]> git.kernelconcepts.de Git - karo-tx-uboot.git/commitdiff
Merge branch 'ext4'
authorTom Rini <trini@ti.com>
Thu, 20 Sep 2012 18:27:35 +0000 (11:27 -0700)
committerTom Rini <trini@ti.com>
Thu, 20 Sep 2012 18:27:35 +0000 (11:27 -0700)
Update Makefile change for LIBS -> LIBS-y change.

Conflicts:
Makefile

Signed-off-by: Tom Rini <trini@ti.com>
21 files changed:
Makefile
common/Makefile
common/cmd_ext2.c
common/cmd_ext4.c [new file with mode: 0644]
common/cmd_ext_common.c [new file with mode: 0644]
doc/README.ext4 [new file with mode: 0644]
fs/Makefile
fs/ext2/dev.c [deleted file]
fs/ext2/ext2fs.c [deleted file]
fs/ext4/Makefile [moved from fs/ext2/Makefile with 86% similarity]
fs/ext4/crc16.c [new file with mode: 0644]
fs/ext4/crc16.h [new file with mode: 0644]
fs/ext4/dev.c [new file with mode: 0644]
fs/ext4/ext4_common.c [new file with mode: 0644]
fs/ext4/ext4_common.h [new file with mode: 0644]
fs/ext4/ext4_journal.c [new file with mode: 0644]
fs/ext4/ext4_journal.h [new file with mode: 0644]
fs/ext4/ext4fs.c [new file with mode: 0644]
include/ext2fs.h [deleted file]
include/ext4fs.h [new file with mode: 0644]
include/ext_common.h [new file with mode: 0644]

index 971513de95f5c81642631d8ff5115c4dfd07d098..bf8165e2f40042b0469db1a28ca398ceee3fc4b0 100644 (file)
--- a/Makefile
+++ b/Makefile
@@ -242,9 +242,15 @@ LIBS-y += arch/arm/cpu/ixp/npe/libnpe.o
 endif
 LIBS-$(CONFIG_OF_EMBED) += dts/libdts.o
 LIBS-y += arch/$(ARCH)/lib/lib$(ARCH).o
-LIBS-y += fs/cramfs/libcramfs.o fs/fat/libfat.o fs/fdos/libfdos.o fs/jffs2/libjffs2.o \
-       fs/reiserfs/libreiserfs.o fs/ext2/libext2fs.o fs/yaffs2/libyaffs2.o \
-       fs/ubifs/libubifs.o fs/zfs/libzfs.o
+LIBS-y += fs/cramfs/libcramfs.o \
+       fs/ext4/libext4fs.o \
+       fs/fat/libfat.o \
+       fs/fdos/libfdos.o \
+       fs/jffs2/libjffs2.o \
+       fs/reiserfs/libreiserfs.o \
+       fs/ubifs/libubifs.o \
+       fs/yaffs2/libyaffs2.o \
+       fs/zfs/libzfs.o
 LIBS-y += net/libnet.o
 LIBS-y += disk/libdisk.o
 LIBS-y += drivers/bios_emulator/libatibiosemu.o
index 57da76f9089f2a341a783422ec30a5b0c5842e2e..22e8a6fb35d2b83a37027085742240efc5fee40e 100644 (file)
@@ -86,7 +86,13 @@ COBJS-$(CONFIG_ENV_IS_IN_EEPROM) += cmd_eeprom.o
 COBJS-$(CONFIG_CMD_EEPROM) += cmd_eeprom.o
 COBJS-$(CONFIG_CMD_ELF) += cmd_elf.o
 COBJS-$(CONFIG_SYS_HUSH_PARSER) += cmd_exit.o
+COBJS-$(CONFIG_CMD_EXT4) += cmd_ext4.o
 COBJS-$(CONFIG_CMD_EXT2) += cmd_ext2.o
+ifdef CONFIG_CMD_EXT4
+COBJS-y        += cmd_ext_common.o
+else
+COBJS-$(CONFIG_CMD_EXT2) += cmd_ext_common.o
+endif
 COBJS-$(CONFIG_CMD_FAT) += cmd_fat.o
 COBJS-$(CONFIG_CMD_FDC)$(CONFIG_CMD_FDOS) += cmd_fdc.o
 COBJS-$(CONFIG_OF_LIBFDT) += cmd_fdt.o fdt_support.o
index 79b1e2fb62f82249b57b8251d3b51c0e9625d969..c27d9c7edeb744442239bde4122de727ed94c84b 100644 (file)
@@ -1,4 +1,9 @@
 /*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+
  * (C) Copyright 2004
  * esd gmbh <www.esd-electronics.com>
  * Reinhard Arlt <reinhard.arlt@esd-electronics.com>
  * Ext2fs support
  */
 #include <common.h>
-#include <part.h>
-#include <config.h>
-#include <command.h>
-#include <image.h>
-#include <linux/ctype.h>
-#include <asm/byteorder.h>
-#include <ext2fs.h>
-#if defined(CONFIG_CMD_USB) && defined(CONFIG_USB_STORAGE)
-#include <usb.h>
-#endif
-
-#if !defined(CONFIG_DOS_PARTITION) && !defined(CONFIG_EFI_PARTITION)
-#error DOS or EFI partition support must be selected
-#endif
-
-/* #define     EXT2_DEBUG */
-
-#ifdef EXT2_DEBUG
-#define        PRINTF(fmt,args...)     printf (fmt ,##args)
-#else
-#define PRINTF(fmt,args...)
-#endif
+#include <ext_common.h>
 
 int do_ext2ls (cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
 {
-       char *filename = "/";
-       int dev=0;
-       int part=1;
-       char *ep;
-       block_dev_desc_t *dev_desc=NULL;
-       int part_length;
-
-       if (argc < 3)
-               return CMD_RET_USAGE;
-
-       dev = (int)simple_strtoul (argv[2], &ep, 16);
-       dev_desc = get_dev(argv[1],dev);
-
-       if (dev_desc == NULL) {
-               printf ("\n** Block device %s %d not supported\n", argv[1], dev);
-               return 1;
-       }
-
-       if (*ep) {
-               if (*ep != ':') {
-                       puts ("\n** Invalid boot device, use `dev[:part]' **\n");
-                       return 1;
-               }
-               part = (int)simple_strtoul(++ep, NULL, 16);
-       }
-
-       if (argc == 4)
-               filename = argv[3];
-
-       PRINTF("Using device %s %d:%d, directory: %s\n", argv[1], dev, part, filename);
-
-       if ((part_length = ext2fs_set_blk_dev(dev_desc, part)) == 0) {
-               printf ("** Bad partition - %s %d:%d **\n",  argv[1], dev, part);
-               ext2fs_close();
-               return 1;
-       }
-
-       if (!ext2fs_mount(part_length)) {
-               printf ("** Bad ext2 partition or disk - %s %d:%d **\n",  argv[1], dev, part);
-               ext2fs_close();
-               return 1;
-       }
-
-       if (ext2fs_ls (filename)) {
-               printf ("** Error ext2fs_ls() **\n");
-               ext2fs_close();
-               return 1;
-       };
-
-       ext2fs_close();
+       if (do_ext_ls(cmdtp, flag, argc, argv))
+               return -1;
 
        return 0;
 }
 
-U_BOOT_CMD(
-       ext2ls, 4,      1,      do_ext2ls,
-       "list files in a directory (default /)",
-       "<interface> <dev[:part]> [directory]\n"
-       "    - list files from 'dev' on 'interface' in a 'directory'"
-);
-
 /******************************************************************************
  * Ext2fs boot command intepreter. Derived from diskboot
  */
 int do_ext2load (cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
 {
-       char *filename = NULL;
-       char *ep;
-       int dev, part = 1;
-       ulong addr = 0, part_length;
-       int filelen;
-       disk_partition_t info;
-       block_dev_desc_t *dev_desc = NULL;
-       char buf [12];
-       unsigned long count;
-       char *addr_str;
-
-       switch (argc) {
-       case 3:
-               addr_str = getenv("loadaddr");
-               if (addr_str != NULL)
-                       addr = simple_strtoul (addr_str, NULL, 16);
-               else
-                       addr = CONFIG_SYS_LOAD_ADDR;
-
-               filename = getenv ("bootfile");
-               count = 0;
-               break;
-       case 4:
-               addr = simple_strtoul (argv[3], NULL, 16);
-               filename = getenv ("bootfile");
-               count = 0;
-               break;
-       case 5:
-               addr = simple_strtoul (argv[3], NULL, 16);
-               filename = argv[4];
-               count = 0;
-               break;
-       case 6:
-               addr = simple_strtoul (argv[3], NULL, 16);
-               filename = argv[4];
-               count = simple_strtoul (argv[5], NULL, 16);
-               break;
-
-       default:
-               return CMD_RET_USAGE;
-       }
-
-       if (!filename) {
-               puts ("** No boot file defined **\n");
-               return 1;
-       }
-
-       dev = (int)simple_strtoul (argv[2], &ep, 16);
-       dev_desc = get_dev(argv[1],dev);
-       if (dev_desc==NULL) {
-               printf ("** Block device %s %d not supported\n", argv[1], dev);
-               return 1;
-       }
-       if (*ep) {
-               if (*ep != ':') {
-                       puts ("** Invalid boot device, use `dev[:part]' **\n");
-                       return 1;
-               }
-               part = (int)simple_strtoul(++ep, NULL, 16);
-       }
-
-       PRINTF("Using device %s%d, partition %d\n", argv[1], dev, part);
-
-       if (part != 0) {
-               if (get_partition_info (dev_desc, part, &info)) {
-                       printf ("** Bad partition %d **\n", part);
-                       return 1;
-               }
-
-               if (strncmp((char *)info.type, BOOT_PART_TYPE, sizeof(info.type)) != 0) {
-                       printf ("** Invalid partition type \"%.32s\""
-                               " (expect \"" BOOT_PART_TYPE "\")\n",
-                               info.type);
-                       return 1;
-               }
-               printf ("Loading file \"%s\" "
-                       "from %s device %d:%d (%.32s)\n",
-                       filename,
-                       argv[1], dev, part, info.name);
-       } else {
-               printf ("Loading file \"%s\" from %s device %d\n",
-                       filename, argv[1], dev);
-       }
-
-
-       if ((part_length = ext2fs_set_blk_dev(dev_desc, part)) == 0) {
-               printf ("** Bad partition - %s %d:%d **\n",  argv[1], dev, part);
-               ext2fs_close();
-               return 1;
-       }
-
-       if (!ext2fs_mount(part_length)) {
-               printf ("** Bad ext2 partition or disk - %s %d:%d **\n",
-                       argv[1], dev, part);
-               ext2fs_close();
-               return 1;
-       }
-
-       filelen = ext2fs_open(filename);
-       if (filelen < 0) {
-               printf("** File not found %s\n", filename);
-               ext2fs_close();
-               return 1;
-       }
-       if ((count < filelen) && (count != 0)) {
-           filelen = count;
-       }
-
-       if (ext2fs_read((char *)addr, filelen) != filelen) {
-               printf("** Unable to read \"%s\" from %s %d:%d **\n",
-                       filename, argv[1], dev, part);
-               ext2fs_close();
-               return 1;
-       }
-
-       ext2fs_close();
-
-       /* Loading ok, update default load address */
-       load_addr = addr;
-
-       printf ("%d bytes read\n", filelen);
-       sprintf(buf, "%X", filelen);
-       setenv("filesize", buf);
+       if (do_ext_load(cmdtp, flag, argc, argv))
+               return -1;
 
        return 0;
 }
 
+U_BOOT_CMD(
+       ext2ls, 4,      1,      do_ext2ls,
+       "list files in a directory (default /)",
+       "<interface> <dev[:part]> [directory]\n"
+       "    - list files from 'dev' on 'interface' in a 'directory'"
+);
+
 U_BOOT_CMD(
        ext2load,       6,      0,      do_ext2load,
        "load binary file from a Ext2 filesystem",
diff --git a/common/cmd_ext4.c b/common/cmd_ext4.c
new file mode 100644 (file)
index 0000000..77094c4
--- /dev/null
@@ -0,0 +1,237 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * Ext4fs support
+ * made from existing cmd_ext2.c file of Uboot
+ *
+ * (C) Copyright 2004
+ * esd gmbh <www.esd-electronics.com>
+ * Reinhard Arlt <reinhard.arlt@esd-electronics.com>
+ *
+ * made from cmd_reiserfs by
+ *
+ * (C) Copyright 2003 - 2004
+ * Sysgo Real-Time Solutions, AG <www.elinos.com>
+ * Pavel Bartusek <pba@sysgo.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
+ * MA 02111-1307 USA
+ *
+ */
+
+/*
+ * Changelog:
+ *     0.1 - Newly created file for ext4fs support. Taken from cmd_ext2.c
+ *             file in uboot. Added ext4fs ls load and write support.
+ */
+
+#include <common.h>
+#include <part.h>
+#include <config.h>
+#include <command.h>
+#include <image.h>
+#include <linux/ctype.h>
+#include <asm/byteorder.h>
+#include <ext_common.h>
+#include <ext4fs.h>
+#include <linux/stat.h>
+#include <malloc.h>
+
+#if defined(CONFIG_CMD_USB) && defined(CONFIG_USB_STORAGE)
+#include <usb.h>
+#endif
+
+#if !defined(CONFIG_DOS_PARTITION) && !defined(CONFIG_EFI_PARTITION)
+#error DOS or EFI partition support must be selected
+#endif
+
+uint64_t total_sector;
+uint64_t part_offset;
+#if defined(CONFIG_CMD_EXT4_WRITE)
+static uint64_t part_size;
+static uint16_t cur_part = 1;
+#endif
+
+#define DOS_PART_MAGIC_OFFSET          0x1fe
+#define DOS_FS_TYPE_OFFSET             0x36
+#define DOS_FS32_TYPE_OFFSET           0x52
+
+int do_ext4_load(cmd_tbl_t *cmdtp, int flag, int argc,
+                                               char *const argv[])
+{
+       if (do_ext_load(cmdtp, flag, argc, argv))
+               return -1;
+
+       return 0;
+}
+
+int do_ext4_ls(cmd_tbl_t *cmdtp, int flag, int argc, char *const argv[])
+{
+       if (do_ext_ls(cmdtp, flag, argc, argv))
+               return -1;
+
+       return 0;
+}
+
+#if defined(CONFIG_CMD_EXT4_WRITE)
+static int ext4_register_device(block_dev_desc_t *dev_desc, int part_no)
+{
+       unsigned char buffer[SECTOR_SIZE];
+       disk_partition_t info;
+
+       if (!dev_desc->block_read)
+               return -1;
+
+       /* check if we have a MBR (on floppies we have only a PBR) */
+       if (dev_desc->block_read(dev_desc->dev, 0, 1, (ulong *) buffer) != 1) {
+               printf("** Can't read from device %d **\n", dev_desc->dev);
+               return -1;
+       }
+       if (buffer[DOS_PART_MAGIC_OFFSET] != 0x55 ||
+           buffer[DOS_PART_MAGIC_OFFSET + 1] != 0xaa) {
+               /* no signature found */
+               return -1;
+       }
+
+       /* First we assume there is a MBR */
+       if (!get_partition_info(dev_desc, part_no, &info)) {
+               part_offset = info.start;
+               cur_part = part_no;
+               part_size = info.size;
+       } else if ((strncmp((char *)&buffer[DOS_FS_TYPE_OFFSET],
+                           "FAT", 3) == 0) || (strncmp((char *)&buffer
+                                                       [DOS_FS32_TYPE_OFFSET],
+                                                       "FAT32", 5) == 0)) {
+               /* ok, we assume we are on a PBR only */
+               cur_part = 1;
+               part_offset = 0;
+       } else {
+               printf("** Partition %d not valid on device %d **\n",
+                      part_no, dev_desc->dev);
+               return -1;
+       }
+
+       return 0;
+}
+
+int do_ext4_write(cmd_tbl_t *cmdtp, int flag, int argc,
+                               char *const argv[])
+{
+       const char *filename = "/";
+       int part_length;
+       unsigned long part = 1;
+       int dev;
+       char *ep;
+       unsigned long ram_address;
+       unsigned long file_size;
+       disk_partition_t info;
+       struct ext_filesystem *fs;
+
+       if (argc < 6)
+               return cmd_usage(cmdtp);
+
+       dev = (int)simple_strtoul(argv[2], &ep, 16);
+       ext4_dev_desc = get_dev(argv[1], dev);
+       if (ext4_dev_desc == NULL) {
+               printf("Block device %s %d not supported\n", argv[1], dev);
+               return 1;
+       }
+       if (init_fs(ext4_dev_desc))
+               return 1;
+
+       fs = get_fs();
+       if (*ep) {
+               if (*ep != ':') {
+                       puts("Invalid boot device, use `dev[:part]'\n");
+                       goto fail;
+               }
+               part = simple_strtoul(++ep, NULL, 16);
+       }
+
+       /* get the filename */
+       filename = argv[3];
+
+       /* get the address in hexadecimal format (string to int) */
+       ram_address = simple_strtoul(argv[4], NULL, 16);
+
+       /* get the filesize in base 10 format */
+       file_size = simple_strtoul(argv[5], NULL, 10);
+
+       /* set the device as block device */
+       part_length = ext4fs_set_blk_dev(fs->dev_desc, part);
+       if (part_length == 0) {
+               printf("Bad partition - %s %d:%lu\n", argv[1], dev, part);
+               goto fail;
+       }
+
+       /* register the device and partition */
+       if (ext4_register_device(fs->dev_desc, part) != 0) {
+               printf("Unable to use %s %d:%lu for fattable\n",
+                      argv[1], dev, part);
+               goto fail;
+       }
+
+       /* get the partition information */
+       if (!get_partition_info(fs->dev_desc, part, &info)) {
+               total_sector = (info.size * info.blksz) / SECTOR_SIZE;
+               fs->total_sect = total_sector;
+       } else {
+               printf("error : get partition info\n");
+               goto fail;
+       }
+
+       /* mount the filesystem */
+       if (!ext4fs_mount(part_length)) {
+               printf("Bad ext4 partition %s %d:%lu\n", argv[1], dev, part);
+               goto fail;
+       }
+
+       /* start write */
+       if (ext4fs_write(filename, (unsigned char *)ram_address, file_size)) {
+               printf("** Error ext4fs_write() **\n");
+               goto fail;
+       }
+       ext4fs_close();
+       deinit_fs(fs->dev_desc);
+
+       return 0;
+
+fail:
+       ext4fs_close();
+       deinit_fs(fs->dev_desc);
+
+       return 1;
+}
+
+U_BOOT_CMD(ext4write, 6, 1, do_ext4_write,
+       "create a file in the root directory",
+       "<interface> <dev[:part]> [Absolute filename path] [Address] [sizebytes]\n"
+       "         - create a file in / directory");
+
+#endif
+
+U_BOOT_CMD(ext4ls, 4, 1, do_ext4_ls,
+          "list files in a directory (default /)",
+          "<interface> <dev[:part]> [directory]\n"
+          "      - list files from 'dev' on 'interface' in a 'directory'");
+
+U_BOOT_CMD(ext4load, 6, 0, do_ext4_load,
+          "load binary file from a Ext4 filesystem",
+          "<interface> <dev[:part]> [addr] [filename] [bytes]\n"
+          "      - load binary file 'filename' from 'dev' on 'interface'\n"
+          "             to address 'addr' from ext4 filesystem");
diff --git a/common/cmd_ext_common.c b/common/cmd_ext_common.c
new file mode 100644 (file)
index 0000000..56ee9a5
--- /dev/null
@@ -0,0 +1,259 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT2/4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * Ext4fs support
+ * made from existing cmd_ext2.c file of Uboot
+ *
+ * (C) Copyright 2004
+ * esd gmbh <www.esd-electronics.com>
+ * Reinhard Arlt <reinhard.arlt@esd-electronics.com>
+ *
+ * made from cmd_reiserfs by
+ *
+ * (C) Copyright 2003 - 2004
+ * Sysgo Real-Time Solutions, AG <www.elinos.com>
+ * Pavel Bartusek <pba@sysgo.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of
+ * the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
+ * MA 02111-1307 USA
+ *
+ */
+
+/*
+ * Changelog:
+ *     0.1 - Newly created file for ext4fs support. Taken from cmd_ext2.c
+ *             file in uboot. Added ext4fs ls load and write support.
+ */
+
+#include <common.h>
+#include <part.h>
+#include <config.h>
+#include <command.h>
+#include <image.h>
+#include <linux/ctype.h>
+#include <asm/byteorder.h>
+#include <ext_common.h>
+#include <ext4fs.h>
+#include <linux/stat.h>
+#include <malloc.h>
+
+#if defined(CONFIG_CMD_USB) && defined(CONFIG_USB_STORAGE)
+#include <usb.h>
+#endif
+
+#if !defined(CONFIG_DOS_PARTITION) && !defined(CONFIG_EFI_PARTITION)
+#error DOS or EFI partition support must be selected
+#endif
+
+#define DOS_PART_MAGIC_OFFSET          0x1fe
+#define DOS_FS_TYPE_OFFSET             0x36
+#define DOS_FS32_TYPE_OFFSET           0x52
+
+int do_ext_load(cmd_tbl_t *cmdtp, int flag, int argc,
+                                               char *const argv[])
+{
+       char *filename = NULL;
+       char *ep;
+       int dev;
+       unsigned long part = 1;
+       ulong addr = 0;
+       ulong part_length;
+       int filelen;
+       disk_partition_t info;
+       struct ext_filesystem *fs;
+       char buf[12];
+       unsigned long count;
+       const char *addr_str;
+
+       count = 0;
+       addr = simple_strtoul(argv[3], NULL, 16);
+       filename = getenv("bootfile");
+       switch (argc) {
+       case 3:
+               addr_str = getenv("loadaddr");
+               if (addr_str != NULL)
+                       addr = simple_strtoul(addr_str, NULL, 16);
+               else
+                       addr = CONFIG_SYS_LOAD_ADDR;
+
+               break;
+       case 4:
+               break;
+       case 5:
+               filename = argv[4];
+               break;
+       case 6:
+               filename = argv[4];
+               count = simple_strtoul(argv[5], NULL, 16);
+               break;
+
+       default:
+               return cmd_usage(cmdtp);
+       }
+
+       if (!filename) {
+               puts("** No boot file defined **\n");
+               return 1;
+       }
+
+       dev = (int)simple_strtoul(argv[2], &ep, 16);
+       ext4_dev_desc = get_dev(argv[1], dev);
+       if (ext4_dev_desc == NULL) {
+               printf("** Block device %s %d not supported\n", argv[1], dev);
+               return 1;
+       }
+       if (init_fs(ext4_dev_desc))
+               return 1;
+
+       fs = get_fs();
+       if (*ep) {
+               if (*ep != ':') {
+                       puts("** Invalid boot device, use `dev[:part]' **\n");
+                       goto fail;
+               }
+               part = simple_strtoul(++ep, NULL, 16);
+       }
+
+       if (part != 0) {
+               if (get_partition_info(fs->dev_desc, part, &info)) {
+                       printf("** Bad partition %lu **\n", part);
+                       goto fail;
+               }
+
+               if (strncmp((char *)info.type, BOOT_PART_TYPE,
+                           strlen(BOOT_PART_TYPE)) != 0) {
+                       printf("** Invalid partition type \"%s\""
+                              " (expect \"" BOOT_PART_TYPE "\")\n", info.type);
+                       goto fail;
+               }
+               printf("Loading file \"%s\" "
+                      "from %s device %d:%lu %s\n",
+                      filename, argv[1], dev, part, info.name);
+       } else {
+               printf("Loading file \"%s\" from %s device %d\n",
+                      filename, argv[1], dev);
+       }
+
+       part_length = ext4fs_set_blk_dev(fs->dev_desc, part);
+       if (part_length == 0) {
+               printf("**Bad partition - %s %d:%lu **\n", argv[1], dev, part);
+               ext4fs_close();
+               goto fail;
+       }
+
+       if (!ext4fs_mount(part_length)) {
+               printf("** Bad ext2 partition or disk - %s %d:%lu **\n",
+                      argv[1], dev, part);
+               ext4fs_close();
+               goto fail;
+       }
+
+       filelen = ext4fs_open(filename);
+       if (filelen < 0) {
+               printf("** File not found %s\n", filename);
+               ext4fs_close();
+               goto fail;
+       }
+       if ((count < filelen) && (count != 0))
+               filelen = count;
+
+       if (ext4fs_read((char *)addr, filelen) != filelen) {
+               printf("** Unable to read \"%s\" from %s %d:%lu **\n",
+                      filename, argv[1], dev, part);
+               ext4fs_close();
+               goto fail;
+       }
+
+       ext4fs_close();
+       deinit_fs(fs->dev_desc);
+       /* Loading ok, update default load address */
+       load_addr = addr;
+
+       printf("%d bytes read\n", filelen);
+       sprintf(buf, "%X", filelen);
+       setenv("filesize", buf);
+
+       return 0;
+fail:
+       deinit_fs(fs->dev_desc);
+       return 1;
+}
+
+int do_ext_ls(cmd_tbl_t *cmdtp, int flag, int argc, char *const argv[])
+{
+       const char *filename = "/";
+       int dev;
+       unsigned long part = 1;
+       char *ep;
+       struct ext_filesystem *fs;
+       int part_length;
+       if (argc < 3)
+               return cmd_usage(cmdtp);
+
+       dev = (int)simple_strtoul(argv[2], &ep, 16);
+
+       ext4_dev_desc = get_dev(argv[1], dev);
+
+       if (ext4_dev_desc == NULL) {
+               printf("\n** Block device %s %d not supported\n", argv[1], dev);
+               return 1;
+       }
+
+       if (init_fs(ext4_dev_desc))
+               return 1;
+
+       fs = get_fs();
+       if (*ep) {
+               if (*ep != ':') {
+                       puts("\n** Invalid boot device, use `dev[:part]' **\n");
+                       goto fail;
+               }
+               part = simple_strtoul(++ep, NULL, 16);
+       }
+
+       if (argc == 4)
+               filename = argv[3];
+
+       part_length = ext4fs_set_blk_dev(fs->dev_desc, part);
+       if (part_length == 0) {
+               printf("** Bad partition - %s %d:%lu **\n", argv[1], dev, part);
+               ext4fs_close();
+               goto fail;
+       }
+
+       if (!ext4fs_mount(part_length)) {
+               printf("** Bad ext2 partition or disk - %s %d:%lu **\n",
+                      argv[1], dev, part);
+               ext4fs_close();
+               goto fail;
+       }
+
+       if (ext4fs_ls(filename)) {
+               printf("** Error extfs_ls() **\n");
+               ext4fs_close();
+               goto fail;
+       };
+
+       ext4fs_close();
+       deinit_fs(fs->dev_desc);
+       return 0;
+
+fail:
+       deinit_fs(fs->dev_desc);
+       return 1;
+}
diff --git a/doc/README.ext4 b/doc/README.ext4
new file mode 100644 (file)
index 0000000..b3ea8b7
--- /dev/null
@@ -0,0 +1,46 @@
+This patch series adds support for ext4 ls,load and write features in uboot
+Journaling is supported for write feature.
+
+To Enable ext2 ls and load commands, modify the board specific config file with
+#define CONFIG_CMD_EXT2
+
+To Enable ext4 ls and load commands, modify the board specific config file with
+#define CONFIG_CMD_EXT4
+
+To enable ext4 write command, modify the board specific config file with
+#define CONFIG_CMD_EXT4
+#define CONFIG_CMD_EXT4_WRITE
+
+Steps to test:
+
+1. After applying the patch, ext4 specific commands can be seen
+   in the boot loader prompt using
+        UBOOT #help
+
+        ext4load- load binary file from a Ext4 file system
+        ext4ls  - list files in a directory (default /)
+        ext4write- create a file in ext4 formatted partition
+
+2. To list the files in ext4 formatted partition, execute
+        ext4ls <interface> <dev[:part]> [directory]
+        For example:
+        UBOOT #ext4ls mmc 0:5 /usr/lib
+
+3. To read and load a file from an ext4 formatted partition to RAM, execute
+        ext4load <interface> <dev[:part]> [addr] [filename] [bytes]
+        For example:
+        UBOOT #ext4load mmc 2:2 0x30007fc0 uImage
+
+4. To write a file to a ext4 formatted partition.
+        a) First load a file to RAM at a particular address for example 0x30007fc0.
+        Now execute ext4write command
+        ext4write <interface> <dev[:part]> [filename] [Address] [sizebytes]
+        For example:
+        UBOOT #ext4write mmc 2:2 /boot/uImage 0x30007fc0 6183120
+        (here 6183120 is the size of the file to be written)
+        Note: Absolute path is required for the file to be written
+
+References :
+       -- ext4 implementation in Linux Kernel
+       -- Uboot existing ext2 load and ls implementation
+       -- Journaling block device JBD2 implementation in linux Kernel
index 28da76e959110de4e41b8d5a4feb38ba8e40026f..901e1894b6a6b5a1fc26e4fe800f75b42db0dae4 100644 (file)
 #
 
