/**
diff --git a/fs/ubifs/misc.h b/fs/ubifs/misc.h
index 87dabf9fe74..4c12a9215d7 100644
--- a/fs/ubifs/misc.h
+++ b/fs/ubifs/misc.h
@@ -284,38 +284,6 @@ static inline void *ubifs_idx_key(const struct ubifs_info *c,
}
/**
- * ubifs_reported_space - calculate reported free space.
- * @c: the UBIFS file-system description object
- * @free: amount of free space
- *
- * This function calculates amount of free space which will be reported to
- * user-space. User-space application tend to expect that if the file-system
- * (e.g., via the 'statfs()' call) reports that it has N bytes available, they
- * are able to write a file of size N. UBIFS attaches node headers to each data
- * node and it has to write indexind nodes as well. This introduces additional
- * overhead, and UBIFS it has to report sligtly less free space to meet the
- * above expectetion.
- *
- * This function assumes free space is made up of uncompressed data nodes and
- * full index nodes (one per data node, doubled because we always allow enough
- * space to write the index twice).
- *
- * Note, the calculation is pessimistic, which means that most of the time
- * UBIFS reports less space than it actually has.
- */
-static inline long long ubifs_reported_space(const struct ubifs_info *c,
- uint64_t free)
-{
- int divisor, factor;
-
- divisor = UBIFS_MAX_DATA_NODE_SZ + (c->max_idx_node_sz * 3);
- factor = UBIFS_MAX_DATA_NODE_SZ - UBIFS_DATA_NODE_SZ;
- do_div(free, divisor);
-
- return free * factor;
-}
-
-/**
* ubifs_current_time - round current time to time granularity.
* @inode: inode
*/
@@ -325,4 +293,21 @@ static inline struct timespec ubifs_current_time(struct inode *inode)
current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
}
+/**
+ * ubifs_tnc_lookup - look up a file-system node.
+ * @c: UBIFS file-system description object
+ * @key: node key to lookup
+ * @node: the node is returned here
+ *
+ * This function look up and reads node with key @key. The caller has to make
+ * sure the @node buffer is large enough to fit the node. Returns zero in case
+ * of success, %-ENOENT if the node was not found, and a negative error code in
+ * case of failure.
+ */
+static inline int ubifs_tnc_lookup(struct ubifs_info *c,
+ const union ubifs_key *key, void *node)
+{
+ return ubifs_tnc_locate(c, key, node, NULL, NULL);
+}
+
#endif /* __UBIFS_MISC_H__ */
diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index f71e6b8822c..3f4902060c7 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -370,8 +370,9 @@ static int ubifs_statfs(struct dentry *dentry, struct kstatfs *buf)
{
struct ubifs_info *c = dentry->d_sb->s_fs_info;
unsigned long long free;
+ __le32 *uuid = (__le32 *)c->uuid;
- free = ubifs_budg_get_free_space(c);
+ free = ubifs_get_free_space(c);
dbg_gen("free space %lld bytes (%lld blocks)",
free, free >> UBIFS_BLOCK_SHIFT);
@@ -386,7 +387,8 @@ static int ubifs_statfs(struct dentry *dentry, struct kstatfs *buf)
buf->f_files = 0;
buf->f_ffree = 0;
buf->f_namelen = UBIFS_MAX_NLEN;
-
+ buf->f_fsid.val[0] = le32_to_cpu(uuid[0]) ^ le32_to_cpu(uuid[2]);
+ buf->f_fsid.val[1] = le32_to_cpu(uuid[1]) ^ le32_to_cpu(uuid[3]);
return 0;
}
@@ -530,6 +532,12 @@ static int init_constants_early(struct ubifs_info *c)
c->dead_wm = ALIGN(MIN_WRITE_SZ, c->min_io_size);
c->dark_wm = ALIGN(UBIFS_MAX_NODE_SZ, c->min_io_size);
+ /*
+ * Calculate how many bytes would be wasted at the end of LEB if it was
+ * fully filled with data nodes of maximum size. This is used in
+ * calculations when reporting free space.
+ */
+ c->leb_overhead = c->leb_size % UBIFS_MAX_DATA_NODE_SZ;
return 0;
}
@@ -647,13 +655,11 @@ static int init_constants_late(struct ubifs_info *c)
* internally because it does not make much sense for UBIFS, but it is
* necessary to report something for the 'statfs()' call.
*
- * Subtract the LEB reserved for GC and the LEB which is reserved for
- * deletions.
- *
- * Review 'ubifs_calc_available()' if changing this calculation.
+ * Subtract the LEB reserved for GC, the LEB which is reserved for
+ * deletions, and assume only one journal head is available.
