Import of Linux/MIPS 2.1.14.2

1 files changed, 65 insertions, 115 deletions

diff --git a/arch/mips/lib/checksum.c b/arch/mips/lib/checksum.c
index 0e04ac5e8..dd1583892 100644
--- a/arch/mips/lib/checksum.c
+++ b/arch/mips/lib/checksum.c
@@ -17,124 +17,76 @@
 #include <net/checksum.h>
 #include <asm/string.h>
 
+static inline unsigned short from32to16(unsigned long x)
+{
+	/* 32 bits --> 16 bits + carry */
+	x = (x & 0xffff) + (x >> 16);
+	/* 16 bits + carry --> 16 bits including carry */
+	x = (x & 0xffff) + (x >> 16);
+	return x;
+}
+
+static inline unsigned long do_csum(const unsigned char * buff, int len)
+{
+	int odd, count;
+	unsigned long result = 0;
+
+	if (len <= 0)
+		goto out;
+	odd = 1 & (unsigned long) buff;
+	if (odd) {
+		result = *buff;
+		len--;
+		buff++;
+	}
+	count = len >> 1;		/* nr of 16-bit words.. */
+	if (count) {
+		if (2 & (unsigned long) buff) {
+			result += *(unsigned short *) buff;
+			count--;
+			len -= 2;
+			buff += 2;
+		}
+		count >>= 1;		/* nr of 32-bit words.. */
+		if (count) {
+		        unsigned long carry = 0;
+			do {
+				unsigned long w = *(unsigned long *) buff;
+				count--;
+				buff += 4;
+				result += carry;
+				result += w;
+				carry = (w > result);
+			} while (count);
+			result += carry;
+			result = (result & 0xffff) + (result >> 16);
+		}
+		if (len & 2) {
+			result += *(unsigned short *) buff;
+			buff += 2;
+		}
+	}
+	if (len & 1)
+		result += (*buff << 8);
+	result = from32to16(result);
+	if (odd)
+		result = ((result >> 8) & 0xff) | ((result & 0xff) << 8);
+out:
+	return result;
+}
+
 /*
  * computes a partial checksum, e.g. for TCP/UDP fragments
  */
 unsigned int csum_partial(const unsigned char *buff, int len, unsigned int sum)
 {
-	unsigned long	scratch1;
-	unsigned long	scratch2;
-
-	/*
-	 * This is for 32-bit MIPS processors.
-	 */
-    __asm__("
-	.set	noreorder
-	.set	noat
-	andi	$1,%4,2		# Check alignment
-	beqz	$1,2f		# Branch if ok
-	nop			# delay slot
-	subu	$1,%3,2		# Alignment uses up two bytes
-	bgez	$1,1f		# Jump if we had at least two bytes
-	move	%3,$1		# delay slot
-	j	4f
-	addiu	%3,2		# delay slot; len was < 2.  Deal with it
-
-1:	lhu	%2,(%4)
-	addiu	%4,2
-	addu	%0,%2
-	sltu	$1,%0,%2
-	addu	%0,$1
-
-2:	srl	%1,%3,5
-	beqz	%1,2f
-	sll	%1,%1,5		# delay slot
-
-	addu	%1,%4
-1:	lw	%2,0(%4)
-	addu	%4,32
-	addu	%0,%2
-	sltu	$1,%0,%2
+	unsigned long result = do_csum(buff, len);
 
-	lw	%2,-28(%4)
-	addu	%0,$1
-	addu	%0,%2
-	sltu	$1,%0,%2
-
-	lw	%2,-24(%4)
-	addu	%0,$1
-	addu	%0,%2
-	sltu	$1,%0,%2
-
-	lw	%2,-20(%4)
-	addu	%0,$1
-	addu	%0,%2
-	sltu	$1,%0,%2
-
-	lw	%2,-16(%4)
-	addu	%0,$1
-	addu	%0,%2
-	sltu	$1,%0,%2
-
-	lw	%2,-12(%4)
-	addu	%0,$1
-	addu	%0,%2
-	sltu	$1,%0,%2
-
-	lw	%2,-8(%4)
-	addu	%0,$1
-	addu	%0,%2
-	sltu	$1,%0,%2
-
-	lw	%2,-4(%4)
-	addu	%0,$1
-	addu	%0,%2
-	sltu	$1,%0,%2
-
-	bne	%4,%1,1b
-	addu	%0,$1		# delay slot
-
-2:	andi	%1,%3,0x1c
-	beqz	%1,4f
-	addu	%1,%4		# delay slot
-3:	lw	%2,0(%4)
-	addu	%4,4
-	addu	%0,%2
-	sltu	$1,%0,%2
-	bne	%4,%1,3b
-	addu	%0,$1		# delay slot
-
-4:	andi	$1,%3,3
-	beqz	$1,7f
-	andi	$1,%3,2		# delay slot
-	beqz	$1,5f
-	move	%2,$0		# delay slot
-	lhu	%2,(%4)
-	addiu	%4,2		# delay slot
-
-5:	andi	$1,%3,1
-	beqz	$1,6f
-	nop			# delay slot
-	lbu	%1,(%4)
-	sll	%2,16\n\t"
-#ifdef __MIPSEB__
-	"sll	%1,8\n\t"
-#endif
-	"or	%2,%1
-6:	addu	%0,%2
-	sltu	$1,%0,%2
-	addu	%0,$1
-7:	.set	at
-	.set	reorder"
-	: "=r"(sum),
-	  "=&r" (scratch1),
-	  "=&r" (scratch2),
-	  "=r" (len),
-	  "=r" (buff)
-	: "0"(sum), "3"(len), "4"(buff)
-	: "$1");
-
-	return sum;
+	/* add in old sum, and carry.. */
+	result += sum;
+	if(sum > result)
+		result += 1;
+	return result;
 }
 
 /*
@@ -144,12 +96,10 @@ unsigned int csum_partial_copy(const char *src, char *dst,
 				  int len, unsigned int sum)
 {
 	/*
-	 * It's 2:30 am and I don't feel like doing it right ...
+	 * It's 2:30 am and I don't feel like doing it real ...
 	 * This is lots slower than the real thing (tm)
 	 *
 	 * XXX This may nuke the kernel for unaligned src addresses!!!
-	 *     (Due to software address error fixing no longer true, but
-	 *     seems to happen very rarely only anyway.)
 	 */
 	sum = csum_partial(src, len, sum);
 	memcpy(dst, src, len);