Hi,
I am not sure if it has already been said, but before trying any
recovery, I would recommend to do a dd of the device to a file and would
also recommend to do recovery tries on a copy of this file mounted as a
loop device, this way you can make sure not to destroy more data by the
recovery tries. I would also recommend passing the -i parameter to the
mount command during recovery tries in order to prevent garbage
collection from running.
Bye,
David Arendt
Jan de Kruyf wrote:
> Hallo,
> here we go again. As a matter of interest, what version of nilfs and
> what distribution are you running? And what processor?
> your endiannes confused me at first.
>
> glad you fixed your hexedit
>
> I have looked at some numbers in the superblock and they look ok. I do
> wonder if you have the garbage collector running.
>
> now for the second superblock. Please pay attention to the exact place
> of it. Because it is of vital
> importance if it in in partition 1 or in the root of the disk.
>
> the first superblock sits as you observed in the root. The flags say
> amongst others that it was unmounted cleanly but errors were detected.
>
> see nilfs2_fs.h - NILFS2 on-disk structures and common declarations.
> in the distribution.
> /**
> * struct nilfs_super_block - structure of super block on disk
> */
> struct nilfs_super_block {
> ....
> translates bit for bit to what you see written in the super block on disk.
>
> Here is an example from a partition of mine on how to discover the
> superblock copy
> it should read the same as the 1st but in your case it might not.
> It is a leftshift - subtract 1 - right shift algorithm.
> go in hexedit to the last data with ">" (shift .)
> note the address of the last byte (the size of the partion)
> in mine:
> 6566B3F0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> ................
> or the status line might give it
> the address of the copy is now at 6566A000
> i.e. 6566B has 1 subtracted from it and the 3 least significant digits
> have been zeroed.
>
> so please dump yours, see if the algorithm works on the 1st part or on
> the root or both,
> so we know where it is.
> And check if it is exactly the same as the first one you send me
> "0000400 0002 0000 0000 3434 0100 0000 b209 5b31"
> etc.
>
> the next thing to discover is where the start of (nilfs)segment 0 is.
> I am not quite sure what is written in it, but the signature is quite
> distinct.
> This is what it looks on my machine:
>
> 00000FC0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> ................
> 00000FD0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> ................
> 00000FE0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> ................
> 00000FF0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> ................
> 00001000 C4 1B 47 5B 51 62 19 13 11 FA AF 1E 38 00 10 00
> ..G[Qb......8...
> 00001010 FB CE 01 00 00 00 00 00 EE BD F1 4A 00 00 00 00
> ...........J....
> 00001020 00 08 00 00 00 00 00 00 FF 07 00 00 32 00 00 00
> ............2...
> 00001030 A0 83 00 00 00 00 00 00 EE 0D 00 00 00 00 00 00
> ................
> 00001040 07 00 00 00 00 00 00 00 7B 00 00 00 7B 00 00 00
> ........{...{...
> 00001050 EA 9E 07 00 00 00 00 00 F8 01 00 00 00 00 00 00
> ................
> 00001060 EB 9E 07 00 00 00 00 00 F9 01 00 00 00 00 00 00
> ................
> 00001070 EC 9E 07 00 00 00 00 00 FA 01 00 00 00 00 00 00
> ................
> 00001080 ED 9E 07 00 00 00 00 00 FB 01 00 00 00 00 00 00
> ................
> 00001090 EE 9E 07 00 00 00 00 00 FC 01 00 00 00 00 00 00
> ................
> 000010A0 EF 9E 07 00 00 00 00 00 FD 01 00 00 00 00 00 00
> ................
> 000010B0 F0 9E 07 00 00 00 00 00 FE 01 00 00 00 00 00 00
> ................
> 000010C0 F2 9E 07 00 00 00 00 00 FF 01 00 00 00 00 00 00
> ................
> 000010D0 F3 9E 07 00 00 00 00 00 00 02 00 00 00 00 00 00
> ................
> 000010E0 F4 9E 07 00 00 00 00 00 01 02 00 00 00 00 00 00
> ................