 subdirs-$(CONFIG_CMD_CRAMFS) := cramfs
-subdirs-$(CONFIG_CMD_EXT2) += ext2
+subdirs-$(CONFIG_CMD_EXT4) += ext4
+ifndef CONFIG_CMD_EXT4
+subdirs-$(CONFIG_CMD_EXT2) += ext4
+endif
 subdirs-$(CONFIG_CMD_FAT) += fat
 subdirs-$(CONFIG_CMD_FDOS) += fdos
 subdirs-$(CONFIG_CMD_JFFS2) += jffs2
diff --git a/fs/ext2/dev.c b/fs/ext2/dev.c
deleted file mode 100644 (file)
index 874e211..0000000
+++ /dev/null
@@ -1,131 +0,0 @@
-/*
- * (C) Copyright 2004
- *  esd gmbh <www.esd-electronics.com>
- *  Reinhard Arlt <reinhard.arlt@esd-electronics.com>
- *
- *  based on code of fs/reiserfs/dev.c by
- *
- *  (C) Copyright 2003 - 2004
- *  Sysgo AG, <www.elinos.com>, Pavel Bartusek <pba@sysgo.com>
- *
- *  This program is free software; you can redistribute it and/or modify
- *  it under the terms of the GNU General Public License as published by
- *  the Free Software Foundation; either version 2 of the License, or
- *  (at your option) any later version.
- *
- *  This program is distributed in the hope that it will be useful,
- *  but WITHOUT ANY WARRANTY; without even the implied warranty of
- *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- *  GNU General Public License for more details.
- *
- *  You should have received a copy of the GNU General Public License
- *  along with this program; if not, write to the Free Software
- *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
- */
-
-
-#include <common.h>
-#include <config.h>
-#include <ext2fs.h>
-
-static block_dev_desc_t *ext2fs_block_dev_desc;
-static disk_partition_t part_info;
-
-int ext2fs_set_blk_dev(block_dev_desc_t *rbdd, int part)
-{
-       ext2fs_block_dev_desc = rbdd;
-
-       if (part == 0) {
-               /* disk doesn't use partition table */
-               part_info.start = 0;
-               part_info.size = rbdd->lba;
-               part_info.blksz = rbdd->blksz;
-       } else {
-               if (get_partition_info
-                   (ext2fs_block_dev_desc, part, &part_info)) {
-                       return 0;
-               }
-       }
-       return part_info.size;
-}
-
-
-int ext2fs_devread(int sector, int byte_offset, int byte_len, char *buf)
-{
-       ALLOC_CACHE_ALIGN_BUFFER(char, sec_buf, SECTOR_SIZE);
-       unsigned sectors;
-
-       /*
-        *  Check partition boundaries
-        */
-       if ((sector < 0) ||
-           ((sector + ((byte_offset + byte_len - 1) >> SECTOR_BITS)) >=
-               part_info.size)) {
-               /* errnum = ERR_OUTSIDE_PART; */
-               printf(" ** %s read outside partition sector %d\n",
-                      __func__,
-                      sector);
-               return 0;
-       }
-
-       /*
-        *  Get the read to the beginning of a partition.
-        */
-       sector += byte_offset >> SECTOR_BITS;
-       byte_offset &= SECTOR_SIZE - 1;
-
-       debug(" <%d, %d, %d>\n", sector, byte_offset, byte_len);
-
-       if (ext2fs_block_dev_desc == NULL) {
-               printf(" ** %s Invalid Block Device Descriptor (NULL)\n",
-                      __func__);
-               return 0;
-       }
-
-       if (byte_offset != 0) {
-               /* read first part which isn't aligned with start of sector */
-               if (ext2fs_block_dev_desc->
-                   block_read(ext2fs_block_dev_desc->dev,
-                              part_info.start + sector, 1,
-                              (unsigned long *) sec_buf) != 1) {
-                       printf(" ** %s read error **\n", __func__);
-                       return 0;
-               }
-               memcpy(buf, sec_buf + byte_offset,
-                      min(SECTOR_SIZE - byte_offset, byte_len));
-               buf += min(SECTOR_SIZE - byte_offset, byte_len);
-               byte_len -= min(SECTOR_SIZE - byte_offset, byte_len);
-               sector++;
-       }
-
-       /*  read sector aligned part */
-       sectors = byte_len / SECTOR_SIZE;
-
-       if (sectors > 0) {
-               if (ext2fs_block_dev_desc->block_read(
-                       ext2fs_block_dev_desc->dev,
-                       part_info.start + sector,
-                       sectors,
-                       (unsigned long *) buf) != sectors) {
-                       printf(" ** %s read error - block\n", __func__);
-                       return 0;
-               }
-
-               buf += sectors * SECTOR_SIZE;
-               byte_len -= sectors * SECTOR_SIZE;
-               sector += sectors;
-       }
-
-       if (byte_len != 0) {
-               /* read rest of data which are not in whole sector */
-               if (ext2fs_block_dev_desc->
-                   block_read(ext2fs_block_dev_desc->dev,
-                              part_info.start + sector, 1,
-                              (unsigned long *) sec_buf) != 1) {
-                       printf(" ** %s read error - last part\n", __func__);
-                       return 0;
-               }
-               memcpy(buf, sec_buf, byte_len);
-       }
-       return 1;
-}
diff --git a/fs/ext2/ext2fs.c b/fs/ext2/ext2fs.c
deleted file mode 100644 (file)
index 418404e..0000000
+++ /dev/null
@@ -1,919 +0,0 @@
-/*
- * (C) Copyright 2004
- *  esd gmbh <www.esd-electronics.com>
- *  Reinhard Arlt <reinhard.arlt@esd-electronics.com>
- *
- *  based on code from grub2 fs/ext2.c and fs/fshelp.c by
- *
- *  GRUB  --  GRand Unified Bootloader
- *  Copyright (C) 2003, 2004  Free Software Foundation, Inc.
- *
- *  This program is free software; you can redistribute it and/or modify
- *  it under the terms of the GNU General Public License as published by
- *  the Free Software Foundation; either version 2 of the License, or
- *  (at your option) any later version.
- *
- *  This program is distributed in the hope that it will be useful,
- *  but WITHOUT ANY WARRANTY; without even the implied warranty of
- *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- *  GNU General Public License for more details.
- *
- *  You should have received a copy of the GNU General Public License
- *  along with this program; if not, write to the Free Software
- *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
- */
-
-#include <common.h>
-#include <ext2fs.h>
-#include <malloc.h>
-#include <asm/byteorder.h>
-
-extern int ext2fs_devread (int sector, int byte_offset, int byte_len,
-                          char *buf);
-
-/* Magic value used to identify an ext2 filesystem.  */
-#define        EXT2_MAGIC              0xEF53
-/* Amount of indirect blocks in an inode.  */
-#define INDIRECT_BLOCKS                12
-/* Maximum lenght of a pathname.  */
-#define EXT2_PATH_MAX          4096
-/* Maximum nesting of symlinks, used to prevent a loop.  */
-#define        EXT2_MAX_SYMLINKCNT     8
-
-/* Filetype used in directory entry.  */
-#define        FILETYPE_UNKNOWN        0
-#define        FILETYPE_REG            1
-#define        FILETYPE_DIRECTORY      2
-#define        FILETYPE_SYMLINK        7
-
-/* Filetype information as used in inodes.  */
-#define FILETYPE_INO_MASK      0170000
-#define FILETYPE_INO_REG       0100000
-#define FILETYPE_INO_DIRECTORY 0040000
-#define FILETYPE_INO_SYMLINK   0120000
-
-/* Bits used as offset in sector */
-#define DISK_SECTOR_BITS        9
-
-/* Log2 size of ext2 block in 512 blocks.  */
-#define LOG2_EXT2_BLOCK_SIZE(data) (__le32_to_cpu (data->sblock.log2_block_size) + 1)
-
-/* Log2 size of ext2 block in bytes.  */
-#define LOG2_BLOCK_SIZE(data)     (__le32_to_cpu (data->sblock.log2_block_size) + 10)
-
-/* The size of an ext2 block in bytes.  */
-#define EXT2_BLOCK_SIZE(data)     (1 << LOG2_BLOCK_SIZE(data))
-
-/* The ext2 superblock.  */
-struct ext2_sblock {
-       uint32_t total_inodes;
-       uint32_t total_blocks;
-       uint32_t reserved_blocks;
-       uint32_t free_blocks;
-       uint32_t free_inodes;
-       uint32_t first_data_block;
-       uint32_t log2_block_size;
-       uint32_t log2_fragment_size;
-       uint32_t blocks_per_group;
-       uint32_t fragments_per_group;
-       uint32_t inodes_per_group;
-       uint32_t mtime;
-       uint32_t utime;
-       uint16_t mnt_count;
-       uint16_t max_mnt_count;
-       uint16_t magic;
-       uint16_t fs_state;
-       uint16_t error_handling;
-       uint16_t minor_revision_level;
-       uint32_t lastcheck;
-       uint32_t checkinterval;
-       uint32_t creator_os;
-       uint32_t revision_level;
-       uint16_t uid_reserved;
-       uint16_t gid_reserved;
-       uint32_t first_inode;
-       uint16_t inode_size;
-       uint16_t block_group_number;
-       uint32_t feature_compatibility;
-       uint32_t feature_incompat;
-       uint32_t feature_ro_compat;
-       uint32_t unique_id[4];
-       char volume_name[16];
-       char last_mounted_on[64];
-       uint32_t compression_info;
-};
-
-/* The ext2 blockgroup.  */
-struct ext2_block_group {
-       uint32_t block_id;
-       uint32_t inode_id;
-       uint32_t inode_table_id;
-       uint16_t free_blocks;
-       uint16_t free_inodes;
-       uint16_t used_dir_cnt;
-       uint32_t reserved[3];
-};
-
-/* The ext2 inode.  */
-struct ext2_inode {
-       uint16_t mode;
-       uint16_t uid;
-       uint32_t size;
-       uint32_t atime;
-       uint32_t ctime;
-       uint32_t mtime;
-       uint32_t dtime;
-       uint16_t gid;
-       uint16_t nlinks;
-       uint32_t blockcnt;      /* Blocks of 512 bytes!! */
-       uint32_t flags;
-       uint32_t osd1;
-       union {
-               struct datablocks {
-                       uint32_t dir_blocks[INDIRECT_BLOCKS];
-                       uint32_t indir_block;
-                       uint32_t double_indir_block;
-                       uint32_t tripple_indir_block;
-               } blocks;
-               char symlink[60];
-       } b;
-       uint32_t version;
-       uint32_t acl;
-       uint32_t dir_acl;
-       uint32_t fragment_addr;
-       uint32_t osd2[3];
-};
-
-/* The header of an ext2 directory entry.  */
-struct ext2_dirent {
-       uint32_t inode;
-       uint16_t direntlen;
-       uint8_t namelen;
-       uint8_t filetype;
-};
-
-struct ext2fs_node {
-       struct ext2_data *data;
-       struct ext2_inode inode;
-       int ino;
-       int inode_read;
-};
-
-/* Information about a "mounted" ext2 filesystem.  */
-struct ext2_data {
-       struct ext2_sblock sblock;
-       struct ext2_inode *inode;
-       struct ext2fs_node diropen;
-};
-
-
-typedef struct ext2fs_node *ext2fs_node_t;
-
-struct ext2_data *ext2fs_root = NULL;
-ext2fs_node_t ext2fs_file = NULL;
-int symlinknest = 0;
-uint32_t *indir1_block = NULL;
-int indir1_size = 0;
-int indir1_blkno = -1;
-uint32_t *indir2_block = NULL;
-int indir2_size = 0;
-int indir2_blkno = -1;
-static unsigned int inode_size;
-
-
-static int ext2fs_blockgroup
-       (struct ext2_data *data, int group, struct ext2_block_group *blkgrp) {
-       unsigned int blkno;
-       unsigned int blkoff;
-       unsigned int desc_per_blk;
-
-       desc_per_blk = EXT2_BLOCK_SIZE(data) / sizeof(struct ext2_block_group);
-
-       blkno = __le32_to_cpu(data->sblock.first_data_block) + 1 +
-       group / desc_per_blk;
-       blkoff = (group % desc_per_blk) * sizeof(struct ext2_block_group);
-#ifdef DEBUG
-       printf ("ext2fs read %d group descriptor (blkno %d blkoff %d)\n",
-               group, blkno, blkoff);
-#endif
-       return (ext2fs_devread (blkno << LOG2_EXT2_BLOCK_SIZE(data),
-               blkoff, sizeof(struct ext2_block_group), (char *)blkgrp));
-
-}
-
-
-static int ext2fs_read_inode
-       (struct ext2_data *data, int ino, struct ext2_inode *inode) {
-       struct ext2_block_group blkgrp;
-       struct ext2_sblock *sblock = &data->sblock;
-       int inodes_per_block;
-       int status;
-
-       unsigned int blkno;
-       unsigned int blkoff;
-
-#ifdef DEBUG
-       printf ("ext2fs read inode %d, inode_size %d\n", ino, inode_size);
-#endif
-       /* It is easier to calculate if the first inode is 0.  */
-       ino--;
-       status = ext2fs_blockgroup (data, ino / __le32_to_cpu
-                                   (sblock->inodes_per_group), &blkgrp);
-       if (status == 0) {
-               return (0);
-       }
-
-       inodes_per_block = EXT2_BLOCK_SIZE(data) / inode_size;
-
-       blkno = __le32_to_cpu (blkgrp.inode_table_id) +
-               (ino % __le32_to_cpu (sblock->inodes_per_group))
-               / inodes_per_block;
-       blkoff = (ino % inodes_per_block) * inode_size;
-#ifdef DEBUG
-       printf ("ext2fs read inode blkno %d blkoff %d\n", blkno, blkoff);
-#endif
-       /* Read the inode.  */
-       status = ext2fs_devread (blkno << LOG2_EXT2_BLOCK_SIZE (data), blkoff,
-                                sizeof (struct ext2_inode), (char *) inode);
-       if (status == 0) {
-               return (0);
-       }
-
-       return (1);
-}
-
-
-void ext2fs_free_node (ext2fs_node_t node, ext2fs_node_t currroot) {
-       if ((node != &ext2fs_root->diropen) && (node != currroot)) {
-               free (node);
-       }
-}
-
-
-static int ext2fs_read_block (ext2fs_node_t node, int fileblock) {
-       struct ext2_data *data = node->data;
-       struct ext2_inode *inode = &node->inode;
-       int blknr;
-       int blksz = EXT2_BLOCK_SIZE (data);
-       int log2_blksz = LOG2_EXT2_BLOCK_SIZE (data);
-       int status;
-
-       /* Direct blocks.  */
-       if (fileblock < INDIRECT_BLOCKS) {
-               blknr = __le32_to_cpu (inode->b.blocks.dir_blocks[fileblock]);
-       }
-       /* Indirect.  */
-       else if (fileblock < (INDIRECT_BLOCKS + (blksz / 4))) {
-               if (indir1_block == NULL) {
-                       indir1_block = (uint32_t *) memalign(ARCH_DMA_MINALIGN,
-                                                            blksz);
-                       if (indir1_block == NULL) {
-                               printf ("** ext2fs read block (indir 1) malloc failed. **\n");
-                               return (-1);
-                       }
-                       indir1_size = blksz;
-                       indir1_blkno = -1;
-               }
-               if (blksz != indir1_size) {
-                       free (indir1_block);
-                       indir1_block = NULL;
-                       indir1_size = 0;
-                       indir1_blkno = -1;
-                       indir1_block = (uint32_t *) memalign(ARCH_DMA_MINALIGN,
-                                                            blksz);
-                       if (indir1_block == NULL) {
-                               printf ("** ext2fs read block (indir 1) malloc failed. **\n");
-                               return (-1);
-                       }
-                       indir1_size = blksz;
-               }
-               if ((__le32_to_cpu (inode->b.blocks.indir_block) <<
-                    log2_blksz) != indir1_blkno) {
-                       status = ext2fs_devread (__le32_to_cpu(inode->b.blocks.indir_block) << log2_blksz,
-                                                0, blksz,
-                                                (char *) indir1_block);
-                       if (status == 0) {
-                               printf ("** ext2fs read block (indir 1) failed. **\n");
-                               return (0);
-                       }
-                       indir1_blkno =
-                               __le32_to_cpu (inode->b.blocks.
-                                              indir_block) << log2_blksz;
-               }
-               blknr = __le32_to_cpu (indir1_block
-                                      [fileblock - INDIRECT_BLOCKS]);
-       }
-       /* Double indirect.  */
-       else if (fileblock <
-                (INDIRECT_BLOCKS + (blksz / 4 * (blksz / 4 + 1)))) {
-               unsigned int perblock = blksz / 4;
-               unsigned int rblock = fileblock - (INDIRECT_BLOCKS
-                                                  + blksz / 4);
-
-               if (indir1_block == NULL) {
-                       indir1_block = (uint32_t *) memalign(ARCH_DMA_MINALIGN,
-                                                            blksz);
-                       if (indir1_block == NULL) {
-                               printf ("** ext2fs read block (indir 2 1) malloc failed. **\n");
-                               return (-1);
-                       }
-                       indir1_size = blksz;
-                       indir1_blkno = -1;
-               }
-               if (blksz != indir1_size) {
-                       free (indir1_block);
-                       indir1_block = NULL;
-                       indir1_size = 0;
-                       indir1_blkno = -1;
-                       indir1_block = (uint32_t *) memalign(ARCH_DMA_MINALIGN,
-                                                            blksz);
-                       if (indir1_block == NULL) {
-                               printf ("** ext2fs read block (indir 2 1) malloc failed. **\n");
-                               return (-1);
-                       }
-                       indir1_size = blksz;
-               }
-               if ((__le32_to_cpu (inode->b.blocks.double_indir_block) <<
-                    log2_blksz) != indir1_blkno) {
-                       status = ext2fs_devread (__le32_to_cpu(inode->b.blocks.double_indir_block) << log2_blksz,
-                                               0, blksz,
-                                               (char *) indir1_block);
-                       if (status == 0) {
-                               printf ("** ext2fs read block (indir 2 1) failed. **\n");
-                               return (-1);
-                       }
-                       indir1_blkno =
-                               __le32_to_cpu (inode->b.blocks.double_indir_block) << log2_blksz;
-               }
-
-               if (indir2_block == NULL) {
-                       indir2_block = (uint32_t *) memalign(ARCH_DMA_MINALIGN,
-                                                            blksz);
-                       if (indir2_block == NULL) {
-                               printf ("** ext2fs read block (indir 2 2) malloc failed. **\n");
-                               return (-1);
-                       }
-                       indir2_size = blksz;
-                       indir2_blkno = -1;
-               }
-               if (blksz != indir2_size) {
-                       free (indir2_block);
-                       indir2_block = NULL;
-                       indir2_size = 0;
-                       indir2_blkno = -1;
-                       indir2_block = (uint32_t *) memalign(ARCH_DMA_MINALIGN,
-                                                            blksz);
-                       if (indir2_block == NULL) {
-                               printf ("** ext2fs read block (indir 2 2) malloc failed. **\n");
-                               return (-1);
-                       }
-                       indir2_size = blksz;
-               }
-               if ((__le32_to_cpu (indir1_block[rblock / perblock]) <<
-                    log2_blksz) != indir2_blkno) {
-                       status = ext2fs_devread (__le32_to_cpu(indir1_block[rblock / perblock]) << log2_blksz,
-                                                0, blksz,
-                                                (char *) indir2_block);
-                       if (status == 0) {
-                               printf ("** ext2fs read block (indir 2 2) failed. **\n");
-                               return (-1);
-                       }
-                       indir2_blkno =
-                               __le32_to_cpu (indir1_block[rblock / perblock]) << log2_blksz;
-               }
-               blknr = __le32_to_cpu (indir2_block[rblock % perblock]);
-       }
-       /* Tripple indirect.  */
-       else {
-               printf ("** ext2fs doesn't support tripple indirect blocks. **\n");
-               return (-1);
-       }
-#ifdef DEBUG
-       printf ("ext2fs_read_block %08x\n", blknr);
-#endif
-       return (blknr);
-}
-
-
-int ext2fs_read_file
-       (ext2fs_node_t node, int pos, unsigned int len, char *buf) {
-       int i;
-       int blockcnt;
-       int log2blocksize = LOG2_EXT2_BLOCK_SIZE (node->data);
-       int blocksize = 1 << (log2blocksize + DISK_SECTOR_BITS);
-       unsigned int filesize = __le32_to_cpu(node->inode.size);
-
-       /* Adjust len so it we can't read past the end of the file.  */
-       if (len > filesize) {
-               len = filesize;
-       }
-       blockcnt = ((len + pos) + blocksize - 1) / blocksize;
-
-       for (i = pos / blocksize; i < blockcnt; i++) {
-               int blknr;
-               int blockoff = pos % blocksize;
-               int blockend = blocksize;
-
-               int skipfirst = 0;
-
-               blknr = ext2fs_read_block (node, i);
-               if (blknr < 0) {
-                       return (-1);
-               }
-
-               /* Last block.  */
-               if (i == blockcnt - 1) {
-                       blockend = (len + pos) % blocksize;
-
-                       /* The last portion is exactly blocksize.  */
-                       if (!blockend) {
-                               blockend = blocksize;
-                       }
-               }
-
-               /* First block.  */
-               if (i == pos / blocksize) {
-                       skipfirst = blockoff;
-                       blockend -= skipfirst;
-               }
-
-               /* grab middle blocks in one go */
-               if (i != pos / blocksize && i < blockcnt - 1 && blockcnt > 3) {
-                       int oldblk = blknr;
-                       int blocknxt = ext2fs_read_block(node, i + 1);
-                       while (i < blockcnt - 1) {
-                               if (blocknxt == (oldblk + 1)) {
-                                       oldblk = blocknxt;
-                                       i++;
-                               } else {
-                                       blocknxt = ext2fs_read_block(node, i);
-                                       break;
-                               }
-                               blocknxt = ext2fs_read_block(node, i);
-                       }
-
-                       if (oldblk == blknr)
-                               blockend = blocksize;
-                       else
-                               blockend = (1 + blocknxt - blknr) * blocksize;
-               }
-
-               blknr = blknr << log2blocksize;
-
-               /* If the block number is 0 this block is not stored on disk but
-                  is zero filled instead.  */
-               if (blknr) {
-                       int status;
-
-                       status = ext2fs_devread (blknr, skipfirst, blockend, buf);
-                       if (status == 0) {
-                               return (-1);
-                       }
-               } else {
-                       memset (buf, 0, blocksize - skipfirst);
-               }
-               buf += blockend - skipfirst;
-       }
-       return (len);
-}
-
-
-static int ext2fs_iterate_dir (ext2fs_node_t dir, char *name, ext2fs_node_t * fnode, int *ftype)
-{
-       unsigned int fpos = 0;
-       int status;
-       struct ext2fs_node *diro = (struct ext2fs_node *) dir;
-
-#ifdef DEBUG
-       if (name != NULL)
-               printf ("Iterate dir %s\n", name);
-#endif /* of DEBUG */
-       if (!diro->inode_read) {
-               status = ext2fs_read_inode (diro->data, diro->ino,
-                                           &diro->inode);
-               if (status == 0) {
-                       return (0);
-               }
-       }
-       /* Search the file.  */
-       while (fpos < __le32_to_cpu (diro->inode.size)) {
-               struct ext2_dirent dirent;
-
-               status = ext2fs_read_file (diro, fpos,
-                                          sizeof (struct ext2_dirent),
-                                          (char *) &dirent);
-               if (status < 1) {
-                       return (0);
-               }
-               if (dirent.namelen != 0) {
-                       char filename[dirent.namelen + 1];
-                       ext2fs_node_t fdiro;
-                       int type = FILETYPE_UNKNOWN;
-
-                       status = ext2fs_read_file (diro,
-                                                  fpos + sizeof (struct ext2_dirent),
-                                                  dirent.namelen, filename);
-                       if (status < 1) {
-                               return (0);
-                       }
-                       fdiro = malloc (sizeof (struct ext2fs_node));
-                       if (!fdiro) {
-                               return (0);
-                       }
-
-                       fdiro->data = diro->data;
-                       fdiro->ino = __le32_to_cpu (dirent.inode);
-
-                       filename[dirent.namelen] = '\0';
-
-                       if (dirent.filetype != FILETYPE_UNKNOWN) {
-                               fdiro->inode_read = 0;
-
-                               if (dirent.filetype == FILETYPE_DIRECTORY) {
-                                       type = FILETYPE_DIRECTORY;
-                               } else if (dirent.filetype ==
-                                          FILETYPE_SYMLINK) {
-                                       type = FILETYPE_SYMLINK;
-                               } else if (dirent.filetype == FILETYPE_REG) {
-                                       type = FILETYPE_REG;
-                               }
-                       } else {
-                               /* The filetype can not be read from the dirent, get it from inode */
-
-                               status = ext2fs_read_inode (diro->data,
-                                                           __le32_to_cpu(dirent.inode),
-                                                           &fdiro->inode);
-                               if (status == 0) {
-                                       free (fdiro);
-                                       return (0);
-                               }
-                               fdiro->inode_read = 1;
-
-                               if ((__le16_to_cpu (fdiro->inode.mode) &
-                                    FILETYPE_INO_MASK) ==
-                                   FILETYPE_INO_DIRECTORY) {
-                                       type = FILETYPE_DIRECTORY;
-                               } else if ((__le16_to_cpu (fdiro->inode.mode)
-                                           & FILETYPE_INO_MASK) ==
-                                          FILETYPE_INO_SYMLINK) {
-                                       type = FILETYPE_SYMLINK;
-                               } else if ((__le16_to_cpu (fdiro->inode.mode)
-                                           & FILETYPE_INO_MASK) ==
-                                          FILETYPE_INO_REG) {
-                                       type = FILETYPE_REG;
-                               }
-                       }
-#ifdef DEBUG
-                       printf ("iterate >%s<\n", filename);
-#endif /* of DEBUG */
-                       if ((name != NULL) && (fnode != NULL)
-                           && (ftype != NULL)) {
-                               if (strcmp (filename, name) == 0) {
-                                       *ftype = type;
-                                       *fnode = fdiro;
-                                       return (1);
-                               }
-                       } else {
-                               if (fdiro->inode_read == 0) {
-                                       status = ext2fs_read_inode (diro->data,
-                                                           __le32_to_cpu (dirent.inode),
-                                                           &fdiro->inode);
-                                       if (status == 0) {
-                                               free (fdiro);
-                                               return (0);
-                                       }
-                                       fdiro->inode_read = 1;
-                               }
-                               switch (type) {
-                               case FILETYPE_DIRECTORY:
-                                       printf ("<DIR> ");
-                                       break;
-                               case FILETYPE_SYMLINK:
-                                       printf ("<SYM> ");
-                                       break;
-                               case FILETYPE_REG:
-                                       printf ("      ");
-                                       break;
-                               default:
-                                       printf ("< ? > ");
-                                       break;
-                               }
-                               printf ("%10d %s\n",
-                                       __le32_to_cpu (fdiro->inode.size),
-                                       filename);
-                       }
-                       free (fdiro);
-               }
-               fpos += __le16_to_cpu (dirent.direntlen);
-       }
-       return (0);
-}
-
-
-static char *ext2fs_read_symlink (ext2fs_node_t node) {
-       char *symlink;
-       struct ext2fs_node *diro = node;
-       int status;
-
-       if (!diro->inode_read) {
-               status = ext2fs_read_inode (diro->data, diro->ino,
-                                           &diro->inode);
-               if (status == 0) {
-                       return (0);
-               }
-       }
-       symlink = malloc (__le32_to_cpu (diro->inode.size) + 1);
-       if (!symlink) {
-               return (0);
-       }
-       /* If the filesize of the symlink is bigger than
-          60 the symlink is stored in a separate block,
-          otherwise it is stored in the inode.  */
-       if (__le32_to_cpu (diro->inode.size) <= 60) {
-               strncpy (symlink, diro->inode.b.symlink,
-                        __le32_to_cpu (diro->inode.size));
-       } else {
-               status = ext2fs_read_file (diro, 0,
-                                          __le32_to_cpu (diro->inode.size),
-                                          symlink);
-               if (status == 0) {
-                       free (symlink);
-                       return (0);
-               }
-       }
-       symlink[__le32_to_cpu (diro->inode.size)] = '\0';
-       return (symlink);
-}
-
-
-int ext2fs_find_file1
-       (const char *currpath,
-        ext2fs_node_t currroot, ext2fs_node_t * currfound, int *foundtype) {
-       char fpath[strlen (currpath) + 1];
-       char *name = fpath;
-       char *next;
-       int status;
-       int type = FILETYPE_DIRECTORY;
-       ext2fs_node_t currnode = currroot;
-       ext2fs_node_t oldnode = currroot;
-
-       strncpy (fpath, currpath, strlen (currpath) + 1);
-
-       /* Remove all leading slashes.  */
-       while (*name == '/') {
-               name++;
-       }
-       if (!*name) {
-               *currfound = currnode;
-               return (1);
-       }
-
-       for (;;) {
-               int found;
-
-               /* Extract the actual part from the pathname.  */
-               next = strchr (name, '/');
-               if (next) {
-                       /* Remove all leading slashes.  */
-                       while (*next == '/') {
-                               *(next++) = '\0';
-                       }
-               }
-
-               /* At this point it is expected that the current node is a directory, check if this is true.  */
-               if (type != FILETYPE_DIRECTORY) {
-                       ext2fs_free_node (currnode, currroot);
-                       return (0);
-               }
-
-               oldnode = currnode;
-
-               /* Iterate over the directory.  */
-               found = ext2fs_iterate_dir (currnode, name, &currnode, &type);
-               if (found == 0) {
-                       return (0);
-               }
-               if (found == -1) {
-                       break;
-               }
-
-               /* Read in the symlink and follow it.  */
-               if (type == FILETYPE_SYMLINK) {
-                       char *symlink;
-
-                       /* Test if the symlink does not loop.  */
-                       if (++symlinknest == 8) {
-                               ext2fs_free_node (currnode, currroot);
-                               ext2fs_free_node (oldnode, currroot);
-                               return (0);
-                       }
-
-                       symlink = ext2fs_read_symlink (currnode);
-                       ext2fs_free_node (currnode, currroot);
-
-                       if (!symlink) {
-                               ext2fs_free_node (oldnode, currroot);
-                               return (0);
-                       }
-#ifdef DEBUG
-                       printf ("Got symlink >%s<\n", symlink);
-#endif /* of DEBUG */
-                       /* The symlink is an absolute path, go back to the root inode.  */
-                       if (symlink[0] == '/') {
-                               ext2fs_free_node (oldnode, currroot);
-                               oldnode = &ext2fs_root->diropen;
-                       }
-
-                       /* Lookup the node the symlink points to.  */
-                       status = ext2fs_find_file1 (symlink, oldnode,
-                                                   &currnode, &type);
-
-                       free (symlink);
-
-                       if (status == 0) {
-                               ext2fs_free_node (oldnode, currroot);
-                               return (0);
-                       }
-               }
-
-               ext2fs_free_node (oldnode, currroot);
-
-               /* Found the node!  */
-               if (!next || *next == '\0') {
-                       *currfound = currnode;
-                       *foundtype = type;
-                       return (1);
-               }
-               name = next;
-       }
-       return (-1);
-}
-
-
-int ext2fs_find_file
-       (const char *path,
-        ext2fs_node_t rootnode, ext2fs_node_t * foundnode, int expecttype) {
-       int status;
-       int foundtype = FILETYPE_DIRECTORY;
-
-
-       symlinknest = 0;
-       if (!path) {
-               return (0);
-       }
-
-       status = ext2fs_find_file1 (path, rootnode, foundnode, &foundtype);
-       if (status == 0) {
-               return (0);
-       }
-       /* Check if the node that was found was of the expected type.  */
-       if ((expecttype == FILETYPE_REG) && (foundtype != expecttype)) {
-               return (0);
-       } else if ((expecttype == FILETYPE_DIRECTORY)
-                  && (foundtype != expecttype)) {
-               return (0);
-       }
-       return (1);
-}
-
-
-int ext2fs_ls (const char *dirname) {
-       ext2fs_node_t dirnode;
-       int status;
-
-       if (ext2fs_root == NULL) {
-               return (0);
-       }
-
-       status = ext2fs_find_file (dirname, &ext2fs_root->diropen, &dirnode,
-                                  FILETYPE_DIRECTORY);
-       if (status != 1) {
-               printf ("** Can not find directory. **\n");
-               return (1);
-       }
-       ext2fs_iterate_dir (dirnode, NULL, NULL, NULL);
-       ext2fs_free_node (dirnode, &ext2fs_root->diropen);
-       return (0);
-}
-
-
-int ext2fs_open (const char *filename) {
-       ext2fs_node_t fdiro = NULL;
-       int status;
-       int len;
-
-       if (ext2fs_root == NULL) {
-               return (-1);
-       }
-       ext2fs_file = NULL;
-       status = ext2fs_find_file (filename, &ext2fs_root->diropen, &fdiro,
-                                  FILETYPE_REG);
-       if (status == 0) {
-               goto fail;
-       }
-       if (!fdiro->inode_read) {
-               status = ext2fs_read_inode (fdiro->data, fdiro->ino,
-                                           &fdiro->inode);
-               if (status == 0) {
-                       goto fail;
-               }
-       }
-       len = __le32_to_cpu (fdiro->inode.size);
-       ext2fs_file = fdiro;
-       return (len);
-
-fail:
-       ext2fs_free_node (fdiro, &ext2fs_root->diropen);
-       return (-1);
-}
-
-
-int ext2fs_close (void
-       ) {
-       if ((ext2fs_file != NULL) && (ext2fs_root != NULL)) {
-               ext2fs_free_node (ext2fs_file, &ext2fs_root->diropen);
-               ext2fs_file = NULL;
-       }
-       if (ext2fs_root != NULL) {
-               free (ext2fs_root);
-               ext2fs_root = NULL;
-       }
-       if (indir1_block != NULL) {
-               free (indir1_block);
-               indir1_block = NULL;
-               indir1_size = 0;
-               indir1_blkno = -1;
-       }
-       if (indir2_block != NULL) {
-               free (indir2_block);
-               indir2_block = NULL;
-               indir2_size = 0;
-               indir2_blkno = -1;
-       }
-       return (0);
-}
-
-
-int ext2fs_read (char *buf, unsigned len) {
-       int status;
-
-       if (ext2fs_root == NULL) {
-               return (0);
-       }
-
-       if (ext2fs_file == NULL) {
-               return (0);
-       }
-
-       status = ext2fs_read_file (ext2fs_file, 0, len, buf);
-       return (status);
-}
-
-
-int ext2fs_mount (unsigned part_length) {
-       struct ext2_data *data;
-       int status;
-
-       data = malloc (sizeof (struct ext2_data));
-       if (!data) {
-               return (0);
-       }
-       /* Read the superblock.  */
-       status = ext2fs_devread (1 * 2, 0, sizeof (struct ext2_sblock),
-                                (char *) &data->sblock);
-       if (status == 0) {
-               goto fail;
-       }
-       /* Make sure this is an ext2 filesystem.  */
-       if (__le16_to_cpu (data->sblock.magic) != EXT2_MAGIC) {
-               goto fail;
-       }
-       if (__le32_to_cpu(data->sblock.revision_level == 0)) {
-               inode_size = 128;
-       } else {
-               inode_size = __le16_to_cpu(data->sblock.inode_size);
-       }
-#ifdef DEBUG
-       printf("EXT2 rev %d, inode_size %d\n",
-                       __le32_to_cpu(data->sblock.revision_level), inode_size);
-#endif
-       data->diropen.data = data;
-       data->diropen.ino = 2;
-       data->diropen.inode_read = 1;
-       data->inode = &data->diropen.inode;
-
-       status = ext2fs_read_inode (data, 2, data->inode);
-       if (status == 0) {
-               goto fail;
-       }
-
-       ext2fs_root = data;
-
-       return (1);
-
-fail:
-       printf ("Failed to mount ext2 filesystem...\n");
-       free (data);
-       ext2fs_root = NULL;
-       return (0);
-}
similarity index 86%
rename from fs/ext2/Makefile
rename to fs/ext4/Makefile
index 3c65d252f582cb597754e7dfd5e6032a65453236..82cd9ae16a58daef3cff28413a3014e88f755558 100644 (file)
 
