12 files changed, 115 insertions, 71 deletions

diff --git a/fs/logfs/dev_bdev.c b/fs/logfs/dev_bdev.c
index 243c00071f76..9bd2ce2a3040 100644
--- a/fs/logfs/dev_bdev.c
+++ b/fs/logfs/dev_bdev.c
@@ -303,6 +303,11 @@ static void bdev_put_device(struct super_block *sb)
         close_bdev_exclusive(logfs_super(sb)->s_bdev, FMODE_READ|FMODE_WRITE);
 }
 
+static int bdev_can_write_buf(struct super_block *sb, u64 ofs)
+{
+        return 0;
+}
+
 static const struct logfs_device_ops bd_devops = {
         .find_first_sb  = bdev_find_first_sb,
         .find_last_sb   = bdev_find_last_sb,
@@ -310,6 +315,7 @@ static const struct logfs_device_ops bd_devops = {
         .readpage       = bdev_readpage,
         .writeseg       = bdev_writeseg,
         .erase          = bdev_erase,
+        .can_write_buf  = bdev_can_write_buf,
         .sync           = bdev_sync,
         .put_device     = bdev_put_device,
 };
diff --git a/fs/logfs/dev_mtd.c b/fs/logfs/dev_mtd.c
index cafb6ef2e05b..a85d47d13e4b 100644
--- a/fs/logfs/dev_mtd.c
+++ b/fs/logfs/dev_mtd.c
@@ -9,6 +9,7 @@
 #include <linux/completion.h>
 #include <linux/mount.h>
 #include <linux/sched.h>
+#include <linux/slab.h>
 
 #define PAGE_OFS(ofs) ((ofs) & (PAGE_SIZE-1))
 
@@ -126,7 +127,8 @@ static int mtd_readpage(void *_sb, struct page *page)
 
         err = mtd_read(sb, page->index << PAGE_SHIFT, PAGE_SIZE,
                         page_address(page));
-        if (err == -EUCLEAN) {
+        if (err == -EUCLEAN || err == -EBADMSG) {
+                /* -EBADMSG happens regularly on power failures */
                 err = 0;
                 /* FIXME: force GC this segment */
         }
@@ -233,12 +235,32 @@ static void mtd_put_device(struct super_block *sb)
         put_mtd_device(logfs_super(sb)->s_mtd);
 }
 
+static int mtd_can_write_buf(struct super_block *sb, u64 ofs)
+{
+        struct logfs_super *super = logfs_super(sb);
+        void *buf;
+        int err;
+
+        buf = kmalloc(super->s_writesize, GFP_KERNEL);
+        if (!buf)
+                return -ENOMEM;
+        err = mtd_read(sb, ofs, super->s_writesize, buf);
+        if (err)
+                goto out;
+        if (memchr_inv(buf, 0xff, super->s_writesize))
+                err = -EIO;
+        kfree(buf);
+out:
+        return err;
+}
+
 static const struct logfs_device_ops mtd_devops = {
         .find_first_sb  = mtd_find_first_sb,
         .find_last_sb   = mtd_find_last_sb,
         .readpage       = mtd_readpage,
         .writeseg       = mtd_writeseg,
         .erase          = mtd_erase,
+        .can_write_buf  = mtd_can_write_buf,
         .sync           = mtd_sync,
         .put_device     = mtd_put_device,
 };
@@ -250,5 +272,7 @@ int logfs_get_sb_mtd(struct file_system_type *type, int flags,
         const struct logfs_device_ops *devops = &mtd_devops;
 
         mtd = get_mtd_device(NULL, mtdnr);
+        if (IS_ERR(mtd))
+                return PTR_ERR(mtd);
         return logfs_get_sb_device(type, flags, mtd, NULL, devops, mnt);
 }
diff --git a/fs/logfs/dir.c b/fs/logfs/dir.c
index 2396a85c0f55..72d1893ddd36 100644
--- a/fs/logfs/dir.c
+++ b/fs/logfs/dir.c
@@ -12,7 +12,7 @@
  * Atomic dir operations
  *
  * Directory operations are by default not atomic.  Dentries and Inodes are
- * created/removed/altered in seperate operations.  Therefore we need to do
+ * created/removed/altered in separate operations.  Therefore we need to do
  * a small amount of journaling.
  *
  * Create, link, mkdir, mknod and symlink all share the same function to do
diff --git a/fs/logfs/file.c b/fs/logfs/file.c
index 370f367a933e..abe1cafbd4c2 100644
--- a/fs/logfs/file.c
+++ b/fs/logfs/file.c
@@ -161,7 +161,17 @@ static int logfs_writepage(struct page *page, struct writeback_control *wbc)
 
