From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Morton Subject: Re: [RESEND PATCH V3] binfmt_elf.c: use get_random_int() to fix entropy depleting Date: Wed, 14 Nov 2012 13:09:09 -0800 Message-ID: <20121114130909.2dcc5b75.akpm@linux-foundation.org> References: <5099F133.5030305@oracle.com> <5099FBAA.6000200@oracle.com> <509A078D.8060705@oracle.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: Jeff Liu , LKML , Andreas Dilger , John Sobecki , "viro@zeniv.linux.org.uk" , Alan Cox , "arnd@arndb.de" , James Morris , "Ted Ts'o" , "gregkh@linuxfoundation.org" , jakub@redhat.com, drepper@redhat.com, "linux-fsdevel@vger.kernel.org" To: Kees Cook Return-path: In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Tue, 6 Nov 2012 23:13:54 -0800 Kees Cook wrote: > On Tue, Nov 6, 2012 at 11:02 PM, Jeff Liu wrote: > > On 11/07/2012 02:21 PM, Kees Cook wrote: > >> I still want to hear at least from Ted about this changes -- we would > >> be potentially increasing the predictability of these bytes... > > > > We would not increasing that if this routine would be used for AT_RANDOM > > only(and if the array keeping aligned to 4 bytes). > > Otherwise, it would be, so let's waiting for further feedbacks. > > get_random_int() comes from a different pool than get_random_bytes(), > IIUC. I'd like to hear some convincing reasoning as to why this change > doesn't compromise predictability. :) But the original "ELF: implement AT_RANDOM for glibc PRNG seeding" compromised predictability. That's the whole point of this patch. What was so important about that patch that justified gobbling down so much of the system's entropy accumulation?