linux-stable/fs/ocfs2/extent_map.c

997 lines
23 KiB
C
Raw Normal View History

/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* extent_map.c
*
* Block/Cluster mapping functions
*
* Copyright (C) 2004 Oracle. All rights reserved.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public
* License, version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*/
#include <linux/fs.h>
#include <linux/init.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <linux/types.h>
#include <linux/fiemap.h>
#include <cluster/masklog.h>
#include "ocfs2.h"
#include "alloc.h"
#include "dlmglue.h"
#include "extent_map.h"
#include "inode.h"
#include "super.h"
#include "symlink.h"
#include "ocfs2_trace.h"
#include "buffer_head_io.h"
/*
* The extent caching implementation is intentionally trivial.
*
* We only cache a small number of extents stored directly on the
* inode, so linear order operations are acceptable. If we ever want
* to increase the size of the extent map, then these algorithms must
* get smarter.
*/
void ocfs2_extent_map_init(struct inode *inode)
{
struct ocfs2_inode_info *oi = OCFS2_I(inode);
oi->ip_extent_map.em_num_items = 0;
INIT_LIST_HEAD(&oi->ip_extent_map.em_list);
}
static void __ocfs2_extent_map_lookup(struct ocfs2_extent_map *em,
unsigned int cpos,
struct ocfs2_extent_map_item **ret_emi)
{
unsigned int range;
struct ocfs2_extent_map_item *emi;
*ret_emi = NULL;
list_for_each_entry(emi, &em->em_list, ei_list) {
range = emi->ei_cpos + emi->ei_clusters;
if (cpos >= emi->ei_cpos && cpos < range) {
list_move(&emi->ei_list, &em->em_list);
*ret_emi = emi;
break;
}
}
}
static int ocfs2_extent_map_lookup(struct inode *inode, unsigned int cpos,
unsigned int *phys, unsigned int *len,
unsigned int *flags)
{
unsigned int coff;
struct ocfs2_inode_info *oi = OCFS2_I(inode);
struct ocfs2_extent_map_item *emi;
spin_lock(&oi->ip_lock);
__ocfs2_extent_map_lookup(&oi->ip_extent_map, cpos, &emi);
if (emi) {
coff = cpos - emi->ei_cpos;
*phys = emi->ei_phys + coff;
if (len)
*len = emi->ei_clusters - coff;
if (flags)
*flags = emi->ei_flags;
}
spin_unlock(&oi->ip_lock);
if (emi == NULL)
return -ENOENT;
return 0;
}
/*
* Forget about all clusters equal to or greater than cpos.
*/
void ocfs2_extent_map_trunc(struct inode *inode, unsigned int cpos)
{
struct ocfs2_extent_map_item *emi, *n;
struct ocfs2_inode_info *oi = OCFS2_I(inode);
struct ocfs2_extent_map *em = &oi->ip_extent_map;
LIST_HEAD(tmp_list);
unsigned int range;
spin_lock(&oi->ip_lock);
list_for_each_entry_safe(emi, n, &em->em_list, ei_list) {
if (emi->ei_cpos >= cpos) {
/* Full truncate of this record. */
list_move(&emi->ei_list, &tmp_list);
BUG_ON(em->em_num_items == 0);
em->em_num_items--;
continue;
}
range = emi->ei_cpos + emi->ei_clusters;
if (range > cpos) {
/* Partial truncate */
emi->ei_clusters = cpos - emi->ei_cpos;
}
}
spin_unlock(&oi->ip_lock);
list_for_each_entry_safe(emi, n, &tmp_list, ei_list) {
list_del(&emi->ei_list);
kfree(emi);
}
}
/*
* Is any part of emi2 contained within emi1
*/
static int ocfs2_ei_is_contained(struct ocfs2_extent_map_item *emi1,
struct ocfs2_extent_map_item *emi2)
{
unsigned int range1, range2;
/*
* Check if logical start of emi2 is inside emi1
*/
range1 = emi1->ei_cpos + emi1->ei_clusters;
if (emi2->ei_cpos >= emi1->ei_cpos && emi2->ei_cpos < range1)
return 1;
/*
* Check if logical end of emi2 is inside emi1
*/
range2 = emi2->ei_cpos + emi2->ei_clusters;
if (range2 > emi1->ei_cpos && range2 <= range1)
return 1;
return 0;
}
static void ocfs2_copy_emi_fields(struct ocfs2_extent_map_item *dest,
struct ocfs2_extent_map_item *src)
{
dest->ei_cpos = src->ei_cpos;
dest->ei_phys = src->ei_phys;
dest->ei_clusters = src->ei_clusters;
dest->ei_flags = src->ei_flags;
}
/*
* Try to merge emi with ins. Returns 1 if merge succeeds, zero
* otherwise.