> 000010F0 F5 9E 07 00 00 00 00 00 02 02 00 00 00 00 00 00
> ................
> 00001100 F6 9E 07 00 00 00 00 00 03 02 00 00 00 00 00 00
> ................
> 00001110 F7 9E 07 00 00 00 00 00 04 02 00 00 00 00 00 00
> ................
> 00001120 F8 9E 07 00 00 00 00 00 05 02 00 00 00 00 00 00
> ................
> 00001130 F9 9E 07 00 00 00 00 00 06 02 00 00 00 00 00 00
> ................
> 00001140 FA 9E 07 00 00 00 00 00 07 02 00 00 00 00 00 00
> ................
> 00001150 FB 9E 07 00 00 00 00 00 08 02 00 00 00 00 00 00
> ................
> 00001160 FC 9E 07 00 00 00 00 00 09 02 00 00 00 00 00 00
> ................
> 00001170 FD 9E 07 00 00 00 00 00 0A 02 00 00 00 00 00 00
> ................
> 00001180 FE 9E 07 00 00 00 00 00 0B 02 00 00 00 00 00 00
> ................
> 00001190 FF 9E 07 00 00 00 00 00 0C 02 00 00 00 00 00 00
> ................
> 000011A0 00 9F 07 00 00 00 00 00 0D 02 00 00 00 00 00 00
> ................
> 000011B0 01 9F 07 00 00 00 00 00 0E 02 00 00 00 00 00 00
> ................
> 000011C0 02 9F 07 00 00 00 00 00 0F 02 00 00 00 00 00 00
> ................
> 000011D0 03 9F 07 00 00 00 00 00 10 02 00 00 00 00 00 00
> ................
> 000011E0 04 9F 07 00 00 00 00 00 11 02 00 00 00 00 00 00
> ................
>
> Segment 0 starts at hex 1000 of a nilfs partition as you can see above
> and carries on for quite a while like this.
>
> so have a look on your disk if it sits at 1000 of the root partition
> or at 1000 of partition 1.
>
> Once we have these things sorted I would say that we are ready to
> plant the _right_ superblock of the 2
> in the right place and see if the partition is recoqnized by nilfs.
> Off course we will save the place where we are going to plant that
> block first.
>
> And now back to the birthday party . . . .
>
> Cheers,
>
> Jan
>
>
>
> On Wed, Nov 4, 2009 at 4:23 AM, Paul L <[email protected]
> <mailto:[email protected]>> wrote:
>
> On 11/3/09, Jan de Kruyf <[email protected]
> <mailto:[email protected]>> wrote:
> > 26 august 98: hexedit 0.9.5 release
> > september 2005:
> > - version 1.2.12 this is the one I am running.
>
> Ah, I must be using the wrong hexedit.. now I've installed the
> same one you use.
>
> > did I hear that you have always had /dev/mmcblk0p1 in fstab??
