From linux-xfs Tue Jan 06 18:04:12 2009 From: "Gergely Soos" Date: Tue, 06 Jan 2009 18:04:12 +0000 To: linux-xfs Subject: Re: need help to repair XFS partition Message-Id: X-MARC-Message: https://marc.info/?l=linux-xfs&m=123126509117829 MIME-Version: 1 Content-Type: multipart/mixed; boundary="--===============4109167692478637782==" --===============4109167692478637782== Content-Type: text/plain Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Content-length: 6783 In my original email there was an attachment that contained the original boot sector, but anyway, here comes the hexdump Eric asked for: 00000000 91 f0 1c 43 90 01 ba bf f7 ee 29 9a 1e 6c d5 aa |.=F0.C..=BA=BF=F7=EE)..l=D5=AA| 00000010 11 5a 12 cb 3b 29 cb ff 39 ce 4e d3 95 ec b9 39 |.Z.=CB;)=CB=FF9=CEN=D3.=EC=B99| 00000020 4a 5f 11 92 0a b5 c8 c0 fd ef 36 11 d7 d4 0a 55 |J_...=B5=C8=C0=FD=EF6.=D7=D4.U| 00000030 f7 f7 54 5e 52 4e 1a 7a 32 72 ad c1 a7 10 2e 28 |=F7=F7T^RN.z2r=AD=C1=A7..(| 00000040 cf 00 fb 5c 95 d3 ea 59 b6 72 51 4e cd 82 b7 7b |=CF.=FB\.=D3=EAY=B6rQN=CD.=B7{| 00000050 3a 13 dd e0 af 01 bd fd 2c 9d b3 a4 ee 59 c3 ea |:.=DD=E0=AF.=BD=FD,.=B3=A4=EEY=C3=EA| 00000060 a7 a1 e5 e9 b0 aa 74 19 14 e4 26 cf 60 3f 4c c9 |=A7=A1=E5=E9=B0=AAt..=E4&=CF`?L=C9| 00000070 51 1f fd 77 df 0e 26 76 43 6b 83 ba e4 f3 6f 56 |Q.=FDw=DF.&vCk.=BA=E4=F3oV| 00000080 7c b2 0a 70 1f 6b a0 a3 27 a6 c5 b2 f7 90 50 e1 ||=B2.p.k =A3'=A6=C5=B2=F7.P=E1| 00000090 e7 7a 2b ab 8a 31 6b 5e 07 b0 48 df e4 cd b0 31 |=E7z+=AB.1k^.=B0H=DF=E4=CD=B01| 000000a0 73 39 7e dd b2 5e 20 15 48 b0 3a bb 40 f0 1b 8b |s9~=DD=B2^ .H=B0:=BB@=F0..| 000000b0 c9 2d a2 bb cf dd bc 91 ee fe fe a8 88 88 92 cc |=C9-=A2=BB=CF=DD=BC.=EE=FE=FE=A8...=CC| 000000c0 99 24 88 08 e5 9c 8b 89 b0 1b df d5 0b d9 d2 79 |.$..=E5...=B0.=DF=D5.=D9=D2y| 000000d0 cb d9 2f 39 d3 b6 c8 74 04 ce 4f 86 d8 50 ea b7 |=CB=D9/9=D3=B6=C8t.=CEO.=D8P=EA=B7| 000000e0 2a c4 d8 8c 11 85 ef 5b 31 2f f2 fa 4e d7 e9 92 |*=C4=D8...=EF[1/=F2=FAN=D7=E9.| 000000f0 ea 06 d5 25 71 d4 22 05 5e dd c8 64 43 02 18 a4 |=EA.=D5%q=D4".^=DD=C8dC..=A4| 00000100 76 01 01 0b fe fc ba d5 74 55 54 c8 bb cd 08 80 |v...=FE=FC=BA=D5tUT=C8=BB=CD..| 00000110 1a 0e 4f d7 17 51 da b5 5e 84 f4 5c 48 b8 63 66 |..O=D7.Q=DA=B5^.=F4\H=B8cf| 00000120 05 19 9c 7b c4 70 3b 37 a7 8c 45 43 ef a8 05 0b |...{=C4p;7=A7.EC=EF=A8..| 00000130 52 d3 cc a4 a7 1e 07 c1 2e 9e 62 0e d1 43 83 4c |R=D3=CC=A4=A7..=C1..b.=D1C.L| 00000140 ab bd e8 c0 e2 72 75 1f 21 b9 08 57 92 a2 e8 b8 |=AB=BD=E8=C0=E2ru.!=B9.W.=A2=E8=B8| 00000150 ee 34 5c 0d 01 55 5e fc 7a a9 07 67 c4 20 48 07 |=EE4\..U^=FCz(= c).g=C4 H.| 00000160 aa 9f 96 d2 49 fd c0 39 35 e1 03 2a ae 93 d8 75 |=AA..=D2I=FD=C095=E1.*(R).=D8u| 00000170 28 0c 00 59 78 e1 28 10 dc c9 10 8c 03 00 24 4a |(..Yx=E1(.=DC=C9....$J| 00000180 04 34 cf 1f 01 f4 cf 12 95 88 ee 39 f6 8c 05 30 |.4=CF..=F4=CF...=EE9=F6..0| 00000190 08 00 92 e0 3e e1 f0 21 83 12 70 93 90 9a 9c 3e |...=E0>=E1=F0!..p....>| 000001a0 23 80 1c 7c 07 f4 c0 f8 10 e1 30 94 5f 8a 43 28 |#..|.=F4=C0=F8.=E10._.C(| 000001b0 78 33 4e 00 52 ef 2b 53 c3 db c3 ae 12 3e 07 6a |x3N.R=EF+S=C3=DB=C3(R).>.j| 000001c0 87 2e c0 16 43 a2 52 ab 65 50 61 d5 f7 df 7d eb |..=C0.C=A2R=ABePa=D5=F7=DF}=EB| 000001d0 7b 57 74 78 0f e9 d2 47 fb 17 b0 33 7b 1a e0 2c |{Wtx.=E9=D2G=FB.=B03{.=E0,| 000001e0 a0 24 b6 09 b1 4b 03 42 56 0f c9 15 2a f6 11 bc | $=B6.=B1K.BV.=C9.*=F6.=BC| 000001f0 06 c1 30 d4 0f 7f 05 2a 0e 34 08 85 2f 54 b4 dd |.=C10=D4...