From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jack Steiner Date: Wed, 15 Feb 2006 15:08:23 +0000 Subject: Re: [PATCH] ia64: simplify and fix udelay() Message-Id: <20060215150823.GA27208@sgi.com> List-Id: References: <20060214184017.20492.48141.sendpatchset@tomahawk.engr.sgi.com> <200602150908.k1F98dg02934@unix-os.sc.intel.com> In-Reply-To: <200602150908.k1F98dg02934@unix-os.sc.intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: "Chen, Kenneth W" Cc: hawkes@sgi.com, Tony Luck , Andrew Morton , linux-ia64@vger.kernel.org, linux-kernel@vger.kernel.org, Robin Holt , Dimitri Sivanich , Jes Sorensen On Wed, Feb 15, 2006 at 01:08:41AM -0800, Chen, Kenneth W wrote: > hawkes@sgi.com wrote on Tuesday, February 14, 2006 10:40 AM > > a preemption and migration to another CPU during the > > while-loop > > Off topic from the subject line a bit, but related: how many Altix > SN2 customers in the field turn on CONFIG_PREEMPT? Redhat EL4 doesn't > turn on preempt, SuSE SLES9 and SLES10 beta don't turn it on either. > Is there a real benefit of turning that option on for SN2? AFAICT, no one at SGI uses or plans to use CONFIG_PREEMPT. Most of our customers use kernels from one of the distros & none at this point enables preemption. The realtime folks here have experimented with CONFIG_PREEMPT but so far have not seen any significant benefit. Regardless, we should fix udelay() to handle unsync'ed ITCs. It would be nice to have it working. -- Jack