import the very fat e2fsck/fsck applets

e2fsprogs/Config.in

@@ -11,6 +11,23 @@ config CONFIG_CHATTR
 	help
 	  chattr changes the file attributes on a second extended file system.
 
+config CONFIG_E2FSCK
+	bool "e2fsck"
+	default n
+	help
+	  e2fsck is used to check Linux second extended file systems (ext2fs).
+	  e2fsck also supports ext2 filesystems countaining a journal (ext3).
+	  The normal compat symlinks 'fsck.ext2' and 'fsck.ext3' are also 
+	  provided.
+
+config CONFIG_FSCK
+	bool "fsck"
+	default n
+	help
+	  fsck is used to check and optionally repair one or more filesystems.
+	  In actuality, fsck is simply a front-end for the various file system 
+	  checkers (fsck.fstype) available under Linux.
+
 config CONFIG_LSATTR
 	bool "lsattr"
 	default n

e2fsprogs/Makefile.in

@@ -6,17 +6,22 @@
 
 E2FSPROGS_AR:=e2fsprogs.a
 
-E2FSPROGS_DIR:=$(top_builddir)/e2fsprogs/
+E2FSPROGS_DIR:=$(top_builddir)/e2fsprogs
 
-srcdir=$(top_srcdir)/e2fsprogs
+E2FSPROGS_CFLAGS := -I$(E2FSPROGS_DIR) -include $(E2FSPROGS_DIR)/e2fsbb.h
-
-E2FSPROGS_CFLAGS := -I$(srcdir) -include $(srcdir)/e2fsbb.h
 
 BLKID_SRC   := cache.c dev.c devname.c devno.c blkid_getsize.c \
                probe.c read.c resolve.c save.c tag.c resolve.c
 BLKID_SRCS  := $(patsubst %,blkid/%, $(BLKID_SRC))
 BLKID_OBJS  := $(patsubst %.c,%.o, $(BLKID_SRCS))
 
+E2FSCK_SRC  := badblocks.c dict.c dirinfo.c dx_dirinfo.c e2fsck.c \
+               ea_refcount.c ehandler.c journal.c message.c pass1.c pass1b.c \
+               pass2.c pass3.c pass4.c pass5.c problem.c recovery.c region.c \
+               rehash.c revoke.c super.c swapfs.c util.c
+E2FSCK_SRCS := $(patsubst %,e2fsck/%, $(E2FSCK_SRC))
+E2FSCK_OBJS := $(patsubst %.c,%.o, $(E2FSCK_SRCS))
+
 E2P_SRC     := fgetsetflags.c fgetsetversion.c pf.c iod.c mntopts.c \
                feature.c ls.c uuid.c pe.c ostype.c ps.c hashstr.c \
                parse_num.c
@@ -29,7 +34,9 @@ EXT2FS_SRC  := gen_bitmap.c bitops.c ismounted.c mkjournal.c unix_io.c \
                openfs.c io_manager.c finddev.c read_bb.c alloc.c badblocks.c \
                getsize.c getsectsize.c alloc_tables.c read_bb_file.c mkdir.c \
                bb_inode.c newdir.c alloc_sb.c lookup.c dirblock.c expanddir.c \
-               dir_iterate.c link.c res_gdt.c
+               dir_iterate.c link.c res_gdt.c icount.c get_pathname.c dblist.c \
+               dirhash.c version.c flushb.c unlink.c check_desc.c valid_blk.c \
+               ext_attr.c bmap.c dblist_dir.c
 EXT2FS_SRCS := $(patsubst %,ext2fs/%, $(EXT2FS_SRC))
 EXT2FS_OBJS := $(patsubst %.c,%.o, $(EXT2FS_SRCS))
 
@@ -40,15 +47,17 @@ UUID_OBJS   := $(patsubst %.c,%.o, $(UUID_SRCS))
 
 E2FSPROGS-:=
 E2FSPROGS-$(CONFIG_CHATTR)     += chattr.o $(E2P_OBJS)
+E2FSPROGS-$(CONFIG_E2FSCK)     += e2fsck.o $(BLKID_OBJS) $(E2FSCK_OBJS)
+E2FSPROGS-$(CONFIG_FSCK)       += fsck.o base_device.o
 E2FSPROGS-$(CONFIG_LSATTR)     += lsattr.o $(E2P_OBJS)
 E2FSPROGS-$(CONFIG_MKE2FS)     += mke2fs.o util.o $(E2P_OBJS) $(BLKID_OBJS) $(EXT2FS_OBJS) $(UUID_OBJS)
 E2FSPROGS-$(CONFIG_TUNE2FS)    += tune2fs.o util.o $(E2P_OBJS) $(BLKID_OBJS) $(EXT2FS_OBJS) $(UUID_OBJS)
 
-libraries-y+=$(E2FSPROGS_DIR)$(E2FSPROGS_AR)
+libraries-y+=$(E2FSPROGS_DIR)/$(E2FSPROGS_AR)
 
 
-$(E2FSPROGS_DIR)$(E2FSPROGS_AR): $(patsubst %,$(E2FSPROGS_DIR)%, $(E2FSPROGS-y))
+$(E2FSPROGS_DIR)/$(E2FSPROGS_AR): $(patsubst %,$(E2FSPROGS_DIR)/%, $(E2FSPROGS-y))
-	$(AR) $(ARFLAGS) $@ $(patsubst %,$(E2FSPROGS_DIR)%, $(E2FSPROGS-y))
+	$(AR) $(ARFLAGS) $@ $(patsubst %,$(E2FSPROGS_DIR)/%, $(E2FSPROGS-y))
 
-$(E2FSPROGS_DIR)%.o: $(srcdir)/%.c
+$(E2FSPROGS_DIR)/%.o: $(E2FSPROGS_DIR)/%.c
 	$(CC) $(CFLAGS) $(EXTRA_CFLAGS) $(E2FSPROGS_CFLAGS) -c -o $@ $<

e2fsprogs/e2fsbb.h

@@ -32,29 +32,33 @@ typedef long errcode_t;
 #define P_(singular, plural, n) ((n) == 1 ? (singular) : (plural))
 
 /* misc crap */
-#define fatal_error(msg, err) bb_error_msg_and_die(msg)
+#define fatal_error(err, msg) bb_error_msg_and_die(msg)
 #define usage() bb_show_usage()
 #define perror(msg) bb_perror_msg(msg)
 
 /* header defines */
+#define ENABLE_HTREE 1
+#define HAVE_DIRENT_H 1
 #define HAVE_ERRNO_H 1
-#define HAVE_UNISTD_H 1
 #define HAVE_EXT2_IOCTLS 1
+#define HAVE_GETOPT_H 1
+#define HAVE_INTTYPES_H 1
+#define HAVE_LINUX_FD_H 1
+#define HAVE_MALLOC_H 1
+#define HAVE_MNTENT_H 1
+#define HAVE_NETINET_IN_H 1
+#define HAVE_NET_IF_H 1
+#define HAVE_SETJMP_H 1
+#define HAVE_SIGNAL_H 1
 #define HAVE_STDLIB_H 1
+#define HAVE_SYS_IOCTL_H 1
+#define HAVE_SYS_MOUNT_H 1
+#define HAVE_SYS_QUEUE_H 1
+#define HAVE_SYS_RESOURCE_H 1
+#define HAVE_SYS_SOCKET_H 1
 #define HAVE_SYS_STAT_H 1
+#define HAVE_SYS_TIME_H 1
 #define HAVE_SYS_TYPES_H 1
 #define HAVE_UNISTD_H 1
-#define HAVE_SYS_IOCTL_H 1
-#define HAVE_SYS_QUEUE_H 1
-#define HAVE_SYS_SOCKET_H 1
-#define HAVE_NETINET_IN_H 1
-#define HAVE_INTTYPES_H 1
-#define HAVE_NET_IF_H 1
-#define HAVE_GETOPT_H 1
-#define HAVE_SYS_RESOURCE_H 1
-#define HAVE_SYS_MOUNT_H 1
-#define HAVE_SYS_TIME_H 1
-#define HAVE_LINUX_FD_H 1
-#define HAVE_MNTENT_H 1
 
 #endif /* __E2FSBB_H__ */

e2fsprogs/e2fsck/badblocks.c

@@ -0,0 +1,136 @@
+/*
+ * badblocks.c --- replace/append bad blocks to the bad block inode
+ * 
+ * Copyright (C) 1993, 1994 Theodore Ts'o.  This file may be
+ * redistributed under the terms of the GNU Public License.
+ */
+
+#include <time.h>
+#ifdef HAVE_ERRNO_H
+#include <errno.h>
+#endif
+
+#include <et/com_err.h>
+#include "e2fsck.h"
+
+static int check_bb_inode_blocks(ext2_filsys fs, blk_t *block_nr, int blockcnt,
+				 void *priv_data);
+
+
+static void invalid_block(ext2_filsys fs EXT2FS_ATTR((unused)), blk_t blk)
+{
+	printf(_("Bad block %u out of range; ignored.\n"), blk);
+	return;
+}
+
+void read_bad_blocks_file(e2fsck_t ctx, const char *bad_blocks_file,
+			  int replace_bad_blocks)
+{
+	ext2_filsys fs = ctx->fs;
+	errcode_t	retval;
+	badblocks_list	bb_list = 0;
+	FILE		*f;
+	char		buf[1024];
+
+	e2fsck_read_bitmaps(ctx);
+
+	/*
+	 * Make sure the bad block inode is sane.  If there are any
+	 * illegal blocks, clear them.
+	 */
+	retval = ext2fs_block_iterate(fs, EXT2_BAD_INO, 0, 0,
+				      check_bb_inode_blocks, 0);
+	if (retval) {
+		com_err("ext2fs_block_iterate", retval,
+			_("while sanity checking the bad blocks inode"));
+		goto fatal;
+	}
+	
+	/*
+	 * If we're appending to the bad blocks inode, read in the
+	 * current bad blocks.
+	 */
+	if (!replace_bad_blocks) {
+		retval = ext2fs_read_bb_inode(fs, &bb_list);
+		if (retval) {
+			com_err("ext2fs_read_bb_inode", retval,
+				_("while reading the bad blocks inode"));
+			goto fatal;
+		}
+	}
+	
+	/*
+	 * Now read in the bad blocks from the file; if
+	 * bad_blocks_file is null, then try to run the badblocks
+	 * command.
+	 */
+	if (bad_blocks_file) {
+		f = fopen(bad_blocks_file, "r");
+		if (!f) {
+			com_err("read_bad_blocks_file", errno,
+				_("while trying to open %s"), bad_blocks_file);
+			goto fatal;
+		}
+	} else {
+		sprintf(buf, "badblocks -b %d %s%s%s %d", fs->blocksize,
+			(ctx->options & E2F_OPT_PREEN) ? "" : "-s ",
+			(ctx->options & E2F_OPT_WRITECHECK) ? "-n " : "",
+			fs->device_name, fs->super->s_blocks_count);
+		f = popen(buf, "r");
+		if (!f) {
+			com_err("read_bad_blocks_file", errno,
+				_("while trying popen '%s'"), buf);
+			goto fatal;
+		}
+	}
+	retval = ext2fs_read_bb_FILE(fs, f, &bb_list, invalid_block);
+	if (bad_blocks_file) 
+		fclose(f);
+	else
+		pclose(f);
+	if (retval) {
+		com_err("ext2fs_read_bb_FILE", retval,
+			_("while reading in list of bad blocks from file"));
+		goto fatal;
+	}
+	
+	/*
+	 * Finally, update the bad blocks from the bad_block_map
+	 */
+	retval = ext2fs_update_bb_inode(fs, bb_list);
+	if (retval) {
+		com_err("ext2fs_update_bb_inode", retval,
+			_("while updating bad block inode"));
+		goto fatal;
+	}
+
+	ext2fs_badblocks_list_free(bb_list);
+	return;
+	
+fatal:
+	ctx->flags |= E2F_FLAG_ABORT;
+	return;
+	
+}
+
+static int check_bb_inode_blocks(ext2_filsys fs, 
+				 blk_t *block_nr, 
+				 int blockcnt EXT2FS_ATTR((unused)),
+				 void *priv_data EXT2FS_ATTR((unused)))
+{
+	if (!*block_nr)
+		return 0;
+
+	/*
+	 * If the block number is outrageous, clear it and ignore it.
+	 */
+	if (*block_nr >= fs->super->s_blocks_count ||
+	    *block_nr < fs->super->s_first_data_block) {
+		printf(_("Warning illegal block %u found in bad block inode.  Cleared.\n"), *block_nr);
+		*block_nr = 0;
+		return BLOCK_CHANGED;
+	}
+
+	return 0;
+}
+

e2fsprogs/e2fsck/dict.h

@@ -0,0 +1,144 @@
+/*
+ * Dictionary Abstract Data Type
+ * Copyright (C) 1997 Kaz Kylheku <kaz@ashi.footprints.net>
+ *
+ * Free Software License:
+ *
+ * All rights are reserved by the author, with the following exceptions:
+ * Permission is granted to freely reproduce and distribute this software,
+ * possibly in exchange for a fee, provided that this copyright notice appears
+ * intact. Permission is also granted to adapt this software to produce
+ * derivative works, as long as the modified versions carry this copyright
+ * notice and additional notices stating that the work has been modified.
+ * This source code may be translated into executable form and incorporated
+ * into proprietary software; there is no requirement for such software to
+ * contain a copyright notice related to this source.
+ *
+ * $Id: dict.h,v 1.22.2.6 2000/11/13 01:36:44 kaz Exp $
+ * $Name: kazlib_1_20 $
+ */
+
+#ifndef DICT_H
+#define DICT_H
+
+#include <limits.h>
+#ifdef KAZLIB_SIDEEFFECT_DEBUG
+#include "sfx.h"
+#endif
+
+/*
+ * Blurb for inclusion into C++ translation units
+ */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef unsigned long dictcount_t;
+#define DICTCOUNT_T_MAX ULONG_MAX
+
+/*
+ * The dictionary is implemented as a red-black tree
+ */
+
+typedef enum { dnode_red, dnode_black } dnode_color_t;
+
+typedef struct dnode_t {
+#if defined(DICT_IMPLEMENTATION) || !defined(KAZLIB_OPAQUE_DEBUG)
+    struct dnode_t *dict_left;
+    struct dnode_t *dict_right;
+    struct dnode_t *dict_parent;
+    dnode_color_t dict_color;
+    const void *dict_key;
+    void *dict_data;
+#else
+    int dict_dummy;
+#endif
+} dnode_t;
+
+typedef int (*dict_comp_t)(const void *, const void *);
+typedef dnode_t *(*dnode_alloc_t)(void *);
+typedef void (*dnode_free_t)(dnode_t *, void *);
+
+typedef struct dict_t {
+#if defined(DICT_IMPLEMENTATION) || !defined(KAZLIB_OPAQUE_DEBUG)
+    dnode_t dict_nilnode;
+    dictcount_t dict_nodecount;
+    dictcount_t dict_maxcount;
+    dict_comp_t dict_compare;
+    dnode_alloc_t dict_allocnode;
+    dnode_free_t dict_freenode;
+    void *dict_context;
+    int dict_dupes;
+#else
+    int dict_dummmy;
+#endif
+} dict_t;
+
+typedef void (*dnode_process_t)(dict_t *, dnode_t *, void *);
+
+typedef struct dict_load_t {
+#if defined(DICT_IMPLEMENTATION) || !defined(KAZLIB_OPAQUE_DEBUG)
+    dict_t *dict_dictptr;
+    dnode_t dict_nilnode;
+#else
+    int dict_dummmy;
+#endif
+} dict_load_t;
+
+extern dict_t *dict_create(dictcount_t, dict_comp_t);
+extern void dict_set_allocator(dict_t *, dnode_alloc_t, dnode_free_t, void *);
+extern void dict_destroy(dict_t *);
+extern void dict_free_nodes(dict_t *);
+extern void dict_free(dict_t *);
+extern dict_t *dict_init(dict_t *, dictcount_t, dict_comp_t);
+extern void dict_init_like(dict_t *, const dict_t *);
+extern int dict_verify(dict_t *);
+extern int dict_similar(const dict_t *, const dict_t *);
+extern dnode_t *dict_lookup(dict_t *, const void *);
+extern dnode_t *dict_lower_bound(dict_t *, const void *);
+extern dnode_t *dict_upper_bound(dict_t *, const void *);
+extern void dict_insert(dict_t *, dnode_t *, const void *);
+extern dnode_t *dict_delete(dict_t *, dnode_t *);
+extern int dict_alloc_insert(dict_t *, const void *, void *);
+extern void dict_delete_free(dict_t *, dnode_t *);
+extern dnode_t *dict_first(dict_t *);
+extern dnode_t *dict_last(dict_t *);
+extern dnode_t *dict_next(dict_t *, dnode_t *);
+extern dnode_t *dict_prev(dict_t *, dnode_t *);
+extern dictcount_t dict_count(dict_t *);
+extern int dict_isempty(dict_t *);
+extern int dict_isfull(dict_t *);
+extern int dict_contains(dict_t *, dnode_t *);
+extern void dict_allow_dupes(dict_t *);
+extern int dnode_is_in_a_dict(dnode_t *);
+extern dnode_t *dnode_create(void *);
+extern dnode_t *dnode_init(dnode_t *, void *);
+extern void dnode_destroy(dnode_t *);
+extern void *dnode_get(dnode_t *);
+extern const void *dnode_getkey(dnode_t *);
+extern void dnode_put(dnode_t *, void *);
+extern void dict_process(dict_t *, void *, dnode_process_t);
+extern void dict_load_begin(dict_load_t *, dict_t *);
+extern void dict_load_next(dict_load_t *, dnode_t *, const void *);
+extern void dict_load_end(dict_load_t *);
+extern void dict_merge(dict_t *, dict_t *);
+
+#if defined(DICT_IMPLEMENTATION) || !defined(KAZLIB_OPAQUE_DEBUG)
+#ifdef KAZLIB_SIDEEFFECT_DEBUG
+#define dict_isfull(D) (SFX_CHECK(D)->dict_nodecount == (D)->dict_maxcount)
+#else
+#define dict_isfull(D) ((D)->dict_nodecount == (D)->dict_maxcount)
+#endif
+#define dict_count(D) ((D)->dict_nodecount)
+#define dict_isempty(D) ((D)->dict_nodecount == 0)
+#define dnode_get(N) ((N)->dict_data)
+#define dnode_getkey(N) ((N)->dict_key)
+#define dnode_put(N, X) ((N)->dict_data = (X))
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

e2fsprogs/e2fsck/dirinfo.c

@@ -0,0 +1,137 @@
+/*
+ * dirinfo.c --- maintains the directory information table for e2fsck.
+ *
+ * Copyright (C) 1993 Theodore Ts'o.  This file may be redistributed
+ * under the terms of the GNU Public License.
+ */
+
+#include "e2fsck.h"
+
+/*
+ * This subroutine is called during pass1 to create a directory info
+ * entry.  During pass1, the passed-in parent is 0; it will get filled
+ * in during pass2.  
+ */
+void e2fsck_add_dir_info(e2fsck_t ctx, ext2_ino_t ino, ext2_ino_t parent)
+{
+	struct dir_info *dir;
+	int		i, j;
+	ext2_ino_t	num_dirs;
+	errcode_t	retval;
+	unsigned long	old_size;
+
+#if 0
+	printf("add_dir_info for inode %lu...\n", ino);
+#endif
+	if (!ctx->dir_info) {
+		ctx->dir_info_count = 0;
+		retval = ext2fs_get_num_dirs(ctx->fs, &num_dirs);
+		if (retval)
+			num_dirs = 1024;	/* Guess */
+		ctx->dir_info_size = num_dirs + 10;
+		ctx->dir_info  = (struct dir_info *)
+			e2fsck_allocate_memory(ctx, ctx->dir_info_size
+					       * sizeof (struct dir_info),
+					       "directory map");
+	}
+	
+	if (ctx->dir_info_count >= ctx->dir_info_size) {
+		old_size = ctx->dir_info_size * sizeof(struct dir_info);
+		ctx->dir_info_size += 10;
+		retval = ext2fs_resize_mem(old_size, ctx->dir_info_size *
+					   sizeof(struct dir_info),
+					   &ctx->dir_info);
+		if (retval) {
+			ctx->dir_info_size -= 10;
+			return;
+		}
+	}
+
+	/*
+	 * Normally, add_dir_info is called with each inode in
+	 * sequential order; but once in a while (like when pass 3
+	 * needs to recreate the root directory or lost+found
+	 * directory) it is called out of order.  In those cases, we
+	 * need to move the dir_info entries down to make room, since
+	 * the dir_info array needs to be sorted by inode number for
+	 * get_dir_info()'s sake.
+	 */
+	if (ctx->dir_info_count &&
+	    ctx->dir_info[ctx->dir_info_count-1].ino >= ino) {
+		for (i = ctx->dir_info_count-1; i > 0; i--)
+			if (ctx->dir_info[i-1].ino < ino)
+				break;
+		dir = &ctx->dir_info[i];
+		if (dir->ino != ino) 
+			for (j = ctx->dir_info_count++; j > i; j--)
+				ctx->dir_info[j] = ctx->dir_info[j-1];
+	} else
+		dir = &ctx->dir_info[ctx->dir_info_count++];
+	
+	dir->ino = ino;
+	dir->dotdot = parent;
+	dir->parent = parent;
+}
+
+/*
+ * get_dir_info() --- given an inode number, try to find the directory
+ * information entry for it.
+ */
+struct dir_info *e2fsck_get_dir_info(e2fsck_t ctx, ext2_ino_t ino)
+{
+	int	low, high, mid;
+
+	low = 0;
+	high = ctx->dir_info_count-1;
+	if (!ctx->dir_info)
+		return 0;
+	if (ino == ctx->dir_info[low].ino)
+		return &ctx->dir_info[low];
+	if  (ino == ctx->dir_info[high].ino)
+		return &ctx->dir_info[high];
+
+	while (low < high) {
+		mid = (low+high)/2;
+		if (mid == low || mid == high)
+			break;
+		if (ino == ctx->dir_info[mid].ino)
+			return &ctx->dir_info[mid];
+		if (ino < ctx->dir_info[mid].ino)
+			high = mid;
+		else
+			low = mid;
+	}
+	return 0;
+}
+
+/*
+ * Free the dir_info structure when it isn't needed any more.
+ */
+void e2fsck_free_dir_info(e2fsck_t ctx)
+{
+	if (ctx->dir_info) {
+		ext2fs_free_mem(&ctx->dir_info);
+		ctx->dir_info = 0;
+	}
+	ctx->dir_info_size = 0;
+	ctx->dir_info_count = 0;
+}
+
+/*
+ * Return the count of number of directories in the dir_info structure
+ */
+int e2fsck_get_num_dirinfo(e2fsck_t ctx)
+{
+	return ctx->dir_info_count;
+}
+
+/*
+ * A simple interator function
+ */
+struct dir_info *e2fsck_dir_info_iter(e2fsck_t ctx, int *control)
+{
+	if (*control >= ctx->dir_info_count)
+		return 0;
+
+	return(ctx->dir_info + (*control)++);
+}

e2fsprogs/e2fsck/dx_dirinfo.c

@@ -0,0 +1,150 @@
+/*
+ * dirinfo.c --- maintains the directory information table for e2fsck.
+ *
+ * Copyright (C) 1993 Theodore Ts'o.  This file may be redistributed
+ * under the terms of the GNU Public License.
+ */
+
+#include "e2fsck.h"
+#ifdef ENABLE_HTREE
+
+/*
+ * This subroutine is called during pass1 to create a directory info
+ * entry.  During pass1, the passed-in parent is 0; it will get filled
+ * in during pass2.  
+ */
+void e2fsck_add_dx_dir(e2fsck_t ctx, ext2_ino_t ino, int num_blocks)
+{
+	struct dx_dir_info *dir;
+	int		i, j;
+	errcode_t	retval;
+	unsigned long	old_size;
+
+#if 0
+	printf("add_dx_dir_info for inode %lu...\n", ino);
+#endif
+	if (!ctx->dx_dir_info) {
+		ctx->dx_dir_info_count = 0;
+		ctx->dx_dir_info_size = 100; /* Guess */
+		ctx->dx_dir_info  = (struct dx_dir_info *)
+			e2fsck_allocate_memory(ctx, ctx->dx_dir_info_size
+					       * sizeof (struct dx_dir_info),
+					       "directory map");
+	}
+	
+	if (ctx->dx_dir_info_count >= ctx->dx_dir_info_size) {
+		old_size = ctx->dx_dir_info_size * sizeof(struct dx_dir_info);
+		ctx->dx_dir_info_size += 10;
+		retval = ext2fs_resize_mem(old_size, ctx->dx_dir_info_size *
+					   sizeof(struct dx_dir_info),
+					   &ctx->dx_dir_info);
+		if (retval) {
+			ctx->dx_dir_info_size -= 10;
+			return;
+		}
+	}
+
+	/*
+	 * Normally, add_dx_dir_info is called with each inode in
+	 * sequential order; but once in a while (like when pass 3
+	 * needs to recreate the root directory or lost+found
+	 * directory) it is called out of order.  In those cases, we
+	 * need to move the dx_dir_info entries down to make room, since
+	 * the dx_dir_info array needs to be sorted by inode number for
+	 * get_dx_dir_info()'s sake.
+	 */
+	if (ctx->dx_dir_info_count &&
+	    ctx->dx_dir_info[ctx->dx_dir_info_count-1].ino >= ino) {
+		for (i = ctx->dx_dir_info_count-1; i > 0; i--)
+			if (ctx->dx_dir_info[i-1].ino < ino)
+				break;
+		dir = &ctx->dx_dir_info[i];
+		if (dir->ino != ino) 
+			for (j = ctx->dx_dir_info_count++; j > i; j--)
+				ctx->dx_dir_info[j] = ctx->dx_dir_info[j-1];
+	} else
+		dir = &ctx->dx_dir_info[ctx->dx_dir_info_count++];
+	
+	dir->ino = ino;
+	dir->numblocks = num_blocks;
+	dir->hashversion = 0;
+	dir->dx_block = e2fsck_allocate_memory(ctx, num_blocks
+				       * sizeof (struct dx_dirblock_info),
+				       "dx_block info array");
+
+}
+
+/*
+ * get_dx_dir_info() --- given an inode number, try to find the directory
+ * information entry for it.
+ */
+struct dx_dir_info *e2fsck_get_dx_dir_info(e2fsck_t ctx, ext2_ino_t ino)
+{
+	int	low, high, mid;
+
+	low = 0;
+	high = ctx->dx_dir_info_count-1;
+	if (!ctx->dx_dir_info)
+		return 0;
+	if (ino == ctx->dx_dir_info[low].ino)
+		return &ctx->dx_dir_info[low];
+	if  (ino == ctx->dx_dir_info[high].ino)
+		return &ctx->dx_dir_info[high];
+
+	while (low < high) {
+		mid = (low+high)/2;
+		if (mid == low || mid == high)
+			break;
+		if (ino == ctx->dx_dir_info[mid].ino)
+			return &ctx->dx_dir_info[mid];
+		if (ino < ctx->dx_dir_info[mid].ino)
+			high = mid;
+		else
+			low = mid;
+	}
+	return 0;
+}
+
+/*
+ * Free the dx_dir_info structure when it isn't needed any more.
+ */
+void e2fsck_free_dx_dir_info(e2fsck_t ctx)
+{
+	int	i;
+	struct dx_dir_info *dir;
+	
+	if (ctx->dx_dir_info) {
+		dir = ctx->dx_dir_info;
+		for (i=0; i < ctx->dx_dir_info_count; i++) {
+			if (dir->dx_block) {
+				ext2fs_free_mem(&dir->dx_block);
+				dir->dx_block = 0;
+			}
+		}
+		ext2fs_free_mem(&ctx->dx_dir_info);
+		ctx->dx_dir_info = 0;
+	}
+	ctx->dx_dir_info_size = 0;
+	ctx->dx_dir_info_count = 0;
+}
+
+/*
+ * Return the count of number of directories in the dx_dir_info structure
+ */
+int e2fsck_get_num_dx_dirinfo(e2fsck_t ctx)
+{
+	return ctx->dx_dir_info_count;
+}
+
+/*
+ * A simple interator function
+ */
+struct dx_dir_info *e2fsck_dx_dir_info_iter(e2fsck_t ctx, int *control)
+{
+	if (*control >= ctx->dx_dir_info_count)
+		return 0;
+
+	return(ctx->dx_dir_info + (*control)++);
+}
+
+#endif /* ENABLE_HTREE */

e2fsprogs/e2fsck/e2fsck.c

@@ -0,0 +1,202 @@
+/*
+ * e2fsck.c - a consistency checker for the new extended file system.
+ * 
+ * Copyright (C) 1993, 1994, 1995, 1996, 1997 Theodore Ts'o.
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ */
+
+#include <errno.h>
+
+#include "e2fsck.h"
+#include "problem.h"
+
+/*
+ * This function allocates an e2fsck context 
+ */
+errcode_t e2fsck_allocate_context(e2fsck_t *ret)
+{
+	e2fsck_t	context;
+	errcode_t	retval;
+
+	retval = ext2fs_get_mem(sizeof(struct e2fsck_struct), &context);
+	if (retval)
+		return retval;
+	
+	memset(context, 0, sizeof(struct e2fsck_struct));
+
+	context->process_inode_size = 256;
+	context->ext_attr_ver = 2;
+
+	*ret = context;
+	return 0;
+}
+
+/*
+ * This function resets an e2fsck context; it is called when e2fsck
+ * needs to be restarted.
+ */
+errcode_t e2fsck_reset_context(e2fsck_t ctx)
+{
+	ctx->flags = 0;
+	ctx->lost_and_found = 0;
+	ctx->bad_lost_and_found = 0;
+	if (ctx->inode_used_map) {
+		ext2fs_free_inode_bitmap(ctx->inode_used_map);
+		ctx->inode_used_map = 0;
+	}
+	if (ctx->inode_dir_map) {
+		ext2fs_free_inode_bitmap(ctx->inode_dir_map);
+		ctx->inode_dir_map = 0;
+	}
+	if (ctx->inode_reg_map) {
+		ext2fs_free_inode_bitmap(ctx->inode_reg_map);
+		ctx->inode_reg_map = 0;
+	}
+	if (ctx->block_found_map) {
+		ext2fs_free_block_bitmap(ctx->block_found_map);
+		ctx->block_found_map = 0;
+	}
+	if (ctx->inode_link_info) {
+		ext2fs_free_icount(ctx->inode_link_info);
+		ctx->inode_link_info = 0;
+	}
+	if (ctx->journal_io) {
+		if (ctx->fs && ctx->fs->io != ctx->journal_io)
+			io_channel_close(ctx->journal_io);
+		ctx->journal_io = 0;
+	}
+	if (ctx->fs && ctx->fs->dblist) {
+		ext2fs_free_dblist(ctx->fs->dblist);
+		ctx->fs->dblist = 0;
+	}
+	e2fsck_free_dir_info(ctx);
+#ifdef ENABLE_HTREE
+	e2fsck_free_dx_dir_info(ctx);
+#endif
+	if (ctx->refcount) {
+		ea_refcount_free(ctx->refcount);
+		ctx->refcount = 0;
+	}
+	if (ctx->refcount_extra) {
+		ea_refcount_free(ctx->refcount_extra);
+		ctx->refcount_extra = 0;
+	}
+	if (ctx->block_dup_map) {
+		ext2fs_free_block_bitmap(ctx->block_dup_map);
+		ctx->block_dup_map = 0;
+	}
+	if (ctx->block_ea_map) {
+		ext2fs_free_block_bitmap(ctx->block_ea_map);
+		ctx->block_ea_map = 0;
+	}
+	if (ctx->inode_bb_map) {
+		ext2fs_free_inode_bitmap(ctx->inode_bb_map);
+		ctx->inode_bb_map = 0;
+	}
+	if (ctx->inode_bad_map) {
+		ext2fs_free_inode_bitmap(ctx->inode_bad_map);
+		ctx->inode_bad_map = 0;
+	}
+	if (ctx->inode_imagic_map) {
+		ext2fs_free_inode_bitmap(ctx->inode_imagic_map);
+		ctx->inode_imagic_map = 0;
+	}
+	if (ctx->dirs_to_hash) {
+		ext2fs_u32_list_free(ctx->dirs_to_hash);
+		ctx->dirs_to_hash = 0;
+	}
+
+	/*
+	 * Clear the array of invalid meta-data flags
+	 */
+	if (ctx->invalid_inode_bitmap_flag) {
+		ext2fs_free_mem(&ctx->invalid_inode_bitmap_flag);
+		ctx->invalid_inode_bitmap_flag = 0;
+	}
+	if (ctx->invalid_block_bitmap_flag) {
+		ext2fs_free_mem(&ctx->invalid_block_bitmap_flag);
+		ctx->invalid_block_bitmap_flag = 0;
+	}
+	if (ctx->invalid_inode_table_flag) {
+		ext2fs_free_mem(&ctx->invalid_inode_table_flag);
+		ctx->invalid_inode_table_flag = 0;
+	}
+
+	/* Clear statistic counters */
+	ctx->fs_directory_count = 0;
+	ctx->fs_regular_count = 0;
+	ctx->fs_blockdev_count = 0;
+	ctx->fs_chardev_count = 0;
+	ctx->fs_links_count = 0;
+	ctx->fs_symlinks_count = 0;
+	ctx->fs_fast_symlinks_count = 0;
+	ctx->fs_fifo_count = 0;
+	ctx->fs_total_count = 0;
+	ctx->fs_badblocks_count = 0;
+	ctx->fs_sockets_count = 0;
+	ctx->fs_ind_count = 0;
+	ctx->fs_dind_count = 0;
+	ctx->fs_tind_count = 0;
+	ctx->fs_fragmented = 0;
+	ctx->large_files = 0;
+
+	/* Reset the superblock to the user's requested value */
+	ctx->superblock = ctx->use_superblock;
+	
+	return 0;
+}
+
+void e2fsck_free_context(e2fsck_t ctx)
+{
+	if (!ctx)
+		return;
+	
+	e2fsck_reset_context(ctx);
+	if (ctx->blkid)
+		blkid_put_cache(ctx->blkid);
+			
+	ext2fs_free_mem(&ctx);
+}
+
+/*
+ * This function runs through the e2fsck passes and calls them all,
+ * returning restart, abort, or cancel as necessary...
+ */
+typedef void (*pass_t)(e2fsck_t ctx);
+
+pass_t e2fsck_passes[] = {
+	e2fsck_pass1, e2fsck_pass2, e2fsck_pass3, e2fsck_pass4,
+	e2fsck_pass5, 0 };
+
+#define E2F_FLAG_RUN_RETURN	(E2F_FLAG_SIGNAL_MASK|E2F_FLAG_RESTART)
+
+int e2fsck_run(e2fsck_t ctx)
+{
+	int	i;
+	pass_t	e2fsck_pass;
+
+#ifdef HAVE_SETJMP_H
+	if (setjmp(ctx->abort_loc)) {
+		ctx->flags &= ~E2F_FLAG_SETJMP_OK;
+		return (ctx->flags & E2F_FLAG_RUN_RETURN);
+	}
+	ctx->flags |= E2F_FLAG_SETJMP_OK;
+#endif
+		
+	for (i=0; (e2fsck_pass = e2fsck_passes[i]); i++) {
+		if (ctx->flags & E2F_FLAG_RUN_RETURN)
+			break;
+		e2fsck_pass(ctx);
+		if (ctx->progress)
+			(void) (ctx->progress)(ctx, 0, 0, 0);
+	}
+	ctx->flags &= ~E2F_FLAG_SETJMP_OK;
+	
+	if (ctx->flags & E2F_FLAG_RUN_RETURN)
+		return (ctx->flags & E2F_FLAG_RUN_RETURN);
+	return 0;
+}

e2fsprogs/e2fsck/e2fsck.h

@@ -0,0 +1,446 @@
+/*
+ * e2fsck.h
+ * 
+ * Copyright (C) 1993, 1994 Theodore Ts'o.  This file may be
+ * redistributed under the terms of the GNU Public License.
+ * 
+ */
+
+#include <stdio.h>
+#include <string.h>
+#ifdef HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+#include <stdlib.h>
+#include <time.h>
+#ifdef HAVE_SYS_TYPES_H
+#include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_TIME_H
+#include <sys/time.h>
+#endif
+#ifdef HAVE_SETJMP_H
+#include <setjmp.h>
+#endif
+
+#if EXT2_FLAT_INCLUDES
+#include "ext2_fs.h"
+#include "ext2fs.h"
+#include "blkid.h"
+#else
+#include "ext2fs/ext2_fs.h"
+#include "ext2fs/ext2fs.h"
+#include "blkid/blkid.h"
+#endif
+
+/*
+ * Exit codes used by fsck-type programs
+ */
+#define FSCK_OK          0	/* No errors */
+#define FSCK_NONDESTRUCT 1	/* File system errors corrected */
+#define FSCK_REBOOT      2	/* System should be rebooted */
+#define FSCK_UNCORRECTED 4	/* File system errors left uncorrected */
+#define FSCK_ERROR       8	/* Operational error */
+#define FSCK_USAGE       16	/* Usage or syntax error */
+#define FSCK_CANCELED	 32	/* Aborted with a signal or ^C */
+#define FSCK_LIBRARY     128	/* Shared library error */
+
+/*
+ * The last ext2fs revision level that this version of e2fsck is able to
+ * support
+ */
+#define E2FSCK_CURRENT_REV	1
+
+/*
+ * The directory information structure; stores directory information
+ * collected in earlier passes, to avoid disk i/o in fetching the
+ * directory information.
+ */
+struct dir_info {
+	ext2_ino_t		ino;	/* Inode number */
+	ext2_ino_t		dotdot;	/* Parent according to '..' */
+	ext2_ino_t		parent; /* Parent according to treewalk */
+};
+
+
+/*
+ * The indexed directory information structure; stores information for
+ * directories which contain a hash tree index.
+ */
+struct dx_dir_info {
+	ext2_ino_t		ino; 		/* Inode number */
+	int			numblocks;	/* number of blocks */
+	int			hashversion;
+	short			depth;		/* depth of tree */
+	struct dx_dirblock_info	*dx_block; 	/* Array of size numblocks */
+};
+
+#define DX_DIRBLOCK_ROOT	1
+#define DX_DIRBLOCK_LEAF	2
+#define DX_DIRBLOCK_NODE	3
+#define DX_DIRBLOCK_CORRUPT	4
+#define DX_DIRBLOCK_CLEARED	8
+
+struct dx_dirblock_info {
+	int		type;
+	blk_t		phys;
+	int		flags;
+	blk_t		parent;
+	ext2_dirhash_t	min_hash; 
+	ext2_dirhash_t	max_hash;
+	ext2_dirhash_t	node_min_hash; 
+	ext2_dirhash_t	node_max_hash;
+};
+
+#define DX_FLAG_REFERENCED	1
+#define DX_FLAG_DUP_REF		2
+#define DX_FLAG_FIRST		4
+#define DX_FLAG_LAST		8
+
+#ifdef RESOURCE_TRACK
+/*
+ * This structure is used for keeping track of how much resources have
+ * been used for a particular pass of e2fsck.
+ */
+struct resource_track {
+	struct timeval time_start;
+	struct timeval user_start;
+	struct timeval system_start;
+	void	*brk_start;
+};
+#endif
+
+/*
+ * E2fsck options
+ */
+#define E2F_OPT_READONLY	0x0001
+#define E2F_OPT_PREEN		0x0002
+#define E2F_OPT_YES		0x0004
+#define E2F_OPT_NO		0x0008
+#define E2F_OPT_TIME		0x0010
+#define E2F_OPT_TIME2		0x0020
+#define E2F_OPT_CHECKBLOCKS	0x0040
+#define E2F_OPT_DEBUG		0x0080
+#define E2F_OPT_FORCE		0x0100
+#define E2F_OPT_WRITECHECK	0x0200
+#define E2F_OPT_COMPRESS_DIRS	0x0400
+
+/*
+ * E2fsck flags
+ */
+#define E2F_FLAG_ABORT		0x0001 /* Abort signaled */
+#define E2F_FLAG_CANCEL		0x0002 /* Cancel signaled */
+#define E2F_FLAG_SIGNAL_MASK	0x0003
+#define E2F_FLAG_RESTART	0x0004 /* Restart signaled */
+
+#define E2F_FLAG_SETJMP_OK	0x0010 /* Setjmp valid for abort */
+
+#define E2F_FLAG_PROG_BAR	0x0020 /* Progress bar on screen */
+#define E2F_FLAG_PROG_SUPPRESS	0x0040 /* Progress suspended */
+#define E2F_FLAG_JOURNAL_INODE	0x0080 /* Create a new ext3 journal inode */
+#define E2F_FLAG_SB_SPECIFIED	0x0100 /* The superblock was explicitly 
+					* specified by the user */
+#define E2F_FLAG_RESTARTED	0x0200 /* E2fsck has been restarted */
+#define E2F_FLAG_RESIZE_INODE	0x0400 /* Request to recreate resize inode */
+
+/*
+ * Defines for indicating the e2fsck pass number
+ */
+#define E2F_PASS_1	1
+#define E2F_PASS_2	2
+#define E2F_PASS_3	3
+#define E2F_PASS_4	4
+#define E2F_PASS_5	5
+#define E2F_PASS_1B	6
+
+/*
+ * Define the extended attribute refcount structure
+ */
+typedef struct ea_refcount *ext2_refcount_t;
+
+/*
+ * This is the global e2fsck structure.
+ */
+typedef struct e2fsck_struct *e2fsck_t;
+
+struct e2fsck_struct {
+	ext2_filsys fs;
+	const char *program_name;
+	char *filesystem_name;
+	char *device_name;
+	char *io_options;
+	int	flags;		/* E2fsck internal flags */
+	int	options;
+	blk_t	use_superblock;	/* sb requested by user */
+	blk_t	superblock;	/* sb used to open fs */
+	int	blocksize;	/* blocksize */
+	blk_t	num_blocks;	/* Total number of blocks */
+	int	mount_flags;
+	blkid_cache blkid;	/* blkid cache */
+
+#ifdef HAVE_SETJMP_H
+	jmp_buf	abort_loc;
+#endif
+	unsigned long abort_code;
+
+	int (*progress)(e2fsck_t ctx, int pass, unsigned long cur,
+			unsigned long max);
+
+	ext2fs_inode_bitmap inode_used_map; /* Inodes which are in use */
+	ext2fs_inode_bitmap inode_bad_map; /* Inodes which are bad somehow */
+	ext2fs_inode_bitmap inode_dir_map; /* Inodes which are directories */
+	ext2fs_inode_bitmap inode_bb_map; /* Inodes which are in bad blocks */
+	ext2fs_inode_bitmap inode_imagic_map; /* AFS inodes */
+	ext2fs_inode_bitmap inode_reg_map; /* Inodes which are regular files*/
+
+	ext2fs_block_bitmap block_found_map; /* Blocks which are in use */
+	ext2fs_block_bitmap block_dup_map; /* Blks referenced more than once */
+	ext2fs_block_bitmap block_ea_map; /* Blocks which are used by EA's */
+
+	/*
+	 * Inode count arrays
+	 */
+	ext2_icount_t	inode_count;
+	ext2_icount_t inode_link_info;
+
+	ext2_refcount_t	refcount;
+	ext2_refcount_t refcount_extra;
+
+	/*
+	 * Array of flags indicating whether an inode bitmap, block
+	 * bitmap, or inode table is invalid
+	 */
+	int *invalid_inode_bitmap_flag;
+	int *invalid_block_bitmap_flag;
+	int *invalid_inode_table_flag;
+	int invalid_bitmaps;	/* There are invalid bitmaps/itable */
+
+	/*
+	 * Block buffer
+	 */
+	char *block_buf;
+
+	/*
+	 * For pass1_check_directory and pass1_get_blocks
+	 */
+	ext2_ino_t stashed_ino;
+	struct ext2_inode *stashed_inode;
+
+	/*
+	 * Location of the lost and found directory
+	 */
+	ext2_ino_t lost_and_found;
+	int bad_lost_and_found;
+	
+	/*
+	 * Directory information
+	 */
+	int		dir_info_count;
+	int		dir_info_size;
+	struct dir_info	*dir_info;
+
+	/*
+	 * Indexed directory information
+	 */
+	int		dx_dir_info_count;
+	int		dx_dir_info_size;
+	struct dx_dir_info *dx_dir_info;
+
+	/*
+	 * Directories to hash
+	 */
+	ext2_u32_list	dirs_to_hash;
+
+	/*
+	 * Tuning parameters
+	 */
+	int process_inode_size;
+	int inode_buffer_blocks;
+
+	/*
+	 * ext3 journal support
+	 */
+	io_channel	journal_io;
+	char	*journal_name;
+
+#ifdef RESOURCE_TRACK
+	/*
+	 * For timing purposes
+	 */
+	struct resource_track	global_rtrack;
+#endif
+
+	/*
+	 * How we display the progress update (for unix)
+	 */
+	int progress_fd;
+	int progress_pos;
+	int progress_last_percent;
+	unsigned int progress_last_time;
+	int interactive;	/* Are we connected directly to a tty? */
+	char start_meta[2], stop_meta[2];
+	
+	/* File counts */
+	int fs_directory_count;
+	int fs_regular_count;
+	int fs_blockdev_count;
+	int fs_chardev_count;
+	int fs_links_count;
+	int fs_symlinks_count;
+	int fs_fast_symlinks_count;
+	int fs_fifo_count;
+	int fs_total_count;
+	int fs_badblocks_count;
+	int fs_sockets_count;
+	int fs_ind_count;
+	int fs_dind_count;
+	int fs_tind_count;
+	int fs_fragmented;
+	int large_files;
+	int fs_ext_attr_inodes;
+	int fs_ext_attr_blocks;
+
+	int ext_attr_ver;
+
+	/*
+	 * For the use of callers of the e2fsck functions; not used by
+	 * e2fsck functions themselves.
+	 */
+	void *priv_data;
+};
+
+/* Used by the region allocation code */
+typedef __u32 region_addr_t;
+typedef struct region_struct *region_t;
+
+#ifndef HAVE_STRNLEN
+#define strnlen(str, x) e2fsck_strnlen((str),(x))
+extern int e2fsck_strnlen(const char * s, int count);
+#endif
+
+/*
+ * Procedure declarations
+ */
+
+extern void e2fsck_pass1(e2fsck_t ctx);
+extern void e2fsck_pass1_dupblocks(e2fsck_t ctx, char *block_buf);
+extern void e2fsck_pass2(e2fsck_t ctx);
+extern void e2fsck_pass3(e2fsck_t ctx);
+extern void e2fsck_pass4(e2fsck_t ctx);
+extern void e2fsck_pass5(e2fsck_t ctx);
+
+/* e2fsck.c */
+extern errcode_t e2fsck_allocate_context(e2fsck_t *ret);
+extern errcode_t e2fsck_reset_context(e2fsck_t ctx);
+extern void e2fsck_free_context(e2fsck_t ctx);
+extern int e2fsck_run(e2fsck_t ctx);
+
+
+/* badblock.c */
+extern void read_bad_blocks_file(e2fsck_t ctx, const char *bad_blocks_file,
+				 int replace_bad_blocks);
+
+/* dirinfo.c */
+extern void e2fsck_add_dir_info(e2fsck_t ctx, ext2_ino_t ino, ext2_ino_t parent);
+extern struct dir_info *e2fsck_get_dir_info(e2fsck_t ctx, ext2_ino_t ino);
+extern void e2fsck_free_dir_info(e2fsck_t ctx);
+extern int e2fsck_get_num_dirinfo(e2fsck_t ctx);
+extern struct dir_info *e2fsck_dir_info_iter(e2fsck_t ctx, int *control);
+
+/* dx_dirinfo.c */
+extern void e2fsck_add_dx_dir(e2fsck_t ctx, ext2_ino_t ino, int num_blocks);
+extern struct dx_dir_info *e2fsck_get_dx_dir_info(e2fsck_t ctx, ext2_ino_t ino);
+extern void e2fsck_free_dx_dir_info(e2fsck_t ctx);
+extern int e2fsck_get_num_dx_dirinfo(e2fsck_t ctx);
+extern struct dx_dir_info *e2fsck_dx_dir_info_iter(e2fsck_t ctx, int *control);
+
+/* ea_refcount.c */
+extern errcode_t ea_refcount_create(int size, ext2_refcount_t *ret);
+extern void ea_refcount_free(ext2_refcount_t refcount);
+extern errcode_t ea_refcount_fetch(ext2_refcount_t refcount, blk_t blk,
+				   int *ret);
+extern errcode_t ea_refcount_increment(ext2_refcount_t refcount,
+				       blk_t blk, int *ret);
+extern errcode_t ea_refcount_decrement(ext2_refcount_t refcount,
+				       blk_t blk, int *ret);
+extern errcode_t ea_refcount_store(ext2_refcount_t refcount,
+				   blk_t blk, int count);
+extern blk_t ext2fs_get_refcount_size(ext2_refcount_t refcount);
+extern void ea_refcount_intr_begin(ext2_refcount_t refcount);
+extern blk_t ea_refcount_intr_next(ext2_refcount_t refcount, int *ret);
+
+/* ehandler.c */
+extern const char *ehandler_operation(const char *op);
+extern void ehandler_init(io_channel channel);
+
+/* journal.c */
+extern int e2fsck_check_ext3_journal(e2fsck_t ctx);
+extern int e2fsck_run_ext3_journal(e2fsck_t ctx);
+extern void e2fsck_move_ext3_journal(e2fsck_t ctx);
+
+/* pass1.c */
+extern void e2fsck_use_inode_shortcuts(e2fsck_t ctx, int bool);
+extern int e2fsck_pass1_check_device_inode(ext2_filsys fs,
+					   struct ext2_inode *inode);
+extern int e2fsck_pass1_check_symlink(ext2_filsys fs,
+				      struct ext2_inode *inode, char *buf);
+
+/* pass2.c */
+extern int e2fsck_process_bad_inode(e2fsck_t ctx, ext2_ino_t dir,
+				    ext2_ino_t ino, char *buf);
+
+/* pass3.c */
+extern int e2fsck_reconnect_file(e2fsck_t ctx, ext2_ino_t inode);
+extern errcode_t e2fsck_expand_directory(e2fsck_t ctx, ext2_ino_t dir,
+					 int num, int gauranteed_size);
+extern ext2_ino_t e2fsck_get_lost_and_found(e2fsck_t ctx, int fix);
+extern errcode_t e2fsck_adjust_inode_count(e2fsck_t ctx, ext2_ino_t ino, 
+					   int adj);
+
+
+/* region.c */
+extern region_t region_create(region_addr_t min, region_addr_t max);
+extern void region_free(region_t region);
+extern int region_allocate(region_t region, region_addr_t start, int n);
+
+/* rehash.c */
+errcode_t e2fsck_rehash_dir(e2fsck_t ctx, ext2_ino_t ino);
+void e2fsck_rehash_directories(e2fsck_t ctx);
+
+/* super.c */
+void check_super_block(e2fsck_t ctx);
+errcode_t e2fsck_get_device_size(e2fsck_t ctx);
+
+/* swapfs.c */
+void swap_filesys(e2fsck_t ctx);
+
+/* util.c */
+extern void *e2fsck_allocate_memory(e2fsck_t ctx, unsigned int size,
+				    const char *description);
+extern int ask(e2fsck_t ctx, const char * string, int def);
+extern int ask_yn(const char * string, int def);
+extern void e2fsck_read_bitmaps(e2fsck_t ctx);
+extern void e2fsck_write_bitmaps(e2fsck_t ctx);
+extern void preenhalt(e2fsck_t ctx);
+extern char *string_copy(e2fsck_t ctx, const char *str, int len);
+#ifdef RESOURCE_TRACK
+extern void print_resource_track(const char *desc,
+				 struct resource_track *track);
+extern void init_resource_track(struct resource_track *track);
+#endif
+extern int inode_has_valid_blocks(struct ext2_inode *inode);
+extern void e2fsck_read_inode(e2fsck_t ctx, unsigned long ino,
+			      struct ext2_inode * inode, const char * proc);
+extern void e2fsck_write_inode(e2fsck_t ctx, unsigned long ino,
+			       struct ext2_inode * inode, const char * proc);
+#ifdef MTRACE
+extern void mtrace_print(char *mesg);
+#endif
+extern blk_t get_backup_sb(e2fsck_t ctx, ext2_filsys fs,
+			   const char *name, io_manager manager);
+extern int ext2_file_type(unsigned int mode);
+
+/* unix.c */
+extern void e2fsck_clear_progbar(e2fsck_t ctx);
+extern int e2fsck_simple_progress(e2fsck_t ctx, const char *label,
+				  float percent, unsigned int dpynum);

