---

yaml
---
r: 138066
b: refs/heads/master
c: 632fe9f
h: refs/heads/master
v: v3

[refs]

@@ -1,2 +1,2 @@
 ---
-refs/heads/master: 7c9e34aaab50a6c8f69ce59816dd76a283090667
+refs/heads/master: 632fe9fe440249642845675d97436c667cbbd21e

trunk/drivers/media/common/saa7146_i2c.c

@@ -293,7 +293,6 @@ static int saa7146_i2c_transfer(struct saa7146_dev *dev, const struct i2c_msg *m
 	int i = 0, count = 0;
 	__le32 *buffer = dev->d_i2c.cpu_addr;
 	int err = 0;
-	int address_err = 0;
 	int short_delay = 0;
 
 	if (mutex_lock_interruptible(&dev->i2c_lock))
@@ -333,17 +332,10 @@ static int saa7146_i2c_transfer(struct saa7146_dev *dev, const struct i2c_msg *m
 				   i2c address probing, however, and address errors indicate that a
 				   device is really *not* there. retrying in that case
 				   increases the time the device needs to probe greatly, so
-				   it should be avoided. because of the fact, that only
-				   analog based cards use irq based i2c transactions (for dvb
-				   cards, this screwes up other interrupt sources), we bail out
-				   completely for analog cards after an address error and trust
-				   the saa7146 address error detection. */
-				if ( -EREMOTEIO == err ) {
-					if( 0 != (SAA7146_USE_I2C_IRQ & dev->ext->flags)) {
-						goto out;
-					}
-					address_err++;
-				}
+				   it should be avoided. So we bail out in irq mode after an
+				   address error and trust the saa7146 address error detection. */
+				if (-EREMOTEIO == err && 0 != (SAA7146_USE_I2C_IRQ & dev->ext->flags))
+					goto out;
 				DEB_I2C(("error while sending message(s). starting again.\n"));
 				break;
 			}
@@ -358,10 +350,9 @@ static int saa7146_i2c_transfer(struct saa7146_dev *dev, const struct i2c_msg *m
 
 	} while (err != num && retries--);
 
-	/* if every retry had an address error, exit right away */
-	if (address_err == retries) {
+	/* quit if any error occurred */
+	if (err != num)
 		goto out;
-	}
 
 	/* if any things had to be read, get the results */
 	if ( 0 != saa7146_i2c_msg_cleanup(msgs, num, buffer)) {