xfs
[Top] [All Lists]

Re: TAKE - xfsdump/restore changes for ia64

To: Timothy Shimmin <tes@xxxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: TAKE - xfsdump/restore changes for ia64
From: Takayuki Sasaki <sasaki@xxxxxxxxxxxxxxxxxx>
Date: Mon, 01 Oct 2001 16:36:34 +0900
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: Your message of Mon, 01 Oct 2001 17:03:29 +1000. <20011001170329.A1372@boing.melbourne.sgi.com>
Sender: owner-linux-xfs@xxxxxxxxxxx
Hi Timothy,

Thank you for your time.

Timothy Shimmin <tes@xxxxxxxxxxxxxxxxxxxxxxx>  wrote:
> Hi Takayuki,
> 
> On Mon, Oct 01, 2001 at 03:06:20PM +0900, Takayuki Sasaki wrote:
> > 
> > FSG QA <fsgqa@xxxxxxxxxxxxxxxxxxxxxxx>  wrote:
> > > Date:  Fri Sep 28 02:49:27 PDT 2001
> > > Workarea:  snort.melbourne.sgi.com:/diskb/build4/fsgqa/isms/2.4.x-xfs
> > (snip)
> > > cmd/xfsdump/common/stkchk.c - 1.2
> > >   - make a size long so comparison works for ia32/ia64 with ptr
> > 
> > I have a question.
> > 
> > The type of sc_sz in struct stkchk is changed to long from int,
> > but get_stacksz() is left declared to return int. 
> > 
> > stkchk.c line 90:
> >     stkchkp->sc_sz = get_stacksz( );
> > 
> > Further more, I'm wondering because it seems that rlim_cur is
> > defined as unsigned long.
> > 
> > [My box is RedHat 7.1]
> > $ grep rlim_cur /usr/include/*/*.h
> > /usr/include/bits/resource.h:    rlim_t rlim_cur;
> > /usr/include/bits/resource.h:    rlim64_t rlim_cur;
> > (snip)
> > /usr/include/linux/resource.h:  unsigned long   rlim_cur;
> > $ grep rlim_t /usr/include/*/*.h
> > /usr/include/bits/resource.h:typedef __rlim_t rlim_t;
> > /usr/include/bits/resource.h:typedef __rlim64_t rlim_t;
> > /usr/include/bits/resource.h:    rlim_t rlim_cur;
> > /usr/include/bits/resource.h:    rlim_t rlim_max;
> > /usr/include/bits/types.h:typedef __u_long __rlim_t; /* Type of resource 
> > counts.  */
> > 
> > Which is correct?
> > 
> Well, from this it seems it should be declared as : unsigned long.
> But I have a better solution forthcoming.
> 
> This abstraction (one of too many in xfsdump) has a stack checking function
> stkchk() which is only called in main.c in a function "stkplay" 
> (it calls itself recursively until it gets stack overflow and outputs
> at what address this happens at...unsure how useful this is)
> which is NEVER compiled in (#ifdef NEVER).

Yes, shkchk() is nerver compiled, however, this is in
stkchk_register()... Could I disregard for this?

Thanks in advance,

Takayuki


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