swiotlb: map_page fix for highmem systems
Becky Bruce [Wed, 8 Apr 2009 14:09:17 +0000 (09:09 -0500)]
The current code calls virt_to_phys() on address that might
be in highmem, which is bad.  This wasn't needed, anyway, because
we already have the physical address we need.

Get rid of the now-unused virtual address as well.

Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org>
Acked-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Cc: jeremy@goop.org
Cc: ian.campbell@citrix.com
LKML-Reference: <1239199761-22886-4-git-send-email-galak@kernel.crashing.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

lib/swiotlb.c

index 4fd6a76..e8a47c8 100644 (file)
@@ -640,7 +640,6 @@ dma_addr_t swiotlb_map_page(struct device *dev, struct page *page,
                            struct dma_attrs *attrs)
 {
        phys_addr_t phys = page_to_phys(page) + offset;
-       void *ptr = page_address(page) + offset;
        dma_addr_t dev_addr = swiotlb_phys_to_bus(dev, phys);
        void *map;
 
@@ -651,7 +650,7 @@ dma_addr_t swiotlb_map_page(struct device *dev, struct page *page,
         * buffering it.
         */
        if (!address_needs_mapping(dev, dev_addr, size) &&
-           !range_needs_mapping(virt_to_phys(ptr), size))
+           !range_needs_mapping(phys, size))
                return dev_addr;
 
        /*