aboutsummaryrefslogtreecommitdiff
path: root/fs
diff options
context:
space:
mode:
authorSimon Glass2012-12-26 09:53:28 +0000
committerTom Rini2013-03-04 14:19:56 -0500
commit293d7fbd47f473d8a6b01a5f80d7faa5cd6b67b3 (patch)
treecd375e2fff0e38133d870c455a8380a25df25265 /fs
parent1c9f47ab2a2e9b62d08d39bfb9c4adc8f8edc5da (diff)
ext4: Split write support into its own file
This code seems to be entirely othogonal, so remove the #ifdef and put the condition in the Makefile instead. Signed-off-by: Simon Glass <sjg@chromium.org>
Diffstat (limited to 'fs')
-rw-r--r--fs/ext4/Makefile2
-rw-r--r--fs/ext4/ext4_write.c996
-rw-r--r--fs/ext4/ext4fs.c963
3 files changed, 997 insertions, 964 deletions
diff --git a/fs/ext4/Makefile b/fs/ext4/Makefile
index bb801f90547..3bde82418e5 100644
--- a/fs/ext4/Makefile
+++ b/fs/ext4/Makefile
@@ -31,7 +31,7 @@ LIB = $(obj)libext4fs.o
AOBJS =
COBJS-$(CONFIG_FS_EXT4) := ext4fs.o ext4_common.o dev.o
-COBJS-$(CONFIG_EXT4_WRITE) += ext4_journal.o crc16.o
+COBJS-$(CONFIG_EXT4_WRITE) += ext4_write.o ext4_journal.o crc16.o
SRCS := $(AOBJS:.o=.S) $(COBJS-y:.o=.c)
OBJS := $(addprefix $(obj),$(AOBJS) $(COBJS-y))
diff --git a/fs/ext4/ext4_write.c b/fs/ext4/ext4_write.c
new file mode 100644
index 00000000000..c4e399ccfae
--- /dev/null
+++ b/fs/ext4/ext4_write.c
@@ -0,0 +1,996 @@
+/*
+ * (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 <linux/stat.h>
+#include <div64.h>
+#include "ext4_common.h"
+
+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->bgd[i].bg_checksum = ext4fs_checksum_update(i);
+ put_ext4((uint64_t)(fs->bgd[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->bgd[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 *bgd = 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 */
+ bgd = (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);
+ bgd[bg_idx].free_blocks++;
+ fs->sb->free_blocks++;
+ /* journal backup */
+ if (prev_bg_bmap_idx != bg_idx) {
+ status =
+ ext4fs_devread(bgd[bg_idx].block_id *
+ fs->sect_perblk, 0, fs->blksz,
+ journal_buffer);
+ if (status == 0)
+ goto fail;
+ if (ext4fs_log_journal
+ (journal_buffer, bgd[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 *bgd = 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 */
+ bgd = (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++;
+ bgd[bg_idx].free_blocks++;
+ fs->sb->free_blocks++;
+ /* journal backup */
+ if (prev_bg_bmap_idx != bg_idx) {
+ status = ext4fs_devread(bgd[bg_idx].block_id
+ * fs->sect_perblk, 0,
+ fs->blksz,
+ journal_buffer);
+ if (status == 0)
+ goto fail;
+
+ if (ext4fs_log_journal(journal_buffer,
+ bgd[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);
+ bgd[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(bgd[bg_idx].block_id *
+ fs->sect_perblk, 0, fs->blksz,
+ journal_buffer);
+ if (status == 0)
+ goto fail;
+
+ if (ext4fs_log_journal(journal_buffer,
+ bgd[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 *bgd = 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 */
+ bgd = (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++;
+ bgd[bg_idx].free_blocks++;
+ fs->sb->free_blocks++;
+ /* journal backup */
+ if (prev_bg_bmap_idx != bg_idx) {
+ status =
+ ext4fs_devread(
+ bgd[bg_idx].block_id *
+ fs->sect_perblk, 0,
+ fs->blksz,
+ journal_buffer);
+ if (status == 0)
+ goto fail;
+
+ if (ext4fs_log_journal(journal_buffer,
+ bgd[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++;
+ bgd[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(bgd[bg_idx].block_id *
+ fs->sect_perblk, 0,
+ fs->blksz, journal_buffer);
+ if (status == 0)
+ goto fail;
+
+ if (ext4fs_log_journal(journal_buffer,
+ bgd[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);
+ bgd[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(bgd[bg_idx].block_id *
+ fs->sect_perblk, 0, fs->blksz,
+ journal_buffer);
+ if (status == 0)
