iio: pressure: zpa2326: Remove always-true check which confuses gcc

With gcc 4.1.2:

    drivers/iio/pressure/zpa2326.c: In function ‘zpa2326_wait_oneshot_completion’:
    drivers/iio/pressure/zpa2326.c:868: warning: ‘ret’ may be used uninitialized in this function

When testing for "timeout < 0", timeout is already guaranteed to be
strict negative, so the branch is always taken, and ret is thus always
initialized.  But (some version of) gcc is not smart enough to notice.

Remove the check to fix this.
As there is no other code in between assigning the error codes and
returning them, the error codes can be returned immediately, and the
intermediate variable can be dropped.
Drop the "else" to please checkpatch.

Fixes: e7215fe4 ("iio: pressure: zpa2326: report interrupted case as failure")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

drivers/iio/pressure/zpa2326.c

@@ -865,7 +865,6 @@ static irqreturn_t zpa2326_handle_threaded_irq(int irq, void *data)
 static int zpa2326_wait_oneshot_completion(const struct iio_dev   *indio_dev,
 					   struct zpa2326_private *private)
 {
-	int          ret;
 	unsigned int val;
 	long     timeout;
 
@@ -887,14 +886,11 @@ static int zpa2326_wait_oneshot_completion(const struct iio_dev   *indio_dev,
 		/* Timed out. */
 		zpa2326_warn(indio_dev, "no one shot interrupt occurred (%ld)",
 			     timeout);
-		ret = -ETIME;
-	} else if (timeout < 0) {
-		zpa2326_warn(indio_dev,
-			     "wait for one shot interrupt cancelled");
-		ret = -ERESTARTSYS;
+		return -ETIME;
 	}
 
-	return ret;
+	zpa2326_warn(indio_dev, "wait for one shot interrupt cancelled");
+	return -ERESTARTSYS;
 }
 
 static int zpa2326_init_managed_irq(struct device          *parent,