e2fsprogs/e2fsck/ea_refcount.c

@@ -0,0 +1,479 @@
+/*
+ * ea_refcount.c
+ * 
+ * Copyright (C) 2001 Theodore Ts'o.  This file may be
+ * redistributed under the terms of the GNU Public License.
+ */
+
+#if HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+#include <string.h>
+#include <stdio.h>
+
+#include "e2fsck.h"
+
+/*
+ * The strategy we use for keeping track of EA refcounts is as
+ * follows.  We keep a sorted array of first EA blocks and its
+ * reference counts.  Once the refcount has dropped to zero, it is
+ * removed from the array to save memory space.  Once the EA block is
+ * checked, its bit is set in the block_ea_map bitmap.  
+ */
+struct ea_refcount_el {
+	blk_t	ea_blk;
+	int	ea_count;
+};
+
+struct ea_refcount {
+	blk_t		count;
+	blk_t		size;
+	blk_t		cursor;
+	struct ea_refcount_el	*list;
+};
+
+void ea_refcount_free(ext2_refcount_t refcount)
+{
+	if (!refcount)
+		return;
+
+	if (refcount->list)
+		ext2fs_free_mem(&refcount->list);
+	ext2fs_free_mem(&refcount);
+}
+
+errcode_t ea_refcount_create(int size, ext2_refcount_t *ret)
+{
+	ext2_refcount_t	refcount;
+	errcode_t	retval;
+	size_t		bytes;
+
+	retval = ext2fs_get_mem(sizeof(struct ea_refcount), &refcount);
+	if (retval)
+		return retval;
+	memset(refcount, 0, sizeof(struct ea_refcount));
+
+	if (!size)
+		size = 500;
+	refcount->size = size;
+	bytes = (size_t) (size * sizeof(struct ea_refcount_el));
+#ifdef DEBUG
+	printf("Refcount allocated %d entries, %d bytes.\n",
+	       refcount->size, bytes);
+#endif
+	retval = ext2fs_get_mem(bytes, &refcount->list);
+	if (retval)
+		goto errout;
+	memset(refcount->list, 0, bytes);
+
+	refcount->count = 0;
+	refcount->cursor = 0;
+
+	*ret = refcount;
+	return 0;
+
+errout:
+	ea_refcount_free(refcount);
+	return(retval);
+}
+
+/*
+ * collapse_refcount() --- go through the refcount array, and get rid
+ * of any count == zero entries
+ */
+static void refcount_collapse(ext2_refcount_t refcount)
+{
+	unsigned int	i, j;
+	struct ea_refcount_el	*list;
+
+	list = refcount->list;
+	for (i = 0, j = 0; i < refcount->count; i++) {
+		if (list[i].ea_count) {
+			if (i != j)
+				list[j] = list[i];
+			j++;
+		}
+	}
+#if defined(DEBUG) || defined(TEST_PROGRAM)
+	printf("Refcount_collapse: size was %d, now %d\n",
+	       refcount->count, j);
+#endif
+	refcount->count = j;
+}
+
+
+/*
+ * insert_refcount_el() --- Insert a new entry into the sorted list at a
+ * 	specified position.
+ */
+static struct ea_refcount_el *insert_refcount_el(ext2_refcount_t refcount,
+						 blk_t blk, int pos)
+{
+	struct ea_refcount_el 	*el;
+	errcode_t		retval;
+	blk_t			new_size = 0;
+	int			num;
+
+	if (refcount->count >= refcount->size) {
+		new_size = refcount->size + 100;
+#ifdef DEBUG
+		printf("Reallocating refcount %d entries...\n", new_size);
+#endif	
+		retval = ext2fs_resize_mem((size_t) refcount->size *
+					   sizeof(struct ea_refcount_el),
+					   (size_t) new_size *
+					   sizeof(struct ea_refcount_el),
+					   &refcount->list);
+		if (retval)
+			return 0;
+		refcount->size = new_size;
+	}
+	num = (int) refcount->count - pos;
+	if (num < 0)
+		return 0;	/* should never happen */
+	if (num) {
+		memmove(&refcount->list[pos+1], &refcount->list[pos],
+			sizeof(struct ea_refcount_el) * num);
+	}
+	refcount->count++;
+	el = &refcount->list[pos];
+	el->ea_count = 0;
+	el->ea_blk = blk;
+	return el;
+}
+
+
+/*
+ * get_refcount_el() --- given an block number, try to find refcount
+ * 	information in the sorted list.  If the create flag is set,
+ * 	and we can't find an entry, create one in the sorted list.
+ */
+static struct ea_refcount_el *get_refcount_el(ext2_refcount_t refcount,
+					      blk_t blk, int create)
+{
+	float	range;
+	int	low, high, mid;
+	blk_t	lowval, highval;
+
+	if (!refcount || !refcount->list)
+		return 0;
+retry:
+	low = 0;
+	high = (int) refcount->count-1;
+	if (create && ((refcount->count == 0) ||
+		       (blk > refcount->list[high].ea_blk))) {
+		if (refcount->count >= refcount->size)
+			refcount_collapse(refcount);
+
+		return insert_refcount_el(refcount, blk,
+					  (unsigned) refcount->count);
+	}
+	if (refcount->count == 0)
+		return 0;
+	
+	if (refcount->cursor >= refcount->count)
+		refcount->cursor = 0;
+	if (blk == refcount->list[refcount->cursor].ea_blk)
+		return &refcount->list[refcount->cursor++];
+#ifdef DEBUG
+	printf("Non-cursor get_refcount_el: %u\n", blk);
+#endif
+	while (low <= high) {
+#if 0
+		mid = (low+high)/2;
+#else
+		if (low == high)
+			mid = low;
+		else {
+			/* Interpolate for efficiency */
+			lowval = refcount->list[low].ea_blk;
+			highval = refcount->list[high].ea_blk;
+
+			if (blk < lowval)
+				range = 0;
+			else if (blk > highval)
+				range = 1;
+			else 
+				range = ((float) (blk - lowval)) /
+					(highval - lowval);
+			mid = low + ((int) (range * (high-low)));
+		}
+#endif
+		if (blk == refcount->list[mid].ea_blk) {
+			refcount->cursor = mid+1;
+			return &refcount->list[mid];
+		}
+		if (blk < refcount->list[mid].ea_blk)
+			high = mid-1;
+		else
+			low = mid+1;
+	}
+	/*
+	 * If we need to create a new entry, it should be right at
+	 * low (where high will be left at low-1).
+	 */
+	if (create) {
+		if (refcount->count >= refcount->size) {
+			refcount_collapse(refcount);
+			if (refcount->count < refcount->size)
+				goto retry;
+		}
+		return insert_refcount_el(refcount, blk, low);
+	}
+	return 0;
+}
+
+errcode_t ea_refcount_fetch(ext2_refcount_t refcount, blk_t blk,
+				int *ret)
+{
+	struct ea_refcount_el	*el;
+	
+	el = get_refcount_el(refcount, blk, 0);
+	if (!el) {
+		*ret = 0;
+		return 0;
+	}
+	*ret = el->ea_count;
+	return 0;
+}
+
+errcode_t ea_refcount_increment(ext2_refcount_t refcount, blk_t blk, int *ret)
+{
+	struct ea_refcount_el	*el;
+
+	el = get_refcount_el(refcount, blk, 1);
+	if (!el)
+		return EXT2_ET_NO_MEMORY;
+	el->ea_count++;
+
+	if (ret)
+		*ret = el->ea_count;
+	return 0;
+}
+
+errcode_t ea_refcount_decrement(ext2_refcount_t refcount, blk_t blk, int *ret)
+{
+	struct ea_refcount_el	*el;
+
+	el = get_refcount_el(refcount, blk, 0);
+	if (!el || el->ea_count == 0)
+		return EXT2_ET_INVALID_ARGUMENT;
+
+	el->ea_count--;
+
+	if (ret)
+		*ret = el->ea_count;
+	return 0;
+}
+
+errcode_t ea_refcount_store(ext2_refcount_t refcount, blk_t blk, int count)
+{
+	struct ea_refcount_el	*el;
+
+	/*
+	 * Get the refcount element
+	 */
+	el = get_refcount_el(refcount, blk, count ? 1 : 0);
+	if (!el)
+		return count ? EXT2_ET_NO_MEMORY : 0;
+	el->ea_count = count;
+	return 0;
+}
+
+blk_t ext2fs_get_refcount_size(ext2_refcount_t refcount)
+{
+	if (!refcount)
+		return 0;
+
+	return refcount->size;
+}
+
+void ea_refcount_intr_begin(ext2_refcount_t refcount)
+{
+	refcount->cursor = 0;
+}
+
+
+blk_t ea_refcount_intr_next(ext2_refcount_t refcount,
+				int *ret)
+{
+	struct ea_refcount_el	*list;
+	
+	while (1) {
+		if (refcount->cursor >= refcount->count)
+			return 0;
+		list = refcount->list;
+		if (list[refcount->cursor].ea_count) {
+			if (ret)
+				*ret = list[refcount->cursor].ea_count;
+			return list[refcount->cursor++].ea_blk;
+		}
+		refcount->cursor++;
+	}
+}
+
+
+#ifdef TEST_PROGRAM
+
+errcode_t ea_refcount_validate(ext2_refcount_t refcount, FILE *out)
+{
+	errcode_t	ret = 0;
+	int		i;
+	const char *bad = "bad refcount";
+	
+	if (refcount->count > refcount->size) {
+		fprintf(out, "%s: count > size\n", bad);
+		return EXT2_ET_INVALID_ARGUMENT;
+	}
+	for (i=1; i < refcount->count; i++) {
+		if (refcount->list[i-1].ea_blk >= refcount->list[i].ea_blk) {
+			fprintf(out, "%s: list[%d].blk=%u, list[%d].blk=%u\n",
+				bad, i-1, refcount->list[i-1].ea_blk,
+				i, refcount->list[i].ea_blk);
+			ret = EXT2_ET_INVALID_ARGUMENT;
+		}
+	}
+	return ret;
+}
+
+#define BCODE_END	0
+#define BCODE_CREATE	1
+#define BCODE_FREE	2
+#define BCODE_STORE	3
+#define BCODE_INCR	4
+#define BCODE_DECR	5
+#define BCODE_FETCH	6
+#define BCODE_VALIDATE	7
+#define BCODE_LIST	8
+#define BCODE_COLLAPSE 9
+
+int bcode_program[] = {
+	BCODE_CREATE, 5,
+	BCODE_STORE, 3, 3,
+	BCODE_STORE, 4, 4,
+	BCODE_STORE, 1, 1,
+	BCODE_STORE, 8, 8,
+	BCODE_STORE, 2, 2,
+	BCODE_STORE, 4, 0,
+	BCODE_STORE, 2, 0,
+	BCODE_STORE, 6, 6,
+	BCODE_VALIDATE,
+	BCODE_STORE, 4, 4,
+	BCODE_STORE, 2, 2,
+	BCODE_FETCH, 1,
+	BCODE_FETCH, 2,
+	BCODE_INCR, 3,
+	BCODE_INCR, 3,
+	BCODE_DECR, 4,
+	BCODE_STORE, 4, 4,
+	BCODE_VALIDATE,
+	BCODE_STORE, 20, 20,
+	BCODE_STORE, 40, 40,
+	BCODE_STORE, 30, 30,
+	BCODE_STORE, 10, 10,
+	BCODE_DECR, 30,
+	BCODE_FETCH, 30,
+	BCODE_DECR, 2,
+	BCODE_DECR, 2,
+	BCODE_COLLAPSE,
+	BCODE_LIST,
+	BCODE_VALIDATE,
+	BCODE_FREE,
+	BCODE_END
+};
+
+int main(int argc, char **argv)
+{
+	int	i = 0;
+	ext2_refcount_t refcount;
+	int		size, arg;
+	blk_t		blk;
+	errcode_t	retval;
+
+	while (1) {
+		switch (bcode_program[i++]) {
+		case BCODE_END:
+			exit(0);
+		case BCODE_CREATE:
+			size = bcode_program[i++];
+			retval = ea_refcount_create(size, &refcount);
+			if (retval) {
+				com_err("ea_refcount_create",
+					retval, "");
+				exit(1);
+			} else
+				printf("Creating refcount with size %d\n",
+				       size);
+			break;
+		case BCODE_FREE:
+			ea_refcount_free(refcount);
+			refcount = 0;
+			printf("Freeing refcount\n");
+			break;
+		case BCODE_STORE:
+			blk = (blk_t) bcode_program[i++];
+			arg = bcode_program[i++];
+			retval = ea_refcount_store(refcount, blk, arg);
+			printf("Storing blk %u with value %d\n", blk, arg);
+			if (retval)
+				com_err("ea_refcount_store", retval, "");
+			break;
+		case BCODE_FETCH:
+			blk = (blk_t) bcode_program[i++];
+			retval = ea_refcount_fetch(refcount, blk, &arg);
+			if (retval)
+				com_err("ea_refcount_fetch", retval, "");
+			else
+				printf("bcode_fetch(%u) returns %d\n",
+				       blk, arg);
+			break;
+		case BCODE_INCR:
+			blk = (blk_t) bcode_program[i++];
+			retval = ea_refcount_increment(refcount, blk,
+							   &arg);
+			if (retval)
+				com_err("ea_refcount_increment", retval,
+					"");
+			else
+				printf("bcode_increment(%u) returns %d\n",
+				       blk, arg);
+			break;
+		case BCODE_DECR:
+			blk = (blk_t) bcode_program[i++];
+			retval = ea_refcount_decrement(refcount, blk,
+							   &arg);
+			if (retval)
+				com_err("ea_refcount_decrement", retval,
+					"while decrementing blk %u", blk);
+			else
+				printf("bcode_decrement(%u) returns %d\n",
+				       blk, arg);
+			break;
+		case BCODE_VALIDATE:
+			retval = ea_refcount_validate(refcount, stderr);
+			if (retval)
+				com_err("ea_refcount_validate",
+					retval, "");
+			else
+				printf("Refcount validation OK.\n");
+			break;
+		case BCODE_LIST:
+			ea_refcount_intr_begin(refcount);
+			while (1) {
+				blk = ea_refcount_intr_next(refcount, 
+								&arg);
+				if (!blk)
+					break;
+				printf("\tblk=%u, count=%d\n", blk,
+				       arg);
+			}
+			break;
+		case BCODE_COLLAPSE:
+			refcount_collapse(refcount);
+			break;
+		}
+		
+	}
+}
+
+#endif

e2fsprogs/e2fsck/ehandler.c

@@ -0,0 +1,124 @@
+/*
+ * ehandler.c --- handle bad block errors which come up during the
+ * 	course of an e2fsck session.
+ * 
+ * Copyright (C) 1994 Theodore Ts'o.  This file may be redistributed
+ * under the terms of the GNU Public License.
+ */
+
+#include <stdlib.h>
+#include <unistd.h>
+#include <string.h>
+#include <ctype.h>
+#include <termios.h>
+
+#include "e2fsck.h"
+
+#include <sys/time.h>
+#include <sys/resource.h>
+
+static const char *operation;
+
+static errcode_t e2fsck_handle_read_error(io_channel channel,
+					  unsigned long block,
+					  int count,
+					  void *data,
+					  size_t size EXT2FS_ATTR((unused)),
+					  int actual EXT2FS_ATTR((unused)),
+					  errcode_t error)
+{
+	int	i;
+	char	*p;
+	ext2_filsys fs = (ext2_filsys) channel->app_data;
+	e2fsck_t ctx;
+
+	ctx = (e2fsck_t) fs->priv_data;
+
+	/*
+	 * If more than one block was read, try reading each block
+	 * separately.  We could use the actual bytes read to figure
+	 * out where to start, but we don't bother.
+	 */
+	if (count > 1) {
+		p = (char *) data;
+		for (i=0; i < count; i++, p += channel->block_size, block++) {
+			error = io_channel_read_blk(channel, block,
+						    1, p);
+			if (error)
+				return error;
+		}
+		return 0;
+	}
+	if (operation)
+		printf(_("Error reading block %lu (%s) while %s.  "), block,
+		       error_message(error), operation);
+	else
+		printf(_("Error reading block %lu (%s).  "), block,
+		       error_message(error));
+	preenhalt(ctx);
+	if (ask(ctx, _("Ignore error"), 1)) {
+		if (ask(ctx, _("Force rewrite"), 1))
+			io_channel_write_blk(channel, block, 1, data);
+		return 0;
+	}
+
+	return error;
+}
+
+static errcode_t e2fsck_handle_write_error(io_channel channel,
+					    unsigned long block,
+					    int count,
+					    const void *data,
+					    size_t size EXT2FS_ATTR((unused)),
+					    int actual EXT2FS_ATTR((unused)),
+					    errcode_t error)
+{
+	int		i;
+	const char	*p;
+	ext2_filsys fs = (ext2_filsys) channel->app_data;
+	e2fsck_t ctx;
+	
+	ctx = (e2fsck_t) fs->priv_data;
+
+	/*
+	 * If more than one block was written, try writing each block
+	 * separately.  We could use the actual bytes read to figure
+	 * out where to start, but we don't bother.
+	 */
+	if (count > 1) {
+		p = (const char *) data;
+		for (i=0; i < count; i++, p += channel->block_size, block++) {
+			error = io_channel_write_blk(channel, block,
+						     1, p);
+			if (error)
+				return error;
+		}
+		return 0;
+	}
+	
+	if (operation)
+		printf(_("Error writing block %lu (%s) while %s.  "), block,
+		       error_message(error), operation);
+	else
+		printf(_("Error writing block %lu (%s).  "), block,
+		       error_message(error));
+	preenhalt(ctx);
+	if (ask(ctx, _("Ignore error"), 1))
+		return 0;
+
+	return error;
+}
+
+const char *ehandler_operation(const char *op)
+{
+	const char *ret = operation;
+
+	operation = op;
+	return ret;
+}
+
+void ehandler_init(io_channel channel)
+{
+	channel->read_error = e2fsck_handle_read_error;
+	channel->write_error = e2fsck_handle_write_error;
+}

e2fsprogs/e2fsck/jfs_user.h

@@ -0,0 +1,120 @@
+/*
+ * Compatibility header file for e2fsck which should be included
+ * instead of linux/jfs.h
+ *
+ * Copyright (C) 2000 Stephen C. Tweedie
+ *
+ * This file may be redistributed under the terms of the
+ * GNU General Public License version 2 or at your discretion
+ * any later version.
+ */
+
+/*
+ * Pull in the definition of the e2fsck context structure
+ */
+#include "e2fsck.h"
+
+struct buffer_head {
+	char		b_data[8192];
+	e2fsck_t	b_ctx;
+	io_channel 	b_io;
+	int	 	b_size;
+	blk_t	 	b_blocknr;
+	int	 	b_dirty;
+	int	 	b_uptodate;
+	int	 	b_err;
+};
+
+struct inode {
+	e2fsck_t	i_ctx;
+	ext2_ino_t	i_ino;
+	struct ext2_inode i_ext2;
+};
+
+struct kdev_s {
+	e2fsck_t	k_ctx;
+	int		k_dev;
+};
+
+#define K_DEV_FS	1
+#define K_DEV_JOURNAL	2
+
+typedef struct kdev_s *kdev_t;
+
+#define lock_buffer(bh) do {} while(0)
+#define unlock_buffer(bh) do {} while(0)
+#define buffer_req(bh) 1
+#define do_readahead(journal, start) do {} while(0)
+	
+extern e2fsck_t e2fsck_global_ctx;  /* Try your very best not to use this! */
+
+typedef struct {
+	int	object_length;
+} kmem_cache_t;
+
+#define kmem_cache_alloc(cache,flags) malloc((cache)->object_length)
+#define kmem_cache_free(cache,obj) free(obj)
+#define kmem_cache_create(name,len,a,b,c,d) do_cache_create(len)
+#define kmem_cache_destroy(cache) do_cache_destroy(cache)
+#define kmalloc(len,flags) malloc(len)
+#define kfree(p) free(p)
+
+/*
+ * We use the standard libext2fs portability tricks for inline
+ * functions.  
+ */
+extern kmem_cache_t * do_cache_create(int len);
+extern void do_cache_destroy(kmem_cache_t *cache);
+	
+#if (defined(E2FSCK_INCLUDE_INLINE_FUNCS) || !defined(NO_INLINE_FUNCS))
+#ifdef E2FSCK_INCLUDE_INLINE_FUNCS
+#define _INLINE_ extern
+#else
+#ifdef __GNUC__
+#define _INLINE_ extern __inline__
+#else				/* For Watcom C */
+#define _INLINE_ extern inline
+#endif
+#endif
+
+_INLINE_ kmem_cache_t * do_cache_create(int len)
+{
+	kmem_cache_t *new_cache;
+	new_cache = malloc(sizeof(*new_cache));
+	if (new_cache)
+		new_cache->object_length = len;
+	return new_cache;
+}
+
+_INLINE_ void do_cache_destroy(kmem_cache_t *cache)
+{
+	free(cache);
+}
+#undef _INLINE_
+#endif
+
+#define __init
+
+/*
+ * Now pull in the real linux/jfs.h definitions.
+ */
+#include <ext2fs/kernel-jbd.h>
+
+/*
+ * Kernel compatibility functions are defined in journal.c
+ */
+int journal_bmap(journal_t *journal, blk_t block, unsigned long *phys);
+struct buffer_head *getblk(kdev_t ctx, blk_t blocknr, int blocksize);
+void sync_blockdev(kdev_t kdev);
+void ll_rw_block(int rw, int dummy, struct buffer_head *bh[]);
+void mark_buffer_dirty(struct buffer_head *bh);
+void mark_buffer_uptodate(struct buffer_head *bh, int val);
+void brelse(struct buffer_head *bh);
+int buffer_uptodate(struct buffer_head *bh);
+void wait_on_buffer(struct buffer_head *bh);
+
+/*
+ * Define newer 2.5 interfaces
+ */
+#define __getblk(dev, blocknr, blocksize) getblk(dev, blocknr, blocksize)
+#define set_buffer_uptodate(bh) mark_buffer_uptodate(bh, 1)

e2fsprogs/e2fsck/journal.c

@@ -0,0 +1,960 @@
+/*
+ * journal.c --- code for handling the "ext3" journal
+ *
+ * Copyright (C) 2000 Andreas Dilger
+ * Copyright (C) 2000 Theodore Ts'o
+ *
+ * Parts of the code are based on fs/jfs/journal.c by Stephen C. Tweedie
+ * Copyright (C) 1999 Red Hat Software
+ *
+ * This file may be redistributed under the terms of the
+ * GNU General Public License version 2 or at your discretion
+ * any later version.
+ */
+
+#ifdef HAVE_SYS_MOUNT_H
+#include <sys/param.h>
+#include <sys/mount.h>
+#define MNT_FL (MS_MGC_VAL | MS_RDONLY)
+#endif
+#ifdef HAVE_SYS_STAT_H
+#include <sys/stat.h>
+#endif
+
+#define E2FSCK_INCLUDE_INLINE_FUNCS
+#include "jfs_user.h"
+#include "problem.h"
+#include "uuid/uuid.h"
+
+#ifdef __CONFIG_JBD_DEBUG__E2FS		/* Enabled by configure --enable-jfs-debug */
+static int bh_count = 0;
+#endif
+
+/*
+ * Define USE_INODE_IO to use the inode_io.c / fileio.c codepaths.
+ * This creates a larger static binary, and a smaller binary using
+ * shared libraries.  It's also probably slightly less CPU-efficient,
+ * which is why it's not on by default.  But, it's a good way of
+ * testing the functions in inode_io.c and fileio.c.
+ */
+#undef USE_INODE_IO
+
+/* Kernel compatibility functions for handling the journal.  These allow us
+ * to use the recovery.c file virtually unchanged from the kernel, so we
+ * don't have to do much to keep kernel and user recovery in sync.
+ */
+int journal_bmap(journal_t *journal, blk_t block, unsigned long *phys)
+{
+#ifdef USE_INODE_IO
+	*phys = block;
+	return 0;
+#else
+	struct inode 	*inode = journal->j_inode;
+	errcode_t	retval;
+	blk_t		pblk;
+
+	if (!inode) {
+		*phys = block;
+		return 0;
+	}
+
+	retval= ext2fs_bmap(inode->i_ctx->fs, inode->i_ino, 
+			    &inode->i_ext2, NULL, 0, block, &pblk);
+	*phys = pblk;
+	return (retval);
+#endif
+}
+
+struct buffer_head *getblk(kdev_t kdev, blk_t blocknr, int blocksize)
+{
+	struct buffer_head *bh;
+
+	bh = e2fsck_allocate_memory(kdev->k_ctx, sizeof(*bh), "block buffer");
+	if (!bh)
+		return NULL;
+
+	jfs_debug(4, "getblk for block %lu (%d bytes)(total %d)\n",
+		  (unsigned long) blocknr, blocksize, ++bh_count);
+
+	bh->b_ctx = kdev->k_ctx;
+	if (kdev->k_dev == K_DEV_FS)
+		bh->b_io = kdev->k_ctx->fs->io;
+	else 
+		bh->b_io = kdev->k_ctx->journal_io;
+	bh->b_size = blocksize;
+	bh->b_blocknr = blocknr;
+
+	return bh;
+}
+
+void sync_blockdev(kdev_t kdev)
+{
+	io_channel	io;
+
+	if (kdev->k_dev == K_DEV_FS)
+		io = kdev->k_ctx->fs->io;
+	else 
+		io = kdev->k_ctx->journal_io;
+
+	io_channel_flush(io);
+}
+
+void ll_rw_block(int rw, int nr, struct buffer_head *bhp[])
+{
+	int retval;
+	struct buffer_head *bh;
+
+	for (; nr > 0; --nr) {
+		bh = *bhp++;
+		if (rw == READ && !bh->b_uptodate) {
+			jfs_debug(3, "reading block %lu/%p\n", 
+				  (unsigned long) bh->b_blocknr, (void *) bh);
+			retval = io_channel_read_blk(bh->b_io, 
+						     bh->b_blocknr,
+						     1, bh->b_data);
+			if (retval) {
+				com_err(bh->b_ctx->device_name, retval,
+					"while reading block %lu\n", 
+					(unsigned long) bh->b_blocknr);
+				bh->b_err = retval;
+				continue;
+			}
+			bh->b_uptodate = 1;
+		} else if (rw == WRITE && bh->b_dirty) {
+			jfs_debug(3, "writing block %lu/%p\n", 
+				  (unsigned long) bh->b_blocknr, (void *) bh);
+			retval = io_channel_write_blk(bh->b_io, 
+						      bh->b_blocknr,
+						      1, bh->b_data);
+			if (retval) {
+				com_err(bh->b_ctx->device_name, retval,
+					"while writing block %lu\n", 
+					(unsigned long) bh->b_blocknr);
+				bh->b_err = retval;
+				continue;
+			}
+			bh->b_dirty = 0;
+			bh->b_uptodate = 1;
+		} else {
+			jfs_debug(3, "no-op %s for block %lu\n",
+				  rw == READ ? "read" : "write", 
+				  (unsigned long) bh->b_blocknr);
+		}
+	}
+}
+
+void mark_buffer_dirty(struct buffer_head *bh)
+{
+	bh->b_dirty = 1;
+}
+
+static void mark_buffer_clean(struct buffer_head * bh)
+{
+	bh->b_dirty = 0;
+}
+
+void brelse(struct buffer_head *bh)
+{
+	if (bh->b_dirty)
+		ll_rw_block(WRITE, 1, &bh);
+	jfs_debug(3, "freeing block %lu/%p (total %d)\n",
+		  (unsigned long) bh->b_blocknr, (void *) bh, --bh_count);
+	ext2fs_free_mem(&bh);
+}
+
+int buffer_uptodate(struct buffer_head *bh)
+{
+	return bh->b_uptodate;
+}
+
+void mark_buffer_uptodate(struct buffer_head *bh, int val)
+{
+	bh->b_uptodate = val;
+}
+
+void wait_on_buffer(struct buffer_head *bh)
+{
+	if (!bh->b_uptodate)
+		ll_rw_block(READ, 1, &bh);
+}
+
+
+static void e2fsck_clear_recover(e2fsck_t ctx, int error)
+{
+	ctx->fs->super->s_feature_incompat &= ~EXT3_FEATURE_INCOMPAT_RECOVER;
+
+	/* if we had an error doing journal recovery, we need a full fsck */
+	if (error)
+		ctx->fs->super->s_state &= ~EXT2_VALID_FS;
+	ext2fs_mark_super_dirty(ctx->fs);
+}
+
+static errcode_t e2fsck_get_journal(e2fsck_t ctx, journal_t **ret_journal)
+{
+	struct ext2_super_block *sb = ctx->fs->super;
+	struct ext2_super_block jsuper;
+	struct problem_context	pctx;
+	struct buffer_head 	*bh;
+	struct inode		*j_inode = NULL;
+	struct kdev_s		*dev_fs = NULL, *dev_journal;
+	const char		*journal_name = 0;
+	journal_t		*journal = NULL;
+	errcode_t		retval = 0;
+	io_manager		io_ptr = 0;
+	unsigned long		start = 0;
+	blk_t			blk;
+	int			ext_journal = 0;
+	int			tried_backup_jnl = 0;
+	int			i;
+		
+	clear_problem_context(&pctx);
+	
+	journal = e2fsck_allocate_memory(ctx, sizeof(journal_t), "journal");
+	if (!journal) {
+		return EXT2_ET_NO_MEMORY;
+	}
+
+	dev_fs = e2fsck_allocate_memory(ctx, 2*sizeof(struct kdev_s), "kdev");
+	if (!dev_fs) {
+		retval = EXT2_ET_NO_MEMORY;
+		goto errout;
+	}
+	dev_journal = dev_fs+1;
+	
+	dev_fs->k_ctx = dev_journal->k_ctx = ctx;
+	dev_fs->k_dev = K_DEV_FS;
+	dev_journal->k_dev = K_DEV_JOURNAL;
+	
+	journal->j_dev = dev_journal;
+	journal->j_fs_dev = dev_fs;
+	journal->j_inode = NULL;
+	journal->j_blocksize = ctx->fs->blocksize;
+
+	if (uuid_is_null(sb->s_journal_uuid)) {
+		if (!sb->s_journal_inum)
+			return EXT2_ET_BAD_INODE_NUM;
+		j_inode = e2fsck_allocate_memory(ctx, sizeof(*j_inode),
+						 "journal inode");
+		if (!j_inode) {
+			retval = EXT2_ET_NO_MEMORY;
+			goto errout;
+		}
+
+		j_inode->i_ctx = ctx;
+		j_inode->i_ino = sb->s_journal_inum;
+		
+		if ((retval = ext2fs_read_inode(ctx->fs,
+						sb->s_journal_inum,
+						&j_inode->i_ext2))) {
+		try_backup_journal:
+			if (sb->s_jnl_backup_type != EXT3_JNL_BACKUP_BLOCKS ||
+			    tried_backup_jnl)
+				goto errout;
+			memset(&j_inode->i_ext2, 0, sizeof(struct ext2_inode));
+			memcpy(&j_inode->i_ext2.i_block[0], sb->s_jnl_blocks, 
+			       EXT2_N_BLOCKS*4);
+			j_inode->i_ext2.i_size = sb->s_jnl_blocks[16];
+			j_inode->i_ext2.i_links_count = 1;
+			j_inode->i_ext2.i_mode = LINUX_S_IFREG | 0600;
+			tried_backup_jnl++;
+		}
+		if (!j_inode->i_ext2.i_links_count ||
+		    !LINUX_S_ISREG(j_inode->i_ext2.i_mode)) {
+			retval = EXT2_ET_NO_JOURNAL;
+			goto try_backup_journal;
+		}
+		if (j_inode->i_ext2.i_size / journal->j_blocksize <
+		    JFS_MIN_JOURNAL_BLOCKS) {
+			retval = EXT2_ET_JOURNAL_TOO_SMALL;
+			goto try_backup_journal;
+		}
+		for (i=0; i < EXT2_N_BLOCKS; i++) {
+			blk = j_inode->i_ext2.i_block[i];
+			if (!blk) {
+				if (i < EXT2_NDIR_BLOCKS) {
+					retval = EXT2_ET_JOURNAL_TOO_SMALL;
+					goto try_backup_journal;
+				}
+				continue;
+			}
+			if (blk < sb->s_first_data_block ||
+			    blk >= sb->s_blocks_count) {
+				retval = EXT2_ET_BAD_BLOCK_NUM;
+				goto try_backup_journal;
+			}
+		}
+		journal->j_maxlen = j_inode->i_ext2.i_size / journal->j_blocksize;
+
+#ifdef USE_INODE_IO
+		retval = ext2fs_inode_io_intern2(ctx->fs, sb->s_journal_inum,
+						 &j_inode->i_ext2,
+						 &journal_name);
+		if (retval)
+			goto errout;
+
+		io_ptr = inode_io_manager;
+#else
+		journal->j_inode = j_inode;
+		ctx->journal_io = ctx->fs->io;
+		if ((retval = journal_bmap(journal, 0, &start)) != 0)
+			goto errout;
+#endif
+	} else {
+		ext_journal = 1;
+		if (!ctx->journal_name) {
+			char uuid[37];
+
+			uuid_unparse(sb->s_journal_uuid, uuid);
+			ctx->journal_name = blkid_get_devname(ctx->blkid,
+							      "UUID", uuid);
+			if (!ctx->journal_name)
+				ctx->journal_name = blkid_devno_to_devname(sb->s_journal_dev);
+		}
+		journal_name = ctx->journal_name;
+			
+		if (!journal_name) {
+			fix_problem(ctx, PR_0_CANT_FIND_JOURNAL, &pctx);
+			return EXT2_ET_LOAD_EXT_JOURNAL;
+		}
+		
+		jfs_debug(1, "Using journal file %s\n", journal_name);
+		io_ptr = unix_io_manager;
+	}
+
+#if 0
+	test_io_backing_manager = io_ptr;
+	io_ptr = test_io_manager;
+#endif
+#ifndef USE_INODE_IO
+	if (ext_journal)
+#endif
+		retval = io_ptr->open(journal_name, IO_FLAG_RW,
+				      &ctx->journal_io);
+	if (retval)
+		goto errout;
+
+	io_channel_set_blksize(ctx->journal_io, ctx->fs->blocksize);
+
+	if (ext_journal) {
+		if (ctx->fs->blocksize == 1024)
+			start = 1;
+		bh = getblk(dev_journal, start, ctx->fs->blocksize);
+		if (!bh) {
+			retval = EXT2_ET_NO_MEMORY;
+			goto errout;
+		}
+		ll_rw_block(READ, 1, &bh);
+		if ((retval = bh->b_err) != 0)
+			goto errout;
+		memcpy(&jsuper, start ? bh->b_data :  bh->b_data + 1024,
+		       sizeof(jsuper));
+		brelse(bh);
+#ifdef EXT2FS_ENABLE_SWAPFS
+		if (jsuper.s_magic == ext2fs_swab16(EXT2_SUPER_MAGIC)) 
+			ext2fs_swap_super(&jsuper);
+#endif
+		if (jsuper.s_magic != EXT2_SUPER_MAGIC ||
+		    !(jsuper.s_feature_incompat & EXT3_FEATURE_INCOMPAT_JOURNAL_DEV)) {
+			fix_problem(ctx, PR_0_EXT_JOURNAL_BAD_SUPER, &pctx);
+			retval = EXT2_ET_LOAD_EXT_JOURNAL;
+			goto errout;
+		}
+		/* Make sure the journal UUID is correct */
+		if (memcmp(jsuper.s_uuid, ctx->fs->super->s_journal_uuid,
+			   sizeof(jsuper.s_uuid))) {
+			fix_problem(ctx, PR_0_JOURNAL_BAD_UUID, &pctx);
+			retval = EXT2_ET_LOAD_EXT_JOURNAL;
+			goto errout;
+		}
+		
+		journal->j_maxlen = jsuper.s_blocks_count;
+		start++;
+	}
+
+	if (!(bh = getblk(dev_journal, start, journal->j_blocksize))) {
+		retval = EXT2_ET_NO_MEMORY;
+		goto errout;
+	}
+	
+	journal->j_sb_buffer = bh;
+	journal->j_superblock = (journal_superblock_t *)bh->b_data;
+	
+#ifdef USE_INODE_IO
+	if (j_inode)
+		ext2fs_free_mem(&j_inode);
+#endif
+
+	*ret_journal = journal;
+	return 0;
+
+errout:
+	if (dev_fs)
+		ext2fs_free_mem(&dev_fs);
+	if (j_inode)
+		ext2fs_free_mem(&j_inode);
+	if (journal)
+		ext2fs_free_mem(&journal);
+	return retval;
+	
+}
+
+static errcode_t e2fsck_journal_fix_bad_inode(e2fsck_t ctx,
+					      struct problem_context *pctx)
+{
+	struct ext2_super_block *sb = ctx->fs->super;
+	int recover = ctx->fs->super->s_feature_incompat &
+		EXT3_FEATURE_INCOMPAT_RECOVER;
+	int has_journal = ctx->fs->super->s_feature_compat &
+		EXT3_FEATURE_COMPAT_HAS_JOURNAL;
+
+	if (has_journal || sb->s_journal_inum) {
+		/* The journal inode is bogus, remove and force full fsck */
+		pctx->ino = sb->s_journal_inum;
+		if (fix_problem(ctx, PR_0_JOURNAL_BAD_INODE, pctx)) {
+			if (has_journal && sb->s_journal_inum)
+				printf("*** ext3 journal has been deleted - "
+				       "filesystem is now ext2 only ***\n\n");
+			sb->s_feature_compat &= ~EXT3_FEATURE_COMPAT_HAS_JOURNAL;
+			sb->s_journal_inum = 0;
+			ctx->flags |= E2F_FLAG_JOURNAL_INODE; /* FIXME: todo */
+			e2fsck_clear_recover(ctx, 1);
+			return 0;
+		}
+		return EXT2_ET_BAD_INODE_NUM;
+	} else if (recover) {
+		if (fix_problem(ctx, PR_0_JOURNAL_RECOVER_SET, pctx)) {
+			e2fsck_clear_recover(ctx, 1);
+			return 0;
+		}
+		return EXT2_ET_UNSUPP_FEATURE;
+	}
+	return 0;
+}
+
+#define V1_SB_SIZE	0x0024
+static void clear_v2_journal_fields(journal_t *journal)
+{
+	e2fsck_t ctx = journal->j_dev->k_ctx;
+	struct problem_context pctx;
+
+	clear_problem_context(&pctx);
+
+	if (!fix_problem(ctx, PR_0_CLEAR_V2_JOURNAL, &pctx))
+		return;
+
+	memset(((char *) journal->j_superblock) + V1_SB_SIZE, 0,
+	       ctx->fs->blocksize-V1_SB_SIZE);
+	mark_buffer_dirty(journal->j_sb_buffer);
+}
+
+
+static errcode_t e2fsck_journal_load(journal_t *journal)
+{
+	e2fsck_t ctx = journal->j_dev->k_ctx;
+	journal_superblock_t *jsb;
+	struct buffer_head *jbh = journal->j_sb_buffer;
+	struct problem_context pctx;
+
+	clear_problem_context(&pctx);
+
+	ll_rw_block(READ, 1, &jbh);
+	if (jbh->b_err) {
+		com_err(ctx->device_name, jbh->b_err,
+			_("reading journal superblock\n"));
+		return jbh->b_err;
+	}
+
+	jsb = journal->j_superblock;
+	/* If we don't even have JFS_MAGIC, we probably have a wrong inode */
+	if (jsb->s_header.h_magic != htonl(JFS_MAGIC_NUMBER))
+		return e2fsck_journal_fix_bad_inode(ctx, &pctx);
+
+	switch (ntohl(jsb->s_header.h_blocktype)) {
+	case JFS_SUPERBLOCK_V1:
+		journal->j_format_version = 1;
+		if (jsb->s_feature_compat ||
+		    jsb->s_feature_incompat ||
+		    jsb->s_feature_ro_compat ||
+		    jsb->s_nr_users)
+			clear_v2_journal_fields(journal);
+		break;
+		
+	case JFS_SUPERBLOCK_V2:
+		journal->j_format_version = 2;
+		if (ntohl(jsb->s_nr_users) > 1 &&
+		    uuid_is_null(ctx->fs->super->s_journal_uuid))
+			clear_v2_journal_fields(journal);
+		if (ntohl(jsb->s_nr_users) > 1) {
+			fix_problem(ctx, PR_0_JOURNAL_UNSUPP_MULTIFS, &pctx);
+			return EXT2_ET_JOURNAL_UNSUPP_VERSION;
+		}
+		break;
+
+	/*
+	 * These should never appear in a journal super block, so if
+	 * they do, the journal is badly corrupted.
+	 */
+	case JFS_DESCRIPTOR_BLOCK:
+	case JFS_COMMIT_BLOCK:
+	case JFS_REVOKE_BLOCK:
+		return EXT2_ET_CORRUPT_SUPERBLOCK;
+		
+	/* If we don't understand the superblock major type, but there
+	 * is a magic number, then it is likely to be a new format we
+	 * just don't understand, so leave it alone. */
+	default:
+		return EXT2_ET_JOURNAL_UNSUPP_VERSION;
+	}
+
+	if (JFS_HAS_INCOMPAT_FEATURE(journal, ~JFS_KNOWN_INCOMPAT_FEATURES))
+		return EXT2_ET_UNSUPP_FEATURE;
+	
+	if (JFS_HAS_RO_COMPAT_FEATURE(journal, ~JFS_KNOWN_ROCOMPAT_FEATURES))
+		return EXT2_ET_RO_UNSUPP_FEATURE;
+
+	/* We have now checked whether we know enough about the journal
+	 * format to be able to proceed safely, so any other checks that
+	 * fail we should attempt to recover from. */
+	if (jsb->s_blocksize != htonl(journal->j_blocksize)) {
+		com_err(ctx->program_name, EXT2_ET_CORRUPT_SUPERBLOCK,
+			_("%s: no valid journal superblock found\n"),
+			ctx->device_name);
+		return EXT2_ET_CORRUPT_SUPERBLOCK;
+	}
+
+	if (ntohl(jsb->s_maxlen) < journal->j_maxlen)
+		journal->j_maxlen = ntohl(jsb->s_maxlen);
+	else if (ntohl(jsb->s_maxlen) > journal->j_maxlen) {
+		com_err(ctx->program_name, EXT2_ET_CORRUPT_SUPERBLOCK,
+			_("%s: journal too short\n"),
+			ctx->device_name);
+		return EXT2_ET_CORRUPT_SUPERBLOCK;
+	}
+
+	journal->j_tail_sequence = ntohl(jsb->s_sequence);
+	journal->j_transaction_sequence = journal->j_tail_sequence;
+	journal->j_tail = ntohl(jsb->s_start);
+	journal->j_first = ntohl(jsb->s_first);
+	journal->j_last = ntohl(jsb->s_maxlen);
+
+	return 0;
+}
+
+static void e2fsck_journal_reset_super(e2fsck_t ctx, journal_superblock_t *jsb,
+				       journal_t *journal)
+{
+	char *p;
+	union {
+		uuid_t uuid;
+		__u32 val[4];
+	} u;
+	__u32 new_seq = 0;
+	int i;
+
+	/* Leave a valid existing V1 superblock signature alone.
+	 * Anything unrecognisable we overwrite with a new V2
+	 * signature. */
+	
+	if (jsb->s_header.h_magic != htonl(JFS_MAGIC_NUMBER) ||
+	    jsb->s_header.h_blocktype != htonl(JFS_SUPERBLOCK_V1)) {
+		jsb->s_header.h_magic = htonl(JFS_MAGIC_NUMBER);
+		jsb->s_header.h_blocktype = htonl(JFS_SUPERBLOCK_V2);
+	}
+
+	/* Zero out everything else beyond the superblock header */
+	
+	p = ((char *) jsb) + sizeof(journal_header_t);
+	memset (p, 0, ctx->fs->blocksize-sizeof(journal_header_t));
+
+	jsb->s_blocksize = htonl(ctx->fs->blocksize);
+	jsb->s_maxlen = htonl(journal->j_maxlen);
+	jsb->s_first = htonl(1);
+
+	/* Initialize the journal sequence number so that there is "no"
+	 * chance we will find old "valid" transactions in the journal.
+	 * This avoids the need to zero the whole journal (slow to do,
+	 * and risky when we are just recovering the filesystem).
+	 */
+	uuid_generate(u.uuid);
+	for (i = 0; i < 4; i ++)
+		new_seq ^= u.val[i];
+	jsb->s_sequence = htonl(new_seq);
+
+	mark_buffer_dirty(journal->j_sb_buffer);
+	ll_rw_block(WRITE, 1, &journal->j_sb_buffer);
+}
+
+static errcode_t e2fsck_journal_fix_corrupt_super(e2fsck_t ctx,
+						  journal_t *journal,
+						  struct problem_context *pctx)
+{
+	struct ext2_super_block *sb = ctx->fs->super;
+	int recover = ctx->fs->super->s_feature_incompat &
+		EXT3_FEATURE_INCOMPAT_RECOVER;
+
+	if (sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL) {
+		if (fix_problem(ctx, PR_0_JOURNAL_BAD_SUPER, pctx)) {
+			e2fsck_journal_reset_super(ctx, journal->j_superblock,
+						   journal);
+			journal->j_transaction_sequence = 1;
+			e2fsck_clear_recover(ctx, recover);
+			return 0;
+		}
+		return EXT2_ET_CORRUPT_SUPERBLOCK;
+	} else if (e2fsck_journal_fix_bad_inode(ctx, pctx))
+		return EXT2_ET_CORRUPT_SUPERBLOCK;
+
+	return 0;
+}
+
+static void e2fsck_journal_release(e2fsck_t ctx, journal_t *journal,
+				   int reset, int drop)
+{
+	journal_superblock_t *jsb;
+
+	if (drop)
+		mark_buffer_clean(journal->j_sb_buffer);
+	else if (!(ctx->options & E2F_OPT_READONLY)) {
+		jsb = journal->j_superblock;
+		jsb->s_sequence = htonl(journal->j_transaction_sequence);
+		if (reset)
+			jsb->s_start = 0; /* this marks the journal as empty */
+		mark_buffer_dirty(journal->j_sb_buffer);
+	}
+	brelse(journal->j_sb_buffer);
+
+	if (ctx->journal_io) {
+		if (ctx->fs && ctx->fs->io != ctx->journal_io)
+			io_channel_close(ctx->journal_io);
+		ctx->journal_io = 0;
+	}
+	
+#ifndef USE_INODE_IO
+	if (journal->j_inode)
+		ext2fs_free_mem(&journal->j_inode);
+#endif
+	if (journal->j_fs_dev)
+		ext2fs_free_mem(&journal->j_fs_dev);
+	ext2fs_free_mem(&journal);
+}
+
+/*
+ * This function makes sure that the superblock fields regarding the
+ * journal are consistent.
+ */
+int e2fsck_check_ext3_journal(e2fsck_t ctx)
+{
+	struct ext2_super_block *sb = ctx->fs->super;
+	journal_t *journal;
+	int recover = ctx->fs->super->s_feature_incompat &
+		EXT3_FEATURE_INCOMPAT_RECOVER;
+	struct problem_context pctx;
+	problem_t problem;
+	int reset = 0, force_fsck = 0;
+	int retval;
+
+	/* If we don't have any journal features, don't do anything more */
+	if (!(sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL) &&
+	    !recover && sb->s_journal_inum == 0 && sb->s_journal_dev == 0 &&
+	    uuid_is_null(sb->s_journal_uuid))
+ 		return 0;
+
+	clear_problem_context(&pctx);
+	pctx.num = sb->s_journal_inum;
+
+	retval = e2fsck_get_journal(ctx, &journal);
+	if (retval) {
+		if ((retval == EXT2_ET_BAD_INODE_NUM) ||
+		    (retval == EXT2_ET_BAD_BLOCK_NUM) ||
+		    (retval == EXT2_ET_JOURNAL_TOO_SMALL) ||
+		    (retval == EXT2_ET_NO_JOURNAL))
+			return e2fsck_journal_fix_bad_inode(ctx, &pctx);
+		return retval;
+	}
+
+	retval = e2fsck_journal_load(journal);
+	if (retval) {
+		if ((retval == EXT2_ET_CORRUPT_SUPERBLOCK) ||
+		    ((retval == EXT2_ET_UNSUPP_FEATURE) &&
+		    (!fix_problem(ctx, PR_0_JOURNAL_UNSUPP_INCOMPAT,
+				  &pctx))) ||
+		    ((retval == EXT2_ET_RO_UNSUPP_FEATURE) &&
+		    (!fix_problem(ctx, PR_0_JOURNAL_UNSUPP_ROCOMPAT,
+				  &pctx))) ||
+		    ((retval == EXT2_ET_JOURNAL_UNSUPP_VERSION) &&
+		    (!fix_problem(ctx, PR_0_JOURNAL_UNSUPP_VERSION, &pctx))))
+			retval = e2fsck_journal_fix_corrupt_super(ctx, journal,
+								  &pctx);
+		e2fsck_journal_release(ctx, journal, 0, 1);
+		return retval;
+	}
+
+	/*
+	 * We want to make the flags consistent here.  We will not leave with
+	 * needs_recovery set but has_journal clear.  We can't get in a loop
+	 * with -y, -n, or -p, only if a user isn't making up their mind.
+	 */
+no_has_journal:
+	if (!(sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL)) {
+		recover = sb->s_feature_incompat & EXT3_FEATURE_INCOMPAT_RECOVER;
+		pctx.str = "inode";
+		if (fix_problem(ctx, PR_0_JOURNAL_HAS_JOURNAL, &pctx)) {
+			if (recover &&
+			    !fix_problem(ctx, PR_0_JOURNAL_RECOVER_SET, &pctx))
+				goto no_has_journal;
+			/*
+			 * Need a full fsck if we are releasing a
+			 * journal stored on a reserved inode.
+			 */
+			force_fsck = recover ||
+				(sb->s_journal_inum < EXT2_FIRST_INODE(sb));
+			/* Clear all of the journal fields */
+			sb->s_journal_inum = 0;
+			sb->s_journal_dev = 0;
+			memset(sb->s_journal_uuid, 0,
+			       sizeof(sb->s_journal_uuid));
+			e2fsck_clear_recover(ctx, force_fsck);
+		} else if (!(ctx->options & E2F_OPT_READONLY)) {
+			sb->s_feature_compat |= EXT3_FEATURE_COMPAT_HAS_JOURNAL;
+			ext2fs_mark_super_dirty(ctx->fs);
+		}
+	}
+
+	if (sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL &&
+	    !(sb->s_feature_incompat & EXT3_FEATURE_INCOMPAT_RECOVER) &&
+	    journal->j_superblock->s_start != 0) {
+		/* Print status information */
+		fix_problem(ctx, PR_0_JOURNAL_RECOVERY_CLEAR, &pctx);
+		if (ctx->superblock)
+			problem = PR_0_JOURNAL_RUN_DEFAULT;
+		else
+			problem = PR_0_JOURNAL_RUN;
+		if (fix_problem(ctx, problem, &pctx)) {
+			ctx->options |= E2F_OPT_FORCE;
+			sb->s_feature_incompat |=
+				EXT3_FEATURE_INCOMPAT_RECOVER;
+			ext2fs_mark_super_dirty(ctx->fs);
+		} else if (fix_problem(ctx,
+				       PR_0_JOURNAL_RESET_JOURNAL, &pctx)) {
+			reset = 1;
+			sb->s_state &= ~EXT2_VALID_FS;
+			ext2fs_mark_super_dirty(ctx->fs);
+		}
+		/*
+		 * If the user answers no to the above question, we
+		 * ignore the fact that journal apparently has data;
+		 * accidentally replaying over valid data would be far
+		 * worse than skipping a questionable recovery.
+		 * 
+		 * XXX should we abort with a fatal error here?  What
+		 * will the ext3 kernel code do if a filesystem with
+		 * !NEEDS_RECOVERY but with a non-zero
+		 * journal->j_superblock->s_start is mounted?
+		 */
+	}
+
+	e2fsck_journal_release(ctx, journal, reset, 0);
+	return retval;
+}
+
+static errcode_t recover_ext3_journal(e2fsck_t ctx)
+{
+	journal_t *journal;
+	int retval;
+
+	journal_init_revoke_caches();
+	retval = e2fsck_get_journal(ctx, &journal);
+	if (retval)
+		return retval;
+
+	retval = e2fsck_journal_load(journal);
+	if (retval)
+		goto errout;
+
+	retval = journal_init_revoke(journal, 1024);
+	if (retval)
+		goto errout;
+	
+	retval = -journal_recover(journal);
+	if (retval)
+		goto errout;
+	
+	if (journal->j_superblock->s_errno) {
+		ctx->fs->super->s_state |= EXT2_ERROR_FS;
+		ext2fs_mark_super_dirty(ctx->fs);
+		journal->j_superblock->s_errno = 0;
+		mark_buffer_dirty(journal->j_sb_buffer);
+	}
+		
+errout:
+	journal_destroy_revoke(journal);
+	journal_destroy_revoke_caches();
+	e2fsck_journal_release(ctx, journal, 1, 0);
+	return retval;
+}
+
+int e2fsck_run_ext3_journal(e2fsck_t ctx)
+{
+	io_manager io_ptr = ctx->fs->io->manager;
+	int blocksize = ctx->fs->blocksize;
+	errcode_t	retval, recover_retval;
+
+	printf(_("%s: recovering journal\n"), ctx->device_name);
+	if (ctx->options & E2F_OPT_READONLY) {
+		printf(_("%s: won't do journal recovery while read-only\n"),
+		       ctx->device_name);
+		return EXT2_ET_FILE_RO;
+	}
+
+	if (ctx->fs->flags & EXT2_FLAG_DIRTY)
+		ext2fs_flush(ctx->fs);	/* Force out any modifications */
+
+	recover_retval = recover_ext3_journal(ctx);
+	
+	/*
+	 * Reload the filesystem context to get up-to-date data from disk
+	 * because journal recovery will change the filesystem under us.
+	 */
+	ext2fs_close(ctx->fs);
+	retval = ext2fs_open(ctx->filesystem_name, EXT2_FLAG_RW,
+			     ctx->superblock, blocksize, io_ptr,
+			     &ctx->fs);
+
+	if (retval) {
+		com_err(ctx->program_name, retval,
+			_("while trying to re-open %s"),
+			ctx->device_name);
+		fatal_error(ctx, 0);
+	}
+	ctx->fs->priv_data = ctx;
+
+	/* Set the superblock flags */
+	e2fsck_clear_recover(ctx, recover_retval);
+	return recover_retval;
+}
+
+/*
+ * This function will move the journal inode from a visible file in
+ * the filesystem directory hierarchy to the reserved inode if necessary.
+ */
+static const char * const journal_names[] = {
+	".journal", "journal", ".journal.dat", "journal.dat", 0 };
+
+void e2fsck_move_ext3_journal(e2fsck_t ctx)
+{
+	struct ext2_super_block *sb = ctx->fs->super;
+	struct problem_context	pctx;
+	struct ext2_inode 	inode;
+	ext2_filsys		fs = ctx->fs;
+	ext2_ino_t		ino;
+	errcode_t		retval;
+	const char * const *	cpp;
+	int			group, mount_flags;
+	
+	clear_problem_context(&pctx);
+
+	/*
+	 * If the filesystem is opened read-only, or there is no
+	 * journal, then do nothing.
+	 */
+	if ((ctx->options & E2F_OPT_READONLY) ||
+	    (sb->s_journal_inum == 0) ||
+	    !(sb->s_feature_compat & EXT3_FEATURE_COMPAT_HAS_JOURNAL))
+		return;
+
+	/*
+	 * Read in the journal inode
+	 */
+	if (ext2fs_read_inode(fs, sb->s_journal_inum, &inode) != 0)
+		return;
+
+	/*
+	 * If it's necessary to backup the journal inode, do so.
+	 */
+	if ((sb->s_jnl_backup_type == 0) ||
+	    ((sb->s_jnl_backup_type == EXT3_JNL_BACKUP_BLOCKS) &&
+	     memcmp(inode.i_block, sb->s_jnl_blocks, EXT2_N_BLOCKS*4))) {
+		if (fix_problem(ctx, PR_0_BACKUP_JNL, &pctx)) {
+			memcpy(sb->s_jnl_blocks, inode.i_block,
+			       EXT2_N_BLOCKS*4);
+			sb->s_jnl_blocks[16] = inode.i_size;
+			sb->s_jnl_backup_type = EXT3_JNL_BACKUP_BLOCKS;
+			ext2fs_mark_super_dirty(fs);
+			fs->flags &= ~EXT2_FLAG_MASTER_SB_ONLY;
+		}
+	}
+
+	/*
+	 * If the journal is already the hidden inode, then do nothing
+	 */
+	if (sb->s_journal_inum == EXT2_JOURNAL_INO)
+		return;
+	
+	/*
+	 * The journal inode had better have only one link and not be readable.
+	 */
+	if (inode.i_links_count != 1)
+		return;
+
+	/*
+	 * If the filesystem is mounted, or we can't tell whether
+	 * or not it's mounted, do nothing.
+	 */
+	retval = ext2fs_check_if_mounted(ctx->filesystem_name, &mount_flags);
+	if (retval || (mount_flags & EXT2_MF_MOUNTED))
+		return;
+
+	/*
+	 * If we can't find the name of the journal inode, then do
+	 * nothing.
+	 */
+	for (cpp = journal_names; *cpp; cpp++) {
+		retval = ext2fs_lookup(fs, EXT2_ROOT_INO, *cpp,
+				       strlen(*cpp), 0, &ino);
+		if ((retval == 0) && (ino == sb->s_journal_inum))
+			break;
+	}
+	if (*cpp == 0)
+		return;
+
+	/* We need the inode bitmap to be loaded */
+	retval = ext2fs_read_bitmaps(fs);
+	if (retval)
+		return;
+
+	pctx.str = *cpp;
+	if (!fix_problem(ctx, PR_0_MOVE_JOURNAL, &pctx))
+		return;
+		
+	/*
+	 * OK, we've done all the checks, let's actually move the
+	 * journal inode.  Errors at this point mean we need to force
+	 * an ext2 filesystem check.
+	 */
+	if ((retval = ext2fs_unlink(fs, EXT2_ROOT_INO, *cpp, ino, 0)) != 0)
+		goto err_out;
+	if ((retval = ext2fs_write_inode(fs, EXT2_JOURNAL_INO, &inode)) != 0)
+		goto err_out;
+	sb->s_journal_inum = EXT2_JOURNAL_INO;
+	ext2fs_mark_super_dirty(fs);
+	fs->flags &= ~EXT2_FLAG_MASTER_SB_ONLY;
+	inode.i_links_count = 0;
+	inode.i_dtime = time(0);
+	if ((retval = ext2fs_write_inode(fs, ino, &inode)) != 0)
+		goto err_out;
+
+	group = ext2fs_group_of_ino(fs, ino);
+	ext2fs_unmark_inode_bitmap(fs->inode_map, ino);
+	ext2fs_mark_ib_dirty(fs);
+	fs->group_desc[group].bg_free_inodes_count++;
+	fs->super->s_free_inodes_count++;
+	return;
+
+err_out:
+	pctx.errcode = retval;
+	fix_problem(ctx, PR_0_ERR_MOVE_JOURNAL, &pctx);
+	fs->super->s_state &= ~EXT2_VALID_FS;
+	ext2fs_mark_super_dirty(fs);
+	return;
+}
+