>
> Yes. What I put there is:
>
> /dev/mmcblk0p1 /home nilfs2 defaults 1 1
>
> > hd -n1536 /dev/mmcblk0 >part.start
> > hd -s8191 -n1536 /dev/mmcblk0 >part1.start
> > an to verify the last line:
> > hd -n1536 /dev/mmcblk0p1 >part1a.start
>
> Here is the output (I use hexdump instead of hd, hopefully they
> are the same)
>
> bash-3.1# cat part.start
> 0000000 0000 0000 0000 0000 0000 0000 0000 0000
> *
> 00001b0 0000 0000 0000 0000 0696 6c22 0000 0100
> 00001c0 0001 0383 ffd0 0010 0000 dff0 01eb 0000
> 00001d0 0000 0000 0000 0000 0000 0000 0000 0000
> *
> 00001f0 0000 0000 0000 0000 0000 0000 0000 aa55
> 0000200 0000 0000 0000 0000 0000 0000 0000 0000
> *
> 0000400 0002 0000 0000 3434 0100 0000 b209 5b31
> 0000410 1183 794c 0002 0000 07af 0000 0000 0000
> 0000420 0000 d780 0003 0000 0001 0000 0000 0000
> 0000430 0800 0000 0005 0000 090b 000e 0000 0000
> 0000440 0710 0035 0000 0000 8502 0008 0000 0000
> 0000450 8800 000b 0000 0000 93c1 493c 0000 0000
> 0000460 013f 4ae2 0000 0000 2a8f 4aef 0000 0000
> 0000470 00a2 0032 0003 0001 93c1 493c 0000 0000
> 0000480 4e00 00ed 0000 0000 0000 0000 000b 0000
> 0000490 0080 0020 00c0 0010 ed2b 04f5 41cb ae48
> 00004a0 7f8a 4849 71ec e7f5 0000 0000 0000 0000
> 00004b0 0000 0000 0000 0000 0000 0000 0000 0000
> *
> 0000600
> bash-3.1# cat part1.start
> 0001fff 0000 0000 0000 0000 0000 0000 0000 0000
> *
> 00025ff
> bash-3.1# cat part1a.start
> 0000000 0000 0000 0000 0000 0000 0000 0000 0000
> *
> 0000600
>
> Regards,
> Paul Liu
>
> > On Tue, Nov 3, 2009 at 9:48 PM, Paul L <[email protected]
> <mailto:[email protected]>> wrote:
> >
> >> Thanks a lot for the instructions! I'm attaching the mbr and
> partition
> >> table with this email.
> >>
> >> I'm pretty sure I had 1 partition on the card, since my /etc/fstab
> >> mounts mmcblk0p1.
> >>
> >> I think something corrupted my disk first, and then what I've
> done to
> >> the disk after noticing the corruption:
> >>
> >> 1. fdisk, it says use "w" will correct the error, so I did. But
> then
> >> the one parition is gone.
> >> 2. fdisk again, create a single partition, then "w"
> >>
> >> My mistake was that I didn't create a backup copy of the MBR. A
> hard
> >> lesson learned :(
> >>
> >> Also, why is that my hexedit doesn't take the "-s" option? It's
> >> version 0.9.7, and can't edit bigger than 4.2GB.
> >>
> >> My SD card is A-DATA brand, class 6, and 16GB.
> >>
> >> I'm using Linux, and fdisk version (util-linux-ng 2.14.1)
> >>
> >> Regards,
> >> Paul Liu
> >>
> >> On 11/3/09, Jan de Kruyf <[email protected]
> <mailto:[email protected]>> wrote:
> >> > hallo,
> >> > Almost sounds like you had only the root master-boot-record
> /dev/mmcblk0
> >> > before and now you have added 1 main partition /dev/mmcblk0p1.
> >> >
> >> > If (and only if) that is the case we have to undo the the 1st
> partition
> >> > +
> >> > check that no nilfs is overwritten.
> >> > and I would have to urgently study fdisk to see exactly what
> it writes
> >> when
> >> > and where.
> >> > The last time I did tricks like these is quit a few years ago.
> >> >
> >> > It is the Linux version of fdisk is it??
> >> >
> >> > So here is the plan of action:
> >> > hexdump the master boot record to file.
> >> > like this:
> >> >
> >> > dd if=/dev/mmcblk0 of=backup-mmcblk0.mbr count=1 bs=512
> >> >
> >> > then dump any partitions of the device in a format useful as
> input to
> >> > sfdisk. For example,
> >> > % sfdisk -d /dev/mmcblk0 > mmcblk0 .out
> >> > sfdisk is a tool provided with the util-linux
> >> > package<http://www.kernel.org/pub/linux/utils/util-linux/>.
> >> >
> >> >
> >> > or you could use hexdump to get machine readable or man
> readable images.
> >> > Here is the man readable version:
> >> > hd -n512 /dev/mmcblk0 > backup-mmcblk0.mbr
> >> > hd -n512 /dev/mmcblk0p1 >backup-mmcblk0p1.mbr
> >> > etc.
> >> >
> >> > By the way boor records always end with '55 AA'.