 static void logfs_invalidatepage(struct page *page, unsigned long offset)
 {
-        move_page_to_btree(page);
+        struct logfs_block *block = logfs_block(page);
+
+        if (block->reserved_bytes) {
+                struct super_block *sb = page->mapping->host->i_sb;
+                struct logfs_super *super = logfs_super(sb);
+
+                super->s_dirty_pages -= block->reserved_bytes;
+                block->ops->free_block(sb, block);
+                BUG_ON(bitmap_weight(block->alias_map, LOGFS_BLOCK_FACTOR));
+        } else
+                move_page_to_btree(page);
         BUG_ON(PagePrivate(page) || page->private);
 }
 
@@ -209,13 +219,11 @@ int logfs_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
         }
 }
 
-int logfs_fsync(struct file *file, struct dentry *dentry, int datasync)
+int logfs_fsync(struct file *file, int datasync)
 {
-        struct super_block *sb = dentry->d_inode->i_sb;
-        struct logfs_super *super = logfs_super(sb);
+        struct super_block *sb = file->f_mapping->host->i_sb;
 
-        /* FIXME: write anchor */
-        super->s_devops->sync(sb);
+        logfs_write_anchor(sb);
         return 0;
 }
 
diff --git a/fs/logfs/gc.c b/fs/logfs/gc.c
index 76c242fbe1b0..caa4419285dc 100644
--- a/fs/logfs/gc.c
+++ b/fs/logfs/gc.c
@@ -122,7 +122,7 @@ static void logfs_cleanse_block(struct super_block *sb, u64 ofs, u64 ino,
         logfs_safe_iput(inode, cookie);
 }
 
-static u32 logfs_gc_segment(struct super_block *sb, u32 segno, u8 dist)
+static u32 logfs_gc_segment(struct super_block *sb, u32 segno)
 {
         struct logfs_super *super = logfs_super(sb);
         struct logfs_segment_header sh;
@@ -401,7 +401,7 @@ static int __logfs_gc_once(struct super_block *sb, struct gc_candidate *cand)
                         segno, (u64)segno << super->s_segshift,
                         dist, no_free_segments(sb), valid,
                         super->s_free_bytes);
-        cleaned = logfs_gc_segment(sb, segno, dist);
+        cleaned = logfs_gc_segment(sb, segno);
         log_gc("GC segment #%02x complete - now %x valid\n", segno,
                         valid - cleaned);
         BUG_ON(cleaned != valid);
@@ -632,38 +632,31 @@ static int check_area(struct super_block *sb, int i)
 {
         struct logfs_super *super = logfs_super(sb);
         struct logfs_area *area = super->s_area[i];
-        struct logfs_object_header oh;
+        gc_level_t gc_level;
+        u32 cleaned, valid, ec;
         u32 segno = area->a_segno;
-        u32 ofs = area->a_used_bytes;
-        __be32 crc;
-        int err;
+        u64 ofs = dev_ofs(sb, area->a_segno, area->a_written_bytes);
 
         if (!area->a_is_open)
                 return 0;
 
-        for (ofs = area->a_used_bytes;
-             ofs <= super->s_segsize - sizeof(oh);
-             ofs += (u32)be16_to_cpu(oh.len) + sizeof(oh)) {
-                err = wbuf_read(sb, dev_ofs(sb, segno, ofs), sizeof(oh), &oh);
-                if (err)
-                        return err;
-
-                if (!memchr_inv(&oh, 0xff, sizeof(oh)))
-                        break;
+        if (super->s_devops->can_write_buf(sb, ofs) == 0)
+                return 0;
 
