Whamcloud - gitweb
b=20672
[fs/lustre-release.git] / lustre / lvfs / fsfilt_ext3.c
index e7655e7..8cfbd00 100644 (file)
 #include <linux/init.h>
 #include <linux/module.h>
 #include <linux/fs.h>
-#include <linux/jbd.h>
 #include <linux/slab.h>
 #include <linux/pagemap.h>
 #include <linux/quotaops.h>
+#ifdef HAVE_EXT4_LDISKFS
+#include <ext4/ext4.h>
+#include <ext4/ext4_jbd2.h>
+#else
+#include <linux/jbd.h>
 #include <linux/ext3_fs.h>
 #include <linux/ext3_jbd.h>
+#endif
 #include <linux/version.h>
 #include <linux/bitops.h>
 #include <linux/quota.h>
 #include <linux/quotaio_v1.h>
 #include <linux/quotaio_v2.h>
+#if defined(HAVE_EXT3_XATTR_H)
 #include <ext3/xattr.h>
+#else
+/* ext3 xattr.h not available in rh style kernel-devel rpm */
+extern int ext3_xattr_get(struct inode *, int, const char *, void *, size_t);
+extern int ext3_xattr_set_handle(handle_t *, struct inode *, int, const char *, const void *, size_t, int);
+#endif
 
 #include <libcfs/libcfs.h>
 #include <lustre_fsfilt.h>
 #include <linux/lprocfs_status.h>
 
 #ifdef EXT3_MULTIBLOCK_ALLOCATOR
+#ifdef HAVE_EXT4_LDISKFS
+#include <ext4/ext4_extents.h>
+#else
 #include <linux/ext3_extents.h>
 #endif
+#endif
 
 #include "lustre_quota_fmt.h"
 
@@ -84,9 +99,6 @@
 #define FSFILT_SINGLEDATA_TRANS_BLOCKS(sb) EXT3_SINGLEDATA_TRANS_BLOCKS
 #endif
 
-#define fsfilt_ext3_journal_start(inode, nblocks) ext3_journal_start(inode, nblocks)
-#define fsfilt_ext3_journal_stop(handle)          ext3_journal_stop(handle)
-
 static cfs_mem_cache_t *fcb_cache;
 
 struct fsfilt_cb_data {
@@ -101,6 +113,19 @@ struct fsfilt_cb_data {
 #define EXT3_XATTR_INDEX_TRUSTED        4
 #endif
 
+#ifdef HAVE_EXT4_LDISKFS
+#define fsfilt_log_start_commit(journal, tid) jbd2_log_start_commit(journal, tid)
+#define fsfilt_log_wait_commit(journal, tid) jbd2_log_wait_commit(journal, tid)
+#define fsfilt_journal_callback_set(handle, func, jcb) jbd2_journal_callback_set(handle, func, jcb)
+#else
+#define fsfilt_log_start_commit(journal, tid) log_start_commit(journal, tid)
+#define fsfilt_log_wait_commit(journal, tid) log_wait_commit(journal, tid)
+#define fsfilt_journal_callback_set(handle, func, jcb) journal_callback_set(handle, func, jcb)
+#define ext_pblock(ex) le32_to_cpu((ex)->ee_start)
+#define ext3_ext_store_pblock(ex, pblock)  ((ex)->ee_start = cpu_to_le32(pblock))
+#define ext3_inode_bitmap(sb,desc) le32_to_cpu((desc)->bg_inode_bitmap)
+#endif
+
 static char *fsfilt_ext3_get_label(struct super_block *sb)
 {
         return EXT3_SB(sb)->s_es->s_volume_name;
@@ -114,7 +139,7 @@ static int fsfilt_ext3_set_label(struct super_block *sb, char *label)
         int err;
 
         journal = EXT3_SB(sb)->s_journal;
-        handle = journal_start(journal, 1);
+        handle = ext3_journal_start_sb(sb, 1);
         if (IS_ERR(handle)) {
                 CERROR("can't start transaction\n");
                 return(PTR_ERR(handle));
@@ -130,7 +155,7 @@ static int fsfilt_ext3_set_label(struct super_block *sb, char *label)
         err = ext3_journal_dirty_metadata(handle, EXT3_SB(sb)->s_sbh);
 
 out:
-        journal_stop(handle);
+        ext3_journal_stop(handle);
 
         return(err);
 }
@@ -141,14 +166,33 @@ static char *fsfilt_ext3_uuid(struct super_block *sb)
 }
 
