From mboxrd@z Thu Jan 1 00:00:00 1970 From: Brent Casavant Date: Tue, 24 Jan 2006 22:41:35 +0000 Subject: RE: [PATCH] SN2 user-MMIO CPU migration Message-Id: <20060124163651.O90635@chenjesu.americas.sgi.com> List-Id: References: <20060118163305.Y42462@chenjesu.americas.sgi.com> In-Reply-To: <20060118163305.Y42462@chenjesu.americas.sgi.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-ia64@vger.kernel.org On Tue, 24 Jan 2006, Chen, Kenneth W wrote: > Brent Casavant wrote on Tuesday, January 24, 2006 2:12 PM > > > On the other hand, it may not be too bad because on kernel exit, flags > > > field in thread_info is accessed for TIF_* and if kernel exit path is > > > taken back-to-back with context switch, the net effect maybe small. > > > > Actually, IA64_HAS_EXTRA_STATE() reads the TIF_* flags, and is called > > in __switch_to(), which is in very close proximity to checking last_cpu. > > Where?? > > IA64_HAS_EXTRA_STATE() uses task->thread.flags, not task->thread_info->flags. > These two flags are in two different places. Bhaa! You are right of course. My mistake. Still, other than moving last_cpu to neighbor cpu (or getting them moved to the thread_struct, as you inquired of Ingo), I'm not sure what can be done to further mitigate the cacheline impact. Brent -- Brent Casavant All music is folk music. I ain't bcasavan@sgi.com never heard a horse sing a song. Silicon Graphics, Inc. -- Louis Armstrong