-                crc = logfs_crc32(&oh, sizeof(oh) - 4, 4);
-                if (crc != oh.crc) {
-                        printk(KERN_INFO "interrupted header at %llx\n",
-                                        dev_ofs(sb, segno, ofs));
-                        return 0;
-                }
-        }
-        if (ofs != area->a_used_bytes) {
-                printk(KERN_INFO "%x bytes unaccounted data found at %llx\n",
-                                ofs - area->a_used_bytes,
-                                dev_ofs(sb, segno, area->a_used_bytes));
-                area->a_used_bytes = ofs;
-        }
+        printk(KERN_INFO"LogFS: Possibly incomplete write at %llx\n", ofs);
+        /*
+         * The device cannot write back the write buffer.  Most likely the
+         * wbuf was already written out and the system crashed at some point
+         * before the journal commit happened.  In that case we wouldn't have
+         * to do anything.  But if the crash happened before the wbuf was
+         * written out correctly, we must GC this segment.  So assume the
+         * worst and always do the GC run.
+         */
+        area->a_is_open = 0;
+        valid = logfs_valid_bytes(sb, segno, &ec, &gc_level);
+        cleaned = logfs_gc_segment(sb, segno);
+        if (cleaned != valid)
+                return -EIO;
         return 0;
 }
 
diff --git a/fs/logfs/inode.c b/fs/logfs/inode.c
index 14ed27274da2..f602e230e162 100644
--- a/fs/logfs/inode.c
+++ b/fs/logfs/inode.c
@@ -193,6 +193,7 @@ static void logfs_init_inode(struct super_block *sb, struct inode *inode)
         inode->i_ctime  = CURRENT_TIME;
         inode->i_mtime  = CURRENT_TIME;
         inode->i_nlink  = 1;
+        li->li_refcount = 1;
         INIT_LIST_HEAD(&li->li_freeing_list);
 
         for (i = 0; i < LOGFS_EMBEDDED_FIELDS; i++)
@@ -326,7 +327,7 @@ static void logfs_set_ino_generation(struct super_block *sb,
         u64 ino;
 
         mutex_lock(&super->s_journal_mutex);
-        ino = logfs_seek_hole(super->s_master_inode, super->s_last_ino);
+        ino = logfs_seek_hole(super->s_master_inode, super->s_last_ino + 1);
         super->s_last_ino = ino;
         super->s_inos_till_wrap--;
         if (super->s_inos_till_wrap < 0) {
@@ -357,14 +358,7 @@ struct inode *logfs_new_inode(struct inode *dir, int mode)
         inode->i_mode = mode;
         logfs_set_ino_generation(sb, inode);
 
-        inode->i_uid = current_fsuid();
-        inode->i_gid = current_fsgid();
-        if (dir->i_mode & S_ISGID) {
-                inode->i_gid = dir->i_gid;
-                if (S_ISDIR(mode))
-                        inode->i_mode |= S_ISGID;
-        }
-
+        inode_init_owner(inode, dir, mode);
         logfs_inode_setops(inode);
         insert_inode_hash(inode);
 
@@ -386,8 +380,7 @@ static void logfs_init_once(void *_li)
 
 static int logfs_sync_fs(struct super_block *sb, int wait)
 {
-        /* FIXME: write anchor */
-        logfs_super(sb)->s_devops->sync(sb);
+        logfs_write_anchor(sb);
         return 0;
 }
 
diff --git a/fs/logfs/journal.c b/fs/logfs/journal.c
index fb0a613f885b..4b0e0616b357 100644
--- a/fs/logfs/journal.c
+++ b/fs/logfs/journal.c
@@ -132,10 +132,9 @@ static int read_area(struct super_block *sb, struct logfs_je_area *a)
 
         ofs = dev_ofs(sb, area->a_segno, area->a_written_bytes);
         if (super->s_writesize > 1)
-                logfs_buf_recover(area, ofs, a + 1, super->s_writesize);
+                return logfs_buf_recover(area, ofs, a + 1, super->s_writesize);
         else
-                logfs_buf_recover(area, ofs, NULL, 0);
-        return 0;
+                return logfs_buf_recover(area, ofs, NULL, 0);
 }
 
