From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Chen, Kenneth W" Date: Tue, 02 May 2006 17:40:37 +0000 Subject: RE: [RFC 1/3] LVHPT - Fault handler modifications Message-Id: <4t153d$t4guc@azsmga001.ch.intel.com> List-Id: In-Reply-To: <20060502052551.8990.16410.sendpatchset@wagner.orchestra.cse.unsw.EDU.AU> References: <20060502052551.8990.16410.sendpatchset@wagner.orchestra.cse.unsw.EDU.AU> In-Reply-To: <20060502052551.8990.16410.sendpatchset@wagner.orchestra.cse.unsw.EDU.AU> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: 'Ian Wienand' , linux-ia64@vger.kernel.org Cc: linux-mm@kvack.org Ian Wienand wrote on Monday, May 01, 2006 10:26 PM > Firstly, we have stripped out common code in ivt.S into assembler > macros in ivt-macro.S. The comments before the macros should explain > what each is doing. I think at current state, it's way too early to extract out the common code into macros, because for long format vhpt, the low level handler is not even optimal. And in the final form, it may not be the same as the short format vhpt (OK, the linux page table walk will be the same, but other part maybe not). I would like to experiment with a few algorithms for lvhpt and best thing to do in my opinion is to have parallel ivt.S (or ivt table to be precise). - Ken