From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephane Eranian Date: Fri, 30 Jun 2006 10:24:35 +0000 Subject: Re: perfmon2 vector argument question Message-Id: <20060630102435.GA21819@frankl.hpl.hp.com> List-Id: References: <20060619204012.GE26378@frankl.hpl.hp.com> <20060628201708.08af034c.akpm@osdl.org> In-Reply-To: <20060628201708.08af034c.akpm@osdl.org> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Andrew Morton Cc: linux-kernel@vger.kernel.org, linux-ia64@vger.kernel.org, perfmon@napali.hpl.hp.com Andrew, On Wed, Jun 28, 2006 at 08:17:08PM -0700, Andrew Morton wrote: > > > > Does someone have something else to propose? > > > > If not, what is your opinion of the two approaches above? > > > > The first approach should be fine - we do that in lots of places, such as > in core_sys_select(). > Ok, that's good to know. I looked at the stack consumption on x86 and it is comparable to what you do for core_sys_select(). > Applications mut be calling this thing at a heck of a rate for kfree() > overhead to matter. I trust CONFIG_DEBUG_SLAB wasn't turned on... That was using a micro-benchmark to stress certain paths in perfmon. CONFIG_DEBUG_SLAB was not turned on. Thanks. -- -Stephane