 include $(TOPDIR)/config.mk
 
-LIB    = $(obj)libext2fs.o
+LIB    = $(obj)libext4fs.o
 
 AOBJS  =
-COBJS-$(CONFIG_CMD_EXT2) := ext2fs.o dev.o
+COBJS-$(CONFIG_CMD_EXT4) := ext4fs.o ext4_common.o dev.o
+ifndef CONFIG_CMD_EXT4
+COBJS-$(CONFIG_CMD_EXT2) := ext4fs.o ext4_common.o dev.o
+endif
+COBJS-$(CONFIG_CMD_EXT4_WRITE) += ext4_journal.o crc16.o
 
 SRCS   := $(AOBJS:.o=.S) $(COBJS-y:.o=.c)
 OBJS   := $(addprefix $(obj),$(AOBJS) $(COBJS-y))
 
-#CPPFLAGS +=
 
 all:   $(LIB) $(AOBJS)
 
diff --git a/fs/ext4/crc16.c b/fs/ext4/crc16.c
new file mode 100644 (file)
index 0000000..3afb34d
--- /dev/null
@@ -0,0 +1,62 @@
+/*
+ *      crc16.c
+ *
+ * This source code is licensed under the GNU General Public License,
+ * Version 2. See the file COPYING for more details.
+ */
+
+#include <common.h>
+#include <asm/byteorder.h>
+#include <linux/stat.h>
+#include "crc16.h"
+
+/** CRC table for the CRC-16. The poly is 0x8005 (x16 + x15 + x2 + 1) */
+static __u16 const crc16_table[256] = {
+       0x0000, 0xC0C1, 0xC181, 0x0140, 0xC301, 0x03C0, 0x0280, 0xC241,
+       0xC601, 0x06C0, 0x0780, 0xC741, 0x0500, 0xC5C1, 0xC481, 0x0440,
+       0xCC01, 0x0CC0, 0x0D80, 0xCD41, 0x0F00, 0xCFC1, 0xCE81, 0x0E40,
+       0x0A00, 0xCAC1, 0xCB81, 0x0B40, 0xC901, 0x09C0, 0x0880, 0xC841,
+       0xD801, 0x18C0, 0x1980, 0xD941, 0x1B00, 0xDBC1, 0xDA81, 0x1A40,
+       0x1E00, 0xDEC1, 0xDF81, 0x1F40, 0xDD01, 0x1DC0, 0x1C80, 0xDC41,
+       0x1400, 0xD4C1, 0xD581, 0x1540, 0xD701, 0x17C0, 0x1680, 0xD641,
+       0xD201, 0x12C0, 0x1380, 0xD341, 0x1100, 0xD1C1, 0xD081, 0x1040,
+       0xF001, 0x30C0, 0x3180, 0xF141, 0x3300, 0xF3C1, 0xF281, 0x3240,
+       0x3600, 0xF6C1, 0xF781, 0x3740, 0xF501, 0x35C0, 0x3480, 0xF441,
+       0x3C00, 0xFCC1, 0xFD81, 0x3D40, 0xFF01, 0x3FC0, 0x3E80, 0xFE41,
+       0xFA01, 0x3AC0, 0x3B80, 0xFB41, 0x3900, 0xF9C1, 0xF881, 0x3840,
+       0x2800, 0xE8C1, 0xE981, 0x2940, 0xEB01, 0x2BC0, 0x2A80, 0xEA41,
+       0xEE01, 0x2EC0, 0x2F80, 0xEF41, 0x2D00, 0xEDC1, 0xEC81, 0x2C40,
+       0xE401, 0x24C0, 0x2580, 0xE541, 0x2700, 0xE7C1, 0xE681, 0x2640,
+       0x2200, 0xE2C1, 0xE381, 0x2340, 0xE101, 0x21C0, 0x2080, 0xE041,
+       0xA001, 0x60C0, 0x6180, 0xA141, 0x6300, 0xA3C1, 0xA281, 0x6240,
+       0x6600, 0xA6C1, 0xA781, 0x6740, 0xA501, 0x65C0, 0x6480, 0xA441,
+       0x6C00, 0xACC1, 0xAD81, 0x6D40, 0xAF01, 0x6FC0, 0x6E80, 0xAE41,
+       0xAA01, 0x6AC0, 0x6B80, 0xAB41, 0x6900, 0xA9C1, 0xA881, 0x6840,
+       0x7800, 0xB8C1, 0xB981, 0x7940, 0xBB01, 0x7BC0, 0x7A80, 0xBA41,
+       0xBE01, 0x7EC0, 0x7F80, 0xBF41, 0x7D00, 0xBDC1, 0xBC81, 0x7C40,
+       0xB401, 0x74C0, 0x7580, 0xB541, 0x7700, 0xB7C1, 0xB681, 0x7640,
+       0x7200, 0xB2C1, 0xB381, 0x7340, 0xB101, 0x71C0, 0x7080, 0xB041,
+       0x5000, 0x90C1, 0x9181, 0x5140, 0x9301, 0x53C0, 0x5280, 0x9241,
+       0x9601, 0x56C0, 0x5780, 0x9741, 0x5500, 0x95C1, 0x9481, 0x5440,
+       0x9C01, 0x5CC0, 0x5D80, 0x9D41, 0x5F00, 0x9FC1, 0x9E81, 0x5E40,
+       0x5A00, 0x9AC1, 0x9B81, 0x5B40, 0x9901, 0x59C0, 0x5880, 0x9841,
+       0x8801, 0x48C0, 0x4980, 0x8941, 0x4B00, 0x8BC1, 0x8A81, 0x4A40,
+       0x4E00, 0x8EC1, 0x8F81, 0x4F40, 0x8D01, 0x4DC0, 0x4C80, 0x8C41,
+       0x4400, 0x84C1, 0x8581, 0x4540, 0x8701, 0x47C0, 0x4680, 0x8641,
+       0x8201, 0x42C0, 0x4380, 0x8341, 0x4100, 0x81C1, 0x8081, 0x4040
+};
+
+/**
+ * Compute the CRC-16 for the data buffer
+*/
+
+unsigned int ext2fs_crc16(unsigned int crc,
+       const void *buffer, unsigned int len)
+{
+       const unsigned char *cp = buffer;
+
+       while (len--)
+               crc = (((crc >> 8) & 0xffU) ^
+                      crc16_table[(crc ^ *cp++) & 0xffU]) & 0x0000ffffU;
+       return crc;
+}
diff --git a/fs/ext4/crc16.h b/fs/ext4/crc16.h
new file mode 100644 (file)
index 0000000..5fd113a
--- /dev/null
@@ -0,0 +1,16 @@
+/*
+ * crc16.h - CRC-16 routine
+ * Implements the standard CRC-16:
+ *  Width 16
+ *  Poly  0x8005 (x16 + x15 + x2 + 1)
+ *  Init  0
+ *
+ * Copyright (c) 2005 Ben Gardner <bgardner@wabtec.com>
+ * This source code is licensed under the GNU General Public License,
+ * Version 2. See the file COPYING for more details.
+ */
+#ifndef __CRC16_H
+#define __CRC16_H
+extern unsigned int ext2fs_crc16(unsigned int crc,
+       const void *buffer, unsigned int len);
+#endif
diff --git a/fs/ext4/dev.c b/fs/ext4/dev.c
new file mode 100644 (file)
index 0000000..9e85228
--- /dev/null
@@ -0,0 +1,145 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * made from existing ext2/dev.c file of Uboot
+ * (C) Copyright 2004
+ * esd gmbh <www.esd-electronics.com>
+ * Reinhard Arlt <reinhard.arlt@esd-electronics.com>
+ *
+ * based on code of fs/reiserfs/dev.c by
+ *
+ * (C) Copyright 2003 - 2004
+ * Sysgo AG, <www.elinos.com>, Pavel Bartusek <pba@sysgo.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ *
+ */
+
+/*
+ * Changelog:
+ *     0.1 - Newly created file for ext4fs support. Taken from
+ *             fs/ext2/dev.c file in uboot.
+ */
+
+#include <common.h>
+#include <config.h>
+#include <ext_common.h>
+
+static block_dev_desc_t *ext4fs_block_dev_desc;
+static disk_partition_t part_info;
+
+int ext4fs_set_blk_dev(block_dev_desc_t *rbdd, int part)
+{
+       ext4fs_block_dev_desc = rbdd;
+
+       if (part == 0) {
+               /* disk doesn't use partition table */
+               part_info.start = 0;
+               part_info.size = rbdd->lba;
+               part_info.blksz = rbdd->blksz;
+       } else {
+               if (get_partition_info(ext4fs_block_dev_desc,
+                                       part, &part_info))
+                       return 0;
+       }
+       return part_info.size;
+}
+
+int ext4fs_devread(int sector, int byte_offset, int byte_len, char *buf)
+{
+       ALLOC_CACHE_ALIGN_BUFFER(char, sec_buf, SECTOR_SIZE);
+       unsigned block_len;
+
+       /* Check partition boundaries */
+       if ((sector < 0)
+           || ((sector + ((byte_offset + byte_len - 1) >> SECTOR_BITS)) >=
+               part_info.size)) {
+               printf("%s read outside partition %d\n", __func__, sector);
+               return 0;
+       }
+
+       /* Get the read to the beginning of a partition */
+       sector += byte_offset >> SECTOR_BITS;
+       byte_offset &= SECTOR_SIZE - 1;
+
+       debug(" <%d, %d, %d>\n", sector, byte_offset, byte_len);
+
+       if (ext4fs_block_dev_desc == NULL) {
+               printf("** Invalid Block Device Descriptor (NULL)\n");
+               return 0;
+       }
+
+       if (byte_offset != 0) {
+               /* read first part which isn't aligned with start of sector */
+               if (ext4fs_block_dev_desc->
+                   block_read(ext4fs_block_dev_desc->dev,
+                               part_info.start + sector, 1,
+                               (unsigned long *) sec_buf) != 1) {
+                       printf(" ** ext2fs_devread() read error **\n");
+                       return 0;
+               }
+               memcpy(buf, sec_buf + byte_offset,
+                       min(SECTOR_SIZE - byte_offset, byte_len));
+               buf += min(SECTOR_SIZE - byte_offset, byte_len);
+               byte_len -= min(SECTOR_SIZE - byte_offset, byte_len);
+               sector++;
+       }
+
+       if (byte_len == 0)
+               return 1;
+
+       /* read sector aligned part */
+       block_len = byte_len & ~(SECTOR_SIZE - 1);
+
+       if (block_len == 0) {
+               ALLOC_CACHE_ALIGN_BUFFER(u8, p, SECTOR_SIZE);
+
+               block_len = SECTOR_SIZE;
+               ext4fs_block_dev_desc->block_read(ext4fs_block_dev_desc->dev,
+                                                 part_info.start + sector,
+                                                 1, (unsigned long *)p);
+               memcpy(buf, p, byte_len);
+               return 1;
+       }
+
+       if (ext4fs_block_dev_desc->block_read(ext4fs_block_dev_desc->dev,
+                                              part_info.start + sector,
+                                              block_len / SECTOR_SIZE,
+                                              (unsigned long *) buf) !=
+                                              block_len / SECTOR_SIZE) {
+               printf(" ** %s read error - block\n", __func__);
+               return 0;
+       }
+       block_len = byte_len & ~(SECTOR_SIZE - 1);
+       buf += block_len;
+       byte_len -= block_len;
+       sector += block_len / SECTOR_SIZE;
+
+       if (byte_len != 0) {
+               /* read rest of data which are not in whole sector */
+               if (ext4fs_block_dev_desc->
+                   block_read(ext4fs_block_dev_desc->dev,
+                               part_info.start + sector, 1,
+                               (unsigned long *) sec_buf) != 1) {
+                       printf("* %s read error - last part\n", __func__);
+                       return 0;
+               }
+               memcpy(buf, sec_buf, byte_len);
+       }
+       return 1;
+}
diff --git a/fs/ext4/ext4_common.c b/fs/ext4/ext4_common.c
new file mode 100644 (file)
index 0000000..3deffd5
--- /dev/null
@@ -0,0 +1,2228 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * ext4ls and ext4load : Based on ext2 ls load support in Uboot.
+ *
+ * (C) Copyright 2004
+ * esd gmbh <www.esd-electronics.com>
+ * Reinhard Arlt <reinhard.arlt@esd-electronics.com>
+ *
+ * based on code from grub2 fs/ext2.c and fs/fshelp.c by
+ * GRUB  --  GRand Unified Bootloader
+ * Copyright (C) 2003, 2004  Free Software Foundation, Inc.
+ *
+ * ext4write : Based on generic ext4 protocol.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <common.h>
+#include <ext_common.h>
+#include <ext4fs.h>
+#include <malloc.h>
+#include <stddef.h>
+#include <linux/stat.h>
+#include <linux/time.h>
+#include <asm/byteorder.h>
+#include "ext4_common.h"
+
+struct ext2_data *ext4fs_root;
+struct ext2fs_node *ext4fs_file;
+uint32_t *ext4fs_indir1_block;
+int ext4fs_indir1_size;
+int ext4fs_indir1_blkno = -1;
+uint32_t *ext4fs_indir2_block;
+int ext4fs_indir2_size;
+int ext4fs_indir2_blkno = -1;
+
+uint32_t *ext4fs_indir3_block;
+int ext4fs_indir3_size;
+int ext4fs_indir3_blkno = -1;
+struct ext2_inode *g_parent_inode;
+static int symlinknest;
+
+#if defined(CONFIG_CMD_EXT4_WRITE)
+uint32_t ext4fs_div_roundup(uint32_t size, uint32_t n)
+{
+       uint32_t res = size / n;
+       if (res * n != size)
+               res++;
+
+       return res;
+}
+
+void put_ext4(uint64_t off, void *buf, uint32_t size)
+{
+       uint64_t startblock;
+       uint64_t remainder;
+       unsigned char *temp_ptr = NULL;
+       ALLOC_CACHE_ALIGN_BUFFER(unsigned char, sec_buf, SECTOR_SIZE);
+       struct ext_filesystem *fs = get_fs();
+
+       startblock = off / (uint64_t)SECTOR_SIZE;
+       startblock += part_offset;
+       remainder = off % (uint64_t)SECTOR_SIZE;
+       remainder &= SECTOR_SIZE - 1;
+
+       if (fs->dev_desc == NULL)
+               return;
+
+       if ((startblock + (size / SECTOR_SIZE)) >
+           (part_offset + fs->total_sect)) {
+               printf("part_offset is %lu\n", part_offset);
+               printf("total_sector is %llu\n", fs->total_sect);
+               printf("error: overflow occurs\n");
+               return;
+       }
+
+       if (remainder) {
+               if (fs->dev_desc->block_read) {
+                       fs->dev_desc->block_read(fs->dev_desc->dev,
+                                                startblock, 1, sec_buf);
+                       temp_ptr = sec_buf;
+                       memcpy((temp_ptr + remainder),
+                              (unsigned char *)buf, size);
+                       fs->dev_desc->block_write(fs->dev_desc->dev,
+                                                 startblock, 1, sec_buf);
+               }
+       } else {
+               if (size / SECTOR_SIZE != 0) {
+                       fs->dev_desc->block_write(fs->dev_desc->dev,
+                                                 startblock,
+                                                 size / SECTOR_SIZE,
+                                                 (unsigned long *)buf);
+               } else {
+                       fs->dev_desc->block_read(fs->dev_desc->dev,
+                                                startblock, 1, sec_buf);
+                       temp_ptr = sec_buf;
+                       memcpy(temp_ptr, buf, size);
+                       fs->dev_desc->block_write(fs->dev_desc->dev,
+                                                 startblock, 1,
+                                                 (unsigned long *)sec_buf);
+               }
+       }
+}
+
+static int _get_new_inode_no(unsigned char *buffer)
+{
+       struct ext_filesystem *fs = get_fs();
+       unsigned char input;
+       int operand, status;
+       int count = 1;
+       int j = 0;
+
+       /* get the blocksize of the filesystem */
+       unsigned char *ptr = buffer;
+       while (*ptr == 255) {
+               ptr++;
+               count += 8;
+               if (count > ext4fs_root->sblock.inodes_per_group)
+                       return -1;
+       }
+
+       for (j = 0; j < fs->blksz; j++) {
+               input = *ptr;
+               int i = 0;
+               while (i <= 7) {
+                       operand = 1 << i;
+                       status = input & operand;
+                       if (status) {
+                               i++;
+                               count++;
+                       } else {
+                               *ptr |= operand;
+                               return count;
+                       }
+               }
+               ptr = ptr + 1;
+       }
+
+       return -1;
+}
+
+static int _get_new_blk_no(unsigned char *buffer)
+{
+       unsigned char input;
+       int operand, status;
+       int count = 0;
+       int j = 0;
+       unsigned char *ptr = buffer;
+       struct ext_filesystem *fs = get_fs();
+
+       if (fs->blksz != 1024)
+               count = 0;
+       else
+               count = 1;
+
+       while (*ptr == 255) {
+               ptr++;
+               count += 8;
+               if (count == (fs->blksz * 8))
+                       return -1;
+       }
+
+       for (j = 0; j < fs->blksz; j++) {
+               input = *ptr;
+               int i = 0;
+               while (i <= 7) {
+                       operand = 1 << i;
+                       status = input & operand;
+                       if (status) {
+                               i++;
+                               count++;
+                       } else {
+                               *ptr |= operand;
+                               return count;
+                       }
+               }
+               ptr = ptr + 1;
+       }
+
+       return -1;
+}
+
+int ext4fs_set_block_bmap(long int blockno, unsigned char *buffer, int index)
+{
+       int i, remainder, status;
+       unsigned char *ptr = buffer;
+       unsigned char operand;
+       i = blockno / 8;
+       remainder = blockno % 8;
+       int blocksize = EXT2_BLOCK_SIZE(ext4fs_root);
+
+       i = i - (index * blocksize);
+       if (blocksize != 1024) {
+               ptr = ptr + i;
+               operand = 1 << remainder;
+               status = *ptr & operand;
+               if (status)
+                       return -1;
+
+               *ptr = *ptr | operand;
+               return 0;
+       } else {
+               if (remainder == 0) {
+                       ptr = ptr + i - 1;
+                       operand = (1 << 7);
+               } else {
+                       ptr = ptr + i;
+                       operand = (1 << (remainder - 1));
+               }
+               status = *ptr & operand;
+               if (status)
+                       return -1;
+
+               *ptr = *ptr | operand;
+               return 0;
+       }
+}
+
+void ext4fs_reset_block_bmap(long int blockno, unsigned char *buffer, int index)
+{
+       int i, remainder, status;
+       unsigned char *ptr = buffer;
+       unsigned char operand;
+       i = blockno / 8;
+       remainder = blockno % 8;
+       int blocksize = EXT2_BLOCK_SIZE(ext4fs_root);
+
+       i = i - (index * blocksize);
+       if (blocksize != 1024) {
+               ptr = ptr + i;
+               operand = (1 << remainder);
+               status = *ptr & operand;
+               if (status)
+                       *ptr = *ptr & ~(operand);
+       } else {
+               if (remainder == 0) {
+                       ptr = ptr + i - 1;
+                       operand = (1 << 7);
+               } else {
+                       ptr = ptr + i;
+                       operand = (1 << (remainder - 1));
+               }
+               status = *ptr & operand;
+               if (status)
+                       *ptr = *ptr & ~(operand);
+       }
+}
+
+int ext4fs_set_inode_bmap(int inode_no, unsigned char *buffer, int index)
+{
+       int i, remainder, status;
+       unsigned char *ptr = buffer;
+       unsigned char operand;
+
+       inode_no -= (index * ext4fs_root->sblock.inodes_per_group);
+       i = inode_no / 8;
+       remainder = inode_no % 8;
+       if (remainder == 0) {
+               ptr = ptr + i - 1;
+               operand = (1 << 7);
+       } else {
+               ptr = ptr + i;
+               operand = (1 << (remainder - 1));
+       }
+       status = *ptr & operand;
+       if (status)
+               return -1;
+
+       *ptr = *ptr | operand;
+
+       return 0;
+}
+
+void ext4fs_reset_inode_bmap(int inode_no, unsigned char *buffer, int index)
+{
+       int i, remainder, status;
+       unsigned char *ptr = buffer;
+       unsigned char operand;
+
+       inode_no -= (index * ext4fs_root->sblock.inodes_per_group);
+       i = inode_no / 8;
+       remainder = inode_no % 8;
+       if (remainder == 0) {
+               ptr = ptr + i - 1;
+               operand = (1 << 7);
+       } else {
+               ptr = ptr + i;
+               operand = (1 << (remainder - 1));
+       }
+       status = *ptr & operand;
+       if (status)
+               *ptr = *ptr & ~(operand);
+}
+
+int ext4fs_checksum_update(unsigned int i)
+{
+       struct ext2_block_group *desc;
+       struct ext_filesystem *fs = get_fs();
+       __u16 crc = 0;
+
+       desc = (struct ext2_block_group *)&fs->gd[i];
+       if (fs->sb->feature_ro_compat & EXT4_FEATURE_RO_COMPAT_GDT_CSUM) {
+               int offset = offsetof(struct ext2_block_group, bg_checksum);
+
+               crc = ext2fs_crc16(~0, fs->sb->unique_id,
+                                  sizeof(fs->sb->unique_id));
+               crc = ext2fs_crc16(crc, &i, sizeof(i));
+               crc = ext2fs_crc16(crc, desc, offset);
+               offset += sizeof(desc->bg_checksum);    /* skip checksum */
+               assert(offset == sizeof(*desc));
+       }
+
+       return crc;
+}
+
+static int check_void_in_dentry(struct ext2_dirent *dir, char *filename)
+{
+       int dentry_length;
+       int sizeof_void_space;
+       int new_entry_byte_reqd;
+       short padding_factor = 0;
+
+       if (dir->namelen % 4 != 0)
+               padding_factor = 4 - (dir->namelen % 4);
+
+       dentry_length = sizeof(struct ext2_dirent) +
+                       dir->namelen + padding_factor;
+       sizeof_void_space = dir->direntlen - dentry_length;
+       if (sizeof_void_space == 0)
+               return 0;
+
+       padding_factor = 0;
+       if (strlen(filename) % 4 != 0)
+               padding_factor = 4 - (strlen(filename) % 4);
+
+       new_entry_byte_reqd = strlen(filename) +
+           sizeof(struct ext2_dirent) + padding_factor;
+       if (sizeof_void_space >= new_entry_byte_reqd) {
+               dir->direntlen = dentry_length;
+               return sizeof_void_space;
+       }
+
+       return 0;
+}
+
+void ext4fs_update_parent_dentry(char *filename, int *p_ino, int file_type)
+{
+       unsigned int *zero_buffer = NULL;
+       char *root_first_block_buffer = NULL;
+       int direct_blk_idx;
+       long int root_blknr;
+       long int first_block_no_of_root = 0;
+       long int previous_blknr = -1;
+       int totalbytes = 0;
+       short int padding_factor = 0;
+       unsigned int new_entry_byte_reqd;
+       unsigned int last_entry_dirlen;
+       int sizeof_void_space = 0;
+       int templength = 0;
+       int inodeno;
+       int status;
+       struct ext_filesystem *fs = get_fs();
+       /* directory entry */
+       struct ext2_dirent *dir;
+       char *ptr = NULL;
+       char *temp_dir = NULL;
+
+       zero_buffer = zalloc(fs->blksz);
+       if (!zero_buffer) {
+               printf("No Memory\n");
+               return;
+       }
+       root_first_block_buffer = zalloc(fs->blksz);
+       if (!root_first_block_buffer) {
+               free(zero_buffer);
+               printf("No Memory\n");
+               return;
+       }
+restart:
+
+       /* read the block no allocated to a file */
+       for (direct_blk_idx = 0; direct_blk_idx < INDIRECT_BLOCKS;
+            direct_blk_idx++) {
+               root_blknr = read_allocated_block(g_parent_inode,
+                                                 direct_blk_idx);
+               if (root_blknr == 0) {
+                       first_block_no_of_root = previous_blknr;
+                       break;
+               }
+               previous_blknr = root_blknr;
+       }
+
+       status = ext4fs_devread(first_block_no_of_root
+                               * fs->sect_perblk,
+                               0, fs->blksz, root_first_block_buffer);
+       if (status == 0)
+               goto fail;
+
+       if (ext4fs_log_journal(root_first_block_buffer, first_block_no_of_root))
+               goto fail;
+       dir = (struct ext2_dirent *)root_first_block_buffer;
+       ptr = (char *)dir;
+       totalbytes = 0;
+       while (dir->direntlen > 0) {
+               /*
+                * blocksize-totalbytes because last directory length
+                * i.e. dir->direntlen is free availble space in the
+                * block that means  it is a last entry of directory
+                * entry
+                */
+
+               /* traversing the each directory entry */
+               if (fs->blksz - totalbytes == dir->direntlen) {
+                       if (strlen(filename) % 4 != 0)
+                               padding_factor = 4 - (strlen(filename) % 4);
+
+                       new_entry_byte_reqd = strlen(filename) +
+                           sizeof(struct ext2_dirent) + padding_factor;
+                       padding_factor = 0;
+                       /*
+                        * update last directory entry length to its
+                        * length because we are creating new directory
+                        * entry
+                        */
+                       if (dir->namelen % 4 != 0)
+                               padding_factor = 4 - (dir->namelen % 4);
+
+                       last_entry_dirlen = dir->namelen +
+                           sizeof(struct ext2_dirent) + padding_factor;
+                       if ((fs->blksz - totalbytes - last_entry_dirlen) <
+                               new_entry_byte_reqd) {
+                               printf("1st Block Full:Allocate new block\n");
+
+                               if (direct_blk_idx == INDIRECT_BLOCKS - 1) {
+                                       printf("Directory exceeds limit\n");
+                                       goto fail;
+                               }
+                               g_parent_inode->b.blocks.dir_blocks
+                                   [direct_blk_idx] = ext4fs_get_new_blk_no();
+                               if (g_parent_inode->b.blocks.dir_blocks
+                                       [direct_blk_idx] == -1) {
+                                       printf("no block left to assign\n");
+                                       goto fail;
+                               }
+                               put_ext4(((uint64_t)
+                                         (g_parent_inode->b.
+                                          blocks.dir_blocks[direct_blk_idx] *
+                                          fs->blksz)), zero_buffer, fs->blksz);
+                               g_parent_inode->size =
+                                   g_parent_inode->size + fs->blksz;
+                               g_parent_inode->blockcnt =
+                                   g_parent_inode->blockcnt + fs->sect_perblk;
+                               if (ext4fs_put_metadata
+                                   (root_first_block_buffer,
+                                    first_block_no_of_root))
+                                       goto fail;
+                               goto restart;
+                       }
+                       dir->direntlen = last_entry_dirlen;
+                       break;
+               }
+
+               templength = dir->direntlen;
+               totalbytes = totalbytes + templength;
+               sizeof_void_space = check_void_in_dentry(dir, filename);
+               if (sizeof_void_space)
+                       break;
+
+               dir = (struct ext2_dirent *)((char *)dir + templength);
+               ptr = (char *)dir;
+       }
+
+       /* make a pointer ready for creating next directory entry */
+       templength = dir->direntlen;
+       totalbytes = totalbytes + templength;
+       dir = (struct ext2_dirent *)((char *)dir + templength);
+       ptr = (char *)dir;
+
+       /* get the next available inode number */
+       inodeno = ext4fs_get_new_inode_no();
+       if (inodeno == -1) {
+               printf("no inode left to assign\n");
+               goto fail;
+       }
+       dir->inode = inodeno;
+       if (sizeof_void_space)
+               dir->direntlen = sizeof_void_space;
+       else
+               dir->direntlen = fs->blksz - totalbytes;
+
+       dir->namelen = strlen(filename);
+       dir->filetype = FILETYPE_REG;   /* regular file */
+       temp_dir = (char *)dir;
+       temp_dir = temp_dir + sizeof(struct ext2_dirent);
+       memcpy(temp_dir, filename, strlen(filename));
+
+       *p_ino = inodeno;
+
+       /* update or write  the 1st block of root inode */
+       if (ext4fs_put_metadata(root_first_block_buffer,
+                               first_block_no_of_root))
+               goto fail;
+
+fail:
+       free(zero_buffer);
+       free(root_first_block_buffer);
+}
+
+static int search_dir(struct ext2_inode *parent_inode, char *dirname)
+{
+       int status;
+       int inodeno;
+       int totalbytes;
+       int templength;
+       int direct_blk_idx;
+       long int blknr;
+       int found = 0;
+       char *ptr = NULL;
+       unsigned char *block_buffer = NULL;
+       struct ext2_dirent *dir = NULL;
+       struct ext2_dirent *previous_dir = NULL;
+       struct ext_filesystem *fs = get_fs();
+
+       /* read the block no allocated to a file */
+       for (direct_blk_idx = 0; direct_blk_idx < INDIRECT_BLOCKS;
+               direct_blk_idx++) {
+               blknr = read_allocated_block(parent_inode, direct_blk_idx);
+               if (blknr == 0)
+                       goto fail;
+
+               /* read the blocks of parenet inode */
+               block_buffer = zalloc(fs->blksz);
+               if (!block_buffer)
+                       goto fail;
+
+               status = ext4fs_devread(blknr * fs->sect_perblk,
+                                       0, fs->blksz, (char *)block_buffer);
+               if (status == 0)
+                       goto fail;
+
+               dir = (struct ext2_dirent *)block_buffer;
+               ptr = (char *)dir;
+               totalbytes = 0;
+               while (dir->direntlen >= 0) {
+                       /*
+                        * blocksize-totalbytes because last directory
+                        * length i.e.,*dir->direntlen is free availble
+                        * space in the block that means
+                        * it is a last entry of directory entry
+                        */
+                       if (strlen(dirname) == dir->namelen) {
+                               if (strncmp(dirname, ptr +
+                                       sizeof(struct ext2_dirent),
+                                       dir->namelen) == 0) {
+                                       previous_dir->direntlen +=
+                                                       dir->direntlen;
+                                       inodeno = dir->inode;
+                                       dir->inode = 0;
+                                       found = 1;
+                                       break;
+                               }
+                       }
+
+                       if (fs->blksz - totalbytes == dir->direntlen)
+                               break;
+
+                       /* traversing the each directory entry */
+                       templength = dir->direntlen;
+                       totalbytes = totalbytes + templength;
+                       previous_dir = dir;
+                       dir = (struct ext2_dirent *)((char *)dir + templength);
+                       ptr = (char *)dir;
+               }
+
+               if (found == 1) {
+                       free(block_buffer);
+                       block_buffer = NULL;
+                       return inodeno;
+               }
+
+               free(block_buffer);
+               block_buffer = NULL;
+       }
+
+fail:
+       free(block_buffer);
+
+       return -1;
+}
+
+static int find_dir_depth(char *dirname)
+{
+       char *token = strtok(dirname, "/");
+       int count = 0;
+       while (token != NULL) {
+               token = strtok(NULL, "/");
+               count++;
+       }
+       return count + 1 + 1;
+       /*
+        * for example  for string /home/temp
+        * depth=home(1)+temp(1)+1 extra for NULL;
+        * so count is 4;
+        */
+}
+
+static int parse_path(char **arr, char *dirname)
+{
+       char *token = strtok(dirname, "/");
+       int i = 0;
+
+       /* add root */
+       arr[i] = zalloc(strlen("/") + 1);
+       if (!arr[i])
+               return -ENOMEM;
+
+       arr[i++] = "/";
+
+       /* add each path entry after root */
+       while (token != NULL) {
+               arr[i] = zalloc(strlen(token) + 1);
+               if (!arr[i])
+                       return -ENOMEM;
+               memcpy(arr[i++], token, strlen(token));
+               token = strtok(NULL, "/");
+       }
+       arr[i] = NULL;
+
+       return 0;
+}
+
+int ext4fs_iget(int inode_no, struct ext2_inode *inode)
+{
+       if (ext4fs_read_inode(ext4fs_root, inode_no, inode) == 0)
+               return -1;
+
+       return 0;
+}
+
+/*
+ * Function: ext4fs_get_parent_inode_num
+ * Return Value: inode Number of the parent directory of  file/Directory to be
+ * created
+ * dirname : Input parmater, input path name of the file/directory to be created