*/
static int ocfs2_try_to_merge_extent_map(struct ocfs2_extent_map_item *emi,
struct ocfs2_extent_map_item *ins)
{
/*
* Handle contiguousness
*/
if (ins->ei_phys == (emi->ei_phys + emi->ei_clusters) &&
ins->ei_cpos == (emi->ei_cpos + emi->ei_clusters) &&
ins->ei_flags == emi->ei_flags) {
emi->ei_clusters += ins->ei_clusters;
return 1;
} else if ((ins->ei_phys + ins->ei_clusters) == emi->ei_phys &&
(ins->ei_cpos + ins->ei_clusters) == emi->ei_cpos &&
ins->ei_flags == emi->ei_flags) {
emi->ei_phys = ins->ei_phys;
emi->ei_cpos = ins->ei_cpos;
emi->ei_clusters += ins->ei_clusters;
return 1;
}
/*
* Overlapping extents - this shouldn't happen unless we've
* split an extent to change it's flags. That is exceedingly
* rare, so there's no sense in trying to optimize it yet.
*/
if (ocfs2_ei_is_contained(emi, ins) ||
ocfs2_ei_is_contained(ins, emi)) {
ocfs2_copy_emi_fields(emi, ins);
return 1;
}
/* No merge was possible. */
return 0;
}
/*
* In order to reduce complexity on the caller, this insert function
* is intentionally liberal in what it will accept.
*
* The only rule is that the truncate call *must* be used whenever
* records have been deleted. This avoids inserting overlapping
* records with different physical mappings.
*/
void ocfs2_extent_map_insert_rec(struct inode *inode,
struct ocfs2_extent_rec *rec)
{
struct ocfs2_inode_info *oi = OCFS2_I(inode);
struct ocfs2_extent_map *em = &oi->ip_extent_map;
struct ocfs2_extent_map_item *emi, *new_emi = NULL;
struct ocfs2_extent_map_item ins;
ins.ei_cpos = le32_to_cpu(rec->e_cpos);
ins.ei_phys = ocfs2_blocks_to_clusters(inode->i_sb,
le64_to_cpu(rec->e_blkno));
ins.ei_clusters = le16_to_cpu(rec->e_leaf_clusters);
ins.ei_flags = rec->e_flags;
search:
spin_lock(&oi->ip_lock);
list_for_each_entry(emi, &em->em_list, ei_list) {
if (ocfs2_try_to_merge_extent_map(emi, &ins)) {
list_move(&emi->ei_list, &em->em_list);
spin_unlock(&oi->ip_lock);
goto out;
}
}
/*
* No item could be merged.
*
* Either allocate and add a new item, or overwrite the last recently
* inserted.
