From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ingo Molnar Subject: Re: [PATCH] mm/x86: AMD Bulldozer ASLR fix Date: Sun, 29 Mar 2015 10:51:22 +0200 Message-ID: <20150329085122.GA25177@gmail.com> References: <20150326190800.GF27751@pd.tnic> <1427456301-3764-1-git-send-email-hecmargi@upv.es> <20150327144438.GA3254@pd.tnic> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Hector Marco-Gisbert , linux-kernel@vger.kernel.org, akpm@linux-foundation.org, Thomas Gleixner , Ingo Molnar , "H. Peter Anvin" , x86@kernel.org, Alexander Viro , Jan-Simon , linux-fsdevel@vger.kernel.org, kees Cook , Ismael Ripoll To: Borislav Petkov Return-path: Content-Disposition: inline In-Reply-To: <20150327144438.GA3254@pd.tnic> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org * Borislav Petkov wrote: > From: Hector Marco-Gisbert > Date: Fri, 27 Mar 2015 12:38:21 +0100 > Subject: [PATCH] x86/mm: Improve AMD Bulldozer ASLR fix > > The ASLR implementation needs to special-case AMD F15h processors by > clearing out bits [14:12] of the virtual address in order to avoid I$ > cross invalidations and thus performance penalty for certain workloads. > For details, see: > > dfb09f9b7ab0 ("x86, amd: Avoid cache aliasing penalties on AMD family 15h") > > This special case reduces the mmapped files entropy by eight. s/reduces the mmapped file's entropy by 3 bits Which does: - a grammar fix - measure it in bits, as later on we are talking about randomness in bits as well. Btw., does this limitation affect both executable and non-executable mmap()s? Because data mmap()s don't need this I$ related workaround, right? So we could relax it for data-mmap()s? Thanks, Ingo