FLEX_BG Kernel support.

PATCH In review Jose R. Santos 2 participants 5 messages Sep 10, 2007 – Sep 12, 2007 ext4 file system

lore

Share & tools

lore

Jose R. Santos

author

2007-09-10 21:25 UTC #0 lore raw

PATCH 1/1 FLEX_BG Kernel support.

From: Jose R. Santos <[email protected]>

FLEX_BG Kernel support.

This feature relaxes check restrictions on where each block groups meta data is located within the storage media. This allows for the allocation of bitmaps or inode tables outside the block group boundaries in cases where bad blocks forces us to look for new blocks which the owning block group can not satisfy. This will also allow for new meta-data allocation schemes to improve performance and scalability.

Signed-off-by Jose R. Santos <[email protected]>

 fs/ext4/super.c         |    7 +++++--
 include/linux/ext4_fs.h |    4 +++-
 2 files changed, 8 insertions(+), 3 deletions(-)

fs/ext4/super.c +5 -2

index 4550b83..902e5c8 100644
@@ -1254,7 +1254,8 @@ static int ext4_check_descriptors (struct super_block * sb)
 
 	for (i = 0; i < sbi->s_groups_count; i++)
 	{
-		if (i == sbi->s_groups_count - 1)
+		if (i == sbi->s_groups_count - 1 || EXT4_HAS_INCOMPAT_FEATURE(sb,
+					EXT4_FEATURE_INCOMPAT_FLEX_BG))
 			last_block = ext4_blocks_count(sbi->s_es) - 1;
 		else
 			last_block = first_block +

@@ -1291,7 +1292,9 @@ static int ext4_check_descriptors (struct super_block * sb)
 				    i, inode_table);
 			return 0;
 		}
-		first_block += EXT4_BLOCKS_PER_GROUP(sb);
+		if (!EXT4_HAS_INCOMPAT_FEATURE(sb,
+					       EXT4_FEATURE_INCOMPAT_FLEX_BG))
+			first_block += EXT4_BLOCKS_PER_GROUP(sb);
 		gdp = (struct ext4_group_desc *)
 			((__u8 *)gdp + EXT4_DESC_SIZE(sb));
 	}

include/linux/ext4_fs.h +3 -1

index cdee7aa..d53e167 100644
@@ -702,13 +702,15 @@ static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
+#define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
 
 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
 					 EXT4_FEATURE_INCOMPAT_META_BG| \
 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
-					 EXT4_FEATURE_INCOMPAT_64BIT)
+					 EXT4_FEATURE_INCOMPAT_64BIT| \
+					 EXT4_FEATURE_INCOMPAT_FLEX_BG)
 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \

Andreas Dilger

2007-09-11 06:04 UTC #1 lore raw

fs/ext4/super.c static int ext4_check_descriptors (struct super_block * sb) · line 1259

Quoted 8 lines · Jose R. Santos

fs/ext4/super.c +2 -1

@@ -1254,7 +1254,8 @@ static int ext4_check_descriptors (struct super_block * sb)
 
        for (i = 0; i < sbi->s_groups_count; i++)
        {
-               if (i == sbi->s_groups_count - 1)
+               if (i == sbi->s_groups_count - 1 || EXT4_HAS_INCOMPAT_FEATURE(sb,
+                                       EXT4_FEATURE_INCOMPAT_FLEX_BG))
                        last_block = ext4_blocks_count(sbi->s_es) - 1;

No need to check this featyre for every group, once at the beginning of the function is enough.