*/
if (em->em_num_items < OCFS2_MAX_EXTENT_MAP_ITEMS) {
if (new_emi == NULL) {
spin_unlock(&oi->ip_lock);
new_emi = kmalloc(sizeof(*new_emi), GFP_NOFS);
if (new_emi == NULL)
goto out;
goto search;
}
ocfs2_copy_emi_fields(new_emi, &ins);
list_add(&new_emi->ei_list, &em->em_list);
em->em_num_items++;
new_emi = NULL;
} else {
BUG_ON(list_empty(&em->em_list) || em->em_num_items == 0);
emi = list_entry(em->em_list.prev,
struct ocfs2_extent_map_item, ei_list);
list_move(&emi->ei_list, &em->em_list);
ocfs2_copy_emi_fields(emi, &ins);
}
spin_unlock(&oi->ip_lock);
out:
kfree(new_emi);
}
static int ocfs2_last_eb_is_empty(struct inode *inode,
struct ocfs2_dinode *di)
{
int ret, next_free;
u64 last_eb_blk = le64_to_cpu(di->i_last_eb_blk);
struct buffer_head *eb_bh = NULL;
struct ocfs2_extent_block *eb;
struct ocfs2_extent_list *el;
ret = ocfs2_read_extent_block(INODE_CACHE(inode), last_eb_blk, &eb_bh);
if (ret) {
mlog_errno(ret);
goto out;
}
eb = (struct ocfs2_extent_block *) eb_bh->b_data;
el = &eb->h_list;
if (el->l_tree_depth) {
ocfs2_error(inode->i_sb,
"Inode %lu has non zero tree depth in leaf block %llu\n",
inode->i_ino,
(unsigned long long)eb_bh->b_blocknr);
ret = -EROFS;
goto out;
}
next_free = le16_to_cpu(el->l_next_free_rec);
if (next_free == 0 ||
(next_free == 1 && ocfs2_is_empty_extent(&el->l_recs[0])))
ret = 1;
out:
brelse(eb_bh);
return ret;
}
/*
* Return the 1st index within el which contains an extent start
* larger than v_cluster.
*/
static int ocfs2_search_for_hole_index(struct ocfs2_extent_list *el,
u32 v_cluster)
{
int i;
struct ocfs2_extent_rec *rec;
for(i = 0; i < le16_to_cpu(el->l_next_free_rec); i++) {
rec = &el->l_recs[i];
if (v_cluster < le32_to_cpu(rec->e_cpos))
break;
}
return i;
}
/*
* Figure out the size of a hole which starts at v_cluster within the given
* extent list.
*
* If there is no more allocation past v_cluster, we return the maximum
* cluster size minus v_cluster.
*
* If we have in-inode extents, then el points to the dinode list and
* eb_bh is NULL. Otherwise, eb_bh should point to the extent block
* containing el.
*/
int ocfs2_figure_hole_clusters(struct ocfs2_caching_info *ci,
struct ocfs2_extent_list *el,
struct buffer_head *eb_bh,
u32 v_cluster,
u32 *num_clusters)
{
int ret, i;
struct buffer_head *next_eb_bh = NULL;
struct ocfs2_extent_block *eb, *next_eb;
i = ocfs2_search_for_hole_index(el, v_cluster);
if (i == le16_to_cpu(el->l_next_free_rec) && eb_bh) {
eb = (struct ocfs2_extent_block *)eb_bh->b_data;
/*
* Check the next leaf for any extents.
*/
if (le64_to_cpu(eb->h_next_leaf_blk) == 0ULL)
goto no_more_extents;
ret = ocfs2_read_extent_block(ci,
le64_to_cpu(eb->h_next_leaf_blk),
&next_eb_bh);
if (ret) {
mlog_errno(ret);
goto out;
}
next_eb = (struct ocfs2_extent_block *)next_eb_bh->b_data;
el = &next_eb->h_list;
i = ocfs2_search_for_hole_index(el, v_cluster);
}
no_more_extents:
if (i == le16_to_cpu(el->l_next_free_rec)) {
/*
* We're at the end of our existing allocation. Just
* return the maximum number of clusters we could
* possibly allocate.