 #ifdef HAVE_DISK_INODE_VERSION
+
+static __u64 get_i_version(struct inode *inode)
+{
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,27)) && defined(HAVE_EXT4_LDISKFS)
+        return inode->i_version;
+#else
+        return EXT3_I(inode)->i_fs_version;
+#endif
+}
+
+static void set_i_version(struct inode *inode, __u64 new_version)
+{
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,27)) && defined(HAVE_EXT4_LDISKFS)
+        inode->i_version = new_version;
+#else
+        (EXT3_I(inode))->i_fs_version = new_version;
+#endif
+}
+
 /*
  * Get the 64-bit version for an inode.
  */
 static __u64 fsfilt_ext3_get_version(struct inode *inode)
 {
         CDEBUG(D_INFO, "Get version "LPX64" for inode %lu\n",
-               EXT3_I(inode)->i_fs_version, inode->i_ino);
-        return EXT3_I(inode)->i_fs_version;
+               get_i_version(inode), inode->i_ino);
+        return get_i_version(inode);
 }
 
 /*
@@ -156,11 +200,11 @@ static __u64 fsfilt_ext3_get_version(struct inode *inode)
  */
 static __u64 fsfilt_ext3_set_version(struct inode *inode, __u64 new_version)
 {
-        __u64 old_version = EXT3_I(inode)->i_fs_version;
+        __u64 old_version = get_i_version(inode);
 
         CDEBUG(D_INFO, "Set version "LPX64" (old "LPX64") for inode %lu\n",
                new_version, old_version, inode->i_ino);
-        (EXT3_I(inode))->i_fs_version = new_version;
+        set_i_version(inode, new_version);
         /* version is set after all inode operations are finished, so we should
          * mark it dirty here */
         inode->i_sb->s_op->dirty_inode(inode);
@@ -206,7 +250,7 @@ static void *fsfilt_ext3_start(struct inode *inode, int op, void *desc_private,
                 nblocks += 3;
                 /* no break */
         case FSFILT_OP_CREATE: {
-#if defined(EXT3_EXTENTS_FL) && defined(EXT3_INDEX_FL)
+#if defined(EXT3_EXTENTS_FL) && defined(EXT3_INDEX_FL) && !defined(HAVE_EXT4_LDISKFS)
                 static int warned;
                 if (!warned) {
                         if (!test_opt(inode->i_sb, EXTENTS)) {
@@ -214,7 +258,8 @@ static void *fsfilt_ext3_start(struct inode *inode, int op, void *desc_private,
                         } else if (((EXT3_I(inode)->i_flags &
                               cpu_to_le32(EXT3_EXTENTS_FL | EXT3_INDEX_FL)) ==
                               cpu_to_le32(EXT3_EXTENTS_FL | EXT3_INDEX_FL))) {
-                                CWARN("extent-mapped directory found - contact "
+                                CWARN("extent-mapped directory found with "
+                                      "ext3-based ldiskfs - contact "
                                       "http://bugzilla.lustre.org/\n");
                                 warned = 1;
                         }
@@ -277,7 +322,7 @@ static void *fsfilt_ext3_start(struct inode *inode, int op, void *desc_private,
 
  journal_start:
         LASSERTF(nblocks > 0, "can't start %d credit transaction\n", nblocks);
-        handle = fsfilt_ext3_journal_start(inode, nblocks);
+        handle = ext3_journal_start(inode, nblocks);
 
         if (!IS_ERR(handle))
                 LASSERT(current->journal_info == handle);
@@ -414,7 +459,7 @@ static void *fsfilt_ext3_brw_start(int objcount, struct fsfilt_objinfo *fso,
         }
 
