[Top] [All Lists]

Re: [PATCH 1/2] xfsrestore: turn off NODECHK

To: aelder@xxxxxxx
Subject: Re: [PATCH 1/2] xfsrestore: turn off NODECHK
From: Bill Kendall <wkendall@xxxxxxx>
Date: Fri, 15 Oct 2010 09:07:31 -0500
Cc: xfs@xxxxxxxxxxx
In-reply-to: <1287081911.2362.574.camel@doink>
References: <20101012215322.749700656@xxxxxxx> <20101012215400.430780299@xxxxxxx> <1287081911.2362.574.camel@doink>
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv: Gecko/20100915 Thunderbird/3.0.8
On 10/14/2010 01:45 PM, Alex Elder wrote:
On Tue, 2010-10-12 at 16:53 -0500, wkendall@xxxxxxx wrote:
plain text document attachment (no_node_check)
The NODECHK macro should only be enabled as needed for
development/debugging.  Having it on limits xfsrestore to
268 million directory entries instead of 4 billion.

I was going to suggest you leave it in the Makefile in
commented-out form, perhaps with an explanation that
it's available for development.  But since NODECHK is
only ever referenced inside "restore/node.c", maybe
such a thing could be done inside that file instead.

I.e., something like:

  * NODECHK can be enabled for development and debugging
  * in order to enforce additional checks on handles
  * when nodes are mapped, unmapped, and freed.  It should
  * not be used for production code, it cuts into the number
  * of directory entries xfsrestore can handle.
/* #define NODECHK */

I'll leave it up to you whether you think that's worth
doing (and if so, let me know).  Either way:

Reviewed-by: Alex Elder<aelder@xxxxxxx>

Sounds like a good idea. I'll add something like this
and repost.


Signed-off-by: Bill Kendall<wkendall@xxxxxxx>

Index: xfsdump-kernel.org/restore/Makefile
--- xfsdump-kernel.org.orig/restore/Makefile
+++ xfsdump-kernel.org/restore/Makefile
@@ -103,7 +103,7 @@ LLDLIBS = $(LIBUUID) $(LIBHANDLE) $(LIBA


  default: depend $(LTCOMMAND)

xfs mailing list

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