+ * dname : Output parameter, to be filled with the name of the directory
+ * extracted from dirname
+ */
+int ext4fs_get_parent_inode_num(const char *dirname, char *dname, int flags)
+{
+       int i;
+       int depth = 0;
+       int matched_inode_no;
+       int result_inode_no = -1;
+       char **ptr = NULL;
+       char *depth_dirname = NULL;
+       char *parse_dirname = NULL;
+       struct ext2_inode *parent_inode = NULL;
+       struct ext2_inode *first_inode = NULL;
+       struct ext2_inode temp_inode;
+
+       if (*dirname != '/') {
+               printf("Please supply Absolute path\n");
+               return -1;
+       }
+
+       /* TODO: input validation make equivalent to linux */
+       depth_dirname = zalloc(strlen(dirname) + 1);
+       if (!depth_dirname)
+               return -ENOMEM;
+
+       memcpy(depth_dirname, dirname, strlen(dirname));
+       depth = find_dir_depth(depth_dirname);
+       parse_dirname = zalloc(strlen(dirname) + 1);
+       if (!parse_dirname)
+               goto fail;
+       memcpy(parse_dirname, dirname, strlen(dirname));
+
+       /* allocate memory for each directory level */
+       ptr = zalloc((depth) * sizeof(char *));
+       if (!ptr)
+               goto fail;
+       if (parse_path(ptr, parse_dirname))
+               goto fail;
+       parent_inode = zalloc(sizeof(struct ext2_inode));
+       if (!parent_inode)
+               goto fail;
+       first_inode = zalloc(sizeof(struct ext2_inode));
+       if (!first_inode)
+               goto fail;
+       memcpy(parent_inode, ext4fs_root->inode, sizeof(struct ext2_inode));
+       memcpy(first_inode, parent_inode, sizeof(struct ext2_inode));
+       if (flags & F_FILE)
+               result_inode_no = EXT2_ROOT_INO;
+       for (i = 1; i < depth; i++) {
+               matched_inode_no = search_dir(parent_inode, ptr[i]);
+               if (matched_inode_no == -1) {
+                       if (ptr[i + 1] == NULL && i == 1) {
+                               result_inode_no = EXT2_ROOT_INO;
+                               goto end;
+                       } else {
+                               if (ptr[i + 1] == NULL)
+                                       break;
+                               printf("Invalid path\n");
+                               result_inode_no = -1;
+                               goto fail;
+                       }
+               } else {
+                       if (ptr[i + 1] != NULL) {
+                               memset(parent_inode, '\0',
+                                      sizeof(struct ext2_inode));
+                               if (ext4fs_iget(matched_inode_no,
+                                               parent_inode)) {
+                                       result_inode_no = -1;
+                                       goto fail;
+                               }
+                               result_inode_no = matched_inode_no;
+                       } else {
+                               break;
+                       }
+               }
+       }
+
+end:
+       if (i == 1)
+               matched_inode_no = search_dir(first_inode, ptr[i]);
+       else
+               matched_inode_no = search_dir(parent_inode, ptr[i]);
+
+       if (matched_inode_no != -1) {
+               ext4fs_iget(matched_inode_no, &temp_inode);
+               if (temp_inode.mode & S_IFDIR) {
+                       printf("It is a Directory\n");
+                       result_inode_no = -1;
+                       goto fail;
+               }
+       }
+
+       if (strlen(ptr[i]) > 256) {
+               result_inode_no = -1;
+               goto fail;
+       }
+       memcpy(dname, ptr[i], strlen(ptr[i]));
+
+fail:
+       free(depth_dirname);
+       free(parse_dirname);
+       free(ptr);
+       free(parent_inode);
+       free(first_inode);
+
+       return result_inode_no;
+}
+
+static int check_filename(char *filename, unsigned int blknr)
+{
+       unsigned int first_block_no_of_root;
+       int totalbytes = 0;
+       int templength = 0;
+       int status, inodeno;
+       int found = 0;
+       char *root_first_block_buffer = NULL;
+       char *root_first_block_addr = NULL;
+       struct ext2_dirent *dir = NULL;
+       struct ext2_dirent *previous_dir = NULL;
+       char *ptr = NULL;
+       struct ext_filesystem *fs = get_fs();
+
+       /* get the first block of root */
+       first_block_no_of_root = blknr;
+       root_first_block_buffer = zalloc(fs->blksz);
+       if (!root_first_block_buffer)
+               return -ENOMEM;
+       root_first_block_addr = root_first_block_buffer;
+       status = ext4fs_devread(first_block_no_of_root *
+                               fs->sect_perblk, 0,
+                               fs->blksz, root_first_block_buffer);
+       if (status == 0)
+               goto fail;
+
+       if (ext4fs_log_journal(root_first_block_buffer, first_block_no_of_root))
+               goto fail;
+       dir = (struct ext2_dirent *)root_first_block_buffer;
+       ptr = (char *)dir;
+       totalbytes = 0;
+       while (dir->direntlen >= 0) {
+               /*
+                * blocksize-totalbytes because last
+                * directory length i.e., *dir->direntlen
+                * is free availble space in the block that
+                * means it is a last entry of directory entry
+                */
+               if (strlen(filename) == dir->namelen) {
+                       if (strncmp(filename, ptr + sizeof(struct ext2_dirent),
+                               dir->namelen) == 0) {
+                               printf("file found deleting\n");
+                               previous_dir->direntlen += dir->direntlen;
+                               inodeno = dir->inode;
+                               dir->inode = 0;
+                               found = 1;
+                               break;
+                       }
+               }
+
+               if (fs->blksz - totalbytes == dir->direntlen)
+                       break;
+
+               /* traversing the each directory entry */
+               templength = dir->direntlen;
+               totalbytes = totalbytes + templength;
+               previous_dir = dir;
+               dir = (struct ext2_dirent *)((char *)dir + templength);
+               ptr = (char *)dir;
+       }
+
+
+       if (found == 1) {
+               if (ext4fs_put_metadata(root_first_block_addr,
+                                       first_block_no_of_root))
+                       goto fail;
+               return inodeno;
+       }
+fail:
+       free(root_first_block_buffer);
+
+       return -1;
+}
+
+int ext4fs_filename_check(char *filename)
+{
+       short direct_blk_idx = 0;
+       long int blknr = -1;
+       int inodeno = -1;
+
+       /* read the block no allocated to a file */
+       for (direct_blk_idx = 0; direct_blk_idx < INDIRECT_BLOCKS;
+               direct_blk_idx++) {
+               blknr = read_allocated_block(g_parent_inode, direct_blk_idx);
+               if (blknr == 0)
+                       break;
+               inodeno = check_filename(filename, blknr);
+               if (inodeno != -1)
+                       return inodeno;
+       }
+
+       return -1;
+}
+
+long int ext4fs_get_new_blk_no(void)
+{
+       short i;
+       short status;
+       int remainder;
+       unsigned int bg_idx;
+       static int prev_bg_bitmap_index = -1;
+       unsigned int blk_per_grp = ext4fs_root->sblock.blocks_per_group;
+       struct ext_filesystem *fs = get_fs();
+       char *journal_buffer = zalloc(fs->blksz);
+       char *zero_buffer = zalloc(fs->blksz);
+       if (!journal_buffer || !zero_buffer)
+               goto fail;
+       struct ext2_block_group *gd = (struct ext2_block_group *)fs->gdtable;
+
+       if (fs->first_pass_bbmap == 0) {
+               for (i = 0; i < fs->no_blkgrp; i++) {
+                       if (gd[i].free_blocks) {
+                               if (gd[i].bg_flags & EXT4_BG_BLOCK_UNINIT) {
+                                       put_ext4(((uint64_t) (gd[i].block_id *
+                                                             fs->blksz)),
+                                                zero_buffer, fs->blksz);
+                                       gd[i].bg_flags =
+                                           gd[i].
+                                           bg_flags & ~EXT4_BG_BLOCK_UNINIT;
+                                       memcpy(fs->blk_bmaps[i], zero_buffer,
+                                              fs->blksz);
+                               }
+                               fs->curr_blkno =
+                                   _get_new_blk_no(fs->blk_bmaps[i]);
+                               if (fs->curr_blkno == -1)
+                                       /* if block bitmap is completely fill */
+                                       continue;
+                               fs->curr_blkno = fs->curr_blkno +
+                                               (i * fs->blksz * 8);
+                               fs->first_pass_bbmap++;
+                               gd[i].free_blocks--;
+                               fs->sb->free_blocks--;
+                               status = ext4fs_devread(gd[i].block_id *
+                                                       fs->sect_perblk, 0,
+                                                       fs->blksz,
+                                                       journal_buffer);
+                               if (status == 0)
+                                       goto fail;
+                               if (ext4fs_log_journal(journal_buffer,
+                                                       gd[i].block_id))
+                                       goto fail;
+                               goto success;
+                       } else {
+                               debug("no space left on block group %d\n", i);
+                       }
+               }
+
+               goto fail;
+       } else {
+restart:
+               fs->curr_blkno++;
+               /* get the blockbitmap index respective to blockno */
+               if (fs->blksz != 1024) {
+                       bg_idx = fs->curr_blkno / blk_per_grp;
+               } else {
+                       bg_idx = fs->curr_blkno / blk_per_grp;
+                       remainder = fs->curr_blkno % blk_per_grp;
+                       if (!remainder)
+                               bg_idx--;
+               }
+
+               /*
+                * To skip completely filled block group bitmaps
+                * Optimize the block allocation
+                */
+               if (bg_idx >= fs->no_blkgrp)
+                       goto fail;
+
+               if (gd[bg_idx].free_blocks == 0) {
+                       debug("block group %u is full. Skipping\n", bg_idx);
+                       fs->curr_blkno = fs->curr_blkno + blk_per_grp;
+                       fs->curr_blkno--;
+                       goto restart;
+               }
+
+               if (gd[bg_idx].bg_flags & EXT4_BG_BLOCK_UNINIT) {
+                       memset(zero_buffer, '\0', fs->blksz);
+                       put_ext4(((uint64_t) (gd[bg_idx].block_id * fs->blksz)),
+                                zero_buffer, fs->blksz);
+                       memcpy(fs->blk_bmaps[bg_idx], zero_buffer, fs->blksz);
+                       gd[bg_idx].bg_flags = gd[bg_idx].bg_flags &
+                                               ~EXT4_BG_BLOCK_UNINIT;
+               }
+
+               if (ext4fs_set_block_bmap(fs->curr_blkno, fs->blk_bmaps[bg_idx],
+                                  bg_idx) != 0) {
+                       debug("going for restart for the block no %ld %u\n",
+                             fs->curr_blkno, bg_idx);
+                       goto restart;
+               }
+
+               /* journal backup */
+               if (prev_bg_bitmap_index != bg_idx) {
+                       memset(journal_buffer, '\0', fs->blksz);
+                       status = ext4fs_devread(gd[bg_idx].block_id
+                                               * fs->sect_perblk,
+                                               0, fs->blksz, journal_buffer);
+                       if (status == 0)
+                               goto fail;
+                       if (ext4fs_log_journal(journal_buffer,
+                                               gd[bg_idx].block_id))
+                               goto fail;
+
+                       prev_bg_bitmap_index = bg_idx;
+               }
+               gd[bg_idx].free_blocks--;
+               fs->sb->free_blocks--;
+               goto success;
+       }
+success:
+       free(journal_buffer);
+       free(zero_buffer);
+
+       return fs->curr_blkno;
+fail:
+       free(journal_buffer);
+       free(zero_buffer);
+
+       return -1;
+}
+
+int ext4fs_get_new_inode_no(void)
+{
+       short i;
+       short status;
+       unsigned int ibmap_idx;
+       static int prev_inode_bitmap_index = -1;
+       unsigned int inodes_per_grp = ext4fs_root->sblock.inodes_per_group;
+       struct ext_filesystem *fs = get_fs();
+       char *journal_buffer = zalloc(fs->blksz);
+       char *zero_buffer = zalloc(fs->blksz);
+       if (!journal_buffer || !zero_buffer)
+               goto fail;
+       struct ext2_block_group *gd = (struct ext2_block_group *)fs->gdtable;
+
+       if (fs->first_pass_ibmap == 0) {
+               for (i = 0; i < fs->no_blkgrp; i++) {
+                       if (gd[i].free_inodes) {
+                               if (gd[i].bg_itable_unused != gd[i].free_inodes)
+                                       gd[i].bg_itable_unused =
+                                               gd[i].free_inodes;
+                               if (gd[i].bg_flags & EXT4_BG_INODE_UNINIT) {
+                                       put_ext4(((uint64_t)
+                                                 (gd[i].inode_id * fs->blksz)),
+                                                zero_buffer, fs->blksz);
+                                       gd[i].bg_flags = gd[i].bg_flags &
+                                                       ~EXT4_BG_INODE_UNINIT;
+                                       memcpy(fs->inode_bmaps[i],
+                                              zero_buffer, fs->blksz);
+                               }
+                               fs->curr_inode_no =
+                                   _get_new_inode_no(fs->inode_bmaps[i]);
+                               if (fs->curr_inode_no == -1)
+                                       /* if block bitmap is completely fill */
+                                       continue;
+                               fs->curr_inode_no = fs->curr_inode_no +
+                                                       (i * inodes_per_grp);
+                               fs->first_pass_ibmap++;
+                               gd[i].free_inodes--;
+                               gd[i].bg_itable_unused--;
+                               fs->sb->free_inodes--;
+                               status = ext4fs_devread(gd[i].inode_id *
+                                                       fs->sect_perblk, 0,
+                                                       fs->blksz,
+                                                       journal_buffer);
+                               if (status == 0)
+                                       goto fail;
+                               if (ext4fs_log_journal(journal_buffer,
+                                                       gd[i].inode_id))
+                                       goto fail;
+                               goto success;
+                       } else
+                               debug("no inode left on block group %d\n", i);
+               }
+               goto fail;
+       } else {
+restart:
+               fs->curr_inode_no++;
+               /* get the blockbitmap index respective to blockno */
+               ibmap_idx = fs->curr_inode_no / inodes_per_grp;
+               if (gd[ibmap_idx].bg_flags & EXT4_BG_INODE_UNINIT) {
+                       memset(zero_buffer, '\0', fs->blksz);
+                       put_ext4(((uint64_t) (gd[ibmap_idx].inode_id *
+                                             fs->blksz)), zero_buffer,
+                                fs->blksz);
+                       gd[ibmap_idx].bg_flags =
+                           gd[ibmap_idx].bg_flags & ~EXT4_BG_INODE_UNINIT;
+                       memcpy(fs->inode_bmaps[ibmap_idx], zero_buffer,
+                               fs->blksz);
+               }
+
+               if (ext4fs_set_inode_bmap(fs->curr_inode_no,
+                                         fs->inode_bmaps[ibmap_idx],
+                                         ibmap_idx) != 0) {
+                       debug("going for restart for the block no %d %u\n",
+                             fs->curr_inode_no, ibmap_idx);
+                       goto restart;
+               }
+
+               /* journal backup */
+               if (prev_inode_bitmap_index != ibmap_idx) {
+                       memset(journal_buffer, '\0', fs->blksz);
+                       status = ext4fs_devread(gd[ibmap_idx].inode_id
+                                               * fs->sect_perblk,
+                                               0, fs->blksz, journal_buffer);
+                       if (status == 0)
+                               goto fail;
+                       if (ext4fs_log_journal(journal_buffer,
+                                               gd[ibmap_idx].inode_id))
+                               goto fail;
+                       prev_inode_bitmap_index = ibmap_idx;
+               }
+               if (gd[ibmap_idx].bg_itable_unused != gd[ibmap_idx].free_inodes)
+                       gd[ibmap_idx].bg_itable_unused =
+                                       gd[ibmap_idx].free_inodes;
+               gd[ibmap_idx].free_inodes--;
+               gd[ibmap_idx].bg_itable_unused--;
+               fs->sb->free_inodes--;
+               goto success;
+       }
+
+success:
+       free(journal_buffer);
+       free(zero_buffer);
+
+       return fs->curr_inode_no;
+fail:
+       free(journal_buffer);
+       free(zero_buffer);
+
+       return -1;
+
+}
+
+
+static void alloc_single_indirect_block(struct ext2_inode *file_inode,
+                                       unsigned int *total_remaining_blocks,
+                                       unsigned int *no_blks_reqd)
+{
+       short i;
+       short status;
+       long int actual_block_no;
+       long int si_blockno;
+       /* si :single indirect */
+       unsigned int *si_buffer = NULL;
+       unsigned int *si_start_addr = NULL;
+       struct ext_filesystem *fs = get_fs();
+
+       if (*total_remaining_blocks != 0) {
+               si_buffer = zalloc(fs->blksz);
+               if (!si_buffer) {
+                       printf("No Memory\n");
+                       return;
+               }
+               si_start_addr = si_buffer;
+               si_blockno = ext4fs_get_new_blk_no();
+               if (si_blockno == -1) {
+                       printf("no block left to assign\n");
+                       goto fail;
+               }
+               (*no_blks_reqd)++;
+               debug("SIPB %ld: %u\n", si_blockno, *total_remaining_blocks);
+
+               status = ext4fs_devread(si_blockno * fs->sect_perblk,
+                                       0, fs->blksz, (char *)si_buffer);
+               memset(si_buffer, '\0', fs->blksz);
+               if (status == 0)
+                       goto fail;
+
+               for (i = 0; i < (fs->blksz / sizeof(int)); i++) {
+                       actual_block_no = ext4fs_get_new_blk_no();
+                       if (actual_block_no == -1) {
+                               printf("no block left to assign\n");
+                               goto fail;
+                       }
+                       *si_buffer = actual_block_no;
+                       debug("SIAB %u: %u\n", *si_buffer,
+                               *total_remaining_blocks);
+
+                       si_buffer++;
+                       (*total_remaining_blocks)--;
+                       if (*total_remaining_blocks == 0)
+                               break;
+               }
+
+               /* write the block to disk */
+               put_ext4(((uint64_t) (si_blockno * fs->blksz)),
+                        si_start_addr, fs->blksz);
+               file_inode->b.blocks.indir_block = si_blockno;
+       }
+fail:
+       free(si_start_addr);
+}
+
+static void alloc_double_indirect_block(struct ext2_inode *file_inode,
+                                       unsigned int *total_remaining_blocks,
+                                       unsigned int *no_blks_reqd)
+{
+       short i;
+       short j;
+       short status;
+       long int actual_block_no;
+       /* di:double indirect */
+       long int di_blockno_parent;
+       long int di_blockno_child;
+       unsigned int *di_parent_buffer = NULL;
+       unsigned int *di_child_buff = NULL;
+       unsigned int *di_block_start_addr = NULL;
+       unsigned int *di_child_buff_start = NULL;
+       struct ext_filesystem *fs = get_fs();
+
+       if (*total_remaining_blocks != 0) {
+               /* double indirect parent block connecting to inode */
+               di_blockno_parent = ext4fs_get_new_blk_no();
+               if (di_blockno_parent == -1) {
+                       printf("no block left to assign\n");
+                       goto fail;
+               }
+               di_parent_buffer = zalloc(fs->blksz);
+               if (!di_parent_buffer)
+                       goto fail;
+
+               di_block_start_addr = di_parent_buffer;
+               (*no_blks_reqd)++;
+               debug("DIPB %ld: %u\n", di_blockno_parent,
+                     *total_remaining_blocks);
+
+               status = ext4fs_devread(di_blockno_parent *
+                                       fs->sect_perblk, 0,
+                                       fs->blksz, (char *)di_parent_buffer);
+               memset(di_parent_buffer, '\0', fs->blksz);
+
+               /*
+                * start:for each double indirect parent
+                * block create one more block
+                */
+               for (i = 0; i < (fs->blksz / sizeof(int)); i++) {
+                       di_blockno_child = ext4fs_get_new_blk_no();
+                       if (di_blockno_child == -1) {
+                               printf("no block left to assign\n");
+                               goto fail;
+                       }
+                       di_child_buff = zalloc(fs->blksz);
+                       if (!di_child_buff)
+                               goto fail;
+
+                       di_child_buff_start = di_child_buff;
+                       *di_parent_buffer = di_blockno_child;
+                       di_parent_buffer++;
+                       (*no_blks_reqd)++;
+                       debug("DICB %ld: %u\n", di_blockno_child,
+                             *total_remaining_blocks);
+
+                       status = ext4fs_devread(di_blockno_child *
+                                               fs->sect_perblk, 0,
+                                               fs->blksz,
+                                               (char *)di_child_buff);
+                       memset(di_child_buff, '\0', fs->blksz);
+                       /* filling of actual datablocks for each child */
+                       for (j = 0; j < (fs->blksz / sizeof(int)); j++) {
+                               actual_block_no = ext4fs_get_new_blk_no();
+                               if (actual_block_no == -1) {
+                                       printf("no block left to assign\n");
+                                       goto fail;
+                               }
+                               *di_child_buff = actual_block_no;
+                               debug("DIAB %ld: %u\n", actual_block_no,
+                                     *total_remaining_blocks);
+
+                               di_child_buff++;
+                               (*total_remaining_blocks)--;
+                               if (*total_remaining_blocks == 0)
+                                       break;
+                       }
+                       /* write the block  table */
+                       put_ext4(((uint64_t) (di_blockno_child * fs->blksz)),
+                                di_child_buff_start, fs->blksz);
+                       free(di_child_buff_start);
+                       di_child_buff_start = NULL;
+
+                       if (*total_remaining_blocks == 0)
+                               break;
+               }
+               put_ext4(((uint64_t) (di_blockno_parent * fs->blksz)),
+                        di_block_start_addr, fs->blksz);
+               file_inode->b.blocks.double_indir_block = di_blockno_parent;
+       }
+fail:
+       free(di_block_start_addr);
+}
+
+static void alloc_triple_indirect_block(struct ext2_inode *file_inode,
+                                       unsigned int *total_remaining_blocks,
+                                       unsigned int *no_blks_reqd)
+{
+       short i;
+       short j;
+       short k;
+       long int actual_block_no;
+       /* ti: Triple Indirect */
+       long int ti_gp_blockno;
+       long int ti_parent_blockno;
+       long int ti_child_blockno;
+       unsigned int *ti_gp_buff = NULL;
+       unsigned int *ti_parent_buff = NULL;
+       unsigned int *ti_child_buff = NULL;
+       unsigned int *ti_gp_buff_start_addr = NULL;
+       unsigned int *ti_pbuff_start_addr = NULL;
+       unsigned int *ti_cbuff_start_addr = NULL;
+       struct ext_filesystem *fs = get_fs();
+       if (*total_remaining_blocks != 0) {
+               /* triple indirect grand parent block connecting to inode */
+               ti_gp_blockno = ext4fs_get_new_blk_no();
+               if (ti_gp_blockno == -1) {
+                       printf("no block left to assign\n");
+                       goto fail;
+               }
+               ti_gp_buff = zalloc(fs->blksz);
+               if (!ti_gp_buff)
+                       goto fail;
+
+               ti_gp_buff_start_addr = ti_gp_buff;
+               (*no_blks_reqd)++;
+               debug("TIGPB %ld: %u\n", ti_gp_blockno,
+                     *total_remaining_blocks);
+
+               /* for each 4 byte grand parent entry create one more block */
+               for (i = 0; i < (fs->blksz / sizeof(int)); i++) {
+                       ti_parent_blockno = ext4fs_get_new_blk_no();
+                       if (ti_parent_blockno == -1) {
+                               printf("no block left to assign\n");
+                               goto fail;
+                       }
+                       ti_parent_buff = zalloc(fs->blksz);
+                       if (!ti_parent_buff)
+                               goto fail;
+
+                       ti_pbuff_start_addr = ti_parent_buff;
+                       *ti_gp_buff = ti_parent_blockno;
+                       ti_gp_buff++;
+                       (*no_blks_reqd)++;
+                       debug("TIPB %ld: %u\n", ti_parent_blockno,
+                             *total_remaining_blocks);
+
+                       /* for each 4 byte entry parent create one more block */
+                       for (j = 0; j < (fs->blksz / sizeof(int)); j++) {
+                               ti_child_blockno = ext4fs_get_new_blk_no();
+                               if (ti_child_blockno == -1) {
+                                       printf("no block left assign\n");
+                                       goto fail;
+                               }
+                               ti_child_buff = zalloc(fs->blksz);
+                               if (!ti_child_buff)
+                                       goto fail;
+
+                               ti_cbuff_start_addr = ti_child_buff;
+                               *ti_parent_buff = ti_child_blockno;
+                               ti_parent_buff++;
+                               (*no_blks_reqd)++;
+                               debug("TICB %ld: %u\n", ti_parent_blockno,
+                                     *total_remaining_blocks);
+
+                               /* fill actual datablocks for each child */
+                               for (k = 0; k < (fs->blksz / sizeof(int));
+                                       k++) {
+                                       actual_block_no =
+                                           ext4fs_get_new_blk_no();
+                                       if (actual_block_no == -1) {
+                                               printf("no block left\n");
+                                               goto fail;
+                                       }
+                                       *ti_child_buff = actual_block_no;
+                                       debug("TIAB %ld: %u\n", actual_block_no,
+                                             *total_remaining_blocks);
+
+                                       ti_child_buff++;
+                                       (*total_remaining_blocks)--;
+                                       if (*total_remaining_blocks == 0)
+                                               break;
+                               }
+                               /* write the child block */
+                               put_ext4(((uint64_t) (ti_child_blockno *
+                                                     fs->blksz)),
+                                        ti_cbuff_start_addr, fs->blksz);
+                               free(ti_cbuff_start_addr);
+
+                               if (*total_remaining_blocks == 0)
+                                       break;
+                       }
+                       /* write the parent block */
+                       put_ext4(((uint64_t) (ti_parent_blockno * fs->blksz)),
+                                ti_pbuff_start_addr, fs->blksz);
+                       free(ti_pbuff_start_addr);
+
+                       if (*total_remaining_blocks == 0)
+                               break;
+               }
+               /* write the grand parent block */
+               put_ext4(((uint64_t) (ti_gp_blockno * fs->blksz)),
+                        ti_gp_buff_start_addr, fs->blksz);
+               file_inode->b.blocks.triple_indir_block = ti_gp_blockno;
+       }
+fail:
+       free(ti_gp_buff_start_addr);
+}
+
+void ext4fs_allocate_blocks(struct ext2_inode *file_inode,
+                               unsigned int total_remaining_blocks,
+                               unsigned int *total_no_of_block)
+{
+       short i;
+       long int direct_blockno;
+       unsigned int no_blks_reqd = 0;