e2fsprogs/e2fsck/message.c

@@ -0,0 +1,466 @@
+/*
+ * message.c --- print e2fsck messages (with compression)
+ *
+ * Copyright 1996, 1997 by Theodore Ts'o
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ *
+ * print_e2fsck_message() prints a message to the user, using
+ * compression techniques and expansions of abbreviations.
+ *
+ * The following % expansions are supported:
+ *
+ * 	%b	<blk>			block number
+ * 	%B	<blkcount>		integer
+ * 	%c	<blk2>			block number
+ * 	%Di	<dirent>->ino		inode number
+ * 	%Dn	<dirent>->name		string
+ * 	%Dr	<dirent>->rec_len
+ * 	%Dl	<dirent>->name_len
+ * 	%Dt	<dirent>->filetype
+ * 	%d	<dir> 			inode number
+ * 	%g	<group>			integer
+ * 	%i	<ino>			inode number
+ * 	%Is	<inode> -> i_size
+ * 	%IS	<inode> -> i_extra_isize
+ * 	%Ib	<inode> -> i_blocks
+ * 	%Il	<inode> -> i_links_count
+ * 	%Im	<inode> -> i_mode
+ * 	%IM	<inode> -> i_mtime
+ * 	%IF	<inode> -> i_faddr
+ * 	%If	<inode> -> i_file_acl
+ * 	%Id	<inode> -> i_dir_acl
+ * 	%Iu	<inode> -> i_uid
+ * 	%Ig	<inode> -> i_gid
+ * 	%j	<ino2>			inode number
+ * 	%m	<com_err error message>
+ * 	%N	<num>
+ *	%p	ext2fs_get_pathname of directory <ino>
+ * 	%P	ext2fs_get_pathname of <dirent>->ino with <ino2> as
+ * 			the containing directory.  (If dirent is NULL
+ * 			then return the pathname of directory <ino2>)
+ * 	%q	ext2fs_get_pathname of directory <dir>
+ * 	%Q	ext2fs_get_pathname of directory <ino> with <dir> as
+ * 			the containing directory.
+ * 	%s	<str>			miscellaneous string
+ * 	%S	backup superblock
+ * 	%X	<num> hexadecimal format
+ *
+ * The following '@' expansions are supported:
+ *
+ * 	@a	extended attribute
+ * 	@A	error allocating
+ * 	@b	block
+ * 	@B	bitmap
+ * 	@c	compress
+ * 	@C	conflicts with some other fs block
+ * 	@D	deleted
+ * 	@d	directory
+ * 	@e	entry
+ * 	@E	Entry '%Dn' in %p (%i)
+ * 	@f	filesystem
+ * 	@F	for @i %i (%Q) is
+ * 	@g	group
+ * 	@h	HTREE directory inode
+ * 	@i	inode
+ * 	@I	illegal
+ * 	@j	journal
+ * 	@l	lost+found
+ * 	@L	is a link
+ * 	@o	orphaned
+ * 	@p	problem in
+ * 	@r	root inode
+ * 	@s	should be 
+ * 	@S	superblock
+ * 	@u	unattached
+ * 	@v	device
+ * 	@z	zero-length
+ */
+
+#include <stdlib.h>
+#include <unistd.h>
+#include <string.h>
+#include <ctype.h>
+#include <termios.h>
+
+#include "e2fsck.h"
+
+#include "problem.h"
+
+#ifdef __GNUC__
+#define _INLINE_ __inline__
+#else
+#define _INLINE_
+#endif
+
+/*
+ * This structure defines the abbreviations used by the text strings
+ * below.  The first character in the string is the index letter.  An
+ * abbreviation of the form '@<i>' is expanded by looking up the index
+ * letter <i> in the table below.
+ */
+static const char *abbrevs[] = {
+	N_("aextended attribute"),
+	N_("Aerror allocating"),
+	N_("bblock"),
+	N_("Bbitmap"),
+	N_("ccompress"),
+	N_("Cconflicts with some other fs @b"),
+	N_("iinode"),
+	N_("Iillegal"),
+	N_("jjournal"),
+	N_("Ddeleted"),
+	N_("ddirectory"),
+	N_("eentry"),
+	N_("E@e '%Dn' in %p (%i)"),
+	N_("ffilesystem"),
+	N_("Ffor @i %i (%Q) is"),
+	N_("ggroup"),
+	N_("hHTREE @d @i"),
+	N_("llost+found"),
+	N_("Lis a link"),
+	N_("oorphaned"),
+	N_("pproblem in"),
+	N_("rroot @i"),
+	N_("sshould be"),
+	N_("Ssuper@b"),
+	N_("uunattached"),
+	N_("vdevice"),
+	N_("zzero-length"),
+	"@@",
+	0
+	};
+
+/*
+ * Give more user friendly names to the "special" inodes.
+ */
+#define num_special_inodes	11
+static const char *special_inode_name[] =
+{
+	N_("<The NULL inode>"),			/* 0 */
+	N_("<The bad blocks inode>"),		/* 1 */
+	"/",					/* 2 */
+	N_("<The ACL index inode>"),		/* 3 */
+	N_("<The ACL data inode>"),		/* 4 */
+	N_("<The boot loader inode>"),		/* 5 */
+	N_("<The undelete directory inode>"),	/* 6 */
+	N_("<The group descriptor inode>"),	/* 7 */
+	N_("<The journal inode>"),		/* 8 */
+	N_("<Reserved inode 9>"),		/* 9 */
+	N_("<Reserved inode 10>"),		/* 10 */
+};
+
+/*
+ * This function does "safe" printing.  It will convert non-printable
+ * ASCII characters using '^' and M- notation.
+ */
+static void safe_print(const char *cp, int len)
+{
+	unsigned char	ch;
+
+	if (len < 0)
+		len = strlen(cp);
+
+	while (len--) {
+		ch = *cp++;
+		if (ch > 128) {
+			fputs("M-", stdout);
+			ch -= 128;
+		}
+		if ((ch < 32) || (ch == 0x7f)) {
+			fputc('^', stdout);
+			ch ^= 0x40; /* ^@, ^A, ^B; ^? for DEL */
+		}
+		fputc(ch, stdout);
+	}
+}
+
+
+/*
+ * This function prints a pathname, using the ext2fs_get_pathname
+ * function
+ */
+static void print_pathname(ext2_filsys fs, ext2_ino_t dir, ext2_ino_t ino)
+{
+	errcode_t	retval;
+	char		*path;
+
+	if (!dir && (ino < num_special_inodes)) {
+		fputs(_(special_inode_name[ino]), stdout);
+		return;
+	}
+	
+	retval = ext2fs_get_pathname(fs, dir, ino, &path);
+	if (retval)
+		fputs("???", stdout);
+	else {
+		safe_print(path, -1);
+		ext2fs_free_mem(&path);
+	}
+}
+
+/*
+ * This function handles the '@' expansion.  We allow recursive
+ * expansion; an @ expression can contain further '@' and '%'
+ * expressions. 
+ */
+static _INLINE_ void expand_at_expression(e2fsck_t ctx, char ch,
+					  struct problem_context *pctx,
+					  int *first)
+{
+	const char **cpp, *str;
+	
+	/* Search for the abbreviation */
+	for (cpp = abbrevs; *cpp; cpp++) {
+		if (ch == *cpp[0])
+			break;
+	}
+	if (*cpp) {
+		str = _(*cpp) + 1;
+		if (*first && islower(*str)) {
+			*first = 0;
+			fputc(toupper(*str++), stdout);
+		}
+		print_e2fsck_message(ctx, str, pctx, *first);
+	} else
+		printf("@%c", ch);
+}
+
+/*
+ * This function expands '%IX' expressions
+ */
+static _INLINE_ void expand_inode_expression(char ch, 
+					     struct problem_context *ctx)
+{
+	struct ext2_inode	*inode;
+	struct ext2_inode_large	*large_inode;
+	char *			time_str;
+	time_t			t;
+	int			do_gmt = -1;
+
+	if (!ctx || !ctx->inode)
+		goto no_inode;
+	
+	inode = ctx->inode;
+	large_inode = (struct ext2_inode_large *) inode;
+
+	switch (ch) {
+	case 's':
+		if (LINUX_S_ISDIR(inode->i_mode))
+			printf("%u", inode->i_size);
+		else {
+#ifdef EXT2_NO_64_TYPE
+			if (inode->i_size_high)
+				printf("0x%x%08x", inode->i_size_high,
+				       inode->i_size);
+			else
+				printf("%u", inode->i_size);
+#else
+			printf("%llu", (inode->i_size | 
+					((__u64) inode->i_size_high << 32)));
+#endif
+		}
+		break;
+	case 'S':
+		printf("%u", large_inode->i_extra_isize);
+		break;
+	case 'b':
+		printf("%u", inode->i_blocks);
+		break;
+	case 'l':
+		printf("%d", inode->i_links_count);
+		break;
+	case 'm':
+		printf("0%o", inode->i_mode);
+		break;
+	case 'M':
+		/* The diet libc doesn't respect the TZ environemnt variable */
+		if (do_gmt == -1) {
+			time_str = getenv("TZ");
+			if (!time_str)
+				time_str = "";
+			do_gmt = !strcmp(time_str, "GMT");
+		}
+		t = inode->i_mtime;
+		time_str = asctime(do_gmt ? gmtime(&t) : localtime(&t));
+		printf("%.24s", time_str);
+		break;
+	case 'F':
+		printf("%u", inode->i_faddr);
+		break;
+	case 'f':
+		printf("%u", inode->i_file_acl);
+		break;
+	case 'd':
+		printf("%u", (LINUX_S_ISDIR(inode->i_mode) ?
+			      inode->i_dir_acl : 0));
+		break;
+	case 'u':
+		printf("%d", (inode->i_uid |
+			      (inode->osd2.linux2.l_i_uid_high << 16)));
+		break;
+	case 'g':
+		printf("%d", (inode->i_gid |
+			      (inode->osd2.linux2.l_i_gid_high << 16)));
+		break;
+	default:
+	no_inode:
+		printf("%%I%c", ch);
+		break;
+	}
+}
+
+/*
+ * This function expands '%dX' expressions
+ */
+static _INLINE_ void expand_dirent_expression(char ch,
+					      struct problem_context *ctx)
+{
+	struct ext2_dir_entry	*dirent;
+	int	len;
+	
+	if (!ctx || !ctx->dirent)
+		goto no_dirent;
+	
+	dirent = ctx->dirent;
+	
+	switch (ch) {
+	case 'i':
+		printf("%u", dirent->inode);
+		break;
+	case 'n':
+		len = dirent->name_len & 0xFF;
+		if (len > EXT2_NAME_LEN)
+			len = EXT2_NAME_LEN;
+		if (len > dirent->rec_len)
+			len = dirent->rec_len;
+		safe_print(dirent->name, len);
+		break;
+	case 'r':
+		printf("%u", dirent->rec_len);
+		break;
+	case 'l':
+		printf("%u", dirent->name_len & 0xFF);
+		break;
+	case 't':
+		printf("%u", dirent->name_len >> 8);
+		break;
+	default:
+	no_dirent:
+		printf("%%D%c", ch);
+		break;
+	}
+}
+
+static _INLINE_ void expand_percent_expression(ext2_filsys fs, char ch,
+					       struct problem_context *ctx)
+{
+	if (!ctx)
+		goto no_context;
+	
+	switch (ch) {
+	case '%':
+		fputc('%', stdout);
+		break;
+	case 'b':
+		printf("%u", ctx->blk);
+		break;
+	case 'B':
+#ifdef EXT2_NO_64_TYPE
+		printf("%d", ctx->blkcount);
+#else
+		printf("%lld", ctx->blkcount);
+#endif
+		break;
+	case 'c':
+		printf("%u", ctx->blk2);
+		break;
+	case 'd':
+		printf("%u", ctx->dir);
+		break;
+	case 'g':
+		printf("%d", ctx->group);
+		break;
+	case 'i':
+		printf("%u", ctx->ino);
+		break;
+	case 'j':
+		printf("%u", ctx->ino2);
+		break;
+	case 'm':
+		printf("%s", error_message(ctx->errcode));
+		break;
+	case 'N':
+#ifdef EXT2_NO_64_TYPE
+		printf("%u", ctx->num);
+#else
+		printf("%llu", ctx->num);
+#endif
+		break;
+	case 'p':
+		print_pathname(fs, ctx->ino, 0);
+		break;
+	case 'P':
+		print_pathname(fs, ctx->ino2,
+			       ctx->dirent ? ctx->dirent->inode : 0);
+		break;
+	case 'q':
+		print_pathname(fs, ctx->dir, 0);
+		break;
+	case 'Q':
+		print_pathname(fs, ctx->dir, ctx->ino);
+		break;
+	case 'S':
+		printf("%d", get_backup_sb(NULL, fs, NULL, NULL));
+		break;
+	case 's':
+		printf("%s", ctx->str ? ctx->str : "NULL");
+		break;
+	case 'X':
+#ifdef EXT2_NO_64_TYPE
+		printf("0x%x", ctx->num);
+#else
+		printf("0x%llx", ctx->num);
+#endif
+		break;
+	default:
+	no_context:
+		printf("%%%c", ch);
+		break;
+	}
+}	
+
+void print_e2fsck_message(e2fsck_t ctx, const char *msg,
+			  struct problem_context *pctx, int first)
+{
+	ext2_filsys fs = ctx->fs;
+	const char *	cp;
+	int		i;
+
+	e2fsck_clear_progbar(ctx);
+	for (cp = msg; *cp; cp++) {
+		if (cp[0] == '@') {
+			cp++;
+			expand_at_expression(ctx, *cp, pctx, &first);
+		} else if (cp[0] == '%' && cp[1] == 'I') {
+			cp += 2;
+			expand_inode_expression(*cp, pctx);
+		} else if (cp[0] == '%' && cp[1] == 'D') {
+			cp += 2;
+			expand_dirent_expression(*cp, pctx);
+		} else if ((cp[0] == '%')) {
+			cp++;
+			expand_percent_expression(fs, *cp, pctx);
+		} else {
+			for (i=0; cp[i]; i++)
+				if ((cp[i] == '@') || cp[i] == '%')
+					break;
+			printf("%.*s", i, cp);
+			cp += i-1;
+		}
+		first = 0;
+	}
+}

e2fsprogs/e2fsck/pass1b.c

@@ -0,0 +1,805 @@
+/*
+ * pass1b.c --- Pass #1b of e2fsck
+ *
+ * This file contains pass1B, pass1C, and pass1D of e2fsck.  They are
+ * only invoked if pass 1 discovered blocks which are in use by more
+ * than one inode.
+ * 
+ * Pass1B scans the data blocks of all the inodes again, generating a
+ * complete list of duplicate blocks and which inodes have claimed
+ * them.
+ *
+ * Pass1C does a tree-traversal of the filesystem, to determine the
+ * parent directories of these inodes.  This step is necessary so that
+ * e2fsck can print out the pathnames of affected inodes.
+ *
+ * Pass1D is a reconciliation pass.  For each inode with duplicate
+ * blocks, the user is prompted if s/he would like to clone the file
+ * (so that the file gets a fresh copy of the duplicated blocks) or
+ * simply to delete the file.
+ * 
+ * Copyright (C) 1993, 1994, 1995, 1996, 1997 Theodore Ts'o.
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ * 
+ */
+
+#include <time.h>
+#ifdef HAVE_ERRNO_H
+#include <errno.h>
+#endif
+
+#ifdef HAVE_INTTYPES_H
+#include <inttypes.h>
+#endif
+
+/* Needed for architectures where sizeof(int) != sizeof(void *) */
+#define INT_TO_VOIDPTR(val)  ((void *)(intptr_t)(val))
+#define VOIDPTR_TO_INT(ptr)  ((int)(intptr_t)(ptr))
+
+#include "e2fsck.h"
+
+#include "problem.h"
+#include "dict.h"
+
+/* Define an extension to the ext2 library's block count information */
+#define BLOCK_COUNT_EXTATTR	(-5)
+
+struct block_el {
+	blk_t	block;
+	struct block_el *next;
+};
+
+struct inode_el {
+	ext2_ino_t	inode;
+	struct inode_el *next;
+};
+
+struct dup_block {
+	int		num_bad;
+	struct inode_el *inode_list;
+};
+
+/*
+ * This structure stores information about a particular inode which
+ * is sharing blocks with other inodes.  This information is collected
+ * to display to the user, so that the user knows what files he or she
+ * is dealing with, when trying to decide how to resolve the conflict
+ * of multiply-claimed blocks.
+ */
+struct dup_inode {
+	ext2_ino_t		dir;
+	int			num_dupblocks;
+	struct ext2_inode	inode;
+	struct block_el		*block_list;
+};
+
+static int process_pass1b_block(ext2_filsys fs, blk_t	*blocknr,
+				e2_blkcnt_t blockcnt, blk_t ref_blk, 
+				int ref_offset, void *priv_data);
+static void delete_file(e2fsck_t ctx, ext2_ino_t ino,
+			struct dup_inode *dp, char *block_buf);
+static int clone_file(e2fsck_t ctx, ext2_ino_t ino,
+		      struct dup_inode *dp, char* block_buf);
+static int check_if_fs_block(e2fsck_t ctx, blk_t test_blk);
+
+static void pass1b(e2fsck_t ctx, char *block_buf);
+static void pass1c(e2fsck_t ctx, char *block_buf);
+static void pass1d(e2fsck_t ctx, char *block_buf);
+
+static int dup_inode_count = 0;
+
+static dict_t blk_dict, ino_dict;
+
+static ext2fs_inode_bitmap inode_dup_map;
+
+static int dict_int_cmp(const void *a, const void *b)
+{
+	intptr_t	ia, ib;
+
+	ia = (intptr_t)a;
+	ib = (intptr_t)b;
+
+	return (ia-ib);
+}
+
+/*
+ * Add a duplicate block record
+ */
+static void add_dupe(e2fsck_t ctx, ext2_ino_t ino, blk_t blk,
+		     struct ext2_inode *inode)
+{
+	dnode_t	*n;
+	struct dup_block	*db;
+	struct dup_inode	*di;
+	struct block_el		*blk_el;
+	struct inode_el 	*ino_el;
+
+	n = dict_lookup(&blk_dict, INT_TO_VOIDPTR(blk));
+	if (n)
+		db = (struct dup_block *) dnode_get(n);
+	else {
+		db = (struct dup_block *) e2fsck_allocate_memory(ctx,
+			 sizeof(struct dup_block), "duplicate block header");
+		db->num_bad = 0;
+		db->inode_list = 0;
+		dict_alloc_insert(&blk_dict, INT_TO_VOIDPTR(blk), db);
+	}
+	ino_el = (struct inode_el *) e2fsck_allocate_memory(ctx,
+			 sizeof(struct inode_el), "inode element");
+	ino_el->inode = ino;
+	ino_el->next = db->inode_list;
+	db->inode_list = ino_el;
+	db->num_bad++;
+
+	n = dict_lookup(&ino_dict, INT_TO_VOIDPTR(ino));
+	if (n)
+		di = (struct dup_inode *) dnode_get(n);
+	else {
+		di = (struct dup_inode *) e2fsck_allocate_memory(ctx,
+			 sizeof(struct dup_inode), "duplicate inode header");
+		di->dir = (ino == EXT2_ROOT_INO) ? EXT2_ROOT_INO : 0 ;
+		di->num_dupblocks = 0;
+		di->block_list = 0;
+		di->inode = *inode;
+		dict_alloc_insert(&ino_dict, INT_TO_VOIDPTR(ino), di);
+	}
+	blk_el = (struct block_el *) e2fsck_allocate_memory(ctx,
+			 sizeof(struct block_el), "block element");
+	blk_el->block = blk;
+	blk_el->next = di->block_list;
+	di->block_list = blk_el;
+	di->num_dupblocks++;
+}
+
+/*
+ * Free a duplicate inode record
+ */
+static void inode_dnode_free(dnode_t *node, 
+			     void *context EXT2FS_ATTR((unused)))
+{
+	struct dup_inode	*di;
+	struct block_el		*p, *next;
+
+	di = (struct dup_inode *) dnode_get(node);
+	for (p = di->block_list; p; p = next) {
+		next = p->next;
+		free(p);
+	}
+	free(node);
+}
+
+/*
+ * Free a duplicate block record
+ */
+static void block_dnode_free(dnode_t *node, 
+			     void *context EXT2FS_ATTR((unused)))
+{
+	struct dup_block	*db;
+	struct inode_el		*p, *next;
+
+	db = (struct dup_block *) dnode_get(node);
+	for (p = db->inode_list; p; p = next) {
+		next = p->next;
+		free(p);
+	}
+	free(node);
+}
+
+
+/*
+ * Main procedure for handling duplicate blocks
+ */
+void e2fsck_pass1_dupblocks(e2fsck_t ctx, char *block_buf)
+{
+	ext2_filsys 		fs = ctx->fs;
+	struct problem_context	pctx;
+
+	clear_problem_context(&pctx);
+	
+	pctx.errcode = ext2fs_allocate_inode_bitmap(fs,
+		      _("multiply claimed inode map"), &inode_dup_map);
+	if (pctx.errcode) {
+		fix_problem(ctx, PR_1B_ALLOCATE_IBITMAP_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+
+	dict_init(&ino_dict, DICTCOUNT_T_MAX, dict_int_cmp);
+	dict_init(&blk_dict, DICTCOUNT_T_MAX, dict_int_cmp);
+	dict_set_allocator(&ino_dict, NULL, inode_dnode_free, NULL);
+	dict_set_allocator(&blk_dict, NULL, block_dnode_free, NULL);
+	
+	pass1b(ctx, block_buf);
+	pass1c(ctx, block_buf);
+	pass1d(ctx, block_buf);
+
+	/*
+	 * Time to free all of the accumulated data structures that we
+	 * don't need anymore.
+	 */
+	dict_free_nodes(&ino_dict);
+	dict_free_nodes(&blk_dict);
+}
+
+/*
+ * Scan the inodes looking for inodes that contain duplicate blocks.
+ */
+struct process_block_struct {
+	e2fsck_t	ctx;
+	ext2_ino_t	ino;
+	int		dup_blocks;
+	struct ext2_inode *inode;
+	struct problem_context *pctx;
+};
+
+static void pass1b(e2fsck_t ctx, char *block_buf)
+{
+	ext2_filsys fs = ctx->fs;
+	ext2_ino_t ino;
+	struct ext2_inode inode;
+	ext2_inode_scan	scan;
+	struct process_block_struct pb;
+	struct problem_context pctx;
+	
+	clear_problem_context(&pctx);
+	
+	if (!(ctx->options & E2F_OPT_PREEN))
+		fix_problem(ctx, PR_1B_PASS_HEADER, &pctx);
+	pctx.errcode = ext2fs_open_inode_scan(fs, ctx->inode_buffer_blocks,
+					      &scan);
+	if (pctx.errcode) {
+		fix_problem(ctx, PR_1B_ISCAN_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+	ctx->stashed_inode = &inode;
+	pb.ctx = ctx;
+	pb.pctx = &pctx;
+	pctx.str = "pass1b";
+	while (1) {
+		pctx.errcode = ext2fs_get_next_inode(scan, &ino, &inode);
+		if (pctx.errcode == EXT2_ET_BAD_BLOCK_IN_INODE_TABLE)
+			continue;
+		if (pctx.errcode) {
+			fix_problem(ctx, PR_1B_ISCAN_ERROR, &pctx);
+			ctx->flags |= E2F_FLAG_ABORT;
+			return;
+		}
+		if (!ino)
+			break;
+		pctx.ino = ctx->stashed_ino = ino;
+		if ((ino != EXT2_BAD_INO) &&
+		    !ext2fs_test_inode_bitmap(ctx->inode_used_map, ino))
+			continue;
+
+		pb.ino = ino;
+		pb.dup_blocks = 0;
+		pb.inode = &inode;
+
+		if (ext2fs_inode_has_valid_blocks(&inode) ||
+		    (ino == EXT2_BAD_INO))
+			pctx.errcode = ext2fs_block_iterate2(fs, ino,
+				     0, block_buf, process_pass1b_block, &pb);
+		if (inode.i_file_acl)
+			process_pass1b_block(fs, &inode.i_file_acl,
+					     BLOCK_COUNT_EXTATTR, 0, 0, &pb);
+		if (pb.dup_blocks) {
+			end_problem_latch(ctx, PR_LATCH_DBLOCK);
+			if (ino >= EXT2_FIRST_INODE(fs->super) ||
+			    ino == EXT2_ROOT_INO)
+				dup_inode_count++;
+		}
+		if (pctx.errcode)
+			fix_problem(ctx, PR_1B_BLOCK_ITERATE, &pctx);
+	}
+	ext2fs_close_inode_scan(scan);
+	e2fsck_use_inode_shortcuts(ctx, 0);
+}
+
+static int process_pass1b_block(ext2_filsys fs EXT2FS_ATTR((unused)),
+				blk_t	*block_nr,
+				e2_blkcnt_t blockcnt EXT2FS_ATTR((unused)),
+				blk_t ref_blk EXT2FS_ATTR((unused)), 
+				int ref_offset EXT2FS_ATTR((unused)),
+				void *priv_data)
+{
+	struct process_block_struct *p;
+	e2fsck_t ctx;
+
+	if (HOLE_BLKADDR(*block_nr))
+		return 0;
+	p = (struct process_block_struct *) priv_data;
+	ctx = p->ctx;
+	
+	if (!ext2fs_test_block_bitmap(ctx->block_dup_map, *block_nr))
+		return 0;
+	
+	/* OK, this is a duplicate block */
+	if (p->ino != EXT2_BAD_INO) {
+		p->pctx->blk = *block_nr;
+		fix_problem(ctx, PR_1B_DUP_BLOCK, p->pctx);
+	}
+	p->dup_blocks++;
+	ext2fs_mark_inode_bitmap(inode_dup_map, p->ino);
+
+	add_dupe(ctx, p->ino, *block_nr, p->inode);
+	
+	return 0;
+}
+
+/*
+ * Pass 1c: Scan directories for inodes with duplicate blocks.  This
+ * is used so that we can print pathnames when prompting the user for
+ * what to do.
+ */
+struct search_dir_struct {
+	int		count;
+	ext2_ino_t	first_inode;
+	ext2_ino_t	max_inode;
+};
+
+static int search_dirent_proc(ext2_ino_t dir, int entry,
+			      struct ext2_dir_entry *dirent,
+			      int offset EXT2FS_ATTR((unused)), 
+			      int blocksize EXT2FS_ATTR((unused)),
+			      char *buf EXT2FS_ATTR((unused)), 
+			      void *priv_data)
+{
+	struct search_dir_struct *sd;
+	struct dup_inode	*p;
+	dnode_t			*n;
+
+	sd = (struct search_dir_struct *) priv_data;
+
+	if (dirent->inode > sd->max_inode)
+		/* Should abort this inode, but not everything */
+		return 0;	
+
+	if ((dirent->inode < sd->first_inode) || (entry < DIRENT_OTHER_FILE) ||
+	    !ext2fs_test_inode_bitmap(inode_dup_map, dirent->inode))
+		return 0;
+
+	n = dict_lookup(&ino_dict, INT_TO_VOIDPTR(dirent->inode));
+	if (!n)
+		return 0;
+	p = (struct dup_inode *) dnode_get(n);
+	p->dir = dir;
+	sd->count--;
+
+	return(sd->count ? 0 : DIRENT_ABORT);
+}
+
+
+static void pass1c(e2fsck_t ctx, char *block_buf)
+{
+	ext2_filsys fs = ctx->fs;
+	struct search_dir_struct sd;
+	struct problem_context pctx;
+
+	clear_problem_context(&pctx);
+
+	if (!(ctx->options & E2F_OPT_PREEN))
+		fix_problem(ctx, PR_1C_PASS_HEADER, &pctx);
+
+	/*
+	 * Search through all directories to translate inodes to names
+	 * (by searching for the containing directory for that inode.)
+	 */
+	sd.count = dup_inode_count;
+	sd.first_inode = EXT2_FIRST_INODE(fs->super);
+	sd.max_inode = fs->super->s_inodes_count;
+	ext2fs_dblist_dir_iterate(fs->dblist, 0, block_buf,
+				  search_dirent_proc, &sd);
+}	
+
+static void pass1d(e2fsck_t ctx, char *block_buf)
+{
+	ext2_filsys fs = ctx->fs;
+	struct dup_inode	*p, *t;
+	struct dup_block	*q;
+	ext2_ino_t		*shared, ino;
+	int	shared_len;
+	int	i;
+	int	file_ok;
+	int	meta_data = 0;
+	struct problem_context pctx;
+	dnode_t	*n, *m;
+	struct block_el	*s;
+	struct inode_el *r;
+	
+	clear_problem_context(&pctx);
+	
+	if (!(ctx->options & E2F_OPT_PREEN))
+		fix_problem(ctx, PR_1D_PASS_HEADER, &pctx);
+	e2fsck_read_bitmaps(ctx);
+
+	pctx.num = dup_inode_count; /* dict_count(&ino_dict); */
+	fix_problem(ctx, PR_1D_NUM_DUP_INODES, &pctx);
+	shared = (ext2_ino_t *) e2fsck_allocate_memory(ctx,
+				sizeof(ext2_ino_t) * dict_count(&ino_dict),
+				"Shared inode list");
+	for (n = dict_first(&ino_dict); n; n = dict_next(&ino_dict, n)) {
+		p = (struct dup_inode *) dnode_get(n);
+		shared_len = 0;
+		file_ok = 1;
+		ino = (ext2_ino_t)VOIDPTR_TO_INT(dnode_getkey(n));
+		if (ino == EXT2_BAD_INO)
+			continue;
+
+		/*
+		 * Find all of the inodes which share blocks with this
+		 * one.  First we find all of the duplicate blocks
+		 * belonging to this inode, and then search each block
+		 * get the list of inodes, and merge them together.
+		 */
+		for (s = p->block_list; s; s = s->next) {
+			m = dict_lookup(&blk_dict, INT_TO_VOIDPTR(s->block));
+			if (!m)
+				continue; /* Should never happen... */
+			q = (struct dup_block *) dnode_get(m);
+			if (q->num_bad > 1)
+				file_ok = 0;
+			if (check_if_fs_block(ctx, s->block)) {
+				file_ok = 0;
+				meta_data = 1;
+			}
+			
+			/*
+			 * Add all inodes used by this block to the
+			 * shared[] --- which is a unique list, so
+			 * if an inode is already in shared[], don't
+			 * add it again.
+			 */
+			for (r = q->inode_list; r; r = r->next) {
+				if (r->inode == ino)
+					continue;
+				for (i = 0; i < shared_len; i++)
+					if (shared[i] == r->inode)
+						break;
+				if (i == shared_len) {
+					shared[shared_len++] = r->inode;
+				}
+			}
+		}
+
+		/*
+		 * Report the inode that we are working on
+		 */
+		pctx.inode = &p->inode;
+		pctx.ino = ino;
+		pctx.dir = p->dir;
+		pctx.blkcount = p->num_dupblocks;
+		pctx.num = meta_data ? shared_len+1 : shared_len;
+		fix_problem(ctx, PR_1D_DUP_FILE, &pctx);
+		pctx.blkcount = 0;
+		pctx.num = 0;
+		
+		if (meta_data)
+			fix_problem(ctx, PR_1D_SHARE_METADATA, &pctx);
+		
+		for (i = 0; i < shared_len; i++) {
+			m = dict_lookup(&ino_dict, INT_TO_VOIDPTR(shared[i]));
+			if (!m)
+				continue; /* should never happen */
+			t = (struct dup_inode *) dnode_get(m);
+			/*
+			 * Report the inode that we are sharing with
+			 */
+			pctx.inode = &t->inode;
+			pctx.ino = shared[i];
+			pctx.dir = t->dir;
+			fix_problem(ctx, PR_1D_DUP_FILE_LIST, &pctx);
+		}
+		if (file_ok) {
+			fix_problem(ctx, PR_1D_DUP_BLOCKS_DEALT, &pctx);
+			continue;
+		}
+		if (fix_problem(ctx, PR_1D_CLONE_QUESTION, &pctx)) {
+			pctx.errcode = clone_file(ctx, ino, p, block_buf);
+			if (pctx.errcode)
+				fix_problem(ctx, PR_1D_CLONE_ERROR, &pctx);
+			else
+				continue;
+		}
+		if (fix_problem(ctx, PR_1D_DELETE_QUESTION, &pctx))
+			delete_file(ctx, ino, p, block_buf);
+		else
+			ext2fs_unmark_valid(fs);
+	}
+	ext2fs_free_mem(&shared);
+}
+
+/*
+ * Drop the refcount on the dup_block structure, and clear the entry
+ * in the block_dup_map if appropriate.
+ */
+static void decrement_badcount(e2fsck_t ctx, blk_t block, struct dup_block *p)
+{
+	p->num_bad--;
+	if (p->num_bad <= 0 ||
+	    (p->num_bad == 1 && !check_if_fs_block(ctx, block)))
+		ext2fs_unmark_block_bitmap(ctx->block_dup_map, block);
+}
+
+static int delete_file_block(ext2_filsys fs,
+			     blk_t	*block_nr,
+			     e2_blkcnt_t blockcnt EXT2FS_ATTR((unused)),
+			     blk_t ref_block EXT2FS_ATTR((unused)),
+			     int ref_offset EXT2FS_ATTR((unused)),
+			     void *priv_data)
+{
+	struct process_block_struct *pb;
+	struct dup_block *p;
+	dnode_t	*n;
+	e2fsck_t ctx;
+
+	pb = (struct process_block_struct *) priv_data;
+	ctx = pb->ctx;
+
+	if (HOLE_BLKADDR(*block_nr))
+		return 0;
+
+	if (ext2fs_test_block_bitmap(ctx->block_dup_map, *block_nr)) {
+		n = dict_lookup(&blk_dict, INT_TO_VOIDPTR(*block_nr));
+		if (n) {
+			p = (struct dup_block *) dnode_get(n);
+			decrement_badcount(ctx, *block_nr, p);
+		} else
+			com_err("delete_file_block", 0,
+			    _("internal error; can't find dup_blk for %d\n"),
+				*block_nr);
+	} else {
+		ext2fs_unmark_block_bitmap(ctx->block_found_map, *block_nr);
+		ext2fs_block_alloc_stats(fs, *block_nr, -1);
+	}
+		
+	return 0;
+}
+		
+static void delete_file(e2fsck_t ctx, ext2_ino_t ino,
+			struct dup_inode *dp, char* block_buf)
+{
+	ext2_filsys fs = ctx->fs;
+	struct process_block_struct pb;
+	struct ext2_inode	inode;
+	struct problem_context	pctx;
+	unsigned int		count;
+
+	clear_problem_context(&pctx);
+	pctx.ino = pb.ino = ino;
+	pb.dup_blocks = dp->num_dupblocks;
+	pb.ctx = ctx;
+	pctx.str = "delete_file";
+
+	e2fsck_read_inode(ctx, ino, &inode, "delete_file");
+	if (ext2fs_inode_has_valid_blocks(&inode))
+		pctx.errcode = ext2fs_block_iterate2(fs, ino, 0, block_buf,
+						     delete_file_block, &pb);
+	if (pctx.errcode)
+		fix_problem(ctx, PR_1B_BLOCK_ITERATE, &pctx);
+	ext2fs_unmark_inode_bitmap(ctx->inode_used_map, ino);
+	ext2fs_unmark_inode_bitmap(ctx->inode_dir_map, ino);
+	if (ctx->inode_bad_map)
+		ext2fs_unmark_inode_bitmap(ctx->inode_bad_map, ino);
+	ext2fs_inode_alloc_stats2(fs, ino, -1, LINUX_S_ISDIR(inode.i_mode));
+
+	/* Inode may have changed by block_iterate, so reread it */
+	e2fsck_read_inode(ctx, ino, &inode, "delete_file");
+	inode.i_links_count = 0;
+	inode.i_dtime = time(0);
+	if (inode.i_file_acl &&
+	    (fs->super->s_feature_compat & EXT2_FEATURE_COMPAT_EXT_ATTR)) {
+		count = 1;
+		pctx.errcode = ext2fs_adjust_ea_refcount(fs, inode.i_file_acl,
+						   block_buf, -1, &count);
+		if (pctx.errcode == EXT2_ET_BAD_EA_BLOCK_NUM) {
+			pctx.errcode = 0;
+			count = 1;
+		}
+		if (pctx.errcode) {
+			pctx.blk = inode.i_file_acl;
+			fix_problem(ctx, PR_1B_ADJ_EA_REFCOUNT, &pctx);
+		}
+		/*
+		 * If the count is zero, then arrange to have the
+		 * block deleted.  If the block is in the block_dup_map,
+		 * also call delete_file_block since it will take care
+		 * of keeping the accounting straight.
+		 */
+		if ((count == 0) ||
+		    ext2fs_test_block_bitmap(ctx->block_dup_map,
+					     inode.i_file_acl)) 
+			delete_file_block(fs, &inode.i_file_acl,
+					  BLOCK_COUNT_EXTATTR, 0, 0, &pb);
+	}
+	e2fsck_write_inode(ctx, ino, &inode, "delete_file");
+}
+
+struct clone_struct {
+	errcode_t	errcode;
+	ext2_ino_t	dir;
+	char	*buf;
+	e2fsck_t ctx;
+};
+
+static int clone_file_block(ext2_filsys fs,
+			    blk_t	*block_nr,
+			    e2_blkcnt_t blockcnt,
+			    blk_t ref_block EXT2FS_ATTR((unused)),
+			    int ref_offset EXT2FS_ATTR((unused)),
+			    void *priv_data)
+{
+	struct dup_block *p;
+	blk_t	new_block;
+	errcode_t	retval;
+	struct clone_struct *cs = (struct clone_struct *) priv_data;
+	dnode_t *n;
+	e2fsck_t ctx;
+
+	ctx = cs->ctx;
+	
+	if (HOLE_BLKADDR(*block_nr))
+		return 0;
+
+	if (ext2fs_test_block_bitmap(ctx->block_dup_map, *block_nr)) {
+		n = dict_lookup(&blk_dict, INT_TO_VOIDPTR(*block_nr));
+		if (n) {
+			p = (struct dup_block *) dnode_get(n);
+			retval = ext2fs_new_block(fs, 0, ctx->block_found_map,
+						  &new_block);
+			if (retval) {
+				cs->errcode = retval;
+				return BLOCK_ABORT;
+			}
+			if (cs->dir && (blockcnt >= 0)) {
+				retval = ext2fs_set_dir_block(fs->dblist,
+				      cs->dir, new_block, blockcnt);
+				if (retval) {
+					cs->errcode = retval;
+					return BLOCK_ABORT;
+				}
+			}
+#if 0
+			printf("Cloning block %u to %u\n", *block_nr,
+			       new_block);
+#endif
+			retval = io_channel_read_blk(fs->io, *block_nr, 1,
+						     cs->buf);
+			if (retval) {
+				cs->errcode = retval;
+				return BLOCK_ABORT;
+			}
+			retval = io_channel_write_blk(fs->io, new_block, 1,
+						      cs->buf);
+			if (retval) {
+				cs->errcode = retval;
+				return BLOCK_ABORT;
+			}
+			decrement_badcount(ctx, *block_nr, p);
+			*block_nr = new_block;
+			ext2fs_mark_block_bitmap(ctx->block_found_map,
+						 new_block);
+			ext2fs_mark_block_bitmap(fs->block_map, new_block);
+			return BLOCK_CHANGED;
+		} else
+			com_err("clone_file_block", 0,
+			    _("internal error; can't find dup_blk for %d\n"),
+				*block_nr);
+	}
+	return 0;
+}
+		
+static int clone_file(e2fsck_t ctx, ext2_ino_t ino,
+		      struct dup_inode *dp, char* block_buf)
+{
+	ext2_filsys fs = ctx->fs;
+	errcode_t	retval;
+	struct clone_struct cs;
+	struct problem_context	pctx;
+	blk_t		blk;
+	dnode_t		*n;
+	struct inode_el	*ino_el;
+	struct dup_block	*db;
+	struct dup_inode	*di;
+
+	clear_problem_context(&pctx);
+	cs.errcode = 0;
+	cs.dir = 0;
+	cs.ctx = ctx;
+	retval = ext2fs_get_mem(fs->blocksize, &cs.buf);
+	if (retval)
+		return retval;
+
+	if (ext2fs_test_inode_bitmap(ctx->inode_dir_map, ino))
+		cs.dir = ino;
+
+	pctx.ino = ino;
+	pctx.str = "clone_file";
+	if (ext2fs_inode_has_valid_blocks(&dp->inode))
+		pctx.errcode = ext2fs_block_iterate2(fs, ino, 0, block_buf,
+						     clone_file_block, &cs);
+	ext2fs_mark_bb_dirty(fs);
+	if (pctx.errcode) {
+		fix_problem(ctx, PR_1B_BLOCK_ITERATE, &pctx);
+		retval = pctx.errcode;
+		goto errout;
+	}
+	if (cs.errcode) {
+		com_err("clone_file", cs.errcode,
+			_("returned from clone_file_block"));
+		retval = cs.errcode;
+		goto errout;
+	}
+	/* The inode may have changed on disk, so we have to re-read it */
+	e2fsck_read_inode(ctx, ino, &dp->inode, "clone file EA");
+	blk = dp->inode.i_file_acl;
+	if (blk && (clone_file_block(fs, &dp->inode.i_file_acl,
+				     BLOCK_COUNT_EXTATTR, 0, 0, &cs) ==
+		    BLOCK_CHANGED)) {
+		e2fsck_write_inode(ctx, ino, &dp->inode, "clone file EA");
+		/*
+		 * If we cloned the EA block, find all other inodes
+		 * which refered to that EA block, and modify
+		 * them to point to the new EA block.
+		 */
+		n = dict_lookup(&blk_dict, INT_TO_VOIDPTR(blk));
+		db = (struct dup_block *) dnode_get(n);
+		for (ino_el = db->inode_list; ino_el; ino_el = ino_el->next) {
+			if (ino_el->inode == ino)
+				continue;
+			n = dict_lookup(&ino_dict, INT_TO_VOIDPTR(ino_el->inode));
+			di = (struct dup_inode *) dnode_get(n);
+			if (di->inode.i_file_acl == blk) {
+				di->inode.i_file_acl = dp->inode.i_file_acl;
+				e2fsck_write_inode(ctx, ino_el->inode,
+					   &di->inode, "clone file EA");
+				decrement_badcount(ctx, blk, db);
+			}
+		}
+	}
+	retval = 0;
+errout:
+	ext2fs_free_mem(&cs.buf);
+	return retval;
+}
+
+/*
+ * This routine returns 1 if a block overlaps with one of the superblocks,
+ * group descriptors, inode bitmaps, or block bitmaps.
+ */
+static int check_if_fs_block(e2fsck_t ctx, blk_t test_block)
+{
+	ext2_filsys fs = ctx->fs;
+	blk_t	block;
+	dgrp_t	i;
+	
+	block = fs->super->s_first_data_block;
+	for (i = 0; i < fs->group_desc_count; i++) {
+
+		/* Check superblocks/block group descriptros */
+		if (ext2fs_bg_has_super(fs, i)) {
+			if (test_block >= block &&
+			    (test_block <= block + fs->desc_blocks))
+				return 1;
+		}
+		
+		/* Check the inode table */
+		if ((fs->group_desc[i].bg_inode_table) &&
+		    (test_block >= fs->group_desc[i].bg_inode_table) &&
+		    (test_block < (fs->group_desc[i].bg_inode_table +
+				   fs->inode_blocks_per_group)))
+			return 1;
+
+		/* Check the bitmap blocks */
+		if ((test_block == fs->group_desc[i].bg_block_bitmap) ||
+		    (test_block == fs->group_desc[i].bg_inode_bitmap))
+			return 1;
+		
+		block += fs->super->s_blocks_per_group;
+	}
+	return 0;
+}

