On Aug 22, 2007 22:41 -0700, Avantika Mathur wrote:
> Below is an updated version of the uninitialized block groups patch,
> ported to the end of the patch queue. There are small changes to the
> previous version:
> - consistent format of the group_desc structure with the e2fsprogs patches
> - do not verify group_desc checksums if the feature is not enabled
Can you please keep the original description intact with the patch, so
it is useful for Andrew when it is submitted upstream:
Keep a high water mark of used inodes for each group to improve e2fsck
time. Block and inode bitmaps can be uninitialized on disk via a flag
in the group descriptor to avoid reading or scanning them at e2fsck time.
A checksum of each group descriptor is used to ensure that corruption in
the group descriptor's bit flags does not cause incorrect operation.
One last note - if this is being applied on top of mballoc it needs one
last patch for that file (against the 2.6.18-rhel5 tree, should work
with s/ext3/ext4/ I think):
Signed-off-by: Andreas Dilger <adilger@xxxxxxxxxxxxx>
Index: linux-rhel5/fs/ext3/mballoc.c
===================================================================
--- linux-rhel5.orig/fs/ext3/mballoc.c 2007-07-18 17:32:04.000000000 +0200
+++ linux-rhel5/fs/ext3/mballoc.c 2007-07-18 17:32:15.000000000 +0200
@@ -36,6 +36,8 @@
#include <linux/seq_file.h>
#include <linux/version.h>
+#include "group.h"
+
/*
* MUSTDO:
* - test ext3_ext_search_left() and ext3_ext_search_right()
@@ -323,6 +325,7 @@ struct ext3_group_info {
unsigned long bb_state;
unsigned long bb_tid;
struct ext3_free_metadata *bb_md_cur;
+ struct ext3_group_desc *bb_gdp;
unsigned short bb_first_free;
unsigned short bb_free;
unsigned short bb_fragments;
@@ -943,10 +946,7 @@ static int ext3_mb_init_cache(struct pag
if (first_group + i >= EXT3_SB(sb)->s_groups_count)
break;
- err = -EIO;
- desc = ext3_get_group_desc(sb, first_group + i, NULL);
- if (desc == NULL)
- goto out;
+ desc = EXT3_GROUP_INFO(sb, first_group + i)->bb_gdp;
err = -ENOMEM;
bh[i] = sb_getblk(sb, le32_to_cpu(desc->bg_block_bitmap));
@@ -961,7 +961,12 @@ static int ext3_mb_init_cache(struct pag
unlock_buffer(bh[i]);
continue;
}
-
+ if (desc->bg_flags & cpu_to_le16(EXT3_BG_BLOCK_UNINIT)) {
+ ext3_init_block_bitmap(sb, bh[i], first_group + i,desc);
+ set_buffer_uptodate(bh[i]);
+ unlock_buffer(bh[i]);
+ continue;
+ }
get_bh(bh[i]);
bh[i]->b_end_io = end_buffer_read_sync;
submit_bh(READ, bh[i]);
@@ -1732,6 +1737,10 @@ static int ext3_mb_good_group(struct ext
switch (cr) {
case 0:
BUG_ON(ac->ac_2order == 0);
+ /* If this group is uninitialized, skip it initially */
+ if (grp->bb_gdp->bg_flags &
+ cpu_to_le16(EXT3_BG_BLOCK_UNINIT))
+ return 0;
bits = ac->ac_sb->s_blocksize_bits + 1;
for (i = ac->ac_2order; i <= bits; i++)
if (grp->bb_counters[i] > 0)
@@ -1825,7 +1834,9 @@ repeat:
}
ac->ac_groups_scanned++;
- if (cr == 0)
+ if (cr == 0 || (e3b.bd_info->bb_gdp->bg_flags &
+ cpu_to_le16(EXT3_BG_BLOCK_UNINIT) &&
+ ac->ac_2order != 0))
ext3_mb_simple_scan_group(ac, &e3b);
else if (cr == 1 && ac->ac_g_ex.fe_len == sbi->s_stripe)
ext3_mb_scan_aligned(ac, &e3b);
@@ -2304,12 +2315,13 @@ int ext3_mb_init_backend(struct super_bl
i--;
goto err_freebuddy;
}
+ memset(meta_group_info[j], 0, len);
desc = ext3_get_group_desc(sb, i, NULL);
+ meta_group_info[j]->bb_gdp = desc;
if (desc == NULL) {
printk(KERN_ERR"EXT3-fs: can't read descriptor %u\n",i);
goto err_freebuddy;
}
- memset(meta_group_info[j], 0, len);
set_bit(EXT3_GROUP_INFO_NEED_INIT_BIT,
&meta_group_info[j]->bb_state);
@@ -2958,9 +2970,17 @@ int ext3_mb_mark_diskspace_used(struct e
mb_set_bits(bitmap_bh->b_data, ac->ac_b_ex.fe_start, ac->ac_b_ex.fe_len);
spin_lock(sb_bgl_lock(sbi, ac->ac_b_ex.fe_group));
+ if (gdp->bg_flags & cpu_to_le16(EXT3_BG_BLOCK_UNINIT)) {
+ gdp->bg_flags &= cpu_to_le16(~EXT3_BG_BLOCK_UNINIT);
+ gdp->bg_free_blocks_count =
+ cpu_to_le16(ext3_free_blocks_after_init(sb,
+ ac->ac_b_ex.fe_group,
+ gdp));
+ }
gdp->bg_free_blocks_count =
cpu_to_le16(le16_to_cpu(gdp->bg_free_blocks_count)
- ac->ac_b_ex.fe_len);
+ gdp->bg_checksum = ext3_group_desc_csum(sbi, ac->ac_b_ex.fe_group, gdp);
spin_unlock(sb_bgl_lock(sbi, ac->ac_b_ex.fe_group));
percpu_counter_mod(&sbi->s_freeblocks_counter, - ac->ac_b_ex.fe_len);
@@ -4346,6 +4366,7 @@ do_more:
spin_lock(sb_bgl_lock(sbi, block_group));
gdp->bg_free_blocks_count =
cpu_to_le16(le16_to_cpu(gdp->bg_free_blocks_count) + count);
+ gdp->bg_checksum = ext3_group_desc_csum(sbi, block_group, gdp);
spin_unlock(sb_bgl_lock(sbi, block_group));
percpu_counter_mod(&sbi->s_freeblocks_counter, count);
Cheers, Andreas
--
Andreas Dilger
Principal Software Engineer
Cluster File Systems, Inc.
-
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
