xfs
[Top] [All Lists]

Re: TAKE - Compile fixes, memory allocation improvements.

To: Daniel Moore <dxm@xxxxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: TAKE - Compile fixes, memory allocation improvements.
From: Steve Lord <lord@xxxxxxx>
Date: Wed, 19 Apr 2000 21:51:17 -0500
Cc: Russell Cattelan <n8994@xxxxxxx>, linux-xfs@xxxxxxxxxxx
In-reply-to: Your message of "Thu, 20 Apr 2000 11:51:35 +1000
Sender: owner-linux-xfs@xxxxxxxxxxx
> 
> Russell Cattelan writes:
> 
>  => linux/fs/xfs/xfs_fsops.c - 1.51
http://oss.sgi.com/cgi-bin/cvsweb.cgi/linux-2.4-xfs/>  => 
linux/fs/xfs/xfs_fsops.c.diff?r1=text&tr1=1.51&r2=text&tr2=1.50&f=h
>  => linux/fs/xfs/xfs_inode.c - 1.280
http://oss.sgi.com/cgi-bin/cvsweb.cgi/linux-2.4-xfs/>  => 
linux/fs/xfs/xfs_inode.c.diff?r1=text&tr1=1.280&r2=text&tr2=1.279&f=h
>  => linux/fs/xfs/xfs_log_recover.c - 1.175
http://oss.sgi.com/cgi-bin/cvsweb.cgi/linux-2.4-xfs/>  => 
linux/fs/xfs/xfs_log_recover.c.diff?r1=text&tr1=1.175&r2=text&tr2=1.174&f=h
>  => linux/fs/xfs/xfs_mount.c - 1.217
http://oss.sgi.com/cgi-bin/cvsweb.cgi/linux-2.4-xfs/>  => 
linux/fs/xfs/xfs_mount.c.diff?r1=text&tr1=1.217&r2=text&tr2=1.216&f=h
>  => linux/fs/xfs/xfs_os_defs.h - 1.5
http://oss.sgi.com/cgi-bin/cvsweb.cgi/linux-2.4-xfs/>  => 
linux/fs/xfs/xfs_os_defs.h.diff?r1=text&tr1=1.5&r2=text&tr2=1.4&f=h
>  =>   - Changed kmem_realloc to include oldsize.
> 
> This change breaks sim. 
> 
> Adding the extra parameter to the definition in sim.random.c
> and removing the SIM conditional define from xfs_os_defs.h fixes
> the problem.
> 
> I haven't made the change because I'm not sure that's what you
> want to do.
> 
> -----------------------------------------------------
>  Daniel Moore                  dxm@xxxxxxx
>  R&D Software Engineer         Phone: +61-3-98348209
>  SGI Performance Tools Group   Fax:   +61-3-98132378
> -----------------------------------------------------


If there are no other calls in user space which do not get confused by this
then go ahead and check in what you described. The kernel needs the extra
parameter to make some realloc cases work. Seems better to keep the same
api in both places than to ifdef header files.

Lets keep broken tree time to a minimum.

Steve


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