xfs
[Top] [All Lists]

Re: BUG REPORT: XFS LOG FORWARD COMPATIBILITY PROBLEM

To: Eric Sandeen <sandeen@xxxxxxxxxxx>
Subject: Re: BUG REPORT: XFS LOG FORWARD COMPATIBILITY PROBLEM
From: Eddy Zhao <eddy.y.zhao@xxxxxxxxx>
Date: Wed, 29 Jul 2009 20:12:22 +0800
Cc: Lachlan McIlroy <lmcilroy@xxxxxxxxxx>, xfs@xxxxxxxxxxx
Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:in-reply-to:references :date:message-id:subject:from:to:cc:content-type; bh=Mzws5jNioXgLO0lMULIAofWPpyauOvmGlIoxxYYbID4=; b=RqLw1CGXAq52ttc2b+n2BXneHTh3A0QbbRn2KXCok+TvI81x8SjgraiatsT0mlxLxf MCTP9RmqQpuZ4Tef/svVxXbfLXJ/gHb0gUO6jq8KVOS/qkIJZT+AfK9awDL26yKbJ6cj EyFsJMrmqLHG5YgNIMZZ2m4PseVtHQ3CLI5wY=
Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; b=K8CQkQVkxg3ws5D65iA+kcNZ5hyPEKFEociFE+ka4j9+BX2tvsQeSDUk/9R5e1E23z Z+FMHUh6+LhjlaR3M1qIhrG2RcAf30NLiE6zWbHsg47/CjkT28iq/SOdMAQdSs0n4E2p gq1qtGKC4ararSAmVo/Wkdj3gfwr9YT2xLV7c=
In-reply-to: <4A6F20D0.4050303@xxxxxxxxxxx>
References: <362522677.1010811248671789619.JavaMail.root@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <1938435153.1010831248671852499.JavaMail.root@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <b094161c0907280550y58cde45fja97b2c413c92a8d4@xxxxxxxxxxxxxx> <4A6F0978.4070709@xxxxxxxxxxx> <b094161c0907280759w6c525201vd4a17e22a8c1d582@xxxxxxxxxxxxxx> <4A6F20D0.4050303@xxxxxxxxxxx>
Hello Eric:
 
>
> A tool that can be very helpful here is "pahole" found in the dwarves
> package:
>

Nice tool.

>
> So just to be clear, on 2.6.28, arm eabi logs replay fine on x86_64?
>

I don't have x86_64 machine at hand. Replay arm eabi log on ia32 machine "seems" success (mount "successfully" returns). Actually I guess that's because 2.6.28 XFS running on ia32 misinterprets unpacked xfs_inode_log_format produced by arm as xfs_inode_log_format_64. And there must be something silently goes wrong on disk after replaying the log.

During fixing the problem, I find it is difficult to change the code to commit packed log item (restricted by xfs_log_iovec_t mechanism). If we could somehow change this piece of code to use packed structure. The endianess problem could be solved by add an byte-swap option to xfs_repair (like e2fsck "-s"). Then we solve the xfs-log-cant-migrate-between-archs problem.
 
Thanks
Eddy
<Prev in Thread] Current Thread [Next in Thread>