+
+       /* allocation of direct blocks */
+       for (i = 0; i < INDIRECT_BLOCKS; i++) {
+               direct_blockno = ext4fs_get_new_blk_no();
+               if (direct_blockno == -1) {
+                       printf("no block left to assign\n");
+                       return;
+               }
+               file_inode->b.blocks.dir_blocks[i] = direct_blockno;
+               debug("DB %ld: %u\n", direct_blockno, total_remaining_blocks);
+
+               total_remaining_blocks--;
+               if (total_remaining_blocks == 0)
+                       break;
+       }
+
+       alloc_single_indirect_block(file_inode, &total_remaining_blocks,
+                                   &no_blks_reqd);
+       alloc_double_indirect_block(file_inode, &total_remaining_blocks,
+                                   &no_blks_reqd);
+       alloc_triple_indirect_block(file_inode, &total_remaining_blocks,
+                                   &no_blks_reqd);
+       *total_no_of_block += no_blks_reqd;
+}
+
+#endif
+
+static struct ext4_extent_header *ext4fs_get_extent_block
+       (struct ext2_data *data, char *buf,
+               struct ext4_extent_header *ext_block,
+               uint32_t fileblock, int log2_blksz)
+{
+       struct ext4_extent_idx *index;
+       unsigned long long block;
+       struct ext_filesystem *fs = get_fs();
+       int i;
+
+       while (1) {
+               index = (struct ext4_extent_idx *)(ext_block + 1);
+
+               if (le32_to_cpu(ext_block->eh_magic) != EXT4_EXT_MAGIC)
+                       return 0;
+
+               if (ext_block->eh_depth == 0)
+                       return ext_block;
+               i = -1;
+               do {
+                       i++;
+                       if (i >= le32_to_cpu(ext_block->eh_entries))
+                               break;
+               } while (fileblock > le32_to_cpu(index[i].ei_block));
+
+               if (--i < 0)
+                       return 0;
+
+               block = le32_to_cpu(index[i].ei_leaf_hi);
+               block = (block << 32) + le32_to_cpu(index[i].ei_leaf_lo);
+
+               if (ext4fs_devread(block << log2_blksz, 0, fs->blksz, buf))
+                       ext_block = (struct ext4_extent_header *)buf;
+               else
+                       return 0;
+       }
+}
+
+static int ext4fs_blockgroup
+       (struct ext2_data *data, int group, struct ext2_block_group *blkgrp)
+{
+       long int blkno;
+       unsigned int blkoff, desc_per_blk;
+
+       desc_per_blk = EXT2_BLOCK_SIZE(data) / sizeof(struct ext2_block_group);
+
+       blkno = __le32_to_cpu(data->sblock.first_data_block) + 1 +
+                       group / desc_per_blk;
+       blkoff = (group % desc_per_blk) * sizeof(struct ext2_block_group);
+
+       debug("ext4fs read %d group descriptor (blkno %ld blkoff %u)\n",
+             group, blkno, blkoff);
+
+       return ext4fs_devread(blkno << LOG2_EXT2_BLOCK_SIZE(data),
+                             blkoff, sizeof(struct ext2_block_group),
+                             (char *)blkgrp);
+}
+
+int ext4fs_read_inode(struct ext2_data *data, int ino, struct ext2_inode *inode)
+{
+       struct ext2_block_group blkgrp;
+       struct ext2_sblock *sblock = &data->sblock;
+       struct ext_filesystem *fs = get_fs();
+       int inodes_per_block, status;
+       long int blkno;
+       unsigned int blkoff;
+
+       /* It is easier to calculate if the first inode is 0. */
+       ino--;
+       status = ext4fs_blockgroup(data, ino / __le32_to_cpu
+                                  (sblock->inodes_per_group), &blkgrp);
+       if (status == 0)
+               return 0;
+
+       inodes_per_block = EXT2_BLOCK_SIZE(data) / fs->inodesz;
+       blkno = __le32_to_cpu(blkgrp.inode_table_id) +
+           (ino % __le32_to_cpu(sblock->inodes_per_group)) / inodes_per_block;
+       blkoff = (ino % inodes_per_block) * fs->inodesz;
+       /* Read the inode. */
+       status = ext4fs_devread(blkno << LOG2_EXT2_BLOCK_SIZE(data), blkoff,
+                               sizeof(struct ext2_inode), (char *)inode);
+       if (status == 0)
+               return 0;
+
+       return 1;
+}
+
+long int read_allocated_block(struct ext2_inode *inode, int fileblock)
+{
+       long int blknr;
+       int blksz;
+       int log2_blksz;
+       int status;
+       long int rblock;
+       long int perblock_parent;
+       long int perblock_child;
+       unsigned long long start;
+       /* get the blocksize of the filesystem */
+       blksz = EXT2_BLOCK_SIZE(ext4fs_root);
+       log2_blksz = LOG2_EXT2_BLOCK_SIZE(ext4fs_root);
+       if (le32_to_cpu(inode->flags) & EXT4_EXTENTS_FL) {
+               char *buf = zalloc(blksz);
+               if (!buf)
+                       return -ENOMEM;
+               struct ext4_extent_header *ext_block;
+               struct ext4_extent *extent;
+               int i = -1;
+               ext_block = ext4fs_get_extent_block(ext4fs_root, buf,
+                                                   (struct ext4_extent_header
+                                                    *)inode->b.
+                                                   blocks.dir_blocks,
+                                                   fileblock, log2_blksz);
+               if (!ext_block) {
+                       printf("invalid extent block\n");
+                       free(buf);
+                       return -EINVAL;
+               }
+
+               extent = (struct ext4_extent *)(ext_block + 1);
+
+               do {
+                       i++;
+                       if (i >= le32_to_cpu(ext_block->eh_entries))
+                               break;
+               } while (fileblock >= le32_to_cpu(extent[i].ee_block));
+               if (--i >= 0) {
+                       fileblock -= le32_to_cpu(extent[i].ee_block);
+                       if (fileblock >= le32_to_cpu(extent[i].ee_len)) {
+                               free(buf);
+                               return 0;
+                       }
+
+                       start = le32_to_cpu(extent[i].ee_start_hi);
+                       start = (start << 32) +
+                                       le32_to_cpu(extent[i].ee_start_lo);
+                       free(buf);
+                       return fileblock + start;
+               }
+
+               printf("Extent Error\n");
+               free(buf);
+               return -1;
+       }
+
+       /* Direct blocks. */
+       if (fileblock < INDIRECT_BLOCKS)
+               blknr = __le32_to_cpu(inode->b.blocks.dir_blocks[fileblock]);
+
+       /* Indirect. */
+       else if (fileblock < (INDIRECT_BLOCKS + (blksz / 4))) {
+               if (ext4fs_indir1_block == NULL) {
+                       ext4fs_indir1_block = zalloc(blksz);
+                       if (ext4fs_indir1_block == NULL) {
+                               printf("** SI ext2fs read block (indir 1)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir1_size = blksz;
+                       ext4fs_indir1_blkno = -1;
+               }
+               if (blksz != ext4fs_indir1_size) {
+                       free(ext4fs_indir1_block);
+                       ext4fs_indir1_block = NULL;
+                       ext4fs_indir1_size = 0;
+                       ext4fs_indir1_blkno = -1;
+                       ext4fs_indir1_block = zalloc(blksz);
+                       if (ext4fs_indir1_block == NULL) {
+                               printf("** SI ext2fs read block (indir 1):"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir1_size = blksz;
+               }
+               if ((__le32_to_cpu(inode->b.blocks.indir_block) <<
+                    log2_blksz) != ext4fs_indir1_blkno) {
+                       status =
+                           ext4fs_devread(__le32_to_cpu
+                                          (inode->b.blocks.
+                                           indir_block) << log2_blksz, 0,
+                                          blksz, (char *)ext4fs_indir1_block);
+                       if (status == 0) {
+                               printf("** SI ext2fs read block (indir 1)"
+                                       "failed. **\n");
+                               return 0;
+                       }
+                       ext4fs_indir1_blkno =
+                               __le32_to_cpu(inode->b.blocks.
+                                              indir_block) << log2_blksz;
+               }
+               blknr = __le32_to_cpu(ext4fs_indir1_block
+                                     [fileblock - INDIRECT_BLOCKS]);
+       }
+       /* Double indirect. */
+       else if (fileblock < (INDIRECT_BLOCKS + (blksz / 4 *
+                                       (blksz / 4 + 1)))) {
+
+               long int perblock = blksz / 4;
+               long int rblock = fileblock - (INDIRECT_BLOCKS + blksz / 4);
+
+               if (ext4fs_indir1_block == NULL) {
+                       ext4fs_indir1_block = zalloc(blksz);
+                       if (ext4fs_indir1_block == NULL) {
+                               printf("** DI ext2fs read block (indir 2 1)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir1_size = blksz;
+                       ext4fs_indir1_blkno = -1;
+               }
+               if (blksz != ext4fs_indir1_size) {
+                       free(ext4fs_indir1_block);
+                       ext4fs_indir1_block = NULL;
+                       ext4fs_indir1_size = 0;
+                       ext4fs_indir1_blkno = -1;
+                       ext4fs_indir1_block = zalloc(blksz);
+                       if (ext4fs_indir1_block == NULL) {
+                               printf("** DI ext2fs read block (indir 2 1)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir1_size = blksz;
+               }
+               if ((__le32_to_cpu(inode->b.blocks.double_indir_block) <<
+                    log2_blksz) != ext4fs_indir1_blkno) {
+                       status =
+                           ext4fs_devread(__le32_to_cpu
+                                          (inode->b.blocks.
+                                           double_indir_block) << log2_blksz,
+                                          0, blksz,
+                                          (char *)ext4fs_indir1_block);
+                       if (status == 0) {
+                               printf("** DI ext2fs read block (indir 2 1)"
+                                       "failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir1_blkno =
+                           __le32_to_cpu(inode->b.blocks.double_indir_block) <<
+                           log2_blksz;
+               }
+
+               if (ext4fs_indir2_block == NULL) {
+                       ext4fs_indir2_block = zalloc(blksz);
+                       if (ext4fs_indir2_block == NULL) {
+                               printf("** DI ext2fs read block (indir 2 2)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir2_size = blksz;
+                       ext4fs_indir2_blkno = -1;
+               }
+               if (blksz != ext4fs_indir2_size) {
+                       free(ext4fs_indir2_block);
+                       ext4fs_indir2_block = NULL;
+                       ext4fs_indir2_size = 0;
+                       ext4fs_indir2_blkno = -1;
+                       ext4fs_indir2_block = zalloc(blksz);
+                       if (ext4fs_indir2_block == NULL) {
+                               printf("** DI ext2fs read block (indir 2 2)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir2_size = blksz;
+               }
+               if ((__le32_to_cpu(ext4fs_indir1_block[rblock / perblock]) <<
+                    log2_blksz) != ext4fs_indir2_blkno) {
+                       status = ext4fs_devread(__le32_to_cpu
+                                               (ext4fs_indir1_block
+                                                [rblock /
+                                                 perblock]) << log2_blksz, 0,
+                                               blksz,
+                                               (char *)ext4fs_indir2_block);
+                       if (status == 0) {
+                               printf("** DI ext2fs read block (indir 2 2)"
+                                       "failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir2_blkno =
+                           __le32_to_cpu(ext4fs_indir1_block[rblock
+                                                             /
+                                                             perblock]) <<
+                           log2_blksz;
+               }
+               blknr = __le32_to_cpu(ext4fs_indir2_block[rblock % perblock]);
+       }
+       /* Tripple indirect. */
+       else {
+               rblock = fileblock - (INDIRECT_BLOCKS + blksz / 4 +
+                                     (blksz / 4 * blksz / 4));
+               perblock_child = blksz / 4;
+               perblock_parent = ((blksz / 4) * (blksz / 4));
+
+               if (ext4fs_indir1_block == NULL) {
+                       ext4fs_indir1_block = zalloc(blksz);
+                       if (ext4fs_indir1_block == NULL) {
+                               printf("** TI ext2fs read block (indir 2 1)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir1_size = blksz;
+                       ext4fs_indir1_blkno = -1;
+               }
+               if (blksz != ext4fs_indir1_size) {
+                       free(ext4fs_indir1_block);
+                       ext4fs_indir1_block = NULL;
+                       ext4fs_indir1_size = 0;
+                       ext4fs_indir1_blkno = -1;
+                       ext4fs_indir1_block = zalloc(blksz);
+                       if (ext4fs_indir1_block == NULL) {
+                               printf("** TI ext2fs read block (indir 2 1)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir1_size = blksz;
+               }
+               if ((__le32_to_cpu(inode->b.blocks.triple_indir_block) <<
+                    log2_blksz) != ext4fs_indir1_blkno) {
+                       status = ext4fs_devread
+                           (__le32_to_cpu(inode->b.blocks.triple_indir_block)
+                            << log2_blksz, 0, blksz,
+                            (char *)ext4fs_indir1_block);
+                       if (status == 0) {
+                               printf("** TI ext2fs read block (indir 2 1)"
+                                       "failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir1_blkno =
+                           __le32_to_cpu(inode->b.blocks.triple_indir_block) <<
+                           log2_blksz;
+               }
+
+               if (ext4fs_indir2_block == NULL) {
+                       ext4fs_indir2_block = zalloc(blksz);
+                       if (ext4fs_indir2_block == NULL) {
+                               printf("** TI ext2fs read block (indir 2 2)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir2_size = blksz;
+                       ext4fs_indir2_blkno = -1;
+               }
+               if (blksz != ext4fs_indir2_size) {
+                       free(ext4fs_indir2_block);
+                       ext4fs_indir2_block = NULL;
+                       ext4fs_indir2_size = 0;
+                       ext4fs_indir2_blkno = -1;
+                       ext4fs_indir2_block = zalloc(blksz);
+                       if (ext4fs_indir2_block == NULL) {
+                               printf("** TI ext2fs read block (indir 2 2)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir2_size = blksz;
+               }
+               if ((__le32_to_cpu(ext4fs_indir1_block[rblock /
+                                                      perblock_parent]) <<
+                    log2_blksz)
+                   != ext4fs_indir2_blkno) {
+                       status = ext4fs_devread(__le32_to_cpu
+                                               (ext4fs_indir1_block
+                                                [rblock /
+                                                 perblock_parent]) <<
+                                               log2_blksz, 0, blksz,
+                                               (char *)ext4fs_indir2_block);
+                       if (status == 0) {
+                               printf("** TI ext2fs read block (indir 2 2)"
+                                       "failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir2_blkno =
+                           __le32_to_cpu(ext4fs_indir1_block[rblock /
+                                                             perblock_parent])
+                           << log2_blksz;
+               }
+
+               if (ext4fs_indir3_block == NULL) {
+                       ext4fs_indir3_block = zalloc(blksz);
+                       if (ext4fs_indir3_block == NULL) {
+                               printf("** TI ext2fs read block (indir 2 2)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir3_size = blksz;
+                       ext4fs_indir3_blkno = -1;
+               }
+               if (blksz != ext4fs_indir3_size) {
+                       free(ext4fs_indir3_block);
+                       ext4fs_indir3_block = NULL;
+                       ext4fs_indir3_size = 0;
+                       ext4fs_indir3_blkno = -1;
+                       ext4fs_indir3_block = zalloc(blksz);
+                       if (ext4fs_indir3_block == NULL) {
+                               printf("** TI ext2fs read block (indir 2 2)"
+                                       "malloc failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir3_size = blksz;
+               }
+               if ((__le32_to_cpu(ext4fs_indir2_block[rblock
+                                                      /
+                                                      perblock_child]) <<
+                    log2_blksz) != ext4fs_indir3_blkno) {
+                       status =
+                           ext4fs_devread(__le32_to_cpu
+                                          (ext4fs_indir2_block
+                                           [(rblock / perblock_child)
+                                            % (blksz / 4)]) << log2_blksz, 0,
+                                          blksz, (char *)ext4fs_indir3_block);
+                       if (status == 0) {
+                               printf("** TI ext2fs read block (indir 2 2)"
+                                      "failed. **\n");
+                               return -1;
+                       }
+                       ext4fs_indir3_blkno =
+                           __le32_to_cpu(ext4fs_indir2_block[(rblock /
+                                                              perblock_child) %
+                                                             (blksz /
+                                                              4)]) <<
+                           log2_blksz;
+               }
+
+               blknr = __le32_to_cpu(ext4fs_indir3_block
+                                     [rblock % perblock_child]);
+       }
+       debug("ext4fs_read_block %ld\n", blknr);
+
+       return blknr;
+}
+
+void ext4fs_close(void)
+{
+       if ((ext4fs_file != NULL) && (ext4fs_root != NULL)) {
+               ext4fs_free_node(ext4fs_file, &ext4fs_root->diropen);
+               ext4fs_file = NULL;
+       }
+       if (ext4fs_root != NULL) {
+               free(ext4fs_root);
+               ext4fs_root = NULL;
+       }
+       if (ext4fs_indir1_block != NULL) {
+               free(ext4fs_indir1_block);
+               ext4fs_indir1_block = NULL;
+               ext4fs_indir1_size = 0;
+               ext4fs_indir1_blkno = -1;
+       }
+       if (ext4fs_indir2_block != NULL) {
+               free(ext4fs_indir2_block);
+               ext4fs_indir2_block = NULL;
+               ext4fs_indir2_size = 0;
+               ext4fs_indir2_blkno = -1;
+       }
+       if (ext4fs_indir3_block != NULL) {
+               free(ext4fs_indir3_block);
+               ext4fs_indir3_block = NULL;
+               ext4fs_indir3_size = 0;
+               ext4fs_indir3_blkno = -1;
+       }
+}
+
+int ext4fs_iterate_dir(struct ext2fs_node *dir, char *name,
+                               struct ext2fs_node **fnode, int *ftype)
+{
+       unsigned int fpos = 0;
+       int status;
+       struct ext2fs_node *diro = (struct ext2fs_node *) dir;
+
+#ifdef DEBUG
+       if (name != NULL)
+               printf("Iterate dir %s\n", name);
+#endif /* of DEBUG */
+       if (!diro->inode_read) {
+               status = ext4fs_read_inode(diro->data, diro->ino, &diro->inode);
+               if (status == 0)
+                       return 0;
+       }
+       /* Search the file.  */
+       while (fpos < __le32_to_cpu(diro->inode.size)) {
+               struct ext2_dirent dirent;
+
+               status = ext4fs_read_file(diro, fpos,
+                                          sizeof(struct ext2_dirent),
+                                          (char *) &dirent);
+               if (status < 1)
+                       return 0;
+
+               if (dirent.namelen != 0) {
+                       char filename[dirent.namelen + 1];
+                       struct ext2fs_node *fdiro;
+                       int type = FILETYPE_UNKNOWN;
+
+                       status = ext4fs_read_file(diro,
+                                                 fpos +
+                                                 sizeof(struct ext2_dirent),
+                                                 dirent.namelen, filename);
+                       if (status < 1)
+                               return 0;
+
+                       fdiro = zalloc(sizeof(struct ext2fs_node));
+                       if (!fdiro)
+                               return 0;
+
+                       fdiro->data = diro->data;
+                       fdiro->ino = __le32_to_cpu(dirent.inode);
+
+                       filename[dirent.namelen] = '\0';
+
+                       if (dirent.filetype != FILETYPE_UNKNOWN) {
+                               fdiro->inode_read = 0;
+
+                               if (dirent.filetype == FILETYPE_DIRECTORY)
+                                       type = FILETYPE_DIRECTORY;
+                               else if (dirent.filetype == FILETYPE_SYMLINK)
+                                       type = FILETYPE_SYMLINK;
+                               else if (dirent.filetype == FILETYPE_REG)
+                                       type = FILETYPE_REG;
+                       } else {
+                               status = ext4fs_read_inode(diro->data,
+                                                          __le32_to_cpu
+                                                          (dirent.inode),
+                                                          &fdiro->inode);
+                               if (status == 0) {
+                                       free(fdiro);
+                                       return 0;
+                               }
+                               fdiro->inode_read = 1;
+
+                               if ((__le16_to_cpu(fdiro->inode.mode) &
+                                    FILETYPE_INO_MASK) ==
+                                   FILETYPE_INO_DIRECTORY) {
+                                       type = FILETYPE_DIRECTORY;
+                               } else if ((__le16_to_cpu(fdiro->inode.mode)
+                                           & FILETYPE_INO_MASK) ==
+                                          FILETYPE_INO_SYMLINK) {
+                                       type = FILETYPE_SYMLINK;
+                               } else if ((__le16_to_cpu(fdiro->inode.mode)
+                                           & FILETYPE_INO_MASK) ==
+                                          FILETYPE_INO_REG) {
+                                       type = FILETYPE_REG;
+                               }
+                       }
+#ifdef DEBUG
+                       printf("iterate >%s<\n", filename);
+#endif /* of DEBUG */
+                       if ((name != NULL) && (fnode != NULL)
+                           && (ftype != NULL)) {
+                               if (strcmp(filename, name) == 0) {
+                                       *ftype = type;
+                                       *fnode = fdiro;
+                                       return 1;
+                               }
+                       } else {
+                               if (fdiro->inode_read == 0) {
+                                       status = ext4fs_read_inode(diro->data,
+                                                                __le32_to_cpu(
+                                                                dirent.inode),
+                                                                &fdiro->inode);
+                                       if (status == 0) {
+                                               free(fdiro);
+                                               return 0;
+                                       }
+                                       fdiro->inode_read = 1;
+                               }
+                               switch (type) {
+                               case FILETYPE_DIRECTORY:
+                                       printf("<DIR> ");
+                                       break;
+                               case FILETYPE_SYMLINK:
+                                       printf("<SYM> ");
+                                       break;
+                               case FILETYPE_REG:
+                                       printf("      ");
+                                       break;
+                               default:
+                                       printf("< ? > ");
+                                       break;
+                               }
+                               printf("%10d %s\n",
+                                       __le32_to_cpu(fdiro->inode.size),
+                                       filename);
+                       }
+                       free(fdiro);
+               }
+               fpos += __le16_to_cpu(dirent.direntlen);
+       }
+       return 0;
+}
+
+static char *ext4fs_read_symlink(struct ext2fs_node *node)
+{
+       char *symlink;
+       struct ext2fs_node *diro = node;
+       int status;
+
+       if (!diro->inode_read) {
+               status = ext4fs_read_inode(diro->data, diro->ino, &diro->inode);
+               if (status == 0)
+                       return 0;
+       }
+       symlink = zalloc(__le32_to_cpu(diro->inode.size) + 1);
+       if (!symlink)
+               return 0;
+
+       if (__le32_to_cpu(diro->inode.size) <= 60) {
+               strncpy(symlink, diro->inode.b.symlink,
+                        __le32_to_cpu(diro->inode.size));
+       } else {
+               status = ext4fs_read_file(diro, 0,
+                                          __le32_to_cpu(diro->inode.size),
+                                          symlink);
+               if (status == 0) {
+                       free(symlink);
+                       return 0;
+               }
+       }
+       symlink[__le32_to_cpu(diro->inode.size)] = '\0';
+       return symlink;
+}
+
+static int ext4fs_find_file1(const char *currpath,
+                            struct ext2fs_node *currroot,
+                            struct ext2fs_node **currfound, int *foundtype)
+{
+       char fpath[strlen(currpath) + 1];
+       char *name = fpath;
+       char *next;
+       int status;
+       int type = FILETYPE_DIRECTORY;
+       struct ext2fs_node *currnode = currroot;
+       struct ext2fs_node *oldnode = currroot;
+
+       strncpy(fpath, currpath, strlen(currpath) + 1);
+
+       /* Remove all leading slashes. */
+       while (*name == '/')
+               name++;
+
+       if (!*name) {
+               *currfound = currnode;
+               return 1;
+       }
+
+       for (;;) {
+               int found;
+
+               /* Extract the actual part from the pathname. */
+               next = strchr(name, '/');
+               if (next) {
+                       /* Remove all leading slashes. */
+                       while (*next == '/')
+                               *(next++) = '\0';
+               }
+
+               if (type != FILETYPE_DIRECTORY) {
+                       ext4fs_free_node(currnode, currroot);
+                       return 0;
+               }
+
+               oldnode = currnode;
+
+               /* Iterate over the directory. */
+               found = ext4fs_iterate_dir(currnode, name, &currnode, &type);
+               if (found == 0)
+                       return 0;
+
+               if (found == -1)
+                       break;
+
+               /* Read in the symlink and follow it. */
+               if (type == FILETYPE_SYMLINK) {
+                       char *symlink;
+
+                       /* Test if the symlink does not loop. */
+                       if (++symlinknest == 8) {
+                               ext4fs_free_node(currnode, currroot);
+                               ext4fs_free_node(oldnode, currroot);
+                               return 0;
+                       }
+
+                       symlink = ext4fs_read_symlink(currnode);
+                       ext4fs_free_node(currnode, currroot);
+
+                       if (!symlink) {
+                               ext4fs_free_node(oldnode, currroot);
+                               return 0;
+                       }
+
+                       debug("Got symlink >%s<\n", symlink);
+
+                       if (symlink[0] == '/') {
+                               ext4fs_free_node(oldnode, currroot);
+                               oldnode = &ext4fs_root->diropen;
+                       }
+
+                       /* Lookup the node the symlink points to. */
+                       status = ext4fs_find_file1(symlink, oldnode,
+                                                   &currnode, &type);
+
+                       free(symlink);
+
+                       if (status == 0) {
+                               ext4fs_free_node(oldnode, currroot);
+                               return 0;
+                       }
+               }
+
+               ext4fs_free_node(oldnode, currroot);
+
+               /* Found the node! */
+               if (!next || *next == '\0') {
+                       *currfound = currnode;
+                       *foundtype = type;
+                       return 1;
+               }
+               name = next;
+       }
+       return -1;
+}
+
+int ext4fs_find_file(const char *path, struct ext2fs_node *rootnode,
+       struct ext2fs_node **foundnode, int expecttype)
+{
+       int status;
+       int foundtype = FILETYPE_DIRECTORY;
+
+       symlinknest = 0;
+       if (!path)
+               return 0;
+
+       status = ext4fs_find_file1(path, rootnode, foundnode, &foundtype);
