> Nikita Danilov wrote:

>> My apology for everyone for missing this bit in the READ.ME:
>>
>> you have to re-create your file-systems with mkfs supplied with this
>> snapshot.
>>

I wrote:
> Is it really mandatory to re-create or will a "fsck.reiser4 --build-fs" 
> be able to fix the filesystem to the new format ? 
> 

(replying to myself) It seems indeed really mandatory... I got a kernel 
oops while fsck-ing my old reiser4 partition with the new 
kernel+userspace snapshot (kernel is 2.6.2-rc1-mm1, reiser4 snapshot is 
2004.01.19.fixed). And now the (unfixed/corrupted) partition causes the 
following oops when I try to mount it:

reiser4[mount(177)]: get_ready_format40 
(fs/reiser4/plugin/disk_format/disk_format40.c:248)[nikita-3228]:
WARNING: Key format mismatch. Only large keys are supported.
Unable to handle kernel NULL pointer dereference at virtual address 00000000
  printing eip:
c01b762d
*pde = 00000000
Oops: 0000 [#1]
PREEMPT
CPU:    0
EIP:    0060:[<c01b762d>]    Not tainted VLI
EFLAGS: 00010246
EIP is at unhash_unformatted_node_nolock+0x1d/0x70
eax: 00000000   ebx: decea7c0   ecx: 00000000   edx: 00000000
esi: dfde6000   edi: 00000000   ebp: c14d0418   esp: dfde7c38
ds: 007b   es: 007b   ss: 0068
Process mount (pid: 177, threadinfo=dfde6000 task=dfdfe6b0)
Stack: dfde6000 dfde6000 c01b769d decea7c0 decea7c0 c01d9437 decea7c0 
decea7c0
        00000000 00000000 00000000 00000000 00000000 00000000 00000000 
00000000
        00000000 00000000 00000000 00000000 00000000 00000000 00000000 
00000000
Call Trace:
  [<c01b769d>] unhash_unformatted_jnode+0x1d/0x40
  [<c01d9437>] reiser4_invalidatepage+0xd7/0x160
  [<c0143ba7>] do_invalidatepage+0x27/0x30
  [<c0143c2b>] truncate_complete_page+0x7b/0x80
  [<c0143e00>] truncate_inode_pages+0x100/0x3c0
  [<c01704ef>] wake_up_inode+0xf/0x30
  [<c01769bb>] write_inode_now+0x4b/0x90
  [<c017000b>] generic_forget_inode+0x14b/0x180
  [<c01700c2>] iput+0x62/0x80
  [<c01cf6b9>] done_formatted_fake+0x59/0x80
  [<c01ddabe>] done_super+0x1e/0x30
  [<c01ddb3a>] reiser4_fill_super+0x6a/0x70
  [<c015d314>] get_sb_bdev+0x124/0x160
  [<c01d712f>] reiser4_get_sb+0x2f/0x40
  [<c01ddad0>] reiser4_fill_super+0x0/0x70
  [<c015d57f>] do_kern_mount+0x5f/0xe0
  [<c0172e08>] do_add_mount+0x78/0x150
  [<c01730f4>] do_mount+0x124/0x170
  [<c0172f60>] copy_mount_options+0x80/0xf0
  [<c01734af>] sys_mount+0xbf/0x140
  [<c02f5ebe>] sysenter_past_esp+0x43/0x65

Code: e9 69 fe ff ff 89 f6 8d bc 27 00 00 00 00 53 83 ec 04 8b 5c 24 0c 
8b 4b 28 8b 53 14 03 53 0c 8b 41 40 48 21 c2 8b 41 3c 8d 0c 90 <8b> 01 
85 c0 74 14 8d 50 1c 0f 18 02 90 39 d8 74 32 8b 40 1c 89
  <6>note: mount[177] exited with preempt_count 2

I also attach the output of debugfs.reiser4 -t in case it could help 
(this is still a bug that mounting a corrupted partition can cause a 
kernel oops)...