matroxfb: fix incorrect use of memcpy_toio()

Screen is completely corrupted since 2.6.34.  Bisection revealed that it's
caused by commit 6175ddf0 ("x86: Clean up mem*io functions.").

H.  Peter Anvin explained that memcpy_toio() does not copy data in 32bit
chunks anymore on x86.
Signed-off-by: NOndrej Zary <linux@rainbow-software.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Petr Vandrovec <vandrove@vc.cvut.cz>
Cc: Jean Delvare <khali@linux-fr.org>
Cc: <stable@kernel.org>		[2.6.34.x, 2.6.35.x]
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

drivers/video/matrox/matroxfb_base.h

@@ -151,13 +151,13 @@ static inline void mga_writel(vaddr_t va, unsigned int offs, u_int32_t value) {
 static inline void mga_memcpy_toio(vaddr_t va, const void* src, int len) {
 #if defined(__alpha__) || defined(__i386__) || defined(__x86_64__)
 	/*
-	 * memcpy_toio works for us if:
+	 * iowrite32_rep works for us if:
 	 *  (1) Copies data as 32bit quantities, not byte after byte,
 	 *  (2) Performs LE ordered stores, and
 	 *  (3) It copes with unaligned source (destination is guaranteed to be page
 	 *      aligned and length is guaranteed to be multiple of 4).
 	 */
-	memcpy_toio(va.vaddr, src, len);
+	iowrite32_rep(va.vaddr, src, len >> 2);
 #else
         u_int32_t __iomem* addr = va.vaddr;