xfs
[Top] [All Lists]

Re: xfsdump on alpha

To: ThH <thh@xxxxxxxxxxxx>
Subject: Re: xfsdump on alpha
From: ivanr@xxxxxxxxxxxxxxxxx (Ivan Rayner)
Date: Mon, 16 Jul 2001 16:41:10 +1000
Cc: <linux-xfs@xxxxxxxxxxx>
In-reply-to: <10107161617.ZM208491@wobbly.melbourne.sgi.com>
Sender: owner-linux-xfs@xxxxxxxxxxx
On Mon, 16 Jul 2001, Nathan Scott wrote:

> hi,
>
> On Jul 16,  7:53am, ThH wrote:
> > Subject: xfsdump on alpha
> > I've tried xfsdump-1.0.11 on my alpha. But there is the Assertion
> > problem again.
> >
> > xfsdump: version 3.0 - Running single-threaded
> > xfsdump: global.c:82 global_hdr_alloc: Assertion `sizeof( time_t ) == 4'
> > failed.
> > Aborted (core dumped)
> >
> > On alpha it is 8 instead of 4.
> >
> > I had corected this one Assertion manualy,
>
> Changing just the assert doesn't seem like the right thing
> to do - this looks like a data structure which gets written
> onto the tape, so it needs to be 4 bytes long.  Will require
> an audit of the xfs{dump/restore} code to check all of the
> uses of time_t, I guess - does that sound right, Ivan/Tim?

Sounds right to me.  If you changed the header to include 8 byte time_t
your dumps would be incompatible with anyone elses.  However, you may not
care about that so if it works, it may be fine for you.

Just be sure not to lose your version of the source and binaries!

If compatibilty is an issue, then you will need to do as Nathan suggests,
and check every use of time_t in the code, and do conversions where
appropriate.

Unfortunately, we don't have the hardware nor people to port or test an
alpha version, although, as always, we'd be happy to consider including
any patches submitted to us. :)

Ivan

-- 
Ivan Rayner
ivanr@xxxxxxxxxxxxxxxxx


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