         LASSERTF(needed > 0, "can't start %d credit transaction\n", needed);
-        handle = fsfilt_ext3_journal_start(fso->fso_dentry->d_inode, needed);
+        handle = ext3_journal_start(fso->fso_dentry->d_inode, needed);
         if (IS_ERR(handle)) {
                 CERROR("can't get handle for %d credits: rc = %ld\n", needed,
                        PTR_ERR(handle));
@@ -438,11 +483,11 @@ static int fsfilt_ext3_extend(struct inode *inode, unsigned int nblocks,void *h)
 
        if (handle->h_buffer_credits > nblocks)
                 return 0;
-       if (journal_extend(handle, nblocks) == 0)
+       if (ext3_journal_extend(handle, nblocks) == 0)
                 return 0;
 
        ext3_mark_inode_dirty(handle, inode);
-       return journal_restart(handle, nblocks);
+       return ext3_journal_restart(handle, nblocks);
 }
 
 static int fsfilt_ext3_commit(struct inode *inode, void *h, int force_sync)
@@ -454,7 +499,7 @@ static int fsfilt_ext3_commit(struct inode *inode, void *h, int force_sync)
         if (force_sync)
                 handle->h_sync = 1; /* recovery likes this */
 
-        rc = fsfilt_ext3_journal_stop(handle);
+        rc = ext3_journal_stop(handle);
 
         return rc;
 }
@@ -475,12 +520,12 @@ static int fsfilt_ext3_commit_async(struct inode *inode, void *h,
         tid = transaction->t_tid;
         /* we don't want to be blocked */
         handle->h_sync = 0;
-        rc = fsfilt_ext3_journal_stop(handle);
+        rc = ext3_journal_stop(handle);
         if (rc) {
                 CERROR("error while stopping transaction: %d\n", rc);
                 return rc;
         }
-        log_start_commit(journal, tid);
+        fsfilt_log_start_commit(journal, tid);
 
         *wait_handle = (void *) tid;
         CDEBUG(D_INODE, "commit async: %lu\n", (unsigned long) tid);
@@ -496,7 +541,7 @@ static int fsfilt_ext3_commit_wait(struct inode *inode, void *h)
         if (unlikely(is_journal_aborted(journal)))
                 return -EIO;
 
-        log_wait_commit(EXT3_JOURNAL(inode), tid);
+        fsfilt_log_wait_commit(EXT3_JOURNAL(inode), tid);
 
         if (unlikely(is_journal_aborted(journal)))
                 return -EIO;
@@ -563,7 +608,7 @@ static int fsfilt_ext3_setattr(struct dentry *dentry, void *handle,
         RETURN(rc);
 }
 
-static int fsfilt_ext3_iocontrol(struct inode * inode, struct file *file,
+static int fsfilt_ext3_iocontrol(struct inode *inode, struct file *file,
                                  unsigned int cmd, unsigned long arg)
 {
         int rc = 0;
@@ -575,8 +620,14 @@ static int fsfilt_ext3_iocontrol(struct inode * inode, struct file *file,
                 RETURN(-EPERM);
         }
 
+#ifdef HAVE_EXT4_LDISKFS
+        /* ext4_ioctl does not have a inode argument */
+        if (inode->i_fop->unlocked_ioctl)
+                rc = inode->i_fop->unlocked_ioctl(file, cmd, arg);
+#else
         if (inode->i_fop->ioctl)
                 rc = inode->i_fop->ioctl(inode, file, cmd, arg);
+#endif
         else
                 RETURN(-ENOTTY);
 
@@ -644,8 +695,8 @@ static ssize_t fsfilt_ext3_readpage(struct file *file, char *buf, size_t count,
                 const int blkbits = inode->i_sb->s_blocksize_bits;
                 const int blksize = inode->i_sb->s_blocksize;
 
-                CDEBUG(D_EXT2, "reading "LPSZ" at dir %lu+%llu\n",
-                       count, inode->i_ino, *off);
+                CDEBUG(D_EXT2, "reading %lu at dir %lu+%llu\n",
+                       (unsigned long)count, inode->i_ino, *off);
                 while (count > 0) {
                         struct buffer_head *bh;
 
@@ -713,8 +764,8 @@ static int fsfilt_ext3_add_journal_cb(struct obd_device *obd, __u64 last_rcvd,
         fcb->cb_data = cb_data;
 
