On Tue, 8 Aug 2006 10:48:50 -0400 Kyle McMartin <[EMAIL PROTECTED]> wrote:
> On second thought, you've compressed two different flushes into > flush_cache_vmap[1] where flush_tlb_all used to be... Yes, that's intentional. I did mention this as one of the differences from i386 in the "implementation" part of this series, but I should perhaps have mentioned it explicitly in each of the arch-specific patches as well? > I don't think this would be a problem because we're creating a > new mapping, but James will probably prove my assumption wrong > as usual. :) That's indeed the reasoning behind the change -- ioremap_pte_range() contains BUG_ON(!pte_none(*pte)) so we should never touch existing mappings. Also, the only difference I can see between ioremap() and vmap() is that ioremap() maps a contiguous range of physical addresses while vmap() maps an array of 'struct page's. So the cache- and tlb-flushing requirements should be the same, right? Haavard - To unsubscribe from this list: send the line "unsubscribe linux-arch" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