+ goto fail;
+
+ if (ext4fs_log_journal(journal_buffer,
+ bgd[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 *bgd = NULL;
+ struct ext_filesystem *fs = get_fs();
+ char *journal_buffer = zalloc(fs->blksz);
+ if (!journal_buffer)
+ return -ENOMEM;
+ /* get the block group descriptor table */
+ bgd = (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);
+
+ bgd[bg_idx].free_blocks++;
+ fs->sb->free_blocks++;
+
+ /* journal backup */
+ if (prev_bg_bmap_idx != bg_idx) {
+ status =
+ ext4fs_devread(bgd[bg_idx].block_id *
+ fs->sect_perblk, 0,
+ fs->blksz, journal_buffer);
+ if (status == 0)
+ goto fail;
+ if (ext4fs_log_journal(journal_buffer,
+ bgd[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);
+
+ bgd[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(bgd[bg_idx].block_id
+ * fs->sect_perblk,
+ 0, fs->blksz,
+ journal_buffer);
+ if (status == 0)
+ goto fail;
+ if (ext4fs_log_journal(journal_buffer,
+ bgd[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(bgd[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);
+ bgd[ibmap_idx].free_inodes++;
+ fs->sb->free_inodes++;
+ /* journal backup */
+ memset(journal_buffer, '\0', fs->blksz);
+ status = ext4fs_devread(bgd[ibmap_idx].inode_id *
+ fs->sect_perblk, 0, fs->blksz, journal_buffer);
+ if (status == 0)
+ goto fail;
+ if (ext4fs_log_journal(journal_buffer, bgd[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->bgd = (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->bgd[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->bgd[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->bgd[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->bgd = 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_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_buf = buf;
+ delayed_next = blknr +
+ (blockend >> SECTOR_BITS);
+ }
+ } else {
+ previous_block_number = blknr;
+ delayed_start = blknr;
+ delayed_extent = blockend;
+ 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 *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();
+ ALLOC_CACHE_ALIGN_BUFFER(char, filename, 256);
+ memset(filename, 0x00, sizeof(filename));
+
+ 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 = lldiv(bytes_reqd_for_file, fs->blksz);
+ if (do_div(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->bgd[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->bgd[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;
+}
diff --git a/fs/ext4/ext4fs.c b/fs/ext4/ext4fs.c
index f02c215ccc2..7fdb4634c79 100644
--- a/fs/ext4/ext4fs.c
+++ b/fs/ext4/ext4fs.c
@@ -34,13 +34,8 @@
*/
#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 <div64.h>
#include "ext4_common.h"
int ext4fs_symlinknest;
@@ -196,961 +191,3 @@ int ext4fs_read(char *buf, unsigned len)
return ext4fs_read_file(ext4fs_file, 0, len, buf);
}
-
-#if defined(CONFIG_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->bgd[i].bg_checksum = ext4fs_checksum_update(i);
- put_ext4((uint64_t)(fs->bgd[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->bgd[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 *bgd = 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 */
- bgd = (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);
- bgd[bg_idx].free_blocks++;
- fs->sb->free_blocks++;
- /* journal backup */
- if (prev_bg_bmap_idx != bg_idx) {
- status =
- ext4fs_devread(bgd[bg_idx].block_id *
- fs->sect_perblk, 0, fs->blksz,
- journal_buffer);
- if (status == 0)
- goto fail;
- if (ext4fs_log_journal
- (journal_buffer, bgd[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 *bgd = 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 */
- bgd = (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++;
- bgd[bg_idx].free_blocks++;
- fs->sb->free_blocks++;
- /* journal backup */
- if (prev_bg_bmap_idx != bg_idx) {