*/
- tmp64 = c->main_lebs - 2;
- tmp64 *= (uint64_t)c->leb_size - c->dark_wm;
+ tmp64 = c->main_lebs - 2 - c->jhead_cnt + 1;
+ tmp64 *= (uint64_t)c->leb_size - c->leb_overhead;
tmp64 = ubifs_reported_space(c, tmp64);
c->block_cnt = tmp64 >> UBIFS_BLOCK_SHIFT;
@@ -1018,14 +1024,13 @@ static int mount_ubifs(struct ubifs_info *c)
goto out_dereg;
}
+ sprintf(c->bgt_name, BGT_NAME_PATTERN, c->vi.ubi_num, c->vi.vol_id);
if (!mounted_read_only) {
err = alloc_wbufs(c);
if (err)
goto out_cbuf;
/* Create background thread */
- sprintf(c->bgt_name, BGT_NAME_PATTERN, c->vi.ubi_num,
- c->vi.vol_id);
c->bgt = kthread_create(ubifs_bg_thread, c, c->bgt_name);
if (!c->bgt)
c->bgt = ERR_PTR(-EINVAL);
diff --git a/fs/ubifs/tnc.c b/fs/ubifs/tnc.c
index e909f4a9644..7634c597088 100644
--- a/fs/ubifs/tnc.c
+++ b/fs/ubifs/tnc.c
@@ -506,7 +506,7 @@ static int fallible_read_node(struct ubifs_info *c, const union ubifs_key *key,
if (keys_cmp(c, key, &node_key) != 0)
ret = 0;
}
- if (ret == 0)
+ if (ret == 0 && c->replaying)
dbg_mnt("dangling branch LEB %d:%d len %d, key %s",
zbr->lnum, zbr->offs, zbr->len, DBGKEY(key));
return ret;
@@ -1382,50 +1382,39 @@ static int lookup_level0_dirty(struct ubifs_info *c, const union ubifs_key *key,
}
/**
- * ubifs_tnc_lookup - look up a file-system node.
+ * maybe_leb_gced - determine if a LEB may have been garbage collected.
* @c: UBIFS file-system description object
- * @key: node key to lookup
- * @node: the node is returned here
+ * @lnum: LEB number
+ * @gc_seq1: garbage collection sequence number
*
- * This function look up and reads node with key @key. The caller has to make
- * sure the @node buffer is large enough to fit the node. Returns zero in case
- * of success, %-ENOENT if the node was not found, and a negative error code in
- * case of failure.
+ * This function determines if @lnum may have been garbage collected since
+ * sequence number @gc_seq1. If it may have been then %1 is returned, otherwise
+ * %0 is returned.
*/
-int ubifs_tnc_lookup(struct ubifs_info *c, const union ubifs_key *key,
- void *node)
+static int maybe_leb_gced(struct ubifs_info *c, int lnum, int gc_seq1)
{
- int found, n, err;
- struct ubifs_znode *znode;
- struct ubifs_zbranch zbr, *zt;
+ int gc_seq2, gced_lnum;
- mutex_lock(&c->tnc_mutex);
- found = ubifs_lookup_level0(c, key, &znode, &n);
- if (!found) {
- err = -ENOENT;
- goto out;
- } else if (found < 0) {
- err = found;
- goto out;
- }
- zt = &znode->zbranch[n];
- if (is_hash_key(c, key)) {
- /*
- * In this case the leaf node cache gets used, so we pass the
- * address of the zbranch and keep the mutex locked
- */
- err = tnc_read_node_nm(c, zt, node);
- goto out;
- }
- zbr = znode->zbranch[n];
- mutex_unlock(&c->tnc_mutex);
-
- err = ubifs_tnc_read_node(c, &zbr, node);
- return err;
-
-out:
- mutex_unlock(&c->tnc_mutex);
- return err;
+ gced_lnum = c->gced_lnum;
+ smp_rmb();
+ gc_seq2 = c->gc_seq;
+ /* Same seq means no GC */
+ if (gc_seq1 == gc_seq2)
+ return 0;
+ /* Different by more than 1 means we don't know */
+ if (gc_seq1 + 1 != gc_seq2)
+ return 1;
+ /*
+ * We have seen the sequence number has increased by 1. Now we need to
+ * be sure we read the right LEB number, so read it again.
+ */
+ smp_rmb();
+ if (gced_lnum != c->gced_lnum)
+ return 1;
+ /* Finally we can check lnum */
+ if (gced_lnum == lnum)
+ return 1;
+ return 0;
}
/**
@@ -1436,16 +1425,19 @@ out:
* @lnum: LEB number is returned here
* @offs: offset is returned here
*
- * This function is the same as 'ubifs_tnc_lookup()' but it returns the node
- * location also. See 'ubifs_tnc_lookup()'.
+ * This function look up and reads node with key @key. The caller has to make
+ * sure the @node buffer is large enough to fit the node. Returns zero in case
+ * of success, %-ENOENT if the node was not found, and a negative error code in
+ * case of failure. The node location can be returned in @lnum and @offs.