e2fsprogs/e2fsck/pass3.c

@@ -0,0 +1,804 @@
+/*
+ * pass3.c -- pass #3 of e2fsck: Check for directory connectivity
+ *
+ * Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999 Theodore Ts'o.
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ * 
+ * Pass #3 assures that all directories are connected to the
+ * filesystem tree, using the following algorithm:
+ *
+ * First, the root directory is checked to make sure it exists; if
+ * not, e2fsck will offer to create a new one.  It is then marked as
+ * "done".
+ * 
+ * Then, pass3 interates over all directory inodes; for each directory
+ * it attempts to trace up the filesystem tree, using dirinfo.parent
+ * until it reaches a directory which has been marked "done".  If it
+ * can not do so, then the directory must be disconnected, and e2fsck
+ * will offer to reconnect it to /lost+found.  While it is chasing
+ * parent pointers up the filesystem tree, if pass3 sees a directory
+ * twice, then it has detected a filesystem loop, and it will again
+ * offer to reconnect the directory to /lost+found in to break the
+ * filesystem loop.
+ * 
+ * Pass 3 also contains the subroutine, e2fsck_reconnect_file() to
+ * reconnect inodes to /lost+found; this subroutine is also used by
+ * pass 4.  e2fsck_reconnect_file() calls get_lost_and_found(), which
+ * is responsible for creating /lost+found if it does not exist.
+ *
+ * Pass 3 frees the following data structures:
+ *     	- The dirinfo directory information cache.
+ */
+
+#ifdef HAVE_ERRNO_H
+#include <errno.h>
+#endif
+
+#include "e2fsck.h"
+#include "problem.h"
+
+static void check_root(e2fsck_t ctx);
+static int check_directory(e2fsck_t ctx, struct dir_info *dir,
+			   struct problem_context *pctx);
+static void fix_dotdot(e2fsck_t ctx, struct dir_info *dir, ext2_ino_t parent);
+
+static ext2fs_inode_bitmap inode_loop_detect = 0;
+static ext2fs_inode_bitmap inode_done_map = 0;
+	
+void e2fsck_pass3(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	int		i;
+#ifdef RESOURCE_TRACK
+	struct resource_track	rtrack;
+#endif
+	struct problem_context	pctx;
+	struct dir_info	*dir;
+	unsigned long maxdirs, count;
+
+#ifdef RESOURCE_TRACK
+	init_resource_track(&rtrack);
+#endif
+
+	clear_problem_context(&pctx);
+
+#ifdef MTRACE
+	mtrace_print("Pass 3");
+#endif
+
+	if (!(ctx->options & E2F_OPT_PREEN))
+		fix_problem(ctx, PR_3_PASS_HEADER, &pctx);
+
+	/*
+	 * Allocate some bitmaps to do loop detection.
+	 */
+	pctx.errcode = ext2fs_allocate_inode_bitmap(fs, _("inode done bitmap"),
+						    &inode_done_map);
+	if (pctx.errcode) {
+		pctx.num = 2;
+		fix_problem(ctx, PR_3_ALLOCATE_IBITMAP_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		goto abort_exit;
+	}
+#ifdef RESOURCE_TRACK
+	if (ctx->options & E2F_OPT_TIME) {
+		e2fsck_clear_progbar(ctx);
+		print_resource_track(_("Peak memory"), &ctx->global_rtrack);
+	}
+#endif
+
+	check_root(ctx);
+	if (ctx->flags & E2F_FLAG_SIGNAL_MASK)
+		goto abort_exit;
+
+	ext2fs_mark_inode_bitmap(inode_done_map, EXT2_ROOT_INO);
+
+	maxdirs = e2fsck_get_num_dirinfo(ctx);
+	count = 1;
+
+	if (ctx->progress)
+		if ((ctx->progress)(ctx, 3, 0, maxdirs))
+			goto abort_exit;
+	
+	for (i=0; (dir = e2fsck_dir_info_iter(ctx, &i)) != 0;) {
+		if (ctx->flags & E2F_FLAG_SIGNAL_MASK)
+			goto abort_exit;
+		if (ctx->progress && (ctx->progress)(ctx, 3, count++, maxdirs))
+			goto abort_exit;
+		if (ext2fs_test_inode_bitmap(ctx->inode_dir_map, dir->ino))
+			if (check_directory(ctx, dir, &pctx))
+				goto abort_exit;
+	}
+
+	/*
+	 * Force the creation of /lost+found if not present
+	 */
+	if ((ctx->flags & E2F_OPT_READONLY) == 0)
+		e2fsck_get_lost_and_found(ctx, 1);
+
+	/*
+	 * If there are any directories that need to be indexed or
+	 * optimized, do it here.
+	 */
+	e2fsck_rehash_directories(ctx);
+	
+abort_exit:
+	e2fsck_free_dir_info(ctx);
+	if (inode_loop_detect) {
+		ext2fs_free_inode_bitmap(inode_loop_detect);
+		inode_loop_detect = 0;
+	}
+	if (inode_done_map) {
+		ext2fs_free_inode_bitmap(inode_done_map);
+		inode_done_map = 0;
+	}
+
+#ifdef RESOURCE_TRACK
+	if (ctx->options & E2F_OPT_TIME2) {
+		e2fsck_clear_progbar(ctx);
+		print_resource_track(_("Pass 3"), &rtrack);
+	}
+#endif
+}
+
+/*
+ * This makes sure the root inode is present; if not, we ask if the
+ * user wants us to create it.  Not creating it is a fatal error.
+ */
+static void check_root(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	blk_t			blk;
+	struct ext2_inode	inode;
+	char *			block;
+	struct problem_context	pctx;
+	
+	clear_problem_context(&pctx);
+	
+	if (ext2fs_test_inode_bitmap(ctx->inode_used_map, EXT2_ROOT_INO)) {
+		/*
+		 * If the root inode is not a directory, die here.  The
+		 * user must have answered 'no' in pass1 when we
+		 * offered to clear it.
+		 */
+		if (!(ext2fs_test_inode_bitmap(ctx->inode_dir_map,
+					       EXT2_ROOT_INO))) {
+			fix_problem(ctx, PR_3_ROOT_NOT_DIR_ABORT, &pctx);
+			ctx->flags |= E2F_FLAG_ABORT;
+		}
+		return;
+	}
+
+	if (!fix_problem(ctx, PR_3_NO_ROOT_INODE, &pctx)) {
+		fix_problem(ctx, PR_3_NO_ROOT_INODE_ABORT, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+
+	e2fsck_read_bitmaps(ctx);
+	
+	/*
+	 * First, find a free block
+	 */
+	pctx.errcode = ext2fs_new_block(fs, 0, ctx->block_found_map, &blk);
+	if (pctx.errcode) {
+		pctx.str = "ext2fs_new_block";
+		fix_problem(ctx, PR_3_CREATE_ROOT_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+	ext2fs_mark_block_bitmap(ctx->block_found_map, blk);
+	ext2fs_mark_block_bitmap(fs->block_map, blk);
+	ext2fs_mark_bb_dirty(fs);
+
+	/*
+	 * Now let's create the actual data block for the inode
+	 */
+	pctx.errcode = ext2fs_new_dir_block(fs, EXT2_ROOT_INO, EXT2_ROOT_INO,
+					    &block);
+	if (pctx.errcode) {
+		pctx.str = "ext2fs_new_dir_block";
+		fix_problem(ctx, PR_3_CREATE_ROOT_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+
+	pctx.errcode = ext2fs_write_dir_block(fs, blk, block);
+	if (pctx.errcode) {
+		pctx.str = "ext2fs_write_dir_block";
+		fix_problem(ctx, PR_3_CREATE_ROOT_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+	ext2fs_free_mem(&block);
+
+	/*
+	 * Set up the inode structure
+	 */
+	memset(&inode, 0, sizeof(inode));
+	inode.i_mode = 040755;
+	inode.i_size = fs->blocksize;
+	inode.i_atime = inode.i_ctime = inode.i_mtime = time(0);
+	inode.i_links_count = 2;
+	inode.i_blocks = fs->blocksize / 512;
+	inode.i_block[0] = blk;
+
+	/*
+	 * Write out the inode.
+	 */
+	pctx.errcode = ext2fs_write_new_inode(fs, EXT2_ROOT_INO, &inode);
+	if (pctx.errcode) {
+		pctx.str = "ext2fs_write_inode";
+		fix_problem(ctx, PR_3_CREATE_ROOT_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+	
+	/*
+	 * Miscellaneous bookkeeping...
+	 */
+	e2fsck_add_dir_info(ctx, EXT2_ROOT_INO, EXT2_ROOT_INO);
+	ext2fs_icount_store(ctx->inode_count, EXT2_ROOT_INO, 2);
+	ext2fs_icount_store(ctx->inode_link_info, EXT2_ROOT_INO, 2);
+
+	ext2fs_mark_inode_bitmap(ctx->inode_used_map, EXT2_ROOT_INO);
+	ext2fs_mark_inode_bitmap(ctx->inode_dir_map, EXT2_ROOT_INO);
+	ext2fs_mark_inode_bitmap(fs->inode_map, EXT2_ROOT_INO);
+	ext2fs_mark_ib_dirty(fs);
+}
+
+/*
+ * This subroutine is responsible for making sure that a particular
+ * directory is connected to the root; if it isn't we trace it up as
+ * far as we can go, and then offer to connect the resulting parent to
+ * the lost+found.  We have to do loop detection; if we ever discover
+ * a loop, we treat that as a disconnected directory and offer to
+ * reparent it to lost+found.
+ * 
+ * However, loop detection is expensive, because for very large
+ * filesystems, the inode_loop_detect bitmap is huge, and clearing it
+ * is non-trivial.  Loops in filesystems are also a rare error case,
+ * and we shouldn't optimize for error cases.  So we try two passes of
+ * the algorithm.  The first time, we ignore loop detection and merely
+ * increment a counter; if the counter exceeds some extreme threshold,
+ * then we try again with the loop detection bitmap enabled.
+ */
+static int check_directory(e2fsck_t ctx, struct dir_info *dir,
+			   struct problem_context *pctx)
+{
+	ext2_filsys 	fs = ctx->fs;
+	struct dir_info *p = dir;
+	int		loop_pass = 0, parent_count = 0;
+
+	if (!p)
+		return 0;
+
+	while (1) {
+		/*
+		 * Mark this inode as being "done"; by the time we
+		 * return from this function, the inode we either be
+		 * verified as being connected to the directory tree,
+		 * or we will have offered to reconnect this to
+		 * lost+found.
+		 *
+		 * If it was marked done already, then we've reached a
+		 * parent we've already checked.
+		 */
+	  	if (ext2fs_mark_inode_bitmap(inode_done_map, p->ino))
+			break;
+
+		/*
+		 * If this directory doesn't have a parent, or we've
+		 * seen the parent once already, then offer to
+		 * reparent it to lost+found
+		 */
+		if (!p->parent ||
+		    (loop_pass && 
+		     (ext2fs_test_inode_bitmap(inode_loop_detect,
+					      p->parent)))) {
+			pctx->ino = p->ino;
+			if (fix_problem(ctx, PR_3_UNCONNECTED_DIR, pctx)) {
+				if (e2fsck_reconnect_file(ctx, pctx->ino))
+					ext2fs_unmark_valid(fs);
+				else {
+					p = e2fsck_get_dir_info(ctx, pctx->ino);
+					p->parent = ctx->lost_and_found;
+					fix_dotdot(ctx, p, ctx->lost_and_found);
+				}
+			}
+			break;
+		}
+		p = e2fsck_get_dir_info(ctx, p->parent);
+		if (!p) {
+			fix_problem(ctx, PR_3_NO_DIRINFO, pctx);
+			return 0;
+		}
+		if (loop_pass) {
+			ext2fs_mark_inode_bitmap(inode_loop_detect,
+						 p->ino);
+		} else if (parent_count++ > 2048) {
+			/*
+			 * If we've run into a path depth that's
+			 * greater than 2048, try again with the inode
+			 * loop bitmap turned on and start from the
+			 * top.
+			 */
+			loop_pass = 1;
+			if (inode_loop_detect)
+				ext2fs_clear_inode_bitmap(inode_loop_detect);
+			else {
+				pctx->errcode = ext2fs_allocate_inode_bitmap(fs, _("inode loop detection bitmap"), &inode_loop_detect);
+				if (pctx->errcode) {
+					pctx->num = 1;
+					fix_problem(ctx, 
+				    PR_3_ALLOCATE_IBITMAP_ERROR, pctx);
+					ctx->flags |= E2F_FLAG_ABORT;
+					return -1;
+				}
+			}
+			p = dir;
+		}
+	}
+
+	/*
+	 * Make sure that .. and the parent directory are the same;
+	 * offer to fix it if not.
+	 */
+	if (dir->parent != dir->dotdot) {
+		pctx->ino = dir->ino;
+		pctx->ino2 = dir->dotdot;
+		pctx->dir = dir->parent;
+		if (fix_problem(ctx, PR_3_BAD_DOT_DOT, pctx))
+			fix_dotdot(ctx, dir, dir->parent);
+	}
+	return 0;
+}
+
+/*
+ * This routine gets the lost_and_found inode, making it a directory
+ * if necessary
+ */
+ext2_ino_t e2fsck_get_lost_and_found(e2fsck_t ctx, int fix)
+{
+	ext2_filsys fs = ctx->fs;
+	ext2_ino_t			ino;
+	blk_t			blk;
+	errcode_t		retval;
+	struct ext2_inode	inode;
+	char *			block;
+	static const char	name[] = "lost+found";
+	struct 	problem_context	pctx;
+	struct dir_info 	*dirinfo;
+
+	if (ctx->lost_and_found)
+		return ctx->lost_and_found;
+
+	clear_problem_context(&pctx);
+	
+	retval = ext2fs_lookup(fs, EXT2_ROOT_INO, name,
+			       sizeof(name)-1, 0, &ino);
+	if (retval && !fix)
+		return 0;
+	if (!retval) {
+		if (ext2fs_test_inode_bitmap(ctx->inode_dir_map, ino)) {
+			ctx->lost_and_found = ino;
+			return ino;
+		}
+				
+		/* Lost+found isn't a directory! */
+		if (!fix)
+			return 0;
+		pctx.ino = ino;
+		if (!fix_problem(ctx, PR_3_LPF_NOTDIR, &pctx))
+			return 0;
+
+		/* OK, unlink the old /lost+found file. */
+		pctx.errcode = ext2fs_unlink(fs, EXT2_ROOT_INO, name, ino, 0);
+		if (pctx.errcode) {
+			pctx.str = "ext2fs_unlink";
+			fix_problem(ctx, PR_3_CREATE_LPF_ERROR, &pctx);
+			return 0;
+		}
+		dirinfo = e2fsck_get_dir_info(ctx, ino);
+		if (dirinfo)
+			dirinfo->parent = 0;
+		e2fsck_adjust_inode_count(ctx, ino, -1);
+	} else if (retval != EXT2_ET_FILE_NOT_FOUND) {
+		pctx.errcode = retval;
+		fix_problem(ctx, PR_3_ERR_FIND_LPF, &pctx);
+	}
+	if (!fix_problem(ctx, PR_3_NO_LF_DIR, 0))
+		return 0;
+
+	/*
+	 * Read the inode and block bitmaps in; we'll be messing with
+	 * them.
+	 */
+	e2fsck_read_bitmaps(ctx);
+	
+	/*
+	 * First, find a free block
+	 */
+	retval = ext2fs_new_block(fs, 0, ctx->block_found_map, &blk);
+	if (retval) {
+		pctx.errcode = retval;
+		fix_problem(ctx, PR_3_ERR_LPF_NEW_BLOCK, &pctx);
+		return 0;
+	}
+	ext2fs_mark_block_bitmap(ctx->block_found_map, blk);
+	ext2fs_block_alloc_stats(fs, blk, +1);
+
+	/*
+	 * Next find a free inode.
+	 */
+	retval = ext2fs_new_inode(fs, EXT2_ROOT_INO, 040700,
+				  ctx->inode_used_map, &ino);
+	if (retval) {
+		pctx.errcode = retval;
+		fix_problem(ctx, PR_3_ERR_LPF_NEW_INODE, &pctx);
+		return 0;
+	}
+	ext2fs_mark_inode_bitmap(ctx->inode_used_map, ino);
+	ext2fs_mark_inode_bitmap(ctx->inode_dir_map, ino);
+	ext2fs_inode_alloc_stats2(fs, ino, +1, 1);
+
+	/*
+	 * Now let's create the actual data block for the inode
+	 */
+	retval = ext2fs_new_dir_block(fs, ino, EXT2_ROOT_INO, &block);
+	if (retval) {
+		pctx.errcode = retval;
+		fix_problem(ctx, PR_3_ERR_LPF_NEW_DIR_BLOCK, &pctx);
+		return 0;
+	}
+
+	retval = ext2fs_write_dir_block(fs, blk, block);
+	ext2fs_free_mem(&block);
+	if (retval) {
+		pctx.errcode = retval;
+		fix_problem(ctx, PR_3_ERR_LPF_WRITE_BLOCK, &pctx);
+		return 0;
+	}
+
+	/*
+	 * Set up the inode structure
+	 */
+	memset(&inode, 0, sizeof(inode));
+	inode.i_mode = 040700;
+	inode.i_size = fs->blocksize;
+	inode.i_atime = inode.i_ctime = inode.i_mtime = time(0);
+	inode.i_links_count = 2;
+	inode.i_blocks = fs->blocksize / 512;
+	inode.i_block[0] = blk;
+
+	/*
+	 * Next, write out the inode.
+	 */
+	pctx.errcode = ext2fs_write_new_inode(fs, ino, &inode);
+	if (pctx.errcode) {
+		pctx.str = "ext2fs_write_inode";
+		fix_problem(ctx, PR_3_CREATE_LPF_ERROR, &pctx);
+		return 0;
+	}
+	/*
+	 * Finally, create the directory link
+	 */
+	pctx.errcode = ext2fs_link(fs, EXT2_ROOT_INO, name, ino, EXT2_FT_DIR);
+	if (pctx.errcode) {
+		pctx.str = "ext2fs_link";
+		fix_problem(ctx, PR_3_CREATE_LPF_ERROR, &pctx);
+		return 0;
+	}
+
+	/*
+	 * Miscellaneous bookkeeping that needs to be kept straight.
+	 */
+	e2fsck_add_dir_info(ctx, ino, EXT2_ROOT_INO);
+	e2fsck_adjust_inode_count(ctx, EXT2_ROOT_INO, 1);
+	ext2fs_icount_store(ctx->inode_count, ino, 2);
+	ext2fs_icount_store(ctx->inode_link_info, ino, 2);
+	ctx->lost_and_found = ino;
+#if 0
+	printf("/lost+found created; inode #%lu\n", ino);
+#endif
+	return ino;
+}
+
+/*
+ * This routine will connect a file to lost+found
+ */
+int e2fsck_reconnect_file(e2fsck_t ctx, ext2_ino_t ino)
+{
+	ext2_filsys fs = ctx->fs;
+	errcode_t	retval;
+	char		name[80];
+	struct problem_context	pctx;
+	struct ext2_inode 	inode;
+	int		file_type = 0;
+
+	clear_problem_context(&pctx);
+	pctx.ino = ino;
+
+	if (!ctx->bad_lost_and_found && !ctx->lost_and_found) {
+		if (e2fsck_get_lost_and_found(ctx, 1) == 0)
+			ctx->bad_lost_and_found++;
+	}
+	if (ctx->bad_lost_and_found) {
+		fix_problem(ctx, PR_3_NO_LPF, &pctx);
+		return 1;
+	}
+	
+	sprintf(name, "#%u", ino);
+	if (ext2fs_read_inode(fs, ino, &inode) == 0)
+		file_type = ext2_file_type(inode.i_mode);
+	retval = ext2fs_link(fs, ctx->lost_and_found, name, ino, file_type);
+	if (retval == EXT2_ET_DIR_NO_SPACE) {
+		if (!fix_problem(ctx, PR_3_EXPAND_LF_DIR, &pctx))
+			return 1;
+		retval = e2fsck_expand_directory(ctx, ctx->lost_and_found, 
+						 1, 0);
+		if (retval) {
+			pctx.errcode = retval;
+			fix_problem(ctx, PR_3_CANT_EXPAND_LPF, &pctx);
+			return 1;
+		}
+		retval = ext2fs_link(fs, ctx->lost_and_found, name,
+				     ino, file_type);
+	}
+	if (retval) {
+		pctx.errcode = retval;
+		fix_problem(ctx, PR_3_CANT_RECONNECT, &pctx);
+		return 1;
+	}
+	e2fsck_adjust_inode_count(ctx, ino, 1);
+
+	return 0;
+}
+
+/*
+ * Utility routine to adjust the inode counts on an inode.
+ */
+errcode_t e2fsck_adjust_inode_count(e2fsck_t ctx, ext2_ino_t ino, int adj)
+{
+	ext2_filsys fs = ctx->fs;
+	errcode_t		retval;
+	struct ext2_inode 	inode;
+	
+	if (!ino)
+		return 0;
+
+	retval = ext2fs_read_inode(fs, ino, &inode);
+	if (retval)
+		return retval;
+
+#if 0
+	printf("Adjusting link count for inode %lu by %d (from %d)\n", ino, adj,
+	       inode.i_links_count);
+#endif
+
+	if (adj == 1) {
+		ext2fs_icount_increment(ctx->inode_count, ino, 0);
+		if (inode.i_links_count == (__u16) ~0)
+			return 0;
+		ext2fs_icount_increment(ctx->inode_link_info, ino, 0);
+		inode.i_links_count++;
+	} else if (adj == -1) {
+		ext2fs_icount_decrement(ctx->inode_count, ino, 0);
+		if (inode.i_links_count == 0)
+			return 0;
+		ext2fs_icount_decrement(ctx->inode_link_info, ino, 0);
+		inode.i_links_count--;
+	}
+	
+	retval = ext2fs_write_inode(fs, ino, &inode);
+	if (retval)
+		return retval;
+
+	return 0;
+}
+
+/*
+ * Fix parent --- this routine fixes up the parent of a directory.
+ */
+struct fix_dotdot_struct {
+	ext2_filsys	fs;
+	ext2_ino_t	parent;
+	int		done;
+	e2fsck_t	ctx;
+};
+
+static int fix_dotdot_proc(struct ext2_dir_entry *dirent,
+			   int	offset EXT2FS_ATTR((unused)),
+			   int	blocksize EXT2FS_ATTR((unused)),
+			   char	*buf EXT2FS_ATTR((unused)),
+			   void	*priv_data)
+{
+	struct fix_dotdot_struct *fp = (struct fix_dotdot_struct *) priv_data;
+	errcode_t	retval;
+	struct problem_context pctx;
+
+	if ((dirent->name_len & 0xFF) != 2)
+		return 0;
+	if (strncmp(dirent->name, "..", 2))
+		return 0;
+
+	clear_problem_context(&pctx);
+	
+	retval = e2fsck_adjust_inode_count(fp->ctx, dirent->inode, -1);
+	if (retval) {
+		pctx.errcode = retval;
+		fix_problem(fp->ctx, PR_3_ADJUST_INODE, &pctx);
+	}
+	retval = e2fsck_adjust_inode_count(fp->ctx, fp->parent, 1);
+	if (retval) {
+		pctx.errcode = retval;
+		fix_problem(fp->ctx, PR_3_ADJUST_INODE, &pctx);
+	}
+	dirent->inode = fp->parent;
+
+	fp->done++;
+	return DIRENT_ABORT | DIRENT_CHANGED;
+}
+
+static void fix_dotdot(e2fsck_t ctx, struct dir_info *dir, ext2_ino_t parent)
+{
+	ext2_filsys fs = ctx->fs;
+	errcode_t	retval;
+	struct fix_dotdot_struct fp;
+	struct problem_context pctx;
+
+	fp.fs = fs;
+	fp.parent = parent;
+	fp.done = 0;
+	fp.ctx = ctx;
+
+#if 0
+	printf("Fixing '..' of inode %lu to be %lu...\n", dir->ino, parent);
+#endif
+	
+	retval = ext2fs_dir_iterate(fs, dir->ino, DIRENT_FLAG_INCLUDE_EMPTY,
+				    0, fix_dotdot_proc, &fp);
+	if (retval || !fp.done) {
+		clear_problem_context(&pctx);
+		pctx.ino = dir->ino;
+		pctx.errcode = retval;
+		fix_problem(ctx, retval ? PR_3_FIX_PARENT_ERR :
+			    PR_3_FIX_PARENT_NOFIND, &pctx);
+		ext2fs_unmark_valid(fs);
+	}
+	dir->dotdot = parent;
+	
+	return;
+}
+
+/*
+ * These routines are responsible for expanding a /lost+found if it is
+ * too small.
+ */
+
+struct expand_dir_struct {
+	int			num;
+	int			guaranteed_size;
+	int			newblocks;
+	int			last_block;
+	errcode_t		err;
+	e2fsck_t		ctx;
+};
+
+static int expand_dir_proc(ext2_filsys fs,
+			   blk_t	*blocknr,
+			   e2_blkcnt_t	blockcnt,
+			   blk_t ref_block EXT2FS_ATTR((unused)),
+			   int ref_offset EXT2FS_ATTR((unused)),
+			   void	*priv_data)
+{
+	struct expand_dir_struct *es = (struct expand_dir_struct *) priv_data;
+	blk_t	new_blk;
+	static blk_t	last_blk = 0;
+	char		*block;
+	errcode_t	retval;
+	e2fsck_t	ctx;
+
+	ctx = es->ctx;
+	
+	if (es->guaranteed_size && blockcnt >= es->guaranteed_size)
+		return BLOCK_ABORT;
+
+	if (blockcnt > 0)
+		es->last_block = blockcnt;
+	if (*blocknr) {
+		last_blk = *blocknr;
+		return 0;
+	}
+	retval = ext2fs_new_block(fs, last_blk, ctx->block_found_map,
+				  &new_blk);
+	if (retval) {
+		es->err = retval;
+		return BLOCK_ABORT;
+	}
+	if (blockcnt > 0) {
+		retval = ext2fs_new_dir_block(fs, 0, 0, &block);
+		if (retval) {
+			es->err = retval;
+			return BLOCK_ABORT;
+		}
+		es->num--;
+		retval = ext2fs_write_dir_block(fs, new_blk, block);
+	} else {
+		retval = ext2fs_get_mem(fs->blocksize, &block);
+		if (retval) {
+			es->err = retval;
+			return BLOCK_ABORT;
+		}
+		memset(block, 0, fs->blocksize);
+		retval = io_channel_write_blk(fs->io, new_blk, 1, block);
+	}	
+	if (retval) {
+		es->err = retval;
+		return BLOCK_ABORT;
+	}
+	ext2fs_free_mem(&block);
+	*blocknr = new_blk;
+	ext2fs_mark_block_bitmap(ctx->block_found_map, new_blk);
+	ext2fs_block_alloc_stats(fs, new_blk, +1);
+	es->newblocks++;
+	
+	if (es->num == 0)
+		return (BLOCK_CHANGED | BLOCK_ABORT);
+	else
+		return BLOCK_CHANGED;
+}
+
+errcode_t e2fsck_expand_directory(e2fsck_t ctx, ext2_ino_t dir,
+				  int num, int guaranteed_size)
+{
+	ext2_filsys fs = ctx->fs;
+	errcode_t	retval;
+	struct expand_dir_struct es;
+	struct ext2_inode	inode;
+	
+	if (!(fs->flags & EXT2_FLAG_RW))
+		return EXT2_ET_RO_FILSYS;
+
+	/*
+	 * Read the inode and block bitmaps in; we'll be messing with
+	 * them.
+	 */
+	e2fsck_read_bitmaps(ctx);
+
+	retval = ext2fs_check_directory(fs, dir);
+	if (retval)
+		return retval;
+	
+	es.num = num;
+	es.guaranteed_size = guaranteed_size;
+	es.last_block = 0;
+	es.err = 0;
+	es.newblocks = 0;
+	es.ctx = ctx;
+	
+	retval = ext2fs_block_iterate2(fs, dir, BLOCK_FLAG_APPEND,
+				       0, expand_dir_proc, &es);
+
+	if (es.err)
+		return es.err;
+
+	/*
+	 * Update the size and block count fields in the inode.
+	 */
+	retval = ext2fs_read_inode(fs, dir, &inode);
+	if (retval)
+		return retval;
+	
+	inode.i_size = (es.last_block + 1) * fs->blocksize;
+	inode.i_blocks += (fs->blocksize / 512) * es.newblocks;
+
+	e2fsck_write_inode(ctx, dir, &inode, "expand_directory");
+
+	return 0;
+}
+

e2fsprogs/e2fsck/pass4.c

@@ -0,0 +1,178 @@
+/*
+ * pass4.c -- pass #4 of e2fsck: Check reference counts
+ *
+ * Copyright (C) 1993, 1994, 1995, 1996, 1997 Theodore Ts'o.
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ *
+ * Pass 4 frees the following data structures:
+ * 	- A bitmap of which inodes are in bad blocks.	(inode_bb_map)
+ * 	- A bitmap of which inodes are imagic inodes.	(inode_imagic_map)
+ */
+
+#include "e2fsck.h"
+#include "problem.h"
+
+/*
+ * This routine is called when an inode is not connected to the
+ * directory tree.
+ * 
+ * This subroutine returns 1 then the caller shouldn't bother with the
+ * rest of the pass 4 tests.
+ */
+static int disconnect_inode(e2fsck_t ctx, ext2_ino_t i)
+{
+	ext2_filsys fs = ctx->fs;
+	struct ext2_inode	inode;
+	struct problem_context	pctx;
+
+	e2fsck_read_inode(ctx, i, &inode, "pass4: disconnect_inode");
+	clear_problem_context(&pctx);
+	pctx.ino = i;
+	pctx.inode = &inode;
+	
+	/*
+	 * Offer to delete any zero-length files that does not have
+	 * blocks.  If there is an EA block, it might have useful
+	 * information, so we won't prompt to delete it, but let it be
+	 * reconnected to lost+found.
+	 */
+	if (!inode.i_blocks && (LINUX_S_ISREG(inode.i_mode) ||
+				LINUX_S_ISDIR(inode.i_mode))) {
+		if (fix_problem(ctx, PR_4_ZERO_LEN_INODE, &pctx)) {
+			ext2fs_icount_store(ctx->inode_link_info, i, 0);
+			inode.i_links_count = 0;
+			inode.i_dtime = time(0);
+			e2fsck_write_inode(ctx, i, &inode,
+					   "disconnect_inode");
+			/*
+			 * Fix up the bitmaps...
+			 */
+			e2fsck_read_bitmaps(ctx);
+			ext2fs_unmark_inode_bitmap(ctx->inode_used_map, i);
+			ext2fs_unmark_inode_bitmap(ctx->inode_dir_map, i);
+			ext2fs_inode_alloc_stats2(fs, i, -1,
+						  LINUX_S_ISDIR(inode.i_mode));
+			return 0;
+		}
+	}
+	
+	/*
+	 * Prompt to reconnect.
+	 */
+	if (fix_problem(ctx, PR_4_UNATTACHED_INODE, &pctx)) {
+		if (e2fsck_reconnect_file(ctx, i))
+			ext2fs_unmark_valid(fs);
+	} else {
+		/*
+		 * If we don't attach the inode, then skip the
+		 * i_links_test since there's no point in trying to
+		 * force i_links_count to zero.
+		 */
+		ext2fs_unmark_valid(fs);
+		return 1;
+	}
+	return 0;
+}
+
+
+void e2fsck_pass4(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	ext2_ino_t	i;
+	struct ext2_inode	inode;
+#ifdef RESOURCE_TRACK
+	struct resource_track	rtrack;
+#endif
+	struct problem_context	pctx;
+	__u16	link_count, link_counted;
+	char	*buf = 0;
+	int	group, maxgroup;
+	
+#ifdef RESOURCE_TRACK
+	init_resource_track(&rtrack);
+#endif
+
+#ifdef MTRACE
+	mtrace_print("Pass 4");
+#endif
+
+	clear_problem_context(&pctx);
+
+	if (!(ctx->options & E2F_OPT_PREEN))
+		fix_problem(ctx, PR_4_PASS_HEADER, &pctx);
+
+	group = 0;
+	maxgroup = fs->group_desc_count;
+	if (ctx->progress)
+		if ((ctx->progress)(ctx, 4, 0, maxgroup))
+			return;
+	
+	for (i=1; i <= fs->super->s_inodes_count; i++) {
+		if (ctx->flags & E2F_FLAG_SIGNAL_MASK)
+			return;
+		if ((i % fs->super->s_inodes_per_group) == 0) {
+			group++;
+			if (ctx->progress)
+				if ((ctx->progress)(ctx, 4, group, maxgroup))
+					return;
+		}
+		if (i == EXT2_BAD_INO ||
+		    (i > EXT2_ROOT_INO && i < EXT2_FIRST_INODE(fs->super)))
+			continue;
+		if (!(ext2fs_test_inode_bitmap(ctx->inode_used_map, i)) ||
+		    (ctx->inode_imagic_map &&
+		     ext2fs_test_inode_bitmap(ctx->inode_imagic_map, i)) ||
+		    (ctx->inode_bb_map &&
+		     ext2fs_test_inode_bitmap(ctx->inode_bb_map, i)))
+			continue;
+		ext2fs_icount_fetch(ctx->inode_link_info, i, &link_count);
+		ext2fs_icount_fetch(ctx->inode_count, i, &link_counted);
+		if (link_counted == 0) {
+			if (!buf)
+				buf = e2fsck_allocate_memory(ctx,
+				     fs->blocksize, "bad_inode buffer");
+			if (e2fsck_process_bad_inode(ctx, 0, i, buf))
+				continue;
+			if (disconnect_inode(ctx, i))
+				continue;
+			ext2fs_icount_fetch(ctx->inode_link_info, i,
+					    &link_count);
+			ext2fs_icount_fetch(ctx->inode_count, i,
+					    &link_counted);
+		}
+		if (link_counted != link_count) {
+			e2fsck_read_inode(ctx, i, &inode, "pass4");
+			pctx.ino = i;
+			pctx.inode = &inode;
+			if (link_count != inode.i_links_count) {
+				pctx.num = link_count;
+				fix_problem(ctx,
+					    PR_4_INCONSISTENT_COUNT, &pctx);
+			}
+			pctx.num = link_counted;
+			if (fix_problem(ctx, PR_4_BAD_REF_COUNT, &pctx)) {
+				inode.i_links_count = link_counted;
+				e2fsck_write_inode(ctx, i, &inode, "pass4");
+			}
+		}
+	}
+	ext2fs_free_icount(ctx->inode_link_info); ctx->inode_link_info = 0;
+	ext2fs_free_icount(ctx->inode_count); ctx->inode_count = 0;
+	ext2fs_free_inode_bitmap(ctx->inode_bb_map);
+	ctx->inode_bb_map = 0;
+	ext2fs_free_inode_bitmap(ctx->inode_imagic_map);
+	ctx->inode_imagic_map = 0;
+	if (buf)
+		ext2fs_free_mem(&buf);
+#ifdef RESOURCE_TRACK
+	if (ctx->options & E2F_OPT_TIME2) {
+		e2fsck_clear_progbar(ctx);
+		print_resource_track(_("Pass 4"), &rtrack);
+	}
+#endif
+}
+