Jose R. Santos

author

2007-09-11 12:27 UTC #2 lore raw

Quoted 13 lines · On Tue, 11 Sep 2007 00:04:43 -0600 "Andreas Dilger" <[email protected]>…

fs/ext4/super.c static int ext4_check_descriptors (struct super_block * sb) · line 1259

Quoted 8 lines · Jose R. Santos

fs/ext4/super.c +2 -1

@@ -1254,7 +1254,8 @@ static int ext4_check_descriptors (struct super_block * sb)
 
        for (i = 0; i < sbi->s_groups_count; i++)
        {
-               if (i == sbi->s_groups_count - 1)
+               if (i == sbi->s_groups_count - 1 || EXT4_HAS_INCOMPAT_FEATURE(sb,
+                                       EXT4_FEATURE_INCOMPAT_FLEX_BG))
                        last_block = ext4_blocks_count(sbi->s_es) - 1;

No need to check this featyre for every group, once at the beginning of the function is enough.

Do you mean something like the original patch?
http://lists.openwall.net/linux-ext4/2007/07/12/20

Wouldn't we need to check all the descriptor for corruption if checksum is not enable on the filesystem?

-JRS

Andreas Dilger

2007-09-11 23:21 UTC #3 lore raw

Quoted 16 lines · Jose R. Santos

Quoted 13 lines · On Tue, 11 Sep 2007 00:04:43 -0600 "Andreas Dilger" <[email protected]>…
fs/ext4/super.c static int ext4_check_descriptors (struct super_block * sb) · line 1259
Quoted 8 lines · Jose R. Santos
fs/ext4/super.c +2 -1
@@ -1254,7 +1254,8 @@ static int ext4_check_descriptors (struct super_block * sb)
 
        for (i = 0; i < sbi->s_groups_count; i++)
        {
-               if (i == sbi->s_groups_count - 1)
+               if (i == sbi->s_groups_count - 1 || EXT4_HAS_INCOMPAT_FEATURE(sb,
+                                       EXT4_FEATURE_INCOMPAT_FLEX_BG))
                        last_block = ext4_blocks_count(sbi->s_es) - 1;
No need to check this featyre for every group, once at the beginning of the function is enough.

Quoted 5 lines

Do you mean something like the original patch?
http://lists.openwall.net/linux-ext4/2007/07/12/20
Wouldn't we need to check all the descriptor for corruption if checksum is not enable on the filesystem?

Yes, I just meant you don't need to have:

EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG))

for each time through the loop. That loop is walked 8000 times per TB at mount, so if we can make it faster we should do so.

Cheers, Andreas
--
Andreas Dilger
Principal Software Engineer
Cluster File Systems, Inc.

Jose R. Santos

author

2007-09-12 00:04 UTC #4 lore raw

Quoted 23 lines · On Tue, 11 Sep 2007 17:21:39 -0600 Andreas Dilger <[email protected]> wrote:

Quoted 16 lines · Jose R. Santos
Quoted 13 lines · On Tue, 11 Sep 2007 00:04:43 -0600 "Andreas Dilger" <[email protected]>…
fs/ext4/super.c static int ext4_check_descriptors (struct super_block * sb) · line 1259
Quoted 8 lines · Jose R. Santos
fs/ext4/super.c +2 -1
@@ -1254,7 +1254,8 @@ static int ext4_check_descriptors (struct super_block * sb)
 
        for (i = 0; i < sbi->s_groups_count; i++)
        {
-               if (i == sbi->s_groups_count - 1)
+               if (i == sbi->s_groups_count - 1 || EXT4_HAS_INCOMPAT_FEATURE(sb,
+                                       EXT4_FEATURE_INCOMPAT_FLEX_BG))
                        last_block = ext4_blocks_count(sbi->s_es) - 1;
No need to check this featyre for every group, once at the beginning of the function is enough.
Quoted 5 lines
Do you mean something like the original patch?
http://lists.openwall.net/linux-ext4/2007/07/12/20
Wouldn't we need to check all the descriptor for corruption if checksum is not enable on the filesystem?

Quoted 6 lines

Yes, I just meant you don't need to have:
EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG))
for each time through the loop. That loop is walked 8000 times per TB at mount, so if we can make it faster we should do so.

Good point, I'll send an updated patch.

-JRS