> >> >
> >> > Keep your files in a safe place in case we mess something we
> can at
> >> > least
> >> go
> >> > back to the present situation.
> >> > If you could dd the whole drive to a file, now that would be
> magic
> >> indeed!
> >> > but you must have the space on a harddrive.
> >> > count=... is the number of sectors in the above line (dd ...)
> that you
> >> dump
> >> > to file.
> >> > Hexedit will tell you the number of sectors is you start it
> with -s
> >> option
> >> > and then go to the last sector.
> >> > DONT stop hexedit with control-x use cntl-c.
> >> > DONT use high level or even midlevel tools on a stuffed disk, it
> >> > normally
> >> > messes more than it solves.
> >> > unless of course you really,really know what you are doing.
> >> > Fiddling the bytes is in general safe and gives results, if a
> man keeps
> >> > a
> >> > cool head.
> >> >
> >> > Please send me the fdisk version, the size of the card, and
> the mbr dump
> >> to
> >> > feast my eyes on.
> >> >
> >> > cheers,
> >> >
> >> > Jan de kruyf.
> >> >
> >> >
> >> >
> >> > On Tue, Nov 3, 2009 at 1:27 AM, Paul L <[email protected]
> <mailto:[email protected]>> wrote:
> >> >
> >> >> just want to add that I've always been using 1 partition on this
> >> >> device, it's actually /dev/mmcblk0p1. But hexedit /dev/mmcblk0p1
> >> >> doesn't show that 34 34 at line begining with 0x400, only
> hexedit
> >> >> /dev/mmcblk0 shows it. Not sure if this is a problem.
> >> >>
> >> >> Any help is greatly appreciated!
> >> >>
> >> >>
> >> >> On 11/2/09, Paul L <[email protected]
> <mailto:[email protected]>> wrote:
> >> >> > Thanks for the tips. When I first used the SD card, I used
> fdisk to
> >> >> > create the partition.
> >> >> >
> >> >> > The device is /dev/mmcblk0, and hexedit -d -s /dev/mmcblk0
> shows that
> >> >> > at the line 0x0400, it is indeed 34 34. What should I do then?
> >> >> >
> >> >> > I tried gparted, but apparently it has no support for nilfs2.
> >> >> >
> >> >> > Regards,
> >> >> > Paul Liu
> >> >> >
> >> >> > On 11/2/09, Jan de Kruyf <[email protected]
> <mailto:[email protected]>> wrote:
> >> >> >> Did you first format this card with fdisk?
> >> >> >> did you give it the exact same info this time around?
> >> >> >>
> >> >> >> Can you read /dev/'sdcard' ? (sdcard being the device in
> the dev
> >> >> >> directory
> >> >> >> where the card lives)
> >> >> >>
> >> >> >> If yes can you run hexedit -s /dev/sdcard1 in a terminal
> as root?
> >> >> >> and go to address 0400 - 04B0 to see if nilfs still exists?
> >> >> >>
> >> >> >> be very careful no to save any data in hexedit, it will
> definitely
> >> and
> >> >> >> finally
> >> >> >> destoy your data.
> >> >> >>
> >> >> >> 0400 looks vagely like this:
> >> >> >> 00000400 02 00 00 00 00 00 34 34 00 01 00 00 D3 56 F0 B9
> >> >> >> ......44.....V..
> >> >> >> 00000410 39 BF D9 73 02 00 00 00 CA 02 00 00 00 00 00 00
> >> >> >> 9..s............
> >> >> >> 00000420 00 B4 66 65 01 00 00 00 01 00 00 00 00 00 00 00
> >> >> >> ..fe............
> >> >> >> 00000430 00 08 00 00 05 00 00 00 01 00 00 00 00 00 00 00
> >> >> >> ................
> >> >> >> 00000440 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> >> >> >> ................
> >> >> >> 00000450 00 48 16 00 00 00 00 00 73 95 DC 4A 00 00 00 00
> >> >> >> .H......s..J....
> >> >> >> 00000460 00 00 00 00 00 00 00 00 73 95 DC 4A 00 00 00 00
> >> >> >> ........s..J....