e2fsprogs/e2fsck/pass5.c

@@ -0,0 +1,547 @@
+/*
+ * pass5.c --- check block and inode bitmaps against on-disk bitmaps
+ * 
+ * Copyright (C) 1993, 1994, 1995, 1996, 1997 Theodore Ts'o.
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ * 
+ */
+
+#include "e2fsck.h"
+#include "problem.h"
+
+static void check_block_bitmaps(e2fsck_t ctx);
+static void check_inode_bitmaps(e2fsck_t ctx);
+static void check_inode_end(e2fsck_t ctx);
+static void check_block_end(e2fsck_t ctx);
+
+void e2fsck_pass5(e2fsck_t ctx)
+{
+#ifdef RESOURCE_TRACK
+	struct resource_track	rtrack;
+#endif
+	struct problem_context	pctx;
+	
+#ifdef MTRACE
+	mtrace_print("Pass 5");
+#endif
+
+#ifdef RESOURCE_TRACK
+	init_resource_track(&rtrack);
+#endif
+	
+	clear_problem_context(&pctx);
+
+	if (!(ctx->options & E2F_OPT_PREEN))
+		fix_problem(ctx, PR_5_PASS_HEADER, &pctx);
+
+	if (ctx->progress)
+		if ((ctx->progress)(ctx, 5, 0, ctx->fs->group_desc_count*2))
+			return;
+
+	e2fsck_read_bitmaps(ctx);
+
+	check_block_bitmaps(ctx);
+	if (ctx->flags & E2F_FLAG_SIGNAL_MASK)
+		return;
+	check_inode_bitmaps(ctx);
+	if (ctx->flags & E2F_FLAG_SIGNAL_MASK)
+		return;
+	check_inode_end(ctx);
+	if (ctx->flags & E2F_FLAG_SIGNAL_MASK)
+		return;
+	check_block_end(ctx);
+	if (ctx->flags & E2F_FLAG_SIGNAL_MASK)
+		return;
+
+	ext2fs_free_inode_bitmap(ctx->inode_used_map);
+	ctx->inode_used_map = 0;
+	ext2fs_free_inode_bitmap(ctx->inode_dir_map);
+	ctx->inode_dir_map = 0;
+	ext2fs_free_block_bitmap(ctx->block_found_map);
+	ctx->block_found_map = 0;
+
+#ifdef RESOURCE_TRACK
+	if (ctx->options & E2F_OPT_TIME2) {
+		e2fsck_clear_progbar(ctx);
+		print_resource_track(_("Pass 5"), &rtrack);
+	}
+#endif
+}
+
+#define NO_BLK ((blk_t) -1)
+
+static void print_bitmap_problem(e2fsck_t ctx, int problem,
+			    struct problem_context *pctx)
+{
+	switch (problem) {
+	case PR_5_BLOCK_UNUSED:
+		if (pctx->blk == pctx->blk2)
+			pctx->blk2 = 0;
+		else
+			problem = PR_5_BLOCK_RANGE_UNUSED;
+		break;
+	case PR_5_BLOCK_USED:
+		if (pctx->blk == pctx->blk2)
+			pctx->blk2 = 0;
+		else
+			problem = PR_5_BLOCK_RANGE_USED;
+		break;
+	case PR_5_INODE_UNUSED:
+		if (pctx->ino == pctx->ino2)
+			pctx->ino2 = 0;
+		else
+			problem = PR_5_INODE_RANGE_UNUSED;
+		break;
+	case PR_5_INODE_USED:
+		if (pctx->ino == pctx->ino2)
+			pctx->ino2 = 0;
+		else
+			problem = PR_5_INODE_RANGE_USED;
+		break;
+	}
+	fix_problem(ctx, problem, pctx);
+	pctx->blk = pctx->blk2 = NO_BLK;
+	pctx->ino = pctx->ino2 = 0;
+}
+	
+static void check_block_bitmaps(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	blk_t	i;
+	int	*free_array;
+	int	group = 0;
+	unsigned int	blocks = 0;
+	unsigned int	free_blocks = 0;
+	int	group_free = 0;
+	int	actual, bitmap;
+	struct problem_context	pctx;
+	int	problem, save_problem, fixit, had_problem;
+	errcode_t	retval;
+	
+	clear_problem_context(&pctx);
+	free_array = (int *) e2fsck_allocate_memory(ctx,
+	    fs->group_desc_count * sizeof(int), "free block count array");
+
+	if ((fs->super->s_first_data_block <
+	     ext2fs_get_block_bitmap_start(ctx->block_found_map)) ||
+	    (fs->super->s_blocks_count-1 >
+	     ext2fs_get_block_bitmap_end(ctx->block_found_map))) {
+		pctx.num = 1;
+		pctx.blk = fs->super->s_first_data_block;
+		pctx.blk2 = fs->super->s_blocks_count -1;
+		pctx.ino = ext2fs_get_block_bitmap_start(ctx->block_found_map);
+		pctx.ino2 = ext2fs_get_block_bitmap_end(ctx->block_found_map);
+		fix_problem(ctx, PR_5_BMAP_ENDPOINTS, &pctx);
+
+		ctx->flags |= E2F_FLAG_ABORT; /* fatal */
+		return;
+	}
+		       
+	if ((fs->super->s_first_data_block <
+	     ext2fs_get_block_bitmap_start(fs->block_map)) ||
+	    (fs->super->s_blocks_count-1 >
+	     ext2fs_get_block_bitmap_end(fs->block_map))) {
+		pctx.num = 2;
+		pctx.blk = fs->super->s_first_data_block;
+		pctx.blk2 = fs->super->s_blocks_count -1;
+		pctx.ino = ext2fs_get_block_bitmap_start(fs->block_map);
+		pctx.ino2 = ext2fs_get_block_bitmap_end(fs->block_map);
+		fix_problem(ctx, PR_5_BMAP_ENDPOINTS, &pctx);
+
+		ctx->flags |= E2F_FLAG_ABORT; /* fatal */
+		return;
+	}
+		       
+redo_counts:
+	had_problem = 0;
+	save_problem = 0;
+	pctx.blk = pctx.blk2 = NO_BLK;
+	for (i = fs->super->s_first_data_block;
+	     i < fs->super->s_blocks_count;
+	     i++) {
+		actual = ext2fs_fast_test_block_bitmap(ctx->block_found_map, i);
+		bitmap = ext2fs_fast_test_block_bitmap(fs->block_map, i);
+		
+		if (actual == bitmap)
+			goto do_counts;
+		
+		if (!actual && bitmap) {
+			/*
+			 * Block not used, but marked in use in the bitmap.
+			 */
+			problem = PR_5_BLOCK_UNUSED;
+		} else {
+			/*
+			 * Block used, but not marked in use in the bitmap.
+			 */
+			problem = PR_5_BLOCK_USED;
+		}
+		if (pctx.blk == NO_BLK) {
+			pctx.blk = pctx.blk2 = i;
+			save_problem = problem;
+		} else {
+			if ((problem == save_problem) &&
+			    (pctx.blk2 == i-1))
+				pctx.blk2++;
+			else {
+				print_bitmap_problem(ctx, save_problem, &pctx);
+				pctx.blk = pctx.blk2 = i;
+				save_problem = problem;
+			}
+		}
+		ctx->flags |= E2F_FLAG_PROG_SUPPRESS;
+		had_problem++;
+		
+	do_counts:
+		if (!bitmap) {
+			group_free++;
+			free_blocks++;
+		}
+		blocks ++;
+		if ((blocks == fs->super->s_blocks_per_group) ||
+		    (i == fs->super->s_blocks_count-1)) {
+			free_array[group] = group_free;
+			group ++;
+			blocks = 0;
+			group_free = 0;
+			if (ctx->progress)
+				if ((ctx->progress)(ctx, 5, group,
+						    fs->group_desc_count*2))
+					return;
+		}
+	}
+	if (pctx.blk != NO_BLK)
+		print_bitmap_problem(ctx, save_problem, &pctx);
+	if (had_problem)
+		fixit = end_problem_latch(ctx, PR_LATCH_BBITMAP);
+	else
+		fixit = -1;
+	ctx->flags &= ~E2F_FLAG_PROG_SUPPRESS;
+	
+	if (fixit == 1) {
+		ext2fs_free_block_bitmap(fs->block_map);
+		retval = ext2fs_copy_bitmap(ctx->block_found_map,
+						  &fs->block_map);
+		if (retval) {
+			clear_problem_context(&pctx);
+			fix_problem(ctx, PR_5_COPY_BBITMAP_ERROR, &pctx);
+			ctx->flags |= E2F_FLAG_ABORT;
+			return;
+		}
+		ext2fs_set_bitmap_padding(fs->block_map);
+		ext2fs_mark_bb_dirty(fs);
+		
+		/* Redo the counts */
+		blocks = 0; free_blocks = 0; group_free = 0; group = 0;
+		memset(free_array, 0, fs->group_desc_count * sizeof(int));
+		goto redo_counts;
+	} else if (fixit == 0)
+		ext2fs_unmark_valid(fs);
+
+	for (i = 0; i < fs->group_desc_count; i++) {
+		if (free_array[i] != fs->group_desc[i].bg_free_blocks_count) {
+			pctx.group = i;
+			pctx.blk = fs->group_desc[i].bg_free_blocks_count;
+			pctx.blk2 = free_array[i];
+
+			if (fix_problem(ctx, PR_5_FREE_BLOCK_COUNT_GROUP,
+					&pctx)) {
+				fs->group_desc[i].bg_free_blocks_count =
+					free_array[i];
+				ext2fs_mark_super_dirty(fs);
+			} else
+				ext2fs_unmark_valid(fs);
+		}
+	}
+	if (free_blocks != fs->super->s_free_blocks_count) {
+		pctx.group = 0;
+		pctx.blk = fs->super->s_free_blocks_count;
+		pctx.blk2 = free_blocks;
+
+		if (fix_problem(ctx, PR_5_FREE_BLOCK_COUNT, &pctx)) {
+			fs->super->s_free_blocks_count = free_blocks;
+			ext2fs_mark_super_dirty(fs);
+		} else
+			ext2fs_unmark_valid(fs);
+	}
+	ext2fs_free_mem(&free_array);
+}
+			
+static void check_inode_bitmaps(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	ext2_ino_t	i;
+	unsigned int	free_inodes = 0;
+	int		group_free = 0;
+	int		dirs_count = 0;
+	int		group = 0;
+	unsigned int	inodes = 0;
+	int		*free_array;
+	int		*dir_array;
+	int		actual, bitmap;
+	errcode_t	retval;
+	struct problem_context	pctx;
+	int		problem, save_problem, fixit, had_problem;
+	
+	clear_problem_context(&pctx);
+	free_array = (int *) e2fsck_allocate_memory(ctx,
+	    fs->group_desc_count * sizeof(int), "free inode count array");
+				     
+	dir_array = (int *) e2fsck_allocate_memory(ctx,
+	   fs->group_desc_count * sizeof(int), "directory count array");
+				     
+	if ((1 < ext2fs_get_inode_bitmap_start(ctx->inode_used_map)) ||
+	    (fs->super->s_inodes_count > 
+	     ext2fs_get_inode_bitmap_end(ctx->inode_used_map))) {
+		pctx.num = 3;
+		pctx.blk = 1;
+		pctx.blk2 = fs->super->s_inodes_count;
+		pctx.ino = ext2fs_get_inode_bitmap_start(ctx->inode_used_map);
+		pctx.ino2 = ext2fs_get_inode_bitmap_end(ctx->inode_used_map);
+		fix_problem(ctx, PR_5_BMAP_ENDPOINTS, &pctx);
+
+		ctx->flags |= E2F_FLAG_ABORT; /* fatal */
+		return;
+	}
+	if ((1 < ext2fs_get_inode_bitmap_start(fs->inode_map)) ||
+	    (fs->super->s_inodes_count > 
+	     ext2fs_get_inode_bitmap_end(fs->inode_map))) {
+		pctx.num = 4;
+		pctx.blk = 1;
+		pctx.blk2 = fs->super->s_inodes_count;
+		pctx.ino = ext2fs_get_inode_bitmap_start(fs->inode_map);
+		pctx.ino2 = ext2fs_get_inode_bitmap_end(fs->inode_map);
+		fix_problem(ctx, PR_5_BMAP_ENDPOINTS, &pctx);
+
+		ctx->flags |= E2F_FLAG_ABORT; /* fatal */
+		return;
+	}
+
+redo_counts:
+	had_problem = 0;
+	save_problem = 0;
+	pctx.ino = pctx.ino2 = 0;
+	for (i = 1; i <= fs->super->s_inodes_count; i++) {
+		actual = ext2fs_fast_test_inode_bitmap(ctx->inode_used_map, i);
+		bitmap = ext2fs_fast_test_inode_bitmap(fs->inode_map, i);
+		
+		if (actual == bitmap)
+			goto do_counts;
+		
+		if (!actual && bitmap) {
+			/*
+			 * Inode wasn't used, but marked in bitmap
+			 */
+			problem = PR_5_INODE_UNUSED;
+		} else /* if (actual && !bitmap) */ {
+			/*
+			 * Inode used, but not in bitmap
+			 */
+			problem = PR_5_INODE_USED;
+		}
+		if (pctx.ino == 0) {
+			pctx.ino = pctx.ino2 = i;
+			save_problem = problem;
+		} else {
+			if ((problem == save_problem) &&
+			    (pctx.ino2 == i-1))
+				pctx.ino2++;
+			else {
+				print_bitmap_problem(ctx, save_problem, &pctx);
+				pctx.ino = pctx.ino2 = i;
+				save_problem = problem;
+			}
+		}
+		ctx->flags |= E2F_FLAG_PROG_SUPPRESS;
+		had_problem++;
+		
+do_counts:
+		if (!bitmap) {
+			group_free++;
+			free_inodes++;
+		} else {
+			if (ext2fs_test_inode_bitmap(ctx->inode_dir_map, i))
+				dirs_count++;
+		}
+		inodes++;
+		if ((inodes == fs->super->s_inodes_per_group) ||
+		    (i == fs->super->s_inodes_count)) {
+			free_array[group] = group_free;
+			dir_array[group] = dirs_count;
+			group ++;
+			inodes = 0;
+			group_free = 0;
+			dirs_count = 0;
+			if (ctx->progress)
+				if ((ctx->progress)(ctx, 5,
+					    group + fs->group_desc_count,
+					    fs->group_desc_count*2))
+					return;
+		}
+	}
+	if (pctx.ino)
+		print_bitmap_problem(ctx, save_problem, &pctx);
+	
+	if (had_problem)
+		fixit = end_problem_latch(ctx, PR_LATCH_IBITMAP);
+	else
+		fixit = -1;
+	ctx->flags &= ~E2F_FLAG_PROG_SUPPRESS;
+	
+	if (fixit == 1) {
+		ext2fs_free_inode_bitmap(fs->inode_map);
+		retval = ext2fs_copy_bitmap(ctx->inode_used_map,
+						  &fs->inode_map);
+		if (retval) {
+			clear_problem_context(&pctx);
+			fix_problem(ctx, PR_5_COPY_IBITMAP_ERROR, &pctx);
+			ctx->flags |= E2F_FLAG_ABORT;
+			return;
+		}
+		ext2fs_set_bitmap_padding(fs->inode_map);
+		ext2fs_mark_ib_dirty(fs);
+
+		/* redo counts */
+		inodes = 0; free_inodes = 0; group_free = 0;
+		dirs_count = 0; group = 0;
+		memset(free_array, 0, fs->group_desc_count * sizeof(int));
+		memset(dir_array, 0, fs->group_desc_count * sizeof(int));
+		goto redo_counts;
+	} else if (fixit == 0)
+		ext2fs_unmark_valid(fs);
+	
+	for (i = 0; i < fs->group_desc_count; i++) {
+		if (free_array[i] != fs->group_desc[i].bg_free_inodes_count) {
+			pctx.group = i;
+			pctx.ino = fs->group_desc[i].bg_free_inodes_count;
+			pctx.ino2 = free_array[i];
+			if (fix_problem(ctx, PR_5_FREE_INODE_COUNT_GROUP,
+					&pctx)) {
+				fs->group_desc[i].bg_free_inodes_count =
+					free_array[i];
+				ext2fs_mark_super_dirty(fs);
+			} else
+				ext2fs_unmark_valid(fs);
+		}
+		if (dir_array[i] != fs->group_desc[i].bg_used_dirs_count) {
+			pctx.group = i;
+			pctx.ino = fs->group_desc[i].bg_used_dirs_count;
+			pctx.ino2 = dir_array[i];
+
+			if (fix_problem(ctx, PR_5_FREE_DIR_COUNT_GROUP,
+					&pctx)) {
+				fs->group_desc[i].bg_used_dirs_count =
+					dir_array[i];
+				ext2fs_mark_super_dirty(fs);
+			} else
+				ext2fs_unmark_valid(fs);
+		}
+	}
+	if (free_inodes != fs->super->s_free_inodes_count) {
+		pctx.group = -1;
+		pctx.ino = fs->super->s_free_inodes_count;
+		pctx.ino2 = free_inodes;
+
+		if (fix_problem(ctx, PR_5_FREE_INODE_COUNT, &pctx)) {
+			fs->super->s_free_inodes_count = free_inodes;
+			ext2fs_mark_super_dirty(fs);
+		} else
+			ext2fs_unmark_valid(fs);
+	}
+	ext2fs_free_mem(&free_array);
+	ext2fs_free_mem(&dir_array);
+}
+
+static void check_inode_end(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	ext2_ino_t	end, save_inodes_count, i;
+	struct problem_context	pctx;
+
+	clear_problem_context(&pctx);
+
+	end = EXT2_INODES_PER_GROUP(fs->super) * fs->group_desc_count;
+	pctx.errcode = ext2fs_fudge_inode_bitmap_end(fs->inode_map, end,
+						     &save_inodes_count);
+	if (pctx.errcode) {
+		pctx.num = 1;
+		fix_problem(ctx, PR_5_FUDGE_BITMAP_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT; /* fatal */
+		return;
+	}
+	if (save_inodes_count == end)
+		return;
+	
+	for (i = save_inodes_count + 1; i <= end; i++) {
+		if (!ext2fs_test_inode_bitmap(fs->inode_map, i)) {
+			if (fix_problem(ctx, PR_5_INODE_BMAP_PADDING, &pctx)) {
+				for (i = save_inodes_count + 1; i <= end; i++)
+					ext2fs_mark_inode_bitmap(fs->inode_map,
+								 i);
+				ext2fs_mark_ib_dirty(fs);
+			} else
+				ext2fs_unmark_valid(fs);
+			break;
+		}
+	}
+
+	pctx.errcode = ext2fs_fudge_inode_bitmap_end(fs->inode_map,
+						     save_inodes_count, 0);
+	if (pctx.errcode) {
+		pctx.num = 2;
+		fix_problem(ctx, PR_5_FUDGE_BITMAP_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT; /* fatal */
+		return;
+	}
+}
+
+static void check_block_end(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	blk_t	end, save_blocks_count, i;
+	struct problem_context	pctx;
+
+	clear_problem_context(&pctx);
+
+	end = fs->block_map->start +
+		(EXT2_BLOCKS_PER_GROUP(fs->super) * fs->group_desc_count) - 1;
+	pctx.errcode = ext2fs_fudge_block_bitmap_end(fs->block_map, end,
+						     &save_blocks_count);
+	if (pctx.errcode) {
+		pctx.num = 3;
+		fix_problem(ctx, PR_5_FUDGE_BITMAP_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT; /* fatal */
+		return;
+	}
+	if (save_blocks_count == end)
+		return;
+	
+	for (i = save_blocks_count + 1; i <= end; i++) {
+		if (!ext2fs_test_block_bitmap(fs->block_map, i)) {
+			if (fix_problem(ctx, PR_5_BLOCK_BMAP_PADDING, &pctx)) {
+				for (i = save_blocks_count + 1; i <= end; i++)
+					ext2fs_mark_block_bitmap(fs->block_map,
+								 i);
+				ext2fs_mark_bb_dirty(fs);
+			} else
+				ext2fs_unmark_valid(fs);
+			break;
+		}
+	}
+
+	pctx.errcode = ext2fs_fudge_block_bitmap_end(fs->block_map,
+						     save_blocks_count, 0);
+	if (pctx.errcode) {
+		pctx.num = 4;
+		fix_problem(ctx, PR_5_FUDGE_BITMAP_ERROR, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT; /* fatal */
+		return;
+	}
+}
+
+
+

e2fsprogs/e2fsck/problem.h

@@ -0,0 +1,897 @@
+/*
+ * problem.h --- e2fsck problem error codes
+ *
+ * Copyright 1996 by Theodore Ts'o
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ */
+
+typedef __u32 problem_t;
+
+struct problem_context {
+	errcode_t	errcode;
+	ext2_ino_t ino, ino2, dir;
+	struct ext2_inode *inode;
+	struct ext2_dir_entry *dirent;
+	blk_t	blk, blk2;
+	e2_blkcnt_t	blkcount;
+	int		group;
+	__u64	num;
+	const char *str;
+};
+
+/*
+ * We define a set of "latch groups"; these are problems which are
+ * handled as a set.  The user answers once for a particular latch
+ * group.
+ */
+#define PR_LATCH_MASK	0x0ff0  /* Latch mask */
+#define PR_LATCH_BLOCK	0x0010	/* Latch for illegal blocks (pass 1) */
+#define PR_LATCH_BBLOCK	0x0020	/* Latch for bad block inode blocks (pass 1) */
+#define PR_LATCH_IBITMAP 0x0030 /* Latch for pass 5 inode bitmap proc. */
+#define PR_LATCH_BBITMAP 0x0040 /* Latch for pass 5 inode bitmap proc. */
+#define PR_LATCH_RELOC	0x0050  /* Latch for superblock relocate hint */
+#define PR_LATCH_DBLOCK	0x0060	/* Latch for pass 1b dup block headers */
+#define PR_LATCH_LOW_DTIME 0x0070 /* Latch for pass1 orphaned list refugees */
+#define PR_LATCH_TOOBIG	0x0080	/* Latch for file to big errors */
+#define PR_LATCH_OPTIMIZE_DIR 0x0090 /* Latch for optimize directories */
+
+#define PR_LATCH(x)	((((x) & PR_LATCH_MASK) >> 4) - 1)
+
+/*
+ * Latch group descriptor flags
+ */
+#define PRL_YES		0x0001	/* Answer yes */
+#define PRL_NO		0x0002	/* Answer no */
+#define PRL_LATCHED	0x0004	/* The latch group is latched */
+#define PRL_SUPPRESS	0x0008	/* Suppress all latch group questions */
+
+#define PRL_VARIABLE	0x000f	/* All the flags that need to be reset */
+
+/*
+ * Pre-Pass 1 errors
+ */
+
+/* Block bitmap not in group */
+#define PR_0_BB_NOT_GROUP	0x000001
+
+/* Inode bitmap not in group */
+#define PR_0_IB_NOT_GROUP	0x000002
+
+/* Inode table not in group */
+#define PR_0_ITABLE_NOT_GROUP	0x000003
+
+/* Superblock corrupt */
+#define PR_0_SB_CORRUPT		0x000004
+
+/* Filesystem size is wrong */
+#define PR_0_FS_SIZE_WRONG	0x000005
+
+/* Fragments not supported */
+#define PR_0_NO_FRAGMENTS	0x000006
+
+/* Bad blocks_per_group */
+#define PR_0_BLOCKS_PER_GROUP	0x000007
+
+/* Bad first_data_block */
+#define PR_0_FIRST_DATA_BLOCK	0x000008
+
+/* Adding UUID to filesystem */
+#define PR_0_ADD_UUID		0x000009
+
+/* Relocate hint */	
+#define PR_0_RELOCATE_HINT	0x00000A
+
+/* Miscellaneous superblock corruption */
+#define PR_0_MISC_CORRUPT_SUPER	0x00000B
+	
+/* Error determing physical device size of filesystem */
+#define PR_0_GETSIZE_ERROR	0x00000C
+
+/* Inode count in the superblock incorrect */
+#define PR_0_INODE_COUNT_WRONG	0x00000D
+
+/* The Hurd does not support the filetype feature */
+#define PR_0_HURD_CLEAR_FILETYPE 0x00000E
+
+/* Journal inode is invalid */
+#define PR_0_JOURNAL_BAD_INODE	0x00000F
+
+/* The external journal has multiple filesystems (which we can't handle yet) */
+#define PR_0_JOURNAL_UNSUPP_MULTIFS 0x000010
+
+/* Can't find external journal */
+#define PR_0_CANT_FIND_JOURNAL	0x000011
+
+/* External journal has bad superblock */
+#define PR_0_EXT_JOURNAL_BAD_SUPER 0x000012
+
+/* Superblock has a bad journal UUID */
+#define PR_0_JOURNAL_BAD_UUID	0x000013
+
+/* Journal has an unknown superblock type */
+#define PR_0_JOURNAL_UNSUPP_SUPER 0x000014
+
+/* Journal superblock is corrupt */
+#define PR_0_JOURNAL_BAD_SUPER	0x000015
+
+/* Journal superblock is corrupt */
+#define PR_0_JOURNAL_HAS_JOURNAL 0x000016
+
+/* Superblock has recovery flag set but no journal */
+#define PR_0_JOURNAL_RECOVER_SET 0x000017
+
+/* Journal has data, but recovery flag is clear */
+#define PR_0_JOURNAL_RECOVERY_CLEAR 0x000018
+
+/* Ask if we should clear the journal */
+#define PR_0_JOURNAL_RESET_JOURNAL 0x000019
+
+/* Filesystem revision is 0, but feature flags are set */
+#define PR_0_FS_REV_LEVEL	0x00001A
+
+/* Clearing orphan inode */
+#define PR_0_ORPHAN_CLEAR_INODE			0x000020
+	
+/* Illegal block found in orphaned inode */
+#define PR_0_ORPHAN_ILLEGAL_BLOCK_NUM		0x000021
+
+/* Already cleared block found in orphaned inode */
+#define PR_0_ORPHAN_ALREADY_CLEARED_BLOCK	0x000022
+	
+/* Illegal orphan inode in superblock */
+#define PR_0_ORPHAN_ILLEGAL_HEAD_INODE		0x000023
+
+/* Illegal inode in orphaned inode list */
+#define PR_0_ORPHAN_ILLEGAL_INODE 		0x000024
+
+/* Journal has unsupported read-only feature - abort */
+#define PR_0_JOURNAL_UNSUPP_ROCOMPAT		0x000025
+
+/* Journal has unsupported incompatible feature - abort */
+#define PR_0_JOURNAL_UNSUPP_INCOMPAT		0x000026
+
+/* Journal has unsupported version number */
+#define PR_0_JOURNAL_UNSUPP_VERSION		0x000027
+
+/* Moving journal to hidden file */
+#define	PR_0_MOVE_JOURNAL			0x000028
+
+/* Error moving journal */
+#define	PR_0_ERR_MOVE_JOURNAL			0x000029
+
+/* Clearing V2 journal superblock */
+#define PR_0_CLEAR_V2_JOURNAL			0x00002A
+
+/* Run journal anyway */
+#define PR_0_JOURNAL_RUN			0x00002B
+
+/* Run journal anyway by default */
+#define PR_0_JOURNAL_RUN_DEFAULT		0x00002C
+
+/* Backup journal inode blocks */
+#define PR_0_BACKUP_JNL				0x00002D
+
+/* Reserved blocks w/o resize_inode */
+#define PR_0_NONZERO_RESERVED_GDT_BLOCKS	0x00002E
+
+/* Resize_inode not enabled, but resize inode is non-zero */
+#define PR_0_CLEAR_RESIZE_INODE			0x00002F
+
+/* Resize inode invalid */
+#define PR_0_RESIZE_INODE_INVALID		0x000030
+
+/*
+ * Pass 1 errors
+ */
+
+/* Pass 1: Checking inodes, blocks, and sizes */
+#define PR_1_PASS_HEADER		0x010000
+
+/* Root directory is not an inode */
+#define PR_1_ROOT_NO_DIR		0x010001
+
+/* Root directory has dtime set */
+#define PR_1_ROOT_DTIME			0x010002
+
+/* Reserved inode has bad mode */
+#define PR_1_RESERVED_BAD_MODE		0x010003
+
+/* Deleted inode has zero dtime */
+#define PR_1_ZERO_DTIME			0x010004
+
+/* Inode in use, but dtime set */
+#define PR_1_SET_DTIME			0x010005
+
+/* Zero-length directory */
+#define PR_1_ZERO_LENGTH_DIR		0x010006
+
+/* Block bitmap conflicts with some other fs block */
+#define PR_1_BB_CONFLICT		0x010007
+
+/* Inode bitmap conflicts with some other fs block */
+#define PR_1_IB_CONFLICT		0x010008
+
+/* Inode table conflicts with some other fs block */
+#define PR_1_ITABLE_CONFLICT		0x010009
+
+/* Block bitmap is on a bad block */
+#define PR_1_BB_BAD_BLOCK		0x01000A
+
+/* Inode bitmap is on a bad block */
+#define PR_1_IB_BAD_BLOCK		0x01000B
+
+/* Inode has incorrect i_size */
+#define PR_1_BAD_I_SIZE			0x01000C
+
+/* Inode has incorrect i_blocks */
+#define PR_1_BAD_I_BLOCKS		0x01000D
+
+/* Illegal block number in inode */
+#define PR_1_ILLEGAL_BLOCK_NUM		0x01000E
+
+/* Block number overlaps fs metadata */
+#define PR_1_BLOCK_OVERLAPS_METADATA	0x01000F
+
+/* Inode has illegal blocks (latch question) */
+#define PR_1_INODE_BLOCK_LATCH		0x010010
+
+/* Too many bad blocks in inode */
+#define	PR_1_TOO_MANY_BAD_BLOCKS 	0x010011
+	
+/* Illegal block number in bad block inode */
+#define PR_1_BB_ILLEGAL_BLOCK_NUM 	0x010012
+
+/* Bad block inode has illegal blocks (latch question) */
+#define PR_1_INODE_BBLOCK_LATCH		0x010013
+
+/* Duplicate or bad blocks in use! */
+#define PR_1_DUP_BLOCKS_PREENSTOP	0x010014
+	
+/* Bad block used as bad block indirect block */	  
+#define PR_1_BBINODE_BAD_METABLOCK	0x010015
+
+/* Inconsistency can't be fixed prompt */
+#define PR_1_BBINODE_BAD_METABLOCK_PROMPT 0x010016
+	
+/* Bad primary block */
+#define PR_1_BAD_PRIMARY_BLOCK		0x010017
+		  
+/* Bad primary block prompt */
+#define PR_1_BAD_PRIMARY_BLOCK_PROMPT	0x010018
+
+/* Bad primary superblock */
+#define PR_1_BAD_PRIMARY_SUPERBLOCK	0x010019
+
+/* Bad primary block group descriptors */
+#define PR_1_BAD_PRIMARY_GROUP_DESCRIPTOR 0x01001A
+
+/* Bad superblock in group */
+#define PR_1_BAD_SUPERBLOCK		0x01001B
+
+/* Bad block group descriptors in group */
+#define PR_1_BAD_GROUP_DESCRIPTORS	0x01001C
+
+/* Block claimed for no reason */	  
+#define PR_1_PROGERR_CLAIMED_BLOCK	0x01001D
+
+/* Error allocating blocks for relocating metadata */
+#define PR_1_RELOC_BLOCK_ALLOCATE	0x01001E
+		
+/* Error allocating block buffer during relocation process */
+#define PR_1_RELOC_MEMORY_ALLOCATE	0x01001F
+		
+/* Relocating metadata group information from X to Y */	
+#define PR_1_RELOC_FROM_TO		0x010020
+		
+/* Relocating metatdata group information to X */
+#define PR_1_RELOC_TO			0x010021
+		
+/* Block read error during relocation process */
+#define PR_1_RELOC_READ_ERR		0x010022
+		
+/* Block write error during relocation process */
+#define PR_1_RELOC_WRITE_ERR		0x010023
+
+/* Error allocating inode bitmap */
+#define PR_1_ALLOCATE_IBITMAP_ERROR	0x010024
+
+/* Error allocating block bitmap */
+#define PR_1_ALLOCATE_BBITMAP_ERROR	0x010025
+
+/* Error allocating icount structure */
+#define PR_1_ALLOCATE_ICOUNT		0x010026
+	
+/* Error allocating dbcount */
+#define PR_1_ALLOCATE_DBCOUNT		0x010027
+
+/* Error while scanning inodes */
+#define PR_1_ISCAN_ERROR		0x010028
+
+/* Error while iterating over blocks */
+#define PR_1_BLOCK_ITERATE		0x010029
+
+/* Error while storing inode count information */	  
+#define PR_1_ICOUNT_STORE		0x01002A
+
+/* Error while storing directory block information */	  
+#define PR_1_ADD_DBLOCK			0x01002B
+
+/* Error while reading inode (for clearing) */
+#define PR_1_READ_INODE			0x01002C
+
+/* Suppress messages prompt */
+#define PR_1_SUPPRESS_MESSAGES		0x01002D
+
+/* Imagic flag set on an inode when filesystem doesn't support it */
+#define PR_1_SET_IMAGIC			0x01002F
+
+/* Immutable flag set on a device or socket inode */
+#define PR_1_SET_IMMUTABLE		0x010030
+
+/* Compression flag set on a non-compressed filesystem */
+#define PR_1_COMPR_SET			0x010031
+
+/* Non-zero size on on device, fifo or socket inode */
+#define PR_1_SET_NONZSIZE		0x010032
+
+/* Filesystem revision is 0, but feature flags are set */
+#define PR_1_FS_REV_LEVEL		0x010033
+
+/* Journal inode not in use, needs clearing */
+#define PR_1_JOURNAL_INODE_NOT_CLEAR	0x010034
+
+/* Journal inode has wrong mode */
+#define PR_1_JOURNAL_BAD_MODE		0x010035
+
+/* Inode that was part of orphan linked list */
+#define PR_1_LOW_DTIME			0x010036
+
+/* Latch question which asks how to deal with low dtime inodes */
+#define PR_1_ORPHAN_LIST_REFUGEES	0x010037
+
+/* Error allocating refcount structure */
+#define PR_1_ALLOCATE_REFCOUNT		0x010038
+	
+/* Error reading Extended Attribute block */
+#define PR_1_READ_EA_BLOCK		0x010039
+
+/* Invalid Extended Attribute block */
+#define PR_1_BAD_EA_BLOCK		0x01003A
+
+/* Error reading Extended Attribute block while fixing refcount -- abort */
+#define PR_1_EXTATTR_READ_ABORT		0x01003B
+
+/* Extended attribute reference count incorrect */
+#define PR_1_EXTATTR_REFCOUNT		0x01003C
+
+/* Error writing Extended Attribute block while fixing refcount */ 
+#define PR_1_EXTATTR_WRITE		0x01003D
+
+/* Multiple EA blocks not supported */
+#define PR_1_EA_MULTI_BLOCK		0x01003E
+
+/* Error allocating EA region allocation structure */
+#define PR_1_EA_ALLOC_REGION		0x01003F
+
+/* Error EA allocation collision */
+#define PR_1_EA_ALLOC_COLLISION		0x010040
+
+/* Bad extended attribute name */
+#define PR_1_EA_BAD_NAME		0x010041
+
+/* Bad extended attribute value */
+#define PR_1_EA_BAD_VALUE		0x010042
+
+/* Inode too big (latch question) */
+#define PR_1_INODE_TOOBIG		0x010043
+
+/* Directory too big */
+#define PR_1_TOOBIG_DIR			0x010044
+
+/* Regular file too big */
+#define PR_1_TOOBIG_REG			0x010045
+
+/* Symlink too big */
+#define PR_1_TOOBIG_SYMLINK		0x010046
+
+/* INDEX_FL flag set on a non-HTREE filesystem */
+#define PR_1_HTREE_SET			0x010047
+
+/* INDEX_FL flag set on a non-directory */	
+#define PR_1_HTREE_NODIR		0x010048
+
+/* Invalid root node in HTREE directory */	
+#define PR_1_HTREE_BADROOT		0x010049
+
+/* Unsupported hash version in HTREE directory */	
+#define PR_1_HTREE_HASHV		0x01004A
+
+/* Incompatible flag in HTREE root node */	
+#define PR_1_HTREE_INCOMPAT		0x01004B
+
+/* HTREE too deep */	
+#define PR_1_HTREE_DEPTH		0x01004C
+
+/* Bad block has indirect block that conflicts with filesystem block */
+#define PR_1_BB_FS_BLOCK		0x01004D
+
+/* Resize inode failed */
+#define PR_1_RESIZE_INODE_CREATE	0x01004E
+
+/* inode->i_size is too long */	
+#define PR_1_EXTRA_ISIZE		0x01004F
+
+/* attribute name is too long */
+#define PR_1_ATTR_NAME_LEN		0x010050
+
+/* wrong EA value offset */
+#define PR_1_ATTR_VALUE_OFFSET		0x010051
+
+/* wrong EA blocknumber */
+#define PR_1_ATTR_VALUE_BLOCK		0x010052
+
+/* wrong EA value size */
+#define PR_1_ATTR_VALUE_SIZE		0x010053
+
+/* wrong EA hash value */
+#define PR_1_ATTR_HASH			0x010054
+
+/*
+ * Pass 1b errors
+ */
+
+/* Pass 1B: Rescan for duplicate/bad blocks */
+#define PR_1B_PASS_HEADER	0x011000
+
+/* Duplicate/bad block(s) header */
+#define PR_1B_DUP_BLOCK_HEADER	0x011001
+
+/* Duplicate/bad block(s) in inode */
+#define PR_1B_DUP_BLOCK		0x011002
+
+/* Duplicate/bad block(s) end */
+#define PR_1B_DUP_BLOCK_END	0x011003
+	
+/* Error while scanning inodes */
+#define PR_1B_ISCAN_ERROR	0x011004
+
+/* Error allocating inode bitmap */
+#define PR_1B_ALLOCATE_IBITMAP_ERROR 0x011005
+
+/* Error while iterating over blocks */
+#define PR_1B_BLOCK_ITERATE	0x0110006
+
+/* Error adjusting EA refcount */
+#define PR_1B_ADJ_EA_REFCOUNT	0x0110007
+
+
+/* Pass 1C: Scan directories for inodes with dup blocks. */
+#define PR_1C_PASS_HEADER	0x012000
+
+
+/* Pass 1D: Reconciling duplicate blocks */
+#define PR_1D_PASS_HEADER	0x013000
+
+/* File has duplicate blocks */
+#define PR_1D_DUP_FILE		0x013001
+
+/* List of files sharing duplicate blocks */	
+#define PR_1D_DUP_FILE_LIST	0x013002
+
+/* File sharing blocks with filesystem metadata  */	
+#define PR_1D_SHARE_METADATA	0x013003
+
+/* Report of how many duplicate/bad inodes */	
+#define PR_1D_NUM_DUP_INODES	0x013004
+
+/* Duplicated blocks already reassigned or cloned. */
+#define PR_1D_DUP_BLOCKS_DEALT	0x013005
+
+/* Clone duplicate/bad blocks? */
+#define PR_1D_CLONE_QUESTION	0x013006
+
+/* Delete file? */
+#define PR_1D_DELETE_QUESTION	0x013007
+
+/* Couldn't clone file (error) */
+#define PR_1D_CLONE_ERROR	0x013008
+		
+/*
+ * Pass 2 errors
+ */
+
+/* Pass 2: Checking directory structure */
+#define PR_2_PASS_HEADER	0x020000
+
+/* Bad inode number for '.' */
+#define PR_2_BAD_INODE_DOT	0x020001
+
+/* Directory entry has bad inode number */
+#define PR_2_BAD_INO		0x020002
+
+/* Directory entry has deleted or unused inode */
+#define PR_2_UNUSED_INODE	0x020003
+
+/* Directry entry is link to '.' */
+#define PR_2_LINK_DOT		0x020004
+
+/* Directory entry points to inode now located in a bad block */
+#define PR_2_BB_INODE		0x020005
+
+/* Directory entry contains a link to a directory */
+#define PR_2_LINK_DIR		0x020006
+
+/* Directory entry contains a link to the root directry */
+#define PR_2_LINK_ROOT		0x020007
+
+/* Directory entry has illegal characters in its name */
+#define PR_2_BAD_NAME		0x020008
+
+/* Missing '.' in directory inode */	  
+#define PR_2_MISSING_DOT	0x020009
+
+/* Missing '..' in directory inode */	  
+#define PR_2_MISSING_DOT_DOT	0x02000A
+
+/* First entry in directory inode doesn't contain '.' */
+#define PR_2_1ST_NOT_DOT	0x02000B
+
+/* Second entry in directory inode doesn't contain '..' */
+#define PR_2_2ND_NOT_DOT_DOT	0x02000C
+
+/* i_faddr should be zero */
+#define PR_2_FADDR_ZERO		0x02000D
+
+/* i_file_acl should be zero */
+#define PR_2_FILE_ACL_ZERO	0x02000E
+
+/* i_dir_acl should be zero */
+#define PR_2_DIR_ACL_ZERO	0x02000F
+
+/* i_frag should be zero */
+#define PR_2_FRAG_ZERO		0x020010
+
+/* i_fsize should be zero */
+#define PR_2_FSIZE_ZERO		0x020011
+		  
+/* inode has bad mode */
+#define PR_2_BAD_MODE		0x020012
+
+/* directory corrupted */
+#define PR_2_DIR_CORRUPTED	0x020013
+		  
+/* filename too long */
+#define PR_2_FILENAME_LONG	0x020014
+		  
+/* Directory inode has a missing block (hole) */
+#define PR_2_DIRECTORY_HOLE	0x020015
+
+/* '.' is not NULL terminated */
+#define PR_2_DOT_NULL_TERM	0x020016
+
+/* '..' is not NULL terminated */
+#define PR_2_DOT_DOT_NULL_TERM	0x020017
+
+/* Illegal character device in inode */
+#define PR_2_BAD_CHAR_DEV	0x020018
+
+/* Illegal block device in inode */
+#define PR_2_BAD_BLOCK_DEV	0x020019
+
+/* Duplicate '.' entry */
+#define PR_2_DUP_DOT		0x02001A
+
+/* Duplicate '..' entry */
+#define PR_2_DUP_DOT_DOT	0x02001B
+	
+/* Internal error: couldn't find dir_info */
+#define PR_2_NO_DIRINFO		0x02001C
+
+/* Final rec_len is wrong */
+#define PR_2_FINAL_RECLEN	0x02001D
+
+/* Error allocating icount structure */
+#define PR_2_ALLOCATE_ICOUNT	0x02001E
+
+/* Error iterating over directory blocks */
+#define PR_2_DBLIST_ITERATE	0x02001F
+
+/* Error reading directory block */
+#define PR_2_READ_DIRBLOCK	0x020020
+
+/* Error writing directory block */
+#define PR_2_WRITE_DIRBLOCK	0x020021
+
+/* Error allocating new directory block */
+#define PR_2_ALLOC_DIRBOCK	0x020022
+
+/* Error deallocating inode */
+#define PR_2_DEALLOC_INODE	0x020023
+
+/* Directory entry for '.' is big.  Split? */
+#define PR_2_SPLIT_DOT		0x020024
+
+/* Illegal FIFO */
+#define PR_2_BAD_FIFO		0x020025
+
+/* Illegal socket */
+#define PR_2_BAD_SOCKET		0x020026
+
+/* Directory filetype not set */
+#define PR_2_SET_FILETYPE	0x020027
+
+/* Directory filetype incorrect */
+#define PR_2_BAD_FILETYPE	0x020028
+
+/* Directory filetype set when it shouldn't be */
+#define PR_2_CLEAR_FILETYPE	0x020029
+
+/* Directory filename can't be zero-length  */
+#define PR_2_NULL_NAME		0x020030
+
+/* Invalid symlink */
+#define PR_2_INVALID_SYMLINK	0x020031
+
+/* i_file_acl (extended attribute) is bad */
+#define PR_2_FILE_ACL_BAD	0x020032
+
+/* Filesystem contains large files, but has no such flag in sb */
+#define PR_2_FEATURE_LARGE_FILES 0x020033
+
+/* Node in HTREE directory not referenced */
+#define PR_2_HTREE_NOTREF	0x020034
+
+/* Node in HTREE directory referenced twice */
+#define PR_2_HTREE_DUPREF	0x020035
+
+/* Node in HTREE directory has bad min hash */
+#define PR_2_HTREE_MIN_HASH	0x020036
+
+/* Node in HTREE directory has bad max hash */
+#define PR_2_HTREE_MAX_HASH	0x020037
+
+/* Clear invalid HTREE directory */
+#define PR_2_HTREE_CLEAR	0x020038
+
+/* Clear the htree flag forcibly */
+/* #define PR_2_HTREE_FCLR	0x020039 */
+
+/* Bad block in htree interior node */
+#define PR_2_HTREE_BADBLK	0x02003A
+
+/* Error adjusting EA refcount */
+#define PR_2_ADJ_EA_REFCOUNT	0x02003B
+
+/* Invalid HTREE root node */
+#define PR_2_HTREE_BAD_ROOT	0x02003C
+
+/* Invalid HTREE limit */
+#define PR_2_HTREE_BAD_LIMIT	0x02003D
+
+/* Invalid HTREE count */
+#define PR_2_HTREE_BAD_COUNT	0x02003E
+
+/* HTREE interior node has out-of-order hashes in table */
+#define PR_2_HTREE_HASH_ORDER	0x02003F
+
+/* Node in HTREE directory has bad depth */
+#define PR_2_HTREE_BAD_DEPTH	0x020040
+
+/* Duplicate directory entry found */
+#define PR_2_DUPLICATE_DIRENT	0x020041
+
+/* Non-unique filename found */
+#define PR_2_NON_UNIQUE_FILE	0x020042
+
+/* Duplicate directory entry found */
+#define PR_2_REPORT_DUP_DIRENT	0x020043
+
+/*
+ * Pass 3 errors
+ */
+
+/* Pass 3: Checking directory connectivity */
+#define PR_3_PASS_HEADER		0x030000
+
+/* Root inode not allocated */
+#define PR_3_NO_ROOT_INODE		0x030001
+
+/* No room in lost+found */
+#define PR_3_EXPAND_LF_DIR		0x030002
+
+/* Unconnected directory inode */
+#define PR_3_UNCONNECTED_DIR		0x030003
+
+/* /lost+found not found */
+#define PR_3_NO_LF_DIR			0x030004
+
+/* .. entry is incorrect */
+#define PR_3_BAD_DOT_DOT		0x030005
+
+/* Bad or non-existent /lost+found.  Cannot reconnect */
+#define PR_3_NO_LPF			0x030006
+
+/* Could not expand /lost+found */
+#define PR_3_CANT_EXPAND_LPF		0x030007
+
+/* Could not reconnect inode */
+#define PR_3_CANT_RECONNECT		0x030008
+
+/* Error while trying to find /lost+found */
+#define PR_3_ERR_FIND_LPF		0x030009
+
+/* Error in ext2fs_new_block while creating /lost+found */
+#define PR_3_ERR_LPF_NEW_BLOCK		0x03000A
+
+/* Error in ext2fs_new_inode while creating /lost+found */
+#define PR_3_ERR_LPF_NEW_INODE		0x03000B
+
+/* Error in ext2fs_new_dir_block while creating /lost+found */	  
+#define PR_3_ERR_LPF_NEW_DIR_BLOCK	0x03000C
+		  
+/* Error while writing directory block for /lost+found */
+#define PR_3_ERR_LPF_WRITE_BLOCK	0x03000D
+
+/* Error while adjusting inode count */
+#define PR_3_ADJUST_INODE		0x03000E
+
+/* Couldn't fix parent directory -- error */
+#define PR_3_FIX_PARENT_ERR		0x03000F
+
+/* Couldn't fix parent directory -- couldn't find it */	  
+#define PR_3_FIX_PARENT_NOFIND		0x030010
+	
+/* Error allocating inode bitmap */
+#define PR_3_ALLOCATE_IBITMAP_ERROR	0x030011
+		  
+/* Error creating root directory */
+#define PR_3_CREATE_ROOT_ERROR		0x030012
+
+/* Error creating lost and found directory */
+#define PR_3_CREATE_LPF_ERROR		0x030013
+
+/* Root inode is not directory; aborting */
+#define PR_3_ROOT_NOT_DIR_ABORT		0x030014
+
+/* Cannot proceed without a root inode. */
+#define PR_3_NO_ROOT_INODE_ABORT	0x030015
+
+/* Internal error: couldn't find dir_info */
+#define PR_3_NO_DIRINFO			0x030016
+
+/* Lost+found is not a directory */
+#define PR_3_LPF_NOTDIR			0x030017
+
+/*
+ * Pass 3a --- rehashing diretories
+ */
+/* Pass 3a: Reindexing directories */
+#define PR_3A_PASS_HEADER		0x031000
+
+/* Error iterating over directories */
+#define PR_3A_OPTIMIZE_ITER		0x031001
+
+/* Error rehash directory */
+#define PR_3A_OPTIMIZE_DIR_ERR		0x031002		
+
+/* Rehashing dir header */
+#define PR_3A_OPTIMIZE_DIR_HEADER		0x031003
+
+/* Rehashing directory %d */
+#define PR_3A_OPTIMIZE_DIR		0x031004
+		  
+/* Rehashing dir end */	  
+#define PR_3A_OPTIMIZE_DIR_END		0x031005
+
+/*
+ * Pass 4 errors
+ */
+
+/* Pass 4: Checking reference counts */
+#define PR_4_PASS_HEADER	0x040000
+
+/* Unattached zero-length inode */
+#define PR_4_ZERO_LEN_INODE	0x040001
+
+/* Unattached inode */
+#define PR_4_UNATTACHED_INODE	0x040002
+
+/* Inode ref count wrong */
+#define PR_4_BAD_REF_COUNT	0x040003
+
+/* Inconsistent inode count information cached */
+#define PR_4_INCONSISTENT_COUNT	0x040004
+
+/*
+ * Pass 5 errors
+ */
+
+/* Pass 5: Checking group summary information */
+#define PR_5_PASS_HEADER		0x050000
+	
+/* Padding at end of inode bitmap is not set. */
+#define PR_5_INODE_BMAP_PADDING		0x050001
+
+/* Padding at end of block bitmap is not set. */
+#define PR_5_BLOCK_BMAP_PADDING		0x050002
+
+/* Block bitmap differences header */
+#define PR_5_BLOCK_BITMAP_HEADER 	0x050003
+
+/* Block not used, but marked in bitmap */
+#define PR_5_BLOCK_UNUSED		0x050004
+	  
+/* Block used, but not marked used in bitmap */
+#define PR_5_BLOCK_USED			0x050005
+
+/* Block bitmap differences end */	  
+#define PR_5_BLOCK_BITMAP_END		0x050006
+
+/* Inode bitmap differences header */
+#define PR_5_INODE_BITMAP_HEADER	0x050007
+
+/* Inode not used, but marked in bitmap */
+#define PR_5_INODE_UNUSED		0x050008
+	  
+/* Inode used, but not marked used in bitmap */
+#define PR_5_INODE_USED			0x050009
+
+/* Inode bitmap differences end */	  
+#define PR_5_INODE_BITMAP_END		0x05000A
+
+/* Free inodes count for group wrong */
+#define PR_5_FREE_INODE_COUNT_GROUP	0x05000B
+
+/* Directories count for group wrong */
+#define PR_5_FREE_DIR_COUNT_GROUP	0x05000C
+
+/* Free inodes count wrong */
+#define PR_5_FREE_INODE_COUNT	0x05000D
+
+/* Free blocks count for group wrong */
+#define PR_5_FREE_BLOCK_COUNT_GROUP	0x05000E
+
+/* Free blocks count wrong */
+#define PR_5_FREE_BLOCK_COUNT		0x05000F
+
+/* Programming error: bitmap endpoints don't match */
+#define PR_5_BMAP_ENDPOINTS		0x050010
+
+/* Internal error: fudging end of bitmap */
+#define PR_5_FUDGE_BITMAP_ERROR		0x050011
+
+/* Error copying in replacement inode bitmap */
+#define PR_5_COPY_IBITMAP_ERROR		0x050012
+
+/* Error copying in replacement block bitmap */
+#define PR_5_COPY_BBITMAP_ERROR		0x050013
+
+/* Block range not used, but marked in bitmap */
+#define PR_5_BLOCK_RANGE_UNUSED		0x050014
+	  
+/* Block range used, but not marked used in bitmap */
+#define PR_5_BLOCK_RANGE_USED		0x050015
+
+/* Inode range not used, but marked in bitmap */
+#define PR_5_INODE_RANGE_UNUSED		0x050016
+	  
+/* Inode rangeused, but not marked used in bitmap */
+#define PR_5_INODE_RANGE_USED		0x050017
+
+/*
+ * Function declarations
+ */
+int fix_problem(e2fsck_t ctx, problem_t code, struct problem_context *pctx);
+int end_problem_latch(e2fsck_t ctx, int mask);
+int set_latch_flags(int mask, int setflags, int clearflags);
+int get_latch_flags(int mask, int *value);
+void clear_problem_context(struct problem_context *ctx);
+
+/* message.c */
+void print_e2fsck_message(e2fsck_t ctx, const char *msg,
+			  struct problem_context *pctx, int first);
+