         CDEBUG(D_EXT2, "set callback for last_rcvd: "LPD64"\n", last_rcvd);
-        journal_callback_set(handle, fsfilt_ext3_cb_func,
-                             (struct journal_callback *)fcb);
+        fsfilt_journal_callback_set(handle, fsfilt_ext3_cb_func,
+                                    (struct journal_callback *)fcb);
 
         return 0;
 }
@@ -757,11 +808,16 @@ static int fsfilt_ext3_sync(struct super_block *sb)
 
 #ifdef EXT3_MULTIBLOCK_ALLOCATOR
 #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17))
-#define ext3_up_truncate_sem(inode)  up(&EXT3_I(inode)->truncate_sem);
-#define ext3_down_truncate_sem(inode)  down(&EXT3_I(inode)->truncate_sem);
+#define fsfilt_up_truncate_sem(inode)  up(&EXT3_I(inode)->truncate_sem);
+#define fsfilt_down_truncate_sem(inode)  down(&EXT3_I(inode)->truncate_sem);
 #else
-#define ext3_up_truncate_sem(inode)  mutex_unlock(&EXT3_I(inode)->truncate_mutex);
-#define ext3_down_truncate_sem(inode)  mutex_lock(&EXT3_I(inode)->truncate_mutex);
+#ifdef HAVE_EXT4_LDISKFS
+#define fsfilt_up_truncate_sem(inode) up_write((&EXT4_I(inode)->i_data_sem));
+#define fsfilt_down_truncate_sem(inode) down_write((&EXT4_I(inode)->i_data_sem));
+#else
+#define fsfilt_up_truncate_sem(inode)  mutex_unlock(&EXT3_I(inode)->truncate_mutex);
+#define fsfilt_down_truncate_sem(inode)  mutex_lock(&EXT3_I(inode)->truncate_mutex);
+#endif
 #endif
 
 #ifndef EXT_ASSERT
@@ -770,10 +826,14 @@ static int fsfilt_ext3_sync(struct super_block *sb)
 
 #ifdef EXT3_EXT_HAS_NO_TREE
 /* for kernels 2.6.18 and later */
+#ifdef HAVE_EXT4_LDISKFS
+#define EXT_GENERATION(inode)           (EXT4_I(inode)->i_ext_generation)
+#else
+#define EXT_GENERATION(inode)           ext_generation(inode)
+#endif
 #define ext3_ext_base                   inode
 #define ext3_ext_base2inode(inode)      (inode)
 #define EXT_DEPTH(inode)                ext_depth(inode)
-#define EXT_GENERATION(inode)           ext_generation(inode)
 #define fsfilt_ext3_ext_walk_space(inode, block, num, cb, cbdata) \
                         ext3_ext_walk_space(inode, block, num, cb, cbdata);
 #else
@@ -784,15 +844,6 @@ static int fsfilt_ext3_sync(struct super_block *sb)
 #endif
 
 #include <linux/lustre_version.h>
-#if EXT3_EXT_MAGIC == 0xf301
-#define ee_start e_start
-#define ee_block e_block
-#define ee_len   e_num
-#endif
-#ifndef EXT3_BB_MAX_BLOCKS
-#define ext3_mb_new_blocks(handle, inode, goal, count, aflags, err) \
-        ext3_new_blocks(handle, inode, count, goal, err)
-#endif
 
 struct bpointers {
         unsigned long *blocks;
@@ -803,8 +854,8 @@ struct bpointers {
         int create;
 };
 
-static int ext3_ext_find_goal(struct inode *inode, struct ext3_ext_path *path,
-                              unsigned long block, int *aflags)
+static long ext3_ext_find_goal(struct inode *inode, struct ext3_ext_path *path,
+                               unsigned long block, int *aflags)
 {
         struct ext3_inode_info *ei = EXT3_I(inode);
         unsigned long bg_start;
@@ -816,18 +867,8 @@ static int ext3_ext_find_goal(struct inode *inode, struct ext3_ext_path *path,
                 depth = path->p_depth;
 
                 /* try to predict block placement */
-                if ((ex = path[depth].p_ext)) {
-#if 0
-                        /* This prefers to eat into a contiguous extent
-                         * rather than find an extent that the whole
-                         * request will fit into.  This can fragment data
-                         * block allocation and prevents our lovely 1M I/Os
-                         * from reaching the disk intact. */
-                        if (ex->ee_block + ex->ee_len == block)
-                                *aflags |= 1;
-#endif
-                        return ex->ee_start + (block - ex->ee_block);
-                }
+                if ((ex = path[depth].p_ext))
+                        return ext_pblock(ex) + (block - le32_to_cpu(ex->ee_block));
 