 static void *unpack(void *from, void *to)
@@ -245,7 +244,7 @@ static int read_je(struct super_block *sb, u64 ofs)
                 read_erasecount(sb, unpack(jh, scratch));
                 break;
         case JE_AREA:
-                read_area(sb, unpack(jh, scratch));
+                err = read_area(sb, unpack(jh, scratch));
                 break;
         case JE_OBJ_ALIAS:
                 err = logfs_load_object_aliases(sb, unpack(jh, scratch),
diff --git a/fs/logfs/logfs.h b/fs/logfs/logfs.h
index 0a3df1a0c936..c838c4d72111 100644
--- a/fs/logfs/logfs.h
+++ b/fs/logfs/logfs.h
@@ -144,6 +144,7 @@ struct logfs_area_ops {
  * @erase:                      erase one segment
  * @read:                       read from the device
  * @erase:                      erase part of the device
+ * @can_write_buf:              decide whether wbuf can be written to ofs
  */
 struct logfs_device_ops {
         struct page *(*find_first_sb)(struct super_block *sb, u64 *ofs);
@@ -153,6 +154,7 @@ struct logfs_device_ops {
         void (*writeseg)(struct super_block *sb, u64 ofs, size_t len);
         int (*erase)(struct super_block *sb, loff_t ofs, size_t len,
                         int ensure_write);
+        int (*can_write_buf)(struct super_block *sb, u64 ofs);
         void (*sync)(struct super_block *sb);
         void (*put_device)(struct super_block *sb);
 };
@@ -394,6 +396,7 @@ struct logfs_super {
         int      s_lock_count;
         mempool_t *s_block_pool;                /* struct logfs_block pool */
         mempool_t *s_shadow_pool;               /* struct logfs_shadow pool */
+        struct list_head s_writeback_list;      /* writeback pages */
         /*
          * Space accounting:
          * - s_used_bytes specifies space used to store valid data objects.
@@ -503,7 +506,7 @@ extern const struct address_space_operations logfs_reg_aops;
 int logfs_readpage(struct file *file, struct page *page);
 int logfs_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
                 unsigned long arg);
-int logfs_fsync(struct file *file, struct dentry *dentry, int datasync);
+int logfs_fsync(struct file *file, int datasync);
 
 /* gc.c */
 u32 get_best_cand(struct super_block *sb, struct candidate_list *list, u32 *ec);
@@ -598,19 +601,19 @@ void freeseg(struct super_block *sb, u32 segno);
 int logfs_init_areas(struct super_block *sb);
 void logfs_cleanup_areas(struct super_block *sb);
 int logfs_open_area(struct logfs_area *area, size_t bytes);
-void __logfs_buf_write(struct logfs_area *area, u64 ofs, void *buf, size_t len,
+int __logfs_buf_write(struct logfs_area *area, u64 ofs, void *buf, size_t len,
                 int use_filler);
 
-static inline void logfs_buf_write(struct logfs_area *area, u64 ofs,
+static inline int logfs_buf_write(struct logfs_area *area, u64 ofs,
                 void *buf, size_t len)
 {
-        __logfs_buf_write(area, ofs, buf, len, 0);
+        return __logfs_buf_write(area, ofs, buf, len, 0);
 }
 
-static inline void logfs_buf_recover(struct logfs_area *area, u64 ofs,
+static inline int logfs_buf_recover(struct logfs_area *area, u64 ofs,
                 void *buf, size_t len)
 {
-        __logfs_buf_write(area, ofs, buf, len, 1);
+        return __logfs_buf_write(area, ofs, buf, len, 1);
 }
 
 /* super.c */
@@ -704,7 +707,7 @@ static inline gc_level_t expand_level(u64 ino, level_t __level)
         u8 level = (__force u8)__level;
 