e2fsprogs/e2fsck/problemP.h

@@ -0,0 +1,42 @@
+/*
+ * problemP.h --- Private header file for fix_problem()
+ *
+ * Copyright 1997 by Theodore Ts'o
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ */
+
+struct e2fsck_problem {
+	problem_t	e2p_code;
+	const char *	e2p_description;
+	char		prompt;
+	int		flags;
+	problem_t	second_code;
+};
+
+struct latch_descr {
+	int		latch_code;
+	problem_t	question;
+	problem_t	end_message;
+	int		flags;
+};
+
+#define PR_PREEN_OK	0x000001 /* Don't need to do preenhalt */
+#define PR_NO_OK	0x000002 /* If user answers no, don't make fs invalid */
+#define PR_NO_DEFAULT	0x000004 /* Default to no */
+#define PR_MSG_ONLY	0x000008 /* Print message only */
+
+/* Bit positions 0x000ff0 are reserved for the PR_LATCH flags */
+
+#define PR_FATAL	0x001000 /* Fatal error */
+#define PR_AFTER_CODE	0x002000 /* After asking the first question, */
+				 /* ask another */
+#define PR_PREEN_NOMSG	0x004000 /* Don't print a message if we're preening */
+#define PR_NOCOLLATE	0x008000 /* Don't collate answers for this latch */
+#define PR_NO_NOMSG	0x010000 /* Don't print a message if e2fsck -n */
+#define PR_PREEN_NO	0x020000 /* Use No as an answer if preening */
+#define PR_PREEN_NOHDR	0x040000 /* Don't print the preen header */
+

e2fsprogs/e2fsck/recovery.c

@@ -0,0 +1,586 @@
+/*
+ * linux/fs/recovery.c
+ * 
+ * Written by Stephen C. Tweedie <sct@redhat.com>, 1999
+ *
+ * Copyright 1999-2000 Red Hat Software --- All Rights Reserved
+ *
+ * This file is part of the Linux kernel and is made available under
+ * the terms of the GNU General Public License, version 2, or at your
+ * option, any later version, incorporated herein by reference.
+ *
+ * Journal recovery routines for the generic filesystem journaling code;
+ * part of the ext2fs journaling system.  
+ */
+
+#ifndef __KERNEL__
+#include "jfs_user.h"
+#else
+#include <linux/time.h>
+#include <linux/fs.h>
+#include <linux/jbd.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+#endif
+
+/*
+ * Maintain information about the progress of the recovery job, so that
+ * the different passes can carry information between them. 
+ */
+struct recovery_info 
+{
+	tid_t		start_transaction;	
+	tid_t		end_transaction;
+	
+	int		nr_replays;
+	int		nr_revokes;
+	int		nr_revoke_hits;
+};
+
+enum passtype {PASS_SCAN, PASS_REVOKE, PASS_REPLAY};
+static int do_one_pass(journal_t *journal,
+				struct recovery_info *info, enum passtype pass);
+static int scan_revoke_records(journal_t *, struct buffer_head *,
+				tid_t, struct recovery_info *);
+
+#ifdef __KERNEL__
+
+/* Release readahead buffers after use */
+void journal_brelse_array(struct buffer_head *b[], int n)
+{
+	while (--n >= 0)
+		brelse (b[n]);
+}
+
+
+/*
+ * When reading from the journal, we are going through the block device
+ * layer directly and so there is no readahead being done for us.  We
+ * need to implement any readahead ourselves if we want it to happen at
+ * all.  Recovery is basically one long sequential read, so make sure we
+ * do the IO in reasonably large chunks.
+ *
+ * This is not so critical that we need to be enormously clever about
+ * the readahead size, though.  128K is a purely arbitrary, good-enough
+ * fixed value.
+ */
+
+#define MAXBUF 8
+static int do_readahead(journal_t *journal, unsigned int start)
+{
+	int err;
+	unsigned int max, nbufs, next;
+	unsigned long blocknr;
+	struct buffer_head *bh;
+	
+	struct buffer_head * bufs[MAXBUF];
+	
+	/* Do up to 128K of readahead */
+	max = start + (128 * 1024 / journal->j_blocksize);
+	if (max > journal->j_maxlen)
+		max = journal->j_maxlen;
+
+	/* Do the readahead itself.  We'll submit MAXBUF buffer_heads at
+	 * a time to the block device IO layer. */
+	
+	nbufs = 0;
+	
+	for (next = start; next < max; next++) {
+		err = journal_bmap(journal, next, &blocknr);
+
+		if (err) {
+			printk (KERN_ERR "JBD: bad block at offset %u\n",
+				next);
+			goto failed;
+		}
+
+		bh = __getblk(journal->j_dev, blocknr, journal->j_blocksize);
+		if (!bh) {
+			err = -ENOMEM;
+			goto failed;
+		}
+
+		if (!buffer_uptodate(bh) && !buffer_locked(bh)) {
+			bufs[nbufs++] = bh;
+			if (nbufs == MAXBUF) {
+				ll_rw_block(READ, nbufs, bufs);
+				journal_brelse_array(bufs, nbufs);
+				nbufs = 0;
+			}
+		} else
+			brelse(bh);
+	}
+
+	if (nbufs)
+		ll_rw_block(READ, nbufs, bufs);
+	err = 0;
+
+failed:	
+	if (nbufs) 
+		journal_brelse_array(bufs, nbufs);
+	return err;
+}
+
+#endif /* __KERNEL__ */
+
+
+/*
+ * Read a block from the journal
+ */
+
+static int jread(struct buffer_head **bhp, journal_t *journal, 
+		 unsigned int offset)
+{
+	int err;
+	unsigned long blocknr;
+	struct buffer_head *bh;
+
+	*bhp = NULL;
+
+	J_ASSERT (offset < journal->j_maxlen);
+	
+	err = journal_bmap(journal, offset, &blocknr);
+
+	if (err) {
+		printk (KERN_ERR "JBD: bad block at offset %u\n",
+			offset);
+		return err;
+	}
+
+	bh = __getblk(journal->j_dev, blocknr, journal->j_blocksize);
+	if (!bh)
+		return -ENOMEM;
+
+	if (!buffer_uptodate(bh)) {
+		/* If this is a brand new buffer, start readahead.
+                   Otherwise, we assume we are already reading it.  */
+		if (!buffer_req(bh))
+			do_readahead(journal, offset);
+		wait_on_buffer(bh);
+	}
+
+	if (!buffer_uptodate(bh)) {
+		printk (KERN_ERR "JBD: Failed to read block at offset %u\n",
+			offset);
+		brelse(bh);
+		return -EIO;
+	}
+
+	*bhp = bh;
+	return 0;
+}
+
+
+/*
+ * Count the number of in-use tags in a journal descriptor block.
+ */
+
+static int count_tags(struct buffer_head *bh, int size)
+{
+	char *			tagp;
+	journal_block_tag_t *	tag;
+	int			nr = 0;
+
+	tagp = &bh->b_data[sizeof(journal_header_t)];
+
+	while ((tagp - bh->b_data + sizeof(journal_block_tag_t)) <= size) {
+		tag = (journal_block_tag_t *) tagp;
+
+		nr++;
+		tagp += sizeof(journal_block_tag_t);
+		if (!(tag->t_flags & htonl(JFS_FLAG_SAME_UUID)))
+			tagp += 16;
+
+		if (tag->t_flags & htonl(JFS_FLAG_LAST_TAG))
+			break;
+	}
+
+	return nr;
+}
+
+
+/* Make sure we wrap around the log correctly! */
+#define wrap(journal, var)						\
+do {									\
+	if (var >= (journal)->j_last)					\
+		var -= ((journal)->j_last - (journal)->j_first);	\
+} while (0)
+
+/**
+ * int journal_recover(journal_t *journal) - recovers a on-disk journal
+ * @journal: the journal to recover
+ *
+ * The primary function for recovering the log contents when mounting a
+ * journaled device.  
+ * 
+ * Recovery is done in three passes.  In the first pass, we look for the
+ * end of the log.  In the second, we assemble the list of revoke
+ * blocks.  In the third and final pass, we replay any un-revoked blocks
+ * in the log.  
+ */
+int journal_recover(journal_t *journal)
+{
+	int			err;
+	journal_superblock_t *	sb;
+
+	struct recovery_info	info;
+	
+	memset(&info, 0, sizeof(info));
+	sb = journal->j_superblock;
+	
+	/* 
+	 * The journal superblock's s_start field (the current log head)
+	 * is always zero if, and only if, the journal was cleanly
+	 * unmounted.  
+	 */
+
+	if (!sb->s_start) {
+		jbd_debug(1, "No recovery required, last transaction %d\n",
+			  ntohl(sb->s_sequence));
+		journal->j_transaction_sequence = ntohl(sb->s_sequence) + 1;
+		return 0;
+	}
+	
+	err = do_one_pass(journal, &info, PASS_SCAN);
+	if (!err)
+		err = do_one_pass(journal, &info, PASS_REVOKE);
+	if (!err)
+		err = do_one_pass(journal, &info, PASS_REPLAY);
+
+	jbd_debug(0, "JBD: recovery, exit status %d, "
+		  "recovered transactions %u to %u\n",
+		  err, info.start_transaction, info.end_transaction);
+	jbd_debug(0, "JBD: Replayed %d and revoked %d/%d blocks\n", 
+		  info.nr_replays, info.nr_revoke_hits, info.nr_revokes);
+
+	/* Restart the log at the next transaction ID, thus invalidating
+	 * any existing commit records in the log. */
+	journal->j_transaction_sequence = ++info.end_transaction;
+		
+	journal_clear_revoke(journal);
+	sync_blockdev(journal->j_fs_dev);
+	return err;
+}
+
+/**
+ * int journal_skip_recovery() - Start journal and wipe exiting records
+ * @journal: journal to startup
+ *
+ * Locate any valid recovery information from the journal and set up the
+ * journal structures in memory to ignore it (presumably because the
+ * caller has evidence that it is out of date).  
+ * This function does'nt appear to be exorted..
+ *
+ * We perform one pass over the journal to allow us to tell the user how
+ * much recovery information is being erased, and to let us initialise
+ * the journal transaction sequence numbers to the next unused ID. 
+ */
+int journal_skip_recovery(journal_t *journal)
+{
+	int			err;
+	journal_superblock_t *	sb;
+
+	struct recovery_info	info;
+	
+	memset (&info, 0, sizeof(info));
+	sb = journal->j_superblock;
+	
+	err = do_one_pass(journal, &info, PASS_SCAN);
+
+	if (err) {
+		printk(KERN_ERR "JBD: error %d scanning journal\n", err);
+		++journal->j_transaction_sequence;
+	} else {
+#ifdef __CONFIG_JBD_DEBUG__E2FS
+		int dropped = info.end_transaction - ntohl(sb->s_sequence);
+#endif
+		jbd_debug(0, 
+			  "JBD: ignoring %d transaction%s from the journal.\n",
+			  dropped, (dropped == 1) ? "" : "s");
+		journal->j_transaction_sequence = ++info.end_transaction;
+	}
+
+	journal->j_tail = 0;
+	return err;
+}
+
+static int do_one_pass(journal_t *journal,
+			struct recovery_info *info, enum passtype pass)
+{
+	unsigned int		first_commit_ID, next_commit_ID;
+	unsigned long		next_log_block;
+	int			err, success = 0;
+	journal_superblock_t *	sb;
+	journal_header_t * 	tmp;
+	struct buffer_head *	bh;
+	unsigned int		sequence;
+	int			blocktype;
+	
+	/* Precompute the maximum metadata descriptors in a descriptor block */
+	int			MAX_BLOCKS_PER_DESC;
+	MAX_BLOCKS_PER_DESC = ((journal->j_blocksize-sizeof(journal_header_t))
+			       / sizeof(journal_block_tag_t));
+
+	/* 
+	 * First thing is to establish what we expect to find in the log
+	 * (in terms of transaction IDs), and where (in terms of log
+	 * block offsets): query the superblock.  
+	 */
+
+	sb = journal->j_superblock;
+	next_commit_ID = ntohl(sb->s_sequence);
+	next_log_block = ntohl(sb->s_start);
+
+	first_commit_ID = next_commit_ID;
+	if (pass == PASS_SCAN)
+		info->start_transaction = first_commit_ID;
+
+	jbd_debug(1, "Starting recovery pass %d\n", pass);
+
+	/*
+	 * Now we walk through the log, transaction by transaction,
+	 * making sure that each transaction has a commit block in the
+	 * expected place.  Each complete transaction gets replayed back
+	 * into the main filesystem. 
+	 */
+
+	while (1) {
+		int			flags;
+		char *			tagp;
+		journal_block_tag_t *	tag;
+		struct buffer_head *	obh;
+		struct buffer_head *	nbh;
+		
+		/* If we already know where to stop the log traversal,
+		 * check right now that we haven't gone past the end of
+		 * the log. */
+		
+		if (pass != PASS_SCAN)
+			if (tid_geq(next_commit_ID, info->end_transaction))
+				break;
+
+		jbd_debug(2, "Scanning for sequence ID %u at %lu/%lu\n",
+			  next_commit_ID, next_log_block, journal->j_last);
+
+		/* Skip over each chunk of the transaction looking
+		 * either the next descriptor block or the final commit
+		 * record. */
+		
+		jbd_debug(3, "JBD: checking block %ld\n", next_log_block);
+		err = jread(&bh, journal, next_log_block);
+		if (err)
+			goto failed;
+
+		next_log_block++;
+		wrap(journal, next_log_block);
+		
+		/* What kind of buffer is it? 
+		 * 
+		 * If it is a descriptor block, check that it has the
+		 * expected sequence number.  Otherwise, we're all done
+		 * here. */
+
+		tmp = (journal_header_t *)bh->b_data;
+		
+		if (tmp->h_magic != htonl(JFS_MAGIC_NUMBER)) {
+			brelse(bh);
+			break;
+		}
+
+		blocktype = ntohl(tmp->h_blocktype);
+		sequence = ntohl(tmp->h_sequence);
+		jbd_debug(3, "Found magic %d, sequence %d\n", 
+			  blocktype, sequence);
+		
+		if (sequence != next_commit_ID) {
+			brelse(bh);
+			break;
+		}
+		
+		/* OK, we have a valid descriptor block which matches
+		 * all of the sequence number checks.  What are we going
+		 * to do with it?  That depends on the pass... */
+
+		switch(blocktype) {
+		case JFS_DESCRIPTOR_BLOCK:
+			/* If it is a valid descriptor block, replay it
+			 * in pass REPLAY; otherwise, just skip over the
+			 * blocks it describes. */
+			if (pass != PASS_REPLAY) {
+				next_log_block +=
+					count_tags(bh, journal->j_blocksize);
+				wrap(journal, next_log_block);
+				brelse(bh);
+				continue;
+			}
+
+			/* A descriptor block: we can now write all of
+			 * the data blocks.  Yay, useful work is finally
+			 * getting done here! */
+
+			tagp = &bh->b_data[sizeof(journal_header_t)];
+			while ((tagp - bh->b_data +sizeof(journal_block_tag_t))
+			       <= journal->j_blocksize) {
+				unsigned long io_block;
+
+				tag = (journal_block_tag_t *) tagp;
+				flags = ntohl(tag->t_flags);
+				
+				io_block = next_log_block++;
+				wrap(journal, next_log_block);
+				err = jread(&obh, journal, io_block);
+				if (err) {
+					/* Recover what we can, but
+					 * report failure at the end. */
+					success = err;
+					printk (KERN_ERR 
+						"JBD: IO error %d recovering "
+						"block %ld in log\n",
+						err, io_block);
+				} else {
+					unsigned long blocknr;
+					
+					J_ASSERT(obh != NULL);
+					blocknr = ntohl(tag->t_blocknr);
+
+					/* If the block has been
+					 * revoked, then we're all done
+					 * here. */
+					if (journal_test_revoke
+					    (journal, blocknr, 
+					     next_commit_ID)) {
+						brelse(obh);
+						++info->nr_revoke_hits;
+						goto skip_write;
+					}
+								
+					/* Find a buffer for the new
+					 * data being restored */
+					nbh = __getblk(journal->j_fs_dev, 
+						       blocknr,
+						     journal->j_blocksize);
+					if (nbh == NULL) {
+						printk(KERN_ERR 
+						       "JBD: Out of memory "
+						       "during recovery.\n");
+						err = -ENOMEM;
+						brelse(bh);
+						brelse(obh);
+						goto failed;
+					}
+
+					lock_buffer(nbh);
+					memcpy(nbh->b_data, obh->b_data,
+							journal->j_blocksize);
+					if (flags & JFS_FLAG_ESCAPE) {
+						*((unsigned int *)bh->b_data) =
+							htonl(JFS_MAGIC_NUMBER);
+					}
+
+					BUFFER_TRACE(nbh, "marking dirty");
+					set_buffer_uptodate(nbh);
+					mark_buffer_dirty(nbh);
+					BUFFER_TRACE(nbh, "marking uptodate");
+					++info->nr_replays;
+					/* ll_rw_block(WRITE, 1, &nbh); */
+					unlock_buffer(nbh);
+					brelse(obh);
+					brelse(nbh);
+				}
+				
+			skip_write:
+				tagp += sizeof(journal_block_tag_t);
+				if (!(flags & JFS_FLAG_SAME_UUID))
+					tagp += 16;
+
+				if (flags & JFS_FLAG_LAST_TAG)
+					break;
+			}
+			
+			brelse(bh);
+			continue;
+
+		case JFS_COMMIT_BLOCK:
+			/* Found an expected commit block: not much to
+			 * do other than move on to the next sequence
+			 * number. */
+			brelse(bh);
+			next_commit_ID++;
+			continue;
+
+		case JFS_REVOKE_BLOCK:
+			/* If we aren't in the REVOKE pass, then we can
+			 * just skip over this block. */
+			if (pass != PASS_REVOKE) {
+				brelse(bh);
+				continue;
+			}
+
+			err = scan_revoke_records(journal, bh,
+						  next_commit_ID, info);
+			brelse(bh);
+			if (err)
+				goto failed;
+			continue;
+
+		default:
+			jbd_debug(3, "Unrecognised magic %d, end of scan.\n",
+				  blocktype);
+			goto done;
+		}
+	}
+
+ done:
+	/* 
+	 * We broke out of the log scan loop: either we came to the
+	 * known end of the log or we found an unexpected block in the
+	 * log.  If the latter happened, then we know that the "current"
+	 * transaction marks the end of the valid log.
+	 */
+	
+	if (pass == PASS_SCAN)
+		info->end_transaction = next_commit_ID;
+	else {
+		/* It's really bad news if different passes end up at
+		 * different places (but possible due to IO errors). */
+		if (info->end_transaction != next_commit_ID) {
+			printk (KERN_ERR "JBD: recovery pass %d ended at "
+				"transaction %u, expected %u\n",
+				pass, next_commit_ID, info->end_transaction);
+			if (!success)
+				success = -EIO;
+		}
+	}
+
+	return success;
+
+ failed:
+	return err;
+}
+
+
+/* Scan a revoke record, marking all blocks mentioned as revoked. */
+
+static int scan_revoke_records(journal_t *journal, struct buffer_head *bh, 
+			       tid_t sequence, struct recovery_info *info)
+{
+	journal_revoke_header_t *header;
+	int offset, max;
+
+	header = (journal_revoke_header_t *) bh->b_data;
+	offset = sizeof(journal_revoke_header_t);
+	max = ntohl(header->r_count);
+	
+	while (offset < max) {
+		unsigned long blocknr;
+		int err;
+		
+		blocknr = ntohl(* ((unsigned int *) (bh->b_data+offset)));
+		offset += 4;
+		err = journal_set_revoke(journal, blocknr, sequence);
+		if (err)
+			return err;
+		++info->nr_revokes;
+	}
+	return 0;
+}

e2fsprogs/e2fsck/region.c

@@ -0,0 +1,204 @@
+/*
+ * region.c --- code which manages allocations within a region.
+ * 
+ * Copyright (C) 2001 Theodore Ts'o.
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ */
+
+#if HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+#include <string.h>
+
+#include "e2fsck.h"
+
+struct region_el {
+	region_addr_t	start;
+	region_addr_t	end;
+	struct region_el *next;
+};
+
+struct region_struct {
+	region_addr_t	min;
+	region_addr_t	max;
+	struct region_el *allocated;
+};
+
+region_t region_create(region_addr_t min, region_addr_t max)
+{
+	region_t	region;
+
+	region = malloc(sizeof(struct region_struct));
+	if (!region)
+		return NULL;
+	memset(region, 0, sizeof(struct region_struct));
+	region->min = min;
+	region->max = max;
+	return region;
+}
+
+void region_free(region_t region)
+{
+	struct region_el	*r, *next;
+
+	for (r = region->allocated; r; r = next) {
+		next = r->next;
+		free(r);
+	}
+	memset(region, 0, sizeof(struct region_struct));
+	free(region);
+}
+
+int region_allocate(region_t region, region_addr_t start, int n)
+{
+	struct region_el	*r, *new_region, *prev, *next;
+	region_addr_t end;
+
+	end = start+n;
+	if ((start < region->min) || (end > region->max))
+		return -1;
+	if (n == 0)
+		return 1;
+
+	/*
+	 * Search through the linked list.  If we find that it
+	 * conflicts witih something that's already allocated, return
+	 * 1; if we can find an existing region which we can grow, do
+	 * so.  Otherwise, stop when we find the appropriate place
+	 * insert a new region element into the linked list.
+	 */
+	for (r = region->allocated, prev=NULL; r; prev = r, r = r->next) {
+		if (((start >= r->start) && (start < r->end)) ||
+		    ((end > r->start) && (end <= r->end)) ||
+		    ((start <= r->start) && (end >= r->end)))
+			return 1;
+		if (end == r->start) {
+			r->start = start;
+			return 0;
+		}
+		if (start == r->end) {
+			if ((next = r->next)) {
+				if (end > next->start)
+					return 1;
+				if (end == next->start) {
+					r->end = next->end;
+					r->next = next->next;
+					free(next);
+					return 0;
+				}
+			}
+			r->end = end;
+			return 0;
+		}
+		if (start < r->start)
+			break;
+	}
+	/*
+	 * Insert a new region element structure into the linked list
+	 */
+	new_region = malloc(sizeof(struct region_el));
+	if (!new_region)
+		return -1;
+	new_region->start = start;
+	new_region->end = start + n;
+	new_region->next = r;
+	if (prev)
+		prev->next = new_region;
+	else
+		region->allocated = new_region;
+	return 0;
+}
+
+#ifdef TEST_PROGRAM
+#include <stdio.h>
+
+#define BCODE_END	0
+#define BCODE_CREATE	1
+#define BCODE_FREE	2
+#define BCODE_ALLOCATE	3
+#define BCODE_PRINT	4
+
+int bcode_program[] = {
+	BCODE_CREATE, 1, 1001,
+	BCODE_PRINT,
+	BCODE_ALLOCATE, 10, 10,
+	BCODE_ALLOCATE, 30, 10,
+	BCODE_PRINT,
+	BCODE_ALLOCATE, 1, 15,
+	BCODE_ALLOCATE, 15, 8,
+	BCODE_ALLOCATE, 1, 20,
+	BCODE_ALLOCATE, 1, 8,
+	BCODE_PRINT,
+	BCODE_ALLOCATE, 40, 10,
+	BCODE_PRINT,
+	BCODE_ALLOCATE, 22, 5,
+	BCODE_PRINT,
+	BCODE_ALLOCATE, 27, 3,
+	BCODE_PRINT,
+	BCODE_ALLOCATE, 20, 2,
+	BCODE_PRINT,
+	BCODE_ALLOCATE, 49, 1,
+	BCODE_ALLOCATE, 50, 5,
+	BCODE_ALLOCATE, 9, 2,
+	BCODE_ALLOCATE, 9, 1,
+	BCODE_PRINT,
+	BCODE_FREE,
+	BCODE_END
+};
+
+void region_print(region_t region, FILE *f)
+{
+	struct region_el	*r;
+	int	i = 0;
+	
+	fprintf(f, "Printing region (min=%d. max=%d)\n\t", region->min,
+		region->max);
+	for (r = region->allocated; r; r = r->next) {
+		fprintf(f, "(%d, %d)  ", r->start, r->end);
+		if (++i >= 8)
+			fprintf(f, "\n\t");
+	}
+	fprintf(f, "\n");
+}
+
+int main(int argc, char **argv)
+{
+	region_t	r;
+	int		pc = 0, ret;
+	region_addr_t	start, end, len;
+
+	
+	while (1) {
+		switch (bcode_program[pc++]) {
+		case BCODE_END:
+			exit(0);
+		case BCODE_CREATE:
+			start = bcode_program[pc++];
+			end = bcode_program[pc++];
+			printf("Creating region with args(%d, %d)\n",
+			       start, end);
+			r = region_create(start, end);
+			if (!r) {
+				fprintf(stderr, "Couldn't create region.\n");
+				exit(1);
+			}
+			break;
+		case BCODE_ALLOCATE:
+			start = bcode_program[pc++];
+			end = bcode_program[pc++];
+			ret = region_allocate(r, start, end);
+			printf("Region_allocate(%d, %d) returns %d\n",
+			       start, end, ret);
+			break;
+		case BCODE_PRINT:
+			region_print(r, stdout);
+			break;
+		}
+	}
+}
+
+#endif /* TEST_PROGRAM */

e2fsprogs/e2fsck/rehash.c

@@ -0,0 +1,840 @@
+/*
+ * rehash.c --- rebuild hash tree directories
+ * 
+ * Copyright (C) 2002 Theodore Ts'o
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ * 
+ * This algorithm is designed for simplicity of implementation and to
+ * pack the directory as much as possible.  It however requires twice
+ * as much memory as the size of the directory.  The maximum size
+ * directory supported using a 4k blocksize is roughly a gigabyte, and
+ * so there may very well be problems with machines that don't have
+ * virtual memory, and obscenely large directories.
+ *
+ * An alternate algorithm which is much more disk intensive could be
+ * written, and probably will need to be written in the future.  The
+ * design goals of such an algorithm are: (a) use (roughly) constant
+ * amounts of memory, no matter how large the directory, (b) the
+ * directory must be safe at all times, even if e2fsck is interrupted
+ * in the middle, (c) we must use minimal amounts of extra disk
+ * blocks.  This pretty much requires an incremental approach, where
+ * we are reading from one part of the directory, and inserting into
+ * the front half.  So the algorithm will have to keep track of a
+ * moving block boundary between the new tree and the old tree, and
+ * files will need to be moved from the old directory and inserted
+ * into the new tree.  If the new directory requires space which isn't
+ * yet available, blocks from the beginning part of the old directory
+ * may need to be moved to the end of the directory to make room for
+ * the new tree:
+ *
+ *    --------------------------------------------------------
+ *    |  new tree   |        | old tree                      |
+ *    --------------------------------------------------------
+ *                  ^ ptr    ^ptr
+ *                tail new   head old
+ * 
+ * This is going to be a pain in the tuckus to implement, and will
+ * require a lot more disk accesses.  So I'm going to skip it for now;
+ * it's only really going to be an issue for really, really big
+ * filesystems (when we reach the level of tens of millions of files
+ * in a single directory).  It will probably be easier to simply
+ * require that e2fsck use VM first.
+ */
+
+#include <string.h>
+#include <ctype.h>
+#include <errno.h>
+#include "e2fsck.h"
+#include "problem.h"
+
+struct fill_dir_struct {
+	char *buf;
+	struct ext2_inode *inode;
+	int err;
+	e2fsck_t ctx;
+	struct hash_entry *harray;
+	int max_array, num_array;
+	int dir_size;
+	int compress;
+	ino_t parent;
+};
+
+struct hash_entry {
+	ext2_dirhash_t	hash;
+	ext2_dirhash_t	minor_hash;
+	struct ext2_dir_entry	*dir;
+};
+
+struct out_dir {
+	int		num;
+	int		max;
+	char		*buf;
+	ext2_dirhash_t	*hashes;
+};
+
+static int fill_dir_block(ext2_filsys fs,
+			  blk_t	*block_nr,
+			  e2_blkcnt_t blockcnt,
+			  blk_t ref_block EXT2FS_ATTR((unused)),
+			  int ref_offset EXT2FS_ATTR((unused)),
+			  void *priv_data)
+{
+	struct fill_dir_struct	*fd = (struct fill_dir_struct *) priv_data;
+	struct hash_entry 	*new_array, *ent;
+	struct ext2_dir_entry 	*dirent;
+	char			*dir;
+	unsigned int		offset, dir_offset;
+	
+	if (blockcnt < 0)
+		return 0;
+
+	offset = blockcnt * fs->blocksize;
+	if (offset + fs->blocksize > fd->inode->i_size) {
+		fd->err = EXT2_ET_DIR_CORRUPTED;
+		return BLOCK_ABORT;
+	}
+	dir = (fd->buf+offset);
+	if (HOLE_BLKADDR(*block_nr)) {
+		memset(dir, 0, fs->blocksize);
+		dirent = (struct ext2_dir_entry *) dir;
+		dirent->rec_len = fs->blocksize;
+	} else {
+		fd->err = ext2fs_read_dir_block(fs, *block_nr, dir);
+		if (fd->err)
+			return BLOCK_ABORT;
+	}
+	/* While the directory block is "hot", index it. */
+	dir_offset = 0;
+	while (dir_offset < fs->blocksize) {
+		dirent = (struct ext2_dir_entry *) (dir + dir_offset);
+		if (((dir_offset + dirent->rec_len) > fs->blocksize) ||
+		    (dirent->rec_len < 8) ||
+		    ((dirent->rec_len % 4) != 0) ||
+		    (((dirent->name_len & 0xFF)+8) > dirent->rec_len)) {
+			fd->err = EXT2_ET_DIR_CORRUPTED;
+			return BLOCK_ABORT;
+		}
+		dir_offset += dirent->rec_len;
+		if (dirent->inode == 0)
+			continue;
+		if (!fd->compress && ((dirent->name_len&0xFF) == 1) &&
+		    (dirent->name[0] == '.'))
+			continue;
+		if (!fd->compress && ((dirent->name_len&0xFF) == 2) &&
+		    (dirent->name[0] == '.') && (dirent->name[1] == '.')) {
+			fd->parent = dirent->inode;
+			continue;
+		}
+		if (fd->num_array >= fd->max_array) {
+			new_array = realloc(fd->harray,
+			    sizeof(struct hash_entry) * (fd->max_array+500));
+			if (!new_array) {
+				fd->err = ENOMEM;
+				return BLOCK_ABORT;
+			}
+			fd->harray = new_array;
+			fd->max_array += 500;
+		}
+		ent = fd->harray + fd->num_array++;
+		ent->dir = dirent;
+		fd->dir_size += EXT2_DIR_REC_LEN(dirent->name_len & 0xFF);
+		if (fd->compress)
+			ent->hash = ent->minor_hash = 0;
+		else {
+			fd->err = ext2fs_dirhash(fs->super->s_def_hash_version,
+						 dirent->name,
+						 dirent->name_len & 0xFF,
+						 fs->super->s_hash_seed,
+						 &ent->hash, &ent->minor_hash);
+			if (fd->err)
+				return BLOCK_ABORT;
+		}
+	}
+	
+	return 0;
+}
+
+/* Used for sorting the hash entry */
+static EXT2_QSORT_TYPE name_cmp(const void *a, const void *b)
+{
+	const struct hash_entry *he_a = (const struct hash_entry *) a;
+	const struct hash_entry *he_b = (const struct hash_entry *) b;
+	int	ret;
+	int	min_len;
+
+	min_len = he_a->dir->name_len;
+	if (min_len > he_b->dir->name_len)
+		min_len = he_b->dir->name_len;
+
+	ret = strncmp(he_a->dir->name, he_b->dir->name, min_len);
+	if (ret == 0) {
+		if (he_a->dir->name_len > he_b->dir->name_len)
+			ret = 1;
+		else if (he_a->dir->name_len < he_b->dir->name_len)
+			ret = -1;
+		else
+			ret = he_b->dir->inode - he_a->dir->inode;
+	}
+	return ret;
+}
+
+/* Used for sorting the hash entry */
+static EXT2_QSORT_TYPE hash_cmp(const void *a, const void *b)
+{
+	const struct hash_entry *he_a = (const struct hash_entry *) a;
+	const struct hash_entry *he_b = (const struct hash_entry *) b;
+	int	ret;
+	
+	if (he_a->hash > he_b->hash)
+		ret = 1;
+	else if (he_a->hash < he_b->hash)
+		ret = -1;
+	else {
+		if (he_a->minor_hash > he_b->minor_hash)
+			ret = 1;
+		else if (he_a->minor_hash < he_b->minor_hash)
+			ret = -1;
+		else
+			ret = name_cmp(a, b);
+	}
+	return ret;
+}
+
+static errcode_t alloc_size_dir(ext2_filsys fs, struct out_dir *outdir, 
+				int blocks)
+{
+	void			*new_mem;
+
+	if (outdir->max) {
+		new_mem = realloc(outdir->buf, blocks * fs->blocksize);
+		if (!new_mem)
+			return ENOMEM;
+		outdir->buf = new_mem;
+		new_mem = realloc(outdir->hashes,
+				  blocks * sizeof(ext2_dirhash_t));
+		if (!new_mem)
+			return ENOMEM;
+		outdir->hashes = new_mem;
+	} else {
+		outdir->buf = malloc(blocks * fs->blocksize);
+		outdir->hashes = malloc(blocks * sizeof(ext2_dirhash_t));
+		outdir->num = 0;
+	}
+	outdir->max = blocks;
+	return 0;
+}
+
+static void free_out_dir(struct out_dir *outdir)
+{
+	if (outdir->buf)
+		free(outdir->buf);
+	if (outdir->hashes)
+		free(outdir->hashes);
+	outdir->max = 0;
+	outdir->num =0;
+}
+
+static errcode_t get_next_block(ext2_filsys fs, struct out_dir *outdir,
+			 char ** ret)
+{
+	errcode_t	retval;
+
+	if (outdir->num >= outdir->max) {
+		retval = alloc_size_dir(fs, outdir, outdir->max + 50);
+		if (retval)
+			return retval;
+	}
+	*ret = outdir->buf + (outdir->num++ * fs->blocksize);
+	memset(*ret, 0, fs->blocksize);
+	return 0;
+}
+
+/*
+ * This function is used to make a unique filename.  We do this by
+ * appending ~0, and then incrementing the number.  However, we cannot
+ * expand the length of the filename beyond the padding available in
+ * the directory entry.
+ */
+static void mutate_name(char *str, __u16 *len)
+{
+	int	i;
+	__u16	l = *len & 0xFF, h = *len & 0xff00;
+	
+	/*
+	 * First check to see if it looks the name has been mutated
+	 * already
+	 */
+	for (i = l-1; i > 0; i--) {
+		if (!isdigit(str[i]))
+			break;
+	}
+	if ((i == l-1) || (str[i] != '~')) {
+		if (((l-1) & 3) < 2)
+			l += 2;
+		else
+			l = (l+3) & ~3;
+		str[l-2] = '~';
+		str[l-1] = '0';
+		*len = l | h;
+		return;
+	}
+	for (i = l-1; i >= 0; i--) {
+		if (isdigit(str[i])) {
+			if (str[i] == '9')
+				str[i] = '0';
+			else {
+				str[i]++;
+				return;
+			}
+			continue;
+		}
+		if (i == 1) {
+			if (str[0] == 'z')
+				str[0] = 'A';
+			else if (str[0] == 'Z') {
+				str[0] = '~';
+				str[1] = '0';
+			} else
+				str[0]++;
+		} else if (i > 0) {
+			str[i] = '1';
+			str[i-1] = '~';
+		} else {
+			if (str[0] == '~')
+				str[0] = 'a';
+			else 
+				str[0]++;
+		}
+		break;
+	}
+}
+
+static int duplicate_search_and_fix(e2fsck_t ctx, ext2_filsys fs,
+				    ext2_ino_t ino,
+				    struct fill_dir_struct *fd)
+{
+	struct problem_context	pctx;
+	struct hash_entry 	*ent, *prev;
+	int			i, j;
+	int			fixed = 0;
+	char			new_name[256];
+	__u16			new_len;
+	
+	clear_problem_context(&pctx);
+	pctx.ino = ino;
+
+	for (i=1; i < fd->num_array; i++) {
+		ent = fd->harray + i;
+		prev = ent - 1;
+		if (!ent->dir->inode ||
+		    ((ent->dir->name_len & 0xFF) !=
+		     (prev->dir->name_len & 0xFF)) ||
+		    (strncmp(ent->dir->name, prev->dir->name,
+			     ent->dir->name_len & 0xFF)))
+			continue;
+		pctx.dirent = ent->dir;
+		if ((ent->dir->inode == prev->dir->inode) &&
+		    fix_problem(ctx, PR_2_DUPLICATE_DIRENT, &pctx)) {
+			e2fsck_adjust_inode_count(ctx, ent->dir->inode, -1);
+			ent->dir->inode = 0;
+			fixed++;
+			continue;
+		}
+		memcpy(new_name, ent->dir->name, ent->dir->name_len & 0xFF);
+		new_len = ent->dir->name_len;
+		mutate_name(new_name, &new_len);
+		for (j=0; j < fd->num_array; j++) {
+			if ((i==j) ||
+			    ((ent->dir->name_len & 0xFF) !=
+			     (fd->harray[j].dir->name_len & 0xFF)) ||
+			    (strncmp(new_name, fd->harray[j].dir->name,
+				     new_len & 0xFF)))
+				continue;
+			mutate_name(new_name, &new_len);
+			
+			j = -1;
+		}
+		new_name[new_len & 0xFF] = 0;
+		pctx.str = new_name;
+		if (fix_problem(ctx, PR_2_NON_UNIQUE_FILE, &pctx)) {
+			memcpy(ent->dir->name, new_name, new_len & 0xFF);
+			ent->dir->name_len = new_len;
+			ext2fs_dirhash(fs->super->s_def_hash_version,
+				       ent->dir->name,
+				       ent->dir->name_len & 0xFF,
+				       fs->super->s_hash_seed,
+				       &ent->hash, &ent->minor_hash);
+			fixed++;
+		}
+	}
+	return fixed;
+}
+
+
+static errcode_t copy_dir_entries(ext2_filsys fs,
+				  struct fill_dir_struct *fd,
+				  struct out_dir *outdir)
+{
+	errcode_t		retval;
+	char			*block_start;
+	struct hash_entry 	*ent;
+	struct ext2_dir_entry	*dirent;
+	int			i, rec_len, left;
+	ext2_dirhash_t		prev_hash;
+	int			offset;
+	
+	outdir->max = 0;
+	retval = alloc_size_dir(fs, outdir,
+				(fd->dir_size / fs->blocksize) + 2);
+	if (retval)
+		return retval;
+	outdir->num = fd->compress ? 0 : 1;
+	offset = 0;
+	outdir->hashes[0] = 0;
+	prev_hash = 1;
+	if ((retval = get_next_block(fs, outdir, &block_start)))
+		return retval;
+	dirent = (struct ext2_dir_entry *) block_start;
+	left = fs->blocksize;
+	for (i=0; i < fd->num_array; i++) {
+		ent = fd->harray + i;
+		if (ent->dir->inode == 0)
+			continue;
+		rec_len = EXT2_DIR_REC_LEN(ent->dir->name_len & 0xFF);
+		if (rec_len > left) {
+			if (left)
+				dirent->rec_len += left;
+			if ((retval = get_next_block(fs, outdir,
+						      &block_start)))
+				return retval;
+			offset = 0;
+		}
+		left = fs->blocksize - offset;
+		dirent = (struct ext2_dir_entry *) (block_start + offset);
+		if (offset == 0) {
+			if (ent->hash == prev_hash)
+				outdir->hashes[outdir->num-1] = ent->hash | 1;
+			else
+				outdir->hashes[outdir->num-1] = ent->hash;
+		}
+		dirent->inode = ent->dir->inode;
+		dirent->name_len = ent->dir->name_len;
+		dirent->rec_len = rec_len;
+		memcpy(dirent->name, ent->dir->name, dirent->name_len & 0xFF);
+		offset += rec_len;
+		left -= rec_len;
+		if (left < 12) {
+			dirent->rec_len += left;
+			offset += left;
+			left = 0;
+		}
+		prev_hash = ent->hash;
+	}
+	if (left)
+		dirent->rec_len += left;
+
+	return 0;
+}
+
+
+static struct ext2_dx_root_info *set_root_node(ext2_filsys fs, char *buf,
+				    ext2_ino_t ino, ext2_ino_t parent)
+{
+	struct ext2_dir_entry 		*dir;
+	struct ext2_dx_root_info  	*root;
+	struct ext2_dx_countlimit	*limits;
+	int				filetype = 0;
+
+	if (fs->super->s_feature_incompat & EXT2_FEATURE_INCOMPAT_FILETYPE)
+		filetype = EXT2_FT_DIR << 8;
+	
+	memset(buf, 0, fs->blocksize);
+	dir = (struct ext2_dir_entry *) buf;
+	dir->inode = ino;
+	dir->name[0] = '.';
+	dir->name_len = 1 | filetype;
+	dir->rec_len = 12;
+	dir = (struct ext2_dir_entry *) (buf + 12);
+	dir->inode = parent;
+	dir->name[0] = '.';
+	dir->name[1] = '.';
+	dir->name_len = 2 | filetype;
+	dir->rec_len = fs->blocksize - 12;
+	
+	root = (struct ext2_dx_root_info *) (buf+24);
+	root->reserved_zero = 0;
+	root->hash_version = fs->super->s_def_hash_version;
+	root->info_length = 8;
+	root->indirect_levels = 0;
+	root->unused_flags = 0;
+
+	limits = (struct ext2_dx_countlimit *) (buf+32);
+	limits->limit = (fs->blocksize - 32) / sizeof(struct ext2_dx_entry);
+	limits->count = 0;
+
+	return root;
+}
+
+
+static struct ext2_dx_entry *set_int_node(ext2_filsys fs, char *buf)
+{
+	struct ext2_dir_entry 		*dir;
+	struct ext2_dx_countlimit	*limits;
+
+	memset(buf, 0, fs->blocksize);
+	dir = (struct ext2_dir_entry *) buf;
+	dir->inode = 0;
+	dir->rec_len = fs->blocksize;
+	
+	limits = (struct ext2_dx_countlimit *) (buf+8);
+	limits->limit = (fs->blocksize - 8) / sizeof(struct ext2_dx_entry);
+	limits->count = 0;
+
+	return (struct ext2_dx_entry *) limits;
+}
+
+/*
+ * This function takes the leaf nodes which have been written in
+ * outdir, and populates the root node and any necessary interior nodes.
+ */
+static errcode_t calculate_tree(ext2_filsys fs,
+				struct out_dir *outdir,
+				ext2_ino_t ino,
+				ext2_ino_t parent)
+{
+	struct ext2_dx_root_info  	*root_info;
+	struct ext2_dx_entry 		*root, *dx_ent = 0;
+	struct ext2_dx_countlimit	*root_limit, *limit;
+	errcode_t			retval;
+	char				* block_start;
+	int				i, c1, c2, nblks;
+	int				limit_offset, root_offset;
+	
+	root_info = set_root_node(fs, outdir->buf, ino, parent);
+	root_offset = limit_offset = ((char *) root_info - outdir->buf) +
+		root_info->info_length;
+	root_limit = (struct ext2_dx_countlimit *) (outdir->buf + limit_offset);
+	c1 = root_limit->limit;
+	nblks = outdir->num;
+
+	/* Write out the pointer blocks */
+	if (nblks-1 <= c1) {
+		/* Just write out the root block, and we're done */
+		root = (struct ext2_dx_entry *) (outdir->buf + root_offset);
+		for (i=1; i < nblks; i++) {
+			root->block = ext2fs_cpu_to_le32(i);
+			if (i != 1)
+				root->hash =
+					ext2fs_cpu_to_le32(outdir->hashes[i]);
+			root++;
+			c1--;
+		}
+	} else {
+		c2 = 0;
+		limit = 0;
+		root_info->indirect_levels = 1;
+		for (i=1; i < nblks; i++) {
+			if (c1 == 0)
+				return ENOSPC;
+			if (c2 == 0) {
+				if (limit)
+					limit->limit = limit->count = 
+		ext2fs_cpu_to_le16(limit->limit);
+				root = (struct ext2_dx_entry *)
+					(outdir->buf + root_offset);
+				root->block = ext2fs_cpu_to_le32(outdir->num);
+				if (i != 1)
+					root->hash =
+			ext2fs_cpu_to_le32(outdir->hashes[i]);
+				if ((retval =  get_next_block(fs, outdir,
+							      &block_start)))
+					return retval;
+				dx_ent = set_int_node(fs, block_start);
+				limit = (struct ext2_dx_countlimit *) dx_ent;
+				c2 = limit->limit;
+				root_offset += sizeof(struct ext2_dx_entry);
+				c1--;
+			}
+			dx_ent->block = ext2fs_cpu_to_le32(i);
+			if (c2 != limit->limit)
+				dx_ent->hash =
+					ext2fs_cpu_to_le32(outdir->hashes[i]);
+			dx_ent++;
+			c2--;
+		}
+		limit->count = ext2fs_cpu_to_le16(limit->limit - c2);
+		limit->limit = ext2fs_cpu_to_le16(limit->limit);
+	}
+	root_limit = (struct ext2_dx_countlimit *) (outdir->buf + limit_offset);
+	root_limit->count = ext2fs_cpu_to_le16(root_limit->limit - c1);
+	root_limit->limit = ext2fs_cpu_to_le16(root_limit->limit);
+
+	return 0;
+}
+
+struct write_dir_struct {
+	struct out_dir *outdir;
+	errcode_t	err;
+	e2fsck_t	ctx;
+	int		cleared;
+};
+
+/*
+ * Helper function which writes out a directory block.
+ */
+static int write_dir_block(ext2_filsys fs,
+			   blk_t	*block_nr,
+			   e2_blkcnt_t blockcnt,
+			   blk_t ref_block EXT2FS_ATTR((unused)),
+			   int ref_offset EXT2FS_ATTR((unused)), 
+			   void *priv_data)
+{
+	struct write_dir_struct	*wd = (struct write_dir_struct *) priv_data;
+	blk_t	blk;
+	char	*dir;
+
+	if (*block_nr == 0)
+		return 0;
+	if (blockcnt >= wd->outdir->num) {
+		e2fsck_read_bitmaps(wd->ctx);
+		blk = *block_nr;
+		ext2fs_unmark_block_bitmap(wd->ctx->block_found_map, blk);
+		ext2fs_block_alloc_stats(fs, blk, -1);
+		*block_nr = 0;
+		wd->cleared++;
+		return BLOCK_CHANGED;
+	}
+	if (blockcnt < 0)
+		return 0;
+
+	dir = wd->outdir->buf + (blockcnt * fs->blocksize);
+	wd->err = ext2fs_write_dir_block(fs, *block_nr, dir);
+	if (wd->err)
+		return BLOCK_ABORT;
+	return 0;
+}
+
+static errcode_t write_directory(e2fsck_t ctx, ext2_filsys fs,
+				 struct out_dir *outdir,
+				 ext2_ino_t ino, int compress)
+{
+	struct write_dir_struct wd;
+	errcode_t	retval;
+	struct ext2_inode 	inode;
+
+	retval = e2fsck_expand_directory(ctx, ino, -1, outdir->num);
+	if (retval)
+		return retval;
+
+	wd.outdir = outdir;
+	wd.err = 0;
+	wd.ctx = ctx;
+	wd.cleared = 0;
+
+	retval = ext2fs_block_iterate2(fs, ino, 0, 0,
+				       write_dir_block, &wd);
+	if (retval)
+		return retval;
+	if (wd.err)
+		return wd.err;
+
+	e2fsck_read_inode(ctx, ino, &inode, "rehash_dir");
+	if (compress)
+		inode.i_flags &= ~EXT2_INDEX_FL;
+	else
+		inode.i_flags |= EXT2_INDEX_FL;
+	inode.i_size = outdir->num * fs->blocksize;
+	inode.i_blocks -= (fs->blocksize / 512) * wd.cleared;
+	e2fsck_write_inode(ctx, ino, &inode, "rehash_dir");
+
+	return 0;
+}
+
+errcode_t e2fsck_rehash_dir(e2fsck_t ctx, ext2_ino_t ino)
+{
+	ext2_filsys 		fs = ctx->fs;
+	errcode_t		retval;
+	struct ext2_inode 	inode;
+	char			*dir_buf = 0;
+	struct fill_dir_struct	fd;
+	struct out_dir		outdir;
+	
+	outdir.max = outdir.num = 0;
+	outdir.buf = 0;
+	outdir.hashes = 0;
+	e2fsck_read_inode(ctx, ino, &inode, "rehash_dir");
+
+	retval = ENOMEM;
+	fd.harray = 0;
+	dir_buf = malloc(inode.i_size);
+	if (!dir_buf)
+		goto errout;
+
+	fd.max_array = inode.i_size / 32;
+	fd.num_array = 0;
+	fd.harray = malloc(fd.max_array * sizeof(struct hash_entry));
+	if (!fd.harray)
+		goto errout;
+
+	fd.ctx = ctx;
+	fd.buf = dir_buf;
+	fd.inode = &inode;
+	fd.err = 0;
+	fd.dir_size = 0;
+	fd.compress = 0;
+	if (!(fs->super->s_feature_compat & EXT2_FEATURE_COMPAT_DIR_INDEX) ||
+	    (inode.i_size / fs->blocksize) < 2)
+		fd.compress = 1;
+	fd.parent = 0;
+
+	/* Read in the entire directory into memory */
+	retval = ext2fs_block_iterate2(fs, ino, 0, 0,
+				       fill_dir_block, &fd);
+	if (fd.err) {
+		retval = fd.err;
+		goto errout;
+	}
+
+#if 0
+	printf("%d entries (%d bytes) found in inode %d\n",
+	       fd.num_array, fd.dir_size, ino);
+#endif
+
+	/* Sort the list */
+resort:
+	if (fd.compress)
+		qsort(fd.harray+2, fd.num_array-2,
+		      sizeof(struct hash_entry), name_cmp);
+	else
+		qsort(fd.harray, fd.num_array,
+		      sizeof(struct hash_entry), hash_cmp);
+
+	/*
+	 * Look for duplicates
+	 */
+	if (duplicate_search_and_fix(ctx, fs, ino, &fd))
+		goto resort;
+
+	if (ctx->options & E2F_OPT_NO) {
+		retval = 0;
+		goto errout;
+	}
+
+	/*
+	 * Copy the directory entries.  In a htree directory these
+	 * will become the leaf nodes.
+	 */
+	retval = copy_dir_entries(fs, &fd, &outdir);
+	if (retval)
+		goto errout;
+	
+	free(dir_buf); dir_buf = 0;
+
+	if (!fd.compress) {
+		/* Calculate the interior nodes */
+		retval = calculate_tree(fs, &outdir, ino, fd.parent);
+		if (retval)
+			goto errout;
+	}
+	
+	retval = write_directory(ctx, fs, &outdir, ino, fd.compress);
+	if (retval)
+		goto errout;
+
+errout:
+	if (dir_buf)
+		free(dir_buf);
+	if (fd.harray)
+		free(fd.harray);
+
+	free_out_dir(&outdir);
+	return retval;
+}
+
+void e2fsck_rehash_directories(e2fsck_t ctx)
+{
+	struct problem_context	pctx;
+#ifdef RESOURCE_TRACK
+	struct resource_track	rtrack;
+#endif
+	struct dir_info		*dir;
+	ext2_u32_iterate 	iter;
+	ext2_ino_t		ino;
+	errcode_t		retval;
+	int			i, cur, max, all_dirs, dir_index, first = 1;
+
+#ifdef RESOURCE_TRACK
+	init_resource_track(&rtrack);
+#endif
+
+	all_dirs = ctx->options & E2F_OPT_COMPRESS_DIRS;
+
+	if (!ctx->dirs_to_hash && !all_dirs)
+		return;
+
+	e2fsck_get_lost_and_found(ctx, 0);
+		
+	clear_problem_context(&pctx);
+
+	dir_index = ctx->fs->super->s_feature_compat & EXT2_FEATURE_COMPAT_DIR_INDEX;
+	cur = 0;
+	if (all_dirs) {
+		i = 0;
+		max = e2fsck_get_num_dirinfo(ctx);
+	} else {
+		retval = ext2fs_u32_list_iterate_begin(ctx->dirs_to_hash, 
+						       &iter);
+		if (retval) {
+			pctx.errcode = retval;
+			fix_problem(ctx, PR_3A_OPTIMIZE_ITER, &pctx);
+			return;
+		}
+		max = ext2fs_u32_list_count(ctx->dirs_to_hash);
+	}
+	while (1) {
+		if (all_dirs) {
+			if ((dir = e2fsck_dir_info_iter(ctx, &i)) == 0)
+				break;
+			ino = dir->ino;
+		} else {
+			if (!ext2fs_u32_list_iterate(iter, &ino))
+				break;
+		}
+		if (ino == ctx->lost_and_found)
+			continue;
+		pctx.dir = ino;
+		if (first) {
+			fix_problem(ctx, PR_3A_PASS_HEADER, &pctx);
+			first = 0;
+		}
+#if 0
+		fix_problem(ctx, PR_3A_OPTIMIZE_DIR, &pctx);
+#endif
+		pctx.errcode = e2fsck_rehash_dir(ctx, ino);
+		if (pctx.errcode) {
+			end_problem_latch(ctx, PR_LATCH_OPTIMIZE_DIR);
+			fix_problem(ctx, PR_3A_OPTIMIZE_DIR_ERR, &pctx);
+		}
+		if (ctx->progress && !ctx->progress_fd)
+			e2fsck_simple_progress(ctx, "Rebuilding directory",
+			       100.0 * (float) (++cur) / (float) max, ino);
+	}
+	end_problem_latch(ctx, PR_LATCH_OPTIMIZE_DIR);
+	if (!all_dirs)
+		ext2fs_u32_list_iterate_end(iter);
+	
+	if (ctx->dirs_to_hash)
+		ext2fs_u32_list_free(ctx->dirs_to_hash);
+	ctx->dirs_to_hash = 0;
+
+#ifdef RESOURCE_TRACK
+	if (ctx->options & E2F_OPT_TIME2) {
+		e2fsck_clear_progbar(ctx);
+		print_resource_track("Pass 3A", &rtrack);
+	}
+#endif
+}

