[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1464121530.31269.86.camel@linux.intel.com>
Date: Tue, 24 May 2016 23:25:30 +0300
From: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
To: Andrew Morton <akpm@...ux-foundation.org>
Cc: Theodore Ts'o <tytso@....edu>, Arnd Bergmann <arnd@...db.de>,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH v1 1/1] sysctl: introduce uuid_le and uuid_be
On Tue, 2016-05-24 at 13:15 -0700, Andrew Morton wrote:
> On Tue, 24 May 2016 20:27:27 +0300 Andy Shevchenko <andriy.shevchenko@
> linux.intel.com> wrote:
>
> > By default the sysctl interface returns random UUID in big endian
> > format.
> > Sometimes it's not suitable, e.g. using generated UUID for EFI
> > variable name.
> > Provide uuid_le and uuid_be to comprehence that interface.
> >
> > ...
> >
> > drivers/char/random.c | 44
> > +++++++++++++++++++++++++++++++++++++++++
> > include/uapi/linux/sysctl.h | 4 +++-
> > kernel/sysctl_binary.c | 48
> > ++++++++++++++++++++++++++++++++++++++++++++
>
> Why does the kernel need to do this? If userspace wants a random uuid
> then it can grab a random number and cook up the UUID itself.
Good question. I dunno why we have uuid in the first place there. At
some point I noticed people who recommend to use that file to get a
random UUID(s), but apparently they don't aware that endianess matters.
P.S. I thought to send this as RFC, but decided to drop the notation in
last minute. So, it's okay to drop it.
--
Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
Intel Finland Oy
Powered by blists - more mailing lists