         if (ino == LOGFS_INO_MASTER) {
-                /* ifile has seperate areas */
+                /* ifile has separate areas */
                 level += LOGFS_MAX_LEVELS;
         }
         return (__force gc_level_t)level;
diff --git a/fs/logfs/logfs_abi.h b/fs/logfs/logfs_abi.h
index f674725663fe..ae960519c54a 100644
--- a/fs/logfs/logfs_abi.h
+++ b/fs/logfs/logfs_abi.h
@@ -50,9 +50,9 @@ static inline void check_##type(void)				\
  * 12   - gc recycled blocks, long-lived data
  * 13   - replacement blocks, short-lived data
  *
- * Levels 1-11 are necessary for robust gc operations and help seperate
+ * Levels 1-11 are necessary for robust gc operations and help separate
  * short-lived metadata from longer-lived file data.  In the future,
- * file data should get seperated into several segments based on simple
+ * file data should get separated into several segments based on simple
  * heuristics.  Old data recycled during gc operation is expected to be
  * long-lived.  New data is of uncertain life expectancy.  New data
  * used to replace older blocks in existing files is expected to be
@@ -117,7 +117,7 @@ static inline void check_##type(void)				\
 #define pure_ofs(ofs) (ofs & ~LOGFS_FULLY_POPULATED)
 
 /*
- * LogFS needs to seperate data into levels.  Each level is defined as the
+ * LogFS needs to separate data into levels.  Each level is defined as the
  * maximal possible distance from the master inode (inode of the inode file).
  * Data blocks reside on level 0, 1x indirect block on level 1, etc.
  * Inodes reside on level 6, indirect blocks for the inode file on levels 7-11.
@@ -204,7 +204,7 @@ SIZE_CHECK(logfs_segment_header, LOGFS_SEGMENT_HEADERSIZE);
  * @ds_crc:                     crc32 of structure starting with the next field
  * @ds_ifile_levels:            maximum number of levels for ifile
  * @ds_iblock_levels:           maximum number of levels for regular files
- * @ds_data_levels:             number of seperate levels for data
+ * @ds_data_levels:             number of separate levels for data
  * @pad0:                       reserved, must be 0
  * @ds_feature_incompat:        incompatible filesystem features
  * @ds_feature_ro_compat:       read-only compatible filesystem features
@@ -456,7 +456,7 @@ enum logfs_vim {
  * @vim:                        life expectancy of data
  *
  * "Areas" are segments currently being used for writing.  There is at least
- * one area per GC level.  Several may be used to seperate long-living from
+ * one area per GC level.  Several may be used to separate long-living from
  * short-living data.  If an area with unknown vim is encountered, it can
  * simply be closed.
  * The write buffer immediately follow this header.
diff --git a/fs/logfs/readwrite.c b/fs/logfs/readwrite.c
index 3159db6958e5..0718d112a1a5 100644
--- a/fs/logfs/readwrite.c
+++ b/fs/logfs/readwrite.c
@@ -892,6 +892,8 @@ u64 logfs_seek_hole(struct inode *inode, u64 bix)
                 return bix;
         else if (li->li_data[INDIRECT_INDEX] & LOGFS_FULLY_POPULATED)
                 bix = maxbix(li->li_height);
+        else if (bix >= maxbix(li->li_height))
+                return bix;
         else {
                 bix = seek_holedata_loop(inode, bix, 0);
                 if (bix < maxbix(li->li_height))
@@ -1093,17 +1095,25 @@ static int logfs_reserve_bytes(struct inode *inode, int bytes)
 int get_page_reserve(struct inode *inode, struct page *page)
 {
         struct logfs_super *super = logfs_super(inode->i_sb);
+        struct logfs_block *block = logfs_block(page);
         int ret;
 
-        if (logfs_block(page) && logfs_block(page)->reserved_bytes)
+        if (block && block->reserved_bytes)
                 return 0;
 
         logfs_get_wblocks(inode->i_sb, page, WF_LOCK);
-        ret = logfs_reserve_bytes(inode, 6 * LOGFS_MAX_OBJECTSIZE);
+        while ((ret = logfs_reserve_bytes(inode, 6 * LOGFS_MAX_OBJECTSIZE)) &&
+                        !list_empty(&super->s_writeback_list)) {
+                block = list_entry(super->s_writeback_list.next,
+                                struct logfs_block, alias_list);
+                block->ops->write_block(block);
+        }
         if (!ret) {
                 alloc_data_block(inode, page);
-                logfs_block(page)->reserved_bytes += 6 * LOGFS_MAX_OBJECTSIZE;
+                block = logfs_block(page);
+                block->reserved_bytes += 6 * LOGFS_MAX_OBJECTSIZE;
                 super->s_dirty_pages += 6 * LOGFS_MAX_OBJECTSIZE;
+                list_move_tail(&block->alias_list, &super->s_writeback_list);
         }
         logfs_put_wblocks(inode->i_sb, page, WF_LOCK);
         return ret;
@@ -1861,7 +1871,7 @@ int logfs_truncate(struct inode *inode, u64 target)
                         size = target;
 