*/
*num_clusters = UINT_MAX - v_cluster;
} else {
*num_clusters = le32_to_cpu(el->l_recs[i].e_cpos) - v_cluster;
}
ret = 0;
out:
brelse(next_eb_bh);
return ret;
}
static int ocfs2_get_clusters_nocache(struct inode *inode,
struct buffer_head *di_bh,
u32 v_cluster, unsigned int *hole_len,
struct ocfs2_extent_rec *ret_rec,
unsigned int *is_last)
{
int i, ret, tree_height, len;
struct ocfs2_dinode *di;
struct ocfs2_extent_block *uninitialized_var(eb);
struct ocfs2_extent_list *el;
struct ocfs2_extent_rec *rec;
struct buffer_head *eb_bh = NULL;
memset(ret_rec, 0, sizeof(*ret_rec));
if (is_last)
*is_last = 0;
di = (struct ocfs2_dinode *) di_bh->b_data;
el = &di->id2.i_list;
tree_height = le16_to_cpu(el->l_tree_depth);
if (tree_height > 0) {
ret = ocfs2_find_leaf(INODE_CACHE(inode), el, v_cluster,
&eb_bh);
if (ret) {
mlog_errno(ret);
goto out;
}
eb = (struct ocfs2_extent_block *) eb_bh->b_data;
el = &eb->h_list;
if (el->l_tree_depth) {
ocfs2_error(inode->i_sb,
"Inode %lu has non zero tree depth in leaf block %llu\n",
inode->i_ino,
(unsigned long long)eb_bh->b_blocknr);
ret = -EROFS;
goto out;
}
}
i = ocfs2_search_extent_list(el, v_cluster);
if (i == -1) {
/*
* Holes can be larger than the maximum size of an
* extent, so we return their lengths in a separate
* field.
*/
if (hole_len) {
ret = ocfs2_figure_hole_clusters(INODE_CACHE(inode),
el, eb_bh,
v_cluster, &len);
if (ret) {
mlog_errno(ret);
goto out;
}
*hole_len = len;
}
goto out_hole;
}
rec = &el->l_recs[i];
BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos));
if (!rec->e_blkno) {
ocfs2_error(inode->i_sb,
"Inode %lu has bad extent record (%u, %u, 0)\n",
inode->i_ino,
le32_to_cpu(rec->e_cpos),
ocfs2_rec_clusters(el, rec));
ret = -EROFS;
goto out;
}
*ret_rec = *rec;
/*
* Checking for last extent is potentially expensive - we
* might have to look at the next leaf over to see if it's
* empty.
*
* The first two checks are to see whether the caller even
* cares for this information, and if the extent is at least
* the last in it's list.
*
* If those hold true, then the extent is last if any of the
* additional conditions hold true:
* - Extent list is in-inode
* - Extent list is right-most
* - Extent list is 2nd to rightmost, with empty right-most
*/
if (is_last) {
if (i == (le16_to_cpu(el->l_next_free_rec) - 1)) {
if (tree_height == 0)
*is_last = 1;
else if (eb->h_blkno == di->i_last_eb_blk)
*is_last = 1;
else if (eb->h_next_leaf_blk == di->i_last_eb_blk) {
ret = ocfs2_last_eb_is_empty(inode, di);
if (ret < 0) {
mlog_errno(ret);
goto out;
}
if (ret == 1)
*is_last = 1;
}
}
}
out_hole:
ret = 0;
out:
brelse(eb_bh);
return ret;
}
static void ocfs2_relative_extent_offsets(struct super_block *sb,
u32 v_cluster,
struct ocfs2_extent_rec *rec,
u32 *p_cluster, u32 *num_clusters)
{
u32 coff = v_cluster - le32_to_cpu(rec->e_cpos);
*p_cluster = ocfs2_blocks_to_clusters(sb, le64_to_cpu(rec->e_blkno));
*p_cluster = *p_cluster + coff;
if (num_clusters)
*num_clusters = le16_to_cpu(rec->e_leaf_clusters) - coff;
}
int ocfs2_xattr_get_clusters(struct inode *inode, u32 v_cluster,
u32 *p_cluster, u32 *num_clusters,
struct ocfs2_extent_list *el,
unsigned int *extent_flags)
{
int ret = 0, i;
struct buffer_head *eb_bh = NULL;
struct ocfs2_extent_block *eb;
struct ocfs2_extent_rec *rec;
u32 coff;
if (el->l_tree_depth) {
