>>>>> On Wed, 25 Feb 2004 17:26:25 -0800, "Chen, Kenneth W" <kenneth.w.chen@intel.com> said: Ken> We believe we have resolved all the remain issues, all critical Ken> speed path has been taken care of, i.e., vhpt hander and Ken> context switch. There should be no performance penalty with Ken> this dynamic hugetlb page size feature. What about huge-page page-faults? Before, HPAGE_SHIFT, HPAGE_SIZE, and HPAGE_MASK all were constant, now they have to be fetched from memory/recalculated each time. I suppose it's rare enough that it's not worth worrying about but did you verify that there is at least not anything really gross going on (like someone doing a modulo operation against HPAGE_SIZE)? How about replacing 0x8000000000000000 with (REGION_HPAGE << REGION_SHIFT) or a manifest constant defined to this value (perhaps HUGETLB_REGION_ADDR)? Why does the region register get initialized both in mm/init.c and hugetlb_setup_sz()? Couldn't this be done once in hugetlb_init()? Other than that, the patch looks good to me. --david - To unsubscribe from this list: send the line "unsubscribe linux-ia64" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.htmlReceived on Thu Feb 26 00:19:21 2004
This archive was generated by hypermail 2.1.8 : 2005-08-02 09:20:22 EST