+       if (status == 0)
+               return 0;
+
+       /* Check if the node that was found was of the expected type. */
+       if ((expecttype == FILETYPE_REG) && (foundtype != expecttype))
+               return 0;
+       else if ((expecttype == FILETYPE_DIRECTORY)
+                  && (foundtype != expecttype))
+               return 0;
+
+       return 1;
+}
+
+int ext4fs_open(const char *filename)
+{
+       struct ext2fs_node *fdiro = NULL;
+       int status;
+       int len;
+
+       if (ext4fs_root == NULL)
+               return -1;
+
+       ext4fs_file = NULL;
+       status = ext4fs_find_file(filename, &ext4fs_root->diropen, &fdiro,
+                                 FILETYPE_REG);
+       if (status == 0)
+               goto fail;
+
+       if (!fdiro->inode_read) {
+               status = ext4fs_read_inode(fdiro->data, fdiro->ino,
+                               &fdiro->inode);
+               if (status == 0)
+                       goto fail;
+       }
+       len = __le32_to_cpu(fdiro->inode.size);
+       ext4fs_file = fdiro;
+
+       return len;
+fail:
+       ext4fs_free_node(fdiro, &ext4fs_root->diropen);
+
+       return -1;
+}
+
+int ext4fs_mount(unsigned part_length)
+{
+       struct ext2_data *data;
+       int status;
+       struct ext_filesystem *fs = get_fs();
+       data = zalloc(sizeof(struct ext2_data));
+       if (!data)
+               return 0;
+
+       /* Read the superblock. */
+       status = ext4fs_devread(1 * 2, 0, sizeof(struct ext2_sblock),
+                               (char *)&data->sblock);
+
+       if (status == 0)
+               goto fail;
+
+       /* Make sure this is an ext2 filesystem. */
+       if (__le16_to_cpu(data->sblock.magic) != EXT2_MAGIC)
+               goto fail;
+
+       if (__le32_to_cpu(data->sblock.revision_level == 0))
+               fs->inodesz = 128;
+       else
+               fs->inodesz = __le16_to_cpu(data->sblock.inode_size);
+
+       debug("EXT2 rev %d, inode_size %d\n",
+              __le32_to_cpu(data->sblock.revision_level), fs->inodesz);
+
+       data->diropen.data = data;
+       data->diropen.ino = 2;
+       data->diropen.inode_read = 1;
+       data->inode = &data->diropen.inode;
+
+       status = ext4fs_read_inode(data, 2, data->inode);
+       if (status == 0)
+               goto fail;
+
+       ext4fs_root = data;
+
+       return 1;
+fail:
+       printf("Failed to mount ext2 filesystem...\n");
+       free(data);
+       ext4fs_root = NULL;
+
+       return 0;
+}
diff --git a/fs/ext4/ext4_common.h b/fs/ext4/ext4_common.h
new file mode 100644 (file)
index 0000000..0af625d
--- /dev/null
@@ -0,0 +1,93 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * ext4ls and ext4load :  based on ext2 ls load support in Uboot.
+ *
+ * (C) Copyright 2004
+ * esd gmbh <www.esd-electronics.com>
+ * Reinhard Arlt <reinhard.arlt@esd-electronics.com>
+ *
+ * based on code from grub2 fs/ext2.c and fs/fshelp.c by
+ * GRUB  --  GRand Unified Bootloader
+ * Copyright (C) 2003, 2004  Free Software Foundation, Inc.
+ *
+ * ext4write : Based on generic ext4 protocol.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef __EXT4_COMMON__
+#define __EXT4_COMMON__
+#include <ext_common.h>
+#include <ext4fs.h>
+#include <malloc.h>
+#include <asm/errno.h>
+#if defined(CONFIG_CMD_EXT4_WRITE)
+#include "ext4_journal.h"
+#include "crc16.h"
+#endif
+
+#define YES            1
+#define NO             0
+#define TRUE           1
+#define FALSE          0
+#define RECOVER        1
+#define SCAN           0
+
+#define S_IFLNK                0120000         /* symbolic link */
+#define BLOCK_NO_ONE           1
+#define SUPERBLOCK_SECTOR      2
+#define SUPERBLOCK_SIZE        1024
+#define F_FILE                 1
+
+static inline void *zalloc(size_t size)
+{
+       void *p = memalign(ARCH_DMA_MINALIGN, size);
+       memset(p, 0, size);
+       return p;
+}
+
+extern unsigned long part_offset;
+int ext4fs_read_inode(struct ext2_data *data, int ino,
+                     struct ext2_inode *inode);
+int ext4fs_read_file(struct ext2fs_node *node, int pos,
+               unsigned int len, char *buf);
+int ext4fs_find_file(const char *path, struct ext2fs_node *rootnode,
+                       struct ext2fs_node **foundnode, int expecttype);
+int ext4fs_iterate_dir(struct ext2fs_node *dir, char *name,
+                       struct ext2fs_node **fnode, int *ftype);
+
+#if defined(CONFIG_CMD_EXT4_WRITE)
+uint32_t ext4fs_div_roundup(uint32_t size, uint32_t n);
+int ext4fs_checksum_update(unsigned int i);
+int ext4fs_get_parent_inode_num(const char *dirname, char *dname, int flags);
+void ext4fs_update_parent_dentry(char *filename, int *p_ino, int file_type);
+long int ext4fs_get_new_blk_no(void);
+int ext4fs_get_new_inode_no(void);
+void ext4fs_reset_block_bmap(long int blockno, unsigned char *buffer,
+                                       int index);
+int ext4fs_set_block_bmap(long int blockno, unsigned char *buffer, int index);
+int ext4fs_set_inode_bmap(int inode_no, unsigned char *buffer, int index);
+void ext4fs_reset_inode_bmap(int inode_no, unsigned char *buffer, int index);
+int ext4fs_iget(int inode_no, struct ext2_inode *inode);
+void ext4fs_allocate_blocks(struct ext2_inode *file_inode,
+                               unsigned int total_remaining_blocks,
+                               unsigned int *total_no_of_block);
+void put_ext4(uint64_t off, void *buf, uint32_t size);
+#endif
+#endif
diff --git a/fs/ext4/ext4_journal.c b/fs/ext4/ext4_journal.c
new file mode 100644 (file)
index 0000000..8a252d6
--- /dev/null
@@ -0,0 +1,667 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * Journal data structures and headers for Journaling feature of ext4
+ * have been referred from JBD2 (Journaling Block device 2)
+ * implementation in Linux Kernel.
+ * Written by Stephen C. Tweedie <sct@redhat.com>
+ *
+ * Copyright 1998-2000 Red Hat, Inc --- 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.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <common.h>
+#include <ext4fs.h>
+#include <malloc.h>
+#include <ext_common.h>
+#include "ext4_common.h"
+
+static struct revoke_blk_list *revk_blk_list;
+static struct revoke_blk_list *prev_node;
+static int first_node = TRUE;
+
+int gindex;
+int gd_index;
+int jrnl_blk_idx;
+struct journal_log *journal_ptr[MAX_JOURNAL_ENTRIES];
+struct dirty_blocks *dirty_block_ptr[MAX_JOURNAL_ENTRIES];
+
+int ext4fs_init_journal(void)
+{
+       int i;
+       char *temp = NULL;
+       struct ext_filesystem *fs = get_fs();
+
+       /* init globals */
+       revk_blk_list = NULL;
+       prev_node = NULL;
+       gindex = 0;
+       gd_index = 0;
+       jrnl_blk_idx = 1;
+
+       for (i = 0; i < MAX_JOURNAL_ENTRIES; i++) {
+               journal_ptr[i] = zalloc(sizeof(struct journal_log));
+               if (!journal_ptr[i])
+                       goto fail;
+               dirty_block_ptr[i] = zalloc(sizeof(struct dirty_blocks));
+               if (!dirty_block_ptr[i])
+                       goto fail;
+               journal_ptr[i]->buf = NULL;
+               journal_ptr[i]->blknr = -1;
+
+               dirty_block_ptr[i]->buf = NULL;
+               dirty_block_ptr[i]->blknr = -1;
+       }
+
+       if (fs->blksz == 4096) {
+               temp = zalloc(fs->blksz);
+               if (!temp)
+                       goto fail;
+               journal_ptr[gindex]->buf = zalloc(fs->blksz);
+               if (!journal_ptr[gindex]->buf)
+                       goto fail;
+               ext4fs_devread(0, 0, fs->blksz, temp);
+               memcpy(temp + SUPERBLOCK_SIZE, fs->sb, SUPERBLOCK_SIZE);
+               memcpy(journal_ptr[gindex]->buf, temp, fs->blksz);
+               journal_ptr[gindex++]->blknr = 0;
+               free(temp);
+       } else {
+               journal_ptr[gindex]->buf = zalloc(fs->blksz);
+               if (!journal_ptr[gindex]->buf)
+                       goto fail;
+               memcpy(journal_ptr[gindex]->buf, fs->sb, SUPERBLOCK_SIZE);
+               journal_ptr[gindex++]->blknr = 1;
+       }
+
+       /* Check the file system state using journal super block */
+       if (ext4fs_check_journal_state(SCAN))
+               goto fail;
+       /* Check the file system state using journal super block */
+       if (ext4fs_check_journal_state(RECOVER))
+               goto fail;
+
+       return 0;
+fail:
+       return -1;
+}
+
+void ext4fs_dump_metadata(void)
+{
+       struct ext_filesystem *fs = get_fs();
+       int i;
+       for (i = 0; i < MAX_JOURNAL_ENTRIES; i++) {
+               if (dirty_block_ptr[i]->blknr == -1)
+                       break;
+               put_ext4((uint64_t) ((uint64_t)dirty_block_ptr[i]->blknr *
+                               (uint64_t)fs->blksz), dirty_block_ptr[i]->buf,
+                                                               fs->blksz);
+       }
+}
+
+void ext4fs_free_journal(void)
+{
+       int i;
+       for (i = 0; i < MAX_JOURNAL_ENTRIES; i++) {
+               if (dirty_block_ptr[i]->blknr == -1)
+                       break;
+               if (dirty_block_ptr[i]->buf)
+                       free(dirty_block_ptr[i]->buf);
+       }
+
+       for (i = 0; i < MAX_JOURNAL_ENTRIES; i++) {
+               if (journal_ptr[i]->blknr == -1)
+                       break;
+               if (journal_ptr[i]->buf)
+                       free(journal_ptr[i]->buf);
+       }
+
+       for (i = 0; i < MAX_JOURNAL_ENTRIES; i++) {
+               if (journal_ptr[i])
+                       free(journal_ptr[i]);
+               if (dirty_block_ptr[i])
+                       free(dirty_block_ptr[i]);
+       }
+       gindex = 0;
+       gd_index = 0;
+       jrnl_blk_idx = 1;
+}
+
+int ext4fs_log_gdt(char *gd_table)
+{
+       struct ext_filesystem *fs = get_fs();
+       short i;
+       long int var = fs->gdtable_blkno;
+       for (i = 0; i < fs->no_blk_pergdt; i++) {
+               journal_ptr[gindex]->buf = zalloc(fs->blksz);
+               if (!journal_ptr[gindex]->buf)
+                       return -ENOMEM;
+               memcpy(journal_ptr[gindex]->buf, gd_table, fs->blksz);
+               gd_table += fs->blksz;
+               journal_ptr[gindex++]->blknr = var++;
+       }
+
+       return 0;
+}
+
+/*
+ * This function stores the backup copy of meta data in RAM
+ * journal_buffer -- Buffer containing meta data
+ * blknr -- Block number on disk of the meta data buffer
+ */
+int ext4fs_log_journal(char *journal_buffer, long int blknr)
+{
+       struct ext_filesystem *fs = get_fs();
+       short i;
+
+       if (!journal_buffer) {
+               printf("Invalid input arguments %s\n", __func__);
+               return -EINVAL;
+       }
+
+       for (i = 0; i < MAX_JOURNAL_ENTRIES; i++) {
+               if (journal_ptr[i]->blknr == -1)
+                       break;
+               if (journal_ptr[i]->blknr == blknr)
+                       return 0;
+       }
+
+       journal_ptr[gindex]->buf = zalloc(fs->blksz);
+       if (!journal_ptr[gindex]->buf)
+               return -ENOMEM;
+
+       memcpy(journal_ptr[gindex]->buf, journal_buffer, fs->blksz);
+       journal_ptr[gindex++]->blknr = blknr;
+
+       return 0;
+}
+
+/*
+ * This function stores the modified meta data in RAM
+ * metadata_buffer -- Buffer containing meta data
+ * blknr -- Block number on disk of the meta data buffer
+ */
+int ext4fs_put_metadata(char *metadata_buffer, long int blknr)
+{
+       struct ext_filesystem *fs = get_fs();
+       if (!metadata_buffer) {
+               printf("Invalid input arguments %s\n", __func__);
+               return -EINVAL;
+       }
+       dirty_block_ptr[gd_index]->buf = zalloc(fs->blksz);
+       if (!dirty_block_ptr[gd_index]->buf)
+               return -ENOMEM;
+       memcpy(dirty_block_ptr[gd_index]->buf, metadata_buffer, fs->blksz);
+       dirty_block_ptr[gd_index++]->blknr = blknr;
+
+       return 0;
+}
+
+void print_revoke_blks(char *revk_blk)
+{
+       int offset;
+       int max;
+       long int blocknr;
+       struct journal_revoke_header_t *header;
+
+       if (revk_blk == NULL)
+               return;
+
+       header = (struct journal_revoke_header_t *) revk_blk;
+       offset = sizeof(struct journal_revoke_header_t);
+       max = be32_to_cpu(header->r_count);
+       printf("total bytes %d\n", max);
+
+       while (offset < max) {
+               blocknr = be32_to_cpu(*((long int *)(revk_blk + offset)));
+               printf("revoke blknr is %ld\n", blocknr);
+               offset += 4;
+       }
+}
+
+static struct revoke_blk_list *_get_node(void)
+{
+       struct revoke_blk_list *tmp_node;
+       tmp_node = zalloc(sizeof(struct revoke_blk_list));
+       if (tmp_node == NULL)
+               return NULL;
+       tmp_node->content = NULL;
+       tmp_node->next = NULL;
+
+       return tmp_node;
+}
+
+void ext4fs_push_revoke_blk(char *buffer)
+{
+       struct revoke_blk_list *node = NULL;
+       struct ext_filesystem *fs = get_fs();
+       if (buffer == NULL) {
+               printf("buffer ptr is NULL\n");
+               return;
+       }
+       node = _get_node();
+       if (!node) {
+               printf("_get_node: malloc failed\n");
+               return;
+       }
+
+       node->content = zalloc(fs->blksz);
+       if (node->content == NULL)
+               return;
+       memcpy(node->content, buffer, fs->blksz);
+
+       if (first_node == TRUE) {
+               revk_blk_list = node;
+               prev_node = node;
+                first_node = FALSE;
+       } else {
+               prev_node->next = node;
+               prev_node = node;
+       }
+}
+
+void ext4fs_free_revoke_blks(void)
+{
+       struct revoke_blk_list *tmp_node = revk_blk_list;
+       struct revoke_blk_list *next_node = NULL;
+
+       while (tmp_node != NULL) {
+               if (tmp_node->content)
+                       free(tmp_node->content);
+               tmp_node = tmp_node->next;
+       }
+
+       tmp_node = revk_blk_list;
+       while (tmp_node != NULL) {
+               next_node = tmp_node->next;
+               free(tmp_node);
+               tmp_node = next_node;
+       }
+
+       revk_blk_list = NULL;
+       prev_node = NULL;
+       first_node = TRUE;
+}
+
+int check_blknr_for_revoke(long int blknr, int sequence_no)
+{
+       struct journal_revoke_header_t *header;
+       int offset;
+       int max;
+       long int blocknr;
+       char *revk_blk;
+       struct revoke_blk_list *tmp_revk_node = revk_blk_list;
+       while (tmp_revk_node != NULL) {
+               revk_blk = tmp_revk_node->content;
+
+               header = (struct journal_revoke_header_t *) revk_blk;
+               if (sequence_no < be32_to_cpu(header->r_header.h_sequence)) {
+                       offset = sizeof(struct journal_revoke_header_t);
+                       max = be32_to_cpu(header->r_count);
+
+                       while (offset < max) {
+                               blocknr = be32_to_cpu(*((long int *)
+                                                 (revk_blk + offset)));
+                               if (blocknr == blknr)
+                                       goto found;
+                               offset += 4;
+                       }
+               }
+               tmp_revk_node = tmp_revk_node->next;
+       }
+
+       return -1;
+
+found:
+       return 0;
+}
+
+/*
+ * This function parses the journal blocks and replays the
+ * suceessful transactions. A transaction is successfull
+ * if commit block is found for a descriptor block
+ * The tags in descriptor block contain the disk block
+ * numbers of the metadata  to be replayed
+ */
+void recover_transaction(int prev_desc_logical_no)
+{
+       struct ext2_inode inode_journal;
+       struct ext_filesystem *fs = get_fs();
+       struct journal_header_t *jdb;
+       long int blknr;
+       char *p_jdb;
+       int ofs, flags;
+       int i;
+       struct ext3_journal_block_tag *tag;
+       char *temp_buff = zalloc(fs->blksz);
+       char *metadata_buff = zalloc(fs->blksz);
+       if (!temp_buff || !metadata_buff)
+               goto fail;
+       i = prev_desc_logical_no;
+       ext4fs_read_inode(ext4fs_root, EXT2_JOURNAL_INO,
+                         (struct ext2_inode *)&inode_journal);
+       blknr = read_allocated_block((struct ext2_inode *)
+                                    &inode_journal, i);
+       ext4fs_devread(blknr * fs->sect_perblk, 0, fs->blksz, temp_buff);
+       p_jdb = (char *)temp_buff;
+       jdb = (struct journal_header_t *) temp_buff;
+       ofs = sizeof(struct journal_header_t);
+
+       do {
+               tag = (struct ext3_journal_block_tag *)&p_jdb[ofs];
+               ofs += sizeof(struct ext3_journal_block_tag);
+
+               if (ofs > fs->blksz)
+                       break;
+
+               flags = be32_to_cpu(tag->flags);
+               if (!(flags & EXT3_JOURNAL_FLAG_SAME_UUID))
+                       ofs += 16;
+
+               i++;
+               debug("\t\ttag %u\n", be32_to_cpu(tag->block));
+               if (revk_blk_list != NULL) {
+                       if (check_blknr_for_revoke(be32_to_cpu(tag->block),
+                               be32_to_cpu(jdb->h_sequence)) == 0)
+                               continue;
+               }
+               blknr = read_allocated_block(&inode_journal, i);
+               ext4fs_devread(blknr * fs->sect_perblk, 0,
+                              fs->blksz, metadata_buff);
+               put_ext4((uint64_t)(be32_to_cpu(tag->block) * fs->blksz),
+                        metadata_buff, (uint32_t) fs->blksz);
+       } while (!(flags & EXT3_JOURNAL_FLAG_LAST_TAG));
+fail:
+       free(temp_buff);
+       free(metadata_buff);
+}
+
+void print_jrnl_status(int recovery_flag)
+{
+       if (recovery_flag == RECOVER)
+               printf("Journal Recovery Completed\n");
+       else
+               printf("Journal Scan Completed\n");
+}
+
+int ext4fs_check_journal_state(int recovery_flag)
+{
+       int i;
+       int DB_FOUND = NO;
+       long int blknr;
+       int transaction_state = TRANSACTION_COMPLETE;
+       int prev_desc_logical_no = 0;
+       int curr_desc_logical_no = 0;
+       int ofs, flags, block;
+       struct ext2_inode inode_journal;
+       struct journal_superblock_t *jsb = NULL;
+       struct journal_header_t *jdb = NULL;
+       char *p_jdb = NULL;
+       struct ext3_journal_block_tag *tag = NULL;
+       char *temp_buff = NULL;
+       char *temp_buff1 = NULL;
+       struct ext_filesystem *fs = get_fs();
+
+       temp_buff = zalloc(fs->blksz);
+       if (!temp_buff)
+               return -ENOMEM;
+       temp_buff1 = zalloc(fs->blksz);
+       if (!temp_buff1) {
+               free(temp_buff);
+               return -ENOMEM;
+       }
+
+       ext4fs_read_inode(ext4fs_root, EXT2_JOURNAL_INO, &inode_journal);
+       blknr = read_allocated_block(&inode_journal, EXT2_JOURNAL_SUPERBLOCK);
+       ext4fs_devread(blknr * fs->sect_perblk, 0, fs->blksz, temp_buff);
+       jsb = (struct journal_superblock_t *) temp_buff;
+
+       if (fs->sb->feature_incompat & EXT3_FEATURE_INCOMPAT_RECOVER) {
+               if (recovery_flag == RECOVER)
+                       printf("Recovery required\n");
+       } else {
+               if (recovery_flag == RECOVER)
+                       printf("File System is consistent\n");
+               goto end;
+       }
+
+       if (be32_to_cpu(jsb->s_start) == 0)
+               goto end;
+
+       if (!(jsb->s_feature_compat &
+                               cpu_to_be32(JBD2_FEATURE_COMPAT_CHECKSUM)))
+               jsb->s_feature_compat |=
+                               cpu_to_be32(JBD2_FEATURE_COMPAT_CHECKSUM);
+
+       i = be32_to_cpu(jsb->s_first);
+       while (1) {
+               block = be32_to_cpu(jsb->s_first);
+               blknr = read_allocated_block(&inode_journal, i);
+               memset(temp_buff1, '\0', fs->blksz);
+               ext4fs_devread(blknr * fs->sect_perblk,
+                              0, fs->blksz, temp_buff1);
+               jdb = (struct journal_header_t *) temp_buff1;
+
+               if (be32_to_cpu(jdb->h_blocktype) ==
+                   EXT3_JOURNAL_DESCRIPTOR_BLOCK) {
+                       if (be32_to_cpu(jdb->h_sequence) !=
+                           be32_to_cpu(jsb->s_sequence)) {
+                               print_jrnl_status(recovery_flag);
+                               break;
+                       }
+
+                       curr_desc_logical_no = i;
+                       if (transaction_state == TRANSACTION_COMPLETE)
+                               transaction_state = TRANSACTION_RUNNING;
+                       else
+                               return -1;
+                       p_jdb = (char *)temp_buff1;
+                       ofs = sizeof(struct journal_header_t);
+                       do {
+                               tag = (struct ext3_journal_block_tag *)
+                                   &p_jdb[ofs];
+                               ofs += sizeof(struct ext3_journal_block_tag);
+                               if (ofs > fs->blksz)
+                                       break;
+                               flags = be32_to_cpu(tag->flags);
+                               if (!(flags & EXT3_JOURNAL_FLAG_SAME_UUID))
+                                       ofs += 16;
+                               i++;
+                               debug("\t\ttag %u\n", be32_to_cpu(tag->block));
+                       } while (!(flags & EXT3_JOURNAL_FLAG_LAST_TAG));
+                       i++;
+                       DB_FOUND = YES;
+               } else if (be32_to_cpu(jdb->h_blocktype) ==
+                               EXT3_JOURNAL_COMMIT_BLOCK) {
+                       if (be32_to_cpu(jdb->h_sequence) !=
+                            be32_to_cpu(jsb->s_sequence)) {
+                               print_jrnl_status(recovery_flag);
+                               break;
+                       }
+
+                       if (transaction_state == TRANSACTION_RUNNING ||
+                                       (DB_FOUND == NO)) {
+                               transaction_state = TRANSACTION_COMPLETE;
+                               i++;
+                               jsb->s_sequence =
+                                       cpu_to_be32(be32_to_cpu(
+                                               jsb->s_sequence) + 1);
+                       }
+                       prev_desc_logical_no = curr_desc_logical_no;
+                       if ((recovery_flag == RECOVER) && (DB_FOUND == YES))
+                               recover_transaction(prev_desc_logical_no);
+
+                       DB_FOUND = NO;
+               } else if (be32_to_cpu(jdb->h_blocktype) ==
+                               EXT3_JOURNAL_REVOKE_BLOCK) {
+                       if (be32_to_cpu(jdb->h_sequence) !=
+                           be32_to_cpu(jsb->s_sequence)) {
+                               print_jrnl_status(recovery_flag);
+                               break;
+                       }
+                       if (recovery_flag == SCAN)
+                               ext4fs_push_revoke_blk((char *)jdb);
+                       i++;
+               } else {
+                       debug("Else Case\n");
+                       if (be32_to_cpu(jdb->h_sequence) !=
+                           be32_to_cpu(jsb->s_sequence)) {
+                               print_jrnl_status(recovery_flag);
+                               break;
+                       }
+               }
+       }
+
+end:
+       if (recovery_flag == RECOVER) {
+               jsb->s_start = cpu_to_be32(1);
+               jsb->s_sequence = cpu_to_be32(be32_to_cpu(jsb->s_sequence) + 1);
+               /* get the superblock */
+               ext4fs_devread(SUPERBLOCK_SECTOR, 0, SUPERBLOCK_SIZE,
+                              (char *)fs->sb);
+               fs->sb->feature_incompat |= EXT3_FEATURE_INCOMPAT_RECOVER;
+
+               /* Update the super block */
+               put_ext4((uint64_t) (SUPERBLOCK_SIZE),
+                        (struct ext2_sblock *)fs->sb,
+                        (uint32_t) SUPERBLOCK_SIZE);
+               ext4fs_devread(SUPERBLOCK_SECTOR, 0, SUPERBLOCK_SIZE,
+                              (char *)fs->sb);
+
+               blknr = read_allocated_block(&inode_journal,
+                                        EXT2_JOURNAL_SUPERBLOCK);
+               put_ext4((uint64_t) (blknr * fs->blksz),
+                        (struct journal_superblock_t *)temp_buff,
+                        (uint32_t) fs->blksz);
+               ext4fs_free_revoke_blks();
+       }
+       free(temp_buff);
+       free(temp_buff1);
+
+       return 0;
+}
+
+static void update_descriptor_block(long int blknr)
+{
+       int i;
+       long int jsb_blknr;
+       struct journal_header_t jdb;
+       struct ext3_journal_block_tag tag;
+       struct ext2_inode inode_journal;
+       struct journal_superblock_t *jsb = NULL;
+       char *buf = NULL;
+       char *temp = NULL;
+       struct ext_filesystem *fs = get_fs();
+       char *temp_buff = zalloc(fs->blksz);
+       if (!temp_buff)
+               return;
+
+       ext4fs_read_inode(ext4fs_root, EXT2_JOURNAL_INO, &inode_journal);
+       jsb_blknr = read_allocated_block(&inode_journal,
+                                        EXT2_JOURNAL_SUPERBLOCK);
+       ext4fs_devread(jsb_blknr * fs->sect_perblk, 0, fs->blksz, temp_buff);
+       jsb = (struct journal_superblock_t *) temp_buff;
+
+       jdb.h_blocktype = cpu_to_be32(EXT3_JOURNAL_DESCRIPTOR_BLOCK);
+       jdb.h_magic = cpu_to_be32(EXT3_JOURNAL_MAGIC_NUMBER);
+       jdb.h_sequence = jsb->s_sequence;
+       buf = zalloc(fs->blksz);
+       if (!buf) {
+               free(temp_buff);
+               return;
+       }
+       temp = buf;
+       memcpy(buf, &jdb, sizeof(struct journal_header_t));
+       temp += sizeof(struct journal_header_t);
+
+       for (i = 0; i < MAX_JOURNAL_ENTRIES; i++) {
+               if (journal_ptr[i]->blknr == -1)
+                       break;
+
+               tag.block = cpu_to_be32(journal_ptr[i]->blknr);
+               tag.flags = cpu_to_be32(EXT3_JOURNAL_FLAG_SAME_UUID);
+               memcpy(temp, &tag, sizeof(struct ext3_journal_block_tag));
+               temp = temp + sizeof(struct ext3_journal_block_tag);
+       }
+
+       tag.block = cpu_to_be32(journal_ptr[--i]->blknr);
+       tag.flags = cpu_to_be32(EXT3_JOURNAL_FLAG_LAST_TAG);
+       memcpy(temp - sizeof(struct ext3_journal_block_tag), &tag,
+              sizeof(struct ext3_journal_block_tag));
+       put_ext4((uint64_t) (blknr * fs->blksz), buf, (uint32_t) fs->blksz);
+
+       free(temp_buff);
+       free(buf);
+}
+
+static void update_commit_block(long int blknr)
+{
+       struct journal_header_t jdb;
+       struct ext_filesystem *fs = get_fs();
+       char *buf = NULL;
+       struct ext2_inode inode_journal;
+       struct journal_superblock_t *jsb;
+       long int jsb_blknr;
+       char *temp_buff = zalloc(fs->blksz);
+       if (!temp_buff)
+               return;
+
+       ext4fs_read_inode(ext4fs_root, EXT2_JOURNAL_INO, &inode_journal);
+       jsb_blknr = read_allocated_block(&inode_journal,
+                                        EXT2_JOURNAL_SUPERBLOCK);
+       ext4fs_devread(jsb_blknr * fs->sect_perblk, 0, fs->blksz, temp_buff);
+       jsb = (struct journal_superblock_t *) temp_buff;
+
+       jdb.h_blocktype = cpu_to_be32(EXT3_JOURNAL_COMMIT_BLOCK);
+       jdb.h_magic = cpu_to_be32(EXT3_JOURNAL_MAGIC_NUMBER);
+       jdb.h_sequence = jsb->s_sequence;
+       buf = zalloc(fs->blksz);
+       if (!buf) {
+               free(temp_buff);
+               return;
+       }
+       memcpy(buf, &jdb, sizeof(struct journal_header_t));
+       put_ext4((uint64_t) (blknr * fs->blksz), buf, (uint32_t) fs->blksz);
+
+       free(temp_buff);
+       free(buf);
+}
+
+void ext4fs_update_journal(void)
+{
+       struct ext2_inode inode_journal;
+       struct ext_filesystem *fs = get_fs();
+       long int blknr;
+       int i;
+       ext4fs_read_inode(ext4fs_root, EXT2_JOURNAL_INO, &inode_journal);
+       blknr = read_allocated_block(&inode_journal, jrnl_blk_idx++);
+       update_descriptor_block(blknr);
+       for (i = 0; i < MAX_JOURNAL_ENTRIES; i++) {
+               if (journal_ptr[i]->blknr == -1)
+                       break;
+               blknr = read_allocated_block(&inode_journal, jrnl_blk_idx++);
+               put_ext4((uint64_t) ((uint64_t)blknr * (uint64_t)fs->blksz),
+                        journal_ptr[i]->buf, fs->blksz);
+       }
+       blknr = read_allocated_block(&inode_journal, jrnl_blk_idx++);
+       update_commit_block(blknr);
+       printf("update journal finished\n");
+}
diff --git a/fs/ext4/ext4_journal.h b/fs/ext4/ext4_journal.h
new file mode 100644 (file)
index 0000000..acc1c51
--- /dev/null
@@ -0,0 +1,141 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * Journal data structures and headers for Journaling feature of ext4
+ * have been referred from JBD2 (Journaling Block device 2)
+ * implementation in Linux Kernel.
+ *
+ * Written by Stephen C. Tweedie <sct@redhat.com>
+ *
+ * Copyright 1998-2000 Red Hat, Inc --- 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.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef __EXT4_JRNL__
+#define __EXT4_JRNL__
+
+#define EXT2_JOURNAL_INO               8       /* Journal inode */
+#define EXT2_JOURNAL_SUPERBLOCK        0       /* Journal  Superblock number */
+
+#define JBD2_FEATURE_COMPAT_CHECKSUM   0x00000001
+#define EXT3_JOURNAL_MAGIC_NUMBER      0xc03b3998U
+#define TRANSACTION_RUNNING            1
+#define TRANSACTION_COMPLETE           0
+#define EXT3_FEATURE_INCOMPAT_RECOVER  0x0004  /* Needs recovery */
+#define EXT3_JOURNAL_DESCRIPTOR_BLOCK  1
+#define EXT3_JOURNAL_COMMIT_BLOCK      2
+#define EXT3_JOURNAL_SUPERBLOCK_V1     3
+#define EXT3_JOURNAL_SUPERBLOCK_V2     4
+#define EXT3_JOURNAL_REVOKE_BLOCK      5
+#define EXT3_JOURNAL_FLAG_ESCAPE       1
+#define EXT3_JOURNAL_FLAG_SAME_UUID    2
+#define EXT3_JOURNAL_FLAG_DELETED      4
+#define EXT3_JOURNAL_FLAG_LAST_TAG     8
+
+/* Maximum entries in 1 journal transaction */
+#define MAX_JOURNAL_ENTRIES 100
+struct journal_log {
+       char *buf;
+       int blknr;
+};
+
+struct dirty_blocks {
+       char *buf;
+       int blknr;
+};
+
+/* Standard header for all descriptor blocks: */
+struct journal_header_t {
+       __u32 h_magic;
+       __u32 h_blocktype;
+       __u32 h_sequence;
+};
+
+/* The journal superblock.  All fields are in big-endian byte order. */
+struct journal_superblock_t {
+       /* 0x0000 */
+       struct journal_header_t s_header;
+
+       /* Static information describing the journal */
+       __u32 s_blocksize;      /* journal device blocksize */