ret = ocfs2_find_leaf(INODE_CACHE(inode), el, v_cluster,
&eb_bh);
if (ret) {
mlog_errno(ret);
goto out;
}
eb = (struct ocfs2_extent_block *) eb_bh->b_data;
el = &eb->h_list;
if (el->l_tree_depth) {
ocfs2_error(inode->i_sb,
"Inode %lu has non zero tree depth in xattr leaf block %llu\n",
inode->i_ino,
(unsigned long long)eb_bh->b_blocknr);
ret = -EROFS;
goto out;
}
}
i = ocfs2_search_extent_list(el, v_cluster);
if (i == -1) {
ret = -EROFS;
mlog_errno(ret);
goto out;
} else {
rec = &el->l_recs[i];
BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos));
if (!rec->e_blkno) {
ocfs2_error(inode->i_sb,
"Inode %lu has bad extent record (%u, %u, 0) in xattr\n",
inode->i_ino,
le32_to_cpu(rec->e_cpos),
ocfs2_rec_clusters(el, rec));
ret = -EROFS;
goto out;
}
coff = v_cluster - le32_to_cpu(rec->e_cpos);
*p_cluster = ocfs2_blocks_to_clusters(inode->i_sb,
le64_to_cpu(rec->e_blkno));
*p_cluster = *p_cluster + coff;
if (num_clusters)
*num_clusters = ocfs2_rec_clusters(el, rec) - coff;
if (extent_flags)
*extent_flags = rec->e_flags;
}
out:
if (eb_bh)
brelse(eb_bh);
return ret;
}
int ocfs2_get_clusters(struct inode *inode, u32 v_cluster,
u32 *p_cluster, u32 *num_clusters,
unsigned int *extent_flags)
{
int ret;
unsigned int uninitialized_var(hole_len), flags = 0;
struct buffer_head *di_bh = NULL;
struct ocfs2_extent_rec rec;
if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
ret = -ERANGE;
mlog_errno(ret);
goto out;
}
ret = ocfs2_extent_map_lookup(inode, v_cluster, p_cluster,
num_clusters, extent_flags);
if (ret == 0)
goto out;
ret = ocfs2_read_inode_block(inode, &di_bh);
if (ret) {
mlog_errno(ret);
goto out;
}
ret = ocfs2_get_clusters_nocache(inode, di_bh, v_cluster, &hole_len,
&rec, NULL);
if (ret) {
mlog_errno(ret);
goto out;
}
if (rec.e_blkno == 0ULL) {
/*
* A hole was found. Return some canned values that
* callers can key on. If asked for, num_clusters will
* be populated with the size of the hole.
*/
*p_cluster = 0;
if (num_clusters) {
*num_clusters = hole_len;
}
} else {
ocfs2_relative_extent_offsets(inode->i_sb, v_cluster, &rec,
p_cluster, num_clusters);
flags = rec.e_flags;
ocfs2_extent_map_insert_rec(inode, &rec);
}
if (extent_flags)
*extent_flags = flags;
out:
brelse(di_bh);
return ret;
}
/*
* This expects alloc_sem to be held. The allocation cannot change at
* all while the map is in the process of being updated.
*/
int ocfs2_extent_map_get_blocks(struct inode *inode, u64 v_blkno, u64 *p_blkno,
u64 *ret_count, unsigned int *extent_flags)
{
int ret;
int bpc = ocfs2_clusters_to_blocks(inode->i_sb, 1);
u32 cpos, num_clusters, p_cluster;
u64 boff = 0;
cpos = ocfs2_blocks_to_clusters(inode->i_sb, v_blkno);
ret = ocfs2_get_clusters(inode, cpos, &p_cluster, &num_clusters,
extent_flags);
if (ret) {
mlog_errno(ret);
goto out;
}
/*
* p_cluster == 0 indicates a hole.
*/
if (p_cluster) {
boff = ocfs2_clusters_to_blocks(inode->i_sb, p_cluster);
boff += (v_blkno & (u64)(bpc - 1));
}
*p_blkno = boff;
if (ret_count) {
*ret_count = ocfs2_clusters_to_blocks(inode->i_sb, num_clusters);
*ret_count -= v_blkno & (u64)(bpc - 1);
}
out:
return ret;
}
/*
* The ocfs2_fiemap_inline() may be a little bit misleading, since
* it not only handles the fiemap for inlined files, but also deals
* with the fast symlink, cause they have no difference for extent
* mapping per se.
