patch "USB: pl2303: fix baud-rate divisor calculations" added to usb-next

This is a note to let you know that I've just added the patch titled

USB: pl2303: fix baud-rate divisor calculations

to my usb git tree which can be found at
git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb.git
in the usb-next branch.

The patch will show up in the next release of the linux-next tree (usually sometime within the next 24 hours during the week.)

The patch will also be merged in the next major kernel release during the merge window.

If you have any questions about this process, please let me know.

From: =?UTF-8?q?Micha=C5=82=20Pecio?= <[email protected]>
Date: Sun, 26 Jul 2015 11:14:34 +0200
Subject: USB: pl2303: fix baud-rate divisor calculations

This commit fixes the following issues:

1. The 9th bit of buf was believed to be the LSB of divisor's exponent, but the hardware interprets it as MSB (9th bit) of the mantissa. The exponent is actually one bit shorter and applies to base 4, not 2 as previously believed.

2. Loop iterations doubled the exponent instead of incrementing.

3. The exponent wasn't checked for overflow.

4. The function returned requested rate instead of actual rate.

Due to issue #2, the old code deviated from the wrong formula described in #1 and actually yielded correct rates when divisor was lower than 4096 by using exponents of 0, 2 or 4 base-2, interpreted as 0, 1, 2 base-4 with the 9th mantissa bit clear. However, at 93.75 kbaud or less the rate turned out too slow due to #2 or too fast due to #2 and #3.

I tested this patch by sending and validating 0x00,0x01,..,0xff to an FTDI dongle at 234, 987, 2401, 9601, 31415, 115199, 250k, 500k, 750k, 1M, 1.5M, 3M+1 baud. All rates passed.

I also used pv to check speed at some rates unsupported by FTDI:
45 (the lowest possible), 2M, 4M, 5M and 6M-1. Looked sane.

rates")

---
 drivers/usb/serial/pl2303.c | 35 ++++++++++++++++++++++++++---------
 1 file changed, 26 insertions(+), 9 deletions(-)

index f5257af33ecf..ae682e4eeaef 100644
@@ -362,21 +362,38 @@ static speed_t pl2303_encode_baud_rate_direct(unsigned char buf[4],
 static speed_t pl2303_encode_baud_rate_divisor(unsigned char buf[4],
 								speed_t baud)
 {
-	unsigned int tmp;
+	unsigned int baseline, mantissa, exponent;
 
 	/*
 	 * Apparently the formula is:
-	 * baudrate = 12M * 32 / (2^buf[1]) / buf[0]
+	 *   baudrate = 12M * 32 / (mantissa * 4^exponent)
+	 * where
+	 *   mantissa = buf[8:0]
+	 *   exponent = buf[11:9]
 	 */
-	tmp = 12000000 * 32 / baud;
+	baseline = 12000000 * 32;
+	mantissa = baseline / baud;
+	if (mantissa == 0)
+		mantissa = 1;	/* Avoid dividing by zero if baud > 32*12M. */
+	exponent = 0;
+	while (mantissa >= 512) {
+		if (exponent < 7) {
+			mantissa >>= 2;	/* divide by 4 */
+			exponent++;
+		} else {
+			/* Exponent is maxed. Trim mantissa and leave. */
+			mantissa = 511;
+			break;
+		}
+	}
+
 	buf[3] = 0x80;
 	buf[2] = 0;
-	buf[1] = (tmp >= 256);
-	while (tmp >= 256) {
-		tmp >>= 2;
-		buf[1] <<= 1;
-	}
-	buf[0] = tmp;
+	buf[1] = exponent << 1 | mantissa >> 8;
+	buf[0] = mantissa & 0xff;
+
+	/* Calculate and return the exact baud rate. */
+	baud = (baseline / mantissa) >> (exponent << 1);
 
 	return baud;
 }

2.5.0