+       __u32 s_maxlen;         /* total blocks in journal file */
+       __u32 s_first;          /* first block of log information */
+
+       /* Dynamic information describing the current state of the log */
+       __u32 s_sequence;       /* first commit ID expected in log */
+       __u32 s_start;          /* blocknr of start of log */
+
+       /* Error value, as set by journal_abort(). */
+       __s32 s_errno;
+
+       /* Remaining fields are only valid in a version-2 superblock */
+       __u32 s_feature_compat; /* compatible feature set */
+       __u32 s_feature_incompat;       /* incompatible feature set */
+       __u32 s_feature_ro_compat;      /* readonly-compatible feature set */
+       /* 0x0030 */
+       __u8 s_uuid[16];        /* 128-bit uuid for journal */
+
+       /* 0x0040 */
+       __u32 s_nr_users;       /* Nr of filesystems sharing log */
+
+       __u32 s_dynsuper;       /* Blocknr of dynamic superblock copy */
+
+       /* 0x0048 */
+       __u32 s_max_transaction;        /* Limit of journal blocks per trans. */
+       __u32 s_max_trans_data; /* Limit of data blocks per trans. */
+
+       /* 0x0050 */
+       __u32 s_padding[44];
+
+       /* 0x0100 */
+       __u8 s_users[16 * 48];  /* ids of all fs'es sharing the log */
+       /* 0x0400 */
+} ;
+
+struct ext3_journal_block_tag {
+       uint32_t block;
+       uint32_t flags;
+};
+
+struct journal_revoke_header_t {
+       struct journal_header_t r_header;
+       int r_count;            /* Count of bytes used in the block */
+};
+
+struct revoke_blk_list {
+       char *content;          /* revoke block itself */
+       struct revoke_blk_list *next;
+};
+
+extern struct ext2_data *ext4fs_root;
+
+int ext4fs_init_journal(void);
+int ext4fs_log_gdt(char *gd_table);
+int ext4fs_check_journal_state(int recovery_flag);
+int ext4fs_log_journal(char *journal_buffer, long int blknr);
+int ext4fs_put_metadata(char *metadata_buffer, long int blknr);
+void ext4fs_update_journal(void);
+void ext4fs_dump_metadata(void);
+void ext4fs_push_revoke_blk(char *buffer);
+void ext4fs_free_journal(void);
+void ext4fs_free_revoke_blks(void);
+#endif
diff --git a/fs/ext4/ext4fs.c b/fs/ext4/ext4fs.c
new file mode 100644 (file)
index 0000000..114c2a2
--- /dev/null
@@ -0,0 +1,1189 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * ext4ls and ext4load : Based on ext2 ls and load support in Uboot.
+ *                    Ext4 read optimization taken from Open-Moko
+ *                    Qi bootloader
+ *
+ * (C) Copyright 2004
+ * esd gmbh <www.esd-electronics.com>
+ * Reinhard Arlt <reinhard.arlt@esd-electronics.com>
+ *
+ * based on code from grub2 fs/ext2.c and fs/fshelp.c by
+ * GRUB  --  GRand Unified Bootloader
+ * Copyright (C) 2003, 2004  Free Software Foundation, Inc.
+ *
+ * ext4write : Based on generic ext4 protocol.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <common.h>
+#include <malloc.h>
+#include <ext_common.h>
+#include <ext4fs.h>
+#include <linux/stat.h>
+#include <linux/time.h>
+#include <asm/byteorder.h>
+#include "ext4_common.h"
+
+int ext4fs_symlinknest;
+block_dev_desc_t *ext4_dev_desc;
+
+struct ext_filesystem *get_fs(void)
+{
+       if (ext4_dev_desc == NULL || ext4_dev_desc->priv == NULL)
+               printf("Invalid Input Arguments %s\n", __func__);
+
+       return ext4_dev_desc->priv;
+}
+
+int init_fs(block_dev_desc_t *dev_desc)
+{
+       struct ext_filesystem *fs;
+       if (dev_desc == NULL) {
+               printf("Invalid Input Arguments %s\n", __func__);
+               return -EINVAL;
+       }
+
+       fs = zalloc(sizeof(struct ext_filesystem));
+       if (fs == NULL) {
+               printf("malloc failed: %s\n", __func__);
+               return -ENOMEM;
+       }
+
+       fs->dev_desc = dev_desc;
+       dev_desc->priv = fs;
+
+       return 0;
+}
+
+void deinit_fs(block_dev_desc_t *dev_desc)
+{
+       if (dev_desc == NULL) {
+               printf("Invalid Input Arguments %s\n", __func__);
+               return;
+       }
+       free(dev_desc->priv);
+       dev_desc->priv = NULL;
+}
+
+void ext4fs_free_node(struct ext2fs_node *node, struct ext2fs_node *currroot)
+{
+       if ((node != &ext4fs_root->diropen) && (node != currroot))
+               free(node);
+}
+
+/*
+ * Taken from openmoko-kernel mailing list: By Andy green
+ * Optimized read file API : collects and defers contiguous sector
+ * reads into one potentially more efficient larger sequential read action
+ */
+int ext4fs_read_file(struct ext2fs_node *node, int pos,
+               unsigned int len, char *buf)
+{
+       int i;
+       int blockcnt;
+       int log2blocksize = LOG2_EXT2_BLOCK_SIZE(node->data);
+       int blocksize = 1 << (log2blocksize + DISK_SECTOR_BITS);
+       unsigned int filesize = __le32_to_cpu(node->inode.size);
+       int previous_block_number = -1;
+       int delayed_start = 0;
+       int delayed_extent = 0;
+       int delayed_skipfirst = 0;
+       int delayed_next = 0;
+       char *delayed_buf = NULL;
+       short status;
+
+       /* Adjust len so it we can't read past the end of the file. */
+       if (len > filesize)
+               len = filesize;
+
+       blockcnt = ((len + pos) + blocksize - 1) / blocksize;
+
+       for (i = pos / blocksize; i < blockcnt; i++) {
+               int blknr;
+               int blockoff = pos % blocksize;
+               int blockend = blocksize;
+               int skipfirst = 0;
+               blknr = read_allocated_block(&(node->inode), i);
+               if (blknr < 0)
+                       return -1;
+
+               blknr = blknr << log2blocksize;
+
+               /* Last block.  */
+               if (i == blockcnt - 1) {
+                       blockend = (len + pos) % blocksize;
+
+                       /* The last portion is exactly blocksize. */
+                       if (!blockend)
+                               blockend = blocksize;
+               }
+
+               /* First block. */
+               if (i == pos / blocksize) {
+                       skipfirst = blockoff;
+                       blockend -= skipfirst;
+               }
+               if (blknr) {
+                       int status;
+
+                       if (previous_block_number != -1) {
+                               if (delayed_next == blknr) {
+                                       delayed_extent += blockend;
+                                       delayed_next += blockend >> SECTOR_BITS;
+                               } else {        /* spill */
+                                       status = ext4fs_devread(delayed_start,
+                                                       delayed_skipfirst,
+                                                       delayed_extent,
+                                                       delayed_buf);
+                                       if (status == 0)
+                                               return -1;
+                                       previous_block_number = blknr;
+                                       delayed_start = blknr;
+                                       delayed_extent = blockend;
+                                       delayed_skipfirst = skipfirst;
+                                       delayed_buf = buf;
+                                       delayed_next = blknr +
+                                               (blockend >> SECTOR_BITS);
+                               }
+                       } else {
+                               previous_block_number = blknr;
+                               delayed_start = blknr;
+                               delayed_extent = blockend;
+                               delayed_skipfirst = skipfirst;
+                               delayed_buf = buf;
+                               delayed_next = blknr +
+                                       (blockend >> SECTOR_BITS);
+                       }
+               } else {
+                       if (previous_block_number != -1) {
+                               /* spill */
+                               status = ext4fs_devread(delayed_start,
+                                                       delayed_skipfirst,
+                                                       delayed_extent,
+                                                       delayed_buf);
+                               if (status == 0)
+                                       return -1;
+                               previous_block_number = -1;
+                       }
+                       memset(buf, 0, blocksize - skipfirst);
+               }
+               buf += blocksize - skipfirst;
+       }
+       if (previous_block_number != -1) {
+               /* spill */
+               status = ext4fs_devread(delayed_start,
+                                       delayed_skipfirst, delayed_extent,
+                                       delayed_buf);
+               if (status == 0)
+                       return -1;
+               previous_block_number = -1;
+       }
+
+       return len;
+}
+
+int ext4fs_ls(const char *dirname)
+{
+       struct ext2fs_node *dirnode;
+       int status;
+
+       if (dirname == NULL)
+               return 0;
+
+       status = ext4fs_find_file(dirname, &ext4fs_root->diropen, &dirnode,
+                                 FILETYPE_DIRECTORY);
+       if (status != 1) {
+               printf("** Can not find directory. **\n");
+               return 1;
+       }
+
+       ext4fs_iterate_dir(dirnode, NULL, NULL, NULL);
+       ext4fs_free_node(dirnode, &ext4fs_root->diropen);
+
+       return 0;
+}
+
+int ext4fs_read(char *buf, unsigned len)
+{
+       if (ext4fs_root == NULL || ext4fs_file == NULL)
+               return 0;
+
+       return ext4fs_read_file(ext4fs_file, 0, len, buf);
+}
+
+#if defined(CONFIG_CMD_EXT4_WRITE)
+static void ext4fs_update(void)
+{
+       short i;
+       ext4fs_update_journal();
+       struct ext_filesystem *fs = get_fs();
+
+       /* update  super block */
+       put_ext4((uint64_t)(SUPERBLOCK_SIZE),
+                (struct ext2_sblock *)fs->sb, (uint32_t)SUPERBLOCK_SIZE);
+
+       /* update block groups */
+       for (i = 0; i < fs->no_blkgrp; i++) {
+               fs->gd[i].bg_checksum = ext4fs_checksum_update(i);
+               put_ext4((uint64_t)(fs->gd[i].block_id * fs->blksz),
+                        fs->blk_bmaps[i], fs->blksz);
+       }
+
+       /* update inode table groups */
+       for (i = 0; i < fs->no_blkgrp; i++) {
+               put_ext4((uint64_t) (fs->gd[i].inode_id * fs->blksz),
+                        fs->inode_bmaps[i], fs->blksz);
+       }
+
+       /* update the block group descriptor table */
+       put_ext4((uint64_t)(fs->gdtable_blkno * fs->blksz),
+                (struct ext2_block_group *)fs->gdtable,
+                (fs->blksz * fs->no_blk_pergdt));
+
+       ext4fs_dump_metadata();
+
+       gindex = 0;
+       gd_index = 0;
+}
+
+int ext4fs_get_bgdtable(void)
+{
+       int status;
+       int grp_desc_size;
+       struct ext_filesystem *fs = get_fs();
+       grp_desc_size = sizeof(struct ext2_block_group);
+       fs->no_blk_pergdt = (fs->no_blkgrp * grp_desc_size) / fs->blksz;
+       if ((fs->no_blkgrp * grp_desc_size) % fs->blksz)
+               fs->no_blk_pergdt++;
+
+       /* allocate memory for gdtable */
+       fs->gdtable = zalloc(fs->blksz * fs->no_blk_pergdt);
+       if (!fs->gdtable)
+               return -ENOMEM;
+       /* read the group descriptor table */
+       status = ext4fs_devread(fs->gdtable_blkno * fs->sect_perblk, 0,
+                               fs->blksz * fs->no_blk_pergdt, fs->gdtable);
+       if (status == 0)
+               goto fail;
+
+       if (ext4fs_log_gdt(fs->gdtable)) {
+               printf("Error in ext4fs_log_gdt\n");
+               return -1;
+       }
+
+       return 0;
+fail:
+       free(fs->gdtable);
+       fs->gdtable = NULL;
+
+       return -1;
+}
+
+static void delete_single_indirect_block(struct ext2_inode *inode)
+{
+       struct ext2_block_group *gd = NULL;
+       static int prev_bg_bmap_idx = -1;
+       long int blknr;
+       int remainder;
+       int bg_idx;
+       int status;
+       unsigned int blk_per_grp = ext4fs_root->sblock.blocks_per_group;
+       struct ext_filesystem *fs = get_fs();
+       char *journal_buffer = zalloc(fs->blksz);
+       if (!journal_buffer) {
+               printf("No memory\n");
+               return;
+       }
+       /* get  block group descriptor table */
+       gd = (struct ext2_block_group *)fs->gdtable;
+
+       /* deleting the single indirect block associated with inode */
+       if (inode->b.blocks.indir_block != 0) {
+               debug("SIPB releasing %u\n", inode->b.blocks.indir_block);
+               blknr = inode->b.blocks.indir_block;
+               if (fs->blksz != 1024) {
+                       bg_idx = blknr / blk_per_grp;
+               } else {
+                       bg_idx = blknr / blk_per_grp;
+                       remainder = blknr % blk_per_grp;
+                       if (!remainder)
+                               bg_idx--;
+               }
+               ext4fs_reset_block_bmap(blknr, fs->blk_bmaps[bg_idx], bg_idx);
+               gd[bg_idx].free_blocks++;
+               fs->sb->free_blocks++;
+               /* journal backup */
+               if (prev_bg_bmap_idx != bg_idx) {
+                       status =
+                           ext4fs_devread(gd[bg_idx].block_id *
+                                          fs->sect_perblk, 0, fs->blksz,
+                                          journal_buffer);
+                       if (status == 0)
+                               goto fail;
+                       if (ext4fs_log_journal
+                           (journal_buffer, gd[bg_idx].block_id))
+                               goto fail;
+                       prev_bg_bmap_idx = bg_idx;
+               }
+       }
+fail:
+       free(journal_buffer);
+}
+
+static void delete_double_indirect_block(struct ext2_inode *inode)
+{
+       int i;
+       short status;
+       static int prev_bg_bmap_idx = -1;
+       long int blknr;
+       int remainder;
+       int bg_idx;
+       unsigned int blk_per_grp = ext4fs_root->sblock.blocks_per_group;
+       unsigned int *di_buffer = NULL;
+       unsigned int *DIB_start_addr = NULL;
+       struct ext2_block_group *gd = NULL;
+       struct ext_filesystem *fs = get_fs();
+       char *journal_buffer = zalloc(fs->blksz);
+       if (!journal_buffer) {
+               printf("No memory\n");
+               return;
+       }
+       /* get the block group descriptor table */
+       gd = (struct ext2_block_group *)fs->gdtable;
+
+       if (inode->b.blocks.double_indir_block != 0) {
+               di_buffer = zalloc(fs->blksz);
+               if (!di_buffer) {
+                       printf("No memory\n");
+                       return;
+               }
+               DIB_start_addr = (unsigned int *)di_buffer;
+               blknr = inode->b.blocks.double_indir_block;
+               status = ext4fs_devread(blknr * fs->sect_perblk, 0, fs->blksz,
+                                       (char *)di_buffer);
+               for (i = 0; i < fs->blksz / sizeof(int); i++) {
+                       if (*di_buffer == 0)
+                               break;
+
+                       debug("DICB releasing %u\n", *di_buffer);
+                       if (fs->blksz != 1024) {
+                               bg_idx = (*di_buffer) / blk_per_grp;
+                       } else {
+                               bg_idx = (*di_buffer) / blk_per_grp;
+                               remainder = (*di_buffer) % blk_per_grp;
+                               if (!remainder)
+                                       bg_idx--;
+                       }
+                       ext4fs_reset_block_bmap(*di_buffer,
+                                       fs->blk_bmaps[bg_idx], bg_idx);
+                       di_buffer++;
+                       gd[bg_idx].free_blocks++;
+                       fs->sb->free_blocks++;
+                       /* journal backup */
+                       if (prev_bg_bmap_idx != bg_idx) {
+                               status = ext4fs_devread(gd[bg_idx].block_id
+                                                       * fs->sect_perblk, 0,
+                                                       fs->blksz,
+                                                       journal_buffer);
+                               if (status == 0)
+                                       goto fail;
+
+                               if (ext4fs_log_journal(journal_buffer,
+                                                       gd[bg_idx].block_id))
+                                       goto fail;
+                               prev_bg_bmap_idx = bg_idx;
+                       }
+               }
+
+               /* removing the parent double indirect block */
+               blknr = inode->b.blocks.double_indir_block;
+               if (fs->blksz != 1024) {
+                       bg_idx = blknr / blk_per_grp;
+               } else {
+                       bg_idx = blknr / blk_per_grp;
+                       remainder = blknr % blk_per_grp;
+                       if (!remainder)
+                               bg_idx--;
+               }
+               ext4fs_reset_block_bmap(blknr, fs->blk_bmaps[bg_idx], bg_idx);
+               gd[bg_idx].free_blocks++;
+               fs->sb->free_blocks++;
+               /* journal backup */
+               if (prev_bg_bmap_idx != bg_idx) {
+                       memset(journal_buffer, '\0', fs->blksz);
+                       status = ext4fs_devread(gd[bg_idx].block_id *
+                                               fs->sect_perblk, 0, fs->blksz,
+                                               journal_buffer);
+                       if (status == 0)
+                               goto fail;
+
+                       if (ext4fs_log_journal(journal_buffer,
+                                               gd[bg_idx].block_id))
+                               goto fail;
+                       prev_bg_bmap_idx = bg_idx;
+               }
+               debug("DIPB releasing %ld\n", blknr);
+       }
+fail:
+       free(DIB_start_addr);
+       free(journal_buffer);
+}
+
+static void delete_triple_indirect_block(struct ext2_inode *inode)
+{
+       int i, j;
+       short status;
+       static int prev_bg_bmap_idx = -1;
+       long int blknr;
+       int remainder;
+       int bg_idx;
+       unsigned int blk_per_grp = ext4fs_root->sblock.blocks_per_group;
+       unsigned int *tigp_buffer = NULL;
+       unsigned int *tib_start_addr = NULL;
+       unsigned int *tip_buffer = NULL;
+       unsigned int *tipb_start_addr = NULL;
+       struct ext2_block_group *gd = NULL;
+       struct ext_filesystem *fs = get_fs();
+       char *journal_buffer = zalloc(fs->blksz);
+       if (!journal_buffer) {
+               printf("No memory\n");
+               return;
+       }
+       /* get block group descriptor table */
+       gd = (struct ext2_block_group *)fs->gdtable;
+
+       if (inode->b.blocks.triple_indir_block != 0) {
+               tigp_buffer = zalloc(fs->blksz);
+               if (!tigp_buffer) {
+                       printf("No memory\n");
+                       return;
+               }
+               tib_start_addr = (unsigned int *)tigp_buffer;
+               blknr = inode->b.blocks.triple_indir_block;
+               status = ext4fs_devread(blknr * fs->sect_perblk, 0, fs->blksz,
+                                       (char *)tigp_buffer);
+               for (i = 0; i < fs->blksz / sizeof(int); i++) {
+                       if (*tigp_buffer == 0)
+                               break;
+                       debug("tigp buffer releasing %u\n", *tigp_buffer);
+
+                       tip_buffer = zalloc(fs->blksz);
+                       if (!tip_buffer)
+                               goto fail;
+                       tipb_start_addr = (unsigned int *)tip_buffer;
+                       status = ext4fs_devread((*tigp_buffer) *
+                                               fs->sect_perblk, 0, fs->blksz,
+                                               (char *)tip_buffer);
+                       for (j = 0; j < fs->blksz / sizeof(int); j++) {
+                               if (*tip_buffer == 0)
+                                       break;
+                               if (fs->blksz != 1024) {
+                                       bg_idx = (*tip_buffer) / blk_per_grp;
+                               } else {
+                                       bg_idx = (*tip_buffer) / blk_per_grp;
+
+                                       remainder = (*tip_buffer) % blk_per_grp;
+                                       if (!remainder)
+                                               bg_idx--;
+                               }
+
+                               ext4fs_reset_block_bmap(*tip_buffer,
+                                                       fs->blk_bmaps[bg_idx],
+                                                       bg_idx);
+
+                               tip_buffer++;
+                               gd[bg_idx].free_blocks++;
+                               fs->sb->free_blocks++;
+                               /* journal backup */
+                               if (prev_bg_bmap_idx != bg_idx) {
+                                       status =
+                                           ext4fs_devread(gd[bg_idx].block_id *
+                                                          fs->sect_perblk, 0,
+                                                          fs->blksz,
+                                                          journal_buffer);
+                                       if (status == 0)
+                                               goto fail;
+
+                                       if (ext4fs_log_journal(journal_buffer,
+                                                              gd[bg_idx].
+                                                              block_id))
+                                               goto fail;
+                                       prev_bg_bmap_idx = bg_idx;
+                               }
+                       }
+                       free(tipb_start_addr);
+                       tipb_start_addr = NULL;
+
+                       /*
+                        * removing the grand parent blocks
+                        * which is connected to inode
+                        */
+                       if (fs->blksz != 1024) {
+                               bg_idx = (*tigp_buffer) / blk_per_grp;
+                       } else {
+                               bg_idx = (*tigp_buffer) / blk_per_grp;
+
+                               remainder = (*tigp_buffer) % blk_per_grp;
+                               if (!remainder)
+                                       bg_idx--;
+                       }
+                       ext4fs_reset_block_bmap(*tigp_buffer,
+                                               fs->blk_bmaps[bg_idx], bg_idx);
+
+                       tigp_buffer++;
+                       gd[bg_idx].free_blocks++;
+                       fs->sb->free_blocks++;
+                       /* journal backup */
+                       if (prev_bg_bmap_idx != bg_idx) {
+                               memset(journal_buffer, '\0', fs->blksz);
+                               status =
+                                   ext4fs_devread(gd[bg_idx].block_id *
+                                                  fs->sect_perblk, 0,
+                                                  fs->blksz, journal_buffer);
+                               if (status == 0)
+                                       goto fail;
+
+                               if (ext4fs_log_journal(journal_buffer,
+                                                       gd[bg_idx].block_id))
+                                       goto fail;
+                               prev_bg_bmap_idx = bg_idx;
+                       }
+               }
+
+               /* removing the grand parent triple indirect block */
+               blknr = inode->b.blocks.triple_indir_block;
+               if (fs->blksz != 1024) {
+                       bg_idx = blknr / blk_per_grp;
+               } else {
+                       bg_idx = blknr / blk_per_grp;
+                       remainder = blknr % blk_per_grp;
+                       if (!remainder)
+                               bg_idx--;
+               }
+               ext4fs_reset_block_bmap(blknr, fs->blk_bmaps[bg_idx], bg_idx);
+               gd[bg_idx].free_blocks++;
+               fs->sb->free_blocks++;
+               /* journal backup */
+               if (prev_bg_bmap_idx != bg_idx) {
+                       memset(journal_buffer, '\0', fs->blksz);
+                       status = ext4fs_devread(gd[bg_idx].block_id *
+                                               fs->sect_perblk, 0, fs->blksz,
+                                               journal_buffer);
+                       if (status == 0)
+                               goto fail;
+
+                       if (ext4fs_log_journal(journal_buffer,
+                                               gd[bg_idx].block_id))
+                               goto fail;
+                       prev_bg_bmap_idx = bg_idx;
+               }
+               debug("tigp buffer itself releasing %ld\n", blknr);
+       }
+fail:
+       free(tib_start_addr);
+       free(tipb_start_addr);
+       free(journal_buffer);
+}
+
+static int ext4fs_delete_file(int inodeno)
+{
+       struct ext2_inode inode;
+       short status;
+       int i;
+       int remainder;
+       long int blknr;
+       int bg_idx;
+       int ibmap_idx;
+       char *read_buffer = NULL;
+       char *start_block_address = NULL;
+       unsigned int no_blocks;
+
+       static int prev_bg_bmap_idx = -1;
+       unsigned int inodes_per_block;
+       long int blkno;
+       unsigned int blkoff;
+       unsigned int blk_per_grp = ext4fs_root->sblock.blocks_per_group;
+       unsigned int inode_per_grp = ext4fs_root->sblock.inodes_per_group;
+       struct ext2_inode *inode_buffer = NULL;
+       struct ext2_block_group *gd = NULL;
+       struct ext_filesystem *fs = get_fs();
+       char *journal_buffer = zalloc(fs->blksz);
+       if (!journal_buffer)
+               return -ENOMEM;
+       /* get the block group descriptor table */
+       gd = (struct ext2_block_group *)fs->gdtable;
+       status = ext4fs_read_inode(ext4fs_root, inodeno, &inode);
+       if (status == 0)
+               goto fail;
+
+       /* read the block no allocated to a file */
+       no_blocks = inode.size / fs->blksz;
+       if (inode.size % fs->blksz)
+               no_blocks++;
+
+       if (le32_to_cpu(inode.flags) & EXT4_EXTENTS_FL) {
+               struct ext2fs_node *node_inode =
+                   zalloc(sizeof(struct ext2fs_node));
+               if (!node_inode)
+                       goto fail;
+               node_inode->data = ext4fs_root;
+               node_inode->ino = inodeno;
+               node_inode->inode_read = 0;
+               memcpy(&(node_inode->inode), &inode, sizeof(struct ext2_inode));
+
+               for (i = 0; i < no_blocks; i++) {
+                       blknr = read_allocated_block(&(node_inode->inode), i);
+                       if (fs->blksz != 1024) {
+                               bg_idx = blknr / blk_per_grp;
+                       } else {
+                               bg_idx = blknr / blk_per_grp;
+                               remainder = blknr % blk_per_grp;
+                               if (!remainder)
+                                       bg_idx--;
+                       }
+                       ext4fs_reset_block_bmap(blknr, fs->blk_bmaps[bg_idx],
+                                               bg_idx);
+                       debug("EXT4_EXTENTS Block releasing %ld: %d\n",
+                             blknr, bg_idx);
+
+                       gd[bg_idx].free_blocks++;
+                       fs->sb->free_blocks++;
+
+                       /* journal backup */
+                       if (prev_bg_bmap_idx != bg_idx) {
+                               status =
+                                   ext4fs_devread(gd[bg_idx].block_id *
+                                                  fs->sect_perblk, 0,
+                                                  fs->blksz, journal_buffer);
+                               if (status == 0)
+                                       goto fail;
+                               if (ext4fs_log_journal(journal_buffer,
+                                                       gd[bg_idx].block_id))
+                                       goto fail;
+                               prev_bg_bmap_idx = bg_idx;
+                       }
+               }
+               if (node_inode) {
+                       free(node_inode);
+                       node_inode = NULL;
+               }
+       } else {
+
+               delete_single_indirect_block(&inode);
+               delete_double_indirect_block(&inode);
+               delete_triple_indirect_block(&inode);
+
+               /* read the block no allocated to a file */
+               no_blocks = inode.size / fs->blksz;
+               if (inode.size % fs->blksz)
+                       no_blocks++;
+               for (i = 0; i < no_blocks; i++) {
+                       blknr = read_allocated_block(&inode, i);
+                       if (fs->blksz != 1024) {
+                               bg_idx = blknr / blk_per_grp;
+                       } else {
+                               bg_idx = blknr / blk_per_grp;
+                               remainder = blknr % blk_per_grp;
+                               if (!remainder)
+                                       bg_idx--;
+                       }
+                       ext4fs_reset_block_bmap(blknr, fs->blk_bmaps[bg_idx],
+                                               bg_idx);
+                       debug("ActualB releasing %ld: %d\n", blknr, bg_idx);
+
+                       gd[bg_idx].free_blocks++;
+                       fs->sb->free_blocks++;
+                       /* journal backup */
+                       if (prev_bg_bmap_idx != bg_idx) {
+                               memset(journal_buffer, '\0', fs->blksz);
+                               status = ext4fs_devread(gd[bg_idx].block_id
+                                                       * fs->sect_perblk,
+                                                       0, fs->blksz,
+                                                       journal_buffer);
+                               if (status == 0)
+                                       goto fail;
+                               if (ext4fs_log_journal(journal_buffer,