> >> >> >> 00000470 00 00 32 00 01 00 01 00 73 95 DC 4A 00 00 00 00
> >> >> >> ..2.....s..J....
> >> >> >> 00000480 00 4E ED 00 00 00 00 00 00 00 00 00 0B 00 00 00
> >> >> >> .N..............
> >> >> >> 00000490 80 00 20 00 C0 00 10 00 4C 73 DD 3D 01 EC 45
> 85 ..
> >> >> >> .....Ls.=..E.
> >> >> >> 000004A0 94 28 44 42 3D F6 EF EC 56 61 72 36 34 00 00 00
> >> >> >> .(DB=...Var64...
> >> >> >> 000004B0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> >> >> >> ................
> >> >> >> 000004C0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> >> >> >> ................
> >> >> >>
> >> >> >> the 34 34 in the top line say this is nilfs.
> >> >> >>
> >> >> >>
> >> >> >> cheers
> >> >> >>
> >> >> >> jan de kruyf
> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >>
> >> >> >> On Mon, Nov 2, 2009 at 9:06 PM, Paul L <[email protected]
> <mailto:[email protected]>> wrote:
> >> >> >>
> >> >> >>> Due to some careless handling of my laptop, the SD card
> popped out
> >> >> >>> when the machine is still running. When I put it back in
> and reboot
> >> >> >>> the machine, it says "partition table error". I then ran
> fdisk and
> >> >> >>> recreated the single partition. Then I can no longer
> mount the
> >> nilfs2
> >> >> >>> partition that was on the SD card!
> >> >> >>>
> >> >> >>> Can any one help to me recover the file system? I
> believe all data
> >> are
> >> >> >>> still there, but just some bits and pieces are missing
> for the
> >> >> >>> mount
> >> >> >>> to work. Any help is greatly appreciated!
> >> >> >>>
> >> >> >>> PS: I've a deadline to meet in 4 hours, not sure if I
> can get my
> >> stuff
> >> >> >>> back in time... so help please!
> >> >> >>>
> >> >> >>> --
> >> >> >>> Regards,
> >> >> >>> Paul Liu
> >> >> >>>
> >> >> >>> Yale Haskell Group
> >> >> >>> http://www.haskell.org/yale
> >> >> >>> _______________________________________________
> >> >> >>> users mailing list
> >> >> >>> [email protected] <mailto:[email protected]>
> >> >> >>> https://www.nilfs.org/mailman/listinfo/users
> >> >> >>>
> >> >> >>
> >> >> >
> >> >> >
> >> >> > --
> >> >> > Regards,
> >> >> > Paul Liu
> >> >> >
> >> >> > Yale Haskell Group
> >> >> > http://www.haskell.org/yale
> >> >> >
> >> >>
> >> >>
> >> >> --
> >> >> Regards,
> >> >> Paul Liu
> >> >>
> >> >> Yale Haskell Group
> >> >> http://www.haskell.org/yale
> >> >> _______________________________________________
> >> >> users mailing list
> >> >> [email protected] <mailto:[email protected]>
> >> >> https://www.nilfs.org/mailman/listinfo/users
> >> >>
> >> >
> >>
> >>
> >> --
> >> Regards,
> >> Paul Liu
> >>
> >> Yale Haskell Group
> >> http://www.haskell.org/yale
> >>
> >> _______________________________________________
> >> users mailing list
> >> [email protected] <mailto:[email protected]>
> >> https://www.nilfs.org/mailman/listinfo/users
> >>
> >>
> >
>
>
> --
> Regards,
> Paul Liu
>
> Yale Haskell Group
> http://www.haskell.org/yale
> _______________________________________________
> users mailing list
> [email protected] <mailto:[email protected]>
> https://www.nilfs.org/mailman/listinfo/users
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> users mailing list
> [email protected]
> https://www.nilfs.org/mailman/listinfo/users
>
_______________________________________________
users mailing list
[email protected]
https://www.nilfs.org/mailman/listinfo/users