                 logfs_get_wblocks(sb, NULL, 1);
-                err = __logfs_truncate(inode, target);
+                err = __logfs_truncate(inode, size);
                 if (!err)
                         err = __logfs_write_inode(inode, 0);
                 logfs_put_wblocks(sb, NULL, 1);
@@ -2249,6 +2259,7 @@ int logfs_init_rw(struct super_block *sb)
         int min_fill = 3 * super->s_no_blocks;
 
         INIT_LIST_HEAD(&super->s_object_alias);
+        INIT_LIST_HEAD(&super->s_writeback_list);
         mutex_init(&super->s_write_mutex);
         super->s_block_pool = mempool_create_kmalloc_pool(min_fill,
                         sizeof(struct logfs_block));
diff --git a/fs/logfs/segment.c b/fs/logfs/segment.c
index f77ce2b470ba..a9657afb70ad 100644
--- a/fs/logfs/segment.c
+++ b/fs/logfs/segment.c
@@ -67,7 +67,7 @@ static struct page *get_mapping_page(struct super_block *sb, pgoff_t index,
         return page;
 }
 
-void __logfs_buf_write(struct logfs_area *area, u64 ofs, void *buf, size_t len,
+int __logfs_buf_write(struct logfs_area *area, u64 ofs, void *buf, size_t len,
                 int use_filler)
 {
         pgoff_t index = ofs >> PAGE_SHIFT;
@@ -81,8 +81,10 @@ void __logfs_buf_write(struct logfs_area *area, u64 ofs, void *buf, size_t len,
                 copylen = min((ulong)len, PAGE_SIZE - offset);
 
                 page = get_mapping_page(area->a_sb, index, use_filler);
-                SetPageUptodate(page);
+                if (IS_ERR(page))
+                        return PTR_ERR(page);
                 BUG_ON(!page); /* FIXME: reserve a pool */
+                SetPageUptodate(page);
                 memcpy(page_address(page) + offset, buf, copylen);
                 SetPagePrivate(page);
                 page_cache_release(page);
@@ -92,6 +94,7 @@ void __logfs_buf_write(struct logfs_area *area, u64 ofs, void *buf, size_t len,
                 offset = 0;
                 index++;
         } while (len);
+        return 0;
 }
 
 static void pad_partial_page(struct logfs_area *area)
diff --git a/fs/logfs/super.c b/fs/logfs/super.c
index d7c23ed8349a..d651e10a1e9c 100644
--- a/fs/logfs/super.c
+++ b/fs/logfs/super.c
@@ -138,10 +138,14 @@ static int logfs_sb_set(struct super_block *sb, void *_super)
         sb->s_fs_info = super;
         sb->s_mtd = super->s_mtd;
         sb->s_bdev = super->s_bdev;
+#ifdef CONFIG_BLOCK
         if (sb->s_bdev)
                 sb->s_bdi = &bdev_get_queue(sb->s_bdev)->backing_dev_info;
+#endif
+#ifdef CONFIG_MTD
         if (sb->s_mtd)
                 sb->s_bdi = sb->s_mtd->backing_dev_info;
+#endif
         return 0;
 }
 
@@ -382,7 +386,7 @@ static struct page *find_super_block(struct super_block *sb)
         if (!first || IS_ERR(first))
                 return NULL;
         last = super->s_devops->find_last_sb(sb, &super->s_sb_ofs[1]);
-        if (!last || IS_ERR(first)) {
+        if (!last || IS_ERR(last)) {
                 page_cache_release(first);
                 return NULL;
         }
@@ -413,7 +417,7 @@ static int __logfs_read_sb(struct super_block *sb)
 
         page = find_super_block(sb);
         if (!page)
-                return -EIO;
+                return -EINVAL;
 
         ds = page_address(page);
         super->s_size = be64_to_cpu(ds->ds_filesystem_size);