*/
int ubifs_tnc_locate(struct ubifs_info *c, const union ubifs_key *key,
void *node, int *lnum, int *offs)
{
- int found, n, err;
+ int found, n, err, safely = 0, gc_seq1;
struct ubifs_znode *znode;
struct ubifs_zbranch zbr, *zt;
+again:
mutex_lock(&c->tnc_mutex);
found = ubifs_lookup_level0(c, key, &znode, &n);
if (!found) {
@@ -1456,24 +1448,43 @@ int ubifs_tnc_locate(struct ubifs_info *c, const union ubifs_key *key,
goto out;
}
zt = &znode->zbranch[n];
+ if (lnum) {
+ *lnum = zt->lnum;
+ *offs = zt->offs;
+ }
if (is_hash_key(c, key)) {
/*
* In this case the leaf node cache gets used, so we pass the
* address of the zbranch and keep the mutex locked
*/
- *lnum = zt->lnum;
- *offs = zt->offs;
err = tnc_read_node_nm(c, zt, node);
goto out;
}
+ if (safely) {
+ err = ubifs_tnc_read_node(c, zt, node);
+ goto out;
+ }
+ /* Drop the TNC mutex prematurely and race with garbage collection */
zbr = znode->zbranch[n];
+ gc_seq1 = c->gc_seq;
mutex_unlock(&c->tnc_mutex);
- *lnum = zbr.lnum;
- *offs = zbr.offs;
+ if (ubifs_get_wbuf(c, zbr.lnum)) {
+ /* We do not GC journal heads */
+ err = ubifs_tnc_read_node(c, &zbr, node);
+ return err;
+ }
- err = ubifs_tnc_read_node(c, &zbr, node);
- return err;
+ err = fallible_read_node(c, key, &zbr, node);
+ if (err <= 0 || maybe_leb_gced(c, zbr.lnum, gc_seq1)) {
+ /*
+ * The node may have been GC'ed out from under us so try again
+ * while keeping the TNC mutex locked.
+ */
+ safely = 1;
+ goto again;
+ }
+ return 0;
out:
mutex_unlock(&c->tnc_mutex);
@@ -1498,7 +1509,6 @@ static int do_lookup_nm(struct ubifs_info *c, const union ubifs_key *key,
{
int found, n, err;
struct ubifs_znode *znode;
- struct ubifs_zbranch zbr;
dbg_tnc("name '%.*s' key %s", nm->len, nm->name, DBGKEY(key));
mutex_lock(&c->tnc_mutex);
@@ -1522,11 +1532,7 @@ static int do_lookup_nm(struct ubifs_info *c, const union ubifs_key *key,
goto out_unlock;
}
- zbr = znode->zbranch[n];
- mutex_unlock(&c->tnc_mutex);
-
- err = tnc_read_node_nm(c, &zbr, node);
- return err;
+ err = tnc_read_node_nm(c, &znode->zbranch[n], node);
out_unlock:
mutex_unlock(&c->tnc_mutex);
diff --git a/fs/ubifs/ubifs-media.h b/fs/ubifs/ubifs-media.h
index bd2121f3426..a9ecbd9af20 100644
--- a/fs/ubifs/ubifs-media.h
+++ b/fs/ubifs/ubifs-media.h
@@ -87,7 +87,7 @@
#define UBIFS_SK_LEN 8
/* Minimum index tree fanout */
-#define UBIFS_MIN_FANOUT 2
+#define UBIFS_MIN_FANOUT 3
/* Maximum number of levels in UBIFS indexing B-tree */
#define UBIFS_MAX_LEVELS 512
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index d7f706f7a30..17c620b93ee 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -995,6 +995,9 @@ struct ubifs_mount_opts {
* @max_idx_node_sz: maximum indexing node aligned on 8-bytes boundary
* @max_inode_sz: maximum possible inode size in bytes
* @max_znode_sz: size of znode in bytes
+ *
+ * @leb_overhead: how many bytes are wasted in an LEB when it is filled with
+ * data nodes of maximum size - used in free space reporting
* @dead_wm: LEB dead space watermark
* @dark_wm: LEB dark space watermark
* @block_cnt: count of 4KiB blocks on the FS
@@ -1028,6 +1031,8 @@ struct ubifs_mount_opts {
* @sbuf: a buffer of LEB size used by GC and replay for scanning
* @idx_gc: list of index LEBs that have been garbage collected
* @idx_gc_cnt: number of elements on the idx_gc list
+ * @gc_seq: incremented for every non-index LEB garbage collected
+ * @gced_lnum: last non-index LEB that was garbage collected
*
* @infos_list: links all 'ubifs_info' objects
* @umount_mutex: serializes shrinker and un-mount
@@ -1224,6 +1229,8 @@ struct ubifs_info {
int max_idx_node_sz;
