From owner-cvs-all Sun Jun 24 12:40:47 2001 Delivered-To: cvs-all@freebsd.org Received: from earth.backplane.com (earth-nat-cw.backplane.com [208.161.114.67]) by hub.freebsd.org (Postfix) with ESMTP id B747137B401; Sun, 24 Jun 2001 12:40:43 -0700 (PDT) (envelope-from dillon@earth.backplane.com) Received: (from dillon@localhost) by earth.backplane.com (8.11.3/8.11.2) id f5OJegE79441; Sun, 24 Jun 2001 12:40:42 -0700 (PDT) (envelope-from dillon) Date: Sun, 24 Jun 2001 12:40:42 -0700 (PDT) From: Matt Dillon Message-Id: <200106241940.f5OJegE79441@earth.backplane.com> To: "Alfred Perlstein" Cc: Mike Silbersack , Mike Silbersack , cvs-committers@FreeBSD.ORG, cvs-all@FreeBSD.ORG, jlemon@FreeBSD.ORG, bmilekic@FreeBSD.ORG Subject: Re: cvs commit: src/sys/netinet tcp_input.c tcp_output.c tcp_subr.c tcp_timer.c tcp_usrreq.c tcp_var.h References: <200106231730.f5NHUNd73058@earth.backplane.com> <20010623131152.I23528-100000@achilles.silby.com> <20010624135828.A64836@sneakerz.org> Sender: owner-cvs-all@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG :bzero seems to be optimized for large areas, perhaps it would help :malloc some if we used some alternative zero'ing function for small :allocations with M_ZERO set? : :-Alfred It would be kinda messy, because 'size' is not a constant inside malloc() so there is no easy way to optimize out the overhead of the bzero() call or the conditionals within it. On the flip side, at least with the kernel malloc you *know* things are aligned and we know things are in chunks of at least sizeof(int), which means we could implement a (non-inline) bzerol() call which does away with a lot of the garbage sitting in i586_bzero(), cutting the overhead for zeroing small allocations in half. -Matt To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message