[Top] [All Lists]

Re: corruption, xfs_repair 3.1.4 segfaults

To: xfs@xxxxxxxxxxx
Subject: Re: corruption, xfs_repair 3.1.4 segfaults
From: Marc Lehmann <schmorp@xxxxxxxxxx>
Date: Fri, 4 Mar 2011 11:29:39 +0100
In-reply-to: <20110304085146.368046e7@xxxxxxxxxxxxxxx>
References: <20110302175818.GA8290@xxxxxxxxxx> <20110302224329.3f62c172@xxxxxxxxxxxxxxx> <20110304071123.GC2316@xxxxxxxxxx> <20110304085146.368046e7@xxxxxxxxxxxxxxx>
On Fri, Mar 04, 2011 at 08:51:46AM +0100, Emmanuel Florac 
<eflorac@xxxxxxxxxxxxxx> wrote:
> Ah, I think the problem may lie in the loop device. Try to run

At least the xfs_repair problem cannot be in the loop device:

From the strace it's obvious that xfs_repair tries to read close to 2**64
bytes, and then crashes when the kernel rightly says that it can't do that.

It also shows that xfs_repair tries to allocate 3gb of memory (which is in
addition to the 1gb it already uses at that point), which is far more then
it should (specifying -m 990 didn't change that), which is another bug in

I think that, no matter what the loop device would do, xfs_repair is buggy
- it simply shouldn't crash, no matter how corrupted the filesystem is.

As a sidenote, I am now about 30% in recovering (copying) and
verifying the data, and it seems the volume isn't corrupted completely
(fortunately), so I can probably recover the important stuff, and reformat
the partition, so this might not turn out to be data loss (fortunately :).

> xfs_repair -f /file/path 
> (not using the loop device).

that will not work, as xfs_repair has no encryption support (which is why
the loop device is used in the first place).

                The choice of a       Deliantra, the free code+content MORPG
      -----==-     _GNU_              http://www.deliantra.net
      ----==-- _       generation
      ---==---(_)__  __ ____  __      Marc Lehmann
      --==---/ / _ \/ // /\ \/ /      schmorp@xxxxxxxxxx
      -=====/_/_//_/\_,_/ /_/\_\

<Prev in Thread] Current Thread [Next in Thread>