ASoC: davinci-mcasp: Correct rx format unit configuration
authorPeter Ujfalusi <peter.ujfalusi@ti.com>
Thu, 4 Sep 2014 07:52:53 +0000 (10:52 +0300)
committerMark Brown <broonie@kernel.org>
Thu, 4 Sep 2014 11:44:49 +0000 (12:44 +0100)
In case of capture we should not use rotation. The reverse and mask is
enough to get the data align correctly from the bus to MCU:
Format   data from bus    after reverse (XRBUF)
S16_LE:  |LSB|MSB|xxx|xxx|  |xxx|xxx|MSB|LSB|
S24_3LE: |LSB|DAT|MSB|xxx|  |xxx|MSB|DAT|LSB|
S24_LE:  |LSB|DAT|MSB|xxx|  |xxx|MSB|DAT|LSB|
S32_LE:  |LSB|DAT|DAT|MSB|  |MSB|DAT|DAT|LSB|

With this patch all supported formats will work for playback and capture.

Reported-by: Jyri Sarha <jsarha@ti.com> (broken S24_3LE capture)
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
sound/soc/davinci/davinci-mcasp.c

index c28508da34cf3ba4354ba6d54de685a03d112ca4..0062601a63c25f140003215878573746460a8b24 100644 (file)
@@ -459,8 +459,17 @@ static int davinci_config_channel_size(struct davinci_mcasp *mcasp,
 {
        u32 fmt;
        u32 tx_rotate = (word_length / 4) & 0x7;
-       u32 rx_rotate = (32 - word_length) / 4;
        u32 mask = (1ULL << word_length) - 1;
+       /*
+        * For captured data we should not rotate, inversion and masking is
+        * enoguh to get the data to the right position:
+        * Format         data from bus         after reverse (XRBUF)
+        * S16_LE:      |LSB|MSB|xxx|xxx|       |xxx|xxx|MSB|LSB|
+        * S24_3LE:     |LSB|DAT|MSB|xxx|       |xxx|MSB|DAT|LSB|
+        * S24_LE:      |LSB|DAT|MSB|xxx|       |xxx|MSB|DAT|LSB|
+        * S32_LE:      |LSB|DAT|DAT|MSB|       |MSB|DAT|DAT|LSB|
+        */
+       u32 rx_rotate = 0;
 
        /*
         * if s BCLK-to-LRCLK ratio has been configured via the set_clkdiv()