long long max_inode_sz;
int max_znode_sz;
+
+ int leb_overhead;
int dead_wm;
int dark_wm;
int block_cnt;
@@ -1257,6 +1264,8 @@ struct ubifs_info {
void *sbuf;
struct list_head idx_gc;
int idx_gc_cnt;
+ volatile int gc_seq;
+ volatile int gced_lnum;
struct list_head infos_list;
struct mutex umount_mutex;
@@ -1434,9 +1443,10 @@ void ubifs_release_ino_dirty(struct ubifs_info *c, struct inode *inode,
struct ubifs_budget_req *req);
void ubifs_cancel_ino_op(struct ubifs_info *c, struct inode *inode,
struct ubifs_budget_req *req);
-long long ubifs_budg_get_free_space(struct ubifs_info *c);
+long long ubifs_get_free_space(struct ubifs_info *c);
int ubifs_calc_min_idx_lebs(struct ubifs_info *c);
void ubifs_convert_page_budget(struct ubifs_info *c);
+long long ubifs_reported_space(const struct ubifs_info *c, uint64_t free);
long long ubifs_calc_available(const struct ubifs_info *c, int min_idx_lebs);
/* find.c */
@@ -1451,8 +1461,6 @@ int ubifs_save_dirty_idx_lnums(struct ubifs_info *c);
/* tnc.c */
int ubifs_lookup_level0(struct ubifs_info *c, const union ubifs_key *key,
struct ubifs_znode **zn, int *n);
-int ubifs_tnc_lookup(struct ubifs_info *c, const union ubifs_key *key,
- void *node);
int ubifs_tnc_lookup_nm(struct ubifs_info *c, const union ubifs_key *key,
void *node, const struct qstr *nm);
int ubifs_tnc_locate(struct ubifs_info *c, const union ubifs_key *key,
diff --git a/fs/udf/file.c b/fs/udf/file.c
index 0ed6e146a0d..eb91f3b7032 100644
--- a/fs/udf/file.c
+++ b/fs/udf/file.c
@@ -211,6 +211,7 @@ const struct file_operations udf_file_operations = {
.release = udf_release_file,
.fsync = udf_fsync_file,
.splice_read = generic_file_splice_read,
+ .llseek = generic_file_llseek,
};
const struct inode_operations udf_file_inode_operations = {
diff --git a/fs/udf/ialloc.c b/fs/udf/ialloc.c
index eb9cfa23dc3..a4f2b3ce45b 100644
--- a/fs/udf/ialloc.c
+++ b/fs/udf/ialloc.c
@@ -76,11 +76,24 @@ struct inode *udf_new_inode(struct inode *dir, int mode, int *err)
*err = -ENOSPC;
iinfo = UDF_I(inode);
- iinfo->i_unique = 0;
- iinfo->i_lenExtents = 0;
- iinfo->i_next_alloc_block = 0;
- iinfo->i_next_alloc_goal = 0;
- iinfo->i_strat4096 = 0;
+ if (UDF_QUERY_FLAG(inode->i_sb, UDF_FLAG_USE_EXTENDED_FE)) {
+ iinfo->i_efe = 1;
+ if (UDF_VERS_USE_EXTENDED_FE > sbi->s_udfrev)
+ sbi->s_udfrev = UDF_VERS_USE_EXTENDED_FE;
+ iinfo->i_ext.i_data = kzalloc(inode->i_sb->s_blocksize -
+ sizeof(struct extendedFileEntry),
+ GFP_KERNEL);
+ } else {
+ iinfo->i_efe = 0;
+ iinfo->i_ext.i_data = kzalloc(inode->i_sb->s_blocksize -
+ sizeof(struct fileEntry),
+ GFP_KERNEL);
+ }
+ if (!iinfo->i_ext.i_data) {
+ iput(inode);
+ *err = -ENOMEM;
+ return NULL;
+ }
block = udf_new_block(dir->i_sb, NULL,
dinfo->i_location.partitionReferenceNum,
@@ -111,6 +124,7 @@ struct inode *udf_new_inode(struct inode *dir, int mode, int *err)
lvhd->uniqueID = cpu_to_le64(uniqueID);
mark_buffer_dirty(sbi->s_lvid_bh);
}
+ mutex_unlock(&sbi->s_alloc_mutex);
inode->i_mode = mode;
inode->i_uid = current->fsuid;
if (dir->i_mode & S_ISGID) {
@@ -129,25 +143,6 @@ struct inode *udf_new_inode(struct inode *dir, int mode, int *err)
iinfo->i_lenEAttr = 0;
iinfo->i_lenAlloc = 0;
iinfo->i_use = 0;
- if (UDF_QUERY_FLAG(inode->i_sb, UDF_FLAG_USE_EXTENDED_FE)) {
- iinfo->i_efe = 1;
- if (UDF_VERS_USE_EXTENDED_FE > sbi->s_udfrev)
- sbi->s_udfrev = UDF_VERS_USE_EXTENDED_FE;
- iinfo->i_ext.i_data = kzalloc(inode->i_sb->s_blocksize -
- sizeof(struct extendedFileEntry),
- GFP_KERNEL);
- } else {
- iinfo->i_efe = 0;
- iinfo->i_ext.i_data = kzalloc(inode->i_sb->s_blocksize -
- sizeof(struct fileEntry),
- GFP_KERNEL);
- }
