xfs
[Top] [All Lists]

Re: xfsprogs: is it one issue?

To: Zhi Yong Wu <zwu.kernel@xxxxxxxxx>
Subject: Re: xfsprogs: is it one issue?
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Mon, 27 May 2013 09:11:22 +1000
Cc: "Michael L. Semon" <mlsemon35@xxxxxxxxx>, xfs@xxxxxxxxxxx
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <CAEH94Lj4_8PBbhdD2f-ZFG=NxgS2uBkGUfaQvct71H4X7giPZA@xxxxxxxxxxxxxx>
References: <CAEH94LjnVn-uD6cfwOcChC4wq1PppcD8BN30F93SD=YjdTbbuw@xxxxxxxxxxxxxx> <51A1144A.4020600@xxxxxxxxx> <CAEH94LgWQcjSMenoT3a81pLWbw2pEYWpKFyGwoxAvDX2FTwbXA@xxxxxxxxxxxxxx> <CAEH94Lj4_8PBbhdD2f-ZFG=NxgS2uBkGUfaQvct71H4X7giPZA@xxxxxxxxxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Sun, May 26, 2013 at 07:11:03AM +0800, Zhi Yong Wu wrote:
> I remove the lines related to h_chksum, then it can work well, so i
> guess that i should send out one patch to fix it.

While it will make the code in loggen work the same (it just writes
0 to the field but the buffer is already zeroed) as it does now,
it doesn't necessarily mean it works correctly. The current kernels
validate the CRC and throw warnings if the CRC isn't correct, and
CRC enabled kernels will simply refuse to mount.

As it is, my test systems don't build loggen because
HAVE_XLOG_ASSIGN_LSN isn't set by configure. It only gets set if
there's an xfs/libxfs.h header file found in an include path. My
guess is that you've run 'make install-qa' in the xfsprogs package
at some time in the past, and that's why loggen is trying to be
built.

So Michael's quick hack to remove it just to make it compile at this
point is probably appropriate...

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx

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