e2fsprogs/e2fsck/revoke.c

@@ -0,0 +1,640 @@
+/*
+ * linux/fs/revoke.c
+ * 
+ * Written by Stephen C. Tweedie <sct@redhat.com>, 2000
+ *
+ * Copyright 2000 Red Hat corp --- All Rights Reserved
+ *
+ * This file is part of the Linux kernel and is made available under
+ * the terms of the GNU General Public License, version 2, or at your
+ * option, any later version, incorporated herein by reference.
+ *
+ * Journal revoke routines for the generic filesystem journaling code;
+ * part of the ext2fs journaling system.
+ *
+ * Revoke is the mechanism used to prevent old log records for deleted
+ * metadata from being replayed on top of newer data using the same
+ * blocks.  The revoke mechanism is used in two separate places:
+ * 
+ * + Commit: during commit we write the entire list of the current
+ *   transaction's revoked blocks to the journal
+ * 
+ * + Recovery: during recovery we record the transaction ID of all
+ *   revoked blocks.  If there are multiple revoke records in the log
+ *   for a single block, only the last one counts, and if there is a log
+ *   entry for a block beyond the last revoke, then that log entry still
+ *   gets replayed.
+ *
+ * We can get interactions between revokes and new log data within a
+ * single transaction:
+ *
+ * Block is revoked and then journaled:
+ *   The desired end result is the journaling of the new block, so we 
+ *   cancel the revoke before the transaction commits.
+ *
+ * Block is journaled and then revoked:
+ *   The revoke must take precedence over the write of the block, so we
+ *   need either to cancel the journal entry or to write the revoke
+ *   later in the log than the log block.  In this case, we choose the
+ *   latter: journaling a block cancels any revoke record for that block
+ *   in the current transaction, so any revoke for that block in the
+ *   transaction must have happened after the block was journaled and so
+ *   the revoke must take precedence.
+ *
+ * Block is revoked and then written as data: 
+ *   The data write is allowed to succeed, but the revoke is _not_
+ *   cancelled.  We still need to prevent old log records from
+ *   overwriting the new data.  We don't even need to clear the revoke
+ *   bit here.
+ *
+ * Revoke information on buffers is a tri-state value:
+ *
+ * RevokeValid clear:	no cached revoke status, need to look it up
+ * RevokeValid set, Revoked clear:
+ *			buffer has not been revoked, and cancel_revoke
+ *			need do nothing.
+ * RevokeValid set, Revoked set:
+ *			buffer has been revoked.  
+ */
+
+#ifndef __KERNEL__
+#include "jfs_user.h"
+#else
+#include <linux/sched.h>
+#include <linux/fs.h>
+#include <linux/jbd.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <linux/locks.h>
+#include <linux/list.h>
+#include <linux/smp_lock.h>
+#include <linux/init.h>
+#endif
+
+static kmem_cache_t *revoke_record_cache;
+static kmem_cache_t *revoke_table_cache;
+
+/* Each revoke record represents one single revoked block.  During
+   journal replay, this involves recording the transaction ID of the
+   last transaction to revoke this block. */
+
+struct jbd_revoke_record_s 
+{
+	struct list_head  hash;
+	tid_t		  sequence;	/* Used for recovery only */
+	unsigned long	  blocknr;	
+};
+
+
+/* The revoke table is just a simple hash table of revoke records. */
+struct jbd_revoke_table_s
+{
+	/* It is conceivable that we might want a larger hash table
+	 * for recovery.  Must be a power of two. */
+	int		  hash_size; 
+	int		  hash_shift; 
+	struct list_head *hash_table;
+};
+
+
+#ifdef __KERNEL__
+static void write_one_revoke_record(journal_t *, transaction_t *,
+				    struct journal_head **, int *,
+				    struct jbd_revoke_record_s *);
+static void flush_descriptor(journal_t *, struct journal_head *, int);
+#endif
+
+/* Utility functions to maintain the revoke table */
+
+/* Borrowed from buffer.c: this is a tried and tested block hash function */
+static inline int hash(journal_t *journal, unsigned long block)
+{
+	struct jbd_revoke_table_s *table = journal->j_revoke;
+	int hash_shift = table->hash_shift;
+	
+	return ((block << (hash_shift - 6)) ^
+		(block >> 13) ^
+		(block << (hash_shift - 12))) & (table->hash_size - 1);
+}
+
+static int insert_revoke_hash(journal_t *journal, unsigned long blocknr,
+			      tid_t seq)
+{
+	struct list_head *hash_list;
+	struct jbd_revoke_record_s *record;
+
+#ifdef __KERNEL__
+repeat:
+#endif
+	record = kmem_cache_alloc(revoke_record_cache, GFP_NOFS);
+	if (!record)
+		goto oom;
+
+	record->sequence = seq;
+	record->blocknr = blocknr;
+	hash_list = &journal->j_revoke->hash_table[hash(journal, blocknr)];
+	list_add(&record->hash, hash_list);
+	return 0;
+
+oom:
+#ifdef __KERNEL__
+	if (!journal_oom_retry)
+		return -ENOMEM;
+	jbd_debug(1, "ENOMEM in " __FUNCTION__ ", retrying.\n");
+	current->policy |= SCHED_YIELD;
+	schedule();
+	goto repeat;
+#else
+	return -ENOMEM;
+#endif
+}
+
+/* Find a revoke record in the journal's hash table. */
+
+static struct jbd_revoke_record_s *find_revoke_record(journal_t *journal,
+						      unsigned long blocknr)
+{
+	struct list_head *hash_list;
+	struct jbd_revoke_record_s *record;
+	
+	hash_list = &journal->j_revoke->hash_table[hash(journal, blocknr)];
+
+	record = (struct jbd_revoke_record_s *) hash_list->next;
+	while (&(record->hash) != hash_list) {
+		if (record->blocknr == blocknr)
+			return record;
+		record = (struct jbd_revoke_record_s *) record->hash.next;
+	}
+	return NULL;
+}
+
+int __init journal_init_revoke_caches(void)
+{
+	revoke_record_cache = kmem_cache_create("revoke_record",
+					   sizeof(struct jbd_revoke_record_s),
+					   0, SLAB_HWCACHE_ALIGN, NULL, NULL);
+	if (revoke_record_cache == 0)
+		return -ENOMEM;
+
+	revoke_table_cache = kmem_cache_create("revoke_table",
+					   sizeof(struct jbd_revoke_table_s),
+					   0, 0, NULL, NULL);
+	if (revoke_table_cache == 0) {
+		kmem_cache_destroy(revoke_record_cache);
+		revoke_record_cache = NULL;
+		return -ENOMEM;
+	}
+	return 0;
+}	
+
+void journal_destroy_revoke_caches(void)
+{
+	kmem_cache_destroy(revoke_record_cache);
+	revoke_record_cache = 0;
+	kmem_cache_destroy(revoke_table_cache);
+	revoke_table_cache = 0;
+}
+
+/* Initialise the revoke table for a given journal to a given size. */
+
+int journal_init_revoke(journal_t *journal, int hash_size)
+{
+	int shift, tmp;
+	
+	J_ASSERT (journal->j_revoke == NULL);
+	
+	journal->j_revoke = kmem_cache_alloc(revoke_table_cache, GFP_KERNEL);
+	if (!journal->j_revoke)
+		return -ENOMEM;
+	
+	/* Check that the hash_size is a power of two */
+	J_ASSERT ((hash_size & (hash_size-1)) == 0);
+
+	journal->j_revoke->hash_size = hash_size;
+
+	shift = 0;
+	tmp = hash_size;
+	while((tmp >>= 1UL) != 0UL)
+		shift++;
+	journal->j_revoke->hash_shift = shift;
+
+	journal->j_revoke->hash_table =
+		kmalloc(hash_size * sizeof(struct list_head), GFP_KERNEL);
+	if (!journal->j_revoke->hash_table) {
+		kmem_cache_free(revoke_table_cache, journal->j_revoke);
+		journal->j_revoke = NULL;
+		return -ENOMEM;
+	}
+	
+	for (tmp = 0; tmp < hash_size; tmp++)
+		INIT_LIST_HEAD(&journal->j_revoke->hash_table[tmp]);
+	
+	return 0;
+}
+
+/* Destoy a journal's revoke table.  The table must already be empty! */
+
+void journal_destroy_revoke(journal_t *journal)
+{
+	struct jbd_revoke_table_s *table;
+	struct list_head *hash_list;
+	int i;
+	
+	table = journal->j_revoke;
+	if (!table)
+		return;
+	
+	for (i=0; i<table->hash_size; i++) {
+		hash_list = &table->hash_table[i];
+		J_ASSERT (list_empty(hash_list));
+	}
+	
+	kfree(table->hash_table);
+	kmem_cache_free(revoke_table_cache, table);
+	journal->j_revoke = NULL;
+}
+
+
+#ifdef __KERNEL__
+
+/* 
+ * journal_revoke: revoke a given buffer_head from the journal.  This
+ * prevents the block from being replayed during recovery if we take a
+ * crash after this current transaction commits.  Any subsequent
+ * metadata writes of the buffer in this transaction cancel the
+ * revoke.  
+ *
+ * Note that this call may block --- it is up to the caller to make
+ * sure that there are no further calls to journal_write_metadata
+ * before the revoke is complete.  In ext3, this implies calling the
+ * revoke before clearing the block bitmap when we are deleting
+ * metadata. 
+ *
+ * Revoke performs a journal_forget on any buffer_head passed in as a
+ * parameter, but does _not_ forget the buffer_head if the bh was only
+ * found implicitly. 
+ *
+ * bh_in may not be a journalled buffer - it may have come off
+ * the hash tables without an attached journal_head.
+ *
+ * If bh_in is non-zero, journal_revoke() will decrement its b_count
+ * by one.
+ */
+
+int journal_revoke(handle_t *handle, unsigned long blocknr, 
+		   struct buffer_head *bh_in)
+{
+	struct buffer_head *bh = NULL;
+	journal_t *journal;
+	kdev_t dev;
+	int err;
+
+	if (bh_in)
+		BUFFER_TRACE(bh_in, "enter");
+
+	journal = handle->h_transaction->t_journal;
+	if (!journal_set_features(journal, 0, 0, JFS_FEATURE_INCOMPAT_REVOKE)){
+		J_ASSERT (!"Cannot set revoke feature!");
+		return -EINVAL;
+	}
+
+	dev = journal->j_fs_dev;
+	bh = bh_in;
+
+	if (!bh) {
+		bh = get_hash_table(dev, blocknr, journal->j_blocksize);
+		if (bh)
+			BUFFER_TRACE(bh, "found on hash");
+	}
+#ifdef JBD_EXPENSIVE_CHECKING
+	else {
+		struct buffer_head *bh2;
+
+		/* If there is a different buffer_head lying around in
+		 * memory anywhere... */
+		bh2 = get_hash_table(dev, blocknr, journal->j_blocksize);
+		if (bh2) {
+			/* ... and it has RevokeValid status... */
+			if ((bh2 != bh) &&
+			    test_bit(BH_RevokeValid, &bh2->b_state))
+				/* ...then it better be revoked too,
+				 * since it's illegal to create a revoke
+				 * record against a buffer_head which is
+				 * not marked revoked --- that would
+				 * risk missing a subsequent revoke
+				 * cancel. */
+				J_ASSERT_BH(bh2, test_bit(BH_Revoked, &
+							  bh2->b_state));
+			__brelse(bh2);
+		}
+	}
+#endif
+
+	/* We really ought not ever to revoke twice in a row without
+           first having the revoke cancelled: it's illegal to free a
+           block twice without allocating it in between! */
+	if (bh) {
+		J_ASSERT_BH(bh, !test_bit(BH_Revoked, &bh->b_state));
+		set_bit(BH_Revoked, &bh->b_state);
+		set_bit(BH_RevokeValid, &bh->b_state);
+		if (bh_in) {
+			BUFFER_TRACE(bh_in, "call journal_forget");
+			journal_forget(handle, bh_in);
+		} else {
+			BUFFER_TRACE(bh, "call brelse");
+			__brelse(bh);
+		}
+	}
+
+	lock_journal(journal);
+	jbd_debug(2, "insert revoke for block %lu, bh_in=%p\n", blocknr, bh_in);
+	err = insert_revoke_hash(journal, blocknr,
+				handle->h_transaction->t_tid);
+	unlock_journal(journal);
+	BUFFER_TRACE(bh_in, "exit");
+	return err;
+}
+
+/*
+ * Cancel an outstanding revoke.  For use only internally by the
+ * journaling code (called from journal_get_write_access).
+ *
+ * We trust the BH_Revoked bit on the buffer if the buffer is already
+ * being journaled: if there is no revoke pending on the buffer, then we
+ * don't do anything here.
+ *
+ * This would break if it were possible for a buffer to be revoked and
+ * discarded, and then reallocated within the same transaction.  In such
+ * a case we would have lost the revoked bit, but when we arrived here
+ * the second time we would still have a pending revoke to cancel.  So,
+ * do not trust the Revoked bit on buffers unless RevokeValid is also
+ * set.
+ *
+ * The caller must have the journal locked.
+ */
+int journal_cancel_revoke(handle_t *handle, struct journal_head *jh)
+{
+	struct jbd_revoke_record_s *record;
+	journal_t *journal = handle->h_transaction->t_journal;
+	int need_cancel;
+	int did_revoke = 0;	/* akpm: debug */
+	struct buffer_head *bh = jh2bh(jh);
+	
+	jbd_debug(4, "journal_head %p, cancelling revoke\n", jh);
+
+	/* Is the existing Revoke bit valid?  If so, we trust it, and
+	 * only perform the full cancel if the revoke bit is set.  If
+	 * not, we can't trust the revoke bit, and we need to do the
+	 * full search for a revoke record. */
+	if (test_and_set_bit(BH_RevokeValid, &bh->b_state))
+		need_cancel = (test_and_clear_bit(BH_Revoked, &bh->b_state));
+	else {
+		need_cancel = 1;
+		clear_bit(BH_Revoked, &bh->b_state);
+	}
+
+	if (need_cancel) {
+		record = find_revoke_record(journal, bh->b_blocknr);
+		if (record) {
+			jbd_debug(4, "cancelled existing revoke on "
+				  "blocknr %lu\n", bh->b_blocknr);
+			list_del(&record->hash);
+			kmem_cache_free(revoke_record_cache, record);
+			did_revoke = 1;
+		}
+	}
+
+#ifdef JBD_EXPENSIVE_CHECKING
+	/* There better not be one left behind by now! */
+	record = find_revoke_record(journal, bh->b_blocknr);
+	J_ASSERT_JH(jh, record == NULL);
+#endif
+
+	/* Finally, have we just cleared revoke on an unhashed
+	 * buffer_head?  If so, we'd better make sure we clear the
+	 * revoked status on any hashed alias too, otherwise the revoke
+	 * state machine will get very upset later on. */
+	if (need_cancel && !bh->b_pprev) {
+		struct buffer_head *bh2;
+		bh2 = get_hash_table(bh->b_dev, bh->b_blocknr, bh->b_size);
+		if (bh2) {
+			clear_bit(BH_Revoked, &bh2->b_state);
+			__brelse(bh2);
+		}
+	}
+	
+	return did_revoke;
+}
+
+
+/*
+ * Write revoke records to the journal for all entries in the current
+ * revoke hash, deleting the entries as we go.
+ *
+ * Called with the journal lock held.
+ */
+
+void journal_write_revoke_records(journal_t *journal, 
+				  transaction_t *transaction)
+{
+	struct journal_head *descriptor;
+	struct jbd_revoke_record_s *record;
+	struct jbd_revoke_table_s *revoke;
+	struct list_head *hash_list;
+	int i, offset, count;
+
+	descriptor = NULL; 
+	offset = 0;
+	count = 0;
+	revoke = journal->j_revoke;
+	
+	for (i = 0; i < revoke->hash_size; i++) {
+		hash_list = &revoke->hash_table[i];
+
+		while (!list_empty(hash_list)) {
+			record = (struct jbd_revoke_record_s *) 
+				hash_list->next;
+			write_one_revoke_record(journal, transaction,
+						&descriptor, &offset, 
+						record);
+			count++;
+			list_del(&record->hash);
+			kmem_cache_free(revoke_record_cache, record);
+		}
+	}
+	if (descriptor) 
+		flush_descriptor(journal, descriptor, offset);
+	jbd_debug(1, "Wrote %d revoke records\n", count);
+}
+
+/* 
+ * Write out one revoke record.  We need to create a new descriptor
+ * block if the old one is full or if we have not already created one.  
+ */
+
+static void write_one_revoke_record(journal_t *journal, 
+				    transaction_t *transaction,
+				    struct journal_head **descriptorp, 
+				    int *offsetp,
+				    struct jbd_revoke_record_s *record)
+{
+	struct journal_head *descriptor;
+	int offset;
+	journal_header_t *header;
+
+	/* If we are already aborting, this all becomes a noop.  We
+           still need to go round the loop in
+           journal_write_revoke_records in order to free all of the
+           revoke records: only the IO to the journal is omitted. */
+	if (is_journal_aborted(journal))
+		return;
+
+	descriptor = *descriptorp;
+	offset = *offsetp;
+
+	/* Make sure we have a descriptor with space left for the record */
+	if (descriptor) {
+		if (offset == journal->j_blocksize) {
+			flush_descriptor(journal, descriptor, offset);
+			descriptor = NULL;
+		}
+	}
+	
+	if (!descriptor) {
+		descriptor = journal_get_descriptor_buffer(journal);
+		if (!descriptor)
+			return;
+		header = (journal_header_t *) &jh2bh(descriptor)->b_data[0];
+		header->h_magic     = htonl(JFS_MAGIC_NUMBER);
+		header->h_blocktype = htonl(JFS_REVOKE_BLOCK);
+		header->h_sequence  = htonl(transaction->t_tid);
+
+		/* Record it so that we can wait for IO completion later */
+		JBUFFER_TRACE(descriptor, "file as BJ_LogCtl");
+		journal_file_buffer(descriptor, transaction, BJ_LogCtl);
+
+		offset = sizeof(journal_revoke_header_t);
+		*descriptorp = descriptor;
+	}
+	
+	* ((unsigned int *)(&jh2bh(descriptor)->b_data[offset])) = 
+		htonl(record->blocknr);
+	offset += 4;
+	*offsetp = offset;
+}
+
+/* 
+ * Flush a revoke descriptor out to the journal.  If we are aborting,
+ * this is a noop; otherwise we are generating a buffer which needs to
+ * be waited for during commit, so it has to go onto the appropriate
+ * journal buffer list.
+ */
+
+static void flush_descriptor(journal_t *journal, 
+			     struct journal_head *descriptor, 
+			     int offset)
+{
+	journal_revoke_header_t *header;
+
+	if (is_journal_aborted(journal)) {
+		JBUFFER_TRACE(descriptor, "brelse");
+		__brelse(jh2bh(descriptor));
+		return;
+	}
+	
+	header = (journal_revoke_header_t *) jh2bh(descriptor)->b_data;
+	header->r_count = htonl(offset);
+	set_bit(BH_JWrite, &jh2bh(descriptor)->b_state);
+	{
+		struct buffer_head *bh = jh2bh(descriptor);
+		BUFFER_TRACE(bh, "write");
+		ll_rw_block (WRITE, 1, &bh);
+	}
+}
+
+#endif
+
+/* 
+ * Revoke support for recovery.
+ *
+ * Recovery needs to be able to:
+ *
+ *  record all revoke records, including the tid of the latest instance
+ *  of each revoke in the journal
+ *
+ *  check whether a given block in a given transaction should be replayed
+ *  (ie. has not been revoked by a revoke record in that or a subsequent
+ *  transaction)
+ * 
+ *  empty the revoke table after recovery.
+ */
+
+/*
+ * First, setting revoke records.  We create a new revoke record for
+ * every block ever revoked in the log as we scan it for recovery, and
+ * we update the existing records if we find multiple revokes for a
+ * single block. 
+ */
+
+int journal_set_revoke(journal_t *journal, 
+		       unsigned long blocknr, 
+		       tid_t sequence)
+{
+	struct jbd_revoke_record_s *record;
+	
+	record = find_revoke_record(journal, blocknr);
+	if (record) {
+		/* If we have multiple occurences, only record the
+		 * latest sequence number in the hashed record */
+		if (tid_gt(sequence, record->sequence))
+			record->sequence = sequence;
+		return 0;
+	} 
+	return insert_revoke_hash(journal, blocknr, sequence);
+}
+
+/* 
+ * Test revoke records.  For a given block referenced in the log, has
+ * that block been revoked?  A revoke record with a given transaction
+ * sequence number revokes all blocks in that transaction and earlier
+ * ones, but later transactions still need replayed.
+ */
+
+int journal_test_revoke(journal_t *journal, 
+			unsigned long blocknr,
+			tid_t sequence)
+{
+	struct jbd_revoke_record_s *record;
+	
+	record = find_revoke_record(journal, blocknr);
+	if (!record)
+		return 0;
+	if (tid_gt(sequence, record->sequence))
+		return 0;
+	return 1;
+}
+
+/*
+ * Finally, once recovery is over, we need to clear the revoke table so
+ * that it can be reused by the running filesystem.
+ */
+
+void journal_clear_revoke(journal_t *journal)
+{
+	int i;
+	struct list_head *hash_list;
+	struct jbd_revoke_record_s *record;
+	struct jbd_revoke_table_s *revoke_var;
+	
+	revoke_var = journal->j_revoke;
+	
+	for (i = 0; i < revoke_var->hash_size; i++) {
+		hash_list = &revoke_var->hash_table[i];
+		while (!list_empty(hash_list)) {
+			record = (struct jbd_revoke_record_s*) hash_list->next;
+			list_del(&record->hash);
+			kmem_cache_free(revoke_record_cache, record);
+		}
+	}
+}
+