                 /* it looks index is empty
                  * try to find starting from index itself */
@@ -952,18 +993,18 @@ static int ext3_ext_new_extent_cb(struct ext3_ext_base *base,
 
         tgen = EXT_GENERATION(base);
         count = ext3_ext_calc_credits_for_insert(base, path);
-        ext3_up_truncate_sem(inode);
+        fsfilt_up_truncate_sem(inode);
 
-        handle = fsfilt_ext3_journal_start(inode, count+EXT3_ALLOC_NEEDED+1);
+        handle = ext3_journal_start(inode, count+EXT3_ALLOC_NEEDED+1);
         if (IS_ERR(handle)) {
-                ext3_down_truncate_sem(inode);
+                fsfilt_down_truncate_sem(inode);
                 return PTR_ERR(handle);
         }
 
-        ext3_down_truncate_sem(inode);
+        fsfilt_down_truncate_sem(inode);
         if (tgen != EXT_GENERATION(base)) {
                 /* the tree has changed. so path can be invalid at moment */
-                fsfilt_ext3_journal_stop(handle);
+                ext3_journal_stop(handle);
                 return EXT_REPEAT;
         }
 
@@ -974,9 +1015,9 @@ static int ext3_ext_new_extent_cb(struct ext3_ext_base *base,
         EXT_ASSERT(count <= cex->ec_len);
 
         /* insert new extent */
-        nex.ee_block = cex->ec_block;
-        nex.ee_start = pblock;
-        nex.ee_len = count;
+        nex.ee_block = cpu_to_le32(cex->ec_block);
+        ext3_ext_store_pblock(&nex, pblock);
+        nex.ee_len = cpu_to_le16(count);
         err = ext3_ext_insert_extent(handle, base, path, &nex);
         if (err) {
                 /* free data blocks we just allocated */
@@ -985,7 +1026,8 @@ static int ext3_ext_new_extent_cb(struct ext3_ext_base *base,
 #ifdef EXT3_MB_HINT_GROUP_ALLOC
                 ext3_mb_discard_inode_preallocations(inode);
 #endif
-                ext3_free_blocks(handle, inode, nex.ee_start, nex.ee_len, 0);
+                ext3_free_blocks(handle, inode, ext_pblock(&nex),
+                                 cpu_to_le16(nex.ee_len), 0);
                 goto out;
         }
 
@@ -994,13 +1036,13 @@ static int ext3_ext_new_extent_cb(struct ext3_ext_base *base,
          * we are asking ext3_ext_walk_space() to continue
          * scaning after that block
          */
-        cex->ec_len = nex.ee_len;
-        cex->ec_start = nex.ee_start;
-        BUG_ON(nex.ee_len == 0);
-        BUG_ON(nex.ee_block != cex->ec_block);
+        cex->ec_len = le16_to_cpu(nex.ee_len);
+        cex->ec_start = ext_pblock(&nex);
+        BUG_ON(le16_to_cpu(nex.ee_len) == 0);
+        BUG_ON(le32_to_cpu(nex.ee_block) != cex->ec_block);
 
 out:
-        fsfilt_ext3_journal_stop(handle);
+        ext3_journal_stop(handle);
 map:
         if (err >= 0) {
                 /* map blocks */
@@ -1008,9 +1050,10 @@ map:
                         CERROR("hmm. why do we find this extent?\n");
                         CERROR("initial space: %lu:%u\n",
                                 bp->start, bp->init_num);
-                        CERROR("current extent: %u/%u/%u %d\n",
+                        CERROR("current extent: %u/%u/%llu %d\n",
                                 cex->ec_block, cex->ec_len,
-                                cex->ec_start, cex->ec_type);
+                                (unsigned long long)cex->ec_start,
+                                cex->ec_type);
                 }
                 i = 0;
                 if (cex->ec_block < bp->start)
@@ -1064,11 +1107,11 @@ int fsfilt_map_nblocks(struct inode *inode, unsigned long block,
         bp.init_num = bp.num = num;
         bp.create = create;
 