- if (!iinfo->i_ext.i_data) {
- iput(inode);
- *err = -ENOMEM;
- mutex_unlock(&sbi->s_alloc_mutex);
- return NULL;
- }
if (UDF_QUERY_FLAG(inode->i_sb, UDF_FLAG_USE_AD_IN_ICB))
iinfo->i_alloc_type = ICBTAG_FLAG_AD_IN_ICB;
else if (UDF_QUERY_FLAG(inode->i_sb, UDF_FLAG_USE_SHORT_AD))
@@ -158,7 +153,6 @@ struct inode *udf_new_inode(struct inode *dir, int mode, int *err)
iinfo->i_crtime = current_fs_time(inode->i_sb);
insert_inode_hash(inode);
mark_inode_dirty(inode);
- mutex_unlock(&sbi->s_alloc_mutex);
if (DQUOT_ALLOC_INODE(inode)) {
DQUOT_DROP(inode);
diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index f42f80a3b1f..a44d68eb50b 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -1338,6 +1338,10 @@ __xfs_get_blocks(
offset = (xfs_off_t)iblock << inode->i_blkbits;
ASSERT(bh_result->b_size >= (1 << inode->i_blkbits));
size = bh_result->b_size;
+
+ if (!create && direct && offset >= i_size_read(inode))
+ return 0;
+
error = xfs_iomap(XFS_I(inode), offset, size,
create ? flags : BMAPI_READ, &iomap, &niomap);
if (error)
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 73c65f19e54..18d3c848783 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -1302,9 +1302,29 @@ xfs_fs_remount(
mp->m_flags &= ~XFS_MOUNT_BARRIER;
break;
default:
+ /*
+ * Logically we would return an error here to prevent
+ * users from believing they might have changed
+ * mount options using remount which can't be changed.
+ *
+ * But unfortunately mount(8) adds all options from
+ * mtab and fstab to the mount arguments in some cases
+ * so we can't blindly reject options, but have to
+ * check for each specified option if it actually
+ * differs from the currently set option and only
+ * reject it if that's the case.
+ *
+ * Until that is implemented we return success for
+ * every remount request, and silently ignore all
+ * options that we can't actually change.
+ */
+#if 0
printk(KERN_INFO
"XFS: mount option \"%s\" not supported for remount\n", p);
return -EINVAL;
+#else
+ return 0;
+#endif
}
}
diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c
index 608c30c3f76..002fc2617c8 100644
--- a/fs/xfs/xfs_buf_item.c
+++ b/fs/xfs/xfs_buf_item.c
@@ -732,6 +732,7 @@ xfs_buf_item_init(
bip->bli_item.li_ops = &xfs_buf_item_ops;
bip->bli_item.li_mountp = mp;
bip->bli_buf = bp;
+ xfs_buf_hold(bp);
bip->bli_format.blf_type = XFS_LI_BUF;
bip->bli_format.blf_blkno = (__int64_t)XFS_BUF_ADDR(bp);
bip->bli_format.blf_len = (ushort)BTOBB(XFS_BUF_COUNT(bp));
@@ -867,6 +868,21 @@ xfs_buf_item_dirty(
return (bip->bli_flags & XFS_BLI_DIRTY);
}
+STATIC void
+xfs_buf_item_free(
+ xfs_buf_log_item_t *bip)
+{
+#ifdef XFS_TRANS_DEBUG
+ kmem_free(bip->bli_orig);
+ kmem_free(bip->bli_logged);
+#endif /* XFS_TRANS_DEBUG */
+
+#ifdef XFS_BLI_TRACE
+ ktrace_free(bip->bli_trace);
+#endif
+ kmem_zone_free(xfs_buf_item_zone, bip);
+}
+
/*
* This is called when the buf log item is no longer needed. It should
* free the buf log item associated with the given buffer and clear
@@ -887,18 +903,8 @@ xfs_buf_item_relse(
(XFS_BUF_IODONE_FUNC(bp) != NULL)) {
XFS_BUF_CLR_IODONE_FUNC(bp);
}
-
-#ifdef XFS_TRANS_DEBUG
- kmem_free(bip->bli_orig);
- bip->bli_orig = NULL;
- kmem_free(bip->bli_logged);
- bip->bli_logged = NULL;
-#endif /* XFS_TRANS_DEBUG */
-
-#ifdef XFS_BLI_TRACE
- ktrace_free(bip->bli_trace);
-#endif
- kmem_zone_free(xfs_buf_item_zone, bip);
+ xfs_buf_rele(bp);
+ xfs_buf_item_free(bip);
}
@@ -1120,6 +1126,7 @@ xfs_buf_iodone(
ASSERT(bip->bli_buf == bp);
+ xfs_buf_rele(bp);
mp = bip->bli_item.li_mountp;
/*
@@ -1136,18 +1143,7 @@ xfs_buf_iodone(
* xfs_trans_delete_ail() drops the AIL lock.