*/
static int ocfs2_fiemap_inline(struct inode *inode, struct buffer_head *di_bh,
struct fiemap_extent_info *fieinfo,
u64 map_start)
{
int ret;
unsigned int id_count;
struct ocfs2_dinode *di;
u64 phys;
u32 flags = FIEMAP_EXTENT_DATA_INLINE|FIEMAP_EXTENT_LAST;
struct ocfs2_inode_info *oi = OCFS2_I(inode);
di = (struct ocfs2_dinode *)di_bh->b_data;
if (ocfs2_inode_is_fast_symlink(inode))
id_count = ocfs2_fast_symlink_chars(inode->i_sb);
else
id_count = le16_to_cpu(di->id2.i_data.id_count);
if (map_start < id_count) {
phys = oi->ip_blkno << inode->i_sb->s_blocksize_bits;
if (ocfs2_inode_is_fast_symlink(inode))
phys += offsetof(struct ocfs2_dinode, id2.i_symlink);
else
phys += offsetof(struct ocfs2_dinode,
id2.i_data.id_data);
ret = fiemap_fill_next_extent(fieinfo, 0, phys, id_count,
flags);
if (ret < 0)
return ret;
}
return 0;
}
#define OCFS2_FIEMAP_FLAGS (FIEMAP_FLAG_SYNC)
int ocfs2_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
u64 map_start, u64 map_len)
{
int ret, is_last;
u32 mapping_end, cpos;
unsigned int hole_size;
struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
u64 len_bytes, phys_bytes, virt_bytes;
struct buffer_head *di_bh = NULL;
struct ocfs2_extent_rec rec;
ret = fiemap_check_flags(fieinfo, OCFS2_FIEMAP_FLAGS);
if (ret)
return ret;
ret = ocfs2_inode_lock(inode, &di_bh, 0);
if (ret) {
mlog_errno(ret);
goto out;
}
down_read(&OCFS2_I(inode)->ip_alloc_sem);
/*
* Handle inline-data and fast symlink separately.
*/
if ((OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) ||
ocfs2_inode_is_fast_symlink(inode)) {
ret = ocfs2_fiemap_inline(inode, di_bh, fieinfo, map_start);
goto out_unlock;
}
cpos = map_start >> osb->s_clustersize_bits;
mapping_end = ocfs2_clusters_for_bytes(inode->i_sb,
map_start + map_len);
is_last = 0;
while (cpos < mapping_end && !is_last) {
u32 fe_flags;
ret = ocfs2_get_clusters_nocache(inode, di_bh, cpos,
&hole_size, &rec, &is_last);
if (ret) {
mlog_errno(ret);
goto out_unlock;
}
if (rec.e_blkno == 0ULL) {
cpos += hole_size;
continue;
}
fe_flags = 0;
if (rec.e_flags & OCFS2_EXT_UNWRITTEN)
fe_flags |= FIEMAP_EXTENT_UNWRITTEN;
if (rec.e_flags & OCFS2_EXT_REFCOUNTED)
fe_flags |= FIEMAP_EXTENT_SHARED;
if (is_last)
fe_flags |= FIEMAP_EXTENT_LAST;
len_bytes = (u64)le16_to_cpu(rec.e_leaf_clusters) << osb->s_clustersize_bits;
phys_bytes = le64_to_cpu(rec.e_blkno) << osb->sb->s_blocksize_bits;
virt_bytes = (u64)le32_to_cpu(rec.e_cpos) << osb->s_clustersize_bits;
ret = fiemap_fill_next_extent(fieinfo, virt_bytes, phys_bytes,
len_bytes, fe_flags);
if (ret)
break;
cpos = le32_to_cpu(rec.e_cpos)+ le16_to_cpu(rec.e_leaf_clusters);
}
if (ret > 0)
ret = 0;
out_unlock:
brelse(di_bh);
up_read(&OCFS2_I(inode)->ip_alloc_sem);
ocfs2_inode_unlock(inode, 0);
out:
return ret;
}
int ocfs2_seek_data_hole_offset(struct file *file, loff_t *offset, int whence)
{
struct inode *inode = file->f_mapping->host;
int ret;
unsigned int is_last = 0, is_data = 0;
u16 cs_bits = OCFS2_SB(inode->i_sb)->s_clustersize_bits;
