Date: Sun, 29 May 2005 15:07:49 +0100 From: Doug Rabson <dfr@nlsystems.com> To: freebsd-arch@freebsd.org Cc: Robert Watson <ratson@freebsd.org>, arch@freebsd.org, Suleiman Souhlal <ssouhlal@freebsd.org> Subject: Re: [PATCH] Stackgap Message-ID: <200505291507.50289.dfr@nlsystems.com> In-Reply-To: <20050528150815.X29776@fledge.watson.org> References: <CC7E6E83-2C2D-46FF-A816-CAD6F16CDA1B@FreeBSD.org> <20050528150815.X29776@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Saturday 28 May 2005 15:10, Robert Watson wrote: > On Fri, 27 May 2005, Suleiman Souhlal wrote: > > You can find an implementation of stackgap from OpenBSD at http:// > > people.freebsd.org/~ssouhlal/testing/stackgap-20050527.diff > > > > You can control the range of the random stack gap with the > > kern.stackgap_random sysctl. A value of 0 disables it. Otherwise, > > it has to be a power of 2 and not too large. The default value is > > 64K. > > > > I've only had the chance to test this on i386. Could anyone test it > > on other architectures as well? > > > > Any comments/objections? > > In the past, substantial performance hits have been measured due to > poor stack alignment. Specifically, in combination with less optimal > compiler behavior, the results have been pretty nasty. Have you > tried micro-benchmarking a series of runs with this stack offset > randomness using floating point on stack arguments to see if there's > a measurable cost to moving the stack around? Hopefull if all is > well, there will be little or no difference, but a small error here > could result in a substantial performance hit... I recently modified the crt code to force the stack alignment to 16 bytes on startup (so that I could safely write code that uses movaps).
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200505291507.50289.dfr>