*.4../T=B4=DD| 00000200 This looks like nothing to me... xfs_repair rejects all superblock candidates and exits saying something like: Sorry, cannot find valid secondary superblock. I'm not sure what a GPT is, but this is an IDE harddisk, I'm using kernel 2.6.20 and my xfs partition is /dev/hdd1 Is there any way xfs_repair would accept the superblock as is and move on with the repairs? Gergely On Tue, Jan 6, 2009 at 6:06 PM, Eric Sandeen wrote: > Gergely Soos wrote: > > Greetings > > > > I heard that people compliment XFS for being the fastest FS these days > > so I thought I give it a try. > > What I did was that I converted my old 107GB FAT32 partition to XFS (I > > created an XFS over it) and I wanted to use it as /home > > Everything went as it should, I moved the files from my old /home then I > > restarted (cleanly!) under windows. > > When I got back from windows I was no longer able to mount the XFS > > partition. > > Which partition was it? I wonder if a gpt label got re-written over it > or something. > > > xfs_repair only said that the superblock is invalid and attempts to find > > secondary superblocks but all candidates fail. > > So I created a file on another computer with the exact same size as the > > partition on my computer and created an XFS on it. > > > Then I extracted the first 512 bytes and copied to my HDD using dd. But > > it complained about the UUID so I corrected it using khexedit. > > Did you keep the original first 512 bytes? It would be worth looking > at; try hexdump -C on it (the original) and send that if you have it. > > > Then xfs_db complained that it cannot find the root inode so I wrote a > > little perl script to search it. > > You can find it in myxfs.tar.bz2. It seems that it found the root inode > > because after I corrected the address using khexedit > > xfs_db started complaining about the realtime bitmap inodes, so I > > corrected those too (using inode numbers rootino+1 and rootino+2). > > And this is where I'm stuck. > > The mount still fails, dmesg says that it cannot find the root inode, > > xfs_repair says the same old thing, xfs_check says a lot of things, > > most of them are messages like this: can't seek in filesystem at bb > > or: ag bad magick number. > > That exceeds my knowledge. > > Please help me restore the partition, It contains lots of programmes I > > wrote over the years and some of them are not backed up, > > some of them are, but only an older version. I also have personal stuff > > on it like my bill records that I cannot recover from anywhere else. > > myxfs.tar.bz2 contains the perl script I mentioned above and the inodes > > it found. The one staring with ri is the root inode, > > the files starting with sb are the suspected superblocks (I also took > > one of those at random and corrected the three inode numbers > > with khexedit but the result is the same) and the file backup_bootsect > > contains the original superblock that was on the disk after I came back > > from windows xp. It does not look like anything to me... > > The numbers after ri and sb are the starting address in bytes of the > > inode on the disk. > > I would also attach the output of xfs_metadump which I created after I > > corrected the three inode numbers in the superblock but even bzip-ed > > it is more than 3MB and the administrator of this list does not accept > it. > > As Chris suggested, I would find a spare 100G somewhere and make a dd > image now, so you have something to experiment with + a backup if things > go wrong. > > It sounds like something under windows clobbered your filesystem, I > guess, unless it's the gpt thing I mentioned, it's very strange. > > -Eric > [[HTML alternate version deleted]] --===============4109167692478637782== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ xfs mailing list xfs@oss.sgi.com http://oss.sgi.com/mailman/listinfo/xfs --===============4109167692478637782==--