-        ext3_down_truncate_sem(inode);
+        fsfilt_down_truncate_sem(inode);
         err = fsfilt_ext3_ext_walk_space(base, block, num,
                                          ext3_ext_new_extent_cb, &bp);
         ext3_ext_invalidate_cache(base);
-        ext3_up_truncate_sem(inode);
+        fsfilt_up_truncate_sem(inode);
 
         return err;
 }
@@ -1301,7 +1344,7 @@ static int fsfilt_ext3_write_record(struct file *file, void *buf, int bufsize,
         block_count = (*offs & (blocksize - 1)) + bufsize;
         block_count = (block_count + blocksize - 1) >> inode->i_blkbits;
 
-        handle = fsfilt_ext3_journal_start(inode,
+        handle = ext3_journal_start(inode,
                                block_count * FSFILT_DATA_TRANS_BLOCKS(inode->i_sb) + 2);
         if (IS_ERR(handle)) {
                 CERROR("can't start transaction for %d blocks (%d bytes)\n",
@@ -1314,7 +1357,7 @@ static int fsfilt_ext3_write_record(struct file *file, void *buf, int bufsize,
         if (!err && force_sync)
                 handle->h_sync = 1; /* recovery likes this */
 
-        fsfilt_ext3_journal_stop(handle);
+        ext3_journal_stop(handle);
 
         return err;
 }
@@ -1671,17 +1714,19 @@ get_group_desc(struct super_block *sb, int group)
         return gdp + desc;
 }
 
+
+#ifndef HAVE_EXT4_LDISKFS
 static inline struct buffer_head *
-read_inode_bitmap(struct super_block *sb, unsigned long group)
+ext3_read_inode_bitmap(struct super_block *sb, unsigned long group)
 {
         struct ext3_group_desc *desc;
         struct buffer_head *bh;
 
         desc = get_group_desc(sb, group);
-        bh = sb_bread(sb, le32_to_cpu(desc->bg_inode_bitmap));
-
+        bh = sb_bread(sb, ext3_inode_bitmap(sb, desc));
         return bh;
 }
+#endif
 
 static inline struct inode *ext3_iget_inuse(struct super_block *sb,
                                      struct buffer_head *bitmap_bh,
@@ -1689,12 +1734,16 @@ static inline struct inode *ext3_iget_inuse(struct super_block *sb,
 {
         struct inode *inode = NULL;
 
+
         if (ext3_test_bit(index, bitmap_bh->b_data))
+#ifdef HAVE_EXT4_LDISKFS
+                inode = ext4_iget(sb, ino);
+                if (IS_ERR(inode))
+                /* Newer kernels return an error instead of a NULL pointer */
+                        inode = NULL;
+#else
                 inode = iget(sb, ino);
-
-        if (IS_ERR(inode))
-               /* Newer kernels return an error instead of a NULL pointer */
-                inode = NULL;
+#endif
         return inode;
 }
 
@@ -1943,10 +1992,23 @@ static int fsfilt_ext3_quotacheck(struct super_block *sb,
 
         /* check quota and update in hash */
         for (group = 0; group < sbi->s_groups_count; group++) {
+                struct ext3_group_desc *desc;
+                desc = get_group_desc(sb, group);
+                if (!desc)
+                        GOTO(out, -EIO);
+
+                spin_lock(sb_bgl_lock(sbi, group));
+                if (desc->bg_flags & cpu_to_le16(EXT3_BG_INODE_UNINIT)) {
+                        /* no inode in use in this group, just skip it */
+                        spin_unlock(sb_bgl_lock(sbi, group));
+                        continue;
+                }
+                spin_unlock(sb_bgl_lock(sbi, group));
+
                 ino = group * sbi->s_inodes_per_group + 1;
-                bitmap_bh = read_inode_bitmap(sb, group);
+                bitmap_bh = ext3_read_inode_bitmap(sb, group);
                 if (!bitmap_bh) {
-                        CERROR("read_inode_bitmap group %d failed", group);
+                        CERROR("ext3_read_inode_bitmap group %d failed", group);
                         GOTO(out, rc = -EIO);
                 }