From mboxrd@z Thu Jan 1 00:00:00 1970 From: Joel Soete Subject: lstrncpy_from_user() additional question [Was: [parisc-linux] test case for pa_memset() ;-)] Date: Sun, 19 Dec 2004 20:34:34 +0000 Message-ID: <41C5E5DA.8000906@tiscali.be> References: <418A80250001032B@mail-2-bnl.tiscali.it> <20041216163417.GL29171@tausq.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Cc: parisc-linux@lists.parisc-linux.org To: Randolph Chung Return-Path: In-Reply-To: <20041216163417.GL29171@tausq.org> List-Id: parisc-linux developers list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: parisc-linux-bounces@lists.parisc-linux.org Randolph, I would like to get your opinion about the idea to re-write this lstrncpy_from_user() in c as you did with pa_memcpy() (may be easiest to maintain?) Thanks, Joel Randolph Chung wrote: >>Any interest? >> >>This tmset3.c is a very draft and all advise will be welcome :-) > > > Joel, i didn't notice earlier that you were working on memset.... i > don't think you'll see much improvement over our current implementation. > memcpy was a target for optimization because it used to copy > *one-byte-at-a-time*, but our memset already knows about byte boundaries > and tries to write more at a time when possible. i suspect you are > seeing slower results with your implementation because of increased code > size and branch hits. you're unlikely to see much improvement compared > to the current version unless you either unroll the loop or do some more > clever instruction scheduling. > > the three routines in arch/parisc/lib/lusercopy.S are probably better > targets for optimization because they are all doing one-byte-at-a-time > operations, so there is more room for improvement. > > randolph _______________________________________________ parisc-linux mailing list parisc-linux@lists.parisc-linux.org http://lists.parisc-linux.org/mailman/listinfo/parisc-linux