u32 cpos, cend, clen, hole_size;
u64 extoff, extlen;
struct buffer_head *di_bh = NULL;
struct ocfs2_extent_rec rec;
BUG_ON(whence != SEEK_DATA && whence != SEEK_HOLE);
ret = ocfs2_inode_lock(inode, &di_bh, 0);
if (ret) {
mlog_errno(ret);
goto out;
}
down_read(&OCFS2_I(inode)->ip_alloc_sem);
if (*offset >= i_size_read(inode)) {
ret = -ENXIO;
goto out_unlock;
}
if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
if (whence == SEEK_HOLE)
*offset = i_size_read(inode);
goto out_unlock;
}
clen = 0;
cpos = *offset >> cs_bits;
cend = ocfs2_clusters_for_bytes(inode->i_sb, i_size_read(inode));
while (cpos < cend && !is_last) {
ret = ocfs2_get_clusters_nocache(inode, di_bh, cpos, &hole_size,
&rec, &is_last);
if (ret) {
mlog_errno(ret);
goto out_unlock;
}
extoff = cpos;
extoff <<= cs_bits;
if (rec.e_blkno == 0ULL) {
clen = hole_size;
is_data = 0;
} else {
clen = le16_to_cpu(rec.e_leaf_clusters) -
(cpos - le32_to_cpu(rec.e_cpos));
is_data = (rec.e_flags & OCFS2_EXT_UNWRITTEN) ? 0 : 1;
}
if ((!is_data && whence == SEEK_HOLE) ||
(is_data && whence == SEEK_DATA)) {
if (extoff > *offset)
*offset = extoff;
goto out_unlock;
}
if (!is_last)
cpos += clen;
}
if (whence == SEEK_HOLE) {
extoff = cpos;
extoff <<= cs_bits;
extlen = clen;
extlen <<= cs_bits;
if ((extoff + extlen) > i_size_read(inode))
extlen = i_size_read(inode) - extoff;
extoff += extlen;
if (extoff > *offset)
*offset = extoff;
goto out_unlock;
}
ret = -ENXIO;
out_unlock:
brelse(di_bh);
up_read(&OCFS2_I(inode)->ip_alloc_sem);
ocfs2_inode_unlock(inode, 0);
out:
return ret;
}
int ocfs2_read_virt_blocks(struct inode *inode, u64 v_block, int nr,
struct buffer_head *bhs[], int flags,
int (*validate)(struct super_block *sb,
struct buffer_head *bh))
{
int rc = 0;
u64 p_block, p_count;
int i, count, done = 0;
trace_ocfs2_read_virt_blocks(
inode, (unsigned long long)v_block, nr, bhs, flags,
validate);
if (((v_block + nr - 1) << inode->i_sb->s_blocksize_bits) >=
i_size_read(inode)) {
BUG_ON(!(flags & OCFS2_BH_READAHEAD));
goto out;
}
while (done < nr) {
down_read(&OCFS2_I(inode)->ip_alloc_sem);
rc = ocfs2_extent_map_get_blocks(inode, v_block + done,
&p_block, &p_count, NULL);
up_read(&OCFS2_I(inode)->ip_alloc_sem);
if (rc) {
mlog_errno(rc);
break;
}
if (!p_block) {
rc = -EIO;
mlog(ML_ERROR,
"Inode #%llu contains a hole at offset %llu\n",
(unsigned long long)OCFS2_I(inode)->ip_blkno,
(unsigned long long)(v_block + done) <<
inode->i_sb->s_blocksize_bits);
break;
}
count = nr - done;
if (p_count < count)
count = p_count;
/*
* If the caller passed us bhs, they should have come
* from a previous readahead call to this function. Thus,
* they should have the right b_blocknr.
*/
for (i = 0; i < count; i++) {
if (!bhs[done + i])
continue;
BUG_ON(bhs[done + i]->b_blocknr != (p_block + i));
}
rc = ocfs2_read_blocks(INODE_CACHE(inode), p_block, count,
bhs + done, flags, validate);
if (rc) {
mlog_errno(rc);
break;
}
done += count;
}
out:
return rc;
}