- status = ext4fs_devread(bgd[bg_idx].block_id
- * fs->sect_perblk, 0,
- fs->blksz,
- journal_buffer);
- if (status == 0)
- goto fail;
-
- if (ext4fs_log_journal(journal_buffer,
- bgd[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);
- bgd[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(bgd[bg_idx].block_id *
- fs->sect_perblk, 0, fs->blksz,
- journal_buffer);
- if (status == 0)
- goto fail;
-
- if (ext4fs_log_journal(journal_buffer,
- bgd[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 *bgd = 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 */
- bgd = (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++;
- bgd[bg_idx].free_blocks++;
- fs->sb->free_blocks++;
- /* journal backup */
- if (prev_bg_bmap_idx != bg_idx) {
- status =
- ext4fs_devread(
- bgd[bg_idx].block_id *
- fs->sect_perblk, 0,
- fs->blksz,
- journal_buffer);
- if (status == 0)
- goto fail;
-
- if (ext4fs_log_journal(journal_buffer,
- bgd[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++;
- bgd[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(bgd[bg_idx].block_id *
- fs->sect_perblk, 0,
- fs->blksz, journal_buffer);
- if (status == 0)
- goto fail;
-
- if (ext4fs_log_journal(journal_buffer,
- bgd[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);
- bgd[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(bgd[bg_idx].block_id *
- fs->sect_perblk, 0, fs->blksz,
- journal_buffer);
- if (status == 0)
- goto fail;
-
- if (ext4fs_log_journal(journal_buffer,
- bgd[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 *bgd = NULL;
- struct ext_filesystem *fs = get_fs();
- char *journal_buffer = zalloc(fs->blksz);
- if (!journal_buffer)
- return -ENOMEM;
- /* get the block group descriptor table */
- bgd = (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);
-
- bgd[bg_idx].free_blocks++;
- fs->sb->free_blocks++;
-
- /* journal backup */
- if (prev_bg_bmap_idx != bg_idx) {
- status =
- ext4fs_devread(bgd[bg_idx].block_id *
- fs->sect_perblk, 0,
- fs->blksz, journal_buffer);
- if (status == 0)
- goto fail;
- if (ext4fs_log_journal(journal_buffer,
- bgd[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);
-
- bgd[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(bgd[bg_idx].block_id
- * fs->sect_perblk,
- 0, fs->blksz,
- journal_buffer);
- if (status == 0)
- goto fail;
- if (ext4fs_log_journal(journal_buffer,
- bgd[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(bgd[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);
- bgd[ibmap_idx].free_inodes++;
- fs->sb->free_inodes++;
- /* journal backup */
- memset(journal_buffer, '\0', fs->blksz);
- status = ext4fs_devread(bgd[ibmap_idx].inode_id *
- fs->sect_perblk, 0, fs->blksz, journal_buffer);
- if (status == 0)
- goto fail;
- if (ext4fs_log_journal(journal_buffer, bgd[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->bgd = (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->bgd[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->bgd[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->bgd[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->bgd = 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_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_buf = buf;
- delayed_next = blknr +
- (blockend >> SECTOR_BITS);
- }
- } else {
- previous_block_number = blknr;
- delayed_start = blknr;
- delayed_extent = blockend;
- 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 *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();
- ALLOC_CACHE_ALIGN_BUFFER(char, filename, 256);
- memset(filename, 0x00, sizeof(filename));
-
- 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 = lldiv(bytes_reqd_for_file, fs->blksz);
- if (do_div(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->bgd[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->bgd[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