*/
xfs_trans_delete_ail(mp, (xfs_log_item_t *)bip);
-
-#ifdef XFS_TRANS_DEBUG
- kmem_free(bip->bli_orig);
- bip->bli_orig = NULL;
- kmem_free(bip->bli_logged);
- bip->bli_logged = NULL;
-#endif /* XFS_TRANS_DEBUG */
-
-#ifdef XFS_BLI_TRACE
- ktrace_free(bip->bli_trace);
-#endif
- kmem_zone_free(xfs_buf_item_zone, bip);
+ xfs_buf_item_free(bip);
}
#if defined(XFS_BLI_TRACE)
diff --git a/fs/xfs/xfs_dfrag.c b/fs/xfs/xfs_dfrag.c
index 760f4c5b516..75b0cd4da0e 100644
--- a/fs/xfs/xfs_dfrag.c
+++ b/fs/xfs/xfs_dfrag.c
@@ -149,7 +149,14 @@ xfs_swap_extents(
sbp = &sxp->sx_stat;
- xfs_lock_two_inodes(ip, tip, lock_flags);
+ /*
+ * we have to do two separate lock calls here to keep lockdep
+ * happy. If we try to get all the locks in one call, lock will
+ * report false positives when we drop the ILOCK and regain them
+ * below.
+ */
+ xfs_lock_two_inodes(ip, tip, XFS_IOLOCK_EXCL);
+ xfs_lock_two_inodes(ip, tip, XFS_ILOCK_EXCL);
locked = 1;
/* Verify that both files have the same format */
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 00e80df9dd9..dbd9cef852e 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -4118,7 +4118,7 @@ xfs_iext_indirect_to_direct(
ASSERT(nextents <= XFS_LINEAR_EXTS);
size = nextents * sizeof(xfs_bmbt_rec_t);
- xfs_iext_irec_compact_full(ifp);
+ xfs_iext_irec_compact_pages(ifp);
ASSERT(ifp->if_real_bytes == XFS_IEXT_BUFSZ);
ep = ifp->if_u1.if_ext_irec->er_extbuf;
@@ -4449,8 +4449,7 @@ xfs_iext_irec_remove(
* compaction policy is as follows:
*
* Full Compaction: Extents fit into a single page (or inline buffer)
- * Full Compaction: Extents occupy less than 10% of allocated space
- * Partial Compaction: Extents occupy > 10% and < 50% of allocated space
+ * Partial Compaction: Extents occupy less than 50% of allocated space
* No Compaction: Extents occupy at least 50% of allocated space
*/
void
@@ -4471,8 +4470,6 @@ xfs_iext_irec_compact(
xfs_iext_direct_to_inline(ifp, nextents);
} else if (nextents <= XFS_LINEAR_EXTS) {
xfs_iext_indirect_to_direct(ifp);
- } else if (nextents < (nlists * XFS_LINEAR_EXTS) >> 3) {
- xfs_iext_irec_compact_full(ifp);
} else if (nextents < (nlists * XFS_LINEAR_EXTS) >> 1) {
xfs_iext_irec_compact_pages(ifp);
}
@@ -4496,7 +4493,7 @@ xfs_iext_irec_compact_pages(
erp_next = erp + 1;
if (erp_next->er_extcount <=
(XFS_LINEAR_EXTS - erp->er_extcount)) {
- memmove(&erp->er_extbuf[erp->er_extcount],
+ memcpy(&erp->er_extbuf[erp->er_extcount],
erp_next->er_extbuf, erp_next->er_extcount *
sizeof(xfs_bmbt_rec_t));
erp->er_extcount += erp_next->er_extcount;
@@ -4516,91 +4513,6 @@ xfs_iext_irec_compact_pages(
}
/*
- * Fully compact the extent records managed by the indirection array.
- */
-void
-xfs_iext_irec_compact_full(
- xfs_ifork_t *ifp) /* inode fork pointer */
-{
- xfs_bmbt_rec_host_t *ep, *ep_next; /* extent record pointers */
- xfs_ext_irec_t *erp, *erp_next; /* extent irec pointers */
- int erp_idx = 0; /* extent irec index */
- int ext_avail; /* empty entries in ex list */
- int ext_diff; /* number of exts to add */
- int nlists; /* number of irec's (ex lists) */
-
- ASSERT(ifp->if_flags & XFS_IFEXTIREC);
-
- nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
- erp = ifp->if_u1.if_ext_irec;
- ep = &erp->er_extbuf[erp->er_extcount];
- erp_next = erp + 1;
- ep_next = erp_next->er_extbuf;
-
- while (erp_idx < nlists - 1) {
- /*
- * Check how many extent records are available in this irec.
- * If there is none skip the whole exercise.
