drivers/memstick/host/r592.c: avoid gcc-6 warning

The r592 driver relies on behavior of the DMA mapping API that is
normally observed but not guaranteed by the API.  Instead it uses a
runtime check to fail transfers if the API ever behaves

When CONFIG_NEED_SG_DMA_LENGTH is not set, one of the checks turns into a
comparison of a variable with itself, which gcc-6.0 now warns about:

drivers/memstick/host/r592.c: In function 'r592_transfer_fifo_dma':
drivers/memstick/host/r592.c:302:31: error: self-comparison always evaluates to false [-Werror=tautological-compare]
    (sg_dma_len(&dev->req->sg) < dev->req->sg.length)) {
                               ^

The check itself is not a problem, so this patch just rephrases the
condition in a way that gcc does not consider an indication of a mistake.
We already know that dev->req->sg.length was initially R592_LFIFO_SIZE, so
we can compare it to that constant again.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Maxim Levitsky <maximlevitsky@gmail.com>
Cc: Quentin Lambert <lambert.quentin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

drivers/memstick/host/r592.c

@@ -298,8 +298,7 @@ static int r592_transfer_fifo_dma(struct r592_device *dev)
 	sg_count = dma_map_sg(&dev->pci_dev->dev, &dev->req->sg, 1, is_write ?
 		PCI_DMA_TODEVICE : PCI_DMA_FROMDEVICE);
 
-	if (sg_count != 1 ||
-			(sg_dma_len(&dev->req->sg) < dev->req->sg.length)) {
+	if (sg_count != 1 || sg_dma_len(&dev->req->sg) < R592_LFIFO_SIZE) {
 		message("problem in dma_map_sg");
 		return -EIO;
 	}