4.9-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Nikita Yushchenko <[email protected]>


[ Upstream commit 602d9858f07c72eab64f5f00e2fae55f9902cfbe ]

Some drivers do depend on page mappings to be page aligned.

Swiotlb already enforces such alignment for mappings greater than page,
extend that to page-sized mappings as well.

Without this fix, nvme hits BUG() in nvme_setup_prps(), because that routine
assumes page-aligned mappings.

Signed-off-by: Nikita Yushchenko <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Signed-off-by: Konrad Rzeszutek Wilk <[email protected]>
Signed-off-by: Sasha Levin <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
---
 lib/swiotlb.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

--- a/lib/swiotlb.c
+++ b/lib/swiotlb.c
@@ -456,11 +456,11 @@ phys_addr_t swiotlb_tbl_map_single(struc
                    : 1UL << (BITS_PER_LONG - IO_TLB_SHIFT);
 
        /*
-        * For mappings greater than a page, we limit the stride (and
-        * hence alignment) to a page size.
+        * For mappings greater than or equal to a page, we limit the stride
+        * (and hence alignment) to a page size.
         */
        nslots = ALIGN(size, 1 << IO_TLB_SHIFT) >> IO_TLB_SHIFT;
-       if (size > PAGE_SIZE)
+       if (size >= PAGE_SIZE)
                stride = (1 << (PAGE_SHIFT - IO_TLB_SHIFT));
        else
                stride = 1;


Reply via email to