- */
- ext_avail = XFS_LINEAR_EXTS - erp->er_extcount;
- if (ext_avail) {
-
- /*
- * Copy over as many as possible extent records into
- * the previous page.
- */
- ext_diff = MIN(ext_avail, erp_next->er_extcount);
- memcpy(ep, ep_next, ext_diff * sizeof(xfs_bmbt_rec_t));
- erp->er_extcount += ext_diff;
- erp_next->er_extcount -= ext_diff;
-
- /*
- * If the next irec is empty now we can simply
- * remove it.
- */
- if (erp_next->er_extcount == 0) {
- /*
- * Free page before removing extent record
- * so er_extoffs don't get modified in
- * xfs_iext_irec_remove.
- */
- kmem_free(erp_next->er_extbuf);
- erp_next->er_extbuf = NULL;
- xfs_iext_irec_remove(ifp, erp_idx + 1);
- erp = &ifp->if_u1.if_ext_irec[erp_idx];
- nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
-
- /*
- * If the next irec is not empty move up the content
- * that has not been copied to the previous page to
- * the beggining of this one.
- */
- } else {
- memmove(erp_next->er_extbuf, &ep_next[ext_diff],
- erp_next->er_extcount *
- sizeof(xfs_bmbt_rec_t));
- ep_next = erp_next->er_extbuf;
- memset(&ep_next[erp_next->er_extcount], 0,
- (XFS_LINEAR_EXTS -
- erp_next->er_extcount) *
- sizeof(xfs_bmbt_rec_t));
- }
- }
-
- if (erp->er_extcount == XFS_LINEAR_EXTS) {
- erp_idx++;
- if (erp_idx < nlists)
- erp = &ifp->if_u1.if_ext_irec[erp_idx];
- else
- break;
- }
- ep = &erp->er_extbuf[erp->er_extcount];
- erp_next = erp + 1;
- ep_next = erp_next->er_extbuf;
- }
-}
-
-/*
* This is called to update the er_extoff field in the indirection
* array when extents have been added or removed from one of the
* extent lists. erp_idx contains the irec index to begin updating
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index ccba14eb9db..503ea89e8b9 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -124,16 +124,27 @@ STATIC void xlog_verify_tail_lsn(xlog_t *log, xlog_in_core_t *iclog,
STATIC int xlog_iclogs_empty(xlog_t *log);
#if defined(XFS_LOG_TRACE)
+
+#define XLOG_TRACE_LOGGRANT_SIZE 2048
+#define XLOG_TRACE_ICLOG_SIZE 256
+
+void
+xlog_trace_loggrant_alloc(xlog_t *log)
+{
+ log->l_grant_trace = ktrace_alloc(XLOG_TRACE_LOGGRANT_SIZE, KM_NOFS);
+}
+
+void
+xlog_trace_loggrant_dealloc(xlog_t *log)
+{
+ ktrace_free(log->l_grant_trace);
+}
+
void
xlog_trace_loggrant(xlog_t *log, xlog_ticket_t *tic, xfs_caddr_t string)
{
unsigned long cnts;
- if (!log->l_grant_trace) {
- log->l_grant_trace = ktrace_alloc(2048, KM_NOSLEEP);
- if (!log->l_grant_trace)
- return;
- }
/* ticket counts are 1 byte each */
cnts = ((unsigned long)tic->t_ocnt) | ((unsigned long)tic->t_cnt) << 8;
@@ -157,10 +168,20 @@ xlog_trace_loggrant(xlog_t *log, xlog_ticket_t *tic, xfs_caddr_t string)
}
void
+xlog_trace_iclog_alloc(xlog_in_core_t *iclog)
+{
+ iclog->ic_trace = ktrace_alloc(XLOG_TRACE_ICLOG_SIZE, KM_NOFS);
+}
+
+void
+xlog_trace_iclog_dealloc(xlog_in_core_t *iclog)
+{
+ ktrace_free(iclog->ic_trace);
+}
+
+void
xlog_trace_iclog(xlog_in_core_t *iclog, uint state)
{
- if (!iclog->ic_trace)
- iclog->ic_trace = ktrace_alloc(256, KM_NOFS);
ktrace_enter(iclog->ic_trace,
(void *)((unsigned long)state),
(void *)((unsigned long)current_pid()),
@@ -170,8 +191,15 @@ xlog_trace_iclog(xlog_in_core_t *iclog, uint state)
(void *)NULL, (void *)NULL);
}
#else
+
+#define xlog_trace_loggrant_alloc(log)
+#define xlog_trace_loggrant_dealloc(log)
#define xlog_trace_loggrant(log,tic,string)
+
+#define xlog_trace_iclog_alloc(iclog)
+#define xlog_trace_iclog_dealloc(iclog)
#define xlog_trace_iclog(iclog,state)
+
#endif /* XFS_LOG_TRACE */
@@ -1009,7 +1037,7 @@ xlog_iodone(xfs_buf_t *bp)
* layer, it means the underlyin device no longer supports
* barrier I/O. Warn loudly and turn off barriers.
