On Thu, Jun 23, 2011 at 04:12, Andrea Gasparini <gasparini@imavis.com> wrote:

is the issue I found and posted a little ago. I never found the time to
correctly send a description of what I found.

In the old version sometimes the size of the block (to be clear the field:
yaffs_packed_tags2_tags_only.n_bytes) isn't written (don't know if it's a
bug or a feature....) while the new check there's a correct size even if
isn't used.

Mark: could you please try to apply this? *IT'S NOT* the right solution,
but just to understand what your problem is.

diff --git a/fs/yaffs2/yaffs_yaffs2.c b/fs/yaffs2/yaffs_yaffs2.c
index 5761e96..5382e02 100644
--- a/fs/yaffs2/yaffs_yaffs2.c
+++ b/fs/yaffs2/yaffs_yaffs2.c
@@ -1024,8 +1024,7 @@ static inline int yaffs2_scan_chunk(struct yaffs_dev
*dev,
} else if (tags.obj_id > YAFFS_MAX_OBJECT_ID ||
tags.chunk_id > YAFFS_MAX_CHUNK_ID ||
tags.obj_id == YAFFS_OBJECTID_SUMMARY ||
- (tags.chunk_id > 0 &&
- tags.n_bytes > dev->data_bytes_per_chunk) ||
+ //(tags.chunk_id > 0 && tags.n_bytes > dev-
>data_bytes_per_chunk) ||
tags.seq_number != bi->seq_number) {
yaffs_trace(YAFFS_TRACE_SCAN,
"Chunk (%d:%d) with bad tags:obj = %d, chunk_id =
%d, n_bytes = %d, ignored"

Thanks Andrea,

I applied the patch to the new tree and got the same result. I also enabled additional msgs in the yaffs_trace_mask. I created the /mnt/card/testdata in the same manner as before and here is the output from the new yaffs2 fs with the patch:

root@homebase:~# mount -t yaffs2 /dev/mtdblock5 /mnt/card/

yaffs: dev is 32505861 name is "mtdblock5" rw

yaffs: passed flags ""

ixed ECC in chunk(17:38), chunk ignored

packed tags obj 261 chunk 868 byte 2048 seq 4110

packed tags obj 261 chunk 804 byte 2048 seq 4109

packed tags obj 261 chunk 676 byte 2048 seq 4107

21 data (null) tags cee65db0

29 data (null) tags cee65db0

65 data (null) tags cee65db0

root@homebase:~# cat /proc/yaffs

Multi-version YAFFS built:Jun 23 2011 10:32:11

Device 0 "system_a"

start_block.......... 0

end_block............ 785

total_bytes_per_chunk 2048

use_nand_ecc......... 1

no_tags_ecc.......... 0

is_yaffs2............ 1

inband_tags.......... 0

empty_lost_n_found... 1

disable_lazy_load.... 0

refresh_period....... 500

n_caches............. 10

n_reserved_blocks.... 5

always_check_erased.. 0

data_bytes_per_chunk. 2048

chunk_grp_bits....... 0

chunk_grp_size....... 1

n_erased_blocks...... 785

blocks_in_checkpt.... 0

n_tnodes............. 0

n_obj................ 4

n_free_chunks........ 50304

n_page_writes........ 0

n_page_reads......... 64

n_erasures........... 1

n_gc_copies.......... 0

all_gcs.............. 1

passive_gc_count..... 1

oldest_dirty_gc_count 0

n_gc_blocks.......... 0

bg_gcs............... 0

n_retired_writes..... 0

n_retired_blocks..... 0

n_ecc_fixed.......... 0

n_ecc_unfixed........ 0

n_tags_ecc_fixed..... 0

n_tags_ecc_unfixed... 0

cache_hits........... 0

n_deleted_files...... 0

n_unlinked_files..... 1

refresh_count........ 1

n_bg_deletions....... 0

tags_used............ 1536

summary_used......... 0

root@homebase:~# ls -/mnt/card/l/mnt/card/ /mnt/card//mnt/card/

drwx------ 1 root root 2048 Jun 23 15:20 lost+found

root@homebase:~# root@homebase:~# ls -l /mnt/card/losroot@homebase:~# ls -l /mnt/card/lost\+found/

root@homebase:~#

testdata still missing.

Thanks a lot for your help. Would it be helpful to look at a nanddump of the partition written using old yaffs? I can also build the old yaffs2 kernel with the additional trace msgs.

bye
--
Andrea Gasparini
---- ImaVis S.r.l. ----
web: www.imavis.com