e2fsprogs/e2fsck/super.c

@@ -0,0 +1,713 @@
+/*
+ * e2fsck.c - superblock checks
+ * 
+ * Copyright (C) 1993, 1994, 1995, 1996, 1997 Theodore Ts'o.
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ */
+
+#ifdef HAVE_ERRNO_H
+#include <errno.h>
+#endif
+
+#ifndef EXT2_SKIP_UUID
+#include "uuid/uuid.h"
+#endif
+#include "e2fsck.h"
+#include "problem.h"
+
+#define MIN_CHECK 1
+#define MAX_CHECK 2
+
+static void check_super_value(e2fsck_t ctx, const char *descr,
+			      unsigned long value, int flags,
+			      unsigned long min_val, unsigned long max_val)
+{
+	struct		problem_context pctx;
+
+	if (((flags & MIN_CHECK) && (value < min_val)) ||
+	    ((flags & MAX_CHECK) && (value > max_val))) {
+		clear_problem_context(&pctx);
+		pctx.num = value;
+		pctx.str = descr;
+		fix_problem(ctx, PR_0_MISC_CORRUPT_SUPER, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT; /* never get here! */
+	}
+}
+
+/*
+ * This routine may get stubbed out in special compilations of the
+ * e2fsck code..
+ */
+#ifndef EXT2_SPECIAL_DEVICE_SIZE
+errcode_t e2fsck_get_device_size(e2fsck_t ctx)
+{
+	return (ext2fs_get_device_size(ctx->filesystem_name,
+				       EXT2_BLOCK_SIZE(ctx->fs->super),
+				       &ctx->num_blocks));
+}
+#endif
+
+/*
+ * helper function to release an inode
+ */
+struct process_block_struct {
+	e2fsck_t 	ctx;
+	char 		*buf;
+	struct problem_context *pctx;
+	int		truncating;
+	int		truncate_offset;
+	e2_blkcnt_t	truncate_block;
+	int		truncated_blocks;
+	int		abort;
+	errcode_t	errcode;
+};
+
+static int release_inode_block(ext2_filsys fs,
+			       blk_t	*block_nr,
+			       e2_blkcnt_t blockcnt,
+			       blk_t	ref_blk EXT2FS_ATTR((unused)),
+			       int	ref_offset EXT2FS_ATTR((unused)),
+			       void *priv_data)
+{
+	struct process_block_struct *pb;
+	e2fsck_t 		ctx;
+	struct problem_context	*pctx;
+	blk_t			blk = *block_nr;
+	int			retval = 0;
+
+	pb = (struct process_block_struct *) priv_data;
+	ctx = pb->ctx;
+	pctx = pb->pctx;
+
+	pctx->blk = blk;
+	pctx->blkcount = blockcnt;
+
+	if (HOLE_BLKADDR(blk))
+		return 0;
+
+	if ((blk < fs->super->s_first_data_block) ||
+	    (blk >= fs->super->s_blocks_count)) {
+		fix_problem(ctx, PR_0_ORPHAN_ILLEGAL_BLOCK_NUM, pctx);
+	return_abort:
+		pb->abort = 1;
+		return BLOCK_ABORT;
+	}
+
+	if (!ext2fs_test_block_bitmap(fs->block_map, blk)) {
+		fix_problem(ctx, PR_0_ORPHAN_ALREADY_CLEARED_BLOCK, pctx);
+		goto return_abort;
+	}
+
+	/*
+	 * If we are deleting an orphan, then we leave the fields alone.
+	 * If we are truncating an orphan, then update the inode fields
+	 * and clean up any partial block data.
+	 */
+	if (pb->truncating) {
+		/*
+		 * We only remove indirect blocks if they are
+		 * completely empty.
+		 */
+		if (blockcnt < 0) {
+			int	i, limit;
+			blk_t	*bp;
+			
+			pb->errcode = io_channel_read_blk(fs->io, blk, 1,
+							pb->buf);
+			if (pb->errcode)
+				goto return_abort;
+
+			limit = fs->blocksize >> 2;
+			for (i = 0, bp = (blk_t *) pb->buf;
+			     i < limit;	 i++, bp++)
+				if (*bp)
+					return 0;
+		}
+		/*
+		 * We don't remove direct blocks until we've reached
+		 * the truncation block.
+		 */
+		if (blockcnt >= 0 && blockcnt < pb->truncate_block)
+			return 0;
+		/*
+		 * If part of the last block needs truncating, we do
+		 * it here.
+		 */
+		if ((blockcnt == pb->truncate_block) && pb->truncate_offset) {
+			pb->errcode = io_channel_read_blk(fs->io, blk, 1,
+							pb->buf);
+			if (pb->errcode)
+				goto return_abort;
+			memset(pb->buf + pb->truncate_offset, 0,
+			       fs->blocksize - pb->truncate_offset);
+			pb->errcode = io_channel_write_blk(fs->io, blk, 1,
+							 pb->buf);
+			if (pb->errcode)
+				goto return_abort;
+		}
+		pb->truncated_blocks++;
+		*block_nr = 0;
+		retval |= BLOCK_CHANGED;
+	}
+	
+	ext2fs_block_alloc_stats(fs, blk, -1);
+	return retval;
+}
+		
+/*
+ * This function releases an inode.  Returns 1 if an inconsistency was
+ * found.  If the inode has a link count, then it is being truncated and
+ * not deleted.
+ */
+static int release_inode_blocks(e2fsck_t ctx, ext2_ino_t ino,
+				struct ext2_inode *inode, char *block_buf,
+				struct problem_context *pctx)
+{
+	struct process_block_struct 	pb;
+	ext2_filsys			fs = ctx->fs;
+	errcode_t			retval;
+	__u32				count;
+
+	if (!ext2fs_inode_has_valid_blocks(inode))
+		return 0;
+
+	pb.buf = block_buf + 3 * ctx->fs->blocksize;
+	pb.ctx = ctx;
+	pb.abort = 0;
+	pb.errcode = 0;
+	pb.pctx = pctx;
+	if (inode->i_links_count) {
+		pb.truncating = 1;
+		pb.truncate_block = (e2_blkcnt_t)
+			((((long long)inode->i_size_high << 32) +
+			  inode->i_size + fs->blocksize - 1) /
+			 fs->blocksize);
+		pb.truncate_offset = inode->i_size % fs->blocksize;
+	} else {
+		pb.truncating = 0;
+		pb.truncate_block = 0;
+		pb.truncate_offset = 0;
+	}
+	pb.truncated_blocks = 0;
+	retval = ext2fs_block_iterate2(fs, ino, BLOCK_FLAG_DEPTH_TRAVERSE, 
+				      block_buf, release_inode_block, &pb);
+	if (retval) {
+		com_err("release_inode_blocks", retval,
+			_("while calling ext2fs_block_iterate for inode %d"),
+			ino);
+		return 1;
+	}
+	if (pb.abort)
+		return 1;
+
+	/* Refresh the inode since ext2fs_block_iterate may have changed it */
+	e2fsck_read_inode(ctx, ino, inode, "release_inode_blocks");
+
+	if (pb.truncated_blocks)
+		inode->i_blocks -= pb.truncated_blocks *
+			(fs->blocksize / 512);
+
+	if (inode->i_file_acl) {
+		retval = ext2fs_adjust_ea_refcount(fs, inode->i_file_acl,
+						   block_buf, -1, &count);
+		if (retval == EXT2_ET_BAD_EA_BLOCK_NUM) {
+			retval = 0;
+			count = 1;
+		}
+		if (retval) {
+			com_err("release_inode_blocks", retval,
+		_("while calling ext2fs_adjust_ea_refocunt for inode %d"),
+				ino);
+			return 1;
+		}
+		if (count == 0)
+			ext2fs_block_alloc_stats(fs, inode->i_file_acl, -1);
+		inode->i_file_acl = 0;
+	}
+	return 0;
+}
+
+/*
+ * This function releases all of the orphan inodes.  It returns 1 if
+ * it hit some error, and 0 on success.
+ */
+static int release_orphan_inodes(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	ext2_ino_t	ino, next_ino;
+	struct ext2_inode inode;
+	struct problem_context pctx;
+	char *block_buf;
+
+	if ((ino = fs->super->s_last_orphan) == 0)
+		return 0;
+
+	/*
+	 * Win or lose, we won't be using the head of the orphan inode
+	 * list again.
+	 */
+	fs->super->s_last_orphan = 0;
+	ext2fs_mark_super_dirty(fs);
+
+	/*
+	 * If the filesystem contains errors, don't run the orphan
+	 * list, since the orphan list can't be trusted; and we're
+	 * going to be running a full e2fsck run anyway...
+	 */
+	if (fs->super->s_state & EXT2_ERROR_FS)
+		return 0;
+	
+	if ((ino < EXT2_FIRST_INODE(fs->super)) ||
+	    (ino > fs->super->s_inodes_count)) {
+		clear_problem_context(&pctx);
+		pctx.ino = ino;
+		fix_problem(ctx, PR_0_ORPHAN_ILLEGAL_HEAD_INODE, &pctx);
+		return 1;
+	}
+
+	block_buf = (char *) e2fsck_allocate_memory(ctx, fs->blocksize * 4,
+						    "block iterate buffer");
+	e2fsck_read_bitmaps(ctx);
+	
+	while (ino) {
+		e2fsck_read_inode(ctx, ino, &inode, "release_orphan_inodes");
+		clear_problem_context(&pctx);
+		pctx.ino = ino;
+		pctx.inode = &inode;
+		pctx.str = inode.i_links_count ? _("Truncating") :
+			_("Clearing");
+
+		fix_problem(ctx, PR_0_ORPHAN_CLEAR_INODE, &pctx);
+
+		next_ino = inode.i_dtime;
+		if (next_ino &&
+		    ((next_ino < EXT2_FIRST_INODE(fs->super)) ||
+		     (next_ino > fs->super->s_inodes_count))) {
+			pctx.ino = next_ino;
+			fix_problem(ctx, PR_0_ORPHAN_ILLEGAL_INODE, &pctx);
+			goto return_abort;
+		}
+
+		if (release_inode_blocks(ctx, ino, &inode, block_buf, &pctx))
+			goto return_abort;
+
+		if (!inode.i_links_count) {
+			ext2fs_inode_alloc_stats2(fs, ino, -1,
+						  LINUX_S_ISDIR(inode.i_mode));
+			inode.i_dtime = time(0);
+		} else {
+			inode.i_dtime = 0;
+		}
+		e2fsck_write_inode(ctx, ino, &inode, "delete_file");
+		ino = next_ino;
+	}
+	ext2fs_free_mem(&block_buf);
+	return 0;
+return_abort:
+	ext2fs_free_mem(&block_buf);
+	return 1;
+}
+
+/*
+ * Check the resize inode to make sure it is sane.  We check both for
+ * the case where on-line resizing is not enabled (in which case the
+ * resize inode should be cleared) as well as the case where on-line
+ * resizing is enabled.
+ */
+void check_resize_inode(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	struct ext2_inode inode;
+	struct problem_context	pctx;
+	int		i, j, gdt_off, ind_off;
+	blk_t		blk, pblk, expect;
+	__u32 		*dind_buf = 0, *ind_buf;
+	errcode_t	retval;
+
+	clear_problem_context(&pctx);
+
+	/* 
+	 * If the resize inode feature isn't set, then 
+	 * s_reserved_gdt_blocks must be zero.
+	 */
+	if (!(fs->super->s_feature_compat & 
+	      EXT2_FEATURE_COMPAT_RESIZE_INODE)) {
+		if (fs->super->s_reserved_gdt_blocks) {
+			pctx.num = fs->super->s_reserved_gdt_blocks;
+			if (fix_problem(ctx, PR_0_NONZERO_RESERVED_GDT_BLOCKS,
+					&pctx)) {
+				fs->super->s_reserved_gdt_blocks = 0;
+				ext2fs_mark_super_dirty(fs);
+			}
+		}
+	}
+
+	/* Read the resizde inode */
+	pctx.ino = EXT2_RESIZE_INO;
+	retval = ext2fs_read_inode(fs, EXT2_RESIZE_INO, &inode);
+	if (retval) {
+		if (fs->super->s_feature_compat & 
+		    EXT2_FEATURE_COMPAT_RESIZE_INODE)
+			ctx->flags |= E2F_FLAG_RESIZE_INODE;
+		return;
+	}
+
+	/* 
+	 * If the resize inode feature isn't set, check to make sure 
+	 * the resize inode is cleared; then we're done.
+	 */
+	if (!(fs->super->s_feature_compat & 
+	      EXT2_FEATURE_COMPAT_RESIZE_INODE)) {
+		for (i=0; i < EXT2_N_BLOCKS; i++) {
+			if (inode.i_block[i])
+				break;
+		}
+		if ((i < EXT2_N_BLOCKS) &&
+		    fix_problem(ctx, PR_0_CLEAR_RESIZE_INODE, &pctx)) {
+			memset(&inode, 0, sizeof(inode));
+			e2fsck_write_inode(ctx, EXT2_RESIZE_INO, &inode,
+					   "clear_resize");
+		}
+		return;
+	}
+
+	/* 
+	 * The resize inode feature is enabled; check to make sure the
+	 * only block in use is the double indirect block
+	 */
+	blk = inode.i_block[EXT2_DIND_BLOCK];
+	for (i=0; i < EXT2_N_BLOCKS; i++) {
+		if (i != EXT2_DIND_BLOCK && inode.i_block[i])
+			break;
+	}
+	if ((i < EXT2_N_BLOCKS) || !blk || !inode.i_links_count ||
+	    !(inode.i_mode & LINUX_S_IFREG) ||
+	    (blk < fs->super->s_first_data_block ||
+	     blk >= fs->super->s_blocks_count)) {
+	resize_inode_invalid:
+		if (fix_problem(ctx, PR_0_RESIZE_INODE_INVALID, &pctx)) {
+			memset(&inode, 0, sizeof(inode));
+			e2fsck_write_inode(ctx, EXT2_RESIZE_INO, &inode,
+					   "clear_resize");
+			ctx->flags |= E2F_FLAG_RESIZE_INODE;
+		}
+		if (!(ctx->options & E2F_OPT_READONLY)) {
+			fs->super->s_state &= ~EXT2_VALID_FS;
+			ext2fs_mark_super_dirty(fs);
+		}
+		goto cleanup;
+	}
+	dind_buf = (__u32 *) e2fsck_allocate_memory(ctx, fs->blocksize * 2,
+						    "resize dind buffer");
+	ind_buf = (__u32 *) ((char *) dind_buf + fs->blocksize);
+
+	retval = ext2fs_read_ind_block(fs, blk, dind_buf);
+	if (retval)
+		goto resize_inode_invalid;
+
+	gdt_off = fs->desc_blocks;
+	pblk = fs->super->s_first_data_block + 1 + fs->desc_blocks;
+	for (i = 0; i < fs->super->s_reserved_gdt_blocks / 4; 
+	     i++, gdt_off++, pblk++) {
+		gdt_off %= fs->blocksize/4;
+		if (dind_buf[gdt_off] != pblk)
+			goto resize_inode_invalid;
+		retval = ext2fs_read_ind_block(fs, pblk, ind_buf);
+		if (retval) 
+			goto resize_inode_invalid;
+		ind_off = 0;
+		for (j = 1; j < fs->group_desc_count; j++) {
+			if (!ext2fs_bg_has_super(fs, j))
+				continue;
+			expect = pblk + (j * fs->super->s_blocks_per_group);
+			if (ind_buf[ind_off] != expect)
+				goto resize_inode_invalid;
+			ind_off++;
+		}
+	}
+
+cleanup:
+	if (dind_buf)
+		ext2fs_free_mem(&dind_buf);
+
+ }
+
+void check_super_block(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	blk_t	first_block, last_block;
+	struct ext2_super_block *sb = fs->super;
+	struct ext2_group_desc *gd;
+	blk_t	blocks_per_group = fs->super->s_blocks_per_group;
+	blk_t	bpg_max;
+	int	inodes_per_block;
+	int	ipg_max;
+	int	inode_size;
+	dgrp_t	i;
+	blk_t	should_be;
+	struct problem_context	pctx;
+	__u32	free_blocks = 0, free_inodes = 0;
+
+	inodes_per_block = EXT2_INODES_PER_BLOCK(fs->super);
+	ipg_max = inodes_per_block * (blocks_per_group - 4);
+	if (ipg_max > EXT2_MAX_INODES_PER_GROUP(sb))
+		ipg_max = EXT2_MAX_INODES_PER_GROUP(sb);
+	bpg_max = 8 * EXT2_BLOCK_SIZE(sb);
+	if (bpg_max > EXT2_MAX_BLOCKS_PER_GROUP(sb))
+		bpg_max = EXT2_MAX_BLOCKS_PER_GROUP(sb);
+
+	ctx->invalid_inode_bitmap_flag = (int *) e2fsck_allocate_memory(ctx,
+		 sizeof(int) * fs->group_desc_count, "invalid_inode_bitmap");
+	ctx->invalid_block_bitmap_flag = (int *) e2fsck_allocate_memory(ctx,
+		 sizeof(int) * fs->group_desc_count, "invalid_block_bitmap");
+	ctx->invalid_inode_table_flag = (int *) e2fsck_allocate_memory(ctx,
+		sizeof(int) * fs->group_desc_count, "invalid_inode_table");
+
+	clear_problem_context(&pctx);
+
+	/*
+	 * Verify the super block constants...
+	 */
+	check_super_value(ctx, "inodes_count", sb->s_inodes_count,
+			  MIN_CHECK, 1, 0);
+	check_super_value(ctx, "blocks_count", sb->s_blocks_count,
+			  MIN_CHECK, 1, 0);
+	check_super_value(ctx, "first_data_block", sb->s_first_data_block,
+			  MAX_CHECK, 0, sb->s_blocks_count);
+	check_super_value(ctx, "log_block_size", sb->s_log_block_size,
+			  MIN_CHECK | MAX_CHECK, 0,
+			  EXT2_MAX_BLOCK_LOG_SIZE - EXT2_MIN_BLOCK_LOG_SIZE);
+	check_super_value(ctx, "log_frag_size", sb->s_log_frag_size,
+			  MIN_CHECK | MAX_CHECK, 0, sb->s_log_block_size);
+	check_super_value(ctx, "frags_per_group", sb->s_frags_per_group,
+			  MIN_CHECK | MAX_CHECK, sb->s_blocks_per_group,
+			  bpg_max);
+	check_super_value(ctx, "blocks_per_group", sb->s_blocks_per_group,
+			  MIN_CHECK | MAX_CHECK, 8, bpg_max);
+	check_super_value(ctx, "inodes_per_group", sb->s_inodes_per_group,
+			  MIN_CHECK | MAX_CHECK, inodes_per_block, ipg_max);
+	check_super_value(ctx, "r_blocks_count", sb->s_r_blocks_count,
+			  MAX_CHECK, 0, sb->s_blocks_count / 2);
+	check_super_value(ctx, "reserved_gdt_blocks", 
+			  sb->s_reserved_gdt_blocks, MAX_CHECK, 0,
+			  fs->blocksize/4);
+	inode_size = EXT2_INODE_SIZE(sb);
+	check_super_value(ctx, "inode_size",
+			  inode_size, MIN_CHECK | MAX_CHECK,
+			  EXT2_GOOD_OLD_INODE_SIZE, fs->blocksize);
+	if (inode_size & (inode_size - 1)) {
+		pctx.num = inode_size;
+		pctx.str = "inode_size";
+		fix_problem(ctx, PR_0_MISC_CORRUPT_SUPER, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT; /* never get here! */
+		return;
+	}
+		
+	if (!ctx->num_blocks) {
+		pctx.errcode = e2fsck_get_device_size(ctx);
+		if (pctx.errcode && pctx.errcode != EXT2_ET_UNIMPLEMENTED) {
+			fix_problem(ctx, PR_0_GETSIZE_ERROR, &pctx);
+			ctx->flags |= E2F_FLAG_ABORT;
+			return;
+		}
+		if ((pctx.errcode != EXT2_ET_UNIMPLEMENTED) &&
+		    (ctx->num_blocks < sb->s_blocks_count)) {
+			pctx.blk = sb->s_blocks_count;
+			pctx.blk2 = ctx->num_blocks;
+			if (fix_problem(ctx, PR_0_FS_SIZE_WRONG, &pctx)) {
+				ctx->flags |= E2F_FLAG_ABORT;
+				return;
+			}
+		}
+	}
+
+	if (sb->s_log_block_size != (__u32) sb->s_log_frag_size) {
+		pctx.blk = EXT2_BLOCK_SIZE(sb);
+		pctx.blk2 = EXT2_FRAG_SIZE(sb);
+		fix_problem(ctx, PR_0_NO_FRAGMENTS, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+
+	should_be = sb->s_frags_per_group >>
+		(sb->s_log_block_size - sb->s_log_frag_size);		
+	if (sb->s_blocks_per_group != should_be) {
+		pctx.blk = sb->s_blocks_per_group;
+		pctx.blk2 = should_be;
+		fix_problem(ctx, PR_0_BLOCKS_PER_GROUP, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+
+	should_be = (sb->s_log_block_size == 0) ? 1 : 0;
+	if (sb->s_first_data_block != should_be) {
+		pctx.blk = sb->s_first_data_block;
+		pctx.blk2 = should_be;
+		fix_problem(ctx, PR_0_FIRST_DATA_BLOCK, &pctx);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+
+	should_be = sb->s_inodes_per_group * fs->group_desc_count;
+	if (sb->s_inodes_count != should_be) {
+		pctx.ino = sb->s_inodes_count;
+		pctx.ino2 = should_be;
+		if (fix_problem(ctx, PR_0_INODE_COUNT_WRONG, &pctx)) {
+			sb->s_inodes_count = should_be;
+			ext2fs_mark_super_dirty(fs);
+		}
+	}
+
+	/*
+	 * Verify the group descriptors....
+	 */
+	first_block =  sb->s_first_data_block;
+	last_block = first_block + blocks_per_group;
+
+	for (i = 0, gd=fs->group_desc; i < fs->group_desc_count; i++, gd++) {
+		pctx.group = i;
+		
+		if (i == fs->group_desc_count - 1)
+			last_block = sb->s_blocks_count;
+		if ((gd->bg_block_bitmap < first_block) ||
+		    (gd->bg_block_bitmap >= last_block)) {
+			pctx.blk = gd->bg_block_bitmap;
+			if (fix_problem(ctx, PR_0_BB_NOT_GROUP, &pctx))
+				gd->bg_block_bitmap = 0;
+		}
+		if (gd->bg_block_bitmap == 0) {
+			ctx->invalid_block_bitmap_flag[i]++;
+			ctx->invalid_bitmaps++;
+		}
+		if ((gd->bg_inode_bitmap < first_block) ||
+		    (gd->bg_inode_bitmap >= last_block)) {
+			pctx.blk = gd->bg_inode_bitmap;
+			if (fix_problem(ctx, PR_0_IB_NOT_GROUP, &pctx))
+				gd->bg_inode_bitmap = 0;
+		}
+		if (gd->bg_inode_bitmap == 0) {
+			ctx->invalid_inode_bitmap_flag[i]++;
+			ctx->invalid_bitmaps++;
+		}
+		if ((gd->bg_inode_table < first_block) ||
+		    ((gd->bg_inode_table +
+		      fs->inode_blocks_per_group - 1) >= last_block)) {
+			pctx.blk = gd->bg_inode_table;
+			if (fix_problem(ctx, PR_0_ITABLE_NOT_GROUP, &pctx))
+				gd->bg_inode_table = 0;
+		}
+		if (gd->bg_inode_table == 0) {
+			ctx->invalid_inode_table_flag[i]++;
+			ctx->invalid_bitmaps++;
+		}
+		free_blocks += gd->bg_free_blocks_count;
+		free_inodes += gd->bg_free_inodes_count;
+		first_block += sb->s_blocks_per_group;
+		last_block += sb->s_blocks_per_group;
+
+		if ((gd->bg_free_blocks_count > sb->s_blocks_per_group) ||
+		    (gd->bg_free_inodes_count > sb->s_inodes_per_group) ||
+		    (gd->bg_used_dirs_count > sb->s_inodes_per_group))
+			ext2fs_unmark_valid(fs);
+
+	}
+
+	/*
+	 * Update the global counts from the block group counts.  This
+	 * is needed for an experimental patch which eliminates
+	 * locking the entire filesystem when allocating blocks or
+	 * inodes; if the filesystem is not unmounted cleanly, the
+	 * global counts may not be accurate.
+	 */
+	if ((free_blocks != sb->s_free_blocks_count) ||
+	    (free_inodes != sb->s_free_inodes_count)) {
+		if (ctx->options & E2F_OPT_READONLY)
+			ext2fs_unmark_valid(fs);
+		else {
+			sb->s_free_blocks_count = free_blocks;
+			sb->s_free_inodes_count = free_inodes;
+			ext2fs_mark_super_dirty(fs);
+		}
+	}
+	
+	if ((sb->s_free_blocks_count > sb->s_blocks_count) ||
+	    (sb->s_free_inodes_count > sb->s_inodes_count))
+		ext2fs_unmark_valid(fs);
+
+
+	/*
+	 * If we have invalid bitmaps, set the error state of the
+	 * filesystem.
+	 */
+	if (ctx->invalid_bitmaps && !(ctx->options & E2F_OPT_READONLY)) {
+		sb->s_state &= ~EXT2_VALID_FS;
+		ext2fs_mark_super_dirty(fs);
+	}
+
+	clear_problem_context(&pctx);
+	
+#ifndef EXT2_SKIP_UUID
+	/*
+	 * If the UUID field isn't assigned, assign it.
+	 */
+	if (!(ctx->options & E2F_OPT_READONLY) && uuid_is_null(sb->s_uuid)) {
+		if (fix_problem(ctx, PR_0_ADD_UUID, &pctx)) {
+			uuid_generate(sb->s_uuid);
+			ext2fs_mark_super_dirty(fs);
+			fs->flags &= ~EXT2_FLAG_MASTER_SB_ONLY;
+		}
+	}
+#endif
+
+	/*
+	 * For the Hurd, check to see if the filetype option is set,
+	 * since it doesn't support it.
+	 */
+	if (!(ctx->options & E2F_OPT_READONLY) &&
+	    fs->super->s_creator_os == EXT2_OS_HURD &&
+	    (fs->super->s_feature_incompat &
+	     EXT2_FEATURE_INCOMPAT_FILETYPE)) {
+		if (fix_problem(ctx, PR_0_HURD_CLEAR_FILETYPE, &pctx)) {
+			fs->super->s_feature_incompat &=
+				~EXT2_FEATURE_INCOMPAT_FILETYPE;
+			ext2fs_mark_super_dirty(fs);
+
+		}
+	}
+
+	/*
+	 * If we have any of the compatibility flags set, we need to have a
+	 * revision 1 filesystem.  Most kernels will not check the flags on
+	 * a rev 0 filesystem and we may have corruption issues because of
+	 * the incompatible changes to the filesystem.
+	 */
+	if (!(ctx->options & E2F_OPT_READONLY) &&
+	    fs->super->s_rev_level == EXT2_GOOD_OLD_REV &&
+	    (fs->super->s_feature_compat ||
+	     fs->super->s_feature_ro_compat ||
+	     fs->super->s_feature_incompat) &&
+	    fix_problem(ctx, PR_0_FS_REV_LEVEL, &pctx)) {
+		ext2fs_update_dynamic_rev(fs);
+		ext2fs_mark_super_dirty(fs);
+	}
+
+	check_resize_inode(ctx);
+
+	/*
+	 * Clean up any orphan inodes, if present.
+	 */
+	if (!(ctx->options & E2F_OPT_READONLY) && release_orphan_inodes(ctx)) {
+		fs->super->s_state &= ~EXT2_VALID_FS;
+		ext2fs_mark_super_dirty(fs);
+	}
+
+	/*
+	 * Move the ext3 journal file, if necessary.
+	 */
+	e2fsck_move_ext3_journal(ctx);
+	return;
+}

e2fsprogs/e2fsck/swapfs.c

@@ -0,0 +1,269 @@
+/*
+ * swapfs.c --- byte-swap an ext2 filesystem
+ *
+ * Copyright 1996, 1997 by Theodore Ts'o
+ * 
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ * 
+ */
+
+#ifdef HAVE_ERRNO_H
+#include <errno.h>
+#endif
+#include <et/com_err.h>
+#include "e2fsck.h"
+
+#ifdef ENABLE_SWAPFS
+
+struct swap_block_struct {
+	ext2_ino_t	ino;
+	int		isdir;
+	errcode_t	errcode;
+	char		*dir_buf;
+	struct ext2_inode *inode;
+};
+
+/*
+ * This is a helper function for block_iterate.  We mark all of the
+ * indirect and direct blocks as changed, so that block_iterate will
+ * write them out.
+ */
+static int swap_block(ext2_filsys fs, blk_t *block_nr, int blockcnt,
+		      void *priv_data)
+{
+	errcode_t	retval;
+	
+	struct swap_block_struct *sb = (struct swap_block_struct *) priv_data;
+
+	if (sb->isdir && (blockcnt >= 0) && *block_nr) {
+		retval = ext2fs_read_dir_block(fs, *block_nr, sb->dir_buf);
+		if (retval) {
+			sb->errcode = retval;
+			return BLOCK_ABORT;
+		}
+		retval = ext2fs_write_dir_block(fs, *block_nr, sb->dir_buf);
+		if (retval) {
+			sb->errcode = retval;
+			return BLOCK_ABORT;
+		}
+	}
+	if (blockcnt >= 0) {
+		if (blockcnt < EXT2_NDIR_BLOCKS)
+			return 0;
+		return BLOCK_CHANGED;
+	}
+	if (blockcnt == BLOCK_COUNT_IND) {
+		if (*block_nr == sb->inode->i_block[EXT2_IND_BLOCK])
+			return 0;
+		return BLOCK_CHANGED;
+	}
+	if (blockcnt == BLOCK_COUNT_DIND) {
+		if (*block_nr == sb->inode->i_block[EXT2_DIND_BLOCK])
+			return 0;
+		return BLOCK_CHANGED;
+	}
+	if (blockcnt == BLOCK_COUNT_TIND) {
+		if (*block_nr == sb->inode->i_block[EXT2_TIND_BLOCK])
+			return 0;
+		return BLOCK_CHANGED;
+	}
+	return BLOCK_CHANGED;
+}
+
+/*
+ * This function is responsible for byte-swapping all of the indirect,
+ * block pointers.  It is also responsible for byte-swapping directories.
+ */
+static void swap_inode_blocks(e2fsck_t ctx, ext2_ino_t ino, char *block_buf,
+			      struct ext2_inode *inode)
+{
+	errcode_t			retval;
+	struct swap_block_struct	sb;
+
+	sb.ino = ino;
+	sb.inode = inode;
+	sb.dir_buf = block_buf + ctx->fs->blocksize*3;
+	sb.errcode = 0;
+	sb.isdir = 0;
+	if (LINUX_S_ISDIR(inode->i_mode))
+		sb.isdir = 1;
+
+	retval = ext2fs_block_iterate(ctx->fs, ino, 0, block_buf,
+				      swap_block, &sb);
+	if (retval) {
+		com_err("swap_inode_blocks", retval,
+			_("while calling ext2fs_block_iterate"));
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+	if (sb.errcode) {
+		com_err("swap_inode_blocks", sb.errcode,
+			_("while calling iterator function"));
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+}
+
+static void swap_inodes(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	dgrp_t			group;
+	unsigned int		i;
+	ext2_ino_t		ino = 1;
+	char 			*buf, *block_buf;
+	errcode_t		retval;
+	struct ext2_inode *	inode;
+
+	e2fsck_use_inode_shortcuts(ctx, 1);
+	
+	retval = ext2fs_get_mem(fs->blocksize * fs->inode_blocks_per_group,
+				&buf);
+	if (retval) {
+		com_err("swap_inodes", retval,
+			_("while allocating inode buffer"));
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+	block_buf = (char *) e2fsck_allocate_memory(ctx, fs->blocksize * 4,
+						    "block interate buffer");
+	for (group = 0; group < fs->group_desc_count; group++) {
+		retval = io_channel_read_blk(fs->io,
+		      fs->group_desc[group].bg_inode_table,
+		      fs->inode_blocks_per_group, buf);
+		if (retval) {
+			com_err("swap_inodes", retval,
+				_("while reading inode table (group %d)"),
+				group);
+			ctx->flags |= E2F_FLAG_ABORT;
+			return;
+		}
+		inode = (struct ext2_inode *) buf;
+		for (i=0; i < fs->super->s_inodes_per_group;
+		     i++, ino++, inode++) {
+			ctx->stashed_ino = ino;
+			ctx->stashed_inode = inode;
+			
+			if (fs->flags & EXT2_FLAG_SWAP_BYTES_READ)
+				ext2fs_swap_inode(fs, inode, inode, 0);
+			
+			/*
+			 * Skip deleted files.
+			 */
+			if (inode->i_links_count == 0)
+				continue;
+			
+			if (LINUX_S_ISDIR(inode->i_mode) ||
+			    ((inode->i_block[EXT2_IND_BLOCK] ||
+			      inode->i_block[EXT2_DIND_BLOCK] ||
+			      inode->i_block[EXT2_TIND_BLOCK]) &&
+			     ext2fs_inode_has_valid_blocks(inode)))
+				swap_inode_blocks(ctx, ino, block_buf, inode);
+
+			if (ctx->flags & E2F_FLAG_SIGNAL_MASK)
+				return;
+			
+			if (fs->flags & EXT2_FLAG_SWAP_BYTES_WRITE)
+				ext2fs_swap_inode(fs, inode, inode, 1);
+		}
+		retval = io_channel_write_blk(fs->io,
+		      fs->group_desc[group].bg_inode_table,
+		      fs->inode_blocks_per_group, buf);
+		if (retval) {
+			com_err("swap_inodes", retval,
+				_("while writing inode table (group %d)"),
+				group);
+			ctx->flags |= E2F_FLAG_ABORT;
+			return;
+		}
+	}
+	ext2fs_free_mem(&buf);
+	ext2fs_free_mem(&block_buf);
+	e2fsck_use_inode_shortcuts(ctx, 0);
+	ext2fs_flush_icache(fs);
+}
+
+#if defined(__powerpc__) && defined(EXT2FS_ENABLE_SWAPFS)
+/*
+ * On the PowerPC, the big-endian variant of the ext2 filesystem
+ * has its bitmaps stored as 32-bit words with bit 0 as the LSB
+ * of each word.  Thus a bitmap with only bit 0 set would be, as
+ * a string of bytes, 00 00 00 01 00 ...
+ * To cope with this, we byte-reverse each word of a bitmap if
+ * we have a big-endian filesystem, that is, if we are *not*
+ * byte-swapping other word-sized numbers.
+ */
+#define EXT2_BIG_ENDIAN_BITMAPS
+#endif
+
+#ifdef EXT2_BIG_ENDIAN_BITMAPS
+static void ext2fs_swap_bitmap(ext2fs_generic_bitmap bmap)
+{
+	__u32 *p = (__u32 *) bmap->bitmap;
+	int n, nbytes = (bmap->end - bmap->start + 7) / 8;
+		
+	for (n = nbytes / sizeof(__u32); n > 0; --n, ++p)
+		*p = ext2fs_swab32(*p);
+}
+#endif
+
+
+void swap_filesys(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+#ifdef RESOURCE_TRACK
+	struct resource_track	rtrack;
+
+	init_resource_track(&rtrack);
+#endif
+
+	if (!(ctx->options & E2F_OPT_PREEN))
+		printf(_("Pass 0: Doing byte-swap of filesystem\n"));
+	
+#ifdef MTRACE
+	mtrace_print("Byte swap");
+#endif
+
+	if (fs->super->s_mnt_count) {
+		fprintf(stderr, _("%s: the filesystem must be freshly "
+			"checked using fsck\n"
+			"and not mounted before trying to "
+			"byte-swap it.\n"), ctx->device_name);
+		ctx->flags |= E2F_FLAG_ABORT;
+		return;
+	}
+	if (fs->flags & EXT2_FLAG_SWAP_BYTES) {
+		fs->flags &= ~(EXT2_FLAG_SWAP_BYTES|
+			       EXT2_FLAG_SWAP_BYTES_WRITE);
+		fs->flags |= EXT2_FLAG_SWAP_BYTES_READ;
+	} else {
+		fs->flags &= ~EXT2_FLAG_SWAP_BYTES_READ;
+		fs->flags |= EXT2_FLAG_SWAP_BYTES_WRITE;
+	}
+	swap_inodes(ctx);
+	if (ctx->flags & E2F_FLAG_SIGNAL_MASK)
+		return;
+	if (fs->flags & EXT2_FLAG_SWAP_BYTES_WRITE)
+		fs->flags |= EXT2_FLAG_SWAP_BYTES;
+	fs->flags &= ~(EXT2_FLAG_SWAP_BYTES_READ|
+		       EXT2_FLAG_SWAP_BYTES_WRITE);
+
+#ifdef EXT2_BIG_ENDIAN_BITMAPS
+	e2fsck_read_bitmaps(ctx);
+	ext2fs_swap_bitmap(fs->inode_map);
+	ext2fs_swap_bitmap(fs->block_map);
+	fs->flags |= EXT2_FLAG_BB_DIRTY | EXT2_FLAG_IB_DIRTY;
+#endif
+	fs->flags &= ~EXT2_FLAG_MASTER_SB_ONLY;
+	ext2fs_flush(fs);
+	fs->flags |= EXT2_FLAG_MASTER_SB_ONLY;
+	
+#ifdef RESOURCE_TRACK
+	if (ctx->options & E2F_OPT_TIME2)
+		print_resource_track(_("Byte swap"), &rtrack);
+#endif
+}
+
+#endif

e2fsprogs/e2fsck/util.c

@@ -0,0 +1,503 @@
+/*
+ * util.c --- miscellaneous utilities
+ * 
+ * Copyright (C) 1993, 1994, 1995, 1996, 1997 Theodore Ts'o.
+ *
+ * %Begin-Header%
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ * %End-Header%
+ */
+
+#include <stdlib.h>
+#include <unistd.h>
+#include <string.h>
+#include <ctype.h>
+
+#ifdef HAVE_CONIO_H
+#undef HAVE_TERMIOS_H
+#include <conio.h>
+#define read_a_char()	getch()
+#else
+#ifdef HAVE_TERMIOS_H
+#include <termios.h>
+#endif
+#include <stdio.h>
+#endif
+
+#ifdef HAVE_MALLOC_H
+#include <malloc.h>
+#endif
+
+#include "e2fsck.h"
+
+extern e2fsck_t e2fsck_global_ctx;   /* Try your very best not to use this! */
+
+#include <sys/time.h>
+#include <sys/resource.h>
+
+#if 0
+void fatal_error(e2fsck_t ctx, const char *msg)
+{
+	if (msg) 
+		fprintf (stderr, "e2fsck: %s\n", msg);
+	if (ctx->fs && ctx->fs->io) {
+		if (ctx->fs->io->magic == EXT2_ET_MAGIC_IO_CHANNEL)
+			io_channel_flush(ctx->fs->io);
+		else
+			fprintf(stderr, "e2fsck: io manager magic bad!\n");
+	}
+	ctx->flags |= E2F_FLAG_ABORT;
+	if (ctx->flags & E2F_FLAG_SETJMP_OK)
+		longjmp(ctx->abort_loc, 1);
+	exit(FSCK_ERROR);
+}
+#endif
+
+void *e2fsck_allocate_memory(e2fsck_t ctx, unsigned int size,
+			     const char *description)
+{
+	void *ret;
+	char buf[256];
+
+#ifdef DEBUG_ALLOCATE_MEMORY
+	printf("Allocating %d bytes for %s...\n", size, description);
+#endif
+	ret = malloc(size);
+	if (!ret) {
+		sprintf(buf, "Can't allocate %s\n", description);
+		fatal_error(ctx, buf);
+	}
+	memset(ret, 0, size);
+	return ret;
+}
+
+char *string_copy(e2fsck_t ctx EXT2FS_ATTR((unused)), 
+		  const char *str, int len)
+{
+	char	*ret;
+	
+	if (!str)
+		return NULL;
+	if (!len)
+		len = strlen(str);
+	ret = malloc(len+1);
+	if (ret) {
+		strncpy(ret, str, len);
+		ret[len] = 0;
+	}
+	return ret;
+}
+
+#ifndef HAVE_STRNLEN
+/*
+ * Incredibly, libc5 doesn't appear to have strnlen.  So we have to
+ * provide our own.
+ */
+int e2fsck_strnlen(const char * s, int count)
+{
+	const char *cp = s;
+
+	while (count-- && *cp)
+		cp++;
+	return cp - s;
+}
+#endif
+
+#ifndef HAVE_CONIO_H
+static int read_a_char(void)
+{
+	char	c;
+	int	r;
+	int	fail = 0;
+
+	while(1) {
+		if (e2fsck_global_ctx &&
+		    (e2fsck_global_ctx->flags & E2F_FLAG_CANCEL)) {
+			return 3;
+		}
+		r = read(0, &c, 1);
+		if (r == 1)
+			return c;
+		if (fail++ > 100)
+			break;
+	}
+	return EOF;
+}
+#endif
+
+int ask_yn(const char * string, int def)
+{
+	int		c;
+	const char	*defstr;
+	const char	*short_yes = _("yY");
+	const char	*short_no = _("nN");
+
+#ifdef HAVE_TERMIOS_H
+	struct termios	termios, tmp;
+
+	tcgetattr (0, &termios);
+	tmp = termios;
+	tmp.c_lflag &= ~(ICANON | ECHO);
+	tmp.c_cc[VMIN] = 1;
+	tmp.c_cc[VTIME] = 0;
+	tcsetattr (0, TCSANOW, &tmp);
+#endif
+
+	if (def == 1)
+		defstr = _(_("<y>"));
+	else if (def == 0)
+		defstr = _(_("<n>"));
+	else
+		defstr = _(" (y/n)");
+	printf("%s%s? ", string, defstr);
+	while (1) {
+		fflush (stdout);
+		if ((c = read_a_char()) == EOF)
+			break;
+		if (c == 3) {
+#ifdef HAVE_TERMIOS_H
+			tcsetattr (0, TCSANOW, &termios);
+#endif
+			if (e2fsck_global_ctx &&
+			    e2fsck_global_ctx->flags & E2F_FLAG_SETJMP_OK) {
+				puts("\n");
+				longjmp(e2fsck_global_ctx->abort_loc, 1);
+			}
+			puts(_("cancelled!\n"));
+			return 0;
+		}
+		if (strchr(short_yes, (char) c)) {
+			def = 1;
+			break;
+		}
+		else if (strchr(short_no, (char) c)) {
+			def = 0;
+			break;
+		}
+		else if ((c == ' ' || c == '\n') && (def != -1))
+			break;
+	}
+	if (def)
+		puts(_("yes\n"));
+	else
+		puts (_("no\n"));
+#ifdef HAVE_TERMIOS_H
+	tcsetattr (0, TCSANOW, &termios);
+#endif
+	return def;
+}
+
+int ask (e2fsck_t ctx, const char * string, int def)
+{
+	if (ctx->options & E2F_OPT_NO) {
+		printf (_("%s? no\n\n"), string);
+		return 0;
+	}
+	if (ctx->options & E2F_OPT_YES) {
+		printf (_("%s? yes\n\n"), string);
+		return 1;
+	}
+	if (ctx->options & E2F_OPT_PREEN) {
+		printf ("%s? %s\n\n", string, def ? _("yes") : _("no"));
+		return def;
+	}
+	return ask_yn(string, def);
+}
+
+void e2fsck_read_bitmaps(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	errcode_t	retval;
+
+	if (ctx->invalid_bitmaps) {
+		com_err(ctx->program_name, 0,
+		    _("e2fsck_read_bitmaps: illegal bitmap block(s) for %s"),
+			ctx->device_name);
+		fatal_error(ctx, 0);
+	}
+
+	ehandler_operation(_("reading inode and block bitmaps"));
+	retval = ext2fs_read_bitmaps(fs);
+	ehandler_operation(0);
+	if (retval) {
+		com_err(ctx->program_name, retval,
+			_("while retrying to read bitmaps for %s"),
+			ctx->device_name);
+		fatal_error(ctx, 0);
+	}
+}
+
+void e2fsck_write_bitmaps(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+	errcode_t	retval;
+
+	if (ext2fs_test_bb_dirty(fs)) {
+		ehandler_operation(_("writing block bitmaps"));
+		retval = ext2fs_write_block_bitmap(fs);
+		ehandler_operation(0);
+		if (retval) {
+			com_err(ctx->program_name, retval,
+			    _("while retrying to write block bitmaps for %s"),
+				ctx->device_name);
+			fatal_error(ctx, 0);
+		}
+	}
+
+	if (ext2fs_test_ib_dirty(fs)) {
+		ehandler_operation(_("writing inode bitmaps"));
+		retval = ext2fs_write_inode_bitmap(fs);
+		ehandler_operation(0);
+		if (retval) {
+			com_err(ctx->program_name, retval,
+			    _("while retrying to write inode bitmaps for %s"),
+				ctx->device_name);
+			fatal_error(ctx, 0);
+		}
+	}
+}
+
+void preenhalt(e2fsck_t ctx)
+{
+	ext2_filsys fs = ctx->fs;
+
+	if (!(ctx->options & E2F_OPT_PREEN))
+		return;
+	fprintf(stderr, _("\n\n%s: UNEXPECTED INCONSISTENCY; "
+		"RUN fsck MANUALLY.\n\t(i.e., without -a or -p options)\n"),
+	       ctx->device_name);
+	if (fs != NULL) {
+		fs->super->s_state |= EXT2_ERROR_FS;
+		ext2fs_mark_super_dirty(fs);
+		ext2fs_close(fs);
+	}
+	exit(FSCK_UNCORRECTED);
+}
+
+#ifdef RESOURCE_TRACK
+void init_resource_track(struct resource_track *track)
+{
+#ifdef HAVE_GETRUSAGE
+	struct rusage r;
+#endif
+	
+	track->brk_start = sbrk(0);
+	gettimeofday(&track->time_start, 0);
+#ifdef HAVE_GETRUSAGE
+#ifdef sun
+	memset(&r, 0, sizeof(struct rusage));
+#endif
+	getrusage(RUSAGE_SELF, &r);
+	track->user_start = r.ru_utime;
+	track->system_start = r.ru_stime;
+#else
+	track->user_start.tv_sec = track->user_start.tv_usec = 0;
+	track->system_start.tv_sec = track->system_start.tv_usec = 0;
+#endif
+}
+
+#ifdef __GNUC__
+#define _INLINE_ __inline__
+#else
+#define _INLINE_
+#endif
+
+static _INLINE_ float timeval_subtract(struct timeval *tv1,
+				       struct timeval *tv2)
+{
+	return ((tv1->tv_sec - tv2->tv_sec) +
+		((float) (tv1->tv_usec - tv2->tv_usec)) / 1000000);
+}
+
+void print_resource_track(const char *desc, struct resource_track *track)
+{
+#ifdef HAVE_GETRUSAGE
+	struct rusage r;
+#endif
+#ifdef HAVE_MALLINFO
+	struct mallinfo	malloc_info;
+#endif
+	struct timeval time_end;
+
+	gettimeofday(&time_end, 0);
+
+	if (desc)
+		printf("%s: ", desc);
+
+#ifdef HAVE_MALLINFO
+#define kbytes(x)	(((x) + 1023) / 1024)
+	
+	malloc_info = mallinfo();
+	printf(_("Memory used: %dk/%dk (%dk/%dk), "),
+	       kbytes(malloc_info.arena), kbytes(malloc_info.hblkhd),
+	       kbytes(malloc_info.uordblks), kbytes(malloc_info.fordblks));
+#else
+	printf(_("Memory used: %d, "),
+	       (int) (((char *) sbrk(0)) - ((char *) track->brk_start)));
+#endif	
+#ifdef HAVE_GETRUSAGE
+	getrusage(RUSAGE_SELF, &r);
+
+	printf(_("time: %5.2f/%5.2f/%5.2f\n"),
+	       timeval_subtract(&time_end, &track->time_start),
+	       timeval_subtract(&r.ru_utime, &track->user_start),
+	       timeval_subtract(&r.ru_stime, &track->system_start));
+#else
+	printf(_("elapsed time: %6.3f\n"),
+	       timeval_subtract(&time_end, &track->time_start));
+#endif
+}
+#endif /* RESOURCE_TRACK */
+
+void e2fsck_read_inode(e2fsck_t ctx, unsigned long ino,
+			      struct ext2_inode * inode, const char *proc)
+{
+	int retval;
+
+	retval = ext2fs_read_inode(ctx->fs, ino, inode);
+	if (retval) {
+		com_err("ext2fs_read_inode", retval,
+			_("while reading inode %ld in %s"), ino, proc);
+		fatal_error(ctx, 0);
+	}
+}
+
+extern void e2fsck_write_inode_full(e2fsck_t ctx, unsigned long ino,
+			       struct ext2_inode * inode, int bufsize,
+			       const char *proc)
+{
+	int retval;
+
+	retval = ext2fs_write_inode_full(ctx->fs, ino, inode, bufsize);
+	if (retval) {
+		com_err("ext2fs_write_inode", retval,
+			_("while writing inode %ld in %s"), ino, proc);
+		fatal_error(ctx, 0);
+	}
+}
+
+extern void e2fsck_write_inode(e2fsck_t ctx, unsigned long ino,
+			       struct ext2_inode * inode, const char *proc)
+{
+	int retval;
+
+	retval = ext2fs_write_inode(ctx->fs, ino, inode);
+	if (retval) {
+		com_err("ext2fs_write_inode", retval,
+			_("while writing inode %ld in %s"), ino, proc);
+		fatal_error(ctx, 0);
+	}
+}
+
+#ifdef MTRACE
+void mtrace_print(char *mesg)
+{
+	FILE	*malloc_get_mallstream();
+	FILE	*f = malloc_get_mallstream();
+
+	if (f)
+		fprintf(f, "============= %s\n", mesg);
+}
+#endif
+
+blk_t get_backup_sb(e2fsck_t ctx, ext2_filsys fs, const char *name,
+		   io_manager manager)
+{
+	struct ext2_super_block *sb;
+	io_channel		io = NULL;
+	void			*buf = NULL;
+	int			blocksize;
+	blk_t			superblock, ret_sb = 8193;
+	
+	if (fs && fs->super) {
+		ret_sb = (fs->super->s_blocks_per_group +
+			  fs->super->s_first_data_block);
+		if (ctx) {
+			ctx->superblock = ret_sb;
+			ctx->blocksize = fs->blocksize;
+		}
+		return ret_sb;
+	}
+		
+	if (ctx) {
+		if (ctx->blocksize) {
+			ret_sb = ctx->blocksize * 8;
+			if (ctx->blocksize == 1024)
+				ret_sb++;
+			ctx->superblock = ret_sb;
+			return ret_sb;
+		}
+		ctx->superblock = ret_sb;
+		ctx->blocksize = 1024;
+	}
+
+	if (!name || !manager)
+		goto cleanup;
+
+	if (manager->open(name, 0, &io) != 0)
+		goto cleanup;
+
+	if (ext2fs_get_mem(SUPERBLOCK_SIZE, &buf))
+		goto cleanup;
+	sb = (struct ext2_super_block *) buf;
+
+	for (blocksize = EXT2_MIN_BLOCK_SIZE;
+	     blocksize <= EXT2_MAX_BLOCK_SIZE ; blocksize *= 2) {
+		superblock = blocksize*8;
+		if (blocksize == 1024)
+			superblock++;
+		io_channel_set_blksize(io, blocksize);
+		if (io_channel_read_blk(io, superblock,
+					-SUPERBLOCK_SIZE, buf))
+			continue;
+#ifdef EXT2FS_ENABLE_SWAPFS
+		if (sb->s_magic == ext2fs_swab16(EXT2_SUPER_MAGIC))
+			ext2fs_swap_super(sb);
+#endif
+		if (sb->s_magic == EXT2_SUPER_MAGIC) {
+			ret_sb = superblock;
+			if (ctx) {
+				ctx->superblock = superblock;
+				ctx->blocksize = blocksize;
+			}
+			break;
+		}
+	}
+
+cleanup:
+	if (io)
+		io_channel_close(io);
+	if (buf)
+		ext2fs_free_mem(&buf);
+	return (ret_sb);
+}
+
+/*
+ * Given a mode, return the ext2 file type
+ */
+int ext2_file_type(unsigned int mode)
+{
+	if (LINUX_S_ISREG(mode))
+		return EXT2_FT_REG_FILE;
+
+	if (LINUX_S_ISDIR(mode))
+		return EXT2_FT_DIR;
+	
+	if (LINUX_S_ISCHR(mode))
+		return EXT2_FT_CHRDEV;
+	
+	if (LINUX_S_ISBLK(mode))
+		return EXT2_FT_BLKDEV;
+	
+	if (LINUX_S_ISLNK(mode))
+		return EXT2_FT_SYMLINK;
+
+	if (LINUX_S_ISFIFO(mode))
+		return EXT2_FT_FIFO;
+	
+	if (LINUX_S_ISSOCK(mode))
+		return EXT2_FT_SOCK;
+	
+	return 0;
+}

e2fsprogs/fsck.h

@@ -0,0 +1,59 @@
+/*
+ * fsck.h
+ */
+
+#include <time.h>
+
+#define FSCK_ATTR(x) __attribute__(x)
+
+
+#ifndef DEFAULT_FSTYPE
+#define DEFAULT_FSTYPE	"ext2"
+#endif
+
+#define MAX_DEVICES 32
+#define MAX_ARGS 32
+
+#define EXIT_OK          0
+#define EXIT_NONDESTRUCT 1
+#define EXIT_DESTRUCT    2
+#define EXIT_UNCORRECTED 4
+#define EXIT_ERROR       8
+#define EXIT_USAGE       16
+#define EXIT_LIBRARY     128
+
+/*
+ * Internal structure for mount tabel entries.
+ */
+
+struct fs_info {
+	char  *device;
+	char  *mountpt;
+	char  *type;
+	char  *opts;
+	int   freq;
+	int   passno;
+	int   flags;
+	struct fs_info *next;
+};
+
+#define FLAG_DONE 1
+#define FLAG_PROGRESS 2
+
+/*
+ * Structure to allow exit codes to be stored
+ */
+struct fsck_instance {
+	int	pid;
+	int	flags;
+	int	exit_status;
+	time_t	start_time;
+	char *	prog;
+	char *	type;
+	char *	device;
+	char *	base_device;
+	struct fsck_instance *next;
+};
+
+extern char *base_device(const char *device);
+extern const char *identify_fs(const char *fs_name, const char *fs_types);

include/applets.h

@@ -182,6 +182,9 @@
 #ifdef CONFIG_DUMPLEASES
 	APPLET(dumpleases, dumpleases_main, _BB_DIR_USR_BIN, _BB_SUID_NEVER)
 #endif
+#ifdef CONFIG_E2FSCK
+	APPLET(e2fsck, e2fsck_main, _BB_DIR_SBIN, _BB_SUID_NEVER)
+#endif
 #ifdef CONFIG_E2LABEL
 	APPLET_NOUSAGE("e2label", tune2fs_main, _BB_DIR_SBIN, _BB_SUID_NEVER)
 #endif
@@ -239,6 +242,13 @@
 #ifdef CONFIG_FREERAMDISK
 	APPLET(freeramdisk, freeramdisk_main, _BB_DIR_SBIN, _BB_SUID_NEVER)
 #endif
+#ifdef CONFIG_FSCK
+	APPLET(fsck, fsck_main, _BB_DIR_SBIN, _BB_SUID_NEVER)
+#endif
+#ifdef CONFIG_E2FSCK
+	APPLET_NOUSAGE("fsck.ext2", e2fsck_main, _BB_DIR_SBIN, _BB_SUID_NEVER)
+	APPLET_NOUSAGE("fsck.ext3", e2fsck_main, _BB_DIR_SBIN, _BB_SUID_NEVER)
+#endif
 #ifdef CONFIG_FSCK_MINIX
 	APPLET_ODDNAME("fsck.minix", fsck_minix_main, _BB_DIR_SBIN, _BB_SUID_NEVER, fsck_minix)
 #endif

include/usage.h

@@ -546,6 +546,26 @@
 	"\t-r,\t--remaining\tInterpret lease times as time remaing\n" \
 	"\t-a,\t--absolute\tInterpret lease times as expire time"
 
+#define e2fsck_trivial_usage \
+	"[-panyrcdfvstDFSV] [-b superblock] [-B blocksize] " \
+	"[-I inode_buffer_blocks] [-P process_inode_size] " \
+	"[-l|-L bad_blocks_file] [-C fd] [-j ext-journal] " \
+	"[-E extended-options] device"
+#define e2fsck_full_usage \
+	"Check a Linux ext2/ext3 file system.\n\n" \
+	"Options:\n" \
+	"\t-p\tAutomatic repair (no questions)\n" \
+	"\t-n\tMake no changes to the filesystem\n" \
+	"\t-y\tAssume 'yes' to all questions\n" \
+	"\t-c\tCheck for bad blocks and add them to the badblock list\n" \
+	"\t-f\tForce checking even if filesystem is marked clean\n" \
+	"\t-v\tBe verbose\n" \
+	"\t-b superblock\tUse alternative superblock\n" \
+	"\t-B blocksize\tForce blocksize when looking for superblock\n" \
+	"\t-j journal\tSet location of the external journal\n" \
+	"\t-l file\tAdd to badblocks list\n" \
+	"\t-L file\tSet badblocks list"
+
 #ifdef CONFIG_FEATURE_FANCY_ECHO
 #  define USAGE_FANCY_ECHO(a) a
 #else
@@ -764,6 +784,20 @@
 #define freeramdisk_example_usage \
 	"$ freeramdisk /dev/ram2\n"
 
+#define fsck_trivial_usage \
+	"[-ANPRTV] [ -C [ fd ] ] [-t fstype] [fs-options] [filesys ...]"
+#define fsck_full_usage \
+	"Check and repair filesystems.\n\n" \
+	"Options:\n" \
+	"\t-A\tWalk /etc/fstab and check all filesystems\n" \
+	"\t-N\tDon't execute, just show what would be done\n" \
+	"\t-P\tWhen using -A, check filesystems in parallel\n" \
+	"\t-R\tWhen using -A, skip the root filesystem\n" \
+	"\t-T\tDon't show title on startup\n" \
+	"\t-V\tVerbose mode\n" \
+	"\t-C\tWrite status information to specified filedescriptor\n" \
+	"\t-t\tList of filesystem types to check"
+
 #define fsck_minix_trivial_usage \
 	"[-larvsmf] /dev/name"
 #define fsck_minix_full_usage \