This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] locks: rename file-private locks to "open file description locks"
- From: Boaz Harrosh <bharrosh at panasas dot com>
- To: Jeff Layton <jlayton at redhat dot com>
- Cc: <linux-fsdevel at vger dot kernel dot org>, <linux-kernel at vger dot kernel dot org>, <samba-technical at lists dot samba dot org>, Ganesha NFS List <nfs-ganesha-devel at lists dot sourceforge dot net>, libc-alpha <libc-alpha at sourceware dot org>, Michael Kerrisk <mtk dot manpages at gmail dot com>, Christoph Hellwig <hch at infradead dot org>, "Carlos O'Donell" <carlos at redhat dot com>, Stefan Metzmacher <metze at samba dot org>, Andy Lutomirski <luto at amacapital dot net>, Frank Filz <ffilzlnx at mindspring dot com>, "Theodore Ts'o" <tytso at mit dot edu>
- Date: Tue, 22 Apr 2014 18:05:13 +0300
- Subject: Re: [PATCH] locks: rename file-private locks to "open file description locks"
- Authentication-results: sourceware.org; auth=none
- References: <1398169383-10743-1-git-send-email-jlayton at redhat dot com> <5356808B dot 3040303 at gmail dot com> <20140422105520 dot 7b09eda5 at tlielax dot poochiereds dot net>
On 04/22/2014 05:55 PM, Jeff Layton wrote:
> On Tue, 22 Apr 2014 17:45:31 +0300
> Boaz Harrosh <openosd@gmail.com> wrote:
>
>
> There's a big difference between the descriptor and the description.
>
> The numerical value you get back from something like open() is a file
> descriptor.
Ha OK sorry I thought that was a file-handle I think in FBSD they
call it that. I guess file-handle is the NFS thing.
I guess the most common name for those is file-No, fn in
code
> The thing that that value points to internally in the
> kernel is the file description.
I did not know that and I completely interchanged these two.
In Kernel this is called plain "struct file" so I've never
seen this name used before.
OK Now it is clear
> It's very important that we do not
> conflate the two here as these locks are associated with the file
> description and not the file descriptor.
>
Sure you are right.
> The best way to illustrate this is the interaction with dup() -- see
> the LWN article on these for a complete overview.
>
I know the difference very well only I got the names mixed
Thanks
Boaz