[U-Boot] [PATCH] EXT2FS: fix inode size for ext2fs rev#0

From: Michael Brandt Michael.Brandt@emsyso.de
extfs.c assumes that there is always a valid inode_size field in the superblock. But this is not true for ext2fs rev 0. Such ext2fs images are for instance generated by genext2fs. Symptoms on ARM machines are messages like: "raise: Signal # 8 caught" This fix checks for rev 0 and uses then 128 bytes as inode size.
Signed-off-by: Michael Brandt Michael.Brandt@emsyso.de --- fs/ext2/ext2fs.c | 20 +++++++++++++------- 1 files changed, 13 insertions(+), 7 deletions(-)
diff --git a/fs/ext2/ext2fs.c b/fs/ext2/ext2fs.c index d54f60b..4b391d6 100644 --- a/fs/ext2/ext2fs.c +++ b/fs/ext2/ext2fs.c @@ -178,6 +178,7 @@ 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 @@ -212,7 +213,7 @@ static int ext2fs_read_inode unsigned int blkoff;
#ifdef DEBUG - printf ("ext2fs read inode %d\n", ino); + 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--; @@ -222,16 +223,12 @@ static int ext2fs_read_inode return (0); }
- inodes_per_block = EXT2_BLOCK_SIZE(data) / __le16_to_cpu(sblock->inode_size); - -#ifdef DEBUG - printf ("ext2fs read inode blkno %d blkoff %d\n", blkno, blkoff); -#endif + 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) * __le16_to_cpu (sblock->inode_size); + blkoff = (ino % inodes_per_block) * inode_size; #ifdef DEBUG printf ("ext2fs read inode blkno %d blkoff %d\n", blkno, blkoff); #endif @@ -863,6 +860,15 @@ int ext2fs_mount (unsigned part_length) { 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;

I tested this patch and it resolved the problems I was having with EXT2 filesystems on my SD card.
On Sun, Nov 22, 2009 at 7:13 AM, Michael Brandt Michael.Brandt@emsyso.de wrote:
From: Michael Brandt Michael.Brandt@emsyso.de
extfs.c assumes that there is always a valid inode_size field in the superblock. But this is not true for ext2fs rev 0. Such ext2fs images are for instance generated by genext2fs. Symptoms on ARM machines are messages like: "raise: Signal # 8 caught" This fix checks for rev 0 and uses then 128 bytes as inode size.
Signed-off-by: Michael Brandt Michael.Brandt@emsyso.de
Thanks! Dan

Dear Michael Brandt,
In message 4B0938F7.8080402@emsyso.de you wrote:
From: Michael Brandt Michael.Brandt@emsyso.de
extfs.c assumes that there is always a valid inode_size field in the superblock. But this is not true for ext2fs rev 0. Such ext2fs images are for instance generated by genext2fs. Symptoms on ARM machines are messages like: "raise: Signal # 8 caught" This fix checks for rev 0 and uses then 128 bytes as inode size.
Signed-off-by: Michael Brandt Michael.Brandt@emsyso.de
fs/ext2/ext2fs.c | 20 +++++++++++++------- 1 files changed, 13 insertions(+), 7 deletions(-)
Tested on TQM5200S.
Applied to "master", thanks.
Best regards,
Wolfgang Denk
participants (3)
-
Dan Christensen
-
Michael Brandt
-
Wolfgang Denk