From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnd Bergmann Subject: Re: [PATCH v2 1/5] add metadata_incore ioctl in vfs Date: Thu, 6 Jan 2011 08:38:48 +0100 Message-ID: <201101060838.49359.arnd@arndb.de> References: <1294119632.1949.366.camel@sli10-conroe> <201101051042.37181.arnd@arndb.de> <1294276437.1949.578.camel@sli10-conroe> Mime-Version: 1.0 Content-Type: Text/Plain; charset="utf-8" Content-Transfer-Encoding: 7bit Cc: "linux-btrfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" , "linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" , Chris Mason , Christoph Hellwig , Andrew Morton , Arjan van de Ven , "Yan, Zheng" , "linux-api-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" , "mtk.manpages-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org" To: Shaohua Li Return-path: In-Reply-To: <1294276437.1949.578.camel@sli10-conroe> Sender: linux-api-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: linux-fsdevel.vger.kernel.org On Thursday 06 January 2011, Shaohua Li wrote: > I don't understand. adding a case statement in compat_sys_ioctl, so we will do > compat_ioctl_check_table(). If I add COMPATIBLE_IOCTL(), then the check > will success, we will go to the found_handler code path and execute > do_vfs_ioctl, which is what we want. if not adding COMPATIBLE_IOCTL(), > the check will fail, and in any case, we will go to the out_fput code > path, so our ioctl does nothing. You are correct, I misremembered the code and did not check properly. > > Two more general comments: > > > > - You probably want to add the ioctls to file_ioctl instead of do_vfs_ioctl, > > so you don't add another case statement to the common path. > > > > - I don't know if there are any rules for what should be an ioctl or an > > fcntl, we're rather inconsistent about this. If you have found a good > > reason for making it an ioctl, just put that into the changelog so we > > can refer to it next time. > it can be applied to a directory too. I thought file_ioctl or fcntl is > for file. Right again, good point! Arnd