*/
- if ((l->l_mp->m_flags & XFS_MOUNT_BARRIER) && !XFS_BUF_ORDERED(bp)) {
+ if ((l->l_mp->m_flags & XFS_MOUNT_BARRIER) && !XFS_BUF_ISORDERED(bp)) {
l->l_mp->m_flags &= ~XFS_MOUNT_BARRIER;
xfs_fs_cmn_err(CE_WARN, l->l_mp,
"xlog_iodone: Barriers are no longer supported"
@@ -1231,6 +1259,7 @@ xlog_alloc_log(xfs_mount_t *mp,
spin_lock_init(&log->l_grant_lock);
sv_init(&log->l_flush_wait, 0, "flush_wait");
+ xlog_trace_loggrant_alloc(log);
/* log record size must be multiple of BBSIZE; see xlog_rec_header_t */
ASSERT((XFS_BUF_SIZE(bp) & BBMASK) == 0);
@@ -1285,6 +1314,8 @@ xlog_alloc_log(xfs_mount_t *mp,
sv_init(&iclog->ic_force_wait, SV_DEFAULT, "iclog-force");
sv_init(&iclog->ic_write_wait, SV_DEFAULT, "iclog-write");
+ xlog_trace_iclog_alloc(iclog);
+
iclogp = &iclog->ic_next;
}
*iclogp = log->l_iclog; /* complete ring */
@@ -1565,11 +1596,7 @@ xlog_dealloc_log(xlog_t *log)
sv_destroy(&iclog->ic_force_wait);
sv_destroy(&iclog->ic_write_wait);
xfs_buf_free(iclog->ic_bp);
-#ifdef XFS_LOG_TRACE
- if (iclog->ic_trace != NULL) {
- ktrace_free(iclog->ic_trace);
- }
-#endif
+ xlog_trace_iclog_dealloc(iclog);
next_iclog = iclog->ic_next;
kmem_free(iclog);
iclog = next_iclog;
@@ -1578,14 +1605,7 @@ xlog_dealloc_log(xlog_t *log)
spinlock_destroy(&log->l_grant_lock);
xfs_buf_free(log->l_xbuf);
-#ifdef XFS_LOG_TRACE
- if (log->l_trace != NULL) {
- ktrace_free(log->l_trace);
- }
- if (log->l_grant_trace != NULL) {
- ktrace_free(log->l_grant_trace);
- }
-#endif
+ xlog_trace_loggrant_dealloc(log);
log->l_mp->m_log = NULL;
kmem_free(log);
} /* xlog_dealloc_log */
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index c8a5b22ee3e..e7d8f84443f 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -448,7 +448,6 @@ typedef struct log {
int l_grant_write_bytes;
#ifdef XFS_LOG_TRACE
- struct ktrace *l_trace;
struct ktrace *l_grant_trace;
#endif
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index aa238c8fbd7..8b6812f66a1 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -1838,6 +1838,12 @@ again:
#endif
}
+/*
+ * xfs_lock_two_inodes() can only be used to lock one type of lock
+ * at a time - the iolock or the ilock, but not both at once. If
+ * we lock both at once, lockdep will report false positives saying
+ * we have violated locking orders.
+ */
void
xfs_lock_two_inodes(
xfs_inode_t *ip0,
@@ -1848,6 +1854,8 @@ xfs_lock_two_inodes(
int attempts = 0;
xfs_log_item_t *lp;
+ if (lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL))
+ ASSERT((lock_mode & (XFS_ILOCK_SHARED|XFS_ILOCK_EXCL)) == 0);
ASSERT(ip0->i_ino != ip1->i_ino);
if (ip0->i_ino > ip1->i_ino) {
@@ -3152,6 +3160,13 @@ error1: /* Just cancel transaction */
/*
* Zero file bytes between startoff and endoff inclusive.
* The iolock is held exclusive and no blocks are buffered.
+ *
+ * This function is used by xfs_free_file_space() to zero
+ * partial blocks when the range to free is not block aligned.
+ * When unreserving space with boundaries that are not block
+ * aligned we round up the start and round down the end
+ * boundaries and then use this function to zero the parts of
+ * the blocks that got dropped during the rounding.
*/
STATIC int
xfs_zero_remaining_bytes(
@@ -3168,6 +3183,17 @@ xfs_zero_remaining_bytes(
int nimap;
int error = 0;
+ /*
+ * Avoid doing I/O beyond eof - it's not necessary
+ * since nothing can read beyond eof. The space will
+ * be zeroed when the file is extended anyway.
+ */
+ if (startoff >= ip->i_size)
+ return 0;
+
+ if (endoff > ip->i_size)
+ endoff = ip->i_size;
+
bp = xfs_buf_get_noaddr(mp->m_sb.sb_blocksize,
XFS_IS_REALTIME_INODE(ip) ?
mp->m_rtdev_targp : mp->m_ddev_targp);