+                                               gd[bg_idx].block_id))
+                                       goto fail;
+                               prev_bg_bmap_idx = bg_idx;
+                       }
+               }
+       }
+
+       /* from the inode no to blockno */
+       inodes_per_block = fs->blksz / fs->inodesz;
+       ibmap_idx = inodeno / inode_per_grp;
+
+       /* get the block no */
+       inodeno--;
+       blkno = __le32_to_cpu(gd[ibmap_idx].inode_table_id) +
+               (inodeno % __le32_to_cpu(inode_per_grp)) / inodes_per_block;
+
+       /* get the offset of the inode */
+       blkoff = ((inodeno) % inodes_per_block) * fs->inodesz;
+
+       /* read the block no containing the inode */
+       read_buffer = zalloc(fs->blksz);
+       if (!read_buffer)
+               goto fail;
+       start_block_address = read_buffer;
+       status = ext4fs_devread(blkno * fs->sect_perblk,
+                               0, fs->blksz, read_buffer);
+       if (status == 0)
+               goto fail;
+
+       if (ext4fs_log_journal(read_buffer, blkno))
+               goto fail;
+
+       read_buffer = read_buffer + blkoff;
+       inode_buffer = (struct ext2_inode *)read_buffer;
+       memset(inode_buffer, '\0', sizeof(struct ext2_inode));
+
+       /* write the inode to original position in inode table */
+       if (ext4fs_put_metadata(start_block_address, blkno))
+               goto fail;
+
+       /* update the respective inode bitmaps */
+       inodeno++;
+       ext4fs_reset_inode_bmap(inodeno, fs->inode_bmaps[ibmap_idx], ibmap_idx);
+       gd[ibmap_idx].free_inodes++;
+       fs->sb->free_inodes++;
+       /* journal backup */
+       memset(journal_buffer, '\0', fs->blksz);
+       status = ext4fs_devread(gd[ibmap_idx].inode_id *
+                               fs->sect_perblk, 0, fs->blksz, journal_buffer);
+       if (status == 0)
+               goto fail;
+       if (ext4fs_log_journal(journal_buffer, gd[ibmap_idx].inode_id))
+               goto fail;
+
+       ext4fs_update();
+       ext4fs_deinit();
+
+       if (ext4fs_init() != 0) {
+               printf("error in File System init\n");
+               goto fail;
+       }
+
+       free(start_block_address);
+       free(journal_buffer);
+
+       return 0;
+fail:
+       free(start_block_address);
+       free(journal_buffer);
+
+       return -1;
+}
+
+int ext4fs_init(void)
+{
+       short status;
+       int i;
+       unsigned int real_free_blocks = 0;
+       struct ext_filesystem *fs = get_fs();
+
+       /* populate fs */
+       fs->blksz = EXT2_BLOCK_SIZE(ext4fs_root);
+       fs->inodesz = INODE_SIZE_FILESYSTEM(ext4fs_root);
+       fs->sect_perblk = fs->blksz / SECTOR_SIZE;
+
+       /* get the superblock */
+       fs->sb = zalloc(SUPERBLOCK_SIZE);
+       if (!fs->sb)
+               return -ENOMEM;
+       if (!ext4fs_devread(SUPERBLOCK_SECTOR, 0, SUPERBLOCK_SIZE,
+                       (char *)fs->sb))
+               goto fail;
+
+       /* init journal */
+       if (ext4fs_init_journal())
+               goto fail;
+
+       /* get total no of blockgroups */
+       fs->no_blkgrp = (uint32_t)ext4fs_div_roundup(
+                       (ext4fs_root->sblock.total_blocks -
+                       ext4fs_root->sblock.first_data_block),
+                       ext4fs_root->sblock.blocks_per_group);
+
+       /* get the block group descriptor table */
+       fs->gdtable_blkno = ((EXT2_MIN_BLOCK_SIZE == fs->blksz) + 1);
+       if (ext4fs_get_bgdtable() == -1) {
+               printf("Error in getting the block group descriptor table\n");
+               goto fail;
+       }
+       fs->gd = (struct ext2_block_group *)fs->gdtable;
+
+       /* load all the available bitmap block of the partition */
+       fs->blk_bmaps = zalloc(fs->no_blkgrp * sizeof(char *));
+       if (!fs->blk_bmaps)
+               goto fail;
+       for (i = 0; i < fs->no_blkgrp; i++) {
+               fs->blk_bmaps[i] = zalloc(fs->blksz);
+               if (!fs->blk_bmaps[i])
+                       goto fail;
+       }
+
+       for (i = 0; i < fs->no_blkgrp; i++) {
+               status =
+                   ext4fs_devread(fs->gd[i].block_id * fs->sect_perblk, 0,
+                                  fs->blksz, (char *)fs->blk_bmaps[i]);
+               if (status == 0)
+                       goto fail;
+       }
+
+       /* load all the available inode bitmap of the partition */
+       fs->inode_bmaps = zalloc(fs->no_blkgrp * sizeof(unsigned char *));
+       if (!fs->inode_bmaps)
+               goto fail;
+       for (i = 0; i < fs->no_blkgrp; i++) {
+               fs->inode_bmaps[i] = zalloc(fs->blksz);
+               if (!fs->inode_bmaps[i])
+                       goto fail;
+       }
+
+       for (i = 0; i < fs->no_blkgrp; i++) {
+               status = ext4fs_devread(fs->gd[i].inode_id * fs->sect_perblk,
+                                       0, fs->blksz,
+                                       (char *)fs->inode_bmaps[i]);
+               if (status == 0)
+                       goto fail;
+       }
+
+       /*
+        * check filesystem consistency with free blocks of file system
+        * some time we observed that superblock freeblocks does not match
+        * with the  blockgroups freeblocks when improper
+        * reboot of a linux kernel
+        */
+       for (i = 0; i < fs->no_blkgrp; i++)
+               real_free_blocks = real_free_blocks + fs->gd[i].free_blocks;
+       if (real_free_blocks != fs->sb->free_blocks)
+               fs->sb->free_blocks = real_free_blocks;
+
+       return 0;
+fail:
+       ext4fs_deinit();
+
+       return -1;
+}
+
+void ext4fs_deinit(void)
+{
+       int i;
+       struct ext2_inode inode_journal;
+       struct journal_superblock_t *jsb;
+       long int blknr;
+       struct ext_filesystem *fs = get_fs();
+
+       /* free journal */
+       char *temp_buff = zalloc(fs->blksz);
+       if (temp_buff) {
+               ext4fs_read_inode(ext4fs_root, EXT2_JOURNAL_INO,
+                                 &inode_journal);
+               blknr = read_allocated_block(&inode_journal,
+                                       EXT2_JOURNAL_SUPERBLOCK);
+               ext4fs_devread(blknr * fs->sect_perblk, 0, fs->blksz,
+                              temp_buff);
+               jsb = (struct journal_superblock_t *)temp_buff;
+               jsb->s_start = cpu_to_be32(0);
+               put_ext4((uint64_t) (blknr * fs->blksz),
+                        (struct journal_superblock_t *)temp_buff, fs->blksz);
+               free(temp_buff);
+       }
+       ext4fs_free_journal();
+
+       /* get the superblock */
+       ext4fs_devread(SUPERBLOCK_SECTOR, 0, SUPERBLOCK_SIZE, (char *)fs->sb);
+       fs->sb->feature_incompat &= ~EXT3_FEATURE_INCOMPAT_RECOVER;
+       put_ext4((uint64_t)(SUPERBLOCK_SIZE),
+                (struct ext2_sblock *)fs->sb, (uint32_t)SUPERBLOCK_SIZE);
+       free(fs->sb);
+       fs->sb = NULL;
+
+       if (fs->blk_bmaps) {
+               for (i = 0; i < fs->no_blkgrp; i++) {
+                       free(fs->blk_bmaps[i]);
+                       fs->blk_bmaps[i] = NULL;
+               }
+               free(fs->blk_bmaps);
+               fs->blk_bmaps = NULL;
+       }
+
+       if (fs->inode_bmaps) {
+               for (i = 0; i < fs->no_blkgrp; i++) {
+                       free(fs->inode_bmaps[i]);
+                       fs->inode_bmaps[i] = NULL;
+               }
+               free(fs->inode_bmaps);
+               fs->inode_bmaps = NULL;
+       }
+
+
+       free(fs->gdtable);
+       fs->gdtable = NULL;
+       fs->gd = NULL;
+       /*
+        * reinitiliazed the global inode and
+        * block bitmap first execution check variables
+        */
+       fs->first_pass_ibmap = 0;
+       fs->first_pass_bbmap = 0;
+       fs->curr_inode_no = 0;
+       fs->curr_blkno = 0;
+}
+
+static int ext4fs_write_file(struct ext2_inode *file_inode,
+                            int pos, unsigned int len, char *buf)
+{
+       int i;
+       int blockcnt;
+       int log2blocksize = LOG2_EXT2_BLOCK_SIZE(ext4fs_root);
+       unsigned int filesize = __le32_to_cpu(file_inode->size);
+       struct ext_filesystem *fs = get_fs();
+       int previous_block_number = -1;
+       int delayed_start = 0;
+       int delayed_extent = 0;
+       int delayed_skipfirst = 0;
+       int delayed_next = 0;
+       char *delayed_buf = NULL;
+
+       /* Adjust len so it we can't read past the end of the file. */
+       if (len > filesize)
+               len = filesize;
+
+       blockcnt = ((len + pos) + fs->blksz - 1) / fs->blksz;
+
+       for (i = pos / fs->blksz; i < blockcnt; i++) {
+               long int blknr;
+               int blockend = fs->blksz;
+               int skipfirst = 0;
+               blknr = read_allocated_block(file_inode, i);
+               if (blknr < 0)
+                       return -1;
+
+               blknr = blknr << log2blocksize;
+
+               if (blknr) {
+                       if (previous_block_number != -1) {
+                               if (delayed_next == blknr) {
+                                       delayed_extent += blockend;
+                                       delayed_next += blockend >> SECTOR_BITS;
+                               } else {        /* spill */
+                                       put_ext4((uint64_t) (delayed_start *
+                                                            SECTOR_SIZE),
+                                                delayed_buf,
+                                                (uint32_t) delayed_extent);
+                                       previous_block_number = blknr;
+                                       delayed_start = blknr;
+                                       delayed_extent = blockend;
+                                       delayed_skipfirst = skipfirst;
+                                       delayed_buf = buf;
+                                       delayed_next = blknr +
+                                           (blockend >> SECTOR_BITS);
+                               }
+                       } else {
+                               previous_block_number = blknr;
+                               delayed_start = blknr;
+                               delayed_extent = blockend;
+                               delayed_skipfirst = skipfirst;
+                               delayed_buf = buf;
+                               delayed_next = blknr +
+                                   (blockend >> SECTOR_BITS);
+                       }
+               } else {
+                       if (previous_block_number != -1) {
+                               /* spill */
+                               put_ext4((uint64_t) (delayed_start *
+                                                    SECTOR_SIZE), delayed_buf,
+                                        (uint32_t) delayed_extent);
+                               previous_block_number = -1;
+                       }
+                       memset(buf, 0, fs->blksz - skipfirst);
+               }
+               buf += fs->blksz - skipfirst;
+       }
+       if (previous_block_number != -1) {
+               /* spill */
+               put_ext4((uint64_t) (delayed_start * SECTOR_SIZE),
+                        delayed_buf, (uint32_t) delayed_extent);
+               previous_block_number = -1;
+       }
+
+       return len;
+}
+
+int ext4fs_write(const char *fname, unsigned char *buffer,
+                                       unsigned long sizebytes)
+{
+       int ret = 0;
+       struct ext2_inode *file_inode = NULL;
+       unsigned char *inode_buffer = NULL;
+       int parent_inodeno;
+       int inodeno;
+       time_t timestamp = 0;
+
+       uint64_t bytes_reqd_for_file;
+       unsigned int blks_reqd_for_file;
+       unsigned int blocks_remaining;
+       int existing_file_inodeno;
+       char filename[256];
+
+       char *temp_ptr = NULL;
+       long int itable_blkno;
+       long int parent_itable_blkno;
+       long int blkoff;
+       struct ext2_sblock *sblock = &(ext4fs_root->sblock);
+       unsigned int inodes_per_block;
+       unsigned int ibmap_idx;
+       struct ext_filesystem *fs = get_fs();
+       g_parent_inode = zalloc(sizeof(struct ext2_inode));
+       if (!g_parent_inode)
+               goto fail;
+
+       if (ext4fs_init() != 0) {
+               printf("error in File System init\n");
+               return -1;
+       }
+       inodes_per_block = fs->blksz / fs->inodesz;
+       parent_inodeno = ext4fs_get_parent_inode_num(fname, filename, F_FILE);
+       if (parent_inodeno == -1)
+               goto fail;
+       if (ext4fs_iget(parent_inodeno, g_parent_inode))
+               goto fail;
+       /* check if the filename is already present in root */
+       existing_file_inodeno = ext4fs_filename_check(filename);
+       if (existing_file_inodeno != -1) {
+               ret = ext4fs_delete_file(existing_file_inodeno);
+               fs->first_pass_bbmap = 0;
+               fs->curr_blkno = 0;
+
+               fs->first_pass_ibmap = 0;
+               fs->curr_inode_no = 0;
+               if (ret)
+                       goto fail;
+       }
+       /* calucalate how many blocks required */
+       bytes_reqd_for_file = sizebytes;
+       blks_reqd_for_file = bytes_reqd_for_file / fs->blksz;
+       if (bytes_reqd_for_file % fs->blksz != 0) {
+               blks_reqd_for_file++;
+               debug("total bytes for a file %u\n", blks_reqd_for_file);
+       }
+       blocks_remaining = blks_reqd_for_file;
+       /* test for available space in partition */
+       if (fs->sb->free_blocks < blks_reqd_for_file) {
+               printf("Not enough space on partition !!!\n");
+               goto fail;
+       }
+
+       ext4fs_update_parent_dentry(filename, &inodeno, FILETYPE_REG);
+       /* prepare file inode */
+       inode_buffer = zalloc(fs->inodesz);
+       if (!inode_buffer)
+               goto fail;
+       file_inode = (struct ext2_inode *)inode_buffer;
+       file_inode->mode = S_IFREG | S_IRWXU |
+           S_IRGRP | S_IROTH | S_IXGRP | S_IXOTH;
+       /* ToDo: Update correct time */
+       file_inode->mtime = timestamp;
+       file_inode->atime = timestamp;
+       file_inode->ctime = timestamp;
+       file_inode->nlinks = 1;
+       file_inode->size = sizebytes;
+
+       /* Allocate data blocks */
+       ext4fs_allocate_blocks(file_inode, blocks_remaining,
+                              &blks_reqd_for_file);
+       file_inode->blockcnt = (blks_reqd_for_file * fs->blksz) / SECTOR_SIZE;
+
+       temp_ptr = zalloc(fs->blksz);
+       if (!temp_ptr)
+               goto fail;
+       ibmap_idx = inodeno / ext4fs_root->sblock.inodes_per_group;
+       inodeno--;
+       itable_blkno = __le32_to_cpu(fs->gd[ibmap_idx].inode_table_id) +
+                       (inodeno % __le32_to_cpu(sblock->inodes_per_group)) /
+                       inodes_per_block;
+       blkoff = (inodeno % inodes_per_block) * fs->inodesz;
+       ext4fs_devread(itable_blkno * fs->sect_perblk, 0, fs->blksz, temp_ptr);
+       if (ext4fs_log_journal(temp_ptr, itable_blkno))
+               goto fail;
+
+       memcpy(temp_ptr + blkoff, inode_buffer, fs->inodesz);
+       if (ext4fs_put_metadata(temp_ptr, itable_blkno))
+               goto fail;
+       /* copy the file content into data blocks */
+       if (ext4fs_write_file(file_inode, 0, sizebytes, (char *)buffer) == -1) {
+               printf("Error in copying content\n");
+               goto fail;
+       }
+       ibmap_idx = parent_inodeno / ext4fs_root->sblock.inodes_per_group;
+       parent_inodeno--;
+       parent_itable_blkno = __le32_to_cpu(fs->gd[ibmap_idx].inode_table_id) +
+           (parent_inodeno %
+            __le32_to_cpu(sblock->inodes_per_group)) / inodes_per_block;
+       blkoff = (parent_inodeno % inodes_per_block) * fs->inodesz;
+       if (parent_itable_blkno != itable_blkno) {
+               memset(temp_ptr, '\0', fs->blksz);
+               ext4fs_devread(parent_itable_blkno * fs->sect_perblk,
+                              0, fs->blksz, temp_ptr);
+               if (ext4fs_log_journal(temp_ptr, parent_itable_blkno))
+                       goto fail;
+
+               memcpy(temp_ptr + blkoff, g_parent_inode,
+                       sizeof(struct ext2_inode));
+               if (ext4fs_put_metadata(temp_ptr, parent_itable_blkno))
+                       goto fail;
+               free(temp_ptr);
+       } else {
+               /*
+                * If parent and child fall in same inode table block
+                * both should be kept in 1 buffer
+                */
+               memcpy(temp_ptr + blkoff, g_parent_inode,
+                      sizeof(struct ext2_inode));
+               gd_index--;
+               if (ext4fs_put_metadata(temp_ptr, itable_blkno))
+                       goto fail;
+               free(temp_ptr);
+       }
+       ext4fs_update();
+       ext4fs_deinit();
+
+       fs->first_pass_bbmap = 0;
+       fs->curr_blkno = 0;
+       fs->first_pass_ibmap = 0;
+       fs->curr_inode_no = 0;
+       free(inode_buffer);
+       free(g_parent_inode);
+       g_parent_inode = NULL;
+
+       return 0;
+fail:
+       ext4fs_deinit();
+       free(inode_buffer);
+       free(g_parent_inode);
+       g_parent_inode = NULL;
+
+       return -1;
+}
+#endif
diff --git a/include/ext2fs.h b/include/ext2fs.h
deleted file mode 100644 (file)
index 163a9bb..0000000
+++ /dev/null
@@ -1,81 +0,0 @@
-/*
- *  GRUB  --  GRand Unified Bootloader
- *  Copyright (C) 2000, 2001  Free Software Foundation, Inc.
- *
- *  (C) Copyright 2003 Sysgo Real-Time Solutions, AG <www.elinos.com>
- *  Pavel Bartusek <pba@sysgo.de>
- *
- *  This program is free software; you can redistribute it and/or modify
- *  it under the terms of the GNU General Public License as published by
- *  the Free Software Foundation; either version 2 of the License, or
- *  (at your option) any later version.
- *
- *  This program is distributed in the hope that it will be useful,
- *  but WITHOUT ANY WARRANTY; without even the implied warranty of
- *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- *  GNU General Public License for more details.
- *
- *  You should have received a copy of the GNU General Public License
- *  along with this program; if not, write to the Free Software
- *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
- */
-
-/* An implementation for the Ext2FS filesystem ported from GRUB.
- * Some parts of this code (mainly the structures and defines) are
- * from the original ext2 fs code, as found in the linux kernel.
- */
-
-
-#define SECTOR_SIZE            0x200
-#define SECTOR_BITS            9
-
-/* Error codes */
-typedef enum
-{
-  ERR_NONE = 0,
-  ERR_BAD_FILENAME,
-  ERR_BAD_FILETYPE,
-  ERR_BAD_GZIP_DATA,
-  ERR_BAD_GZIP_HEADER,
-  ERR_BAD_PART_TABLE,
-  ERR_BAD_VERSION,
-  ERR_BELOW_1MB,
-  ERR_BOOT_COMMAND,
-  ERR_BOOT_FAILURE,
-  ERR_BOOT_FEATURES,
-  ERR_DEV_FORMAT,
-  ERR_DEV_VALUES,
-  ERR_EXEC_FORMAT,
-  ERR_FILELENGTH,
-  ERR_FILE_NOT_FOUND,
-  ERR_FSYS_CORRUPT,
-  ERR_FSYS_MOUNT,
-  ERR_GEOM,
-  ERR_NEED_LX_KERNEL,
-  ERR_NEED_MB_KERNEL,
-  ERR_NO_DISK,
-  ERR_NO_PART,
-  ERR_NUMBER_PARSING,
-  ERR_OUTSIDE_PART,
-  ERR_READ,
-  ERR_SYMLINK_LOOP,
-  ERR_UNRECOGNIZED,
-  ERR_WONT_FIT,
-  ERR_WRITE,
-  ERR_BAD_ARGUMENT,
-  ERR_UNALIGNED,
-  ERR_PRIVILEGED,
-  ERR_DEV_NEED_INIT,
-  ERR_NO_DISK_SPACE,
-  ERR_NUMBER_OVERFLOW,
-
-  MAX_ERR_NUM
-} ext2fs_error_t;
-
-
-extern int ext2fs_set_blk_dev(block_dev_desc_t *rbdd, int part);
-extern int ext2fs_ls (const char *dirname);
-extern int ext2fs_open (const char *filename);
-extern int ext2fs_read (char *buf, unsigned len);
-extern int ext2fs_mount (unsigned part_length);
-extern int ext2fs_close(void);
diff --git a/include/ext4fs.h b/include/ext4fs.h
new file mode 100644 (file)
index 0000000..ab2983c
--- /dev/null
@@ -0,0 +1,144 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * Ext4 Extent data structures are taken from  original ext4 fs code
+ * as found in the linux kernel.
+ *
+ * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
+ * Written by Alex Tomas <alex@clusterfs.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef __EXT4__
+#define __EXT4__
+#include <ext_common.h>
+
+#define EXT4_EXTENTS_FL                0x00080000 /* Inode uses extents */
+#define EXT4_EXT_MAGIC                 0xf30a
+#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM        0x0010
+#define EXT4_FEATURE_INCOMPAT_EXTENTS  0x0040
+#define EXT4_INDIRECT_BLOCKS           12
+
+#define EXT4_BG_INODE_UNINIT           0x0001
+#define EXT4_BG_BLOCK_UNINIT           0x0002
+#define EXT4_BG_INODE_ZEROED           0x0004
+
+/*
+ * ext4_inode has i_block array (60 bytes total).
+ * The first 12 bytes store ext4_extent_header;
+ * the remainder stores an array of ext4_extent.
+ */
+
+/*
+ * This is the extent on-disk structure.
+ * It's used at the bottom of the tree.
+ */
+struct ext4_extent {
+       __le32  ee_block;       /* first logical block extent covers */
+       __le16  ee_len;         /* number of blocks covered by extent */
+       __le16  ee_start_hi;    /* high 16 bits of physical block */
+       __le32  ee_start_lo;    /* low 32 bits of physical block */
+};
+
+/*
+ * This is index on-disk structure.
+ * It's used at all the levels except the bottom.
+ */
+struct ext4_extent_idx {
+       __le32  ei_block;       /* index covers logical blocks from 'block' */
+       __le32  ei_leaf_lo;     /* pointer to the physical block of the next *
+                                * level. leaf or next index could be there */
+       __le16  ei_leaf_hi;     /* high 16 bits of physical block */
+       __u16   ei_unused;
+};
+
+/* Each block (leaves and indexes), even inode-stored has header. */
+struct ext4_extent_header {
+       __le16  eh_magic;       /* probably will support different formats */
+       __le16  eh_entries;     /* number of valid entries */
+       __le16  eh_max;         /* capacity of store in entries */
+       __le16  eh_depth;       /* has tree real underlying blocks? */
+       __le32  eh_generation;  /* generation of the tree */
+};
+
+struct ext_filesystem {
+       /* Total Sector of partition */
+       uint64_t total_sect;
+       /* Block size  of partition */
+       uint32_t blksz;
+       /* Inode size of partition */
+       uint32_t inodesz;
+       /* Sectors per Block */
+       uint32_t sect_perblk;
+       /* Group Descriptor Block Number */
+       uint32_t gdtable_blkno;
+       /* Total block groups of partition */
+       uint32_t no_blkgrp;
+       /* No of blocks required for bgdtable */
+       uint32_t no_blk_pergdt;
+       /* Superblock */
+       struct ext2_sblock *sb;
+       /* Block group descritpor table */
+       struct ext2_block_group *gd;
+       char *gdtable;
+
+       /* Block Bitmap Related */
+       unsigned char **blk_bmaps;
+       long int curr_blkno;
+       uint16_t first_pass_bbmap;
+
+       /* Inode Bitmap Related */
+       unsigned char **inode_bmaps;
+       int curr_inode_no;
+       uint16_t first_pass_ibmap;
+
+       /* Journal Related */
+
+       /* Block Device Descriptor */
+       block_dev_desc_t *dev_desc;
+};
+
+extern block_dev_desc_t *ext4_dev_desc;
+extern struct ext2_data *ext4fs_root;
+extern struct ext2fs_node *ext4fs_file;
+
+#if defined(CONFIG_CMD_EXT4_WRITE)
+extern struct ext2_inode *g_parent_inode;
+extern int gd_index;
+extern int gindex;
+
+int ext4fs_init(void);
+void ext4fs_deinit(void);
+int ext4fs_filename_check(char *filename);
+int ext4fs_write(const char *fname, unsigned char *buffer,
+                               unsigned long sizebytes);
+#endif
+
+struct ext_filesystem *get_fs(void);
+int init_fs(block_dev_desc_t *dev_desc);
+void deinit_fs(block_dev_desc_t *dev_desc);
+int ext4fs_open(const char *filename);
+int ext4fs_read(char *buf, unsigned len);
+int ext4fs_mount(unsigned part_length);
+void ext4fs_close(void);
+int ext4fs_ls(const char *dirname);
+void ext4fs_free_node(struct ext2fs_node *node, struct ext2fs_node *currroot);
+int ext4fs_devread(int sector, int byte_offset, int byte_len, char *buf);
+int ext4fs_set_blk_dev(block_dev_desc_t *rbdd, int part);
+long int read_allocated_block(struct ext2_inode *inode, int fileblock);
+#endif
diff --git a/include/ext_common.h b/include/ext_common.h
new file mode 100644 (file)
index 0000000..9b97522
--- /dev/null
@@ -0,0 +1,199 @@
+/*
+ * (C) Copyright 2011 - 2012 Samsung Electronics
+ * EXT4 filesystem implementation in Uboot by
+ * Uma Shankar <uma.shankar@samsung.com>
+ * Manjunatha C Achar <a.manjunatha@samsung.com>
+ *
+ * Data structures and headers for ext4 support have been taken from
+ * ext2 ls load support in Uboot
+ *
+ * (C) Copyright 2004
+ * esd gmbh <www.esd-electronics.com>
+ * Reinhard Arlt <reinhard.arlt@esd-electronics.com>
+ *
+ * based on code from grub2 fs/ext2.c and fs/fshelp.c by
+ * GRUB  --  GRand Unified Bootloader
+ * Copyright (C) 2003, 2004  Free Software Foundation, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef __EXT_COMMON__
+#define __EXT_COMMON__
+#include <command.h>
+#define SECTOR_SIZE            0x200
+#define SECTOR_BITS            9
+
+/* Magic value used to identify an ext2 filesystem.  */
+#define        EXT2_MAGIC                      0xEF53
+/* Amount of indirect blocks in an inode.  */
+#define INDIRECT_BLOCKS                        12
+/* Maximum lenght of a pathname.  */
+#define EXT2_PATH_MAX                          4096
+/* Maximum nesting of symlinks, used to prevent a loop.  */
+#define        EXT2_MAX_SYMLINKCNT             8
+
+/* Filetype used in directory entry.  */
+#define        FILETYPE_UNKNOWN                0
+#define        FILETYPE_REG                    1
+#define        FILETYPE_DIRECTORY              2
+#define        FILETYPE_SYMLINK                7
+
+/* Filetype information as used in inodes.  */
+#define FILETYPE_INO_MASK              0170000
+#define FILETYPE_INO_REG               0100000
+#define FILETYPE_INO_DIRECTORY         0040000
+#define FILETYPE_INO_SYMLINK           0120000
+#define EXT2_ROOT_INO                  2 /* Root inode */
+
+/* Bits used as offset in sector */
+#define DISK_SECTOR_BITS               9
+/* The size of an ext2 block in bytes.  */
+#define EXT2_BLOCK_SIZE(data)     (1 << LOG2_BLOCK_SIZE(data))
+
+/* Log2 size of ext2 block in 512 blocks.  */
+#define LOG2_EXT2_BLOCK_SIZE(data) (__le32_to_cpu \
+                               (data->sblock.log2_block_size) + 1)
+
+/* Log2 size of ext2 block in bytes.  */
+#define LOG2_BLOCK_SIZE(data)     (__le32_to_cpu \
+               (data->sblock.log2_block_size) + 10)
+#define INODE_SIZE_FILESYSTEM(data)    (__le32_to_cpu \
+                       (data->sblock.inode_size))
+
+#define EXT2_FT_DIR    2
+#define SUCCESS        1
+
+/* Macro-instructions used to manage several block sizes  */
+#define EXT2_MIN_BLOCK_LOG_SIZE        10 /* 1024 */
+#define EXT2_MAX_BLOCK_LOG_SIZE        16 /* 65536 */
+#define EXT2_MIN_BLOCK_SIZE            (1 << EXT2_MIN_BLOCK_LOG_SIZE)
+#define EXT2_MAX_BLOCK_SIZE            (1 << EXT2_MAX_BLOCK_LOG_SIZE)
+
+/* The ext2 superblock.  */
+struct ext2_sblock {
+       uint32_t total_inodes;
+       uint32_t total_blocks;
+       uint32_t reserved_blocks;
+       uint32_t free_blocks;
+       uint32_t free_inodes;
+       uint32_t first_data_block;
+       uint32_t log2_block_size;
+       uint32_t log2_fragment_size;
+       uint32_t blocks_per_group;
+       uint32_t fragments_per_group;
+       uint32_t inodes_per_group;
+       uint32_t mtime;
+       uint32_t utime;
+       uint16_t mnt_count;
+       uint16_t max_mnt_count;
+       uint16_t magic;
+       uint16_t fs_state;
+       uint16_t error_handling;
+       uint16_t minor_revision_level;
+       uint32_t lastcheck;
+       uint32_t checkinterval;
+       uint32_t creator_os;
+       uint32_t revision_level;
+       uint16_t uid_reserved;
+       uint16_t gid_reserved;
+       uint32_t first_inode;
+       uint16_t inode_size;
+       uint16_t block_group_number;
+       uint32_t feature_compatibility;
+       uint32_t feature_incompat;
+       uint32_t feature_ro_compat;
+       uint32_t unique_id[4];
+       char volume_name[16];
+       char last_mounted_on[64];
+       uint32_t compression_info;
+};
+
+struct ext2_block_group {
+       __u32 block_id; /* Blocks bitmap block */
+       __u32 inode_id; /* Inodes bitmap block */
+       __u32 inode_table_id;   /* Inodes table block */
+       __u16 free_blocks;      /* Free blocks count */
+       __u16 free_inodes;      /* Free inodes count */
+       __u16 used_dir_cnt;     /* Directories count */
+       __u16 bg_flags;
+       __u32 bg_reserved[2];
+       __u16 bg_itable_unused; /* Unused inodes count */
+       __u16 bg_checksum;      /* crc16(s_uuid+grouo_num+group_desc)*/
+};
+
+/* The ext2 inode. */
+struct ext2_inode {
+       uint16_t mode;
+       uint16_t uid;
+       uint32_t size;
+       uint32_t atime;
+       uint32_t ctime;
+       uint32_t mtime;
+       uint32_t dtime;
+       uint16_t gid;
+       uint16_t nlinks;
+       uint32_t blockcnt;      /* Blocks of 512 bytes!! */
+       uint32_t flags;
+       uint32_t osd1;
+       union {
+               struct datablocks {
+                       uint32_t dir_blocks[INDIRECT_BLOCKS];
+                       uint32_t indir_block;
+                       uint32_t double_indir_block;
+                       uint32_t triple_indir_block;
+               } blocks;
+               char symlink[60];
+       } b;
+       uint32_t version;
+       uint32_t acl;
+       uint32_t dir_acl;
+       uint32_t fragment_addr;
+       uint32_t osd2[3];
+};
+
+/* The header of an ext2 directory entry. */
+struct ext2_dirent {
+       uint32_t inode;
+       uint16_t direntlen;
+       uint8_t namelen;
+       uint8_t filetype;
+};
+
+struct ext2fs_node {
+       struct ext2_data *data;
+       struct ext2_inode inode;
+       int ino;
+       int inode_read;
+};
+
+/* Information about a "mounted" ext2 filesystem. */
+struct ext2_data {
+       struct ext2_sblock sblock;
+       struct ext2_inode *inode;
+       struct ext2fs_node diropen;
+};
+
+int do_ext2ls(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]);
+int do_ext2load(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]);
+int do_ext4_load(cmd_tbl_t *cmdtp, int flag, int argc,
+                                       char *const argv[]);
+int do_ext4_ls(cmd_tbl_t *cmdtp, int flag, int argc, char *const argv[]);
+int do_ext4_write(cmd_tbl_t *cmdtp, int flag, int argc,
+                               char *const argv[]);
+int do_ext_load(cmd_tbl_t *cmdtp, int flag, int argc,
+                                       char *const argv[]);
+int do_ext_ls(cmd_tbl_t *cmdtp, int flag, int argc, char *const argv[]);
+#endif